{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.0191822311963654,
  "eval_steps": 500,
  "global_step": 44000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 4.589050525446285e-05,
      "grad_norm": 1.3403533697128296,
      "learning_rate": 7.645259938837922e-09,
      "loss": 0.5188,
      "step": 1
    },
    {
      "epoch": 9.17810105089257e-05,
      "grad_norm": 1.7611314058303833,
      "learning_rate": 1.5290519877675843e-08,
      "loss": 0.5457,
      "step": 2
    },
    {
      "epoch": 0.00013767151576338856,
      "grad_norm": 0.9028812050819397,
      "learning_rate": 2.2935779816513765e-08,
      "loss": 0.4036,
      "step": 3
    },
    {
      "epoch": 0.0001835620210178514,
      "grad_norm": 0.4214492440223694,
      "learning_rate": 3.0581039755351686e-08,
      "loss": 0.2921,
      "step": 4
    },
    {
      "epoch": 0.00022945252627231425,
      "grad_norm": 0.8581069707870483,
      "learning_rate": 3.8226299694189604e-08,
      "loss": 0.4235,
      "step": 5
    },
    {
      "epoch": 0.0002753430315267771,
      "grad_norm": 0.9656532406806946,
      "learning_rate": 4.587155963302753e-08,
      "loss": 0.4195,
      "step": 6
    },
    {
      "epoch": 0.00032123353678124,
      "grad_norm": 2.2578439712524414,
      "learning_rate": 5.351681957186545e-08,
      "loss": 0.6178,
      "step": 7
    },
    {
      "epoch": 0.0003671240420357028,
      "grad_norm": 2.229050636291504,
      "learning_rate": 6.116207951070337e-08,
      "loss": 0.5685,
      "step": 8
    },
    {
      "epoch": 0.0004130145472901657,
      "grad_norm": 1.2387725114822388,
      "learning_rate": 6.88073394495413e-08,
      "loss": 0.5159,
      "step": 9
    },
    {
      "epoch": 0.0004589050525446285,
      "grad_norm": 1.74851393699646,
      "learning_rate": 7.645259938837921e-08,
      "loss": 0.5221,
      "step": 10
    },
    {
      "epoch": 0.0005047955577990914,
      "grad_norm": 3.0931506156921387,
      "learning_rate": 8.409785932721712e-08,
      "loss": 0.7711,
      "step": 11
    },
    {
      "epoch": 0.0005506860630535542,
      "grad_norm": 0.7829667925834656,
      "learning_rate": 9.174311926605506e-08,
      "loss": 0.4439,
      "step": 12
    },
    {
      "epoch": 0.0005965765683080171,
      "grad_norm": 2.237008571624756,
      "learning_rate": 9.938837920489297e-08,
      "loss": 0.5605,
      "step": 13
    },
    {
      "epoch": 0.00064246707356248,
      "grad_norm": 1.2005207538604736,
      "learning_rate": 1.070336391437309e-07,
      "loss": 0.5013,
      "step": 14
    },
    {
      "epoch": 0.0006883575788169427,
      "grad_norm": 1.6844494342803955,
      "learning_rate": 1.1467889908256882e-07,
      "loss": 0.551,
      "step": 15
    },
    {
      "epoch": 0.0007342480840714056,
      "grad_norm": 1.7097495794296265,
      "learning_rate": 1.2232415902140674e-07,
      "loss": 0.5089,
      "step": 16
    },
    {
      "epoch": 0.0007801385893258685,
      "grad_norm": 1.409144639968872,
      "learning_rate": 1.2996941896024464e-07,
      "loss": 0.4896,
      "step": 17
    },
    {
      "epoch": 0.0008260290945803313,
      "grad_norm": 3.0614712238311768,
      "learning_rate": 1.376146788990826e-07,
      "loss": 0.7174,
      "step": 18
    },
    {
      "epoch": 0.0008719195998347942,
      "grad_norm": 0.8959612250328064,
      "learning_rate": 1.452599388379205e-07,
      "loss": 0.4232,
      "step": 19
    },
    {
      "epoch": 0.000917810105089257,
      "grad_norm": 1.8896371126174927,
      "learning_rate": 1.5290519877675842e-07,
      "loss": 0.5492,
      "step": 20
    },
    {
      "epoch": 0.0009637006103437199,
      "grad_norm": 2.0700249671936035,
      "learning_rate": 1.6055045871559634e-07,
      "loss": 0.601,
      "step": 21
    },
    {
      "epoch": 0.0010095911155981827,
      "grad_norm": 2.7334303855895996,
      "learning_rate": 1.6819571865443424e-07,
      "loss": 0.6509,
      "step": 22
    },
    {
      "epoch": 0.0010554816208526456,
      "grad_norm": 1.7926756143569946,
      "learning_rate": 1.758409785932722e-07,
      "loss": 0.5735,
      "step": 23
    },
    {
      "epoch": 0.0011013721261071085,
      "grad_norm": 2.588320255279541,
      "learning_rate": 1.8348623853211012e-07,
      "loss": 0.6498,
      "step": 24
    },
    {
      "epoch": 0.0011472626313615713,
      "grad_norm": 0.9466708302497864,
      "learning_rate": 1.9113149847094802e-07,
      "loss": 0.4235,
      "step": 25
    },
    {
      "epoch": 0.0011931531366160342,
      "grad_norm": 2.035421371459961,
      "learning_rate": 1.9877675840978594e-07,
      "loss": 0.5886,
      "step": 26
    },
    {
      "epoch": 0.001239043641870497,
      "grad_norm": 1.3734371662139893,
      "learning_rate": 2.064220183486239e-07,
      "loss": 0.5093,
      "step": 27
    },
    {
      "epoch": 0.00128493414712496,
      "grad_norm": 2.2274961471557617,
      "learning_rate": 2.140672782874618e-07,
      "loss": 0.6198,
      "step": 28
    },
    {
      "epoch": 0.0013308246523794226,
      "grad_norm": 0.7679215669631958,
      "learning_rate": 2.2171253822629971e-07,
      "loss": 0.4081,
      "step": 29
    },
    {
      "epoch": 0.0013767151576338855,
      "grad_norm": 2.2280468940734863,
      "learning_rate": 2.2935779816513764e-07,
      "loss": 0.5711,
      "step": 30
    },
    {
      "epoch": 0.0014226056628883483,
      "grad_norm": 1.821223258972168,
      "learning_rate": 2.3700305810397554e-07,
      "loss": 0.5237,
      "step": 31
    },
    {
      "epoch": 0.0014684961681428112,
      "grad_norm": 1.4683890342712402,
      "learning_rate": 2.446483180428135e-07,
      "loss": 0.5886,
      "step": 32
    },
    {
      "epoch": 0.001514386673397274,
      "grad_norm": 1.8166035413742065,
      "learning_rate": 2.522935779816514e-07,
      "loss": 0.5846,
      "step": 33
    },
    {
      "epoch": 0.001560277178651737,
      "grad_norm": 2.1721227169036865,
      "learning_rate": 2.599388379204893e-07,
      "loss": 0.5962,
      "step": 34
    },
    {
      "epoch": 0.0016061676839061998,
      "grad_norm": 2.8068950176239014,
      "learning_rate": 2.6758409785932726e-07,
      "loss": 0.7705,
      "step": 35
    },
    {
      "epoch": 0.0016520581891606627,
      "grad_norm": 0.9062060117721558,
      "learning_rate": 2.752293577981652e-07,
      "loss": 0.3827,
      "step": 36
    },
    {
      "epoch": 0.0016979486944151256,
      "grad_norm": 2.020232677459717,
      "learning_rate": 2.8287461773700306e-07,
      "loss": 0.6352,
      "step": 37
    },
    {
      "epoch": 0.0017438391996695884,
      "grad_norm": 0.4772844612598419,
      "learning_rate": 2.90519877675841e-07,
      "loss": 0.3533,
      "step": 38
    },
    {
      "epoch": 0.0017897297049240513,
      "grad_norm": 1.5592962503433228,
      "learning_rate": 2.9816513761467896e-07,
      "loss": 0.5481,
      "step": 39
    },
    {
      "epoch": 0.001835620210178514,
      "grad_norm": 1.48556387424469,
      "learning_rate": 3.0581039755351683e-07,
      "loss": 0.519,
      "step": 40
    },
    {
      "epoch": 0.0018815107154329768,
      "grad_norm": 1.432593584060669,
      "learning_rate": 3.1345565749235476e-07,
      "loss": 0.5216,
      "step": 41
    },
    {
      "epoch": 0.0019274012206874397,
      "grad_norm": 1.971846580505371,
      "learning_rate": 3.211009174311927e-07,
      "loss": 0.5861,
      "step": 42
    },
    {
      "epoch": 0.001973291725941903,
      "grad_norm": 1.8548887968063354,
      "learning_rate": 3.287461773700306e-07,
      "loss": 0.5152,
      "step": 43
    },
    {
      "epoch": 0.0020191822311963654,
      "grad_norm": 1.3518904447555542,
      "learning_rate": 3.363914373088685e-07,
      "loss": 0.4928,
      "step": 44
    },
    {
      "epoch": 0.0020650727364508285,
      "grad_norm": 1.132017970085144,
      "learning_rate": 3.4403669724770646e-07,
      "loss": 0.4917,
      "step": 45
    },
    {
      "epoch": 0.002110963241705291,
      "grad_norm": 1.535001516342163,
      "learning_rate": 3.516819571865444e-07,
      "loss": 0.4899,
      "step": 46
    },
    {
      "epoch": 0.002156853746959754,
      "grad_norm": 0.6526742577552795,
      "learning_rate": 3.5932721712538226e-07,
      "loss": 0.4051,
      "step": 47
    },
    {
      "epoch": 0.002202744252214217,
      "grad_norm": 1.3923543691635132,
      "learning_rate": 3.6697247706422023e-07,
      "loss": 0.5483,
      "step": 48
    },
    {
      "epoch": 0.0022486347574686796,
      "grad_norm": 0.6318294405937195,
      "learning_rate": 3.7461773700305816e-07,
      "loss": 0.406,
      "step": 49
    },
    {
      "epoch": 0.0022945252627231427,
      "grad_norm": 2.7209126949310303,
      "learning_rate": 3.8226299694189603e-07,
      "loss": 0.687,
      "step": 50
    },
    {
      "epoch": 0.0023404157679776053,
      "grad_norm": 1.347872018814087,
      "learning_rate": 3.89908256880734e-07,
      "loss": 0.5432,
      "step": 51
    },
    {
      "epoch": 0.0023863062732320684,
      "grad_norm": 1.8943320512771606,
      "learning_rate": 3.975535168195719e-07,
      "loss": 0.6615,
      "step": 52
    },
    {
      "epoch": 0.002432196778486531,
      "grad_norm": 0.7322803139686584,
      "learning_rate": 4.051987767584098e-07,
      "loss": 0.4356,
      "step": 53
    },
    {
      "epoch": 0.002478087283740994,
      "grad_norm": 1.2317309379577637,
      "learning_rate": 4.128440366972478e-07,
      "loss": 0.547,
      "step": 54
    },
    {
      "epoch": 0.002523977788995457,
      "grad_norm": 2.214301586151123,
      "learning_rate": 4.2048929663608565e-07,
      "loss": 0.5578,
      "step": 55
    },
    {
      "epoch": 0.00256986829424992,
      "grad_norm": 1.2360845804214478,
      "learning_rate": 4.281345565749236e-07,
      "loss": 0.5632,
      "step": 56
    },
    {
      "epoch": 0.0026157587995043825,
      "grad_norm": 2.4533538818359375,
      "learning_rate": 4.357798165137615e-07,
      "loss": 0.6488,
      "step": 57
    },
    {
      "epoch": 0.002661649304758845,
      "grad_norm": 2.7331459522247314,
      "learning_rate": 4.4342507645259943e-07,
      "loss": 0.6527,
      "step": 58
    },
    {
      "epoch": 0.0027075398100133083,
      "grad_norm": 1.1070164442062378,
      "learning_rate": 4.510703363914373e-07,
      "loss": 0.4592,
      "step": 59
    },
    {
      "epoch": 0.002753430315267771,
      "grad_norm": 0.45862826704978943,
      "learning_rate": 4.587155963302753e-07,
      "loss": 0.3563,
      "step": 60
    },
    {
      "epoch": 0.002799320820522234,
      "grad_norm": 1.9747591018676758,
      "learning_rate": 4.663608562691132e-07,
      "loss": 0.6012,
      "step": 61
    },
    {
      "epoch": 0.0028452113257766967,
      "grad_norm": 1.7752922773361206,
      "learning_rate": 4.740061162079511e-07,
      "loss": 0.6039,
      "step": 62
    },
    {
      "epoch": 0.0028911018310311598,
      "grad_norm": 0.45932909846305847,
      "learning_rate": 4.816513761467891e-07,
      "loss": 0.339,
      "step": 63
    },
    {
      "epoch": 0.0029369923362856224,
      "grad_norm": 1.701511263847351,
      "learning_rate": 4.89296636085627e-07,
      "loss": 0.6563,
      "step": 64
    },
    {
      "epoch": 0.0029828828415400855,
      "grad_norm": 1.5318350791931152,
      "learning_rate": 4.969418960244648e-07,
      "loss": 0.5504,
      "step": 65
    },
    {
      "epoch": 0.003028773346794548,
      "grad_norm": 2.573740243911743,
      "learning_rate": 5.045871559633028e-07,
      "loss": 0.7237,
      "step": 66
    },
    {
      "epoch": 0.0030746638520490113,
      "grad_norm": 1.351897120475769,
      "learning_rate": 5.122324159021407e-07,
      "loss": 0.5439,
      "step": 67
    },
    {
      "epoch": 0.003120554357303474,
      "grad_norm": 1.41916024684906,
      "learning_rate": 5.198776758409786e-07,
      "loss": 0.5839,
      "step": 68
    },
    {
      "epoch": 0.0031664448625579366,
      "grad_norm": 0.7206918597221375,
      "learning_rate": 5.275229357798165e-07,
      "loss": 0.4373,
      "step": 69
    },
    {
      "epoch": 0.0032123353678123997,
      "grad_norm": 1.529764175415039,
      "learning_rate": 5.351681957186545e-07,
      "loss": 0.5101,
      "step": 70
    },
    {
      "epoch": 0.0032582258730668623,
      "grad_norm": 0.5984582304954529,
      "learning_rate": 5.428134556574924e-07,
      "loss": 0.3715,
      "step": 71
    },
    {
      "epoch": 0.0033041163783213254,
      "grad_norm": 0.830768346786499,
      "learning_rate": 5.504587155963304e-07,
      "loss": 0.4262,
      "step": 72
    },
    {
      "epoch": 0.003350006883575788,
      "grad_norm": 0.5834783315658569,
      "learning_rate": 5.581039755351682e-07,
      "loss": 0.4524,
      "step": 73
    },
    {
      "epoch": 0.003395897388830251,
      "grad_norm": 0.6707944273948669,
      "learning_rate": 5.657492354740061e-07,
      "loss": 0.3892,
      "step": 74
    },
    {
      "epoch": 0.003441787894084714,
      "grad_norm": 0.5306781530380249,
      "learning_rate": 5.733944954128441e-07,
      "loss": 0.3801,
      "step": 75
    },
    {
      "epoch": 0.003487678399339177,
      "grad_norm": 0.5925319194793701,
      "learning_rate": 5.81039755351682e-07,
      "loss": 0.4101,
      "step": 76
    },
    {
      "epoch": 0.0035335689045936395,
      "grad_norm": 0.5854651927947998,
      "learning_rate": 5.8868501529052e-07,
      "loss": 0.4071,
      "step": 77
    },
    {
      "epoch": 0.0035794594098481026,
      "grad_norm": 1.4466856718063354,
      "learning_rate": 5.963302752293579e-07,
      "loss": 0.5946,
      "step": 78
    },
    {
      "epoch": 0.0036253499151025653,
      "grad_norm": 2.0078835487365723,
      "learning_rate": 6.039755351681958e-07,
      "loss": 0.6802,
      "step": 79
    },
    {
      "epoch": 0.003671240420357028,
      "grad_norm": 0.6159325838088989,
      "learning_rate": 6.116207951070337e-07,
      "loss": 0.3883,
      "step": 80
    },
    {
      "epoch": 0.003717130925611491,
      "grad_norm": 0.44337207078933716,
      "learning_rate": 6.192660550458716e-07,
      "loss": 0.2945,
      "step": 81
    },
    {
      "epoch": 0.0037630214308659537,
      "grad_norm": 0.550020158290863,
      "learning_rate": 6.269113149847095e-07,
      "loss": 0.3542,
      "step": 82
    },
    {
      "epoch": 0.0038089119361204168,
      "grad_norm": 1.473622441291809,
      "learning_rate": 6.345565749235474e-07,
      "loss": 0.6035,
      "step": 83
    },
    {
      "epoch": 0.0038548024413748794,
      "grad_norm": 1.674747347831726,
      "learning_rate": 6.422018348623854e-07,
      "loss": 0.535,
      "step": 84
    },
    {
      "epoch": 0.0039006929466293425,
      "grad_norm": 1.0093748569488525,
      "learning_rate": 6.498470948012232e-07,
      "loss": 0.5078,
      "step": 85
    },
    {
      "epoch": 0.003946583451883806,
      "grad_norm": 0.6665521264076233,
      "learning_rate": 6.574923547400612e-07,
      "loss": 0.3994,
      "step": 86
    },
    {
      "epoch": 0.003992473957138268,
      "grad_norm": 1.0105335712432861,
      "learning_rate": 6.651376146788992e-07,
      "loss": 0.4706,
      "step": 87
    },
    {
      "epoch": 0.004038364462392731,
      "grad_norm": 0.9650800824165344,
      "learning_rate": 6.72782874617737e-07,
      "loss": 0.4861,
      "step": 88
    },
    {
      "epoch": 0.004084254967647194,
      "grad_norm": 0.7468169331550598,
      "learning_rate": 6.804281345565749e-07,
      "loss": 0.4496,
      "step": 89
    },
    {
      "epoch": 0.004130145472901657,
      "grad_norm": 0.9022457003593445,
      "learning_rate": 6.880733944954129e-07,
      "loss": 0.5108,
      "step": 90
    },
    {
      "epoch": 0.004176035978156119,
      "grad_norm": 0.5702598690986633,
      "learning_rate": 6.957186544342508e-07,
      "loss": 0.438,
      "step": 91
    },
    {
      "epoch": 0.004221926483410582,
      "grad_norm": 1.3481727838516235,
      "learning_rate": 7.033639143730888e-07,
      "loss": 0.5674,
      "step": 92
    },
    {
      "epoch": 0.0042678169886650455,
      "grad_norm": 0.4537692666053772,
      "learning_rate": 7.110091743119267e-07,
      "loss": 0.3574,
      "step": 93
    },
    {
      "epoch": 0.004313707493919508,
      "grad_norm": 1.1757780313491821,
      "learning_rate": 7.186544342507645e-07,
      "loss": 0.5351,
      "step": 94
    },
    {
      "epoch": 0.004359597999173971,
      "grad_norm": 1.3105450868606567,
      "learning_rate": 7.262996941896025e-07,
      "loss": 0.5779,
      "step": 95
    },
    {
      "epoch": 0.004405488504428434,
      "grad_norm": 0.6526632308959961,
      "learning_rate": 7.339449541284405e-07,
      "loss": 0.4459,
      "step": 96
    },
    {
      "epoch": 0.004451379009682897,
      "grad_norm": 0.807191789150238,
      "learning_rate": 7.415902140672783e-07,
      "loss": 0.5109,
      "step": 97
    },
    {
      "epoch": 0.004497269514937359,
      "grad_norm": 0.8935018181800842,
      "learning_rate": 7.492354740061163e-07,
      "loss": 0.4762,
      "step": 98
    },
    {
      "epoch": 0.004543160020191822,
      "grad_norm": 0.8139580488204956,
      "learning_rate": 7.568807339449542e-07,
      "loss": 0.4826,
      "step": 99
    },
    {
      "epoch": 0.004589050525446285,
      "grad_norm": 0.6782995462417603,
      "learning_rate": 7.645259938837921e-07,
      "loss": 0.4629,
      "step": 100
    },
    {
      "epoch": 0.004634941030700748,
      "grad_norm": 0.8879847526550293,
      "learning_rate": 7.7217125382263e-07,
      "loss": 0.5781,
      "step": 101
    },
    {
      "epoch": 0.004680831535955211,
      "grad_norm": 1.026992678642273,
      "learning_rate": 7.79816513761468e-07,
      "loss": 0.6328,
      "step": 102
    },
    {
      "epoch": 0.004726722041209674,
      "grad_norm": 0.5849438905715942,
      "learning_rate": 7.874617737003058e-07,
      "loss": 0.3326,
      "step": 103
    },
    {
      "epoch": 0.004772612546464137,
      "grad_norm": 0.5503666400909424,
      "learning_rate": 7.951070336391438e-07,
      "loss": 0.4173,
      "step": 104
    },
    {
      "epoch": 0.004818503051718599,
      "grad_norm": 0.5570368766784668,
      "learning_rate": 8.027522935779817e-07,
      "loss": 0.4654,
      "step": 105
    },
    {
      "epoch": 0.004864393556973062,
      "grad_norm": 0.661638081073761,
      "learning_rate": 8.103975535168196e-07,
      "loss": 0.5453,
      "step": 106
    },
    {
      "epoch": 0.004910284062227525,
      "grad_norm": 0.6135708093643188,
      "learning_rate": 8.180428134556576e-07,
      "loss": 0.5167,
      "step": 107
    },
    {
      "epoch": 0.004956174567481988,
      "grad_norm": 0.4996219873428345,
      "learning_rate": 8.256880733944956e-07,
      "loss": 0.3413,
      "step": 108
    },
    {
      "epoch": 0.0050020650727364505,
      "grad_norm": 0.6201441287994385,
      "learning_rate": 8.333333333333333e-07,
      "loss": 0.506,
      "step": 109
    },
    {
      "epoch": 0.005047955577990914,
      "grad_norm": 0.6837106943130493,
      "learning_rate": 8.409785932721713e-07,
      "loss": 0.5386,
      "step": 110
    },
    {
      "epoch": 0.005093846083245377,
      "grad_norm": 0.6054534316062927,
      "learning_rate": 8.486238532110093e-07,
      "loss": 0.4014,
      "step": 111
    },
    {
      "epoch": 0.00513973658849984,
      "grad_norm": 0.874347448348999,
      "learning_rate": 8.562691131498472e-07,
      "loss": 0.6087,
      "step": 112
    },
    {
      "epoch": 0.005185627093754302,
      "grad_norm": 0.5703540444374084,
      "learning_rate": 8.639143730886851e-07,
      "loss": 0.4674,
      "step": 113
    },
    {
      "epoch": 0.005231517599008765,
      "grad_norm": 0.5193082094192505,
      "learning_rate": 8.71559633027523e-07,
      "loss": 0.3745,
      "step": 114
    },
    {
      "epoch": 0.005277408104263228,
      "grad_norm": 0.5911059379577637,
      "learning_rate": 8.792048929663609e-07,
      "loss": 0.4971,
      "step": 115
    },
    {
      "epoch": 0.00532329860951769,
      "grad_norm": 0.6464957594871521,
      "learning_rate": 8.868501529051989e-07,
      "loss": 0.5544,
      "step": 116
    },
    {
      "epoch": 0.0053691891147721535,
      "grad_norm": 0.5820969343185425,
      "learning_rate": 8.944954128440368e-07,
      "loss": 0.4245,
      "step": 117
    },
    {
      "epoch": 0.005415079620026617,
      "grad_norm": 0.6394116878509521,
      "learning_rate": 9.021406727828746e-07,
      "loss": 0.3375,
      "step": 118
    },
    {
      "epoch": 0.00546097012528108,
      "grad_norm": 0.568695604801178,
      "learning_rate": 9.097859327217126e-07,
      "loss": 0.4163,
      "step": 119
    },
    {
      "epoch": 0.005506860630535542,
      "grad_norm": 0.7871958613395691,
      "learning_rate": 9.174311926605506e-07,
      "loss": 0.52,
      "step": 120
    },
    {
      "epoch": 0.005552751135790005,
      "grad_norm": 0.5058826804161072,
      "learning_rate": 9.250764525993884e-07,
      "loss": 0.4539,
      "step": 121
    },
    {
      "epoch": 0.005598641641044468,
      "grad_norm": 0.6198352575302124,
      "learning_rate": 9.327217125382264e-07,
      "loss": 0.5701,
      "step": 122
    },
    {
      "epoch": 0.005644532146298931,
      "grad_norm": 0.6333051919937134,
      "learning_rate": 9.403669724770644e-07,
      "loss": 0.5923,
      "step": 123
    },
    {
      "epoch": 0.005690422651553393,
      "grad_norm": 0.49544772505760193,
      "learning_rate": 9.480122324159022e-07,
      "loss": 0.3582,
      "step": 124
    },
    {
      "epoch": 0.0057363131568078565,
      "grad_norm": 0.6826655268669128,
      "learning_rate": 9.556574923547401e-07,
      "loss": 0.4117,
      "step": 125
    },
    {
      "epoch": 0.0057822036620623195,
      "grad_norm": 0.49462851881980896,
      "learning_rate": 9.633027522935782e-07,
      "loss": 0.4171,
      "step": 126
    },
    {
      "epoch": 0.005828094167316782,
      "grad_norm": 0.5096755027770996,
      "learning_rate": 9.709480122324159e-07,
      "loss": 0.4312,
      "step": 127
    },
    {
      "epoch": 0.005873984672571245,
      "grad_norm": 0.5185785293579102,
      "learning_rate": 9.78593272171254e-07,
      "loss": 0.4836,
      "step": 128
    },
    {
      "epoch": 0.005919875177825708,
      "grad_norm": 0.4647371768951416,
      "learning_rate": 9.862385321100918e-07,
      "loss": 0.3823,
      "step": 129
    },
    {
      "epoch": 0.005965765683080171,
      "grad_norm": 0.4872678518295288,
      "learning_rate": 9.938837920489297e-07,
      "loss": 0.4208,
      "step": 130
    },
    {
      "epoch": 0.006011656188334633,
      "grad_norm": 0.5332506895065308,
      "learning_rate": 1.0015290519877676e-06,
      "loss": 0.5208,
      "step": 131
    },
    {
      "epoch": 0.006057546693589096,
      "grad_norm": 0.4972846806049347,
      "learning_rate": 1.0091743119266057e-06,
      "loss": 0.4175,
      "step": 132
    },
    {
      "epoch": 0.006103437198843559,
      "grad_norm": 0.47214367985725403,
      "learning_rate": 1.0168195718654435e-06,
      "loss": 0.4308,
      "step": 133
    },
    {
      "epoch": 0.0061493277040980225,
      "grad_norm": 0.6262720823287964,
      "learning_rate": 1.0244648318042814e-06,
      "loss": 0.5003,
      "step": 134
    },
    {
      "epoch": 0.006195218209352485,
      "grad_norm": 0.4969519376754761,
      "learning_rate": 1.0321100917431195e-06,
      "loss": 0.4497,
      "step": 135
    },
    {
      "epoch": 0.006241108714606948,
      "grad_norm": 0.5266798734664917,
      "learning_rate": 1.0397553516819571e-06,
      "loss": 0.4646,
      "step": 136
    },
    {
      "epoch": 0.006286999219861411,
      "grad_norm": 0.5022170543670654,
      "learning_rate": 1.0474006116207952e-06,
      "loss": 0.5034,
      "step": 137
    },
    {
      "epoch": 0.006332889725115873,
      "grad_norm": 0.6192761063575745,
      "learning_rate": 1.055045871559633e-06,
      "loss": 0.4547,
      "step": 138
    },
    {
      "epoch": 0.006378780230370336,
      "grad_norm": 0.456028550863266,
      "learning_rate": 1.062691131498471e-06,
      "loss": 0.3919,
      "step": 139
    },
    {
      "epoch": 0.006424670735624799,
      "grad_norm": 0.5489574670791626,
      "learning_rate": 1.070336391437309e-06,
      "loss": 0.4536,
      "step": 140
    },
    {
      "epoch": 0.006470561240879262,
      "grad_norm": 0.6313809156417847,
      "learning_rate": 1.077981651376147e-06,
      "loss": 0.5674,
      "step": 141
    },
    {
      "epoch": 0.006516451746133725,
      "grad_norm": 0.5987564921379089,
      "learning_rate": 1.0856269113149848e-06,
      "loss": 0.6084,
      "step": 142
    },
    {
      "epoch": 0.006562342251388188,
      "grad_norm": 0.4713725745677948,
      "learning_rate": 1.0932721712538227e-06,
      "loss": 0.3861,
      "step": 143
    },
    {
      "epoch": 0.006608232756642651,
      "grad_norm": 0.5778293609619141,
      "learning_rate": 1.1009174311926608e-06,
      "loss": 0.4871,
      "step": 144
    },
    {
      "epoch": 0.006654123261897114,
      "grad_norm": 0.4850349426269531,
      "learning_rate": 1.1085626911314986e-06,
      "loss": 0.3813,
      "step": 145
    },
    {
      "epoch": 0.006700013767151576,
      "grad_norm": 0.4866754710674286,
      "learning_rate": 1.1162079510703365e-06,
      "loss": 0.4618,
      "step": 146
    },
    {
      "epoch": 0.006745904272406039,
      "grad_norm": 0.6613770723342896,
      "learning_rate": 1.1238532110091744e-06,
      "loss": 0.5501,
      "step": 147
    },
    {
      "epoch": 0.006791794777660502,
      "grad_norm": 0.4495118260383606,
      "learning_rate": 1.1314984709480122e-06,
      "loss": 0.3879,
      "step": 148
    },
    {
      "epoch": 0.0068376852829149645,
      "grad_norm": 0.6176930069923401,
      "learning_rate": 1.1391437308868503e-06,
      "loss": 0.4349,
      "step": 149
    },
    {
      "epoch": 0.006883575788169428,
      "grad_norm": 0.5159872174263,
      "learning_rate": 1.1467889908256882e-06,
      "loss": 0.4828,
      "step": 150
    },
    {
      "epoch": 0.006929466293423891,
      "grad_norm": 0.4736764430999756,
      "learning_rate": 1.154434250764526e-06,
      "loss": 0.3564,
      "step": 151
    },
    {
      "epoch": 0.006975356798678354,
      "grad_norm": 0.6415404081344604,
      "learning_rate": 1.162079510703364e-06,
      "loss": 0.6007,
      "step": 152
    },
    {
      "epoch": 0.007021247303932816,
      "grad_norm": 0.5175949335098267,
      "learning_rate": 1.169724770642202e-06,
      "loss": 0.5334,
      "step": 153
    },
    {
      "epoch": 0.007067137809187279,
      "grad_norm": 0.46300917863845825,
      "learning_rate": 1.17737003058104e-06,
      "loss": 0.413,
      "step": 154
    },
    {
      "epoch": 0.007113028314441742,
      "grad_norm": 0.5009661912918091,
      "learning_rate": 1.1850152905198778e-06,
      "loss": 0.5139,
      "step": 155
    },
    {
      "epoch": 0.007158918819696205,
      "grad_norm": 0.5458509922027588,
      "learning_rate": 1.1926605504587159e-06,
      "loss": 0.4877,
      "step": 156
    },
    {
      "epoch": 0.0072048093249506675,
      "grad_norm": 0.5963668823242188,
      "learning_rate": 1.2003058103975535e-06,
      "loss": 0.5502,
      "step": 157
    },
    {
      "epoch": 0.0072506998302051305,
      "grad_norm": 0.4685239791870117,
      "learning_rate": 1.2079510703363916e-06,
      "loss": 0.396,
      "step": 158
    },
    {
      "epoch": 0.007296590335459594,
      "grad_norm": 0.47896337509155273,
      "learning_rate": 1.2155963302752295e-06,
      "loss": 0.481,
      "step": 159
    },
    {
      "epoch": 0.007342480840714056,
      "grad_norm": 0.5194567441940308,
      "learning_rate": 1.2232415902140673e-06,
      "loss": 0.5662,
      "step": 160
    },
    {
      "epoch": 0.007388371345968519,
      "grad_norm": 0.48354044556617737,
      "learning_rate": 1.2308868501529054e-06,
      "loss": 0.402,
      "step": 161
    },
    {
      "epoch": 0.007434261851222982,
      "grad_norm": 0.4972841441631317,
      "learning_rate": 1.2385321100917433e-06,
      "loss": 0.4685,
      "step": 162
    },
    {
      "epoch": 0.007480152356477445,
      "grad_norm": 0.4714357256889343,
      "learning_rate": 1.2461773700305812e-06,
      "loss": 0.4032,
      "step": 163
    },
    {
      "epoch": 0.007526042861731907,
      "grad_norm": 0.5029146075248718,
      "learning_rate": 1.253822629969419e-06,
      "loss": 0.4139,
      "step": 164
    },
    {
      "epoch": 0.00757193336698637,
      "grad_norm": 0.4872240126132965,
      "learning_rate": 1.261467889908257e-06,
      "loss": 0.4753,
      "step": 165
    },
    {
      "epoch": 0.0076178238722408335,
      "grad_norm": 0.464991956949234,
      "learning_rate": 1.2691131498470948e-06,
      "loss": 0.3451,
      "step": 166
    },
    {
      "epoch": 0.007663714377495297,
      "grad_norm": 0.535402774810791,
      "learning_rate": 1.2767584097859329e-06,
      "loss": 0.5317,
      "step": 167
    },
    {
      "epoch": 0.007709604882749759,
      "grad_norm": 0.5238930583000183,
      "learning_rate": 1.2844036697247707e-06,
      "loss": 0.5113,
      "step": 168
    },
    {
      "epoch": 0.007755495388004222,
      "grad_norm": 0.4925292730331421,
      "learning_rate": 1.2920489296636088e-06,
      "loss": 0.4153,
      "step": 169
    },
    {
      "epoch": 0.007801385893258685,
      "grad_norm": 0.5098105669021606,
      "learning_rate": 1.2996941896024465e-06,
      "loss": 0.4934,
      "step": 170
    },
    {
      "epoch": 0.007847276398513148,
      "grad_norm": 0.5790193676948547,
      "learning_rate": 1.3073394495412844e-06,
      "loss": 0.3271,
      "step": 171
    },
    {
      "epoch": 0.007893166903767611,
      "grad_norm": 0.5277330875396729,
      "learning_rate": 1.3149847094801224e-06,
      "loss": 0.4385,
      "step": 172
    },
    {
      "epoch": 0.007939057409022073,
      "grad_norm": 0.45835790038108826,
      "learning_rate": 1.3226299694189603e-06,
      "loss": 0.4397,
      "step": 173
    },
    {
      "epoch": 0.007984947914276536,
      "grad_norm": 0.46754586696624756,
      "learning_rate": 1.3302752293577984e-06,
      "loss": 0.417,
      "step": 174
    },
    {
      "epoch": 0.008030838419530999,
      "grad_norm": 0.6357254981994629,
      "learning_rate": 1.3379204892966363e-06,
      "loss": 0.7048,
      "step": 175
    },
    {
      "epoch": 0.008076728924785462,
      "grad_norm": 0.4879402220249176,
      "learning_rate": 1.345565749235474e-06,
      "loss": 0.3694,
      "step": 176
    },
    {
      "epoch": 0.008122619430039925,
      "grad_norm": 0.5214935541152954,
      "learning_rate": 1.353211009174312e-06,
      "loss": 0.4588,
      "step": 177
    },
    {
      "epoch": 0.008168509935294388,
      "grad_norm": 0.5096695423126221,
      "learning_rate": 1.3608562691131499e-06,
      "loss": 0.4498,
      "step": 178
    },
    {
      "epoch": 0.008214400440548851,
      "grad_norm": 0.44374576210975647,
      "learning_rate": 1.368501529051988e-06,
      "loss": 0.3125,
      "step": 179
    },
    {
      "epoch": 0.008260290945803314,
      "grad_norm": 0.49549832940101624,
      "learning_rate": 1.3761467889908258e-06,
      "loss": 0.4494,
      "step": 180
    },
    {
      "epoch": 0.008306181451057775,
      "grad_norm": 0.4328993558883667,
      "learning_rate": 1.383792048929664e-06,
      "loss": 0.3694,
      "step": 181
    },
    {
      "epoch": 0.008352071956312239,
      "grad_norm": 0.4350336790084839,
      "learning_rate": 1.3914373088685016e-06,
      "loss": 0.3625,
      "step": 182
    },
    {
      "epoch": 0.008397962461566702,
      "grad_norm": 0.46577188372612,
      "learning_rate": 1.3990825688073395e-06,
      "loss": 0.4098,
      "step": 183
    },
    {
      "epoch": 0.008443852966821165,
      "grad_norm": 0.553417444229126,
      "learning_rate": 1.4067278287461775e-06,
      "loss": 0.5746,
      "step": 184
    },
    {
      "epoch": 0.008489743472075628,
      "grad_norm": 0.4803412854671478,
      "learning_rate": 1.4143730886850154e-06,
      "loss": 0.4149,
      "step": 185
    },
    {
      "epoch": 0.008535633977330091,
      "grad_norm": 0.42834556102752686,
      "learning_rate": 1.4220183486238535e-06,
      "loss": 0.3503,
      "step": 186
    },
    {
      "epoch": 0.008581524482584554,
      "grad_norm": 0.500589907169342,
      "learning_rate": 1.4296636085626914e-06,
      "loss": 0.4777,
      "step": 187
    },
    {
      "epoch": 0.008627414987839015,
      "grad_norm": 0.4265582859516144,
      "learning_rate": 1.437308868501529e-06,
      "loss": 0.3326,
      "step": 188
    },
    {
      "epoch": 0.008673305493093478,
      "grad_norm": 0.6069602370262146,
      "learning_rate": 1.4449541284403671e-06,
      "loss": 0.6107,
      "step": 189
    },
    {
      "epoch": 0.008719195998347942,
      "grad_norm": 0.4830178916454315,
      "learning_rate": 1.452599388379205e-06,
      "loss": 0.4705,
      "step": 190
    },
    {
      "epoch": 0.008765086503602405,
      "grad_norm": 0.5238247513771057,
      "learning_rate": 1.460244648318043e-06,
      "loss": 0.5562,
      "step": 191
    },
    {
      "epoch": 0.008810977008856868,
      "grad_norm": 0.4872019588947296,
      "learning_rate": 1.467889908256881e-06,
      "loss": 0.4079,
      "step": 192
    },
    {
      "epoch": 0.00885686751411133,
      "grad_norm": 0.5381096005439758,
      "learning_rate": 1.4755351681957188e-06,
      "loss": 0.4264,
      "step": 193
    },
    {
      "epoch": 0.008902758019365794,
      "grad_norm": 0.5009563565254211,
      "learning_rate": 1.4831804281345567e-06,
      "loss": 0.4761,
      "step": 194
    },
    {
      "epoch": 0.008948648524620255,
      "grad_norm": 0.5903949737548828,
      "learning_rate": 1.4908256880733945e-06,
      "loss": 0.6371,
      "step": 195
    },
    {
      "epoch": 0.008994539029874718,
      "grad_norm": 0.5457831621170044,
      "learning_rate": 1.4984709480122326e-06,
      "loss": 0.4981,
      "step": 196
    },
    {
      "epoch": 0.009040429535129181,
      "grad_norm": 0.5037556290626526,
      "learning_rate": 1.5061162079510705e-06,
      "loss": 0.429,
      "step": 197
    },
    {
      "epoch": 0.009086320040383645,
      "grad_norm": 0.4773487150669098,
      "learning_rate": 1.5137614678899084e-06,
      "loss": 0.4345,
      "step": 198
    },
    {
      "epoch": 0.009132210545638108,
      "grad_norm": 0.5097615718841553,
      "learning_rate": 1.5214067278287465e-06,
      "loss": 0.5077,
      "step": 199
    },
    {
      "epoch": 0.00917810105089257,
      "grad_norm": 0.4838576912879944,
      "learning_rate": 1.5290519877675841e-06,
      "loss": 0.4384,
      "step": 200
    },
    {
      "epoch": 0.009223991556147034,
      "grad_norm": 0.4566458761692047,
      "learning_rate": 1.536697247706422e-06,
      "loss": 0.3858,
      "step": 201
    },
    {
      "epoch": 0.009269882061401497,
      "grad_norm": 0.6419386863708496,
      "learning_rate": 1.54434250764526e-06,
      "loss": 0.5308,
      "step": 202
    },
    {
      "epoch": 0.009315772566655958,
      "grad_norm": 0.5053452849388123,
      "learning_rate": 1.551987767584098e-06,
      "loss": 0.5171,
      "step": 203
    },
    {
      "epoch": 0.009361663071910421,
      "grad_norm": 0.4292317032814026,
      "learning_rate": 1.559633027522936e-06,
      "loss": 0.3162,
      "step": 204
    },
    {
      "epoch": 0.009407553577164884,
      "grad_norm": 0.5141228437423706,
      "learning_rate": 1.567278287461774e-06,
      "loss": 0.4805,
      "step": 205
    },
    {
      "epoch": 0.009453444082419347,
      "grad_norm": 0.5513278841972351,
      "learning_rate": 1.5749235474006116e-06,
      "loss": 0.5149,
      "step": 206
    },
    {
      "epoch": 0.00949933458767381,
      "grad_norm": 0.46986624598503113,
      "learning_rate": 1.5825688073394496e-06,
      "loss": 0.4009,
      "step": 207
    },
    {
      "epoch": 0.009545225092928274,
      "grad_norm": 0.46508821845054626,
      "learning_rate": 1.5902140672782875e-06,
      "loss": 0.4486,
      "step": 208
    },
    {
      "epoch": 0.009591115598182737,
      "grad_norm": 0.5256361961364746,
      "learning_rate": 1.5978593272171256e-06,
      "loss": 0.4443,
      "step": 209
    },
    {
      "epoch": 0.009637006103437198,
      "grad_norm": 0.4129773676395416,
      "learning_rate": 1.6055045871559635e-06,
      "loss": 0.2691,
      "step": 210
    },
    {
      "epoch": 0.009682896608691661,
      "grad_norm": 0.5132619738578796,
      "learning_rate": 1.6131498470948016e-06,
      "loss": 0.4946,
      "step": 211
    },
    {
      "epoch": 0.009728787113946124,
      "grad_norm": 0.6025416254997253,
      "learning_rate": 1.6207951070336392e-06,
      "loss": 0.4538,
      "step": 212
    },
    {
      "epoch": 0.009774677619200587,
      "grad_norm": 0.5104288458824158,
      "learning_rate": 1.628440366972477e-06,
      "loss": 0.512,
      "step": 213
    },
    {
      "epoch": 0.00982056812445505,
      "grad_norm": 0.4239733815193176,
      "learning_rate": 1.6360856269113152e-06,
      "loss": 0.3331,
      "step": 214
    },
    {
      "epoch": 0.009866458629709514,
      "grad_norm": 0.5097137093544006,
      "learning_rate": 1.643730886850153e-06,
      "loss": 0.518,
      "step": 215
    },
    {
      "epoch": 0.009912349134963977,
      "grad_norm": 0.5095618963241577,
      "learning_rate": 1.6513761467889911e-06,
      "loss": 0.5671,
      "step": 216
    },
    {
      "epoch": 0.009958239640218438,
      "grad_norm": 0.5642290115356445,
      "learning_rate": 1.659021406727829e-06,
      "loss": 0.5115,
      "step": 217
    },
    {
      "epoch": 0.010004130145472901,
      "grad_norm": 0.4698328375816345,
      "learning_rate": 1.6666666666666667e-06,
      "loss": 0.4595,
      "step": 218
    },
    {
      "epoch": 0.010050020650727364,
      "grad_norm": 0.42536523938179016,
      "learning_rate": 1.6743119266055047e-06,
      "loss": 0.3607,
      "step": 219
    },
    {
      "epoch": 0.010095911155981827,
      "grad_norm": 0.4711594879627228,
      "learning_rate": 1.6819571865443426e-06,
      "loss": 0.4012,
      "step": 220
    },
    {
      "epoch": 0.01014180166123629,
      "grad_norm": 0.4453487694263458,
      "learning_rate": 1.6896024464831807e-06,
      "loss": 0.3845,
      "step": 221
    },
    {
      "epoch": 0.010187692166490753,
      "grad_norm": 0.4480073153972626,
      "learning_rate": 1.6972477064220186e-06,
      "loss": 0.4128,
      "step": 222
    },
    {
      "epoch": 0.010233582671745216,
      "grad_norm": 0.4990222454071045,
      "learning_rate": 1.7048929663608562e-06,
      "loss": 0.4794,
      "step": 223
    },
    {
      "epoch": 0.01027947317699968,
      "grad_norm": 0.5442683696746826,
      "learning_rate": 1.7125382262996943e-06,
      "loss": 0.537,
      "step": 224
    },
    {
      "epoch": 0.010325363682254141,
      "grad_norm": 0.4410986304283142,
      "learning_rate": 1.7201834862385322e-06,
      "loss": 0.4166,
      "step": 225
    },
    {
      "epoch": 0.010371254187508604,
      "grad_norm": 0.43682900071144104,
      "learning_rate": 1.7278287461773703e-06,
      "loss": 0.3537,
      "step": 226
    },
    {
      "epoch": 0.010417144692763067,
      "grad_norm": 0.4223477840423584,
      "learning_rate": 1.7354740061162081e-06,
      "loss": 0.3292,
      "step": 227
    },
    {
      "epoch": 0.01046303519801753,
      "grad_norm": 0.4678100347518921,
      "learning_rate": 1.743119266055046e-06,
      "loss": 0.467,
      "step": 228
    },
    {
      "epoch": 0.010508925703271993,
      "grad_norm": 0.49103376269340515,
      "learning_rate": 1.7507645259938839e-06,
      "loss": 0.4286,
      "step": 229
    },
    {
      "epoch": 0.010554816208526456,
      "grad_norm": 0.5196321606636047,
      "learning_rate": 1.7584097859327218e-06,
      "loss": 0.4914,
      "step": 230
    },
    {
      "epoch": 0.01060070671378092,
      "grad_norm": 0.46310898661613464,
      "learning_rate": 1.7660550458715596e-06,
      "loss": 0.3717,
      "step": 231
    },
    {
      "epoch": 0.01064659721903538,
      "grad_norm": 0.4803982079029083,
      "learning_rate": 1.7737003058103977e-06,
      "loss": 0.5182,
      "step": 232
    },
    {
      "epoch": 0.010692487724289844,
      "grad_norm": 0.49229687452316284,
      "learning_rate": 1.7813455657492356e-06,
      "loss": 0.4578,
      "step": 233
    },
    {
      "epoch": 0.010738378229544307,
      "grad_norm": 0.5128414034843445,
      "learning_rate": 1.7889908256880737e-06,
      "loss": 0.473,
      "step": 234
    },
    {
      "epoch": 0.01078426873479877,
      "grad_norm": 0.5010553598403931,
      "learning_rate": 1.7966360856269113e-06,
      "loss": 0.475,
      "step": 235
    },
    {
      "epoch": 0.010830159240053233,
      "grad_norm": 0.4786660075187683,
      "learning_rate": 1.8042813455657492e-06,
      "loss": 0.4771,
      "step": 236
    },
    {
      "epoch": 0.010876049745307696,
      "grad_norm": 0.4521401822566986,
      "learning_rate": 1.8119266055045873e-06,
      "loss": 0.3874,
      "step": 237
    },
    {
      "epoch": 0.01092194025056216,
      "grad_norm": 0.4972446858882904,
      "learning_rate": 1.8195718654434252e-06,
      "loss": 0.5038,
      "step": 238
    },
    {
      "epoch": 0.01096783075581662,
      "grad_norm": 0.4640282094478607,
      "learning_rate": 1.8272171253822632e-06,
      "loss": 0.4197,
      "step": 239
    },
    {
      "epoch": 0.011013721261071084,
      "grad_norm": 0.5040390491485596,
      "learning_rate": 1.8348623853211011e-06,
      "loss": 0.4846,
      "step": 240
    },
    {
      "epoch": 0.011059611766325547,
      "grad_norm": 0.43553870916366577,
      "learning_rate": 1.8425076452599388e-06,
      "loss": 0.3355,
      "step": 241
    },
    {
      "epoch": 0.01110550227158001,
      "grad_norm": 0.47265294194221497,
      "learning_rate": 1.8501529051987769e-06,
      "loss": 0.4562,
      "step": 242
    },
    {
      "epoch": 0.011151392776834473,
      "grad_norm": 0.4227467179298401,
      "learning_rate": 1.8577981651376147e-06,
      "loss": 0.3322,
      "step": 243
    },
    {
      "epoch": 0.011197283282088936,
      "grad_norm": 0.5771182179450989,
      "learning_rate": 1.8654434250764528e-06,
      "loss": 0.5562,
      "step": 244
    },
    {
      "epoch": 0.0112431737873434,
      "grad_norm": 0.42815539240837097,
      "learning_rate": 1.8730886850152907e-06,
      "loss": 0.3546,
      "step": 245
    },
    {
      "epoch": 0.011289064292597862,
      "grad_norm": 0.5375799536705017,
      "learning_rate": 1.8807339449541288e-06,
      "loss": 0.4165,
      "step": 246
    },
    {
      "epoch": 0.011334954797852324,
      "grad_norm": 0.4885975122451782,
      "learning_rate": 1.8883792048929664e-06,
      "loss": 0.4749,
      "step": 247
    },
    {
      "epoch": 0.011380845303106787,
      "grad_norm": 0.43277713656425476,
      "learning_rate": 1.8960244648318043e-06,
      "loss": 0.3485,
      "step": 248
    },
    {
      "epoch": 0.01142673580836125,
      "grad_norm": 0.42765310406684875,
      "learning_rate": 1.9036697247706424e-06,
      "loss": 0.3523,
      "step": 249
    },
    {
      "epoch": 0.011472626313615713,
      "grad_norm": 0.4612452983856201,
      "learning_rate": 1.9113149847094803e-06,
      "loss": 0.4482,
      "step": 250
    },
    {
      "epoch": 0.011518516818870176,
      "grad_norm": 0.43229803442955017,
      "learning_rate": 1.918960244648318e-06,
      "loss": 0.3545,
      "step": 251
    },
    {
      "epoch": 0.011564407324124639,
      "grad_norm": 0.46931692957878113,
      "learning_rate": 1.9266055045871564e-06,
      "loss": 0.4344,
      "step": 252
    },
    {
      "epoch": 0.011610297829379102,
      "grad_norm": 0.428820937871933,
      "learning_rate": 1.934250764525994e-06,
      "loss": 0.3404,
      "step": 253
    },
    {
      "epoch": 0.011656188334633564,
      "grad_norm": 0.5306540727615356,
      "learning_rate": 1.9418960244648317e-06,
      "loss": 0.5357,
      "step": 254
    },
    {
      "epoch": 0.011702078839888027,
      "grad_norm": 0.4210246801376343,
      "learning_rate": 1.94954128440367e-06,
      "loss": 0.304,
      "step": 255
    },
    {
      "epoch": 0.01174796934514249,
      "grad_norm": 0.5046725869178772,
      "learning_rate": 1.957186544342508e-06,
      "loss": 0.4961,
      "step": 256
    },
    {
      "epoch": 0.011793859850396953,
      "grad_norm": 0.45517218112945557,
      "learning_rate": 1.9648318042813458e-06,
      "loss": 0.4069,
      "step": 257
    },
    {
      "epoch": 0.011839750355651416,
      "grad_norm": 0.4402482807636261,
      "learning_rate": 1.9724770642201837e-06,
      "loss": 0.419,
      "step": 258
    },
    {
      "epoch": 0.011885640860905879,
      "grad_norm": 0.4569067656993866,
      "learning_rate": 1.9801223241590215e-06,
      "loss": 0.4055,
      "step": 259
    },
    {
      "epoch": 0.011931531366160342,
      "grad_norm": 0.4946267604827881,
      "learning_rate": 1.9877675840978594e-06,
      "loss": 0.407,
      "step": 260
    },
    {
      "epoch": 0.011977421871414803,
      "grad_norm": 0.44158419966697693,
      "learning_rate": 1.9954128440366973e-06,
      "loss": 0.4159,
      "step": 261
    },
    {
      "epoch": 0.012023312376669266,
      "grad_norm": 0.4439694881439209,
      "learning_rate": 2.003058103975535e-06,
      "loss": 0.3569,
      "step": 262
    },
    {
      "epoch": 0.01206920288192373,
      "grad_norm": 0.48543256521224976,
      "learning_rate": 2.0107033639143734e-06,
      "loss": 0.4774,
      "step": 263
    },
    {
      "epoch": 0.012115093387178193,
      "grad_norm": 0.501968264579773,
      "learning_rate": 2.0183486238532113e-06,
      "loss": 0.5504,
      "step": 264
    },
    {
      "epoch": 0.012160983892432656,
      "grad_norm": 0.45566222071647644,
      "learning_rate": 2.025993883792049e-06,
      "loss": 0.3634,
      "step": 265
    },
    {
      "epoch": 0.012206874397687119,
      "grad_norm": 0.565933346748352,
      "learning_rate": 2.033639143730887e-06,
      "loss": 0.5008,
      "step": 266
    },
    {
      "epoch": 0.012252764902941582,
      "grad_norm": 0.5437031984329224,
      "learning_rate": 2.041284403669725e-06,
      "loss": 0.537,
      "step": 267
    },
    {
      "epoch": 0.012298655408196045,
      "grad_norm": 0.443162739276886,
      "learning_rate": 2.048929663608563e-06,
      "loss": 0.3894,
      "step": 268
    },
    {
      "epoch": 0.012344545913450506,
      "grad_norm": 0.533871054649353,
      "learning_rate": 2.0565749235474007e-06,
      "loss": 0.5715,
      "step": 269
    },
    {
      "epoch": 0.01239043641870497,
      "grad_norm": 0.4236867427825928,
      "learning_rate": 2.064220183486239e-06,
      "loss": 0.3289,
      "step": 270
    },
    {
      "epoch": 0.012436326923959433,
      "grad_norm": 0.5032171607017517,
      "learning_rate": 2.0718654434250764e-06,
      "loss": 0.5618,
      "step": 271
    },
    {
      "epoch": 0.012482217429213896,
      "grad_norm": 0.5658868551254272,
      "learning_rate": 2.0795107033639143e-06,
      "loss": 0.5318,
      "step": 272
    },
    {
      "epoch": 0.012528107934468359,
      "grad_norm": 0.6331881880760193,
      "learning_rate": 2.0871559633027526e-06,
      "loss": 0.5536,
      "step": 273
    },
    {
      "epoch": 0.012573998439722822,
      "grad_norm": 0.4568878710269928,
      "learning_rate": 2.0948012232415905e-06,
      "loss": 0.407,
      "step": 274
    },
    {
      "epoch": 0.012619888944977285,
      "grad_norm": 0.4961564540863037,
      "learning_rate": 2.1024464831804283e-06,
      "loss": 0.4665,
      "step": 275
    },
    {
      "epoch": 0.012665779450231746,
      "grad_norm": 0.48572826385498047,
      "learning_rate": 2.110091743119266e-06,
      "loss": 0.4487,
      "step": 276
    },
    {
      "epoch": 0.01271166995548621,
      "grad_norm": 0.47618433833122253,
      "learning_rate": 2.117737003058104e-06,
      "loss": 0.4825,
      "step": 277
    },
    {
      "epoch": 0.012757560460740672,
      "grad_norm": 0.4763575494289398,
      "learning_rate": 2.125382262996942e-06,
      "loss": 0.4608,
      "step": 278
    },
    {
      "epoch": 0.012803450965995136,
      "grad_norm": 0.43774715065956116,
      "learning_rate": 2.13302752293578e-06,
      "loss": 0.4085,
      "step": 279
    },
    {
      "epoch": 0.012849341471249599,
      "grad_norm": 0.47849181294441223,
      "learning_rate": 2.140672782874618e-06,
      "loss": 0.391,
      "step": 280
    },
    {
      "epoch": 0.012895231976504062,
      "grad_norm": 0.5424180626869202,
      "learning_rate": 2.148318042813456e-06,
      "loss": 0.5967,
      "step": 281
    },
    {
      "epoch": 0.012941122481758525,
      "grad_norm": 0.4189923405647278,
      "learning_rate": 2.155963302752294e-06,
      "loss": 0.3116,
      "step": 282
    },
    {
      "epoch": 0.012987012987012988,
      "grad_norm": 0.48549145460128784,
      "learning_rate": 2.1636085626911317e-06,
      "loss": 0.513,
      "step": 283
    },
    {
      "epoch": 0.01303290349226745,
      "grad_norm": 0.5055712461471558,
      "learning_rate": 2.1712538226299696e-06,
      "loss": 0.4366,
      "step": 284
    },
    {
      "epoch": 0.013078793997521912,
      "grad_norm": 0.46339482069015503,
      "learning_rate": 2.1788990825688075e-06,
      "loss": 0.4272,
      "step": 285
    },
    {
      "epoch": 0.013124684502776375,
      "grad_norm": 0.42901337146759033,
      "learning_rate": 2.1865443425076453e-06,
      "loss": 0.3895,
      "step": 286
    },
    {
      "epoch": 0.013170575008030838,
      "grad_norm": 0.46788305044174194,
      "learning_rate": 2.1941896024464836e-06,
      "loss": 0.4193,
      "step": 287
    },
    {
      "epoch": 0.013216465513285302,
      "grad_norm": 0.43930676579475403,
      "learning_rate": 2.2018348623853215e-06,
      "loss": 0.3832,
      "step": 288
    },
    {
      "epoch": 0.013262356018539765,
      "grad_norm": 0.5346788763999939,
      "learning_rate": 2.209480122324159e-06,
      "loss": 0.4841,
      "step": 289
    },
    {
      "epoch": 0.013308246523794228,
      "grad_norm": 0.41882237792015076,
      "learning_rate": 2.2171253822629973e-06,
      "loss": 0.3196,
      "step": 290
    },
    {
      "epoch": 0.013354137029048689,
      "grad_norm": 0.45628708600997925,
      "learning_rate": 2.224770642201835e-06,
      "loss": 0.4657,
      "step": 291
    },
    {
      "epoch": 0.013400027534303152,
      "grad_norm": 0.49518153071403503,
      "learning_rate": 2.232415902140673e-06,
      "loss": 0.5638,
      "step": 292
    },
    {
      "epoch": 0.013445918039557615,
      "grad_norm": 0.4087590277194977,
      "learning_rate": 2.240061162079511e-06,
      "loss": 0.3098,
      "step": 293
    },
    {
      "epoch": 0.013491808544812078,
      "grad_norm": 0.5218579769134521,
      "learning_rate": 2.2477064220183487e-06,
      "loss": 0.4815,
      "step": 294
    },
    {
      "epoch": 0.013537699050066541,
      "grad_norm": 0.44696488976478577,
      "learning_rate": 2.2553516819571866e-06,
      "loss": 0.4104,
      "step": 295
    },
    {
      "epoch": 0.013583589555321005,
      "grad_norm": 0.4772244691848755,
      "learning_rate": 2.2629969418960245e-06,
      "loss": 0.4269,
      "step": 296
    },
    {
      "epoch": 0.013629480060575468,
      "grad_norm": 0.45414552092552185,
      "learning_rate": 2.2706422018348624e-06,
      "loss": 0.4195,
      "step": 297
    },
    {
      "epoch": 0.013675370565829929,
      "grad_norm": 0.4902088940143585,
      "learning_rate": 2.2782874617737006e-06,
      "loss": 0.3963,
      "step": 298
    },
    {
      "epoch": 0.013721261071084392,
      "grad_norm": 0.4744234085083008,
      "learning_rate": 2.2859327217125385e-06,
      "loss": 0.4487,
      "step": 299
    },
    {
      "epoch": 0.013767151576338855,
      "grad_norm": 0.46651050448417664,
      "learning_rate": 2.2935779816513764e-06,
      "loss": 0.4291,
      "step": 300
    },
    {
      "epoch": 0.013813042081593318,
      "grad_norm": 0.4345714747905731,
      "learning_rate": 2.3012232415902143e-06,
      "loss": 0.3359,
      "step": 301
    },
    {
      "epoch": 0.013858932586847781,
      "grad_norm": 0.5217494368553162,
      "learning_rate": 2.308868501529052e-06,
      "loss": 0.5498,
      "step": 302
    },
    {
      "epoch": 0.013904823092102244,
      "grad_norm": 0.4570966362953186,
      "learning_rate": 2.31651376146789e-06,
      "loss": 0.4328,
      "step": 303
    },
    {
      "epoch": 0.013950713597356708,
      "grad_norm": 0.48730015754699707,
      "learning_rate": 2.324159021406728e-06,
      "loss": 0.4382,
      "step": 304
    },
    {
      "epoch": 0.01399660410261117,
      "grad_norm": 0.41138121485710144,
      "learning_rate": 2.331804281345566e-06,
      "loss": 0.3028,
      "step": 305
    },
    {
      "epoch": 0.014042494607865632,
      "grad_norm": 0.46273642778396606,
      "learning_rate": 2.339449541284404e-06,
      "loss": 0.4055,
      "step": 306
    },
    {
      "epoch": 0.014088385113120095,
      "grad_norm": 0.6203993558883667,
      "learning_rate": 2.3470948012232415e-06,
      "loss": 0.4562,
      "step": 307
    },
    {
      "epoch": 0.014134275618374558,
      "grad_norm": 0.45870843529701233,
      "learning_rate": 2.35474006116208e-06,
      "loss": 0.4438,
      "step": 308
    },
    {
      "epoch": 0.014180166123629021,
      "grad_norm": 0.4719725251197815,
      "learning_rate": 2.3623853211009177e-06,
      "loss": 0.3947,
      "step": 309
    },
    {
      "epoch": 0.014226056628883484,
      "grad_norm": 0.48564717173576355,
      "learning_rate": 2.3700305810397555e-06,
      "loss": 0.4698,
      "step": 310
    },
    {
      "epoch": 0.014271947134137947,
      "grad_norm": 0.49916261434555054,
      "learning_rate": 2.3776758409785934e-06,
      "loss": 0.5144,
      "step": 311
    },
    {
      "epoch": 0.01431783763939241,
      "grad_norm": 0.48892831802368164,
      "learning_rate": 2.3853211009174317e-06,
      "loss": 0.4824,
      "step": 312
    },
    {
      "epoch": 0.014363728144646872,
      "grad_norm": 0.5027477145195007,
      "learning_rate": 2.392966360856269e-06,
      "loss": 0.4271,
      "step": 313
    },
    {
      "epoch": 0.014409618649901335,
      "grad_norm": 0.43566134572029114,
      "learning_rate": 2.400611620795107e-06,
      "loss": 0.3792,
      "step": 314
    },
    {
      "epoch": 0.014455509155155798,
      "grad_norm": 0.46502846479415894,
      "learning_rate": 2.4082568807339453e-06,
      "loss": 0.3786,
      "step": 315
    },
    {
      "epoch": 0.014501399660410261,
      "grad_norm": 0.46683400869369507,
      "learning_rate": 2.415902140672783e-06,
      "loss": 0.4447,
      "step": 316
    },
    {
      "epoch": 0.014547290165664724,
      "grad_norm": 0.5267835855484009,
      "learning_rate": 2.423547400611621e-06,
      "loss": 0.5284,
      "step": 317
    },
    {
      "epoch": 0.014593180670919187,
      "grad_norm": 0.48118987679481506,
      "learning_rate": 2.431192660550459e-06,
      "loss": 0.4114,
      "step": 318
    },
    {
      "epoch": 0.01463907117617365,
      "grad_norm": 0.4025678038597107,
      "learning_rate": 2.438837920489297e-06,
      "loss": 0.2934,
      "step": 319
    },
    {
      "epoch": 0.014684961681428112,
      "grad_norm": 0.3976534307003021,
      "learning_rate": 2.4464831804281347e-06,
      "loss": 0.28,
      "step": 320
    },
    {
      "epoch": 0.014730852186682575,
      "grad_norm": 0.4465334415435791,
      "learning_rate": 2.4541284403669725e-06,
      "loss": 0.3942,
      "step": 321
    },
    {
      "epoch": 0.014776742691937038,
      "grad_norm": 0.48250025510787964,
      "learning_rate": 2.461773700305811e-06,
      "loss": 0.4059,
      "step": 322
    },
    {
      "epoch": 0.014822633197191501,
      "grad_norm": 0.5287503600120544,
      "learning_rate": 2.4694189602446487e-06,
      "loss": 0.5658,
      "step": 323
    },
    {
      "epoch": 0.014868523702445964,
      "grad_norm": 0.45731842517852783,
      "learning_rate": 2.4770642201834866e-06,
      "loss": 0.4149,
      "step": 324
    },
    {
      "epoch": 0.014914414207700427,
      "grad_norm": 0.4629984200000763,
      "learning_rate": 2.4847094801223245e-06,
      "loss": 0.3735,
      "step": 325
    },
    {
      "epoch": 0.01496030471295489,
      "grad_norm": 0.5468152761459351,
      "learning_rate": 2.4923547400611623e-06,
      "loss": 0.5926,
      "step": 326
    },
    {
      "epoch": 0.015006195218209353,
      "grad_norm": 0.45142504572868347,
      "learning_rate": 2.5e-06,
      "loss": 0.3972,
      "step": 327
    },
    {
      "epoch": 0.015052085723463815,
      "grad_norm": 1.0734608173370361,
      "learning_rate": 2.507645259938838e-06,
      "loss": 0.5062,
      "step": 328
    },
    {
      "epoch": 0.015097976228718278,
      "grad_norm": 0.4371100664138794,
      "learning_rate": 2.515290519877676e-06,
      "loss": 0.3681,
      "step": 329
    },
    {
      "epoch": 0.01514386673397274,
      "grad_norm": 0.4611198604106903,
      "learning_rate": 2.522935779816514e-06,
      "loss": 0.3832,
      "step": 330
    },
    {
      "epoch": 0.015189757239227204,
      "grad_norm": 0.5871808528900146,
      "learning_rate": 2.530581039755352e-06,
      "loss": 0.3449,
      "step": 331
    },
    {
      "epoch": 0.015235647744481667,
      "grad_norm": 0.43380486965179443,
      "learning_rate": 2.5382262996941896e-06,
      "loss": 0.3395,
      "step": 332
    },
    {
      "epoch": 0.01528153824973613,
      "grad_norm": 0.4677191972732544,
      "learning_rate": 2.5458715596330274e-06,
      "loss": 0.4951,
      "step": 333
    },
    {
      "epoch": 0.015327428754990593,
      "grad_norm": 0.48182302713394165,
      "learning_rate": 2.5535168195718657e-06,
      "loss": 0.5163,
      "step": 334
    },
    {
      "epoch": 0.015373319260245055,
      "grad_norm": 0.563805878162384,
      "learning_rate": 2.5611620795107036e-06,
      "loss": 0.5234,
      "step": 335
    },
    {
      "epoch": 0.015419209765499518,
      "grad_norm": 0.512263834476471,
      "learning_rate": 2.5688073394495415e-06,
      "loss": 0.5025,
      "step": 336
    },
    {
      "epoch": 0.01546510027075398,
      "grad_norm": 0.43746107816696167,
      "learning_rate": 2.5764525993883793e-06,
      "loss": 0.3421,
      "step": 337
    },
    {
      "epoch": 0.015510990776008444,
      "grad_norm": 0.6869626641273499,
      "learning_rate": 2.5840978593272176e-06,
      "loss": 0.5717,
      "step": 338
    },
    {
      "epoch": 0.015556881281262907,
      "grad_norm": 0.44178250432014465,
      "learning_rate": 2.591743119266055e-06,
      "loss": 0.3735,
      "step": 339
    },
    {
      "epoch": 0.01560277178651737,
      "grad_norm": 0.4314460754394531,
      "learning_rate": 2.599388379204893e-06,
      "loss": 0.3144,
      "step": 340
    },
    {
      "epoch": 0.015648662291771833,
      "grad_norm": 0.48751088976860046,
      "learning_rate": 2.6070336391437313e-06,
      "loss": 0.476,
      "step": 341
    },
    {
      "epoch": 0.015694552797026296,
      "grad_norm": 0.4440633952617645,
      "learning_rate": 2.6146788990825687e-06,
      "loss": 0.3794,
      "step": 342
    },
    {
      "epoch": 0.01574044330228076,
      "grad_norm": 0.932782769203186,
      "learning_rate": 2.622324159021407e-06,
      "loss": 0.5274,
      "step": 343
    },
    {
      "epoch": 0.015786333807535222,
      "grad_norm": 0.50304114818573,
      "learning_rate": 2.629969418960245e-06,
      "loss": 0.4773,
      "step": 344
    },
    {
      "epoch": 0.015832224312789685,
      "grad_norm": 0.4992387890815735,
      "learning_rate": 2.6376146788990823e-06,
      "loss": 0.5398,
      "step": 345
    },
    {
      "epoch": 0.015878114818044145,
      "grad_norm": 0.4897677004337311,
      "learning_rate": 2.6452599388379206e-06,
      "loss": 0.4276,
      "step": 346
    },
    {
      "epoch": 0.015924005323298608,
      "grad_norm": 0.441156268119812,
      "learning_rate": 2.6529051987767585e-06,
      "loss": 0.3767,
      "step": 347
    },
    {
      "epoch": 0.01596989582855307,
      "grad_norm": 0.4375210702419281,
      "learning_rate": 2.6605504587155968e-06,
      "loss": 0.3923,
      "step": 348
    },
    {
      "epoch": 0.016015786333807534,
      "grad_norm": 0.4848385453224182,
      "learning_rate": 2.6681957186544342e-06,
      "loss": 0.3759,
      "step": 349
    },
    {
      "epoch": 0.016061676839061997,
      "grad_norm": 0.48047348856925964,
      "learning_rate": 2.6758409785932725e-06,
      "loss": 0.4541,
      "step": 350
    },
    {
      "epoch": 0.01610756734431646,
      "grad_norm": 0.48511138558387756,
      "learning_rate": 2.6834862385321104e-06,
      "loss": 0.4383,
      "step": 351
    },
    {
      "epoch": 0.016153457849570924,
      "grad_norm": 0.44000443816185,
      "learning_rate": 2.691131498470948e-06,
      "loss": 0.3572,
      "step": 352
    },
    {
      "epoch": 0.016199348354825387,
      "grad_norm": 0.5879843235015869,
      "learning_rate": 2.698776758409786e-06,
      "loss": 0.5802,
      "step": 353
    },
    {
      "epoch": 0.01624523886007985,
      "grad_norm": 0.5100506544113159,
      "learning_rate": 2.706422018348624e-06,
      "loss": 0.4733,
      "step": 354
    },
    {
      "epoch": 0.016291129365334313,
      "grad_norm": 0.4778331518173218,
      "learning_rate": 2.7140672782874623e-06,
      "loss": 0.4595,
      "step": 355
    },
    {
      "epoch": 0.016337019870588776,
      "grad_norm": 0.4732493460178375,
      "learning_rate": 2.7217125382262998e-06,
      "loss": 0.4049,
      "step": 356
    },
    {
      "epoch": 0.01638291037584324,
      "grad_norm": 0.4520963728427887,
      "learning_rate": 2.7293577981651376e-06,
      "loss": 0.4055,
      "step": 357
    },
    {
      "epoch": 0.016428800881097702,
      "grad_norm": 0.4668768346309662,
      "learning_rate": 2.737003058103976e-06,
      "loss": 0.4399,
      "step": 358
    },
    {
      "epoch": 0.016474691386352165,
      "grad_norm": 0.46692126989364624,
      "learning_rate": 2.7446483180428134e-06,
      "loss": 0.4235,
      "step": 359
    },
    {
      "epoch": 0.01652058189160663,
      "grad_norm": 0.5359144806861877,
      "learning_rate": 2.7522935779816517e-06,
      "loss": 0.5436,
      "step": 360
    },
    {
      "epoch": 0.016566472396861088,
      "grad_norm": 0.4567144811153412,
      "learning_rate": 2.7599388379204895e-06,
      "loss": 0.3992,
      "step": 361
    },
    {
      "epoch": 0.01661236290211555,
      "grad_norm": 0.4288550317287445,
      "learning_rate": 2.767584097859328e-06,
      "loss": 0.3498,
      "step": 362
    },
    {
      "epoch": 0.016658253407370014,
      "grad_norm": 0.4583730399608612,
      "learning_rate": 2.7752293577981653e-06,
      "loss": 0.3955,
      "step": 363
    },
    {
      "epoch": 0.016704143912624477,
      "grad_norm": 0.49381697177886963,
      "learning_rate": 2.782874617737003e-06,
      "loss": 0.4863,
      "step": 364
    },
    {
      "epoch": 0.01675003441787894,
      "grad_norm": 0.48270758986473083,
      "learning_rate": 2.7905198776758415e-06,
      "loss": 0.4584,
      "step": 365
    },
    {
      "epoch": 0.016795924923133403,
      "grad_norm": 0.4611847400665283,
      "learning_rate": 2.798165137614679e-06,
      "loss": 0.3522,
      "step": 366
    },
    {
      "epoch": 0.016841815428387866,
      "grad_norm": 0.4648805260658264,
      "learning_rate": 2.805810397553517e-06,
      "loss": 0.4502,
      "step": 367
    },
    {
      "epoch": 0.01688770593364233,
      "grad_norm": 0.49289172887802124,
      "learning_rate": 2.813455657492355e-06,
      "loss": 0.4656,
      "step": 368
    },
    {
      "epoch": 0.016933596438896793,
      "grad_norm": 0.4598466157913208,
      "learning_rate": 2.8211009174311925e-06,
      "loss": 0.3899,
      "step": 369
    },
    {
      "epoch": 0.016979486944151256,
      "grad_norm": 0.5077979564666748,
      "learning_rate": 2.828746177370031e-06,
      "loss": 0.3977,
      "step": 370
    },
    {
      "epoch": 0.01702537744940572,
      "grad_norm": 0.45774930715560913,
      "learning_rate": 2.8363914373088687e-06,
      "loss": 0.3997,
      "step": 371
    },
    {
      "epoch": 0.017071267954660182,
      "grad_norm": 0.43321484327316284,
      "learning_rate": 2.844036697247707e-06,
      "loss": 0.3554,
      "step": 372
    },
    {
      "epoch": 0.017117158459914645,
      "grad_norm": 0.43385207653045654,
      "learning_rate": 2.8516819571865444e-06,
      "loss": 0.341,
      "step": 373
    },
    {
      "epoch": 0.017163048965169108,
      "grad_norm": 0.5218827128410339,
      "learning_rate": 2.8593272171253827e-06,
      "loss": 0.5407,
      "step": 374
    },
    {
      "epoch": 0.017208939470423568,
      "grad_norm": 0.5167555809020996,
      "learning_rate": 2.8669724770642206e-06,
      "loss": 0.4346,
      "step": 375
    },
    {
      "epoch": 0.01725482997567803,
      "grad_norm": 0.510474443435669,
      "learning_rate": 2.874617737003058e-06,
      "loss": 0.5492,
      "step": 376
    },
    {
      "epoch": 0.017300720480932494,
      "grad_norm": 0.4777190685272217,
      "learning_rate": 2.8822629969418963e-06,
      "loss": 0.4308,
      "step": 377
    },
    {
      "epoch": 0.017346610986186957,
      "grad_norm": 0.48560985922813416,
      "learning_rate": 2.8899082568807342e-06,
      "loss": 0.4206,
      "step": 378
    },
    {
      "epoch": 0.01739250149144142,
      "grad_norm": 0.4789966642856598,
      "learning_rate": 2.8975535168195725e-06,
      "loss": 0.4558,
      "step": 379
    },
    {
      "epoch": 0.017438391996695883,
      "grad_norm": 0.4180901348590851,
      "learning_rate": 2.90519877675841e-06,
      "loss": 0.3134,
      "step": 380
    },
    {
      "epoch": 0.017484282501950346,
      "grad_norm": 0.4317328631877899,
      "learning_rate": 2.912844036697248e-06,
      "loss": 0.3532,
      "step": 381
    },
    {
      "epoch": 0.01753017300720481,
      "grad_norm": 0.5050603151321411,
      "learning_rate": 2.920489296636086e-06,
      "loss": 0.5576,
      "step": 382
    },
    {
      "epoch": 0.017576063512459272,
      "grad_norm": 0.49497556686401367,
      "learning_rate": 2.9281345565749236e-06,
      "loss": 0.5347,
      "step": 383
    },
    {
      "epoch": 0.017621954017713735,
      "grad_norm": 0.5131199359893799,
      "learning_rate": 2.935779816513762e-06,
      "loss": 0.4114,
      "step": 384
    },
    {
      "epoch": 0.0176678445229682,
      "grad_norm": 0.4344429671764374,
      "learning_rate": 2.9434250764525997e-06,
      "loss": 0.3587,
      "step": 385
    },
    {
      "epoch": 0.01771373502822266,
      "grad_norm": 0.45279866456985474,
      "learning_rate": 2.9510703363914376e-06,
      "loss": 0.3789,
      "step": 386
    },
    {
      "epoch": 0.017759625533477125,
      "grad_norm": 0.5409918427467346,
      "learning_rate": 2.9587155963302755e-06,
      "loss": 0.3814,
      "step": 387
    },
    {
      "epoch": 0.017805516038731588,
      "grad_norm": 0.5497090220451355,
      "learning_rate": 2.9663608562691134e-06,
      "loss": 0.5602,
      "step": 388
    },
    {
      "epoch": 0.01785140654398605,
      "grad_norm": 0.47578442096710205,
      "learning_rate": 2.9740061162079512e-06,
      "loss": 0.4324,
      "step": 389
    },
    {
      "epoch": 0.01789729704924051,
      "grad_norm": 0.47323742508888245,
      "learning_rate": 2.981651376146789e-06,
      "loss": 0.4478,
      "step": 390
    },
    {
      "epoch": 0.017943187554494974,
      "grad_norm": 0.4371729791164398,
      "learning_rate": 2.9892966360856274e-06,
      "loss": 0.3612,
      "step": 391
    },
    {
      "epoch": 0.017989078059749437,
      "grad_norm": 0.6900670528411865,
      "learning_rate": 2.9969418960244653e-06,
      "loss": 0.6762,
      "step": 392
    },
    {
      "epoch": 0.0180349685650039,
      "grad_norm": 0.5032302141189575,
      "learning_rate": 3.0045871559633027e-06,
      "loss": 0.5237,
      "step": 393
    },
    {
      "epoch": 0.018080859070258363,
      "grad_norm": 0.45847564935684204,
      "learning_rate": 3.012232415902141e-06,
      "loss": 0.357,
      "step": 394
    },
    {
      "epoch": 0.018126749575512826,
      "grad_norm": 0.48517248034477234,
      "learning_rate": 3.019877675840979e-06,
      "loss": 0.4645,
      "step": 395
    },
    {
      "epoch": 0.01817264008076729,
      "grad_norm": 0.47780027985572815,
      "learning_rate": 3.0275229357798168e-06,
      "loss": 0.4434,
      "step": 396
    },
    {
      "epoch": 0.018218530586021752,
      "grad_norm": 0.455100417137146,
      "learning_rate": 3.0351681957186546e-06,
      "loss": 0.4364,
      "step": 397
    },
    {
      "epoch": 0.018264421091276215,
      "grad_norm": 0.43008488416671753,
      "learning_rate": 3.042813455657493e-06,
      "loss": 0.3327,
      "step": 398
    },
    {
      "epoch": 0.01831031159653068,
      "grad_norm": 0.5193257331848145,
      "learning_rate": 3.0504587155963304e-06,
      "loss": 0.4522,
      "step": 399
    },
    {
      "epoch": 0.01835620210178514,
      "grad_norm": 0.45401597023010254,
      "learning_rate": 3.0581039755351682e-06,
      "loss": 0.4047,
      "step": 400
    },
    {
      "epoch": 0.018402092607039604,
      "grad_norm": 0.43108049035072327,
      "learning_rate": 3.0657492354740065e-06,
      "loss": 0.3774,
      "step": 401
    },
    {
      "epoch": 0.018447983112294068,
      "grad_norm": 0.46105724573135376,
      "learning_rate": 3.073394495412844e-06,
      "loss": 0.3984,
      "step": 402
    },
    {
      "epoch": 0.01849387361754853,
      "grad_norm": 0.502966582775116,
      "learning_rate": 3.0810397553516823e-06,
      "loss": 0.4595,
      "step": 403
    },
    {
      "epoch": 0.018539764122802994,
      "grad_norm": 0.48311692476272583,
      "learning_rate": 3.08868501529052e-06,
      "loss": 0.4275,
      "step": 404
    },
    {
      "epoch": 0.018585654628057453,
      "grad_norm": 0.4644903838634491,
      "learning_rate": 3.0963302752293576e-06,
      "loss": 0.3781,
      "step": 405
    },
    {
      "epoch": 0.018631545133311916,
      "grad_norm": 0.5004028081893921,
      "learning_rate": 3.103975535168196e-06,
      "loss": 0.4339,
      "step": 406
    },
    {
      "epoch": 0.01867743563856638,
      "grad_norm": 0.4505075514316559,
      "learning_rate": 3.1116207951070338e-06,
      "loss": 0.3918,
      "step": 407
    },
    {
      "epoch": 0.018723326143820843,
      "grad_norm": 0.4375948905944824,
      "learning_rate": 3.119266055045872e-06,
      "loss": 0.362,
      "step": 408
    },
    {
      "epoch": 0.018769216649075306,
      "grad_norm": 0.4854259490966797,
      "learning_rate": 3.1269113149847095e-06,
      "loss": 0.4676,
      "step": 409
    },
    {
      "epoch": 0.01881510715432977,
      "grad_norm": 0.4625663161277771,
      "learning_rate": 3.134556574923548e-06,
      "loss": 0.4561,
      "step": 410
    },
    {
      "epoch": 0.018860997659584232,
      "grad_norm": 0.46960076689720154,
      "learning_rate": 3.1422018348623857e-06,
      "loss": 0.4868,
      "step": 411
    },
    {
      "epoch": 0.018906888164838695,
      "grad_norm": 0.5641589164733887,
      "learning_rate": 3.149847094801223e-06,
      "loss": 0.4322,
      "step": 412
    },
    {
      "epoch": 0.018952778670093158,
      "grad_norm": 0.5606700778007507,
      "learning_rate": 3.1574923547400614e-06,
      "loss": 0.4834,
      "step": 413
    },
    {
      "epoch": 0.01899866917534762,
      "grad_norm": 0.576024055480957,
      "learning_rate": 3.1651376146788993e-06,
      "loss": 0.5227,
      "step": 414
    },
    {
      "epoch": 0.019044559680602084,
      "grad_norm": 0.4867952764034271,
      "learning_rate": 3.1727828746177376e-06,
      "loss": 0.4878,
      "step": 415
    },
    {
      "epoch": 0.019090450185856547,
      "grad_norm": 0.4988221228122711,
      "learning_rate": 3.180428134556575e-06,
      "loss": 0.4424,
      "step": 416
    },
    {
      "epoch": 0.01913634069111101,
      "grad_norm": 0.4742036461830139,
      "learning_rate": 3.188073394495413e-06,
      "loss": 0.4811,
      "step": 417
    },
    {
      "epoch": 0.019182231196365473,
      "grad_norm": 0.4699513614177704,
      "learning_rate": 3.195718654434251e-06,
      "loss": 0.4354,
      "step": 418
    },
    {
      "epoch": 0.019228121701619937,
      "grad_norm": 0.46931612491607666,
      "learning_rate": 3.2033639143730887e-06,
      "loss": 0.3925,
      "step": 419
    },
    {
      "epoch": 0.019274012206874396,
      "grad_norm": 0.4854743182659149,
      "learning_rate": 3.211009174311927e-06,
      "loss": 0.3773,
      "step": 420
    },
    {
      "epoch": 0.01931990271212886,
      "grad_norm": 0.5192313194274902,
      "learning_rate": 3.218654434250765e-06,
      "loss": 0.5547,
      "step": 421
    },
    {
      "epoch": 0.019365793217383322,
      "grad_norm": 0.47898295521736145,
      "learning_rate": 3.226299694189603e-06,
      "loss": 0.4937,
      "step": 422
    },
    {
      "epoch": 0.019411683722637785,
      "grad_norm": 0.46824750304222107,
      "learning_rate": 3.2339449541284406e-06,
      "loss": 0.4389,
      "step": 423
    },
    {
      "epoch": 0.01945757422789225,
      "grad_norm": 0.5154852867126465,
      "learning_rate": 3.2415902140672784e-06,
      "loss": 0.4951,
      "step": 424
    },
    {
      "epoch": 0.01950346473314671,
      "grad_norm": 0.4472064971923828,
      "learning_rate": 3.2492354740061167e-06,
      "loss": 0.3962,
      "step": 425
    },
    {
      "epoch": 0.019549355238401175,
      "grad_norm": 0.4869037866592407,
      "learning_rate": 3.256880733944954e-06,
      "loss": 0.448,
      "step": 426
    },
    {
      "epoch": 0.019595245743655638,
      "grad_norm": 0.436694473028183,
      "learning_rate": 3.2645259938837925e-06,
      "loss": 0.3469,
      "step": 427
    },
    {
      "epoch": 0.0196411362489101,
      "grad_norm": 0.5162097215652466,
      "learning_rate": 3.2721712538226303e-06,
      "loss": 0.4408,
      "step": 428
    },
    {
      "epoch": 0.019687026754164564,
      "grad_norm": 0.5034533143043518,
      "learning_rate": 3.279816513761468e-06,
      "loss": 0.4719,
      "step": 429
    },
    {
      "epoch": 0.019732917259419027,
      "grad_norm": 0.4772058427333832,
      "learning_rate": 3.287461773700306e-06,
      "loss": 0.417,
      "step": 430
    },
    {
      "epoch": 0.01977880776467349,
      "grad_norm": 0.5042054057121277,
      "learning_rate": 3.295107033639144e-06,
      "loss": 0.4666,
      "step": 431
    },
    {
      "epoch": 0.019824698269927953,
      "grad_norm": 0.4831267297267914,
      "learning_rate": 3.3027522935779823e-06,
      "loss": 0.4666,
      "step": 432
    },
    {
      "epoch": 0.019870588775182416,
      "grad_norm": 0.462446004152298,
      "learning_rate": 3.3103975535168197e-06,
      "loss": 0.3896,
      "step": 433
    },
    {
      "epoch": 0.019916479280436876,
      "grad_norm": 0.4908536374568939,
      "learning_rate": 3.318042813455658e-06,
      "loss": 0.5126,
      "step": 434
    },
    {
      "epoch": 0.01996236978569134,
      "grad_norm": 0.5101654529571533,
      "learning_rate": 3.325688073394496e-06,
      "loss": 0.5114,
      "step": 435
    },
    {
      "epoch": 0.020008260290945802,
      "grad_norm": 0.4652755856513977,
      "learning_rate": 3.3333333333333333e-06,
      "loss": 0.4303,
      "step": 436
    },
    {
      "epoch": 0.020054150796200265,
      "grad_norm": 0.4511263370513916,
      "learning_rate": 3.3409785932721716e-06,
      "loss": 0.4182,
      "step": 437
    },
    {
      "epoch": 0.02010004130145473,
      "grad_norm": 0.4430512487888336,
      "learning_rate": 3.3486238532110095e-06,
      "loss": 0.3437,
      "step": 438
    },
    {
      "epoch": 0.02014593180670919,
      "grad_norm": 0.5460695624351501,
      "learning_rate": 3.3562691131498478e-06,
      "loss": 0.5211,
      "step": 439
    },
    {
      "epoch": 0.020191822311963654,
      "grad_norm": 0.46814751625061035,
      "learning_rate": 3.3639143730886852e-06,
      "loss": 0.4979,
      "step": 440
    },
    {
      "epoch": 0.020237712817218118,
      "grad_norm": 0.451825886964798,
      "learning_rate": 3.371559633027523e-06,
      "loss": 0.358,
      "step": 441
    },
    {
      "epoch": 0.02028360332247258,
      "grad_norm": 0.48225319385528564,
      "learning_rate": 3.3792048929663614e-06,
      "loss": 0.4531,
      "step": 442
    },
    {
      "epoch": 0.020329493827727044,
      "grad_norm": 0.441541850566864,
      "learning_rate": 3.386850152905199e-06,
      "loss": 0.3809,
      "step": 443
    },
    {
      "epoch": 0.020375384332981507,
      "grad_norm": 0.9595709443092346,
      "learning_rate": 3.394495412844037e-06,
      "loss": 0.4303,
      "step": 444
    },
    {
      "epoch": 0.02042127483823597,
      "grad_norm": 0.6199988126754761,
      "learning_rate": 3.402140672782875e-06,
      "loss": 0.6407,
      "step": 445
    },
    {
      "epoch": 0.020467165343490433,
      "grad_norm": 0.5241382718086243,
      "learning_rate": 3.4097859327217125e-06,
      "loss": 0.5388,
      "step": 446
    },
    {
      "epoch": 0.020513055848744896,
      "grad_norm": 0.4802880883216858,
      "learning_rate": 3.4174311926605508e-06,
      "loss": 0.4601,
      "step": 447
    },
    {
      "epoch": 0.02055894635399936,
      "grad_norm": 0.462083101272583,
      "learning_rate": 3.4250764525993886e-06,
      "loss": 0.4682,
      "step": 448
    },
    {
      "epoch": 0.02060483685925382,
      "grad_norm": 0.43879276514053345,
      "learning_rate": 3.432721712538227e-06,
      "loss": 0.3861,
      "step": 449
    },
    {
      "epoch": 0.020650727364508282,
      "grad_norm": 0.5051079988479614,
      "learning_rate": 3.4403669724770644e-06,
      "loss": 0.5445,
      "step": 450
    },
    {
      "epoch": 0.020696617869762745,
      "grad_norm": 0.46442314982414246,
      "learning_rate": 3.4480122324159027e-06,
      "loss": 0.4028,
      "step": 451
    },
    {
      "epoch": 0.020742508375017208,
      "grad_norm": 0.44685596227645874,
      "learning_rate": 3.4556574923547405e-06,
      "loss": 0.3928,
      "step": 452
    },
    {
      "epoch": 0.02078839888027167,
      "grad_norm": 0.4670540988445282,
      "learning_rate": 3.463302752293578e-06,
      "loss": 0.3846,
      "step": 453
    },
    {
      "epoch": 0.020834289385526134,
      "grad_norm": 0.5065159797668457,
      "learning_rate": 3.4709480122324163e-06,
      "loss": 0.426,
      "step": 454
    },
    {
      "epoch": 0.020880179890780597,
      "grad_norm": 0.45759421586990356,
      "learning_rate": 3.478593272171254e-06,
      "loss": 0.3778,
      "step": 455
    },
    {
      "epoch": 0.02092607039603506,
      "grad_norm": 0.526719331741333,
      "learning_rate": 3.486238532110092e-06,
      "loss": 0.4252,
      "step": 456
    },
    {
      "epoch": 0.020971960901289523,
      "grad_norm": 0.4285305142402649,
      "learning_rate": 3.49388379204893e-06,
      "loss": 0.3334,
      "step": 457
    },
    {
      "epoch": 0.021017851406543987,
      "grad_norm": 0.5938563346862793,
      "learning_rate": 3.5015290519877678e-06,
      "loss": 0.4861,
      "step": 458
    },
    {
      "epoch": 0.02106374191179845,
      "grad_norm": 0.5004566311836243,
      "learning_rate": 3.5091743119266056e-06,
      "loss": 0.4418,
      "step": 459
    },
    {
      "epoch": 0.021109632417052913,
      "grad_norm": 0.497659832239151,
      "learning_rate": 3.5168195718654435e-06,
      "loss": 0.4686,
      "step": 460
    },
    {
      "epoch": 0.021155522922307376,
      "grad_norm": 0.4815540909767151,
      "learning_rate": 3.524464831804282e-06,
      "loss": 0.4217,
      "step": 461
    },
    {
      "epoch": 0.02120141342756184,
      "grad_norm": 0.43580955266952515,
      "learning_rate": 3.5321100917431193e-06,
      "loss": 0.295,
      "step": 462
    },
    {
      "epoch": 0.021247303932816302,
      "grad_norm": 0.46454644203186035,
      "learning_rate": 3.5397553516819576e-06,
      "loss": 0.4484,
      "step": 463
    },
    {
      "epoch": 0.02129319443807076,
      "grad_norm": 0.5093151926994324,
      "learning_rate": 3.5474006116207954e-06,
      "loss": 0.3842,
      "step": 464
    },
    {
      "epoch": 0.021339084943325225,
      "grad_norm": 0.4791293144226074,
      "learning_rate": 3.5550458715596333e-06,
      "loss": 0.4639,
      "step": 465
    },
    {
      "epoch": 0.021384975448579688,
      "grad_norm": 0.47072818875312805,
      "learning_rate": 3.562691131498471e-06,
      "loss": 0.4182,
      "step": 466
    },
    {
      "epoch": 0.02143086595383415,
      "grad_norm": 0.49291640520095825,
      "learning_rate": 3.570336391437309e-06,
      "loss": 0.4588,
      "step": 467
    },
    {
      "epoch": 0.021476756459088614,
      "grad_norm": 0.46959802508354187,
      "learning_rate": 3.5779816513761473e-06,
      "loss": 0.4093,
      "step": 468
    },
    {
      "epoch": 0.021522646964343077,
      "grad_norm": 0.4998926818370819,
      "learning_rate": 3.585626911314985e-06,
      "loss": 0.4976,
      "step": 469
    },
    {
      "epoch": 0.02156853746959754,
      "grad_norm": 0.43465107679367065,
      "learning_rate": 3.5932721712538227e-06,
      "loss": 0.3593,
      "step": 470
    },
    {
      "epoch": 0.021614427974852003,
      "grad_norm": 0.5242964029312134,
      "learning_rate": 3.600917431192661e-06,
      "loss": 0.5659,
      "step": 471
    },
    {
      "epoch": 0.021660318480106466,
      "grad_norm": 0.4925539493560791,
      "learning_rate": 3.6085626911314984e-06,
      "loss": 0.4773,
      "step": 472
    },
    {
      "epoch": 0.02170620898536093,
      "grad_norm": 0.44556525349617004,
      "learning_rate": 3.6162079510703367e-06,
      "loss": 0.3939,
      "step": 473
    },
    {
      "epoch": 0.021752099490615392,
      "grad_norm": 0.46809321641921997,
      "learning_rate": 3.6238532110091746e-06,
      "loss": 0.3871,
      "step": 474
    },
    {
      "epoch": 0.021797989995869856,
      "grad_norm": 0.5198566317558289,
      "learning_rate": 3.631498470948013e-06,
      "loss": 0.4929,
      "step": 475
    },
    {
      "epoch": 0.02184388050112432,
      "grad_norm": 0.4656904339790344,
      "learning_rate": 3.6391437308868503e-06,
      "loss": 0.3858,
      "step": 476
    },
    {
      "epoch": 0.021889771006378782,
      "grad_norm": 0.6396805047988892,
      "learning_rate": 3.646788990825688e-06,
      "loss": 0.5824,
      "step": 477
    },
    {
      "epoch": 0.02193566151163324,
      "grad_norm": 0.5012964606285095,
      "learning_rate": 3.6544342507645265e-06,
      "loss": 0.4931,
      "step": 478
    },
    {
      "epoch": 0.021981552016887704,
      "grad_norm": 0.5147992372512817,
      "learning_rate": 3.662079510703364e-06,
      "loss": 0.557,
      "step": 479
    },
    {
      "epoch": 0.022027442522142168,
      "grad_norm": 0.5051838159561157,
      "learning_rate": 3.6697247706422022e-06,
      "loss": 0.5182,
      "step": 480
    },
    {
      "epoch": 0.02207333302739663,
      "grad_norm": 0.4526509940624237,
      "learning_rate": 3.67737003058104e-06,
      "loss": 0.3788,
      "step": 481
    },
    {
      "epoch": 0.022119223532651094,
      "grad_norm": 0.4833812713623047,
      "learning_rate": 3.6850152905198775e-06,
      "loss": 0.4043,
      "step": 482
    },
    {
      "epoch": 0.022165114037905557,
      "grad_norm": 0.450137197971344,
      "learning_rate": 3.692660550458716e-06,
      "loss": 0.3674,
      "step": 483
    },
    {
      "epoch": 0.02221100454316002,
      "grad_norm": 0.472493976354599,
      "learning_rate": 3.7003058103975537e-06,
      "loss": 0.4229,
      "step": 484
    },
    {
      "epoch": 0.022256895048414483,
      "grad_norm": 0.460699200630188,
      "learning_rate": 3.707951070336392e-06,
      "loss": 0.386,
      "step": 485
    },
    {
      "epoch": 0.022302785553668946,
      "grad_norm": 0.43607982993125916,
      "learning_rate": 3.7155963302752295e-06,
      "loss": 0.3474,
      "step": 486
    },
    {
      "epoch": 0.02234867605892341,
      "grad_norm": 0.4386248290538788,
      "learning_rate": 3.7232415902140678e-06,
      "loss": 0.4147,
      "step": 487
    },
    {
      "epoch": 0.022394566564177872,
      "grad_norm": 0.4909244775772095,
      "learning_rate": 3.7308868501529056e-06,
      "loss": 0.5338,
      "step": 488
    },
    {
      "epoch": 0.022440457069432335,
      "grad_norm": 0.45259836316108704,
      "learning_rate": 3.738532110091743e-06,
      "loss": 0.3678,
      "step": 489
    },
    {
      "epoch": 0.0224863475746868,
      "grad_norm": 0.5172081589698792,
      "learning_rate": 3.7461773700305814e-06,
      "loss": 0.5185,
      "step": 490
    },
    {
      "epoch": 0.02253223807994126,
      "grad_norm": 0.45696645975112915,
      "learning_rate": 3.7538226299694192e-06,
      "loss": 0.4019,
      "step": 491
    },
    {
      "epoch": 0.022578128585195725,
      "grad_norm": 0.489422082901001,
      "learning_rate": 3.7614678899082575e-06,
      "loss": 0.4353,
      "step": 492
    },
    {
      "epoch": 0.022624019090450184,
      "grad_norm": 0.4883096516132355,
      "learning_rate": 3.769113149847095e-06,
      "loss": 0.4958,
      "step": 493
    },
    {
      "epoch": 0.022669909595704647,
      "grad_norm": 0.48254308104515076,
      "learning_rate": 3.776758409785933e-06,
      "loss": 0.4422,
      "step": 494
    },
    {
      "epoch": 0.02271580010095911,
      "grad_norm": 0.4923512041568756,
      "learning_rate": 3.784403669724771e-06,
      "loss": 0.4321,
      "step": 495
    },
    {
      "epoch": 0.022761690606213573,
      "grad_norm": 0.5353966355323792,
      "learning_rate": 3.7920489296636086e-06,
      "loss": 0.5279,
      "step": 496
    },
    {
      "epoch": 0.022807581111468037,
      "grad_norm": 0.49044910073280334,
      "learning_rate": 3.799694189602447e-06,
      "loss": 0.4661,
      "step": 497
    },
    {
      "epoch": 0.0228534716167225,
      "grad_norm": 0.44412946701049805,
      "learning_rate": 3.8073394495412848e-06,
      "loss": 0.359,
      "step": 498
    },
    {
      "epoch": 0.022899362121976963,
      "grad_norm": 0.5708341002464294,
      "learning_rate": 3.814984709480123e-06,
      "loss": 0.4945,
      "step": 499
    },
    {
      "epoch": 0.022945252627231426,
      "grad_norm": 0.45246854424476624,
      "learning_rate": 3.8226299694189605e-06,
      "loss": 0.4112,
      "step": 500
    },
    {
      "epoch": 0.02299114313248589,
      "grad_norm": 0.4905444085597992,
      "learning_rate": 3.830275229357798e-06,
      "loss": 0.4717,
      "step": 501
    },
    {
      "epoch": 0.023037033637740352,
      "grad_norm": 0.447233110666275,
      "learning_rate": 3.837920489296636e-06,
      "loss": 0.367,
      "step": 502
    },
    {
      "epoch": 0.023082924142994815,
      "grad_norm": 0.5000691413879395,
      "learning_rate": 3.8455657492354746e-06,
      "loss": 0.4311,
      "step": 503
    },
    {
      "epoch": 0.023128814648249278,
      "grad_norm": 0.4629581570625305,
      "learning_rate": 3.853211009174313e-06,
      "loss": 0.4196,
      "step": 504
    },
    {
      "epoch": 0.02317470515350374,
      "grad_norm": 0.42086851596832275,
      "learning_rate": 3.86085626911315e-06,
      "loss": 0.3129,
      "step": 505
    },
    {
      "epoch": 0.023220595658758204,
      "grad_norm": 1.1491636037826538,
      "learning_rate": 3.868501529051988e-06,
      "loss": 0.5884,
      "step": 506
    },
    {
      "epoch": 0.023266486164012667,
      "grad_norm": 0.4791019856929779,
      "learning_rate": 3.876146788990826e-06,
      "loss": 0.4413,
      "step": 507
    },
    {
      "epoch": 0.023312376669267127,
      "grad_norm": 0.5129106640815735,
      "learning_rate": 3.8837920489296635e-06,
      "loss": 0.5112,
      "step": 508
    },
    {
      "epoch": 0.02335826717452159,
      "grad_norm": 0.4540897309780121,
      "learning_rate": 3.891437308868502e-06,
      "loss": 0.4214,
      "step": 509
    },
    {
      "epoch": 0.023404157679776053,
      "grad_norm": 0.4633723795413971,
      "learning_rate": 3.89908256880734e-06,
      "loss": 0.4294,
      "step": 510
    },
    {
      "epoch": 0.023450048185030516,
      "grad_norm": 0.45631659030914307,
      "learning_rate": 3.906727828746178e-06,
      "loss": 0.355,
      "step": 511
    },
    {
      "epoch": 0.02349593869028498,
      "grad_norm": 0.7669097185134888,
      "learning_rate": 3.914373088685016e-06,
      "loss": 0.527,
      "step": 512
    },
    {
      "epoch": 0.023541829195539443,
      "grad_norm": 0.5041212439537048,
      "learning_rate": 3.922018348623853e-06,
      "loss": 0.4816,
      "step": 513
    },
    {
      "epoch": 0.023587719700793906,
      "grad_norm": 0.49600949883461,
      "learning_rate": 3.9296636085626916e-06,
      "loss": 0.5172,
      "step": 514
    },
    {
      "epoch": 0.02363361020604837,
      "grad_norm": 0.43761149048805237,
      "learning_rate": 3.937308868501529e-06,
      "loss": 0.3567,
      "step": 515
    },
    {
      "epoch": 0.023679500711302832,
      "grad_norm": 0.44402939081192017,
      "learning_rate": 3.944954128440367e-06,
      "loss": 0.4105,
      "step": 516
    },
    {
      "epoch": 0.023725391216557295,
      "grad_norm": 0.4629693925380707,
      "learning_rate": 3.952599388379206e-06,
      "loss": 0.3719,
      "step": 517
    },
    {
      "epoch": 0.023771281721811758,
      "grad_norm": 0.4541211426258087,
      "learning_rate": 3.960244648318043e-06,
      "loss": 0.4061,
      "step": 518
    },
    {
      "epoch": 0.02381717222706622,
      "grad_norm": 0.5160866379737854,
      "learning_rate": 3.967889908256881e-06,
      "loss": 0.4723,
      "step": 519
    },
    {
      "epoch": 0.023863062732320684,
      "grad_norm": 0.4838929772377014,
      "learning_rate": 3.975535168195719e-06,
      "loss": 0.4531,
      "step": 520
    },
    {
      "epoch": 0.023908953237575147,
      "grad_norm": 0.4735890030860901,
      "learning_rate": 3.983180428134557e-06,
      "loss": 0.461,
      "step": 521
    },
    {
      "epoch": 0.023954843742829607,
      "grad_norm": 0.46348604559898376,
      "learning_rate": 3.9908256880733945e-06,
      "loss": 0.4249,
      "step": 522
    },
    {
      "epoch": 0.02400073424808407,
      "grad_norm": 0.6115723848342896,
      "learning_rate": 3.998470948012233e-06,
      "loss": 0.6942,
      "step": 523
    },
    {
      "epoch": 0.024046624753338533,
      "grad_norm": 0.5280125737190247,
      "learning_rate": 4.00611620795107e-06,
      "loss": 0.5101,
      "step": 524
    },
    {
      "epoch": 0.024092515258592996,
      "grad_norm": 0.4911775290966034,
      "learning_rate": 4.013761467889909e-06,
      "loss": 0.403,
      "step": 525
    },
    {
      "epoch": 0.02413840576384746,
      "grad_norm": 0.5467319488525391,
      "learning_rate": 4.021406727828747e-06,
      "loss": 0.4585,
      "step": 526
    },
    {
      "epoch": 0.024184296269101922,
      "grad_norm": 0.4691978096961975,
      "learning_rate": 4.029051987767584e-06,
      "loss": 0.4003,
      "step": 527
    },
    {
      "epoch": 0.024230186774356385,
      "grad_norm": 0.523053765296936,
      "learning_rate": 4.036697247706423e-06,
      "loss": 0.5734,
      "step": 528
    },
    {
      "epoch": 0.02427607727961085,
      "grad_norm": 0.45209935307502747,
      "learning_rate": 4.04434250764526e-06,
      "loss": 0.3715,
      "step": 529
    },
    {
      "epoch": 0.02432196778486531,
      "grad_norm": 0.48233136534690857,
      "learning_rate": 4.051987767584098e-06,
      "loss": 0.4148,
      "step": 530
    },
    {
      "epoch": 0.024367858290119775,
      "grad_norm": 0.4820379316806793,
      "learning_rate": 4.059633027522936e-06,
      "loss": 0.3949,
      "step": 531
    },
    {
      "epoch": 0.024413748795374238,
      "grad_norm": 0.5382937788963318,
      "learning_rate": 4.067278287461774e-06,
      "loss": 0.5764,
      "step": 532
    },
    {
      "epoch": 0.0244596393006287,
      "grad_norm": 0.4763702154159546,
      "learning_rate": 4.074923547400612e-06,
      "loss": 0.4179,
      "step": 533
    },
    {
      "epoch": 0.024505529805883164,
      "grad_norm": 0.5012703537940979,
      "learning_rate": 4.08256880733945e-06,
      "loss": 0.4761,
      "step": 534
    },
    {
      "epoch": 0.024551420311137627,
      "grad_norm": 0.5251275300979614,
      "learning_rate": 4.090214067278288e-06,
      "loss": 0.4383,
      "step": 535
    },
    {
      "epoch": 0.02459731081639209,
      "grad_norm": 0.4684962332248688,
      "learning_rate": 4.097859327217126e-06,
      "loss": 0.401,
      "step": 536
    },
    {
      "epoch": 0.02464320132164655,
      "grad_norm": 0.5177205801010132,
      "learning_rate": 4.105504587155963e-06,
      "loss": 0.5061,
      "step": 537
    },
    {
      "epoch": 0.024689091826901013,
      "grad_norm": 0.4861942529678345,
      "learning_rate": 4.113149847094801e-06,
      "loss": 0.4672,
      "step": 538
    },
    {
      "epoch": 0.024734982332155476,
      "grad_norm": 0.5345531702041626,
      "learning_rate": 4.12079510703364e-06,
      "loss": 0.4102,
      "step": 539
    },
    {
      "epoch": 0.02478087283740994,
      "grad_norm": 0.506618857383728,
      "learning_rate": 4.128440366972478e-06,
      "loss": 0.4757,
      "step": 540
    },
    {
      "epoch": 0.024826763342664402,
      "grad_norm": 0.4715940058231354,
      "learning_rate": 4.136085626911315e-06,
      "loss": 0.3911,
      "step": 541
    },
    {
      "epoch": 0.024872653847918865,
      "grad_norm": 0.4640718400478363,
      "learning_rate": 4.143730886850153e-06,
      "loss": 0.3523,
      "step": 542
    },
    {
      "epoch": 0.024918544353173328,
      "grad_norm": 0.5020986199378967,
      "learning_rate": 4.151376146788991e-06,
      "loss": 0.4585,
      "step": 543
    },
    {
      "epoch": 0.02496443485842779,
      "grad_norm": 0.4291110634803772,
      "learning_rate": 4.1590214067278286e-06,
      "loss": 0.329,
      "step": 544
    },
    {
      "epoch": 0.025010325363682254,
      "grad_norm": 0.4956706762313843,
      "learning_rate": 4.166666666666667e-06,
      "loss": 0.4538,
      "step": 545
    },
    {
      "epoch": 0.025056215868936717,
      "grad_norm": 0.4470667839050293,
      "learning_rate": 4.174311926605505e-06,
      "loss": 0.3776,
      "step": 546
    },
    {
      "epoch": 0.02510210637419118,
      "grad_norm": 0.4709324538707733,
      "learning_rate": 4.181957186544343e-06,
      "loss": 0.4152,
      "step": 547
    },
    {
      "epoch": 0.025147996879445644,
      "grad_norm": 0.47205275297164917,
      "learning_rate": 4.189602446483181e-06,
      "loss": 0.3729,
      "step": 548
    },
    {
      "epoch": 0.025193887384700107,
      "grad_norm": 0.48534488677978516,
      "learning_rate": 4.197247706422018e-06,
      "loss": 0.4385,
      "step": 549
    },
    {
      "epoch": 0.02523977788995457,
      "grad_norm": 0.498604416847229,
      "learning_rate": 4.204892966360857e-06,
      "loss": 0.4402,
      "step": 550
    },
    {
      "epoch": 0.025285668395209033,
      "grad_norm": 0.49241143465042114,
      "learning_rate": 4.212538226299694e-06,
      "loss": 0.4474,
      "step": 551
    },
    {
      "epoch": 0.025331558900463493,
      "grad_norm": 0.4935033321380615,
      "learning_rate": 4.220183486238532e-06,
      "loss": 0.4455,
      "step": 552
    },
    {
      "epoch": 0.025377449405717956,
      "grad_norm": 0.5777140259742737,
      "learning_rate": 4.227828746177371e-06,
      "loss": 0.5521,
      "step": 553
    },
    {
      "epoch": 0.02542333991097242,
      "grad_norm": 0.4227786064147949,
      "learning_rate": 4.235474006116208e-06,
      "loss": 0.2976,
      "step": 554
    },
    {
      "epoch": 0.025469230416226882,
      "grad_norm": 0.41813260316848755,
      "learning_rate": 4.2431192660550464e-06,
      "loss": 0.2944,
      "step": 555
    },
    {
      "epoch": 0.025515120921481345,
      "grad_norm": 0.4449165165424347,
      "learning_rate": 4.250764525993884e-06,
      "loss": 0.3895,
      "step": 556
    },
    {
      "epoch": 0.025561011426735808,
      "grad_norm": 0.5028536915779114,
      "learning_rate": 4.258409785932722e-06,
      "loss": 0.4556,
      "step": 557
    },
    {
      "epoch": 0.02560690193199027,
      "grad_norm": 0.4760664999485016,
      "learning_rate": 4.26605504587156e-06,
      "loss": 0.433,
      "step": 558
    },
    {
      "epoch": 0.025652792437244734,
      "grad_norm": 0.4297028183937073,
      "learning_rate": 4.273700305810398e-06,
      "loss": 0.3351,
      "step": 559
    },
    {
      "epoch": 0.025698682942499197,
      "grad_norm": 0.48132580518722534,
      "learning_rate": 4.281345565749236e-06,
      "loss": 0.3972,
      "step": 560
    },
    {
      "epoch": 0.02574457344775366,
      "grad_norm": 0.46537891030311584,
      "learning_rate": 4.288990825688074e-06,
      "loss": 0.3974,
      "step": 561
    },
    {
      "epoch": 0.025790463953008123,
      "grad_norm": 0.43812325596809387,
      "learning_rate": 4.296636085626912e-06,
      "loss": 0.3828,
      "step": 562
    },
    {
      "epoch": 0.025836354458262586,
      "grad_norm": 0.4738851487636566,
      "learning_rate": 4.304281345565749e-06,
      "loss": 0.3836,
      "step": 563
    },
    {
      "epoch": 0.02588224496351705,
      "grad_norm": 0.5260022878646851,
      "learning_rate": 4.311926605504588e-06,
      "loss": 0.5371,
      "step": 564
    },
    {
      "epoch": 0.025928135468771513,
      "grad_norm": 0.4480191767215729,
      "learning_rate": 4.319571865443425e-06,
      "loss": 0.3949,
      "step": 565
    },
    {
      "epoch": 0.025974025974025976,
      "grad_norm": 0.5282284617424011,
      "learning_rate": 4.3272171253822634e-06,
      "loss": 0.4548,
      "step": 566
    },
    {
      "epoch": 0.026019916479280435,
      "grad_norm": 0.44939708709716797,
      "learning_rate": 4.334862385321102e-06,
      "loss": 0.3321,
      "step": 567
    },
    {
      "epoch": 0.0260658069845349,
      "grad_norm": 0.4996342062950134,
      "learning_rate": 4.342507645259939e-06,
      "loss": 0.4423,
      "step": 568
    },
    {
      "epoch": 0.02611169748978936,
      "grad_norm": 0.5004560947418213,
      "learning_rate": 4.3501529051987775e-06,
      "loss": 0.4239,
      "step": 569
    },
    {
      "epoch": 0.026157587995043825,
      "grad_norm": 0.5284087061882019,
      "learning_rate": 4.357798165137615e-06,
      "loss": 0.4846,
      "step": 570
    },
    {
      "epoch": 0.026203478500298288,
      "grad_norm": 0.4797920882701874,
      "learning_rate": 4.365443425076452e-06,
      "loss": 0.4571,
      "step": 571
    },
    {
      "epoch": 0.02624936900555275,
      "grad_norm": 0.4827975630760193,
      "learning_rate": 4.373088685015291e-06,
      "loss": 0.3913,
      "step": 572
    },
    {
      "epoch": 0.026295259510807214,
      "grad_norm": 0.4639168679714203,
      "learning_rate": 4.380733944954129e-06,
      "loss": 0.4207,
      "step": 573
    },
    {
      "epoch": 0.026341150016061677,
      "grad_norm": 0.44913971424102783,
      "learning_rate": 4.388379204892967e-06,
      "loss": 0.3854,
      "step": 574
    },
    {
      "epoch": 0.02638704052131614,
      "grad_norm": 0.44848018884658813,
      "learning_rate": 4.396024464831805e-06,
      "loss": 0.3791,
      "step": 575
    },
    {
      "epoch": 0.026432931026570603,
      "grad_norm": 0.4293183386325836,
      "learning_rate": 4.403669724770643e-06,
      "loss": 0.3616,
      "step": 576
    },
    {
      "epoch": 0.026478821531825066,
      "grad_norm": 0.47283944487571716,
      "learning_rate": 4.4113149847094805e-06,
      "loss": 0.4652,
      "step": 577
    },
    {
      "epoch": 0.02652471203707953,
      "grad_norm": 0.5692253112792969,
      "learning_rate": 4.418960244648318e-06,
      "loss": 0.5863,
      "step": 578
    },
    {
      "epoch": 0.026570602542333992,
      "grad_norm": 0.504953145980835,
      "learning_rate": 4.426605504587156e-06,
      "loss": 0.5659,
      "step": 579
    },
    {
      "epoch": 0.026616493047588455,
      "grad_norm": 0.510410726070404,
      "learning_rate": 4.4342507645259945e-06,
      "loss": 0.4835,
      "step": 580
    },
    {
      "epoch": 0.026662383552842915,
      "grad_norm": 0.5205755829811096,
      "learning_rate": 4.441896024464832e-06,
      "loss": 0.5909,
      "step": 581
    },
    {
      "epoch": 0.026708274058097378,
      "grad_norm": 0.4615248441696167,
      "learning_rate": 4.44954128440367e-06,
      "loss": 0.4298,
      "step": 582
    },
    {
      "epoch": 0.02675416456335184,
      "grad_norm": 0.48772287368774414,
      "learning_rate": 4.457186544342508e-06,
      "loss": 0.4506,
      "step": 583
    },
    {
      "epoch": 0.026800055068606304,
      "grad_norm": 0.4559618830680847,
      "learning_rate": 4.464831804281346e-06,
      "loss": 0.4039,
      "step": 584
    },
    {
      "epoch": 0.026845945573860767,
      "grad_norm": 0.4673352539539337,
      "learning_rate": 4.4724770642201834e-06,
      "loss": 0.4271,
      "step": 585
    },
    {
      "epoch": 0.02689183607911523,
      "grad_norm": 0.46606773138046265,
      "learning_rate": 4.480122324159022e-06,
      "loss": 0.4269,
      "step": 586
    },
    {
      "epoch": 0.026937726584369694,
      "grad_norm": 0.4551379382610321,
      "learning_rate": 4.48776758409786e-06,
      "loss": 0.3897,
      "step": 587
    },
    {
      "epoch": 0.026983617089624157,
      "grad_norm": 0.4167824983596802,
      "learning_rate": 4.4954128440366975e-06,
      "loss": 0.3339,
      "step": 588
    },
    {
      "epoch": 0.02702950759487862,
      "grad_norm": 0.4326254427433014,
      "learning_rate": 4.503058103975536e-06,
      "loss": 0.335,
      "step": 589
    },
    {
      "epoch": 0.027075398100133083,
      "grad_norm": 0.4331264793872833,
      "learning_rate": 4.510703363914373e-06,
      "loss": 0.3609,
      "step": 590
    },
    {
      "epoch": 0.027121288605387546,
      "grad_norm": 0.5396191477775574,
      "learning_rate": 4.5183486238532115e-06,
      "loss": 0.4927,
      "step": 591
    },
    {
      "epoch": 0.02716717911064201,
      "grad_norm": 0.4724072217941284,
      "learning_rate": 4.525993883792049e-06,
      "loss": 0.4395,
      "step": 592
    },
    {
      "epoch": 0.027213069615896472,
      "grad_norm": 0.43360739946365356,
      "learning_rate": 4.533639143730887e-06,
      "loss": 0.3339,
      "step": 593
    },
    {
      "epoch": 0.027258960121150935,
      "grad_norm": 0.5597114562988281,
      "learning_rate": 4.541284403669725e-06,
      "loss": 0.5304,
      "step": 594
    },
    {
      "epoch": 0.0273048506264054,
      "grad_norm": 0.45619672536849976,
      "learning_rate": 4.548929663608563e-06,
      "loss": 0.3814,
      "step": 595
    },
    {
      "epoch": 0.027350741131659858,
      "grad_norm": 0.4712444543838501,
      "learning_rate": 4.556574923547401e-06,
      "loss": 0.408,
      "step": 596
    },
    {
      "epoch": 0.02739663163691432,
      "grad_norm": 0.4540334641933441,
      "learning_rate": 4.564220183486239e-06,
      "loss": 0.3752,
      "step": 597
    },
    {
      "epoch": 0.027442522142168784,
      "grad_norm": 0.4880044460296631,
      "learning_rate": 4.571865443425077e-06,
      "loss": 0.4574,
      "step": 598
    },
    {
      "epoch": 0.027488412647423247,
      "grad_norm": 0.444128155708313,
      "learning_rate": 4.5795107033639145e-06,
      "loss": 0.3243,
      "step": 599
    },
    {
      "epoch": 0.02753430315267771,
      "grad_norm": 0.43998053669929504,
      "learning_rate": 4.587155963302753e-06,
      "loss": 0.3342,
      "step": 600
    },
    {
      "epoch": 0.027580193657932173,
      "grad_norm": 0.5074039101600647,
      "learning_rate": 4.59480122324159e-06,
      "loss": 0.4704,
      "step": 601
    },
    {
      "epoch": 0.027626084163186636,
      "grad_norm": 0.5305527448654175,
      "learning_rate": 4.6024464831804285e-06,
      "loss": 0.4954,
      "step": 602
    },
    {
      "epoch": 0.0276719746684411,
      "grad_norm": 0.5054502487182617,
      "learning_rate": 4.610091743119267e-06,
      "loss": 0.422,
      "step": 603
    },
    {
      "epoch": 0.027717865173695563,
      "grad_norm": 0.5841851234436035,
      "learning_rate": 4.617737003058104e-06,
      "loss": 0.4957,
      "step": 604
    },
    {
      "epoch": 0.027763755678950026,
      "grad_norm": 0.4930340349674225,
      "learning_rate": 4.6253822629969426e-06,
      "loss": 0.3941,
      "step": 605
    },
    {
      "epoch": 0.02780964618420449,
      "grad_norm": 0.5147585868835449,
      "learning_rate": 4.63302752293578e-06,
      "loss": 0.4628,
      "step": 606
    },
    {
      "epoch": 0.027855536689458952,
      "grad_norm": 0.5381789207458496,
      "learning_rate": 4.6406727828746175e-06,
      "loss": 0.4589,
      "step": 607
    },
    {
      "epoch": 0.027901427194713415,
      "grad_norm": 0.4610876739025116,
      "learning_rate": 4.648318042813456e-06,
      "loss": 0.317,
      "step": 608
    },
    {
      "epoch": 0.027947317699967878,
      "grad_norm": 0.5606909990310669,
      "learning_rate": 4.655963302752294e-06,
      "loss": 0.4998,
      "step": 609
    },
    {
      "epoch": 0.02799320820522234,
      "grad_norm": 0.4477992057800293,
      "learning_rate": 4.663608562691132e-06,
      "loss": 0.3569,
      "step": 610
    },
    {
      "epoch": 0.0280390987104768,
      "grad_norm": 0.4859997034072876,
      "learning_rate": 4.67125382262997e-06,
      "loss": 0.4474,
      "step": 611
    },
    {
      "epoch": 0.028084989215731264,
      "grad_norm": 0.5378237962722778,
      "learning_rate": 4.678899082568808e-06,
      "loss": 0.5315,
      "step": 612
    },
    {
      "epoch": 0.028130879720985727,
      "grad_norm": 0.48265835642814636,
      "learning_rate": 4.6865443425076455e-06,
      "loss": 0.4979,
      "step": 613
    },
    {
      "epoch": 0.02817677022624019,
      "grad_norm": 0.4947357773780823,
      "learning_rate": 4.694189602446483e-06,
      "loss": 0.5025,
      "step": 614
    },
    {
      "epoch": 0.028222660731494653,
      "grad_norm": 0.5121883153915405,
      "learning_rate": 4.701834862385321e-06,
      "loss": 0.5109,
      "step": 615
    },
    {
      "epoch": 0.028268551236749116,
      "grad_norm": 0.5024587512016296,
      "learning_rate": 4.70948012232416e-06,
      "loss": 0.444,
      "step": 616
    },
    {
      "epoch": 0.02831444174200358,
      "grad_norm": 0.48505645990371704,
      "learning_rate": 4.717125382262998e-06,
      "loss": 0.4286,
      "step": 617
    },
    {
      "epoch": 0.028360332247258042,
      "grad_norm": 0.5032019019126892,
      "learning_rate": 4.724770642201835e-06,
      "loss": 0.5055,
      "step": 618
    },
    {
      "epoch": 0.028406222752512506,
      "grad_norm": 0.5626063942909241,
      "learning_rate": 4.732415902140673e-06,
      "loss": 0.534,
      "step": 619
    },
    {
      "epoch": 0.02845211325776697,
      "grad_norm": 0.47117555141448975,
      "learning_rate": 4.740061162079511e-06,
      "loss": 0.3794,
      "step": 620
    },
    {
      "epoch": 0.02849800376302143,
      "grad_norm": 0.5280769467353821,
      "learning_rate": 4.7477064220183485e-06,
      "loss": 0.5664,
      "step": 621
    },
    {
      "epoch": 0.028543894268275895,
      "grad_norm": 0.47849786281585693,
      "learning_rate": 4.755351681957187e-06,
      "loss": 0.3908,
      "step": 622
    },
    {
      "epoch": 0.028589784773530358,
      "grad_norm": 0.4524434506893158,
      "learning_rate": 4.762996941896025e-06,
      "loss": 0.3959,
      "step": 623
    },
    {
      "epoch": 0.02863567527878482,
      "grad_norm": 0.5147923231124878,
      "learning_rate": 4.770642201834863e-06,
      "loss": 0.4882,
      "step": 624
    },
    {
      "epoch": 0.02868156578403928,
      "grad_norm": 0.47584009170532227,
      "learning_rate": 4.778287461773701e-06,
      "loss": 0.4213,
      "step": 625
    },
    {
      "epoch": 0.028727456289293744,
      "grad_norm": 0.47318753600120544,
      "learning_rate": 4.785932721712538e-06,
      "loss": 0.4043,
      "step": 626
    },
    {
      "epoch": 0.028773346794548207,
      "grad_norm": 0.442760169506073,
      "learning_rate": 4.793577981651377e-06,
      "loss": 0.3256,
      "step": 627
    },
    {
      "epoch": 0.02881923729980267,
      "grad_norm": 0.4752501845359802,
      "learning_rate": 4.801223241590214e-06,
      "loss": 0.3863,
      "step": 628
    },
    {
      "epoch": 0.028865127805057133,
      "grad_norm": 0.4469601809978485,
      "learning_rate": 4.808868501529052e-06,
      "loss": 0.3454,
      "step": 629
    },
    {
      "epoch": 0.028911018310311596,
      "grad_norm": 0.5374414920806885,
      "learning_rate": 4.816513761467891e-06,
      "loss": 0.5175,
      "step": 630
    },
    {
      "epoch": 0.02895690881556606,
      "grad_norm": 0.44401299953460693,
      "learning_rate": 4.824159021406728e-06,
      "loss": 0.375,
      "step": 631
    },
    {
      "epoch": 0.029002799320820522,
      "grad_norm": 0.5041106939315796,
      "learning_rate": 4.831804281345566e-06,
      "loss": 0.4823,
      "step": 632
    },
    {
      "epoch": 0.029048689826074985,
      "grad_norm": 0.5558406114578247,
      "learning_rate": 4.839449541284404e-06,
      "loss": 0.6037,
      "step": 633
    },
    {
      "epoch": 0.02909458033132945,
      "grad_norm": 0.5479883551597595,
      "learning_rate": 4.847094801223242e-06,
      "loss": 0.4702,
      "step": 634
    },
    {
      "epoch": 0.02914047083658391,
      "grad_norm": 0.47405314445495605,
      "learning_rate": 4.8547400611620796e-06,
      "loss": 0.3948,
      "step": 635
    },
    {
      "epoch": 0.029186361341838375,
      "grad_norm": 0.4791420102119446,
      "learning_rate": 4.862385321100918e-06,
      "loss": 0.3764,
      "step": 636
    },
    {
      "epoch": 0.029232251847092838,
      "grad_norm": 0.47440847754478455,
      "learning_rate": 4.870030581039756e-06,
      "loss": 0.3936,
      "step": 637
    },
    {
      "epoch": 0.0292781423523473,
      "grad_norm": 0.47720280289649963,
      "learning_rate": 4.877675840978594e-06,
      "loss": 0.4285,
      "step": 638
    },
    {
      "epoch": 0.029324032857601764,
      "grad_norm": 0.48758217692375183,
      "learning_rate": 4.885321100917432e-06,
      "loss": 0.4051,
      "step": 639
    },
    {
      "epoch": 0.029369923362856223,
      "grad_norm": 0.5307623744010925,
      "learning_rate": 4.892966360856269e-06,
      "loss": 0.4783,
      "step": 640
    },
    {
      "epoch": 0.029415813868110686,
      "grad_norm": 0.47263261675834656,
      "learning_rate": 4.900611620795108e-06,
      "loss": 0.4024,
      "step": 641
    },
    {
      "epoch": 0.02946170437336515,
      "grad_norm": 0.49153900146484375,
      "learning_rate": 4.908256880733945e-06,
      "loss": 0.4511,
      "step": 642
    },
    {
      "epoch": 0.029507594878619613,
      "grad_norm": 0.4701794385910034,
      "learning_rate": 4.915902140672783e-06,
      "loss": 0.3819,
      "step": 643
    },
    {
      "epoch": 0.029553485383874076,
      "grad_norm": 0.47918131947517395,
      "learning_rate": 4.923547400611622e-06,
      "loss": 0.3981,
      "step": 644
    },
    {
      "epoch": 0.02959937588912854,
      "grad_norm": 0.51798015832901,
      "learning_rate": 4.931192660550459e-06,
      "loss": 0.4983,
      "step": 645
    },
    {
      "epoch": 0.029645266394383002,
      "grad_norm": 0.4869624078273773,
      "learning_rate": 4.9388379204892974e-06,
      "loss": 0.4772,
      "step": 646
    },
    {
      "epoch": 0.029691156899637465,
      "grad_norm": 0.4917526841163635,
      "learning_rate": 4.946483180428135e-06,
      "loss": 0.427,
      "step": 647
    },
    {
      "epoch": 0.029737047404891928,
      "grad_norm": 0.5127080678939819,
      "learning_rate": 4.954128440366973e-06,
      "loss": 0.434,
      "step": 648
    },
    {
      "epoch": 0.02978293791014639,
      "grad_norm": 0.4668504297733307,
      "learning_rate": 4.961773700305811e-06,
      "loss": 0.4045,
      "step": 649
    },
    {
      "epoch": 0.029828828415400854,
      "grad_norm": 0.44504514336586,
      "learning_rate": 4.969418960244649e-06,
      "loss": 0.3242,
      "step": 650
    },
    {
      "epoch": 0.029874718920655317,
      "grad_norm": 0.4559177756309509,
      "learning_rate": 4.977064220183486e-06,
      "loss": 0.3539,
      "step": 651
    },
    {
      "epoch": 0.02992060942590978,
      "grad_norm": 0.4884553551673889,
      "learning_rate": 4.984709480122325e-06,
      "loss": 0.4261,
      "step": 652
    },
    {
      "epoch": 0.029966499931164244,
      "grad_norm": 0.5162047743797302,
      "learning_rate": 4.992354740061163e-06,
      "loss": 0.4672,
      "step": 653
    },
    {
      "epoch": 0.030012390436418707,
      "grad_norm": 0.46461427211761475,
      "learning_rate": 5e-06,
      "loss": 0.3678,
      "step": 654
    },
    {
      "epoch": 0.030058280941673166,
      "grad_norm": 0.44757315516471863,
      "learning_rate": 5.007645259938838e-06,
      "loss": 0.3639,
      "step": 655
    },
    {
      "epoch": 0.03010417144692763,
      "grad_norm": 0.45951172709465027,
      "learning_rate": 5.015290519877676e-06,
      "loss": 0.3985,
      "step": 656
    },
    {
      "epoch": 0.030150061952182092,
      "grad_norm": 0.47510138154029846,
      "learning_rate": 5.0229357798165144e-06,
      "loss": 0.3896,
      "step": 657
    },
    {
      "epoch": 0.030195952457436556,
      "grad_norm": 0.5010815858840942,
      "learning_rate": 5.030581039755352e-06,
      "loss": 0.4675,
      "step": 658
    },
    {
      "epoch": 0.03024184296269102,
      "grad_norm": 0.49433425068855286,
      "learning_rate": 5.03822629969419e-06,
      "loss": 0.3865,
      "step": 659
    },
    {
      "epoch": 0.03028773346794548,
      "grad_norm": 0.4810725450515747,
      "learning_rate": 5.045871559633028e-06,
      "loss": 0.3971,
      "step": 660
    },
    {
      "epoch": 0.030333623973199945,
      "grad_norm": 0.5787946581840515,
      "learning_rate": 5.053516819571865e-06,
      "loss": 0.5114,
      "step": 661
    },
    {
      "epoch": 0.030379514478454408,
      "grad_norm": 0.4574304521083832,
      "learning_rate": 5.061162079510704e-06,
      "loss": 0.3848,
      "step": 662
    },
    {
      "epoch": 0.03042540498370887,
      "grad_norm": 0.48120027780532837,
      "learning_rate": 5.068807339449542e-06,
      "loss": 0.4127,
      "step": 663
    },
    {
      "epoch": 0.030471295488963334,
      "grad_norm": 0.4440484046936035,
      "learning_rate": 5.076452599388379e-06,
      "loss": 0.3293,
      "step": 664
    },
    {
      "epoch": 0.030517185994217797,
      "grad_norm": 0.44411030411720276,
      "learning_rate": 5.084097859327217e-06,
      "loss": 0.306,
      "step": 665
    },
    {
      "epoch": 0.03056307649947226,
      "grad_norm": 0.5139580965042114,
      "learning_rate": 5.091743119266055e-06,
      "loss": 0.4569,
      "step": 666
    },
    {
      "epoch": 0.030608967004726723,
      "grad_norm": 0.5118184089660645,
      "learning_rate": 5.099388379204894e-06,
      "loss": 0.4527,
      "step": 667
    },
    {
      "epoch": 0.030654857509981186,
      "grad_norm": 0.504111647605896,
      "learning_rate": 5.1070336391437315e-06,
      "loss": 0.4701,
      "step": 668
    },
    {
      "epoch": 0.03070074801523565,
      "grad_norm": 0.45596060156822205,
      "learning_rate": 5.114678899082569e-06,
      "loss": 0.2945,
      "step": 669
    },
    {
      "epoch": 0.03074663852049011,
      "grad_norm": 0.5010163187980652,
      "learning_rate": 5.122324159021407e-06,
      "loss": 0.3896,
      "step": 670
    },
    {
      "epoch": 0.030792529025744572,
      "grad_norm": 0.47708919644355774,
      "learning_rate": 5.1299694189602455e-06,
      "loss": 0.3956,
      "step": 671
    },
    {
      "epoch": 0.030838419530999035,
      "grad_norm": 0.47731465101242065,
      "learning_rate": 5.137614678899083e-06,
      "loss": 0.4272,
      "step": 672
    },
    {
      "epoch": 0.0308843100362535,
      "grad_norm": 0.5047810673713684,
      "learning_rate": 5.145259938837921e-06,
      "loss": 0.4789,
      "step": 673
    },
    {
      "epoch": 0.03093020054150796,
      "grad_norm": 0.4855414032936096,
      "learning_rate": 5.152905198776759e-06,
      "loss": 0.402,
      "step": 674
    },
    {
      "epoch": 0.030976091046762425,
      "grad_norm": 0.48568686842918396,
      "learning_rate": 5.160550458715596e-06,
      "loss": 0.4531,
      "step": 675
    },
    {
      "epoch": 0.031021981552016888,
      "grad_norm": 0.4904922842979431,
      "learning_rate": 5.168195718654435e-06,
      "loss": 0.4363,
      "step": 676
    },
    {
      "epoch": 0.03106787205727135,
      "grad_norm": 0.45971164107322693,
      "learning_rate": 5.175840978593273e-06,
      "loss": 0.3517,
      "step": 677
    },
    {
      "epoch": 0.031113762562525814,
      "grad_norm": 0.4890196621417999,
      "learning_rate": 5.18348623853211e-06,
      "loss": 0.3709,
      "step": 678
    },
    {
      "epoch": 0.031159653067780277,
      "grad_norm": 0.4717274010181427,
      "learning_rate": 5.1911314984709485e-06,
      "loss": 0.4312,
      "step": 679
    },
    {
      "epoch": 0.03120554357303474,
      "grad_norm": 0.45426157116889954,
      "learning_rate": 5.198776758409786e-06,
      "loss": 0.4003,
      "step": 680
    },
    {
      "epoch": 0.0312514340782892,
      "grad_norm": 0.4568299949169159,
      "learning_rate": 5.206422018348625e-06,
      "loss": 0.398,
      "step": 681
    },
    {
      "epoch": 0.031297324583543666,
      "grad_norm": 0.4921552538871765,
      "learning_rate": 5.2140672782874625e-06,
      "loss": 0.4375,
      "step": 682
    },
    {
      "epoch": 0.03134321508879813,
      "grad_norm": 0.511883556842804,
      "learning_rate": 5.2217125382263e-06,
      "loss": 0.4425,
      "step": 683
    },
    {
      "epoch": 0.03138910559405259,
      "grad_norm": 0.47729989886283875,
      "learning_rate": 5.229357798165137e-06,
      "loss": 0.4291,
      "step": 684
    },
    {
      "epoch": 0.031434996099307055,
      "grad_norm": 0.5338758826255798,
      "learning_rate": 5.237003058103976e-06,
      "loss": 0.4765,
      "step": 685
    },
    {
      "epoch": 0.03148088660456152,
      "grad_norm": 0.5013654828071594,
      "learning_rate": 5.244648318042814e-06,
      "loss": 0.4556,
      "step": 686
    },
    {
      "epoch": 0.03152677710981598,
      "grad_norm": 0.48483702540397644,
      "learning_rate": 5.252293577981652e-06,
      "loss": 0.389,
      "step": 687
    },
    {
      "epoch": 0.031572667615070445,
      "grad_norm": 0.4565930962562561,
      "learning_rate": 5.25993883792049e-06,
      "loss": 0.3939,
      "step": 688
    },
    {
      "epoch": 0.03161855812032491,
      "grad_norm": 0.4636811912059784,
      "learning_rate": 5.267584097859327e-06,
      "loss": 0.3682,
      "step": 689
    },
    {
      "epoch": 0.03166444862557937,
      "grad_norm": 0.44330793619155884,
      "learning_rate": 5.275229357798165e-06,
      "loss": 0.3359,
      "step": 690
    },
    {
      "epoch": 0.031710339130833834,
      "grad_norm": 0.47419053316116333,
      "learning_rate": 5.282874617737004e-06,
      "loss": 0.4596,
      "step": 691
    },
    {
      "epoch": 0.03175622963608829,
      "grad_norm": 0.41946354508399963,
      "learning_rate": 5.290519877675841e-06,
      "loss": 0.3305,
      "step": 692
    },
    {
      "epoch": 0.03180212014134275,
      "grad_norm": 0.510080099105835,
      "learning_rate": 5.2981651376146795e-06,
      "loss": 0.4226,
      "step": 693
    },
    {
      "epoch": 0.031848010646597216,
      "grad_norm": 0.4508179724216461,
      "learning_rate": 5.305810397553517e-06,
      "loss": 0.4033,
      "step": 694
    },
    {
      "epoch": 0.03189390115185168,
      "grad_norm": 0.47535669803619385,
      "learning_rate": 5.313455657492356e-06,
      "loss": 0.4062,
      "step": 695
    },
    {
      "epoch": 0.03193979165710614,
      "grad_norm": 0.4734593629837036,
      "learning_rate": 5.3211009174311936e-06,
      "loss": 0.3946,
      "step": 696
    },
    {
      "epoch": 0.031985682162360606,
      "grad_norm": 0.41853395104408264,
      "learning_rate": 5.328746177370031e-06,
      "loss": 0.2861,
      "step": 697
    },
    {
      "epoch": 0.03203157266761507,
      "grad_norm": 0.5116884708404541,
      "learning_rate": 5.3363914373088685e-06,
      "loss": 0.4876,
      "step": 698
    },
    {
      "epoch": 0.03207746317286953,
      "grad_norm": 0.43127205967903137,
      "learning_rate": 5.344036697247707e-06,
      "loss": 0.3202,
      "step": 699
    },
    {
      "epoch": 0.032123353678123995,
      "grad_norm": 0.45206716656684875,
      "learning_rate": 5.351681957186545e-06,
      "loss": 0.3521,
      "step": 700
    },
    {
      "epoch": 0.03216924418337846,
      "grad_norm": 0.49567705392837524,
      "learning_rate": 5.359327217125383e-06,
      "loss": 0.4851,
      "step": 701
    },
    {
      "epoch": 0.03221513468863292,
      "grad_norm": 0.4594910740852356,
      "learning_rate": 5.366972477064221e-06,
      "loss": 0.3726,
      "step": 702
    },
    {
      "epoch": 0.032261025193887384,
      "grad_norm": 0.4929613769054413,
      "learning_rate": 5.374617737003058e-06,
      "loss": 0.402,
      "step": 703
    },
    {
      "epoch": 0.03230691569914185,
      "grad_norm": 0.5152514576911926,
      "learning_rate": 5.382262996941896e-06,
      "loss": 0.4337,
      "step": 704
    },
    {
      "epoch": 0.03235280620439631,
      "grad_norm": 0.4473576545715332,
      "learning_rate": 5.389908256880735e-06,
      "loss": 0.3202,
      "step": 705
    },
    {
      "epoch": 0.03239869670965077,
      "grad_norm": 0.4397395849227905,
      "learning_rate": 5.397553516819572e-06,
      "loss": 0.3618,
      "step": 706
    },
    {
      "epoch": 0.032444587214905236,
      "grad_norm": 0.4591197073459625,
      "learning_rate": 5.405198776758411e-06,
      "loss": 0.3665,
      "step": 707
    },
    {
      "epoch": 0.0324904777201597,
      "grad_norm": 0.4718942940235138,
      "learning_rate": 5.412844036697248e-06,
      "loss": 0.4327,
      "step": 708
    },
    {
      "epoch": 0.03253636822541416,
      "grad_norm": 0.4826303720474243,
      "learning_rate": 5.4204892966360855e-06,
      "loss": 0.4296,
      "step": 709
    },
    {
      "epoch": 0.032582258730668626,
      "grad_norm": 0.48603740334510803,
      "learning_rate": 5.428134556574925e-06,
      "loss": 0.4174,
      "step": 710
    },
    {
      "epoch": 0.03262814923592309,
      "grad_norm": 0.4769875109195709,
      "learning_rate": 5.435779816513762e-06,
      "loss": 0.4157,
      "step": 711
    },
    {
      "epoch": 0.03267403974117755,
      "grad_norm": 0.5343371629714966,
      "learning_rate": 5.4434250764525995e-06,
      "loss": 0.5321,
      "step": 712
    },
    {
      "epoch": 0.032719930246432015,
      "grad_norm": 0.5503456592559814,
      "learning_rate": 5.451070336391438e-06,
      "loss": 0.5236,
      "step": 713
    },
    {
      "epoch": 0.03276582075168648,
      "grad_norm": 0.4854121804237366,
      "learning_rate": 5.458715596330275e-06,
      "loss": 0.4325,
      "step": 714
    },
    {
      "epoch": 0.03281171125694094,
      "grad_norm": 0.43479588627815247,
      "learning_rate": 5.4663608562691136e-06,
      "loss": 0.3029,
      "step": 715
    },
    {
      "epoch": 0.032857601762195404,
      "grad_norm": 0.5901136994361877,
      "learning_rate": 5.474006116207952e-06,
      "loss": 0.5026,
      "step": 716
    },
    {
      "epoch": 0.03290349226744987,
      "grad_norm": 0.502284049987793,
      "learning_rate": 5.481651376146789e-06,
      "loss": 0.4487,
      "step": 717
    },
    {
      "epoch": 0.03294938277270433,
      "grad_norm": 0.5088910460472107,
      "learning_rate": 5.489296636085627e-06,
      "loss": 0.5032,
      "step": 718
    },
    {
      "epoch": 0.03299527327795879,
      "grad_norm": 0.4692699611186981,
      "learning_rate": 5.496941896024466e-06,
      "loss": 0.4325,
      "step": 719
    },
    {
      "epoch": 0.03304116378321326,
      "grad_norm": 0.4397878348827362,
      "learning_rate": 5.504587155963303e-06,
      "loss": 0.3201,
      "step": 720
    },
    {
      "epoch": 0.03308705428846771,
      "grad_norm": 0.4541064500808716,
      "learning_rate": 5.512232415902141e-06,
      "loss": 0.3928,
      "step": 721
    },
    {
      "epoch": 0.033132944793722176,
      "grad_norm": 0.4665283262729645,
      "learning_rate": 5.519877675840979e-06,
      "loss": 0.3678,
      "step": 722
    },
    {
      "epoch": 0.03317883529897664,
      "grad_norm": 0.5041980147361755,
      "learning_rate": 5.5275229357798165e-06,
      "loss": 0.5136,
      "step": 723
    },
    {
      "epoch": 0.0332247258042311,
      "grad_norm": 0.4926072061061859,
      "learning_rate": 5.535168195718656e-06,
      "loss": 0.417,
      "step": 724
    },
    {
      "epoch": 0.033270616309485565,
      "grad_norm": 0.4644889235496521,
      "learning_rate": 5.542813455657493e-06,
      "loss": 0.4224,
      "step": 725
    },
    {
      "epoch": 0.03331650681474003,
      "grad_norm": 0.5296972393989563,
      "learning_rate": 5.5504587155963306e-06,
      "loss": 0.4644,
      "step": 726
    },
    {
      "epoch": 0.03336239731999449,
      "grad_norm": 0.47378745675086975,
      "learning_rate": 5.558103975535169e-06,
      "loss": 0.3869,
      "step": 727
    },
    {
      "epoch": 0.033408287825248954,
      "grad_norm": 0.4840254485607147,
      "learning_rate": 5.565749235474006e-06,
      "loss": 0.4706,
      "step": 728
    },
    {
      "epoch": 0.03345417833050342,
      "grad_norm": 0.4293925166130066,
      "learning_rate": 5.573394495412845e-06,
      "loss": 0.3216,
      "step": 729
    },
    {
      "epoch": 0.03350006883575788,
      "grad_norm": 0.4357646405696869,
      "learning_rate": 5.581039755351683e-06,
      "loss": 0.3394,
      "step": 730
    },
    {
      "epoch": 0.033545959341012344,
      "grad_norm": 0.5204983353614807,
      "learning_rate": 5.58868501529052e-06,
      "loss": 0.4744,
      "step": 731
    },
    {
      "epoch": 0.03359184984626681,
      "grad_norm": 0.4658423066139221,
      "learning_rate": 5.596330275229358e-06,
      "loss": 0.39,
      "step": 732
    },
    {
      "epoch": 0.03363774035152127,
      "grad_norm": 0.47554606199264526,
      "learning_rate": 5.603975535168196e-06,
      "loss": 0.3871,
      "step": 733
    },
    {
      "epoch": 0.03368363085677573,
      "grad_norm": 0.4657820463180542,
      "learning_rate": 5.611620795107034e-06,
      "loss": 0.378,
      "step": 734
    },
    {
      "epoch": 0.033729521362030196,
      "grad_norm": 0.46907272934913635,
      "learning_rate": 5.619266055045872e-06,
      "loss": 0.4027,
      "step": 735
    },
    {
      "epoch": 0.03377541186728466,
      "grad_norm": 0.4753458797931671,
      "learning_rate": 5.62691131498471e-06,
      "loss": 0.466,
      "step": 736
    },
    {
      "epoch": 0.03382130237253912,
      "grad_norm": 0.5024524927139282,
      "learning_rate": 5.634556574923548e-06,
      "loss": 0.4678,
      "step": 737
    },
    {
      "epoch": 0.033867192877793585,
      "grad_norm": 0.40047597885131836,
      "learning_rate": 5.642201834862385e-06,
      "loss": 0.2765,
      "step": 738
    },
    {
      "epoch": 0.03391308338304805,
      "grad_norm": 0.4640674293041229,
      "learning_rate": 5.649847094801224e-06,
      "loss": 0.4219,
      "step": 739
    },
    {
      "epoch": 0.03395897388830251,
      "grad_norm": 0.5290677547454834,
      "learning_rate": 5.657492354740062e-06,
      "loss": 0.4627,
      "step": 740
    },
    {
      "epoch": 0.034004864393556974,
      "grad_norm": 0.4202752709388733,
      "learning_rate": 5.665137614678899e-06,
      "loss": 0.3275,
      "step": 741
    },
    {
      "epoch": 0.03405075489881144,
      "grad_norm": 0.5070269107818604,
      "learning_rate": 5.672782874617737e-06,
      "loss": 0.5058,
      "step": 742
    },
    {
      "epoch": 0.0340966454040659,
      "grad_norm": 0.544164776802063,
      "learning_rate": 5.680428134556576e-06,
      "loss": 0.5202,
      "step": 743
    },
    {
      "epoch": 0.034142535909320364,
      "grad_norm": 0.4559119641780853,
      "learning_rate": 5.688073394495414e-06,
      "loss": 0.39,
      "step": 744
    },
    {
      "epoch": 0.03418842641457483,
      "grad_norm": 0.5321412682533264,
      "learning_rate": 5.695718654434251e-06,
      "loss": 0.5447,
      "step": 745
    },
    {
      "epoch": 0.03423431691982929,
      "grad_norm": 0.47550147771835327,
      "learning_rate": 5.703363914373089e-06,
      "loss": 0.4263,
      "step": 746
    },
    {
      "epoch": 0.03428020742508375,
      "grad_norm": 0.4727621078491211,
      "learning_rate": 5.711009174311926e-06,
      "loss": 0.3601,
      "step": 747
    },
    {
      "epoch": 0.034326097930338216,
      "grad_norm": 0.4610954821109772,
      "learning_rate": 5.7186544342507654e-06,
      "loss": 0.3784,
      "step": 748
    },
    {
      "epoch": 0.03437198843559268,
      "grad_norm": 0.48836562037467957,
      "learning_rate": 5.726299694189603e-06,
      "loss": 0.4206,
      "step": 749
    },
    {
      "epoch": 0.034417878940847135,
      "grad_norm": 0.507480800151825,
      "learning_rate": 5.733944954128441e-06,
      "loss": 0.4779,
      "step": 750
    },
    {
      "epoch": 0.0344637694461016,
      "grad_norm": 0.552983820438385,
      "learning_rate": 5.741590214067279e-06,
      "loss": 0.5358,
      "step": 751
    },
    {
      "epoch": 0.03450965995135606,
      "grad_norm": 0.4471736252307892,
      "learning_rate": 5.749235474006116e-06,
      "loss": 0.3624,
      "step": 752
    },
    {
      "epoch": 0.034555550456610525,
      "grad_norm": 0.4889674484729767,
      "learning_rate": 5.756880733944955e-06,
      "loss": 0.4329,
      "step": 753
    },
    {
      "epoch": 0.03460144096186499,
      "grad_norm": 0.505480945110321,
      "learning_rate": 5.764525993883793e-06,
      "loss": 0.4043,
      "step": 754
    },
    {
      "epoch": 0.03464733146711945,
      "grad_norm": 0.4882463812828064,
      "learning_rate": 5.77217125382263e-06,
      "loss": 0.4151,
      "step": 755
    },
    {
      "epoch": 0.034693221972373914,
      "grad_norm": 0.4720045328140259,
      "learning_rate": 5.7798165137614684e-06,
      "loss": 0.3869,
      "step": 756
    },
    {
      "epoch": 0.03473911247762838,
      "grad_norm": 0.5206193327903748,
      "learning_rate": 5.787461773700306e-06,
      "loss": 0.5136,
      "step": 757
    },
    {
      "epoch": 0.03478500298288284,
      "grad_norm": 0.46404632925987244,
      "learning_rate": 5.795107033639145e-06,
      "loss": 0.3856,
      "step": 758
    },
    {
      "epoch": 0.0348308934881373,
      "grad_norm": 0.4719366431236267,
      "learning_rate": 5.8027522935779825e-06,
      "loss": 0.3643,
      "step": 759
    },
    {
      "epoch": 0.034876783993391766,
      "grad_norm": 0.4552661180496216,
      "learning_rate": 5.81039755351682e-06,
      "loss": 0.328,
      "step": 760
    },
    {
      "epoch": 0.03492267449864623,
      "grad_norm": 0.503520667552948,
      "learning_rate": 5.818042813455657e-06,
      "loss": 0.4676,
      "step": 761
    },
    {
      "epoch": 0.03496856500390069,
      "grad_norm": 0.490029513835907,
      "learning_rate": 5.825688073394496e-06,
      "loss": 0.4581,
      "step": 762
    },
    {
      "epoch": 0.035014455509155155,
      "grad_norm": 0.42828088998794556,
      "learning_rate": 5.833333333333334e-06,
      "loss": 0.3307,
      "step": 763
    },
    {
      "epoch": 0.03506034601440962,
      "grad_norm": 0.5580511689186096,
      "learning_rate": 5.840978593272172e-06,
      "loss": 0.3878,
      "step": 764
    },
    {
      "epoch": 0.03510623651966408,
      "grad_norm": 0.4630361497402191,
      "learning_rate": 5.84862385321101e-06,
      "loss": 0.4286,
      "step": 765
    },
    {
      "epoch": 0.035152127024918545,
      "grad_norm": 0.4680016040802002,
      "learning_rate": 5.856269113149847e-06,
      "loss": 0.4324,
      "step": 766
    },
    {
      "epoch": 0.03519801753017301,
      "grad_norm": 0.4838612675666809,
      "learning_rate": 5.863914373088685e-06,
      "loss": 0.491,
      "step": 767
    },
    {
      "epoch": 0.03524390803542747,
      "grad_norm": 0.47573021054267883,
      "learning_rate": 5.871559633027524e-06,
      "loss": 0.4217,
      "step": 768
    },
    {
      "epoch": 0.035289798540681934,
      "grad_norm": 0.6087118983268738,
      "learning_rate": 5.879204892966361e-06,
      "loss": 0.4436,
      "step": 769
    },
    {
      "epoch": 0.0353356890459364,
      "grad_norm": 0.4900757968425751,
      "learning_rate": 5.8868501529051995e-06,
      "loss": 0.406,
      "step": 770
    },
    {
      "epoch": 0.03538157955119086,
      "grad_norm": 0.44781285524368286,
      "learning_rate": 5.894495412844037e-06,
      "loss": 0.377,
      "step": 771
    },
    {
      "epoch": 0.03542747005644532,
      "grad_norm": 0.4651029407978058,
      "learning_rate": 5.902140672782875e-06,
      "loss": 0.3313,
      "step": 772
    },
    {
      "epoch": 0.035473360561699786,
      "grad_norm": 0.46212464570999146,
      "learning_rate": 5.9097859327217135e-06,
      "loss": 0.3985,
      "step": 773
    },
    {
      "epoch": 0.03551925106695425,
      "grad_norm": 0.49656569957733154,
      "learning_rate": 5.917431192660551e-06,
      "loss": 0.4624,
      "step": 774
    },
    {
      "epoch": 0.03556514157220871,
      "grad_norm": 0.514012336730957,
      "learning_rate": 5.925076452599388e-06,
      "loss": 0.5059,
      "step": 775
    },
    {
      "epoch": 0.035611032077463176,
      "grad_norm": 0.48806434869766235,
      "learning_rate": 5.932721712538227e-06,
      "loss": 0.4592,
      "step": 776
    },
    {
      "epoch": 0.03565692258271764,
      "grad_norm": 0.4882490038871765,
      "learning_rate": 5.940366972477065e-06,
      "loss": 0.4285,
      "step": 777
    },
    {
      "epoch": 0.0357028130879721,
      "grad_norm": 0.46129822731018066,
      "learning_rate": 5.9480122324159025e-06,
      "loss": 0.3848,
      "step": 778
    },
    {
      "epoch": 0.035748703593226565,
      "grad_norm": 0.4811941385269165,
      "learning_rate": 5.955657492354741e-06,
      "loss": 0.4294,
      "step": 779
    },
    {
      "epoch": 0.03579459409848102,
      "grad_norm": 0.47969749569892883,
      "learning_rate": 5.963302752293578e-06,
      "loss": 0.4201,
      "step": 780
    },
    {
      "epoch": 0.035840484603735484,
      "grad_norm": 0.5206419229507446,
      "learning_rate": 5.970948012232416e-06,
      "loss": 0.4679,
      "step": 781
    },
    {
      "epoch": 0.03588637510898995,
      "grad_norm": 0.5180150270462036,
      "learning_rate": 5.978593272171255e-06,
      "loss": 0.5091,
      "step": 782
    },
    {
      "epoch": 0.03593226561424441,
      "grad_norm": 0.4841292202472687,
      "learning_rate": 5.986238532110092e-06,
      "loss": 0.4233,
      "step": 783
    },
    {
      "epoch": 0.03597815611949887,
      "grad_norm": 0.43384209275245667,
      "learning_rate": 5.9938837920489305e-06,
      "loss": 0.3337,
      "step": 784
    },
    {
      "epoch": 0.036024046624753336,
      "grad_norm": 0.45960533618927,
      "learning_rate": 6.001529051987768e-06,
      "loss": 0.3717,
      "step": 785
    },
    {
      "epoch": 0.0360699371300078,
      "grad_norm": 0.4859628677368164,
      "learning_rate": 6.0091743119266054e-06,
      "loss": 0.45,
      "step": 786
    },
    {
      "epoch": 0.03611582763526226,
      "grad_norm": 0.41695070266723633,
      "learning_rate": 6.0168195718654446e-06,
      "loss": 0.3128,
      "step": 787
    },
    {
      "epoch": 0.036161718140516726,
      "grad_norm": 0.4612447917461395,
      "learning_rate": 6.024464831804282e-06,
      "loss": 0.3968,
      "step": 788
    },
    {
      "epoch": 0.03620760864577119,
      "grad_norm": 0.5911262035369873,
      "learning_rate": 6.0321100917431195e-06,
      "loss": 0.5441,
      "step": 789
    },
    {
      "epoch": 0.03625349915102565,
      "grad_norm": 0.43984946608543396,
      "learning_rate": 6.039755351681958e-06,
      "loss": 0.3262,
      "step": 790
    },
    {
      "epoch": 0.036299389656280115,
      "grad_norm": 0.4176672101020813,
      "learning_rate": 6.047400611620795e-06,
      "loss": 0.2796,
      "step": 791
    },
    {
      "epoch": 0.03634528016153458,
      "grad_norm": 0.49719345569610596,
      "learning_rate": 6.0550458715596335e-06,
      "loss": 0.4723,
      "step": 792
    },
    {
      "epoch": 0.03639117066678904,
      "grad_norm": 0.4714670479297638,
      "learning_rate": 6.062691131498472e-06,
      "loss": 0.4089,
      "step": 793
    },
    {
      "epoch": 0.036437061172043504,
      "grad_norm": 0.5643766522407532,
      "learning_rate": 6.070336391437309e-06,
      "loss": 0.5232,
      "step": 794
    },
    {
      "epoch": 0.03648295167729797,
      "grad_norm": 0.5200351476669312,
      "learning_rate": 6.077981651376147e-06,
      "loss": 0.432,
      "step": 795
    },
    {
      "epoch": 0.03652884218255243,
      "grad_norm": 0.48379260301589966,
      "learning_rate": 6.085626911314986e-06,
      "loss": 0.4007,
      "step": 796
    },
    {
      "epoch": 0.03657473268780689,
      "grad_norm": 0.450309693813324,
      "learning_rate": 6.093272171253823e-06,
      "loss": 0.3764,
      "step": 797
    },
    {
      "epoch": 0.03662062319306136,
      "grad_norm": 0.4852868318557739,
      "learning_rate": 6.100917431192661e-06,
      "loss": 0.4812,
      "step": 798
    },
    {
      "epoch": 0.03666651369831582,
      "grad_norm": 0.48230257630348206,
      "learning_rate": 6.108562691131499e-06,
      "loss": 0.4429,
      "step": 799
    },
    {
      "epoch": 0.03671240420357028,
      "grad_norm": 0.4992111623287201,
      "learning_rate": 6.1162079510703365e-06,
      "loss": 0.4462,
      "step": 800
    },
    {
      "epoch": 0.036758294708824746,
      "grad_norm": 0.45000842213630676,
      "learning_rate": 6.123853211009176e-06,
      "loss": 0.3441,
      "step": 801
    },
    {
      "epoch": 0.03680418521407921,
      "grad_norm": 0.5243483185768127,
      "learning_rate": 6.131498470948013e-06,
      "loss": 0.4953,
      "step": 802
    },
    {
      "epoch": 0.03685007571933367,
      "grad_norm": 0.5090179443359375,
      "learning_rate": 6.1391437308868505e-06,
      "loss": 0.4669,
      "step": 803
    },
    {
      "epoch": 0.036895966224588135,
      "grad_norm": 0.465063214302063,
      "learning_rate": 6.146788990825688e-06,
      "loss": 0.3666,
      "step": 804
    },
    {
      "epoch": 0.0369418567298426,
      "grad_norm": 0.5193577408790588,
      "learning_rate": 6.154434250764526e-06,
      "loss": 0.4688,
      "step": 805
    },
    {
      "epoch": 0.03698774723509706,
      "grad_norm": 0.5085948705673218,
      "learning_rate": 6.1620795107033646e-06,
      "loss": 0.4257,
      "step": 806
    },
    {
      "epoch": 0.037033637740351524,
      "grad_norm": 0.44270625710487366,
      "learning_rate": 6.169724770642203e-06,
      "loss": 0.3476,
      "step": 807
    },
    {
      "epoch": 0.03707952824560599,
      "grad_norm": 0.4461106061935425,
      "learning_rate": 6.17737003058104e-06,
      "loss": 0.375,
      "step": 808
    },
    {
      "epoch": 0.037125418750860444,
      "grad_norm": 0.4571780264377594,
      "learning_rate": 6.185015290519878e-06,
      "loss": 0.3259,
      "step": 809
    },
    {
      "epoch": 0.03717130925611491,
      "grad_norm": 0.44006818532943726,
      "learning_rate": 6.192660550458715e-06,
      "loss": 0.3335,
      "step": 810
    },
    {
      "epoch": 0.03721719976136937,
      "grad_norm": 0.4510287046432495,
      "learning_rate": 6.200305810397554e-06,
      "loss": 0.3969,
      "step": 811
    },
    {
      "epoch": 0.03726309026662383,
      "grad_norm": 0.4678497612476349,
      "learning_rate": 6.207951070336392e-06,
      "loss": 0.4056,
      "step": 812
    },
    {
      "epoch": 0.037308980771878296,
      "grad_norm": 0.5567219257354736,
      "learning_rate": 6.21559633027523e-06,
      "loss": 0.4224,
      "step": 813
    },
    {
      "epoch": 0.03735487127713276,
      "grad_norm": 0.48783227801322937,
      "learning_rate": 6.2232415902140675e-06,
      "loss": 0.4979,
      "step": 814
    },
    {
      "epoch": 0.03740076178238722,
      "grad_norm": 0.4996022582054138,
      "learning_rate": 6.230886850152905e-06,
      "loss": 0.4312,
      "step": 815
    },
    {
      "epoch": 0.037446652287641685,
      "grad_norm": 0.648202657699585,
      "learning_rate": 6.238532110091744e-06,
      "loss": 0.6078,
      "step": 816
    },
    {
      "epoch": 0.03749254279289615,
      "grad_norm": 0.49291545152664185,
      "learning_rate": 6.2461773700305816e-06,
      "loss": 0.4469,
      "step": 817
    },
    {
      "epoch": 0.03753843329815061,
      "grad_norm": 0.45443353056907654,
      "learning_rate": 6.253822629969419e-06,
      "loss": 0.3117,
      "step": 818
    },
    {
      "epoch": 0.037584323803405074,
      "grad_norm": 0.4558636546134949,
      "learning_rate": 6.261467889908257e-06,
      "loss": 0.3608,
      "step": 819
    },
    {
      "epoch": 0.03763021430865954,
      "grad_norm": 0.4326160252094269,
      "learning_rate": 6.269113149847096e-06,
      "loss": 0.3597,
      "step": 820
    },
    {
      "epoch": 0.037676104813914,
      "grad_norm": 0.514356255531311,
      "learning_rate": 6.276758409785934e-06,
      "loss": 0.4521,
      "step": 821
    },
    {
      "epoch": 0.037721995319168464,
      "grad_norm": 0.5081436038017273,
      "learning_rate": 6.284403669724771e-06,
      "loss": 0.4222,
      "step": 822
    },
    {
      "epoch": 0.03776788582442293,
      "grad_norm": 0.474962055683136,
      "learning_rate": 6.292048929663609e-06,
      "loss": 0.386,
      "step": 823
    },
    {
      "epoch": 0.03781377632967739,
      "grad_norm": 0.5017170310020447,
      "learning_rate": 6.299694189602446e-06,
      "loss": 0.4958,
      "step": 824
    },
    {
      "epoch": 0.03785966683493185,
      "grad_norm": 0.4880979359149933,
      "learning_rate": 6.307339449541285e-06,
      "loss": 0.4353,
      "step": 825
    },
    {
      "epoch": 0.037905557340186316,
      "grad_norm": 0.45421984791755676,
      "learning_rate": 6.314984709480123e-06,
      "loss": 0.3656,
      "step": 826
    },
    {
      "epoch": 0.03795144784544078,
      "grad_norm": 0.5236946940422058,
      "learning_rate": 6.322629969418961e-06,
      "loss": 0.4898,
      "step": 827
    },
    {
      "epoch": 0.03799733835069524,
      "grad_norm": 0.4378328323364258,
      "learning_rate": 6.330275229357799e-06,
      "loss": 0.3456,
      "step": 828
    },
    {
      "epoch": 0.038043228855949705,
      "grad_norm": 0.4880081117153168,
      "learning_rate": 6.337920489296636e-06,
      "loss": 0.4003,
      "step": 829
    },
    {
      "epoch": 0.03808911936120417,
      "grad_norm": 0.5049906969070435,
      "learning_rate": 6.345565749235475e-06,
      "loss": 0.4344,
      "step": 830
    },
    {
      "epoch": 0.03813500986645863,
      "grad_norm": 0.47188711166381836,
      "learning_rate": 6.353211009174313e-06,
      "loss": 0.3879,
      "step": 831
    },
    {
      "epoch": 0.038180900371713095,
      "grad_norm": 0.5207975506782532,
      "learning_rate": 6.36085626911315e-06,
      "loss": 0.5027,
      "step": 832
    },
    {
      "epoch": 0.03822679087696756,
      "grad_norm": 0.4640098512172699,
      "learning_rate": 6.368501529051988e-06,
      "loss": 0.3845,
      "step": 833
    },
    {
      "epoch": 0.03827268138222202,
      "grad_norm": 0.49735915660858154,
      "learning_rate": 6.376146788990826e-06,
      "loss": 0.4481,
      "step": 834
    },
    {
      "epoch": 0.038318571887476484,
      "grad_norm": 0.4384276866912842,
      "learning_rate": 6.383792048929664e-06,
      "loss": 0.3684,
      "step": 835
    },
    {
      "epoch": 0.03836446239273095,
      "grad_norm": 0.5249605178833008,
      "learning_rate": 6.391437308868502e-06,
      "loss": 0.5716,
      "step": 836
    },
    {
      "epoch": 0.03841035289798541,
      "grad_norm": 0.4621300995349884,
      "learning_rate": 6.39908256880734e-06,
      "loss": 0.4427,
      "step": 837
    },
    {
      "epoch": 0.03845624340323987,
      "grad_norm": 0.488831490278244,
      "learning_rate": 6.406727828746177e-06,
      "loss": 0.4785,
      "step": 838
    },
    {
      "epoch": 0.03850213390849433,
      "grad_norm": 0.5020730495452881,
      "learning_rate": 6.414373088685016e-06,
      "loss": 0.4396,
      "step": 839
    },
    {
      "epoch": 0.03854802441374879,
      "grad_norm": 0.508812665939331,
      "learning_rate": 6.422018348623854e-06,
      "loss": 0.3891,
      "step": 840
    },
    {
      "epoch": 0.038593914919003255,
      "grad_norm": 0.5025983452796936,
      "learning_rate": 6.429663608562692e-06,
      "loss": 0.4385,
      "step": 841
    },
    {
      "epoch": 0.03863980542425772,
      "grad_norm": 0.45254746079444885,
      "learning_rate": 6.43730886850153e-06,
      "loss": 0.3807,
      "step": 842
    },
    {
      "epoch": 0.03868569592951218,
      "grad_norm": 0.4926454424858093,
      "learning_rate": 6.444954128440367e-06,
      "loss": 0.3821,
      "step": 843
    },
    {
      "epoch": 0.038731586434766645,
      "grad_norm": 0.5796123147010803,
      "learning_rate": 6.452599388379206e-06,
      "loss": 0.5965,
      "step": 844
    },
    {
      "epoch": 0.03877747694002111,
      "grad_norm": 0.45580214262008667,
      "learning_rate": 6.460244648318044e-06,
      "loss": 0.3457,
      "step": 845
    },
    {
      "epoch": 0.03882336744527557,
      "grad_norm": 0.4710889756679535,
      "learning_rate": 6.467889908256881e-06,
      "loss": 0.3665,
      "step": 846
    },
    {
      "epoch": 0.038869257950530034,
      "grad_norm": 0.4428875744342804,
      "learning_rate": 6.4755351681957194e-06,
      "loss": 0.3398,
      "step": 847
    },
    {
      "epoch": 0.0389151484557845,
      "grad_norm": 0.4860277473926544,
      "learning_rate": 6.483180428134557e-06,
      "loss": 0.4059,
      "step": 848
    },
    {
      "epoch": 0.03896103896103896,
      "grad_norm": 0.519835352897644,
      "learning_rate": 6.490825688073395e-06,
      "loss": 0.466,
      "step": 849
    },
    {
      "epoch": 0.03900692946629342,
      "grad_norm": 0.5390611290931702,
      "learning_rate": 6.4984709480122335e-06,
      "loss": 0.5381,
      "step": 850
    },
    {
      "epoch": 0.039052819971547886,
      "grad_norm": 0.4673818349838257,
      "learning_rate": 6.506116207951071e-06,
      "loss": 0.3762,
      "step": 851
    },
    {
      "epoch": 0.03909871047680235,
      "grad_norm": 0.5004668235778809,
      "learning_rate": 6.513761467889908e-06,
      "loss": 0.4299,
      "step": 852
    },
    {
      "epoch": 0.03914460098205681,
      "grad_norm": 0.4638606905937195,
      "learning_rate": 6.521406727828747e-06,
      "loss": 0.3703,
      "step": 853
    },
    {
      "epoch": 0.039190491487311276,
      "grad_norm": 0.47947144508361816,
      "learning_rate": 6.529051987767585e-06,
      "loss": 0.4012,
      "step": 854
    },
    {
      "epoch": 0.03923638199256574,
      "grad_norm": 0.5222530364990234,
      "learning_rate": 6.536697247706422e-06,
      "loss": 0.4813,
      "step": 855
    },
    {
      "epoch": 0.0392822724978202,
      "grad_norm": 0.48211199045181274,
      "learning_rate": 6.544342507645261e-06,
      "loss": 0.3946,
      "step": 856
    },
    {
      "epoch": 0.039328163003074665,
      "grad_norm": 0.48552441596984863,
      "learning_rate": 6.551987767584098e-06,
      "loss": 0.3975,
      "step": 857
    },
    {
      "epoch": 0.03937405350832913,
      "grad_norm": 0.4373522996902466,
      "learning_rate": 6.559633027522936e-06,
      "loss": 0.3136,
      "step": 858
    },
    {
      "epoch": 0.03941994401358359,
      "grad_norm": 0.45665863156318665,
      "learning_rate": 6.567278287461775e-06,
      "loss": 0.4093,
      "step": 859
    },
    {
      "epoch": 0.039465834518838054,
      "grad_norm": 0.48821836709976196,
      "learning_rate": 6.574923547400612e-06,
      "loss": 0.4146,
      "step": 860
    },
    {
      "epoch": 0.03951172502409252,
      "grad_norm": 0.44661468267440796,
      "learning_rate": 6.58256880733945e-06,
      "loss": 0.3337,
      "step": 861
    },
    {
      "epoch": 0.03955761552934698,
      "grad_norm": 0.46406564116477966,
      "learning_rate": 6.590214067278288e-06,
      "loss": 0.3904,
      "step": 862
    },
    {
      "epoch": 0.03960350603460144,
      "grad_norm": 0.4877103567123413,
      "learning_rate": 6.597859327217125e-06,
      "loss": 0.4084,
      "step": 863
    },
    {
      "epoch": 0.039649396539855906,
      "grad_norm": 0.46920427680015564,
      "learning_rate": 6.6055045871559645e-06,
      "loss": 0.3958,
      "step": 864
    },
    {
      "epoch": 0.03969528704511037,
      "grad_norm": 0.5075839757919312,
      "learning_rate": 6.613149847094802e-06,
      "loss": 0.4671,
      "step": 865
    },
    {
      "epoch": 0.03974117755036483,
      "grad_norm": 0.5123674273490906,
      "learning_rate": 6.620795107033639e-06,
      "loss": 0.4602,
      "step": 866
    },
    {
      "epoch": 0.039787068055619296,
      "grad_norm": 0.5872593522071838,
      "learning_rate": 6.628440366972477e-06,
      "loss": 0.4539,
      "step": 867
    },
    {
      "epoch": 0.03983295856087375,
      "grad_norm": 0.5298520922660828,
      "learning_rate": 6.636085626911316e-06,
      "loss": 0.4833,
      "step": 868
    },
    {
      "epoch": 0.039878849066128215,
      "grad_norm": 0.5340838432312012,
      "learning_rate": 6.6437308868501535e-06,
      "loss": 0.5307,
      "step": 869
    },
    {
      "epoch": 0.03992473957138268,
      "grad_norm": 0.46776753664016724,
      "learning_rate": 6.651376146788992e-06,
      "loss": 0.3792,
      "step": 870
    },
    {
      "epoch": 0.03997063007663714,
      "grad_norm": 0.5008481740951538,
      "learning_rate": 6.659021406727829e-06,
      "loss": 0.4161,
      "step": 871
    },
    {
      "epoch": 0.040016520581891604,
      "grad_norm": 0.5117496252059937,
      "learning_rate": 6.666666666666667e-06,
      "loss": 0.4842,
      "step": 872
    },
    {
      "epoch": 0.04006241108714607,
      "grad_norm": 0.4765634834766388,
      "learning_rate": 6.674311926605506e-06,
      "loss": 0.3831,
      "step": 873
    },
    {
      "epoch": 0.04010830159240053,
      "grad_norm": 0.421880304813385,
      "learning_rate": 6.681957186544343e-06,
      "loss": 0.3088,
      "step": 874
    },
    {
      "epoch": 0.040154192097654993,
      "grad_norm": 0.4803207516670227,
      "learning_rate": 6.689602446483181e-06,
      "loss": 0.4326,
      "step": 875
    },
    {
      "epoch": 0.04020008260290946,
      "grad_norm": 0.4485001266002655,
      "learning_rate": 6.697247706422019e-06,
      "loss": 0.3705,
      "step": 876
    },
    {
      "epoch": 0.04024597310816392,
      "grad_norm": 0.6779983639717102,
      "learning_rate": 6.7048929663608564e-06,
      "loss": 0.4751,
      "step": 877
    },
    {
      "epoch": 0.04029186361341838,
      "grad_norm": 0.48208385705947876,
      "learning_rate": 6.7125382262996956e-06,
      "loss": 0.3827,
      "step": 878
    },
    {
      "epoch": 0.040337754118672846,
      "grad_norm": 0.4891941547393799,
      "learning_rate": 6.720183486238533e-06,
      "loss": 0.4127,
      "step": 879
    },
    {
      "epoch": 0.04038364462392731,
      "grad_norm": 0.5813407897949219,
      "learning_rate": 6.7278287461773705e-06,
      "loss": 0.5615,
      "step": 880
    },
    {
      "epoch": 0.04042953512918177,
      "grad_norm": 0.48525410890579224,
      "learning_rate": 6.735474006116208e-06,
      "loss": 0.4173,
      "step": 881
    },
    {
      "epoch": 0.040475425634436235,
      "grad_norm": 0.4815974831581116,
      "learning_rate": 6.743119266055046e-06,
      "loss": 0.4394,
      "step": 882
    },
    {
      "epoch": 0.0405213161396907,
      "grad_norm": 0.4465784430503845,
      "learning_rate": 6.7507645259938845e-06,
      "loss": 0.3633,
      "step": 883
    },
    {
      "epoch": 0.04056720664494516,
      "grad_norm": 0.5458753705024719,
      "learning_rate": 6.758409785932723e-06,
      "loss": 0.4662,
      "step": 884
    },
    {
      "epoch": 0.040613097150199624,
      "grad_norm": 0.5036852955818176,
      "learning_rate": 6.76605504587156e-06,
      "loss": 0.4756,
      "step": 885
    },
    {
      "epoch": 0.04065898765545409,
      "grad_norm": 0.5536975860595703,
      "learning_rate": 6.773700305810398e-06,
      "loss": 0.5082,
      "step": 886
    },
    {
      "epoch": 0.04070487816070855,
      "grad_norm": 0.5013907551765442,
      "learning_rate": 6.781345565749235e-06,
      "loss": 0.4929,
      "step": 887
    },
    {
      "epoch": 0.040750768665963014,
      "grad_norm": 0.6268038749694824,
      "learning_rate": 6.788990825688074e-06,
      "loss": 0.5717,
      "step": 888
    },
    {
      "epoch": 0.04079665917121748,
      "grad_norm": 0.5077654123306274,
      "learning_rate": 6.796636085626912e-06,
      "loss": 0.508,
      "step": 889
    },
    {
      "epoch": 0.04084254967647194,
      "grad_norm": 0.5034418106079102,
      "learning_rate": 6.80428134556575e-06,
      "loss": 0.4734,
      "step": 890
    },
    {
      "epoch": 0.0408884401817264,
      "grad_norm": 0.5471553802490234,
      "learning_rate": 6.8119266055045875e-06,
      "loss": 0.4197,
      "step": 891
    },
    {
      "epoch": 0.040934330686980866,
      "grad_norm": 0.503756582736969,
      "learning_rate": 6.819571865443425e-06,
      "loss": 0.4441,
      "step": 892
    },
    {
      "epoch": 0.04098022119223533,
      "grad_norm": 0.5133985877037048,
      "learning_rate": 6.827217125382264e-06,
      "loss": 0.4309,
      "step": 893
    },
    {
      "epoch": 0.04102611169748979,
      "grad_norm": 0.5441486835479736,
      "learning_rate": 6.8348623853211015e-06,
      "loss": 0.5227,
      "step": 894
    },
    {
      "epoch": 0.041072002202744255,
      "grad_norm": 0.48359623551368713,
      "learning_rate": 6.842507645259939e-06,
      "loss": 0.3764,
      "step": 895
    },
    {
      "epoch": 0.04111789270799872,
      "grad_norm": 0.46526890993118286,
      "learning_rate": 6.850152905198777e-06,
      "loss": 0.3697,
      "step": 896
    },
    {
      "epoch": 0.041163783213253174,
      "grad_norm": 0.4271526038646698,
      "learning_rate": 6.8577981651376156e-06,
      "loss": 0.3467,
      "step": 897
    },
    {
      "epoch": 0.04120967371850764,
      "grad_norm": 0.4491538107395172,
      "learning_rate": 6.865443425076454e-06,
      "loss": 0.3525,
      "step": 898
    },
    {
      "epoch": 0.0412555642237621,
      "grad_norm": 0.4674430191516876,
      "learning_rate": 6.873088685015291e-06,
      "loss": 0.376,
      "step": 899
    },
    {
      "epoch": 0.041301454729016564,
      "grad_norm": 0.4818445146083832,
      "learning_rate": 6.880733944954129e-06,
      "loss": 0.398,
      "step": 900
    },
    {
      "epoch": 0.04134734523427103,
      "grad_norm": 0.4773513674736023,
      "learning_rate": 6.888379204892966e-06,
      "loss": 0.4041,
      "step": 901
    },
    {
      "epoch": 0.04139323573952549,
      "grad_norm": 0.5189515352249146,
      "learning_rate": 6.896024464831805e-06,
      "loss": 0.5368,
      "step": 902
    },
    {
      "epoch": 0.04143912624477995,
      "grad_norm": 0.5067550539970398,
      "learning_rate": 6.903669724770643e-06,
      "loss": 0.4413,
      "step": 903
    },
    {
      "epoch": 0.041485016750034416,
      "grad_norm": 0.4592413306236267,
      "learning_rate": 6.911314984709481e-06,
      "loss": 0.3912,
      "step": 904
    },
    {
      "epoch": 0.04153090725528888,
      "grad_norm": 0.451252818107605,
      "learning_rate": 6.9189602446483185e-06,
      "loss": 0.3298,
      "step": 905
    },
    {
      "epoch": 0.04157679776054334,
      "grad_norm": 0.4569626748561859,
      "learning_rate": 6.926605504587156e-06,
      "loss": 0.3555,
      "step": 906
    },
    {
      "epoch": 0.041622688265797805,
      "grad_norm": 0.5089961886405945,
      "learning_rate": 6.934250764525995e-06,
      "loss": 0.4276,
      "step": 907
    },
    {
      "epoch": 0.04166857877105227,
      "grad_norm": 0.6535539627075195,
      "learning_rate": 6.941896024464833e-06,
      "loss": 0.6558,
      "step": 908
    },
    {
      "epoch": 0.04171446927630673,
      "grad_norm": 0.4889644980430603,
      "learning_rate": 6.94954128440367e-06,
      "loss": 0.384,
      "step": 909
    },
    {
      "epoch": 0.041760359781561195,
      "grad_norm": 0.49032333493232727,
      "learning_rate": 6.957186544342508e-06,
      "loss": 0.4779,
      "step": 910
    },
    {
      "epoch": 0.04180625028681566,
      "grad_norm": 0.5432859659194946,
      "learning_rate": 6.964831804281346e-06,
      "loss": 0.4565,
      "step": 911
    },
    {
      "epoch": 0.04185214079207012,
      "grad_norm": 0.4391240179538727,
      "learning_rate": 6.972477064220184e-06,
      "loss": 0.3484,
      "step": 912
    },
    {
      "epoch": 0.041898031297324584,
      "grad_norm": 0.5433114767074585,
      "learning_rate": 6.980122324159022e-06,
      "loss": 0.542,
      "step": 913
    },
    {
      "epoch": 0.04194392180257905,
      "grad_norm": 0.5047166347503662,
      "learning_rate": 6.98776758409786e-06,
      "loss": 0.4734,
      "step": 914
    },
    {
      "epoch": 0.04198981230783351,
      "grad_norm": 0.4411572515964508,
      "learning_rate": 6.995412844036697e-06,
      "loss": 0.3259,
      "step": 915
    },
    {
      "epoch": 0.04203570281308797,
      "grad_norm": 0.476979523897171,
      "learning_rate": 7.0030581039755356e-06,
      "loss": 0.4139,
      "step": 916
    },
    {
      "epoch": 0.042081593318342436,
      "grad_norm": 0.5029655694961548,
      "learning_rate": 7.010703363914374e-06,
      "loss": 0.4258,
      "step": 917
    },
    {
      "epoch": 0.0421274838235969,
      "grad_norm": 0.4646511375904083,
      "learning_rate": 7.018348623853211e-06,
      "loss": 0.3987,
      "step": 918
    },
    {
      "epoch": 0.04217337432885136,
      "grad_norm": 0.5069194436073303,
      "learning_rate": 7.02599388379205e-06,
      "loss": 0.3741,
      "step": 919
    },
    {
      "epoch": 0.042219264834105825,
      "grad_norm": 0.5072588920593262,
      "learning_rate": 7.033639143730887e-06,
      "loss": 0.4191,
      "step": 920
    },
    {
      "epoch": 0.04226515533936029,
      "grad_norm": 0.5100658535957336,
      "learning_rate": 7.041284403669726e-06,
      "loss": 0.4241,
      "step": 921
    },
    {
      "epoch": 0.04231104584461475,
      "grad_norm": 0.4676949679851532,
      "learning_rate": 7.048929663608564e-06,
      "loss": 0.4464,
      "step": 922
    },
    {
      "epoch": 0.042356936349869215,
      "grad_norm": 0.48504266142845154,
      "learning_rate": 7.056574923547401e-06,
      "loss": 0.3581,
      "step": 923
    },
    {
      "epoch": 0.04240282685512368,
      "grad_norm": 0.46454373002052307,
      "learning_rate": 7.0642201834862385e-06,
      "loss": 0.3942,
      "step": 924
    },
    {
      "epoch": 0.04244871736037814,
      "grad_norm": 0.44111472368240356,
      "learning_rate": 7.071865443425077e-06,
      "loss": 0.3617,
      "step": 925
    },
    {
      "epoch": 0.042494607865632604,
      "grad_norm": 0.4450927972793579,
      "learning_rate": 7.079510703363915e-06,
      "loss": 0.3347,
      "step": 926
    },
    {
      "epoch": 0.04254049837088706,
      "grad_norm": 0.5157008171081543,
      "learning_rate": 7.087155963302753e-06,
      "loss": 0.5074,
      "step": 927
    },
    {
      "epoch": 0.04258638887614152,
      "grad_norm": 0.4555475413799286,
      "learning_rate": 7.094801223241591e-06,
      "loss": 0.3985,
      "step": 928
    },
    {
      "epoch": 0.042632279381395986,
      "grad_norm": 0.5189474821090698,
      "learning_rate": 7.102446483180428e-06,
      "loss": 0.4656,
      "step": 929
    },
    {
      "epoch": 0.04267816988665045,
      "grad_norm": 0.44826528429985046,
      "learning_rate": 7.110091743119267e-06,
      "loss": 0.3747,
      "step": 930
    },
    {
      "epoch": 0.04272406039190491,
      "grad_norm": 0.4239514172077179,
      "learning_rate": 7.117737003058105e-06,
      "loss": 0.3316,
      "step": 931
    },
    {
      "epoch": 0.042769950897159376,
      "grad_norm": 0.5833219289779663,
      "learning_rate": 7.125382262996942e-06,
      "loss": 0.5697,
      "step": 932
    },
    {
      "epoch": 0.04281584140241384,
      "grad_norm": 0.45834407210350037,
      "learning_rate": 7.133027522935781e-06,
      "loss": 0.3672,
      "step": 933
    },
    {
      "epoch": 0.0428617319076683,
      "grad_norm": 0.5336815714836121,
      "learning_rate": 7.140672782874618e-06,
      "loss": 0.506,
      "step": 934
    },
    {
      "epoch": 0.042907622412922765,
      "grad_norm": 0.4766632318496704,
      "learning_rate": 7.1483180428134555e-06,
      "loss": 0.4341,
      "step": 935
    },
    {
      "epoch": 0.04295351291817723,
      "grad_norm": 0.4502406716346741,
      "learning_rate": 7.155963302752295e-06,
      "loss": 0.3182,
      "step": 936
    },
    {
      "epoch": 0.04299940342343169,
      "grad_norm": 0.5229837894439697,
      "learning_rate": 7.163608562691132e-06,
      "loss": 0.4083,
      "step": 937
    },
    {
      "epoch": 0.043045293928686154,
      "grad_norm": 0.510887086391449,
      "learning_rate": 7.17125382262997e-06,
      "loss": 0.4283,
      "step": 938
    },
    {
      "epoch": 0.04309118443394062,
      "grad_norm": 0.5132585167884827,
      "learning_rate": 7.178899082568808e-06,
      "loss": 0.4038,
      "step": 939
    },
    {
      "epoch": 0.04313707493919508,
      "grad_norm": 0.5235181450843811,
      "learning_rate": 7.186544342507645e-06,
      "loss": 0.5071,
      "step": 940
    },
    {
      "epoch": 0.04318296544444954,
      "grad_norm": 0.40967270731925964,
      "learning_rate": 7.1941896024464845e-06,
      "loss": 0.3069,
      "step": 941
    },
    {
      "epoch": 0.043228855949704006,
      "grad_norm": 0.5366747975349426,
      "learning_rate": 7.201834862385322e-06,
      "loss": 0.4964,
      "step": 942
    },
    {
      "epoch": 0.04327474645495847,
      "grad_norm": 0.4693954586982727,
      "learning_rate": 7.209480122324159e-06,
      "loss": 0.3575,
      "step": 943
    },
    {
      "epoch": 0.04332063696021293,
      "grad_norm": 0.48921796679496765,
      "learning_rate": 7.217125382262997e-06,
      "loss": 0.395,
      "step": 944
    },
    {
      "epoch": 0.043366527465467396,
      "grad_norm": 0.40874385833740234,
      "learning_rate": 7.224770642201836e-06,
      "loss": 0.2886,
      "step": 945
    },
    {
      "epoch": 0.04341241797072186,
      "grad_norm": 0.48684173822402954,
      "learning_rate": 7.232415902140673e-06,
      "loss": 0.3695,
      "step": 946
    },
    {
      "epoch": 0.04345830847597632,
      "grad_norm": 0.6032975912094116,
      "learning_rate": 7.240061162079512e-06,
      "loss": 0.4136,
      "step": 947
    },
    {
      "epoch": 0.043504198981230785,
      "grad_norm": 0.48327934741973877,
      "learning_rate": 7.247706422018349e-06,
      "loss": 0.3671,
      "step": 948
    },
    {
      "epoch": 0.04355008948648525,
      "grad_norm": 0.4743063151836395,
      "learning_rate": 7.255351681957187e-06,
      "loss": 0.4433,
      "step": 949
    },
    {
      "epoch": 0.04359597999173971,
      "grad_norm": 0.4645388126373291,
      "learning_rate": 7.262996941896026e-06,
      "loss": 0.3776,
      "step": 950
    },
    {
      "epoch": 0.043641870496994174,
      "grad_norm": 0.5633676648139954,
      "learning_rate": 7.270642201834863e-06,
      "loss": 0.3596,
      "step": 951
    },
    {
      "epoch": 0.04368776100224864,
      "grad_norm": 0.5181424021720886,
      "learning_rate": 7.278287461773701e-06,
      "loss": 0.5116,
      "step": 952
    },
    {
      "epoch": 0.0437336515075031,
      "grad_norm": 0.4617640972137451,
      "learning_rate": 7.285932721712539e-06,
      "loss": 0.3885,
      "step": 953
    },
    {
      "epoch": 0.043779542012757564,
      "grad_norm": 0.4297509789466858,
      "learning_rate": 7.293577981651376e-06,
      "loss": 0.3777,
      "step": 954
    },
    {
      "epoch": 0.04382543251801203,
      "grad_norm": 0.48908212780952454,
      "learning_rate": 7.3012232415902155e-06,
      "loss": 0.3916,
      "step": 955
    },
    {
      "epoch": 0.04387132302326648,
      "grad_norm": 0.4835954010486603,
      "learning_rate": 7.308868501529053e-06,
      "loss": 0.442,
      "step": 956
    },
    {
      "epoch": 0.043917213528520946,
      "grad_norm": 0.6452763080596924,
      "learning_rate": 7.31651376146789e-06,
      "loss": 0.4267,
      "step": 957
    },
    {
      "epoch": 0.04396310403377541,
      "grad_norm": 0.48812511563301086,
      "learning_rate": 7.324159021406728e-06,
      "loss": 0.397,
      "step": 958
    },
    {
      "epoch": 0.04400899453902987,
      "grad_norm": 0.552649974822998,
      "learning_rate": 7.331804281345566e-06,
      "loss": 0.4923,
      "step": 959
    },
    {
      "epoch": 0.044054885044284335,
      "grad_norm": 0.5882543921470642,
      "learning_rate": 7.3394495412844045e-06,
      "loss": 0.4857,
      "step": 960
    },
    {
      "epoch": 0.0441007755495388,
      "grad_norm": 0.47771885991096497,
      "learning_rate": 7.347094801223243e-06,
      "loss": 0.3856,
      "step": 961
    },
    {
      "epoch": 0.04414666605479326,
      "grad_norm": 0.47342777252197266,
      "learning_rate": 7.35474006116208e-06,
      "loss": 0.4174,
      "step": 962
    },
    {
      "epoch": 0.044192556560047724,
      "grad_norm": 0.47401317954063416,
      "learning_rate": 7.362385321100918e-06,
      "loss": 0.4041,
      "step": 963
    },
    {
      "epoch": 0.04423844706530219,
      "grad_norm": 0.4642093777656555,
      "learning_rate": 7.370030581039755e-06,
      "loss": 0.3239,
      "step": 964
    },
    {
      "epoch": 0.04428433757055665,
      "grad_norm": 0.4429576098918915,
      "learning_rate": 7.377675840978594e-06,
      "loss": 0.3217,
      "step": 965
    },
    {
      "epoch": 0.044330228075811114,
      "grad_norm": 0.5023338794708252,
      "learning_rate": 7.385321100917432e-06,
      "loss": 0.4977,
      "step": 966
    },
    {
      "epoch": 0.04437611858106558,
      "grad_norm": 0.463932603597641,
      "learning_rate": 7.39296636085627e-06,
      "loss": 0.3722,
      "step": 967
    },
    {
      "epoch": 0.04442200908632004,
      "grad_norm": 0.4808919131755829,
      "learning_rate": 7.4006116207951074e-06,
      "loss": 0.4459,
      "step": 968
    },
    {
      "epoch": 0.0444678995915745,
      "grad_norm": 0.5225597023963928,
      "learning_rate": 7.408256880733946e-06,
      "loss": 0.4114,
      "step": 969
    },
    {
      "epoch": 0.044513790096828966,
      "grad_norm": 0.49161815643310547,
      "learning_rate": 7.415902140672784e-06,
      "loss": 0.3953,
      "step": 970
    },
    {
      "epoch": 0.04455968060208343,
      "grad_norm": 0.46979379653930664,
      "learning_rate": 7.4235474006116215e-06,
      "loss": 0.3508,
      "step": 971
    },
    {
      "epoch": 0.04460557110733789,
      "grad_norm": 0.4807756543159485,
      "learning_rate": 7.431192660550459e-06,
      "loss": 0.4344,
      "step": 972
    },
    {
      "epoch": 0.044651461612592355,
      "grad_norm": 0.45674464106559753,
      "learning_rate": 7.438837920489297e-06,
      "loss": 0.3935,
      "step": 973
    },
    {
      "epoch": 0.04469735211784682,
      "grad_norm": 0.4198451340198517,
      "learning_rate": 7.4464831804281355e-06,
      "loss": 0.3077,
      "step": 974
    },
    {
      "epoch": 0.04474324262310128,
      "grad_norm": 0.4758407473564148,
      "learning_rate": 7.454128440366973e-06,
      "loss": 0.4641,
      "step": 975
    },
    {
      "epoch": 0.044789133128355745,
      "grad_norm": 0.44405484199523926,
      "learning_rate": 7.461773700305811e-06,
      "loss": 0.3356,
      "step": 976
    },
    {
      "epoch": 0.04483502363361021,
      "grad_norm": 0.492264062166214,
      "learning_rate": 7.469418960244649e-06,
      "loss": 0.3786,
      "step": 977
    },
    {
      "epoch": 0.04488091413886467,
      "grad_norm": 0.5459489226341248,
      "learning_rate": 7.477064220183486e-06,
      "loss": 0.5428,
      "step": 978
    },
    {
      "epoch": 0.044926804644119134,
      "grad_norm": 0.4653685688972473,
      "learning_rate": 7.484709480122325e-06,
      "loss": 0.4453,
      "step": 979
    },
    {
      "epoch": 0.0449726951493736,
      "grad_norm": 0.44355902075767517,
      "learning_rate": 7.492354740061163e-06,
      "loss": 0.351,
      "step": 980
    },
    {
      "epoch": 0.04501858565462806,
      "grad_norm": 0.47885799407958984,
      "learning_rate": 7.500000000000001e-06,
      "loss": 0.3902,
      "step": 981
    },
    {
      "epoch": 0.04506447615988252,
      "grad_norm": 0.48619183897972107,
      "learning_rate": 7.5076452599388385e-06,
      "loss": 0.4266,
      "step": 982
    },
    {
      "epoch": 0.045110366665136986,
      "grad_norm": 0.5208702683448792,
      "learning_rate": 7.515290519877676e-06,
      "loss": 0.4965,
      "step": 983
    },
    {
      "epoch": 0.04515625717039145,
      "grad_norm": 0.48644906282424927,
      "learning_rate": 7.522935779816515e-06,
      "loss": 0.4389,
      "step": 984
    },
    {
      "epoch": 0.04520214767564591,
      "grad_norm": 0.5237307548522949,
      "learning_rate": 7.5305810397553525e-06,
      "loss": 0.5208,
      "step": 985
    },
    {
      "epoch": 0.04524803818090037,
      "grad_norm": 0.5199766159057617,
      "learning_rate": 7.53822629969419e-06,
      "loss": 0.5159,
      "step": 986
    },
    {
      "epoch": 0.04529392868615483,
      "grad_norm": 0.4719480276107788,
      "learning_rate": 7.545871559633028e-06,
      "loss": 0.4001,
      "step": 987
    },
    {
      "epoch": 0.045339819191409295,
      "grad_norm": 0.5043044090270996,
      "learning_rate": 7.553516819571866e-06,
      "loss": 0.4523,
      "step": 988
    },
    {
      "epoch": 0.04538570969666376,
      "grad_norm": 0.5083537101745605,
      "learning_rate": 7.561162079510704e-06,
      "loss": 0.4894,
      "step": 989
    },
    {
      "epoch": 0.04543160020191822,
      "grad_norm": 0.4714929163455963,
      "learning_rate": 7.568807339449542e-06,
      "loss": 0.3466,
      "step": 990
    },
    {
      "epoch": 0.045477490707172684,
      "grad_norm": 0.5016339421272278,
      "learning_rate": 7.57645259938838e-06,
      "loss": 0.4748,
      "step": 991
    },
    {
      "epoch": 0.04552338121242715,
      "grad_norm": 0.496962308883667,
      "learning_rate": 7.584097859327217e-06,
      "loss": 0.4705,
      "step": 992
    },
    {
      "epoch": 0.04556927171768161,
      "grad_norm": 0.4911184310913086,
      "learning_rate": 7.5917431192660555e-06,
      "loss": 0.4765,
      "step": 993
    },
    {
      "epoch": 0.04561516222293607,
      "grad_norm": 0.5251611471176147,
      "learning_rate": 7.599388379204894e-06,
      "loss": 0.507,
      "step": 994
    },
    {
      "epoch": 0.045661052728190536,
      "grad_norm": 0.5036035776138306,
      "learning_rate": 7.607033639143731e-06,
      "loss": 0.4383,
      "step": 995
    },
    {
      "epoch": 0.045706943233445,
      "grad_norm": 0.5988549590110779,
      "learning_rate": 7.6146788990825695e-06,
      "loss": 0.5687,
      "step": 996
    },
    {
      "epoch": 0.04575283373869946,
      "grad_norm": 0.519762396812439,
      "learning_rate": 7.622324159021407e-06,
      "loss": 0.3997,
      "step": 997
    },
    {
      "epoch": 0.045798724243953925,
      "grad_norm": 0.4208580553531647,
      "learning_rate": 7.629969418960246e-06,
      "loss": 0.3384,
      "step": 998
    },
    {
      "epoch": 0.04584461474920839,
      "grad_norm": 0.5738177299499512,
      "learning_rate": 7.637614678899084e-06,
      "loss": 0.5505,
      "step": 999
    },
    {
      "epoch": 0.04589050525446285,
      "grad_norm": 0.4675780236721039,
      "learning_rate": 7.645259938837921e-06,
      "loss": 0.4085,
      "step": 1000
    },
    {
      "epoch": 0.045936395759717315,
      "grad_norm": 0.5079315900802612,
      "learning_rate": 7.652905198776758e-06,
      "loss": 0.4242,
      "step": 1001
    },
    {
      "epoch": 0.04598228626497178,
      "grad_norm": 0.45595890283584595,
      "learning_rate": 7.660550458715596e-06,
      "loss": 0.351,
      "step": 1002
    },
    {
      "epoch": 0.04602817677022624,
      "grad_norm": 0.4902017414569855,
      "learning_rate": 7.668195718654435e-06,
      "loss": 0.412,
      "step": 1003
    },
    {
      "epoch": 0.046074067275480704,
      "grad_norm": 0.506516695022583,
      "learning_rate": 7.675840978593273e-06,
      "loss": 0.4548,
      "step": 1004
    },
    {
      "epoch": 0.04611995778073517,
      "grad_norm": 0.4780597984790802,
      "learning_rate": 7.68348623853211e-06,
      "loss": 0.4037,
      "step": 1005
    },
    {
      "epoch": 0.04616584828598963,
      "grad_norm": 0.5067412257194519,
      "learning_rate": 7.691131498470949e-06,
      "loss": 0.433,
      "step": 1006
    },
    {
      "epoch": 0.04621173879124409,
      "grad_norm": 0.46468278765678406,
      "learning_rate": 7.698776758409787e-06,
      "loss": 0.3715,
      "step": 1007
    },
    {
      "epoch": 0.046257629296498556,
      "grad_norm": 0.4942983090877533,
      "learning_rate": 7.706422018348626e-06,
      "loss": 0.4499,
      "step": 1008
    },
    {
      "epoch": 0.04630351980175302,
      "grad_norm": 0.46590471267700195,
      "learning_rate": 7.714067278287463e-06,
      "loss": 0.3844,
      "step": 1009
    },
    {
      "epoch": 0.04634941030700748,
      "grad_norm": 0.5226325988769531,
      "learning_rate": 7.7217125382263e-06,
      "loss": 0.4663,
      "step": 1010
    },
    {
      "epoch": 0.046395300812261946,
      "grad_norm": 0.5061750411987305,
      "learning_rate": 7.729357798165138e-06,
      "loss": 0.4606,
      "step": 1011
    },
    {
      "epoch": 0.04644119131751641,
      "grad_norm": 0.5064273476600647,
      "learning_rate": 7.737003058103975e-06,
      "loss": 0.4086,
      "step": 1012
    },
    {
      "epoch": 0.04648708182277087,
      "grad_norm": 0.46252965927124023,
      "learning_rate": 7.744648318042815e-06,
      "loss": 0.4199,
      "step": 1013
    },
    {
      "epoch": 0.046532972328025335,
      "grad_norm": 0.4614937901496887,
      "learning_rate": 7.752293577981652e-06,
      "loss": 0.396,
      "step": 1014
    },
    {
      "epoch": 0.04657886283327979,
      "grad_norm": 0.4680635929107666,
      "learning_rate": 7.75993883792049e-06,
      "loss": 0.3866,
      "step": 1015
    },
    {
      "epoch": 0.046624753338534254,
      "grad_norm": 0.45811694860458374,
      "learning_rate": 7.767584097859327e-06,
      "loss": 0.354,
      "step": 1016
    },
    {
      "epoch": 0.04667064384378872,
      "grad_norm": 0.4698444604873657,
      "learning_rate": 7.775229357798164e-06,
      "loss": 0.381,
      "step": 1017
    },
    {
      "epoch": 0.04671653434904318,
      "grad_norm": 0.4797396659851074,
      "learning_rate": 7.782874617737004e-06,
      "loss": 0.3426,
      "step": 1018
    },
    {
      "epoch": 0.04676242485429764,
      "grad_norm": 0.4425613284111023,
      "learning_rate": 7.790519877675841e-06,
      "loss": 0.3473,
      "step": 1019
    },
    {
      "epoch": 0.046808315359552106,
      "grad_norm": 0.4574027359485626,
      "learning_rate": 7.79816513761468e-06,
      "loss": 0.3769,
      "step": 1020
    },
    {
      "epoch": 0.04685420586480657,
      "grad_norm": 0.4499163031578064,
      "learning_rate": 7.805810397553518e-06,
      "loss": 0.3748,
      "step": 1021
    },
    {
      "epoch": 0.04690009637006103,
      "grad_norm": 0.48446786403656006,
      "learning_rate": 7.813455657492357e-06,
      "loss": 0.447,
      "step": 1022
    },
    {
      "epoch": 0.046945986875315496,
      "grad_norm": 0.49225109815597534,
      "learning_rate": 7.821100917431194e-06,
      "loss": 0.3767,
      "step": 1023
    },
    {
      "epoch": 0.04699187738056996,
      "grad_norm": 0.4963989555835724,
      "learning_rate": 7.828746177370032e-06,
      "loss": 0.4705,
      "step": 1024
    },
    {
      "epoch": 0.04703776788582442,
      "grad_norm": 0.4896707236766815,
      "learning_rate": 7.836391437308869e-06,
      "loss": 0.4356,
      "step": 1025
    },
    {
      "epoch": 0.047083658391078885,
      "grad_norm": 0.4789886474609375,
      "learning_rate": 7.844036697247707e-06,
      "loss": 0.343,
      "step": 1026
    },
    {
      "epoch": 0.04712954889633335,
      "grad_norm": 0.46077170968055725,
      "learning_rate": 7.851681957186546e-06,
      "loss": 0.343,
      "step": 1027
    },
    {
      "epoch": 0.04717543940158781,
      "grad_norm": 0.4612993597984314,
      "learning_rate": 7.859327217125383e-06,
      "loss": 0.3912,
      "step": 1028
    },
    {
      "epoch": 0.047221329906842274,
      "grad_norm": 0.447415828704834,
      "learning_rate": 7.86697247706422e-06,
      "loss": 0.3323,
      "step": 1029
    },
    {
      "epoch": 0.04726722041209674,
      "grad_norm": 0.4708919823169708,
      "learning_rate": 7.874617737003058e-06,
      "loss": 0.3577,
      "step": 1030
    },
    {
      "epoch": 0.0473131109173512,
      "grad_norm": 0.448274165391922,
      "learning_rate": 7.882262996941895e-06,
      "loss": 0.3848,
      "step": 1031
    },
    {
      "epoch": 0.047359001422605664,
      "grad_norm": 0.4808759093284607,
      "learning_rate": 7.889908256880735e-06,
      "loss": 0.4683,
      "step": 1032
    },
    {
      "epoch": 0.04740489192786013,
      "grad_norm": 0.5238559246063232,
      "learning_rate": 7.897553516819572e-06,
      "loss": 0.4138,
      "step": 1033
    },
    {
      "epoch": 0.04745078243311459,
      "grad_norm": 0.44173386693000793,
      "learning_rate": 7.905198776758411e-06,
      "loss": 0.3381,
      "step": 1034
    },
    {
      "epoch": 0.04749667293836905,
      "grad_norm": 0.42944467067718506,
      "learning_rate": 7.912844036697249e-06,
      "loss": 0.3466,
      "step": 1035
    },
    {
      "epoch": 0.047542563443623516,
      "grad_norm": 0.48287513852119446,
      "learning_rate": 7.920489296636086e-06,
      "loss": 0.3962,
      "step": 1036
    },
    {
      "epoch": 0.04758845394887798,
      "grad_norm": 0.5754697918891907,
      "learning_rate": 7.928134556574925e-06,
      "loss": 0.5859,
      "step": 1037
    },
    {
      "epoch": 0.04763434445413244,
      "grad_norm": 0.5029233694076538,
      "learning_rate": 7.935779816513763e-06,
      "loss": 0.5072,
      "step": 1038
    },
    {
      "epoch": 0.047680234959386905,
      "grad_norm": 0.476535439491272,
      "learning_rate": 7.9434250764526e-06,
      "loss": 0.3594,
      "step": 1039
    },
    {
      "epoch": 0.04772612546464137,
      "grad_norm": 0.4771139621734619,
      "learning_rate": 7.951070336391438e-06,
      "loss": 0.4406,
      "step": 1040
    },
    {
      "epoch": 0.04777201596989583,
      "grad_norm": 0.4652896821498871,
      "learning_rate": 7.958715596330275e-06,
      "loss": 0.3709,
      "step": 1041
    },
    {
      "epoch": 0.047817906475150294,
      "grad_norm": 0.4450419247150421,
      "learning_rate": 7.966360856269114e-06,
      "loss": 0.3827,
      "step": 1042
    },
    {
      "epoch": 0.04786379698040476,
      "grad_norm": 0.48332735896110535,
      "learning_rate": 7.974006116207952e-06,
      "loss": 0.4561,
      "step": 1043
    },
    {
      "epoch": 0.047909687485659214,
      "grad_norm": 0.5330381989479065,
      "learning_rate": 7.981651376146789e-06,
      "loss": 0.4423,
      "step": 1044
    },
    {
      "epoch": 0.04795557799091368,
      "grad_norm": 0.45897090435028076,
      "learning_rate": 7.989296636085627e-06,
      "loss": 0.3191,
      "step": 1045
    },
    {
      "epoch": 0.04800146849616814,
      "grad_norm": 0.47465670108795166,
      "learning_rate": 7.996941896024466e-06,
      "loss": 0.4216,
      "step": 1046
    },
    {
      "epoch": 0.0480473590014226,
      "grad_norm": 0.48171624541282654,
      "learning_rate": 8.004587155963303e-06,
      "loss": 0.4003,
      "step": 1047
    },
    {
      "epoch": 0.048093249506677066,
      "grad_norm": 0.47438323497772217,
      "learning_rate": 8.01223241590214e-06,
      "loss": 0.388,
      "step": 1048
    },
    {
      "epoch": 0.04813914001193153,
      "grad_norm": 0.6740774512290955,
      "learning_rate": 8.01987767584098e-06,
      "loss": 0.4373,
      "step": 1049
    },
    {
      "epoch": 0.04818503051718599,
      "grad_norm": 0.4899618327617645,
      "learning_rate": 8.027522935779817e-06,
      "loss": 0.4915,
      "step": 1050
    },
    {
      "epoch": 0.048230921022440455,
      "grad_norm": 0.4588882625102997,
      "learning_rate": 8.035168195718656e-06,
      "loss": 0.4083,
      "step": 1051
    },
    {
      "epoch": 0.04827681152769492,
      "grad_norm": 0.4788903594017029,
      "learning_rate": 8.042813455657494e-06,
      "loss": 0.4449,
      "step": 1052
    },
    {
      "epoch": 0.04832270203294938,
      "grad_norm": 0.44267287850379944,
      "learning_rate": 8.050458715596331e-06,
      "loss": 0.3238,
      "step": 1053
    },
    {
      "epoch": 0.048368592538203845,
      "grad_norm": 0.48027700185775757,
      "learning_rate": 8.058103975535169e-06,
      "loss": 0.3214,
      "step": 1054
    },
    {
      "epoch": 0.04841448304345831,
      "grad_norm": 0.4961937963962555,
      "learning_rate": 8.065749235474006e-06,
      "loss": 0.4385,
      "step": 1055
    },
    {
      "epoch": 0.04846037354871277,
      "grad_norm": 0.45443207025527954,
      "learning_rate": 8.073394495412845e-06,
      "loss": 0.3663,
      "step": 1056
    },
    {
      "epoch": 0.048506264053967234,
      "grad_norm": 0.4791746437549591,
      "learning_rate": 8.081039755351683e-06,
      "loss": 0.4246,
      "step": 1057
    },
    {
      "epoch": 0.0485521545592217,
      "grad_norm": 0.41660791635513306,
      "learning_rate": 8.08868501529052e-06,
      "loss": 0.2961,
      "step": 1058
    },
    {
      "epoch": 0.04859804506447616,
      "grad_norm": 0.6231920123100281,
      "learning_rate": 8.096330275229358e-06,
      "loss": 0.4865,
      "step": 1059
    },
    {
      "epoch": 0.04864393556973062,
      "grad_norm": 0.4434593617916107,
      "learning_rate": 8.103975535168197e-06,
      "loss": 0.328,
      "step": 1060
    },
    {
      "epoch": 0.048689826074985086,
      "grad_norm": 0.5092295408248901,
      "learning_rate": 8.111620795107034e-06,
      "loss": 0.4893,
      "step": 1061
    },
    {
      "epoch": 0.04873571658023955,
      "grad_norm": 1.117873191833496,
      "learning_rate": 8.119266055045872e-06,
      "loss": 0.3935,
      "step": 1062
    },
    {
      "epoch": 0.04878160708549401,
      "grad_norm": 0.5893769860267639,
      "learning_rate": 8.12691131498471e-06,
      "loss": 0.5127,
      "step": 1063
    },
    {
      "epoch": 0.048827497590748475,
      "grad_norm": 0.4874710440635681,
      "learning_rate": 8.134556574923548e-06,
      "loss": 0.369,
      "step": 1064
    },
    {
      "epoch": 0.04887338809600294,
      "grad_norm": 0.4883679747581482,
      "learning_rate": 8.142201834862386e-06,
      "loss": 0.3504,
      "step": 1065
    },
    {
      "epoch": 0.0489192786012574,
      "grad_norm": 0.6136423945426941,
      "learning_rate": 8.149847094801225e-06,
      "loss": 0.5323,
      "step": 1066
    },
    {
      "epoch": 0.048965169106511865,
      "grad_norm": 0.526439368724823,
      "learning_rate": 8.157492354740062e-06,
      "loss": 0.469,
      "step": 1067
    },
    {
      "epoch": 0.04901105961176633,
      "grad_norm": 0.4777195155620575,
      "learning_rate": 8.1651376146789e-06,
      "loss": 0.3699,
      "step": 1068
    },
    {
      "epoch": 0.04905695011702079,
      "grad_norm": 0.57276451587677,
      "learning_rate": 8.172782874617737e-06,
      "loss": 0.5136,
      "step": 1069
    },
    {
      "epoch": 0.049102840622275254,
      "grad_norm": 0.48893672227859497,
      "learning_rate": 8.180428134556576e-06,
      "loss": 0.4174,
      "step": 1070
    },
    {
      "epoch": 0.04914873112752972,
      "grad_norm": 0.5125523805618286,
      "learning_rate": 8.188073394495414e-06,
      "loss": 0.4925,
      "step": 1071
    },
    {
      "epoch": 0.04919462163278418,
      "grad_norm": 0.463162362575531,
      "learning_rate": 8.195718654434251e-06,
      "loss": 0.4295,
      "step": 1072
    },
    {
      "epoch": 0.04924051213803864,
      "grad_norm": 0.434839129447937,
      "learning_rate": 8.203363914373089e-06,
      "loss": 0.3128,
      "step": 1073
    },
    {
      "epoch": 0.0492864026432931,
      "grad_norm": 0.45718950033187866,
      "learning_rate": 8.211009174311926e-06,
      "loss": 0.3063,
      "step": 1074
    },
    {
      "epoch": 0.04933229314854756,
      "grad_norm": 0.6839661002159119,
      "learning_rate": 8.218654434250765e-06,
      "loss": 0.6405,
      "step": 1075
    },
    {
      "epoch": 0.049378183653802026,
      "grad_norm": 0.49980488419532776,
      "learning_rate": 8.226299694189603e-06,
      "loss": 0.465,
      "step": 1076
    },
    {
      "epoch": 0.04942407415905649,
      "grad_norm": 0.46241894364356995,
      "learning_rate": 8.233944954128442e-06,
      "loss": 0.3392,
      "step": 1077
    },
    {
      "epoch": 0.04946996466431095,
      "grad_norm": 0.4796869456768036,
      "learning_rate": 8.24159021406728e-06,
      "loss": 0.4357,
      "step": 1078
    },
    {
      "epoch": 0.049515855169565415,
      "grad_norm": 0.4757649600505829,
      "learning_rate": 8.249235474006117e-06,
      "loss": 0.3805,
      "step": 1079
    },
    {
      "epoch": 0.04956174567481988,
      "grad_norm": 0.49290862679481506,
      "learning_rate": 8.256880733944956e-06,
      "loss": 0.4157,
      "step": 1080
    },
    {
      "epoch": 0.04960763618007434,
      "grad_norm": 0.5197950005531311,
      "learning_rate": 8.264525993883793e-06,
      "loss": 0.4078,
      "step": 1081
    },
    {
      "epoch": 0.049653526685328804,
      "grad_norm": 0.4865533709526062,
      "learning_rate": 8.27217125382263e-06,
      "loss": 0.4001,
      "step": 1082
    },
    {
      "epoch": 0.04969941719058327,
      "grad_norm": 0.5275459289550781,
      "learning_rate": 8.279816513761468e-06,
      "loss": 0.4909,
      "step": 1083
    },
    {
      "epoch": 0.04974530769583773,
      "grad_norm": 0.5157485008239746,
      "learning_rate": 8.287461773700306e-06,
      "loss": 0.4823,
      "step": 1084
    },
    {
      "epoch": 0.04979119820109219,
      "grad_norm": 0.44135355949401855,
      "learning_rate": 8.295107033639145e-06,
      "loss": 0.3462,
      "step": 1085
    },
    {
      "epoch": 0.049837088706346656,
      "grad_norm": 0.47907930612564087,
      "learning_rate": 8.302752293577982e-06,
      "loss": 0.4662,
      "step": 1086
    },
    {
      "epoch": 0.04988297921160112,
      "grad_norm": 0.4304979741573334,
      "learning_rate": 8.31039755351682e-06,
      "loss": 0.3433,
      "step": 1087
    },
    {
      "epoch": 0.04992886971685558,
      "grad_norm": 0.5039060115814209,
      "learning_rate": 8.318042813455657e-06,
      "loss": 0.4455,
      "step": 1088
    },
    {
      "epoch": 0.049974760222110046,
      "grad_norm": 0.48326313495635986,
      "learning_rate": 8.325688073394496e-06,
      "loss": 0.4794,
      "step": 1089
    },
    {
      "epoch": 0.05002065072736451,
      "grad_norm": 0.4881344437599182,
      "learning_rate": 8.333333333333334e-06,
      "loss": 0.3763,
      "step": 1090
    },
    {
      "epoch": 0.05006654123261897,
      "grad_norm": 0.5004485845565796,
      "learning_rate": 8.340978593272173e-06,
      "loss": 0.5007,
      "step": 1091
    },
    {
      "epoch": 0.050112431737873435,
      "grad_norm": 0.4826316833496094,
      "learning_rate": 8.34862385321101e-06,
      "loss": 0.3967,
      "step": 1092
    },
    {
      "epoch": 0.0501583222431279,
      "grad_norm": 0.49310603737831116,
      "learning_rate": 8.356269113149848e-06,
      "loss": 0.3985,
      "step": 1093
    },
    {
      "epoch": 0.05020421274838236,
      "grad_norm": 0.45969587564468384,
      "learning_rate": 8.363914373088685e-06,
      "loss": 0.3666,
      "step": 1094
    },
    {
      "epoch": 0.050250103253636824,
      "grad_norm": 0.49299341440200806,
      "learning_rate": 8.371559633027524e-06,
      "loss": 0.4684,
      "step": 1095
    },
    {
      "epoch": 0.05029599375889129,
      "grad_norm": 0.5132526159286499,
      "learning_rate": 8.379204892966362e-06,
      "loss": 0.5313,
      "step": 1096
    },
    {
      "epoch": 0.05034188426414575,
      "grad_norm": 0.45509016513824463,
      "learning_rate": 8.3868501529052e-06,
      "loss": 0.4106,
      "step": 1097
    },
    {
      "epoch": 0.05038777476940021,
      "grad_norm": 0.48494666814804077,
      "learning_rate": 8.394495412844037e-06,
      "loss": 0.4712,
      "step": 1098
    },
    {
      "epoch": 0.050433665274654677,
      "grad_norm": 0.5069655179977417,
      "learning_rate": 8.402140672782876e-06,
      "loss": 0.5029,
      "step": 1099
    },
    {
      "epoch": 0.05047955577990914,
      "grad_norm": 0.4696100354194641,
      "learning_rate": 8.409785932721713e-06,
      "loss": 0.4046,
      "step": 1100
    },
    {
      "epoch": 0.0505254462851636,
      "grad_norm": 0.4602081775665283,
      "learning_rate": 8.41743119266055e-06,
      "loss": 0.3858,
      "step": 1101
    },
    {
      "epoch": 0.050571336790418066,
      "grad_norm": 0.44474631547927856,
      "learning_rate": 8.425076452599388e-06,
      "loss": 0.3764,
      "step": 1102
    },
    {
      "epoch": 0.05061722729567252,
      "grad_norm": 0.42140090465545654,
      "learning_rate": 8.432721712538227e-06,
      "loss": 0.3266,
      "step": 1103
    },
    {
      "epoch": 0.050663117800926985,
      "grad_norm": 0.5296817421913147,
      "learning_rate": 8.440366972477065e-06,
      "loss": 0.4808,
      "step": 1104
    },
    {
      "epoch": 0.05070900830618145,
      "grad_norm": 0.5089552402496338,
      "learning_rate": 8.448012232415902e-06,
      "loss": 0.5265,
      "step": 1105
    },
    {
      "epoch": 0.05075489881143591,
      "grad_norm": 0.6396613717079163,
      "learning_rate": 8.455657492354741e-06,
      "loss": 0.5338,
      "step": 1106
    },
    {
      "epoch": 0.050800789316690374,
      "grad_norm": 0.5080808997154236,
      "learning_rate": 8.463302752293579e-06,
      "loss": 0.427,
      "step": 1107
    },
    {
      "epoch": 0.05084667982194484,
      "grad_norm": 0.5155501961708069,
      "learning_rate": 8.470948012232416e-06,
      "loss": 0.4954,
      "step": 1108
    },
    {
      "epoch": 0.0508925703271993,
      "grad_norm": 0.4518275558948517,
      "learning_rate": 8.478593272171255e-06,
      "loss": 0.3507,
      "step": 1109
    },
    {
      "epoch": 0.050938460832453764,
      "grad_norm": 0.5265688300132751,
      "learning_rate": 8.486238532110093e-06,
      "loss": 0.4542,
      "step": 1110
    },
    {
      "epoch": 0.05098435133770823,
      "grad_norm": 0.46242907643318176,
      "learning_rate": 8.49388379204893e-06,
      "loss": 0.3573,
      "step": 1111
    },
    {
      "epoch": 0.05103024184296269,
      "grad_norm": 0.535705029964447,
      "learning_rate": 8.501529051987768e-06,
      "loss": 0.5078,
      "step": 1112
    },
    {
      "epoch": 0.05107613234821715,
      "grad_norm": 0.532334566116333,
      "learning_rate": 8.509174311926605e-06,
      "loss": 0.5033,
      "step": 1113
    },
    {
      "epoch": 0.051122022853471616,
      "grad_norm": 0.4832940697669983,
      "learning_rate": 8.516819571865444e-06,
      "loss": 0.494,
      "step": 1114
    },
    {
      "epoch": 0.05116791335872608,
      "grad_norm": 0.5339627265930176,
      "learning_rate": 8.524464831804282e-06,
      "loss": 0.4468,
      "step": 1115
    },
    {
      "epoch": 0.05121380386398054,
      "grad_norm": 0.4926060140132904,
      "learning_rate": 8.53211009174312e-06,
      "loss": 0.4503,
      "step": 1116
    },
    {
      "epoch": 0.051259694369235005,
      "grad_norm": 0.4966335594654083,
      "learning_rate": 8.539755351681958e-06,
      "loss": 0.3704,
      "step": 1117
    },
    {
      "epoch": 0.05130558487448947,
      "grad_norm": 0.5581846237182617,
      "learning_rate": 8.547400611620796e-06,
      "loss": 0.5624,
      "step": 1118
    },
    {
      "epoch": 0.05135147537974393,
      "grad_norm": 0.5431637167930603,
      "learning_rate": 8.555045871559633e-06,
      "loss": 0.5087,
      "step": 1119
    },
    {
      "epoch": 0.051397365884998394,
      "grad_norm": 0.5035592913627625,
      "learning_rate": 8.562691131498472e-06,
      "loss": 0.43,
      "step": 1120
    },
    {
      "epoch": 0.05144325639025286,
      "grad_norm": 0.5313109755516052,
      "learning_rate": 8.57033639143731e-06,
      "loss": 0.4949,
      "step": 1121
    },
    {
      "epoch": 0.05148914689550732,
      "grad_norm": 0.5008397102355957,
      "learning_rate": 8.577981651376147e-06,
      "loss": 0.5005,
      "step": 1122
    },
    {
      "epoch": 0.051535037400761784,
      "grad_norm": 0.48177602887153625,
      "learning_rate": 8.585626911314986e-06,
      "loss": 0.4194,
      "step": 1123
    },
    {
      "epoch": 0.05158092790601625,
      "grad_norm": 0.48903730511665344,
      "learning_rate": 8.593272171253824e-06,
      "loss": 0.5026,
      "step": 1124
    },
    {
      "epoch": 0.05162681841127071,
      "grad_norm": 0.561829149723053,
      "learning_rate": 8.600917431192661e-06,
      "loss": 0.4368,
      "step": 1125
    },
    {
      "epoch": 0.05167270891652517,
      "grad_norm": 0.508267343044281,
      "learning_rate": 8.608562691131499e-06,
      "loss": 0.4451,
      "step": 1126
    },
    {
      "epoch": 0.051718599421779636,
      "grad_norm": 0.4770166575908661,
      "learning_rate": 8.616207951070336e-06,
      "loss": 0.4039,
      "step": 1127
    },
    {
      "epoch": 0.0517644899270341,
      "grad_norm": 0.458354115486145,
      "learning_rate": 8.623853211009175e-06,
      "loss": 0.379,
      "step": 1128
    },
    {
      "epoch": 0.05181038043228856,
      "grad_norm": 0.5431506037712097,
      "learning_rate": 8.631498470948013e-06,
      "loss": 0.6012,
      "step": 1129
    },
    {
      "epoch": 0.051856270937543025,
      "grad_norm": 0.5154648423194885,
      "learning_rate": 8.63914373088685e-06,
      "loss": 0.4835,
      "step": 1130
    },
    {
      "epoch": 0.05190216144279749,
      "grad_norm": 0.494040310382843,
      "learning_rate": 8.646788990825688e-06,
      "loss": 0.3786,
      "step": 1131
    },
    {
      "epoch": 0.05194805194805195,
      "grad_norm": 0.48181861639022827,
      "learning_rate": 8.654434250764527e-06,
      "loss": 0.4693,
      "step": 1132
    },
    {
      "epoch": 0.05199394245330641,
      "grad_norm": 0.45175430178642273,
      "learning_rate": 8.662079510703364e-06,
      "loss": 0.329,
      "step": 1133
    },
    {
      "epoch": 0.05203983295856087,
      "grad_norm": 0.48362353444099426,
      "learning_rate": 8.669724770642203e-06,
      "loss": 0.4633,
      "step": 1134
    },
    {
      "epoch": 0.052085723463815334,
      "grad_norm": 0.532863199710846,
      "learning_rate": 8.677370030581041e-06,
      "loss": 0.5269,
      "step": 1135
    },
    {
      "epoch": 0.0521316139690698,
      "grad_norm": 0.46966904401779175,
      "learning_rate": 8.685015290519878e-06,
      "loss": 0.3886,
      "step": 1136
    },
    {
      "epoch": 0.05217750447432426,
      "grad_norm": 0.7479985356330872,
      "learning_rate": 8.692660550458716e-06,
      "loss": 0.4961,
      "step": 1137
    },
    {
      "epoch": 0.05222339497957872,
      "grad_norm": 0.49896538257598877,
      "learning_rate": 8.700305810397555e-06,
      "loss": 0.4537,
      "step": 1138
    },
    {
      "epoch": 0.052269285484833186,
      "grad_norm": 0.45643138885498047,
      "learning_rate": 8.707951070336392e-06,
      "loss": 0.3589,
      "step": 1139
    },
    {
      "epoch": 0.05231517599008765,
      "grad_norm": 0.5456955432891846,
      "learning_rate": 8.71559633027523e-06,
      "loss": 0.4759,
      "step": 1140
    },
    {
      "epoch": 0.05236106649534211,
      "grad_norm": 0.47742560505867004,
      "learning_rate": 8.723241590214067e-06,
      "loss": 0.4256,
      "step": 1141
    },
    {
      "epoch": 0.052406957000596575,
      "grad_norm": 0.4854537546634674,
      "learning_rate": 8.730886850152905e-06,
      "loss": 0.4826,
      "step": 1142
    },
    {
      "epoch": 0.05245284750585104,
      "grad_norm": 0.4848230183124542,
      "learning_rate": 8.738532110091744e-06,
      "loss": 0.3266,
      "step": 1143
    },
    {
      "epoch": 0.0524987380111055,
      "grad_norm": 0.4757491648197174,
      "learning_rate": 8.746177370030581e-06,
      "loss": 0.421,
      "step": 1144
    },
    {
      "epoch": 0.052544628516359965,
      "grad_norm": 0.45205485820770264,
      "learning_rate": 8.753822629969419e-06,
      "loss": 0.3446,
      "step": 1145
    },
    {
      "epoch": 0.05259051902161443,
      "grad_norm": 0.514741063117981,
      "learning_rate": 8.761467889908258e-06,
      "loss": 0.4837,
      "step": 1146
    },
    {
      "epoch": 0.05263640952686889,
      "grad_norm": 0.5387755632400513,
      "learning_rate": 8.769113149847095e-06,
      "loss": 0.477,
      "step": 1147
    },
    {
      "epoch": 0.052682300032123354,
      "grad_norm": 0.500912070274353,
      "learning_rate": 8.776758409785935e-06,
      "loss": 0.5217,
      "step": 1148
    },
    {
      "epoch": 0.05272819053737782,
      "grad_norm": 0.5131441354751587,
      "learning_rate": 8.784403669724772e-06,
      "loss": 0.4837,
      "step": 1149
    },
    {
      "epoch": 0.05277408104263228,
      "grad_norm": 0.4618057608604431,
      "learning_rate": 8.79204892966361e-06,
      "loss": 0.3989,
      "step": 1150
    },
    {
      "epoch": 0.05281997154788674,
      "grad_norm": 0.4999014735221863,
      "learning_rate": 8.799694189602447e-06,
      "loss": 0.4462,
      "step": 1151
    },
    {
      "epoch": 0.052865862053141206,
      "grad_norm": 0.5007792115211487,
      "learning_rate": 8.807339449541286e-06,
      "loss": 0.4841,
      "step": 1152
    },
    {
      "epoch": 0.05291175255839567,
      "grad_norm": 0.4365804195404053,
      "learning_rate": 8.814984709480123e-06,
      "loss": 0.3408,
      "step": 1153
    },
    {
      "epoch": 0.05295764306365013,
      "grad_norm": 0.46871837973594666,
      "learning_rate": 8.822629969418961e-06,
      "loss": 0.3868,
      "step": 1154
    },
    {
      "epoch": 0.053003533568904596,
      "grad_norm": 0.4836746156215668,
      "learning_rate": 8.830275229357798e-06,
      "loss": 0.3902,
      "step": 1155
    },
    {
      "epoch": 0.05304942407415906,
      "grad_norm": 0.48468664288520813,
      "learning_rate": 8.837920489296636e-06,
      "loss": 0.412,
      "step": 1156
    },
    {
      "epoch": 0.05309531457941352,
      "grad_norm": 0.49635201692581177,
      "learning_rate": 8.845565749235475e-06,
      "loss": 0.3932,
      "step": 1157
    },
    {
      "epoch": 0.053141205084667985,
      "grad_norm": 0.4699406921863556,
      "learning_rate": 8.853211009174312e-06,
      "loss": 0.4021,
      "step": 1158
    },
    {
      "epoch": 0.05318709558992245,
      "grad_norm": 0.4714055359363556,
      "learning_rate": 8.86085626911315e-06,
      "loss": 0.4062,
      "step": 1159
    },
    {
      "epoch": 0.05323298609517691,
      "grad_norm": 0.48148754239082336,
      "learning_rate": 8.868501529051989e-06,
      "loss": 0.3957,
      "step": 1160
    },
    {
      "epoch": 0.053278876600431374,
      "grad_norm": 0.4738316535949707,
      "learning_rate": 8.876146788990826e-06,
      "loss": 0.3604,
      "step": 1161
    },
    {
      "epoch": 0.05332476710568583,
      "grad_norm": 0.5054348111152649,
      "learning_rate": 8.883792048929664e-06,
      "loss": 0.4233,
      "step": 1162
    },
    {
      "epoch": 0.05337065761094029,
      "grad_norm": 0.48546209931373596,
      "learning_rate": 8.891437308868503e-06,
      "loss": 0.4192,
      "step": 1163
    },
    {
      "epoch": 0.053416548116194756,
      "grad_norm": 0.4655562937259674,
      "learning_rate": 8.89908256880734e-06,
      "loss": 0.3959,
      "step": 1164
    },
    {
      "epoch": 0.05346243862144922,
      "grad_norm": 0.4764060080051422,
      "learning_rate": 8.906727828746178e-06,
      "loss": 0.4215,
      "step": 1165
    },
    {
      "epoch": 0.05350832912670368,
      "grad_norm": 0.5141769051551819,
      "learning_rate": 8.914373088685015e-06,
      "loss": 0.4131,
      "step": 1166
    },
    {
      "epoch": 0.053554219631958146,
      "grad_norm": 0.48103976249694824,
      "learning_rate": 8.922018348623855e-06,
      "loss": 0.3503,
      "step": 1167
    },
    {
      "epoch": 0.05360011013721261,
      "grad_norm": 0.4506582021713257,
      "learning_rate": 8.929663608562692e-06,
      "loss": 0.3688,
      "step": 1168
    },
    {
      "epoch": 0.05364600064246707,
      "grad_norm": 0.5111414194107056,
      "learning_rate": 8.93730886850153e-06,
      "loss": 0.4357,
      "step": 1169
    },
    {
      "epoch": 0.053691891147721535,
      "grad_norm": 0.4592117667198181,
      "learning_rate": 8.944954128440367e-06,
      "loss": 0.3724,
      "step": 1170
    },
    {
      "epoch": 0.053737781652976,
      "grad_norm": 0.45227017998695374,
      "learning_rate": 8.952599388379206e-06,
      "loss": 0.3607,
      "step": 1171
    },
    {
      "epoch": 0.05378367215823046,
      "grad_norm": 0.6733904480934143,
      "learning_rate": 8.960244648318043e-06,
      "loss": 0.5963,
      "step": 1172
    },
    {
      "epoch": 0.053829562663484924,
      "grad_norm": 0.517734169960022,
      "learning_rate": 8.967889908256881e-06,
      "loss": 0.5069,
      "step": 1173
    },
    {
      "epoch": 0.05387545316873939,
      "grad_norm": 0.5107107758522034,
      "learning_rate": 8.97553516819572e-06,
      "loss": 0.4816,
      "step": 1174
    },
    {
      "epoch": 0.05392134367399385,
      "grad_norm": 0.48223477602005005,
      "learning_rate": 8.983180428134558e-06,
      "loss": 0.341,
      "step": 1175
    },
    {
      "epoch": 0.05396723417924831,
      "grad_norm": 0.5055661201477051,
      "learning_rate": 8.990825688073395e-06,
      "loss": 0.4068,
      "step": 1176
    },
    {
      "epoch": 0.05401312468450278,
      "grad_norm": 0.44902464747428894,
      "learning_rate": 8.998470948012234e-06,
      "loss": 0.3577,
      "step": 1177
    },
    {
      "epoch": 0.05405901518975724,
      "grad_norm": 0.47143155336380005,
      "learning_rate": 9.006116207951072e-06,
      "loss": 0.3626,
      "step": 1178
    },
    {
      "epoch": 0.0541049056950117,
      "grad_norm": 0.4975210726261139,
      "learning_rate": 9.013761467889909e-06,
      "loss": 0.3911,
      "step": 1179
    },
    {
      "epoch": 0.054150796200266166,
      "grad_norm": 0.46615511178970337,
      "learning_rate": 9.021406727828746e-06,
      "loss": 0.3652,
      "step": 1180
    },
    {
      "epoch": 0.05419668670552063,
      "grad_norm": 0.5239560008049011,
      "learning_rate": 9.029051987767586e-06,
      "loss": 0.4994,
      "step": 1181
    },
    {
      "epoch": 0.05424257721077509,
      "grad_norm": 0.46002134680747986,
      "learning_rate": 9.036697247706423e-06,
      "loss": 0.346,
      "step": 1182
    },
    {
      "epoch": 0.054288467716029555,
      "grad_norm": 0.5116428136825562,
      "learning_rate": 9.04434250764526e-06,
      "loss": 0.4643,
      "step": 1183
    },
    {
      "epoch": 0.05433435822128402,
      "grad_norm": 0.5150168538093567,
      "learning_rate": 9.051987767584098e-06,
      "loss": 0.4952,
      "step": 1184
    },
    {
      "epoch": 0.05438024872653848,
      "grad_norm": 0.5286275148391724,
      "learning_rate": 9.059633027522935e-06,
      "loss": 0.476,
      "step": 1185
    },
    {
      "epoch": 0.054426139231792944,
      "grad_norm": 0.5359267592430115,
      "learning_rate": 9.067278287461775e-06,
      "loss": 0.544,
      "step": 1186
    },
    {
      "epoch": 0.05447202973704741,
      "grad_norm": 0.5676963329315186,
      "learning_rate": 9.074923547400612e-06,
      "loss": 0.5112,
      "step": 1187
    },
    {
      "epoch": 0.05451792024230187,
      "grad_norm": 0.48615550994873047,
      "learning_rate": 9.08256880733945e-06,
      "loss": 0.4037,
      "step": 1188
    },
    {
      "epoch": 0.054563810747556334,
      "grad_norm": 0.46548187732696533,
      "learning_rate": 9.090214067278289e-06,
      "loss": 0.3574,
      "step": 1189
    },
    {
      "epoch": 0.0546097012528108,
      "grad_norm": 0.552119791507721,
      "learning_rate": 9.097859327217126e-06,
      "loss": 0.4812,
      "step": 1190
    },
    {
      "epoch": 0.05465559175806526,
      "grad_norm": 0.5597963333129883,
      "learning_rate": 9.105504587155965e-06,
      "loss": 0.4233,
      "step": 1191
    },
    {
      "epoch": 0.054701482263319716,
      "grad_norm": 0.5239797234535217,
      "learning_rate": 9.113149847094803e-06,
      "loss": 0.4471,
      "step": 1192
    },
    {
      "epoch": 0.05474737276857418,
      "grad_norm": 0.58037269115448,
      "learning_rate": 9.12079510703364e-06,
      "loss": 0.5173,
      "step": 1193
    },
    {
      "epoch": 0.05479326327382864,
      "grad_norm": 0.5209234356880188,
      "learning_rate": 9.128440366972477e-06,
      "loss": 0.4571,
      "step": 1194
    },
    {
      "epoch": 0.054839153779083105,
      "grad_norm": 0.6197280287742615,
      "learning_rate": 9.136085626911317e-06,
      "loss": 0.444,
      "step": 1195
    },
    {
      "epoch": 0.05488504428433757,
      "grad_norm": 0.49763816595077515,
      "learning_rate": 9.143730886850154e-06,
      "loss": 0.3774,
      "step": 1196
    },
    {
      "epoch": 0.05493093478959203,
      "grad_norm": 0.48769843578338623,
      "learning_rate": 9.151376146788992e-06,
      "loss": 0.4328,
      "step": 1197
    },
    {
      "epoch": 0.054976825294846494,
      "grad_norm": 0.47927042841911316,
      "learning_rate": 9.159021406727829e-06,
      "loss": 0.3704,
      "step": 1198
    },
    {
      "epoch": 0.05502271580010096,
      "grad_norm": 0.47584062814712524,
      "learning_rate": 9.166666666666666e-06,
      "loss": 0.4078,
      "step": 1199
    },
    {
      "epoch": 0.05506860630535542,
      "grad_norm": 0.5033342838287354,
      "learning_rate": 9.174311926605506e-06,
      "loss": 0.4563,
      "step": 1200
    },
    {
      "epoch": 0.055114496810609884,
      "grad_norm": 0.5032165050506592,
      "learning_rate": 9.181957186544343e-06,
      "loss": 0.3943,
      "step": 1201
    },
    {
      "epoch": 0.05516038731586435,
      "grad_norm": 0.4666723310947418,
      "learning_rate": 9.18960244648318e-06,
      "loss": 0.4284,
      "step": 1202
    },
    {
      "epoch": 0.05520627782111881,
      "grad_norm": 0.543606162071228,
      "learning_rate": 9.19724770642202e-06,
      "loss": 0.472,
      "step": 1203
    },
    {
      "epoch": 0.05525216832637327,
      "grad_norm": 0.46025291085243225,
      "learning_rate": 9.204892966360857e-06,
      "loss": 0.3949,
      "step": 1204
    },
    {
      "epoch": 0.055298058831627736,
      "grad_norm": 0.4508424997329712,
      "learning_rate": 9.212538226299696e-06,
      "loss": 0.3562,
      "step": 1205
    },
    {
      "epoch": 0.0553439493368822,
      "grad_norm": 0.4536934494972229,
      "learning_rate": 9.220183486238534e-06,
      "loss": 0.35,
      "step": 1206
    },
    {
      "epoch": 0.05538983984213666,
      "grad_norm": 0.502681314945221,
      "learning_rate": 9.227828746177371e-06,
      "loss": 0.431,
      "step": 1207
    },
    {
      "epoch": 0.055435730347391125,
      "grad_norm": 0.4763748049736023,
      "learning_rate": 9.235474006116209e-06,
      "loss": 0.4265,
      "step": 1208
    },
    {
      "epoch": 0.05548162085264559,
      "grad_norm": 0.4737585186958313,
      "learning_rate": 9.243119266055046e-06,
      "loss": 0.4079,
      "step": 1209
    },
    {
      "epoch": 0.05552751135790005,
      "grad_norm": 0.5160509347915649,
      "learning_rate": 9.250764525993885e-06,
      "loss": 0.4793,
      "step": 1210
    },
    {
      "epoch": 0.055573401863154515,
      "grad_norm": 0.44092419743537903,
      "learning_rate": 9.258409785932723e-06,
      "loss": 0.3411,
      "step": 1211
    },
    {
      "epoch": 0.05561929236840898,
      "grad_norm": 0.5281226634979248,
      "learning_rate": 9.26605504587156e-06,
      "loss": 0.4619,
      "step": 1212
    },
    {
      "epoch": 0.05566518287366344,
      "grad_norm": 0.5943772792816162,
      "learning_rate": 9.273700305810397e-06,
      "loss": 0.3918,
      "step": 1213
    },
    {
      "epoch": 0.055711073378917904,
      "grad_norm": 0.49317190051078796,
      "learning_rate": 9.281345565749235e-06,
      "loss": 0.4246,
      "step": 1214
    },
    {
      "epoch": 0.05575696388417237,
      "grad_norm": 0.496288925409317,
      "learning_rate": 9.288990825688074e-06,
      "loss": 0.4581,
      "step": 1215
    },
    {
      "epoch": 0.05580285438942683,
      "grad_norm": 0.4898359775543213,
      "learning_rate": 9.296636085626912e-06,
      "loss": 0.3975,
      "step": 1216
    },
    {
      "epoch": 0.05584874489468129,
      "grad_norm": 0.5074936747550964,
      "learning_rate": 9.30428134556575e-06,
      "loss": 0.422,
      "step": 1217
    },
    {
      "epoch": 0.055894635399935756,
      "grad_norm": 0.6084580421447754,
      "learning_rate": 9.311926605504588e-06,
      "loss": 0.5219,
      "step": 1218
    },
    {
      "epoch": 0.05594052590519022,
      "grad_norm": 0.4942367970943451,
      "learning_rate": 9.319571865443426e-06,
      "loss": 0.5041,
      "step": 1219
    },
    {
      "epoch": 0.05598641641044468,
      "grad_norm": 0.5235260725021362,
      "learning_rate": 9.327217125382265e-06,
      "loss": 0.4732,
      "step": 1220
    },
    {
      "epoch": 0.05603230691569914,
      "grad_norm": 0.46256786584854126,
      "learning_rate": 9.334862385321102e-06,
      "loss": 0.3409,
      "step": 1221
    },
    {
      "epoch": 0.0560781974209536,
      "grad_norm": 0.49697116017341614,
      "learning_rate": 9.34250764525994e-06,
      "loss": 0.3842,
      "step": 1222
    },
    {
      "epoch": 0.056124087926208065,
      "grad_norm": 0.5227659344673157,
      "learning_rate": 9.350152905198777e-06,
      "loss": 0.47,
      "step": 1223
    },
    {
      "epoch": 0.05616997843146253,
      "grad_norm": 0.5589193105697632,
      "learning_rate": 9.357798165137616e-06,
      "loss": 0.5295,
      "step": 1224
    },
    {
      "epoch": 0.05621586893671699,
      "grad_norm": 0.44184815883636475,
      "learning_rate": 9.365443425076454e-06,
      "loss": 0.3563,
      "step": 1225
    },
    {
      "epoch": 0.056261759441971454,
      "grad_norm": 0.48488152027130127,
      "learning_rate": 9.373088685015291e-06,
      "loss": 0.4578,
      "step": 1226
    },
    {
      "epoch": 0.05630764994722592,
      "grad_norm": 0.598098874092102,
      "learning_rate": 9.380733944954129e-06,
      "loss": 0.4705,
      "step": 1227
    },
    {
      "epoch": 0.05635354045248038,
      "grad_norm": 0.48798874020576477,
      "learning_rate": 9.388379204892966e-06,
      "loss": 0.3858,
      "step": 1228
    },
    {
      "epoch": 0.05639943095773484,
      "grad_norm": 0.5060775876045227,
      "learning_rate": 9.396024464831805e-06,
      "loss": 0.4714,
      "step": 1229
    },
    {
      "epoch": 0.056445321462989306,
      "grad_norm": 0.44378143548965454,
      "learning_rate": 9.403669724770643e-06,
      "loss": 0.358,
      "step": 1230
    },
    {
      "epoch": 0.05649121196824377,
      "grad_norm": 0.472260981798172,
      "learning_rate": 9.411314984709482e-06,
      "loss": 0.3976,
      "step": 1231
    },
    {
      "epoch": 0.05653710247349823,
      "grad_norm": 0.5407238602638245,
      "learning_rate": 9.41896024464832e-06,
      "loss": 0.44,
      "step": 1232
    },
    {
      "epoch": 0.056582992978752696,
      "grad_norm": 0.5374550819396973,
      "learning_rate": 9.426605504587157e-06,
      "loss": 0.5074,
      "step": 1233
    },
    {
      "epoch": 0.05662888348400716,
      "grad_norm": 0.48143428564071655,
      "learning_rate": 9.434250764525996e-06,
      "loss": 0.4651,
      "step": 1234
    },
    {
      "epoch": 0.05667477398926162,
      "grad_norm": 0.5040050148963928,
      "learning_rate": 9.441896024464833e-06,
      "loss": 0.3929,
      "step": 1235
    },
    {
      "epoch": 0.056720664494516085,
      "grad_norm": 0.4760933518409729,
      "learning_rate": 9.44954128440367e-06,
      "loss": 0.3841,
      "step": 1236
    },
    {
      "epoch": 0.05676655499977055,
      "grad_norm": 0.48517096042633057,
      "learning_rate": 9.457186544342508e-06,
      "loss": 0.3814,
      "step": 1237
    },
    {
      "epoch": 0.05681244550502501,
      "grad_norm": 0.4878256320953369,
      "learning_rate": 9.464831804281346e-06,
      "loss": 0.3942,
      "step": 1238
    },
    {
      "epoch": 0.056858336010279474,
      "grad_norm": 0.48731502890586853,
      "learning_rate": 9.472477064220185e-06,
      "loss": 0.4196,
      "step": 1239
    },
    {
      "epoch": 0.05690422651553394,
      "grad_norm": 0.572546124458313,
      "learning_rate": 9.480122324159022e-06,
      "loss": 0.4353,
      "step": 1240
    },
    {
      "epoch": 0.0569501170207884,
      "grad_norm": 0.5423152446746826,
      "learning_rate": 9.48776758409786e-06,
      "loss": 0.5241,
      "step": 1241
    },
    {
      "epoch": 0.05699600752604286,
      "grad_norm": 0.4827650785446167,
      "learning_rate": 9.495412844036697e-06,
      "loss": 0.356,
      "step": 1242
    },
    {
      "epoch": 0.057041898031297326,
      "grad_norm": 0.5505901575088501,
      "learning_rate": 9.503058103975536e-06,
      "loss": 0.5216,
      "step": 1243
    },
    {
      "epoch": 0.05708778853655179,
      "grad_norm": 0.4636831283569336,
      "learning_rate": 9.510703363914374e-06,
      "loss": 0.3311,
      "step": 1244
    },
    {
      "epoch": 0.05713367904180625,
      "grad_norm": 0.4764443039894104,
      "learning_rate": 9.518348623853211e-06,
      "loss": 0.3514,
      "step": 1245
    },
    {
      "epoch": 0.057179569547060716,
      "grad_norm": 0.47547975182533264,
      "learning_rate": 9.52599388379205e-06,
      "loss": 0.4474,
      "step": 1246
    },
    {
      "epoch": 0.05722546005231518,
      "grad_norm": 0.4269947409629822,
      "learning_rate": 9.533639143730888e-06,
      "loss": 0.3651,
      "step": 1247
    },
    {
      "epoch": 0.05727135055756964,
      "grad_norm": 0.4777385890483856,
      "learning_rate": 9.541284403669727e-06,
      "loss": 0.3895,
      "step": 1248
    },
    {
      "epoch": 0.057317241062824105,
      "grad_norm": 0.5050618052482605,
      "learning_rate": 9.548929663608564e-06,
      "loss": 0.4435,
      "step": 1249
    },
    {
      "epoch": 0.05736313156807856,
      "grad_norm": 0.4309498965740204,
      "learning_rate": 9.556574923547402e-06,
      "loss": 0.3661,
      "step": 1250
    },
    {
      "epoch": 0.057409022073333024,
      "grad_norm": 0.4313931465148926,
      "learning_rate": 9.56422018348624e-06,
      "loss": 0.3227,
      "step": 1251
    },
    {
      "epoch": 0.05745491257858749,
      "grad_norm": 0.5016030073165894,
      "learning_rate": 9.571865443425077e-06,
      "loss": 0.4767,
      "step": 1252
    },
    {
      "epoch": 0.05750080308384195,
      "grad_norm": 0.49609053134918213,
      "learning_rate": 9.579510703363916e-06,
      "loss": 0.4073,
      "step": 1253
    },
    {
      "epoch": 0.05754669358909641,
      "grad_norm": 0.42334234714508057,
      "learning_rate": 9.587155963302753e-06,
      "loss": 0.3122,
      "step": 1254
    },
    {
      "epoch": 0.05759258409435088,
      "grad_norm": 0.45502781867980957,
      "learning_rate": 9.59480122324159e-06,
      "loss": 0.4212,
      "step": 1255
    },
    {
      "epoch": 0.05763847459960534,
      "grad_norm": 0.47657495737075806,
      "learning_rate": 9.602446483180428e-06,
      "loss": 0.3769,
      "step": 1256
    },
    {
      "epoch": 0.0576843651048598,
      "grad_norm": 0.4626166522502899,
      "learning_rate": 9.610091743119267e-06,
      "loss": 0.3816,
      "step": 1257
    },
    {
      "epoch": 0.057730255610114266,
      "grad_norm": 0.4420531392097473,
      "learning_rate": 9.617737003058105e-06,
      "loss": 0.3481,
      "step": 1258
    },
    {
      "epoch": 0.05777614611536873,
      "grad_norm": 0.47691357135772705,
      "learning_rate": 9.625382262996942e-06,
      "loss": 0.4452,
      "step": 1259
    },
    {
      "epoch": 0.05782203662062319,
      "grad_norm": 0.5204956531524658,
      "learning_rate": 9.633027522935781e-06,
      "loss": 0.505,
      "step": 1260
    },
    {
      "epoch": 0.057867927125877655,
      "grad_norm": 0.4837752878665924,
      "learning_rate": 9.640672782874619e-06,
      "loss": 0.3862,
      "step": 1261
    },
    {
      "epoch": 0.05791381763113212,
      "grad_norm": 0.49503791332244873,
      "learning_rate": 9.648318042813456e-06,
      "loss": 0.3766,
      "step": 1262
    },
    {
      "epoch": 0.05795970813638658,
      "grad_norm": 0.46857956051826477,
      "learning_rate": 9.655963302752295e-06,
      "loss": 0.3909,
      "step": 1263
    },
    {
      "epoch": 0.058005598641641044,
      "grad_norm": 0.49671974778175354,
      "learning_rate": 9.663608562691133e-06,
      "loss": 0.404,
      "step": 1264
    },
    {
      "epoch": 0.05805148914689551,
      "grad_norm": 0.4362523555755615,
      "learning_rate": 9.67125382262997e-06,
      "loss": 0.3093,
      "step": 1265
    },
    {
      "epoch": 0.05809737965214997,
      "grad_norm": 0.4402255713939667,
      "learning_rate": 9.678899082568808e-06,
      "loss": 0.3176,
      "step": 1266
    },
    {
      "epoch": 0.058143270157404434,
      "grad_norm": 0.516869843006134,
      "learning_rate": 9.686544342507645e-06,
      "loss": 0.4923,
      "step": 1267
    },
    {
      "epoch": 0.0581891606626589,
      "grad_norm": 0.5135446190834045,
      "learning_rate": 9.694189602446484e-06,
      "loss": 0.4981,
      "step": 1268
    },
    {
      "epoch": 0.05823505116791336,
      "grad_norm": 0.4470340609550476,
      "learning_rate": 9.701834862385322e-06,
      "loss": 0.3615,
      "step": 1269
    },
    {
      "epoch": 0.05828094167316782,
      "grad_norm": 0.43490803241729736,
      "learning_rate": 9.709480122324159e-06,
      "loss": 0.3296,
      "step": 1270
    },
    {
      "epoch": 0.058326832178422286,
      "grad_norm": 0.4599528908729553,
      "learning_rate": 9.717125382262997e-06,
      "loss": 0.4019,
      "step": 1271
    },
    {
      "epoch": 0.05837272268367675,
      "grad_norm": 0.4902098476886749,
      "learning_rate": 9.724770642201836e-06,
      "loss": 0.4476,
      "step": 1272
    },
    {
      "epoch": 0.05841861318893121,
      "grad_norm": 0.458825945854187,
      "learning_rate": 9.732415902140673e-06,
      "loss": 0.3687,
      "step": 1273
    },
    {
      "epoch": 0.058464503694185675,
      "grad_norm": 0.497462660074234,
      "learning_rate": 9.740061162079512e-06,
      "loss": 0.4179,
      "step": 1274
    },
    {
      "epoch": 0.05851039419944014,
      "grad_norm": 0.49030202627182007,
      "learning_rate": 9.74770642201835e-06,
      "loss": 0.4212,
      "step": 1275
    },
    {
      "epoch": 0.0585562847046946,
      "grad_norm": 0.5183060169219971,
      "learning_rate": 9.755351681957187e-06,
      "loss": 0.4335,
      "step": 1276
    },
    {
      "epoch": 0.058602175209949064,
      "grad_norm": 0.4406938850879669,
      "learning_rate": 9.762996941896026e-06,
      "loss": 0.3798,
      "step": 1277
    },
    {
      "epoch": 0.05864806571520353,
      "grad_norm": 0.49757298827171326,
      "learning_rate": 9.770642201834864e-06,
      "loss": 0.4439,
      "step": 1278
    },
    {
      "epoch": 0.05869395622045799,
      "grad_norm": 0.5338001847267151,
      "learning_rate": 9.778287461773701e-06,
      "loss": 0.4791,
      "step": 1279
    },
    {
      "epoch": 0.05873984672571245,
      "grad_norm": 0.451894611120224,
      "learning_rate": 9.785932721712539e-06,
      "loss": 0.3487,
      "step": 1280
    },
    {
      "epoch": 0.05878573723096691,
      "grad_norm": 0.5234604477882385,
      "learning_rate": 9.793577981651376e-06,
      "loss": 0.4761,
      "step": 1281
    },
    {
      "epoch": 0.05883162773622137,
      "grad_norm": 0.46757015585899353,
      "learning_rate": 9.801223241590215e-06,
      "loss": 0.3682,
      "step": 1282
    },
    {
      "epoch": 0.058877518241475836,
      "grad_norm": 0.45961517095565796,
      "learning_rate": 9.808868501529053e-06,
      "loss": 0.3523,
      "step": 1283
    },
    {
      "epoch": 0.0589234087467303,
      "grad_norm": 0.4827962815761566,
      "learning_rate": 9.81651376146789e-06,
      "loss": 0.3985,
      "step": 1284
    },
    {
      "epoch": 0.05896929925198476,
      "grad_norm": 0.48937731981277466,
      "learning_rate": 9.824159021406728e-06,
      "loss": 0.4421,
      "step": 1285
    },
    {
      "epoch": 0.059015189757239225,
      "grad_norm": 0.48009026050567627,
      "learning_rate": 9.831804281345567e-06,
      "loss": 0.4754,
      "step": 1286
    },
    {
      "epoch": 0.05906108026249369,
      "grad_norm": 0.4757213592529297,
      "learning_rate": 9.839449541284404e-06,
      "loss": 0.4376,
      "step": 1287
    },
    {
      "epoch": 0.05910697076774815,
      "grad_norm": 0.5103703141212463,
      "learning_rate": 9.847094801223243e-06,
      "loss": 0.4093,
      "step": 1288
    },
    {
      "epoch": 0.059152861273002615,
      "grad_norm": 0.4713591933250427,
      "learning_rate": 9.85474006116208e-06,
      "loss": 0.3903,
      "step": 1289
    },
    {
      "epoch": 0.05919875177825708,
      "grad_norm": 0.4589911699295044,
      "learning_rate": 9.862385321100918e-06,
      "loss": 0.4012,
      "step": 1290
    },
    {
      "epoch": 0.05924464228351154,
      "grad_norm": 0.4538578391075134,
      "learning_rate": 9.870030581039756e-06,
      "loss": 0.3765,
      "step": 1291
    },
    {
      "epoch": 0.059290532788766004,
      "grad_norm": 0.5259283781051636,
      "learning_rate": 9.877675840978595e-06,
      "loss": 0.5194,
      "step": 1292
    },
    {
      "epoch": 0.05933642329402047,
      "grad_norm": 0.47437596321105957,
      "learning_rate": 9.885321100917432e-06,
      "loss": 0.4035,
      "step": 1293
    },
    {
      "epoch": 0.05938231379927493,
      "grad_norm": 0.4400242865085602,
      "learning_rate": 9.89296636085627e-06,
      "loss": 0.3607,
      "step": 1294
    },
    {
      "epoch": 0.05942820430452939,
      "grad_norm": 0.5560067892074585,
      "learning_rate": 9.900611620795107e-06,
      "loss": 0.4591,
      "step": 1295
    },
    {
      "epoch": 0.059474094809783856,
      "grad_norm": 0.5343872308731079,
      "learning_rate": 9.908256880733946e-06,
      "loss": 0.4959,
      "step": 1296
    },
    {
      "epoch": 0.05951998531503832,
      "grad_norm": 0.5153381824493408,
      "learning_rate": 9.915902140672784e-06,
      "loss": 0.4717,
      "step": 1297
    },
    {
      "epoch": 0.05956587582029278,
      "grad_norm": 0.44272804260253906,
      "learning_rate": 9.923547400611621e-06,
      "loss": 0.3202,
      "step": 1298
    },
    {
      "epoch": 0.059611766325547245,
      "grad_norm": 0.4977368712425232,
      "learning_rate": 9.931192660550459e-06,
      "loss": 0.4114,
      "step": 1299
    },
    {
      "epoch": 0.05965765683080171,
      "grad_norm": 0.46729445457458496,
      "learning_rate": 9.938837920489298e-06,
      "loss": 0.4007,
      "step": 1300
    },
    {
      "epoch": 0.05970354733605617,
      "grad_norm": 0.4667612910270691,
      "learning_rate": 9.946483180428135e-06,
      "loss": 0.3873,
      "step": 1301
    },
    {
      "epoch": 0.059749437841310635,
      "grad_norm": 0.5385662317276001,
      "learning_rate": 9.954128440366973e-06,
      "loss": 0.4531,
      "step": 1302
    },
    {
      "epoch": 0.0597953283465651,
      "grad_norm": 0.4886770248413086,
      "learning_rate": 9.961773700305812e-06,
      "loss": 0.4461,
      "step": 1303
    },
    {
      "epoch": 0.05984121885181956,
      "grad_norm": 0.4952675700187683,
      "learning_rate": 9.96941896024465e-06,
      "loss": 0.3985,
      "step": 1304
    },
    {
      "epoch": 0.059887109357074024,
      "grad_norm": 0.46684184670448303,
      "learning_rate": 9.977064220183487e-06,
      "loss": 0.3868,
      "step": 1305
    },
    {
      "epoch": 0.05993299986232849,
      "grad_norm": 0.453785240650177,
      "learning_rate": 9.984709480122326e-06,
      "loss": 0.3615,
      "step": 1306
    },
    {
      "epoch": 0.05997889036758295,
      "grad_norm": 0.4815285801887512,
      "learning_rate": 9.992354740061163e-06,
      "loss": 0.3839,
      "step": 1307
    },
    {
      "epoch": 0.06002478087283741,
      "grad_norm": 0.4813125431537628,
      "learning_rate": 1e-05,
      "loss": 0.4371,
      "step": 1308
    },
    {
      "epoch": 0.06007067137809187,
      "grad_norm": 0.5038076639175415,
      "learning_rate": 9.99999999398829e-06,
      "loss": 0.4438,
      "step": 1309
    },
    {
      "epoch": 0.06011656188334633,
      "grad_norm": 0.4282680153846741,
      "learning_rate": 9.999999975953156e-06,
      "loss": 0.3444,
      "step": 1310
    },
    {
      "epoch": 0.060162452388600796,
      "grad_norm": 0.44697362184524536,
      "learning_rate": 9.999999945894602e-06,
      "loss": 0.305,
      "step": 1311
    },
    {
      "epoch": 0.06020834289385526,
      "grad_norm": 0.484749972820282,
      "learning_rate": 9.999999903812625e-06,
      "loss": 0.4041,
      "step": 1312
    },
    {
      "epoch": 0.06025423339910972,
      "grad_norm": 0.4960935115814209,
      "learning_rate": 9.999999849707227e-06,
      "loss": 0.4118,
      "step": 1313
    },
    {
      "epoch": 0.060300123904364185,
      "grad_norm": 0.4719673991203308,
      "learning_rate": 9.999999783578407e-06,
      "loss": 0.4299,
      "step": 1314
    },
    {
      "epoch": 0.06034601440961865,
      "grad_norm": 0.4409143924713135,
      "learning_rate": 9.999999705426165e-06,
      "loss": 0.3409,
      "step": 1315
    },
    {
      "epoch": 0.06039190491487311,
      "grad_norm": 0.5070991516113281,
      "learning_rate": 9.999999615250502e-06,
      "loss": 0.4469,
      "step": 1316
    },
    {
      "epoch": 0.060437795420127574,
      "grad_norm": 0.47174718976020813,
      "learning_rate": 9.99999951305142e-06,
      "loss": 0.4125,
      "step": 1317
    },
    {
      "epoch": 0.06048368592538204,
      "grad_norm": 0.47255584597587585,
      "learning_rate": 9.999999398828914e-06,
      "loss": 0.4568,
      "step": 1318
    },
    {
      "epoch": 0.0605295764306365,
      "grad_norm": 0.45447489619255066,
      "learning_rate": 9.99999927258299e-06,
      "loss": 0.3368,
      "step": 1319
    },
    {
      "epoch": 0.06057546693589096,
      "grad_norm": 0.45103153586387634,
      "learning_rate": 9.999999134313645e-06,
      "loss": 0.3629,
      "step": 1320
    },
    {
      "epoch": 0.060621357441145426,
      "grad_norm": 0.4483513832092285,
      "learning_rate": 9.99999898402088e-06,
      "loss": 0.3971,
      "step": 1321
    },
    {
      "epoch": 0.06066724794639989,
      "grad_norm": 0.47159498929977417,
      "learning_rate": 9.999998821704696e-06,
      "loss": 0.4108,
      "step": 1322
    },
    {
      "epoch": 0.06071313845165435,
      "grad_norm": 0.4768999516963959,
      "learning_rate": 9.999998647365093e-06,
      "loss": 0.3864,
      "step": 1323
    },
    {
      "epoch": 0.060759028956908816,
      "grad_norm": 0.48006561398506165,
      "learning_rate": 9.99999846100207e-06,
      "loss": 0.3576,
      "step": 1324
    },
    {
      "epoch": 0.06080491946216328,
      "grad_norm": 0.5075549483299255,
      "learning_rate": 9.999998262615628e-06,
      "loss": 0.4452,
      "step": 1325
    },
    {
      "epoch": 0.06085080996741774,
      "grad_norm": 0.4682963788509369,
      "learning_rate": 9.99999805220577e-06,
      "loss": 0.3618,
      "step": 1326
    },
    {
      "epoch": 0.060896700472672205,
      "grad_norm": 0.49834227561950684,
      "learning_rate": 9.999997829772494e-06,
      "loss": 0.4272,
      "step": 1327
    },
    {
      "epoch": 0.06094259097792667,
      "grad_norm": 0.5462979674339294,
      "learning_rate": 9.999997595315802e-06,
      "loss": 0.4809,
      "step": 1328
    },
    {
      "epoch": 0.06098848148318113,
      "grad_norm": 0.4988521933555603,
      "learning_rate": 9.999997348835694e-06,
      "loss": 0.4348,
      "step": 1329
    },
    {
      "epoch": 0.061034371988435594,
      "grad_norm": 0.4651349186897278,
      "learning_rate": 9.99999709033217e-06,
      "loss": 0.3178,
      "step": 1330
    },
    {
      "epoch": 0.06108026249369006,
      "grad_norm": 0.4642215371131897,
      "learning_rate": 9.99999681980523e-06,
      "loss": 0.3708,
      "step": 1331
    },
    {
      "epoch": 0.06112615299894452,
      "grad_norm": 0.4974886476993561,
      "learning_rate": 9.999996537254877e-06,
      "loss": 0.4474,
      "step": 1332
    },
    {
      "epoch": 0.061172043504198984,
      "grad_norm": 0.5218479633331299,
      "learning_rate": 9.999996242681109e-06,
      "loss": 0.5018,
      "step": 1333
    },
    {
      "epoch": 0.06121793400945345,
      "grad_norm": 0.4917214512825012,
      "learning_rate": 9.99999593608393e-06,
      "loss": 0.4078,
      "step": 1334
    },
    {
      "epoch": 0.06126382451470791,
      "grad_norm": 0.5320242643356323,
      "learning_rate": 9.999995617463338e-06,
      "loss": 0.5073,
      "step": 1335
    },
    {
      "epoch": 0.06130971501996237,
      "grad_norm": 0.5273686051368713,
      "learning_rate": 9.999995286819334e-06,
      "loss": 0.4865,
      "step": 1336
    },
    {
      "epoch": 0.061355605525216836,
      "grad_norm": 0.5050492882728577,
      "learning_rate": 9.99999494415192e-06,
      "loss": 0.3848,
      "step": 1337
    },
    {
      "epoch": 0.0614014960304713,
      "grad_norm": 0.5004304051399231,
      "learning_rate": 9.999994589461097e-06,
      "loss": 0.5303,
      "step": 1338
    },
    {
      "epoch": 0.061447386535725755,
      "grad_norm": 0.59356689453125,
      "learning_rate": 9.999994222746862e-06,
      "loss": 0.4351,
      "step": 1339
    },
    {
      "epoch": 0.06149327704098022,
      "grad_norm": 0.47034984827041626,
      "learning_rate": 9.999993844009221e-06,
      "loss": 0.3963,
      "step": 1340
    },
    {
      "epoch": 0.06153916754623468,
      "grad_norm": 0.47136425971984863,
      "learning_rate": 9.999993453248174e-06,
      "loss": 0.4423,
      "step": 1341
    },
    {
      "epoch": 0.061585058051489144,
      "grad_norm": 0.5834726095199585,
      "learning_rate": 9.99999305046372e-06,
      "loss": 0.5866,
      "step": 1342
    },
    {
      "epoch": 0.06163094855674361,
      "grad_norm": 0.4277462661266327,
      "learning_rate": 9.999992635655859e-06,
      "loss": 0.3772,
      "step": 1343
    },
    {
      "epoch": 0.06167683906199807,
      "grad_norm": 0.4777684211730957,
      "learning_rate": 9.999992208824596e-06,
      "loss": 0.4126,
      "step": 1344
    },
    {
      "epoch": 0.061722729567252534,
      "grad_norm": 0.520283043384552,
      "learning_rate": 9.999991769969928e-06,
      "loss": 0.5277,
      "step": 1345
    },
    {
      "epoch": 0.061768620072507,
      "grad_norm": 0.5171157717704773,
      "learning_rate": 9.99999131909186e-06,
      "loss": 0.4842,
      "step": 1346
    },
    {
      "epoch": 0.06181451057776146,
      "grad_norm": 0.5312116742134094,
      "learning_rate": 9.999990856190388e-06,
      "loss": 0.5218,
      "step": 1347
    },
    {
      "epoch": 0.06186040108301592,
      "grad_norm": 0.4500662386417389,
      "learning_rate": 9.999990381265518e-06,
      "loss": 0.345,
      "step": 1348
    },
    {
      "epoch": 0.061906291588270386,
      "grad_norm": 0.4788573682308197,
      "learning_rate": 9.99998989431725e-06,
      "loss": 0.4695,
      "step": 1349
    },
    {
      "epoch": 0.06195218209352485,
      "grad_norm": 0.520769476890564,
      "learning_rate": 9.999989395345583e-06,
      "loss": 0.4041,
      "step": 1350
    },
    {
      "epoch": 0.06199807259877931,
      "grad_norm": 0.4782469570636749,
      "learning_rate": 9.99998888435052e-06,
      "loss": 0.3781,
      "step": 1351
    },
    {
      "epoch": 0.062043963104033775,
      "grad_norm": 0.46850380301475525,
      "learning_rate": 9.999988361332063e-06,
      "loss": 0.3572,
      "step": 1352
    },
    {
      "epoch": 0.06208985360928824,
      "grad_norm": 0.4841806888580322,
      "learning_rate": 9.99998782629021e-06,
      "loss": 0.4375,
      "step": 1353
    },
    {
      "epoch": 0.0621357441145427,
      "grad_norm": 0.4694962203502655,
      "learning_rate": 9.999987279224964e-06,
      "loss": 0.4303,
      "step": 1354
    },
    {
      "epoch": 0.062181634619797164,
      "grad_norm": 0.44295835494995117,
      "learning_rate": 9.999986720136328e-06,
      "loss": 0.3407,
      "step": 1355
    },
    {
      "epoch": 0.06222752512505163,
      "grad_norm": 0.4706135392189026,
      "learning_rate": 9.999986149024301e-06,
      "loss": 0.449,
      "step": 1356
    },
    {
      "epoch": 0.06227341563030609,
      "grad_norm": 0.5196651816368103,
      "learning_rate": 9.999985565888885e-06,
      "loss": 0.5021,
      "step": 1357
    },
    {
      "epoch": 0.062319306135560554,
      "grad_norm": 0.5165014863014221,
      "learning_rate": 9.999984970730083e-06,
      "loss": 0.4387,
      "step": 1358
    },
    {
      "epoch": 0.06236519664081502,
      "grad_norm": 0.5116204023361206,
      "learning_rate": 9.999984363547895e-06,
      "loss": 0.4128,
      "step": 1359
    },
    {
      "epoch": 0.06241108714606948,
      "grad_norm": 0.47467002272605896,
      "learning_rate": 9.999983744342323e-06,
      "loss": 0.3345,
      "step": 1360
    },
    {
      "epoch": 0.06245697765132394,
      "grad_norm": 0.49090775847435,
      "learning_rate": 9.999983113113367e-06,
      "loss": 0.4033,
      "step": 1361
    },
    {
      "epoch": 0.0625028681565784,
      "grad_norm": 0.5150254368782043,
      "learning_rate": 9.99998246986103e-06,
      "loss": 0.4605,
      "step": 1362
    },
    {
      "epoch": 0.06254875866183286,
      "grad_norm": 0.48796579241752625,
      "learning_rate": 9.999981814585314e-06,
      "loss": 0.4349,
      "step": 1363
    },
    {
      "epoch": 0.06259464916708733,
      "grad_norm": 0.5245981812477112,
      "learning_rate": 9.99998114728622e-06,
      "loss": 0.4697,
      "step": 1364
    },
    {
      "epoch": 0.06264053967234179,
      "grad_norm": 0.4576440751552582,
      "learning_rate": 9.999980467963748e-06,
      "loss": 0.4114,
      "step": 1365
    },
    {
      "epoch": 0.06268643017759626,
      "grad_norm": 0.6070294976234436,
      "learning_rate": 9.999979776617902e-06,
      "loss": 0.4911,
      "step": 1366
    },
    {
      "epoch": 0.06273232068285071,
      "grad_norm": 0.4996192753314972,
      "learning_rate": 9.999979073248682e-06,
      "loss": 0.525,
      "step": 1367
    },
    {
      "epoch": 0.06277821118810518,
      "grad_norm": 0.4975182116031647,
      "learning_rate": 9.999978357856091e-06,
      "loss": 0.4964,
      "step": 1368
    },
    {
      "epoch": 0.06282410169335964,
      "grad_norm": 0.4779447019100189,
      "learning_rate": 9.99997763044013e-06,
      "loss": 0.342,
      "step": 1369
    },
    {
      "epoch": 0.06286999219861411,
      "grad_norm": 0.4856947362422943,
      "learning_rate": 9.9999768910008e-06,
      "loss": 0.3754,
      "step": 1370
    },
    {
      "epoch": 0.06291588270386857,
      "grad_norm": 0.4655861556529999,
      "learning_rate": 9.999976139538104e-06,
      "loss": 0.4096,
      "step": 1371
    },
    {
      "epoch": 0.06296177320912304,
      "grad_norm": 0.4430520832538605,
      "learning_rate": 9.999975376052045e-06,
      "loss": 0.3258,
      "step": 1372
    },
    {
      "epoch": 0.0630076637143775,
      "grad_norm": 0.4377870559692383,
      "learning_rate": 9.99997460054262e-06,
      "loss": 0.3093,
      "step": 1373
    },
    {
      "epoch": 0.06305355421963196,
      "grad_norm": 0.4862578809261322,
      "learning_rate": 9.999973813009838e-06,
      "loss": 0.4123,
      "step": 1374
    },
    {
      "epoch": 0.06309944472488642,
      "grad_norm": 0.4473819434642792,
      "learning_rate": 9.999973013453694e-06,
      "loss": 0.3162,
      "step": 1375
    },
    {
      "epoch": 0.06314533523014089,
      "grad_norm": 0.4855145215988159,
      "learning_rate": 9.999972201874193e-06,
      "loss": 0.4235,
      "step": 1376
    },
    {
      "epoch": 0.06319122573539535,
      "grad_norm": 0.46401524543762207,
      "learning_rate": 9.99997137827134e-06,
      "loss": 0.3717,
      "step": 1377
    },
    {
      "epoch": 0.06323711624064982,
      "grad_norm": 0.5210104584693909,
      "learning_rate": 9.99997054264513e-06,
      "loss": 0.482,
      "step": 1378
    },
    {
      "epoch": 0.06328300674590427,
      "grad_norm": 0.4522438049316406,
      "learning_rate": 9.999969694995572e-06,
      "loss": 0.3534,
      "step": 1379
    },
    {
      "epoch": 0.06332889725115874,
      "grad_norm": 0.49172359704971313,
      "learning_rate": 9.999968835322662e-06,
      "loss": 0.5015,
      "step": 1380
    },
    {
      "epoch": 0.0633747877564132,
      "grad_norm": 0.49448615312576294,
      "learning_rate": 9.999967963626406e-06,
      "loss": 0.3983,
      "step": 1381
    },
    {
      "epoch": 0.06342067826166767,
      "grad_norm": 0.4784165620803833,
      "learning_rate": 9.999967079906807e-06,
      "loss": 0.4444,
      "step": 1382
    },
    {
      "epoch": 0.06346656876692212,
      "grad_norm": 0.4576626121997833,
      "learning_rate": 9.999966184163865e-06,
      "loss": 0.4187,
      "step": 1383
    },
    {
      "epoch": 0.06351245927217658,
      "grad_norm": 0.4514203369617462,
      "learning_rate": 9.999965276397581e-06,
      "loss": 0.3193,
      "step": 1384
    },
    {
      "epoch": 0.06355834977743105,
      "grad_norm": 0.4750826060771942,
      "learning_rate": 9.999964356607958e-06,
      "loss": 0.3896,
      "step": 1385
    },
    {
      "epoch": 0.0636042402826855,
      "grad_norm": 0.6570529937744141,
      "learning_rate": 9.999963424795e-06,
      "loss": 0.4737,
      "step": 1386
    },
    {
      "epoch": 0.06365013078793998,
      "grad_norm": 0.47984176874160767,
      "learning_rate": 9.999962480958708e-06,
      "loss": 0.4061,
      "step": 1387
    },
    {
      "epoch": 0.06369602129319443,
      "grad_norm": 0.5021207332611084,
      "learning_rate": 9.999961525099083e-06,
      "loss": 0.4647,
      "step": 1388
    },
    {
      "epoch": 0.0637419117984489,
      "grad_norm": 0.5036116242408752,
      "learning_rate": 9.999960557216128e-06,
      "loss": 0.4665,
      "step": 1389
    },
    {
      "epoch": 0.06378780230370336,
      "grad_norm": 0.4725930392742157,
      "learning_rate": 9.999959577309847e-06,
      "loss": 0.3911,
      "step": 1390
    },
    {
      "epoch": 0.06383369280895783,
      "grad_norm": 0.5347336530685425,
      "learning_rate": 9.999958585380242e-06,
      "loss": 0.572,
      "step": 1391
    },
    {
      "epoch": 0.06387958331421228,
      "grad_norm": 0.4795195460319519,
      "learning_rate": 9.999957581427315e-06,
      "loss": 0.388,
      "step": 1392
    },
    {
      "epoch": 0.06392547381946675,
      "grad_norm": 0.5054691433906555,
      "learning_rate": 9.999956565451066e-06,
      "loss": 0.4247,
      "step": 1393
    },
    {
      "epoch": 0.06397136432472121,
      "grad_norm": 0.46716243028640747,
      "learning_rate": 9.999955537451502e-06,
      "loss": 0.3481,
      "step": 1394
    },
    {
      "epoch": 0.06401725482997568,
      "grad_norm": 0.5037909746170044,
      "learning_rate": 9.999954497428621e-06,
      "loss": 0.3781,
      "step": 1395
    },
    {
      "epoch": 0.06406314533523014,
      "grad_norm": 0.481431782245636,
      "learning_rate": 9.999953445382428e-06,
      "loss": 0.4591,
      "step": 1396
    },
    {
      "epoch": 0.06410903584048461,
      "grad_norm": 0.4764237105846405,
      "learning_rate": 9.999952381312924e-06,
      "loss": 0.4518,
      "step": 1397
    },
    {
      "epoch": 0.06415492634573906,
      "grad_norm": 0.5006119608879089,
      "learning_rate": 9.999951305220115e-06,
      "loss": 0.4727,
      "step": 1398
    },
    {
      "epoch": 0.06420081685099353,
      "grad_norm": 0.4904876947402954,
      "learning_rate": 9.999950217104e-06,
      "loss": 0.4317,
      "step": 1399
    },
    {
      "epoch": 0.06424670735624799,
      "grad_norm": 0.49891409277915955,
      "learning_rate": 9.999949116964582e-06,
      "loss": 0.4191,
      "step": 1400
    },
    {
      "epoch": 0.06429259786150246,
      "grad_norm": 0.48268988728523254,
      "learning_rate": 9.999948004801866e-06,
      "loss": 0.4,
      "step": 1401
    },
    {
      "epoch": 0.06433848836675692,
      "grad_norm": 0.4835261404514313,
      "learning_rate": 9.99994688061585e-06,
      "loss": 0.43,
      "step": 1402
    },
    {
      "epoch": 0.06438437887201139,
      "grad_norm": 0.4731764495372772,
      "learning_rate": 9.999945744406542e-06,
      "loss": 0.4271,
      "step": 1403
    },
    {
      "epoch": 0.06443026937726584,
      "grad_norm": 0.5208236575126648,
      "learning_rate": 9.999944596173943e-06,
      "loss": 0.4904,
      "step": 1404
    },
    {
      "epoch": 0.06447615988252031,
      "grad_norm": 0.48828527331352234,
      "learning_rate": 9.999943435918054e-06,
      "loss": 0.395,
      "step": 1405
    },
    {
      "epoch": 0.06452205038777477,
      "grad_norm": 0.4636908173561096,
      "learning_rate": 9.99994226363888e-06,
      "loss": 0.4136,
      "step": 1406
    },
    {
      "epoch": 0.06456794089302924,
      "grad_norm": 0.486735999584198,
      "learning_rate": 9.999941079336422e-06,
      "loss": 0.3574,
      "step": 1407
    },
    {
      "epoch": 0.0646138313982837,
      "grad_norm": 0.4457509219646454,
      "learning_rate": 9.999939883010686e-06,
      "loss": 0.3711,
      "step": 1408
    },
    {
      "epoch": 0.06465972190353816,
      "grad_norm": 0.5154923796653748,
      "learning_rate": 9.99993867466167e-06,
      "loss": 0.4917,
      "step": 1409
    },
    {
      "epoch": 0.06470561240879262,
      "grad_norm": 0.5085795521736145,
      "learning_rate": 9.99993745428938e-06,
      "loss": 0.5071,
      "step": 1410
    },
    {
      "epoch": 0.06475150291404709,
      "grad_norm": 0.45585259795188904,
      "learning_rate": 9.999936221893819e-06,
      "loss": 0.346,
      "step": 1411
    },
    {
      "epoch": 0.06479739341930155,
      "grad_norm": 0.5325887799263,
      "learning_rate": 9.999934977474988e-06,
      "loss": 0.4801,
      "step": 1412
    },
    {
      "epoch": 0.064843283924556,
      "grad_norm": 0.48388898372650146,
      "learning_rate": 9.999933721032892e-06,
      "loss": 0.3864,
      "step": 1413
    },
    {
      "epoch": 0.06488917442981047,
      "grad_norm": 0.46632200479507446,
      "learning_rate": 9.999932452567535e-06,
      "loss": 0.3283,
      "step": 1414
    },
    {
      "epoch": 0.06493506493506493,
      "grad_norm": 0.49091312289237976,
      "learning_rate": 9.999931172078918e-06,
      "loss": 0.3574,
      "step": 1415
    },
    {
      "epoch": 0.0649809554403194,
      "grad_norm": 0.4777570962905884,
      "learning_rate": 9.999929879567044e-06,
      "loss": 0.3807,
      "step": 1416
    },
    {
      "epoch": 0.06502684594557386,
      "grad_norm": 0.6797553300857544,
      "learning_rate": 9.999928575031917e-06,
      "loss": 0.6496,
      "step": 1417
    },
    {
      "epoch": 0.06507273645082833,
      "grad_norm": 0.48289597034454346,
      "learning_rate": 9.99992725847354e-06,
      "loss": 0.414,
      "step": 1418
    },
    {
      "epoch": 0.06511862695608278,
      "grad_norm": 0.49677640199661255,
      "learning_rate": 9.999925929891915e-06,
      "loss": 0.4638,
      "step": 1419
    },
    {
      "epoch": 0.06516451746133725,
      "grad_norm": 0.5317786335945129,
      "learning_rate": 9.999924589287047e-06,
      "loss": 0.3898,
      "step": 1420
    },
    {
      "epoch": 0.06521040796659171,
      "grad_norm": 0.48046281933784485,
      "learning_rate": 9.999923236658937e-06,
      "loss": 0.3705,
      "step": 1421
    },
    {
      "epoch": 0.06525629847184618,
      "grad_norm": 0.5125747919082642,
      "learning_rate": 9.999921872007591e-06,
      "loss": 0.44,
      "step": 1422
    },
    {
      "epoch": 0.06530218897710063,
      "grad_norm": 0.4760970175266266,
      "learning_rate": 9.999920495333011e-06,
      "loss": 0.3075,
      "step": 1423
    },
    {
      "epoch": 0.0653480794823551,
      "grad_norm": 0.49629834294319153,
      "learning_rate": 9.9999191066352e-06,
      "loss": 0.4229,
      "step": 1424
    },
    {
      "epoch": 0.06539396998760956,
      "grad_norm": 0.5172837972640991,
      "learning_rate": 9.999917705914162e-06,
      "loss": 0.4229,
      "step": 1425
    },
    {
      "epoch": 0.06543986049286403,
      "grad_norm": 0.5421844720840454,
      "learning_rate": 9.999916293169898e-06,
      "loss": 0.4791,
      "step": 1426
    },
    {
      "epoch": 0.06548575099811849,
      "grad_norm": 0.566855788230896,
      "learning_rate": 9.999914868402415e-06,
      "loss": 0.556,
      "step": 1427
    },
    {
      "epoch": 0.06553164150337296,
      "grad_norm": 0.5387606620788574,
      "learning_rate": 9.999913431611714e-06,
      "loss": 0.4659,
      "step": 1428
    },
    {
      "epoch": 0.06557753200862741,
      "grad_norm": 0.5093755722045898,
      "learning_rate": 9.999911982797801e-06,
      "loss": 0.3959,
      "step": 1429
    },
    {
      "epoch": 0.06562342251388188,
      "grad_norm": 0.4817783236503601,
      "learning_rate": 9.999910521960677e-06,
      "loss": 0.3666,
      "step": 1430
    },
    {
      "epoch": 0.06566931301913634,
      "grad_norm": 0.4898487329483032,
      "learning_rate": 9.999909049100345e-06,
      "loss": 0.4227,
      "step": 1431
    },
    {
      "epoch": 0.06571520352439081,
      "grad_norm": 0.45877912640571594,
      "learning_rate": 9.999907564216808e-06,
      "loss": 0.3669,
      "step": 1432
    },
    {
      "epoch": 0.06576109402964526,
      "grad_norm": 0.5108855366706848,
      "learning_rate": 9.999906067310075e-06,
      "loss": 0.4327,
      "step": 1433
    },
    {
      "epoch": 0.06580698453489973,
      "grad_norm": 0.4705347716808319,
      "learning_rate": 9.999904558380145e-06,
      "loss": 0.3935,
      "step": 1434
    },
    {
      "epoch": 0.06585287504015419,
      "grad_norm": 0.4982197880744934,
      "learning_rate": 9.999903037427022e-06,
      "loss": 0.4312,
      "step": 1435
    },
    {
      "epoch": 0.06589876554540866,
      "grad_norm": 0.5360764861106873,
      "learning_rate": 9.99990150445071e-06,
      "loss": 0.4989,
      "step": 1436
    },
    {
      "epoch": 0.06594465605066312,
      "grad_norm": 0.44408050179481506,
      "learning_rate": 9.999899959451214e-06,
      "loss": 0.3254,
      "step": 1437
    },
    {
      "epoch": 0.06599054655591759,
      "grad_norm": 0.44298574328422546,
      "learning_rate": 9.999898402428535e-06,
      "loss": 0.365,
      "step": 1438
    },
    {
      "epoch": 0.06603643706117204,
      "grad_norm": 0.48327207565307617,
      "learning_rate": 9.999896833382679e-06,
      "loss": 0.4417,
      "step": 1439
    },
    {
      "epoch": 0.06608232756642651,
      "grad_norm": 0.4832613170146942,
      "learning_rate": 9.99989525231365e-06,
      "loss": 0.357,
      "step": 1440
    },
    {
      "epoch": 0.06612821807168097,
      "grad_norm": 0.452497273683548,
      "learning_rate": 9.99989365922145e-06,
      "loss": 0.3956,
      "step": 1441
    },
    {
      "epoch": 0.06617410857693543,
      "grad_norm": 0.5015016794204712,
      "learning_rate": 9.999892054106083e-06,
      "loss": 0.4752,
      "step": 1442
    },
    {
      "epoch": 0.0662199990821899,
      "grad_norm": 0.541102945804596,
      "learning_rate": 9.999890436967555e-06,
      "loss": 0.4629,
      "step": 1443
    },
    {
      "epoch": 0.06626588958744435,
      "grad_norm": 0.44886717200279236,
      "learning_rate": 9.999888807805868e-06,
      "loss": 0.33,
      "step": 1444
    },
    {
      "epoch": 0.06631178009269882,
      "grad_norm": 0.4742789566516876,
      "learning_rate": 9.999887166621027e-06,
      "loss": 0.4154,
      "step": 1445
    },
    {
      "epoch": 0.06635767059795328,
      "grad_norm": 0.522352933883667,
      "learning_rate": 9.999885513413035e-06,
      "loss": 0.4191,
      "step": 1446
    },
    {
      "epoch": 0.06640356110320775,
      "grad_norm": 0.5095981955528259,
      "learning_rate": 9.999883848181896e-06,
      "loss": 0.4686,
      "step": 1447
    },
    {
      "epoch": 0.0664494516084622,
      "grad_norm": 0.5130565762519836,
      "learning_rate": 9.999882170927615e-06,
      "loss": 0.4693,
      "step": 1448
    },
    {
      "epoch": 0.06649534211371667,
      "grad_norm": 0.5398648977279663,
      "learning_rate": 9.999880481650196e-06,
      "loss": 0.4936,
      "step": 1449
    },
    {
      "epoch": 0.06654123261897113,
      "grad_norm": 0.47104206681251526,
      "learning_rate": 9.999878780349642e-06,
      "loss": 0.4079,
      "step": 1450
    },
    {
      "epoch": 0.0665871231242256,
      "grad_norm": 0.5229426622390747,
      "learning_rate": 9.999877067025955e-06,
      "loss": 0.4416,
      "step": 1451
    },
    {
      "epoch": 0.06663301362948006,
      "grad_norm": 0.4794468283653259,
      "learning_rate": 9.999875341679144e-06,
      "loss": 0.34,
      "step": 1452
    },
    {
      "epoch": 0.06667890413473453,
      "grad_norm": 0.5375478267669678,
      "learning_rate": 9.99987360430921e-06,
      "loss": 0.5231,
      "step": 1453
    },
    {
      "epoch": 0.06672479463998898,
      "grad_norm": 0.48184657096862793,
      "learning_rate": 9.999871854916159e-06,
      "loss": 0.3922,
      "step": 1454
    },
    {
      "epoch": 0.06677068514524345,
      "grad_norm": 0.5267612934112549,
      "learning_rate": 9.999870093499993e-06,
      "loss": 0.4748,
      "step": 1455
    },
    {
      "epoch": 0.06681657565049791,
      "grad_norm": 0.5034180283546448,
      "learning_rate": 9.99986832006072e-06,
      "loss": 0.4159,
      "step": 1456
    },
    {
      "epoch": 0.06686246615575238,
      "grad_norm": 0.49324777722358704,
      "learning_rate": 9.999866534598339e-06,
      "loss": 0.3985,
      "step": 1457
    },
    {
      "epoch": 0.06690835666100683,
      "grad_norm": 0.522338330745697,
      "learning_rate": 9.999864737112857e-06,
      "loss": 0.4403,
      "step": 1458
    },
    {
      "epoch": 0.0669542471662613,
      "grad_norm": 0.4822441637516022,
      "learning_rate": 9.99986292760428e-06,
      "loss": 0.4341,
      "step": 1459
    },
    {
      "epoch": 0.06700013767151576,
      "grad_norm": 0.4854448437690735,
      "learning_rate": 9.999861106072608e-06,
      "loss": 0.4203,
      "step": 1460
    },
    {
      "epoch": 0.06704602817677023,
      "grad_norm": 0.4899885356426239,
      "learning_rate": 9.999859272517852e-06,
      "loss": 0.4546,
      "step": 1461
    },
    {
      "epoch": 0.06709191868202469,
      "grad_norm": 0.489889919757843,
      "learning_rate": 9.99985742694001e-06,
      "loss": 0.3858,
      "step": 1462
    },
    {
      "epoch": 0.06713780918727916,
      "grad_norm": 0.50341796875,
      "learning_rate": 9.999855569339088e-06,
      "loss": 0.436,
      "step": 1463
    },
    {
      "epoch": 0.06718369969253361,
      "grad_norm": 0.4324839413166046,
      "learning_rate": 9.999853699715094e-06,
      "loss": 0.3398,
      "step": 1464
    },
    {
      "epoch": 0.06722959019778808,
      "grad_norm": 0.4206226170063019,
      "learning_rate": 9.999851818068027e-06,
      "loss": 0.3115,
      "step": 1465
    },
    {
      "epoch": 0.06727548070304254,
      "grad_norm": 0.4473482668399811,
      "learning_rate": 9.999849924397895e-06,
      "loss": 0.3311,
      "step": 1466
    },
    {
      "epoch": 0.06732137120829701,
      "grad_norm": 0.5379679203033447,
      "learning_rate": 9.999848018704703e-06,
      "loss": 0.5851,
      "step": 1467
    },
    {
      "epoch": 0.06736726171355147,
      "grad_norm": 0.46014630794525146,
      "learning_rate": 9.999846100988455e-06,
      "loss": 0.3913,
      "step": 1468
    },
    {
      "epoch": 0.06741315221880594,
      "grad_norm": 0.4650583565235138,
      "learning_rate": 9.999844171249153e-06,
      "loss": 0.425,
      "step": 1469
    },
    {
      "epoch": 0.06745904272406039,
      "grad_norm": 0.4622094929218292,
      "learning_rate": 9.999842229486806e-06,
      "loss": 0.3487,
      "step": 1470
    },
    {
      "epoch": 0.06750493322931485,
      "grad_norm": 0.50398850440979,
      "learning_rate": 9.999840275701416e-06,
      "loss": 0.4285,
      "step": 1471
    },
    {
      "epoch": 0.06755082373456932,
      "grad_norm": 0.5965123176574707,
      "learning_rate": 9.999838309892988e-06,
      "loss": 0.5413,
      "step": 1472
    },
    {
      "epoch": 0.06759671423982377,
      "grad_norm": 0.49533459544181824,
      "learning_rate": 9.999836332061526e-06,
      "loss": 0.4413,
      "step": 1473
    },
    {
      "epoch": 0.06764260474507824,
      "grad_norm": 0.5203729867935181,
      "learning_rate": 9.999834342207035e-06,
      "loss": 0.4864,
      "step": 1474
    },
    {
      "epoch": 0.0676884952503327,
      "grad_norm": 0.46969038248062134,
      "learning_rate": 9.999832340329523e-06,
      "loss": 0.3856,
      "step": 1475
    },
    {
      "epoch": 0.06773438575558717,
      "grad_norm": 0.4709130525588989,
      "learning_rate": 9.99983032642899e-06,
      "loss": 0.3594,
      "step": 1476
    },
    {
      "epoch": 0.06778027626084163,
      "grad_norm": 0.5098454356193542,
      "learning_rate": 9.999828300505443e-06,
      "loss": 0.3973,
      "step": 1477
    },
    {
      "epoch": 0.0678261667660961,
      "grad_norm": 0.5962406396865845,
      "learning_rate": 9.999826262558889e-06,
      "loss": 0.4966,
      "step": 1478
    },
    {
      "epoch": 0.06787205727135055,
      "grad_norm": 0.48592934012413025,
      "learning_rate": 9.999824212589328e-06,
      "loss": 0.445,
      "step": 1479
    },
    {
      "epoch": 0.06791794777660502,
      "grad_norm": 0.5435962080955505,
      "learning_rate": 9.99982215059677e-06,
      "loss": 0.482,
      "step": 1480
    },
    {
      "epoch": 0.06796383828185948,
      "grad_norm": 0.5200150609016418,
      "learning_rate": 9.999820076581218e-06,
      "loss": 0.3642,
      "step": 1481
    },
    {
      "epoch": 0.06800972878711395,
      "grad_norm": 0.4944203197956085,
      "learning_rate": 9.999817990542675e-06,
      "loss": 0.3487,
      "step": 1482
    },
    {
      "epoch": 0.0680556192923684,
      "grad_norm": 0.4612996280193329,
      "learning_rate": 9.999815892481149e-06,
      "loss": 0.3388,
      "step": 1483
    },
    {
      "epoch": 0.06810150979762288,
      "grad_norm": 0.5335448384284973,
      "learning_rate": 9.999813782396642e-06,
      "loss": 0.4182,
      "step": 1484
    },
    {
      "epoch": 0.06814740030287733,
      "grad_norm": 0.6675297021865845,
      "learning_rate": 9.999811660289163e-06,
      "loss": 0.6561,
      "step": 1485
    },
    {
      "epoch": 0.0681932908081318,
      "grad_norm": 0.49340584874153137,
      "learning_rate": 9.999809526158713e-06,
      "loss": 0.4091,
      "step": 1486
    },
    {
      "epoch": 0.06823918131338626,
      "grad_norm": 0.5457292199134827,
      "learning_rate": 9.9998073800053e-06,
      "loss": 0.3674,
      "step": 1487
    },
    {
      "epoch": 0.06828507181864073,
      "grad_norm": 0.48735201358795166,
      "learning_rate": 9.999805221828929e-06,
      "loss": 0.3718,
      "step": 1488
    },
    {
      "epoch": 0.06833096232389518,
      "grad_norm": 0.5068523287773132,
      "learning_rate": 9.999803051629604e-06,
      "loss": 0.3869,
      "step": 1489
    },
    {
      "epoch": 0.06837685282914965,
      "grad_norm": 0.5415776371955872,
      "learning_rate": 9.99980086940733e-06,
      "loss": 0.5328,
      "step": 1490
    },
    {
      "epoch": 0.06842274333440411,
      "grad_norm": 0.4694725275039673,
      "learning_rate": 9.999798675162114e-06,
      "loss": 0.4555,
      "step": 1491
    },
    {
      "epoch": 0.06846863383965858,
      "grad_norm": 0.4799397587776184,
      "learning_rate": 9.99979646889396e-06,
      "loss": 0.3815,
      "step": 1492
    },
    {
      "epoch": 0.06851452434491304,
      "grad_norm": 0.5082115530967712,
      "learning_rate": 9.999794250602872e-06,
      "loss": 0.3927,
      "step": 1493
    },
    {
      "epoch": 0.0685604148501675,
      "grad_norm": 0.4930780827999115,
      "learning_rate": 9.999792020288857e-06,
      "loss": 0.4178,
      "step": 1494
    },
    {
      "epoch": 0.06860630535542196,
      "grad_norm": 0.509013295173645,
      "learning_rate": 9.99978977795192e-06,
      "loss": 0.3992,
      "step": 1495
    },
    {
      "epoch": 0.06865219586067643,
      "grad_norm": 0.5124661326408386,
      "learning_rate": 9.999787523592068e-06,
      "loss": 0.4708,
      "step": 1496
    },
    {
      "epoch": 0.06869808636593089,
      "grad_norm": 0.5059641003608704,
      "learning_rate": 9.999785257209304e-06,
      "loss": 0.4204,
      "step": 1497
    },
    {
      "epoch": 0.06874397687118536,
      "grad_norm": 0.5025946497917175,
      "learning_rate": 9.999782978803636e-06,
      "loss": 0.4101,
      "step": 1498
    },
    {
      "epoch": 0.06878986737643981,
      "grad_norm": 0.501882016658783,
      "learning_rate": 9.999780688375068e-06,
      "loss": 0.4057,
      "step": 1499
    },
    {
      "epoch": 0.06883575788169427,
      "grad_norm": 0.5084762573242188,
      "learning_rate": 9.999778385923602e-06,
      "loss": 0.4067,
      "step": 1500
    },
    {
      "epoch": 0.06888164838694874,
      "grad_norm": 0.5197412967681885,
      "learning_rate": 9.99977607144925e-06,
      "loss": 0.4178,
      "step": 1501
    },
    {
      "epoch": 0.0689275388922032,
      "grad_norm": 0.4817604720592499,
      "learning_rate": 9.999773744952013e-06,
      "loss": 0.378,
      "step": 1502
    },
    {
      "epoch": 0.06897342939745767,
      "grad_norm": 0.5243045687675476,
      "learning_rate": 9.9997714064319e-06,
      "loss": 0.5335,
      "step": 1503
    },
    {
      "epoch": 0.06901931990271212,
      "grad_norm": 0.4980696439743042,
      "learning_rate": 9.999769055888913e-06,
      "loss": 0.4237,
      "step": 1504
    },
    {
      "epoch": 0.06906521040796659,
      "grad_norm": 0.57636958360672,
      "learning_rate": 9.999766693323058e-06,
      "loss": 0.5621,
      "step": 1505
    },
    {
      "epoch": 0.06911110091322105,
      "grad_norm": 0.47423747181892395,
      "learning_rate": 9.999764318734344e-06,
      "loss": 0.3721,
      "step": 1506
    },
    {
      "epoch": 0.06915699141847552,
      "grad_norm": 0.5168173909187317,
      "learning_rate": 9.999761932122776e-06,
      "loss": 0.5384,
      "step": 1507
    },
    {
      "epoch": 0.06920288192372998,
      "grad_norm": 0.47357746958732605,
      "learning_rate": 9.999759533488359e-06,
      "loss": 0.418,
      "step": 1508
    },
    {
      "epoch": 0.06924877242898445,
      "grad_norm": 0.5696466565132141,
      "learning_rate": 9.999757122831095e-06,
      "loss": 0.4814,
      "step": 1509
    },
    {
      "epoch": 0.0692946629342389,
      "grad_norm": 0.45018795132637024,
      "learning_rate": 9.999754700150995e-06,
      "loss": 0.3467,
      "step": 1510
    },
    {
      "epoch": 0.06934055343949337,
      "grad_norm": 0.4679853320121765,
      "learning_rate": 9.999752265448061e-06,
      "loss": 0.4159,
      "step": 1511
    },
    {
      "epoch": 0.06938644394474783,
      "grad_norm": 0.48860302567481995,
      "learning_rate": 9.999749818722303e-06,
      "loss": 0.4303,
      "step": 1512
    },
    {
      "epoch": 0.0694323344500023,
      "grad_norm": 0.4883895218372345,
      "learning_rate": 9.999747359973724e-06,
      "loss": 0.3779,
      "step": 1513
    },
    {
      "epoch": 0.06947822495525675,
      "grad_norm": 0.6044162511825562,
      "learning_rate": 9.99974488920233e-06,
      "loss": 0.5678,
      "step": 1514
    },
    {
      "epoch": 0.06952411546051122,
      "grad_norm": 0.46854501962661743,
      "learning_rate": 9.999742406408128e-06,
      "loss": 0.4126,
      "step": 1515
    },
    {
      "epoch": 0.06957000596576568,
      "grad_norm": 0.5346769690513611,
      "learning_rate": 9.999739911591124e-06,
      "loss": 0.5383,
      "step": 1516
    },
    {
      "epoch": 0.06961589647102015,
      "grad_norm": 0.4709138572216034,
      "learning_rate": 9.999737404751324e-06,
      "loss": 0.4031,
      "step": 1517
    },
    {
      "epoch": 0.0696617869762746,
      "grad_norm": 0.5270324945449829,
      "learning_rate": 9.999734885888732e-06,
      "loss": 0.4744,
      "step": 1518
    },
    {
      "epoch": 0.06970767748152908,
      "grad_norm": 0.534546971321106,
      "learning_rate": 9.999732355003356e-06,
      "loss": 0.4899,
      "step": 1519
    },
    {
      "epoch": 0.06975356798678353,
      "grad_norm": 0.47926053404808044,
      "learning_rate": 9.999729812095203e-06,
      "loss": 0.4319,
      "step": 1520
    },
    {
      "epoch": 0.069799458492038,
      "grad_norm": 0.5181357264518738,
      "learning_rate": 9.999727257164276e-06,
      "loss": 0.4303,
      "step": 1521
    },
    {
      "epoch": 0.06984534899729246,
      "grad_norm": 0.4206531345844269,
      "learning_rate": 9.999724690210583e-06,
      "loss": 0.3178,
      "step": 1522
    },
    {
      "epoch": 0.06989123950254693,
      "grad_norm": 0.6375095248222351,
      "learning_rate": 9.99972211123413e-06,
      "loss": 0.5252,
      "step": 1523
    },
    {
      "epoch": 0.06993713000780138,
      "grad_norm": 0.48132389783859253,
      "learning_rate": 9.999719520234924e-06,
      "loss": 0.4108,
      "step": 1524
    },
    {
      "epoch": 0.06998302051305585,
      "grad_norm": 0.48003053665161133,
      "learning_rate": 9.99971691721297e-06,
      "loss": 0.4281,
      "step": 1525
    },
    {
      "epoch": 0.07002891101831031,
      "grad_norm": 0.478013277053833,
      "learning_rate": 9.999714302168276e-06,
      "loss": 0.4039,
      "step": 1526
    },
    {
      "epoch": 0.07007480152356478,
      "grad_norm": 0.45703446865081787,
      "learning_rate": 9.999711675100845e-06,
      "loss": 0.3867,
      "step": 1527
    },
    {
      "epoch": 0.07012069202881924,
      "grad_norm": 0.4633644223213196,
      "learning_rate": 9.999709036010687e-06,
      "loss": 0.3881,
      "step": 1528
    },
    {
      "epoch": 0.07016658253407371,
      "grad_norm": 0.5113197565078735,
      "learning_rate": 9.999706384897805e-06,
      "loss": 0.4683,
      "step": 1529
    },
    {
      "epoch": 0.07021247303932816,
      "grad_norm": 0.45382463932037354,
      "learning_rate": 9.999703721762209e-06,
      "loss": 0.3756,
      "step": 1530
    },
    {
      "epoch": 0.07025836354458262,
      "grad_norm": 0.47707948088645935,
      "learning_rate": 9.999701046603903e-06,
      "loss": 0.3911,
      "step": 1531
    },
    {
      "epoch": 0.07030425404983709,
      "grad_norm": 0.5260950922966003,
      "learning_rate": 9.999698359422893e-06,
      "loss": 0.5267,
      "step": 1532
    },
    {
      "epoch": 0.07035014455509155,
      "grad_norm": 0.5094505548477173,
      "learning_rate": 9.999695660219188e-06,
      "loss": 0.4628,
      "step": 1533
    },
    {
      "epoch": 0.07039603506034602,
      "grad_norm": 0.507024347782135,
      "learning_rate": 9.999692948992792e-06,
      "loss": 0.4448,
      "step": 1534
    },
    {
      "epoch": 0.07044192556560047,
      "grad_norm": 0.4874345064163208,
      "learning_rate": 9.999690225743712e-06,
      "loss": 0.374,
      "step": 1535
    },
    {
      "epoch": 0.07048781607085494,
      "grad_norm": 0.4903397560119629,
      "learning_rate": 9.999687490471956e-06,
      "loss": 0.4031,
      "step": 1536
    },
    {
      "epoch": 0.0705337065761094,
      "grad_norm": 0.4693142771720886,
      "learning_rate": 9.99968474317753e-06,
      "loss": 0.3749,
      "step": 1537
    },
    {
      "epoch": 0.07057959708136387,
      "grad_norm": 0.4820583760738373,
      "learning_rate": 9.999681983860439e-06,
      "loss": 0.4276,
      "step": 1538
    },
    {
      "epoch": 0.07062548758661832,
      "grad_norm": 0.49902161955833435,
      "learning_rate": 9.99967921252069e-06,
      "loss": 0.4251,
      "step": 1539
    },
    {
      "epoch": 0.0706713780918728,
      "grad_norm": 0.542457103729248,
      "learning_rate": 9.999676429158292e-06,
      "loss": 0.4836,
      "step": 1540
    },
    {
      "epoch": 0.07071726859712725,
      "grad_norm": 0.516501247882843,
      "learning_rate": 9.999673633773248e-06,
      "loss": 0.5087,
      "step": 1541
    },
    {
      "epoch": 0.07076315910238172,
      "grad_norm": 0.4509405493736267,
      "learning_rate": 9.999670826365568e-06,
      "loss": 0.3612,
      "step": 1542
    },
    {
      "epoch": 0.07080904960763618,
      "grad_norm": 0.4723970293998718,
      "learning_rate": 9.99966800693526e-06,
      "loss": 0.4065,
      "step": 1543
    },
    {
      "epoch": 0.07085494011289065,
      "grad_norm": 0.46913012862205505,
      "learning_rate": 9.999665175482326e-06,
      "loss": 0.3502,
      "step": 1544
    },
    {
      "epoch": 0.0709008306181451,
      "grad_norm": 0.5791012644767761,
      "learning_rate": 9.999662332006774e-06,
      "loss": 0.4714,
      "step": 1545
    },
    {
      "epoch": 0.07094672112339957,
      "grad_norm": 0.4996468126773834,
      "learning_rate": 9.999659476508615e-06,
      "loss": 0.4715,
      "step": 1546
    },
    {
      "epoch": 0.07099261162865403,
      "grad_norm": 0.4880896806716919,
      "learning_rate": 9.99965660898785e-06,
      "loss": 0.4798,
      "step": 1547
    },
    {
      "epoch": 0.0710385021339085,
      "grad_norm": 0.480820894241333,
      "learning_rate": 9.99965372944449e-06,
      "loss": 0.3996,
      "step": 1548
    },
    {
      "epoch": 0.07108439263916295,
      "grad_norm": 0.5169656276702881,
      "learning_rate": 9.99965083787854e-06,
      "loss": 0.4567,
      "step": 1549
    },
    {
      "epoch": 0.07113028314441742,
      "grad_norm": 0.44148287177085876,
      "learning_rate": 9.99964793429001e-06,
      "loss": 0.3508,
      "step": 1550
    },
    {
      "epoch": 0.07117617364967188,
      "grad_norm": 0.45293569564819336,
      "learning_rate": 9.999645018678902e-06,
      "loss": 0.3841,
      "step": 1551
    },
    {
      "epoch": 0.07122206415492635,
      "grad_norm": 0.4716408848762512,
      "learning_rate": 9.999642091045227e-06,
      "loss": 0.3899,
      "step": 1552
    },
    {
      "epoch": 0.07126795466018081,
      "grad_norm": 0.4980045258998871,
      "learning_rate": 9.99963915138899e-06,
      "loss": 0.3948,
      "step": 1553
    },
    {
      "epoch": 0.07131384516543528,
      "grad_norm": 0.4896238446235657,
      "learning_rate": 9.999636199710201e-06,
      "loss": 0.4369,
      "step": 1554
    },
    {
      "epoch": 0.07135973567068973,
      "grad_norm": 0.4903663396835327,
      "learning_rate": 9.999633236008863e-06,
      "loss": 0.3711,
      "step": 1555
    },
    {
      "epoch": 0.0714056261759442,
      "grad_norm": 0.5232402682304382,
      "learning_rate": 9.999630260284985e-06,
      "loss": 0.4723,
      "step": 1556
    },
    {
      "epoch": 0.07145151668119866,
      "grad_norm": 0.4787043333053589,
      "learning_rate": 9.999627272538574e-06,
      "loss": 0.3576,
      "step": 1557
    },
    {
      "epoch": 0.07149740718645313,
      "grad_norm": 0.4703802466392517,
      "learning_rate": 9.999624272769638e-06,
      "loss": 0.381,
      "step": 1558
    },
    {
      "epoch": 0.07154329769170759,
      "grad_norm": 0.47272223234176636,
      "learning_rate": 9.999621260978184e-06,
      "loss": 0.4201,
      "step": 1559
    },
    {
      "epoch": 0.07158918819696204,
      "grad_norm": 0.4860653579235077,
      "learning_rate": 9.999618237164217e-06,
      "loss": 0.3986,
      "step": 1560
    },
    {
      "epoch": 0.07163507870221651,
      "grad_norm": 0.43770676851272583,
      "learning_rate": 9.999615201327747e-06,
      "loss": 0.3059,
      "step": 1561
    },
    {
      "epoch": 0.07168096920747097,
      "grad_norm": 0.5277722477912903,
      "learning_rate": 9.999612153468781e-06,
      "loss": 0.4428,
      "step": 1562
    },
    {
      "epoch": 0.07172685971272544,
      "grad_norm": 0.5587084889411926,
      "learning_rate": 9.999609093587325e-06,
      "loss": 0.5141,
      "step": 1563
    },
    {
      "epoch": 0.0717727502179799,
      "grad_norm": 0.47455084323883057,
      "learning_rate": 9.999606021683388e-06,
      "loss": 0.4299,
      "step": 1564
    },
    {
      "epoch": 0.07181864072323436,
      "grad_norm": 0.4605918228626251,
      "learning_rate": 9.999602937756975e-06,
      "loss": 0.4025,
      "step": 1565
    },
    {
      "epoch": 0.07186453122848882,
      "grad_norm": 0.5553957223892212,
      "learning_rate": 9.999599841808095e-06,
      "loss": 0.5459,
      "step": 1566
    },
    {
      "epoch": 0.07191042173374329,
      "grad_norm": 0.4999491572380066,
      "learning_rate": 9.999596733836756e-06,
      "loss": 0.4325,
      "step": 1567
    },
    {
      "epoch": 0.07195631223899775,
      "grad_norm": 0.5644165277481079,
      "learning_rate": 9.999593613842963e-06,
      "loss": 0.4809,
      "step": 1568
    },
    {
      "epoch": 0.07200220274425222,
      "grad_norm": 0.5210996866226196,
      "learning_rate": 9.99959048182673e-06,
      "loss": 0.4506,
      "step": 1569
    },
    {
      "epoch": 0.07204809324950667,
      "grad_norm": 0.5275651216506958,
      "learning_rate": 9.999587337788055e-06,
      "loss": 0.4472,
      "step": 1570
    },
    {
      "epoch": 0.07209398375476114,
      "grad_norm": 0.46705251932144165,
      "learning_rate": 9.99958418172695e-06,
      "loss": 0.3575,
      "step": 1571
    },
    {
      "epoch": 0.0721398742600156,
      "grad_norm": 0.496692955493927,
      "learning_rate": 9.999581013643427e-06,
      "loss": 0.3859,
      "step": 1572
    },
    {
      "epoch": 0.07218576476527007,
      "grad_norm": 0.5298336744308472,
      "learning_rate": 9.999577833537487e-06,
      "loss": 0.41,
      "step": 1573
    },
    {
      "epoch": 0.07223165527052453,
      "grad_norm": 0.4614573121070862,
      "learning_rate": 9.99957464140914e-06,
      "loss": 0.3891,
      "step": 1574
    },
    {
      "epoch": 0.072277545775779,
      "grad_norm": 0.5605151653289795,
      "learning_rate": 9.999571437258395e-06,
      "loss": 0.514,
      "step": 1575
    },
    {
      "epoch": 0.07232343628103345,
      "grad_norm": 0.4622315466403961,
      "learning_rate": 9.999568221085258e-06,
      "loss": 0.4008,
      "step": 1576
    },
    {
      "epoch": 0.07236932678628792,
      "grad_norm": 0.49091848731040955,
      "learning_rate": 9.999564992889739e-06,
      "loss": 0.4182,
      "step": 1577
    },
    {
      "epoch": 0.07241521729154238,
      "grad_norm": 0.5226359367370605,
      "learning_rate": 9.999561752671842e-06,
      "loss": 0.4755,
      "step": 1578
    },
    {
      "epoch": 0.07246110779679685,
      "grad_norm": 0.5172597765922546,
      "learning_rate": 9.999558500431578e-06,
      "loss": 0.39,
      "step": 1579
    },
    {
      "epoch": 0.0725069983020513,
      "grad_norm": 0.4999792277812958,
      "learning_rate": 9.999555236168954e-06,
      "loss": 0.4659,
      "step": 1580
    },
    {
      "epoch": 0.07255288880730577,
      "grad_norm": 0.5266298055648804,
      "learning_rate": 9.999551959883976e-06,
      "loss": 0.4457,
      "step": 1581
    },
    {
      "epoch": 0.07259877931256023,
      "grad_norm": 0.44880470633506775,
      "learning_rate": 9.999548671576655e-06,
      "loss": 0.3589,
      "step": 1582
    },
    {
      "epoch": 0.0726446698178147,
      "grad_norm": 0.49409666657447815,
      "learning_rate": 9.999545371246997e-06,
      "loss": 0.4132,
      "step": 1583
    },
    {
      "epoch": 0.07269056032306916,
      "grad_norm": 0.49254390597343445,
      "learning_rate": 9.999542058895009e-06,
      "loss": 0.3911,
      "step": 1584
    },
    {
      "epoch": 0.07273645082832363,
      "grad_norm": 0.5089792609214783,
      "learning_rate": 9.999538734520702e-06,
      "loss": 0.3997,
      "step": 1585
    },
    {
      "epoch": 0.07278234133357808,
      "grad_norm": 0.5009763836860657,
      "learning_rate": 9.99953539812408e-06,
      "loss": 0.3708,
      "step": 1586
    },
    {
      "epoch": 0.07282823183883255,
      "grad_norm": 0.4541475474834442,
      "learning_rate": 9.999532049705156e-06,
      "loss": 0.3192,
      "step": 1587
    },
    {
      "epoch": 0.07287412234408701,
      "grad_norm": 0.4755299687385559,
      "learning_rate": 9.999528689263933e-06,
      "loss": 0.4193,
      "step": 1588
    },
    {
      "epoch": 0.07292001284934146,
      "grad_norm": 0.4521019756793976,
      "learning_rate": 9.999525316800422e-06,
      "loss": 0.3407,
      "step": 1589
    },
    {
      "epoch": 0.07296590335459593,
      "grad_norm": 0.4645295739173889,
      "learning_rate": 9.999521932314632e-06,
      "loss": 0.3181,
      "step": 1590
    },
    {
      "epoch": 0.07301179385985039,
      "grad_norm": 0.500056266784668,
      "learning_rate": 9.999518535806568e-06,
      "loss": 0.4333,
      "step": 1591
    },
    {
      "epoch": 0.07305768436510486,
      "grad_norm": 0.4169314205646515,
      "learning_rate": 9.99951512727624e-06,
      "loss": 0.3031,
      "step": 1592
    },
    {
      "epoch": 0.07310357487035932,
      "grad_norm": 0.5232287645339966,
      "learning_rate": 9.999511706723657e-06,
      "loss": 0.4911,
      "step": 1593
    },
    {
      "epoch": 0.07314946537561379,
      "grad_norm": 0.5228270888328552,
      "learning_rate": 9.999508274148825e-06,
      "loss": 0.5135,
      "step": 1594
    },
    {
      "epoch": 0.07319535588086824,
      "grad_norm": 0.5087874531745911,
      "learning_rate": 9.999504829551755e-06,
      "loss": 0.4426,
      "step": 1595
    },
    {
      "epoch": 0.07324124638612271,
      "grad_norm": 0.5080674886703491,
      "learning_rate": 9.999501372932453e-06,
      "loss": 0.4309,
      "step": 1596
    },
    {
      "epoch": 0.07328713689137717,
      "grad_norm": 0.5303704142570496,
      "learning_rate": 9.999497904290928e-06,
      "loss": 0.5957,
      "step": 1597
    },
    {
      "epoch": 0.07333302739663164,
      "grad_norm": 0.5428318977355957,
      "learning_rate": 9.99949442362719e-06,
      "loss": 0.4981,
      "step": 1598
    },
    {
      "epoch": 0.0733789179018861,
      "grad_norm": 0.49112948775291443,
      "learning_rate": 9.999490930941243e-06,
      "loss": 0.4595,
      "step": 1599
    },
    {
      "epoch": 0.07342480840714057,
      "grad_norm": 0.4157106280326843,
      "learning_rate": 9.999487426233099e-06,
      "loss": 0.3169,
      "step": 1600
    },
    {
      "epoch": 0.07347069891239502,
      "grad_norm": 0.4652913510799408,
      "learning_rate": 9.999483909502767e-06,
      "loss": 0.392,
      "step": 1601
    },
    {
      "epoch": 0.07351658941764949,
      "grad_norm": 0.4506971538066864,
      "learning_rate": 9.999480380750253e-06,
      "loss": 0.3363,
      "step": 1602
    },
    {
      "epoch": 0.07356247992290395,
      "grad_norm": 0.5362367630004883,
      "learning_rate": 9.999476839975566e-06,
      "loss": 0.5811,
      "step": 1603
    },
    {
      "epoch": 0.07360837042815842,
      "grad_norm": 0.5057291388511658,
      "learning_rate": 9.999473287178715e-06,
      "loss": 0.4575,
      "step": 1604
    },
    {
      "epoch": 0.07365426093341287,
      "grad_norm": 0.48182326555252075,
      "learning_rate": 9.99946972235971e-06,
      "loss": 0.4301,
      "step": 1605
    },
    {
      "epoch": 0.07370015143866734,
      "grad_norm": 0.4803471565246582,
      "learning_rate": 9.999466145518558e-06,
      "loss": 0.4645,
      "step": 1606
    },
    {
      "epoch": 0.0737460419439218,
      "grad_norm": 0.5345284342765808,
      "learning_rate": 9.999462556655266e-06,
      "loss": 0.388,
      "step": 1607
    },
    {
      "epoch": 0.07379193244917627,
      "grad_norm": 0.47404131293296814,
      "learning_rate": 9.999458955769848e-06,
      "loss": 0.4262,
      "step": 1608
    },
    {
      "epoch": 0.07383782295443073,
      "grad_norm": 0.4565417468547821,
      "learning_rate": 9.999455342862305e-06,
      "loss": 0.3921,
      "step": 1609
    },
    {
      "epoch": 0.0738837134596852,
      "grad_norm": 0.40013477206230164,
      "learning_rate": 9.999451717932652e-06,
      "loss": 0.297,
      "step": 1610
    },
    {
      "epoch": 0.07392960396493965,
      "grad_norm": 0.4709477424621582,
      "learning_rate": 9.999448080980896e-06,
      "loss": 0.4011,
      "step": 1611
    },
    {
      "epoch": 0.07397549447019412,
      "grad_norm": 0.494823157787323,
      "learning_rate": 9.999444432007044e-06,
      "loss": 0.4454,
      "step": 1612
    },
    {
      "epoch": 0.07402138497544858,
      "grad_norm": 0.4898694157600403,
      "learning_rate": 9.999440771011106e-06,
      "loss": 0.4206,
      "step": 1613
    },
    {
      "epoch": 0.07406727548070305,
      "grad_norm": 0.4711502492427826,
      "learning_rate": 9.999437097993092e-06,
      "loss": 0.4097,
      "step": 1614
    },
    {
      "epoch": 0.0741131659859575,
      "grad_norm": 0.4736449420452118,
      "learning_rate": 9.999433412953009e-06,
      "loss": 0.4286,
      "step": 1615
    },
    {
      "epoch": 0.07415905649121197,
      "grad_norm": 0.48827502131462097,
      "learning_rate": 9.999429715890867e-06,
      "loss": 0.4054,
      "step": 1616
    },
    {
      "epoch": 0.07420494699646643,
      "grad_norm": 0.44157078862190247,
      "learning_rate": 9.999426006806671e-06,
      "loss": 0.3429,
      "step": 1617
    },
    {
      "epoch": 0.07425083750172089,
      "grad_norm": 0.4292319416999817,
      "learning_rate": 9.999422285700437e-06,
      "loss": 0.3686,
      "step": 1618
    },
    {
      "epoch": 0.07429672800697536,
      "grad_norm": 0.5420536398887634,
      "learning_rate": 9.99941855257217e-06,
      "loss": 0.4866,
      "step": 1619
    },
    {
      "epoch": 0.07434261851222981,
      "grad_norm": 0.488696426153183,
      "learning_rate": 9.99941480742188e-06,
      "loss": 0.4742,
      "step": 1620
    },
    {
      "epoch": 0.07438850901748428,
      "grad_norm": 0.4898790121078491,
      "learning_rate": 9.999411050249572e-06,
      "loss": 0.4159,
      "step": 1621
    },
    {
      "epoch": 0.07443439952273874,
      "grad_norm": 0.49666571617126465,
      "learning_rate": 9.99940728105526e-06,
      "loss": 0.3826,
      "step": 1622
    },
    {
      "epoch": 0.07448029002799321,
      "grad_norm": 0.5466395020484924,
      "learning_rate": 9.999403499838953e-06,
      "loss": 0.4834,
      "step": 1623
    },
    {
      "epoch": 0.07452618053324767,
      "grad_norm": 0.4694845676422119,
      "learning_rate": 9.999399706600656e-06,
      "loss": 0.383,
      "step": 1624
    },
    {
      "epoch": 0.07457207103850214,
      "grad_norm": 0.45946910977363586,
      "learning_rate": 9.99939590134038e-06,
      "loss": 0.4152,
      "step": 1625
    },
    {
      "epoch": 0.07461796154375659,
      "grad_norm": 0.47266894578933716,
      "learning_rate": 9.999392084058138e-06,
      "loss": 0.3846,
      "step": 1626
    },
    {
      "epoch": 0.07466385204901106,
      "grad_norm": 0.4746292233467102,
      "learning_rate": 9.999388254753934e-06,
      "loss": 0.4409,
      "step": 1627
    },
    {
      "epoch": 0.07470974255426552,
      "grad_norm": 0.44414758682250977,
      "learning_rate": 9.99938441342778e-06,
      "loss": 0.3678,
      "step": 1628
    },
    {
      "epoch": 0.07475563305951999,
      "grad_norm": 0.4766943156719208,
      "learning_rate": 9.999380560079682e-06,
      "loss": 0.4166,
      "step": 1629
    },
    {
      "epoch": 0.07480152356477444,
      "grad_norm": 0.7866554856300354,
      "learning_rate": 9.999376694709654e-06,
      "loss": 0.6537,
      "step": 1630
    },
    {
      "epoch": 0.07484741407002891,
      "grad_norm": 0.44019708037376404,
      "learning_rate": 9.999372817317701e-06,
      "loss": 0.3407,
      "step": 1631
    },
    {
      "epoch": 0.07489330457528337,
      "grad_norm": 0.5445822477340698,
      "learning_rate": 9.999368927903837e-06,
      "loss": 0.4334,
      "step": 1632
    },
    {
      "epoch": 0.07493919508053784,
      "grad_norm": 0.49358266592025757,
      "learning_rate": 9.999365026468066e-06,
      "loss": 0.4076,
      "step": 1633
    },
    {
      "epoch": 0.0749850855857923,
      "grad_norm": 0.5387043952941895,
      "learning_rate": 9.9993611130104e-06,
      "loss": 0.4127,
      "step": 1634
    },
    {
      "epoch": 0.07503097609104677,
      "grad_norm": 0.47857046127319336,
      "learning_rate": 9.99935718753085e-06,
      "loss": 0.4418,
      "step": 1635
    },
    {
      "epoch": 0.07507686659630122,
      "grad_norm": 0.5949978828430176,
      "learning_rate": 9.999353250029422e-06,
      "loss": 0.5505,
      "step": 1636
    },
    {
      "epoch": 0.07512275710155569,
      "grad_norm": 0.5482997894287109,
      "learning_rate": 9.999349300506129e-06,
      "loss": 0.4237,
      "step": 1637
    },
    {
      "epoch": 0.07516864760681015,
      "grad_norm": 0.47320905327796936,
      "learning_rate": 9.999345338960977e-06,
      "loss": 0.3847,
      "step": 1638
    },
    {
      "epoch": 0.07521453811206462,
      "grad_norm": 0.4918878674507141,
      "learning_rate": 9.999341365393979e-06,
      "loss": 0.3908,
      "step": 1639
    },
    {
      "epoch": 0.07526042861731908,
      "grad_norm": 0.5108606815338135,
      "learning_rate": 9.999337379805142e-06,
      "loss": 0.3645,
      "step": 1640
    },
    {
      "epoch": 0.07530631912257355,
      "grad_norm": 0.5490548014640808,
      "learning_rate": 9.999333382194476e-06,
      "loss": 0.4629,
      "step": 1641
    },
    {
      "epoch": 0.075352209627828,
      "grad_norm": 0.466717392206192,
      "learning_rate": 9.999329372561992e-06,
      "loss": 0.3738,
      "step": 1642
    },
    {
      "epoch": 0.07539810013308247,
      "grad_norm": 0.4765077233314514,
      "learning_rate": 9.999325350907697e-06,
      "loss": 0.3859,
      "step": 1643
    },
    {
      "epoch": 0.07544399063833693,
      "grad_norm": 0.5266830325126648,
      "learning_rate": 9.999321317231604e-06,
      "loss": 0.415,
      "step": 1644
    },
    {
      "epoch": 0.0754898811435914,
      "grad_norm": 0.690089762210846,
      "learning_rate": 9.999317271533722e-06,
      "loss": 0.4348,
      "step": 1645
    },
    {
      "epoch": 0.07553577164884585,
      "grad_norm": 0.47478124499320984,
      "learning_rate": 9.999313213814058e-06,
      "loss": 0.4176,
      "step": 1646
    },
    {
      "epoch": 0.07558166215410031,
      "grad_norm": 0.46994373202323914,
      "learning_rate": 9.999309144072624e-06,
      "loss": 0.3919,
      "step": 1647
    },
    {
      "epoch": 0.07562755265935478,
      "grad_norm": 0.500823438167572,
      "learning_rate": 9.99930506230943e-06,
      "loss": 0.3488,
      "step": 1648
    },
    {
      "epoch": 0.07567344316460924,
      "grad_norm": 0.5512404441833496,
      "learning_rate": 9.999300968524485e-06,
      "loss": 0.5014,
      "step": 1649
    },
    {
      "epoch": 0.0757193336698637,
      "grad_norm": 0.5216124057769775,
      "learning_rate": 9.999296862717799e-06,
      "loss": 0.423,
      "step": 1650
    },
    {
      "epoch": 0.07576522417511816,
      "grad_norm": 0.4777539372444153,
      "learning_rate": 9.99929274488938e-06,
      "loss": 0.3984,
      "step": 1651
    },
    {
      "epoch": 0.07581111468037263,
      "grad_norm": 0.47649621963500977,
      "learning_rate": 9.999288615039242e-06,
      "loss": 0.3922,
      "step": 1652
    },
    {
      "epoch": 0.07585700518562709,
      "grad_norm": 0.4488726556301117,
      "learning_rate": 9.999284473167392e-06,
      "loss": 0.376,
      "step": 1653
    },
    {
      "epoch": 0.07590289569088156,
      "grad_norm": 0.5228418111801147,
      "learning_rate": 9.999280319273841e-06,
      "loss": 0.4731,
      "step": 1654
    },
    {
      "epoch": 0.07594878619613601,
      "grad_norm": 0.5244832038879395,
      "learning_rate": 9.999276153358599e-06,
      "loss": 0.5163,
      "step": 1655
    },
    {
      "epoch": 0.07599467670139048,
      "grad_norm": 0.5477701425552368,
      "learning_rate": 9.999271975421673e-06,
      "loss": 0.4904,
      "step": 1656
    },
    {
      "epoch": 0.07604056720664494,
      "grad_norm": 0.5430203676223755,
      "learning_rate": 9.999267785463078e-06,
      "loss": 0.5424,
      "step": 1657
    },
    {
      "epoch": 0.07608645771189941,
      "grad_norm": 0.4881824553012848,
      "learning_rate": 9.999263583482821e-06,
      "loss": 0.3996,
      "step": 1658
    },
    {
      "epoch": 0.07613234821715387,
      "grad_norm": 0.5316872596740723,
      "learning_rate": 9.999259369480914e-06,
      "loss": 0.4728,
      "step": 1659
    },
    {
      "epoch": 0.07617823872240834,
      "grad_norm": 0.4986788034439087,
      "learning_rate": 9.999255143457365e-06,
      "loss": 0.4431,
      "step": 1660
    },
    {
      "epoch": 0.07622412922766279,
      "grad_norm": 0.4756495952606201,
      "learning_rate": 9.999250905412185e-06,
      "loss": 0.4193,
      "step": 1661
    },
    {
      "epoch": 0.07627001973291726,
      "grad_norm": 0.5116938352584839,
      "learning_rate": 9.999246655345386e-06,
      "loss": 0.4547,
      "step": 1662
    },
    {
      "epoch": 0.07631591023817172,
      "grad_norm": 0.46892058849334717,
      "learning_rate": 9.999242393256976e-06,
      "loss": 0.3562,
      "step": 1663
    },
    {
      "epoch": 0.07636180074342619,
      "grad_norm": 0.542378842830658,
      "learning_rate": 9.999238119146964e-06,
      "loss": 0.5119,
      "step": 1664
    },
    {
      "epoch": 0.07640769124868065,
      "grad_norm": 0.44915884733200073,
      "learning_rate": 9.999233833015364e-06,
      "loss": 0.3532,
      "step": 1665
    },
    {
      "epoch": 0.07645358175393512,
      "grad_norm": 0.4248468577861786,
      "learning_rate": 9.999229534862185e-06,
      "loss": 0.3035,
      "step": 1666
    },
    {
      "epoch": 0.07649947225918957,
      "grad_norm": 0.4087550938129425,
      "learning_rate": 9.999225224687435e-06,
      "loss": 0.3022,
      "step": 1667
    },
    {
      "epoch": 0.07654536276444404,
      "grad_norm": 0.4679374694824219,
      "learning_rate": 9.999220902491127e-06,
      "loss": 0.4054,
      "step": 1668
    },
    {
      "epoch": 0.0765912532696985,
      "grad_norm": 0.501705527305603,
      "learning_rate": 9.99921656827327e-06,
      "loss": 0.4336,
      "step": 1669
    },
    {
      "epoch": 0.07663714377495297,
      "grad_norm": 0.5003342628479004,
      "learning_rate": 9.999212222033877e-06,
      "loss": 0.5029,
      "step": 1670
    },
    {
      "epoch": 0.07668303428020742,
      "grad_norm": 0.5207168459892273,
      "learning_rate": 9.999207863772952e-06,
      "loss": 0.4937,
      "step": 1671
    },
    {
      "epoch": 0.0767289247854619,
      "grad_norm": 0.44004499912261963,
      "learning_rate": 9.999203493490513e-06,
      "loss": 0.3677,
      "step": 1672
    },
    {
      "epoch": 0.07677481529071635,
      "grad_norm": 0.46720001101493835,
      "learning_rate": 9.999199111186569e-06,
      "loss": 0.3987,
      "step": 1673
    },
    {
      "epoch": 0.07682070579597082,
      "grad_norm": 0.5164360404014587,
      "learning_rate": 9.999194716861125e-06,
      "loss": 0.4235,
      "step": 1674
    },
    {
      "epoch": 0.07686659630122528,
      "grad_norm": 0.448982298374176,
      "learning_rate": 9.999190310514198e-06,
      "loss": 0.3607,
      "step": 1675
    },
    {
      "epoch": 0.07691248680647975,
      "grad_norm": 0.44107311964035034,
      "learning_rate": 9.999185892145795e-06,
      "loss": 0.3519,
      "step": 1676
    },
    {
      "epoch": 0.0769583773117342,
      "grad_norm": 0.5092818737030029,
      "learning_rate": 9.999181461755929e-06,
      "loss": 0.4563,
      "step": 1677
    },
    {
      "epoch": 0.07700426781698866,
      "grad_norm": 0.5030384659767151,
      "learning_rate": 9.999177019344607e-06,
      "loss": 0.5062,
      "step": 1678
    },
    {
      "epoch": 0.07705015832224313,
      "grad_norm": 0.5244572758674622,
      "learning_rate": 9.999172564911845e-06,
      "loss": 0.4991,
      "step": 1679
    },
    {
      "epoch": 0.07709604882749758,
      "grad_norm": 0.46394649147987366,
      "learning_rate": 9.99916809845765e-06,
      "loss": 0.435,
      "step": 1680
    },
    {
      "epoch": 0.07714193933275205,
      "grad_norm": 0.4839397370815277,
      "learning_rate": 9.999163619982033e-06,
      "loss": 0.3944,
      "step": 1681
    },
    {
      "epoch": 0.07718782983800651,
      "grad_norm": 0.4912411868572235,
      "learning_rate": 9.999159129485004e-06,
      "loss": 0.4847,
      "step": 1682
    },
    {
      "epoch": 0.07723372034326098,
      "grad_norm": 0.5226150751113892,
      "learning_rate": 9.999154626966576e-06,
      "loss": 0.4746,
      "step": 1683
    },
    {
      "epoch": 0.07727961084851544,
      "grad_norm": 0.540511965751648,
      "learning_rate": 9.99915011242676e-06,
      "loss": 0.493,
      "step": 1684
    },
    {
      "epoch": 0.07732550135376991,
      "grad_norm": 0.5299343466758728,
      "learning_rate": 9.999145585865565e-06,
      "loss": 0.4382,
      "step": 1685
    },
    {
      "epoch": 0.07737139185902436,
      "grad_norm": 0.5273798704147339,
      "learning_rate": 9.999141047283003e-06,
      "loss": 0.5291,
      "step": 1686
    },
    {
      "epoch": 0.07741728236427883,
      "grad_norm": 0.5626891851425171,
      "learning_rate": 9.999136496679082e-06,
      "loss": 0.531,
      "step": 1687
    },
    {
      "epoch": 0.07746317286953329,
      "grad_norm": 0.5069789290428162,
      "learning_rate": 9.999131934053818e-06,
      "loss": 0.5268,
      "step": 1688
    },
    {
      "epoch": 0.07750906337478776,
      "grad_norm": 0.4648505747318268,
      "learning_rate": 9.99912735940722e-06,
      "loss": 0.3655,
      "step": 1689
    },
    {
      "epoch": 0.07755495388004222,
      "grad_norm": 0.4760192334651947,
      "learning_rate": 9.999122772739299e-06,
      "loss": 0.4099,
      "step": 1690
    },
    {
      "epoch": 0.07760084438529669,
      "grad_norm": 0.47898706793785095,
      "learning_rate": 9.999118174050063e-06,
      "loss": 0.418,
      "step": 1691
    },
    {
      "epoch": 0.07764673489055114,
      "grad_norm": 0.5100760459899902,
      "learning_rate": 9.999113563339527e-06,
      "loss": 0.5028,
      "step": 1692
    },
    {
      "epoch": 0.07769262539580561,
      "grad_norm": 0.4605445861816406,
      "learning_rate": 9.9991089406077e-06,
      "loss": 0.4021,
      "step": 1693
    },
    {
      "epoch": 0.07773851590106007,
      "grad_norm": 0.5177266001701355,
      "learning_rate": 9.999104305854595e-06,
      "loss": 0.4719,
      "step": 1694
    },
    {
      "epoch": 0.07778440640631454,
      "grad_norm": 0.474923700094223,
      "learning_rate": 9.99909965908022e-06,
      "loss": 0.347,
      "step": 1695
    },
    {
      "epoch": 0.077830296911569,
      "grad_norm": 0.4658982753753662,
      "learning_rate": 9.999095000284588e-06,
      "loss": 0.3567,
      "step": 1696
    },
    {
      "epoch": 0.07787618741682346,
      "grad_norm": 0.4969722628593445,
      "learning_rate": 9.999090329467712e-06,
      "loss": 0.3779,
      "step": 1697
    },
    {
      "epoch": 0.07792207792207792,
      "grad_norm": 0.4199715256690979,
      "learning_rate": 9.9990856466296e-06,
      "loss": 0.3473,
      "step": 1698
    },
    {
      "epoch": 0.07796796842733239,
      "grad_norm": 0.5353632569313049,
      "learning_rate": 9.999080951770266e-06,
      "loss": 0.5526,
      "step": 1699
    },
    {
      "epoch": 0.07801385893258685,
      "grad_norm": 0.5019856095314026,
      "learning_rate": 9.99907624488972e-06,
      "loss": 0.4686,
      "step": 1700
    },
    {
      "epoch": 0.07805974943784132,
      "grad_norm": 0.48264744877815247,
      "learning_rate": 9.999071525987973e-06,
      "loss": 0.356,
      "step": 1701
    },
    {
      "epoch": 0.07810563994309577,
      "grad_norm": 0.4562687277793884,
      "learning_rate": 9.999066795065037e-06,
      "loss": 0.3239,
      "step": 1702
    },
    {
      "epoch": 0.07815153044835024,
      "grad_norm": 0.5122276544570923,
      "learning_rate": 9.999062052120922e-06,
      "loss": 0.4815,
      "step": 1703
    },
    {
      "epoch": 0.0781974209536047,
      "grad_norm": 0.505256175994873,
      "learning_rate": 9.999057297155642e-06,
      "loss": 0.4016,
      "step": 1704
    },
    {
      "epoch": 0.07824331145885917,
      "grad_norm": 0.46977514028549194,
      "learning_rate": 9.999052530169207e-06,
      "loss": 0.3919,
      "step": 1705
    },
    {
      "epoch": 0.07828920196411362,
      "grad_norm": 0.4590289890766144,
      "learning_rate": 9.999047751161629e-06,
      "loss": 0.3469,
      "step": 1706
    },
    {
      "epoch": 0.07833509246936808,
      "grad_norm": 0.4907650649547577,
      "learning_rate": 9.999042960132917e-06,
      "loss": 0.3969,
      "step": 1707
    },
    {
      "epoch": 0.07838098297462255,
      "grad_norm": 0.421985387802124,
      "learning_rate": 9.999038157083084e-06,
      "loss": 0.2934,
      "step": 1708
    },
    {
      "epoch": 0.07842687347987701,
      "grad_norm": 0.5014013648033142,
      "learning_rate": 9.999033342012143e-06,
      "loss": 0.3988,
      "step": 1709
    },
    {
      "epoch": 0.07847276398513148,
      "grad_norm": 0.43560871481895447,
      "learning_rate": 9.999028514920104e-06,
      "loss": 0.3443,
      "step": 1710
    },
    {
      "epoch": 0.07851865449038593,
      "grad_norm": 0.49493759870529175,
      "learning_rate": 9.99902367580698e-06,
      "loss": 0.443,
      "step": 1711
    },
    {
      "epoch": 0.0785645449956404,
      "grad_norm": 0.4802602529525757,
      "learning_rate": 9.999018824672781e-06,
      "loss": 0.4454,
      "step": 1712
    },
    {
      "epoch": 0.07861043550089486,
      "grad_norm": 0.4706229269504547,
      "learning_rate": 9.999013961517521e-06,
      "loss": 0.3732,
      "step": 1713
    },
    {
      "epoch": 0.07865632600614933,
      "grad_norm": 0.5777989029884338,
      "learning_rate": 9.999009086341209e-06,
      "loss": 0.5767,
      "step": 1714
    },
    {
      "epoch": 0.07870221651140379,
      "grad_norm": 0.4825112819671631,
      "learning_rate": 9.999004199143858e-06,
      "loss": 0.3597,
      "step": 1715
    },
    {
      "epoch": 0.07874810701665826,
      "grad_norm": 0.5652618408203125,
      "learning_rate": 9.99899929992548e-06,
      "loss": 0.5087,
      "step": 1716
    },
    {
      "epoch": 0.07879399752191271,
      "grad_norm": 0.5355689525604248,
      "learning_rate": 9.998994388686086e-06,
      "loss": 0.4749,
      "step": 1717
    },
    {
      "epoch": 0.07883988802716718,
      "grad_norm": 0.4698161482810974,
      "learning_rate": 9.998989465425689e-06,
      "loss": 0.3432,
      "step": 1718
    },
    {
      "epoch": 0.07888577853242164,
      "grad_norm": 0.4778725802898407,
      "learning_rate": 9.9989845301443e-06,
      "loss": 0.4306,
      "step": 1719
    },
    {
      "epoch": 0.07893166903767611,
      "grad_norm": 0.4972658157348633,
      "learning_rate": 9.99897958284193e-06,
      "loss": 0.4225,
      "step": 1720
    },
    {
      "epoch": 0.07897755954293056,
      "grad_norm": 0.48646584153175354,
      "learning_rate": 9.998974623518592e-06,
      "loss": 0.4428,
      "step": 1721
    },
    {
      "epoch": 0.07902345004818503,
      "grad_norm": 0.5065861940383911,
      "learning_rate": 9.998969652174297e-06,
      "loss": 0.4162,
      "step": 1722
    },
    {
      "epoch": 0.07906934055343949,
      "grad_norm": 0.5074966549873352,
      "learning_rate": 9.998964668809058e-06,
      "loss": 0.4307,
      "step": 1723
    },
    {
      "epoch": 0.07911523105869396,
      "grad_norm": 0.5008343458175659,
      "learning_rate": 9.998959673422887e-06,
      "loss": 0.4689,
      "step": 1724
    },
    {
      "epoch": 0.07916112156394842,
      "grad_norm": 0.5037351250648499,
      "learning_rate": 9.998954666015797e-06,
      "loss": 0.4365,
      "step": 1725
    },
    {
      "epoch": 0.07920701206920289,
      "grad_norm": 0.4351848363876343,
      "learning_rate": 9.998949646587798e-06,
      "loss": 0.3675,
      "step": 1726
    },
    {
      "epoch": 0.07925290257445734,
      "grad_norm": 0.5463142991065979,
      "learning_rate": 9.998944615138903e-06,
      "loss": 0.4844,
      "step": 1727
    },
    {
      "epoch": 0.07929879307971181,
      "grad_norm": 0.5145556926727295,
      "learning_rate": 9.998939571669123e-06,
      "loss": 0.5189,
      "step": 1728
    },
    {
      "epoch": 0.07934468358496627,
      "grad_norm": 0.493591845035553,
      "learning_rate": 9.998934516178471e-06,
      "loss": 0.4383,
      "step": 1729
    },
    {
      "epoch": 0.07939057409022074,
      "grad_norm": 0.4987589120864868,
      "learning_rate": 9.998929448666962e-06,
      "loss": 0.4852,
      "step": 1730
    },
    {
      "epoch": 0.0794364645954752,
      "grad_norm": 0.46000444889068604,
      "learning_rate": 9.998924369134601e-06,
      "loss": 0.3748,
      "step": 1731
    },
    {
      "epoch": 0.07948235510072967,
      "grad_norm": 0.4645262062549591,
      "learning_rate": 9.998919277581406e-06,
      "loss": 0.3641,
      "step": 1732
    },
    {
      "epoch": 0.07952824560598412,
      "grad_norm": 0.46884673833847046,
      "learning_rate": 9.99891417400739e-06,
      "loss": 0.3751,
      "step": 1733
    },
    {
      "epoch": 0.07957413611123859,
      "grad_norm": 0.51549232006073,
      "learning_rate": 9.99890905841256e-06,
      "loss": 0.4764,
      "step": 1734
    },
    {
      "epoch": 0.07962002661649305,
      "grad_norm": 0.45478355884552,
      "learning_rate": 9.998903930796935e-06,
      "loss": 0.368,
      "step": 1735
    },
    {
      "epoch": 0.0796659171217475,
      "grad_norm": 0.5080839395523071,
      "learning_rate": 9.99889879116052e-06,
      "loss": 0.4487,
      "step": 1736
    },
    {
      "epoch": 0.07971180762700197,
      "grad_norm": 0.4810938835144043,
      "learning_rate": 9.998893639503334e-06,
      "loss": 0.4262,
      "step": 1737
    },
    {
      "epoch": 0.07975769813225643,
      "grad_norm": 0.4891236424446106,
      "learning_rate": 9.998888475825385e-06,
      "loss": 0.3694,
      "step": 1738
    },
    {
      "epoch": 0.0798035886375109,
      "grad_norm": 0.506584107875824,
      "learning_rate": 9.998883300126687e-06,
      "loss": 0.4447,
      "step": 1739
    },
    {
      "epoch": 0.07984947914276536,
      "grad_norm": 0.5336281061172485,
      "learning_rate": 9.998878112407254e-06,
      "loss": 0.4713,
      "step": 1740
    },
    {
      "epoch": 0.07989536964801983,
      "grad_norm": 0.5062115788459778,
      "learning_rate": 9.998872912667094e-06,
      "loss": 0.5001,
      "step": 1741
    },
    {
      "epoch": 0.07994126015327428,
      "grad_norm": 0.4527471363544464,
      "learning_rate": 9.998867700906225e-06,
      "loss": 0.3423,
      "step": 1742
    },
    {
      "epoch": 0.07998715065852875,
      "grad_norm": 0.4902063310146332,
      "learning_rate": 9.998862477124656e-06,
      "loss": 0.4471,
      "step": 1743
    },
    {
      "epoch": 0.08003304116378321,
      "grad_norm": 0.5510109663009644,
      "learning_rate": 9.9988572413224e-06,
      "loss": 0.4737,
      "step": 1744
    },
    {
      "epoch": 0.08007893166903768,
      "grad_norm": 0.4556294083595276,
      "learning_rate": 9.99885199349947e-06,
      "loss": 0.3612,
      "step": 1745
    },
    {
      "epoch": 0.08012482217429213,
      "grad_norm": 0.4711126387119293,
      "learning_rate": 9.998846733655879e-06,
      "loss": 0.4123,
      "step": 1746
    },
    {
      "epoch": 0.0801707126795466,
      "grad_norm": 0.5133438110351562,
      "learning_rate": 9.99884146179164e-06,
      "loss": 0.437,
      "step": 1747
    },
    {
      "epoch": 0.08021660318480106,
      "grad_norm": 0.5058325529098511,
      "learning_rate": 9.998836177906763e-06,
      "loss": 0.385,
      "step": 1748
    },
    {
      "epoch": 0.08026249369005553,
      "grad_norm": 0.45551466941833496,
      "learning_rate": 9.998830882001263e-06,
      "loss": 0.3553,
      "step": 1749
    },
    {
      "epoch": 0.08030838419530999,
      "grad_norm": 0.4666111469268799,
      "learning_rate": 9.998825574075154e-06,
      "loss": 0.3602,
      "step": 1750
    },
    {
      "epoch": 0.08035427470056446,
      "grad_norm": 0.45222869515419006,
      "learning_rate": 9.998820254128446e-06,
      "loss": 0.3608,
      "step": 1751
    },
    {
      "epoch": 0.08040016520581891,
      "grad_norm": 0.4758358299732208,
      "learning_rate": 9.998814922161153e-06,
      "loss": 0.4731,
      "step": 1752
    },
    {
      "epoch": 0.08044605571107338,
      "grad_norm": 0.45364636182785034,
      "learning_rate": 9.998809578173288e-06,
      "loss": 0.3556,
      "step": 1753
    },
    {
      "epoch": 0.08049194621632784,
      "grad_norm": 0.4644677937030792,
      "learning_rate": 9.998804222164864e-06,
      "loss": 0.3477,
      "step": 1754
    },
    {
      "epoch": 0.08053783672158231,
      "grad_norm": 0.5056809782981873,
      "learning_rate": 9.998798854135894e-06,
      "loss": 0.4897,
      "step": 1755
    },
    {
      "epoch": 0.08058372722683677,
      "grad_norm": 0.4946051836013794,
      "learning_rate": 9.99879347408639e-06,
      "loss": 0.4263,
      "step": 1756
    },
    {
      "epoch": 0.08062961773209124,
      "grad_norm": 0.4351504445075989,
      "learning_rate": 9.998788082016364e-06,
      "loss": 0.3642,
      "step": 1757
    },
    {
      "epoch": 0.08067550823734569,
      "grad_norm": 0.48808377981185913,
      "learning_rate": 9.998782677925832e-06,
      "loss": 0.398,
      "step": 1758
    },
    {
      "epoch": 0.08072139874260016,
      "grad_norm": 0.4811568558216095,
      "learning_rate": 9.998777261814804e-06,
      "loss": 0.4454,
      "step": 1759
    },
    {
      "epoch": 0.08076728924785462,
      "grad_norm": 0.46869370341300964,
      "learning_rate": 9.998771833683294e-06,
      "loss": 0.3588,
      "step": 1760
    },
    {
      "epoch": 0.08081317975310909,
      "grad_norm": 0.45088866353034973,
      "learning_rate": 9.998766393531318e-06,
      "loss": 0.3422,
      "step": 1761
    },
    {
      "epoch": 0.08085907025836354,
      "grad_norm": 0.5257999897003174,
      "learning_rate": 9.998760941358884e-06,
      "loss": 0.5018,
      "step": 1762
    },
    {
      "epoch": 0.08090496076361801,
      "grad_norm": 0.49018165469169617,
      "learning_rate": 9.99875547716601e-06,
      "loss": 0.4011,
      "step": 1763
    },
    {
      "epoch": 0.08095085126887247,
      "grad_norm": 0.47185319662094116,
      "learning_rate": 9.998750000952703e-06,
      "loss": 0.4002,
      "step": 1764
    },
    {
      "epoch": 0.08099674177412693,
      "grad_norm": 0.4556969404220581,
      "learning_rate": 9.998744512718984e-06,
      "loss": 0.3721,
      "step": 1765
    },
    {
      "epoch": 0.0810426322793814,
      "grad_norm": 0.496563583612442,
      "learning_rate": 9.998739012464859e-06,
      "loss": 0.4128,
      "step": 1766
    },
    {
      "epoch": 0.08108852278463585,
      "grad_norm": 0.4539939761161804,
      "learning_rate": 9.998733500190345e-06,
      "loss": 0.3447,
      "step": 1767
    },
    {
      "epoch": 0.08113441328989032,
      "grad_norm": 0.4825995862483978,
      "learning_rate": 9.998727975895454e-06,
      "loss": 0.4731,
      "step": 1768
    },
    {
      "epoch": 0.08118030379514478,
      "grad_norm": 0.47168925404548645,
      "learning_rate": 9.998722439580203e-06,
      "loss": 0.4643,
      "step": 1769
    },
    {
      "epoch": 0.08122619430039925,
      "grad_norm": 0.5266677141189575,
      "learning_rate": 9.9987168912446e-06,
      "loss": 0.5071,
      "step": 1770
    },
    {
      "epoch": 0.0812720848056537,
      "grad_norm": 0.4776930809020996,
      "learning_rate": 9.99871133088866e-06,
      "loss": 0.4359,
      "step": 1771
    },
    {
      "epoch": 0.08131797531090817,
      "grad_norm": 0.476874440908432,
      "learning_rate": 9.998705758512397e-06,
      "loss": 0.453,
      "step": 1772
    },
    {
      "epoch": 0.08136386581616263,
      "grad_norm": 0.48035696148872375,
      "learning_rate": 9.998700174115825e-06,
      "loss": 0.4577,
      "step": 1773
    },
    {
      "epoch": 0.0814097563214171,
      "grad_norm": 0.5677808523178101,
      "learning_rate": 9.998694577698957e-06,
      "loss": 0.3808,
      "step": 1774
    },
    {
      "epoch": 0.08145564682667156,
      "grad_norm": 0.5057392120361328,
      "learning_rate": 9.998688969261806e-06,
      "loss": 0.4445,
      "step": 1775
    },
    {
      "epoch": 0.08150153733192603,
      "grad_norm": 0.46602535247802734,
      "learning_rate": 9.998683348804386e-06,
      "loss": 0.3993,
      "step": 1776
    },
    {
      "epoch": 0.08154742783718048,
      "grad_norm": 0.5494076609611511,
      "learning_rate": 9.998677716326707e-06,
      "loss": 0.5327,
      "step": 1777
    },
    {
      "epoch": 0.08159331834243495,
      "grad_norm": 0.4570852220058441,
      "learning_rate": 9.99867207182879e-06,
      "loss": 0.3816,
      "step": 1778
    },
    {
      "epoch": 0.08163920884768941,
      "grad_norm": 0.5257952809333801,
      "learning_rate": 9.998666415310642e-06,
      "loss": 0.5145,
      "step": 1779
    },
    {
      "epoch": 0.08168509935294388,
      "grad_norm": 0.5287887454032898,
      "learning_rate": 9.99866074677228e-06,
      "loss": 0.525,
      "step": 1780
    },
    {
      "epoch": 0.08173098985819834,
      "grad_norm": 0.46291640400886536,
      "learning_rate": 9.998655066213716e-06,
      "loss": 0.339,
      "step": 1781
    },
    {
      "epoch": 0.0817768803634528,
      "grad_norm": 0.48628246784210205,
      "learning_rate": 9.998649373634963e-06,
      "loss": 0.4235,
      "step": 1782
    },
    {
      "epoch": 0.08182277086870726,
      "grad_norm": 0.46783190965652466,
      "learning_rate": 9.998643669036038e-06,
      "loss": 0.4047,
      "step": 1783
    },
    {
      "epoch": 0.08186866137396173,
      "grad_norm": 0.4377851188182831,
      "learning_rate": 9.998637952416952e-06,
      "loss": 0.3505,
      "step": 1784
    },
    {
      "epoch": 0.08191455187921619,
      "grad_norm": 0.5219147801399231,
      "learning_rate": 9.998632223777718e-06,
      "loss": 0.4853,
      "step": 1785
    },
    {
      "epoch": 0.08196044238447066,
      "grad_norm": 0.5514872670173645,
      "learning_rate": 9.998626483118352e-06,
      "loss": 0.5059,
      "step": 1786
    },
    {
      "epoch": 0.08200633288972511,
      "grad_norm": 0.48154398798942566,
      "learning_rate": 9.998620730438868e-06,
      "loss": 0.3723,
      "step": 1787
    },
    {
      "epoch": 0.08205222339497958,
      "grad_norm": 0.4764332175254822,
      "learning_rate": 9.998614965739278e-06,
      "loss": 0.4291,
      "step": 1788
    },
    {
      "epoch": 0.08209811390023404,
      "grad_norm": 0.49761006236076355,
      "learning_rate": 9.998609189019594e-06,
      "loss": 0.4565,
      "step": 1789
    },
    {
      "epoch": 0.08214400440548851,
      "grad_norm": 0.49671006202697754,
      "learning_rate": 9.998603400279837e-06,
      "loss": 0.4208,
      "step": 1790
    },
    {
      "epoch": 0.08218989491074297,
      "grad_norm": 0.47068002820014954,
      "learning_rate": 9.998597599520013e-06,
      "loss": 0.3703,
      "step": 1791
    },
    {
      "epoch": 0.08223578541599744,
      "grad_norm": 0.4687693417072296,
      "learning_rate": 9.998591786740142e-06,
      "loss": 0.4059,
      "step": 1792
    },
    {
      "epoch": 0.08228167592125189,
      "grad_norm": 0.46863317489624023,
      "learning_rate": 9.998585961940232e-06,
      "loss": 0.3568,
      "step": 1793
    },
    {
      "epoch": 0.08232756642650635,
      "grad_norm": 0.453643798828125,
      "learning_rate": 9.998580125120303e-06,
      "loss": 0.3677,
      "step": 1794
    },
    {
      "epoch": 0.08237345693176082,
      "grad_norm": 0.4481407403945923,
      "learning_rate": 9.998574276280367e-06,
      "loss": 0.3202,
      "step": 1795
    },
    {
      "epoch": 0.08241934743701528,
      "grad_norm": 0.5013203024864197,
      "learning_rate": 9.998568415420435e-06,
      "loss": 0.4275,
      "step": 1796
    },
    {
      "epoch": 0.08246523794226975,
      "grad_norm": 0.5677851438522339,
      "learning_rate": 9.998562542540526e-06,
      "loss": 0.5211,
      "step": 1797
    },
    {
      "epoch": 0.0825111284475242,
      "grad_norm": 0.47754719853401184,
      "learning_rate": 9.99855665764065e-06,
      "loss": 0.4155,
      "step": 1798
    },
    {
      "epoch": 0.08255701895277867,
      "grad_norm": 0.49698135256767273,
      "learning_rate": 9.998550760720823e-06,
      "loss": 0.391,
      "step": 1799
    },
    {
      "epoch": 0.08260290945803313,
      "grad_norm": 0.5295330286026001,
      "learning_rate": 9.99854485178106e-06,
      "loss": 0.4011,
      "step": 1800
    },
    {
      "epoch": 0.0826487999632876,
      "grad_norm": 0.4918323755264282,
      "learning_rate": 9.998538930821373e-06,
      "loss": 0.3698,
      "step": 1801
    },
    {
      "epoch": 0.08269469046854205,
      "grad_norm": 0.5112568736076355,
      "learning_rate": 9.998532997841778e-06,
      "loss": 0.425,
      "step": 1802
    },
    {
      "epoch": 0.08274058097379652,
      "grad_norm": 0.5109966993331909,
      "learning_rate": 9.998527052842289e-06,
      "loss": 0.4505,
      "step": 1803
    },
    {
      "epoch": 0.08278647147905098,
      "grad_norm": 0.5352508425712585,
      "learning_rate": 9.99852109582292e-06,
      "loss": 0.4678,
      "step": 1804
    },
    {
      "epoch": 0.08283236198430545,
      "grad_norm": 0.5146759748458862,
      "learning_rate": 9.998515126783685e-06,
      "loss": 0.4879,
      "step": 1805
    },
    {
      "epoch": 0.0828782524895599,
      "grad_norm": 0.4801282286643982,
      "learning_rate": 9.998509145724597e-06,
      "loss": 0.4034,
      "step": 1806
    },
    {
      "epoch": 0.08292414299481438,
      "grad_norm": 0.48487892746925354,
      "learning_rate": 9.998503152645675e-06,
      "loss": 0.4516,
      "step": 1807
    },
    {
      "epoch": 0.08297003350006883,
      "grad_norm": 0.5353267788887024,
      "learning_rate": 9.99849714754693e-06,
      "loss": 0.4417,
      "step": 1808
    },
    {
      "epoch": 0.0830159240053233,
      "grad_norm": 0.5028082132339478,
      "learning_rate": 9.998491130428376e-06,
      "loss": 0.4558,
      "step": 1809
    },
    {
      "epoch": 0.08306181451057776,
      "grad_norm": 0.49690622091293335,
      "learning_rate": 9.998485101290028e-06,
      "loss": 0.4304,
      "step": 1810
    },
    {
      "epoch": 0.08310770501583223,
      "grad_norm": 0.4832867980003357,
      "learning_rate": 9.9984790601319e-06,
      "loss": 0.3484,
      "step": 1811
    },
    {
      "epoch": 0.08315359552108668,
      "grad_norm": 0.46231919527053833,
      "learning_rate": 9.99847300695401e-06,
      "loss": 0.3558,
      "step": 1812
    },
    {
      "epoch": 0.08319948602634115,
      "grad_norm": 0.4444623589515686,
      "learning_rate": 9.99846694175637e-06,
      "loss": 0.3532,
      "step": 1813
    },
    {
      "epoch": 0.08324537653159561,
      "grad_norm": 0.493055135011673,
      "learning_rate": 9.998460864538992e-06,
      "loss": 0.4317,
      "step": 1814
    },
    {
      "epoch": 0.08329126703685008,
      "grad_norm": 0.47260966897010803,
      "learning_rate": 9.998454775301896e-06,
      "loss": 0.402,
      "step": 1815
    },
    {
      "epoch": 0.08333715754210454,
      "grad_norm": 0.49794045090675354,
      "learning_rate": 9.998448674045093e-06,
      "loss": 0.4521,
      "step": 1816
    },
    {
      "epoch": 0.083383048047359,
      "grad_norm": 0.47017616033554077,
      "learning_rate": 9.998442560768597e-06,
      "loss": 0.401,
      "step": 1817
    },
    {
      "epoch": 0.08342893855261346,
      "grad_norm": 0.5197916030883789,
      "learning_rate": 9.998436435472426e-06,
      "loss": 0.4634,
      "step": 1818
    },
    {
      "epoch": 0.08347482905786793,
      "grad_norm": 0.4560147821903229,
      "learning_rate": 9.998430298156591e-06,
      "loss": 0.4195,
      "step": 1819
    },
    {
      "epoch": 0.08352071956312239,
      "grad_norm": 0.49685657024383545,
      "learning_rate": 9.99842414882111e-06,
      "loss": 0.4512,
      "step": 1820
    },
    {
      "epoch": 0.08356661006837686,
      "grad_norm": 0.49072790145874023,
      "learning_rate": 9.998417987465998e-06,
      "loss": 0.4078,
      "step": 1821
    },
    {
      "epoch": 0.08361250057363132,
      "grad_norm": 0.5176905393600464,
      "learning_rate": 9.998411814091266e-06,
      "loss": 0.4949,
      "step": 1822
    },
    {
      "epoch": 0.08365839107888579,
      "grad_norm": 0.4655018150806427,
      "learning_rate": 9.998405628696931e-06,
      "loss": 0.3251,
      "step": 1823
    },
    {
      "epoch": 0.08370428158414024,
      "grad_norm": 0.48832565546035767,
      "learning_rate": 9.99839943128301e-06,
      "loss": 0.4302,
      "step": 1824
    },
    {
      "epoch": 0.0837501720893947,
      "grad_norm": 0.5239647030830383,
      "learning_rate": 9.998393221849514e-06,
      "loss": 0.4541,
      "step": 1825
    },
    {
      "epoch": 0.08379606259464917,
      "grad_norm": 0.5271238684654236,
      "learning_rate": 9.99838700039646e-06,
      "loss": 0.5473,
      "step": 1826
    },
    {
      "epoch": 0.08384195309990362,
      "grad_norm": 0.47936439514160156,
      "learning_rate": 9.998380766923864e-06,
      "loss": 0.4519,
      "step": 1827
    },
    {
      "epoch": 0.0838878436051581,
      "grad_norm": 0.4982088506221771,
      "learning_rate": 9.998374521431739e-06,
      "loss": 0.3711,
      "step": 1828
    },
    {
      "epoch": 0.08393373411041255,
      "grad_norm": 0.5050623416900635,
      "learning_rate": 9.9983682639201e-06,
      "loss": 0.402,
      "step": 1829
    },
    {
      "epoch": 0.08397962461566702,
      "grad_norm": 0.42724063992500305,
      "learning_rate": 9.998361994388964e-06,
      "loss": 0.358,
      "step": 1830
    },
    {
      "epoch": 0.08402551512092148,
      "grad_norm": 0.4541013836860657,
      "learning_rate": 9.998355712838345e-06,
      "loss": 0.3718,
      "step": 1831
    },
    {
      "epoch": 0.08407140562617595,
      "grad_norm": 0.42599621415138245,
      "learning_rate": 9.998349419268257e-06,
      "loss": 0.3117,
      "step": 1832
    },
    {
      "epoch": 0.0841172961314304,
      "grad_norm": 0.5127098560333252,
      "learning_rate": 9.998343113678716e-06,
      "loss": 0.4095,
      "step": 1833
    },
    {
      "epoch": 0.08416318663668487,
      "grad_norm": 0.5799052119255066,
      "learning_rate": 9.99833679606974e-06,
      "loss": 0.4023,
      "step": 1834
    },
    {
      "epoch": 0.08420907714193933,
      "grad_norm": 0.5320877432823181,
      "learning_rate": 9.99833046644134e-06,
      "loss": 0.458,
      "step": 1835
    },
    {
      "epoch": 0.0842549676471938,
      "grad_norm": 0.5162080526351929,
      "learning_rate": 9.998324124793531e-06,
      "loss": 0.5065,
      "step": 1836
    },
    {
      "epoch": 0.08430085815244825,
      "grad_norm": 0.5115873217582703,
      "learning_rate": 9.998317771126332e-06,
      "loss": 0.4371,
      "step": 1837
    },
    {
      "epoch": 0.08434674865770272,
      "grad_norm": 0.5874608159065247,
      "learning_rate": 9.998311405439756e-06,
      "loss": 0.4531,
      "step": 1838
    },
    {
      "epoch": 0.08439263916295718,
      "grad_norm": 0.4815157651901245,
      "learning_rate": 9.99830502773382e-06,
      "loss": 0.429,
      "step": 1839
    },
    {
      "epoch": 0.08443852966821165,
      "grad_norm": 0.48297929763793945,
      "learning_rate": 9.998298638008535e-06,
      "loss": 0.4392,
      "step": 1840
    },
    {
      "epoch": 0.08448442017346611,
      "grad_norm": 0.4752102792263031,
      "learning_rate": 9.99829223626392e-06,
      "loss": 0.3882,
      "step": 1841
    },
    {
      "epoch": 0.08453031067872058,
      "grad_norm": 0.4992143213748932,
      "learning_rate": 9.998285822499991e-06,
      "loss": 0.3998,
      "step": 1842
    },
    {
      "epoch": 0.08457620118397503,
      "grad_norm": 0.4586814343929291,
      "learning_rate": 9.99827939671676e-06,
      "loss": 0.4086,
      "step": 1843
    },
    {
      "epoch": 0.0846220916892295,
      "grad_norm": 0.46860653162002563,
      "learning_rate": 9.998272958914245e-06,
      "loss": 0.4088,
      "step": 1844
    },
    {
      "epoch": 0.08466798219448396,
      "grad_norm": 0.45195892453193665,
      "learning_rate": 9.998266509092464e-06,
      "loss": 0.342,
      "step": 1845
    },
    {
      "epoch": 0.08471387269973843,
      "grad_norm": 0.5009403228759766,
      "learning_rate": 9.998260047251427e-06,
      "loss": 0.4964,
      "step": 1846
    },
    {
      "epoch": 0.08475976320499289,
      "grad_norm": 0.5115389227867126,
      "learning_rate": 9.998253573391153e-06,
      "loss": 0.4795,
      "step": 1847
    },
    {
      "epoch": 0.08480565371024736,
      "grad_norm": 0.5265485644340515,
      "learning_rate": 9.998247087511655e-06,
      "loss": 0.5418,
      "step": 1848
    },
    {
      "epoch": 0.08485154421550181,
      "grad_norm": 0.4805338680744171,
      "learning_rate": 9.998240589612951e-06,
      "loss": 0.3822,
      "step": 1849
    },
    {
      "epoch": 0.08489743472075628,
      "grad_norm": 0.614960253238678,
      "learning_rate": 9.998234079695056e-06,
      "loss": 0.4183,
      "step": 1850
    },
    {
      "epoch": 0.08494332522601074,
      "grad_norm": 0.45898082852363586,
      "learning_rate": 9.998227557757986e-06,
      "loss": 0.32,
      "step": 1851
    },
    {
      "epoch": 0.08498921573126521,
      "grad_norm": 0.48527857661247253,
      "learning_rate": 9.998221023801757e-06,
      "loss": 0.4043,
      "step": 1852
    },
    {
      "epoch": 0.08503510623651966,
      "grad_norm": 0.4501403272151947,
      "learning_rate": 9.998214477826382e-06,
      "loss": 0.3178,
      "step": 1853
    },
    {
      "epoch": 0.08508099674177412,
      "grad_norm": 0.4642013609409332,
      "learning_rate": 9.998207919831881e-06,
      "loss": 0.3559,
      "step": 1854
    },
    {
      "epoch": 0.08512688724702859,
      "grad_norm": 0.5030194520950317,
      "learning_rate": 9.998201349818266e-06,
      "loss": 0.4838,
      "step": 1855
    },
    {
      "epoch": 0.08517277775228305,
      "grad_norm": 0.7251027226448059,
      "learning_rate": 9.998194767785553e-06,
      "loss": 0.4935,
      "step": 1856
    },
    {
      "epoch": 0.08521866825753752,
      "grad_norm": 0.4636625647544861,
      "learning_rate": 9.99818817373376e-06,
      "loss": 0.3633,
      "step": 1857
    },
    {
      "epoch": 0.08526455876279197,
      "grad_norm": 0.48684799671173096,
      "learning_rate": 9.998181567662904e-06,
      "loss": 0.4078,
      "step": 1858
    },
    {
      "epoch": 0.08531044926804644,
      "grad_norm": 0.5248304605484009,
      "learning_rate": 9.998174949572997e-06,
      "loss": 0.4351,
      "step": 1859
    },
    {
      "epoch": 0.0853563397733009,
      "grad_norm": 0.4800574779510498,
      "learning_rate": 9.998168319464057e-06,
      "loss": 0.4359,
      "step": 1860
    },
    {
      "epoch": 0.08540223027855537,
      "grad_norm": 0.436203271150589,
      "learning_rate": 9.998161677336098e-06,
      "loss": 0.3411,
      "step": 1861
    },
    {
      "epoch": 0.08544812078380983,
      "grad_norm": 0.5083044767379761,
      "learning_rate": 9.99815502318914e-06,
      "loss": 0.4054,
      "step": 1862
    },
    {
      "epoch": 0.0854940112890643,
      "grad_norm": 0.4430050849914551,
      "learning_rate": 9.998148357023196e-06,
      "loss": 0.3311,
      "step": 1863
    },
    {
      "epoch": 0.08553990179431875,
      "grad_norm": 0.543315052986145,
      "learning_rate": 9.998141678838283e-06,
      "loss": 0.497,
      "step": 1864
    },
    {
      "epoch": 0.08558579229957322,
      "grad_norm": 0.5219607353210449,
      "learning_rate": 9.998134988634414e-06,
      "loss": 0.5208,
      "step": 1865
    },
    {
      "epoch": 0.08563168280482768,
      "grad_norm": 0.4644080698490143,
      "learning_rate": 9.998128286411611e-06,
      "loss": 0.3786,
      "step": 1866
    },
    {
      "epoch": 0.08567757331008215,
      "grad_norm": 0.4564545750617981,
      "learning_rate": 9.998121572169886e-06,
      "loss": 0.3495,
      "step": 1867
    },
    {
      "epoch": 0.0857234638153366,
      "grad_norm": 0.4753549098968506,
      "learning_rate": 9.998114845909255e-06,
      "loss": 0.3941,
      "step": 1868
    },
    {
      "epoch": 0.08576935432059107,
      "grad_norm": 0.42606431245803833,
      "learning_rate": 9.998108107629736e-06,
      "loss": 0.3202,
      "step": 1869
    },
    {
      "epoch": 0.08581524482584553,
      "grad_norm": 0.4720016419887543,
      "learning_rate": 9.998101357331345e-06,
      "loss": 0.4031,
      "step": 1870
    },
    {
      "epoch": 0.0858611353311,
      "grad_norm": 0.5052058100700378,
      "learning_rate": 9.998094595014098e-06,
      "loss": 0.4904,
      "step": 1871
    },
    {
      "epoch": 0.08590702583635446,
      "grad_norm": 0.5914888978004456,
      "learning_rate": 9.99808782067801e-06,
      "loss": 0.3381,
      "step": 1872
    },
    {
      "epoch": 0.08595291634160893,
      "grad_norm": 0.4664536714553833,
      "learning_rate": 9.998081034323098e-06,
      "loss": 0.4378,
      "step": 1873
    },
    {
      "epoch": 0.08599880684686338,
      "grad_norm": 0.4502689838409424,
      "learning_rate": 9.99807423594938e-06,
      "loss": 0.3745,
      "step": 1874
    },
    {
      "epoch": 0.08604469735211785,
      "grad_norm": 0.4514409303665161,
      "learning_rate": 9.998067425556869e-06,
      "loss": 0.3836,
      "step": 1875
    },
    {
      "epoch": 0.08609058785737231,
      "grad_norm": 0.509328305721283,
      "learning_rate": 9.998060603145584e-06,
      "loss": 0.5113,
      "step": 1876
    },
    {
      "epoch": 0.08613647836262678,
      "grad_norm": 0.4698704779148102,
      "learning_rate": 9.99805376871554e-06,
      "loss": 0.3526,
      "step": 1877
    },
    {
      "epoch": 0.08618236886788123,
      "grad_norm": 0.5010122656822205,
      "learning_rate": 9.998046922266755e-06,
      "loss": 0.4637,
      "step": 1878
    },
    {
      "epoch": 0.0862282593731357,
      "grad_norm": 0.5294129848480225,
      "learning_rate": 9.998040063799246e-06,
      "loss": 0.5403,
      "step": 1879
    },
    {
      "epoch": 0.08627414987839016,
      "grad_norm": 0.5025519728660583,
      "learning_rate": 9.998033193313026e-06,
      "loss": 0.5209,
      "step": 1880
    },
    {
      "epoch": 0.08632004038364463,
      "grad_norm": 0.5287100076675415,
      "learning_rate": 9.998026310808114e-06,
      "loss": 0.4951,
      "step": 1881
    },
    {
      "epoch": 0.08636593088889909,
      "grad_norm": 0.4639873802661896,
      "learning_rate": 9.998019416284527e-06,
      "loss": 0.4193,
      "step": 1882
    },
    {
      "epoch": 0.08641182139415354,
      "grad_norm": 0.5831707119941711,
      "learning_rate": 9.99801250974228e-06,
      "loss": 0.5465,
      "step": 1883
    },
    {
      "epoch": 0.08645771189940801,
      "grad_norm": 0.46959373354911804,
      "learning_rate": 9.99800559118139e-06,
      "loss": 0.3839,
      "step": 1884
    },
    {
      "epoch": 0.08650360240466247,
      "grad_norm": 0.4728989005088806,
      "learning_rate": 9.997998660601876e-06,
      "loss": 0.4195,
      "step": 1885
    },
    {
      "epoch": 0.08654949290991694,
      "grad_norm": 0.4978037178516388,
      "learning_rate": 9.99799171800375e-06,
      "loss": 0.4834,
      "step": 1886
    },
    {
      "epoch": 0.0865953834151714,
      "grad_norm": 0.5176402926445007,
      "learning_rate": 9.997984763387032e-06,
      "loss": 0.4728,
      "step": 1887
    },
    {
      "epoch": 0.08664127392042587,
      "grad_norm": 0.4842921495437622,
      "learning_rate": 9.997977796751739e-06,
      "loss": 0.4632,
      "step": 1888
    },
    {
      "epoch": 0.08668716442568032,
      "grad_norm": 0.4526640474796295,
      "learning_rate": 9.997970818097886e-06,
      "loss": 0.374,
      "step": 1889
    },
    {
      "epoch": 0.08673305493093479,
      "grad_norm": 0.5068642497062683,
      "learning_rate": 9.99796382742549e-06,
      "loss": 0.4516,
      "step": 1890
    },
    {
      "epoch": 0.08677894543618925,
      "grad_norm": 0.4842502176761627,
      "learning_rate": 9.997956824734569e-06,
      "loss": 0.4543,
      "step": 1891
    },
    {
      "epoch": 0.08682483594144372,
      "grad_norm": 0.4825424253940582,
      "learning_rate": 9.99794981002514e-06,
      "loss": 0.4322,
      "step": 1892
    },
    {
      "epoch": 0.08687072644669817,
      "grad_norm": 0.4785364866256714,
      "learning_rate": 9.997942783297219e-06,
      "loss": 0.4509,
      "step": 1893
    },
    {
      "epoch": 0.08691661695195264,
      "grad_norm": 0.47149842977523804,
      "learning_rate": 9.997935744550823e-06,
      "loss": 0.396,
      "step": 1894
    },
    {
      "epoch": 0.0869625074572071,
      "grad_norm": 0.51544189453125,
      "learning_rate": 9.997928693785967e-06,
      "loss": 0.5019,
      "step": 1895
    },
    {
      "epoch": 0.08700839796246157,
      "grad_norm": 0.4829908609390259,
      "learning_rate": 9.997921631002671e-06,
      "loss": 0.479,
      "step": 1896
    },
    {
      "epoch": 0.08705428846771603,
      "grad_norm": 0.48497772216796875,
      "learning_rate": 9.997914556200952e-06,
      "loss": 0.3736,
      "step": 1897
    },
    {
      "epoch": 0.0871001789729705,
      "grad_norm": 0.5008928775787354,
      "learning_rate": 9.997907469380825e-06,
      "loss": 0.416,
      "step": 1898
    },
    {
      "epoch": 0.08714606947822495,
      "grad_norm": 0.525834858417511,
      "learning_rate": 9.997900370542308e-06,
      "loss": 0.4766,
      "step": 1899
    },
    {
      "epoch": 0.08719195998347942,
      "grad_norm": 0.5160179734230042,
      "learning_rate": 9.997893259685418e-06,
      "loss": 0.5139,
      "step": 1900
    },
    {
      "epoch": 0.08723785048873388,
      "grad_norm": 0.5106474161148071,
      "learning_rate": 9.997886136810173e-06,
      "loss": 0.4794,
      "step": 1901
    },
    {
      "epoch": 0.08728374099398835,
      "grad_norm": 0.45013463497161865,
      "learning_rate": 9.997879001916589e-06,
      "loss": 0.3253,
      "step": 1902
    },
    {
      "epoch": 0.0873296314992428,
      "grad_norm": 0.4547373056411743,
      "learning_rate": 9.997871855004681e-06,
      "loss": 0.3554,
      "step": 1903
    },
    {
      "epoch": 0.08737552200449727,
      "grad_norm": 0.5522216558456421,
      "learning_rate": 9.99786469607447e-06,
      "loss": 0.4373,
      "step": 1904
    },
    {
      "epoch": 0.08742141250975173,
      "grad_norm": 0.5390852093696594,
      "learning_rate": 9.997857525125974e-06,
      "loss": 0.4526,
      "step": 1905
    },
    {
      "epoch": 0.0874673030150062,
      "grad_norm": 0.4166499674320221,
      "learning_rate": 9.997850342159206e-06,
      "loss": 0.3042,
      "step": 1906
    },
    {
      "epoch": 0.08751319352026066,
      "grad_norm": 0.5154944658279419,
      "learning_rate": 9.997843147174184e-06,
      "loss": 0.4139,
      "step": 1907
    },
    {
      "epoch": 0.08755908402551513,
      "grad_norm": 0.4787823259830475,
      "learning_rate": 9.997835940170928e-06,
      "loss": 0.4075,
      "step": 1908
    },
    {
      "epoch": 0.08760497453076958,
      "grad_norm": 0.48952722549438477,
      "learning_rate": 9.997828721149454e-06,
      "loss": 0.414,
      "step": 1909
    },
    {
      "epoch": 0.08765086503602405,
      "grad_norm": 0.517336368560791,
      "learning_rate": 9.997821490109779e-06,
      "loss": 0.4964,
      "step": 1910
    },
    {
      "epoch": 0.08769675554127851,
      "grad_norm": 0.47285598516464233,
      "learning_rate": 9.99781424705192e-06,
      "loss": 0.4142,
      "step": 1911
    },
    {
      "epoch": 0.08774264604653297,
      "grad_norm": 0.4789881110191345,
      "learning_rate": 9.997806991975896e-06,
      "loss": 0.4791,
      "step": 1912
    },
    {
      "epoch": 0.08778853655178744,
      "grad_norm": 0.44119635224342346,
      "learning_rate": 9.997799724881725e-06,
      "loss": 0.336,
      "step": 1913
    },
    {
      "epoch": 0.08783442705704189,
      "grad_norm": 0.48285871744155884,
      "learning_rate": 9.99779244576942e-06,
      "loss": 0.4268,
      "step": 1914
    },
    {
      "epoch": 0.08788031756229636,
      "grad_norm": 0.4547489881515503,
      "learning_rate": 9.997785154639004e-06,
      "loss": 0.3245,
      "step": 1915
    },
    {
      "epoch": 0.08792620806755082,
      "grad_norm": 0.5265078544616699,
      "learning_rate": 9.997777851490492e-06,
      "loss": 0.3546,
      "step": 1916
    },
    {
      "epoch": 0.08797209857280529,
      "grad_norm": 0.4521884322166443,
      "learning_rate": 9.9977705363239e-06,
      "loss": 0.3176,
      "step": 1917
    },
    {
      "epoch": 0.08801798907805974,
      "grad_norm": 0.47659745812416077,
      "learning_rate": 9.997763209139248e-06,
      "loss": 0.3128,
      "step": 1918
    },
    {
      "epoch": 0.08806387958331421,
      "grad_norm": 0.45969158411026,
      "learning_rate": 9.997755869936552e-06,
      "loss": 0.4127,
      "step": 1919
    },
    {
      "epoch": 0.08810977008856867,
      "grad_norm": 0.46879997849464417,
      "learning_rate": 9.99774851871583e-06,
      "loss": 0.3375,
      "step": 1920
    },
    {
      "epoch": 0.08815566059382314,
      "grad_norm": 0.4860617220401764,
      "learning_rate": 9.997741155477103e-06,
      "loss": 0.4345,
      "step": 1921
    },
    {
      "epoch": 0.0882015510990776,
      "grad_norm": 0.4547886252403259,
      "learning_rate": 9.997733780220384e-06,
      "loss": 0.3451,
      "step": 1922
    },
    {
      "epoch": 0.08824744160433207,
      "grad_norm": 0.5249549150466919,
      "learning_rate": 9.997726392945693e-06,
      "loss": 0.5316,
      "step": 1923
    },
    {
      "epoch": 0.08829333210958652,
      "grad_norm": 0.5182896256446838,
      "learning_rate": 9.997718993653048e-06,
      "loss": 0.5236,
      "step": 1924
    },
    {
      "epoch": 0.08833922261484099,
      "grad_norm": 0.6074200868606567,
      "learning_rate": 9.997711582342464e-06,
      "loss": 0.4571,
      "step": 1925
    },
    {
      "epoch": 0.08838511312009545,
      "grad_norm": 0.4656011760234833,
      "learning_rate": 9.997704159013961e-06,
      "loss": 0.3757,
      "step": 1926
    },
    {
      "epoch": 0.08843100362534992,
      "grad_norm": 0.46809640526771545,
      "learning_rate": 9.997696723667559e-06,
      "loss": 0.4061,
      "step": 1927
    },
    {
      "epoch": 0.08847689413060437,
      "grad_norm": 0.46747127175331116,
      "learning_rate": 9.997689276303273e-06,
      "loss": 0.4115,
      "step": 1928
    },
    {
      "epoch": 0.08852278463585884,
      "grad_norm": 0.56339430809021,
      "learning_rate": 9.997681816921121e-06,
      "loss": 0.5307,
      "step": 1929
    },
    {
      "epoch": 0.0885686751411133,
      "grad_norm": 0.4788985848426819,
      "learning_rate": 9.997674345521122e-06,
      "loss": 0.4362,
      "step": 1930
    },
    {
      "epoch": 0.08861456564636777,
      "grad_norm": 0.48429176211357117,
      "learning_rate": 9.997666862103294e-06,
      "loss": 0.4138,
      "step": 1931
    },
    {
      "epoch": 0.08866045615162223,
      "grad_norm": 0.42795562744140625,
      "learning_rate": 9.997659366667653e-06,
      "loss": 0.3495,
      "step": 1932
    },
    {
      "epoch": 0.0887063466568767,
      "grad_norm": 0.5259376764297485,
      "learning_rate": 9.99765185921422e-06,
      "loss": 0.4898,
      "step": 1933
    },
    {
      "epoch": 0.08875223716213115,
      "grad_norm": 0.5304051637649536,
      "learning_rate": 9.997644339743012e-06,
      "loss": 0.4647,
      "step": 1934
    },
    {
      "epoch": 0.08879812766738562,
      "grad_norm": 0.5148341655731201,
      "learning_rate": 9.997636808254045e-06,
      "loss": 0.495,
      "step": 1935
    },
    {
      "epoch": 0.08884401817264008,
      "grad_norm": 0.4492471218109131,
      "learning_rate": 9.997629264747339e-06,
      "loss": 0.4242,
      "step": 1936
    },
    {
      "epoch": 0.08888990867789455,
      "grad_norm": 0.48096388578414917,
      "learning_rate": 9.997621709222912e-06,
      "loss": 0.3927,
      "step": 1937
    },
    {
      "epoch": 0.088935799183149,
      "grad_norm": 0.5156227946281433,
      "learning_rate": 9.997614141680783e-06,
      "loss": 0.4571,
      "step": 1938
    },
    {
      "epoch": 0.08898168968840348,
      "grad_norm": 0.45808881521224976,
      "learning_rate": 9.997606562120968e-06,
      "loss": 0.3624,
      "step": 1939
    },
    {
      "epoch": 0.08902758019365793,
      "grad_norm": 0.5129602551460266,
      "learning_rate": 9.997598970543487e-06,
      "loss": 0.4317,
      "step": 1940
    },
    {
      "epoch": 0.08907347069891239,
      "grad_norm": 0.4733436107635498,
      "learning_rate": 9.997591366948358e-06,
      "loss": 0.4045,
      "step": 1941
    },
    {
      "epoch": 0.08911936120416686,
      "grad_norm": 0.45796629786491394,
      "learning_rate": 9.9975837513356e-06,
      "loss": 0.4217,
      "step": 1942
    },
    {
      "epoch": 0.08916525170942131,
      "grad_norm": 0.4778556823730469,
      "learning_rate": 9.997576123705228e-06,
      "loss": 0.3812,
      "step": 1943
    },
    {
      "epoch": 0.08921114221467578,
      "grad_norm": 0.4669647812843323,
      "learning_rate": 9.997568484057266e-06,
      "loss": 0.3546,
      "step": 1944
    },
    {
      "epoch": 0.08925703271993024,
      "grad_norm": 0.4887997508049011,
      "learning_rate": 9.997560832391726e-06,
      "loss": 0.4292,
      "step": 1945
    },
    {
      "epoch": 0.08930292322518471,
      "grad_norm": 0.48549383878707886,
      "learning_rate": 9.997553168708631e-06,
      "loss": 0.393,
      "step": 1946
    },
    {
      "epoch": 0.08934881373043917,
      "grad_norm": 0.48522454500198364,
      "learning_rate": 9.997545493007997e-06,
      "loss": 0.482,
      "step": 1947
    },
    {
      "epoch": 0.08939470423569364,
      "grad_norm": 0.5656195878982544,
      "learning_rate": 9.997537805289844e-06,
      "loss": 0.6164,
      "step": 1948
    },
    {
      "epoch": 0.08944059474094809,
      "grad_norm": 0.47381073236465454,
      "learning_rate": 9.997530105554189e-06,
      "loss": 0.3774,
      "step": 1949
    },
    {
      "epoch": 0.08948648524620256,
      "grad_norm": 0.4426414966583252,
      "learning_rate": 9.997522393801054e-06,
      "loss": 0.3824,
      "step": 1950
    },
    {
      "epoch": 0.08953237575145702,
      "grad_norm": 0.4970719516277313,
      "learning_rate": 9.997514670030453e-06,
      "loss": 0.497,
      "step": 1951
    },
    {
      "epoch": 0.08957826625671149,
      "grad_norm": 0.4999612867832184,
      "learning_rate": 9.997506934242406e-06,
      "loss": 0.4459,
      "step": 1952
    },
    {
      "epoch": 0.08962415676196595,
      "grad_norm": 0.522359311580658,
      "learning_rate": 9.997499186436932e-06,
      "loss": 0.5015,
      "step": 1953
    },
    {
      "epoch": 0.08967004726722042,
      "grad_norm": 0.482906699180603,
      "learning_rate": 9.99749142661405e-06,
      "loss": 0.429,
      "step": 1954
    },
    {
      "epoch": 0.08971593777247487,
      "grad_norm": 0.47550684213638306,
      "learning_rate": 9.997483654773779e-06,
      "loss": 0.3637,
      "step": 1955
    },
    {
      "epoch": 0.08976182827772934,
      "grad_norm": 0.45095688104629517,
      "learning_rate": 9.997475870916138e-06,
      "loss": 0.3617,
      "step": 1956
    },
    {
      "epoch": 0.0898077187829838,
      "grad_norm": 0.5475599765777588,
      "learning_rate": 9.997468075041142e-06,
      "loss": 0.5413,
      "step": 1957
    },
    {
      "epoch": 0.08985360928823827,
      "grad_norm": 0.4633029103279114,
      "learning_rate": 9.997460267148816e-06,
      "loss": 0.4015,
      "step": 1958
    },
    {
      "epoch": 0.08989949979349272,
      "grad_norm": 0.46834853291511536,
      "learning_rate": 9.997452447239172e-06,
      "loss": 0.4489,
      "step": 1959
    },
    {
      "epoch": 0.0899453902987472,
      "grad_norm": 0.5147649645805359,
      "learning_rate": 9.997444615312233e-06,
      "loss": 0.4254,
      "step": 1960
    },
    {
      "epoch": 0.08999128080400165,
      "grad_norm": 0.4903944432735443,
      "learning_rate": 9.997436771368018e-06,
      "loss": 0.4514,
      "step": 1961
    },
    {
      "epoch": 0.09003717130925612,
      "grad_norm": 0.5154668092727661,
      "learning_rate": 9.997428915406545e-06,
      "loss": 0.3971,
      "step": 1962
    },
    {
      "epoch": 0.09008306181451058,
      "grad_norm": 0.5075410008430481,
      "learning_rate": 9.99742104742783e-06,
      "loss": 0.4101,
      "step": 1963
    },
    {
      "epoch": 0.09012895231976505,
      "grad_norm": 0.49641481041908264,
      "learning_rate": 9.997413167431897e-06,
      "loss": 0.4142,
      "step": 1964
    },
    {
      "epoch": 0.0901748428250195,
      "grad_norm": 0.4650253355503082,
      "learning_rate": 9.997405275418762e-06,
      "loss": 0.4014,
      "step": 1965
    },
    {
      "epoch": 0.09022073333027397,
      "grad_norm": 0.4805956482887268,
      "learning_rate": 9.997397371388443e-06,
      "loss": 0.3422,
      "step": 1966
    },
    {
      "epoch": 0.09026662383552843,
      "grad_norm": 0.5101807713508606,
      "learning_rate": 9.997389455340965e-06,
      "loss": 0.436,
      "step": 1967
    },
    {
      "epoch": 0.0903125143407829,
      "grad_norm": 0.5229566097259521,
      "learning_rate": 9.99738152727634e-06,
      "loss": 0.4953,
      "step": 1968
    },
    {
      "epoch": 0.09035840484603735,
      "grad_norm": 0.4792730212211609,
      "learning_rate": 9.997373587194587e-06,
      "loss": 0.4073,
      "step": 1969
    },
    {
      "epoch": 0.09040429535129182,
      "grad_norm": 0.519981324672699,
      "learning_rate": 9.997365635095732e-06,
      "loss": 0.3603,
      "step": 1970
    },
    {
      "epoch": 0.09045018585654628,
      "grad_norm": 0.5242164731025696,
      "learning_rate": 9.997357670979788e-06,
      "loss": 0.5586,
      "step": 1971
    },
    {
      "epoch": 0.09049607636180074,
      "grad_norm": 0.48712316155433655,
      "learning_rate": 9.997349694846775e-06,
      "loss": 0.4531,
      "step": 1972
    },
    {
      "epoch": 0.0905419668670552,
      "grad_norm": 0.49038171768188477,
      "learning_rate": 9.997341706696715e-06,
      "loss": 0.5264,
      "step": 1973
    },
    {
      "epoch": 0.09058785737230966,
      "grad_norm": 0.45785385370254517,
      "learning_rate": 9.997333706529624e-06,
      "loss": 0.3475,
      "step": 1974
    },
    {
      "epoch": 0.09063374787756413,
      "grad_norm": 0.46331408619880676,
      "learning_rate": 9.997325694345523e-06,
      "loss": 0.4144,
      "step": 1975
    },
    {
      "epoch": 0.09067963838281859,
      "grad_norm": 0.4868534207344055,
      "learning_rate": 9.997317670144431e-06,
      "loss": 0.4183,
      "step": 1976
    },
    {
      "epoch": 0.09072552888807306,
      "grad_norm": 0.4846562147140503,
      "learning_rate": 9.997309633926367e-06,
      "loss": 0.4039,
      "step": 1977
    },
    {
      "epoch": 0.09077141939332752,
      "grad_norm": 0.5201089978218079,
      "learning_rate": 9.997301585691351e-06,
      "loss": 0.5501,
      "step": 1978
    },
    {
      "epoch": 0.09081730989858199,
      "grad_norm": 0.5357211232185364,
      "learning_rate": 9.997293525439402e-06,
      "loss": 0.4765,
      "step": 1979
    },
    {
      "epoch": 0.09086320040383644,
      "grad_norm": 0.4953075051307678,
      "learning_rate": 9.99728545317054e-06,
      "loss": 0.411,
      "step": 1980
    },
    {
      "epoch": 0.09090909090909091,
      "grad_norm": 0.4356902539730072,
      "learning_rate": 9.997277368884782e-06,
      "loss": 0.3312,
      "step": 1981
    },
    {
      "epoch": 0.09095498141434537,
      "grad_norm": 0.49398893117904663,
      "learning_rate": 9.99726927258215e-06,
      "loss": 0.4406,
      "step": 1982
    },
    {
      "epoch": 0.09100087191959984,
      "grad_norm": 0.5011243224143982,
      "learning_rate": 9.997261164262662e-06,
      "loss": 0.4356,
      "step": 1983
    },
    {
      "epoch": 0.0910467624248543,
      "grad_norm": 0.4967057704925537,
      "learning_rate": 9.997253043926339e-06,
      "loss": 0.4436,
      "step": 1984
    },
    {
      "epoch": 0.09109265293010876,
      "grad_norm": 0.44922012090682983,
      "learning_rate": 9.997244911573199e-06,
      "loss": 0.3491,
      "step": 1985
    },
    {
      "epoch": 0.09113854343536322,
      "grad_norm": 0.5510815978050232,
      "learning_rate": 9.997236767203261e-06,
      "loss": 0.5051,
      "step": 1986
    },
    {
      "epoch": 0.09118443394061769,
      "grad_norm": 0.4843643009662628,
      "learning_rate": 9.997228610816547e-06,
      "loss": 0.3963,
      "step": 1987
    },
    {
      "epoch": 0.09123032444587215,
      "grad_norm": 0.5763829946517944,
      "learning_rate": 9.997220442413075e-06,
      "loss": 0.4622,
      "step": 1988
    },
    {
      "epoch": 0.09127621495112662,
      "grad_norm": 0.5644519329071045,
      "learning_rate": 9.997212261992865e-06,
      "loss": 0.562,
      "step": 1989
    },
    {
      "epoch": 0.09132210545638107,
      "grad_norm": 0.4677918255329132,
      "learning_rate": 9.997204069555939e-06,
      "loss": 0.3484,
      "step": 1990
    },
    {
      "epoch": 0.09136799596163554,
      "grad_norm": 0.5098165273666382,
      "learning_rate": 9.997195865102311e-06,
      "loss": 0.361,
      "step": 1991
    },
    {
      "epoch": 0.09141388646689,
      "grad_norm": 0.6736876368522644,
      "learning_rate": 9.997187648632006e-06,
      "loss": 0.4992,
      "step": 1992
    },
    {
      "epoch": 0.09145977697214447,
      "grad_norm": 0.440369576215744,
      "learning_rate": 9.997179420145043e-06,
      "loss": 0.3183,
      "step": 1993
    },
    {
      "epoch": 0.09150566747739892,
      "grad_norm": 0.4916366934776306,
      "learning_rate": 9.997171179641438e-06,
      "loss": 0.3439,
      "step": 1994
    },
    {
      "epoch": 0.0915515579826534,
      "grad_norm": 0.5069000124931335,
      "learning_rate": 9.997162927121213e-06,
      "loss": 0.4629,
      "step": 1995
    },
    {
      "epoch": 0.09159744848790785,
      "grad_norm": 0.4618993401527405,
      "learning_rate": 9.99715466258439e-06,
      "loss": 0.3542,
      "step": 1996
    },
    {
      "epoch": 0.09164333899316232,
      "grad_norm": 0.5310156345367432,
      "learning_rate": 9.997146386030989e-06,
      "loss": 0.5221,
      "step": 1997
    },
    {
      "epoch": 0.09168922949841678,
      "grad_norm": 0.5467902421951294,
      "learning_rate": 9.997138097461026e-06,
      "loss": 0.4811,
      "step": 1998
    },
    {
      "epoch": 0.09173512000367125,
      "grad_norm": 0.5695247650146484,
      "learning_rate": 9.997129796874521e-06,
      "loss": 0.4704,
      "step": 1999
    },
    {
      "epoch": 0.0917810105089257,
      "grad_norm": 0.45439305901527405,
      "learning_rate": 9.9971214842715e-06,
      "loss": 0.372,
      "step": 2000
    },
    {
      "epoch": 0.09182690101418016,
      "grad_norm": 0.5147867798805237,
      "learning_rate": 9.997113159651977e-06,
      "loss": 0.4309,
      "step": 2001
    },
    {
      "epoch": 0.09187279151943463,
      "grad_norm": 0.452332466840744,
      "learning_rate": 9.997104823015972e-06,
      "loss": 0.3857,
      "step": 2002
    },
    {
      "epoch": 0.09191868202468909,
      "grad_norm": 0.4936886429786682,
      "learning_rate": 9.99709647436351e-06,
      "loss": 0.442,
      "step": 2003
    },
    {
      "epoch": 0.09196457252994356,
      "grad_norm": 0.5347589254379272,
      "learning_rate": 9.997088113694607e-06,
      "loss": 0.4539,
      "step": 2004
    },
    {
      "epoch": 0.09201046303519801,
      "grad_norm": 0.5151339173316956,
      "learning_rate": 9.997079741009285e-06,
      "loss": 0.4736,
      "step": 2005
    },
    {
      "epoch": 0.09205635354045248,
      "grad_norm": 0.5258619785308838,
      "learning_rate": 9.997071356307562e-06,
      "loss": 0.4598,
      "step": 2006
    },
    {
      "epoch": 0.09210224404570694,
      "grad_norm": 0.45975634455680847,
      "learning_rate": 9.99706295958946e-06,
      "loss": 0.3403,
      "step": 2007
    },
    {
      "epoch": 0.09214813455096141,
      "grad_norm": 0.47857576608657837,
      "learning_rate": 9.997054550855e-06,
      "loss": 0.4264,
      "step": 2008
    },
    {
      "epoch": 0.09219402505621586,
      "grad_norm": 0.6320284605026245,
      "learning_rate": 9.9970461301042e-06,
      "loss": 0.5996,
      "step": 2009
    },
    {
      "epoch": 0.09223991556147033,
      "grad_norm": 0.46689409017562866,
      "learning_rate": 9.99703769733708e-06,
      "loss": 0.3845,
      "step": 2010
    },
    {
      "epoch": 0.09228580606672479,
      "grad_norm": 0.4759824573993683,
      "learning_rate": 9.997029252553663e-06,
      "loss": 0.417,
      "step": 2011
    },
    {
      "epoch": 0.09233169657197926,
      "grad_norm": 0.4591982066631317,
      "learning_rate": 9.997020795753966e-06,
      "loss": 0.3989,
      "step": 2012
    },
    {
      "epoch": 0.09237758707723372,
      "grad_norm": 0.4557945132255554,
      "learning_rate": 9.997012326938013e-06,
      "loss": 0.389,
      "step": 2013
    },
    {
      "epoch": 0.09242347758248819,
      "grad_norm": 0.49194812774658203,
      "learning_rate": 9.997003846105821e-06,
      "loss": 0.4788,
      "step": 2014
    },
    {
      "epoch": 0.09246936808774264,
      "grad_norm": 0.4524982273578644,
      "learning_rate": 9.996995353257414e-06,
      "loss": 0.3415,
      "step": 2015
    },
    {
      "epoch": 0.09251525859299711,
      "grad_norm": 0.48286381363868713,
      "learning_rate": 9.996986848392809e-06,
      "loss": 0.4012,
      "step": 2016
    },
    {
      "epoch": 0.09256114909825157,
      "grad_norm": 0.47071242332458496,
      "learning_rate": 9.996978331512027e-06,
      "loss": 0.3686,
      "step": 2017
    },
    {
      "epoch": 0.09260703960350604,
      "grad_norm": 0.49137750267982483,
      "learning_rate": 9.99696980261509e-06,
      "loss": 0.4518,
      "step": 2018
    },
    {
      "epoch": 0.0926529301087605,
      "grad_norm": 0.45813873410224915,
      "learning_rate": 9.996961261702018e-06,
      "loss": 0.3902,
      "step": 2019
    },
    {
      "epoch": 0.09269882061401497,
      "grad_norm": 0.5015468597412109,
      "learning_rate": 9.99695270877283e-06,
      "loss": 0.4394,
      "step": 2020
    },
    {
      "epoch": 0.09274471111926942,
      "grad_norm": 0.5339084267616272,
      "learning_rate": 9.99694414382755e-06,
      "loss": 0.5767,
      "step": 2021
    },
    {
      "epoch": 0.09279060162452389,
      "grad_norm": 0.5212705135345459,
      "learning_rate": 9.996935566866193e-06,
      "loss": 0.4934,
      "step": 2022
    },
    {
      "epoch": 0.09283649212977835,
      "grad_norm": 0.47062548995018005,
      "learning_rate": 9.996926977888786e-06,
      "loss": 0.4076,
      "step": 2023
    },
    {
      "epoch": 0.09288238263503282,
      "grad_norm": 0.4327714443206787,
      "learning_rate": 9.996918376895347e-06,
      "loss": 0.3372,
      "step": 2024
    },
    {
      "epoch": 0.09292827314028727,
      "grad_norm": 0.4736381471157074,
      "learning_rate": 9.996909763885893e-06,
      "loss": 0.4838,
      "step": 2025
    },
    {
      "epoch": 0.09297416364554174,
      "grad_norm": 0.46666479110717773,
      "learning_rate": 9.99690113886045e-06,
      "loss": 0.3819,
      "step": 2026
    },
    {
      "epoch": 0.0930200541507962,
      "grad_norm": 0.4831838309764862,
      "learning_rate": 9.996892501819037e-06,
      "loss": 0.4729,
      "step": 2027
    },
    {
      "epoch": 0.09306594465605067,
      "grad_norm": 0.45430436730384827,
      "learning_rate": 9.996883852761676e-06,
      "loss": 0.3528,
      "step": 2028
    },
    {
      "epoch": 0.09311183516130513,
      "grad_norm": 0.5162645578384399,
      "learning_rate": 9.996875191688385e-06,
      "loss": 0.4576,
      "step": 2029
    },
    {
      "epoch": 0.09315772566655958,
      "grad_norm": 0.48196762800216675,
      "learning_rate": 9.996866518599186e-06,
      "loss": 0.4552,
      "step": 2030
    },
    {
      "epoch": 0.09320361617181405,
      "grad_norm": 0.4665124714374542,
      "learning_rate": 9.9968578334941e-06,
      "loss": 0.4446,
      "step": 2031
    },
    {
      "epoch": 0.09324950667706851,
      "grad_norm": 0.45105865597724915,
      "learning_rate": 9.996849136373148e-06,
      "loss": 0.3795,
      "step": 2032
    },
    {
      "epoch": 0.09329539718232298,
      "grad_norm": 0.4254264533519745,
      "learning_rate": 9.996840427236351e-06,
      "loss": 0.3304,
      "step": 2033
    },
    {
      "epoch": 0.09334128768757743,
      "grad_norm": 0.46485844254493713,
      "learning_rate": 9.996831706083731e-06,
      "loss": 0.4124,
      "step": 2034
    },
    {
      "epoch": 0.0933871781928319,
      "grad_norm": 0.4854772388935089,
      "learning_rate": 9.996822972915305e-06,
      "loss": 0.4252,
      "step": 2035
    },
    {
      "epoch": 0.09343306869808636,
      "grad_norm": 0.47920531034469604,
      "learning_rate": 9.9968142277311e-06,
      "loss": 0.486,
      "step": 2036
    },
    {
      "epoch": 0.09347895920334083,
      "grad_norm": 0.47636234760284424,
      "learning_rate": 9.996805470531133e-06,
      "loss": 0.4332,
      "step": 2037
    },
    {
      "epoch": 0.09352484970859529,
      "grad_norm": 0.46059149503707886,
      "learning_rate": 9.996796701315424e-06,
      "loss": 0.4391,
      "step": 2038
    },
    {
      "epoch": 0.09357074021384976,
      "grad_norm": 0.4492204487323761,
      "learning_rate": 9.996787920083998e-06,
      "loss": 0.3421,
      "step": 2039
    },
    {
      "epoch": 0.09361663071910421,
      "grad_norm": 0.5381326675415039,
      "learning_rate": 9.996779126836873e-06,
      "loss": 0.4947,
      "step": 2040
    },
    {
      "epoch": 0.09366252122435868,
      "grad_norm": 0.6390119194984436,
      "learning_rate": 9.996770321574071e-06,
      "loss": 0.358,
      "step": 2041
    },
    {
      "epoch": 0.09370841172961314,
      "grad_norm": 0.4528900682926178,
      "learning_rate": 9.996761504295614e-06,
      "loss": 0.4001,
      "step": 2042
    },
    {
      "epoch": 0.09375430223486761,
      "grad_norm": 0.49994057416915894,
      "learning_rate": 9.996752675001524e-06,
      "loss": 0.3863,
      "step": 2043
    },
    {
      "epoch": 0.09380019274012207,
      "grad_norm": 0.46004506945610046,
      "learning_rate": 9.996743833691817e-06,
      "loss": 0.3099,
      "step": 2044
    },
    {
      "epoch": 0.09384608324537654,
      "grad_norm": 0.48346641659736633,
      "learning_rate": 9.996734980366523e-06,
      "loss": 0.3334,
      "step": 2045
    },
    {
      "epoch": 0.09389197375063099,
      "grad_norm": 0.48080554604530334,
      "learning_rate": 9.996726115025654e-06,
      "loss": 0.4218,
      "step": 2046
    },
    {
      "epoch": 0.09393786425588546,
      "grad_norm": 0.5070620775222778,
      "learning_rate": 9.996717237669237e-06,
      "loss": 0.4295,
      "step": 2047
    },
    {
      "epoch": 0.09398375476113992,
      "grad_norm": 0.4773145318031311,
      "learning_rate": 9.996708348297296e-06,
      "loss": 0.3935,
      "step": 2048
    },
    {
      "epoch": 0.09402964526639439,
      "grad_norm": 0.5115299820899963,
      "learning_rate": 9.996699446909844e-06,
      "loss": 0.4566,
      "step": 2049
    },
    {
      "epoch": 0.09407553577164884,
      "grad_norm": 0.4422728419303894,
      "learning_rate": 9.996690533506908e-06,
      "loss": 0.3483,
      "step": 2050
    },
    {
      "epoch": 0.09412142627690331,
      "grad_norm": 0.47135084867477417,
      "learning_rate": 9.996681608088507e-06,
      "loss": 0.3561,
      "step": 2051
    },
    {
      "epoch": 0.09416731678215777,
      "grad_norm": 0.4875592887401581,
      "learning_rate": 9.996672670654666e-06,
      "loss": 0.4294,
      "step": 2052
    },
    {
      "epoch": 0.09421320728741224,
      "grad_norm": 0.5143119692802429,
      "learning_rate": 9.996663721205402e-06,
      "loss": 0.5118,
      "step": 2053
    },
    {
      "epoch": 0.0942590977926667,
      "grad_norm": 0.4918496608734131,
      "learning_rate": 9.99665475974074e-06,
      "loss": 0.5125,
      "step": 2054
    },
    {
      "epoch": 0.09430498829792117,
      "grad_norm": 0.47512781620025635,
      "learning_rate": 9.9966457862607e-06,
      "loss": 0.3891,
      "step": 2055
    },
    {
      "epoch": 0.09435087880317562,
      "grad_norm": 0.4893255829811096,
      "learning_rate": 9.996636800765304e-06,
      "loss": 0.4618,
      "step": 2056
    },
    {
      "epoch": 0.09439676930843009,
      "grad_norm": 0.4645330011844635,
      "learning_rate": 9.996627803254574e-06,
      "loss": 0.3717,
      "step": 2057
    },
    {
      "epoch": 0.09444265981368455,
      "grad_norm": 0.480774462223053,
      "learning_rate": 9.99661879372853e-06,
      "loss": 0.4081,
      "step": 2058
    },
    {
      "epoch": 0.094488550318939,
      "grad_norm": 0.6540485620498657,
      "learning_rate": 9.996609772187196e-06,
      "loss": 0.5681,
      "step": 2059
    },
    {
      "epoch": 0.09453444082419347,
      "grad_norm": 0.499470591545105,
      "learning_rate": 9.996600738630592e-06,
      "loss": 0.4294,
      "step": 2060
    },
    {
      "epoch": 0.09458033132944793,
      "grad_norm": 0.50464928150177,
      "learning_rate": 9.99659169305874e-06,
      "loss": 0.4999,
      "step": 2061
    },
    {
      "epoch": 0.0946262218347024,
      "grad_norm": 0.5615844130516052,
      "learning_rate": 9.996582635471662e-06,
      "loss": 0.3745,
      "step": 2062
    },
    {
      "epoch": 0.09467211233995686,
      "grad_norm": 0.5331128239631653,
      "learning_rate": 9.99657356586938e-06,
      "loss": 0.5311,
      "step": 2063
    },
    {
      "epoch": 0.09471800284521133,
      "grad_norm": 0.5935415625572205,
      "learning_rate": 9.996564484251916e-06,
      "loss": 0.4087,
      "step": 2064
    },
    {
      "epoch": 0.09476389335046578,
      "grad_norm": 0.49952229857444763,
      "learning_rate": 9.99655539061929e-06,
      "loss": 0.4899,
      "step": 2065
    },
    {
      "epoch": 0.09480978385572025,
      "grad_norm": 0.4531458616256714,
      "learning_rate": 9.996546284971525e-06,
      "loss": 0.3466,
      "step": 2066
    },
    {
      "epoch": 0.09485567436097471,
      "grad_norm": 0.4564852714538574,
      "learning_rate": 9.996537167308644e-06,
      "loss": 0.3841,
      "step": 2067
    },
    {
      "epoch": 0.09490156486622918,
      "grad_norm": 0.529405415058136,
      "learning_rate": 9.996528037630668e-06,
      "loss": 0.4519,
      "step": 2068
    },
    {
      "epoch": 0.09494745537148364,
      "grad_norm": 0.4793223440647125,
      "learning_rate": 9.996518895937618e-06,
      "loss": 0.3978,
      "step": 2069
    },
    {
      "epoch": 0.0949933458767381,
      "grad_norm": 0.512132465839386,
      "learning_rate": 9.99650974222952e-06,
      "loss": 0.4337,
      "step": 2070
    },
    {
      "epoch": 0.09503923638199256,
      "grad_norm": 0.4214874804019928,
      "learning_rate": 9.996500576506392e-06,
      "loss": 0.3104,
      "step": 2071
    },
    {
      "epoch": 0.09508512688724703,
      "grad_norm": 0.49968063831329346,
      "learning_rate": 9.996491398768255e-06,
      "loss": 0.4654,
      "step": 2072
    },
    {
      "epoch": 0.09513101739250149,
      "grad_norm": 0.4648069143295288,
      "learning_rate": 9.996482209015135e-06,
      "loss": 0.4839,
      "step": 2073
    },
    {
      "epoch": 0.09517690789775596,
      "grad_norm": 0.513130247592926,
      "learning_rate": 9.996473007247052e-06,
      "loss": 0.4561,
      "step": 2074
    },
    {
      "epoch": 0.09522279840301041,
      "grad_norm": 0.4773569703102112,
      "learning_rate": 9.996463793464026e-06,
      "loss": 0.4036,
      "step": 2075
    },
    {
      "epoch": 0.09526868890826488,
      "grad_norm": 0.4816969335079193,
      "learning_rate": 9.996454567666085e-06,
      "loss": 0.4538,
      "step": 2076
    },
    {
      "epoch": 0.09531457941351934,
      "grad_norm": 0.49606969952583313,
      "learning_rate": 9.996445329853245e-06,
      "loss": 0.4317,
      "step": 2077
    },
    {
      "epoch": 0.09536046991877381,
      "grad_norm": 0.4426336884498596,
      "learning_rate": 9.996436080025533e-06,
      "loss": 0.367,
      "step": 2078
    },
    {
      "epoch": 0.09540636042402827,
      "grad_norm": 0.4468148350715637,
      "learning_rate": 9.996426818182967e-06,
      "loss": 0.3741,
      "step": 2079
    },
    {
      "epoch": 0.09545225092928274,
      "grad_norm": 0.5305752754211426,
      "learning_rate": 9.996417544325574e-06,
      "loss": 0.499,
      "step": 2080
    },
    {
      "epoch": 0.09549814143453719,
      "grad_norm": 0.5080432295799255,
      "learning_rate": 9.996408258453372e-06,
      "loss": 0.3881,
      "step": 2081
    },
    {
      "epoch": 0.09554403193979166,
      "grad_norm": 0.4637645184993744,
      "learning_rate": 9.996398960566385e-06,
      "loss": 0.3281,
      "step": 2082
    },
    {
      "epoch": 0.09558992244504612,
      "grad_norm": 0.4604986310005188,
      "learning_rate": 9.996389650664635e-06,
      "loss": 0.3345,
      "step": 2083
    },
    {
      "epoch": 0.09563581295030059,
      "grad_norm": 0.5082672238349915,
      "learning_rate": 9.996380328748147e-06,
      "loss": 0.4351,
      "step": 2084
    },
    {
      "epoch": 0.09568170345555504,
      "grad_norm": 0.4816417396068573,
      "learning_rate": 9.99637099481694e-06,
      "loss": 0.4196,
      "step": 2085
    },
    {
      "epoch": 0.09572759396080951,
      "grad_norm": 0.4824964702129364,
      "learning_rate": 9.996361648871036e-06,
      "loss": 0.4231,
      "step": 2086
    },
    {
      "epoch": 0.09577348446606397,
      "grad_norm": 0.4349449574947357,
      "learning_rate": 9.99635229091046e-06,
      "loss": 0.2979,
      "step": 2087
    },
    {
      "epoch": 0.09581937497131843,
      "grad_norm": 0.5159413814544678,
      "learning_rate": 9.996342920935236e-06,
      "loss": 0.5134,
      "step": 2088
    },
    {
      "epoch": 0.0958652654765729,
      "grad_norm": 0.5171272158622742,
      "learning_rate": 9.996333538945382e-06,
      "loss": 0.5119,
      "step": 2089
    },
    {
      "epoch": 0.09591115598182735,
      "grad_norm": 0.4492175579071045,
      "learning_rate": 9.996324144940924e-06,
      "loss": 0.3254,
      "step": 2090
    },
    {
      "epoch": 0.09595704648708182,
      "grad_norm": 0.4594343602657318,
      "learning_rate": 9.996314738921881e-06,
      "loss": 0.3976,
      "step": 2091
    },
    {
      "epoch": 0.09600293699233628,
      "grad_norm": 0.5824548006057739,
      "learning_rate": 9.99630532088828e-06,
      "loss": 0.4315,
      "step": 2092
    },
    {
      "epoch": 0.09604882749759075,
      "grad_norm": 0.4677094519138336,
      "learning_rate": 9.99629589084014e-06,
      "loss": 0.3412,
      "step": 2093
    },
    {
      "epoch": 0.0960947180028452,
      "grad_norm": 0.4852971136569977,
      "learning_rate": 9.996286448777486e-06,
      "loss": 0.3897,
      "step": 2094
    },
    {
      "epoch": 0.09614060850809968,
      "grad_norm": 0.4925045073032379,
      "learning_rate": 9.99627699470034e-06,
      "loss": 0.4314,
      "step": 2095
    },
    {
      "epoch": 0.09618649901335413,
      "grad_norm": 0.4866385757923126,
      "learning_rate": 9.996267528608725e-06,
      "loss": 0.4103,
      "step": 2096
    },
    {
      "epoch": 0.0962323895186086,
      "grad_norm": 0.48117971420288086,
      "learning_rate": 9.996258050502664e-06,
      "loss": 0.3878,
      "step": 2097
    },
    {
      "epoch": 0.09627828002386306,
      "grad_norm": 0.5057356357574463,
      "learning_rate": 9.996248560382177e-06,
      "loss": 0.4645,
      "step": 2098
    },
    {
      "epoch": 0.09632417052911753,
      "grad_norm": 0.48927298188209534,
      "learning_rate": 9.996239058247292e-06,
      "loss": 0.4218,
      "step": 2099
    },
    {
      "epoch": 0.09637006103437198,
      "grad_norm": 0.4779609441757202,
      "learning_rate": 9.996229544098028e-06,
      "loss": 0.3482,
      "step": 2100
    },
    {
      "epoch": 0.09641595153962645,
      "grad_norm": 0.5371790528297424,
      "learning_rate": 9.996220017934409e-06,
      "loss": 0.5009,
      "step": 2101
    },
    {
      "epoch": 0.09646184204488091,
      "grad_norm": 0.4955594539642334,
      "learning_rate": 9.996210479756456e-06,
      "loss": 0.3913,
      "step": 2102
    },
    {
      "epoch": 0.09650773255013538,
      "grad_norm": 0.4752846360206604,
      "learning_rate": 9.996200929564195e-06,
      "loss": 0.392,
      "step": 2103
    },
    {
      "epoch": 0.09655362305538984,
      "grad_norm": 0.4950645864009857,
      "learning_rate": 9.996191367357647e-06,
      "loss": 0.4475,
      "step": 2104
    },
    {
      "epoch": 0.0965995135606443,
      "grad_norm": 0.5252650380134583,
      "learning_rate": 9.996181793136836e-06,
      "loss": 0.4614,
      "step": 2105
    },
    {
      "epoch": 0.09664540406589876,
      "grad_norm": 0.47076714038848877,
      "learning_rate": 9.996172206901785e-06,
      "loss": 0.3954,
      "step": 2106
    },
    {
      "epoch": 0.09669129457115323,
      "grad_norm": 0.5345096588134766,
      "learning_rate": 9.996162608652516e-06,
      "loss": 0.469,
      "step": 2107
    },
    {
      "epoch": 0.09673718507640769,
      "grad_norm": 0.44931167364120483,
      "learning_rate": 9.996152998389052e-06,
      "loss": 0.3419,
      "step": 2108
    },
    {
      "epoch": 0.09678307558166216,
      "grad_norm": 0.5248048901557922,
      "learning_rate": 9.996143376111418e-06,
      "loss": 0.4938,
      "step": 2109
    },
    {
      "epoch": 0.09682896608691662,
      "grad_norm": 0.5095524787902832,
      "learning_rate": 9.996133741819635e-06,
      "loss": 0.4481,
      "step": 2110
    },
    {
      "epoch": 0.09687485659217109,
      "grad_norm": 0.46126413345336914,
      "learning_rate": 9.996124095513729e-06,
      "loss": 0.3248,
      "step": 2111
    },
    {
      "epoch": 0.09692074709742554,
      "grad_norm": 0.4844275712966919,
      "learning_rate": 9.99611443719372e-06,
      "loss": 0.4377,
      "step": 2112
    },
    {
      "epoch": 0.09696663760268001,
      "grad_norm": 0.4996603727340698,
      "learning_rate": 9.996104766859634e-06,
      "loss": 0.4815,
      "step": 2113
    },
    {
      "epoch": 0.09701252810793447,
      "grad_norm": 0.4958064556121826,
      "learning_rate": 9.996095084511491e-06,
      "loss": 0.4438,
      "step": 2114
    },
    {
      "epoch": 0.09705841861318894,
      "grad_norm": 0.4917685091495514,
      "learning_rate": 9.996085390149316e-06,
      "loss": 0.3995,
      "step": 2115
    },
    {
      "epoch": 0.0971043091184434,
      "grad_norm": 0.482188880443573,
      "learning_rate": 9.996075683773134e-06,
      "loss": 0.4524,
      "step": 2116
    },
    {
      "epoch": 0.09715019962369786,
      "grad_norm": 0.47922563552856445,
      "learning_rate": 9.996065965382967e-06,
      "loss": 0.3988,
      "step": 2117
    },
    {
      "epoch": 0.09719609012895232,
      "grad_norm": 0.5032529234886169,
      "learning_rate": 9.996056234978837e-06,
      "loss": 0.4015,
      "step": 2118
    },
    {
      "epoch": 0.09724198063420678,
      "grad_norm": 0.4822005033493042,
      "learning_rate": 9.99604649256077e-06,
      "loss": 0.3739,
      "step": 2119
    },
    {
      "epoch": 0.09728787113946125,
      "grad_norm": 0.5190449357032776,
      "learning_rate": 9.996036738128787e-06,
      "loss": 0.4614,
      "step": 2120
    },
    {
      "epoch": 0.0973337616447157,
      "grad_norm": 0.5280289053916931,
      "learning_rate": 9.996026971682912e-06,
      "loss": 0.5566,
      "step": 2121
    },
    {
      "epoch": 0.09737965214997017,
      "grad_norm": 0.4400560259819031,
      "learning_rate": 9.996017193223169e-06,
      "loss": 0.3012,
      "step": 2122
    },
    {
      "epoch": 0.09742554265522463,
      "grad_norm": 0.46748727560043335,
      "learning_rate": 9.996007402749582e-06,
      "loss": 0.3512,
      "step": 2123
    },
    {
      "epoch": 0.0974714331604791,
      "grad_norm": 0.5129647850990295,
      "learning_rate": 9.995997600262172e-06,
      "loss": 0.4995,
      "step": 2124
    },
    {
      "epoch": 0.09751732366573355,
      "grad_norm": 0.5150854587554932,
      "learning_rate": 9.995987785760967e-06,
      "loss": 0.5015,
      "step": 2125
    },
    {
      "epoch": 0.09756321417098802,
      "grad_norm": 0.46464788913726807,
      "learning_rate": 9.995977959245988e-06,
      "loss": 0.409,
      "step": 2126
    },
    {
      "epoch": 0.09760910467624248,
      "grad_norm": 0.46803930401802063,
      "learning_rate": 9.995968120717258e-06,
      "loss": 0.3685,
      "step": 2127
    },
    {
      "epoch": 0.09765499518149695,
      "grad_norm": 0.5460236668586731,
      "learning_rate": 9.995958270174801e-06,
      "loss": 0.462,
      "step": 2128
    },
    {
      "epoch": 0.09770088568675141,
      "grad_norm": 0.4722246825695038,
      "learning_rate": 9.995948407618643e-06,
      "loss": 0.3668,
      "step": 2129
    },
    {
      "epoch": 0.09774677619200588,
      "grad_norm": 0.497059166431427,
      "learning_rate": 9.995938533048805e-06,
      "loss": 0.4719,
      "step": 2130
    },
    {
      "epoch": 0.09779266669726033,
      "grad_norm": 0.5201273560523987,
      "learning_rate": 9.99592864646531e-06,
      "loss": 0.4974,
      "step": 2131
    },
    {
      "epoch": 0.0978385572025148,
      "grad_norm": 0.44994765520095825,
      "learning_rate": 9.995918747868186e-06,
      "loss": 0.3272,
      "step": 2132
    },
    {
      "epoch": 0.09788444770776926,
      "grad_norm": 0.5128187537193298,
      "learning_rate": 9.995908837257452e-06,
      "loss": 0.4933,
      "step": 2133
    },
    {
      "epoch": 0.09793033821302373,
      "grad_norm": 0.5345513820648193,
      "learning_rate": 9.995898914633135e-06,
      "loss": 0.4939,
      "step": 2134
    },
    {
      "epoch": 0.09797622871827819,
      "grad_norm": 0.4516834318637848,
      "learning_rate": 9.995888979995257e-06,
      "loss": 0.3654,
      "step": 2135
    },
    {
      "epoch": 0.09802211922353266,
      "grad_norm": 0.4768800139427185,
      "learning_rate": 9.995879033343844e-06,
      "loss": 0.4255,
      "step": 2136
    },
    {
      "epoch": 0.09806800972878711,
      "grad_norm": 0.5497641563415527,
      "learning_rate": 9.995869074678918e-06,
      "loss": 0.4979,
      "step": 2137
    },
    {
      "epoch": 0.09811390023404158,
      "grad_norm": 0.5317090153694153,
      "learning_rate": 9.995859104000502e-06,
      "loss": 0.5097,
      "step": 2138
    },
    {
      "epoch": 0.09815979073929604,
      "grad_norm": 0.5085233449935913,
      "learning_rate": 9.995849121308626e-06,
      "loss": 0.4393,
      "step": 2139
    },
    {
      "epoch": 0.09820568124455051,
      "grad_norm": 0.4718397557735443,
      "learning_rate": 9.995839126603306e-06,
      "loss": 0.4096,
      "step": 2140
    },
    {
      "epoch": 0.09825157174980496,
      "grad_norm": 0.47357964515686035,
      "learning_rate": 9.99582911988457e-06,
      "loss": 0.3901,
      "step": 2141
    },
    {
      "epoch": 0.09829746225505943,
      "grad_norm": 0.49856793880462646,
      "learning_rate": 9.995819101152441e-06,
      "loss": 0.4724,
      "step": 2142
    },
    {
      "epoch": 0.09834335276031389,
      "grad_norm": 0.5811507701873779,
      "learning_rate": 9.995809070406945e-06,
      "loss": 0.5367,
      "step": 2143
    },
    {
      "epoch": 0.09838924326556836,
      "grad_norm": 0.5023209452629089,
      "learning_rate": 9.995799027648105e-06,
      "loss": 0.4118,
      "step": 2144
    },
    {
      "epoch": 0.09843513377082282,
      "grad_norm": 0.5006213784217834,
      "learning_rate": 9.995788972875947e-06,
      "loss": 0.4306,
      "step": 2145
    },
    {
      "epoch": 0.09848102427607729,
      "grad_norm": 0.4715560972690582,
      "learning_rate": 9.99577890609049e-06,
      "loss": 0.3177,
      "step": 2146
    },
    {
      "epoch": 0.09852691478133174,
      "grad_norm": 0.47375479340553284,
      "learning_rate": 9.995768827291763e-06,
      "loss": 0.3851,
      "step": 2147
    },
    {
      "epoch": 0.0985728052865862,
      "grad_norm": 0.5320106148719788,
      "learning_rate": 9.995758736479788e-06,
      "loss": 0.5271,
      "step": 2148
    },
    {
      "epoch": 0.09861869579184067,
      "grad_norm": 0.4591202437877655,
      "learning_rate": 9.99574863365459e-06,
      "loss": 0.3905,
      "step": 2149
    },
    {
      "epoch": 0.09866458629709512,
      "grad_norm": 0.4693296253681183,
      "learning_rate": 9.995738518816193e-06,
      "loss": 0.4113,
      "step": 2150
    },
    {
      "epoch": 0.0987104768023496,
      "grad_norm": 0.4697159230709076,
      "learning_rate": 9.995728391964623e-06,
      "loss": 0.4133,
      "step": 2151
    },
    {
      "epoch": 0.09875636730760405,
      "grad_norm": 0.6639912128448486,
      "learning_rate": 9.995718253099902e-06,
      "loss": 0.4733,
      "step": 2152
    },
    {
      "epoch": 0.09880225781285852,
      "grad_norm": 0.9481196403503418,
      "learning_rate": 9.995708102222055e-06,
      "loss": 0.4575,
      "step": 2153
    },
    {
      "epoch": 0.09884814831811298,
      "grad_norm": 0.4139403700828552,
      "learning_rate": 9.995697939331105e-06,
      "loss": 0.2906,
      "step": 2154
    },
    {
      "epoch": 0.09889403882336745,
      "grad_norm": 0.4694135785102844,
      "learning_rate": 9.995687764427082e-06,
      "loss": 0.4363,
      "step": 2155
    },
    {
      "epoch": 0.0989399293286219,
      "grad_norm": 0.5318163633346558,
      "learning_rate": 9.995677577510005e-06,
      "loss": 0.5405,
      "step": 2156
    },
    {
      "epoch": 0.09898581983387637,
      "grad_norm": 0.4968397915363312,
      "learning_rate": 9.995667378579898e-06,
      "loss": 0.5101,
      "step": 2157
    },
    {
      "epoch": 0.09903171033913083,
      "grad_norm": 0.5037608742713928,
      "learning_rate": 9.995657167636791e-06,
      "loss": 0.4637,
      "step": 2158
    },
    {
      "epoch": 0.0990776008443853,
      "grad_norm": 0.462295800447464,
      "learning_rate": 9.995646944680703e-06,
      "loss": 0.3995,
      "step": 2159
    },
    {
      "epoch": 0.09912349134963976,
      "grad_norm": 0.5383294224739075,
      "learning_rate": 9.995636709711662e-06,
      "loss": 0.4927,
      "step": 2160
    },
    {
      "epoch": 0.09916938185489423,
      "grad_norm": 0.4859579801559448,
      "learning_rate": 9.99562646272969e-06,
      "loss": 0.4202,
      "step": 2161
    },
    {
      "epoch": 0.09921527236014868,
      "grad_norm": 0.5066967010498047,
      "learning_rate": 9.995616203734816e-06,
      "loss": 0.451,
      "step": 2162
    },
    {
      "epoch": 0.09926116286540315,
      "grad_norm": 0.5146127939224243,
      "learning_rate": 9.995605932727057e-06,
      "loss": 0.4443,
      "step": 2163
    },
    {
      "epoch": 0.09930705337065761,
      "grad_norm": 0.48541831970214844,
      "learning_rate": 9.995595649706445e-06,
      "loss": 0.4243,
      "step": 2164
    },
    {
      "epoch": 0.09935294387591208,
      "grad_norm": 0.4704222083091736,
      "learning_rate": 9.995585354673004e-06,
      "loss": 0.4556,
      "step": 2165
    },
    {
      "epoch": 0.09939883438116653,
      "grad_norm": 0.47955968976020813,
      "learning_rate": 9.995575047626753e-06,
      "loss": 0.3764,
      "step": 2166
    },
    {
      "epoch": 0.099444724886421,
      "grad_norm": 0.5205246210098267,
      "learning_rate": 9.995564728567723e-06,
      "loss": 0.5056,
      "step": 2167
    },
    {
      "epoch": 0.09949061539167546,
      "grad_norm": 0.4942520260810852,
      "learning_rate": 9.995554397495937e-06,
      "loss": 0.517,
      "step": 2168
    },
    {
      "epoch": 0.09953650589692993,
      "grad_norm": 0.4720578193664551,
      "learning_rate": 9.995544054411418e-06,
      "loss": 0.4101,
      "step": 2169
    },
    {
      "epoch": 0.09958239640218439,
      "grad_norm": 0.4484683573246002,
      "learning_rate": 9.995533699314192e-06,
      "loss": 0.3906,
      "step": 2170
    },
    {
      "epoch": 0.09962828690743886,
      "grad_norm": 0.4514347016811371,
      "learning_rate": 9.995523332204285e-06,
      "loss": 0.4033,
      "step": 2171
    },
    {
      "epoch": 0.09967417741269331,
      "grad_norm": 0.5216149687767029,
      "learning_rate": 9.99551295308172e-06,
      "loss": 0.536,
      "step": 2172
    },
    {
      "epoch": 0.09972006791794778,
      "grad_norm": 0.4413753151893616,
      "learning_rate": 9.995502561946524e-06,
      "loss": 0.406,
      "step": 2173
    },
    {
      "epoch": 0.09976595842320224,
      "grad_norm": 0.44576913118362427,
      "learning_rate": 9.99549215879872e-06,
      "loss": 0.4069,
      "step": 2174
    },
    {
      "epoch": 0.09981184892845671,
      "grad_norm": 0.4599571228027344,
      "learning_rate": 9.995481743638335e-06,
      "loss": 0.3899,
      "step": 2175
    },
    {
      "epoch": 0.09985773943371117,
      "grad_norm": 0.48537591099739075,
      "learning_rate": 9.995471316465392e-06,
      "loss": 0.4125,
      "step": 2176
    },
    {
      "epoch": 0.09990362993896562,
      "grad_norm": 0.4500022530555725,
      "learning_rate": 9.995460877279918e-06,
      "loss": 0.4163,
      "step": 2177
    },
    {
      "epoch": 0.09994952044422009,
      "grad_norm": 0.4591972529888153,
      "learning_rate": 9.995450426081936e-06,
      "loss": 0.439,
      "step": 2178
    },
    {
      "epoch": 0.09999541094947455,
      "grad_norm": 0.5077624917030334,
      "learning_rate": 9.995439962871473e-06,
      "loss": 0.4161,
      "step": 2179
    },
    {
      "epoch": 0.10004130145472902,
      "grad_norm": 0.45666274428367615,
      "learning_rate": 9.995429487648555e-06,
      "loss": 0.436,
      "step": 2180
    },
    {
      "epoch": 0.10008719195998347,
      "grad_norm": 0.4778783321380615,
      "learning_rate": 9.995419000413204e-06,
      "loss": 0.4074,
      "step": 2181
    },
    {
      "epoch": 0.10013308246523794,
      "grad_norm": 0.47730910778045654,
      "learning_rate": 9.995408501165447e-06,
      "loss": 0.4468,
      "step": 2182
    },
    {
      "epoch": 0.1001789729704924,
      "grad_norm": 0.4342380166053772,
      "learning_rate": 9.99539798990531e-06,
      "loss": 0.3603,
      "step": 2183
    },
    {
      "epoch": 0.10022486347574687,
      "grad_norm": 0.4938841760158539,
      "learning_rate": 9.995387466632816e-06,
      "loss": 0.4539,
      "step": 2184
    },
    {
      "epoch": 0.10027075398100133,
      "grad_norm": 0.5060190558433533,
      "learning_rate": 9.995376931347993e-06,
      "loss": 0.3947,
      "step": 2185
    },
    {
      "epoch": 0.1003166444862558,
      "grad_norm": 0.4662055969238281,
      "learning_rate": 9.995366384050865e-06,
      "loss": 0.4299,
      "step": 2186
    },
    {
      "epoch": 0.10036253499151025,
      "grad_norm": 0.45805132389068604,
      "learning_rate": 9.995355824741457e-06,
      "loss": 0.3403,
      "step": 2187
    },
    {
      "epoch": 0.10040842549676472,
      "grad_norm": 0.506393313407898,
      "learning_rate": 9.995345253419795e-06,
      "loss": 0.335,
      "step": 2188
    },
    {
      "epoch": 0.10045431600201918,
      "grad_norm": 0.4764336347579956,
      "learning_rate": 9.995334670085905e-06,
      "loss": 0.3535,
      "step": 2189
    },
    {
      "epoch": 0.10050020650727365,
      "grad_norm": 0.4943844676017761,
      "learning_rate": 9.99532407473981e-06,
      "loss": 0.4677,
      "step": 2190
    },
    {
      "epoch": 0.1005460970125281,
      "grad_norm": 0.47232797741889954,
      "learning_rate": 9.995313467381538e-06,
      "loss": 0.3643,
      "step": 2191
    },
    {
      "epoch": 0.10059198751778257,
      "grad_norm": 0.49308115243911743,
      "learning_rate": 9.995302848011114e-06,
      "loss": 0.4105,
      "step": 2192
    },
    {
      "epoch": 0.10063787802303703,
      "grad_norm": 0.45486244559288025,
      "learning_rate": 9.995292216628563e-06,
      "loss": 0.4052,
      "step": 2193
    },
    {
      "epoch": 0.1006837685282915,
      "grad_norm": 0.5041335821151733,
      "learning_rate": 9.99528157323391e-06,
      "loss": 0.4164,
      "step": 2194
    },
    {
      "epoch": 0.10072965903354596,
      "grad_norm": 0.47231435775756836,
      "learning_rate": 9.995270917827182e-06,
      "loss": 0.4312,
      "step": 2195
    },
    {
      "epoch": 0.10077554953880043,
      "grad_norm": 0.4928642213344574,
      "learning_rate": 9.995260250408406e-06,
      "loss": 0.4261,
      "step": 2196
    },
    {
      "epoch": 0.10082144004405488,
      "grad_norm": 0.7827200889587402,
      "learning_rate": 9.995249570977603e-06,
      "loss": 0.4144,
      "step": 2197
    },
    {
      "epoch": 0.10086733054930935,
      "grad_norm": 0.5145460963249207,
      "learning_rate": 9.9952388795348e-06,
      "loss": 0.4065,
      "step": 2198
    },
    {
      "epoch": 0.10091322105456381,
      "grad_norm": 0.4500555992126465,
      "learning_rate": 9.995228176080027e-06,
      "loss": 0.3574,
      "step": 2199
    },
    {
      "epoch": 0.10095911155981828,
      "grad_norm": 0.5063720345497131,
      "learning_rate": 9.995217460613307e-06,
      "loss": 0.456,
      "step": 2200
    },
    {
      "epoch": 0.10100500206507274,
      "grad_norm": 0.47892463207244873,
      "learning_rate": 9.995206733134663e-06,
      "loss": 0.4054,
      "step": 2201
    },
    {
      "epoch": 0.1010508925703272,
      "grad_norm": 0.4448998272418976,
      "learning_rate": 9.995195993644125e-06,
      "loss": 0.3883,
      "step": 2202
    },
    {
      "epoch": 0.10109678307558166,
      "grad_norm": 0.4948670566082001,
      "learning_rate": 9.995185242141718e-06,
      "loss": 0.4721,
      "step": 2203
    },
    {
      "epoch": 0.10114267358083613,
      "grad_norm": 0.541731595993042,
      "learning_rate": 9.995174478627465e-06,
      "loss": 0.5679,
      "step": 2204
    },
    {
      "epoch": 0.10118856408609059,
      "grad_norm": 0.39994311332702637,
      "learning_rate": 9.995163703101395e-06,
      "loss": 0.2985,
      "step": 2205
    },
    {
      "epoch": 0.10123445459134504,
      "grad_norm": 0.5325403213500977,
      "learning_rate": 9.995152915563533e-06,
      "loss": 0.5079,
      "step": 2206
    },
    {
      "epoch": 0.10128034509659951,
      "grad_norm": 0.47382909059524536,
      "learning_rate": 9.995142116013903e-06,
      "loss": 0.4287,
      "step": 2207
    },
    {
      "epoch": 0.10132623560185397,
      "grad_norm": 0.46792659163475037,
      "learning_rate": 9.995131304452535e-06,
      "loss": 0.3958,
      "step": 2208
    },
    {
      "epoch": 0.10137212610710844,
      "grad_norm": 0.46098190546035767,
      "learning_rate": 9.995120480879452e-06,
      "loss": 0.3795,
      "step": 2209
    },
    {
      "epoch": 0.1014180166123629,
      "grad_norm": 0.5036364197731018,
      "learning_rate": 9.99510964529468e-06,
      "loss": 0.4578,
      "step": 2210
    },
    {
      "epoch": 0.10146390711761737,
      "grad_norm": 0.470348596572876,
      "learning_rate": 9.995098797698247e-06,
      "loss": 0.3892,
      "step": 2211
    },
    {
      "epoch": 0.10150979762287182,
      "grad_norm": 0.4469543397426605,
      "learning_rate": 9.995087938090178e-06,
      "loss": 0.3101,
      "step": 2212
    },
    {
      "epoch": 0.10155568812812629,
      "grad_norm": 0.4983699321746826,
      "learning_rate": 9.995077066470498e-06,
      "loss": 0.4663,
      "step": 2213
    },
    {
      "epoch": 0.10160157863338075,
      "grad_norm": 0.4524115025997162,
      "learning_rate": 9.995066182839233e-06,
      "loss": 0.3654,
      "step": 2214
    },
    {
      "epoch": 0.10164746913863522,
      "grad_norm": 0.4622112512588501,
      "learning_rate": 9.995055287196413e-06,
      "loss": 0.3746,
      "step": 2215
    },
    {
      "epoch": 0.10169335964388967,
      "grad_norm": 0.44869089126586914,
      "learning_rate": 9.99504437954206e-06,
      "loss": 0.3533,
      "step": 2216
    },
    {
      "epoch": 0.10173925014914414,
      "grad_norm": 0.4604414999485016,
      "learning_rate": 9.995033459876202e-06,
      "loss": 0.4453,
      "step": 2217
    },
    {
      "epoch": 0.1017851406543986,
      "grad_norm": 0.4404522776603699,
      "learning_rate": 9.995022528198865e-06,
      "loss": 0.3684,
      "step": 2218
    },
    {
      "epoch": 0.10183103115965307,
      "grad_norm": 0.4539918303489685,
      "learning_rate": 9.995011584510074e-06,
      "loss": 0.3738,
      "step": 2219
    },
    {
      "epoch": 0.10187692166490753,
      "grad_norm": 0.5033751130104065,
      "learning_rate": 9.995000628809859e-06,
      "loss": 0.4241,
      "step": 2220
    },
    {
      "epoch": 0.101922812170162,
      "grad_norm": 0.5245758891105652,
      "learning_rate": 9.994989661098243e-06,
      "loss": 0.3629,
      "step": 2221
    },
    {
      "epoch": 0.10196870267541645,
      "grad_norm": 0.4730580449104309,
      "learning_rate": 9.994978681375254e-06,
      "loss": 0.3834,
      "step": 2222
    },
    {
      "epoch": 0.10201459318067092,
      "grad_norm": 0.46525928378105164,
      "learning_rate": 9.994967689640915e-06,
      "loss": 0.3916,
      "step": 2223
    },
    {
      "epoch": 0.10206048368592538,
      "grad_norm": 0.46430933475494385,
      "learning_rate": 9.994956685895259e-06,
      "loss": 0.3255,
      "step": 2224
    },
    {
      "epoch": 0.10210637419117985,
      "grad_norm": 0.67746901512146,
      "learning_rate": 9.994945670138305e-06,
      "loss": 0.3333,
      "step": 2225
    },
    {
      "epoch": 0.1021522646964343,
      "grad_norm": 0.5671274662017822,
      "learning_rate": 9.994934642370085e-06,
      "loss": 0.4657,
      "step": 2226
    },
    {
      "epoch": 0.10219815520168878,
      "grad_norm": 0.48215028643608093,
      "learning_rate": 9.994923602590626e-06,
      "loss": 0.4336,
      "step": 2227
    },
    {
      "epoch": 0.10224404570694323,
      "grad_norm": 0.5110505819320679,
      "learning_rate": 9.994912550799948e-06,
      "loss": 0.4637,
      "step": 2228
    },
    {
      "epoch": 0.1022899362121977,
      "grad_norm": 0.5171206593513489,
      "learning_rate": 9.994901486998084e-06,
      "loss": 0.4985,
      "step": 2229
    },
    {
      "epoch": 0.10233582671745216,
      "grad_norm": 0.5020104646682739,
      "learning_rate": 9.994890411185057e-06,
      "loss": 0.4134,
      "step": 2230
    },
    {
      "epoch": 0.10238171722270663,
      "grad_norm": 0.515200138092041,
      "learning_rate": 9.994879323360895e-06,
      "loss": 0.392,
      "step": 2231
    },
    {
      "epoch": 0.10242760772796108,
      "grad_norm": 0.5204353928565979,
      "learning_rate": 9.994868223525627e-06,
      "loss": 0.4528,
      "step": 2232
    },
    {
      "epoch": 0.10247349823321555,
      "grad_norm": 0.439077764749527,
      "learning_rate": 9.994857111679275e-06,
      "loss": 0.3284,
      "step": 2233
    },
    {
      "epoch": 0.10251938873847001,
      "grad_norm": 0.4522359371185303,
      "learning_rate": 9.994845987821869e-06,
      "loss": 0.3765,
      "step": 2234
    },
    {
      "epoch": 0.10256527924372447,
      "grad_norm": 0.5172908306121826,
      "learning_rate": 9.994834851953434e-06,
      "loss": 0.4447,
      "step": 2235
    },
    {
      "epoch": 0.10261116974897894,
      "grad_norm": 0.5548928380012512,
      "learning_rate": 9.994823704073999e-06,
      "loss": 0.5318,
      "step": 2236
    },
    {
      "epoch": 0.10265706025423339,
      "grad_norm": 0.45871788263320923,
      "learning_rate": 9.994812544183587e-06,
      "loss": 0.3571,
      "step": 2237
    },
    {
      "epoch": 0.10270295075948786,
      "grad_norm": 0.47312793135643005,
      "learning_rate": 9.99480137228223e-06,
      "loss": 0.3511,
      "step": 2238
    },
    {
      "epoch": 0.10274884126474232,
      "grad_norm": 0.4848744571208954,
      "learning_rate": 9.994790188369952e-06,
      "loss": 0.4607,
      "step": 2239
    },
    {
      "epoch": 0.10279473176999679,
      "grad_norm": 0.47760719060897827,
      "learning_rate": 9.994778992446779e-06,
      "loss": 0.3909,
      "step": 2240
    },
    {
      "epoch": 0.10284062227525124,
      "grad_norm": 0.47932136058807373,
      "learning_rate": 9.994767784512738e-06,
      "loss": 0.3861,
      "step": 2241
    },
    {
      "epoch": 0.10288651278050572,
      "grad_norm": 0.49739232659339905,
      "learning_rate": 9.994756564567858e-06,
      "loss": 0.394,
      "step": 2242
    },
    {
      "epoch": 0.10293240328576017,
      "grad_norm": 0.47455066442489624,
      "learning_rate": 9.994745332612165e-06,
      "loss": 0.379,
      "step": 2243
    },
    {
      "epoch": 0.10297829379101464,
      "grad_norm": 0.5340999364852905,
      "learning_rate": 9.994734088645685e-06,
      "loss": 0.4073,
      "step": 2244
    },
    {
      "epoch": 0.1030241842962691,
      "grad_norm": 0.5435153841972351,
      "learning_rate": 9.994722832668447e-06,
      "loss": 0.5173,
      "step": 2245
    },
    {
      "epoch": 0.10307007480152357,
      "grad_norm": 0.5073051452636719,
      "learning_rate": 9.994711564680475e-06,
      "loss": 0.465,
      "step": 2246
    },
    {
      "epoch": 0.10311596530677802,
      "grad_norm": 0.505617082118988,
      "learning_rate": 9.9947002846818e-06,
      "loss": 0.4141,
      "step": 2247
    },
    {
      "epoch": 0.1031618558120325,
      "grad_norm": 0.5151702165603638,
      "learning_rate": 9.994688992672446e-06,
      "loss": 0.4138,
      "step": 2248
    },
    {
      "epoch": 0.10320774631728695,
      "grad_norm": 0.4773789942264557,
      "learning_rate": 9.994677688652443e-06,
      "loss": 0.3628,
      "step": 2249
    },
    {
      "epoch": 0.10325363682254142,
      "grad_norm": 0.48059847950935364,
      "learning_rate": 9.994666372621816e-06,
      "loss": 0.4127,
      "step": 2250
    },
    {
      "epoch": 0.10329952732779588,
      "grad_norm": 0.4865242540836334,
      "learning_rate": 9.99465504458059e-06,
      "loss": 0.4151,
      "step": 2251
    },
    {
      "epoch": 0.10334541783305035,
      "grad_norm": 0.5266616344451904,
      "learning_rate": 9.994643704528799e-06,
      "loss": 0.3796,
      "step": 2252
    },
    {
      "epoch": 0.1033913083383048,
      "grad_norm": 0.47762367129325867,
      "learning_rate": 9.994632352466464e-06,
      "loss": 0.3826,
      "step": 2253
    },
    {
      "epoch": 0.10343719884355927,
      "grad_norm": 0.5382362008094788,
      "learning_rate": 9.994620988393616e-06,
      "loss": 0.5113,
      "step": 2254
    },
    {
      "epoch": 0.10348308934881373,
      "grad_norm": 0.4840710461139679,
      "learning_rate": 9.994609612310279e-06,
      "loss": 0.4081,
      "step": 2255
    },
    {
      "epoch": 0.1035289798540682,
      "grad_norm": 0.4215961694717407,
      "learning_rate": 9.994598224216482e-06,
      "loss": 0.3472,
      "step": 2256
    },
    {
      "epoch": 0.10357487035932265,
      "grad_norm": 0.4542910158634186,
      "learning_rate": 9.994586824112255e-06,
      "loss": 0.3611,
      "step": 2257
    },
    {
      "epoch": 0.10362076086457712,
      "grad_norm": 0.4572446644306183,
      "learning_rate": 9.99457541199762e-06,
      "loss": 0.4029,
      "step": 2258
    },
    {
      "epoch": 0.10366665136983158,
      "grad_norm": 0.4768829345703125,
      "learning_rate": 9.99456398787261e-06,
      "loss": 0.4332,
      "step": 2259
    },
    {
      "epoch": 0.10371254187508605,
      "grad_norm": 0.474031001329422,
      "learning_rate": 9.99455255173725e-06,
      "loss": 0.4178,
      "step": 2260
    },
    {
      "epoch": 0.1037584323803405,
      "grad_norm": 0.48735129833221436,
      "learning_rate": 9.994541103591566e-06,
      "loss": 0.4297,
      "step": 2261
    },
    {
      "epoch": 0.10380432288559498,
      "grad_norm": 0.48689502477645874,
      "learning_rate": 9.994529643435588e-06,
      "loss": 0.3712,
      "step": 2262
    },
    {
      "epoch": 0.10385021339084943,
      "grad_norm": 0.44529426097869873,
      "learning_rate": 9.994518171269342e-06,
      "loss": 0.3496,
      "step": 2263
    },
    {
      "epoch": 0.1038961038961039,
      "grad_norm": 0.49490877985954285,
      "learning_rate": 9.994506687092856e-06,
      "loss": 0.4297,
      "step": 2264
    },
    {
      "epoch": 0.10394199440135836,
      "grad_norm": 0.473677396774292,
      "learning_rate": 9.994495190906158e-06,
      "loss": 0.4025,
      "step": 2265
    },
    {
      "epoch": 0.10398788490661282,
      "grad_norm": 0.49198609590530396,
      "learning_rate": 9.994483682709275e-06,
      "loss": 0.466,
      "step": 2266
    },
    {
      "epoch": 0.10403377541186729,
      "grad_norm": 0.6021363735198975,
      "learning_rate": 9.994472162502235e-06,
      "loss": 0.4573,
      "step": 2267
    },
    {
      "epoch": 0.10407966591712174,
      "grad_norm": 0.4464619755744934,
      "learning_rate": 9.994460630285066e-06,
      "loss": 0.327,
      "step": 2268
    },
    {
      "epoch": 0.10412555642237621,
      "grad_norm": 0.5164344310760498,
      "learning_rate": 9.994449086057796e-06,
      "loss": 0.4582,
      "step": 2269
    },
    {
      "epoch": 0.10417144692763067,
      "grad_norm": 0.5113261938095093,
      "learning_rate": 9.994437529820452e-06,
      "loss": 0.4951,
      "step": 2270
    },
    {
      "epoch": 0.10421733743288514,
      "grad_norm": 0.4675207734107971,
      "learning_rate": 9.994425961573063e-06,
      "loss": 0.3984,
      "step": 2271
    },
    {
      "epoch": 0.1042632279381396,
      "grad_norm": 0.483003169298172,
      "learning_rate": 9.994414381315654e-06,
      "loss": 0.457,
      "step": 2272
    },
    {
      "epoch": 0.10430911844339406,
      "grad_norm": 0.5005264282226562,
      "learning_rate": 9.994402789048254e-06,
      "loss": 0.4435,
      "step": 2273
    },
    {
      "epoch": 0.10435500894864852,
      "grad_norm": 0.45094045996665955,
      "learning_rate": 9.994391184770894e-06,
      "loss": 0.3642,
      "step": 2274
    },
    {
      "epoch": 0.10440089945390299,
      "grad_norm": 0.4513227343559265,
      "learning_rate": 9.994379568483599e-06,
      "loss": 0.3488,
      "step": 2275
    },
    {
      "epoch": 0.10444678995915745,
      "grad_norm": 0.45290058851242065,
      "learning_rate": 9.994367940186395e-06,
      "loss": 0.3874,
      "step": 2276
    },
    {
      "epoch": 0.10449268046441192,
      "grad_norm": 0.46660661697387695,
      "learning_rate": 9.994356299879315e-06,
      "loss": 0.4454,
      "step": 2277
    },
    {
      "epoch": 0.10453857096966637,
      "grad_norm": 0.5158165693283081,
      "learning_rate": 9.994344647562384e-06,
      "loss": 0.4185,
      "step": 2278
    },
    {
      "epoch": 0.10458446147492084,
      "grad_norm": 0.45892372727394104,
      "learning_rate": 9.99433298323563e-06,
      "loss": 0.3642,
      "step": 2279
    },
    {
      "epoch": 0.1046303519801753,
      "grad_norm": 0.48131415247917175,
      "learning_rate": 9.99432130689908e-06,
      "loss": 0.4475,
      "step": 2280
    },
    {
      "epoch": 0.10467624248542977,
      "grad_norm": 0.45579615235328674,
      "learning_rate": 9.994309618552767e-06,
      "loss": 0.3903,
      "step": 2281
    },
    {
      "epoch": 0.10472213299068422,
      "grad_norm": 0.4769909679889679,
      "learning_rate": 9.994297918196712e-06,
      "loss": 0.3563,
      "step": 2282
    },
    {
      "epoch": 0.1047680234959387,
      "grad_norm": 0.5031367540359497,
      "learning_rate": 9.99428620583095e-06,
      "loss": 0.4604,
      "step": 2283
    },
    {
      "epoch": 0.10481391400119315,
      "grad_norm": 0.499083012342453,
      "learning_rate": 9.994274481455503e-06,
      "loss": 0.4311,
      "step": 2284
    },
    {
      "epoch": 0.10485980450644762,
      "grad_norm": 0.45621082186698914,
      "learning_rate": 9.994262745070404e-06,
      "loss": 0.3708,
      "step": 2285
    },
    {
      "epoch": 0.10490569501170208,
      "grad_norm": 0.5472161173820496,
      "learning_rate": 9.994250996675678e-06,
      "loss": 0.481,
      "step": 2286
    },
    {
      "epoch": 0.10495158551695655,
      "grad_norm": 0.47954538464546204,
      "learning_rate": 9.994239236271355e-06,
      "loss": 0.4298,
      "step": 2287
    },
    {
      "epoch": 0.104997476022211,
      "grad_norm": 0.48620352149009705,
      "learning_rate": 9.994227463857462e-06,
      "loss": 0.41,
      "step": 2288
    },
    {
      "epoch": 0.10504336652746547,
      "grad_norm": 0.4758157432079315,
      "learning_rate": 9.994215679434029e-06,
      "loss": 0.4177,
      "step": 2289
    },
    {
      "epoch": 0.10508925703271993,
      "grad_norm": 0.46810030937194824,
      "learning_rate": 9.994203883001086e-06,
      "loss": 0.3778,
      "step": 2290
    },
    {
      "epoch": 0.1051351475379744,
      "grad_norm": 0.46195679903030396,
      "learning_rate": 9.994192074558655e-06,
      "loss": 0.3535,
      "step": 2291
    },
    {
      "epoch": 0.10518103804322886,
      "grad_norm": 0.47153595089912415,
      "learning_rate": 9.99418025410677e-06,
      "loss": 0.4338,
      "step": 2292
    },
    {
      "epoch": 0.10522692854848333,
      "grad_norm": 0.4545343816280365,
      "learning_rate": 9.994168421645458e-06,
      "loss": 0.3429,
      "step": 2293
    },
    {
      "epoch": 0.10527281905373778,
      "grad_norm": 0.499397337436676,
      "learning_rate": 9.994156577174747e-06,
      "loss": 0.4304,
      "step": 2294
    },
    {
      "epoch": 0.10531870955899224,
      "grad_norm": 0.5031590461730957,
      "learning_rate": 9.994144720694665e-06,
      "loss": 0.4586,
      "step": 2295
    },
    {
      "epoch": 0.10536460006424671,
      "grad_norm": 0.5150892734527588,
      "learning_rate": 9.994132852205241e-06,
      "loss": 0.5245,
      "step": 2296
    },
    {
      "epoch": 0.10541049056950116,
      "grad_norm": 0.5018406510353088,
      "learning_rate": 9.994120971706505e-06,
      "loss": 0.4291,
      "step": 2297
    },
    {
      "epoch": 0.10545638107475563,
      "grad_norm": 0.48349466919898987,
      "learning_rate": 9.994109079198485e-06,
      "loss": 0.4885,
      "step": 2298
    },
    {
      "epoch": 0.10550227158001009,
      "grad_norm": 0.4850239157676697,
      "learning_rate": 9.994097174681206e-06,
      "loss": 0.4081,
      "step": 2299
    },
    {
      "epoch": 0.10554816208526456,
      "grad_norm": 0.5362891554832458,
      "learning_rate": 9.9940852581547e-06,
      "loss": 0.4581,
      "step": 2300
    },
    {
      "epoch": 0.10559405259051902,
      "grad_norm": 0.46285656094551086,
      "learning_rate": 9.994073329618996e-06,
      "loss": 0.3888,
      "step": 2301
    },
    {
      "epoch": 0.10563994309577349,
      "grad_norm": 0.5105830430984497,
      "learning_rate": 9.994061389074121e-06,
      "loss": 0.5244,
      "step": 2302
    },
    {
      "epoch": 0.10568583360102794,
      "grad_norm": 0.46407097578048706,
      "learning_rate": 9.994049436520105e-06,
      "loss": 0.3871,
      "step": 2303
    },
    {
      "epoch": 0.10573172410628241,
      "grad_norm": 0.492532342672348,
      "learning_rate": 9.994037471956977e-06,
      "loss": 0.4071,
      "step": 2304
    },
    {
      "epoch": 0.10577761461153687,
      "grad_norm": 0.4274005591869354,
      "learning_rate": 9.994025495384764e-06,
      "loss": 0.3901,
      "step": 2305
    },
    {
      "epoch": 0.10582350511679134,
      "grad_norm": 0.4736476242542267,
      "learning_rate": 9.994013506803497e-06,
      "loss": 0.4085,
      "step": 2306
    },
    {
      "epoch": 0.1058693956220458,
      "grad_norm": 0.45817533135414124,
      "learning_rate": 9.994001506213204e-06,
      "loss": 0.3445,
      "step": 2307
    },
    {
      "epoch": 0.10591528612730026,
      "grad_norm": 0.5188115239143372,
      "learning_rate": 9.993989493613913e-06,
      "loss": 0.5088,
      "step": 2308
    },
    {
      "epoch": 0.10596117663255472,
      "grad_norm": 0.46297457814216614,
      "learning_rate": 9.993977469005653e-06,
      "loss": 0.3574,
      "step": 2309
    },
    {
      "epoch": 0.10600706713780919,
      "grad_norm": 0.5024999976158142,
      "learning_rate": 9.993965432388453e-06,
      "loss": 0.4564,
      "step": 2310
    },
    {
      "epoch": 0.10605295764306365,
      "grad_norm": 0.4582570791244507,
      "learning_rate": 9.993953383762342e-06,
      "loss": 0.3533,
      "step": 2311
    },
    {
      "epoch": 0.10609884814831812,
      "grad_norm": 0.46465012431144714,
      "learning_rate": 9.99394132312735e-06,
      "loss": 0.3691,
      "step": 2312
    },
    {
      "epoch": 0.10614473865357257,
      "grad_norm": 0.4867328405380249,
      "learning_rate": 9.993929250483504e-06,
      "loss": 0.4135,
      "step": 2313
    },
    {
      "epoch": 0.10619062915882704,
      "grad_norm": 0.5091094374656677,
      "learning_rate": 9.993917165830837e-06,
      "loss": 0.4449,
      "step": 2314
    },
    {
      "epoch": 0.1062365196640815,
      "grad_norm": 0.4904663562774658,
      "learning_rate": 9.993905069169374e-06,
      "loss": 0.3909,
      "step": 2315
    },
    {
      "epoch": 0.10628241016933597,
      "grad_norm": 0.4993605613708496,
      "learning_rate": 9.993892960499146e-06,
      "loss": 0.402,
      "step": 2316
    },
    {
      "epoch": 0.10632830067459043,
      "grad_norm": 0.5261427164077759,
      "learning_rate": 9.99388083982018e-06,
      "loss": 0.509,
      "step": 2317
    },
    {
      "epoch": 0.1063741911798449,
      "grad_norm": 0.5238061547279358,
      "learning_rate": 9.993868707132508e-06,
      "loss": 0.4949,
      "step": 2318
    },
    {
      "epoch": 0.10642008168509935,
      "grad_norm": 0.4810526967048645,
      "learning_rate": 9.993856562436158e-06,
      "loss": 0.3733,
      "step": 2319
    },
    {
      "epoch": 0.10646597219035382,
      "grad_norm": 0.5121590495109558,
      "learning_rate": 9.993844405731156e-06,
      "loss": 0.4591,
      "step": 2320
    },
    {
      "epoch": 0.10651186269560828,
      "grad_norm": 0.5276305079460144,
      "learning_rate": 9.993832237017539e-06,
      "loss": 0.5269,
      "step": 2321
    },
    {
      "epoch": 0.10655775320086275,
      "grad_norm": 0.4516938328742981,
      "learning_rate": 9.993820056295328e-06,
      "loss": 0.3456,
      "step": 2322
    },
    {
      "epoch": 0.1066036437061172,
      "grad_norm": 0.4453715980052948,
      "learning_rate": 9.993807863564557e-06,
      "loss": 0.3187,
      "step": 2323
    },
    {
      "epoch": 0.10664953421137166,
      "grad_norm": 0.4927120804786682,
      "learning_rate": 9.993795658825254e-06,
      "loss": 0.4846,
      "step": 2324
    },
    {
      "epoch": 0.10669542471662613,
      "grad_norm": 0.4254423975944519,
      "learning_rate": 9.99378344207745e-06,
      "loss": 0.3249,
      "step": 2325
    },
    {
      "epoch": 0.10674131522188059,
      "grad_norm": 0.48758989572525024,
      "learning_rate": 9.99377121332117e-06,
      "loss": 0.4594,
      "step": 2326
    },
    {
      "epoch": 0.10678720572713506,
      "grad_norm": 0.459878146648407,
      "learning_rate": 9.99375897255645e-06,
      "loss": 0.3656,
      "step": 2327
    },
    {
      "epoch": 0.10683309623238951,
      "grad_norm": 0.5259808301925659,
      "learning_rate": 9.993746719783313e-06,
      "loss": 0.5098,
      "step": 2328
    },
    {
      "epoch": 0.10687898673764398,
      "grad_norm": 0.4410202205181122,
      "learning_rate": 9.993734455001791e-06,
      "loss": 0.3947,
      "step": 2329
    },
    {
      "epoch": 0.10692487724289844,
      "grad_norm": 0.4721528887748718,
      "learning_rate": 9.993722178211916e-06,
      "loss": 0.4093,
      "step": 2330
    },
    {
      "epoch": 0.10697076774815291,
      "grad_norm": 0.48999449610710144,
      "learning_rate": 9.993709889413714e-06,
      "loss": 0.3596,
      "step": 2331
    },
    {
      "epoch": 0.10701665825340737,
      "grad_norm": 0.45347708463668823,
      "learning_rate": 9.993697588607216e-06,
      "loss": 0.3595,
      "step": 2332
    },
    {
      "epoch": 0.10706254875866184,
      "grad_norm": 0.4915441572666168,
      "learning_rate": 9.99368527579245e-06,
      "loss": 0.4159,
      "step": 2333
    },
    {
      "epoch": 0.10710843926391629,
      "grad_norm": 0.5159887075424194,
      "learning_rate": 9.993672950969449e-06,
      "loss": 0.4007,
      "step": 2334
    },
    {
      "epoch": 0.10715432976917076,
      "grad_norm": 0.4823332130908966,
      "learning_rate": 9.99366061413824e-06,
      "loss": 0.3842,
      "step": 2335
    },
    {
      "epoch": 0.10720022027442522,
      "grad_norm": 0.5124900341033936,
      "learning_rate": 9.993648265298852e-06,
      "loss": 0.5253,
      "step": 2336
    },
    {
      "epoch": 0.10724611077967969,
      "grad_norm": 0.5371134877204895,
      "learning_rate": 9.993635904451318e-06,
      "loss": 0.4226,
      "step": 2337
    },
    {
      "epoch": 0.10729200128493414,
      "grad_norm": 0.45335957407951355,
      "learning_rate": 9.993623531595663e-06,
      "loss": 0.3406,
      "step": 2338
    },
    {
      "epoch": 0.10733789179018861,
      "grad_norm": 0.4715803563594818,
      "learning_rate": 9.993611146731921e-06,
      "loss": 0.3892,
      "step": 2339
    },
    {
      "epoch": 0.10738378229544307,
      "grad_norm": 0.5375993847846985,
      "learning_rate": 9.993598749860121e-06,
      "loss": 0.4932,
      "step": 2340
    },
    {
      "epoch": 0.10742967280069754,
      "grad_norm": 0.5276365876197815,
      "learning_rate": 9.993586340980291e-06,
      "loss": 0.4471,
      "step": 2341
    },
    {
      "epoch": 0.107475563305952,
      "grad_norm": 0.5309797525405884,
      "learning_rate": 9.993573920092463e-06,
      "loss": 0.4651,
      "step": 2342
    },
    {
      "epoch": 0.10752145381120647,
      "grad_norm": 0.5680058598518372,
      "learning_rate": 9.993561487196663e-06,
      "loss": 0.4458,
      "step": 2343
    },
    {
      "epoch": 0.10756734431646092,
      "grad_norm": 0.4932131767272949,
      "learning_rate": 9.993549042292927e-06,
      "loss": 0.386,
      "step": 2344
    },
    {
      "epoch": 0.10761323482171539,
      "grad_norm": 0.47958359122276306,
      "learning_rate": 9.99353658538128e-06,
      "loss": 0.4351,
      "step": 2345
    },
    {
      "epoch": 0.10765912532696985,
      "grad_norm": 0.4573841989040375,
      "learning_rate": 9.993524116461752e-06,
      "loss": 0.3812,
      "step": 2346
    },
    {
      "epoch": 0.10770501583222432,
      "grad_norm": 0.4970970153808594,
      "learning_rate": 9.993511635534377e-06,
      "loss": 0.3835,
      "step": 2347
    },
    {
      "epoch": 0.10775090633747877,
      "grad_norm": 0.4605258107185364,
      "learning_rate": 9.993499142599182e-06,
      "loss": 0.3937,
      "step": 2348
    },
    {
      "epoch": 0.10779679684273324,
      "grad_norm": 0.4642235338687897,
      "learning_rate": 9.993486637656196e-06,
      "loss": 0.4232,
      "step": 2349
    },
    {
      "epoch": 0.1078426873479877,
      "grad_norm": 0.43981191515922546,
      "learning_rate": 9.993474120705453e-06,
      "loss": 0.3081,
      "step": 2350
    },
    {
      "epoch": 0.10788857785324217,
      "grad_norm": 0.5256198644638062,
      "learning_rate": 9.993461591746978e-06,
      "loss": 0.3888,
      "step": 2351
    },
    {
      "epoch": 0.10793446835849663,
      "grad_norm": 0.5696247220039368,
      "learning_rate": 9.993449050780807e-06,
      "loss": 0.4899,
      "step": 2352
    },
    {
      "epoch": 0.10798035886375108,
      "grad_norm": 0.4480229616165161,
      "learning_rate": 9.993436497806965e-06,
      "loss": 0.3523,
      "step": 2353
    },
    {
      "epoch": 0.10802624936900555,
      "grad_norm": 0.4560874402523041,
      "learning_rate": 9.993423932825485e-06,
      "loss": 0.294,
      "step": 2354
    },
    {
      "epoch": 0.10807213987426001,
      "grad_norm": 0.509881854057312,
      "learning_rate": 9.993411355836395e-06,
      "loss": 0.4791,
      "step": 2355
    },
    {
      "epoch": 0.10811803037951448,
      "grad_norm": 0.4819093942642212,
      "learning_rate": 9.993398766839728e-06,
      "loss": 0.4125,
      "step": 2356
    },
    {
      "epoch": 0.10816392088476894,
      "grad_norm": 0.489335834980011,
      "learning_rate": 9.993386165835512e-06,
      "loss": 0.4444,
      "step": 2357
    },
    {
      "epoch": 0.1082098113900234,
      "grad_norm": 0.475861519575119,
      "learning_rate": 9.993373552823779e-06,
      "loss": 0.429,
      "step": 2358
    },
    {
      "epoch": 0.10825570189527786,
      "grad_norm": 0.5080271363258362,
      "learning_rate": 9.993360927804558e-06,
      "loss": 0.4447,
      "step": 2359
    },
    {
      "epoch": 0.10830159240053233,
      "grad_norm": 0.49017783999443054,
      "learning_rate": 9.99334829077788e-06,
      "loss": 0.4698,
      "step": 2360
    },
    {
      "epoch": 0.10834748290578679,
      "grad_norm": 0.4478916525840759,
      "learning_rate": 9.993335641743776e-06,
      "loss": 0.3644,
      "step": 2361
    },
    {
      "epoch": 0.10839337341104126,
      "grad_norm": 0.45496076345443726,
      "learning_rate": 9.993322980702276e-06,
      "loss": 0.3563,
      "step": 2362
    },
    {
      "epoch": 0.10843926391629571,
      "grad_norm": 0.5220422148704529,
      "learning_rate": 9.99331030765341e-06,
      "loss": 0.5412,
      "step": 2363
    },
    {
      "epoch": 0.10848515442155018,
      "grad_norm": 0.5577587485313416,
      "learning_rate": 9.993297622597207e-06,
      "loss": 0.5075,
      "step": 2364
    },
    {
      "epoch": 0.10853104492680464,
      "grad_norm": 0.49074456095695496,
      "learning_rate": 9.9932849255337e-06,
      "loss": 0.4133,
      "step": 2365
    },
    {
      "epoch": 0.10857693543205911,
      "grad_norm": 0.4843331575393677,
      "learning_rate": 9.993272216462918e-06,
      "loss": 0.4323,
      "step": 2366
    },
    {
      "epoch": 0.10862282593731357,
      "grad_norm": 0.4688464105129242,
      "learning_rate": 9.993259495384894e-06,
      "loss": 0.3456,
      "step": 2367
    },
    {
      "epoch": 0.10866871644256804,
      "grad_norm": 0.4599507451057434,
      "learning_rate": 9.993246762299656e-06,
      "loss": 0.352,
      "step": 2368
    },
    {
      "epoch": 0.10871460694782249,
      "grad_norm": 0.5474652647972107,
      "learning_rate": 9.993234017207236e-06,
      "loss": 0.391,
      "step": 2369
    },
    {
      "epoch": 0.10876049745307696,
      "grad_norm": 0.5002933740615845,
      "learning_rate": 9.993221260107664e-06,
      "loss": 0.3981,
      "step": 2370
    },
    {
      "epoch": 0.10880638795833142,
      "grad_norm": 0.5205653309822083,
      "learning_rate": 9.993208491000969e-06,
      "loss": 0.4138,
      "step": 2371
    },
    {
      "epoch": 0.10885227846358589,
      "grad_norm": 0.44890275597572327,
      "learning_rate": 9.993195709887186e-06,
      "loss": 0.3429,
      "step": 2372
    },
    {
      "epoch": 0.10889816896884034,
      "grad_norm": 0.4605308771133423,
      "learning_rate": 9.993182916766343e-06,
      "loss": 0.3621,
      "step": 2373
    },
    {
      "epoch": 0.10894405947409481,
      "grad_norm": 0.4829280376434326,
      "learning_rate": 9.993170111638469e-06,
      "loss": 0.4611,
      "step": 2374
    },
    {
      "epoch": 0.10898994997934927,
      "grad_norm": 0.4615125060081482,
      "learning_rate": 9.993157294503596e-06,
      "loss": 0.4265,
      "step": 2375
    },
    {
      "epoch": 0.10903584048460374,
      "grad_norm": 0.48623043298721313,
      "learning_rate": 9.99314446536176e-06,
      "loss": 0.4458,
      "step": 2376
    },
    {
      "epoch": 0.1090817309898582,
      "grad_norm": 0.48533710837364197,
      "learning_rate": 9.993131624212984e-06,
      "loss": 0.449,
      "step": 2377
    },
    {
      "epoch": 0.10912762149511267,
      "grad_norm": 0.4932247996330261,
      "learning_rate": 9.993118771057302e-06,
      "loss": 0.392,
      "step": 2378
    },
    {
      "epoch": 0.10917351200036712,
      "grad_norm": 0.44672098755836487,
      "learning_rate": 9.993105905894747e-06,
      "loss": 0.3543,
      "step": 2379
    },
    {
      "epoch": 0.1092194025056216,
      "grad_norm": 0.46219930052757263,
      "learning_rate": 9.993093028725349e-06,
      "loss": 0.3476,
      "step": 2380
    },
    {
      "epoch": 0.10926529301087605,
      "grad_norm": 0.49842169880867004,
      "learning_rate": 9.993080139549137e-06,
      "loss": 0.372,
      "step": 2381
    },
    {
      "epoch": 0.10931118351613052,
      "grad_norm": 0.47868072986602783,
      "learning_rate": 9.993067238366141e-06,
      "loss": 0.3983,
      "step": 2382
    },
    {
      "epoch": 0.10935707402138498,
      "grad_norm": 0.4512401819229126,
      "learning_rate": 9.993054325176397e-06,
      "loss": 0.3671,
      "step": 2383
    },
    {
      "epoch": 0.10940296452663943,
      "grad_norm": 0.5758407115936279,
      "learning_rate": 9.993041399979931e-06,
      "loss": 0.4839,
      "step": 2384
    },
    {
      "epoch": 0.1094488550318939,
      "grad_norm": 0.4671958088874817,
      "learning_rate": 9.993028462776777e-06,
      "loss": 0.3999,
      "step": 2385
    },
    {
      "epoch": 0.10949474553714836,
      "grad_norm": 0.4457875192165375,
      "learning_rate": 9.993015513566965e-06,
      "loss": 0.3492,
      "step": 2386
    },
    {
      "epoch": 0.10954063604240283,
      "grad_norm": 0.49180343747138977,
      "learning_rate": 9.99300255235053e-06,
      "loss": 0.4583,
      "step": 2387
    },
    {
      "epoch": 0.10958652654765728,
      "grad_norm": 0.5329729318618774,
      "learning_rate": 9.992989579127497e-06,
      "loss": 0.5426,
      "step": 2388
    },
    {
      "epoch": 0.10963241705291175,
      "grad_norm": 0.454342782497406,
      "learning_rate": 9.992976593897898e-06,
      "loss": 0.3241,
      "step": 2389
    },
    {
      "epoch": 0.10967830755816621,
      "grad_norm": 0.466134637594223,
      "learning_rate": 9.99296359666177e-06,
      "loss": 0.381,
      "step": 2390
    },
    {
      "epoch": 0.10972419806342068,
      "grad_norm": 0.5188995599746704,
      "learning_rate": 9.992950587419137e-06,
      "loss": 0.4044,
      "step": 2391
    },
    {
      "epoch": 0.10977008856867514,
      "grad_norm": 0.5016509294509888,
      "learning_rate": 9.992937566170036e-06,
      "loss": 0.438,
      "step": 2392
    },
    {
      "epoch": 0.1098159790739296,
      "grad_norm": 0.4966766834259033,
      "learning_rate": 9.992924532914493e-06,
      "loss": 0.4513,
      "step": 2393
    },
    {
      "epoch": 0.10986186957918406,
      "grad_norm": 0.480093777179718,
      "learning_rate": 9.992911487652545e-06,
      "loss": 0.4445,
      "step": 2394
    },
    {
      "epoch": 0.10990776008443853,
      "grad_norm": 0.45770004391670227,
      "learning_rate": 9.992898430384222e-06,
      "loss": 0.3784,
      "step": 2395
    },
    {
      "epoch": 0.10995365058969299,
      "grad_norm": 0.46298423409461975,
      "learning_rate": 9.99288536110955e-06,
      "loss": 0.3823,
      "step": 2396
    },
    {
      "epoch": 0.10999954109494746,
      "grad_norm": 0.5011987686157227,
      "learning_rate": 9.992872279828566e-06,
      "loss": 0.4484,
      "step": 2397
    },
    {
      "epoch": 0.11004543160020192,
      "grad_norm": 0.4758690893650055,
      "learning_rate": 9.992859186541302e-06,
      "loss": 0.4126,
      "step": 2398
    },
    {
      "epoch": 0.11009132210545639,
      "grad_norm": 0.4889727830886841,
      "learning_rate": 9.992846081247786e-06,
      "loss": 0.4645,
      "step": 2399
    },
    {
      "epoch": 0.11013721261071084,
      "grad_norm": 0.4708193838596344,
      "learning_rate": 9.992832963948052e-06,
      "loss": 0.3602,
      "step": 2400
    },
    {
      "epoch": 0.11018310311596531,
      "grad_norm": 0.42878010869026184,
      "learning_rate": 9.99281983464213e-06,
      "loss": 0.3197,
      "step": 2401
    },
    {
      "epoch": 0.11022899362121977,
      "grad_norm": 0.5145881175994873,
      "learning_rate": 9.99280669333005e-06,
      "loss": 0.4186,
      "step": 2402
    },
    {
      "epoch": 0.11027488412647424,
      "grad_norm": 0.5047385692596436,
      "learning_rate": 9.992793540011847e-06,
      "loss": 0.3386,
      "step": 2403
    },
    {
      "epoch": 0.1103207746317287,
      "grad_norm": 0.504069447517395,
      "learning_rate": 9.992780374687552e-06,
      "loss": 0.4812,
      "step": 2404
    },
    {
      "epoch": 0.11036666513698316,
      "grad_norm": 0.49924415349960327,
      "learning_rate": 9.992767197357196e-06,
      "loss": 0.4426,
      "step": 2405
    },
    {
      "epoch": 0.11041255564223762,
      "grad_norm": 0.45040181279182434,
      "learning_rate": 9.99275400802081e-06,
      "loss": 0.3412,
      "step": 2406
    },
    {
      "epoch": 0.11045844614749209,
      "grad_norm": 0.46537142992019653,
      "learning_rate": 9.992740806678427e-06,
      "loss": 0.4099,
      "step": 2407
    },
    {
      "epoch": 0.11050433665274655,
      "grad_norm": 0.521116316318512,
      "learning_rate": 9.992727593330077e-06,
      "loss": 0.5319,
      "step": 2408
    },
    {
      "epoch": 0.11055022715800102,
      "grad_norm": 0.49834680557250977,
      "learning_rate": 9.992714367975793e-06,
      "loss": 0.4444,
      "step": 2409
    },
    {
      "epoch": 0.11059611766325547,
      "grad_norm": 0.4632551074028015,
      "learning_rate": 9.992701130615607e-06,
      "loss": 0.3695,
      "step": 2410
    },
    {
      "epoch": 0.11064200816850994,
      "grad_norm": 0.5194320678710938,
      "learning_rate": 9.992687881249552e-06,
      "loss": 0.5201,
      "step": 2411
    },
    {
      "epoch": 0.1106878986737644,
      "grad_norm": 0.5251150131225586,
      "learning_rate": 9.992674619877656e-06,
      "loss": 0.494,
      "step": 2412
    },
    {
      "epoch": 0.11073378917901885,
      "grad_norm": 0.5041511654853821,
      "learning_rate": 9.992661346499956e-06,
      "loss": 0.48,
      "step": 2413
    },
    {
      "epoch": 0.11077967968427332,
      "grad_norm": 0.5411491394042969,
      "learning_rate": 9.99264806111648e-06,
      "loss": 0.5058,
      "step": 2414
    },
    {
      "epoch": 0.11082557018952778,
      "grad_norm": 0.529249370098114,
      "learning_rate": 9.99263476372726e-06,
      "loss": 0.4878,
      "step": 2415
    },
    {
      "epoch": 0.11087146069478225,
      "grad_norm": 0.5474193096160889,
      "learning_rate": 9.992621454332329e-06,
      "loss": 0.4853,
      "step": 2416
    },
    {
      "epoch": 0.1109173512000367,
      "grad_norm": 0.5019587874412537,
      "learning_rate": 9.992608132931721e-06,
      "loss": 0.5294,
      "step": 2417
    },
    {
      "epoch": 0.11096324170529118,
      "grad_norm": 0.46557125449180603,
      "learning_rate": 9.992594799525466e-06,
      "loss": 0.3972,
      "step": 2418
    },
    {
      "epoch": 0.11100913221054563,
      "grad_norm": 0.4534105956554413,
      "learning_rate": 9.992581454113595e-06,
      "loss": 0.3619,
      "step": 2419
    },
    {
      "epoch": 0.1110550227158001,
      "grad_norm": 0.45725610852241516,
      "learning_rate": 9.99256809669614e-06,
      "loss": 0.3296,
      "step": 2420
    },
    {
      "epoch": 0.11110091322105456,
      "grad_norm": 0.4957942068576813,
      "learning_rate": 9.992554727273136e-06,
      "loss": 0.3492,
      "step": 2421
    },
    {
      "epoch": 0.11114680372630903,
      "grad_norm": 0.4696333408355713,
      "learning_rate": 9.992541345844615e-06,
      "loss": 0.4115,
      "step": 2422
    },
    {
      "epoch": 0.11119269423156349,
      "grad_norm": 0.45193302631378174,
      "learning_rate": 9.992527952410607e-06,
      "loss": 0.4146,
      "step": 2423
    },
    {
      "epoch": 0.11123858473681796,
      "grad_norm": 0.5438513159751892,
      "learning_rate": 9.992514546971144e-06,
      "loss": 0.5345,
      "step": 2424
    },
    {
      "epoch": 0.11128447524207241,
      "grad_norm": 0.4902760088443756,
      "learning_rate": 9.99250112952626e-06,
      "loss": 0.46,
      "step": 2425
    },
    {
      "epoch": 0.11133036574732688,
      "grad_norm": 0.5059555172920227,
      "learning_rate": 9.992487700075986e-06,
      "loss": 0.4454,
      "step": 2426
    },
    {
      "epoch": 0.11137625625258134,
      "grad_norm": 0.48046404123306274,
      "learning_rate": 9.992474258620353e-06,
      "loss": 0.4067,
      "step": 2427
    },
    {
      "epoch": 0.11142214675783581,
      "grad_norm": 0.4960084557533264,
      "learning_rate": 9.9924608051594e-06,
      "loss": 0.4689,
      "step": 2428
    },
    {
      "epoch": 0.11146803726309026,
      "grad_norm": 0.4739789664745331,
      "learning_rate": 9.99244733969315e-06,
      "loss": 0.3776,
      "step": 2429
    },
    {
      "epoch": 0.11151392776834473,
      "grad_norm": 0.4622792601585388,
      "learning_rate": 9.99243386222164e-06,
      "loss": 0.3687,
      "step": 2430
    },
    {
      "epoch": 0.11155981827359919,
      "grad_norm": 0.46889933943748474,
      "learning_rate": 9.992420372744905e-06,
      "loss": 0.3734,
      "step": 2431
    },
    {
      "epoch": 0.11160570877885366,
      "grad_norm": 0.5245449542999268,
      "learning_rate": 9.992406871262971e-06,
      "loss": 0.4792,
      "step": 2432
    },
    {
      "epoch": 0.11165159928410812,
      "grad_norm": 0.500846266746521,
      "learning_rate": 9.992393357775877e-06,
      "loss": 0.4371,
      "step": 2433
    },
    {
      "epoch": 0.11169748978936259,
      "grad_norm": 0.4570404887199402,
      "learning_rate": 9.992379832283653e-06,
      "loss": 0.3525,
      "step": 2434
    },
    {
      "epoch": 0.11174338029461704,
      "grad_norm": 0.506668210029602,
      "learning_rate": 9.992366294786329e-06,
      "loss": 0.4516,
      "step": 2435
    },
    {
      "epoch": 0.11178927079987151,
      "grad_norm": 0.5264373421669006,
      "learning_rate": 9.99235274528394e-06,
      "loss": 0.4056,
      "step": 2436
    },
    {
      "epoch": 0.11183516130512597,
      "grad_norm": 0.4368688464164734,
      "learning_rate": 9.992339183776518e-06,
      "loss": 0.3442,
      "step": 2437
    },
    {
      "epoch": 0.11188105181038044,
      "grad_norm": 0.5090033411979675,
      "learning_rate": 9.992325610264097e-06,
      "loss": 0.5376,
      "step": 2438
    },
    {
      "epoch": 0.1119269423156349,
      "grad_norm": 0.5053332448005676,
      "learning_rate": 9.992312024746708e-06,
      "loss": 0.5406,
      "step": 2439
    },
    {
      "epoch": 0.11197283282088936,
      "grad_norm": 0.511529266834259,
      "learning_rate": 9.992298427224384e-06,
      "loss": 0.4448,
      "step": 2440
    },
    {
      "epoch": 0.11201872332614382,
      "grad_norm": 0.46909859776496887,
      "learning_rate": 9.99228481769716e-06,
      "loss": 0.3173,
      "step": 2441
    },
    {
      "epoch": 0.11206461383139828,
      "grad_norm": 0.4972430169582367,
      "learning_rate": 9.992271196165065e-06,
      "loss": 0.4328,
      "step": 2442
    },
    {
      "epoch": 0.11211050433665275,
      "grad_norm": 0.48291459679603577,
      "learning_rate": 9.992257562628133e-06,
      "loss": 0.4198,
      "step": 2443
    },
    {
      "epoch": 0.1121563948419072,
      "grad_norm": 0.4549403786659241,
      "learning_rate": 9.992243917086397e-06,
      "loss": 0.3854,
      "step": 2444
    },
    {
      "epoch": 0.11220228534716167,
      "grad_norm": 0.43886083364486694,
      "learning_rate": 9.992230259539892e-06,
      "loss": 0.357,
      "step": 2445
    },
    {
      "epoch": 0.11224817585241613,
      "grad_norm": 0.5201106071472168,
      "learning_rate": 9.992216589988647e-06,
      "loss": 0.5056,
      "step": 2446
    },
    {
      "epoch": 0.1122940663576706,
      "grad_norm": 0.4949498176574707,
      "learning_rate": 9.992202908432696e-06,
      "loss": 0.4781,
      "step": 2447
    },
    {
      "epoch": 0.11233995686292506,
      "grad_norm": 0.4632495641708374,
      "learning_rate": 9.992189214872074e-06,
      "loss": 0.3345,
      "step": 2448
    },
    {
      "epoch": 0.11238584736817953,
      "grad_norm": 0.5333968997001648,
      "learning_rate": 9.992175509306812e-06,
      "loss": 0.5341,
      "step": 2449
    },
    {
      "epoch": 0.11243173787343398,
      "grad_norm": 0.5001994371414185,
      "learning_rate": 9.992161791736945e-06,
      "loss": 0.3955,
      "step": 2450
    },
    {
      "epoch": 0.11247762837868845,
      "grad_norm": 0.5235245227813721,
      "learning_rate": 9.992148062162503e-06,
      "loss": 0.5332,
      "step": 2451
    },
    {
      "epoch": 0.11252351888394291,
      "grad_norm": 0.5011329054832458,
      "learning_rate": 9.99213432058352e-06,
      "loss": 0.4945,
      "step": 2452
    },
    {
      "epoch": 0.11256940938919738,
      "grad_norm": 0.529055655002594,
      "learning_rate": 9.99212056700003e-06,
      "loss": 0.5399,
      "step": 2453
    },
    {
      "epoch": 0.11261529989445183,
      "grad_norm": 0.4809805452823639,
      "learning_rate": 9.992106801412065e-06,
      "loss": 0.3989,
      "step": 2454
    },
    {
      "epoch": 0.1126611903997063,
      "grad_norm": 0.4782566428184509,
      "learning_rate": 9.99209302381966e-06,
      "loss": 0.3819,
      "step": 2455
    },
    {
      "epoch": 0.11270708090496076,
      "grad_norm": 0.49292415380477905,
      "learning_rate": 9.992079234222846e-06,
      "loss": 0.4182,
      "step": 2456
    },
    {
      "epoch": 0.11275297141021523,
      "grad_norm": 0.45130375027656555,
      "learning_rate": 9.992065432621657e-06,
      "loss": 0.3879,
      "step": 2457
    },
    {
      "epoch": 0.11279886191546969,
      "grad_norm": 0.46563634276390076,
      "learning_rate": 9.992051619016126e-06,
      "loss": 0.4124,
      "step": 2458
    },
    {
      "epoch": 0.11284475242072416,
      "grad_norm": 0.43359073996543884,
      "learning_rate": 9.992037793406287e-06,
      "loss": 0.3345,
      "step": 2459
    },
    {
      "epoch": 0.11289064292597861,
      "grad_norm": 0.5066606402397156,
      "learning_rate": 9.992023955792172e-06,
      "loss": 0.4848,
      "step": 2460
    },
    {
      "epoch": 0.11293653343123308,
      "grad_norm": 0.4598350524902344,
      "learning_rate": 9.992010106173816e-06,
      "loss": 0.3666,
      "step": 2461
    },
    {
      "epoch": 0.11298242393648754,
      "grad_norm": 0.46526703238487244,
      "learning_rate": 9.99199624455125e-06,
      "loss": 0.4273,
      "step": 2462
    },
    {
      "epoch": 0.11302831444174201,
      "grad_norm": 0.4622974693775177,
      "learning_rate": 9.991982370924509e-06,
      "loss": 0.3942,
      "step": 2463
    },
    {
      "epoch": 0.11307420494699646,
      "grad_norm": 0.450998991727829,
      "learning_rate": 9.991968485293627e-06,
      "loss": 0.3573,
      "step": 2464
    },
    {
      "epoch": 0.11312009545225093,
      "grad_norm": 0.447102814912796,
      "learning_rate": 9.991954587658634e-06,
      "loss": 0.3541,
      "step": 2465
    },
    {
      "epoch": 0.11316598595750539,
      "grad_norm": 0.4572606086730957,
      "learning_rate": 9.991940678019569e-06,
      "loss": 0.3263,
      "step": 2466
    },
    {
      "epoch": 0.11321187646275986,
      "grad_norm": 0.5254818201065063,
      "learning_rate": 9.99192675637646e-06,
      "loss": 0.479,
      "step": 2467
    },
    {
      "epoch": 0.11325776696801432,
      "grad_norm": 0.4464569389820099,
      "learning_rate": 9.991912822729342e-06,
      "loss": 0.3968,
      "step": 2468
    },
    {
      "epoch": 0.11330365747326879,
      "grad_norm": 0.48612359166145325,
      "learning_rate": 9.99189887707825e-06,
      "loss": 0.3998,
      "step": 2469
    },
    {
      "epoch": 0.11334954797852324,
      "grad_norm": 0.49350661039352417,
      "learning_rate": 9.991884919423217e-06,
      "loss": 0.4847,
      "step": 2470
    },
    {
      "epoch": 0.1133954384837777,
      "grad_norm": 0.470751017332077,
      "learning_rate": 9.991870949764276e-06,
      "loss": 0.4253,
      "step": 2471
    },
    {
      "epoch": 0.11344132898903217,
      "grad_norm": 0.4237576723098755,
      "learning_rate": 9.991856968101462e-06,
      "loss": 0.2862,
      "step": 2472
    },
    {
      "epoch": 0.11348721949428663,
      "grad_norm": 0.448932021856308,
      "learning_rate": 9.991842974434805e-06,
      "loss": 0.3259,
      "step": 2473
    },
    {
      "epoch": 0.1135331099995411,
      "grad_norm": 0.49851247668266296,
      "learning_rate": 9.991828968764343e-06,
      "loss": 0.4279,
      "step": 2474
    },
    {
      "epoch": 0.11357900050479555,
      "grad_norm": 0.48448216915130615,
      "learning_rate": 9.991814951090106e-06,
      "loss": 0.4553,
      "step": 2475
    },
    {
      "epoch": 0.11362489101005002,
      "grad_norm": 0.5331076383590698,
      "learning_rate": 9.991800921412132e-06,
      "loss": 0.4624,
      "step": 2476
    },
    {
      "epoch": 0.11367078151530448,
      "grad_norm": 0.47850388288497925,
      "learning_rate": 9.99178687973045e-06,
      "loss": 0.4941,
      "step": 2477
    },
    {
      "epoch": 0.11371667202055895,
      "grad_norm": 0.5628966689109802,
      "learning_rate": 9.991772826045097e-06,
      "loss": 0.526,
      "step": 2478
    },
    {
      "epoch": 0.1137625625258134,
      "grad_norm": 0.44776520133018494,
      "learning_rate": 9.991758760356105e-06,
      "loss": 0.368,
      "step": 2479
    },
    {
      "epoch": 0.11380845303106787,
      "grad_norm": 0.4841279983520508,
      "learning_rate": 9.99174468266351e-06,
      "loss": 0.4209,
      "step": 2480
    },
    {
      "epoch": 0.11385434353632233,
      "grad_norm": 0.4831261932849884,
      "learning_rate": 9.991730592967344e-06,
      "loss": 0.4023,
      "step": 2481
    },
    {
      "epoch": 0.1139002340415768,
      "grad_norm": 0.47709277272224426,
      "learning_rate": 9.99171649126764e-06,
      "loss": 0.441,
      "step": 2482
    },
    {
      "epoch": 0.11394612454683126,
      "grad_norm": 0.5035306811332703,
      "learning_rate": 9.991702377564435e-06,
      "loss": 0.543,
      "step": 2483
    },
    {
      "epoch": 0.11399201505208573,
      "grad_norm": 0.45184990763664246,
      "learning_rate": 9.991688251857761e-06,
      "loss": 0.3407,
      "step": 2484
    },
    {
      "epoch": 0.11403790555734018,
      "grad_norm": 0.4857518970966339,
      "learning_rate": 9.991674114147651e-06,
      "loss": 0.432,
      "step": 2485
    },
    {
      "epoch": 0.11408379606259465,
      "grad_norm": 0.45240625739097595,
      "learning_rate": 9.991659964434142e-06,
      "loss": 0.4001,
      "step": 2486
    },
    {
      "epoch": 0.11412968656784911,
      "grad_norm": 0.5007497668266296,
      "learning_rate": 9.991645802717265e-06,
      "loss": 0.4381,
      "step": 2487
    },
    {
      "epoch": 0.11417557707310358,
      "grad_norm": 0.4801659882068634,
      "learning_rate": 9.991631628997055e-06,
      "loss": 0.3593,
      "step": 2488
    },
    {
      "epoch": 0.11422146757835804,
      "grad_norm": 0.4322071969509125,
      "learning_rate": 9.991617443273547e-06,
      "loss": 0.3531,
      "step": 2489
    },
    {
      "epoch": 0.1142673580836125,
      "grad_norm": 0.49512311816215515,
      "learning_rate": 9.991603245546775e-06,
      "loss": 0.4317,
      "step": 2490
    },
    {
      "epoch": 0.11431324858886696,
      "grad_norm": 0.4706094563007355,
      "learning_rate": 9.99158903581677e-06,
      "loss": 0.4298,
      "step": 2491
    },
    {
      "epoch": 0.11435913909412143,
      "grad_norm": 0.49296361207962036,
      "learning_rate": 9.991574814083572e-06,
      "loss": 0.5042,
      "step": 2492
    },
    {
      "epoch": 0.11440502959937589,
      "grad_norm": 0.46887606382369995,
      "learning_rate": 9.99156058034721e-06,
      "loss": 0.3861,
      "step": 2493
    },
    {
      "epoch": 0.11445092010463036,
      "grad_norm": 0.529515266418457,
      "learning_rate": 9.991546334607721e-06,
      "loss": 0.4387,
      "step": 2494
    },
    {
      "epoch": 0.11449681060988481,
      "grad_norm": 0.49518662691116333,
      "learning_rate": 9.991532076865138e-06,
      "loss": 0.4588,
      "step": 2495
    },
    {
      "epoch": 0.11454270111513928,
      "grad_norm": 0.6056028008460999,
      "learning_rate": 9.991517807119497e-06,
      "loss": 0.5964,
      "step": 2496
    },
    {
      "epoch": 0.11458859162039374,
      "grad_norm": 0.46793216466903687,
      "learning_rate": 9.99150352537083e-06,
      "loss": 0.383,
      "step": 2497
    },
    {
      "epoch": 0.11463448212564821,
      "grad_norm": 0.47859108448028564,
      "learning_rate": 9.991489231619171e-06,
      "loss": 0.4231,
      "step": 2498
    },
    {
      "epoch": 0.11468037263090267,
      "grad_norm": 0.5051490664482117,
      "learning_rate": 9.991474925864557e-06,
      "loss": 0.4568,
      "step": 2499
    },
    {
      "epoch": 0.11472626313615712,
      "grad_norm": 0.4955484867095947,
      "learning_rate": 9.991460608107023e-06,
      "loss": 0.4417,
      "step": 2500
    },
    {
      "epoch": 0.11477215364141159,
      "grad_norm": 0.49946221709251404,
      "learning_rate": 9.9914462783466e-06,
      "loss": 0.3824,
      "step": 2501
    },
    {
      "epoch": 0.11481804414666605,
      "grad_norm": 0.46977555751800537,
      "learning_rate": 9.991431936583322e-06,
      "loss": 0.3355,
      "step": 2502
    },
    {
      "epoch": 0.11486393465192052,
      "grad_norm": 0.5135558247566223,
      "learning_rate": 9.991417582817229e-06,
      "loss": 0.4024,
      "step": 2503
    },
    {
      "epoch": 0.11490982515717497,
      "grad_norm": 0.4645470082759857,
      "learning_rate": 9.99140321704835e-06,
      "loss": 0.4053,
      "step": 2504
    },
    {
      "epoch": 0.11495571566242944,
      "grad_norm": 0.4802762567996979,
      "learning_rate": 9.991388839276722e-06,
      "loss": 0.3575,
      "step": 2505
    },
    {
      "epoch": 0.1150016061676839,
      "grad_norm": 0.4936521351337433,
      "learning_rate": 9.991374449502378e-06,
      "loss": 0.4517,
      "step": 2506
    },
    {
      "epoch": 0.11504749667293837,
      "grad_norm": 0.48894599080085754,
      "learning_rate": 9.991360047725356e-06,
      "loss": 0.404,
      "step": 2507
    },
    {
      "epoch": 0.11509338717819283,
      "grad_norm": 0.4959702491760254,
      "learning_rate": 9.991345633945689e-06,
      "loss": 0.4437,
      "step": 2508
    },
    {
      "epoch": 0.1151392776834473,
      "grad_norm": 0.4830079972743988,
      "learning_rate": 9.991331208163409e-06,
      "loss": 0.3637,
      "step": 2509
    },
    {
      "epoch": 0.11518516818870175,
      "grad_norm": 0.5489052534103394,
      "learning_rate": 9.991316770378552e-06,
      "loss": 0.5557,
      "step": 2510
    },
    {
      "epoch": 0.11523105869395622,
      "grad_norm": 0.5327038764953613,
      "learning_rate": 9.991302320591155e-06,
      "loss": 0.4937,
      "step": 2511
    },
    {
      "epoch": 0.11527694919921068,
      "grad_norm": 0.5096604228019714,
      "learning_rate": 9.99128785880125e-06,
      "loss": 0.4203,
      "step": 2512
    },
    {
      "epoch": 0.11532283970446515,
      "grad_norm": 0.483519583940506,
      "learning_rate": 9.991273385008875e-06,
      "loss": 0.3665,
      "step": 2513
    },
    {
      "epoch": 0.1153687302097196,
      "grad_norm": 0.4577787518501282,
      "learning_rate": 9.991258899214062e-06,
      "loss": 0.3305,
      "step": 2514
    },
    {
      "epoch": 0.11541462071497408,
      "grad_norm": 0.44032642245292664,
      "learning_rate": 9.991244401416846e-06,
      "loss": 0.349,
      "step": 2515
    },
    {
      "epoch": 0.11546051122022853,
      "grad_norm": 0.46372753381729126,
      "learning_rate": 9.991229891617263e-06,
      "loss": 0.3502,
      "step": 2516
    },
    {
      "epoch": 0.115506401725483,
      "grad_norm": 0.5140289068222046,
      "learning_rate": 9.991215369815347e-06,
      "loss": 0.4978,
      "step": 2517
    },
    {
      "epoch": 0.11555229223073746,
      "grad_norm": 0.4673903286457062,
      "learning_rate": 9.991200836011133e-06,
      "loss": 0.3771,
      "step": 2518
    },
    {
      "epoch": 0.11559818273599193,
      "grad_norm": 0.5397522449493408,
      "learning_rate": 9.991186290204657e-06,
      "loss": 0.4595,
      "step": 2519
    },
    {
      "epoch": 0.11564407324124638,
      "grad_norm": 0.5108731985092163,
      "learning_rate": 9.991171732395953e-06,
      "loss": 0.4213,
      "step": 2520
    },
    {
      "epoch": 0.11568996374650085,
      "grad_norm": 0.5284456014633179,
      "learning_rate": 9.991157162585056e-06,
      "loss": 0.4834,
      "step": 2521
    },
    {
      "epoch": 0.11573585425175531,
      "grad_norm": 0.5100253820419312,
      "learning_rate": 9.991142580772003e-06,
      "loss": 0.5323,
      "step": 2522
    },
    {
      "epoch": 0.11578174475700978,
      "grad_norm": 0.6847053170204163,
      "learning_rate": 9.991127986956826e-06,
      "loss": 0.3686,
      "step": 2523
    },
    {
      "epoch": 0.11582763526226424,
      "grad_norm": 0.493842750787735,
      "learning_rate": 9.991113381139561e-06,
      "loss": 0.3763,
      "step": 2524
    },
    {
      "epoch": 0.1158735257675187,
      "grad_norm": 0.4907994568347931,
      "learning_rate": 9.991098763320243e-06,
      "loss": 0.4254,
      "step": 2525
    },
    {
      "epoch": 0.11591941627277316,
      "grad_norm": 0.4895491302013397,
      "learning_rate": 9.991084133498909e-06,
      "loss": 0.4286,
      "step": 2526
    },
    {
      "epoch": 0.11596530677802763,
      "grad_norm": 0.4673796594142914,
      "learning_rate": 9.991069491675592e-06,
      "loss": 0.4089,
      "step": 2527
    },
    {
      "epoch": 0.11601119728328209,
      "grad_norm": 0.4665570855140686,
      "learning_rate": 9.99105483785033e-06,
      "loss": 0.3868,
      "step": 2528
    },
    {
      "epoch": 0.11605708778853656,
      "grad_norm": 0.4687078893184662,
      "learning_rate": 9.991040172023155e-06,
      "loss": 0.3428,
      "step": 2529
    },
    {
      "epoch": 0.11610297829379101,
      "grad_norm": 0.588813304901123,
      "learning_rate": 9.991025494194102e-06,
      "loss": 0.5647,
      "step": 2530
    },
    {
      "epoch": 0.11614886879904547,
      "grad_norm": 0.500828742980957,
      "learning_rate": 9.99101080436321e-06,
      "loss": 0.425,
      "step": 2531
    },
    {
      "epoch": 0.11619475930429994,
      "grad_norm": 0.540797233581543,
      "learning_rate": 9.990996102530511e-06,
      "loss": 0.4804,
      "step": 2532
    },
    {
      "epoch": 0.1162406498095544,
      "grad_norm": 0.45883938670158386,
      "learning_rate": 9.990981388696043e-06,
      "loss": 0.4273,
      "step": 2533
    },
    {
      "epoch": 0.11628654031480887,
      "grad_norm": 0.5134001970291138,
      "learning_rate": 9.99096666285984e-06,
      "loss": 0.4881,
      "step": 2534
    },
    {
      "epoch": 0.11633243082006332,
      "grad_norm": 0.5219642519950867,
      "learning_rate": 9.990951925021936e-06,
      "loss": 0.4852,
      "step": 2535
    },
    {
      "epoch": 0.1163783213253178,
      "grad_norm": 0.49288153648376465,
      "learning_rate": 9.990937175182367e-06,
      "loss": 0.4311,
      "step": 2536
    },
    {
      "epoch": 0.11642421183057225,
      "grad_norm": 0.5010160803794861,
      "learning_rate": 9.990922413341172e-06,
      "loss": 0.4391,
      "step": 2537
    },
    {
      "epoch": 0.11647010233582672,
      "grad_norm": 0.5057856440544128,
      "learning_rate": 9.990907639498381e-06,
      "loss": 0.3806,
      "step": 2538
    },
    {
      "epoch": 0.11651599284108118,
      "grad_norm": 0.5091994404792786,
      "learning_rate": 9.990892853654035e-06,
      "loss": 0.3987,
      "step": 2539
    },
    {
      "epoch": 0.11656188334633565,
      "grad_norm": 0.49505552649497986,
      "learning_rate": 9.990878055808166e-06,
      "loss": 0.4096,
      "step": 2540
    },
    {
      "epoch": 0.1166077738515901,
      "grad_norm": 0.5190210342407227,
      "learning_rate": 9.990863245960809e-06,
      "loss": 0.4279,
      "step": 2541
    },
    {
      "epoch": 0.11665366435684457,
      "grad_norm": 0.5601308345794678,
      "learning_rate": 9.990848424112003e-06,
      "loss": 0.4448,
      "step": 2542
    },
    {
      "epoch": 0.11669955486209903,
      "grad_norm": 0.4967178404331207,
      "learning_rate": 9.99083359026178e-06,
      "loss": 0.3949,
      "step": 2543
    },
    {
      "epoch": 0.1167454453673535,
      "grad_norm": 0.4323314428329468,
      "learning_rate": 9.990818744410178e-06,
      "loss": 0.3785,
      "step": 2544
    },
    {
      "epoch": 0.11679133587260795,
      "grad_norm": 0.47798097133636475,
      "learning_rate": 9.990803886557232e-06,
      "loss": 0.4007,
      "step": 2545
    },
    {
      "epoch": 0.11683722637786242,
      "grad_norm": 0.43997934460639954,
      "learning_rate": 9.990789016702979e-06,
      "loss": 0.3193,
      "step": 2546
    },
    {
      "epoch": 0.11688311688311688,
      "grad_norm": 0.44810405373573303,
      "learning_rate": 9.990774134847452e-06,
      "loss": 0.3171,
      "step": 2547
    },
    {
      "epoch": 0.11692900738837135,
      "grad_norm": 0.5047373175621033,
      "learning_rate": 9.99075924099069e-06,
      "loss": 0.4914,
      "step": 2548
    },
    {
      "epoch": 0.1169748978936258,
      "grad_norm": 0.4475233852863312,
      "learning_rate": 9.990744335132725e-06,
      "loss": 0.3559,
      "step": 2549
    },
    {
      "epoch": 0.11702078839888028,
      "grad_norm": 0.4465727210044861,
      "learning_rate": 9.990729417273597e-06,
      "loss": 0.3767,
      "step": 2550
    },
    {
      "epoch": 0.11706667890413473,
      "grad_norm": 0.4620459973812103,
      "learning_rate": 9.990714487413338e-06,
      "loss": 0.3695,
      "step": 2551
    },
    {
      "epoch": 0.1171125694093892,
      "grad_norm": 0.5415677428245544,
      "learning_rate": 9.990699545551988e-06,
      "loss": 0.505,
      "step": 2552
    },
    {
      "epoch": 0.11715845991464366,
      "grad_norm": 0.4978041648864746,
      "learning_rate": 9.990684591689578e-06,
      "loss": 0.4121,
      "step": 2553
    },
    {
      "epoch": 0.11720435041989813,
      "grad_norm": 0.4712114632129669,
      "learning_rate": 9.990669625826149e-06,
      "loss": 0.392,
      "step": 2554
    },
    {
      "epoch": 0.11725024092515259,
      "grad_norm": 0.5263304114341736,
      "learning_rate": 9.990654647961732e-06,
      "loss": 0.4467,
      "step": 2555
    },
    {
      "epoch": 0.11729613143040706,
      "grad_norm": 0.46179741621017456,
      "learning_rate": 9.990639658096369e-06,
      "loss": 0.3736,
      "step": 2556
    },
    {
      "epoch": 0.11734202193566151,
      "grad_norm": 0.46357646584510803,
      "learning_rate": 9.99062465623009e-06,
      "loss": 0.3265,
      "step": 2557
    },
    {
      "epoch": 0.11738791244091598,
      "grad_norm": 0.5010532140731812,
      "learning_rate": 9.990609642362936e-06,
      "loss": 0.4042,
      "step": 2558
    },
    {
      "epoch": 0.11743380294617044,
      "grad_norm": 0.44892188906669617,
      "learning_rate": 9.99059461649494e-06,
      "loss": 0.3619,
      "step": 2559
    },
    {
      "epoch": 0.1174796934514249,
      "grad_norm": 0.5023994445800781,
      "learning_rate": 9.990579578626138e-06,
      "loss": 0.4238,
      "step": 2560
    },
    {
      "epoch": 0.11752558395667936,
      "grad_norm": 0.47415891289711,
      "learning_rate": 9.990564528756567e-06,
      "loss": 0.419,
      "step": 2561
    },
    {
      "epoch": 0.11757147446193382,
      "grad_norm": 0.4921308755874634,
      "learning_rate": 9.990549466886266e-06,
      "loss": 0.4009,
      "step": 2562
    },
    {
      "epoch": 0.11761736496718829,
      "grad_norm": 0.4659929871559143,
      "learning_rate": 9.990534393015266e-06,
      "loss": 0.3781,
      "step": 2563
    },
    {
      "epoch": 0.11766325547244275,
      "grad_norm": 0.4787804186344147,
      "learning_rate": 9.990519307143607e-06,
      "loss": 0.4515,
      "step": 2564
    },
    {
      "epoch": 0.11770914597769722,
      "grad_norm": 0.4983351230621338,
      "learning_rate": 9.990504209271324e-06,
      "loss": 0.4243,
      "step": 2565
    },
    {
      "epoch": 0.11775503648295167,
      "grad_norm": 0.4443513751029968,
      "learning_rate": 9.990489099398454e-06,
      "loss": 0.3249,
      "step": 2566
    },
    {
      "epoch": 0.11780092698820614,
      "grad_norm": 0.532058835029602,
      "learning_rate": 9.990473977525032e-06,
      "loss": 0.4121,
      "step": 2567
    },
    {
      "epoch": 0.1178468174934606,
      "grad_norm": 0.5136082768440247,
      "learning_rate": 9.990458843651096e-06,
      "loss": 0.4726,
      "step": 2568
    },
    {
      "epoch": 0.11789270799871507,
      "grad_norm": 0.4879734516143799,
      "learning_rate": 9.99044369777668e-06,
      "loss": 0.4131,
      "step": 2569
    },
    {
      "epoch": 0.11793859850396952,
      "grad_norm": 0.5160858035087585,
      "learning_rate": 9.990428539901821e-06,
      "loss": 0.4291,
      "step": 2570
    },
    {
      "epoch": 0.117984489009224,
      "grad_norm": 0.49275532364845276,
      "learning_rate": 9.990413370026559e-06,
      "loss": 0.3801,
      "step": 2571
    },
    {
      "epoch": 0.11803037951447845,
      "grad_norm": 0.4487593173980713,
      "learning_rate": 9.990398188150927e-06,
      "loss": 0.3677,
      "step": 2572
    },
    {
      "epoch": 0.11807627001973292,
      "grad_norm": 0.453101247549057,
      "learning_rate": 9.990382994274964e-06,
      "loss": 0.3956,
      "step": 2573
    },
    {
      "epoch": 0.11812216052498738,
      "grad_norm": 0.44491493701934814,
      "learning_rate": 9.990367788398702e-06,
      "loss": 0.3494,
      "step": 2574
    },
    {
      "epoch": 0.11816805103024185,
      "grad_norm": 0.4446192681789398,
      "learning_rate": 9.990352570522181e-06,
      "loss": 0.3465,
      "step": 2575
    },
    {
      "epoch": 0.1182139415354963,
      "grad_norm": 0.4731144905090332,
      "learning_rate": 9.990337340645438e-06,
      "loss": 0.3828,
      "step": 2576
    },
    {
      "epoch": 0.11825983204075077,
      "grad_norm": 0.5044993758201599,
      "learning_rate": 9.990322098768509e-06,
      "loss": 0.4299,
      "step": 2577
    },
    {
      "epoch": 0.11830572254600523,
      "grad_norm": 0.48940330743789673,
      "learning_rate": 9.990306844891429e-06,
      "loss": 0.3868,
      "step": 2578
    },
    {
      "epoch": 0.1183516130512597,
      "grad_norm": 0.4931536018848419,
      "learning_rate": 9.990291579014237e-06,
      "loss": 0.4411,
      "step": 2579
    },
    {
      "epoch": 0.11839750355651416,
      "grad_norm": 0.48019281029701233,
      "learning_rate": 9.99027630113697e-06,
      "loss": 0.3771,
      "step": 2580
    },
    {
      "epoch": 0.11844339406176863,
      "grad_norm": 0.4962136447429657,
      "learning_rate": 9.990261011259661e-06,
      "loss": 0.406,
      "step": 2581
    },
    {
      "epoch": 0.11848928456702308,
      "grad_norm": 0.47679248452186584,
      "learning_rate": 9.990245709382351e-06,
      "loss": 0.4009,
      "step": 2582
    },
    {
      "epoch": 0.11853517507227755,
      "grad_norm": 0.4609931409358978,
      "learning_rate": 9.990230395505074e-06,
      "loss": 0.3559,
      "step": 2583
    },
    {
      "epoch": 0.11858106557753201,
      "grad_norm": 0.4899750053882599,
      "learning_rate": 9.990215069627869e-06,
      "loss": 0.4558,
      "step": 2584
    },
    {
      "epoch": 0.11862695608278648,
      "grad_norm": 0.5094980001449585,
      "learning_rate": 9.990199731750769e-06,
      "loss": 0.5164,
      "step": 2585
    },
    {
      "epoch": 0.11867284658804093,
      "grad_norm": 0.48061394691467285,
      "learning_rate": 9.990184381873816e-06,
      "loss": 0.3473,
      "step": 2586
    },
    {
      "epoch": 0.1187187370932954,
      "grad_norm": 0.44875308871269226,
      "learning_rate": 9.990169019997045e-06,
      "loss": 0.3931,
      "step": 2587
    },
    {
      "epoch": 0.11876462759854986,
      "grad_norm": 0.4578744173049927,
      "learning_rate": 9.99015364612049e-06,
      "loss": 0.4115,
      "step": 2588
    },
    {
      "epoch": 0.11881051810380432,
      "grad_norm": 0.4992080628871918,
      "learning_rate": 9.990138260244193e-06,
      "loss": 0.4382,
      "step": 2589
    },
    {
      "epoch": 0.11885640860905879,
      "grad_norm": 0.4635515809059143,
      "learning_rate": 9.990122862368187e-06,
      "loss": 0.3502,
      "step": 2590
    },
    {
      "epoch": 0.11890229911431324,
      "grad_norm": 0.45318344235420227,
      "learning_rate": 9.99010745249251e-06,
      "loss": 0.3755,
      "step": 2591
    },
    {
      "epoch": 0.11894818961956771,
      "grad_norm": 0.4907990097999573,
      "learning_rate": 9.990092030617201e-06,
      "loss": 0.4112,
      "step": 2592
    },
    {
      "epoch": 0.11899408012482217,
      "grad_norm": 0.527769923210144,
      "learning_rate": 9.990076596742294e-06,
      "loss": 0.5634,
      "step": 2593
    },
    {
      "epoch": 0.11903997063007664,
      "grad_norm": 0.40479812026023865,
      "learning_rate": 9.99006115086783e-06,
      "loss": 0.309,
      "step": 2594
    },
    {
      "epoch": 0.1190858611353311,
      "grad_norm": 0.4663656949996948,
      "learning_rate": 9.990045692993842e-06,
      "loss": 0.4433,
      "step": 2595
    },
    {
      "epoch": 0.11913175164058556,
      "grad_norm": 0.46576741337776184,
      "learning_rate": 9.990030223120369e-06,
      "loss": 0.3417,
      "step": 2596
    },
    {
      "epoch": 0.11917764214584002,
      "grad_norm": 0.49073708057403564,
      "learning_rate": 9.990014741247449e-06,
      "loss": 0.4298,
      "step": 2597
    },
    {
      "epoch": 0.11922353265109449,
      "grad_norm": 0.49052125215530396,
      "learning_rate": 9.989999247375117e-06,
      "loss": 0.4487,
      "step": 2598
    },
    {
      "epoch": 0.11926942315634895,
      "grad_norm": 0.4916822910308838,
      "learning_rate": 9.989983741503412e-06,
      "loss": 0.4126,
      "step": 2599
    },
    {
      "epoch": 0.11931531366160342,
      "grad_norm": 0.46278244256973267,
      "learning_rate": 9.989968223632373e-06,
      "loss": 0.3623,
      "step": 2600
    },
    {
      "epoch": 0.11936120416685787,
      "grad_norm": 0.509976863861084,
      "learning_rate": 9.989952693762033e-06,
      "loss": 0.4513,
      "step": 2601
    },
    {
      "epoch": 0.11940709467211234,
      "grad_norm": 0.5073480010032654,
      "learning_rate": 9.989937151892434e-06,
      "loss": 0.477,
      "step": 2602
    },
    {
      "epoch": 0.1194529851773668,
      "grad_norm": 0.5486426949501038,
      "learning_rate": 9.989921598023609e-06,
      "loss": 0.4428,
      "step": 2603
    },
    {
      "epoch": 0.11949887568262127,
      "grad_norm": 0.4678741991519928,
      "learning_rate": 9.989906032155599e-06,
      "loss": 0.4013,
      "step": 2604
    },
    {
      "epoch": 0.11954476618787573,
      "grad_norm": 0.5263512134552002,
      "learning_rate": 9.989890454288439e-06,
      "loss": 0.5139,
      "step": 2605
    },
    {
      "epoch": 0.1195906566931302,
      "grad_norm": 0.4683694839477539,
      "learning_rate": 9.989874864422167e-06,
      "loss": 0.4018,
      "step": 2606
    },
    {
      "epoch": 0.11963654719838465,
      "grad_norm": 0.45887887477874756,
      "learning_rate": 9.98985926255682e-06,
      "loss": 0.4354,
      "step": 2607
    },
    {
      "epoch": 0.11968243770363912,
      "grad_norm": 0.6490362882614136,
      "learning_rate": 9.989843648692437e-06,
      "loss": 0.5797,
      "step": 2608
    },
    {
      "epoch": 0.11972832820889358,
      "grad_norm": 0.4808391034603119,
      "learning_rate": 9.989828022829056e-06,
      "loss": 0.3606,
      "step": 2609
    },
    {
      "epoch": 0.11977421871414805,
      "grad_norm": 0.46937552094459534,
      "learning_rate": 9.989812384966713e-06,
      "loss": 0.4311,
      "step": 2610
    },
    {
      "epoch": 0.1198201092194025,
      "grad_norm": 0.5063654184341431,
      "learning_rate": 9.989796735105446e-06,
      "loss": 0.4181,
      "step": 2611
    },
    {
      "epoch": 0.11986599972465697,
      "grad_norm": 0.4380093216896057,
      "learning_rate": 9.989781073245292e-06,
      "loss": 0.3394,
      "step": 2612
    },
    {
      "epoch": 0.11991189022991143,
      "grad_norm": 0.4629599452018738,
      "learning_rate": 9.98976539938629e-06,
      "loss": 0.3732,
      "step": 2613
    },
    {
      "epoch": 0.1199577807351659,
      "grad_norm": 0.45319634675979614,
      "learning_rate": 9.989749713528475e-06,
      "loss": 0.3796,
      "step": 2614
    },
    {
      "epoch": 0.12000367124042036,
      "grad_norm": 0.47024860978126526,
      "learning_rate": 9.98973401567189e-06,
      "loss": 0.3873,
      "step": 2615
    },
    {
      "epoch": 0.12004956174567483,
      "grad_norm": 0.47099870443344116,
      "learning_rate": 9.989718305816568e-06,
      "loss": 0.3683,
      "step": 2616
    },
    {
      "epoch": 0.12009545225092928,
      "grad_norm": 0.5088024139404297,
      "learning_rate": 9.989702583962548e-06,
      "loss": 0.4209,
      "step": 2617
    },
    {
      "epoch": 0.12014134275618374,
      "grad_norm": 0.44578009843826294,
      "learning_rate": 9.98968685010987e-06,
      "loss": 0.3672,
      "step": 2618
    },
    {
      "epoch": 0.12018723326143821,
      "grad_norm": 0.48774299025535583,
      "learning_rate": 9.989671104258567e-06,
      "loss": 0.4517,
      "step": 2619
    },
    {
      "epoch": 0.12023312376669266,
      "grad_norm": 0.4851672649383545,
      "learning_rate": 9.989655346408681e-06,
      "loss": 0.4784,
      "step": 2620
    },
    {
      "epoch": 0.12027901427194714,
      "grad_norm": 0.46987080574035645,
      "learning_rate": 9.989639576560248e-06,
      "loss": 0.3549,
      "step": 2621
    },
    {
      "epoch": 0.12032490477720159,
      "grad_norm": 0.4889931380748749,
      "learning_rate": 9.989623794713307e-06,
      "loss": 0.4615,
      "step": 2622
    },
    {
      "epoch": 0.12037079528245606,
      "grad_norm": 0.4562816023826599,
      "learning_rate": 9.989608000867896e-06,
      "loss": 0.3668,
      "step": 2623
    },
    {
      "epoch": 0.12041668578771052,
      "grad_norm": 0.5622779726982117,
      "learning_rate": 9.989592195024054e-06,
      "loss": 0.5258,
      "step": 2624
    },
    {
      "epoch": 0.12046257629296499,
      "grad_norm": 0.4577435255050659,
      "learning_rate": 9.989576377181816e-06,
      "loss": 0.3997,
      "step": 2625
    },
    {
      "epoch": 0.12050846679821944,
      "grad_norm": 0.47942185401916504,
      "learning_rate": 9.98956054734122e-06,
      "loss": 0.4302,
      "step": 2626
    },
    {
      "epoch": 0.12055435730347391,
      "grad_norm": 0.4641333222389221,
      "learning_rate": 9.989544705502307e-06,
      "loss": 0.39,
      "step": 2627
    },
    {
      "epoch": 0.12060024780872837,
      "grad_norm": 0.4787479639053345,
      "learning_rate": 9.989528851665114e-06,
      "loss": 0.3977,
      "step": 2628
    },
    {
      "epoch": 0.12064613831398284,
      "grad_norm": 0.47346392273902893,
      "learning_rate": 9.98951298582968e-06,
      "loss": 0.4281,
      "step": 2629
    },
    {
      "epoch": 0.1206920288192373,
      "grad_norm": 0.4877626299858093,
      "learning_rate": 9.98949710799604e-06,
      "loss": 0.4898,
      "step": 2630
    },
    {
      "epoch": 0.12073791932449177,
      "grad_norm": 0.4503246247768402,
      "learning_rate": 9.989481218164235e-06,
      "loss": 0.3747,
      "step": 2631
    },
    {
      "epoch": 0.12078380982974622,
      "grad_norm": 0.4832894206047058,
      "learning_rate": 9.989465316334303e-06,
      "loss": 0.4471,
      "step": 2632
    },
    {
      "epoch": 0.12082970033500069,
      "grad_norm": 0.465067595243454,
      "learning_rate": 9.98944940250628e-06,
      "loss": 0.3676,
      "step": 2633
    },
    {
      "epoch": 0.12087559084025515,
      "grad_norm": 0.43929052352905273,
      "learning_rate": 9.989433476680207e-06,
      "loss": 0.3595,
      "step": 2634
    },
    {
      "epoch": 0.12092148134550962,
      "grad_norm": 0.5205087661743164,
      "learning_rate": 9.989417538856121e-06,
      "loss": 0.478,
      "step": 2635
    },
    {
      "epoch": 0.12096737185076407,
      "grad_norm": 0.496156245470047,
      "learning_rate": 9.98940158903406e-06,
      "loss": 0.4612,
      "step": 2636
    },
    {
      "epoch": 0.12101326235601854,
      "grad_norm": 0.47536590695381165,
      "learning_rate": 9.989385627214064e-06,
      "loss": 0.3773,
      "step": 2637
    },
    {
      "epoch": 0.121059152861273,
      "grad_norm": 0.4678027629852295,
      "learning_rate": 9.98936965339617e-06,
      "loss": 0.423,
      "step": 2638
    },
    {
      "epoch": 0.12110504336652747,
      "grad_norm": 0.5094507932662964,
      "learning_rate": 9.989353667580418e-06,
      "loss": 0.4631,
      "step": 2639
    },
    {
      "epoch": 0.12115093387178193,
      "grad_norm": 0.4905732572078705,
      "learning_rate": 9.989337669766841e-06,
      "loss": 0.3958,
      "step": 2640
    },
    {
      "epoch": 0.1211968243770364,
      "grad_norm": 0.45689842104911804,
      "learning_rate": 9.989321659955484e-06,
      "loss": 0.3389,
      "step": 2641
    },
    {
      "epoch": 0.12124271488229085,
      "grad_norm": 0.46184709668159485,
      "learning_rate": 9.989305638146384e-06,
      "loss": 0.4432,
      "step": 2642
    },
    {
      "epoch": 0.12128860538754532,
      "grad_norm": 0.48346012830734253,
      "learning_rate": 9.989289604339577e-06,
      "loss": 0.4496,
      "step": 2643
    },
    {
      "epoch": 0.12133449589279978,
      "grad_norm": 0.4688853919506073,
      "learning_rate": 9.989273558535105e-06,
      "loss": 0.4094,
      "step": 2644
    },
    {
      "epoch": 0.12138038639805425,
      "grad_norm": 0.49808549880981445,
      "learning_rate": 9.989257500733003e-06,
      "loss": 0.4532,
      "step": 2645
    },
    {
      "epoch": 0.1214262769033087,
      "grad_norm": 0.43277788162231445,
      "learning_rate": 9.989241430933312e-06,
      "loss": 0.3305,
      "step": 2646
    },
    {
      "epoch": 0.12147216740856316,
      "grad_norm": 0.4733690619468689,
      "learning_rate": 9.98922534913607e-06,
      "loss": 0.4502,
      "step": 2647
    },
    {
      "epoch": 0.12151805791381763,
      "grad_norm": 0.4406627118587494,
      "learning_rate": 9.989209255341316e-06,
      "loss": 0.3209,
      "step": 2648
    },
    {
      "epoch": 0.12156394841907209,
      "grad_norm": 0.4449348747730255,
      "learning_rate": 9.989193149549088e-06,
      "loss": 0.3765,
      "step": 2649
    },
    {
      "epoch": 0.12160983892432656,
      "grad_norm": 0.4798898994922638,
      "learning_rate": 9.989177031759425e-06,
      "loss": 0.4042,
      "step": 2650
    },
    {
      "epoch": 0.12165572942958101,
      "grad_norm": 0.4864308834075928,
      "learning_rate": 9.989160901972366e-06,
      "loss": 0.3935,
      "step": 2651
    },
    {
      "epoch": 0.12170161993483548,
      "grad_norm": 0.46155428886413574,
      "learning_rate": 9.98914476018795e-06,
      "loss": 0.2942,
      "step": 2652
    },
    {
      "epoch": 0.12174751044008994,
      "grad_norm": 0.4384247958660126,
      "learning_rate": 9.989128606406215e-06,
      "loss": 0.3186,
      "step": 2653
    },
    {
      "epoch": 0.12179340094534441,
      "grad_norm": 0.45776841044425964,
      "learning_rate": 9.989112440627202e-06,
      "loss": 0.3595,
      "step": 2654
    },
    {
      "epoch": 0.12183929145059887,
      "grad_norm": 0.5123647451400757,
      "learning_rate": 9.989096262850945e-06,
      "loss": 0.4416,
      "step": 2655
    },
    {
      "epoch": 0.12188518195585334,
      "grad_norm": 0.42523425817489624,
      "learning_rate": 9.989080073077489e-06,
      "loss": 0.2779,
      "step": 2656
    },
    {
      "epoch": 0.12193107246110779,
      "grad_norm": 0.46643584966659546,
      "learning_rate": 9.98906387130687e-06,
      "loss": 0.3834,
      "step": 2657
    },
    {
      "epoch": 0.12197696296636226,
      "grad_norm": 0.44290754199028015,
      "learning_rate": 9.989047657539125e-06,
      "loss": 0.3423,
      "step": 2658
    },
    {
      "epoch": 0.12202285347161672,
      "grad_norm": 0.4556252360343933,
      "learning_rate": 9.989031431774295e-06,
      "loss": 0.3335,
      "step": 2659
    },
    {
      "epoch": 0.12206874397687119,
      "grad_norm": 0.5164615511894226,
      "learning_rate": 9.98901519401242e-06,
      "loss": 0.4986,
      "step": 2660
    },
    {
      "epoch": 0.12211463448212564,
      "grad_norm": 0.4679234027862549,
      "learning_rate": 9.988998944253539e-06,
      "loss": 0.4101,
      "step": 2661
    },
    {
      "epoch": 0.12216052498738011,
      "grad_norm": 0.522869348526001,
      "learning_rate": 9.988982682497688e-06,
      "loss": 0.4316,
      "step": 2662
    },
    {
      "epoch": 0.12220641549263457,
      "grad_norm": 0.48769786953926086,
      "learning_rate": 9.98896640874491e-06,
      "loss": 0.4261,
      "step": 2663
    },
    {
      "epoch": 0.12225230599788904,
      "grad_norm": 0.4540833532810211,
      "learning_rate": 9.98895012299524e-06,
      "loss": 0.4039,
      "step": 2664
    },
    {
      "epoch": 0.1222981965031435,
      "grad_norm": 0.4306601881980896,
      "learning_rate": 9.98893382524872e-06,
      "loss": 0.335,
      "step": 2665
    },
    {
      "epoch": 0.12234408700839797,
      "grad_norm": 0.534954309463501,
      "learning_rate": 9.988917515505392e-06,
      "loss": 0.4612,
      "step": 2666
    },
    {
      "epoch": 0.12238997751365242,
      "grad_norm": 0.47996991872787476,
      "learning_rate": 9.988901193765288e-06,
      "loss": 0.4145,
      "step": 2667
    },
    {
      "epoch": 0.1224358680189069,
      "grad_norm": 0.4783177077770233,
      "learning_rate": 9.988884860028453e-06,
      "loss": 0.3981,
      "step": 2668
    },
    {
      "epoch": 0.12248175852416135,
      "grad_norm": 0.4888615906238556,
      "learning_rate": 9.988868514294925e-06,
      "loss": 0.4878,
      "step": 2669
    },
    {
      "epoch": 0.12252764902941582,
      "grad_norm": 0.5233306884765625,
      "learning_rate": 9.98885215656474e-06,
      "loss": 0.4703,
      "step": 2670
    },
    {
      "epoch": 0.12257353953467028,
      "grad_norm": 0.47517794370651245,
      "learning_rate": 9.988835786837945e-06,
      "loss": 0.3881,
      "step": 2671
    },
    {
      "epoch": 0.12261943003992475,
      "grad_norm": 0.4643119275569916,
      "learning_rate": 9.98881940511457e-06,
      "loss": 0.3692,
      "step": 2672
    },
    {
      "epoch": 0.1226653205451792,
      "grad_norm": 0.4417555630207062,
      "learning_rate": 9.988803011394661e-06,
      "loss": 0.3455,
      "step": 2673
    },
    {
      "epoch": 0.12271121105043367,
      "grad_norm": 0.47155094146728516,
      "learning_rate": 9.988786605678254e-06,
      "loss": 0.3854,
      "step": 2674
    },
    {
      "epoch": 0.12275710155568813,
      "grad_norm": 0.4446570873260498,
      "learning_rate": 9.988770187965391e-06,
      "loss": 0.369,
      "step": 2675
    },
    {
      "epoch": 0.1228029920609426,
      "grad_norm": 0.473347932100296,
      "learning_rate": 9.988753758256108e-06,
      "loss": 0.4111,
      "step": 2676
    },
    {
      "epoch": 0.12284888256619705,
      "grad_norm": 0.555674135684967,
      "learning_rate": 9.98873731655045e-06,
      "loss": 0.5086,
      "step": 2677
    },
    {
      "epoch": 0.12289477307145151,
      "grad_norm": 0.6599460244178772,
      "learning_rate": 9.988720862848451e-06,
      "loss": 0.4837,
      "step": 2678
    },
    {
      "epoch": 0.12294066357670598,
      "grad_norm": 0.5217191576957703,
      "learning_rate": 9.988704397150155e-06,
      "loss": 0.4923,
      "step": 2679
    },
    {
      "epoch": 0.12298655408196044,
      "grad_norm": 0.5135862827301025,
      "learning_rate": 9.988687919455597e-06,
      "loss": 0.4923,
      "step": 2680
    },
    {
      "epoch": 0.1230324445872149,
      "grad_norm": 0.5184143781661987,
      "learning_rate": 9.988671429764821e-06,
      "loss": 0.4831,
      "step": 2681
    },
    {
      "epoch": 0.12307833509246936,
      "grad_norm": 0.5583066940307617,
      "learning_rate": 9.988654928077862e-06,
      "loss": 0.4151,
      "step": 2682
    },
    {
      "epoch": 0.12312422559772383,
      "grad_norm": 0.4416000247001648,
      "learning_rate": 9.988638414394764e-06,
      "loss": 0.371,
      "step": 2683
    },
    {
      "epoch": 0.12317011610297829,
      "grad_norm": 0.5057492256164551,
      "learning_rate": 9.988621888715564e-06,
      "loss": 0.4177,
      "step": 2684
    },
    {
      "epoch": 0.12321600660823276,
      "grad_norm": 0.5240790247917175,
      "learning_rate": 9.988605351040304e-06,
      "loss": 0.4063,
      "step": 2685
    },
    {
      "epoch": 0.12326189711348721,
      "grad_norm": 0.4482614994049072,
      "learning_rate": 9.988588801369023e-06,
      "loss": 0.3951,
      "step": 2686
    },
    {
      "epoch": 0.12330778761874168,
      "grad_norm": 0.427959144115448,
      "learning_rate": 9.988572239701758e-06,
      "loss": 0.3258,
      "step": 2687
    },
    {
      "epoch": 0.12335367812399614,
      "grad_norm": 0.5692232847213745,
      "learning_rate": 9.988555666038556e-06,
      "loss": 0.5131,
      "step": 2688
    },
    {
      "epoch": 0.12339956862925061,
      "grad_norm": 0.4918581247329712,
      "learning_rate": 9.988539080379447e-06,
      "loss": 0.4351,
      "step": 2689
    },
    {
      "epoch": 0.12344545913450507,
      "grad_norm": 0.45957711338996887,
      "learning_rate": 9.98852248272448e-06,
      "loss": 0.4242,
      "step": 2690
    },
    {
      "epoch": 0.12349134963975954,
      "grad_norm": 0.480080783367157,
      "learning_rate": 9.988505873073686e-06,
      "loss": 0.3804,
      "step": 2691
    },
    {
      "epoch": 0.123537240145014,
      "grad_norm": 0.4867088198661804,
      "learning_rate": 9.988489251427113e-06,
      "loss": 0.4472,
      "step": 2692
    },
    {
      "epoch": 0.12358313065026846,
      "grad_norm": 0.49736061692237854,
      "learning_rate": 9.988472617784797e-06,
      "loss": 0.4333,
      "step": 2693
    },
    {
      "epoch": 0.12362902115552292,
      "grad_norm": 0.4782455265522003,
      "learning_rate": 9.98845597214678e-06,
      "loss": 0.37,
      "step": 2694
    },
    {
      "epoch": 0.12367491166077739,
      "grad_norm": 0.4965982437133789,
      "learning_rate": 9.9884393145131e-06,
      "loss": 0.4735,
      "step": 2695
    },
    {
      "epoch": 0.12372080216603185,
      "grad_norm": 0.4879864454269409,
      "learning_rate": 9.988422644883797e-06,
      "loss": 0.433,
      "step": 2696
    },
    {
      "epoch": 0.12376669267128632,
      "grad_norm": 0.4590333104133606,
      "learning_rate": 9.988405963258913e-06,
      "loss": 0.4145,
      "step": 2697
    },
    {
      "epoch": 0.12381258317654077,
      "grad_norm": 0.4210829436779022,
      "learning_rate": 9.988389269638487e-06,
      "loss": 0.2989,
      "step": 2698
    },
    {
      "epoch": 0.12385847368179524,
      "grad_norm": 0.4543023407459259,
      "learning_rate": 9.988372564022559e-06,
      "loss": 0.3524,
      "step": 2699
    },
    {
      "epoch": 0.1239043641870497,
      "grad_norm": 0.7114877104759216,
      "learning_rate": 9.988355846411169e-06,
      "loss": 0.5302,
      "step": 2700
    },
    {
      "epoch": 0.12395025469230417,
      "grad_norm": 0.4773704707622528,
      "learning_rate": 9.988339116804358e-06,
      "loss": 0.3911,
      "step": 2701
    },
    {
      "epoch": 0.12399614519755862,
      "grad_norm": 0.488718718290329,
      "learning_rate": 9.988322375202166e-06,
      "loss": 0.4934,
      "step": 2702
    },
    {
      "epoch": 0.1240420357028131,
      "grad_norm": 0.47661206126213074,
      "learning_rate": 9.988305621604631e-06,
      "loss": 0.4276,
      "step": 2703
    },
    {
      "epoch": 0.12408792620806755,
      "grad_norm": 0.48617932200431824,
      "learning_rate": 9.988288856011797e-06,
      "loss": 0.3923,
      "step": 2704
    },
    {
      "epoch": 0.12413381671332202,
      "grad_norm": 0.5129133462905884,
      "learning_rate": 9.988272078423703e-06,
      "loss": 0.4549,
      "step": 2705
    },
    {
      "epoch": 0.12417970721857648,
      "grad_norm": 0.484331876039505,
      "learning_rate": 9.988255288840389e-06,
      "loss": 0.4466,
      "step": 2706
    },
    {
      "epoch": 0.12422559772383093,
      "grad_norm": 0.464389443397522,
      "learning_rate": 9.988238487261894e-06,
      "loss": 0.3958,
      "step": 2707
    },
    {
      "epoch": 0.1242714882290854,
      "grad_norm": 0.5085384249687195,
      "learning_rate": 9.98822167368826e-06,
      "loss": 0.5681,
      "step": 2708
    },
    {
      "epoch": 0.12431737873433986,
      "grad_norm": 0.493242084980011,
      "learning_rate": 9.98820484811953e-06,
      "loss": 0.4812,
      "step": 2709
    },
    {
      "epoch": 0.12436326923959433,
      "grad_norm": 0.48836490511894226,
      "learning_rate": 9.988188010555739e-06,
      "loss": 0.4301,
      "step": 2710
    },
    {
      "epoch": 0.12440915974484879,
      "grad_norm": 0.41648826003074646,
      "learning_rate": 9.98817116099693e-06,
      "loss": 0.3083,
      "step": 2711
    },
    {
      "epoch": 0.12445505025010326,
      "grad_norm": 0.4782598614692688,
      "learning_rate": 9.988154299443145e-06,
      "loss": 0.4008,
      "step": 2712
    },
    {
      "epoch": 0.12450094075535771,
      "grad_norm": 0.4725387394428253,
      "learning_rate": 9.988137425894422e-06,
      "loss": 0.4191,
      "step": 2713
    },
    {
      "epoch": 0.12454683126061218,
      "grad_norm": 0.5006836652755737,
      "learning_rate": 9.988120540350804e-06,
      "loss": 0.4333,
      "step": 2714
    },
    {
      "epoch": 0.12459272176586664,
      "grad_norm": 0.4685843288898468,
      "learning_rate": 9.98810364281233e-06,
      "loss": 0.4105,
      "step": 2715
    },
    {
      "epoch": 0.12463861227112111,
      "grad_norm": 0.4412849545478821,
      "learning_rate": 9.988086733279042e-06,
      "loss": 0.3268,
      "step": 2716
    },
    {
      "epoch": 0.12468450277637556,
      "grad_norm": 0.579803466796875,
      "learning_rate": 9.988069811750978e-06,
      "loss": 0.564,
      "step": 2717
    },
    {
      "epoch": 0.12473039328163003,
      "grad_norm": 0.49471402168273926,
      "learning_rate": 9.98805287822818e-06,
      "loss": 0.4621,
      "step": 2718
    },
    {
      "epoch": 0.12477628378688449,
      "grad_norm": 0.4630710482597351,
      "learning_rate": 9.98803593271069e-06,
      "loss": 0.3895,
      "step": 2719
    },
    {
      "epoch": 0.12482217429213896,
      "grad_norm": 0.47090473771095276,
      "learning_rate": 9.988018975198549e-06,
      "loss": 0.3714,
      "step": 2720
    },
    {
      "epoch": 0.12486806479739342,
      "grad_norm": 0.46573033928871155,
      "learning_rate": 9.988002005691794e-06,
      "loss": 0.3511,
      "step": 2721
    },
    {
      "epoch": 0.12491395530264789,
      "grad_norm": 0.5408806800842285,
      "learning_rate": 9.98798502419047e-06,
      "loss": 0.4806,
      "step": 2722
    },
    {
      "epoch": 0.12495984580790234,
      "grad_norm": 0.44541439414024353,
      "learning_rate": 9.987968030694616e-06,
      "loss": 0.3738,
      "step": 2723
    },
    {
      "epoch": 0.1250057363131568,
      "grad_norm": 0.46675705909729004,
      "learning_rate": 9.987951025204274e-06,
      "loss": 0.4484,
      "step": 2724
    },
    {
      "epoch": 0.12505162681841128,
      "grad_norm": 0.47927436232566833,
      "learning_rate": 9.987934007719485e-06,
      "loss": 0.4382,
      "step": 2725
    },
    {
      "epoch": 0.12509751732366572,
      "grad_norm": 0.46207964420318604,
      "learning_rate": 9.987916978240286e-06,
      "loss": 0.4188,
      "step": 2726
    },
    {
      "epoch": 0.1251434078289202,
      "grad_norm": 0.4308772385120392,
      "learning_rate": 9.987899936766724e-06,
      "loss": 0.3191,
      "step": 2727
    },
    {
      "epoch": 0.12518929833417466,
      "grad_norm": 0.45977357029914856,
      "learning_rate": 9.987882883298834e-06,
      "loss": 0.3623,
      "step": 2728
    },
    {
      "epoch": 0.12523518883942913,
      "grad_norm": 0.44675368070602417,
      "learning_rate": 9.987865817836659e-06,
      "loss": 0.3188,
      "step": 2729
    },
    {
      "epoch": 0.12528107934468358,
      "grad_norm": 0.508658230304718,
      "learning_rate": 9.987848740380243e-06,
      "loss": 0.4487,
      "step": 2730
    },
    {
      "epoch": 0.12532696984993805,
      "grad_norm": 0.5319099426269531,
      "learning_rate": 9.987831650929625e-06,
      "loss": 0.4188,
      "step": 2731
    },
    {
      "epoch": 0.12537286035519252,
      "grad_norm": 0.5304564237594604,
      "learning_rate": 9.987814549484846e-06,
      "loss": 0.5167,
      "step": 2732
    },
    {
      "epoch": 0.125418750860447,
      "grad_norm": 0.5082815289497375,
      "learning_rate": 9.987797436045949e-06,
      "loss": 0.504,
      "step": 2733
    },
    {
      "epoch": 0.12546464136570143,
      "grad_norm": 0.5002908110618591,
      "learning_rate": 9.987780310612972e-06,
      "loss": 0.4142,
      "step": 2734
    },
    {
      "epoch": 0.1255105318709559,
      "grad_norm": 0.4971759021282196,
      "learning_rate": 9.987763173185956e-06,
      "loss": 0.3695,
      "step": 2735
    },
    {
      "epoch": 0.12555642237621037,
      "grad_norm": 0.48756933212280273,
      "learning_rate": 9.987746023764944e-06,
      "loss": 0.3668,
      "step": 2736
    },
    {
      "epoch": 0.1256023128814648,
      "grad_norm": 0.45884984731674194,
      "learning_rate": 9.987728862349979e-06,
      "loss": 0.3895,
      "step": 2737
    },
    {
      "epoch": 0.12564820338671928,
      "grad_norm": 0.5038509964942932,
      "learning_rate": 9.987711688941098e-06,
      "loss": 0.4321,
      "step": 2738
    },
    {
      "epoch": 0.12569409389197375,
      "grad_norm": 0.5076050162315369,
      "learning_rate": 9.987694503538347e-06,
      "loss": 0.463,
      "step": 2739
    },
    {
      "epoch": 0.12573998439722822,
      "grad_norm": 0.44172585010528564,
      "learning_rate": 9.987677306141763e-06,
      "loss": 0.3518,
      "step": 2740
    },
    {
      "epoch": 0.12578587490248266,
      "grad_norm": 0.5345978736877441,
      "learning_rate": 9.98766009675139e-06,
      "loss": 0.5107,
      "step": 2741
    },
    {
      "epoch": 0.12583176540773713,
      "grad_norm": 0.4749714136123657,
      "learning_rate": 9.987642875367269e-06,
      "loss": 0.3612,
      "step": 2742
    },
    {
      "epoch": 0.1258776559129916,
      "grad_norm": 0.5910397171974182,
      "learning_rate": 9.98762564198944e-06,
      "loss": 0.4715,
      "step": 2743
    },
    {
      "epoch": 0.12592354641824607,
      "grad_norm": 0.4772932529449463,
      "learning_rate": 9.987608396617946e-06,
      "loss": 0.3904,
      "step": 2744
    },
    {
      "epoch": 0.12596943692350052,
      "grad_norm": 0.460406631231308,
      "learning_rate": 9.987591139252828e-06,
      "loss": 0.3764,
      "step": 2745
    },
    {
      "epoch": 0.126015327428755,
      "grad_norm": 0.44393283128738403,
      "learning_rate": 9.987573869894128e-06,
      "loss": 0.3805,
      "step": 2746
    },
    {
      "epoch": 0.12606121793400946,
      "grad_norm": 0.5282798409461975,
      "learning_rate": 9.987556588541884e-06,
      "loss": 0.4113,
      "step": 2747
    },
    {
      "epoch": 0.12610710843926393,
      "grad_norm": 0.46061545610427856,
      "learning_rate": 9.987539295196144e-06,
      "loss": 0.3199,
      "step": 2748
    },
    {
      "epoch": 0.12615299894451837,
      "grad_norm": 0.48865440487861633,
      "learning_rate": 9.987521989856944e-06,
      "loss": 0.3675,
      "step": 2749
    },
    {
      "epoch": 0.12619888944977284,
      "grad_norm": 0.47560811042785645,
      "learning_rate": 9.98750467252433e-06,
      "loss": 0.4288,
      "step": 2750
    },
    {
      "epoch": 0.1262447799550273,
      "grad_norm": 0.48677363991737366,
      "learning_rate": 9.98748734319834e-06,
      "loss": 0.4112,
      "step": 2751
    },
    {
      "epoch": 0.12629067046028178,
      "grad_norm": 0.44393691420555115,
      "learning_rate": 9.987470001879016e-06,
      "loss": 0.3558,
      "step": 2752
    },
    {
      "epoch": 0.12633656096553622,
      "grad_norm": 0.5380996465682983,
      "learning_rate": 9.987452648566401e-06,
      "loss": 0.418,
      "step": 2753
    },
    {
      "epoch": 0.1263824514707907,
      "grad_norm": 0.5498951077461243,
      "learning_rate": 9.987435283260538e-06,
      "loss": 0.4486,
      "step": 2754
    },
    {
      "epoch": 0.12642834197604516,
      "grad_norm": 0.48008909821510315,
      "learning_rate": 9.987417905961467e-06,
      "loss": 0.3517,
      "step": 2755
    },
    {
      "epoch": 0.12647423248129963,
      "grad_norm": 0.5095076560974121,
      "learning_rate": 9.987400516669227e-06,
      "loss": 0.4982,
      "step": 2756
    },
    {
      "epoch": 0.12652012298655407,
      "grad_norm": 0.420227587223053,
      "learning_rate": 9.987383115383867e-06,
      "loss": 0.2704,
      "step": 2757
    },
    {
      "epoch": 0.12656601349180854,
      "grad_norm": 0.44439947605133057,
      "learning_rate": 9.987365702105423e-06,
      "loss": 0.344,
      "step": 2758
    },
    {
      "epoch": 0.126611903997063,
      "grad_norm": 0.44056349992752075,
      "learning_rate": 9.987348276833938e-06,
      "loss": 0.3255,
      "step": 2759
    },
    {
      "epoch": 0.12665779450231748,
      "grad_norm": 0.4823189675807953,
      "learning_rate": 9.987330839569456e-06,
      "loss": 0.3681,
      "step": 2760
    },
    {
      "epoch": 0.12670368500757193,
      "grad_norm": 0.4581528604030609,
      "learning_rate": 9.987313390312015e-06,
      "loss": 0.3541,
      "step": 2761
    },
    {
      "epoch": 0.1267495755128264,
      "grad_norm": 0.48997363448143005,
      "learning_rate": 9.98729592906166e-06,
      "loss": 0.3867,
      "step": 2762
    },
    {
      "epoch": 0.12679546601808087,
      "grad_norm": 0.5013061165809631,
      "learning_rate": 9.987278455818434e-06,
      "loss": 0.4401,
      "step": 2763
    },
    {
      "epoch": 0.12684135652333534,
      "grad_norm": 0.48018598556518555,
      "learning_rate": 9.987260970582376e-06,
      "loss": 0.3603,
      "step": 2764
    },
    {
      "epoch": 0.12688724702858978,
      "grad_norm": 0.5071067214012146,
      "learning_rate": 9.98724347335353e-06,
      "loss": 0.4919,
      "step": 2765
    },
    {
      "epoch": 0.12693313753384425,
      "grad_norm": 0.4889163076877594,
      "learning_rate": 9.987225964131937e-06,
      "loss": 0.4011,
      "step": 2766
    },
    {
      "epoch": 0.12697902803909872,
      "grad_norm": 0.47455519437789917,
      "learning_rate": 9.98720844291764e-06,
      "loss": 0.4128,
      "step": 2767
    },
    {
      "epoch": 0.12702491854435316,
      "grad_norm": 0.5203969478607178,
      "learning_rate": 9.98719090971068e-06,
      "loss": 0.4459,
      "step": 2768
    },
    {
      "epoch": 0.12707080904960763,
      "grad_norm": 0.4268333911895752,
      "learning_rate": 9.9871733645111e-06,
      "loss": 0.3224,
      "step": 2769
    },
    {
      "epoch": 0.1271166995548621,
      "grad_norm": 0.4883555769920349,
      "learning_rate": 9.987155807318942e-06,
      "loss": 0.3919,
      "step": 2770
    },
    {
      "epoch": 0.12716259006011657,
      "grad_norm": 0.47102999687194824,
      "learning_rate": 9.98713823813425e-06,
      "loss": 0.4466,
      "step": 2771
    },
    {
      "epoch": 0.127208480565371,
      "grad_norm": 0.5064966082572937,
      "learning_rate": 9.987120656957063e-06,
      "loss": 0.4034,
      "step": 2772
    },
    {
      "epoch": 0.12725437107062548,
      "grad_norm": 0.4775254428386688,
      "learning_rate": 9.987103063787425e-06,
      "loss": 0.4264,
      "step": 2773
    },
    {
      "epoch": 0.12730026157587995,
      "grad_norm": 0.4880734384059906,
      "learning_rate": 9.987085458625378e-06,
      "loss": 0.4213,
      "step": 2774
    },
    {
      "epoch": 0.12734615208113442,
      "grad_norm": 0.5105839967727661,
      "learning_rate": 9.987067841470964e-06,
      "loss": 0.522,
      "step": 2775
    },
    {
      "epoch": 0.12739204258638887,
      "grad_norm": 0.46990150213241577,
      "learning_rate": 9.987050212324227e-06,
      "loss": 0.4288,
      "step": 2776
    },
    {
      "epoch": 0.12743793309164334,
      "grad_norm": 0.4751972556114197,
      "learning_rate": 9.987032571185206e-06,
      "loss": 0.4182,
      "step": 2777
    },
    {
      "epoch": 0.1274838235968978,
      "grad_norm": 0.4855301082134247,
      "learning_rate": 9.987014918053947e-06,
      "loss": 0.4451,
      "step": 2778
    },
    {
      "epoch": 0.12752971410215228,
      "grad_norm": 0.4894592761993408,
      "learning_rate": 9.986997252930492e-06,
      "loss": 0.4717,
      "step": 2779
    },
    {
      "epoch": 0.12757560460740672,
      "grad_norm": 0.45551395416259766,
      "learning_rate": 9.986979575814882e-06,
      "loss": 0.3876,
      "step": 2780
    },
    {
      "epoch": 0.1276214951126612,
      "grad_norm": 0.5171364545822144,
      "learning_rate": 9.98696188670716e-06,
      "loss": 0.4956,
      "step": 2781
    },
    {
      "epoch": 0.12766738561791566,
      "grad_norm": 0.48302924633026123,
      "learning_rate": 9.986944185607368e-06,
      "loss": 0.3774,
      "step": 2782
    },
    {
      "epoch": 0.12771327612317013,
      "grad_norm": 0.5062499046325684,
      "learning_rate": 9.98692647251555e-06,
      "loss": 0.4683,
      "step": 2783
    },
    {
      "epoch": 0.12775916662842457,
      "grad_norm": 0.4605990946292877,
      "learning_rate": 9.986908747431747e-06,
      "loss": 0.3978,
      "step": 2784
    },
    {
      "epoch": 0.12780505713367904,
      "grad_norm": 0.5306394100189209,
      "learning_rate": 9.986891010356004e-06,
      "loss": 0.4384,
      "step": 2785
    },
    {
      "epoch": 0.1278509476389335,
      "grad_norm": 0.47338321805000305,
      "learning_rate": 9.986873261288358e-06,
      "loss": 0.416,
      "step": 2786
    },
    {
      "epoch": 0.12789683814418798,
      "grad_norm": 0.512321949005127,
      "learning_rate": 9.98685550022886e-06,
      "loss": 0.5109,
      "step": 2787
    },
    {
      "epoch": 0.12794272864944242,
      "grad_norm": 0.5197054743766785,
      "learning_rate": 9.986837727177547e-06,
      "loss": 0.4975,
      "step": 2788
    },
    {
      "epoch": 0.1279886191546969,
      "grad_norm": 0.45874637365341187,
      "learning_rate": 9.986819942134462e-06,
      "loss": 0.446,
      "step": 2789
    },
    {
      "epoch": 0.12803450965995136,
      "grad_norm": 0.49551525712013245,
      "learning_rate": 9.98680214509965e-06,
      "loss": 0.4219,
      "step": 2790
    },
    {
      "epoch": 0.12808040016520583,
      "grad_norm": 0.48614075779914856,
      "learning_rate": 9.986784336073153e-06,
      "loss": 0.3977,
      "step": 2791
    },
    {
      "epoch": 0.12812629067046027,
      "grad_norm": 0.5329895615577698,
      "learning_rate": 9.986766515055014e-06,
      "loss": 0.5341,
      "step": 2792
    },
    {
      "epoch": 0.12817218117571474,
      "grad_norm": 0.4470343291759491,
      "learning_rate": 9.986748682045275e-06,
      "loss": 0.3473,
      "step": 2793
    },
    {
      "epoch": 0.12821807168096921,
      "grad_norm": 0.4502967894077301,
      "learning_rate": 9.986730837043977e-06,
      "loss": 0.4066,
      "step": 2794
    },
    {
      "epoch": 0.12826396218622366,
      "grad_norm": 0.5154269933700562,
      "learning_rate": 9.986712980051167e-06,
      "loss": 0.4548,
      "step": 2795
    },
    {
      "epoch": 0.12830985269147813,
      "grad_norm": 0.5148901343345642,
      "learning_rate": 9.986695111066887e-06,
      "loss": 0.4738,
      "step": 2796
    },
    {
      "epoch": 0.1283557431967326,
      "grad_norm": 0.4601181745529175,
      "learning_rate": 9.986677230091178e-06,
      "loss": 0.3941,
      "step": 2797
    },
    {
      "epoch": 0.12840163370198707,
      "grad_norm": 0.48027870059013367,
      "learning_rate": 9.986659337124085e-06,
      "loss": 0.4309,
      "step": 2798
    },
    {
      "epoch": 0.1284475242072415,
      "grad_norm": 0.45475712418556213,
      "learning_rate": 9.98664143216565e-06,
      "loss": 0.3625,
      "step": 2799
    },
    {
      "epoch": 0.12849341471249598,
      "grad_norm": 0.42265060544013977,
      "learning_rate": 9.986623515215915e-06,
      "loss": 0.354,
      "step": 2800
    },
    {
      "epoch": 0.12853930521775045,
      "grad_norm": 0.4429236352443695,
      "learning_rate": 9.986605586274924e-06,
      "loss": 0.3569,
      "step": 2801
    },
    {
      "epoch": 0.12858519572300492,
      "grad_norm": 0.443429172039032,
      "learning_rate": 9.986587645342722e-06,
      "loss": 0.3551,
      "step": 2802
    },
    {
      "epoch": 0.12863108622825936,
      "grad_norm": 0.5227746367454529,
      "learning_rate": 9.98656969241935e-06,
      "loss": 0.488,
      "step": 2803
    },
    {
      "epoch": 0.12867697673351383,
      "grad_norm": 0.48553329706192017,
      "learning_rate": 9.986551727504851e-06,
      "loss": 0.3797,
      "step": 2804
    },
    {
      "epoch": 0.1287228672387683,
      "grad_norm": 0.48094120621681213,
      "learning_rate": 9.98653375059927e-06,
      "loss": 0.4062,
      "step": 2805
    },
    {
      "epoch": 0.12876875774402277,
      "grad_norm": 0.5202357769012451,
      "learning_rate": 9.98651576170265e-06,
      "loss": 0.4871,
      "step": 2806
    },
    {
      "epoch": 0.1288146482492772,
      "grad_norm": 0.4971047639846802,
      "learning_rate": 9.98649776081503e-06,
      "loss": 0.4294,
      "step": 2807
    },
    {
      "epoch": 0.12886053875453168,
      "grad_norm": 0.4567926228046417,
      "learning_rate": 9.98647974793646e-06,
      "loss": 0.4046,
      "step": 2808
    },
    {
      "epoch": 0.12890642925978615,
      "grad_norm": 0.45385095477104187,
      "learning_rate": 9.986461723066978e-06,
      "loss": 0.3881,
      "step": 2809
    },
    {
      "epoch": 0.12895231976504062,
      "grad_norm": 0.48508352041244507,
      "learning_rate": 9.986443686206631e-06,
      "loss": 0.387,
      "step": 2810
    },
    {
      "epoch": 0.12899821027029507,
      "grad_norm": 0.4662885069847107,
      "learning_rate": 9.986425637355459e-06,
      "loss": 0.3979,
      "step": 2811
    },
    {
      "epoch": 0.12904410077554954,
      "grad_norm": 0.4589211344718933,
      "learning_rate": 9.986407576513508e-06,
      "loss": 0.3599,
      "step": 2812
    },
    {
      "epoch": 0.129089991280804,
      "grad_norm": 0.49436479806900024,
      "learning_rate": 9.98638950368082e-06,
      "loss": 0.4318,
      "step": 2813
    },
    {
      "epoch": 0.12913588178605848,
      "grad_norm": 0.5392730236053467,
      "learning_rate": 9.986371418857439e-06,
      "loss": 0.4702,
      "step": 2814
    },
    {
      "epoch": 0.12918177229131292,
      "grad_norm": 0.44370704889297485,
      "learning_rate": 9.98635332204341e-06,
      "loss": 0.3399,
      "step": 2815
    },
    {
      "epoch": 0.1292276627965674,
      "grad_norm": 0.4396328330039978,
      "learning_rate": 9.986335213238773e-06,
      "loss": 0.3396,
      "step": 2816
    },
    {
      "epoch": 0.12927355330182186,
      "grad_norm": 0.47519561648368835,
      "learning_rate": 9.986317092443574e-06,
      "loss": 0.3803,
      "step": 2817
    },
    {
      "epoch": 0.12931944380707633,
      "grad_norm": 0.4754610061645508,
      "learning_rate": 9.986298959657857e-06,
      "loss": 0.3875,
      "step": 2818
    },
    {
      "epoch": 0.12936533431233077,
      "grad_norm": 0.45905134081840515,
      "learning_rate": 9.986280814881664e-06,
      "loss": 0.3814,
      "step": 2819
    },
    {
      "epoch": 0.12941122481758524,
      "grad_norm": 0.5036689639091492,
      "learning_rate": 9.98626265811504e-06,
      "loss": 0.3898,
      "step": 2820
    },
    {
      "epoch": 0.1294571153228397,
      "grad_norm": 0.4956029951572418,
      "learning_rate": 9.986244489358027e-06,
      "loss": 0.4421,
      "step": 2821
    },
    {
      "epoch": 0.12950300582809418,
      "grad_norm": 0.4815804064273834,
      "learning_rate": 9.98622630861067e-06,
      "loss": 0.3907,
      "step": 2822
    },
    {
      "epoch": 0.12954889633334862,
      "grad_norm": 0.44697144627571106,
      "learning_rate": 9.986208115873012e-06,
      "loss": 0.3483,
      "step": 2823
    },
    {
      "epoch": 0.1295947868386031,
      "grad_norm": 0.4544214904308319,
      "learning_rate": 9.986189911145099e-06,
      "loss": 0.3447,
      "step": 2824
    },
    {
      "epoch": 0.12964067734385756,
      "grad_norm": 0.46760818362236023,
      "learning_rate": 9.986171694426972e-06,
      "loss": 0.4243,
      "step": 2825
    },
    {
      "epoch": 0.129686567849112,
      "grad_norm": 0.46960702538490295,
      "learning_rate": 9.986153465718675e-06,
      "loss": 0.3831,
      "step": 2826
    },
    {
      "epoch": 0.12973245835436648,
      "grad_norm": 0.426664263010025,
      "learning_rate": 9.986135225020252e-06,
      "loss": 0.3425,
      "step": 2827
    },
    {
      "epoch": 0.12977834885962095,
      "grad_norm": 0.43156060576438904,
      "learning_rate": 9.986116972331749e-06,
      "loss": 0.3025,
      "step": 2828
    },
    {
      "epoch": 0.12982423936487542,
      "grad_norm": 0.47947800159454346,
      "learning_rate": 9.986098707653209e-06,
      "loss": 0.3699,
      "step": 2829
    },
    {
      "epoch": 0.12987012987012986,
      "grad_norm": 0.4837958514690399,
      "learning_rate": 9.986080430984674e-06,
      "loss": 0.4231,
      "step": 2830
    },
    {
      "epoch": 0.12991602037538433,
      "grad_norm": 0.479728639125824,
      "learning_rate": 9.986062142326189e-06,
      "loss": 0.4314,
      "step": 2831
    },
    {
      "epoch": 0.1299619108806388,
      "grad_norm": 0.4612409472465515,
      "learning_rate": 9.9860438416778e-06,
      "loss": 0.4022,
      "step": 2832
    },
    {
      "epoch": 0.13000780138589327,
      "grad_norm": 0.4356297254562378,
      "learning_rate": 9.986025529039547e-06,
      "loss": 0.3633,
      "step": 2833
    },
    {
      "epoch": 0.1300536918911477,
      "grad_norm": 0.4393591284751892,
      "learning_rate": 9.986007204411476e-06,
      "loss": 0.3702,
      "step": 2834
    },
    {
      "epoch": 0.13009958239640218,
      "grad_norm": 0.4903585612773895,
      "learning_rate": 9.985988867793633e-06,
      "loss": 0.4499,
      "step": 2835
    },
    {
      "epoch": 0.13014547290165665,
      "grad_norm": 0.4818597435951233,
      "learning_rate": 9.98597051918606e-06,
      "loss": 0.4053,
      "step": 2836
    },
    {
      "epoch": 0.13019136340691112,
      "grad_norm": 0.47199276089668274,
      "learning_rate": 9.9859521585888e-06,
      "loss": 0.3754,
      "step": 2837
    },
    {
      "epoch": 0.13023725391216556,
      "grad_norm": 0.47593963146209717,
      "learning_rate": 9.985933786001902e-06,
      "loss": 0.3988,
      "step": 2838
    },
    {
      "epoch": 0.13028314441742003,
      "grad_norm": 0.4384521245956421,
      "learning_rate": 9.985915401425403e-06,
      "loss": 0.3253,
      "step": 2839
    },
    {
      "epoch": 0.1303290349226745,
      "grad_norm": 0.4463512599468231,
      "learning_rate": 9.985897004859354e-06,
      "loss": 0.3491,
      "step": 2840
    },
    {
      "epoch": 0.13037492542792897,
      "grad_norm": 0.47416383028030396,
      "learning_rate": 9.985878596303795e-06,
      "loss": 0.4332,
      "step": 2841
    },
    {
      "epoch": 0.13042081593318341,
      "grad_norm": 0.44360268115997314,
      "learning_rate": 9.985860175758772e-06,
      "loss": 0.3497,
      "step": 2842
    },
    {
      "epoch": 0.13046670643843788,
      "grad_norm": 0.4880979359149933,
      "learning_rate": 9.985841743224329e-06,
      "loss": 0.4226,
      "step": 2843
    },
    {
      "epoch": 0.13051259694369235,
      "grad_norm": 0.4397323429584503,
      "learning_rate": 9.985823298700509e-06,
      "loss": 0.3382,
      "step": 2844
    },
    {
      "epoch": 0.13055848744894683,
      "grad_norm": 0.47623541951179504,
      "learning_rate": 9.98580484218736e-06,
      "loss": 0.4189,
      "step": 2845
    },
    {
      "epoch": 0.13060437795420127,
      "grad_norm": 0.4655338525772095,
      "learning_rate": 9.98578637368492e-06,
      "loss": 0.4045,
      "step": 2846
    },
    {
      "epoch": 0.13065026845945574,
      "grad_norm": 0.4542866051197052,
      "learning_rate": 9.98576789319324e-06,
      "loss": 0.3466,
      "step": 2847
    },
    {
      "epoch": 0.1306961589647102,
      "grad_norm": 0.4580006003379822,
      "learning_rate": 9.985749400712362e-06,
      "loss": 0.3737,
      "step": 2848
    },
    {
      "epoch": 0.13074204946996468,
      "grad_norm": 0.4561164975166321,
      "learning_rate": 9.98573089624233e-06,
      "loss": 0.4018,
      "step": 2849
    },
    {
      "epoch": 0.13078793997521912,
      "grad_norm": 0.47935986518859863,
      "learning_rate": 9.985712379783188e-06,
      "loss": 0.4262,
      "step": 2850
    },
    {
      "epoch": 0.1308338304804736,
      "grad_norm": 0.4901215136051178,
      "learning_rate": 9.985693851334981e-06,
      "loss": 0.3723,
      "step": 2851
    },
    {
      "epoch": 0.13087972098572806,
      "grad_norm": 0.4891490340232849,
      "learning_rate": 9.985675310897755e-06,
      "loss": 0.4574,
      "step": 2852
    },
    {
      "epoch": 0.1309256114909825,
      "grad_norm": 0.49904853105545044,
      "learning_rate": 9.985656758471551e-06,
      "loss": 0.4673,
      "step": 2853
    },
    {
      "epoch": 0.13097150199623697,
      "grad_norm": 0.4225284457206726,
      "learning_rate": 9.98563819405642e-06,
      "loss": 0.342,
      "step": 2854
    },
    {
      "epoch": 0.13101739250149144,
      "grad_norm": 0.4977639317512512,
      "learning_rate": 9.9856196176524e-06,
      "loss": 0.486,
      "step": 2855
    },
    {
      "epoch": 0.1310632830067459,
      "grad_norm": 0.4584922194480896,
      "learning_rate": 9.985601029259538e-06,
      "loss": 0.375,
      "step": 2856
    },
    {
      "epoch": 0.13110917351200035,
      "grad_norm": 0.42476415634155273,
      "learning_rate": 9.985582428877881e-06,
      "loss": 0.3345,
      "step": 2857
    },
    {
      "epoch": 0.13115506401725482,
      "grad_norm": 0.4525851905345917,
      "learning_rate": 9.985563816507469e-06,
      "loss": 0.3431,
      "step": 2858
    },
    {
      "epoch": 0.1312009545225093,
      "grad_norm": 0.4731519818305969,
      "learning_rate": 9.985545192148351e-06,
      "loss": 0.4038,
      "step": 2859
    },
    {
      "epoch": 0.13124684502776376,
      "grad_norm": 0.6694402694702148,
      "learning_rate": 9.985526555800571e-06,
      "loss": 0.4627,
      "step": 2860
    },
    {
      "epoch": 0.1312927355330182,
      "grad_norm": 0.5048166513442993,
      "learning_rate": 9.985507907464173e-06,
      "loss": 0.4525,
      "step": 2861
    },
    {
      "epoch": 0.13133862603827268,
      "grad_norm": 0.4961734712123871,
      "learning_rate": 9.9854892471392e-06,
      "loss": 0.4833,
      "step": 2862
    },
    {
      "epoch": 0.13138451654352715,
      "grad_norm": 0.4771215617656708,
      "learning_rate": 9.9854705748257e-06,
      "loss": 0.4171,
      "step": 2863
    },
    {
      "epoch": 0.13143040704878162,
      "grad_norm": 0.4323614537715912,
      "learning_rate": 9.985451890523719e-06,
      "loss": 0.3657,
      "step": 2864
    },
    {
      "epoch": 0.13147629755403606,
      "grad_norm": 0.48595213890075684,
      "learning_rate": 9.985433194233296e-06,
      "loss": 0.4805,
      "step": 2865
    },
    {
      "epoch": 0.13152218805929053,
      "grad_norm": 0.5306482315063477,
      "learning_rate": 9.985414485954482e-06,
      "loss": 0.561,
      "step": 2866
    },
    {
      "epoch": 0.131568078564545,
      "grad_norm": 0.4797099828720093,
      "learning_rate": 9.985395765687319e-06,
      "loss": 0.3897,
      "step": 2867
    },
    {
      "epoch": 0.13161396906979947,
      "grad_norm": 0.45678552985191345,
      "learning_rate": 9.985377033431851e-06,
      "loss": 0.36,
      "step": 2868
    },
    {
      "epoch": 0.1316598595750539,
      "grad_norm": 0.504903256893158,
      "learning_rate": 9.985358289188127e-06,
      "loss": 0.4316,
      "step": 2869
    },
    {
      "epoch": 0.13170575008030838,
      "grad_norm": 0.49582263827323914,
      "learning_rate": 9.985339532956189e-06,
      "loss": 0.4183,
      "step": 2870
    },
    {
      "epoch": 0.13175164058556285,
      "grad_norm": 0.518846869468689,
      "learning_rate": 9.985320764736082e-06,
      "loss": 0.4101,
      "step": 2871
    },
    {
      "epoch": 0.13179753109081732,
      "grad_norm": 0.4616560935974121,
      "learning_rate": 9.985301984527853e-06,
      "loss": 0.3848,
      "step": 2872
    },
    {
      "epoch": 0.13184342159607176,
      "grad_norm": 0.47033193707466125,
      "learning_rate": 9.985283192331546e-06,
      "loss": 0.3859,
      "step": 2873
    },
    {
      "epoch": 0.13188931210132623,
      "grad_norm": 0.46414950489997864,
      "learning_rate": 9.985264388147207e-06,
      "loss": 0.3802,
      "step": 2874
    },
    {
      "epoch": 0.1319352026065807,
      "grad_norm": 0.49761998653411865,
      "learning_rate": 9.98524557197488e-06,
      "loss": 0.4782,
      "step": 2875
    },
    {
      "epoch": 0.13198109311183517,
      "grad_norm": 0.4831656813621521,
      "learning_rate": 9.98522674381461e-06,
      "loss": 0.4049,
      "step": 2876
    },
    {
      "epoch": 0.13202698361708962,
      "grad_norm": 0.4910261631011963,
      "learning_rate": 9.985207903666442e-06,
      "loss": 0.4116,
      "step": 2877
    },
    {
      "epoch": 0.13207287412234409,
      "grad_norm": 0.529514729976654,
      "learning_rate": 9.985189051530425e-06,
      "loss": 0.4655,
      "step": 2878
    },
    {
      "epoch": 0.13211876462759856,
      "grad_norm": 0.5253849625587463,
      "learning_rate": 9.9851701874066e-06,
      "loss": 0.4962,
      "step": 2879
    },
    {
      "epoch": 0.13216465513285303,
      "grad_norm": 0.4822869300842285,
      "learning_rate": 9.985151311295016e-06,
      "loss": 0.4606,
      "step": 2880
    },
    {
      "epoch": 0.13221054563810747,
      "grad_norm": 0.5626682043075562,
      "learning_rate": 9.985132423195715e-06,
      "loss": 0.4697,
      "step": 2881
    },
    {
      "epoch": 0.13225643614336194,
      "grad_norm": 0.4947010576725006,
      "learning_rate": 9.985113523108743e-06,
      "loss": 0.4634,
      "step": 2882
    },
    {
      "epoch": 0.1323023266486164,
      "grad_norm": 0.4874163866043091,
      "learning_rate": 9.985094611034148e-06,
      "loss": 0.4035,
      "step": 2883
    },
    {
      "epoch": 0.13234821715387085,
      "grad_norm": 0.5082656741142273,
      "learning_rate": 9.985075686971974e-06,
      "loss": 0.4759,
      "step": 2884
    },
    {
      "epoch": 0.13239410765912532,
      "grad_norm": 0.4486599564552307,
      "learning_rate": 9.985056750922263e-06,
      "loss": 0.3524,
      "step": 2885
    },
    {
      "epoch": 0.1324399981643798,
      "grad_norm": 0.47667229175567627,
      "learning_rate": 9.985037802885068e-06,
      "loss": 0.4013,
      "step": 2886
    },
    {
      "epoch": 0.13248588866963426,
      "grad_norm": 0.48025259375572205,
      "learning_rate": 9.985018842860428e-06,
      "loss": 0.3661,
      "step": 2887
    },
    {
      "epoch": 0.1325317791748887,
      "grad_norm": 0.42884561419487,
      "learning_rate": 9.984999870848392e-06,
      "loss": 0.3575,
      "step": 2888
    },
    {
      "epoch": 0.13257766968014317,
      "grad_norm": 0.49767404794692993,
      "learning_rate": 9.984980886849005e-06,
      "loss": 0.4659,
      "step": 2889
    },
    {
      "epoch": 0.13262356018539764,
      "grad_norm": 0.502864420413971,
      "learning_rate": 9.984961890862311e-06,
      "loss": 0.4605,
      "step": 2890
    },
    {
      "epoch": 0.1326694506906521,
      "grad_norm": 0.4371437728404999,
      "learning_rate": 9.984942882888358e-06,
      "loss": 0.3454,
      "step": 2891
    },
    {
      "epoch": 0.13271534119590656,
      "grad_norm": 0.5332352519035339,
      "learning_rate": 9.98492386292719e-06,
      "loss": 0.5113,
      "step": 2892
    },
    {
      "epoch": 0.13276123170116103,
      "grad_norm": 0.4774024486541748,
      "learning_rate": 9.984904830978855e-06,
      "loss": 0.4557,
      "step": 2893
    },
    {
      "epoch": 0.1328071222064155,
      "grad_norm": 0.4903777837753296,
      "learning_rate": 9.984885787043394e-06,
      "loss": 0.4565,
      "step": 2894
    },
    {
      "epoch": 0.13285301271166997,
      "grad_norm": 0.46722671389579773,
      "learning_rate": 9.984866731120858e-06,
      "loss": 0.392,
      "step": 2895
    },
    {
      "epoch": 0.1328989032169244,
      "grad_norm": 0.5084421634674072,
      "learning_rate": 9.98484766321129e-06,
      "loss": 0.4594,
      "step": 2896
    },
    {
      "epoch": 0.13294479372217888,
      "grad_norm": 0.42572417855262756,
      "learning_rate": 9.984828583314738e-06,
      "loss": 0.3296,
      "step": 2897
    },
    {
      "epoch": 0.13299068422743335,
      "grad_norm": 0.5137713551521301,
      "learning_rate": 9.984809491431246e-06,
      "loss": 0.4231,
      "step": 2898
    },
    {
      "epoch": 0.13303657473268782,
      "grad_norm": 0.540494978427887,
      "learning_rate": 9.984790387560859e-06,
      "loss": 0.4796,
      "step": 2899
    },
    {
      "epoch": 0.13308246523794226,
      "grad_norm": 0.5203617811203003,
      "learning_rate": 9.984771271703626e-06,
      "loss": 0.465,
      "step": 2900
    },
    {
      "epoch": 0.13312835574319673,
      "grad_norm": 0.45151373744010925,
      "learning_rate": 9.984752143859591e-06,
      "loss": 0.4146,
      "step": 2901
    },
    {
      "epoch": 0.1331742462484512,
      "grad_norm": 0.49115505814552307,
      "learning_rate": 9.9847330040288e-06,
      "loss": 0.4722,
      "step": 2902
    },
    {
      "epoch": 0.13322013675370567,
      "grad_norm": 0.49026769399642944,
      "learning_rate": 9.9847138522113e-06,
      "loss": 0.4013,
      "step": 2903
    },
    {
      "epoch": 0.1332660272589601,
      "grad_norm": 0.48324286937713623,
      "learning_rate": 9.984694688407135e-06,
      "loss": 0.3783,
      "step": 2904
    },
    {
      "epoch": 0.13331191776421458,
      "grad_norm": 0.4451691210269928,
      "learning_rate": 9.984675512616353e-06,
      "loss": 0.3456,
      "step": 2905
    },
    {
      "epoch": 0.13335780826946905,
      "grad_norm": 0.5046260356903076,
      "learning_rate": 9.984656324839001e-06,
      "loss": 0.4426,
      "step": 2906
    },
    {
      "epoch": 0.13340369877472352,
      "grad_norm": 0.47953924536705017,
      "learning_rate": 9.984637125075123e-06,
      "loss": 0.4526,
      "step": 2907
    },
    {
      "epoch": 0.13344958927997796,
      "grad_norm": 0.4949834942817688,
      "learning_rate": 9.984617913324765e-06,
      "loss": 0.4242,
      "step": 2908
    },
    {
      "epoch": 0.13349547978523243,
      "grad_norm": 0.4905979633331299,
      "learning_rate": 9.984598689587975e-06,
      "loss": 0.376,
      "step": 2909
    },
    {
      "epoch": 0.1335413702904869,
      "grad_norm": 0.47098323702812195,
      "learning_rate": 9.984579453864799e-06,
      "loss": 0.4382,
      "step": 2910
    },
    {
      "epoch": 0.13358726079574137,
      "grad_norm": 0.49535703659057617,
      "learning_rate": 9.98456020615528e-06,
      "loss": 0.4273,
      "step": 2911
    },
    {
      "epoch": 0.13363315130099582,
      "grad_norm": 0.43479371070861816,
      "learning_rate": 9.984540946459469e-06,
      "loss": 0.3242,
      "step": 2912
    },
    {
      "epoch": 0.1336790418062503,
      "grad_norm": 0.4440033733844757,
      "learning_rate": 9.98452167477741e-06,
      "loss": 0.3853,
      "step": 2913
    },
    {
      "epoch": 0.13372493231150476,
      "grad_norm": 0.4634888172149658,
      "learning_rate": 9.984502391109148e-06,
      "loss": 0.3898,
      "step": 2914
    },
    {
      "epoch": 0.1337708228167592,
      "grad_norm": 0.46266263723373413,
      "learning_rate": 9.984483095454734e-06,
      "loss": 0.4323,
      "step": 2915
    },
    {
      "epoch": 0.13381671332201367,
      "grad_norm": 0.5111375451087952,
      "learning_rate": 9.984463787814209e-06,
      "loss": 0.5356,
      "step": 2916
    },
    {
      "epoch": 0.13386260382726814,
      "grad_norm": 0.536014199256897,
      "learning_rate": 9.984444468187622e-06,
      "loss": 0.4703,
      "step": 2917
    },
    {
      "epoch": 0.1339084943325226,
      "grad_norm": 0.4837159514427185,
      "learning_rate": 9.984425136575019e-06,
      "loss": 0.4142,
      "step": 2918
    },
    {
      "epoch": 0.13395438483777705,
      "grad_norm": 0.575469434261322,
      "learning_rate": 9.984405792976447e-06,
      "loss": 0.5177,
      "step": 2919
    },
    {
      "epoch": 0.13400027534303152,
      "grad_norm": 0.5213784575462341,
      "learning_rate": 9.984386437391951e-06,
      "loss": 0.5453,
      "step": 2920
    },
    {
      "epoch": 0.134046165848286,
      "grad_norm": 0.46826115250587463,
      "learning_rate": 9.984367069821582e-06,
      "loss": 0.3794,
      "step": 2921
    },
    {
      "epoch": 0.13409205635354046,
      "grad_norm": 0.5002729296684265,
      "learning_rate": 9.98434769026538e-06,
      "loss": 0.4654,
      "step": 2922
    },
    {
      "epoch": 0.1341379468587949,
      "grad_norm": 0.46221449971199036,
      "learning_rate": 9.984328298723396e-06,
      "loss": 0.3405,
      "step": 2923
    },
    {
      "epoch": 0.13418383736404937,
      "grad_norm": 0.46960917115211487,
      "learning_rate": 9.984308895195677e-06,
      "loss": 0.3443,
      "step": 2924
    },
    {
      "epoch": 0.13422972786930384,
      "grad_norm": 0.5330900549888611,
      "learning_rate": 9.984289479682265e-06,
      "loss": 0.5086,
      "step": 2925
    },
    {
      "epoch": 0.13427561837455831,
      "grad_norm": 0.5278019309043884,
      "learning_rate": 9.984270052183213e-06,
      "loss": 0.4222,
      "step": 2926
    },
    {
      "epoch": 0.13432150887981276,
      "grad_norm": 0.4523743987083435,
      "learning_rate": 9.984250612698564e-06,
      "loss": 0.387,
      "step": 2927
    },
    {
      "epoch": 0.13436739938506723,
      "grad_norm": 0.5056136250495911,
      "learning_rate": 9.984231161228364e-06,
      "loss": 0.4466,
      "step": 2928
    },
    {
      "epoch": 0.1344132898903217,
      "grad_norm": 0.5223401188850403,
      "learning_rate": 9.984211697772662e-06,
      "loss": 0.469,
      "step": 2929
    },
    {
      "epoch": 0.13445918039557617,
      "grad_norm": 0.4763677716255188,
      "learning_rate": 9.984192222331503e-06,
      "loss": 0.3786,
      "step": 2930
    },
    {
      "epoch": 0.1345050709008306,
      "grad_norm": 0.4712555408477783,
      "learning_rate": 9.984172734904937e-06,
      "loss": 0.4073,
      "step": 2931
    },
    {
      "epoch": 0.13455096140608508,
      "grad_norm": 0.505085289478302,
      "learning_rate": 9.984153235493006e-06,
      "loss": 0.471,
      "step": 2932
    },
    {
      "epoch": 0.13459685191133955,
      "grad_norm": 0.4289167523384094,
      "learning_rate": 9.984133724095761e-06,
      "loss": 0.2825,
      "step": 2933
    },
    {
      "epoch": 0.13464274241659402,
      "grad_norm": 0.540083110332489,
      "learning_rate": 9.984114200713247e-06,
      "loss": 0.5823,
      "step": 2934
    },
    {
      "epoch": 0.13468863292184846,
      "grad_norm": 0.45814406871795654,
      "learning_rate": 9.984094665345512e-06,
      "loss": 0.4116,
      "step": 2935
    },
    {
      "epoch": 0.13473452342710293,
      "grad_norm": 0.4792840778827667,
      "learning_rate": 9.984075117992602e-06,
      "loss": 0.378,
      "step": 2936
    },
    {
      "epoch": 0.1347804139323574,
      "grad_norm": 0.458863765001297,
      "learning_rate": 9.984055558654565e-06,
      "loss": 0.4024,
      "step": 2937
    },
    {
      "epoch": 0.13482630443761187,
      "grad_norm": 0.5246399641036987,
      "learning_rate": 9.984035987331446e-06,
      "loss": 0.4475,
      "step": 2938
    },
    {
      "epoch": 0.1348721949428663,
      "grad_norm": 0.49317172169685364,
      "learning_rate": 9.984016404023293e-06,
      "loss": 0.4132,
      "step": 2939
    },
    {
      "epoch": 0.13491808544812078,
      "grad_norm": 0.4826701581478119,
      "learning_rate": 9.983996808730157e-06,
      "loss": 0.3769,
      "step": 2940
    },
    {
      "epoch": 0.13496397595337525,
      "grad_norm": 0.543339192867279,
      "learning_rate": 9.983977201452077e-06,
      "loss": 0.4822,
      "step": 2941
    },
    {
      "epoch": 0.1350098664586297,
      "grad_norm": 0.46195462346076965,
      "learning_rate": 9.983957582189108e-06,
      "loss": 0.331,
      "step": 2942
    },
    {
      "epoch": 0.13505575696388417,
      "grad_norm": 0.47714972496032715,
      "learning_rate": 9.983937950941293e-06,
      "loss": 0.4408,
      "step": 2943
    },
    {
      "epoch": 0.13510164746913864,
      "grad_norm": 0.44046348333358765,
      "learning_rate": 9.98391830770868e-06,
      "loss": 0.3491,
      "step": 2944
    },
    {
      "epoch": 0.1351475379743931,
      "grad_norm": 0.4799294173717499,
      "learning_rate": 9.983898652491318e-06,
      "loss": 0.4402,
      "step": 2945
    },
    {
      "epoch": 0.13519342847964755,
      "grad_norm": 0.48560038208961487,
      "learning_rate": 9.98387898528925e-06,
      "loss": 0.439,
      "step": 2946
    },
    {
      "epoch": 0.13523931898490202,
      "grad_norm": 0.4312838613986969,
      "learning_rate": 9.983859306102527e-06,
      "loss": 0.3,
      "step": 2947
    },
    {
      "epoch": 0.1352852094901565,
      "grad_norm": 0.46086788177490234,
      "learning_rate": 9.983839614931197e-06,
      "loss": 0.3927,
      "step": 2948
    },
    {
      "epoch": 0.13533109999541096,
      "grad_norm": 0.44949251413345337,
      "learning_rate": 9.983819911775305e-06,
      "loss": 0.4055,
      "step": 2949
    },
    {
      "epoch": 0.1353769905006654,
      "grad_norm": 0.5050535202026367,
      "learning_rate": 9.983800196634898e-06,
      "loss": 0.4866,
      "step": 2950
    },
    {
      "epoch": 0.13542288100591987,
      "grad_norm": 0.4712943434715271,
      "learning_rate": 9.983780469510025e-06,
      "loss": 0.4055,
      "step": 2951
    },
    {
      "epoch": 0.13546877151117434,
      "grad_norm": 0.44342565536499023,
      "learning_rate": 9.983760730400733e-06,
      "loss": 0.3747,
      "step": 2952
    },
    {
      "epoch": 0.1355146620164288,
      "grad_norm": 0.5054129362106323,
      "learning_rate": 9.98374097930707e-06,
      "loss": 0.4137,
      "step": 2953
    },
    {
      "epoch": 0.13556055252168325,
      "grad_norm": 0.5286996364593506,
      "learning_rate": 9.983721216229081e-06,
      "loss": 0.5068,
      "step": 2954
    },
    {
      "epoch": 0.13560644302693772,
      "grad_norm": 0.4568662643432617,
      "learning_rate": 9.983701441166817e-06,
      "loss": 0.3382,
      "step": 2955
    },
    {
      "epoch": 0.1356523335321922,
      "grad_norm": 0.5419427752494812,
      "learning_rate": 9.983681654120323e-06,
      "loss": 0.5546,
      "step": 2956
    },
    {
      "epoch": 0.13569822403744666,
      "grad_norm": 0.5187088847160339,
      "learning_rate": 9.983661855089647e-06,
      "loss": 0.4198,
      "step": 2957
    },
    {
      "epoch": 0.1357441145427011,
      "grad_norm": 0.47891896963119507,
      "learning_rate": 9.98364204407484e-06,
      "loss": 0.3829,
      "step": 2958
    },
    {
      "epoch": 0.13579000504795558,
      "grad_norm": 0.5072096586227417,
      "learning_rate": 9.983622221075944e-06,
      "loss": 0.4828,
      "step": 2959
    },
    {
      "epoch": 0.13583589555321005,
      "grad_norm": 0.43741774559020996,
      "learning_rate": 9.98360238609301e-06,
      "loss": 0.3563,
      "step": 2960
    },
    {
      "epoch": 0.13588178605846452,
      "grad_norm": 0.47707393765449524,
      "learning_rate": 9.983582539126086e-06,
      "loss": 0.4248,
      "step": 2961
    },
    {
      "epoch": 0.13592767656371896,
      "grad_norm": 0.4735635221004486,
      "learning_rate": 9.983562680175219e-06,
      "loss": 0.3808,
      "step": 2962
    },
    {
      "epoch": 0.13597356706897343,
      "grad_norm": 0.5682708024978638,
      "learning_rate": 9.983542809240455e-06,
      "loss": 0.4385,
      "step": 2963
    },
    {
      "epoch": 0.1360194575742279,
      "grad_norm": 0.49021559953689575,
      "learning_rate": 9.983522926321845e-06,
      "loss": 0.4929,
      "step": 2964
    },
    {
      "epoch": 0.13606534807948237,
      "grad_norm": 0.4581962823867798,
      "learning_rate": 9.983503031419434e-06,
      "loss": 0.3794,
      "step": 2965
    },
    {
      "epoch": 0.1361112385847368,
      "grad_norm": 0.45535171031951904,
      "learning_rate": 9.983483124533271e-06,
      "loss": 0.4255,
      "step": 2966
    },
    {
      "epoch": 0.13615712908999128,
      "grad_norm": 0.4496273994445801,
      "learning_rate": 9.983463205663403e-06,
      "loss": 0.3651,
      "step": 2967
    },
    {
      "epoch": 0.13620301959524575,
      "grad_norm": 0.4613458514213562,
      "learning_rate": 9.983443274809881e-06,
      "loss": 0.3727,
      "step": 2968
    },
    {
      "epoch": 0.13624891010050022,
      "grad_norm": 0.4700455665588379,
      "learning_rate": 9.98342333197275e-06,
      "loss": 0.4458,
      "step": 2969
    },
    {
      "epoch": 0.13629480060575466,
      "grad_norm": 0.44079285860061646,
      "learning_rate": 9.98340337715206e-06,
      "loss": 0.376,
      "step": 2970
    },
    {
      "epoch": 0.13634069111100913,
      "grad_norm": 0.4573451280593872,
      "learning_rate": 9.983383410347856e-06,
      "loss": 0.4004,
      "step": 2971
    },
    {
      "epoch": 0.1363865816162636,
      "grad_norm": 0.4634217619895935,
      "learning_rate": 9.983363431560189e-06,
      "loss": 0.3977,
      "step": 2972
    },
    {
      "epoch": 0.13643247212151804,
      "grad_norm": 0.5109870433807373,
      "learning_rate": 9.983343440789104e-06,
      "loss": 0.3594,
      "step": 2973
    },
    {
      "epoch": 0.13647836262677251,
      "grad_norm": 0.4920806884765625,
      "learning_rate": 9.983323438034652e-06,
      "loss": 0.4794,
      "step": 2974
    },
    {
      "epoch": 0.13652425313202698,
      "grad_norm": 0.4754081070423126,
      "learning_rate": 9.98330342329688e-06,
      "loss": 0.3923,
      "step": 2975
    },
    {
      "epoch": 0.13657014363728145,
      "grad_norm": 0.49609583616256714,
      "learning_rate": 9.983283396575835e-06,
      "loss": 0.4968,
      "step": 2976
    },
    {
      "epoch": 0.1366160341425359,
      "grad_norm": 0.4833639860153198,
      "learning_rate": 9.983263357871566e-06,
      "loss": 0.4471,
      "step": 2977
    },
    {
      "epoch": 0.13666192464779037,
      "grad_norm": 0.47659772634506226,
      "learning_rate": 9.983243307184122e-06,
      "loss": 0.454,
      "step": 2978
    },
    {
      "epoch": 0.13670781515304484,
      "grad_norm": 0.48343032598495483,
      "learning_rate": 9.98322324451355e-06,
      "loss": 0.3838,
      "step": 2979
    },
    {
      "epoch": 0.1367537056582993,
      "grad_norm": 0.45296165347099304,
      "learning_rate": 9.9832031698599e-06,
      "loss": 0.4094,
      "step": 2980
    },
    {
      "epoch": 0.13679959616355375,
      "grad_norm": 0.49433180689811707,
      "learning_rate": 9.98318308322322e-06,
      "loss": 0.423,
      "step": 2981
    },
    {
      "epoch": 0.13684548666880822,
      "grad_norm": 0.5197073221206665,
      "learning_rate": 9.983162984603557e-06,
      "loss": 0.5168,
      "step": 2982
    },
    {
      "epoch": 0.1368913771740627,
      "grad_norm": 0.487143337726593,
      "learning_rate": 9.983142874000958e-06,
      "loss": 0.4387,
      "step": 2983
    },
    {
      "epoch": 0.13693726767931716,
      "grad_norm": 0.4529087245464325,
      "learning_rate": 9.983122751415475e-06,
      "loss": 0.3518,
      "step": 2984
    },
    {
      "epoch": 0.1369831581845716,
      "grad_norm": 0.47358787059783936,
      "learning_rate": 9.983102616847153e-06,
      "loss": 0.3994,
      "step": 2985
    },
    {
      "epoch": 0.13702904868982607,
      "grad_norm": 0.42946839332580566,
      "learning_rate": 9.983082470296042e-06,
      "loss": 0.3157,
      "step": 2986
    },
    {
      "epoch": 0.13707493919508054,
      "grad_norm": 0.4564993977546692,
      "learning_rate": 9.983062311762191e-06,
      "loss": 0.3439,
      "step": 2987
    },
    {
      "epoch": 0.137120829700335,
      "grad_norm": 0.48037704825401306,
      "learning_rate": 9.983042141245648e-06,
      "loss": 0.3779,
      "step": 2988
    },
    {
      "epoch": 0.13716672020558945,
      "grad_norm": 0.4633772373199463,
      "learning_rate": 9.98302195874646e-06,
      "loss": 0.3894,
      "step": 2989
    },
    {
      "epoch": 0.13721261071084392,
      "grad_norm": 0.49246591329574585,
      "learning_rate": 9.98300176426468e-06,
      "loss": 0.5057,
      "step": 2990
    },
    {
      "epoch": 0.1372585012160984,
      "grad_norm": 0.5284117460250854,
      "learning_rate": 9.98298155780035e-06,
      "loss": 0.4869,
      "step": 2991
    },
    {
      "epoch": 0.13730439172135286,
      "grad_norm": 0.4986521005630493,
      "learning_rate": 9.982961339353522e-06,
      "loss": 0.3958,
      "step": 2992
    },
    {
      "epoch": 0.1373502822266073,
      "grad_norm": 0.493929922580719,
      "learning_rate": 9.982941108924247e-06,
      "loss": 0.4244,
      "step": 2993
    },
    {
      "epoch": 0.13739617273186178,
      "grad_norm": 0.4537135064601898,
      "learning_rate": 9.98292086651257e-06,
      "loss": 0.4015,
      "step": 2994
    },
    {
      "epoch": 0.13744206323711625,
      "grad_norm": 0.4772287607192993,
      "learning_rate": 9.982900612118543e-06,
      "loss": 0.4243,
      "step": 2995
    },
    {
      "epoch": 0.13748795374237072,
      "grad_norm": 0.44992128014564514,
      "learning_rate": 9.982880345742209e-06,
      "loss": 0.4048,
      "step": 2996
    },
    {
      "epoch": 0.13753384424762516,
      "grad_norm": 0.47454380989074707,
      "learning_rate": 9.982860067383624e-06,
      "loss": 0.3977,
      "step": 2997
    },
    {
      "epoch": 0.13757973475287963,
      "grad_norm": 0.4821033775806427,
      "learning_rate": 9.98283977704283e-06,
      "loss": 0.4167,
      "step": 2998
    },
    {
      "epoch": 0.1376256252581341,
      "grad_norm": 0.5374534726142883,
      "learning_rate": 9.982819474719882e-06,
      "loss": 0.5086,
      "step": 2999
    },
    {
      "epoch": 0.13767151576338854,
      "grad_norm": 0.452519029378891,
      "learning_rate": 9.982799160414824e-06,
      "loss": 0.3634,
      "step": 3000
    },
    {
      "epoch": 0.137717406268643,
      "grad_norm": 0.44751930236816406,
      "learning_rate": 9.982778834127706e-06,
      "loss": 0.4112,
      "step": 3001
    },
    {
      "epoch": 0.13776329677389748,
      "grad_norm": 0.5348258018493652,
      "learning_rate": 9.982758495858578e-06,
      "loss": 0.4842,
      "step": 3002
    },
    {
      "epoch": 0.13780918727915195,
      "grad_norm": 0.494220495223999,
      "learning_rate": 9.982738145607489e-06,
      "loss": 0.4673,
      "step": 3003
    },
    {
      "epoch": 0.1378550777844064,
      "grad_norm": 0.45482999086380005,
      "learning_rate": 9.982717783374488e-06,
      "loss": 0.3896,
      "step": 3004
    },
    {
      "epoch": 0.13790096828966086,
      "grad_norm": 0.4997367858886719,
      "learning_rate": 9.982697409159623e-06,
      "loss": 0.4042,
      "step": 3005
    },
    {
      "epoch": 0.13794685879491533,
      "grad_norm": 0.4939901828765869,
      "learning_rate": 9.982677022962942e-06,
      "loss": 0.4215,
      "step": 3006
    },
    {
      "epoch": 0.1379927493001698,
      "grad_norm": 0.5064058899879456,
      "learning_rate": 9.982656624784497e-06,
      "loss": 0.5187,
      "step": 3007
    },
    {
      "epoch": 0.13803863980542425,
      "grad_norm": 0.4413110315799713,
      "learning_rate": 9.982636214624334e-06,
      "loss": 0.359,
      "step": 3008
    },
    {
      "epoch": 0.13808453031067872,
      "grad_norm": 0.46107569336891174,
      "learning_rate": 9.982615792482503e-06,
      "loss": 0.3349,
      "step": 3009
    },
    {
      "epoch": 0.13813042081593319,
      "grad_norm": 0.5945149660110474,
      "learning_rate": 9.982595358359055e-06,
      "loss": 0.4809,
      "step": 3010
    },
    {
      "epoch": 0.13817631132118766,
      "grad_norm": 0.46925055980682373,
      "learning_rate": 9.982574912254039e-06,
      "loss": 0.4304,
      "step": 3011
    },
    {
      "epoch": 0.1382222018264421,
      "grad_norm": 0.4706903398036957,
      "learning_rate": 9.9825544541675e-06,
      "loss": 0.4235,
      "step": 3012
    },
    {
      "epoch": 0.13826809233169657,
      "grad_norm": 0.4697916805744171,
      "learning_rate": 9.982533984099492e-06,
      "loss": 0.3537,
      "step": 3013
    },
    {
      "epoch": 0.13831398283695104,
      "grad_norm": 0.5078604221343994,
      "learning_rate": 9.982513502050062e-06,
      "loss": 0.4434,
      "step": 3014
    },
    {
      "epoch": 0.1383598733422055,
      "grad_norm": 0.43013402819633484,
      "learning_rate": 9.982493008019258e-06,
      "loss": 0.3062,
      "step": 3015
    },
    {
      "epoch": 0.13840576384745995,
      "grad_norm": 0.48810437321662903,
      "learning_rate": 9.982472502007132e-06,
      "loss": 0.382,
      "step": 3016
    },
    {
      "epoch": 0.13845165435271442,
      "grad_norm": 0.493046373128891,
      "learning_rate": 9.982451984013732e-06,
      "loss": 0.4408,
      "step": 3017
    },
    {
      "epoch": 0.1384975448579689,
      "grad_norm": 0.4840654730796814,
      "learning_rate": 9.982431454039107e-06,
      "loss": 0.4071,
      "step": 3018
    },
    {
      "epoch": 0.13854343536322336,
      "grad_norm": 0.465982049703598,
      "learning_rate": 9.98241091208331e-06,
      "loss": 0.4104,
      "step": 3019
    },
    {
      "epoch": 0.1385893258684778,
      "grad_norm": 0.44012361764907837,
      "learning_rate": 9.982390358146383e-06,
      "loss": 0.3719,
      "step": 3020
    },
    {
      "epoch": 0.13863521637373227,
      "grad_norm": 0.4710675776004791,
      "learning_rate": 9.982369792228382e-06,
      "loss": 0.4165,
      "step": 3021
    },
    {
      "epoch": 0.13868110687898674,
      "grad_norm": 0.603533923625946,
      "learning_rate": 9.982349214329353e-06,
      "loss": 0.5535,
      "step": 3022
    },
    {
      "epoch": 0.1387269973842412,
      "grad_norm": 0.4514589011669159,
      "learning_rate": 9.982328624449347e-06,
      "loss": 0.3429,
      "step": 3023
    },
    {
      "epoch": 0.13877288788949566,
      "grad_norm": 0.48825907707214355,
      "learning_rate": 9.982308022588413e-06,
      "loss": 0.4228,
      "step": 3024
    },
    {
      "epoch": 0.13881877839475013,
      "grad_norm": 0.47944125533103943,
      "learning_rate": 9.982287408746601e-06,
      "loss": 0.4445,
      "step": 3025
    },
    {
      "epoch": 0.1388646689000046,
      "grad_norm": 0.4505082964897156,
      "learning_rate": 9.98226678292396e-06,
      "loss": 0.3558,
      "step": 3026
    },
    {
      "epoch": 0.13891055940525907,
      "grad_norm": 0.5942498445510864,
      "learning_rate": 9.98224614512054e-06,
      "loss": 0.4874,
      "step": 3027
    },
    {
      "epoch": 0.1389564499105135,
      "grad_norm": 0.496677041053772,
      "learning_rate": 9.982225495336391e-06,
      "loss": 0.4097,
      "step": 3028
    },
    {
      "epoch": 0.13900234041576798,
      "grad_norm": 0.455191969871521,
      "learning_rate": 9.982204833571563e-06,
      "loss": 0.3526,
      "step": 3029
    },
    {
      "epoch": 0.13904823092102245,
      "grad_norm": 0.4483187794685364,
      "learning_rate": 9.982184159826103e-06,
      "loss": 0.3606,
      "step": 3030
    },
    {
      "epoch": 0.1390941214262769,
      "grad_norm": 0.5057973861694336,
      "learning_rate": 9.982163474100061e-06,
      "loss": 0.4309,
      "step": 3031
    },
    {
      "epoch": 0.13914001193153136,
      "grad_norm": 0.46668848395347595,
      "learning_rate": 9.982142776393492e-06,
      "loss": 0.3847,
      "step": 3032
    },
    {
      "epoch": 0.13918590243678583,
      "grad_norm": 0.5688084363937378,
      "learning_rate": 9.982122066706439e-06,
      "loss": 0.5048,
      "step": 3033
    },
    {
      "epoch": 0.1392317929420403,
      "grad_norm": 0.46805140376091003,
      "learning_rate": 9.982101345038957e-06,
      "loss": 0.3865,
      "step": 3034
    },
    {
      "epoch": 0.13927768344729474,
      "grad_norm": 0.470278263092041,
      "learning_rate": 9.982080611391091e-06,
      "loss": 0.4083,
      "step": 3035
    },
    {
      "epoch": 0.1393235739525492,
      "grad_norm": 0.47664496302604675,
      "learning_rate": 9.982059865762897e-06,
      "loss": 0.4615,
      "step": 3036
    },
    {
      "epoch": 0.13936946445780368,
      "grad_norm": 0.4638785421848297,
      "learning_rate": 9.98203910815442e-06,
      "loss": 0.3479,
      "step": 3037
    },
    {
      "epoch": 0.13941535496305815,
      "grad_norm": 0.5027391910552979,
      "learning_rate": 9.98201833856571e-06,
      "loss": 0.46,
      "step": 3038
    },
    {
      "epoch": 0.1394612454683126,
      "grad_norm": 0.4439915120601654,
      "learning_rate": 9.981997556996818e-06,
      "loss": 0.3558,
      "step": 3039
    },
    {
      "epoch": 0.13950713597356706,
      "grad_norm": 0.48414838314056396,
      "learning_rate": 9.981976763447797e-06,
      "loss": 0.4048,
      "step": 3040
    },
    {
      "epoch": 0.13955302647882153,
      "grad_norm": 0.4496571719646454,
      "learning_rate": 9.981955957918692e-06,
      "loss": 0.3261,
      "step": 3041
    },
    {
      "epoch": 0.139598916984076,
      "grad_norm": 0.45858386158943176,
      "learning_rate": 9.981935140409556e-06,
      "loss": 0.3721,
      "step": 3042
    },
    {
      "epoch": 0.13964480748933045,
      "grad_norm": 0.5063408613204956,
      "learning_rate": 9.981914310920438e-06,
      "loss": 0.3553,
      "step": 3043
    },
    {
      "epoch": 0.13969069799458492,
      "grad_norm": 0.5631232857704163,
      "learning_rate": 9.98189346945139e-06,
      "loss": 0.4224,
      "step": 3044
    },
    {
      "epoch": 0.1397365884998394,
      "grad_norm": 0.4210375249385834,
      "learning_rate": 9.98187261600246e-06,
      "loss": 0.299,
      "step": 3045
    },
    {
      "epoch": 0.13978247900509386,
      "grad_norm": 0.46979600191116333,
      "learning_rate": 9.981851750573697e-06,
      "loss": 0.4476,
      "step": 3046
    },
    {
      "epoch": 0.1398283695103483,
      "grad_norm": 0.48486843705177307,
      "learning_rate": 9.981830873165154e-06,
      "loss": 0.3981,
      "step": 3047
    },
    {
      "epoch": 0.13987426001560277,
      "grad_norm": 0.45954447984695435,
      "learning_rate": 9.98180998377688e-06,
      "loss": 0.3512,
      "step": 3048
    },
    {
      "epoch": 0.13992015052085724,
      "grad_norm": 0.4965614378452301,
      "learning_rate": 9.981789082408925e-06,
      "loss": 0.4775,
      "step": 3049
    },
    {
      "epoch": 0.1399660410261117,
      "grad_norm": 0.5440787672996521,
      "learning_rate": 9.98176816906134e-06,
      "loss": 0.5307,
      "step": 3050
    },
    {
      "epoch": 0.14001193153136615,
      "grad_norm": 0.48101547360420227,
      "learning_rate": 9.981747243734174e-06,
      "loss": 0.3847,
      "step": 3051
    },
    {
      "epoch": 0.14005782203662062,
      "grad_norm": 0.4872550070285797,
      "learning_rate": 9.981726306427478e-06,
      "loss": 0.4246,
      "step": 3052
    },
    {
      "epoch": 0.1401037125418751,
      "grad_norm": 0.49027836322784424,
      "learning_rate": 9.981705357141303e-06,
      "loss": 0.404,
      "step": 3053
    },
    {
      "epoch": 0.14014960304712956,
      "grad_norm": 0.4657130837440491,
      "learning_rate": 9.981684395875699e-06,
      "loss": 0.4031,
      "step": 3054
    },
    {
      "epoch": 0.140195493552384,
      "grad_norm": 0.4884653091430664,
      "learning_rate": 9.981663422630718e-06,
      "loss": 0.4748,
      "step": 3055
    },
    {
      "epoch": 0.14024138405763847,
      "grad_norm": 0.512407660484314,
      "learning_rate": 9.981642437406407e-06,
      "loss": 0.5276,
      "step": 3056
    },
    {
      "epoch": 0.14028727456289294,
      "grad_norm": 0.4834372103214264,
      "learning_rate": 9.981621440202818e-06,
      "loss": 0.4412,
      "step": 3057
    },
    {
      "epoch": 0.14033316506814741,
      "grad_norm": 0.4676523506641388,
      "learning_rate": 9.981600431020003e-06,
      "loss": 0.4183,
      "step": 3058
    },
    {
      "epoch": 0.14037905557340186,
      "grad_norm": 0.4383641481399536,
      "learning_rate": 9.981579409858008e-06,
      "loss": 0.3281,
      "step": 3059
    },
    {
      "epoch": 0.14042494607865633,
      "grad_norm": 0.5031691193580627,
      "learning_rate": 9.98155837671689e-06,
      "loss": 0.4985,
      "step": 3060
    },
    {
      "epoch": 0.1404708365839108,
      "grad_norm": 0.4581845700740814,
      "learning_rate": 9.981537331596696e-06,
      "loss": 0.3907,
      "step": 3061
    },
    {
      "epoch": 0.14051672708916524,
      "grad_norm": 0.5085744857788086,
      "learning_rate": 9.981516274497476e-06,
      "loss": 0.4928,
      "step": 3062
    },
    {
      "epoch": 0.1405626175944197,
      "grad_norm": 0.48630714416503906,
      "learning_rate": 9.981495205419282e-06,
      "loss": 0.4323,
      "step": 3063
    },
    {
      "epoch": 0.14060850809967418,
      "grad_norm": 0.4920274019241333,
      "learning_rate": 9.981474124362163e-06,
      "loss": 0.4801,
      "step": 3064
    },
    {
      "epoch": 0.14065439860492865,
      "grad_norm": 0.4722157120704651,
      "learning_rate": 9.98145303132617e-06,
      "loss": 0.3693,
      "step": 3065
    },
    {
      "epoch": 0.1407002891101831,
      "grad_norm": 0.5346464514732361,
      "learning_rate": 9.981431926311359e-06,
      "loss": 0.5228,
      "step": 3066
    },
    {
      "epoch": 0.14074617961543756,
      "grad_norm": 0.46221181750297546,
      "learning_rate": 9.981410809317773e-06,
      "loss": 0.3349,
      "step": 3067
    },
    {
      "epoch": 0.14079207012069203,
      "grad_norm": 0.4616551995277405,
      "learning_rate": 9.981389680345468e-06,
      "loss": 0.3898,
      "step": 3068
    },
    {
      "epoch": 0.1408379606259465,
      "grad_norm": 0.4333873987197876,
      "learning_rate": 9.981368539394492e-06,
      "loss": 0.3396,
      "step": 3069
    },
    {
      "epoch": 0.14088385113120094,
      "grad_norm": 0.45387235283851624,
      "learning_rate": 9.981347386464897e-06,
      "loss": 0.3525,
      "step": 3070
    },
    {
      "epoch": 0.1409297416364554,
      "grad_norm": 0.4792275130748749,
      "learning_rate": 9.981326221556734e-06,
      "loss": 0.4041,
      "step": 3071
    },
    {
      "epoch": 0.14097563214170988,
      "grad_norm": 0.4632616639137268,
      "learning_rate": 9.981305044670054e-06,
      "loss": 0.3335,
      "step": 3072
    },
    {
      "epoch": 0.14102152264696435,
      "grad_norm": 0.5189899206161499,
      "learning_rate": 9.981283855804906e-06,
      "loss": 0.4768,
      "step": 3073
    },
    {
      "epoch": 0.1410674131522188,
      "grad_norm": 0.47905048727989197,
      "learning_rate": 9.981262654961343e-06,
      "loss": 0.4456,
      "step": 3074
    },
    {
      "epoch": 0.14111330365747327,
      "grad_norm": 0.4708541929721832,
      "learning_rate": 9.981241442139415e-06,
      "loss": 0.4004,
      "step": 3075
    },
    {
      "epoch": 0.14115919416272774,
      "grad_norm": 0.4787214398384094,
      "learning_rate": 9.981220217339176e-06,
      "loss": 0.426,
      "step": 3076
    },
    {
      "epoch": 0.1412050846679822,
      "grad_norm": 0.4904378652572632,
      "learning_rate": 9.981198980560671e-06,
      "loss": 0.375,
      "step": 3077
    },
    {
      "epoch": 0.14125097517323665,
      "grad_norm": 0.5384431481361389,
      "learning_rate": 9.981177731803956e-06,
      "loss": 0.4761,
      "step": 3078
    },
    {
      "epoch": 0.14129686567849112,
      "grad_norm": 0.47541213035583496,
      "learning_rate": 9.981156471069083e-06,
      "loss": 0.4076,
      "step": 3079
    },
    {
      "epoch": 0.1413427561837456,
      "grad_norm": 0.6860296726226807,
      "learning_rate": 9.981135198356098e-06,
      "loss": 0.466,
      "step": 3080
    },
    {
      "epoch": 0.14138864668900006,
      "grad_norm": 0.4394347369670868,
      "learning_rate": 9.981113913665056e-06,
      "loss": 0.297,
      "step": 3081
    },
    {
      "epoch": 0.1414345371942545,
      "grad_norm": 0.5016234517097473,
      "learning_rate": 9.981092616996007e-06,
      "loss": 0.4169,
      "step": 3082
    },
    {
      "epoch": 0.14148042769950897,
      "grad_norm": 0.4984942078590393,
      "learning_rate": 9.981071308349001e-06,
      "loss": 0.4473,
      "step": 3083
    },
    {
      "epoch": 0.14152631820476344,
      "grad_norm": 0.5313014984130859,
      "learning_rate": 9.981049987724092e-06,
      "loss": 0.524,
      "step": 3084
    },
    {
      "epoch": 0.1415722087100179,
      "grad_norm": 0.4959547221660614,
      "learning_rate": 9.981028655121329e-06,
      "loss": 0.3705,
      "step": 3085
    },
    {
      "epoch": 0.14161809921527235,
      "grad_norm": 0.49455925822257996,
      "learning_rate": 9.981007310540765e-06,
      "loss": 0.3202,
      "step": 3086
    },
    {
      "epoch": 0.14166398972052682,
      "grad_norm": 0.4558359682559967,
      "learning_rate": 9.98098595398245e-06,
      "loss": 0.3861,
      "step": 3087
    },
    {
      "epoch": 0.1417098802257813,
      "grad_norm": 0.5051929950714111,
      "learning_rate": 9.980964585446435e-06,
      "loss": 0.4286,
      "step": 3088
    },
    {
      "epoch": 0.14175577073103574,
      "grad_norm": 0.4991109371185303,
      "learning_rate": 9.980943204932773e-06,
      "loss": 0.4307,
      "step": 3089
    },
    {
      "epoch": 0.1418016612362902,
      "grad_norm": 0.5207083821296692,
      "learning_rate": 9.980921812441515e-06,
      "loss": 0.4935,
      "step": 3090
    },
    {
      "epoch": 0.14184755174154468,
      "grad_norm": 0.4758206307888031,
      "learning_rate": 9.980900407972714e-06,
      "loss": 0.4257,
      "step": 3091
    },
    {
      "epoch": 0.14189344224679915,
      "grad_norm": 0.47634357213974,
      "learning_rate": 9.980878991526417e-06,
      "loss": 0.4223,
      "step": 3092
    },
    {
      "epoch": 0.1419393327520536,
      "grad_norm": 0.6730012893676758,
      "learning_rate": 9.980857563102677e-06,
      "loss": 0.5096,
      "step": 3093
    },
    {
      "epoch": 0.14198522325730806,
      "grad_norm": 0.45978879928588867,
      "learning_rate": 9.98083612270155e-06,
      "loss": 0.3773,
      "step": 3094
    },
    {
      "epoch": 0.14203111376256253,
      "grad_norm": 0.5018622875213623,
      "learning_rate": 9.98081467032308e-06,
      "loss": 0.4277,
      "step": 3095
    },
    {
      "epoch": 0.142077004267817,
      "grad_norm": 0.4870125651359558,
      "learning_rate": 9.980793205967325e-06,
      "loss": 0.3581,
      "step": 3096
    },
    {
      "epoch": 0.14212289477307144,
      "grad_norm": 0.4462220370769501,
      "learning_rate": 9.980771729634335e-06,
      "loss": 0.3673,
      "step": 3097
    },
    {
      "epoch": 0.1421687852783259,
      "grad_norm": 0.4655051529407501,
      "learning_rate": 9.98075024132416e-06,
      "loss": 0.3897,
      "step": 3098
    },
    {
      "epoch": 0.14221467578358038,
      "grad_norm": 0.542564868927002,
      "learning_rate": 9.980728741036853e-06,
      "loss": 0.4938,
      "step": 3099
    },
    {
      "epoch": 0.14226056628883485,
      "grad_norm": 0.4483783543109894,
      "learning_rate": 9.980707228772464e-06,
      "loss": 0.3459,
      "step": 3100
    },
    {
      "epoch": 0.1423064567940893,
      "grad_norm": 0.4641178250312805,
      "learning_rate": 9.980685704531048e-06,
      "loss": 0.3285,
      "step": 3101
    },
    {
      "epoch": 0.14235234729934376,
      "grad_norm": 0.5129979848861694,
      "learning_rate": 9.980664168312654e-06,
      "loss": 0.4006,
      "step": 3102
    },
    {
      "epoch": 0.14239823780459823,
      "grad_norm": 0.4599338471889496,
      "learning_rate": 9.980642620117334e-06,
      "loss": 0.3767,
      "step": 3103
    },
    {
      "epoch": 0.1424441283098527,
      "grad_norm": 0.49075594544410706,
      "learning_rate": 9.980621059945141e-06,
      "loss": 0.457,
      "step": 3104
    },
    {
      "epoch": 0.14249001881510714,
      "grad_norm": 0.48812124133110046,
      "learning_rate": 9.980599487796126e-06,
      "loss": 0.4538,
      "step": 3105
    },
    {
      "epoch": 0.14253590932036161,
      "grad_norm": 0.551496684551239,
      "learning_rate": 9.980577903670342e-06,
      "loss": 0.5404,
      "step": 3106
    },
    {
      "epoch": 0.14258179982561608,
      "grad_norm": 0.46660685539245605,
      "learning_rate": 9.98055630756784e-06,
      "loss": 0.3679,
      "step": 3107
    },
    {
      "epoch": 0.14262769033087055,
      "grad_norm": 0.46349766850471497,
      "learning_rate": 9.980534699488671e-06,
      "loss": 0.425,
      "step": 3108
    },
    {
      "epoch": 0.142673580836125,
      "grad_norm": 0.49127018451690674,
      "learning_rate": 9.980513079432889e-06,
      "loss": 0.4616,
      "step": 3109
    },
    {
      "epoch": 0.14271947134137947,
      "grad_norm": 0.5036889314651489,
      "learning_rate": 9.980491447400544e-06,
      "loss": 0.4524,
      "step": 3110
    },
    {
      "epoch": 0.14276536184663394,
      "grad_norm": 0.49507656693458557,
      "learning_rate": 9.98046980339169e-06,
      "loss": 0.4538,
      "step": 3111
    },
    {
      "epoch": 0.1428112523518884,
      "grad_norm": 0.4613012969493866,
      "learning_rate": 9.980448147406376e-06,
      "loss": 0.4542,
      "step": 3112
    },
    {
      "epoch": 0.14285714285714285,
      "grad_norm": 0.4675821363925934,
      "learning_rate": 9.980426479444658e-06,
      "loss": 0.4006,
      "step": 3113
    },
    {
      "epoch": 0.14290303336239732,
      "grad_norm": 0.43902119994163513,
      "learning_rate": 9.980404799506585e-06,
      "loss": 0.3266,
      "step": 3114
    },
    {
      "epoch": 0.1429489238676518,
      "grad_norm": 0.46223410964012146,
      "learning_rate": 9.98038310759221e-06,
      "loss": 0.3844,
      "step": 3115
    },
    {
      "epoch": 0.14299481437290626,
      "grad_norm": 0.546207070350647,
      "learning_rate": 9.980361403701588e-06,
      "loss": 0.4387,
      "step": 3116
    },
    {
      "epoch": 0.1430407048781607,
      "grad_norm": 0.4951692819595337,
      "learning_rate": 9.980339687834767e-06,
      "loss": 0.4257,
      "step": 3117
    },
    {
      "epoch": 0.14308659538341517,
      "grad_norm": 0.4896540343761444,
      "learning_rate": 9.9803179599918e-06,
      "loss": 0.4065,
      "step": 3118
    },
    {
      "epoch": 0.14313248588866964,
      "grad_norm": 0.45201075077056885,
      "learning_rate": 9.980296220172742e-06,
      "loss": 0.3398,
      "step": 3119
    },
    {
      "epoch": 0.14317837639392408,
      "grad_norm": 0.4665740728378296,
      "learning_rate": 9.980274468377641e-06,
      "loss": 0.3576,
      "step": 3120
    },
    {
      "epoch": 0.14322426689917855,
      "grad_norm": 0.48130103945732117,
      "learning_rate": 9.980252704606555e-06,
      "loss": 0.4094,
      "step": 3121
    },
    {
      "epoch": 0.14327015740443302,
      "grad_norm": 0.44395318627357483,
      "learning_rate": 9.980230928859532e-06,
      "loss": 0.3734,
      "step": 3122
    },
    {
      "epoch": 0.1433160479096875,
      "grad_norm": 0.4439608156681061,
      "learning_rate": 9.980209141136624e-06,
      "loss": 0.3335,
      "step": 3123
    },
    {
      "epoch": 0.14336193841494194,
      "grad_norm": 0.48598894476890564,
      "learning_rate": 9.980187341437886e-06,
      "loss": 0.4086,
      "step": 3124
    },
    {
      "epoch": 0.1434078289201964,
      "grad_norm": 0.4441927373409271,
      "learning_rate": 9.980165529763368e-06,
      "loss": 0.3999,
      "step": 3125
    },
    {
      "epoch": 0.14345371942545088,
      "grad_norm": 0.44016677141189575,
      "learning_rate": 9.980143706113125e-06,
      "loss": 0.3391,
      "step": 3126
    },
    {
      "epoch": 0.14349960993070535,
      "grad_norm": 0.44359922409057617,
      "learning_rate": 9.980121870487207e-06,
      "loss": 0.3427,
      "step": 3127
    },
    {
      "epoch": 0.1435455004359598,
      "grad_norm": 0.4892541170120239,
      "learning_rate": 9.980100022885668e-06,
      "loss": 0.4538,
      "step": 3128
    },
    {
      "epoch": 0.14359139094121426,
      "grad_norm": 0.5039497017860413,
      "learning_rate": 9.980078163308561e-06,
      "loss": 0.4661,
      "step": 3129
    },
    {
      "epoch": 0.14363728144646873,
      "grad_norm": 0.4830937385559082,
      "learning_rate": 9.980056291755937e-06,
      "loss": 0.4725,
      "step": 3130
    },
    {
      "epoch": 0.1436831719517232,
      "grad_norm": 0.43063437938690186,
      "learning_rate": 9.98003440822785e-06,
      "loss": 0.3218,
      "step": 3131
    },
    {
      "epoch": 0.14372906245697764,
      "grad_norm": 0.4961840808391571,
      "learning_rate": 9.980012512724353e-06,
      "loss": 0.4388,
      "step": 3132
    },
    {
      "epoch": 0.1437749529622321,
      "grad_norm": 0.43553978204727173,
      "learning_rate": 9.979990605245495e-06,
      "loss": 0.3254,
      "step": 3133
    },
    {
      "epoch": 0.14382084346748658,
      "grad_norm": 0.4516310691833496,
      "learning_rate": 9.979968685791332e-06,
      "loss": 0.3654,
      "step": 3134
    },
    {
      "epoch": 0.14386673397274105,
      "grad_norm": 0.4878760874271393,
      "learning_rate": 9.979946754361918e-06,
      "loss": 0.4295,
      "step": 3135
    },
    {
      "epoch": 0.1439126244779955,
      "grad_norm": 0.4790392816066742,
      "learning_rate": 9.979924810957302e-06,
      "loss": 0.4528,
      "step": 3136
    },
    {
      "epoch": 0.14395851498324996,
      "grad_norm": 0.5120466947555542,
      "learning_rate": 9.979902855577539e-06,
      "loss": 0.4706,
      "step": 3137
    },
    {
      "epoch": 0.14400440548850443,
      "grad_norm": 0.4738479256629944,
      "learning_rate": 9.979880888222682e-06,
      "loss": 0.3859,
      "step": 3138
    },
    {
      "epoch": 0.1440502959937589,
      "grad_norm": 0.4464550018310547,
      "learning_rate": 9.979858908892782e-06,
      "loss": 0.3758,
      "step": 3139
    },
    {
      "epoch": 0.14409618649901335,
      "grad_norm": 0.5037957429885864,
      "learning_rate": 9.979836917587892e-06,
      "loss": 0.4588,
      "step": 3140
    },
    {
      "epoch": 0.14414207700426782,
      "grad_norm": 0.4527125954627991,
      "learning_rate": 9.979814914308069e-06,
      "loss": 0.3471,
      "step": 3141
    },
    {
      "epoch": 0.14418796750952229,
      "grad_norm": 0.44850632548332214,
      "learning_rate": 9.979792899053361e-06,
      "loss": 0.3834,
      "step": 3142
    },
    {
      "epoch": 0.14423385801477676,
      "grad_norm": 0.4473056495189667,
      "learning_rate": 9.979770871823823e-06,
      "loss": 0.3564,
      "step": 3143
    },
    {
      "epoch": 0.1442797485200312,
      "grad_norm": 0.4869324564933777,
      "learning_rate": 9.979748832619508e-06,
      "loss": 0.4307,
      "step": 3144
    },
    {
      "epoch": 0.14432563902528567,
      "grad_norm": 0.4606243371963501,
      "learning_rate": 9.979726781440469e-06,
      "loss": 0.4164,
      "step": 3145
    },
    {
      "epoch": 0.14437152953054014,
      "grad_norm": 0.503044605255127,
      "learning_rate": 9.979704718286756e-06,
      "loss": 0.4867,
      "step": 3146
    },
    {
      "epoch": 0.14441742003579458,
      "grad_norm": 0.5367587804794312,
      "learning_rate": 9.979682643158428e-06,
      "loss": 0.5154,
      "step": 3147
    },
    {
      "epoch": 0.14446331054104905,
      "grad_norm": 0.48371759057044983,
      "learning_rate": 9.979660556055533e-06,
      "loss": 0.4645,
      "step": 3148
    },
    {
      "epoch": 0.14450920104630352,
      "grad_norm": 0.4573518633842468,
      "learning_rate": 9.979638456978126e-06,
      "loss": 0.4268,
      "step": 3149
    },
    {
      "epoch": 0.144555091551558,
      "grad_norm": 0.514528751373291,
      "learning_rate": 9.979616345926262e-06,
      "loss": 0.4475,
      "step": 3150
    },
    {
      "epoch": 0.14460098205681243,
      "grad_norm": 0.5610572695732117,
      "learning_rate": 9.979594222899992e-06,
      "loss": 0.4988,
      "step": 3151
    },
    {
      "epoch": 0.1446468725620669,
      "grad_norm": 0.4925415515899658,
      "learning_rate": 9.979572087899369e-06,
      "loss": 0.3891,
      "step": 3152
    },
    {
      "epoch": 0.14469276306732137,
      "grad_norm": 0.4756164848804474,
      "learning_rate": 9.979549940924446e-06,
      "loss": 0.369,
      "step": 3153
    },
    {
      "epoch": 0.14473865357257584,
      "grad_norm": 0.47581610083580017,
      "learning_rate": 9.979527781975278e-06,
      "loss": 0.4067,
      "step": 3154
    },
    {
      "epoch": 0.14478454407783029,
      "grad_norm": 0.49081236124038696,
      "learning_rate": 9.979505611051916e-06,
      "loss": 0.3976,
      "step": 3155
    },
    {
      "epoch": 0.14483043458308476,
      "grad_norm": 0.5139479041099548,
      "learning_rate": 9.979483428154414e-06,
      "loss": 0.5148,
      "step": 3156
    },
    {
      "epoch": 0.14487632508833923,
      "grad_norm": 0.47468358278274536,
      "learning_rate": 9.979461233282828e-06,
      "loss": 0.3882,
      "step": 3157
    },
    {
      "epoch": 0.1449222155935937,
      "grad_norm": 0.4881764352321625,
      "learning_rate": 9.97943902643721e-06,
      "loss": 0.4197,
      "step": 3158
    },
    {
      "epoch": 0.14496810609884814,
      "grad_norm": 0.5144550204277039,
      "learning_rate": 9.979416807617611e-06,
      "loss": 0.4181,
      "step": 3159
    },
    {
      "epoch": 0.1450139966041026,
      "grad_norm": 0.42927655577659607,
      "learning_rate": 9.979394576824085e-06,
      "loss": 0.3798,
      "step": 3160
    },
    {
      "epoch": 0.14505988710935708,
      "grad_norm": 0.4579162895679474,
      "learning_rate": 9.979372334056687e-06,
      "loss": 0.3593,
      "step": 3161
    },
    {
      "epoch": 0.14510577761461155,
      "grad_norm": 0.4813615083694458,
      "learning_rate": 9.979350079315472e-06,
      "loss": 0.3868,
      "step": 3162
    },
    {
      "epoch": 0.145151668119866,
      "grad_norm": 0.5369012355804443,
      "learning_rate": 9.97932781260049e-06,
      "loss": 0.4931,
      "step": 3163
    },
    {
      "epoch": 0.14519755862512046,
      "grad_norm": 0.47141051292419434,
      "learning_rate": 9.979305533911797e-06,
      "loss": 0.4274,
      "step": 3164
    },
    {
      "epoch": 0.14524344913037493,
      "grad_norm": 0.48149245977401733,
      "learning_rate": 9.979283243249446e-06,
      "loss": 0.3994,
      "step": 3165
    },
    {
      "epoch": 0.1452893396356294,
      "grad_norm": 0.5012102723121643,
      "learning_rate": 9.97926094061349e-06,
      "loss": 0.4275,
      "step": 3166
    },
    {
      "epoch": 0.14533523014088384,
      "grad_norm": 0.5880632400512695,
      "learning_rate": 9.979238626003982e-06,
      "loss": 0.4164,
      "step": 3167
    },
    {
      "epoch": 0.1453811206461383,
      "grad_norm": 0.4732213616371155,
      "learning_rate": 9.979216299420978e-06,
      "loss": 0.3799,
      "step": 3168
    },
    {
      "epoch": 0.14542701115139278,
      "grad_norm": 0.4795583486557007,
      "learning_rate": 9.979193960864528e-06,
      "loss": 0.4202,
      "step": 3169
    },
    {
      "epoch": 0.14547290165664725,
      "grad_norm": 0.49628451466560364,
      "learning_rate": 9.979171610334691e-06,
      "loss": 0.4166,
      "step": 3170
    },
    {
      "epoch": 0.1455187921619017,
      "grad_norm": 0.468717098236084,
      "learning_rate": 9.979149247831515e-06,
      "loss": 0.3935,
      "step": 3171
    },
    {
      "epoch": 0.14556468266715616,
      "grad_norm": 0.5007694959640503,
      "learning_rate": 9.979126873355059e-06,
      "loss": 0.463,
      "step": 3172
    },
    {
      "epoch": 0.14561057317241063,
      "grad_norm": 0.49003463983535767,
      "learning_rate": 9.979104486905372e-06,
      "loss": 0.4481,
      "step": 3173
    },
    {
      "epoch": 0.1456564636776651,
      "grad_norm": 0.4339561462402344,
      "learning_rate": 9.979082088482512e-06,
      "loss": 0.3073,
      "step": 3174
    },
    {
      "epoch": 0.14570235418291955,
      "grad_norm": 0.5013194680213928,
      "learning_rate": 9.979059678086529e-06,
      "loss": 0.4631,
      "step": 3175
    },
    {
      "epoch": 0.14574824468817402,
      "grad_norm": 0.4873732030391693,
      "learning_rate": 9.979037255717482e-06,
      "loss": 0.4555,
      "step": 3176
    },
    {
      "epoch": 0.1457941351934285,
      "grad_norm": 0.4566386342048645,
      "learning_rate": 9.979014821375419e-06,
      "loss": 0.3723,
      "step": 3177
    },
    {
      "epoch": 0.14584002569868293,
      "grad_norm": 0.4844808876514435,
      "learning_rate": 9.978992375060397e-06,
      "loss": 0.4172,
      "step": 3178
    },
    {
      "epoch": 0.1458859162039374,
      "grad_norm": 0.49305760860443115,
      "learning_rate": 9.978969916772472e-06,
      "loss": 0.4111,
      "step": 3179
    },
    {
      "epoch": 0.14593180670919187,
      "grad_norm": 0.42815062403678894,
      "learning_rate": 9.978947446511692e-06,
      "loss": 0.3426,
      "step": 3180
    },
    {
      "epoch": 0.14597769721444634,
      "grad_norm": 0.5468980669975281,
      "learning_rate": 9.978924964278119e-06,
      "loss": 0.4926,
      "step": 3181
    },
    {
      "epoch": 0.14602358771970078,
      "grad_norm": 0.44816407561302185,
      "learning_rate": 9.9789024700718e-06,
      "loss": 0.3472,
      "step": 3182
    },
    {
      "epoch": 0.14606947822495525,
      "grad_norm": 0.49575862288475037,
      "learning_rate": 9.978879963892794e-06,
      "loss": 0.4554,
      "step": 3183
    },
    {
      "epoch": 0.14611536873020972,
      "grad_norm": 0.4600825011730194,
      "learning_rate": 9.978857445741152e-06,
      "loss": 0.3694,
      "step": 3184
    },
    {
      "epoch": 0.1461612592354642,
      "grad_norm": 0.5010223984718323,
      "learning_rate": 9.978834915616929e-06,
      "loss": 0.4739,
      "step": 3185
    },
    {
      "epoch": 0.14620714974071863,
      "grad_norm": 0.5065903067588806,
      "learning_rate": 9.978812373520179e-06,
      "loss": 0.4736,
      "step": 3186
    },
    {
      "epoch": 0.1462530402459731,
      "grad_norm": 0.47557079792022705,
      "learning_rate": 9.978789819450957e-06,
      "loss": 0.4275,
      "step": 3187
    },
    {
      "epoch": 0.14629893075122757,
      "grad_norm": 0.5065678954124451,
      "learning_rate": 9.978767253409318e-06,
      "loss": 0.4784,
      "step": 3188
    },
    {
      "epoch": 0.14634482125648204,
      "grad_norm": 0.5704526305198669,
      "learning_rate": 9.978744675395314e-06,
      "loss": 0.4233,
      "step": 3189
    },
    {
      "epoch": 0.1463907117617365,
      "grad_norm": 0.5078095197677612,
      "learning_rate": 9.978722085409e-06,
      "loss": 0.3907,
      "step": 3190
    },
    {
      "epoch": 0.14643660226699096,
      "grad_norm": 0.48235201835632324,
      "learning_rate": 9.978699483450432e-06,
      "loss": 0.3985,
      "step": 3191
    },
    {
      "epoch": 0.14648249277224543,
      "grad_norm": 0.49654367566108704,
      "learning_rate": 9.978676869519662e-06,
      "loss": 0.4956,
      "step": 3192
    },
    {
      "epoch": 0.1465283832774999,
      "grad_norm": 0.48536545038223267,
      "learning_rate": 9.978654243616746e-06,
      "loss": 0.4643,
      "step": 3193
    },
    {
      "epoch": 0.14657427378275434,
      "grad_norm": 0.5111007690429688,
      "learning_rate": 9.978631605741737e-06,
      "loss": 0.5157,
      "step": 3194
    },
    {
      "epoch": 0.1466201642880088,
      "grad_norm": 0.445444792509079,
      "learning_rate": 9.97860895589469e-06,
      "loss": 0.314,
      "step": 3195
    },
    {
      "epoch": 0.14666605479326328,
      "grad_norm": 0.47209522128105164,
      "learning_rate": 9.978586294075661e-06,
      "loss": 0.4301,
      "step": 3196
    },
    {
      "epoch": 0.14671194529851775,
      "grad_norm": 0.521162211894989,
      "learning_rate": 9.978563620284704e-06,
      "loss": 0.4361,
      "step": 3197
    },
    {
      "epoch": 0.1467578358037722,
      "grad_norm": 0.4355694651603699,
      "learning_rate": 9.978540934521871e-06,
      "loss": 0.3996,
      "step": 3198
    },
    {
      "epoch": 0.14680372630902666,
      "grad_norm": 0.47809621691703796,
      "learning_rate": 9.978518236787219e-06,
      "loss": 0.4249,
      "step": 3199
    },
    {
      "epoch": 0.14684961681428113,
      "grad_norm": 0.4769158959388733,
      "learning_rate": 9.978495527080803e-06,
      "loss": 0.4532,
      "step": 3200
    },
    {
      "epoch": 0.1468955073195356,
      "grad_norm": 0.47216641902923584,
      "learning_rate": 9.978472805402675e-06,
      "loss": 0.4076,
      "step": 3201
    },
    {
      "epoch": 0.14694139782479004,
      "grad_norm": 0.4564381241798401,
      "learning_rate": 9.97845007175289e-06,
      "loss": 0.41,
      "step": 3202
    },
    {
      "epoch": 0.1469872883300445,
      "grad_norm": 0.5419045090675354,
      "learning_rate": 9.978427326131506e-06,
      "loss": 0.4873,
      "step": 3203
    },
    {
      "epoch": 0.14703317883529898,
      "grad_norm": 0.5343883037567139,
      "learning_rate": 9.978404568538575e-06,
      "loss": 0.424,
      "step": 3204
    },
    {
      "epoch": 0.14707906934055345,
      "grad_norm": 0.4494619369506836,
      "learning_rate": 9.978381798974153e-06,
      "loss": 0.3195,
      "step": 3205
    },
    {
      "epoch": 0.1471249598458079,
      "grad_norm": 0.5076942443847656,
      "learning_rate": 9.978359017438293e-06,
      "loss": 0.4641,
      "step": 3206
    },
    {
      "epoch": 0.14717085035106237,
      "grad_norm": 0.4370076060295105,
      "learning_rate": 9.978336223931051e-06,
      "loss": 0.3307,
      "step": 3207
    },
    {
      "epoch": 0.14721674085631684,
      "grad_norm": 0.5028615593910217,
      "learning_rate": 9.978313418452482e-06,
      "loss": 0.4516,
      "step": 3208
    },
    {
      "epoch": 0.14726263136157128,
      "grad_norm": 0.4506158232688904,
      "learning_rate": 9.978290601002638e-06,
      "loss": 0.3574,
      "step": 3209
    },
    {
      "epoch": 0.14730852186682575,
      "grad_norm": 0.44715920090675354,
      "learning_rate": 9.978267771581579e-06,
      "loss": 0.335,
      "step": 3210
    },
    {
      "epoch": 0.14735441237208022,
      "grad_norm": 0.4608115553855896,
      "learning_rate": 9.978244930189356e-06,
      "loss": 0.339,
      "step": 3211
    },
    {
      "epoch": 0.1474003028773347,
      "grad_norm": 0.5308574438095093,
      "learning_rate": 9.978222076826026e-06,
      "loss": 0.4133,
      "step": 3212
    },
    {
      "epoch": 0.14744619338258913,
      "grad_norm": 0.4709831476211548,
      "learning_rate": 9.978199211491643e-06,
      "loss": 0.4215,
      "step": 3213
    },
    {
      "epoch": 0.1474920838878436,
      "grad_norm": 0.4851463735103607,
      "learning_rate": 9.978176334186261e-06,
      "loss": 0.4392,
      "step": 3214
    },
    {
      "epoch": 0.14753797439309807,
      "grad_norm": 0.43764227628707886,
      "learning_rate": 9.978153444909938e-06,
      "loss": 0.3467,
      "step": 3215
    },
    {
      "epoch": 0.14758386489835254,
      "grad_norm": 0.4525946378707886,
      "learning_rate": 9.978130543662726e-06,
      "loss": 0.4284,
      "step": 3216
    },
    {
      "epoch": 0.14762975540360698,
      "grad_norm": 0.4712614417076111,
      "learning_rate": 9.97810763044468e-06,
      "loss": 0.4149,
      "step": 3217
    },
    {
      "epoch": 0.14767564590886145,
      "grad_norm": 0.46015509963035583,
      "learning_rate": 9.978084705255857e-06,
      "loss": 0.3853,
      "step": 3218
    },
    {
      "epoch": 0.14772153641411592,
      "grad_norm": 0.48580053448677063,
      "learning_rate": 9.978061768096312e-06,
      "loss": 0.3521,
      "step": 3219
    },
    {
      "epoch": 0.1477674269193704,
      "grad_norm": 0.49456503987312317,
      "learning_rate": 9.9780388189661e-06,
      "loss": 0.43,
      "step": 3220
    },
    {
      "epoch": 0.14781331742462483,
      "grad_norm": 0.485686719417572,
      "learning_rate": 9.978015857865275e-06,
      "loss": 0.4268,
      "step": 3221
    },
    {
      "epoch": 0.1478592079298793,
      "grad_norm": 0.4638362228870392,
      "learning_rate": 9.977992884793893e-06,
      "loss": 0.4398,
      "step": 3222
    },
    {
      "epoch": 0.14790509843513377,
      "grad_norm": 0.46629539132118225,
      "learning_rate": 9.977969899752009e-06,
      "loss": 0.4198,
      "step": 3223
    },
    {
      "epoch": 0.14795098894038824,
      "grad_norm": 0.5020266771316528,
      "learning_rate": 9.977946902739679e-06,
      "loss": 0.4318,
      "step": 3224
    },
    {
      "epoch": 0.1479968794456427,
      "grad_norm": 0.48321160674095154,
      "learning_rate": 9.977923893756958e-06,
      "loss": 0.4115,
      "step": 3225
    },
    {
      "epoch": 0.14804276995089716,
      "grad_norm": 0.5496830940246582,
      "learning_rate": 9.9779008728039e-06,
      "loss": 0.4595,
      "step": 3226
    },
    {
      "epoch": 0.14808866045615163,
      "grad_norm": 0.5111088156700134,
      "learning_rate": 9.977877839880562e-06,
      "loss": 0.4378,
      "step": 3227
    },
    {
      "epoch": 0.1481345509614061,
      "grad_norm": 0.5269816517829895,
      "learning_rate": 9.977854794986998e-06,
      "loss": 0.5112,
      "step": 3228
    },
    {
      "epoch": 0.14818044146666054,
      "grad_norm": 0.5188043713569641,
      "learning_rate": 9.977831738123266e-06,
      "loss": 0.4758,
      "step": 3229
    },
    {
      "epoch": 0.148226331971915,
      "grad_norm": 0.510813295841217,
      "learning_rate": 9.977808669289419e-06,
      "loss": 0.4725,
      "step": 3230
    },
    {
      "epoch": 0.14827222247716948,
      "grad_norm": 0.521381676197052,
      "learning_rate": 9.977785588485514e-06,
      "loss": 0.5073,
      "step": 3231
    },
    {
      "epoch": 0.14831811298242395,
      "grad_norm": 0.49391964077949524,
      "learning_rate": 9.977762495711605e-06,
      "loss": 0.4603,
      "step": 3232
    },
    {
      "epoch": 0.1483640034876784,
      "grad_norm": 0.47634458541870117,
      "learning_rate": 9.977739390967746e-06,
      "loss": 0.4183,
      "step": 3233
    },
    {
      "epoch": 0.14840989399293286,
      "grad_norm": 0.4440281093120575,
      "learning_rate": 9.977716274253998e-06,
      "loss": 0.3587,
      "step": 3234
    },
    {
      "epoch": 0.14845578449818733,
      "grad_norm": 0.43542689085006714,
      "learning_rate": 9.977693145570413e-06,
      "loss": 0.3481,
      "step": 3235
    },
    {
      "epoch": 0.14850167500344177,
      "grad_norm": 0.5680211186408997,
      "learning_rate": 9.977670004917046e-06,
      "loss": 0.3766,
      "step": 3236
    },
    {
      "epoch": 0.14854756550869624,
      "grad_norm": 0.6538944244384766,
      "learning_rate": 9.977646852293954e-06,
      "loss": 0.4768,
      "step": 3237
    },
    {
      "epoch": 0.14859345601395071,
      "grad_norm": 0.4630160331726074,
      "learning_rate": 9.977623687701192e-06,
      "loss": 0.4119,
      "step": 3238
    },
    {
      "epoch": 0.14863934651920518,
      "grad_norm": 0.45330503582954407,
      "learning_rate": 9.977600511138816e-06,
      "loss": 0.3413,
      "step": 3239
    },
    {
      "epoch": 0.14868523702445963,
      "grad_norm": 0.4710831642150879,
      "learning_rate": 9.977577322606882e-06,
      "loss": 0.3961,
      "step": 3240
    },
    {
      "epoch": 0.1487311275297141,
      "grad_norm": 0.48167937994003296,
      "learning_rate": 9.977554122105446e-06,
      "loss": 0.4179,
      "step": 3241
    },
    {
      "epoch": 0.14877701803496857,
      "grad_norm": 0.5230624079704285,
      "learning_rate": 9.977530909634561e-06,
      "loss": 0.4514,
      "step": 3242
    },
    {
      "epoch": 0.14882290854022304,
      "grad_norm": 0.47600480914115906,
      "learning_rate": 9.977507685194287e-06,
      "loss": 0.3741,
      "step": 3243
    },
    {
      "epoch": 0.14886879904547748,
      "grad_norm": 0.46319711208343506,
      "learning_rate": 9.97748444878468e-06,
      "loss": 0.3849,
      "step": 3244
    },
    {
      "epoch": 0.14891468955073195,
      "grad_norm": 0.46935534477233887,
      "learning_rate": 9.97746120040579e-06,
      "loss": 0.407,
      "step": 3245
    },
    {
      "epoch": 0.14896058005598642,
      "grad_norm": 0.47236064076423645,
      "learning_rate": 9.97743794005768e-06,
      "loss": 0.4032,
      "step": 3246
    },
    {
      "epoch": 0.1490064705612409,
      "grad_norm": 0.49963274598121643,
      "learning_rate": 9.9774146677404e-06,
      "loss": 0.4263,
      "step": 3247
    },
    {
      "epoch": 0.14905236106649533,
      "grad_norm": 0.515075147151947,
      "learning_rate": 9.97739138345401e-06,
      "loss": 0.4636,
      "step": 3248
    },
    {
      "epoch": 0.1490982515717498,
      "grad_norm": 0.4450444281101227,
      "learning_rate": 9.977368087198565e-06,
      "loss": 0.4166,
      "step": 3249
    },
    {
      "epoch": 0.14914414207700427,
      "grad_norm": 0.4645858108997345,
      "learning_rate": 9.977344778974118e-06,
      "loss": 0.397,
      "step": 3250
    },
    {
      "epoch": 0.14919003258225874,
      "grad_norm": 0.4821338653564453,
      "learning_rate": 9.977321458780731e-06,
      "loss": 0.3795,
      "step": 3251
    },
    {
      "epoch": 0.14923592308751318,
      "grad_norm": 0.47550302743911743,
      "learning_rate": 9.977298126618456e-06,
      "loss": 0.4161,
      "step": 3252
    },
    {
      "epoch": 0.14928181359276765,
      "grad_norm": 0.5013076663017273,
      "learning_rate": 9.977274782487348e-06,
      "loss": 0.4643,
      "step": 3253
    },
    {
      "epoch": 0.14932770409802212,
      "grad_norm": 0.42847785353660583,
      "learning_rate": 9.977251426387466e-06,
      "loss": 0.3158,
      "step": 3254
    },
    {
      "epoch": 0.1493735946032766,
      "grad_norm": 0.49158450961112976,
      "learning_rate": 9.977228058318867e-06,
      "loss": 0.4821,
      "step": 3255
    },
    {
      "epoch": 0.14941948510853104,
      "grad_norm": 0.5497332215309143,
      "learning_rate": 9.977204678281603e-06,
      "loss": 0.4187,
      "step": 3256
    },
    {
      "epoch": 0.1494653756137855,
      "grad_norm": 0.49648115038871765,
      "learning_rate": 9.977181286275735e-06,
      "loss": 0.3362,
      "step": 3257
    },
    {
      "epoch": 0.14951126611903998,
      "grad_norm": 0.5207563042640686,
      "learning_rate": 9.977157882301314e-06,
      "loss": 0.5135,
      "step": 3258
    },
    {
      "epoch": 0.14955715662429445,
      "grad_norm": 0.49305441975593567,
      "learning_rate": 9.9771344663584e-06,
      "loss": 0.3643,
      "step": 3259
    },
    {
      "epoch": 0.1496030471295489,
      "grad_norm": 0.45614635944366455,
      "learning_rate": 9.977111038447049e-06,
      "loss": 0.4039,
      "step": 3260
    },
    {
      "epoch": 0.14964893763480336,
      "grad_norm": 0.4625093340873718,
      "learning_rate": 9.977087598567317e-06,
      "loss": 0.3974,
      "step": 3261
    },
    {
      "epoch": 0.14969482814005783,
      "grad_norm": 0.43722766637802124,
      "learning_rate": 9.97706414671926e-06,
      "loss": 0.3446,
      "step": 3262
    },
    {
      "epoch": 0.1497407186453123,
      "grad_norm": 0.5063542127609253,
      "learning_rate": 9.977040682902935e-06,
      "loss": 0.4738,
      "step": 3263
    },
    {
      "epoch": 0.14978660915056674,
      "grad_norm": 0.47105082869529724,
      "learning_rate": 9.977017207118398e-06,
      "loss": 0.3969,
      "step": 3264
    },
    {
      "epoch": 0.1498324996558212,
      "grad_norm": 0.4708927571773529,
      "learning_rate": 9.976993719365704e-06,
      "loss": 0.3844,
      "step": 3265
    },
    {
      "epoch": 0.14987839016107568,
      "grad_norm": 0.48262327909469604,
      "learning_rate": 9.976970219644912e-06,
      "loss": 0.4316,
      "step": 3266
    },
    {
      "epoch": 0.14992428066633012,
      "grad_norm": 0.453774631023407,
      "learning_rate": 9.976946707956077e-06,
      "loss": 0.3611,
      "step": 3267
    },
    {
      "epoch": 0.1499701711715846,
      "grad_norm": 0.5171927809715271,
      "learning_rate": 9.976923184299257e-06,
      "loss": 0.4728,
      "step": 3268
    },
    {
      "epoch": 0.15001606167683906,
      "grad_norm": 0.41328203678131104,
      "learning_rate": 9.976899648674506e-06,
      "loss": 0.3226,
      "step": 3269
    },
    {
      "epoch": 0.15006195218209353,
      "grad_norm": 0.4921911954879761,
      "learning_rate": 9.976876101081883e-06,
      "loss": 0.4472,
      "step": 3270
    },
    {
      "epoch": 0.15010784268734798,
      "grad_norm": 0.47238001227378845,
      "learning_rate": 9.976852541521443e-06,
      "loss": 0.3968,
      "step": 3271
    },
    {
      "epoch": 0.15015373319260245,
      "grad_norm": 0.4779175817966461,
      "learning_rate": 9.976828969993243e-06,
      "loss": 0.441,
      "step": 3272
    },
    {
      "epoch": 0.15019962369785692,
      "grad_norm": 0.5207765698432922,
      "learning_rate": 9.976805386497342e-06,
      "loss": 0.4787,
      "step": 3273
    },
    {
      "epoch": 0.15024551420311139,
      "grad_norm": 0.4468748867511749,
      "learning_rate": 9.976781791033792e-06,
      "loss": 0.3815,
      "step": 3274
    },
    {
      "epoch": 0.15029140470836583,
      "grad_norm": 0.4599221646785736,
      "learning_rate": 9.976758183602656e-06,
      "loss": 0.4145,
      "step": 3275
    },
    {
      "epoch": 0.1503372952136203,
      "grad_norm": 0.5256581902503967,
      "learning_rate": 9.976734564203984e-06,
      "loss": 0.4799,
      "step": 3276
    },
    {
      "epoch": 0.15038318571887477,
      "grad_norm": 0.49408283829689026,
      "learning_rate": 9.976710932837838e-06,
      "loss": 0.4687,
      "step": 3277
    },
    {
      "epoch": 0.15042907622412924,
      "grad_norm": 0.5246826410293579,
      "learning_rate": 9.976687289504272e-06,
      "loss": 0.5185,
      "step": 3278
    },
    {
      "epoch": 0.15047496672938368,
      "grad_norm": 0.47444865107536316,
      "learning_rate": 9.976663634203344e-06,
      "loss": 0.3545,
      "step": 3279
    },
    {
      "epoch": 0.15052085723463815,
      "grad_norm": 0.4859449863433838,
      "learning_rate": 9.976639966935111e-06,
      "loss": 0.4716,
      "step": 3280
    },
    {
      "epoch": 0.15056674773989262,
      "grad_norm": 0.4734502136707306,
      "learning_rate": 9.976616287699629e-06,
      "loss": 0.382,
      "step": 3281
    },
    {
      "epoch": 0.1506126382451471,
      "grad_norm": 0.4478846490383148,
      "learning_rate": 9.976592596496956e-06,
      "loss": 0.3778,
      "step": 3282
    },
    {
      "epoch": 0.15065852875040153,
      "grad_norm": 0.5079098343849182,
      "learning_rate": 9.976568893327148e-06,
      "loss": 0.3747,
      "step": 3283
    },
    {
      "epoch": 0.150704419255656,
      "grad_norm": 0.46207308769226074,
      "learning_rate": 9.976545178190262e-06,
      "loss": 0.4172,
      "step": 3284
    },
    {
      "epoch": 0.15075030976091047,
      "grad_norm": 0.49411800503730774,
      "learning_rate": 9.976521451086357e-06,
      "loss": 0.4661,
      "step": 3285
    },
    {
      "epoch": 0.15079620026616494,
      "grad_norm": 0.4663957953453064,
      "learning_rate": 9.976497712015487e-06,
      "loss": 0.429,
      "step": 3286
    },
    {
      "epoch": 0.15084209077141938,
      "grad_norm": 0.5326253175735474,
      "learning_rate": 9.976473960977711e-06,
      "loss": 0.5189,
      "step": 3287
    },
    {
      "epoch": 0.15088798127667385,
      "grad_norm": 0.43387115001678467,
      "learning_rate": 9.976450197973087e-06,
      "loss": 0.3104,
      "step": 3288
    },
    {
      "epoch": 0.15093387178192832,
      "grad_norm": 0.4897501468658447,
      "learning_rate": 9.97642642300167e-06,
      "loss": 0.4416,
      "step": 3289
    },
    {
      "epoch": 0.1509797622871828,
      "grad_norm": 0.46450138092041016,
      "learning_rate": 9.976402636063519e-06,
      "loss": 0.3624,
      "step": 3290
    },
    {
      "epoch": 0.15102565279243724,
      "grad_norm": 0.4670571982860565,
      "learning_rate": 9.97637883715869e-06,
      "loss": 0.3467,
      "step": 3291
    },
    {
      "epoch": 0.1510715432976917,
      "grad_norm": 0.5098255276679993,
      "learning_rate": 9.976355026287237e-06,
      "loss": 0.4609,
      "step": 3292
    },
    {
      "epoch": 0.15111743380294618,
      "grad_norm": 0.4938158392906189,
      "learning_rate": 9.976331203449224e-06,
      "loss": 0.437,
      "step": 3293
    },
    {
      "epoch": 0.15116332430820062,
      "grad_norm": 0.5252211093902588,
      "learning_rate": 9.976307368644704e-06,
      "loss": 0.5031,
      "step": 3294
    },
    {
      "epoch": 0.1512092148134551,
      "grad_norm": 0.4526505470275879,
      "learning_rate": 9.976283521873738e-06,
      "loss": 0.3591,
      "step": 3295
    },
    {
      "epoch": 0.15125510531870956,
      "grad_norm": 0.4161567687988281,
      "learning_rate": 9.976259663136378e-06,
      "loss": 0.3301,
      "step": 3296
    },
    {
      "epoch": 0.15130099582396403,
      "grad_norm": 0.4712582528591156,
      "learning_rate": 9.976235792432685e-06,
      "loss": 0.42,
      "step": 3297
    },
    {
      "epoch": 0.15134688632921847,
      "grad_norm": 0.42962464690208435,
      "learning_rate": 9.976211909762714e-06,
      "loss": 0.3086,
      "step": 3298
    },
    {
      "epoch": 0.15139277683447294,
      "grad_norm": 0.43894869089126587,
      "learning_rate": 9.976188015126526e-06,
      "loss": 0.3597,
      "step": 3299
    },
    {
      "epoch": 0.1514386673397274,
      "grad_norm": 0.4999287724494934,
      "learning_rate": 9.976164108524175e-06,
      "loss": 0.4506,
      "step": 3300
    },
    {
      "epoch": 0.15148455784498188,
      "grad_norm": 0.48164981603622437,
      "learning_rate": 9.97614018995572e-06,
      "loss": 0.4152,
      "step": 3301
    },
    {
      "epoch": 0.15153044835023632,
      "grad_norm": 0.4818864166736603,
      "learning_rate": 9.976116259421218e-06,
      "loss": 0.398,
      "step": 3302
    },
    {
      "epoch": 0.1515763388554908,
      "grad_norm": 0.4519684314727783,
      "learning_rate": 9.976092316920729e-06,
      "loss": 0.3498,
      "step": 3303
    },
    {
      "epoch": 0.15162222936074526,
      "grad_norm": 0.48560863733291626,
      "learning_rate": 9.976068362454306e-06,
      "loss": 0.4594,
      "step": 3304
    },
    {
      "epoch": 0.15166811986599973,
      "grad_norm": 0.476107656955719,
      "learning_rate": 9.97604439602201e-06,
      "loss": 0.4122,
      "step": 3305
    },
    {
      "epoch": 0.15171401037125418,
      "grad_norm": 0.46209830045700073,
      "learning_rate": 9.976020417623898e-06,
      "loss": 0.3688,
      "step": 3306
    },
    {
      "epoch": 0.15175990087650865,
      "grad_norm": 0.4434915781021118,
      "learning_rate": 9.975996427260027e-06,
      "loss": 0.3113,
      "step": 3307
    },
    {
      "epoch": 0.15180579138176312,
      "grad_norm": 0.4564756453037262,
      "learning_rate": 9.975972424930455e-06,
      "loss": 0.3615,
      "step": 3308
    },
    {
      "epoch": 0.1518516818870176,
      "grad_norm": 0.46090400218963623,
      "learning_rate": 9.97594841063524e-06,
      "loss": 0.3829,
      "step": 3309
    },
    {
      "epoch": 0.15189757239227203,
      "grad_norm": 0.419474720954895,
      "learning_rate": 9.97592438437444e-06,
      "loss": 0.3067,
      "step": 3310
    },
    {
      "epoch": 0.1519434628975265,
      "grad_norm": 0.5025521516799927,
      "learning_rate": 9.97590034614811e-06,
      "loss": 0.5174,
      "step": 3311
    },
    {
      "epoch": 0.15198935340278097,
      "grad_norm": 0.4800359010696411,
      "learning_rate": 9.975876295956313e-06,
      "loss": 0.4037,
      "step": 3312
    },
    {
      "epoch": 0.15203524390803544,
      "grad_norm": 0.4942193031311035,
      "learning_rate": 9.975852233799104e-06,
      "loss": 0.4936,
      "step": 3313
    },
    {
      "epoch": 0.15208113441328988,
      "grad_norm": 0.4991069436073303,
      "learning_rate": 9.97582815967654e-06,
      "loss": 0.4341,
      "step": 3314
    },
    {
      "epoch": 0.15212702491854435,
      "grad_norm": 0.468336284160614,
      "learning_rate": 9.975804073588677e-06,
      "loss": 0.4085,
      "step": 3315
    },
    {
      "epoch": 0.15217291542379882,
      "grad_norm": 0.6517894864082336,
      "learning_rate": 9.975779975535578e-06,
      "loss": 0.4395,
      "step": 3316
    },
    {
      "epoch": 0.1522188059290533,
      "grad_norm": 0.47369495034217834,
      "learning_rate": 9.975755865517299e-06,
      "loss": 0.4105,
      "step": 3317
    },
    {
      "epoch": 0.15226469643430773,
      "grad_norm": 0.4783988893032074,
      "learning_rate": 9.975731743533898e-06,
      "loss": 0.4524,
      "step": 3318
    },
    {
      "epoch": 0.1523105869395622,
      "grad_norm": 0.5697835087776184,
      "learning_rate": 9.97570760958543e-06,
      "loss": 0.4298,
      "step": 3319
    },
    {
      "epoch": 0.15235647744481667,
      "grad_norm": 0.46827632188796997,
      "learning_rate": 9.97568346367196e-06,
      "loss": 0.3887,
      "step": 3320
    },
    {
      "epoch": 0.15240236795007114,
      "grad_norm": 0.4631456434726715,
      "learning_rate": 9.975659305793537e-06,
      "loss": 0.3773,
      "step": 3321
    },
    {
      "epoch": 0.15244825845532559,
      "grad_norm": 0.5249735116958618,
      "learning_rate": 9.975635135950225e-06,
      "loss": 0.5099,
      "step": 3322
    },
    {
      "epoch": 0.15249414896058006,
      "grad_norm": 0.4569174647331238,
      "learning_rate": 9.975610954142082e-06,
      "loss": 0.3858,
      "step": 3323
    },
    {
      "epoch": 0.15254003946583453,
      "grad_norm": 0.46604296565055847,
      "learning_rate": 9.975586760369164e-06,
      "loss": 0.429,
      "step": 3324
    },
    {
      "epoch": 0.15258592997108897,
      "grad_norm": 0.4511224925518036,
      "learning_rate": 9.97556255463153e-06,
      "loss": 0.3165,
      "step": 3325
    },
    {
      "epoch": 0.15263182047634344,
      "grad_norm": 0.47166356444358826,
      "learning_rate": 9.97553833692924e-06,
      "loss": 0.4136,
      "step": 3326
    },
    {
      "epoch": 0.1526777109815979,
      "grad_norm": 0.48687249422073364,
      "learning_rate": 9.97551410726235e-06,
      "loss": 0.4023,
      "step": 3327
    },
    {
      "epoch": 0.15272360148685238,
      "grad_norm": 0.5155811905860901,
      "learning_rate": 9.975489865630919e-06,
      "loss": 0.5007,
      "step": 3328
    },
    {
      "epoch": 0.15276949199210682,
      "grad_norm": 0.46205323934555054,
      "learning_rate": 9.975465612035003e-06,
      "loss": 0.4261,
      "step": 3329
    },
    {
      "epoch": 0.1528153824973613,
      "grad_norm": 0.44857993721961975,
      "learning_rate": 9.975441346474665e-06,
      "loss": 0.345,
      "step": 3330
    },
    {
      "epoch": 0.15286127300261576,
      "grad_norm": 0.4689047336578369,
      "learning_rate": 9.97541706894996e-06,
      "loss": 0.3974,
      "step": 3331
    },
    {
      "epoch": 0.15290716350787023,
      "grad_norm": 0.5168011784553528,
      "learning_rate": 9.975392779460947e-06,
      "loss": 0.5219,
      "step": 3332
    },
    {
      "epoch": 0.15295305401312467,
      "grad_norm": 0.5148208141326904,
      "learning_rate": 9.975368478007686e-06,
      "loss": 0.5021,
      "step": 3333
    },
    {
      "epoch": 0.15299894451837914,
      "grad_norm": 0.46601489186286926,
      "learning_rate": 9.975344164590233e-06,
      "loss": 0.3969,
      "step": 3334
    },
    {
      "epoch": 0.1530448350236336,
      "grad_norm": 0.5498623251914978,
      "learning_rate": 9.975319839208646e-06,
      "loss": 0.4889,
      "step": 3335
    },
    {
      "epoch": 0.15309072552888808,
      "grad_norm": 0.47389283776283264,
      "learning_rate": 9.975295501862987e-06,
      "loss": 0.3962,
      "step": 3336
    },
    {
      "epoch": 0.15313661603414253,
      "grad_norm": 0.42044904828071594,
      "learning_rate": 9.975271152553312e-06,
      "loss": 0.2977,
      "step": 3337
    },
    {
      "epoch": 0.153182506539397,
      "grad_norm": 0.4859681725502014,
      "learning_rate": 9.97524679127968e-06,
      "loss": 0.4673,
      "step": 3338
    },
    {
      "epoch": 0.15322839704465147,
      "grad_norm": 0.50887131690979,
      "learning_rate": 9.97522241804215e-06,
      "loss": 0.4669,
      "step": 3339
    },
    {
      "epoch": 0.15327428754990594,
      "grad_norm": 0.48404571413993835,
      "learning_rate": 9.97519803284078e-06,
      "loss": 0.4461,
      "step": 3340
    },
    {
      "epoch": 0.15332017805516038,
      "grad_norm": 0.4372429847717285,
      "learning_rate": 9.97517363567563e-06,
      "loss": 0.3436,
      "step": 3341
    },
    {
      "epoch": 0.15336606856041485,
      "grad_norm": 0.48688292503356934,
      "learning_rate": 9.975149226546756e-06,
      "loss": 0.4416,
      "step": 3342
    },
    {
      "epoch": 0.15341195906566932,
      "grad_norm": 0.4545394480228424,
      "learning_rate": 9.975124805454219e-06,
      "loss": 0.3806,
      "step": 3343
    },
    {
      "epoch": 0.1534578495709238,
      "grad_norm": 0.437548965215683,
      "learning_rate": 9.975100372398078e-06,
      "loss": 0.3484,
      "step": 3344
    },
    {
      "epoch": 0.15350374007617823,
      "grad_norm": 0.43340831995010376,
      "learning_rate": 9.97507592737839e-06,
      "loss": 0.2852,
      "step": 3345
    },
    {
      "epoch": 0.1535496305814327,
      "grad_norm": 0.49539417028427124,
      "learning_rate": 9.975051470395214e-06,
      "loss": 0.4105,
      "step": 3346
    },
    {
      "epoch": 0.15359552108668717,
      "grad_norm": 0.5326932072639465,
      "learning_rate": 9.97502700144861e-06,
      "loss": 0.5791,
      "step": 3347
    },
    {
      "epoch": 0.15364141159194164,
      "grad_norm": 0.5001088380813599,
      "learning_rate": 9.975002520538637e-06,
      "loss": 0.5075,
      "step": 3348
    },
    {
      "epoch": 0.15368730209719608,
      "grad_norm": 0.5102899074554443,
      "learning_rate": 9.974978027665351e-06,
      "loss": 0.4281,
      "step": 3349
    },
    {
      "epoch": 0.15373319260245055,
      "grad_norm": 0.4571594297885895,
      "learning_rate": 9.974953522828815e-06,
      "loss": 0.4102,
      "step": 3350
    },
    {
      "epoch": 0.15377908310770502,
      "grad_norm": 0.4282696545124054,
      "learning_rate": 9.974929006029087e-06,
      "loss": 0.3064,
      "step": 3351
    },
    {
      "epoch": 0.1538249736129595,
      "grad_norm": 0.5148342847824097,
      "learning_rate": 9.974904477266222e-06,
      "loss": 0.4558,
      "step": 3352
    },
    {
      "epoch": 0.15387086411821393,
      "grad_norm": 0.468266099691391,
      "learning_rate": 9.974879936540283e-06,
      "loss": 0.362,
      "step": 3353
    },
    {
      "epoch": 0.1539167546234684,
      "grad_norm": 0.4335395395755768,
      "learning_rate": 9.974855383851327e-06,
      "loss": 0.3431,
      "step": 3354
    },
    {
      "epoch": 0.15396264512872287,
      "grad_norm": 0.4697313606739044,
      "learning_rate": 9.974830819199414e-06,
      "loss": 0.4196,
      "step": 3355
    },
    {
      "epoch": 0.15400853563397732,
      "grad_norm": 0.4735826551914215,
      "learning_rate": 9.974806242584605e-06,
      "loss": 0.3677,
      "step": 3356
    },
    {
      "epoch": 0.1540544261392318,
      "grad_norm": 0.5247685313224792,
      "learning_rate": 9.974781654006955e-06,
      "loss": 0.5134,
      "step": 3357
    },
    {
      "epoch": 0.15410031664448626,
      "grad_norm": 0.5189568400382996,
      "learning_rate": 9.974757053466526e-06,
      "loss": 0.5135,
      "step": 3358
    },
    {
      "epoch": 0.15414620714974073,
      "grad_norm": 0.46164682507514954,
      "learning_rate": 9.974732440963377e-06,
      "loss": 0.377,
      "step": 3359
    },
    {
      "epoch": 0.15419209765499517,
      "grad_norm": 0.44032758474349976,
      "learning_rate": 9.974707816497566e-06,
      "loss": 0.3886,
      "step": 3360
    },
    {
      "epoch": 0.15423798816024964,
      "grad_norm": 0.49407023191452026,
      "learning_rate": 9.974683180069152e-06,
      "loss": 0.4566,
      "step": 3361
    },
    {
      "epoch": 0.1542838786655041,
      "grad_norm": 0.5214409828186035,
      "learning_rate": 9.974658531678196e-06,
      "loss": 0.529,
      "step": 3362
    },
    {
      "epoch": 0.15432976917075858,
      "grad_norm": 0.41360002756118774,
      "learning_rate": 9.974633871324754e-06,
      "loss": 0.31,
      "step": 3363
    },
    {
      "epoch": 0.15437565967601302,
      "grad_norm": 0.506309986114502,
      "learning_rate": 9.97460919900889e-06,
      "loss": 0.4304,
      "step": 3364
    },
    {
      "epoch": 0.1544215501812675,
      "grad_norm": 0.467813640832901,
      "learning_rate": 9.97458451473066e-06,
      "loss": 0.3867,
      "step": 3365
    },
    {
      "epoch": 0.15446744068652196,
      "grad_norm": 0.46149975061416626,
      "learning_rate": 9.974559818490126e-06,
      "loss": 0.346,
      "step": 3366
    },
    {
      "epoch": 0.15451333119177643,
      "grad_norm": 0.4611506760120392,
      "learning_rate": 9.974535110287344e-06,
      "loss": 0.3927,
      "step": 3367
    },
    {
      "epoch": 0.15455922169703087,
      "grad_norm": 0.45857664942741394,
      "learning_rate": 9.974510390122376e-06,
      "loss": 0.3892,
      "step": 3368
    },
    {
      "epoch": 0.15460511220228534,
      "grad_norm": 0.4210169017314911,
      "learning_rate": 9.97448565799528e-06,
      "loss": 0.3153,
      "step": 3369
    },
    {
      "epoch": 0.15465100270753981,
      "grad_norm": 0.4962354004383087,
      "learning_rate": 9.974460913906115e-06,
      "loss": 0.4937,
      "step": 3370
    },
    {
      "epoch": 0.15469689321279428,
      "grad_norm": 0.4946744441986084,
      "learning_rate": 9.974436157854943e-06,
      "loss": 0.4492,
      "step": 3371
    },
    {
      "epoch": 0.15474278371804873,
      "grad_norm": 0.5004401206970215,
      "learning_rate": 9.974411389841822e-06,
      "loss": 0.5147,
      "step": 3372
    },
    {
      "epoch": 0.1547886742233032,
      "grad_norm": 0.44354549050331116,
      "learning_rate": 9.974386609866811e-06,
      "loss": 0.3313,
      "step": 3373
    },
    {
      "epoch": 0.15483456472855767,
      "grad_norm": 0.4852031469345093,
      "learning_rate": 9.97436181792997e-06,
      "loss": 0.4147,
      "step": 3374
    },
    {
      "epoch": 0.15488045523381214,
      "grad_norm": 0.45381179451942444,
      "learning_rate": 9.974337014031358e-06,
      "loss": 0.3414,
      "step": 3375
    },
    {
      "epoch": 0.15492634573906658,
      "grad_norm": 0.49262282252311707,
      "learning_rate": 9.974312198171037e-06,
      "loss": 0.4475,
      "step": 3376
    },
    {
      "epoch": 0.15497223624432105,
      "grad_norm": 0.5245398283004761,
      "learning_rate": 9.974287370349066e-06,
      "loss": 0.5369,
      "step": 3377
    },
    {
      "epoch": 0.15501812674957552,
      "grad_norm": 0.5146275162696838,
      "learning_rate": 9.9742625305655e-06,
      "loss": 0.4949,
      "step": 3378
    },
    {
      "epoch": 0.15506401725483,
      "grad_norm": 0.47069868445396423,
      "learning_rate": 9.974237678820405e-06,
      "loss": 0.3461,
      "step": 3379
    },
    {
      "epoch": 0.15510990776008443,
      "grad_norm": 0.4811304211616516,
      "learning_rate": 9.974212815113838e-06,
      "loss": 0.3857,
      "step": 3380
    },
    {
      "epoch": 0.1551557982653389,
      "grad_norm": 0.4665345847606659,
      "learning_rate": 9.97418793944586e-06,
      "loss": 0.335,
      "step": 3381
    },
    {
      "epoch": 0.15520168877059337,
      "grad_norm": 0.48886385560035706,
      "learning_rate": 9.97416305181653e-06,
      "loss": 0.4225,
      "step": 3382
    },
    {
      "epoch": 0.1552475792758478,
      "grad_norm": 0.4920494556427002,
      "learning_rate": 9.974138152225906e-06,
      "loss": 0.3834,
      "step": 3383
    },
    {
      "epoch": 0.15529346978110228,
      "grad_norm": 0.5018653869628906,
      "learning_rate": 9.97411324067405e-06,
      "loss": 0.4707,
      "step": 3384
    },
    {
      "epoch": 0.15533936028635675,
      "grad_norm": 0.46046924591064453,
      "learning_rate": 9.974088317161022e-06,
      "loss": 0.3816,
      "step": 3385
    },
    {
      "epoch": 0.15538525079161122,
      "grad_norm": 0.471876323223114,
      "learning_rate": 9.974063381686882e-06,
      "loss": 0.393,
      "step": 3386
    },
    {
      "epoch": 0.15543114129686567,
      "grad_norm": 0.5284046530723572,
      "learning_rate": 9.974038434251688e-06,
      "loss": 0.4763,
      "step": 3387
    },
    {
      "epoch": 0.15547703180212014,
      "grad_norm": 0.45570287108421326,
      "learning_rate": 9.974013474855502e-06,
      "loss": 0.4046,
      "step": 3388
    },
    {
      "epoch": 0.1555229223073746,
      "grad_norm": 0.5144028067588806,
      "learning_rate": 9.973988503498385e-06,
      "loss": 0.3695,
      "step": 3389
    },
    {
      "epoch": 0.15556881281262908,
      "grad_norm": 0.5176663398742676,
      "learning_rate": 9.973963520180394e-06,
      "loss": 0.4461,
      "step": 3390
    },
    {
      "epoch": 0.15561470331788352,
      "grad_norm": 0.4735722839832306,
      "learning_rate": 9.97393852490159e-06,
      "loss": 0.38,
      "step": 3391
    },
    {
      "epoch": 0.155660593823138,
      "grad_norm": 0.5099039673805237,
      "learning_rate": 9.973913517662034e-06,
      "loss": 0.486,
      "step": 3392
    },
    {
      "epoch": 0.15570648432839246,
      "grad_norm": 0.4798545241355896,
      "learning_rate": 9.973888498461786e-06,
      "loss": 0.3874,
      "step": 3393
    },
    {
      "epoch": 0.15575237483364693,
      "grad_norm": 0.4520500600337982,
      "learning_rate": 9.973863467300908e-06,
      "loss": 0.3966,
      "step": 3394
    },
    {
      "epoch": 0.15579826533890137,
      "grad_norm": 0.526831328868866,
      "learning_rate": 9.973838424179456e-06,
      "loss": 0.4817,
      "step": 3395
    },
    {
      "epoch": 0.15584415584415584,
      "grad_norm": 0.45720869302749634,
      "learning_rate": 9.973813369097493e-06,
      "loss": 0.4067,
      "step": 3396
    },
    {
      "epoch": 0.1558900463494103,
      "grad_norm": 0.45186010003089905,
      "learning_rate": 9.973788302055078e-06,
      "loss": 0.3877,
      "step": 3397
    },
    {
      "epoch": 0.15593593685466478,
      "grad_norm": 0.4824221730232239,
      "learning_rate": 9.973763223052272e-06,
      "loss": 0.4087,
      "step": 3398
    },
    {
      "epoch": 0.15598182735991922,
      "grad_norm": 0.47514066100120544,
      "learning_rate": 9.973738132089135e-06,
      "loss": 0.4592,
      "step": 3399
    },
    {
      "epoch": 0.1560277178651737,
      "grad_norm": 0.5052584409713745,
      "learning_rate": 9.973713029165728e-06,
      "loss": 0.3943,
      "step": 3400
    },
    {
      "epoch": 0.15607360837042816,
      "grad_norm": 0.4857885241508484,
      "learning_rate": 9.97368791428211e-06,
      "loss": 0.4136,
      "step": 3401
    },
    {
      "epoch": 0.15611949887568263,
      "grad_norm": 0.544730007648468,
      "learning_rate": 9.973662787438345e-06,
      "loss": 0.4757,
      "step": 3402
    },
    {
      "epoch": 0.15616538938093708,
      "grad_norm": 0.49583789706230164,
      "learning_rate": 9.973637648634489e-06,
      "loss": 0.4761,
      "step": 3403
    },
    {
      "epoch": 0.15621127988619155,
      "grad_norm": 0.4893946349620819,
      "learning_rate": 9.973612497870605e-06,
      "loss": 0.4811,
      "step": 3404
    },
    {
      "epoch": 0.15625717039144602,
      "grad_norm": 0.4920247793197632,
      "learning_rate": 9.973587335146751e-06,
      "loss": 0.4511,
      "step": 3405
    },
    {
      "epoch": 0.15630306089670049,
      "grad_norm": 0.511658787727356,
      "learning_rate": 9.973562160462989e-06,
      "loss": 0.4338,
      "step": 3406
    },
    {
      "epoch": 0.15634895140195493,
      "grad_norm": 0.5208668112754822,
      "learning_rate": 9.973536973819382e-06,
      "loss": 0.4519,
      "step": 3407
    },
    {
      "epoch": 0.1563948419072094,
      "grad_norm": 0.4765201807022095,
      "learning_rate": 9.973511775215987e-06,
      "loss": 0.4218,
      "step": 3408
    },
    {
      "epoch": 0.15644073241246387,
      "grad_norm": 0.45064905285835266,
      "learning_rate": 9.973486564652866e-06,
      "loss": 0.3699,
      "step": 3409
    },
    {
      "epoch": 0.15648662291771834,
      "grad_norm": 0.4541752338409424,
      "learning_rate": 9.973461342130079e-06,
      "loss": 0.3394,
      "step": 3410
    },
    {
      "epoch": 0.15653251342297278,
      "grad_norm": 0.47415247559547424,
      "learning_rate": 9.973436107647688e-06,
      "loss": 0.4073,
      "step": 3411
    },
    {
      "epoch": 0.15657840392822725,
      "grad_norm": 0.4652484953403473,
      "learning_rate": 9.97341086120575e-06,
      "loss": 0.4058,
      "step": 3412
    },
    {
      "epoch": 0.15662429443348172,
      "grad_norm": 0.47529393434524536,
      "learning_rate": 9.973385602804333e-06,
      "loss": 0.434,
      "step": 3413
    },
    {
      "epoch": 0.15667018493873616,
      "grad_norm": 0.4971560835838318,
      "learning_rate": 9.97336033244349e-06,
      "loss": 0.4805,
      "step": 3414
    },
    {
      "epoch": 0.15671607544399063,
      "grad_norm": 0.5403168201446533,
      "learning_rate": 9.973335050123285e-06,
      "loss": 0.4962,
      "step": 3415
    },
    {
      "epoch": 0.1567619659492451,
      "grad_norm": 0.4820096492767334,
      "learning_rate": 9.97330975584378e-06,
      "loss": 0.4088,
      "step": 3416
    },
    {
      "epoch": 0.15680785645449957,
      "grad_norm": 0.51247638463974,
      "learning_rate": 9.973284449605037e-06,
      "loss": 0.5114,
      "step": 3417
    },
    {
      "epoch": 0.15685374695975401,
      "grad_norm": 0.4764186441898346,
      "learning_rate": 9.973259131407112e-06,
      "loss": 0.3556,
      "step": 3418
    },
    {
      "epoch": 0.15689963746500848,
      "grad_norm": 0.44877660274505615,
      "learning_rate": 9.97323380125007e-06,
      "loss": 0.364,
      "step": 3419
    },
    {
      "epoch": 0.15694552797026295,
      "grad_norm": 0.4653012156486511,
      "learning_rate": 9.973208459133967e-06,
      "loss": 0.404,
      "step": 3420
    },
    {
      "epoch": 0.15699141847551742,
      "grad_norm": 0.5108131170272827,
      "learning_rate": 9.973183105058868e-06,
      "loss": 0.4285,
      "step": 3421
    },
    {
      "epoch": 0.15703730898077187,
      "grad_norm": 0.4447953999042511,
      "learning_rate": 9.973157739024834e-06,
      "loss": 0.3819,
      "step": 3422
    },
    {
      "epoch": 0.15708319948602634,
      "grad_norm": 0.4533008635044098,
      "learning_rate": 9.973132361031926e-06,
      "loss": 0.3846,
      "step": 3423
    },
    {
      "epoch": 0.1571290899912808,
      "grad_norm": 0.5040102005004883,
      "learning_rate": 9.973106971080205e-06,
      "loss": 0.3853,
      "step": 3424
    },
    {
      "epoch": 0.15717498049653528,
      "grad_norm": 0.5119069814682007,
      "learning_rate": 9.973081569169729e-06,
      "loss": 0.5325,
      "step": 3425
    },
    {
      "epoch": 0.15722087100178972,
      "grad_norm": 0.451943039894104,
      "learning_rate": 9.973056155300563e-06,
      "loss": 0.3256,
      "step": 3426
    },
    {
      "epoch": 0.1572667615070442,
      "grad_norm": 0.5147818922996521,
      "learning_rate": 9.973030729472766e-06,
      "loss": 0.4459,
      "step": 3427
    },
    {
      "epoch": 0.15731265201229866,
      "grad_norm": 0.4829035699367523,
      "learning_rate": 9.9730052916864e-06,
      "loss": 0.508,
      "step": 3428
    },
    {
      "epoch": 0.15735854251755313,
      "grad_norm": 0.49244794249534607,
      "learning_rate": 9.972979841941525e-06,
      "loss": 0.4083,
      "step": 3429
    },
    {
      "epoch": 0.15740443302280757,
      "grad_norm": 0.4974241852760315,
      "learning_rate": 9.972954380238203e-06,
      "loss": 0.4255,
      "step": 3430
    },
    {
      "epoch": 0.15745032352806204,
      "grad_norm": 0.4844846725463867,
      "learning_rate": 9.972928906576496e-06,
      "loss": 0.4384,
      "step": 3431
    },
    {
      "epoch": 0.1574962140333165,
      "grad_norm": 0.4852088391780853,
      "learning_rate": 9.972903420956463e-06,
      "loss": 0.4507,
      "step": 3432
    },
    {
      "epoch": 0.15754210453857098,
      "grad_norm": 0.47295406460762024,
      "learning_rate": 9.972877923378168e-06,
      "loss": 0.412,
      "step": 3433
    },
    {
      "epoch": 0.15758799504382542,
      "grad_norm": 0.47541558742523193,
      "learning_rate": 9.972852413841671e-06,
      "loss": 0.4161,
      "step": 3434
    },
    {
      "epoch": 0.1576338855490799,
      "grad_norm": 0.47602322697639465,
      "learning_rate": 9.972826892347035e-06,
      "loss": 0.4245,
      "step": 3435
    },
    {
      "epoch": 0.15767977605433436,
      "grad_norm": 0.48176711797714233,
      "learning_rate": 9.972801358894317e-06,
      "loss": 0.343,
      "step": 3436
    },
    {
      "epoch": 0.15772566655958883,
      "grad_norm": 0.5384151935577393,
      "learning_rate": 9.972775813483582e-06,
      "loss": 0.5104,
      "step": 3437
    },
    {
      "epoch": 0.15777155706484328,
      "grad_norm": 0.5327281951904297,
      "learning_rate": 9.972750256114891e-06,
      "loss": 0.5175,
      "step": 3438
    },
    {
      "epoch": 0.15781744757009775,
      "grad_norm": 0.5229492783546448,
      "learning_rate": 9.972724686788304e-06,
      "loss": 0.4321,
      "step": 3439
    },
    {
      "epoch": 0.15786333807535222,
      "grad_norm": 0.5037031173706055,
      "learning_rate": 9.972699105503885e-06,
      "loss": 0.446,
      "step": 3440
    },
    {
      "epoch": 0.15790922858060666,
      "grad_norm": 0.5082417130470276,
      "learning_rate": 9.972673512261694e-06,
      "loss": 0.4699,
      "step": 3441
    },
    {
      "epoch": 0.15795511908586113,
      "grad_norm": 0.4872379004955292,
      "learning_rate": 9.972647907061791e-06,
      "loss": 0.4487,
      "step": 3442
    },
    {
      "epoch": 0.1580010095911156,
      "grad_norm": 0.44630125164985657,
      "learning_rate": 9.972622289904241e-06,
      "loss": 0.3654,
      "step": 3443
    },
    {
      "epoch": 0.15804690009637007,
      "grad_norm": 0.5068696737289429,
      "learning_rate": 9.972596660789104e-06,
      "loss": 0.5295,
      "step": 3444
    },
    {
      "epoch": 0.1580927906016245,
      "grad_norm": 0.4432818591594696,
      "learning_rate": 9.97257101971644e-06,
      "loss": 0.3641,
      "step": 3445
    },
    {
      "epoch": 0.15813868110687898,
      "grad_norm": 0.4711827039718628,
      "learning_rate": 9.972545366686314e-06,
      "loss": 0.4242,
      "step": 3446
    },
    {
      "epoch": 0.15818457161213345,
      "grad_norm": 0.5113137364387512,
      "learning_rate": 9.972519701698784e-06,
      "loss": 0.4778,
      "step": 3447
    },
    {
      "epoch": 0.15823046211738792,
      "grad_norm": 0.42560282349586487,
      "learning_rate": 9.972494024753915e-06,
      "loss": 0.311,
      "step": 3448
    },
    {
      "epoch": 0.15827635262264236,
      "grad_norm": 0.46892601251602173,
      "learning_rate": 9.972468335851766e-06,
      "loss": 0.4211,
      "step": 3449
    },
    {
      "epoch": 0.15832224312789683,
      "grad_norm": 0.5339772701263428,
      "learning_rate": 9.972442634992399e-06,
      "loss": 0.5147,
      "step": 3450
    },
    {
      "epoch": 0.1583681336331513,
      "grad_norm": 0.4870809018611908,
      "learning_rate": 9.972416922175878e-06,
      "loss": 0.4442,
      "step": 3451
    },
    {
      "epoch": 0.15841402413840577,
      "grad_norm": 0.4548492133617401,
      "learning_rate": 9.972391197402265e-06,
      "loss": 0.3703,
      "step": 3452
    },
    {
      "epoch": 0.15845991464366022,
      "grad_norm": 0.4818779230117798,
      "learning_rate": 9.972365460671618e-06,
      "loss": 0.4731,
      "step": 3453
    },
    {
      "epoch": 0.15850580514891469,
      "grad_norm": 0.5016392469406128,
      "learning_rate": 9.972339711984002e-06,
      "loss": 0.4718,
      "step": 3454
    },
    {
      "epoch": 0.15855169565416916,
      "grad_norm": 0.4378732740879059,
      "learning_rate": 9.972313951339481e-06,
      "loss": 0.3528,
      "step": 3455
    },
    {
      "epoch": 0.15859758615942363,
      "grad_norm": 0.45574894547462463,
      "learning_rate": 9.972288178738112e-06,
      "loss": 0.3417,
      "step": 3456
    },
    {
      "epoch": 0.15864347666467807,
      "grad_norm": 0.4504535496234894,
      "learning_rate": 9.972262394179959e-06,
      "loss": 0.407,
      "step": 3457
    },
    {
      "epoch": 0.15868936716993254,
      "grad_norm": 0.5264883041381836,
      "learning_rate": 9.972236597665085e-06,
      "loss": 0.3318,
      "step": 3458
    },
    {
      "epoch": 0.158735257675187,
      "grad_norm": 0.45955637097358704,
      "learning_rate": 9.97221078919355e-06,
      "loss": 0.3648,
      "step": 3459
    },
    {
      "epoch": 0.15878114818044148,
      "grad_norm": 0.44068920612335205,
      "learning_rate": 9.97218496876542e-06,
      "loss": 0.3676,
      "step": 3460
    },
    {
      "epoch": 0.15882703868569592,
      "grad_norm": 0.4735561013221741,
      "learning_rate": 9.972159136380751e-06,
      "loss": 0.378,
      "step": 3461
    },
    {
      "epoch": 0.1588729291909504,
      "grad_norm": 0.5666905045509338,
      "learning_rate": 9.97213329203961e-06,
      "loss": 0.4883,
      "step": 3462
    },
    {
      "epoch": 0.15891881969620486,
      "grad_norm": 0.4964545667171478,
      "learning_rate": 9.972107435742057e-06,
      "loss": 0.3688,
      "step": 3463
    },
    {
      "epoch": 0.15896471020145933,
      "grad_norm": 0.49066397547721863,
      "learning_rate": 9.972081567488156e-06,
      "loss": 0.4053,
      "step": 3464
    },
    {
      "epoch": 0.15901060070671377,
      "grad_norm": 0.4651356637477875,
      "learning_rate": 9.972055687277969e-06,
      "loss": 0.3821,
      "step": 3465
    },
    {
      "epoch": 0.15905649121196824,
      "grad_norm": 0.45859792828559875,
      "learning_rate": 9.972029795111557e-06,
      "loss": 0.3823,
      "step": 3466
    },
    {
      "epoch": 0.1591023817172227,
      "grad_norm": 0.5831094980239868,
      "learning_rate": 9.972003890988982e-06,
      "loss": 0.481,
      "step": 3467
    },
    {
      "epoch": 0.15914827222247718,
      "grad_norm": 0.46102064847946167,
      "learning_rate": 9.971977974910305e-06,
      "loss": 0.3656,
      "step": 3468
    },
    {
      "epoch": 0.15919416272773163,
      "grad_norm": 0.4691809117794037,
      "learning_rate": 9.971952046875592e-06,
      "loss": 0.4027,
      "step": 3469
    },
    {
      "epoch": 0.1592400532329861,
      "grad_norm": 0.4473104774951935,
      "learning_rate": 9.971926106884904e-06,
      "loss": 0.3324,
      "step": 3470
    },
    {
      "epoch": 0.15928594373824057,
      "grad_norm": 0.45411378145217896,
      "learning_rate": 9.971900154938301e-06,
      "loss": 0.3414,
      "step": 3471
    },
    {
      "epoch": 0.159331834243495,
      "grad_norm": 0.6522583961486816,
      "learning_rate": 9.97187419103585e-06,
      "loss": 0.6009,
      "step": 3472
    },
    {
      "epoch": 0.15937772474874948,
      "grad_norm": 0.48917824029922485,
      "learning_rate": 9.971848215177608e-06,
      "loss": 0.4148,
      "step": 3473
    },
    {
      "epoch": 0.15942361525400395,
      "grad_norm": 0.46714523434638977,
      "learning_rate": 9.971822227363642e-06,
      "loss": 0.3926,
      "step": 3474
    },
    {
      "epoch": 0.15946950575925842,
      "grad_norm": 0.4685218632221222,
      "learning_rate": 9.971796227594011e-06,
      "loss": 0.4305,
      "step": 3475
    },
    {
      "epoch": 0.15951539626451286,
      "grad_norm": 0.5073850154876709,
      "learning_rate": 9.97177021586878e-06,
      "loss": 0.4215,
      "step": 3476
    },
    {
      "epoch": 0.15956128676976733,
      "grad_norm": 0.4946824014186859,
      "learning_rate": 9.971744192188012e-06,
      "loss": 0.3863,
      "step": 3477
    },
    {
      "epoch": 0.1596071772750218,
      "grad_norm": 0.5422838926315308,
      "learning_rate": 9.971718156551767e-06,
      "loss": 0.4859,
      "step": 3478
    },
    {
      "epoch": 0.15965306778027627,
      "grad_norm": 0.47773876786231995,
      "learning_rate": 9.971692108960109e-06,
      "loss": 0.3946,
      "step": 3479
    },
    {
      "epoch": 0.1596989582855307,
      "grad_norm": 0.5057024955749512,
      "learning_rate": 9.971666049413102e-06,
      "loss": 0.4902,
      "step": 3480
    },
    {
      "epoch": 0.15974484879078518,
      "grad_norm": 0.4997882843017578,
      "learning_rate": 9.971639977910806e-06,
      "loss": 0.4698,
      "step": 3481
    },
    {
      "epoch": 0.15979073929603965,
      "grad_norm": 0.4545903205871582,
      "learning_rate": 9.971613894453284e-06,
      "loss": 0.3522,
      "step": 3482
    },
    {
      "epoch": 0.15983662980129412,
      "grad_norm": 0.4788646996021271,
      "learning_rate": 9.971587799040602e-06,
      "loss": 0.3968,
      "step": 3483
    },
    {
      "epoch": 0.15988252030654856,
      "grad_norm": 0.4190256595611572,
      "learning_rate": 9.971561691672818e-06,
      "loss": 0.3483,
      "step": 3484
    },
    {
      "epoch": 0.15992841081180303,
      "grad_norm": 0.6223978400230408,
      "learning_rate": 9.97153557235e-06,
      "loss": 0.4374,
      "step": 3485
    },
    {
      "epoch": 0.1599743013170575,
      "grad_norm": 0.4866560697555542,
      "learning_rate": 9.971509441072206e-06,
      "loss": 0.4325,
      "step": 3486
    },
    {
      "epoch": 0.16002019182231197,
      "grad_norm": 0.4690292179584503,
      "learning_rate": 9.971483297839501e-06,
      "loss": 0.4232,
      "step": 3487
    },
    {
      "epoch": 0.16006608232756642,
      "grad_norm": 0.5000173449516296,
      "learning_rate": 9.971457142651948e-06,
      "loss": 0.5088,
      "step": 3488
    },
    {
      "epoch": 0.1601119728328209,
      "grad_norm": 0.45145830512046814,
      "learning_rate": 9.97143097550961e-06,
      "loss": 0.3997,
      "step": 3489
    },
    {
      "epoch": 0.16015786333807536,
      "grad_norm": 0.48280757665634155,
      "learning_rate": 9.971404796412549e-06,
      "loss": 0.4424,
      "step": 3490
    },
    {
      "epoch": 0.16020375384332983,
      "grad_norm": 0.463607519865036,
      "learning_rate": 9.97137860536083e-06,
      "loss": 0.3572,
      "step": 3491
    },
    {
      "epoch": 0.16024964434858427,
      "grad_norm": 0.4859597980976105,
      "learning_rate": 9.97135240235451e-06,
      "loss": 0.4294,
      "step": 3492
    },
    {
      "epoch": 0.16029553485383874,
      "grad_norm": 0.4947422444820404,
      "learning_rate": 9.971326187393661e-06,
      "loss": 0.4047,
      "step": 3493
    },
    {
      "epoch": 0.1603414253590932,
      "grad_norm": 0.508838415145874,
      "learning_rate": 9.97129996047834e-06,
      "loss": 0.4525,
      "step": 3494
    },
    {
      "epoch": 0.16038731586434768,
      "grad_norm": 0.4381849467754364,
      "learning_rate": 9.971273721608611e-06,
      "loss": 0.3566,
      "step": 3495
    },
    {
      "epoch": 0.16043320636960212,
      "grad_norm": 0.4655773639678955,
      "learning_rate": 9.971247470784538e-06,
      "loss": 0.3967,
      "step": 3496
    },
    {
      "epoch": 0.1604790968748566,
      "grad_norm": 0.4902365803718567,
      "learning_rate": 9.971221208006185e-06,
      "loss": 0.4305,
      "step": 3497
    },
    {
      "epoch": 0.16052498738011106,
      "grad_norm": 0.4625941812992096,
      "learning_rate": 9.971194933273614e-06,
      "loss": 0.348,
      "step": 3498
    },
    {
      "epoch": 0.16057087788536553,
      "grad_norm": 0.4561326503753662,
      "learning_rate": 9.971168646586886e-06,
      "loss": 0.3368,
      "step": 3499
    },
    {
      "epoch": 0.16061676839061997,
      "grad_norm": 0.45755860209465027,
      "learning_rate": 9.971142347946068e-06,
      "loss": 0.3352,
      "step": 3500
    },
    {
      "epoch": 0.16066265889587444,
      "grad_norm": 0.5377780199050903,
      "learning_rate": 9.971116037351221e-06,
      "loss": 0.525,
      "step": 3501
    },
    {
      "epoch": 0.16070854940112891,
      "grad_norm": 0.5142645835876465,
      "learning_rate": 9.971089714802409e-06,
      "loss": 0.4804,
      "step": 3502
    },
    {
      "epoch": 0.16075443990638336,
      "grad_norm": 0.4570762813091278,
      "learning_rate": 9.971063380299695e-06,
      "loss": 0.3607,
      "step": 3503
    },
    {
      "epoch": 0.16080033041163783,
      "grad_norm": 0.5035489201545715,
      "learning_rate": 9.971037033843143e-06,
      "loss": 0.4369,
      "step": 3504
    },
    {
      "epoch": 0.1608462209168923,
      "grad_norm": 0.4712914824485779,
      "learning_rate": 9.971010675432816e-06,
      "loss": 0.3548,
      "step": 3505
    },
    {
      "epoch": 0.16089211142214677,
      "grad_norm": 0.48235970735549927,
      "learning_rate": 9.970984305068777e-06,
      "loss": 0.4269,
      "step": 3506
    },
    {
      "epoch": 0.1609380019274012,
      "grad_norm": 0.48529648780822754,
      "learning_rate": 9.970957922751088e-06,
      "loss": 0.4613,
      "step": 3507
    },
    {
      "epoch": 0.16098389243265568,
      "grad_norm": 0.45862317085266113,
      "learning_rate": 9.970931528479815e-06,
      "loss": 0.354,
      "step": 3508
    },
    {
      "epoch": 0.16102978293791015,
      "grad_norm": 0.4915834665298462,
      "learning_rate": 9.970905122255022e-06,
      "loss": 0.4101,
      "step": 3509
    },
    {
      "epoch": 0.16107567344316462,
      "grad_norm": 0.4911457300186157,
      "learning_rate": 9.97087870407677e-06,
      "loss": 0.457,
      "step": 3510
    },
    {
      "epoch": 0.16112156394841906,
      "grad_norm": 0.4485325515270233,
      "learning_rate": 9.970852273945124e-06,
      "loss": 0.3288,
      "step": 3511
    },
    {
      "epoch": 0.16116745445367353,
      "grad_norm": 0.4835895001888275,
      "learning_rate": 9.970825831860147e-06,
      "loss": 0.4256,
      "step": 3512
    },
    {
      "epoch": 0.161213344958928,
      "grad_norm": 0.4708102345466614,
      "learning_rate": 9.970799377821902e-06,
      "loss": 0.4208,
      "step": 3513
    },
    {
      "epoch": 0.16125923546418247,
      "grad_norm": 0.45050904154777527,
      "learning_rate": 9.970772911830452e-06,
      "loss": 0.3896,
      "step": 3514
    },
    {
      "epoch": 0.1613051259694369,
      "grad_norm": 0.4893935024738312,
      "learning_rate": 9.970746433885865e-06,
      "loss": 0.4551,
      "step": 3515
    },
    {
      "epoch": 0.16135101647469138,
      "grad_norm": 0.4802362024784088,
      "learning_rate": 9.9707199439882e-06,
      "loss": 0.4072,
      "step": 3516
    },
    {
      "epoch": 0.16139690697994585,
      "grad_norm": 0.4859634041786194,
      "learning_rate": 9.970693442137524e-06,
      "loss": 0.472,
      "step": 3517
    },
    {
      "epoch": 0.16144279748520032,
      "grad_norm": 0.4686821401119232,
      "learning_rate": 9.970666928333898e-06,
      "loss": 0.4034,
      "step": 3518
    },
    {
      "epoch": 0.16148868799045477,
      "grad_norm": 0.46538659930229187,
      "learning_rate": 9.970640402577386e-06,
      "loss": 0.4106,
      "step": 3519
    },
    {
      "epoch": 0.16153457849570924,
      "grad_norm": 0.5193923115730286,
      "learning_rate": 9.970613864868052e-06,
      "loss": 0.4924,
      "step": 3520
    },
    {
      "epoch": 0.1615804690009637,
      "grad_norm": 0.4929127097129822,
      "learning_rate": 9.970587315205963e-06,
      "loss": 0.4591,
      "step": 3521
    },
    {
      "epoch": 0.16162635950621818,
      "grad_norm": 0.46408167481422424,
      "learning_rate": 9.970560753591179e-06,
      "loss": 0.4327,
      "step": 3522
    },
    {
      "epoch": 0.16167225001147262,
      "grad_norm": 0.4852508008480072,
      "learning_rate": 9.970534180023765e-06,
      "loss": 0.3999,
      "step": 3523
    },
    {
      "epoch": 0.1617181405167271,
      "grad_norm": 0.5348367094993591,
      "learning_rate": 9.970507594503784e-06,
      "loss": 0.4923,
      "step": 3524
    },
    {
      "epoch": 0.16176403102198156,
      "grad_norm": 0.6300472021102905,
      "learning_rate": 9.970480997031302e-06,
      "loss": 0.4225,
      "step": 3525
    },
    {
      "epoch": 0.16180992152723603,
      "grad_norm": 0.4214268624782562,
      "learning_rate": 9.970454387606383e-06,
      "loss": 0.3489,
      "step": 3526
    },
    {
      "epoch": 0.16185581203249047,
      "grad_norm": 0.5366626977920532,
      "learning_rate": 9.97042776622909e-06,
      "loss": 0.4199,
      "step": 3527
    },
    {
      "epoch": 0.16190170253774494,
      "grad_norm": 0.5000843405723572,
      "learning_rate": 9.970401132899486e-06,
      "loss": 0.417,
      "step": 3528
    },
    {
      "epoch": 0.1619475930429994,
      "grad_norm": 0.49688413739204407,
      "learning_rate": 9.970374487617634e-06,
      "loss": 0.4798,
      "step": 3529
    },
    {
      "epoch": 0.16199348354825385,
      "grad_norm": 0.44794321060180664,
      "learning_rate": 9.970347830383602e-06,
      "loss": 0.344,
      "step": 3530
    },
    {
      "epoch": 0.16203937405350832,
      "grad_norm": 0.4808467626571655,
      "learning_rate": 9.970321161197452e-06,
      "loss": 0.4158,
      "step": 3531
    },
    {
      "epoch": 0.1620852645587628,
      "grad_norm": 0.46551400423049927,
      "learning_rate": 9.970294480059249e-06,
      "loss": 0.411,
      "step": 3532
    },
    {
      "epoch": 0.16213115506401726,
      "grad_norm": 0.4455750286579132,
      "learning_rate": 9.970267786969056e-06,
      "loss": 0.3465,
      "step": 3533
    },
    {
      "epoch": 0.1621770455692717,
      "grad_norm": 0.4404858350753784,
      "learning_rate": 9.970241081926936e-06,
      "loss": 0.3182,
      "step": 3534
    },
    {
      "epoch": 0.16222293607452618,
      "grad_norm": 0.49429038166999817,
      "learning_rate": 9.970214364932958e-06,
      "loss": 0.405,
      "step": 3535
    },
    {
      "epoch": 0.16226882657978065,
      "grad_norm": 0.4389179050922394,
      "learning_rate": 9.970187635987181e-06,
      "loss": 0.3637,
      "step": 3536
    },
    {
      "epoch": 0.16231471708503512,
      "grad_norm": 0.49280327558517456,
      "learning_rate": 9.97016089508967e-06,
      "loss": 0.4417,
      "step": 3537
    },
    {
      "epoch": 0.16236060759028956,
      "grad_norm": 0.4841049909591675,
      "learning_rate": 9.970134142240492e-06,
      "loss": 0.4105,
      "step": 3538
    },
    {
      "epoch": 0.16240649809554403,
      "grad_norm": 0.4690589904785156,
      "learning_rate": 9.970107377439713e-06,
      "loss": 0.4488,
      "step": 3539
    },
    {
      "epoch": 0.1624523886007985,
      "grad_norm": 0.48826152086257935,
      "learning_rate": 9.97008060068739e-06,
      "loss": 0.4522,
      "step": 3540
    },
    {
      "epoch": 0.16249827910605297,
      "grad_norm": 0.4803142249584198,
      "learning_rate": 9.970053811983594e-06,
      "loss": 0.409,
      "step": 3541
    },
    {
      "epoch": 0.1625441696113074,
      "grad_norm": 0.5035238862037659,
      "learning_rate": 9.970027011328386e-06,
      "loss": 0.4819,
      "step": 3542
    },
    {
      "epoch": 0.16259006011656188,
      "grad_norm": 0.44776809215545654,
      "learning_rate": 9.970000198721835e-06,
      "loss": 0.3522,
      "step": 3543
    },
    {
      "epoch": 0.16263595062181635,
      "grad_norm": 0.5222958326339722,
      "learning_rate": 9.969973374163998e-06,
      "loss": 0.4357,
      "step": 3544
    },
    {
      "epoch": 0.16268184112707082,
      "grad_norm": 0.5011484026908875,
      "learning_rate": 9.969946537654944e-06,
      "loss": 0.4468,
      "step": 3545
    },
    {
      "epoch": 0.16272773163232526,
      "grad_norm": 0.5149145722389221,
      "learning_rate": 9.969919689194739e-06,
      "loss": 0.4858,
      "step": 3546
    },
    {
      "epoch": 0.16277362213757973,
      "grad_norm": 0.4738142192363739,
      "learning_rate": 9.969892828783445e-06,
      "loss": 0.4285,
      "step": 3547
    },
    {
      "epoch": 0.1628195126428342,
      "grad_norm": 0.49899822473526,
      "learning_rate": 9.969865956421127e-06,
      "loss": 0.4527,
      "step": 3548
    },
    {
      "epoch": 0.16286540314808867,
      "grad_norm": 0.4843798279762268,
      "learning_rate": 9.96983907210785e-06,
      "loss": 0.4493,
      "step": 3549
    },
    {
      "epoch": 0.16291129365334311,
      "grad_norm": 0.5192020535469055,
      "learning_rate": 9.96981217584368e-06,
      "loss": 0.4716,
      "step": 3550
    },
    {
      "epoch": 0.16295718415859758,
      "grad_norm": 0.46311327815055847,
      "learning_rate": 9.969785267628677e-06,
      "loss": 0.391,
      "step": 3551
    },
    {
      "epoch": 0.16300307466385205,
      "grad_norm": 0.5039772987365723,
      "learning_rate": 9.969758347462912e-06,
      "loss": 0.472,
      "step": 3552
    },
    {
      "epoch": 0.16304896516910652,
      "grad_norm": 0.4776366651058197,
      "learning_rate": 9.969731415346445e-06,
      "loss": 0.3838,
      "step": 3553
    },
    {
      "epoch": 0.16309485567436097,
      "grad_norm": 0.4233502149581909,
      "learning_rate": 9.969704471279342e-06,
      "loss": 0.3524,
      "step": 3554
    },
    {
      "epoch": 0.16314074617961544,
      "grad_norm": 0.48174601793289185,
      "learning_rate": 9.96967751526167e-06,
      "loss": 0.3853,
      "step": 3555
    },
    {
      "epoch": 0.1631866366848699,
      "grad_norm": 0.48675301671028137,
      "learning_rate": 9.969650547293492e-06,
      "loss": 0.4591,
      "step": 3556
    },
    {
      "epoch": 0.16323252719012438,
      "grad_norm": 0.4992932975292206,
      "learning_rate": 9.96962356737487e-06,
      "loss": 0.412,
      "step": 3557
    },
    {
      "epoch": 0.16327841769537882,
      "grad_norm": 0.48553457856178284,
      "learning_rate": 9.969596575505876e-06,
      "loss": 0.4594,
      "step": 3558
    },
    {
      "epoch": 0.1633243082006333,
      "grad_norm": 0.519641637802124,
      "learning_rate": 9.969569571686569e-06,
      "loss": 0.5074,
      "step": 3559
    },
    {
      "epoch": 0.16337019870588776,
      "grad_norm": 0.4730679988861084,
      "learning_rate": 9.969542555917013e-06,
      "loss": 0.4667,
      "step": 3560
    },
    {
      "epoch": 0.1634160892111422,
      "grad_norm": 0.49281245470046997,
      "learning_rate": 9.969515528197279e-06,
      "loss": 0.4372,
      "step": 3561
    },
    {
      "epoch": 0.16346197971639667,
      "grad_norm": 0.46238622069358826,
      "learning_rate": 9.969488488527429e-06,
      "loss": 0.395,
      "step": 3562
    },
    {
      "epoch": 0.16350787022165114,
      "grad_norm": 0.4411563277244568,
      "learning_rate": 9.969461436907524e-06,
      "loss": 0.3493,
      "step": 3563
    },
    {
      "epoch": 0.1635537607269056,
      "grad_norm": 0.4830007553100586,
      "learning_rate": 9.969434373337635e-06,
      "loss": 0.3951,
      "step": 3564
    },
    {
      "epoch": 0.16359965123216005,
      "grad_norm": 0.4694591462612152,
      "learning_rate": 9.969407297817823e-06,
      "loss": 0.3695,
      "step": 3565
    },
    {
      "epoch": 0.16364554173741452,
      "grad_norm": 0.45552733540534973,
      "learning_rate": 9.969380210348157e-06,
      "loss": 0.3376,
      "step": 3566
    },
    {
      "epoch": 0.163691432242669,
      "grad_norm": 0.46346917748451233,
      "learning_rate": 9.9693531109287e-06,
      "loss": 0.3801,
      "step": 3567
    },
    {
      "epoch": 0.16373732274792346,
      "grad_norm": 0.443578839302063,
      "learning_rate": 9.969325999559516e-06,
      "loss": 0.3775,
      "step": 3568
    },
    {
      "epoch": 0.1637832132531779,
      "grad_norm": 0.43470194935798645,
      "learning_rate": 9.969298876240672e-06,
      "loss": 0.3593,
      "step": 3569
    },
    {
      "epoch": 0.16382910375843238,
      "grad_norm": 0.4560805857181549,
      "learning_rate": 9.969271740972232e-06,
      "loss": 0.3197,
      "step": 3570
    },
    {
      "epoch": 0.16387499426368685,
      "grad_norm": 0.47402840852737427,
      "learning_rate": 9.969244593754261e-06,
      "loss": 0.4427,
      "step": 3571
    },
    {
      "epoch": 0.16392088476894132,
      "grad_norm": 0.4482603073120117,
      "learning_rate": 9.969217434586826e-06,
      "loss": 0.3807,
      "step": 3572
    },
    {
      "epoch": 0.16396677527419576,
      "grad_norm": 0.4365561306476593,
      "learning_rate": 9.969190263469992e-06,
      "loss": 0.3573,
      "step": 3573
    },
    {
      "epoch": 0.16401266577945023,
      "grad_norm": 0.45426681637763977,
      "learning_rate": 9.969163080403823e-06,
      "loss": 0.3594,
      "step": 3574
    },
    {
      "epoch": 0.1640585562847047,
      "grad_norm": 0.44736793637275696,
      "learning_rate": 9.969135885388386e-06,
      "loss": 0.3356,
      "step": 3575
    },
    {
      "epoch": 0.16410444678995917,
      "grad_norm": 0.4486168622970581,
      "learning_rate": 9.969108678423744e-06,
      "loss": 0.3529,
      "step": 3576
    },
    {
      "epoch": 0.1641503372952136,
      "grad_norm": 0.5961507558822632,
      "learning_rate": 9.969081459509965e-06,
      "loss": 0.4484,
      "step": 3577
    },
    {
      "epoch": 0.16419622780046808,
      "grad_norm": 0.47634968161582947,
      "learning_rate": 9.969054228647113e-06,
      "loss": 0.427,
      "step": 3578
    },
    {
      "epoch": 0.16424211830572255,
      "grad_norm": 0.4267920255661011,
      "learning_rate": 9.969026985835256e-06,
      "loss": 0.3241,
      "step": 3579
    },
    {
      "epoch": 0.16428800881097702,
      "grad_norm": 0.4745945930480957,
      "learning_rate": 9.968999731074453e-06,
      "loss": 0.4324,
      "step": 3580
    },
    {
      "epoch": 0.16433389931623146,
      "grad_norm": 0.5914706587791443,
      "learning_rate": 9.968972464364777e-06,
      "loss": 0.5544,
      "step": 3581
    },
    {
      "epoch": 0.16437978982148593,
      "grad_norm": 0.5118852853775024,
      "learning_rate": 9.96894518570629e-06,
      "loss": 0.4806,
      "step": 3582
    },
    {
      "epoch": 0.1644256803267404,
      "grad_norm": 0.47302529215812683,
      "learning_rate": 9.968917895099057e-06,
      "loss": 0.326,
      "step": 3583
    },
    {
      "epoch": 0.16447157083199487,
      "grad_norm": 0.4752989411354065,
      "learning_rate": 9.968890592543145e-06,
      "loss": 0.4096,
      "step": 3584
    },
    {
      "epoch": 0.16451746133724932,
      "grad_norm": 0.49046263098716736,
      "learning_rate": 9.96886327803862e-06,
      "loss": 0.4083,
      "step": 3585
    },
    {
      "epoch": 0.16456335184250379,
      "grad_norm": 0.5011879205703735,
      "learning_rate": 9.968835951585549e-06,
      "loss": 0.4752,
      "step": 3586
    },
    {
      "epoch": 0.16460924234775826,
      "grad_norm": 0.46065208315849304,
      "learning_rate": 9.968808613183993e-06,
      "loss": 0.3789,
      "step": 3587
    },
    {
      "epoch": 0.1646551328530127,
      "grad_norm": 0.46373555064201355,
      "learning_rate": 9.96878126283402e-06,
      "loss": 0.391,
      "step": 3588
    },
    {
      "epoch": 0.16470102335826717,
      "grad_norm": 0.49008578062057495,
      "learning_rate": 9.968753900535699e-06,
      "loss": 0.4234,
      "step": 3589
    },
    {
      "epoch": 0.16474691386352164,
      "grad_norm": 0.49280592799186707,
      "learning_rate": 9.968726526289091e-06,
      "loss": 0.4676,
      "step": 3590
    },
    {
      "epoch": 0.1647928043687761,
      "grad_norm": 0.4429382085800171,
      "learning_rate": 9.968699140094264e-06,
      "loss": 0.3359,
      "step": 3591
    },
    {
      "epoch": 0.16483869487403055,
      "grad_norm": 0.4706316590309143,
      "learning_rate": 9.968671741951284e-06,
      "loss": 0.4487,
      "step": 3592
    },
    {
      "epoch": 0.16488458537928502,
      "grad_norm": 0.4552987515926361,
      "learning_rate": 9.968644331860216e-06,
      "loss": 0.3573,
      "step": 3593
    },
    {
      "epoch": 0.1649304758845395,
      "grad_norm": 0.5117678642272949,
      "learning_rate": 9.968616909821128e-06,
      "loss": 0.4636,
      "step": 3594
    },
    {
      "epoch": 0.16497636638979396,
      "grad_norm": 0.48572269082069397,
      "learning_rate": 9.968589475834083e-06,
      "loss": 0.4353,
      "step": 3595
    },
    {
      "epoch": 0.1650222568950484,
      "grad_norm": 0.41882574558258057,
      "learning_rate": 9.968562029899149e-06,
      "loss": 0.3105,
      "step": 3596
    },
    {
      "epoch": 0.16506814740030287,
      "grad_norm": 0.49018704891204834,
      "learning_rate": 9.968534572016393e-06,
      "loss": 0.4289,
      "step": 3597
    },
    {
      "epoch": 0.16511403790555734,
      "grad_norm": 0.4399867355823517,
      "learning_rate": 9.968507102185879e-06,
      "loss": 0.3976,
      "step": 3598
    },
    {
      "epoch": 0.1651599284108118,
      "grad_norm": 0.4994634985923767,
      "learning_rate": 9.968479620407672e-06,
      "loss": 0.4735,
      "step": 3599
    },
    {
      "epoch": 0.16520581891606625,
      "grad_norm": 0.4570196866989136,
      "learning_rate": 9.96845212668184e-06,
      "loss": 0.341,
      "step": 3600
    },
    {
      "epoch": 0.16525170942132072,
      "grad_norm": 0.49720463156700134,
      "learning_rate": 9.968424621008448e-06,
      "loss": 0.4065,
      "step": 3601
    },
    {
      "epoch": 0.1652975999265752,
      "grad_norm": 0.4903962314128876,
      "learning_rate": 9.968397103387564e-06,
      "loss": 0.4717,
      "step": 3602
    },
    {
      "epoch": 0.16534349043182966,
      "grad_norm": 0.4884265959262848,
      "learning_rate": 9.968369573819253e-06,
      "loss": 0.4325,
      "step": 3603
    },
    {
      "epoch": 0.1653893809370841,
      "grad_norm": 0.4561275541782379,
      "learning_rate": 9.96834203230358e-06,
      "loss": 0.4037,
      "step": 3604
    },
    {
      "epoch": 0.16543527144233858,
      "grad_norm": 0.4439562261104584,
      "learning_rate": 9.968314478840614e-06,
      "loss": 0.3302,
      "step": 3605
    },
    {
      "epoch": 0.16548116194759305,
      "grad_norm": 0.4430003762245178,
      "learning_rate": 9.968286913430419e-06,
      "loss": 0.3572,
      "step": 3606
    },
    {
      "epoch": 0.16552705245284752,
      "grad_norm": 0.54210364818573,
      "learning_rate": 9.968259336073063e-06,
      "loss": 0.4807,
      "step": 3607
    },
    {
      "epoch": 0.16557294295810196,
      "grad_norm": 0.42343929409980774,
      "learning_rate": 9.96823174676861e-06,
      "loss": 0.3588,
      "step": 3608
    },
    {
      "epoch": 0.16561883346335643,
      "grad_norm": 0.4443022906780243,
      "learning_rate": 9.968204145517128e-06,
      "loss": 0.3409,
      "step": 3609
    },
    {
      "epoch": 0.1656647239686109,
      "grad_norm": 0.47555527091026306,
      "learning_rate": 9.968176532318682e-06,
      "loss": 0.4034,
      "step": 3610
    },
    {
      "epoch": 0.16571061447386537,
      "grad_norm": 0.47132250666618347,
      "learning_rate": 9.968148907173344e-06,
      "loss": 0.3995,
      "step": 3611
    },
    {
      "epoch": 0.1657565049791198,
      "grad_norm": 0.551042377948761,
      "learning_rate": 9.968121270081171e-06,
      "loss": 0.4833,
      "step": 3612
    },
    {
      "epoch": 0.16580239548437428,
      "grad_norm": 0.4281960427761078,
      "learning_rate": 9.968093621042236e-06,
      "loss": 0.3566,
      "step": 3613
    },
    {
      "epoch": 0.16584828598962875,
      "grad_norm": 0.48665499687194824,
      "learning_rate": 9.968065960056603e-06,
      "loss": 0.401,
      "step": 3614
    },
    {
      "epoch": 0.16589417649488322,
      "grad_norm": 0.5198683142662048,
      "learning_rate": 9.96803828712434e-06,
      "loss": 0.4607,
      "step": 3615
    },
    {
      "epoch": 0.16594006700013766,
      "grad_norm": 0.48868244886398315,
      "learning_rate": 9.968010602245514e-06,
      "loss": 0.4556,
      "step": 3616
    },
    {
      "epoch": 0.16598595750539213,
      "grad_norm": 0.4600757658481598,
      "learning_rate": 9.967982905420188e-06,
      "loss": 0.3886,
      "step": 3617
    },
    {
      "epoch": 0.1660318480106466,
      "grad_norm": 0.4531581997871399,
      "learning_rate": 9.967955196648433e-06,
      "loss": 0.3932,
      "step": 3618
    },
    {
      "epoch": 0.16607773851590105,
      "grad_norm": 0.511566698551178,
      "learning_rate": 9.967927475930314e-06,
      "loss": 0.4625,
      "step": 3619
    },
    {
      "epoch": 0.16612362902115552,
      "grad_norm": 0.5202604532241821,
      "learning_rate": 9.967899743265896e-06,
      "loss": 0.5543,
      "step": 3620
    },
    {
      "epoch": 0.16616951952641,
      "grad_norm": 0.46145981550216675,
      "learning_rate": 9.967871998655247e-06,
      "loss": 0.3985,
      "step": 3621
    },
    {
      "epoch": 0.16621541003166446,
      "grad_norm": 0.4441681206226349,
      "learning_rate": 9.967844242098435e-06,
      "loss": 0.4123,
      "step": 3622
    },
    {
      "epoch": 0.1662613005369189,
      "grad_norm": 0.5212213397026062,
      "learning_rate": 9.967816473595525e-06,
      "loss": 0.4368,
      "step": 3623
    },
    {
      "epoch": 0.16630719104217337,
      "grad_norm": 0.4769091308116913,
      "learning_rate": 9.967788693146583e-06,
      "loss": 0.4398,
      "step": 3624
    },
    {
      "epoch": 0.16635308154742784,
      "grad_norm": 0.5438241362571716,
      "learning_rate": 9.967760900751679e-06,
      "loss": 0.5716,
      "step": 3625
    },
    {
      "epoch": 0.1663989720526823,
      "grad_norm": 0.5117270350456238,
      "learning_rate": 9.967733096410877e-06,
      "loss": 0.5079,
      "step": 3626
    },
    {
      "epoch": 0.16644486255793675,
      "grad_norm": 0.4526447653770447,
      "learning_rate": 9.967705280124244e-06,
      "loss": 0.3317,
      "step": 3627
    },
    {
      "epoch": 0.16649075306319122,
      "grad_norm": 0.48207688331604004,
      "learning_rate": 9.967677451891849e-06,
      "loss": 0.4355,
      "step": 3628
    },
    {
      "epoch": 0.1665366435684457,
      "grad_norm": 0.5575090646743774,
      "learning_rate": 9.967649611713757e-06,
      "loss": 0.4618,
      "step": 3629
    },
    {
      "epoch": 0.16658253407370016,
      "grad_norm": 0.48127418756484985,
      "learning_rate": 9.967621759590036e-06,
      "loss": 0.4136,
      "step": 3630
    },
    {
      "epoch": 0.1666284245789546,
      "grad_norm": 0.44972941279411316,
      "learning_rate": 9.96759389552075e-06,
      "loss": 0.351,
      "step": 3631
    },
    {
      "epoch": 0.16667431508420907,
      "grad_norm": 0.4519535005092621,
      "learning_rate": 9.967566019505972e-06,
      "loss": 0.3936,
      "step": 3632
    },
    {
      "epoch": 0.16672020558946354,
      "grad_norm": 0.48320138454437256,
      "learning_rate": 9.967538131545765e-06,
      "loss": 0.422,
      "step": 3633
    },
    {
      "epoch": 0.166766096094718,
      "grad_norm": 0.4901284873485565,
      "learning_rate": 9.967510231640194e-06,
      "loss": 0.4414,
      "step": 3634
    },
    {
      "epoch": 0.16681198659997246,
      "grad_norm": 0.6204987168312073,
      "learning_rate": 9.96748231978933e-06,
      "loss": 0.4445,
      "step": 3635
    },
    {
      "epoch": 0.16685787710522693,
      "grad_norm": 0.42596858739852905,
      "learning_rate": 9.967454395993239e-06,
      "loss": 0.3173,
      "step": 3636
    },
    {
      "epoch": 0.1669037676104814,
      "grad_norm": 0.4553443193435669,
      "learning_rate": 9.96742646025199e-06,
      "loss": 0.3544,
      "step": 3637
    },
    {
      "epoch": 0.16694965811573587,
      "grad_norm": 0.4198419749736786,
      "learning_rate": 9.967398512565645e-06,
      "loss": 0.345,
      "step": 3638
    },
    {
      "epoch": 0.1669955486209903,
      "grad_norm": 0.4592035114765167,
      "learning_rate": 9.967370552934275e-06,
      "loss": 0.3388,
      "step": 3639
    },
    {
      "epoch": 0.16704143912624478,
      "grad_norm": 0.4720841646194458,
      "learning_rate": 9.967342581357948e-06,
      "loss": 0.3526,
      "step": 3640
    },
    {
      "epoch": 0.16708732963149925,
      "grad_norm": 0.449739009141922,
      "learning_rate": 9.967314597836729e-06,
      "loss": 0.3367,
      "step": 3641
    },
    {
      "epoch": 0.16713322013675372,
      "grad_norm": 0.45591211318969727,
      "learning_rate": 9.967286602370684e-06,
      "loss": 0.346,
      "step": 3642
    },
    {
      "epoch": 0.16717911064200816,
      "grad_norm": 0.477683961391449,
      "learning_rate": 9.967258594959885e-06,
      "loss": 0.3681,
      "step": 3643
    },
    {
      "epoch": 0.16722500114726263,
      "grad_norm": 0.4451040029525757,
      "learning_rate": 9.967230575604397e-06,
      "loss": 0.3441,
      "step": 3644
    },
    {
      "epoch": 0.1672708916525171,
      "grad_norm": 0.5009549856185913,
      "learning_rate": 9.967202544304286e-06,
      "loss": 0.4363,
      "step": 3645
    },
    {
      "epoch": 0.16731678215777157,
      "grad_norm": 0.49261751770973206,
      "learning_rate": 9.967174501059622e-06,
      "loss": 0.4321,
      "step": 3646
    },
    {
      "epoch": 0.167362672663026,
      "grad_norm": 0.48177361488342285,
      "learning_rate": 9.96714644587047e-06,
      "loss": 0.423,
      "step": 3647
    },
    {
      "epoch": 0.16740856316828048,
      "grad_norm": 0.48607900738716125,
      "learning_rate": 9.967118378736899e-06,
      "loss": 0.459,
      "step": 3648
    },
    {
      "epoch": 0.16745445367353495,
      "grad_norm": 0.49827563762664795,
      "learning_rate": 9.967090299658975e-06,
      "loss": 0.4954,
      "step": 3649
    },
    {
      "epoch": 0.1675003441787894,
      "grad_norm": 0.5257965922355652,
      "learning_rate": 9.967062208636767e-06,
      "loss": 0.4866,
      "step": 3650
    },
    {
      "epoch": 0.16754623468404387,
      "grad_norm": 0.4925329089164734,
      "learning_rate": 9.967034105670341e-06,
      "loss": 0.3914,
      "step": 3651
    },
    {
      "epoch": 0.16759212518929834,
      "grad_norm": 0.4688475430011749,
      "learning_rate": 9.967005990759768e-06,
      "loss": 0.3363,
      "step": 3652
    },
    {
      "epoch": 0.1676380156945528,
      "grad_norm": 0.4857305884361267,
      "learning_rate": 9.966977863905112e-06,
      "loss": 0.4413,
      "step": 3653
    },
    {
      "epoch": 0.16768390619980725,
      "grad_norm": 0.4764745533466339,
      "learning_rate": 9.96694972510644e-06,
      "loss": 0.4698,
      "step": 3654
    },
    {
      "epoch": 0.16772979670506172,
      "grad_norm": 0.5319045782089233,
      "learning_rate": 9.966921574363824e-06,
      "loss": 0.4475,
      "step": 3655
    },
    {
      "epoch": 0.1677756872103162,
      "grad_norm": 0.48479700088500977,
      "learning_rate": 9.966893411677328e-06,
      "loss": 0.4159,
      "step": 3656
    },
    {
      "epoch": 0.16782157771557066,
      "grad_norm": 0.46323296427726746,
      "learning_rate": 9.966865237047022e-06,
      "loss": 0.3678,
      "step": 3657
    },
    {
      "epoch": 0.1678674682208251,
      "grad_norm": 0.5238285660743713,
      "learning_rate": 9.966837050472973e-06,
      "loss": 0.4789,
      "step": 3658
    },
    {
      "epoch": 0.16791335872607957,
      "grad_norm": 0.5209804177284241,
      "learning_rate": 9.966808851955248e-06,
      "loss": 0.4723,
      "step": 3659
    },
    {
      "epoch": 0.16795924923133404,
      "grad_norm": 0.4425673484802246,
      "learning_rate": 9.966780641493914e-06,
      "loss": 0.3276,
      "step": 3660
    },
    {
      "epoch": 0.1680051397365885,
      "grad_norm": 0.4626409411430359,
      "learning_rate": 9.96675241908904e-06,
      "loss": 0.3701,
      "step": 3661
    },
    {
      "epoch": 0.16805103024184295,
      "grad_norm": 0.4914664924144745,
      "learning_rate": 9.966724184740694e-06,
      "loss": 0.4388,
      "step": 3662
    },
    {
      "epoch": 0.16809692074709742,
      "grad_norm": 0.4645136892795563,
      "learning_rate": 9.966695938448946e-06,
      "loss": 0.3894,
      "step": 3663
    },
    {
      "epoch": 0.1681428112523519,
      "grad_norm": 0.48512527346611023,
      "learning_rate": 9.966667680213861e-06,
      "loss": 0.4511,
      "step": 3664
    },
    {
      "epoch": 0.16818870175760636,
      "grad_norm": 0.48779502511024475,
      "learning_rate": 9.966639410035508e-06,
      "loss": 0.4428,
      "step": 3665
    },
    {
      "epoch": 0.1682345922628608,
      "grad_norm": 0.45460015535354614,
      "learning_rate": 9.966611127913953e-06,
      "loss": 0.3389,
      "step": 3666
    },
    {
      "epoch": 0.16828048276811527,
      "grad_norm": 0.4733645021915436,
      "learning_rate": 9.966582833849267e-06,
      "loss": 0.3401,
      "step": 3667
    },
    {
      "epoch": 0.16832637327336974,
      "grad_norm": 0.5052480101585388,
      "learning_rate": 9.966554527841517e-06,
      "loss": 0.3982,
      "step": 3668
    },
    {
      "epoch": 0.16837226377862421,
      "grad_norm": 0.4637988805770874,
      "learning_rate": 9.96652620989077e-06,
      "loss": 0.4158,
      "step": 3669
    },
    {
      "epoch": 0.16841815428387866,
      "grad_norm": 0.48965662717819214,
      "learning_rate": 9.966497879997097e-06,
      "loss": 0.4883,
      "step": 3670
    },
    {
      "epoch": 0.16846404478913313,
      "grad_norm": 0.5050069689750671,
      "learning_rate": 9.96646953816056e-06,
      "loss": 0.4985,
      "step": 3671
    },
    {
      "epoch": 0.1685099352943876,
      "grad_norm": 0.4886408746242523,
      "learning_rate": 9.966441184381235e-06,
      "loss": 0.5171,
      "step": 3672
    },
    {
      "epoch": 0.16855582579964207,
      "grad_norm": 0.5022122263908386,
      "learning_rate": 9.966412818659186e-06,
      "loss": 0.4599,
      "step": 3673
    },
    {
      "epoch": 0.1686017163048965,
      "grad_norm": 0.4836370646953583,
      "learning_rate": 9.96638444099448e-06,
      "loss": 0.4248,
      "step": 3674
    },
    {
      "epoch": 0.16864760681015098,
      "grad_norm": 0.4565540850162506,
      "learning_rate": 9.966356051387186e-06,
      "loss": 0.4002,
      "step": 3675
    },
    {
      "epoch": 0.16869349731540545,
      "grad_norm": 0.5088211297988892,
      "learning_rate": 9.966327649837376e-06,
      "loss": 0.4823,
      "step": 3676
    },
    {
      "epoch": 0.1687393878206599,
      "grad_norm": 0.4447498023509979,
      "learning_rate": 9.966299236345113e-06,
      "loss": 0.3399,
      "step": 3677
    },
    {
      "epoch": 0.16878527832591436,
      "grad_norm": 0.46449801325798035,
      "learning_rate": 9.96627081091047e-06,
      "loss": 0.3688,
      "step": 3678
    },
    {
      "epoch": 0.16883116883116883,
      "grad_norm": 0.5423493981361389,
      "learning_rate": 9.966242373533513e-06,
      "loss": 0.5368,
      "step": 3679
    },
    {
      "epoch": 0.1688770593364233,
      "grad_norm": 0.4940808415412903,
      "learning_rate": 9.966213924214308e-06,
      "loss": 0.4598,
      "step": 3680
    },
    {
      "epoch": 0.16892294984167774,
      "grad_norm": 0.4560638964176178,
      "learning_rate": 9.966185462952929e-06,
      "loss": 0.3924,
      "step": 3681
    },
    {
      "epoch": 0.16896884034693221,
      "grad_norm": 0.4676917493343353,
      "learning_rate": 9.966156989749438e-06,
      "loss": 0.3839,
      "step": 3682
    },
    {
      "epoch": 0.16901473085218668,
      "grad_norm": 0.4943281412124634,
      "learning_rate": 9.96612850460391e-06,
      "loss": 0.4358,
      "step": 3683
    },
    {
      "epoch": 0.16906062135744115,
      "grad_norm": 0.48242637515068054,
      "learning_rate": 9.966100007516409e-06,
      "loss": 0.3759,
      "step": 3684
    },
    {
      "epoch": 0.1691065118626956,
      "grad_norm": 0.5070915222167969,
      "learning_rate": 9.966071498487004e-06,
      "loss": 0.4628,
      "step": 3685
    },
    {
      "epoch": 0.16915240236795007,
      "grad_norm": 0.49251246452331543,
      "learning_rate": 9.966042977515765e-06,
      "loss": 0.45,
      "step": 3686
    },
    {
      "epoch": 0.16919829287320454,
      "grad_norm": 0.5031691193580627,
      "learning_rate": 9.96601444460276e-06,
      "loss": 0.4651,
      "step": 3687
    },
    {
      "epoch": 0.169244183378459,
      "grad_norm": 0.43061497807502747,
      "learning_rate": 9.965985899748058e-06,
      "loss": 0.3171,
      "step": 3688
    },
    {
      "epoch": 0.16929007388371345,
      "grad_norm": 0.4535612165927887,
      "learning_rate": 9.965957342951726e-06,
      "loss": 0.3392,
      "step": 3689
    },
    {
      "epoch": 0.16933596438896792,
      "grad_norm": 0.46733710169792175,
      "learning_rate": 9.965928774213835e-06,
      "loss": 0.3838,
      "step": 3690
    },
    {
      "epoch": 0.1693818548942224,
      "grad_norm": 0.4973802864551544,
      "learning_rate": 9.965900193534452e-06,
      "loss": 0.3731,
      "step": 3691
    },
    {
      "epoch": 0.16942774539947686,
      "grad_norm": 0.501643180847168,
      "learning_rate": 9.965871600913646e-06,
      "loss": 0.4404,
      "step": 3692
    },
    {
      "epoch": 0.1694736359047313,
      "grad_norm": 0.5125111937522888,
      "learning_rate": 9.965842996351487e-06,
      "loss": 0.4333,
      "step": 3693
    },
    {
      "epoch": 0.16951952640998577,
      "grad_norm": 0.475371778011322,
      "learning_rate": 9.965814379848043e-06,
      "loss": 0.4084,
      "step": 3694
    },
    {
      "epoch": 0.16956541691524024,
      "grad_norm": 0.4575386345386505,
      "learning_rate": 9.965785751403383e-06,
      "loss": 0.3676,
      "step": 3695
    },
    {
      "epoch": 0.1696113074204947,
      "grad_norm": 0.45974910259246826,
      "learning_rate": 9.965757111017574e-06,
      "loss": 0.3816,
      "step": 3696
    },
    {
      "epoch": 0.16965719792574915,
      "grad_norm": 0.4485854506492615,
      "learning_rate": 9.965728458690687e-06,
      "loss": 0.3579,
      "step": 3697
    },
    {
      "epoch": 0.16970308843100362,
      "grad_norm": 0.5038155317306519,
      "learning_rate": 9.96569979442279e-06,
      "loss": 0.4234,
      "step": 3698
    },
    {
      "epoch": 0.1697489789362581,
      "grad_norm": 0.45121437311172485,
      "learning_rate": 9.965671118213954e-06,
      "loss": 0.4015,
      "step": 3699
    },
    {
      "epoch": 0.16979486944151256,
      "grad_norm": 0.4932691156864166,
      "learning_rate": 9.965642430064245e-06,
      "loss": 0.4586,
      "step": 3700
    },
    {
      "epoch": 0.169840759946767,
      "grad_norm": 0.4838521480560303,
      "learning_rate": 9.965613729973731e-06,
      "loss": 0.4324,
      "step": 3701
    },
    {
      "epoch": 0.16988665045202148,
      "grad_norm": 0.4346429705619812,
      "learning_rate": 9.965585017942484e-06,
      "loss": 0.3272,
      "step": 3702
    },
    {
      "epoch": 0.16993254095727595,
      "grad_norm": 0.5014439225196838,
      "learning_rate": 9.965556293970575e-06,
      "loss": 0.4157,
      "step": 3703
    },
    {
      "epoch": 0.16997843146253042,
      "grad_norm": 0.5021127462387085,
      "learning_rate": 9.965527558058067e-06,
      "loss": 0.4221,
      "step": 3704
    },
    {
      "epoch": 0.17002432196778486,
      "grad_norm": 0.4837076961994171,
      "learning_rate": 9.965498810205035e-06,
      "loss": 0.4338,
      "step": 3705
    },
    {
      "epoch": 0.17007021247303933,
      "grad_norm": 0.4779275953769684,
      "learning_rate": 9.965470050411545e-06,
      "loss": 0.4311,
      "step": 3706
    },
    {
      "epoch": 0.1701161029782938,
      "grad_norm": 0.5239678025245667,
      "learning_rate": 9.965441278677665e-06,
      "loss": 0.4465,
      "step": 3707
    },
    {
      "epoch": 0.17016199348354824,
      "grad_norm": 0.4581717848777771,
      "learning_rate": 9.965412495003468e-06,
      "loss": 0.4253,
      "step": 3708
    },
    {
      "epoch": 0.1702078839888027,
      "grad_norm": 0.4749666750431061,
      "learning_rate": 9.965383699389019e-06,
      "loss": 0.4275,
      "step": 3709
    },
    {
      "epoch": 0.17025377449405718,
      "grad_norm": 0.47646042704582214,
      "learning_rate": 9.96535489183439e-06,
      "loss": 0.3938,
      "step": 3710
    },
    {
      "epoch": 0.17029966499931165,
      "grad_norm": 0.5016754269599915,
      "learning_rate": 9.965326072339652e-06,
      "loss": 0.4693,
      "step": 3711
    },
    {
      "epoch": 0.1703455555045661,
      "grad_norm": 0.5509760975837708,
      "learning_rate": 9.96529724090487e-06,
      "loss": 0.5276,
      "step": 3712
    },
    {
      "epoch": 0.17039144600982056,
      "grad_norm": 0.5025253891944885,
      "learning_rate": 9.965268397530114e-06,
      "loss": 0.4199,
      "step": 3713
    },
    {
      "epoch": 0.17043733651507503,
      "grad_norm": 0.4659944772720337,
      "learning_rate": 9.965239542215457e-06,
      "loss": 0.3768,
      "step": 3714
    },
    {
      "epoch": 0.1704832270203295,
      "grad_norm": 0.4489624500274658,
      "learning_rate": 9.965210674960965e-06,
      "loss": 0.3799,
      "step": 3715
    },
    {
      "epoch": 0.17052911752558395,
      "grad_norm": 0.47398844361305237,
      "learning_rate": 9.965181795766707e-06,
      "loss": 0.4186,
      "step": 3716
    },
    {
      "epoch": 0.17057500803083842,
      "grad_norm": 0.48437100648880005,
      "learning_rate": 9.965152904632756e-06,
      "loss": 0.3324,
      "step": 3717
    },
    {
      "epoch": 0.17062089853609289,
      "grad_norm": 0.49503275752067566,
      "learning_rate": 9.965124001559177e-06,
      "loss": 0.4116,
      "step": 3718
    },
    {
      "epoch": 0.17066678904134736,
      "grad_norm": 0.5525345802307129,
      "learning_rate": 9.965095086546044e-06,
      "loss": 0.4694,
      "step": 3719
    },
    {
      "epoch": 0.1707126795466018,
      "grad_norm": 0.46774229407310486,
      "learning_rate": 9.965066159593424e-06,
      "loss": 0.4353,
      "step": 3720
    },
    {
      "epoch": 0.17075857005185627,
      "grad_norm": 0.4906098544597626,
      "learning_rate": 9.965037220701385e-06,
      "loss": 0.4723,
      "step": 3721
    },
    {
      "epoch": 0.17080446055711074,
      "grad_norm": 0.4435950815677643,
      "learning_rate": 9.96500826987e-06,
      "loss": 0.3678,
      "step": 3722
    },
    {
      "epoch": 0.1708503510623652,
      "grad_norm": 0.4562053382396698,
      "learning_rate": 9.964979307099337e-06,
      "loss": 0.3785,
      "step": 3723
    },
    {
      "epoch": 0.17089624156761965,
      "grad_norm": 0.5205396413803101,
      "learning_rate": 9.964950332389466e-06,
      "loss": 0.4253,
      "step": 3724
    },
    {
      "epoch": 0.17094213207287412,
      "grad_norm": 0.4992291033267975,
      "learning_rate": 9.964921345740457e-06,
      "loss": 0.5003,
      "step": 3725
    },
    {
      "epoch": 0.1709880225781286,
      "grad_norm": 0.49866196513175964,
      "learning_rate": 9.964892347152377e-06,
      "loss": 0.4234,
      "step": 3726
    },
    {
      "epoch": 0.17103391308338306,
      "grad_norm": 0.463359534740448,
      "learning_rate": 9.964863336625298e-06,
      "loss": 0.3827,
      "step": 3727
    },
    {
      "epoch": 0.1710798035886375,
      "grad_norm": 0.441690057516098,
      "learning_rate": 9.964834314159293e-06,
      "loss": 0.3189,
      "step": 3728
    },
    {
      "epoch": 0.17112569409389197,
      "grad_norm": 0.5331432223320007,
      "learning_rate": 9.964805279754425e-06,
      "loss": 0.4308,
      "step": 3729
    },
    {
      "epoch": 0.17117158459914644,
      "grad_norm": 0.5610324144363403,
      "learning_rate": 9.964776233410768e-06,
      "loss": 0.4735,
      "step": 3730
    },
    {
      "epoch": 0.1712174751044009,
      "grad_norm": 0.4304240643978119,
      "learning_rate": 9.964747175128393e-06,
      "loss": 0.3354,
      "step": 3731
    },
    {
      "epoch": 0.17126336560965535,
      "grad_norm": 0.4620472192764282,
      "learning_rate": 9.964718104907366e-06,
      "loss": 0.365,
      "step": 3732
    },
    {
      "epoch": 0.17130925611490982,
      "grad_norm": 0.44709792733192444,
      "learning_rate": 9.96468902274776e-06,
      "loss": 0.329,
      "step": 3733
    },
    {
      "epoch": 0.1713551466201643,
      "grad_norm": 0.4359210431575775,
      "learning_rate": 9.964659928649643e-06,
      "loss": 0.3637,
      "step": 3734
    },
    {
      "epoch": 0.17140103712541874,
      "grad_norm": 0.48796746134757996,
      "learning_rate": 9.964630822613085e-06,
      "loss": 0.4465,
      "step": 3735
    },
    {
      "epoch": 0.1714469276306732,
      "grad_norm": 0.48671266436576843,
      "learning_rate": 9.964601704638158e-06,
      "loss": 0.4265,
      "step": 3736
    },
    {
      "epoch": 0.17149281813592768,
      "grad_norm": 0.4906500279903412,
      "learning_rate": 9.96457257472493e-06,
      "loss": 0.4478,
      "step": 3737
    },
    {
      "epoch": 0.17153870864118215,
      "grad_norm": 0.46206560730934143,
      "learning_rate": 9.964543432873472e-06,
      "loss": 0.3835,
      "step": 3738
    },
    {
      "epoch": 0.1715845991464366,
      "grad_norm": 0.4673829674720764,
      "learning_rate": 9.964514279083855e-06,
      "loss": 0.4125,
      "step": 3739
    },
    {
      "epoch": 0.17163048965169106,
      "grad_norm": 0.4572446942329407,
      "learning_rate": 9.964485113356147e-06,
      "loss": 0.3757,
      "step": 3740
    },
    {
      "epoch": 0.17167638015694553,
      "grad_norm": 0.4328352212905884,
      "learning_rate": 9.96445593569042e-06,
      "loss": 0.3632,
      "step": 3741
    },
    {
      "epoch": 0.1717222706622,
      "grad_norm": 0.519863486289978,
      "learning_rate": 9.964426746086742e-06,
      "loss": 0.3733,
      "step": 3742
    },
    {
      "epoch": 0.17176816116745444,
      "grad_norm": 0.47234734892845154,
      "learning_rate": 9.964397544545185e-06,
      "loss": 0.3922,
      "step": 3743
    },
    {
      "epoch": 0.1718140516727089,
      "grad_norm": 0.4189852476119995,
      "learning_rate": 9.964368331065819e-06,
      "loss": 0.3346,
      "step": 3744
    },
    {
      "epoch": 0.17185994217796338,
      "grad_norm": 0.481424480676651,
      "learning_rate": 9.964339105648714e-06,
      "loss": 0.4253,
      "step": 3745
    },
    {
      "epoch": 0.17190583268321785,
      "grad_norm": 0.43244752287864685,
      "learning_rate": 9.96430986829394e-06,
      "loss": 0.3178,
      "step": 3746
    },
    {
      "epoch": 0.1719517231884723,
      "grad_norm": 0.50309157371521,
      "learning_rate": 9.964280619001567e-06,
      "loss": 0.4308,
      "step": 3747
    },
    {
      "epoch": 0.17199761369372676,
      "grad_norm": 0.4675474464893341,
      "learning_rate": 9.964251357771667e-06,
      "loss": 0.3889,
      "step": 3748
    },
    {
      "epoch": 0.17204350419898123,
      "grad_norm": 0.5053369998931885,
      "learning_rate": 9.96422208460431e-06,
      "loss": 0.4544,
      "step": 3749
    },
    {
      "epoch": 0.1720893947042357,
      "grad_norm": 0.4964584708213806,
      "learning_rate": 9.964192799499564e-06,
      "loss": 0.4483,
      "step": 3750
    },
    {
      "epoch": 0.17213528520949015,
      "grad_norm": 0.491644024848938,
      "learning_rate": 9.9641635024575e-06,
      "loss": 0.4092,
      "step": 3751
    },
    {
      "epoch": 0.17218117571474462,
      "grad_norm": 0.4788534939289093,
      "learning_rate": 9.964134193478191e-06,
      "loss": 0.375,
      "step": 3752
    },
    {
      "epoch": 0.1722270662199991,
      "grad_norm": 0.4660608172416687,
      "learning_rate": 9.964104872561706e-06,
      "loss": 0.3623,
      "step": 3753
    },
    {
      "epoch": 0.17227295672525356,
      "grad_norm": 0.4548393785953522,
      "learning_rate": 9.964075539708116e-06,
      "loss": 0.3486,
      "step": 3754
    },
    {
      "epoch": 0.172318847230508,
      "grad_norm": 0.5075151920318604,
      "learning_rate": 9.964046194917491e-06,
      "loss": 0.4984,
      "step": 3755
    },
    {
      "epoch": 0.17236473773576247,
      "grad_norm": 0.4920263886451721,
      "learning_rate": 9.9640168381899e-06,
      "loss": 0.4409,
      "step": 3756
    },
    {
      "epoch": 0.17241062824101694,
      "grad_norm": 0.4510353207588196,
      "learning_rate": 9.963987469525415e-06,
      "loss": 0.3959,
      "step": 3757
    },
    {
      "epoch": 0.1724565187462714,
      "grad_norm": 0.49358996748924255,
      "learning_rate": 9.963958088924109e-06,
      "loss": 0.4045,
      "step": 3758
    },
    {
      "epoch": 0.17250240925152585,
      "grad_norm": 0.4904539883136749,
      "learning_rate": 9.96392869638605e-06,
      "loss": 0.4491,
      "step": 3759
    },
    {
      "epoch": 0.17254829975678032,
      "grad_norm": 0.45569685101509094,
      "learning_rate": 9.963899291911308e-06,
      "loss": 0.4318,
      "step": 3760
    },
    {
      "epoch": 0.1725941902620348,
      "grad_norm": 0.4705761671066284,
      "learning_rate": 9.963869875499956e-06,
      "loss": 0.3506,
      "step": 3761
    },
    {
      "epoch": 0.17264008076728926,
      "grad_norm": 0.4552665650844574,
      "learning_rate": 9.963840447152063e-06,
      "loss": 0.4337,
      "step": 3762
    },
    {
      "epoch": 0.1726859712725437,
      "grad_norm": 0.4712662696838379,
      "learning_rate": 9.9638110068677e-06,
      "loss": 0.3753,
      "step": 3763
    },
    {
      "epoch": 0.17273186177779817,
      "grad_norm": 0.44561973214149475,
      "learning_rate": 9.96378155464694e-06,
      "loss": 0.3517,
      "step": 3764
    },
    {
      "epoch": 0.17277775228305264,
      "grad_norm": 0.50220787525177,
      "learning_rate": 9.96375209048985e-06,
      "loss": 0.4737,
      "step": 3765
    },
    {
      "epoch": 0.17282364278830709,
      "grad_norm": 0.459413081407547,
      "learning_rate": 9.963722614396503e-06,
      "loss": 0.4116,
      "step": 3766
    },
    {
      "epoch": 0.17286953329356156,
      "grad_norm": 0.4793335497379303,
      "learning_rate": 9.963693126366972e-06,
      "loss": 0.3843,
      "step": 3767
    },
    {
      "epoch": 0.17291542379881603,
      "grad_norm": 0.4663759171962738,
      "learning_rate": 9.963663626401323e-06,
      "loss": 0.4197,
      "step": 3768
    },
    {
      "epoch": 0.1729613143040705,
      "grad_norm": 0.4310791492462158,
      "learning_rate": 9.963634114499629e-06,
      "loss": 0.358,
      "step": 3769
    },
    {
      "epoch": 0.17300720480932494,
      "grad_norm": 0.4253799617290497,
      "learning_rate": 9.963604590661964e-06,
      "loss": 0.3484,
      "step": 3770
    },
    {
      "epoch": 0.1730530953145794,
      "grad_norm": 0.5067339539527893,
      "learning_rate": 9.963575054888395e-06,
      "loss": 0.4284,
      "step": 3771
    },
    {
      "epoch": 0.17309898581983388,
      "grad_norm": 0.5011279582977295,
      "learning_rate": 9.963545507178995e-06,
      "loss": 0.41,
      "step": 3772
    },
    {
      "epoch": 0.17314487632508835,
      "grad_norm": 0.45513442158699036,
      "learning_rate": 9.963515947533835e-06,
      "loss": 0.4058,
      "step": 3773
    },
    {
      "epoch": 0.1731907668303428,
      "grad_norm": 0.5227753520011902,
      "learning_rate": 9.963486375952984e-06,
      "loss": 0.4831,
      "step": 3774
    },
    {
      "epoch": 0.17323665733559726,
      "grad_norm": 0.4769015610218048,
      "learning_rate": 9.963456792436517e-06,
      "loss": 0.3781,
      "step": 3775
    },
    {
      "epoch": 0.17328254784085173,
      "grad_norm": 0.4975082278251648,
      "learning_rate": 9.9634271969845e-06,
      "loss": 0.364,
      "step": 3776
    },
    {
      "epoch": 0.1733284383461062,
      "grad_norm": 0.4561043381690979,
      "learning_rate": 9.96339758959701e-06,
      "loss": 0.392,
      "step": 3777
    },
    {
      "epoch": 0.17337432885136064,
      "grad_norm": 0.5077850222587585,
      "learning_rate": 9.963367970274115e-06,
      "loss": 0.4662,
      "step": 3778
    },
    {
      "epoch": 0.1734202193566151,
      "grad_norm": 0.5776070952415466,
      "learning_rate": 9.963338339015886e-06,
      "loss": 0.492,
      "step": 3779
    },
    {
      "epoch": 0.17346610986186958,
      "grad_norm": 0.49255937337875366,
      "learning_rate": 9.963308695822394e-06,
      "loss": 0.448,
      "step": 3780
    },
    {
      "epoch": 0.17351200036712405,
      "grad_norm": 0.4622495472431183,
      "learning_rate": 9.963279040693712e-06,
      "loss": 0.4243,
      "step": 3781
    },
    {
      "epoch": 0.1735578908723785,
      "grad_norm": 0.45348840951919556,
      "learning_rate": 9.96324937362991e-06,
      "loss": 0.3295,
      "step": 3782
    },
    {
      "epoch": 0.17360378137763297,
      "grad_norm": 0.46781226992607117,
      "learning_rate": 9.96321969463106e-06,
      "loss": 0.4001,
      "step": 3783
    },
    {
      "epoch": 0.17364967188288744,
      "grad_norm": 0.5318712592124939,
      "learning_rate": 9.963190003697232e-06,
      "loss": 0.5383,
      "step": 3784
    },
    {
      "epoch": 0.1736955623881419,
      "grad_norm": 0.520931601524353,
      "learning_rate": 9.963160300828498e-06,
      "loss": 0.4828,
      "step": 3785
    },
    {
      "epoch": 0.17374145289339635,
      "grad_norm": 0.4373350143432617,
      "learning_rate": 9.963130586024932e-06,
      "loss": 0.3647,
      "step": 3786
    },
    {
      "epoch": 0.17378734339865082,
      "grad_norm": 0.468069851398468,
      "learning_rate": 9.963100859286603e-06,
      "loss": 0.4065,
      "step": 3787
    },
    {
      "epoch": 0.1738332339039053,
      "grad_norm": 0.44646987318992615,
      "learning_rate": 9.963071120613582e-06,
      "loss": 0.3698,
      "step": 3788
    },
    {
      "epoch": 0.17387912440915976,
      "grad_norm": 0.4330134689807892,
      "learning_rate": 9.963041370005941e-06,
      "loss": 0.3641,
      "step": 3789
    },
    {
      "epoch": 0.1739250149144142,
      "grad_norm": 0.4948446452617645,
      "learning_rate": 9.963011607463754e-06,
      "loss": 0.4567,
      "step": 3790
    },
    {
      "epoch": 0.17397090541966867,
      "grad_norm": 0.47151556611061096,
      "learning_rate": 9.962981832987088e-06,
      "loss": 0.4281,
      "step": 3791
    },
    {
      "epoch": 0.17401679592492314,
      "grad_norm": 0.6274124383926392,
      "learning_rate": 9.962952046576017e-06,
      "loss": 0.4597,
      "step": 3792
    },
    {
      "epoch": 0.1740626864301776,
      "grad_norm": 0.46481117606163025,
      "learning_rate": 9.962922248230616e-06,
      "loss": 0.3707,
      "step": 3793
    },
    {
      "epoch": 0.17410857693543205,
      "grad_norm": 0.4411012828350067,
      "learning_rate": 9.96289243795095e-06,
      "loss": 0.3539,
      "step": 3794
    },
    {
      "epoch": 0.17415446744068652,
      "grad_norm": 0.49497660994529724,
      "learning_rate": 9.962862615737094e-06,
      "loss": 0.4638,
      "step": 3795
    },
    {
      "epoch": 0.174200357945941,
      "grad_norm": 0.4659208655357361,
      "learning_rate": 9.962832781589123e-06,
      "loss": 0.4386,
      "step": 3796
    },
    {
      "epoch": 0.17424624845119543,
      "grad_norm": 0.4618721008300781,
      "learning_rate": 9.962802935507102e-06,
      "loss": 0.3876,
      "step": 3797
    },
    {
      "epoch": 0.1742921389564499,
      "grad_norm": 0.460696280002594,
      "learning_rate": 9.962773077491107e-06,
      "loss": 0.3844,
      "step": 3798
    },
    {
      "epoch": 0.17433802946170437,
      "grad_norm": 0.48609527945518494,
      "learning_rate": 9.96274320754121e-06,
      "loss": 0.4433,
      "step": 3799
    },
    {
      "epoch": 0.17438391996695884,
      "grad_norm": 0.5098398923873901,
      "learning_rate": 9.96271332565748e-06,
      "loss": 0.4815,
      "step": 3800
    },
    {
      "epoch": 0.1744298104722133,
      "grad_norm": 0.40706196427345276,
      "learning_rate": 9.962683431839994e-06,
      "loss": 0.2788,
      "step": 3801
    },
    {
      "epoch": 0.17447570097746776,
      "grad_norm": 0.5089811682701111,
      "learning_rate": 9.96265352608882e-06,
      "loss": 0.4467,
      "step": 3802
    },
    {
      "epoch": 0.17452159148272223,
      "grad_norm": 0.4700194299221039,
      "learning_rate": 9.962623608404028e-06,
      "loss": 0.4038,
      "step": 3803
    },
    {
      "epoch": 0.1745674819879767,
      "grad_norm": 0.5290846824645996,
      "learning_rate": 9.962593678785693e-06,
      "loss": 0.5268,
      "step": 3804
    },
    {
      "epoch": 0.17461337249323114,
      "grad_norm": 0.4735446870326996,
      "learning_rate": 9.962563737233888e-06,
      "loss": 0.3945,
      "step": 3805
    },
    {
      "epoch": 0.1746592629984856,
      "grad_norm": 0.49044692516326904,
      "learning_rate": 9.962533783748682e-06,
      "loss": 0.4818,
      "step": 3806
    },
    {
      "epoch": 0.17470515350374008,
      "grad_norm": 0.43118026852607727,
      "learning_rate": 9.96250381833015e-06,
      "loss": 0.3217,
      "step": 3807
    },
    {
      "epoch": 0.17475104400899455,
      "grad_norm": 0.49126771092414856,
      "learning_rate": 9.962473840978362e-06,
      "loss": 0.4709,
      "step": 3808
    },
    {
      "epoch": 0.174796934514249,
      "grad_norm": 0.46345171332359314,
      "learning_rate": 9.96244385169339e-06,
      "loss": 0.4144,
      "step": 3809
    },
    {
      "epoch": 0.17484282501950346,
      "grad_norm": 0.4336242377758026,
      "learning_rate": 9.962413850475306e-06,
      "loss": 0.3282,
      "step": 3810
    },
    {
      "epoch": 0.17488871552475793,
      "grad_norm": 0.48221275210380554,
      "learning_rate": 9.962383837324184e-06,
      "loss": 0.4674,
      "step": 3811
    },
    {
      "epoch": 0.1749346060300124,
      "grad_norm": 0.4423587918281555,
      "learning_rate": 9.962353812240096e-06,
      "loss": 0.3312,
      "step": 3812
    },
    {
      "epoch": 0.17498049653526684,
      "grad_norm": 0.47093838453292847,
      "learning_rate": 9.962323775223111e-06,
      "loss": 0.3988,
      "step": 3813
    },
    {
      "epoch": 0.17502638704052131,
      "grad_norm": 0.4595012366771698,
      "learning_rate": 9.962293726273305e-06,
      "loss": 0.4156,
      "step": 3814
    },
    {
      "epoch": 0.17507227754577578,
      "grad_norm": 0.4803919792175293,
      "learning_rate": 9.962263665390748e-06,
      "loss": 0.4951,
      "step": 3815
    },
    {
      "epoch": 0.17511816805103025,
      "grad_norm": 0.4672667980194092,
      "learning_rate": 9.962233592575515e-06,
      "loss": 0.4091,
      "step": 3816
    },
    {
      "epoch": 0.1751640585562847,
      "grad_norm": 0.4833678901195526,
      "learning_rate": 9.962203507827674e-06,
      "loss": 0.3736,
      "step": 3817
    },
    {
      "epoch": 0.17520994906153917,
      "grad_norm": 0.4316823482513428,
      "learning_rate": 9.9621734111473e-06,
      "loss": 0.3018,
      "step": 3818
    },
    {
      "epoch": 0.17525583956679364,
      "grad_norm": 0.47848308086395264,
      "learning_rate": 9.962143302534466e-06,
      "loss": 0.4222,
      "step": 3819
    },
    {
      "epoch": 0.1753017300720481,
      "grad_norm": 0.5713353157043457,
      "learning_rate": 9.962113181989243e-06,
      "loss": 0.4904,
      "step": 3820
    },
    {
      "epoch": 0.17534762057730255,
      "grad_norm": 0.5381240248680115,
      "learning_rate": 9.962083049511704e-06,
      "loss": 0.4708,
      "step": 3821
    },
    {
      "epoch": 0.17539351108255702,
      "grad_norm": 0.47154226899147034,
      "learning_rate": 9.96205290510192e-06,
      "loss": 0.4344,
      "step": 3822
    },
    {
      "epoch": 0.1754394015878115,
      "grad_norm": 0.4654196500778198,
      "learning_rate": 9.962022748759967e-06,
      "loss": 0.4286,
      "step": 3823
    },
    {
      "epoch": 0.17548529209306593,
      "grad_norm": 0.4943636357784271,
      "learning_rate": 9.961992580485913e-06,
      "loss": 0.4273,
      "step": 3824
    },
    {
      "epoch": 0.1755311825983204,
      "grad_norm": 0.4802487790584564,
      "learning_rate": 9.961962400279834e-06,
      "loss": 0.4116,
      "step": 3825
    },
    {
      "epoch": 0.17557707310357487,
      "grad_norm": 0.4972319006919861,
      "learning_rate": 9.961932208141802e-06,
      "loss": 0.4426,
      "step": 3826
    },
    {
      "epoch": 0.17562296360882934,
      "grad_norm": 0.4343493580818176,
      "learning_rate": 9.96190200407189e-06,
      "loss": 0.3951,
      "step": 3827
    },
    {
      "epoch": 0.17566885411408378,
      "grad_norm": 0.4805845618247986,
      "learning_rate": 9.961871788070168e-06,
      "loss": 0.404,
      "step": 3828
    },
    {
      "epoch": 0.17571474461933825,
      "grad_norm": 0.45114660263061523,
      "learning_rate": 9.961841560136712e-06,
      "loss": 0.3356,
      "step": 3829
    },
    {
      "epoch": 0.17576063512459272,
      "grad_norm": 0.49260181188583374,
      "learning_rate": 9.96181132027159e-06,
      "loss": 0.4555,
      "step": 3830
    },
    {
      "epoch": 0.1758065256298472,
      "grad_norm": 0.48737210035324097,
      "learning_rate": 9.96178106847488e-06,
      "loss": 0.3995,
      "step": 3831
    },
    {
      "epoch": 0.17585241613510164,
      "grad_norm": 0.4794608950614929,
      "learning_rate": 9.961750804746654e-06,
      "loss": 0.3866,
      "step": 3832
    },
    {
      "epoch": 0.1758983066403561,
      "grad_norm": 0.4778745472431183,
      "learning_rate": 9.96172052908698e-06,
      "loss": 0.4369,
      "step": 3833
    },
    {
      "epoch": 0.17594419714561058,
      "grad_norm": 0.5258398652076721,
      "learning_rate": 9.961690241495936e-06,
      "loss": 0.4323,
      "step": 3834
    },
    {
      "epoch": 0.17599008765086505,
      "grad_norm": 0.45457005500793457,
      "learning_rate": 9.961659941973592e-06,
      "loss": 0.3441,
      "step": 3835
    },
    {
      "epoch": 0.1760359781561195,
      "grad_norm": 0.46597379446029663,
      "learning_rate": 9.961629630520022e-06,
      "loss": 0.355,
      "step": 3836
    },
    {
      "epoch": 0.17608186866137396,
      "grad_norm": 0.5124558806419373,
      "learning_rate": 9.961599307135299e-06,
      "loss": 0.4844,
      "step": 3837
    },
    {
      "epoch": 0.17612775916662843,
      "grad_norm": 0.46185001730918884,
      "learning_rate": 9.961568971819495e-06,
      "loss": 0.3821,
      "step": 3838
    },
    {
      "epoch": 0.1761736496718829,
      "grad_norm": 0.44716793298721313,
      "learning_rate": 9.961538624572685e-06,
      "loss": 0.3264,
      "step": 3839
    },
    {
      "epoch": 0.17621954017713734,
      "grad_norm": 0.49822792410850525,
      "learning_rate": 9.96150826539494e-06,
      "loss": 0.4027,
      "step": 3840
    },
    {
      "epoch": 0.1762654306823918,
      "grad_norm": 0.4766508638858795,
      "learning_rate": 9.961477894286332e-06,
      "loss": 0.4056,
      "step": 3841
    },
    {
      "epoch": 0.17631132118764628,
      "grad_norm": 0.4933946132659912,
      "learning_rate": 9.961447511246936e-06,
      "loss": 0.4253,
      "step": 3842
    },
    {
      "epoch": 0.17635721169290075,
      "grad_norm": 0.4556927978992462,
      "learning_rate": 9.961417116276826e-06,
      "loss": 0.3572,
      "step": 3843
    },
    {
      "epoch": 0.1764031021981552,
      "grad_norm": 0.5002210140228271,
      "learning_rate": 9.961386709376072e-06,
      "loss": 0.4649,
      "step": 3844
    },
    {
      "epoch": 0.17644899270340966,
      "grad_norm": 0.4519858658313751,
      "learning_rate": 9.961356290544751e-06,
      "loss": 0.3925,
      "step": 3845
    },
    {
      "epoch": 0.17649488320866413,
      "grad_norm": 0.4758327603340149,
      "learning_rate": 9.961325859782933e-06,
      "loss": 0.3982,
      "step": 3846
    },
    {
      "epoch": 0.1765407737139186,
      "grad_norm": 0.4676792323589325,
      "learning_rate": 9.96129541709069e-06,
      "loss": 0.3477,
      "step": 3847
    },
    {
      "epoch": 0.17658666421917305,
      "grad_norm": 0.4864857494831085,
      "learning_rate": 9.9612649624681e-06,
      "loss": 0.3598,
      "step": 3848
    },
    {
      "epoch": 0.17663255472442752,
      "grad_norm": 0.49274805188179016,
      "learning_rate": 9.961234495915233e-06,
      "loss": 0.463,
      "step": 3849
    },
    {
      "epoch": 0.17667844522968199,
      "grad_norm": 0.4697285294532776,
      "learning_rate": 9.961204017432163e-06,
      "loss": 0.4385,
      "step": 3850
    },
    {
      "epoch": 0.17672433573493646,
      "grad_norm": 0.46031564474105835,
      "learning_rate": 9.961173527018963e-06,
      "loss": 0.3431,
      "step": 3851
    },
    {
      "epoch": 0.1767702262401909,
      "grad_norm": 0.4652436375617981,
      "learning_rate": 9.961143024675706e-06,
      "loss": 0.4305,
      "step": 3852
    },
    {
      "epoch": 0.17681611674544537,
      "grad_norm": 0.4586910307407379,
      "learning_rate": 9.961112510402467e-06,
      "loss": 0.4002,
      "step": 3853
    },
    {
      "epoch": 0.17686200725069984,
      "grad_norm": 0.46300554275512695,
      "learning_rate": 9.961081984199316e-06,
      "loss": 0.4463,
      "step": 3854
    },
    {
      "epoch": 0.17690789775595428,
      "grad_norm": 0.4883333444595337,
      "learning_rate": 9.96105144606633e-06,
      "loss": 0.4351,
      "step": 3855
    },
    {
      "epoch": 0.17695378826120875,
      "grad_norm": 0.4559125006198883,
      "learning_rate": 9.96102089600358e-06,
      "loss": 0.3931,
      "step": 3856
    },
    {
      "epoch": 0.17699967876646322,
      "grad_norm": 0.5298438668251038,
      "learning_rate": 9.960990334011142e-06,
      "loss": 0.4868,
      "step": 3857
    },
    {
      "epoch": 0.1770455692717177,
      "grad_norm": 0.43969202041625977,
      "learning_rate": 9.960959760089089e-06,
      "loss": 0.386,
      "step": 3858
    },
    {
      "epoch": 0.17709145977697213,
      "grad_norm": 0.4842301905155182,
      "learning_rate": 9.96092917423749e-06,
      "loss": 0.4328,
      "step": 3859
    },
    {
      "epoch": 0.1771373502822266,
      "grad_norm": 0.4557982087135315,
      "learning_rate": 9.960898576456425e-06,
      "loss": 0.3619,
      "step": 3860
    },
    {
      "epoch": 0.17718324078748107,
      "grad_norm": 0.48352086544036865,
      "learning_rate": 9.960867966745965e-06,
      "loss": 0.4453,
      "step": 3861
    },
    {
      "epoch": 0.17722913129273554,
      "grad_norm": 0.5317992568016052,
      "learning_rate": 9.96083734510618e-06,
      "loss": 0.4751,
      "step": 3862
    },
    {
      "epoch": 0.17727502179798998,
      "grad_norm": 0.4392800033092499,
      "learning_rate": 9.960806711537148e-06,
      "loss": 0.3603,
      "step": 3863
    },
    {
      "epoch": 0.17732091230324445,
      "grad_norm": 0.45197445154190063,
      "learning_rate": 9.960776066038943e-06,
      "loss": 0.388,
      "step": 3864
    },
    {
      "epoch": 0.17736680280849892,
      "grad_norm": 0.5064261555671692,
      "learning_rate": 9.960745408611635e-06,
      "loss": 0.4131,
      "step": 3865
    },
    {
      "epoch": 0.1774126933137534,
      "grad_norm": 0.4442690312862396,
      "learning_rate": 9.960714739255302e-06,
      "loss": 0.3958,
      "step": 3866
    },
    {
      "epoch": 0.17745858381900784,
      "grad_norm": 0.4621271789073944,
      "learning_rate": 9.960684057970016e-06,
      "loss": 0.4308,
      "step": 3867
    },
    {
      "epoch": 0.1775044743242623,
      "grad_norm": 0.48386305570602417,
      "learning_rate": 9.960653364755849e-06,
      "loss": 0.4659,
      "step": 3868
    },
    {
      "epoch": 0.17755036482951678,
      "grad_norm": 0.4527401328086853,
      "learning_rate": 9.960622659612877e-06,
      "loss": 0.3876,
      "step": 3869
    },
    {
      "epoch": 0.17759625533477125,
      "grad_norm": 0.4745809733867645,
      "learning_rate": 9.960591942541173e-06,
      "loss": 0.458,
      "step": 3870
    },
    {
      "epoch": 0.1776421458400257,
      "grad_norm": 0.4915264844894409,
      "learning_rate": 9.960561213540812e-06,
      "loss": 0.5,
      "step": 3871
    },
    {
      "epoch": 0.17768803634528016,
      "grad_norm": 0.4707813560962677,
      "learning_rate": 9.960530472611865e-06,
      "loss": 0.4334,
      "step": 3872
    },
    {
      "epoch": 0.17773392685053463,
      "grad_norm": 0.4973183274269104,
      "learning_rate": 9.960499719754408e-06,
      "loss": 0.4591,
      "step": 3873
    },
    {
      "epoch": 0.1777798173557891,
      "grad_norm": 0.4460873603820801,
      "learning_rate": 9.960468954968516e-06,
      "loss": 0.3125,
      "step": 3874
    },
    {
      "epoch": 0.17782570786104354,
      "grad_norm": 0.51996910572052,
      "learning_rate": 9.960438178254261e-06,
      "loss": 0.5235,
      "step": 3875
    },
    {
      "epoch": 0.177871598366298,
      "grad_norm": 0.45582249760627747,
      "learning_rate": 9.960407389611718e-06,
      "loss": 0.3608,
      "step": 3876
    },
    {
      "epoch": 0.17791748887155248,
      "grad_norm": 0.470163494348526,
      "learning_rate": 9.960376589040962e-06,
      "loss": 0.4331,
      "step": 3877
    },
    {
      "epoch": 0.17796337937680695,
      "grad_norm": 0.49415069818496704,
      "learning_rate": 9.960345776542062e-06,
      "loss": 0.4714,
      "step": 3878
    },
    {
      "epoch": 0.1780092698820614,
      "grad_norm": 0.5034629106521606,
      "learning_rate": 9.9603149521151e-06,
      "loss": 0.514,
      "step": 3879
    },
    {
      "epoch": 0.17805516038731586,
      "grad_norm": 0.4771920144557953,
      "learning_rate": 9.960284115760144e-06,
      "loss": 0.3596,
      "step": 3880
    },
    {
      "epoch": 0.17810105089257033,
      "grad_norm": 0.47345420718193054,
      "learning_rate": 9.96025326747727e-06,
      "loss": 0.4061,
      "step": 3881
    },
    {
      "epoch": 0.17814694139782478,
      "grad_norm": 0.44490885734558105,
      "learning_rate": 9.960222407266553e-06,
      "loss": 0.3568,
      "step": 3882
    },
    {
      "epoch": 0.17819283190307925,
      "grad_norm": 0.5291013121604919,
      "learning_rate": 9.960191535128067e-06,
      "loss": 0.4777,
      "step": 3883
    },
    {
      "epoch": 0.17823872240833372,
      "grad_norm": 0.4483031630516052,
      "learning_rate": 9.960160651061886e-06,
      "loss": 0.3457,
      "step": 3884
    },
    {
      "epoch": 0.1782846129135882,
      "grad_norm": 0.48007747530937195,
      "learning_rate": 9.960129755068084e-06,
      "loss": 0.439,
      "step": 3885
    },
    {
      "epoch": 0.17833050341884263,
      "grad_norm": 0.44911623001098633,
      "learning_rate": 9.960098847146734e-06,
      "loss": 0.3728,
      "step": 3886
    },
    {
      "epoch": 0.1783763939240971,
      "grad_norm": 0.46962645649909973,
      "learning_rate": 9.960067927297912e-06,
      "loss": 0.3574,
      "step": 3887
    },
    {
      "epoch": 0.17842228442935157,
      "grad_norm": 0.44819939136505127,
      "learning_rate": 9.960036995521694e-06,
      "loss": 0.3183,
      "step": 3888
    },
    {
      "epoch": 0.17846817493460604,
      "grad_norm": 0.471474826335907,
      "learning_rate": 9.960006051818151e-06,
      "loss": 0.4022,
      "step": 3889
    },
    {
      "epoch": 0.17851406543986048,
      "grad_norm": 0.4894413948059082,
      "learning_rate": 9.959975096187358e-06,
      "loss": 0.4282,
      "step": 3890
    },
    {
      "epoch": 0.17855995594511495,
      "grad_norm": 0.47600311040878296,
      "learning_rate": 9.959944128629393e-06,
      "loss": 0.4284,
      "step": 3891
    },
    {
      "epoch": 0.17860584645036942,
      "grad_norm": 0.4736107587814331,
      "learning_rate": 9.959913149144325e-06,
      "loss": 0.4115,
      "step": 3892
    },
    {
      "epoch": 0.1786517369556239,
      "grad_norm": 0.48581165075302124,
      "learning_rate": 9.959882157732233e-06,
      "loss": 0.4666,
      "step": 3893
    },
    {
      "epoch": 0.17869762746087833,
      "grad_norm": 0.6128531098365784,
      "learning_rate": 9.95985115439319e-06,
      "loss": 0.4254,
      "step": 3894
    },
    {
      "epoch": 0.1787435179661328,
      "grad_norm": 0.4596128463745117,
      "learning_rate": 9.95982013912727e-06,
      "loss": 0.353,
      "step": 3895
    },
    {
      "epoch": 0.17878940847138727,
      "grad_norm": 0.4830702543258667,
      "learning_rate": 9.959789111934546e-06,
      "loss": 0.3903,
      "step": 3896
    },
    {
      "epoch": 0.17883529897664174,
      "grad_norm": 0.4682193696498871,
      "learning_rate": 9.959758072815097e-06,
      "loss": 0.3927,
      "step": 3897
    },
    {
      "epoch": 0.17888118948189619,
      "grad_norm": 0.6033740043640137,
      "learning_rate": 9.959727021768993e-06,
      "loss": 0.4599,
      "step": 3898
    },
    {
      "epoch": 0.17892707998715066,
      "grad_norm": 0.4633723795413971,
      "learning_rate": 9.959695958796312e-06,
      "loss": 0.3477,
      "step": 3899
    },
    {
      "epoch": 0.17897297049240513,
      "grad_norm": 0.4755842089653015,
      "learning_rate": 9.95966488389713e-06,
      "loss": 0.4074,
      "step": 3900
    },
    {
      "epoch": 0.1790188609976596,
      "grad_norm": 0.4634113311767578,
      "learning_rate": 9.959633797071516e-06,
      "loss": 0.362,
      "step": 3901
    },
    {
      "epoch": 0.17906475150291404,
      "grad_norm": 0.9781712293624878,
      "learning_rate": 9.959602698319548e-06,
      "loss": 0.4318,
      "step": 3902
    },
    {
      "epoch": 0.1791106420081685,
      "grad_norm": 0.4694291651248932,
      "learning_rate": 9.959571587641302e-06,
      "loss": 0.4251,
      "step": 3903
    },
    {
      "epoch": 0.17915653251342298,
      "grad_norm": 0.48279836773872375,
      "learning_rate": 9.959540465036853e-06,
      "loss": 0.4014,
      "step": 3904
    },
    {
      "epoch": 0.17920242301867745,
      "grad_norm": 0.47537532448768616,
      "learning_rate": 9.959509330506272e-06,
      "loss": 0.4127,
      "step": 3905
    },
    {
      "epoch": 0.1792483135239319,
      "grad_norm": 0.46119365096092224,
      "learning_rate": 9.959478184049636e-06,
      "loss": 0.4297,
      "step": 3906
    },
    {
      "epoch": 0.17929420402918636,
      "grad_norm": 0.44106653332710266,
      "learning_rate": 9.959447025667021e-06,
      "loss": 0.3989,
      "step": 3907
    },
    {
      "epoch": 0.17934009453444083,
      "grad_norm": 0.4463611841201782,
      "learning_rate": 9.959415855358503e-06,
      "loss": 0.3282,
      "step": 3908
    },
    {
      "epoch": 0.1793859850396953,
      "grad_norm": 0.4646354615688324,
      "learning_rate": 9.959384673124154e-06,
      "loss": 0.3682,
      "step": 3909
    },
    {
      "epoch": 0.17943187554494974,
      "grad_norm": 0.45753049850463867,
      "learning_rate": 9.95935347896405e-06,
      "loss": 0.3966,
      "step": 3910
    },
    {
      "epoch": 0.1794777660502042,
      "grad_norm": 0.4642256796360016,
      "learning_rate": 9.959322272878264e-06,
      "loss": 0.3782,
      "step": 3911
    },
    {
      "epoch": 0.17952365655545868,
      "grad_norm": 0.46932828426361084,
      "learning_rate": 9.959291054866876e-06,
      "loss": 0.3421,
      "step": 3912
    },
    {
      "epoch": 0.17956954706071312,
      "grad_norm": 0.4701588451862335,
      "learning_rate": 9.959259824929957e-06,
      "loss": 0.4411,
      "step": 3913
    },
    {
      "epoch": 0.1796154375659676,
      "grad_norm": 0.4810388386249542,
      "learning_rate": 9.959228583067583e-06,
      "loss": 0.4355,
      "step": 3914
    },
    {
      "epoch": 0.17966132807122207,
      "grad_norm": 0.4653016924858093,
      "learning_rate": 9.959197329279831e-06,
      "loss": 0.3771,
      "step": 3915
    },
    {
      "epoch": 0.17970721857647654,
      "grad_norm": 0.5119547843933105,
      "learning_rate": 9.959166063566771e-06,
      "loss": 0.5317,
      "step": 3916
    },
    {
      "epoch": 0.17975310908173098,
      "grad_norm": 0.44539543986320496,
      "learning_rate": 9.959134785928486e-06,
      "loss": 0.3539,
      "step": 3917
    },
    {
      "epoch": 0.17979899958698545,
      "grad_norm": 0.4559662938117981,
      "learning_rate": 9.959103496365045e-06,
      "loss": 0.3529,
      "step": 3918
    },
    {
      "epoch": 0.17984489009223992,
      "grad_norm": 0.5066208243370056,
      "learning_rate": 9.959072194876526e-06,
      "loss": 0.4485,
      "step": 3919
    },
    {
      "epoch": 0.1798907805974944,
      "grad_norm": 0.4939250349998474,
      "learning_rate": 9.959040881463002e-06,
      "loss": 0.4229,
      "step": 3920
    },
    {
      "epoch": 0.17993667110274883,
      "grad_norm": 0.42819133400917053,
      "learning_rate": 9.959009556124552e-06,
      "loss": 0.3316,
      "step": 3921
    },
    {
      "epoch": 0.1799825616080033,
      "grad_norm": 0.439520925283432,
      "learning_rate": 9.958978218861249e-06,
      "loss": 0.3549,
      "step": 3922
    },
    {
      "epoch": 0.18002845211325777,
      "grad_norm": 0.5061075091362,
      "learning_rate": 9.958946869673165e-06,
      "loss": 0.4384,
      "step": 3923
    },
    {
      "epoch": 0.18007434261851224,
      "grad_norm": 0.4657272696495056,
      "learning_rate": 9.958915508560382e-06,
      "loss": 0.3991,
      "step": 3924
    },
    {
      "epoch": 0.18012023312376668,
      "grad_norm": 0.5086844563484192,
      "learning_rate": 9.958884135522971e-06,
      "loss": 0.4468,
      "step": 3925
    },
    {
      "epoch": 0.18016612362902115,
      "grad_norm": 0.4692036211490631,
      "learning_rate": 9.95885275056101e-06,
      "loss": 0.3922,
      "step": 3926
    },
    {
      "epoch": 0.18021201413427562,
      "grad_norm": 0.509840190410614,
      "learning_rate": 9.958821353674573e-06,
      "loss": 0.5127,
      "step": 3927
    },
    {
      "epoch": 0.1802579046395301,
      "grad_norm": 0.4572141766548157,
      "learning_rate": 9.958789944863735e-06,
      "loss": 0.3947,
      "step": 3928
    },
    {
      "epoch": 0.18030379514478453,
      "grad_norm": 0.47842952609062195,
      "learning_rate": 9.958758524128572e-06,
      "loss": 0.3683,
      "step": 3929
    },
    {
      "epoch": 0.180349685650039,
      "grad_norm": 0.4818190932273865,
      "learning_rate": 9.95872709146916e-06,
      "loss": 0.4286,
      "step": 3930
    },
    {
      "epoch": 0.18039557615529347,
      "grad_norm": 0.4555456340312958,
      "learning_rate": 9.958695646885575e-06,
      "loss": 0.3942,
      "step": 3931
    },
    {
      "epoch": 0.18044146666054794,
      "grad_norm": 0.4639066159725189,
      "learning_rate": 9.958664190377892e-06,
      "loss": 0.4262,
      "step": 3932
    },
    {
      "epoch": 0.1804873571658024,
      "grad_norm": 0.5069988965988159,
      "learning_rate": 9.958632721946185e-06,
      "loss": 0.517,
      "step": 3933
    },
    {
      "epoch": 0.18053324767105686,
      "grad_norm": 0.4582611918449402,
      "learning_rate": 9.958601241590533e-06,
      "loss": 0.3656,
      "step": 3934
    },
    {
      "epoch": 0.18057913817631133,
      "grad_norm": 0.4548102021217346,
      "learning_rate": 9.958569749311009e-06,
      "loss": 0.4117,
      "step": 3935
    },
    {
      "epoch": 0.1806250286815658,
      "grad_norm": 0.4673822224140167,
      "learning_rate": 9.958538245107691e-06,
      "loss": 0.4217,
      "step": 3936
    },
    {
      "epoch": 0.18067091918682024,
      "grad_norm": 0.5158348679542542,
      "learning_rate": 9.958506728980652e-06,
      "loss": 0.4531,
      "step": 3937
    },
    {
      "epoch": 0.1807168096920747,
      "grad_norm": 0.4394562542438507,
      "learning_rate": 9.95847520092997e-06,
      "loss": 0.3664,
      "step": 3938
    },
    {
      "epoch": 0.18076270019732918,
      "grad_norm": 0.5189195275306702,
      "learning_rate": 9.95844366095572e-06,
      "loss": 0.4463,
      "step": 3939
    },
    {
      "epoch": 0.18080859070258365,
      "grad_norm": 0.479986310005188,
      "learning_rate": 9.958412109057978e-06,
      "loss": 0.5212,
      "step": 3940
    },
    {
      "epoch": 0.1808544812078381,
      "grad_norm": 0.4612562954425812,
      "learning_rate": 9.958380545236819e-06,
      "loss": 0.3943,
      "step": 3941
    },
    {
      "epoch": 0.18090037171309256,
      "grad_norm": 0.4503750503063202,
      "learning_rate": 9.958348969492321e-06,
      "loss": 0.3546,
      "step": 3942
    },
    {
      "epoch": 0.18094626221834703,
      "grad_norm": 0.4488202929496765,
      "learning_rate": 9.958317381824559e-06,
      "loss": 0.3579,
      "step": 3943
    },
    {
      "epoch": 0.18099215272360147,
      "grad_norm": 0.5286926627159119,
      "learning_rate": 9.958285782233607e-06,
      "loss": 0.406,
      "step": 3944
    },
    {
      "epoch": 0.18103804322885594,
      "grad_norm": 0.48888880014419556,
      "learning_rate": 9.958254170719543e-06,
      "loss": 0.4059,
      "step": 3945
    },
    {
      "epoch": 0.1810839337341104,
      "grad_norm": 0.5178829431533813,
      "learning_rate": 9.958222547282442e-06,
      "loss": 0.4931,
      "step": 3946
    },
    {
      "epoch": 0.18112982423936488,
      "grad_norm": 0.480390727519989,
      "learning_rate": 9.958190911922381e-06,
      "loss": 0.4159,
      "step": 3947
    },
    {
      "epoch": 0.18117571474461933,
      "grad_norm": 0.44573745131492615,
      "learning_rate": 9.958159264639437e-06,
      "loss": 0.3759,
      "step": 3948
    },
    {
      "epoch": 0.1812216052498738,
      "grad_norm": 0.502204179763794,
      "learning_rate": 9.958127605433685e-06,
      "loss": 0.4147,
      "step": 3949
    },
    {
      "epoch": 0.18126749575512827,
      "grad_norm": 0.4709887206554413,
      "learning_rate": 9.958095934305198e-06,
      "loss": 0.4046,
      "step": 3950
    },
    {
      "epoch": 0.18131338626038274,
      "grad_norm": 0.43431344628334045,
      "learning_rate": 9.958064251254059e-06,
      "loss": 0.3325,
      "step": 3951
    },
    {
      "epoch": 0.18135927676563718,
      "grad_norm": 0.5602325201034546,
      "learning_rate": 9.958032556280338e-06,
      "loss": 0.5714,
      "step": 3952
    },
    {
      "epoch": 0.18140516727089165,
      "grad_norm": 0.49465030431747437,
      "learning_rate": 9.958000849384113e-06,
      "loss": 0.3943,
      "step": 3953
    },
    {
      "epoch": 0.18145105777614612,
      "grad_norm": 0.5021671056747437,
      "learning_rate": 9.957969130565462e-06,
      "loss": 0.5217,
      "step": 3954
    },
    {
      "epoch": 0.1814969482814006,
      "grad_norm": 0.4833013713359833,
      "learning_rate": 9.957937399824458e-06,
      "loss": 0.3885,
      "step": 3955
    },
    {
      "epoch": 0.18154283878665503,
      "grad_norm": 0.4508078396320343,
      "learning_rate": 9.957905657161182e-06,
      "loss": 0.4219,
      "step": 3956
    },
    {
      "epoch": 0.1815887292919095,
      "grad_norm": 0.47512227296829224,
      "learning_rate": 9.957873902575706e-06,
      "loss": 0.4352,
      "step": 3957
    },
    {
      "epoch": 0.18163461979716397,
      "grad_norm": 0.4449382722377777,
      "learning_rate": 9.95784213606811e-06,
      "loss": 0.3642,
      "step": 3958
    },
    {
      "epoch": 0.18168051030241844,
      "grad_norm": 0.4501861333847046,
      "learning_rate": 9.957810357638466e-06,
      "loss": 0.3856,
      "step": 3959
    },
    {
      "epoch": 0.18172640080767288,
      "grad_norm": 0.5231090784072876,
      "learning_rate": 9.957778567286853e-06,
      "loss": 0.4758,
      "step": 3960
    },
    {
      "epoch": 0.18177229131292735,
      "grad_norm": 0.4330017864704132,
      "learning_rate": 9.957746765013348e-06,
      "loss": 0.3635,
      "step": 3961
    },
    {
      "epoch": 0.18181818181818182,
      "grad_norm": 0.45830225944519043,
      "learning_rate": 9.957714950818027e-06,
      "loss": 0.4342,
      "step": 3962
    },
    {
      "epoch": 0.1818640723234363,
      "grad_norm": 0.44776225090026855,
      "learning_rate": 9.957683124700966e-06,
      "loss": 0.3411,
      "step": 3963
    },
    {
      "epoch": 0.18190996282869074,
      "grad_norm": 0.4855400025844574,
      "learning_rate": 9.957651286662242e-06,
      "loss": 0.4379,
      "step": 3964
    },
    {
      "epoch": 0.1819558533339452,
      "grad_norm": 0.459383487701416,
      "learning_rate": 9.95761943670193e-06,
      "loss": 0.3399,
      "step": 3965
    },
    {
      "epoch": 0.18200174383919968,
      "grad_norm": 0.45101475715637207,
      "learning_rate": 9.95758757482011e-06,
      "loss": 0.3845,
      "step": 3966
    },
    {
      "epoch": 0.18204763434445415,
      "grad_norm": 0.46671363711357117,
      "learning_rate": 9.957555701016855e-06,
      "loss": 0.4073,
      "step": 3967
    },
    {
      "epoch": 0.1820935248497086,
      "grad_norm": 0.4768611490726471,
      "learning_rate": 9.957523815292243e-06,
      "loss": 0.4268,
      "step": 3968
    },
    {
      "epoch": 0.18213941535496306,
      "grad_norm": 0.4484013319015503,
      "learning_rate": 9.957491917646353e-06,
      "loss": 0.3557,
      "step": 3969
    },
    {
      "epoch": 0.18218530586021753,
      "grad_norm": 0.5023492574691772,
      "learning_rate": 9.957460008079257e-06,
      "loss": 0.4171,
      "step": 3970
    },
    {
      "epoch": 0.18223119636547197,
      "grad_norm": 0.513933002948761,
      "learning_rate": 9.957428086591037e-06,
      "loss": 0.5274,
      "step": 3971
    },
    {
      "epoch": 0.18227708687072644,
      "grad_norm": 0.46191951632499695,
      "learning_rate": 9.957396153181764e-06,
      "loss": 0.4412,
      "step": 3972
    },
    {
      "epoch": 0.1823229773759809,
      "grad_norm": 0.46174636483192444,
      "learning_rate": 9.957364207851521e-06,
      "loss": 0.3979,
      "step": 3973
    },
    {
      "epoch": 0.18236886788123538,
      "grad_norm": 0.48360326886177063,
      "learning_rate": 9.95733225060038e-06,
      "loss": 0.4211,
      "step": 3974
    },
    {
      "epoch": 0.18241475838648982,
      "grad_norm": 0.4586115777492523,
      "learning_rate": 9.95730028142842e-06,
      "loss": 0.3738,
      "step": 3975
    },
    {
      "epoch": 0.1824606488917443,
      "grad_norm": 0.47462958097457886,
      "learning_rate": 9.957268300335716e-06,
      "loss": 0.4475,
      "step": 3976
    },
    {
      "epoch": 0.18250653939699876,
      "grad_norm": 0.4541427195072174,
      "learning_rate": 9.957236307322346e-06,
      "loss": 0.3834,
      "step": 3977
    },
    {
      "epoch": 0.18255242990225323,
      "grad_norm": 0.4755142033100128,
      "learning_rate": 9.95720430238839e-06,
      "loss": 0.463,
      "step": 3978
    },
    {
      "epoch": 0.18259832040750767,
      "grad_norm": 0.4314226806163788,
      "learning_rate": 9.95717228553392e-06,
      "loss": 0.3392,
      "step": 3979
    },
    {
      "epoch": 0.18264421091276214,
      "grad_norm": 0.4666334092617035,
      "learning_rate": 9.957140256759016e-06,
      "loss": 0.4551,
      "step": 3980
    },
    {
      "epoch": 0.18269010141801661,
      "grad_norm": 0.4947948455810547,
      "learning_rate": 9.957108216063755e-06,
      "loss": 0.4132,
      "step": 3981
    },
    {
      "epoch": 0.18273599192327108,
      "grad_norm": 0.48704156279563904,
      "learning_rate": 9.957076163448212e-06,
      "loss": 0.4595,
      "step": 3982
    },
    {
      "epoch": 0.18278188242852553,
      "grad_norm": 0.48995763063430786,
      "learning_rate": 9.957044098912466e-06,
      "loss": 0.4638,
      "step": 3983
    },
    {
      "epoch": 0.18282777293378,
      "grad_norm": 0.47832152247428894,
      "learning_rate": 9.957012022456591e-06,
      "loss": 0.4046,
      "step": 3984
    },
    {
      "epoch": 0.18287366343903447,
      "grad_norm": 0.5112481713294983,
      "learning_rate": 9.95697993408067e-06,
      "loss": 0.3786,
      "step": 3985
    },
    {
      "epoch": 0.18291955394428894,
      "grad_norm": 0.487212598323822,
      "learning_rate": 9.956947833784776e-06,
      "loss": 0.4781,
      "step": 3986
    },
    {
      "epoch": 0.18296544444954338,
      "grad_norm": 0.4842546284198761,
      "learning_rate": 9.956915721568984e-06,
      "loss": 0.4206,
      "step": 3987
    },
    {
      "epoch": 0.18301133495479785,
      "grad_norm": 0.4674060344696045,
      "learning_rate": 9.956883597433378e-06,
      "loss": 0.4173,
      "step": 3988
    },
    {
      "epoch": 0.18305722546005232,
      "grad_norm": 0.5002169609069824,
      "learning_rate": 9.95685146137803e-06,
      "loss": 0.4517,
      "step": 3989
    },
    {
      "epoch": 0.1831031159653068,
      "grad_norm": 0.49130091071128845,
      "learning_rate": 9.956819313403018e-06,
      "loss": 0.4897,
      "step": 3990
    },
    {
      "epoch": 0.18314900647056123,
      "grad_norm": 0.46109530329704285,
      "learning_rate": 9.95678715350842e-06,
      "loss": 0.3323,
      "step": 3991
    },
    {
      "epoch": 0.1831948969758157,
      "grad_norm": 0.48416006565093994,
      "learning_rate": 9.956754981694315e-06,
      "loss": 0.4193,
      "step": 3992
    },
    {
      "epoch": 0.18324078748107017,
      "grad_norm": 0.5197775959968567,
      "learning_rate": 9.956722797960775e-06,
      "loss": 0.4927,
      "step": 3993
    },
    {
      "epoch": 0.18328667798632464,
      "grad_norm": 0.4878934919834137,
      "learning_rate": 9.956690602307885e-06,
      "loss": 0.4042,
      "step": 3994
    },
    {
      "epoch": 0.18333256849157908,
      "grad_norm": 0.5160571932792664,
      "learning_rate": 9.956658394735716e-06,
      "loss": 0.4567,
      "step": 3995
    },
    {
      "epoch": 0.18337845899683355,
      "grad_norm": 0.4523058235645294,
      "learning_rate": 9.95662617524435e-06,
      "loss": 0.353,
      "step": 3996
    },
    {
      "epoch": 0.18342434950208802,
      "grad_norm": 0.4639807343482971,
      "learning_rate": 9.956593943833861e-06,
      "loss": 0.3531,
      "step": 3997
    },
    {
      "epoch": 0.1834702400073425,
      "grad_norm": 0.4653351902961731,
      "learning_rate": 9.956561700504329e-06,
      "loss": 0.3933,
      "step": 3998
    },
    {
      "epoch": 0.18351613051259694,
      "grad_norm": 0.4264218807220459,
      "learning_rate": 9.95652944525583e-06,
      "loss": 0.3293,
      "step": 3999
    },
    {
      "epoch": 0.1835620210178514,
      "grad_norm": 0.6798258423805237,
      "learning_rate": 9.95649717808844e-06,
      "loss": 0.4556,
      "step": 4000
    },
    {
      "epoch": 0.18360791152310588,
      "grad_norm": 0.4980906844139099,
      "learning_rate": 9.95646489900224e-06,
      "loss": 0.4259,
      "step": 4001
    },
    {
      "epoch": 0.18365380202836032,
      "grad_norm": 0.48687276244163513,
      "learning_rate": 9.956432607997308e-06,
      "loss": 0.4209,
      "step": 4002
    },
    {
      "epoch": 0.1836996925336148,
      "grad_norm": 0.5508676767349243,
      "learning_rate": 9.956400305073717e-06,
      "loss": 0.533,
      "step": 4003
    },
    {
      "epoch": 0.18374558303886926,
      "grad_norm": 0.4933052361011505,
      "learning_rate": 9.95636799023155e-06,
      "loss": 0.3412,
      "step": 4004
    },
    {
      "epoch": 0.18379147354412373,
      "grad_norm": 0.5059148073196411,
      "learning_rate": 9.956335663470881e-06,
      "loss": 0.4447,
      "step": 4005
    },
    {
      "epoch": 0.18383736404937817,
      "grad_norm": 0.5061053037643433,
      "learning_rate": 9.95630332479179e-06,
      "loss": 0.4818,
      "step": 4006
    },
    {
      "epoch": 0.18388325455463264,
      "grad_norm": 0.4864663779735565,
      "learning_rate": 9.956270974194352e-06,
      "loss": 0.3813,
      "step": 4007
    },
    {
      "epoch": 0.1839291450598871,
      "grad_norm": 0.4997391700744629,
      "learning_rate": 9.956238611678648e-06,
      "loss": 0.4266,
      "step": 4008
    },
    {
      "epoch": 0.18397503556514158,
      "grad_norm": 0.47904402017593384,
      "learning_rate": 9.956206237244753e-06,
      "loss": 0.3989,
      "step": 4009
    },
    {
      "epoch": 0.18402092607039602,
      "grad_norm": 0.5069399476051331,
      "learning_rate": 9.956173850892747e-06,
      "loss": 0.4303,
      "step": 4010
    },
    {
      "epoch": 0.1840668165756505,
      "grad_norm": 0.4637645184993744,
      "learning_rate": 9.956141452622708e-06,
      "loss": 0.4011,
      "step": 4011
    },
    {
      "epoch": 0.18411270708090496,
      "grad_norm": 0.46542516350746155,
      "learning_rate": 9.956109042434712e-06,
      "loss": 0.3793,
      "step": 4012
    },
    {
      "epoch": 0.18415859758615943,
      "grad_norm": 0.48267537355422974,
      "learning_rate": 9.956076620328837e-06,
      "loss": 0.4578,
      "step": 4013
    },
    {
      "epoch": 0.18420448809141388,
      "grad_norm": 0.4549073278903961,
      "learning_rate": 9.956044186305164e-06,
      "loss": 0.3814,
      "step": 4014
    },
    {
      "epoch": 0.18425037859666835,
      "grad_norm": 0.5556150674819946,
      "learning_rate": 9.956011740363767e-06,
      "loss": 0.5544,
      "step": 4015
    },
    {
      "epoch": 0.18429626910192282,
      "grad_norm": 0.514133870601654,
      "learning_rate": 9.955979282504728e-06,
      "loss": 0.4205,
      "step": 4016
    },
    {
      "epoch": 0.18434215960717729,
      "grad_norm": 0.42726820707321167,
      "learning_rate": 9.955946812728121e-06,
      "loss": 0.3124,
      "step": 4017
    },
    {
      "epoch": 0.18438805011243173,
      "grad_norm": 0.4487590193748474,
      "learning_rate": 9.955914331034027e-06,
      "loss": 0.3693,
      "step": 4018
    },
    {
      "epoch": 0.1844339406176862,
      "grad_norm": 0.460810124874115,
      "learning_rate": 9.955881837422524e-06,
      "loss": 0.4237,
      "step": 4019
    },
    {
      "epoch": 0.18447983112294067,
      "grad_norm": 0.5685785412788391,
      "learning_rate": 9.955849331893688e-06,
      "loss": 0.5322,
      "step": 4020
    },
    {
      "epoch": 0.18452572162819514,
      "grad_norm": 0.4468558430671692,
      "learning_rate": 9.955816814447599e-06,
      "loss": 0.3774,
      "step": 4021
    },
    {
      "epoch": 0.18457161213344958,
      "grad_norm": 0.5469991564750671,
      "learning_rate": 9.955784285084334e-06,
      "loss": 0.5207,
      "step": 4022
    },
    {
      "epoch": 0.18461750263870405,
      "grad_norm": 0.48401325941085815,
      "learning_rate": 9.955751743803971e-06,
      "loss": 0.3991,
      "step": 4023
    },
    {
      "epoch": 0.18466339314395852,
      "grad_norm": 0.49764013290405273,
      "learning_rate": 9.955719190606591e-06,
      "loss": 0.3764,
      "step": 4024
    },
    {
      "epoch": 0.184709283649213,
      "grad_norm": 0.4192434549331665,
      "learning_rate": 9.955686625492271e-06,
      "loss": 0.2729,
      "step": 4025
    },
    {
      "epoch": 0.18475517415446743,
      "grad_norm": 0.5076939463615417,
      "learning_rate": 9.955654048461087e-06,
      "loss": 0.4315,
      "step": 4026
    },
    {
      "epoch": 0.1848010646597219,
      "grad_norm": 0.4300815463066101,
      "learning_rate": 9.95562145951312e-06,
      "loss": 0.3584,
      "step": 4027
    },
    {
      "epoch": 0.18484695516497637,
      "grad_norm": 0.48613524436950684,
      "learning_rate": 9.955588858648446e-06,
      "loss": 0.4223,
      "step": 4028
    },
    {
      "epoch": 0.18489284567023082,
      "grad_norm": 0.44557058811187744,
      "learning_rate": 9.955556245867147e-06,
      "loss": 0.2963,
      "step": 4029
    },
    {
      "epoch": 0.18493873617548529,
      "grad_norm": 0.4597506821155548,
      "learning_rate": 9.955523621169299e-06,
      "loss": 0.3922,
      "step": 4030
    },
    {
      "epoch": 0.18498462668073976,
      "grad_norm": 0.5106930136680603,
      "learning_rate": 9.955490984554978e-06,
      "loss": 0.3743,
      "step": 4031
    },
    {
      "epoch": 0.18503051718599423,
      "grad_norm": 0.49481138586997986,
      "learning_rate": 9.955458336024269e-06,
      "loss": 0.4458,
      "step": 4032
    },
    {
      "epoch": 0.18507640769124867,
      "grad_norm": 0.4394301474094391,
      "learning_rate": 9.955425675577244e-06,
      "loss": 0.3203,
      "step": 4033
    },
    {
      "epoch": 0.18512229819650314,
      "grad_norm": 0.4792270064353943,
      "learning_rate": 9.955393003213984e-06,
      "loss": 0.3465,
      "step": 4034
    },
    {
      "epoch": 0.1851681887017576,
      "grad_norm": 0.50316321849823,
      "learning_rate": 9.955360318934571e-06,
      "loss": 0.4486,
      "step": 4035
    },
    {
      "epoch": 0.18521407920701208,
      "grad_norm": 0.4733767807483673,
      "learning_rate": 9.955327622739077e-06,
      "loss": 0.3962,
      "step": 4036
    },
    {
      "epoch": 0.18525996971226652,
      "grad_norm": 0.5152449011802673,
      "learning_rate": 9.955294914627584e-06,
      "loss": 0.3869,
      "step": 4037
    },
    {
      "epoch": 0.185305860217521,
      "grad_norm": 0.5111094117164612,
      "learning_rate": 9.955262194600172e-06,
      "loss": 0.4123,
      "step": 4038
    },
    {
      "epoch": 0.18535175072277546,
      "grad_norm": 0.5918365716934204,
      "learning_rate": 9.95522946265692e-06,
      "loss": 0.4543,
      "step": 4039
    },
    {
      "epoch": 0.18539764122802993,
      "grad_norm": 0.5309789776802063,
      "learning_rate": 9.955196718797902e-06,
      "loss": 0.513,
      "step": 4040
    },
    {
      "epoch": 0.18544353173328437,
      "grad_norm": 0.4646918475627899,
      "learning_rate": 9.955163963023203e-06,
      "loss": 0.3561,
      "step": 4041
    },
    {
      "epoch": 0.18548942223853884,
      "grad_norm": 0.44384002685546875,
      "learning_rate": 9.955131195332894e-06,
      "loss": 0.3161,
      "step": 4042
    },
    {
      "epoch": 0.1855353127437933,
      "grad_norm": 0.491471529006958,
      "learning_rate": 9.955098415727062e-06,
      "loss": 0.4716,
      "step": 4043
    },
    {
      "epoch": 0.18558120324904778,
      "grad_norm": 0.48043712973594666,
      "learning_rate": 9.955065624205782e-06,
      "loss": 0.429,
      "step": 4044
    },
    {
      "epoch": 0.18562709375430222,
      "grad_norm": 0.46590757369995117,
      "learning_rate": 9.955032820769132e-06,
      "loss": 0.3984,
      "step": 4045
    },
    {
      "epoch": 0.1856729842595567,
      "grad_norm": 0.45795169472694397,
      "learning_rate": 9.955000005417191e-06,
      "loss": 0.3536,
      "step": 4046
    },
    {
      "epoch": 0.18571887476481116,
      "grad_norm": 0.4461454153060913,
      "learning_rate": 9.954967178150041e-06,
      "loss": 0.3526,
      "step": 4047
    },
    {
      "epoch": 0.18576476527006563,
      "grad_norm": 0.4559987187385559,
      "learning_rate": 9.954934338967757e-06,
      "loss": 0.3503,
      "step": 4048
    },
    {
      "epoch": 0.18581065577532008,
      "grad_norm": 0.43757522106170654,
      "learning_rate": 9.954901487870421e-06,
      "loss": 0.3774,
      "step": 4049
    },
    {
      "epoch": 0.18585654628057455,
      "grad_norm": 0.44442030787467957,
      "learning_rate": 9.954868624858109e-06,
      "loss": 0.4096,
      "step": 4050
    },
    {
      "epoch": 0.18590243678582902,
      "grad_norm": 0.5150233507156372,
      "learning_rate": 9.954835749930905e-06,
      "loss": 0.4486,
      "step": 4051
    },
    {
      "epoch": 0.1859483272910835,
      "grad_norm": 0.4833717942237854,
      "learning_rate": 9.95480286308888e-06,
      "loss": 0.384,
      "step": 4052
    },
    {
      "epoch": 0.18599421779633793,
      "grad_norm": 0.49683383107185364,
      "learning_rate": 9.954769964332123e-06,
      "loss": 0.3969,
      "step": 4053
    },
    {
      "epoch": 0.1860401083015924,
      "grad_norm": 0.44091925024986267,
      "learning_rate": 9.954737053660705e-06,
      "loss": 0.3131,
      "step": 4054
    },
    {
      "epoch": 0.18608599880684687,
      "grad_norm": 0.6123804450035095,
      "learning_rate": 9.954704131074709e-06,
      "loss": 0.4667,
      "step": 4055
    },
    {
      "epoch": 0.18613188931210134,
      "grad_norm": 0.45483189821243286,
      "learning_rate": 9.954671196574213e-06,
      "loss": 0.3564,
      "step": 4056
    },
    {
      "epoch": 0.18617777981735578,
      "grad_norm": 0.5285971164703369,
      "learning_rate": 9.954638250159297e-06,
      "loss": 0.5051,
      "step": 4057
    },
    {
      "epoch": 0.18622367032261025,
      "grad_norm": 0.4830685257911682,
      "learning_rate": 9.95460529183004e-06,
      "loss": 0.3996,
      "step": 4058
    },
    {
      "epoch": 0.18626956082786472,
      "grad_norm": 0.4549676179885864,
      "learning_rate": 9.95457232158652e-06,
      "loss": 0.3661,
      "step": 4059
    },
    {
      "epoch": 0.18631545133311916,
      "grad_norm": 0.4777267575263977,
      "learning_rate": 9.954539339428817e-06,
      "loss": 0.3395,
      "step": 4060
    },
    {
      "epoch": 0.18636134183837363,
      "grad_norm": 0.44971832633018494,
      "learning_rate": 9.95450634535701e-06,
      "loss": 0.3558,
      "step": 4061
    },
    {
      "epoch": 0.1864072323436281,
      "grad_norm": 0.45441970229148865,
      "learning_rate": 9.954473339371181e-06,
      "loss": 0.335,
      "step": 4062
    },
    {
      "epoch": 0.18645312284888257,
      "grad_norm": 0.5144253969192505,
      "learning_rate": 9.954440321471408e-06,
      "loss": 0.4626,
      "step": 4063
    },
    {
      "epoch": 0.18649901335413702,
      "grad_norm": 0.48288124799728394,
      "learning_rate": 9.954407291657767e-06,
      "loss": 0.4476,
      "step": 4064
    },
    {
      "epoch": 0.1865449038593915,
      "grad_norm": 0.4835912585258484,
      "learning_rate": 9.95437424993034e-06,
      "loss": 0.3409,
      "step": 4065
    },
    {
      "epoch": 0.18659079436464596,
      "grad_norm": 0.4531983733177185,
      "learning_rate": 9.954341196289209e-06,
      "loss": 0.3627,
      "step": 4066
    },
    {
      "epoch": 0.18663668486990043,
      "grad_norm": 0.4809619188308716,
      "learning_rate": 9.95430813073445e-06,
      "loss": 0.3838,
      "step": 4067
    },
    {
      "epoch": 0.18668257537515487,
      "grad_norm": 0.47172215580940247,
      "learning_rate": 9.954275053266143e-06,
      "loss": 0.4215,
      "step": 4068
    },
    {
      "epoch": 0.18672846588040934,
      "grad_norm": 0.46933355927467346,
      "learning_rate": 9.954241963884369e-06,
      "loss": 0.3847,
      "step": 4069
    },
    {
      "epoch": 0.1867743563856638,
      "grad_norm": 0.44868549704551697,
      "learning_rate": 9.954208862589206e-06,
      "loss": 0.3685,
      "step": 4070
    },
    {
      "epoch": 0.18682024689091828,
      "grad_norm": 0.484976202249527,
      "learning_rate": 9.954175749380735e-06,
      "loss": 0.3832,
      "step": 4071
    },
    {
      "epoch": 0.18686613739617272,
      "grad_norm": 0.45270830392837524,
      "learning_rate": 9.954142624259035e-06,
      "loss": 0.324,
      "step": 4072
    },
    {
      "epoch": 0.1869120279014272,
      "grad_norm": 0.45380428433418274,
      "learning_rate": 9.954109487224185e-06,
      "loss": 0.3614,
      "step": 4073
    },
    {
      "epoch": 0.18695791840668166,
      "grad_norm": 0.5037969350814819,
      "learning_rate": 9.954076338276265e-06,
      "loss": 0.5168,
      "step": 4074
    },
    {
      "epoch": 0.18700380891193613,
      "grad_norm": 0.4648370146751404,
      "learning_rate": 9.954043177415355e-06,
      "loss": 0.3109,
      "step": 4075
    },
    {
      "epoch": 0.18704969941719057,
      "grad_norm": 0.4700336158275604,
      "learning_rate": 9.954010004641534e-06,
      "loss": 0.3497,
      "step": 4076
    },
    {
      "epoch": 0.18709558992244504,
      "grad_norm": 0.4684875011444092,
      "learning_rate": 9.953976819954885e-06,
      "loss": 0.3808,
      "step": 4077
    },
    {
      "epoch": 0.1871414804276995,
      "grad_norm": 0.4918420612812042,
      "learning_rate": 9.953943623355482e-06,
      "loss": 0.3775,
      "step": 4078
    },
    {
      "epoch": 0.18718737093295398,
      "grad_norm": 0.5494481921195984,
      "learning_rate": 9.95391041484341e-06,
      "loss": 0.4873,
      "step": 4079
    },
    {
      "epoch": 0.18723326143820843,
      "grad_norm": 0.479916512966156,
      "learning_rate": 9.953877194418747e-06,
      "loss": 0.3805,
      "step": 4080
    },
    {
      "epoch": 0.1872791519434629,
      "grad_norm": 0.4575951099395752,
      "learning_rate": 9.953843962081574e-06,
      "loss": 0.4386,
      "step": 4081
    },
    {
      "epoch": 0.18732504244871737,
      "grad_norm": 0.5033926963806152,
      "learning_rate": 9.953810717831966e-06,
      "loss": 0.4555,
      "step": 4082
    },
    {
      "epoch": 0.18737093295397184,
      "grad_norm": 0.46825897693634033,
      "learning_rate": 9.95377746167001e-06,
      "loss": 0.3692,
      "step": 4083
    },
    {
      "epoch": 0.18741682345922628,
      "grad_norm": 0.46704164147377014,
      "learning_rate": 9.953744193595782e-06,
      "loss": 0.3737,
      "step": 4084
    },
    {
      "epoch": 0.18746271396448075,
      "grad_norm": 0.4961448609828949,
      "learning_rate": 9.953710913609362e-06,
      "loss": 0.5158,
      "step": 4085
    },
    {
      "epoch": 0.18750860446973522,
      "grad_norm": 0.46118593215942383,
      "learning_rate": 9.953677621710831e-06,
      "loss": 0.3884,
      "step": 4086
    },
    {
      "epoch": 0.1875544949749897,
      "grad_norm": 0.47987639904022217,
      "learning_rate": 9.953644317900269e-06,
      "loss": 0.4246,
      "step": 4087
    },
    {
      "epoch": 0.18760038548024413,
      "grad_norm": 0.4968966245651245,
      "learning_rate": 9.953611002177755e-06,
      "loss": 0.3639,
      "step": 4088
    },
    {
      "epoch": 0.1876462759854986,
      "grad_norm": 0.4718018174171448,
      "learning_rate": 9.95357767454337e-06,
      "loss": 0.3913,
      "step": 4089
    },
    {
      "epoch": 0.18769216649075307,
      "grad_norm": 0.4618558883666992,
      "learning_rate": 9.953544334997196e-06,
      "loss": 0.3679,
      "step": 4090
    },
    {
      "epoch": 0.1877380569960075,
      "grad_norm": 0.5178093314170837,
      "learning_rate": 9.95351098353931e-06,
      "loss": 0.4624,
      "step": 4091
    },
    {
      "epoch": 0.18778394750126198,
      "grad_norm": 0.44858354330062866,
      "learning_rate": 9.953477620169793e-06,
      "loss": 0.3908,
      "step": 4092
    },
    {
      "epoch": 0.18782983800651645,
      "grad_norm": 0.4788037836551666,
      "learning_rate": 9.953444244888725e-06,
      "loss": 0.4495,
      "step": 4093
    },
    {
      "epoch": 0.18787572851177092,
      "grad_norm": 0.4847019612789154,
      "learning_rate": 9.953410857696188e-06,
      "loss": 0.4005,
      "step": 4094
    },
    {
      "epoch": 0.18792161901702537,
      "grad_norm": 0.530505359172821,
      "learning_rate": 9.953377458592261e-06,
      "loss": 0.5364,
      "step": 4095
    },
    {
      "epoch": 0.18796750952227984,
      "grad_norm": 0.4931606948375702,
      "learning_rate": 9.953344047577025e-06,
      "loss": 0.4736,
      "step": 4096
    },
    {
      "epoch": 0.1880134000275343,
      "grad_norm": 0.4111191928386688,
      "learning_rate": 9.95331062465056e-06,
      "loss": 0.2957,
      "step": 4097
    },
    {
      "epoch": 0.18805929053278878,
      "grad_norm": 0.43773943185806274,
      "learning_rate": 9.953277189812945e-06,
      "loss": 0.3478,
      "step": 4098
    },
    {
      "epoch": 0.18810518103804322,
      "grad_norm": 0.4870125353336334,
      "learning_rate": 9.953243743064263e-06,
      "loss": 0.3839,
      "step": 4099
    },
    {
      "epoch": 0.1881510715432977,
      "grad_norm": 0.4274459779262543,
      "learning_rate": 9.953210284404592e-06,
      "loss": 0.3513,
      "step": 4100
    },
    {
      "epoch": 0.18819696204855216,
      "grad_norm": 0.49045252799987793,
      "learning_rate": 9.953176813834015e-06,
      "loss": 0.4276,
      "step": 4101
    },
    {
      "epoch": 0.18824285255380663,
      "grad_norm": 0.5296708941459656,
      "learning_rate": 9.95314333135261e-06,
      "loss": 0.4135,
      "step": 4102
    },
    {
      "epoch": 0.18828874305906107,
      "grad_norm": 0.5185618996620178,
      "learning_rate": 9.95310983696046e-06,
      "loss": 0.4544,
      "step": 4103
    },
    {
      "epoch": 0.18833463356431554,
      "grad_norm": 0.4736163318157196,
      "learning_rate": 9.953076330657641e-06,
      "loss": 0.3828,
      "step": 4104
    },
    {
      "epoch": 0.18838052406957,
      "grad_norm": 0.4864441454410553,
      "learning_rate": 9.95304281244424e-06,
      "loss": 0.4292,
      "step": 4105
    },
    {
      "epoch": 0.18842641457482448,
      "grad_norm": 0.4908764958381653,
      "learning_rate": 9.953009282320332e-06,
      "loss": 0.4628,
      "step": 4106
    },
    {
      "epoch": 0.18847230508007892,
      "grad_norm": 0.44502270221710205,
      "learning_rate": 9.952975740286e-06,
      "loss": 0.4036,
      "step": 4107
    },
    {
      "epoch": 0.1885181955853334,
      "grad_norm": 0.5089215040206909,
      "learning_rate": 9.952942186341325e-06,
      "loss": 0.5021,
      "step": 4108
    },
    {
      "epoch": 0.18856408609058786,
      "grad_norm": 0.4454263150691986,
      "learning_rate": 9.952908620486387e-06,
      "loss": 0.3538,
      "step": 4109
    },
    {
      "epoch": 0.18860997659584233,
      "grad_norm": 0.46301963925361633,
      "learning_rate": 9.952875042721268e-06,
      "loss": 0.3976,
      "step": 4110
    },
    {
      "epoch": 0.18865586710109677,
      "grad_norm": 0.4358971118927002,
      "learning_rate": 9.952841453046047e-06,
      "loss": 0.3036,
      "step": 4111
    },
    {
      "epoch": 0.18870175760635124,
      "grad_norm": 0.47368836402893066,
      "learning_rate": 9.952807851460807e-06,
      "loss": 0.4217,
      "step": 4112
    },
    {
      "epoch": 0.18874764811160571,
      "grad_norm": 0.550803542137146,
      "learning_rate": 9.952774237965624e-06,
      "loss": 0.5123,
      "step": 4113
    },
    {
      "epoch": 0.18879353861686018,
      "grad_norm": 0.4942846894264221,
      "learning_rate": 9.952740612560585e-06,
      "loss": 0.4708,
      "step": 4114
    },
    {
      "epoch": 0.18883942912211463,
      "grad_norm": 0.47420063614845276,
      "learning_rate": 9.952706975245767e-06,
      "loss": 0.3872,
      "step": 4115
    },
    {
      "epoch": 0.1888853196273691,
      "grad_norm": 0.5260906219482422,
      "learning_rate": 9.952673326021252e-06,
      "loss": 0.4132,
      "step": 4116
    },
    {
      "epoch": 0.18893121013262357,
      "grad_norm": 0.6215383410453796,
      "learning_rate": 9.95263966488712e-06,
      "loss": 0.4234,
      "step": 4117
    },
    {
      "epoch": 0.188977100637878,
      "grad_norm": 0.5035076141357422,
      "learning_rate": 9.952605991843452e-06,
      "loss": 0.4147,
      "step": 4118
    },
    {
      "epoch": 0.18902299114313248,
      "grad_norm": 0.4822065532207489,
      "learning_rate": 9.952572306890332e-06,
      "loss": 0.3967,
      "step": 4119
    },
    {
      "epoch": 0.18906888164838695,
      "grad_norm": 0.5079449415206909,
      "learning_rate": 9.952538610027837e-06,
      "loss": 0.4548,
      "step": 4120
    },
    {
      "epoch": 0.18911477215364142,
      "grad_norm": 0.42507266998291016,
      "learning_rate": 9.95250490125605e-06,
      "loss": 0.3327,
      "step": 4121
    },
    {
      "epoch": 0.18916066265889586,
      "grad_norm": 0.49618425965309143,
      "learning_rate": 9.952471180575052e-06,
      "loss": 0.4404,
      "step": 4122
    },
    {
      "epoch": 0.18920655316415033,
      "grad_norm": 0.5208756923675537,
      "learning_rate": 9.952437447984924e-06,
      "loss": 0.4517,
      "step": 4123
    },
    {
      "epoch": 0.1892524436694048,
      "grad_norm": 0.5111962556838989,
      "learning_rate": 9.952403703485746e-06,
      "loss": 0.4436,
      "step": 4124
    },
    {
      "epoch": 0.18929833417465927,
      "grad_norm": 0.43555372953414917,
      "learning_rate": 9.952369947077603e-06,
      "loss": 0.3401,
      "step": 4125
    },
    {
      "epoch": 0.18934422467991371,
      "grad_norm": 0.493084579706192,
      "learning_rate": 9.95233617876057e-06,
      "loss": 0.4378,
      "step": 4126
    },
    {
      "epoch": 0.18939011518516818,
      "grad_norm": 0.490662544965744,
      "learning_rate": 9.952302398534733e-06,
      "loss": 0.4177,
      "step": 4127
    },
    {
      "epoch": 0.18943600569042265,
      "grad_norm": 0.4406987130641937,
      "learning_rate": 9.95226860640017e-06,
      "loss": 0.364,
      "step": 4128
    },
    {
      "epoch": 0.18948189619567712,
      "grad_norm": 0.47129300236701965,
      "learning_rate": 9.952234802356966e-06,
      "loss": 0.3863,
      "step": 4129
    },
    {
      "epoch": 0.18952778670093157,
      "grad_norm": 0.5045755505561829,
      "learning_rate": 9.952200986405201e-06,
      "loss": 0.3848,
      "step": 4130
    },
    {
      "epoch": 0.18957367720618604,
      "grad_norm": 0.48028242588043213,
      "learning_rate": 9.952167158544955e-06,
      "loss": 0.4149,
      "step": 4131
    },
    {
      "epoch": 0.1896195677114405,
      "grad_norm": 0.5005995631217957,
      "learning_rate": 9.952133318776307e-06,
      "loss": 0.4154,
      "step": 4132
    },
    {
      "epoch": 0.18966545821669498,
      "grad_norm": 0.45134297013282776,
      "learning_rate": 9.952099467099344e-06,
      "loss": 0.3423,
      "step": 4133
    },
    {
      "epoch": 0.18971134872194942,
      "grad_norm": 0.483934223651886,
      "learning_rate": 9.952065603514145e-06,
      "loss": 0.5167,
      "step": 4134
    },
    {
      "epoch": 0.1897572392272039,
      "grad_norm": 0.45987600088119507,
      "learning_rate": 9.95203172802079e-06,
      "loss": 0.3426,
      "step": 4135
    },
    {
      "epoch": 0.18980312973245836,
      "grad_norm": 0.4913298785686493,
      "learning_rate": 9.951997840619362e-06,
      "loss": 0.4623,
      "step": 4136
    },
    {
      "epoch": 0.18984902023771283,
      "grad_norm": 0.46920305490493774,
      "learning_rate": 9.951963941309941e-06,
      "loss": 0.3779,
      "step": 4137
    },
    {
      "epoch": 0.18989491074296727,
      "grad_norm": 0.4733751714229584,
      "learning_rate": 9.951930030092611e-06,
      "loss": 0.3656,
      "step": 4138
    },
    {
      "epoch": 0.18994080124822174,
      "grad_norm": 0.49268674850463867,
      "learning_rate": 9.951896106967452e-06,
      "loss": 0.3912,
      "step": 4139
    },
    {
      "epoch": 0.1899866917534762,
      "grad_norm": 0.4943540096282959,
      "learning_rate": 9.951862171934545e-06,
      "loss": 0.4087,
      "step": 4140
    },
    {
      "epoch": 0.19003258225873068,
      "grad_norm": 0.4985579252243042,
      "learning_rate": 9.951828224993974e-06,
      "loss": 0.4165,
      "step": 4141
    },
    {
      "epoch": 0.19007847276398512,
      "grad_norm": 0.5130932331085205,
      "learning_rate": 9.951794266145818e-06,
      "loss": 0.4843,
      "step": 4142
    },
    {
      "epoch": 0.1901243632692396,
      "grad_norm": 0.5205416679382324,
      "learning_rate": 9.951760295390158e-06,
      "loss": 0.4239,
      "step": 4143
    },
    {
      "epoch": 0.19017025377449406,
      "grad_norm": 0.48263564705848694,
      "learning_rate": 9.95172631272708e-06,
      "loss": 0.3825,
      "step": 4144
    },
    {
      "epoch": 0.19021614427974853,
      "grad_norm": 0.47723209857940674,
      "learning_rate": 9.951692318156662e-06,
      "loss": 0.4286,
      "step": 4145
    },
    {
      "epoch": 0.19026203478500298,
      "grad_norm": 0.4813656210899353,
      "learning_rate": 9.951658311678987e-06,
      "loss": 0.3568,
      "step": 4146
    },
    {
      "epoch": 0.19030792529025745,
      "grad_norm": 0.5103087425231934,
      "learning_rate": 9.951624293294135e-06,
      "loss": 0.4905,
      "step": 4147
    },
    {
      "epoch": 0.19035381579551192,
      "grad_norm": 0.5263637900352478,
      "learning_rate": 9.951590263002191e-06,
      "loss": 0.4505,
      "step": 4148
    },
    {
      "epoch": 0.19039970630076636,
      "grad_norm": 0.46938657760620117,
      "learning_rate": 9.951556220803233e-06,
      "loss": 0.437,
      "step": 4149
    },
    {
      "epoch": 0.19044559680602083,
      "grad_norm": 0.5023797750473022,
      "learning_rate": 9.951522166697347e-06,
      "loss": 0.4919,
      "step": 4150
    },
    {
      "epoch": 0.1904914873112753,
      "grad_norm": 0.5201046466827393,
      "learning_rate": 9.951488100684613e-06,
      "loss": 0.4952,
      "step": 4151
    },
    {
      "epoch": 0.19053737781652977,
      "grad_norm": 0.4829038381576538,
      "learning_rate": 9.951454022765112e-06,
      "loss": 0.3966,
      "step": 4152
    },
    {
      "epoch": 0.1905832683217842,
      "grad_norm": 0.4435252547264099,
      "learning_rate": 9.951419932938928e-06,
      "loss": 0.3429,
      "step": 4153
    },
    {
      "epoch": 0.19062915882703868,
      "grad_norm": 0.4686864912509918,
      "learning_rate": 9.951385831206142e-06,
      "loss": 0.4335,
      "step": 4154
    },
    {
      "epoch": 0.19067504933229315,
      "grad_norm": 0.4584447145462036,
      "learning_rate": 9.951351717566835e-06,
      "loss": 0.3807,
      "step": 4155
    },
    {
      "epoch": 0.19072093983754762,
      "grad_norm": 0.46828722953796387,
      "learning_rate": 9.951317592021089e-06,
      "loss": 0.3829,
      "step": 4156
    },
    {
      "epoch": 0.19076683034280206,
      "grad_norm": 0.5308449268341064,
      "learning_rate": 9.951283454568987e-06,
      "loss": 0.4734,
      "step": 4157
    },
    {
      "epoch": 0.19081272084805653,
      "grad_norm": 0.4835406243801117,
      "learning_rate": 9.951249305210612e-06,
      "loss": 0.3929,
      "step": 4158
    },
    {
      "epoch": 0.190858611353311,
      "grad_norm": 0.4220728874206543,
      "learning_rate": 9.951215143946044e-06,
      "loss": 0.3082,
      "step": 4159
    },
    {
      "epoch": 0.19090450185856547,
      "grad_norm": 0.47217854857444763,
      "learning_rate": 9.951180970775366e-06,
      "loss": 0.356,
      "step": 4160
    },
    {
      "epoch": 0.19095039236381992,
      "grad_norm": 0.5041155815124512,
      "learning_rate": 9.951146785698663e-06,
      "loss": 0.4472,
      "step": 4161
    },
    {
      "epoch": 0.19099628286907439,
      "grad_norm": 0.49931737780570984,
      "learning_rate": 9.951112588716013e-06,
      "loss": 0.4941,
      "step": 4162
    },
    {
      "epoch": 0.19104217337432886,
      "grad_norm": 0.4517405331134796,
      "learning_rate": 9.951078379827499e-06,
      "loss": 0.3641,
      "step": 4163
    },
    {
      "epoch": 0.19108806387958333,
      "grad_norm": 0.5002760887145996,
      "learning_rate": 9.951044159033204e-06,
      "loss": 0.3661,
      "step": 4164
    },
    {
      "epoch": 0.19113395438483777,
      "grad_norm": 0.493163526058197,
      "learning_rate": 9.951009926333212e-06,
      "loss": 0.4541,
      "step": 4165
    },
    {
      "epoch": 0.19117984489009224,
      "grad_norm": 0.4570077955722809,
      "learning_rate": 9.950975681727602e-06,
      "loss": 0.3872,
      "step": 4166
    },
    {
      "epoch": 0.1912257353953467,
      "grad_norm": 0.4414304494857788,
      "learning_rate": 9.950941425216459e-06,
      "loss": 0.3718,
      "step": 4167
    },
    {
      "epoch": 0.19127162590060118,
      "grad_norm": 0.4730222523212433,
      "learning_rate": 9.950907156799865e-06,
      "loss": 0.3382,
      "step": 4168
    },
    {
      "epoch": 0.19131751640585562,
      "grad_norm": 0.4675235450267792,
      "learning_rate": 9.950872876477901e-06,
      "loss": 0.4547,
      "step": 4169
    },
    {
      "epoch": 0.1913634069111101,
      "grad_norm": 0.4699147939682007,
      "learning_rate": 9.95083858425065e-06,
      "loss": 0.3982,
      "step": 4170
    },
    {
      "epoch": 0.19140929741636456,
      "grad_norm": 0.4556610584259033,
      "learning_rate": 9.950804280118196e-06,
      "loss": 0.3722,
      "step": 4171
    },
    {
      "epoch": 0.19145518792161903,
      "grad_norm": 0.46198368072509766,
      "learning_rate": 9.950769964080618e-06,
      "loss": 0.4223,
      "step": 4172
    },
    {
      "epoch": 0.19150107842687347,
      "grad_norm": 0.5001542568206787,
      "learning_rate": 9.950735636138003e-06,
      "loss": 0.4736,
      "step": 4173
    },
    {
      "epoch": 0.19154696893212794,
      "grad_norm": 0.5015148520469666,
      "learning_rate": 9.95070129629043e-06,
      "loss": 0.3649,
      "step": 4174
    },
    {
      "epoch": 0.1915928594373824,
      "grad_norm": 0.5021859407424927,
      "learning_rate": 9.950666944537983e-06,
      "loss": 0.5189,
      "step": 4175
    },
    {
      "epoch": 0.19163874994263685,
      "grad_norm": 0.4707670509815216,
      "learning_rate": 9.950632580880745e-06,
      "loss": 0.4479,
      "step": 4176
    },
    {
      "epoch": 0.19168464044789132,
      "grad_norm": 0.5313079953193665,
      "learning_rate": 9.950598205318798e-06,
      "loss": 0.4702,
      "step": 4177
    },
    {
      "epoch": 0.1917305309531458,
      "grad_norm": 0.5079951286315918,
      "learning_rate": 9.950563817852225e-06,
      "loss": 0.5436,
      "step": 4178
    },
    {
      "epoch": 0.19177642145840026,
      "grad_norm": 0.49826809763908386,
      "learning_rate": 9.950529418481107e-06,
      "loss": 0.505,
      "step": 4179
    },
    {
      "epoch": 0.1918223119636547,
      "grad_norm": 0.4891725480556488,
      "learning_rate": 9.95049500720553e-06,
      "loss": 0.4788,
      "step": 4180
    },
    {
      "epoch": 0.19186820246890918,
      "grad_norm": 0.5809705257415771,
      "learning_rate": 9.950460584025574e-06,
      "loss": 0.4472,
      "step": 4181
    },
    {
      "epoch": 0.19191409297416365,
      "grad_norm": 0.48403993248939514,
      "learning_rate": 9.950426148941323e-06,
      "loss": 0.3733,
      "step": 4182
    },
    {
      "epoch": 0.19195998347941812,
      "grad_norm": 0.6793543100357056,
      "learning_rate": 9.95039170195286e-06,
      "loss": 0.4737,
      "step": 4183
    },
    {
      "epoch": 0.19200587398467256,
      "grad_norm": 0.4881032705307007,
      "learning_rate": 9.950357243060266e-06,
      "loss": 0.4271,
      "step": 4184
    },
    {
      "epoch": 0.19205176448992703,
      "grad_norm": 0.4615977704524994,
      "learning_rate": 9.950322772263627e-06,
      "loss": 0.417,
      "step": 4185
    },
    {
      "epoch": 0.1920976549951815,
      "grad_norm": 0.47727513313293457,
      "learning_rate": 9.950288289563023e-06,
      "loss": 0.4553,
      "step": 4186
    },
    {
      "epoch": 0.19214354550043597,
      "grad_norm": 0.5171818137168884,
      "learning_rate": 9.950253794958537e-06,
      "loss": 0.4944,
      "step": 4187
    },
    {
      "epoch": 0.1921894360056904,
      "grad_norm": 0.4399084746837616,
      "learning_rate": 9.950219288450255e-06,
      "loss": 0.2992,
      "step": 4188
    },
    {
      "epoch": 0.19223532651094488,
      "grad_norm": 0.47041231393814087,
      "learning_rate": 9.950184770038257e-06,
      "loss": 0.4007,
      "step": 4189
    },
    {
      "epoch": 0.19228121701619935,
      "grad_norm": 0.44988352060317993,
      "learning_rate": 9.950150239722626e-06,
      "loss": 0.3841,
      "step": 4190
    },
    {
      "epoch": 0.19232710752145382,
      "grad_norm": 0.5435940027236938,
      "learning_rate": 9.950115697503449e-06,
      "loss": 0.5663,
      "step": 4191
    },
    {
      "epoch": 0.19237299802670826,
      "grad_norm": 0.555151104927063,
      "learning_rate": 9.950081143380804e-06,
      "loss": 0.4795,
      "step": 4192
    },
    {
      "epoch": 0.19241888853196273,
      "grad_norm": 0.4963894486427307,
      "learning_rate": 9.950046577354774e-06,
      "loss": 0.4438,
      "step": 4193
    },
    {
      "epoch": 0.1924647790372172,
      "grad_norm": 0.46585705876350403,
      "learning_rate": 9.950011999425448e-06,
      "loss": 0.3749,
      "step": 4194
    },
    {
      "epoch": 0.19251066954247167,
      "grad_norm": 0.43702268600463867,
      "learning_rate": 9.949977409592904e-06,
      "loss": 0.3037,
      "step": 4195
    },
    {
      "epoch": 0.19255656004772612,
      "grad_norm": 0.5150640606880188,
      "learning_rate": 9.949942807857225e-06,
      "loss": 0.4286,
      "step": 4196
    },
    {
      "epoch": 0.1926024505529806,
      "grad_norm": 0.512639582157135,
      "learning_rate": 9.949908194218498e-06,
      "loss": 0.4699,
      "step": 4197
    },
    {
      "epoch": 0.19264834105823506,
      "grad_norm": 0.49953705072402954,
      "learning_rate": 9.949873568676801e-06,
      "loss": 0.4439,
      "step": 4198
    },
    {
      "epoch": 0.19269423156348953,
      "grad_norm": 0.49101316928863525,
      "learning_rate": 9.949838931232225e-06,
      "loss": 0.4182,
      "step": 4199
    },
    {
      "epoch": 0.19274012206874397,
      "grad_norm": 0.5629458427429199,
      "learning_rate": 9.949804281884845e-06,
      "loss": 0.5942,
      "step": 4200
    },
    {
      "epoch": 0.19278601257399844,
      "grad_norm": 0.46506381034851074,
      "learning_rate": 9.94976962063475e-06,
      "loss": 0.3641,
      "step": 4201
    },
    {
      "epoch": 0.1928319030792529,
      "grad_norm": 0.46575990319252014,
      "learning_rate": 9.949734947482019e-06,
      "loss": 0.3955,
      "step": 4202
    },
    {
      "epoch": 0.19287779358450738,
      "grad_norm": 0.4566187858581543,
      "learning_rate": 9.94970026242674e-06,
      "loss": 0.3614,
      "step": 4203
    },
    {
      "epoch": 0.19292368408976182,
      "grad_norm": 0.5021764039993286,
      "learning_rate": 9.949665565468994e-06,
      "loss": 0.4061,
      "step": 4204
    },
    {
      "epoch": 0.1929695745950163,
      "grad_norm": 0.5028994083404541,
      "learning_rate": 9.949630856608862e-06,
      "loss": 0.4135,
      "step": 4205
    },
    {
      "epoch": 0.19301546510027076,
      "grad_norm": 0.48667213320732117,
      "learning_rate": 9.94959613584643e-06,
      "loss": 0.4098,
      "step": 4206
    },
    {
      "epoch": 0.1930613556055252,
      "grad_norm": 0.45400986075401306,
      "learning_rate": 9.949561403181785e-06,
      "loss": 0.3744,
      "step": 4207
    },
    {
      "epoch": 0.19310724611077967,
      "grad_norm": 0.45983439683914185,
      "learning_rate": 9.949526658615004e-06,
      "loss": 0.4092,
      "step": 4208
    },
    {
      "epoch": 0.19315313661603414,
      "grad_norm": 0.49159324169158936,
      "learning_rate": 9.949491902146175e-06,
      "loss": 0.4313,
      "step": 4209
    },
    {
      "epoch": 0.1931990271212886,
      "grad_norm": 0.46645256876945496,
      "learning_rate": 9.949457133775379e-06,
      "loss": 0.4934,
      "step": 4210
    },
    {
      "epoch": 0.19324491762654306,
      "grad_norm": 0.4616362154483795,
      "learning_rate": 9.9494223535027e-06,
      "loss": 0.4111,
      "step": 4211
    },
    {
      "epoch": 0.19329080813179753,
      "grad_norm": 0.45995163917541504,
      "learning_rate": 9.949387561328224e-06,
      "loss": 0.4165,
      "step": 4212
    },
    {
      "epoch": 0.193336698637052,
      "grad_norm": 0.607218861579895,
      "learning_rate": 9.949352757252033e-06,
      "loss": 0.3802,
      "step": 4213
    },
    {
      "epoch": 0.19338258914230647,
      "grad_norm": 0.5580732226371765,
      "learning_rate": 9.949317941274209e-06,
      "loss": 0.5255,
      "step": 4214
    },
    {
      "epoch": 0.1934284796475609,
      "grad_norm": 0.4484957754611969,
      "learning_rate": 9.949283113394838e-06,
      "loss": 0.3946,
      "step": 4215
    },
    {
      "epoch": 0.19347437015281538,
      "grad_norm": 0.44185206294059753,
      "learning_rate": 9.949248273614005e-06,
      "loss": 0.3651,
      "step": 4216
    },
    {
      "epoch": 0.19352026065806985,
      "grad_norm": 0.4761306047439575,
      "learning_rate": 9.949213421931789e-06,
      "loss": 0.4186,
      "step": 4217
    },
    {
      "epoch": 0.19356615116332432,
      "grad_norm": 0.47021451592445374,
      "learning_rate": 9.949178558348279e-06,
      "loss": 0.3737,
      "step": 4218
    },
    {
      "epoch": 0.19361204166857876,
      "grad_norm": 0.48406535387039185,
      "learning_rate": 9.949143682863556e-06,
      "loss": 0.4257,
      "step": 4219
    },
    {
      "epoch": 0.19365793217383323,
      "grad_norm": 0.4199129343032837,
      "learning_rate": 9.949108795477704e-06,
      "loss": 0.287,
      "step": 4220
    },
    {
      "epoch": 0.1937038226790877,
      "grad_norm": 0.46197354793548584,
      "learning_rate": 9.949073896190808e-06,
      "loss": 0.3748,
      "step": 4221
    },
    {
      "epoch": 0.19374971318434217,
      "grad_norm": 0.47738486528396606,
      "learning_rate": 9.94903898500295e-06,
      "loss": 0.4203,
      "step": 4222
    },
    {
      "epoch": 0.1937956036895966,
      "grad_norm": 0.49528124928474426,
      "learning_rate": 9.949004061914216e-06,
      "loss": 0.4721,
      "step": 4223
    },
    {
      "epoch": 0.19384149419485108,
      "grad_norm": 0.4897059500217438,
      "learning_rate": 9.94896912692469e-06,
      "loss": 0.4441,
      "step": 4224
    },
    {
      "epoch": 0.19388738470010555,
      "grad_norm": 0.4783026874065399,
      "learning_rate": 9.948934180034453e-06,
      "loss": 0.4509,
      "step": 4225
    },
    {
      "epoch": 0.19393327520536002,
      "grad_norm": 0.47514715790748596,
      "learning_rate": 9.948899221243592e-06,
      "loss": 0.4035,
      "step": 4226
    },
    {
      "epoch": 0.19397916571061447,
      "grad_norm": 0.4776126444339752,
      "learning_rate": 9.948864250552192e-06,
      "loss": 0.3831,
      "step": 4227
    },
    {
      "epoch": 0.19402505621586894,
      "grad_norm": 0.49811041355133057,
      "learning_rate": 9.948829267960335e-06,
      "loss": 0.3906,
      "step": 4228
    },
    {
      "epoch": 0.1940709467211234,
      "grad_norm": 0.47314468026161194,
      "learning_rate": 9.948794273468105e-06,
      "loss": 0.416,
      "step": 4229
    },
    {
      "epoch": 0.19411683722637788,
      "grad_norm": 0.4896462857723236,
      "learning_rate": 9.948759267075585e-06,
      "loss": 0.4495,
      "step": 4230
    },
    {
      "epoch": 0.19416272773163232,
      "grad_norm": 0.45454174280166626,
      "learning_rate": 9.948724248782863e-06,
      "loss": 0.4122,
      "step": 4231
    },
    {
      "epoch": 0.1942086182368868,
      "grad_norm": 0.44305551052093506,
      "learning_rate": 9.94868921859002e-06,
      "loss": 0.3236,
      "step": 4232
    },
    {
      "epoch": 0.19425450874214126,
      "grad_norm": 0.4922850430011749,
      "learning_rate": 9.948654176497142e-06,
      "loss": 0.4557,
      "step": 4233
    },
    {
      "epoch": 0.19430039924739573,
      "grad_norm": 0.47722384333610535,
      "learning_rate": 9.948619122504312e-06,
      "loss": 0.4404,
      "step": 4234
    },
    {
      "epoch": 0.19434628975265017,
      "grad_norm": 0.4752442538738251,
      "learning_rate": 9.948584056611616e-06,
      "loss": 0.4523,
      "step": 4235
    },
    {
      "epoch": 0.19439218025790464,
      "grad_norm": 0.42401692271232605,
      "learning_rate": 9.948548978819135e-06,
      "loss": 0.3058,
      "step": 4236
    },
    {
      "epoch": 0.1944380707631591,
      "grad_norm": 0.4936096668243408,
      "learning_rate": 9.948513889126956e-06,
      "loss": 0.4493,
      "step": 4237
    },
    {
      "epoch": 0.19448396126841355,
      "grad_norm": 0.5123659372329712,
      "learning_rate": 9.948478787535166e-06,
      "loss": 0.473,
      "step": 4238
    },
    {
      "epoch": 0.19452985177366802,
      "grad_norm": 0.4875311851501465,
      "learning_rate": 9.948443674043841e-06,
      "loss": 0.4571,
      "step": 4239
    },
    {
      "epoch": 0.1945757422789225,
      "grad_norm": 0.430096834897995,
      "learning_rate": 9.948408548653074e-06,
      "loss": 0.3582,
      "step": 4240
    },
    {
      "epoch": 0.19462163278417696,
      "grad_norm": 0.5173108577728271,
      "learning_rate": 9.948373411362946e-06,
      "loss": 0.5267,
      "step": 4241
    },
    {
      "epoch": 0.1946675232894314,
      "grad_norm": 0.48662590980529785,
      "learning_rate": 9.948338262173543e-06,
      "loss": 0.467,
      "step": 4242
    },
    {
      "epoch": 0.19471341379468587,
      "grad_norm": 0.4564754068851471,
      "learning_rate": 9.948303101084947e-06,
      "loss": 0.3897,
      "step": 4243
    },
    {
      "epoch": 0.19475930429994034,
      "grad_norm": 0.4806521236896515,
      "learning_rate": 9.948267928097243e-06,
      "loss": 0.4413,
      "step": 4244
    },
    {
      "epoch": 0.19480519480519481,
      "grad_norm": 0.45602428913116455,
      "learning_rate": 9.948232743210516e-06,
      "loss": 0.385,
      "step": 4245
    },
    {
      "epoch": 0.19485108531044926,
      "grad_norm": 0.5353230834007263,
      "learning_rate": 9.948197546424851e-06,
      "loss": 0.4468,
      "step": 4246
    },
    {
      "epoch": 0.19489697581570373,
      "grad_norm": 0.4799993634223938,
      "learning_rate": 9.948162337740334e-06,
      "loss": 0.3978,
      "step": 4247
    },
    {
      "epoch": 0.1949428663209582,
      "grad_norm": 0.44750842452049255,
      "learning_rate": 9.948127117157048e-06,
      "loss": 0.39,
      "step": 4248
    },
    {
      "epoch": 0.19498875682621267,
      "grad_norm": 0.43253499269485474,
      "learning_rate": 9.948091884675077e-06,
      "loss": 0.3331,
      "step": 4249
    },
    {
      "epoch": 0.1950346473314671,
      "grad_norm": 0.47233980894088745,
      "learning_rate": 9.948056640294507e-06,
      "loss": 0.4055,
      "step": 4250
    },
    {
      "epoch": 0.19508053783672158,
      "grad_norm": 0.4375777542591095,
      "learning_rate": 9.948021384015424e-06,
      "loss": 0.3446,
      "step": 4251
    },
    {
      "epoch": 0.19512642834197605,
      "grad_norm": 0.5303660035133362,
      "learning_rate": 9.94798611583791e-06,
      "loss": 0.4573,
      "step": 4252
    },
    {
      "epoch": 0.19517231884723052,
      "grad_norm": 0.46682602167129517,
      "learning_rate": 9.947950835762051e-06,
      "loss": 0.4336,
      "step": 4253
    },
    {
      "epoch": 0.19521820935248496,
      "grad_norm": 0.48022106289863586,
      "learning_rate": 9.947915543787931e-06,
      "loss": 0.4261,
      "step": 4254
    },
    {
      "epoch": 0.19526409985773943,
      "grad_norm": 0.47179487347602844,
      "learning_rate": 9.947880239915637e-06,
      "loss": 0.474,
      "step": 4255
    },
    {
      "epoch": 0.1953099903629939,
      "grad_norm": 0.4498111307621002,
      "learning_rate": 9.947844924145253e-06,
      "loss": 0.367,
      "step": 4256
    },
    {
      "epoch": 0.19535588086824837,
      "grad_norm": 0.44921138882637024,
      "learning_rate": 9.947809596476862e-06,
      "loss": 0.3785,
      "step": 4257
    },
    {
      "epoch": 0.19540177137350281,
      "grad_norm": 0.47249871492385864,
      "learning_rate": 9.947774256910552e-06,
      "loss": 0.4707,
      "step": 4258
    },
    {
      "epoch": 0.19544766187875728,
      "grad_norm": 0.7527539134025574,
      "learning_rate": 9.947738905446405e-06,
      "loss": 0.4634,
      "step": 4259
    },
    {
      "epoch": 0.19549355238401175,
      "grad_norm": 0.4666089713573456,
      "learning_rate": 9.947703542084508e-06,
      "loss": 0.3485,
      "step": 4260
    },
    {
      "epoch": 0.19553944288926622,
      "grad_norm": 0.5009175539016724,
      "learning_rate": 9.947668166824947e-06,
      "loss": 0.4518,
      "step": 4261
    },
    {
      "epoch": 0.19558533339452067,
      "grad_norm": 0.5116076469421387,
      "learning_rate": 9.947632779667802e-06,
      "loss": 0.491,
      "step": 4262
    },
    {
      "epoch": 0.19563122389977514,
      "grad_norm": 0.4593135118484497,
      "learning_rate": 9.947597380613165e-06,
      "loss": 0.3643,
      "step": 4263
    },
    {
      "epoch": 0.1956771144050296,
      "grad_norm": 0.47461822628974915,
      "learning_rate": 9.947561969661117e-06,
      "loss": 0.4328,
      "step": 4264
    },
    {
      "epoch": 0.19572300491028405,
      "grad_norm": 0.47299692034721375,
      "learning_rate": 9.947526546811743e-06,
      "loss": 0.4174,
      "step": 4265
    },
    {
      "epoch": 0.19576889541553852,
      "grad_norm": 0.45413699746131897,
      "learning_rate": 9.94749111206513e-06,
      "loss": 0.3684,
      "step": 4266
    },
    {
      "epoch": 0.195814785920793,
      "grad_norm": 0.5074641704559326,
      "learning_rate": 9.947455665421361e-06,
      "loss": 0.3983,
      "step": 4267
    },
    {
      "epoch": 0.19586067642604746,
      "grad_norm": 0.535548746585846,
      "learning_rate": 9.947420206880525e-06,
      "loss": 0.5404,
      "step": 4268
    },
    {
      "epoch": 0.1959065669313019,
      "grad_norm": 0.44411972165107727,
      "learning_rate": 9.947384736442705e-06,
      "loss": 0.3346,
      "step": 4269
    },
    {
      "epoch": 0.19595245743655637,
      "grad_norm": 0.5077770948410034,
      "learning_rate": 9.947349254107983e-06,
      "loss": 0.476,
      "step": 4270
    },
    {
      "epoch": 0.19599834794181084,
      "grad_norm": 0.48464199900627136,
      "learning_rate": 9.94731375987645e-06,
      "loss": 0.4549,
      "step": 4271
    },
    {
      "epoch": 0.1960442384470653,
      "grad_norm": 0.42790719866752625,
      "learning_rate": 9.947278253748188e-06,
      "loss": 0.3358,
      "step": 4272
    },
    {
      "epoch": 0.19609012895231975,
      "grad_norm": 0.48634856939315796,
      "learning_rate": 9.947242735723284e-06,
      "loss": 0.3404,
      "step": 4273
    },
    {
      "epoch": 0.19613601945757422,
      "grad_norm": 0.5001307725906372,
      "learning_rate": 9.947207205801821e-06,
      "loss": 0.4408,
      "step": 4274
    },
    {
      "epoch": 0.1961819099628287,
      "grad_norm": 0.5179278254508972,
      "learning_rate": 9.947171663983889e-06,
      "loss": 0.4575,
      "step": 4275
    },
    {
      "epoch": 0.19622780046808316,
      "grad_norm": 0.4798543453216553,
      "learning_rate": 9.947136110269566e-06,
      "loss": 0.4852,
      "step": 4276
    },
    {
      "epoch": 0.1962736909733376,
      "grad_norm": 0.4629528224468231,
      "learning_rate": 9.947100544658944e-06,
      "loss": 0.4064,
      "step": 4277
    },
    {
      "epoch": 0.19631958147859208,
      "grad_norm": 0.5005196928977966,
      "learning_rate": 9.947064967152107e-06,
      "loss": 0.4438,
      "step": 4278
    },
    {
      "epoch": 0.19636547198384655,
      "grad_norm": 0.5028572678565979,
      "learning_rate": 9.947029377749141e-06,
      "loss": 0.4699,
      "step": 4279
    },
    {
      "epoch": 0.19641136248910102,
      "grad_norm": 0.5163561105728149,
      "learning_rate": 9.94699377645013e-06,
      "loss": 0.457,
      "step": 4280
    },
    {
      "epoch": 0.19645725299435546,
      "grad_norm": 0.45968979597091675,
      "learning_rate": 9.94695816325516e-06,
      "loss": 0.4245,
      "step": 4281
    },
    {
      "epoch": 0.19650314349960993,
      "grad_norm": 0.42907556891441345,
      "learning_rate": 9.946922538164316e-06,
      "loss": 0.3038,
      "step": 4282
    },
    {
      "epoch": 0.1965490340048644,
      "grad_norm": 0.46068835258483887,
      "learning_rate": 9.946886901177687e-06,
      "loss": 0.4054,
      "step": 4283
    },
    {
      "epoch": 0.19659492451011887,
      "grad_norm": 0.4411108195781708,
      "learning_rate": 9.946851252295356e-06,
      "loss": 0.358,
      "step": 4284
    },
    {
      "epoch": 0.1966408150153733,
      "grad_norm": 0.44611024856567383,
      "learning_rate": 9.946815591517407e-06,
      "loss": 0.373,
      "step": 4285
    },
    {
      "epoch": 0.19668670552062778,
      "grad_norm": 0.40401241183280945,
      "learning_rate": 9.946779918843928e-06,
      "loss": 0.2614,
      "step": 4286
    },
    {
      "epoch": 0.19673259602588225,
      "grad_norm": 0.4756178855895996,
      "learning_rate": 9.946744234275005e-06,
      "loss": 0.3595,
      "step": 4287
    },
    {
      "epoch": 0.19677848653113672,
      "grad_norm": 0.485768586397171,
      "learning_rate": 9.946708537810723e-06,
      "loss": 0.4089,
      "step": 4288
    },
    {
      "epoch": 0.19682437703639116,
      "grad_norm": 0.5151448845863342,
      "learning_rate": 9.94667282945117e-06,
      "loss": 0.4467,
      "step": 4289
    },
    {
      "epoch": 0.19687026754164563,
      "grad_norm": 0.48313891887664795,
      "learning_rate": 9.946637109196428e-06,
      "loss": 0.419,
      "step": 4290
    },
    {
      "epoch": 0.1969161580469001,
      "grad_norm": 0.5669131278991699,
      "learning_rate": 9.946601377046586e-06,
      "loss": 0.5904,
      "step": 4291
    },
    {
      "epoch": 0.19696204855215457,
      "grad_norm": 0.49674496054649353,
      "learning_rate": 9.94656563300173e-06,
      "loss": 0.4869,
      "step": 4292
    },
    {
      "epoch": 0.19700793905740902,
      "grad_norm": 0.4247874319553375,
      "learning_rate": 9.946529877061942e-06,
      "loss": 0.2985,
      "step": 4293
    },
    {
      "epoch": 0.19705382956266349,
      "grad_norm": 0.46898841857910156,
      "learning_rate": 9.946494109227313e-06,
      "loss": 0.3638,
      "step": 4294
    },
    {
      "epoch": 0.19709972006791796,
      "grad_norm": 0.49758079648017883,
      "learning_rate": 9.946458329497926e-06,
      "loss": 0.4661,
      "step": 4295
    },
    {
      "epoch": 0.1971456105731724,
      "grad_norm": 0.483528196811676,
      "learning_rate": 9.946422537873867e-06,
      "loss": 0.426,
      "step": 4296
    },
    {
      "epoch": 0.19719150107842687,
      "grad_norm": 0.42439407110214233,
      "learning_rate": 9.946386734355224e-06,
      "loss": 0.3243,
      "step": 4297
    },
    {
      "epoch": 0.19723739158368134,
      "grad_norm": 0.48264139890670776,
      "learning_rate": 9.946350918942082e-06,
      "loss": 0.4228,
      "step": 4298
    },
    {
      "epoch": 0.1972832820889358,
      "grad_norm": 0.5076059699058533,
      "learning_rate": 9.946315091634527e-06,
      "loss": 0.4116,
      "step": 4299
    },
    {
      "epoch": 0.19732917259419025,
      "grad_norm": 0.47341057658195496,
      "learning_rate": 9.946279252432646e-06,
      "loss": 0.3503,
      "step": 4300
    },
    {
      "epoch": 0.19737506309944472,
      "grad_norm": 0.48038241267204285,
      "learning_rate": 9.946243401336524e-06,
      "loss": 0.4433,
      "step": 4301
    },
    {
      "epoch": 0.1974209536046992,
      "grad_norm": 0.5301975011825562,
      "learning_rate": 9.946207538346247e-06,
      "loss": 0.4859,
      "step": 4302
    },
    {
      "epoch": 0.19746684410995366,
      "grad_norm": 0.4284140467643738,
      "learning_rate": 9.946171663461904e-06,
      "loss": 0.3626,
      "step": 4303
    },
    {
      "epoch": 0.1975127346152081,
      "grad_norm": 0.4876924753189087,
      "learning_rate": 9.946135776683576e-06,
      "loss": 0.4888,
      "step": 4304
    },
    {
      "epoch": 0.19755862512046257,
      "grad_norm": 0.5320894122123718,
      "learning_rate": 9.946099878011355e-06,
      "loss": 0.4871,
      "step": 4305
    },
    {
      "epoch": 0.19760451562571704,
      "grad_norm": 0.47058752179145813,
      "learning_rate": 9.946063967445325e-06,
      "loss": 0.4272,
      "step": 4306
    },
    {
      "epoch": 0.1976504061309715,
      "grad_norm": 0.42561522126197815,
      "learning_rate": 9.94602804498557e-06,
      "loss": 0.3609,
      "step": 4307
    },
    {
      "epoch": 0.19769629663622595,
      "grad_norm": 0.46116769313812256,
      "learning_rate": 9.94599211063218e-06,
      "loss": 0.411,
      "step": 4308
    },
    {
      "epoch": 0.19774218714148042,
      "grad_norm": 0.44296127557754517,
      "learning_rate": 9.94595616438524e-06,
      "loss": 0.395,
      "step": 4309
    },
    {
      "epoch": 0.1977880776467349,
      "grad_norm": 0.5113347768783569,
      "learning_rate": 9.945920206244837e-06,
      "loss": 0.4722,
      "step": 4310
    },
    {
      "epoch": 0.19783396815198936,
      "grad_norm": 0.50044184923172,
      "learning_rate": 9.945884236211055e-06,
      "loss": 0.4721,
      "step": 4311
    },
    {
      "epoch": 0.1978798586572438,
      "grad_norm": 0.4492280185222626,
      "learning_rate": 9.945848254283984e-06,
      "loss": 0.3816,
      "step": 4312
    },
    {
      "epoch": 0.19792574916249828,
      "grad_norm": 0.4583589732646942,
      "learning_rate": 9.945812260463707e-06,
      "loss": 0.3461,
      "step": 4313
    },
    {
      "epoch": 0.19797163966775275,
      "grad_norm": 0.4791368842124939,
      "learning_rate": 9.945776254750314e-06,
      "loss": 0.4349,
      "step": 4314
    },
    {
      "epoch": 0.19801753017300722,
      "grad_norm": 0.477237731218338,
      "learning_rate": 9.94574023714389e-06,
      "loss": 0.4792,
      "step": 4315
    },
    {
      "epoch": 0.19806342067826166,
      "grad_norm": 0.47030895948410034,
      "learning_rate": 9.945704207644523e-06,
      "loss": 0.3728,
      "step": 4316
    },
    {
      "epoch": 0.19810931118351613,
      "grad_norm": 0.4306229054927826,
      "learning_rate": 9.945668166252296e-06,
      "loss": 0.3525,
      "step": 4317
    },
    {
      "epoch": 0.1981552016887706,
      "grad_norm": 0.5677311420440674,
      "learning_rate": 9.945632112967298e-06,
      "loss": 0.4834,
      "step": 4318
    },
    {
      "epoch": 0.19820109219402507,
      "grad_norm": 0.4629094898700714,
      "learning_rate": 9.945596047789617e-06,
      "loss": 0.3952,
      "step": 4319
    },
    {
      "epoch": 0.1982469826992795,
      "grad_norm": 0.5262762308120728,
      "learning_rate": 9.945559970719339e-06,
      "loss": 0.5055,
      "step": 4320
    },
    {
      "epoch": 0.19829287320453398,
      "grad_norm": 0.4672391712665558,
      "learning_rate": 9.945523881756547e-06,
      "loss": 0.3768,
      "step": 4321
    },
    {
      "epoch": 0.19833876370978845,
      "grad_norm": 0.47864776849746704,
      "learning_rate": 9.945487780901334e-06,
      "loss": 0.4429,
      "step": 4322
    },
    {
      "epoch": 0.1983846542150429,
      "grad_norm": 0.4701748788356781,
      "learning_rate": 9.945451668153783e-06,
      "loss": 0.424,
      "step": 4323
    },
    {
      "epoch": 0.19843054472029736,
      "grad_norm": 0.4563615620136261,
      "learning_rate": 9.94541554351398e-06,
      "loss": 0.3921,
      "step": 4324
    },
    {
      "epoch": 0.19847643522555183,
      "grad_norm": 0.48338043689727783,
      "learning_rate": 9.945379406982015e-06,
      "loss": 0.4768,
      "step": 4325
    },
    {
      "epoch": 0.1985223257308063,
      "grad_norm": 0.4890962839126587,
      "learning_rate": 9.945343258557973e-06,
      "loss": 0.4172,
      "step": 4326
    },
    {
      "epoch": 0.19856821623606075,
      "grad_norm": 0.5320715308189392,
      "learning_rate": 9.945307098241941e-06,
      "loss": 0.3265,
      "step": 4327
    },
    {
      "epoch": 0.19861410674131522,
      "grad_norm": 0.4402206540107727,
      "learning_rate": 9.945270926034007e-06,
      "loss": 0.3586,
      "step": 4328
    },
    {
      "epoch": 0.1986599972465697,
      "grad_norm": 0.5281245708465576,
      "learning_rate": 9.945234741934255e-06,
      "loss": 0.4942,
      "step": 4329
    },
    {
      "epoch": 0.19870588775182416,
      "grad_norm": 0.5047677755355835,
      "learning_rate": 9.945198545942778e-06,
      "loss": 0.489,
      "step": 4330
    },
    {
      "epoch": 0.1987517782570786,
      "grad_norm": 0.4735115170478821,
      "learning_rate": 9.945162338059655e-06,
      "loss": 0.3846,
      "step": 4331
    },
    {
      "epoch": 0.19879766876233307,
      "grad_norm": 0.49209386110305786,
      "learning_rate": 9.94512611828498e-06,
      "loss": 0.4111,
      "step": 4332
    },
    {
      "epoch": 0.19884355926758754,
      "grad_norm": 0.5027639865875244,
      "learning_rate": 9.945089886618838e-06,
      "loss": 0.4653,
      "step": 4333
    },
    {
      "epoch": 0.198889449772842,
      "grad_norm": 0.4580379128456116,
      "learning_rate": 9.945053643061314e-06,
      "loss": 0.3582,
      "step": 4334
    },
    {
      "epoch": 0.19893534027809645,
      "grad_norm": 0.5238893628120422,
      "learning_rate": 9.945017387612497e-06,
      "loss": 0.4862,
      "step": 4335
    },
    {
      "epoch": 0.19898123078335092,
      "grad_norm": 0.4721764922142029,
      "learning_rate": 9.944981120272472e-06,
      "loss": 0.3906,
      "step": 4336
    },
    {
      "epoch": 0.1990271212886054,
      "grad_norm": 0.5390194058418274,
      "learning_rate": 9.94494484104133e-06,
      "loss": 0.395,
      "step": 4337
    },
    {
      "epoch": 0.19907301179385986,
      "grad_norm": 0.4833170771598816,
      "learning_rate": 9.944908549919157e-06,
      "loss": 0.4092,
      "step": 4338
    },
    {
      "epoch": 0.1991189022991143,
      "grad_norm": 0.4220992624759674,
      "learning_rate": 9.944872246906039e-06,
      "loss": 0.3035,
      "step": 4339
    },
    {
      "epoch": 0.19916479280436877,
      "grad_norm": 0.46783629059791565,
      "learning_rate": 9.944835932002064e-06,
      "loss": 0.3739,
      "step": 4340
    },
    {
      "epoch": 0.19921068330962324,
      "grad_norm": 0.5067040324211121,
      "learning_rate": 9.944799605207318e-06,
      "loss": 0.4772,
      "step": 4341
    },
    {
      "epoch": 0.1992565738148777,
      "grad_norm": 0.4913345277309418,
      "learning_rate": 9.94476326652189e-06,
      "loss": 0.4322,
      "step": 4342
    },
    {
      "epoch": 0.19930246432013216,
      "grad_norm": 0.4785483777523041,
      "learning_rate": 9.944726915945868e-06,
      "loss": 0.4452,
      "step": 4343
    },
    {
      "epoch": 0.19934835482538663,
      "grad_norm": 0.5592127442359924,
      "learning_rate": 9.94469055347934e-06,
      "loss": 0.4301,
      "step": 4344
    },
    {
      "epoch": 0.1993942453306411,
      "grad_norm": 0.47358712553977966,
      "learning_rate": 9.944654179122388e-06,
      "loss": 0.4043,
      "step": 4345
    },
    {
      "epoch": 0.19944013583589557,
      "grad_norm": 0.4482409358024597,
      "learning_rate": 9.944617792875105e-06,
      "loss": 0.3777,
      "step": 4346
    },
    {
      "epoch": 0.19948602634115,
      "grad_norm": 0.45446881651878357,
      "learning_rate": 9.944581394737576e-06,
      "loss": 0.3884,
      "step": 4347
    },
    {
      "epoch": 0.19953191684640448,
      "grad_norm": 0.4457714259624481,
      "learning_rate": 9.944544984709892e-06,
      "loss": 0.3887,
      "step": 4348
    },
    {
      "epoch": 0.19957780735165895,
      "grad_norm": 0.5872363448143005,
      "learning_rate": 9.944508562792134e-06,
      "loss": 0.4737,
      "step": 4349
    },
    {
      "epoch": 0.19962369785691342,
      "grad_norm": 0.478573203086853,
      "learning_rate": 9.944472128984396e-06,
      "loss": 0.4174,
      "step": 4350
    },
    {
      "epoch": 0.19966958836216786,
      "grad_norm": 0.47778937220573425,
      "learning_rate": 9.944435683286762e-06,
      "loss": 0.3973,
      "step": 4351
    },
    {
      "epoch": 0.19971547886742233,
      "grad_norm": 0.47848138213157654,
      "learning_rate": 9.944399225699321e-06,
      "loss": 0.4402,
      "step": 4352
    },
    {
      "epoch": 0.1997613693726768,
      "grad_norm": 0.5309686660766602,
      "learning_rate": 9.94436275622216e-06,
      "loss": 0.4762,
      "step": 4353
    },
    {
      "epoch": 0.19980725987793124,
      "grad_norm": 0.4534291923046112,
      "learning_rate": 9.94432627485537e-06,
      "loss": 0.3333,
      "step": 4354
    },
    {
      "epoch": 0.1998531503831857,
      "grad_norm": 0.46538591384887695,
      "learning_rate": 9.944289781599033e-06,
      "loss": 0.3901,
      "step": 4355
    },
    {
      "epoch": 0.19989904088844018,
      "grad_norm": 0.8436093926429749,
      "learning_rate": 9.944253276453238e-06,
      "loss": 0.4983,
      "step": 4356
    },
    {
      "epoch": 0.19994493139369465,
      "grad_norm": 0.46009090542793274,
      "learning_rate": 9.944216759418078e-06,
      "loss": 0.3762,
      "step": 4357
    },
    {
      "epoch": 0.1999908218989491,
      "grad_norm": 0.4920811057090759,
      "learning_rate": 9.944180230493634e-06,
      "loss": 0.4398,
      "step": 4358
    },
    {
      "epoch": 0.20003671240420356,
      "grad_norm": 0.4614555537700653,
      "learning_rate": 9.944143689679999e-06,
      "loss": 0.4109,
      "step": 4359
    },
    {
      "epoch": 0.20008260290945803,
      "grad_norm": 0.45403623580932617,
      "learning_rate": 9.944107136977256e-06,
      "loss": 0.3989,
      "step": 4360
    },
    {
      "epoch": 0.2001284934147125,
      "grad_norm": 0.4971005916595459,
      "learning_rate": 9.944070572385499e-06,
      "loss": 0.4265,
      "step": 4361
    },
    {
      "epoch": 0.20017438391996695,
      "grad_norm": 0.47762706875801086,
      "learning_rate": 9.944033995904812e-06,
      "loss": 0.3909,
      "step": 4362
    },
    {
      "epoch": 0.20022027442522142,
      "grad_norm": 0.4471714496612549,
      "learning_rate": 9.943997407535283e-06,
      "loss": 0.3805,
      "step": 4363
    },
    {
      "epoch": 0.2002661649304759,
      "grad_norm": 0.4476510286331177,
      "learning_rate": 9.943960807277e-06,
      "loss": 0.4116,
      "step": 4364
    },
    {
      "epoch": 0.20031205543573036,
      "grad_norm": 0.4717048406600952,
      "learning_rate": 9.943924195130052e-06,
      "loss": 0.4657,
      "step": 4365
    },
    {
      "epoch": 0.2003579459409848,
      "grad_norm": 0.4262549579143524,
      "learning_rate": 9.943887571094529e-06,
      "loss": 0.353,
      "step": 4366
    },
    {
      "epoch": 0.20040383644623927,
      "grad_norm": 0.4871671795845032,
      "learning_rate": 9.943850935170513e-06,
      "loss": 0.4918,
      "step": 4367
    },
    {
      "epoch": 0.20044972695149374,
      "grad_norm": 0.4724729061126709,
      "learning_rate": 9.943814287358099e-06,
      "loss": 0.4267,
      "step": 4368
    },
    {
      "epoch": 0.2004956174567482,
      "grad_norm": 0.45711904764175415,
      "learning_rate": 9.94377762765737e-06,
      "loss": 0.3473,
      "step": 4369
    },
    {
      "epoch": 0.20054150796200265,
      "grad_norm": 0.44956108927726746,
      "learning_rate": 9.943740956068416e-06,
      "loss": 0.313,
      "step": 4370
    },
    {
      "epoch": 0.20058739846725712,
      "grad_norm": 0.49146658182144165,
      "learning_rate": 9.943704272591325e-06,
      "loss": 0.3787,
      "step": 4371
    },
    {
      "epoch": 0.2006332889725116,
      "grad_norm": 0.4804185628890991,
      "learning_rate": 9.943667577226188e-06,
      "loss": 0.4134,
      "step": 4372
    },
    {
      "epoch": 0.20067917947776606,
      "grad_norm": 0.501727283000946,
      "learning_rate": 9.94363086997309e-06,
      "loss": 0.4848,
      "step": 4373
    },
    {
      "epoch": 0.2007250699830205,
      "grad_norm": 0.4757954180240631,
      "learning_rate": 9.943594150832117e-06,
      "loss": 0.3426,
      "step": 4374
    },
    {
      "epoch": 0.20077096048827497,
      "grad_norm": 0.5328327417373657,
      "learning_rate": 9.943557419803362e-06,
      "loss": 0.5253,
      "step": 4375
    },
    {
      "epoch": 0.20081685099352944,
      "grad_norm": 0.4832775890827179,
      "learning_rate": 9.943520676886914e-06,
      "loss": 0.4047,
      "step": 4376
    },
    {
      "epoch": 0.20086274149878391,
      "grad_norm": 0.4592137932777405,
      "learning_rate": 9.943483922082857e-06,
      "loss": 0.3798,
      "step": 4377
    },
    {
      "epoch": 0.20090863200403836,
      "grad_norm": 0.49707669019699097,
      "learning_rate": 9.943447155391281e-06,
      "loss": 0.452,
      "step": 4378
    },
    {
      "epoch": 0.20095452250929283,
      "grad_norm": 0.45820289850234985,
      "learning_rate": 9.943410376812274e-06,
      "loss": 0.3892,
      "step": 4379
    },
    {
      "epoch": 0.2010004130145473,
      "grad_norm": 0.47397100925445557,
      "learning_rate": 9.943373586345928e-06,
      "loss": 0.4722,
      "step": 4380
    },
    {
      "epoch": 0.20104630351980177,
      "grad_norm": 0.46828311681747437,
      "learning_rate": 9.943336783992327e-06,
      "loss": 0.4242,
      "step": 4381
    },
    {
      "epoch": 0.2010921940250562,
      "grad_norm": 0.45665767788887024,
      "learning_rate": 9.94329996975156e-06,
      "loss": 0.3851,
      "step": 4382
    },
    {
      "epoch": 0.20113808453031068,
      "grad_norm": 0.4675580561161041,
      "learning_rate": 9.943263143623719e-06,
      "loss": 0.3954,
      "step": 4383
    },
    {
      "epoch": 0.20118397503556515,
      "grad_norm": 0.45406094193458557,
      "learning_rate": 9.943226305608889e-06,
      "loss": 0.4383,
      "step": 4384
    },
    {
      "epoch": 0.2012298655408196,
      "grad_norm": 0.44739025831222534,
      "learning_rate": 9.943189455707161e-06,
      "loss": 0.3752,
      "step": 4385
    },
    {
      "epoch": 0.20127575604607406,
      "grad_norm": 0.5119508504867554,
      "learning_rate": 9.943152593918622e-06,
      "loss": 0.4103,
      "step": 4386
    },
    {
      "epoch": 0.20132164655132853,
      "grad_norm": 0.49356532096862793,
      "learning_rate": 9.943115720243361e-06,
      "loss": 0.3819,
      "step": 4387
    },
    {
      "epoch": 0.201367537056583,
      "grad_norm": 0.4203157424926758,
      "learning_rate": 9.943078834681465e-06,
      "loss": 0.3681,
      "step": 4388
    },
    {
      "epoch": 0.20141342756183744,
      "grad_norm": 0.5196059942245483,
      "learning_rate": 9.943041937233027e-06,
      "loss": 0.5161,
      "step": 4389
    },
    {
      "epoch": 0.2014593180670919,
      "grad_norm": 0.4554559290409088,
      "learning_rate": 9.943005027898131e-06,
      "loss": 0.3392,
      "step": 4390
    },
    {
      "epoch": 0.20150520857234638,
      "grad_norm": 0.4803728759288788,
      "learning_rate": 9.942968106676871e-06,
      "loss": 0.3597,
      "step": 4391
    },
    {
      "epoch": 0.20155109907760085,
      "grad_norm": 0.5130305886268616,
      "learning_rate": 9.94293117356933e-06,
      "loss": 0.4558,
      "step": 4392
    },
    {
      "epoch": 0.2015969895828553,
      "grad_norm": 0.4634862542152405,
      "learning_rate": 9.942894228575601e-06,
      "loss": 0.405,
      "step": 4393
    },
    {
      "epoch": 0.20164288008810977,
      "grad_norm": 0.4780745804309845,
      "learning_rate": 9.942857271695772e-06,
      "loss": 0.4258,
      "step": 4394
    },
    {
      "epoch": 0.20168877059336424,
      "grad_norm": 0.4796124994754791,
      "learning_rate": 9.94282030292993e-06,
      "loss": 0.4351,
      "step": 4395
    },
    {
      "epoch": 0.2017346610986187,
      "grad_norm": 0.4890759289264679,
      "learning_rate": 9.942783322278165e-06,
      "loss": 0.5176,
      "step": 4396
    },
    {
      "epoch": 0.20178055160387315,
      "grad_norm": 0.4720701277256012,
      "learning_rate": 9.942746329740566e-06,
      "loss": 0.3985,
      "step": 4397
    },
    {
      "epoch": 0.20182644210912762,
      "grad_norm": 0.47165292501449585,
      "learning_rate": 9.942709325317224e-06,
      "loss": 0.383,
      "step": 4398
    },
    {
      "epoch": 0.2018723326143821,
      "grad_norm": 0.43780720233917236,
      "learning_rate": 9.942672309008223e-06,
      "loss": 0.3484,
      "step": 4399
    },
    {
      "epoch": 0.20191822311963656,
      "grad_norm": 0.4558258652687073,
      "learning_rate": 9.942635280813658e-06,
      "loss": 0.3921,
      "step": 4400
    },
    {
      "epoch": 0.201964113624891,
      "grad_norm": 0.475436270236969,
      "learning_rate": 9.942598240733613e-06,
      "loss": 0.4596,
      "step": 4401
    },
    {
      "epoch": 0.20201000413014547,
      "grad_norm": 0.4847857654094696,
      "learning_rate": 9.94256118876818e-06,
      "loss": 0.4741,
      "step": 4402
    },
    {
      "epoch": 0.20205589463539994,
      "grad_norm": 0.467704176902771,
      "learning_rate": 9.942524124917447e-06,
      "loss": 0.3736,
      "step": 4403
    },
    {
      "epoch": 0.2021017851406544,
      "grad_norm": 0.47344571352005005,
      "learning_rate": 9.942487049181504e-06,
      "loss": 0.4159,
      "step": 4404
    },
    {
      "epoch": 0.20214767564590885,
      "grad_norm": 0.44193920493125916,
      "learning_rate": 9.942449961560439e-06,
      "loss": 0.3511,
      "step": 4405
    },
    {
      "epoch": 0.20219356615116332,
      "grad_norm": 0.5351225137710571,
      "learning_rate": 9.94241286205434e-06,
      "loss": 0.3977,
      "step": 4406
    },
    {
      "epoch": 0.2022394566564178,
      "grad_norm": 0.45293131470680237,
      "learning_rate": 9.942375750663302e-06,
      "loss": 0.3726,
      "step": 4407
    },
    {
      "epoch": 0.20228534716167226,
      "grad_norm": 0.4660302698612213,
      "learning_rate": 9.942338627387408e-06,
      "loss": 0.3984,
      "step": 4408
    },
    {
      "epoch": 0.2023312376669267,
      "grad_norm": 0.4848549962043762,
      "learning_rate": 9.942301492226749e-06,
      "loss": 0.4585,
      "step": 4409
    },
    {
      "epoch": 0.20237712817218118,
      "grad_norm": 0.49276554584503174,
      "learning_rate": 9.942264345181414e-06,
      "loss": 0.4164,
      "step": 4410
    },
    {
      "epoch": 0.20242301867743565,
      "grad_norm": 0.4952569305896759,
      "learning_rate": 9.942227186251495e-06,
      "loss": 0.4355,
      "step": 4411
    },
    {
      "epoch": 0.2024689091826901,
      "grad_norm": 0.608250617980957,
      "learning_rate": 9.94219001543708e-06,
      "loss": 0.5878,
      "step": 4412
    },
    {
      "epoch": 0.20251479968794456,
      "grad_norm": 0.4380188584327698,
      "learning_rate": 9.942152832738255e-06,
      "loss": 0.3362,
      "step": 4413
    },
    {
      "epoch": 0.20256069019319903,
      "grad_norm": 0.4906007647514343,
      "learning_rate": 9.942115638155114e-06,
      "loss": 0.445,
      "step": 4414
    },
    {
      "epoch": 0.2026065806984535,
      "grad_norm": 0.45186811685562134,
      "learning_rate": 9.942078431687744e-06,
      "loss": 0.3808,
      "step": 4415
    },
    {
      "epoch": 0.20265247120370794,
      "grad_norm": 0.4381430745124817,
      "learning_rate": 9.942041213336235e-06,
      "loss": 0.3271,
      "step": 4416
    },
    {
      "epoch": 0.2026983617089624,
      "grad_norm": 0.48071515560150146,
      "learning_rate": 9.942003983100678e-06,
      "loss": 0.4215,
      "step": 4417
    },
    {
      "epoch": 0.20274425221421688,
      "grad_norm": 0.4494588375091553,
      "learning_rate": 9.941966740981159e-06,
      "loss": 0.327,
      "step": 4418
    },
    {
      "epoch": 0.20279014271947135,
      "grad_norm": 0.4773856997489929,
      "learning_rate": 9.941929486977773e-06,
      "loss": 0.4035,
      "step": 4419
    },
    {
      "epoch": 0.2028360332247258,
      "grad_norm": 0.45517271757125854,
      "learning_rate": 9.941892221090604e-06,
      "loss": 0.3514,
      "step": 4420
    },
    {
      "epoch": 0.20288192372998026,
      "grad_norm": 0.4729015827178955,
      "learning_rate": 9.941854943319744e-06,
      "loss": 0.383,
      "step": 4421
    },
    {
      "epoch": 0.20292781423523473,
      "grad_norm": 0.5019860863685608,
      "learning_rate": 9.941817653665282e-06,
      "loss": 0.4457,
      "step": 4422
    },
    {
      "epoch": 0.2029737047404892,
      "grad_norm": 0.4792742431163788,
      "learning_rate": 9.94178035212731e-06,
      "loss": 0.4153,
      "step": 4423
    },
    {
      "epoch": 0.20301959524574364,
      "grad_norm": 0.44474557042121887,
      "learning_rate": 9.941743038705914e-06,
      "loss": 0.376,
      "step": 4424
    },
    {
      "epoch": 0.20306548575099811,
      "grad_norm": 0.4339393973350525,
      "learning_rate": 9.941705713401186e-06,
      "loss": 0.3541,
      "step": 4425
    },
    {
      "epoch": 0.20311137625625258,
      "grad_norm": 0.49014732241630554,
      "learning_rate": 9.941668376213216e-06,
      "loss": 0.4878,
      "step": 4426
    },
    {
      "epoch": 0.20315726676150705,
      "grad_norm": 0.4729020893573761,
      "learning_rate": 9.941631027142093e-06,
      "loss": 0.3974,
      "step": 4427
    },
    {
      "epoch": 0.2032031572667615,
      "grad_norm": 0.4551442861557007,
      "learning_rate": 9.941593666187907e-06,
      "loss": 0.3624,
      "step": 4428
    },
    {
      "epoch": 0.20324904777201597,
      "grad_norm": 0.44830113649368286,
      "learning_rate": 9.94155629335075e-06,
      "loss": 0.4464,
      "step": 4429
    },
    {
      "epoch": 0.20329493827727044,
      "grad_norm": 0.47621333599090576,
      "learning_rate": 9.941518908630707e-06,
      "loss": 0.4121,
      "step": 4430
    },
    {
      "epoch": 0.2033408287825249,
      "grad_norm": 0.4960765540599823,
      "learning_rate": 9.94148151202787e-06,
      "loss": 0.4427,
      "step": 4431
    },
    {
      "epoch": 0.20338671928777935,
      "grad_norm": 0.45448818802833557,
      "learning_rate": 9.941444103542332e-06,
      "loss": 0.3995,
      "step": 4432
    },
    {
      "epoch": 0.20343260979303382,
      "grad_norm": 0.4503958225250244,
      "learning_rate": 9.941406683174179e-06,
      "loss": 0.37,
      "step": 4433
    },
    {
      "epoch": 0.2034785002982883,
      "grad_norm": 0.5781404376029968,
      "learning_rate": 9.941369250923503e-06,
      "loss": 0.4018,
      "step": 4434
    },
    {
      "epoch": 0.20352439080354276,
      "grad_norm": 0.4687477648258209,
      "learning_rate": 9.941331806790392e-06,
      "loss": 0.3866,
      "step": 4435
    },
    {
      "epoch": 0.2035702813087972,
      "grad_norm": 0.42383551597595215,
      "learning_rate": 9.94129435077494e-06,
      "loss": 0.316,
      "step": 4436
    },
    {
      "epoch": 0.20361617181405167,
      "grad_norm": 0.46412205696105957,
      "learning_rate": 9.941256882877234e-06,
      "loss": 0.3539,
      "step": 4437
    },
    {
      "epoch": 0.20366206231930614,
      "grad_norm": 0.47226232290267944,
      "learning_rate": 9.941219403097365e-06,
      "loss": 0.3929,
      "step": 4438
    },
    {
      "epoch": 0.2037079528245606,
      "grad_norm": 0.4539547562599182,
      "learning_rate": 9.94118191143542e-06,
      "loss": 0.3962,
      "step": 4439
    },
    {
      "epoch": 0.20375384332981505,
      "grad_norm": 0.4528510570526123,
      "learning_rate": 9.941144407891495e-06,
      "loss": 0.3672,
      "step": 4440
    },
    {
      "epoch": 0.20379973383506952,
      "grad_norm": 0.464222252368927,
      "learning_rate": 9.941106892465675e-06,
      "loss": 0.3846,
      "step": 4441
    },
    {
      "epoch": 0.203845624340324,
      "grad_norm": 0.45024967193603516,
      "learning_rate": 9.941069365158054e-06,
      "loss": 0.3263,
      "step": 4442
    },
    {
      "epoch": 0.20389151484557844,
      "grad_norm": 0.466435045003891,
      "learning_rate": 9.941031825968722e-06,
      "loss": 0.4019,
      "step": 4443
    },
    {
      "epoch": 0.2039374053508329,
      "grad_norm": 0.49162209033966064,
      "learning_rate": 9.940994274897766e-06,
      "loss": 0.5259,
      "step": 4444
    },
    {
      "epoch": 0.20398329585608738,
      "grad_norm": 0.47819215059280396,
      "learning_rate": 9.940956711945278e-06,
      "loss": 0.4076,
      "step": 4445
    },
    {
      "epoch": 0.20402918636134185,
      "grad_norm": 0.4855642020702362,
      "learning_rate": 9.940919137111348e-06,
      "loss": 0.4123,
      "step": 4446
    },
    {
      "epoch": 0.2040750768665963,
      "grad_norm": 0.49447107315063477,
      "learning_rate": 9.940881550396068e-06,
      "loss": 0.4585,
      "step": 4447
    },
    {
      "epoch": 0.20412096737185076,
      "grad_norm": 0.4881538450717926,
      "learning_rate": 9.940843951799526e-06,
      "loss": 0.4688,
      "step": 4448
    },
    {
      "epoch": 0.20416685787710523,
      "grad_norm": 0.40869176387786865,
      "learning_rate": 9.940806341321816e-06,
      "loss": 0.3085,
      "step": 4449
    },
    {
      "epoch": 0.2042127483823597,
      "grad_norm": 0.4979475140571594,
      "learning_rate": 9.940768718963023e-06,
      "loss": 0.4415,
      "step": 4450
    },
    {
      "epoch": 0.20425863888761414,
      "grad_norm": 0.47777259349823,
      "learning_rate": 9.940731084723243e-06,
      "loss": 0.4227,
      "step": 4451
    },
    {
      "epoch": 0.2043045293928686,
      "grad_norm": 0.4737626612186432,
      "learning_rate": 9.940693438602563e-06,
      "loss": 0.4377,
      "step": 4452
    },
    {
      "epoch": 0.20435041989812308,
      "grad_norm": 0.504940390586853,
      "learning_rate": 9.940655780601075e-06,
      "loss": 0.4374,
      "step": 4453
    },
    {
      "epoch": 0.20439631040337755,
      "grad_norm": 0.4666500985622406,
      "learning_rate": 9.94061811071887e-06,
      "loss": 0.4497,
      "step": 4454
    },
    {
      "epoch": 0.204442200908632,
      "grad_norm": 0.5264491438865662,
      "learning_rate": 9.940580428956038e-06,
      "loss": 0.4791,
      "step": 4455
    },
    {
      "epoch": 0.20448809141388646,
      "grad_norm": 0.525458037853241,
      "learning_rate": 9.940542735312667e-06,
      "loss": 0.486,
      "step": 4456
    },
    {
      "epoch": 0.20453398191914093,
      "grad_norm": 0.4548773169517517,
      "learning_rate": 9.94050502978885e-06,
      "loss": 0.3781,
      "step": 4457
    },
    {
      "epoch": 0.2045798724243954,
      "grad_norm": 0.48906221985816956,
      "learning_rate": 9.940467312384679e-06,
      "loss": 0.4616,
      "step": 4458
    },
    {
      "epoch": 0.20462576292964985,
      "grad_norm": 0.5151055455207825,
      "learning_rate": 9.940429583100244e-06,
      "loss": 0.4066,
      "step": 4459
    },
    {
      "epoch": 0.20467165343490432,
      "grad_norm": 0.4916255474090576,
      "learning_rate": 9.940391841935634e-06,
      "loss": 0.3986,
      "step": 4460
    },
    {
      "epoch": 0.20471754394015879,
      "grad_norm": 0.5003378391265869,
      "learning_rate": 9.940354088890942e-06,
      "loss": 0.4381,
      "step": 4461
    },
    {
      "epoch": 0.20476343444541326,
      "grad_norm": 0.49715420603752136,
      "learning_rate": 9.940316323966257e-06,
      "loss": 0.3912,
      "step": 4462
    },
    {
      "epoch": 0.2048093249506677,
      "grad_norm": 0.4934089183807373,
      "learning_rate": 9.94027854716167e-06,
      "loss": 0.4619,
      "step": 4463
    },
    {
      "epoch": 0.20485521545592217,
      "grad_norm": 0.4589710831642151,
      "learning_rate": 9.940240758477273e-06,
      "loss": 0.424,
      "step": 4464
    },
    {
      "epoch": 0.20490110596117664,
      "grad_norm": 0.43391865491867065,
      "learning_rate": 9.940202957913155e-06,
      "loss": 0.33,
      "step": 4465
    },
    {
      "epoch": 0.2049469964664311,
      "grad_norm": 0.45424923300743103,
      "learning_rate": 9.940165145469409e-06,
      "loss": 0.378,
      "step": 4466
    },
    {
      "epoch": 0.20499288697168555,
      "grad_norm": 0.4214986562728882,
      "learning_rate": 9.940127321146125e-06,
      "loss": 0.3123,
      "step": 4467
    },
    {
      "epoch": 0.20503877747694002,
      "grad_norm": 0.47325822710990906,
      "learning_rate": 9.940089484943394e-06,
      "loss": 0.4446,
      "step": 4468
    },
    {
      "epoch": 0.2050846679821945,
      "grad_norm": 0.4586750268936157,
      "learning_rate": 9.940051636861306e-06,
      "loss": 0.4123,
      "step": 4469
    },
    {
      "epoch": 0.20513055848744893,
      "grad_norm": 0.46702393889427185,
      "learning_rate": 9.940013776899954e-06,
      "loss": 0.4035,
      "step": 4470
    },
    {
      "epoch": 0.2051764489927034,
      "grad_norm": 0.4892576336860657,
      "learning_rate": 9.939975905059427e-06,
      "loss": 0.4395,
      "step": 4471
    },
    {
      "epoch": 0.20522233949795787,
      "grad_norm": 0.4993913769721985,
      "learning_rate": 9.939938021339818e-06,
      "loss": 0.4919,
      "step": 4472
    },
    {
      "epoch": 0.20526823000321234,
      "grad_norm": 0.46500056982040405,
      "learning_rate": 9.939900125741217e-06,
      "loss": 0.3916,
      "step": 4473
    },
    {
      "epoch": 0.20531412050846679,
      "grad_norm": 0.540874183177948,
      "learning_rate": 9.939862218263715e-06,
      "loss": 0.5311,
      "step": 4474
    },
    {
      "epoch": 0.20536001101372126,
      "grad_norm": 0.4383721947669983,
      "learning_rate": 9.939824298907403e-06,
      "loss": 0.3548,
      "step": 4475
    },
    {
      "epoch": 0.20540590151897573,
      "grad_norm": 0.4833926558494568,
      "learning_rate": 9.939786367672373e-06,
      "loss": 0.4557,
      "step": 4476
    },
    {
      "epoch": 0.2054517920242302,
      "grad_norm": 0.47873058915138245,
      "learning_rate": 9.939748424558716e-06,
      "loss": 0.3945,
      "step": 4477
    },
    {
      "epoch": 0.20549768252948464,
      "grad_norm": 0.4713362157344818,
      "learning_rate": 9.939710469566524e-06,
      "loss": 0.4332,
      "step": 4478
    },
    {
      "epoch": 0.2055435730347391,
      "grad_norm": 0.4947201609611511,
      "learning_rate": 9.939672502695884e-06,
      "loss": 0.4073,
      "step": 4479
    },
    {
      "epoch": 0.20558946353999358,
      "grad_norm": 0.4919685125350952,
      "learning_rate": 9.939634523946895e-06,
      "loss": 0.4273,
      "step": 4480
    },
    {
      "epoch": 0.20563535404524805,
      "grad_norm": 0.4664944112300873,
      "learning_rate": 9.939596533319643e-06,
      "loss": 0.3558,
      "step": 4481
    },
    {
      "epoch": 0.2056812445505025,
      "grad_norm": 0.5513607263565063,
      "learning_rate": 9.939558530814217e-06,
      "loss": 0.4335,
      "step": 4482
    },
    {
      "epoch": 0.20572713505575696,
      "grad_norm": 0.49874311685562134,
      "learning_rate": 9.939520516430714e-06,
      "loss": 0.4148,
      "step": 4483
    },
    {
      "epoch": 0.20577302556101143,
      "grad_norm": 0.4702422618865967,
      "learning_rate": 9.939482490169223e-06,
      "loss": 0.4007,
      "step": 4484
    },
    {
      "epoch": 0.2058189160662659,
      "grad_norm": 0.48694729804992676,
      "learning_rate": 9.939444452029836e-06,
      "loss": 0.4521,
      "step": 4485
    },
    {
      "epoch": 0.20586480657152034,
      "grad_norm": 0.48092415928840637,
      "learning_rate": 9.939406402012643e-06,
      "loss": 0.4273,
      "step": 4486
    },
    {
      "epoch": 0.2059106970767748,
      "grad_norm": 0.49558618664741516,
      "learning_rate": 9.93936834011774e-06,
      "loss": 0.4962,
      "step": 4487
    },
    {
      "epoch": 0.20595658758202928,
      "grad_norm": 0.4936516284942627,
      "learning_rate": 9.939330266345211e-06,
      "loss": 0.5067,
      "step": 4488
    },
    {
      "epoch": 0.20600247808728375,
      "grad_norm": 0.46066442131996155,
      "learning_rate": 9.939292180695153e-06,
      "loss": 0.4153,
      "step": 4489
    },
    {
      "epoch": 0.2060483685925382,
      "grad_norm": 0.45661619305610657,
      "learning_rate": 9.939254083167655e-06,
      "loss": 0.3974,
      "step": 4490
    },
    {
      "epoch": 0.20609425909779266,
      "grad_norm": 0.44079911708831787,
      "learning_rate": 9.93921597376281e-06,
      "loss": 0.3302,
      "step": 4491
    },
    {
      "epoch": 0.20614014960304713,
      "grad_norm": 0.46443063020706177,
      "learning_rate": 9.939177852480713e-06,
      "loss": 0.3847,
      "step": 4492
    },
    {
      "epoch": 0.2061860401083016,
      "grad_norm": 0.46124452352523804,
      "learning_rate": 9.939139719321449e-06,
      "loss": 0.3719,
      "step": 4493
    },
    {
      "epoch": 0.20623193061355605,
      "grad_norm": 0.43286919593811035,
      "learning_rate": 9.939101574285113e-06,
      "loss": 0.3668,
      "step": 4494
    },
    {
      "epoch": 0.20627782111881052,
      "grad_norm": 0.46169283986091614,
      "learning_rate": 9.939063417371797e-06,
      "loss": 0.3729,
      "step": 4495
    },
    {
      "epoch": 0.206323711624065,
      "grad_norm": 0.45073485374450684,
      "learning_rate": 9.93902524858159e-06,
      "loss": 0.3499,
      "step": 4496
    },
    {
      "epoch": 0.20636960212931946,
      "grad_norm": 0.5126428008079529,
      "learning_rate": 9.938987067914589e-06,
      "loss": 0.4592,
      "step": 4497
    },
    {
      "epoch": 0.2064154926345739,
      "grad_norm": 0.49929559230804443,
      "learning_rate": 9.938948875370882e-06,
      "loss": 0.5132,
      "step": 4498
    },
    {
      "epoch": 0.20646138313982837,
      "grad_norm": 0.4436069130897522,
      "learning_rate": 9.938910670950562e-06,
      "loss": 0.4044,
      "step": 4499
    },
    {
      "epoch": 0.20650727364508284,
      "grad_norm": 0.4961375892162323,
      "learning_rate": 9.93887245465372e-06,
      "loss": 0.4579,
      "step": 4500
    },
    {
      "epoch": 0.20655316415033728,
      "grad_norm": 0.5659322738647461,
      "learning_rate": 9.938834226480449e-06,
      "loss": 0.4344,
      "step": 4501
    },
    {
      "epoch": 0.20659905465559175,
      "grad_norm": 0.5714291930198669,
      "learning_rate": 9.93879598643084e-06,
      "loss": 0.435,
      "step": 4502
    },
    {
      "epoch": 0.20664494516084622,
      "grad_norm": 0.5349580645561218,
      "learning_rate": 9.938757734504987e-06,
      "loss": 0.5182,
      "step": 4503
    },
    {
      "epoch": 0.2066908356661007,
      "grad_norm": 0.4456104338169098,
      "learning_rate": 9.938719470702977e-06,
      "loss": 0.3345,
      "step": 4504
    },
    {
      "epoch": 0.20673672617135513,
      "grad_norm": 0.48803824186325073,
      "learning_rate": 9.938681195024908e-06,
      "loss": 0.4485,
      "step": 4505
    },
    {
      "epoch": 0.2067826166766096,
      "grad_norm": 0.4779717028141022,
      "learning_rate": 9.93864290747087e-06,
      "loss": 0.4255,
      "step": 4506
    },
    {
      "epoch": 0.20682850718186407,
      "grad_norm": 0.5309612154960632,
      "learning_rate": 9.938604608040952e-06,
      "loss": 0.4291,
      "step": 4507
    },
    {
      "epoch": 0.20687439768711854,
      "grad_norm": 0.4602609872817993,
      "learning_rate": 9.93856629673525e-06,
      "loss": 0.4199,
      "step": 4508
    },
    {
      "epoch": 0.206920288192373,
      "grad_norm": 0.4964911639690399,
      "learning_rate": 9.938527973553856e-06,
      "loss": 0.4335,
      "step": 4509
    },
    {
      "epoch": 0.20696617869762746,
      "grad_norm": 0.47558078169822693,
      "learning_rate": 9.938489638496859e-06,
      "loss": 0.3687,
      "step": 4510
    },
    {
      "epoch": 0.20701206920288193,
      "grad_norm": 0.5527694225311279,
      "learning_rate": 9.938451291564352e-06,
      "loss": 0.4992,
      "step": 4511
    },
    {
      "epoch": 0.2070579597081364,
      "grad_norm": 0.4722549617290497,
      "learning_rate": 9.938412932756431e-06,
      "loss": 0.3937,
      "step": 4512
    },
    {
      "epoch": 0.20710385021339084,
      "grad_norm": 0.4794664680957794,
      "learning_rate": 9.938374562073185e-06,
      "loss": 0.3461,
      "step": 4513
    },
    {
      "epoch": 0.2071497407186453,
      "grad_norm": 0.5057462453842163,
      "learning_rate": 9.938336179514704e-06,
      "loss": 0.4597,
      "step": 4514
    },
    {
      "epoch": 0.20719563122389978,
      "grad_norm": 0.483297735452652,
      "learning_rate": 9.938297785081086e-06,
      "loss": 0.4136,
      "step": 4515
    },
    {
      "epoch": 0.20724152172915425,
      "grad_norm": 0.43480783700942993,
      "learning_rate": 9.93825937877242e-06,
      "loss": 0.3352,
      "step": 4516
    },
    {
      "epoch": 0.2072874122344087,
      "grad_norm": 0.48694080114364624,
      "learning_rate": 9.938220960588799e-06,
      "loss": 0.4332,
      "step": 4517
    },
    {
      "epoch": 0.20733330273966316,
      "grad_norm": 0.4349701404571533,
      "learning_rate": 9.938182530530313e-06,
      "loss": 0.3012,
      "step": 4518
    },
    {
      "epoch": 0.20737919324491763,
      "grad_norm": 0.462764710187912,
      "learning_rate": 9.938144088597059e-06,
      "loss": 0.3018,
      "step": 4519
    },
    {
      "epoch": 0.2074250837501721,
      "grad_norm": 0.5145973563194275,
      "learning_rate": 9.938105634789126e-06,
      "loss": 0.5471,
      "step": 4520
    },
    {
      "epoch": 0.20747097425542654,
      "grad_norm": 0.4801902174949646,
      "learning_rate": 9.938067169106606e-06,
      "loss": 0.3622,
      "step": 4521
    },
    {
      "epoch": 0.207516864760681,
      "grad_norm": 0.41820165514945984,
      "learning_rate": 9.938028691549594e-06,
      "loss": 0.2819,
      "step": 4522
    },
    {
      "epoch": 0.20756275526593548,
      "grad_norm": 0.5365082025527954,
      "learning_rate": 9.937990202118183e-06,
      "loss": 0.412,
      "step": 4523
    },
    {
      "epoch": 0.20760864577118995,
      "grad_norm": 0.5280576944351196,
      "learning_rate": 9.937951700812463e-06,
      "loss": 0.5134,
      "step": 4524
    },
    {
      "epoch": 0.2076545362764444,
      "grad_norm": 0.4733649790287018,
      "learning_rate": 9.937913187632528e-06,
      "loss": 0.4668,
      "step": 4525
    },
    {
      "epoch": 0.20770042678169887,
      "grad_norm": 0.48566192388534546,
      "learning_rate": 9.93787466257847e-06,
      "loss": 0.4427,
      "step": 4526
    },
    {
      "epoch": 0.20774631728695334,
      "grad_norm": 0.46069398522377014,
      "learning_rate": 9.937836125650382e-06,
      "loss": 0.4217,
      "step": 4527
    },
    {
      "epoch": 0.2077922077922078,
      "grad_norm": 0.4433155357837677,
      "learning_rate": 9.937797576848356e-06,
      "loss": 0.3643,
      "step": 4528
    },
    {
      "epoch": 0.20783809829746225,
      "grad_norm": 0.5018631815910339,
      "learning_rate": 9.937759016172486e-06,
      "loss": 0.4516,
      "step": 4529
    },
    {
      "epoch": 0.20788398880271672,
      "grad_norm": 0.4642772972583771,
      "learning_rate": 9.937720443622862e-06,
      "loss": 0.3228,
      "step": 4530
    },
    {
      "epoch": 0.2079298793079712,
      "grad_norm": 0.43649399280548096,
      "learning_rate": 9.93768185919958e-06,
      "loss": 0.3555,
      "step": 4531
    },
    {
      "epoch": 0.20797576981322563,
      "grad_norm": 0.9431378245353699,
      "learning_rate": 9.937643262902732e-06,
      "loss": 0.4603,
      "step": 4532
    },
    {
      "epoch": 0.2080216603184801,
      "grad_norm": 0.47615882754325867,
      "learning_rate": 9.93760465473241e-06,
      "loss": 0.3854,
      "step": 4533
    },
    {
      "epoch": 0.20806755082373457,
      "grad_norm": 0.48449355363845825,
      "learning_rate": 9.937566034688708e-06,
      "loss": 0.446,
      "step": 4534
    },
    {
      "epoch": 0.20811344132898904,
      "grad_norm": 0.4414539039134979,
      "learning_rate": 9.937527402771717e-06,
      "loss": 0.3785,
      "step": 4535
    },
    {
      "epoch": 0.20815933183424348,
      "grad_norm": 0.48071879148483276,
      "learning_rate": 9.937488758981531e-06,
      "loss": 0.3833,
      "step": 4536
    },
    {
      "epoch": 0.20820522233949795,
      "grad_norm": 0.4578094482421875,
      "learning_rate": 9.937450103318242e-06,
      "loss": 0.3959,
      "step": 4537
    },
    {
      "epoch": 0.20825111284475242,
      "grad_norm": 0.4642413556575775,
      "learning_rate": 9.937411435781945e-06,
      "loss": 0.4447,
      "step": 4538
    },
    {
      "epoch": 0.2082970033500069,
      "grad_norm": 0.4551283121109009,
      "learning_rate": 9.937372756372732e-06,
      "loss": 0.3615,
      "step": 4539
    },
    {
      "epoch": 0.20834289385526134,
      "grad_norm": 0.4219685196876526,
      "learning_rate": 9.937334065090694e-06,
      "loss": 0.356,
      "step": 4540
    },
    {
      "epoch": 0.2083887843605158,
      "grad_norm": 0.47293487191200256,
      "learning_rate": 9.937295361935928e-06,
      "loss": 0.4564,
      "step": 4541
    },
    {
      "epoch": 0.20843467486577028,
      "grad_norm": 0.4208858907222748,
      "learning_rate": 9.937256646908524e-06,
      "loss": 0.3348,
      "step": 4542
    },
    {
      "epoch": 0.20848056537102475,
      "grad_norm": 0.4567595422267914,
      "learning_rate": 9.937217920008576e-06,
      "loss": 0.3791,
      "step": 4543
    },
    {
      "epoch": 0.2085264558762792,
      "grad_norm": 0.3732040524482727,
      "learning_rate": 9.937179181236177e-06,
      "loss": 0.2618,
      "step": 4544
    },
    {
      "epoch": 0.20857234638153366,
      "grad_norm": 0.4999716281890869,
      "learning_rate": 9.937140430591421e-06,
      "loss": 0.4753,
      "step": 4545
    },
    {
      "epoch": 0.20861823688678813,
      "grad_norm": 0.45168647170066833,
      "learning_rate": 9.937101668074399e-06,
      "loss": 0.4372,
      "step": 4546
    },
    {
      "epoch": 0.2086641273920426,
      "grad_norm": 0.476558119058609,
      "learning_rate": 9.937062893685207e-06,
      "loss": 0.4153,
      "step": 4547
    },
    {
      "epoch": 0.20871001789729704,
      "grad_norm": 0.4637129008769989,
      "learning_rate": 9.937024107423936e-06,
      "loss": 0.4591,
      "step": 4548
    },
    {
      "epoch": 0.2087559084025515,
      "grad_norm": 0.45690834522247314,
      "learning_rate": 9.936985309290681e-06,
      "loss": 0.4383,
      "step": 4549
    },
    {
      "epoch": 0.20880179890780598,
      "grad_norm": 0.46199309825897217,
      "learning_rate": 9.936946499285535e-06,
      "loss": 0.4008,
      "step": 4550
    },
    {
      "epoch": 0.20884768941306045,
      "grad_norm": 0.45231765508651733,
      "learning_rate": 9.93690767740859e-06,
      "loss": 0.377,
      "step": 4551
    },
    {
      "epoch": 0.2088935799183149,
      "grad_norm": 0.5034369230270386,
      "learning_rate": 9.936868843659939e-06,
      "loss": 0.4892,
      "step": 4552
    },
    {
      "epoch": 0.20893947042356936,
      "grad_norm": 0.46708884835243225,
      "learning_rate": 9.93682999803968e-06,
      "loss": 0.3877,
      "step": 4553
    },
    {
      "epoch": 0.20898536092882383,
      "grad_norm": 0.46299293637275696,
      "learning_rate": 9.936791140547899e-06,
      "loss": 0.3464,
      "step": 4554
    },
    {
      "epoch": 0.2090312514340783,
      "grad_norm": 0.4468526244163513,
      "learning_rate": 9.936752271184696e-06,
      "loss": 0.3677,
      "step": 4555
    },
    {
      "epoch": 0.20907714193933274,
      "grad_norm": 0.44090819358825684,
      "learning_rate": 9.93671338995016e-06,
      "loss": 0.3098,
      "step": 4556
    },
    {
      "epoch": 0.20912303244458721,
      "grad_norm": 0.4699522852897644,
      "learning_rate": 9.936674496844387e-06,
      "loss": 0.376,
      "step": 4557
    },
    {
      "epoch": 0.20916892294984168,
      "grad_norm": 0.53037428855896,
      "learning_rate": 9.936635591867472e-06,
      "loss": 0.4554,
      "step": 4558
    },
    {
      "epoch": 0.20921481345509613,
      "grad_norm": 0.4777945578098297,
      "learning_rate": 9.936596675019504e-06,
      "loss": 0.4197,
      "step": 4559
    },
    {
      "epoch": 0.2092607039603506,
      "grad_norm": 0.45397379994392395,
      "learning_rate": 9.936557746300579e-06,
      "loss": 0.4053,
      "step": 4560
    },
    {
      "epoch": 0.20930659446560507,
      "grad_norm": 0.492824524641037,
      "learning_rate": 9.936518805710792e-06,
      "loss": 0.4595,
      "step": 4561
    },
    {
      "epoch": 0.20935248497085954,
      "grad_norm": 0.47252020239830017,
      "learning_rate": 9.936479853250235e-06,
      "loss": 0.4659,
      "step": 4562
    },
    {
      "epoch": 0.20939837547611398,
      "grad_norm": 0.47343355417251587,
      "learning_rate": 9.936440888919e-06,
      "loss": 0.4794,
      "step": 4563
    },
    {
      "epoch": 0.20944426598136845,
      "grad_norm": 0.4707490801811218,
      "learning_rate": 9.936401912717186e-06,
      "loss": 0.3712,
      "step": 4564
    },
    {
      "epoch": 0.20949015648662292,
      "grad_norm": 0.45677340030670166,
      "learning_rate": 9.936362924644882e-06,
      "loss": 0.3616,
      "step": 4565
    },
    {
      "epoch": 0.2095360469918774,
      "grad_norm": 0.48813164234161377,
      "learning_rate": 9.936323924702182e-06,
      "loss": 0.4205,
      "step": 4566
    },
    {
      "epoch": 0.20958193749713183,
      "grad_norm": 0.45545223355293274,
      "learning_rate": 9.93628491288918e-06,
      "loss": 0.3896,
      "step": 4567
    },
    {
      "epoch": 0.2096278280023863,
      "grad_norm": 0.44278645515441895,
      "learning_rate": 9.936245889205974e-06,
      "loss": 0.3779,
      "step": 4568
    },
    {
      "epoch": 0.20967371850764077,
      "grad_norm": 0.4800693392753601,
      "learning_rate": 9.936206853652652e-06,
      "loss": 0.4588,
      "step": 4569
    },
    {
      "epoch": 0.20971960901289524,
      "grad_norm": 0.43922656774520874,
      "learning_rate": 9.936167806229313e-06,
      "loss": 0.3885,
      "step": 4570
    },
    {
      "epoch": 0.20976549951814968,
      "grad_norm": 0.4473532438278198,
      "learning_rate": 9.936128746936046e-06,
      "loss": 0.3586,
      "step": 4571
    },
    {
      "epoch": 0.20981139002340415,
      "grad_norm": 0.5110666155815125,
      "learning_rate": 9.93608967577295e-06,
      "loss": 0.4732,
      "step": 4572
    },
    {
      "epoch": 0.20985728052865862,
      "grad_norm": 0.4744645357131958,
      "learning_rate": 9.936050592740112e-06,
      "loss": 0.4136,
      "step": 4573
    },
    {
      "epoch": 0.2099031710339131,
      "grad_norm": 0.44420140981674194,
      "learning_rate": 9.936011497837632e-06,
      "loss": 0.321,
      "step": 4574
    },
    {
      "epoch": 0.20994906153916754,
      "grad_norm": 0.4751008450984955,
      "learning_rate": 9.935972391065602e-06,
      "loss": 0.406,
      "step": 4575
    },
    {
      "epoch": 0.209994952044422,
      "grad_norm": 0.4755615293979645,
      "learning_rate": 9.935933272424118e-06,
      "loss": 0.4121,
      "step": 4576
    },
    {
      "epoch": 0.21004084254967648,
      "grad_norm": 0.5103150606155396,
      "learning_rate": 9.935894141913272e-06,
      "loss": 0.4388,
      "step": 4577
    },
    {
      "epoch": 0.21008673305493095,
      "grad_norm": 0.46369847655296326,
      "learning_rate": 9.935854999533157e-06,
      "loss": 0.4117,
      "step": 4578
    },
    {
      "epoch": 0.2101326235601854,
      "grad_norm": 0.4254971444606781,
      "learning_rate": 9.93581584528387e-06,
      "loss": 0.3513,
      "step": 4579
    },
    {
      "epoch": 0.21017851406543986,
      "grad_norm": 0.4889409840106964,
      "learning_rate": 9.935776679165501e-06,
      "loss": 0.4657,
      "step": 4580
    },
    {
      "epoch": 0.21022440457069433,
      "grad_norm": 0.5066528916358948,
      "learning_rate": 9.935737501178148e-06,
      "loss": 0.4835,
      "step": 4581
    },
    {
      "epoch": 0.2102702950759488,
      "grad_norm": 0.4572798013687134,
      "learning_rate": 9.935698311321906e-06,
      "loss": 0.3487,
      "step": 4582
    },
    {
      "epoch": 0.21031618558120324,
      "grad_norm": 0.4852089583873749,
      "learning_rate": 9.935659109596865e-06,
      "loss": 0.4318,
      "step": 4583
    },
    {
      "epoch": 0.2103620760864577,
      "grad_norm": 0.45804309844970703,
      "learning_rate": 9.935619896003124e-06,
      "loss": 0.418,
      "step": 4584
    },
    {
      "epoch": 0.21040796659171218,
      "grad_norm": 0.45562052726745605,
      "learning_rate": 9.935580670540772e-06,
      "loss": 0.4115,
      "step": 4585
    },
    {
      "epoch": 0.21045385709696665,
      "grad_norm": 0.4580683410167694,
      "learning_rate": 9.935541433209907e-06,
      "loss": 0.3851,
      "step": 4586
    },
    {
      "epoch": 0.2104997476022211,
      "grad_norm": 0.4790094494819641,
      "learning_rate": 9.935502184010624e-06,
      "loss": 0.4472,
      "step": 4587
    },
    {
      "epoch": 0.21054563810747556,
      "grad_norm": 0.46903327107429504,
      "learning_rate": 9.935462922943016e-06,
      "loss": 0.4229,
      "step": 4588
    },
    {
      "epoch": 0.21059152861273003,
      "grad_norm": 0.4421257972717285,
      "learning_rate": 9.935423650007176e-06,
      "loss": 0.3426,
      "step": 4589
    },
    {
      "epoch": 0.21063741911798448,
      "grad_norm": 0.4814123511314392,
      "learning_rate": 9.9353843652032e-06,
      "loss": 0.4254,
      "step": 4590
    },
    {
      "epoch": 0.21068330962323895,
      "grad_norm": 0.4628497064113617,
      "learning_rate": 9.935345068531182e-06,
      "loss": 0.3915,
      "step": 4591
    },
    {
      "epoch": 0.21072920012849342,
      "grad_norm": 0.5278543829917908,
      "learning_rate": 9.935305759991218e-06,
      "loss": 0.4866,
      "step": 4592
    },
    {
      "epoch": 0.21077509063374789,
      "grad_norm": 0.5155585408210754,
      "learning_rate": 9.9352664395834e-06,
      "loss": 0.516,
      "step": 4593
    },
    {
      "epoch": 0.21082098113900233,
      "grad_norm": 0.4477238655090332,
      "learning_rate": 9.935227107307824e-06,
      "loss": 0.4218,
      "step": 4594
    },
    {
      "epoch": 0.2108668716442568,
      "grad_norm": 0.4337116479873657,
      "learning_rate": 9.935187763164585e-06,
      "loss": 0.3589,
      "step": 4595
    },
    {
      "epoch": 0.21091276214951127,
      "grad_norm": 0.44227463006973267,
      "learning_rate": 9.935148407153776e-06,
      "loss": 0.3419,
      "step": 4596
    },
    {
      "epoch": 0.21095865265476574,
      "grad_norm": 0.49109065532684326,
      "learning_rate": 9.935109039275495e-06,
      "loss": 0.4522,
      "step": 4597
    },
    {
      "epoch": 0.21100454316002018,
      "grad_norm": 0.43904322385787964,
      "learning_rate": 9.935069659529832e-06,
      "loss": 0.4285,
      "step": 4598
    },
    {
      "epoch": 0.21105043366527465,
      "grad_norm": 0.49477484822273254,
      "learning_rate": 9.935030267916885e-06,
      "loss": 0.4747,
      "step": 4599
    },
    {
      "epoch": 0.21109632417052912,
      "grad_norm": 0.5089412927627563,
      "learning_rate": 9.934990864436748e-06,
      "loss": 0.5179,
      "step": 4600
    },
    {
      "epoch": 0.2111422146757836,
      "grad_norm": 0.44941091537475586,
      "learning_rate": 9.934951449089515e-06,
      "loss": 0.3687,
      "step": 4601
    },
    {
      "epoch": 0.21118810518103803,
      "grad_norm": 0.4486393630504608,
      "learning_rate": 9.93491202187528e-06,
      "loss": 0.3638,
      "step": 4602
    },
    {
      "epoch": 0.2112339956862925,
      "grad_norm": 0.45564234256744385,
      "learning_rate": 9.934872582794142e-06,
      "loss": 0.3335,
      "step": 4603
    },
    {
      "epoch": 0.21127988619154697,
      "grad_norm": 0.4690791964530945,
      "learning_rate": 9.934833131846191e-06,
      "loss": 0.3744,
      "step": 4604
    },
    {
      "epoch": 0.21132577669680144,
      "grad_norm": 0.5168153643608093,
      "learning_rate": 9.934793669031524e-06,
      "loss": 0.4792,
      "step": 4605
    },
    {
      "epoch": 0.21137166720205589,
      "grad_norm": 0.41905689239501953,
      "learning_rate": 9.934754194350235e-06,
      "loss": 0.3344,
      "step": 4606
    },
    {
      "epoch": 0.21141755770731036,
      "grad_norm": 0.5103574991226196,
      "learning_rate": 9.934714707802421e-06,
      "loss": 0.4452,
      "step": 4607
    },
    {
      "epoch": 0.21146344821256483,
      "grad_norm": 0.43421080708503723,
      "learning_rate": 9.934675209388174e-06,
      "loss": 0.2992,
      "step": 4608
    },
    {
      "epoch": 0.2115093387178193,
      "grad_norm": 0.4528566598892212,
      "learning_rate": 9.934635699107593e-06,
      "loss": 0.3869,
      "step": 4609
    },
    {
      "epoch": 0.21155522922307374,
      "grad_norm": 0.4664683938026428,
      "learning_rate": 9.934596176960769e-06,
      "loss": 0.3571,
      "step": 4610
    },
    {
      "epoch": 0.2116011197283282,
      "grad_norm": 0.4770258367061615,
      "learning_rate": 9.934556642947798e-06,
      "loss": 0.4233,
      "step": 4611
    },
    {
      "epoch": 0.21164701023358268,
      "grad_norm": 0.465563029050827,
      "learning_rate": 9.934517097068777e-06,
      "loss": 0.415,
      "step": 4612
    },
    {
      "epoch": 0.21169290073883715,
      "grad_norm": 0.4937037527561188,
      "learning_rate": 9.934477539323799e-06,
      "loss": 0.4737,
      "step": 4613
    },
    {
      "epoch": 0.2117387912440916,
      "grad_norm": 0.47488823533058167,
      "learning_rate": 9.93443796971296e-06,
      "loss": 0.4374,
      "step": 4614
    },
    {
      "epoch": 0.21178468174934606,
      "grad_norm": 0.512753427028656,
      "learning_rate": 9.934398388236354e-06,
      "loss": 0.4745,
      "step": 4615
    },
    {
      "epoch": 0.21183057225460053,
      "grad_norm": 0.4905538260936737,
      "learning_rate": 9.93435879489408e-06,
      "loss": 0.4449,
      "step": 4616
    },
    {
      "epoch": 0.21187646275985497,
      "grad_norm": 0.4729050397872925,
      "learning_rate": 9.934319189686228e-06,
      "loss": 0.3873,
      "step": 4617
    },
    {
      "epoch": 0.21192235326510944,
      "grad_norm": 0.4916042983531952,
      "learning_rate": 9.934279572612895e-06,
      "loss": 0.4925,
      "step": 4618
    },
    {
      "epoch": 0.2119682437703639,
      "grad_norm": 0.47090786695480347,
      "learning_rate": 9.93423994367418e-06,
      "loss": 0.3606,
      "step": 4619
    },
    {
      "epoch": 0.21201413427561838,
      "grad_norm": 0.45816871523857117,
      "learning_rate": 9.934200302870173e-06,
      "loss": 0.3507,
      "step": 4620
    },
    {
      "epoch": 0.21206002478087282,
      "grad_norm": 0.4796963632106781,
      "learning_rate": 9.934160650200972e-06,
      "loss": 0.4552,
      "step": 4621
    },
    {
      "epoch": 0.2121059152861273,
      "grad_norm": 0.42053475975990295,
      "learning_rate": 9.934120985666672e-06,
      "loss": 0.3078,
      "step": 4622
    },
    {
      "epoch": 0.21215180579138176,
      "grad_norm": 0.4671541452407837,
      "learning_rate": 9.934081309267369e-06,
      "loss": 0.3931,
      "step": 4623
    },
    {
      "epoch": 0.21219769629663623,
      "grad_norm": 0.48593783378601074,
      "learning_rate": 9.934041621003156e-06,
      "loss": 0.3703,
      "step": 4624
    },
    {
      "epoch": 0.21224358680189068,
      "grad_norm": 0.5138500332832336,
      "learning_rate": 9.934001920874132e-06,
      "loss": 0.5261,
      "step": 4625
    },
    {
      "epoch": 0.21228947730714515,
      "grad_norm": 0.5050431489944458,
      "learning_rate": 9.93396220888039e-06,
      "loss": 0.4324,
      "step": 4626
    },
    {
      "epoch": 0.21233536781239962,
      "grad_norm": 0.4766155481338501,
      "learning_rate": 9.933922485022025e-06,
      "loss": 0.4238,
      "step": 4627
    },
    {
      "epoch": 0.2123812583176541,
      "grad_norm": 0.467413991689682,
      "learning_rate": 9.933882749299134e-06,
      "loss": 0.3753,
      "step": 4628
    },
    {
      "epoch": 0.21242714882290853,
      "grad_norm": 0.4811027944087982,
      "learning_rate": 9.933843001711812e-06,
      "loss": 0.4998,
      "step": 4629
    },
    {
      "epoch": 0.212473039328163,
      "grad_norm": 0.44246906042099,
      "learning_rate": 9.933803242260155e-06,
      "loss": 0.3488,
      "step": 4630
    },
    {
      "epoch": 0.21251892983341747,
      "grad_norm": 0.4997427463531494,
      "learning_rate": 9.933763470944259e-06,
      "loss": 0.4519,
      "step": 4631
    },
    {
      "epoch": 0.21256482033867194,
      "grad_norm": 0.46644890308380127,
      "learning_rate": 9.933723687764218e-06,
      "loss": 0.3926,
      "step": 4632
    },
    {
      "epoch": 0.21261071084392638,
      "grad_norm": 0.48944222927093506,
      "learning_rate": 9.93368389272013e-06,
      "loss": 0.3765,
      "step": 4633
    },
    {
      "epoch": 0.21265660134918085,
      "grad_norm": 0.4626849591732025,
      "learning_rate": 9.93364408581209e-06,
      "loss": 0.3711,
      "step": 4634
    },
    {
      "epoch": 0.21270249185443532,
      "grad_norm": 0.6008056998252869,
      "learning_rate": 9.933604267040191e-06,
      "loss": 0.5543,
      "step": 4635
    },
    {
      "epoch": 0.2127483823596898,
      "grad_norm": 0.4939439296722412,
      "learning_rate": 9.933564436404531e-06,
      "loss": 0.5149,
      "step": 4636
    },
    {
      "epoch": 0.21279427286494423,
      "grad_norm": 0.4714864194393158,
      "learning_rate": 9.933524593905206e-06,
      "loss": 0.3824,
      "step": 4637
    },
    {
      "epoch": 0.2128401633701987,
      "grad_norm": 0.4945215880870819,
      "learning_rate": 9.933484739542312e-06,
      "loss": 0.4661,
      "step": 4638
    },
    {
      "epoch": 0.21288605387545317,
      "grad_norm": 0.47502630949020386,
      "learning_rate": 9.933444873315943e-06,
      "loss": 0.3845,
      "step": 4639
    },
    {
      "epoch": 0.21293194438070764,
      "grad_norm": 0.4932652413845062,
      "learning_rate": 9.933404995226198e-06,
      "loss": 0.3947,
      "step": 4640
    },
    {
      "epoch": 0.2129778348859621,
      "grad_norm": 0.5257059335708618,
      "learning_rate": 9.93336510527317e-06,
      "loss": 0.5189,
      "step": 4641
    },
    {
      "epoch": 0.21302372539121656,
      "grad_norm": 0.4927190840244293,
      "learning_rate": 9.933325203456957e-06,
      "loss": 0.4563,
      "step": 4642
    },
    {
      "epoch": 0.21306961589647103,
      "grad_norm": 0.4890120327472687,
      "learning_rate": 9.933285289777651e-06,
      "loss": 0.3684,
      "step": 4643
    },
    {
      "epoch": 0.2131155064017255,
      "grad_norm": 0.47681906819343567,
      "learning_rate": 9.933245364235353e-06,
      "loss": 0.4273,
      "step": 4644
    },
    {
      "epoch": 0.21316139690697994,
      "grad_norm": 0.4440349042415619,
      "learning_rate": 9.933205426830157e-06,
      "loss": 0.3612,
      "step": 4645
    },
    {
      "epoch": 0.2132072874122344,
      "grad_norm": 0.46425560116767883,
      "learning_rate": 9.933165477562159e-06,
      "loss": 0.3829,
      "step": 4646
    },
    {
      "epoch": 0.21325317791748888,
      "grad_norm": 0.42245426774024963,
      "learning_rate": 9.933125516431454e-06,
      "loss": 0.3467,
      "step": 4647
    },
    {
      "epoch": 0.21329906842274332,
      "grad_norm": 0.46113356947898865,
      "learning_rate": 9.933085543438138e-06,
      "loss": 0.3626,
      "step": 4648
    },
    {
      "epoch": 0.2133449589279978,
      "grad_norm": 0.4818299412727356,
      "learning_rate": 9.93304555858231e-06,
      "loss": 0.4237,
      "step": 4649
    },
    {
      "epoch": 0.21339084943325226,
      "grad_norm": 0.47459596395492554,
      "learning_rate": 9.933005561864064e-06,
      "loss": 0.4305,
      "step": 4650
    },
    {
      "epoch": 0.21343673993850673,
      "grad_norm": 0.49322208762168884,
      "learning_rate": 9.932965553283497e-06,
      "loss": 0.4904,
      "step": 4651
    },
    {
      "epoch": 0.21348263044376117,
      "grad_norm": 0.4633965492248535,
      "learning_rate": 9.932925532840705e-06,
      "loss": 0.433,
      "step": 4652
    },
    {
      "epoch": 0.21352852094901564,
      "grad_norm": 0.5234594345092773,
      "learning_rate": 9.932885500535782e-06,
      "loss": 0.4421,
      "step": 4653
    },
    {
      "epoch": 0.2135744114542701,
      "grad_norm": 0.4683118760585785,
      "learning_rate": 9.93284545636883e-06,
      "loss": 0.3587,
      "step": 4654
    },
    {
      "epoch": 0.21362030195952458,
      "grad_norm": 0.4599636495113373,
      "learning_rate": 9.932805400339937e-06,
      "loss": 0.3466,
      "step": 4655
    },
    {
      "epoch": 0.21366619246477903,
      "grad_norm": 0.44249051809310913,
      "learning_rate": 9.932765332449205e-06,
      "loss": 0.3813,
      "step": 4656
    },
    {
      "epoch": 0.2137120829700335,
      "grad_norm": 0.4925571382045746,
      "learning_rate": 9.93272525269673e-06,
      "loss": 0.4616,
      "step": 4657
    },
    {
      "epoch": 0.21375797347528797,
      "grad_norm": 0.49963822960853577,
      "learning_rate": 9.93268516108261e-06,
      "loss": 0.4401,
      "step": 4658
    },
    {
      "epoch": 0.21380386398054244,
      "grad_norm": 0.49291783571243286,
      "learning_rate": 9.932645057606936e-06,
      "loss": 0.4371,
      "step": 4659
    },
    {
      "epoch": 0.21384975448579688,
      "grad_norm": 0.4303255081176758,
      "learning_rate": 9.932604942269808e-06,
      "loss": 0.344,
      "step": 4660
    },
    {
      "epoch": 0.21389564499105135,
      "grad_norm": 0.5120735168457031,
      "learning_rate": 9.932564815071321e-06,
      "loss": 0.4922,
      "step": 4661
    },
    {
      "epoch": 0.21394153549630582,
      "grad_norm": 0.42673927545547485,
      "learning_rate": 9.932524676011575e-06,
      "loss": 0.3694,
      "step": 4662
    },
    {
      "epoch": 0.2139874260015603,
      "grad_norm": 0.4549776315689087,
      "learning_rate": 9.932484525090662e-06,
      "loss": 0.3945,
      "step": 4663
    },
    {
      "epoch": 0.21403331650681473,
      "grad_norm": 0.45083972811698914,
      "learning_rate": 9.932444362308679e-06,
      "loss": 0.3774,
      "step": 4664
    },
    {
      "epoch": 0.2140792070120692,
      "grad_norm": 0.46399959921836853,
      "learning_rate": 9.932404187665727e-06,
      "loss": 0.4022,
      "step": 4665
    },
    {
      "epoch": 0.21412509751732367,
      "grad_norm": 0.4504632353782654,
      "learning_rate": 9.932364001161899e-06,
      "loss": 0.3848,
      "step": 4666
    },
    {
      "epoch": 0.21417098802257814,
      "grad_norm": 0.48397886753082275,
      "learning_rate": 9.93232380279729e-06,
      "loss": 0.4029,
      "step": 4667
    },
    {
      "epoch": 0.21421687852783258,
      "grad_norm": 0.4804795980453491,
      "learning_rate": 9.932283592572e-06,
      "loss": 0.3998,
      "step": 4668
    },
    {
      "epoch": 0.21426276903308705,
      "grad_norm": 0.45309773087501526,
      "learning_rate": 9.932243370486127e-06,
      "loss": 0.4238,
      "step": 4669
    },
    {
      "epoch": 0.21430865953834152,
      "grad_norm": 0.5145482420921326,
      "learning_rate": 9.932203136539762e-06,
      "loss": 0.4925,
      "step": 4670
    },
    {
      "epoch": 0.214354550043596,
      "grad_norm": 0.48443666100502014,
      "learning_rate": 9.932162890733007e-06,
      "loss": 0.3905,
      "step": 4671
    },
    {
      "epoch": 0.21440044054885044,
      "grad_norm": 0.4662611484527588,
      "learning_rate": 9.932122633065956e-06,
      "loss": 0.3916,
      "step": 4672
    },
    {
      "epoch": 0.2144463310541049,
      "grad_norm": 0.4925994575023651,
      "learning_rate": 9.93208236353871e-06,
      "loss": 0.44,
      "step": 4673
    },
    {
      "epoch": 0.21449222155935938,
      "grad_norm": 0.5234211087226868,
      "learning_rate": 9.932042082151358e-06,
      "loss": 0.5185,
      "step": 4674
    },
    {
      "epoch": 0.21453811206461385,
      "grad_norm": 0.4390690326690674,
      "learning_rate": 9.932001788904004e-06,
      "loss": 0.3463,
      "step": 4675
    },
    {
      "epoch": 0.2145840025698683,
      "grad_norm": 0.47881796956062317,
      "learning_rate": 9.931961483796741e-06,
      "loss": 0.4358,
      "step": 4676
    },
    {
      "epoch": 0.21462989307512276,
      "grad_norm": 0.5072889924049377,
      "learning_rate": 9.931921166829668e-06,
      "loss": 0.4383,
      "step": 4677
    },
    {
      "epoch": 0.21467578358037723,
      "grad_norm": 0.48179998993873596,
      "learning_rate": 9.93188083800288e-06,
      "loss": 0.3891,
      "step": 4678
    },
    {
      "epoch": 0.21472167408563167,
      "grad_norm": 0.47286829352378845,
      "learning_rate": 9.931840497316475e-06,
      "loss": 0.4204,
      "step": 4679
    },
    {
      "epoch": 0.21476756459088614,
      "grad_norm": 0.44967126846313477,
      "learning_rate": 9.931800144770552e-06,
      "loss": 0.3522,
      "step": 4680
    },
    {
      "epoch": 0.2148134550961406,
      "grad_norm": 0.502183198928833,
      "learning_rate": 9.931759780365206e-06,
      "loss": 0.4726,
      "step": 4681
    },
    {
      "epoch": 0.21485934560139508,
      "grad_norm": 0.4539750814437866,
      "learning_rate": 9.931719404100532e-06,
      "loss": 0.4267,
      "step": 4682
    },
    {
      "epoch": 0.21490523610664952,
      "grad_norm": 0.47905218601226807,
      "learning_rate": 9.931679015976631e-06,
      "loss": 0.4516,
      "step": 4683
    },
    {
      "epoch": 0.214951126611904,
      "grad_norm": 0.47632932662963867,
      "learning_rate": 9.9316386159936e-06,
      "loss": 0.3734,
      "step": 4684
    },
    {
      "epoch": 0.21499701711715846,
      "grad_norm": 0.41714102029800415,
      "learning_rate": 9.931598204151533e-06,
      "loss": 0.3123,
      "step": 4685
    },
    {
      "epoch": 0.21504290762241293,
      "grad_norm": 0.4931911528110504,
      "learning_rate": 9.931557780450529e-06,
      "loss": 0.4126,
      "step": 4686
    },
    {
      "epoch": 0.21508879812766737,
      "grad_norm": 0.626720666885376,
      "learning_rate": 9.931517344890684e-06,
      "loss": 0.4795,
      "step": 4687
    },
    {
      "epoch": 0.21513468863292184,
      "grad_norm": 0.44990983605384827,
      "learning_rate": 9.931476897472096e-06,
      "loss": 0.367,
      "step": 4688
    },
    {
      "epoch": 0.21518057913817631,
      "grad_norm": 0.5085331797599792,
      "learning_rate": 9.931436438194863e-06,
      "loss": 0.4798,
      "step": 4689
    },
    {
      "epoch": 0.21522646964343078,
      "grad_norm": 0.4893498420715332,
      "learning_rate": 9.931395967059084e-06,
      "loss": 0.4724,
      "step": 4690
    },
    {
      "epoch": 0.21527236014868523,
      "grad_norm": 0.47430935502052307,
      "learning_rate": 9.93135548406485e-06,
      "loss": 0.4613,
      "step": 4691
    },
    {
      "epoch": 0.2153182506539397,
      "grad_norm": 0.44587236642837524,
      "learning_rate": 9.931314989212267e-06,
      "loss": 0.3774,
      "step": 4692
    },
    {
      "epoch": 0.21536414115919417,
      "grad_norm": 0.44930770993232727,
      "learning_rate": 9.931274482501424e-06,
      "loss": 0.3579,
      "step": 4693
    },
    {
      "epoch": 0.21541003166444864,
      "grad_norm": 0.4720441997051239,
      "learning_rate": 9.931233963932423e-06,
      "loss": 0.3846,
      "step": 4694
    },
    {
      "epoch": 0.21545592216970308,
      "grad_norm": 0.48503828048706055,
      "learning_rate": 9.931193433505362e-06,
      "loss": 0.4111,
      "step": 4695
    },
    {
      "epoch": 0.21550181267495755,
      "grad_norm": 0.5085614919662476,
      "learning_rate": 9.931152891220336e-06,
      "loss": 0.5087,
      "step": 4696
    },
    {
      "epoch": 0.21554770318021202,
      "grad_norm": 0.4879940152168274,
      "learning_rate": 9.931112337077444e-06,
      "loss": 0.4084,
      "step": 4697
    },
    {
      "epoch": 0.2155935936854665,
      "grad_norm": 0.48777949810028076,
      "learning_rate": 9.931071771076783e-06,
      "loss": 0.396,
      "step": 4698
    },
    {
      "epoch": 0.21563948419072093,
      "grad_norm": 0.4808366894721985,
      "learning_rate": 9.931031193218453e-06,
      "loss": 0.4375,
      "step": 4699
    },
    {
      "epoch": 0.2156853746959754,
      "grad_norm": 0.4755386412143707,
      "learning_rate": 9.930990603502547e-06,
      "loss": 0.4587,
      "step": 4700
    },
    {
      "epoch": 0.21573126520122987,
      "grad_norm": 0.4220426678657532,
      "learning_rate": 9.930950001929163e-06,
      "loss": 0.3619,
      "step": 4701
    },
    {
      "epoch": 0.21577715570648434,
      "grad_norm": 0.41686415672302246,
      "learning_rate": 9.930909388498403e-06,
      "loss": 0.3155,
      "step": 4702
    },
    {
      "epoch": 0.21582304621173878,
      "grad_norm": 0.46974438428878784,
      "learning_rate": 9.930868763210361e-06,
      "loss": 0.3552,
      "step": 4703
    },
    {
      "epoch": 0.21586893671699325,
      "grad_norm": 0.42856499552726746,
      "learning_rate": 9.930828126065138e-06,
      "loss": 0.3378,
      "step": 4704
    },
    {
      "epoch": 0.21591482722224772,
      "grad_norm": 0.44362497329711914,
      "learning_rate": 9.930787477062828e-06,
      "loss": 0.3648,
      "step": 4705
    },
    {
      "epoch": 0.21596071772750217,
      "grad_norm": 0.4723816215991974,
      "learning_rate": 9.930746816203529e-06,
      "loss": 0.3962,
      "step": 4706
    },
    {
      "epoch": 0.21600660823275664,
      "grad_norm": 0.4279741942882538,
      "learning_rate": 9.930706143487343e-06,
      "loss": 0.315,
      "step": 4707
    },
    {
      "epoch": 0.2160524987380111,
      "grad_norm": 0.504045307636261,
      "learning_rate": 9.930665458914362e-06,
      "loss": 0.4924,
      "step": 4708
    },
    {
      "epoch": 0.21609838924326558,
      "grad_norm": 0.45611247420310974,
      "learning_rate": 9.930624762484688e-06,
      "loss": 0.4316,
      "step": 4709
    },
    {
      "epoch": 0.21614427974852002,
      "grad_norm": 0.49936166405677795,
      "learning_rate": 9.930584054198419e-06,
      "loss": 0.4028,
      "step": 4710
    },
    {
      "epoch": 0.2161901702537745,
      "grad_norm": 0.46898287534713745,
      "learning_rate": 9.93054333405565e-06,
      "loss": 0.4093,
      "step": 4711
    },
    {
      "epoch": 0.21623606075902896,
      "grad_norm": 0.47403737902641296,
      "learning_rate": 9.93050260205648e-06,
      "loss": 0.3809,
      "step": 4712
    },
    {
      "epoch": 0.21628195126428343,
      "grad_norm": 0.5165886282920837,
      "learning_rate": 9.930461858201008e-06,
      "loss": 0.4791,
      "step": 4713
    },
    {
      "epoch": 0.21632784176953787,
      "grad_norm": 0.47071903944015503,
      "learning_rate": 9.930421102489331e-06,
      "loss": 0.4067,
      "step": 4714
    },
    {
      "epoch": 0.21637373227479234,
      "grad_norm": 0.4974483847618103,
      "learning_rate": 9.930380334921548e-06,
      "loss": 0.4711,
      "step": 4715
    },
    {
      "epoch": 0.2164196227800468,
      "grad_norm": 0.48442625999450684,
      "learning_rate": 9.930339555497756e-06,
      "loss": 0.4497,
      "step": 4716
    },
    {
      "epoch": 0.21646551328530128,
      "grad_norm": 0.4645082950592041,
      "learning_rate": 9.930298764218053e-06,
      "loss": 0.3787,
      "step": 4717
    },
    {
      "epoch": 0.21651140379055572,
      "grad_norm": 0.5022499561309814,
      "learning_rate": 9.930257961082538e-06,
      "loss": 0.4822,
      "step": 4718
    },
    {
      "epoch": 0.2165572942958102,
      "grad_norm": 0.4625333249568939,
      "learning_rate": 9.93021714609131e-06,
      "loss": 0.4244,
      "step": 4719
    },
    {
      "epoch": 0.21660318480106466,
      "grad_norm": 0.45841649174690247,
      "learning_rate": 9.930176319244464e-06,
      "loss": 0.3745,
      "step": 4720
    },
    {
      "epoch": 0.21664907530631913,
      "grad_norm": 0.4419494867324829,
      "learning_rate": 9.9301354805421e-06,
      "loss": 0.3729,
      "step": 4721
    },
    {
      "epoch": 0.21669496581157358,
      "grad_norm": 0.4620323181152344,
      "learning_rate": 9.930094629984317e-06,
      "loss": 0.3547,
      "step": 4722
    },
    {
      "epoch": 0.21674085631682805,
      "grad_norm": 0.49037355184555054,
      "learning_rate": 9.930053767571212e-06,
      "loss": 0.3814,
      "step": 4723
    },
    {
      "epoch": 0.21678674682208252,
      "grad_norm": 0.44606703519821167,
      "learning_rate": 9.930012893302884e-06,
      "loss": 0.36,
      "step": 4724
    },
    {
      "epoch": 0.21683263732733699,
      "grad_norm": 0.4860285222530365,
      "learning_rate": 9.929972007179429e-06,
      "loss": 0.448,
      "step": 4725
    },
    {
      "epoch": 0.21687852783259143,
      "grad_norm": 0.4884079694747925,
      "learning_rate": 9.92993110920095e-06,
      "loss": 0.4032,
      "step": 4726
    },
    {
      "epoch": 0.2169244183378459,
      "grad_norm": 0.4397047460079193,
      "learning_rate": 9.92989019936754e-06,
      "loss": 0.3398,
      "step": 4727
    },
    {
      "epoch": 0.21697030884310037,
      "grad_norm": 0.4717492461204529,
      "learning_rate": 9.929849277679303e-06,
      "loss": 0.4437,
      "step": 4728
    },
    {
      "epoch": 0.21701619934835484,
      "grad_norm": 0.45874541997909546,
      "learning_rate": 9.929808344136332e-06,
      "loss": 0.4141,
      "step": 4729
    },
    {
      "epoch": 0.21706208985360928,
      "grad_norm": 0.44129058718681335,
      "learning_rate": 9.929767398738728e-06,
      "loss": 0.3837,
      "step": 4730
    },
    {
      "epoch": 0.21710798035886375,
      "grad_norm": 0.5274844169616699,
      "learning_rate": 9.92972644148659e-06,
      "loss": 0.5665,
      "step": 4731
    },
    {
      "epoch": 0.21715387086411822,
      "grad_norm": 0.47592389583587646,
      "learning_rate": 9.929685472380015e-06,
      "loss": 0.4163,
      "step": 4732
    },
    {
      "epoch": 0.2171997613693727,
      "grad_norm": 0.4594455361366272,
      "learning_rate": 9.929644491419103e-06,
      "loss": 0.3405,
      "step": 4733
    },
    {
      "epoch": 0.21724565187462713,
      "grad_norm": 0.4659575819969177,
      "learning_rate": 9.929603498603953e-06,
      "loss": 0.3718,
      "step": 4734
    },
    {
      "epoch": 0.2172915423798816,
      "grad_norm": 0.47570160031318665,
      "learning_rate": 9.929562493934661e-06,
      "loss": 0.3864,
      "step": 4735
    },
    {
      "epoch": 0.21733743288513607,
      "grad_norm": 0.423115611076355,
      "learning_rate": 9.929521477411326e-06,
      "loss": 0.328,
      "step": 4736
    },
    {
      "epoch": 0.21738332339039051,
      "grad_norm": 0.46982258558273315,
      "learning_rate": 9.92948044903405e-06,
      "loss": 0.4263,
      "step": 4737
    },
    {
      "epoch": 0.21742921389564498,
      "grad_norm": 0.5065194368362427,
      "learning_rate": 9.929439408802927e-06,
      "loss": 0.4478,
      "step": 4738
    },
    {
      "epoch": 0.21747510440089945,
      "grad_norm": 0.5010607838630676,
      "learning_rate": 9.929398356718061e-06,
      "loss": 0.5078,
      "step": 4739
    },
    {
      "epoch": 0.21752099490615392,
      "grad_norm": 0.45388680696487427,
      "learning_rate": 9.929357292779545e-06,
      "loss": 0.408,
      "step": 4740
    },
    {
      "epoch": 0.21756688541140837,
      "grad_norm": 0.4765114188194275,
      "learning_rate": 9.929316216987482e-06,
      "loss": 0.3863,
      "step": 4741
    },
    {
      "epoch": 0.21761277591666284,
      "grad_norm": 0.46761444211006165,
      "learning_rate": 9.929275129341968e-06,
      "loss": 0.355,
      "step": 4742
    },
    {
      "epoch": 0.2176586664219173,
      "grad_norm": 0.49700087308883667,
      "learning_rate": 9.929234029843103e-06,
      "loss": 0.3955,
      "step": 4743
    },
    {
      "epoch": 0.21770455692717178,
      "grad_norm": 0.43094325065612793,
      "learning_rate": 9.929192918490986e-06,
      "loss": 0.3257,
      "step": 4744
    },
    {
      "epoch": 0.21775044743242622,
      "grad_norm": 0.48548972606658936,
      "learning_rate": 9.929151795285718e-06,
      "loss": 0.4836,
      "step": 4745
    },
    {
      "epoch": 0.2177963379376807,
      "grad_norm": 0.45151907205581665,
      "learning_rate": 9.929110660227392e-06,
      "loss": 0.3847,
      "step": 4746
    },
    {
      "epoch": 0.21784222844293516,
      "grad_norm": 0.4795612096786499,
      "learning_rate": 9.929069513316115e-06,
      "loss": 0.4237,
      "step": 4747
    },
    {
      "epoch": 0.21788811894818963,
      "grad_norm": 0.465311199426651,
      "learning_rate": 9.929028354551977e-06,
      "loss": 0.4508,
      "step": 4748
    },
    {
      "epoch": 0.21793400945344407,
      "grad_norm": 0.44016215205192566,
      "learning_rate": 9.928987183935084e-06,
      "loss": 0.3319,
      "step": 4749
    },
    {
      "epoch": 0.21797989995869854,
      "grad_norm": 0.49470531940460205,
      "learning_rate": 9.928946001465531e-06,
      "loss": 0.4994,
      "step": 4750
    },
    {
      "epoch": 0.218025790463953,
      "grad_norm": 0.5167783498764038,
      "learning_rate": 9.92890480714342e-06,
      "loss": 0.4838,
      "step": 4751
    },
    {
      "epoch": 0.21807168096920748,
      "grad_norm": 0.5040650963783264,
      "learning_rate": 9.928863600968847e-06,
      "loss": 0.4089,
      "step": 4752
    },
    {
      "epoch": 0.21811757147446192,
      "grad_norm": 0.43722257018089294,
      "learning_rate": 9.928822382941914e-06,
      "loss": 0.3165,
      "step": 4753
    },
    {
      "epoch": 0.2181634619797164,
      "grad_norm": 0.45732051134109497,
      "learning_rate": 9.928781153062718e-06,
      "loss": 0.3594,
      "step": 4754
    },
    {
      "epoch": 0.21820935248497086,
      "grad_norm": 0.4525343179702759,
      "learning_rate": 9.92873991133136e-06,
      "loss": 0.3525,
      "step": 4755
    },
    {
      "epoch": 0.21825524299022533,
      "grad_norm": 0.4378473460674286,
      "learning_rate": 9.928698657747936e-06,
      "loss": 0.3292,
      "step": 4756
    },
    {
      "epoch": 0.21830113349547978,
      "grad_norm": 0.5118082761764526,
      "learning_rate": 9.928657392312548e-06,
      "loss": 0.4048,
      "step": 4757
    },
    {
      "epoch": 0.21834702400073425,
      "grad_norm": 0.4829464852809906,
      "learning_rate": 9.928616115025296e-06,
      "loss": 0.3909,
      "step": 4758
    },
    {
      "epoch": 0.21839291450598872,
      "grad_norm": 0.5080723762512207,
      "learning_rate": 9.928574825886277e-06,
      "loss": 0.4956,
      "step": 4759
    },
    {
      "epoch": 0.2184388050112432,
      "grad_norm": 0.473085880279541,
      "learning_rate": 9.928533524895589e-06,
      "loss": 0.4321,
      "step": 4760
    },
    {
      "epoch": 0.21848469551649763,
      "grad_norm": 0.450904905796051,
      "learning_rate": 9.928492212053336e-06,
      "loss": 0.3879,
      "step": 4761
    },
    {
      "epoch": 0.2185305860217521,
      "grad_norm": 0.4535611569881439,
      "learning_rate": 9.928450887359613e-06,
      "loss": 0.3482,
      "step": 4762
    },
    {
      "epoch": 0.21857647652700657,
      "grad_norm": 0.5056604146957397,
      "learning_rate": 9.928409550814524e-06,
      "loss": 0.4549,
      "step": 4763
    },
    {
      "epoch": 0.21862236703226104,
      "grad_norm": 0.5012725591659546,
      "learning_rate": 9.92836820241816e-06,
      "loss": 0.4525,
      "step": 4764
    },
    {
      "epoch": 0.21866825753751548,
      "grad_norm": 0.5008786916732788,
      "learning_rate": 9.928326842170631e-06,
      "loss": 0.5025,
      "step": 4765
    },
    {
      "epoch": 0.21871414804276995,
      "grad_norm": 0.528501033782959,
      "learning_rate": 9.92828547007203e-06,
      "loss": 0.4971,
      "step": 4766
    },
    {
      "epoch": 0.21876003854802442,
      "grad_norm": 0.4721131920814514,
      "learning_rate": 9.928244086122457e-06,
      "loss": 0.4287,
      "step": 4767
    },
    {
      "epoch": 0.21880592905327886,
      "grad_norm": 0.4517940282821655,
      "learning_rate": 9.928202690322013e-06,
      "loss": 0.357,
      "step": 4768
    },
    {
      "epoch": 0.21885181955853333,
      "grad_norm": 0.4652464985847473,
      "learning_rate": 9.928161282670796e-06,
      "loss": 0.3721,
      "step": 4769
    },
    {
      "epoch": 0.2188977100637878,
      "grad_norm": 0.47898006439208984,
      "learning_rate": 9.928119863168907e-06,
      "loss": 0.4176,
      "step": 4770
    },
    {
      "epoch": 0.21894360056904227,
      "grad_norm": 0.4960164725780487,
      "learning_rate": 9.928078431816445e-06,
      "loss": 0.4252,
      "step": 4771
    },
    {
      "epoch": 0.21898949107429672,
      "grad_norm": 0.5123987793922424,
      "learning_rate": 9.928036988613512e-06,
      "loss": 0.4848,
      "step": 4772
    },
    {
      "epoch": 0.21903538157955119,
      "grad_norm": 0.5303327441215515,
      "learning_rate": 9.927995533560202e-06,
      "loss": 0.5233,
      "step": 4773
    },
    {
      "epoch": 0.21908127208480566,
      "grad_norm": 0.4559354782104492,
      "learning_rate": 9.92795406665662e-06,
      "loss": 0.3632,
      "step": 4774
    },
    {
      "epoch": 0.21912716259006013,
      "grad_norm": 0.4832862317562103,
      "learning_rate": 9.927912587902863e-06,
      "loss": 0.4975,
      "step": 4775
    },
    {
      "epoch": 0.21917305309531457,
      "grad_norm": 0.8387763500213623,
      "learning_rate": 9.927871097299033e-06,
      "loss": 0.4251,
      "step": 4776
    },
    {
      "epoch": 0.21921894360056904,
      "grad_norm": 0.45203593373298645,
      "learning_rate": 9.927829594845226e-06,
      "loss": 0.3637,
      "step": 4777
    },
    {
      "epoch": 0.2192648341058235,
      "grad_norm": 0.4715596139431,
      "learning_rate": 9.927788080541546e-06,
      "loss": 0.4197,
      "step": 4778
    },
    {
      "epoch": 0.21931072461107798,
      "grad_norm": 0.44730404019355774,
      "learning_rate": 9.927746554388091e-06,
      "loss": 0.3367,
      "step": 4779
    },
    {
      "epoch": 0.21935661511633242,
      "grad_norm": 0.4531867504119873,
      "learning_rate": 9.927705016384958e-06,
      "loss": 0.3688,
      "step": 4780
    },
    {
      "epoch": 0.2194025056215869,
      "grad_norm": 0.43707215785980225,
      "learning_rate": 9.927663466532253e-06,
      "loss": 0.3331,
      "step": 4781
    },
    {
      "epoch": 0.21944839612684136,
      "grad_norm": 0.4890100359916687,
      "learning_rate": 9.92762190483007e-06,
      "loss": 0.4518,
      "step": 4782
    },
    {
      "epoch": 0.21949428663209583,
      "grad_norm": 0.4831465482711792,
      "learning_rate": 9.927580331278512e-06,
      "loss": 0.4518,
      "step": 4783
    },
    {
      "epoch": 0.21954017713735027,
      "grad_norm": 0.46450120210647583,
      "learning_rate": 9.92753874587768e-06,
      "loss": 0.4191,
      "step": 4784
    },
    {
      "epoch": 0.21958606764260474,
      "grad_norm": 0.4956110119819641,
      "learning_rate": 9.927497148627671e-06,
      "loss": 0.4453,
      "step": 4785
    },
    {
      "epoch": 0.2196319581478592,
      "grad_norm": 0.45824500918388367,
      "learning_rate": 9.927455539528588e-06,
      "loss": 0.3729,
      "step": 4786
    },
    {
      "epoch": 0.21967784865311368,
      "grad_norm": 0.4803895354270935,
      "learning_rate": 9.92741391858053e-06,
      "loss": 0.4795,
      "step": 4787
    },
    {
      "epoch": 0.21972373915836813,
      "grad_norm": 0.4855837821960449,
      "learning_rate": 9.927372285783595e-06,
      "loss": 0.51,
      "step": 4788
    },
    {
      "epoch": 0.2197696296636226,
      "grad_norm": 0.41429024934768677,
      "learning_rate": 9.927330641137884e-06,
      "loss": 0.3029,
      "step": 4789
    },
    {
      "epoch": 0.21981552016887707,
      "grad_norm": 0.47005584836006165,
      "learning_rate": 9.9272889846435e-06,
      "loss": 0.4218,
      "step": 4790
    },
    {
      "epoch": 0.21986141067413154,
      "grad_norm": 0.4601205885410309,
      "learning_rate": 9.92724731630054e-06,
      "loss": 0.393,
      "step": 4791
    },
    {
      "epoch": 0.21990730117938598,
      "grad_norm": 0.4288758933544159,
      "learning_rate": 9.927205636109105e-06,
      "loss": 0.3036,
      "step": 4792
    },
    {
      "epoch": 0.21995319168464045,
      "grad_norm": 0.4515363574028015,
      "learning_rate": 9.927163944069295e-06,
      "loss": 0.3551,
      "step": 4793
    },
    {
      "epoch": 0.21999908218989492,
      "grad_norm": 0.4707154333591461,
      "learning_rate": 9.927122240181212e-06,
      "loss": 0.3927,
      "step": 4794
    },
    {
      "epoch": 0.22004497269514936,
      "grad_norm": 0.44049072265625,
      "learning_rate": 9.927080524444956e-06,
      "loss": 0.4036,
      "step": 4795
    },
    {
      "epoch": 0.22009086320040383,
      "grad_norm": 0.5246038436889648,
      "learning_rate": 9.927038796860623e-06,
      "loss": 0.4199,
      "step": 4796
    },
    {
      "epoch": 0.2201367537056583,
      "grad_norm": 0.45726901292800903,
      "learning_rate": 9.92699705742832e-06,
      "loss": 0.3726,
      "step": 4797
    },
    {
      "epoch": 0.22018264421091277,
      "grad_norm": 0.46941837668418884,
      "learning_rate": 9.926955306148145e-06,
      "loss": 0.3908,
      "step": 4798
    },
    {
      "epoch": 0.2202285347161672,
      "grad_norm": 0.46446794271469116,
      "learning_rate": 9.926913543020194e-06,
      "loss": 0.4257,
      "step": 4799
    },
    {
      "epoch": 0.22027442522142168,
      "grad_norm": 0.4293504059314728,
      "learning_rate": 9.926871768044572e-06,
      "loss": 0.3154,
      "step": 4800
    },
    {
      "epoch": 0.22032031572667615,
      "grad_norm": 0.4713653028011322,
      "learning_rate": 9.92682998122138e-06,
      "loss": 0.4542,
      "step": 4801
    },
    {
      "epoch": 0.22036620623193062,
      "grad_norm": 0.4862630069255829,
      "learning_rate": 9.926788182550715e-06,
      "loss": 0.4627,
      "step": 4802
    },
    {
      "epoch": 0.22041209673718506,
      "grad_norm": 0.44805315136909485,
      "learning_rate": 9.92674637203268e-06,
      "loss": 0.3545,
      "step": 4803
    },
    {
      "epoch": 0.22045798724243953,
      "grad_norm": 0.4568629860877991,
      "learning_rate": 9.926704549667374e-06,
      "loss": 0.4017,
      "step": 4804
    },
    {
      "epoch": 0.220503877747694,
      "grad_norm": 0.5366540551185608,
      "learning_rate": 9.9266627154549e-06,
      "loss": 0.495,
      "step": 4805
    },
    {
      "epoch": 0.22054976825294847,
      "grad_norm": 0.4854573607444763,
      "learning_rate": 9.926620869395356e-06,
      "loss": 0.4367,
      "step": 4806
    },
    {
      "epoch": 0.22059565875820292,
      "grad_norm": 0.49729904532432556,
      "learning_rate": 9.926579011488843e-06,
      "loss": 0.5154,
      "step": 4807
    },
    {
      "epoch": 0.2206415492634574,
      "grad_norm": 0.482592910528183,
      "learning_rate": 9.926537141735463e-06,
      "loss": 0.3894,
      "step": 4808
    },
    {
      "epoch": 0.22068743976871186,
      "grad_norm": 1.0421732664108276,
      "learning_rate": 9.926495260135318e-06,
      "loss": 0.4382,
      "step": 4809
    },
    {
      "epoch": 0.22073333027396633,
      "grad_norm": 0.4814826250076294,
      "learning_rate": 9.926453366688505e-06,
      "loss": 0.421,
      "step": 4810
    },
    {
      "epoch": 0.22077922077922077,
      "grad_norm": 0.5726569294929504,
      "learning_rate": 9.926411461395127e-06,
      "loss": 0.5295,
      "step": 4811
    },
    {
      "epoch": 0.22082511128447524,
      "grad_norm": 0.5161497592926025,
      "learning_rate": 9.926369544255283e-06,
      "loss": 0.5035,
      "step": 4812
    },
    {
      "epoch": 0.2208710017897297,
      "grad_norm": 0.46062567830085754,
      "learning_rate": 9.926327615269077e-06,
      "loss": 0.3891,
      "step": 4813
    },
    {
      "epoch": 0.22091689229498418,
      "grad_norm": 0.5229039192199707,
      "learning_rate": 9.926285674436607e-06,
      "loss": 0.4578,
      "step": 4814
    },
    {
      "epoch": 0.22096278280023862,
      "grad_norm": 0.5289216637611389,
      "learning_rate": 9.926243721757975e-06,
      "loss": 0.5183,
      "step": 4815
    },
    {
      "epoch": 0.2210086733054931,
      "grad_norm": 0.4917519986629486,
      "learning_rate": 9.92620175723328e-06,
      "loss": 0.3812,
      "step": 4816
    },
    {
      "epoch": 0.22105456381074756,
      "grad_norm": 0.4435589909553528,
      "learning_rate": 9.926159780862626e-06,
      "loss": 0.3763,
      "step": 4817
    },
    {
      "epoch": 0.22110045431600203,
      "grad_norm": 0.4811856746673584,
      "learning_rate": 9.926117792646113e-06,
      "loss": 0.4062,
      "step": 4818
    },
    {
      "epoch": 0.22114634482125647,
      "grad_norm": 1.2258802652359009,
      "learning_rate": 9.92607579258384e-06,
      "loss": 0.5963,
      "step": 4819
    },
    {
      "epoch": 0.22119223532651094,
      "grad_norm": 0.4955314099788666,
      "learning_rate": 9.92603378067591e-06,
      "loss": 0.4613,
      "step": 4820
    },
    {
      "epoch": 0.22123812583176541,
      "grad_norm": 0.4535846710205078,
      "learning_rate": 9.925991756922424e-06,
      "loss": 0.3626,
      "step": 4821
    },
    {
      "epoch": 0.22128401633701988,
      "grad_norm": 0.499987930059433,
      "learning_rate": 9.925949721323483e-06,
      "loss": 0.3775,
      "step": 4822
    },
    {
      "epoch": 0.22132990684227433,
      "grad_norm": 0.520010232925415,
      "learning_rate": 9.925907673879184e-06,
      "loss": 0.4285,
      "step": 4823
    },
    {
      "epoch": 0.2213757973475288,
      "grad_norm": 0.5386976599693298,
      "learning_rate": 9.925865614589636e-06,
      "loss": 0.5437,
      "step": 4824
    },
    {
      "epoch": 0.22142168785278327,
      "grad_norm": 0.5095916986465454,
      "learning_rate": 9.925823543454933e-06,
      "loss": 0.5072,
      "step": 4825
    },
    {
      "epoch": 0.2214675783580377,
      "grad_norm": 0.4769110381603241,
      "learning_rate": 9.925781460475181e-06,
      "loss": 0.411,
      "step": 4826
    },
    {
      "epoch": 0.22151346886329218,
      "grad_norm": 0.45860767364501953,
      "learning_rate": 9.925739365650478e-06,
      "loss": 0.4112,
      "step": 4827
    },
    {
      "epoch": 0.22155935936854665,
      "grad_norm": 0.4927826523780823,
      "learning_rate": 9.925697258980927e-06,
      "loss": 0.4174,
      "step": 4828
    },
    {
      "epoch": 0.22160524987380112,
      "grad_norm": 0.4682226777076721,
      "learning_rate": 9.925655140466629e-06,
      "loss": 0.3407,
      "step": 4829
    },
    {
      "epoch": 0.22165114037905556,
      "grad_norm": 0.49554941058158875,
      "learning_rate": 9.925613010107684e-06,
      "loss": 0.4167,
      "step": 4830
    },
    {
      "epoch": 0.22169703088431003,
      "grad_norm": 0.48000916838645935,
      "learning_rate": 9.925570867904195e-06,
      "loss": 0.3714,
      "step": 4831
    },
    {
      "epoch": 0.2217429213895645,
      "grad_norm": 0.5282401442527771,
      "learning_rate": 9.925528713856263e-06,
      "loss": 0.4281,
      "step": 4832
    },
    {
      "epoch": 0.22178881189481897,
      "grad_norm": 0.4732506573200226,
      "learning_rate": 9.925486547963988e-06,
      "loss": 0.4233,
      "step": 4833
    },
    {
      "epoch": 0.2218347024000734,
      "grad_norm": 0.4619845747947693,
      "learning_rate": 9.925444370227473e-06,
      "loss": 0.4208,
      "step": 4834
    },
    {
      "epoch": 0.22188059290532788,
      "grad_norm": 0.4698489308357239,
      "learning_rate": 9.925402180646819e-06,
      "loss": 0.3937,
      "step": 4835
    },
    {
      "epoch": 0.22192648341058235,
      "grad_norm": 0.43977537751197815,
      "learning_rate": 9.925359979222126e-06,
      "loss": 0.3232,
      "step": 4836
    },
    {
      "epoch": 0.22197237391583682,
      "grad_norm": 0.5040274262428284,
      "learning_rate": 9.925317765953498e-06,
      "loss": 0.4913,
      "step": 4837
    },
    {
      "epoch": 0.22201826442109127,
      "grad_norm": 0.44031843543052673,
      "learning_rate": 9.925275540841035e-06,
      "loss": 0.3401,
      "step": 4838
    },
    {
      "epoch": 0.22206415492634574,
      "grad_norm": 0.502856969833374,
      "learning_rate": 9.925233303884838e-06,
      "loss": 0.4747,
      "step": 4839
    },
    {
      "epoch": 0.2221100454316002,
      "grad_norm": 0.47573596239089966,
      "learning_rate": 9.92519105508501e-06,
      "loss": 0.3765,
      "step": 4840
    },
    {
      "epoch": 0.22215593593685468,
      "grad_norm": 0.5229244828224182,
      "learning_rate": 9.925148794441653e-06,
      "loss": 0.568,
      "step": 4841
    },
    {
      "epoch": 0.22220182644210912,
      "grad_norm": 0.7730299234390259,
      "learning_rate": 9.925106521954866e-06,
      "loss": 0.4092,
      "step": 4842
    },
    {
      "epoch": 0.2222477169473636,
      "grad_norm": 0.488257497549057,
      "learning_rate": 9.925064237624754e-06,
      "loss": 0.4716,
      "step": 4843
    },
    {
      "epoch": 0.22229360745261806,
      "grad_norm": 0.5342119932174683,
      "learning_rate": 9.925021941451413e-06,
      "loss": 0.4693,
      "step": 4844
    },
    {
      "epoch": 0.22233949795787253,
      "grad_norm": 0.4660462439060211,
      "learning_rate": 9.924979633434953e-06,
      "loss": 0.3616,
      "step": 4845
    },
    {
      "epoch": 0.22238538846312697,
      "grad_norm": 0.5627459287643433,
      "learning_rate": 9.92493731357547e-06,
      "loss": 0.4949,
      "step": 4846
    },
    {
      "epoch": 0.22243127896838144,
      "grad_norm": 0.5174509286880493,
      "learning_rate": 9.924894981873065e-06,
      "loss": 0.3901,
      "step": 4847
    },
    {
      "epoch": 0.2224771694736359,
      "grad_norm": 0.508909285068512,
      "learning_rate": 9.924852638327845e-06,
      "loss": 0.4304,
      "step": 4848
    },
    {
      "epoch": 0.22252305997889038,
      "grad_norm": 0.4349357485771179,
      "learning_rate": 9.924810282939907e-06,
      "loss": 0.3564,
      "step": 4849
    },
    {
      "epoch": 0.22256895048414482,
      "grad_norm": 0.5553564429283142,
      "learning_rate": 9.924767915709355e-06,
      "loss": 0.4513,
      "step": 4850
    },
    {
      "epoch": 0.2226148409893993,
      "grad_norm": 0.4676671624183655,
      "learning_rate": 9.924725536636289e-06,
      "loss": 0.3673,
      "step": 4851
    },
    {
      "epoch": 0.22266073149465376,
      "grad_norm": 0.45374375581741333,
      "learning_rate": 9.924683145720814e-06,
      "loss": 0.3632,
      "step": 4852
    },
    {
      "epoch": 0.2227066219999082,
      "grad_norm": 0.4324459433555603,
      "learning_rate": 9.92464074296303e-06,
      "loss": 0.3566,
      "step": 4853
    },
    {
      "epoch": 0.22275251250516268,
      "grad_norm": 0.46625950932502747,
      "learning_rate": 9.92459832836304e-06,
      "loss": 0.3997,
      "step": 4854
    },
    {
      "epoch": 0.22279840301041715,
      "grad_norm": 0.48104342818260193,
      "learning_rate": 9.924555901920945e-06,
      "loss": 0.4583,
      "step": 4855
    },
    {
      "epoch": 0.22284429351567162,
      "grad_norm": 0.48989519476890564,
      "learning_rate": 9.924513463636847e-06,
      "loss": 0.4017,
      "step": 4856
    },
    {
      "epoch": 0.22289018402092606,
      "grad_norm": 0.48691120743751526,
      "learning_rate": 9.924471013510848e-06,
      "loss": 0.335,
      "step": 4857
    },
    {
      "epoch": 0.22293607452618053,
      "grad_norm": 0.47887545824050903,
      "learning_rate": 9.92442855154305e-06,
      "loss": 0.433,
      "step": 4858
    },
    {
      "epoch": 0.222981965031435,
      "grad_norm": 0.44519808888435364,
      "learning_rate": 9.924386077733556e-06,
      "loss": 0.3675,
      "step": 4859
    },
    {
      "epoch": 0.22302785553668947,
      "grad_norm": 0.45259690284729004,
      "learning_rate": 9.924343592082468e-06,
      "loss": 0.3812,
      "step": 4860
    },
    {
      "epoch": 0.2230737460419439,
      "grad_norm": 0.4651898443698883,
      "learning_rate": 9.924301094589887e-06,
      "loss": 0.3783,
      "step": 4861
    },
    {
      "epoch": 0.22311963654719838,
      "grad_norm": 0.46688616275787354,
      "learning_rate": 9.924258585255917e-06,
      "loss": 0.3688,
      "step": 4862
    },
    {
      "epoch": 0.22316552705245285,
      "grad_norm": 0.48918551206588745,
      "learning_rate": 9.92421606408066e-06,
      "loss": 0.4354,
      "step": 4863
    },
    {
      "epoch": 0.22321141755770732,
      "grad_norm": 0.4666862189769745,
      "learning_rate": 9.924173531064217e-06,
      "loss": 0.4221,
      "step": 4864
    },
    {
      "epoch": 0.22325730806296176,
      "grad_norm": 0.47541213035583496,
      "learning_rate": 9.924130986206689e-06,
      "loss": 0.4155,
      "step": 4865
    },
    {
      "epoch": 0.22330319856821623,
      "grad_norm": 0.4850527346134186,
      "learning_rate": 9.92408842950818e-06,
      "loss": 0.4539,
      "step": 4866
    },
    {
      "epoch": 0.2233490890734707,
      "grad_norm": 0.4666447043418884,
      "learning_rate": 9.924045860968796e-06,
      "loss": 0.3746,
      "step": 4867
    },
    {
      "epoch": 0.22339497957872517,
      "grad_norm": 0.4348229765892029,
      "learning_rate": 9.924003280588632e-06,
      "loss": 0.3793,
      "step": 4868
    },
    {
      "epoch": 0.22344087008397961,
      "grad_norm": 0.4767991900444031,
      "learning_rate": 9.923960688367795e-06,
      "loss": 0.4597,
      "step": 4869
    },
    {
      "epoch": 0.22348676058923408,
      "grad_norm": 0.46462228894233704,
      "learning_rate": 9.923918084306388e-06,
      "loss": 0.3848,
      "step": 4870
    },
    {
      "epoch": 0.22353265109448855,
      "grad_norm": 0.4375239312648773,
      "learning_rate": 9.923875468404512e-06,
      "loss": 0.3367,
      "step": 4871
    },
    {
      "epoch": 0.22357854159974302,
      "grad_norm": 0.5722892880439758,
      "learning_rate": 9.923832840662269e-06,
      "loss": 0.5179,
      "step": 4872
    },
    {
      "epoch": 0.22362443210499747,
      "grad_norm": 0.48885923624038696,
      "learning_rate": 9.923790201079761e-06,
      "loss": 0.488,
      "step": 4873
    },
    {
      "epoch": 0.22367032261025194,
      "grad_norm": 0.5007164478302002,
      "learning_rate": 9.923747549657093e-06,
      "loss": 0.411,
      "step": 4874
    },
    {
      "epoch": 0.2237162131155064,
      "grad_norm": 0.4516223669052124,
      "learning_rate": 9.923704886394366e-06,
      "loss": 0.3612,
      "step": 4875
    },
    {
      "epoch": 0.22376210362076088,
      "grad_norm": 0.4912409782409668,
      "learning_rate": 9.923662211291682e-06,
      "loss": 0.4297,
      "step": 4876
    },
    {
      "epoch": 0.22380799412601532,
      "grad_norm": 0.4462112784385681,
      "learning_rate": 9.923619524349144e-06,
      "loss": 0.3832,
      "step": 4877
    },
    {
      "epoch": 0.2238538846312698,
      "grad_norm": 0.4445367753505707,
      "learning_rate": 9.923576825566856e-06,
      "loss": 0.3615,
      "step": 4878
    },
    {
      "epoch": 0.22389977513652426,
      "grad_norm": 0.4949045777320862,
      "learning_rate": 9.923534114944918e-06,
      "loss": 0.3948,
      "step": 4879
    },
    {
      "epoch": 0.22394566564177873,
      "grad_norm": 0.48120051622390747,
      "learning_rate": 9.923491392483436e-06,
      "loss": 0.3816,
      "step": 4880
    },
    {
      "epoch": 0.22399155614703317,
      "grad_norm": 0.44552674889564514,
      "learning_rate": 9.92344865818251e-06,
      "loss": 0.37,
      "step": 4881
    },
    {
      "epoch": 0.22403744665228764,
      "grad_norm": 0.4595223069190979,
      "learning_rate": 9.923405912042246e-06,
      "loss": 0.3798,
      "step": 4882
    },
    {
      "epoch": 0.2240833371575421,
      "grad_norm": 0.4784606993198395,
      "learning_rate": 9.923363154062741e-06,
      "loss": 0.4145,
      "step": 4883
    },
    {
      "epoch": 0.22412922766279655,
      "grad_norm": 0.4554690718650818,
      "learning_rate": 9.923320384244104e-06,
      "loss": 0.3452,
      "step": 4884
    },
    {
      "epoch": 0.22417511816805102,
      "grad_norm": 0.48759523034095764,
      "learning_rate": 9.923277602586436e-06,
      "loss": 0.5028,
      "step": 4885
    },
    {
      "epoch": 0.2242210086733055,
      "grad_norm": 0.4825500547885895,
      "learning_rate": 9.923234809089839e-06,
      "loss": 0.3734,
      "step": 4886
    },
    {
      "epoch": 0.22426689917855996,
      "grad_norm": 0.5292503237724304,
      "learning_rate": 9.923192003754414e-06,
      "loss": 0.5135,
      "step": 4887
    },
    {
      "epoch": 0.2243127896838144,
      "grad_norm": 0.4699976444244385,
      "learning_rate": 9.923149186580268e-06,
      "loss": 0.3411,
      "step": 4888
    },
    {
      "epoch": 0.22435868018906888,
      "grad_norm": 0.5008219480514526,
      "learning_rate": 9.9231063575675e-06,
      "loss": 0.437,
      "step": 4889
    },
    {
      "epoch": 0.22440457069432335,
      "grad_norm": 0.5333161354064941,
      "learning_rate": 9.923063516716217e-06,
      "loss": 0.4769,
      "step": 4890
    },
    {
      "epoch": 0.22445046119957782,
      "grad_norm": 0.47741106152534485,
      "learning_rate": 9.92302066402652e-06,
      "loss": 0.3862,
      "step": 4891
    },
    {
      "epoch": 0.22449635170483226,
      "grad_norm": 0.48897621035575867,
      "learning_rate": 9.92297779949851e-06,
      "loss": 0.4839,
      "step": 4892
    },
    {
      "epoch": 0.22454224221008673,
      "grad_norm": 0.48338815569877625,
      "learning_rate": 9.922934923132294e-06,
      "loss": 0.4277,
      "step": 4893
    },
    {
      "epoch": 0.2245881327153412,
      "grad_norm": 0.4981870949268341,
      "learning_rate": 9.922892034927974e-06,
      "loss": 0.4592,
      "step": 4894
    },
    {
      "epoch": 0.22463402322059567,
      "grad_norm": 0.5111919641494751,
      "learning_rate": 9.92284913488565e-06,
      "loss": 0.5358,
      "step": 4895
    },
    {
      "epoch": 0.2246799137258501,
      "grad_norm": 0.5054614543914795,
      "learning_rate": 9.922806223005427e-06,
      "loss": 0.4364,
      "step": 4896
    },
    {
      "epoch": 0.22472580423110458,
      "grad_norm": 0.5211411118507385,
      "learning_rate": 9.922763299287413e-06,
      "loss": 0.4487,
      "step": 4897
    },
    {
      "epoch": 0.22477169473635905,
      "grad_norm": 0.4872068464756012,
      "learning_rate": 9.922720363731704e-06,
      "loss": 0.4725,
      "step": 4898
    },
    {
      "epoch": 0.22481758524161352,
      "grad_norm": 0.45341187715530396,
      "learning_rate": 9.922677416338406e-06,
      "loss": 0.3843,
      "step": 4899
    },
    {
      "epoch": 0.22486347574686796,
      "grad_norm": 0.4295167624950409,
      "learning_rate": 9.922634457107622e-06,
      "loss": 0.3519,
      "step": 4900
    },
    {
      "epoch": 0.22490936625212243,
      "grad_norm": 0.4651036560535431,
      "learning_rate": 9.922591486039456e-06,
      "loss": 0.3863,
      "step": 4901
    },
    {
      "epoch": 0.2249552567573769,
      "grad_norm": 0.48001551628112793,
      "learning_rate": 9.922548503134013e-06,
      "loss": 0.4453,
      "step": 4902
    },
    {
      "epoch": 0.22500114726263137,
      "grad_norm": 0.4280970096588135,
      "learning_rate": 9.922505508391393e-06,
      "loss": 0.3189,
      "step": 4903
    },
    {
      "epoch": 0.22504703776788582,
      "grad_norm": 0.44235193729400635,
      "learning_rate": 9.922462501811701e-06,
      "loss": 0.3418,
      "step": 4904
    },
    {
      "epoch": 0.22509292827314029,
      "grad_norm": 0.4464187026023865,
      "learning_rate": 9.92241948339504e-06,
      "loss": 0.3817,
      "step": 4905
    },
    {
      "epoch": 0.22513881877839476,
      "grad_norm": 0.4962049424648285,
      "learning_rate": 9.922376453141515e-06,
      "loss": 0.4429,
      "step": 4906
    },
    {
      "epoch": 0.22518470928364923,
      "grad_norm": 0.4880332350730896,
      "learning_rate": 9.922333411051225e-06,
      "loss": 0.4166,
      "step": 4907
    },
    {
      "epoch": 0.22523059978890367,
      "grad_norm": 0.4541180431842804,
      "learning_rate": 9.92229035712428e-06,
      "loss": 0.4276,
      "step": 4908
    },
    {
      "epoch": 0.22527649029415814,
      "grad_norm": 0.4275414049625397,
      "learning_rate": 9.922247291360779e-06,
      "loss": 0.3274,
      "step": 4909
    },
    {
      "epoch": 0.2253223807994126,
      "grad_norm": 0.4537696838378906,
      "learning_rate": 9.922204213760827e-06,
      "loss": 0.3627,
      "step": 4910
    },
    {
      "epoch": 0.22536827130466708,
      "grad_norm": 0.4656986892223358,
      "learning_rate": 9.922161124324528e-06,
      "loss": 0.408,
      "step": 4911
    },
    {
      "epoch": 0.22541416180992152,
      "grad_norm": 0.49436402320861816,
      "learning_rate": 9.922118023051983e-06,
      "loss": 0.4223,
      "step": 4912
    },
    {
      "epoch": 0.225460052315176,
      "grad_norm": 0.43175196647644043,
      "learning_rate": 9.9220749099433e-06,
      "loss": 0.3672,
      "step": 4913
    },
    {
      "epoch": 0.22550594282043046,
      "grad_norm": 0.45036956667900085,
      "learning_rate": 9.922031784998577e-06,
      "loss": 0.3082,
      "step": 4914
    },
    {
      "epoch": 0.2255518333256849,
      "grad_norm": 0.4689321517944336,
      "learning_rate": 9.921988648217924e-06,
      "loss": 0.3918,
      "step": 4915
    },
    {
      "epoch": 0.22559772383093937,
      "grad_norm": 0.4360283613204956,
      "learning_rate": 9.921945499601442e-06,
      "loss": 0.3363,
      "step": 4916
    },
    {
      "epoch": 0.22564361433619384,
      "grad_norm": 0.4614073634147644,
      "learning_rate": 9.921902339149232e-06,
      "loss": 0.4051,
      "step": 4917
    },
    {
      "epoch": 0.2256895048414483,
      "grad_norm": 0.44617804884910583,
      "learning_rate": 9.921859166861401e-06,
      "loss": 0.4,
      "step": 4918
    },
    {
      "epoch": 0.22573539534670276,
      "grad_norm": 0.5219184756278992,
      "learning_rate": 9.921815982738054e-06,
      "loss": 0.5138,
      "step": 4919
    },
    {
      "epoch": 0.22578128585195723,
      "grad_norm": 0.44831669330596924,
      "learning_rate": 9.921772786779291e-06,
      "loss": 0.3405,
      "step": 4920
    },
    {
      "epoch": 0.2258271763572117,
      "grad_norm": 0.4779570698738098,
      "learning_rate": 9.921729578985218e-06,
      "loss": 0.4236,
      "step": 4921
    },
    {
      "epoch": 0.22587306686246617,
      "grad_norm": 0.42781367897987366,
      "learning_rate": 9.92168635935594e-06,
      "loss": 0.336,
      "step": 4922
    },
    {
      "epoch": 0.2259189573677206,
      "grad_norm": 0.43575751781463623,
      "learning_rate": 9.921643127891558e-06,
      "loss": 0.3422,
      "step": 4923
    },
    {
      "epoch": 0.22596484787297508,
      "grad_norm": 0.47726768255233765,
      "learning_rate": 9.92159988459218e-06,
      "loss": 0.4122,
      "step": 4924
    },
    {
      "epoch": 0.22601073837822955,
      "grad_norm": 0.5359604954719543,
      "learning_rate": 9.921556629457904e-06,
      "loss": 0.573,
      "step": 4925
    },
    {
      "epoch": 0.22605662888348402,
      "grad_norm": 0.4933781027793884,
      "learning_rate": 9.92151336248884e-06,
      "loss": 0.4153,
      "step": 4926
    },
    {
      "epoch": 0.22610251938873846,
      "grad_norm": 0.43301907181739807,
      "learning_rate": 9.92147008368509e-06,
      "loss": 0.3239,
      "step": 4927
    },
    {
      "epoch": 0.22614840989399293,
      "grad_norm": 0.47758033871650696,
      "learning_rate": 9.921426793046756e-06,
      "loss": 0.3941,
      "step": 4928
    },
    {
      "epoch": 0.2261943003992474,
      "grad_norm": 0.5660764575004578,
      "learning_rate": 9.921383490573944e-06,
      "loss": 0.5265,
      "step": 4929
    },
    {
      "epoch": 0.22624019090450187,
      "grad_norm": 0.46754592657089233,
      "learning_rate": 9.921340176266759e-06,
      "loss": 0.4155,
      "step": 4930
    },
    {
      "epoch": 0.2262860814097563,
      "grad_norm": 0.45041424036026,
      "learning_rate": 9.921296850125304e-06,
      "loss": 0.4033,
      "step": 4931
    },
    {
      "epoch": 0.22633197191501078,
      "grad_norm": 0.4429803192615509,
      "learning_rate": 9.921253512149682e-06,
      "loss": 0.3396,
      "step": 4932
    },
    {
      "epoch": 0.22637786242026525,
      "grad_norm": 0.42263364791870117,
      "learning_rate": 9.92121016234e-06,
      "loss": 0.3202,
      "step": 4933
    },
    {
      "epoch": 0.22642375292551972,
      "grad_norm": 0.49394991993904114,
      "learning_rate": 9.92116680069636e-06,
      "loss": 0.4353,
      "step": 4934
    },
    {
      "epoch": 0.22646964343077416,
      "grad_norm": 0.48084592819213867,
      "learning_rate": 9.921123427218867e-06,
      "loss": 0.4636,
      "step": 4935
    },
    {
      "epoch": 0.22651553393602863,
      "grad_norm": 0.5092942714691162,
      "learning_rate": 9.921080041907625e-06,
      "loss": 0.4916,
      "step": 4936
    },
    {
      "epoch": 0.2265614244412831,
      "grad_norm": 0.6865168213844299,
      "learning_rate": 9.92103664476274e-06,
      "loss": 0.5119,
      "step": 4937
    },
    {
      "epoch": 0.22660731494653757,
      "grad_norm": 0.47005903720855713,
      "learning_rate": 9.920993235784314e-06,
      "loss": 0.4308,
      "step": 4938
    },
    {
      "epoch": 0.22665320545179202,
      "grad_norm": 0.545192301273346,
      "learning_rate": 9.920949814972453e-06,
      "loss": 0.4275,
      "step": 4939
    },
    {
      "epoch": 0.2266990959570465,
      "grad_norm": 0.5689237713813782,
      "learning_rate": 9.92090638232726e-06,
      "loss": 0.4499,
      "step": 4940
    },
    {
      "epoch": 0.22674498646230096,
      "grad_norm": 0.46823644638061523,
      "learning_rate": 9.920862937848841e-06,
      "loss": 0.3879,
      "step": 4941
    },
    {
      "epoch": 0.2267908769675554,
      "grad_norm": 0.450050950050354,
      "learning_rate": 9.920819481537299e-06,
      "loss": 0.3849,
      "step": 4942
    },
    {
      "epoch": 0.22683676747280987,
      "grad_norm": 0.46422719955444336,
      "learning_rate": 9.920776013392739e-06,
      "loss": 0.4221,
      "step": 4943
    },
    {
      "epoch": 0.22688265797806434,
      "grad_norm": 0.48257312178611755,
      "learning_rate": 9.920732533415265e-06,
      "loss": 0.4016,
      "step": 4944
    },
    {
      "epoch": 0.2269285484833188,
      "grad_norm": 0.4819818437099457,
      "learning_rate": 9.920689041604986e-06,
      "loss": 0.4961,
      "step": 4945
    },
    {
      "epoch": 0.22697443898857325,
      "grad_norm": 0.46663084626197815,
      "learning_rate": 9.920645537962e-06,
      "loss": 0.3989,
      "step": 4946
    },
    {
      "epoch": 0.22702032949382772,
      "grad_norm": 0.4643586575984955,
      "learning_rate": 9.920602022486414e-06,
      "loss": 0.3691,
      "step": 4947
    },
    {
      "epoch": 0.2270662199990822,
      "grad_norm": 0.4569918215274811,
      "learning_rate": 9.920558495178334e-06,
      "loss": 0.3682,
      "step": 4948
    },
    {
      "epoch": 0.22711211050433666,
      "grad_norm": 0.44335222244262695,
      "learning_rate": 9.920514956037864e-06,
      "loss": 0.353,
      "step": 4949
    },
    {
      "epoch": 0.2271580010095911,
      "grad_norm": 0.5647660493850708,
      "learning_rate": 9.920471405065108e-06,
      "loss": 0.5335,
      "step": 4950
    },
    {
      "epoch": 0.22720389151484557,
      "grad_norm": 0.4861886203289032,
      "learning_rate": 9.92042784226017e-06,
      "loss": 0.4031,
      "step": 4951
    },
    {
      "epoch": 0.22724978202010004,
      "grad_norm": 0.43190282583236694,
      "learning_rate": 9.920384267623158e-06,
      "loss": 0.337,
      "step": 4952
    },
    {
      "epoch": 0.22729567252535451,
      "grad_norm": 0.5032286047935486,
      "learning_rate": 9.920340681154175e-06,
      "loss": 0.4178,
      "step": 4953
    },
    {
      "epoch": 0.22734156303060896,
      "grad_norm": 0.5101658701896667,
      "learning_rate": 9.920297082853325e-06,
      "loss": 0.4177,
      "step": 4954
    },
    {
      "epoch": 0.22738745353586343,
      "grad_norm": 0.5094603896141052,
      "learning_rate": 9.920253472720713e-06,
      "loss": 0.3953,
      "step": 4955
    },
    {
      "epoch": 0.2274333440411179,
      "grad_norm": 0.512938916683197,
      "learning_rate": 9.920209850756443e-06,
      "loss": 0.4765,
      "step": 4956
    },
    {
      "epoch": 0.22747923454637237,
      "grad_norm": 0.4783475995063782,
      "learning_rate": 9.920166216960622e-06,
      "loss": 0.3806,
      "step": 4957
    },
    {
      "epoch": 0.2275251250516268,
      "grad_norm": 0.5001879930496216,
      "learning_rate": 9.920122571333356e-06,
      "loss": 0.4572,
      "step": 4958
    },
    {
      "epoch": 0.22757101555688128,
      "grad_norm": 0.46075478196144104,
      "learning_rate": 9.920078913874746e-06,
      "loss": 0.3807,
      "step": 4959
    },
    {
      "epoch": 0.22761690606213575,
      "grad_norm": 0.4371754229068756,
      "learning_rate": 9.9200352445849e-06,
      "loss": 0.3531,
      "step": 4960
    },
    {
      "epoch": 0.22766279656739022,
      "grad_norm": 0.49161723256111145,
      "learning_rate": 9.919991563463922e-06,
      "loss": 0.3624,
      "step": 4961
    },
    {
      "epoch": 0.22770868707264466,
      "grad_norm": 0.5151509642601013,
      "learning_rate": 9.919947870511916e-06,
      "loss": 0.5005,
      "step": 4962
    },
    {
      "epoch": 0.22775457757789913,
      "grad_norm": 0.43065324425697327,
      "learning_rate": 9.919904165728992e-06,
      "loss": 0.3367,
      "step": 4963
    },
    {
      "epoch": 0.2278004680831536,
      "grad_norm": 0.47946444153785706,
      "learning_rate": 9.919860449115246e-06,
      "loss": 0.4089,
      "step": 4964
    },
    {
      "epoch": 0.22784635858840807,
      "grad_norm": 0.48312655091285706,
      "learning_rate": 9.91981672067079e-06,
      "loss": 0.3934,
      "step": 4965
    },
    {
      "epoch": 0.2278922490936625,
      "grad_norm": 0.4617455303668976,
      "learning_rate": 9.91977298039573e-06,
      "loss": 0.3329,
      "step": 4966
    },
    {
      "epoch": 0.22793813959891698,
      "grad_norm": 0.5390293002128601,
      "learning_rate": 9.919729228290165e-06,
      "loss": 0.5246,
      "step": 4967
    },
    {
      "epoch": 0.22798403010417145,
      "grad_norm": 0.46788522601127625,
      "learning_rate": 9.919685464354207e-06,
      "loss": 0.4028,
      "step": 4968
    },
    {
      "epoch": 0.22802992060942592,
      "grad_norm": 0.451246052980423,
      "learning_rate": 9.919641688587956e-06,
      "loss": 0.3774,
      "step": 4969
    },
    {
      "epoch": 0.22807581111468037,
      "grad_norm": 0.46292322874069214,
      "learning_rate": 9.919597900991522e-06,
      "loss": 0.4005,
      "step": 4970
    },
    {
      "epoch": 0.22812170161993484,
      "grad_norm": 0.46303513646125793,
      "learning_rate": 9.919554101565005e-06,
      "loss": 0.3767,
      "step": 4971
    },
    {
      "epoch": 0.2281675921251893,
      "grad_norm": 0.5459726452827454,
      "learning_rate": 9.919510290308514e-06,
      "loss": 0.5472,
      "step": 4972
    },
    {
      "epoch": 0.22821348263044375,
      "grad_norm": 0.4538825452327728,
      "learning_rate": 9.919466467222155e-06,
      "loss": 0.3638,
      "step": 4973
    },
    {
      "epoch": 0.22825937313569822,
      "grad_norm": 0.4757990539073944,
      "learning_rate": 9.919422632306028e-06,
      "loss": 0.42,
      "step": 4974
    },
    {
      "epoch": 0.2283052636409527,
      "grad_norm": 0.4459306597709656,
      "learning_rate": 9.919378785560246e-06,
      "loss": 0.3834,
      "step": 4975
    },
    {
      "epoch": 0.22835115414620716,
      "grad_norm": 0.45102763175964355,
      "learning_rate": 9.919334926984909e-06,
      "loss": 0.3641,
      "step": 4976
    },
    {
      "epoch": 0.2283970446514616,
      "grad_norm": 0.4604094326496124,
      "learning_rate": 9.919291056580124e-06,
      "loss": 0.3718,
      "step": 4977
    },
    {
      "epoch": 0.22844293515671607,
      "grad_norm": 0.4630272388458252,
      "learning_rate": 9.919247174345996e-06,
      "loss": 0.4185,
      "step": 4978
    },
    {
      "epoch": 0.22848882566197054,
      "grad_norm": 0.4595909118652344,
      "learning_rate": 9.919203280282633e-06,
      "loss": 0.3717,
      "step": 4979
    },
    {
      "epoch": 0.228534716167225,
      "grad_norm": 0.4760896861553192,
      "learning_rate": 9.919159374390138e-06,
      "loss": 0.4476,
      "step": 4980
    },
    {
      "epoch": 0.22858060667247945,
      "grad_norm": 0.48160412907600403,
      "learning_rate": 9.919115456668615e-06,
      "loss": 0.4438,
      "step": 4981
    },
    {
      "epoch": 0.22862649717773392,
      "grad_norm": 0.5180866718292236,
      "learning_rate": 9.919071527118173e-06,
      "loss": 0.34,
      "step": 4982
    },
    {
      "epoch": 0.2286723876829884,
      "grad_norm": 0.4597753882408142,
      "learning_rate": 9.91902758573892e-06,
      "loss": 0.4243,
      "step": 4983
    },
    {
      "epoch": 0.22871827818824286,
      "grad_norm": 0.4963390529155731,
      "learning_rate": 9.918983632530953e-06,
      "loss": 0.4177,
      "step": 4984
    },
    {
      "epoch": 0.2287641686934973,
      "grad_norm": 0.4787067770957947,
      "learning_rate": 9.918939667494386e-06,
      "loss": 0.4149,
      "step": 4985
    },
    {
      "epoch": 0.22881005919875178,
      "grad_norm": 0.44735705852508545,
      "learning_rate": 9.91889569062932e-06,
      "loss": 0.3726,
      "step": 4986
    },
    {
      "epoch": 0.22885594970400625,
      "grad_norm": 0.4647885262966156,
      "learning_rate": 9.918851701935863e-06,
      "loss": 0.4184,
      "step": 4987
    },
    {
      "epoch": 0.22890184020926072,
      "grad_norm": 0.4930536150932312,
      "learning_rate": 9.918807701414121e-06,
      "loss": 0.4056,
      "step": 4988
    },
    {
      "epoch": 0.22894773071451516,
      "grad_norm": 0.48790839314460754,
      "learning_rate": 9.918763689064197e-06,
      "loss": 0.4461,
      "step": 4989
    },
    {
      "epoch": 0.22899362121976963,
      "grad_norm": 0.4574865698814392,
      "learning_rate": 9.9187196648862e-06,
      "loss": 0.4118,
      "step": 4990
    },
    {
      "epoch": 0.2290395117250241,
      "grad_norm": 0.4391997158527374,
      "learning_rate": 9.918675628880234e-06,
      "loss": 0.3646,
      "step": 4991
    },
    {
      "epoch": 0.22908540223027857,
      "grad_norm": 0.5032287836074829,
      "learning_rate": 9.918631581046406e-06,
      "loss": 0.3845,
      "step": 4992
    },
    {
      "epoch": 0.229131292735533,
      "grad_norm": 0.5309714674949646,
      "learning_rate": 9.918587521384822e-06,
      "loss": 0.4641,
      "step": 4993
    },
    {
      "epoch": 0.22917718324078748,
      "grad_norm": 0.48825475573539734,
      "learning_rate": 9.918543449895586e-06,
      "loss": 0.4098,
      "step": 4994
    },
    {
      "epoch": 0.22922307374604195,
      "grad_norm": 0.48650339245796204,
      "learning_rate": 9.918499366578806e-06,
      "loss": 0.432,
      "step": 4995
    },
    {
      "epoch": 0.22926896425129642,
      "grad_norm": 0.49815139174461365,
      "learning_rate": 9.918455271434586e-06,
      "loss": 0.4667,
      "step": 4996
    },
    {
      "epoch": 0.22931485475655086,
      "grad_norm": 0.4991191327571869,
      "learning_rate": 9.918411164463036e-06,
      "loss": 0.3274,
      "step": 4997
    },
    {
      "epoch": 0.22936074526180533,
      "grad_norm": 0.47008630633354187,
      "learning_rate": 9.918367045664257e-06,
      "loss": 0.3997,
      "step": 4998
    },
    {
      "epoch": 0.2294066357670598,
      "grad_norm": 0.5059807896614075,
      "learning_rate": 9.91832291503836e-06,
      "loss": 0.4651,
      "step": 4999
    },
    {
      "epoch": 0.22945252627231424,
      "grad_norm": 0.48933085799217224,
      "learning_rate": 9.918278772585446e-06,
      "loss": 0.4763,
      "step": 5000
    },
    {
      "epoch": 0.22949841677756871,
      "grad_norm": 0.5065524578094482,
      "learning_rate": 9.918234618305626e-06,
      "loss": 0.4039,
      "step": 5001
    },
    {
      "epoch": 0.22954430728282318,
      "grad_norm": 0.47976741194725037,
      "learning_rate": 9.918190452199002e-06,
      "loss": 0.4145,
      "step": 5002
    },
    {
      "epoch": 0.22959019778807765,
      "grad_norm": 0.5124884843826294,
      "learning_rate": 9.918146274265684e-06,
      "loss": 0.5291,
      "step": 5003
    },
    {
      "epoch": 0.2296360882933321,
      "grad_norm": 0.504642128944397,
      "learning_rate": 9.918102084505776e-06,
      "loss": 0.4208,
      "step": 5004
    },
    {
      "epoch": 0.22968197879858657,
      "grad_norm": 0.49537789821624756,
      "learning_rate": 9.918057882919383e-06,
      "loss": 0.4488,
      "step": 5005
    },
    {
      "epoch": 0.22972786930384104,
      "grad_norm": 0.48464325070381165,
      "learning_rate": 9.918013669506615e-06,
      "loss": 0.4553,
      "step": 5006
    },
    {
      "epoch": 0.2297737598090955,
      "grad_norm": 0.502467691898346,
      "learning_rate": 9.917969444267575e-06,
      "loss": 0.44,
      "step": 5007
    },
    {
      "epoch": 0.22981965031434995,
      "grad_norm": 0.5058987140655518,
      "learning_rate": 9.917925207202372e-06,
      "loss": 0.4285,
      "step": 5008
    },
    {
      "epoch": 0.22986554081960442,
      "grad_norm": 0.4840467870235443,
      "learning_rate": 9.91788095831111e-06,
      "loss": 0.3878,
      "step": 5009
    },
    {
      "epoch": 0.2299114313248589,
      "grad_norm": 0.4543435275554657,
      "learning_rate": 9.917836697593896e-06,
      "loss": 0.4097,
      "step": 5010
    },
    {
      "epoch": 0.22995732183011336,
      "grad_norm": 0.43923765420913696,
      "learning_rate": 9.917792425050838e-06,
      "loss": 0.3778,
      "step": 5011
    },
    {
      "epoch": 0.2300032123353678,
      "grad_norm": 0.5077973008155823,
      "learning_rate": 9.917748140682042e-06,
      "loss": 0.4911,
      "step": 5012
    },
    {
      "epoch": 0.23004910284062227,
      "grad_norm": 0.4553045630455017,
      "learning_rate": 9.917703844487611e-06,
      "loss": 0.3817,
      "step": 5013
    },
    {
      "epoch": 0.23009499334587674,
      "grad_norm": 0.4960514008998871,
      "learning_rate": 9.917659536467656e-06,
      "loss": 0.4597,
      "step": 5014
    },
    {
      "epoch": 0.2301408838511312,
      "grad_norm": 0.4994480311870575,
      "learning_rate": 9.917615216622284e-06,
      "loss": 0.5244,
      "step": 5015
    },
    {
      "epoch": 0.23018677435638565,
      "grad_norm": 0.4767090678215027,
      "learning_rate": 9.917570884951596e-06,
      "loss": 0.4246,
      "step": 5016
    },
    {
      "epoch": 0.23023266486164012,
      "grad_norm": 0.44151371717453003,
      "learning_rate": 9.917526541455704e-06,
      "loss": 0.3646,
      "step": 5017
    },
    {
      "epoch": 0.2302785553668946,
      "grad_norm": 0.4284513592720032,
      "learning_rate": 9.917482186134711e-06,
      "loss": 0.3836,
      "step": 5018
    },
    {
      "epoch": 0.23032444587214906,
      "grad_norm": 0.48593762516975403,
      "learning_rate": 9.917437818988727e-06,
      "loss": 0.4625,
      "step": 5019
    },
    {
      "epoch": 0.2303703363774035,
      "grad_norm": 0.4842730760574341,
      "learning_rate": 9.917393440017856e-06,
      "loss": 0.4597,
      "step": 5020
    },
    {
      "epoch": 0.23041622688265798,
      "grad_norm": 0.6369385123252869,
      "learning_rate": 9.917349049222207e-06,
      "loss": 0.4711,
      "step": 5021
    },
    {
      "epoch": 0.23046211738791245,
      "grad_norm": 0.4542660713195801,
      "learning_rate": 9.917304646601884e-06,
      "loss": 0.3852,
      "step": 5022
    },
    {
      "epoch": 0.23050800789316692,
      "grad_norm": 0.4675392508506775,
      "learning_rate": 9.917260232156995e-06,
      "loss": 0.3547,
      "step": 5023
    },
    {
      "epoch": 0.23055389839842136,
      "grad_norm": 0.4550405442714691,
      "learning_rate": 9.91721580588765e-06,
      "loss": 0.355,
      "step": 5024
    },
    {
      "epoch": 0.23059978890367583,
      "grad_norm": 0.47967174649238586,
      "learning_rate": 9.91717136779395e-06,
      "loss": 0.435,
      "step": 5025
    },
    {
      "epoch": 0.2306456794089303,
      "grad_norm": 0.454787939786911,
      "learning_rate": 9.917126917876006e-06,
      "loss": 0.4212,
      "step": 5026
    },
    {
      "epoch": 0.23069156991418477,
      "grad_norm": 0.4577714502811432,
      "learning_rate": 9.917082456133923e-06,
      "loss": 0.4459,
      "step": 5027
    },
    {
      "epoch": 0.2307374604194392,
      "grad_norm": 0.4787876307964325,
      "learning_rate": 9.91703798256781e-06,
      "loss": 0.4197,
      "step": 5028
    },
    {
      "epoch": 0.23078335092469368,
      "grad_norm": 0.4666035771369934,
      "learning_rate": 9.91699349717777e-06,
      "loss": 0.3405,
      "step": 5029
    },
    {
      "epoch": 0.23082924142994815,
      "grad_norm": 0.48483747243881226,
      "learning_rate": 9.916948999963915e-06,
      "loss": 0.4135,
      "step": 5030
    },
    {
      "epoch": 0.2308751319352026,
      "grad_norm": 0.49244368076324463,
      "learning_rate": 9.916904490926348e-06,
      "loss": 0.4789,
      "step": 5031
    },
    {
      "epoch": 0.23092102244045706,
      "grad_norm": 0.47838544845581055,
      "learning_rate": 9.916859970065178e-06,
      "loss": 0.39,
      "step": 5032
    },
    {
      "epoch": 0.23096691294571153,
      "grad_norm": 0.4705907702445984,
      "learning_rate": 9.916815437380511e-06,
      "loss": 0.3959,
      "step": 5033
    },
    {
      "epoch": 0.231012803450966,
      "grad_norm": 0.5116842985153198,
      "learning_rate": 9.916770892872455e-06,
      "loss": 0.4381,
      "step": 5034
    },
    {
      "epoch": 0.23105869395622045,
      "grad_norm": 0.4437456727027893,
      "learning_rate": 9.916726336541117e-06,
      "loss": 0.4053,
      "step": 5035
    },
    {
      "epoch": 0.23110458446147492,
      "grad_norm": 0.4988843500614166,
      "learning_rate": 9.916681768386605e-06,
      "loss": 0.4513,
      "step": 5036
    },
    {
      "epoch": 0.23115047496672939,
      "grad_norm": 0.4737136662006378,
      "learning_rate": 9.916637188409024e-06,
      "loss": 0.4184,
      "step": 5037
    },
    {
      "epoch": 0.23119636547198386,
      "grad_norm": 0.46205082535743713,
      "learning_rate": 9.91659259660848e-06,
      "loss": 0.395,
      "step": 5038
    },
    {
      "epoch": 0.2312422559772383,
      "grad_norm": 0.4780011475086212,
      "learning_rate": 9.916547992985087e-06,
      "loss": 0.4331,
      "step": 5039
    },
    {
      "epoch": 0.23128814648249277,
      "grad_norm": 0.42814281582832336,
      "learning_rate": 9.916503377538945e-06,
      "loss": 0.3221,
      "step": 5040
    },
    {
      "epoch": 0.23133403698774724,
      "grad_norm": 0.4663562774658203,
      "learning_rate": 9.916458750270164e-06,
      "loss": 0.3745,
      "step": 5041
    },
    {
      "epoch": 0.2313799274930017,
      "grad_norm": 0.448881059885025,
      "learning_rate": 9.916414111178853e-06,
      "loss": 0.3841,
      "step": 5042
    },
    {
      "epoch": 0.23142581799825615,
      "grad_norm": 0.45631641149520874,
      "learning_rate": 9.916369460265116e-06,
      "loss": 0.399,
      "step": 5043
    },
    {
      "epoch": 0.23147170850351062,
      "grad_norm": 0.4652758836746216,
      "learning_rate": 9.916324797529063e-06,
      "loss": 0.3973,
      "step": 5044
    },
    {
      "epoch": 0.2315175990087651,
      "grad_norm": 0.42635032534599304,
      "learning_rate": 9.9162801229708e-06,
      "loss": 0.3452,
      "step": 5045
    },
    {
      "epoch": 0.23156348951401956,
      "grad_norm": 0.5374182462692261,
      "learning_rate": 9.916235436590436e-06,
      "loss": 0.5509,
      "step": 5046
    },
    {
      "epoch": 0.231609380019274,
      "grad_norm": 0.4726683497428894,
      "learning_rate": 9.916190738388076e-06,
      "loss": 0.3692,
      "step": 5047
    },
    {
      "epoch": 0.23165527052452847,
      "grad_norm": 0.5173768401145935,
      "learning_rate": 9.91614602836383e-06,
      "loss": 0.4784,
      "step": 5048
    },
    {
      "epoch": 0.23170116102978294,
      "grad_norm": 0.4761207103729248,
      "learning_rate": 9.916101306517804e-06,
      "loss": 0.4431,
      "step": 5049
    },
    {
      "epoch": 0.2317470515350374,
      "grad_norm": 0.46522706747055054,
      "learning_rate": 9.916056572850106e-06,
      "loss": 0.4243,
      "step": 5050
    },
    {
      "epoch": 0.23179294204029185,
      "grad_norm": 0.5308298468589783,
      "learning_rate": 9.916011827360842e-06,
      "loss": 0.4571,
      "step": 5051
    },
    {
      "epoch": 0.23183883254554633,
      "grad_norm": 0.45010480284690857,
      "learning_rate": 9.915967070050124e-06,
      "loss": 0.3421,
      "step": 5052
    },
    {
      "epoch": 0.2318847230508008,
      "grad_norm": 0.5172383189201355,
      "learning_rate": 9.915922300918054e-06,
      "loss": 0.4745,
      "step": 5053
    },
    {
      "epoch": 0.23193061355605527,
      "grad_norm": 0.4733390808105469,
      "learning_rate": 9.915877519964744e-06,
      "loss": 0.3964,
      "step": 5054
    },
    {
      "epoch": 0.2319765040613097,
      "grad_norm": 0.45765385031700134,
      "learning_rate": 9.9158327271903e-06,
      "loss": 0.3607,
      "step": 5055
    },
    {
      "epoch": 0.23202239456656418,
      "grad_norm": 0.5071452260017395,
      "learning_rate": 9.915787922594829e-06,
      "loss": 0.4789,
      "step": 5056
    },
    {
      "epoch": 0.23206828507181865,
      "grad_norm": 0.45090365409851074,
      "learning_rate": 9.91574310617844e-06,
      "loss": 0.3889,
      "step": 5057
    },
    {
      "epoch": 0.23211417557707312,
      "grad_norm": 0.457913339138031,
      "learning_rate": 9.91569827794124e-06,
      "loss": 0.4029,
      "step": 5058
    },
    {
      "epoch": 0.23216006608232756,
      "grad_norm": 0.458809494972229,
      "learning_rate": 9.915653437883337e-06,
      "loss": 0.3796,
      "step": 5059
    },
    {
      "epoch": 0.23220595658758203,
      "grad_norm": 0.5112406015396118,
      "learning_rate": 9.915608586004839e-06,
      "loss": 0.5472,
      "step": 5060
    },
    {
      "epoch": 0.2322518470928365,
      "grad_norm": 0.527134120464325,
      "learning_rate": 9.915563722305854e-06,
      "loss": 0.511,
      "step": 5061
    },
    {
      "epoch": 0.23229773759809094,
      "grad_norm": 0.424339234828949,
      "learning_rate": 9.915518846786489e-06,
      "loss": 0.3261,
      "step": 5062
    },
    {
      "epoch": 0.2323436281033454,
      "grad_norm": 0.5623648166656494,
      "learning_rate": 9.915473959446853e-06,
      "loss": 0.3969,
      "step": 5063
    },
    {
      "epoch": 0.23238951860859988,
      "grad_norm": 0.4731033444404602,
      "learning_rate": 9.915429060287054e-06,
      "loss": 0.381,
      "step": 5064
    },
    {
      "epoch": 0.23243540911385435,
      "grad_norm": 0.4561477303504944,
      "learning_rate": 9.915384149307198e-06,
      "loss": 0.3852,
      "step": 5065
    },
    {
      "epoch": 0.2324812996191088,
      "grad_norm": 0.4622589647769928,
      "learning_rate": 9.915339226507396e-06,
      "loss": 0.3907,
      "step": 5066
    },
    {
      "epoch": 0.23252719012436326,
      "grad_norm": 0.4760129451751709,
      "learning_rate": 9.915294291887755e-06,
      "loss": 0.385,
      "step": 5067
    },
    {
      "epoch": 0.23257308062961773,
      "grad_norm": 0.5139806270599365,
      "learning_rate": 9.915249345448381e-06,
      "loss": 0.4594,
      "step": 5068
    },
    {
      "epoch": 0.2326189711348722,
      "grad_norm": 0.4360876977443695,
      "learning_rate": 9.915204387189384e-06,
      "loss": 0.3597,
      "step": 5069
    },
    {
      "epoch": 0.23266486164012665,
      "grad_norm": 0.44654330611228943,
      "learning_rate": 9.915159417110874e-06,
      "loss": 0.3742,
      "step": 5070
    },
    {
      "epoch": 0.23271075214538112,
      "grad_norm": 0.48947635293006897,
      "learning_rate": 9.915114435212954e-06,
      "loss": 0.4181,
      "step": 5071
    },
    {
      "epoch": 0.2327566426506356,
      "grad_norm": 0.5250278115272522,
      "learning_rate": 9.915069441495735e-06,
      "loss": 0.4752,
      "step": 5072
    },
    {
      "epoch": 0.23280253315589006,
      "grad_norm": 0.4782872200012207,
      "learning_rate": 9.915024435959326e-06,
      "loss": 0.424,
      "step": 5073
    },
    {
      "epoch": 0.2328484236611445,
      "grad_norm": 0.4785747826099396,
      "learning_rate": 9.914979418603833e-06,
      "loss": 0.4427,
      "step": 5074
    },
    {
      "epoch": 0.23289431416639897,
      "grad_norm": 0.6122748255729675,
      "learning_rate": 9.91493438942937e-06,
      "loss": 0.5118,
      "step": 5075
    },
    {
      "epoch": 0.23294020467165344,
      "grad_norm": 0.5223903656005859,
      "learning_rate": 9.914889348436036e-06,
      "loss": 0.4699,
      "step": 5076
    },
    {
      "epoch": 0.2329860951769079,
      "grad_norm": 0.4806162118911743,
      "learning_rate": 9.914844295623947e-06,
      "loss": 0.466,
      "step": 5077
    },
    {
      "epoch": 0.23303198568216235,
      "grad_norm": 0.45232102274894714,
      "learning_rate": 9.914799230993209e-06,
      "loss": 0.3862,
      "step": 5078
    },
    {
      "epoch": 0.23307787618741682,
      "grad_norm": 0.47773313522338867,
      "learning_rate": 9.914754154543929e-06,
      "loss": 0.3989,
      "step": 5079
    },
    {
      "epoch": 0.2331237666926713,
      "grad_norm": 0.4781469404697418,
      "learning_rate": 9.914709066276216e-06,
      "loss": 0.4233,
      "step": 5080
    },
    {
      "epoch": 0.23316965719792576,
      "grad_norm": 0.4485608637332916,
      "learning_rate": 9.91466396619018e-06,
      "loss": 0.3653,
      "step": 5081
    },
    {
      "epoch": 0.2332155477031802,
      "grad_norm": 0.4454772174358368,
      "learning_rate": 9.914618854285927e-06,
      "loss": 0.3705,
      "step": 5082
    },
    {
      "epoch": 0.23326143820843467,
      "grad_norm": 0.464932918548584,
      "learning_rate": 9.914573730563568e-06,
      "loss": 0.3975,
      "step": 5083
    },
    {
      "epoch": 0.23330732871368914,
      "grad_norm": 0.48395225405693054,
      "learning_rate": 9.91452859502321e-06,
      "loss": 0.4764,
      "step": 5084
    },
    {
      "epoch": 0.2333532192189436,
      "grad_norm": 0.5508474707603455,
      "learning_rate": 9.914483447664961e-06,
      "loss": 0.5212,
      "step": 5085
    },
    {
      "epoch": 0.23339910972419806,
      "grad_norm": 0.4481509327888489,
      "learning_rate": 9.91443828848893e-06,
      "loss": 0.3757,
      "step": 5086
    },
    {
      "epoch": 0.23344500022945253,
      "grad_norm": 0.46165764331817627,
      "learning_rate": 9.914393117495227e-06,
      "loss": 0.3676,
      "step": 5087
    },
    {
      "epoch": 0.233490890734707,
      "grad_norm": 0.4828527271747589,
      "learning_rate": 9.91434793468396e-06,
      "loss": 0.4699,
      "step": 5088
    },
    {
      "epoch": 0.23353678123996144,
      "grad_norm": 0.45526665449142456,
      "learning_rate": 9.914302740055237e-06,
      "loss": 0.3505,
      "step": 5089
    },
    {
      "epoch": 0.2335826717452159,
      "grad_norm": 0.4978918135166168,
      "learning_rate": 9.914257533609166e-06,
      "loss": 0.399,
      "step": 5090
    },
    {
      "epoch": 0.23362856225047038,
      "grad_norm": 0.4557889997959137,
      "learning_rate": 9.914212315345857e-06,
      "loss": 0.4146,
      "step": 5091
    },
    {
      "epoch": 0.23367445275572485,
      "grad_norm": 0.46067050099372864,
      "learning_rate": 9.914167085265418e-06,
      "loss": 0.3605,
      "step": 5092
    },
    {
      "epoch": 0.2337203432609793,
      "grad_norm": 0.42339664697647095,
      "learning_rate": 9.914121843367959e-06,
      "loss": 0.3508,
      "step": 5093
    },
    {
      "epoch": 0.23376623376623376,
      "grad_norm": 0.47236618399620056,
      "learning_rate": 9.914076589653587e-06,
      "loss": 0.408,
      "step": 5094
    },
    {
      "epoch": 0.23381212427148823,
      "grad_norm": 0.46094775199890137,
      "learning_rate": 9.914031324122412e-06,
      "loss": 0.3546,
      "step": 5095
    },
    {
      "epoch": 0.2338580147767427,
      "grad_norm": 0.4532855153083801,
      "learning_rate": 9.913986046774544e-06,
      "loss": 0.3817,
      "step": 5096
    },
    {
      "epoch": 0.23390390528199714,
      "grad_norm": 0.5269132256507874,
      "learning_rate": 9.91394075761009e-06,
      "loss": 0.4936,
      "step": 5097
    },
    {
      "epoch": 0.2339497957872516,
      "grad_norm": 0.6242097616195679,
      "learning_rate": 9.913895456629156e-06,
      "loss": 0.4235,
      "step": 5098
    },
    {
      "epoch": 0.23399568629250608,
      "grad_norm": 0.43184566497802734,
      "learning_rate": 9.913850143831858e-06,
      "loss": 0.322,
      "step": 5099
    },
    {
      "epoch": 0.23404157679776055,
      "grad_norm": 0.4632769823074341,
      "learning_rate": 9.9138048192183e-06,
      "loss": 0.3936,
      "step": 5100
    },
    {
      "epoch": 0.234087467303015,
      "grad_norm": 0.5015445947647095,
      "learning_rate": 9.913759482788593e-06,
      "loss": 0.4405,
      "step": 5101
    },
    {
      "epoch": 0.23413335780826947,
      "grad_norm": 0.49522823095321655,
      "learning_rate": 9.913714134542845e-06,
      "loss": 0.4501,
      "step": 5102
    },
    {
      "epoch": 0.23417924831352394,
      "grad_norm": 0.5072798132896423,
      "learning_rate": 9.913668774481164e-06,
      "loss": 0.4807,
      "step": 5103
    },
    {
      "epoch": 0.2342251388187784,
      "grad_norm": 0.5007641315460205,
      "learning_rate": 9.913623402603663e-06,
      "loss": 0.4056,
      "step": 5104
    },
    {
      "epoch": 0.23427102932403285,
      "grad_norm": 0.4772675931453705,
      "learning_rate": 9.913578018910445e-06,
      "loss": 0.3997,
      "step": 5105
    },
    {
      "epoch": 0.23431691982928732,
      "grad_norm": 0.4053618609905243,
      "learning_rate": 9.913532623401624e-06,
      "loss": 0.2955,
      "step": 5106
    },
    {
      "epoch": 0.2343628103345418,
      "grad_norm": 0.483697772026062,
      "learning_rate": 9.91348721607731e-06,
      "loss": 0.3812,
      "step": 5107
    },
    {
      "epoch": 0.23440870083979626,
      "grad_norm": 0.5483150482177734,
      "learning_rate": 9.913441796937609e-06,
      "loss": 0.4784,
      "step": 5108
    },
    {
      "epoch": 0.2344545913450507,
      "grad_norm": 0.44203898310661316,
      "learning_rate": 9.913396365982631e-06,
      "loss": 0.3345,
      "step": 5109
    },
    {
      "epoch": 0.23450048185030517,
      "grad_norm": 0.4321436882019043,
      "learning_rate": 9.913350923212484e-06,
      "loss": 0.3443,
      "step": 5110
    },
    {
      "epoch": 0.23454637235555964,
      "grad_norm": 0.525423526763916,
      "learning_rate": 9.913305468627279e-06,
      "loss": 0.482,
      "step": 5111
    },
    {
      "epoch": 0.2345922628608141,
      "grad_norm": 0.43955475091934204,
      "learning_rate": 9.913260002227125e-06,
      "loss": 0.3484,
      "step": 5112
    },
    {
      "epoch": 0.23463815336606855,
      "grad_norm": 0.49478378891944885,
      "learning_rate": 9.913214524012134e-06,
      "loss": 0.4614,
      "step": 5113
    },
    {
      "epoch": 0.23468404387132302,
      "grad_norm": 0.5197994709014893,
      "learning_rate": 9.91316903398241e-06,
      "loss": 0.4403,
      "step": 5114
    },
    {
      "epoch": 0.2347299343765775,
      "grad_norm": 0.45656052231788635,
      "learning_rate": 9.913123532138066e-06,
      "loss": 0.3948,
      "step": 5115
    },
    {
      "epoch": 0.23477582488183196,
      "grad_norm": 0.5082331299781799,
      "learning_rate": 9.91307801847921e-06,
      "loss": 0.4409,
      "step": 5116
    },
    {
      "epoch": 0.2348217153870864,
      "grad_norm": 0.49327245354652405,
      "learning_rate": 9.913032493005953e-06,
      "loss": 0.3817,
      "step": 5117
    },
    {
      "epoch": 0.23486760589234087,
      "grad_norm": 0.4587899446487427,
      "learning_rate": 9.912986955718403e-06,
      "loss": 0.3432,
      "step": 5118
    },
    {
      "epoch": 0.23491349639759534,
      "grad_norm": 0.4807136356830597,
      "learning_rate": 9.912941406616669e-06,
      "loss": 0.4259,
      "step": 5119
    },
    {
      "epoch": 0.2349593869028498,
      "grad_norm": 0.5040963888168335,
      "learning_rate": 9.912895845700863e-06,
      "loss": 0.4801,
      "step": 5120
    },
    {
      "epoch": 0.23500527740810426,
      "grad_norm": 0.44389253854751587,
      "learning_rate": 9.912850272971093e-06,
      "loss": 0.3594,
      "step": 5121
    },
    {
      "epoch": 0.23505116791335873,
      "grad_norm": 0.4951320290565491,
      "learning_rate": 9.912804688427468e-06,
      "loss": 0.4914,
      "step": 5122
    },
    {
      "epoch": 0.2350970584186132,
      "grad_norm": 0.5539964437484741,
      "learning_rate": 9.912759092070098e-06,
      "loss": 0.5691,
      "step": 5123
    },
    {
      "epoch": 0.23514294892386764,
      "grad_norm": 0.44868677854537964,
      "learning_rate": 9.912713483899092e-06,
      "loss": 0.4059,
      "step": 5124
    },
    {
      "epoch": 0.2351888394291221,
      "grad_norm": 0.4907565414905548,
      "learning_rate": 9.912667863914562e-06,
      "loss": 0.4315,
      "step": 5125
    },
    {
      "epoch": 0.23523472993437658,
      "grad_norm": 0.47038042545318604,
      "learning_rate": 9.912622232116617e-06,
      "loss": 0.3852,
      "step": 5126
    },
    {
      "epoch": 0.23528062043963105,
      "grad_norm": 0.5518418550491333,
      "learning_rate": 9.912576588505364e-06,
      "loss": 0.2998,
      "step": 5127
    },
    {
      "epoch": 0.2353265109448855,
      "grad_norm": 0.4782343804836273,
      "learning_rate": 9.912530933080916e-06,
      "loss": 0.3463,
      "step": 5128
    },
    {
      "epoch": 0.23537240145013996,
      "grad_norm": 0.43094146251678467,
      "learning_rate": 9.912485265843382e-06,
      "loss": 0.3668,
      "step": 5129
    },
    {
      "epoch": 0.23541829195539443,
      "grad_norm": 0.45232853293418884,
      "learning_rate": 9.91243958679287e-06,
      "loss": 0.4059,
      "step": 5130
    },
    {
      "epoch": 0.2354641824606489,
      "grad_norm": 0.4581269919872284,
      "learning_rate": 9.912393895929492e-06,
      "loss": 0.4202,
      "step": 5131
    },
    {
      "epoch": 0.23551007296590334,
      "grad_norm": 0.44661420583724976,
      "learning_rate": 9.912348193253357e-06,
      "loss": 0.3487,
      "step": 5132
    },
    {
      "epoch": 0.23555596347115781,
      "grad_norm": 0.4853942096233368,
      "learning_rate": 9.912302478764574e-06,
      "loss": 0.4126,
      "step": 5133
    },
    {
      "epoch": 0.23560185397641228,
      "grad_norm": 0.4832753539085388,
      "learning_rate": 9.912256752463256e-06,
      "loss": 0.4062,
      "step": 5134
    },
    {
      "epoch": 0.23564774448166675,
      "grad_norm": 0.5102874040603638,
      "learning_rate": 9.91221101434951e-06,
      "loss": 0.5426,
      "step": 5135
    },
    {
      "epoch": 0.2356936349869212,
      "grad_norm": 0.5339062809944153,
      "learning_rate": 9.912165264423446e-06,
      "loss": 0.4436,
      "step": 5136
    },
    {
      "epoch": 0.23573952549217567,
      "grad_norm": 0.4484342634677887,
      "learning_rate": 9.912119502685175e-06,
      "loss": 0.3705,
      "step": 5137
    },
    {
      "epoch": 0.23578541599743014,
      "grad_norm": 0.4828857183456421,
      "learning_rate": 9.912073729134808e-06,
      "loss": 0.4456,
      "step": 5138
    },
    {
      "epoch": 0.2358313065026846,
      "grad_norm": 0.4640159606933594,
      "learning_rate": 9.912027943772453e-06,
      "loss": 0.3925,
      "step": 5139
    },
    {
      "epoch": 0.23587719700793905,
      "grad_norm": 0.46920710802078247,
      "learning_rate": 9.911982146598222e-06,
      "loss": 0.4533,
      "step": 5140
    },
    {
      "epoch": 0.23592308751319352,
      "grad_norm": 0.47482946515083313,
      "learning_rate": 9.911936337612223e-06,
      "loss": 0.4311,
      "step": 5141
    },
    {
      "epoch": 0.235968978018448,
      "grad_norm": 0.4582137167453766,
      "learning_rate": 9.911890516814568e-06,
      "loss": 0.4379,
      "step": 5142
    },
    {
      "epoch": 0.23601486852370246,
      "grad_norm": 0.47792595624923706,
      "learning_rate": 9.911844684205367e-06,
      "loss": 0.4349,
      "step": 5143
    },
    {
      "epoch": 0.2360607590289569,
      "grad_norm": 0.49188026785850525,
      "learning_rate": 9.911798839784729e-06,
      "loss": 0.4055,
      "step": 5144
    },
    {
      "epoch": 0.23610664953421137,
      "grad_norm": 0.45212605595588684,
      "learning_rate": 9.911752983552765e-06,
      "loss": 0.34,
      "step": 5145
    },
    {
      "epoch": 0.23615254003946584,
      "grad_norm": 0.4533120393753052,
      "learning_rate": 9.911707115509586e-06,
      "loss": 0.3739,
      "step": 5146
    },
    {
      "epoch": 0.23619843054472028,
      "grad_norm": 0.48456940054893494,
      "learning_rate": 9.911661235655302e-06,
      "loss": 0.4001,
      "step": 5147
    },
    {
      "epoch": 0.23624432104997475,
      "grad_norm": 0.46132054924964905,
      "learning_rate": 9.91161534399002e-06,
      "loss": 0.3723,
      "step": 5148
    },
    {
      "epoch": 0.23629021155522922,
      "grad_norm": 0.5226472616195679,
      "learning_rate": 9.911569440513855e-06,
      "loss": 0.4996,
      "step": 5149
    },
    {
      "epoch": 0.2363361020604837,
      "grad_norm": 0.5238263607025146,
      "learning_rate": 9.911523525226915e-06,
      "loss": 0.4161,
      "step": 5150
    },
    {
      "epoch": 0.23638199256573814,
      "grad_norm": 0.43530258536338806,
      "learning_rate": 9.911477598129312e-06,
      "loss": 0.3656,
      "step": 5151
    },
    {
      "epoch": 0.2364278830709926,
      "grad_norm": 0.4978402256965637,
      "learning_rate": 9.911431659221156e-06,
      "loss": 0.4373,
      "step": 5152
    },
    {
      "epoch": 0.23647377357624708,
      "grad_norm": 0.5485527515411377,
      "learning_rate": 9.911385708502557e-06,
      "loss": 0.4785,
      "step": 5153
    },
    {
      "epoch": 0.23651966408150155,
      "grad_norm": 0.4689266085624695,
      "learning_rate": 9.911339745973625e-06,
      "loss": 0.3464,
      "step": 5154
    },
    {
      "epoch": 0.236565554586756,
      "grad_norm": 0.46060195565223694,
      "learning_rate": 9.91129377163447e-06,
      "loss": 0.3971,
      "step": 5155
    },
    {
      "epoch": 0.23661144509201046,
      "grad_norm": 0.48276910185813904,
      "learning_rate": 9.911247785485205e-06,
      "loss": 0.4089,
      "step": 5156
    },
    {
      "epoch": 0.23665733559726493,
      "grad_norm": 0.43599194288253784,
      "learning_rate": 9.911201787525939e-06,
      "loss": 0.3701,
      "step": 5157
    },
    {
      "epoch": 0.2367032261025194,
      "grad_norm": 0.49651390314102173,
      "learning_rate": 9.911155777756782e-06,
      "loss": 0.4417,
      "step": 5158
    },
    {
      "epoch": 0.23674911660777384,
      "grad_norm": 0.44457516074180603,
      "learning_rate": 9.911109756177844e-06,
      "loss": 0.3784,
      "step": 5159
    },
    {
      "epoch": 0.2367950071130283,
      "grad_norm": 0.4829283356666565,
      "learning_rate": 9.91106372278924e-06,
      "loss": 0.4389,
      "step": 5160
    },
    {
      "epoch": 0.23684089761828278,
      "grad_norm": 0.4659006893634796,
      "learning_rate": 9.911017677591074e-06,
      "loss": 0.4041,
      "step": 5161
    },
    {
      "epoch": 0.23688678812353725,
      "grad_norm": 0.4855808913707733,
      "learning_rate": 9.910971620583466e-06,
      "loss": 0.4141,
      "step": 5162
    },
    {
      "epoch": 0.2369326786287917,
      "grad_norm": 0.4902080297470093,
      "learning_rate": 9.910925551766515e-06,
      "loss": 0.4841,
      "step": 5163
    },
    {
      "epoch": 0.23697856913404616,
      "grad_norm": 0.4838698208332062,
      "learning_rate": 9.910879471140342e-06,
      "loss": 0.4061,
      "step": 5164
    },
    {
      "epoch": 0.23702445963930063,
      "grad_norm": 0.4559089243412018,
      "learning_rate": 9.910833378705053e-06,
      "loss": 0.3783,
      "step": 5165
    },
    {
      "epoch": 0.2370703501445551,
      "grad_norm": 0.4579033851623535,
      "learning_rate": 9.910787274460758e-06,
      "loss": 0.3984,
      "step": 5166
    },
    {
      "epoch": 0.23711624064980955,
      "grad_norm": 0.4385758936405182,
      "learning_rate": 9.91074115840757e-06,
      "loss": 0.3563,
      "step": 5167
    },
    {
      "epoch": 0.23716213115506402,
      "grad_norm": 0.4908759593963623,
      "learning_rate": 9.910695030545602e-06,
      "loss": 0.4099,
      "step": 5168
    },
    {
      "epoch": 0.23720802166031849,
      "grad_norm": 0.4790913462638855,
      "learning_rate": 9.91064889087496e-06,
      "loss": 0.4249,
      "step": 5169
    },
    {
      "epoch": 0.23725391216557296,
      "grad_norm": 0.4737749397754669,
      "learning_rate": 9.910602739395758e-06,
      "loss": 0.3913,
      "step": 5170
    },
    {
      "epoch": 0.2372998026708274,
      "grad_norm": 0.450844407081604,
      "learning_rate": 9.910556576108106e-06,
      "loss": 0.4062,
      "step": 5171
    },
    {
      "epoch": 0.23734569317608187,
      "grad_norm": 0.43245989084243774,
      "learning_rate": 9.910510401012114e-06,
      "loss": 0.3329,
      "step": 5172
    },
    {
      "epoch": 0.23739158368133634,
      "grad_norm": 0.4363330602645874,
      "learning_rate": 9.910464214107897e-06,
      "loss": 0.3513,
      "step": 5173
    },
    {
      "epoch": 0.2374374741865908,
      "grad_norm": 0.42545145750045776,
      "learning_rate": 9.910418015395563e-06,
      "loss": 0.3447,
      "step": 5174
    },
    {
      "epoch": 0.23748336469184525,
      "grad_norm": 0.4734691083431244,
      "learning_rate": 9.910371804875221e-06,
      "loss": 0.3764,
      "step": 5175
    },
    {
      "epoch": 0.23752925519709972,
      "grad_norm": 0.5080201625823975,
      "learning_rate": 9.910325582546986e-06,
      "loss": 0.454,
      "step": 5176
    },
    {
      "epoch": 0.2375751457023542,
      "grad_norm": 0.5100452899932861,
      "learning_rate": 9.910279348410967e-06,
      "loss": 0.5139,
      "step": 5177
    },
    {
      "epoch": 0.23762103620760863,
      "grad_norm": 0.500399112701416,
      "learning_rate": 9.910233102467277e-06,
      "loss": 0.5094,
      "step": 5178
    },
    {
      "epoch": 0.2376669267128631,
      "grad_norm": 0.4432986378669739,
      "learning_rate": 9.910186844716025e-06,
      "loss": 0.3335,
      "step": 5179
    },
    {
      "epoch": 0.23771281721811757,
      "grad_norm": 0.5063264966011047,
      "learning_rate": 9.910140575157324e-06,
      "loss": 0.4478,
      "step": 5180
    },
    {
      "epoch": 0.23775870772337204,
      "grad_norm": 0.4564862847328186,
      "learning_rate": 9.910094293791284e-06,
      "loss": 0.3658,
      "step": 5181
    },
    {
      "epoch": 0.23780459822862648,
      "grad_norm": 0.5089392066001892,
      "learning_rate": 9.910048000618019e-06,
      "loss": 0.4701,
      "step": 5182
    },
    {
      "epoch": 0.23785048873388095,
      "grad_norm": 0.518581211566925,
      "learning_rate": 9.910001695637635e-06,
      "loss": 0.4543,
      "step": 5183
    },
    {
      "epoch": 0.23789637923913542,
      "grad_norm": 0.5212941765785217,
      "learning_rate": 9.90995537885025e-06,
      "loss": 0.5024,
      "step": 5184
    },
    {
      "epoch": 0.2379422697443899,
      "grad_norm": 0.46136966347694397,
      "learning_rate": 9.909909050255968e-06,
      "loss": 0.3411,
      "step": 5185
    },
    {
      "epoch": 0.23798816024964434,
      "grad_norm": 0.48377469182014465,
      "learning_rate": 9.909862709854907e-06,
      "loss": 0.4455,
      "step": 5186
    },
    {
      "epoch": 0.2380340507548988,
      "grad_norm": 0.5080629587173462,
      "learning_rate": 9.909816357647175e-06,
      "loss": 0.4416,
      "step": 5187
    },
    {
      "epoch": 0.23807994126015328,
      "grad_norm": 0.45300838351249695,
      "learning_rate": 9.909769993632885e-06,
      "loss": 0.406,
      "step": 5188
    },
    {
      "epoch": 0.23812583176540775,
      "grad_norm": 0.46028393507003784,
      "learning_rate": 9.909723617812148e-06,
      "loss": 0.3942,
      "step": 5189
    },
    {
      "epoch": 0.2381717222706622,
      "grad_norm": 0.44495072960853577,
      "learning_rate": 9.909677230185072e-06,
      "loss": 0.354,
      "step": 5190
    },
    {
      "epoch": 0.23821761277591666,
      "grad_norm": 0.5082409977912903,
      "learning_rate": 9.909630830751776e-06,
      "loss": 0.4953,
      "step": 5191
    },
    {
      "epoch": 0.23826350328117113,
      "grad_norm": 0.4716769754886627,
      "learning_rate": 9.909584419512366e-06,
      "loss": 0.399,
      "step": 5192
    },
    {
      "epoch": 0.2383093937864256,
      "grad_norm": 0.4795176684856415,
      "learning_rate": 9.909537996466954e-06,
      "loss": 0.3919,
      "step": 5193
    },
    {
      "epoch": 0.23835528429168004,
      "grad_norm": 0.4324151575565338,
      "learning_rate": 9.909491561615654e-06,
      "loss": 0.3471,
      "step": 5194
    },
    {
      "epoch": 0.2384011747969345,
      "grad_norm": 0.47427061200141907,
      "learning_rate": 9.909445114958577e-06,
      "loss": 0.435,
      "step": 5195
    },
    {
      "epoch": 0.23844706530218898,
      "grad_norm": 0.43681710958480835,
      "learning_rate": 9.90939865649583e-06,
      "loss": 0.3474,
      "step": 5196
    },
    {
      "epoch": 0.23849295580744345,
      "grad_norm": 0.4783264994621277,
      "learning_rate": 9.909352186227533e-06,
      "loss": 0.3484,
      "step": 5197
    },
    {
      "epoch": 0.2385388463126979,
      "grad_norm": 0.42034077644348145,
      "learning_rate": 9.90930570415379e-06,
      "loss": 0.3434,
      "step": 5198
    },
    {
      "epoch": 0.23858473681795236,
      "grad_norm": 0.444210022687912,
      "learning_rate": 9.909259210274718e-06,
      "loss": 0.362,
      "step": 5199
    },
    {
      "epoch": 0.23863062732320683,
      "grad_norm": 0.4768187403678894,
      "learning_rate": 9.909212704590428e-06,
      "loss": 0.3682,
      "step": 5200
    },
    {
      "epoch": 0.2386765178284613,
      "grad_norm": 0.45087242126464844,
      "learning_rate": 9.909166187101028e-06,
      "loss": 0.3906,
      "step": 5201
    },
    {
      "epoch": 0.23872240833371575,
      "grad_norm": 0.443838894367218,
      "learning_rate": 9.909119657806636e-06,
      "loss": 0.3508,
      "step": 5202
    },
    {
      "epoch": 0.23876829883897022,
      "grad_norm": 0.4832899272441864,
      "learning_rate": 9.909073116707358e-06,
      "loss": 0.4142,
      "step": 5203
    },
    {
      "epoch": 0.2388141893442247,
      "grad_norm": 0.45838668942451477,
      "learning_rate": 9.90902656380331e-06,
      "loss": 0.334,
      "step": 5204
    },
    {
      "epoch": 0.23886007984947916,
      "grad_norm": 0.4528806805610657,
      "learning_rate": 9.908979999094601e-06,
      "loss": 0.397,
      "step": 5205
    },
    {
      "epoch": 0.2389059703547336,
      "grad_norm": 0.5295097827911377,
      "learning_rate": 9.908933422581345e-06,
      "loss": 0.4663,
      "step": 5206
    },
    {
      "epoch": 0.23895186085998807,
      "grad_norm": 0.47119978070259094,
      "learning_rate": 9.908886834263654e-06,
      "loss": 0.3863,
      "step": 5207
    },
    {
      "epoch": 0.23899775136524254,
      "grad_norm": 0.5063113570213318,
      "learning_rate": 9.908840234141638e-06,
      "loss": 0.4808,
      "step": 5208
    },
    {
      "epoch": 0.23904364187049698,
      "grad_norm": 0.4429151117801666,
      "learning_rate": 9.908793622215414e-06,
      "loss": 0.3421,
      "step": 5209
    },
    {
      "epoch": 0.23908953237575145,
      "grad_norm": 0.4345877170562744,
      "learning_rate": 9.908746998485087e-06,
      "loss": 0.3644,
      "step": 5210
    },
    {
      "epoch": 0.23913542288100592,
      "grad_norm": 0.4474395215511322,
      "learning_rate": 9.908700362950774e-06,
      "loss": 0.3679,
      "step": 5211
    },
    {
      "epoch": 0.2391813133862604,
      "grad_norm": 0.45323947072029114,
      "learning_rate": 9.908653715612586e-06,
      "loss": 0.3864,
      "step": 5212
    },
    {
      "epoch": 0.23922720389151483,
      "grad_norm": 0.5114389061927795,
      "learning_rate": 9.908607056470636e-06,
      "loss": 0.4715,
      "step": 5213
    },
    {
      "epoch": 0.2392730943967693,
      "grad_norm": 0.5005764961242676,
      "learning_rate": 9.908560385525033e-06,
      "loss": 0.4753,
      "step": 5214
    },
    {
      "epoch": 0.23931898490202377,
      "grad_norm": 0.5477319955825806,
      "learning_rate": 9.908513702775893e-06,
      "loss": 0.4803,
      "step": 5215
    },
    {
      "epoch": 0.23936487540727824,
      "grad_norm": 0.4974982440471649,
      "learning_rate": 9.908467008223326e-06,
      "loss": 0.4279,
      "step": 5216
    },
    {
      "epoch": 0.23941076591253269,
      "grad_norm": 0.4765852987766266,
      "learning_rate": 9.908420301867444e-06,
      "loss": 0.4022,
      "step": 5217
    },
    {
      "epoch": 0.23945665641778716,
      "grad_norm": 0.47904282808303833,
      "learning_rate": 9.908373583708364e-06,
      "loss": 0.4516,
      "step": 5218
    },
    {
      "epoch": 0.23950254692304163,
      "grad_norm": 0.4584992825984955,
      "learning_rate": 9.90832685374619e-06,
      "loss": 0.3646,
      "step": 5219
    },
    {
      "epoch": 0.2395484374282961,
      "grad_norm": 0.498783677816391,
      "learning_rate": 9.908280111981044e-06,
      "loss": 0.4333,
      "step": 5220
    },
    {
      "epoch": 0.23959432793355054,
      "grad_norm": 0.4474789500236511,
      "learning_rate": 9.908233358413028e-06,
      "loss": 0.3875,
      "step": 5221
    },
    {
      "epoch": 0.239640218438805,
      "grad_norm": 0.4636237919330597,
      "learning_rate": 9.908186593042264e-06,
      "loss": 0.3933,
      "step": 5222
    },
    {
      "epoch": 0.23968610894405948,
      "grad_norm": 0.5215755105018616,
      "learning_rate": 9.90813981586886e-06,
      "loss": 0.4842,
      "step": 5223
    },
    {
      "epoch": 0.23973199944931395,
      "grad_norm": 0.44562146067619324,
      "learning_rate": 9.908093026892927e-06,
      "loss": 0.3968,
      "step": 5224
    },
    {
      "epoch": 0.2397778899545684,
      "grad_norm": 0.5041016340255737,
      "learning_rate": 9.90804622611458e-06,
      "loss": 0.4594,
      "step": 5225
    },
    {
      "epoch": 0.23982378045982286,
      "grad_norm": 0.4785310924053192,
      "learning_rate": 9.90799941353393e-06,
      "loss": 0.4668,
      "step": 5226
    },
    {
      "epoch": 0.23986967096507733,
      "grad_norm": 0.46815717220306396,
      "learning_rate": 9.907952589151093e-06,
      "loss": 0.4368,
      "step": 5227
    },
    {
      "epoch": 0.2399155614703318,
      "grad_norm": 0.5085462331771851,
      "learning_rate": 9.907905752966177e-06,
      "loss": 0.4655,
      "step": 5228
    },
    {
      "epoch": 0.23996145197558624,
      "grad_norm": 0.5794103741645813,
      "learning_rate": 9.907858904979297e-06,
      "loss": 0.4431,
      "step": 5229
    },
    {
      "epoch": 0.2400073424808407,
      "grad_norm": 0.4709904193878174,
      "learning_rate": 9.907812045190563e-06,
      "loss": 0.4685,
      "step": 5230
    },
    {
      "epoch": 0.24005323298609518,
      "grad_norm": 0.4938027560710907,
      "learning_rate": 9.907765173600093e-06,
      "loss": 0.4597,
      "step": 5231
    },
    {
      "epoch": 0.24009912349134965,
      "grad_norm": 0.44817906618118286,
      "learning_rate": 9.907718290207997e-06,
      "loss": 0.3518,
      "step": 5232
    },
    {
      "epoch": 0.2401450139966041,
      "grad_norm": 0.4672658443450928,
      "learning_rate": 9.907671395014385e-06,
      "loss": 0.4,
      "step": 5233
    },
    {
      "epoch": 0.24019090450185857,
      "grad_norm": 0.48939913511276245,
      "learning_rate": 9.907624488019375e-06,
      "loss": 0.4524,
      "step": 5234
    },
    {
      "epoch": 0.24023679500711304,
      "grad_norm": 0.43702808022499084,
      "learning_rate": 9.907577569223075e-06,
      "loss": 0.3652,
      "step": 5235
    },
    {
      "epoch": 0.24028268551236748,
      "grad_norm": 0.47192588448524475,
      "learning_rate": 9.9075306386256e-06,
      "loss": 0.4474,
      "step": 5236
    },
    {
      "epoch": 0.24032857601762195,
      "grad_norm": 0.4448605179786682,
      "learning_rate": 9.907483696227062e-06,
      "loss": 0.3433,
      "step": 5237
    },
    {
      "epoch": 0.24037446652287642,
      "grad_norm": 0.45337679982185364,
      "learning_rate": 9.907436742027577e-06,
      "loss": 0.4256,
      "step": 5238
    },
    {
      "epoch": 0.2404203570281309,
      "grad_norm": 0.47401705384254456,
      "learning_rate": 9.907389776027254e-06,
      "loss": 0.4088,
      "step": 5239
    },
    {
      "epoch": 0.24046624753338533,
      "grad_norm": 0.4362351894378662,
      "learning_rate": 9.907342798226207e-06,
      "loss": 0.3789,
      "step": 5240
    },
    {
      "epoch": 0.2405121380386398,
      "grad_norm": 0.4289877712726593,
      "learning_rate": 9.90729580862455e-06,
      "loss": 0.3608,
      "step": 5241
    },
    {
      "epoch": 0.24055802854389427,
      "grad_norm": 0.470234215259552,
      "learning_rate": 9.907248807222396e-06,
      "loss": 0.4006,
      "step": 5242
    },
    {
      "epoch": 0.24060391904914874,
      "grad_norm": 0.46503838896751404,
      "learning_rate": 9.907201794019855e-06,
      "loss": 0.4155,
      "step": 5243
    },
    {
      "epoch": 0.24064980955440318,
      "grad_norm": 0.5531212687492371,
      "learning_rate": 9.907154769017045e-06,
      "loss": 0.4964,
      "step": 5244
    },
    {
      "epoch": 0.24069570005965765,
      "grad_norm": 0.46351608633995056,
      "learning_rate": 9.907107732214075e-06,
      "loss": 0.3674,
      "step": 5245
    },
    {
      "epoch": 0.24074159056491212,
      "grad_norm": 0.470418781042099,
      "learning_rate": 9.90706068361106e-06,
      "loss": 0.3869,
      "step": 5246
    },
    {
      "epoch": 0.2407874810701666,
      "grad_norm": 0.4828798174858093,
      "learning_rate": 9.907013623208112e-06,
      "loss": 0.4118,
      "step": 5247
    },
    {
      "epoch": 0.24083337157542103,
      "grad_norm": 0.5748075246810913,
      "learning_rate": 9.906966551005348e-06,
      "loss": 0.3887,
      "step": 5248
    },
    {
      "epoch": 0.2408792620806755,
      "grad_norm": 0.446044385433197,
      "learning_rate": 9.906919467002876e-06,
      "loss": 0.3558,
      "step": 5249
    },
    {
      "epoch": 0.24092515258592997,
      "grad_norm": 0.5182888507843018,
      "learning_rate": 9.90687237120081e-06,
      "loss": 0.4417,
      "step": 5250
    },
    {
      "epoch": 0.24097104309118444,
      "grad_norm": 0.4966528117656708,
      "learning_rate": 9.906825263599267e-06,
      "loss": 0.4883,
      "step": 5251
    },
    {
      "epoch": 0.2410169335964389,
      "grad_norm": 0.43832606077194214,
      "learning_rate": 9.906778144198357e-06,
      "loss": 0.3195,
      "step": 5252
    },
    {
      "epoch": 0.24106282410169336,
      "grad_norm": 0.5073176026344299,
      "learning_rate": 9.906731012998194e-06,
      "loss": 0.5308,
      "step": 5253
    },
    {
      "epoch": 0.24110871460694783,
      "grad_norm": 0.5341662764549255,
      "learning_rate": 9.906683869998892e-06,
      "loss": 0.4406,
      "step": 5254
    },
    {
      "epoch": 0.2411546051122023,
      "grad_norm": 0.42646026611328125,
      "learning_rate": 9.906636715200563e-06,
      "loss": 0.3218,
      "step": 5255
    },
    {
      "epoch": 0.24120049561745674,
      "grad_norm": 0.46547940373420715,
      "learning_rate": 9.906589548603322e-06,
      "loss": 0.4193,
      "step": 5256
    },
    {
      "epoch": 0.2412463861227112,
      "grad_norm": 0.46058008074760437,
      "learning_rate": 9.906542370207282e-06,
      "loss": 0.416,
      "step": 5257
    },
    {
      "epoch": 0.24129227662796568,
      "grad_norm": 0.562469482421875,
      "learning_rate": 9.906495180012554e-06,
      "loss": 0.5718,
      "step": 5258
    },
    {
      "epoch": 0.24133816713322015,
      "grad_norm": 0.4656757414340973,
      "learning_rate": 9.906447978019256e-06,
      "loss": 0.3832,
      "step": 5259
    },
    {
      "epoch": 0.2413840576384746,
      "grad_norm": 0.4760354161262512,
      "learning_rate": 9.906400764227496e-06,
      "loss": 0.3877,
      "step": 5260
    },
    {
      "epoch": 0.24142994814372906,
      "grad_norm": 0.4418363869190216,
      "learning_rate": 9.906353538637394e-06,
      "loss": 0.3989,
      "step": 5261
    },
    {
      "epoch": 0.24147583864898353,
      "grad_norm": 0.504486083984375,
      "learning_rate": 9.90630630124906e-06,
      "loss": 0.4498,
      "step": 5262
    },
    {
      "epoch": 0.241521729154238,
      "grad_norm": 0.43661609292030334,
      "learning_rate": 9.906259052062607e-06,
      "loss": 0.3419,
      "step": 5263
    },
    {
      "epoch": 0.24156761965949244,
      "grad_norm": 0.49924150109291077,
      "learning_rate": 9.90621179107815e-06,
      "loss": 0.4639,
      "step": 5264
    },
    {
      "epoch": 0.24161351016474691,
      "grad_norm": 0.4788973033428192,
      "learning_rate": 9.906164518295801e-06,
      "loss": 0.4391,
      "step": 5265
    },
    {
      "epoch": 0.24165940067000138,
      "grad_norm": 0.5613639950752258,
      "learning_rate": 9.906117233715675e-06,
      "loss": 0.5097,
      "step": 5266
    },
    {
      "epoch": 0.24170529117525583,
      "grad_norm": 0.4763454496860504,
      "learning_rate": 9.906069937337886e-06,
      "loss": 0.3964,
      "step": 5267
    },
    {
      "epoch": 0.2417511816805103,
      "grad_norm": 0.4812481105327606,
      "learning_rate": 9.906022629162548e-06,
      "loss": 0.4186,
      "step": 5268
    },
    {
      "epoch": 0.24179707218576477,
      "grad_norm": 0.45494502782821655,
      "learning_rate": 9.905975309189773e-06,
      "loss": 0.3889,
      "step": 5269
    },
    {
      "epoch": 0.24184296269101924,
      "grad_norm": 0.4880490303039551,
      "learning_rate": 9.905927977419676e-06,
      "loss": 0.3954,
      "step": 5270
    },
    {
      "epoch": 0.24188885319627368,
      "grad_norm": 0.4997728765010834,
      "learning_rate": 9.905880633852369e-06,
      "loss": 0.4384,
      "step": 5271
    },
    {
      "epoch": 0.24193474370152815,
      "grad_norm": 0.46307483315467834,
      "learning_rate": 9.905833278487968e-06,
      "loss": 0.3614,
      "step": 5272
    },
    {
      "epoch": 0.24198063420678262,
      "grad_norm": 0.41685691475868225,
      "learning_rate": 9.905785911326588e-06,
      "loss": 0.3399,
      "step": 5273
    },
    {
      "epoch": 0.2420265247120371,
      "grad_norm": 0.45186159014701843,
      "learning_rate": 9.90573853236834e-06,
      "loss": 0.4201,
      "step": 5274
    },
    {
      "epoch": 0.24207241521729153,
      "grad_norm": 0.4620940387248993,
      "learning_rate": 9.905691141613339e-06,
      "loss": 0.3656,
      "step": 5275
    },
    {
      "epoch": 0.242118305722546,
      "grad_norm": 0.4886024594306946,
      "learning_rate": 9.9056437390617e-06,
      "loss": 0.4279,
      "step": 5276
    },
    {
      "epoch": 0.24216419622780047,
      "grad_norm": 0.521808922290802,
      "learning_rate": 9.905596324713536e-06,
      "loss": 0.5023,
      "step": 5277
    },
    {
      "epoch": 0.24221008673305494,
      "grad_norm": 0.46379438042640686,
      "learning_rate": 9.90554889856896e-06,
      "loss": 0.3682,
      "step": 5278
    },
    {
      "epoch": 0.24225597723830938,
      "grad_norm": 0.4650763273239136,
      "learning_rate": 9.905501460628089e-06,
      "loss": 0.3743,
      "step": 5279
    },
    {
      "epoch": 0.24230186774356385,
      "grad_norm": 0.4888210892677307,
      "learning_rate": 9.905454010891034e-06,
      "loss": 0.4759,
      "step": 5280
    },
    {
      "epoch": 0.24234775824881832,
      "grad_norm": 0.46028730273246765,
      "learning_rate": 9.905406549357909e-06,
      "loss": 0.3395,
      "step": 5281
    },
    {
      "epoch": 0.2423936487540728,
      "grad_norm": 0.49124062061309814,
      "learning_rate": 9.905359076028832e-06,
      "loss": 0.418,
      "step": 5282
    },
    {
      "epoch": 0.24243953925932724,
      "grad_norm": 0.45967069268226624,
      "learning_rate": 9.905311590903913e-06,
      "loss": 0.4075,
      "step": 5283
    },
    {
      "epoch": 0.2424854297645817,
      "grad_norm": 0.48415204882621765,
      "learning_rate": 9.905264093983267e-06,
      "loss": 0.485,
      "step": 5284
    },
    {
      "epoch": 0.24253132026983618,
      "grad_norm": 0.4569302499294281,
      "learning_rate": 9.90521658526701e-06,
      "loss": 0.3483,
      "step": 5285
    },
    {
      "epoch": 0.24257721077509065,
      "grad_norm": 0.47105672955513,
      "learning_rate": 9.905169064755255e-06,
      "loss": 0.4113,
      "step": 5286
    },
    {
      "epoch": 0.2426231012803451,
      "grad_norm": 0.45132601261138916,
      "learning_rate": 9.905121532448116e-06,
      "loss": 0.354,
      "step": 5287
    },
    {
      "epoch": 0.24266899178559956,
      "grad_norm": 0.44951263070106506,
      "learning_rate": 9.905073988345711e-06,
      "loss": 0.3862,
      "step": 5288
    },
    {
      "epoch": 0.24271488229085403,
      "grad_norm": 0.46571624279022217,
      "learning_rate": 9.905026432448148e-06,
      "loss": 0.4024,
      "step": 5289
    },
    {
      "epoch": 0.2427607727961085,
      "grad_norm": 0.46681633591651917,
      "learning_rate": 9.904978864755544e-06,
      "loss": 0.4392,
      "step": 5290
    },
    {
      "epoch": 0.24280666330136294,
      "grad_norm": 0.48999905586242676,
      "learning_rate": 9.904931285268015e-06,
      "loss": 0.3859,
      "step": 5291
    },
    {
      "epoch": 0.2428525538066174,
      "grad_norm": 0.5016316771507263,
      "learning_rate": 9.904883693985675e-06,
      "loss": 0.4552,
      "step": 5292
    },
    {
      "epoch": 0.24289844431187188,
      "grad_norm": 0.4712497889995575,
      "learning_rate": 9.904836090908638e-06,
      "loss": 0.4361,
      "step": 5293
    },
    {
      "epoch": 0.24294433481712632,
      "grad_norm": 0.4656883478164673,
      "learning_rate": 9.904788476037017e-06,
      "loss": 0.3875,
      "step": 5294
    },
    {
      "epoch": 0.2429902253223808,
      "grad_norm": 0.4602237045764923,
      "learning_rate": 9.904740849370929e-06,
      "loss": 0.4351,
      "step": 5295
    },
    {
      "epoch": 0.24303611582763526,
      "grad_norm": 0.4638361930847168,
      "learning_rate": 9.904693210910485e-06,
      "loss": 0.4131,
      "step": 5296
    },
    {
      "epoch": 0.24308200633288973,
      "grad_norm": 0.4269849359989166,
      "learning_rate": 9.904645560655804e-06,
      "loss": 0.3709,
      "step": 5297
    },
    {
      "epoch": 0.24312789683814418,
      "grad_norm": 0.46226897835731506,
      "learning_rate": 9.904597898606996e-06,
      "loss": 0.3598,
      "step": 5298
    },
    {
      "epoch": 0.24317378734339865,
      "grad_norm": 0.49389010667800903,
      "learning_rate": 9.90455022476418e-06,
      "loss": 0.417,
      "step": 5299
    },
    {
      "epoch": 0.24321967784865312,
      "grad_norm": 0.44024893641471863,
      "learning_rate": 9.904502539127469e-06,
      "loss": 0.3864,
      "step": 5300
    },
    {
      "epoch": 0.24326556835390759,
      "grad_norm": 0.44661688804626465,
      "learning_rate": 9.904454841696976e-06,
      "loss": 0.3553,
      "step": 5301
    },
    {
      "epoch": 0.24331145885916203,
      "grad_norm": 0.4814205467700958,
      "learning_rate": 9.904407132472818e-06,
      "loss": 0.4291,
      "step": 5302
    },
    {
      "epoch": 0.2433573493644165,
      "grad_norm": 0.42537546157836914,
      "learning_rate": 9.904359411455108e-06,
      "loss": 0.3558,
      "step": 5303
    },
    {
      "epoch": 0.24340323986967097,
      "grad_norm": 0.48626184463500977,
      "learning_rate": 9.90431167864396e-06,
      "loss": 0.3951,
      "step": 5304
    },
    {
      "epoch": 0.24344913037492544,
      "grad_norm": 0.4893483817577362,
      "learning_rate": 9.904263934039494e-06,
      "loss": 0.4642,
      "step": 5305
    },
    {
      "epoch": 0.24349502088017988,
      "grad_norm": 0.45457032322883606,
      "learning_rate": 9.90421617764182e-06,
      "loss": 0.3421,
      "step": 5306
    },
    {
      "epoch": 0.24354091138543435,
      "grad_norm": 0.48694470524787903,
      "learning_rate": 9.904168409451051e-06,
      "loss": 0.4423,
      "step": 5307
    },
    {
      "epoch": 0.24358680189068882,
      "grad_norm": 0.48867183923721313,
      "learning_rate": 9.904120629467307e-06,
      "loss": 0.4286,
      "step": 5308
    },
    {
      "epoch": 0.2436326923959433,
      "grad_norm": 0.42153680324554443,
      "learning_rate": 9.9040728376907e-06,
      "loss": 0.3241,
      "step": 5309
    },
    {
      "epoch": 0.24367858290119773,
      "grad_norm": 0.45671263337135315,
      "learning_rate": 9.904025034121345e-06,
      "loss": 0.4572,
      "step": 5310
    },
    {
      "epoch": 0.2437244734064522,
      "grad_norm": 0.49713119864463806,
      "learning_rate": 9.90397721875936e-06,
      "loss": 0.4098,
      "step": 5311
    },
    {
      "epoch": 0.24377036391170667,
      "grad_norm": 0.4736146330833435,
      "learning_rate": 9.903929391604856e-06,
      "loss": 0.3934,
      "step": 5312
    },
    {
      "epoch": 0.24381625441696114,
      "grad_norm": 0.47610053420066833,
      "learning_rate": 9.903881552657949e-06,
      "loss": 0.438,
      "step": 5313
    },
    {
      "epoch": 0.24386214492221558,
      "grad_norm": 0.47151419520378113,
      "learning_rate": 9.903833701918755e-06,
      "loss": 0.3813,
      "step": 5314
    },
    {
      "epoch": 0.24390803542747005,
      "grad_norm": 0.5233833193778992,
      "learning_rate": 9.903785839387388e-06,
      "loss": 0.4563,
      "step": 5315
    },
    {
      "epoch": 0.24395392593272452,
      "grad_norm": 0.4568115770816803,
      "learning_rate": 9.903737965063966e-06,
      "loss": 0.426,
      "step": 5316
    },
    {
      "epoch": 0.243999816437979,
      "grad_norm": 0.4647948145866394,
      "learning_rate": 9.9036900789486e-06,
      "loss": 0.3909,
      "step": 5317
    },
    {
      "epoch": 0.24404570694323344,
      "grad_norm": 0.4955919086933136,
      "learning_rate": 9.903642181041406e-06,
      "loss": 0.5079,
      "step": 5318
    },
    {
      "epoch": 0.2440915974484879,
      "grad_norm": 0.49967601895332336,
      "learning_rate": 9.903594271342503e-06,
      "loss": 0.4424,
      "step": 5319
    },
    {
      "epoch": 0.24413748795374238,
      "grad_norm": 0.5169775485992432,
      "learning_rate": 9.903546349852002e-06,
      "loss": 0.4342,
      "step": 5320
    },
    {
      "epoch": 0.24418337845899685,
      "grad_norm": 0.4782702922821045,
      "learning_rate": 9.903498416570018e-06,
      "loss": 0.4022,
      "step": 5321
    },
    {
      "epoch": 0.2442292689642513,
      "grad_norm": 0.45459210872650146,
      "learning_rate": 9.903450471496671e-06,
      "loss": 0.4304,
      "step": 5322
    },
    {
      "epoch": 0.24427515946950576,
      "grad_norm": 0.478246808052063,
      "learning_rate": 9.903402514632072e-06,
      "loss": 0.4448,
      "step": 5323
    },
    {
      "epoch": 0.24432104997476023,
      "grad_norm": 0.44923147559165955,
      "learning_rate": 9.903354545976336e-06,
      "loss": 0.3719,
      "step": 5324
    },
    {
      "epoch": 0.24436694048001467,
      "grad_norm": 0.48403874039649963,
      "learning_rate": 9.90330656552958e-06,
      "loss": 0.438,
      "step": 5325
    },
    {
      "epoch": 0.24441283098526914,
      "grad_norm": 0.49797770380973816,
      "learning_rate": 9.90325857329192e-06,
      "loss": 0.5376,
      "step": 5326
    },
    {
      "epoch": 0.2444587214905236,
      "grad_norm": 0.4303591251373291,
      "learning_rate": 9.903210569263472e-06,
      "loss": 0.3172,
      "step": 5327
    },
    {
      "epoch": 0.24450461199577808,
      "grad_norm": 0.5149800777435303,
      "learning_rate": 9.90316255344435e-06,
      "loss": 0.5387,
      "step": 5328
    },
    {
      "epoch": 0.24455050250103252,
      "grad_norm": 0.46345871686935425,
      "learning_rate": 9.903114525834667e-06,
      "loss": 0.3917,
      "step": 5329
    },
    {
      "epoch": 0.244596393006287,
      "grad_norm": 0.4298897385597229,
      "learning_rate": 9.903066486434542e-06,
      "loss": 0.3343,
      "step": 5330
    },
    {
      "epoch": 0.24464228351154146,
      "grad_norm": 0.4684180021286011,
      "learning_rate": 9.903018435244091e-06,
      "loss": 0.4486,
      "step": 5331
    },
    {
      "epoch": 0.24468817401679593,
      "grad_norm": 0.5151140093803406,
      "learning_rate": 9.902970372263428e-06,
      "loss": 0.5135,
      "step": 5332
    },
    {
      "epoch": 0.24473406452205038,
      "grad_norm": 0.4247035086154938,
      "learning_rate": 9.902922297492667e-06,
      "loss": 0.3458,
      "step": 5333
    },
    {
      "epoch": 0.24477995502730485,
      "grad_norm": 0.44648730754852295,
      "learning_rate": 9.902874210931928e-06,
      "loss": 0.3905,
      "step": 5334
    },
    {
      "epoch": 0.24482584553255932,
      "grad_norm": 0.4940642714500427,
      "learning_rate": 9.902826112581321e-06,
      "loss": 0.4026,
      "step": 5335
    },
    {
      "epoch": 0.2448717360378138,
      "grad_norm": 0.5639809370040894,
      "learning_rate": 9.902778002440965e-06,
      "loss": 0.4083,
      "step": 5336
    },
    {
      "epoch": 0.24491762654306823,
      "grad_norm": 0.43036025762557983,
      "learning_rate": 9.902729880510978e-06,
      "loss": 0.3583,
      "step": 5337
    },
    {
      "epoch": 0.2449635170483227,
      "grad_norm": 0.460487425327301,
      "learning_rate": 9.902681746791469e-06,
      "loss": 0.3922,
      "step": 5338
    },
    {
      "epoch": 0.24500940755357717,
      "grad_norm": 0.48581740260124207,
      "learning_rate": 9.902633601282561e-06,
      "loss": 0.4569,
      "step": 5339
    },
    {
      "epoch": 0.24505529805883164,
      "grad_norm": 0.493158757686615,
      "learning_rate": 9.902585443984365e-06,
      "loss": 0.4104,
      "step": 5340
    },
    {
      "epoch": 0.24510118856408608,
      "grad_norm": 0.44886279106140137,
      "learning_rate": 9.902537274896998e-06,
      "loss": 0.4001,
      "step": 5341
    },
    {
      "epoch": 0.24514707906934055,
      "grad_norm": 0.4597589671611786,
      "learning_rate": 9.902489094020576e-06,
      "loss": 0.4364,
      "step": 5342
    },
    {
      "epoch": 0.24519296957459502,
      "grad_norm": 0.45080989599227905,
      "learning_rate": 9.902440901355216e-06,
      "loss": 0.417,
      "step": 5343
    },
    {
      "epoch": 0.2452388600798495,
      "grad_norm": 0.4921380877494812,
      "learning_rate": 9.902392696901031e-06,
      "loss": 0.4396,
      "step": 5344
    },
    {
      "epoch": 0.24528475058510393,
      "grad_norm": 0.4710673689842224,
      "learning_rate": 9.902344480658142e-06,
      "loss": 0.4282,
      "step": 5345
    },
    {
      "epoch": 0.2453306410903584,
      "grad_norm": 0.4434351325035095,
      "learning_rate": 9.902296252626659e-06,
      "loss": 0.3867,
      "step": 5346
    },
    {
      "epoch": 0.24537653159561287,
      "grad_norm": 0.5592572093009949,
      "learning_rate": 9.902248012806702e-06,
      "loss": 0.6233,
      "step": 5347
    },
    {
      "epoch": 0.24542242210086734,
      "grad_norm": 0.4374180734157562,
      "learning_rate": 9.902199761198386e-06,
      "loss": 0.3276,
      "step": 5348
    },
    {
      "epoch": 0.24546831260612179,
      "grad_norm": 0.4640013575553894,
      "learning_rate": 9.902151497801826e-06,
      "loss": 0.3984,
      "step": 5349
    },
    {
      "epoch": 0.24551420311137626,
      "grad_norm": 0.6440848708152771,
      "learning_rate": 9.90210322261714e-06,
      "loss": 0.5029,
      "step": 5350
    },
    {
      "epoch": 0.24556009361663073,
      "grad_norm": 0.43278902769088745,
      "learning_rate": 9.90205493564444e-06,
      "loss": 0.3086,
      "step": 5351
    },
    {
      "epoch": 0.2456059841218852,
      "grad_norm": 0.5207533240318298,
      "learning_rate": 9.902006636883847e-06,
      "loss": 0.5387,
      "step": 5352
    },
    {
      "epoch": 0.24565187462713964,
      "grad_norm": 0.48598361015319824,
      "learning_rate": 9.901958326335475e-06,
      "loss": 0.4106,
      "step": 5353
    },
    {
      "epoch": 0.2456977651323941,
      "grad_norm": 0.5017309188842773,
      "learning_rate": 9.901910003999442e-06,
      "loss": 0.4588,
      "step": 5354
    },
    {
      "epoch": 0.24574365563764858,
      "grad_norm": 0.4472849667072296,
      "learning_rate": 9.901861669875861e-06,
      "loss": 0.3667,
      "step": 5355
    },
    {
      "epoch": 0.24578954614290302,
      "grad_norm": 0.5211498737335205,
      "learning_rate": 9.901813323964848e-06,
      "loss": 0.5066,
      "step": 5356
    },
    {
      "epoch": 0.2458354366481575,
      "grad_norm": 0.4777042865753174,
      "learning_rate": 9.901764966266523e-06,
      "loss": 0.4603,
      "step": 5357
    },
    {
      "epoch": 0.24588132715341196,
      "grad_norm": 0.4984428584575653,
      "learning_rate": 9.901716596781001e-06,
      "loss": 0.4901,
      "step": 5358
    },
    {
      "epoch": 0.24592721765866643,
      "grad_norm": 0.45887160301208496,
      "learning_rate": 9.901668215508397e-06,
      "loss": 0.3685,
      "step": 5359
    },
    {
      "epoch": 0.24597310816392087,
      "grad_norm": 0.46002840995788574,
      "learning_rate": 9.901619822448826e-06,
      "loss": 0.4,
      "step": 5360
    },
    {
      "epoch": 0.24601899866917534,
      "grad_norm": 0.5057162642478943,
      "learning_rate": 9.90157141760241e-06,
      "loss": 0.4418,
      "step": 5361
    },
    {
      "epoch": 0.2460648891744298,
      "grad_norm": 0.739828884601593,
      "learning_rate": 9.90152300096926e-06,
      "loss": 0.4511,
      "step": 5362
    },
    {
      "epoch": 0.24611077967968428,
      "grad_norm": 0.43112438917160034,
      "learning_rate": 9.901474572549492e-06,
      "loss": 0.3474,
      "step": 5363
    },
    {
      "epoch": 0.24615667018493873,
      "grad_norm": 0.490246444940567,
      "learning_rate": 9.901426132343227e-06,
      "loss": 0.3951,
      "step": 5364
    },
    {
      "epoch": 0.2462025606901932,
      "grad_norm": 0.4832803010940552,
      "learning_rate": 9.901377680350579e-06,
      "loss": 0.4264,
      "step": 5365
    },
    {
      "epoch": 0.24624845119544767,
      "grad_norm": 0.46472403407096863,
      "learning_rate": 9.901329216571663e-06,
      "loss": 0.4087,
      "step": 5366
    },
    {
      "epoch": 0.24629434170070214,
      "grad_norm": 0.4616561830043793,
      "learning_rate": 9.901280741006597e-06,
      "loss": 0.392,
      "step": 5367
    },
    {
      "epoch": 0.24634023220595658,
      "grad_norm": 0.48619821667671204,
      "learning_rate": 9.9012322536555e-06,
      "loss": 0.4162,
      "step": 5368
    },
    {
      "epoch": 0.24638612271121105,
      "grad_norm": 0.46516546607017517,
      "learning_rate": 9.901183754518487e-06,
      "loss": 0.3789,
      "step": 5369
    },
    {
      "epoch": 0.24643201321646552,
      "grad_norm": 0.4606388509273529,
      "learning_rate": 9.90113524359567e-06,
      "loss": 0.3692,
      "step": 5370
    },
    {
      "epoch": 0.24647790372172,
      "grad_norm": 0.49570757150650024,
      "learning_rate": 9.901086720887172e-06,
      "loss": 0.4847,
      "step": 5371
    },
    {
      "epoch": 0.24652379422697443,
      "grad_norm": 0.4897167682647705,
      "learning_rate": 9.901038186393108e-06,
      "loss": 0.3975,
      "step": 5372
    },
    {
      "epoch": 0.2465696847322289,
      "grad_norm": 0.5077126622200012,
      "learning_rate": 9.900989640113592e-06,
      "loss": 0.524,
      "step": 5373
    },
    {
      "epoch": 0.24661557523748337,
      "grad_norm": 0.45957979559898376,
      "learning_rate": 9.900941082048743e-06,
      "loss": 0.3744,
      "step": 5374
    },
    {
      "epoch": 0.24666146574273784,
      "grad_norm": 0.41891229152679443,
      "learning_rate": 9.900892512198679e-06,
      "loss": 0.3375,
      "step": 5375
    },
    {
      "epoch": 0.24670735624799228,
      "grad_norm": 0.46823734045028687,
      "learning_rate": 9.900843930563516e-06,
      "loss": 0.4704,
      "step": 5376
    },
    {
      "epoch": 0.24675324675324675,
      "grad_norm": 0.46074435114860535,
      "learning_rate": 9.900795337143367e-06,
      "loss": 0.3884,
      "step": 5377
    },
    {
      "epoch": 0.24679913725850122,
      "grad_norm": 0.4602743685245514,
      "learning_rate": 9.900746731938353e-06,
      "loss": 0.359,
      "step": 5378
    },
    {
      "epoch": 0.2468450277637557,
      "grad_norm": 0.43587926030158997,
      "learning_rate": 9.900698114948591e-06,
      "loss": 0.3424,
      "step": 5379
    },
    {
      "epoch": 0.24689091826901013,
      "grad_norm": 0.4460495114326477,
      "learning_rate": 9.900649486174196e-06,
      "loss": 0.3973,
      "step": 5380
    },
    {
      "epoch": 0.2469368087742646,
      "grad_norm": 0.42193976044654846,
      "learning_rate": 9.900600845615287e-06,
      "loss": 0.2964,
      "step": 5381
    },
    {
      "epoch": 0.24698269927951907,
      "grad_norm": 0.5385944247245789,
      "learning_rate": 9.900552193271977e-06,
      "loss": 0.5448,
      "step": 5382
    },
    {
      "epoch": 0.24702858978477352,
      "grad_norm": 0.43305185437202454,
      "learning_rate": 9.900503529144387e-06,
      "loss": 0.3894,
      "step": 5383
    },
    {
      "epoch": 0.247074480290028,
      "grad_norm": 0.44950735569000244,
      "learning_rate": 9.900454853232632e-06,
      "loss": 0.3361,
      "step": 5384
    },
    {
      "epoch": 0.24712037079528246,
      "grad_norm": 0.4558485746383667,
      "learning_rate": 9.90040616553683e-06,
      "loss": 0.4135,
      "step": 5385
    },
    {
      "epoch": 0.24716626130053693,
      "grad_norm": 0.4376548230648041,
      "learning_rate": 9.9003574660571e-06,
      "loss": 0.3542,
      "step": 5386
    },
    {
      "epoch": 0.24721215180579137,
      "grad_norm": 0.45413148403167725,
      "learning_rate": 9.900308754793556e-06,
      "loss": 0.3734,
      "step": 5387
    },
    {
      "epoch": 0.24725804231104584,
      "grad_norm": 0.4224342703819275,
      "learning_rate": 9.900260031746313e-06,
      "loss": 0.3484,
      "step": 5388
    },
    {
      "epoch": 0.2473039328163003,
      "grad_norm": 0.46938297152519226,
      "learning_rate": 9.900211296915494e-06,
      "loss": 0.3974,
      "step": 5389
    },
    {
      "epoch": 0.24734982332155478,
      "grad_norm": 0.5534500479698181,
      "learning_rate": 9.900162550301212e-06,
      "loss": 0.5464,
      "step": 5390
    },
    {
      "epoch": 0.24739571382680922,
      "grad_norm": 0.4902797341346741,
      "learning_rate": 9.900113791903587e-06,
      "loss": 0.4971,
      "step": 5391
    },
    {
      "epoch": 0.2474416043320637,
      "grad_norm": 0.4419653117656708,
      "learning_rate": 9.900065021722734e-06,
      "loss": 0.3469,
      "step": 5392
    },
    {
      "epoch": 0.24748749483731816,
      "grad_norm": 0.4650265574455261,
      "learning_rate": 9.900016239758771e-06,
      "loss": 0.3608,
      "step": 5393
    },
    {
      "epoch": 0.24753338534257263,
      "grad_norm": 0.5291844606399536,
      "learning_rate": 9.899967446011815e-06,
      "loss": 0.4684,
      "step": 5394
    },
    {
      "epoch": 0.24757927584782707,
      "grad_norm": 0.42976513504981995,
      "learning_rate": 9.899918640481985e-06,
      "loss": 0.3599,
      "step": 5395
    },
    {
      "epoch": 0.24762516635308154,
      "grad_norm": 0.46596038341522217,
      "learning_rate": 9.899869823169397e-06,
      "loss": 0.4153,
      "step": 5396
    },
    {
      "epoch": 0.24767105685833601,
      "grad_norm": 0.48037806153297424,
      "learning_rate": 9.899820994074168e-06,
      "loss": 0.4672,
      "step": 5397
    },
    {
      "epoch": 0.24771694736359048,
      "grad_norm": 0.4876294732093811,
      "learning_rate": 9.899772153196415e-06,
      "loss": 0.4749,
      "step": 5398
    },
    {
      "epoch": 0.24776283786884493,
      "grad_norm": 0.4756849408149719,
      "learning_rate": 9.899723300536257e-06,
      "loss": 0.4704,
      "step": 5399
    },
    {
      "epoch": 0.2478087283740994,
      "grad_norm": 0.5220165848731995,
      "learning_rate": 9.899674436093812e-06,
      "loss": 0.538,
      "step": 5400
    },
    {
      "epoch": 0.24785461887935387,
      "grad_norm": 0.4860324263572693,
      "learning_rate": 9.899625559869194e-06,
      "loss": 0.4719,
      "step": 5401
    },
    {
      "epoch": 0.24790050938460834,
      "grad_norm": 0.45706814527511597,
      "learning_rate": 9.899576671862526e-06,
      "loss": 0.3734,
      "step": 5402
    },
    {
      "epoch": 0.24794639988986278,
      "grad_norm": 0.4765598773956299,
      "learning_rate": 9.89952777207392e-06,
      "loss": 0.4416,
      "step": 5403
    },
    {
      "epoch": 0.24799229039511725,
      "grad_norm": 0.46519675850868225,
      "learning_rate": 9.899478860503497e-06,
      "loss": 0.3998,
      "step": 5404
    },
    {
      "epoch": 0.24803818090037172,
      "grad_norm": 0.4826446771621704,
      "learning_rate": 9.899429937151374e-06,
      "loss": 0.4689,
      "step": 5405
    },
    {
      "epoch": 0.2480840714056262,
      "grad_norm": 0.46294093132019043,
      "learning_rate": 9.89938100201767e-06,
      "loss": 0.4155,
      "step": 5406
    },
    {
      "epoch": 0.24812996191088063,
      "grad_norm": 0.44682589173316956,
      "learning_rate": 9.899332055102498e-06,
      "loss": 0.313,
      "step": 5407
    },
    {
      "epoch": 0.2481758524161351,
      "grad_norm": 0.4746490716934204,
      "learning_rate": 9.899283096405979e-06,
      "loss": 0.4808,
      "step": 5408
    },
    {
      "epoch": 0.24822174292138957,
      "grad_norm": 0.4911626875400543,
      "learning_rate": 9.899234125928232e-06,
      "loss": 0.4485,
      "step": 5409
    },
    {
      "epoch": 0.24826763342664404,
      "grad_norm": 0.48363322019577026,
      "learning_rate": 9.899185143669372e-06,
      "loss": 0.4176,
      "step": 5410
    },
    {
      "epoch": 0.24831352393189848,
      "grad_norm": 0.49399086833000183,
      "learning_rate": 9.899136149629519e-06,
      "loss": 0.3879,
      "step": 5411
    },
    {
      "epoch": 0.24835941443715295,
      "grad_norm": 0.5003026723861694,
      "learning_rate": 9.899087143808787e-06,
      "loss": 0.4122,
      "step": 5412
    },
    {
      "epoch": 0.24840530494240742,
      "grad_norm": 0.43921470642089844,
      "learning_rate": 9.8990381262073e-06,
      "loss": 0.3523,
      "step": 5413
    },
    {
      "epoch": 0.24845119544766187,
      "grad_norm": 0.5052942037582397,
      "learning_rate": 9.898989096825171e-06,
      "loss": 0.4764,
      "step": 5414
    },
    {
      "epoch": 0.24849708595291634,
      "grad_norm": 0.4983556866645813,
      "learning_rate": 9.898940055662518e-06,
      "loss": 0.5089,
      "step": 5415
    },
    {
      "epoch": 0.2485429764581708,
      "grad_norm": 0.5024328231811523,
      "learning_rate": 9.898891002719462e-06,
      "loss": 0.5261,
      "step": 5416
    },
    {
      "epoch": 0.24858886696342528,
      "grad_norm": 0.4489428699016571,
      "learning_rate": 9.89884193799612e-06,
      "loss": 0.376,
      "step": 5417
    },
    {
      "epoch": 0.24863475746867972,
      "grad_norm": 0.4745553433895111,
      "learning_rate": 9.898792861492608e-06,
      "loss": 0.436,
      "step": 5418
    },
    {
      "epoch": 0.2486806479739342,
      "grad_norm": 0.43664437532424927,
      "learning_rate": 9.898743773209045e-06,
      "loss": 0.3235,
      "step": 5419
    },
    {
      "epoch": 0.24872653847918866,
      "grad_norm": 0.4805891811847687,
      "learning_rate": 9.898694673145551e-06,
      "loss": 0.4153,
      "step": 5420
    },
    {
      "epoch": 0.24877242898444313,
      "grad_norm": 0.4661649465560913,
      "learning_rate": 9.89864556130224e-06,
      "loss": 0.386,
      "step": 5421
    },
    {
      "epoch": 0.24881831948969757,
      "grad_norm": 0.46313953399658203,
      "learning_rate": 9.898596437679233e-06,
      "loss": 0.433,
      "step": 5422
    },
    {
      "epoch": 0.24886420999495204,
      "grad_norm": 0.46011000871658325,
      "learning_rate": 9.898547302276649e-06,
      "loss": 0.4261,
      "step": 5423
    },
    {
      "epoch": 0.2489101005002065,
      "grad_norm": 0.4639430344104767,
      "learning_rate": 9.898498155094605e-06,
      "loss": 0.3763,
      "step": 5424
    },
    {
      "epoch": 0.24895599100546098,
      "grad_norm": 0.4886471629142761,
      "learning_rate": 9.898448996133217e-06,
      "loss": 0.4665,
      "step": 5425
    },
    {
      "epoch": 0.24900188151071542,
      "grad_norm": 0.4836229085922241,
      "learning_rate": 9.898399825392606e-06,
      "loss": 0.401,
      "step": 5426
    },
    {
      "epoch": 0.2490477720159699,
      "grad_norm": 0.4913848042488098,
      "learning_rate": 9.89835064287289e-06,
      "loss": 0.449,
      "step": 5427
    },
    {
      "epoch": 0.24909366252122436,
      "grad_norm": 0.45920923352241516,
      "learning_rate": 9.898301448574186e-06,
      "loss": 0.368,
      "step": 5428
    },
    {
      "epoch": 0.24913955302647883,
      "grad_norm": 0.38684335350990295,
      "learning_rate": 9.898252242496614e-06,
      "loss": 0.2848,
      "step": 5429
    },
    {
      "epoch": 0.24918544353173327,
      "grad_norm": 0.44047191739082336,
      "learning_rate": 9.898203024640291e-06,
      "loss": 0.3122,
      "step": 5430
    },
    {
      "epoch": 0.24923133403698775,
      "grad_norm": 0.5357457995414734,
      "learning_rate": 9.898153795005333e-06,
      "loss": 0.5125,
      "step": 5431
    },
    {
      "epoch": 0.24927722454224222,
      "grad_norm": 0.483256459236145,
      "learning_rate": 9.898104553591866e-06,
      "loss": 0.4326,
      "step": 5432
    },
    {
      "epoch": 0.24932311504749669,
      "grad_norm": 0.4657694399356842,
      "learning_rate": 9.898055300399999e-06,
      "loss": 0.4537,
      "step": 5433
    },
    {
      "epoch": 0.24936900555275113,
      "grad_norm": 0.4349009096622467,
      "learning_rate": 9.898006035429857e-06,
      "loss": 0.3362,
      "step": 5434
    },
    {
      "epoch": 0.2494148960580056,
      "grad_norm": 0.4730532467365265,
      "learning_rate": 9.897956758681555e-06,
      "loss": 0.4016,
      "step": 5435
    },
    {
      "epoch": 0.24946078656326007,
      "grad_norm": 0.45093175768852234,
      "learning_rate": 9.897907470155215e-06,
      "loss": 0.3748,
      "step": 5436
    },
    {
      "epoch": 0.24950667706851454,
      "grad_norm": 0.4878894090652466,
      "learning_rate": 9.897858169850952e-06,
      "loss": 0.4483,
      "step": 5437
    },
    {
      "epoch": 0.24955256757376898,
      "grad_norm": 0.441736102104187,
      "learning_rate": 9.897808857768885e-06,
      "loss": 0.3487,
      "step": 5438
    },
    {
      "epoch": 0.24959845807902345,
      "grad_norm": 0.4571661949157715,
      "learning_rate": 9.897759533909134e-06,
      "loss": 0.3553,
      "step": 5439
    },
    {
      "epoch": 0.24964434858427792,
      "grad_norm": 0.47505679726600647,
      "learning_rate": 9.897710198271818e-06,
      "loss": 0.4569,
      "step": 5440
    },
    {
      "epoch": 0.24969023908953236,
      "grad_norm": 0.4656061828136444,
      "learning_rate": 9.897660850857054e-06,
      "loss": 0.4285,
      "step": 5441
    },
    {
      "epoch": 0.24973612959478683,
      "grad_norm": 0.4578312337398529,
      "learning_rate": 9.897611491664962e-06,
      "loss": 0.3531,
      "step": 5442
    },
    {
      "epoch": 0.2497820201000413,
      "grad_norm": 0.44951313734054565,
      "learning_rate": 9.89756212069566e-06,
      "loss": 0.3796,
      "step": 5443
    },
    {
      "epoch": 0.24982791060529577,
      "grad_norm": 0.5055868625640869,
      "learning_rate": 9.897512737949267e-06,
      "loss": 0.4526,
      "step": 5444
    },
    {
      "epoch": 0.24987380111055021,
      "grad_norm": 0.48434147238731384,
      "learning_rate": 9.8974633434259e-06,
      "loss": 0.4565,
      "step": 5445
    },
    {
      "epoch": 0.24991969161580468,
      "grad_norm": 0.49553877115249634,
      "learning_rate": 9.897413937125682e-06,
      "loss": 0.3681,
      "step": 5446
    },
    {
      "epoch": 0.24996558212105915,
      "grad_norm": 0.5233808159828186,
      "learning_rate": 9.897364519048726e-06,
      "loss": 0.3203,
      "step": 5447
    },
    {
      "epoch": 0.2500114726263136,
      "grad_norm": 0.47207024693489075,
      "learning_rate": 9.897315089195156e-06,
      "loss": 0.4163,
      "step": 5448
    },
    {
      "epoch": 0.25005736313156807,
      "grad_norm": 0.467523992061615,
      "learning_rate": 9.89726564756509e-06,
      "loss": 0.4103,
      "step": 5449
    },
    {
      "epoch": 0.25010325363682256,
      "grad_norm": 0.5478648543357849,
      "learning_rate": 9.897216194158643e-06,
      "loss": 0.4499,
      "step": 5450
    },
    {
      "epoch": 0.250149144142077,
      "grad_norm": 0.4833225607872009,
      "learning_rate": 9.897166728975937e-06,
      "loss": 0.4268,
      "step": 5451
    },
    {
      "epoch": 0.25019503464733145,
      "grad_norm": 0.4996795952320099,
      "learning_rate": 9.897117252017091e-06,
      "loss": 0.4351,
      "step": 5452
    },
    {
      "epoch": 0.25024092515258595,
      "grad_norm": 0.4672998785972595,
      "learning_rate": 9.897067763282226e-06,
      "loss": 0.3898,
      "step": 5453
    },
    {
      "epoch": 0.2502868156578404,
      "grad_norm": 0.4596433639526367,
      "learning_rate": 9.897018262771456e-06,
      "loss": 0.3618,
      "step": 5454
    },
    {
      "epoch": 0.25033270616309483,
      "grad_norm": 0.44598808884620667,
      "learning_rate": 9.896968750484902e-06,
      "loss": 0.3583,
      "step": 5455
    },
    {
      "epoch": 0.25037859666834933,
      "grad_norm": 0.48249292373657227,
      "learning_rate": 9.896919226422685e-06,
      "loss": 0.4144,
      "step": 5456
    },
    {
      "epoch": 0.25042448717360377,
      "grad_norm": 0.4608473479747772,
      "learning_rate": 9.896869690584924e-06,
      "loss": 0.415,
      "step": 5457
    },
    {
      "epoch": 0.25047037767885827,
      "grad_norm": 0.5003462433815002,
      "learning_rate": 9.896820142971734e-06,
      "loss": 0.4949,
      "step": 5458
    },
    {
      "epoch": 0.2505162681841127,
      "grad_norm": 0.4707450568675995,
      "learning_rate": 9.89677058358324e-06,
      "loss": 0.4355,
      "step": 5459
    },
    {
      "epoch": 0.25056215868936715,
      "grad_norm": 0.47938671708106995,
      "learning_rate": 9.896721012419557e-06,
      "loss": 0.4784,
      "step": 5460
    },
    {
      "epoch": 0.25060804919462165,
      "grad_norm": 0.49174219369888306,
      "learning_rate": 9.896671429480806e-06,
      "loss": 0.5469,
      "step": 5461
    },
    {
      "epoch": 0.2506539396998761,
      "grad_norm": 0.4595566391944885,
      "learning_rate": 9.896621834767104e-06,
      "loss": 0.3622,
      "step": 5462
    },
    {
      "epoch": 0.25069983020513054,
      "grad_norm": 0.5035466551780701,
      "learning_rate": 9.896572228278574e-06,
      "loss": 0.4716,
      "step": 5463
    },
    {
      "epoch": 0.25074572071038503,
      "grad_norm": 0.4592345952987671,
      "learning_rate": 9.896522610015332e-06,
      "loss": 0.3313,
      "step": 5464
    },
    {
      "epoch": 0.2507916112156395,
      "grad_norm": 0.4709053635597229,
      "learning_rate": 9.896472979977498e-06,
      "loss": 0.3501,
      "step": 5465
    },
    {
      "epoch": 0.250837501720894,
      "grad_norm": 0.5008472204208374,
      "learning_rate": 9.896423338165192e-06,
      "loss": 0.4512,
      "step": 5466
    },
    {
      "epoch": 0.2508833922261484,
      "grad_norm": 0.5236533284187317,
      "learning_rate": 9.896373684578533e-06,
      "loss": 0.4884,
      "step": 5467
    },
    {
      "epoch": 0.25092928273140286,
      "grad_norm": 0.514059841632843,
      "learning_rate": 9.896324019217642e-06,
      "loss": 0.5125,
      "step": 5468
    },
    {
      "epoch": 0.25097517323665736,
      "grad_norm": 0.449461430311203,
      "learning_rate": 9.896274342082637e-06,
      "loss": 0.3511,
      "step": 5469
    },
    {
      "epoch": 0.2510210637419118,
      "grad_norm": 0.5872984528541565,
      "learning_rate": 9.896224653173636e-06,
      "loss": 0.5086,
      "step": 5470
    },
    {
      "epoch": 0.25106695424716624,
      "grad_norm": 0.510627269744873,
      "learning_rate": 9.89617495249076e-06,
      "loss": 0.4619,
      "step": 5471
    },
    {
      "epoch": 0.25111284475242074,
      "grad_norm": 0.5474761128425598,
      "learning_rate": 9.89612524003413e-06,
      "loss": 0.5013,
      "step": 5472
    },
    {
      "epoch": 0.2511587352576752,
      "grad_norm": 0.6970511674880981,
      "learning_rate": 9.896075515803864e-06,
      "loss": 0.524,
      "step": 5473
    },
    {
      "epoch": 0.2512046257629296,
      "grad_norm": 0.4522456228733063,
      "learning_rate": 9.89602577980008e-06,
      "loss": 0.3851,
      "step": 5474
    },
    {
      "epoch": 0.2512505162681841,
      "grad_norm": 0.4794396162033081,
      "learning_rate": 9.8959760320229e-06,
      "loss": 0.3983,
      "step": 5475
    },
    {
      "epoch": 0.25129640677343856,
      "grad_norm": 0.4617340564727783,
      "learning_rate": 9.895926272472444e-06,
      "loss": 0.4012,
      "step": 5476
    },
    {
      "epoch": 0.25134229727869306,
      "grad_norm": 0.49735814332962036,
      "learning_rate": 9.89587650114883e-06,
      "loss": 0.4215,
      "step": 5477
    },
    {
      "epoch": 0.2513881877839475,
      "grad_norm": 0.49297571182250977,
      "learning_rate": 9.895826718052178e-06,
      "loss": 0.4557,
      "step": 5478
    },
    {
      "epoch": 0.25143407828920195,
      "grad_norm": 0.5293284058570862,
      "learning_rate": 9.895776923182607e-06,
      "loss": 0.5061,
      "step": 5479
    },
    {
      "epoch": 0.25147996879445644,
      "grad_norm": 0.4811141788959503,
      "learning_rate": 9.895727116540239e-06,
      "loss": 0.3875,
      "step": 5480
    },
    {
      "epoch": 0.2515258592997109,
      "grad_norm": 0.48295044898986816,
      "learning_rate": 9.895677298125192e-06,
      "loss": 0.3987,
      "step": 5481
    },
    {
      "epoch": 0.25157174980496533,
      "grad_norm": 0.516815185546875,
      "learning_rate": 9.895627467937585e-06,
      "loss": 0.4116,
      "step": 5482
    },
    {
      "epoch": 0.2516176403102198,
      "grad_norm": 0.44515928626060486,
      "learning_rate": 9.89557762597754e-06,
      "loss": 0.3499,
      "step": 5483
    },
    {
      "epoch": 0.25166353081547427,
      "grad_norm": 0.43094825744628906,
      "learning_rate": 9.895527772245177e-06,
      "loss": 0.3346,
      "step": 5484
    },
    {
      "epoch": 0.25170942132072877,
      "grad_norm": 0.4834476411342621,
      "learning_rate": 9.895477906740614e-06,
      "loss": 0.4147,
      "step": 5485
    },
    {
      "epoch": 0.2517553118259832,
      "grad_norm": 0.45588424801826477,
      "learning_rate": 9.89542802946397e-06,
      "loss": 0.3709,
      "step": 5486
    },
    {
      "epoch": 0.25180120233123765,
      "grad_norm": 0.47425857186317444,
      "learning_rate": 9.895378140415369e-06,
      "loss": 0.4195,
      "step": 5487
    },
    {
      "epoch": 0.25184709283649215,
      "grad_norm": 0.4501340985298157,
      "learning_rate": 9.895328239594928e-06,
      "loss": 0.3498,
      "step": 5488
    },
    {
      "epoch": 0.2518929833417466,
      "grad_norm": 0.4549058675765991,
      "learning_rate": 9.895278327002766e-06,
      "loss": 0.4422,
      "step": 5489
    },
    {
      "epoch": 0.25193887384700103,
      "grad_norm": 0.457581102848053,
      "learning_rate": 9.895228402639005e-06,
      "loss": 0.3769,
      "step": 5490
    },
    {
      "epoch": 0.25198476435225553,
      "grad_norm": 0.43253082036972046,
      "learning_rate": 9.895178466503768e-06,
      "loss": 0.3287,
      "step": 5491
    },
    {
      "epoch": 0.25203065485751,
      "grad_norm": 0.45106038451194763,
      "learning_rate": 9.895128518597168e-06,
      "loss": 0.3674,
      "step": 5492
    },
    {
      "epoch": 0.25207654536276447,
      "grad_norm": 0.4881967008113861,
      "learning_rate": 9.895078558919331e-06,
      "loss": 0.459,
      "step": 5493
    },
    {
      "epoch": 0.2521224358680189,
      "grad_norm": 0.48361441493034363,
      "learning_rate": 9.895028587470374e-06,
      "loss": 0.4003,
      "step": 5494
    },
    {
      "epoch": 0.25216832637327335,
      "grad_norm": 0.5004562735557556,
      "learning_rate": 9.894978604250419e-06,
      "loss": 0.4629,
      "step": 5495
    },
    {
      "epoch": 0.25221421687852785,
      "grad_norm": 0.4870319962501526,
      "learning_rate": 9.894928609259585e-06,
      "loss": 0.4475,
      "step": 5496
    },
    {
      "epoch": 0.2522601073837823,
      "grad_norm": 0.4632299244403839,
      "learning_rate": 9.89487860249799e-06,
      "loss": 0.3918,
      "step": 5497
    },
    {
      "epoch": 0.25230599788903674,
      "grad_norm": 0.46857255697250366,
      "learning_rate": 9.89482858396576e-06,
      "loss": 0.3941,
      "step": 5498
    },
    {
      "epoch": 0.25235188839429123,
      "grad_norm": 0.7058764100074768,
      "learning_rate": 9.894778553663012e-06,
      "loss": 0.4038,
      "step": 5499
    },
    {
      "epoch": 0.2523977788995457,
      "grad_norm": 0.4546481966972351,
      "learning_rate": 9.894728511589864e-06,
      "loss": 0.4424,
      "step": 5500
    },
    {
      "epoch": 0.2524436694048001,
      "grad_norm": 0.47095194458961487,
      "learning_rate": 9.89467845774644e-06,
      "loss": 0.3491,
      "step": 5501
    },
    {
      "epoch": 0.2524895599100546,
      "grad_norm": 0.5268166065216064,
      "learning_rate": 9.89462839213286e-06,
      "loss": 0.4758,
      "step": 5502
    },
    {
      "epoch": 0.25253545041530906,
      "grad_norm": 0.4952910542488098,
      "learning_rate": 9.894578314749244e-06,
      "loss": 0.4379,
      "step": 5503
    },
    {
      "epoch": 0.25258134092056356,
      "grad_norm": 0.46238887310028076,
      "learning_rate": 9.894528225595709e-06,
      "loss": 0.356,
      "step": 5504
    },
    {
      "epoch": 0.252627231425818,
      "grad_norm": 0.46178463101387024,
      "learning_rate": 9.894478124672382e-06,
      "loss": 0.3836,
      "step": 5505
    },
    {
      "epoch": 0.25267312193107244,
      "grad_norm": 0.49608203768730164,
      "learning_rate": 9.894428011979378e-06,
      "loss": 0.4863,
      "step": 5506
    },
    {
      "epoch": 0.25271901243632694,
      "grad_norm": 0.4876364469528198,
      "learning_rate": 9.89437788751682e-06,
      "loss": 0.3791,
      "step": 5507
    },
    {
      "epoch": 0.2527649029415814,
      "grad_norm": 0.4905228614807129,
      "learning_rate": 9.894327751284827e-06,
      "loss": 0.4258,
      "step": 5508
    },
    {
      "epoch": 0.2528107934468358,
      "grad_norm": 0.46615681052207947,
      "learning_rate": 9.89427760328352e-06,
      "loss": 0.4357,
      "step": 5509
    },
    {
      "epoch": 0.2528566839520903,
      "grad_norm": 0.5354674458503723,
      "learning_rate": 9.89422744351302e-06,
      "loss": 0.4851,
      "step": 5510
    },
    {
      "epoch": 0.25290257445734476,
      "grad_norm": 0.4339059889316559,
      "learning_rate": 9.89417727197345e-06,
      "loss": 0.3217,
      "step": 5511
    },
    {
      "epoch": 0.25294846496259926,
      "grad_norm": 0.4238944947719574,
      "learning_rate": 9.894127088664927e-06,
      "loss": 0.3389,
      "step": 5512
    },
    {
      "epoch": 0.2529943554678537,
      "grad_norm": 0.5029617547988892,
      "learning_rate": 9.894076893587572e-06,
      "loss": 0.4502,
      "step": 5513
    },
    {
      "epoch": 0.25304024597310815,
      "grad_norm": 0.4667750298976898,
      "learning_rate": 9.894026686741508e-06,
      "loss": 0.3406,
      "step": 5514
    },
    {
      "epoch": 0.25308613647836264,
      "grad_norm": 0.4833560883998871,
      "learning_rate": 9.893976468126853e-06,
      "loss": 0.4507,
      "step": 5515
    },
    {
      "epoch": 0.2531320269836171,
      "grad_norm": 0.4052206575870514,
      "learning_rate": 9.893926237743732e-06,
      "loss": 0.3219,
      "step": 5516
    },
    {
      "epoch": 0.25317791748887153,
      "grad_norm": 0.4374043345451355,
      "learning_rate": 9.893875995592259e-06,
      "loss": 0.3163,
      "step": 5517
    },
    {
      "epoch": 0.253223807994126,
      "grad_norm": 0.5027537941932678,
      "learning_rate": 9.893825741672562e-06,
      "loss": 0.4739,
      "step": 5518
    },
    {
      "epoch": 0.25326969849938047,
      "grad_norm": 0.5046042799949646,
      "learning_rate": 9.893775475984757e-06,
      "loss": 0.4676,
      "step": 5519
    },
    {
      "epoch": 0.25331558900463497,
      "grad_norm": 0.5012478828430176,
      "learning_rate": 9.893725198528968e-06,
      "loss": 0.3771,
      "step": 5520
    },
    {
      "epoch": 0.2533614795098894,
      "grad_norm": 0.45714667439460754,
      "learning_rate": 9.893674909305312e-06,
      "loss": 0.4008,
      "step": 5521
    },
    {
      "epoch": 0.25340737001514385,
      "grad_norm": 0.44811463356018066,
      "learning_rate": 9.893624608313915e-06,
      "loss": 0.37,
      "step": 5522
    },
    {
      "epoch": 0.25345326052039835,
      "grad_norm": 0.4998549520969391,
      "learning_rate": 9.893574295554893e-06,
      "loss": 0.414,
      "step": 5523
    },
    {
      "epoch": 0.2534991510256528,
      "grad_norm": 0.5043789744377136,
      "learning_rate": 9.893523971028371e-06,
      "loss": 0.4726,
      "step": 5524
    },
    {
      "epoch": 0.25354504153090723,
      "grad_norm": 0.4747314751148224,
      "learning_rate": 9.893473634734467e-06,
      "loss": 0.4017,
      "step": 5525
    },
    {
      "epoch": 0.25359093203616173,
      "grad_norm": 0.4761166274547577,
      "learning_rate": 9.893423286673306e-06,
      "loss": 0.407,
      "step": 5526
    },
    {
      "epoch": 0.2536368225414162,
      "grad_norm": 0.44375160336494446,
      "learning_rate": 9.893372926845002e-06,
      "loss": 0.3302,
      "step": 5527
    },
    {
      "epoch": 0.25368271304667067,
      "grad_norm": 0.43359482288360596,
      "learning_rate": 9.893322555249683e-06,
      "loss": 0.3575,
      "step": 5528
    },
    {
      "epoch": 0.2537286035519251,
      "grad_norm": 0.45954516530036926,
      "learning_rate": 9.893272171887467e-06,
      "loss": 0.3575,
      "step": 5529
    },
    {
      "epoch": 0.25377449405717956,
      "grad_norm": 0.4840056300163269,
      "learning_rate": 9.893221776758477e-06,
      "loss": 0.4109,
      "step": 5530
    },
    {
      "epoch": 0.25382038456243405,
      "grad_norm": 0.49138200283050537,
      "learning_rate": 9.893171369862833e-06,
      "loss": 0.46,
      "step": 5531
    },
    {
      "epoch": 0.2538662750676885,
      "grad_norm": 0.474150687456131,
      "learning_rate": 9.893120951200654e-06,
      "loss": 0.3854,
      "step": 5532
    },
    {
      "epoch": 0.25391216557294294,
      "grad_norm": 0.4818645119667053,
      "learning_rate": 9.893070520772066e-06,
      "loss": 0.496,
      "step": 5533
    },
    {
      "epoch": 0.25395805607819744,
      "grad_norm": 0.44986072182655334,
      "learning_rate": 9.893020078577186e-06,
      "loss": 0.3641,
      "step": 5534
    },
    {
      "epoch": 0.2540039465834519,
      "grad_norm": 0.4712718427181244,
      "learning_rate": 9.892969624616138e-06,
      "loss": 0.4439,
      "step": 5535
    },
    {
      "epoch": 0.2540498370887063,
      "grad_norm": 0.4735841751098633,
      "learning_rate": 9.89291915888904e-06,
      "loss": 0.4061,
      "step": 5536
    },
    {
      "epoch": 0.2540957275939608,
      "grad_norm": 0.46155858039855957,
      "learning_rate": 9.89286868139602e-06,
      "loss": 0.3522,
      "step": 5537
    },
    {
      "epoch": 0.25414161809921526,
      "grad_norm": 0.5060663819313049,
      "learning_rate": 9.892818192137192e-06,
      "loss": 0.485,
      "step": 5538
    },
    {
      "epoch": 0.25418750860446976,
      "grad_norm": 0.5239194631576538,
      "learning_rate": 9.89276769111268e-06,
      "loss": 0.5012,
      "step": 5539
    },
    {
      "epoch": 0.2542333991097242,
      "grad_norm": 0.48788994550704956,
      "learning_rate": 9.892717178322606e-06,
      "loss": 0.4553,
      "step": 5540
    },
    {
      "epoch": 0.25427928961497864,
      "grad_norm": 1.0392351150512695,
      "learning_rate": 9.892666653767093e-06,
      "loss": 0.3119,
      "step": 5541
    },
    {
      "epoch": 0.25432518012023314,
      "grad_norm": 0.4582270681858063,
      "learning_rate": 9.892616117446261e-06,
      "loss": 0.4056,
      "step": 5542
    },
    {
      "epoch": 0.2543710706254876,
      "grad_norm": 0.4743838608264923,
      "learning_rate": 9.892565569360231e-06,
      "loss": 0.4256,
      "step": 5543
    },
    {
      "epoch": 0.254416961130742,
      "grad_norm": 0.60857754945755,
      "learning_rate": 9.892515009509122e-06,
      "loss": 0.4651,
      "step": 5544
    },
    {
      "epoch": 0.2544628516359965,
      "grad_norm": 0.4752808213233948,
      "learning_rate": 9.892464437893063e-06,
      "loss": 0.3833,
      "step": 5545
    },
    {
      "epoch": 0.25450874214125097,
      "grad_norm": 0.4716651737689972,
      "learning_rate": 9.892413854512168e-06,
      "loss": 0.3562,
      "step": 5546
    },
    {
      "epoch": 0.25455463264650546,
      "grad_norm": 0.44940945506095886,
      "learning_rate": 9.892363259366565e-06,
      "loss": 0.3903,
      "step": 5547
    },
    {
      "epoch": 0.2546005231517599,
      "grad_norm": 0.44334253668785095,
      "learning_rate": 9.892312652456369e-06,
      "loss": 0.3756,
      "step": 5548
    },
    {
      "epoch": 0.25464641365701435,
      "grad_norm": 0.4882247745990753,
      "learning_rate": 9.892262033781706e-06,
      "loss": 0.4503,
      "step": 5549
    },
    {
      "epoch": 0.25469230416226885,
      "grad_norm": 0.47684043645858765,
      "learning_rate": 9.892211403342697e-06,
      "loss": 0.386,
      "step": 5550
    },
    {
      "epoch": 0.2547381946675233,
      "grad_norm": 0.5005866885185242,
      "learning_rate": 9.892160761139465e-06,
      "loss": 0.4564,
      "step": 5551
    },
    {
      "epoch": 0.25478408517277773,
      "grad_norm": 0.4876570999622345,
      "learning_rate": 9.89211010717213e-06,
      "loss": 0.4046,
      "step": 5552
    },
    {
      "epoch": 0.25482997567803223,
      "grad_norm": 0.5085650682449341,
      "learning_rate": 9.892059441440813e-06,
      "loss": 0.3926,
      "step": 5553
    },
    {
      "epoch": 0.25487586618328667,
      "grad_norm": 0.4553479254245758,
      "learning_rate": 9.892008763945637e-06,
      "loss": 0.3673,
      "step": 5554
    },
    {
      "epoch": 0.25492175668854117,
      "grad_norm": 0.4442640244960785,
      "learning_rate": 9.891958074686725e-06,
      "loss": 0.3453,
      "step": 5555
    },
    {
      "epoch": 0.2549676471937956,
      "grad_norm": 0.46716830134391785,
      "learning_rate": 9.891907373664197e-06,
      "loss": 0.3529,
      "step": 5556
    },
    {
      "epoch": 0.25501353769905005,
      "grad_norm": 0.49357402324676514,
      "learning_rate": 9.891856660878176e-06,
      "loss": 0.3627,
      "step": 5557
    },
    {
      "epoch": 0.25505942820430455,
      "grad_norm": 0.4599742591381073,
      "learning_rate": 9.891805936328783e-06,
      "loss": 0.4236,
      "step": 5558
    },
    {
      "epoch": 0.255105318709559,
      "grad_norm": 0.47392216324806213,
      "learning_rate": 9.89175520001614e-06,
      "loss": 0.4066,
      "step": 5559
    },
    {
      "epoch": 0.25515120921481343,
      "grad_norm": 0.41979286074638367,
      "learning_rate": 9.891704451940372e-06,
      "loss": 0.3507,
      "step": 5560
    },
    {
      "epoch": 0.25519709972006793,
      "grad_norm": 0.4431149363517761,
      "learning_rate": 9.891653692101597e-06,
      "loss": 0.3008,
      "step": 5561
    },
    {
      "epoch": 0.2552429902253224,
      "grad_norm": 0.45472508668899536,
      "learning_rate": 9.891602920499939e-06,
      "loss": 0.3621,
      "step": 5562
    },
    {
      "epoch": 0.2552888807305768,
      "grad_norm": 0.5023146271705627,
      "learning_rate": 9.89155213713552e-06,
      "loss": 0.3844,
      "step": 5563
    },
    {
      "epoch": 0.2553347712358313,
      "grad_norm": 0.4400866627693176,
      "learning_rate": 9.891501342008463e-06,
      "loss": 0.3393,
      "step": 5564
    },
    {
      "epoch": 0.25538066174108576,
      "grad_norm": 0.49070653319358826,
      "learning_rate": 9.891450535118887e-06,
      "loss": 0.4588,
      "step": 5565
    },
    {
      "epoch": 0.25542655224634025,
      "grad_norm": 0.4746112823486328,
      "learning_rate": 9.891399716466919e-06,
      "loss": 0.3997,
      "step": 5566
    },
    {
      "epoch": 0.2554724427515947,
      "grad_norm": 0.46234557032585144,
      "learning_rate": 9.891348886052675e-06,
      "loss": 0.365,
      "step": 5567
    },
    {
      "epoch": 0.25551833325684914,
      "grad_norm": 0.46192625164985657,
      "learning_rate": 9.891298043876282e-06,
      "loss": 0.366,
      "step": 5568
    },
    {
      "epoch": 0.25556422376210364,
      "grad_norm": 0.5020145773887634,
      "learning_rate": 9.891247189937863e-06,
      "loss": 0.5286,
      "step": 5569
    },
    {
      "epoch": 0.2556101142673581,
      "grad_norm": 0.5359102487564087,
      "learning_rate": 9.891196324237536e-06,
      "loss": 0.4536,
      "step": 5570
    },
    {
      "epoch": 0.2556560047726125,
      "grad_norm": 0.4720180332660675,
      "learning_rate": 9.891145446775425e-06,
      "loss": 0.3753,
      "step": 5571
    },
    {
      "epoch": 0.255701895277867,
      "grad_norm": 0.45280662178993225,
      "learning_rate": 9.891094557551654e-06,
      "loss": 0.3654,
      "step": 5572
    },
    {
      "epoch": 0.25574778578312146,
      "grad_norm": 0.45483464002609253,
      "learning_rate": 9.891043656566346e-06,
      "loss": 0.3865,
      "step": 5573
    },
    {
      "epoch": 0.25579367628837596,
      "grad_norm": 0.46961984038352966,
      "learning_rate": 9.890992743819618e-06,
      "loss": 0.4235,
      "step": 5574
    },
    {
      "epoch": 0.2558395667936304,
      "grad_norm": 0.4911828637123108,
      "learning_rate": 9.8909418193116e-06,
      "loss": 0.4326,
      "step": 5575
    },
    {
      "epoch": 0.25588545729888484,
      "grad_norm": 0.48026853799819946,
      "learning_rate": 9.890890883042408e-06,
      "loss": 0.4484,
      "step": 5576
    },
    {
      "epoch": 0.25593134780413934,
      "grad_norm": 0.49577564001083374,
      "learning_rate": 9.890839935012166e-06,
      "loss": 0.5035,
      "step": 5577
    },
    {
      "epoch": 0.2559772383093938,
      "grad_norm": 0.4562341272830963,
      "learning_rate": 9.890788975221e-06,
      "loss": 0.3576,
      "step": 5578
    },
    {
      "epoch": 0.2560231288146482,
      "grad_norm": 0.557658314704895,
      "learning_rate": 9.890738003669029e-06,
      "loss": 0.5264,
      "step": 5579
    },
    {
      "epoch": 0.2560690193199027,
      "grad_norm": 0.4654678702354431,
      "learning_rate": 9.890687020356378e-06,
      "loss": 0.3988,
      "step": 5580
    },
    {
      "epoch": 0.25611490982515717,
      "grad_norm": 0.5333075523376465,
      "learning_rate": 9.890636025283166e-06,
      "loss": 0.4855,
      "step": 5581
    },
    {
      "epoch": 0.25616080033041166,
      "grad_norm": 0.45178472995758057,
      "learning_rate": 9.890585018449518e-06,
      "loss": 0.3296,
      "step": 5582
    },
    {
      "epoch": 0.2562066908356661,
      "grad_norm": 0.4532388746738434,
      "learning_rate": 9.890533999855558e-06,
      "loss": 0.3558,
      "step": 5583
    },
    {
      "epoch": 0.25625258134092055,
      "grad_norm": 0.4403430223464966,
      "learning_rate": 9.890482969501408e-06,
      "loss": 0.3978,
      "step": 5584
    },
    {
      "epoch": 0.25629847184617505,
      "grad_norm": 0.46092748641967773,
      "learning_rate": 9.890431927387188e-06,
      "loss": 0.398,
      "step": 5585
    },
    {
      "epoch": 0.2563443623514295,
      "grad_norm": 0.693153440952301,
      "learning_rate": 9.890380873513023e-06,
      "loss": 0.5726,
      "step": 5586
    },
    {
      "epoch": 0.25639025285668393,
      "grad_norm": 0.48726266622543335,
      "learning_rate": 9.890329807879037e-06,
      "loss": 0.408,
      "step": 5587
    },
    {
      "epoch": 0.25643614336193843,
      "grad_norm": 0.49542340636253357,
      "learning_rate": 9.89027873048535e-06,
      "loss": 0.4368,
      "step": 5588
    },
    {
      "epoch": 0.25648203386719287,
      "grad_norm": 0.5103986263275146,
      "learning_rate": 9.890227641332086e-06,
      "loss": 0.5059,
      "step": 5589
    },
    {
      "epoch": 0.2565279243724473,
      "grad_norm": 0.4401119351387024,
      "learning_rate": 9.890176540419367e-06,
      "loss": 0.3144,
      "step": 5590
    },
    {
      "epoch": 0.2565738148777018,
      "grad_norm": 0.47371500730514526,
      "learning_rate": 9.890125427747319e-06,
      "loss": 0.4317,
      "step": 5591
    },
    {
      "epoch": 0.25661970538295625,
      "grad_norm": 0.5025570392608643,
      "learning_rate": 9.890074303316062e-06,
      "loss": 0.4695,
      "step": 5592
    },
    {
      "epoch": 0.25666559588821075,
      "grad_norm": 0.5215364694595337,
      "learning_rate": 9.89002316712572e-06,
      "loss": 0.4513,
      "step": 5593
    },
    {
      "epoch": 0.2567114863934652,
      "grad_norm": 0.4691109359264374,
      "learning_rate": 9.889972019176415e-06,
      "loss": 0.3736,
      "step": 5594
    },
    {
      "epoch": 0.25675737689871964,
      "grad_norm": 0.4602471590042114,
      "learning_rate": 9.88992085946827e-06,
      "loss": 0.4056,
      "step": 5595
    },
    {
      "epoch": 0.25680326740397413,
      "grad_norm": 0.47685331106185913,
      "learning_rate": 9.889869688001407e-06,
      "loss": 0.4077,
      "step": 5596
    },
    {
      "epoch": 0.2568491579092286,
      "grad_norm": 0.4990473687648773,
      "learning_rate": 9.889818504775955e-06,
      "loss": 0.4436,
      "step": 5597
    },
    {
      "epoch": 0.256895048414483,
      "grad_norm": 0.496969610452652,
      "learning_rate": 9.889767309792029e-06,
      "loss": 0.4826,
      "step": 5598
    },
    {
      "epoch": 0.2569409389197375,
      "grad_norm": 0.45757797360420227,
      "learning_rate": 9.889716103049757e-06,
      "loss": 0.409,
      "step": 5599
    },
    {
      "epoch": 0.25698682942499196,
      "grad_norm": 0.45495250821113586,
      "learning_rate": 9.889664884549262e-06,
      "loss": 0.4045,
      "step": 5600
    },
    {
      "epoch": 0.25703271993024646,
      "grad_norm": 0.5110076665878296,
      "learning_rate": 9.889613654290666e-06,
      "loss": 0.4376,
      "step": 5601
    },
    {
      "epoch": 0.2570786104355009,
      "grad_norm": 0.47988972067832947,
      "learning_rate": 9.889562412274092e-06,
      "loss": 0.4244,
      "step": 5602
    },
    {
      "epoch": 0.25712450094075534,
      "grad_norm": 0.48515453934669495,
      "learning_rate": 9.889511158499664e-06,
      "loss": 0.4985,
      "step": 5603
    },
    {
      "epoch": 0.25717039144600984,
      "grad_norm": 0.4759584665298462,
      "learning_rate": 9.889459892967504e-06,
      "loss": 0.3717,
      "step": 5604
    },
    {
      "epoch": 0.2572162819512643,
      "grad_norm": 0.4437233805656433,
      "learning_rate": 9.889408615677737e-06,
      "loss": 0.3377,
      "step": 5605
    },
    {
      "epoch": 0.2572621724565187,
      "grad_norm": 0.45227745175361633,
      "learning_rate": 9.889357326630485e-06,
      "loss": 0.3572,
      "step": 5606
    },
    {
      "epoch": 0.2573080629617732,
      "grad_norm": 0.42537635564804077,
      "learning_rate": 9.88930602582587e-06,
      "loss": 0.2736,
      "step": 5607
    },
    {
      "epoch": 0.25735395346702766,
      "grad_norm": 0.4980013370513916,
      "learning_rate": 9.88925471326402e-06,
      "loss": 0.4433,
      "step": 5608
    },
    {
      "epoch": 0.25739984397228216,
      "grad_norm": 0.5872184634208679,
      "learning_rate": 9.889203388945054e-06,
      "loss": 0.4934,
      "step": 5609
    },
    {
      "epoch": 0.2574457344775366,
      "grad_norm": 0.561522364616394,
      "learning_rate": 9.889152052869098e-06,
      "loss": 0.5086,
      "step": 5610
    },
    {
      "epoch": 0.25749162498279105,
      "grad_norm": 0.5021909475326538,
      "learning_rate": 9.889100705036274e-06,
      "loss": 0.4543,
      "step": 5611
    },
    {
      "epoch": 0.25753751548804554,
      "grad_norm": 0.4873514473438263,
      "learning_rate": 9.889049345446706e-06,
      "loss": 0.3693,
      "step": 5612
    },
    {
      "epoch": 0.2575834059933,
      "grad_norm": 0.5251417756080627,
      "learning_rate": 9.888997974100517e-06,
      "loss": 0.4893,
      "step": 5613
    },
    {
      "epoch": 0.2576292964985544,
      "grad_norm": 0.48143884539604187,
      "learning_rate": 9.88894659099783e-06,
      "loss": 0.3885,
      "step": 5614
    },
    {
      "epoch": 0.2576751870038089,
      "grad_norm": 0.49281200766563416,
      "learning_rate": 9.88889519613877e-06,
      "loss": 0.4234,
      "step": 5615
    },
    {
      "epoch": 0.25772107750906337,
      "grad_norm": 0.4757404625415802,
      "learning_rate": 9.888843789523461e-06,
      "loss": 0.4158,
      "step": 5616
    },
    {
      "epoch": 0.2577669680143178,
      "grad_norm": 0.44728314876556396,
      "learning_rate": 9.888792371152023e-06,
      "loss": 0.3709,
      "step": 5617
    },
    {
      "epoch": 0.2578128585195723,
      "grad_norm": 0.4950680136680603,
      "learning_rate": 9.888740941024585e-06,
      "loss": 0.4795,
      "step": 5618
    },
    {
      "epoch": 0.25785874902482675,
      "grad_norm": 0.4776691198348999,
      "learning_rate": 9.888689499141268e-06,
      "loss": 0.4224,
      "step": 5619
    },
    {
      "epoch": 0.25790463953008125,
      "grad_norm": 0.4736878573894501,
      "learning_rate": 9.888638045502194e-06,
      "loss": 0.3926,
      "step": 5620
    },
    {
      "epoch": 0.2579505300353357,
      "grad_norm": 0.42839500308036804,
      "learning_rate": 9.888586580107489e-06,
      "loss": 0.2935,
      "step": 5621
    },
    {
      "epoch": 0.25799642054059013,
      "grad_norm": 0.4897996485233307,
      "learning_rate": 9.888535102957278e-06,
      "loss": 0.3817,
      "step": 5622
    },
    {
      "epoch": 0.25804231104584463,
      "grad_norm": 0.4382639527320862,
      "learning_rate": 9.88848361405168e-06,
      "loss": 0.3819,
      "step": 5623
    },
    {
      "epoch": 0.2580882015510991,
      "grad_norm": 0.48136210441589355,
      "learning_rate": 9.888432113390823e-06,
      "loss": 0.3766,
      "step": 5624
    },
    {
      "epoch": 0.2581340920563535,
      "grad_norm": 0.4868128299713135,
      "learning_rate": 9.88838060097483e-06,
      "loss": 0.3944,
      "step": 5625
    },
    {
      "epoch": 0.258179982561608,
      "grad_norm": 0.5675439238548279,
      "learning_rate": 9.888329076803824e-06,
      "loss": 0.4392,
      "step": 5626
    },
    {
      "epoch": 0.25822587306686245,
      "grad_norm": 0.46386784315109253,
      "learning_rate": 9.88827754087793e-06,
      "loss": 0.4326,
      "step": 5627
    },
    {
      "epoch": 0.25827176357211695,
      "grad_norm": 0.47001662850379944,
      "learning_rate": 9.88822599319727e-06,
      "loss": 0.4033,
      "step": 5628
    },
    {
      "epoch": 0.2583176540773714,
      "grad_norm": 0.4964858889579773,
      "learning_rate": 9.888174433761971e-06,
      "loss": 0.4086,
      "step": 5629
    },
    {
      "epoch": 0.25836354458262584,
      "grad_norm": 0.48577457666397095,
      "learning_rate": 9.888122862572154e-06,
      "loss": 0.4221,
      "step": 5630
    },
    {
      "epoch": 0.25840943508788033,
      "grad_norm": 0.4313344955444336,
      "learning_rate": 9.888071279627944e-06,
      "loss": 0.3411,
      "step": 5631
    },
    {
      "epoch": 0.2584553255931348,
      "grad_norm": 0.5158870816230774,
      "learning_rate": 9.888019684929467e-06,
      "loss": 0.4409,
      "step": 5632
    },
    {
      "epoch": 0.2585012160983892,
      "grad_norm": 0.5176045298576355,
      "learning_rate": 9.887968078476844e-06,
      "loss": 0.4836,
      "step": 5633
    },
    {
      "epoch": 0.2585471066036437,
      "grad_norm": 0.45540517568588257,
      "learning_rate": 9.8879164602702e-06,
      "loss": 0.379,
      "step": 5634
    },
    {
      "epoch": 0.25859299710889816,
      "grad_norm": 0.470023512840271,
      "learning_rate": 9.88786483030966e-06,
      "loss": 0.4302,
      "step": 5635
    },
    {
      "epoch": 0.25863888761415266,
      "grad_norm": 0.4943717122077942,
      "learning_rate": 9.887813188595348e-06,
      "loss": 0.4839,
      "step": 5636
    },
    {
      "epoch": 0.2586847781194071,
      "grad_norm": 0.4424592852592468,
      "learning_rate": 9.887761535127388e-06,
      "loss": 0.3031,
      "step": 5637
    },
    {
      "epoch": 0.25873066862466154,
      "grad_norm": 0.489914208650589,
      "learning_rate": 9.887709869905903e-06,
      "loss": 0.419,
      "step": 5638
    },
    {
      "epoch": 0.25877655912991604,
      "grad_norm": 0.4437110126018524,
      "learning_rate": 9.88765819293102e-06,
      "loss": 0.324,
      "step": 5639
    },
    {
      "epoch": 0.2588224496351705,
      "grad_norm": 0.47663143277168274,
      "learning_rate": 9.887606504202862e-06,
      "loss": 0.4116,
      "step": 5640
    },
    {
      "epoch": 0.2588683401404249,
      "grad_norm": 0.4884633719921112,
      "learning_rate": 9.887554803721551e-06,
      "loss": 0.4195,
      "step": 5641
    },
    {
      "epoch": 0.2589142306456794,
      "grad_norm": 0.44610312581062317,
      "learning_rate": 9.887503091487215e-06,
      "loss": 0.3988,
      "step": 5642
    },
    {
      "epoch": 0.25896012115093386,
      "grad_norm": 0.4350149631500244,
      "learning_rate": 9.887451367499977e-06,
      "loss": 0.3877,
      "step": 5643
    },
    {
      "epoch": 0.25900601165618836,
      "grad_norm": 0.4546768367290497,
      "learning_rate": 9.88739963175996e-06,
      "loss": 0.3519,
      "step": 5644
    },
    {
      "epoch": 0.2590519021614428,
      "grad_norm": 0.46952152252197266,
      "learning_rate": 9.887347884267287e-06,
      "loss": 0.3849,
      "step": 5645
    },
    {
      "epoch": 0.25909779266669725,
      "grad_norm": 0.50528484582901,
      "learning_rate": 9.887296125022089e-06,
      "loss": 0.468,
      "step": 5646
    },
    {
      "epoch": 0.25914368317195174,
      "grad_norm": 0.4472591280937195,
      "learning_rate": 9.887244354024484e-06,
      "loss": 0.3703,
      "step": 5647
    },
    {
      "epoch": 0.2591895736772062,
      "grad_norm": 0.477812260389328,
      "learning_rate": 9.887192571274597e-06,
      "loss": 0.4241,
      "step": 5648
    },
    {
      "epoch": 0.25923546418246063,
      "grad_norm": 0.4468366801738739,
      "learning_rate": 9.887140776772557e-06,
      "loss": 0.3831,
      "step": 5649
    },
    {
      "epoch": 0.2592813546877151,
      "grad_norm": 0.47129249572753906,
      "learning_rate": 9.887088970518485e-06,
      "loss": 0.4195,
      "step": 5650
    },
    {
      "epoch": 0.25932724519296957,
      "grad_norm": 0.5563701391220093,
      "learning_rate": 9.887037152512507e-06,
      "loss": 0.4239,
      "step": 5651
    },
    {
      "epoch": 0.259373135698224,
      "grad_norm": 0.47246047854423523,
      "learning_rate": 9.886985322754748e-06,
      "loss": 0.4148,
      "step": 5652
    },
    {
      "epoch": 0.2594190262034785,
      "grad_norm": 0.45770326256752014,
      "learning_rate": 9.88693348124533e-06,
      "loss": 0.3352,
      "step": 5653
    },
    {
      "epoch": 0.25946491670873295,
      "grad_norm": 0.4647235572338104,
      "learning_rate": 9.886881627984379e-06,
      "loss": 0.4179,
      "step": 5654
    },
    {
      "epoch": 0.25951080721398745,
      "grad_norm": 0.4490934908390045,
      "learning_rate": 9.886829762972021e-06,
      "loss": 0.3971,
      "step": 5655
    },
    {
      "epoch": 0.2595566977192419,
      "grad_norm": 0.4440559148788452,
      "learning_rate": 9.886777886208379e-06,
      "loss": 0.3211,
      "step": 5656
    },
    {
      "epoch": 0.25960258822449633,
      "grad_norm": 0.44536978006362915,
      "learning_rate": 9.886725997693581e-06,
      "loss": 0.3703,
      "step": 5657
    },
    {
      "epoch": 0.25964847872975083,
      "grad_norm": 0.5159117579460144,
      "learning_rate": 9.886674097427746e-06,
      "loss": 0.4096,
      "step": 5658
    },
    {
      "epoch": 0.2596943692350053,
      "grad_norm": 0.4857231378555298,
      "learning_rate": 9.886622185411004e-06,
      "loss": 0.396,
      "step": 5659
    },
    {
      "epoch": 0.2597402597402597,
      "grad_norm": 0.4726279079914093,
      "learning_rate": 9.886570261643477e-06,
      "loss": 0.3939,
      "step": 5660
    },
    {
      "epoch": 0.2597861502455142,
      "grad_norm": 0.4712117314338684,
      "learning_rate": 9.886518326125291e-06,
      "loss": 0.4464,
      "step": 5661
    },
    {
      "epoch": 0.25983204075076866,
      "grad_norm": 0.47619110345840454,
      "learning_rate": 9.886466378856571e-06,
      "loss": 0.4226,
      "step": 5662
    },
    {
      "epoch": 0.25987793125602315,
      "grad_norm": 0.47308769822120667,
      "learning_rate": 9.886414419837441e-06,
      "loss": 0.4048,
      "step": 5663
    },
    {
      "epoch": 0.2599238217612776,
      "grad_norm": 0.4963769316673279,
      "learning_rate": 9.886362449068027e-06,
      "loss": 0.4494,
      "step": 5664
    },
    {
      "epoch": 0.25996971226653204,
      "grad_norm": 0.45801129937171936,
      "learning_rate": 9.886310466548454e-06,
      "loss": 0.3756,
      "step": 5665
    },
    {
      "epoch": 0.26001560277178654,
      "grad_norm": 0.46514126658439636,
      "learning_rate": 9.886258472278845e-06,
      "loss": 0.3812,
      "step": 5666
    },
    {
      "epoch": 0.260061493277041,
      "grad_norm": 0.5091173052787781,
      "learning_rate": 9.886206466259327e-06,
      "loss": 0.428,
      "step": 5667
    },
    {
      "epoch": 0.2601073837822954,
      "grad_norm": 0.49231746792793274,
      "learning_rate": 9.886154448490025e-06,
      "loss": 0.4484,
      "step": 5668
    },
    {
      "epoch": 0.2601532742875499,
      "grad_norm": 0.4699980318546295,
      "learning_rate": 9.886102418971063e-06,
      "loss": 0.3719,
      "step": 5669
    },
    {
      "epoch": 0.26019916479280436,
      "grad_norm": 0.4968722462654114,
      "learning_rate": 9.886050377702567e-06,
      "loss": 0.4844,
      "step": 5670
    },
    {
      "epoch": 0.26024505529805886,
      "grad_norm": 0.4719417989253998,
      "learning_rate": 9.885998324684664e-06,
      "loss": 0.4785,
      "step": 5671
    },
    {
      "epoch": 0.2602909458033133,
      "grad_norm": 0.47552016377449036,
      "learning_rate": 9.885946259917475e-06,
      "loss": 0.4222,
      "step": 5672
    },
    {
      "epoch": 0.26033683630856774,
      "grad_norm": 0.49827972054481506,
      "learning_rate": 9.885894183401127e-06,
      "loss": 0.4506,
      "step": 5673
    },
    {
      "epoch": 0.26038272681382224,
      "grad_norm": 0.4808642268180847,
      "learning_rate": 9.885842095135747e-06,
      "loss": 0.4455,
      "step": 5674
    },
    {
      "epoch": 0.2604286173190767,
      "grad_norm": 0.43016520142555237,
      "learning_rate": 9.885789995121458e-06,
      "loss": 0.3076,
      "step": 5675
    },
    {
      "epoch": 0.2604745078243311,
      "grad_norm": 0.44965511560440063,
      "learning_rate": 9.885737883358386e-06,
      "loss": 0.3776,
      "step": 5676
    },
    {
      "epoch": 0.2605203983295856,
      "grad_norm": 0.455857515335083,
      "learning_rate": 9.885685759846657e-06,
      "loss": 0.4175,
      "step": 5677
    },
    {
      "epoch": 0.26056628883484007,
      "grad_norm": 0.5128337144851685,
      "learning_rate": 9.885633624586394e-06,
      "loss": 0.471,
      "step": 5678
    },
    {
      "epoch": 0.2606121793400945,
      "grad_norm": 0.5024634599685669,
      "learning_rate": 9.885581477577725e-06,
      "loss": 0.4149,
      "step": 5679
    },
    {
      "epoch": 0.260658069845349,
      "grad_norm": 0.4706532657146454,
      "learning_rate": 9.885529318820775e-06,
      "loss": 0.3852,
      "step": 5680
    },
    {
      "epoch": 0.26070396035060345,
      "grad_norm": 0.46730953454971313,
      "learning_rate": 9.885477148315668e-06,
      "loss": 0.3759,
      "step": 5681
    },
    {
      "epoch": 0.26074985085585795,
      "grad_norm": 0.49785250425338745,
      "learning_rate": 9.885424966062531e-06,
      "loss": 0.4406,
      "step": 5682
    },
    {
      "epoch": 0.2607957413611124,
      "grad_norm": 0.5237879753112793,
      "learning_rate": 9.885372772061489e-06,
      "loss": 0.471,
      "step": 5683
    },
    {
      "epoch": 0.26084163186636683,
      "grad_norm": 0.48327532410621643,
      "learning_rate": 9.885320566312667e-06,
      "loss": 0.4935,
      "step": 5684
    },
    {
      "epoch": 0.2608875223716213,
      "grad_norm": 0.46478649973869324,
      "learning_rate": 9.88526834881619e-06,
      "loss": 0.3727,
      "step": 5685
    },
    {
      "epoch": 0.26093341287687577,
      "grad_norm": 0.44101977348327637,
      "learning_rate": 9.885216119572187e-06,
      "loss": 0.344,
      "step": 5686
    },
    {
      "epoch": 0.2609793033821302,
      "grad_norm": 0.4429907202720642,
      "learning_rate": 9.88516387858078e-06,
      "loss": 0.359,
      "step": 5687
    },
    {
      "epoch": 0.2610251938873847,
      "grad_norm": 0.4707849621772766,
      "learning_rate": 9.885111625842094e-06,
      "loss": 0.3665,
      "step": 5688
    },
    {
      "epoch": 0.26107108439263915,
      "grad_norm": 0.47951963543891907,
      "learning_rate": 9.885059361356257e-06,
      "loss": 0.3379,
      "step": 5689
    },
    {
      "epoch": 0.26111697489789365,
      "grad_norm": 0.4259636700153351,
      "learning_rate": 9.885007085123394e-06,
      "loss": 0.3238,
      "step": 5690
    },
    {
      "epoch": 0.2611628654031481,
      "grad_norm": 0.46246758103370667,
      "learning_rate": 9.884954797143634e-06,
      "loss": 0.422,
      "step": 5691
    },
    {
      "epoch": 0.26120875590840253,
      "grad_norm": 0.5283334255218506,
      "learning_rate": 9.884902497417094e-06,
      "loss": 0.5073,
      "step": 5692
    },
    {
      "epoch": 0.26125464641365703,
      "grad_norm": 0.4169554114341736,
      "learning_rate": 9.884850185943909e-06,
      "loss": 0.3214,
      "step": 5693
    },
    {
      "epoch": 0.2613005369189115,
      "grad_norm": 0.4948008060455322,
      "learning_rate": 9.8847978627242e-06,
      "loss": 0.4556,
      "step": 5694
    },
    {
      "epoch": 0.2613464274241659,
      "grad_norm": 0.4492760896682739,
      "learning_rate": 9.884745527758095e-06,
      "loss": 0.376,
      "step": 5695
    },
    {
      "epoch": 0.2613923179294204,
      "grad_norm": 0.4468342363834381,
      "learning_rate": 9.884693181045716e-06,
      "loss": 0.3981,
      "step": 5696
    },
    {
      "epoch": 0.26143820843467486,
      "grad_norm": 0.4745164215564728,
      "learning_rate": 9.884640822587193e-06,
      "loss": 0.4631,
      "step": 5697
    },
    {
      "epoch": 0.26148409893992935,
      "grad_norm": 0.4509073495864868,
      "learning_rate": 9.88458845238265e-06,
      "loss": 0.3812,
      "step": 5698
    },
    {
      "epoch": 0.2615299894451838,
      "grad_norm": 0.4773392677307129,
      "learning_rate": 9.884536070432214e-06,
      "loss": 0.4344,
      "step": 5699
    },
    {
      "epoch": 0.26157587995043824,
      "grad_norm": 0.43625348806381226,
      "learning_rate": 9.884483676736011e-06,
      "loss": 0.31,
      "step": 5700
    },
    {
      "epoch": 0.26162177045569274,
      "grad_norm": 0.4980311393737793,
      "learning_rate": 9.884431271294166e-06,
      "loss": 0.4344,
      "step": 5701
    },
    {
      "epoch": 0.2616676609609472,
      "grad_norm": 0.4493265748023987,
      "learning_rate": 9.884378854106804e-06,
      "loss": 0.3803,
      "step": 5702
    },
    {
      "epoch": 0.2617135514662016,
      "grad_norm": 0.4348578155040741,
      "learning_rate": 9.884326425174053e-06,
      "loss": 0.3679,
      "step": 5703
    },
    {
      "epoch": 0.2617594419714561,
      "grad_norm": 0.4575294852256775,
      "learning_rate": 9.88427398449604e-06,
      "loss": 0.3482,
      "step": 5704
    },
    {
      "epoch": 0.26180533247671056,
      "grad_norm": 0.49547243118286133,
      "learning_rate": 9.884221532072887e-06,
      "loss": 0.457,
      "step": 5705
    },
    {
      "epoch": 0.261851222981965,
      "grad_norm": 0.4817352890968323,
      "learning_rate": 9.884169067904725e-06,
      "loss": 0.4634,
      "step": 5706
    },
    {
      "epoch": 0.2618971134872195,
      "grad_norm": 0.44964274764060974,
      "learning_rate": 9.884116591991677e-06,
      "loss": 0.3597,
      "step": 5707
    },
    {
      "epoch": 0.26194300399247394,
      "grad_norm": 0.44966939091682434,
      "learning_rate": 9.88406410433387e-06,
      "loss": 0.3964,
      "step": 5708
    },
    {
      "epoch": 0.26198889449772844,
      "grad_norm": 0.4539758563041687,
      "learning_rate": 9.88401160493143e-06,
      "loss": 0.3721,
      "step": 5709
    },
    {
      "epoch": 0.2620347850029829,
      "grad_norm": 0.4753742218017578,
      "learning_rate": 9.883959093784482e-06,
      "loss": 0.4485,
      "step": 5710
    },
    {
      "epoch": 0.2620806755082373,
      "grad_norm": 0.4653816223144531,
      "learning_rate": 9.883906570893158e-06,
      "loss": 0.4017,
      "step": 5711
    },
    {
      "epoch": 0.2621265660134918,
      "grad_norm": 0.513113796710968,
      "learning_rate": 9.883854036257577e-06,
      "loss": 0.5253,
      "step": 5712
    },
    {
      "epoch": 0.26217245651874627,
      "grad_norm": 0.45812666416168213,
      "learning_rate": 9.883801489877869e-06,
      "loss": 0.3516,
      "step": 5713
    },
    {
      "epoch": 0.2622183470240007,
      "grad_norm": 0.47847726941108704,
      "learning_rate": 9.88374893175416e-06,
      "loss": 0.4634,
      "step": 5714
    },
    {
      "epoch": 0.2622642375292552,
      "grad_norm": 0.507748007774353,
      "learning_rate": 9.883696361886577e-06,
      "loss": 0.5075,
      "step": 5715
    },
    {
      "epoch": 0.26231012803450965,
      "grad_norm": 0.44268330931663513,
      "learning_rate": 9.883643780275246e-06,
      "loss": 0.3411,
      "step": 5716
    },
    {
      "epoch": 0.26235601853976415,
      "grad_norm": 0.46061259508132935,
      "learning_rate": 9.88359118692029e-06,
      "loss": 0.382,
      "step": 5717
    },
    {
      "epoch": 0.2624019090450186,
      "grad_norm": 0.45478954911231995,
      "learning_rate": 9.883538581821841e-06,
      "loss": 0.441,
      "step": 5718
    },
    {
      "epoch": 0.26244779955027303,
      "grad_norm": 0.4721737504005432,
      "learning_rate": 9.883485964980022e-06,
      "loss": 0.3561,
      "step": 5719
    },
    {
      "epoch": 0.26249369005552753,
      "grad_norm": 0.46608126163482666,
      "learning_rate": 9.88343333639496e-06,
      "loss": 0.3668,
      "step": 5720
    },
    {
      "epoch": 0.26253958056078197,
      "grad_norm": 0.4553476572036743,
      "learning_rate": 9.883380696066783e-06,
      "loss": 0.3346,
      "step": 5721
    },
    {
      "epoch": 0.2625854710660364,
      "grad_norm": 0.4680670499801636,
      "learning_rate": 9.883328043995617e-06,
      "loss": 0.3955,
      "step": 5722
    },
    {
      "epoch": 0.2626313615712909,
      "grad_norm": 0.4598741829395294,
      "learning_rate": 9.883275380181588e-06,
      "loss": 0.3939,
      "step": 5723
    },
    {
      "epoch": 0.26267725207654535,
      "grad_norm": 0.4326989948749542,
      "learning_rate": 9.883222704624822e-06,
      "loss": 0.3301,
      "step": 5724
    },
    {
      "epoch": 0.26272314258179985,
      "grad_norm": 0.49923720955848694,
      "learning_rate": 9.883170017325447e-06,
      "loss": 0.4002,
      "step": 5725
    },
    {
      "epoch": 0.2627690330870543,
      "grad_norm": 0.47960779070854187,
      "learning_rate": 9.88311731828359e-06,
      "loss": 0.412,
      "step": 5726
    },
    {
      "epoch": 0.26281492359230874,
      "grad_norm": 0.45729711651802063,
      "learning_rate": 9.883064607499377e-06,
      "loss": 0.3864,
      "step": 5727
    },
    {
      "epoch": 0.26286081409756323,
      "grad_norm": 0.4778943657875061,
      "learning_rate": 9.883011884972934e-06,
      "loss": 0.3802,
      "step": 5728
    },
    {
      "epoch": 0.2629067046028177,
      "grad_norm": 0.5121933817863464,
      "learning_rate": 9.88295915070439e-06,
      "loss": 0.493,
      "step": 5729
    },
    {
      "epoch": 0.2629525951080721,
      "grad_norm": 0.4572793245315552,
      "learning_rate": 9.882906404693868e-06,
      "loss": 0.3748,
      "step": 5730
    },
    {
      "epoch": 0.2629984856133266,
      "grad_norm": 0.48729655146598816,
      "learning_rate": 9.8828536469415e-06,
      "loss": 0.387,
      "step": 5731
    },
    {
      "epoch": 0.26304437611858106,
      "grad_norm": 0.4959196150302887,
      "learning_rate": 9.882800877447406e-06,
      "loss": 0.459,
      "step": 5732
    },
    {
      "epoch": 0.26309026662383556,
      "grad_norm": 0.46004343032836914,
      "learning_rate": 9.88274809621172e-06,
      "loss": 0.3487,
      "step": 5733
    },
    {
      "epoch": 0.26313615712909,
      "grad_norm": 0.44322940707206726,
      "learning_rate": 9.882695303234567e-06,
      "loss": 0.3927,
      "step": 5734
    },
    {
      "epoch": 0.26318204763434444,
      "grad_norm": 0.4589797258377075,
      "learning_rate": 9.88264249851607e-06,
      "loss": 0.3698,
      "step": 5735
    },
    {
      "epoch": 0.26322793813959894,
      "grad_norm": 0.49748119711875916,
      "learning_rate": 9.882589682056361e-06,
      "loss": 0.4289,
      "step": 5736
    },
    {
      "epoch": 0.2632738286448534,
      "grad_norm": 0.4449711740016937,
      "learning_rate": 9.882536853855563e-06,
      "loss": 0.3617,
      "step": 5737
    },
    {
      "epoch": 0.2633197191501078,
      "grad_norm": 0.4532252252101898,
      "learning_rate": 9.882484013913804e-06,
      "loss": 0.3536,
      "step": 5738
    },
    {
      "epoch": 0.2633656096553623,
      "grad_norm": 0.4398576021194458,
      "learning_rate": 9.882431162231214e-06,
      "loss": 0.3379,
      "step": 5739
    },
    {
      "epoch": 0.26341150016061676,
      "grad_norm": 0.4621528387069702,
      "learning_rate": 9.882378298807918e-06,
      "loss": 0.3954,
      "step": 5740
    },
    {
      "epoch": 0.2634573906658712,
      "grad_norm": 0.4594041705131531,
      "learning_rate": 9.882325423644042e-06,
      "loss": 0.3696,
      "step": 5741
    },
    {
      "epoch": 0.2635032811711257,
      "grad_norm": 0.5007659792900085,
      "learning_rate": 9.882272536739715e-06,
      "loss": 0.4275,
      "step": 5742
    },
    {
      "epoch": 0.26354917167638015,
      "grad_norm": 0.48232123255729675,
      "learning_rate": 9.882219638095063e-06,
      "loss": 0.3846,
      "step": 5743
    },
    {
      "epoch": 0.26359506218163464,
      "grad_norm": 0.538147509098053,
      "learning_rate": 9.882166727710214e-06,
      "loss": 0.4239,
      "step": 5744
    },
    {
      "epoch": 0.2636409526868891,
      "grad_norm": 0.511863648891449,
      "learning_rate": 9.882113805585296e-06,
      "loss": 0.4344,
      "step": 5745
    },
    {
      "epoch": 0.2636868431921435,
      "grad_norm": 0.45896872878074646,
      "learning_rate": 9.882060871720434e-06,
      "loss": 0.3709,
      "step": 5746
    },
    {
      "epoch": 0.263732733697398,
      "grad_norm": 0.5214148163795471,
      "learning_rate": 9.882007926115755e-06,
      "loss": 0.4948,
      "step": 5747
    },
    {
      "epoch": 0.26377862420265247,
      "grad_norm": 0.46154317259788513,
      "learning_rate": 9.88195496877139e-06,
      "loss": 0.3642,
      "step": 5748
    },
    {
      "epoch": 0.2638245147079069,
      "grad_norm": 0.44234949350357056,
      "learning_rate": 9.881901999687463e-06,
      "loss": 0.3731,
      "step": 5749
    },
    {
      "epoch": 0.2638704052131614,
      "grad_norm": 0.5097475647926331,
      "learning_rate": 9.881849018864103e-06,
      "loss": 0.4874,
      "step": 5750
    },
    {
      "epoch": 0.26391629571841585,
      "grad_norm": 0.47292381525039673,
      "learning_rate": 9.881796026301435e-06,
      "loss": 0.4313,
      "step": 5751
    },
    {
      "epoch": 0.26396218622367035,
      "grad_norm": 0.43647173047065735,
      "learning_rate": 9.881743021999592e-06,
      "loss": 0.321,
      "step": 5752
    },
    {
      "epoch": 0.2640080767289248,
      "grad_norm": 0.41725602746009827,
      "learning_rate": 9.881690005958695e-06,
      "loss": 0.3319,
      "step": 5753
    },
    {
      "epoch": 0.26405396723417923,
      "grad_norm": 0.4627762734889984,
      "learning_rate": 9.881636978178874e-06,
      "loss": 0.3703,
      "step": 5754
    },
    {
      "epoch": 0.26409985773943373,
      "grad_norm": 0.423604816198349,
      "learning_rate": 9.881583938660257e-06,
      "loss": 0.3589,
      "step": 5755
    },
    {
      "epoch": 0.26414574824468817,
      "grad_norm": 0.45560187101364136,
      "learning_rate": 9.881530887402972e-06,
      "loss": 0.3751,
      "step": 5756
    },
    {
      "epoch": 0.2641916387499426,
      "grad_norm": 0.44976869225502014,
      "learning_rate": 9.881477824407146e-06,
      "loss": 0.4118,
      "step": 5757
    },
    {
      "epoch": 0.2642375292551971,
      "grad_norm": 0.4750780165195465,
      "learning_rate": 9.881424749672907e-06,
      "loss": 0.451,
      "step": 5758
    },
    {
      "epoch": 0.26428341976045155,
      "grad_norm": 0.42854002118110657,
      "learning_rate": 9.881371663200381e-06,
      "loss": 0.3403,
      "step": 5759
    },
    {
      "epoch": 0.26432931026570605,
      "grad_norm": 0.45952582359313965,
      "learning_rate": 9.881318564989698e-06,
      "loss": 0.4611,
      "step": 5760
    },
    {
      "epoch": 0.2643752007709605,
      "grad_norm": 0.48808106780052185,
      "learning_rate": 9.881265455040982e-06,
      "loss": 0.4437,
      "step": 5761
    },
    {
      "epoch": 0.26442109127621494,
      "grad_norm": 0.514225423336029,
      "learning_rate": 9.881212333354366e-06,
      "loss": 0.5328,
      "step": 5762
    },
    {
      "epoch": 0.26446698178146943,
      "grad_norm": 0.479399710893631,
      "learning_rate": 9.881159199929973e-06,
      "loss": 0.4818,
      "step": 5763
    },
    {
      "epoch": 0.2645128722867239,
      "grad_norm": 0.478593647480011,
      "learning_rate": 9.881106054767933e-06,
      "loss": 0.4653,
      "step": 5764
    },
    {
      "epoch": 0.2645587627919783,
      "grad_norm": 0.5184587240219116,
      "learning_rate": 9.881052897868375e-06,
      "loss": 0.567,
      "step": 5765
    },
    {
      "epoch": 0.2646046532972328,
      "grad_norm": 0.4631170630455017,
      "learning_rate": 9.880999729231423e-06,
      "loss": 0.3712,
      "step": 5766
    },
    {
      "epoch": 0.26465054380248726,
      "grad_norm": 0.4283306300640106,
      "learning_rate": 9.880946548857207e-06,
      "loss": 0.3075,
      "step": 5767
    },
    {
      "epoch": 0.2646964343077417,
      "grad_norm": 0.4697890281677246,
      "learning_rate": 9.880893356745857e-06,
      "loss": 0.4248,
      "step": 5768
    },
    {
      "epoch": 0.2647423248129962,
      "grad_norm": 0.49365106225013733,
      "learning_rate": 9.880840152897497e-06,
      "loss": 0.4529,
      "step": 5769
    },
    {
      "epoch": 0.26478821531825064,
      "grad_norm": 0.4644263982772827,
      "learning_rate": 9.880786937312259e-06,
      "loss": 0.4098,
      "step": 5770
    },
    {
      "epoch": 0.26483410582350514,
      "grad_norm": 0.4478089213371277,
      "learning_rate": 9.880733709990267e-06,
      "loss": 0.4018,
      "step": 5771
    },
    {
      "epoch": 0.2648799963287596,
      "grad_norm": 0.5120177865028381,
      "learning_rate": 9.880680470931652e-06,
      "loss": 0.4748,
      "step": 5772
    },
    {
      "epoch": 0.264925886834014,
      "grad_norm": 0.4669431746006012,
      "learning_rate": 9.88062722013654e-06,
      "loss": 0.3958,
      "step": 5773
    },
    {
      "epoch": 0.2649717773392685,
      "grad_norm": 0.45079270005226135,
      "learning_rate": 9.880573957605061e-06,
      "loss": 0.3406,
      "step": 5774
    },
    {
      "epoch": 0.26501766784452296,
      "grad_norm": 0.4865074157714844,
      "learning_rate": 9.880520683337341e-06,
      "loss": 0.489,
      "step": 5775
    },
    {
      "epoch": 0.2650635583497774,
      "grad_norm": 0.4581310451030731,
      "learning_rate": 9.880467397333509e-06,
      "loss": 0.4049,
      "step": 5776
    },
    {
      "epoch": 0.2651094488550319,
      "grad_norm": 0.44578155875205994,
      "learning_rate": 9.880414099593693e-06,
      "loss": 0.3336,
      "step": 5777
    },
    {
      "epoch": 0.26515533936028635,
      "grad_norm": 0.44198092818260193,
      "learning_rate": 9.880360790118024e-06,
      "loss": 0.3614,
      "step": 5778
    },
    {
      "epoch": 0.26520122986554084,
      "grad_norm": 0.46466711163520813,
      "learning_rate": 9.880307468906625e-06,
      "loss": 0.3915,
      "step": 5779
    },
    {
      "epoch": 0.2652471203707953,
      "grad_norm": 0.5257967710494995,
      "learning_rate": 9.880254135959628e-06,
      "loss": 0.5765,
      "step": 5780
    },
    {
      "epoch": 0.26529301087604973,
      "grad_norm": 0.4531961679458618,
      "learning_rate": 9.880200791277158e-06,
      "loss": 0.4019,
      "step": 5781
    },
    {
      "epoch": 0.2653389013813042,
      "grad_norm": 0.47527703642845154,
      "learning_rate": 9.880147434859348e-06,
      "loss": 0.4204,
      "step": 5782
    },
    {
      "epoch": 0.26538479188655867,
      "grad_norm": 0.4635518789291382,
      "learning_rate": 9.880094066706322e-06,
      "loss": 0.3913,
      "step": 5783
    },
    {
      "epoch": 0.2654306823918131,
      "grad_norm": 0.4819350242614746,
      "learning_rate": 9.880040686818212e-06,
      "loss": 0.318,
      "step": 5784
    },
    {
      "epoch": 0.2654765728970676,
      "grad_norm": 0.4777471721172333,
      "learning_rate": 9.87998729519514e-06,
      "loss": 0.4002,
      "step": 5785
    },
    {
      "epoch": 0.26552246340232205,
      "grad_norm": 0.4680997133255005,
      "learning_rate": 9.879933891837243e-06,
      "loss": 0.4685,
      "step": 5786
    },
    {
      "epoch": 0.26556835390757655,
      "grad_norm": 0.484408438205719,
      "learning_rate": 9.879880476744643e-06,
      "loss": 0.4164,
      "step": 5787
    },
    {
      "epoch": 0.265614244412831,
      "grad_norm": 0.48880401253700256,
      "learning_rate": 9.879827049917471e-06,
      "loss": 0.4097,
      "step": 5788
    },
    {
      "epoch": 0.26566013491808543,
      "grad_norm": 0.4620544910430908,
      "learning_rate": 9.879773611355856e-06,
      "loss": 0.3871,
      "step": 5789
    },
    {
      "epoch": 0.26570602542333993,
      "grad_norm": 0.5591461658477783,
      "learning_rate": 9.879720161059924e-06,
      "loss": 0.4778,
      "step": 5790
    },
    {
      "epoch": 0.2657519159285944,
      "grad_norm": 0.4735066890716553,
      "learning_rate": 9.879666699029806e-06,
      "loss": 0.3572,
      "step": 5791
    },
    {
      "epoch": 0.2657978064338488,
      "grad_norm": 0.4814257323741913,
      "learning_rate": 9.87961322526563e-06,
      "loss": 0.4181,
      "step": 5792
    },
    {
      "epoch": 0.2658436969391033,
      "grad_norm": 0.424625039100647,
      "learning_rate": 9.879559739767524e-06,
      "loss": 0.2849,
      "step": 5793
    },
    {
      "epoch": 0.26588958744435776,
      "grad_norm": 0.485614538192749,
      "learning_rate": 9.879506242535617e-06,
      "loss": 0.4594,
      "step": 5794
    },
    {
      "epoch": 0.2659354779496122,
      "grad_norm": 0.4749804735183716,
      "learning_rate": 9.879452733570037e-06,
      "loss": 0.4785,
      "step": 5795
    },
    {
      "epoch": 0.2659813684548667,
      "grad_norm": 0.5107864141464233,
      "learning_rate": 9.879399212870914e-06,
      "loss": 0.4985,
      "step": 5796
    },
    {
      "epoch": 0.26602725896012114,
      "grad_norm": 0.4326646625995636,
      "learning_rate": 9.879345680438375e-06,
      "loss": 0.3177,
      "step": 5797
    },
    {
      "epoch": 0.26607314946537564,
      "grad_norm": 0.44445276260375977,
      "learning_rate": 9.87929213627255e-06,
      "loss": 0.343,
      "step": 5798
    },
    {
      "epoch": 0.2661190399706301,
      "grad_norm": 0.7286478877067566,
      "learning_rate": 9.879238580373566e-06,
      "loss": 0.5213,
      "step": 5799
    },
    {
      "epoch": 0.2661649304758845,
      "grad_norm": 0.4173383414745331,
      "learning_rate": 9.879185012741555e-06,
      "loss": 0.3196,
      "step": 5800
    },
    {
      "epoch": 0.266210820981139,
      "grad_norm": 0.5137373208999634,
      "learning_rate": 9.879131433376643e-06,
      "loss": 0.4986,
      "step": 5801
    },
    {
      "epoch": 0.26625671148639346,
      "grad_norm": 0.46538951992988586,
      "learning_rate": 9.879077842278962e-06,
      "loss": 0.413,
      "step": 5802
    },
    {
      "epoch": 0.2663026019916479,
      "grad_norm": 0.45981860160827637,
      "learning_rate": 9.879024239448635e-06,
      "loss": 0.4034,
      "step": 5803
    },
    {
      "epoch": 0.2663484924969024,
      "grad_norm": 0.5179405212402344,
      "learning_rate": 9.878970624885798e-06,
      "loss": 0.4053,
      "step": 5804
    },
    {
      "epoch": 0.26639438300215684,
      "grad_norm": 0.4923546314239502,
      "learning_rate": 9.878916998590574e-06,
      "loss": 0.4159,
      "step": 5805
    },
    {
      "epoch": 0.26644027350741134,
      "grad_norm": 0.46567609906196594,
      "learning_rate": 9.878863360563096e-06,
      "loss": 0.4509,
      "step": 5806
    },
    {
      "epoch": 0.2664861640126658,
      "grad_norm": 0.5032748579978943,
      "learning_rate": 9.87880971080349e-06,
      "loss": 0.4966,
      "step": 5807
    },
    {
      "epoch": 0.2665320545179202,
      "grad_norm": 0.5178773403167725,
      "learning_rate": 9.878756049311887e-06,
      "loss": 0.4739,
      "step": 5808
    },
    {
      "epoch": 0.2665779450231747,
      "grad_norm": 0.49333029985427856,
      "learning_rate": 9.878702376088417e-06,
      "loss": 0.4365,
      "step": 5809
    },
    {
      "epoch": 0.26662383552842917,
      "grad_norm": 0.4771004617214203,
      "learning_rate": 9.878648691133206e-06,
      "loss": 0.4197,
      "step": 5810
    },
    {
      "epoch": 0.2666697260336836,
      "grad_norm": 0.49757635593414307,
      "learning_rate": 9.878594994446385e-06,
      "loss": 0.4898,
      "step": 5811
    },
    {
      "epoch": 0.2667156165389381,
      "grad_norm": 0.46155011653900146,
      "learning_rate": 9.878541286028082e-06,
      "loss": 0.3847,
      "step": 5812
    },
    {
      "epoch": 0.26676150704419255,
      "grad_norm": 0.4896308481693268,
      "learning_rate": 9.878487565878427e-06,
      "loss": 0.431,
      "step": 5813
    },
    {
      "epoch": 0.26680739754944705,
      "grad_norm": 0.45312026143074036,
      "learning_rate": 9.87843383399755e-06,
      "loss": 0.3733,
      "step": 5814
    },
    {
      "epoch": 0.2668532880547015,
      "grad_norm": 0.4678769111633301,
      "learning_rate": 9.878380090385578e-06,
      "loss": 0.399,
      "step": 5815
    },
    {
      "epoch": 0.26689917855995593,
      "grad_norm": 0.4658665060997009,
      "learning_rate": 9.878326335042643e-06,
      "loss": 0.4442,
      "step": 5816
    },
    {
      "epoch": 0.2669450690652104,
      "grad_norm": 0.504565417766571,
      "learning_rate": 9.878272567968871e-06,
      "loss": 0.396,
      "step": 5817
    },
    {
      "epoch": 0.26699095957046487,
      "grad_norm": 0.46499332785606384,
      "learning_rate": 9.878218789164396e-06,
      "loss": 0.4007,
      "step": 5818
    },
    {
      "epoch": 0.2670368500757193,
      "grad_norm": 0.47522398829460144,
      "learning_rate": 9.878164998629342e-06,
      "loss": 0.4289,
      "step": 5819
    },
    {
      "epoch": 0.2670827405809738,
      "grad_norm": 0.44240060448646545,
      "learning_rate": 9.878111196363843e-06,
      "loss": 0.3217,
      "step": 5820
    },
    {
      "epoch": 0.26712863108622825,
      "grad_norm": 0.4531952738761902,
      "learning_rate": 9.878057382368025e-06,
      "loss": 0.3433,
      "step": 5821
    },
    {
      "epoch": 0.26717452159148275,
      "grad_norm": 0.478973925113678,
      "learning_rate": 9.878003556642017e-06,
      "loss": 0.4489,
      "step": 5822
    },
    {
      "epoch": 0.2672204120967372,
      "grad_norm": 0.49470970034599304,
      "learning_rate": 9.877949719185951e-06,
      "loss": 0.4187,
      "step": 5823
    },
    {
      "epoch": 0.26726630260199163,
      "grad_norm": 0.4251794219017029,
      "learning_rate": 9.877895869999956e-06,
      "loss": 0.3585,
      "step": 5824
    },
    {
      "epoch": 0.26731219310724613,
      "grad_norm": 0.4552548825740814,
      "learning_rate": 9.877842009084163e-06,
      "loss": 0.4083,
      "step": 5825
    },
    {
      "epoch": 0.2673580836125006,
      "grad_norm": 0.46127310395240784,
      "learning_rate": 9.877788136438696e-06,
      "loss": 0.4271,
      "step": 5826
    },
    {
      "epoch": 0.267403974117755,
      "grad_norm": 0.4486490488052368,
      "learning_rate": 9.877734252063688e-06,
      "loss": 0.3841,
      "step": 5827
    },
    {
      "epoch": 0.2674498646230095,
      "grad_norm": 0.4565608501434326,
      "learning_rate": 9.877680355959271e-06,
      "loss": 0.3767,
      "step": 5828
    },
    {
      "epoch": 0.26749575512826396,
      "grad_norm": 0.49532851576805115,
      "learning_rate": 9.87762644812557e-06,
      "loss": 0.4108,
      "step": 5829
    },
    {
      "epoch": 0.2675416456335184,
      "grad_norm": 0.4849661886692047,
      "learning_rate": 9.877572528562718e-06,
      "loss": 0.4619,
      "step": 5830
    },
    {
      "epoch": 0.2675875361387729,
      "grad_norm": 0.4641399383544922,
      "learning_rate": 9.877518597270843e-06,
      "loss": 0.3853,
      "step": 5831
    },
    {
      "epoch": 0.26763342664402734,
      "grad_norm": 0.5094941258430481,
      "learning_rate": 9.877464654250074e-06,
      "loss": 0.483,
      "step": 5832
    },
    {
      "epoch": 0.26767931714928184,
      "grad_norm": 0.4141210913658142,
      "learning_rate": 9.877410699500545e-06,
      "loss": 0.3292,
      "step": 5833
    },
    {
      "epoch": 0.2677252076545363,
      "grad_norm": 0.4716409742832184,
      "learning_rate": 9.877356733022381e-06,
      "loss": 0.3975,
      "step": 5834
    },
    {
      "epoch": 0.2677710981597907,
      "grad_norm": 0.44029295444488525,
      "learning_rate": 9.877302754815713e-06,
      "loss": 0.3804,
      "step": 5835
    },
    {
      "epoch": 0.2678169886650452,
      "grad_norm": 0.4616566300392151,
      "learning_rate": 9.87724876488067e-06,
      "loss": 0.3806,
      "step": 5836
    },
    {
      "epoch": 0.26786287917029966,
      "grad_norm": 0.5055999755859375,
      "learning_rate": 9.877194763217387e-06,
      "loss": 0.4449,
      "step": 5837
    },
    {
      "epoch": 0.2679087696755541,
      "grad_norm": 0.45195257663726807,
      "learning_rate": 9.877140749825986e-06,
      "loss": 0.411,
      "step": 5838
    },
    {
      "epoch": 0.2679546601808086,
      "grad_norm": 0.45655742287635803,
      "learning_rate": 9.877086724706601e-06,
      "loss": 0.4326,
      "step": 5839
    },
    {
      "epoch": 0.26800055068606304,
      "grad_norm": 0.4880322515964508,
      "learning_rate": 9.877032687859364e-06,
      "loss": 0.487,
      "step": 5840
    },
    {
      "epoch": 0.26804644119131754,
      "grad_norm": 0.45049306750297546,
      "learning_rate": 9.876978639284402e-06,
      "loss": 0.374,
      "step": 5841
    },
    {
      "epoch": 0.268092331696572,
      "grad_norm": 0.45101916790008545,
      "learning_rate": 9.876924578981845e-06,
      "loss": 0.3572,
      "step": 5842
    },
    {
      "epoch": 0.2681382222018264,
      "grad_norm": 0.4441739022731781,
      "learning_rate": 9.876870506951823e-06,
      "loss": 0.4055,
      "step": 5843
    },
    {
      "epoch": 0.2681841127070809,
      "grad_norm": 0.5254185795783997,
      "learning_rate": 9.876816423194468e-06,
      "loss": 0.4843,
      "step": 5844
    },
    {
      "epoch": 0.26823000321233537,
      "grad_norm": 0.5126960277557373,
      "learning_rate": 9.876762327709908e-06,
      "loss": 0.448,
      "step": 5845
    },
    {
      "epoch": 0.2682758937175898,
      "grad_norm": 0.426925390958786,
      "learning_rate": 9.876708220498272e-06,
      "loss": 0.3594,
      "step": 5846
    },
    {
      "epoch": 0.2683217842228443,
      "grad_norm": 0.6193928718566895,
      "learning_rate": 9.876654101559695e-06,
      "loss": 0.5764,
      "step": 5847
    },
    {
      "epoch": 0.26836767472809875,
      "grad_norm": 0.4437464773654938,
      "learning_rate": 9.876599970894303e-06,
      "loss": 0.3566,
      "step": 5848
    },
    {
      "epoch": 0.26841356523335325,
      "grad_norm": 0.4854471683502197,
      "learning_rate": 9.876545828502227e-06,
      "loss": 0.3888,
      "step": 5849
    },
    {
      "epoch": 0.2684594557386077,
      "grad_norm": 0.4578111469745636,
      "learning_rate": 9.876491674383595e-06,
      "loss": 0.3887,
      "step": 5850
    },
    {
      "epoch": 0.26850534624386213,
      "grad_norm": 0.4687778353691101,
      "learning_rate": 9.876437508538543e-06,
      "loss": 0.4203,
      "step": 5851
    },
    {
      "epoch": 0.26855123674911663,
      "grad_norm": 0.4990726709365845,
      "learning_rate": 9.876383330967197e-06,
      "loss": 0.388,
      "step": 5852
    },
    {
      "epoch": 0.26859712725437107,
      "grad_norm": 0.5319591164588928,
      "learning_rate": 9.876329141669687e-06,
      "loss": 0.484,
      "step": 5853
    },
    {
      "epoch": 0.2686430177596255,
      "grad_norm": 0.4489472508430481,
      "learning_rate": 9.876274940646144e-06,
      "loss": 0.4378,
      "step": 5854
    },
    {
      "epoch": 0.26868890826488,
      "grad_norm": 0.4428797662258148,
      "learning_rate": 9.876220727896701e-06,
      "loss": 0.3298,
      "step": 5855
    },
    {
      "epoch": 0.26873479877013445,
      "grad_norm": 0.444782018661499,
      "learning_rate": 9.876166503421485e-06,
      "loss": 0.3861,
      "step": 5856
    },
    {
      "epoch": 0.2687806892753889,
      "grad_norm": 0.48130282759666443,
      "learning_rate": 9.876112267220627e-06,
      "loss": 0.419,
      "step": 5857
    },
    {
      "epoch": 0.2688265797806434,
      "grad_norm": 0.4787299931049347,
      "learning_rate": 9.876058019294257e-06,
      "loss": 0.4544,
      "step": 5858
    },
    {
      "epoch": 0.26887247028589784,
      "grad_norm": 0.4317842423915863,
      "learning_rate": 9.87600375964251e-06,
      "loss": 0.3802,
      "step": 5859
    },
    {
      "epoch": 0.26891836079115233,
      "grad_norm": 0.48446449637413025,
      "learning_rate": 9.875949488265509e-06,
      "loss": 0.3837,
      "step": 5860
    },
    {
      "epoch": 0.2689642512964068,
      "grad_norm": 0.46203431487083435,
      "learning_rate": 9.87589520516339e-06,
      "loss": 0.4113,
      "step": 5861
    },
    {
      "epoch": 0.2690101418016612,
      "grad_norm": 0.4458620548248291,
      "learning_rate": 9.875840910336282e-06,
      "loss": 0.3601,
      "step": 5862
    },
    {
      "epoch": 0.2690560323069157,
      "grad_norm": 0.49450910091400146,
      "learning_rate": 9.875786603784317e-06,
      "loss": 0.3779,
      "step": 5863
    },
    {
      "epoch": 0.26910192281217016,
      "grad_norm": 0.4460464119911194,
      "learning_rate": 9.875732285507622e-06,
      "loss": 0.3369,
      "step": 5864
    },
    {
      "epoch": 0.2691478133174246,
      "grad_norm": 0.468208372592926,
      "learning_rate": 9.875677955506329e-06,
      "loss": 0.3797,
      "step": 5865
    },
    {
      "epoch": 0.2691937038226791,
      "grad_norm": 0.4886190891265869,
      "learning_rate": 9.875623613780572e-06,
      "loss": 0.3437,
      "step": 5866
    },
    {
      "epoch": 0.26923959432793354,
      "grad_norm": 0.4577125012874603,
      "learning_rate": 9.875569260330476e-06,
      "loss": 0.4853,
      "step": 5867
    },
    {
      "epoch": 0.26928548483318804,
      "grad_norm": 0.4669910669326782,
      "learning_rate": 9.875514895156176e-06,
      "loss": 0.429,
      "step": 5868
    },
    {
      "epoch": 0.2693313753384425,
      "grad_norm": 0.4648025333881378,
      "learning_rate": 9.875460518257803e-06,
      "loss": 0.3617,
      "step": 5869
    },
    {
      "epoch": 0.2693772658436969,
      "grad_norm": 0.44554293155670166,
      "learning_rate": 9.875406129635484e-06,
      "loss": 0.369,
      "step": 5870
    },
    {
      "epoch": 0.2694231563489514,
      "grad_norm": 0.4423474073410034,
      "learning_rate": 9.875351729289353e-06,
      "loss": 0.3745,
      "step": 5871
    },
    {
      "epoch": 0.26946904685420586,
      "grad_norm": 0.5084224939346313,
      "learning_rate": 9.87529731721954e-06,
      "loss": 0.4447,
      "step": 5872
    },
    {
      "epoch": 0.2695149373594603,
      "grad_norm": 0.45994704961776733,
      "learning_rate": 9.875242893426175e-06,
      "loss": 0.3462,
      "step": 5873
    },
    {
      "epoch": 0.2695608278647148,
      "grad_norm": 0.47925201058387756,
      "learning_rate": 9.87518845790939e-06,
      "loss": 0.4285,
      "step": 5874
    },
    {
      "epoch": 0.26960671836996924,
      "grad_norm": 0.6300848126411438,
      "learning_rate": 9.875134010669314e-06,
      "loss": 0.3816,
      "step": 5875
    },
    {
      "epoch": 0.26965260887522374,
      "grad_norm": 0.4699513018131256,
      "learning_rate": 9.875079551706081e-06,
      "loss": 0.4287,
      "step": 5876
    },
    {
      "epoch": 0.2696984993804782,
      "grad_norm": 0.4392528533935547,
      "learning_rate": 9.87502508101982e-06,
      "loss": 0.3791,
      "step": 5877
    },
    {
      "epoch": 0.2697443898857326,
      "grad_norm": 0.46475958824157715,
      "learning_rate": 9.874970598610661e-06,
      "loss": 0.3803,
      "step": 5878
    },
    {
      "epoch": 0.2697902803909871,
      "grad_norm": 0.47555962204933167,
      "learning_rate": 9.874916104478737e-06,
      "loss": 0.443,
      "step": 5879
    },
    {
      "epoch": 0.26983617089624157,
      "grad_norm": 0.4785141944885254,
      "learning_rate": 9.874861598624179e-06,
      "loss": 0.3838,
      "step": 5880
    },
    {
      "epoch": 0.269882061401496,
      "grad_norm": 0.48035871982574463,
      "learning_rate": 9.874807081047118e-06,
      "loss": 0.4224,
      "step": 5881
    },
    {
      "epoch": 0.2699279519067505,
      "grad_norm": 0.4470847547054291,
      "learning_rate": 9.874752551747685e-06,
      "loss": 0.402,
      "step": 5882
    },
    {
      "epoch": 0.26997384241200495,
      "grad_norm": 0.45945751667022705,
      "learning_rate": 9.874698010726007e-06,
      "loss": 0.4107,
      "step": 5883
    },
    {
      "epoch": 0.2700197329172594,
      "grad_norm": 0.4976825714111328,
      "learning_rate": 9.874643457982222e-06,
      "loss": 0.432,
      "step": 5884
    },
    {
      "epoch": 0.2700656234225139,
      "grad_norm": 0.4576255679130554,
      "learning_rate": 9.874588893516458e-06,
      "loss": 0.4294,
      "step": 5885
    },
    {
      "epoch": 0.27011151392776833,
      "grad_norm": 0.45117494463920593,
      "learning_rate": 9.874534317328844e-06,
      "loss": 0.3383,
      "step": 5886
    },
    {
      "epoch": 0.27015740443302283,
      "grad_norm": 0.4742228388786316,
      "learning_rate": 9.874479729419515e-06,
      "loss": 0.4182,
      "step": 5887
    },
    {
      "epoch": 0.27020329493827727,
      "grad_norm": 0.4628211557865143,
      "learning_rate": 9.8744251297886e-06,
      "loss": 0.3856,
      "step": 5888
    },
    {
      "epoch": 0.2702491854435317,
      "grad_norm": 0.4697141647338867,
      "learning_rate": 9.874370518436232e-06,
      "loss": 0.3867,
      "step": 5889
    },
    {
      "epoch": 0.2702950759487862,
      "grad_norm": 0.5377175807952881,
      "learning_rate": 9.874315895362539e-06,
      "loss": 0.5391,
      "step": 5890
    },
    {
      "epoch": 0.27034096645404065,
      "grad_norm": 0.47244906425476074,
      "learning_rate": 9.874261260567656e-06,
      "loss": 0.4338,
      "step": 5891
    },
    {
      "epoch": 0.2703868569592951,
      "grad_norm": 0.4866069555282593,
      "learning_rate": 9.874206614051713e-06,
      "loss": 0.406,
      "step": 5892
    },
    {
      "epoch": 0.2704327474645496,
      "grad_norm": 0.44471365213394165,
      "learning_rate": 9.874151955814843e-06,
      "loss": 0.4138,
      "step": 5893
    },
    {
      "epoch": 0.27047863796980404,
      "grad_norm": 0.45400270819664,
      "learning_rate": 9.874097285857173e-06,
      "loss": 0.3691,
      "step": 5894
    },
    {
      "epoch": 0.27052452847505853,
      "grad_norm": 0.5032569169998169,
      "learning_rate": 9.874042604178839e-06,
      "loss": 0.536,
      "step": 5895
    },
    {
      "epoch": 0.270570418980313,
      "grad_norm": 0.47621259093284607,
      "learning_rate": 9.873987910779971e-06,
      "loss": 0.3677,
      "step": 5896
    },
    {
      "epoch": 0.2706163094855674,
      "grad_norm": 0.4404245913028717,
      "learning_rate": 9.873933205660701e-06,
      "loss": 0.3465,
      "step": 5897
    },
    {
      "epoch": 0.2706621999908219,
      "grad_norm": 0.4451431334018707,
      "learning_rate": 9.873878488821157e-06,
      "loss": 0.4091,
      "step": 5898
    },
    {
      "epoch": 0.27070809049607636,
      "grad_norm": 0.524995744228363,
      "learning_rate": 9.873823760261476e-06,
      "loss": 0.43,
      "step": 5899
    },
    {
      "epoch": 0.2707539810013308,
      "grad_norm": 0.49389970302581787,
      "learning_rate": 9.873769019981787e-06,
      "loss": 0.4485,
      "step": 5900
    },
    {
      "epoch": 0.2707998715065853,
      "grad_norm": 0.899696946144104,
      "learning_rate": 9.873714267982222e-06,
      "loss": 0.4961,
      "step": 5901
    },
    {
      "epoch": 0.27084576201183974,
      "grad_norm": 0.5391474366188049,
      "learning_rate": 9.873659504262911e-06,
      "loss": 0.4806,
      "step": 5902
    },
    {
      "epoch": 0.27089165251709424,
      "grad_norm": 0.4806353449821472,
      "learning_rate": 9.873604728823988e-06,
      "loss": 0.4258,
      "step": 5903
    },
    {
      "epoch": 0.2709375430223487,
      "grad_norm": 0.47695058584213257,
      "learning_rate": 9.873549941665583e-06,
      "loss": 0.3707,
      "step": 5904
    },
    {
      "epoch": 0.2709834335276031,
      "grad_norm": 0.5217123031616211,
      "learning_rate": 9.87349514278783e-06,
      "loss": 0.4838,
      "step": 5905
    },
    {
      "epoch": 0.2710293240328576,
      "grad_norm": 0.4487816393375397,
      "learning_rate": 9.873440332190857e-06,
      "loss": 0.3887,
      "step": 5906
    },
    {
      "epoch": 0.27107521453811206,
      "grad_norm": 0.47574514150619507,
      "learning_rate": 9.8733855098748e-06,
      "loss": 0.3999,
      "step": 5907
    },
    {
      "epoch": 0.2711211050433665,
      "grad_norm": 0.48733723163604736,
      "learning_rate": 9.873330675839787e-06,
      "loss": 0.4528,
      "step": 5908
    },
    {
      "epoch": 0.271166995548621,
      "grad_norm": 0.4945688843727112,
      "learning_rate": 9.873275830085953e-06,
      "loss": 0.4224,
      "step": 5909
    },
    {
      "epoch": 0.27121288605387545,
      "grad_norm": 0.4674447476863861,
      "learning_rate": 9.87322097261343e-06,
      "loss": 0.3548,
      "step": 5910
    },
    {
      "epoch": 0.2712587765591299,
      "grad_norm": 0.445092111825943,
      "learning_rate": 9.873166103422348e-06,
      "loss": 0.3592,
      "step": 5911
    },
    {
      "epoch": 0.2713046670643844,
      "grad_norm": 0.4449789226055145,
      "learning_rate": 9.873111222512838e-06,
      "loss": 0.3496,
      "step": 5912
    },
    {
      "epoch": 0.27135055756963883,
      "grad_norm": 0.5025137066841125,
      "learning_rate": 9.873056329885033e-06,
      "loss": 0.3531,
      "step": 5913
    },
    {
      "epoch": 0.2713964480748933,
      "grad_norm": 0.4014642536640167,
      "learning_rate": 9.873001425539068e-06,
      "loss": 0.3382,
      "step": 5914
    },
    {
      "epoch": 0.27144233858014777,
      "grad_norm": 0.45683935284614563,
      "learning_rate": 9.872946509475072e-06,
      "loss": 0.3925,
      "step": 5915
    },
    {
      "epoch": 0.2714882290854022,
      "grad_norm": 0.507476270198822,
      "learning_rate": 9.872891581693175e-06,
      "loss": 0.4704,
      "step": 5916
    },
    {
      "epoch": 0.2715341195906567,
      "grad_norm": 0.45307326316833496,
      "learning_rate": 9.872836642193514e-06,
      "loss": 0.4062,
      "step": 5917
    },
    {
      "epoch": 0.27158001009591115,
      "grad_norm": 0.45223546028137207,
      "learning_rate": 9.87278169097622e-06,
      "loss": 0.3459,
      "step": 5918
    },
    {
      "epoch": 0.2716259006011656,
      "grad_norm": 0.49584969878196716,
      "learning_rate": 9.872726728041422e-06,
      "loss": 0.4939,
      "step": 5919
    },
    {
      "epoch": 0.2716717911064201,
      "grad_norm": 0.45140790939331055,
      "learning_rate": 9.872671753389254e-06,
      "loss": 0.358,
      "step": 5920
    },
    {
      "epoch": 0.27171768161167453,
      "grad_norm": 0.4547443389892578,
      "learning_rate": 9.87261676701985e-06,
      "loss": 0.4085,
      "step": 5921
    },
    {
      "epoch": 0.27176357211692903,
      "grad_norm": 0.4753084182739258,
      "learning_rate": 9.87256176893334e-06,
      "loss": 0.4668,
      "step": 5922
    },
    {
      "epoch": 0.2718094626221835,
      "grad_norm": 0.45439091324806213,
      "learning_rate": 9.872506759129856e-06,
      "loss": 0.3493,
      "step": 5923
    },
    {
      "epoch": 0.2718553531274379,
      "grad_norm": 0.45839372277259827,
      "learning_rate": 9.87245173760953e-06,
      "loss": 0.3633,
      "step": 5924
    },
    {
      "epoch": 0.2719012436326924,
      "grad_norm": 0.5432732105255127,
      "learning_rate": 9.872396704372497e-06,
      "loss": 0.474,
      "step": 5925
    },
    {
      "epoch": 0.27194713413794686,
      "grad_norm": 0.474185585975647,
      "learning_rate": 9.872341659418887e-06,
      "loss": 0.3808,
      "step": 5926
    },
    {
      "epoch": 0.2719930246432013,
      "grad_norm": 0.47217246890068054,
      "learning_rate": 9.872286602748834e-06,
      "loss": 0.432,
      "step": 5927
    },
    {
      "epoch": 0.2720389151484558,
      "grad_norm": 0.47725555300712585,
      "learning_rate": 9.872231534362468e-06,
      "loss": 0.396,
      "step": 5928
    },
    {
      "epoch": 0.27208480565371024,
      "grad_norm": 0.504806637763977,
      "learning_rate": 9.872176454259924e-06,
      "loss": 0.4386,
      "step": 5929
    },
    {
      "epoch": 0.27213069615896474,
      "grad_norm": 0.4668581485748291,
      "learning_rate": 9.872121362441331e-06,
      "loss": 0.3863,
      "step": 5930
    },
    {
      "epoch": 0.2721765866642192,
      "grad_norm": 0.48398470878601074,
      "learning_rate": 9.872066258906827e-06,
      "loss": 0.3986,
      "step": 5931
    },
    {
      "epoch": 0.2722224771694736,
      "grad_norm": 0.482972115278244,
      "learning_rate": 9.87201114365654e-06,
      "loss": 0.3975,
      "step": 5932
    },
    {
      "epoch": 0.2722683676747281,
      "grad_norm": 0.45084044337272644,
      "learning_rate": 9.871956016690604e-06,
      "loss": 0.4015,
      "step": 5933
    },
    {
      "epoch": 0.27231425817998256,
      "grad_norm": 0.46074631810188293,
      "learning_rate": 9.87190087800915e-06,
      "loss": 0.4036,
      "step": 5934
    },
    {
      "epoch": 0.272360148685237,
      "grad_norm": 0.5191996097564697,
      "learning_rate": 9.871845727612316e-06,
      "loss": 0.4689,
      "step": 5935
    },
    {
      "epoch": 0.2724060391904915,
      "grad_norm": 0.5497806668281555,
      "learning_rate": 9.871790565500226e-06,
      "loss": 0.525,
      "step": 5936
    },
    {
      "epoch": 0.27245192969574594,
      "grad_norm": 0.4974724054336548,
      "learning_rate": 9.87173539167302e-06,
      "loss": 0.4359,
      "step": 5937
    },
    {
      "epoch": 0.27249782020100044,
      "grad_norm": 0.46319347620010376,
      "learning_rate": 9.871680206130827e-06,
      "loss": 0.4271,
      "step": 5938
    },
    {
      "epoch": 0.2725437107062549,
      "grad_norm": 0.4652736485004425,
      "learning_rate": 9.87162500887378e-06,
      "loss": 0.3636,
      "step": 5939
    },
    {
      "epoch": 0.2725896012115093,
      "grad_norm": 0.49670347571372986,
      "learning_rate": 9.871569799902014e-06,
      "loss": 0.4716,
      "step": 5940
    },
    {
      "epoch": 0.2726354917167638,
      "grad_norm": 0.4571923613548279,
      "learning_rate": 9.87151457921566e-06,
      "loss": 0.3567,
      "step": 5941
    },
    {
      "epoch": 0.27268138222201826,
      "grad_norm": 0.45072683691978455,
      "learning_rate": 9.871459346814848e-06,
      "loss": 0.4093,
      "step": 5942
    },
    {
      "epoch": 0.2727272727272727,
      "grad_norm": 0.4656715393066406,
      "learning_rate": 9.871404102699717e-06,
      "loss": 0.393,
      "step": 5943
    },
    {
      "epoch": 0.2727731632325272,
      "grad_norm": 0.46393948793411255,
      "learning_rate": 9.871348846870396e-06,
      "loss": 0.4274,
      "step": 5944
    },
    {
      "epoch": 0.27281905373778165,
      "grad_norm": 0.45287051796913147,
      "learning_rate": 9.871293579327018e-06,
      "loss": 0.3387,
      "step": 5945
    },
    {
      "epoch": 0.2728649442430361,
      "grad_norm": 0.484746515750885,
      "learning_rate": 9.871238300069716e-06,
      "loss": 0.4234,
      "step": 5946
    },
    {
      "epoch": 0.2729108347482906,
      "grad_norm": 0.4816627502441406,
      "learning_rate": 9.871183009098624e-06,
      "loss": 0.419,
      "step": 5947
    },
    {
      "epoch": 0.27295672525354503,
      "grad_norm": 0.509975790977478,
      "learning_rate": 9.871127706413874e-06,
      "loss": 0.4628,
      "step": 5948
    },
    {
      "epoch": 0.2730026157587995,
      "grad_norm": 0.49830472469329834,
      "learning_rate": 9.871072392015599e-06,
      "loss": 0.5383,
      "step": 5949
    },
    {
      "epoch": 0.27304850626405397,
      "grad_norm": 0.43803879618644714,
      "learning_rate": 9.871017065903934e-06,
      "loss": 0.3549,
      "step": 5950
    },
    {
      "epoch": 0.2730943967693084,
      "grad_norm": 0.46389350295066833,
      "learning_rate": 9.870961728079007e-06,
      "loss": 0.3929,
      "step": 5951
    },
    {
      "epoch": 0.2731402872745629,
      "grad_norm": 0.43711918592453003,
      "learning_rate": 9.870906378540956e-06,
      "loss": 0.3382,
      "step": 5952
    },
    {
      "epoch": 0.27318617777981735,
      "grad_norm": 0.456473171710968,
      "learning_rate": 9.870851017289913e-06,
      "loss": 0.4041,
      "step": 5953
    },
    {
      "epoch": 0.2732320682850718,
      "grad_norm": 0.4576350152492523,
      "learning_rate": 9.87079564432601e-06,
      "loss": 0.359,
      "step": 5954
    },
    {
      "epoch": 0.2732779587903263,
      "grad_norm": 0.5189813375473022,
      "learning_rate": 9.870740259649381e-06,
      "loss": 0.5216,
      "step": 5955
    },
    {
      "epoch": 0.27332384929558073,
      "grad_norm": 0.4453928470611572,
      "learning_rate": 9.87068486326016e-06,
      "loss": 0.341,
      "step": 5956
    },
    {
      "epoch": 0.27336973980083523,
      "grad_norm": 0.44720685482025146,
      "learning_rate": 9.870629455158478e-06,
      "loss": 0.3877,
      "step": 5957
    },
    {
      "epoch": 0.2734156303060897,
      "grad_norm": 0.4457598924636841,
      "learning_rate": 9.87057403534447e-06,
      "loss": 0.3428,
      "step": 5958
    },
    {
      "epoch": 0.2734615208113441,
      "grad_norm": 0.492818683385849,
      "learning_rate": 9.870518603818267e-06,
      "loss": 0.4015,
      "step": 5959
    },
    {
      "epoch": 0.2735074113165986,
      "grad_norm": 0.48952701687812805,
      "learning_rate": 9.870463160580005e-06,
      "loss": 0.4396,
      "step": 5960
    },
    {
      "epoch": 0.27355330182185306,
      "grad_norm": 0.4919719994068146,
      "learning_rate": 9.870407705629816e-06,
      "loss": 0.4305,
      "step": 5961
    },
    {
      "epoch": 0.2735991923271075,
      "grad_norm": 0.48953744769096375,
      "learning_rate": 9.870352238967834e-06,
      "loss": 0.4562,
      "step": 5962
    },
    {
      "epoch": 0.273645082832362,
      "grad_norm": 0.5031746625900269,
      "learning_rate": 9.87029676059419e-06,
      "loss": 0.434,
      "step": 5963
    },
    {
      "epoch": 0.27369097333761644,
      "grad_norm": 0.6283437013626099,
      "learning_rate": 9.870241270509023e-06,
      "loss": 0.5591,
      "step": 5964
    },
    {
      "epoch": 0.27373686384287094,
      "grad_norm": 0.4421950876712799,
      "learning_rate": 9.87018576871246e-06,
      "loss": 0.3357,
      "step": 5965
    },
    {
      "epoch": 0.2737827543481254,
      "grad_norm": 0.4651041626930237,
      "learning_rate": 9.87013025520464e-06,
      "loss": 0.3752,
      "step": 5966
    },
    {
      "epoch": 0.2738286448533798,
      "grad_norm": 0.5725246667861938,
      "learning_rate": 9.870074729985692e-06,
      "loss": 0.4919,
      "step": 5967
    },
    {
      "epoch": 0.2738745353586343,
      "grad_norm": 0.5128257870674133,
      "learning_rate": 9.870019193055752e-06,
      "loss": 0.4658,
      "step": 5968
    },
    {
      "epoch": 0.27392042586388876,
      "grad_norm": 0.4714244604110718,
      "learning_rate": 9.869963644414952e-06,
      "loss": 0.3826,
      "step": 5969
    },
    {
      "epoch": 0.2739663163691432,
      "grad_norm": 0.45479047298431396,
      "learning_rate": 9.869908084063426e-06,
      "loss": 0.3877,
      "step": 5970
    },
    {
      "epoch": 0.2740122068743977,
      "grad_norm": 0.6153403520584106,
      "learning_rate": 9.869852512001309e-06,
      "loss": 0.4738,
      "step": 5971
    },
    {
      "epoch": 0.27405809737965214,
      "grad_norm": 0.4815972149372101,
      "learning_rate": 9.869796928228733e-06,
      "loss": 0.3445,
      "step": 5972
    },
    {
      "epoch": 0.2741039878849066,
      "grad_norm": 0.4348076283931732,
      "learning_rate": 9.869741332745833e-06,
      "loss": 0.3542,
      "step": 5973
    },
    {
      "epoch": 0.2741498783901611,
      "grad_norm": 0.47442400455474854,
      "learning_rate": 9.869685725552743e-06,
      "loss": 0.3773,
      "step": 5974
    },
    {
      "epoch": 0.2741957688954155,
      "grad_norm": 0.48906534910202026,
      "learning_rate": 9.869630106649593e-06,
      "loss": 0.5121,
      "step": 5975
    },
    {
      "epoch": 0.27424165940067,
      "grad_norm": 0.4835379719734192,
      "learning_rate": 9.869574476036521e-06,
      "loss": 0.4297,
      "step": 5976
    },
    {
      "epoch": 0.27428754990592447,
      "grad_norm": 0.49576854705810547,
      "learning_rate": 9.869518833713661e-06,
      "loss": 0.4579,
      "step": 5977
    },
    {
      "epoch": 0.2743334404111789,
      "grad_norm": 0.45131343603134155,
      "learning_rate": 9.869463179681143e-06,
      "loss": 0.4173,
      "step": 5978
    },
    {
      "epoch": 0.2743793309164334,
      "grad_norm": 0.5188154578208923,
      "learning_rate": 9.869407513939102e-06,
      "loss": 0.4362,
      "step": 5979
    },
    {
      "epoch": 0.27442522142168785,
      "grad_norm": 0.4979189336299896,
      "learning_rate": 9.869351836487675e-06,
      "loss": 0.4805,
      "step": 5980
    },
    {
      "epoch": 0.2744711119269423,
      "grad_norm": 0.4617033898830414,
      "learning_rate": 9.869296147326991e-06,
      "loss": 0.4108,
      "step": 5981
    },
    {
      "epoch": 0.2745170024321968,
      "grad_norm": 0.4562774896621704,
      "learning_rate": 9.86924044645719e-06,
      "loss": 0.3971,
      "step": 5982
    },
    {
      "epoch": 0.27456289293745123,
      "grad_norm": 0.4326223134994507,
      "learning_rate": 9.8691847338784e-06,
      "loss": 0.3449,
      "step": 5983
    },
    {
      "epoch": 0.27460878344270573,
      "grad_norm": 0.45280131697654724,
      "learning_rate": 9.869129009590759e-06,
      "loss": 0.4136,
      "step": 5984
    },
    {
      "epoch": 0.27465467394796017,
      "grad_norm": 0.4537236988544464,
      "learning_rate": 9.869073273594398e-06,
      "loss": 0.3549,
      "step": 5985
    },
    {
      "epoch": 0.2747005644532146,
      "grad_norm": 0.44010278582572937,
      "learning_rate": 9.869017525889455e-06,
      "loss": 0.4083,
      "step": 5986
    },
    {
      "epoch": 0.2747464549584691,
      "grad_norm": 0.44472166895866394,
      "learning_rate": 9.86896176647606e-06,
      "loss": 0.3514,
      "step": 5987
    },
    {
      "epoch": 0.27479234546372355,
      "grad_norm": 0.4771850109100342,
      "learning_rate": 9.868905995354347e-06,
      "loss": 0.3602,
      "step": 5988
    },
    {
      "epoch": 0.274838235968978,
      "grad_norm": 0.4713582694530487,
      "learning_rate": 9.868850212524454e-06,
      "loss": 0.3913,
      "step": 5989
    },
    {
      "epoch": 0.2748841264742325,
      "grad_norm": 0.48855167627334595,
      "learning_rate": 9.868794417986512e-06,
      "loss": 0.4371,
      "step": 5990
    },
    {
      "epoch": 0.27493001697948694,
      "grad_norm": 0.4928397834300995,
      "learning_rate": 9.868738611740655e-06,
      "loss": 0.4089,
      "step": 5991
    },
    {
      "epoch": 0.27497590748474143,
      "grad_norm": 0.48676592111587524,
      "learning_rate": 9.86868279378702e-06,
      "loss": 0.4136,
      "step": 5992
    },
    {
      "epoch": 0.2750217979899959,
      "grad_norm": 0.4864250123500824,
      "learning_rate": 9.868626964125738e-06,
      "loss": 0.3802,
      "step": 5993
    },
    {
      "epoch": 0.2750676884952503,
      "grad_norm": 0.44911256432533264,
      "learning_rate": 9.868571122756948e-06,
      "loss": 0.4073,
      "step": 5994
    },
    {
      "epoch": 0.2751135790005048,
      "grad_norm": 0.5066159963607788,
      "learning_rate": 9.868515269680777e-06,
      "loss": 0.4672,
      "step": 5995
    },
    {
      "epoch": 0.27515946950575926,
      "grad_norm": 0.4743802547454834,
      "learning_rate": 9.868459404897366e-06,
      "loss": 0.4571,
      "step": 5996
    },
    {
      "epoch": 0.2752053600110137,
      "grad_norm": 0.44951435923576355,
      "learning_rate": 9.868403528406846e-06,
      "loss": 0.3973,
      "step": 5997
    },
    {
      "epoch": 0.2752512505162682,
      "grad_norm": 0.4462641775608063,
      "learning_rate": 9.86834764020935e-06,
      "loss": 0.3708,
      "step": 5998
    },
    {
      "epoch": 0.27529714102152264,
      "grad_norm": 0.452903687953949,
      "learning_rate": 9.868291740305016e-06,
      "loss": 0.3918,
      "step": 5999
    },
    {
      "epoch": 0.2753430315267771,
      "grad_norm": 0.48993316292762756,
      "learning_rate": 9.868235828693976e-06,
      "loss": 0.4715,
      "step": 6000
    },
    {
      "epoch": 0.2753889220320316,
      "grad_norm": 0.46686631441116333,
      "learning_rate": 9.868179905376367e-06,
      "loss": 0.3712,
      "step": 6001
    },
    {
      "epoch": 0.275434812537286,
      "grad_norm": 0.45403245091438293,
      "learning_rate": 9.86812397035232e-06,
      "loss": 0.4324,
      "step": 6002
    },
    {
      "epoch": 0.2754807030425405,
      "grad_norm": 0.43723514676094055,
      "learning_rate": 9.868068023621974e-06,
      "loss": 0.358,
      "step": 6003
    },
    {
      "epoch": 0.27552659354779496,
      "grad_norm": 0.4659489691257477,
      "learning_rate": 9.868012065185459e-06,
      "loss": 0.3852,
      "step": 6004
    },
    {
      "epoch": 0.2755724840530494,
      "grad_norm": 0.46562132239341736,
      "learning_rate": 9.867956095042911e-06,
      "loss": 0.3726,
      "step": 6005
    },
    {
      "epoch": 0.2756183745583039,
      "grad_norm": 0.4423086941242218,
      "learning_rate": 9.867900113194467e-06,
      "loss": 0.3031,
      "step": 6006
    },
    {
      "epoch": 0.27566426506355834,
      "grad_norm": 0.46174925565719604,
      "learning_rate": 9.867844119640257e-06,
      "loss": 0.3695,
      "step": 6007
    },
    {
      "epoch": 0.2757101555688128,
      "grad_norm": 0.43979412317276,
      "learning_rate": 9.867788114380418e-06,
      "loss": 0.3412,
      "step": 6008
    },
    {
      "epoch": 0.2757560460740673,
      "grad_norm": 0.44853121042251587,
      "learning_rate": 9.867732097415087e-06,
      "loss": 0.3702,
      "step": 6009
    },
    {
      "epoch": 0.2758019365793217,
      "grad_norm": 0.47380557656288147,
      "learning_rate": 9.867676068744396e-06,
      "loss": 0.3724,
      "step": 6010
    },
    {
      "epoch": 0.2758478270845762,
      "grad_norm": 0.45100897550582886,
      "learning_rate": 9.86762002836848e-06,
      "loss": 0.3805,
      "step": 6011
    },
    {
      "epoch": 0.27589371758983067,
      "grad_norm": 0.47433117032051086,
      "learning_rate": 9.867563976287474e-06,
      "loss": 0.4351,
      "step": 6012
    },
    {
      "epoch": 0.2759396080950851,
      "grad_norm": 0.4508553147315979,
      "learning_rate": 9.867507912501512e-06,
      "loss": 0.3911,
      "step": 6013
    },
    {
      "epoch": 0.2759854986003396,
      "grad_norm": 0.455611914396286,
      "learning_rate": 9.867451837010732e-06,
      "loss": 0.41,
      "step": 6014
    },
    {
      "epoch": 0.27603138910559405,
      "grad_norm": 0.48005589842796326,
      "learning_rate": 9.867395749815265e-06,
      "loss": 0.3878,
      "step": 6015
    },
    {
      "epoch": 0.2760772796108485,
      "grad_norm": 0.4138370156288147,
      "learning_rate": 9.867339650915248e-06,
      "loss": 0.3232,
      "step": 6016
    },
    {
      "epoch": 0.276123170116103,
      "grad_norm": 0.48386910557746887,
      "learning_rate": 9.867283540310814e-06,
      "loss": 0.3999,
      "step": 6017
    },
    {
      "epoch": 0.27616906062135743,
      "grad_norm": 0.44743290543556213,
      "learning_rate": 9.867227418002101e-06,
      "loss": 0.3728,
      "step": 6018
    },
    {
      "epoch": 0.27621495112661193,
      "grad_norm": 0.4604841470718384,
      "learning_rate": 9.86717128398924e-06,
      "loss": 0.3521,
      "step": 6019
    },
    {
      "epoch": 0.27626084163186637,
      "grad_norm": 0.4918363690376282,
      "learning_rate": 9.867115138272372e-06,
      "loss": 0.4684,
      "step": 6020
    },
    {
      "epoch": 0.2763067321371208,
      "grad_norm": 0.4465678036212921,
      "learning_rate": 9.867058980851625e-06,
      "loss": 0.3749,
      "step": 6021
    },
    {
      "epoch": 0.2763526226423753,
      "grad_norm": 0.514955997467041,
      "learning_rate": 9.867002811727138e-06,
      "loss": 0.4157,
      "step": 6022
    },
    {
      "epoch": 0.27639851314762975,
      "grad_norm": 0.45760083198547363,
      "learning_rate": 9.866946630899046e-06,
      "loss": 0.4192,
      "step": 6023
    },
    {
      "epoch": 0.2764444036528842,
      "grad_norm": 0.46829721331596375,
      "learning_rate": 9.866890438367482e-06,
      "loss": 0.429,
      "step": 6024
    },
    {
      "epoch": 0.2764902941581387,
      "grad_norm": 0.4232097566127777,
      "learning_rate": 9.866834234132583e-06,
      "loss": 0.3617,
      "step": 6025
    },
    {
      "epoch": 0.27653618466339314,
      "grad_norm": 0.4737151861190796,
      "learning_rate": 9.866778018194484e-06,
      "loss": 0.4451,
      "step": 6026
    },
    {
      "epoch": 0.27658207516864763,
      "grad_norm": 0.4926268458366394,
      "learning_rate": 9.86672179055332e-06,
      "loss": 0.5006,
      "step": 6027
    },
    {
      "epoch": 0.2766279656739021,
      "grad_norm": 0.4679906368255615,
      "learning_rate": 9.866665551209227e-06,
      "loss": 0.4562,
      "step": 6028
    },
    {
      "epoch": 0.2766738561791565,
      "grad_norm": 0.4732656180858612,
      "learning_rate": 9.866609300162338e-06,
      "loss": 0.4293,
      "step": 6029
    },
    {
      "epoch": 0.276719746684411,
      "grad_norm": 0.458100289106369,
      "learning_rate": 9.866553037412789e-06,
      "loss": 0.4312,
      "step": 6030
    },
    {
      "epoch": 0.27676563718966546,
      "grad_norm": 0.4775184988975525,
      "learning_rate": 9.866496762960718e-06,
      "loss": 0.4759,
      "step": 6031
    },
    {
      "epoch": 0.2768115276949199,
      "grad_norm": 0.427941232919693,
      "learning_rate": 9.866440476806256e-06,
      "loss": 0.3221,
      "step": 6032
    },
    {
      "epoch": 0.2768574182001744,
      "grad_norm": 0.44261130690574646,
      "learning_rate": 9.866384178949543e-06,
      "loss": 0.2905,
      "step": 6033
    },
    {
      "epoch": 0.27690330870542884,
      "grad_norm": 0.5109956860542297,
      "learning_rate": 9.866327869390708e-06,
      "loss": 0.381,
      "step": 6034
    },
    {
      "epoch": 0.2769491992106833,
      "grad_norm": 0.47376441955566406,
      "learning_rate": 9.866271548129894e-06,
      "loss": 0.3962,
      "step": 6035
    },
    {
      "epoch": 0.2769950897159378,
      "grad_norm": 0.46055057644844055,
      "learning_rate": 9.866215215167233e-06,
      "loss": 0.4346,
      "step": 6036
    },
    {
      "epoch": 0.2770409802211922,
      "grad_norm": 0.46654534339904785,
      "learning_rate": 9.866158870502857e-06,
      "loss": 0.4225,
      "step": 6037
    },
    {
      "epoch": 0.2770868707264467,
      "grad_norm": 0.46061649918556213,
      "learning_rate": 9.866102514136906e-06,
      "loss": 0.343,
      "step": 6038
    },
    {
      "epoch": 0.27713276123170116,
      "grad_norm": 0.4611658751964569,
      "learning_rate": 9.866046146069515e-06,
      "loss": 0.3944,
      "step": 6039
    },
    {
      "epoch": 0.2771786517369556,
      "grad_norm": 0.4746161103248596,
      "learning_rate": 9.865989766300819e-06,
      "loss": 0.3886,
      "step": 6040
    },
    {
      "epoch": 0.2772245422422101,
      "grad_norm": 0.49253445863723755,
      "learning_rate": 9.865933374830951e-06,
      "loss": 0.4248,
      "step": 6041
    },
    {
      "epoch": 0.27727043274746455,
      "grad_norm": 0.42558035254478455,
      "learning_rate": 9.865876971660052e-06,
      "loss": 0.2707,
      "step": 6042
    },
    {
      "epoch": 0.277316323252719,
      "grad_norm": 0.47400254011154175,
      "learning_rate": 9.865820556788252e-06,
      "loss": 0.4119,
      "step": 6043
    },
    {
      "epoch": 0.2773622137579735,
      "grad_norm": 0.5436603426933289,
      "learning_rate": 9.86576413021569e-06,
      "loss": 0.5675,
      "step": 6044
    },
    {
      "epoch": 0.27740810426322793,
      "grad_norm": 0.4616427719593048,
      "learning_rate": 9.865707691942504e-06,
      "loss": 0.3613,
      "step": 6045
    },
    {
      "epoch": 0.2774539947684824,
      "grad_norm": 0.4473024606704712,
      "learning_rate": 9.865651241968823e-06,
      "loss": 0.3783,
      "step": 6046
    },
    {
      "epoch": 0.27749988527373687,
      "grad_norm": 0.5100050568580627,
      "learning_rate": 9.865594780294789e-06,
      "loss": 0.4904,
      "step": 6047
    },
    {
      "epoch": 0.2775457757789913,
      "grad_norm": 0.45534130930900574,
      "learning_rate": 9.865538306920533e-06,
      "loss": 0.3858,
      "step": 6048
    },
    {
      "epoch": 0.2775916662842458,
      "grad_norm": 0.49166369438171387,
      "learning_rate": 9.865481821846194e-06,
      "loss": 0.431,
      "step": 6049
    },
    {
      "epoch": 0.27763755678950025,
      "grad_norm": 0.43643680214881897,
      "learning_rate": 9.865425325071906e-06,
      "loss": 0.3599,
      "step": 6050
    },
    {
      "epoch": 0.2776834472947547,
      "grad_norm": 0.4726827144622803,
      "learning_rate": 9.865368816597806e-06,
      "loss": 0.4008,
      "step": 6051
    },
    {
      "epoch": 0.2777293378000092,
      "grad_norm": 0.4844609797000885,
      "learning_rate": 9.86531229642403e-06,
      "loss": 0.4984,
      "step": 6052
    },
    {
      "epoch": 0.27777522830526363,
      "grad_norm": 0.4752030074596405,
      "learning_rate": 9.865255764550714e-06,
      "loss": 0.407,
      "step": 6053
    },
    {
      "epoch": 0.27782111881051813,
      "grad_norm": 0.4702875316143036,
      "learning_rate": 9.865199220977993e-06,
      "loss": 0.3394,
      "step": 6054
    },
    {
      "epoch": 0.2778670093157726,
      "grad_norm": 0.4608556032180786,
      "learning_rate": 9.865142665706002e-06,
      "loss": 0.3854,
      "step": 6055
    },
    {
      "epoch": 0.277912899821027,
      "grad_norm": 0.4405447244644165,
      "learning_rate": 9.86508609873488e-06,
      "loss": 0.3974,
      "step": 6056
    },
    {
      "epoch": 0.2779587903262815,
      "grad_norm": 0.4800127446651459,
      "learning_rate": 9.865029520064762e-06,
      "loss": 0.4155,
      "step": 6057
    },
    {
      "epoch": 0.27800468083153596,
      "grad_norm": 0.4713591933250427,
      "learning_rate": 9.864972929695782e-06,
      "loss": 0.4152,
      "step": 6058
    },
    {
      "epoch": 0.2780505713367904,
      "grad_norm": 0.46247512102127075,
      "learning_rate": 9.864916327628078e-06,
      "loss": 0.4498,
      "step": 6059
    },
    {
      "epoch": 0.2780964618420449,
      "grad_norm": 0.47576457262039185,
      "learning_rate": 9.864859713861786e-06,
      "loss": 0.4022,
      "step": 6060
    },
    {
      "epoch": 0.27814235234729934,
      "grad_norm": 0.4680711627006531,
      "learning_rate": 9.86480308839704e-06,
      "loss": 0.4515,
      "step": 6061
    },
    {
      "epoch": 0.2781882428525538,
      "grad_norm": 0.47682899236679077,
      "learning_rate": 9.86474645123398e-06,
      "loss": 0.4855,
      "step": 6062
    },
    {
      "epoch": 0.2782341333578083,
      "grad_norm": 0.4476422071456909,
      "learning_rate": 9.864689802372738e-06,
      "loss": 0.3862,
      "step": 6063
    },
    {
      "epoch": 0.2782800238630627,
      "grad_norm": 0.4934159517288208,
      "learning_rate": 9.864633141813455e-06,
      "loss": 0.4421,
      "step": 6064
    },
    {
      "epoch": 0.2783259143683172,
      "grad_norm": 0.45127177238464355,
      "learning_rate": 9.864576469556263e-06,
      "loss": 0.414,
      "step": 6065
    },
    {
      "epoch": 0.27837180487357166,
      "grad_norm": 0.4764055013656616,
      "learning_rate": 9.8645197856013e-06,
      "loss": 0.4681,
      "step": 6066
    },
    {
      "epoch": 0.2784176953788261,
      "grad_norm": 0.4488907754421234,
      "learning_rate": 9.864463089948701e-06,
      "loss": 0.3914,
      "step": 6067
    },
    {
      "epoch": 0.2784635858840806,
      "grad_norm": 0.486619770526886,
      "learning_rate": 9.864406382598605e-06,
      "loss": 0.449,
      "step": 6068
    },
    {
      "epoch": 0.27850947638933504,
      "grad_norm": 0.46068549156188965,
      "learning_rate": 9.864349663551147e-06,
      "loss": 0.3418,
      "step": 6069
    },
    {
      "epoch": 0.2785553668945895,
      "grad_norm": 0.5222238302230835,
      "learning_rate": 9.864292932806464e-06,
      "loss": 0.5161,
      "step": 6070
    },
    {
      "epoch": 0.278601257399844,
      "grad_norm": 0.43491387367248535,
      "learning_rate": 9.864236190364691e-06,
      "loss": 0.4342,
      "step": 6071
    },
    {
      "epoch": 0.2786471479050984,
      "grad_norm": 0.44954827427864075,
      "learning_rate": 9.864179436225965e-06,
      "loss": 0.366,
      "step": 6072
    },
    {
      "epoch": 0.2786930384103529,
      "grad_norm": 0.5231358408927917,
      "learning_rate": 9.864122670390424e-06,
      "loss": 0.4041,
      "step": 6073
    },
    {
      "epoch": 0.27873892891560736,
      "grad_norm": 0.46889233589172363,
      "learning_rate": 9.8640658928582e-06,
      "loss": 0.4475,
      "step": 6074
    },
    {
      "epoch": 0.2787848194208618,
      "grad_norm": 0.469134122133255,
      "learning_rate": 9.864009103629435e-06,
      "loss": 0.3953,
      "step": 6075
    },
    {
      "epoch": 0.2788307099261163,
      "grad_norm": 0.4443010687828064,
      "learning_rate": 9.863952302704262e-06,
      "loss": 0.3408,
      "step": 6076
    },
    {
      "epoch": 0.27887660043137075,
      "grad_norm": 0.46828216314315796,
      "learning_rate": 9.86389549008282e-06,
      "loss": 0.4016,
      "step": 6077
    },
    {
      "epoch": 0.2789224909366252,
      "grad_norm": 0.440003365278244,
      "learning_rate": 9.863838665765245e-06,
      "loss": 0.366,
      "step": 6078
    },
    {
      "epoch": 0.2789683814418797,
      "grad_norm": 0.5009279847145081,
      "learning_rate": 9.863781829751671e-06,
      "loss": 0.3238,
      "step": 6079
    },
    {
      "epoch": 0.27901427194713413,
      "grad_norm": 0.4994819164276123,
      "learning_rate": 9.863724982042238e-06,
      "loss": 0.3638,
      "step": 6080
    },
    {
      "epoch": 0.2790601624523886,
      "grad_norm": 0.6921394467353821,
      "learning_rate": 9.863668122637082e-06,
      "loss": 0.4238,
      "step": 6081
    },
    {
      "epoch": 0.27910605295764307,
      "grad_norm": 0.4951264560222626,
      "learning_rate": 9.863611251536339e-06,
      "loss": 0.4093,
      "step": 6082
    },
    {
      "epoch": 0.2791519434628975,
      "grad_norm": 0.43739986419677734,
      "learning_rate": 9.863554368740145e-06,
      "loss": 0.3596,
      "step": 6083
    },
    {
      "epoch": 0.279197833968152,
      "grad_norm": 0.511479377746582,
      "learning_rate": 9.863497474248638e-06,
      "loss": 0.4941,
      "step": 6084
    },
    {
      "epoch": 0.27924372447340645,
      "grad_norm": 0.43493038415908813,
      "learning_rate": 9.863440568061954e-06,
      "loss": 0.3226,
      "step": 6085
    },
    {
      "epoch": 0.2792896149786609,
      "grad_norm": 0.5051162242889404,
      "learning_rate": 9.863383650180232e-06,
      "loss": 0.4715,
      "step": 6086
    },
    {
      "epoch": 0.2793355054839154,
      "grad_norm": 0.47087568044662476,
      "learning_rate": 9.863326720603605e-06,
      "loss": 0.3864,
      "step": 6087
    },
    {
      "epoch": 0.27938139598916983,
      "grad_norm": 0.4549236595630646,
      "learning_rate": 9.863269779332214e-06,
      "loss": 0.3953,
      "step": 6088
    },
    {
      "epoch": 0.2794272864944243,
      "grad_norm": 0.42424309253692627,
      "learning_rate": 9.863212826366192e-06,
      "loss": 0.3251,
      "step": 6089
    },
    {
      "epoch": 0.2794731769996788,
      "grad_norm": 0.508939266204834,
      "learning_rate": 9.86315586170568e-06,
      "loss": 0.4594,
      "step": 6090
    },
    {
      "epoch": 0.2795190675049332,
      "grad_norm": 0.415714293718338,
      "learning_rate": 9.863098885350812e-06,
      "loss": 0.3422,
      "step": 6091
    },
    {
      "epoch": 0.2795649580101877,
      "grad_norm": 0.4551631510257721,
      "learning_rate": 9.863041897301728e-06,
      "loss": 0.3628,
      "step": 6092
    },
    {
      "epoch": 0.27961084851544216,
      "grad_norm": 0.5171768665313721,
      "learning_rate": 9.86298489755856e-06,
      "loss": 0.4421,
      "step": 6093
    },
    {
      "epoch": 0.2796567390206966,
      "grad_norm": 0.4643041789531708,
      "learning_rate": 9.862927886121452e-06,
      "loss": 0.4118,
      "step": 6094
    },
    {
      "epoch": 0.2797026295259511,
      "grad_norm": 0.5115693211555481,
      "learning_rate": 9.862870862990534e-06,
      "loss": 0.5214,
      "step": 6095
    },
    {
      "epoch": 0.27974852003120554,
      "grad_norm": 0.42760568857192993,
      "learning_rate": 9.862813828165948e-06,
      "loss": 0.3509,
      "step": 6096
    },
    {
      "epoch": 0.27979441053646,
      "grad_norm": 0.41440504789352417,
      "learning_rate": 9.862756781647829e-06,
      "loss": 0.318,
      "step": 6097
    },
    {
      "epoch": 0.2798403010417145,
      "grad_norm": 0.43415525555610657,
      "learning_rate": 9.862699723436314e-06,
      "loss": 0.3542,
      "step": 6098
    },
    {
      "epoch": 0.2798861915469689,
      "grad_norm": 0.4744921624660492,
      "learning_rate": 9.86264265353154e-06,
      "loss": 0.3865,
      "step": 6099
    },
    {
      "epoch": 0.2799320820522234,
      "grad_norm": 0.471827894449234,
      "learning_rate": 9.862585571933648e-06,
      "loss": 0.3922,
      "step": 6100
    },
    {
      "epoch": 0.27997797255747786,
      "grad_norm": 0.5131245851516724,
      "learning_rate": 9.86252847864277e-06,
      "loss": 0.4893,
      "step": 6101
    },
    {
      "epoch": 0.2800238630627323,
      "grad_norm": 0.42059680819511414,
      "learning_rate": 9.862471373659045e-06,
      "loss": 0.3381,
      "step": 6102
    },
    {
      "epoch": 0.2800697535679868,
      "grad_norm": 0.5674087405204773,
      "learning_rate": 9.862414256982614e-06,
      "loss": 0.3941,
      "step": 6103
    },
    {
      "epoch": 0.28011564407324124,
      "grad_norm": 0.48303478956222534,
      "learning_rate": 9.86235712861361e-06,
      "loss": 0.4399,
      "step": 6104
    },
    {
      "epoch": 0.2801615345784957,
      "grad_norm": 0.477804571390152,
      "learning_rate": 9.86229998855217e-06,
      "loss": 0.436,
      "step": 6105
    },
    {
      "epoch": 0.2802074250837502,
      "grad_norm": 0.5001541376113892,
      "learning_rate": 9.862242836798436e-06,
      "loss": 0.4081,
      "step": 6106
    },
    {
      "epoch": 0.2802533155890046,
      "grad_norm": 0.4856053292751312,
      "learning_rate": 9.86218567335254e-06,
      "loss": 0.426,
      "step": 6107
    },
    {
      "epoch": 0.2802992060942591,
      "grad_norm": 0.5164276361465454,
      "learning_rate": 9.862128498214625e-06,
      "loss": 0.3316,
      "step": 6108
    },
    {
      "epoch": 0.28034509659951357,
      "grad_norm": 0.49896273016929626,
      "learning_rate": 9.862071311384821e-06,
      "loss": 0.4582,
      "step": 6109
    },
    {
      "epoch": 0.280390987104768,
      "grad_norm": 0.4719243347644806,
      "learning_rate": 9.862014112863274e-06,
      "loss": 0.4441,
      "step": 6110
    },
    {
      "epoch": 0.2804368776100225,
      "grad_norm": 0.4493636190891266,
      "learning_rate": 9.861956902650116e-06,
      "loss": 0.3212,
      "step": 6111
    },
    {
      "epoch": 0.28048276811527695,
      "grad_norm": 0.46457400918006897,
      "learning_rate": 9.861899680745487e-06,
      "loss": 0.4405,
      "step": 6112
    },
    {
      "epoch": 0.2805286586205314,
      "grad_norm": 0.4559255838394165,
      "learning_rate": 9.861842447149524e-06,
      "loss": 0.3922,
      "step": 6113
    },
    {
      "epoch": 0.2805745491257859,
      "grad_norm": 0.5552449822425842,
      "learning_rate": 9.861785201862363e-06,
      "loss": 0.3847,
      "step": 6114
    },
    {
      "epoch": 0.28062043963104033,
      "grad_norm": 0.5308847427368164,
      "learning_rate": 9.861727944884143e-06,
      "loss": 0.4579,
      "step": 6115
    },
    {
      "epoch": 0.28066633013629483,
      "grad_norm": 0.42697668075561523,
      "learning_rate": 9.861670676215003e-06,
      "loss": 0.3741,
      "step": 6116
    },
    {
      "epoch": 0.28071222064154927,
      "grad_norm": 0.47981858253479004,
      "learning_rate": 9.861613395855079e-06,
      "loss": 0.4689,
      "step": 6117
    },
    {
      "epoch": 0.2807581111468037,
      "grad_norm": 0.4542839825153351,
      "learning_rate": 9.861556103804508e-06,
      "loss": 0.3981,
      "step": 6118
    },
    {
      "epoch": 0.2808040016520582,
      "grad_norm": 0.5141986608505249,
      "learning_rate": 9.861498800063429e-06,
      "loss": 0.513,
      "step": 6119
    },
    {
      "epoch": 0.28084989215731265,
      "grad_norm": 0.45149800181388855,
      "learning_rate": 9.86144148463198e-06,
      "loss": 0.3897,
      "step": 6120
    },
    {
      "epoch": 0.2808957826625671,
      "grad_norm": 0.4892216920852661,
      "learning_rate": 9.8613841575103e-06,
      "loss": 0.454,
      "step": 6121
    },
    {
      "epoch": 0.2809416731678216,
      "grad_norm": 0.4294017553329468,
      "learning_rate": 9.861326818698524e-06,
      "loss": 0.3142,
      "step": 6122
    },
    {
      "epoch": 0.28098756367307604,
      "grad_norm": 0.4647076725959778,
      "learning_rate": 9.861269468196791e-06,
      "loss": 0.4483,
      "step": 6123
    },
    {
      "epoch": 0.2810334541783305,
      "grad_norm": 0.5382829904556274,
      "learning_rate": 9.86121210600524e-06,
      "loss": 0.474,
      "step": 6124
    },
    {
      "epoch": 0.281079344683585,
      "grad_norm": 0.46035826206207275,
      "learning_rate": 9.861154732124008e-06,
      "loss": 0.4208,
      "step": 6125
    },
    {
      "epoch": 0.2811252351888394,
      "grad_norm": 0.47033190727233887,
      "learning_rate": 9.861097346553232e-06,
      "loss": 0.3954,
      "step": 6126
    },
    {
      "epoch": 0.2811711256940939,
      "grad_norm": 0.44361451268196106,
      "learning_rate": 9.861039949293052e-06,
      "loss": 0.3444,
      "step": 6127
    },
    {
      "epoch": 0.28121701619934836,
      "grad_norm": 0.4546622633934021,
      "learning_rate": 9.860982540343606e-06,
      "loss": 0.3728,
      "step": 6128
    },
    {
      "epoch": 0.2812629067046028,
      "grad_norm": 0.4699499309062958,
      "learning_rate": 9.86092511970503e-06,
      "loss": 0.4348,
      "step": 6129
    },
    {
      "epoch": 0.2813087972098573,
      "grad_norm": 0.4492476284503937,
      "learning_rate": 9.860867687377464e-06,
      "loss": 0.398,
      "step": 6130
    },
    {
      "epoch": 0.28135468771511174,
      "grad_norm": 0.48728132247924805,
      "learning_rate": 9.860810243361046e-06,
      "loss": 0.4207,
      "step": 6131
    },
    {
      "epoch": 0.2814005782203662,
      "grad_norm": 0.47781768441200256,
      "learning_rate": 9.860752787655911e-06,
      "loss": 0.4069,
      "step": 6132
    },
    {
      "epoch": 0.2814464687256207,
      "grad_norm": 0.4726647436618805,
      "learning_rate": 9.860695320262202e-06,
      "loss": 0.4453,
      "step": 6133
    },
    {
      "epoch": 0.2814923592308751,
      "grad_norm": 0.5571734309196472,
      "learning_rate": 9.860637841180054e-06,
      "loss": 0.5334,
      "step": 6134
    },
    {
      "epoch": 0.2815382497361296,
      "grad_norm": 0.7431037425994873,
      "learning_rate": 9.860580350409606e-06,
      "loss": 0.3719,
      "step": 6135
    },
    {
      "epoch": 0.28158414024138406,
      "grad_norm": 0.46557629108428955,
      "learning_rate": 9.860522847950995e-06,
      "loss": 0.4142,
      "step": 6136
    },
    {
      "epoch": 0.2816300307466385,
      "grad_norm": 0.43953147530555725,
      "learning_rate": 9.860465333804364e-06,
      "loss": 0.3497,
      "step": 6137
    },
    {
      "epoch": 0.281675921251893,
      "grad_norm": 0.4810963273048401,
      "learning_rate": 9.860407807969845e-06,
      "loss": 0.4325,
      "step": 6138
    },
    {
      "epoch": 0.28172181175714744,
      "grad_norm": 0.44177404046058655,
      "learning_rate": 9.860350270447579e-06,
      "loss": 0.3463,
      "step": 6139
    },
    {
      "epoch": 0.2817677022624019,
      "grad_norm": 0.4617029130458832,
      "learning_rate": 9.860292721237705e-06,
      "loss": 0.402,
      "step": 6140
    },
    {
      "epoch": 0.2818135927676564,
      "grad_norm": 0.40834078192710876,
      "learning_rate": 9.86023516034036e-06,
      "loss": 0.3023,
      "step": 6141
    },
    {
      "epoch": 0.2818594832729108,
      "grad_norm": 0.5194640755653381,
      "learning_rate": 9.860177587755686e-06,
      "loss": 0.4701,
      "step": 6142
    },
    {
      "epoch": 0.2819053737781653,
      "grad_norm": 0.4561716318130493,
      "learning_rate": 9.860120003483818e-06,
      "loss": 0.3613,
      "step": 6143
    },
    {
      "epoch": 0.28195126428341977,
      "grad_norm": 0.46242067217826843,
      "learning_rate": 9.860062407524895e-06,
      "loss": 0.3776,
      "step": 6144
    },
    {
      "epoch": 0.2819971547886742,
      "grad_norm": 0.481309711933136,
      "learning_rate": 9.860004799879056e-06,
      "loss": 0.4168,
      "step": 6145
    },
    {
      "epoch": 0.2820430452939287,
      "grad_norm": 0.488971471786499,
      "learning_rate": 9.859947180546439e-06,
      "loss": 0.3861,
      "step": 6146
    },
    {
      "epoch": 0.28208893579918315,
      "grad_norm": 0.4441189467906952,
      "learning_rate": 9.859889549527183e-06,
      "loss": 0.3681,
      "step": 6147
    },
    {
      "epoch": 0.2821348263044376,
      "grad_norm": 0.49475425481796265,
      "learning_rate": 9.859831906821427e-06,
      "loss": 0.4366,
      "step": 6148
    },
    {
      "epoch": 0.2821807168096921,
      "grad_norm": 0.4932027757167816,
      "learning_rate": 9.859774252429309e-06,
      "loss": 0.5011,
      "step": 6149
    },
    {
      "epoch": 0.28222660731494653,
      "grad_norm": 0.43192771077156067,
      "learning_rate": 9.859716586350965e-06,
      "loss": 0.3367,
      "step": 6150
    },
    {
      "epoch": 0.282272497820201,
      "grad_norm": 0.5084330439567566,
      "learning_rate": 9.85965890858654e-06,
      "loss": 0.4483,
      "step": 6151
    },
    {
      "epoch": 0.28231838832545547,
      "grad_norm": 0.472550630569458,
      "learning_rate": 9.859601219136167e-06,
      "loss": 0.4775,
      "step": 6152
    },
    {
      "epoch": 0.2823642788307099,
      "grad_norm": 0.465170681476593,
      "learning_rate": 9.859543517999987e-06,
      "loss": 0.4077,
      "step": 6153
    },
    {
      "epoch": 0.2824101693359644,
      "grad_norm": 0.45738843083381653,
      "learning_rate": 9.859485805178139e-06,
      "loss": 0.4112,
      "step": 6154
    },
    {
      "epoch": 0.28245605984121885,
      "grad_norm": 0.4472235143184662,
      "learning_rate": 9.859428080670761e-06,
      "loss": 0.3617,
      "step": 6155
    },
    {
      "epoch": 0.2825019503464733,
      "grad_norm": 0.42663973569869995,
      "learning_rate": 9.859370344477993e-06,
      "loss": 0.3578,
      "step": 6156
    },
    {
      "epoch": 0.2825478408517278,
      "grad_norm": 0.44951340556144714,
      "learning_rate": 9.859312596599972e-06,
      "loss": 0.3795,
      "step": 6157
    },
    {
      "epoch": 0.28259373135698224,
      "grad_norm": 0.4800887405872345,
      "learning_rate": 9.859254837036838e-06,
      "loss": 0.4393,
      "step": 6158
    },
    {
      "epoch": 0.2826396218622367,
      "grad_norm": 0.4766150116920471,
      "learning_rate": 9.859197065788731e-06,
      "loss": 0.434,
      "step": 6159
    },
    {
      "epoch": 0.2826855123674912,
      "grad_norm": 0.44755032658576965,
      "learning_rate": 9.859139282855787e-06,
      "loss": 0.3704,
      "step": 6160
    },
    {
      "epoch": 0.2827314028727456,
      "grad_norm": 0.48971590399742126,
      "learning_rate": 9.859081488238147e-06,
      "loss": 0.4697,
      "step": 6161
    },
    {
      "epoch": 0.2827772933780001,
      "grad_norm": 0.4429587423801422,
      "learning_rate": 9.859023681935951e-06,
      "loss": 0.39,
      "step": 6162
    },
    {
      "epoch": 0.28282318388325456,
      "grad_norm": 0.46537846326828003,
      "learning_rate": 9.858965863949335e-06,
      "loss": 0.4017,
      "step": 6163
    },
    {
      "epoch": 0.282869074388509,
      "grad_norm": 0.4563058912754059,
      "learning_rate": 9.85890803427844e-06,
      "loss": 0.3872,
      "step": 6164
    },
    {
      "epoch": 0.2829149648937635,
      "grad_norm": 0.47252127528190613,
      "learning_rate": 9.858850192923405e-06,
      "loss": 0.4008,
      "step": 6165
    },
    {
      "epoch": 0.28296085539901794,
      "grad_norm": 0.4733385741710663,
      "learning_rate": 9.858792339884368e-06,
      "loss": 0.4248,
      "step": 6166
    },
    {
      "epoch": 0.2830067459042724,
      "grad_norm": 0.48564577102661133,
      "learning_rate": 9.85873447516147e-06,
      "loss": 0.4165,
      "step": 6167
    },
    {
      "epoch": 0.2830526364095269,
      "grad_norm": 0.4254152774810791,
      "learning_rate": 9.858676598754848e-06,
      "loss": 0.3245,
      "step": 6168
    },
    {
      "epoch": 0.2830985269147813,
      "grad_norm": 0.4674239456653595,
      "learning_rate": 9.858618710664644e-06,
      "loss": 0.4076,
      "step": 6169
    },
    {
      "epoch": 0.2831444174200358,
      "grad_norm": 0.4380985498428345,
      "learning_rate": 9.858560810890993e-06,
      "loss": 0.3581,
      "step": 6170
    },
    {
      "epoch": 0.28319030792529026,
      "grad_norm": 0.45615044236183167,
      "learning_rate": 9.858502899434038e-06,
      "loss": 0.3923,
      "step": 6171
    },
    {
      "epoch": 0.2832361984305447,
      "grad_norm": 0.5834977030754089,
      "learning_rate": 9.858444976293918e-06,
      "loss": 0.2977,
      "step": 6172
    },
    {
      "epoch": 0.2832820889357992,
      "grad_norm": 0.4499538540840149,
      "learning_rate": 9.85838704147077e-06,
      "loss": 0.32,
      "step": 6173
    },
    {
      "epoch": 0.28332797944105365,
      "grad_norm": 0.4485560357570648,
      "learning_rate": 9.858329094964734e-06,
      "loss": 0.3897,
      "step": 6174
    },
    {
      "epoch": 0.2833738699463081,
      "grad_norm": 0.491111159324646,
      "learning_rate": 9.85827113677595e-06,
      "loss": 0.4323,
      "step": 6175
    },
    {
      "epoch": 0.2834197604515626,
      "grad_norm": 0.44551801681518555,
      "learning_rate": 9.858213166904558e-06,
      "loss": 0.3833,
      "step": 6176
    },
    {
      "epoch": 0.28346565095681703,
      "grad_norm": 0.44683682918548584,
      "learning_rate": 9.858155185350696e-06,
      "loss": 0.3739,
      "step": 6177
    },
    {
      "epoch": 0.28351154146207147,
      "grad_norm": 0.46934232115745544,
      "learning_rate": 9.858097192114504e-06,
      "loss": 0.3867,
      "step": 6178
    },
    {
      "epoch": 0.28355743196732597,
      "grad_norm": 0.4553755223751068,
      "learning_rate": 9.858039187196122e-06,
      "loss": 0.3992,
      "step": 6179
    },
    {
      "epoch": 0.2836033224725804,
      "grad_norm": 0.45895448327064514,
      "learning_rate": 9.857981170595689e-06,
      "loss": 0.431,
      "step": 6180
    },
    {
      "epoch": 0.2836492129778349,
      "grad_norm": 0.4690587520599365,
      "learning_rate": 9.857923142313343e-06,
      "loss": 0.4161,
      "step": 6181
    },
    {
      "epoch": 0.28369510348308935,
      "grad_norm": 0.493902325630188,
      "learning_rate": 9.857865102349227e-06,
      "loss": 0.464,
      "step": 6182
    },
    {
      "epoch": 0.2837409939883438,
      "grad_norm": 0.47692567110061646,
      "learning_rate": 9.857807050703478e-06,
      "loss": 0.378,
      "step": 6183
    },
    {
      "epoch": 0.2837868844935983,
      "grad_norm": 0.47036850452423096,
      "learning_rate": 9.857748987376237e-06,
      "loss": 0.3987,
      "step": 6184
    },
    {
      "epoch": 0.28383277499885273,
      "grad_norm": 0.5170320868492126,
      "learning_rate": 9.85769091236764e-06,
      "loss": 0.4314,
      "step": 6185
    },
    {
      "epoch": 0.2838786655041072,
      "grad_norm": 0.5118705630302429,
      "learning_rate": 9.857632825677832e-06,
      "loss": 0.4361,
      "step": 6186
    },
    {
      "epoch": 0.2839245560093617,
      "grad_norm": 0.46199196577072144,
      "learning_rate": 9.85757472730695e-06,
      "loss": 0.3972,
      "step": 6187
    },
    {
      "epoch": 0.2839704465146161,
      "grad_norm": 0.43804553151130676,
      "learning_rate": 9.857516617255134e-06,
      "loss": 0.3553,
      "step": 6188
    },
    {
      "epoch": 0.2840163370198706,
      "grad_norm": 0.48519113659858704,
      "learning_rate": 9.857458495522523e-06,
      "loss": 0.454,
      "step": 6189
    },
    {
      "epoch": 0.28406222752512506,
      "grad_norm": 0.4630095958709717,
      "learning_rate": 9.857400362109258e-06,
      "loss": 0.4249,
      "step": 6190
    },
    {
      "epoch": 0.2841081180303795,
      "grad_norm": 0.45949694514274597,
      "learning_rate": 9.857342217015477e-06,
      "loss": 0.3727,
      "step": 6191
    },
    {
      "epoch": 0.284154008535634,
      "grad_norm": 0.438872367143631,
      "learning_rate": 9.857284060241321e-06,
      "loss": 0.3853,
      "step": 6192
    },
    {
      "epoch": 0.28419989904088844,
      "grad_norm": 0.4650721251964569,
      "learning_rate": 9.857225891786932e-06,
      "loss": 0.3775,
      "step": 6193
    },
    {
      "epoch": 0.2842457895461429,
      "grad_norm": 0.4481979310512543,
      "learning_rate": 9.857167711652445e-06,
      "loss": 0.3502,
      "step": 6194
    },
    {
      "epoch": 0.2842916800513974,
      "grad_norm": 0.4544903635978699,
      "learning_rate": 9.857109519838005e-06,
      "loss": 0.4039,
      "step": 6195
    },
    {
      "epoch": 0.2843375705566518,
      "grad_norm": 0.531489372253418,
      "learning_rate": 9.857051316343748e-06,
      "loss": 0.4356,
      "step": 6196
    },
    {
      "epoch": 0.2843834610619063,
      "grad_norm": 0.482708215713501,
      "learning_rate": 9.856993101169817e-06,
      "loss": 0.4393,
      "step": 6197
    },
    {
      "epoch": 0.28442935156716076,
      "grad_norm": 0.44754308462142944,
      "learning_rate": 9.856934874316348e-06,
      "loss": 0.3202,
      "step": 6198
    },
    {
      "epoch": 0.2844752420724152,
      "grad_norm": 0.5014321804046631,
      "learning_rate": 9.856876635783484e-06,
      "loss": 0.435,
      "step": 6199
    },
    {
      "epoch": 0.2845211325776697,
      "grad_norm": 0.43194690346717834,
      "learning_rate": 9.856818385571366e-06,
      "loss": 0.3564,
      "step": 6200
    },
    {
      "epoch": 0.28456702308292414,
      "grad_norm": 0.5064954161643982,
      "learning_rate": 9.856760123680133e-06,
      "loss": 0.4168,
      "step": 6201
    },
    {
      "epoch": 0.2846129135881786,
      "grad_norm": 0.4354548752307892,
      "learning_rate": 9.856701850109923e-06,
      "loss": 0.3646,
      "step": 6202
    },
    {
      "epoch": 0.2846588040934331,
      "grad_norm": 0.5071757435798645,
      "learning_rate": 9.85664356486088e-06,
      "loss": 0.4773,
      "step": 6203
    },
    {
      "epoch": 0.2847046945986875,
      "grad_norm": 0.5185554623603821,
      "learning_rate": 9.85658526793314e-06,
      "loss": 0.4839,
      "step": 6204
    },
    {
      "epoch": 0.28475058510394197,
      "grad_norm": 0.4655497372150421,
      "learning_rate": 9.856526959326845e-06,
      "loss": 0.4514,
      "step": 6205
    },
    {
      "epoch": 0.28479647560919646,
      "grad_norm": 0.4587070941925049,
      "learning_rate": 9.856468639042137e-06,
      "loss": 0.3853,
      "step": 6206
    },
    {
      "epoch": 0.2848423661144509,
      "grad_norm": 0.46329328417778015,
      "learning_rate": 9.856410307079154e-06,
      "loss": 0.4049,
      "step": 6207
    },
    {
      "epoch": 0.2848882566197054,
      "grad_norm": 0.4677334129810333,
      "learning_rate": 9.856351963438038e-06,
      "loss": 0.416,
      "step": 6208
    },
    {
      "epoch": 0.28493414712495985,
      "grad_norm": 0.4836731553077698,
      "learning_rate": 9.856293608118926e-06,
      "loss": 0.4301,
      "step": 6209
    },
    {
      "epoch": 0.2849800376302143,
      "grad_norm": 0.48330870270729065,
      "learning_rate": 9.856235241121962e-06,
      "loss": 0.4022,
      "step": 6210
    },
    {
      "epoch": 0.2850259281354688,
      "grad_norm": 0.46648284792900085,
      "learning_rate": 9.856176862447285e-06,
      "loss": 0.4227,
      "step": 6211
    },
    {
      "epoch": 0.28507181864072323,
      "grad_norm": 0.47135964035987854,
      "learning_rate": 9.856118472095036e-06,
      "loss": 0.4559,
      "step": 6212
    },
    {
      "epoch": 0.28511770914597767,
      "grad_norm": 0.43633779883384705,
      "learning_rate": 9.856060070065352e-06,
      "loss": 0.4009,
      "step": 6213
    },
    {
      "epoch": 0.28516359965123217,
      "grad_norm": 0.47200459241867065,
      "learning_rate": 9.856001656358378e-06,
      "loss": 0.4159,
      "step": 6214
    },
    {
      "epoch": 0.2852094901564866,
      "grad_norm": 0.466999888420105,
      "learning_rate": 9.855943230974252e-06,
      "loss": 0.3784,
      "step": 6215
    },
    {
      "epoch": 0.2852553806617411,
      "grad_norm": 0.4383961260318756,
      "learning_rate": 9.855884793913117e-06,
      "loss": 0.3757,
      "step": 6216
    },
    {
      "epoch": 0.28530127116699555,
      "grad_norm": 0.4501575827598572,
      "learning_rate": 9.85582634517511e-06,
      "loss": 0.3763,
      "step": 6217
    },
    {
      "epoch": 0.28534716167225,
      "grad_norm": 0.48830845952033997,
      "learning_rate": 9.855767884760374e-06,
      "loss": 0.4774,
      "step": 6218
    },
    {
      "epoch": 0.2853930521775045,
      "grad_norm": 0.4896560609340668,
      "learning_rate": 9.855709412669047e-06,
      "loss": 0.4351,
      "step": 6219
    },
    {
      "epoch": 0.28543894268275893,
      "grad_norm": 0.4627843499183655,
      "learning_rate": 9.855650928901273e-06,
      "loss": 0.394,
      "step": 6220
    },
    {
      "epoch": 0.2854848331880134,
      "grad_norm": 0.4609028697013855,
      "learning_rate": 9.85559243345719e-06,
      "loss": 0.3892,
      "step": 6221
    },
    {
      "epoch": 0.2855307236932679,
      "grad_norm": 0.4463323950767517,
      "learning_rate": 9.85553392633694e-06,
      "loss": 0.3555,
      "step": 6222
    },
    {
      "epoch": 0.2855766141985223,
      "grad_norm": 0.46091026067733765,
      "learning_rate": 9.855475407540664e-06,
      "loss": 0.3662,
      "step": 6223
    },
    {
      "epoch": 0.2856225047037768,
      "grad_norm": 0.5015755295753479,
      "learning_rate": 9.855416877068503e-06,
      "loss": 0.5323,
      "step": 6224
    },
    {
      "epoch": 0.28566839520903126,
      "grad_norm": 0.47849828004837036,
      "learning_rate": 9.855358334920596e-06,
      "loss": 0.419,
      "step": 6225
    },
    {
      "epoch": 0.2857142857142857,
      "grad_norm": 0.472225159406662,
      "learning_rate": 9.855299781097083e-06,
      "loss": 0.428,
      "step": 6226
    },
    {
      "epoch": 0.2857601762195402,
      "grad_norm": 0.482185423374176,
      "learning_rate": 9.855241215598107e-06,
      "loss": 0.482,
      "step": 6227
    },
    {
      "epoch": 0.28580606672479464,
      "grad_norm": 0.530707836151123,
      "learning_rate": 9.855182638423809e-06,
      "loss": 0.5243,
      "step": 6228
    },
    {
      "epoch": 0.2858519572300491,
      "grad_norm": 0.4378359317779541,
      "learning_rate": 9.855124049574328e-06,
      "loss": 0.3606,
      "step": 6229
    },
    {
      "epoch": 0.2858978477353036,
      "grad_norm": 0.43571460247039795,
      "learning_rate": 9.855065449049808e-06,
      "loss": 0.3494,
      "step": 6230
    },
    {
      "epoch": 0.285943738240558,
      "grad_norm": 0.4354548454284668,
      "learning_rate": 9.855006836850386e-06,
      "loss": 0.3529,
      "step": 6231
    },
    {
      "epoch": 0.2859896287458125,
      "grad_norm": 0.4659758508205414,
      "learning_rate": 9.854948212976205e-06,
      "loss": 0.4743,
      "step": 6232
    },
    {
      "epoch": 0.28603551925106696,
      "grad_norm": 0.4602943956851959,
      "learning_rate": 9.854889577427406e-06,
      "loss": 0.387,
      "step": 6233
    },
    {
      "epoch": 0.2860814097563214,
      "grad_norm": 0.4563579261302948,
      "learning_rate": 9.854830930204132e-06,
      "loss": 0.4165,
      "step": 6234
    },
    {
      "epoch": 0.2861273002615759,
      "grad_norm": 0.4972166419029236,
      "learning_rate": 9.85477227130652e-06,
      "loss": 0.4951,
      "step": 6235
    },
    {
      "epoch": 0.28617319076683034,
      "grad_norm": 0.4556387960910797,
      "learning_rate": 9.85471360073471e-06,
      "loss": 0.3638,
      "step": 6236
    },
    {
      "epoch": 0.2862190812720848,
      "grad_norm": 0.477286696434021,
      "learning_rate": 9.854654918488852e-06,
      "loss": 0.4429,
      "step": 6237
    },
    {
      "epoch": 0.2862649717773393,
      "grad_norm": 0.45831719040870667,
      "learning_rate": 9.854596224569076e-06,
      "loss": 0.3959,
      "step": 6238
    },
    {
      "epoch": 0.2863108622825937,
      "grad_norm": 0.4485675096511841,
      "learning_rate": 9.854537518975531e-06,
      "loss": 0.3725,
      "step": 6239
    },
    {
      "epoch": 0.28635675278784817,
      "grad_norm": 0.5275249481201172,
      "learning_rate": 9.854478801708353e-06,
      "loss": 0.4309,
      "step": 6240
    },
    {
      "epoch": 0.28640264329310267,
      "grad_norm": 0.46288010478019714,
      "learning_rate": 9.854420072767689e-06,
      "loss": 0.3662,
      "step": 6241
    },
    {
      "epoch": 0.2864485337983571,
      "grad_norm": 0.4517335295677185,
      "learning_rate": 9.854361332153675e-06,
      "loss": 0.3587,
      "step": 6242
    },
    {
      "epoch": 0.2864944243036116,
      "grad_norm": 0.45703309774398804,
      "learning_rate": 9.854302579866454e-06,
      "loss": 0.3633,
      "step": 6243
    },
    {
      "epoch": 0.28654031480886605,
      "grad_norm": 0.5704601407051086,
      "learning_rate": 9.854243815906168e-06,
      "loss": 0.5165,
      "step": 6244
    },
    {
      "epoch": 0.2865862053141205,
      "grad_norm": 0.4818931519985199,
      "learning_rate": 9.854185040272956e-06,
      "loss": 0.3881,
      "step": 6245
    },
    {
      "epoch": 0.286632095819375,
      "grad_norm": 0.45618441700935364,
      "learning_rate": 9.854126252966963e-06,
      "loss": 0.4299,
      "step": 6246
    },
    {
      "epoch": 0.28667798632462943,
      "grad_norm": 0.46862849593162537,
      "learning_rate": 9.854067453988327e-06,
      "loss": 0.3777,
      "step": 6247
    },
    {
      "epoch": 0.2867238768298839,
      "grad_norm": 0.49320071935653687,
      "learning_rate": 9.854008643337191e-06,
      "loss": 0.4403,
      "step": 6248
    },
    {
      "epoch": 0.28676976733513837,
      "grad_norm": 0.5146374106407166,
      "learning_rate": 9.853949821013697e-06,
      "loss": 0.4296,
      "step": 6249
    },
    {
      "epoch": 0.2868156578403928,
      "grad_norm": 0.493019163608551,
      "learning_rate": 9.853890987017984e-06,
      "loss": 0.4529,
      "step": 6250
    },
    {
      "epoch": 0.2868615483456473,
      "grad_norm": 0.4376666247844696,
      "learning_rate": 9.853832141350196e-06,
      "loss": 0.3588,
      "step": 6251
    },
    {
      "epoch": 0.28690743885090175,
      "grad_norm": 0.5039815306663513,
      "learning_rate": 9.853773284010476e-06,
      "loss": 0.5084,
      "step": 6252
    },
    {
      "epoch": 0.2869533293561562,
      "grad_norm": 0.43810781836509705,
      "learning_rate": 9.85371441499896e-06,
      "loss": 0.381,
      "step": 6253
    },
    {
      "epoch": 0.2869992198614107,
      "grad_norm": 0.48453664779663086,
      "learning_rate": 9.853655534315793e-06,
      "loss": 0.4326,
      "step": 6254
    },
    {
      "epoch": 0.28704511036666513,
      "grad_norm": 0.47260886430740356,
      "learning_rate": 9.853596641961118e-06,
      "loss": 0.4076,
      "step": 6255
    },
    {
      "epoch": 0.2870910008719196,
      "grad_norm": 0.4749000370502472,
      "learning_rate": 9.853537737935072e-06,
      "loss": 0.3791,
      "step": 6256
    },
    {
      "epoch": 0.2871368913771741,
      "grad_norm": 0.48497769236564636,
      "learning_rate": 9.853478822237804e-06,
      "loss": 0.3888,
      "step": 6257
    },
    {
      "epoch": 0.2871827818824285,
      "grad_norm": 0.4510013461112976,
      "learning_rate": 9.853419894869446e-06,
      "loss": 0.3574,
      "step": 6258
    },
    {
      "epoch": 0.287228672387683,
      "grad_norm": 0.4828437268733978,
      "learning_rate": 9.853360955830149e-06,
      "loss": 0.4116,
      "step": 6259
    },
    {
      "epoch": 0.28727456289293746,
      "grad_norm": 0.483196884393692,
      "learning_rate": 9.85330200512005e-06,
      "loss": 0.4448,
      "step": 6260
    },
    {
      "epoch": 0.2873204533981919,
      "grad_norm": 0.4670977294445038,
      "learning_rate": 9.853243042739289e-06,
      "loss": 0.4238,
      "step": 6261
    },
    {
      "epoch": 0.2873663439034464,
      "grad_norm": 0.5728627443313599,
      "learning_rate": 9.853184068688013e-06,
      "loss": 0.5693,
      "step": 6262
    },
    {
      "epoch": 0.28741223440870084,
      "grad_norm": 0.4972585141658783,
      "learning_rate": 9.85312508296636e-06,
      "loss": 0.5224,
      "step": 6263
    },
    {
      "epoch": 0.2874581249139553,
      "grad_norm": 0.46810635924339294,
      "learning_rate": 9.853066085574472e-06,
      "loss": 0.4353,
      "step": 6264
    },
    {
      "epoch": 0.2875040154192098,
      "grad_norm": 0.4771280884742737,
      "learning_rate": 9.853007076512492e-06,
      "loss": 0.4605,
      "step": 6265
    },
    {
      "epoch": 0.2875499059244642,
      "grad_norm": 0.47825321555137634,
      "learning_rate": 9.852948055780562e-06,
      "loss": 0.413,
      "step": 6266
    },
    {
      "epoch": 0.28759579642971866,
      "grad_norm": 0.45670366287231445,
      "learning_rate": 9.852889023378824e-06,
      "loss": 0.359,
      "step": 6267
    },
    {
      "epoch": 0.28764168693497316,
      "grad_norm": 0.4528508484363556,
      "learning_rate": 9.85282997930742e-06,
      "loss": 0.3876,
      "step": 6268
    },
    {
      "epoch": 0.2876875774402276,
      "grad_norm": 0.4421367347240448,
      "learning_rate": 9.85277092356649e-06,
      "loss": 0.3386,
      "step": 6269
    },
    {
      "epoch": 0.2877334679454821,
      "grad_norm": 0.5153143405914307,
      "learning_rate": 9.852711856156178e-06,
      "loss": 0.5255,
      "step": 6270
    },
    {
      "epoch": 0.28777935845073654,
      "grad_norm": 0.473999559879303,
      "learning_rate": 9.852652777076627e-06,
      "loss": 0.4572,
      "step": 6271
    },
    {
      "epoch": 0.287825248955991,
      "grad_norm": 0.5037572383880615,
      "learning_rate": 9.852593686327975e-06,
      "loss": 0.4231,
      "step": 6272
    },
    {
      "epoch": 0.2878711394612455,
      "grad_norm": 0.4410327076911926,
      "learning_rate": 9.852534583910369e-06,
      "loss": 0.28,
      "step": 6273
    },
    {
      "epoch": 0.2879170299664999,
      "grad_norm": 0.44536107778549194,
      "learning_rate": 9.852475469823947e-06,
      "loss": 0.3573,
      "step": 6274
    },
    {
      "epoch": 0.28796292047175437,
      "grad_norm": 0.5043711066246033,
      "learning_rate": 9.852416344068855e-06,
      "loss": 0.4452,
      "step": 6275
    },
    {
      "epoch": 0.28800881097700887,
      "grad_norm": 0.4719536006450653,
      "learning_rate": 9.852357206645232e-06,
      "loss": 0.4123,
      "step": 6276
    },
    {
      "epoch": 0.2880547014822633,
      "grad_norm": 0.4324806034564972,
      "learning_rate": 9.852298057553222e-06,
      "loss": 0.3518,
      "step": 6277
    },
    {
      "epoch": 0.2881005919875178,
      "grad_norm": 0.4597470462322235,
      "learning_rate": 9.852238896792964e-06,
      "loss": 0.3938,
      "step": 6278
    },
    {
      "epoch": 0.28814648249277225,
      "grad_norm": 0.490261971950531,
      "learning_rate": 9.852179724364607e-06,
      "loss": 0.454,
      "step": 6279
    },
    {
      "epoch": 0.2881923729980267,
      "grad_norm": 0.43131354451179504,
      "learning_rate": 9.852120540268286e-06,
      "loss": 0.3061,
      "step": 6280
    },
    {
      "epoch": 0.2882382635032812,
      "grad_norm": 0.4604109525680542,
      "learning_rate": 9.852061344504148e-06,
      "loss": 0.4241,
      "step": 6281
    },
    {
      "epoch": 0.28828415400853563,
      "grad_norm": 0.4556732773780823,
      "learning_rate": 9.852002137072334e-06,
      "loss": 0.3855,
      "step": 6282
    },
    {
      "epoch": 0.2883300445137901,
      "grad_norm": 0.4630882143974304,
      "learning_rate": 9.851942917972986e-06,
      "loss": 0.3837,
      "step": 6283
    },
    {
      "epoch": 0.28837593501904457,
      "grad_norm": 0.45042598247528076,
      "learning_rate": 9.851883687206244e-06,
      "loss": 0.3589,
      "step": 6284
    },
    {
      "epoch": 0.288421825524299,
      "grad_norm": 0.47822022438049316,
      "learning_rate": 9.851824444772256e-06,
      "loss": 0.3997,
      "step": 6285
    },
    {
      "epoch": 0.2884677160295535,
      "grad_norm": 0.47928518056869507,
      "learning_rate": 9.85176519067116e-06,
      "loss": 0.4617,
      "step": 6286
    },
    {
      "epoch": 0.28851360653480795,
      "grad_norm": 0.4958309233188629,
      "learning_rate": 9.851705924903101e-06,
      "loss": 0.4961,
      "step": 6287
    },
    {
      "epoch": 0.2885594970400624,
      "grad_norm": 0.4729976952075958,
      "learning_rate": 9.85164664746822e-06,
      "loss": 0.4363,
      "step": 6288
    },
    {
      "epoch": 0.2886053875453169,
      "grad_norm": 0.4796024560928345,
      "learning_rate": 9.851587358366662e-06,
      "loss": 0.4429,
      "step": 6289
    },
    {
      "epoch": 0.28865127805057134,
      "grad_norm": 0.4620402753353119,
      "learning_rate": 9.851528057598564e-06,
      "loss": 0.3999,
      "step": 6290
    },
    {
      "epoch": 0.2886971685558258,
      "grad_norm": 0.4856235384941101,
      "learning_rate": 9.851468745164074e-06,
      "loss": 0.397,
      "step": 6291
    },
    {
      "epoch": 0.2887430590610803,
      "grad_norm": 0.4951448440551758,
      "learning_rate": 9.851409421063334e-06,
      "loss": 0.455,
      "step": 6292
    },
    {
      "epoch": 0.2887889495663347,
      "grad_norm": 0.5033533573150635,
      "learning_rate": 9.851350085296484e-06,
      "loss": 0.4282,
      "step": 6293
    },
    {
      "epoch": 0.28883484007158916,
      "grad_norm": 0.45773935317993164,
      "learning_rate": 9.851290737863668e-06,
      "loss": 0.4021,
      "step": 6294
    },
    {
      "epoch": 0.28888073057684366,
      "grad_norm": 0.4243152141571045,
      "learning_rate": 9.851231378765029e-06,
      "loss": 0.3325,
      "step": 6295
    },
    {
      "epoch": 0.2889266210820981,
      "grad_norm": 0.4444849193096161,
      "learning_rate": 9.85117200800071e-06,
      "loss": 0.3756,
      "step": 6296
    },
    {
      "epoch": 0.2889725115873526,
      "grad_norm": 0.48350393772125244,
      "learning_rate": 9.851112625570853e-06,
      "loss": 0.4251,
      "step": 6297
    },
    {
      "epoch": 0.28901840209260704,
      "grad_norm": 0.46487635374069214,
      "learning_rate": 9.851053231475603e-06,
      "loss": 0.3936,
      "step": 6298
    },
    {
      "epoch": 0.2890642925978615,
      "grad_norm": 0.47869253158569336,
      "learning_rate": 9.8509938257151e-06,
      "loss": 0.4332,
      "step": 6299
    },
    {
      "epoch": 0.289110183103116,
      "grad_norm": 0.45086222887039185,
      "learning_rate": 9.850934408289488e-06,
      "loss": 0.3187,
      "step": 6300
    },
    {
      "epoch": 0.2891560736083704,
      "grad_norm": 0.5071315169334412,
      "learning_rate": 9.850874979198908e-06,
      "loss": 0.4334,
      "step": 6301
    },
    {
      "epoch": 0.28920196411362487,
      "grad_norm": 0.4443237781524658,
      "learning_rate": 9.850815538443505e-06,
      "loss": 0.3971,
      "step": 6302
    },
    {
      "epoch": 0.28924785461887936,
      "grad_norm": 0.47912678122520447,
      "learning_rate": 9.850756086023423e-06,
      "loss": 0.4374,
      "step": 6303
    },
    {
      "epoch": 0.2892937451241338,
      "grad_norm": 0.4409722685813904,
      "learning_rate": 9.850696621938803e-06,
      "loss": 0.3711,
      "step": 6304
    },
    {
      "epoch": 0.2893396356293883,
      "grad_norm": 0.5260995626449585,
      "learning_rate": 9.850637146189789e-06,
      "loss": 0.4472,
      "step": 6305
    },
    {
      "epoch": 0.28938552613464275,
      "grad_norm": 0.45590466260910034,
      "learning_rate": 9.850577658776523e-06,
      "loss": 0.4021,
      "step": 6306
    },
    {
      "epoch": 0.2894314166398972,
      "grad_norm": 0.4203667640686035,
      "learning_rate": 9.850518159699149e-06,
      "loss": 0.3079,
      "step": 6307
    },
    {
      "epoch": 0.2894773071451517,
      "grad_norm": 0.4631643295288086,
      "learning_rate": 9.85045864895781e-06,
      "loss": 0.4046,
      "step": 6308
    },
    {
      "epoch": 0.28952319765040613,
      "grad_norm": 0.46179717779159546,
      "learning_rate": 9.850399126552649e-06,
      "loss": 0.3801,
      "step": 6309
    },
    {
      "epoch": 0.28956908815566057,
      "grad_norm": 0.4564787745475769,
      "learning_rate": 9.850339592483807e-06,
      "loss": 0.3699,
      "step": 6310
    },
    {
      "epoch": 0.28961497866091507,
      "grad_norm": 0.44304466247558594,
      "learning_rate": 9.850280046751431e-06,
      "loss": 0.3717,
      "step": 6311
    },
    {
      "epoch": 0.2896608691661695,
      "grad_norm": 0.45484358072280884,
      "learning_rate": 9.85022048935566e-06,
      "loss": 0.4025,
      "step": 6312
    },
    {
      "epoch": 0.289706759671424,
      "grad_norm": 0.5319861173629761,
      "learning_rate": 9.85016092029664e-06,
      "loss": 0.5365,
      "step": 6313
    },
    {
      "epoch": 0.28975265017667845,
      "grad_norm": 0.4413909614086151,
      "learning_rate": 9.850101339574516e-06,
      "loss": 0.3917,
      "step": 6314
    },
    {
      "epoch": 0.2897985406819329,
      "grad_norm": 0.43781182169914246,
      "learning_rate": 9.850041747189428e-06,
      "loss": 0.3902,
      "step": 6315
    },
    {
      "epoch": 0.2898444311871874,
      "grad_norm": 0.4508521854877472,
      "learning_rate": 9.84998214314152e-06,
      "loss": 0.4106,
      "step": 6316
    },
    {
      "epoch": 0.28989032169244183,
      "grad_norm": 0.4941001832485199,
      "learning_rate": 9.849922527430937e-06,
      "loss": 0.5126,
      "step": 6317
    },
    {
      "epoch": 0.2899362121976963,
      "grad_norm": 0.41639837622642517,
      "learning_rate": 9.849862900057818e-06,
      "loss": 0.3229,
      "step": 6318
    },
    {
      "epoch": 0.2899821027029508,
      "grad_norm": 0.49247151613235474,
      "learning_rate": 9.849803261022313e-06,
      "loss": 0.4942,
      "step": 6319
    },
    {
      "epoch": 0.2900279932082052,
      "grad_norm": 1.1304585933685303,
      "learning_rate": 9.849743610324558e-06,
      "loss": 0.4759,
      "step": 6320
    },
    {
      "epoch": 0.2900738837134597,
      "grad_norm": 0.4426749050617218,
      "learning_rate": 9.849683947964704e-06,
      "loss": 0.3618,
      "step": 6321
    },
    {
      "epoch": 0.29011977421871415,
      "grad_norm": 0.46332091093063354,
      "learning_rate": 9.849624273942888e-06,
      "loss": 0.3773,
      "step": 6322
    },
    {
      "epoch": 0.2901656647239686,
      "grad_norm": 0.4430350661277771,
      "learning_rate": 9.849564588259257e-06,
      "loss": 0.4225,
      "step": 6323
    },
    {
      "epoch": 0.2902115552292231,
      "grad_norm": 0.44035014510154724,
      "learning_rate": 9.849504890913955e-06,
      "loss": 0.4053,
      "step": 6324
    },
    {
      "epoch": 0.29025744573447754,
      "grad_norm": 0.50107741355896,
      "learning_rate": 9.849445181907123e-06,
      "loss": 0.5099,
      "step": 6325
    },
    {
      "epoch": 0.290303336239732,
      "grad_norm": 0.5155110955238342,
      "learning_rate": 9.849385461238905e-06,
      "loss": 0.4322,
      "step": 6326
    },
    {
      "epoch": 0.2903492267449865,
      "grad_norm": 0.46884647011756897,
      "learning_rate": 9.849325728909447e-06,
      "loss": 0.3444,
      "step": 6327
    },
    {
      "epoch": 0.2903951172502409,
      "grad_norm": 0.5182893872261047,
      "learning_rate": 9.84926598491889e-06,
      "loss": 0.5378,
      "step": 6328
    },
    {
      "epoch": 0.29044100775549536,
      "grad_norm": 0.48908722400665283,
      "learning_rate": 9.849206229267379e-06,
      "loss": 0.4469,
      "step": 6329
    },
    {
      "epoch": 0.29048689826074986,
      "grad_norm": 0.4591538906097412,
      "learning_rate": 9.849146461955059e-06,
      "loss": 0.4338,
      "step": 6330
    },
    {
      "epoch": 0.2905327887660043,
      "grad_norm": 0.38945627212524414,
      "learning_rate": 9.84908668298207e-06,
      "loss": 0.3045,
      "step": 6331
    },
    {
      "epoch": 0.2905786792712588,
      "grad_norm": 0.49531134963035583,
      "learning_rate": 9.849026892348559e-06,
      "loss": 0.5334,
      "step": 6332
    },
    {
      "epoch": 0.29062456977651324,
      "grad_norm": 0.4673563539981842,
      "learning_rate": 9.848967090054669e-06,
      "loss": 0.3642,
      "step": 6333
    },
    {
      "epoch": 0.2906704602817677,
      "grad_norm": 0.46669042110443115,
      "learning_rate": 9.848907276100544e-06,
      "loss": 0.3777,
      "step": 6334
    },
    {
      "epoch": 0.2907163507870222,
      "grad_norm": 0.5021378993988037,
      "learning_rate": 9.848847450486324e-06,
      "loss": 0.4991,
      "step": 6335
    },
    {
      "epoch": 0.2907622412922766,
      "grad_norm": 0.4477701187133789,
      "learning_rate": 9.848787613212161e-06,
      "loss": 0.3844,
      "step": 6336
    },
    {
      "epoch": 0.29080813179753107,
      "grad_norm": 0.7059426307678223,
      "learning_rate": 9.848727764278192e-06,
      "loss": 0.4781,
      "step": 6337
    },
    {
      "epoch": 0.29085402230278556,
      "grad_norm": 0.46379926800727844,
      "learning_rate": 9.848667903684563e-06,
      "loss": 0.3711,
      "step": 6338
    },
    {
      "epoch": 0.29089991280804,
      "grad_norm": 0.5228615999221802,
      "learning_rate": 9.848608031431417e-06,
      "loss": 0.4622,
      "step": 6339
    },
    {
      "epoch": 0.2909458033132945,
      "grad_norm": 0.5413428544998169,
      "learning_rate": 9.8485481475189e-06,
      "loss": 0.5449,
      "step": 6340
    },
    {
      "epoch": 0.29099169381854895,
      "grad_norm": 0.47650569677352905,
      "learning_rate": 9.848488251947154e-06,
      "loss": 0.4415,
      "step": 6341
    },
    {
      "epoch": 0.2910375843238034,
      "grad_norm": 0.491497665643692,
      "learning_rate": 9.848428344716325e-06,
      "loss": 0.4288,
      "step": 6342
    },
    {
      "epoch": 0.2910834748290579,
      "grad_norm": 0.4416244924068451,
      "learning_rate": 9.848368425826557e-06,
      "loss": 0.339,
      "step": 6343
    },
    {
      "epoch": 0.29112936533431233,
      "grad_norm": 0.496927946805954,
      "learning_rate": 9.848308495277991e-06,
      "loss": 0.4707,
      "step": 6344
    },
    {
      "epoch": 0.29117525583956677,
      "grad_norm": 0.47234657406806946,
      "learning_rate": 9.848248553070776e-06,
      "loss": 0.4215,
      "step": 6345
    },
    {
      "epoch": 0.29122114634482127,
      "grad_norm": 0.4572868049144745,
      "learning_rate": 9.848188599205051e-06,
      "loss": 0.367,
      "step": 6346
    },
    {
      "epoch": 0.2912670368500757,
      "grad_norm": 0.45265915989875793,
      "learning_rate": 9.848128633680962e-06,
      "loss": 0.4051,
      "step": 6347
    },
    {
      "epoch": 0.2913129273553302,
      "grad_norm": 0.4708366394042969,
      "learning_rate": 9.848068656498655e-06,
      "loss": 0.3957,
      "step": 6348
    },
    {
      "epoch": 0.29135881786058465,
      "grad_norm": 0.5093710422515869,
      "learning_rate": 9.848008667658274e-06,
      "loss": 0.4993,
      "step": 6349
    },
    {
      "epoch": 0.2914047083658391,
      "grad_norm": 0.4938952922821045,
      "learning_rate": 9.84794866715996e-06,
      "loss": 0.4665,
      "step": 6350
    },
    {
      "epoch": 0.2914505988710936,
      "grad_norm": 0.4694634675979614,
      "learning_rate": 9.847888655003861e-06,
      "loss": 0.4194,
      "step": 6351
    },
    {
      "epoch": 0.29149648937634803,
      "grad_norm": 0.4938065707683563,
      "learning_rate": 9.84782863119012e-06,
      "loss": 0.3888,
      "step": 6352
    },
    {
      "epoch": 0.2915423798816025,
      "grad_norm": 0.48033443093299866,
      "learning_rate": 9.84776859571888e-06,
      "loss": 0.385,
      "step": 6353
    },
    {
      "epoch": 0.291588270386857,
      "grad_norm": 0.4907785654067993,
      "learning_rate": 9.847708548590286e-06,
      "loss": 0.4851,
      "step": 6354
    },
    {
      "epoch": 0.2916341608921114,
      "grad_norm": 0.4656744599342346,
      "learning_rate": 9.847648489804484e-06,
      "loss": 0.4369,
      "step": 6355
    },
    {
      "epoch": 0.29168005139736586,
      "grad_norm": 0.4629974067211151,
      "learning_rate": 9.847588419361617e-06,
      "loss": 0.4166,
      "step": 6356
    },
    {
      "epoch": 0.29172594190262036,
      "grad_norm": 0.46845969557762146,
      "learning_rate": 9.847528337261831e-06,
      "loss": 0.4363,
      "step": 6357
    },
    {
      "epoch": 0.2917718324078748,
      "grad_norm": 0.42425650358200073,
      "learning_rate": 9.84746824350527e-06,
      "loss": 0.3541,
      "step": 6358
    },
    {
      "epoch": 0.2918177229131293,
      "grad_norm": 0.511037290096283,
      "learning_rate": 9.847408138092075e-06,
      "loss": 0.4267,
      "step": 6359
    },
    {
      "epoch": 0.29186361341838374,
      "grad_norm": 0.46509623527526855,
      "learning_rate": 9.847348021022395e-06,
      "loss": 0.3967,
      "step": 6360
    },
    {
      "epoch": 0.2919095039236382,
      "grad_norm": 0.4297862946987152,
      "learning_rate": 9.847287892296373e-06,
      "loss": 0.3803,
      "step": 6361
    },
    {
      "epoch": 0.2919553944288927,
      "grad_norm": 0.4749957323074341,
      "learning_rate": 9.847227751914153e-06,
      "loss": 0.4744,
      "step": 6362
    },
    {
      "epoch": 0.2920012849341471,
      "grad_norm": 0.4495031237602234,
      "learning_rate": 9.847167599875878e-06,
      "loss": 0.3885,
      "step": 6363
    },
    {
      "epoch": 0.29204717543940156,
      "grad_norm": 0.46682700514793396,
      "learning_rate": 9.847107436181698e-06,
      "loss": 0.4055,
      "step": 6364
    },
    {
      "epoch": 0.29209306594465606,
      "grad_norm": 0.49244606494903564,
      "learning_rate": 9.847047260831752e-06,
      "loss": 0.4439,
      "step": 6365
    },
    {
      "epoch": 0.2921389564499105,
      "grad_norm": 0.4242806136608124,
      "learning_rate": 9.846987073826189e-06,
      "loss": 0.3485,
      "step": 6366
    },
    {
      "epoch": 0.292184846955165,
      "grad_norm": 0.44537290930747986,
      "learning_rate": 9.846926875165151e-06,
      "loss": 0.3951,
      "step": 6367
    },
    {
      "epoch": 0.29223073746041944,
      "grad_norm": 0.5811315178871155,
      "learning_rate": 9.846866664848784e-06,
      "loss": 0.3869,
      "step": 6368
    },
    {
      "epoch": 0.2922766279656739,
      "grad_norm": 0.4557437002658844,
      "learning_rate": 9.846806442877231e-06,
      "loss": 0.3864,
      "step": 6369
    },
    {
      "epoch": 0.2923225184709284,
      "grad_norm": 0.4330872893333435,
      "learning_rate": 9.846746209250639e-06,
      "loss": 0.3092,
      "step": 6370
    },
    {
      "epoch": 0.2923684089761828,
      "grad_norm": 0.48558738827705383,
      "learning_rate": 9.846685963969153e-06,
      "loss": 0.4185,
      "step": 6371
    },
    {
      "epoch": 0.29241429948143727,
      "grad_norm": 0.46199607849121094,
      "learning_rate": 9.846625707032916e-06,
      "loss": 0.3748,
      "step": 6372
    },
    {
      "epoch": 0.29246018998669177,
      "grad_norm": 0.4925325810909271,
      "learning_rate": 9.846565438442074e-06,
      "loss": 0.4596,
      "step": 6373
    },
    {
      "epoch": 0.2925060804919462,
      "grad_norm": 0.44381558895111084,
      "learning_rate": 9.846505158196772e-06,
      "loss": 0.4229,
      "step": 6374
    },
    {
      "epoch": 0.2925519709972007,
      "grad_norm": 0.44838079810142517,
      "learning_rate": 9.846444866297155e-06,
      "loss": 0.3692,
      "step": 6375
    },
    {
      "epoch": 0.29259786150245515,
      "grad_norm": 0.42445579171180725,
      "learning_rate": 9.846384562743368e-06,
      "loss": 0.3145,
      "step": 6376
    },
    {
      "epoch": 0.2926437520077096,
      "grad_norm": 0.43952476978302,
      "learning_rate": 9.846324247535553e-06,
      "loss": 0.3756,
      "step": 6377
    },
    {
      "epoch": 0.2926896425129641,
      "grad_norm": 0.4262857139110565,
      "learning_rate": 9.84626392067386e-06,
      "loss": 0.2983,
      "step": 6378
    },
    {
      "epoch": 0.29273553301821853,
      "grad_norm": 0.41948673129081726,
      "learning_rate": 9.846203582158432e-06,
      "loss": 0.3118,
      "step": 6379
    },
    {
      "epoch": 0.292781423523473,
      "grad_norm": 0.4843999445438385,
      "learning_rate": 9.846143231989414e-06,
      "loss": 0.4254,
      "step": 6380
    },
    {
      "epoch": 0.29282731402872747,
      "grad_norm": 0.4298352003097534,
      "learning_rate": 9.846082870166948e-06,
      "loss": 0.3348,
      "step": 6381
    },
    {
      "epoch": 0.2928732045339819,
      "grad_norm": 0.48264268040657043,
      "learning_rate": 9.846022496691186e-06,
      "loss": 0.434,
      "step": 6382
    },
    {
      "epoch": 0.29291909503923635,
      "grad_norm": 0.4493555724620819,
      "learning_rate": 9.845962111562267e-06,
      "loss": 0.3571,
      "step": 6383
    },
    {
      "epoch": 0.29296498554449085,
      "grad_norm": 0.4615822434425354,
      "learning_rate": 9.84590171478034e-06,
      "loss": 0.4247,
      "step": 6384
    },
    {
      "epoch": 0.2930108760497453,
      "grad_norm": 0.4548717439174652,
      "learning_rate": 9.845841306345547e-06,
      "loss": 0.3291,
      "step": 6385
    },
    {
      "epoch": 0.2930567665549998,
      "grad_norm": 0.4489175081253052,
      "learning_rate": 9.845780886258037e-06,
      "loss": 0.3643,
      "step": 6386
    },
    {
      "epoch": 0.29310265706025423,
      "grad_norm": 0.49297118186950684,
      "learning_rate": 9.845720454517952e-06,
      "loss": 0.444,
      "step": 6387
    },
    {
      "epoch": 0.2931485475655087,
      "grad_norm": 0.43482524156570435,
      "learning_rate": 9.845660011125439e-06,
      "loss": 0.3757,
      "step": 6388
    },
    {
      "epoch": 0.2931944380707632,
      "grad_norm": 0.4555683434009552,
      "learning_rate": 9.845599556080643e-06,
      "loss": 0.3224,
      "step": 6389
    },
    {
      "epoch": 0.2932403285760176,
      "grad_norm": 0.4259699285030365,
      "learning_rate": 9.84553908938371e-06,
      "loss": 0.3416,
      "step": 6390
    },
    {
      "epoch": 0.29328621908127206,
      "grad_norm": 0.47314390540122986,
      "learning_rate": 9.845478611034785e-06,
      "loss": 0.3927,
      "step": 6391
    },
    {
      "epoch": 0.29333210958652656,
      "grad_norm": 0.4683067798614502,
      "learning_rate": 9.845418121034012e-06,
      "loss": 0.36,
      "step": 6392
    },
    {
      "epoch": 0.293378000091781,
      "grad_norm": 0.5327858328819275,
      "learning_rate": 9.845357619381538e-06,
      "loss": 0.5268,
      "step": 6393
    },
    {
      "epoch": 0.2934238905970355,
      "grad_norm": 0.4776647984981537,
      "learning_rate": 9.845297106077508e-06,
      "loss": 0.4274,
      "step": 6394
    },
    {
      "epoch": 0.29346978110228994,
      "grad_norm": 0.4570990204811096,
      "learning_rate": 9.845236581122068e-06,
      "loss": 0.3529,
      "step": 6395
    },
    {
      "epoch": 0.2935156716075444,
      "grad_norm": 0.46448954939842224,
      "learning_rate": 9.845176044515364e-06,
      "loss": 0.4181,
      "step": 6396
    },
    {
      "epoch": 0.2935615621127989,
      "grad_norm": 0.4757956266403198,
      "learning_rate": 9.84511549625754e-06,
      "loss": 0.4223,
      "step": 6397
    },
    {
      "epoch": 0.2936074526180533,
      "grad_norm": 0.48264947533607483,
      "learning_rate": 9.845054936348742e-06,
      "loss": 0.3668,
      "step": 6398
    },
    {
      "epoch": 0.29365334312330776,
      "grad_norm": 0.4112507402896881,
      "learning_rate": 9.844994364789116e-06,
      "loss": 0.3455,
      "step": 6399
    },
    {
      "epoch": 0.29369923362856226,
      "grad_norm": 0.462025910615921,
      "learning_rate": 9.84493378157881e-06,
      "loss": 0.4124,
      "step": 6400
    },
    {
      "epoch": 0.2937451241338167,
      "grad_norm": 0.5162631273269653,
      "learning_rate": 9.844873186717965e-06,
      "loss": 0.4927,
      "step": 6401
    },
    {
      "epoch": 0.2937910146390712,
      "grad_norm": 0.4539836049079895,
      "learning_rate": 9.844812580206729e-06,
      "loss": 0.3894,
      "step": 6402
    },
    {
      "epoch": 0.29383690514432564,
      "grad_norm": 0.4653432369232178,
      "learning_rate": 9.84475196204525e-06,
      "loss": 0.3955,
      "step": 6403
    },
    {
      "epoch": 0.2938827956495801,
      "grad_norm": 0.44719579815864563,
      "learning_rate": 9.844691332233669e-06,
      "loss": 0.3909,
      "step": 6404
    },
    {
      "epoch": 0.2939286861548346,
      "grad_norm": 0.45951777696609497,
      "learning_rate": 9.844630690772134e-06,
      "loss": 0.4455,
      "step": 6405
    },
    {
      "epoch": 0.293974576660089,
      "grad_norm": 0.4864802360534668,
      "learning_rate": 9.844570037660793e-06,
      "loss": 0.4325,
      "step": 6406
    },
    {
      "epoch": 0.29402046716534347,
      "grad_norm": 0.4538343548774719,
      "learning_rate": 9.84450937289979e-06,
      "loss": 0.3527,
      "step": 6407
    },
    {
      "epoch": 0.29406635767059797,
      "grad_norm": 0.4541611075401306,
      "learning_rate": 9.84444869648927e-06,
      "loss": 0.389,
      "step": 6408
    },
    {
      "epoch": 0.2941122481758524,
      "grad_norm": 0.4638700783252716,
      "learning_rate": 9.844388008429381e-06,
      "loss": 0.3964,
      "step": 6409
    },
    {
      "epoch": 0.2941581386811069,
      "grad_norm": 0.47902151942253113,
      "learning_rate": 9.844327308720267e-06,
      "loss": 0.4073,
      "step": 6410
    },
    {
      "epoch": 0.29420402918636135,
      "grad_norm": 0.47087058424949646,
      "learning_rate": 9.844266597362075e-06,
      "loss": 0.4433,
      "step": 6411
    },
    {
      "epoch": 0.2942499196916158,
      "grad_norm": 0.4772723913192749,
      "learning_rate": 9.84420587435495e-06,
      "loss": 0.4385,
      "step": 6412
    },
    {
      "epoch": 0.2942958101968703,
      "grad_norm": 0.4500417113304138,
      "learning_rate": 9.84414513969904e-06,
      "loss": 0.3983,
      "step": 6413
    },
    {
      "epoch": 0.29434170070212473,
      "grad_norm": 0.4624321460723877,
      "learning_rate": 9.844084393394489e-06,
      "loss": 0.3858,
      "step": 6414
    },
    {
      "epoch": 0.2943875912073792,
      "grad_norm": 0.479070246219635,
      "learning_rate": 9.844023635441444e-06,
      "loss": 0.3847,
      "step": 6415
    },
    {
      "epoch": 0.29443348171263367,
      "grad_norm": 0.46939918398857117,
      "learning_rate": 9.84396286584005e-06,
      "loss": 0.4602,
      "step": 6416
    },
    {
      "epoch": 0.2944793722178881,
      "grad_norm": 0.44934970140457153,
      "learning_rate": 9.843902084590456e-06,
      "loss": 0.3572,
      "step": 6417
    },
    {
      "epoch": 0.29452526272314256,
      "grad_norm": 0.45539265871047974,
      "learning_rate": 9.843841291692807e-06,
      "loss": 0.4148,
      "step": 6418
    },
    {
      "epoch": 0.29457115322839705,
      "grad_norm": 0.4768684208393097,
      "learning_rate": 9.843780487147245e-06,
      "loss": 0.4083,
      "step": 6419
    },
    {
      "epoch": 0.2946170437336515,
      "grad_norm": 0.45186012983322144,
      "learning_rate": 9.843719670953922e-06,
      "loss": 0.4194,
      "step": 6420
    },
    {
      "epoch": 0.294662934238906,
      "grad_norm": 0.4309633672237396,
      "learning_rate": 9.843658843112983e-06,
      "loss": 0.3339,
      "step": 6421
    },
    {
      "epoch": 0.29470882474416044,
      "grad_norm": 0.45534074306488037,
      "learning_rate": 9.84359800362457e-06,
      "loss": 0.3801,
      "step": 6422
    },
    {
      "epoch": 0.2947547152494149,
      "grad_norm": 0.5023715496063232,
      "learning_rate": 9.843537152488834e-06,
      "loss": 0.4925,
      "step": 6423
    },
    {
      "epoch": 0.2948006057546694,
      "grad_norm": 0.4333205223083496,
      "learning_rate": 9.84347628970592e-06,
      "loss": 0.3989,
      "step": 6424
    },
    {
      "epoch": 0.2948464962599238,
      "grad_norm": 0.49474775791168213,
      "learning_rate": 9.843415415275974e-06,
      "loss": 0.4363,
      "step": 6425
    },
    {
      "epoch": 0.29489238676517826,
      "grad_norm": 0.4935716688632965,
      "learning_rate": 9.843354529199144e-06,
      "loss": 0.4546,
      "step": 6426
    },
    {
      "epoch": 0.29493827727043276,
      "grad_norm": 0.45932480692863464,
      "learning_rate": 9.843293631475571e-06,
      "loss": 0.4189,
      "step": 6427
    },
    {
      "epoch": 0.2949841677756872,
      "grad_norm": 0.4477168321609497,
      "learning_rate": 9.84323272210541e-06,
      "loss": 0.3724,
      "step": 6428
    },
    {
      "epoch": 0.2950300582809417,
      "grad_norm": 0.4423210918903351,
      "learning_rate": 9.8431718010888e-06,
      "loss": 0.3546,
      "step": 6429
    },
    {
      "epoch": 0.29507594878619614,
      "grad_norm": 0.4845663607120514,
      "learning_rate": 9.84311086842589e-06,
      "loss": 0.4647,
      "step": 6430
    },
    {
      "epoch": 0.2951218392914506,
      "grad_norm": 0.4656390845775604,
      "learning_rate": 9.84304992411683e-06,
      "loss": 0.3645,
      "step": 6431
    },
    {
      "epoch": 0.2951677297967051,
      "grad_norm": 0.4559401273727417,
      "learning_rate": 9.842988968161762e-06,
      "loss": 0.3815,
      "step": 6432
    },
    {
      "epoch": 0.2952136203019595,
      "grad_norm": 0.507779061794281,
      "learning_rate": 9.842928000560833e-06,
      "loss": 0.4388,
      "step": 6433
    },
    {
      "epoch": 0.29525951080721397,
      "grad_norm": 0.4545881748199463,
      "learning_rate": 9.842867021314192e-06,
      "loss": 0.3696,
      "step": 6434
    },
    {
      "epoch": 0.29530540131246846,
      "grad_norm": 0.49496033787727356,
      "learning_rate": 9.842806030421983e-06,
      "loss": 0.44,
      "step": 6435
    },
    {
      "epoch": 0.2953512918177229,
      "grad_norm": 0.4435413181781769,
      "learning_rate": 9.842745027884353e-06,
      "loss": 0.3548,
      "step": 6436
    },
    {
      "epoch": 0.2953971823229774,
      "grad_norm": 0.45298701524734497,
      "learning_rate": 9.842684013701452e-06,
      "loss": 0.3755,
      "step": 6437
    },
    {
      "epoch": 0.29544307282823185,
      "grad_norm": 0.5006486773490906,
      "learning_rate": 9.842622987873423e-06,
      "loss": 0.5108,
      "step": 6438
    },
    {
      "epoch": 0.2954889633334863,
      "grad_norm": 0.41208890080451965,
      "learning_rate": 9.842561950400414e-06,
      "loss": 0.3358,
      "step": 6439
    },
    {
      "epoch": 0.2955348538387408,
      "grad_norm": 0.4643336832523346,
      "learning_rate": 9.842500901282573e-06,
      "loss": 0.4702,
      "step": 6440
    },
    {
      "epoch": 0.2955807443439952,
      "grad_norm": 0.5686439871788025,
      "learning_rate": 9.842439840520046e-06,
      "loss": 0.5074,
      "step": 6441
    },
    {
      "epoch": 0.29562663484924967,
      "grad_norm": 0.497657835483551,
      "learning_rate": 9.842378768112977e-06,
      "loss": 0.4649,
      "step": 6442
    },
    {
      "epoch": 0.29567252535450417,
      "grad_norm": 0.4624097943305969,
      "learning_rate": 9.842317684061518e-06,
      "loss": 0.3705,
      "step": 6443
    },
    {
      "epoch": 0.2957184158597586,
      "grad_norm": 0.5089067220687866,
      "learning_rate": 9.842256588365811e-06,
      "loss": 0.4939,
      "step": 6444
    },
    {
      "epoch": 0.29576430636501305,
      "grad_norm": 0.45058178901672363,
      "learning_rate": 9.842195481026006e-06,
      "loss": 0.3532,
      "step": 6445
    },
    {
      "epoch": 0.29581019687026755,
      "grad_norm": 0.5498747229576111,
      "learning_rate": 9.842134362042251e-06,
      "loss": 0.4785,
      "step": 6446
    },
    {
      "epoch": 0.295856087375522,
      "grad_norm": 0.46191641688346863,
      "learning_rate": 9.842073231414688e-06,
      "loss": 0.4063,
      "step": 6447
    },
    {
      "epoch": 0.2959019778807765,
      "grad_norm": 0.5215046405792236,
      "learning_rate": 9.84201208914347e-06,
      "loss": 0.4213,
      "step": 6448
    },
    {
      "epoch": 0.29594786838603093,
      "grad_norm": 0.44170433282852173,
      "learning_rate": 9.841950935228738e-06,
      "loss": 0.3179,
      "step": 6449
    },
    {
      "epoch": 0.2959937588912854,
      "grad_norm": 0.4883936047554016,
      "learning_rate": 9.841889769670644e-06,
      "loss": 0.4437,
      "step": 6450
    },
    {
      "epoch": 0.2960396493965399,
      "grad_norm": 0.5236921310424805,
      "learning_rate": 9.841828592469334e-06,
      "loss": 0.4653,
      "step": 6451
    },
    {
      "epoch": 0.2960855399017943,
      "grad_norm": 0.482817143201828,
      "learning_rate": 9.841767403624953e-06,
      "loss": 0.3997,
      "step": 6452
    },
    {
      "epoch": 0.29613143040704876,
      "grad_norm": 0.44491875171661377,
      "learning_rate": 9.84170620313765e-06,
      "loss": 0.361,
      "step": 6453
    },
    {
      "epoch": 0.29617732091230325,
      "grad_norm": 0.5162143111228943,
      "learning_rate": 9.84164499100757e-06,
      "loss": 0.4497,
      "step": 6454
    },
    {
      "epoch": 0.2962232114175577,
      "grad_norm": 0.4418305456638336,
      "learning_rate": 9.841583767234865e-06,
      "loss": 0.3561,
      "step": 6455
    },
    {
      "epoch": 0.2962691019228122,
      "grad_norm": 0.5333799719810486,
      "learning_rate": 9.841522531819677e-06,
      "loss": 0.4374,
      "step": 6456
    },
    {
      "epoch": 0.29631499242806664,
      "grad_norm": 0.5086747407913208,
      "learning_rate": 9.841461284762155e-06,
      "loss": 0.4472,
      "step": 6457
    },
    {
      "epoch": 0.2963608829333211,
      "grad_norm": 0.46974003314971924,
      "learning_rate": 9.841400026062449e-06,
      "loss": 0.4156,
      "step": 6458
    },
    {
      "epoch": 0.2964067734385756,
      "grad_norm": 0.5830226540565491,
      "learning_rate": 9.841338755720702e-06,
      "loss": 0.5177,
      "step": 6459
    },
    {
      "epoch": 0.29645266394383,
      "grad_norm": 0.45398828387260437,
      "learning_rate": 9.841277473737063e-06,
      "loss": 0.4129,
      "step": 6460
    },
    {
      "epoch": 0.29649855444908446,
      "grad_norm": 0.5102665424346924,
      "learning_rate": 9.841216180111682e-06,
      "loss": 0.5214,
      "step": 6461
    },
    {
      "epoch": 0.29654444495433896,
      "grad_norm": 0.4631696343421936,
      "learning_rate": 9.841154874844702e-06,
      "loss": 0.3102,
      "step": 6462
    },
    {
      "epoch": 0.2965903354595934,
      "grad_norm": 0.4801292419433594,
      "learning_rate": 9.841093557936273e-06,
      "loss": 0.3703,
      "step": 6463
    },
    {
      "epoch": 0.2966362259648479,
      "grad_norm": 0.45674824714660645,
      "learning_rate": 9.84103222938654e-06,
      "loss": 0.4139,
      "step": 6464
    },
    {
      "epoch": 0.29668211647010234,
      "grad_norm": 0.47155240178108215,
      "learning_rate": 9.840970889195656e-06,
      "loss": 0.419,
      "step": 6465
    },
    {
      "epoch": 0.2967280069753568,
      "grad_norm": 0.5416005253791809,
      "learning_rate": 9.840909537363762e-06,
      "loss": 0.5438,
      "step": 6466
    },
    {
      "epoch": 0.2967738974806113,
      "grad_norm": 0.5135039687156677,
      "learning_rate": 9.84084817389101e-06,
      "loss": 0.5158,
      "step": 6467
    },
    {
      "epoch": 0.2968197879858657,
      "grad_norm": 0.452438622713089,
      "learning_rate": 9.840786798777544e-06,
      "loss": 0.3799,
      "step": 6468
    },
    {
      "epoch": 0.29686567849112017,
      "grad_norm": 0.4335687756538391,
      "learning_rate": 9.840725412023514e-06,
      "loss": 0.3331,
      "step": 6469
    },
    {
      "epoch": 0.29691156899637466,
      "grad_norm": 0.4449588358402252,
      "learning_rate": 9.840664013629069e-06,
      "loss": 0.3797,
      "step": 6470
    },
    {
      "epoch": 0.2969574595016291,
      "grad_norm": 0.4790562391281128,
      "learning_rate": 9.840602603594354e-06,
      "loss": 0.399,
      "step": 6471
    },
    {
      "epoch": 0.29700335000688355,
      "grad_norm": 0.4763764441013336,
      "learning_rate": 9.840541181919518e-06,
      "loss": 0.3955,
      "step": 6472
    },
    {
      "epoch": 0.29704924051213805,
      "grad_norm": 0.4998711347579956,
      "learning_rate": 9.840479748604706e-06,
      "loss": 0.4513,
      "step": 6473
    },
    {
      "epoch": 0.2970951310173925,
      "grad_norm": 0.45951715111732483,
      "learning_rate": 9.84041830365007e-06,
      "loss": 0.413,
      "step": 6474
    },
    {
      "epoch": 0.297141021522647,
      "grad_norm": 0.4189939498901367,
      "learning_rate": 9.840356847055754e-06,
      "loss": 0.3119,
      "step": 6475
    },
    {
      "epoch": 0.29718691202790143,
      "grad_norm": 0.4655221998691559,
      "learning_rate": 9.840295378821909e-06,
      "loss": 0.3727,
      "step": 6476
    },
    {
      "epoch": 0.29723280253315587,
      "grad_norm": 0.439113587141037,
      "learning_rate": 9.84023389894868e-06,
      "loss": 0.3901,
      "step": 6477
    },
    {
      "epoch": 0.29727869303841037,
      "grad_norm": 0.49309042096138,
      "learning_rate": 9.840172407436217e-06,
      "loss": 0.4295,
      "step": 6478
    },
    {
      "epoch": 0.2973245835436648,
      "grad_norm": 0.5152106881141663,
      "learning_rate": 9.840110904284668e-06,
      "loss": 0.5105,
      "step": 6479
    },
    {
      "epoch": 0.29737047404891925,
      "grad_norm": 0.49067193269729614,
      "learning_rate": 9.840049389494177e-06,
      "loss": 0.3915,
      "step": 6480
    },
    {
      "epoch": 0.29741636455417375,
      "grad_norm": 0.5083408355712891,
      "learning_rate": 9.839987863064897e-06,
      "loss": 0.5256,
      "step": 6481
    },
    {
      "epoch": 0.2974622550594282,
      "grad_norm": 0.4649994969367981,
      "learning_rate": 9.839926324996974e-06,
      "loss": 0.4401,
      "step": 6482
    },
    {
      "epoch": 0.2975081455646827,
      "grad_norm": 0.49472588300704956,
      "learning_rate": 9.839864775290555e-06,
      "loss": 0.5231,
      "step": 6483
    },
    {
      "epoch": 0.29755403606993713,
      "grad_norm": 0.5070353746414185,
      "learning_rate": 9.839803213945788e-06,
      "loss": 0.5542,
      "step": 6484
    },
    {
      "epoch": 0.2975999265751916,
      "grad_norm": 0.48984256386756897,
      "learning_rate": 9.839741640962822e-06,
      "loss": 0.4659,
      "step": 6485
    },
    {
      "epoch": 0.2976458170804461,
      "grad_norm": 0.4643564522266388,
      "learning_rate": 9.839680056341805e-06,
      "loss": 0.4231,
      "step": 6486
    },
    {
      "epoch": 0.2976917075857005,
      "grad_norm": 0.4469192922115326,
      "learning_rate": 9.839618460082885e-06,
      "loss": 0.3666,
      "step": 6487
    },
    {
      "epoch": 0.29773759809095496,
      "grad_norm": 0.41152966022491455,
      "learning_rate": 9.839556852186209e-06,
      "loss": 0.3313,
      "step": 6488
    },
    {
      "epoch": 0.29778348859620946,
      "grad_norm": 0.4438681900501251,
      "learning_rate": 9.839495232651928e-06,
      "loss": 0.3721,
      "step": 6489
    },
    {
      "epoch": 0.2978293791014639,
      "grad_norm": 0.4528467655181885,
      "learning_rate": 9.839433601480188e-06,
      "loss": 0.4153,
      "step": 6490
    },
    {
      "epoch": 0.2978752696067184,
      "grad_norm": 0.45829904079437256,
      "learning_rate": 9.839371958671138e-06,
      "loss": 0.4018,
      "step": 6491
    },
    {
      "epoch": 0.29792116011197284,
      "grad_norm": 0.48633503913879395,
      "learning_rate": 9.839310304224925e-06,
      "loss": 0.4758,
      "step": 6492
    },
    {
      "epoch": 0.2979670506172273,
      "grad_norm": 0.47346365451812744,
      "learning_rate": 9.839248638141697e-06,
      "loss": 0.4716,
      "step": 6493
    },
    {
      "epoch": 0.2980129411224818,
      "grad_norm": 0.4712088108062744,
      "learning_rate": 9.839186960421605e-06,
      "loss": 0.4176,
      "step": 6494
    },
    {
      "epoch": 0.2980588316277362,
      "grad_norm": 0.46418920159339905,
      "learning_rate": 9.839125271064794e-06,
      "loss": 0.3809,
      "step": 6495
    },
    {
      "epoch": 0.29810472213299066,
      "grad_norm": 0.5189914703369141,
      "learning_rate": 9.839063570071416e-06,
      "loss": 0.513,
      "step": 6496
    },
    {
      "epoch": 0.29815061263824516,
      "grad_norm": 0.4563591480255127,
      "learning_rate": 9.839001857441616e-06,
      "loss": 0.3524,
      "step": 6497
    },
    {
      "epoch": 0.2981965031434996,
      "grad_norm": 0.4932597577571869,
      "learning_rate": 9.838940133175544e-06,
      "loss": 0.4692,
      "step": 6498
    },
    {
      "epoch": 0.2982423936487541,
      "grad_norm": 0.49317753314971924,
      "learning_rate": 9.838878397273348e-06,
      "loss": 0.4512,
      "step": 6499
    },
    {
      "epoch": 0.29828828415400854,
      "grad_norm": 0.4801498055458069,
      "learning_rate": 9.838816649735177e-06,
      "loss": 0.4833,
      "step": 6500
    },
    {
      "epoch": 0.298334174659263,
      "grad_norm": 0.4859009385108948,
      "learning_rate": 9.83875489056118e-06,
      "loss": 0.415,
      "step": 6501
    },
    {
      "epoch": 0.2983800651645175,
      "grad_norm": 0.4693339765071869,
      "learning_rate": 9.838693119751504e-06,
      "loss": 0.3941,
      "step": 6502
    },
    {
      "epoch": 0.2984259556697719,
      "grad_norm": 0.4729110300540924,
      "learning_rate": 9.838631337306296e-06,
      "loss": 0.4608,
      "step": 6503
    },
    {
      "epoch": 0.29847184617502637,
      "grad_norm": 0.4753352105617523,
      "learning_rate": 9.838569543225708e-06,
      "loss": 0.4244,
      "step": 6504
    },
    {
      "epoch": 0.29851773668028087,
      "grad_norm": 0.436207115650177,
      "learning_rate": 9.83850773750989e-06,
      "loss": 0.339,
      "step": 6505
    },
    {
      "epoch": 0.2985636271855353,
      "grad_norm": 0.4982065260410309,
      "learning_rate": 9.838445920158986e-06,
      "loss": 0.446,
      "step": 6506
    },
    {
      "epoch": 0.29860951769078975,
      "grad_norm": 0.4677481949329376,
      "learning_rate": 9.838384091173146e-06,
      "loss": 0.45,
      "step": 6507
    },
    {
      "epoch": 0.29865540819604425,
      "grad_norm": 0.46237245202064514,
      "learning_rate": 9.83832225055252e-06,
      "loss": 0.385,
      "step": 6508
    },
    {
      "epoch": 0.2987012987012987,
      "grad_norm": 0.43586406111717224,
      "learning_rate": 9.838260398297256e-06,
      "loss": 0.3582,
      "step": 6509
    },
    {
      "epoch": 0.2987471892065532,
      "grad_norm": 0.48615148663520813,
      "learning_rate": 9.838198534407503e-06,
      "loss": 0.4371,
      "step": 6510
    },
    {
      "epoch": 0.29879307971180763,
      "grad_norm": 0.45605766773223877,
      "learning_rate": 9.838136658883408e-06,
      "loss": 0.3935,
      "step": 6511
    },
    {
      "epoch": 0.29883897021706207,
      "grad_norm": 0.4805205762386322,
      "learning_rate": 9.838074771725124e-06,
      "loss": 0.4107,
      "step": 6512
    },
    {
      "epoch": 0.29888486072231657,
      "grad_norm": 0.46869051456451416,
      "learning_rate": 9.838012872932793e-06,
      "loss": 0.4279,
      "step": 6513
    },
    {
      "epoch": 0.298930751227571,
      "grad_norm": 0.4750407040119171,
      "learning_rate": 9.837950962506573e-06,
      "loss": 0.435,
      "step": 6514
    },
    {
      "epoch": 0.29897664173282545,
      "grad_norm": 0.4712446630001068,
      "learning_rate": 9.837889040446604e-06,
      "loss": 0.376,
      "step": 6515
    },
    {
      "epoch": 0.29902253223807995,
      "grad_norm": 0.48265543580055237,
      "learning_rate": 9.837827106753041e-06,
      "loss": 0.4212,
      "step": 6516
    },
    {
      "epoch": 0.2990684227433344,
      "grad_norm": 0.4499903619289398,
      "learning_rate": 9.83776516142603e-06,
      "loss": 0.4286,
      "step": 6517
    },
    {
      "epoch": 0.2991143132485889,
      "grad_norm": 0.46939411759376526,
      "learning_rate": 9.837703204465719e-06,
      "loss": 0.4197,
      "step": 6518
    },
    {
      "epoch": 0.29916020375384333,
      "grad_norm": 0.43522733449935913,
      "learning_rate": 9.837641235872261e-06,
      "loss": 0.3497,
      "step": 6519
    },
    {
      "epoch": 0.2992060942590978,
      "grad_norm": 0.46146097779273987,
      "learning_rate": 9.837579255645802e-06,
      "loss": 0.3878,
      "step": 6520
    },
    {
      "epoch": 0.2992519847643523,
      "grad_norm": 0.4577597975730896,
      "learning_rate": 9.837517263786492e-06,
      "loss": 0.3928,
      "step": 6521
    },
    {
      "epoch": 0.2992978752696067,
      "grad_norm": 0.45282700657844543,
      "learning_rate": 9.837455260294477e-06,
      "loss": 0.3904,
      "step": 6522
    },
    {
      "epoch": 0.29934376577486116,
      "grad_norm": 0.4805585443973541,
      "learning_rate": 9.837393245169911e-06,
      "loss": 0.4518,
      "step": 6523
    },
    {
      "epoch": 0.29938965628011566,
      "grad_norm": 0.46631982922554016,
      "learning_rate": 9.837331218412942e-06,
      "loss": 0.4321,
      "step": 6524
    },
    {
      "epoch": 0.2994355467853701,
      "grad_norm": 0.4699938893318176,
      "learning_rate": 9.837269180023717e-06,
      "loss": 0.3663,
      "step": 6525
    },
    {
      "epoch": 0.2994814372906246,
      "grad_norm": 0.4217190742492676,
      "learning_rate": 9.837207130002387e-06,
      "loss": 0.3209,
      "step": 6526
    },
    {
      "epoch": 0.29952732779587904,
      "grad_norm": 0.6820508241653442,
      "learning_rate": 9.837145068349099e-06,
      "loss": 0.56,
      "step": 6527
    },
    {
      "epoch": 0.2995732183011335,
      "grad_norm": 0.5178728699684143,
      "learning_rate": 9.837082995064005e-06,
      "loss": 0.4858,
      "step": 6528
    },
    {
      "epoch": 0.299619108806388,
      "grad_norm": 0.4568500518798828,
      "learning_rate": 9.837020910147251e-06,
      "loss": 0.3856,
      "step": 6529
    },
    {
      "epoch": 0.2996649993116424,
      "grad_norm": 0.4973050057888031,
      "learning_rate": 9.83695881359899e-06,
      "loss": 0.4751,
      "step": 6530
    },
    {
      "epoch": 0.29971088981689686,
      "grad_norm": 0.45793575048446655,
      "learning_rate": 9.83689670541937e-06,
      "loss": 0.4092,
      "step": 6531
    },
    {
      "epoch": 0.29975678032215136,
      "grad_norm": 0.4828096032142639,
      "learning_rate": 9.836834585608539e-06,
      "loss": 0.4164,
      "step": 6532
    },
    {
      "epoch": 0.2998026708274058,
      "grad_norm": 0.49118292331695557,
      "learning_rate": 9.83677245416665e-06,
      "loss": 0.5064,
      "step": 6533
    },
    {
      "epoch": 0.29984856133266025,
      "grad_norm": 0.4733990728855133,
      "learning_rate": 9.836710311093847e-06,
      "loss": 0.4062,
      "step": 6534
    },
    {
      "epoch": 0.29989445183791474,
      "grad_norm": 0.4406806528568268,
      "learning_rate": 9.836648156390283e-06,
      "loss": 0.3573,
      "step": 6535
    },
    {
      "epoch": 0.2999403423431692,
      "grad_norm": 0.43260520696640015,
      "learning_rate": 9.836585990056106e-06,
      "loss": 0.3823,
      "step": 6536
    },
    {
      "epoch": 0.2999862328484237,
      "grad_norm": 0.450859397649765,
      "learning_rate": 9.836523812091467e-06,
      "loss": 0.383,
      "step": 6537
    },
    {
      "epoch": 0.3000321233536781,
      "grad_norm": 0.4424304962158203,
      "learning_rate": 9.836461622496513e-06,
      "loss": 0.3363,
      "step": 6538
    },
    {
      "epoch": 0.30007801385893257,
      "grad_norm": 0.430082768201828,
      "learning_rate": 9.836399421271398e-06,
      "loss": 0.3244,
      "step": 6539
    },
    {
      "epoch": 0.30012390436418707,
      "grad_norm": 0.45263203978538513,
      "learning_rate": 9.836337208416267e-06,
      "loss": 0.3847,
      "step": 6540
    },
    {
      "epoch": 0.3001697948694415,
      "grad_norm": 0.43024125695228577,
      "learning_rate": 9.836274983931271e-06,
      "loss": 0.3556,
      "step": 6541
    },
    {
      "epoch": 0.30021568537469595,
      "grad_norm": 0.49892905354499817,
      "learning_rate": 9.836212747816561e-06,
      "loss": 0.4458,
      "step": 6542
    },
    {
      "epoch": 0.30026157587995045,
      "grad_norm": 0.46596211194992065,
      "learning_rate": 9.836150500072286e-06,
      "loss": 0.3829,
      "step": 6543
    },
    {
      "epoch": 0.3003074663852049,
      "grad_norm": 0.4795933961868286,
      "learning_rate": 9.836088240698596e-06,
      "loss": 0.425,
      "step": 6544
    },
    {
      "epoch": 0.3003533568904594,
      "grad_norm": 0.49768438935279846,
      "learning_rate": 9.836025969695638e-06,
      "loss": 0.4604,
      "step": 6545
    },
    {
      "epoch": 0.30039924739571383,
      "grad_norm": 0.4352327585220337,
      "learning_rate": 9.835963687063566e-06,
      "loss": 0.3119,
      "step": 6546
    },
    {
      "epoch": 0.3004451379009683,
      "grad_norm": 0.47003909945487976,
      "learning_rate": 9.835901392802527e-06,
      "loss": 0.3984,
      "step": 6547
    },
    {
      "epoch": 0.30049102840622277,
      "grad_norm": 0.46133220195770264,
      "learning_rate": 9.83583908691267e-06,
      "loss": 0.4355,
      "step": 6548
    },
    {
      "epoch": 0.3005369189114772,
      "grad_norm": 0.46906328201293945,
      "learning_rate": 9.835776769394147e-06,
      "loss": 0.4026,
      "step": 6549
    },
    {
      "epoch": 0.30058280941673166,
      "grad_norm": 0.4672048091888428,
      "learning_rate": 9.835714440247107e-06,
      "loss": 0.4234,
      "step": 6550
    },
    {
      "epoch": 0.30062869992198615,
      "grad_norm": 0.5263383984565735,
      "learning_rate": 9.8356520994717e-06,
      "loss": 0.5082,
      "step": 6551
    },
    {
      "epoch": 0.3006745904272406,
      "grad_norm": 0.47042831778526306,
      "learning_rate": 9.835589747068077e-06,
      "loss": 0.4367,
      "step": 6552
    },
    {
      "epoch": 0.3007204809324951,
      "grad_norm": 0.4628260135650635,
      "learning_rate": 9.835527383036384e-06,
      "loss": 0.4198,
      "step": 6553
    },
    {
      "epoch": 0.30076637143774954,
      "grad_norm": 0.46939021348953247,
      "learning_rate": 9.835465007376776e-06,
      "loss": 0.3787,
      "step": 6554
    },
    {
      "epoch": 0.300812261943004,
      "grad_norm": 0.49608689546585083,
      "learning_rate": 9.835402620089401e-06,
      "loss": 0.4822,
      "step": 6555
    },
    {
      "epoch": 0.3008581524482585,
      "grad_norm": 0.47997990250587463,
      "learning_rate": 9.835340221174409e-06,
      "loss": 0.4409,
      "step": 6556
    },
    {
      "epoch": 0.3009040429535129,
      "grad_norm": 0.4424269497394562,
      "learning_rate": 9.83527781063195e-06,
      "loss": 0.3396,
      "step": 6557
    },
    {
      "epoch": 0.30094993345876736,
      "grad_norm": 0.4835622012615204,
      "learning_rate": 9.835215388462172e-06,
      "loss": 0.4149,
      "step": 6558
    },
    {
      "epoch": 0.30099582396402186,
      "grad_norm": 0.44946855306625366,
      "learning_rate": 9.835152954665228e-06,
      "loss": 0.3803,
      "step": 6559
    },
    {
      "epoch": 0.3010417144692763,
      "grad_norm": 0.4765688180923462,
      "learning_rate": 9.835090509241267e-06,
      "loss": 0.4577,
      "step": 6560
    },
    {
      "epoch": 0.30108760497453074,
      "grad_norm": 0.4615131616592407,
      "learning_rate": 9.83502805219044e-06,
      "loss": 0.4384,
      "step": 6561
    },
    {
      "epoch": 0.30113349547978524,
      "grad_norm": 0.428415447473526,
      "learning_rate": 9.834965583512895e-06,
      "loss": 0.4073,
      "step": 6562
    },
    {
      "epoch": 0.3011793859850397,
      "grad_norm": 0.4402190148830414,
      "learning_rate": 9.834903103208786e-06,
      "loss": 0.376,
      "step": 6563
    },
    {
      "epoch": 0.3012252764902942,
      "grad_norm": 0.43763837218284607,
      "learning_rate": 9.834840611278259e-06,
      "loss": 0.3386,
      "step": 6564
    },
    {
      "epoch": 0.3012711669955486,
      "grad_norm": 0.49497467279434204,
      "learning_rate": 9.834778107721469e-06,
      "loss": 0.4624,
      "step": 6565
    },
    {
      "epoch": 0.30131705750080306,
      "grad_norm": 0.4978835880756378,
      "learning_rate": 9.83471559253856e-06,
      "loss": 0.4564,
      "step": 6566
    },
    {
      "epoch": 0.30136294800605756,
      "grad_norm": 0.43127644062042236,
      "learning_rate": 9.834653065729686e-06,
      "loss": 0.3442,
      "step": 6567
    },
    {
      "epoch": 0.301408838511312,
      "grad_norm": 0.4543454945087433,
      "learning_rate": 9.834590527295001e-06,
      "loss": 0.3691,
      "step": 6568
    },
    {
      "epoch": 0.30145472901656645,
      "grad_norm": 0.4683040678501129,
      "learning_rate": 9.834527977234648e-06,
      "loss": 0.458,
      "step": 6569
    },
    {
      "epoch": 0.30150061952182095,
      "grad_norm": 0.46243152022361755,
      "learning_rate": 9.834465415548782e-06,
      "loss": 0.342,
      "step": 6570
    },
    {
      "epoch": 0.3015465100270754,
      "grad_norm": 0.5035633444786072,
      "learning_rate": 9.834402842237554e-06,
      "loss": 0.3959,
      "step": 6571
    },
    {
      "epoch": 0.3015924005323299,
      "grad_norm": 0.48306867480278015,
      "learning_rate": 9.83434025730111e-06,
      "loss": 0.454,
      "step": 6572
    },
    {
      "epoch": 0.3016382910375843,
      "grad_norm": 0.5091403722763062,
      "learning_rate": 9.834277660739606e-06,
      "loss": 0.42,
      "step": 6573
    },
    {
      "epoch": 0.30168418154283877,
      "grad_norm": 0.5102655291557312,
      "learning_rate": 9.83421505255319e-06,
      "loss": 0.3819,
      "step": 6574
    },
    {
      "epoch": 0.30173007204809327,
      "grad_norm": 0.4413202106952667,
      "learning_rate": 9.83415243274201e-06,
      "loss": 0.3708,
      "step": 6575
    },
    {
      "epoch": 0.3017759625533477,
      "grad_norm": 0.5381954908370972,
      "learning_rate": 9.834089801306222e-06,
      "loss": 0.5789,
      "step": 6576
    },
    {
      "epoch": 0.30182185305860215,
      "grad_norm": 0.4343145489692688,
      "learning_rate": 9.834027158245973e-06,
      "loss": 0.3496,
      "step": 6577
    },
    {
      "epoch": 0.30186774356385665,
      "grad_norm": 0.4530371427536011,
      "learning_rate": 9.833964503561413e-06,
      "loss": 0.4166,
      "step": 6578
    },
    {
      "epoch": 0.3019136340691111,
      "grad_norm": 0.4598515033721924,
      "learning_rate": 9.833901837252695e-06,
      "loss": 0.3957,
      "step": 6579
    },
    {
      "epoch": 0.3019595245743656,
      "grad_norm": 0.44292208552360535,
      "learning_rate": 9.83383915931997e-06,
      "loss": 0.3571,
      "step": 6580
    },
    {
      "epoch": 0.30200541507962003,
      "grad_norm": 0.4871571958065033,
      "learning_rate": 9.833776469763385e-06,
      "loss": 0.3955,
      "step": 6581
    },
    {
      "epoch": 0.3020513055848745,
      "grad_norm": 0.44335129857063293,
      "learning_rate": 9.833713768583093e-06,
      "loss": 0.3983,
      "step": 6582
    },
    {
      "epoch": 0.30209719609012897,
      "grad_norm": 0.4381604492664337,
      "learning_rate": 9.833651055779247e-06,
      "loss": 0.3204,
      "step": 6583
    },
    {
      "epoch": 0.3021430865953834,
      "grad_norm": 0.46537187695503235,
      "learning_rate": 9.833588331351995e-06,
      "loss": 0.37,
      "step": 6584
    },
    {
      "epoch": 0.30218897710063786,
      "grad_norm": 0.4677359163761139,
      "learning_rate": 9.833525595301488e-06,
      "loss": 0.3702,
      "step": 6585
    },
    {
      "epoch": 0.30223486760589235,
      "grad_norm": 0.4967990219593048,
      "learning_rate": 9.833462847627877e-06,
      "loss": 0.4279,
      "step": 6586
    },
    {
      "epoch": 0.3022807581111468,
      "grad_norm": 0.4469231963157654,
      "learning_rate": 9.833400088331314e-06,
      "loss": 0.4083,
      "step": 6587
    },
    {
      "epoch": 0.30232664861640124,
      "grad_norm": 0.44363659620285034,
      "learning_rate": 9.83333731741195e-06,
      "loss": 0.3688,
      "step": 6588
    },
    {
      "epoch": 0.30237253912165574,
      "grad_norm": 0.4754067361354828,
      "learning_rate": 9.833274534869934e-06,
      "loss": 0.4374,
      "step": 6589
    },
    {
      "epoch": 0.3024184296269102,
      "grad_norm": 0.44701388478279114,
      "learning_rate": 9.83321174070542e-06,
      "loss": 0.3807,
      "step": 6590
    },
    {
      "epoch": 0.3024643201321647,
      "grad_norm": 0.4392775893211365,
      "learning_rate": 9.833148934918555e-06,
      "loss": 0.3418,
      "step": 6591
    },
    {
      "epoch": 0.3025102106374191,
      "grad_norm": 0.47023794054985046,
      "learning_rate": 9.833086117509493e-06,
      "loss": 0.3818,
      "step": 6592
    },
    {
      "epoch": 0.30255610114267356,
      "grad_norm": 0.46759796142578125,
      "learning_rate": 9.833023288478385e-06,
      "loss": 0.3714,
      "step": 6593
    },
    {
      "epoch": 0.30260199164792806,
      "grad_norm": 0.4507737159729004,
      "learning_rate": 9.832960447825382e-06,
      "loss": 0.4002,
      "step": 6594
    },
    {
      "epoch": 0.3026478821531825,
      "grad_norm": 0.47006985545158386,
      "learning_rate": 9.832897595550633e-06,
      "loss": 0.4088,
      "step": 6595
    },
    {
      "epoch": 0.30269377265843694,
      "grad_norm": 0.46786174178123474,
      "learning_rate": 9.83283473165429e-06,
      "loss": 0.4398,
      "step": 6596
    },
    {
      "epoch": 0.30273966316369144,
      "grad_norm": 0.4522979259490967,
      "learning_rate": 9.832771856136506e-06,
      "loss": 0.3766,
      "step": 6597
    },
    {
      "epoch": 0.3027855536689459,
      "grad_norm": 0.5230292677879333,
      "learning_rate": 9.832708968997432e-06,
      "loss": 0.4698,
      "step": 6598
    },
    {
      "epoch": 0.3028314441742004,
      "grad_norm": 0.44311752915382385,
      "learning_rate": 9.832646070237214e-06,
      "loss": 0.3761,
      "step": 6599
    },
    {
      "epoch": 0.3028773346794548,
      "grad_norm": 0.524550199508667,
      "learning_rate": 9.832583159856012e-06,
      "loss": 0.4986,
      "step": 6600
    },
    {
      "epoch": 0.30292322518470927,
      "grad_norm": 0.42996567487716675,
      "learning_rate": 9.832520237853971e-06,
      "loss": 0.3615,
      "step": 6601
    },
    {
      "epoch": 0.30296911568996376,
      "grad_norm": 0.8597694039344788,
      "learning_rate": 9.832457304231243e-06,
      "loss": 0.3857,
      "step": 6602
    },
    {
      "epoch": 0.3030150061952182,
      "grad_norm": 0.47751420736312866,
      "learning_rate": 9.832394358987982e-06,
      "loss": 0.4148,
      "step": 6603
    },
    {
      "epoch": 0.30306089670047265,
      "grad_norm": 0.47136661410331726,
      "learning_rate": 9.832331402124337e-06,
      "loss": 0.3816,
      "step": 6604
    },
    {
      "epoch": 0.30310678720572715,
      "grad_norm": 0.4670959413051605,
      "learning_rate": 9.83226843364046e-06,
      "loss": 0.3492,
      "step": 6605
    },
    {
      "epoch": 0.3031526777109816,
      "grad_norm": 0.42425355315208435,
      "learning_rate": 9.832205453536501e-06,
      "loss": 0.3333,
      "step": 6606
    },
    {
      "epoch": 0.3031985682162361,
      "grad_norm": 0.46095117926597595,
      "learning_rate": 9.832142461812615e-06,
      "loss": 0.3674,
      "step": 6607
    },
    {
      "epoch": 0.30324445872149053,
      "grad_norm": 0.4839651584625244,
      "learning_rate": 9.832079458468952e-06,
      "loss": 0.4645,
      "step": 6608
    },
    {
      "epoch": 0.30329034922674497,
      "grad_norm": 0.4896933138370514,
      "learning_rate": 9.83201644350566e-06,
      "loss": 0.4454,
      "step": 6609
    },
    {
      "epoch": 0.30333623973199947,
      "grad_norm": 0.44667983055114746,
      "learning_rate": 9.831953416922896e-06,
      "loss": 0.426,
      "step": 6610
    },
    {
      "epoch": 0.3033821302372539,
      "grad_norm": 0.49069464206695557,
      "learning_rate": 9.83189037872081e-06,
      "loss": 0.5117,
      "step": 6611
    },
    {
      "epoch": 0.30342802074250835,
      "grad_norm": 0.46525976061820984,
      "learning_rate": 9.83182732889955e-06,
      "loss": 0.402,
      "step": 6612
    },
    {
      "epoch": 0.30347391124776285,
      "grad_norm": 0.47777456045150757,
      "learning_rate": 9.831764267459272e-06,
      "loss": 0.4385,
      "step": 6613
    },
    {
      "epoch": 0.3035198017530173,
      "grad_norm": 0.4671141505241394,
      "learning_rate": 9.831701194400125e-06,
      "loss": 0.364,
      "step": 6614
    },
    {
      "epoch": 0.3035656922582718,
      "grad_norm": 0.4922887086868286,
      "learning_rate": 9.831638109722261e-06,
      "loss": 0.4119,
      "step": 6615
    },
    {
      "epoch": 0.30361158276352623,
      "grad_norm": 0.43810832500457764,
      "learning_rate": 9.831575013425834e-06,
      "loss": 0.3676,
      "step": 6616
    },
    {
      "epoch": 0.3036574732687807,
      "grad_norm": 0.48343196511268616,
      "learning_rate": 9.831511905510993e-06,
      "loss": 0.4167,
      "step": 6617
    },
    {
      "epoch": 0.3037033637740352,
      "grad_norm": 0.4384100139141083,
      "learning_rate": 9.831448785977891e-06,
      "loss": 0.3578,
      "step": 6618
    },
    {
      "epoch": 0.3037492542792896,
      "grad_norm": 0.4542798101902008,
      "learning_rate": 9.83138565482668e-06,
      "loss": 0.3735,
      "step": 6619
    },
    {
      "epoch": 0.30379514478454406,
      "grad_norm": 0.5416396260261536,
      "learning_rate": 9.831322512057509e-06,
      "loss": 0.5325,
      "step": 6620
    },
    {
      "epoch": 0.30384103528979856,
      "grad_norm": 0.45069217681884766,
      "learning_rate": 9.831259357670536e-06,
      "loss": 0.3483,
      "step": 6621
    },
    {
      "epoch": 0.303886925795053,
      "grad_norm": 0.4554518163204193,
      "learning_rate": 9.831196191665905e-06,
      "loss": 0.4149,
      "step": 6622
    },
    {
      "epoch": 0.30393281630030744,
      "grad_norm": 0.4834083616733551,
      "learning_rate": 9.831133014043774e-06,
      "loss": 0.3852,
      "step": 6623
    },
    {
      "epoch": 0.30397870680556194,
      "grad_norm": 0.5047228336334229,
      "learning_rate": 9.831069824804292e-06,
      "loss": 0.4666,
      "step": 6624
    },
    {
      "epoch": 0.3040245973108164,
      "grad_norm": 0.4621022641658783,
      "learning_rate": 9.831006623947613e-06,
      "loss": 0.3605,
      "step": 6625
    },
    {
      "epoch": 0.3040704878160709,
      "grad_norm": 0.46323683857917786,
      "learning_rate": 9.830943411473888e-06,
      "loss": 0.4444,
      "step": 6626
    },
    {
      "epoch": 0.3041163783213253,
      "grad_norm": 0.4874735176563263,
      "learning_rate": 9.830880187383268e-06,
      "loss": 0.5466,
      "step": 6627
    },
    {
      "epoch": 0.30416226882657976,
      "grad_norm": 0.4583863317966461,
      "learning_rate": 9.830816951675904e-06,
      "loss": 0.359,
      "step": 6628
    },
    {
      "epoch": 0.30420815933183426,
      "grad_norm": 0.45843398571014404,
      "learning_rate": 9.830753704351952e-06,
      "loss": 0.3759,
      "step": 6629
    },
    {
      "epoch": 0.3042540498370887,
      "grad_norm": 0.4736770689487457,
      "learning_rate": 9.830690445411562e-06,
      "loss": 0.3644,
      "step": 6630
    },
    {
      "epoch": 0.30429994034234314,
      "grad_norm": 0.46205586194992065,
      "learning_rate": 9.830627174854886e-06,
      "loss": 0.3539,
      "step": 6631
    },
    {
      "epoch": 0.30434583084759764,
      "grad_norm": 0.4209558367729187,
      "learning_rate": 9.830563892682076e-06,
      "loss": 0.3371,
      "step": 6632
    },
    {
      "epoch": 0.3043917213528521,
      "grad_norm": 0.5072474479675293,
      "learning_rate": 9.830500598893285e-06,
      "loss": 0.4838,
      "step": 6633
    },
    {
      "epoch": 0.3044376118581066,
      "grad_norm": 0.49206194281578064,
      "learning_rate": 9.830437293488662e-06,
      "loss": 0.4432,
      "step": 6634
    },
    {
      "epoch": 0.304483502363361,
      "grad_norm": 0.5380978584289551,
      "learning_rate": 9.830373976468365e-06,
      "loss": 0.4835,
      "step": 6635
    },
    {
      "epoch": 0.30452939286861547,
      "grad_norm": 0.4803902506828308,
      "learning_rate": 9.830310647832542e-06,
      "loss": 0.4243,
      "step": 6636
    },
    {
      "epoch": 0.30457528337386996,
      "grad_norm": 0.47719717025756836,
      "learning_rate": 9.830247307581346e-06,
      "loss": 0.397,
      "step": 6637
    },
    {
      "epoch": 0.3046211738791244,
      "grad_norm": 0.458240807056427,
      "learning_rate": 9.83018395571493e-06,
      "loss": 0.3738,
      "step": 6638
    },
    {
      "epoch": 0.30466706438437885,
      "grad_norm": 0.48805272579193115,
      "learning_rate": 9.830120592233445e-06,
      "loss": 0.3818,
      "step": 6639
    },
    {
      "epoch": 0.30471295488963335,
      "grad_norm": 0.4603116512298584,
      "learning_rate": 9.830057217137045e-06,
      "loss": 0.3588,
      "step": 6640
    },
    {
      "epoch": 0.3047588453948878,
      "grad_norm": 0.51239013671875,
      "learning_rate": 9.829993830425884e-06,
      "loss": 0.4586,
      "step": 6641
    },
    {
      "epoch": 0.3048047359001423,
      "grad_norm": 0.46400687098503113,
      "learning_rate": 9.82993043210011e-06,
      "loss": 0.3897,
      "step": 6642
    },
    {
      "epoch": 0.30485062640539673,
      "grad_norm": 0.5125425457954407,
      "learning_rate": 9.829867022159878e-06,
      "loss": 0.4143,
      "step": 6643
    },
    {
      "epoch": 0.30489651691065117,
      "grad_norm": 0.4582475423812866,
      "learning_rate": 9.82980360060534e-06,
      "loss": 0.4024,
      "step": 6644
    },
    {
      "epoch": 0.30494240741590567,
      "grad_norm": 0.4980606436729431,
      "learning_rate": 9.82974016743665e-06,
      "loss": 0.433,
      "step": 6645
    },
    {
      "epoch": 0.3049882979211601,
      "grad_norm": 0.478557825088501,
      "learning_rate": 9.829676722653958e-06,
      "loss": 0.456,
      "step": 6646
    },
    {
      "epoch": 0.30503418842641455,
      "grad_norm": 0.48435866832733154,
      "learning_rate": 9.829613266257418e-06,
      "loss": 0.4389,
      "step": 6647
    },
    {
      "epoch": 0.30508007893166905,
      "grad_norm": 0.46086686849594116,
      "learning_rate": 9.829549798247182e-06,
      "loss": 0.4359,
      "step": 6648
    },
    {
      "epoch": 0.3051259694369235,
      "grad_norm": 0.4501870572566986,
      "learning_rate": 9.829486318623403e-06,
      "loss": 0.3573,
      "step": 6649
    },
    {
      "epoch": 0.30517185994217794,
      "grad_norm": 0.4945095181465149,
      "learning_rate": 9.829422827386233e-06,
      "loss": 0.511,
      "step": 6650
    },
    {
      "epoch": 0.30521775044743243,
      "grad_norm": 0.5074198842048645,
      "learning_rate": 9.829359324535828e-06,
      "loss": 0.3607,
      "step": 6651
    },
    {
      "epoch": 0.3052636409526869,
      "grad_norm": 0.48221343755722046,
      "learning_rate": 9.829295810072335e-06,
      "loss": 0.413,
      "step": 6652
    },
    {
      "epoch": 0.3053095314579414,
      "grad_norm": 0.4983444809913635,
      "learning_rate": 9.82923228399591e-06,
      "loss": 0.4966,
      "step": 6653
    },
    {
      "epoch": 0.3053554219631958,
      "grad_norm": 0.49983564019203186,
      "learning_rate": 9.829168746306708e-06,
      "loss": 0.5255,
      "step": 6654
    },
    {
      "epoch": 0.30540131246845026,
      "grad_norm": 0.43073901534080505,
      "learning_rate": 9.829105197004879e-06,
      "loss": 0.3646,
      "step": 6655
    },
    {
      "epoch": 0.30544720297370476,
      "grad_norm": 0.4418764114379883,
      "learning_rate": 9.829041636090573e-06,
      "loss": 0.3333,
      "step": 6656
    },
    {
      "epoch": 0.3054930934789592,
      "grad_norm": 0.4820026159286499,
      "learning_rate": 9.828978063563948e-06,
      "loss": 0.3755,
      "step": 6657
    },
    {
      "epoch": 0.30553898398421364,
      "grad_norm": 0.5043435096740723,
      "learning_rate": 9.828914479425156e-06,
      "loss": 0.4748,
      "step": 6658
    },
    {
      "epoch": 0.30558487448946814,
      "grad_norm": 0.4486979842185974,
      "learning_rate": 9.828850883674346e-06,
      "loss": 0.4015,
      "step": 6659
    },
    {
      "epoch": 0.3056307649947226,
      "grad_norm": 0.5034217238426208,
      "learning_rate": 9.828787276311676e-06,
      "loss": 0.4817,
      "step": 6660
    },
    {
      "epoch": 0.3056766554999771,
      "grad_norm": 0.45763203501701355,
      "learning_rate": 9.828723657337295e-06,
      "loss": 0.3262,
      "step": 6661
    },
    {
      "epoch": 0.3057225460052315,
      "grad_norm": 0.5401304364204407,
      "learning_rate": 9.828660026751357e-06,
      "loss": 0.4488,
      "step": 6662
    },
    {
      "epoch": 0.30576843651048596,
      "grad_norm": 0.4717482924461365,
      "learning_rate": 9.828596384554019e-06,
      "loss": 0.4169,
      "step": 6663
    },
    {
      "epoch": 0.30581432701574046,
      "grad_norm": 0.4565146267414093,
      "learning_rate": 9.828532730745429e-06,
      "loss": 0.417,
      "step": 6664
    },
    {
      "epoch": 0.3058602175209949,
      "grad_norm": 0.4420337975025177,
      "learning_rate": 9.82846906532574e-06,
      "loss": 0.3572,
      "step": 6665
    },
    {
      "epoch": 0.30590610802624935,
      "grad_norm": 0.4823428988456726,
      "learning_rate": 9.828405388295108e-06,
      "loss": 0.4379,
      "step": 6666
    },
    {
      "epoch": 0.30595199853150384,
      "grad_norm": 0.4770444631576538,
      "learning_rate": 9.828341699653684e-06,
      "loss": 0.4272,
      "step": 6667
    },
    {
      "epoch": 0.3059978890367583,
      "grad_norm": 0.4345322251319885,
      "learning_rate": 9.828277999401624e-06,
      "loss": 0.3315,
      "step": 6668
    },
    {
      "epoch": 0.3060437795420128,
      "grad_norm": 0.47918233275413513,
      "learning_rate": 9.828214287539078e-06,
      "loss": 0.4397,
      "step": 6669
    },
    {
      "epoch": 0.3060896700472672,
      "grad_norm": 0.4709562063217163,
      "learning_rate": 9.8281505640662e-06,
      "loss": 0.4066,
      "step": 6670
    },
    {
      "epoch": 0.30613556055252167,
      "grad_norm": 0.4455477297306061,
      "learning_rate": 9.828086828983146e-06,
      "loss": 0.3195,
      "step": 6671
    },
    {
      "epoch": 0.30618145105777617,
      "grad_norm": 0.4749097526073456,
      "learning_rate": 9.828023082290065e-06,
      "loss": 0.4371,
      "step": 6672
    },
    {
      "epoch": 0.3062273415630306,
      "grad_norm": 0.4528983533382416,
      "learning_rate": 9.827959323987113e-06,
      "loss": 0.3659,
      "step": 6673
    },
    {
      "epoch": 0.30627323206828505,
      "grad_norm": 0.4834461808204651,
      "learning_rate": 9.827895554074443e-06,
      "loss": 0.4083,
      "step": 6674
    },
    {
      "epoch": 0.30631912257353955,
      "grad_norm": 0.44200068712234497,
      "learning_rate": 9.827831772552206e-06,
      "loss": 0.3884,
      "step": 6675
    },
    {
      "epoch": 0.306365013078794,
      "grad_norm": 0.48811590671539307,
      "learning_rate": 9.827767979420559e-06,
      "loss": 0.443,
      "step": 6676
    },
    {
      "epoch": 0.30641090358404843,
      "grad_norm": 0.4491792917251587,
      "learning_rate": 9.827704174679653e-06,
      "loss": 0.3705,
      "step": 6677
    },
    {
      "epoch": 0.30645679408930293,
      "grad_norm": 0.47117140889167786,
      "learning_rate": 9.827640358329641e-06,
      "loss": 0.4032,
      "step": 6678
    },
    {
      "epoch": 0.3065026845945574,
      "grad_norm": 0.4772627055644989,
      "learning_rate": 9.82757653037068e-06,
      "loss": 0.4043,
      "step": 6679
    },
    {
      "epoch": 0.30654857509981187,
      "grad_norm": 0.5154597163200378,
      "learning_rate": 9.82751269080292e-06,
      "loss": 0.495,
      "step": 6680
    },
    {
      "epoch": 0.3065944656050663,
      "grad_norm": 0.4699918329715729,
      "learning_rate": 9.827448839626516e-06,
      "loss": 0.3969,
      "step": 6681
    },
    {
      "epoch": 0.30664035611032076,
      "grad_norm": 0.5239112377166748,
      "learning_rate": 9.82738497684162e-06,
      "loss": 0.4475,
      "step": 6682
    },
    {
      "epoch": 0.30668624661557525,
      "grad_norm": 0.5227832198143005,
      "learning_rate": 9.827321102448387e-06,
      "loss": 0.5436,
      "step": 6683
    },
    {
      "epoch": 0.3067321371208297,
      "grad_norm": 0.4827341139316559,
      "learning_rate": 9.82725721644697e-06,
      "loss": 0.4618,
      "step": 6684
    },
    {
      "epoch": 0.30677802762608414,
      "grad_norm": 0.4586300551891327,
      "learning_rate": 9.827193318837525e-06,
      "loss": 0.3699,
      "step": 6685
    },
    {
      "epoch": 0.30682391813133864,
      "grad_norm": 0.4359295070171356,
      "learning_rate": 9.827129409620202e-06,
      "loss": 0.3373,
      "step": 6686
    },
    {
      "epoch": 0.3068698086365931,
      "grad_norm": 0.5202856659889221,
      "learning_rate": 9.827065488795155e-06,
      "loss": 0.4576,
      "step": 6687
    },
    {
      "epoch": 0.3069156991418476,
      "grad_norm": 0.4542224705219269,
      "learning_rate": 9.827001556362541e-06,
      "loss": 0.3763,
      "step": 6688
    },
    {
      "epoch": 0.306961589647102,
      "grad_norm": 0.5091878771781921,
      "learning_rate": 9.826937612322512e-06,
      "loss": 0.4471,
      "step": 6689
    },
    {
      "epoch": 0.30700748015235646,
      "grad_norm": 0.48083943128585815,
      "learning_rate": 9.82687365667522e-06,
      "loss": 0.4775,
      "step": 6690
    },
    {
      "epoch": 0.30705337065761096,
      "grad_norm": 0.4575759470462799,
      "learning_rate": 9.82680968942082e-06,
      "loss": 0.4145,
      "step": 6691
    },
    {
      "epoch": 0.3070992611628654,
      "grad_norm": 0.44299009442329407,
      "learning_rate": 9.826745710559468e-06,
      "loss": 0.4012,
      "step": 6692
    },
    {
      "epoch": 0.30714515166811984,
      "grad_norm": 0.5044476389884949,
      "learning_rate": 9.826681720091316e-06,
      "loss": 0.4123,
      "step": 6693
    },
    {
      "epoch": 0.30719104217337434,
      "grad_norm": 0.44683578610420227,
      "learning_rate": 9.826617718016516e-06,
      "loss": 0.3326,
      "step": 6694
    },
    {
      "epoch": 0.3072369326786288,
      "grad_norm": 0.4374343156814575,
      "learning_rate": 9.826553704335224e-06,
      "loss": 0.3638,
      "step": 6695
    },
    {
      "epoch": 0.3072828231838833,
      "grad_norm": 0.47219011187553406,
      "learning_rate": 9.826489679047594e-06,
      "loss": 0.3738,
      "step": 6696
    },
    {
      "epoch": 0.3073287136891377,
      "grad_norm": 0.4438578486442566,
      "learning_rate": 9.82642564215378e-06,
      "loss": 0.391,
      "step": 6697
    },
    {
      "epoch": 0.30737460419439216,
      "grad_norm": 0.49390602111816406,
      "learning_rate": 9.826361593653935e-06,
      "loss": 0.4326,
      "step": 6698
    },
    {
      "epoch": 0.30742049469964666,
      "grad_norm": 0.5040885806083679,
      "learning_rate": 9.826297533548213e-06,
      "loss": 0.462,
      "step": 6699
    },
    {
      "epoch": 0.3074663852049011,
      "grad_norm": 0.4489750266075134,
      "learning_rate": 9.826233461836769e-06,
      "loss": 0.3422,
      "step": 6700
    },
    {
      "epoch": 0.30751227571015555,
      "grad_norm": 0.47171011567115784,
      "learning_rate": 9.826169378519759e-06,
      "loss": 0.3885,
      "step": 6701
    },
    {
      "epoch": 0.30755816621541004,
      "grad_norm": 0.5421554446220398,
      "learning_rate": 9.826105283597331e-06,
      "loss": 0.3882,
      "step": 6702
    },
    {
      "epoch": 0.3076040567206645,
      "grad_norm": 0.4394133985042572,
      "learning_rate": 9.826041177069645e-06,
      "loss": 0.3487,
      "step": 6703
    },
    {
      "epoch": 0.307649947225919,
      "grad_norm": 0.45673462748527527,
      "learning_rate": 9.825977058936855e-06,
      "loss": 0.4245,
      "step": 6704
    },
    {
      "epoch": 0.3076958377311734,
      "grad_norm": 0.4650372862815857,
      "learning_rate": 9.82591292919911e-06,
      "loss": 0.3721,
      "step": 6705
    },
    {
      "epoch": 0.30774172823642787,
      "grad_norm": 0.44464802742004395,
      "learning_rate": 9.825848787856569e-06,
      "loss": 0.3382,
      "step": 6706
    },
    {
      "epoch": 0.30778761874168237,
      "grad_norm": 0.4994361102581024,
      "learning_rate": 9.825784634909385e-06,
      "loss": 0.4831,
      "step": 6707
    },
    {
      "epoch": 0.3078335092469368,
      "grad_norm": 0.4889013171195984,
      "learning_rate": 9.825720470357712e-06,
      "loss": 0.4627,
      "step": 6708
    },
    {
      "epoch": 0.30787939975219125,
      "grad_norm": 0.4376967251300812,
      "learning_rate": 9.825656294201705e-06,
      "loss": 0.3755,
      "step": 6709
    },
    {
      "epoch": 0.30792529025744575,
      "grad_norm": 0.43729689717292786,
      "learning_rate": 9.825592106441515e-06,
      "loss": 0.3339,
      "step": 6710
    },
    {
      "epoch": 0.3079711807627002,
      "grad_norm": 0.47534286975860596,
      "learning_rate": 9.825527907077302e-06,
      "loss": 0.3803,
      "step": 6711
    },
    {
      "epoch": 0.30801707126795463,
      "grad_norm": 0.5077545046806335,
      "learning_rate": 9.825463696109215e-06,
      "loss": 0.5131,
      "step": 6712
    },
    {
      "epoch": 0.30806296177320913,
      "grad_norm": 0.42200082540512085,
      "learning_rate": 9.825399473537411e-06,
      "loss": 0.3106,
      "step": 6713
    },
    {
      "epoch": 0.3081088522784636,
      "grad_norm": 0.4354619085788727,
      "learning_rate": 9.825335239362044e-06,
      "loss": 0.3895,
      "step": 6714
    },
    {
      "epoch": 0.30815474278371807,
      "grad_norm": 0.4706278145313263,
      "learning_rate": 9.825270993583271e-06,
      "loss": 0.4428,
      "step": 6715
    },
    {
      "epoch": 0.3082006332889725,
      "grad_norm": 0.44295641779899597,
      "learning_rate": 9.825206736201243e-06,
      "loss": 0.3724,
      "step": 6716
    },
    {
      "epoch": 0.30824652379422696,
      "grad_norm": 0.48312485218048096,
      "learning_rate": 9.825142467216115e-06,
      "loss": 0.4525,
      "step": 6717
    },
    {
      "epoch": 0.30829241429948145,
      "grad_norm": 0.4753396511077881,
      "learning_rate": 9.825078186628043e-06,
      "loss": 0.4604,
      "step": 6718
    },
    {
      "epoch": 0.3083383048047359,
      "grad_norm": 0.5158783197402954,
      "learning_rate": 9.82501389443718e-06,
      "loss": 0.5569,
      "step": 6719
    },
    {
      "epoch": 0.30838419530999034,
      "grad_norm": 0.5148168206214905,
      "learning_rate": 9.824949590643683e-06,
      "loss": 0.574,
      "step": 6720
    },
    {
      "epoch": 0.30843008581524484,
      "grad_norm": 0.44234567880630493,
      "learning_rate": 9.824885275247702e-06,
      "loss": 0.4117,
      "step": 6721
    },
    {
      "epoch": 0.3084759763204993,
      "grad_norm": 0.4514029026031494,
      "learning_rate": 9.824820948249398e-06,
      "loss": 0.354,
      "step": 6722
    },
    {
      "epoch": 0.3085218668257538,
      "grad_norm": 0.520056962966919,
      "learning_rate": 9.82475660964892e-06,
      "loss": 0.437,
      "step": 6723
    },
    {
      "epoch": 0.3085677573310082,
      "grad_norm": 0.4706914722919464,
      "learning_rate": 9.824692259446426e-06,
      "loss": 0.3694,
      "step": 6724
    },
    {
      "epoch": 0.30861364783626266,
      "grad_norm": 0.4695865511894226,
      "learning_rate": 9.82462789764207e-06,
      "loss": 0.4361,
      "step": 6725
    },
    {
      "epoch": 0.30865953834151716,
      "grad_norm": 0.4542319178581238,
      "learning_rate": 9.824563524236006e-06,
      "loss": 0.39,
      "step": 6726
    },
    {
      "epoch": 0.3087054288467716,
      "grad_norm": 0.4493260383605957,
      "learning_rate": 9.82449913922839e-06,
      "loss": 0.3702,
      "step": 6727
    },
    {
      "epoch": 0.30875131935202604,
      "grad_norm": 0.5305804014205933,
      "learning_rate": 9.824434742619377e-06,
      "loss": 0.454,
      "step": 6728
    },
    {
      "epoch": 0.30879720985728054,
      "grad_norm": 0.47414857149124146,
      "learning_rate": 9.824370334409119e-06,
      "loss": 0.4491,
      "step": 6729
    },
    {
      "epoch": 0.308843100362535,
      "grad_norm": 0.4635559916496277,
      "learning_rate": 9.824305914597774e-06,
      "loss": 0.4394,
      "step": 6730
    },
    {
      "epoch": 0.3088889908677895,
      "grad_norm": 0.5106761455535889,
      "learning_rate": 9.824241483185497e-06,
      "loss": 0.4769,
      "step": 6731
    },
    {
      "epoch": 0.3089348813730439,
      "grad_norm": 0.48257845640182495,
      "learning_rate": 9.82417704017244e-06,
      "loss": 0.3978,
      "step": 6732
    },
    {
      "epoch": 0.30898077187829837,
      "grad_norm": 0.45212891697883606,
      "learning_rate": 9.824112585558761e-06,
      "loss": 0.3702,
      "step": 6733
    },
    {
      "epoch": 0.30902666238355286,
      "grad_norm": 0.4260169267654419,
      "learning_rate": 9.824048119344614e-06,
      "loss": 0.2992,
      "step": 6734
    },
    {
      "epoch": 0.3090725528888073,
      "grad_norm": 0.43211981654167175,
      "learning_rate": 9.823983641530153e-06,
      "loss": 0.3891,
      "step": 6735
    },
    {
      "epoch": 0.30911844339406175,
      "grad_norm": 0.4793911278247833,
      "learning_rate": 9.823919152115533e-06,
      "loss": 0.427,
      "step": 6736
    },
    {
      "epoch": 0.30916433389931625,
      "grad_norm": 0.4465513229370117,
      "learning_rate": 9.823854651100911e-06,
      "loss": 0.3903,
      "step": 6737
    },
    {
      "epoch": 0.3092102244045707,
      "grad_norm": 0.4502408802509308,
      "learning_rate": 9.823790138486443e-06,
      "loss": 0.3753,
      "step": 6738
    },
    {
      "epoch": 0.30925611490982513,
      "grad_norm": 0.5921756625175476,
      "learning_rate": 9.82372561427228e-06,
      "loss": 0.5627,
      "step": 6739
    },
    {
      "epoch": 0.30930200541507963,
      "grad_norm": 0.4430526793003082,
      "learning_rate": 9.823661078458577e-06,
      "loss": 0.3759,
      "step": 6740
    },
    {
      "epoch": 0.30934789592033407,
      "grad_norm": 0.45313969254493713,
      "learning_rate": 9.823596531045495e-06,
      "loss": 0.3884,
      "step": 6741
    },
    {
      "epoch": 0.30939378642558857,
      "grad_norm": 0.5367785692214966,
      "learning_rate": 9.823531972033184e-06,
      "loss": 0.3975,
      "step": 6742
    },
    {
      "epoch": 0.309439676930843,
      "grad_norm": 0.4326730966567993,
      "learning_rate": 9.823467401421802e-06,
      "loss": 0.3607,
      "step": 6743
    },
    {
      "epoch": 0.30948556743609745,
      "grad_norm": 0.4941154420375824,
      "learning_rate": 9.823402819211503e-06,
      "loss": 0.4385,
      "step": 6744
    },
    {
      "epoch": 0.30953145794135195,
      "grad_norm": 0.46483370661735535,
      "learning_rate": 9.823338225402442e-06,
      "loss": 0.3745,
      "step": 6745
    },
    {
      "epoch": 0.3095773484466064,
      "grad_norm": 0.4239407181739807,
      "learning_rate": 9.823273619994776e-06,
      "loss": 0.3263,
      "step": 6746
    },
    {
      "epoch": 0.30962323895186084,
      "grad_norm": 0.47472575306892395,
      "learning_rate": 9.823209002988657e-06,
      "loss": 0.4236,
      "step": 6747
    },
    {
      "epoch": 0.30966912945711533,
      "grad_norm": 0.4964427351951599,
      "learning_rate": 9.823144374384244e-06,
      "loss": 0.4787,
      "step": 6748
    },
    {
      "epoch": 0.3097150199623698,
      "grad_norm": 0.4612743556499481,
      "learning_rate": 9.82307973418169e-06,
      "loss": 0.3665,
      "step": 6749
    },
    {
      "epoch": 0.3097609104676243,
      "grad_norm": 0.46500423550605774,
      "learning_rate": 9.823015082381154e-06,
      "loss": 0.4377,
      "step": 6750
    },
    {
      "epoch": 0.3098068009728787,
      "grad_norm": 0.43587544560432434,
      "learning_rate": 9.822950418982788e-06,
      "loss": 0.344,
      "step": 6751
    },
    {
      "epoch": 0.30985269147813316,
      "grad_norm": 0.46367624402046204,
      "learning_rate": 9.822885743986747e-06,
      "loss": 0.4613,
      "step": 6752
    },
    {
      "epoch": 0.30989858198338766,
      "grad_norm": 0.44159722328186035,
      "learning_rate": 9.822821057393188e-06,
      "loss": 0.3268,
      "step": 6753
    },
    {
      "epoch": 0.3099444724886421,
      "grad_norm": 0.47241494059562683,
      "learning_rate": 9.822756359202268e-06,
      "loss": 0.3823,
      "step": 6754
    },
    {
      "epoch": 0.30999036299389654,
      "grad_norm": 0.48571261763572693,
      "learning_rate": 9.82269164941414e-06,
      "loss": 0.3809,
      "step": 6755
    },
    {
      "epoch": 0.31003625349915104,
      "grad_norm": 0.46417245268821716,
      "learning_rate": 9.82262692802896e-06,
      "loss": 0.3901,
      "step": 6756
    },
    {
      "epoch": 0.3100821440044055,
      "grad_norm": 0.47183191776275635,
      "learning_rate": 9.822562195046886e-06,
      "loss": 0.4448,
      "step": 6757
    },
    {
      "epoch": 0.31012803450966,
      "grad_norm": 0.8337662816047668,
      "learning_rate": 9.82249745046807e-06,
      "loss": 0.4436,
      "step": 6758
    },
    {
      "epoch": 0.3101739250149144,
      "grad_norm": 0.4456123113632202,
      "learning_rate": 9.822432694292671e-06,
      "loss": 0.3869,
      "step": 6759
    },
    {
      "epoch": 0.31021981552016886,
      "grad_norm": 0.516566812992096,
      "learning_rate": 9.822367926520844e-06,
      "loss": 0.5755,
      "step": 6760
    },
    {
      "epoch": 0.31026570602542336,
      "grad_norm": 0.49161389470100403,
      "learning_rate": 9.822303147152743e-06,
      "loss": 0.4532,
      "step": 6761
    },
    {
      "epoch": 0.3103115965306778,
      "grad_norm": 0.46822506189346313,
      "learning_rate": 9.822238356188525e-06,
      "loss": 0.4379,
      "step": 6762
    },
    {
      "epoch": 0.31035748703593224,
      "grad_norm": 0.4389892518520355,
      "learning_rate": 9.822173553628348e-06,
      "loss": 0.3266,
      "step": 6763
    },
    {
      "epoch": 0.31040337754118674,
      "grad_norm": 0.4525942802429199,
      "learning_rate": 9.822108739472362e-06,
      "loss": 0.4083,
      "step": 6764
    },
    {
      "epoch": 0.3104492680464412,
      "grad_norm": 0.497713565826416,
      "learning_rate": 9.822043913720726e-06,
      "loss": 0.4315,
      "step": 6765
    },
    {
      "epoch": 0.3104951585516956,
      "grad_norm": 0.43228432536125183,
      "learning_rate": 9.821979076373601e-06,
      "loss": 0.3552,
      "step": 6766
    },
    {
      "epoch": 0.3105410490569501,
      "grad_norm": 0.469587117433548,
      "learning_rate": 9.821914227431135e-06,
      "loss": 0.4346,
      "step": 6767
    },
    {
      "epoch": 0.31058693956220457,
      "grad_norm": 0.4590999484062195,
      "learning_rate": 9.821849366893485e-06,
      "loss": 0.3984,
      "step": 6768
    },
    {
      "epoch": 0.31063283006745906,
      "grad_norm": 0.4557521641254425,
      "learning_rate": 9.821784494760814e-06,
      "loss": 0.3753,
      "step": 6769
    },
    {
      "epoch": 0.3106787205727135,
      "grad_norm": 0.48752713203430176,
      "learning_rate": 9.82171961103327e-06,
      "loss": 0.4944,
      "step": 6770
    },
    {
      "epoch": 0.31072461107796795,
      "grad_norm": 0.4644518196582794,
      "learning_rate": 9.821654715711013e-06,
      "loss": 0.4506,
      "step": 6771
    },
    {
      "epoch": 0.31077050158322245,
      "grad_norm": 0.48173952102661133,
      "learning_rate": 9.821589808794196e-06,
      "loss": 0.4288,
      "step": 6772
    },
    {
      "epoch": 0.3108163920884769,
      "grad_norm": 0.4833815097808838,
      "learning_rate": 9.82152489028298e-06,
      "loss": 0.4272,
      "step": 6773
    },
    {
      "epoch": 0.31086228259373133,
      "grad_norm": 0.4809361398220062,
      "learning_rate": 9.821459960177518e-06,
      "loss": 0.4431,
      "step": 6774
    },
    {
      "epoch": 0.31090817309898583,
      "grad_norm": 0.5219428539276123,
      "learning_rate": 9.821395018477965e-06,
      "loss": 0.4928,
      "step": 6775
    },
    {
      "epoch": 0.31095406360424027,
      "grad_norm": 0.472053587436676,
      "learning_rate": 9.82133006518448e-06,
      "loss": 0.4697,
      "step": 6776
    },
    {
      "epoch": 0.31099995410949477,
      "grad_norm": 0.5082793831825256,
      "learning_rate": 9.821265100297217e-06,
      "loss": 0.4758,
      "step": 6777
    },
    {
      "epoch": 0.3110458446147492,
      "grad_norm": 0.4976414442062378,
      "learning_rate": 9.821200123816334e-06,
      "loss": 0.4297,
      "step": 6778
    },
    {
      "epoch": 0.31109173512000365,
      "grad_norm": 0.4456143081188202,
      "learning_rate": 9.821135135741986e-06,
      "loss": 0.3396,
      "step": 6779
    },
    {
      "epoch": 0.31113762562525815,
      "grad_norm": 0.5019738674163818,
      "learning_rate": 9.82107013607433e-06,
      "loss": 0.4587,
      "step": 6780
    },
    {
      "epoch": 0.3111835161305126,
      "grad_norm": 0.42361247539520264,
      "learning_rate": 9.82100512481352e-06,
      "loss": 0.3399,
      "step": 6781
    },
    {
      "epoch": 0.31122940663576704,
      "grad_norm": 0.500214159488678,
      "learning_rate": 9.820940101959715e-06,
      "loss": 0.3793,
      "step": 6782
    },
    {
      "epoch": 0.31127529714102153,
      "grad_norm": 0.4825003743171692,
      "learning_rate": 9.820875067513072e-06,
      "loss": 0.4559,
      "step": 6783
    },
    {
      "epoch": 0.311321187646276,
      "grad_norm": 0.457776814699173,
      "learning_rate": 9.820810021473745e-06,
      "loss": 0.3326,
      "step": 6784
    },
    {
      "epoch": 0.3113670781515305,
      "grad_norm": 0.45118215680122375,
      "learning_rate": 9.820744963841892e-06,
      "loss": 0.3781,
      "step": 6785
    },
    {
      "epoch": 0.3114129686567849,
      "grad_norm": 0.4419827461242676,
      "learning_rate": 9.820679894617669e-06,
      "loss": 0.3899,
      "step": 6786
    },
    {
      "epoch": 0.31145885916203936,
      "grad_norm": 0.44874539971351624,
      "learning_rate": 9.82061481380123e-06,
      "loss": 0.3887,
      "step": 6787
    },
    {
      "epoch": 0.31150474966729386,
      "grad_norm": 0.4241480231285095,
      "learning_rate": 9.820549721392736e-06,
      "loss": 0.3616,
      "step": 6788
    },
    {
      "epoch": 0.3115506401725483,
      "grad_norm": 0.4799668788909912,
      "learning_rate": 9.82048461739234e-06,
      "loss": 0.3893,
      "step": 6789
    },
    {
      "epoch": 0.31159653067780274,
      "grad_norm": 0.5291328430175781,
      "learning_rate": 9.820419501800201e-06,
      "loss": 0.4716,
      "step": 6790
    },
    {
      "epoch": 0.31164242118305724,
      "grad_norm": 0.47475743293762207,
      "learning_rate": 9.820354374616475e-06,
      "loss": 0.4743,
      "step": 6791
    },
    {
      "epoch": 0.3116883116883117,
      "grad_norm": 0.449655145406723,
      "learning_rate": 9.820289235841316e-06,
      "loss": 0.4246,
      "step": 6792
    },
    {
      "epoch": 0.3117342021935662,
      "grad_norm": 0.4789987802505493,
      "learning_rate": 9.820224085474884e-06,
      "loss": 0.47,
      "step": 6793
    },
    {
      "epoch": 0.3117800926988206,
      "grad_norm": 0.43304553627967834,
      "learning_rate": 9.820158923517335e-06,
      "loss": 0.3307,
      "step": 6794
    },
    {
      "epoch": 0.31182598320407506,
      "grad_norm": 0.4332766532897949,
      "learning_rate": 9.820093749968825e-06,
      "loss": 0.3732,
      "step": 6795
    },
    {
      "epoch": 0.31187187370932956,
      "grad_norm": 0.43455588817596436,
      "learning_rate": 9.820028564829509e-06,
      "loss": 0.3597,
      "step": 6796
    },
    {
      "epoch": 0.311917764214584,
      "grad_norm": 0.45282676815986633,
      "learning_rate": 9.819963368099548e-06,
      "loss": 0.3876,
      "step": 6797
    },
    {
      "epoch": 0.31196365471983845,
      "grad_norm": 0.5092303156852722,
      "learning_rate": 9.819898159779093e-06,
      "loss": 0.4611,
      "step": 6798
    },
    {
      "epoch": 0.31200954522509294,
      "grad_norm": 0.47585415840148926,
      "learning_rate": 9.819832939868307e-06,
      "loss": 0.413,
      "step": 6799
    },
    {
      "epoch": 0.3120554357303474,
      "grad_norm": 0.48095840215682983,
      "learning_rate": 9.819767708367342e-06,
      "loss": 0.4087,
      "step": 6800
    },
    {
      "epoch": 0.31210132623560183,
      "grad_norm": 0.42886844277381897,
      "learning_rate": 9.819702465276359e-06,
      "loss": 0.3632,
      "step": 6801
    },
    {
      "epoch": 0.3121472167408563,
      "grad_norm": 0.43350598216056824,
      "learning_rate": 9.819637210595513e-06,
      "loss": 0.411,
      "step": 6802
    },
    {
      "epoch": 0.31219310724611077,
      "grad_norm": 0.4743501543998718,
      "learning_rate": 9.819571944324958e-06,
      "loss": 0.4221,
      "step": 6803
    },
    {
      "epoch": 0.31223899775136527,
      "grad_norm": 0.4626290798187256,
      "learning_rate": 9.819506666464855e-06,
      "loss": 0.4594,
      "step": 6804
    },
    {
      "epoch": 0.3122848882566197,
      "grad_norm": 0.4297126829624176,
      "learning_rate": 9.81944137701536e-06,
      "loss": 0.3511,
      "step": 6805
    },
    {
      "epoch": 0.31233077876187415,
      "grad_norm": 0.47251227498054504,
      "learning_rate": 9.819376075976628e-06,
      "loss": 0.3901,
      "step": 6806
    },
    {
      "epoch": 0.31237666926712865,
      "grad_norm": 0.46752774715423584,
      "learning_rate": 9.81931076334882e-06,
      "loss": 0.434,
      "step": 6807
    },
    {
      "epoch": 0.3124225597723831,
      "grad_norm": 0.4704759120941162,
      "learning_rate": 9.819245439132087e-06,
      "loss": 0.3896,
      "step": 6808
    },
    {
      "epoch": 0.31246845027763753,
      "grad_norm": 0.4572508633136749,
      "learning_rate": 9.819180103326592e-06,
      "loss": 0.3852,
      "step": 6809
    },
    {
      "epoch": 0.31251434078289203,
      "grad_norm": 0.4954196810722351,
      "learning_rate": 9.819114755932491e-06,
      "loss": 0.5039,
      "step": 6810
    },
    {
      "epoch": 0.3125602312881465,
      "grad_norm": 0.447406142950058,
      "learning_rate": 9.819049396949939e-06,
      "loss": 0.3404,
      "step": 6811
    },
    {
      "epoch": 0.31260612179340097,
      "grad_norm": 0.46359962224960327,
      "learning_rate": 9.818984026379093e-06,
      "loss": 0.3905,
      "step": 6812
    },
    {
      "epoch": 0.3126520122986554,
      "grad_norm": 0.489170104265213,
      "learning_rate": 9.818918644220113e-06,
      "loss": 0.4083,
      "step": 6813
    },
    {
      "epoch": 0.31269790280390986,
      "grad_norm": 0.5148153305053711,
      "learning_rate": 9.818853250473153e-06,
      "loss": 0.4992,
      "step": 6814
    },
    {
      "epoch": 0.31274379330916435,
      "grad_norm": 0.43294501304626465,
      "learning_rate": 9.818787845138373e-06,
      "loss": 0.3482,
      "step": 6815
    },
    {
      "epoch": 0.3127896838144188,
      "grad_norm": 0.4783685803413391,
      "learning_rate": 9.818722428215927e-06,
      "loss": 0.4115,
      "step": 6816
    },
    {
      "epoch": 0.31283557431967324,
      "grad_norm": 0.45074763894081116,
      "learning_rate": 9.818656999705977e-06,
      "loss": 0.3764,
      "step": 6817
    },
    {
      "epoch": 0.31288146482492774,
      "grad_norm": 0.4706534147262573,
      "learning_rate": 9.818591559608676e-06,
      "loss": 0.4095,
      "step": 6818
    },
    {
      "epoch": 0.3129273553301822,
      "grad_norm": 0.43691709637641907,
      "learning_rate": 9.818526107924185e-06,
      "loss": 0.357,
      "step": 6819
    },
    {
      "epoch": 0.3129732458354367,
      "grad_norm": 0.4433576762676239,
      "learning_rate": 9.818460644652658e-06,
      "loss": 0.3544,
      "step": 6820
    },
    {
      "epoch": 0.3130191363406911,
      "grad_norm": 0.39241915941238403,
      "learning_rate": 9.818395169794253e-06,
      "loss": 0.2927,
      "step": 6821
    },
    {
      "epoch": 0.31306502684594556,
      "grad_norm": 0.4818485975265503,
      "learning_rate": 9.81832968334913e-06,
      "loss": 0.3995,
      "step": 6822
    },
    {
      "epoch": 0.31311091735120006,
      "grad_norm": 0.5008805394172668,
      "learning_rate": 9.818264185317445e-06,
      "loss": 0.4308,
      "step": 6823
    },
    {
      "epoch": 0.3131568078564545,
      "grad_norm": 0.5207908749580383,
      "learning_rate": 9.818198675699354e-06,
      "loss": 0.4304,
      "step": 6824
    },
    {
      "epoch": 0.31320269836170894,
      "grad_norm": 0.5156873464584351,
      "learning_rate": 9.818133154495017e-06,
      "loss": 0.5269,
      "step": 6825
    },
    {
      "epoch": 0.31324858886696344,
      "grad_norm": 0.4534202814102173,
      "learning_rate": 9.81806762170459e-06,
      "loss": 0.3903,
      "step": 6826
    },
    {
      "epoch": 0.3132944793722179,
      "grad_norm": 0.4930480122566223,
      "learning_rate": 9.81800207732823e-06,
      "loss": 0.4219,
      "step": 6827
    },
    {
      "epoch": 0.3133403698774723,
      "grad_norm": 0.4778401255607605,
      "learning_rate": 9.817936521366097e-06,
      "loss": 0.4259,
      "step": 6828
    },
    {
      "epoch": 0.3133862603827268,
      "grad_norm": 0.4639195203781128,
      "learning_rate": 9.817870953818348e-06,
      "loss": 0.4056,
      "step": 6829
    },
    {
      "epoch": 0.31343215088798126,
      "grad_norm": 0.4899890720844269,
      "learning_rate": 9.817805374685137e-06,
      "loss": 0.4101,
      "step": 6830
    },
    {
      "epoch": 0.31347804139323576,
      "grad_norm": 0.4542444944381714,
      "learning_rate": 9.817739783966627e-06,
      "loss": 0.404,
      "step": 6831
    },
    {
      "epoch": 0.3135239318984902,
      "grad_norm": 0.4820408523082733,
      "learning_rate": 9.817674181662973e-06,
      "loss": 0.429,
      "step": 6832
    },
    {
      "epoch": 0.31356982240374465,
      "grad_norm": 0.440933495759964,
      "learning_rate": 9.817608567774332e-06,
      "loss": 0.4221,
      "step": 6833
    },
    {
      "epoch": 0.31361571290899914,
      "grad_norm": 0.4212203323841095,
      "learning_rate": 9.817542942300864e-06,
      "loss": 0.3191,
      "step": 6834
    },
    {
      "epoch": 0.3136616034142536,
      "grad_norm": 0.4479084610939026,
      "learning_rate": 9.817477305242725e-06,
      "loss": 0.3881,
      "step": 6835
    },
    {
      "epoch": 0.31370749391950803,
      "grad_norm": 0.4258545935153961,
      "learning_rate": 9.817411656600073e-06,
      "loss": 0.3374,
      "step": 6836
    },
    {
      "epoch": 0.3137533844247625,
      "grad_norm": 0.4910781681537628,
      "learning_rate": 9.817345996373068e-06,
      "loss": 0.4312,
      "step": 6837
    },
    {
      "epoch": 0.31379927493001697,
      "grad_norm": 0.5013216137886047,
      "learning_rate": 9.817280324561867e-06,
      "loss": 0.5181,
      "step": 6838
    },
    {
      "epoch": 0.31384516543527147,
      "grad_norm": 0.469016432762146,
      "learning_rate": 9.817214641166624e-06,
      "loss": 0.4246,
      "step": 6839
    },
    {
      "epoch": 0.3138910559405259,
      "grad_norm": 0.43212074041366577,
      "learning_rate": 9.817148946187502e-06,
      "loss": 0.3472,
      "step": 6840
    },
    {
      "epoch": 0.31393694644578035,
      "grad_norm": 0.41999009251594543,
      "learning_rate": 9.817083239624657e-06,
      "loss": 0.3224,
      "step": 6841
    },
    {
      "epoch": 0.31398283695103485,
      "grad_norm": 0.4577331840991974,
      "learning_rate": 9.817017521478247e-06,
      "loss": 0.418,
      "step": 6842
    },
    {
      "epoch": 0.3140287274562893,
      "grad_norm": 0.46392470598220825,
      "learning_rate": 9.81695179174843e-06,
      "loss": 0.3958,
      "step": 6843
    },
    {
      "epoch": 0.31407461796154373,
      "grad_norm": 0.4637294411659241,
      "learning_rate": 9.816886050435366e-06,
      "loss": 0.4212,
      "step": 6844
    },
    {
      "epoch": 0.31412050846679823,
      "grad_norm": 0.5168417692184448,
      "learning_rate": 9.816820297539208e-06,
      "loss": 0.4553,
      "step": 6845
    },
    {
      "epoch": 0.3141663989720527,
      "grad_norm": 0.4453659653663635,
      "learning_rate": 9.81675453306012e-06,
      "loss": 0.4034,
      "step": 6846
    },
    {
      "epoch": 0.31421228947730717,
      "grad_norm": 0.47412481904029846,
      "learning_rate": 9.816688756998255e-06,
      "loss": 0.4703,
      "step": 6847
    },
    {
      "epoch": 0.3142581799825616,
      "grad_norm": 0.45725494623184204,
      "learning_rate": 9.816622969353777e-06,
      "loss": 0.3651,
      "step": 6848
    },
    {
      "epoch": 0.31430407048781606,
      "grad_norm": 0.420184463262558,
      "learning_rate": 9.81655717012684e-06,
      "loss": 0.3373,
      "step": 6849
    },
    {
      "epoch": 0.31434996099307055,
      "grad_norm": 0.46347182989120483,
      "learning_rate": 9.816491359317602e-06,
      "loss": 0.4177,
      "step": 6850
    },
    {
      "epoch": 0.314395851498325,
      "grad_norm": 0.43666452169418335,
      "learning_rate": 9.816425536926223e-06,
      "loss": 0.3647,
      "step": 6851
    },
    {
      "epoch": 0.31444174200357944,
      "grad_norm": 0.4481038451194763,
      "learning_rate": 9.81635970295286e-06,
      "loss": 0.4295,
      "step": 6852
    },
    {
      "epoch": 0.31448763250883394,
      "grad_norm": 0.4232562184333801,
      "learning_rate": 9.816293857397675e-06,
      "loss": 0.372,
      "step": 6853
    },
    {
      "epoch": 0.3145335230140884,
      "grad_norm": 0.44476720690727234,
      "learning_rate": 9.81622800026082e-06,
      "loss": 0.3777,
      "step": 6854
    },
    {
      "epoch": 0.3145794135193428,
      "grad_norm": 0.4472750723361969,
      "learning_rate": 9.816162131542457e-06,
      "loss": 0.3674,
      "step": 6855
    },
    {
      "epoch": 0.3146253040245973,
      "grad_norm": 0.508522629737854,
      "learning_rate": 9.816096251242745e-06,
      "loss": 0.4854,
      "step": 6856
    },
    {
      "epoch": 0.31467119452985176,
      "grad_norm": 0.4416707754135132,
      "learning_rate": 9.816030359361842e-06,
      "loss": 0.3662,
      "step": 6857
    },
    {
      "epoch": 0.31471708503510626,
      "grad_norm": 0.44998955726623535,
      "learning_rate": 9.815964455899904e-06,
      "loss": 0.3842,
      "step": 6858
    },
    {
      "epoch": 0.3147629755403607,
      "grad_norm": 0.4699195325374603,
      "learning_rate": 9.815898540857093e-06,
      "loss": 0.4812,
      "step": 6859
    },
    {
      "epoch": 0.31480886604561514,
      "grad_norm": 0.4703376293182373,
      "learning_rate": 9.815832614233566e-06,
      "loss": 0.4087,
      "step": 6860
    },
    {
      "epoch": 0.31485475655086964,
      "grad_norm": 0.49357494711875916,
      "learning_rate": 9.815766676029481e-06,
      "loss": 0.5339,
      "step": 6861
    },
    {
      "epoch": 0.3149006470561241,
      "grad_norm": 0.4573034942150116,
      "learning_rate": 9.815700726244997e-06,
      "loss": 0.3443,
      "step": 6862
    },
    {
      "epoch": 0.3149465375613785,
      "grad_norm": 0.495026171207428,
      "learning_rate": 9.815634764880273e-06,
      "loss": 0.474,
      "step": 6863
    },
    {
      "epoch": 0.314992428066633,
      "grad_norm": 0.45636898279190063,
      "learning_rate": 9.815568791935467e-06,
      "loss": 0.4166,
      "step": 6864
    },
    {
      "epoch": 0.31503831857188747,
      "grad_norm": 0.5281257629394531,
      "learning_rate": 9.815502807410736e-06,
      "loss": 0.5191,
      "step": 6865
    },
    {
      "epoch": 0.31508420907714196,
      "grad_norm": 0.4458218216896057,
      "learning_rate": 9.815436811306245e-06,
      "loss": 0.3546,
      "step": 6866
    },
    {
      "epoch": 0.3151300995823964,
      "grad_norm": 0.49445730447769165,
      "learning_rate": 9.815370803622145e-06,
      "loss": 0.4454,
      "step": 6867
    },
    {
      "epoch": 0.31517599008765085,
      "grad_norm": 0.46657928824424744,
      "learning_rate": 9.8153047843586e-06,
      "loss": 0.4503,
      "step": 6868
    },
    {
      "epoch": 0.31522188059290535,
      "grad_norm": 0.445287823677063,
      "learning_rate": 9.815238753515764e-06,
      "loss": 0.3806,
      "step": 6869
    },
    {
      "epoch": 0.3152677710981598,
      "grad_norm": 0.4831392168998718,
      "learning_rate": 9.8151727110938e-06,
      "loss": 0.4556,
      "step": 6870
    },
    {
      "epoch": 0.31531366160341423,
      "grad_norm": 0.46448814868927,
      "learning_rate": 9.815106657092866e-06,
      "loss": 0.3889,
      "step": 6871
    },
    {
      "epoch": 0.31535955210866873,
      "grad_norm": 0.4358893036842346,
      "learning_rate": 9.81504059151312e-06,
      "loss": 0.3642,
      "step": 6872
    },
    {
      "epoch": 0.31540544261392317,
      "grad_norm": 0.49335893988609314,
      "learning_rate": 9.81497451435472e-06,
      "loss": 0.3775,
      "step": 6873
    },
    {
      "epoch": 0.31545133311917767,
      "grad_norm": 0.47260338068008423,
      "learning_rate": 9.814908425617827e-06,
      "loss": 0.365,
      "step": 6874
    },
    {
      "epoch": 0.3154972236244321,
      "grad_norm": 0.45433372259140015,
      "learning_rate": 9.814842325302598e-06,
      "loss": 0.3685,
      "step": 6875
    },
    {
      "epoch": 0.31554311412968655,
      "grad_norm": 0.44888371229171753,
      "learning_rate": 9.814776213409194e-06,
      "loss": 0.3977,
      "step": 6876
    },
    {
      "epoch": 0.31558900463494105,
      "grad_norm": 0.451658695936203,
      "learning_rate": 9.81471008993777e-06,
      "loss": 0.3366,
      "step": 6877
    },
    {
      "epoch": 0.3156348951401955,
      "grad_norm": 0.4468802511692047,
      "learning_rate": 9.814643954888492e-06,
      "loss": 0.398,
      "step": 6878
    },
    {
      "epoch": 0.31568078564544994,
      "grad_norm": 0.503246009349823,
      "learning_rate": 9.814577808261511e-06,
      "loss": 0.4326,
      "step": 6879
    },
    {
      "epoch": 0.31572667615070443,
      "grad_norm": 0.41447919607162476,
      "learning_rate": 9.814511650056993e-06,
      "loss": 0.3594,
      "step": 6880
    },
    {
      "epoch": 0.3157725666559589,
      "grad_norm": 0.44382134079933167,
      "learning_rate": 9.814445480275093e-06,
      "loss": 0.3424,
      "step": 6881
    },
    {
      "epoch": 0.3158184571612133,
      "grad_norm": 0.45402440428733826,
      "learning_rate": 9.81437929891597e-06,
      "loss": 0.3828,
      "step": 6882
    },
    {
      "epoch": 0.3158643476664678,
      "grad_norm": 0.42428353428840637,
      "learning_rate": 9.814313105979784e-06,
      "loss": 0.3192,
      "step": 6883
    },
    {
      "epoch": 0.31591023817172226,
      "grad_norm": 0.4662676155567169,
      "learning_rate": 9.814246901466697e-06,
      "loss": 0.3875,
      "step": 6884
    },
    {
      "epoch": 0.31595612867697676,
      "grad_norm": 0.4538199007511139,
      "learning_rate": 9.814180685376864e-06,
      "loss": 0.3851,
      "step": 6885
    },
    {
      "epoch": 0.3160020191822312,
      "grad_norm": 0.4505821764469147,
      "learning_rate": 9.814114457710445e-06,
      "loss": 0.3876,
      "step": 6886
    },
    {
      "epoch": 0.31604790968748564,
      "grad_norm": 0.48920732736587524,
      "learning_rate": 9.814048218467601e-06,
      "loss": 0.4925,
      "step": 6887
    },
    {
      "epoch": 0.31609380019274014,
      "grad_norm": 0.43079760670661926,
      "learning_rate": 9.81398196764849e-06,
      "loss": 0.323,
      "step": 6888
    },
    {
      "epoch": 0.3161396906979946,
      "grad_norm": 0.4810311496257782,
      "learning_rate": 9.813915705253271e-06,
      "loss": 0.4861,
      "step": 6889
    },
    {
      "epoch": 0.316185581203249,
      "grad_norm": 0.4472326636314392,
      "learning_rate": 9.813849431282107e-06,
      "loss": 0.4089,
      "step": 6890
    },
    {
      "epoch": 0.3162314717085035,
      "grad_norm": 0.4294922351837158,
      "learning_rate": 9.813783145735152e-06,
      "loss": 0.3728,
      "step": 6891
    },
    {
      "epoch": 0.31627736221375796,
      "grad_norm": 0.5034098625183105,
      "learning_rate": 9.813716848612568e-06,
      "loss": 0.4398,
      "step": 6892
    },
    {
      "epoch": 0.31632325271901246,
      "grad_norm": 0.4683046042919159,
      "learning_rate": 9.813650539914514e-06,
      "loss": 0.4353,
      "step": 6893
    },
    {
      "epoch": 0.3163691432242669,
      "grad_norm": 0.4813278019428253,
      "learning_rate": 9.81358421964115e-06,
      "loss": 0.4366,
      "step": 6894
    },
    {
      "epoch": 0.31641503372952134,
      "grad_norm": 0.5084885954856873,
      "learning_rate": 9.813517887792637e-06,
      "loss": 0.425,
      "step": 6895
    },
    {
      "epoch": 0.31646092423477584,
      "grad_norm": 0.45638856291770935,
      "learning_rate": 9.81345154436913e-06,
      "loss": 0.3871,
      "step": 6896
    },
    {
      "epoch": 0.3165068147400303,
      "grad_norm": 0.4983183741569519,
      "learning_rate": 9.813385189370792e-06,
      "loss": 0.4615,
      "step": 6897
    },
    {
      "epoch": 0.3165527052452847,
      "grad_norm": 0.458781898021698,
      "learning_rate": 9.813318822797783e-06,
      "loss": 0.4101,
      "step": 6898
    },
    {
      "epoch": 0.3165985957505392,
      "grad_norm": 0.4356752634048462,
      "learning_rate": 9.81325244465026e-06,
      "loss": 0.3081,
      "step": 6899
    },
    {
      "epoch": 0.31664448625579367,
      "grad_norm": 0.45443177223205566,
      "learning_rate": 9.813186054928384e-06,
      "loss": 0.3498,
      "step": 6900
    },
    {
      "epoch": 0.31669037676104816,
      "grad_norm": 0.4583868086338043,
      "learning_rate": 9.813119653632314e-06,
      "loss": 0.3936,
      "step": 6901
    },
    {
      "epoch": 0.3167362672663026,
      "grad_norm": 0.44368353486061096,
      "learning_rate": 9.813053240762213e-06,
      "loss": 0.3739,
      "step": 6902
    },
    {
      "epoch": 0.31678215777155705,
      "grad_norm": 0.476655513048172,
      "learning_rate": 9.812986816318236e-06,
      "loss": 0.4399,
      "step": 6903
    },
    {
      "epoch": 0.31682804827681155,
      "grad_norm": 0.4905949831008911,
      "learning_rate": 9.812920380300545e-06,
      "loss": 0.4837,
      "step": 6904
    },
    {
      "epoch": 0.316873938782066,
      "grad_norm": 0.48626771569252014,
      "learning_rate": 9.812853932709298e-06,
      "loss": 0.4532,
      "step": 6905
    },
    {
      "epoch": 0.31691982928732043,
      "grad_norm": 0.4806963801383972,
      "learning_rate": 9.81278747354466e-06,
      "loss": 0.4484,
      "step": 6906
    },
    {
      "epoch": 0.31696571979257493,
      "grad_norm": 0.4692946672439575,
      "learning_rate": 9.812721002806783e-06,
      "loss": 0.4232,
      "step": 6907
    },
    {
      "epoch": 0.31701161029782937,
      "grad_norm": 0.5127679705619812,
      "learning_rate": 9.812654520495833e-06,
      "loss": 0.4092,
      "step": 6908
    },
    {
      "epoch": 0.31705750080308387,
      "grad_norm": 0.5175108909606934,
      "learning_rate": 9.812588026611968e-06,
      "loss": 0.4258,
      "step": 6909
    },
    {
      "epoch": 0.3171033913083383,
      "grad_norm": 0.4932761788368225,
      "learning_rate": 9.812521521155348e-06,
      "loss": 0.4263,
      "step": 6910
    },
    {
      "epoch": 0.31714928181359275,
      "grad_norm": 0.4879387319087982,
      "learning_rate": 9.812455004126131e-06,
      "loss": 0.4454,
      "step": 6911
    },
    {
      "epoch": 0.31719517231884725,
      "grad_norm": 0.46224719285964966,
      "learning_rate": 9.81238847552448e-06,
      "loss": 0.4198,
      "step": 6912
    },
    {
      "epoch": 0.3172410628241017,
      "grad_norm": 0.4714747965335846,
      "learning_rate": 9.812321935350554e-06,
      "loss": 0.3862,
      "step": 6913
    },
    {
      "epoch": 0.31728695332935614,
      "grad_norm": 0.4764077961444855,
      "learning_rate": 9.81225538360451e-06,
      "loss": 0.4279,
      "step": 6914
    },
    {
      "epoch": 0.31733284383461063,
      "grad_norm": 0.4578917622566223,
      "learning_rate": 9.812188820286513e-06,
      "loss": 0.4449,
      "step": 6915
    },
    {
      "epoch": 0.3173787343398651,
      "grad_norm": 0.39900583028793335,
      "learning_rate": 9.81212224539672e-06,
      "loss": 0.2864,
      "step": 6916
    },
    {
      "epoch": 0.3174246248451195,
      "grad_norm": 0.4593629837036133,
      "learning_rate": 9.81205565893529e-06,
      "loss": 0.4134,
      "step": 6917
    },
    {
      "epoch": 0.317470515350374,
      "grad_norm": 0.4725753366947174,
      "learning_rate": 9.811989060902387e-06,
      "loss": 0.4788,
      "step": 6918
    },
    {
      "epoch": 0.31751640585562846,
      "grad_norm": 0.47312673926353455,
      "learning_rate": 9.811922451298169e-06,
      "loss": 0.4613,
      "step": 6919
    },
    {
      "epoch": 0.31756229636088296,
      "grad_norm": 0.48358017206192017,
      "learning_rate": 9.811855830122795e-06,
      "loss": 0.4758,
      "step": 6920
    },
    {
      "epoch": 0.3176081868661374,
      "grad_norm": 0.46165886521339417,
      "learning_rate": 9.811789197376427e-06,
      "loss": 0.3356,
      "step": 6921
    },
    {
      "epoch": 0.31765407737139184,
      "grad_norm": 0.4359027147293091,
      "learning_rate": 9.811722553059224e-06,
      "loss": 0.3635,
      "step": 6922
    },
    {
      "epoch": 0.31769996787664634,
      "grad_norm": 0.45390981435775757,
      "learning_rate": 9.811655897171348e-06,
      "loss": 0.3932,
      "step": 6923
    },
    {
      "epoch": 0.3177458583819008,
      "grad_norm": 0.4628724753856659,
      "learning_rate": 9.81158922971296e-06,
      "loss": 0.4826,
      "step": 6924
    },
    {
      "epoch": 0.3177917488871552,
      "grad_norm": 0.4714392125606537,
      "learning_rate": 9.811522550684214e-06,
      "loss": 0.3806,
      "step": 6925
    },
    {
      "epoch": 0.3178376393924097,
      "grad_norm": 0.485016405582428,
      "learning_rate": 9.811455860085277e-06,
      "loss": 0.4665,
      "step": 6926
    },
    {
      "epoch": 0.31788352989766416,
      "grad_norm": 0.4822465777397156,
      "learning_rate": 9.811389157916307e-06,
      "loss": 0.4384,
      "step": 6927
    },
    {
      "epoch": 0.31792942040291866,
      "grad_norm": 0.46231627464294434,
      "learning_rate": 9.811322444177464e-06,
      "loss": 0.3518,
      "step": 6928
    },
    {
      "epoch": 0.3179753109081731,
      "grad_norm": 0.42591583728790283,
      "learning_rate": 9.811255718868911e-06,
      "loss": 0.3583,
      "step": 6929
    },
    {
      "epoch": 0.31802120141342755,
      "grad_norm": 0.4956151247024536,
      "learning_rate": 9.811188981990804e-06,
      "loss": 0.4451,
      "step": 6930
    },
    {
      "epoch": 0.31806709191868204,
      "grad_norm": 0.43963199853897095,
      "learning_rate": 9.81112223354331e-06,
      "loss": 0.3631,
      "step": 6931
    },
    {
      "epoch": 0.3181129824239365,
      "grad_norm": 0.43790316581726074,
      "learning_rate": 9.811055473526581e-06,
      "loss": 0.3464,
      "step": 6932
    },
    {
      "epoch": 0.31815887292919093,
      "grad_norm": 0.46583110094070435,
      "learning_rate": 9.810988701940784e-06,
      "loss": 0.3773,
      "step": 6933
    },
    {
      "epoch": 0.3182047634344454,
      "grad_norm": 0.4792909622192383,
      "learning_rate": 9.810921918786078e-06,
      "loss": 0.3611,
      "step": 6934
    },
    {
      "epoch": 0.31825065393969987,
      "grad_norm": 0.43365955352783203,
      "learning_rate": 9.810855124062622e-06,
      "loss": 0.3427,
      "step": 6935
    },
    {
      "epoch": 0.31829654444495437,
      "grad_norm": 0.4479745924472809,
      "learning_rate": 9.810788317770578e-06,
      "loss": 0.3929,
      "step": 6936
    },
    {
      "epoch": 0.3183424349502088,
      "grad_norm": 0.45520591735839844,
      "learning_rate": 9.810721499910106e-06,
      "loss": 0.3704,
      "step": 6937
    },
    {
      "epoch": 0.31838832545546325,
      "grad_norm": 0.4315906763076782,
      "learning_rate": 9.810654670481367e-06,
      "loss": 0.3478,
      "step": 6938
    },
    {
      "epoch": 0.31843421596071775,
      "grad_norm": 0.504904568195343,
      "learning_rate": 9.810587829484524e-06,
      "loss": 0.4925,
      "step": 6939
    },
    {
      "epoch": 0.3184801064659722,
      "grad_norm": 0.4537879228591919,
      "learning_rate": 9.810520976919734e-06,
      "loss": 0.3708,
      "step": 6940
    },
    {
      "epoch": 0.31852599697122663,
      "grad_norm": 0.4790514409542084,
      "learning_rate": 9.81045411278716e-06,
      "loss": 0.36,
      "step": 6941
    },
    {
      "epoch": 0.31857188747648113,
      "grad_norm": 0.44858425855636597,
      "learning_rate": 9.810387237086961e-06,
      "loss": 0.3354,
      "step": 6942
    },
    {
      "epoch": 0.3186177779817356,
      "grad_norm": 0.4829810559749603,
      "learning_rate": 9.8103203498193e-06,
      "loss": 0.4302,
      "step": 6943
    },
    {
      "epoch": 0.31866366848699,
      "grad_norm": 0.48292991518974304,
      "learning_rate": 9.810253450984336e-06,
      "loss": 0.3937,
      "step": 6944
    },
    {
      "epoch": 0.3187095589922445,
      "grad_norm": 0.4910351037979126,
      "learning_rate": 9.810186540582229e-06,
      "loss": 0.3995,
      "step": 6945
    },
    {
      "epoch": 0.31875544949749895,
      "grad_norm": 0.4695226848125458,
      "learning_rate": 9.810119618613144e-06,
      "loss": 0.3814,
      "step": 6946
    },
    {
      "epoch": 0.31880134000275345,
      "grad_norm": 0.4637891948223114,
      "learning_rate": 9.810052685077239e-06,
      "loss": 0.3707,
      "step": 6947
    },
    {
      "epoch": 0.3188472305080079,
      "grad_norm": 0.45026934146881104,
      "learning_rate": 9.809985739974674e-06,
      "loss": 0.363,
      "step": 6948
    },
    {
      "epoch": 0.31889312101326234,
      "grad_norm": 0.5132016539573669,
      "learning_rate": 9.809918783305614e-06,
      "loss": 0.4293,
      "step": 6949
    },
    {
      "epoch": 0.31893901151851684,
      "grad_norm": 0.4351544976234436,
      "learning_rate": 9.809851815070215e-06,
      "loss": 0.3798,
      "step": 6950
    },
    {
      "epoch": 0.3189849020237713,
      "grad_norm": 0.48168808221817017,
      "learning_rate": 9.80978483526864e-06,
      "loss": 0.4532,
      "step": 6951
    },
    {
      "epoch": 0.3190307925290257,
      "grad_norm": 0.4781895577907562,
      "learning_rate": 9.809717843901052e-06,
      "loss": 0.4552,
      "step": 6952
    },
    {
      "epoch": 0.3190766830342802,
      "grad_norm": 0.48530498147010803,
      "learning_rate": 9.80965084096761e-06,
      "loss": 0.4408,
      "step": 6953
    },
    {
      "epoch": 0.31912257353953466,
      "grad_norm": 0.4523056745529175,
      "learning_rate": 9.809583826468477e-06,
      "loss": 0.3825,
      "step": 6954
    },
    {
      "epoch": 0.31916846404478916,
      "grad_norm": 0.4712919294834137,
      "learning_rate": 9.809516800403811e-06,
      "loss": 0.4002,
      "step": 6955
    },
    {
      "epoch": 0.3192143545500436,
      "grad_norm": 0.47801870107650757,
      "learning_rate": 9.809449762773776e-06,
      "loss": 0.4436,
      "step": 6956
    },
    {
      "epoch": 0.31926024505529804,
      "grad_norm": 0.45333749055862427,
      "learning_rate": 9.809382713578533e-06,
      "loss": 0.367,
      "step": 6957
    },
    {
      "epoch": 0.31930613556055254,
      "grad_norm": 0.4778808355331421,
      "learning_rate": 9.80931565281824e-06,
      "loss": 0.4332,
      "step": 6958
    },
    {
      "epoch": 0.319352026065807,
      "grad_norm": 0.47561392188072205,
      "learning_rate": 9.809248580493062e-06,
      "loss": 0.3716,
      "step": 6959
    },
    {
      "epoch": 0.3193979165710614,
      "grad_norm": 0.4420454204082489,
      "learning_rate": 9.80918149660316e-06,
      "loss": 0.3823,
      "step": 6960
    },
    {
      "epoch": 0.3194438070763159,
      "grad_norm": 0.48482653498649597,
      "learning_rate": 9.809114401148694e-06,
      "loss": 0.3927,
      "step": 6961
    },
    {
      "epoch": 0.31948969758157036,
      "grad_norm": 0.47465482354164124,
      "learning_rate": 9.809047294129825e-06,
      "loss": 0.4422,
      "step": 6962
    },
    {
      "epoch": 0.31953558808682486,
      "grad_norm": 0.48826512694358826,
      "learning_rate": 9.808980175546715e-06,
      "loss": 0.4382,
      "step": 6963
    },
    {
      "epoch": 0.3195814785920793,
      "grad_norm": 0.4856005609035492,
      "learning_rate": 9.808913045399527e-06,
      "loss": 0.4235,
      "step": 6964
    },
    {
      "epoch": 0.31962736909733375,
      "grad_norm": 0.4142557382583618,
      "learning_rate": 9.808845903688418e-06,
      "loss": 0.3019,
      "step": 6965
    },
    {
      "epoch": 0.31967325960258824,
      "grad_norm": 0.42200809717178345,
      "learning_rate": 9.808778750413554e-06,
      "loss": 0.3034,
      "step": 6966
    },
    {
      "epoch": 0.3197191501078427,
      "grad_norm": 0.5138015151023865,
      "learning_rate": 9.808711585575094e-06,
      "loss": 0.4655,
      "step": 6967
    },
    {
      "epoch": 0.31976504061309713,
      "grad_norm": 0.41709649562835693,
      "learning_rate": 9.808644409173202e-06,
      "loss": 0.3122,
      "step": 6968
    },
    {
      "epoch": 0.3198109311183516,
      "grad_norm": 0.438488245010376,
      "learning_rate": 9.808577221208036e-06,
      "loss": 0.4023,
      "step": 6969
    },
    {
      "epoch": 0.31985682162360607,
      "grad_norm": 0.46840208768844604,
      "learning_rate": 9.808510021679761e-06,
      "loss": 0.4192,
      "step": 6970
    },
    {
      "epoch": 0.3199027121288605,
      "grad_norm": 0.47444257140159607,
      "learning_rate": 9.808442810588537e-06,
      "loss": 0.4053,
      "step": 6971
    },
    {
      "epoch": 0.319948602634115,
      "grad_norm": 0.46105051040649414,
      "learning_rate": 9.808375587934524e-06,
      "loss": 0.362,
      "step": 6972
    },
    {
      "epoch": 0.31999449313936945,
      "grad_norm": 0.4323471188545227,
      "learning_rate": 9.808308353717887e-06,
      "loss": 0.327,
      "step": 6973
    },
    {
      "epoch": 0.32004038364462395,
      "grad_norm": 0.48899421095848083,
      "learning_rate": 9.808241107938783e-06,
      "loss": 0.4483,
      "step": 6974
    },
    {
      "epoch": 0.3200862741498784,
      "grad_norm": 0.4337638020515442,
      "learning_rate": 9.80817385059738e-06,
      "loss": 0.382,
      "step": 6975
    },
    {
      "epoch": 0.32013216465513283,
      "grad_norm": 0.4616484045982361,
      "learning_rate": 9.808106581693835e-06,
      "loss": 0.4113,
      "step": 6976
    },
    {
      "epoch": 0.32017805516038733,
      "grad_norm": 0.45937931537628174,
      "learning_rate": 9.808039301228311e-06,
      "loss": 0.3923,
      "step": 6977
    },
    {
      "epoch": 0.3202239456656418,
      "grad_norm": 0.44955921173095703,
      "learning_rate": 9.80797200920097e-06,
      "loss": 0.3513,
      "step": 6978
    },
    {
      "epoch": 0.3202698361708962,
      "grad_norm": 0.4731086790561676,
      "learning_rate": 9.807904705611975e-06,
      "loss": 0.4254,
      "step": 6979
    },
    {
      "epoch": 0.3203157266761507,
      "grad_norm": 0.4315633475780487,
      "learning_rate": 9.807837390461486e-06,
      "loss": 0.3215,
      "step": 6980
    },
    {
      "epoch": 0.32036161718140516,
      "grad_norm": 0.449117511510849,
      "learning_rate": 9.807770063749664e-06,
      "loss": 0.3426,
      "step": 6981
    },
    {
      "epoch": 0.32040750768665965,
      "grad_norm": 0.4739314615726471,
      "learning_rate": 9.807702725476673e-06,
      "loss": 0.4126,
      "step": 6982
    },
    {
      "epoch": 0.3204533981919141,
      "grad_norm": 0.4968165457248688,
      "learning_rate": 9.807635375642676e-06,
      "loss": 0.4268,
      "step": 6983
    },
    {
      "epoch": 0.32049928869716854,
      "grad_norm": 0.47270330786705017,
      "learning_rate": 9.807568014247831e-06,
      "loss": 0.42,
      "step": 6984
    },
    {
      "epoch": 0.32054517920242304,
      "grad_norm": 0.43920719623565674,
      "learning_rate": 9.807500641292304e-06,
      "loss": 0.3528,
      "step": 6985
    },
    {
      "epoch": 0.3205910697076775,
      "grad_norm": 0.4495665431022644,
      "learning_rate": 9.807433256776255e-06,
      "loss": 0.3774,
      "step": 6986
    },
    {
      "epoch": 0.3206369602129319,
      "grad_norm": 0.4770635664463043,
      "learning_rate": 9.807365860699844e-06,
      "loss": 0.4482,
      "step": 6987
    },
    {
      "epoch": 0.3206828507181864,
      "grad_norm": 0.4722910225391388,
      "learning_rate": 9.807298453063239e-06,
      "loss": 0.3922,
      "step": 6988
    },
    {
      "epoch": 0.32072874122344086,
      "grad_norm": 0.5004497170448303,
      "learning_rate": 9.807231033866596e-06,
      "loss": 0.4724,
      "step": 6989
    },
    {
      "epoch": 0.32077463172869536,
      "grad_norm": 0.44961291551589966,
      "learning_rate": 9.80716360311008e-06,
      "loss": 0.3485,
      "step": 6990
    },
    {
      "epoch": 0.3208205222339498,
      "grad_norm": 0.5045326948165894,
      "learning_rate": 9.807096160793852e-06,
      "loss": 0.5059,
      "step": 6991
    },
    {
      "epoch": 0.32086641273920424,
      "grad_norm": 0.4946959614753723,
      "learning_rate": 9.807028706918076e-06,
      "loss": 0.3704,
      "step": 6992
    },
    {
      "epoch": 0.32091230324445874,
      "grad_norm": 0.44438186287879944,
      "learning_rate": 9.806961241482913e-06,
      "loss": 0.3803,
      "step": 6993
    },
    {
      "epoch": 0.3209581937497132,
      "grad_norm": 0.46101340651512146,
      "learning_rate": 9.806893764488526e-06,
      "loss": 0.4262,
      "step": 6994
    },
    {
      "epoch": 0.3210040842549676,
      "grad_norm": 0.44207680225372314,
      "learning_rate": 9.806826275935076e-06,
      "loss": 0.3462,
      "step": 6995
    },
    {
      "epoch": 0.3210499747602221,
      "grad_norm": 0.5115715265274048,
      "learning_rate": 9.806758775822724e-06,
      "loss": 0.4651,
      "step": 6996
    },
    {
      "epoch": 0.32109586526547657,
      "grad_norm": 0.45746099948883057,
      "learning_rate": 9.806691264151637e-06,
      "loss": 0.3775,
      "step": 6997
    },
    {
      "epoch": 0.32114175577073106,
      "grad_norm": 0.4725476801395416,
      "learning_rate": 9.806623740921974e-06,
      "loss": 0.4832,
      "step": 6998
    },
    {
      "epoch": 0.3211876462759855,
      "grad_norm": 0.5354151725769043,
      "learning_rate": 9.806556206133897e-06,
      "loss": 0.4192,
      "step": 6999
    },
    {
      "epoch": 0.32123353678123995,
      "grad_norm": 0.4477815330028534,
      "learning_rate": 9.80648865978757e-06,
      "loss": 0.3633,
      "step": 7000
    },
    {
      "epoch": 0.32127942728649445,
      "grad_norm": 0.49112147092819214,
      "learning_rate": 9.806421101883155e-06,
      "loss": 0.4279,
      "step": 7001
    },
    {
      "epoch": 0.3213253177917489,
      "grad_norm": 0.40987682342529297,
      "learning_rate": 9.806353532420812e-06,
      "loss": 0.315,
      "step": 7002
    },
    {
      "epoch": 0.32137120829700333,
      "grad_norm": 0.48975443840026855,
      "learning_rate": 9.806285951400707e-06,
      "loss": 0.4083,
      "step": 7003
    },
    {
      "epoch": 0.32141709880225783,
      "grad_norm": 0.4901741147041321,
      "learning_rate": 9.806218358823003e-06,
      "loss": 0.4654,
      "step": 7004
    },
    {
      "epoch": 0.32146298930751227,
      "grad_norm": 0.47973528504371643,
      "learning_rate": 9.806150754687858e-06,
      "loss": 0.3852,
      "step": 7005
    },
    {
      "epoch": 0.3215088798127667,
      "grad_norm": 0.497993141412735,
      "learning_rate": 9.80608313899544e-06,
      "loss": 0.471,
      "step": 7006
    },
    {
      "epoch": 0.3215547703180212,
      "grad_norm": 0.4888631999492645,
      "learning_rate": 9.806015511745906e-06,
      "loss": 0.4445,
      "step": 7007
    },
    {
      "epoch": 0.32160066082327565,
      "grad_norm": 0.5411748886108398,
      "learning_rate": 9.805947872939424e-06,
      "loss": 0.5165,
      "step": 7008
    },
    {
      "epoch": 0.32164655132853015,
      "grad_norm": 0.5264907479286194,
      "learning_rate": 9.805880222576151e-06,
      "loss": 0.5388,
      "step": 7009
    },
    {
      "epoch": 0.3216924418337846,
      "grad_norm": 0.4764230251312256,
      "learning_rate": 9.805812560656256e-06,
      "loss": 0.4247,
      "step": 7010
    },
    {
      "epoch": 0.32173833233903903,
      "grad_norm": 0.4643279016017914,
      "learning_rate": 9.805744887179897e-06,
      "loss": 0.3476,
      "step": 7011
    },
    {
      "epoch": 0.32178422284429353,
      "grad_norm": 0.47345560789108276,
      "learning_rate": 9.805677202147239e-06,
      "loss": 0.4037,
      "step": 7012
    },
    {
      "epoch": 0.321830113349548,
      "grad_norm": 0.4348125457763672,
      "learning_rate": 9.805609505558444e-06,
      "loss": 0.3175,
      "step": 7013
    },
    {
      "epoch": 0.3218760038548024,
      "grad_norm": 0.5154327750205994,
      "learning_rate": 9.805541797413674e-06,
      "loss": 0.4469,
      "step": 7014
    },
    {
      "epoch": 0.3219218943600569,
      "grad_norm": 0.45275041460990906,
      "learning_rate": 9.805474077713094e-06,
      "loss": 0.3966,
      "step": 7015
    },
    {
      "epoch": 0.32196778486531136,
      "grad_norm": 0.49300527572631836,
      "learning_rate": 9.805406346456864e-06,
      "loss": 0.4317,
      "step": 7016
    },
    {
      "epoch": 0.32201367537056585,
      "grad_norm": 0.460650235414505,
      "learning_rate": 9.80533860364515e-06,
      "loss": 0.3902,
      "step": 7017
    },
    {
      "epoch": 0.3220595658758203,
      "grad_norm": 0.5048458576202393,
      "learning_rate": 9.805270849278112e-06,
      "loss": 0.4507,
      "step": 7018
    },
    {
      "epoch": 0.32210545638107474,
      "grad_norm": 0.5237061381340027,
      "learning_rate": 9.805203083355915e-06,
      "loss": 0.5366,
      "step": 7019
    },
    {
      "epoch": 0.32215134688632924,
      "grad_norm": 0.5383538007736206,
      "learning_rate": 9.805135305878721e-06,
      "loss": 0.5092,
      "step": 7020
    },
    {
      "epoch": 0.3221972373915837,
      "grad_norm": 0.450731486082077,
      "learning_rate": 9.805067516846694e-06,
      "loss": 0.3733,
      "step": 7021
    },
    {
      "epoch": 0.3222431278968381,
      "grad_norm": 0.44247156381607056,
      "learning_rate": 9.804999716259994e-06,
      "loss": 0.3733,
      "step": 7022
    },
    {
      "epoch": 0.3222890184020926,
      "grad_norm": 0.4735422432422638,
      "learning_rate": 9.804931904118787e-06,
      "loss": 0.3995,
      "step": 7023
    },
    {
      "epoch": 0.32233490890734706,
      "grad_norm": 0.4424068033695221,
      "learning_rate": 9.804864080423235e-06,
      "loss": 0.3455,
      "step": 7024
    },
    {
      "epoch": 0.32238079941260156,
      "grad_norm": 0.46388566493988037,
      "learning_rate": 9.804796245173502e-06,
      "loss": 0.4523,
      "step": 7025
    },
    {
      "epoch": 0.322426689917856,
      "grad_norm": 0.4692797064781189,
      "learning_rate": 9.80472839836975e-06,
      "loss": 0.4066,
      "step": 7026
    },
    {
      "epoch": 0.32247258042311044,
      "grad_norm": 0.4479910731315613,
      "learning_rate": 9.804660540012144e-06,
      "loss": 0.3168,
      "step": 7027
    },
    {
      "epoch": 0.32251847092836494,
      "grad_norm": 0.46002689003944397,
      "learning_rate": 9.804592670100844e-06,
      "loss": 0.3653,
      "step": 7028
    },
    {
      "epoch": 0.3225643614336194,
      "grad_norm": 0.491105854511261,
      "learning_rate": 9.804524788636015e-06,
      "loss": 0.4993,
      "step": 7029
    },
    {
      "epoch": 0.3226102519388738,
      "grad_norm": 0.45007118582725525,
      "learning_rate": 9.804456895617822e-06,
      "loss": 0.3991,
      "step": 7030
    },
    {
      "epoch": 0.3226561424441283,
      "grad_norm": 0.48447445034980774,
      "learning_rate": 9.804388991046425e-06,
      "loss": 0.4252,
      "step": 7031
    },
    {
      "epoch": 0.32270203294938277,
      "grad_norm": 0.4327069818973541,
      "learning_rate": 9.80432107492199e-06,
      "loss": 0.3564,
      "step": 7032
    },
    {
      "epoch": 0.3227479234546372,
      "grad_norm": 0.47352153062820435,
      "learning_rate": 9.804253147244677e-06,
      "loss": 0.5026,
      "step": 7033
    },
    {
      "epoch": 0.3227938139598917,
      "grad_norm": 0.474072128534317,
      "learning_rate": 9.804185208014653e-06,
      "loss": 0.4239,
      "step": 7034
    },
    {
      "epoch": 0.32283970446514615,
      "grad_norm": 0.484637588262558,
      "learning_rate": 9.80411725723208e-06,
      "loss": 0.395,
      "step": 7035
    },
    {
      "epoch": 0.32288559497040065,
      "grad_norm": 0.5270601511001587,
      "learning_rate": 9.804049294897121e-06,
      "loss": 0.4765,
      "step": 7036
    },
    {
      "epoch": 0.3229314854756551,
      "grad_norm": 0.5001643300056458,
      "learning_rate": 9.803981321009938e-06,
      "loss": 0.4406,
      "step": 7037
    },
    {
      "epoch": 0.32297737598090953,
      "grad_norm": 0.5076088309288025,
      "learning_rate": 9.8039133355707e-06,
      "loss": 0.4903,
      "step": 7038
    },
    {
      "epoch": 0.32302326648616403,
      "grad_norm": 0.43508297204971313,
      "learning_rate": 9.803845338579563e-06,
      "loss": 0.3405,
      "step": 7039
    },
    {
      "epoch": 0.32306915699141847,
      "grad_norm": 0.4845043122768402,
      "learning_rate": 9.803777330036696e-06,
      "loss": 0.4611,
      "step": 7040
    },
    {
      "epoch": 0.3231150474966729,
      "grad_norm": 0.4102739989757538,
      "learning_rate": 9.80370930994226e-06,
      "loss": 0.2978,
      "step": 7041
    },
    {
      "epoch": 0.3231609380019274,
      "grad_norm": 0.49328944087028503,
      "learning_rate": 9.803641278296418e-06,
      "loss": 0.4564,
      "step": 7042
    },
    {
      "epoch": 0.32320682850718185,
      "grad_norm": 0.4268563985824585,
      "learning_rate": 9.803573235099336e-06,
      "loss": 0.3137,
      "step": 7043
    },
    {
      "epoch": 0.32325271901243635,
      "grad_norm": 0.44580456614494324,
      "learning_rate": 9.803505180351176e-06,
      "loss": 0.3794,
      "step": 7044
    },
    {
      "epoch": 0.3232986095176908,
      "grad_norm": 0.5148495435714722,
      "learning_rate": 9.803437114052103e-06,
      "loss": 0.4334,
      "step": 7045
    },
    {
      "epoch": 0.32334450002294524,
      "grad_norm": 0.4483284652233124,
      "learning_rate": 9.80336903620228e-06,
      "loss": 0.3684,
      "step": 7046
    },
    {
      "epoch": 0.32339039052819973,
      "grad_norm": 0.49176308512687683,
      "learning_rate": 9.803300946801868e-06,
      "loss": 0.4482,
      "step": 7047
    },
    {
      "epoch": 0.3234362810334542,
      "grad_norm": 0.42327481508255005,
      "learning_rate": 9.803232845851037e-06,
      "loss": 0.3303,
      "step": 7048
    },
    {
      "epoch": 0.3234821715387086,
      "grad_norm": 0.4596683382987976,
      "learning_rate": 9.803164733349944e-06,
      "loss": 0.4131,
      "step": 7049
    },
    {
      "epoch": 0.3235280620439631,
      "grad_norm": 0.43727219104766846,
      "learning_rate": 9.803096609298756e-06,
      "loss": 0.3492,
      "step": 7050
    },
    {
      "epoch": 0.32357395254921756,
      "grad_norm": 0.46289801597595215,
      "learning_rate": 9.803028473697637e-06,
      "loss": 0.3989,
      "step": 7051
    },
    {
      "epoch": 0.32361984305447206,
      "grad_norm": 0.5220200419425964,
      "learning_rate": 9.80296032654675e-06,
      "loss": 0.5431,
      "step": 7052
    },
    {
      "epoch": 0.3236657335597265,
      "grad_norm": 0.4585173428058624,
      "learning_rate": 9.80289216784626e-06,
      "loss": 0.4008,
      "step": 7053
    },
    {
      "epoch": 0.32371162406498094,
      "grad_norm": 0.46890947222709656,
      "learning_rate": 9.802823997596328e-06,
      "loss": 0.4512,
      "step": 7054
    },
    {
      "epoch": 0.32375751457023544,
      "grad_norm": 0.4436309039592743,
      "learning_rate": 9.802755815797124e-06,
      "loss": 0.3756,
      "step": 7055
    },
    {
      "epoch": 0.3238034050754899,
      "grad_norm": 0.454496294260025,
      "learning_rate": 9.802687622448805e-06,
      "loss": 0.4038,
      "step": 7056
    },
    {
      "epoch": 0.3238492955807443,
      "grad_norm": 0.46063926815986633,
      "learning_rate": 9.802619417551539e-06,
      "loss": 0.4051,
      "step": 7057
    },
    {
      "epoch": 0.3238951860859988,
      "grad_norm": 0.43899863958358765,
      "learning_rate": 9.80255120110549e-06,
      "loss": 0.3332,
      "step": 7058
    },
    {
      "epoch": 0.32394107659125326,
      "grad_norm": 0.4674995541572571,
      "learning_rate": 9.80248297311082e-06,
      "loss": 0.4234,
      "step": 7059
    },
    {
      "epoch": 0.3239869670965077,
      "grad_norm": 0.4729859530925751,
      "learning_rate": 9.802414733567695e-06,
      "loss": 0.4144,
      "step": 7060
    },
    {
      "epoch": 0.3240328576017622,
      "grad_norm": 0.476251482963562,
      "learning_rate": 9.802346482476277e-06,
      "loss": 0.4748,
      "step": 7061
    },
    {
      "epoch": 0.32407874810701665,
      "grad_norm": 0.435566782951355,
      "learning_rate": 9.802278219836731e-06,
      "loss": 0.3687,
      "step": 7062
    },
    {
      "epoch": 0.32412463861227114,
      "grad_norm": 0.46455636620521545,
      "learning_rate": 9.802209945649224e-06,
      "loss": 0.4097,
      "step": 7063
    },
    {
      "epoch": 0.3241705291175256,
      "grad_norm": 0.4642910957336426,
      "learning_rate": 9.802141659913916e-06,
      "loss": 0.4161,
      "step": 7064
    },
    {
      "epoch": 0.32421641962278,
      "grad_norm": 0.478520929813385,
      "learning_rate": 9.802073362630973e-06,
      "loss": 0.4581,
      "step": 7065
    },
    {
      "epoch": 0.3242623101280345,
      "grad_norm": 0.5269163846969604,
      "learning_rate": 9.802005053800559e-06,
      "loss": 0.4577,
      "step": 7066
    },
    {
      "epoch": 0.32430820063328897,
      "grad_norm": 0.4635535180568695,
      "learning_rate": 9.801936733422839e-06,
      "loss": 0.401,
      "step": 7067
    },
    {
      "epoch": 0.3243540911385434,
      "grad_norm": 0.4734187126159668,
      "learning_rate": 9.801868401497975e-06,
      "loss": 0.4034,
      "step": 7068
    },
    {
      "epoch": 0.3243999816437979,
      "grad_norm": 0.45452624559402466,
      "learning_rate": 9.801800058026136e-06,
      "loss": 0.4311,
      "step": 7069
    },
    {
      "epoch": 0.32444587214905235,
      "grad_norm": 0.47766679525375366,
      "learning_rate": 9.801731703007481e-06,
      "loss": 0.4112,
      "step": 7070
    },
    {
      "epoch": 0.32449176265430685,
      "grad_norm": 0.48639923334121704,
      "learning_rate": 9.801663336442178e-06,
      "loss": 0.4079,
      "step": 7071
    },
    {
      "epoch": 0.3245376531595613,
      "grad_norm": 0.478443443775177,
      "learning_rate": 9.80159495833039e-06,
      "loss": 0.4015,
      "step": 7072
    },
    {
      "epoch": 0.32458354366481573,
      "grad_norm": 0.5137795805931091,
      "learning_rate": 9.80152656867228e-06,
      "loss": 0.4385,
      "step": 7073
    },
    {
      "epoch": 0.32462943417007023,
      "grad_norm": 0.5265718102455139,
      "learning_rate": 9.801458167468015e-06,
      "loss": 0.3669,
      "step": 7074
    },
    {
      "epoch": 0.3246753246753247,
      "grad_norm": 0.4520326852798462,
      "learning_rate": 9.801389754717758e-06,
      "loss": 0.366,
      "step": 7075
    },
    {
      "epoch": 0.3247212151805791,
      "grad_norm": 0.4636530578136444,
      "learning_rate": 9.801321330421676e-06,
      "loss": 0.3618,
      "step": 7076
    },
    {
      "epoch": 0.3247671056858336,
      "grad_norm": 0.46683430671691895,
      "learning_rate": 9.80125289457993e-06,
      "loss": 0.354,
      "step": 7077
    },
    {
      "epoch": 0.32481299619108805,
      "grad_norm": 0.5233639478683472,
      "learning_rate": 9.801184447192686e-06,
      "loss": 0.4779,
      "step": 7078
    },
    {
      "epoch": 0.32485888669634255,
      "grad_norm": 0.4272993803024292,
      "learning_rate": 9.801115988260109e-06,
      "loss": 0.3138,
      "step": 7079
    },
    {
      "epoch": 0.324904777201597,
      "grad_norm": 0.46075162291526794,
      "learning_rate": 9.801047517782362e-06,
      "loss": 0.3997,
      "step": 7080
    },
    {
      "epoch": 0.32495066770685144,
      "grad_norm": 0.4669494330883026,
      "learning_rate": 9.800979035759614e-06,
      "loss": 0.4474,
      "step": 7081
    },
    {
      "epoch": 0.32499655821210593,
      "grad_norm": 0.48696446418762207,
      "learning_rate": 9.800910542192024e-06,
      "loss": 0.4384,
      "step": 7082
    },
    {
      "epoch": 0.3250424487173604,
      "grad_norm": 0.4249604344367981,
      "learning_rate": 9.800842037079761e-06,
      "loss": 0.3286,
      "step": 7083
    },
    {
      "epoch": 0.3250883392226148,
      "grad_norm": 0.44789379835128784,
      "learning_rate": 9.800773520422988e-06,
      "loss": 0.381,
      "step": 7084
    },
    {
      "epoch": 0.3251342297278693,
      "grad_norm": 0.45104047656059265,
      "learning_rate": 9.800704992221867e-06,
      "loss": 0.3582,
      "step": 7085
    },
    {
      "epoch": 0.32518012023312376,
      "grad_norm": 0.4855601489543915,
      "learning_rate": 9.800636452476568e-06,
      "loss": 0.436,
      "step": 7086
    },
    {
      "epoch": 0.32522601073837826,
      "grad_norm": 0.48952972888946533,
      "learning_rate": 9.800567901187254e-06,
      "loss": 0.4434,
      "step": 7087
    },
    {
      "epoch": 0.3252719012436327,
      "grad_norm": 0.45446062088012695,
      "learning_rate": 9.80049933835409e-06,
      "loss": 0.4054,
      "step": 7088
    },
    {
      "epoch": 0.32531779174888714,
      "grad_norm": 0.42279475927352905,
      "learning_rate": 9.800430763977237e-06,
      "loss": 0.3123,
      "step": 7089
    },
    {
      "epoch": 0.32536368225414164,
      "grad_norm": 0.4780943691730499,
      "learning_rate": 9.800362178056866e-06,
      "loss": 0.4539,
      "step": 7090
    },
    {
      "epoch": 0.3254095727593961,
      "grad_norm": 0.4395153820514679,
      "learning_rate": 9.800293580593137e-06,
      "loss": 0.3645,
      "step": 7091
    },
    {
      "epoch": 0.3254554632646505,
      "grad_norm": 0.509714663028717,
      "learning_rate": 9.800224971586219e-06,
      "loss": 0.4953,
      "step": 7092
    },
    {
      "epoch": 0.325501353769905,
      "grad_norm": 0.5116782784461975,
      "learning_rate": 9.800156351036274e-06,
      "loss": 0.4945,
      "step": 7093
    },
    {
      "epoch": 0.32554724427515946,
      "grad_norm": 0.4845845401287079,
      "learning_rate": 9.800087718943467e-06,
      "loss": 0.4493,
      "step": 7094
    },
    {
      "epoch": 0.3255931347804139,
      "grad_norm": 0.4175945520401001,
      "learning_rate": 9.800019075307964e-06,
      "loss": 0.3195,
      "step": 7095
    },
    {
      "epoch": 0.3256390252856684,
      "grad_norm": 0.4356710910797119,
      "learning_rate": 9.799950420129932e-06,
      "loss": 0.3676,
      "step": 7096
    },
    {
      "epoch": 0.32568491579092285,
      "grad_norm": 0.4709814786911011,
      "learning_rate": 9.799881753409532e-06,
      "loss": 0.467,
      "step": 7097
    },
    {
      "epoch": 0.32573080629617734,
      "grad_norm": 0.4607873857021332,
      "learning_rate": 9.799813075146933e-06,
      "loss": 0.3508,
      "step": 7098
    },
    {
      "epoch": 0.3257766968014318,
      "grad_norm": 0.541644275188446,
      "learning_rate": 9.799744385342297e-06,
      "loss": 0.5019,
      "step": 7099
    },
    {
      "epoch": 0.32582258730668623,
      "grad_norm": 0.4374476373195648,
      "learning_rate": 9.79967568399579e-06,
      "loss": 0.3826,
      "step": 7100
    },
    {
      "epoch": 0.3258684778119407,
      "grad_norm": 0.422495037317276,
      "learning_rate": 9.79960697110758e-06,
      "loss": 0.3211,
      "step": 7101
    },
    {
      "epoch": 0.32591436831719517,
      "grad_norm": 0.43906939029693604,
      "learning_rate": 9.79953824667783e-06,
      "loss": 0.3528,
      "step": 7102
    },
    {
      "epoch": 0.3259602588224496,
      "grad_norm": 0.4160074293613434,
      "learning_rate": 9.799469510706703e-06,
      "loss": 0.3348,
      "step": 7103
    },
    {
      "epoch": 0.3260061493277041,
      "grad_norm": 0.47561705112457275,
      "learning_rate": 9.79940076319437e-06,
      "loss": 0.3957,
      "step": 7104
    },
    {
      "epoch": 0.32605203983295855,
      "grad_norm": 0.46520480513572693,
      "learning_rate": 9.799332004140989e-06,
      "loss": 0.3878,
      "step": 7105
    },
    {
      "epoch": 0.32609793033821305,
      "grad_norm": 0.5143229365348816,
      "learning_rate": 9.799263233546731e-06,
      "loss": 0.4957,
      "step": 7106
    },
    {
      "epoch": 0.3261438208434675,
      "grad_norm": 0.456076979637146,
      "learning_rate": 9.79919445141176e-06,
      "loss": 0.3233,
      "step": 7107
    },
    {
      "epoch": 0.32618971134872193,
      "grad_norm": 0.439850389957428,
      "learning_rate": 9.799125657736241e-06,
      "loss": 0.3266,
      "step": 7108
    },
    {
      "epoch": 0.32623560185397643,
      "grad_norm": 0.4379720985889435,
      "learning_rate": 9.799056852520339e-06,
      "loss": 0.3678,
      "step": 7109
    },
    {
      "epoch": 0.3262814923592309,
      "grad_norm": 0.45695942640304565,
      "learning_rate": 9.798988035764219e-06,
      "loss": 0.3551,
      "step": 7110
    },
    {
      "epoch": 0.3263273828644853,
      "grad_norm": 0.4401470124721527,
      "learning_rate": 9.798919207468048e-06,
      "loss": 0.3616,
      "step": 7111
    },
    {
      "epoch": 0.3263732733697398,
      "grad_norm": 0.4338792860507965,
      "learning_rate": 9.798850367631991e-06,
      "loss": 0.3577,
      "step": 7112
    },
    {
      "epoch": 0.32641916387499426,
      "grad_norm": 0.469707190990448,
      "learning_rate": 9.798781516256215e-06,
      "loss": 0.3933,
      "step": 7113
    },
    {
      "epoch": 0.32646505438024875,
      "grad_norm": 0.4474703371524811,
      "learning_rate": 9.798712653340882e-06,
      "loss": 0.3433,
      "step": 7114
    },
    {
      "epoch": 0.3265109448855032,
      "grad_norm": 0.45166924595832825,
      "learning_rate": 9.79864377888616e-06,
      "loss": 0.3616,
      "step": 7115
    },
    {
      "epoch": 0.32655683539075764,
      "grad_norm": 0.4328984022140503,
      "learning_rate": 9.798574892892214e-06,
      "loss": 0.3671,
      "step": 7116
    },
    {
      "epoch": 0.32660272589601214,
      "grad_norm": 0.4833146035671234,
      "learning_rate": 9.79850599535921e-06,
      "loss": 0.4224,
      "step": 7117
    },
    {
      "epoch": 0.3266486164012666,
      "grad_norm": 0.4277803897857666,
      "learning_rate": 9.798437086287316e-06,
      "loss": 0.3241,
      "step": 7118
    },
    {
      "epoch": 0.326694506906521,
      "grad_norm": 0.5145024657249451,
      "learning_rate": 9.798368165676693e-06,
      "loss": 0.4964,
      "step": 7119
    },
    {
      "epoch": 0.3267403974117755,
      "grad_norm": 0.42128852009773254,
      "learning_rate": 9.798299233527508e-06,
      "loss": 0.3438,
      "step": 7120
    },
    {
      "epoch": 0.32678628791702996,
      "grad_norm": 0.4736366271972656,
      "learning_rate": 9.79823028983993e-06,
      "loss": 0.4334,
      "step": 7121
    },
    {
      "epoch": 0.3268321784222844,
      "grad_norm": 0.42067641019821167,
      "learning_rate": 9.79816133461412e-06,
      "loss": 0.328,
      "step": 7122
    },
    {
      "epoch": 0.3268780689275389,
      "grad_norm": 0.45127958059310913,
      "learning_rate": 9.798092367850248e-06,
      "loss": 0.3417,
      "step": 7123
    },
    {
      "epoch": 0.32692395943279334,
      "grad_norm": 0.4792923033237457,
      "learning_rate": 9.798023389548478e-06,
      "loss": 0.4227,
      "step": 7124
    },
    {
      "epoch": 0.32696984993804784,
      "grad_norm": 0.5275638699531555,
      "learning_rate": 9.797954399708975e-06,
      "loss": 0.4821,
      "step": 7125
    },
    {
      "epoch": 0.3270157404433023,
      "grad_norm": 0.48499158024787903,
      "learning_rate": 9.797885398331908e-06,
      "loss": 0.43,
      "step": 7126
    },
    {
      "epoch": 0.3270616309485567,
      "grad_norm": 0.6616424322128296,
      "learning_rate": 9.797816385417438e-06,
      "loss": 0.3699,
      "step": 7127
    },
    {
      "epoch": 0.3271075214538112,
      "grad_norm": 0.47677940130233765,
      "learning_rate": 9.797747360965736e-06,
      "loss": 0.4036,
      "step": 7128
    },
    {
      "epoch": 0.32715341195906567,
      "grad_norm": 0.5152416229248047,
      "learning_rate": 9.797678324976966e-06,
      "loss": 0.4422,
      "step": 7129
    },
    {
      "epoch": 0.3271993024643201,
      "grad_norm": 0.4563952088356018,
      "learning_rate": 9.797609277451294e-06,
      "loss": 0.3685,
      "step": 7130
    },
    {
      "epoch": 0.3272451929695746,
      "grad_norm": 0.44995447993278503,
      "learning_rate": 9.797540218388884e-06,
      "loss": 0.374,
      "step": 7131
    },
    {
      "epoch": 0.32729108347482905,
      "grad_norm": 0.4183056354522705,
      "learning_rate": 9.797471147789905e-06,
      "loss": 0.3169,
      "step": 7132
    },
    {
      "epoch": 0.32733697398008355,
      "grad_norm": 0.45963019132614136,
      "learning_rate": 9.797402065654522e-06,
      "loss": 0.3875,
      "step": 7133
    },
    {
      "epoch": 0.327382864485338,
      "grad_norm": 0.48143985867500305,
      "learning_rate": 9.797332971982902e-06,
      "loss": 0.4199,
      "step": 7134
    },
    {
      "epoch": 0.32742875499059243,
      "grad_norm": 0.4704936146736145,
      "learning_rate": 9.797263866775209e-06,
      "loss": 0.4185,
      "step": 7135
    },
    {
      "epoch": 0.3274746454958469,
      "grad_norm": 0.4663526713848114,
      "learning_rate": 9.79719475003161e-06,
      "loss": 0.3888,
      "step": 7136
    },
    {
      "epoch": 0.32752053600110137,
      "grad_norm": 0.45433521270751953,
      "learning_rate": 9.797125621752273e-06,
      "loss": 0.386,
      "step": 7137
    },
    {
      "epoch": 0.3275664265063558,
      "grad_norm": 0.4518144130706787,
      "learning_rate": 9.797056481937362e-06,
      "loss": 0.3491,
      "step": 7138
    },
    {
      "epoch": 0.3276123170116103,
      "grad_norm": 0.5075174570083618,
      "learning_rate": 9.796987330587045e-06,
      "loss": 0.4592,
      "step": 7139
    },
    {
      "epoch": 0.32765820751686475,
      "grad_norm": 0.46694523096084595,
      "learning_rate": 9.796918167701487e-06,
      "loss": 0.4264,
      "step": 7140
    },
    {
      "epoch": 0.32770409802211925,
      "grad_norm": 0.4850994944572449,
      "learning_rate": 9.796848993280853e-06,
      "loss": 0.4002,
      "step": 7141
    },
    {
      "epoch": 0.3277499885273737,
      "grad_norm": 0.4872625172138214,
      "learning_rate": 9.796779807325313e-06,
      "loss": 0.3883,
      "step": 7142
    },
    {
      "epoch": 0.32779587903262813,
      "grad_norm": 0.5141732692718506,
      "learning_rate": 9.796710609835031e-06,
      "loss": 0.4552,
      "step": 7143
    },
    {
      "epoch": 0.32784176953788263,
      "grad_norm": 0.4589981138706207,
      "learning_rate": 9.796641400810176e-06,
      "loss": 0.3665,
      "step": 7144
    },
    {
      "epoch": 0.3278876600431371,
      "grad_norm": 0.5192262530326843,
      "learning_rate": 9.796572180250909e-06,
      "loss": 0.4587,
      "step": 7145
    },
    {
      "epoch": 0.3279335505483915,
      "grad_norm": 0.4339587390422821,
      "learning_rate": 9.796502948157403e-06,
      "loss": 0.3545,
      "step": 7146
    },
    {
      "epoch": 0.327979441053646,
      "grad_norm": 0.5236297249794006,
      "learning_rate": 9.796433704529818e-06,
      "loss": 0.4239,
      "step": 7147
    },
    {
      "epoch": 0.32802533155890046,
      "grad_norm": 0.44440963864326477,
      "learning_rate": 9.796364449368326e-06,
      "loss": 0.3881,
      "step": 7148
    },
    {
      "epoch": 0.3280712220641549,
      "grad_norm": 0.47269517183303833,
      "learning_rate": 9.79629518267309e-06,
      "loss": 0.419,
      "step": 7149
    },
    {
      "epoch": 0.3281171125694094,
      "grad_norm": 0.4474486708641052,
      "learning_rate": 9.79622590444428e-06,
      "loss": 0.3476,
      "step": 7150
    },
    {
      "epoch": 0.32816300307466384,
      "grad_norm": 0.4910537600517273,
      "learning_rate": 9.796156614682058e-06,
      "loss": 0.4651,
      "step": 7151
    },
    {
      "epoch": 0.32820889357991834,
      "grad_norm": 0.4842228591442108,
      "learning_rate": 9.796087313386594e-06,
      "loss": 0.4837,
      "step": 7152
    },
    {
      "epoch": 0.3282547840851728,
      "grad_norm": 0.42983606457710266,
      "learning_rate": 9.796018000558052e-06,
      "loss": 0.3841,
      "step": 7153
    },
    {
      "epoch": 0.3283006745904272,
      "grad_norm": 0.48621898889541626,
      "learning_rate": 9.795948676196602e-06,
      "loss": 0.5052,
      "step": 7154
    },
    {
      "epoch": 0.3283465650956817,
      "grad_norm": 0.5249859690666199,
      "learning_rate": 9.79587934030241e-06,
      "loss": 0.4923,
      "step": 7155
    },
    {
      "epoch": 0.32839245560093616,
      "grad_norm": 0.4727841019630432,
      "learning_rate": 9.79580999287564e-06,
      "loss": 0.3707,
      "step": 7156
    },
    {
      "epoch": 0.3284383461061906,
      "grad_norm": 0.46034321188926697,
      "learning_rate": 9.795740633916462e-06,
      "loss": 0.4326,
      "step": 7157
    },
    {
      "epoch": 0.3284842366114451,
      "grad_norm": 0.478100061416626,
      "learning_rate": 9.79567126342504e-06,
      "loss": 0.3785,
      "step": 7158
    },
    {
      "epoch": 0.32853012711669954,
      "grad_norm": 0.4651584327220917,
      "learning_rate": 9.795601881401544e-06,
      "loss": 0.3653,
      "step": 7159
    },
    {
      "epoch": 0.32857601762195404,
      "grad_norm": 0.4878479838371277,
      "learning_rate": 9.795532487846138e-06,
      "loss": 0.4823,
      "step": 7160
    },
    {
      "epoch": 0.3286219081272085,
      "grad_norm": 0.4737392067909241,
      "learning_rate": 9.79546308275899e-06,
      "loss": 0.4219,
      "step": 7161
    },
    {
      "epoch": 0.3286677986324629,
      "grad_norm": 0.41405847668647766,
      "learning_rate": 9.795393666140264e-06,
      "loss": 0.3403,
      "step": 7162
    },
    {
      "epoch": 0.3287136891377174,
      "grad_norm": 0.45254451036453247,
      "learning_rate": 9.795324237990134e-06,
      "loss": 0.3722,
      "step": 7163
    },
    {
      "epoch": 0.32875957964297187,
      "grad_norm": 0.5143303275108337,
      "learning_rate": 9.79525479830876e-06,
      "loss": 0.456,
      "step": 7164
    },
    {
      "epoch": 0.3288054701482263,
      "grad_norm": 0.5072575211524963,
      "learning_rate": 9.795185347096312e-06,
      "loss": 0.4478,
      "step": 7165
    },
    {
      "epoch": 0.3288513606534808,
      "grad_norm": 0.7139179110527039,
      "learning_rate": 9.795115884352958e-06,
      "loss": 0.4245,
      "step": 7166
    },
    {
      "epoch": 0.32889725115873525,
      "grad_norm": 0.4818011522293091,
      "learning_rate": 9.795046410078863e-06,
      "loss": 0.4641,
      "step": 7167
    },
    {
      "epoch": 0.32894314166398975,
      "grad_norm": 0.4635820686817169,
      "learning_rate": 9.794976924274195e-06,
      "loss": 0.3921,
      "step": 7168
    },
    {
      "epoch": 0.3289890321692442,
      "grad_norm": 0.4369891583919525,
      "learning_rate": 9.79490742693912e-06,
      "loss": 0.3806,
      "step": 7169
    },
    {
      "epoch": 0.32903492267449863,
      "grad_norm": 0.5386903285980225,
      "learning_rate": 9.794837918073807e-06,
      "loss": 0.5265,
      "step": 7170
    },
    {
      "epoch": 0.32908081317975313,
      "grad_norm": 0.4369814693927765,
      "learning_rate": 9.79476839767842e-06,
      "loss": 0.317,
      "step": 7171
    },
    {
      "epoch": 0.32912670368500757,
      "grad_norm": 0.5245389938354492,
      "learning_rate": 9.79469886575313e-06,
      "loss": 0.5484,
      "step": 7172
    },
    {
      "epoch": 0.329172594190262,
      "grad_norm": 0.480056494474411,
      "learning_rate": 9.794629322298102e-06,
      "loss": 0.4701,
      "step": 7173
    },
    {
      "epoch": 0.3292184846955165,
      "grad_norm": 0.4815145432949066,
      "learning_rate": 9.794559767313505e-06,
      "loss": 0.4118,
      "step": 7174
    },
    {
      "epoch": 0.32926437520077095,
      "grad_norm": 0.4611259400844574,
      "learning_rate": 9.794490200799504e-06,
      "loss": 0.3807,
      "step": 7175
    },
    {
      "epoch": 0.3293102657060254,
      "grad_norm": 0.45196035504341125,
      "learning_rate": 9.794420622756267e-06,
      "loss": 0.3288,
      "step": 7176
    },
    {
      "epoch": 0.3293561562112799,
      "grad_norm": 0.4814394414424896,
      "learning_rate": 9.794351033183962e-06,
      "loss": 0.4466,
      "step": 7177
    },
    {
      "epoch": 0.32940204671653434,
      "grad_norm": 0.4677477478981018,
      "learning_rate": 9.794281432082755e-06,
      "loss": 0.4389,
      "step": 7178
    },
    {
      "epoch": 0.32944793722178883,
      "grad_norm": 0.4827670156955719,
      "learning_rate": 9.794211819452814e-06,
      "loss": 0.4288,
      "step": 7179
    },
    {
      "epoch": 0.3294938277270433,
      "grad_norm": 0.489328533411026,
      "learning_rate": 9.794142195294308e-06,
      "loss": 0.4772,
      "step": 7180
    },
    {
      "epoch": 0.3295397182322977,
      "grad_norm": 0.4538560211658478,
      "learning_rate": 9.794072559607404e-06,
      "loss": 0.4263,
      "step": 7181
    },
    {
      "epoch": 0.3295856087375522,
      "grad_norm": 0.457256942987442,
      "learning_rate": 9.794002912392266e-06,
      "loss": 0.3747,
      "step": 7182
    },
    {
      "epoch": 0.32963149924280666,
      "grad_norm": 0.4503881633281708,
      "learning_rate": 9.793933253649066e-06,
      "loss": 0.4264,
      "step": 7183
    },
    {
      "epoch": 0.3296773897480611,
      "grad_norm": 0.5007647275924683,
      "learning_rate": 9.793863583377969e-06,
      "loss": 0.4302,
      "step": 7184
    },
    {
      "epoch": 0.3297232802533156,
      "grad_norm": 0.43484941124916077,
      "learning_rate": 9.793793901579143e-06,
      "loss": 0.3447,
      "step": 7185
    },
    {
      "epoch": 0.32976917075857004,
      "grad_norm": 0.4231371283531189,
      "learning_rate": 9.793724208252755e-06,
      "loss": 0.3184,
      "step": 7186
    },
    {
      "epoch": 0.32981506126382454,
      "grad_norm": 0.4640555679798126,
      "learning_rate": 9.793654503398973e-06,
      "loss": 0.433,
      "step": 7187
    },
    {
      "epoch": 0.329860951769079,
      "grad_norm": 0.47930943965911865,
      "learning_rate": 9.793584787017966e-06,
      "loss": 0.4666,
      "step": 7188
    },
    {
      "epoch": 0.3299068422743334,
      "grad_norm": 0.4279690384864807,
      "learning_rate": 9.793515059109899e-06,
      "loss": 0.2937,
      "step": 7189
    },
    {
      "epoch": 0.3299527327795879,
      "grad_norm": 0.5087365508079529,
      "learning_rate": 9.793445319674944e-06,
      "loss": 0.4128,
      "step": 7190
    },
    {
      "epoch": 0.32999862328484236,
      "grad_norm": 0.4626550078392029,
      "learning_rate": 9.793375568713264e-06,
      "loss": 0.3579,
      "step": 7191
    },
    {
      "epoch": 0.3300445137900968,
      "grad_norm": 0.4642115533351898,
      "learning_rate": 9.793305806225028e-06,
      "loss": 0.4468,
      "step": 7192
    },
    {
      "epoch": 0.3300904042953513,
      "grad_norm": 0.46554937958717346,
      "learning_rate": 9.793236032210405e-06,
      "loss": 0.3858,
      "step": 7193
    },
    {
      "epoch": 0.33013629480060575,
      "grad_norm": 0.48201119899749756,
      "learning_rate": 9.79316624666956e-06,
      "loss": 0.3998,
      "step": 7194
    },
    {
      "epoch": 0.33018218530586024,
      "grad_norm": 0.49105948209762573,
      "learning_rate": 9.793096449602666e-06,
      "loss": 0.4233,
      "step": 7195
    },
    {
      "epoch": 0.3302280758111147,
      "grad_norm": 0.44804123044013977,
      "learning_rate": 9.793026641009888e-06,
      "loss": 0.3812,
      "step": 7196
    },
    {
      "epoch": 0.3302739663163691,
      "grad_norm": 0.47043225169181824,
      "learning_rate": 9.792956820891392e-06,
      "loss": 0.3847,
      "step": 7197
    },
    {
      "epoch": 0.3303198568216236,
      "grad_norm": 0.47808828949928284,
      "learning_rate": 9.792886989247347e-06,
      "loss": 0.4185,
      "step": 7198
    },
    {
      "epoch": 0.33036574732687807,
      "grad_norm": 0.5053940415382385,
      "learning_rate": 9.792817146077923e-06,
      "loss": 0.3971,
      "step": 7199
    },
    {
      "epoch": 0.3304116378321325,
      "grad_norm": 0.49025535583496094,
      "learning_rate": 9.792747291383287e-06,
      "loss": 0.4273,
      "step": 7200
    },
    {
      "epoch": 0.330457528337387,
      "grad_norm": 0.46630677580833435,
      "learning_rate": 9.792677425163604e-06,
      "loss": 0.4232,
      "step": 7201
    },
    {
      "epoch": 0.33050341884264145,
      "grad_norm": 0.4766657054424286,
      "learning_rate": 9.792607547419046e-06,
      "loss": 0.4236,
      "step": 7202
    },
    {
      "epoch": 0.33054930934789595,
      "grad_norm": 0.42645126581192017,
      "learning_rate": 9.792537658149778e-06,
      "loss": 0.3416,
      "step": 7203
    },
    {
      "epoch": 0.3305951998531504,
      "grad_norm": 0.47774538397789,
      "learning_rate": 9.792467757355973e-06,
      "loss": 0.3896,
      "step": 7204
    },
    {
      "epoch": 0.33064109035840483,
      "grad_norm": 0.46343567967414856,
      "learning_rate": 9.792397845037793e-06,
      "loss": 0.4005,
      "step": 7205
    },
    {
      "epoch": 0.33068698086365933,
      "grad_norm": 0.4334917962551117,
      "learning_rate": 9.792327921195408e-06,
      "loss": 0.3492,
      "step": 7206
    },
    {
      "epoch": 0.3307328713689138,
      "grad_norm": 0.4725760817527771,
      "learning_rate": 9.792257985828988e-06,
      "loss": 0.4258,
      "step": 7207
    },
    {
      "epoch": 0.3307787618741682,
      "grad_norm": 0.49206680059432983,
      "learning_rate": 9.7921880389387e-06,
      "loss": 0.4647,
      "step": 7208
    },
    {
      "epoch": 0.3308246523794227,
      "grad_norm": 0.5222856402397156,
      "learning_rate": 9.792118080524713e-06,
      "loss": 0.4053,
      "step": 7209
    },
    {
      "epoch": 0.33087054288467715,
      "grad_norm": 0.473874568939209,
      "learning_rate": 9.792048110587194e-06,
      "loss": 0.3893,
      "step": 7210
    },
    {
      "epoch": 0.3309164333899316,
      "grad_norm": 0.495440274477005,
      "learning_rate": 9.791978129126309e-06,
      "loss": 0.4391,
      "step": 7211
    },
    {
      "epoch": 0.3309623238951861,
      "grad_norm": 0.4714395999908447,
      "learning_rate": 9.791908136142232e-06,
      "loss": 0.3767,
      "step": 7212
    },
    {
      "epoch": 0.33100821440044054,
      "grad_norm": 0.46735507249832153,
      "learning_rate": 9.791838131635128e-06,
      "loss": 0.392,
      "step": 7213
    },
    {
      "epoch": 0.33105410490569503,
      "grad_norm": 0.4522331655025482,
      "learning_rate": 9.791768115605165e-06,
      "loss": 0.3825,
      "step": 7214
    },
    {
      "epoch": 0.3310999954109495,
      "grad_norm": 0.4642760157585144,
      "learning_rate": 9.791698088052513e-06,
      "loss": 0.3896,
      "step": 7215
    },
    {
      "epoch": 0.3311458859162039,
      "grad_norm": 0.45967569947242737,
      "learning_rate": 9.791628048977337e-06,
      "loss": 0.3795,
      "step": 7216
    },
    {
      "epoch": 0.3311917764214584,
      "grad_norm": 0.46094194054603577,
      "learning_rate": 9.79155799837981e-06,
      "loss": 0.38,
      "step": 7217
    },
    {
      "epoch": 0.33123766692671286,
      "grad_norm": 0.4372478425502777,
      "learning_rate": 9.791487936260098e-06,
      "loss": 0.3637,
      "step": 7218
    },
    {
      "epoch": 0.3312835574319673,
      "grad_norm": 0.5078421831130981,
      "learning_rate": 9.79141786261837e-06,
      "loss": 0.4898,
      "step": 7219
    },
    {
      "epoch": 0.3313294479372218,
      "grad_norm": 0.42087045311927795,
      "learning_rate": 9.791347777454792e-06,
      "loss": 0.3459,
      "step": 7220
    },
    {
      "epoch": 0.33137533844247624,
      "grad_norm": 0.4824684262275696,
      "learning_rate": 9.791277680769537e-06,
      "loss": 0.4553,
      "step": 7221
    },
    {
      "epoch": 0.33142122894773074,
      "grad_norm": 0.45815813541412354,
      "learning_rate": 9.79120757256277e-06,
      "loss": 0.4095,
      "step": 7222
    },
    {
      "epoch": 0.3314671194529852,
      "grad_norm": 0.6095938682556152,
      "learning_rate": 9.791137452834662e-06,
      "loss": 0.4104,
      "step": 7223
    },
    {
      "epoch": 0.3315130099582396,
      "grad_norm": 0.5119222402572632,
      "learning_rate": 9.79106732158538e-06,
      "loss": 0.5002,
      "step": 7224
    },
    {
      "epoch": 0.3315589004634941,
      "grad_norm": 0.494268000125885,
      "learning_rate": 9.790997178815094e-06,
      "loss": 0.4571,
      "step": 7225
    },
    {
      "epoch": 0.33160479096874856,
      "grad_norm": 0.5009903311729431,
      "learning_rate": 9.790927024523972e-06,
      "loss": 0.4527,
      "step": 7226
    },
    {
      "epoch": 0.331650681474003,
      "grad_norm": 0.4653403162956238,
      "learning_rate": 9.79085685871218e-06,
      "loss": 0.4065,
      "step": 7227
    },
    {
      "epoch": 0.3316965719792575,
      "grad_norm": 0.4519183337688446,
      "learning_rate": 9.790786681379894e-06,
      "loss": 0.3558,
      "step": 7228
    },
    {
      "epoch": 0.33174246248451195,
      "grad_norm": 0.49175065755844116,
      "learning_rate": 9.790716492527275e-06,
      "loss": 0.4353,
      "step": 7229
    },
    {
      "epoch": 0.33178835298976644,
      "grad_norm": 0.4947044253349304,
      "learning_rate": 9.790646292154494e-06,
      "loss": 0.4518,
      "step": 7230
    },
    {
      "epoch": 0.3318342434950209,
      "grad_norm": 0.44338518381118774,
      "learning_rate": 9.790576080261722e-06,
      "loss": 0.3291,
      "step": 7231
    },
    {
      "epoch": 0.33188013400027533,
      "grad_norm": 0.44278407096862793,
      "learning_rate": 9.790505856849127e-06,
      "loss": 0.355,
      "step": 7232
    },
    {
      "epoch": 0.3319260245055298,
      "grad_norm": 0.5269693732261658,
      "learning_rate": 9.790435621916877e-06,
      "loss": 0.3901,
      "step": 7233
    },
    {
      "epoch": 0.33197191501078427,
      "grad_norm": 0.43144699931144714,
      "learning_rate": 9.79036537546514e-06,
      "loss": 0.3214,
      "step": 7234
    },
    {
      "epoch": 0.3320178055160387,
      "grad_norm": 0.46735692024230957,
      "learning_rate": 9.790295117494086e-06,
      "loss": 0.4135,
      "step": 7235
    },
    {
      "epoch": 0.3320636960212932,
      "grad_norm": 0.4459136426448822,
      "learning_rate": 9.790224848003885e-06,
      "loss": 0.3713,
      "step": 7236
    },
    {
      "epoch": 0.33210958652654765,
      "grad_norm": 0.511888325214386,
      "learning_rate": 9.790154566994706e-06,
      "loss": 0.4826,
      "step": 7237
    },
    {
      "epoch": 0.3321554770318021,
      "grad_norm": 0.519389271736145,
      "learning_rate": 9.790084274466716e-06,
      "loss": 0.4835,
      "step": 7238
    },
    {
      "epoch": 0.3322013675370566,
      "grad_norm": 0.49302688241004944,
      "learning_rate": 9.790013970420084e-06,
      "loss": 0.4773,
      "step": 7239
    },
    {
      "epoch": 0.33224725804231103,
      "grad_norm": 0.5046329498291016,
      "learning_rate": 9.789943654854983e-06,
      "loss": 0.4823,
      "step": 7240
    },
    {
      "epoch": 0.33229314854756553,
      "grad_norm": 0.46639305353164673,
      "learning_rate": 9.789873327771578e-06,
      "loss": 0.4214,
      "step": 7241
    },
    {
      "epoch": 0.33233903905282,
      "grad_norm": 0.5145673751831055,
      "learning_rate": 9.789802989170039e-06,
      "loss": 0.4655,
      "step": 7242
    },
    {
      "epoch": 0.3323849295580744,
      "grad_norm": 0.4844233989715576,
      "learning_rate": 9.789732639050536e-06,
      "loss": 0.4006,
      "step": 7243
    },
    {
      "epoch": 0.3324308200633289,
      "grad_norm": 0.5160293579101562,
      "learning_rate": 9.789662277413236e-06,
      "loss": 0.5178,
      "step": 7244
    },
    {
      "epoch": 0.33247671056858336,
      "grad_norm": 0.46510404348373413,
      "learning_rate": 9.789591904258311e-06,
      "loss": 0.4424,
      "step": 7245
    },
    {
      "epoch": 0.3325226010738378,
      "grad_norm": 0.490192711353302,
      "learning_rate": 9.78952151958593e-06,
      "loss": 0.4831,
      "step": 7246
    },
    {
      "epoch": 0.3325684915790923,
      "grad_norm": 0.46661701798439026,
      "learning_rate": 9.78945112339626e-06,
      "loss": 0.3739,
      "step": 7247
    },
    {
      "epoch": 0.33261438208434674,
      "grad_norm": 0.4853288531303406,
      "learning_rate": 9.789380715689472e-06,
      "loss": 0.4102,
      "step": 7248
    },
    {
      "epoch": 0.33266027258960124,
      "grad_norm": 0.49862831830978394,
      "learning_rate": 9.789310296465737e-06,
      "loss": 0.4497,
      "step": 7249
    },
    {
      "epoch": 0.3327061630948557,
      "grad_norm": 1.1059547662734985,
      "learning_rate": 9.789239865725219e-06,
      "loss": 0.4951,
      "step": 7250
    },
    {
      "epoch": 0.3327520536001101,
      "grad_norm": 0.48300477862358093,
      "learning_rate": 9.789169423468092e-06,
      "loss": 0.491,
      "step": 7251
    },
    {
      "epoch": 0.3327979441053646,
      "grad_norm": 0.4315713942050934,
      "learning_rate": 9.789098969694525e-06,
      "loss": 0.3301,
      "step": 7252
    },
    {
      "epoch": 0.33284383461061906,
      "grad_norm": 0.48701056838035583,
      "learning_rate": 9.789028504404685e-06,
      "loss": 0.3717,
      "step": 7253
    },
    {
      "epoch": 0.3328897251158735,
      "grad_norm": 0.4320523738861084,
      "learning_rate": 9.788958027598745e-06,
      "loss": 0.3022,
      "step": 7254
    },
    {
      "epoch": 0.332935615621128,
      "grad_norm": 0.4756210744380951,
      "learning_rate": 9.78888753927687e-06,
      "loss": 0.3696,
      "step": 7255
    },
    {
      "epoch": 0.33298150612638244,
      "grad_norm": 0.49892500042915344,
      "learning_rate": 9.788817039439233e-06,
      "loss": 0.4624,
      "step": 7256
    },
    {
      "epoch": 0.33302739663163694,
      "grad_norm": 0.5936259627342224,
      "learning_rate": 9.788746528086002e-06,
      "loss": 0.481,
      "step": 7257
    },
    {
      "epoch": 0.3330732871368914,
      "grad_norm": 0.48534366488456726,
      "learning_rate": 9.788676005217348e-06,
      "loss": 0.4262,
      "step": 7258
    },
    {
      "epoch": 0.3331191776421458,
      "grad_norm": 0.4955044388771057,
      "learning_rate": 9.788605470833438e-06,
      "loss": 0.398,
      "step": 7259
    },
    {
      "epoch": 0.3331650681474003,
      "grad_norm": 0.503099262714386,
      "learning_rate": 9.788534924934445e-06,
      "loss": 0.4667,
      "step": 7260
    },
    {
      "epoch": 0.33321095865265477,
      "grad_norm": 0.5332765579223633,
      "learning_rate": 9.788464367520536e-06,
      "loss": 0.4635,
      "step": 7261
    },
    {
      "epoch": 0.3332568491579092,
      "grad_norm": 0.4678882956504822,
      "learning_rate": 9.78839379859188e-06,
      "loss": 0.361,
      "step": 7262
    },
    {
      "epoch": 0.3333027396631637,
      "grad_norm": 0.49001142382621765,
      "learning_rate": 9.78832321814865e-06,
      "loss": 0.4849,
      "step": 7263
    },
    {
      "epoch": 0.33334863016841815,
      "grad_norm": 0.4585002064704895,
      "learning_rate": 9.788252626191013e-06,
      "loss": 0.4336,
      "step": 7264
    },
    {
      "epoch": 0.3333945206736726,
      "grad_norm": 0.430682897567749,
      "learning_rate": 9.788182022719141e-06,
      "loss": 0.3388,
      "step": 7265
    },
    {
      "epoch": 0.3334404111789271,
      "grad_norm": 0.5248414278030396,
      "learning_rate": 9.788111407733201e-06,
      "loss": 0.4567,
      "step": 7266
    },
    {
      "epoch": 0.33348630168418153,
      "grad_norm": 0.4665341377258301,
      "learning_rate": 9.788040781233364e-06,
      "loss": 0.407,
      "step": 7267
    },
    {
      "epoch": 0.333532192189436,
      "grad_norm": 0.4517952501773834,
      "learning_rate": 9.787970143219802e-06,
      "loss": 0.3562,
      "step": 7268
    },
    {
      "epoch": 0.33357808269469047,
      "grad_norm": 0.4285184442996979,
      "learning_rate": 9.787899493692681e-06,
      "loss": 0.3468,
      "step": 7269
    },
    {
      "epoch": 0.3336239731999449,
      "grad_norm": 0.4653485417366028,
      "learning_rate": 9.787828832652172e-06,
      "loss": 0.395,
      "step": 7270
    },
    {
      "epoch": 0.3336698637051994,
      "grad_norm": 0.49116840958595276,
      "learning_rate": 9.787758160098448e-06,
      "loss": 0.4869,
      "step": 7271
    },
    {
      "epoch": 0.33371575421045385,
      "grad_norm": 0.44094935059547424,
      "learning_rate": 9.787687476031678e-06,
      "loss": 0.37,
      "step": 7272
    },
    {
      "epoch": 0.3337616447157083,
      "grad_norm": 0.4531886875629425,
      "learning_rate": 9.787616780452027e-06,
      "loss": 0.3624,
      "step": 7273
    },
    {
      "epoch": 0.3338075352209628,
      "grad_norm": 0.49113717675209045,
      "learning_rate": 9.78754607335967e-06,
      "loss": 0.4738,
      "step": 7274
    },
    {
      "epoch": 0.33385342572621723,
      "grad_norm": 0.450723797082901,
      "learning_rate": 9.787475354754777e-06,
      "loss": 0.4091,
      "step": 7275
    },
    {
      "epoch": 0.33389931623147173,
      "grad_norm": 0.5710439682006836,
      "learning_rate": 9.787404624637515e-06,
      "loss": 0.4463,
      "step": 7276
    },
    {
      "epoch": 0.3339452067367262,
      "grad_norm": 0.4603712856769562,
      "learning_rate": 9.787333883008057e-06,
      "loss": 0.3562,
      "step": 7277
    },
    {
      "epoch": 0.3339910972419806,
      "grad_norm": 0.4705801010131836,
      "learning_rate": 9.78726312986657e-06,
      "loss": 0.3933,
      "step": 7278
    },
    {
      "epoch": 0.3340369877472351,
      "grad_norm": 0.5127866864204407,
      "learning_rate": 9.787192365213229e-06,
      "loss": 0.4299,
      "step": 7279
    },
    {
      "epoch": 0.33408287825248956,
      "grad_norm": 0.47350046038627625,
      "learning_rate": 9.7871215890482e-06,
      "loss": 0.401,
      "step": 7280
    },
    {
      "epoch": 0.334128768757744,
      "grad_norm": 0.47064611315727234,
      "learning_rate": 9.787050801371654e-06,
      "loss": 0.4301,
      "step": 7281
    },
    {
      "epoch": 0.3341746592629985,
      "grad_norm": 0.514356255531311,
      "learning_rate": 9.786980002183761e-06,
      "loss": 0.5029,
      "step": 7282
    },
    {
      "epoch": 0.33422054976825294,
      "grad_norm": 0.5017508268356323,
      "learning_rate": 9.786909191484692e-06,
      "loss": 0.4608,
      "step": 7283
    },
    {
      "epoch": 0.33426644027350744,
      "grad_norm": 0.48175665736198425,
      "learning_rate": 9.78683836927462e-06,
      "loss": 0.4587,
      "step": 7284
    },
    {
      "epoch": 0.3343123307787619,
      "grad_norm": 0.4773286283016205,
      "learning_rate": 9.78676753555371e-06,
      "loss": 0.4436,
      "step": 7285
    },
    {
      "epoch": 0.3343582212840163,
      "grad_norm": 0.4366253614425659,
      "learning_rate": 9.786696690322135e-06,
      "loss": 0.3282,
      "step": 7286
    },
    {
      "epoch": 0.3344041117892708,
      "grad_norm": 0.47915419936180115,
      "learning_rate": 9.786625833580066e-06,
      "loss": 0.4337,
      "step": 7287
    },
    {
      "epoch": 0.33445000229452526,
      "grad_norm": 0.47704559564590454,
      "learning_rate": 9.786554965327671e-06,
      "loss": 0.4453,
      "step": 7288
    },
    {
      "epoch": 0.3344958927997797,
      "grad_norm": 0.43671613931655884,
      "learning_rate": 9.786484085565123e-06,
      "loss": 0.3287,
      "step": 7289
    },
    {
      "epoch": 0.3345417833050342,
      "grad_norm": 0.4726767838001251,
      "learning_rate": 9.786413194292593e-06,
      "loss": 0.4294,
      "step": 7290
    },
    {
      "epoch": 0.33458767381028864,
      "grad_norm": 0.5310972929000854,
      "learning_rate": 9.786342291510248e-06,
      "loss": 0.5696,
      "step": 7291
    },
    {
      "epoch": 0.33463356431554314,
      "grad_norm": 0.5059199333190918,
      "learning_rate": 9.78627137721826e-06,
      "loss": 0.4087,
      "step": 7292
    },
    {
      "epoch": 0.3346794548207976,
      "grad_norm": 0.4523943066596985,
      "learning_rate": 9.7862004514168e-06,
      "loss": 0.4209,
      "step": 7293
    },
    {
      "epoch": 0.334725345326052,
      "grad_norm": 0.46609583497047424,
      "learning_rate": 9.78612951410604e-06,
      "loss": 0.4482,
      "step": 7294
    },
    {
      "epoch": 0.3347712358313065,
      "grad_norm": 0.46719783544540405,
      "learning_rate": 9.786058565286148e-06,
      "loss": 0.3592,
      "step": 7295
    },
    {
      "epoch": 0.33481712633656097,
      "grad_norm": 0.4765491783618927,
      "learning_rate": 9.785987604957296e-06,
      "loss": 0.372,
      "step": 7296
    },
    {
      "epoch": 0.3348630168418154,
      "grad_norm": 0.5152177214622498,
      "learning_rate": 9.785916633119655e-06,
      "loss": 0.4995,
      "step": 7297
    },
    {
      "epoch": 0.3349089073470699,
      "grad_norm": 0.4777143895626068,
      "learning_rate": 9.785845649773394e-06,
      "loss": 0.4153,
      "step": 7298
    },
    {
      "epoch": 0.33495479785232435,
      "grad_norm": 0.4916500151157379,
      "learning_rate": 9.785774654918684e-06,
      "loss": 0.5133,
      "step": 7299
    },
    {
      "epoch": 0.3350006883575788,
      "grad_norm": 0.44841763377189636,
      "learning_rate": 9.785703648555698e-06,
      "loss": 0.3407,
      "step": 7300
    },
    {
      "epoch": 0.3350465788628333,
      "grad_norm": 0.46499189734458923,
      "learning_rate": 9.785632630684604e-06,
      "loss": 0.3804,
      "step": 7301
    },
    {
      "epoch": 0.33509246936808773,
      "grad_norm": 0.48407605290412903,
      "learning_rate": 9.785561601305574e-06,
      "loss": 0.4472,
      "step": 7302
    },
    {
      "epoch": 0.33513835987334223,
      "grad_norm": 0.5582600235939026,
      "learning_rate": 9.78549056041878e-06,
      "loss": 0.5371,
      "step": 7303
    },
    {
      "epoch": 0.33518425037859667,
      "grad_norm": 0.46563172340393066,
      "learning_rate": 9.78541950802439e-06,
      "loss": 0.3987,
      "step": 7304
    },
    {
      "epoch": 0.3352301408838511,
      "grad_norm": 0.4615235924720764,
      "learning_rate": 9.785348444122577e-06,
      "loss": 0.4229,
      "step": 7305
    },
    {
      "epoch": 0.3352760313891056,
      "grad_norm": 0.4656515419483185,
      "learning_rate": 9.785277368713511e-06,
      "loss": 0.4111,
      "step": 7306
    },
    {
      "epoch": 0.33532192189436005,
      "grad_norm": 0.49970924854278564,
      "learning_rate": 9.785206281797364e-06,
      "loss": 0.43,
      "step": 7307
    },
    {
      "epoch": 0.3353678123996145,
      "grad_norm": 0.4648034870624542,
      "learning_rate": 9.785135183374305e-06,
      "loss": 0.4315,
      "step": 7308
    },
    {
      "epoch": 0.335413702904869,
      "grad_norm": 0.4694409966468811,
      "learning_rate": 9.785064073444506e-06,
      "loss": 0.4015,
      "step": 7309
    },
    {
      "epoch": 0.33545959341012344,
      "grad_norm": 0.4168854355812073,
      "learning_rate": 9.78499295200814e-06,
      "loss": 0.3232,
      "step": 7310
    },
    {
      "epoch": 0.33550548391537793,
      "grad_norm": 0.4731890857219696,
      "learning_rate": 9.784921819065374e-06,
      "loss": 0.3618,
      "step": 7311
    },
    {
      "epoch": 0.3355513744206324,
      "grad_norm": 0.47363045811653137,
      "learning_rate": 9.784850674616382e-06,
      "loss": 0.4217,
      "step": 7312
    },
    {
      "epoch": 0.3355972649258868,
      "grad_norm": 0.5044989585876465,
      "learning_rate": 9.784779518661334e-06,
      "loss": 0.4715,
      "step": 7313
    },
    {
      "epoch": 0.3356431554311413,
      "grad_norm": 0.44733887910842896,
      "learning_rate": 9.784708351200402e-06,
      "loss": 0.3689,
      "step": 7314
    },
    {
      "epoch": 0.33568904593639576,
      "grad_norm": 0.45569586753845215,
      "learning_rate": 9.784637172233757e-06,
      "loss": 0.4305,
      "step": 7315
    },
    {
      "epoch": 0.3357349364416502,
      "grad_norm": 0.470212459564209,
      "learning_rate": 9.78456598176157e-06,
      "loss": 0.4393,
      "step": 7316
    },
    {
      "epoch": 0.3357808269469047,
      "grad_norm": 0.45726487040519714,
      "learning_rate": 9.784494779784011e-06,
      "loss": 0.4165,
      "step": 7317
    },
    {
      "epoch": 0.33582671745215914,
      "grad_norm": 0.45731082558631897,
      "learning_rate": 9.784423566301252e-06,
      "loss": 0.4185,
      "step": 7318
    },
    {
      "epoch": 0.33587260795741364,
      "grad_norm": 0.46821147203445435,
      "learning_rate": 9.784352341313464e-06,
      "loss": 0.4396,
      "step": 7319
    },
    {
      "epoch": 0.3359184984626681,
      "grad_norm": 0.4776060879230499,
      "learning_rate": 9.784281104820819e-06,
      "loss": 0.4332,
      "step": 7320
    },
    {
      "epoch": 0.3359643889679225,
      "grad_norm": 0.43097859621047974,
      "learning_rate": 9.784209856823486e-06,
      "loss": 0.3142,
      "step": 7321
    },
    {
      "epoch": 0.336010279473177,
      "grad_norm": 0.4558699131011963,
      "learning_rate": 9.784138597321641e-06,
      "loss": 0.3547,
      "step": 7322
    },
    {
      "epoch": 0.33605616997843146,
      "grad_norm": 0.4798987805843353,
      "learning_rate": 9.784067326315451e-06,
      "loss": 0.4673,
      "step": 7323
    },
    {
      "epoch": 0.3361020604836859,
      "grad_norm": 0.45970186591148376,
      "learning_rate": 9.78399604380509e-06,
      "loss": 0.3984,
      "step": 7324
    },
    {
      "epoch": 0.3361479509889404,
      "grad_norm": 0.4498862326145172,
      "learning_rate": 9.783924749790728e-06,
      "loss": 0.3628,
      "step": 7325
    },
    {
      "epoch": 0.33619384149419484,
      "grad_norm": 0.4933110475540161,
      "learning_rate": 9.783853444272538e-06,
      "loss": 0.4336,
      "step": 7326
    },
    {
      "epoch": 0.3362397319994493,
      "grad_norm": 0.4886942505836487,
      "learning_rate": 9.783782127250686e-06,
      "loss": 0.502,
      "step": 7327
    },
    {
      "epoch": 0.3362856225047038,
      "grad_norm": 0.4729424715042114,
      "learning_rate": 9.783710798725352e-06,
      "loss": 0.4605,
      "step": 7328
    },
    {
      "epoch": 0.3363315130099582,
      "grad_norm": 0.5042920112609863,
      "learning_rate": 9.783639458696704e-06,
      "loss": 0.4152,
      "step": 7329
    },
    {
      "epoch": 0.3363774035152127,
      "grad_norm": 0.4635804295539856,
      "learning_rate": 9.78356810716491e-06,
      "loss": 0.381,
      "step": 7330
    },
    {
      "epoch": 0.33642329402046717,
      "grad_norm": 0.5181156992912292,
      "learning_rate": 9.783496744130145e-06,
      "loss": 0.5257,
      "step": 7331
    },
    {
      "epoch": 0.3364691845257216,
      "grad_norm": 0.4616307318210602,
      "learning_rate": 9.783425369592581e-06,
      "loss": 0.4525,
      "step": 7332
    },
    {
      "epoch": 0.3365150750309761,
      "grad_norm": 0.4939948320388794,
      "learning_rate": 9.783353983552388e-06,
      "loss": 0.4589,
      "step": 7333
    },
    {
      "epoch": 0.33656096553623055,
      "grad_norm": 0.4589994251728058,
      "learning_rate": 9.783282586009738e-06,
      "loss": 0.4259,
      "step": 7334
    },
    {
      "epoch": 0.336606856041485,
      "grad_norm": 0.45634517073631287,
      "learning_rate": 9.783211176964805e-06,
      "loss": 0.3807,
      "step": 7335
    },
    {
      "epoch": 0.3366527465467395,
      "grad_norm": 0.46722501516342163,
      "learning_rate": 9.783139756417757e-06,
      "loss": 0.4704,
      "step": 7336
    },
    {
      "epoch": 0.33669863705199393,
      "grad_norm": 0.4854772984981537,
      "learning_rate": 9.783068324368767e-06,
      "loss": 0.4293,
      "step": 7337
    },
    {
      "epoch": 0.33674452755724843,
      "grad_norm": 0.45999211072921753,
      "learning_rate": 9.782996880818008e-06,
      "loss": 0.3623,
      "step": 7338
    },
    {
      "epoch": 0.33679041806250287,
      "grad_norm": 0.4588179588317871,
      "learning_rate": 9.782925425765652e-06,
      "loss": 0.3652,
      "step": 7339
    },
    {
      "epoch": 0.3368363085677573,
      "grad_norm": 0.46099618077278137,
      "learning_rate": 9.782853959211869e-06,
      "loss": 0.4297,
      "step": 7340
    },
    {
      "epoch": 0.3368821990730118,
      "grad_norm": 0.4382844567298889,
      "learning_rate": 9.782782481156832e-06,
      "loss": 0.3752,
      "step": 7341
    },
    {
      "epoch": 0.33692808957826625,
      "grad_norm": 0.4364856481552124,
      "learning_rate": 9.782710991600712e-06,
      "loss": 0.3582,
      "step": 7342
    },
    {
      "epoch": 0.3369739800835207,
      "grad_norm": 0.5402190089225769,
      "learning_rate": 9.782639490543682e-06,
      "loss": 0.5645,
      "step": 7343
    },
    {
      "epoch": 0.3370198705887752,
      "grad_norm": 0.5003296732902527,
      "learning_rate": 9.782567977985913e-06,
      "loss": 0.4358,
      "step": 7344
    },
    {
      "epoch": 0.33706576109402964,
      "grad_norm": 0.46167850494384766,
      "learning_rate": 9.782496453927578e-06,
      "loss": 0.3983,
      "step": 7345
    },
    {
      "epoch": 0.33711165159928413,
      "grad_norm": 0.4567517042160034,
      "learning_rate": 9.78242491836885e-06,
      "loss": 0.3937,
      "step": 7346
    },
    {
      "epoch": 0.3371575421045386,
      "grad_norm": 0.4671885371208191,
      "learning_rate": 9.782353371309896e-06,
      "loss": 0.391,
      "step": 7347
    },
    {
      "epoch": 0.337203432609793,
      "grad_norm": 0.4809706509113312,
      "learning_rate": 9.782281812750894e-06,
      "loss": 0.4286,
      "step": 7348
    },
    {
      "epoch": 0.3372493231150475,
      "grad_norm": 0.4873986840248108,
      "learning_rate": 9.782210242692012e-06,
      "loss": 0.4265,
      "step": 7349
    },
    {
      "epoch": 0.33729521362030196,
      "grad_norm": 0.46643853187561035,
      "learning_rate": 9.782138661133425e-06,
      "loss": 0.3784,
      "step": 7350
    },
    {
      "epoch": 0.3373411041255564,
      "grad_norm": 0.4696289598941803,
      "learning_rate": 9.782067068075303e-06,
      "loss": 0.3923,
      "step": 7351
    },
    {
      "epoch": 0.3373869946308109,
      "grad_norm": 0.4900300204753876,
      "learning_rate": 9.78199546351782e-06,
      "loss": 0.408,
      "step": 7352
    },
    {
      "epoch": 0.33743288513606534,
      "grad_norm": 0.4681951403617859,
      "learning_rate": 9.781923847461146e-06,
      "loss": 0.3547,
      "step": 7353
    },
    {
      "epoch": 0.3374787756413198,
      "grad_norm": 0.5273519158363342,
      "learning_rate": 9.781852219905454e-06,
      "loss": 0.5323,
      "step": 7354
    },
    {
      "epoch": 0.3375246661465743,
      "grad_norm": 0.4322478771209717,
      "learning_rate": 9.781780580850918e-06,
      "loss": 0.3641,
      "step": 7355
    },
    {
      "epoch": 0.3375705566518287,
      "grad_norm": 0.482887327671051,
      "learning_rate": 9.781708930297707e-06,
      "loss": 0.4685,
      "step": 7356
    },
    {
      "epoch": 0.3376164471570832,
      "grad_norm": 0.4598298966884613,
      "learning_rate": 9.781637268245998e-06,
      "loss": 0.435,
      "step": 7357
    },
    {
      "epoch": 0.33766233766233766,
      "grad_norm": 0.46907344460487366,
      "learning_rate": 9.781565594695958e-06,
      "loss": 0.4061,
      "step": 7358
    },
    {
      "epoch": 0.3377082281675921,
      "grad_norm": 0.4659236669540405,
      "learning_rate": 9.781493909647763e-06,
      "loss": 0.3572,
      "step": 7359
    },
    {
      "epoch": 0.3377541186728466,
      "grad_norm": 0.4964559078216553,
      "learning_rate": 9.781422213101583e-06,
      "loss": 0.4637,
      "step": 7360
    },
    {
      "epoch": 0.33780000917810105,
      "grad_norm": 0.4614852964878082,
      "learning_rate": 9.781350505057594e-06,
      "loss": 0.4129,
      "step": 7361
    },
    {
      "epoch": 0.3378458996833555,
      "grad_norm": 0.4575325548648834,
      "learning_rate": 9.781278785515964e-06,
      "loss": 0.3943,
      "step": 7362
    },
    {
      "epoch": 0.33789179018861,
      "grad_norm": 0.4790130853652954,
      "learning_rate": 9.781207054476868e-06,
      "loss": 0.3808,
      "step": 7363
    },
    {
      "epoch": 0.33793768069386443,
      "grad_norm": 0.47188782691955566,
      "learning_rate": 9.781135311940479e-06,
      "loss": 0.4109,
      "step": 7364
    },
    {
      "epoch": 0.3379835711991189,
      "grad_norm": 0.4460643529891968,
      "learning_rate": 9.781063557906967e-06,
      "loss": 0.384,
      "step": 7365
    },
    {
      "epoch": 0.33802946170437337,
      "grad_norm": 0.4549757242202759,
      "learning_rate": 9.780991792376507e-06,
      "loss": 0.3717,
      "step": 7366
    },
    {
      "epoch": 0.3380753522096278,
      "grad_norm": 0.4437856674194336,
      "learning_rate": 9.780920015349272e-06,
      "loss": 0.28,
      "step": 7367
    },
    {
      "epoch": 0.3381212427148823,
      "grad_norm": 0.48330235481262207,
      "learning_rate": 9.780848226825431e-06,
      "loss": 0.4135,
      "step": 7368
    },
    {
      "epoch": 0.33816713322013675,
      "grad_norm": 0.46891406178474426,
      "learning_rate": 9.78077642680516e-06,
      "loss": 0.3988,
      "step": 7369
    },
    {
      "epoch": 0.3382130237253912,
      "grad_norm": 0.5104764699935913,
      "learning_rate": 9.78070461528863e-06,
      "loss": 0.4929,
      "step": 7370
    },
    {
      "epoch": 0.3382589142306457,
      "grad_norm": 0.47054728865623474,
      "learning_rate": 9.780632792276013e-06,
      "loss": 0.4311,
      "step": 7371
    },
    {
      "epoch": 0.33830480473590013,
      "grad_norm": 0.463750958442688,
      "learning_rate": 9.780560957767485e-06,
      "loss": 0.3682,
      "step": 7372
    },
    {
      "epoch": 0.33835069524115463,
      "grad_norm": 0.48049643635749817,
      "learning_rate": 9.780489111763216e-06,
      "loss": 0.4498,
      "step": 7373
    },
    {
      "epoch": 0.3383965857464091,
      "grad_norm": 0.46395596861839294,
      "learning_rate": 9.780417254263379e-06,
      "loss": 0.3965,
      "step": 7374
    },
    {
      "epoch": 0.3384424762516635,
      "grad_norm": 0.48248291015625,
      "learning_rate": 9.780345385268149e-06,
      "loss": 0.3944,
      "step": 7375
    },
    {
      "epoch": 0.338488366756918,
      "grad_norm": 0.4716425836086273,
      "learning_rate": 9.780273504777695e-06,
      "loss": 0.3583,
      "step": 7376
    },
    {
      "epoch": 0.33853425726217246,
      "grad_norm": 0.4617311358451843,
      "learning_rate": 9.780201612792192e-06,
      "loss": 0.4188,
      "step": 7377
    },
    {
      "epoch": 0.3385801477674269,
      "grad_norm": 0.4742300808429718,
      "learning_rate": 9.780129709311815e-06,
      "loss": 0.3849,
      "step": 7378
    },
    {
      "epoch": 0.3386260382726814,
      "grad_norm": 0.4515502154827118,
      "learning_rate": 9.780057794336733e-06,
      "loss": 0.4069,
      "step": 7379
    },
    {
      "epoch": 0.33867192877793584,
      "grad_norm": 0.4433266818523407,
      "learning_rate": 9.77998586786712e-06,
      "loss": 0.3805,
      "step": 7380
    },
    {
      "epoch": 0.33871781928319034,
      "grad_norm": 0.4708617329597473,
      "learning_rate": 9.77991392990315e-06,
      "loss": 0.3756,
      "step": 7381
    },
    {
      "epoch": 0.3387637097884448,
      "grad_norm": 0.4890288710594177,
      "learning_rate": 9.779841980444997e-06,
      "loss": 0.4572,
      "step": 7382
    },
    {
      "epoch": 0.3388096002936992,
      "grad_norm": 0.4627659320831299,
      "learning_rate": 9.779770019492832e-06,
      "loss": 0.3503,
      "step": 7383
    },
    {
      "epoch": 0.3388554907989537,
      "grad_norm": 0.46248921751976013,
      "learning_rate": 9.779698047046829e-06,
      "loss": 0.3711,
      "step": 7384
    },
    {
      "epoch": 0.33890138130420816,
      "grad_norm": 0.45156726241111755,
      "learning_rate": 9.77962606310716e-06,
      "loss": 0.4164,
      "step": 7385
    },
    {
      "epoch": 0.3389472718094626,
      "grad_norm": 0.5241009593009949,
      "learning_rate": 9.779554067673998e-06,
      "loss": 0.4313,
      "step": 7386
    },
    {
      "epoch": 0.3389931623147171,
      "grad_norm": 0.4329284727573395,
      "learning_rate": 9.779482060747518e-06,
      "loss": 0.351,
      "step": 7387
    },
    {
      "epoch": 0.33903905281997154,
      "grad_norm": 0.472797691822052,
      "learning_rate": 9.779410042327892e-06,
      "loss": 0.4198,
      "step": 7388
    },
    {
      "epoch": 0.339084943325226,
      "grad_norm": 0.468295693397522,
      "learning_rate": 9.779338012415294e-06,
      "loss": 0.3968,
      "step": 7389
    },
    {
      "epoch": 0.3391308338304805,
      "grad_norm": 0.43124014139175415,
      "learning_rate": 9.779265971009894e-06,
      "loss": 0.3049,
      "step": 7390
    },
    {
      "epoch": 0.3391767243357349,
      "grad_norm": 0.4837438762187958,
      "learning_rate": 9.779193918111872e-06,
      "loss": 0.4124,
      "step": 7391
    },
    {
      "epoch": 0.3392226148409894,
      "grad_norm": 0.4602155089378357,
      "learning_rate": 9.779121853721392e-06,
      "loss": 0.4022,
      "step": 7392
    },
    {
      "epoch": 0.33926850534624386,
      "grad_norm": 0.4885324239730835,
      "learning_rate": 9.779049777838636e-06,
      "loss": 0.5222,
      "step": 7393
    },
    {
      "epoch": 0.3393143958514983,
      "grad_norm": 0.43079251050949097,
      "learning_rate": 9.778977690463772e-06,
      "loss": 0.3498,
      "step": 7394
    },
    {
      "epoch": 0.3393602863567528,
      "grad_norm": 0.4765048921108246,
      "learning_rate": 9.778905591596975e-06,
      "loss": 0.4488,
      "step": 7395
    },
    {
      "epoch": 0.33940617686200725,
      "grad_norm": 0.48552173376083374,
      "learning_rate": 9.778833481238419e-06,
      "loss": 0.4377,
      "step": 7396
    },
    {
      "epoch": 0.3394520673672617,
      "grad_norm": 0.4365825951099396,
      "learning_rate": 9.778761359388275e-06,
      "loss": 0.4156,
      "step": 7397
    },
    {
      "epoch": 0.3394979578725162,
      "grad_norm": 0.4548175632953644,
      "learning_rate": 9.77868922604672e-06,
      "loss": 0.376,
      "step": 7398
    },
    {
      "epoch": 0.33954384837777063,
      "grad_norm": 0.49979400634765625,
      "learning_rate": 9.778617081213925e-06,
      "loss": 0.4958,
      "step": 7399
    },
    {
      "epoch": 0.3395897388830251,
      "grad_norm": 0.47651001811027527,
      "learning_rate": 9.778544924890064e-06,
      "loss": 0.4356,
      "step": 7400
    },
    {
      "epoch": 0.33963562938827957,
      "grad_norm": 0.4891176223754883,
      "learning_rate": 9.77847275707531e-06,
      "loss": 0.4214,
      "step": 7401
    },
    {
      "epoch": 0.339681519893534,
      "grad_norm": 0.47928473353385925,
      "learning_rate": 9.77840057776984e-06,
      "loss": 0.4383,
      "step": 7402
    },
    {
      "epoch": 0.3397274103987885,
      "grad_norm": 0.5314728021621704,
      "learning_rate": 9.778328386973821e-06,
      "loss": 0.6025,
      "step": 7403
    },
    {
      "epoch": 0.33977330090404295,
      "grad_norm": 0.46432963013648987,
      "learning_rate": 9.778256184687432e-06,
      "loss": 0.4022,
      "step": 7404
    },
    {
      "epoch": 0.3398191914092974,
      "grad_norm": 0.42920351028442383,
      "learning_rate": 9.778183970910846e-06,
      "loss": 0.3159,
      "step": 7405
    },
    {
      "epoch": 0.3398650819145519,
      "grad_norm": 0.4578544497489929,
      "learning_rate": 9.778111745644234e-06,
      "loss": 0.4192,
      "step": 7406
    },
    {
      "epoch": 0.33991097241980633,
      "grad_norm": 0.5304063558578491,
      "learning_rate": 9.778039508887771e-06,
      "loss": 0.4783,
      "step": 7407
    },
    {
      "epoch": 0.33995686292506083,
      "grad_norm": 0.4821876585483551,
      "learning_rate": 9.777967260641631e-06,
      "loss": 0.4547,
      "step": 7408
    },
    {
      "epoch": 0.3400027534303153,
      "grad_norm": 0.42716121673583984,
      "learning_rate": 9.777895000905988e-06,
      "loss": 0.3188,
      "step": 7409
    },
    {
      "epoch": 0.3400486439355697,
      "grad_norm": 0.47628113627433777,
      "learning_rate": 9.777822729681017e-06,
      "loss": 0.4419,
      "step": 7410
    },
    {
      "epoch": 0.3400945344408242,
      "grad_norm": 0.48569783568382263,
      "learning_rate": 9.777750446966889e-06,
      "loss": 0.4191,
      "step": 7411
    },
    {
      "epoch": 0.34014042494607866,
      "grad_norm": 0.49629464745521545,
      "learning_rate": 9.777678152763779e-06,
      "loss": 0.4734,
      "step": 7412
    },
    {
      "epoch": 0.3401863154513331,
      "grad_norm": 0.4593360722064972,
      "learning_rate": 9.77760584707186e-06,
      "loss": 0.3641,
      "step": 7413
    },
    {
      "epoch": 0.3402322059565876,
      "grad_norm": 0.4248492419719696,
      "learning_rate": 9.777533529891309e-06,
      "loss": 0.3361,
      "step": 7414
    },
    {
      "epoch": 0.34027809646184204,
      "grad_norm": 0.4597489535808563,
      "learning_rate": 9.777461201222296e-06,
      "loss": 0.3887,
      "step": 7415
    },
    {
      "epoch": 0.3403239869670965,
      "grad_norm": 0.45317384600639343,
      "learning_rate": 9.777388861064998e-06,
      "loss": 0.3714,
      "step": 7416
    },
    {
      "epoch": 0.340369877472351,
      "grad_norm": 0.4567270576953888,
      "learning_rate": 9.777316509419586e-06,
      "loss": 0.4214,
      "step": 7417
    },
    {
      "epoch": 0.3404157679776054,
      "grad_norm": 0.4635644853115082,
      "learning_rate": 9.777244146286236e-06,
      "loss": 0.403,
      "step": 7418
    },
    {
      "epoch": 0.3404616584828599,
      "grad_norm": 0.44588348269462585,
      "learning_rate": 9.777171771665122e-06,
      "loss": 0.3936,
      "step": 7419
    },
    {
      "epoch": 0.34050754898811436,
      "grad_norm": 0.4618290066719055,
      "learning_rate": 9.777099385556418e-06,
      "loss": 0.3959,
      "step": 7420
    },
    {
      "epoch": 0.3405534394933688,
      "grad_norm": 0.4413503408432007,
      "learning_rate": 9.777026987960296e-06,
      "loss": 0.3685,
      "step": 7421
    },
    {
      "epoch": 0.3405993299986233,
      "grad_norm": 0.4958931505680084,
      "learning_rate": 9.776954578876933e-06,
      "loss": 0.5016,
      "step": 7422
    },
    {
      "epoch": 0.34064522050387774,
      "grad_norm": 0.4384816586971283,
      "learning_rate": 9.776882158306502e-06,
      "loss": 0.326,
      "step": 7423
    },
    {
      "epoch": 0.3406911110091322,
      "grad_norm": 0.4481096863746643,
      "learning_rate": 9.776809726249177e-06,
      "loss": 0.3409,
      "step": 7424
    },
    {
      "epoch": 0.3407370015143867,
      "grad_norm": 0.5002566576004028,
      "learning_rate": 9.776737282705134e-06,
      "loss": 0.4446,
      "step": 7425
    },
    {
      "epoch": 0.3407828920196411,
      "grad_norm": 0.49928605556488037,
      "learning_rate": 9.776664827674542e-06,
      "loss": 0.5451,
      "step": 7426
    },
    {
      "epoch": 0.3408287825248956,
      "grad_norm": 0.4513339102268219,
      "learning_rate": 9.77659236115758e-06,
      "loss": 0.4233,
      "step": 7427
    },
    {
      "epoch": 0.34087467303015007,
      "grad_norm": 0.47260984778404236,
      "learning_rate": 9.77651988315442e-06,
      "loss": 0.3183,
      "step": 7428
    },
    {
      "epoch": 0.3409205635354045,
      "grad_norm": 0.4430692195892334,
      "learning_rate": 9.776447393665239e-06,
      "loss": 0.3944,
      "step": 7429
    },
    {
      "epoch": 0.340966454040659,
      "grad_norm": 0.5236510038375854,
      "learning_rate": 9.776374892690209e-06,
      "loss": 0.4704,
      "step": 7430
    },
    {
      "epoch": 0.34101234454591345,
      "grad_norm": 0.5233997106552124,
      "learning_rate": 9.776302380229506e-06,
      "loss": 0.4617,
      "step": 7431
    },
    {
      "epoch": 0.3410582350511679,
      "grad_norm": 0.472449392080307,
      "learning_rate": 9.776229856283302e-06,
      "loss": 0.3668,
      "step": 7432
    },
    {
      "epoch": 0.3411041255564224,
      "grad_norm": 0.4406496286392212,
      "learning_rate": 9.776157320851771e-06,
      "loss": 0.3931,
      "step": 7433
    },
    {
      "epoch": 0.34115001606167683,
      "grad_norm": 0.4443661570549011,
      "learning_rate": 9.77608477393509e-06,
      "loss": 0.3288,
      "step": 7434
    },
    {
      "epoch": 0.34119590656693133,
      "grad_norm": 0.43557700514793396,
      "learning_rate": 9.776012215533434e-06,
      "loss": 0.3403,
      "step": 7435
    },
    {
      "epoch": 0.34124179707218577,
      "grad_norm": 0.48722273111343384,
      "learning_rate": 9.775939645646975e-06,
      "loss": 0.4342,
      "step": 7436
    },
    {
      "epoch": 0.3412876875774402,
      "grad_norm": 0.43956923484802246,
      "learning_rate": 9.775867064275888e-06,
      "loss": 0.3229,
      "step": 7437
    },
    {
      "epoch": 0.3413335780826947,
      "grad_norm": 0.4262770712375641,
      "learning_rate": 9.775794471420348e-06,
      "loss": 0.3336,
      "step": 7438
    },
    {
      "epoch": 0.34137946858794915,
      "grad_norm": 0.4993220567703247,
      "learning_rate": 9.77572186708053e-06,
      "loss": 0.4723,
      "step": 7439
    },
    {
      "epoch": 0.3414253590932036,
      "grad_norm": 0.4437876343727112,
      "learning_rate": 9.775649251256609e-06,
      "loss": 0.4048,
      "step": 7440
    },
    {
      "epoch": 0.3414712495984581,
      "grad_norm": 0.4749569892883301,
      "learning_rate": 9.775576623948758e-06,
      "loss": 0.3839,
      "step": 7441
    },
    {
      "epoch": 0.34151714010371254,
      "grad_norm": 0.4711102843284607,
      "learning_rate": 9.775503985157153e-06,
      "loss": 0.446,
      "step": 7442
    },
    {
      "epoch": 0.341563030608967,
      "grad_norm": 0.5104339718818665,
      "learning_rate": 9.775431334881966e-06,
      "loss": 0.5018,
      "step": 7443
    },
    {
      "epoch": 0.3416089211142215,
      "grad_norm": 0.5113730430603027,
      "learning_rate": 9.775358673123376e-06,
      "loss": 0.5161,
      "step": 7444
    },
    {
      "epoch": 0.3416548116194759,
      "grad_norm": 0.5042397379875183,
      "learning_rate": 9.775285999881554e-06,
      "loss": 0.4196,
      "step": 7445
    },
    {
      "epoch": 0.3417007021247304,
      "grad_norm": 0.4670191705226898,
      "learning_rate": 9.775213315156677e-06,
      "loss": 0.3446,
      "step": 7446
    },
    {
      "epoch": 0.34174659262998486,
      "grad_norm": 0.49151450395584106,
      "learning_rate": 9.77514061894892e-06,
      "loss": 0.4573,
      "step": 7447
    },
    {
      "epoch": 0.3417924831352393,
      "grad_norm": 0.4387458264827728,
      "learning_rate": 9.775067911258455e-06,
      "loss": 0.4224,
      "step": 7448
    },
    {
      "epoch": 0.3418383736404938,
      "grad_norm": 0.4623500108718872,
      "learning_rate": 9.774995192085461e-06,
      "loss": 0.3767,
      "step": 7449
    },
    {
      "epoch": 0.34188426414574824,
      "grad_norm": 0.46996501088142395,
      "learning_rate": 9.77492246143011e-06,
      "loss": 0.4493,
      "step": 7450
    },
    {
      "epoch": 0.3419301546510027,
      "grad_norm": 0.4974955916404724,
      "learning_rate": 9.774849719292577e-06,
      "loss": 0.4775,
      "step": 7451
    },
    {
      "epoch": 0.3419760451562572,
      "grad_norm": 0.5088875889778137,
      "learning_rate": 9.774776965673036e-06,
      "loss": 0.4994,
      "step": 7452
    },
    {
      "epoch": 0.3420219356615116,
      "grad_norm": 0.49207568168640137,
      "learning_rate": 9.774704200571665e-06,
      "loss": 0.4823,
      "step": 7453
    },
    {
      "epoch": 0.3420678261667661,
      "grad_norm": 0.48566123843193054,
      "learning_rate": 9.774631423988638e-06,
      "loss": 0.3988,
      "step": 7454
    },
    {
      "epoch": 0.34211371667202056,
      "grad_norm": 0.4807029962539673,
      "learning_rate": 9.774558635924127e-06,
      "loss": 0.435,
      "step": 7455
    },
    {
      "epoch": 0.342159607177275,
      "grad_norm": 0.467380166053772,
      "learning_rate": 9.774485836378311e-06,
      "loss": 0.3394,
      "step": 7456
    },
    {
      "epoch": 0.3422054976825295,
      "grad_norm": 0.459309458732605,
      "learning_rate": 9.774413025351364e-06,
      "loss": 0.4007,
      "step": 7457
    },
    {
      "epoch": 0.34225138818778394,
      "grad_norm": 0.4063344895839691,
      "learning_rate": 9.774340202843459e-06,
      "loss": 0.3032,
      "step": 7458
    },
    {
      "epoch": 0.3422972786930384,
      "grad_norm": 0.4463087320327759,
      "learning_rate": 9.774267368854775e-06,
      "loss": 0.3259,
      "step": 7459
    },
    {
      "epoch": 0.3423431691982929,
      "grad_norm": 0.4934060871601105,
      "learning_rate": 9.774194523385482e-06,
      "loss": 0.4326,
      "step": 7460
    },
    {
      "epoch": 0.3423890597035473,
      "grad_norm": 0.4699383080005646,
      "learning_rate": 9.77412166643576e-06,
      "loss": 0.4409,
      "step": 7461
    },
    {
      "epoch": 0.3424349502088018,
      "grad_norm": 0.45647281408309937,
      "learning_rate": 9.77404879800578e-06,
      "loss": 0.3772,
      "step": 7462
    },
    {
      "epoch": 0.34248084071405627,
      "grad_norm": 0.44446781277656555,
      "learning_rate": 9.773975918095722e-06,
      "loss": 0.4095,
      "step": 7463
    },
    {
      "epoch": 0.3425267312193107,
      "grad_norm": 0.4419586956501007,
      "learning_rate": 9.773903026705756e-06,
      "loss": 0.3584,
      "step": 7464
    },
    {
      "epoch": 0.3425726217245652,
      "grad_norm": 0.46052002906799316,
      "learning_rate": 9.773830123836061e-06,
      "loss": 0.3994,
      "step": 7465
    },
    {
      "epoch": 0.34261851222981965,
      "grad_norm": 0.5097686648368835,
      "learning_rate": 9.773757209486811e-06,
      "loss": 0.4763,
      "step": 7466
    },
    {
      "epoch": 0.3426644027350741,
      "grad_norm": 0.4929233193397522,
      "learning_rate": 9.773684283658182e-06,
      "loss": 0.4249,
      "step": 7467
    },
    {
      "epoch": 0.3427102932403286,
      "grad_norm": 0.4204244017601013,
      "learning_rate": 9.773611346350348e-06,
      "loss": 0.3391,
      "step": 7468
    },
    {
      "epoch": 0.34275618374558303,
      "grad_norm": 0.6642082929611206,
      "learning_rate": 9.773538397563487e-06,
      "loss": 0.4263,
      "step": 7469
    },
    {
      "epoch": 0.3428020742508375,
      "grad_norm": 0.470769464969635,
      "learning_rate": 9.773465437297772e-06,
      "loss": 0.3931,
      "step": 7470
    },
    {
      "epoch": 0.34284796475609197,
      "grad_norm": 0.510100245475769,
      "learning_rate": 9.773392465553379e-06,
      "loss": 0.4977,
      "step": 7471
    },
    {
      "epoch": 0.3428938552613464,
      "grad_norm": 0.4964597821235657,
      "learning_rate": 9.773319482330483e-06,
      "loss": 0.4551,
      "step": 7472
    },
    {
      "epoch": 0.3429397457666009,
      "grad_norm": 0.5028560161590576,
      "learning_rate": 9.773246487629262e-06,
      "loss": 0.4151,
      "step": 7473
    },
    {
      "epoch": 0.34298563627185535,
      "grad_norm": 0.5038970708847046,
      "learning_rate": 9.77317348144989e-06,
      "loss": 0.3966,
      "step": 7474
    },
    {
      "epoch": 0.3430315267771098,
      "grad_norm": 0.5261569023132324,
      "learning_rate": 9.77310046379254e-06,
      "loss": 0.4499,
      "step": 7475
    },
    {
      "epoch": 0.3430774172823643,
      "grad_norm": 0.4408392906188965,
      "learning_rate": 9.77302743465739e-06,
      "loss": 0.3591,
      "step": 7476
    },
    {
      "epoch": 0.34312330778761874,
      "grad_norm": 0.4895802140235901,
      "learning_rate": 9.772954394044616e-06,
      "loss": 0.4207,
      "step": 7477
    },
    {
      "epoch": 0.3431691982928732,
      "grad_norm": 0.6594173312187195,
      "learning_rate": 9.772881341954393e-06,
      "loss": 0.4924,
      "step": 7478
    },
    {
      "epoch": 0.3432150887981277,
      "grad_norm": 0.46530622243881226,
      "learning_rate": 9.772808278386896e-06,
      "loss": 0.4068,
      "step": 7479
    },
    {
      "epoch": 0.3432609793033821,
      "grad_norm": 0.4976832866668701,
      "learning_rate": 9.772735203342303e-06,
      "loss": 0.4494,
      "step": 7480
    },
    {
      "epoch": 0.3433068698086366,
      "grad_norm": 0.4775434732437134,
      "learning_rate": 9.772662116820788e-06,
      "loss": 0.4023,
      "step": 7481
    },
    {
      "epoch": 0.34335276031389106,
      "grad_norm": 0.5039447546005249,
      "learning_rate": 9.772589018822524e-06,
      "loss": 0.4189,
      "step": 7482
    },
    {
      "epoch": 0.3433986508191455,
      "grad_norm": 0.535153865814209,
      "learning_rate": 9.772515909347694e-06,
      "loss": 0.5446,
      "step": 7483
    },
    {
      "epoch": 0.3434445413244,
      "grad_norm": 0.472126305103302,
      "learning_rate": 9.772442788396466e-06,
      "loss": 0.4248,
      "step": 7484
    },
    {
      "epoch": 0.34349043182965444,
      "grad_norm": 0.483585387468338,
      "learning_rate": 9.772369655969021e-06,
      "loss": 0.4117,
      "step": 7485
    },
    {
      "epoch": 0.3435363223349089,
      "grad_norm": 0.4720342755317688,
      "learning_rate": 9.772296512065534e-06,
      "loss": 0.4871,
      "step": 7486
    },
    {
      "epoch": 0.3435822128401634,
      "grad_norm": 0.4842703640460968,
      "learning_rate": 9.77222335668618e-06,
      "loss": 0.4475,
      "step": 7487
    },
    {
      "epoch": 0.3436281033454178,
      "grad_norm": 0.44866934418678284,
      "learning_rate": 9.772150189831133e-06,
      "loss": 0.3661,
      "step": 7488
    },
    {
      "epoch": 0.3436739938506723,
      "grad_norm": 0.4554039239883423,
      "learning_rate": 9.77207701150057e-06,
      "loss": 0.4475,
      "step": 7489
    },
    {
      "epoch": 0.34371988435592676,
      "grad_norm": 0.4484030306339264,
      "learning_rate": 9.77200382169467e-06,
      "loss": 0.4091,
      "step": 7490
    },
    {
      "epoch": 0.3437657748611812,
      "grad_norm": 0.5099649429321289,
      "learning_rate": 9.771930620413608e-06,
      "loss": 0.5278,
      "step": 7491
    },
    {
      "epoch": 0.3438116653664357,
      "grad_norm": 0.5197622179985046,
      "learning_rate": 9.771857407657558e-06,
      "loss": 0.4491,
      "step": 7492
    },
    {
      "epoch": 0.34385755587169015,
      "grad_norm": 0.4589492678642273,
      "learning_rate": 9.771784183426695e-06,
      "loss": 0.4519,
      "step": 7493
    },
    {
      "epoch": 0.3439034463769446,
      "grad_norm": 0.46764957904815674,
      "learning_rate": 9.7717109477212e-06,
      "loss": 0.4426,
      "step": 7494
    },
    {
      "epoch": 0.3439493368821991,
      "grad_norm": 0.4547639787197113,
      "learning_rate": 9.771637700541246e-06,
      "loss": 0.3956,
      "step": 7495
    },
    {
      "epoch": 0.34399522738745353,
      "grad_norm": 0.5215696692466736,
      "learning_rate": 9.771564441887007e-06,
      "loss": 0.4559,
      "step": 7496
    },
    {
      "epoch": 0.344041117892708,
      "grad_norm": 0.4635773003101349,
      "learning_rate": 9.771491171758663e-06,
      "loss": 0.4242,
      "step": 7497
    },
    {
      "epoch": 0.34408700839796247,
      "grad_norm": 0.4817538857460022,
      "learning_rate": 9.771417890156388e-06,
      "loss": 0.494,
      "step": 7498
    },
    {
      "epoch": 0.3441328989032169,
      "grad_norm": 0.45221272110939026,
      "learning_rate": 9.77134459708036e-06,
      "loss": 0.3741,
      "step": 7499
    },
    {
      "epoch": 0.3441787894084714,
      "grad_norm": 0.47556164860725403,
      "learning_rate": 9.771271292530753e-06,
      "loss": 0.4498,
      "step": 7500
    },
    {
      "epoch": 0.34422467991372585,
      "grad_norm": 0.5360143780708313,
      "learning_rate": 9.771197976507745e-06,
      "loss": 0.4889,
      "step": 7501
    },
    {
      "epoch": 0.3442705704189803,
      "grad_norm": 0.4882326126098633,
      "learning_rate": 9.77112464901151e-06,
      "loss": 0.3097,
      "step": 7502
    },
    {
      "epoch": 0.3443164609242348,
      "grad_norm": 0.5099873542785645,
      "learning_rate": 9.771051310042227e-06,
      "loss": 0.458,
      "step": 7503
    },
    {
      "epoch": 0.34436235142948923,
      "grad_norm": 0.5317543745040894,
      "learning_rate": 9.770977959600072e-06,
      "loss": 0.5257,
      "step": 7504
    },
    {
      "epoch": 0.3444082419347437,
      "grad_norm": 0.4684421420097351,
      "learning_rate": 9.770904597685222e-06,
      "loss": 0.4284,
      "step": 7505
    },
    {
      "epoch": 0.3444541324399982,
      "grad_norm": 0.4865078926086426,
      "learning_rate": 9.77083122429785e-06,
      "loss": 0.3681,
      "step": 7506
    },
    {
      "epoch": 0.3445000229452526,
      "grad_norm": 0.5494277477264404,
      "learning_rate": 9.770757839438135e-06,
      "loss": 0.4223,
      "step": 7507
    },
    {
      "epoch": 0.3445459134505071,
      "grad_norm": 0.45232754945755005,
      "learning_rate": 9.770684443106254e-06,
      "loss": 0.3585,
      "step": 7508
    },
    {
      "epoch": 0.34459180395576156,
      "grad_norm": 0.4730750024318695,
      "learning_rate": 9.770611035302383e-06,
      "loss": 0.4006,
      "step": 7509
    },
    {
      "epoch": 0.344637694461016,
      "grad_norm": 0.45557841658592224,
      "learning_rate": 9.770537616026697e-06,
      "loss": 0.375,
      "step": 7510
    },
    {
      "epoch": 0.3446835849662705,
      "grad_norm": 0.4844571053981781,
      "learning_rate": 9.770464185279374e-06,
      "loss": 0.4796,
      "step": 7511
    },
    {
      "epoch": 0.34472947547152494,
      "grad_norm": 0.4509688913822174,
      "learning_rate": 9.77039074306059e-06,
      "loss": 0.397,
      "step": 7512
    },
    {
      "epoch": 0.3447753659767794,
      "grad_norm": 0.4386076033115387,
      "learning_rate": 9.77031728937052e-06,
      "loss": 0.3526,
      "step": 7513
    },
    {
      "epoch": 0.3448212564820339,
      "grad_norm": 0.4587988555431366,
      "learning_rate": 9.770243824209346e-06,
      "loss": 0.4188,
      "step": 7514
    },
    {
      "epoch": 0.3448671469872883,
      "grad_norm": 0.44408902525901794,
      "learning_rate": 9.770170347577239e-06,
      "loss": 0.3547,
      "step": 7515
    },
    {
      "epoch": 0.3449130374925428,
      "grad_norm": 0.4457341730594635,
      "learning_rate": 9.770096859474378e-06,
      "loss": 0.3843,
      "step": 7516
    },
    {
      "epoch": 0.34495892799779726,
      "grad_norm": 0.4639625549316406,
      "learning_rate": 9.77002335990094e-06,
      "loss": 0.385,
      "step": 7517
    },
    {
      "epoch": 0.3450048185030517,
      "grad_norm": 0.42840731143951416,
      "learning_rate": 9.7699498488571e-06,
      "loss": 0.3852,
      "step": 7518
    },
    {
      "epoch": 0.3450507090083062,
      "grad_norm": 0.443139910697937,
      "learning_rate": 9.769876326343038e-06,
      "loss": 0.4059,
      "step": 7519
    },
    {
      "epoch": 0.34509659951356064,
      "grad_norm": 0.4856567680835724,
      "learning_rate": 9.769802792358929e-06,
      "loss": 0.4479,
      "step": 7520
    },
    {
      "epoch": 0.3451424900188151,
      "grad_norm": 0.47379782795906067,
      "learning_rate": 9.769729246904949e-06,
      "loss": 0.4116,
      "step": 7521
    },
    {
      "epoch": 0.3451883805240696,
      "grad_norm": 0.4640233814716339,
      "learning_rate": 9.769655689981274e-06,
      "loss": 0.4809,
      "step": 7522
    },
    {
      "epoch": 0.345234271029324,
      "grad_norm": 0.459486722946167,
      "learning_rate": 9.769582121588084e-06,
      "loss": 0.4004,
      "step": 7523
    },
    {
      "epoch": 0.3452801615345785,
      "grad_norm": 0.4644634425640106,
      "learning_rate": 9.769508541725555e-06,
      "loss": 0.3768,
      "step": 7524
    },
    {
      "epoch": 0.34532605203983296,
      "grad_norm": 0.46861618757247925,
      "learning_rate": 9.769434950393862e-06,
      "loss": 0.4209,
      "step": 7525
    },
    {
      "epoch": 0.3453719425450874,
      "grad_norm": 0.4838896691799164,
      "learning_rate": 9.769361347593182e-06,
      "loss": 0.5021,
      "step": 7526
    },
    {
      "epoch": 0.3454178330503419,
      "grad_norm": 0.4666491150856018,
      "learning_rate": 9.769287733323696e-06,
      "loss": 0.4339,
      "step": 7527
    },
    {
      "epoch": 0.34546372355559635,
      "grad_norm": 0.42675402760505676,
      "learning_rate": 9.769214107585576e-06,
      "loss": 0.3755,
      "step": 7528
    },
    {
      "epoch": 0.3455096140608508,
      "grad_norm": 0.5362486839294434,
      "learning_rate": 9.769140470379002e-06,
      "loss": 0.5166,
      "step": 7529
    },
    {
      "epoch": 0.3455555045661053,
      "grad_norm": 0.5078483819961548,
      "learning_rate": 9.769066821704151e-06,
      "loss": 0.4289,
      "step": 7530
    },
    {
      "epoch": 0.34560139507135973,
      "grad_norm": 0.46759679913520813,
      "learning_rate": 9.7689931615612e-06,
      "loss": 0.4006,
      "step": 7531
    },
    {
      "epoch": 0.34564728557661417,
      "grad_norm": 0.48502886295318604,
      "learning_rate": 9.768919489950324e-06,
      "loss": 0.4283,
      "step": 7532
    },
    {
      "epoch": 0.34569317608186867,
      "grad_norm": 0.4736391007900238,
      "learning_rate": 9.768845806871704e-06,
      "loss": 0.4122,
      "step": 7533
    },
    {
      "epoch": 0.3457390665871231,
      "grad_norm": 0.5229737162590027,
      "learning_rate": 9.76877211232551e-06,
      "loss": 0.4758,
      "step": 7534
    },
    {
      "epoch": 0.3457849570923776,
      "grad_norm": 0.40566006302833557,
      "learning_rate": 9.76869840631193e-06,
      "loss": 0.317,
      "step": 7535
    },
    {
      "epoch": 0.34583084759763205,
      "grad_norm": 0.4811519682407379,
      "learning_rate": 9.768624688831132e-06,
      "loss": 0.4311,
      "step": 7536
    },
    {
      "epoch": 0.3458767381028865,
      "grad_norm": 0.4791409969329834,
      "learning_rate": 9.768550959883296e-06,
      "loss": 0.3871,
      "step": 7537
    },
    {
      "epoch": 0.345922628608141,
      "grad_norm": 0.46593907475471497,
      "learning_rate": 9.768477219468602e-06,
      "loss": 0.3786,
      "step": 7538
    },
    {
      "epoch": 0.34596851911339543,
      "grad_norm": 0.45886313915252686,
      "learning_rate": 9.768403467587225e-06,
      "loss": 0.3734,
      "step": 7539
    },
    {
      "epoch": 0.3460144096186499,
      "grad_norm": 0.4441809058189392,
      "learning_rate": 9.768329704239342e-06,
      "loss": 0.3372,
      "step": 7540
    },
    {
      "epoch": 0.3460603001239044,
      "grad_norm": 0.5206360816955566,
      "learning_rate": 9.76825592942513e-06,
      "loss": 0.4563,
      "step": 7541
    },
    {
      "epoch": 0.3461061906291588,
      "grad_norm": 0.4262565076351166,
      "learning_rate": 9.76818214314477e-06,
      "loss": 0.3201,
      "step": 7542
    },
    {
      "epoch": 0.3461520811344133,
      "grad_norm": 0.4973420202732086,
      "learning_rate": 9.768108345398436e-06,
      "loss": 0.4333,
      "step": 7543
    },
    {
      "epoch": 0.34619797163966776,
      "grad_norm": 0.44933706521987915,
      "learning_rate": 9.768034536186304e-06,
      "loss": 0.3556,
      "step": 7544
    },
    {
      "epoch": 0.3462438621449222,
      "grad_norm": 0.4575684368610382,
      "learning_rate": 9.767960715508557e-06,
      "loss": 0.3872,
      "step": 7545
    },
    {
      "epoch": 0.3462897526501767,
      "grad_norm": 0.4663599729537964,
      "learning_rate": 9.767886883365367e-06,
      "loss": 0.3769,
      "step": 7546
    },
    {
      "epoch": 0.34633564315543114,
      "grad_norm": 0.45803332328796387,
      "learning_rate": 9.767813039756915e-06,
      "loss": 0.4193,
      "step": 7547
    },
    {
      "epoch": 0.3463815336606856,
      "grad_norm": 0.48722875118255615,
      "learning_rate": 9.767739184683377e-06,
      "loss": 0.4366,
      "step": 7548
    },
    {
      "epoch": 0.3464274241659401,
      "grad_norm": 0.4429193437099457,
      "learning_rate": 9.767665318144932e-06,
      "loss": 0.3692,
      "step": 7549
    },
    {
      "epoch": 0.3464733146711945,
      "grad_norm": 0.46046680212020874,
      "learning_rate": 9.767591440141755e-06,
      "loss": 0.3868,
      "step": 7550
    },
    {
      "epoch": 0.346519205176449,
      "grad_norm": 0.44924381375312805,
      "learning_rate": 9.767517550674026e-06,
      "loss": 0.313,
      "step": 7551
    },
    {
      "epoch": 0.34656509568170346,
      "grad_norm": 0.4696842133998871,
      "learning_rate": 9.767443649741921e-06,
      "loss": 0.4281,
      "step": 7552
    },
    {
      "epoch": 0.3466109861869579,
      "grad_norm": 0.47532036900520325,
      "learning_rate": 9.76736973734562e-06,
      "loss": 0.4298,
      "step": 7553
    },
    {
      "epoch": 0.3466568766922124,
      "grad_norm": 0.47718507051467896,
      "learning_rate": 9.7672958134853e-06,
      "loss": 0.4406,
      "step": 7554
    },
    {
      "epoch": 0.34670276719746684,
      "grad_norm": 0.47704941034317017,
      "learning_rate": 9.767221878161136e-06,
      "loss": 0.4042,
      "step": 7555
    },
    {
      "epoch": 0.3467486577027213,
      "grad_norm": 0.4445193111896515,
      "learning_rate": 9.767147931373309e-06,
      "loss": 0.3546,
      "step": 7556
    },
    {
      "epoch": 0.3467945482079758,
      "grad_norm": 0.4629325270652771,
      "learning_rate": 9.767073973121997e-06,
      "loss": 0.3604,
      "step": 7557
    },
    {
      "epoch": 0.3468404387132302,
      "grad_norm": 0.4590432345867157,
      "learning_rate": 9.767000003407375e-06,
      "loss": 0.3934,
      "step": 7558
    },
    {
      "epoch": 0.34688632921848467,
      "grad_norm": 0.451778382062912,
      "learning_rate": 9.766926022229623e-06,
      "loss": 0.3826,
      "step": 7559
    },
    {
      "epoch": 0.34693221972373917,
      "grad_norm": 0.44265544414520264,
      "learning_rate": 9.766852029588917e-06,
      "loss": 0.4331,
      "step": 7560
    },
    {
      "epoch": 0.3469781102289936,
      "grad_norm": 0.46604517102241516,
      "learning_rate": 9.76677802548544e-06,
      "loss": 0.501,
      "step": 7561
    },
    {
      "epoch": 0.3470240007342481,
      "grad_norm": 0.4993465542793274,
      "learning_rate": 9.766704009919364e-06,
      "loss": 0.4682,
      "step": 7562
    },
    {
      "epoch": 0.34706989123950255,
      "grad_norm": 0.4657362103462219,
      "learning_rate": 9.766629982890869e-06,
      "loss": 0.4157,
      "step": 7563
    },
    {
      "epoch": 0.347115781744757,
      "grad_norm": 0.5058092474937439,
      "learning_rate": 9.766555944400134e-06,
      "loss": 0.4851,
      "step": 7564
    },
    {
      "epoch": 0.3471616722500115,
      "grad_norm": 0.43176037073135376,
      "learning_rate": 9.766481894447334e-06,
      "loss": 0.3812,
      "step": 7565
    },
    {
      "epoch": 0.34720756275526593,
      "grad_norm": 0.506644606590271,
      "learning_rate": 9.766407833032652e-06,
      "loss": 0.5175,
      "step": 7566
    },
    {
      "epoch": 0.3472534532605204,
      "grad_norm": 0.4476737380027771,
      "learning_rate": 9.766333760156263e-06,
      "loss": 0.3327,
      "step": 7567
    },
    {
      "epoch": 0.34729934376577487,
      "grad_norm": 0.48623380064964294,
      "learning_rate": 9.766259675818345e-06,
      "loss": 0.3904,
      "step": 7568
    },
    {
      "epoch": 0.3473452342710293,
      "grad_norm": 0.45771583914756775,
      "learning_rate": 9.766185580019077e-06,
      "loss": 0.3821,
      "step": 7569
    },
    {
      "epoch": 0.3473911247762838,
      "grad_norm": 0.46506422758102417,
      "learning_rate": 9.766111472758637e-06,
      "loss": 0.4236,
      "step": 7570
    },
    {
      "epoch": 0.34743701528153825,
      "grad_norm": 0.4524925649166107,
      "learning_rate": 9.766037354037202e-06,
      "loss": 0.3786,
      "step": 7571
    },
    {
      "epoch": 0.3474829057867927,
      "grad_norm": 0.4716859459877014,
      "learning_rate": 9.765963223854955e-06,
      "loss": 0.4596,
      "step": 7572
    },
    {
      "epoch": 0.3475287962920472,
      "grad_norm": 0.4681797921657562,
      "learning_rate": 9.765889082212068e-06,
      "loss": 0.4272,
      "step": 7573
    },
    {
      "epoch": 0.34757468679730164,
      "grad_norm": 0.47106581926345825,
      "learning_rate": 9.765814929108721e-06,
      "loss": 0.453,
      "step": 7574
    },
    {
      "epoch": 0.3476205773025561,
      "grad_norm": 0.47277775406837463,
      "learning_rate": 9.765740764545096e-06,
      "loss": 0.4526,
      "step": 7575
    },
    {
      "epoch": 0.3476664678078106,
      "grad_norm": 0.48712101578712463,
      "learning_rate": 9.765666588521366e-06,
      "loss": 0.4492,
      "step": 7576
    },
    {
      "epoch": 0.347712358313065,
      "grad_norm": 0.41602057218551636,
      "learning_rate": 9.765592401037713e-06,
      "loss": 0.3263,
      "step": 7577
    },
    {
      "epoch": 0.3477582488183195,
      "grad_norm": 0.4814780056476593,
      "learning_rate": 9.765518202094313e-06,
      "loss": 0.4366,
      "step": 7578
    },
    {
      "epoch": 0.34780413932357396,
      "grad_norm": 0.46363043785095215,
      "learning_rate": 9.765443991691347e-06,
      "loss": 0.417,
      "step": 7579
    },
    {
      "epoch": 0.3478500298288284,
      "grad_norm": 0.44514644145965576,
      "learning_rate": 9.765369769828994e-06,
      "loss": 0.3675,
      "step": 7580
    },
    {
      "epoch": 0.3478959203340829,
      "grad_norm": 0.42944419384002686,
      "learning_rate": 9.765295536507429e-06,
      "loss": 0.3409,
      "step": 7581
    },
    {
      "epoch": 0.34794181083933734,
      "grad_norm": 0.46850618720054626,
      "learning_rate": 9.765221291726833e-06,
      "loss": 0.4455,
      "step": 7582
    },
    {
      "epoch": 0.3479877013445918,
      "grad_norm": 0.4708006978034973,
      "learning_rate": 9.765147035487381e-06,
      "loss": 0.404,
      "step": 7583
    },
    {
      "epoch": 0.3480335918498463,
      "grad_norm": 0.4630293846130371,
      "learning_rate": 9.765072767789257e-06,
      "loss": 0.3902,
      "step": 7584
    },
    {
      "epoch": 0.3480794823551007,
      "grad_norm": 0.48770028352737427,
      "learning_rate": 9.764998488632635e-06,
      "loss": 0.5014,
      "step": 7585
    },
    {
      "epoch": 0.3481253728603552,
      "grad_norm": 0.4378030002117157,
      "learning_rate": 9.764924198017696e-06,
      "loss": 0.3357,
      "step": 7586
    },
    {
      "epoch": 0.34817126336560966,
      "grad_norm": 0.43347692489624023,
      "learning_rate": 9.76484989594462e-06,
      "loss": 0.3606,
      "step": 7587
    },
    {
      "epoch": 0.3482171538708641,
      "grad_norm": 0.48702043294906616,
      "learning_rate": 9.764775582413581e-06,
      "loss": 0.4239,
      "step": 7588
    },
    {
      "epoch": 0.3482630443761186,
      "grad_norm": 0.4599001109600067,
      "learning_rate": 9.764701257424764e-06,
      "loss": 0.4388,
      "step": 7589
    },
    {
      "epoch": 0.34830893488137304,
      "grad_norm": 0.43673670291900635,
      "learning_rate": 9.764626920978342e-06,
      "loss": 0.3221,
      "step": 7590
    },
    {
      "epoch": 0.3483548253866275,
      "grad_norm": 0.4766339063644409,
      "learning_rate": 9.764552573074496e-06,
      "loss": 0.4216,
      "step": 7591
    },
    {
      "epoch": 0.348400715891882,
      "grad_norm": 0.4625072479248047,
      "learning_rate": 9.764478213713405e-06,
      "loss": 0.405,
      "step": 7592
    },
    {
      "epoch": 0.3484466063971364,
      "grad_norm": 0.4607468545436859,
      "learning_rate": 9.764403842895246e-06,
      "loss": 0.401,
      "step": 7593
    },
    {
      "epoch": 0.34849249690239087,
      "grad_norm": 0.4497205913066864,
      "learning_rate": 9.764329460620203e-06,
      "loss": 0.3526,
      "step": 7594
    },
    {
      "epoch": 0.34853838740764537,
      "grad_norm": 0.47558921575546265,
      "learning_rate": 9.76425506688845e-06,
      "loss": 0.4318,
      "step": 7595
    },
    {
      "epoch": 0.3485842779128998,
      "grad_norm": 0.4221167266368866,
      "learning_rate": 9.764180661700167e-06,
      "loss": 0.3213,
      "step": 7596
    },
    {
      "epoch": 0.3486301684181543,
      "grad_norm": 0.495102196931839,
      "learning_rate": 9.76410624505553e-06,
      "loss": 0.4529,
      "step": 7597
    },
    {
      "epoch": 0.34867605892340875,
      "grad_norm": 0.4522758722305298,
      "learning_rate": 9.764031816954726e-06,
      "loss": 0.4188,
      "step": 7598
    },
    {
      "epoch": 0.3487219494286632,
      "grad_norm": 0.4883578419685364,
      "learning_rate": 9.763957377397927e-06,
      "loss": 0.4327,
      "step": 7599
    },
    {
      "epoch": 0.3487678399339177,
      "grad_norm": 0.481063574552536,
      "learning_rate": 9.763882926385314e-06,
      "loss": 0.4431,
      "step": 7600
    },
    {
      "epoch": 0.34881373043917213,
      "grad_norm": 0.4394407570362091,
      "learning_rate": 9.763808463917067e-06,
      "loss": 0.336,
      "step": 7601
    },
    {
      "epoch": 0.3488596209444266,
      "grad_norm": 0.47730714082717896,
      "learning_rate": 9.763733989993362e-06,
      "loss": 0.4527,
      "step": 7602
    },
    {
      "epoch": 0.34890551144968107,
      "grad_norm": 0.4249570071697235,
      "learning_rate": 9.763659504614382e-06,
      "loss": 0.3607,
      "step": 7603
    },
    {
      "epoch": 0.3489514019549355,
      "grad_norm": 0.43443503975868225,
      "learning_rate": 9.763585007780302e-06,
      "loss": 0.3925,
      "step": 7604
    },
    {
      "epoch": 0.34899729246019,
      "grad_norm": 0.4808576703071594,
      "learning_rate": 9.763510499491307e-06,
      "loss": 0.488,
      "step": 7605
    },
    {
      "epoch": 0.34904318296544445,
      "grad_norm": 0.42800503969192505,
      "learning_rate": 9.763435979747572e-06,
      "loss": 0.3532,
      "step": 7606
    },
    {
      "epoch": 0.3490890734706989,
      "grad_norm": 0.4606592059135437,
      "learning_rate": 9.763361448549276e-06,
      "loss": 0.4119,
      "step": 7607
    },
    {
      "epoch": 0.3491349639759534,
      "grad_norm": 0.46513354778289795,
      "learning_rate": 9.7632869058966e-06,
      "loss": 0.392,
      "step": 7608
    },
    {
      "epoch": 0.34918085448120784,
      "grad_norm": 0.48260772228240967,
      "learning_rate": 9.76321235178972e-06,
      "loss": 0.4427,
      "step": 7609
    },
    {
      "epoch": 0.3492267449864623,
      "grad_norm": 0.46555444598197937,
      "learning_rate": 9.763137786228819e-06,
      "loss": 0.4097,
      "step": 7610
    },
    {
      "epoch": 0.3492726354917168,
      "grad_norm": 0.4430099129676819,
      "learning_rate": 9.763063209214074e-06,
      "loss": 0.3802,
      "step": 7611
    },
    {
      "epoch": 0.3493185259969712,
      "grad_norm": 0.48142874240875244,
      "learning_rate": 9.762988620745666e-06,
      "loss": 0.4731,
      "step": 7612
    },
    {
      "epoch": 0.3493644165022257,
      "grad_norm": 0.4732450842857361,
      "learning_rate": 9.762914020823772e-06,
      "loss": 0.4733,
      "step": 7613
    },
    {
      "epoch": 0.34941030700748016,
      "grad_norm": 0.42975950241088867,
      "learning_rate": 9.762839409448575e-06,
      "loss": 0.4021,
      "step": 7614
    },
    {
      "epoch": 0.3494561975127346,
      "grad_norm": 0.44928207993507385,
      "learning_rate": 9.762764786620254e-06,
      "loss": 0.4275,
      "step": 7615
    },
    {
      "epoch": 0.3495020880179891,
      "grad_norm": 0.4540664851665497,
      "learning_rate": 9.762690152338984e-06,
      "loss": 0.3657,
      "step": 7616
    },
    {
      "epoch": 0.34954797852324354,
      "grad_norm": 0.46384164690971375,
      "learning_rate": 9.762615506604947e-06,
      "loss": 0.4554,
      "step": 7617
    },
    {
      "epoch": 0.349593869028498,
      "grad_norm": 0.47765716910362244,
      "learning_rate": 9.762540849418325e-06,
      "loss": 0.4297,
      "step": 7618
    },
    {
      "epoch": 0.3496397595337525,
      "grad_norm": 0.4374117851257324,
      "learning_rate": 9.762466180779295e-06,
      "loss": 0.3794,
      "step": 7619
    },
    {
      "epoch": 0.3496856500390069,
      "grad_norm": 0.4981018900871277,
      "learning_rate": 9.762391500688035e-06,
      "loss": 0.4546,
      "step": 7620
    },
    {
      "epoch": 0.34973154054426137,
      "grad_norm": 0.44146597385406494,
      "learning_rate": 9.762316809144728e-06,
      "loss": 0.4038,
      "step": 7621
    },
    {
      "epoch": 0.34977743104951586,
      "grad_norm": 0.46910905838012695,
      "learning_rate": 9.762242106149553e-06,
      "loss": 0.4241,
      "step": 7622
    },
    {
      "epoch": 0.3498233215547703,
      "grad_norm": 0.45875227451324463,
      "learning_rate": 9.762167391702688e-06,
      "loss": 0.41,
      "step": 7623
    },
    {
      "epoch": 0.3498692120600248,
      "grad_norm": 0.48692455887794495,
      "learning_rate": 9.762092665804313e-06,
      "loss": 0.4912,
      "step": 7624
    },
    {
      "epoch": 0.34991510256527925,
      "grad_norm": 0.46123653650283813,
      "learning_rate": 9.762017928454608e-06,
      "loss": 0.4409,
      "step": 7625
    },
    {
      "epoch": 0.3499609930705337,
      "grad_norm": 0.4709842801094055,
      "learning_rate": 9.761943179653752e-06,
      "loss": 0.3886,
      "step": 7626
    },
    {
      "epoch": 0.3500068835757882,
      "grad_norm": 0.4526197910308838,
      "learning_rate": 9.761868419401928e-06,
      "loss": 0.3963,
      "step": 7627
    },
    {
      "epoch": 0.35005277408104263,
      "grad_norm": 0.44349461793899536,
      "learning_rate": 9.761793647699311e-06,
      "loss": 0.3245,
      "step": 7628
    },
    {
      "epoch": 0.35009866458629707,
      "grad_norm": 0.5441825985908508,
      "learning_rate": 9.761718864546083e-06,
      "loss": 0.3911,
      "step": 7629
    },
    {
      "epoch": 0.35014455509155157,
      "grad_norm": 0.45233970880508423,
      "learning_rate": 9.761644069942426e-06,
      "loss": 0.3987,
      "step": 7630
    },
    {
      "epoch": 0.350190445596806,
      "grad_norm": 0.48667091131210327,
      "learning_rate": 9.761569263888515e-06,
      "loss": 0.4647,
      "step": 7631
    },
    {
      "epoch": 0.3502363361020605,
      "grad_norm": 0.4708206355571747,
      "learning_rate": 9.761494446384537e-06,
      "loss": 0.43,
      "step": 7632
    },
    {
      "epoch": 0.35028222660731495,
      "grad_norm": 0.4773523807525635,
      "learning_rate": 9.761419617430663e-06,
      "loss": 0.4534,
      "step": 7633
    },
    {
      "epoch": 0.3503281171125694,
      "grad_norm": 0.4876518249511719,
      "learning_rate": 9.761344777027081e-06,
      "loss": 0.4542,
      "step": 7634
    },
    {
      "epoch": 0.3503740076178239,
      "grad_norm": 0.46373939514160156,
      "learning_rate": 9.761269925173965e-06,
      "loss": 0.3625,
      "step": 7635
    },
    {
      "epoch": 0.35041989812307833,
      "grad_norm": 0.493642121553421,
      "learning_rate": 9.7611950618715e-06,
      "loss": 0.4174,
      "step": 7636
    },
    {
      "epoch": 0.3504657886283328,
      "grad_norm": 0.4417996406555176,
      "learning_rate": 9.761120187119863e-06,
      "loss": 0.3771,
      "step": 7637
    },
    {
      "epoch": 0.3505116791335873,
      "grad_norm": 0.4484770894050598,
      "learning_rate": 9.761045300919234e-06,
      "loss": 0.3818,
      "step": 7638
    },
    {
      "epoch": 0.3505575696388417,
      "grad_norm": 0.44239673018455505,
      "learning_rate": 9.760970403269794e-06,
      "loss": 0.3683,
      "step": 7639
    },
    {
      "epoch": 0.3506034601440962,
      "grad_norm": 0.4582446217536926,
      "learning_rate": 9.760895494171721e-06,
      "loss": 0.4083,
      "step": 7640
    },
    {
      "epoch": 0.35064935064935066,
      "grad_norm": 0.475626140832901,
      "learning_rate": 9.760820573625199e-06,
      "loss": 0.4032,
      "step": 7641
    },
    {
      "epoch": 0.3506952411546051,
      "grad_norm": 0.46180489659309387,
      "learning_rate": 9.760745641630405e-06,
      "loss": 0.3767,
      "step": 7642
    },
    {
      "epoch": 0.3507411316598596,
      "grad_norm": 0.46454235911369324,
      "learning_rate": 9.760670698187521e-06,
      "loss": 0.3942,
      "step": 7643
    },
    {
      "epoch": 0.35078702216511404,
      "grad_norm": 0.4947590231895447,
      "learning_rate": 9.760595743296726e-06,
      "loss": 0.4151,
      "step": 7644
    },
    {
      "epoch": 0.3508329126703685,
      "grad_norm": 0.43232065439224243,
      "learning_rate": 9.760520776958202e-06,
      "loss": 0.3852,
      "step": 7645
    },
    {
      "epoch": 0.350878803175623,
      "grad_norm": 0.44100961089134216,
      "learning_rate": 9.760445799172125e-06,
      "loss": 0.3756,
      "step": 7646
    },
    {
      "epoch": 0.3509246936808774,
      "grad_norm": 0.47151312232017517,
      "learning_rate": 9.760370809938681e-06,
      "loss": 0.4546,
      "step": 7647
    },
    {
      "epoch": 0.35097058418613186,
      "grad_norm": 0.4657520651817322,
      "learning_rate": 9.760295809258048e-06,
      "loss": 0.4345,
      "step": 7648
    },
    {
      "epoch": 0.35101647469138636,
      "grad_norm": 0.5069671869277954,
      "learning_rate": 9.760220797130404e-06,
      "loss": 0.4271,
      "step": 7649
    },
    {
      "epoch": 0.3510623651966408,
      "grad_norm": 0.4813474118709564,
      "learning_rate": 9.76014577355593e-06,
      "loss": 0.4842,
      "step": 7650
    },
    {
      "epoch": 0.3511082557018953,
      "grad_norm": 0.4638098180294037,
      "learning_rate": 9.760070738534813e-06,
      "loss": 0.4126,
      "step": 7651
    },
    {
      "epoch": 0.35115414620714974,
      "grad_norm": 0.501701831817627,
      "learning_rate": 9.759995692067224e-06,
      "loss": 0.3932,
      "step": 7652
    },
    {
      "epoch": 0.3512000367124042,
      "grad_norm": 0.4455215632915497,
      "learning_rate": 9.759920634153348e-06,
      "loss": 0.3458,
      "step": 7653
    },
    {
      "epoch": 0.3512459272176587,
      "grad_norm": 0.4620753526687622,
      "learning_rate": 9.759845564793367e-06,
      "loss": 0.4218,
      "step": 7654
    },
    {
      "epoch": 0.3512918177229131,
      "grad_norm": 0.6094580292701721,
      "learning_rate": 9.759770483987459e-06,
      "loss": 0.4281,
      "step": 7655
    },
    {
      "epoch": 0.35133770822816757,
      "grad_norm": 0.44705837965011597,
      "learning_rate": 9.759695391735805e-06,
      "loss": 0.3885,
      "step": 7656
    },
    {
      "epoch": 0.35138359873342206,
      "grad_norm": 0.444299578666687,
      "learning_rate": 9.759620288038585e-06,
      "loss": 0.3727,
      "step": 7657
    },
    {
      "epoch": 0.3514294892386765,
      "grad_norm": 0.4475720226764679,
      "learning_rate": 9.75954517289598e-06,
      "loss": 0.3575,
      "step": 7658
    },
    {
      "epoch": 0.351475379743931,
      "grad_norm": 0.47326457500457764,
      "learning_rate": 9.759470046308172e-06,
      "loss": 0.4009,
      "step": 7659
    },
    {
      "epoch": 0.35152127024918545,
      "grad_norm": 0.4519866108894348,
      "learning_rate": 9.75939490827534e-06,
      "loss": 0.3705,
      "step": 7660
    },
    {
      "epoch": 0.3515671607544399,
      "grad_norm": 0.4684807062149048,
      "learning_rate": 9.759319758797668e-06,
      "loss": 0.4154,
      "step": 7661
    },
    {
      "epoch": 0.3516130512596944,
      "grad_norm": 0.49699339270591736,
      "learning_rate": 9.759244597875331e-06,
      "loss": 0.4872,
      "step": 7662
    },
    {
      "epoch": 0.35165894176494883,
      "grad_norm": 0.4818732440471649,
      "learning_rate": 9.759169425508514e-06,
      "loss": 0.464,
      "step": 7663
    },
    {
      "epoch": 0.35170483227020327,
      "grad_norm": 0.4806285500526428,
      "learning_rate": 9.759094241697395e-06,
      "loss": 0.383,
      "step": 7664
    },
    {
      "epoch": 0.35175072277545777,
      "grad_norm": 0.4402625262737274,
      "learning_rate": 9.75901904644216e-06,
      "loss": 0.4168,
      "step": 7665
    },
    {
      "epoch": 0.3517966132807122,
      "grad_norm": 0.44949936866760254,
      "learning_rate": 9.758943839742982e-06,
      "loss": 0.4153,
      "step": 7666
    },
    {
      "epoch": 0.3518425037859667,
      "grad_norm": 0.42444729804992676,
      "learning_rate": 9.758868621600048e-06,
      "loss": 0.348,
      "step": 7667
    },
    {
      "epoch": 0.35188839429122115,
      "grad_norm": 0.44733190536499023,
      "learning_rate": 9.758793392013538e-06,
      "loss": 0.4005,
      "step": 7668
    },
    {
      "epoch": 0.3519342847964756,
      "grad_norm": 0.4605877101421356,
      "learning_rate": 9.75871815098363e-06,
      "loss": 0.4127,
      "step": 7669
    },
    {
      "epoch": 0.3519801753017301,
      "grad_norm": 0.48094403743743896,
      "learning_rate": 9.758642898510509e-06,
      "loss": 0.4713,
      "step": 7670
    },
    {
      "epoch": 0.35202606580698453,
      "grad_norm": 0.45184120535850525,
      "learning_rate": 9.758567634594351e-06,
      "loss": 0.3871,
      "step": 7671
    },
    {
      "epoch": 0.352071956312239,
      "grad_norm": 0.5021167993545532,
      "learning_rate": 9.758492359235341e-06,
      "loss": 0.5071,
      "step": 7672
    },
    {
      "epoch": 0.3521178468174935,
      "grad_norm": 0.44302135705947876,
      "learning_rate": 9.75841707243366e-06,
      "loss": 0.3998,
      "step": 7673
    },
    {
      "epoch": 0.3521637373227479,
      "grad_norm": 0.49103161692619324,
      "learning_rate": 9.758341774189484e-06,
      "loss": 0.379,
      "step": 7674
    },
    {
      "epoch": 0.3522096278280024,
      "grad_norm": 0.433456152677536,
      "learning_rate": 9.758266464503002e-06,
      "loss": 0.3236,
      "step": 7675
    },
    {
      "epoch": 0.35225551833325686,
      "grad_norm": 0.44734370708465576,
      "learning_rate": 9.758191143374388e-06,
      "loss": 0.4047,
      "step": 7676
    },
    {
      "epoch": 0.3523014088385113,
      "grad_norm": 0.5386069416999817,
      "learning_rate": 9.758115810803827e-06,
      "loss": 0.5584,
      "step": 7677
    },
    {
      "epoch": 0.3523472993437658,
      "grad_norm": 0.48044177889823914,
      "learning_rate": 9.758040466791499e-06,
      "loss": 0.5166,
      "step": 7678
    },
    {
      "epoch": 0.35239318984902024,
      "grad_norm": 0.4861387610435486,
      "learning_rate": 9.757965111337586e-06,
      "loss": 0.3863,
      "step": 7679
    },
    {
      "epoch": 0.3524390803542747,
      "grad_norm": 0.4583209753036499,
      "learning_rate": 9.757889744442268e-06,
      "loss": 0.3483,
      "step": 7680
    },
    {
      "epoch": 0.3524849708595292,
      "grad_norm": 0.4502546489238739,
      "learning_rate": 9.757814366105727e-06,
      "loss": 0.3488,
      "step": 7681
    },
    {
      "epoch": 0.3525308613647836,
      "grad_norm": 0.47257286310195923,
      "learning_rate": 9.757738976328144e-06,
      "loss": 0.4316,
      "step": 7682
    },
    {
      "epoch": 0.35257675187003806,
      "grad_norm": 0.44128793478012085,
      "learning_rate": 9.7576635751097e-06,
      "loss": 0.3447,
      "step": 7683
    },
    {
      "epoch": 0.35262264237529256,
      "grad_norm": 0.46079134941101074,
      "learning_rate": 9.757588162450577e-06,
      "loss": 0.3577,
      "step": 7684
    },
    {
      "epoch": 0.352668532880547,
      "grad_norm": 0.4709286391735077,
      "learning_rate": 9.757512738350957e-06,
      "loss": 0.422,
      "step": 7685
    },
    {
      "epoch": 0.3527144233858015,
      "grad_norm": 0.49472177028656006,
      "learning_rate": 9.75743730281102e-06,
      "loss": 0.4517,
      "step": 7686
    },
    {
      "epoch": 0.35276031389105594,
      "grad_norm": 0.45751404762268066,
      "learning_rate": 9.757361855830945e-06,
      "loss": 0.3875,
      "step": 7687
    },
    {
      "epoch": 0.3528062043963104,
      "grad_norm": 0.47235843539237976,
      "learning_rate": 9.757286397410919e-06,
      "loss": 0.4256,
      "step": 7688
    },
    {
      "epoch": 0.3528520949015649,
      "grad_norm": 0.47484615445137024,
      "learning_rate": 9.75721092755112e-06,
      "loss": 0.4535,
      "step": 7689
    },
    {
      "epoch": 0.3528979854068193,
      "grad_norm": 0.46440690755844116,
      "learning_rate": 9.757135446251732e-06,
      "loss": 0.3657,
      "step": 7690
    },
    {
      "epoch": 0.35294387591207377,
      "grad_norm": 0.5233825445175171,
      "learning_rate": 9.75705995351293e-06,
      "loss": 0.4062,
      "step": 7691
    },
    {
      "epoch": 0.35298976641732827,
      "grad_norm": 0.46515750885009766,
      "learning_rate": 9.756984449334905e-06,
      "loss": 0.4359,
      "step": 7692
    },
    {
      "epoch": 0.3530356569225827,
      "grad_norm": 0.4670040011405945,
      "learning_rate": 9.75690893371783e-06,
      "loss": 0.4315,
      "step": 7693
    },
    {
      "epoch": 0.3530815474278372,
      "grad_norm": 0.4625621736049652,
      "learning_rate": 9.756833406661892e-06,
      "loss": 0.374,
      "step": 7694
    },
    {
      "epoch": 0.35312743793309165,
      "grad_norm": 0.4773547649383545,
      "learning_rate": 9.756757868167273e-06,
      "loss": 0.4316,
      "step": 7695
    },
    {
      "epoch": 0.3531733284383461,
      "grad_norm": 0.48153454065322876,
      "learning_rate": 9.75668231823415e-06,
      "loss": 0.3806,
      "step": 7696
    },
    {
      "epoch": 0.3532192189436006,
      "grad_norm": 0.47327664494514465,
      "learning_rate": 9.756606756862708e-06,
      "loss": 0.4027,
      "step": 7697
    },
    {
      "epoch": 0.35326510944885503,
      "grad_norm": 0.4210459887981415,
      "learning_rate": 9.756531184053128e-06,
      "loss": 0.333,
      "step": 7698
    },
    {
      "epoch": 0.3533109999541095,
      "grad_norm": 0.4121096134185791,
      "learning_rate": 9.756455599805591e-06,
      "loss": 0.3196,
      "step": 7699
    },
    {
      "epoch": 0.35335689045936397,
      "grad_norm": 0.4854283630847931,
      "learning_rate": 9.75638000412028e-06,
      "loss": 0.4747,
      "step": 7700
    },
    {
      "epoch": 0.3534027809646184,
      "grad_norm": 0.5128400325775146,
      "learning_rate": 9.756304396997376e-06,
      "loss": 0.4958,
      "step": 7701
    },
    {
      "epoch": 0.3534486714698729,
      "grad_norm": 0.47235602140426636,
      "learning_rate": 9.756228778437061e-06,
      "loss": 0.4246,
      "step": 7702
    },
    {
      "epoch": 0.35349456197512735,
      "grad_norm": 0.4501318335533142,
      "learning_rate": 9.756153148439518e-06,
      "loss": 0.3874,
      "step": 7703
    },
    {
      "epoch": 0.3535404524803818,
      "grad_norm": 0.4421451985836029,
      "learning_rate": 9.756077507004927e-06,
      "loss": 0.358,
      "step": 7704
    },
    {
      "epoch": 0.3535863429856363,
      "grad_norm": 0.4623696208000183,
      "learning_rate": 9.75600185413347e-06,
      "loss": 0.4097,
      "step": 7705
    },
    {
      "epoch": 0.35363223349089073,
      "grad_norm": 0.4466571807861328,
      "learning_rate": 9.75592618982533e-06,
      "loss": 0.3721,
      "step": 7706
    },
    {
      "epoch": 0.3536781239961452,
      "grad_norm": 0.509890079498291,
      "learning_rate": 9.75585051408069e-06,
      "loss": 0.5093,
      "step": 7707
    },
    {
      "epoch": 0.3537240145013997,
      "grad_norm": 0.4390869140625,
      "learning_rate": 9.75577482689973e-06,
      "loss": 0.3904,
      "step": 7708
    },
    {
      "epoch": 0.3537699050066541,
      "grad_norm": 0.4590704143047333,
      "learning_rate": 9.755699128282631e-06,
      "loss": 0.4314,
      "step": 7709
    },
    {
      "epoch": 0.35381579551190856,
      "grad_norm": 0.5847171545028687,
      "learning_rate": 9.755623418229577e-06,
      "loss": 0.4429,
      "step": 7710
    },
    {
      "epoch": 0.35386168601716306,
      "grad_norm": 0.44052451848983765,
      "learning_rate": 9.755547696740751e-06,
      "loss": 0.4015,
      "step": 7711
    },
    {
      "epoch": 0.3539075765224175,
      "grad_norm": 0.43914538621902466,
      "learning_rate": 9.755471963816331e-06,
      "loss": 0.3486,
      "step": 7712
    },
    {
      "epoch": 0.353953467027672,
      "grad_norm": 0.44835731387138367,
      "learning_rate": 9.755396219456503e-06,
      "loss": 0.4374,
      "step": 7713
    },
    {
      "epoch": 0.35399935753292644,
      "grad_norm": 0.4606400430202484,
      "learning_rate": 9.755320463661449e-06,
      "loss": 0.4279,
      "step": 7714
    },
    {
      "epoch": 0.3540452480381809,
      "grad_norm": 0.4537215232849121,
      "learning_rate": 9.75524469643135e-06,
      "loss": 0.3369,
      "step": 7715
    },
    {
      "epoch": 0.3540911385434354,
      "grad_norm": 0.4954622983932495,
      "learning_rate": 9.755168917766387e-06,
      "loss": 0.3899,
      "step": 7716
    },
    {
      "epoch": 0.3541370290486898,
      "grad_norm": 0.47796550393104553,
      "learning_rate": 9.755093127666746e-06,
      "loss": 0.4118,
      "step": 7717
    },
    {
      "epoch": 0.35418291955394426,
      "grad_norm": 0.43439486622810364,
      "learning_rate": 9.755017326132603e-06,
      "loss": 0.3434,
      "step": 7718
    },
    {
      "epoch": 0.35422881005919876,
      "grad_norm": 0.4611569344997406,
      "learning_rate": 9.754941513164146e-06,
      "loss": 0.3655,
      "step": 7719
    },
    {
      "epoch": 0.3542747005644532,
      "grad_norm": 0.46086350083351135,
      "learning_rate": 9.754865688761556e-06,
      "loss": 0.3756,
      "step": 7720
    },
    {
      "epoch": 0.3543205910697077,
      "grad_norm": 0.45835474133491516,
      "learning_rate": 9.754789852925015e-06,
      "loss": 0.4078,
      "step": 7721
    },
    {
      "epoch": 0.35436648157496214,
      "grad_norm": 0.4975372850894928,
      "learning_rate": 9.754714005654704e-06,
      "loss": 0.4873,
      "step": 7722
    },
    {
      "epoch": 0.3544123720802166,
      "grad_norm": 0.7710484862327576,
      "learning_rate": 9.754638146950806e-06,
      "loss": 0.4407,
      "step": 7723
    },
    {
      "epoch": 0.3544582625854711,
      "grad_norm": 0.49734094738960266,
      "learning_rate": 9.754562276813505e-06,
      "loss": 0.4624,
      "step": 7724
    },
    {
      "epoch": 0.3545041530907255,
      "grad_norm": 0.40934163331985474,
      "learning_rate": 9.754486395242981e-06,
      "loss": 0.2775,
      "step": 7725
    },
    {
      "epoch": 0.35455004359597997,
      "grad_norm": 0.49207213521003723,
      "learning_rate": 9.75441050223942e-06,
      "loss": 0.4016,
      "step": 7726
    },
    {
      "epoch": 0.35459593410123447,
      "grad_norm": 0.5139963626861572,
      "learning_rate": 9.754334597803002e-06,
      "loss": 0.4973,
      "step": 7727
    },
    {
      "epoch": 0.3546418246064889,
      "grad_norm": 0.45089462399482727,
      "learning_rate": 9.754258681933907e-06,
      "loss": 0.4124,
      "step": 7728
    },
    {
      "epoch": 0.3546877151117434,
      "grad_norm": 0.398920476436615,
      "learning_rate": 9.754182754632322e-06,
      "loss": 0.3441,
      "step": 7729
    },
    {
      "epoch": 0.35473360561699785,
      "grad_norm": 0.46400922536849976,
      "learning_rate": 9.754106815898427e-06,
      "loss": 0.4032,
      "step": 7730
    },
    {
      "epoch": 0.3547794961222523,
      "grad_norm": 0.45743653178215027,
      "learning_rate": 9.754030865732408e-06,
      "loss": 0.3683,
      "step": 7731
    },
    {
      "epoch": 0.3548253866275068,
      "grad_norm": 0.5279353857040405,
      "learning_rate": 9.753954904134444e-06,
      "loss": 0.5009,
      "step": 7732
    },
    {
      "epoch": 0.35487127713276123,
      "grad_norm": 0.49791938066482544,
      "learning_rate": 9.75387893110472e-06,
      "loss": 0.3971,
      "step": 7733
    },
    {
      "epoch": 0.3549171676380157,
      "grad_norm": 0.48502689599990845,
      "learning_rate": 9.753802946643414e-06,
      "loss": 0.4203,
      "step": 7734
    },
    {
      "epoch": 0.35496305814327017,
      "grad_norm": 0.5431489944458008,
      "learning_rate": 9.753726950750714e-06,
      "loss": 0.4644,
      "step": 7735
    },
    {
      "epoch": 0.3550089486485246,
      "grad_norm": 0.5550913214683533,
      "learning_rate": 9.753650943426802e-06,
      "loss": 0.4091,
      "step": 7736
    },
    {
      "epoch": 0.35505483915377906,
      "grad_norm": 0.5059146285057068,
      "learning_rate": 9.75357492467186e-06,
      "loss": 0.5144,
      "step": 7737
    },
    {
      "epoch": 0.35510072965903355,
      "grad_norm": 0.4584423899650574,
      "learning_rate": 9.75349889448607e-06,
      "loss": 0.3674,
      "step": 7738
    },
    {
      "epoch": 0.355146620164288,
      "grad_norm": 0.4817931354045868,
      "learning_rate": 9.753422852869615e-06,
      "loss": 0.3707,
      "step": 7739
    },
    {
      "epoch": 0.3551925106695425,
      "grad_norm": 0.47430333495140076,
      "learning_rate": 9.753346799822678e-06,
      "loss": 0.3851,
      "step": 7740
    },
    {
      "epoch": 0.35523840117479694,
      "grad_norm": 0.4733114242553711,
      "learning_rate": 9.753270735345441e-06,
      "loss": 0.3991,
      "step": 7741
    },
    {
      "epoch": 0.3552842916800514,
      "grad_norm": 0.5028703212738037,
      "learning_rate": 9.75319465943809e-06,
      "loss": 0.5023,
      "step": 7742
    },
    {
      "epoch": 0.3553301821853059,
      "grad_norm": 0.473196417093277,
      "learning_rate": 9.753118572100805e-06,
      "loss": 0.433,
      "step": 7743
    },
    {
      "epoch": 0.3553760726905603,
      "grad_norm": 0.560771644115448,
      "learning_rate": 9.75304247333377e-06,
      "loss": 0.4082,
      "step": 7744
    },
    {
      "epoch": 0.35542196319581476,
      "grad_norm": 0.5072433352470398,
      "learning_rate": 9.75296636313717e-06,
      "loss": 0.4519,
      "step": 7745
    },
    {
      "epoch": 0.35546785370106926,
      "grad_norm": 0.4948049485683441,
      "learning_rate": 9.752890241511184e-06,
      "loss": 0.4845,
      "step": 7746
    },
    {
      "epoch": 0.3555137442063237,
      "grad_norm": 0.4300050735473633,
      "learning_rate": 9.752814108455998e-06,
      "loss": 0.3378,
      "step": 7747
    },
    {
      "epoch": 0.3555596347115782,
      "grad_norm": 0.42569500207901,
      "learning_rate": 9.752737963971794e-06,
      "loss": 0.2983,
      "step": 7748
    },
    {
      "epoch": 0.35560552521683264,
      "grad_norm": 0.4897690713405609,
      "learning_rate": 9.752661808058755e-06,
      "loss": 0.5083,
      "step": 7749
    },
    {
      "epoch": 0.3556514157220871,
      "grad_norm": 0.4683336913585663,
      "learning_rate": 9.752585640717064e-06,
      "loss": 0.349,
      "step": 7750
    },
    {
      "epoch": 0.3556973062273416,
      "grad_norm": 0.4705594778060913,
      "learning_rate": 9.752509461946904e-06,
      "loss": 0.4051,
      "step": 7751
    },
    {
      "epoch": 0.355743196732596,
      "grad_norm": 0.4658791720867157,
      "learning_rate": 9.752433271748461e-06,
      "loss": 0.4634,
      "step": 7752
    },
    {
      "epoch": 0.35578908723785047,
      "grad_norm": 0.4247986674308777,
      "learning_rate": 9.752357070121915e-06,
      "loss": 0.3238,
      "step": 7753
    },
    {
      "epoch": 0.35583497774310496,
      "grad_norm": 0.46490156650543213,
      "learning_rate": 9.752280857067447e-06,
      "loss": 0.4495,
      "step": 7754
    },
    {
      "epoch": 0.3558808682483594,
      "grad_norm": 0.4360750615596771,
      "learning_rate": 9.752204632585248e-06,
      "loss": 0.3791,
      "step": 7755
    },
    {
      "epoch": 0.3559267587536139,
      "grad_norm": 0.485958456993103,
      "learning_rate": 9.752128396675493e-06,
      "loss": 0.4704,
      "step": 7756
    },
    {
      "epoch": 0.35597264925886835,
      "grad_norm": 0.4442468583583832,
      "learning_rate": 9.752052149338371e-06,
      "loss": 0.3851,
      "step": 7757
    },
    {
      "epoch": 0.3560185397641228,
      "grad_norm": 0.467679888010025,
      "learning_rate": 9.751975890574062e-06,
      "loss": 0.4581,
      "step": 7758
    },
    {
      "epoch": 0.3560644302693773,
      "grad_norm": 0.4374620318412781,
      "learning_rate": 9.751899620382753e-06,
      "loss": 0.3498,
      "step": 7759
    },
    {
      "epoch": 0.35611032077463173,
      "grad_norm": 0.49808269739151,
      "learning_rate": 9.751823338764624e-06,
      "loss": 0.4232,
      "step": 7760
    },
    {
      "epoch": 0.35615621127988617,
      "grad_norm": 0.417671799659729,
      "learning_rate": 9.751747045719859e-06,
      "loss": 0.3399,
      "step": 7761
    },
    {
      "epoch": 0.35620210178514067,
      "grad_norm": 0.4990861713886261,
      "learning_rate": 9.751670741248642e-06,
      "loss": 0.4508,
      "step": 7762
    },
    {
      "epoch": 0.3562479922903951,
      "grad_norm": 0.46671056747436523,
      "learning_rate": 9.751594425351156e-06,
      "loss": 0.3873,
      "step": 7763
    },
    {
      "epoch": 0.35629388279564955,
      "grad_norm": 0.45195499062538147,
      "learning_rate": 9.751518098027586e-06,
      "loss": 0.3415,
      "step": 7764
    },
    {
      "epoch": 0.35633977330090405,
      "grad_norm": 0.46051305532455444,
      "learning_rate": 9.751441759278115e-06,
      "loss": 0.4018,
      "step": 7765
    },
    {
      "epoch": 0.3563856638061585,
      "grad_norm": 0.4263821244239807,
      "learning_rate": 9.751365409102926e-06,
      "loss": 0.3276,
      "step": 7766
    },
    {
      "epoch": 0.356431554311413,
      "grad_norm": 0.427919864654541,
      "learning_rate": 9.751289047502201e-06,
      "loss": 0.3558,
      "step": 7767
    },
    {
      "epoch": 0.35647744481666743,
      "grad_norm": 0.47430238127708435,
      "learning_rate": 9.751212674476129e-06,
      "loss": 0.4146,
      "step": 7768
    },
    {
      "epoch": 0.3565233353219219,
      "grad_norm": 0.4680333733558655,
      "learning_rate": 9.751136290024888e-06,
      "loss": 0.3563,
      "step": 7769
    },
    {
      "epoch": 0.3565692258271764,
      "grad_norm": 0.46310386061668396,
      "learning_rate": 9.751059894148662e-06,
      "loss": 0.4163,
      "step": 7770
    },
    {
      "epoch": 0.3566151163324308,
      "grad_norm": 0.44909363985061646,
      "learning_rate": 9.750983486847638e-06,
      "loss": 0.3926,
      "step": 7771
    },
    {
      "epoch": 0.35666100683768526,
      "grad_norm": 0.4788450002670288,
      "learning_rate": 9.750907068121999e-06,
      "loss": 0.4853,
      "step": 7772
    },
    {
      "epoch": 0.35670689734293975,
      "grad_norm": 0.47060585021972656,
      "learning_rate": 9.750830637971928e-06,
      "loss": 0.4284,
      "step": 7773
    },
    {
      "epoch": 0.3567527878481942,
      "grad_norm": 0.4730147123336792,
      "learning_rate": 9.750754196397607e-06,
      "loss": 0.403,
      "step": 7774
    },
    {
      "epoch": 0.3567986783534487,
      "grad_norm": 0.4556496739387512,
      "learning_rate": 9.750677743399223e-06,
      "loss": 0.3824,
      "step": 7775
    },
    {
      "epoch": 0.35684456885870314,
      "grad_norm": 0.45856332778930664,
      "learning_rate": 9.750601278976958e-06,
      "loss": 0.3533,
      "step": 7776
    },
    {
      "epoch": 0.3568904593639576,
      "grad_norm": 0.4799988567829132,
      "learning_rate": 9.750524803130996e-06,
      "loss": 0.4198,
      "step": 7777
    },
    {
      "epoch": 0.3569363498692121,
      "grad_norm": 0.46967068314552307,
      "learning_rate": 9.75044831586152e-06,
      "loss": 0.3939,
      "step": 7778
    },
    {
      "epoch": 0.3569822403744665,
      "grad_norm": 0.44241034984588623,
      "learning_rate": 9.750371817168718e-06,
      "loss": 0.3768,
      "step": 7779
    },
    {
      "epoch": 0.35702813087972096,
      "grad_norm": 0.4487215280532837,
      "learning_rate": 9.750295307052769e-06,
      "loss": 0.381,
      "step": 7780
    },
    {
      "epoch": 0.35707402138497546,
      "grad_norm": 0.4378577470779419,
      "learning_rate": 9.750218785513859e-06,
      "loss": 0.3629,
      "step": 7781
    },
    {
      "epoch": 0.3571199118902299,
      "grad_norm": 0.42855754494667053,
      "learning_rate": 9.750142252552172e-06,
      "loss": 0.3382,
      "step": 7782
    },
    {
      "epoch": 0.3571658023954844,
      "grad_norm": 0.47133708000183105,
      "learning_rate": 9.750065708167894e-06,
      "loss": 0.3997,
      "step": 7783
    },
    {
      "epoch": 0.35721169290073884,
      "grad_norm": 0.4432305693626404,
      "learning_rate": 9.749989152361205e-06,
      "loss": 0.3429,
      "step": 7784
    },
    {
      "epoch": 0.3572575834059933,
      "grad_norm": 0.44957345724105835,
      "learning_rate": 9.749912585132292e-06,
      "loss": 0.4048,
      "step": 7785
    },
    {
      "epoch": 0.3573034739112478,
      "grad_norm": 0.44443538784980774,
      "learning_rate": 9.749836006481338e-06,
      "loss": 0.35,
      "step": 7786
    },
    {
      "epoch": 0.3573493644165022,
      "grad_norm": 0.470224529504776,
      "learning_rate": 9.749759416408528e-06,
      "loss": 0.4396,
      "step": 7787
    },
    {
      "epoch": 0.35739525492175667,
      "grad_norm": 0.47980430722236633,
      "learning_rate": 9.749682814914046e-06,
      "loss": 0.4742,
      "step": 7788
    },
    {
      "epoch": 0.35744114542701116,
      "grad_norm": 0.45683225989341736,
      "learning_rate": 9.749606201998075e-06,
      "loss": 0.4464,
      "step": 7789
    },
    {
      "epoch": 0.3574870359322656,
      "grad_norm": 0.5236878991127014,
      "learning_rate": 9.7495295776608e-06,
      "loss": 0.4779,
      "step": 7790
    },
    {
      "epoch": 0.3575329264375201,
      "grad_norm": 0.5000109672546387,
      "learning_rate": 9.749452941902405e-06,
      "loss": 0.4626,
      "step": 7791
    },
    {
      "epoch": 0.35757881694277455,
      "grad_norm": 0.4409351944923401,
      "learning_rate": 9.749376294723076e-06,
      "loss": 0.3602,
      "step": 7792
    },
    {
      "epoch": 0.357624707448029,
      "grad_norm": 0.45808207988739014,
      "learning_rate": 9.749299636122996e-06,
      "loss": 0.3734,
      "step": 7793
    },
    {
      "epoch": 0.3576705979532835,
      "grad_norm": 0.41938477754592896,
      "learning_rate": 9.749222966102347e-06,
      "loss": 0.3579,
      "step": 7794
    },
    {
      "epoch": 0.35771648845853793,
      "grad_norm": 0.43687528371810913,
      "learning_rate": 9.749146284661318e-06,
      "loss": 0.3778,
      "step": 7795
    },
    {
      "epoch": 0.35776237896379237,
      "grad_norm": 0.8167298436164856,
      "learning_rate": 9.74906959180009e-06,
      "loss": 0.411,
      "step": 7796
    },
    {
      "epoch": 0.35780826946904687,
      "grad_norm": 0.49253273010253906,
      "learning_rate": 9.74899288751885e-06,
      "loss": 0.4197,
      "step": 7797
    },
    {
      "epoch": 0.3578541599743013,
      "grad_norm": 0.47256723046302795,
      "learning_rate": 9.74891617181778e-06,
      "loss": 0.4295,
      "step": 7798
    },
    {
      "epoch": 0.35790005047955575,
      "grad_norm": 0.45953455567359924,
      "learning_rate": 9.748839444697065e-06,
      "loss": 0.4315,
      "step": 7799
    },
    {
      "epoch": 0.35794594098481025,
      "grad_norm": 0.45818227529525757,
      "learning_rate": 9.748762706156889e-06,
      "loss": 0.4282,
      "step": 7800
    },
    {
      "epoch": 0.3579918314900647,
      "grad_norm": 0.4343860149383545,
      "learning_rate": 9.74868595619744e-06,
      "loss": 0.365,
      "step": 7801
    },
    {
      "epoch": 0.3580377219953192,
      "grad_norm": 0.44190502166748047,
      "learning_rate": 9.748609194818897e-06,
      "loss": 0.341,
      "step": 7802
    },
    {
      "epoch": 0.35808361250057363,
      "grad_norm": 0.47795814275741577,
      "learning_rate": 9.74853242202145e-06,
      "loss": 0.4164,
      "step": 7803
    },
    {
      "epoch": 0.3581295030058281,
      "grad_norm": 0.5763908624649048,
      "learning_rate": 9.748455637805279e-06,
      "loss": 0.413,
      "step": 7804
    },
    {
      "epoch": 0.3581753935110826,
      "grad_norm": 0.4751807451248169,
      "learning_rate": 9.748378842170573e-06,
      "loss": 0.4114,
      "step": 7805
    },
    {
      "epoch": 0.358221284016337,
      "grad_norm": 0.5041840076446533,
      "learning_rate": 9.748302035117513e-06,
      "loss": 0.4775,
      "step": 7806
    },
    {
      "epoch": 0.35826717452159146,
      "grad_norm": 0.46125584840774536,
      "learning_rate": 9.748225216646287e-06,
      "loss": 0.3733,
      "step": 7807
    },
    {
      "epoch": 0.35831306502684596,
      "grad_norm": 0.4681585431098938,
      "learning_rate": 9.748148386757077e-06,
      "loss": 0.3873,
      "step": 7808
    },
    {
      "epoch": 0.3583589555321004,
      "grad_norm": 0.4707982540130615,
      "learning_rate": 9.748071545450068e-06,
      "loss": 0.3958,
      "step": 7809
    },
    {
      "epoch": 0.3584048460373549,
      "grad_norm": 0.4922158718109131,
      "learning_rate": 9.747994692725446e-06,
      "loss": 0.4335,
      "step": 7810
    },
    {
      "epoch": 0.35845073654260934,
      "grad_norm": 0.5148698687553406,
      "learning_rate": 9.747917828583396e-06,
      "loss": 0.4249,
      "step": 7811
    },
    {
      "epoch": 0.3584966270478638,
      "grad_norm": 0.5053566098213196,
      "learning_rate": 9.747840953024102e-06,
      "loss": 0.4917,
      "step": 7812
    },
    {
      "epoch": 0.3585425175531183,
      "grad_norm": 0.5171065330505371,
      "learning_rate": 9.747764066047746e-06,
      "loss": 0.5039,
      "step": 7813
    },
    {
      "epoch": 0.3585884080583727,
      "grad_norm": 0.4390053451061249,
      "learning_rate": 9.74768716765452e-06,
      "loss": 0.3239,
      "step": 7814
    },
    {
      "epoch": 0.35863429856362716,
      "grad_norm": 0.46710720658302307,
      "learning_rate": 9.747610257844602e-06,
      "loss": 0.4435,
      "step": 7815
    },
    {
      "epoch": 0.35868018906888166,
      "grad_norm": 0.4621230661869049,
      "learning_rate": 9.74753333661818e-06,
      "loss": 0.3884,
      "step": 7816
    },
    {
      "epoch": 0.3587260795741361,
      "grad_norm": 0.4727157652378082,
      "learning_rate": 9.747456403975441e-06,
      "loss": 0.4455,
      "step": 7817
    },
    {
      "epoch": 0.3587719700793906,
      "grad_norm": 0.4456329345703125,
      "learning_rate": 9.747379459916566e-06,
      "loss": 0.4084,
      "step": 7818
    },
    {
      "epoch": 0.35881786058464504,
      "grad_norm": 0.5070009827613831,
      "learning_rate": 9.747302504441742e-06,
      "loss": 0.4945,
      "step": 7819
    },
    {
      "epoch": 0.3588637510898995,
      "grad_norm": 0.4655400216579437,
      "learning_rate": 9.747225537551153e-06,
      "loss": 0.4418,
      "step": 7820
    },
    {
      "epoch": 0.358909641595154,
      "grad_norm": 0.435698002576828,
      "learning_rate": 9.747148559244986e-06,
      "loss": 0.3099,
      "step": 7821
    },
    {
      "epoch": 0.3589555321004084,
      "grad_norm": 0.4887925982475281,
      "learning_rate": 9.747071569523423e-06,
      "loss": 0.4227,
      "step": 7822
    },
    {
      "epoch": 0.35900142260566287,
      "grad_norm": 0.47531765699386597,
      "learning_rate": 9.746994568386653e-06,
      "loss": 0.3816,
      "step": 7823
    },
    {
      "epoch": 0.35904731311091737,
      "grad_norm": 0.48864513635635376,
      "learning_rate": 9.746917555834858e-06,
      "loss": 0.4512,
      "step": 7824
    },
    {
      "epoch": 0.3590932036161718,
      "grad_norm": 0.46279168128967285,
      "learning_rate": 9.746840531868228e-06,
      "loss": 0.3494,
      "step": 7825
    },
    {
      "epoch": 0.35913909412142625,
      "grad_norm": 0.5045490264892578,
      "learning_rate": 9.746763496486939e-06,
      "loss": 0.4512,
      "step": 7826
    },
    {
      "epoch": 0.35918498462668075,
      "grad_norm": 0.4746337831020355,
      "learning_rate": 9.746686449691185e-06,
      "loss": 0.4295,
      "step": 7827
    },
    {
      "epoch": 0.3592308751319352,
      "grad_norm": 0.4743708074092865,
      "learning_rate": 9.74660939148115e-06,
      "loss": 0.418,
      "step": 7828
    },
    {
      "epoch": 0.3592767656371897,
      "grad_norm": 0.4391838014125824,
      "learning_rate": 9.746532321857014e-06,
      "loss": 0.3247,
      "step": 7829
    },
    {
      "epoch": 0.35932265614244413,
      "grad_norm": 0.4868067800998688,
      "learning_rate": 9.746455240818967e-06,
      "loss": 0.4604,
      "step": 7830
    },
    {
      "epoch": 0.3593685466476986,
      "grad_norm": 0.4518764615058899,
      "learning_rate": 9.746378148367194e-06,
      "loss": 0.3991,
      "step": 7831
    },
    {
      "epoch": 0.35941443715295307,
      "grad_norm": 0.485799640417099,
      "learning_rate": 9.746301044501877e-06,
      "loss": 0.461,
      "step": 7832
    },
    {
      "epoch": 0.3594603276582075,
      "grad_norm": 0.44626227021217346,
      "learning_rate": 9.746223929223208e-06,
      "loss": 0.379,
      "step": 7833
    },
    {
      "epoch": 0.35950621816346195,
      "grad_norm": 0.45438483357429504,
      "learning_rate": 9.746146802531367e-06,
      "loss": 0.4061,
      "step": 7834
    },
    {
      "epoch": 0.35955210866871645,
      "grad_norm": 0.46258604526519775,
      "learning_rate": 9.746069664426539e-06,
      "loss": 0.3958,
      "step": 7835
    },
    {
      "epoch": 0.3595979991739709,
      "grad_norm": 0.47691595554351807,
      "learning_rate": 9.745992514908912e-06,
      "loss": 0.4187,
      "step": 7836
    },
    {
      "epoch": 0.3596438896792254,
      "grad_norm": 0.5064228773117065,
      "learning_rate": 9.745915353978672e-06,
      "loss": 0.4322,
      "step": 7837
    },
    {
      "epoch": 0.35968978018447983,
      "grad_norm": 0.5110331773757935,
      "learning_rate": 9.745838181636003e-06,
      "loss": 0.5039,
      "step": 7838
    },
    {
      "epoch": 0.3597356706897343,
      "grad_norm": 0.505457878112793,
      "learning_rate": 9.74576099788109e-06,
      "loss": 0.522,
      "step": 7839
    },
    {
      "epoch": 0.3597815611949888,
      "grad_norm": 0.45371755957603455,
      "learning_rate": 9.74568380271412e-06,
      "loss": 0.389,
      "step": 7840
    },
    {
      "epoch": 0.3598274517002432,
      "grad_norm": 0.483137845993042,
      "learning_rate": 9.74560659613528e-06,
      "loss": 0.4681,
      "step": 7841
    },
    {
      "epoch": 0.35987334220549766,
      "grad_norm": 0.446307510137558,
      "learning_rate": 9.745529378144752e-06,
      "loss": 0.3258,
      "step": 7842
    },
    {
      "epoch": 0.35991923271075216,
      "grad_norm": 0.49514302611351013,
      "learning_rate": 9.745452148742725e-06,
      "loss": 0.498,
      "step": 7843
    },
    {
      "epoch": 0.3599651232160066,
      "grad_norm": 0.48106908798217773,
      "learning_rate": 9.745374907929382e-06,
      "loss": 0.4203,
      "step": 7844
    },
    {
      "epoch": 0.3600110137212611,
      "grad_norm": 0.4576026201248169,
      "learning_rate": 9.745297655704911e-06,
      "loss": 0.4265,
      "step": 7845
    },
    {
      "epoch": 0.36005690422651554,
      "grad_norm": 0.4870549440383911,
      "learning_rate": 9.745220392069497e-06,
      "loss": 0.421,
      "step": 7846
    },
    {
      "epoch": 0.36010279473177,
      "grad_norm": 0.4612952768802643,
      "learning_rate": 9.745143117023323e-06,
      "loss": 0.409,
      "step": 7847
    },
    {
      "epoch": 0.3601486852370245,
      "grad_norm": 0.46829164028167725,
      "learning_rate": 9.74506583056658e-06,
      "loss": 0.3973,
      "step": 7848
    },
    {
      "epoch": 0.3601945757422789,
      "grad_norm": 0.4556679427623749,
      "learning_rate": 9.74498853269945e-06,
      "loss": 0.3589,
      "step": 7849
    },
    {
      "epoch": 0.36024046624753336,
      "grad_norm": 0.4898398816585541,
      "learning_rate": 9.744911223422122e-06,
      "loss": 0.4219,
      "step": 7850
    },
    {
      "epoch": 0.36028635675278786,
      "grad_norm": 0.4399726986885071,
      "learning_rate": 9.744833902734778e-06,
      "loss": 0.3717,
      "step": 7851
    },
    {
      "epoch": 0.3603322472580423,
      "grad_norm": 0.469087690114975,
      "learning_rate": 9.744756570637608e-06,
      "loss": 0.4642,
      "step": 7852
    },
    {
      "epoch": 0.36037813776329675,
      "grad_norm": 0.5035542249679565,
      "learning_rate": 9.744679227130793e-06,
      "loss": 0.4944,
      "step": 7853
    },
    {
      "epoch": 0.36042402826855124,
      "grad_norm": 0.4788002371788025,
      "learning_rate": 9.744601872214525e-06,
      "loss": 0.4799,
      "step": 7854
    },
    {
      "epoch": 0.3604699187738057,
      "grad_norm": 0.46393391489982605,
      "learning_rate": 9.744524505888985e-06,
      "loss": 0.4026,
      "step": 7855
    },
    {
      "epoch": 0.3605158092790602,
      "grad_norm": 0.4808913469314575,
      "learning_rate": 9.744447128154362e-06,
      "loss": 0.3827,
      "step": 7856
    },
    {
      "epoch": 0.3605616997843146,
      "grad_norm": 0.4478481411933899,
      "learning_rate": 9.74436973901084e-06,
      "loss": 0.3692,
      "step": 7857
    },
    {
      "epoch": 0.36060759028956907,
      "grad_norm": 0.47368311882019043,
      "learning_rate": 9.744292338458606e-06,
      "loss": 0.4095,
      "step": 7858
    },
    {
      "epoch": 0.36065348079482357,
      "grad_norm": 0.4372916519641876,
      "learning_rate": 9.744214926497849e-06,
      "loss": 0.3465,
      "step": 7859
    },
    {
      "epoch": 0.360699371300078,
      "grad_norm": 0.467933714389801,
      "learning_rate": 9.74413750312875e-06,
      "loss": 0.4185,
      "step": 7860
    },
    {
      "epoch": 0.36074526180533245,
      "grad_norm": 0.4576081335544586,
      "learning_rate": 9.744060068351498e-06,
      "loss": 0.4262,
      "step": 7861
    },
    {
      "epoch": 0.36079115231058695,
      "grad_norm": 0.48531052470207214,
      "learning_rate": 9.743982622166277e-06,
      "loss": 0.412,
      "step": 7862
    },
    {
      "epoch": 0.3608370428158414,
      "grad_norm": 0.4853992164134979,
      "learning_rate": 9.743905164573278e-06,
      "loss": 0.4408,
      "step": 7863
    },
    {
      "epoch": 0.3608829333210959,
      "grad_norm": 0.44082745909690857,
      "learning_rate": 9.743827695572682e-06,
      "loss": 0.3457,
      "step": 7864
    },
    {
      "epoch": 0.36092882382635033,
      "grad_norm": 0.45144349336624146,
      "learning_rate": 9.74375021516468e-06,
      "loss": 0.3713,
      "step": 7865
    },
    {
      "epoch": 0.3609747143316048,
      "grad_norm": 0.5377853512763977,
      "learning_rate": 9.743672723349453e-06,
      "loss": 0.4886,
      "step": 7866
    },
    {
      "epoch": 0.36102060483685927,
      "grad_norm": 0.46922898292541504,
      "learning_rate": 9.743595220127194e-06,
      "loss": 0.3851,
      "step": 7867
    },
    {
      "epoch": 0.3610664953421137,
      "grad_norm": 0.46638351678848267,
      "learning_rate": 9.743517705498083e-06,
      "loss": 0.3693,
      "step": 7868
    },
    {
      "epoch": 0.36111238584736816,
      "grad_norm": 0.45565682649612427,
      "learning_rate": 9.743440179462309e-06,
      "loss": 0.3911,
      "step": 7869
    },
    {
      "epoch": 0.36115827635262265,
      "grad_norm": 0.49627816677093506,
      "learning_rate": 9.743362642020059e-06,
      "loss": 0.4815,
      "step": 7870
    },
    {
      "epoch": 0.3612041668578771,
      "grad_norm": 0.46992817521095276,
      "learning_rate": 9.743285093171518e-06,
      "loss": 0.4337,
      "step": 7871
    },
    {
      "epoch": 0.3612500573631316,
      "grad_norm": 0.4777149260044098,
      "learning_rate": 9.743207532916873e-06,
      "loss": 0.3943,
      "step": 7872
    },
    {
      "epoch": 0.36129594786838604,
      "grad_norm": 0.47868648171424866,
      "learning_rate": 9.743129961256315e-06,
      "loss": 0.4345,
      "step": 7873
    },
    {
      "epoch": 0.3613418383736405,
      "grad_norm": 0.509576141834259,
      "learning_rate": 9.743052378190021e-06,
      "loss": 0.5466,
      "step": 7874
    },
    {
      "epoch": 0.361387728878895,
      "grad_norm": 0.45552173256874084,
      "learning_rate": 9.742974783718187e-06,
      "loss": 0.4081,
      "step": 7875
    },
    {
      "epoch": 0.3614336193841494,
      "grad_norm": 0.5026187300682068,
      "learning_rate": 9.742897177840992e-06,
      "loss": 0.4541,
      "step": 7876
    },
    {
      "epoch": 0.36147950988940386,
      "grad_norm": 0.4564884305000305,
      "learning_rate": 9.74281956055863e-06,
      "loss": 0.3673,
      "step": 7877
    },
    {
      "epoch": 0.36152540039465836,
      "grad_norm": 0.4623177945613861,
      "learning_rate": 9.742741931871282e-06,
      "loss": 0.3979,
      "step": 7878
    },
    {
      "epoch": 0.3615712908999128,
      "grad_norm": 0.47553497552871704,
      "learning_rate": 9.742664291779137e-06,
      "loss": 0.3921,
      "step": 7879
    },
    {
      "epoch": 0.3616171814051673,
      "grad_norm": 0.48133906722068787,
      "learning_rate": 9.742586640282382e-06,
      "loss": 0.4832,
      "step": 7880
    },
    {
      "epoch": 0.36166307191042174,
      "grad_norm": 0.4605104327201843,
      "learning_rate": 9.7425089773812e-06,
      "loss": 0.3594,
      "step": 7881
    },
    {
      "epoch": 0.3617089624156762,
      "grad_norm": 0.4825238287448883,
      "learning_rate": 9.742431303075783e-06,
      "loss": 0.4512,
      "step": 7882
    },
    {
      "epoch": 0.3617548529209307,
      "grad_norm": 0.48937928676605225,
      "learning_rate": 9.742353617366315e-06,
      "loss": 0.439,
      "step": 7883
    },
    {
      "epoch": 0.3618007434261851,
      "grad_norm": 0.4327901303768158,
      "learning_rate": 9.742275920252982e-06,
      "loss": 0.3527,
      "step": 7884
    },
    {
      "epoch": 0.36184663393143957,
      "grad_norm": 0.4557718336582184,
      "learning_rate": 9.742198211735974e-06,
      "loss": 0.3619,
      "step": 7885
    },
    {
      "epoch": 0.36189252443669406,
      "grad_norm": 0.48134058713912964,
      "learning_rate": 9.742120491815475e-06,
      "loss": 0.4231,
      "step": 7886
    },
    {
      "epoch": 0.3619384149419485,
      "grad_norm": 0.47641870379447937,
      "learning_rate": 9.742042760491673e-06,
      "loss": 0.3821,
      "step": 7887
    },
    {
      "epoch": 0.36198430544720295,
      "grad_norm": 0.47012123465538025,
      "learning_rate": 9.741965017764756e-06,
      "loss": 0.4714,
      "step": 7888
    },
    {
      "epoch": 0.36203019595245745,
      "grad_norm": 0.4664852023124695,
      "learning_rate": 9.741887263634909e-06,
      "loss": 0.4504,
      "step": 7889
    },
    {
      "epoch": 0.3620760864577119,
      "grad_norm": 0.46899867057800293,
      "learning_rate": 9.741809498102319e-06,
      "loss": 0.4597,
      "step": 7890
    },
    {
      "epoch": 0.3621219769629664,
      "grad_norm": 0.4533671736717224,
      "learning_rate": 9.741731721167173e-06,
      "loss": 0.3454,
      "step": 7891
    },
    {
      "epoch": 0.3621678674682208,
      "grad_norm": 0.4495394229888916,
      "learning_rate": 9.74165393282966e-06,
      "loss": 0.3776,
      "step": 7892
    },
    {
      "epoch": 0.36221375797347527,
      "grad_norm": 0.4663272798061371,
      "learning_rate": 9.741576133089967e-06,
      "loss": 0.3986,
      "step": 7893
    },
    {
      "epoch": 0.36225964847872977,
      "grad_norm": 0.4765564203262329,
      "learning_rate": 9.741498321948278e-06,
      "loss": 0.4229,
      "step": 7894
    },
    {
      "epoch": 0.3623055389839842,
      "grad_norm": 0.45327427983283997,
      "learning_rate": 9.741420499404782e-06,
      "loss": 0.4103,
      "step": 7895
    },
    {
      "epoch": 0.36235142948923865,
      "grad_norm": 0.44875141978263855,
      "learning_rate": 9.741342665459667e-06,
      "loss": 0.398,
      "step": 7896
    },
    {
      "epoch": 0.36239731999449315,
      "grad_norm": 0.47597238421440125,
      "learning_rate": 9.74126482011312e-06,
      "loss": 0.43,
      "step": 7897
    },
    {
      "epoch": 0.3624432104997476,
      "grad_norm": 0.49926310777664185,
      "learning_rate": 9.741186963365326e-06,
      "loss": 0.4118,
      "step": 7898
    },
    {
      "epoch": 0.3624891010050021,
      "grad_norm": 0.4713848829269409,
      "learning_rate": 9.741109095216473e-06,
      "loss": 0.409,
      "step": 7899
    },
    {
      "epoch": 0.36253499151025653,
      "grad_norm": 0.4256654381752014,
      "learning_rate": 9.74103121566675e-06,
      "loss": 0.3286,
      "step": 7900
    },
    {
      "epoch": 0.362580882015511,
      "grad_norm": 0.41809964179992676,
      "learning_rate": 9.740953324716345e-06,
      "loss": 0.3188,
      "step": 7901
    },
    {
      "epoch": 0.3626267725207655,
      "grad_norm": 0.4738941788673401,
      "learning_rate": 9.74087542236544e-06,
      "loss": 0.4476,
      "step": 7902
    },
    {
      "epoch": 0.3626726630260199,
      "grad_norm": 0.6595299243927002,
      "learning_rate": 9.740797508614228e-06,
      "loss": 0.3794,
      "step": 7903
    },
    {
      "epoch": 0.36271855353127436,
      "grad_norm": 0.44186052680015564,
      "learning_rate": 9.740719583462894e-06,
      "loss": 0.3737,
      "step": 7904
    },
    {
      "epoch": 0.36276444403652885,
      "grad_norm": 0.48513317108154297,
      "learning_rate": 9.740641646911627e-06,
      "loss": 0.3911,
      "step": 7905
    },
    {
      "epoch": 0.3628103345417833,
      "grad_norm": 0.4776601493358612,
      "learning_rate": 9.74056369896061e-06,
      "loss": 0.4289,
      "step": 7906
    },
    {
      "epoch": 0.3628562250470378,
      "grad_norm": 0.4962489902973175,
      "learning_rate": 9.740485739610037e-06,
      "loss": 0.4146,
      "step": 7907
    },
    {
      "epoch": 0.36290211555229224,
      "grad_norm": 0.4357406795024872,
      "learning_rate": 9.74040776886009e-06,
      "loss": 0.3859,
      "step": 7908
    },
    {
      "epoch": 0.3629480060575467,
      "grad_norm": 0.5032374262809753,
      "learning_rate": 9.740329786710959e-06,
      "loss": 0.4832,
      "step": 7909
    },
    {
      "epoch": 0.3629938965628012,
      "grad_norm": 0.45092374086380005,
      "learning_rate": 9.740251793162829e-06,
      "loss": 0.3764,
      "step": 7910
    },
    {
      "epoch": 0.3630397870680556,
      "grad_norm": 0.4660680294036865,
      "learning_rate": 9.74017378821589e-06,
      "loss": 0.4938,
      "step": 7911
    },
    {
      "epoch": 0.36308567757331006,
      "grad_norm": 0.4867881238460541,
      "learning_rate": 9.74009577187033e-06,
      "loss": 0.4193,
      "step": 7912
    },
    {
      "epoch": 0.36313156807856456,
      "grad_norm": 0.44200262427330017,
      "learning_rate": 9.740017744126337e-06,
      "loss": 0.3725,
      "step": 7913
    },
    {
      "epoch": 0.363177458583819,
      "grad_norm": 0.45537450909614563,
      "learning_rate": 9.739939704984095e-06,
      "loss": 0.3911,
      "step": 7914
    },
    {
      "epoch": 0.36322334908907344,
      "grad_norm": 0.49914830923080444,
      "learning_rate": 9.739861654443796e-06,
      "loss": 0.504,
      "step": 7915
    },
    {
      "epoch": 0.36326923959432794,
      "grad_norm": 0.4785454273223877,
      "learning_rate": 9.739783592505625e-06,
      "loss": 0.4042,
      "step": 7916
    },
    {
      "epoch": 0.3633151300995824,
      "grad_norm": 0.4819380044937134,
      "learning_rate": 9.73970551916977e-06,
      "loss": 0.4635,
      "step": 7917
    },
    {
      "epoch": 0.3633610206048369,
      "grad_norm": 0.43958884477615356,
      "learning_rate": 9.739627434436419e-06,
      "loss": 0.3693,
      "step": 7918
    },
    {
      "epoch": 0.3634069111100913,
      "grad_norm": 0.47859689593315125,
      "learning_rate": 9.73954933830576e-06,
      "loss": 0.4124,
      "step": 7919
    },
    {
      "epoch": 0.36345280161534577,
      "grad_norm": 0.49066272377967834,
      "learning_rate": 9.73947123077798e-06,
      "loss": 0.4944,
      "step": 7920
    },
    {
      "epoch": 0.36349869212060026,
      "grad_norm": 0.47038328647613525,
      "learning_rate": 9.739393111853268e-06,
      "loss": 0.4648,
      "step": 7921
    },
    {
      "epoch": 0.3635445826258547,
      "grad_norm": 0.4451914429664612,
      "learning_rate": 9.739314981531811e-06,
      "loss": 0.3687,
      "step": 7922
    },
    {
      "epoch": 0.36359047313110915,
      "grad_norm": 0.458329975605011,
      "learning_rate": 9.739236839813799e-06,
      "loss": 0.4393,
      "step": 7923
    },
    {
      "epoch": 0.36363636363636365,
      "grad_norm": 0.4524228274822235,
      "learning_rate": 9.739158686699417e-06,
      "loss": 0.3649,
      "step": 7924
    },
    {
      "epoch": 0.3636822541416181,
      "grad_norm": 0.4961557388305664,
      "learning_rate": 9.739080522188854e-06,
      "loss": 0.4529,
      "step": 7925
    },
    {
      "epoch": 0.3637281446468726,
      "grad_norm": 0.46224445104599,
      "learning_rate": 9.739002346282298e-06,
      "loss": 0.4069,
      "step": 7926
    },
    {
      "epoch": 0.36377403515212703,
      "grad_norm": 0.4642353057861328,
      "learning_rate": 9.738924158979937e-06,
      "loss": 0.3756,
      "step": 7927
    },
    {
      "epoch": 0.36381992565738147,
      "grad_norm": 0.523690402507782,
      "learning_rate": 9.73884596028196e-06,
      "loss": 0.4648,
      "step": 7928
    },
    {
      "epoch": 0.36386581616263597,
      "grad_norm": 0.4337020516395569,
      "learning_rate": 9.738767750188554e-06,
      "loss": 0.3333,
      "step": 7929
    },
    {
      "epoch": 0.3639117066678904,
      "grad_norm": 0.4346909523010254,
      "learning_rate": 9.738689528699906e-06,
      "loss": 0.3229,
      "step": 7930
    },
    {
      "epoch": 0.36395759717314485,
      "grad_norm": 0.4247654676437378,
      "learning_rate": 9.738611295816205e-06,
      "loss": 0.3347,
      "step": 7931
    },
    {
      "epoch": 0.36400348767839935,
      "grad_norm": 0.4586181640625,
      "learning_rate": 9.73853305153764e-06,
      "loss": 0.4501,
      "step": 7932
    },
    {
      "epoch": 0.3640493781836538,
      "grad_norm": 0.4227321147918701,
      "learning_rate": 9.738454795864398e-06,
      "loss": 0.3257,
      "step": 7933
    },
    {
      "epoch": 0.3640952686889083,
      "grad_norm": 0.8993532657623291,
      "learning_rate": 9.73837652879667e-06,
      "loss": 0.4287,
      "step": 7934
    },
    {
      "epoch": 0.36414115919416273,
      "grad_norm": 0.4714774489402771,
      "learning_rate": 9.738298250334639e-06,
      "loss": 0.398,
      "step": 7935
    },
    {
      "epoch": 0.3641870496994172,
      "grad_norm": 0.45705947279930115,
      "learning_rate": 9.738219960478499e-06,
      "loss": 0.4146,
      "step": 7936
    },
    {
      "epoch": 0.3642329402046717,
      "grad_norm": 0.449897825717926,
      "learning_rate": 9.738141659228433e-06,
      "loss": 0.3609,
      "step": 7937
    },
    {
      "epoch": 0.3642788307099261,
      "grad_norm": 0.45911169052124023,
      "learning_rate": 9.738063346584631e-06,
      "loss": 0.36,
      "step": 7938
    },
    {
      "epoch": 0.36432472121518056,
      "grad_norm": 0.46098849177360535,
      "learning_rate": 9.737985022547285e-06,
      "loss": 0.3791,
      "step": 7939
    },
    {
      "epoch": 0.36437061172043506,
      "grad_norm": 0.44855889678001404,
      "learning_rate": 9.737906687116578e-06,
      "loss": 0.375,
      "step": 7940
    },
    {
      "epoch": 0.3644165022256895,
      "grad_norm": 0.479921817779541,
      "learning_rate": 9.737828340292703e-06,
      "loss": 0.4512,
      "step": 7941
    },
    {
      "epoch": 0.36446239273094394,
      "grad_norm": 0.4942409098148346,
      "learning_rate": 9.737749982075844e-06,
      "loss": 0.4476,
      "step": 7942
    },
    {
      "epoch": 0.36450828323619844,
      "grad_norm": 0.5237117409706116,
      "learning_rate": 9.737671612466192e-06,
      "loss": 0.409,
      "step": 7943
    },
    {
      "epoch": 0.3645541737414529,
      "grad_norm": 0.45513567328453064,
      "learning_rate": 9.737593231463937e-06,
      "loss": 0.3522,
      "step": 7944
    },
    {
      "epoch": 0.3646000642467074,
      "grad_norm": 0.5181505680084229,
      "learning_rate": 9.737514839069263e-06,
      "loss": 0.5133,
      "step": 7945
    },
    {
      "epoch": 0.3646459547519618,
      "grad_norm": 0.4719248116016388,
      "learning_rate": 9.737436435282362e-06,
      "loss": 0.3977,
      "step": 7946
    },
    {
      "epoch": 0.36469184525721626,
      "grad_norm": 0.5288018584251404,
      "learning_rate": 9.737358020103422e-06,
      "loss": 0.534,
      "step": 7947
    },
    {
      "epoch": 0.36473773576247076,
      "grad_norm": 0.4950268268585205,
      "learning_rate": 9.73727959353263e-06,
      "loss": 0.4654,
      "step": 7948
    },
    {
      "epoch": 0.3647836262677252,
      "grad_norm": 0.44992390275001526,
      "learning_rate": 9.737201155570177e-06,
      "loss": 0.4266,
      "step": 7949
    },
    {
      "epoch": 0.36482951677297965,
      "grad_norm": 0.4925025403499603,
      "learning_rate": 9.73712270621625e-06,
      "loss": 0.4857,
      "step": 7950
    },
    {
      "epoch": 0.36487540727823414,
      "grad_norm": 0.5202951431274414,
      "learning_rate": 9.737044245471036e-06,
      "loss": 0.4753,
      "step": 7951
    },
    {
      "epoch": 0.3649212977834886,
      "grad_norm": 0.4630596935749054,
      "learning_rate": 9.736965773334727e-06,
      "loss": 0.4286,
      "step": 7952
    },
    {
      "epoch": 0.3649671882887431,
      "grad_norm": 0.4617745280265808,
      "learning_rate": 9.736887289807511e-06,
      "loss": 0.4045,
      "step": 7953
    },
    {
      "epoch": 0.3650130787939975,
      "grad_norm": 0.4143560230731964,
      "learning_rate": 9.736808794889577e-06,
      "loss": 0.2965,
      "step": 7954
    },
    {
      "epoch": 0.36505896929925197,
      "grad_norm": 0.5091694593429565,
      "learning_rate": 9.736730288581111e-06,
      "loss": 0.4693,
      "step": 7955
    },
    {
      "epoch": 0.36510485980450647,
      "grad_norm": 0.4266557991504669,
      "learning_rate": 9.736651770882304e-06,
      "loss": 0.3079,
      "step": 7956
    },
    {
      "epoch": 0.3651507503097609,
      "grad_norm": 0.5005029439926147,
      "learning_rate": 9.736573241793345e-06,
      "loss": 0.4579,
      "step": 7957
    },
    {
      "epoch": 0.36519664081501535,
      "grad_norm": 0.4957512319087982,
      "learning_rate": 9.736494701314421e-06,
      "loss": 0.4787,
      "step": 7958
    },
    {
      "epoch": 0.36524253132026985,
      "grad_norm": 0.4641667306423187,
      "learning_rate": 9.736416149445724e-06,
      "loss": 0.3949,
      "step": 7959
    },
    {
      "epoch": 0.3652884218255243,
      "grad_norm": 0.4726676642894745,
      "learning_rate": 9.736337586187438e-06,
      "loss": 0.3537,
      "step": 7960
    },
    {
      "epoch": 0.3653343123307788,
      "grad_norm": 0.505027174949646,
      "learning_rate": 9.736259011539758e-06,
      "loss": 0.452,
      "step": 7961
    },
    {
      "epoch": 0.36538020283603323,
      "grad_norm": 0.4358639717102051,
      "learning_rate": 9.73618042550287e-06,
      "loss": 0.3539,
      "step": 7962
    },
    {
      "epoch": 0.36542609334128767,
      "grad_norm": 0.47085753083229065,
      "learning_rate": 9.73610182807696e-06,
      "loss": 0.4127,
      "step": 7963
    },
    {
      "epoch": 0.36547198384654217,
      "grad_norm": 0.5275803208351135,
      "learning_rate": 9.736023219262223e-06,
      "loss": 0.4694,
      "step": 7964
    },
    {
      "epoch": 0.3655178743517966,
      "grad_norm": 1.7177000045776367,
      "learning_rate": 9.735944599058842e-06,
      "loss": 0.3908,
      "step": 7965
    },
    {
      "epoch": 0.36556376485705105,
      "grad_norm": 0.4439907670021057,
      "learning_rate": 9.735865967467011e-06,
      "loss": 0.3698,
      "step": 7966
    },
    {
      "epoch": 0.36560965536230555,
      "grad_norm": 0.4635026454925537,
      "learning_rate": 9.735787324486916e-06,
      "loss": 0.3921,
      "step": 7967
    },
    {
      "epoch": 0.36565554586756,
      "grad_norm": 0.6005432605743408,
      "learning_rate": 9.735708670118748e-06,
      "loss": 0.5426,
      "step": 7968
    },
    {
      "epoch": 0.3657014363728145,
      "grad_norm": 0.47062450647354126,
      "learning_rate": 9.735630004362695e-06,
      "loss": 0.3658,
      "step": 7969
    },
    {
      "epoch": 0.36574732687806893,
      "grad_norm": 0.46867257356643677,
      "learning_rate": 9.735551327218946e-06,
      "loss": 0.411,
      "step": 7970
    },
    {
      "epoch": 0.3657932173833234,
      "grad_norm": 0.5301876664161682,
      "learning_rate": 9.735472638687691e-06,
      "loss": 0.5041,
      "step": 7971
    },
    {
      "epoch": 0.3658391078885779,
      "grad_norm": 0.4820837080478668,
      "learning_rate": 9.735393938769118e-06,
      "loss": 0.3985,
      "step": 7972
    },
    {
      "epoch": 0.3658849983938323,
      "grad_norm": 0.45955127477645874,
      "learning_rate": 9.735315227463418e-06,
      "loss": 0.3785,
      "step": 7973
    },
    {
      "epoch": 0.36593088889908676,
      "grad_norm": 0.4670088589191437,
      "learning_rate": 9.73523650477078e-06,
      "loss": 0.3965,
      "step": 7974
    },
    {
      "epoch": 0.36597677940434126,
      "grad_norm": 0.4967685639858246,
      "learning_rate": 9.735157770691391e-06,
      "loss": 0.4598,
      "step": 7975
    },
    {
      "epoch": 0.3660226699095957,
      "grad_norm": 0.49250465631484985,
      "learning_rate": 9.735079025225443e-06,
      "loss": 0.459,
      "step": 7976
    },
    {
      "epoch": 0.36606856041485014,
      "grad_norm": 0.47690075635910034,
      "learning_rate": 9.735000268373125e-06,
      "loss": 0.396,
      "step": 7977
    },
    {
      "epoch": 0.36611445092010464,
      "grad_norm": 0.4766157269477844,
      "learning_rate": 9.734921500134623e-06,
      "loss": 0.4069,
      "step": 7978
    },
    {
      "epoch": 0.3661603414253591,
      "grad_norm": 0.45853912830352783,
      "learning_rate": 9.734842720510132e-06,
      "loss": 0.3928,
      "step": 7979
    },
    {
      "epoch": 0.3662062319306136,
      "grad_norm": 0.4406105577945709,
      "learning_rate": 9.734763929499839e-06,
      "loss": 0.4138,
      "step": 7980
    },
    {
      "epoch": 0.366252122435868,
      "grad_norm": 0.49110254645347595,
      "learning_rate": 9.734685127103932e-06,
      "loss": 0.4184,
      "step": 7981
    },
    {
      "epoch": 0.36629801294112246,
      "grad_norm": 0.4742239713668823,
      "learning_rate": 9.734606313322601e-06,
      "loss": 0.3729,
      "step": 7982
    },
    {
      "epoch": 0.36634390344637696,
      "grad_norm": 0.486428827047348,
      "learning_rate": 9.734527488156037e-06,
      "loss": 0.4794,
      "step": 7983
    },
    {
      "epoch": 0.3663897939516314,
      "grad_norm": 0.4881732165813446,
      "learning_rate": 9.734448651604427e-06,
      "loss": 0.5084,
      "step": 7984
    },
    {
      "epoch": 0.36643568445688585,
      "grad_norm": 0.4637189507484436,
      "learning_rate": 9.734369803667965e-06,
      "loss": 0.4171,
      "step": 7985
    },
    {
      "epoch": 0.36648157496214034,
      "grad_norm": 0.5018987655639648,
      "learning_rate": 9.734290944346837e-06,
      "loss": 0.439,
      "step": 7986
    },
    {
      "epoch": 0.3665274654673948,
      "grad_norm": 0.497918963432312,
      "learning_rate": 9.734212073641231e-06,
      "loss": 0.4926,
      "step": 7987
    },
    {
      "epoch": 0.3665733559726493,
      "grad_norm": 0.42361322045326233,
      "learning_rate": 9.734133191551342e-06,
      "loss": 0.3401,
      "step": 7988
    },
    {
      "epoch": 0.3666192464779037,
      "grad_norm": 0.4947502613067627,
      "learning_rate": 9.734054298077355e-06,
      "loss": 0.3858,
      "step": 7989
    },
    {
      "epoch": 0.36666513698315817,
      "grad_norm": 0.43561750650405884,
      "learning_rate": 9.733975393219462e-06,
      "loss": 0.3477,
      "step": 7990
    },
    {
      "epoch": 0.36671102748841267,
      "grad_norm": 0.45899152755737305,
      "learning_rate": 9.733896476977853e-06,
      "loss": 0.414,
      "step": 7991
    },
    {
      "epoch": 0.3667569179936671,
      "grad_norm": 0.45467379689216614,
      "learning_rate": 9.733817549352716e-06,
      "loss": 0.39,
      "step": 7992
    },
    {
      "epoch": 0.36680280849892155,
      "grad_norm": 0.4570685625076294,
      "learning_rate": 9.733738610344244e-06,
      "loss": 0.3725,
      "step": 7993
    },
    {
      "epoch": 0.36684869900417605,
      "grad_norm": 0.4900420904159546,
      "learning_rate": 9.733659659952621e-06,
      "loss": 0.4643,
      "step": 7994
    },
    {
      "epoch": 0.3668945895094305,
      "grad_norm": 0.4762815535068512,
      "learning_rate": 9.733580698178045e-06,
      "loss": 0.409,
      "step": 7995
    },
    {
      "epoch": 0.366940480014685,
      "grad_norm": 0.44649678468704224,
      "learning_rate": 9.733501725020698e-06,
      "loss": 0.3446,
      "step": 7996
    },
    {
      "epoch": 0.36698637051993943,
      "grad_norm": 0.4945557415485382,
      "learning_rate": 9.733422740480774e-06,
      "loss": 0.4329,
      "step": 7997
    },
    {
      "epoch": 0.3670322610251939,
      "grad_norm": 0.4655408561229706,
      "learning_rate": 9.733343744558462e-06,
      "loss": 0.4047,
      "step": 7998
    },
    {
      "epoch": 0.36707815153044837,
      "grad_norm": 0.46442753076553345,
      "learning_rate": 9.733264737253954e-06,
      "loss": 0.4269,
      "step": 7999
    },
    {
      "epoch": 0.3671240420357028,
      "grad_norm": 0.49488914012908936,
      "learning_rate": 9.733185718567435e-06,
      "loss": 0.4925,
      "step": 8000
    },
    {
      "epoch": 0.36716993254095726,
      "grad_norm": 0.45052027702331543,
      "learning_rate": 9.733106688499101e-06,
      "loss": 0.3999,
      "step": 8001
    },
    {
      "epoch": 0.36721582304621175,
      "grad_norm": 0.4414416551589966,
      "learning_rate": 9.733027647049138e-06,
      "loss": 0.4014,
      "step": 8002
    },
    {
      "epoch": 0.3672617135514662,
      "grad_norm": 0.4591817259788513,
      "learning_rate": 9.732948594217737e-06,
      "loss": 0.4023,
      "step": 8003
    },
    {
      "epoch": 0.36730760405672064,
      "grad_norm": 0.4903697669506073,
      "learning_rate": 9.73286953000509e-06,
      "loss": 0.419,
      "step": 8004
    },
    {
      "epoch": 0.36735349456197514,
      "grad_norm": 0.45190897583961487,
      "learning_rate": 9.732790454411384e-06,
      "loss": 0.3792,
      "step": 8005
    },
    {
      "epoch": 0.3673993850672296,
      "grad_norm": 0.4193302392959595,
      "learning_rate": 9.732711367436812e-06,
      "loss": 0.3274,
      "step": 8006
    },
    {
      "epoch": 0.3674452755724841,
      "grad_norm": 0.4411812722682953,
      "learning_rate": 9.73263226908156e-06,
      "loss": 0.3432,
      "step": 8007
    },
    {
      "epoch": 0.3674911660777385,
      "grad_norm": 0.44763392210006714,
      "learning_rate": 9.732553159345824e-06,
      "loss": 0.3955,
      "step": 8008
    },
    {
      "epoch": 0.36753705658299296,
      "grad_norm": 0.4963451325893402,
      "learning_rate": 9.73247403822979e-06,
      "loss": 0.3898,
      "step": 8009
    },
    {
      "epoch": 0.36758294708824746,
      "grad_norm": 0.44995906949043274,
      "learning_rate": 9.732394905733649e-06,
      "loss": 0.3448,
      "step": 8010
    },
    {
      "epoch": 0.3676288375935019,
      "grad_norm": 0.42833393812179565,
      "learning_rate": 9.732315761857594e-06,
      "loss": 0.3531,
      "step": 8011
    },
    {
      "epoch": 0.36767472809875634,
      "grad_norm": 0.48283034563064575,
      "learning_rate": 9.73223660660181e-06,
      "loss": 0.4453,
      "step": 8012
    },
    {
      "epoch": 0.36772061860401084,
      "grad_norm": 0.5067793726921082,
      "learning_rate": 9.732157439966494e-06,
      "loss": 0.5105,
      "step": 8013
    },
    {
      "epoch": 0.3677665091092653,
      "grad_norm": 0.46469753980636597,
      "learning_rate": 9.73207826195183e-06,
      "loss": 0.4428,
      "step": 8014
    },
    {
      "epoch": 0.3678123996145198,
      "grad_norm": 0.518373966217041,
      "learning_rate": 9.731999072558014e-06,
      "loss": 0.4657,
      "step": 8015
    },
    {
      "epoch": 0.3678582901197742,
      "grad_norm": 0.4774475395679474,
      "learning_rate": 9.731919871785231e-06,
      "loss": 0.4674,
      "step": 8016
    },
    {
      "epoch": 0.36790418062502867,
      "grad_norm": 0.4660840928554535,
      "learning_rate": 9.731840659633675e-06,
      "loss": 0.4101,
      "step": 8017
    },
    {
      "epoch": 0.36795007113028316,
      "grad_norm": 0.4802314043045044,
      "learning_rate": 9.731761436103536e-06,
      "loss": 0.4052,
      "step": 8018
    },
    {
      "epoch": 0.3679959616355376,
      "grad_norm": 0.48872867226600647,
      "learning_rate": 9.731682201195005e-06,
      "loss": 0.4028,
      "step": 8019
    },
    {
      "epoch": 0.36804185214079205,
      "grad_norm": 0.4654121696949005,
      "learning_rate": 9.73160295490827e-06,
      "loss": 0.4347,
      "step": 8020
    },
    {
      "epoch": 0.36808774264604655,
      "grad_norm": 0.4395206868648529,
      "learning_rate": 9.731523697243523e-06,
      "loss": 0.3254,
      "step": 8021
    },
    {
      "epoch": 0.368133633151301,
      "grad_norm": 0.5039449334144592,
      "learning_rate": 9.731444428200957e-06,
      "loss": 0.5234,
      "step": 8022
    },
    {
      "epoch": 0.3681795236565555,
      "grad_norm": 0.4755762219429016,
      "learning_rate": 9.73136514778076e-06,
      "loss": 0.481,
      "step": 8023
    },
    {
      "epoch": 0.3682254141618099,
      "grad_norm": 0.4889451563358307,
      "learning_rate": 9.731285855983121e-06,
      "loss": 0.3608,
      "step": 8024
    },
    {
      "epoch": 0.36827130466706437,
      "grad_norm": 0.47623538970947266,
      "learning_rate": 9.731206552808234e-06,
      "loss": 0.4812,
      "step": 8025
    },
    {
      "epoch": 0.36831719517231887,
      "grad_norm": 0.4962598979473114,
      "learning_rate": 9.731127238256288e-06,
      "loss": 0.474,
      "step": 8026
    },
    {
      "epoch": 0.3683630856775733,
      "grad_norm": 0.4775947034358978,
      "learning_rate": 9.731047912327475e-06,
      "loss": 0.4698,
      "step": 8027
    },
    {
      "epoch": 0.36840897618282775,
      "grad_norm": 0.4568853974342346,
      "learning_rate": 9.730968575021985e-06,
      "loss": 0.4123,
      "step": 8028
    },
    {
      "epoch": 0.36845486668808225,
      "grad_norm": 0.4713488817214966,
      "learning_rate": 9.730889226340007e-06,
      "loss": 0.4179,
      "step": 8029
    },
    {
      "epoch": 0.3685007571933367,
      "grad_norm": 0.4416423439979553,
      "learning_rate": 9.730809866281737e-06,
      "loss": 0.3621,
      "step": 8030
    },
    {
      "epoch": 0.36854664769859113,
      "grad_norm": 0.46662425994873047,
      "learning_rate": 9.73073049484736e-06,
      "loss": 0.3324,
      "step": 8031
    },
    {
      "epoch": 0.36859253820384563,
      "grad_norm": 0.4579404592514038,
      "learning_rate": 9.730651112037069e-06,
      "loss": 0.4361,
      "step": 8032
    },
    {
      "epoch": 0.3686384287091001,
      "grad_norm": 0.4758075475692749,
      "learning_rate": 9.730571717851055e-06,
      "loss": 0.3991,
      "step": 8033
    },
    {
      "epoch": 0.36868431921435457,
      "grad_norm": 0.45792677998542786,
      "learning_rate": 9.73049231228951e-06,
      "loss": 0.3729,
      "step": 8034
    },
    {
      "epoch": 0.368730209719609,
      "grad_norm": 0.4402088224887848,
      "learning_rate": 9.730412895352623e-06,
      "loss": 0.42,
      "step": 8035
    },
    {
      "epoch": 0.36877610022486346,
      "grad_norm": 0.5387215614318848,
      "learning_rate": 9.730333467040587e-06,
      "loss": 0.4696,
      "step": 8036
    },
    {
      "epoch": 0.36882199073011795,
      "grad_norm": 0.45912817120552063,
      "learning_rate": 9.730254027353592e-06,
      "loss": 0.4138,
      "step": 8037
    },
    {
      "epoch": 0.3688678812353724,
      "grad_norm": 0.5113257169723511,
      "learning_rate": 9.730174576291828e-06,
      "loss": 0.4988,
      "step": 8038
    },
    {
      "epoch": 0.36891377174062684,
      "grad_norm": 0.47522401809692383,
      "learning_rate": 9.730095113855487e-06,
      "loss": 0.4432,
      "step": 8039
    },
    {
      "epoch": 0.36895966224588134,
      "grad_norm": 0.4611656069755554,
      "learning_rate": 9.73001564004476e-06,
      "loss": 0.3948,
      "step": 8040
    },
    {
      "epoch": 0.3690055527511358,
      "grad_norm": 0.4605605900287628,
      "learning_rate": 9.729936154859839e-06,
      "loss": 0.4,
      "step": 8041
    },
    {
      "epoch": 0.3690514432563903,
      "grad_norm": 0.5019216537475586,
      "learning_rate": 9.729856658300914e-06,
      "loss": 0.4318,
      "step": 8042
    },
    {
      "epoch": 0.3690973337616447,
      "grad_norm": 0.4993398189544678,
      "learning_rate": 9.729777150368177e-06,
      "loss": 0.4073,
      "step": 8043
    },
    {
      "epoch": 0.36914322426689916,
      "grad_norm": 0.41586291790008545,
      "learning_rate": 9.729697631061819e-06,
      "loss": 0.3428,
      "step": 8044
    },
    {
      "epoch": 0.36918911477215366,
      "grad_norm": 0.4678192436695099,
      "learning_rate": 9.72961810038203e-06,
      "loss": 0.3977,
      "step": 8045
    },
    {
      "epoch": 0.3692350052774081,
      "grad_norm": 0.8091902732849121,
      "learning_rate": 9.729538558329002e-06,
      "loss": 0.5191,
      "step": 8046
    },
    {
      "epoch": 0.36928089578266254,
      "grad_norm": 0.45810991525650024,
      "learning_rate": 9.729459004902927e-06,
      "loss": 0.3857,
      "step": 8047
    },
    {
      "epoch": 0.36932678628791704,
      "grad_norm": 0.46541959047317505,
      "learning_rate": 9.729379440103996e-06,
      "loss": 0.4144,
      "step": 8048
    },
    {
      "epoch": 0.3693726767931715,
      "grad_norm": 0.49193358421325684,
      "learning_rate": 9.7292998639324e-06,
      "loss": 0.3851,
      "step": 8049
    },
    {
      "epoch": 0.369418567298426,
      "grad_norm": 0.4602566957473755,
      "learning_rate": 9.72922027638833e-06,
      "loss": 0.3705,
      "step": 8050
    },
    {
      "epoch": 0.3694644578036804,
      "grad_norm": 0.46869969367980957,
      "learning_rate": 9.729140677471978e-06,
      "loss": 0.401,
      "step": 8051
    },
    {
      "epoch": 0.36951034830893487,
      "grad_norm": 0.4285276234149933,
      "learning_rate": 9.729061067183534e-06,
      "loss": 0.336,
      "step": 8052
    },
    {
      "epoch": 0.36955623881418936,
      "grad_norm": 0.45687344670295715,
      "learning_rate": 9.728981445523193e-06,
      "loss": 0.3566,
      "step": 8053
    },
    {
      "epoch": 0.3696021293194438,
      "grad_norm": 0.47448474168777466,
      "learning_rate": 9.728901812491142e-06,
      "loss": 0.3981,
      "step": 8054
    },
    {
      "epoch": 0.36964801982469825,
      "grad_norm": 0.44643843173980713,
      "learning_rate": 9.728822168087577e-06,
      "loss": 0.3645,
      "step": 8055
    },
    {
      "epoch": 0.36969391032995275,
      "grad_norm": 0.42920738458633423,
      "learning_rate": 9.728742512312686e-06,
      "loss": 0.3616,
      "step": 8056
    },
    {
      "epoch": 0.3697398008352072,
      "grad_norm": 0.46367207169532776,
      "learning_rate": 9.728662845166661e-06,
      "loss": 0.4826,
      "step": 8057
    },
    {
      "epoch": 0.36978569134046163,
      "grad_norm": 0.482564777135849,
      "learning_rate": 9.728583166649694e-06,
      "loss": 0.3564,
      "step": 8058
    },
    {
      "epoch": 0.36983158184571613,
      "grad_norm": 0.5418311953544617,
      "learning_rate": 9.728503476761978e-06,
      "loss": 0.5025,
      "step": 8059
    },
    {
      "epoch": 0.36987747235097057,
      "grad_norm": 0.4505927562713623,
      "learning_rate": 9.728423775503703e-06,
      "loss": 0.399,
      "step": 8060
    },
    {
      "epoch": 0.36992336285622507,
      "grad_norm": 0.4269683361053467,
      "learning_rate": 9.728344062875061e-06,
      "loss": 0.3556,
      "step": 8061
    },
    {
      "epoch": 0.3699692533614795,
      "grad_norm": 0.5010607242584229,
      "learning_rate": 9.728264338876245e-06,
      "loss": 0.3637,
      "step": 8062
    },
    {
      "epoch": 0.37001514386673395,
      "grad_norm": 0.4785168468952179,
      "learning_rate": 9.728184603507445e-06,
      "loss": 0.4574,
      "step": 8063
    },
    {
      "epoch": 0.37006103437198845,
      "grad_norm": 0.4363779127597809,
      "learning_rate": 9.728104856768853e-06,
      "loss": 0.3622,
      "step": 8064
    },
    {
      "epoch": 0.3701069248772429,
      "grad_norm": 0.4038386046886444,
      "learning_rate": 9.72802509866066e-06,
      "loss": 0.3071,
      "step": 8065
    },
    {
      "epoch": 0.37015281538249734,
      "grad_norm": 0.4707005023956299,
      "learning_rate": 9.72794532918306e-06,
      "loss": 0.4239,
      "step": 8066
    },
    {
      "epoch": 0.37019870588775183,
      "grad_norm": 0.4823363125324249,
      "learning_rate": 9.727865548336245e-06,
      "loss": 0.4239,
      "step": 8067
    },
    {
      "epoch": 0.3702445963930063,
      "grad_norm": 0.450297474861145,
      "learning_rate": 9.727785756120402e-06,
      "loss": 0.3783,
      "step": 8068
    },
    {
      "epoch": 0.3702904868982608,
      "grad_norm": 0.5109229683876038,
      "learning_rate": 9.72770595253573e-06,
      "loss": 0.4809,
      "step": 8069
    },
    {
      "epoch": 0.3703363774035152,
      "grad_norm": 0.4691585600376129,
      "learning_rate": 9.727626137582416e-06,
      "loss": 0.3848,
      "step": 8070
    },
    {
      "epoch": 0.37038226790876966,
      "grad_norm": 0.46388256549835205,
      "learning_rate": 9.727546311260651e-06,
      "loss": 0.4518,
      "step": 8071
    },
    {
      "epoch": 0.37042815841402416,
      "grad_norm": 0.47018569707870483,
      "learning_rate": 9.727466473570632e-06,
      "loss": 0.3919,
      "step": 8072
    },
    {
      "epoch": 0.3704740489192786,
      "grad_norm": 0.43438851833343506,
      "learning_rate": 9.727386624512546e-06,
      "loss": 0.3402,
      "step": 8073
    },
    {
      "epoch": 0.37051993942453304,
      "grad_norm": 0.4712080955505371,
      "learning_rate": 9.727306764086591e-06,
      "loss": 0.4324,
      "step": 8074
    },
    {
      "epoch": 0.37056582992978754,
      "grad_norm": 0.48610761761665344,
      "learning_rate": 9.727226892292952e-06,
      "loss": 0.4226,
      "step": 8075
    },
    {
      "epoch": 0.370611720435042,
      "grad_norm": 0.4481717050075531,
      "learning_rate": 9.727147009131824e-06,
      "loss": 0.3749,
      "step": 8076
    },
    {
      "epoch": 0.3706576109402965,
      "grad_norm": 0.4812678098678589,
      "learning_rate": 9.727067114603402e-06,
      "loss": 0.4426,
      "step": 8077
    },
    {
      "epoch": 0.3707035014455509,
      "grad_norm": 0.44761893153190613,
      "learning_rate": 9.726987208707873e-06,
      "loss": 0.3825,
      "step": 8078
    },
    {
      "epoch": 0.37074939195080536,
      "grad_norm": 0.4644887149333954,
      "learning_rate": 9.72690729144543e-06,
      "loss": 0.4108,
      "step": 8079
    },
    {
      "epoch": 0.37079528245605986,
      "grad_norm": 0.4892721176147461,
      "learning_rate": 9.726827362816272e-06,
      "loss": 0.4612,
      "step": 8080
    },
    {
      "epoch": 0.3708411729613143,
      "grad_norm": 0.47281813621520996,
      "learning_rate": 9.726747422820582e-06,
      "loss": 0.3941,
      "step": 8081
    },
    {
      "epoch": 0.37088706346656874,
      "grad_norm": 0.4691774249076843,
      "learning_rate": 9.726667471458558e-06,
      "loss": 0.4512,
      "step": 8082
    },
    {
      "epoch": 0.37093295397182324,
      "grad_norm": 0.4633810818195343,
      "learning_rate": 9.726587508730389e-06,
      "loss": 0.4224,
      "step": 8083
    },
    {
      "epoch": 0.3709788444770777,
      "grad_norm": 0.45314016938209534,
      "learning_rate": 9.726507534636271e-06,
      "loss": 0.4136,
      "step": 8084
    },
    {
      "epoch": 0.3710247349823322,
      "grad_norm": 0.45336097478866577,
      "learning_rate": 9.726427549176393e-06,
      "loss": 0.4142,
      "step": 8085
    },
    {
      "epoch": 0.3710706254875866,
      "grad_norm": 0.4467010796070099,
      "learning_rate": 9.726347552350948e-06,
      "loss": 0.3816,
      "step": 8086
    },
    {
      "epoch": 0.37111651599284107,
      "grad_norm": 0.45809540152549744,
      "learning_rate": 9.72626754416013e-06,
      "loss": 0.4042,
      "step": 8087
    },
    {
      "epoch": 0.37116240649809557,
      "grad_norm": 0.526657223701477,
      "learning_rate": 9.72618752460413e-06,
      "loss": 0.3665,
      "step": 8088
    },
    {
      "epoch": 0.37120829700335,
      "grad_norm": 0.48596763610839844,
      "learning_rate": 9.726107493683139e-06,
      "loss": 0.4757,
      "step": 8089
    },
    {
      "epoch": 0.37125418750860445,
      "grad_norm": 0.4452357590198517,
      "learning_rate": 9.726027451397351e-06,
      "loss": 0.4281,
      "step": 8090
    },
    {
      "epoch": 0.37130007801385895,
      "grad_norm": 0.441429078578949,
      "learning_rate": 9.725947397746961e-06,
      "loss": 0.349,
      "step": 8091
    },
    {
      "epoch": 0.3713459685191134,
      "grad_norm": 0.517781674861908,
      "learning_rate": 9.725867332732157e-06,
      "loss": 0.5259,
      "step": 8092
    },
    {
      "epoch": 0.37139185902436783,
      "grad_norm": 0.487958699464798,
      "learning_rate": 9.725787256353134e-06,
      "loss": 0.4962,
      "step": 8093
    },
    {
      "epoch": 0.37143774952962233,
      "grad_norm": 0.46005934476852417,
      "learning_rate": 9.725707168610085e-06,
      "loss": 0.3649,
      "step": 8094
    },
    {
      "epoch": 0.37148364003487677,
      "grad_norm": 0.48686784505844116,
      "learning_rate": 9.725627069503201e-06,
      "loss": 0.42,
      "step": 8095
    },
    {
      "epoch": 0.37152953054013127,
      "grad_norm": 0.45250093936920166,
      "learning_rate": 9.725546959032676e-06,
      "loss": 0.3806,
      "step": 8096
    },
    {
      "epoch": 0.3715754210453857,
      "grad_norm": 0.4368855953216553,
      "learning_rate": 9.725466837198702e-06,
      "loss": 0.3566,
      "step": 8097
    },
    {
      "epoch": 0.37162131155064015,
      "grad_norm": 0.49231866002082825,
      "learning_rate": 9.725386704001471e-06,
      "loss": 0.451,
      "step": 8098
    },
    {
      "epoch": 0.37166720205589465,
      "grad_norm": 0.45820948481559753,
      "learning_rate": 9.725306559441177e-06,
      "loss": 0.3865,
      "step": 8099
    },
    {
      "epoch": 0.3717130925611491,
      "grad_norm": 0.4619249999523163,
      "learning_rate": 9.725226403518011e-06,
      "loss": 0.4132,
      "step": 8100
    },
    {
      "epoch": 0.37175898306640354,
      "grad_norm": 0.4838564097881317,
      "learning_rate": 9.725146236232168e-06,
      "loss": 0.4788,
      "step": 8101
    },
    {
      "epoch": 0.37180487357165803,
      "grad_norm": 0.43584656715393066,
      "learning_rate": 9.725066057583839e-06,
      "loss": 0.3823,
      "step": 8102
    },
    {
      "epoch": 0.3718507640769125,
      "grad_norm": 0.43821296095848083,
      "learning_rate": 9.724985867573218e-06,
      "loss": 0.3435,
      "step": 8103
    },
    {
      "epoch": 0.371896654582167,
      "grad_norm": 0.4871059060096741,
      "learning_rate": 9.724905666200496e-06,
      "loss": 0.431,
      "step": 8104
    },
    {
      "epoch": 0.3719425450874214,
      "grad_norm": 0.47945672273635864,
      "learning_rate": 9.724825453465867e-06,
      "loss": 0.4234,
      "step": 8105
    },
    {
      "epoch": 0.37198843559267586,
      "grad_norm": 0.46462875604629517,
      "learning_rate": 9.724745229369526e-06,
      "loss": 0.3931,
      "step": 8106
    },
    {
      "epoch": 0.37203432609793036,
      "grad_norm": 0.4686892330646515,
      "learning_rate": 9.724664993911662e-06,
      "loss": 0.3871,
      "step": 8107
    },
    {
      "epoch": 0.3720802166031848,
      "grad_norm": 0.4590422213077545,
      "learning_rate": 9.724584747092471e-06,
      "loss": 0.3704,
      "step": 8108
    },
    {
      "epoch": 0.37212610710843924,
      "grad_norm": 0.43814969062805176,
      "learning_rate": 9.724504488912144e-06,
      "loss": 0.3134,
      "step": 8109
    },
    {
      "epoch": 0.37217199761369374,
      "grad_norm": 0.4633161723613739,
      "learning_rate": 9.724424219370876e-06,
      "loss": 0.3491,
      "step": 8110
    },
    {
      "epoch": 0.3722178881189482,
      "grad_norm": 0.4515477418899536,
      "learning_rate": 9.724343938468857e-06,
      "loss": 0.3521,
      "step": 8111
    },
    {
      "epoch": 0.3722637786242027,
      "grad_norm": 0.4554186761379242,
      "learning_rate": 9.724263646206284e-06,
      "loss": 0.4451,
      "step": 8112
    },
    {
      "epoch": 0.3723096691294571,
      "grad_norm": 0.47241196036338806,
      "learning_rate": 9.724183342583347e-06,
      "loss": 0.3444,
      "step": 8113
    },
    {
      "epoch": 0.37235555963471156,
      "grad_norm": 0.44862687587738037,
      "learning_rate": 9.724103027600239e-06,
      "loss": 0.3639,
      "step": 8114
    },
    {
      "epoch": 0.37240145013996606,
      "grad_norm": 0.4485190510749817,
      "learning_rate": 9.724022701257155e-06,
      "loss": 0.3434,
      "step": 8115
    },
    {
      "epoch": 0.3724473406452205,
      "grad_norm": 0.4773459732532501,
      "learning_rate": 9.723942363554288e-06,
      "loss": 0.4407,
      "step": 8116
    },
    {
      "epoch": 0.37249323115047495,
      "grad_norm": 0.46536871790885925,
      "learning_rate": 9.723862014491829e-06,
      "loss": 0.4085,
      "step": 8117
    },
    {
      "epoch": 0.37253912165572944,
      "grad_norm": 0.4543834924697876,
      "learning_rate": 9.723781654069973e-06,
      "loss": 0.334,
      "step": 8118
    },
    {
      "epoch": 0.3725850121609839,
      "grad_norm": 0.4743187725543976,
      "learning_rate": 9.723701282288914e-06,
      "loss": 0.4437,
      "step": 8119
    },
    {
      "epoch": 0.37263090266623833,
      "grad_norm": 0.5024212002754211,
      "learning_rate": 9.723620899148844e-06,
      "loss": 0.4955,
      "step": 8120
    },
    {
      "epoch": 0.3726767931714928,
      "grad_norm": 0.47235605120658875,
      "learning_rate": 9.723540504649957e-06,
      "loss": 0.4826,
      "step": 8121
    },
    {
      "epoch": 0.37272268367674727,
      "grad_norm": 0.45504891872406006,
      "learning_rate": 9.723460098792446e-06,
      "loss": 0.354,
      "step": 8122
    },
    {
      "epoch": 0.37276857418200177,
      "grad_norm": 0.46430978178977966,
      "learning_rate": 9.723379681576504e-06,
      "loss": 0.4321,
      "step": 8123
    },
    {
      "epoch": 0.3728144646872562,
      "grad_norm": 0.4982248842716217,
      "learning_rate": 9.723299253002324e-06,
      "loss": 0.4926,
      "step": 8124
    },
    {
      "epoch": 0.37286035519251065,
      "grad_norm": 0.44990095496177673,
      "learning_rate": 9.723218813070099e-06,
      "loss": 0.3483,
      "step": 8125
    },
    {
      "epoch": 0.37290624569776515,
      "grad_norm": 0.42564651370048523,
      "learning_rate": 9.723138361780024e-06,
      "loss": 0.3379,
      "step": 8126
    },
    {
      "epoch": 0.3729521362030196,
      "grad_norm": 0.427450031042099,
      "learning_rate": 9.723057899132293e-06,
      "loss": 0.3139,
      "step": 8127
    },
    {
      "epoch": 0.37299802670827403,
      "grad_norm": 0.4523105025291443,
      "learning_rate": 9.722977425127099e-06,
      "loss": 0.438,
      "step": 8128
    },
    {
      "epoch": 0.37304391721352853,
      "grad_norm": 0.4399386942386627,
      "learning_rate": 9.722896939764634e-06,
      "loss": 0.3582,
      "step": 8129
    },
    {
      "epoch": 0.373089807718783,
      "grad_norm": 0.5143728256225586,
      "learning_rate": 9.722816443045091e-06,
      "loss": 0.4195,
      "step": 8130
    },
    {
      "epoch": 0.37313569822403747,
      "grad_norm": 0.4279252588748932,
      "learning_rate": 9.722735934968667e-06,
      "loss": 0.3418,
      "step": 8131
    },
    {
      "epoch": 0.3731815887292919,
      "grad_norm": 0.44557589292526245,
      "learning_rate": 9.722655415535554e-06,
      "loss": 0.3673,
      "step": 8132
    },
    {
      "epoch": 0.37322747923454636,
      "grad_norm": 0.4313417673110962,
      "learning_rate": 9.722574884745944e-06,
      "loss": 0.3499,
      "step": 8133
    },
    {
      "epoch": 0.37327336973980085,
      "grad_norm": 0.45331770181655884,
      "learning_rate": 9.72249434260003e-06,
      "loss": 0.4123,
      "step": 8134
    },
    {
      "epoch": 0.3733192602450553,
      "grad_norm": 0.49247288703918457,
      "learning_rate": 9.72241378909801e-06,
      "loss": 0.497,
      "step": 8135
    },
    {
      "epoch": 0.37336515075030974,
      "grad_norm": 0.5737965703010559,
      "learning_rate": 9.722333224240074e-06,
      "loss": 0.3849,
      "step": 8136
    },
    {
      "epoch": 0.37341104125556424,
      "grad_norm": 0.48814693093299866,
      "learning_rate": 9.722252648026421e-06,
      "loss": 0.4537,
      "step": 8137
    },
    {
      "epoch": 0.3734569317608187,
      "grad_norm": 0.4780582785606384,
      "learning_rate": 9.722172060457236e-06,
      "loss": 0.4296,
      "step": 8138
    },
    {
      "epoch": 0.3735028222660732,
      "grad_norm": 0.4632754921913147,
      "learning_rate": 9.722091461532718e-06,
      "loss": 0.3281,
      "step": 8139
    },
    {
      "epoch": 0.3735487127713276,
      "grad_norm": 0.46364662051200867,
      "learning_rate": 9.722010851253064e-06,
      "loss": 0.3693,
      "step": 8140
    },
    {
      "epoch": 0.37359460327658206,
      "grad_norm": 0.45502471923828125,
      "learning_rate": 9.721930229618461e-06,
      "loss": 0.3474,
      "step": 8141
    },
    {
      "epoch": 0.37364049378183656,
      "grad_norm": 0.5249725580215454,
      "learning_rate": 9.721849596629106e-06,
      "loss": 0.4211,
      "step": 8142
    },
    {
      "epoch": 0.373686384287091,
      "grad_norm": 0.4458678364753723,
      "learning_rate": 9.721768952285195e-06,
      "loss": 0.3686,
      "step": 8143
    },
    {
      "epoch": 0.37373227479234544,
      "grad_norm": 0.46885237097740173,
      "learning_rate": 9.721688296586917e-06,
      "loss": 0.3746,
      "step": 8144
    },
    {
      "epoch": 0.37377816529759994,
      "grad_norm": 0.4442320764064789,
      "learning_rate": 9.721607629534473e-06,
      "loss": 0.3726,
      "step": 8145
    },
    {
      "epoch": 0.3738240558028544,
      "grad_norm": 0.4607435464859009,
      "learning_rate": 9.72152695112805e-06,
      "loss": 0.3858,
      "step": 8146
    },
    {
      "epoch": 0.3738699463081088,
      "grad_norm": 0.45024415850639343,
      "learning_rate": 9.721446261367844e-06,
      "loss": 0.3547,
      "step": 8147
    },
    {
      "epoch": 0.3739158368133633,
      "grad_norm": 0.4726341962814331,
      "learning_rate": 9.721365560254052e-06,
      "loss": 0.4038,
      "step": 8148
    },
    {
      "epoch": 0.37396172731861776,
      "grad_norm": 0.4787328243255615,
      "learning_rate": 9.721284847786865e-06,
      "loss": 0.3894,
      "step": 8149
    },
    {
      "epoch": 0.37400761782387226,
      "grad_norm": 0.4535009562969208,
      "learning_rate": 9.721204123966479e-06,
      "loss": 0.4018,
      "step": 8150
    },
    {
      "epoch": 0.3740535083291267,
      "grad_norm": 0.4782507121562958,
      "learning_rate": 9.721123388793085e-06,
      "loss": 0.4268,
      "step": 8151
    },
    {
      "epoch": 0.37409939883438115,
      "grad_norm": 0.481377512216568,
      "learning_rate": 9.72104264226688e-06,
      "loss": 0.3855,
      "step": 8152
    },
    {
      "epoch": 0.37414528933963564,
      "grad_norm": 0.47644633054733276,
      "learning_rate": 9.720961884388058e-06,
      "loss": 0.3527,
      "step": 8153
    },
    {
      "epoch": 0.3741911798448901,
      "grad_norm": 0.44543230533599854,
      "learning_rate": 9.720881115156813e-06,
      "loss": 0.3561,
      "step": 8154
    },
    {
      "epoch": 0.37423707035014453,
      "grad_norm": 0.4528518319129944,
      "learning_rate": 9.720800334573339e-06,
      "loss": 0.346,
      "step": 8155
    },
    {
      "epoch": 0.374282960855399,
      "grad_norm": 0.4602299630641937,
      "learning_rate": 9.72071954263783e-06,
      "loss": 0.3263,
      "step": 8156
    },
    {
      "epoch": 0.37432885136065347,
      "grad_norm": 0.4406090974807739,
      "learning_rate": 9.720638739350479e-06,
      "loss": 0.3746,
      "step": 8157
    },
    {
      "epoch": 0.37437474186590797,
      "grad_norm": 0.42417415976524353,
      "learning_rate": 9.720557924711483e-06,
      "loss": 0.3472,
      "step": 8158
    },
    {
      "epoch": 0.3744206323711624,
      "grad_norm": 0.41825437545776367,
      "learning_rate": 9.720477098721034e-06,
      "loss": 0.3116,
      "step": 8159
    },
    {
      "epoch": 0.37446652287641685,
      "grad_norm": 0.48889315128326416,
      "learning_rate": 9.72039626137933e-06,
      "loss": 0.422,
      "step": 8160
    },
    {
      "epoch": 0.37451241338167135,
      "grad_norm": 0.4643462002277374,
      "learning_rate": 9.720315412686561e-06,
      "loss": 0.3873,
      "step": 8161
    },
    {
      "epoch": 0.3745583038869258,
      "grad_norm": 0.4061015248298645,
      "learning_rate": 9.720234552642921e-06,
      "loss": 0.3423,
      "step": 8162
    },
    {
      "epoch": 0.37460419439218023,
      "grad_norm": 0.4500768780708313,
      "learning_rate": 9.720153681248612e-06,
      "loss": 0.3727,
      "step": 8163
    },
    {
      "epoch": 0.37465008489743473,
      "grad_norm": 0.4545755088329315,
      "learning_rate": 9.72007279850382e-06,
      "loss": 0.3921,
      "step": 8164
    },
    {
      "epoch": 0.3746959754026892,
      "grad_norm": 0.49224841594696045,
      "learning_rate": 9.71999190440874e-06,
      "loss": 0.4213,
      "step": 8165
    },
    {
      "epoch": 0.37474186590794367,
      "grad_norm": 0.48643729090690613,
      "learning_rate": 9.719910998963574e-06,
      "loss": 0.4964,
      "step": 8166
    },
    {
      "epoch": 0.3747877564131981,
      "grad_norm": 0.5026622414588928,
      "learning_rate": 9.719830082168508e-06,
      "loss": 0.5113,
      "step": 8167
    },
    {
      "epoch": 0.37483364691845256,
      "grad_norm": 0.45205119252204895,
      "learning_rate": 9.719749154023743e-06,
      "loss": 0.3411,
      "step": 8168
    },
    {
      "epoch": 0.37487953742370705,
      "grad_norm": 0.46454983949661255,
      "learning_rate": 9.719668214529468e-06,
      "loss": 0.3569,
      "step": 8169
    },
    {
      "epoch": 0.3749254279289615,
      "grad_norm": 0.44269686937332153,
      "learning_rate": 9.719587263685882e-06,
      "loss": 0.3974,
      "step": 8170
    },
    {
      "epoch": 0.37497131843421594,
      "grad_norm": 0.5285033583641052,
      "learning_rate": 9.719506301493179e-06,
      "loss": 0.4325,
      "step": 8171
    },
    {
      "epoch": 0.37501720893947044,
      "grad_norm": 0.45297661423683167,
      "learning_rate": 9.719425327951552e-06,
      "loss": 0.3534,
      "step": 8172
    },
    {
      "epoch": 0.3750630994447249,
      "grad_norm": 0.4935588240623474,
      "learning_rate": 9.719344343061197e-06,
      "loss": 0.4017,
      "step": 8173
    },
    {
      "epoch": 0.3751089899499794,
      "grad_norm": 0.4304918646812439,
      "learning_rate": 9.719263346822308e-06,
      "loss": 0.3535,
      "step": 8174
    },
    {
      "epoch": 0.3751548804552338,
      "grad_norm": 0.4949699640274048,
      "learning_rate": 9.71918233923508e-06,
      "loss": 0.4339,
      "step": 8175
    },
    {
      "epoch": 0.37520077096048826,
      "grad_norm": 0.4781631827354431,
      "learning_rate": 9.719101320299708e-06,
      "loss": 0.4488,
      "step": 8176
    },
    {
      "epoch": 0.37524666146574276,
      "grad_norm": 0.4888651669025421,
      "learning_rate": 9.719020290016386e-06,
      "loss": 0.4003,
      "step": 8177
    },
    {
      "epoch": 0.3752925519709972,
      "grad_norm": 0.4683087170124054,
      "learning_rate": 9.718939248385311e-06,
      "loss": 0.4143,
      "step": 8178
    },
    {
      "epoch": 0.37533844247625164,
      "grad_norm": 0.46588730812072754,
      "learning_rate": 9.718858195406675e-06,
      "loss": 0.388,
      "step": 8179
    },
    {
      "epoch": 0.37538433298150614,
      "grad_norm": 0.4829517602920532,
      "learning_rate": 9.718777131080676e-06,
      "loss": 0.4694,
      "step": 8180
    },
    {
      "epoch": 0.3754302234867606,
      "grad_norm": 0.4469701945781708,
      "learning_rate": 9.718696055407506e-06,
      "loss": 0.3786,
      "step": 8181
    },
    {
      "epoch": 0.375476113992015,
      "grad_norm": 0.5979227423667908,
      "learning_rate": 9.718614968387361e-06,
      "loss": 0.5714,
      "step": 8182
    },
    {
      "epoch": 0.3755220044972695,
      "grad_norm": 0.45635634660720825,
      "learning_rate": 9.718533870020436e-06,
      "loss": 0.3049,
      "step": 8183
    },
    {
      "epoch": 0.37556789500252397,
      "grad_norm": 0.45526766777038574,
      "learning_rate": 9.718452760306928e-06,
      "loss": 0.3959,
      "step": 8184
    },
    {
      "epoch": 0.37561378550777846,
      "grad_norm": 0.4375819265842438,
      "learning_rate": 9.718371639247031e-06,
      "loss": 0.332,
      "step": 8185
    },
    {
      "epoch": 0.3756596760130329,
      "grad_norm": 0.49725911021232605,
      "learning_rate": 9.718290506840937e-06,
      "loss": 0.4112,
      "step": 8186
    },
    {
      "epoch": 0.37570556651828735,
      "grad_norm": 0.5130476355552673,
      "learning_rate": 9.718209363088844e-06,
      "loss": 0.3664,
      "step": 8187
    },
    {
      "epoch": 0.37575145702354185,
      "grad_norm": 0.48302939534187317,
      "learning_rate": 9.718128207990947e-06,
      "loss": 0.3991,
      "step": 8188
    },
    {
      "epoch": 0.3757973475287963,
      "grad_norm": 0.4441150426864624,
      "learning_rate": 9.71804704154744e-06,
      "loss": 0.3049,
      "step": 8189
    },
    {
      "epoch": 0.37584323803405073,
      "grad_norm": 0.43720924854278564,
      "learning_rate": 9.717965863758521e-06,
      "loss": 0.3473,
      "step": 8190
    },
    {
      "epoch": 0.37588912853930523,
      "grad_norm": 0.45557448267936707,
      "learning_rate": 9.71788467462438e-06,
      "loss": 0.3919,
      "step": 8191
    },
    {
      "epoch": 0.37593501904455967,
      "grad_norm": 0.46237924695014954,
      "learning_rate": 9.717803474145217e-06,
      "loss": 0.4151,
      "step": 8192
    },
    {
      "epoch": 0.37598090954981417,
      "grad_norm": 0.4438866078853607,
      "learning_rate": 9.717722262321227e-06,
      "loss": 0.3338,
      "step": 8193
    },
    {
      "epoch": 0.3760268000550686,
      "grad_norm": 0.5185517072677612,
      "learning_rate": 9.717641039152603e-06,
      "loss": 0.5049,
      "step": 8194
    },
    {
      "epoch": 0.37607269056032305,
      "grad_norm": 0.4654289186000824,
      "learning_rate": 9.71755980463954e-06,
      "loss": 0.3356,
      "step": 8195
    },
    {
      "epoch": 0.37611858106557755,
      "grad_norm": 0.47097015380859375,
      "learning_rate": 9.717478558782236e-06,
      "loss": 0.3879,
      "step": 8196
    },
    {
      "epoch": 0.376164471570832,
      "grad_norm": 0.5115286707878113,
      "learning_rate": 9.717397301580885e-06,
      "loss": 0.5489,
      "step": 8197
    },
    {
      "epoch": 0.37621036207608644,
      "grad_norm": 0.4676591157913208,
      "learning_rate": 9.717316033035679e-06,
      "loss": 0.3835,
      "step": 8198
    },
    {
      "epoch": 0.37625625258134093,
      "grad_norm": 0.44990721344947815,
      "learning_rate": 9.71723475314682e-06,
      "loss": 0.3384,
      "step": 8199
    },
    {
      "epoch": 0.3763021430865954,
      "grad_norm": 0.48706746101379395,
      "learning_rate": 9.7171534619145e-06,
      "loss": 0.4235,
      "step": 8200
    },
    {
      "epoch": 0.3763480335918499,
      "grad_norm": 0.48613953590393066,
      "learning_rate": 9.717072159338914e-06,
      "loss": 0.4891,
      "step": 8201
    },
    {
      "epoch": 0.3763939240971043,
      "grad_norm": 0.5020565390586853,
      "learning_rate": 9.716990845420259e-06,
      "loss": 0.4717,
      "step": 8202
    },
    {
      "epoch": 0.37643981460235876,
      "grad_norm": 0.4507388770580292,
      "learning_rate": 9.71690952015873e-06,
      "loss": 0.3763,
      "step": 8203
    },
    {
      "epoch": 0.37648570510761326,
      "grad_norm": 0.47889065742492676,
      "learning_rate": 9.716828183554521e-06,
      "loss": 0.4376,
      "step": 8204
    },
    {
      "epoch": 0.3765315956128677,
      "grad_norm": 0.460827112197876,
      "learning_rate": 9.716746835607829e-06,
      "loss": 0.3815,
      "step": 8205
    },
    {
      "epoch": 0.37657748611812214,
      "grad_norm": 0.4715343117713928,
      "learning_rate": 9.716665476318851e-06,
      "loss": 0.3768,
      "step": 8206
    },
    {
      "epoch": 0.37662337662337664,
      "grad_norm": 0.449844628572464,
      "learning_rate": 9.71658410568778e-06,
      "loss": 0.3932,
      "step": 8207
    },
    {
      "epoch": 0.3766692671286311,
      "grad_norm": 0.46386027336120605,
      "learning_rate": 9.716502723714813e-06,
      "loss": 0.4176,
      "step": 8208
    },
    {
      "epoch": 0.3767151576338855,
      "grad_norm": 0.4416763484477997,
      "learning_rate": 9.716421330400146e-06,
      "loss": 0.3637,
      "step": 8209
    },
    {
      "epoch": 0.37676104813914,
      "grad_norm": 0.44814369082450867,
      "learning_rate": 9.716339925743972e-06,
      "loss": 0.3526,
      "step": 8210
    },
    {
      "epoch": 0.37680693864439446,
      "grad_norm": 0.48289379477500916,
      "learning_rate": 9.716258509746492e-06,
      "loss": 0.4202,
      "step": 8211
    },
    {
      "epoch": 0.37685282914964896,
      "grad_norm": 0.49049267172813416,
      "learning_rate": 9.716177082407897e-06,
      "loss": 0.4116,
      "step": 8212
    },
    {
      "epoch": 0.3768987196549034,
      "grad_norm": 0.4465479850769043,
      "learning_rate": 9.716095643728385e-06,
      "loss": 0.3506,
      "step": 8213
    },
    {
      "epoch": 0.37694461016015784,
      "grad_norm": 0.47375378012657166,
      "learning_rate": 9.716014193708153e-06,
      "loss": 0.4015,
      "step": 8214
    },
    {
      "epoch": 0.37699050066541234,
      "grad_norm": 0.47693172097206116,
      "learning_rate": 9.715932732347392e-06,
      "loss": 0.4315,
      "step": 8215
    },
    {
      "epoch": 0.3770363911706668,
      "grad_norm": 0.47348132729530334,
      "learning_rate": 9.715851259646302e-06,
      "loss": 0.4407,
      "step": 8216
    },
    {
      "epoch": 0.3770822816759212,
      "grad_norm": 0.48790687322616577,
      "learning_rate": 9.715769775605081e-06,
      "loss": 0.4909,
      "step": 8217
    },
    {
      "epoch": 0.3771281721811757,
      "grad_norm": 0.418795645236969,
      "learning_rate": 9.71568828022392e-06,
      "loss": 0.338,
      "step": 8218
    },
    {
      "epoch": 0.37717406268643017,
      "grad_norm": 0.4578612446784973,
      "learning_rate": 9.715606773503017e-06,
      "loss": 0.3925,
      "step": 8219
    },
    {
      "epoch": 0.37721995319168466,
      "grad_norm": 0.45661234855651855,
      "learning_rate": 9.715525255442569e-06,
      "loss": 0.3501,
      "step": 8220
    },
    {
      "epoch": 0.3772658436969391,
      "grad_norm": 0.4631122350692749,
      "learning_rate": 9.715443726042767e-06,
      "loss": 0.4087,
      "step": 8221
    },
    {
      "epoch": 0.37731173420219355,
      "grad_norm": 0.44347864389419556,
      "learning_rate": 9.715362185303817e-06,
      "loss": 0.3735,
      "step": 8222
    },
    {
      "epoch": 0.37735762470744805,
      "grad_norm": 0.4790317118167877,
      "learning_rate": 9.715280633225906e-06,
      "loss": 0.4146,
      "step": 8223
    },
    {
      "epoch": 0.3774035152127025,
      "grad_norm": 0.4675402045249939,
      "learning_rate": 9.715199069809234e-06,
      "loss": 0.4484,
      "step": 8224
    },
    {
      "epoch": 0.37744940571795693,
      "grad_norm": 0.4524977505207062,
      "learning_rate": 9.715117495053996e-06,
      "loss": 0.3488,
      "step": 8225
    },
    {
      "epoch": 0.37749529622321143,
      "grad_norm": 0.43830931186676025,
      "learning_rate": 9.715035908960387e-06,
      "loss": 0.3731,
      "step": 8226
    },
    {
      "epoch": 0.37754118672846587,
      "grad_norm": 0.46418023109436035,
      "learning_rate": 9.714954311528608e-06,
      "loss": 0.4149,
      "step": 8227
    },
    {
      "epoch": 0.37758707723372037,
      "grad_norm": 0.5092841982841492,
      "learning_rate": 9.714872702758848e-06,
      "loss": 0.4191,
      "step": 8228
    },
    {
      "epoch": 0.3776329677389748,
      "grad_norm": 0.4970175325870514,
      "learning_rate": 9.714791082651311e-06,
      "loss": 0.4023,
      "step": 8229
    },
    {
      "epoch": 0.37767885824422925,
      "grad_norm": 0.4844402074813843,
      "learning_rate": 9.714709451206187e-06,
      "loss": 0.4583,
      "step": 8230
    },
    {
      "epoch": 0.37772474874948375,
      "grad_norm": 0.4584631025791168,
      "learning_rate": 9.714627808423675e-06,
      "loss": 0.3986,
      "step": 8231
    },
    {
      "epoch": 0.3777706392547382,
      "grad_norm": 0.4738438129425049,
      "learning_rate": 9.714546154303972e-06,
      "loss": 0.3893,
      "step": 8232
    },
    {
      "epoch": 0.37781652975999264,
      "grad_norm": 0.4953833818435669,
      "learning_rate": 9.714464488847274e-06,
      "loss": 0.4488,
      "step": 8233
    },
    {
      "epoch": 0.37786242026524713,
      "grad_norm": 0.4003993570804596,
      "learning_rate": 9.714382812053777e-06,
      "loss": 0.281,
      "step": 8234
    },
    {
      "epoch": 0.3779083107705016,
      "grad_norm": 0.5398240685462952,
      "learning_rate": 9.714301123923675e-06,
      "loss": 0.4551,
      "step": 8235
    },
    {
      "epoch": 0.377954201275756,
      "grad_norm": 0.5457197427749634,
      "learning_rate": 9.71421942445717e-06,
      "loss": 0.4189,
      "step": 8236
    },
    {
      "epoch": 0.3780000917810105,
      "grad_norm": 0.4626288414001465,
      "learning_rate": 9.714137713654452e-06,
      "loss": 0.3555,
      "step": 8237
    },
    {
      "epoch": 0.37804598228626496,
      "grad_norm": 0.41474246978759766,
      "learning_rate": 9.714055991515722e-06,
      "loss": 0.3186,
      "step": 8238
    },
    {
      "epoch": 0.37809187279151946,
      "grad_norm": 0.456434428691864,
      "learning_rate": 9.713974258041175e-06,
      "loss": 0.4133,
      "step": 8239
    },
    {
      "epoch": 0.3781377632967739,
      "grad_norm": 0.40926459431648254,
      "learning_rate": 9.713892513231007e-06,
      "loss": 0.2841,
      "step": 8240
    },
    {
      "epoch": 0.37818365380202834,
      "grad_norm": 0.4456164538860321,
      "learning_rate": 9.713810757085415e-06,
      "loss": 0.3693,
      "step": 8241
    },
    {
      "epoch": 0.37822954430728284,
      "grad_norm": 0.43123140931129456,
      "learning_rate": 9.713728989604598e-06,
      "loss": 0.3622,
      "step": 8242
    },
    {
      "epoch": 0.3782754348125373,
      "grad_norm": 0.4761943221092224,
      "learning_rate": 9.71364721078875e-06,
      "loss": 0.4717,
      "step": 8243
    },
    {
      "epoch": 0.3783213253177917,
      "grad_norm": 0.4528585374355316,
      "learning_rate": 9.713565420638067e-06,
      "loss": 0.4023,
      "step": 8244
    },
    {
      "epoch": 0.3783672158230462,
      "grad_norm": 0.49487781524658203,
      "learning_rate": 9.713483619152746e-06,
      "loss": 0.4649,
      "step": 8245
    },
    {
      "epoch": 0.37841310632830066,
      "grad_norm": 0.4666244685649872,
      "learning_rate": 9.713401806332986e-06,
      "loss": 0.365,
      "step": 8246
    },
    {
      "epoch": 0.37845899683355516,
      "grad_norm": 0.46442312002182007,
      "learning_rate": 9.713319982178981e-06,
      "loss": 0.3467,
      "step": 8247
    },
    {
      "epoch": 0.3785048873388096,
      "grad_norm": 0.4666978120803833,
      "learning_rate": 9.71323814669093e-06,
      "loss": 0.439,
      "step": 8248
    },
    {
      "epoch": 0.37855077784406405,
      "grad_norm": 0.464617520570755,
      "learning_rate": 9.713156299869028e-06,
      "loss": 0.4004,
      "step": 8249
    },
    {
      "epoch": 0.37859666834931854,
      "grad_norm": 0.5115286111831665,
      "learning_rate": 9.713074441713475e-06,
      "loss": 0.4607,
      "step": 8250
    },
    {
      "epoch": 0.378642558854573,
      "grad_norm": 0.4645697772502899,
      "learning_rate": 9.712992572224463e-06,
      "loss": 0.4062,
      "step": 8251
    },
    {
      "epoch": 0.37868844935982743,
      "grad_norm": 0.52003014087677,
      "learning_rate": 9.712910691402194e-06,
      "loss": 0.498,
      "step": 8252
    },
    {
      "epoch": 0.3787343398650819,
      "grad_norm": 0.440660297870636,
      "learning_rate": 9.712828799246858e-06,
      "loss": 0.3562,
      "step": 8253
    },
    {
      "epoch": 0.37878023037033637,
      "grad_norm": 0.472378134727478,
      "learning_rate": 9.71274689575866e-06,
      "loss": 0.4419,
      "step": 8254
    },
    {
      "epoch": 0.37882612087559087,
      "grad_norm": 0.45117640495300293,
      "learning_rate": 9.712664980937791e-06,
      "loss": 0.4224,
      "step": 8255
    },
    {
      "epoch": 0.3788720113808453,
      "grad_norm": 0.5475156903266907,
      "learning_rate": 9.71258305478445e-06,
      "loss": 0.5655,
      "step": 8256
    },
    {
      "epoch": 0.37891790188609975,
      "grad_norm": 0.5003858208656311,
      "learning_rate": 9.712501117298835e-06,
      "loss": 0.4582,
      "step": 8257
    },
    {
      "epoch": 0.37896379239135425,
      "grad_norm": 0.4168681800365448,
      "learning_rate": 9.712419168481142e-06,
      "loss": 0.3305,
      "step": 8258
    },
    {
      "epoch": 0.3790096828966087,
      "grad_norm": 0.5392186045646667,
      "learning_rate": 9.712337208331568e-06,
      "loss": 0.4708,
      "step": 8259
    },
    {
      "epoch": 0.37905557340186313,
      "grad_norm": 0.49951058626174927,
      "learning_rate": 9.71225523685031e-06,
      "loss": 0.4676,
      "step": 8260
    },
    {
      "epoch": 0.37910146390711763,
      "grad_norm": 0.4714858829975128,
      "learning_rate": 9.712173254037567e-06,
      "loss": 0.4164,
      "step": 8261
    },
    {
      "epoch": 0.3791473544123721,
      "grad_norm": 0.5070141553878784,
      "learning_rate": 9.712091259893532e-06,
      "loss": 0.4381,
      "step": 8262
    },
    {
      "epoch": 0.37919324491762657,
      "grad_norm": 0.4539435803890228,
      "learning_rate": 9.712009254418407e-06,
      "loss": 0.3811,
      "step": 8263
    },
    {
      "epoch": 0.379239135422881,
      "grad_norm": 0.4992663264274597,
      "learning_rate": 9.711927237612386e-06,
      "loss": 0.4175,
      "step": 8264
    },
    {
      "epoch": 0.37928502592813546,
      "grad_norm": 0.46159428358078003,
      "learning_rate": 9.711845209475665e-06,
      "loss": 0.3727,
      "step": 8265
    },
    {
      "epoch": 0.37933091643338995,
      "grad_norm": 0.44057777523994446,
      "learning_rate": 9.711763170008447e-06,
      "loss": 0.3485,
      "step": 8266
    },
    {
      "epoch": 0.3793768069386444,
      "grad_norm": 0.46408411860466003,
      "learning_rate": 9.711681119210923e-06,
      "loss": 0.36,
      "step": 8267
    },
    {
      "epoch": 0.37942269744389884,
      "grad_norm": 0.471848726272583,
      "learning_rate": 9.711599057083294e-06,
      "loss": 0.3848,
      "step": 8268
    },
    {
      "epoch": 0.37946858794915334,
      "grad_norm": 0.4797298312187195,
      "learning_rate": 9.711516983625757e-06,
      "loss": 0.3684,
      "step": 8269
    },
    {
      "epoch": 0.3795144784544078,
      "grad_norm": 0.8952170610427856,
      "learning_rate": 9.711434898838508e-06,
      "loss": 0.4252,
      "step": 8270
    },
    {
      "epoch": 0.3795603689596622,
      "grad_norm": 0.4285070300102234,
      "learning_rate": 9.711352802721744e-06,
      "loss": 0.3429,
      "step": 8271
    },
    {
      "epoch": 0.3796062594649167,
      "grad_norm": 0.4940952658653259,
      "learning_rate": 9.711270695275666e-06,
      "loss": 0.4656,
      "step": 8272
    },
    {
      "epoch": 0.37965214997017116,
      "grad_norm": 0.5163105130195618,
      "learning_rate": 9.711188576500465e-06,
      "loss": 0.4868,
      "step": 8273
    },
    {
      "epoch": 0.37969804047542566,
      "grad_norm": 0.4435252547264099,
      "learning_rate": 9.711106446396345e-06,
      "loss": 0.3222,
      "step": 8274
    },
    {
      "epoch": 0.3797439309806801,
      "grad_norm": 0.4323263466358185,
      "learning_rate": 9.7110243049635e-06,
      "loss": 0.3481,
      "step": 8275
    },
    {
      "epoch": 0.37978982148593454,
      "grad_norm": 0.45938247442245483,
      "learning_rate": 9.71094215220213e-06,
      "loss": 0.3663,
      "step": 8276
    },
    {
      "epoch": 0.37983571199118904,
      "grad_norm": 0.5354594588279724,
      "learning_rate": 9.710859988112429e-06,
      "loss": 0.4911,
      "step": 8277
    },
    {
      "epoch": 0.3798816024964435,
      "grad_norm": 0.479200541973114,
      "learning_rate": 9.710777812694598e-06,
      "loss": 0.4471,
      "step": 8278
    },
    {
      "epoch": 0.3799274930016979,
      "grad_norm": 0.44190752506256104,
      "learning_rate": 9.710695625948832e-06,
      "loss": 0.3578,
      "step": 8279
    },
    {
      "epoch": 0.3799733835069524,
      "grad_norm": 0.45690789818763733,
      "learning_rate": 9.710613427875331e-06,
      "loss": 0.3934,
      "step": 8280
    },
    {
      "epoch": 0.38001927401220686,
      "grad_norm": 0.47667229175567627,
      "learning_rate": 9.710531218474291e-06,
      "loss": 0.4304,
      "step": 8281
    },
    {
      "epoch": 0.38006516451746136,
      "grad_norm": 0.4858250319957733,
      "learning_rate": 9.71044899774591e-06,
      "loss": 0.4131,
      "step": 8282
    },
    {
      "epoch": 0.3801110550227158,
      "grad_norm": 0.4979327321052551,
      "learning_rate": 9.710366765690384e-06,
      "loss": 0.4392,
      "step": 8283
    },
    {
      "epoch": 0.38015694552797025,
      "grad_norm": 0.49216383695602417,
      "learning_rate": 9.710284522307914e-06,
      "loss": 0.4238,
      "step": 8284
    },
    {
      "epoch": 0.38020283603322474,
      "grad_norm": 0.41371646523475647,
      "learning_rate": 9.710202267598697e-06,
      "loss": 0.3432,
      "step": 8285
    },
    {
      "epoch": 0.3802487265384792,
      "grad_norm": 0.4837815761566162,
      "learning_rate": 9.71012000156293e-06,
      "loss": 0.4219,
      "step": 8286
    },
    {
      "epoch": 0.38029461704373363,
      "grad_norm": 0.45832619071006775,
      "learning_rate": 9.71003772420081e-06,
      "loss": 0.352,
      "step": 8287
    },
    {
      "epoch": 0.3803405075489881,
      "grad_norm": 0.4829460382461548,
      "learning_rate": 9.709955435512538e-06,
      "loss": 0.4202,
      "step": 8288
    },
    {
      "epoch": 0.38038639805424257,
      "grad_norm": 0.4729141891002655,
      "learning_rate": 9.709873135498307e-06,
      "loss": 0.4027,
      "step": 8289
    },
    {
      "epoch": 0.38043228855949707,
      "grad_norm": 0.5346100926399231,
      "learning_rate": 9.709790824158318e-06,
      "loss": 0.4894,
      "step": 8290
    },
    {
      "epoch": 0.3804781790647515,
      "grad_norm": 0.5437172055244446,
      "learning_rate": 9.70970850149277e-06,
      "loss": 0.5451,
      "step": 8291
    },
    {
      "epoch": 0.38052406957000595,
      "grad_norm": 0.45044800639152527,
      "learning_rate": 9.709626167501858e-06,
      "loss": 0.3225,
      "step": 8292
    },
    {
      "epoch": 0.38056996007526045,
      "grad_norm": 0.4613552689552307,
      "learning_rate": 9.709543822185781e-06,
      "loss": 0.4099,
      "step": 8293
    },
    {
      "epoch": 0.3806158505805149,
      "grad_norm": 0.47125422954559326,
      "learning_rate": 9.70946146554474e-06,
      "loss": 0.2868,
      "step": 8294
    },
    {
      "epoch": 0.38066174108576933,
      "grad_norm": 0.4469430446624756,
      "learning_rate": 9.709379097578929e-06,
      "loss": 0.3564,
      "step": 8295
    },
    {
      "epoch": 0.38070763159102383,
      "grad_norm": 0.5077095627784729,
      "learning_rate": 9.709296718288548e-06,
      "loss": 0.4828,
      "step": 8296
    },
    {
      "epoch": 0.3807535220962783,
      "grad_norm": 0.4666096866130829,
      "learning_rate": 9.709214327673795e-06,
      "loss": 0.4056,
      "step": 8297
    },
    {
      "epoch": 0.3807994126015327,
      "grad_norm": 0.4739914834499359,
      "learning_rate": 9.709131925734866e-06,
      "loss": 0.3924,
      "step": 8298
    },
    {
      "epoch": 0.3808453031067872,
      "grad_norm": 0.46668705344200134,
      "learning_rate": 9.709049512471962e-06,
      "loss": 0.3901,
      "step": 8299
    },
    {
      "epoch": 0.38089119361204166,
      "grad_norm": 0.4913577735424042,
      "learning_rate": 9.70896708788528e-06,
      "loss": 0.4143,
      "step": 8300
    },
    {
      "epoch": 0.38093708411729615,
      "grad_norm": 0.4827563762664795,
      "learning_rate": 9.70888465197502e-06,
      "loss": 0.4219,
      "step": 8301
    },
    {
      "epoch": 0.3809829746225506,
      "grad_norm": 0.44982078671455383,
      "learning_rate": 9.708802204741377e-06,
      "loss": 0.3535,
      "step": 8302
    },
    {
      "epoch": 0.38102886512780504,
      "grad_norm": 0.4478099048137665,
      "learning_rate": 9.70871974618455e-06,
      "loss": 0.4027,
      "step": 8303
    },
    {
      "epoch": 0.38107475563305954,
      "grad_norm": 0.48432034254074097,
      "learning_rate": 9.70863727630474e-06,
      "loss": 0.4367,
      "step": 8304
    },
    {
      "epoch": 0.381120646138314,
      "grad_norm": 0.4797355532646179,
      "learning_rate": 9.708554795102141e-06,
      "loss": 0.4354,
      "step": 8305
    },
    {
      "epoch": 0.3811665366435684,
      "grad_norm": 0.4524461030960083,
      "learning_rate": 9.708472302576957e-06,
      "loss": 0.4401,
      "step": 8306
    },
    {
      "epoch": 0.3812124271488229,
      "grad_norm": 0.45739978551864624,
      "learning_rate": 9.70838979872938e-06,
      "loss": 0.3756,
      "step": 8307
    },
    {
      "epoch": 0.38125831765407736,
      "grad_norm": 0.48678842186927795,
      "learning_rate": 9.708307283559614e-06,
      "loss": 0.483,
      "step": 8308
    },
    {
      "epoch": 0.38130420815933186,
      "grad_norm": 0.4844021499156952,
      "learning_rate": 9.708224757067854e-06,
      "loss": 0.4392,
      "step": 8309
    },
    {
      "epoch": 0.3813500986645863,
      "grad_norm": 0.5173817276954651,
      "learning_rate": 9.7081422192543e-06,
      "loss": 0.4988,
      "step": 8310
    },
    {
      "epoch": 0.38139598916984074,
      "grad_norm": 0.4789606034755707,
      "learning_rate": 9.70805967011915e-06,
      "loss": 0.4551,
      "step": 8311
    },
    {
      "epoch": 0.38144187967509524,
      "grad_norm": 0.42956599593162537,
      "learning_rate": 9.7079771096626e-06,
      "loss": 0.3246,
      "step": 8312
    },
    {
      "epoch": 0.3814877701803497,
      "grad_norm": 0.49112486839294434,
      "learning_rate": 9.707894537884854e-06,
      "loss": 0.4657,
      "step": 8313
    },
    {
      "epoch": 0.3815336606856041,
      "grad_norm": 0.45202475786209106,
      "learning_rate": 9.707811954786106e-06,
      "loss": 0.4416,
      "step": 8314
    },
    {
      "epoch": 0.3815795511908586,
      "grad_norm": 0.4294533133506775,
      "learning_rate": 9.707729360366555e-06,
      "loss": 0.3074,
      "step": 8315
    },
    {
      "epoch": 0.38162544169611307,
      "grad_norm": 0.4875158965587616,
      "learning_rate": 9.707646754626404e-06,
      "loss": 0.4598,
      "step": 8316
    },
    {
      "epoch": 0.38167133220136756,
      "grad_norm": 0.4753047227859497,
      "learning_rate": 9.707564137565845e-06,
      "loss": 0.4554,
      "step": 8317
    },
    {
      "epoch": 0.381717222706622,
      "grad_norm": 0.44825419783592224,
      "learning_rate": 9.707481509185082e-06,
      "loss": 0.4122,
      "step": 8318
    },
    {
      "epoch": 0.38176311321187645,
      "grad_norm": 0.43962204456329346,
      "learning_rate": 9.70739886948431e-06,
      "loss": 0.3636,
      "step": 8319
    },
    {
      "epoch": 0.38180900371713095,
      "grad_norm": 0.4450395405292511,
      "learning_rate": 9.70731621846373e-06,
      "loss": 0.3952,
      "step": 8320
    },
    {
      "epoch": 0.3818548942223854,
      "grad_norm": 0.4353902041912079,
      "learning_rate": 9.707233556123542e-06,
      "loss": 0.3541,
      "step": 8321
    },
    {
      "epoch": 0.38190078472763983,
      "grad_norm": 0.43953320384025574,
      "learning_rate": 9.70715088246394e-06,
      "loss": 0.3376,
      "step": 8322
    },
    {
      "epoch": 0.38194667523289433,
      "grad_norm": 0.4470181167125702,
      "learning_rate": 9.707068197485128e-06,
      "loss": 0.3843,
      "step": 8323
    },
    {
      "epoch": 0.38199256573814877,
      "grad_norm": 0.47696831822395325,
      "learning_rate": 9.706985501187302e-06,
      "loss": 0.4273,
      "step": 8324
    },
    {
      "epoch": 0.3820384562434032,
      "grad_norm": 0.49272671341896057,
      "learning_rate": 9.706902793570661e-06,
      "loss": 0.5135,
      "step": 8325
    },
    {
      "epoch": 0.3820843467486577,
      "grad_norm": 0.4752046465873718,
      "learning_rate": 9.706820074635405e-06,
      "loss": 0.4341,
      "step": 8326
    },
    {
      "epoch": 0.38213023725391215,
      "grad_norm": 0.4485021233558655,
      "learning_rate": 9.706737344381732e-06,
      "loss": 0.3836,
      "step": 8327
    },
    {
      "epoch": 0.38217612775916665,
      "grad_norm": 0.44785410165786743,
      "learning_rate": 9.706654602809839e-06,
      "loss": 0.3488,
      "step": 8328
    },
    {
      "epoch": 0.3822220182644211,
      "grad_norm": 0.464387983083725,
      "learning_rate": 9.70657184991993e-06,
      "loss": 0.396,
      "step": 8329
    },
    {
      "epoch": 0.38226790876967554,
      "grad_norm": 0.48688462376594543,
      "learning_rate": 9.7064890857122e-06,
      "loss": 0.4341,
      "step": 8330
    },
    {
      "epoch": 0.38231379927493003,
      "grad_norm": 0.42955482006073,
      "learning_rate": 9.70640631018685e-06,
      "loss": 0.3605,
      "step": 8331
    },
    {
      "epoch": 0.3823596897801845,
      "grad_norm": 0.434162974357605,
      "learning_rate": 9.706323523344077e-06,
      "loss": 0.3592,
      "step": 8332
    },
    {
      "epoch": 0.3824055802854389,
      "grad_norm": 0.49073097109794617,
      "learning_rate": 9.706240725184083e-06,
      "loss": 0.4413,
      "step": 8333
    },
    {
      "epoch": 0.3824514707906934,
      "grad_norm": 0.4110638201236725,
      "learning_rate": 9.706157915707064e-06,
      "loss": 0.2926,
      "step": 8334
    },
    {
      "epoch": 0.38249736129594786,
      "grad_norm": 0.4811594486236572,
      "learning_rate": 9.706075094913222e-06,
      "loss": 0.4186,
      "step": 8335
    },
    {
      "epoch": 0.38254325180120236,
      "grad_norm": 0.4778303802013397,
      "learning_rate": 9.705992262802753e-06,
      "loss": 0.4189,
      "step": 8336
    },
    {
      "epoch": 0.3825891423064568,
      "grad_norm": 0.4732328951358795,
      "learning_rate": 9.705909419375859e-06,
      "loss": 0.4317,
      "step": 8337
    },
    {
      "epoch": 0.38263503281171124,
      "grad_norm": 0.45637059211730957,
      "learning_rate": 9.705826564632736e-06,
      "loss": 0.3987,
      "step": 8338
    },
    {
      "epoch": 0.38268092331696574,
      "grad_norm": 0.44102656841278076,
      "learning_rate": 9.705743698573586e-06,
      "loss": 0.3642,
      "step": 8339
    },
    {
      "epoch": 0.3827268138222202,
      "grad_norm": 0.4993564188480377,
      "learning_rate": 9.70566082119861e-06,
      "loss": 0.4381,
      "step": 8340
    },
    {
      "epoch": 0.3827727043274746,
      "grad_norm": 0.4486505687236786,
      "learning_rate": 9.705577932508004e-06,
      "loss": 0.3664,
      "step": 8341
    },
    {
      "epoch": 0.3828185948327291,
      "grad_norm": 0.4665676951408386,
      "learning_rate": 9.705495032501966e-06,
      "loss": 0.3959,
      "step": 8342
    },
    {
      "epoch": 0.38286448533798356,
      "grad_norm": 0.5160298347473145,
      "learning_rate": 9.705412121180699e-06,
      "loss": 0.5282,
      "step": 8343
    },
    {
      "epoch": 0.38291037584323806,
      "grad_norm": 0.4632991850376129,
      "learning_rate": 9.7053291985444e-06,
      "loss": 0.4255,
      "step": 8344
    },
    {
      "epoch": 0.3829562663484925,
      "grad_norm": 0.46649202704429626,
      "learning_rate": 9.705246264593271e-06,
      "loss": 0.4196,
      "step": 8345
    },
    {
      "epoch": 0.38300215685374694,
      "grad_norm": 0.4724917411804199,
      "learning_rate": 9.705163319327508e-06,
      "loss": 0.4375,
      "step": 8346
    },
    {
      "epoch": 0.38304804735900144,
      "grad_norm": 0.4525358974933624,
      "learning_rate": 9.705080362747314e-06,
      "loss": 0.3612,
      "step": 8347
    },
    {
      "epoch": 0.3830939378642559,
      "grad_norm": 0.443567156791687,
      "learning_rate": 9.704997394852887e-06,
      "loss": 0.3399,
      "step": 8348
    },
    {
      "epoch": 0.3831398283695103,
      "grad_norm": 0.4305554926395416,
      "learning_rate": 9.704914415644425e-06,
      "loss": 0.3343,
      "step": 8349
    },
    {
      "epoch": 0.3831857188747648,
      "grad_norm": 0.44661766290664673,
      "learning_rate": 9.704831425122128e-06,
      "loss": 0.3465,
      "step": 8350
    },
    {
      "epoch": 0.38323160938001927,
      "grad_norm": 0.4470031261444092,
      "learning_rate": 9.704748423286196e-06,
      "loss": 0.3771,
      "step": 8351
    },
    {
      "epoch": 0.3832774998852737,
      "grad_norm": 0.4797007441520691,
      "learning_rate": 9.70466541013683e-06,
      "loss": 0.4049,
      "step": 8352
    },
    {
      "epoch": 0.3833233903905282,
      "grad_norm": 0.49246424436569214,
      "learning_rate": 9.70458238567423e-06,
      "loss": 0.4165,
      "step": 8353
    },
    {
      "epoch": 0.38336928089578265,
      "grad_norm": 0.448569118976593,
      "learning_rate": 9.704499349898593e-06,
      "loss": 0.3709,
      "step": 8354
    },
    {
      "epoch": 0.38341517140103715,
      "grad_norm": 0.43227648735046387,
      "learning_rate": 9.704416302810118e-06,
      "loss": 0.3323,
      "step": 8355
    },
    {
      "epoch": 0.3834610619062916,
      "grad_norm": 0.5418111681938171,
      "learning_rate": 9.70433324440901e-06,
      "loss": 0.4384,
      "step": 8356
    },
    {
      "epoch": 0.38350695241154603,
      "grad_norm": 0.5218394994735718,
      "learning_rate": 9.704250174695463e-06,
      "loss": 0.4795,
      "step": 8357
    },
    {
      "epoch": 0.38355284291680053,
      "grad_norm": 0.4497133493423462,
      "learning_rate": 9.704167093669678e-06,
      "loss": 0.3759,
      "step": 8358
    },
    {
      "epoch": 0.38359873342205497,
      "grad_norm": 0.47627773880958557,
      "learning_rate": 9.704084001331858e-06,
      "loss": 0.395,
      "step": 8359
    },
    {
      "epoch": 0.3836446239273094,
      "grad_norm": 0.49893948435783386,
      "learning_rate": 9.7040008976822e-06,
      "loss": 0.4328,
      "step": 8360
    },
    {
      "epoch": 0.3836905144325639,
      "grad_norm": 0.46872392296791077,
      "learning_rate": 9.703917782720904e-06,
      "loss": 0.4086,
      "step": 8361
    },
    {
      "epoch": 0.38373640493781835,
      "grad_norm": 0.4398308992385864,
      "learning_rate": 9.70383465644817e-06,
      "loss": 0.3391,
      "step": 8362
    },
    {
      "epoch": 0.38378229544307285,
      "grad_norm": 0.43760430812835693,
      "learning_rate": 9.703751518864199e-06,
      "loss": 0.3523,
      "step": 8363
    },
    {
      "epoch": 0.3838281859483273,
      "grad_norm": 0.4730702340602875,
      "learning_rate": 9.703668369969191e-06,
      "loss": 0.3992,
      "step": 8364
    },
    {
      "epoch": 0.38387407645358174,
      "grad_norm": 0.45412975549697876,
      "learning_rate": 9.703585209763344e-06,
      "loss": 0.4617,
      "step": 8365
    },
    {
      "epoch": 0.38391996695883623,
      "grad_norm": 0.4601110816001892,
      "learning_rate": 9.70350203824686e-06,
      "loss": 0.3668,
      "step": 8366
    },
    {
      "epoch": 0.3839658574640907,
      "grad_norm": 0.4649891257286072,
      "learning_rate": 9.703418855419937e-06,
      "loss": 0.4254,
      "step": 8367
    },
    {
      "epoch": 0.3840117479693451,
      "grad_norm": 0.47691312432289124,
      "learning_rate": 9.703335661282776e-06,
      "loss": 0.4247,
      "step": 8368
    },
    {
      "epoch": 0.3840576384745996,
      "grad_norm": 0.4574812948703766,
      "learning_rate": 9.703252455835576e-06,
      "loss": 0.333,
      "step": 8369
    },
    {
      "epoch": 0.38410352897985406,
      "grad_norm": 0.46248960494995117,
      "learning_rate": 9.70316923907854e-06,
      "loss": 0.415,
      "step": 8370
    },
    {
      "epoch": 0.38414941948510856,
      "grad_norm": 0.5206280946731567,
      "learning_rate": 9.703086011011867e-06,
      "loss": 0.4653,
      "step": 8371
    },
    {
      "epoch": 0.384195309990363,
      "grad_norm": 0.47378215193748474,
      "learning_rate": 9.703002771635755e-06,
      "loss": 0.4273,
      "step": 8372
    },
    {
      "epoch": 0.38424120049561744,
      "grad_norm": 0.5185543298721313,
      "learning_rate": 9.702919520950406e-06,
      "loss": 0.481,
      "step": 8373
    },
    {
      "epoch": 0.38428709100087194,
      "grad_norm": 0.48190850019454956,
      "learning_rate": 9.70283625895602e-06,
      "loss": 0.4519,
      "step": 8374
    },
    {
      "epoch": 0.3843329815061264,
      "grad_norm": 0.4631403982639313,
      "learning_rate": 9.702752985652798e-06,
      "loss": 0.4326,
      "step": 8375
    },
    {
      "epoch": 0.3843788720113808,
      "grad_norm": 0.4552483558654785,
      "learning_rate": 9.702669701040938e-06,
      "loss": 0.3611,
      "step": 8376
    },
    {
      "epoch": 0.3844247625166353,
      "grad_norm": 0.4930436909198761,
      "learning_rate": 9.702586405120642e-06,
      "loss": 0.385,
      "step": 8377
    },
    {
      "epoch": 0.38447065302188976,
      "grad_norm": 0.429897278547287,
      "learning_rate": 9.702503097892112e-06,
      "loss": 0.3455,
      "step": 8378
    },
    {
      "epoch": 0.38451654352714426,
      "grad_norm": 0.4308724105358124,
      "learning_rate": 9.702419779355542e-06,
      "loss": 0.3798,
      "step": 8379
    },
    {
      "epoch": 0.3845624340323987,
      "grad_norm": 0.49082064628601074,
      "learning_rate": 9.702336449511138e-06,
      "loss": 0.4063,
      "step": 8380
    },
    {
      "epoch": 0.38460832453765315,
      "grad_norm": 0.47112369537353516,
      "learning_rate": 9.7022531083591e-06,
      "loss": 0.4269,
      "step": 8381
    },
    {
      "epoch": 0.38465421504290764,
      "grad_norm": 0.45445263385772705,
      "learning_rate": 9.702169755899627e-06,
      "loss": 0.3778,
      "step": 8382
    },
    {
      "epoch": 0.3847001055481621,
      "grad_norm": 0.45491957664489746,
      "learning_rate": 9.70208639213292e-06,
      "loss": 0.4068,
      "step": 8383
    },
    {
      "epoch": 0.38474599605341653,
      "grad_norm": 0.47046777606010437,
      "learning_rate": 9.702003017059179e-06,
      "loss": 0.3934,
      "step": 8384
    },
    {
      "epoch": 0.384791886558671,
      "grad_norm": 0.47704777121543884,
      "learning_rate": 9.701919630678604e-06,
      "loss": 0.4147,
      "step": 8385
    },
    {
      "epoch": 0.38483777706392547,
      "grad_norm": 0.4478757679462433,
      "learning_rate": 9.701836232991397e-06,
      "loss": 0.3877,
      "step": 8386
    },
    {
      "epoch": 0.3848836675691799,
      "grad_norm": 0.4703952670097351,
      "learning_rate": 9.701752823997759e-06,
      "loss": 0.4235,
      "step": 8387
    },
    {
      "epoch": 0.3849295580744344,
      "grad_norm": 0.46203720569610596,
      "learning_rate": 9.701669403697887e-06,
      "loss": 0.441,
      "step": 8388
    },
    {
      "epoch": 0.38497544857968885,
      "grad_norm": 0.4345662295818329,
      "learning_rate": 9.701585972091984e-06,
      "loss": 0.4014,
      "step": 8389
    },
    {
      "epoch": 0.38502133908494335,
      "grad_norm": 0.4861225485801697,
      "learning_rate": 9.701502529180253e-06,
      "loss": 0.4328,
      "step": 8390
    },
    {
      "epoch": 0.3850672295901978,
      "grad_norm": 0.5345048308372498,
      "learning_rate": 9.70141907496289e-06,
      "loss": 0.5577,
      "step": 8391
    },
    {
      "epoch": 0.38511312009545223,
      "grad_norm": 0.46646860241889954,
      "learning_rate": 9.7013356094401e-06,
      "loss": 0.3742,
      "step": 8392
    },
    {
      "epoch": 0.38515901060070673,
      "grad_norm": 0.46382835507392883,
      "learning_rate": 9.70125213261208e-06,
      "loss": 0.4333,
      "step": 8393
    },
    {
      "epoch": 0.3852049011059612,
      "grad_norm": 0.4557649791240692,
      "learning_rate": 9.701168644479033e-06,
      "loss": 0.3744,
      "step": 8394
    },
    {
      "epoch": 0.3852507916112156,
      "grad_norm": 0.44920802116394043,
      "learning_rate": 9.70108514504116e-06,
      "loss": 0.3411,
      "step": 8395
    },
    {
      "epoch": 0.3852966821164701,
      "grad_norm": 0.49356749653816223,
      "learning_rate": 9.70100163429866e-06,
      "loss": 0.4903,
      "step": 8396
    },
    {
      "epoch": 0.38534257262172456,
      "grad_norm": 0.44686180353164673,
      "learning_rate": 9.700918112251733e-06,
      "loss": 0.4096,
      "step": 8397
    },
    {
      "epoch": 0.38538846312697905,
      "grad_norm": 0.48263344168663025,
      "learning_rate": 9.700834578900583e-06,
      "loss": 0.4149,
      "step": 8398
    },
    {
      "epoch": 0.3854343536322335,
      "grad_norm": 0.49760544300079346,
      "learning_rate": 9.70075103424541e-06,
      "loss": 0.5004,
      "step": 8399
    },
    {
      "epoch": 0.38548024413748794,
      "grad_norm": 0.4403175413608551,
      "learning_rate": 9.700667478286415e-06,
      "loss": 0.3843,
      "step": 8400
    },
    {
      "epoch": 0.38552613464274244,
      "grad_norm": 0.4831075370311737,
      "learning_rate": 9.700583911023798e-06,
      "loss": 0.3931,
      "step": 8401
    },
    {
      "epoch": 0.3855720251479969,
      "grad_norm": 0.47512587904930115,
      "learning_rate": 9.700500332457758e-06,
      "loss": 0.4417,
      "step": 8402
    },
    {
      "epoch": 0.3856179156532513,
      "grad_norm": 0.47712501883506775,
      "learning_rate": 9.700416742588501e-06,
      "loss": 0.4216,
      "step": 8403
    },
    {
      "epoch": 0.3856638061585058,
      "grad_norm": 0.48236021399497986,
      "learning_rate": 9.700333141416224e-06,
      "loss": 0.4912,
      "step": 8404
    },
    {
      "epoch": 0.38570969666376026,
      "grad_norm": 0.4740074574947357,
      "learning_rate": 9.70024952894113e-06,
      "loss": 0.4072,
      "step": 8405
    },
    {
      "epoch": 0.38575558716901476,
      "grad_norm": 0.4570198357105255,
      "learning_rate": 9.700165905163419e-06,
      "loss": 0.3705,
      "step": 8406
    },
    {
      "epoch": 0.3858014776742692,
      "grad_norm": 0.45355919003486633,
      "learning_rate": 9.70008227008329e-06,
      "loss": 0.3839,
      "step": 8407
    },
    {
      "epoch": 0.38584736817952364,
      "grad_norm": 0.45618879795074463,
      "learning_rate": 9.69999862370095e-06,
      "loss": 0.4008,
      "step": 8408
    },
    {
      "epoch": 0.38589325868477814,
      "grad_norm": 0.48290762305259705,
      "learning_rate": 9.699914966016595e-06,
      "loss": 0.3982,
      "step": 8409
    },
    {
      "epoch": 0.3859391491900326,
      "grad_norm": 0.47827664017677307,
      "learning_rate": 9.699831297030429e-06,
      "loss": 0.4636,
      "step": 8410
    },
    {
      "epoch": 0.385985039695287,
      "grad_norm": 0.46036893129348755,
      "learning_rate": 9.699747616742651e-06,
      "loss": 0.4115,
      "step": 8411
    },
    {
      "epoch": 0.3860309302005415,
      "grad_norm": 0.4618206024169922,
      "learning_rate": 9.699663925153463e-06,
      "loss": 0.418,
      "step": 8412
    },
    {
      "epoch": 0.38607682070579596,
      "grad_norm": 0.4307831823825836,
      "learning_rate": 9.699580222263068e-06,
      "loss": 0.3225,
      "step": 8413
    },
    {
      "epoch": 0.3861227112110504,
      "grad_norm": 0.44958049058914185,
      "learning_rate": 9.699496508071665e-06,
      "loss": 0.351,
      "step": 8414
    },
    {
      "epoch": 0.3861686017163049,
      "grad_norm": 0.4616737961769104,
      "learning_rate": 9.699412782579454e-06,
      "loss": 0.4309,
      "step": 8415
    },
    {
      "epoch": 0.38621449222155935,
      "grad_norm": 0.4471718668937683,
      "learning_rate": 9.69932904578664e-06,
      "loss": 0.3834,
      "step": 8416
    },
    {
      "epoch": 0.38626038272681384,
      "grad_norm": 0.4755763113498688,
      "learning_rate": 9.699245297693424e-06,
      "loss": 0.4484,
      "step": 8417
    },
    {
      "epoch": 0.3863062732320683,
      "grad_norm": 0.42185693979263306,
      "learning_rate": 9.699161538300006e-06,
      "loss": 0.3188,
      "step": 8418
    },
    {
      "epoch": 0.38635216373732273,
      "grad_norm": 0.4859219789505005,
      "learning_rate": 9.699077767606587e-06,
      "loss": 0.4504,
      "step": 8419
    },
    {
      "epoch": 0.3863980542425772,
      "grad_norm": 0.4808337688446045,
      "learning_rate": 9.698993985613367e-06,
      "loss": 0.3962,
      "step": 8420
    },
    {
      "epoch": 0.38644394474783167,
      "grad_norm": 0.4514528810977936,
      "learning_rate": 9.698910192320552e-06,
      "loss": 0.3964,
      "step": 8421
    },
    {
      "epoch": 0.3864898352530861,
      "grad_norm": 0.44106897711753845,
      "learning_rate": 9.698826387728341e-06,
      "loss": 0.3675,
      "step": 8422
    },
    {
      "epoch": 0.3865357257583406,
      "grad_norm": 0.4566441476345062,
      "learning_rate": 9.698742571836934e-06,
      "loss": 0.4168,
      "step": 8423
    },
    {
      "epoch": 0.38658161626359505,
      "grad_norm": 0.4769282937049866,
      "learning_rate": 9.698658744646536e-06,
      "loss": 0.4725,
      "step": 8424
    },
    {
      "epoch": 0.38662750676884955,
      "grad_norm": 0.5071470141410828,
      "learning_rate": 9.698574906157346e-06,
      "loss": 0.4504,
      "step": 8425
    },
    {
      "epoch": 0.386673397274104,
      "grad_norm": 0.47008535265922546,
      "learning_rate": 9.698491056369566e-06,
      "loss": 0.3983,
      "step": 8426
    },
    {
      "epoch": 0.38671928777935843,
      "grad_norm": 0.48331886529922485,
      "learning_rate": 9.698407195283396e-06,
      "loss": 0.4095,
      "step": 8427
    },
    {
      "epoch": 0.38676517828461293,
      "grad_norm": 0.46814489364624023,
      "learning_rate": 9.698323322899044e-06,
      "loss": 0.4155,
      "step": 8428
    },
    {
      "epoch": 0.3868110687898674,
      "grad_norm": 0.41404491662979126,
      "learning_rate": 9.698239439216703e-06,
      "loss": 0.3112,
      "step": 8429
    },
    {
      "epoch": 0.3868569592951218,
      "grad_norm": 0.4392063319683075,
      "learning_rate": 9.69815554423658e-06,
      "loss": 0.375,
      "step": 8430
    },
    {
      "epoch": 0.3869028498003763,
      "grad_norm": 0.5238365530967712,
      "learning_rate": 9.698071637958877e-06,
      "loss": 0.4396,
      "step": 8431
    },
    {
      "epoch": 0.38694874030563076,
      "grad_norm": 0.44936665892601013,
      "learning_rate": 9.697987720383795e-06,
      "loss": 0.3931,
      "step": 8432
    },
    {
      "epoch": 0.38699463081088525,
      "grad_norm": 0.4627906382083893,
      "learning_rate": 9.697903791511534e-06,
      "loss": 0.4581,
      "step": 8433
    },
    {
      "epoch": 0.3870405213161397,
      "grad_norm": 0.45458608865737915,
      "learning_rate": 9.697819851342296e-06,
      "loss": 0.3489,
      "step": 8434
    },
    {
      "epoch": 0.38708641182139414,
      "grad_norm": 0.4643838107585907,
      "learning_rate": 9.697735899876286e-06,
      "loss": 0.4042,
      "step": 8435
    },
    {
      "epoch": 0.38713230232664864,
      "grad_norm": 0.45791617035865784,
      "learning_rate": 9.697651937113703e-06,
      "loss": 0.3752,
      "step": 8436
    },
    {
      "epoch": 0.3871781928319031,
      "grad_norm": 0.4958685040473938,
      "learning_rate": 9.697567963054748e-06,
      "loss": 0.4622,
      "step": 8437
    },
    {
      "epoch": 0.3872240833371575,
      "grad_norm": 0.4513900578022003,
      "learning_rate": 9.697483977699626e-06,
      "loss": 0.351,
      "step": 8438
    },
    {
      "epoch": 0.387269973842412,
      "grad_norm": 0.4998222589492798,
      "learning_rate": 9.697399981048539e-06,
      "loss": 0.4314,
      "step": 8439
    },
    {
      "epoch": 0.38731586434766646,
      "grad_norm": 0.5018230676651001,
      "learning_rate": 9.697315973101686e-06,
      "loss": 0.4611,
      "step": 8440
    },
    {
      "epoch": 0.3873617548529209,
      "grad_norm": 0.4948044419288635,
      "learning_rate": 9.697231953859271e-06,
      "loss": 0.5221,
      "step": 8441
    },
    {
      "epoch": 0.3874076453581754,
      "grad_norm": 0.5019229054450989,
      "learning_rate": 9.697147923321496e-06,
      "loss": 0.4145,
      "step": 8442
    },
    {
      "epoch": 0.38745353586342984,
      "grad_norm": 0.4509308338165283,
      "learning_rate": 9.697063881488561e-06,
      "loss": 0.3844,
      "step": 8443
    },
    {
      "epoch": 0.38749942636868434,
      "grad_norm": 0.4653271734714508,
      "learning_rate": 9.696979828360673e-06,
      "loss": 0.4791,
      "step": 8444
    },
    {
      "epoch": 0.3875453168739388,
      "grad_norm": 0.5491061210632324,
      "learning_rate": 9.696895763938027e-06,
      "loss": 0.5448,
      "step": 8445
    },
    {
      "epoch": 0.3875912073791932,
      "grad_norm": 0.49758028984069824,
      "learning_rate": 9.696811688220833e-06,
      "loss": 0.4972,
      "step": 8446
    },
    {
      "epoch": 0.3876370978844477,
      "grad_norm": 0.4888775050640106,
      "learning_rate": 9.696727601209287e-06,
      "loss": 0.492,
      "step": 8447
    },
    {
      "epoch": 0.38768298838970217,
      "grad_norm": 0.4525541663169861,
      "learning_rate": 9.696643502903594e-06,
      "loss": 0.3484,
      "step": 8448
    },
    {
      "epoch": 0.3877288788949566,
      "grad_norm": 0.4701160192489624,
      "learning_rate": 9.696559393303954e-06,
      "loss": 0.4824,
      "step": 8449
    },
    {
      "epoch": 0.3877747694002111,
      "grad_norm": 0.45301681756973267,
      "learning_rate": 9.696475272410574e-06,
      "loss": 0.4303,
      "step": 8450
    },
    {
      "epoch": 0.38782065990546555,
      "grad_norm": 0.49915072321891785,
      "learning_rate": 9.696391140223651e-06,
      "loss": 0.4596,
      "step": 8451
    },
    {
      "epoch": 0.38786655041072005,
      "grad_norm": 0.4563049376010895,
      "learning_rate": 9.69630699674339e-06,
      "loss": 0.3533,
      "step": 8452
    },
    {
      "epoch": 0.3879124409159745,
      "grad_norm": 0.44588980078697205,
      "learning_rate": 9.696222841969993e-06,
      "loss": 0.3923,
      "step": 8453
    },
    {
      "epoch": 0.38795833142122893,
      "grad_norm": 0.47425606846809387,
      "learning_rate": 9.696138675903663e-06,
      "loss": 0.3236,
      "step": 8454
    },
    {
      "epoch": 0.38800422192648343,
      "grad_norm": 0.48490166664123535,
      "learning_rate": 9.6960544985446e-06,
      "loss": 0.4331,
      "step": 8455
    },
    {
      "epoch": 0.38805011243173787,
      "grad_norm": 0.48784756660461426,
      "learning_rate": 9.69597030989301e-06,
      "loss": 0.439,
      "step": 8456
    },
    {
      "epoch": 0.3880960029369923,
      "grad_norm": 0.48858046531677246,
      "learning_rate": 9.695886109949093e-06,
      "loss": 0.4961,
      "step": 8457
    },
    {
      "epoch": 0.3881418934422468,
      "grad_norm": 0.48567572236061096,
      "learning_rate": 9.69580189871305e-06,
      "loss": 0.4434,
      "step": 8458
    },
    {
      "epoch": 0.38818778394750125,
      "grad_norm": 0.4231758713722229,
      "learning_rate": 9.695717676185088e-06,
      "loss": 0.3258,
      "step": 8459
    },
    {
      "epoch": 0.38823367445275575,
      "grad_norm": 0.5134083032608032,
      "learning_rate": 9.695633442365406e-06,
      "loss": 0.4745,
      "step": 8460
    },
    {
      "epoch": 0.3882795649580102,
      "grad_norm": 0.45244917273521423,
      "learning_rate": 9.695549197254208e-06,
      "loss": 0.3711,
      "step": 8461
    },
    {
      "epoch": 0.38832545546326463,
      "grad_norm": 0.48285987973213196,
      "learning_rate": 9.695464940851696e-06,
      "loss": 0.4354,
      "step": 8462
    },
    {
      "epoch": 0.38837134596851913,
      "grad_norm": 0.4743655323982239,
      "learning_rate": 9.695380673158072e-06,
      "loss": 0.4325,
      "step": 8463
    },
    {
      "epoch": 0.3884172364737736,
      "grad_norm": 0.49074769020080566,
      "learning_rate": 9.695296394173539e-06,
      "loss": 0.4708,
      "step": 8464
    },
    {
      "epoch": 0.388463126979028,
      "grad_norm": 0.4766613841056824,
      "learning_rate": 9.6952121038983e-06,
      "loss": 0.4008,
      "step": 8465
    },
    {
      "epoch": 0.3885090174842825,
      "grad_norm": 0.4954339265823364,
      "learning_rate": 9.69512780233256e-06,
      "loss": 0.4165,
      "step": 8466
    },
    {
      "epoch": 0.38855490798953696,
      "grad_norm": 0.44261690974235535,
      "learning_rate": 9.695043489476517e-06,
      "loss": 0.3501,
      "step": 8467
    },
    {
      "epoch": 0.38860079849479146,
      "grad_norm": 0.4264901280403137,
      "learning_rate": 9.694959165330378e-06,
      "loss": 0.3331,
      "step": 8468
    },
    {
      "epoch": 0.3886466890000459,
      "grad_norm": 0.5148507952690125,
      "learning_rate": 9.694874829894343e-06,
      "loss": 0.4608,
      "step": 8469
    },
    {
      "epoch": 0.38869257950530034,
      "grad_norm": 0.48227229714393616,
      "learning_rate": 9.694790483168617e-06,
      "loss": 0.4412,
      "step": 8470
    },
    {
      "epoch": 0.38873847001055484,
      "grad_norm": 0.4578356444835663,
      "learning_rate": 9.694706125153399e-06,
      "loss": 0.3351,
      "step": 8471
    },
    {
      "epoch": 0.3887843605158093,
      "grad_norm": 0.4463386535644531,
      "learning_rate": 9.694621755848896e-06,
      "loss": 0.3749,
      "step": 8472
    },
    {
      "epoch": 0.3888302510210637,
      "grad_norm": 0.48371225595474243,
      "learning_rate": 9.69453737525531e-06,
      "loss": 0.4609,
      "step": 8473
    },
    {
      "epoch": 0.3888761415263182,
      "grad_norm": 0.47581446170806885,
      "learning_rate": 9.694452983372841e-06,
      "loss": 0.37,
      "step": 8474
    },
    {
      "epoch": 0.38892203203157266,
      "grad_norm": 0.49874043464660645,
      "learning_rate": 9.694368580201698e-06,
      "loss": 0.4723,
      "step": 8475
    },
    {
      "epoch": 0.3889679225368271,
      "grad_norm": 0.4929570257663727,
      "learning_rate": 9.694284165742077e-06,
      "loss": 0.4854,
      "step": 8476
    },
    {
      "epoch": 0.3890138130420816,
      "grad_norm": 0.4662057161331177,
      "learning_rate": 9.694199739994185e-06,
      "loss": 0.374,
      "step": 8477
    },
    {
      "epoch": 0.38905970354733604,
      "grad_norm": 0.4705458879470825,
      "learning_rate": 9.694115302958225e-06,
      "loss": 0.4088,
      "step": 8478
    },
    {
      "epoch": 0.38910559405259054,
      "grad_norm": 0.5173704624176025,
      "learning_rate": 9.694030854634398e-06,
      "loss": 0.4742,
      "step": 8479
    },
    {
      "epoch": 0.389151484557845,
      "grad_norm": 0.4635837972164154,
      "learning_rate": 9.693946395022908e-06,
      "loss": 0.4277,
      "step": 8480
    },
    {
      "epoch": 0.3891973750630994,
      "grad_norm": 0.46322330832481384,
      "learning_rate": 9.693861924123959e-06,
      "loss": 0.3846,
      "step": 8481
    },
    {
      "epoch": 0.3892432655683539,
      "grad_norm": 0.4779599905014038,
      "learning_rate": 9.693777441937755e-06,
      "loss": 0.4831,
      "step": 8482
    },
    {
      "epoch": 0.38928915607360837,
      "grad_norm": 0.4560511112213135,
      "learning_rate": 9.693692948464495e-06,
      "loss": 0.4042,
      "step": 8483
    },
    {
      "epoch": 0.3893350465788628,
      "grad_norm": 0.5237135887145996,
      "learning_rate": 9.693608443704388e-06,
      "loss": 0.4776,
      "step": 8484
    },
    {
      "epoch": 0.3893809370841173,
      "grad_norm": 0.5197299718856812,
      "learning_rate": 9.69352392765763e-06,
      "loss": 0.6065,
      "step": 8485
    },
    {
      "epoch": 0.38942682758937175,
      "grad_norm": 0.4203716218471527,
      "learning_rate": 9.693439400324431e-06,
      "loss": 0.3024,
      "step": 8486
    },
    {
      "epoch": 0.38947271809462625,
      "grad_norm": 0.4687536060810089,
      "learning_rate": 9.693354861704991e-06,
      "loss": 0.4207,
      "step": 8487
    },
    {
      "epoch": 0.3895186085998807,
      "grad_norm": 0.45304107666015625,
      "learning_rate": 9.693270311799515e-06,
      "loss": 0.3672,
      "step": 8488
    },
    {
      "epoch": 0.38956449910513513,
      "grad_norm": 0.4355490505695343,
      "learning_rate": 9.693185750608204e-06,
      "loss": 0.3401,
      "step": 8489
    },
    {
      "epoch": 0.38961038961038963,
      "grad_norm": 0.43713536858558655,
      "learning_rate": 9.69310117813126e-06,
      "loss": 0.375,
      "step": 8490
    },
    {
      "epoch": 0.38965628011564407,
      "grad_norm": 0.439841091632843,
      "learning_rate": 9.693016594368893e-06,
      "loss": 0.3499,
      "step": 8491
    },
    {
      "epoch": 0.3897021706208985,
      "grad_norm": 0.49400460720062256,
      "learning_rate": 9.6929319993213e-06,
      "loss": 0.4991,
      "step": 8492
    },
    {
      "epoch": 0.389748061126153,
      "grad_norm": 0.4501924514770508,
      "learning_rate": 9.692847392988686e-06,
      "loss": 0.3877,
      "step": 8493
    },
    {
      "epoch": 0.38979395163140745,
      "grad_norm": 0.5039815306663513,
      "learning_rate": 9.692762775371259e-06,
      "loss": 0.4918,
      "step": 8494
    },
    {
      "epoch": 0.38983984213666195,
      "grad_norm": 0.4276253581047058,
      "learning_rate": 9.692678146469214e-06,
      "loss": 0.329,
      "step": 8495
    },
    {
      "epoch": 0.3898857326419164,
      "grad_norm": 0.43700525164604187,
      "learning_rate": 9.692593506282762e-06,
      "loss": 0.3587,
      "step": 8496
    },
    {
      "epoch": 0.38993162314717084,
      "grad_norm": 0.4485294818878174,
      "learning_rate": 9.692508854812102e-06,
      "loss": 0.3581,
      "step": 8497
    },
    {
      "epoch": 0.38997751365242533,
      "grad_norm": 0.47604647278785706,
      "learning_rate": 9.692424192057441e-06,
      "loss": 0.3977,
      "step": 8498
    },
    {
      "epoch": 0.3900234041576798,
      "grad_norm": 0.4553724229335785,
      "learning_rate": 9.692339518018978e-06,
      "loss": 0.3736,
      "step": 8499
    },
    {
      "epoch": 0.3900692946629342,
      "grad_norm": 0.45458129048347473,
      "learning_rate": 9.692254832696921e-06,
      "loss": 0.3968,
      "step": 8500
    },
    {
      "epoch": 0.3901151851681887,
      "grad_norm": 0.46565794944763184,
      "learning_rate": 9.692170136091472e-06,
      "loss": 0.4006,
      "step": 8501
    },
    {
      "epoch": 0.39016107567344316,
      "grad_norm": 0.5006628632545471,
      "learning_rate": 9.692085428202835e-06,
      "loss": 0.4566,
      "step": 8502
    },
    {
      "epoch": 0.3902069661786976,
      "grad_norm": 0.48373353481292725,
      "learning_rate": 9.692000709031214e-06,
      "loss": 0.4101,
      "step": 8503
    },
    {
      "epoch": 0.3902528566839521,
      "grad_norm": 0.48382923007011414,
      "learning_rate": 9.691915978576812e-06,
      "loss": 0.4183,
      "step": 8504
    },
    {
      "epoch": 0.39029874718920654,
      "grad_norm": 0.4850400984287262,
      "learning_rate": 9.691831236839832e-06,
      "loss": 0.4086,
      "step": 8505
    },
    {
      "epoch": 0.39034463769446104,
      "grad_norm": 0.46354368329048157,
      "learning_rate": 9.69174648382048e-06,
      "loss": 0.4245,
      "step": 8506
    },
    {
      "epoch": 0.3903905281997155,
      "grad_norm": 0.47254428267478943,
      "learning_rate": 9.691661719518955e-06,
      "loss": 0.4764,
      "step": 8507
    },
    {
      "epoch": 0.3904364187049699,
      "grad_norm": 0.46487656235694885,
      "learning_rate": 9.691576943935468e-06,
      "loss": 0.4001,
      "step": 8508
    },
    {
      "epoch": 0.3904823092102244,
      "grad_norm": 0.4596210718154907,
      "learning_rate": 9.691492157070217e-06,
      "loss": 0.3776,
      "step": 8509
    },
    {
      "epoch": 0.39052819971547886,
      "grad_norm": 0.4168436527252197,
      "learning_rate": 9.691407358923409e-06,
      "loss": 0.3023,
      "step": 8510
    },
    {
      "epoch": 0.3905740902207333,
      "grad_norm": 0.45442837476730347,
      "learning_rate": 9.691322549495247e-06,
      "loss": 0.3876,
      "step": 8511
    },
    {
      "epoch": 0.3906199807259878,
      "grad_norm": 0.4610825777053833,
      "learning_rate": 9.691237728785934e-06,
      "loss": 0.4181,
      "step": 8512
    },
    {
      "epoch": 0.39066587123124225,
      "grad_norm": 0.4814831614494324,
      "learning_rate": 9.691152896795676e-06,
      "loss": 0.4046,
      "step": 8513
    },
    {
      "epoch": 0.39071176173649674,
      "grad_norm": 0.43903297185897827,
      "learning_rate": 9.691068053524675e-06,
      "loss": 0.4178,
      "step": 8514
    },
    {
      "epoch": 0.3907576522417512,
      "grad_norm": 0.4849167466163635,
      "learning_rate": 9.690983198973136e-06,
      "loss": 0.4721,
      "step": 8515
    },
    {
      "epoch": 0.39080354274700563,
      "grad_norm": 0.48738500475883484,
      "learning_rate": 9.690898333141264e-06,
      "loss": 0.3764,
      "step": 8516
    },
    {
      "epoch": 0.3908494332522601,
      "grad_norm": 0.45710647106170654,
      "learning_rate": 9.690813456029262e-06,
      "loss": 0.4057,
      "step": 8517
    },
    {
      "epoch": 0.39089532375751457,
      "grad_norm": 0.46582651138305664,
      "learning_rate": 9.690728567637332e-06,
      "loss": 0.4071,
      "step": 8518
    },
    {
      "epoch": 0.390941214262769,
      "grad_norm": 0.4486205577850342,
      "learning_rate": 9.690643667965682e-06,
      "loss": 0.3692,
      "step": 8519
    },
    {
      "epoch": 0.3909871047680235,
      "grad_norm": 0.44895729422569275,
      "learning_rate": 9.690558757014514e-06,
      "loss": 0.3793,
      "step": 8520
    },
    {
      "epoch": 0.39103299527327795,
      "grad_norm": 0.515789270401001,
      "learning_rate": 9.690473834784032e-06,
      "loss": 0.4626,
      "step": 8521
    },
    {
      "epoch": 0.39107888577853245,
      "grad_norm": 0.43202105164527893,
      "learning_rate": 9.690388901274442e-06,
      "loss": 0.3591,
      "step": 8522
    },
    {
      "epoch": 0.3911247762837869,
      "grad_norm": 0.46173349022865295,
      "learning_rate": 9.690303956485946e-06,
      "loss": 0.4006,
      "step": 8523
    },
    {
      "epoch": 0.39117066678904133,
      "grad_norm": 0.4614937901496887,
      "learning_rate": 9.69021900041875e-06,
      "loss": 0.3967,
      "step": 8524
    },
    {
      "epoch": 0.39121655729429583,
      "grad_norm": 0.4738675355911255,
      "learning_rate": 9.690134033073056e-06,
      "loss": 0.4006,
      "step": 8525
    },
    {
      "epoch": 0.3912624477995503,
      "grad_norm": 0.4683053493499756,
      "learning_rate": 9.690049054449071e-06,
      "loss": 0.4394,
      "step": 8526
    },
    {
      "epoch": 0.3913083383048047,
      "grad_norm": 0.4362429678440094,
      "learning_rate": 9.689964064546999e-06,
      "loss": 0.3818,
      "step": 8527
    },
    {
      "epoch": 0.3913542288100592,
      "grad_norm": 0.431314617395401,
      "learning_rate": 9.689879063367042e-06,
      "loss": 0.3552,
      "step": 8528
    },
    {
      "epoch": 0.39140011931531365,
      "grad_norm": 0.4697848856449127,
      "learning_rate": 9.689794050909408e-06,
      "loss": 0.5138,
      "step": 8529
    },
    {
      "epoch": 0.3914460098205681,
      "grad_norm": 0.4638991951942444,
      "learning_rate": 9.689709027174299e-06,
      "loss": 0.435,
      "step": 8530
    },
    {
      "epoch": 0.3914919003258226,
      "grad_norm": 0.44568994641304016,
      "learning_rate": 9.689623992161918e-06,
      "loss": 0.3829,
      "step": 8531
    },
    {
      "epoch": 0.39153779083107704,
      "grad_norm": 0.4965907335281372,
      "learning_rate": 9.689538945872473e-06,
      "loss": 0.3303,
      "step": 8532
    },
    {
      "epoch": 0.39158368133633153,
      "grad_norm": 0.47154876589775085,
      "learning_rate": 9.689453888306167e-06,
      "loss": 0.3669,
      "step": 8533
    },
    {
      "epoch": 0.391629571841586,
      "grad_norm": 0.47644510865211487,
      "learning_rate": 9.689368819463204e-06,
      "loss": 0.4193,
      "step": 8534
    },
    {
      "epoch": 0.3916754623468404,
      "grad_norm": 0.43990036845207214,
      "learning_rate": 9.68928373934379e-06,
      "loss": 0.3319,
      "step": 8535
    },
    {
      "epoch": 0.3917213528520949,
      "grad_norm": 0.47022491693496704,
      "learning_rate": 9.689198647948126e-06,
      "loss": 0.4163,
      "step": 8536
    },
    {
      "epoch": 0.39176724335734936,
      "grad_norm": 0.4544909596443176,
      "learning_rate": 9.68911354527642e-06,
      "loss": 0.4216,
      "step": 8537
    },
    {
      "epoch": 0.3918131338626038,
      "grad_norm": 0.4939616322517395,
      "learning_rate": 9.689028431328877e-06,
      "loss": 0.4723,
      "step": 8538
    },
    {
      "epoch": 0.3918590243678583,
      "grad_norm": 0.46473363041877747,
      "learning_rate": 9.6889433061057e-06,
      "loss": 0.4063,
      "step": 8539
    },
    {
      "epoch": 0.39190491487311274,
      "grad_norm": 0.5054740309715271,
      "learning_rate": 9.688858169607095e-06,
      "loss": 0.4681,
      "step": 8540
    },
    {
      "epoch": 0.39195080537836724,
      "grad_norm": 0.43713995814323425,
      "learning_rate": 9.688773021833266e-06,
      "loss": 0.3545,
      "step": 8541
    },
    {
      "epoch": 0.3919966958836217,
      "grad_norm": 0.47141167521476746,
      "learning_rate": 9.688687862784416e-06,
      "loss": 0.4337,
      "step": 8542
    },
    {
      "epoch": 0.3920425863888761,
      "grad_norm": 0.5062408447265625,
      "learning_rate": 9.688602692460754e-06,
      "loss": 0.5004,
      "step": 8543
    },
    {
      "epoch": 0.3920884768941306,
      "grad_norm": 0.5107776522636414,
      "learning_rate": 9.688517510862481e-06,
      "loss": 0.5179,
      "step": 8544
    },
    {
      "epoch": 0.39213436739938506,
      "grad_norm": 0.477706640958786,
      "learning_rate": 9.688432317989802e-06,
      "loss": 0.4731,
      "step": 8545
    },
    {
      "epoch": 0.3921802579046395,
      "grad_norm": 0.4755861759185791,
      "learning_rate": 9.688347113842924e-06,
      "loss": 0.4613,
      "step": 8546
    },
    {
      "epoch": 0.392226148409894,
      "grad_norm": 0.44089922308921814,
      "learning_rate": 9.688261898422052e-06,
      "loss": 0.3978,
      "step": 8547
    },
    {
      "epoch": 0.39227203891514845,
      "grad_norm": 0.4510617256164551,
      "learning_rate": 9.688176671727387e-06,
      "loss": 0.4135,
      "step": 8548
    },
    {
      "epoch": 0.39231792942040294,
      "grad_norm": 0.4418678879737854,
      "learning_rate": 9.68809143375914e-06,
      "loss": 0.3326,
      "step": 8549
    },
    {
      "epoch": 0.3923638199256574,
      "grad_norm": 0.46288397908210754,
      "learning_rate": 9.68800618451751e-06,
      "loss": 0.3966,
      "step": 8550
    },
    {
      "epoch": 0.39240971043091183,
      "grad_norm": 0.4673866331577301,
      "learning_rate": 9.687920924002705e-06,
      "loss": 0.426,
      "step": 8551
    },
    {
      "epoch": 0.3924556009361663,
      "grad_norm": 0.46155846118927,
      "learning_rate": 9.687835652214932e-06,
      "loss": 0.3853,
      "step": 8552
    },
    {
      "epoch": 0.39250149144142077,
      "grad_norm": 0.494147926568985,
      "learning_rate": 9.687750369154391e-06,
      "loss": 0.4917,
      "step": 8553
    },
    {
      "epoch": 0.3925473819466752,
      "grad_norm": 0.48230621218681335,
      "learning_rate": 9.687665074821291e-06,
      "loss": 0.4006,
      "step": 8554
    },
    {
      "epoch": 0.3925932724519297,
      "grad_norm": 0.473418265581131,
      "learning_rate": 9.687579769215837e-06,
      "loss": 0.4253,
      "step": 8555
    },
    {
      "epoch": 0.39263916295718415,
      "grad_norm": 0.5901590585708618,
      "learning_rate": 9.687494452338232e-06,
      "loss": 0.3793,
      "step": 8556
    },
    {
      "epoch": 0.39268505346243865,
      "grad_norm": 0.46326932311058044,
      "learning_rate": 9.687409124188682e-06,
      "loss": 0.399,
      "step": 8557
    },
    {
      "epoch": 0.3927309439676931,
      "grad_norm": 0.4534607231616974,
      "learning_rate": 9.687323784767394e-06,
      "loss": 0.4032,
      "step": 8558
    },
    {
      "epoch": 0.39277683447294753,
      "grad_norm": 0.42637670040130615,
      "learning_rate": 9.68723843407457e-06,
      "loss": 0.3167,
      "step": 8559
    },
    {
      "epoch": 0.39282272497820203,
      "grad_norm": 0.4752059280872345,
      "learning_rate": 9.687153072110418e-06,
      "loss": 0.3901,
      "step": 8560
    },
    {
      "epoch": 0.3928686154834565,
      "grad_norm": 0.4701560139656067,
      "learning_rate": 9.68706769887514e-06,
      "loss": 0.4818,
      "step": 8561
    },
    {
      "epoch": 0.3929145059887109,
      "grad_norm": 0.4895174205303192,
      "learning_rate": 9.686982314368946e-06,
      "loss": 0.5422,
      "step": 8562
    },
    {
      "epoch": 0.3929603964939654,
      "grad_norm": 0.435077041387558,
      "learning_rate": 9.686896918592037e-06,
      "loss": 0.3287,
      "step": 8563
    },
    {
      "epoch": 0.39300628699921986,
      "grad_norm": 0.4535768926143646,
      "learning_rate": 9.686811511544622e-06,
      "loss": 0.3589,
      "step": 8564
    },
    {
      "epoch": 0.3930521775044743,
      "grad_norm": 0.4414762556552887,
      "learning_rate": 9.686726093226904e-06,
      "loss": 0.4104,
      "step": 8565
    },
    {
      "epoch": 0.3930980680097288,
      "grad_norm": 0.5018694996833801,
      "learning_rate": 9.686640663639087e-06,
      "loss": 0.4713,
      "step": 8566
    },
    {
      "epoch": 0.39314395851498324,
      "grad_norm": 0.45490220189094543,
      "learning_rate": 9.68655522278138e-06,
      "loss": 0.4358,
      "step": 8567
    },
    {
      "epoch": 0.39318984902023774,
      "grad_norm": 0.5319716930389404,
      "learning_rate": 9.686469770653987e-06,
      "loss": 0.3918,
      "step": 8568
    },
    {
      "epoch": 0.3932357395254922,
      "grad_norm": 0.4732978940010071,
      "learning_rate": 9.686384307257113e-06,
      "loss": 0.429,
      "step": 8569
    },
    {
      "epoch": 0.3932816300307466,
      "grad_norm": 0.47988659143447876,
      "learning_rate": 9.686298832590963e-06,
      "loss": 0.432,
      "step": 8570
    },
    {
      "epoch": 0.3933275205360011,
      "grad_norm": 0.444713830947876,
      "learning_rate": 9.686213346655745e-06,
      "loss": 0.3807,
      "step": 8571
    },
    {
      "epoch": 0.39337341104125556,
      "grad_norm": 0.44191858172416687,
      "learning_rate": 9.686127849451661e-06,
      "loss": 0.3998,
      "step": 8572
    },
    {
      "epoch": 0.39341930154651,
      "grad_norm": 0.49021926522254944,
      "learning_rate": 9.68604234097892e-06,
      "loss": 0.4851,
      "step": 8573
    },
    {
      "epoch": 0.3934651920517645,
      "grad_norm": 0.4527379274368286,
      "learning_rate": 9.685956821237726e-06,
      "loss": 0.4,
      "step": 8574
    },
    {
      "epoch": 0.39351108255701894,
      "grad_norm": 0.45258426666259766,
      "learning_rate": 9.685871290228284e-06,
      "loss": 0.3384,
      "step": 8575
    },
    {
      "epoch": 0.39355697306227344,
      "grad_norm": 0.4111131727695465,
      "learning_rate": 9.685785747950801e-06,
      "loss": 0.3023,
      "step": 8576
    },
    {
      "epoch": 0.3936028635675279,
      "grad_norm": 0.4864082932472229,
      "learning_rate": 9.685700194405482e-06,
      "loss": 0.4327,
      "step": 8577
    },
    {
      "epoch": 0.3936487540727823,
      "grad_norm": 0.4577995538711548,
      "learning_rate": 9.685614629592533e-06,
      "loss": 0.3697,
      "step": 8578
    },
    {
      "epoch": 0.3936946445780368,
      "grad_norm": 0.45337432622909546,
      "learning_rate": 9.685529053512159e-06,
      "loss": 0.4147,
      "step": 8579
    },
    {
      "epoch": 0.39374053508329127,
      "grad_norm": 0.43586283922195435,
      "learning_rate": 9.685443466164567e-06,
      "loss": 0.3508,
      "step": 8580
    },
    {
      "epoch": 0.3937864255885457,
      "grad_norm": 0.43408092856407166,
      "learning_rate": 9.685357867549963e-06,
      "loss": 0.3576,
      "step": 8581
    },
    {
      "epoch": 0.3938323160938002,
      "grad_norm": 0.5230141282081604,
      "learning_rate": 9.68527225766855e-06,
      "loss": 0.5352,
      "step": 8582
    },
    {
      "epoch": 0.39387820659905465,
      "grad_norm": 0.4405350387096405,
      "learning_rate": 9.685186636520538e-06,
      "loss": 0.3796,
      "step": 8583
    },
    {
      "epoch": 0.39392409710430915,
      "grad_norm": 0.45898011326789856,
      "learning_rate": 9.68510100410613e-06,
      "loss": 0.412,
      "step": 8584
    },
    {
      "epoch": 0.3939699876095636,
      "grad_norm": 0.5374167561531067,
      "learning_rate": 9.685015360425532e-06,
      "loss": 0.4964,
      "step": 8585
    },
    {
      "epoch": 0.39401587811481803,
      "grad_norm": 0.4404294788837433,
      "learning_rate": 9.684929705478951e-06,
      "loss": 0.34,
      "step": 8586
    },
    {
      "epoch": 0.39406176862007253,
      "grad_norm": 0.8178809881210327,
      "learning_rate": 9.684844039266594e-06,
      "loss": 0.4853,
      "step": 8587
    },
    {
      "epoch": 0.39410765912532697,
      "grad_norm": 0.45292502641677856,
      "learning_rate": 9.684758361788664e-06,
      "loss": 0.3611,
      "step": 8588
    },
    {
      "epoch": 0.3941535496305814,
      "grad_norm": 0.47277024388313293,
      "learning_rate": 9.68467267304537e-06,
      "loss": 0.4334,
      "step": 8589
    },
    {
      "epoch": 0.3941994401358359,
      "grad_norm": 0.4607960283756256,
      "learning_rate": 9.684586973036915e-06,
      "loss": 0.3686,
      "step": 8590
    },
    {
      "epoch": 0.39424533064109035,
      "grad_norm": 0.4308449923992157,
      "learning_rate": 9.684501261763509e-06,
      "loss": 0.3531,
      "step": 8591
    },
    {
      "epoch": 0.3942912211463448,
      "grad_norm": 0.48661860823631287,
      "learning_rate": 9.684415539225355e-06,
      "loss": 0.4151,
      "step": 8592
    },
    {
      "epoch": 0.3943371116515993,
      "grad_norm": 0.4583309292793274,
      "learning_rate": 9.68432980542266e-06,
      "loss": 0.373,
      "step": 8593
    },
    {
      "epoch": 0.39438300215685373,
      "grad_norm": 0.49229541420936584,
      "learning_rate": 9.684244060355629e-06,
      "loss": 0.4696,
      "step": 8594
    },
    {
      "epoch": 0.39442889266210823,
      "grad_norm": 0.4471207559108734,
      "learning_rate": 9.684158304024472e-06,
      "loss": 0.3559,
      "step": 8595
    },
    {
      "epoch": 0.3944747831673627,
      "grad_norm": 0.4282625615596771,
      "learning_rate": 9.684072536429392e-06,
      "loss": 0.3334,
      "step": 8596
    },
    {
      "epoch": 0.3945206736726171,
      "grad_norm": 0.4677380323410034,
      "learning_rate": 9.683986757570594e-06,
      "loss": 0.4136,
      "step": 8597
    },
    {
      "epoch": 0.3945665641778716,
      "grad_norm": 0.4334053695201874,
      "learning_rate": 9.683900967448288e-06,
      "loss": 0.3456,
      "step": 8598
    },
    {
      "epoch": 0.39461245468312606,
      "grad_norm": 0.5461660623550415,
      "learning_rate": 9.683815166062678e-06,
      "loss": 0.5009,
      "step": 8599
    },
    {
      "epoch": 0.3946583451883805,
      "grad_norm": 0.48494023084640503,
      "learning_rate": 9.683729353413971e-06,
      "loss": 0.4338,
      "step": 8600
    },
    {
      "epoch": 0.394704235693635,
      "grad_norm": 0.47498902678489685,
      "learning_rate": 9.683643529502372e-06,
      "loss": 0.4112,
      "step": 8601
    },
    {
      "epoch": 0.39475012619888944,
      "grad_norm": 0.4361850917339325,
      "learning_rate": 9.68355769432809e-06,
      "loss": 0.2788,
      "step": 8602
    },
    {
      "epoch": 0.39479601670414394,
      "grad_norm": 0.4567633867263794,
      "learning_rate": 9.683471847891327e-06,
      "loss": 0.3805,
      "step": 8603
    },
    {
      "epoch": 0.3948419072093984,
      "grad_norm": 0.5143073201179504,
      "learning_rate": 9.683385990192295e-06,
      "loss": 0.5381,
      "step": 8604
    },
    {
      "epoch": 0.3948877977146528,
      "grad_norm": 0.46747028827667236,
      "learning_rate": 9.683300121231199e-06,
      "loss": 0.4511,
      "step": 8605
    },
    {
      "epoch": 0.3949336882199073,
      "grad_norm": 0.47865384817123413,
      "learning_rate": 9.683214241008242e-06,
      "loss": 0.4808,
      "step": 8606
    },
    {
      "epoch": 0.39497957872516176,
      "grad_norm": 0.4456637501716614,
      "learning_rate": 9.683128349523633e-06,
      "loss": 0.3755,
      "step": 8607
    },
    {
      "epoch": 0.3950254692304162,
      "grad_norm": 0.4276733994483948,
      "learning_rate": 9.683042446777579e-06,
      "loss": 0.3112,
      "step": 8608
    },
    {
      "epoch": 0.3950713597356707,
      "grad_norm": 0.47264665365219116,
      "learning_rate": 9.682956532770286e-06,
      "loss": 0.4235,
      "step": 8609
    },
    {
      "epoch": 0.39511725024092514,
      "grad_norm": 0.4751649498939514,
      "learning_rate": 9.68287060750196e-06,
      "loss": 0.3473,
      "step": 8610
    },
    {
      "epoch": 0.39516314074617964,
      "grad_norm": 0.48298001289367676,
      "learning_rate": 9.682784670972808e-06,
      "loss": 0.3829,
      "step": 8611
    },
    {
      "epoch": 0.3952090312514341,
      "grad_norm": 0.5119484663009644,
      "learning_rate": 9.682698723183036e-06,
      "loss": 0.5422,
      "step": 8612
    },
    {
      "epoch": 0.3952549217566885,
      "grad_norm": 0.44060173630714417,
      "learning_rate": 9.682612764132854e-06,
      "loss": 0.3846,
      "step": 8613
    },
    {
      "epoch": 0.395300812261943,
      "grad_norm": 0.45631831884384155,
      "learning_rate": 9.682526793822465e-06,
      "loss": 0.4581,
      "step": 8614
    },
    {
      "epoch": 0.39534670276719747,
      "grad_norm": 0.4429319500923157,
      "learning_rate": 9.682440812252077e-06,
      "loss": 0.3627,
      "step": 8615
    },
    {
      "epoch": 0.3953925932724519,
      "grad_norm": 0.4876386225223541,
      "learning_rate": 9.682354819421896e-06,
      "loss": 0.5217,
      "step": 8616
    },
    {
      "epoch": 0.3954384837777064,
      "grad_norm": 0.436872273683548,
      "learning_rate": 9.68226881533213e-06,
      "loss": 0.3166,
      "step": 8617
    },
    {
      "epoch": 0.39548437428296085,
      "grad_norm": 0.5447903275489807,
      "learning_rate": 9.682182799982985e-06,
      "loss": 0.5565,
      "step": 8618
    },
    {
      "epoch": 0.3955302647882153,
      "grad_norm": 0.44715312123298645,
      "learning_rate": 9.682096773374668e-06,
      "loss": 0.3737,
      "step": 8619
    },
    {
      "epoch": 0.3955761552934698,
      "grad_norm": 0.4636843204498291,
      "learning_rate": 9.682010735507387e-06,
      "loss": 0.4234,
      "step": 8620
    },
    {
      "epoch": 0.39562204579872423,
      "grad_norm": 0.49817508459091187,
      "learning_rate": 9.681924686381348e-06,
      "loss": 0.552,
      "step": 8621
    },
    {
      "epoch": 0.39566793630397873,
      "grad_norm": 0.4809209704399109,
      "learning_rate": 9.681838625996755e-06,
      "loss": 0.4036,
      "step": 8622
    },
    {
      "epoch": 0.39571382680923317,
      "grad_norm": 0.43769770860671997,
      "learning_rate": 9.68175255435382e-06,
      "loss": 0.3464,
      "step": 8623
    },
    {
      "epoch": 0.3957597173144876,
      "grad_norm": 0.4978189170360565,
      "learning_rate": 9.681666471452748e-06,
      "loss": 0.3743,
      "step": 8624
    },
    {
      "epoch": 0.3958056078197421,
      "grad_norm": 0.42524853348731995,
      "learning_rate": 9.681580377293744e-06,
      "loss": 0.3209,
      "step": 8625
    },
    {
      "epoch": 0.39585149832499655,
      "grad_norm": 0.4592619836330414,
      "learning_rate": 9.68149427187702e-06,
      "loss": 0.3811,
      "step": 8626
    },
    {
      "epoch": 0.395897388830251,
      "grad_norm": 0.49542543292045593,
      "learning_rate": 9.681408155202776e-06,
      "loss": 0.4291,
      "step": 8627
    },
    {
      "epoch": 0.3959432793355055,
      "grad_norm": 0.47730204463005066,
      "learning_rate": 9.681322027271225e-06,
      "loss": 0.4148,
      "step": 8628
    },
    {
      "epoch": 0.39598916984075994,
      "grad_norm": 0.46925047039985657,
      "learning_rate": 9.68123588808257e-06,
      "loss": 0.3972,
      "step": 8629
    },
    {
      "epoch": 0.39603506034601443,
      "grad_norm": 0.47826483845710754,
      "learning_rate": 9.681149737637023e-06,
      "loss": 0.404,
      "step": 8630
    },
    {
      "epoch": 0.3960809508512689,
      "grad_norm": 0.47326555848121643,
      "learning_rate": 9.681063575934786e-06,
      "loss": 0.4716,
      "step": 8631
    },
    {
      "epoch": 0.3961268413565233,
      "grad_norm": 0.46842607855796814,
      "learning_rate": 9.68097740297607e-06,
      "loss": 0.3945,
      "step": 8632
    },
    {
      "epoch": 0.3961727318617778,
      "grad_norm": 0.444974809885025,
      "learning_rate": 9.68089121876108e-06,
      "loss": 0.3945,
      "step": 8633
    },
    {
      "epoch": 0.39621862236703226,
      "grad_norm": 0.4635046720504761,
      "learning_rate": 9.680805023290022e-06,
      "loss": 0.3563,
      "step": 8634
    },
    {
      "epoch": 0.3962645128722867,
      "grad_norm": 0.6594986915588379,
      "learning_rate": 9.680718816563108e-06,
      "loss": 0.452,
      "step": 8635
    },
    {
      "epoch": 0.3963104033775412,
      "grad_norm": 0.4828979969024658,
      "learning_rate": 9.680632598580543e-06,
      "loss": 0.3869,
      "step": 8636
    },
    {
      "epoch": 0.39635629388279564,
      "grad_norm": 0.4403561055660248,
      "learning_rate": 9.680546369342532e-06,
      "loss": 0.3422,
      "step": 8637
    },
    {
      "epoch": 0.39640218438805014,
      "grad_norm": 0.4655972421169281,
      "learning_rate": 9.680460128849285e-06,
      "loss": 0.4474,
      "step": 8638
    },
    {
      "epoch": 0.3964480748933046,
      "grad_norm": 0.5102384686470032,
      "learning_rate": 9.680373877101008e-06,
      "loss": 0.5371,
      "step": 8639
    },
    {
      "epoch": 0.396493965398559,
      "grad_norm": 0.5012203454971313,
      "learning_rate": 9.68028761409791e-06,
      "loss": 0.4112,
      "step": 8640
    },
    {
      "epoch": 0.3965398559038135,
      "grad_norm": 0.49299463629722595,
      "learning_rate": 9.680201339840197e-06,
      "loss": 0.4381,
      "step": 8641
    },
    {
      "epoch": 0.39658574640906796,
      "grad_norm": 0.45829281210899353,
      "learning_rate": 9.680115054328077e-06,
      "loss": 0.4173,
      "step": 8642
    },
    {
      "epoch": 0.3966316369143224,
      "grad_norm": 0.4389673173427582,
      "learning_rate": 9.680028757561757e-06,
      "loss": 0.3092,
      "step": 8643
    },
    {
      "epoch": 0.3966775274195769,
      "grad_norm": 0.4829205870628357,
      "learning_rate": 9.679942449541444e-06,
      "loss": 0.4208,
      "step": 8644
    },
    {
      "epoch": 0.39672341792483135,
      "grad_norm": 0.45412907004356384,
      "learning_rate": 9.679856130267348e-06,
      "loss": 0.3746,
      "step": 8645
    },
    {
      "epoch": 0.3967693084300858,
      "grad_norm": 0.44100072979927063,
      "learning_rate": 9.679769799739674e-06,
      "loss": 0.3858,
      "step": 8646
    },
    {
      "epoch": 0.3968151989353403,
      "grad_norm": 0.4650171399116516,
      "learning_rate": 9.67968345795863e-06,
      "loss": 0.3736,
      "step": 8647
    },
    {
      "epoch": 0.3968610894405947,
      "grad_norm": 0.4790247678756714,
      "learning_rate": 9.679597104924426e-06,
      "loss": 0.436,
      "step": 8648
    },
    {
      "epoch": 0.3969069799458492,
      "grad_norm": 0.43356242775917053,
      "learning_rate": 9.679510740637268e-06,
      "loss": 0.35,
      "step": 8649
    },
    {
      "epoch": 0.39695287045110367,
      "grad_norm": 0.47927725315093994,
      "learning_rate": 9.679424365097363e-06,
      "loss": 0.4029,
      "step": 8650
    },
    {
      "epoch": 0.3969987609563581,
      "grad_norm": 0.46591562032699585,
      "learning_rate": 9.67933797830492e-06,
      "loss": 0.4259,
      "step": 8651
    },
    {
      "epoch": 0.3970446514616126,
      "grad_norm": 0.551018476486206,
      "learning_rate": 9.679251580260142e-06,
      "loss": 0.4364,
      "step": 8652
    },
    {
      "epoch": 0.39709054196686705,
      "grad_norm": 0.5083770155906677,
      "learning_rate": 9.679165170963244e-06,
      "loss": 0.5282,
      "step": 8653
    },
    {
      "epoch": 0.3971364324721215,
      "grad_norm": 0.4843112826347351,
      "learning_rate": 9.679078750414432e-06,
      "loss": 0.4272,
      "step": 8654
    },
    {
      "epoch": 0.397182322977376,
      "grad_norm": 0.4888418912887573,
      "learning_rate": 9.67899231861391e-06,
      "loss": 0.4052,
      "step": 8655
    },
    {
      "epoch": 0.39722821348263043,
      "grad_norm": 0.449928879737854,
      "learning_rate": 9.67890587556189e-06,
      "loss": 0.3592,
      "step": 8656
    },
    {
      "epoch": 0.39727410398788493,
      "grad_norm": 0.43793001770973206,
      "learning_rate": 9.678819421258577e-06,
      "loss": 0.3789,
      "step": 8657
    },
    {
      "epoch": 0.3973199944931394,
      "grad_norm": 0.47794607281684875,
      "learning_rate": 9.67873295570418e-06,
      "loss": 0.4464,
      "step": 8658
    },
    {
      "epoch": 0.3973658849983938,
      "grad_norm": 0.43191906809806824,
      "learning_rate": 9.678646478898905e-06,
      "loss": 0.3627,
      "step": 8659
    },
    {
      "epoch": 0.3974117755036483,
      "grad_norm": 0.44276878237724304,
      "learning_rate": 9.678559990842964e-06,
      "loss": 0.3922,
      "step": 8660
    },
    {
      "epoch": 0.39745766600890275,
      "grad_norm": 0.49213507771492004,
      "learning_rate": 9.678473491536563e-06,
      "loss": 0.4544,
      "step": 8661
    },
    {
      "epoch": 0.3975035565141572,
      "grad_norm": 0.4666652977466583,
      "learning_rate": 9.678386980979909e-06,
      "loss": 0.4494,
      "step": 8662
    },
    {
      "epoch": 0.3975494470194117,
      "grad_norm": 0.4337114691734314,
      "learning_rate": 9.678300459173212e-06,
      "loss": 0.3423,
      "step": 8663
    },
    {
      "epoch": 0.39759533752466614,
      "grad_norm": 0.47524869441986084,
      "learning_rate": 9.678213926116677e-06,
      "loss": 0.4516,
      "step": 8664
    },
    {
      "epoch": 0.39764122802992063,
      "grad_norm": 0.4766837954521179,
      "learning_rate": 9.678127381810516e-06,
      "loss": 0.3835,
      "step": 8665
    },
    {
      "epoch": 0.3976871185351751,
      "grad_norm": 0.4313611686229706,
      "learning_rate": 9.678040826254935e-06,
      "loss": 0.3573,
      "step": 8666
    },
    {
      "epoch": 0.3977330090404295,
      "grad_norm": 0.4806932210922241,
      "learning_rate": 9.67795425945014e-06,
      "loss": 0.4467,
      "step": 8667
    },
    {
      "epoch": 0.397778899545684,
      "grad_norm": 0.46154314279556274,
      "learning_rate": 9.677867681396343e-06,
      "loss": 0.3718,
      "step": 8668
    },
    {
      "epoch": 0.39782479005093846,
      "grad_norm": 0.46274620294570923,
      "learning_rate": 9.677781092093749e-06,
      "loss": 0.3601,
      "step": 8669
    },
    {
      "epoch": 0.3978706805561929,
      "grad_norm": 0.49439704418182373,
      "learning_rate": 9.677694491542571e-06,
      "loss": 0.4287,
      "step": 8670
    },
    {
      "epoch": 0.3979165710614474,
      "grad_norm": 0.4502599537372589,
      "learning_rate": 9.677607879743011e-06,
      "loss": 0.3675,
      "step": 8671
    },
    {
      "epoch": 0.39796246156670184,
      "grad_norm": 0.46441757678985596,
      "learning_rate": 9.677521256695282e-06,
      "loss": 0.3799,
      "step": 8672
    },
    {
      "epoch": 0.39800835207195634,
      "grad_norm": 0.4544665813446045,
      "learning_rate": 9.67743462239959e-06,
      "loss": 0.3737,
      "step": 8673
    },
    {
      "epoch": 0.3980542425772108,
      "grad_norm": 0.4876054525375366,
      "learning_rate": 9.677347976856145e-06,
      "loss": 0.4533,
      "step": 8674
    },
    {
      "epoch": 0.3981001330824652,
      "grad_norm": 0.43368858098983765,
      "learning_rate": 9.677261320065155e-06,
      "loss": 0.3311,
      "step": 8675
    },
    {
      "epoch": 0.3981460235877197,
      "grad_norm": 0.486748069524765,
      "learning_rate": 9.677174652026825e-06,
      "loss": 0.4155,
      "step": 8676
    },
    {
      "epoch": 0.39819191409297416,
      "grad_norm": 0.4976727068424225,
      "learning_rate": 9.677087972741368e-06,
      "loss": 0.4726,
      "step": 8677
    },
    {
      "epoch": 0.3982378045982286,
      "grad_norm": 0.433361291885376,
      "learning_rate": 9.67700128220899e-06,
      "loss": 0.3223,
      "step": 8678
    },
    {
      "epoch": 0.3982836951034831,
      "grad_norm": 0.4959779381752014,
      "learning_rate": 9.676914580429898e-06,
      "loss": 0.4811,
      "step": 8679
    },
    {
      "epoch": 0.39832958560873755,
      "grad_norm": 0.5422683954238892,
      "learning_rate": 9.676827867404306e-06,
      "loss": 0.4972,
      "step": 8680
    },
    {
      "epoch": 0.398375476113992,
      "grad_norm": 0.4244866669178009,
      "learning_rate": 9.676741143132417e-06,
      "loss": 0.3469,
      "step": 8681
    },
    {
      "epoch": 0.3984213666192465,
      "grad_norm": 0.5025492310523987,
      "learning_rate": 9.676654407614442e-06,
      "loss": 0.4004,
      "step": 8682
    },
    {
      "epoch": 0.39846725712450093,
      "grad_norm": 0.5294975638389587,
      "learning_rate": 9.67656766085059e-06,
      "loss": 0.49,
      "step": 8683
    },
    {
      "epoch": 0.3985131476297554,
      "grad_norm": 0.4405219852924347,
      "learning_rate": 9.676480902841068e-06,
      "loss": 0.4164,
      "step": 8684
    },
    {
      "epoch": 0.39855903813500987,
      "grad_norm": 0.48667651414871216,
      "learning_rate": 9.676394133586086e-06,
      "loss": 0.3533,
      "step": 8685
    },
    {
      "epoch": 0.3986049286402643,
      "grad_norm": 0.5046839714050293,
      "learning_rate": 9.67630735308585e-06,
      "loss": 0.4598,
      "step": 8686
    },
    {
      "epoch": 0.3986508191455188,
      "grad_norm": 0.4614804983139038,
      "learning_rate": 9.676220561340573e-06,
      "loss": 0.4477,
      "step": 8687
    },
    {
      "epoch": 0.39869670965077325,
      "grad_norm": 0.4799266457557678,
      "learning_rate": 9.676133758350462e-06,
      "loss": 0.4523,
      "step": 8688
    },
    {
      "epoch": 0.3987426001560277,
      "grad_norm": 0.5135653018951416,
      "learning_rate": 9.676046944115723e-06,
      "loss": 0.4636,
      "step": 8689
    },
    {
      "epoch": 0.3987884906612822,
      "grad_norm": 0.45647597312927246,
      "learning_rate": 9.675960118636567e-06,
      "loss": 0.3858,
      "step": 8690
    },
    {
      "epoch": 0.39883438116653663,
      "grad_norm": 0.46854186058044434,
      "learning_rate": 9.675873281913205e-06,
      "loss": 0.3809,
      "step": 8691
    },
    {
      "epoch": 0.39888027167179113,
      "grad_norm": 0.4501729905605316,
      "learning_rate": 9.67578643394584e-06,
      "loss": 0.3927,
      "step": 8692
    },
    {
      "epoch": 0.3989261621770456,
      "grad_norm": 0.4693083167076111,
      "learning_rate": 9.675699574734686e-06,
      "loss": 0.3742,
      "step": 8693
    },
    {
      "epoch": 0.3989720526823,
      "grad_norm": 0.44620537757873535,
      "learning_rate": 9.67561270427995e-06,
      "loss": 0.3338,
      "step": 8694
    },
    {
      "epoch": 0.3990179431875545,
      "grad_norm": 0.4076438248157501,
      "learning_rate": 9.675525822581841e-06,
      "loss": 0.3323,
      "step": 8695
    },
    {
      "epoch": 0.39906383369280896,
      "grad_norm": 0.48204880952835083,
      "learning_rate": 9.675438929640568e-06,
      "loss": 0.3734,
      "step": 8696
    },
    {
      "epoch": 0.3991097241980634,
      "grad_norm": 0.5527828335762024,
      "learning_rate": 9.675352025456341e-06,
      "loss": 0.4371,
      "step": 8697
    },
    {
      "epoch": 0.3991556147033179,
      "grad_norm": 0.4597312808036804,
      "learning_rate": 9.675265110029368e-06,
      "loss": 0.415,
      "step": 8698
    },
    {
      "epoch": 0.39920150520857234,
      "grad_norm": 0.47532808780670166,
      "learning_rate": 9.675178183359858e-06,
      "loss": 0.4687,
      "step": 8699
    },
    {
      "epoch": 0.39924739571382684,
      "grad_norm": 0.5203988552093506,
      "learning_rate": 9.675091245448019e-06,
      "loss": 0.4401,
      "step": 8700
    },
    {
      "epoch": 0.3992932862190813,
      "grad_norm": 0.4371298849582672,
      "learning_rate": 9.675004296294062e-06,
      "loss": 0.416,
      "step": 8701
    },
    {
      "epoch": 0.3993391767243357,
      "grad_norm": 0.4408140480518341,
      "learning_rate": 9.674917335898194e-06,
      "loss": 0.3721,
      "step": 8702
    },
    {
      "epoch": 0.3993850672295902,
      "grad_norm": 0.4458071291446686,
      "learning_rate": 9.674830364260625e-06,
      "loss": 0.3857,
      "step": 8703
    },
    {
      "epoch": 0.39943095773484466,
      "grad_norm": 0.4959181249141693,
      "learning_rate": 9.674743381381567e-06,
      "loss": 0.4705,
      "step": 8704
    },
    {
      "epoch": 0.3994768482400991,
      "grad_norm": 0.5098963975906372,
      "learning_rate": 9.674656387261224e-06,
      "loss": 0.4585,
      "step": 8705
    },
    {
      "epoch": 0.3995227387453536,
      "grad_norm": 0.49235400557518005,
      "learning_rate": 9.67456938189981e-06,
      "loss": 0.4994,
      "step": 8706
    },
    {
      "epoch": 0.39956862925060804,
      "grad_norm": 0.4554671347141266,
      "learning_rate": 9.674482365297529e-06,
      "loss": 0.3529,
      "step": 8707
    },
    {
      "epoch": 0.3996145197558625,
      "grad_norm": 0.5342531204223633,
      "learning_rate": 9.674395337454594e-06,
      "loss": 0.4535,
      "step": 8708
    },
    {
      "epoch": 0.399660410261117,
      "grad_norm": 0.44364133477211,
      "learning_rate": 9.674308298371215e-06,
      "loss": 0.3678,
      "step": 8709
    },
    {
      "epoch": 0.3997063007663714,
      "grad_norm": 0.47711804509162903,
      "learning_rate": 9.674221248047599e-06,
      "loss": 0.4373,
      "step": 8710
    },
    {
      "epoch": 0.3997521912716259,
      "grad_norm": 0.4616168439388275,
      "learning_rate": 9.674134186483957e-06,
      "loss": 0.3191,
      "step": 8711
    },
    {
      "epoch": 0.39979808177688037,
      "grad_norm": 0.469154417514801,
      "learning_rate": 9.674047113680496e-06,
      "loss": 0.3913,
      "step": 8712
    },
    {
      "epoch": 0.3998439722821348,
      "grad_norm": 0.5800721645355225,
      "learning_rate": 9.673960029637429e-06,
      "loss": 0.5304,
      "step": 8713
    },
    {
      "epoch": 0.3998898627873893,
      "grad_norm": 0.4438539147377014,
      "learning_rate": 9.67387293435496e-06,
      "loss": 0.4213,
      "step": 8714
    },
    {
      "epoch": 0.39993575329264375,
      "grad_norm": 0.4614817202091217,
      "learning_rate": 9.673785827833303e-06,
      "loss": 0.4235,
      "step": 8715
    },
    {
      "epoch": 0.3999816437978982,
      "grad_norm": 0.4509181082248688,
      "learning_rate": 9.673698710072668e-06,
      "loss": 0.4535,
      "step": 8716
    },
    {
      "epoch": 0.4000275343031527,
      "grad_norm": 0.45474687218666077,
      "learning_rate": 9.673611581073262e-06,
      "loss": 0.3889,
      "step": 8717
    },
    {
      "epoch": 0.40007342480840713,
      "grad_norm": 0.4203234314918518,
      "learning_rate": 9.673524440835294e-06,
      "loss": 0.3204,
      "step": 8718
    },
    {
      "epoch": 0.4001193153136616,
      "grad_norm": 0.4642297625541687,
      "learning_rate": 9.673437289358975e-06,
      "loss": 0.4379,
      "step": 8719
    },
    {
      "epoch": 0.40016520581891607,
      "grad_norm": 0.5028543472290039,
      "learning_rate": 9.673350126644513e-06,
      "loss": 0.5018,
      "step": 8720
    },
    {
      "epoch": 0.4002110963241705,
      "grad_norm": 0.44732141494750977,
      "learning_rate": 9.673262952692122e-06,
      "loss": 0.3907,
      "step": 8721
    },
    {
      "epoch": 0.400256986829425,
      "grad_norm": 0.4514525532722473,
      "learning_rate": 9.673175767502006e-06,
      "loss": 0.4066,
      "step": 8722
    },
    {
      "epoch": 0.40030287733467945,
      "grad_norm": 0.5037662386894226,
      "learning_rate": 9.673088571074378e-06,
      "loss": 0.4965,
      "step": 8723
    },
    {
      "epoch": 0.4003487678399339,
      "grad_norm": 0.4511241316795349,
      "learning_rate": 9.673001363409447e-06,
      "loss": 0.3874,
      "step": 8724
    },
    {
      "epoch": 0.4003946583451884,
      "grad_norm": 0.4596615731716156,
      "learning_rate": 9.672914144507423e-06,
      "loss": 0.4276,
      "step": 8725
    },
    {
      "epoch": 0.40044054885044283,
      "grad_norm": 0.4645096957683563,
      "learning_rate": 9.672826914368515e-06,
      "loss": 0.3884,
      "step": 8726
    },
    {
      "epoch": 0.40048643935569733,
      "grad_norm": 0.49575841426849365,
      "learning_rate": 9.672739672992933e-06,
      "loss": 0.4398,
      "step": 8727
    },
    {
      "epoch": 0.4005323298609518,
      "grad_norm": 0.4681076407432556,
      "learning_rate": 9.672652420380887e-06,
      "loss": 0.3918,
      "step": 8728
    },
    {
      "epoch": 0.4005782203662062,
      "grad_norm": 0.5176234245300293,
      "learning_rate": 9.672565156532585e-06,
      "loss": 0.4654,
      "step": 8729
    },
    {
      "epoch": 0.4006241108714607,
      "grad_norm": 0.4722677767276764,
      "learning_rate": 9.672477881448241e-06,
      "loss": 0.4331,
      "step": 8730
    },
    {
      "epoch": 0.40067000137671516,
      "grad_norm": 0.48463112115859985,
      "learning_rate": 9.672390595128061e-06,
      "loss": 0.445,
      "step": 8731
    },
    {
      "epoch": 0.4007158918819696,
      "grad_norm": 0.46203476190567017,
      "learning_rate": 9.672303297572256e-06,
      "loss": 0.3796,
      "step": 8732
    },
    {
      "epoch": 0.4007617823872241,
      "grad_norm": 0.4735702574253082,
      "learning_rate": 9.672215988781038e-06,
      "loss": 0.4534,
      "step": 8733
    },
    {
      "epoch": 0.40080767289247854,
      "grad_norm": 0.49243053793907166,
      "learning_rate": 9.672128668754613e-06,
      "loss": 0.4332,
      "step": 8734
    },
    {
      "epoch": 0.400853563397733,
      "grad_norm": 0.43404310941696167,
      "learning_rate": 9.672041337493194e-06,
      "loss": 0.372,
      "step": 8735
    },
    {
      "epoch": 0.4008994539029875,
      "grad_norm": 0.7410294413566589,
      "learning_rate": 9.67195399499699e-06,
      "loss": 0.3726,
      "step": 8736
    },
    {
      "epoch": 0.4009453444082419,
      "grad_norm": 0.43253272771835327,
      "learning_rate": 9.671866641266212e-06,
      "loss": 0.3663,
      "step": 8737
    },
    {
      "epoch": 0.4009912349134964,
      "grad_norm": 0.4867757260799408,
      "learning_rate": 9.671779276301067e-06,
      "loss": 0.4927,
      "step": 8738
    },
    {
      "epoch": 0.40103712541875086,
      "grad_norm": 0.46213769912719727,
      "learning_rate": 9.67169190010177e-06,
      "loss": 0.4026,
      "step": 8739
    },
    {
      "epoch": 0.4010830159240053,
      "grad_norm": 0.4710848927497864,
      "learning_rate": 9.671604512668526e-06,
      "loss": 0.4045,
      "step": 8740
    },
    {
      "epoch": 0.4011289064292598,
      "grad_norm": 0.4646565020084381,
      "learning_rate": 9.67151711400155e-06,
      "loss": 0.4326,
      "step": 8741
    },
    {
      "epoch": 0.40117479693451424,
      "grad_norm": 0.4739971160888672,
      "learning_rate": 9.671429704101048e-06,
      "loss": 0.386,
      "step": 8742
    },
    {
      "epoch": 0.4012206874397687,
      "grad_norm": 0.46422308683395386,
      "learning_rate": 9.671342282967233e-06,
      "loss": 0.3734,
      "step": 8743
    },
    {
      "epoch": 0.4012665779450232,
      "grad_norm": 0.4172900915145874,
      "learning_rate": 9.671254850600314e-06,
      "loss": 0.3392,
      "step": 8744
    },
    {
      "epoch": 0.4013124684502776,
      "grad_norm": 0.4752531945705414,
      "learning_rate": 9.6711674070005e-06,
      "loss": 0.376,
      "step": 8745
    },
    {
      "epoch": 0.4013583589555321,
      "grad_norm": 0.4242478311061859,
      "learning_rate": 9.671079952168004e-06,
      "loss": 0.3683,
      "step": 8746
    },
    {
      "epoch": 0.40140424946078657,
      "grad_norm": 0.46339577436447144,
      "learning_rate": 9.670992486103034e-06,
      "loss": 0.4493,
      "step": 8747
    },
    {
      "epoch": 0.401450139966041,
      "grad_norm": 0.45612382888793945,
      "learning_rate": 9.670905008805801e-06,
      "loss": 0.4149,
      "step": 8748
    },
    {
      "epoch": 0.4014960304712955,
      "grad_norm": 0.471419095993042,
      "learning_rate": 9.670817520276517e-06,
      "loss": 0.4376,
      "step": 8749
    },
    {
      "epoch": 0.40154192097654995,
      "grad_norm": 0.45914721488952637,
      "learning_rate": 9.670730020515391e-06,
      "loss": 0.391,
      "step": 8750
    },
    {
      "epoch": 0.4015878114818044,
      "grad_norm": 0.4708364009857178,
      "learning_rate": 9.670642509522632e-06,
      "loss": 0.3835,
      "step": 8751
    },
    {
      "epoch": 0.4016337019870589,
      "grad_norm": 0.5328819155693054,
      "learning_rate": 9.670554987298453e-06,
      "loss": 0.5547,
      "step": 8752
    },
    {
      "epoch": 0.40167959249231333,
      "grad_norm": 0.4802168607711792,
      "learning_rate": 9.670467453843062e-06,
      "loss": 0.4126,
      "step": 8753
    },
    {
      "epoch": 0.40172548299756783,
      "grad_norm": 0.474074125289917,
      "learning_rate": 9.670379909156674e-06,
      "loss": 0.3766,
      "step": 8754
    },
    {
      "epoch": 0.40177137350282227,
      "grad_norm": 0.4740825891494751,
      "learning_rate": 9.670292353239493e-06,
      "loss": 0.4881,
      "step": 8755
    },
    {
      "epoch": 0.4018172640080767,
      "grad_norm": 0.7397067546844482,
      "learning_rate": 9.670204786091734e-06,
      "loss": 0.4164,
      "step": 8756
    },
    {
      "epoch": 0.4018631545133312,
      "grad_norm": 0.43916967511177063,
      "learning_rate": 9.670117207713606e-06,
      "loss": 0.3594,
      "step": 8757
    },
    {
      "epoch": 0.40190904501858565,
      "grad_norm": 0.44764959812164307,
      "learning_rate": 9.67002961810532e-06,
      "loss": 0.4159,
      "step": 8758
    },
    {
      "epoch": 0.4019549355238401,
      "grad_norm": 0.5043580532073975,
      "learning_rate": 9.669942017267087e-06,
      "loss": 0.4939,
      "step": 8759
    },
    {
      "epoch": 0.4020008260290946,
      "grad_norm": 0.5208820700645447,
      "learning_rate": 9.669854405199118e-06,
      "loss": 0.4505,
      "step": 8760
    },
    {
      "epoch": 0.40204671653434904,
      "grad_norm": 0.5180500745773315,
      "learning_rate": 9.669766781901623e-06,
      "loss": 0.5205,
      "step": 8761
    },
    {
      "epoch": 0.40209260703960353,
      "grad_norm": 0.49732547998428345,
      "learning_rate": 9.669679147374813e-06,
      "loss": 0.4714,
      "step": 8762
    },
    {
      "epoch": 0.402138497544858,
      "grad_norm": 0.48268479108810425,
      "learning_rate": 9.669591501618897e-06,
      "loss": 0.4714,
      "step": 8763
    },
    {
      "epoch": 0.4021843880501124,
      "grad_norm": 1.4454619884490967,
      "learning_rate": 9.669503844634087e-06,
      "loss": 0.5187,
      "step": 8764
    },
    {
      "epoch": 0.4022302785553669,
      "grad_norm": 0.4634663164615631,
      "learning_rate": 9.669416176420595e-06,
      "loss": 0.4357,
      "step": 8765
    },
    {
      "epoch": 0.40227616906062136,
      "grad_norm": 0.4607052803039551,
      "learning_rate": 9.669328496978632e-06,
      "loss": 0.3477,
      "step": 8766
    },
    {
      "epoch": 0.4023220595658758,
      "grad_norm": 0.4328441023826599,
      "learning_rate": 9.669240806308406e-06,
      "loss": 0.3492,
      "step": 8767
    },
    {
      "epoch": 0.4023679500711303,
      "grad_norm": 0.42733681201934814,
      "learning_rate": 9.66915310441013e-06,
      "loss": 0.348,
      "step": 8768
    },
    {
      "epoch": 0.40241384057638474,
      "grad_norm": 0.5779761672019958,
      "learning_rate": 9.669065391284013e-06,
      "loss": 0.4813,
      "step": 8769
    },
    {
      "epoch": 0.4024597310816392,
      "grad_norm": 0.48003625869750977,
      "learning_rate": 9.668977666930269e-06,
      "loss": 0.4887,
      "step": 8770
    },
    {
      "epoch": 0.4025056215868937,
      "grad_norm": 0.455434650182724,
      "learning_rate": 9.668889931349107e-06,
      "loss": 0.4662,
      "step": 8771
    },
    {
      "epoch": 0.4025515120921481,
      "grad_norm": 0.47613126039505005,
      "learning_rate": 9.668802184540737e-06,
      "loss": 0.3716,
      "step": 8772
    },
    {
      "epoch": 0.4025974025974026,
      "grad_norm": 0.4700065851211548,
      "learning_rate": 9.668714426505373e-06,
      "loss": 0.371,
      "step": 8773
    },
    {
      "epoch": 0.40264329310265706,
      "grad_norm": 0.47429323196411133,
      "learning_rate": 9.668626657243223e-06,
      "loss": 0.4006,
      "step": 8774
    },
    {
      "epoch": 0.4026891836079115,
      "grad_norm": 0.48311731219291687,
      "learning_rate": 9.668538876754501e-06,
      "loss": 0.4503,
      "step": 8775
    },
    {
      "epoch": 0.402735074113166,
      "grad_norm": 0.49212297797203064,
      "learning_rate": 9.668451085039416e-06,
      "loss": 0.4663,
      "step": 8776
    },
    {
      "epoch": 0.40278096461842045,
      "grad_norm": 0.45360979437828064,
      "learning_rate": 9.668363282098178e-06,
      "loss": 0.4231,
      "step": 8777
    },
    {
      "epoch": 0.4028268551236749,
      "grad_norm": 0.4916783571243286,
      "learning_rate": 9.668275467931e-06,
      "loss": 0.3849,
      "step": 8778
    },
    {
      "epoch": 0.4028727456289294,
      "grad_norm": 0.4797839820384979,
      "learning_rate": 9.668187642538095e-06,
      "loss": 0.3894,
      "step": 8779
    },
    {
      "epoch": 0.4029186361341838,
      "grad_norm": 0.5041285157203674,
      "learning_rate": 9.668099805919671e-06,
      "loss": 0.3729,
      "step": 8780
    },
    {
      "epoch": 0.4029645266394383,
      "grad_norm": 0.4648650288581848,
      "learning_rate": 9.668011958075939e-06,
      "loss": 0.3926,
      "step": 8781
    },
    {
      "epoch": 0.40301041714469277,
      "grad_norm": 0.4691621959209442,
      "learning_rate": 9.667924099007113e-06,
      "loss": 0.4071,
      "step": 8782
    },
    {
      "epoch": 0.4030563076499472,
      "grad_norm": 0.44032666087150574,
      "learning_rate": 9.667836228713401e-06,
      "loss": 0.3243,
      "step": 8783
    },
    {
      "epoch": 0.4031021981552017,
      "grad_norm": 0.47945475578308105,
      "learning_rate": 9.667748347195019e-06,
      "loss": 0.4283,
      "step": 8784
    },
    {
      "epoch": 0.40314808866045615,
      "grad_norm": 0.48606324195861816,
      "learning_rate": 9.667660454452173e-06,
      "loss": 0.4477,
      "step": 8785
    },
    {
      "epoch": 0.4031939791657106,
      "grad_norm": 0.4703650176525116,
      "learning_rate": 9.667572550485076e-06,
      "loss": 0.3507,
      "step": 8786
    },
    {
      "epoch": 0.4032398696709651,
      "grad_norm": 0.44076669216156006,
      "learning_rate": 9.667484635293943e-06,
      "loss": 0.3461,
      "step": 8787
    },
    {
      "epoch": 0.40328576017621953,
      "grad_norm": 0.47779592871665955,
      "learning_rate": 9.66739670887898e-06,
      "loss": 0.4538,
      "step": 8788
    },
    {
      "epoch": 0.40333165068147403,
      "grad_norm": 0.4094685912132263,
      "learning_rate": 9.667308771240402e-06,
      "loss": 0.3259,
      "step": 8789
    },
    {
      "epoch": 0.40337754118672847,
      "grad_norm": 0.5172468423843384,
      "learning_rate": 9.66722082237842e-06,
      "loss": 0.5105,
      "step": 8790
    },
    {
      "epoch": 0.4034234316919829,
      "grad_norm": 0.48521021008491516,
      "learning_rate": 9.667132862293245e-06,
      "loss": 0.4166,
      "step": 8791
    },
    {
      "epoch": 0.4034693221972374,
      "grad_norm": 0.5032215118408203,
      "learning_rate": 9.667044890985088e-06,
      "loss": 0.4611,
      "step": 8792
    },
    {
      "epoch": 0.40351521270249185,
      "grad_norm": 0.4498281478881836,
      "learning_rate": 9.666956908454161e-06,
      "loss": 0.4297,
      "step": 8793
    },
    {
      "epoch": 0.4035611032077463,
      "grad_norm": 0.43463563919067383,
      "learning_rate": 9.666868914700676e-06,
      "loss": 0.3188,
      "step": 8794
    },
    {
      "epoch": 0.4036069937130008,
      "grad_norm": 0.43224555253982544,
      "learning_rate": 9.666780909724843e-06,
      "loss": 0.3612,
      "step": 8795
    },
    {
      "epoch": 0.40365288421825524,
      "grad_norm": 0.4897977411746979,
      "learning_rate": 9.666692893526874e-06,
      "loss": 0.4817,
      "step": 8796
    },
    {
      "epoch": 0.4036987747235097,
      "grad_norm": 0.4552345275878906,
      "learning_rate": 9.666604866106983e-06,
      "loss": 0.382,
      "step": 8797
    },
    {
      "epoch": 0.4037446652287642,
      "grad_norm": 0.5175498723983765,
      "learning_rate": 9.66651682746538e-06,
      "loss": 0.515,
      "step": 8798
    },
    {
      "epoch": 0.4037905557340186,
      "grad_norm": 0.5001868009567261,
      "learning_rate": 9.666428777602278e-06,
      "loss": 0.4702,
      "step": 8799
    },
    {
      "epoch": 0.4038364462392731,
      "grad_norm": 0.4322042763233185,
      "learning_rate": 9.666340716517886e-06,
      "loss": 0.3908,
      "step": 8800
    },
    {
      "epoch": 0.40388233674452756,
      "grad_norm": 0.4481142461299896,
      "learning_rate": 9.666252644212417e-06,
      "loss": 0.4312,
      "step": 8801
    },
    {
      "epoch": 0.403928227249782,
      "grad_norm": 0.4381375014781952,
      "learning_rate": 9.666164560686084e-06,
      "loss": 0.3569,
      "step": 8802
    },
    {
      "epoch": 0.4039741177550365,
      "grad_norm": 0.4626046121120453,
      "learning_rate": 9.666076465939099e-06,
      "loss": 0.4468,
      "step": 8803
    },
    {
      "epoch": 0.40402000826029094,
      "grad_norm": 0.4291236698627472,
      "learning_rate": 9.665988359971671e-06,
      "loss": 0.3875,
      "step": 8804
    },
    {
      "epoch": 0.4040658987655454,
      "grad_norm": 0.4435580372810364,
      "learning_rate": 9.665900242784014e-06,
      "loss": 0.3764,
      "step": 8805
    },
    {
      "epoch": 0.4041117892707999,
      "grad_norm": 0.476551353931427,
      "learning_rate": 9.665812114376339e-06,
      "loss": 0.4777,
      "step": 8806
    },
    {
      "epoch": 0.4041576797760543,
      "grad_norm": 0.44785547256469727,
      "learning_rate": 9.665723974748859e-06,
      "loss": 0.3756,
      "step": 8807
    },
    {
      "epoch": 0.4042035702813088,
      "grad_norm": 0.4439668357372284,
      "learning_rate": 9.665635823901787e-06,
      "loss": 0.3895,
      "step": 8808
    },
    {
      "epoch": 0.40424946078656326,
      "grad_norm": 0.4501616954803467,
      "learning_rate": 9.66554766183533e-06,
      "loss": 0.3971,
      "step": 8809
    },
    {
      "epoch": 0.4042953512918177,
      "grad_norm": 0.46380820870399475,
      "learning_rate": 9.665459488549707e-06,
      "loss": 0.4009,
      "step": 8810
    },
    {
      "epoch": 0.4043412417970722,
      "grad_norm": 0.4688146412372589,
      "learning_rate": 9.665371304045124e-06,
      "loss": 0.3796,
      "step": 8811
    },
    {
      "epoch": 0.40438713230232665,
      "grad_norm": 0.44879040122032166,
      "learning_rate": 9.665283108321796e-06,
      "loss": 0.3576,
      "step": 8812
    },
    {
      "epoch": 0.4044330228075811,
      "grad_norm": 0.4635031223297119,
      "learning_rate": 9.665194901379934e-06,
      "loss": 0.4298,
      "step": 8813
    },
    {
      "epoch": 0.4044789133128356,
      "grad_norm": 0.5085549354553223,
      "learning_rate": 9.665106683219751e-06,
      "loss": 0.4608,
      "step": 8814
    },
    {
      "epoch": 0.40452480381809003,
      "grad_norm": 0.5086290836334229,
      "learning_rate": 9.66501845384146e-06,
      "loss": 0.4306,
      "step": 8815
    },
    {
      "epoch": 0.4045706943233445,
      "grad_norm": 0.46002769470214844,
      "learning_rate": 9.66493021324527e-06,
      "loss": 0.3792,
      "step": 8816
    },
    {
      "epoch": 0.40461658482859897,
      "grad_norm": 0.4461497366428375,
      "learning_rate": 9.664841961431396e-06,
      "loss": 0.3697,
      "step": 8817
    },
    {
      "epoch": 0.4046624753338534,
      "grad_norm": 0.4822692275047302,
      "learning_rate": 9.664753698400048e-06,
      "loss": 0.3977,
      "step": 8818
    },
    {
      "epoch": 0.4047083658391079,
      "grad_norm": 0.4605186879634857,
      "learning_rate": 9.664665424151442e-06,
      "loss": 0.3969,
      "step": 8819
    },
    {
      "epoch": 0.40475425634436235,
      "grad_norm": 0.46697866916656494,
      "learning_rate": 9.664577138685786e-06,
      "loss": 0.3928,
      "step": 8820
    },
    {
      "epoch": 0.4048001468496168,
      "grad_norm": 0.506500244140625,
      "learning_rate": 9.664488842003297e-06,
      "loss": 0.5086,
      "step": 8821
    },
    {
      "epoch": 0.4048460373548713,
      "grad_norm": 0.411395788192749,
      "learning_rate": 9.66440053410418e-06,
      "loss": 0.3056,
      "step": 8822
    },
    {
      "epoch": 0.40489192786012573,
      "grad_norm": 0.47621437907218933,
      "learning_rate": 9.664312214988656e-06,
      "loss": 0.4473,
      "step": 8823
    },
    {
      "epoch": 0.4049378183653802,
      "grad_norm": 0.4494498670101166,
      "learning_rate": 9.664223884656932e-06,
      "loss": 0.3808,
      "step": 8824
    },
    {
      "epoch": 0.4049837088706347,
      "grad_norm": 0.47695329785346985,
      "learning_rate": 9.664135543109222e-06,
      "loss": 0.4421,
      "step": 8825
    },
    {
      "epoch": 0.4050295993758891,
      "grad_norm": 0.5242432355880737,
      "learning_rate": 9.664047190345736e-06,
      "loss": 0.5188,
      "step": 8826
    },
    {
      "epoch": 0.4050754898811436,
      "grad_norm": 0.4797147512435913,
      "learning_rate": 9.663958826366691e-06,
      "loss": 0.4586,
      "step": 8827
    },
    {
      "epoch": 0.40512138038639806,
      "grad_norm": 0.45458757877349854,
      "learning_rate": 9.663870451172295e-06,
      "loss": 0.4057,
      "step": 8828
    },
    {
      "epoch": 0.4051672708916525,
      "grad_norm": 0.45657458901405334,
      "learning_rate": 9.663782064762763e-06,
      "loss": 0.3862,
      "step": 8829
    },
    {
      "epoch": 0.405213161396907,
      "grad_norm": 0.4464982748031616,
      "learning_rate": 9.663693667138307e-06,
      "loss": 0.3651,
      "step": 8830
    },
    {
      "epoch": 0.40525905190216144,
      "grad_norm": 0.4419625699520111,
      "learning_rate": 9.66360525829914e-06,
      "loss": 0.3426,
      "step": 8831
    },
    {
      "epoch": 0.4053049424074159,
      "grad_norm": 0.4469248354434967,
      "learning_rate": 9.663516838245475e-06,
      "loss": 0.3777,
      "step": 8832
    },
    {
      "epoch": 0.4053508329126704,
      "grad_norm": 0.45046088099479675,
      "learning_rate": 9.663428406977521e-06,
      "loss": 0.3817,
      "step": 8833
    },
    {
      "epoch": 0.4053967234179248,
      "grad_norm": 0.42754629254341125,
      "learning_rate": 9.663339964495496e-06,
      "loss": 0.3373,
      "step": 8834
    },
    {
      "epoch": 0.4054426139231793,
      "grad_norm": 0.5050519108772278,
      "learning_rate": 9.663251510799611e-06,
      "loss": 0.4941,
      "step": 8835
    },
    {
      "epoch": 0.40548850442843376,
      "grad_norm": 0.4624300003051758,
      "learning_rate": 9.663163045890076e-06,
      "loss": 0.4599,
      "step": 8836
    },
    {
      "epoch": 0.4055343949336882,
      "grad_norm": 0.4650493562221527,
      "learning_rate": 9.663074569767106e-06,
      "loss": 0.434,
      "step": 8837
    },
    {
      "epoch": 0.4055802854389427,
      "grad_norm": 0.48047754168510437,
      "learning_rate": 9.662986082430914e-06,
      "loss": 0.422,
      "step": 8838
    },
    {
      "epoch": 0.40562617594419714,
      "grad_norm": 0.48945415019989014,
      "learning_rate": 9.662897583881712e-06,
      "loss": 0.4009,
      "step": 8839
    },
    {
      "epoch": 0.4056720664494516,
      "grad_norm": 0.5040608048439026,
      "learning_rate": 9.662809074119713e-06,
      "loss": 0.5,
      "step": 8840
    },
    {
      "epoch": 0.4057179569547061,
      "grad_norm": 0.43869921565055847,
      "learning_rate": 9.66272055314513e-06,
      "loss": 0.3554,
      "step": 8841
    },
    {
      "epoch": 0.4057638474599605,
      "grad_norm": 0.46926331520080566,
      "learning_rate": 9.662632020958174e-06,
      "loss": 0.3831,
      "step": 8842
    },
    {
      "epoch": 0.405809737965215,
      "grad_norm": 0.4961845576763153,
      "learning_rate": 9.662543477559062e-06,
      "loss": 0.4681,
      "step": 8843
    },
    {
      "epoch": 0.40585562847046946,
      "grad_norm": 0.440050333738327,
      "learning_rate": 9.662454922948004e-06,
      "loss": 0.3751,
      "step": 8844
    },
    {
      "epoch": 0.4059015189757239,
      "grad_norm": 0.4777848422527313,
      "learning_rate": 9.66236635712521e-06,
      "loss": 0.4162,
      "step": 8845
    },
    {
      "epoch": 0.4059474094809784,
      "grad_norm": 0.4768102765083313,
      "learning_rate": 9.662277780090901e-06,
      "loss": 0.421,
      "step": 8846
    },
    {
      "epoch": 0.40599329998623285,
      "grad_norm": 0.5259941816329956,
      "learning_rate": 9.662189191845284e-06,
      "loss": 0.531,
      "step": 8847
    },
    {
      "epoch": 0.4060391904914873,
      "grad_norm": 0.4325702488422394,
      "learning_rate": 9.662100592388572e-06,
      "loss": 0.379,
      "step": 8848
    },
    {
      "epoch": 0.4060850809967418,
      "grad_norm": 0.4171409010887146,
      "learning_rate": 9.66201198172098e-06,
      "loss": 0.3519,
      "step": 8849
    },
    {
      "epoch": 0.40613097150199623,
      "grad_norm": 0.49433276057243347,
      "learning_rate": 9.661923359842721e-06,
      "loss": 0.3644,
      "step": 8850
    },
    {
      "epoch": 0.4061768620072507,
      "grad_norm": 0.44720640778541565,
      "learning_rate": 9.661834726754007e-06,
      "loss": 0.3493,
      "step": 8851
    },
    {
      "epoch": 0.40622275251250517,
      "grad_norm": 0.7170031070709229,
      "learning_rate": 9.661746082455053e-06,
      "loss": 0.4727,
      "step": 8852
    },
    {
      "epoch": 0.4062686430177596,
      "grad_norm": 0.4148782789707184,
      "learning_rate": 9.661657426946071e-06,
      "loss": 0.3584,
      "step": 8853
    },
    {
      "epoch": 0.4063145335230141,
      "grad_norm": 0.47836795449256897,
      "learning_rate": 9.661568760227273e-06,
      "loss": 0.4052,
      "step": 8854
    },
    {
      "epoch": 0.40636042402826855,
      "grad_norm": 0.4515821039676666,
      "learning_rate": 9.661480082298872e-06,
      "loss": 0.414,
      "step": 8855
    },
    {
      "epoch": 0.406406314533523,
      "grad_norm": 0.46457260847091675,
      "learning_rate": 9.661391393161085e-06,
      "loss": 0.3409,
      "step": 8856
    },
    {
      "epoch": 0.4064522050387775,
      "grad_norm": 0.4624975323677063,
      "learning_rate": 9.661302692814123e-06,
      "loss": 0.395,
      "step": 8857
    },
    {
      "epoch": 0.40649809554403193,
      "grad_norm": 0.48750466108322144,
      "learning_rate": 9.6612139812582e-06,
      "loss": 0.4434,
      "step": 8858
    },
    {
      "epoch": 0.4065439860492864,
      "grad_norm": 0.5120342969894409,
      "learning_rate": 9.661125258493527e-06,
      "loss": 0.436,
      "step": 8859
    },
    {
      "epoch": 0.4065898765545409,
      "grad_norm": 0.45142799615859985,
      "learning_rate": 9.661036524520317e-06,
      "loss": 0.4019,
      "step": 8860
    },
    {
      "epoch": 0.4066357670597953,
      "grad_norm": 0.5066254138946533,
      "learning_rate": 9.660947779338787e-06,
      "loss": 0.4435,
      "step": 8861
    },
    {
      "epoch": 0.4066816575650498,
      "grad_norm": 0.45592033863067627,
      "learning_rate": 9.660859022949149e-06,
      "loss": 0.4223,
      "step": 8862
    },
    {
      "epoch": 0.40672754807030426,
      "grad_norm": 0.4346959888935089,
      "learning_rate": 9.660770255351616e-06,
      "loss": 0.3779,
      "step": 8863
    },
    {
      "epoch": 0.4067734385755587,
      "grad_norm": 0.45302462577819824,
      "learning_rate": 9.660681476546401e-06,
      "loss": 0.3893,
      "step": 8864
    },
    {
      "epoch": 0.4068193290808132,
      "grad_norm": 0.46152085065841675,
      "learning_rate": 9.660592686533719e-06,
      "loss": 0.3773,
      "step": 8865
    },
    {
      "epoch": 0.40686521958606764,
      "grad_norm": 0.4670110046863556,
      "learning_rate": 9.660503885313781e-06,
      "loss": 0.3767,
      "step": 8866
    },
    {
      "epoch": 0.4069111100913221,
      "grad_norm": 0.47388020157814026,
      "learning_rate": 9.660415072886805e-06,
      "loss": 0.426,
      "step": 8867
    },
    {
      "epoch": 0.4069570005965766,
      "grad_norm": 0.5018221139907837,
      "learning_rate": 9.660326249252998e-06,
      "loss": 0.4972,
      "step": 8868
    },
    {
      "epoch": 0.407002891101831,
      "grad_norm": 0.4418553411960602,
      "learning_rate": 9.66023741441258e-06,
      "loss": 0.3659,
      "step": 8869
    },
    {
      "epoch": 0.4070487816070855,
      "grad_norm": 0.5037994980812073,
      "learning_rate": 9.660148568365759e-06,
      "loss": 0.5028,
      "step": 8870
    },
    {
      "epoch": 0.40709467211233996,
      "grad_norm": 0.45036739110946655,
      "learning_rate": 9.660059711112754e-06,
      "loss": 0.3592,
      "step": 8871
    },
    {
      "epoch": 0.4071405626175944,
      "grad_norm": 0.47340333461761475,
      "learning_rate": 9.659970842653774e-06,
      "loss": 0.3885,
      "step": 8872
    },
    {
      "epoch": 0.4071864531228489,
      "grad_norm": 0.46310076117515564,
      "learning_rate": 9.659881962989037e-06,
      "loss": 0.4387,
      "step": 8873
    },
    {
      "epoch": 0.40723234362810334,
      "grad_norm": 0.4902832508087158,
      "learning_rate": 9.659793072118753e-06,
      "loss": 0.4212,
      "step": 8874
    },
    {
      "epoch": 0.4072782341333578,
      "grad_norm": 0.45954540371894836,
      "learning_rate": 9.659704170043137e-06,
      "loss": 0.3867,
      "step": 8875
    },
    {
      "epoch": 0.4073241246386123,
      "grad_norm": 0.47316038608551025,
      "learning_rate": 9.659615256762404e-06,
      "loss": 0.4236,
      "step": 8876
    },
    {
      "epoch": 0.4073700151438667,
      "grad_norm": 0.5014979243278503,
      "learning_rate": 9.659526332276767e-06,
      "loss": 0.4968,
      "step": 8877
    },
    {
      "epoch": 0.4074159056491212,
      "grad_norm": 0.4483209550380707,
      "learning_rate": 9.65943739658644e-06,
      "loss": 0.3236,
      "step": 8878
    },
    {
      "epoch": 0.40746179615437567,
      "grad_norm": 0.5122126936912537,
      "learning_rate": 9.659348449691637e-06,
      "loss": 0.5046,
      "step": 8879
    },
    {
      "epoch": 0.4075076866596301,
      "grad_norm": 0.4964318871498108,
      "learning_rate": 9.659259491592571e-06,
      "loss": 0.4214,
      "step": 8880
    },
    {
      "epoch": 0.4075535771648846,
      "grad_norm": 0.43395060300827026,
      "learning_rate": 9.659170522289457e-06,
      "loss": 0.369,
      "step": 8881
    },
    {
      "epoch": 0.40759946767013905,
      "grad_norm": 0.48802536725997925,
      "learning_rate": 9.659081541782507e-06,
      "loss": 0.4032,
      "step": 8882
    },
    {
      "epoch": 0.4076453581753935,
      "grad_norm": 0.4703458547592163,
      "learning_rate": 9.658992550071937e-06,
      "loss": 0.4511,
      "step": 8883
    },
    {
      "epoch": 0.407691248680648,
      "grad_norm": 0.45683303475379944,
      "learning_rate": 9.658903547157961e-06,
      "loss": 0.3857,
      "step": 8884
    },
    {
      "epoch": 0.40773713918590243,
      "grad_norm": 0.4692900776863098,
      "learning_rate": 9.65881453304079e-06,
      "loss": 0.4329,
      "step": 8885
    },
    {
      "epoch": 0.4077830296911569,
      "grad_norm": 0.4641232192516327,
      "learning_rate": 9.658725507720644e-06,
      "loss": 0.4199,
      "step": 8886
    },
    {
      "epoch": 0.40782892019641137,
      "grad_norm": 0.46960094571113586,
      "learning_rate": 9.658636471197731e-06,
      "loss": 0.3754,
      "step": 8887
    },
    {
      "epoch": 0.4078748107016658,
      "grad_norm": 0.5141364932060242,
      "learning_rate": 9.658547423472268e-06,
      "loss": 0.4944,
      "step": 8888
    },
    {
      "epoch": 0.4079207012069203,
      "grad_norm": 0.47083625197410583,
      "learning_rate": 9.658458364544469e-06,
      "loss": 0.4758,
      "step": 8889
    },
    {
      "epoch": 0.40796659171217475,
      "grad_norm": 0.4520561397075653,
      "learning_rate": 9.658369294414547e-06,
      "loss": 0.3993,
      "step": 8890
    },
    {
      "epoch": 0.4080124822174292,
      "grad_norm": 0.4793906509876251,
      "learning_rate": 9.658280213082717e-06,
      "loss": 0.3692,
      "step": 8891
    },
    {
      "epoch": 0.4080583727226837,
      "grad_norm": 0.4123370051383972,
      "learning_rate": 9.658191120549196e-06,
      "loss": 0.3098,
      "step": 8892
    },
    {
      "epoch": 0.40810426322793814,
      "grad_norm": 0.44171273708343506,
      "learning_rate": 9.658102016814192e-06,
      "loss": 0.3776,
      "step": 8893
    },
    {
      "epoch": 0.4081501537331926,
      "grad_norm": 0.4890217185020447,
      "learning_rate": 9.658012901877925e-06,
      "loss": 0.3832,
      "step": 8894
    },
    {
      "epoch": 0.4081960442384471,
      "grad_norm": 0.44416534900665283,
      "learning_rate": 9.657923775740607e-06,
      "loss": 0.3525,
      "step": 8895
    },
    {
      "epoch": 0.4082419347437015,
      "grad_norm": 0.4593847692012787,
      "learning_rate": 9.657834638402453e-06,
      "loss": 0.3494,
      "step": 8896
    },
    {
      "epoch": 0.408287825248956,
      "grad_norm": 0.43646371364593506,
      "learning_rate": 9.657745489863675e-06,
      "loss": 0.3672,
      "step": 8897
    },
    {
      "epoch": 0.40833371575421046,
      "grad_norm": 0.47590407729148865,
      "learning_rate": 9.65765633012449e-06,
      "loss": 0.4208,
      "step": 8898
    },
    {
      "epoch": 0.4083796062594649,
      "grad_norm": 0.4629414975643158,
      "learning_rate": 9.657567159185111e-06,
      "loss": 0.3724,
      "step": 8899
    },
    {
      "epoch": 0.4084254967647194,
      "grad_norm": 0.43548890948295593,
      "learning_rate": 9.657477977045753e-06,
      "loss": 0.3367,
      "step": 8900
    },
    {
      "epoch": 0.40847138726997384,
      "grad_norm": 0.4981532692909241,
      "learning_rate": 9.657388783706632e-06,
      "loss": 0.3757,
      "step": 8901
    },
    {
      "epoch": 0.4085172777752283,
      "grad_norm": 0.5075209140777588,
      "learning_rate": 9.657299579167958e-06,
      "loss": 0.4951,
      "step": 8902
    },
    {
      "epoch": 0.4085631682804828,
      "grad_norm": 0.448089063167572,
      "learning_rate": 9.657210363429949e-06,
      "loss": 0.3878,
      "step": 8903
    },
    {
      "epoch": 0.4086090587857372,
      "grad_norm": 0.4841988980770111,
      "learning_rate": 9.657121136492822e-06,
      "loss": 0.44,
      "step": 8904
    },
    {
      "epoch": 0.4086549492909917,
      "grad_norm": 0.49064114689826965,
      "learning_rate": 9.657031898356785e-06,
      "loss": 0.4792,
      "step": 8905
    },
    {
      "epoch": 0.40870083979624616,
      "grad_norm": 0.5222851037979126,
      "learning_rate": 9.656942649022059e-06,
      "loss": 0.5046,
      "step": 8906
    },
    {
      "epoch": 0.4087467303015006,
      "grad_norm": 0.4698149561882019,
      "learning_rate": 9.656853388488855e-06,
      "loss": 0.3668,
      "step": 8907
    },
    {
      "epoch": 0.4087926208067551,
      "grad_norm": 0.45221269130706787,
      "learning_rate": 9.656764116757387e-06,
      "loss": 0.3472,
      "step": 8908
    },
    {
      "epoch": 0.40883851131200954,
      "grad_norm": 0.47569355368614197,
      "learning_rate": 9.65667483382787e-06,
      "loss": 0.4016,
      "step": 8909
    },
    {
      "epoch": 0.408884401817264,
      "grad_norm": 0.44347915053367615,
      "learning_rate": 9.656585539700523e-06,
      "loss": 0.345,
      "step": 8910
    },
    {
      "epoch": 0.4089302923225185,
      "grad_norm": 0.45892956852912903,
      "learning_rate": 9.656496234375554e-06,
      "loss": 0.4293,
      "step": 8911
    },
    {
      "epoch": 0.4089761828277729,
      "grad_norm": 0.5029703974723816,
      "learning_rate": 9.656406917853183e-06,
      "loss": 0.4494,
      "step": 8912
    },
    {
      "epoch": 0.40902207333302737,
      "grad_norm": 0.4619259536266327,
      "learning_rate": 9.656317590133624e-06,
      "loss": 0.3803,
      "step": 8913
    },
    {
      "epoch": 0.40906796383828187,
      "grad_norm": 0.5027748346328735,
      "learning_rate": 9.65622825121709e-06,
      "loss": 0.4413,
      "step": 8914
    },
    {
      "epoch": 0.4091138543435363,
      "grad_norm": 0.43719255924224854,
      "learning_rate": 9.656138901103795e-06,
      "loss": 0.3505,
      "step": 8915
    },
    {
      "epoch": 0.4091597448487908,
      "grad_norm": 0.49089735746383667,
      "learning_rate": 9.656049539793958e-06,
      "loss": 0.4474,
      "step": 8916
    },
    {
      "epoch": 0.40920563535404525,
      "grad_norm": 0.43796178698539734,
      "learning_rate": 9.655960167287788e-06,
      "loss": 0.3338,
      "step": 8917
    },
    {
      "epoch": 0.4092515258592997,
      "grad_norm": 0.48327669501304626,
      "learning_rate": 9.655870783585505e-06,
      "loss": 0.4916,
      "step": 8918
    },
    {
      "epoch": 0.4092974163645542,
      "grad_norm": 0.5065589547157288,
      "learning_rate": 9.655781388687324e-06,
      "loss": 0.5143,
      "step": 8919
    },
    {
      "epoch": 0.40934330686980863,
      "grad_norm": 0.6615613102912903,
      "learning_rate": 9.655691982593455e-06,
      "loss": 0.4162,
      "step": 8920
    },
    {
      "epoch": 0.4093891973750631,
      "grad_norm": 0.4805823564529419,
      "learning_rate": 9.655602565304119e-06,
      "loss": 0.4021,
      "step": 8921
    },
    {
      "epoch": 0.40943508788031757,
      "grad_norm": 0.47744014859199524,
      "learning_rate": 9.655513136819528e-06,
      "loss": 0.4232,
      "step": 8922
    },
    {
      "epoch": 0.409480978385572,
      "grad_norm": 0.4743710160255432,
      "learning_rate": 9.655423697139897e-06,
      "loss": 0.3534,
      "step": 8923
    },
    {
      "epoch": 0.4095268688908265,
      "grad_norm": 0.47602227330207825,
      "learning_rate": 9.65533424626544e-06,
      "loss": 0.3696,
      "step": 8924
    },
    {
      "epoch": 0.40957275939608095,
      "grad_norm": 0.4405398964881897,
      "learning_rate": 9.655244784196373e-06,
      "loss": 0.3676,
      "step": 8925
    },
    {
      "epoch": 0.4096186499013354,
      "grad_norm": 0.4972655773162842,
      "learning_rate": 9.655155310932914e-06,
      "loss": 0.4846,
      "step": 8926
    },
    {
      "epoch": 0.4096645404065899,
      "grad_norm": 0.44851911067962646,
      "learning_rate": 9.655065826475272e-06,
      "loss": 0.4186,
      "step": 8927
    },
    {
      "epoch": 0.40971043091184434,
      "grad_norm": 0.45139557123184204,
      "learning_rate": 9.654976330823669e-06,
      "loss": 0.3977,
      "step": 8928
    },
    {
      "epoch": 0.4097563214170988,
      "grad_norm": 0.5475239753723145,
      "learning_rate": 9.654886823978316e-06,
      "loss": 0.5654,
      "step": 8929
    },
    {
      "epoch": 0.4098022119223533,
      "grad_norm": 0.45348426699638367,
      "learning_rate": 9.65479730593943e-06,
      "loss": 0.4009,
      "step": 8930
    },
    {
      "epoch": 0.4098481024276077,
      "grad_norm": 0.4306449592113495,
      "learning_rate": 9.654707776707224e-06,
      "loss": 0.3363,
      "step": 8931
    },
    {
      "epoch": 0.4098939929328622,
      "grad_norm": 0.42864498496055603,
      "learning_rate": 9.654618236281916e-06,
      "loss": 0.33,
      "step": 8932
    },
    {
      "epoch": 0.40993988343811666,
      "grad_norm": 0.4650707542896271,
      "learning_rate": 9.65452868466372e-06,
      "loss": 0.4095,
      "step": 8933
    },
    {
      "epoch": 0.4099857739433711,
      "grad_norm": 0.499290406703949,
      "learning_rate": 9.654439121852852e-06,
      "loss": 0.4491,
      "step": 8934
    },
    {
      "epoch": 0.4100316644486256,
      "grad_norm": 0.46717971563339233,
      "learning_rate": 9.654349547849527e-06,
      "loss": 0.4191,
      "step": 8935
    },
    {
      "epoch": 0.41007755495388004,
      "grad_norm": 0.47403067350387573,
      "learning_rate": 9.65425996265396e-06,
      "loss": 0.3985,
      "step": 8936
    },
    {
      "epoch": 0.4101234454591345,
      "grad_norm": 0.4315500557422638,
      "learning_rate": 9.654170366266367e-06,
      "loss": 0.3364,
      "step": 8937
    },
    {
      "epoch": 0.410169335964389,
      "grad_norm": 0.49038535356521606,
      "learning_rate": 9.654080758686962e-06,
      "loss": 0.458,
      "step": 8938
    },
    {
      "epoch": 0.4102152264696434,
      "grad_norm": 0.5044361352920532,
      "learning_rate": 9.653991139915961e-06,
      "loss": 0.4644,
      "step": 8939
    },
    {
      "epoch": 0.41026111697489787,
      "grad_norm": 0.505952775478363,
      "learning_rate": 9.653901509953582e-06,
      "loss": 0.4936,
      "step": 8940
    },
    {
      "epoch": 0.41030700748015236,
      "grad_norm": 0.4509478807449341,
      "learning_rate": 9.653811868800038e-06,
      "loss": 0.4125,
      "step": 8941
    },
    {
      "epoch": 0.4103528979854068,
      "grad_norm": 0.45194777846336365,
      "learning_rate": 9.653722216455545e-06,
      "loss": 0.3793,
      "step": 8942
    },
    {
      "epoch": 0.4103987884906613,
      "grad_norm": 0.45763230323791504,
      "learning_rate": 9.65363255292032e-06,
      "loss": 0.388,
      "step": 8943
    },
    {
      "epoch": 0.41044467899591575,
      "grad_norm": 0.5256699919700623,
      "learning_rate": 9.653542878194578e-06,
      "loss": 0.5126,
      "step": 8944
    },
    {
      "epoch": 0.4104905695011702,
      "grad_norm": 0.47213244438171387,
      "learning_rate": 9.653453192278532e-06,
      "loss": 0.4268,
      "step": 8945
    },
    {
      "epoch": 0.4105364600064247,
      "grad_norm": 0.4997270107269287,
      "learning_rate": 9.6533634951724e-06,
      "loss": 0.4214,
      "step": 8946
    },
    {
      "epoch": 0.41058235051167913,
      "grad_norm": 0.483229398727417,
      "learning_rate": 9.653273786876399e-06,
      "loss": 0.4796,
      "step": 8947
    },
    {
      "epoch": 0.41062824101693357,
      "grad_norm": 0.4692580997943878,
      "learning_rate": 9.653184067390741e-06,
      "loss": 0.4213,
      "step": 8948
    },
    {
      "epoch": 0.41067413152218807,
      "grad_norm": 0.4542502164840698,
      "learning_rate": 9.653094336715645e-06,
      "loss": 0.3964,
      "step": 8949
    },
    {
      "epoch": 0.4107200220274425,
      "grad_norm": 0.48807159066200256,
      "learning_rate": 9.653004594851325e-06,
      "loss": 0.4329,
      "step": 8950
    },
    {
      "epoch": 0.410765912532697,
      "grad_norm": 0.4923957586288452,
      "learning_rate": 9.652914841797998e-06,
      "loss": 0.4287,
      "step": 8951
    },
    {
      "epoch": 0.41081180303795145,
      "grad_norm": 0.49464336037635803,
      "learning_rate": 9.65282507755588e-06,
      "loss": 0.4271,
      "step": 8952
    },
    {
      "epoch": 0.4108576935432059,
      "grad_norm": 0.4886745512485504,
      "learning_rate": 9.652735302125186e-06,
      "loss": 0.4384,
      "step": 8953
    },
    {
      "epoch": 0.4109035840484604,
      "grad_norm": 0.449872761964798,
      "learning_rate": 9.652645515506132e-06,
      "loss": 0.389,
      "step": 8954
    },
    {
      "epoch": 0.41094947455371483,
      "grad_norm": 0.4313916563987732,
      "learning_rate": 9.652555717698933e-06,
      "loss": 0.4083,
      "step": 8955
    },
    {
      "epoch": 0.4109953650589693,
      "grad_norm": 0.4735181927680969,
      "learning_rate": 9.652465908703807e-06,
      "loss": 0.3884,
      "step": 8956
    },
    {
      "epoch": 0.4110412555642238,
      "grad_norm": 0.4801724851131439,
      "learning_rate": 9.652376088520967e-06,
      "loss": 0.4158,
      "step": 8957
    },
    {
      "epoch": 0.4110871460694782,
      "grad_norm": 0.44262146949768066,
      "learning_rate": 9.652286257150633e-06,
      "loss": 0.4059,
      "step": 8958
    },
    {
      "epoch": 0.4111330365747327,
      "grad_norm": 0.4646487236022949,
      "learning_rate": 9.652196414593018e-06,
      "loss": 0.3928,
      "step": 8959
    },
    {
      "epoch": 0.41117892707998716,
      "grad_norm": 0.5538948774337769,
      "learning_rate": 9.65210656084834e-06,
      "loss": 0.6343,
      "step": 8960
    },
    {
      "epoch": 0.4112248175852416,
      "grad_norm": 0.4833605885505676,
      "learning_rate": 9.652016695916813e-06,
      "loss": 0.4251,
      "step": 8961
    },
    {
      "epoch": 0.4112707080904961,
      "grad_norm": 0.4419155418872833,
      "learning_rate": 9.651926819798652e-06,
      "loss": 0.4092,
      "step": 8962
    },
    {
      "epoch": 0.41131659859575054,
      "grad_norm": 0.6728895306587219,
      "learning_rate": 9.651836932494079e-06,
      "loss": 0.3942,
      "step": 8963
    },
    {
      "epoch": 0.411362489101005,
      "grad_norm": 0.42597320675849915,
      "learning_rate": 9.651747034003302e-06,
      "loss": 0.3767,
      "step": 8964
    },
    {
      "epoch": 0.4114083796062595,
      "grad_norm": 0.4316471815109253,
      "learning_rate": 9.651657124326546e-06,
      "loss": 0.3967,
      "step": 8965
    },
    {
      "epoch": 0.4114542701115139,
      "grad_norm": 0.4501246511936188,
      "learning_rate": 9.651567203464021e-06,
      "loss": 0.3691,
      "step": 8966
    },
    {
      "epoch": 0.4115001606167684,
      "grad_norm": 0.432313472032547,
      "learning_rate": 9.651477271415944e-06,
      "loss": 0.3508,
      "step": 8967
    },
    {
      "epoch": 0.41154605112202286,
      "grad_norm": 0.4484502077102661,
      "learning_rate": 9.651387328182532e-06,
      "loss": 0.3657,
      "step": 8968
    },
    {
      "epoch": 0.4115919416272773,
      "grad_norm": 0.5222544074058533,
      "learning_rate": 9.651297373764003e-06,
      "loss": 0.4591,
      "step": 8969
    },
    {
      "epoch": 0.4116378321325318,
      "grad_norm": 0.44669589400291443,
      "learning_rate": 9.651207408160572e-06,
      "loss": 0.3488,
      "step": 8970
    },
    {
      "epoch": 0.41168372263778624,
      "grad_norm": 0.4638034701347351,
      "learning_rate": 9.651117431372455e-06,
      "loss": 0.3627,
      "step": 8971
    },
    {
      "epoch": 0.4117296131430407,
      "grad_norm": 0.5077080726623535,
      "learning_rate": 9.651027443399866e-06,
      "loss": 0.4146,
      "step": 8972
    },
    {
      "epoch": 0.4117755036482952,
      "grad_norm": 0.4447862505912781,
      "learning_rate": 9.650937444243027e-06,
      "loss": 0.3864,
      "step": 8973
    },
    {
      "epoch": 0.4118213941535496,
      "grad_norm": 0.4611743688583374,
      "learning_rate": 9.650847433902149e-06,
      "loss": 0.3561,
      "step": 8974
    },
    {
      "epoch": 0.41186728465880407,
      "grad_norm": 0.4529739022254944,
      "learning_rate": 9.650757412377452e-06,
      "loss": 0.3564,
      "step": 8975
    },
    {
      "epoch": 0.41191317516405856,
      "grad_norm": 0.4539554715156555,
      "learning_rate": 9.650667379669151e-06,
      "loss": 0.4295,
      "step": 8976
    },
    {
      "epoch": 0.411959065669313,
      "grad_norm": 0.46761810779571533,
      "learning_rate": 9.650577335777463e-06,
      "loss": 0.3738,
      "step": 8977
    },
    {
      "epoch": 0.4120049561745675,
      "grad_norm": 0.46458902955055237,
      "learning_rate": 9.650487280702603e-06,
      "loss": 0.3613,
      "step": 8978
    },
    {
      "epoch": 0.41205084667982195,
      "grad_norm": 0.4038749039173126,
      "learning_rate": 9.650397214444792e-06,
      "loss": 0.2992,
      "step": 8979
    },
    {
      "epoch": 0.4120967371850764,
      "grad_norm": 0.467714786529541,
      "learning_rate": 9.65030713700424e-06,
      "loss": 0.366,
      "step": 8980
    },
    {
      "epoch": 0.4121426276903309,
      "grad_norm": 0.4536033868789673,
      "learning_rate": 9.65021704838117e-06,
      "loss": 0.3437,
      "step": 8981
    },
    {
      "epoch": 0.41218851819558533,
      "grad_norm": 0.4523737132549286,
      "learning_rate": 9.650126948575793e-06,
      "loss": 0.3393,
      "step": 8982
    },
    {
      "epoch": 0.41223440870083977,
      "grad_norm": 0.46972012519836426,
      "learning_rate": 9.65003683758833e-06,
      "loss": 0.4243,
      "step": 8983
    },
    {
      "epoch": 0.41228029920609427,
      "grad_norm": 0.462490439414978,
      "learning_rate": 9.649946715418994e-06,
      "loss": 0.3413,
      "step": 8984
    },
    {
      "epoch": 0.4123261897113487,
      "grad_norm": 0.4883803725242615,
      "learning_rate": 9.649856582068006e-06,
      "loss": 0.459,
      "step": 8985
    },
    {
      "epoch": 0.4123720802166032,
      "grad_norm": 0.47559982538223267,
      "learning_rate": 9.64976643753558e-06,
      "loss": 0.4373,
      "step": 8986
    },
    {
      "epoch": 0.41241797072185765,
      "grad_norm": 0.4913952946662903,
      "learning_rate": 9.649676281821932e-06,
      "loss": 0.409,
      "step": 8987
    },
    {
      "epoch": 0.4124638612271121,
      "grad_norm": 0.46557748317718506,
      "learning_rate": 9.64958611492728e-06,
      "loss": 0.3908,
      "step": 8988
    },
    {
      "epoch": 0.4125097517323666,
      "grad_norm": 0.45035800337791443,
      "learning_rate": 9.649495936851843e-06,
      "loss": 0.4594,
      "step": 8989
    },
    {
      "epoch": 0.41255564223762103,
      "grad_norm": 0.47361722588539124,
      "learning_rate": 9.649405747595834e-06,
      "loss": 0.3937,
      "step": 8990
    },
    {
      "epoch": 0.4126015327428755,
      "grad_norm": 0.47661158442497253,
      "learning_rate": 9.64931554715947e-06,
      "loss": 0.432,
      "step": 8991
    },
    {
      "epoch": 0.41264742324813,
      "grad_norm": 0.5061835050582886,
      "learning_rate": 9.649225335542974e-06,
      "loss": 0.4547,
      "step": 8992
    },
    {
      "epoch": 0.4126933137533844,
      "grad_norm": 0.4640252888202667,
      "learning_rate": 9.649135112746554e-06,
      "loss": 0.4239,
      "step": 8993
    },
    {
      "epoch": 0.4127392042586389,
      "grad_norm": 0.48689496517181396,
      "learning_rate": 9.649044878770432e-06,
      "loss": 0.414,
      "step": 8994
    },
    {
      "epoch": 0.41278509476389336,
      "grad_norm": 0.48968270421028137,
      "learning_rate": 9.648954633614827e-06,
      "loss": 0.3765,
      "step": 8995
    },
    {
      "epoch": 0.4128309852691478,
      "grad_norm": 0.45853233337402344,
      "learning_rate": 9.648864377279952e-06,
      "loss": 0.3139,
      "step": 8996
    },
    {
      "epoch": 0.4128768757744023,
      "grad_norm": 0.5381628274917603,
      "learning_rate": 9.648774109766024e-06,
      "loss": 0.5339,
      "step": 8997
    },
    {
      "epoch": 0.41292276627965674,
      "grad_norm": 0.4467434287071228,
      "learning_rate": 9.648683831073264e-06,
      "loss": 0.3395,
      "step": 8998
    },
    {
      "epoch": 0.4129686567849112,
      "grad_norm": 0.4439462125301361,
      "learning_rate": 9.648593541201885e-06,
      "loss": 0.3316,
      "step": 8999
    },
    {
      "epoch": 0.4130145472901657,
      "grad_norm": 0.4597897529602051,
      "learning_rate": 9.648503240152105e-06,
      "loss": 0.3633,
      "step": 9000
    },
    {
      "epoch": 0.4130604377954201,
      "grad_norm": 0.45608511567115784,
      "learning_rate": 9.648412927924143e-06,
      "loss": 0.3649,
      "step": 9001
    },
    {
      "epoch": 0.41310632830067456,
      "grad_norm": 0.4434874653816223,
      "learning_rate": 9.648322604518215e-06,
      "loss": 0.3797,
      "step": 9002
    },
    {
      "epoch": 0.41315221880592906,
      "grad_norm": 0.5455503463745117,
      "learning_rate": 9.64823226993454e-06,
      "loss": 0.5055,
      "step": 9003
    },
    {
      "epoch": 0.4131981093111835,
      "grad_norm": 0.4706743657588959,
      "learning_rate": 9.64814192417333e-06,
      "loss": 0.3866,
      "step": 9004
    },
    {
      "epoch": 0.413243999816438,
      "grad_norm": 0.4459630846977234,
      "learning_rate": 9.648051567234806e-06,
      "loss": 0.4126,
      "step": 9005
    },
    {
      "epoch": 0.41328989032169244,
      "grad_norm": 0.4447766840457916,
      "learning_rate": 9.647961199119187e-06,
      "loss": 0.3818,
      "step": 9006
    },
    {
      "epoch": 0.4133357808269469,
      "grad_norm": 0.46116307377815247,
      "learning_rate": 9.647870819826685e-06,
      "loss": 0.3769,
      "step": 9007
    },
    {
      "epoch": 0.4133816713322014,
      "grad_norm": 0.466164231300354,
      "learning_rate": 9.647780429357524e-06,
      "loss": 0.4502,
      "step": 9008
    },
    {
      "epoch": 0.4134275618374558,
      "grad_norm": 0.4497338831424713,
      "learning_rate": 9.647690027711916e-06,
      "loss": 0.3357,
      "step": 9009
    },
    {
      "epoch": 0.41347345234271027,
      "grad_norm": 0.4638221859931946,
      "learning_rate": 9.647599614890081e-06,
      "loss": 0.3774,
      "step": 9010
    },
    {
      "epoch": 0.41351934284796477,
      "grad_norm": 0.4725976586341858,
      "learning_rate": 9.647509190892237e-06,
      "loss": 0.4332,
      "step": 9011
    },
    {
      "epoch": 0.4135652333532192,
      "grad_norm": 0.4279395341873169,
      "learning_rate": 9.647418755718599e-06,
      "loss": 0.372,
      "step": 9012
    },
    {
      "epoch": 0.4136111238584737,
      "grad_norm": 0.4862496256828308,
      "learning_rate": 9.647328309369385e-06,
      "loss": 0.4566,
      "step": 9013
    },
    {
      "epoch": 0.41365701436372815,
      "grad_norm": 0.4215000569820404,
      "learning_rate": 9.647237851844814e-06,
      "loss": 0.3125,
      "step": 9014
    },
    {
      "epoch": 0.4137029048689826,
      "grad_norm": 0.4699840843677521,
      "learning_rate": 9.647147383145103e-06,
      "loss": 0.4562,
      "step": 9015
    },
    {
      "epoch": 0.4137487953742371,
      "grad_norm": 0.4386625289916992,
      "learning_rate": 9.647056903270469e-06,
      "loss": 0.3528,
      "step": 9016
    },
    {
      "epoch": 0.41379468587949153,
      "grad_norm": 0.43476006388664246,
      "learning_rate": 9.646966412221128e-06,
      "loss": 0.3503,
      "step": 9017
    },
    {
      "epoch": 0.413840576384746,
      "grad_norm": 0.4651428163051605,
      "learning_rate": 9.646875909997301e-06,
      "loss": 0.355,
      "step": 9018
    },
    {
      "epoch": 0.41388646689000047,
      "grad_norm": 0.4806995391845703,
      "learning_rate": 9.646785396599206e-06,
      "loss": 0.4185,
      "step": 9019
    },
    {
      "epoch": 0.4139323573952549,
      "grad_norm": 0.4785299599170685,
      "learning_rate": 9.646694872027055e-06,
      "loss": 0.4414,
      "step": 9020
    },
    {
      "epoch": 0.4139782479005094,
      "grad_norm": 0.4907771348953247,
      "learning_rate": 9.646604336281072e-06,
      "loss": 0.3659,
      "step": 9021
    },
    {
      "epoch": 0.41402413840576385,
      "grad_norm": 0.5225874185562134,
      "learning_rate": 9.64651378936147e-06,
      "loss": 0.5623,
      "step": 9022
    },
    {
      "epoch": 0.4140700289110183,
      "grad_norm": 0.4605194628238678,
      "learning_rate": 9.646423231268471e-06,
      "loss": 0.3689,
      "step": 9023
    },
    {
      "epoch": 0.4141159194162728,
      "grad_norm": 0.4852149784564972,
      "learning_rate": 9.646332662002288e-06,
      "loss": 0.5034,
      "step": 9024
    },
    {
      "epoch": 0.41416180992152724,
      "grad_norm": 0.5123778581619263,
      "learning_rate": 9.646242081563144e-06,
      "loss": 0.5514,
      "step": 9025
    },
    {
      "epoch": 0.4142077004267817,
      "grad_norm": 0.5407106280326843,
      "learning_rate": 9.646151489951252e-06,
      "loss": 0.4827,
      "step": 9026
    },
    {
      "epoch": 0.4142535909320362,
      "grad_norm": 0.4818034768104553,
      "learning_rate": 9.646060887166833e-06,
      "loss": 0.432,
      "step": 9027
    },
    {
      "epoch": 0.4142994814372906,
      "grad_norm": 0.48240476846694946,
      "learning_rate": 9.645970273210105e-06,
      "loss": 0.4935,
      "step": 9028
    },
    {
      "epoch": 0.41434537194254506,
      "grad_norm": 0.5163688659667969,
      "learning_rate": 9.645879648081284e-06,
      "loss": 0.4852,
      "step": 9029
    },
    {
      "epoch": 0.41439126244779956,
      "grad_norm": 0.4567596912384033,
      "learning_rate": 9.645789011780588e-06,
      "loss": 0.3627,
      "step": 9030
    },
    {
      "epoch": 0.414437152953054,
      "grad_norm": 0.5446053743362427,
      "learning_rate": 9.645698364308236e-06,
      "loss": 0.497,
      "step": 9031
    },
    {
      "epoch": 0.4144830434583085,
      "grad_norm": 0.46646878123283386,
      "learning_rate": 9.645607705664446e-06,
      "loss": 0.4309,
      "step": 9032
    },
    {
      "epoch": 0.41452893396356294,
      "grad_norm": 0.44527146220207214,
      "learning_rate": 9.645517035849435e-06,
      "loss": 0.3512,
      "step": 9033
    },
    {
      "epoch": 0.4145748244688174,
      "grad_norm": 0.4800756573677063,
      "learning_rate": 9.645426354863424e-06,
      "loss": 0.4695,
      "step": 9034
    },
    {
      "epoch": 0.4146207149740719,
      "grad_norm": 0.4417162239551544,
      "learning_rate": 9.645335662706626e-06,
      "loss": 0.3374,
      "step": 9035
    },
    {
      "epoch": 0.4146666054793263,
      "grad_norm": 0.4372383654117584,
      "learning_rate": 9.645244959379263e-06,
      "loss": 0.3465,
      "step": 9036
    },
    {
      "epoch": 0.41471249598458076,
      "grad_norm": 0.49670517444610596,
      "learning_rate": 9.645154244881551e-06,
      "loss": 0.4593,
      "step": 9037
    },
    {
      "epoch": 0.41475838648983526,
      "grad_norm": 0.5172892212867737,
      "learning_rate": 9.645063519213711e-06,
      "loss": 0.461,
      "step": 9038
    },
    {
      "epoch": 0.4148042769950897,
      "grad_norm": 0.461757093667984,
      "learning_rate": 9.644972782375956e-06,
      "loss": 0.4068,
      "step": 9039
    },
    {
      "epoch": 0.4148501675003442,
      "grad_norm": 0.49969741702079773,
      "learning_rate": 9.644882034368511e-06,
      "loss": 0.468,
      "step": 9040
    },
    {
      "epoch": 0.41489605800559864,
      "grad_norm": 0.4867217540740967,
      "learning_rate": 9.644791275191589e-06,
      "loss": 0.3975,
      "step": 9041
    },
    {
      "epoch": 0.4149419485108531,
      "grad_norm": 0.4218943417072296,
      "learning_rate": 9.64470050484541e-06,
      "loss": 0.3978,
      "step": 9042
    },
    {
      "epoch": 0.4149878390161076,
      "grad_norm": 0.4625298082828522,
      "learning_rate": 9.644609723330193e-06,
      "loss": 0.3579,
      "step": 9043
    },
    {
      "epoch": 0.415033729521362,
      "grad_norm": 0.44674423336982727,
      "learning_rate": 9.644518930646154e-06,
      "loss": 0.3903,
      "step": 9044
    },
    {
      "epoch": 0.41507962002661647,
      "grad_norm": 0.48712071776390076,
      "learning_rate": 9.644428126793514e-06,
      "loss": 0.5092,
      "step": 9045
    },
    {
      "epoch": 0.41512551053187097,
      "grad_norm": 0.4480718970298767,
      "learning_rate": 9.64433731177249e-06,
      "loss": 0.3893,
      "step": 9046
    },
    {
      "epoch": 0.4151714010371254,
      "grad_norm": 0.46100783348083496,
      "learning_rate": 9.644246485583301e-06,
      "loss": 0.4156,
      "step": 9047
    },
    {
      "epoch": 0.4152172915423799,
      "grad_norm": 0.482197105884552,
      "learning_rate": 9.644155648226163e-06,
      "loss": 0.4626,
      "step": 9048
    },
    {
      "epoch": 0.41526318204763435,
      "grad_norm": 0.5053380727767944,
      "learning_rate": 9.644064799701298e-06,
      "loss": 0.4673,
      "step": 9049
    },
    {
      "epoch": 0.4153090725528888,
      "grad_norm": 0.4576834738254547,
      "learning_rate": 9.643973940008924e-06,
      "loss": 0.3836,
      "step": 9050
    },
    {
      "epoch": 0.4153549630581433,
      "grad_norm": 0.425549179315567,
      "learning_rate": 9.643883069149255e-06,
      "loss": 0.3438,
      "step": 9051
    },
    {
      "epoch": 0.41540085356339773,
      "grad_norm": 0.4614979028701782,
      "learning_rate": 9.643792187122516e-06,
      "loss": 0.3793,
      "step": 9052
    },
    {
      "epoch": 0.4154467440686522,
      "grad_norm": 0.48205986618995667,
      "learning_rate": 9.643701293928922e-06,
      "loss": 0.4028,
      "step": 9053
    },
    {
      "epoch": 0.41549263457390667,
      "grad_norm": 0.48817959427833557,
      "learning_rate": 9.643610389568692e-06,
      "loss": 0.4144,
      "step": 9054
    },
    {
      "epoch": 0.4155385250791611,
      "grad_norm": 0.448244571685791,
      "learning_rate": 9.643519474042042e-06,
      "loss": 0.4054,
      "step": 9055
    },
    {
      "epoch": 0.4155844155844156,
      "grad_norm": 0.5007070302963257,
      "learning_rate": 9.643428547349195e-06,
      "loss": 0.4517,
      "step": 9056
    },
    {
      "epoch": 0.41563030608967005,
      "grad_norm": 0.4293174147605896,
      "learning_rate": 9.64333760949037e-06,
      "loss": 0.3692,
      "step": 9057
    },
    {
      "epoch": 0.4156761965949245,
      "grad_norm": 0.4343571066856384,
      "learning_rate": 9.64324666046578e-06,
      "loss": 0.3502,
      "step": 9058
    },
    {
      "epoch": 0.415722087100179,
      "grad_norm": 0.4221457839012146,
      "learning_rate": 9.643155700275647e-06,
      "loss": 0.3291,
      "step": 9059
    },
    {
      "epoch": 0.41576797760543344,
      "grad_norm": 0.46484214067459106,
      "learning_rate": 9.643064728920192e-06,
      "loss": 0.4107,
      "step": 9060
    },
    {
      "epoch": 0.4158138681106879,
      "grad_norm": 0.5128991603851318,
      "learning_rate": 9.642973746399633e-06,
      "loss": 0.5017,
      "step": 9061
    },
    {
      "epoch": 0.4158597586159424,
      "grad_norm": 0.5106416940689087,
      "learning_rate": 9.642882752714187e-06,
      "loss": 0.479,
      "step": 9062
    },
    {
      "epoch": 0.4159056491211968,
      "grad_norm": 0.4736003577709198,
      "learning_rate": 9.64279174786407e-06,
      "loss": 0.3928,
      "step": 9063
    },
    {
      "epoch": 0.41595153962645126,
      "grad_norm": 0.49389421939849854,
      "learning_rate": 9.642700731849507e-06,
      "loss": 0.4692,
      "step": 9064
    },
    {
      "epoch": 0.41599743013170576,
      "grad_norm": 0.4896695017814636,
      "learning_rate": 9.642609704670713e-06,
      "loss": 0.4646,
      "step": 9065
    },
    {
      "epoch": 0.4160433206369602,
      "grad_norm": 0.6674607396125793,
      "learning_rate": 9.64251866632791e-06,
      "loss": 0.3912,
      "step": 9066
    },
    {
      "epoch": 0.4160892111422147,
      "grad_norm": 0.4925616681575775,
      "learning_rate": 9.642427616821313e-06,
      "loss": 0.4665,
      "step": 9067
    },
    {
      "epoch": 0.41613510164746914,
      "grad_norm": 0.46515190601348877,
      "learning_rate": 9.642336556151141e-06,
      "loss": 0.3909,
      "step": 9068
    },
    {
      "epoch": 0.4161809921527236,
      "grad_norm": 0.45093727111816406,
      "learning_rate": 9.642245484317619e-06,
      "loss": 0.3877,
      "step": 9069
    },
    {
      "epoch": 0.4162268826579781,
      "grad_norm": 0.5061994791030884,
      "learning_rate": 9.642154401320958e-06,
      "loss": 0.4438,
      "step": 9070
    },
    {
      "epoch": 0.4162727731632325,
      "grad_norm": 0.49421414732933044,
      "learning_rate": 9.642063307161383e-06,
      "loss": 0.3592,
      "step": 9071
    },
    {
      "epoch": 0.41631866366848697,
      "grad_norm": 0.5362238883972168,
      "learning_rate": 9.641972201839109e-06,
      "loss": 0.4679,
      "step": 9072
    },
    {
      "epoch": 0.41636455417374146,
      "grad_norm": 0.4559922516345978,
      "learning_rate": 9.641881085354358e-06,
      "loss": 0.4111,
      "step": 9073
    },
    {
      "epoch": 0.4164104446789959,
      "grad_norm": 0.4967464804649353,
      "learning_rate": 9.641789957707348e-06,
      "loss": 0.4857,
      "step": 9074
    },
    {
      "epoch": 0.4164563351842504,
      "grad_norm": 0.4805856943130493,
      "learning_rate": 9.641698818898298e-06,
      "loss": 0.4534,
      "step": 9075
    },
    {
      "epoch": 0.41650222568950485,
      "grad_norm": 0.46756449341773987,
      "learning_rate": 9.641607668927426e-06,
      "loss": 0.419,
      "step": 9076
    },
    {
      "epoch": 0.4165481161947593,
      "grad_norm": 0.451858252286911,
      "learning_rate": 9.641516507794955e-06,
      "loss": 0.3637,
      "step": 9077
    },
    {
      "epoch": 0.4165940067000138,
      "grad_norm": 0.4719327390193939,
      "learning_rate": 9.6414253355011e-06,
      "loss": 0.416,
      "step": 9078
    },
    {
      "epoch": 0.41663989720526823,
      "grad_norm": 0.44034528732299805,
      "learning_rate": 9.641334152046082e-06,
      "loss": 0.3617,
      "step": 9079
    },
    {
      "epoch": 0.41668578771052267,
      "grad_norm": 0.5202441811561584,
      "learning_rate": 9.64124295743012e-06,
      "loss": 0.5065,
      "step": 9080
    },
    {
      "epoch": 0.41673167821577717,
      "grad_norm": 0.47937485575675964,
      "learning_rate": 9.641151751653435e-06,
      "loss": 0.4285,
      "step": 9081
    },
    {
      "epoch": 0.4167775687210316,
      "grad_norm": 0.47166207432746887,
      "learning_rate": 9.641060534716245e-06,
      "loss": 0.436,
      "step": 9082
    },
    {
      "epoch": 0.4168234592262861,
      "grad_norm": 0.4816824793815613,
      "learning_rate": 9.640969306618768e-06,
      "loss": 0.4389,
      "step": 9083
    },
    {
      "epoch": 0.41686934973154055,
      "grad_norm": 0.45850440859794617,
      "learning_rate": 9.640878067361224e-06,
      "loss": 0.3283,
      "step": 9084
    },
    {
      "epoch": 0.416915240236795,
      "grad_norm": 0.4667164385318756,
      "learning_rate": 9.640786816943834e-06,
      "loss": 0.4307,
      "step": 9085
    },
    {
      "epoch": 0.4169611307420495,
      "grad_norm": 0.43857523798942566,
      "learning_rate": 9.640695555366817e-06,
      "loss": 0.3858,
      "step": 9086
    },
    {
      "epoch": 0.41700702124730393,
      "grad_norm": 0.4880298674106598,
      "learning_rate": 9.64060428263039e-06,
      "loss": 0.4606,
      "step": 9087
    },
    {
      "epoch": 0.4170529117525584,
      "grad_norm": 0.4753496050834656,
      "learning_rate": 9.640512998734774e-06,
      "loss": 0.4369,
      "step": 9088
    },
    {
      "epoch": 0.4170988022578129,
      "grad_norm": 0.4901658296585083,
      "learning_rate": 9.640421703680191e-06,
      "loss": 0.4804,
      "step": 9089
    },
    {
      "epoch": 0.4171446927630673,
      "grad_norm": 0.44069623947143555,
      "learning_rate": 9.640330397466856e-06,
      "loss": 0.3321,
      "step": 9090
    },
    {
      "epoch": 0.41719058326832176,
      "grad_norm": 0.46980416774749756,
      "learning_rate": 9.640239080094995e-06,
      "loss": 0.4239,
      "step": 9091
    },
    {
      "epoch": 0.41723647377357626,
      "grad_norm": 0.5105448365211487,
      "learning_rate": 9.640147751564819e-06,
      "loss": 0.521,
      "step": 9092
    },
    {
      "epoch": 0.4172823642788307,
      "grad_norm": 0.5242206454277039,
      "learning_rate": 9.640056411876554e-06,
      "loss": 0.4081,
      "step": 9093
    },
    {
      "epoch": 0.4173282547840852,
      "grad_norm": 0.45659470558166504,
      "learning_rate": 9.639965061030418e-06,
      "loss": 0.4055,
      "step": 9094
    },
    {
      "epoch": 0.41737414528933964,
      "grad_norm": 0.4579678773880005,
      "learning_rate": 9.63987369902663e-06,
      "loss": 0.3584,
      "step": 9095
    },
    {
      "epoch": 0.4174200357945941,
      "grad_norm": 0.4552818238735199,
      "learning_rate": 9.63978232586541e-06,
      "loss": 0.3537,
      "step": 9096
    },
    {
      "epoch": 0.4174659262998486,
      "grad_norm": 0.47954949736595154,
      "learning_rate": 9.63969094154698e-06,
      "loss": 0.4565,
      "step": 9097
    },
    {
      "epoch": 0.417511816805103,
      "grad_norm": 0.47457948327064514,
      "learning_rate": 9.639599546071554e-06,
      "loss": 0.3638,
      "step": 9098
    },
    {
      "epoch": 0.41755770731035746,
      "grad_norm": 0.49206864833831787,
      "learning_rate": 9.639508139439358e-06,
      "loss": 0.5135,
      "step": 9099
    },
    {
      "epoch": 0.41760359781561196,
      "grad_norm": 0.5259010791778564,
      "learning_rate": 9.639416721650609e-06,
      "loss": 0.5508,
      "step": 9100
    },
    {
      "epoch": 0.4176494883208664,
      "grad_norm": 0.6092624664306641,
      "learning_rate": 9.639325292705526e-06,
      "loss": 0.6375,
      "step": 9101
    },
    {
      "epoch": 0.4176953788261209,
      "grad_norm": 0.4506482779979706,
      "learning_rate": 9.639233852604331e-06,
      "loss": 0.368,
      "step": 9102
    },
    {
      "epoch": 0.41774126933137534,
      "grad_norm": 0.4400605261325836,
      "learning_rate": 9.639142401347242e-06,
      "loss": 0.3434,
      "step": 9103
    },
    {
      "epoch": 0.4177871598366298,
      "grad_norm": 0.5243903994560242,
      "learning_rate": 9.63905093893448e-06,
      "loss": 0.4622,
      "step": 9104
    },
    {
      "epoch": 0.4178330503418843,
      "grad_norm": 0.47471651434898376,
      "learning_rate": 9.638959465366264e-06,
      "loss": 0.4276,
      "step": 9105
    },
    {
      "epoch": 0.4178789408471387,
      "grad_norm": 0.4387849271297455,
      "learning_rate": 9.638867980642815e-06,
      "loss": 0.3598,
      "step": 9106
    },
    {
      "epoch": 0.41792483135239317,
      "grad_norm": 0.4695335030555725,
      "learning_rate": 9.638776484764351e-06,
      "loss": 0.3938,
      "step": 9107
    },
    {
      "epoch": 0.41797072185764766,
      "grad_norm": 0.45224127173423767,
      "learning_rate": 9.638684977731096e-06,
      "loss": 0.3562,
      "step": 9108
    },
    {
      "epoch": 0.4180166123629021,
      "grad_norm": 0.5024235844612122,
      "learning_rate": 9.638593459543265e-06,
      "loss": 0.4783,
      "step": 9109
    },
    {
      "epoch": 0.4180625028681566,
      "grad_norm": 0.45329779386520386,
      "learning_rate": 9.638501930201083e-06,
      "loss": 0.4151,
      "step": 9110
    },
    {
      "epoch": 0.41810839337341105,
      "grad_norm": 0.4989628493785858,
      "learning_rate": 9.638410389704768e-06,
      "loss": 0.462,
      "step": 9111
    },
    {
      "epoch": 0.4181542838786655,
      "grad_norm": 0.42240384221076965,
      "learning_rate": 9.638318838054538e-06,
      "loss": 0.2958,
      "step": 9112
    },
    {
      "epoch": 0.41820017438392,
      "grad_norm": 0.46871256828308105,
      "learning_rate": 9.638227275250618e-06,
      "loss": 0.4324,
      "step": 9113
    },
    {
      "epoch": 0.41824606488917443,
      "grad_norm": 0.4595646262168884,
      "learning_rate": 9.638135701293223e-06,
      "loss": 0.4537,
      "step": 9114
    },
    {
      "epoch": 0.41829195539442887,
      "grad_norm": 0.4329849183559418,
      "learning_rate": 9.638044116182577e-06,
      "loss": 0.3948,
      "step": 9115
    },
    {
      "epoch": 0.41833784589968337,
      "grad_norm": 0.5041369199752808,
      "learning_rate": 9.637952519918898e-06,
      "loss": 0.4718,
      "step": 9116
    },
    {
      "epoch": 0.4183837364049378,
      "grad_norm": 0.4626680016517639,
      "learning_rate": 9.637860912502406e-06,
      "loss": 0.4633,
      "step": 9117
    },
    {
      "epoch": 0.41842962691019225,
      "grad_norm": 0.45026925206184387,
      "learning_rate": 9.637769293933324e-06,
      "loss": 0.3674,
      "step": 9118
    },
    {
      "epoch": 0.41847551741544675,
      "grad_norm": 0.5466172695159912,
      "learning_rate": 9.637677664211868e-06,
      "loss": 0.4562,
      "step": 9119
    },
    {
      "epoch": 0.4185214079207012,
      "grad_norm": 0.46066147089004517,
      "learning_rate": 9.637586023338264e-06,
      "loss": 0.4645,
      "step": 9120
    },
    {
      "epoch": 0.4185672984259557,
      "grad_norm": 0.45379945635795593,
      "learning_rate": 9.637494371312726e-06,
      "loss": 0.4386,
      "step": 9121
    },
    {
      "epoch": 0.41861318893121013,
      "grad_norm": 0.4638664126396179,
      "learning_rate": 9.63740270813548e-06,
      "loss": 0.3932,
      "step": 9122
    },
    {
      "epoch": 0.4186590794364646,
      "grad_norm": 0.4590039551258087,
      "learning_rate": 9.637311033806744e-06,
      "loss": 0.3518,
      "step": 9123
    },
    {
      "epoch": 0.4187049699417191,
      "grad_norm": 0.49647775292396545,
      "learning_rate": 9.637219348326738e-06,
      "loss": 0.577,
      "step": 9124
    },
    {
      "epoch": 0.4187508604469735,
      "grad_norm": 0.5052891969680786,
      "learning_rate": 9.637127651695684e-06,
      "loss": 0.4724,
      "step": 9125
    },
    {
      "epoch": 0.41879675095222796,
      "grad_norm": 0.49633723497390747,
      "learning_rate": 9.6370359439138e-06,
      "loss": 0.4776,
      "step": 9126
    },
    {
      "epoch": 0.41884264145748246,
      "grad_norm": 0.5114006996154785,
      "learning_rate": 9.636944224981308e-06,
      "loss": 0.5153,
      "step": 9127
    },
    {
      "epoch": 0.4188885319627369,
      "grad_norm": 0.42441409826278687,
      "learning_rate": 9.63685249489843e-06,
      "loss": 0.3575,
      "step": 9128
    },
    {
      "epoch": 0.4189344224679914,
      "grad_norm": 0.4883800745010376,
      "learning_rate": 9.636760753665384e-06,
      "loss": 0.4491,
      "step": 9129
    },
    {
      "epoch": 0.41898031297324584,
      "grad_norm": 0.4798250198364258,
      "learning_rate": 9.636669001282393e-06,
      "loss": 0.3934,
      "step": 9130
    },
    {
      "epoch": 0.4190262034785003,
      "grad_norm": 0.4608350694179535,
      "learning_rate": 9.636577237749676e-06,
      "loss": 0.3904,
      "step": 9131
    },
    {
      "epoch": 0.4190720939837548,
      "grad_norm": 0.5018261671066284,
      "learning_rate": 9.636485463067453e-06,
      "loss": 0.5107,
      "step": 9132
    },
    {
      "epoch": 0.4191179844890092,
      "grad_norm": 0.4269445538520813,
      "learning_rate": 9.636393677235946e-06,
      "loss": 0.3553,
      "step": 9133
    },
    {
      "epoch": 0.41916387499426366,
      "grad_norm": 0.48879799246788025,
      "learning_rate": 9.636301880255376e-06,
      "loss": 0.4282,
      "step": 9134
    },
    {
      "epoch": 0.41920976549951816,
      "grad_norm": 0.4451883137226105,
      "learning_rate": 9.636210072125962e-06,
      "loss": 0.4078,
      "step": 9135
    },
    {
      "epoch": 0.4192556560047726,
      "grad_norm": 0.4177812933921814,
      "learning_rate": 9.636118252847929e-06,
      "loss": 0.3452,
      "step": 9136
    },
    {
      "epoch": 0.4193015465100271,
      "grad_norm": 0.480985552072525,
      "learning_rate": 9.636026422421492e-06,
      "loss": 0.371,
      "step": 9137
    },
    {
      "epoch": 0.41934743701528154,
      "grad_norm": 0.46856895089149475,
      "learning_rate": 9.635934580846875e-06,
      "loss": 0.4205,
      "step": 9138
    },
    {
      "epoch": 0.419393327520536,
      "grad_norm": 0.44325849413871765,
      "learning_rate": 9.6358427281243e-06,
      "loss": 0.3488,
      "step": 9139
    },
    {
      "epoch": 0.4194392180257905,
      "grad_norm": 0.4497213363647461,
      "learning_rate": 9.635750864253986e-06,
      "loss": 0.3653,
      "step": 9140
    },
    {
      "epoch": 0.4194851085310449,
      "grad_norm": 0.46376290917396545,
      "learning_rate": 9.635658989236154e-06,
      "loss": 0.4238,
      "step": 9141
    },
    {
      "epoch": 0.41953099903629937,
      "grad_norm": 0.4417935609817505,
      "learning_rate": 9.635567103071024e-06,
      "loss": 0.3823,
      "step": 9142
    },
    {
      "epoch": 0.41957688954155387,
      "grad_norm": 0.47996753454208374,
      "learning_rate": 9.63547520575882e-06,
      "loss": 0.4199,
      "step": 9143
    },
    {
      "epoch": 0.4196227800468083,
      "grad_norm": 0.47316473722457886,
      "learning_rate": 9.63538329729976e-06,
      "loss": 0.4307,
      "step": 9144
    },
    {
      "epoch": 0.4196686705520628,
      "grad_norm": 0.46204081177711487,
      "learning_rate": 9.635291377694066e-06,
      "loss": 0.3833,
      "step": 9145
    },
    {
      "epoch": 0.41971456105731725,
      "grad_norm": 0.4463976323604584,
      "learning_rate": 9.63519944694196e-06,
      "loss": 0.3962,
      "step": 9146
    },
    {
      "epoch": 0.4197604515625717,
      "grad_norm": 0.4870917499065399,
      "learning_rate": 9.635107505043662e-06,
      "loss": 0.4372,
      "step": 9147
    },
    {
      "epoch": 0.4198063420678262,
      "grad_norm": 0.5178536772727966,
      "learning_rate": 9.63501555199939e-06,
      "loss": 0.5154,
      "step": 9148
    },
    {
      "epoch": 0.41985223257308063,
      "grad_norm": 0.44570448994636536,
      "learning_rate": 9.634923587809374e-06,
      "loss": 0.3513,
      "step": 9149
    },
    {
      "epoch": 0.4198981230783351,
      "grad_norm": 0.47698888182640076,
      "learning_rate": 9.634831612473828e-06,
      "loss": 0.3736,
      "step": 9150
    },
    {
      "epoch": 0.41994401358358957,
      "grad_norm": 0.4657828211784363,
      "learning_rate": 9.634739625992972e-06,
      "loss": 0.4458,
      "step": 9151
    },
    {
      "epoch": 0.419989904088844,
      "grad_norm": 0.4672933518886566,
      "learning_rate": 9.634647628367033e-06,
      "loss": 0.4115,
      "step": 9152
    },
    {
      "epoch": 0.42003579459409845,
      "grad_norm": 0.4705240726470947,
      "learning_rate": 9.634555619596228e-06,
      "loss": 0.4102,
      "step": 9153
    },
    {
      "epoch": 0.42008168509935295,
      "grad_norm": 0.4999903440475464,
      "learning_rate": 9.63446359968078e-06,
      "loss": 0.4769,
      "step": 9154
    },
    {
      "epoch": 0.4201275756046074,
      "grad_norm": 0.521174967288971,
      "learning_rate": 9.634371568620908e-06,
      "loss": 0.539,
      "step": 9155
    },
    {
      "epoch": 0.4201734661098619,
      "grad_norm": 0.47343260049819946,
      "learning_rate": 9.634279526416836e-06,
      "loss": 0.4257,
      "step": 9156
    },
    {
      "epoch": 0.42021935661511634,
      "grad_norm": 0.4652255177497864,
      "learning_rate": 9.634187473068786e-06,
      "loss": 0.3951,
      "step": 9157
    },
    {
      "epoch": 0.4202652471203708,
      "grad_norm": 0.4572575092315674,
      "learning_rate": 9.634095408576975e-06,
      "loss": 0.3682,
      "step": 9158
    },
    {
      "epoch": 0.4203111376256253,
      "grad_norm": 0.4927283525466919,
      "learning_rate": 9.634003332941629e-06,
      "loss": 0.4164,
      "step": 9159
    },
    {
      "epoch": 0.4203570281308797,
      "grad_norm": 0.499971866607666,
      "learning_rate": 9.633911246162966e-06,
      "loss": 0.3992,
      "step": 9160
    },
    {
      "epoch": 0.42040291863613416,
      "grad_norm": 0.4839976727962494,
      "learning_rate": 9.63381914824121e-06,
      "loss": 0.4543,
      "step": 9161
    },
    {
      "epoch": 0.42044880914138866,
      "grad_norm": 0.45867499709129333,
      "learning_rate": 9.633727039176583e-06,
      "loss": 0.3848,
      "step": 9162
    },
    {
      "epoch": 0.4204946996466431,
      "grad_norm": 0.42866039276123047,
      "learning_rate": 9.633634918969304e-06,
      "loss": 0.3316,
      "step": 9163
    },
    {
      "epoch": 0.4205405901518976,
      "grad_norm": 0.44626402854919434,
      "learning_rate": 9.633542787619593e-06,
      "loss": 0.3421,
      "step": 9164
    },
    {
      "epoch": 0.42058648065715204,
      "grad_norm": 0.45159152150154114,
      "learning_rate": 9.633450645127676e-06,
      "loss": 0.4199,
      "step": 9165
    },
    {
      "epoch": 0.4206323711624065,
      "grad_norm": 0.4681631922721863,
      "learning_rate": 9.633358491493771e-06,
      "loss": 0.3974,
      "step": 9166
    },
    {
      "epoch": 0.420678261667661,
      "grad_norm": 0.4973943829536438,
      "learning_rate": 9.633266326718104e-06,
      "loss": 0.434,
      "step": 9167
    },
    {
      "epoch": 0.4207241521729154,
      "grad_norm": 0.5066848993301392,
      "learning_rate": 9.633174150800893e-06,
      "loss": 0.4319,
      "step": 9168
    },
    {
      "epoch": 0.42077004267816986,
      "grad_norm": 0.4840337634086609,
      "learning_rate": 9.633081963742358e-06,
      "loss": 0.3641,
      "step": 9169
    },
    {
      "epoch": 0.42081593318342436,
      "grad_norm": 0.47228360176086426,
      "learning_rate": 9.632989765542725e-06,
      "loss": 0.4501,
      "step": 9170
    },
    {
      "epoch": 0.4208618236886788,
      "grad_norm": 0.46816036105155945,
      "learning_rate": 9.632897556202214e-06,
      "loss": 0.4044,
      "step": 9171
    },
    {
      "epoch": 0.4209077141939333,
      "grad_norm": 0.4344419836997986,
      "learning_rate": 9.632805335721047e-06,
      "loss": 0.3297,
      "step": 9172
    },
    {
      "epoch": 0.42095360469918774,
      "grad_norm": 0.469348281621933,
      "learning_rate": 9.632713104099442e-06,
      "loss": 0.3914,
      "step": 9173
    },
    {
      "epoch": 0.4209994952044422,
      "grad_norm": 0.47385624051094055,
      "learning_rate": 9.632620861337627e-06,
      "loss": 0.3633,
      "step": 9174
    },
    {
      "epoch": 0.4210453857096967,
      "grad_norm": 0.4502538740634918,
      "learning_rate": 9.632528607435822e-06,
      "loss": 0.397,
      "step": 9175
    },
    {
      "epoch": 0.4210912762149511,
      "grad_norm": 0.4432981610298157,
      "learning_rate": 9.632436342394246e-06,
      "loss": 0.3933,
      "step": 9176
    },
    {
      "epoch": 0.42113716672020557,
      "grad_norm": 0.44716376066207886,
      "learning_rate": 9.632344066213122e-06,
      "loss": 0.3806,
      "step": 9177
    },
    {
      "epoch": 0.42118305722546007,
      "grad_norm": 0.46810510754585266,
      "learning_rate": 9.632251778892674e-06,
      "loss": 0.4201,
      "step": 9178
    },
    {
      "epoch": 0.4212289477307145,
      "grad_norm": 0.42006680369377136,
      "learning_rate": 9.632159480433122e-06,
      "loss": 0.3457,
      "step": 9179
    },
    {
      "epoch": 0.42127483823596895,
      "grad_norm": 0.4443822503089905,
      "learning_rate": 9.632067170834687e-06,
      "loss": 0.3501,
      "step": 9180
    },
    {
      "epoch": 0.42132072874122345,
      "grad_norm": 0.47762802243232727,
      "learning_rate": 9.631974850097595e-06,
      "loss": 0.3165,
      "step": 9181
    },
    {
      "epoch": 0.4213666192464779,
      "grad_norm": 0.4917104244232178,
      "learning_rate": 9.631882518222064e-06,
      "loss": 0.4505,
      "step": 9182
    },
    {
      "epoch": 0.4214125097517324,
      "grad_norm": 0.4593959152698517,
      "learning_rate": 9.631790175208317e-06,
      "loss": 0.3495,
      "step": 9183
    },
    {
      "epoch": 0.42145840025698683,
      "grad_norm": 0.5756860971450806,
      "learning_rate": 9.631697821056578e-06,
      "loss": 0.3994,
      "step": 9184
    },
    {
      "epoch": 0.4215042907622413,
      "grad_norm": 0.44248494505882263,
      "learning_rate": 9.631605455767066e-06,
      "loss": 0.3622,
      "step": 9185
    },
    {
      "epoch": 0.42155018126749577,
      "grad_norm": 0.4316681921482086,
      "learning_rate": 9.631513079340006e-06,
      "loss": 0.3223,
      "step": 9186
    },
    {
      "epoch": 0.4215960717727502,
      "grad_norm": 0.5344674587249756,
      "learning_rate": 9.631420691775617e-06,
      "loss": 0.4768,
      "step": 9187
    },
    {
      "epoch": 0.42164196227800466,
      "grad_norm": 0.45224109292030334,
      "learning_rate": 9.631328293074123e-06,
      "loss": 0.3535,
      "step": 9188
    },
    {
      "epoch": 0.42168785278325915,
      "grad_norm": 0.4437558948993683,
      "learning_rate": 9.631235883235748e-06,
      "loss": 0.3745,
      "step": 9189
    },
    {
      "epoch": 0.4217337432885136,
      "grad_norm": 0.49107006192207336,
      "learning_rate": 9.63114346226071e-06,
      "loss": 0.4793,
      "step": 9190
    },
    {
      "epoch": 0.4217796337937681,
      "grad_norm": 0.4538659453392029,
      "learning_rate": 9.631051030149236e-06,
      "loss": 0.3959,
      "step": 9191
    },
    {
      "epoch": 0.42182552429902254,
      "grad_norm": 0.47982704639434814,
      "learning_rate": 9.630958586901545e-06,
      "loss": 0.4302,
      "step": 9192
    },
    {
      "epoch": 0.421871414804277,
      "grad_norm": 0.4702761173248291,
      "learning_rate": 9.63086613251786e-06,
      "loss": 0.4455,
      "step": 9193
    },
    {
      "epoch": 0.4219173053095315,
      "grad_norm": 0.43576353788375854,
      "learning_rate": 9.630773666998404e-06,
      "loss": 0.3259,
      "step": 9194
    },
    {
      "epoch": 0.4219631958147859,
      "grad_norm": 0.4703406095504761,
      "learning_rate": 9.630681190343398e-06,
      "loss": 0.4793,
      "step": 9195
    },
    {
      "epoch": 0.42200908632004036,
      "grad_norm": 0.4469132423400879,
      "learning_rate": 9.630588702553065e-06,
      "loss": 0.4254,
      "step": 9196
    },
    {
      "epoch": 0.42205497682529486,
      "grad_norm": 0.48336511850357056,
      "learning_rate": 9.630496203627628e-06,
      "loss": 0.4489,
      "step": 9197
    },
    {
      "epoch": 0.4221008673305493,
      "grad_norm": 0.46181246638298035,
      "learning_rate": 9.63040369356731e-06,
      "loss": 0.4561,
      "step": 9198
    },
    {
      "epoch": 0.4221467578358038,
      "grad_norm": 0.48599866032600403,
      "learning_rate": 9.630311172372331e-06,
      "loss": 0.4654,
      "step": 9199
    },
    {
      "epoch": 0.42219264834105824,
      "grad_norm": 0.5396682620048523,
      "learning_rate": 9.630218640042917e-06,
      "loss": 0.3289,
      "step": 9200
    },
    {
      "epoch": 0.4222385388463127,
      "grad_norm": 0.4663090109825134,
      "learning_rate": 9.630126096579287e-06,
      "loss": 0.4402,
      "step": 9201
    },
    {
      "epoch": 0.4222844293515672,
      "grad_norm": 0.4351460039615631,
      "learning_rate": 9.630033541981664e-06,
      "loss": 0.3594,
      "step": 9202
    },
    {
      "epoch": 0.4223303198568216,
      "grad_norm": 0.41514748334884644,
      "learning_rate": 9.629940976250273e-06,
      "loss": 0.3182,
      "step": 9203
    },
    {
      "epoch": 0.42237621036207607,
      "grad_norm": 0.4399113655090332,
      "learning_rate": 9.629848399385335e-06,
      "loss": 0.3948,
      "step": 9204
    },
    {
      "epoch": 0.42242210086733056,
      "grad_norm": 0.5144763588905334,
      "learning_rate": 9.629755811387074e-06,
      "loss": 0.459,
      "step": 9205
    },
    {
      "epoch": 0.422467991372585,
      "grad_norm": 0.45055699348449707,
      "learning_rate": 9.62966321225571e-06,
      "loss": 0.3578,
      "step": 9206
    },
    {
      "epoch": 0.42251388187783945,
      "grad_norm": 0.46767714619636536,
      "learning_rate": 9.629570601991467e-06,
      "loss": 0.4006,
      "step": 9207
    },
    {
      "epoch": 0.42255977238309395,
      "grad_norm": 0.4586738348007202,
      "learning_rate": 9.62947798059457e-06,
      "loss": 0.3616,
      "step": 9208
    },
    {
      "epoch": 0.4226056628883484,
      "grad_norm": 0.4546884298324585,
      "learning_rate": 9.629385348065237e-06,
      "loss": 0.4148,
      "step": 9209
    },
    {
      "epoch": 0.4226515533936029,
      "grad_norm": 0.46073809266090393,
      "learning_rate": 9.629292704403694e-06,
      "loss": 0.3829,
      "step": 9210
    },
    {
      "epoch": 0.42269744389885733,
      "grad_norm": 0.4920588433742523,
      "learning_rate": 9.629200049610163e-06,
      "loss": 0.3888,
      "step": 9211
    },
    {
      "epoch": 0.42274333440411177,
      "grad_norm": 0.44922736287117004,
      "learning_rate": 9.629107383684868e-06,
      "loss": 0.4143,
      "step": 9212
    },
    {
      "epoch": 0.42278922490936627,
      "grad_norm": 0.46897852420806885,
      "learning_rate": 9.629014706628031e-06,
      "loss": 0.3605,
      "step": 9213
    },
    {
      "epoch": 0.4228351154146207,
      "grad_norm": 0.4589211940765381,
      "learning_rate": 9.628922018439872e-06,
      "loss": 0.3584,
      "step": 9214
    },
    {
      "epoch": 0.42288100591987515,
      "grad_norm": 0.5066733360290527,
      "learning_rate": 9.628829319120618e-06,
      "loss": 0.5265,
      "step": 9215
    },
    {
      "epoch": 0.42292689642512965,
      "grad_norm": 0.4305788278579712,
      "learning_rate": 9.62873660867049e-06,
      "loss": 0.3751,
      "step": 9216
    },
    {
      "epoch": 0.4229727869303841,
      "grad_norm": 0.4506034851074219,
      "learning_rate": 9.628643887089712e-06,
      "loss": 0.3819,
      "step": 9217
    },
    {
      "epoch": 0.4230186774356386,
      "grad_norm": 0.5699796080589294,
      "learning_rate": 9.628551154378508e-06,
      "loss": 0.4476,
      "step": 9218
    },
    {
      "epoch": 0.42306456794089303,
      "grad_norm": 0.47635558247566223,
      "learning_rate": 9.628458410537096e-06,
      "loss": 0.3653,
      "step": 9219
    },
    {
      "epoch": 0.4231104584461475,
      "grad_norm": 0.48368704319000244,
      "learning_rate": 9.628365655565703e-06,
      "loss": 0.4219,
      "step": 9220
    },
    {
      "epoch": 0.423156348951402,
      "grad_norm": 0.4511227607727051,
      "learning_rate": 9.628272889464554e-06,
      "loss": 0.4061,
      "step": 9221
    },
    {
      "epoch": 0.4232022394566564,
      "grad_norm": 0.42040878534317017,
      "learning_rate": 9.628180112233867e-06,
      "loss": 0.3503,
      "step": 9222
    },
    {
      "epoch": 0.42324812996191086,
      "grad_norm": 0.43288448452949524,
      "learning_rate": 9.628087323873867e-06,
      "loss": 0.3426,
      "step": 9223
    },
    {
      "epoch": 0.42329402046716536,
      "grad_norm": 0.4799714982509613,
      "learning_rate": 9.627994524384779e-06,
      "loss": 0.4161,
      "step": 9224
    },
    {
      "epoch": 0.4233399109724198,
      "grad_norm": 0.4229944944381714,
      "learning_rate": 9.627901713766825e-06,
      "loss": 0.3422,
      "step": 9225
    },
    {
      "epoch": 0.4233858014776743,
      "grad_norm": 0.43074628710746765,
      "learning_rate": 9.627808892020227e-06,
      "loss": 0.3511,
      "step": 9226
    },
    {
      "epoch": 0.42343169198292874,
      "grad_norm": 0.44593873620033264,
      "learning_rate": 9.627716059145209e-06,
      "loss": 0.411,
      "step": 9227
    },
    {
      "epoch": 0.4234775824881832,
      "grad_norm": 0.4194604456424713,
      "learning_rate": 9.627623215141995e-06,
      "loss": 0.3466,
      "step": 9228
    },
    {
      "epoch": 0.4235234729934377,
      "grad_norm": 0.4713863730430603,
      "learning_rate": 9.627530360010808e-06,
      "loss": 0.4614,
      "step": 9229
    },
    {
      "epoch": 0.4235693634986921,
      "grad_norm": 0.5067123770713806,
      "learning_rate": 9.627437493751871e-06,
      "loss": 0.5093,
      "step": 9230
    },
    {
      "epoch": 0.42361525400394656,
      "grad_norm": 0.4608362317085266,
      "learning_rate": 9.627344616365407e-06,
      "loss": 0.4213,
      "step": 9231
    },
    {
      "epoch": 0.42366114450920106,
      "grad_norm": 0.4694509208202362,
      "learning_rate": 9.627251727851638e-06,
      "loss": 0.369,
      "step": 9232
    },
    {
      "epoch": 0.4237070350144555,
      "grad_norm": 0.5229730606079102,
      "learning_rate": 9.627158828210793e-06,
      "loss": 0.4363,
      "step": 9233
    },
    {
      "epoch": 0.42375292551970994,
      "grad_norm": 0.4506397843360901,
      "learning_rate": 9.627065917443088e-06,
      "loss": 0.3655,
      "step": 9234
    },
    {
      "epoch": 0.42379881602496444,
      "grad_norm": 0.4641627073287964,
      "learning_rate": 9.626972995548751e-06,
      "loss": 0.4375,
      "step": 9235
    },
    {
      "epoch": 0.4238447065302189,
      "grad_norm": 0.44912189245224,
      "learning_rate": 9.626880062528004e-06,
      "loss": 0.4098,
      "step": 9236
    },
    {
      "epoch": 0.4238905970354734,
      "grad_norm": 0.4723019301891327,
      "learning_rate": 9.62678711838107e-06,
      "loss": 0.3611,
      "step": 9237
    },
    {
      "epoch": 0.4239364875407278,
      "grad_norm": 0.40141505002975464,
      "learning_rate": 9.626694163108175e-06,
      "loss": 0.2889,
      "step": 9238
    },
    {
      "epoch": 0.42398237804598227,
      "grad_norm": 0.5123217105865479,
      "learning_rate": 9.626601196709539e-06,
      "loss": 0.4302,
      "step": 9239
    },
    {
      "epoch": 0.42402826855123676,
      "grad_norm": 0.4635584354400635,
      "learning_rate": 9.626508219185388e-06,
      "loss": 0.4033,
      "step": 9240
    },
    {
      "epoch": 0.4240741590564912,
      "grad_norm": 0.43918484449386597,
      "learning_rate": 9.626415230535947e-06,
      "loss": 0.3181,
      "step": 9241
    },
    {
      "epoch": 0.42412004956174565,
      "grad_norm": 0.46661925315856934,
      "learning_rate": 9.626322230761434e-06,
      "loss": 0.4369,
      "step": 9242
    },
    {
      "epoch": 0.42416594006700015,
      "grad_norm": 0.4900226891040802,
      "learning_rate": 9.626229219862078e-06,
      "loss": 0.4706,
      "step": 9243
    },
    {
      "epoch": 0.4242118305722546,
      "grad_norm": 0.5329292416572571,
      "learning_rate": 9.6261361978381e-06,
      "loss": 0.4179,
      "step": 9244
    },
    {
      "epoch": 0.4242577210775091,
      "grad_norm": 0.4939621388912201,
      "learning_rate": 9.626043164689726e-06,
      "loss": 0.3436,
      "step": 9245
    },
    {
      "epoch": 0.42430361158276353,
      "grad_norm": 0.4607374966144562,
      "learning_rate": 9.625950120417177e-06,
      "loss": 0.4247,
      "step": 9246
    },
    {
      "epoch": 0.42434950208801797,
      "grad_norm": 0.46404802799224854,
      "learning_rate": 9.625857065020678e-06,
      "loss": 0.3985,
      "step": 9247
    },
    {
      "epoch": 0.42439539259327247,
      "grad_norm": 0.4763891398906708,
      "learning_rate": 9.625763998500452e-06,
      "loss": 0.4214,
      "step": 9248
    },
    {
      "epoch": 0.4244412830985269,
      "grad_norm": 0.4844805598258972,
      "learning_rate": 9.625670920856725e-06,
      "loss": 0.3892,
      "step": 9249
    },
    {
      "epoch": 0.42448717360378135,
      "grad_norm": 0.4685736298561096,
      "learning_rate": 9.62557783208972e-06,
      "loss": 0.4103,
      "step": 9250
    },
    {
      "epoch": 0.42453306410903585,
      "grad_norm": 0.8583676218986511,
      "learning_rate": 9.625484732199656e-06,
      "loss": 0.5084,
      "step": 9251
    },
    {
      "epoch": 0.4245789546142903,
      "grad_norm": 0.4483937621116638,
      "learning_rate": 9.625391621186767e-06,
      "loss": 0.3639,
      "step": 9252
    },
    {
      "epoch": 0.4246248451195448,
      "grad_norm": 0.4661558270454407,
      "learning_rate": 9.625298499051267e-06,
      "loss": 0.401,
      "step": 9253
    },
    {
      "epoch": 0.42467073562479923,
      "grad_norm": 0.4873422384262085,
      "learning_rate": 9.625205365793385e-06,
      "loss": 0.4293,
      "step": 9254
    },
    {
      "epoch": 0.4247166261300537,
      "grad_norm": 0.5073668956756592,
      "learning_rate": 9.625112221413343e-06,
      "loss": 0.3772,
      "step": 9255
    },
    {
      "epoch": 0.4247625166353082,
      "grad_norm": 0.47060680389404297,
      "learning_rate": 9.625019065911368e-06,
      "loss": 0.392,
      "step": 9256
    },
    {
      "epoch": 0.4248084071405626,
      "grad_norm": 0.5233715772628784,
      "learning_rate": 9.624925899287679e-06,
      "loss": 0.4704,
      "step": 9257
    },
    {
      "epoch": 0.42485429764581706,
      "grad_norm": 0.4383015036582947,
      "learning_rate": 9.624832721542504e-06,
      "loss": 0.3227,
      "step": 9258
    },
    {
      "epoch": 0.42490018815107156,
      "grad_norm": 0.4938104450702667,
      "learning_rate": 9.624739532676065e-06,
      "loss": 0.4378,
      "step": 9259
    },
    {
      "epoch": 0.424946078656326,
      "grad_norm": 0.4555651843547821,
      "learning_rate": 9.62464633268859e-06,
      "loss": 0.4051,
      "step": 9260
    },
    {
      "epoch": 0.4249919691615805,
      "grad_norm": 0.4442608654499054,
      "learning_rate": 9.624553121580298e-06,
      "loss": 0.3711,
      "step": 9261
    },
    {
      "epoch": 0.42503785966683494,
      "grad_norm": 0.47801512479782104,
      "learning_rate": 9.624459899351416e-06,
      "loss": 0.4486,
      "step": 9262
    },
    {
      "epoch": 0.4250837501720894,
      "grad_norm": 0.4152389466762543,
      "learning_rate": 9.624366666002168e-06,
      "loss": 0.3147,
      "step": 9263
    },
    {
      "epoch": 0.4251296406773439,
      "grad_norm": 0.44813215732574463,
      "learning_rate": 9.624273421532776e-06,
      "loss": 0.3317,
      "step": 9264
    },
    {
      "epoch": 0.4251755311825983,
      "grad_norm": 0.5241290926933289,
      "learning_rate": 9.624180165943467e-06,
      "loss": 0.5044,
      "step": 9265
    },
    {
      "epoch": 0.42522142168785276,
      "grad_norm": 0.8696032762527466,
      "learning_rate": 9.624086899234465e-06,
      "loss": 0.5278,
      "step": 9266
    },
    {
      "epoch": 0.42526731219310726,
      "grad_norm": 0.47406888008117676,
      "learning_rate": 9.62399362140599e-06,
      "loss": 0.4308,
      "step": 9267
    },
    {
      "epoch": 0.4253132026983617,
      "grad_norm": 0.49069467186927795,
      "learning_rate": 9.623900332458273e-06,
      "loss": 0.4554,
      "step": 9268
    },
    {
      "epoch": 0.42535909320361615,
      "grad_norm": 0.47076311707496643,
      "learning_rate": 9.623807032391533e-06,
      "loss": 0.4368,
      "step": 9269
    },
    {
      "epoch": 0.42540498370887064,
      "grad_norm": 0.4652973413467407,
      "learning_rate": 9.623713721205998e-06,
      "loss": 0.4413,
      "step": 9270
    },
    {
      "epoch": 0.4254508742141251,
      "grad_norm": 0.45212292671203613,
      "learning_rate": 9.62362039890189e-06,
      "loss": 0.3793,
      "step": 9271
    },
    {
      "epoch": 0.4254967647193796,
      "grad_norm": 0.5371978878974915,
      "learning_rate": 9.623527065479434e-06,
      "loss": 0.5501,
      "step": 9272
    },
    {
      "epoch": 0.425542655224634,
      "grad_norm": 0.5082307457923889,
      "learning_rate": 9.623433720938854e-06,
      "loss": 0.3749,
      "step": 9273
    },
    {
      "epoch": 0.42558854572988847,
      "grad_norm": 0.5133089423179626,
      "learning_rate": 9.623340365280378e-06,
      "loss": 0.3987,
      "step": 9274
    },
    {
      "epoch": 0.42563443623514297,
      "grad_norm": 0.47863346338272095,
      "learning_rate": 9.623246998504226e-06,
      "loss": 0.4279,
      "step": 9275
    },
    {
      "epoch": 0.4256803267403974,
      "grad_norm": 0.4552866220474243,
      "learning_rate": 9.623153620610623e-06,
      "loss": 0.3884,
      "step": 9276
    },
    {
      "epoch": 0.42572621724565185,
      "grad_norm": 0.45817872881889343,
      "learning_rate": 9.623060231599796e-06,
      "loss": 0.3832,
      "step": 9277
    },
    {
      "epoch": 0.42577210775090635,
      "grad_norm": 0.45008647441864014,
      "learning_rate": 9.622966831471967e-06,
      "loss": 0.3713,
      "step": 9278
    },
    {
      "epoch": 0.4258179982561608,
      "grad_norm": 0.4441007375717163,
      "learning_rate": 9.622873420227363e-06,
      "loss": 0.3495,
      "step": 9279
    },
    {
      "epoch": 0.4258638887614153,
      "grad_norm": 0.46405982971191406,
      "learning_rate": 9.622779997866208e-06,
      "loss": 0.3886,
      "step": 9280
    },
    {
      "epoch": 0.42590977926666973,
      "grad_norm": 0.49822187423706055,
      "learning_rate": 9.622686564388724e-06,
      "loss": 0.3776,
      "step": 9281
    },
    {
      "epoch": 0.4259556697719242,
      "grad_norm": 0.4638362526893616,
      "learning_rate": 9.622593119795139e-06,
      "loss": 0.4053,
      "step": 9282
    },
    {
      "epoch": 0.42600156027717867,
      "grad_norm": 0.4604857861995697,
      "learning_rate": 9.622499664085676e-06,
      "loss": 0.4014,
      "step": 9283
    },
    {
      "epoch": 0.4260474507824331,
      "grad_norm": 0.5440638065338135,
      "learning_rate": 9.62240619726056e-06,
      "loss": 0.5325,
      "step": 9284
    },
    {
      "epoch": 0.42609334128768755,
      "grad_norm": 0.46557363867759705,
      "learning_rate": 9.622312719320017e-06,
      "loss": 0.387,
      "step": 9285
    },
    {
      "epoch": 0.42613923179294205,
      "grad_norm": 0.4589885175228119,
      "learning_rate": 9.62221923026427e-06,
      "loss": 0.4085,
      "step": 9286
    },
    {
      "epoch": 0.4261851222981965,
      "grad_norm": 0.6126189827919006,
      "learning_rate": 9.622125730093546e-06,
      "loss": 0.5568,
      "step": 9287
    },
    {
      "epoch": 0.426231012803451,
      "grad_norm": 0.4489326477050781,
      "learning_rate": 9.622032218808066e-06,
      "loss": 0.3142,
      "step": 9288
    },
    {
      "epoch": 0.42627690330870543,
      "grad_norm": 0.4269963502883911,
      "learning_rate": 9.621938696408059e-06,
      "loss": 0.3284,
      "step": 9289
    },
    {
      "epoch": 0.4263227938139599,
      "grad_norm": 0.45649832487106323,
      "learning_rate": 9.621845162893748e-06,
      "loss": 0.4132,
      "step": 9290
    },
    {
      "epoch": 0.4263686843192144,
      "grad_norm": 0.46363863348960876,
      "learning_rate": 9.621751618265356e-06,
      "loss": 0.4587,
      "step": 9291
    },
    {
      "epoch": 0.4264145748244688,
      "grad_norm": 0.471809446811676,
      "learning_rate": 9.621658062523113e-06,
      "loss": 0.4442,
      "step": 9292
    },
    {
      "epoch": 0.42646046532972326,
      "grad_norm": 0.4662055969238281,
      "learning_rate": 9.62156449566724e-06,
      "loss": 0.4064,
      "step": 9293
    },
    {
      "epoch": 0.42650635583497776,
      "grad_norm": 0.5097374320030212,
      "learning_rate": 9.621470917697962e-06,
      "loss": 0.4923,
      "step": 9294
    },
    {
      "epoch": 0.4265522463402322,
      "grad_norm": 0.44528651237487793,
      "learning_rate": 9.621377328615505e-06,
      "loss": 0.3651,
      "step": 9295
    },
    {
      "epoch": 0.42659813684548664,
      "grad_norm": 0.4545913636684418,
      "learning_rate": 9.621283728420095e-06,
      "loss": 0.397,
      "step": 9296
    },
    {
      "epoch": 0.42664402735074114,
      "grad_norm": 0.4809154272079468,
      "learning_rate": 9.621190117111956e-06,
      "loss": 0.4121,
      "step": 9297
    },
    {
      "epoch": 0.4266899178559956,
      "grad_norm": 0.465979665517807,
      "learning_rate": 9.621096494691313e-06,
      "loss": 0.3999,
      "step": 9298
    },
    {
      "epoch": 0.4267358083612501,
      "grad_norm": 0.4242551326751709,
      "learning_rate": 9.621002861158393e-06,
      "loss": 0.3317,
      "step": 9299
    },
    {
      "epoch": 0.4267816988665045,
      "grad_norm": 0.41084274649620056,
      "learning_rate": 9.620909216513418e-06,
      "loss": 0.2963,
      "step": 9300
    },
    {
      "epoch": 0.42682758937175896,
      "grad_norm": 0.44750308990478516,
      "learning_rate": 9.620815560756615e-06,
      "loss": 0.341,
      "step": 9301
    },
    {
      "epoch": 0.42687347987701346,
      "grad_norm": 0.45712295174598694,
      "learning_rate": 9.620721893888208e-06,
      "loss": 0.4317,
      "step": 9302
    },
    {
      "epoch": 0.4269193703822679,
      "grad_norm": 0.47288960218429565,
      "learning_rate": 9.620628215908424e-06,
      "loss": 0.4755,
      "step": 9303
    },
    {
      "epoch": 0.42696526088752235,
      "grad_norm": 0.5106072425842285,
      "learning_rate": 9.620534526817488e-06,
      "loss": 0.4701,
      "step": 9304
    },
    {
      "epoch": 0.42701115139277684,
      "grad_norm": 0.4548209607601166,
      "learning_rate": 9.620440826615624e-06,
      "loss": 0.3885,
      "step": 9305
    },
    {
      "epoch": 0.4270570418980313,
      "grad_norm": 0.4438328444957733,
      "learning_rate": 9.62034711530306e-06,
      "loss": 0.3867,
      "step": 9306
    },
    {
      "epoch": 0.4271029324032858,
      "grad_norm": 0.46092694997787476,
      "learning_rate": 9.620253392880017e-06,
      "loss": 0.4293,
      "step": 9307
    },
    {
      "epoch": 0.4271488229085402,
      "grad_norm": 0.42802494764328003,
      "learning_rate": 9.620159659346722e-06,
      "loss": 0.362,
      "step": 9308
    },
    {
      "epoch": 0.42719471341379467,
      "grad_norm": 0.4766315221786499,
      "learning_rate": 9.620065914703403e-06,
      "loss": 0.4794,
      "step": 9309
    },
    {
      "epoch": 0.42724060391904917,
      "grad_norm": 0.4250241816043854,
      "learning_rate": 9.619972158950284e-06,
      "loss": 0.3648,
      "step": 9310
    },
    {
      "epoch": 0.4272864944243036,
      "grad_norm": 0.4395454525947571,
      "learning_rate": 9.61987839208759e-06,
      "loss": 0.3661,
      "step": 9311
    },
    {
      "epoch": 0.42733238492955805,
      "grad_norm": 0.44705507159233093,
      "learning_rate": 9.619784614115545e-06,
      "loss": 0.3744,
      "step": 9312
    },
    {
      "epoch": 0.42737827543481255,
      "grad_norm": 0.44999632239341736,
      "learning_rate": 9.619690825034376e-06,
      "loss": 0.3945,
      "step": 9313
    },
    {
      "epoch": 0.427424165940067,
      "grad_norm": 0.46289411187171936,
      "learning_rate": 9.61959702484431e-06,
      "loss": 0.4242,
      "step": 9314
    },
    {
      "epoch": 0.4274700564453215,
      "grad_norm": 0.5039185285568237,
      "learning_rate": 9.619503213545571e-06,
      "loss": 0.4496,
      "step": 9315
    },
    {
      "epoch": 0.42751594695057593,
      "grad_norm": 0.46557557582855225,
      "learning_rate": 9.619409391138384e-06,
      "loss": 0.4462,
      "step": 9316
    },
    {
      "epoch": 0.4275618374558304,
      "grad_norm": 0.44910427927970886,
      "learning_rate": 9.619315557622977e-06,
      "loss": 0.4513,
      "step": 9317
    },
    {
      "epoch": 0.42760772796108487,
      "grad_norm": 0.4621914327144623,
      "learning_rate": 9.619221712999572e-06,
      "loss": 0.4038,
      "step": 9318
    },
    {
      "epoch": 0.4276536184663393,
      "grad_norm": 0.43913882970809937,
      "learning_rate": 9.619127857268398e-06,
      "loss": 0.3546,
      "step": 9319
    },
    {
      "epoch": 0.42769950897159376,
      "grad_norm": 0.4547036588191986,
      "learning_rate": 9.619033990429678e-06,
      "loss": 0.4286,
      "step": 9320
    },
    {
      "epoch": 0.42774539947684825,
      "grad_norm": 0.4780140817165375,
      "learning_rate": 9.618940112483639e-06,
      "loss": 0.382,
      "step": 9321
    },
    {
      "epoch": 0.4277912899821027,
      "grad_norm": 0.4566645622253418,
      "learning_rate": 9.618846223430508e-06,
      "loss": 0.365,
      "step": 9322
    },
    {
      "epoch": 0.42783718048735714,
      "grad_norm": 0.4559045135974884,
      "learning_rate": 9.61875232327051e-06,
      "loss": 0.4314,
      "step": 9323
    },
    {
      "epoch": 0.42788307099261164,
      "grad_norm": 0.4224149286746979,
      "learning_rate": 9.61865841200387e-06,
      "loss": 0.3271,
      "step": 9324
    },
    {
      "epoch": 0.4279289614978661,
      "grad_norm": 0.48331961035728455,
      "learning_rate": 9.618564489630813e-06,
      "loss": 0.4475,
      "step": 9325
    },
    {
      "epoch": 0.4279748520031206,
      "grad_norm": 0.4809322655200958,
      "learning_rate": 9.618470556151566e-06,
      "loss": 0.3672,
      "step": 9326
    },
    {
      "epoch": 0.428020742508375,
      "grad_norm": 0.4940967559814453,
      "learning_rate": 9.618376611566355e-06,
      "loss": 0.4648,
      "step": 9327
    },
    {
      "epoch": 0.42806663301362946,
      "grad_norm": 0.4519594609737396,
      "learning_rate": 9.618282655875409e-06,
      "loss": 0.3651,
      "step": 9328
    },
    {
      "epoch": 0.42811252351888396,
      "grad_norm": 0.4870782494544983,
      "learning_rate": 9.618188689078947e-06,
      "loss": 0.4205,
      "step": 9329
    },
    {
      "epoch": 0.4281584140241384,
      "grad_norm": 0.4744412899017334,
      "learning_rate": 9.618094711177201e-06,
      "loss": 0.4473,
      "step": 9330
    },
    {
      "epoch": 0.42820430452939284,
      "grad_norm": 0.43273308873176575,
      "learning_rate": 9.618000722170392e-06,
      "loss": 0.3325,
      "step": 9331
    },
    {
      "epoch": 0.42825019503464734,
      "grad_norm": 0.538695752620697,
      "learning_rate": 9.617906722058751e-06,
      "loss": 0.4849,
      "step": 9332
    },
    {
      "epoch": 0.4282960855399018,
      "grad_norm": 0.5047197937965393,
      "learning_rate": 9.617812710842502e-06,
      "loss": 0.4393,
      "step": 9333
    },
    {
      "epoch": 0.4283419760451563,
      "grad_norm": 0.45876622200012207,
      "learning_rate": 9.617718688521872e-06,
      "loss": 0.4118,
      "step": 9334
    },
    {
      "epoch": 0.4283878665504107,
      "grad_norm": 0.5221706628799438,
      "learning_rate": 9.617624655097084e-06,
      "loss": 0.4447,
      "step": 9335
    },
    {
      "epoch": 0.42843375705566517,
      "grad_norm": 0.4644910991191864,
      "learning_rate": 9.617530610568365e-06,
      "loss": 0.4468,
      "step": 9336
    },
    {
      "epoch": 0.42847964756091966,
      "grad_norm": 0.47872182726860046,
      "learning_rate": 9.617436554935943e-06,
      "loss": 0.4131,
      "step": 9337
    },
    {
      "epoch": 0.4285255380661741,
      "grad_norm": 0.43918725848197937,
      "learning_rate": 9.617342488200045e-06,
      "loss": 0.3405,
      "step": 9338
    },
    {
      "epoch": 0.42857142857142855,
      "grad_norm": 0.4436374008655548,
      "learning_rate": 9.617248410360895e-06,
      "loss": 0.365,
      "step": 9339
    },
    {
      "epoch": 0.42861731907668305,
      "grad_norm": 0.45412611961364746,
      "learning_rate": 9.617154321418718e-06,
      "loss": 0.4159,
      "step": 9340
    },
    {
      "epoch": 0.4286632095819375,
      "grad_norm": 0.4963136315345764,
      "learning_rate": 9.617060221373743e-06,
      "loss": 0.4397,
      "step": 9341
    },
    {
      "epoch": 0.428709100087192,
      "grad_norm": 0.43968525528907776,
      "learning_rate": 9.616966110226196e-06,
      "loss": 0.3423,
      "step": 9342
    },
    {
      "epoch": 0.4287549905924464,
      "grad_norm": 0.41697973012924194,
      "learning_rate": 9.616871987976302e-06,
      "loss": 0.3143,
      "step": 9343
    },
    {
      "epoch": 0.42880088109770087,
      "grad_norm": 0.4994734525680542,
      "learning_rate": 9.616777854624288e-06,
      "loss": 0.5338,
      "step": 9344
    },
    {
      "epoch": 0.42884677160295537,
      "grad_norm": 0.436317503452301,
      "learning_rate": 9.616683710170383e-06,
      "loss": 0.3557,
      "step": 9345
    },
    {
      "epoch": 0.4288926621082098,
      "grad_norm": 0.4690050482749939,
      "learning_rate": 9.616589554614807e-06,
      "loss": 0.4209,
      "step": 9346
    },
    {
      "epoch": 0.42893855261346425,
      "grad_norm": 0.5130156874656677,
      "learning_rate": 9.616495387957791e-06,
      "loss": 0.4243,
      "step": 9347
    },
    {
      "epoch": 0.42898444311871875,
      "grad_norm": 0.4453177750110626,
      "learning_rate": 9.616401210199562e-06,
      "loss": 0.3717,
      "step": 9348
    },
    {
      "epoch": 0.4290303336239732,
      "grad_norm": 0.4651983380317688,
      "learning_rate": 9.616307021340345e-06,
      "loss": 0.4732,
      "step": 9349
    },
    {
      "epoch": 0.4290762241292277,
      "grad_norm": 0.4337430000305176,
      "learning_rate": 9.616212821380367e-06,
      "loss": 0.4131,
      "step": 9350
    },
    {
      "epoch": 0.42912211463448213,
      "grad_norm": 0.4609726667404175,
      "learning_rate": 9.616118610319851e-06,
      "loss": 0.3984,
      "step": 9351
    },
    {
      "epoch": 0.4291680051397366,
      "grad_norm": 0.47717565298080444,
      "learning_rate": 9.61602438815903e-06,
      "loss": 0.4544,
      "step": 9352
    },
    {
      "epoch": 0.4292138956449911,
      "grad_norm": 0.4874224364757538,
      "learning_rate": 9.615930154898126e-06,
      "loss": 0.4817,
      "step": 9353
    },
    {
      "epoch": 0.4292597861502455,
      "grad_norm": 0.5027182102203369,
      "learning_rate": 9.615835910537365e-06,
      "loss": 0.4354,
      "step": 9354
    },
    {
      "epoch": 0.42930567665549996,
      "grad_norm": 0.4737550616264343,
      "learning_rate": 9.615741655076978e-06,
      "loss": 0.3733,
      "step": 9355
    },
    {
      "epoch": 0.42935156716075445,
      "grad_norm": 0.4546607434749603,
      "learning_rate": 9.615647388517188e-06,
      "loss": 0.4453,
      "step": 9356
    },
    {
      "epoch": 0.4293974576660089,
      "grad_norm": 0.46273118257522583,
      "learning_rate": 9.615553110858222e-06,
      "loss": 0.3763,
      "step": 9357
    },
    {
      "epoch": 0.42944334817126334,
      "grad_norm": 0.43113821744918823,
      "learning_rate": 9.61545882210031e-06,
      "loss": 0.3462,
      "step": 9358
    },
    {
      "epoch": 0.42948923867651784,
      "grad_norm": 0.4763573408126831,
      "learning_rate": 9.615364522243674e-06,
      "loss": 0.4053,
      "step": 9359
    },
    {
      "epoch": 0.4295351291817723,
      "grad_norm": 0.48686087131500244,
      "learning_rate": 9.615270211288544e-06,
      "loss": 0.4547,
      "step": 9360
    },
    {
      "epoch": 0.4295810196870268,
      "grad_norm": 0.41065531969070435,
      "learning_rate": 9.615175889235145e-06,
      "loss": 0.2928,
      "step": 9361
    },
    {
      "epoch": 0.4296269101922812,
      "grad_norm": 0.437703400850296,
      "learning_rate": 9.615081556083705e-06,
      "loss": 0.3571,
      "step": 9362
    },
    {
      "epoch": 0.42967280069753566,
      "grad_norm": 0.4673965275287628,
      "learning_rate": 9.61498721183445e-06,
      "loss": 0.3705,
      "step": 9363
    },
    {
      "epoch": 0.42971869120279016,
      "grad_norm": 0.43763649463653564,
      "learning_rate": 9.614892856487608e-06,
      "loss": 0.3493,
      "step": 9364
    },
    {
      "epoch": 0.4297645817080446,
      "grad_norm": 0.514441728591919,
      "learning_rate": 9.614798490043405e-06,
      "loss": 0.502,
      "step": 9365
    },
    {
      "epoch": 0.42981047221329904,
      "grad_norm": 0.45282578468322754,
      "learning_rate": 9.614704112502066e-06,
      "loss": 0.41,
      "step": 9366
    },
    {
      "epoch": 0.42985636271855354,
      "grad_norm": 0.4755251109600067,
      "learning_rate": 9.614609723863824e-06,
      "loss": 0.3975,
      "step": 9367
    },
    {
      "epoch": 0.429902253223808,
      "grad_norm": 0.5186518430709839,
      "learning_rate": 9.614515324128899e-06,
      "loss": 0.5104,
      "step": 9368
    },
    {
      "epoch": 0.4299481437290625,
      "grad_norm": 0.46767401695251465,
      "learning_rate": 9.614420913297523e-06,
      "loss": 0.4207,
      "step": 9369
    },
    {
      "epoch": 0.4299940342343169,
      "grad_norm": 0.4562342166900635,
      "learning_rate": 9.614326491369918e-06,
      "loss": 0.4023,
      "step": 9370
    },
    {
      "epoch": 0.43003992473957137,
      "grad_norm": 0.5214709639549255,
      "learning_rate": 9.614232058346317e-06,
      "loss": 0.5009,
      "step": 9371
    },
    {
      "epoch": 0.43008581524482586,
      "grad_norm": 0.43774959444999695,
      "learning_rate": 9.614137614226944e-06,
      "loss": 0.3507,
      "step": 9372
    },
    {
      "epoch": 0.4301317057500803,
      "grad_norm": 0.4659118950366974,
      "learning_rate": 9.614043159012025e-06,
      "loss": 0.4417,
      "step": 9373
    },
    {
      "epoch": 0.43017759625533475,
      "grad_norm": 0.4844622313976288,
      "learning_rate": 9.613948692701788e-06,
      "loss": 0.4637,
      "step": 9374
    },
    {
      "epoch": 0.43022348676058925,
      "grad_norm": 0.4660356640815735,
      "learning_rate": 9.613854215296463e-06,
      "loss": 0.4089,
      "step": 9375
    },
    {
      "epoch": 0.4302693772658437,
      "grad_norm": 0.5010260343551636,
      "learning_rate": 9.613759726796274e-06,
      "loss": 0.4641,
      "step": 9376
    },
    {
      "epoch": 0.4303152677710982,
      "grad_norm": 0.45488470792770386,
      "learning_rate": 9.613665227201448e-06,
      "loss": 0.4029,
      "step": 9377
    },
    {
      "epoch": 0.43036115827635263,
      "grad_norm": 0.5302449464797974,
      "learning_rate": 9.613570716512214e-06,
      "loss": 0.6093,
      "step": 9378
    },
    {
      "epoch": 0.43040704878160707,
      "grad_norm": 0.45343151688575745,
      "learning_rate": 9.613476194728798e-06,
      "loss": 0.3532,
      "step": 9379
    },
    {
      "epoch": 0.43045293928686157,
      "grad_norm": 0.42666423320770264,
      "learning_rate": 9.613381661851428e-06,
      "loss": 0.3315,
      "step": 9380
    },
    {
      "epoch": 0.430498829792116,
      "grad_norm": 0.40548163652420044,
      "learning_rate": 9.613287117880331e-06,
      "loss": 0.3252,
      "step": 9381
    },
    {
      "epoch": 0.43054472029737045,
      "grad_norm": 0.5159562230110168,
      "learning_rate": 9.613192562815734e-06,
      "loss": 0.4775,
      "step": 9382
    },
    {
      "epoch": 0.43059061080262495,
      "grad_norm": 0.4368624687194824,
      "learning_rate": 9.613097996657866e-06,
      "loss": 0.3283,
      "step": 9383
    },
    {
      "epoch": 0.4306365013078794,
      "grad_norm": 0.4870242476463318,
      "learning_rate": 9.613003419406954e-06,
      "loss": 0.4814,
      "step": 9384
    },
    {
      "epoch": 0.43068239181313384,
      "grad_norm": 0.4523240029811859,
      "learning_rate": 9.612908831063223e-06,
      "loss": 0.4047,
      "step": 9385
    },
    {
      "epoch": 0.43072828231838833,
      "grad_norm": 0.4423651397228241,
      "learning_rate": 9.612814231626903e-06,
      "loss": 0.3166,
      "step": 9386
    },
    {
      "epoch": 0.4307741728236428,
      "grad_norm": 0.4728807508945465,
      "learning_rate": 9.612719621098219e-06,
      "loss": 0.4265,
      "step": 9387
    },
    {
      "epoch": 0.4308200633288973,
      "grad_norm": 0.4348207116127014,
      "learning_rate": 9.612624999477403e-06,
      "loss": 0.3906,
      "step": 9388
    },
    {
      "epoch": 0.4308659538341517,
      "grad_norm": 0.47058945894241333,
      "learning_rate": 9.612530366764677e-06,
      "loss": 0.4664,
      "step": 9389
    },
    {
      "epoch": 0.43091184433940616,
      "grad_norm": 0.4843747317790985,
      "learning_rate": 9.612435722960271e-06,
      "loss": 0.4426,
      "step": 9390
    },
    {
      "epoch": 0.43095773484466066,
      "grad_norm": 0.46338361501693726,
      "learning_rate": 9.612341068064415e-06,
      "loss": 0.3525,
      "step": 9391
    },
    {
      "epoch": 0.4310036253499151,
      "grad_norm": 0.4496888518333435,
      "learning_rate": 9.612246402077335e-06,
      "loss": 0.3451,
      "step": 9392
    },
    {
      "epoch": 0.43104951585516954,
      "grad_norm": 0.4750821888446808,
      "learning_rate": 9.612151724999254e-06,
      "loss": 0.4109,
      "step": 9393
    },
    {
      "epoch": 0.43109540636042404,
      "grad_norm": 0.47097453474998474,
      "learning_rate": 9.612057036830407e-06,
      "loss": 0.4209,
      "step": 9394
    },
    {
      "epoch": 0.4311412968656785,
      "grad_norm": 0.45879799127578735,
      "learning_rate": 9.611962337571018e-06,
      "loss": 0.3614,
      "step": 9395
    },
    {
      "epoch": 0.431187187370933,
      "grad_norm": 0.43468353152275085,
      "learning_rate": 9.611867627221314e-06,
      "loss": 0.3744,
      "step": 9396
    },
    {
      "epoch": 0.4312330778761874,
      "grad_norm": 0.4649869501590729,
      "learning_rate": 9.611772905781526e-06,
      "loss": 0.4359,
      "step": 9397
    },
    {
      "epoch": 0.43127896838144186,
      "grad_norm": 0.47029486298561096,
      "learning_rate": 9.611678173251876e-06,
      "loss": 0.4371,
      "step": 9398
    },
    {
      "epoch": 0.43132485888669636,
      "grad_norm": 0.45058026909828186,
      "learning_rate": 9.6115834296326e-06,
      "loss": 0.3673,
      "step": 9399
    },
    {
      "epoch": 0.4313707493919508,
      "grad_norm": 0.5006216764450073,
      "learning_rate": 9.61148867492392e-06,
      "loss": 0.5004,
      "step": 9400
    },
    {
      "epoch": 0.43141663989720525,
      "grad_norm": 0.504587709903717,
      "learning_rate": 9.611393909126063e-06,
      "loss": 0.4078,
      "step": 9401
    },
    {
      "epoch": 0.43146253040245974,
      "grad_norm": 0.46144551038742065,
      "learning_rate": 9.61129913223926e-06,
      "loss": 0.4672,
      "step": 9402
    },
    {
      "epoch": 0.4315084209077142,
      "grad_norm": 0.4192473590373993,
      "learning_rate": 9.611204344263739e-06,
      "loss": 0.323,
      "step": 9403
    },
    {
      "epoch": 0.4315543114129687,
      "grad_norm": 0.45751240849494934,
      "learning_rate": 9.611109545199726e-06,
      "loss": 0.3884,
      "step": 9404
    },
    {
      "epoch": 0.4316002019182231,
      "grad_norm": 0.49888166785240173,
      "learning_rate": 9.61101473504745e-06,
      "loss": 0.4231,
      "step": 9405
    },
    {
      "epoch": 0.43164609242347757,
      "grad_norm": 0.455972284078598,
      "learning_rate": 9.61091991380714e-06,
      "loss": 0.4122,
      "step": 9406
    },
    {
      "epoch": 0.43169198292873207,
      "grad_norm": 0.4863723814487457,
      "learning_rate": 9.610825081479021e-06,
      "loss": 0.4429,
      "step": 9407
    },
    {
      "epoch": 0.4317378734339865,
      "grad_norm": 0.44898533821105957,
      "learning_rate": 9.610730238063324e-06,
      "loss": 0.3827,
      "step": 9408
    },
    {
      "epoch": 0.43178376393924095,
      "grad_norm": 0.5116301774978638,
      "learning_rate": 9.610635383560276e-06,
      "loss": 0.4845,
      "step": 9409
    },
    {
      "epoch": 0.43182965444449545,
      "grad_norm": 0.463850200176239,
      "learning_rate": 9.610540517970105e-06,
      "loss": 0.382,
      "step": 9410
    },
    {
      "epoch": 0.4318755449497499,
      "grad_norm": 0.4139117896556854,
      "learning_rate": 9.610445641293037e-06,
      "loss": 0.3493,
      "step": 9411
    },
    {
      "epoch": 0.43192143545500433,
      "grad_norm": 0.43511831760406494,
      "learning_rate": 9.610350753529306e-06,
      "loss": 0.3891,
      "step": 9412
    },
    {
      "epoch": 0.43196732596025883,
      "grad_norm": 0.5639481544494629,
      "learning_rate": 9.610255854679134e-06,
      "loss": 0.6202,
      "step": 9413
    },
    {
      "epoch": 0.4320132164655133,
      "grad_norm": 0.4561227560043335,
      "learning_rate": 9.610160944742752e-06,
      "loss": 0.364,
      "step": 9414
    },
    {
      "epoch": 0.43205910697076777,
      "grad_norm": 0.48013460636138916,
      "learning_rate": 9.61006602372039e-06,
      "loss": 0.4019,
      "step": 9415
    },
    {
      "epoch": 0.4321049974760222,
      "grad_norm": 0.4754844009876251,
      "learning_rate": 9.609971091612272e-06,
      "loss": 0.3712,
      "step": 9416
    },
    {
      "epoch": 0.43215088798127665,
      "grad_norm": 0.48912543058395386,
      "learning_rate": 9.60987614841863e-06,
      "loss": 0.4797,
      "step": 9417
    },
    {
      "epoch": 0.43219677848653115,
      "grad_norm": 0.4525083303451538,
      "learning_rate": 9.60978119413969e-06,
      "loss": 0.4012,
      "step": 9418
    },
    {
      "epoch": 0.4322426689917856,
      "grad_norm": 0.4672132730484009,
      "learning_rate": 9.609686228775682e-06,
      "loss": 0.4231,
      "step": 9419
    },
    {
      "epoch": 0.43228855949704004,
      "grad_norm": 0.45324018597602844,
      "learning_rate": 9.609591252326831e-06,
      "loss": 0.397,
      "step": 9420
    },
    {
      "epoch": 0.43233445000229453,
      "grad_norm": 0.49847131967544556,
      "learning_rate": 9.60949626479337e-06,
      "loss": 0.5175,
      "step": 9421
    },
    {
      "epoch": 0.432380340507549,
      "grad_norm": 0.4509052336215973,
      "learning_rate": 9.609401266175524e-06,
      "loss": 0.3989,
      "step": 9422
    },
    {
      "epoch": 0.4324262310128035,
      "grad_norm": 0.45550236105918884,
      "learning_rate": 9.609306256473522e-06,
      "loss": 0.3819,
      "step": 9423
    },
    {
      "epoch": 0.4324721215180579,
      "grad_norm": 0.41421663761138916,
      "learning_rate": 9.609211235687595e-06,
      "loss": 0.3217,
      "step": 9424
    },
    {
      "epoch": 0.43251801202331236,
      "grad_norm": 0.4731671214103699,
      "learning_rate": 9.60911620381797e-06,
      "loss": 0.4454,
      "step": 9425
    },
    {
      "epoch": 0.43256390252856686,
      "grad_norm": 0.44386017322540283,
      "learning_rate": 9.609021160864873e-06,
      "loss": 0.3974,
      "step": 9426
    },
    {
      "epoch": 0.4326097930338213,
      "grad_norm": 0.4222288131713867,
      "learning_rate": 9.608926106828537e-06,
      "loss": 0.3068,
      "step": 9427
    },
    {
      "epoch": 0.43265568353907574,
      "grad_norm": 0.4673410654067993,
      "learning_rate": 9.608831041709188e-06,
      "loss": 0.3837,
      "step": 9428
    },
    {
      "epoch": 0.43270157404433024,
      "grad_norm": 0.4927833676338196,
      "learning_rate": 9.608735965507055e-06,
      "loss": 0.4184,
      "step": 9429
    },
    {
      "epoch": 0.4327474645495847,
      "grad_norm": 0.46952155232429504,
      "learning_rate": 9.608640878222367e-06,
      "loss": 0.4006,
      "step": 9430
    },
    {
      "epoch": 0.4327933550548392,
      "grad_norm": 0.4492369592189789,
      "learning_rate": 9.60854577985535e-06,
      "loss": 0.3567,
      "step": 9431
    },
    {
      "epoch": 0.4328392455600936,
      "grad_norm": 0.49307572841644287,
      "learning_rate": 9.608450670406235e-06,
      "loss": 0.4771,
      "step": 9432
    },
    {
      "epoch": 0.43288513606534806,
      "grad_norm": 0.9059242606163025,
      "learning_rate": 9.608355549875252e-06,
      "loss": 0.3824,
      "step": 9433
    },
    {
      "epoch": 0.43293102657060256,
      "grad_norm": 0.44346991181373596,
      "learning_rate": 9.608260418262628e-06,
      "loss": 0.3705,
      "step": 9434
    },
    {
      "epoch": 0.432976917075857,
      "grad_norm": 0.4912658631801605,
      "learning_rate": 9.608165275568593e-06,
      "loss": 0.4065,
      "step": 9435
    },
    {
      "epoch": 0.43302280758111145,
      "grad_norm": 0.4544029235839844,
      "learning_rate": 9.608070121793373e-06,
      "loss": 0.4197,
      "step": 9436
    },
    {
      "epoch": 0.43306869808636594,
      "grad_norm": 0.4280281364917755,
      "learning_rate": 9.6079749569372e-06,
      "loss": 0.3548,
      "step": 9437
    },
    {
      "epoch": 0.4331145885916204,
      "grad_norm": 0.46020108461380005,
      "learning_rate": 9.607879781000303e-06,
      "loss": 0.4141,
      "step": 9438
    },
    {
      "epoch": 0.4331604790968749,
      "grad_norm": 0.48227033019065857,
      "learning_rate": 9.607784593982906e-06,
      "loss": 0.4336,
      "step": 9439
    },
    {
      "epoch": 0.4332063696021293,
      "grad_norm": 0.5099912881851196,
      "learning_rate": 9.607689395885244e-06,
      "loss": 0.4812,
      "step": 9440
    },
    {
      "epoch": 0.43325226010738377,
      "grad_norm": 0.5185948014259338,
      "learning_rate": 9.607594186707542e-06,
      "loss": 0.425,
      "step": 9441
    },
    {
      "epoch": 0.43329815061263827,
      "grad_norm": 0.4974026381969452,
      "learning_rate": 9.607498966450031e-06,
      "loss": 0.4347,
      "step": 9442
    },
    {
      "epoch": 0.4333440411178927,
      "grad_norm": 0.46012112498283386,
      "learning_rate": 9.607403735112938e-06,
      "loss": 0.3862,
      "step": 9443
    },
    {
      "epoch": 0.43338993162314715,
      "grad_norm": 0.5346178412437439,
      "learning_rate": 9.607308492696494e-06,
      "loss": 0.5133,
      "step": 9444
    },
    {
      "epoch": 0.43343582212840165,
      "grad_norm": 0.43566057085990906,
      "learning_rate": 9.607213239200926e-06,
      "loss": 0.372,
      "step": 9445
    },
    {
      "epoch": 0.4334817126336561,
      "grad_norm": 0.46320590376853943,
      "learning_rate": 9.607117974626465e-06,
      "loss": 0.4043,
      "step": 9446
    },
    {
      "epoch": 0.43352760313891053,
      "grad_norm": 0.5251379013061523,
      "learning_rate": 9.607022698973341e-06,
      "loss": 0.4748,
      "step": 9447
    },
    {
      "epoch": 0.43357349364416503,
      "grad_norm": 0.4413606524467468,
      "learning_rate": 9.60692741224178e-06,
      "loss": 0.4233,
      "step": 9448
    },
    {
      "epoch": 0.4336193841494195,
      "grad_norm": 0.4972039759159088,
      "learning_rate": 9.606832114432012e-06,
      "loss": 0.4904,
      "step": 9449
    },
    {
      "epoch": 0.43366527465467397,
      "grad_norm": 0.4795592725276947,
      "learning_rate": 9.606736805544266e-06,
      "loss": 0.4393,
      "step": 9450
    },
    {
      "epoch": 0.4337111651599284,
      "grad_norm": 0.9199171662330627,
      "learning_rate": 9.606641485578773e-06,
      "loss": 0.5734,
      "step": 9451
    },
    {
      "epoch": 0.43375705566518286,
      "grad_norm": 0.4427761733531952,
      "learning_rate": 9.606546154535763e-06,
      "loss": 0.3629,
      "step": 9452
    },
    {
      "epoch": 0.43380294617043735,
      "grad_norm": 0.4971867501735687,
      "learning_rate": 9.606450812415461e-06,
      "loss": 0.4502,
      "step": 9453
    },
    {
      "epoch": 0.4338488366756918,
      "grad_norm": 0.4610409438610077,
      "learning_rate": 9.6063554592181e-06,
      "loss": 0.3953,
      "step": 9454
    },
    {
      "epoch": 0.43389472718094624,
      "grad_norm": 0.540839433670044,
      "learning_rate": 9.606260094943907e-06,
      "loss": 0.5324,
      "step": 9455
    },
    {
      "epoch": 0.43394061768620074,
      "grad_norm": 0.46103745698928833,
      "learning_rate": 9.606164719593111e-06,
      "loss": 0.3814,
      "step": 9456
    },
    {
      "epoch": 0.4339865081914552,
      "grad_norm": 0.4704090356826782,
      "learning_rate": 9.606069333165945e-06,
      "loss": 0.4242,
      "step": 9457
    },
    {
      "epoch": 0.4340323986967097,
      "grad_norm": 0.4563893675804138,
      "learning_rate": 9.605973935662635e-06,
      "loss": 0.3891,
      "step": 9458
    },
    {
      "epoch": 0.4340782892019641,
      "grad_norm": 0.5015137791633606,
      "learning_rate": 9.605878527083411e-06,
      "loss": 0.3907,
      "step": 9459
    },
    {
      "epoch": 0.43412417970721856,
      "grad_norm": 0.4614434242248535,
      "learning_rate": 9.605783107428503e-06,
      "loss": 0.3574,
      "step": 9460
    },
    {
      "epoch": 0.43417007021247306,
      "grad_norm": 0.44149383902549744,
      "learning_rate": 9.605687676698141e-06,
      "loss": 0.3515,
      "step": 9461
    },
    {
      "epoch": 0.4342159607177275,
      "grad_norm": 0.4034615457057953,
      "learning_rate": 9.605592234892553e-06,
      "loss": 0.2984,
      "step": 9462
    },
    {
      "epoch": 0.43426185122298194,
      "grad_norm": 0.49119535088539124,
      "learning_rate": 9.60549678201197e-06,
      "loss": 0.4548,
      "step": 9463
    },
    {
      "epoch": 0.43430774172823644,
      "grad_norm": 0.4659927487373352,
      "learning_rate": 9.60540131805662e-06,
      "loss": 0.3694,
      "step": 9464
    },
    {
      "epoch": 0.4343536322334909,
      "grad_norm": 0.4542892873287201,
      "learning_rate": 9.605305843026734e-06,
      "loss": 0.4375,
      "step": 9465
    },
    {
      "epoch": 0.4343995227387454,
      "grad_norm": 0.4805331826210022,
      "learning_rate": 9.605210356922538e-06,
      "loss": 0.466,
      "step": 9466
    },
    {
      "epoch": 0.4344454132439998,
      "grad_norm": 0.46991226077079773,
      "learning_rate": 9.60511485974427e-06,
      "loss": 0.4049,
      "step": 9467
    },
    {
      "epoch": 0.43449130374925427,
      "grad_norm": 0.48190444707870483,
      "learning_rate": 9.60501935149215e-06,
      "loss": 0.432,
      "step": 9468
    },
    {
      "epoch": 0.43453719425450876,
      "grad_norm": 0.5730679035186768,
      "learning_rate": 9.604923832166413e-06,
      "loss": 0.4493,
      "step": 9469
    },
    {
      "epoch": 0.4345830847597632,
      "grad_norm": 0.45229271054267883,
      "learning_rate": 9.604828301767286e-06,
      "loss": 0.367,
      "step": 9470
    },
    {
      "epoch": 0.43462897526501765,
      "grad_norm": 0.4923461079597473,
      "learning_rate": 9.604732760295004e-06,
      "loss": 0.4005,
      "step": 9471
    },
    {
      "epoch": 0.43467486577027215,
      "grad_norm": 0.4399645924568176,
      "learning_rate": 9.60463720774979e-06,
      "loss": 0.3317,
      "step": 9472
    },
    {
      "epoch": 0.4347207562755266,
      "grad_norm": 0.4814951419830322,
      "learning_rate": 9.604541644131878e-06,
      "loss": 0.3795,
      "step": 9473
    },
    {
      "epoch": 0.43476664678078103,
      "grad_norm": 0.43064799904823303,
      "learning_rate": 9.604446069441496e-06,
      "loss": 0.348,
      "step": 9474
    },
    {
      "epoch": 0.4348125372860355,
      "grad_norm": 0.4646523892879486,
      "learning_rate": 9.604350483678875e-06,
      "loss": 0.3811,
      "step": 9475
    },
    {
      "epoch": 0.43485842779128997,
      "grad_norm": 0.47420167922973633,
      "learning_rate": 9.604254886844243e-06,
      "loss": 0.4106,
      "step": 9476
    },
    {
      "epoch": 0.43490431829654447,
      "grad_norm": 0.5082074403762817,
      "learning_rate": 9.604159278937833e-06,
      "loss": 0.46,
      "step": 9477
    },
    {
      "epoch": 0.4349502088017989,
      "grad_norm": 0.4871568977832794,
      "learning_rate": 9.60406365995987e-06,
      "loss": 0.4379,
      "step": 9478
    },
    {
      "epoch": 0.43499609930705335,
      "grad_norm": 0.4797271192073822,
      "learning_rate": 9.603968029910589e-06,
      "loss": 0.3855,
      "step": 9479
    },
    {
      "epoch": 0.43504198981230785,
      "grad_norm": 0.43549373745918274,
      "learning_rate": 9.603872388790218e-06,
      "loss": 0.3664,
      "step": 9480
    },
    {
      "epoch": 0.4350878803175623,
      "grad_norm": 0.4948456287384033,
      "learning_rate": 9.603776736598988e-06,
      "loss": 0.4371,
      "step": 9481
    },
    {
      "epoch": 0.43513377082281673,
      "grad_norm": 0.4821363389492035,
      "learning_rate": 9.603681073337126e-06,
      "loss": 0.434,
      "step": 9482
    },
    {
      "epoch": 0.43517966132807123,
      "grad_norm": 0.45945584774017334,
      "learning_rate": 9.603585399004866e-06,
      "loss": 0.3649,
      "step": 9483
    },
    {
      "epoch": 0.4352255518333257,
      "grad_norm": 0.4420243203639984,
      "learning_rate": 9.603489713602434e-06,
      "loss": 0.3929,
      "step": 9484
    },
    {
      "epoch": 0.4352714423385802,
      "grad_norm": 0.457217812538147,
      "learning_rate": 9.603394017130063e-06,
      "loss": 0.3579,
      "step": 9485
    },
    {
      "epoch": 0.4353173328438346,
      "grad_norm": 0.4531521201133728,
      "learning_rate": 9.603298309587982e-06,
      "loss": 0.3579,
      "step": 9486
    },
    {
      "epoch": 0.43536322334908906,
      "grad_norm": 0.4106457531452179,
      "learning_rate": 9.60320259097642e-06,
      "loss": 0.3202,
      "step": 9487
    },
    {
      "epoch": 0.43540911385434355,
      "grad_norm": 0.460921972990036,
      "learning_rate": 9.603106861295613e-06,
      "loss": 0.4365,
      "step": 9488
    },
    {
      "epoch": 0.435455004359598,
      "grad_norm": 0.4275890588760376,
      "learning_rate": 9.603011120545783e-06,
      "loss": 0.3401,
      "step": 9489
    },
    {
      "epoch": 0.43550089486485244,
      "grad_norm": 0.46876978874206543,
      "learning_rate": 9.602915368727166e-06,
      "loss": 0.4684,
      "step": 9490
    },
    {
      "epoch": 0.43554678537010694,
      "grad_norm": 0.4690188765525818,
      "learning_rate": 9.602819605839988e-06,
      "loss": 0.4397,
      "step": 9491
    },
    {
      "epoch": 0.4355926758753614,
      "grad_norm": 0.48514583706855774,
      "learning_rate": 9.602723831884485e-06,
      "loss": 0.4298,
      "step": 9492
    },
    {
      "epoch": 0.4356385663806159,
      "grad_norm": 0.4991267919540405,
      "learning_rate": 9.60262804686088e-06,
      "loss": 0.4446,
      "step": 9493
    },
    {
      "epoch": 0.4356844568858703,
      "grad_norm": 0.4434119462966919,
      "learning_rate": 9.602532250769409e-06,
      "loss": 0.3573,
      "step": 9494
    },
    {
      "epoch": 0.43573034739112476,
      "grad_norm": 0.4420779049396515,
      "learning_rate": 9.6024364436103e-06,
      "loss": 0.3557,
      "step": 9495
    },
    {
      "epoch": 0.43577623789637926,
      "grad_norm": 0.4561111330986023,
      "learning_rate": 9.602340625383785e-06,
      "loss": 0.4103,
      "step": 9496
    },
    {
      "epoch": 0.4358221284016337,
      "grad_norm": 0.5396824479103088,
      "learning_rate": 9.602244796090092e-06,
      "loss": 0.438,
      "step": 9497
    },
    {
      "epoch": 0.43586801890688814,
      "grad_norm": 0.5025354623794556,
      "learning_rate": 9.602148955729454e-06,
      "loss": 0.4624,
      "step": 9498
    },
    {
      "epoch": 0.43591390941214264,
      "grad_norm": 0.4464099407196045,
      "learning_rate": 9.602053104302101e-06,
      "loss": 0.353,
      "step": 9499
    },
    {
      "epoch": 0.4359597999173971,
      "grad_norm": 0.47576573491096497,
      "learning_rate": 9.60195724180826e-06,
      "loss": 0.3878,
      "step": 9500
    },
    {
      "epoch": 0.4360056904226515,
      "grad_norm": 0.4834880232810974,
      "learning_rate": 9.601861368248165e-06,
      "loss": 0.444,
      "step": 9501
    },
    {
      "epoch": 0.436051580927906,
      "grad_norm": 0.4710325002670288,
      "learning_rate": 9.601765483622048e-06,
      "loss": 0.48,
      "step": 9502
    },
    {
      "epoch": 0.43609747143316047,
      "grad_norm": 0.4765704572200775,
      "learning_rate": 9.601669587930134e-06,
      "loss": 0.4363,
      "step": 9503
    },
    {
      "epoch": 0.43614336193841496,
      "grad_norm": 0.47389891743659973,
      "learning_rate": 9.60157368117266e-06,
      "loss": 0.4445,
      "step": 9504
    },
    {
      "epoch": 0.4361892524436694,
      "grad_norm": 0.4966842830181122,
      "learning_rate": 9.601477763349853e-06,
      "loss": 0.4541,
      "step": 9505
    },
    {
      "epoch": 0.43623514294892385,
      "grad_norm": 0.503948986530304,
      "learning_rate": 9.601381834461942e-06,
      "loss": 0.4241,
      "step": 9506
    },
    {
      "epoch": 0.43628103345417835,
      "grad_norm": 0.44321370124816895,
      "learning_rate": 9.601285894509162e-06,
      "loss": 0.3601,
      "step": 9507
    },
    {
      "epoch": 0.4363269239594328,
      "grad_norm": 0.48795321583747864,
      "learning_rate": 9.601189943491742e-06,
      "loss": 0.4362,
      "step": 9508
    },
    {
      "epoch": 0.43637281446468723,
      "grad_norm": 0.5016379952430725,
      "learning_rate": 9.601093981409911e-06,
      "loss": 0.4245,
      "step": 9509
    },
    {
      "epoch": 0.43641870496994173,
      "grad_norm": 0.4635215103626251,
      "learning_rate": 9.600998008263902e-06,
      "loss": 0.4033,
      "step": 9510
    },
    {
      "epoch": 0.43646459547519617,
      "grad_norm": 0.46343687176704407,
      "learning_rate": 9.600902024053944e-06,
      "loss": 0.3665,
      "step": 9511
    },
    {
      "epoch": 0.43651048598045067,
      "grad_norm": 0.46449822187423706,
      "learning_rate": 9.60080602878027e-06,
      "loss": 0.3782,
      "step": 9512
    },
    {
      "epoch": 0.4365563764857051,
      "grad_norm": 0.5395718216896057,
      "learning_rate": 9.600710022443109e-06,
      "loss": 0.5637,
      "step": 9513
    },
    {
      "epoch": 0.43660226699095955,
      "grad_norm": 0.4605312943458557,
      "learning_rate": 9.600614005042693e-06,
      "loss": 0.387,
      "step": 9514
    },
    {
      "epoch": 0.43664815749621405,
      "grad_norm": 0.47935715317726135,
      "learning_rate": 9.600517976579251e-06,
      "loss": 0.4268,
      "step": 9515
    },
    {
      "epoch": 0.4366940480014685,
      "grad_norm": 0.48694777488708496,
      "learning_rate": 9.600421937053015e-06,
      "loss": 0.3775,
      "step": 9516
    },
    {
      "epoch": 0.43673993850672294,
      "grad_norm": 0.48647522926330566,
      "learning_rate": 9.600325886464217e-06,
      "loss": 0.4895,
      "step": 9517
    },
    {
      "epoch": 0.43678582901197743,
      "grad_norm": 0.5025939345359802,
      "learning_rate": 9.600229824813087e-06,
      "loss": 0.457,
      "step": 9518
    },
    {
      "epoch": 0.4368317195172319,
      "grad_norm": 0.4780937433242798,
      "learning_rate": 9.600133752099856e-06,
      "loss": 0.3984,
      "step": 9519
    },
    {
      "epoch": 0.4368776100224864,
      "grad_norm": 0.4750468134880066,
      "learning_rate": 9.600037668324756e-06,
      "loss": 0.4292,
      "step": 9520
    },
    {
      "epoch": 0.4369235005277408,
      "grad_norm": 0.46017226576805115,
      "learning_rate": 9.599941573488016e-06,
      "loss": 0.44,
      "step": 9521
    },
    {
      "epoch": 0.43696939103299526,
      "grad_norm": 0.49469447135925293,
      "learning_rate": 9.59984546758987e-06,
      "loss": 0.4668,
      "step": 9522
    },
    {
      "epoch": 0.43701528153824976,
      "grad_norm": 0.42648306488990784,
      "learning_rate": 9.599749350630547e-06,
      "loss": 0.3352,
      "step": 9523
    },
    {
      "epoch": 0.4370611720435042,
      "grad_norm": 0.5039868354797363,
      "learning_rate": 9.599653222610278e-06,
      "loss": 0.5306,
      "step": 9524
    },
    {
      "epoch": 0.43710706254875864,
      "grad_norm": 0.4735119640827179,
      "learning_rate": 9.599557083529295e-06,
      "loss": 0.4487,
      "step": 9525
    },
    {
      "epoch": 0.43715295305401314,
      "grad_norm": 0.4705061912536621,
      "learning_rate": 9.599460933387828e-06,
      "loss": 0.4126,
      "step": 9526
    },
    {
      "epoch": 0.4371988435592676,
      "grad_norm": 0.47537651658058167,
      "learning_rate": 9.599364772186109e-06,
      "loss": 0.4323,
      "step": 9527
    },
    {
      "epoch": 0.4372447340645221,
      "grad_norm": 0.4777722656726837,
      "learning_rate": 9.59926859992437e-06,
      "loss": 0.4177,
      "step": 9528
    },
    {
      "epoch": 0.4372906245697765,
      "grad_norm": 0.4627701938152313,
      "learning_rate": 9.599172416602843e-06,
      "loss": 0.4882,
      "step": 9529
    },
    {
      "epoch": 0.43733651507503096,
      "grad_norm": 0.41366854310035706,
      "learning_rate": 9.599076222221755e-06,
      "loss": 0.3336,
      "step": 9530
    },
    {
      "epoch": 0.43738240558028546,
      "grad_norm": 0.4612693786621094,
      "learning_rate": 9.59898001678134e-06,
      "loss": 0.4008,
      "step": 9531
    },
    {
      "epoch": 0.4374282960855399,
      "grad_norm": 0.44290438294410706,
      "learning_rate": 9.598883800281833e-06,
      "loss": 0.345,
      "step": 9532
    },
    {
      "epoch": 0.43747418659079434,
      "grad_norm": 0.46944841742515564,
      "learning_rate": 9.59878757272346e-06,
      "loss": 0.3832,
      "step": 9533
    },
    {
      "epoch": 0.43752007709604884,
      "grad_norm": 0.520319402217865,
      "learning_rate": 9.598691334106453e-06,
      "loss": 0.5183,
      "step": 9534
    },
    {
      "epoch": 0.4375659676013033,
      "grad_norm": 0.45137467980384827,
      "learning_rate": 9.598595084431046e-06,
      "loss": 0.3357,
      "step": 9535
    },
    {
      "epoch": 0.4376118581065577,
      "grad_norm": 0.4630063474178314,
      "learning_rate": 9.59849882369747e-06,
      "loss": 0.4132,
      "step": 9536
    },
    {
      "epoch": 0.4376577486118122,
      "grad_norm": 0.4519720673561096,
      "learning_rate": 9.598402551905955e-06,
      "loss": 0.387,
      "step": 9537
    },
    {
      "epoch": 0.43770363911706667,
      "grad_norm": 0.4464069902896881,
      "learning_rate": 9.598306269056733e-06,
      "loss": 0.3556,
      "step": 9538
    },
    {
      "epoch": 0.43774952962232117,
      "grad_norm": 0.4828057289123535,
      "learning_rate": 9.598209975150035e-06,
      "loss": 0.4224,
      "step": 9539
    },
    {
      "epoch": 0.4377954201275756,
      "grad_norm": 0.48324036598205566,
      "learning_rate": 9.598113670186096e-06,
      "loss": 0.4474,
      "step": 9540
    },
    {
      "epoch": 0.43784131063283005,
      "grad_norm": 0.4475082755088806,
      "learning_rate": 9.598017354165141e-06,
      "loss": 0.3249,
      "step": 9541
    },
    {
      "epoch": 0.43788720113808455,
      "grad_norm": 0.4720956087112427,
      "learning_rate": 9.597921027087408e-06,
      "loss": 0.4081,
      "step": 9542
    },
    {
      "epoch": 0.437933091643339,
      "grad_norm": 0.4305553436279297,
      "learning_rate": 9.597824688953125e-06,
      "loss": 0.3958,
      "step": 9543
    },
    {
      "epoch": 0.43797898214859343,
      "grad_norm": 0.4112338721752167,
      "learning_rate": 9.597728339762526e-06,
      "loss": 0.3558,
      "step": 9544
    },
    {
      "epoch": 0.43802487265384793,
      "grad_norm": 0.46954888105392456,
      "learning_rate": 9.59763197951584e-06,
      "loss": 0.4346,
      "step": 9545
    },
    {
      "epoch": 0.43807076315910237,
      "grad_norm": 0.4459943473339081,
      "learning_rate": 9.5975356082133e-06,
      "loss": 0.3925,
      "step": 9546
    },
    {
      "epoch": 0.43811665366435687,
      "grad_norm": 0.48806485533714294,
      "learning_rate": 9.597439225855139e-06,
      "loss": 0.4316,
      "step": 9547
    },
    {
      "epoch": 0.4381625441696113,
      "grad_norm": 0.46897292137145996,
      "learning_rate": 9.597342832441587e-06,
      "loss": 0.4863,
      "step": 9548
    },
    {
      "epoch": 0.43820843467486575,
      "grad_norm": 0.4176226854324341,
      "learning_rate": 9.597246427972878e-06,
      "loss": 0.3357,
      "step": 9549
    },
    {
      "epoch": 0.43825432518012025,
      "grad_norm": 0.46972376108169556,
      "learning_rate": 9.59715001244924e-06,
      "loss": 0.3881,
      "step": 9550
    },
    {
      "epoch": 0.4383002156853747,
      "grad_norm": 0.4630694091320038,
      "learning_rate": 9.597053585870909e-06,
      "loss": 0.4316,
      "step": 9551
    },
    {
      "epoch": 0.43834610619062914,
      "grad_norm": 0.506456732749939,
      "learning_rate": 9.596957148238113e-06,
      "loss": 0.5058,
      "step": 9552
    },
    {
      "epoch": 0.43839199669588363,
      "grad_norm": 0.44118791818618774,
      "learning_rate": 9.596860699551087e-06,
      "loss": 0.3391,
      "step": 9553
    },
    {
      "epoch": 0.4384378872011381,
      "grad_norm": 0.48329615592956543,
      "learning_rate": 9.596764239810061e-06,
      "loss": 0.4639,
      "step": 9554
    },
    {
      "epoch": 0.4384837777063926,
      "grad_norm": 0.47136417031288147,
      "learning_rate": 9.596667769015269e-06,
      "loss": 0.4262,
      "step": 9555
    },
    {
      "epoch": 0.438529668211647,
      "grad_norm": 0.4728092849254608,
      "learning_rate": 9.596571287166942e-06,
      "loss": 0.4678,
      "step": 9556
    },
    {
      "epoch": 0.43857555871690146,
      "grad_norm": 0.4291680157184601,
      "learning_rate": 9.59647479426531e-06,
      "loss": 0.351,
      "step": 9557
    },
    {
      "epoch": 0.43862144922215596,
      "grad_norm": 0.47856372594833374,
      "learning_rate": 9.596378290310609e-06,
      "loss": 0.4745,
      "step": 9558
    },
    {
      "epoch": 0.4386673397274104,
      "grad_norm": 0.4719753861427307,
      "learning_rate": 9.596281775303066e-06,
      "loss": 0.4166,
      "step": 9559
    },
    {
      "epoch": 0.43871323023266484,
      "grad_norm": 0.4732882082462311,
      "learning_rate": 9.596185249242918e-06,
      "loss": 0.4545,
      "step": 9560
    },
    {
      "epoch": 0.43875912073791934,
      "grad_norm": 0.4735700190067291,
      "learning_rate": 9.596088712130396e-06,
      "loss": 0.4461,
      "step": 9561
    },
    {
      "epoch": 0.4388050112431738,
      "grad_norm": 0.4803396463394165,
      "learning_rate": 9.59599216396573e-06,
      "loss": 0.4025,
      "step": 9562
    },
    {
      "epoch": 0.4388509017484282,
      "grad_norm": 0.7367607951164246,
      "learning_rate": 9.595895604749153e-06,
      "loss": 0.5485,
      "step": 9563
    },
    {
      "epoch": 0.4388967922536827,
      "grad_norm": 0.4681912660598755,
      "learning_rate": 9.595799034480898e-06,
      "loss": 0.4451,
      "step": 9564
    },
    {
      "epoch": 0.43894268275893716,
      "grad_norm": 0.44611018896102905,
      "learning_rate": 9.595702453161198e-06,
      "loss": 0.3619,
      "step": 9565
    },
    {
      "epoch": 0.43898857326419166,
      "grad_norm": 0.4557933807373047,
      "learning_rate": 9.595605860790282e-06,
      "loss": 0.4256,
      "step": 9566
    },
    {
      "epoch": 0.4390344637694461,
      "grad_norm": 0.4834311604499817,
      "learning_rate": 9.595509257368386e-06,
      "loss": 0.4134,
      "step": 9567
    },
    {
      "epoch": 0.43908035427470055,
      "grad_norm": 0.5164036154747009,
      "learning_rate": 9.595412642895738e-06,
      "loss": 0.5809,
      "step": 9568
    },
    {
      "epoch": 0.43912624477995504,
      "grad_norm": 0.48303714394569397,
      "learning_rate": 9.595316017372576e-06,
      "loss": 0.4757,
      "step": 9569
    },
    {
      "epoch": 0.4391721352852095,
      "grad_norm": 0.47512075304985046,
      "learning_rate": 9.595219380799128e-06,
      "loss": 0.3888,
      "step": 9570
    },
    {
      "epoch": 0.43921802579046393,
      "grad_norm": 0.4493614137172699,
      "learning_rate": 9.595122733175626e-06,
      "loss": 0.3954,
      "step": 9571
    },
    {
      "epoch": 0.4392639162957184,
      "grad_norm": 0.4429350197315216,
      "learning_rate": 9.595026074502307e-06,
      "loss": 0.4018,
      "step": 9572
    },
    {
      "epoch": 0.43930980680097287,
      "grad_norm": 0.47695687413215637,
      "learning_rate": 9.594929404779398e-06,
      "loss": 0.4396,
      "step": 9573
    },
    {
      "epoch": 0.43935569730622737,
      "grad_norm": 0.4400349259376526,
      "learning_rate": 9.594832724007137e-06,
      "loss": 0.367,
      "step": 9574
    },
    {
      "epoch": 0.4394015878114818,
      "grad_norm": 0.4291749894618988,
      "learning_rate": 9.594736032185751e-06,
      "loss": 0.3441,
      "step": 9575
    },
    {
      "epoch": 0.43944747831673625,
      "grad_norm": 0.48809078335762024,
      "learning_rate": 9.594639329315476e-06,
      "loss": 0.457,
      "step": 9576
    },
    {
      "epoch": 0.43949336882199075,
      "grad_norm": 0.47735485434532166,
      "learning_rate": 9.594542615396542e-06,
      "loss": 0.4483,
      "step": 9577
    },
    {
      "epoch": 0.4395392593272452,
      "grad_norm": 0.4509314298629761,
      "learning_rate": 9.594445890429185e-06,
      "loss": 0.3491,
      "step": 9578
    },
    {
      "epoch": 0.43958514983249963,
      "grad_norm": 0.47399163246154785,
      "learning_rate": 9.594349154413635e-06,
      "loss": 0.4207,
      "step": 9579
    },
    {
      "epoch": 0.43963104033775413,
      "grad_norm": 0.44592374563217163,
      "learning_rate": 9.594252407350126e-06,
      "loss": 0.3543,
      "step": 9580
    },
    {
      "epoch": 0.4396769308430086,
      "grad_norm": 0.4471324384212494,
      "learning_rate": 9.59415564923889e-06,
      "loss": 0.3725,
      "step": 9581
    },
    {
      "epoch": 0.43972282134826307,
      "grad_norm": 0.5123616456985474,
      "learning_rate": 9.594058880080157e-06,
      "loss": 0.5099,
      "step": 9582
    },
    {
      "epoch": 0.4397687118535175,
      "grad_norm": 0.4294447898864746,
      "learning_rate": 9.593962099874166e-06,
      "loss": 0.3653,
      "step": 9583
    },
    {
      "epoch": 0.43981460235877196,
      "grad_norm": 0.41964370012283325,
      "learning_rate": 9.593865308621144e-06,
      "loss": 0.3352,
      "step": 9584
    },
    {
      "epoch": 0.43986049286402645,
      "grad_norm": 0.4615685045719147,
      "learning_rate": 9.593768506321326e-06,
      "loss": 0.4276,
      "step": 9585
    },
    {
      "epoch": 0.4399063833692809,
      "grad_norm": 0.47215625643730164,
      "learning_rate": 9.593671692974944e-06,
      "loss": 0.4478,
      "step": 9586
    },
    {
      "epoch": 0.43995227387453534,
      "grad_norm": 0.44181594252586365,
      "learning_rate": 9.593574868582232e-06,
      "loss": 0.3995,
      "step": 9587
    },
    {
      "epoch": 0.43999816437978984,
      "grad_norm": 0.4748929738998413,
      "learning_rate": 9.593478033143422e-06,
      "loss": 0.4311,
      "step": 9588
    },
    {
      "epoch": 0.4400440548850443,
      "grad_norm": 0.4751283526420593,
      "learning_rate": 9.593381186658748e-06,
      "loss": 0.4445,
      "step": 9589
    },
    {
      "epoch": 0.4400899453902987,
      "grad_norm": 0.438077449798584,
      "learning_rate": 9.593284329128441e-06,
      "loss": 0.3401,
      "step": 9590
    },
    {
      "epoch": 0.4401358358955532,
      "grad_norm": 0.4550577700138092,
      "learning_rate": 9.593187460552734e-06,
      "loss": 0.4084,
      "step": 9591
    },
    {
      "epoch": 0.44018172640080766,
      "grad_norm": 0.46328914165496826,
      "learning_rate": 9.593090580931862e-06,
      "loss": 0.3899,
      "step": 9592
    },
    {
      "epoch": 0.44022761690606216,
      "grad_norm": 0.5900828242301941,
      "learning_rate": 9.592993690266058e-06,
      "loss": 0.3853,
      "step": 9593
    },
    {
      "epoch": 0.4402735074113166,
      "grad_norm": 0.4365132451057434,
      "learning_rate": 9.592896788555553e-06,
      "loss": 0.3784,
      "step": 9594
    },
    {
      "epoch": 0.44031939791657104,
      "grad_norm": 0.45450422167778015,
      "learning_rate": 9.59279987580058e-06,
      "loss": 0.3958,
      "step": 9595
    },
    {
      "epoch": 0.44036528842182554,
      "grad_norm": 0.4881937503814697,
      "learning_rate": 9.592702952001372e-06,
      "loss": 0.3977,
      "step": 9596
    },
    {
      "epoch": 0.44041117892708,
      "grad_norm": 0.4581640362739563,
      "learning_rate": 9.592606017158165e-06,
      "loss": 0.3548,
      "step": 9597
    },
    {
      "epoch": 0.4404570694323344,
      "grad_norm": 0.47902989387512207,
      "learning_rate": 9.592509071271188e-06,
      "loss": 0.415,
      "step": 9598
    },
    {
      "epoch": 0.4405029599375889,
      "grad_norm": 0.4827384650707245,
      "learning_rate": 9.592412114340677e-06,
      "loss": 0.401,
      "step": 9599
    },
    {
      "epoch": 0.44054885044284336,
      "grad_norm": 0.48581916093826294,
      "learning_rate": 9.592315146366865e-06,
      "loss": 0.4201,
      "step": 9600
    },
    {
      "epoch": 0.44059474094809786,
      "grad_norm": 0.4750884771347046,
      "learning_rate": 9.592218167349985e-06,
      "loss": 0.4305,
      "step": 9601
    },
    {
      "epoch": 0.4406406314533523,
      "grad_norm": 0.47014978528022766,
      "learning_rate": 9.592121177290267e-06,
      "loss": 0.3972,
      "step": 9602
    },
    {
      "epoch": 0.44068652195860675,
      "grad_norm": 0.44357892870903015,
      "learning_rate": 9.592024176187948e-06,
      "loss": 0.4019,
      "step": 9603
    },
    {
      "epoch": 0.44073241246386125,
      "grad_norm": 0.47168901562690735,
      "learning_rate": 9.591927164043263e-06,
      "loss": 0.4105,
      "step": 9604
    },
    {
      "epoch": 0.4407783029691157,
      "grad_norm": 0.45505568385124207,
      "learning_rate": 9.591830140856438e-06,
      "loss": 0.3724,
      "step": 9605
    },
    {
      "epoch": 0.44082419347437013,
      "grad_norm": 0.4602077007293701,
      "learning_rate": 9.591733106627713e-06,
      "loss": 0.4212,
      "step": 9606
    },
    {
      "epoch": 0.4408700839796246,
      "grad_norm": 0.4131883978843689,
      "learning_rate": 9.591636061357318e-06,
      "loss": 0.296,
      "step": 9607
    },
    {
      "epoch": 0.44091597448487907,
      "grad_norm": 0.4569132626056671,
      "learning_rate": 9.591539005045489e-06,
      "loss": 0.4232,
      "step": 9608
    },
    {
      "epoch": 0.44096186499013357,
      "grad_norm": 0.4728423058986664,
      "learning_rate": 9.591441937692457e-06,
      "loss": 0.4557,
      "step": 9609
    },
    {
      "epoch": 0.441007755495388,
      "grad_norm": 0.48723241686820984,
      "learning_rate": 9.591344859298456e-06,
      "loss": 0.4776,
      "step": 9610
    },
    {
      "epoch": 0.44105364600064245,
      "grad_norm": 0.469016432762146,
      "learning_rate": 9.59124776986372e-06,
      "loss": 0.3925,
      "step": 9611
    },
    {
      "epoch": 0.44109953650589695,
      "grad_norm": 0.46609246730804443,
      "learning_rate": 9.591150669388482e-06,
      "loss": 0.392,
      "step": 9612
    },
    {
      "epoch": 0.4411454270111514,
      "grad_norm": 0.4937361478805542,
      "learning_rate": 9.591053557872975e-06,
      "loss": 0.4476,
      "step": 9613
    },
    {
      "epoch": 0.44119131751640583,
      "grad_norm": 0.49860453605651855,
      "learning_rate": 9.590956435317436e-06,
      "loss": 0.4802,
      "step": 9614
    },
    {
      "epoch": 0.44123720802166033,
      "grad_norm": 0.47362595796585083,
      "learning_rate": 9.590859301722093e-06,
      "loss": 0.4158,
      "step": 9615
    },
    {
      "epoch": 0.4412830985269148,
      "grad_norm": 0.48048558831214905,
      "learning_rate": 9.590762157087182e-06,
      "loss": 0.4388,
      "step": 9616
    },
    {
      "epoch": 0.4413289890321692,
      "grad_norm": 0.5018331408500671,
      "learning_rate": 9.590665001412938e-06,
      "loss": 0.4515,
      "step": 9617
    },
    {
      "epoch": 0.4413748795374237,
      "grad_norm": 0.45968523621559143,
      "learning_rate": 9.590567834699593e-06,
      "loss": 0.4522,
      "step": 9618
    },
    {
      "epoch": 0.44142077004267816,
      "grad_norm": 0.445666640996933,
      "learning_rate": 9.590470656947382e-06,
      "loss": 0.3746,
      "step": 9619
    },
    {
      "epoch": 0.44146666054793265,
      "grad_norm": 0.4866577088832855,
      "learning_rate": 9.590373468156538e-06,
      "loss": 0.4338,
      "step": 9620
    },
    {
      "epoch": 0.4415125510531871,
      "grad_norm": 0.4516773819923401,
      "learning_rate": 9.590276268327295e-06,
      "loss": 0.399,
      "step": 9621
    },
    {
      "epoch": 0.44155844155844154,
      "grad_norm": 0.4766620695590973,
      "learning_rate": 9.590179057459884e-06,
      "loss": 0.4171,
      "step": 9622
    },
    {
      "epoch": 0.44160433206369604,
      "grad_norm": 0.3945341408252716,
      "learning_rate": 9.590081835554542e-06,
      "loss": 0.29,
      "step": 9623
    },
    {
      "epoch": 0.4416502225689505,
      "grad_norm": 0.5096600651741028,
      "learning_rate": 9.589984602611501e-06,
      "loss": 0.4799,
      "step": 9624
    },
    {
      "epoch": 0.4416961130742049,
      "grad_norm": 0.4714215397834778,
      "learning_rate": 9.589887358630998e-06,
      "loss": 0.4063,
      "step": 9625
    },
    {
      "epoch": 0.4417420035794594,
      "grad_norm": 0.47318416833877563,
      "learning_rate": 9.589790103613263e-06,
      "loss": 0.4342,
      "step": 9626
    },
    {
      "epoch": 0.44178789408471386,
      "grad_norm": 0.4896180033683777,
      "learning_rate": 9.589692837558534e-06,
      "loss": 0.4109,
      "step": 9627
    },
    {
      "epoch": 0.44183378458996836,
      "grad_norm": 0.4745769202709198,
      "learning_rate": 9.589595560467037e-06,
      "loss": 0.4208,
      "step": 9628
    },
    {
      "epoch": 0.4418796750952228,
      "grad_norm": 0.44588232040405273,
      "learning_rate": 9.589498272339016e-06,
      "loss": 0.3602,
      "step": 9629
    },
    {
      "epoch": 0.44192556560047724,
      "grad_norm": 0.46436935663223267,
      "learning_rate": 9.589400973174698e-06,
      "loss": 0.4071,
      "step": 9630
    },
    {
      "epoch": 0.44197145610573174,
      "grad_norm": 0.48432761430740356,
      "learning_rate": 9.589303662974319e-06,
      "loss": 0.3974,
      "step": 9631
    },
    {
      "epoch": 0.4420173466109862,
      "grad_norm": 0.5417534112930298,
      "learning_rate": 9.589206341738113e-06,
      "loss": 0.5975,
      "step": 9632
    },
    {
      "epoch": 0.4420632371162406,
      "grad_norm": 0.45078712701797485,
      "learning_rate": 9.589109009466316e-06,
      "loss": 0.4189,
      "step": 9633
    },
    {
      "epoch": 0.4421091276214951,
      "grad_norm": 0.45233264565467834,
      "learning_rate": 9.589011666159159e-06,
      "loss": 0.4251,
      "step": 9634
    },
    {
      "epoch": 0.44215501812674957,
      "grad_norm": 0.521564781665802,
      "learning_rate": 9.588914311816877e-06,
      "loss": 0.4916,
      "step": 9635
    },
    {
      "epoch": 0.44220090863200406,
      "grad_norm": 0.5074800252914429,
      "learning_rate": 9.588816946439705e-06,
      "loss": 0.5138,
      "step": 9636
    },
    {
      "epoch": 0.4422467991372585,
      "grad_norm": 0.45381203293800354,
      "learning_rate": 9.588719570027875e-06,
      "loss": 0.3815,
      "step": 9637
    },
    {
      "epoch": 0.44229268964251295,
      "grad_norm": 0.480557918548584,
      "learning_rate": 9.588622182581625e-06,
      "loss": 0.4628,
      "step": 9638
    },
    {
      "epoch": 0.44233858014776745,
      "grad_norm": 0.477324903011322,
      "learning_rate": 9.588524784101185e-06,
      "loss": 0.4028,
      "step": 9639
    },
    {
      "epoch": 0.4423844706530219,
      "grad_norm": 0.4725293219089508,
      "learning_rate": 9.588427374586792e-06,
      "loss": 0.4111,
      "step": 9640
    },
    {
      "epoch": 0.44243036115827633,
      "grad_norm": 0.4444078803062439,
      "learning_rate": 9.588329954038679e-06,
      "loss": 0.3974,
      "step": 9641
    },
    {
      "epoch": 0.44247625166353083,
      "grad_norm": 0.45349982380867004,
      "learning_rate": 9.588232522457081e-06,
      "loss": 0.3895,
      "step": 9642
    },
    {
      "epoch": 0.44252214216878527,
      "grad_norm": 0.44622552394866943,
      "learning_rate": 9.588135079842232e-06,
      "loss": 0.4195,
      "step": 9643
    },
    {
      "epoch": 0.44256803267403977,
      "grad_norm": 0.4689968526363373,
      "learning_rate": 9.588037626194365e-06,
      "loss": 0.419,
      "step": 9644
    },
    {
      "epoch": 0.4426139231792942,
      "grad_norm": 0.46988803148269653,
      "learning_rate": 9.587940161513716e-06,
      "loss": 0.3757,
      "step": 9645
    },
    {
      "epoch": 0.44265981368454865,
      "grad_norm": 0.4673752188682556,
      "learning_rate": 9.587842685800519e-06,
      "loss": 0.4555,
      "step": 9646
    },
    {
      "epoch": 0.44270570418980315,
      "grad_norm": 0.43658646941185,
      "learning_rate": 9.587745199055009e-06,
      "loss": 0.3212,
      "step": 9647
    },
    {
      "epoch": 0.4427515946950576,
      "grad_norm": 0.39917293190956116,
      "learning_rate": 9.587647701277418e-06,
      "loss": 0.2871,
      "step": 9648
    },
    {
      "epoch": 0.44279748520031204,
      "grad_norm": 0.4589487314224243,
      "learning_rate": 9.587550192467984e-06,
      "loss": 0.4084,
      "step": 9649
    },
    {
      "epoch": 0.44284337570556653,
      "grad_norm": 0.45076534152030945,
      "learning_rate": 9.58745267262694e-06,
      "loss": 0.3552,
      "step": 9650
    },
    {
      "epoch": 0.442889266210821,
      "grad_norm": 0.5135691165924072,
      "learning_rate": 9.587355141754519e-06,
      "loss": 0.5182,
      "step": 9651
    },
    {
      "epoch": 0.4429351567160754,
      "grad_norm": 0.47803282737731934,
      "learning_rate": 9.587257599850957e-06,
      "loss": 0.4357,
      "step": 9652
    },
    {
      "epoch": 0.4429810472213299,
      "grad_norm": 0.441232293844223,
      "learning_rate": 9.587160046916488e-06,
      "loss": 0.3583,
      "step": 9653
    },
    {
      "epoch": 0.44302693772658436,
      "grad_norm": 0.4099386930465698,
      "learning_rate": 9.587062482951345e-06,
      "loss": 0.2851,
      "step": 9654
    },
    {
      "epoch": 0.44307282823183886,
      "grad_norm": 0.42625892162323,
      "learning_rate": 9.586964907955768e-06,
      "loss": 0.3393,
      "step": 9655
    },
    {
      "epoch": 0.4431187187370933,
      "grad_norm": 0.5025972127914429,
      "learning_rate": 9.586867321929987e-06,
      "loss": 0.4622,
      "step": 9656
    },
    {
      "epoch": 0.44316460924234774,
      "grad_norm": 0.4824918210506439,
      "learning_rate": 9.586769724874236e-06,
      "loss": 0.4678,
      "step": 9657
    },
    {
      "epoch": 0.44321049974760224,
      "grad_norm": 0.4923332631587982,
      "learning_rate": 9.586672116788753e-06,
      "loss": 0.4995,
      "step": 9658
    },
    {
      "epoch": 0.4432563902528567,
      "grad_norm": 0.4466186463832855,
      "learning_rate": 9.58657449767377e-06,
      "loss": 0.3339,
      "step": 9659
    },
    {
      "epoch": 0.4433022807581111,
      "grad_norm": 0.4873707592487335,
      "learning_rate": 9.586476867529525e-06,
      "loss": 0.4248,
      "step": 9660
    },
    {
      "epoch": 0.4433481712633656,
      "grad_norm": 0.5259816646575928,
      "learning_rate": 9.58637922635625e-06,
      "loss": 0.4431,
      "step": 9661
    },
    {
      "epoch": 0.44339406176862006,
      "grad_norm": 0.42020919919013977,
      "learning_rate": 9.58628157415418e-06,
      "loss": 0.372,
      "step": 9662
    },
    {
      "epoch": 0.44343995227387456,
      "grad_norm": 0.4119405448436737,
      "learning_rate": 9.58618391092355e-06,
      "loss": 0.3331,
      "step": 9663
    },
    {
      "epoch": 0.443485842779129,
      "grad_norm": 0.47012048959732056,
      "learning_rate": 9.586086236664595e-06,
      "loss": 0.4227,
      "step": 9664
    },
    {
      "epoch": 0.44353173328438344,
      "grad_norm": 0.8761385679244995,
      "learning_rate": 9.58598855137755e-06,
      "loss": 0.4267,
      "step": 9665
    },
    {
      "epoch": 0.44357762378963794,
      "grad_norm": 0.49415332078933716,
      "learning_rate": 9.58589085506265e-06,
      "loss": 0.4361,
      "step": 9666
    },
    {
      "epoch": 0.4436235142948924,
      "grad_norm": 0.4912410080432892,
      "learning_rate": 9.58579314772013e-06,
      "loss": 0.3832,
      "step": 9667
    },
    {
      "epoch": 0.4436694048001468,
      "grad_norm": 0.46939942240715027,
      "learning_rate": 9.585695429350225e-06,
      "loss": 0.38,
      "step": 9668
    },
    {
      "epoch": 0.4437152953054013,
      "grad_norm": 0.5038236379623413,
      "learning_rate": 9.585597699953169e-06,
      "loss": 0.4197,
      "step": 9669
    },
    {
      "epoch": 0.44376118581065577,
      "grad_norm": 0.4478367865085602,
      "learning_rate": 9.5854999595292e-06,
      "loss": 0.3755,
      "step": 9670
    },
    {
      "epoch": 0.44380707631591026,
      "grad_norm": 0.44914358854293823,
      "learning_rate": 9.585402208078548e-06,
      "loss": 0.4188,
      "step": 9671
    },
    {
      "epoch": 0.4438529668211647,
      "grad_norm": 0.5287940502166748,
      "learning_rate": 9.585304445601454e-06,
      "loss": 0.3888,
      "step": 9672
    },
    {
      "epoch": 0.44389885732641915,
      "grad_norm": 0.46928712725639343,
      "learning_rate": 9.585206672098145e-06,
      "loss": 0.363,
      "step": 9673
    },
    {
      "epoch": 0.44394474783167365,
      "grad_norm": 0.49741995334625244,
      "learning_rate": 9.585108887568865e-06,
      "loss": 0.4185,
      "step": 9674
    },
    {
      "epoch": 0.4439906383369281,
      "grad_norm": 0.4508717358112335,
      "learning_rate": 9.585011092013845e-06,
      "loss": 0.3754,
      "step": 9675
    },
    {
      "epoch": 0.44403652884218253,
      "grad_norm": 0.5743631720542908,
      "learning_rate": 9.58491328543332e-06,
      "loss": 0.3973,
      "step": 9676
    },
    {
      "epoch": 0.44408241934743703,
      "grad_norm": 0.4918135106563568,
      "learning_rate": 9.584815467827525e-06,
      "loss": 0.42,
      "step": 9677
    },
    {
      "epoch": 0.44412830985269147,
      "grad_norm": 0.4639934003353119,
      "learning_rate": 9.584717639196696e-06,
      "loss": 0.389,
      "step": 9678
    },
    {
      "epoch": 0.4441742003579459,
      "grad_norm": 0.4706595838069916,
      "learning_rate": 9.584619799541069e-06,
      "loss": 0.4092,
      "step": 9679
    },
    {
      "epoch": 0.4442200908632004,
      "grad_norm": 0.4610098898410797,
      "learning_rate": 9.584521948860878e-06,
      "loss": 0.4672,
      "step": 9680
    },
    {
      "epoch": 0.44426598136845485,
      "grad_norm": 0.43970662355422974,
      "learning_rate": 9.584424087156358e-06,
      "loss": 0.3839,
      "step": 9681
    },
    {
      "epoch": 0.44431187187370935,
      "grad_norm": 0.4475744664669037,
      "learning_rate": 9.584326214427744e-06,
      "loss": 0.3711,
      "step": 9682
    },
    {
      "epoch": 0.4443577623789638,
      "grad_norm": 0.41338759660720825,
      "learning_rate": 9.584228330675273e-06,
      "loss": 0.4046,
      "step": 9683
    },
    {
      "epoch": 0.44440365288421824,
      "grad_norm": 0.4646712839603424,
      "learning_rate": 9.584130435899181e-06,
      "loss": 0.3965,
      "step": 9684
    },
    {
      "epoch": 0.44444954338947273,
      "grad_norm": 0.4934080243110657,
      "learning_rate": 9.584032530099701e-06,
      "loss": 0.4669,
      "step": 9685
    },
    {
      "epoch": 0.4444954338947272,
      "grad_norm": 0.5081644058227539,
      "learning_rate": 9.583934613277071e-06,
      "loss": 0.49,
      "step": 9686
    },
    {
      "epoch": 0.4445413243999816,
      "grad_norm": 0.4256170988082886,
      "learning_rate": 9.583836685431524e-06,
      "loss": 0.3619,
      "step": 9687
    },
    {
      "epoch": 0.4445872149052361,
      "grad_norm": 0.4500099718570709,
      "learning_rate": 9.583738746563296e-06,
      "loss": 0.322,
      "step": 9688
    },
    {
      "epoch": 0.44463310541049056,
      "grad_norm": 0.4437118172645569,
      "learning_rate": 9.583640796672622e-06,
      "loss": 0.4048,
      "step": 9689
    },
    {
      "epoch": 0.44467899591574506,
      "grad_norm": 0.436017245054245,
      "learning_rate": 9.58354283575974e-06,
      "loss": 0.3937,
      "step": 9690
    },
    {
      "epoch": 0.4447248864209995,
      "grad_norm": 0.4234682023525238,
      "learning_rate": 9.583444863824885e-06,
      "loss": 0.3169,
      "step": 9691
    },
    {
      "epoch": 0.44477077692625394,
      "grad_norm": 0.45163267850875854,
      "learning_rate": 9.58334688086829e-06,
      "loss": 0.3767,
      "step": 9692
    },
    {
      "epoch": 0.44481666743150844,
      "grad_norm": 0.4595990478992462,
      "learning_rate": 9.583248886890192e-06,
      "loss": 0.3929,
      "step": 9693
    },
    {
      "epoch": 0.4448625579367629,
      "grad_norm": 0.4723793864250183,
      "learning_rate": 9.583150881890828e-06,
      "loss": 0.4233,
      "step": 9694
    },
    {
      "epoch": 0.4449084484420173,
      "grad_norm": 0.4567941725254059,
      "learning_rate": 9.583052865870433e-06,
      "loss": 0.3847,
      "step": 9695
    },
    {
      "epoch": 0.4449543389472718,
      "grad_norm": 0.4242376685142517,
      "learning_rate": 9.582954838829242e-06,
      "loss": 0.3075,
      "step": 9696
    },
    {
      "epoch": 0.44500022945252626,
      "grad_norm": 0.45160433650016785,
      "learning_rate": 9.58285680076749e-06,
      "loss": 0.3498,
      "step": 9697
    },
    {
      "epoch": 0.44504611995778076,
      "grad_norm": 0.42402368783950806,
      "learning_rate": 9.582758751685416e-06,
      "loss": 0.3124,
      "step": 9698
    },
    {
      "epoch": 0.4450920104630352,
      "grad_norm": 0.4305524528026581,
      "learning_rate": 9.582660691583251e-06,
      "loss": 0.3335,
      "step": 9699
    },
    {
      "epoch": 0.44513790096828965,
      "grad_norm": 0.4245362877845764,
      "learning_rate": 9.582562620461234e-06,
      "loss": 0.322,
      "step": 9700
    },
    {
      "epoch": 0.44518379147354414,
      "grad_norm": 0.45082196593284607,
      "learning_rate": 9.5824645383196e-06,
      "loss": 0.3475,
      "step": 9701
    },
    {
      "epoch": 0.4452296819787986,
      "grad_norm": 0.44884252548217773,
      "learning_rate": 9.582366445158586e-06,
      "loss": 0.3881,
      "step": 9702
    },
    {
      "epoch": 0.44527557248405303,
      "grad_norm": 0.42796602845191956,
      "learning_rate": 9.582268340978426e-06,
      "loss": 0.364,
      "step": 9703
    },
    {
      "epoch": 0.4453214629893075,
      "grad_norm": 0.5100491642951965,
      "learning_rate": 9.582170225779359e-06,
      "loss": 0.5006,
      "step": 9704
    },
    {
      "epoch": 0.44536735349456197,
      "grad_norm": 0.4466865360736847,
      "learning_rate": 9.582072099561617e-06,
      "loss": 0.313,
      "step": 9705
    },
    {
      "epoch": 0.4454132439998164,
      "grad_norm": 0.4920097589492798,
      "learning_rate": 9.581973962325438e-06,
      "loss": 0.423,
      "step": 9706
    },
    {
      "epoch": 0.4454591345050709,
      "grad_norm": 0.48109546303749084,
      "learning_rate": 9.581875814071057e-06,
      "loss": 0.4311,
      "step": 9707
    },
    {
      "epoch": 0.44550502501032535,
      "grad_norm": 0.4446703791618347,
      "learning_rate": 9.581777654798712e-06,
      "loss": 0.4136,
      "step": 9708
    },
    {
      "epoch": 0.44555091551557985,
      "grad_norm": 0.4389550983905792,
      "learning_rate": 9.581679484508637e-06,
      "loss": 0.401,
      "step": 9709
    },
    {
      "epoch": 0.4455968060208343,
      "grad_norm": 0.8718675971031189,
      "learning_rate": 9.581581303201069e-06,
      "loss": 0.5176,
      "step": 9710
    },
    {
      "epoch": 0.44564269652608873,
      "grad_norm": 0.4206651747226715,
      "learning_rate": 9.581483110876243e-06,
      "loss": 0.2982,
      "step": 9711
    },
    {
      "epoch": 0.44568858703134323,
      "grad_norm": 0.4238147437572479,
      "learning_rate": 9.581384907534399e-06,
      "loss": 0.3167,
      "step": 9712
    },
    {
      "epoch": 0.4457344775365977,
      "grad_norm": 0.4877219796180725,
      "learning_rate": 9.58128669317577e-06,
      "loss": 0.4763,
      "step": 9713
    },
    {
      "epoch": 0.4457803680418521,
      "grad_norm": 0.4501168131828308,
      "learning_rate": 9.581188467800589e-06,
      "loss": 0.393,
      "step": 9714
    },
    {
      "epoch": 0.4458262585471066,
      "grad_norm": 0.4622237980365753,
      "learning_rate": 9.581090231409098e-06,
      "loss": 0.3658,
      "step": 9715
    },
    {
      "epoch": 0.44587214905236106,
      "grad_norm": 0.45816364884376526,
      "learning_rate": 9.58099198400153e-06,
      "loss": 0.4164,
      "step": 9716
    },
    {
      "epoch": 0.44591803955761555,
      "grad_norm": 0.47169187664985657,
      "learning_rate": 9.580893725578122e-06,
      "loss": 0.3562,
      "step": 9717
    },
    {
      "epoch": 0.44596393006287,
      "grad_norm": 0.46834126114845276,
      "learning_rate": 9.58079545613911e-06,
      "loss": 0.4316,
      "step": 9718
    },
    {
      "epoch": 0.44600982056812444,
      "grad_norm": 0.50584876537323,
      "learning_rate": 9.580697175684733e-06,
      "loss": 0.4767,
      "step": 9719
    },
    {
      "epoch": 0.44605571107337894,
      "grad_norm": 0.45500805974006653,
      "learning_rate": 9.580598884215223e-06,
      "loss": 0.3598,
      "step": 9720
    },
    {
      "epoch": 0.4461016015786334,
      "grad_norm": 0.46660587191581726,
      "learning_rate": 9.580500581730818e-06,
      "loss": 0.3811,
      "step": 9721
    },
    {
      "epoch": 0.4461474920838878,
      "grad_norm": 0.41970840096473694,
      "learning_rate": 9.580402268231756e-06,
      "loss": 0.3358,
      "step": 9722
    },
    {
      "epoch": 0.4461933825891423,
      "grad_norm": 0.43053188920021057,
      "learning_rate": 9.58030394371827e-06,
      "loss": 0.3876,
      "step": 9723
    },
    {
      "epoch": 0.44623927309439676,
      "grad_norm": 0.4564133286476135,
      "learning_rate": 9.5802056081906e-06,
      "loss": 0.4138,
      "step": 9724
    },
    {
      "epoch": 0.44628516359965126,
      "grad_norm": 0.45494577288627625,
      "learning_rate": 9.580107261648983e-06,
      "loss": 0.4176,
      "step": 9725
    },
    {
      "epoch": 0.4463310541049057,
      "grad_norm": 0.4483310580253601,
      "learning_rate": 9.580008904093652e-06,
      "loss": 0.3746,
      "step": 9726
    },
    {
      "epoch": 0.44637694461016014,
      "grad_norm": 0.46922314167022705,
      "learning_rate": 9.579910535524843e-06,
      "loss": 0.4402,
      "step": 9727
    },
    {
      "epoch": 0.44642283511541464,
      "grad_norm": 0.49520522356033325,
      "learning_rate": 9.579812155942798e-06,
      "loss": 0.4259,
      "step": 9728
    },
    {
      "epoch": 0.4464687256206691,
      "grad_norm": 0.4410932958126068,
      "learning_rate": 9.57971376534775e-06,
      "loss": 0.3668,
      "step": 9729
    },
    {
      "epoch": 0.4465146161259235,
      "grad_norm": 0.44762057065963745,
      "learning_rate": 9.579615363739934e-06,
      "loss": 0.3704,
      "step": 9730
    },
    {
      "epoch": 0.446560506631178,
      "grad_norm": 0.4395647943019867,
      "learning_rate": 9.579516951119587e-06,
      "loss": 0.3822,
      "step": 9731
    },
    {
      "epoch": 0.44660639713643246,
      "grad_norm": 0.4863123893737793,
      "learning_rate": 9.57941852748695e-06,
      "loss": 0.4338,
      "step": 9732
    },
    {
      "epoch": 0.44665228764168696,
      "grad_norm": 0.4585941731929779,
      "learning_rate": 9.579320092842254e-06,
      "loss": 0.4042,
      "step": 9733
    },
    {
      "epoch": 0.4466981781469414,
      "grad_norm": 0.5000231862068176,
      "learning_rate": 9.57922164718574e-06,
      "loss": 0.4862,
      "step": 9734
    },
    {
      "epoch": 0.44674406865219585,
      "grad_norm": 0.46738728880882263,
      "learning_rate": 9.579123190517645e-06,
      "loss": 0.3494,
      "step": 9735
    },
    {
      "epoch": 0.44678995915745034,
      "grad_norm": 0.5346416234970093,
      "learning_rate": 9.579024722838202e-06,
      "loss": 0.542,
      "step": 9736
    },
    {
      "epoch": 0.4468358496627048,
      "grad_norm": 0.43725869059562683,
      "learning_rate": 9.57892624414765e-06,
      "loss": 0.3412,
      "step": 9737
    },
    {
      "epoch": 0.44688174016795923,
      "grad_norm": 0.466680109500885,
      "learning_rate": 9.578827754446226e-06,
      "loss": 0.4109,
      "step": 9738
    },
    {
      "epoch": 0.4469276306732137,
      "grad_norm": 0.44553911685943604,
      "learning_rate": 9.578729253734165e-06,
      "loss": 0.3729,
      "step": 9739
    },
    {
      "epoch": 0.44697352117846817,
      "grad_norm": 0.4994473159313202,
      "learning_rate": 9.578630742011708e-06,
      "loss": 0.5378,
      "step": 9740
    },
    {
      "epoch": 0.4470194116837226,
      "grad_norm": 0.4441971480846405,
      "learning_rate": 9.578532219279087e-06,
      "loss": 0.3986,
      "step": 9741
    },
    {
      "epoch": 0.4470653021889771,
      "grad_norm": 0.46104422211647034,
      "learning_rate": 9.578433685536541e-06,
      "loss": 0.4017,
      "step": 9742
    },
    {
      "epoch": 0.44711119269423155,
      "grad_norm": 0.45616480708122253,
      "learning_rate": 9.578335140784308e-06,
      "loss": 0.4138,
      "step": 9743
    },
    {
      "epoch": 0.44715708319948605,
      "grad_norm": 0.43608972430229187,
      "learning_rate": 9.578236585022622e-06,
      "loss": 0.3771,
      "step": 9744
    },
    {
      "epoch": 0.4472029737047405,
      "grad_norm": 0.42397522926330566,
      "learning_rate": 9.578138018251723e-06,
      "loss": 0.3225,
      "step": 9745
    },
    {
      "epoch": 0.44724886420999493,
      "grad_norm": 0.4208241403102875,
      "learning_rate": 9.578039440471848e-06,
      "loss": 0.3517,
      "step": 9746
    },
    {
      "epoch": 0.44729475471524943,
      "grad_norm": 0.47919583320617676,
      "learning_rate": 9.577940851683232e-06,
      "loss": 0.4654,
      "step": 9747
    },
    {
      "epoch": 0.4473406452205039,
      "grad_norm": 0.40554603934288025,
      "learning_rate": 9.577842251886114e-06,
      "loss": 0.2959,
      "step": 9748
    },
    {
      "epoch": 0.4473865357257583,
      "grad_norm": 0.4671021103858948,
      "learning_rate": 9.577743641080729e-06,
      "loss": 0.4055,
      "step": 9749
    },
    {
      "epoch": 0.4474324262310128,
      "grad_norm": 0.4165404438972473,
      "learning_rate": 9.577645019267315e-06,
      "loss": 0.3016,
      "step": 9750
    },
    {
      "epoch": 0.44747831673626726,
      "grad_norm": 0.42552512884140015,
      "learning_rate": 9.57754638644611e-06,
      "loss": 0.3225,
      "step": 9751
    },
    {
      "epoch": 0.44752420724152175,
      "grad_norm": 0.4466932415962219,
      "learning_rate": 9.577447742617352e-06,
      "loss": 0.3876,
      "step": 9752
    },
    {
      "epoch": 0.4475700977467762,
      "grad_norm": 0.49270591139793396,
      "learning_rate": 9.577349087781277e-06,
      "loss": 0.3693,
      "step": 9753
    },
    {
      "epoch": 0.44761598825203064,
      "grad_norm": 0.4785556495189667,
      "learning_rate": 9.57725042193812e-06,
      "loss": 0.3814,
      "step": 9754
    },
    {
      "epoch": 0.44766187875728514,
      "grad_norm": 0.5062205195426941,
      "learning_rate": 9.57715174508812e-06,
      "loss": 0.458,
      "step": 9755
    },
    {
      "epoch": 0.4477077692625396,
      "grad_norm": 0.4693039357662201,
      "learning_rate": 9.577053057231517e-06,
      "loss": 0.4285,
      "step": 9756
    },
    {
      "epoch": 0.447753659767794,
      "grad_norm": 0.45526188611984253,
      "learning_rate": 9.576954358368545e-06,
      "loss": 0.3891,
      "step": 9757
    },
    {
      "epoch": 0.4477995502730485,
      "grad_norm": 0.4523662328720093,
      "learning_rate": 9.576855648499441e-06,
      "loss": 0.394,
      "step": 9758
    },
    {
      "epoch": 0.44784544077830296,
      "grad_norm": 0.49156418442726135,
      "learning_rate": 9.576756927624445e-06,
      "loss": 0.437,
      "step": 9759
    },
    {
      "epoch": 0.44789133128355746,
      "grad_norm": 0.489103227853775,
      "learning_rate": 9.576658195743793e-06,
      "loss": 0.4246,
      "step": 9760
    },
    {
      "epoch": 0.4479372217888119,
      "grad_norm": 0.4653508961200714,
      "learning_rate": 9.576559452857724e-06,
      "loss": 0.4119,
      "step": 9761
    },
    {
      "epoch": 0.44798311229406634,
      "grad_norm": 0.4668499529361725,
      "learning_rate": 9.576460698966471e-06,
      "loss": 0.4023,
      "step": 9762
    },
    {
      "epoch": 0.44802900279932084,
      "grad_norm": 0.4694982171058655,
      "learning_rate": 9.576361934070275e-06,
      "loss": 0.4737,
      "step": 9763
    },
    {
      "epoch": 0.4480748933045753,
      "grad_norm": 0.5218654870986938,
      "learning_rate": 9.576263158169375e-06,
      "loss": 0.5114,
      "step": 9764
    },
    {
      "epoch": 0.4481207838098297,
      "grad_norm": 0.473928302526474,
      "learning_rate": 9.576164371264004e-06,
      "loss": 0.4333,
      "step": 9765
    },
    {
      "epoch": 0.4481666743150842,
      "grad_norm": 0.4562360644340515,
      "learning_rate": 9.576065573354402e-06,
      "loss": 0.4024,
      "step": 9766
    },
    {
      "epoch": 0.44821256482033867,
      "grad_norm": 0.47578755021095276,
      "learning_rate": 9.575966764440809e-06,
      "loss": 0.3949,
      "step": 9767
    },
    {
      "epoch": 0.4482584553255931,
      "grad_norm": 0.4578250050544739,
      "learning_rate": 9.575867944523459e-06,
      "loss": 0.4231,
      "step": 9768
    },
    {
      "epoch": 0.4483043458308476,
      "grad_norm": 0.4514824151992798,
      "learning_rate": 9.57576911360259e-06,
      "loss": 0.3135,
      "step": 9769
    },
    {
      "epoch": 0.44835023633610205,
      "grad_norm": 0.4930398166179657,
      "learning_rate": 9.57567027167844e-06,
      "loss": 0.4383,
      "step": 9770
    },
    {
      "epoch": 0.44839612684135655,
      "grad_norm": 0.44422447681427,
      "learning_rate": 9.575571418751248e-06,
      "loss": 0.336,
      "step": 9771
    },
    {
      "epoch": 0.448442017346611,
      "grad_norm": 0.49810612201690674,
      "learning_rate": 9.575472554821251e-06,
      "loss": 0.4321,
      "step": 9772
    },
    {
      "epoch": 0.44848790785186543,
      "grad_norm": 0.4803476333618164,
      "learning_rate": 9.575373679888687e-06,
      "loss": 0.4501,
      "step": 9773
    },
    {
      "epoch": 0.44853379835711993,
      "grad_norm": 0.5777461528778076,
      "learning_rate": 9.575274793953792e-06,
      "loss": 0.3456,
      "step": 9774
    },
    {
      "epoch": 0.44857968886237437,
      "grad_norm": 0.514886736869812,
      "learning_rate": 9.575175897016806e-06,
      "loss": 0.4582,
      "step": 9775
    },
    {
      "epoch": 0.4486255793676288,
      "grad_norm": 0.48452407121658325,
      "learning_rate": 9.575076989077965e-06,
      "loss": 0.4077,
      "step": 9776
    },
    {
      "epoch": 0.4486714698728833,
      "grad_norm": 0.467336505651474,
      "learning_rate": 9.57497807013751e-06,
      "loss": 0.3709,
      "step": 9777
    },
    {
      "epoch": 0.44871736037813775,
      "grad_norm": 0.4870755672454834,
      "learning_rate": 9.574879140195674e-06,
      "loss": 0.4272,
      "step": 9778
    },
    {
      "epoch": 0.44876325088339225,
      "grad_norm": 0.4655212163925171,
      "learning_rate": 9.574780199252699e-06,
      "loss": 0.407,
      "step": 9779
    },
    {
      "epoch": 0.4488091413886467,
      "grad_norm": 0.4765538275241852,
      "learning_rate": 9.574681247308821e-06,
      "loss": 0.3801,
      "step": 9780
    },
    {
      "epoch": 0.44885503189390114,
      "grad_norm": 0.5006573796272278,
      "learning_rate": 9.574582284364279e-06,
      "loss": 0.4318,
      "step": 9781
    },
    {
      "epoch": 0.44890092239915563,
      "grad_norm": 0.46182742714881897,
      "learning_rate": 9.57448331041931e-06,
      "loss": 0.3596,
      "step": 9782
    },
    {
      "epoch": 0.4489468129044101,
      "grad_norm": 0.5996143817901611,
      "learning_rate": 9.574384325474152e-06,
      "loss": 0.3796,
      "step": 9783
    },
    {
      "epoch": 0.4489927034096645,
      "grad_norm": 0.4312472343444824,
      "learning_rate": 9.574285329529044e-06,
      "loss": 0.3661,
      "step": 9784
    },
    {
      "epoch": 0.449038593914919,
      "grad_norm": 0.44791820645332336,
      "learning_rate": 9.574186322584223e-06,
      "loss": 0.3896,
      "step": 9785
    },
    {
      "epoch": 0.44908448442017346,
      "grad_norm": 0.4694032371044159,
      "learning_rate": 9.574087304639928e-06,
      "loss": 0.4215,
      "step": 9786
    },
    {
      "epoch": 0.44913037492542796,
      "grad_norm": 0.4973786473274231,
      "learning_rate": 9.573988275696396e-06,
      "loss": 0.5019,
      "step": 9787
    },
    {
      "epoch": 0.4491762654306824,
      "grad_norm": 0.5054702758789062,
      "learning_rate": 9.573889235753865e-06,
      "loss": 0.5471,
      "step": 9788
    },
    {
      "epoch": 0.44922215593593684,
      "grad_norm": 0.507432222366333,
      "learning_rate": 9.573790184812576e-06,
      "loss": 0.4276,
      "step": 9789
    },
    {
      "epoch": 0.44926804644119134,
      "grad_norm": 0.41484472155570984,
      "learning_rate": 9.573691122872766e-06,
      "loss": 0.3224,
      "step": 9790
    },
    {
      "epoch": 0.4493139369464458,
      "grad_norm": 0.46572345495224,
      "learning_rate": 9.57359204993467e-06,
      "loss": 0.4246,
      "step": 9791
    },
    {
      "epoch": 0.4493598274517002,
      "grad_norm": 0.47469544410705566,
      "learning_rate": 9.573492965998529e-06,
      "loss": 0.3825,
      "step": 9792
    },
    {
      "epoch": 0.4494057179569547,
      "grad_norm": 0.47324785590171814,
      "learning_rate": 9.573393871064581e-06,
      "loss": 0.3947,
      "step": 9793
    },
    {
      "epoch": 0.44945160846220916,
      "grad_norm": 0.445686936378479,
      "learning_rate": 9.573294765133064e-06,
      "loss": 0.4203,
      "step": 9794
    },
    {
      "epoch": 0.4494974989674636,
      "grad_norm": 0.4478321671485901,
      "learning_rate": 9.573195648204216e-06,
      "loss": 0.3496,
      "step": 9795
    },
    {
      "epoch": 0.4495433894727181,
      "grad_norm": 0.4292794167995453,
      "learning_rate": 9.573096520278277e-06,
      "loss": 0.3274,
      "step": 9796
    },
    {
      "epoch": 0.44958927997797254,
      "grad_norm": 0.45690569281578064,
      "learning_rate": 9.572997381355485e-06,
      "loss": 0.3853,
      "step": 9797
    },
    {
      "epoch": 0.44963517048322704,
      "grad_norm": 0.5141229033470154,
      "learning_rate": 9.572898231436077e-06,
      "loss": 0.4877,
      "step": 9798
    },
    {
      "epoch": 0.4496810609884815,
      "grad_norm": 0.4356003701686859,
      "learning_rate": 9.57279907052029e-06,
      "loss": 0.3691,
      "step": 9799
    },
    {
      "epoch": 0.4497269514937359,
      "grad_norm": 0.4536591172218323,
      "learning_rate": 9.572699898608366e-06,
      "loss": 0.3818,
      "step": 9800
    },
    {
      "epoch": 0.4497728419989904,
      "grad_norm": 0.4678402543067932,
      "learning_rate": 9.572600715700542e-06,
      "loss": 0.446,
      "step": 9801
    },
    {
      "epoch": 0.44981873250424487,
      "grad_norm": 0.4600459635257721,
      "learning_rate": 9.572501521797056e-06,
      "loss": 0.4137,
      "step": 9802
    },
    {
      "epoch": 0.4498646230094993,
      "grad_norm": 0.4637428820133209,
      "learning_rate": 9.572402316898147e-06,
      "loss": 0.3978,
      "step": 9803
    },
    {
      "epoch": 0.4499105135147538,
      "grad_norm": 0.470689594745636,
      "learning_rate": 9.572303101004052e-06,
      "loss": 0.4394,
      "step": 9804
    },
    {
      "epoch": 0.44995640402000825,
      "grad_norm": 0.5019899010658264,
      "learning_rate": 9.572203874115014e-06,
      "loss": 0.5465,
      "step": 9805
    },
    {
      "epoch": 0.45000229452526275,
      "grad_norm": 0.47686177492141724,
      "learning_rate": 9.572104636231267e-06,
      "loss": 0.4644,
      "step": 9806
    },
    {
      "epoch": 0.4500481850305172,
      "grad_norm": 0.45432478189468384,
      "learning_rate": 9.572005387353052e-06,
      "loss": 0.3873,
      "step": 9807
    },
    {
      "epoch": 0.45009407553577163,
      "grad_norm": 0.4617711901664734,
      "learning_rate": 9.571906127480605e-06,
      "loss": 0.3755,
      "step": 9808
    },
    {
      "epoch": 0.45013996604102613,
      "grad_norm": 0.4422835111618042,
      "learning_rate": 9.571806856614169e-06,
      "loss": 0.3231,
      "step": 9809
    },
    {
      "epoch": 0.45018585654628057,
      "grad_norm": 0.47981172800064087,
      "learning_rate": 9.57170757475398e-06,
      "loss": 0.4628,
      "step": 9810
    },
    {
      "epoch": 0.450231747051535,
      "grad_norm": 0.43930914998054504,
      "learning_rate": 9.571608281900278e-06,
      "loss": 0.3544,
      "step": 9811
    },
    {
      "epoch": 0.4502776375567895,
      "grad_norm": 0.4524284899234772,
      "learning_rate": 9.5715089780533e-06,
      "loss": 0.4185,
      "step": 9812
    },
    {
      "epoch": 0.45032352806204395,
      "grad_norm": 0.46251779794692993,
      "learning_rate": 9.571409663213285e-06,
      "loss": 0.4108,
      "step": 9813
    },
    {
      "epoch": 0.45036941856729845,
      "grad_norm": 0.4838874638080597,
      "learning_rate": 9.571310337380474e-06,
      "loss": 0.4312,
      "step": 9814
    },
    {
      "epoch": 0.4504153090725529,
      "grad_norm": 0.46627694368362427,
      "learning_rate": 9.571211000555102e-06,
      "loss": 0.4183,
      "step": 9815
    },
    {
      "epoch": 0.45046119957780734,
      "grad_norm": 0.4588378667831421,
      "learning_rate": 9.571111652737413e-06,
      "loss": 0.4412,
      "step": 9816
    },
    {
      "epoch": 0.45050709008306183,
      "grad_norm": 0.4505850076675415,
      "learning_rate": 9.571012293927642e-06,
      "loss": 0.408,
      "step": 9817
    },
    {
      "epoch": 0.4505529805883163,
      "grad_norm": 0.5060086250305176,
      "learning_rate": 9.570912924126028e-06,
      "loss": 0.5212,
      "step": 9818
    },
    {
      "epoch": 0.4505988710935707,
      "grad_norm": 0.45375683903694153,
      "learning_rate": 9.570813543332813e-06,
      "loss": 0.3957,
      "step": 9819
    },
    {
      "epoch": 0.4506447615988252,
      "grad_norm": 0.44547316431999207,
      "learning_rate": 9.570714151548233e-06,
      "loss": 0.3901,
      "step": 9820
    },
    {
      "epoch": 0.45069065210407966,
      "grad_norm": 0.46824169158935547,
      "learning_rate": 9.570614748772529e-06,
      "loss": 0.4406,
      "step": 9821
    },
    {
      "epoch": 0.45073654260933416,
      "grad_norm": 0.5276828408241272,
      "learning_rate": 9.570515335005937e-06,
      "loss": 0.4566,
      "step": 9822
    },
    {
      "epoch": 0.4507824331145886,
      "grad_norm": 0.45947471261024475,
      "learning_rate": 9.5704159102487e-06,
      "loss": 0.4443,
      "step": 9823
    },
    {
      "epoch": 0.45082832361984304,
      "grad_norm": 0.46542027592658997,
      "learning_rate": 9.570316474501055e-06,
      "loss": 0.398,
      "step": 9824
    },
    {
      "epoch": 0.45087421412509754,
      "grad_norm": 0.4361060857772827,
      "learning_rate": 9.57021702776324e-06,
      "loss": 0.3282,
      "step": 9825
    },
    {
      "epoch": 0.450920104630352,
      "grad_norm": 0.41409507393836975,
      "learning_rate": 9.570117570035496e-06,
      "loss": 0.3092,
      "step": 9826
    },
    {
      "epoch": 0.4509659951356064,
      "grad_norm": 0.44705623388290405,
      "learning_rate": 9.570018101318062e-06,
      "loss": 0.341,
      "step": 9827
    },
    {
      "epoch": 0.4510118856408609,
      "grad_norm": 0.46038907766342163,
      "learning_rate": 9.569918621611177e-06,
      "loss": 0.3685,
      "step": 9828
    },
    {
      "epoch": 0.45105777614611536,
      "grad_norm": 0.44764235615730286,
      "learning_rate": 9.56981913091508e-06,
      "loss": 0.3608,
      "step": 9829
    },
    {
      "epoch": 0.4511036666513698,
      "grad_norm": 0.5124440789222717,
      "learning_rate": 9.56971962923001e-06,
      "loss": 0.4526,
      "step": 9830
    },
    {
      "epoch": 0.4511495571566243,
      "grad_norm": 0.49712926149368286,
      "learning_rate": 9.569620116556206e-06,
      "loss": 0.4519,
      "step": 9831
    },
    {
      "epoch": 0.45119544766187875,
      "grad_norm": 0.42241939902305603,
      "learning_rate": 9.569520592893908e-06,
      "loss": 0.3309,
      "step": 9832
    },
    {
      "epoch": 0.45124133816713324,
      "grad_norm": 0.4900880455970764,
      "learning_rate": 9.569421058243355e-06,
      "loss": 0.4743,
      "step": 9833
    },
    {
      "epoch": 0.4512872286723877,
      "grad_norm": 0.46104082465171814,
      "learning_rate": 9.569321512604788e-06,
      "loss": 0.3815,
      "step": 9834
    },
    {
      "epoch": 0.45133311917764213,
      "grad_norm": 0.4504721462726593,
      "learning_rate": 9.569221955978444e-06,
      "loss": 0.3956,
      "step": 9835
    },
    {
      "epoch": 0.4513790096828966,
      "grad_norm": 0.4503343105316162,
      "learning_rate": 9.569122388364564e-06,
      "loss": 0.3717,
      "step": 9836
    },
    {
      "epoch": 0.45142490018815107,
      "grad_norm": 0.4914402365684509,
      "learning_rate": 9.569022809763385e-06,
      "loss": 0.475,
      "step": 9837
    },
    {
      "epoch": 0.4514707906934055,
      "grad_norm": 0.4153030812740326,
      "learning_rate": 9.56892322017515e-06,
      "loss": 0.3246,
      "step": 9838
    },
    {
      "epoch": 0.45151668119866,
      "grad_norm": 0.4414408504962921,
      "learning_rate": 9.568823619600095e-06,
      "loss": 0.3837,
      "step": 9839
    },
    {
      "epoch": 0.45156257170391445,
      "grad_norm": 0.4685708284378052,
      "learning_rate": 9.56872400803846e-06,
      "loss": 0.3944,
      "step": 9840
    },
    {
      "epoch": 0.45160846220916895,
      "grad_norm": 0.5370137691497803,
      "learning_rate": 9.56862438549049e-06,
      "loss": 0.4182,
      "step": 9841
    },
    {
      "epoch": 0.4516543527144234,
      "grad_norm": 0.42223286628723145,
      "learning_rate": 9.568524751956417e-06,
      "loss": 0.3517,
      "step": 9842
    },
    {
      "epoch": 0.45170024321967783,
      "grad_norm": 0.45244261622428894,
      "learning_rate": 9.568425107436485e-06,
      "loss": 0.427,
      "step": 9843
    },
    {
      "epoch": 0.45174613372493233,
      "grad_norm": 0.4518932104110718,
      "learning_rate": 9.568325451930932e-06,
      "loss": 0.3763,
      "step": 9844
    },
    {
      "epoch": 0.4517920242301868,
      "grad_norm": 0.46881550550460815,
      "learning_rate": 9.568225785439999e-06,
      "loss": 0.439,
      "step": 9845
    },
    {
      "epoch": 0.4518379147354412,
      "grad_norm": 0.42553529143333435,
      "learning_rate": 9.568126107963924e-06,
      "loss": 0.3393,
      "step": 9846
    },
    {
      "epoch": 0.4518838052406957,
      "grad_norm": 0.4750100076198578,
      "learning_rate": 9.568026419502947e-06,
      "loss": 0.478,
      "step": 9847
    },
    {
      "epoch": 0.45192969574595016,
      "grad_norm": 0.48248741030693054,
      "learning_rate": 9.567926720057308e-06,
      "loss": 0.4431,
      "step": 9848
    },
    {
      "epoch": 0.45197558625120465,
      "grad_norm": 0.45480674505233765,
      "learning_rate": 9.567827009627248e-06,
      "loss": 0.3805,
      "step": 9849
    },
    {
      "epoch": 0.4520214767564591,
      "grad_norm": 0.4876725673675537,
      "learning_rate": 9.567727288213005e-06,
      "loss": 0.4161,
      "step": 9850
    },
    {
      "epoch": 0.45206736726171354,
      "grad_norm": 0.47479158639907837,
      "learning_rate": 9.567627555814821e-06,
      "loss": 0.394,
      "step": 9851
    },
    {
      "epoch": 0.45211325776696804,
      "grad_norm": 0.4589460790157318,
      "learning_rate": 9.567527812432932e-06,
      "loss": 0.3451,
      "step": 9852
    },
    {
      "epoch": 0.4521591482722225,
      "grad_norm": 0.46952447295188904,
      "learning_rate": 9.56742805806758e-06,
      "loss": 0.4182,
      "step": 9853
    },
    {
      "epoch": 0.4522050387774769,
      "grad_norm": 0.4585420489311218,
      "learning_rate": 9.567328292719007e-06,
      "loss": 0.3487,
      "step": 9854
    },
    {
      "epoch": 0.4522509292827314,
      "grad_norm": 0.4362148642539978,
      "learning_rate": 9.56722851638745e-06,
      "loss": 0.3445,
      "step": 9855
    },
    {
      "epoch": 0.45229681978798586,
      "grad_norm": 0.4377683103084564,
      "learning_rate": 9.56712872907315e-06,
      "loss": 0.3838,
      "step": 9856
    },
    {
      "epoch": 0.4523427102932403,
      "grad_norm": 0.4936470687389374,
      "learning_rate": 9.567028930776347e-06,
      "loss": 0.4501,
      "step": 9857
    },
    {
      "epoch": 0.4523886007984948,
      "grad_norm": 0.46923792362213135,
      "learning_rate": 9.566929121497282e-06,
      "loss": 0.4205,
      "step": 9858
    },
    {
      "epoch": 0.45243449130374924,
      "grad_norm": 0.44804394245147705,
      "learning_rate": 9.56682930123619e-06,
      "loss": 0.3546,
      "step": 9859
    },
    {
      "epoch": 0.45248038180900374,
      "grad_norm": 0.481635183095932,
      "learning_rate": 9.56672946999332e-06,
      "loss": 0.455,
      "step": 9860
    },
    {
      "epoch": 0.4525262723142582,
      "grad_norm": 0.48465442657470703,
      "learning_rate": 9.566629627768904e-06,
      "loss": 0.4507,
      "step": 9861
    },
    {
      "epoch": 0.4525721628195126,
      "grad_norm": 0.46866318583488464,
      "learning_rate": 9.566529774563185e-06,
      "loss": 0.4042,
      "step": 9862
    },
    {
      "epoch": 0.4526180533247671,
      "grad_norm": 0.48285502195358276,
      "learning_rate": 9.566429910376402e-06,
      "loss": 0.4134,
      "step": 9863
    },
    {
      "epoch": 0.45266394383002156,
      "grad_norm": 0.483559787273407,
      "learning_rate": 9.566330035208799e-06,
      "loss": 0.4087,
      "step": 9864
    },
    {
      "epoch": 0.452709834335276,
      "grad_norm": 0.46345236897468567,
      "learning_rate": 9.566230149060611e-06,
      "loss": 0.3767,
      "step": 9865
    },
    {
      "epoch": 0.4527557248405305,
      "grad_norm": 0.4565286636352539,
      "learning_rate": 9.566130251932081e-06,
      "loss": 0.4095,
      "step": 9866
    },
    {
      "epoch": 0.45280161534578495,
      "grad_norm": 0.4623097777366638,
      "learning_rate": 9.566030343823452e-06,
      "loss": 0.3953,
      "step": 9867
    },
    {
      "epoch": 0.45284750585103944,
      "grad_norm": 0.4674161374568939,
      "learning_rate": 9.565930424734959e-06,
      "loss": 0.4443,
      "step": 9868
    },
    {
      "epoch": 0.4528933963562939,
      "grad_norm": 0.46204760670661926,
      "learning_rate": 9.565830494666843e-06,
      "loss": 0.3836,
      "step": 9869
    },
    {
      "epoch": 0.45293928686154833,
      "grad_norm": 0.49224013090133667,
      "learning_rate": 9.565730553619348e-06,
      "loss": 0.446,
      "step": 9870
    },
    {
      "epoch": 0.4529851773668028,
      "grad_norm": 0.47723156213760376,
      "learning_rate": 9.56563060159271e-06,
      "loss": 0.3997,
      "step": 9871
    },
    {
      "epoch": 0.45303106787205727,
      "grad_norm": 0.4757527709007263,
      "learning_rate": 9.565530638587172e-06,
      "loss": 0.3889,
      "step": 9872
    },
    {
      "epoch": 0.4530769583773117,
      "grad_norm": 0.5213149785995483,
      "learning_rate": 9.565430664602974e-06,
      "loss": 0.4369,
      "step": 9873
    },
    {
      "epoch": 0.4531228488825662,
      "grad_norm": 0.47971388697624207,
      "learning_rate": 9.565330679640357e-06,
      "loss": 0.4328,
      "step": 9874
    },
    {
      "epoch": 0.45316873938782065,
      "grad_norm": 0.4483637809753418,
      "learning_rate": 9.565230683699559e-06,
      "loss": 0.3169,
      "step": 9875
    },
    {
      "epoch": 0.45321462989307515,
      "grad_norm": 0.4834587872028351,
      "learning_rate": 9.565130676780822e-06,
      "loss": 0.4584,
      "step": 9876
    },
    {
      "epoch": 0.4532605203983296,
      "grad_norm": 0.4770866334438324,
      "learning_rate": 9.565030658884388e-06,
      "loss": 0.4218,
      "step": 9877
    },
    {
      "epoch": 0.45330641090358403,
      "grad_norm": 0.4676073491573334,
      "learning_rate": 9.564930630010495e-06,
      "loss": 0.4257,
      "step": 9878
    },
    {
      "epoch": 0.45335230140883853,
      "grad_norm": 0.5158021450042725,
      "learning_rate": 9.564830590159383e-06,
      "loss": 0.404,
      "step": 9879
    },
    {
      "epoch": 0.453398191914093,
      "grad_norm": 0.4540286660194397,
      "learning_rate": 9.564730539331297e-06,
      "loss": 0.3928,
      "step": 9880
    },
    {
      "epoch": 0.4534440824193474,
      "grad_norm": 0.42820119857788086,
      "learning_rate": 9.564630477526474e-06,
      "loss": 0.2926,
      "step": 9881
    },
    {
      "epoch": 0.4534899729246019,
      "grad_norm": 0.44111403822898865,
      "learning_rate": 9.564530404745156e-06,
      "loss": 0.3866,
      "step": 9882
    },
    {
      "epoch": 0.45353586342985636,
      "grad_norm": 0.4540165364742279,
      "learning_rate": 9.56443032098758e-06,
      "loss": 0.3712,
      "step": 9883
    },
    {
      "epoch": 0.4535817539351108,
      "grad_norm": 0.7026460766792297,
      "learning_rate": 9.564330226253992e-06,
      "loss": 0.3399,
      "step": 9884
    },
    {
      "epoch": 0.4536276444403653,
      "grad_norm": 0.4685457944869995,
      "learning_rate": 9.564230120544631e-06,
      "loss": 0.4083,
      "step": 9885
    },
    {
      "epoch": 0.45367353494561974,
      "grad_norm": 0.5072273015975952,
      "learning_rate": 9.564130003859735e-06,
      "loss": 0.4957,
      "step": 9886
    },
    {
      "epoch": 0.45371942545087424,
      "grad_norm": 0.4591566324234009,
      "learning_rate": 9.564029876199549e-06,
      "loss": 0.4106,
      "step": 9887
    },
    {
      "epoch": 0.4537653159561287,
      "grad_norm": 0.4739809036254883,
      "learning_rate": 9.563929737564311e-06,
      "loss": 0.496,
      "step": 9888
    },
    {
      "epoch": 0.4538112064613831,
      "grad_norm": 0.43732163310050964,
      "learning_rate": 9.56382958795426e-06,
      "loss": 0.4069,
      "step": 9889
    },
    {
      "epoch": 0.4538570969666376,
      "grad_norm": 0.49638670682907104,
      "learning_rate": 9.56372942736964e-06,
      "loss": 0.4804,
      "step": 9890
    },
    {
      "epoch": 0.45390298747189206,
      "grad_norm": 0.5043736100196838,
      "learning_rate": 9.563629255810693e-06,
      "loss": 0.481,
      "step": 9891
    },
    {
      "epoch": 0.4539488779771465,
      "grad_norm": 0.44765156507492065,
      "learning_rate": 9.563529073277659e-06,
      "loss": 0.3717,
      "step": 9892
    },
    {
      "epoch": 0.453994768482401,
      "grad_norm": 0.42711880803108215,
      "learning_rate": 9.563428879770775e-06,
      "loss": 0.3024,
      "step": 9893
    },
    {
      "epoch": 0.45404065898765544,
      "grad_norm": 0.4751996397972107,
      "learning_rate": 9.563328675290287e-06,
      "loss": 0.4168,
      "step": 9894
    },
    {
      "epoch": 0.45408654949290994,
      "grad_norm": 0.4565356969833374,
      "learning_rate": 9.563228459836432e-06,
      "loss": 0.3708,
      "step": 9895
    },
    {
      "epoch": 0.4541324399981644,
      "grad_norm": 0.4587240517139435,
      "learning_rate": 9.563128233409453e-06,
      "loss": 0.4059,
      "step": 9896
    },
    {
      "epoch": 0.4541783305034188,
      "grad_norm": 0.47915181517601013,
      "learning_rate": 9.56302799600959e-06,
      "loss": 0.4249,
      "step": 9897
    },
    {
      "epoch": 0.4542242210086733,
      "grad_norm": 0.5254428386688232,
      "learning_rate": 9.562927747637087e-06,
      "loss": 0.4322,
      "step": 9898
    },
    {
      "epoch": 0.45427011151392777,
      "grad_norm": 0.43688803911209106,
      "learning_rate": 9.562827488292181e-06,
      "loss": 0.3284,
      "step": 9899
    },
    {
      "epoch": 0.4543160020191822,
      "grad_norm": 0.5806544423103333,
      "learning_rate": 9.562727217975116e-06,
      "loss": 0.5325,
      "step": 9900
    },
    {
      "epoch": 0.4543618925244367,
      "grad_norm": 0.4851280152797699,
      "learning_rate": 9.562626936686133e-06,
      "loss": 0.4538,
      "step": 9901
    },
    {
      "epoch": 0.45440778302969115,
      "grad_norm": 0.4443572461605072,
      "learning_rate": 9.56252664442547e-06,
      "loss": 0.3763,
      "step": 9902
    },
    {
      "epoch": 0.45445367353494565,
      "grad_norm": 0.45183512568473816,
      "learning_rate": 9.562426341193372e-06,
      "loss": 0.3728,
      "step": 9903
    },
    {
      "epoch": 0.4544995640402001,
      "grad_norm": 0.47146010398864746,
      "learning_rate": 9.562326026990077e-06,
      "loss": 0.3879,
      "step": 9904
    },
    {
      "epoch": 0.45454545454545453,
      "grad_norm": 0.5086910128593445,
      "learning_rate": 9.56222570181583e-06,
      "loss": 0.4595,
      "step": 9905
    },
    {
      "epoch": 0.45459134505070903,
      "grad_norm": 0.4530327022075653,
      "learning_rate": 9.562125365670869e-06,
      "loss": 0.3851,
      "step": 9906
    },
    {
      "epoch": 0.45463723555596347,
      "grad_norm": 0.4583710730075836,
      "learning_rate": 9.562025018555437e-06,
      "loss": 0.4029,
      "step": 9907
    },
    {
      "epoch": 0.4546831260612179,
      "grad_norm": 0.48848941922187805,
      "learning_rate": 9.561924660469773e-06,
      "loss": 0.4501,
      "step": 9908
    },
    {
      "epoch": 0.4547290165664724,
      "grad_norm": 0.4487016201019287,
      "learning_rate": 9.561824291414121e-06,
      "loss": 0.3288,
      "step": 9909
    },
    {
      "epoch": 0.45477490707172685,
      "grad_norm": 0.47554028034210205,
      "learning_rate": 9.56172391138872e-06,
      "loss": 0.4681,
      "step": 9910
    },
    {
      "epoch": 0.4548207975769813,
      "grad_norm": 0.48326393961906433,
      "learning_rate": 9.561623520393815e-06,
      "loss": 0.3799,
      "step": 9911
    },
    {
      "epoch": 0.4548666880822358,
      "grad_norm": 0.4342602789402008,
      "learning_rate": 9.561523118429642e-06,
      "loss": 0.364,
      "step": 9912
    },
    {
      "epoch": 0.45491257858749024,
      "grad_norm": 0.4346505403518677,
      "learning_rate": 9.561422705496449e-06,
      "loss": 0.3384,
      "step": 9913
    },
    {
      "epoch": 0.45495846909274473,
      "grad_norm": 0.45098307728767395,
      "learning_rate": 9.561322281594473e-06,
      "loss": 0.352,
      "step": 9914
    },
    {
      "epoch": 0.4550043595979992,
      "grad_norm": 0.4688247740268707,
      "learning_rate": 9.561221846723956e-06,
      "loss": 0.3906,
      "step": 9915
    },
    {
      "epoch": 0.4550502501032536,
      "grad_norm": 0.46275565028190613,
      "learning_rate": 9.56112140088514e-06,
      "loss": 0.4409,
      "step": 9916
    },
    {
      "epoch": 0.4550961406085081,
      "grad_norm": 0.4721828103065491,
      "learning_rate": 9.561020944078266e-06,
      "loss": 0.3938,
      "step": 9917
    },
    {
      "epoch": 0.45514203111376256,
      "grad_norm": 0.42936527729034424,
      "learning_rate": 9.560920476303577e-06,
      "loss": 0.3459,
      "step": 9918
    },
    {
      "epoch": 0.455187921619017,
      "grad_norm": 0.48279470205307007,
      "learning_rate": 9.560819997561314e-06,
      "loss": 0.4704,
      "step": 9919
    },
    {
      "epoch": 0.4552338121242715,
      "grad_norm": 0.47656986117362976,
      "learning_rate": 9.560719507851717e-06,
      "loss": 0.4162,
      "step": 9920
    },
    {
      "epoch": 0.45527970262952594,
      "grad_norm": 0.43818405270576477,
      "learning_rate": 9.560619007175031e-06,
      "loss": 0.3781,
      "step": 9921
    },
    {
      "epoch": 0.45532559313478044,
      "grad_norm": 0.4752810597419739,
      "learning_rate": 9.560518495531495e-06,
      "loss": 0.4059,
      "step": 9922
    },
    {
      "epoch": 0.4553714836400349,
      "grad_norm": 0.47806331515312195,
      "learning_rate": 9.56041797292135e-06,
      "loss": 0.3955,
      "step": 9923
    },
    {
      "epoch": 0.4554173741452893,
      "grad_norm": 0.4881223142147064,
      "learning_rate": 9.56031743934484e-06,
      "loss": 0.4556,
      "step": 9924
    },
    {
      "epoch": 0.4554632646505438,
      "grad_norm": 0.8381417989730835,
      "learning_rate": 9.560216894802206e-06,
      "loss": 0.4527,
      "step": 9925
    },
    {
      "epoch": 0.45550915515579826,
      "grad_norm": 0.48614510893821716,
      "learning_rate": 9.56011633929369e-06,
      "loss": 0.438,
      "step": 9926
    },
    {
      "epoch": 0.4555550456610527,
      "grad_norm": 0.45397496223449707,
      "learning_rate": 9.560015772819533e-06,
      "loss": 0.3989,
      "step": 9927
    },
    {
      "epoch": 0.4556009361663072,
      "grad_norm": 0.4825592339038849,
      "learning_rate": 9.559915195379976e-06,
      "loss": 0.4072,
      "step": 9928
    },
    {
      "epoch": 0.45564682667156164,
      "grad_norm": 0.4833358824253082,
      "learning_rate": 9.559814606975265e-06,
      "loss": 0.4568,
      "step": 9929
    },
    {
      "epoch": 0.45569271717681614,
      "grad_norm": 0.49016672372817993,
      "learning_rate": 9.559714007605637e-06,
      "loss": 0.4164,
      "step": 9930
    },
    {
      "epoch": 0.4557386076820706,
      "grad_norm": 0.48718762397766113,
      "learning_rate": 9.559613397271335e-06,
      "loss": 0.4291,
      "step": 9931
    },
    {
      "epoch": 0.455784498187325,
      "grad_norm": 0.538737416267395,
      "learning_rate": 9.559512775972604e-06,
      "loss": 0.589,
      "step": 9932
    },
    {
      "epoch": 0.4558303886925795,
      "grad_norm": 0.5791155099868774,
      "learning_rate": 9.559412143709681e-06,
      "loss": 0.5949,
      "step": 9933
    },
    {
      "epoch": 0.45587627919783397,
      "grad_norm": 0.48282018303871155,
      "learning_rate": 9.559311500482814e-06,
      "loss": 0.4559,
      "step": 9934
    },
    {
      "epoch": 0.4559221697030884,
      "grad_norm": 0.486346572637558,
      "learning_rate": 9.55921084629224e-06,
      "loss": 0.3804,
      "step": 9935
    },
    {
      "epoch": 0.4559680602083429,
      "grad_norm": 0.468128502368927,
      "learning_rate": 9.559110181138203e-06,
      "loss": 0.4217,
      "step": 9936
    },
    {
      "epoch": 0.45601395071359735,
      "grad_norm": 0.45098164677619934,
      "learning_rate": 9.559009505020944e-06,
      "loss": 0.3715,
      "step": 9937
    },
    {
      "epoch": 0.45605984121885185,
      "grad_norm": 0.4745773673057556,
      "learning_rate": 9.558908817940707e-06,
      "loss": 0.4245,
      "step": 9938
    },
    {
      "epoch": 0.4561057317241063,
      "grad_norm": 0.4722738564014435,
      "learning_rate": 9.558808119897733e-06,
      "loss": 0.4258,
      "step": 9939
    },
    {
      "epoch": 0.45615162222936073,
      "grad_norm": 0.4512943625450134,
      "learning_rate": 9.558707410892265e-06,
      "loss": 0.3406,
      "step": 9940
    },
    {
      "epoch": 0.45619751273461523,
      "grad_norm": 0.4353412687778473,
      "learning_rate": 9.558606690924543e-06,
      "loss": 0.354,
      "step": 9941
    },
    {
      "epoch": 0.45624340323986967,
      "grad_norm": 0.4160791039466858,
      "learning_rate": 9.558505959994813e-06,
      "loss": 0.2931,
      "step": 9942
    },
    {
      "epoch": 0.4562892937451241,
      "grad_norm": 0.45235979557037354,
      "learning_rate": 9.558405218103313e-06,
      "loss": 0.4079,
      "step": 9943
    },
    {
      "epoch": 0.4563351842503786,
      "grad_norm": 0.4443121552467346,
      "learning_rate": 9.558304465250287e-06,
      "loss": 0.3518,
      "step": 9944
    },
    {
      "epoch": 0.45638107475563305,
      "grad_norm": 0.519679844379425,
      "learning_rate": 9.558203701435979e-06,
      "loss": 0.4479,
      "step": 9945
    },
    {
      "epoch": 0.4564269652608875,
      "grad_norm": 0.46667245030403137,
      "learning_rate": 9.558102926660629e-06,
      "loss": 0.3999,
      "step": 9946
    },
    {
      "epoch": 0.456472855766142,
      "grad_norm": 0.4934587776660919,
      "learning_rate": 9.55800214092448e-06,
      "loss": 0.4411,
      "step": 9947
    },
    {
      "epoch": 0.45651874627139644,
      "grad_norm": 0.5010746121406555,
      "learning_rate": 9.557901344227774e-06,
      "loss": 0.5098,
      "step": 9948
    },
    {
      "epoch": 0.45656463677665093,
      "grad_norm": 0.46531081199645996,
      "learning_rate": 9.557800536570754e-06,
      "loss": 0.452,
      "step": 9949
    },
    {
      "epoch": 0.4566105272819054,
      "grad_norm": 0.4719448983669281,
      "learning_rate": 9.557699717953661e-06,
      "loss": 0.4586,
      "step": 9950
    },
    {
      "epoch": 0.4566564177871598,
      "grad_norm": 0.5225871801376343,
      "learning_rate": 9.55759888837674e-06,
      "loss": 0.4823,
      "step": 9951
    },
    {
      "epoch": 0.4567023082924143,
      "grad_norm": 0.4800194203853607,
      "learning_rate": 9.557498047840232e-06,
      "loss": 0.4404,
      "step": 9952
    },
    {
      "epoch": 0.45674819879766876,
      "grad_norm": 0.45608821511268616,
      "learning_rate": 9.55739719634438e-06,
      "loss": 0.4175,
      "step": 9953
    },
    {
      "epoch": 0.4567940893029232,
      "grad_norm": 0.45326918363571167,
      "learning_rate": 9.557296333889425e-06,
      "loss": 0.3586,
      "step": 9954
    },
    {
      "epoch": 0.4568399798081777,
      "grad_norm": 0.4494493901729584,
      "learning_rate": 9.55719546047561e-06,
      "loss": 0.3591,
      "step": 9955
    },
    {
      "epoch": 0.45688587031343214,
      "grad_norm": 0.4518440365791321,
      "learning_rate": 9.557094576103181e-06,
      "loss": 0.4096,
      "step": 9956
    },
    {
      "epoch": 0.45693176081868664,
      "grad_norm": 0.520563006401062,
      "learning_rate": 9.556993680772375e-06,
      "loss": 0.5272,
      "step": 9957
    },
    {
      "epoch": 0.4569776513239411,
      "grad_norm": 0.507922887802124,
      "learning_rate": 9.55689277448344e-06,
      "loss": 0.4813,
      "step": 9958
    },
    {
      "epoch": 0.4570235418291955,
      "grad_norm": 0.4522140622138977,
      "learning_rate": 9.556791857236613e-06,
      "loss": 0.4013,
      "step": 9959
    },
    {
      "epoch": 0.45706943233445,
      "grad_norm": 0.4685266315937042,
      "learning_rate": 9.556690929032142e-06,
      "loss": 0.4915,
      "step": 9960
    },
    {
      "epoch": 0.45711532283970446,
      "grad_norm": 0.4657999873161316,
      "learning_rate": 9.556589989870265e-06,
      "loss": 0.4688,
      "step": 9961
    },
    {
      "epoch": 0.4571612133449589,
      "grad_norm": 0.4579140841960907,
      "learning_rate": 9.556489039751231e-06,
      "loss": 0.413,
      "step": 9962
    },
    {
      "epoch": 0.4572071038502134,
      "grad_norm": 0.45779508352279663,
      "learning_rate": 9.556388078675276e-06,
      "loss": 0.4171,
      "step": 9963
    },
    {
      "epoch": 0.45725299435546785,
      "grad_norm": 0.42638781666755676,
      "learning_rate": 9.556287106642647e-06,
      "loss": 0.3325,
      "step": 9964
    },
    {
      "epoch": 0.45729888486072234,
      "grad_norm": 0.47515642642974854,
      "learning_rate": 9.556186123653587e-06,
      "loss": 0.4014,
      "step": 9965
    },
    {
      "epoch": 0.4573447753659768,
      "grad_norm": 0.47173863649368286,
      "learning_rate": 9.556085129708334e-06,
      "loss": 0.4416,
      "step": 9966
    },
    {
      "epoch": 0.45739066587123123,
      "grad_norm": 0.4339618384838104,
      "learning_rate": 9.555984124807136e-06,
      "loss": 0.3945,
      "step": 9967
    },
    {
      "epoch": 0.4574365563764857,
      "grad_norm": 0.4697257876396179,
      "learning_rate": 9.555883108950234e-06,
      "loss": 0.4495,
      "step": 9968
    },
    {
      "epoch": 0.45748244688174017,
      "grad_norm": 0.45135417580604553,
      "learning_rate": 9.55578208213787e-06,
      "loss": 0.3734,
      "step": 9969
    },
    {
      "epoch": 0.4575283373869946,
      "grad_norm": 0.4891314208507538,
      "learning_rate": 9.555681044370288e-06,
      "loss": 0.4063,
      "step": 9970
    },
    {
      "epoch": 0.4575742278922491,
      "grad_norm": 0.47307920455932617,
      "learning_rate": 9.555579995647732e-06,
      "loss": 0.4372,
      "step": 9971
    },
    {
      "epoch": 0.45762011839750355,
      "grad_norm": 0.4372274577617645,
      "learning_rate": 9.555478935970443e-06,
      "loss": 0.4241,
      "step": 9972
    },
    {
      "epoch": 0.457666008902758,
      "grad_norm": 0.47360700368881226,
      "learning_rate": 9.555377865338666e-06,
      "loss": 0.4516,
      "step": 9973
    },
    {
      "epoch": 0.4577118994080125,
      "grad_norm": 0.45366328954696655,
      "learning_rate": 9.555276783752643e-06,
      "loss": 0.3862,
      "step": 9974
    },
    {
      "epoch": 0.45775778991326693,
      "grad_norm": 0.4219350218772888,
      "learning_rate": 9.555175691212617e-06,
      "loss": 0.3428,
      "step": 9975
    },
    {
      "epoch": 0.45780368041852143,
      "grad_norm": 0.48012861609458923,
      "learning_rate": 9.55507458771883e-06,
      "loss": 0.4389,
      "step": 9976
    },
    {
      "epoch": 0.4578495709237759,
      "grad_norm": 0.4771484136581421,
      "learning_rate": 9.554973473271526e-06,
      "loss": 0.3932,
      "step": 9977
    },
    {
      "epoch": 0.4578954614290303,
      "grad_norm": 0.43656250834465027,
      "learning_rate": 9.55487234787095e-06,
      "loss": 0.3869,
      "step": 9978
    },
    {
      "epoch": 0.4579413519342848,
      "grad_norm": 0.4799874722957611,
      "learning_rate": 9.554771211517343e-06,
      "loss": 0.3874,
      "step": 9979
    },
    {
      "epoch": 0.45798724243953925,
      "grad_norm": 0.4342607855796814,
      "learning_rate": 9.554670064210948e-06,
      "loss": 0.3585,
      "step": 9980
    },
    {
      "epoch": 0.4580331329447937,
      "grad_norm": 0.47616249322891235,
      "learning_rate": 9.55456890595201e-06,
      "loss": 0.4598,
      "step": 9981
    },
    {
      "epoch": 0.4580790234500482,
      "grad_norm": 0.4475398361682892,
      "learning_rate": 9.554467736740771e-06,
      "loss": 0.3992,
      "step": 9982
    },
    {
      "epoch": 0.45812491395530264,
      "grad_norm": 0.4943822920322418,
      "learning_rate": 9.554366556577474e-06,
      "loss": 0.5044,
      "step": 9983
    },
    {
      "epoch": 0.45817080446055714,
      "grad_norm": 0.5043686032295227,
      "learning_rate": 9.554265365462364e-06,
      "loss": 0.5225,
      "step": 9984
    },
    {
      "epoch": 0.4582166949658116,
      "grad_norm": 0.44076502323150635,
      "learning_rate": 9.55416416339568e-06,
      "loss": 0.36,
      "step": 9985
    },
    {
      "epoch": 0.458262585471066,
      "grad_norm": 0.4140150249004364,
      "learning_rate": 9.554062950377672e-06,
      "loss": 0.3421,
      "step": 9986
    },
    {
      "epoch": 0.4583084759763205,
      "grad_norm": 0.46010980010032654,
      "learning_rate": 9.553961726408578e-06,
      "loss": 0.3757,
      "step": 9987
    },
    {
      "epoch": 0.45835436648157496,
      "grad_norm": 0.44241753220558167,
      "learning_rate": 9.553860491488644e-06,
      "loss": 0.3274,
      "step": 9988
    },
    {
      "epoch": 0.4584002569868294,
      "grad_norm": 0.43740734457969666,
      "learning_rate": 9.553759245618112e-06,
      "loss": 0.3575,
      "step": 9989
    },
    {
      "epoch": 0.4584461474920839,
      "grad_norm": 1.5007730722427368,
      "learning_rate": 9.553657988797228e-06,
      "loss": 0.5251,
      "step": 9990
    },
    {
      "epoch": 0.45849203799733834,
      "grad_norm": 0.5382006764411926,
      "learning_rate": 9.553556721026233e-06,
      "loss": 0.4733,
      "step": 9991
    },
    {
      "epoch": 0.45853792850259284,
      "grad_norm": 0.44896650314331055,
      "learning_rate": 9.55345544230537e-06,
      "loss": 0.3813,
      "step": 9992
    },
    {
      "epoch": 0.4585838190078473,
      "grad_norm": 0.49121126532554626,
      "learning_rate": 9.553354152634886e-06,
      "loss": 0.4258,
      "step": 9993
    },
    {
      "epoch": 0.4586297095131017,
      "grad_norm": 0.42866453528404236,
      "learning_rate": 9.55325285201502e-06,
      "loss": 0.3065,
      "step": 9994
    },
    {
      "epoch": 0.4586756000183562,
      "grad_norm": 0.5034466981887817,
      "learning_rate": 9.553151540446019e-06,
      "loss": 0.462,
      "step": 9995
    },
    {
      "epoch": 0.45872149052361066,
      "grad_norm": 0.48079365491867065,
      "learning_rate": 9.553050217928126e-06,
      "loss": 0.4196,
      "step": 9996
    },
    {
      "epoch": 0.4587673810288651,
      "grad_norm": 0.48676690459251404,
      "learning_rate": 9.552948884461584e-06,
      "loss": 0.393,
      "step": 9997
    },
    {
      "epoch": 0.4588132715341196,
      "grad_norm": 0.4471967816352844,
      "learning_rate": 9.552847540046636e-06,
      "loss": 0.414,
      "step": 9998
    },
    {
      "epoch": 0.45885916203937405,
      "grad_norm": 0.4555295705795288,
      "learning_rate": 9.552746184683528e-06,
      "loss": 0.3607,
      "step": 9999
    },
    {
      "epoch": 0.4589050525446285,
      "grad_norm": 0.49966102838516235,
      "learning_rate": 9.552644818372501e-06,
      "loss": 0.4345,
      "step": 10000
    },
    {
      "epoch": 0.458950943049883,
      "grad_norm": 0.4645128548145294,
      "learning_rate": 9.552543441113801e-06,
      "loss": 0.3502,
      "step": 10001
    },
    {
      "epoch": 0.45899683355513743,
      "grad_norm": 0.5004390478134155,
      "learning_rate": 9.552442052907672e-06,
      "loss": 0.4586,
      "step": 10002
    },
    {
      "epoch": 0.4590427240603919,
      "grad_norm": 0.5824493765830994,
      "learning_rate": 9.552340653754353e-06,
      "loss": 0.4122,
      "step": 10003
    },
    {
      "epoch": 0.45908861456564637,
      "grad_norm": 0.42458784580230713,
      "learning_rate": 9.552239243654094e-06,
      "loss": 0.3161,
      "step": 10004
    },
    {
      "epoch": 0.4591345050709008,
      "grad_norm": 0.5094610452651978,
      "learning_rate": 9.552137822607135e-06,
      "loss": 0.4611,
      "step": 10005
    },
    {
      "epoch": 0.4591803955761553,
      "grad_norm": 0.5777984261512756,
      "learning_rate": 9.552036390613724e-06,
      "loss": 0.5842,
      "step": 10006
    },
    {
      "epoch": 0.45922628608140975,
      "grad_norm": 0.4628232717514038,
      "learning_rate": 9.551934947674099e-06,
      "loss": 0.4088,
      "step": 10007
    },
    {
      "epoch": 0.4592721765866642,
      "grad_norm": 0.4700241982936859,
      "learning_rate": 9.55183349378851e-06,
      "loss": 0.3928,
      "step": 10008
    },
    {
      "epoch": 0.4593180670919187,
      "grad_norm": 0.4583022892475128,
      "learning_rate": 9.551732028957195e-06,
      "loss": 0.3695,
      "step": 10009
    },
    {
      "epoch": 0.45936395759717313,
      "grad_norm": 0.43629127740859985,
      "learning_rate": 9.551630553180401e-06,
      "loss": 0.3745,
      "step": 10010
    },
    {
      "epoch": 0.45940984810242763,
      "grad_norm": 0.4497649669647217,
      "learning_rate": 9.551529066458375e-06,
      "loss": 0.3664,
      "step": 10011
    },
    {
      "epoch": 0.4594557386076821,
      "grad_norm": 0.43785974383354187,
      "learning_rate": 9.551427568791357e-06,
      "loss": 0.328,
      "step": 10012
    },
    {
      "epoch": 0.4595016291129365,
      "grad_norm": 0.44024258852005005,
      "learning_rate": 9.551326060179593e-06,
      "loss": 0.3499,
      "step": 10013
    },
    {
      "epoch": 0.459547519618191,
      "grad_norm": 0.4628494381904602,
      "learning_rate": 9.551224540623324e-06,
      "loss": 0.4056,
      "step": 10014
    },
    {
      "epoch": 0.45959341012344546,
      "grad_norm": 0.4891040623188019,
      "learning_rate": 9.551123010122798e-06,
      "loss": 0.4342,
      "step": 10015
    },
    {
      "epoch": 0.4596393006286999,
      "grad_norm": 0.4771755039691925,
      "learning_rate": 9.551021468678256e-06,
      "loss": 0.4586,
      "step": 10016
    },
    {
      "epoch": 0.4596851911339544,
      "grad_norm": 0.450152724981308,
      "learning_rate": 9.550919916289945e-06,
      "loss": 0.3853,
      "step": 10017
    },
    {
      "epoch": 0.45973108163920884,
      "grad_norm": 0.4463956654071808,
      "learning_rate": 9.550818352958108e-06,
      "loss": 0.3803,
      "step": 10018
    },
    {
      "epoch": 0.45977697214446334,
      "grad_norm": 0.5033095479011536,
      "learning_rate": 9.55071677868299e-06,
      "loss": 0.5156,
      "step": 10019
    },
    {
      "epoch": 0.4598228626497178,
      "grad_norm": 0.4749075472354889,
      "learning_rate": 9.550615193464831e-06,
      "loss": 0.4763,
      "step": 10020
    },
    {
      "epoch": 0.4598687531549722,
      "grad_norm": 0.4581708610057831,
      "learning_rate": 9.55051359730388e-06,
      "loss": 0.3961,
      "step": 10021
    },
    {
      "epoch": 0.4599146436602267,
      "grad_norm": 0.4889867603778839,
      "learning_rate": 9.550411990200381e-06,
      "loss": 0.4524,
      "step": 10022
    },
    {
      "epoch": 0.45996053416548116,
      "grad_norm": 0.46329137682914734,
      "learning_rate": 9.550310372154579e-06,
      "loss": 0.3885,
      "step": 10023
    },
    {
      "epoch": 0.4600064246707356,
      "grad_norm": 0.455548495054245,
      "learning_rate": 9.550208743166715e-06,
      "loss": 0.4026,
      "step": 10024
    },
    {
      "epoch": 0.4600523151759901,
      "grad_norm": 0.4738123416900635,
      "learning_rate": 9.550107103237034e-06,
      "loss": 0.4446,
      "step": 10025
    },
    {
      "epoch": 0.46009820568124454,
      "grad_norm": 0.5240274667739868,
      "learning_rate": 9.550005452365783e-06,
      "loss": 0.4979,
      "step": 10026
    },
    {
      "epoch": 0.46014409618649904,
      "grad_norm": 0.43227049708366394,
      "learning_rate": 9.549903790553203e-06,
      "loss": 0.3814,
      "step": 10027
    },
    {
      "epoch": 0.4601899866917535,
      "grad_norm": 0.46555179357528687,
      "learning_rate": 9.549802117799542e-06,
      "loss": 0.4336,
      "step": 10028
    },
    {
      "epoch": 0.4602358771970079,
      "grad_norm": 0.4735663831233978,
      "learning_rate": 9.549700434105042e-06,
      "loss": 0.4299,
      "step": 10029
    },
    {
      "epoch": 0.4602817677022624,
      "grad_norm": 0.45129549503326416,
      "learning_rate": 9.54959873946995e-06,
      "loss": 0.3625,
      "step": 10030
    },
    {
      "epoch": 0.46032765820751687,
      "grad_norm": 0.5759739875793457,
      "learning_rate": 9.549497033894509e-06,
      "loss": 0.3094,
      "step": 10031
    },
    {
      "epoch": 0.4603735487127713,
      "grad_norm": 0.4574528932571411,
      "learning_rate": 9.549395317378961e-06,
      "loss": 0.4236,
      "step": 10032
    },
    {
      "epoch": 0.4604194392180258,
      "grad_norm": 0.4939922094345093,
      "learning_rate": 9.549293589923554e-06,
      "loss": 0.4679,
      "step": 10033
    },
    {
      "epoch": 0.46046532972328025,
      "grad_norm": 0.4289783239364624,
      "learning_rate": 9.549191851528533e-06,
      "loss": 0.3391,
      "step": 10034
    },
    {
      "epoch": 0.4605112202285347,
      "grad_norm": 0.4712502360343933,
      "learning_rate": 9.549090102194141e-06,
      "loss": 0.4229,
      "step": 10035
    },
    {
      "epoch": 0.4605571107337892,
      "grad_norm": 0.4966989755630493,
      "learning_rate": 9.548988341920623e-06,
      "loss": 0.4486,
      "step": 10036
    },
    {
      "epoch": 0.46060300123904363,
      "grad_norm": 0.4693969190120697,
      "learning_rate": 9.54888657070822e-06,
      "loss": 0.3969,
      "step": 10037
    },
    {
      "epoch": 0.46064889174429813,
      "grad_norm": 0.45313915610313416,
      "learning_rate": 9.548784788557186e-06,
      "loss": 0.3803,
      "step": 10038
    },
    {
      "epoch": 0.46069478224955257,
      "grad_norm": 0.8935025930404663,
      "learning_rate": 9.548682995467758e-06,
      "loss": 0.4941,
      "step": 10039
    },
    {
      "epoch": 0.460740672754807,
      "grad_norm": 0.4878140687942505,
      "learning_rate": 9.548581191440183e-06,
      "loss": 0.395,
      "step": 10040
    },
    {
      "epoch": 0.4607865632600615,
      "grad_norm": 0.5029117465019226,
      "learning_rate": 9.548479376474704e-06,
      "loss": 0.4907,
      "step": 10041
    },
    {
      "epoch": 0.46083245376531595,
      "grad_norm": 0.45163223147392273,
      "learning_rate": 9.54837755057157e-06,
      "loss": 0.341,
      "step": 10042
    },
    {
      "epoch": 0.4608783442705704,
      "grad_norm": 0.44898027181625366,
      "learning_rate": 9.548275713731023e-06,
      "loss": 0.3525,
      "step": 10043
    },
    {
      "epoch": 0.4609242347758249,
      "grad_norm": 0.46235594153404236,
      "learning_rate": 9.54817386595331e-06,
      "loss": 0.3689,
      "step": 10044
    },
    {
      "epoch": 0.46097012528107933,
      "grad_norm": 0.48003360629081726,
      "learning_rate": 9.548072007238672e-06,
      "loss": 0.4639,
      "step": 10045
    },
    {
      "epoch": 0.46101601578633383,
      "grad_norm": 0.4563031494617462,
      "learning_rate": 9.547970137587355e-06,
      "loss": 0.415,
      "step": 10046
    },
    {
      "epoch": 0.4610619062915883,
      "grad_norm": 0.4135788381099701,
      "learning_rate": 9.547868256999608e-06,
      "loss": 0.3119,
      "step": 10047
    },
    {
      "epoch": 0.4611077967968427,
      "grad_norm": 0.4257953464984894,
      "learning_rate": 9.547766365475673e-06,
      "loss": 0.339,
      "step": 10048
    },
    {
      "epoch": 0.4611536873020972,
      "grad_norm": 0.4664488434791565,
      "learning_rate": 9.547664463015793e-06,
      "loss": 0.4404,
      "step": 10049
    },
    {
      "epoch": 0.46119957780735166,
      "grad_norm": 0.43049153685569763,
      "learning_rate": 9.547562549620217e-06,
      "loss": 0.3518,
      "step": 10050
    },
    {
      "epoch": 0.4612454683126061,
      "grad_norm": 0.4459921419620514,
      "learning_rate": 9.547460625289188e-06,
      "loss": 0.3748,
      "step": 10051
    },
    {
      "epoch": 0.4612913588178606,
      "grad_norm": 0.45962008833885193,
      "learning_rate": 9.547358690022952e-06,
      "loss": 0.3849,
      "step": 10052
    },
    {
      "epoch": 0.46133724932311504,
      "grad_norm": 0.46512913703918457,
      "learning_rate": 9.547256743821753e-06,
      "loss": 0.4191,
      "step": 10053
    },
    {
      "epoch": 0.46138313982836954,
      "grad_norm": 0.4879626929759979,
      "learning_rate": 9.547154786685836e-06,
      "loss": 0.4551,
      "step": 10054
    },
    {
      "epoch": 0.461429030333624,
      "grad_norm": 0.4872152805328369,
      "learning_rate": 9.547052818615448e-06,
      "loss": 0.467,
      "step": 10055
    },
    {
      "epoch": 0.4614749208388784,
      "grad_norm": 0.4161750376224518,
      "learning_rate": 9.546950839610831e-06,
      "loss": 0.3203,
      "step": 10056
    },
    {
      "epoch": 0.4615208113441329,
      "grad_norm": 0.4910643398761749,
      "learning_rate": 9.546848849672235e-06,
      "loss": 0.4535,
      "step": 10057
    },
    {
      "epoch": 0.46156670184938736,
      "grad_norm": 0.4327855408191681,
      "learning_rate": 9.5467468487999e-06,
      "loss": 0.3518,
      "step": 10058
    },
    {
      "epoch": 0.4616125923546418,
      "grad_norm": 0.44933268427848816,
      "learning_rate": 9.546644836994077e-06,
      "loss": 0.3708,
      "step": 10059
    },
    {
      "epoch": 0.4616584828598963,
      "grad_norm": 0.47097790241241455,
      "learning_rate": 9.546542814255006e-06,
      "loss": 0.4399,
      "step": 10060
    },
    {
      "epoch": 0.46170437336515074,
      "grad_norm": 0.5800557732582092,
      "learning_rate": 9.546440780582935e-06,
      "loss": 0.4243,
      "step": 10061
    },
    {
      "epoch": 0.4617502638704052,
      "grad_norm": 0.4938318729400635,
      "learning_rate": 9.546338735978108e-06,
      "loss": 0.4192,
      "step": 10062
    },
    {
      "epoch": 0.4617961543756597,
      "grad_norm": 0.508450984954834,
      "learning_rate": 9.546236680440772e-06,
      "loss": 0.5059,
      "step": 10063
    },
    {
      "epoch": 0.4618420448809141,
      "grad_norm": 0.4875849783420563,
      "learning_rate": 9.546134613971172e-06,
      "loss": 0.4661,
      "step": 10064
    },
    {
      "epoch": 0.4618879353861686,
      "grad_norm": 0.45299094915390015,
      "learning_rate": 9.546032536569552e-06,
      "loss": 0.3789,
      "step": 10065
    },
    {
      "epoch": 0.46193382589142307,
      "grad_norm": 0.41441550850868225,
      "learning_rate": 9.545930448236158e-06,
      "loss": 0.3209,
      "step": 10066
    },
    {
      "epoch": 0.4619797163966775,
      "grad_norm": 0.4659954309463501,
      "learning_rate": 9.545828348971238e-06,
      "loss": 0.4362,
      "step": 10067
    },
    {
      "epoch": 0.462025606901932,
      "grad_norm": 0.4999721050262451,
      "learning_rate": 9.545726238775036e-06,
      "loss": 0.4306,
      "step": 10068
    },
    {
      "epoch": 0.46207149740718645,
      "grad_norm": 0.44158822298049927,
      "learning_rate": 9.545624117647794e-06,
      "loss": 0.3768,
      "step": 10069
    },
    {
      "epoch": 0.4621173879124409,
      "grad_norm": 0.45205965638160706,
      "learning_rate": 9.545521985589763e-06,
      "loss": 0.3594,
      "step": 10070
    },
    {
      "epoch": 0.4621632784176954,
      "grad_norm": 0.4988010823726654,
      "learning_rate": 9.545419842601185e-06,
      "loss": 0.5267,
      "step": 10071
    },
    {
      "epoch": 0.46220916892294983,
      "grad_norm": 0.43586310744285583,
      "learning_rate": 9.545317688682308e-06,
      "loss": 0.3744,
      "step": 10072
    },
    {
      "epoch": 0.46225505942820433,
      "grad_norm": 0.5058405995368958,
      "learning_rate": 9.545215523833375e-06,
      "loss": 0.4243,
      "step": 10073
    },
    {
      "epoch": 0.46230094993345877,
      "grad_norm": 0.45228537917137146,
      "learning_rate": 9.545113348054634e-06,
      "loss": 0.3641,
      "step": 10074
    },
    {
      "epoch": 0.4623468404387132,
      "grad_norm": 0.4844028353691101,
      "learning_rate": 9.54501116134633e-06,
      "loss": 0.4161,
      "step": 10075
    },
    {
      "epoch": 0.4623927309439677,
      "grad_norm": 0.5068084597587585,
      "learning_rate": 9.544908963708708e-06,
      "loss": 0.4654,
      "step": 10076
    },
    {
      "epoch": 0.46243862144922215,
      "grad_norm": 0.4577224850654602,
      "learning_rate": 9.544806755142016e-06,
      "loss": 0.3672,
      "step": 10077
    },
    {
      "epoch": 0.4624845119544766,
      "grad_norm": 0.4572412371635437,
      "learning_rate": 9.544704535646497e-06,
      "loss": 0.4306,
      "step": 10078
    },
    {
      "epoch": 0.4625304024597311,
      "grad_norm": 0.519980251789093,
      "learning_rate": 9.544602305222397e-06,
      "loss": 0.4995,
      "step": 10079
    },
    {
      "epoch": 0.46257629296498554,
      "grad_norm": 0.44654372334480286,
      "learning_rate": 9.544500063869964e-06,
      "loss": 0.4219,
      "step": 10080
    },
    {
      "epoch": 0.46262218347024003,
      "grad_norm": 0.5094678401947021,
      "learning_rate": 9.544397811589441e-06,
      "loss": 0.4552,
      "step": 10081
    },
    {
      "epoch": 0.4626680739754945,
      "grad_norm": 0.526958703994751,
      "learning_rate": 9.544295548381077e-06,
      "loss": 0.4772,
      "step": 10082
    },
    {
      "epoch": 0.4627139644807489,
      "grad_norm": 0.5276633501052856,
      "learning_rate": 9.544193274245118e-06,
      "loss": 0.4472,
      "step": 10083
    },
    {
      "epoch": 0.4627598549860034,
      "grad_norm": 0.4546002149581909,
      "learning_rate": 9.544090989181806e-06,
      "loss": 0.387,
      "step": 10084
    },
    {
      "epoch": 0.46280574549125786,
      "grad_norm": 0.45366042852401733,
      "learning_rate": 9.543988693191389e-06,
      "loss": 0.3801,
      "step": 10085
    },
    {
      "epoch": 0.4628516359965123,
      "grad_norm": 0.4777795970439911,
      "learning_rate": 9.543886386274116e-06,
      "loss": 0.4108,
      "step": 10086
    },
    {
      "epoch": 0.4628975265017668,
      "grad_norm": 0.4432874619960785,
      "learning_rate": 9.543784068430228e-06,
      "loss": 0.378,
      "step": 10087
    },
    {
      "epoch": 0.46294341700702124,
      "grad_norm": 0.5480496883392334,
      "learning_rate": 9.543681739659974e-06,
      "loss": 0.5073,
      "step": 10088
    },
    {
      "epoch": 0.4629893075122757,
      "grad_norm": 0.5434205532073975,
      "learning_rate": 9.543579399963598e-06,
      "loss": 0.4025,
      "step": 10089
    },
    {
      "epoch": 0.4630351980175302,
      "grad_norm": 0.46511849761009216,
      "learning_rate": 9.54347704934135e-06,
      "loss": 0.3929,
      "step": 10090
    },
    {
      "epoch": 0.4630810885227846,
      "grad_norm": 0.46351978182792664,
      "learning_rate": 9.543374687793472e-06,
      "loss": 0.4117,
      "step": 10091
    },
    {
      "epoch": 0.4631269790280391,
      "grad_norm": 0.46285396814346313,
      "learning_rate": 9.543272315320213e-06,
      "loss": 0.4013,
      "step": 10092
    },
    {
      "epoch": 0.46317286953329356,
      "grad_norm": 0.45155471563339233,
      "learning_rate": 9.543169931921817e-06,
      "loss": 0.3618,
      "step": 10093
    },
    {
      "epoch": 0.463218760038548,
      "grad_norm": 0.4584038555622101,
      "learning_rate": 9.543067537598532e-06,
      "loss": 0.3709,
      "step": 10094
    },
    {
      "epoch": 0.4632646505438025,
      "grad_norm": 0.48299309611320496,
      "learning_rate": 9.542965132350603e-06,
      "loss": 0.4702,
      "step": 10095
    },
    {
      "epoch": 0.46331054104905695,
      "grad_norm": 0.49806708097457886,
      "learning_rate": 9.542862716178275e-06,
      "loss": 0.4814,
      "step": 10096
    },
    {
      "epoch": 0.4633564315543114,
      "grad_norm": 0.48939889669418335,
      "learning_rate": 9.5427602890818e-06,
      "loss": 0.4914,
      "step": 10097
    },
    {
      "epoch": 0.4634023220595659,
      "grad_norm": 0.4154219329357147,
      "learning_rate": 9.542657851061417e-06,
      "loss": 0.3038,
      "step": 10098
    },
    {
      "epoch": 0.4634482125648203,
      "grad_norm": 0.43353354930877686,
      "learning_rate": 9.542555402117376e-06,
      "loss": 0.381,
      "step": 10099
    },
    {
      "epoch": 0.4634941030700748,
      "grad_norm": 0.43858015537261963,
      "learning_rate": 9.542452942249924e-06,
      "loss": 0.3336,
      "step": 10100
    },
    {
      "epoch": 0.46353999357532927,
      "grad_norm": 0.47354984283447266,
      "learning_rate": 9.542350471459304e-06,
      "loss": 0.3993,
      "step": 10101
    },
    {
      "epoch": 0.4635858840805837,
      "grad_norm": 0.4408109784126282,
      "learning_rate": 9.542247989745768e-06,
      "loss": 0.3207,
      "step": 10102
    },
    {
      "epoch": 0.4636317745858382,
      "grad_norm": 0.47126561403274536,
      "learning_rate": 9.542145497109557e-06,
      "loss": 0.4249,
      "step": 10103
    },
    {
      "epoch": 0.46367766509109265,
      "grad_norm": 0.4833935797214508,
      "learning_rate": 9.54204299355092e-06,
      "loss": 0.3718,
      "step": 10104
    },
    {
      "epoch": 0.4637235555963471,
      "grad_norm": 0.4889027774333954,
      "learning_rate": 9.541940479070104e-06,
      "loss": 0.4329,
      "step": 10105
    },
    {
      "epoch": 0.4637694461016016,
      "grad_norm": 0.4404420256614685,
      "learning_rate": 9.541837953667353e-06,
      "loss": 0.3362,
      "step": 10106
    },
    {
      "epoch": 0.46381533660685603,
      "grad_norm": 0.45375174283981323,
      "learning_rate": 9.541735417342916e-06,
      "loss": 0.3671,
      "step": 10107
    },
    {
      "epoch": 0.46386122711211053,
      "grad_norm": 0.48985207080841064,
      "learning_rate": 9.541632870097039e-06,
      "loss": 0.4456,
      "step": 10108
    },
    {
      "epoch": 0.463907117617365,
      "grad_norm": 0.42800143361091614,
      "learning_rate": 9.541530311929968e-06,
      "loss": 0.3287,
      "step": 10109
    },
    {
      "epoch": 0.4639530081226194,
      "grad_norm": 0.4616200923919678,
      "learning_rate": 9.54142774284195e-06,
      "loss": 0.4028,
      "step": 10110
    },
    {
      "epoch": 0.4639988986278739,
      "grad_norm": 0.4615591764450073,
      "learning_rate": 9.541325162833232e-06,
      "loss": 0.3619,
      "step": 10111
    },
    {
      "epoch": 0.46404478913312835,
      "grad_norm": 0.5035247206687927,
      "learning_rate": 9.54122257190406e-06,
      "loss": 0.3915,
      "step": 10112
    },
    {
      "epoch": 0.4640906796383828,
      "grad_norm": 0.40776196122169495,
      "learning_rate": 9.54111997005468e-06,
      "loss": 0.3218,
      "step": 10113
    },
    {
      "epoch": 0.4641365701436373,
      "grad_norm": 0.504343569278717,
      "learning_rate": 9.541017357285342e-06,
      "loss": 0.4219,
      "step": 10114
    },
    {
      "epoch": 0.46418246064889174,
      "grad_norm": 0.5200358629226685,
      "learning_rate": 9.54091473359629e-06,
      "loss": 0.5024,
      "step": 10115
    },
    {
      "epoch": 0.46422835115414623,
      "grad_norm": 0.46448755264282227,
      "learning_rate": 9.540812098987768e-06,
      "loss": 0.3645,
      "step": 10116
    },
    {
      "epoch": 0.4642742416594007,
      "grad_norm": 0.49726325273513794,
      "learning_rate": 9.540709453460028e-06,
      "loss": 0.4559,
      "step": 10117
    },
    {
      "epoch": 0.4643201321646551,
      "grad_norm": 0.48025810718536377,
      "learning_rate": 9.540606797013317e-06,
      "loss": 0.4823,
      "step": 10118
    },
    {
      "epoch": 0.4643660226699096,
      "grad_norm": 0.5020802021026611,
      "learning_rate": 9.540504129647878e-06,
      "loss": 0.4306,
      "step": 10119
    },
    {
      "epoch": 0.46441191317516406,
      "grad_norm": 0.4576500654220581,
      "learning_rate": 9.54040145136396e-06,
      "loss": 0.3812,
      "step": 10120
    },
    {
      "epoch": 0.4644578036804185,
      "grad_norm": 0.4354941248893738,
      "learning_rate": 9.540298762161808e-06,
      "loss": 0.3377,
      "step": 10121
    },
    {
      "epoch": 0.464503694185673,
      "grad_norm": 0.4891124665737152,
      "learning_rate": 9.540196062041672e-06,
      "loss": 0.4525,
      "step": 10122
    },
    {
      "epoch": 0.46454958469092744,
      "grad_norm": 0.48238879442214966,
      "learning_rate": 9.540093351003797e-06,
      "loss": 0.3812,
      "step": 10123
    },
    {
      "epoch": 0.4645954751961819,
      "grad_norm": 0.4946328401565552,
      "learning_rate": 9.539990629048428e-06,
      "loss": 0.437,
      "step": 10124
    },
    {
      "epoch": 0.4646413657014364,
      "grad_norm": 0.4605189263820648,
      "learning_rate": 9.539887896175816e-06,
      "loss": 0.3833,
      "step": 10125
    },
    {
      "epoch": 0.4646872562066908,
      "grad_norm": 0.455402672290802,
      "learning_rate": 9.539785152386208e-06,
      "loss": 0.4505,
      "step": 10126
    },
    {
      "epoch": 0.4647331467119453,
      "grad_norm": 0.4873480200767517,
      "learning_rate": 9.53968239767985e-06,
      "loss": 0.3642,
      "step": 10127
    },
    {
      "epoch": 0.46477903721719976,
      "grad_norm": 0.41558223962783813,
      "learning_rate": 9.539579632056985e-06,
      "loss": 0.3062,
      "step": 10128
    },
    {
      "epoch": 0.4648249277224542,
      "grad_norm": 0.43191471695899963,
      "learning_rate": 9.539476855517866e-06,
      "loss": 0.337,
      "step": 10129
    },
    {
      "epoch": 0.4648708182277087,
      "grad_norm": 0.4598858952522278,
      "learning_rate": 9.539374068062739e-06,
      "loss": 0.3958,
      "step": 10130
    },
    {
      "epoch": 0.46491670873296315,
      "grad_norm": 0.424194872379303,
      "learning_rate": 9.539271269691847e-06,
      "loss": 0.3619,
      "step": 10131
    },
    {
      "epoch": 0.4649625992382176,
      "grad_norm": 0.4774252772331238,
      "learning_rate": 9.539168460405444e-06,
      "loss": 0.4249,
      "step": 10132
    },
    {
      "epoch": 0.4650084897434721,
      "grad_norm": 0.4371560215950012,
      "learning_rate": 9.53906564020377e-06,
      "loss": 0.4019,
      "step": 10133
    },
    {
      "epoch": 0.46505438024872653,
      "grad_norm": 0.4690553843975067,
      "learning_rate": 9.538962809087078e-06,
      "loss": 0.4406,
      "step": 10134
    },
    {
      "epoch": 0.465100270753981,
      "grad_norm": 0.47569596767425537,
      "learning_rate": 9.53885996705561e-06,
      "loss": 0.3947,
      "step": 10135
    },
    {
      "epoch": 0.46514616125923547,
      "grad_norm": 0.4826330542564392,
      "learning_rate": 9.53875711410962e-06,
      "loss": 0.4479,
      "step": 10136
    },
    {
      "epoch": 0.4651920517644899,
      "grad_norm": 0.5102675557136536,
      "learning_rate": 9.53865425024935e-06,
      "loss": 0.4498,
      "step": 10137
    },
    {
      "epoch": 0.4652379422697444,
      "grad_norm": 0.5118948817253113,
      "learning_rate": 9.53855137547505e-06,
      "loss": 0.4493,
      "step": 10138
    },
    {
      "epoch": 0.46528383277499885,
      "grad_norm": 0.4750429093837738,
      "learning_rate": 9.538448489786964e-06,
      "loss": 0.4453,
      "step": 10139
    },
    {
      "epoch": 0.4653297232802533,
      "grad_norm": 0.47009220719337463,
      "learning_rate": 9.538345593185343e-06,
      "loss": 0.4075,
      "step": 10140
    },
    {
      "epoch": 0.4653756137855078,
      "grad_norm": 0.48859694600105286,
      "learning_rate": 9.538242685670434e-06,
      "loss": 0.4574,
      "step": 10141
    },
    {
      "epoch": 0.46542150429076223,
      "grad_norm": 0.46785372495651245,
      "learning_rate": 9.538139767242482e-06,
      "loss": 0.4051,
      "step": 10142
    },
    {
      "epoch": 0.46546739479601673,
      "grad_norm": 0.48553353548049927,
      "learning_rate": 9.538036837901737e-06,
      "loss": 0.4099,
      "step": 10143
    },
    {
      "epoch": 0.4655132853012712,
      "grad_norm": 0.47608351707458496,
      "learning_rate": 9.537933897648445e-06,
      "loss": 0.4638,
      "step": 10144
    },
    {
      "epoch": 0.4655591758065256,
      "grad_norm": 0.42500007152557373,
      "learning_rate": 9.537830946482856e-06,
      "loss": 0.346,
      "step": 10145
    },
    {
      "epoch": 0.4656050663117801,
      "grad_norm": 0.4008198082447052,
      "learning_rate": 9.537727984405214e-06,
      "loss": 0.2771,
      "step": 10146
    },
    {
      "epoch": 0.46565095681703456,
      "grad_norm": 0.4721977710723877,
      "learning_rate": 9.53762501141577e-06,
      "loss": 0.4136,
      "step": 10147
    },
    {
      "epoch": 0.465696847322289,
      "grad_norm": 0.44315576553344727,
      "learning_rate": 9.53752202751477e-06,
      "loss": 0.3474,
      "step": 10148
    },
    {
      "epoch": 0.4657427378275435,
      "grad_norm": 0.5014727115631104,
      "learning_rate": 9.53741903270246e-06,
      "loss": 0.4673,
      "step": 10149
    },
    {
      "epoch": 0.46578862833279794,
      "grad_norm": 0.44032490253448486,
      "learning_rate": 9.537316026979088e-06,
      "loss": 0.3678,
      "step": 10150
    },
    {
      "epoch": 0.4658345188380524,
      "grad_norm": 0.48446351289749146,
      "learning_rate": 9.537213010344906e-06,
      "loss": 0.4935,
      "step": 10151
    },
    {
      "epoch": 0.4658804093433069,
      "grad_norm": 0.4451989233493805,
      "learning_rate": 9.537109982800157e-06,
      "loss": 0.3976,
      "step": 10152
    },
    {
      "epoch": 0.4659262998485613,
      "grad_norm": 0.4543994069099426,
      "learning_rate": 9.537006944345092e-06,
      "loss": 0.3456,
      "step": 10153
    },
    {
      "epoch": 0.4659721903538158,
      "grad_norm": 0.46358808875083923,
      "learning_rate": 9.536903894979955e-06,
      "loss": 0.4122,
      "step": 10154
    },
    {
      "epoch": 0.46601808085907026,
      "grad_norm": 0.4269695281982422,
      "learning_rate": 9.536800834704998e-06,
      "loss": 0.3582,
      "step": 10155
    },
    {
      "epoch": 0.4660639713643247,
      "grad_norm": 0.41859170794487,
      "learning_rate": 9.536697763520465e-06,
      "loss": 0.3425,
      "step": 10156
    },
    {
      "epoch": 0.4661098618695792,
      "grad_norm": 0.49543049931526184,
      "learning_rate": 9.536594681426608e-06,
      "loss": 0.4629,
      "step": 10157
    },
    {
      "epoch": 0.46615575237483364,
      "grad_norm": 0.41522207856178284,
      "learning_rate": 9.536491588423671e-06,
      "loss": 0.3175,
      "step": 10158
    },
    {
      "epoch": 0.4662016428800881,
      "grad_norm": 0.41735395789146423,
      "learning_rate": 9.536388484511902e-06,
      "loss": 0.3178,
      "step": 10159
    },
    {
      "epoch": 0.4662475333853426,
      "grad_norm": 0.4680722951889038,
      "learning_rate": 9.536285369691554e-06,
      "loss": 0.3714,
      "step": 10160
    },
    {
      "epoch": 0.466293423890597,
      "grad_norm": 0.47423529624938965,
      "learning_rate": 9.53618224396287e-06,
      "loss": 0.3745,
      "step": 10161
    },
    {
      "epoch": 0.4663393143958515,
      "grad_norm": 0.4534178376197815,
      "learning_rate": 9.536079107326099e-06,
      "loss": 0.3747,
      "step": 10162
    },
    {
      "epoch": 0.46638520490110597,
      "grad_norm": 0.4683798849582672,
      "learning_rate": 9.535975959781489e-06,
      "loss": 0.3924,
      "step": 10163
    },
    {
      "epoch": 0.4664310954063604,
      "grad_norm": 0.48483073711395264,
      "learning_rate": 9.535872801329289e-06,
      "loss": 0.4296,
      "step": 10164
    },
    {
      "epoch": 0.4664769859116149,
      "grad_norm": 0.47956332564353943,
      "learning_rate": 9.535769631969745e-06,
      "loss": 0.4267,
      "step": 10165
    },
    {
      "epoch": 0.46652287641686935,
      "grad_norm": 0.491567462682724,
      "learning_rate": 9.535666451703109e-06,
      "loss": 0.4816,
      "step": 10166
    },
    {
      "epoch": 0.4665687669221238,
      "grad_norm": 0.45649975538253784,
      "learning_rate": 9.535563260529624e-06,
      "loss": 0.4043,
      "step": 10167
    },
    {
      "epoch": 0.4666146574273783,
      "grad_norm": 0.4739347994327545,
      "learning_rate": 9.535460058449543e-06,
      "loss": 0.4243,
      "step": 10168
    },
    {
      "epoch": 0.46666054793263273,
      "grad_norm": 0.4462469816207886,
      "learning_rate": 9.53535684546311e-06,
      "loss": 0.4154,
      "step": 10169
    },
    {
      "epoch": 0.4667064384378872,
      "grad_norm": 0.5060902833938599,
      "learning_rate": 9.535253621570578e-06,
      "loss": 0.413,
      "step": 10170
    },
    {
      "epoch": 0.46675232894314167,
      "grad_norm": 0.46325913071632385,
      "learning_rate": 9.53515038677219e-06,
      "loss": 0.3567,
      "step": 10171
    },
    {
      "epoch": 0.4667982194483961,
      "grad_norm": 0.4650494456291199,
      "learning_rate": 9.535047141068199e-06,
      "loss": 0.3631,
      "step": 10172
    },
    {
      "epoch": 0.4668441099536506,
      "grad_norm": 0.4779733419418335,
      "learning_rate": 9.534943884458847e-06,
      "loss": 0.4563,
      "step": 10173
    },
    {
      "epoch": 0.46689000045890505,
      "grad_norm": 0.45985865592956543,
      "learning_rate": 9.534840616944389e-06,
      "loss": 0.3753,
      "step": 10174
    },
    {
      "epoch": 0.4669358909641595,
      "grad_norm": 0.4448172450065613,
      "learning_rate": 9.53473733852507e-06,
      "loss": 0.3499,
      "step": 10175
    },
    {
      "epoch": 0.466981781469414,
      "grad_norm": 0.4870678782463074,
      "learning_rate": 9.534634049201142e-06,
      "loss": 0.4437,
      "step": 10176
    },
    {
      "epoch": 0.46702767197466843,
      "grad_norm": 0.502816915512085,
      "learning_rate": 9.534530748972846e-06,
      "loss": 0.4155,
      "step": 10177
    },
    {
      "epoch": 0.4670735624799229,
      "grad_norm": 0.452913373708725,
      "learning_rate": 9.534427437840436e-06,
      "loss": 0.3465,
      "step": 10178
    },
    {
      "epoch": 0.4671194529851774,
      "grad_norm": 0.42497164011001587,
      "learning_rate": 9.53432411580416e-06,
      "loss": 0.3269,
      "step": 10179
    },
    {
      "epoch": 0.4671653434904318,
      "grad_norm": 0.49143680930137634,
      "learning_rate": 9.534220782864264e-06,
      "loss": 0.4783,
      "step": 10180
    },
    {
      "epoch": 0.4672112339956863,
      "grad_norm": 0.44599488377571106,
      "learning_rate": 9.534117439020999e-06,
      "loss": 0.3457,
      "step": 10181
    },
    {
      "epoch": 0.46725712450094076,
      "grad_norm": 0.4745642840862274,
      "learning_rate": 9.534014084274612e-06,
      "loss": 0.4089,
      "step": 10182
    },
    {
      "epoch": 0.4673030150061952,
      "grad_norm": 0.5003796219825745,
      "learning_rate": 9.533910718625354e-06,
      "loss": 0.4759,
      "step": 10183
    },
    {
      "epoch": 0.4673489055114497,
      "grad_norm": 0.43620696663856506,
      "learning_rate": 9.533807342073469e-06,
      "loss": 0.3626,
      "step": 10184
    },
    {
      "epoch": 0.46739479601670414,
      "grad_norm": 0.4845469295978546,
      "learning_rate": 9.53370395461921e-06,
      "loss": 0.4311,
      "step": 10185
    },
    {
      "epoch": 0.4674406865219586,
      "grad_norm": 0.47875112295150757,
      "learning_rate": 9.533600556262823e-06,
      "loss": 0.4171,
      "step": 10186
    },
    {
      "epoch": 0.4674865770272131,
      "grad_norm": 0.4743548035621643,
      "learning_rate": 9.533497147004558e-06,
      "loss": 0.4694,
      "step": 10187
    },
    {
      "epoch": 0.4675324675324675,
      "grad_norm": 0.4402513802051544,
      "learning_rate": 9.533393726844662e-06,
      "loss": 0.3554,
      "step": 10188
    },
    {
      "epoch": 0.467578358037722,
      "grad_norm": 0.43777450919151306,
      "learning_rate": 9.533290295783387e-06,
      "loss": 0.3824,
      "step": 10189
    },
    {
      "epoch": 0.46762424854297646,
      "grad_norm": 0.4674980044364929,
      "learning_rate": 9.533186853820978e-06,
      "loss": 0.4826,
      "step": 10190
    },
    {
      "epoch": 0.4676701390482309,
      "grad_norm": 0.4540131986141205,
      "learning_rate": 9.533083400957686e-06,
      "loss": 0.3677,
      "step": 10191
    },
    {
      "epoch": 0.4677160295534854,
      "grad_norm": 0.43238818645477295,
      "learning_rate": 9.532979937193758e-06,
      "loss": 0.3242,
      "step": 10192
    },
    {
      "epoch": 0.46776192005873984,
      "grad_norm": 0.49364128708839417,
      "learning_rate": 9.532876462529445e-06,
      "loss": 0.4689,
      "step": 10193
    },
    {
      "epoch": 0.4678078105639943,
      "grad_norm": 0.4342409372329712,
      "learning_rate": 9.532772976964994e-06,
      "loss": 0.2977,
      "step": 10194
    },
    {
      "epoch": 0.4678537010692488,
      "grad_norm": 0.427198201417923,
      "learning_rate": 9.532669480500655e-06,
      "loss": 0.3365,
      "step": 10195
    },
    {
      "epoch": 0.4678995915745032,
      "grad_norm": 0.4457751512527466,
      "learning_rate": 9.532565973136678e-06,
      "loss": 0.3594,
      "step": 10196
    },
    {
      "epoch": 0.4679454820797577,
      "grad_norm": 0.5074992179870605,
      "learning_rate": 9.532462454873308e-06,
      "loss": 0.415,
      "step": 10197
    },
    {
      "epoch": 0.46799137258501217,
      "grad_norm": 0.43412360548973083,
      "learning_rate": 9.532358925710799e-06,
      "loss": 0.3679,
      "step": 10198
    },
    {
      "epoch": 0.4680372630902666,
      "grad_norm": 0.4655688405036926,
      "learning_rate": 9.532255385649394e-06,
      "loss": 0.4313,
      "step": 10199
    },
    {
      "epoch": 0.4680831535955211,
      "grad_norm": 0.46152544021606445,
      "learning_rate": 9.532151834689347e-06,
      "loss": 0.4158,
      "step": 10200
    },
    {
      "epoch": 0.46812904410077555,
      "grad_norm": 0.4291996657848358,
      "learning_rate": 9.532048272830906e-06,
      "loss": 0.3904,
      "step": 10201
    },
    {
      "epoch": 0.46817493460603,
      "grad_norm": 0.44603246450424194,
      "learning_rate": 9.531944700074316e-06,
      "loss": 0.3269,
      "step": 10202
    },
    {
      "epoch": 0.4682208251112845,
      "grad_norm": 0.48220697045326233,
      "learning_rate": 9.531841116419831e-06,
      "loss": 0.4626,
      "step": 10203
    },
    {
      "epoch": 0.46826671561653893,
      "grad_norm": 0.46037623286247253,
      "learning_rate": 9.5317375218677e-06,
      "loss": 0.4074,
      "step": 10204
    },
    {
      "epoch": 0.4683126061217934,
      "grad_norm": 0.47353342175483704,
      "learning_rate": 9.531633916418168e-06,
      "loss": 0.4856,
      "step": 10205
    },
    {
      "epoch": 0.46835849662704787,
      "grad_norm": 0.4755588173866272,
      "learning_rate": 9.53153030007149e-06,
      "loss": 0.4093,
      "step": 10206
    },
    {
      "epoch": 0.4684043871323023,
      "grad_norm": 0.4176962077617645,
      "learning_rate": 9.531426672827909e-06,
      "loss": 0.2916,
      "step": 10207
    },
    {
      "epoch": 0.4684502776375568,
      "grad_norm": 0.4827055335044861,
      "learning_rate": 9.531323034687679e-06,
      "loss": 0.4857,
      "step": 10208
    },
    {
      "epoch": 0.46849616814281125,
      "grad_norm": 0.43406835198402405,
      "learning_rate": 9.531219385651045e-06,
      "loss": 0.3524,
      "step": 10209
    },
    {
      "epoch": 0.4685420586480657,
      "grad_norm": 0.4584280252456665,
      "learning_rate": 9.531115725718261e-06,
      "loss": 0.4167,
      "step": 10210
    },
    {
      "epoch": 0.4685879491533202,
      "grad_norm": 0.49388375878334045,
      "learning_rate": 9.531012054889572e-06,
      "loss": 0.4283,
      "step": 10211
    },
    {
      "epoch": 0.46863383965857464,
      "grad_norm": 0.4825495183467865,
      "learning_rate": 9.53090837316523e-06,
      "loss": 0.3636,
      "step": 10212
    },
    {
      "epoch": 0.4686797301638291,
      "grad_norm": 0.46548572182655334,
      "learning_rate": 9.530804680545482e-06,
      "loss": 0.364,
      "step": 10213
    },
    {
      "epoch": 0.4687256206690836,
      "grad_norm": 0.4603481590747833,
      "learning_rate": 9.530700977030578e-06,
      "loss": 0.4266,
      "step": 10214
    },
    {
      "epoch": 0.468771511174338,
      "grad_norm": 0.4530159831047058,
      "learning_rate": 9.53059726262077e-06,
      "loss": 0.3879,
      "step": 10215
    },
    {
      "epoch": 0.4688174016795925,
      "grad_norm": 0.4745256304740906,
      "learning_rate": 9.530493537316306e-06,
      "loss": 0.4746,
      "step": 10216
    },
    {
      "epoch": 0.46886329218484696,
      "grad_norm": 0.5069655776023865,
      "learning_rate": 9.530389801117432e-06,
      "loss": 0.4944,
      "step": 10217
    },
    {
      "epoch": 0.4689091826901014,
      "grad_norm": 0.4569256007671356,
      "learning_rate": 9.530286054024403e-06,
      "loss": 0.3674,
      "step": 10218
    },
    {
      "epoch": 0.4689550731953559,
      "grad_norm": 0.46480974555015564,
      "learning_rate": 9.530182296037464e-06,
      "loss": 0.4174,
      "step": 10219
    },
    {
      "epoch": 0.46900096370061034,
      "grad_norm": 0.5011894702911377,
      "learning_rate": 9.530078527156869e-06,
      "loss": 0.4719,
      "step": 10220
    },
    {
      "epoch": 0.4690468542058648,
      "grad_norm": 0.4615441858768463,
      "learning_rate": 9.529974747382862e-06,
      "loss": 0.4254,
      "step": 10221
    },
    {
      "epoch": 0.4690927447111193,
      "grad_norm": 0.5117265582084656,
      "learning_rate": 9.529870956715697e-06,
      "loss": 0.4158,
      "step": 10222
    },
    {
      "epoch": 0.4691386352163737,
      "grad_norm": 0.48618847131729126,
      "learning_rate": 9.529767155155622e-06,
      "loss": 0.4272,
      "step": 10223
    },
    {
      "epoch": 0.4691845257216282,
      "grad_norm": 0.40197300910949707,
      "learning_rate": 9.529663342702886e-06,
      "loss": 0.3239,
      "step": 10224
    },
    {
      "epoch": 0.46923041622688266,
      "grad_norm": 0.43460094928741455,
      "learning_rate": 9.52955951935774e-06,
      "loss": 0.4016,
      "step": 10225
    },
    {
      "epoch": 0.4692763067321371,
      "grad_norm": 0.46816888451576233,
      "learning_rate": 9.529455685120432e-06,
      "loss": 0.4068,
      "step": 10226
    },
    {
      "epoch": 0.4693221972373916,
      "grad_norm": 0.4750359058380127,
      "learning_rate": 9.529351839991215e-06,
      "loss": 0.4378,
      "step": 10227
    },
    {
      "epoch": 0.46936808774264605,
      "grad_norm": 0.45918720960617065,
      "learning_rate": 9.529247983970333e-06,
      "loss": 0.3774,
      "step": 10228
    },
    {
      "epoch": 0.4694139782479005,
      "grad_norm": 0.5184150338172913,
      "learning_rate": 9.529144117058041e-06,
      "loss": 0.3301,
      "step": 10229
    },
    {
      "epoch": 0.469459868753155,
      "grad_norm": 0.478085458278656,
      "learning_rate": 9.529040239254587e-06,
      "loss": 0.3812,
      "step": 10230
    },
    {
      "epoch": 0.4695057592584094,
      "grad_norm": 0.4576238691806793,
      "learning_rate": 9.52893635056022e-06,
      "loss": 0.3832,
      "step": 10231
    },
    {
      "epoch": 0.4695516497636639,
      "grad_norm": 0.44175854325294495,
      "learning_rate": 9.52883245097519e-06,
      "loss": 0.373,
      "step": 10232
    },
    {
      "epoch": 0.46959754026891837,
      "grad_norm": 0.4565293490886688,
      "learning_rate": 9.528728540499747e-06,
      "loss": 0.4443,
      "step": 10233
    },
    {
      "epoch": 0.4696434307741728,
      "grad_norm": 0.4721430242061615,
      "learning_rate": 9.528624619134143e-06,
      "loss": 0.427,
      "step": 10234
    },
    {
      "epoch": 0.4696893212794273,
      "grad_norm": 0.4381866455078125,
      "learning_rate": 9.528520686878624e-06,
      "loss": 0.3493,
      "step": 10235
    },
    {
      "epoch": 0.46973521178468175,
      "grad_norm": 0.5155766010284424,
      "learning_rate": 9.528416743733445e-06,
      "loss": 0.4975,
      "step": 10236
    },
    {
      "epoch": 0.4697811022899362,
      "grad_norm": 0.44764846563339233,
      "learning_rate": 9.52831278969885e-06,
      "loss": 0.3981,
      "step": 10237
    },
    {
      "epoch": 0.4698269927951907,
      "grad_norm": 0.4705737233161926,
      "learning_rate": 9.528208824775093e-06,
      "loss": 0.4123,
      "step": 10238
    },
    {
      "epoch": 0.46987288330044513,
      "grad_norm": 0.5041414499282837,
      "learning_rate": 9.528104848962423e-06,
      "loss": 0.4097,
      "step": 10239
    },
    {
      "epoch": 0.4699187738056996,
      "grad_norm": 0.4294908046722412,
      "learning_rate": 9.52800086226109e-06,
      "loss": 0.3758,
      "step": 10240
    },
    {
      "epoch": 0.46996466431095407,
      "grad_norm": 0.46984410285949707,
      "learning_rate": 9.527896864671343e-06,
      "loss": 0.4259,
      "step": 10241
    },
    {
      "epoch": 0.4700105548162085,
      "grad_norm": 0.4776791036128998,
      "learning_rate": 9.527792856193434e-06,
      "loss": 0.4033,
      "step": 10242
    },
    {
      "epoch": 0.470056445321463,
      "grad_norm": 0.4898414611816406,
      "learning_rate": 9.52768883682761e-06,
      "loss": 0.4652,
      "step": 10243
    },
    {
      "epoch": 0.47010233582671745,
      "grad_norm": 0.46027126908302307,
      "learning_rate": 9.527584806574127e-06,
      "loss": 0.3806,
      "step": 10244
    },
    {
      "epoch": 0.4701482263319719,
      "grad_norm": 0.4511317014694214,
      "learning_rate": 9.52748076543323e-06,
      "loss": 0.4283,
      "step": 10245
    },
    {
      "epoch": 0.4701941168372264,
      "grad_norm": 0.4421726167201996,
      "learning_rate": 9.527376713405169e-06,
      "loss": 0.3843,
      "step": 10246
    },
    {
      "epoch": 0.47024000734248084,
      "grad_norm": 0.4418502748012543,
      "learning_rate": 9.527272650490198e-06,
      "loss": 0.3694,
      "step": 10247
    },
    {
      "epoch": 0.4702858978477353,
      "grad_norm": 0.4714941084384918,
      "learning_rate": 9.527168576688565e-06,
      "loss": 0.3938,
      "step": 10248
    },
    {
      "epoch": 0.4703317883529898,
      "grad_norm": 0.49469122290611267,
      "learning_rate": 9.52706449200052e-06,
      "loss": 0.4831,
      "step": 10249
    },
    {
      "epoch": 0.4703776788582442,
      "grad_norm": 0.4973783493041992,
      "learning_rate": 9.526960396426313e-06,
      "loss": 0.4478,
      "step": 10250
    },
    {
      "epoch": 0.4704235693634987,
      "grad_norm": 0.48669660091400146,
      "learning_rate": 9.526856289966196e-06,
      "loss": 0.4281,
      "step": 10251
    },
    {
      "epoch": 0.47046945986875316,
      "grad_norm": 0.46653491258621216,
      "learning_rate": 9.526752172620417e-06,
      "loss": 0.446,
      "step": 10252
    },
    {
      "epoch": 0.4705153503740076,
      "grad_norm": 0.48437002301216125,
      "learning_rate": 9.52664804438923e-06,
      "loss": 0.4706,
      "step": 10253
    },
    {
      "epoch": 0.4705612408792621,
      "grad_norm": 0.4409119188785553,
      "learning_rate": 9.526543905272882e-06,
      "loss": 0.3761,
      "step": 10254
    },
    {
      "epoch": 0.47060713138451654,
      "grad_norm": 0.46961262822151184,
      "learning_rate": 9.526439755271624e-06,
      "loss": 0.4061,
      "step": 10255
    },
    {
      "epoch": 0.470653021889771,
      "grad_norm": 0.4741775095462799,
      "learning_rate": 9.526335594385707e-06,
      "loss": 0.469,
      "step": 10256
    },
    {
      "epoch": 0.4706989123950255,
      "grad_norm": 0.4921030104160309,
      "learning_rate": 9.526231422615382e-06,
      "loss": 0.426,
      "step": 10257
    },
    {
      "epoch": 0.4707448029002799,
      "grad_norm": 0.44580015540122986,
      "learning_rate": 9.526127239960899e-06,
      "loss": 0.4274,
      "step": 10258
    },
    {
      "epoch": 0.4707906934055344,
      "grad_norm": 0.4783933758735657,
      "learning_rate": 9.526023046422508e-06,
      "loss": 0.4075,
      "step": 10259
    },
    {
      "epoch": 0.47083658391078886,
      "grad_norm": 0.4656791090965271,
      "learning_rate": 9.52591884200046e-06,
      "loss": 0.4411,
      "step": 10260
    },
    {
      "epoch": 0.4708824744160433,
      "grad_norm": 0.4863322377204895,
      "learning_rate": 9.525814626695005e-06,
      "loss": 0.4826,
      "step": 10261
    },
    {
      "epoch": 0.4709283649212978,
      "grad_norm": 0.45109036564826965,
      "learning_rate": 9.525710400506395e-06,
      "loss": 0.421,
      "step": 10262
    },
    {
      "epoch": 0.47097425542655225,
      "grad_norm": 0.4539772868156433,
      "learning_rate": 9.525606163434879e-06,
      "loss": 0.3467,
      "step": 10263
    },
    {
      "epoch": 0.4710201459318067,
      "grad_norm": 0.4710644781589508,
      "learning_rate": 9.52550191548071e-06,
      "loss": 0.4561,
      "step": 10264
    },
    {
      "epoch": 0.4710660364370612,
      "grad_norm": 0.5383159518241882,
      "learning_rate": 9.525397656644137e-06,
      "loss": 0.5325,
      "step": 10265
    },
    {
      "epoch": 0.47111192694231563,
      "grad_norm": 0.5490121245384216,
      "learning_rate": 9.52529338692541e-06,
      "loss": 0.5469,
      "step": 10266
    },
    {
      "epoch": 0.47115781744757007,
      "grad_norm": 0.4448584020137787,
      "learning_rate": 9.525189106324781e-06,
      "loss": 0.402,
      "step": 10267
    },
    {
      "epoch": 0.47120370795282457,
      "grad_norm": 0.5430421829223633,
      "learning_rate": 9.5250848148425e-06,
      "loss": 0.4333,
      "step": 10268
    },
    {
      "epoch": 0.471249598458079,
      "grad_norm": 0.47605228424072266,
      "learning_rate": 9.524980512478819e-06,
      "loss": 0.4455,
      "step": 10269
    },
    {
      "epoch": 0.4712954889633335,
      "grad_norm": 0.4852641224861145,
      "learning_rate": 9.524876199233988e-06,
      "loss": 0.4392,
      "step": 10270
    },
    {
      "epoch": 0.47134137946858795,
      "grad_norm": 0.5136424899101257,
      "learning_rate": 9.524771875108255e-06,
      "loss": 0.4962,
      "step": 10271
    },
    {
      "epoch": 0.4713872699738424,
      "grad_norm": 0.4403326213359833,
      "learning_rate": 9.524667540101877e-06,
      "loss": 0.3098,
      "step": 10272
    },
    {
      "epoch": 0.4714331604790969,
      "grad_norm": 0.4392525255680084,
      "learning_rate": 9.524563194215101e-06,
      "loss": 0.3844,
      "step": 10273
    },
    {
      "epoch": 0.47147905098435133,
      "grad_norm": 0.4410269260406494,
      "learning_rate": 9.524458837448178e-06,
      "loss": 0.3493,
      "step": 10274
    },
    {
      "epoch": 0.4715249414896058,
      "grad_norm": 0.40997621417045593,
      "learning_rate": 9.52435446980136e-06,
      "loss": 0.3285,
      "step": 10275
    },
    {
      "epoch": 0.4715708319948603,
      "grad_norm": 0.4391801059246063,
      "learning_rate": 9.524250091274898e-06,
      "loss": 0.3405,
      "step": 10276
    },
    {
      "epoch": 0.4716167225001147,
      "grad_norm": 0.46179893612861633,
      "learning_rate": 9.524145701869041e-06,
      "loss": 0.391,
      "step": 10277
    },
    {
      "epoch": 0.4716626130053692,
      "grad_norm": 0.4987298250198364,
      "learning_rate": 9.524041301584042e-06,
      "loss": 0.4279,
      "step": 10278
    },
    {
      "epoch": 0.47170850351062366,
      "grad_norm": 0.42264726758003235,
      "learning_rate": 9.52393689042015e-06,
      "loss": 0.3186,
      "step": 10279
    },
    {
      "epoch": 0.4717543940158781,
      "grad_norm": 0.4643816351890564,
      "learning_rate": 9.523832468377621e-06,
      "loss": 0.4201,
      "step": 10280
    },
    {
      "epoch": 0.4718002845211326,
      "grad_norm": 0.4555171728134155,
      "learning_rate": 9.5237280354567e-06,
      "loss": 0.367,
      "step": 10281
    },
    {
      "epoch": 0.47184617502638704,
      "grad_norm": 0.4316999018192291,
      "learning_rate": 9.523623591657643e-06,
      "loss": 0.3305,
      "step": 10282
    },
    {
      "epoch": 0.4718920655316415,
      "grad_norm": 0.5543466210365295,
      "learning_rate": 9.523519136980696e-06,
      "loss": 0.6002,
      "step": 10283
    },
    {
      "epoch": 0.471937956036896,
      "grad_norm": 0.46315932273864746,
      "learning_rate": 9.523414671426115e-06,
      "loss": 0.3714,
      "step": 10284
    },
    {
      "epoch": 0.4719838465421504,
      "grad_norm": 0.45800137519836426,
      "learning_rate": 9.523310194994148e-06,
      "loss": 0.3558,
      "step": 10285
    },
    {
      "epoch": 0.4720297370474049,
      "grad_norm": 0.43640342354774475,
      "learning_rate": 9.523205707685049e-06,
      "loss": 0.3481,
      "step": 10286
    },
    {
      "epoch": 0.47207562755265936,
      "grad_norm": 0.5094229578971863,
      "learning_rate": 9.523101209499068e-06,
      "loss": 0.4948,
      "step": 10287
    },
    {
      "epoch": 0.4721215180579138,
      "grad_norm": 0.44438436627388,
      "learning_rate": 9.522996700436456e-06,
      "loss": 0.3976,
      "step": 10288
    },
    {
      "epoch": 0.4721674085631683,
      "grad_norm": 0.4255135655403137,
      "learning_rate": 9.522892180497462e-06,
      "loss": 0.3419,
      "step": 10289
    },
    {
      "epoch": 0.47221329906842274,
      "grad_norm": 0.44764408469200134,
      "learning_rate": 9.522787649682342e-06,
      "loss": 0.322,
      "step": 10290
    },
    {
      "epoch": 0.4722591895736772,
      "grad_norm": 0.43707945942878723,
      "learning_rate": 9.522683107991345e-06,
      "loss": 0.3792,
      "step": 10291
    },
    {
      "epoch": 0.4723050800789317,
      "grad_norm": 0.458928644657135,
      "learning_rate": 9.522578555424722e-06,
      "loss": 0.4478,
      "step": 10292
    },
    {
      "epoch": 0.4723509705841861,
      "grad_norm": 0.4472564160823822,
      "learning_rate": 9.522473991982726e-06,
      "loss": 0.4004,
      "step": 10293
    },
    {
      "epoch": 0.47239686108944057,
      "grad_norm": 0.5485993027687073,
      "learning_rate": 9.522369417665606e-06,
      "loss": 0.4856,
      "step": 10294
    },
    {
      "epoch": 0.47244275159469507,
      "grad_norm": 0.4359759986400604,
      "learning_rate": 9.522264832473615e-06,
      "loss": 0.3413,
      "step": 10295
    },
    {
      "epoch": 0.4724886420999495,
      "grad_norm": 0.4886433184146881,
      "learning_rate": 9.522160236407005e-06,
      "loss": 0.4406,
      "step": 10296
    },
    {
      "epoch": 0.472534532605204,
      "grad_norm": 0.4512285590171814,
      "learning_rate": 9.522055629466028e-06,
      "loss": 0.3789,
      "step": 10297
    },
    {
      "epoch": 0.47258042311045845,
      "grad_norm": 0.5037420988082886,
      "learning_rate": 9.521951011650932e-06,
      "loss": 0.499,
      "step": 10298
    },
    {
      "epoch": 0.4726263136157129,
      "grad_norm": 0.5104064345359802,
      "learning_rate": 9.521846382961973e-06,
      "loss": 0.3686,
      "step": 10299
    },
    {
      "epoch": 0.4726722041209674,
      "grad_norm": 0.44444432854652405,
      "learning_rate": 9.521741743399399e-06,
      "loss": 0.3436,
      "step": 10300
    },
    {
      "epoch": 0.47271809462622183,
      "grad_norm": 0.4367820918560028,
      "learning_rate": 9.521637092963465e-06,
      "loss": 0.3725,
      "step": 10301
    },
    {
      "epoch": 0.47276398513147627,
      "grad_norm": 0.47296854853630066,
      "learning_rate": 9.521532431654418e-06,
      "loss": 0.4431,
      "step": 10302
    },
    {
      "epoch": 0.47280987563673077,
      "grad_norm": 0.43497464060783386,
      "learning_rate": 9.521427759472516e-06,
      "loss": 0.3652,
      "step": 10303
    },
    {
      "epoch": 0.4728557661419852,
      "grad_norm": 0.4277838468551636,
      "learning_rate": 9.521323076418006e-06,
      "loss": 0.3574,
      "step": 10304
    },
    {
      "epoch": 0.4729016566472397,
      "grad_norm": 0.45304176211357117,
      "learning_rate": 9.52121838249114e-06,
      "loss": 0.3415,
      "step": 10305
    },
    {
      "epoch": 0.47294754715249415,
      "grad_norm": 0.5273206233978271,
      "learning_rate": 9.52111367769217e-06,
      "loss": 0.4002,
      "step": 10306
    },
    {
      "epoch": 0.4729934376577486,
      "grad_norm": 0.4759596288204193,
      "learning_rate": 9.52100896202135e-06,
      "loss": 0.4662,
      "step": 10307
    },
    {
      "epoch": 0.4730393281630031,
      "grad_norm": 0.44349730014801025,
      "learning_rate": 9.520904235478932e-06,
      "loss": 0.329,
      "step": 10308
    },
    {
      "epoch": 0.47308521866825753,
      "grad_norm": 0.4337792694568634,
      "learning_rate": 9.520799498065162e-06,
      "loss": 0.374,
      "step": 10309
    },
    {
      "epoch": 0.473131109173512,
      "grad_norm": 0.43011367321014404,
      "learning_rate": 9.520694749780299e-06,
      "loss": 0.3557,
      "step": 10310
    },
    {
      "epoch": 0.4731769996787665,
      "grad_norm": 0.4629124104976654,
      "learning_rate": 9.520589990624591e-06,
      "loss": 0.442,
      "step": 10311
    },
    {
      "epoch": 0.4732228901840209,
      "grad_norm": 0.42186805605888367,
      "learning_rate": 9.520485220598291e-06,
      "loss": 0.3164,
      "step": 10312
    },
    {
      "epoch": 0.4732687806892754,
      "grad_norm": 0.4764348268508911,
      "learning_rate": 9.52038043970165e-06,
      "loss": 0.4198,
      "step": 10313
    },
    {
      "epoch": 0.47331467119452986,
      "grad_norm": 0.47077757120132446,
      "learning_rate": 9.520275647934922e-06,
      "loss": 0.4076,
      "step": 10314
    },
    {
      "epoch": 0.4733605616997843,
      "grad_norm": 0.443193256855011,
      "learning_rate": 9.520170845298358e-06,
      "loss": 0.3577,
      "step": 10315
    },
    {
      "epoch": 0.4734064522050388,
      "grad_norm": 0.4883959889411926,
      "learning_rate": 9.520066031792209e-06,
      "loss": 0.4595,
      "step": 10316
    },
    {
      "epoch": 0.47345234271029324,
      "grad_norm": 0.45426440238952637,
      "learning_rate": 9.519961207416727e-06,
      "loss": 0.3288,
      "step": 10317
    },
    {
      "epoch": 0.4734982332155477,
      "grad_norm": 0.44965294003486633,
      "learning_rate": 9.519856372172165e-06,
      "loss": 0.4036,
      "step": 10318
    },
    {
      "epoch": 0.4735441237208022,
      "grad_norm": 0.45954135060310364,
      "learning_rate": 9.519751526058776e-06,
      "loss": 0.4165,
      "step": 10319
    },
    {
      "epoch": 0.4735900142260566,
      "grad_norm": 0.45515939593315125,
      "learning_rate": 9.51964666907681e-06,
      "loss": 0.3682,
      "step": 10320
    },
    {
      "epoch": 0.4736359047313111,
      "grad_norm": 0.5606639385223389,
      "learning_rate": 9.519541801226522e-06,
      "loss": 0.5414,
      "step": 10321
    },
    {
      "epoch": 0.47368179523656556,
      "grad_norm": 0.5108808279037476,
      "learning_rate": 9.519436922508161e-06,
      "loss": 0.4288,
      "step": 10322
    },
    {
      "epoch": 0.47372768574182,
      "grad_norm": 0.4635060131549835,
      "learning_rate": 9.51933203292198e-06,
      "loss": 0.3776,
      "step": 10323
    },
    {
      "epoch": 0.4737735762470745,
      "grad_norm": 0.43096524477005005,
      "learning_rate": 9.519227132468232e-06,
      "loss": 0.336,
      "step": 10324
    },
    {
      "epoch": 0.47381946675232894,
      "grad_norm": 0.42057979106903076,
      "learning_rate": 9.51912222114717e-06,
      "loss": 0.3425,
      "step": 10325
    },
    {
      "epoch": 0.4738653572575834,
      "grad_norm": 0.45328864455223083,
      "learning_rate": 9.519017298959045e-06,
      "loss": 0.3719,
      "step": 10326
    },
    {
      "epoch": 0.4739112477628379,
      "grad_norm": 0.4315711259841919,
      "learning_rate": 9.518912365904109e-06,
      "loss": 0.3673,
      "step": 10327
    },
    {
      "epoch": 0.4739571382680923,
      "grad_norm": 0.4419160783290863,
      "learning_rate": 9.518807421982616e-06,
      "loss": 0.3609,
      "step": 10328
    },
    {
      "epoch": 0.47400302877334677,
      "grad_norm": 0.44125065207481384,
      "learning_rate": 9.518702467194816e-06,
      "loss": 0.4238,
      "step": 10329
    },
    {
      "epoch": 0.47404891927860127,
      "grad_norm": 0.445504367351532,
      "learning_rate": 9.518597501540965e-06,
      "loss": 0.3638,
      "step": 10330
    },
    {
      "epoch": 0.4740948097838557,
      "grad_norm": 0.46772706508636475,
      "learning_rate": 9.51849252502131e-06,
      "loss": 0.4668,
      "step": 10331
    },
    {
      "epoch": 0.4741407002891102,
      "grad_norm": 0.48574432730674744,
      "learning_rate": 9.51838753763611e-06,
      "loss": 0.4196,
      "step": 10332
    },
    {
      "epoch": 0.47418659079436465,
      "grad_norm": 0.44626715779304504,
      "learning_rate": 9.518282539385611e-06,
      "loss": 0.3901,
      "step": 10333
    },
    {
      "epoch": 0.4742324812996191,
      "grad_norm": 0.42279523611068726,
      "learning_rate": 9.518177530270069e-06,
      "loss": 0.3583,
      "step": 10334
    },
    {
      "epoch": 0.4742783718048736,
      "grad_norm": 0.45353010296821594,
      "learning_rate": 9.518072510289737e-06,
      "loss": 0.4176,
      "step": 10335
    },
    {
      "epoch": 0.47432426231012803,
      "grad_norm": 0.43583959341049194,
      "learning_rate": 9.517967479444867e-06,
      "loss": 0.3751,
      "step": 10336
    },
    {
      "epoch": 0.4743701528153825,
      "grad_norm": 0.4660477340221405,
      "learning_rate": 9.517862437735709e-06,
      "loss": 0.4001,
      "step": 10337
    },
    {
      "epoch": 0.47441604332063697,
      "grad_norm": 0.4626091420650482,
      "learning_rate": 9.51775738516252e-06,
      "loss": 0.3803,
      "step": 10338
    },
    {
      "epoch": 0.4744619338258914,
      "grad_norm": 0.44223880767822266,
      "learning_rate": 9.51765232172555e-06,
      "loss": 0.3913,
      "step": 10339
    },
    {
      "epoch": 0.4745078243311459,
      "grad_norm": 0.44942721724510193,
      "learning_rate": 9.517547247425051e-06,
      "loss": 0.4065,
      "step": 10340
    },
    {
      "epoch": 0.47455371483640035,
      "grad_norm": 0.4664980471134186,
      "learning_rate": 9.517442162261277e-06,
      "loss": 0.4304,
      "step": 10341
    },
    {
      "epoch": 0.4745996053416548,
      "grad_norm": 0.433765172958374,
      "learning_rate": 9.51733706623448e-06,
      "loss": 0.3561,
      "step": 10342
    },
    {
      "epoch": 0.4746454958469093,
      "grad_norm": 0.49178147315979004,
      "learning_rate": 9.517231959344914e-06,
      "loss": 0.4427,
      "step": 10343
    },
    {
      "epoch": 0.47469138635216374,
      "grad_norm": 0.43830791115760803,
      "learning_rate": 9.51712684159283e-06,
      "loss": 0.3729,
      "step": 10344
    },
    {
      "epoch": 0.4747372768574182,
      "grad_norm": 0.4705291986465454,
      "learning_rate": 9.517021712978482e-06,
      "loss": 0.4293,
      "step": 10345
    },
    {
      "epoch": 0.4747831673626727,
      "grad_norm": 0.4658603370189667,
      "learning_rate": 9.516916573502121e-06,
      "loss": 0.4178,
      "step": 10346
    },
    {
      "epoch": 0.4748290578679271,
      "grad_norm": 0.49700331687927246,
      "learning_rate": 9.516811423164003e-06,
      "loss": 0.4768,
      "step": 10347
    },
    {
      "epoch": 0.4748749483731816,
      "grad_norm": 0.45584195852279663,
      "learning_rate": 9.51670626196438e-06,
      "loss": 0.3917,
      "step": 10348
    },
    {
      "epoch": 0.47492083887843606,
      "grad_norm": 0.5152073502540588,
      "learning_rate": 9.516601089903502e-06,
      "loss": 0.5275,
      "step": 10349
    },
    {
      "epoch": 0.4749667293836905,
      "grad_norm": 0.41748711466789246,
      "learning_rate": 9.516495906981624e-06,
      "loss": 0.3217,
      "step": 10350
    },
    {
      "epoch": 0.475012619888945,
      "grad_norm": 0.4511125981807709,
      "learning_rate": 9.516390713198999e-06,
      "loss": 0.4312,
      "step": 10351
    },
    {
      "epoch": 0.47505851039419944,
      "grad_norm": 0.4734383523464203,
      "learning_rate": 9.51628550855588e-06,
      "loss": 0.3898,
      "step": 10352
    },
    {
      "epoch": 0.4751044008994539,
      "grad_norm": 0.49554720520973206,
      "learning_rate": 9.51618029305252e-06,
      "loss": 0.4827,
      "step": 10353
    },
    {
      "epoch": 0.4751502914047084,
      "grad_norm": 0.43140262365341187,
      "learning_rate": 9.516075066689169e-06,
      "loss": 0.3552,
      "step": 10354
    },
    {
      "epoch": 0.4751961819099628,
      "grad_norm": 0.4343739151954651,
      "learning_rate": 9.515969829466087e-06,
      "loss": 0.3541,
      "step": 10355
    },
    {
      "epoch": 0.47524207241521726,
      "grad_norm": 0.49409523606300354,
      "learning_rate": 9.515864581383521e-06,
      "loss": 0.447,
      "step": 10356
    },
    {
      "epoch": 0.47528796292047176,
      "grad_norm": 0.4462185502052307,
      "learning_rate": 9.515759322441727e-06,
      "loss": 0.3996,
      "step": 10357
    },
    {
      "epoch": 0.4753338534257262,
      "grad_norm": 0.450898140668869,
      "learning_rate": 9.515654052640955e-06,
      "loss": 0.4085,
      "step": 10358
    },
    {
      "epoch": 0.4753797439309807,
      "grad_norm": 0.5207261443138123,
      "learning_rate": 9.515548771981461e-06,
      "loss": 0.4161,
      "step": 10359
    },
    {
      "epoch": 0.47542563443623514,
      "grad_norm": 0.4385071396827698,
      "learning_rate": 9.515443480463498e-06,
      "loss": 0.3645,
      "step": 10360
    },
    {
      "epoch": 0.4754715249414896,
      "grad_norm": 0.5149874687194824,
      "learning_rate": 9.515338178087316e-06,
      "loss": 0.5163,
      "step": 10361
    },
    {
      "epoch": 0.4755174154467441,
      "grad_norm": 0.4715368151664734,
      "learning_rate": 9.515232864853175e-06,
      "loss": 0.4108,
      "step": 10362
    },
    {
      "epoch": 0.4755633059519985,
      "grad_norm": 0.4490543603897095,
      "learning_rate": 9.515127540761321e-06,
      "loss": 0.3611,
      "step": 10363
    },
    {
      "epoch": 0.47560919645725297,
      "grad_norm": 0.45235684514045715,
      "learning_rate": 9.515022205812011e-06,
      "loss": 0.4326,
      "step": 10364
    },
    {
      "epoch": 0.47565508696250747,
      "grad_norm": 0.45228105783462524,
      "learning_rate": 9.514916860005497e-06,
      "loss": 0.35,
      "step": 10365
    },
    {
      "epoch": 0.4757009774677619,
      "grad_norm": 0.4305766522884369,
      "learning_rate": 9.514811503342033e-06,
      "loss": 0.3919,
      "step": 10366
    },
    {
      "epoch": 0.4757468679730164,
      "grad_norm": 0.4573293924331665,
      "learning_rate": 9.514706135821874e-06,
      "loss": 0.3812,
      "step": 10367
    },
    {
      "epoch": 0.47579275847827085,
      "grad_norm": 0.45627254247665405,
      "learning_rate": 9.51460075744527e-06,
      "loss": 0.3904,
      "step": 10368
    },
    {
      "epoch": 0.4758386489835253,
      "grad_norm": 0.4555344879627228,
      "learning_rate": 9.514495368212478e-06,
      "loss": 0.4001,
      "step": 10369
    },
    {
      "epoch": 0.4758845394887798,
      "grad_norm": 0.48171466588974,
      "learning_rate": 9.514389968123747e-06,
      "loss": 0.4767,
      "step": 10370
    },
    {
      "epoch": 0.47593042999403423,
      "grad_norm": 0.47452256083488464,
      "learning_rate": 9.514284557179334e-06,
      "loss": 0.4533,
      "step": 10371
    },
    {
      "epoch": 0.4759763204992887,
      "grad_norm": 0.4739775061607361,
      "learning_rate": 9.51417913537949e-06,
      "loss": 0.4212,
      "step": 10372
    },
    {
      "epoch": 0.47602221100454317,
      "grad_norm": 0.4628604054450989,
      "learning_rate": 9.514073702724473e-06,
      "loss": 0.417,
      "step": 10373
    },
    {
      "epoch": 0.4760681015097976,
      "grad_norm": 0.4621262550354004,
      "learning_rate": 9.513968259214531e-06,
      "loss": 0.3887,
      "step": 10374
    },
    {
      "epoch": 0.4761139920150521,
      "grad_norm": 0.45815667510032654,
      "learning_rate": 9.51386280484992e-06,
      "loss": 0.4496,
      "step": 10375
    },
    {
      "epoch": 0.47615988252030655,
      "grad_norm": 0.47607067227363586,
      "learning_rate": 9.513757339630895e-06,
      "loss": 0.4337,
      "step": 10376
    },
    {
      "epoch": 0.476205773025561,
      "grad_norm": 0.4208063781261444,
      "learning_rate": 9.513651863557707e-06,
      "loss": 0.324,
      "step": 10377
    },
    {
      "epoch": 0.4762516635308155,
      "grad_norm": 0.41242238879203796,
      "learning_rate": 9.513546376630612e-06,
      "loss": 0.3007,
      "step": 10378
    },
    {
      "epoch": 0.47629755403606994,
      "grad_norm": 0.46094265580177307,
      "learning_rate": 9.51344087884986e-06,
      "loss": 0.3832,
      "step": 10379
    },
    {
      "epoch": 0.4763434445413244,
      "grad_norm": 0.4570407569408417,
      "learning_rate": 9.513335370215709e-06,
      "loss": 0.4495,
      "step": 10380
    },
    {
      "epoch": 0.4763893350465789,
      "grad_norm": 0.451535165309906,
      "learning_rate": 9.513229850728411e-06,
      "loss": 0.4328,
      "step": 10381
    },
    {
      "epoch": 0.4764352255518333,
      "grad_norm": 0.4502072334289551,
      "learning_rate": 9.513124320388218e-06,
      "loss": 0.4257,
      "step": 10382
    },
    {
      "epoch": 0.47648111605708776,
      "grad_norm": 0.49615052342414856,
      "learning_rate": 9.513018779195387e-06,
      "loss": 0.4467,
      "step": 10383
    },
    {
      "epoch": 0.47652700656234226,
      "grad_norm": 0.43216705322265625,
      "learning_rate": 9.512913227150171e-06,
      "loss": 0.3653,
      "step": 10384
    },
    {
      "epoch": 0.4765728970675967,
      "grad_norm": 0.4520290493965149,
      "learning_rate": 9.512807664252821e-06,
      "loss": 0.363,
      "step": 10385
    },
    {
      "epoch": 0.4766187875728512,
      "grad_norm": 0.4575624465942383,
      "learning_rate": 9.512702090503593e-06,
      "loss": 0.4093,
      "step": 10386
    },
    {
      "epoch": 0.47666467807810564,
      "grad_norm": 0.4549427032470703,
      "learning_rate": 9.512596505902742e-06,
      "loss": 0.4027,
      "step": 10387
    },
    {
      "epoch": 0.4767105685833601,
      "grad_norm": 0.44714000821113586,
      "learning_rate": 9.512490910450521e-06,
      "loss": 0.3318,
      "step": 10388
    },
    {
      "epoch": 0.4767564590886146,
      "grad_norm": 0.474368691444397,
      "learning_rate": 9.512385304147183e-06,
      "loss": 0.4436,
      "step": 10389
    },
    {
      "epoch": 0.476802349593869,
      "grad_norm": 0.4662497639656067,
      "learning_rate": 9.512279686992982e-06,
      "loss": 0.3604,
      "step": 10390
    },
    {
      "epoch": 0.47684824009912347,
      "grad_norm": 0.4281292259693146,
      "learning_rate": 9.512174058988171e-06,
      "loss": 0.304,
      "step": 10391
    },
    {
      "epoch": 0.47689413060437796,
      "grad_norm": 0.4604879915714264,
      "learning_rate": 9.512068420133006e-06,
      "loss": 0.3846,
      "step": 10392
    },
    {
      "epoch": 0.4769400211096324,
      "grad_norm": 0.49535027146339417,
      "learning_rate": 9.511962770427742e-06,
      "loss": 0.3673,
      "step": 10393
    },
    {
      "epoch": 0.4769859116148869,
      "grad_norm": 0.5369259715080261,
      "learning_rate": 9.511857109872633e-06,
      "loss": 0.4326,
      "step": 10394
    },
    {
      "epoch": 0.47703180212014135,
      "grad_norm": 0.4738069474697113,
      "learning_rate": 9.511751438467928e-06,
      "loss": 0.3636,
      "step": 10395
    },
    {
      "epoch": 0.4770776926253958,
      "grad_norm": 0.46687376499176025,
      "learning_rate": 9.511645756213887e-06,
      "loss": 0.4106,
      "step": 10396
    },
    {
      "epoch": 0.4771235831306503,
      "grad_norm": 0.46414706110954285,
      "learning_rate": 9.511540063110763e-06,
      "loss": 0.4249,
      "step": 10397
    },
    {
      "epoch": 0.47716947363590473,
      "grad_norm": 0.453397661447525,
      "learning_rate": 9.511434359158807e-06,
      "loss": 0.3995,
      "step": 10398
    },
    {
      "epoch": 0.47721536414115917,
      "grad_norm": 0.44518303871154785,
      "learning_rate": 9.511328644358275e-06,
      "loss": 0.3732,
      "step": 10399
    },
    {
      "epoch": 0.47726125464641367,
      "grad_norm": 0.46980151534080505,
      "learning_rate": 9.511222918709422e-06,
      "loss": 0.4471,
      "step": 10400
    },
    {
      "epoch": 0.4773071451516681,
      "grad_norm": 0.46569621562957764,
      "learning_rate": 9.511117182212502e-06,
      "loss": 0.3176,
      "step": 10401
    },
    {
      "epoch": 0.4773530356569226,
      "grad_norm": 0.4824908673763275,
      "learning_rate": 9.511011434867768e-06,
      "loss": 0.4412,
      "step": 10402
    },
    {
      "epoch": 0.47739892616217705,
      "grad_norm": 0.44572189450263977,
      "learning_rate": 9.510905676675476e-06,
      "loss": 0.3869,
      "step": 10403
    },
    {
      "epoch": 0.4774448166674315,
      "grad_norm": 0.45096150040626526,
      "learning_rate": 9.510799907635879e-06,
      "loss": 0.3817,
      "step": 10404
    },
    {
      "epoch": 0.477490707172686,
      "grad_norm": 0.46703019738197327,
      "learning_rate": 9.510694127749232e-06,
      "loss": 0.4425,
      "step": 10405
    },
    {
      "epoch": 0.47753659767794043,
      "grad_norm": 0.47779399156570435,
      "learning_rate": 9.510588337015789e-06,
      "loss": 0.4049,
      "step": 10406
    },
    {
      "epoch": 0.4775824881831949,
      "grad_norm": 0.49011799693107605,
      "learning_rate": 9.510482535435805e-06,
      "loss": 0.439,
      "step": 10407
    },
    {
      "epoch": 0.4776283786884494,
      "grad_norm": 0.4545629620552063,
      "learning_rate": 9.510376723009535e-06,
      "loss": 0.3812,
      "step": 10408
    },
    {
      "epoch": 0.4776742691937038,
      "grad_norm": 0.41130968928337097,
      "learning_rate": 9.51027089973723e-06,
      "loss": 0.3595,
      "step": 10409
    },
    {
      "epoch": 0.4777201596989583,
      "grad_norm": 0.37521857023239136,
      "learning_rate": 9.51016506561915e-06,
      "loss": 0.2613,
      "step": 10410
    },
    {
      "epoch": 0.47776605020421276,
      "grad_norm": 0.4440804719924927,
      "learning_rate": 9.510059220655546e-06,
      "loss": 0.3196,
      "step": 10411
    },
    {
      "epoch": 0.4778119407094672,
      "grad_norm": 0.4269392192363739,
      "learning_rate": 9.50995336484667e-06,
      "loss": 0.3093,
      "step": 10412
    },
    {
      "epoch": 0.4778578312147217,
      "grad_norm": 0.4689266085624695,
      "learning_rate": 9.509847498192782e-06,
      "loss": 0.3996,
      "step": 10413
    },
    {
      "epoch": 0.47790372171997614,
      "grad_norm": 0.5067508816719055,
      "learning_rate": 9.509741620694135e-06,
      "loss": 0.4835,
      "step": 10414
    },
    {
      "epoch": 0.4779496122252306,
      "grad_norm": 0.5038485527038574,
      "learning_rate": 9.509635732350982e-06,
      "loss": 0.4798,
      "step": 10415
    },
    {
      "epoch": 0.4779955027304851,
      "grad_norm": 0.5294581055641174,
      "learning_rate": 9.509529833163576e-06,
      "loss": 0.5106,
      "step": 10416
    },
    {
      "epoch": 0.4780413932357395,
      "grad_norm": 0.46726858615875244,
      "learning_rate": 9.509423923132177e-06,
      "loss": 0.4051,
      "step": 10417
    },
    {
      "epoch": 0.47808728374099396,
      "grad_norm": 0.5157381296157837,
      "learning_rate": 9.509318002257036e-06,
      "loss": 0.485,
      "step": 10418
    },
    {
      "epoch": 0.47813317424624846,
      "grad_norm": 0.46835318207740784,
      "learning_rate": 9.509212070538407e-06,
      "loss": 0.3957,
      "step": 10419
    },
    {
      "epoch": 0.4781790647515029,
      "grad_norm": 0.47847121953964233,
      "learning_rate": 9.509106127976546e-06,
      "loss": 0.4101,
      "step": 10420
    },
    {
      "epoch": 0.4782249552567574,
      "grad_norm": 0.502334475517273,
      "learning_rate": 9.50900017457171e-06,
      "loss": 0.5169,
      "step": 10421
    },
    {
      "epoch": 0.47827084576201184,
      "grad_norm": 0.45988163352012634,
      "learning_rate": 9.50889421032415e-06,
      "loss": 0.4024,
      "step": 10422
    },
    {
      "epoch": 0.4783167362672663,
      "grad_norm": 0.496215283870697,
      "learning_rate": 9.508788235234123e-06,
      "loss": 0.4842,
      "step": 10423
    },
    {
      "epoch": 0.4783626267725208,
      "grad_norm": 0.43937432765960693,
      "learning_rate": 9.508682249301883e-06,
      "loss": 0.3463,
      "step": 10424
    },
    {
      "epoch": 0.4784085172777752,
      "grad_norm": 0.4643916189670563,
      "learning_rate": 9.508576252527686e-06,
      "loss": 0.4199,
      "step": 10425
    },
    {
      "epoch": 0.47845440778302967,
      "grad_norm": 0.49748629331588745,
      "learning_rate": 9.508470244911784e-06,
      "loss": 0.4291,
      "step": 10426
    },
    {
      "epoch": 0.47850029828828416,
      "grad_norm": 0.4719739556312561,
      "learning_rate": 9.508364226454436e-06,
      "loss": 0.3515,
      "step": 10427
    },
    {
      "epoch": 0.4785461887935386,
      "grad_norm": 0.46450793743133545,
      "learning_rate": 9.508258197155896e-06,
      "loss": 0.3917,
      "step": 10428
    },
    {
      "epoch": 0.4785920792987931,
      "grad_norm": 0.4392354190349579,
      "learning_rate": 9.508152157016415e-06,
      "loss": 0.3933,
      "step": 10429
    },
    {
      "epoch": 0.47863796980404755,
      "grad_norm": 0.47227659821510315,
      "learning_rate": 9.508046106036253e-06,
      "loss": 0.4181,
      "step": 10430
    },
    {
      "epoch": 0.478683860309302,
      "grad_norm": 0.4628347158432007,
      "learning_rate": 9.50794004421566e-06,
      "loss": 0.4008,
      "step": 10431
    },
    {
      "epoch": 0.4787297508145565,
      "grad_norm": 0.5100054144859314,
      "learning_rate": 9.507833971554898e-06,
      "loss": 0.4806,
      "step": 10432
    },
    {
      "epoch": 0.47877564131981093,
      "grad_norm": 0.4490429759025574,
      "learning_rate": 9.507727888054215e-06,
      "loss": 0.3962,
      "step": 10433
    },
    {
      "epoch": 0.47882153182506537,
      "grad_norm": 0.4828034043312073,
      "learning_rate": 9.507621793713871e-06,
      "loss": 0.462,
      "step": 10434
    },
    {
      "epoch": 0.47886742233031987,
      "grad_norm": 0.4738355576992035,
      "learning_rate": 9.507515688534119e-06,
      "loss": 0.4342,
      "step": 10435
    },
    {
      "epoch": 0.4789133128355743,
      "grad_norm": 0.4522918164730072,
      "learning_rate": 9.507409572515211e-06,
      "loss": 0.3792,
      "step": 10436
    },
    {
      "epoch": 0.4789592033408288,
      "grad_norm": 0.4803236126899719,
      "learning_rate": 9.50730344565741e-06,
      "loss": 0.4594,
      "step": 10437
    },
    {
      "epoch": 0.47900509384608325,
      "grad_norm": 0.433921217918396,
      "learning_rate": 9.507197307960966e-06,
      "loss": 0.311,
      "step": 10438
    },
    {
      "epoch": 0.4790509843513377,
      "grad_norm": 0.5136846899986267,
      "learning_rate": 9.507091159426134e-06,
      "loss": 0.4486,
      "step": 10439
    },
    {
      "epoch": 0.4790968748565922,
      "grad_norm": 0.4746669828891754,
      "learning_rate": 9.506985000053168e-06,
      "loss": 0.4328,
      "step": 10440
    },
    {
      "epoch": 0.47914276536184663,
      "grad_norm": 0.4642753601074219,
      "learning_rate": 9.506878829842329e-06,
      "loss": 0.3835,
      "step": 10441
    },
    {
      "epoch": 0.4791886558671011,
      "grad_norm": 0.45964664220809937,
      "learning_rate": 9.506772648793866e-06,
      "loss": 0.3879,
      "step": 10442
    },
    {
      "epoch": 0.4792345463723556,
      "grad_norm": 0.48215681314468384,
      "learning_rate": 9.50666645690804e-06,
      "loss": 0.4792,
      "step": 10443
    },
    {
      "epoch": 0.47928043687761,
      "grad_norm": 0.4566616714000702,
      "learning_rate": 9.506560254185103e-06,
      "loss": 0.3838,
      "step": 10444
    },
    {
      "epoch": 0.47932632738286446,
      "grad_norm": 0.4943268895149231,
      "learning_rate": 9.50645404062531e-06,
      "loss": 0.4459,
      "step": 10445
    },
    {
      "epoch": 0.47937221788811896,
      "grad_norm": 0.4451558291912079,
      "learning_rate": 9.506347816228918e-06,
      "loss": 0.3927,
      "step": 10446
    },
    {
      "epoch": 0.4794181083933734,
      "grad_norm": 0.4583683907985687,
      "learning_rate": 9.506241580996181e-06,
      "loss": 0.4457,
      "step": 10447
    },
    {
      "epoch": 0.4794639988986279,
      "grad_norm": 0.44078630208969116,
      "learning_rate": 9.506135334927355e-06,
      "loss": 0.3949,
      "step": 10448
    },
    {
      "epoch": 0.47950988940388234,
      "grad_norm": 0.41707995533943176,
      "learning_rate": 9.506029078022696e-06,
      "loss": 0.3386,
      "step": 10449
    },
    {
      "epoch": 0.4795557799091368,
      "grad_norm": 0.4449442923069,
      "learning_rate": 9.50592281028246e-06,
      "loss": 0.3752,
      "step": 10450
    },
    {
      "epoch": 0.4796016704143913,
      "grad_norm": 0.3976631164550781,
      "learning_rate": 9.505816531706901e-06,
      "loss": 0.27,
      "step": 10451
    },
    {
      "epoch": 0.4796475609196457,
      "grad_norm": 0.45820823311805725,
      "learning_rate": 9.505710242296275e-06,
      "loss": 0.378,
      "step": 10452
    },
    {
      "epoch": 0.47969345142490016,
      "grad_norm": 0.49073418974876404,
      "learning_rate": 9.505603942050839e-06,
      "loss": 0.4245,
      "step": 10453
    },
    {
      "epoch": 0.47973934193015466,
      "grad_norm": 0.4454165995121002,
      "learning_rate": 9.505497630970845e-06,
      "loss": 0.3778,
      "step": 10454
    },
    {
      "epoch": 0.4797852324354091,
      "grad_norm": 0.5252615809440613,
      "learning_rate": 9.505391309056555e-06,
      "loss": 0.5217,
      "step": 10455
    },
    {
      "epoch": 0.4798311229406636,
      "grad_norm": 0.49363264441490173,
      "learning_rate": 9.505284976308218e-06,
      "loss": 0.5159,
      "step": 10456
    },
    {
      "epoch": 0.47987701344591804,
      "grad_norm": 0.4703807830810547,
      "learning_rate": 9.505178632726093e-06,
      "loss": 0.4229,
      "step": 10457
    },
    {
      "epoch": 0.4799229039511725,
      "grad_norm": 0.4552096426486969,
      "learning_rate": 9.505072278310437e-06,
      "loss": 0.436,
      "step": 10458
    },
    {
      "epoch": 0.479968794456427,
      "grad_norm": 0.45829489827156067,
      "learning_rate": 9.5049659130615e-06,
      "loss": 0.3731,
      "step": 10459
    },
    {
      "epoch": 0.4800146849616814,
      "grad_norm": 0.445065438747406,
      "learning_rate": 9.504859536979546e-06,
      "loss": 0.3966,
      "step": 10460
    },
    {
      "epoch": 0.48006057546693587,
      "grad_norm": 0.5642270445823669,
      "learning_rate": 9.504753150064825e-06,
      "loss": 0.4826,
      "step": 10461
    },
    {
      "epoch": 0.48010646597219037,
      "grad_norm": 0.464288592338562,
      "learning_rate": 9.504646752317594e-06,
      "loss": 0.4127,
      "step": 10462
    },
    {
      "epoch": 0.4801523564774448,
      "grad_norm": 0.44272831082344055,
      "learning_rate": 9.504540343738111e-06,
      "loss": 0.4207,
      "step": 10463
    },
    {
      "epoch": 0.4801982469826993,
      "grad_norm": 0.43861475586891174,
      "learning_rate": 9.504433924326628e-06,
      "loss": 0.3548,
      "step": 10464
    },
    {
      "epoch": 0.48024413748795375,
      "grad_norm": 0.466901957988739,
      "learning_rate": 9.504327494083405e-06,
      "loss": 0.4029,
      "step": 10465
    },
    {
      "epoch": 0.4802900279932082,
      "grad_norm": 0.46454697847366333,
      "learning_rate": 9.504221053008694e-06,
      "loss": 0.3889,
      "step": 10466
    },
    {
      "epoch": 0.4803359184984627,
      "grad_norm": 0.43511927127838135,
      "learning_rate": 9.504114601102755e-06,
      "loss": 0.3384,
      "step": 10467
    },
    {
      "epoch": 0.48038180900371713,
      "grad_norm": 0.4771687388420105,
      "learning_rate": 9.50400813836584e-06,
      "loss": 0.4094,
      "step": 10468
    },
    {
      "epoch": 0.4804276995089716,
      "grad_norm": 0.4551457166671753,
      "learning_rate": 9.503901664798208e-06,
      "loss": 0.3711,
      "step": 10469
    },
    {
      "epoch": 0.48047359001422607,
      "grad_norm": 0.4445315897464752,
      "learning_rate": 9.503795180400113e-06,
      "loss": 0.3907,
      "step": 10470
    },
    {
      "epoch": 0.4805194805194805,
      "grad_norm": 0.48052436113357544,
      "learning_rate": 9.503688685171813e-06,
      "loss": 0.4886,
      "step": 10471
    },
    {
      "epoch": 0.48056537102473496,
      "grad_norm": 0.48993992805480957,
      "learning_rate": 9.503582179113562e-06,
      "loss": 0.38,
      "step": 10472
    },
    {
      "epoch": 0.48061126152998945,
      "grad_norm": 0.4691629707813263,
      "learning_rate": 9.503475662225618e-06,
      "loss": 0.3833,
      "step": 10473
    },
    {
      "epoch": 0.4806571520352439,
      "grad_norm": 0.4672612249851227,
      "learning_rate": 9.503369134508237e-06,
      "loss": 0.411,
      "step": 10474
    },
    {
      "epoch": 0.4807030425404984,
      "grad_norm": 0.4373226761817932,
      "learning_rate": 9.503262595961676e-06,
      "loss": 0.4001,
      "step": 10475
    },
    {
      "epoch": 0.48074893304575284,
      "grad_norm": 0.4269385039806366,
      "learning_rate": 9.503156046586186e-06,
      "loss": 0.3212,
      "step": 10476
    },
    {
      "epoch": 0.4807948235510073,
      "grad_norm": 0.4578581154346466,
      "learning_rate": 9.503049486382028e-06,
      "loss": 0.4274,
      "step": 10477
    },
    {
      "epoch": 0.4808407140562618,
      "grad_norm": 0.48944443464279175,
      "learning_rate": 9.50294291534946e-06,
      "loss": 0.3831,
      "step": 10478
    },
    {
      "epoch": 0.4808866045615162,
      "grad_norm": 0.46075639128685,
      "learning_rate": 9.502836333488732e-06,
      "loss": 0.4088,
      "step": 10479
    },
    {
      "epoch": 0.48093249506677066,
      "grad_norm": 0.4391716420650482,
      "learning_rate": 9.502729740800105e-06,
      "loss": 0.3774,
      "step": 10480
    },
    {
      "epoch": 0.48097838557202516,
      "grad_norm": 0.48299962282180786,
      "learning_rate": 9.502623137283835e-06,
      "loss": 0.4833,
      "step": 10481
    },
    {
      "epoch": 0.4810242760772796,
      "grad_norm": 0.4231841266155243,
      "learning_rate": 9.502516522940176e-06,
      "loss": 0.3107,
      "step": 10482
    },
    {
      "epoch": 0.4810701665825341,
      "grad_norm": 0.47765496373176575,
      "learning_rate": 9.502409897769388e-06,
      "loss": 0.4071,
      "step": 10483
    },
    {
      "epoch": 0.48111605708778854,
      "grad_norm": 0.4894293248653412,
      "learning_rate": 9.502303261771721e-06,
      "loss": 0.3773,
      "step": 10484
    },
    {
      "epoch": 0.481161947593043,
      "grad_norm": 0.44916993379592896,
      "learning_rate": 9.502196614947439e-06,
      "loss": 0.3895,
      "step": 10485
    },
    {
      "epoch": 0.4812078380982975,
      "grad_norm": 0.5385277271270752,
      "learning_rate": 9.502089957296794e-06,
      "loss": 0.5212,
      "step": 10486
    },
    {
      "epoch": 0.4812537286035519,
      "grad_norm": 0.4333556592464447,
      "learning_rate": 9.501983288820044e-06,
      "loss": 0.3637,
      "step": 10487
    },
    {
      "epoch": 0.48129961910880636,
      "grad_norm": 0.43650519847869873,
      "learning_rate": 9.501876609517444e-06,
      "loss": 0.3821,
      "step": 10488
    },
    {
      "epoch": 0.48134550961406086,
      "grad_norm": 0.4622440040111542,
      "learning_rate": 9.501769919389252e-06,
      "loss": 0.4373,
      "step": 10489
    },
    {
      "epoch": 0.4813914001193153,
      "grad_norm": 0.4766353666782379,
      "learning_rate": 9.501663218435726e-06,
      "loss": 0.4478,
      "step": 10490
    },
    {
      "epoch": 0.4814372906245698,
      "grad_norm": 0.4908713400363922,
      "learning_rate": 9.50155650665712e-06,
      "loss": 0.5197,
      "step": 10491
    },
    {
      "epoch": 0.48148318112982424,
      "grad_norm": 0.4828174412250519,
      "learning_rate": 9.50144978405369e-06,
      "loss": 0.4673,
      "step": 10492
    },
    {
      "epoch": 0.4815290716350787,
      "grad_norm": 0.4744407832622528,
      "learning_rate": 9.501343050625695e-06,
      "loss": 0.4248,
      "step": 10493
    },
    {
      "epoch": 0.4815749621403332,
      "grad_norm": 0.4782252013683319,
      "learning_rate": 9.501236306373389e-06,
      "loss": 0.4626,
      "step": 10494
    },
    {
      "epoch": 0.4816208526455876,
      "grad_norm": 0.4637541174888611,
      "learning_rate": 9.501129551297032e-06,
      "loss": 0.4179,
      "step": 10495
    },
    {
      "epoch": 0.48166674315084207,
      "grad_norm": 0.4931493401527405,
      "learning_rate": 9.501022785396878e-06,
      "loss": 0.4008,
      "step": 10496
    },
    {
      "epoch": 0.48171263365609657,
      "grad_norm": 0.48418864607810974,
      "learning_rate": 9.500916008673187e-06,
      "loss": 0.4416,
      "step": 10497
    },
    {
      "epoch": 0.481758524161351,
      "grad_norm": 0.47030889987945557,
      "learning_rate": 9.50080922112621e-06,
      "loss": 0.4116,
      "step": 10498
    },
    {
      "epoch": 0.48180441466660545,
      "grad_norm": 0.45532140135765076,
      "learning_rate": 9.50070242275621e-06,
      "loss": 0.3612,
      "step": 10499
    },
    {
      "epoch": 0.48185030517185995,
      "grad_norm": 0.4816846549510956,
      "learning_rate": 9.500595613563441e-06,
      "loss": 0.4048,
      "step": 10500
    },
    {
      "epoch": 0.4818961956771144,
      "grad_norm": 0.44284936785697937,
      "learning_rate": 9.50048879354816e-06,
      "loss": 0.3405,
      "step": 10501
    },
    {
      "epoch": 0.4819420861823689,
      "grad_norm": 0.5109581351280212,
      "learning_rate": 9.500381962710624e-06,
      "loss": 0.4817,
      "step": 10502
    },
    {
      "epoch": 0.48198797668762333,
      "grad_norm": 0.44986802339553833,
      "learning_rate": 9.500275121051087e-06,
      "loss": 0.3545,
      "step": 10503
    },
    {
      "epoch": 0.4820338671928778,
      "grad_norm": 0.5242726802825928,
      "learning_rate": 9.500168268569812e-06,
      "loss": 0.4571,
      "step": 10504
    },
    {
      "epoch": 0.48207975769813227,
      "grad_norm": 0.4705042243003845,
      "learning_rate": 9.50006140526705e-06,
      "loss": 0.3932,
      "step": 10505
    },
    {
      "epoch": 0.4821256482033867,
      "grad_norm": 0.4624444544315338,
      "learning_rate": 9.499954531143061e-06,
      "loss": 0.3465,
      "step": 10506
    },
    {
      "epoch": 0.48217153870864116,
      "grad_norm": 0.4773663580417633,
      "learning_rate": 9.499847646198104e-06,
      "loss": 0.4933,
      "step": 10507
    },
    {
      "epoch": 0.48221742921389565,
      "grad_norm": 0.49081405997276306,
      "learning_rate": 9.499740750432432e-06,
      "loss": 0.3814,
      "step": 10508
    },
    {
      "epoch": 0.4822633197191501,
      "grad_norm": 0.4599322974681854,
      "learning_rate": 9.499633843846303e-06,
      "loss": 0.3385,
      "step": 10509
    },
    {
      "epoch": 0.4823092102244046,
      "grad_norm": 0.4541787803173065,
      "learning_rate": 9.499526926439974e-06,
      "loss": 0.3435,
      "step": 10510
    },
    {
      "epoch": 0.48235510072965904,
      "grad_norm": 0.4409223198890686,
      "learning_rate": 9.499419998213706e-06,
      "loss": 0.3293,
      "step": 10511
    },
    {
      "epoch": 0.4824009912349135,
      "grad_norm": 0.46895596385002136,
      "learning_rate": 9.49931305916775e-06,
      "loss": 0.452,
      "step": 10512
    },
    {
      "epoch": 0.482446881740168,
      "grad_norm": 0.47766542434692383,
      "learning_rate": 9.499206109302368e-06,
      "loss": 0.3616,
      "step": 10513
    },
    {
      "epoch": 0.4824927722454224,
      "grad_norm": 0.48215818405151367,
      "learning_rate": 9.499099148617815e-06,
      "loss": 0.4581,
      "step": 10514
    },
    {
      "epoch": 0.48253866275067686,
      "grad_norm": 0.4555628001689911,
      "learning_rate": 9.498992177114347e-06,
      "loss": 0.4677,
      "step": 10515
    },
    {
      "epoch": 0.48258455325593136,
      "grad_norm": 0.4614100158214569,
      "learning_rate": 9.498885194792225e-06,
      "loss": 0.421,
      "step": 10516
    },
    {
      "epoch": 0.4826304437611858,
      "grad_norm": 0.49830806255340576,
      "learning_rate": 9.498778201651702e-06,
      "loss": 0.3402,
      "step": 10517
    },
    {
      "epoch": 0.4826763342664403,
      "grad_norm": 0.48189035058021545,
      "learning_rate": 9.498671197693038e-06,
      "loss": 0.3849,
      "step": 10518
    },
    {
      "epoch": 0.48272222477169474,
      "grad_norm": 0.45526474714279175,
      "learning_rate": 9.49856418291649e-06,
      "loss": 0.3397,
      "step": 10519
    },
    {
      "epoch": 0.4827681152769492,
      "grad_norm": 0.45288729667663574,
      "learning_rate": 9.498457157322314e-06,
      "loss": 0.4046,
      "step": 10520
    },
    {
      "epoch": 0.4828140057822037,
      "grad_norm": 0.46893730759620667,
      "learning_rate": 9.498350120910769e-06,
      "loss": 0.3723,
      "step": 10521
    },
    {
      "epoch": 0.4828598962874581,
      "grad_norm": 0.48233258724212646,
      "learning_rate": 9.498243073682111e-06,
      "loss": 0.4054,
      "step": 10522
    },
    {
      "epoch": 0.48290578679271257,
      "grad_norm": 0.46372032165527344,
      "learning_rate": 9.498136015636599e-06,
      "loss": 0.3859,
      "step": 10523
    },
    {
      "epoch": 0.48295167729796706,
      "grad_norm": 0.4478050470352173,
      "learning_rate": 9.49802894677449e-06,
      "loss": 0.3784,
      "step": 10524
    },
    {
      "epoch": 0.4829975678032215,
      "grad_norm": 0.5011600255966187,
      "learning_rate": 9.497921867096039e-06,
      "loss": 0.4296,
      "step": 10525
    },
    {
      "epoch": 0.483043458308476,
      "grad_norm": 0.4738364517688751,
      "learning_rate": 9.497814776601506e-06,
      "loss": 0.4038,
      "step": 10526
    },
    {
      "epoch": 0.48308934881373045,
      "grad_norm": 0.4783861041069031,
      "learning_rate": 9.49770767529115e-06,
      "loss": 0.4039,
      "step": 10527
    },
    {
      "epoch": 0.4831352393189849,
      "grad_norm": 0.512788712978363,
      "learning_rate": 9.497600563165224e-06,
      "loss": 0.4844,
      "step": 10528
    },
    {
      "epoch": 0.4831811298242394,
      "grad_norm": 0.4720323979854584,
      "learning_rate": 9.497493440223991e-06,
      "loss": 0.3864,
      "step": 10529
    },
    {
      "epoch": 0.48322702032949383,
      "grad_norm": 0.4891703426837921,
      "learning_rate": 9.497386306467703e-06,
      "loss": 0.4253,
      "step": 10530
    },
    {
      "epoch": 0.48327291083474827,
      "grad_norm": 0.46546056866645813,
      "learning_rate": 9.497279161896621e-06,
      "loss": 0.4531,
      "step": 10531
    },
    {
      "epoch": 0.48331880134000277,
      "grad_norm": 0.47407132387161255,
      "learning_rate": 9.497172006511003e-06,
      "loss": 0.4052,
      "step": 10532
    },
    {
      "epoch": 0.4833646918452572,
      "grad_norm": 0.47243431210517883,
      "learning_rate": 9.497064840311105e-06,
      "loss": 0.3758,
      "step": 10533
    },
    {
      "epoch": 0.48341058235051165,
      "grad_norm": 0.495991587638855,
      "learning_rate": 9.496957663297186e-06,
      "loss": 0.4253,
      "step": 10534
    },
    {
      "epoch": 0.48345647285576615,
      "grad_norm": 0.48500409722328186,
      "learning_rate": 9.496850475469502e-06,
      "loss": 0.4046,
      "step": 10535
    },
    {
      "epoch": 0.4835023633610206,
      "grad_norm": 0.42323896288871765,
      "learning_rate": 9.496743276828313e-06,
      "loss": 0.3403,
      "step": 10536
    },
    {
      "epoch": 0.4835482538662751,
      "grad_norm": 0.4268997609615326,
      "learning_rate": 9.496636067373875e-06,
      "loss": 0.357,
      "step": 10537
    },
    {
      "epoch": 0.48359414437152953,
      "grad_norm": 0.46194788813591003,
      "learning_rate": 9.496528847106447e-06,
      "loss": 0.4241,
      "step": 10538
    },
    {
      "epoch": 0.483640034876784,
      "grad_norm": 0.41879960894584656,
      "learning_rate": 9.496421616026285e-06,
      "loss": 0.3563,
      "step": 10539
    },
    {
      "epoch": 0.4836859253820385,
      "grad_norm": 0.47671058773994446,
      "learning_rate": 9.496314374133649e-06,
      "loss": 0.3823,
      "step": 10540
    },
    {
      "epoch": 0.4837318158872929,
      "grad_norm": 0.5220320820808411,
      "learning_rate": 9.496207121428795e-06,
      "loss": 0.5725,
      "step": 10541
    },
    {
      "epoch": 0.48377770639254736,
      "grad_norm": 0.4554789364337921,
      "learning_rate": 9.496099857911984e-06,
      "loss": 0.4219,
      "step": 10542
    },
    {
      "epoch": 0.48382359689780186,
      "grad_norm": 0.42972588539123535,
      "learning_rate": 9.495992583583471e-06,
      "loss": 0.3613,
      "step": 10543
    },
    {
      "epoch": 0.4838694874030563,
      "grad_norm": 0.4622933566570282,
      "learning_rate": 9.495885298443515e-06,
      "loss": 0.3811,
      "step": 10544
    },
    {
      "epoch": 0.4839153779083108,
      "grad_norm": 0.5069906711578369,
      "learning_rate": 9.495778002492374e-06,
      "loss": 0.4743,
      "step": 10545
    },
    {
      "epoch": 0.48396126841356524,
      "grad_norm": 0.5044757127761841,
      "learning_rate": 9.495670695730304e-06,
      "loss": 0.5002,
      "step": 10546
    },
    {
      "epoch": 0.4840071589188197,
      "grad_norm": 0.5277027487754822,
      "learning_rate": 9.495563378157564e-06,
      "loss": 0.5758,
      "step": 10547
    },
    {
      "epoch": 0.4840530494240742,
      "grad_norm": 0.4643923044204712,
      "learning_rate": 9.495456049774416e-06,
      "loss": 0.4404,
      "step": 10548
    },
    {
      "epoch": 0.4840989399293286,
      "grad_norm": 0.49092474579811096,
      "learning_rate": 9.495348710581113e-06,
      "loss": 0.4437,
      "step": 10549
    },
    {
      "epoch": 0.48414483043458306,
      "grad_norm": 0.48186805844306946,
      "learning_rate": 9.495241360577916e-06,
      "loss": 0.3862,
      "step": 10550
    },
    {
      "epoch": 0.48419072093983756,
      "grad_norm": 0.49741244316101074,
      "learning_rate": 9.495133999765083e-06,
      "loss": 0.4555,
      "step": 10551
    },
    {
      "epoch": 0.484236611445092,
      "grad_norm": 0.447372704744339,
      "learning_rate": 9.495026628142872e-06,
      "loss": 0.3559,
      "step": 10552
    },
    {
      "epoch": 0.4842825019503465,
      "grad_norm": 0.4247765839099884,
      "learning_rate": 9.494919245711539e-06,
      "loss": 0.3681,
      "step": 10553
    },
    {
      "epoch": 0.48432839245560094,
      "grad_norm": 0.45308107137680054,
      "learning_rate": 9.494811852471344e-06,
      "loss": 0.3715,
      "step": 10554
    },
    {
      "epoch": 0.4843742829608554,
      "grad_norm": 0.5200189352035522,
      "learning_rate": 9.494704448422544e-06,
      "loss": 0.4981,
      "step": 10555
    },
    {
      "epoch": 0.4844201734661099,
      "grad_norm": 0.45749276876449585,
      "learning_rate": 9.4945970335654e-06,
      "loss": 0.3655,
      "step": 10556
    },
    {
      "epoch": 0.4844660639713643,
      "grad_norm": 0.4648534059524536,
      "learning_rate": 9.494489607900168e-06,
      "loss": 0.3552,
      "step": 10557
    },
    {
      "epoch": 0.48451195447661877,
      "grad_norm": 0.4545661509037018,
      "learning_rate": 9.494382171427108e-06,
      "loss": 0.3629,
      "step": 10558
    },
    {
      "epoch": 0.48455784498187326,
      "grad_norm": 0.555922269821167,
      "learning_rate": 9.494274724146477e-06,
      "loss": 0.5095,
      "step": 10559
    },
    {
      "epoch": 0.4846037354871277,
      "grad_norm": 0.4687761068344116,
      "learning_rate": 9.494167266058534e-06,
      "loss": 0.3892,
      "step": 10560
    },
    {
      "epoch": 0.48464962599238215,
      "grad_norm": 0.45815131068229675,
      "learning_rate": 9.494059797163536e-06,
      "loss": 0.3839,
      "step": 10561
    },
    {
      "epoch": 0.48469551649763665,
      "grad_norm": 0.4146079421043396,
      "learning_rate": 9.493952317461743e-06,
      "loss": 0.3361,
      "step": 10562
    },
    {
      "epoch": 0.4847414070028911,
      "grad_norm": 0.4430672228336334,
      "learning_rate": 9.493844826953414e-06,
      "loss": 0.3602,
      "step": 10563
    },
    {
      "epoch": 0.4847872975081456,
      "grad_norm": 0.4650081396102905,
      "learning_rate": 9.493737325638804e-06,
      "loss": 0.3953,
      "step": 10564
    },
    {
      "epoch": 0.48483318801340003,
      "grad_norm": 0.4787168502807617,
      "learning_rate": 9.493629813518175e-06,
      "loss": 0.4182,
      "step": 10565
    },
    {
      "epoch": 0.48487907851865447,
      "grad_norm": 0.5862655639648438,
      "learning_rate": 9.493522290591786e-06,
      "loss": 0.4181,
      "step": 10566
    },
    {
      "epoch": 0.48492496902390897,
      "grad_norm": 0.4384766221046448,
      "learning_rate": 9.493414756859893e-06,
      "loss": 0.3364,
      "step": 10567
    },
    {
      "epoch": 0.4849708595291634,
      "grad_norm": 0.5203227400779724,
      "learning_rate": 9.493307212322756e-06,
      "loss": 0.5114,
      "step": 10568
    },
    {
      "epoch": 0.48501675003441785,
      "grad_norm": 0.44314152002334595,
      "learning_rate": 9.493199656980634e-06,
      "loss": 0.3823,
      "step": 10569
    },
    {
      "epoch": 0.48506264053967235,
      "grad_norm": 0.45190590620040894,
      "learning_rate": 9.493092090833782e-06,
      "loss": 0.3966,
      "step": 10570
    },
    {
      "epoch": 0.4851085310449268,
      "grad_norm": 0.49951040744781494,
      "learning_rate": 9.492984513882465e-06,
      "loss": 0.5049,
      "step": 10571
    },
    {
      "epoch": 0.4851544215501813,
      "grad_norm": 0.4859614670276642,
      "learning_rate": 9.492876926126937e-06,
      "loss": 0.4033,
      "step": 10572
    },
    {
      "epoch": 0.48520031205543573,
      "grad_norm": 0.4782193601131439,
      "learning_rate": 9.492769327567456e-06,
      "loss": 0.4687,
      "step": 10573
    },
    {
      "epoch": 0.4852462025606902,
      "grad_norm": 0.46321114897727966,
      "learning_rate": 9.492661718204286e-06,
      "loss": 0.413,
      "step": 10574
    },
    {
      "epoch": 0.4852920930659447,
      "grad_norm": 0.4493845999240875,
      "learning_rate": 9.49255409803768e-06,
      "loss": 0.3359,
      "step": 10575
    },
    {
      "epoch": 0.4853379835711991,
      "grad_norm": 0.47707003355026245,
      "learning_rate": 9.492446467067902e-06,
      "loss": 0.4981,
      "step": 10576
    },
    {
      "epoch": 0.48538387407645356,
      "grad_norm": 0.443598210811615,
      "learning_rate": 9.492338825295207e-06,
      "loss": 0.4085,
      "step": 10577
    },
    {
      "epoch": 0.48542976458170806,
      "grad_norm": 0.45336687564849854,
      "learning_rate": 9.492231172719854e-06,
      "loss": 0.3886,
      "step": 10578
    },
    {
      "epoch": 0.4854756550869625,
      "grad_norm": 0.45578062534332275,
      "learning_rate": 9.492123509342104e-06,
      "loss": 0.414,
      "step": 10579
    },
    {
      "epoch": 0.485521545592217,
      "grad_norm": 0.4789424240589142,
      "learning_rate": 9.492015835162213e-06,
      "loss": 0.452,
      "step": 10580
    },
    {
      "epoch": 0.48556743609747144,
      "grad_norm": 0.46922335028648376,
      "learning_rate": 9.491908150180442e-06,
      "loss": 0.409,
      "step": 10581
    },
    {
      "epoch": 0.4856133266027259,
      "grad_norm": 0.46665775775909424,
      "learning_rate": 9.491800454397051e-06,
      "loss": 0.4309,
      "step": 10582
    },
    {
      "epoch": 0.4856592171079804,
      "grad_norm": 0.4767134487628937,
      "learning_rate": 9.491692747812296e-06,
      "loss": 0.4193,
      "step": 10583
    },
    {
      "epoch": 0.4857051076132348,
      "grad_norm": 0.4366234540939331,
      "learning_rate": 9.49158503042644e-06,
      "loss": 0.3475,
      "step": 10584
    },
    {
      "epoch": 0.48575099811848926,
      "grad_norm": 0.42120498418807983,
      "learning_rate": 9.491477302239737e-06,
      "loss": 0.3228,
      "step": 10585
    },
    {
      "epoch": 0.48579688862374376,
      "grad_norm": 0.4392586350440979,
      "learning_rate": 9.49136956325245e-06,
      "loss": 0.3343,
      "step": 10586
    },
    {
      "epoch": 0.4858427791289982,
      "grad_norm": 0.47664669156074524,
      "learning_rate": 9.491261813464837e-06,
      "loss": 0.4366,
      "step": 10587
    },
    {
      "epoch": 0.48588866963425265,
      "grad_norm": 0.4301667809486389,
      "learning_rate": 9.491154052877154e-06,
      "loss": 0.3805,
      "step": 10588
    },
    {
      "epoch": 0.48593456013950714,
      "grad_norm": 0.4611995816230774,
      "learning_rate": 9.491046281489667e-06,
      "loss": 0.3998,
      "step": 10589
    },
    {
      "epoch": 0.4859804506447616,
      "grad_norm": 0.4516323506832123,
      "learning_rate": 9.490938499302627e-06,
      "loss": 0.4378,
      "step": 10590
    },
    {
      "epoch": 0.4860263411500161,
      "grad_norm": 0.47884055972099304,
      "learning_rate": 9.4908307063163e-06,
      "loss": 0.4323,
      "step": 10591
    },
    {
      "epoch": 0.4860722316552705,
      "grad_norm": 0.5051462054252625,
      "learning_rate": 9.490722902530943e-06,
      "loss": 0.4771,
      "step": 10592
    },
    {
      "epoch": 0.48611812216052497,
      "grad_norm": 0.47771400213241577,
      "learning_rate": 9.490615087946812e-06,
      "loss": 0.4078,
      "step": 10593
    },
    {
      "epoch": 0.48616401266577947,
      "grad_norm": 0.47820213437080383,
      "learning_rate": 9.490507262564172e-06,
      "loss": 0.4374,
      "step": 10594
    },
    {
      "epoch": 0.4862099031710339,
      "grad_norm": 0.45286113023757935,
      "learning_rate": 9.490399426383277e-06,
      "loss": 0.4117,
      "step": 10595
    },
    {
      "epoch": 0.48625579367628835,
      "grad_norm": 0.5176018476486206,
      "learning_rate": 9.49029157940439e-06,
      "loss": 0.4869,
      "step": 10596
    },
    {
      "epoch": 0.48630168418154285,
      "grad_norm": 0.4529692828655243,
      "learning_rate": 9.490183721627767e-06,
      "loss": 0.3432,
      "step": 10597
    },
    {
      "epoch": 0.4863475746867973,
      "grad_norm": 0.49915948510169983,
      "learning_rate": 9.490075853053673e-06,
      "loss": 0.4814,
      "step": 10598
    },
    {
      "epoch": 0.4863934651920518,
      "grad_norm": 0.48007404804229736,
      "learning_rate": 9.48996797368236e-06,
      "loss": 0.4559,
      "step": 10599
    },
    {
      "epoch": 0.48643935569730623,
      "grad_norm": 0.46602025628089905,
      "learning_rate": 9.489860083514094e-06,
      "loss": 0.3751,
      "step": 10600
    },
    {
      "epoch": 0.4864852462025607,
      "grad_norm": 0.46526169776916504,
      "learning_rate": 9.48975218254913e-06,
      "loss": 0.404,
      "step": 10601
    },
    {
      "epoch": 0.48653113670781517,
      "grad_norm": 0.4285773038864136,
      "learning_rate": 9.489644270787728e-06,
      "loss": 0.3255,
      "step": 10602
    },
    {
      "epoch": 0.4865770272130696,
      "grad_norm": 0.4511367976665497,
      "learning_rate": 9.48953634823015e-06,
      "loss": 0.4547,
      "step": 10603
    },
    {
      "epoch": 0.48662291771832406,
      "grad_norm": 0.4331052899360657,
      "learning_rate": 9.489428414876654e-06,
      "loss": 0.3531,
      "step": 10604
    },
    {
      "epoch": 0.48666880822357855,
      "grad_norm": 0.4630088210105896,
      "learning_rate": 9.489320470727499e-06,
      "loss": 0.4049,
      "step": 10605
    },
    {
      "epoch": 0.486714698728833,
      "grad_norm": 0.45251232385635376,
      "learning_rate": 9.489212515782946e-06,
      "loss": 0.401,
      "step": 10606
    },
    {
      "epoch": 0.4867605892340875,
      "grad_norm": 0.4439692497253418,
      "learning_rate": 9.489104550043251e-06,
      "loss": 0.3855,
      "step": 10607
    },
    {
      "epoch": 0.48680647973934194,
      "grad_norm": 0.43586409091949463,
      "learning_rate": 9.488996573508679e-06,
      "loss": 0.3428,
      "step": 10608
    },
    {
      "epoch": 0.4868523702445964,
      "grad_norm": 0.5121430158615112,
      "learning_rate": 9.488888586179487e-06,
      "loss": 0.4365,
      "step": 10609
    },
    {
      "epoch": 0.4868982607498509,
      "grad_norm": 0.4647507965564728,
      "learning_rate": 9.488780588055936e-06,
      "loss": 0.4203,
      "step": 10610
    },
    {
      "epoch": 0.4869441512551053,
      "grad_norm": 0.4112704396247864,
      "learning_rate": 9.48867257913828e-06,
      "loss": 0.2968,
      "step": 10611
    },
    {
      "epoch": 0.48699004176035976,
      "grad_norm": 0.46082592010498047,
      "learning_rate": 9.488564559426787e-06,
      "loss": 0.4453,
      "step": 10612
    },
    {
      "epoch": 0.48703593226561426,
      "grad_norm": 0.48928314447402954,
      "learning_rate": 9.48845652892171e-06,
      "loss": 0.4928,
      "step": 10613
    },
    {
      "epoch": 0.4870818227708687,
      "grad_norm": 0.45143210887908936,
      "learning_rate": 9.488348487623313e-06,
      "loss": 0.4082,
      "step": 10614
    },
    {
      "epoch": 0.4871277132761232,
      "grad_norm": 0.46791449189186096,
      "learning_rate": 9.488240435531854e-06,
      "loss": 0.4305,
      "step": 10615
    },
    {
      "epoch": 0.48717360378137764,
      "grad_norm": 0.45897534489631653,
      "learning_rate": 9.488132372647596e-06,
      "loss": 0.4325,
      "step": 10616
    },
    {
      "epoch": 0.4872194942866321,
      "grad_norm": 0.48604848980903625,
      "learning_rate": 9.488024298970792e-06,
      "loss": 0.4039,
      "step": 10617
    },
    {
      "epoch": 0.4872653847918866,
      "grad_norm": 0.521153450012207,
      "learning_rate": 9.487916214501709e-06,
      "loss": 0.4017,
      "step": 10618
    },
    {
      "epoch": 0.487311275297141,
      "grad_norm": 0.4453768730163574,
      "learning_rate": 9.487808119240602e-06,
      "loss": 0.4487,
      "step": 10619
    },
    {
      "epoch": 0.48735716580239546,
      "grad_norm": 0.4747428894042969,
      "learning_rate": 9.487700013187734e-06,
      "loss": 0.4266,
      "step": 10620
    },
    {
      "epoch": 0.48740305630764996,
      "grad_norm": 0.44069811701774597,
      "learning_rate": 9.487591896343364e-06,
      "loss": 0.3339,
      "step": 10621
    },
    {
      "epoch": 0.4874489468129044,
      "grad_norm": 0.5257259607315063,
      "learning_rate": 9.487483768707753e-06,
      "loss": 0.5003,
      "step": 10622
    },
    {
      "epoch": 0.48749483731815885,
      "grad_norm": 0.4492780864238739,
      "learning_rate": 9.487375630281156e-06,
      "loss": 0.3613,
      "step": 10623
    },
    {
      "epoch": 0.48754072782341334,
      "grad_norm": 0.48792123794555664,
      "learning_rate": 9.487267481063839e-06,
      "loss": 0.4324,
      "step": 10624
    },
    {
      "epoch": 0.4875866183286678,
      "grad_norm": 0.47915229201316833,
      "learning_rate": 9.48715932105606e-06,
      "loss": 0.4347,
      "step": 10625
    },
    {
      "epoch": 0.4876325088339223,
      "grad_norm": 0.452625572681427,
      "learning_rate": 9.48705115025808e-06,
      "loss": 0.41,
      "step": 10626
    },
    {
      "epoch": 0.4876783993391767,
      "grad_norm": 0.5111326575279236,
      "learning_rate": 9.486942968670157e-06,
      "loss": 0.4949,
      "step": 10627
    },
    {
      "epoch": 0.48772428984443117,
      "grad_norm": 0.5307044982910156,
      "learning_rate": 9.486834776292552e-06,
      "loss": 0.5728,
      "step": 10628
    },
    {
      "epoch": 0.48777018034968567,
      "grad_norm": 0.4536900818347931,
      "learning_rate": 9.486726573125526e-06,
      "loss": 0.3901,
      "step": 10629
    },
    {
      "epoch": 0.4878160708549401,
      "grad_norm": 0.4589734375476837,
      "learning_rate": 9.48661835916934e-06,
      "loss": 0.3918,
      "step": 10630
    },
    {
      "epoch": 0.48786196136019455,
      "grad_norm": 0.4557384252548218,
      "learning_rate": 9.486510134424251e-06,
      "loss": 0.4146,
      "step": 10631
    },
    {
      "epoch": 0.48790785186544905,
      "grad_norm": 0.44064903259277344,
      "learning_rate": 9.48640189889052e-06,
      "loss": 0.3445,
      "step": 10632
    },
    {
      "epoch": 0.4879537423707035,
      "grad_norm": 0.440535306930542,
      "learning_rate": 9.486293652568413e-06,
      "loss": 0.3971,
      "step": 10633
    },
    {
      "epoch": 0.487999632875958,
      "grad_norm": 0.43332499265670776,
      "learning_rate": 9.48618539545818e-06,
      "loss": 0.4,
      "step": 10634
    },
    {
      "epoch": 0.48804552338121243,
      "grad_norm": 0.4713751971721649,
      "learning_rate": 9.48607712756009e-06,
      "loss": 0.4134,
      "step": 10635
    },
    {
      "epoch": 0.4880914138864669,
      "grad_norm": 0.43473419547080994,
      "learning_rate": 9.485968848874401e-06,
      "loss": 0.3481,
      "step": 10636
    },
    {
      "epoch": 0.48813730439172137,
      "grad_norm": 0.4714726507663727,
      "learning_rate": 9.485860559401372e-06,
      "loss": 0.4216,
      "step": 10637
    },
    {
      "epoch": 0.4881831948969758,
      "grad_norm": 0.44355708360671997,
      "learning_rate": 9.485752259141264e-06,
      "loss": 0.3728,
      "step": 10638
    },
    {
      "epoch": 0.48822908540223026,
      "grad_norm": 0.4756859242916107,
      "learning_rate": 9.485643948094336e-06,
      "loss": 0.4963,
      "step": 10639
    },
    {
      "epoch": 0.48827497590748475,
      "grad_norm": 0.46483999490737915,
      "learning_rate": 9.485535626260852e-06,
      "loss": 0.3987,
      "step": 10640
    },
    {
      "epoch": 0.4883208664127392,
      "grad_norm": 0.48093196749687195,
      "learning_rate": 9.485427293641069e-06,
      "loss": 0.4531,
      "step": 10641
    },
    {
      "epoch": 0.4883667569179937,
      "grad_norm": 0.48590660095214844,
      "learning_rate": 9.485318950235252e-06,
      "loss": 0.476,
      "step": 10642
    },
    {
      "epoch": 0.48841264742324814,
      "grad_norm": 0.487311452627182,
      "learning_rate": 9.485210596043655e-06,
      "loss": 0.4561,
      "step": 10643
    },
    {
      "epoch": 0.4884585379285026,
      "grad_norm": 0.47221386432647705,
      "learning_rate": 9.485102231066545e-06,
      "loss": 0.4587,
      "step": 10644
    },
    {
      "epoch": 0.4885044284337571,
      "grad_norm": 0.43011292815208435,
      "learning_rate": 9.484993855304177e-06,
      "loss": 0.3676,
      "step": 10645
    },
    {
      "epoch": 0.4885503189390115,
      "grad_norm": 0.5184840559959412,
      "learning_rate": 9.484885468756815e-06,
      "loss": 0.5442,
      "step": 10646
    },
    {
      "epoch": 0.48859620944426596,
      "grad_norm": 0.4347049593925476,
      "learning_rate": 9.48477707142472e-06,
      "loss": 0.3366,
      "step": 10647
    },
    {
      "epoch": 0.48864209994952046,
      "grad_norm": 0.4783165454864502,
      "learning_rate": 9.48466866330815e-06,
      "loss": 0.413,
      "step": 10648
    },
    {
      "epoch": 0.4886879904547749,
      "grad_norm": 0.4483320713043213,
      "learning_rate": 9.484560244407369e-06,
      "loss": 0.3751,
      "step": 10649
    },
    {
      "epoch": 0.48873388096002934,
      "grad_norm": 0.4035055339336395,
      "learning_rate": 9.484451814722635e-06,
      "loss": 0.293,
      "step": 10650
    },
    {
      "epoch": 0.48877977146528384,
      "grad_norm": 0.46935850381851196,
      "learning_rate": 9.48434337425421e-06,
      "loss": 0.4051,
      "step": 10651
    },
    {
      "epoch": 0.4888256619705383,
      "grad_norm": 0.49506109952926636,
      "learning_rate": 9.484234923002356e-06,
      "loss": 0.4865,
      "step": 10652
    },
    {
      "epoch": 0.4888715524757928,
      "grad_norm": 0.5216750502586365,
      "learning_rate": 9.48412646096733e-06,
      "loss": 0.6004,
      "step": 10653
    },
    {
      "epoch": 0.4889174429810472,
      "grad_norm": 0.4630841612815857,
      "learning_rate": 9.484017988149396e-06,
      "loss": 0.3929,
      "step": 10654
    },
    {
      "epoch": 0.48896333348630167,
      "grad_norm": 0.45899176597595215,
      "learning_rate": 9.483909504548814e-06,
      "loss": 0.4132,
      "step": 10655
    },
    {
      "epoch": 0.48900922399155616,
      "grad_norm": 0.4924745261669159,
      "learning_rate": 9.483801010165844e-06,
      "loss": 0.4558,
      "step": 10656
    },
    {
      "epoch": 0.4890551144968106,
      "grad_norm": 0.42792290449142456,
      "learning_rate": 9.483692505000749e-06,
      "loss": 0.3295,
      "step": 10657
    },
    {
      "epoch": 0.48910100500206505,
      "grad_norm": 0.49291712045669556,
      "learning_rate": 9.483583989053788e-06,
      "loss": 0.5214,
      "step": 10658
    },
    {
      "epoch": 0.48914689550731955,
      "grad_norm": 0.4470479488372803,
      "learning_rate": 9.483475462325221e-06,
      "loss": 0.3972,
      "step": 10659
    },
    {
      "epoch": 0.489192786012574,
      "grad_norm": 0.43819937109947205,
      "learning_rate": 9.483366924815312e-06,
      "loss": 0.3487,
      "step": 10660
    },
    {
      "epoch": 0.4892386765178285,
      "grad_norm": 0.4256936013698578,
      "learning_rate": 9.48325837652432e-06,
      "loss": 0.3338,
      "step": 10661
    },
    {
      "epoch": 0.48928456702308293,
      "grad_norm": 0.8580310940742493,
      "learning_rate": 9.483149817452508e-06,
      "loss": 0.4282,
      "step": 10662
    },
    {
      "epoch": 0.48933045752833737,
      "grad_norm": 0.505344569683075,
      "learning_rate": 9.483041247600135e-06,
      "loss": 0.5029,
      "step": 10663
    },
    {
      "epoch": 0.48937634803359187,
      "grad_norm": 0.44550150632858276,
      "learning_rate": 9.48293266696746e-06,
      "loss": 0.3384,
      "step": 10664
    },
    {
      "epoch": 0.4894222385388463,
      "grad_norm": 0.41006457805633545,
      "learning_rate": 9.48282407555475e-06,
      "loss": 0.314,
      "step": 10665
    },
    {
      "epoch": 0.48946812904410075,
      "grad_norm": 0.5112209320068359,
      "learning_rate": 9.482715473362263e-06,
      "loss": 0.4939,
      "step": 10666
    },
    {
      "epoch": 0.48951401954935525,
      "grad_norm": 0.44809049367904663,
      "learning_rate": 9.48260686039026e-06,
      "loss": 0.3873,
      "step": 10667
    },
    {
      "epoch": 0.4895599100546097,
      "grad_norm": 0.43120285868644714,
      "learning_rate": 9.482498236639002e-06,
      "loss": 0.362,
      "step": 10668
    },
    {
      "epoch": 0.4896058005598642,
      "grad_norm": 0.45697617530822754,
      "learning_rate": 9.48238960210875e-06,
      "loss": 0.3977,
      "step": 10669
    },
    {
      "epoch": 0.48965169106511863,
      "grad_norm": 0.4931700527667999,
      "learning_rate": 9.482280956799765e-06,
      "loss": 0.4953,
      "step": 10670
    },
    {
      "epoch": 0.4896975815703731,
      "grad_norm": 0.471383273601532,
      "learning_rate": 9.48217230071231e-06,
      "loss": 0.4039,
      "step": 10671
    },
    {
      "epoch": 0.4897434720756276,
      "grad_norm": 0.44583895802497864,
      "learning_rate": 9.482063633846642e-06,
      "loss": 0.4038,
      "step": 10672
    },
    {
      "epoch": 0.489789362580882,
      "grad_norm": 0.45217204093933105,
      "learning_rate": 9.48195495620303e-06,
      "loss": 0.3917,
      "step": 10673
    },
    {
      "epoch": 0.48983525308613646,
      "grad_norm": 0.5157490968704224,
      "learning_rate": 9.48184626778173e-06,
      "loss": 0.4222,
      "step": 10674
    },
    {
      "epoch": 0.48988114359139096,
      "grad_norm": 0.5463132262229919,
      "learning_rate": 9.481737568583003e-06,
      "loss": 0.4731,
      "step": 10675
    },
    {
      "epoch": 0.4899270340966454,
      "grad_norm": 0.48229992389678955,
      "learning_rate": 9.481628858607111e-06,
      "loss": 0.3891,
      "step": 10676
    },
    {
      "epoch": 0.48997292460189984,
      "grad_norm": 0.5011181831359863,
      "learning_rate": 9.481520137854318e-06,
      "loss": 0.4392,
      "step": 10677
    },
    {
      "epoch": 0.49001881510715434,
      "grad_norm": 0.48628997802734375,
      "learning_rate": 9.481411406324883e-06,
      "loss": 0.4644,
      "step": 10678
    },
    {
      "epoch": 0.4900647056124088,
      "grad_norm": 0.4589502811431885,
      "learning_rate": 9.481302664019067e-06,
      "loss": 0.4061,
      "step": 10679
    },
    {
      "epoch": 0.4901105961176633,
      "grad_norm": 0.4357941448688507,
      "learning_rate": 9.481193910937132e-06,
      "loss": 0.329,
      "step": 10680
    },
    {
      "epoch": 0.4901564866229177,
      "grad_norm": 0.4721980094909668,
      "learning_rate": 9.48108514707934e-06,
      "loss": 0.4175,
      "step": 10681
    },
    {
      "epoch": 0.49020237712817216,
      "grad_norm": 0.47352492809295654,
      "learning_rate": 9.480976372445954e-06,
      "loss": 0.41,
      "step": 10682
    },
    {
      "epoch": 0.49024826763342666,
      "grad_norm": 0.45808619260787964,
      "learning_rate": 9.480867587037235e-06,
      "loss": 0.41,
      "step": 10683
    },
    {
      "epoch": 0.4902941581386811,
      "grad_norm": 0.47846418619155884,
      "learning_rate": 9.480758790853442e-06,
      "loss": 0.4784,
      "step": 10684
    },
    {
      "epoch": 0.49034004864393554,
      "grad_norm": 0.4267057478427887,
      "learning_rate": 9.480649983894838e-06,
      "loss": 0.3286,
      "step": 10685
    },
    {
      "epoch": 0.49038593914919004,
      "grad_norm": 0.4665498733520508,
      "learning_rate": 9.480541166161685e-06,
      "loss": 0.476,
      "step": 10686
    },
    {
      "epoch": 0.4904318296544445,
      "grad_norm": 0.45767974853515625,
      "learning_rate": 9.480432337654244e-06,
      "loss": 0.3493,
      "step": 10687
    },
    {
      "epoch": 0.490477720159699,
      "grad_norm": 0.5075933933258057,
      "learning_rate": 9.48032349837278e-06,
      "loss": 0.5108,
      "step": 10688
    },
    {
      "epoch": 0.4905236106649534,
      "grad_norm": 0.4356735050678253,
      "learning_rate": 9.48021464831755e-06,
      "loss": 0.3335,
      "step": 10689
    },
    {
      "epoch": 0.49056950117020787,
      "grad_norm": 0.4836972653865814,
      "learning_rate": 9.480105787488818e-06,
      "loss": 0.4355,
      "step": 10690
    },
    {
      "epoch": 0.49061539167546236,
      "grad_norm": 0.46491938829421997,
      "learning_rate": 9.479996915886844e-06,
      "loss": 0.4266,
      "step": 10691
    },
    {
      "epoch": 0.4906612821807168,
      "grad_norm": 0.43625667691230774,
      "learning_rate": 9.479888033511892e-06,
      "loss": 0.3522,
      "step": 10692
    },
    {
      "epoch": 0.49070717268597125,
      "grad_norm": 0.46028441190719604,
      "learning_rate": 9.479779140364225e-06,
      "loss": 0.4166,
      "step": 10693
    },
    {
      "epoch": 0.49075306319122575,
      "grad_norm": 0.4954717457294464,
      "learning_rate": 9.479670236444104e-06,
      "loss": 0.5178,
      "step": 10694
    },
    {
      "epoch": 0.4907989536964802,
      "grad_norm": 0.45140647888183594,
      "learning_rate": 9.479561321751787e-06,
      "loss": 0.3438,
      "step": 10695
    },
    {
      "epoch": 0.4908448442017347,
      "grad_norm": 0.4738351106643677,
      "learning_rate": 9.479452396287539e-06,
      "loss": 0.4298,
      "step": 10696
    },
    {
      "epoch": 0.49089073470698913,
      "grad_norm": 0.45133888721466064,
      "learning_rate": 9.479343460051622e-06,
      "loss": 0.4058,
      "step": 10697
    },
    {
      "epoch": 0.49093662521224357,
      "grad_norm": 0.435988187789917,
      "learning_rate": 9.479234513044299e-06,
      "loss": 0.3259,
      "step": 10698
    },
    {
      "epoch": 0.49098251571749807,
      "grad_norm": 0.4549507796764374,
      "learning_rate": 9.479125555265829e-06,
      "loss": 0.3961,
      "step": 10699
    },
    {
      "epoch": 0.4910284062227525,
      "grad_norm": 0.46372732520103455,
      "learning_rate": 9.479016586716476e-06,
      "loss": 0.3804,
      "step": 10700
    },
    {
      "epoch": 0.49107429672800695,
      "grad_norm": 0.40998783707618713,
      "learning_rate": 9.478907607396504e-06,
      "loss": 0.3145,
      "step": 10701
    },
    {
      "epoch": 0.49112018723326145,
      "grad_norm": 0.4988030791282654,
      "learning_rate": 9.478798617306171e-06,
      "loss": 0.4894,
      "step": 10702
    },
    {
      "epoch": 0.4911660777385159,
      "grad_norm": 0.46360355615615845,
      "learning_rate": 9.47868961644574e-06,
      "loss": 0.3937,
      "step": 10703
    },
    {
      "epoch": 0.4912119682437704,
      "grad_norm": 0.5051340460777283,
      "learning_rate": 9.478580604815475e-06,
      "loss": 0.4093,
      "step": 10704
    },
    {
      "epoch": 0.49125785874902483,
      "grad_norm": 0.47017911076545715,
      "learning_rate": 9.478471582415638e-06,
      "loss": 0.4045,
      "step": 10705
    },
    {
      "epoch": 0.4913037492542793,
      "grad_norm": 0.469716340303421,
      "learning_rate": 9.478362549246488e-06,
      "loss": 0.4113,
      "step": 10706
    },
    {
      "epoch": 0.4913496397595338,
      "grad_norm": 0.4343683123588562,
      "learning_rate": 9.478253505308291e-06,
      "loss": 0.386,
      "step": 10707
    },
    {
      "epoch": 0.4913955302647882,
      "grad_norm": 0.4531431794166565,
      "learning_rate": 9.478144450601308e-06,
      "loss": 0.3852,
      "step": 10708
    },
    {
      "epoch": 0.49144142077004266,
      "grad_norm": 0.46602678298950195,
      "learning_rate": 9.478035385125802e-06,
      "loss": 0.4207,
      "step": 10709
    },
    {
      "epoch": 0.49148731127529716,
      "grad_norm": 0.4904780983924866,
      "learning_rate": 9.47792630888203e-06,
      "loss": 0.4395,
      "step": 10710
    },
    {
      "epoch": 0.4915332017805516,
      "grad_norm": 0.4122215509414673,
      "learning_rate": 9.477817221870263e-06,
      "loss": 0.3141,
      "step": 10711
    },
    {
      "epoch": 0.49157909228580604,
      "grad_norm": 0.4632011950016022,
      "learning_rate": 9.477708124090756e-06,
      "loss": 0.452,
      "step": 10712
    },
    {
      "epoch": 0.49162498279106054,
      "grad_norm": 0.4378023147583008,
      "learning_rate": 9.477599015543774e-06,
      "loss": 0.41,
      "step": 10713
    },
    {
      "epoch": 0.491670873296315,
      "grad_norm": 0.49732813239097595,
      "learning_rate": 9.477489896229583e-06,
      "loss": 0.4237,
      "step": 10714
    },
    {
      "epoch": 0.4917167638015695,
      "grad_norm": 0.4453243613243103,
      "learning_rate": 9.477380766148438e-06,
      "loss": 0.3801,
      "step": 10715
    },
    {
      "epoch": 0.4917626543068239,
      "grad_norm": 0.44639262557029724,
      "learning_rate": 9.477271625300608e-06,
      "loss": 0.3914,
      "step": 10716
    },
    {
      "epoch": 0.49180854481207836,
      "grad_norm": 0.45876771211624146,
      "learning_rate": 9.477162473686353e-06,
      "loss": 0.4365,
      "step": 10717
    },
    {
      "epoch": 0.49185443531733286,
      "grad_norm": 0.4284830689430237,
      "learning_rate": 9.477053311305932e-06,
      "loss": 0.3459,
      "step": 10718
    },
    {
      "epoch": 0.4919003258225873,
      "grad_norm": 0.4694904386997223,
      "learning_rate": 9.476944138159615e-06,
      "loss": 0.4294,
      "step": 10719
    },
    {
      "epoch": 0.49194621632784175,
      "grad_norm": 0.4418640434741974,
      "learning_rate": 9.476834954247657e-06,
      "loss": 0.3728,
      "step": 10720
    },
    {
      "epoch": 0.49199210683309624,
      "grad_norm": 0.47429221868515015,
      "learning_rate": 9.476725759570324e-06,
      "loss": 0.4423,
      "step": 10721
    },
    {
      "epoch": 0.4920379973383507,
      "grad_norm": 0.4893615245819092,
      "learning_rate": 9.47661655412788e-06,
      "loss": 0.5006,
      "step": 10722
    },
    {
      "epoch": 0.4920838878436052,
      "grad_norm": 0.48861318826675415,
      "learning_rate": 9.476507337920586e-06,
      "loss": 0.4628,
      "step": 10723
    },
    {
      "epoch": 0.4921297783488596,
      "grad_norm": 0.4213911294937134,
      "learning_rate": 9.476398110948703e-06,
      "loss": 0.336,
      "step": 10724
    },
    {
      "epoch": 0.49217566885411407,
      "grad_norm": 0.46832531690597534,
      "learning_rate": 9.476288873212498e-06,
      "loss": 0.4058,
      "step": 10725
    },
    {
      "epoch": 0.49222155935936857,
      "grad_norm": 0.4661771059036255,
      "learning_rate": 9.47617962471223e-06,
      "loss": 0.3812,
      "step": 10726
    },
    {
      "epoch": 0.492267449864623,
      "grad_norm": 0.46775874495506287,
      "learning_rate": 9.476070365448162e-06,
      "loss": 0.4248,
      "step": 10727
    },
    {
      "epoch": 0.49231334036987745,
      "grad_norm": 0.43332505226135254,
      "learning_rate": 9.475961095420557e-06,
      "loss": 0.3788,
      "step": 10728
    },
    {
      "epoch": 0.49235923087513195,
      "grad_norm": 0.4617195427417755,
      "learning_rate": 9.47585181462968e-06,
      "loss": 0.4286,
      "step": 10729
    },
    {
      "epoch": 0.4924051213803864,
      "grad_norm": 0.43549850583076477,
      "learning_rate": 9.47574252307579e-06,
      "loss": 0.3835,
      "step": 10730
    },
    {
      "epoch": 0.4924510118856409,
      "grad_norm": 0.46854934096336365,
      "learning_rate": 9.475633220759153e-06,
      "loss": 0.4168,
      "step": 10731
    },
    {
      "epoch": 0.49249690239089533,
      "grad_norm": 0.4645409882068634,
      "learning_rate": 9.47552390768003e-06,
      "loss": 0.3976,
      "step": 10732
    },
    {
      "epoch": 0.4925427928961498,
      "grad_norm": 0.47584426403045654,
      "learning_rate": 9.475414583838686e-06,
      "loss": 0.4163,
      "step": 10733
    },
    {
      "epoch": 0.49258868340140427,
      "grad_norm": 0.48771488666534424,
      "learning_rate": 9.47530524923538e-06,
      "loss": 0.4566,
      "step": 10734
    },
    {
      "epoch": 0.4926345739066587,
      "grad_norm": 0.5110533237457275,
      "learning_rate": 9.47519590387038e-06,
      "loss": 0.5134,
      "step": 10735
    },
    {
      "epoch": 0.49268046441191315,
      "grad_norm": 0.4774998128414154,
      "learning_rate": 9.475086547743945e-06,
      "loss": 0.4421,
      "step": 10736
    },
    {
      "epoch": 0.49272635491716765,
      "grad_norm": 0.4370076358318329,
      "learning_rate": 9.474977180856339e-06,
      "loss": 0.3821,
      "step": 10737
    },
    {
      "epoch": 0.4927722454224221,
      "grad_norm": 0.46614742279052734,
      "learning_rate": 9.474867803207826e-06,
      "loss": 0.3748,
      "step": 10738
    },
    {
      "epoch": 0.49281813592767654,
      "grad_norm": 0.42606380581855774,
      "learning_rate": 9.474758414798668e-06,
      "loss": 0.3502,
      "step": 10739
    },
    {
      "epoch": 0.49286402643293103,
      "grad_norm": 0.46804261207580566,
      "learning_rate": 9.474649015629126e-06,
      "loss": 0.3631,
      "step": 10740
    },
    {
      "epoch": 0.4929099169381855,
      "grad_norm": 0.48882976174354553,
      "learning_rate": 9.474539605699468e-06,
      "loss": 0.3939,
      "step": 10741
    },
    {
      "epoch": 0.49295580744344,
      "grad_norm": 0.4956061542034149,
      "learning_rate": 9.474430185009953e-06,
      "loss": 0.4319,
      "step": 10742
    },
    {
      "epoch": 0.4930016979486944,
      "grad_norm": 0.4858354330062866,
      "learning_rate": 9.474320753560849e-06,
      "loss": 0.4353,
      "step": 10743
    },
    {
      "epoch": 0.49304758845394886,
      "grad_norm": 0.4654906392097473,
      "learning_rate": 9.474211311352411e-06,
      "loss": 0.438,
      "step": 10744
    },
    {
      "epoch": 0.49309347895920336,
      "grad_norm": 0.4981384873390198,
      "learning_rate": 9.47410185838491e-06,
      "loss": 0.4793,
      "step": 10745
    },
    {
      "epoch": 0.4931393694644578,
      "grad_norm": 0.5082470774650574,
      "learning_rate": 9.473992394658604e-06,
      "loss": 0.3928,
      "step": 10746
    },
    {
      "epoch": 0.49318525996971224,
      "grad_norm": 0.5102579593658447,
      "learning_rate": 9.473882920173761e-06,
      "loss": 0.3757,
      "step": 10747
    },
    {
      "epoch": 0.49323115047496674,
      "grad_norm": 0.4583779573440552,
      "learning_rate": 9.473773434930639e-06,
      "loss": 0.4549,
      "step": 10748
    },
    {
      "epoch": 0.4932770409802212,
      "grad_norm": 0.471036821603775,
      "learning_rate": 9.473663938929504e-06,
      "loss": 0.3451,
      "step": 10749
    },
    {
      "epoch": 0.4933229314854757,
      "grad_norm": 0.43662160634994507,
      "learning_rate": 9.47355443217062e-06,
      "loss": 0.3752,
      "step": 10750
    },
    {
      "epoch": 0.4933688219907301,
      "grad_norm": 0.4624191224575043,
      "learning_rate": 9.47344491465425e-06,
      "loss": 0.3744,
      "step": 10751
    },
    {
      "epoch": 0.49341471249598456,
      "grad_norm": 0.4741438031196594,
      "learning_rate": 9.473335386380657e-06,
      "loss": 0.3788,
      "step": 10752
    },
    {
      "epoch": 0.49346060300123906,
      "grad_norm": 0.422494500875473,
      "learning_rate": 9.473225847350103e-06,
      "loss": 0.3237,
      "step": 10753
    },
    {
      "epoch": 0.4935064935064935,
      "grad_norm": 0.48873844742774963,
      "learning_rate": 9.473116297562854e-06,
      "loss": 0.4374,
      "step": 10754
    },
    {
      "epoch": 0.49355238401174795,
      "grad_norm": 0.5160603523254395,
      "learning_rate": 9.47300673701917e-06,
      "loss": 0.4394,
      "step": 10755
    },
    {
      "epoch": 0.49359827451700244,
      "grad_norm": 0.4478933811187744,
      "learning_rate": 9.472897165719318e-06,
      "loss": 0.3853,
      "step": 10756
    },
    {
      "epoch": 0.4936441650222569,
      "grad_norm": 0.4606477916240692,
      "learning_rate": 9.47278758366356e-06,
      "loss": 0.4096,
      "step": 10757
    },
    {
      "epoch": 0.4936900555275114,
      "grad_norm": 0.5313938856124878,
      "learning_rate": 9.472677990852159e-06,
      "loss": 0.4168,
      "step": 10758
    },
    {
      "epoch": 0.4937359460327658,
      "grad_norm": 0.4804835021495819,
      "learning_rate": 9.472568387285378e-06,
      "loss": 0.4234,
      "step": 10759
    },
    {
      "epoch": 0.49378183653802027,
      "grad_norm": 0.46497994661331177,
      "learning_rate": 9.472458772963485e-06,
      "loss": 0.4534,
      "step": 10760
    },
    {
      "epoch": 0.49382772704327477,
      "grad_norm": 0.4685143828392029,
      "learning_rate": 9.472349147886736e-06,
      "loss": 0.453,
      "step": 10761
    },
    {
      "epoch": 0.4938736175485292,
      "grad_norm": 0.43738800287246704,
      "learning_rate": 9.472239512055402e-06,
      "loss": 0.3224,
      "step": 10762
    },
    {
      "epoch": 0.49391950805378365,
      "grad_norm": 0.4840410351753235,
      "learning_rate": 9.472129865469742e-06,
      "loss": 0.3878,
      "step": 10763
    },
    {
      "epoch": 0.49396539855903815,
      "grad_norm": 0.44155141711235046,
      "learning_rate": 9.472020208130022e-06,
      "loss": 0.3661,
      "step": 10764
    },
    {
      "epoch": 0.4940112890642926,
      "grad_norm": 0.45112836360931396,
      "learning_rate": 9.471910540036503e-06,
      "loss": 0.374,
      "step": 10765
    },
    {
      "epoch": 0.49405717956954703,
      "grad_norm": 0.47623029351234436,
      "learning_rate": 9.471800861189453e-06,
      "loss": 0.4311,
      "step": 10766
    },
    {
      "epoch": 0.49410307007480153,
      "grad_norm": 0.4813067615032196,
      "learning_rate": 9.471691171589132e-06,
      "loss": 0.4031,
      "step": 10767
    },
    {
      "epoch": 0.494148960580056,
      "grad_norm": 0.47675156593322754,
      "learning_rate": 9.471581471235805e-06,
      "loss": 0.4343,
      "step": 10768
    },
    {
      "epoch": 0.49419485108531047,
      "grad_norm": 0.4491693377494812,
      "learning_rate": 9.471471760129735e-06,
      "loss": 0.433,
      "step": 10769
    },
    {
      "epoch": 0.4942407415905649,
      "grad_norm": 0.4628542363643646,
      "learning_rate": 9.47136203827119e-06,
      "loss": 0.456,
      "step": 10770
    },
    {
      "epoch": 0.49428663209581936,
      "grad_norm": 0.46257779002189636,
      "learning_rate": 9.471252305660429e-06,
      "loss": 0.43,
      "step": 10771
    },
    {
      "epoch": 0.49433252260107385,
      "grad_norm": 0.49147915840148926,
      "learning_rate": 9.471142562297714e-06,
      "loss": 0.4763,
      "step": 10772
    },
    {
      "epoch": 0.4943784131063283,
      "grad_norm": 0.4747050404548645,
      "learning_rate": 9.471032808183316e-06,
      "loss": 0.4369,
      "step": 10773
    },
    {
      "epoch": 0.49442430361158274,
      "grad_norm": 0.4625003933906555,
      "learning_rate": 9.470923043317493e-06,
      "loss": 0.3201,
      "step": 10774
    },
    {
      "epoch": 0.49447019411683724,
      "grad_norm": 0.48790210485458374,
      "learning_rate": 9.470813267700512e-06,
      "loss": 0.4353,
      "step": 10775
    },
    {
      "epoch": 0.4945160846220917,
      "grad_norm": 0.4906661808490753,
      "learning_rate": 9.470703481332638e-06,
      "loss": 0.4954,
      "step": 10776
    },
    {
      "epoch": 0.4945619751273462,
      "grad_norm": 0.5141158699989319,
      "learning_rate": 9.47059368421413e-06,
      "loss": 0.5236,
      "step": 10777
    },
    {
      "epoch": 0.4946078656326006,
      "grad_norm": 0.46352094411849976,
      "learning_rate": 9.470483876345256e-06,
      "loss": 0.4038,
      "step": 10778
    },
    {
      "epoch": 0.49465375613785506,
      "grad_norm": 0.4780634641647339,
      "learning_rate": 9.47037405772628e-06,
      "loss": 0.4054,
      "step": 10779
    },
    {
      "epoch": 0.49469964664310956,
      "grad_norm": 0.4802604615688324,
      "learning_rate": 9.470264228357465e-06,
      "loss": 0.4441,
      "step": 10780
    },
    {
      "epoch": 0.494745537148364,
      "grad_norm": 0.49443528056144714,
      "learning_rate": 9.470154388239076e-06,
      "loss": 0.5114,
      "step": 10781
    },
    {
      "epoch": 0.49479142765361844,
      "grad_norm": 0.4826546609401703,
      "learning_rate": 9.470044537371376e-06,
      "loss": 0.4254,
      "step": 10782
    },
    {
      "epoch": 0.49483731815887294,
      "grad_norm": 0.5457146167755127,
      "learning_rate": 9.469934675754629e-06,
      "loss": 0.4759,
      "step": 10783
    },
    {
      "epoch": 0.4948832086641274,
      "grad_norm": 0.47386622428894043,
      "learning_rate": 9.4698248033891e-06,
      "loss": 0.4428,
      "step": 10784
    },
    {
      "epoch": 0.4949290991693819,
      "grad_norm": 0.4439961910247803,
      "learning_rate": 9.469714920275055e-06,
      "loss": 0.3831,
      "step": 10785
    },
    {
      "epoch": 0.4949749896746363,
      "grad_norm": 0.4569987654685974,
      "learning_rate": 9.469605026412754e-06,
      "loss": 0.3715,
      "step": 10786
    },
    {
      "epoch": 0.49502088017989077,
      "grad_norm": 0.49769458174705505,
      "learning_rate": 9.469495121802463e-06,
      "loss": 0.5095,
      "step": 10787
    },
    {
      "epoch": 0.49506677068514526,
      "grad_norm": 0.5076336860656738,
      "learning_rate": 9.469385206444449e-06,
      "loss": 0.4933,
      "step": 10788
    },
    {
      "epoch": 0.4951126611903997,
      "grad_norm": 0.4258926808834076,
      "learning_rate": 9.469275280338974e-06,
      "loss": 0.333,
      "step": 10789
    },
    {
      "epoch": 0.49515855169565415,
      "grad_norm": 0.4661965072154999,
      "learning_rate": 9.469165343486302e-06,
      "loss": 0.389,
      "step": 10790
    },
    {
      "epoch": 0.49520444220090865,
      "grad_norm": 0.46809253096580505,
      "learning_rate": 9.469055395886697e-06,
      "loss": 0.4094,
      "step": 10791
    },
    {
      "epoch": 0.4952503327061631,
      "grad_norm": 0.4882303476333618,
      "learning_rate": 9.468945437540424e-06,
      "loss": 0.3824,
      "step": 10792
    },
    {
      "epoch": 0.49529622321141753,
      "grad_norm": 0.4730532765388489,
      "learning_rate": 9.46883546844775e-06,
      "loss": 0.3591,
      "step": 10793
    },
    {
      "epoch": 0.49534211371667203,
      "grad_norm": 0.48045167326927185,
      "learning_rate": 9.468725488608936e-06,
      "loss": 0.3237,
      "step": 10794
    },
    {
      "epoch": 0.49538800422192647,
      "grad_norm": 0.5047442317008972,
      "learning_rate": 9.468615498024247e-06,
      "loss": 0.4381,
      "step": 10795
    },
    {
      "epoch": 0.49543389472718097,
      "grad_norm": 0.49691057205200195,
      "learning_rate": 9.468505496693947e-06,
      "loss": 0.4573,
      "step": 10796
    },
    {
      "epoch": 0.4954797852324354,
      "grad_norm": 0.424167662858963,
      "learning_rate": 9.468395484618305e-06,
      "loss": 0.3622,
      "step": 10797
    },
    {
      "epoch": 0.49552567573768985,
      "grad_norm": 0.47169479727745056,
      "learning_rate": 9.46828546179758e-06,
      "loss": 0.3843,
      "step": 10798
    },
    {
      "epoch": 0.49557156624294435,
      "grad_norm": 0.5038683414459229,
      "learning_rate": 9.468175428232038e-06,
      "loss": 0.4184,
      "step": 10799
    },
    {
      "epoch": 0.4956174567481988,
      "grad_norm": 0.4351760745048523,
      "learning_rate": 9.468065383921945e-06,
      "loss": 0.342,
      "step": 10800
    },
    {
      "epoch": 0.49566334725345323,
      "grad_norm": 0.4330252408981323,
      "learning_rate": 9.467955328867564e-06,
      "loss": 0.3461,
      "step": 10801
    },
    {
      "epoch": 0.49570923775870773,
      "grad_norm": 0.4928853511810303,
      "learning_rate": 9.467845263069163e-06,
      "loss": 0.449,
      "step": 10802
    },
    {
      "epoch": 0.4957551282639622,
      "grad_norm": 0.43306809663772583,
      "learning_rate": 9.467735186527002e-06,
      "loss": 0.3363,
      "step": 10803
    },
    {
      "epoch": 0.4958010187692167,
      "grad_norm": 0.4502425193786621,
      "learning_rate": 9.467625099241349e-06,
      "loss": 0.3988,
      "step": 10804
    },
    {
      "epoch": 0.4958469092744711,
      "grad_norm": 0.44554266333580017,
      "learning_rate": 9.467515001212466e-06,
      "loss": 0.3805,
      "step": 10805
    },
    {
      "epoch": 0.49589279977972556,
      "grad_norm": 0.5150051712989807,
      "learning_rate": 9.467404892440621e-06,
      "loss": 0.5178,
      "step": 10806
    },
    {
      "epoch": 0.49593869028498005,
      "grad_norm": 0.4569566249847412,
      "learning_rate": 9.467294772926077e-06,
      "loss": 0.3887,
      "step": 10807
    },
    {
      "epoch": 0.4959845807902345,
      "grad_norm": 0.44354185461997986,
      "learning_rate": 9.467184642669099e-06,
      "loss": 0.3555,
      "step": 10808
    },
    {
      "epoch": 0.49603047129548894,
      "grad_norm": 0.46612823009490967,
      "learning_rate": 9.467074501669952e-06,
      "loss": 0.4138,
      "step": 10809
    },
    {
      "epoch": 0.49607636180074344,
      "grad_norm": 0.4554126560688019,
      "learning_rate": 9.4669643499289e-06,
      "loss": 0.3604,
      "step": 10810
    },
    {
      "epoch": 0.4961222523059979,
      "grad_norm": 0.43482455611228943,
      "learning_rate": 9.466854187446211e-06,
      "loss": 0.4205,
      "step": 10811
    },
    {
      "epoch": 0.4961681428112524,
      "grad_norm": 0.442158579826355,
      "learning_rate": 9.466744014222145e-06,
      "loss": 0.3966,
      "step": 10812
    },
    {
      "epoch": 0.4962140333165068,
      "grad_norm": 0.4627056121826172,
      "learning_rate": 9.466633830256969e-06,
      "loss": 0.4574,
      "step": 10813
    },
    {
      "epoch": 0.49625992382176126,
      "grad_norm": 0.46705663204193115,
      "learning_rate": 9.46652363555095e-06,
      "loss": 0.4315,
      "step": 10814
    },
    {
      "epoch": 0.49630581432701576,
      "grad_norm": 0.44923144578933716,
      "learning_rate": 9.46641343010435e-06,
      "loss": 0.3555,
      "step": 10815
    },
    {
      "epoch": 0.4963517048322702,
      "grad_norm": 0.4127691388130188,
      "learning_rate": 9.466303213917437e-06,
      "loss": 0.3065,
      "step": 10816
    },
    {
      "epoch": 0.49639759533752464,
      "grad_norm": 0.448308527469635,
      "learning_rate": 9.466192986990472e-06,
      "loss": 0.4083,
      "step": 10817
    },
    {
      "epoch": 0.49644348584277914,
      "grad_norm": 0.4670823812484741,
      "learning_rate": 9.466082749323724e-06,
      "loss": 0.4168,
      "step": 10818
    },
    {
      "epoch": 0.4964893763480336,
      "grad_norm": 0.42175057530403137,
      "learning_rate": 9.465972500917456e-06,
      "loss": 0.3077,
      "step": 10819
    },
    {
      "epoch": 0.4965352668532881,
      "grad_norm": 0.48070865869522095,
      "learning_rate": 9.465862241771932e-06,
      "loss": 0.4887,
      "step": 10820
    },
    {
      "epoch": 0.4965811573585425,
      "grad_norm": 0.43024805188179016,
      "learning_rate": 9.465751971887422e-06,
      "loss": 0.3409,
      "step": 10821
    },
    {
      "epoch": 0.49662704786379697,
      "grad_norm": 0.429681658744812,
      "learning_rate": 9.465641691264186e-06,
      "loss": 0.3514,
      "step": 10822
    },
    {
      "epoch": 0.49667293836905146,
      "grad_norm": 0.4429875314235687,
      "learning_rate": 9.465531399902492e-06,
      "loss": 0.3687,
      "step": 10823
    },
    {
      "epoch": 0.4967188288743059,
      "grad_norm": 0.5693091750144958,
      "learning_rate": 9.465421097802603e-06,
      "loss": 0.4011,
      "step": 10824
    },
    {
      "epoch": 0.49676471937956035,
      "grad_norm": 0.45220044255256653,
      "learning_rate": 9.465310784964787e-06,
      "loss": 0.3353,
      "step": 10825
    },
    {
      "epoch": 0.49681060988481485,
      "grad_norm": 0.5061473250389099,
      "learning_rate": 9.465200461389308e-06,
      "loss": 0.4995,
      "step": 10826
    },
    {
      "epoch": 0.4968565003900693,
      "grad_norm": 0.48667842149734497,
      "learning_rate": 9.46509012707643e-06,
      "loss": 0.412,
      "step": 10827
    },
    {
      "epoch": 0.49690239089532373,
      "grad_norm": 0.6104325652122498,
      "learning_rate": 9.46497978202642e-06,
      "loss": 0.3776,
      "step": 10828
    },
    {
      "epoch": 0.49694828140057823,
      "grad_norm": 0.4795345366001129,
      "learning_rate": 9.464869426239544e-06,
      "loss": 0.4681,
      "step": 10829
    },
    {
      "epoch": 0.49699417190583267,
      "grad_norm": 0.557748019695282,
      "learning_rate": 9.464759059716064e-06,
      "loss": 0.4729,
      "step": 10830
    },
    {
      "epoch": 0.49704006241108717,
      "grad_norm": 0.4785155951976776,
      "learning_rate": 9.46464868245625e-06,
      "loss": 0.4545,
      "step": 10831
    },
    {
      "epoch": 0.4970859529163416,
      "grad_norm": 0.47897031903266907,
      "learning_rate": 9.464538294460363e-06,
      "loss": 0.3708,
      "step": 10832
    },
    {
      "epoch": 0.49713184342159605,
      "grad_norm": 0.5082811117172241,
      "learning_rate": 9.464427895728672e-06,
      "loss": 0.5203,
      "step": 10833
    },
    {
      "epoch": 0.49717773392685055,
      "grad_norm": 0.435053288936615,
      "learning_rate": 9.464317486261439e-06,
      "loss": 0.3852,
      "step": 10834
    },
    {
      "epoch": 0.497223624432105,
      "grad_norm": 0.44608721137046814,
      "learning_rate": 9.464207066058933e-06,
      "loss": 0.3908,
      "step": 10835
    },
    {
      "epoch": 0.49726951493735944,
      "grad_norm": 0.4874432682991028,
      "learning_rate": 9.464096635121417e-06,
      "loss": 0.4701,
      "step": 10836
    },
    {
      "epoch": 0.49731540544261393,
      "grad_norm": 0.47423055768013,
      "learning_rate": 9.463986193449157e-06,
      "loss": 0.3958,
      "step": 10837
    },
    {
      "epoch": 0.4973612959478684,
      "grad_norm": 0.4569908082485199,
      "learning_rate": 9.463875741042421e-06,
      "loss": 0.3969,
      "step": 10838
    },
    {
      "epoch": 0.4974071864531229,
      "grad_norm": 0.4475950300693512,
      "learning_rate": 9.463765277901471e-06,
      "loss": 0.3766,
      "step": 10839
    },
    {
      "epoch": 0.4974530769583773,
      "grad_norm": 0.4636659324169159,
      "learning_rate": 9.463654804026575e-06,
      "loss": 0.4222,
      "step": 10840
    },
    {
      "epoch": 0.49749896746363176,
      "grad_norm": 0.4321160912513733,
      "learning_rate": 9.463544319417999e-06,
      "loss": 0.3752,
      "step": 10841
    },
    {
      "epoch": 0.49754485796888626,
      "grad_norm": 0.48935502767562866,
      "learning_rate": 9.463433824076007e-06,
      "loss": 0.4471,
      "step": 10842
    },
    {
      "epoch": 0.4975907484741407,
      "grad_norm": 0.46794259548187256,
      "learning_rate": 9.463323318000864e-06,
      "loss": 0.4017,
      "step": 10843
    },
    {
      "epoch": 0.49763663897939514,
      "grad_norm": 0.5219445824623108,
      "learning_rate": 9.463212801192838e-06,
      "loss": 0.5316,
      "step": 10844
    },
    {
      "epoch": 0.49768252948464964,
      "grad_norm": 0.4497554302215576,
      "learning_rate": 9.463102273652194e-06,
      "loss": 0.379,
      "step": 10845
    },
    {
      "epoch": 0.4977284199899041,
      "grad_norm": 0.4437973201274872,
      "learning_rate": 9.462991735379196e-06,
      "loss": 0.402,
      "step": 10846
    },
    {
      "epoch": 0.4977743104951586,
      "grad_norm": 0.4793253242969513,
      "learning_rate": 9.462881186374115e-06,
      "loss": 0.4369,
      "step": 10847
    },
    {
      "epoch": 0.497820201000413,
      "grad_norm": 0.46814313530921936,
      "learning_rate": 9.46277062663721e-06,
      "loss": 0.4732,
      "step": 10848
    },
    {
      "epoch": 0.49786609150566746,
      "grad_norm": 0.4704726040363312,
      "learning_rate": 9.462660056168753e-06,
      "loss": 0.4177,
      "step": 10849
    },
    {
      "epoch": 0.49791198201092196,
      "grad_norm": 0.47194790840148926,
      "learning_rate": 9.462549474969004e-06,
      "loss": 0.4517,
      "step": 10850
    },
    {
      "epoch": 0.4979578725161764,
      "grad_norm": 0.5083450078964233,
      "learning_rate": 9.462438883038234e-06,
      "loss": 0.3601,
      "step": 10851
    },
    {
      "epoch": 0.49800376302143085,
      "grad_norm": 0.4686483144760132,
      "learning_rate": 9.462328280376705e-06,
      "loss": 0.4115,
      "step": 10852
    },
    {
      "epoch": 0.49804965352668534,
      "grad_norm": 0.4468414783477783,
      "learning_rate": 9.462217666984687e-06,
      "loss": 0.3686,
      "step": 10853
    },
    {
      "epoch": 0.4980955440319398,
      "grad_norm": 0.4344910681247711,
      "learning_rate": 9.462107042862443e-06,
      "loss": 0.3288,
      "step": 10854
    },
    {
      "epoch": 0.4981414345371942,
      "grad_norm": 0.4448080062866211,
      "learning_rate": 9.461996408010239e-06,
      "loss": 0.403,
      "step": 10855
    },
    {
      "epoch": 0.4981873250424487,
      "grad_norm": 0.5039364099502563,
      "learning_rate": 9.461885762428343e-06,
      "loss": 0.4173,
      "step": 10856
    },
    {
      "epoch": 0.49823321554770317,
      "grad_norm": 0.4316632151603699,
      "learning_rate": 9.46177510611702e-06,
      "loss": 0.3294,
      "step": 10857
    },
    {
      "epoch": 0.49827910605295767,
      "grad_norm": 0.45275360345840454,
      "learning_rate": 9.461664439076534e-06,
      "loss": 0.3848,
      "step": 10858
    },
    {
      "epoch": 0.4983249965582121,
      "grad_norm": 0.45665857195854187,
      "learning_rate": 9.461553761307156e-06,
      "loss": 0.3552,
      "step": 10859
    },
    {
      "epoch": 0.49837088706346655,
      "grad_norm": 0.4466211199760437,
      "learning_rate": 9.461443072809146e-06,
      "loss": 0.412,
      "step": 10860
    },
    {
      "epoch": 0.49841677756872105,
      "grad_norm": 0.4732917249202728,
      "learning_rate": 9.461332373582776e-06,
      "loss": 0.4232,
      "step": 10861
    },
    {
      "epoch": 0.4984626680739755,
      "grad_norm": 0.5152777433395386,
      "learning_rate": 9.461221663628311e-06,
      "loss": 0.5253,
      "step": 10862
    },
    {
      "epoch": 0.49850855857922993,
      "grad_norm": 0.4891108572483063,
      "learning_rate": 9.461110942946014e-06,
      "loss": 0.4449,
      "step": 10863
    },
    {
      "epoch": 0.49855444908448443,
      "grad_norm": 0.4732539653778076,
      "learning_rate": 9.461000211536153e-06,
      "loss": 0.4368,
      "step": 10864
    },
    {
      "epoch": 0.4986003395897389,
      "grad_norm": 0.4699980318546295,
      "learning_rate": 9.460889469398994e-06,
      "loss": 0.414,
      "step": 10865
    },
    {
      "epoch": 0.49864623009499337,
      "grad_norm": 0.456589013338089,
      "learning_rate": 9.460778716534803e-06,
      "loss": 0.3833,
      "step": 10866
    },
    {
      "epoch": 0.4986921206002478,
      "grad_norm": 0.48536157608032227,
      "learning_rate": 9.46066795294385e-06,
      "loss": 0.4275,
      "step": 10867
    },
    {
      "epoch": 0.49873801110550225,
      "grad_norm": 0.4452584385871887,
      "learning_rate": 9.460557178626397e-06,
      "loss": 0.3514,
      "step": 10868
    },
    {
      "epoch": 0.49878390161075675,
      "grad_norm": 0.4469469487667084,
      "learning_rate": 9.460446393582711e-06,
      "loss": 0.3919,
      "step": 10869
    },
    {
      "epoch": 0.4988297921160112,
      "grad_norm": 0.5035808682441711,
      "learning_rate": 9.46033559781306e-06,
      "loss": 0.4527,
      "step": 10870
    },
    {
      "epoch": 0.49887568262126564,
      "grad_norm": 0.48639756441116333,
      "learning_rate": 9.460224791317709e-06,
      "loss": 0.4663,
      "step": 10871
    },
    {
      "epoch": 0.49892157312652013,
      "grad_norm": 0.5119649767875671,
      "learning_rate": 9.460113974096925e-06,
      "loss": 0.4516,
      "step": 10872
    },
    {
      "epoch": 0.4989674636317746,
      "grad_norm": 0.48175814747810364,
      "learning_rate": 9.460003146150975e-06,
      "loss": 0.3247,
      "step": 10873
    },
    {
      "epoch": 0.4990133541370291,
      "grad_norm": 0.5030218958854675,
      "learning_rate": 9.459892307480126e-06,
      "loss": 0.4418,
      "step": 10874
    },
    {
      "epoch": 0.4990592446422835,
      "grad_norm": 0.4917619228363037,
      "learning_rate": 9.459781458084642e-06,
      "loss": 0.4748,
      "step": 10875
    },
    {
      "epoch": 0.49910513514753796,
      "grad_norm": 0.44831913709640503,
      "learning_rate": 9.459670597964793e-06,
      "loss": 0.3791,
      "step": 10876
    },
    {
      "epoch": 0.49915102565279246,
      "grad_norm": 0.45432212948799133,
      "learning_rate": 9.459559727120842e-06,
      "loss": 0.3857,
      "step": 10877
    },
    {
      "epoch": 0.4991969161580469,
      "grad_norm": 0.4490716755390167,
      "learning_rate": 9.459448845553058e-06,
      "loss": 0.354,
      "step": 10878
    },
    {
      "epoch": 0.49924280666330134,
      "grad_norm": 0.45428746938705444,
      "learning_rate": 9.459337953261706e-06,
      "loss": 0.4084,
      "step": 10879
    },
    {
      "epoch": 0.49928869716855584,
      "grad_norm": 0.5144004821777344,
      "learning_rate": 9.459227050247055e-06,
      "loss": 0.4153,
      "step": 10880
    },
    {
      "epoch": 0.4993345876738103,
      "grad_norm": 0.4629228711128235,
      "learning_rate": 9.459116136509369e-06,
      "loss": 0.4218,
      "step": 10881
    },
    {
      "epoch": 0.4993804781790647,
      "grad_norm": 0.44414469599723816,
      "learning_rate": 9.459005212048917e-06,
      "loss": 0.3562,
      "step": 10882
    },
    {
      "epoch": 0.4994263686843192,
      "grad_norm": 0.4481891095638275,
      "learning_rate": 9.458894276865965e-06,
      "loss": 0.3429,
      "step": 10883
    },
    {
      "epoch": 0.49947225918957366,
      "grad_norm": 0.4644007086753845,
      "learning_rate": 9.458783330960779e-06,
      "loss": 0.3752,
      "step": 10884
    },
    {
      "epoch": 0.49951814969482816,
      "grad_norm": 0.43031781911849976,
      "learning_rate": 9.458672374333626e-06,
      "loss": 0.3614,
      "step": 10885
    },
    {
      "epoch": 0.4995640402000826,
      "grad_norm": 0.5275810360908508,
      "learning_rate": 9.458561406984774e-06,
      "loss": 0.4922,
      "step": 10886
    },
    {
      "epoch": 0.49960993070533705,
      "grad_norm": 0.4725988507270813,
      "learning_rate": 9.45845042891449e-06,
      "loss": 0.4443,
      "step": 10887
    },
    {
      "epoch": 0.49965582121059154,
      "grad_norm": 0.4356434941291809,
      "learning_rate": 9.458339440123036e-06,
      "loss": 0.3544,
      "step": 10888
    },
    {
      "epoch": 0.499701711715846,
      "grad_norm": 0.466789573431015,
      "learning_rate": 9.458228440610686e-06,
      "loss": 0.4302,
      "step": 10889
    },
    {
      "epoch": 0.49974760222110043,
      "grad_norm": 0.45808231830596924,
      "learning_rate": 9.458117430377703e-06,
      "loss": 0.3713,
      "step": 10890
    },
    {
      "epoch": 0.4997934927263549,
      "grad_norm": 0.5023537278175354,
      "learning_rate": 9.458006409424354e-06,
      "loss": 0.491,
      "step": 10891
    },
    {
      "epoch": 0.49983938323160937,
      "grad_norm": 0.4618149995803833,
      "learning_rate": 9.457895377750907e-06,
      "loss": 0.431,
      "step": 10892
    },
    {
      "epoch": 0.49988527373686387,
      "grad_norm": 0.4783206880092621,
      "learning_rate": 9.457784335357629e-06,
      "loss": 0.4146,
      "step": 10893
    },
    {
      "epoch": 0.4999311642421183,
      "grad_norm": 0.4852701723575592,
      "learning_rate": 9.457673282244785e-06,
      "loss": 0.4225,
      "step": 10894
    },
    {
      "epoch": 0.49997705474737275,
      "grad_norm": 0.4660445749759674,
      "learning_rate": 9.457562218412646e-06,
      "loss": 0.3648,
      "step": 10895
    },
    {
      "epoch": 0.5000229452526272,
      "grad_norm": 0.44535401463508606,
      "learning_rate": 9.457451143861477e-06,
      "loss": 0.3388,
      "step": 10896
    },
    {
      "epoch": 0.5000688357578817,
      "grad_norm": 0.4588528573513031,
      "learning_rate": 9.457340058591542e-06,
      "loss": 0.3814,
      "step": 10897
    },
    {
      "epoch": 0.5001147262631361,
      "grad_norm": 0.4969598948955536,
      "learning_rate": 9.457228962603112e-06,
      "loss": 0.4511,
      "step": 10898
    },
    {
      "epoch": 0.5001606167683906,
      "grad_norm": 0.4603307843208313,
      "learning_rate": 9.457117855896454e-06,
      "loss": 0.4057,
      "step": 10899
    },
    {
      "epoch": 0.5002065072736451,
      "grad_norm": 0.4576098322868347,
      "learning_rate": 9.457006738471832e-06,
      "loss": 0.4008,
      "step": 10900
    },
    {
      "epoch": 0.5002523977788995,
      "grad_norm": 0.45939192175865173,
      "learning_rate": 9.456895610329518e-06,
      "loss": 0.3751,
      "step": 10901
    },
    {
      "epoch": 0.500298288284154,
      "grad_norm": 0.49289727210998535,
      "learning_rate": 9.456784471469775e-06,
      "loss": 0.4668,
      "step": 10902
    },
    {
      "epoch": 0.5003441787894085,
      "grad_norm": 0.4463818371295929,
      "learning_rate": 9.456673321892872e-06,
      "loss": 0.3387,
      "step": 10903
    },
    {
      "epoch": 0.5003900692946629,
      "grad_norm": 0.4723441004753113,
      "learning_rate": 9.456562161599078e-06,
      "loss": 0.4563,
      "step": 10904
    },
    {
      "epoch": 0.5004359597999174,
      "grad_norm": 0.4962775707244873,
      "learning_rate": 9.456450990588657e-06,
      "loss": 0.4172,
      "step": 10905
    },
    {
      "epoch": 0.5004818503051719,
      "grad_norm": 0.46568378806114197,
      "learning_rate": 9.456339808861878e-06,
      "loss": 0.3816,
      "step": 10906
    },
    {
      "epoch": 0.5005277408104263,
      "grad_norm": 0.47389382123947144,
      "learning_rate": 9.45622861641901e-06,
      "loss": 0.4341,
      "step": 10907
    },
    {
      "epoch": 0.5005736313156808,
      "grad_norm": 0.4549764394760132,
      "learning_rate": 9.456117413260316e-06,
      "loss": 0.4217,
      "step": 10908
    },
    {
      "epoch": 0.5006195218209353,
      "grad_norm": 0.44340980052948,
      "learning_rate": 9.456006199386067e-06,
      "loss": 0.3703,
      "step": 10909
    },
    {
      "epoch": 0.5006654123261897,
      "grad_norm": 0.5011385083198547,
      "learning_rate": 9.45589497479653e-06,
      "loss": 0.4669,
      "step": 10910
    },
    {
      "epoch": 0.5007113028314442,
      "grad_norm": 0.49119576811790466,
      "learning_rate": 9.455783739491971e-06,
      "loss": 0.4572,
      "step": 10911
    },
    {
      "epoch": 0.5007571933366987,
      "grad_norm": 0.46003612875938416,
      "learning_rate": 9.455672493472659e-06,
      "loss": 0.3605,
      "step": 10912
    },
    {
      "epoch": 0.500803083841953,
      "grad_norm": 0.4600478410720825,
      "learning_rate": 9.455561236738861e-06,
      "loss": 0.3545,
      "step": 10913
    },
    {
      "epoch": 0.5008489743472075,
      "grad_norm": 0.4378669559955597,
      "learning_rate": 9.455449969290844e-06,
      "loss": 0.3451,
      "step": 10914
    },
    {
      "epoch": 0.500894864852462,
      "grad_norm": 0.4849533438682556,
      "learning_rate": 9.455338691128878e-06,
      "loss": 0.4146,
      "step": 10915
    },
    {
      "epoch": 0.5009407553577165,
      "grad_norm": 0.5508562922477722,
      "learning_rate": 9.455227402253225e-06,
      "loss": 0.5187,
      "step": 10916
    },
    {
      "epoch": 0.5009866458629709,
      "grad_norm": 0.4776171147823334,
      "learning_rate": 9.45511610266416e-06,
      "loss": 0.3663,
      "step": 10917
    },
    {
      "epoch": 0.5010325363682254,
      "grad_norm": 0.474141001701355,
      "learning_rate": 9.455004792361944e-06,
      "loss": 0.3803,
      "step": 10918
    },
    {
      "epoch": 0.5010784268734799,
      "grad_norm": 0.4400337040424347,
      "learning_rate": 9.454893471346848e-06,
      "loss": 0.3934,
      "step": 10919
    },
    {
      "epoch": 0.5011243173787343,
      "grad_norm": 0.46708279848098755,
      "learning_rate": 9.454782139619138e-06,
      "loss": 0.4168,
      "step": 10920
    },
    {
      "epoch": 0.5011702078839888,
      "grad_norm": 0.4862530827522278,
      "learning_rate": 9.454670797179086e-06,
      "loss": 0.4147,
      "step": 10921
    },
    {
      "epoch": 0.5012160983892433,
      "grad_norm": 0.46721550822257996,
      "learning_rate": 9.454559444026955e-06,
      "loss": 0.4531,
      "step": 10922
    },
    {
      "epoch": 0.5012619888944977,
      "grad_norm": 0.44835421442985535,
      "learning_rate": 9.454448080163015e-06,
      "loss": 0.3639,
      "step": 10923
    },
    {
      "epoch": 0.5013078793997522,
      "grad_norm": 0.43290725350379944,
      "learning_rate": 9.454336705587532e-06,
      "loss": 0.3455,
      "step": 10924
    },
    {
      "epoch": 0.5013537699050067,
      "grad_norm": 0.45366838574409485,
      "learning_rate": 9.454225320300776e-06,
      "loss": 0.4542,
      "step": 10925
    },
    {
      "epoch": 0.5013996604102611,
      "grad_norm": 0.4568560719490051,
      "learning_rate": 9.454113924303015e-06,
      "loss": 0.4013,
      "step": 10926
    },
    {
      "epoch": 0.5014455509155156,
      "grad_norm": 0.450371116399765,
      "learning_rate": 9.454002517594514e-06,
      "loss": 0.3984,
      "step": 10927
    },
    {
      "epoch": 0.5014914414207701,
      "grad_norm": 0.6074205040931702,
      "learning_rate": 9.453891100175542e-06,
      "loss": 0.3823,
      "step": 10928
    },
    {
      "epoch": 0.5015373319260245,
      "grad_norm": 0.4859667122364044,
      "learning_rate": 9.45377967204637e-06,
      "loss": 0.4463,
      "step": 10929
    },
    {
      "epoch": 0.501583222431279,
      "grad_norm": 0.4499225914478302,
      "learning_rate": 9.453668233207262e-06,
      "loss": 0.3986,
      "step": 10930
    },
    {
      "epoch": 0.5016291129365335,
      "grad_norm": 0.4626608192920685,
      "learning_rate": 9.453556783658488e-06,
      "loss": 0.39,
      "step": 10931
    },
    {
      "epoch": 0.501675003441788,
      "grad_norm": 0.4317147731781006,
      "learning_rate": 9.453445323400315e-06,
      "loss": 0.3243,
      "step": 10932
    },
    {
      "epoch": 0.5017208939470423,
      "grad_norm": 0.4683949947357178,
      "learning_rate": 9.453333852433013e-06,
      "loss": 0.4438,
      "step": 10933
    },
    {
      "epoch": 0.5017667844522968,
      "grad_norm": 0.45582839846611023,
      "learning_rate": 9.453222370756848e-06,
      "loss": 0.3785,
      "step": 10934
    },
    {
      "epoch": 0.5018126749575513,
      "grad_norm": 0.45699912309646606,
      "learning_rate": 9.453110878372087e-06,
      "loss": 0.3962,
      "step": 10935
    },
    {
      "epoch": 0.5018585654628057,
      "grad_norm": 0.4745451509952545,
      "learning_rate": 9.452999375279002e-06,
      "loss": 0.4242,
      "step": 10936
    },
    {
      "epoch": 0.5019044559680602,
      "grad_norm": 0.4583762288093567,
      "learning_rate": 9.452887861477858e-06,
      "loss": 0.4034,
      "step": 10937
    },
    {
      "epoch": 0.5019503464733147,
      "grad_norm": 0.4873026907444,
      "learning_rate": 9.452776336968924e-06,
      "loss": 0.4433,
      "step": 10938
    },
    {
      "epoch": 0.5019962369785691,
      "grad_norm": 0.46590790152549744,
      "learning_rate": 9.452664801752467e-06,
      "loss": 0.373,
      "step": 10939
    },
    {
      "epoch": 0.5020421274838236,
      "grad_norm": 0.44053980708122253,
      "learning_rate": 9.452553255828758e-06,
      "loss": 0.3849,
      "step": 10940
    },
    {
      "epoch": 0.5020880179890781,
      "grad_norm": 0.4903087317943573,
      "learning_rate": 9.452441699198063e-06,
      "loss": 0.4812,
      "step": 10941
    },
    {
      "epoch": 0.5021339084943325,
      "grad_norm": 0.4747794568538666,
      "learning_rate": 9.452330131860653e-06,
      "loss": 0.3791,
      "step": 10942
    },
    {
      "epoch": 0.502179798999587,
      "grad_norm": 0.43665972352027893,
      "learning_rate": 9.45221855381679e-06,
      "loss": 0.3842,
      "step": 10943
    },
    {
      "epoch": 0.5022256895048415,
      "grad_norm": 0.45434272289276123,
      "learning_rate": 9.45210696506675e-06,
      "loss": 0.3805,
      "step": 10944
    },
    {
      "epoch": 0.5022715800100959,
      "grad_norm": 0.46483394503593445,
      "learning_rate": 9.451995365610797e-06,
      "loss": 0.4074,
      "step": 10945
    },
    {
      "epoch": 0.5023174705153504,
      "grad_norm": 0.44043317437171936,
      "learning_rate": 9.4518837554492e-06,
      "loss": 0.3921,
      "step": 10946
    },
    {
      "epoch": 0.5023633610206049,
      "grad_norm": 0.5069910883903503,
      "learning_rate": 9.451772134582229e-06,
      "loss": 0.5197,
      "step": 10947
    },
    {
      "epoch": 0.5024092515258592,
      "grad_norm": 0.5411222577095032,
      "learning_rate": 9.45166050301015e-06,
      "loss": 0.5132,
      "step": 10948
    },
    {
      "epoch": 0.5024551420311137,
      "grad_norm": 0.4758906364440918,
      "learning_rate": 9.451548860733231e-06,
      "loss": 0.379,
      "step": 10949
    },
    {
      "epoch": 0.5025010325363682,
      "grad_norm": 0.46093040704727173,
      "learning_rate": 9.451437207751744e-06,
      "loss": 0.3877,
      "step": 10950
    },
    {
      "epoch": 0.5025469230416227,
      "grad_norm": 0.4234349727630615,
      "learning_rate": 9.451325544065954e-06,
      "loss": 0.3541,
      "step": 10951
    },
    {
      "epoch": 0.5025928135468771,
      "grad_norm": 0.4991777837276459,
      "learning_rate": 9.451213869676133e-06,
      "loss": 0.4991,
      "step": 10952
    },
    {
      "epoch": 0.5026387040521316,
      "grad_norm": 0.4620932340621948,
      "learning_rate": 9.451102184582546e-06,
      "loss": 0.4213,
      "step": 10953
    },
    {
      "epoch": 0.5026845945573861,
      "grad_norm": 0.4639440178871155,
      "learning_rate": 9.450990488785464e-06,
      "loss": 0.4008,
      "step": 10954
    },
    {
      "epoch": 0.5027304850626405,
      "grad_norm": 0.48175880312919617,
      "learning_rate": 9.450878782285152e-06,
      "loss": 0.4694,
      "step": 10955
    },
    {
      "epoch": 0.502776375567895,
      "grad_norm": 0.46049076318740845,
      "learning_rate": 9.450767065081882e-06,
      "loss": 0.3939,
      "step": 10956
    },
    {
      "epoch": 0.5028222660731495,
      "grad_norm": 0.4564223289489746,
      "learning_rate": 9.450655337175925e-06,
      "loss": 0.3992,
      "step": 10957
    },
    {
      "epoch": 0.5028681565784039,
      "grad_norm": 0.4506520926952362,
      "learning_rate": 9.450543598567545e-06,
      "loss": 0.3645,
      "step": 10958
    },
    {
      "epoch": 0.5029140470836584,
      "grad_norm": 0.45811915397644043,
      "learning_rate": 9.450431849257012e-06,
      "loss": 0.4137,
      "step": 10959
    },
    {
      "epoch": 0.5029599375889129,
      "grad_norm": 0.49642693996429443,
      "learning_rate": 9.450320089244594e-06,
      "loss": 0.5011,
      "step": 10960
    },
    {
      "epoch": 0.5030058280941673,
      "grad_norm": 0.4784826636314392,
      "learning_rate": 9.450208318530562e-06,
      "loss": 0.4272,
      "step": 10961
    },
    {
      "epoch": 0.5030517185994218,
      "grad_norm": 0.4685262441635132,
      "learning_rate": 9.450096537115183e-06,
      "loss": 0.4068,
      "step": 10962
    },
    {
      "epoch": 0.5030976091046763,
      "grad_norm": 0.48505303263664246,
      "learning_rate": 9.449984744998725e-06,
      "loss": 0.4724,
      "step": 10963
    },
    {
      "epoch": 0.5031434996099307,
      "grad_norm": 0.47323817014694214,
      "learning_rate": 9.44987294218146e-06,
      "loss": 0.4041,
      "step": 10964
    },
    {
      "epoch": 0.5031893901151852,
      "grad_norm": 0.446487158536911,
      "learning_rate": 9.449761128663655e-06,
      "loss": 0.3674,
      "step": 10965
    },
    {
      "epoch": 0.5032352806204397,
      "grad_norm": 0.48941147327423096,
      "learning_rate": 9.449649304445579e-06,
      "loss": 0.4872,
      "step": 10966
    },
    {
      "epoch": 0.5032811711256941,
      "grad_norm": 0.41921108961105347,
      "learning_rate": 9.4495374695275e-06,
      "loss": 0.2918,
      "step": 10967
    },
    {
      "epoch": 0.5033270616309485,
      "grad_norm": 0.47301679849624634,
      "learning_rate": 9.449425623909686e-06,
      "loss": 0.4287,
      "step": 10968
    },
    {
      "epoch": 0.503372952136203,
      "grad_norm": 0.4526398181915283,
      "learning_rate": 9.449313767592412e-06,
      "loss": 0.3616,
      "step": 10969
    },
    {
      "epoch": 0.5034188426414575,
      "grad_norm": 0.45543360710144043,
      "learning_rate": 9.44920190057594e-06,
      "loss": 0.3561,
      "step": 10970
    },
    {
      "epoch": 0.5034647331467119,
      "grad_norm": 0.4691105782985687,
      "learning_rate": 9.44909002286054e-06,
      "loss": 0.4069,
      "step": 10971
    },
    {
      "epoch": 0.5035106236519664,
      "grad_norm": 0.4809856116771698,
      "learning_rate": 9.448978134446486e-06,
      "loss": 0.4643,
      "step": 10972
    },
    {
      "epoch": 0.5035565141572209,
      "grad_norm": 0.49662455916404724,
      "learning_rate": 9.448866235334043e-06,
      "loss": 0.4942,
      "step": 10973
    },
    {
      "epoch": 0.5036024046624753,
      "grad_norm": 0.4130520820617676,
      "learning_rate": 9.448754325523479e-06,
      "loss": 0.3395,
      "step": 10974
    },
    {
      "epoch": 0.5036482951677298,
      "grad_norm": 0.5070728659629822,
      "learning_rate": 9.448642405015066e-06,
      "loss": 0.5039,
      "step": 10975
    },
    {
      "epoch": 0.5036941856729843,
      "grad_norm": 0.41832247376441956,
      "learning_rate": 9.448530473809072e-06,
      "loss": 0.3049,
      "step": 10976
    },
    {
      "epoch": 0.5037400761782387,
      "grad_norm": 0.45334652066230774,
      "learning_rate": 9.448418531905766e-06,
      "loss": 0.3939,
      "step": 10977
    },
    {
      "epoch": 0.5037859666834932,
      "grad_norm": 0.4927808940410614,
      "learning_rate": 9.448306579305418e-06,
      "loss": 0.4129,
      "step": 10978
    },
    {
      "epoch": 0.5038318571887477,
      "grad_norm": 0.4754318594932556,
      "learning_rate": 9.448194616008296e-06,
      "loss": 0.4267,
      "step": 10979
    },
    {
      "epoch": 0.5038777476940021,
      "grad_norm": 0.4737919569015503,
      "learning_rate": 9.44808264201467e-06,
      "loss": 0.4352,
      "step": 10980
    },
    {
      "epoch": 0.5039236381992566,
      "grad_norm": 0.4747816026210785,
      "learning_rate": 9.447970657324809e-06,
      "loss": 0.4402,
      "step": 10981
    },
    {
      "epoch": 0.5039695287045111,
      "grad_norm": 0.462495893239975,
      "learning_rate": 9.447858661938982e-06,
      "loss": 0.3933,
      "step": 10982
    },
    {
      "epoch": 0.5040154192097654,
      "grad_norm": 0.44550761580467224,
      "learning_rate": 9.447746655857459e-06,
      "loss": 0.3714,
      "step": 10983
    },
    {
      "epoch": 0.50406130971502,
      "grad_norm": 0.4744863510131836,
      "learning_rate": 9.44763463908051e-06,
      "loss": 0.4387,
      "step": 10984
    },
    {
      "epoch": 0.5041072002202744,
      "grad_norm": 0.49458232522010803,
      "learning_rate": 9.447522611608404e-06,
      "loss": 0.4542,
      "step": 10985
    },
    {
      "epoch": 0.5041530907255289,
      "grad_norm": 0.5333617329597473,
      "learning_rate": 9.447410573441408e-06,
      "loss": 0.4722,
      "step": 10986
    },
    {
      "epoch": 0.5041989812307833,
      "grad_norm": 0.4457358121871948,
      "learning_rate": 9.447298524579793e-06,
      "loss": 0.3642,
      "step": 10987
    },
    {
      "epoch": 0.5042448717360378,
      "grad_norm": 0.4825538694858551,
      "learning_rate": 9.447186465023829e-06,
      "loss": 0.4365,
      "step": 10988
    },
    {
      "epoch": 0.5042907622412923,
      "grad_norm": 0.45974797010421753,
      "learning_rate": 9.447074394773786e-06,
      "loss": 0.4263,
      "step": 10989
    },
    {
      "epoch": 0.5043366527465467,
      "grad_norm": 0.4827609360218048,
      "learning_rate": 9.446962313829933e-06,
      "loss": 0.4573,
      "step": 10990
    },
    {
      "epoch": 0.5043825432518012,
      "grad_norm": 0.44754886627197266,
      "learning_rate": 9.446850222192537e-06,
      "loss": 0.3979,
      "step": 10991
    },
    {
      "epoch": 0.5044284337570557,
      "grad_norm": 0.4962870180606842,
      "learning_rate": 9.446738119861872e-06,
      "loss": 0.4441,
      "step": 10992
    },
    {
      "epoch": 0.5044743242623101,
      "grad_norm": 0.5338955521583557,
      "learning_rate": 9.446626006838203e-06,
      "loss": 0.488,
      "step": 10993
    },
    {
      "epoch": 0.5045202147675646,
      "grad_norm": 0.44123077392578125,
      "learning_rate": 9.446513883121805e-06,
      "loss": 0.401,
      "step": 10994
    },
    {
      "epoch": 0.5045661052728191,
      "grad_norm": 0.42539265751838684,
      "learning_rate": 9.446401748712942e-06,
      "loss": 0.3241,
      "step": 10995
    },
    {
      "epoch": 0.5046119957780735,
      "grad_norm": 0.47869786620140076,
      "learning_rate": 9.446289603611887e-06,
      "loss": 0.3962,
      "step": 10996
    },
    {
      "epoch": 0.504657886283328,
      "grad_norm": 0.4272017478942871,
      "learning_rate": 9.446177447818908e-06,
      "loss": 0.3219,
      "step": 10997
    },
    {
      "epoch": 0.5047037767885825,
      "grad_norm": 0.4733532965183258,
      "learning_rate": 9.446065281334275e-06,
      "loss": 0.4628,
      "step": 10998
    },
    {
      "epoch": 0.5047496672938369,
      "grad_norm": 0.4485187530517578,
      "learning_rate": 9.445953104158258e-06,
      "loss": 0.3655,
      "step": 10999
    },
    {
      "epoch": 0.5047955577990914,
      "grad_norm": 0.4721904695034027,
      "learning_rate": 9.44584091629113e-06,
      "loss": 0.4482,
      "step": 11000
    },
    {
      "epoch": 0.5048414483043459,
      "grad_norm": 0.45469409227371216,
      "learning_rate": 9.445728717733156e-06,
      "loss": 0.4143,
      "step": 11001
    },
    {
      "epoch": 0.5048873388096002,
      "grad_norm": 0.4424894452095032,
      "learning_rate": 9.445616508484607e-06,
      "loss": 0.3927,
      "step": 11002
    },
    {
      "epoch": 0.5049332293148547,
      "grad_norm": 0.4687498211860657,
      "learning_rate": 9.445504288545753e-06,
      "loss": 0.4065,
      "step": 11003
    },
    {
      "epoch": 0.5049791198201092,
      "grad_norm": 0.46504124999046326,
      "learning_rate": 9.445392057916865e-06,
      "loss": 0.3621,
      "step": 11004
    },
    {
      "epoch": 0.5050250103253637,
      "grad_norm": 0.4430292844772339,
      "learning_rate": 9.445279816598212e-06,
      "loss": 0.3743,
      "step": 11005
    },
    {
      "epoch": 0.5050709008306181,
      "grad_norm": 0.4403539299964905,
      "learning_rate": 9.445167564590065e-06,
      "loss": 0.3877,
      "step": 11006
    },
    {
      "epoch": 0.5051167913358726,
      "grad_norm": 0.5480190515518188,
      "learning_rate": 9.445055301892692e-06,
      "loss": 0.5021,
      "step": 11007
    },
    {
      "epoch": 0.5051626818411271,
      "grad_norm": 0.4245948791503906,
      "learning_rate": 9.444943028506363e-06,
      "loss": 0.3349,
      "step": 11008
    },
    {
      "epoch": 0.5052085723463815,
      "grad_norm": 0.5123680233955383,
      "learning_rate": 9.44483074443135e-06,
      "loss": 0.5073,
      "step": 11009
    },
    {
      "epoch": 0.505254462851636,
      "grad_norm": 0.4700130820274353,
      "learning_rate": 9.444718449667922e-06,
      "loss": 0.4152,
      "step": 11010
    },
    {
      "epoch": 0.5053003533568905,
      "grad_norm": 0.4817167818546295,
      "learning_rate": 9.444606144216349e-06,
      "loss": 0.4252,
      "step": 11011
    },
    {
      "epoch": 0.5053462438621449,
      "grad_norm": 0.46350380778312683,
      "learning_rate": 9.4444938280769e-06,
      "loss": 0.3156,
      "step": 11012
    },
    {
      "epoch": 0.5053921343673994,
      "grad_norm": 0.5053478479385376,
      "learning_rate": 9.444381501249846e-06,
      "loss": 0.4161,
      "step": 11013
    },
    {
      "epoch": 0.5054380248726539,
      "grad_norm": 0.47810742259025574,
      "learning_rate": 9.444269163735459e-06,
      "loss": 0.4633,
      "step": 11014
    },
    {
      "epoch": 0.5054839153779083,
      "grad_norm": 0.4782871901988983,
      "learning_rate": 9.444156815534006e-06,
      "loss": 0.4407,
      "step": 11015
    },
    {
      "epoch": 0.5055298058831628,
      "grad_norm": 0.4952266216278076,
      "learning_rate": 9.444044456645758e-06,
      "loss": 0.4181,
      "step": 11016
    },
    {
      "epoch": 0.5055756963884173,
      "grad_norm": 0.41617321968078613,
      "learning_rate": 9.443932087070988e-06,
      "loss": 0.3305,
      "step": 11017
    },
    {
      "epoch": 0.5056215868936716,
      "grad_norm": 0.4508669376373291,
      "learning_rate": 9.443819706809961e-06,
      "loss": 0.3839,
      "step": 11018
    },
    {
      "epoch": 0.5056674773989261,
      "grad_norm": 0.4599497616291046,
      "learning_rate": 9.443707315862952e-06,
      "loss": 0.3817,
      "step": 11019
    },
    {
      "epoch": 0.5057133679041806,
      "grad_norm": 1.0045057535171509,
      "learning_rate": 9.443594914230228e-06,
      "loss": 0.4503,
      "step": 11020
    },
    {
      "epoch": 0.5057592584094351,
      "grad_norm": 0.4739968478679657,
      "learning_rate": 9.44348250191206e-06,
      "loss": 0.4313,
      "step": 11021
    },
    {
      "epoch": 0.5058051489146895,
      "grad_norm": 0.47547319531440735,
      "learning_rate": 9.443370078908722e-06,
      "loss": 0.3381,
      "step": 11022
    },
    {
      "epoch": 0.505851039419944,
      "grad_norm": 0.4720299541950226,
      "learning_rate": 9.443257645220479e-06,
      "loss": 0.3905,
      "step": 11023
    },
    {
      "epoch": 0.5058969299251985,
      "grad_norm": 0.44392523169517517,
      "learning_rate": 9.443145200847605e-06,
      "loss": 0.3638,
      "step": 11024
    },
    {
      "epoch": 0.5059428204304529,
      "grad_norm": 0.5037426948547363,
      "learning_rate": 9.443032745790369e-06,
      "loss": 0.4775,
      "step": 11025
    },
    {
      "epoch": 0.5059887109357074,
      "grad_norm": 0.49098989367485046,
      "learning_rate": 9.44292028004904e-06,
      "loss": 0.4033,
      "step": 11026
    },
    {
      "epoch": 0.5060346014409619,
      "grad_norm": 0.46203309297561646,
      "learning_rate": 9.44280780362389e-06,
      "loss": 0.4478,
      "step": 11027
    },
    {
      "epoch": 0.5060804919462163,
      "grad_norm": 0.4806912839412689,
      "learning_rate": 9.44269531651519e-06,
      "loss": 0.4354,
      "step": 11028
    },
    {
      "epoch": 0.5061263824514708,
      "grad_norm": 0.4335838556289673,
      "learning_rate": 9.442582818723211e-06,
      "loss": 0.3283,
      "step": 11029
    },
    {
      "epoch": 0.5061722729567253,
      "grad_norm": 0.46188610792160034,
      "learning_rate": 9.442470310248222e-06,
      "loss": 0.4047,
      "step": 11030
    },
    {
      "epoch": 0.5062181634619797,
      "grad_norm": 0.42918816208839417,
      "learning_rate": 9.442357791090494e-06,
      "loss": 0.3459,
      "step": 11031
    },
    {
      "epoch": 0.5062640539672342,
      "grad_norm": 0.5138610601425171,
      "learning_rate": 9.442245261250298e-06,
      "loss": 0.5251,
      "step": 11032
    },
    {
      "epoch": 0.5063099444724887,
      "grad_norm": 0.5044010877609253,
      "learning_rate": 9.442132720727904e-06,
      "loss": 0.477,
      "step": 11033
    },
    {
      "epoch": 0.5063558349777431,
      "grad_norm": 0.556893527507782,
      "learning_rate": 9.44202016952358e-06,
      "loss": 0.5587,
      "step": 11034
    },
    {
      "epoch": 0.5064017254829976,
      "grad_norm": 0.5577366948127747,
      "learning_rate": 9.441907607637602e-06,
      "loss": 0.5195,
      "step": 11035
    },
    {
      "epoch": 0.506447615988252,
      "grad_norm": 0.4409179389476776,
      "learning_rate": 9.441795035070238e-06,
      "loss": 0.3331,
      "step": 11036
    },
    {
      "epoch": 0.5064935064935064,
      "grad_norm": 0.49706846475601196,
      "learning_rate": 9.441682451821759e-06,
      "loss": 0.4677,
      "step": 11037
    },
    {
      "epoch": 0.5065393969987609,
      "grad_norm": 0.4645450711250305,
      "learning_rate": 9.441569857892435e-06,
      "loss": 0.4131,
      "step": 11038
    },
    {
      "epoch": 0.5065852875040154,
      "grad_norm": 0.501076340675354,
      "learning_rate": 9.441457253282538e-06,
      "loss": 0.4945,
      "step": 11039
    },
    {
      "epoch": 0.5066311780092699,
      "grad_norm": 0.5052675008773804,
      "learning_rate": 9.441344637992337e-06,
      "loss": 0.4626,
      "step": 11040
    },
    {
      "epoch": 0.5066770685145243,
      "grad_norm": 0.4847705662250519,
      "learning_rate": 9.441232012022104e-06,
      "loss": 0.4113,
      "step": 11041
    },
    {
      "epoch": 0.5067229590197788,
      "grad_norm": 0.4488966464996338,
      "learning_rate": 9.441119375372109e-06,
      "loss": 0.3954,
      "step": 11042
    },
    {
      "epoch": 0.5067688495250333,
      "grad_norm": 0.45276081562042236,
      "learning_rate": 9.441006728042626e-06,
      "loss": 0.3949,
      "step": 11043
    },
    {
      "epoch": 0.5068147400302877,
      "grad_norm": 0.43426671624183655,
      "learning_rate": 9.44089407003392e-06,
      "loss": 0.3778,
      "step": 11044
    },
    {
      "epoch": 0.5068606305355422,
      "grad_norm": 0.5115455985069275,
      "learning_rate": 9.440781401346268e-06,
      "loss": 0.4712,
      "step": 11045
    },
    {
      "epoch": 0.5069065210407967,
      "grad_norm": 0.45821547508239746,
      "learning_rate": 9.440668721979939e-06,
      "loss": 0.4115,
      "step": 11046
    },
    {
      "epoch": 0.5069524115460511,
      "grad_norm": 0.5105917453765869,
      "learning_rate": 9.440556031935202e-06,
      "loss": 0.4245,
      "step": 11047
    },
    {
      "epoch": 0.5069983020513056,
      "grad_norm": 0.4881758987903595,
      "learning_rate": 9.440443331212328e-06,
      "loss": 0.4939,
      "step": 11048
    },
    {
      "epoch": 0.5070441925565601,
      "grad_norm": 0.4476834535598755,
      "learning_rate": 9.44033061981159e-06,
      "loss": 0.4147,
      "step": 11049
    },
    {
      "epoch": 0.5070900830618145,
      "grad_norm": 0.4782133102416992,
      "learning_rate": 9.44021789773326e-06,
      "loss": 0.4147,
      "step": 11050
    },
    {
      "epoch": 0.507135973567069,
      "grad_norm": 0.4676351845264435,
      "learning_rate": 9.440105164977605e-06,
      "loss": 0.4056,
      "step": 11051
    },
    {
      "epoch": 0.5071818640723235,
      "grad_norm": 0.44782114028930664,
      "learning_rate": 9.439992421544899e-06,
      "loss": 0.3467,
      "step": 11052
    },
    {
      "epoch": 0.5072277545775778,
      "grad_norm": 0.4424588978290558,
      "learning_rate": 9.439879667435415e-06,
      "loss": 0.4033,
      "step": 11053
    },
    {
      "epoch": 0.5072736450828323,
      "grad_norm": 0.48231878876686096,
      "learning_rate": 9.43976690264942e-06,
      "loss": 0.4728,
      "step": 11054
    },
    {
      "epoch": 0.5073195355880868,
      "grad_norm": 0.4335916340351105,
      "learning_rate": 9.439654127187184e-06,
      "loss": 0.4288,
      "step": 11055
    },
    {
      "epoch": 0.5073654260933413,
      "grad_norm": 0.4402496814727783,
      "learning_rate": 9.439541341048984e-06,
      "loss": 0.3749,
      "step": 11056
    },
    {
      "epoch": 0.5074113165985957,
      "grad_norm": 0.45048797130584717,
      "learning_rate": 9.439428544235087e-06,
      "loss": 0.3642,
      "step": 11057
    },
    {
      "epoch": 0.5074572071038502,
      "grad_norm": 0.518036425113678,
      "learning_rate": 9.439315736745767e-06,
      "loss": 0.4899,
      "step": 11058
    },
    {
      "epoch": 0.5075030976091047,
      "grad_norm": 0.4576689600944519,
      "learning_rate": 9.439202918581293e-06,
      "loss": 0.4694,
      "step": 11059
    },
    {
      "epoch": 0.5075489881143591,
      "grad_norm": 0.45883145928382874,
      "learning_rate": 9.439090089741936e-06,
      "loss": 0.4037,
      "step": 11060
    },
    {
      "epoch": 0.5075948786196136,
      "grad_norm": 0.4919106662273407,
      "learning_rate": 9.43897725022797e-06,
      "loss": 0.5226,
      "step": 11061
    },
    {
      "epoch": 0.5076407691248681,
      "grad_norm": 0.47049248218536377,
      "learning_rate": 9.438864400039663e-06,
      "loss": 0.423,
      "step": 11062
    },
    {
      "epoch": 0.5076866596301225,
      "grad_norm": 0.43477338552474976,
      "learning_rate": 9.438751539177289e-06,
      "loss": 0.3594,
      "step": 11063
    },
    {
      "epoch": 0.507732550135377,
      "grad_norm": 0.4439249634742737,
      "learning_rate": 9.438638667641119e-06,
      "loss": 0.3314,
      "step": 11064
    },
    {
      "epoch": 0.5077784406406315,
      "grad_norm": 0.4799375832080841,
      "learning_rate": 9.438525785431422e-06,
      "loss": 0.4011,
      "step": 11065
    },
    {
      "epoch": 0.5078243311458859,
      "grad_norm": 0.4653036296367645,
      "learning_rate": 9.438412892548475e-06,
      "loss": 0.3585,
      "step": 11066
    },
    {
      "epoch": 0.5078702216511404,
      "grad_norm": 0.4222617447376251,
      "learning_rate": 9.438299988992543e-06,
      "loss": 0.3557,
      "step": 11067
    },
    {
      "epoch": 0.5079161121563949,
      "grad_norm": 0.48153528571128845,
      "learning_rate": 9.4381870747639e-06,
      "loss": 0.4467,
      "step": 11068
    },
    {
      "epoch": 0.5079620026616493,
      "grad_norm": 0.48363277316093445,
      "learning_rate": 9.43807414986282e-06,
      "loss": 0.4085,
      "step": 11069
    },
    {
      "epoch": 0.5080078931669038,
      "grad_norm": 0.441182941198349,
      "learning_rate": 9.437961214289571e-06,
      "loss": 0.3777,
      "step": 11070
    },
    {
      "epoch": 0.5080537836721583,
      "grad_norm": 0.506866455078125,
      "learning_rate": 9.437848268044424e-06,
      "loss": 0.4978,
      "step": 11071
    },
    {
      "epoch": 0.5080996741774126,
      "grad_norm": 0.4856387972831726,
      "learning_rate": 9.437735311127655e-06,
      "loss": 0.4605,
      "step": 11072
    },
    {
      "epoch": 0.5081455646826671,
      "grad_norm": 0.44156569242477417,
      "learning_rate": 9.437622343539532e-06,
      "loss": 0.3731,
      "step": 11073
    },
    {
      "epoch": 0.5081914551879216,
      "grad_norm": 0.4477945566177368,
      "learning_rate": 9.437509365280328e-06,
      "loss": 0.4327,
      "step": 11074
    },
    {
      "epoch": 0.5082373456931761,
      "grad_norm": 0.4663015305995941,
      "learning_rate": 9.437396376350316e-06,
      "loss": 0.4586,
      "step": 11075
    },
    {
      "epoch": 0.5082832361984305,
      "grad_norm": 0.43222957849502563,
      "learning_rate": 9.437283376749766e-06,
      "loss": 0.3475,
      "step": 11076
    },
    {
      "epoch": 0.508329126703685,
      "grad_norm": 0.4556671977043152,
      "learning_rate": 9.437170366478948e-06,
      "loss": 0.4169,
      "step": 11077
    },
    {
      "epoch": 0.5083750172089395,
      "grad_norm": 0.49400678277015686,
      "learning_rate": 9.437057345538137e-06,
      "loss": 0.4747,
      "step": 11078
    },
    {
      "epoch": 0.5084209077141939,
      "grad_norm": 0.4953756332397461,
      "learning_rate": 9.436944313927603e-06,
      "loss": 0.4709,
      "step": 11079
    },
    {
      "epoch": 0.5084667982194484,
      "grad_norm": 0.4935978353023529,
      "learning_rate": 9.436831271647619e-06,
      "loss": 0.4369,
      "step": 11080
    },
    {
      "epoch": 0.5085126887247029,
      "grad_norm": 0.43096497654914856,
      "learning_rate": 9.436718218698455e-06,
      "loss": 0.3543,
      "step": 11081
    },
    {
      "epoch": 0.5085585792299573,
      "grad_norm": 0.5037040114402771,
      "learning_rate": 9.436605155080383e-06,
      "loss": 0.4801,
      "step": 11082
    },
    {
      "epoch": 0.5086044697352118,
      "grad_norm": 0.47201764583587646,
      "learning_rate": 9.436492080793677e-06,
      "loss": 0.4253,
      "step": 11083
    },
    {
      "epoch": 0.5086503602404663,
      "grad_norm": 0.4751473069190979,
      "learning_rate": 9.43637899583861e-06,
      "loss": 0.4651,
      "step": 11084
    },
    {
      "epoch": 0.5086962507457207,
      "grad_norm": 0.52275550365448,
      "learning_rate": 9.436265900215448e-06,
      "loss": 0.5488,
      "step": 11085
    },
    {
      "epoch": 0.5087421412509752,
      "grad_norm": 0.4285326600074768,
      "learning_rate": 9.43615279392447e-06,
      "loss": 0.3608,
      "step": 11086
    },
    {
      "epoch": 0.5087880317562297,
      "grad_norm": 0.43743857741355896,
      "learning_rate": 9.436039676965942e-06,
      "loss": 0.347,
      "step": 11087
    },
    {
      "epoch": 0.508833922261484,
      "grad_norm": 0.4144723415374756,
      "learning_rate": 9.43592654934014e-06,
      "loss": 0.313,
      "step": 11088
    },
    {
      "epoch": 0.5088798127667385,
      "grad_norm": 0.43682458996772766,
      "learning_rate": 9.435813411047335e-06,
      "loss": 0.35,
      "step": 11089
    },
    {
      "epoch": 0.508925703271993,
      "grad_norm": 0.48336261510849,
      "learning_rate": 9.435700262087796e-06,
      "loss": 0.445,
      "step": 11090
    },
    {
      "epoch": 0.5089715937772474,
      "grad_norm": 0.45538023114204407,
      "learning_rate": 9.4355871024618e-06,
      "loss": 0.3784,
      "step": 11091
    },
    {
      "epoch": 0.5090174842825019,
      "grad_norm": 0.4261086881160736,
      "learning_rate": 9.435473932169619e-06,
      "loss": 0.3645,
      "step": 11092
    },
    {
      "epoch": 0.5090633747877564,
      "grad_norm": 0.47664251923561096,
      "learning_rate": 9.43536075121152e-06,
      "loss": 0.4192,
      "step": 11093
    },
    {
      "epoch": 0.5091092652930109,
      "grad_norm": 0.458617627620697,
      "learning_rate": 9.43524755958778e-06,
      "loss": 0.3755,
      "step": 11094
    },
    {
      "epoch": 0.5091551557982653,
      "grad_norm": 0.45727846026420593,
      "learning_rate": 9.435134357298667e-06,
      "loss": 0.4144,
      "step": 11095
    },
    {
      "epoch": 0.5092010463035198,
      "grad_norm": 0.416154682636261,
      "learning_rate": 9.435021144344458e-06,
      "loss": 0.3371,
      "step": 11096
    },
    {
      "epoch": 0.5092469368087743,
      "grad_norm": 0.4496779441833496,
      "learning_rate": 9.434907920725421e-06,
      "loss": 0.3086,
      "step": 11097
    },
    {
      "epoch": 0.5092928273140287,
      "grad_norm": 0.4601019322872162,
      "learning_rate": 9.434794686441831e-06,
      "loss": 0.4058,
      "step": 11098
    },
    {
      "epoch": 0.5093387178192832,
      "grad_norm": 0.4619458317756653,
      "learning_rate": 9.43468144149396e-06,
      "loss": 0.3592,
      "step": 11099
    },
    {
      "epoch": 0.5093846083245377,
      "grad_norm": 0.4989260733127594,
      "learning_rate": 9.43456818588208e-06,
      "loss": 0.4978,
      "step": 11100
    },
    {
      "epoch": 0.5094304988297921,
      "grad_norm": 0.4481591284275055,
      "learning_rate": 9.434454919606461e-06,
      "loss": 0.3988,
      "step": 11101
    },
    {
      "epoch": 0.5094763893350466,
      "grad_norm": 0.4668935537338257,
      "learning_rate": 9.43434164266738e-06,
      "loss": 0.4377,
      "step": 11102
    },
    {
      "epoch": 0.5095222798403011,
      "grad_norm": 0.4743530750274658,
      "learning_rate": 9.434228355065105e-06,
      "loss": 0.4418,
      "step": 11103
    },
    {
      "epoch": 0.5095681703455555,
      "grad_norm": 0.4634436368942261,
      "learning_rate": 9.434115056799911e-06,
      "loss": 0.3932,
      "step": 11104
    },
    {
      "epoch": 0.50961406085081,
      "grad_norm": 0.4736510217189789,
      "learning_rate": 9.43400174787207e-06,
      "loss": 0.4718,
      "step": 11105
    },
    {
      "epoch": 0.5096599513560645,
      "grad_norm": 0.45432233810424805,
      "learning_rate": 9.433888428281854e-06,
      "loss": 0.3718,
      "step": 11106
    },
    {
      "epoch": 0.5097058418613188,
      "grad_norm": 0.43205001950263977,
      "learning_rate": 9.433775098029536e-06,
      "loss": 0.4024,
      "step": 11107
    },
    {
      "epoch": 0.5097517323665733,
      "grad_norm": 0.4442681670188904,
      "learning_rate": 9.433661757115388e-06,
      "loss": 0.3669,
      "step": 11108
    },
    {
      "epoch": 0.5097976228718278,
      "grad_norm": 0.6076908707618713,
      "learning_rate": 9.433548405539683e-06,
      "loss": 0.4564,
      "step": 11109
    },
    {
      "epoch": 0.5098435133770823,
      "grad_norm": 0.4501707851886749,
      "learning_rate": 9.433435043302694e-06,
      "loss": 0.3678,
      "step": 11110
    },
    {
      "epoch": 0.5098894038823367,
      "grad_norm": 0.4663463234901428,
      "learning_rate": 9.433321670404691e-06,
      "loss": 0.4278,
      "step": 11111
    },
    {
      "epoch": 0.5099352943875912,
      "grad_norm": 0.4603365957736969,
      "learning_rate": 9.433208286845949e-06,
      "loss": 0.4038,
      "step": 11112
    },
    {
      "epoch": 0.5099811848928457,
      "grad_norm": 0.4779384434223175,
      "learning_rate": 9.433094892626742e-06,
      "loss": 0.4395,
      "step": 11113
    },
    {
      "epoch": 0.5100270753981001,
      "grad_norm": 0.47363585233688354,
      "learning_rate": 9.432981487747338e-06,
      "loss": 0.4425,
      "step": 11114
    },
    {
      "epoch": 0.5100729659033546,
      "grad_norm": 0.4191666543483734,
      "learning_rate": 9.432868072208015e-06,
      "loss": 0.3398,
      "step": 11115
    },
    {
      "epoch": 0.5101188564086091,
      "grad_norm": 0.4613591432571411,
      "learning_rate": 9.432754646009043e-06,
      "loss": 0.3888,
      "step": 11116
    },
    {
      "epoch": 0.5101647469138635,
      "grad_norm": 0.4800584018230438,
      "learning_rate": 9.432641209150694e-06,
      "loss": 0.4865,
      "step": 11117
    },
    {
      "epoch": 0.510210637419118,
      "grad_norm": 0.4228001534938812,
      "learning_rate": 9.432527761633243e-06,
      "loss": 0.3847,
      "step": 11118
    },
    {
      "epoch": 0.5102565279243725,
      "grad_norm": 0.5002844333648682,
      "learning_rate": 9.432414303456963e-06,
      "loss": 0.4685,
      "step": 11119
    },
    {
      "epoch": 0.5103024184296269,
      "grad_norm": 0.4854716658592224,
      "learning_rate": 9.432300834622123e-06,
      "loss": 0.4367,
      "step": 11120
    },
    {
      "epoch": 0.5103483089348814,
      "grad_norm": 0.48742276430130005,
      "learning_rate": 9.432187355128999e-06,
      "loss": 0.469,
      "step": 11121
    },
    {
      "epoch": 0.5103941994401359,
      "grad_norm": 0.4191174805164337,
      "learning_rate": 9.432073864977864e-06,
      "loss": 0.3568,
      "step": 11122
    },
    {
      "epoch": 0.5104400899453903,
      "grad_norm": 0.476814329624176,
      "learning_rate": 9.431960364168991e-06,
      "loss": 0.4237,
      "step": 11123
    },
    {
      "epoch": 0.5104859804506447,
      "grad_norm": 0.41378504037857056,
      "learning_rate": 9.431846852702648e-06,
      "loss": 0.3246,
      "step": 11124
    },
    {
      "epoch": 0.5105318709558992,
      "grad_norm": 0.49643921852111816,
      "learning_rate": 9.431733330579116e-06,
      "loss": 0.4809,
      "step": 11125
    },
    {
      "epoch": 0.5105777614611536,
      "grad_norm": 0.4422619938850403,
      "learning_rate": 9.431619797798664e-06,
      "loss": 0.3348,
      "step": 11126
    },
    {
      "epoch": 0.5106236519664081,
      "grad_norm": 0.486173540353775,
      "learning_rate": 9.431506254361562e-06,
      "loss": 0.3969,
      "step": 11127
    },
    {
      "epoch": 0.5106695424716626,
      "grad_norm": 0.49294888973236084,
      "learning_rate": 9.43139270026809e-06,
      "loss": 0.4826,
      "step": 11128
    },
    {
      "epoch": 0.5107154329769171,
      "grad_norm": 0.40683048963546753,
      "learning_rate": 9.431279135518514e-06,
      "loss": 0.3064,
      "step": 11129
    },
    {
      "epoch": 0.5107613234821715,
      "grad_norm": 0.44529032707214355,
      "learning_rate": 9.431165560113111e-06,
      "loss": 0.3702,
      "step": 11130
    },
    {
      "epoch": 0.510807213987426,
      "grad_norm": 0.44968166947364807,
      "learning_rate": 9.431051974052154e-06,
      "loss": 0.3575,
      "step": 11131
    },
    {
      "epoch": 0.5108531044926805,
      "grad_norm": 0.46065571904182434,
      "learning_rate": 9.430938377335916e-06,
      "loss": 0.3516,
      "step": 11132
    },
    {
      "epoch": 0.5108989949979349,
      "grad_norm": 0.4441080093383789,
      "learning_rate": 9.430824769964669e-06,
      "loss": 0.3351,
      "step": 11133
    },
    {
      "epoch": 0.5109448855031894,
      "grad_norm": 0.44278252124786377,
      "learning_rate": 9.430711151938686e-06,
      "loss": 0.4041,
      "step": 11134
    },
    {
      "epoch": 0.5109907760084439,
      "grad_norm": 0.48169058561325073,
      "learning_rate": 9.43059752325824e-06,
      "loss": 0.424,
      "step": 11135
    },
    {
      "epoch": 0.5110366665136983,
      "grad_norm": 0.4215027391910553,
      "learning_rate": 9.430483883923608e-06,
      "loss": 0.3204,
      "step": 11136
    },
    {
      "epoch": 0.5110825570189528,
      "grad_norm": 0.458992600440979,
      "learning_rate": 9.43037023393506e-06,
      "loss": 0.4212,
      "step": 11137
    },
    {
      "epoch": 0.5111284475242073,
      "grad_norm": 0.4420967102050781,
      "learning_rate": 9.43025657329287e-06,
      "loss": 0.3699,
      "step": 11138
    },
    {
      "epoch": 0.5111743380294617,
      "grad_norm": 0.48791202902793884,
      "learning_rate": 9.43014290199731e-06,
      "loss": 0.4388,
      "step": 11139
    },
    {
      "epoch": 0.5112202285347162,
      "grad_norm": 0.4718987047672272,
      "learning_rate": 9.430029220048655e-06,
      "loss": 0.3936,
      "step": 11140
    },
    {
      "epoch": 0.5112661190399707,
      "grad_norm": 0.4414863884449005,
      "learning_rate": 9.429915527447178e-06,
      "loss": 0.3878,
      "step": 11141
    },
    {
      "epoch": 0.511312009545225,
      "grad_norm": 0.4793201684951782,
      "learning_rate": 9.429801824193153e-06,
      "loss": 0.4288,
      "step": 11142
    },
    {
      "epoch": 0.5113579000504795,
      "grad_norm": 0.48404479026794434,
      "learning_rate": 9.42968811028685e-06,
      "loss": 0.483,
      "step": 11143
    },
    {
      "epoch": 0.511403790555734,
      "grad_norm": 0.4797365665435791,
      "learning_rate": 9.429574385728547e-06,
      "loss": 0.4312,
      "step": 11144
    },
    {
      "epoch": 0.5114496810609885,
      "grad_norm": 0.45008593797683716,
      "learning_rate": 9.429460650518516e-06,
      "loss": 0.3769,
      "step": 11145
    },
    {
      "epoch": 0.5114955715662429,
      "grad_norm": 0.44718408584594727,
      "learning_rate": 9.429346904657031e-06,
      "loss": 0.392,
      "step": 11146
    },
    {
      "epoch": 0.5115414620714974,
      "grad_norm": 0.4396439790725708,
      "learning_rate": 9.429233148144363e-06,
      "loss": 0.3595,
      "step": 11147
    },
    {
      "epoch": 0.5115873525767519,
      "grad_norm": 0.4328153431415558,
      "learning_rate": 9.429119380980788e-06,
      "loss": 0.3461,
      "step": 11148
    },
    {
      "epoch": 0.5116332430820063,
      "grad_norm": 0.48482391238212585,
      "learning_rate": 9.429005603166577e-06,
      "loss": 0.4111,
      "step": 11149
    },
    {
      "epoch": 0.5116791335872608,
      "grad_norm": 0.41375404596328735,
      "learning_rate": 9.428891814702009e-06,
      "loss": 0.3204,
      "step": 11150
    },
    {
      "epoch": 0.5117250240925153,
      "grad_norm": 0.47852614521980286,
      "learning_rate": 9.42877801558735e-06,
      "loss": 0.4461,
      "step": 11151
    },
    {
      "epoch": 0.5117709145977697,
      "grad_norm": 0.4818577170372009,
      "learning_rate": 9.428664205822879e-06,
      "loss": 0.4554,
      "step": 11152
    },
    {
      "epoch": 0.5118168051030242,
      "grad_norm": 0.47473928332328796,
      "learning_rate": 9.42855038540887e-06,
      "loss": 0.4177,
      "step": 11153
    },
    {
      "epoch": 0.5118626956082787,
      "grad_norm": 0.4590417146682739,
      "learning_rate": 9.428436554345592e-06,
      "loss": 0.4494,
      "step": 11154
    },
    {
      "epoch": 0.5119085861135331,
      "grad_norm": 0.4606229364871979,
      "learning_rate": 9.428322712633324e-06,
      "loss": 0.4102,
      "step": 11155
    },
    {
      "epoch": 0.5119544766187876,
      "grad_norm": 0.4564058184623718,
      "learning_rate": 9.428208860272337e-06,
      "loss": 0.398,
      "step": 11156
    },
    {
      "epoch": 0.5120003671240421,
      "grad_norm": 0.46197837591171265,
      "learning_rate": 9.428094997262906e-06,
      "loss": 0.4134,
      "step": 11157
    },
    {
      "epoch": 0.5120462576292965,
      "grad_norm": 0.4781615734100342,
      "learning_rate": 9.427981123605304e-06,
      "loss": 0.4224,
      "step": 11158
    },
    {
      "epoch": 0.512092148134551,
      "grad_norm": 0.4690605700016022,
      "learning_rate": 9.427867239299804e-06,
      "loss": 0.4236,
      "step": 11159
    },
    {
      "epoch": 0.5121380386398054,
      "grad_norm": 0.49025338888168335,
      "learning_rate": 9.42775334434668e-06,
      "loss": 0.4675,
      "step": 11160
    },
    {
      "epoch": 0.5121839291450598,
      "grad_norm": 0.45674118399620056,
      "learning_rate": 9.427639438746208e-06,
      "loss": 0.384,
      "step": 11161
    },
    {
      "epoch": 0.5122298196503143,
      "grad_norm": 0.4760828912258148,
      "learning_rate": 9.42752552249866e-06,
      "loss": 0.3817,
      "step": 11162
    },
    {
      "epoch": 0.5122757101555688,
      "grad_norm": 0.4247443377971649,
      "learning_rate": 9.42741159560431e-06,
      "loss": 0.3024,
      "step": 11163
    },
    {
      "epoch": 0.5123216006608233,
      "grad_norm": 0.46078816056251526,
      "learning_rate": 9.427297658063433e-06,
      "loss": 0.4108,
      "step": 11164
    },
    {
      "epoch": 0.5123674911660777,
      "grad_norm": 0.48824793100357056,
      "learning_rate": 9.427183709876302e-06,
      "loss": 0.4855,
      "step": 11165
    },
    {
      "epoch": 0.5124133816713322,
      "grad_norm": 0.40844792127609253,
      "learning_rate": 9.427069751043193e-06,
      "loss": 0.335,
      "step": 11166
    },
    {
      "epoch": 0.5124592721765867,
      "grad_norm": 0.4416067600250244,
      "learning_rate": 9.426955781564378e-06,
      "loss": 0.3825,
      "step": 11167
    },
    {
      "epoch": 0.5125051626818411,
      "grad_norm": 0.4281124770641327,
      "learning_rate": 9.42684180144013e-06,
      "loss": 0.3098,
      "step": 11168
    },
    {
      "epoch": 0.5125510531870956,
      "grad_norm": 0.47789958119392395,
      "learning_rate": 9.426727810670725e-06,
      "loss": 0.4523,
      "step": 11169
    },
    {
      "epoch": 0.5125969436923501,
      "grad_norm": 0.465964138507843,
      "learning_rate": 9.42661380925644e-06,
      "loss": 0.4595,
      "step": 11170
    },
    {
      "epoch": 0.5126428341976045,
      "grad_norm": 0.4484297037124634,
      "learning_rate": 9.426499797197541e-06,
      "loss": 0.4043,
      "step": 11171
    },
    {
      "epoch": 0.512688724702859,
      "grad_norm": 0.4079855978488922,
      "learning_rate": 9.426385774494308e-06,
      "loss": 0.3055,
      "step": 11172
    },
    {
      "epoch": 0.5127346152081135,
      "grad_norm": 0.4222286343574524,
      "learning_rate": 9.426271741147018e-06,
      "loss": 0.3721,
      "step": 11173
    },
    {
      "epoch": 0.5127805057133679,
      "grad_norm": 0.5693210959434509,
      "learning_rate": 9.426157697155938e-06,
      "loss": 0.3347,
      "step": 11174
    },
    {
      "epoch": 0.5128263962186224,
      "grad_norm": 0.438080757856369,
      "learning_rate": 9.426043642521347e-06,
      "loss": 0.3475,
      "step": 11175
    },
    {
      "epoch": 0.5128722867238769,
      "grad_norm": 0.4451926648616791,
      "learning_rate": 9.425929577243516e-06,
      "loss": 0.3451,
      "step": 11176
    },
    {
      "epoch": 0.5129181772291312,
      "grad_norm": 0.4349431097507477,
      "learning_rate": 9.425815501322724e-06,
      "loss": 0.3319,
      "step": 11177
    },
    {
      "epoch": 0.5129640677343857,
      "grad_norm": 0.43615883588790894,
      "learning_rate": 9.42570141475924e-06,
      "loss": 0.3544,
      "step": 11178
    },
    {
      "epoch": 0.5130099582396402,
      "grad_norm": 0.45016226172447205,
      "learning_rate": 9.425587317553343e-06,
      "loss": 0.3658,
      "step": 11179
    },
    {
      "epoch": 0.5130558487448946,
      "grad_norm": 0.4785882830619812,
      "learning_rate": 9.425473209705304e-06,
      "loss": 0.443,
      "step": 11180
    },
    {
      "epoch": 0.5131017392501491,
      "grad_norm": 0.460958868265152,
      "learning_rate": 9.4253590912154e-06,
      "loss": 0.4202,
      "step": 11181
    },
    {
      "epoch": 0.5131476297554036,
      "grad_norm": 0.44740164279937744,
      "learning_rate": 9.425244962083902e-06,
      "loss": 0.4001,
      "step": 11182
    },
    {
      "epoch": 0.5131935202606581,
      "grad_norm": 0.41863784193992615,
      "learning_rate": 9.425130822311089e-06,
      "loss": 0.3213,
      "step": 11183
    },
    {
      "epoch": 0.5132394107659125,
      "grad_norm": 0.4627070426940918,
      "learning_rate": 9.425016671897231e-06,
      "loss": 0.3605,
      "step": 11184
    },
    {
      "epoch": 0.513285301271167,
      "grad_norm": 0.45955419540405273,
      "learning_rate": 9.424902510842605e-06,
      "loss": 0.388,
      "step": 11185
    },
    {
      "epoch": 0.5133311917764215,
      "grad_norm": 0.47954294085502625,
      "learning_rate": 9.424788339147485e-06,
      "loss": 0.4674,
      "step": 11186
    },
    {
      "epoch": 0.5133770822816759,
      "grad_norm": 0.4473467469215393,
      "learning_rate": 9.424674156812147e-06,
      "loss": 0.4135,
      "step": 11187
    },
    {
      "epoch": 0.5134229727869304,
      "grad_norm": 0.46484407782554626,
      "learning_rate": 9.424559963836862e-06,
      "loss": 0.4595,
      "step": 11188
    },
    {
      "epoch": 0.5134688632921849,
      "grad_norm": 0.49906080961227417,
      "learning_rate": 9.424445760221908e-06,
      "loss": 0.4543,
      "step": 11189
    },
    {
      "epoch": 0.5135147537974393,
      "grad_norm": 0.5169631242752075,
      "learning_rate": 9.424331545967557e-06,
      "loss": 0.5245,
      "step": 11190
    },
    {
      "epoch": 0.5135606443026938,
      "grad_norm": 0.4689655005931854,
      "learning_rate": 9.424217321074085e-06,
      "loss": 0.4127,
      "step": 11191
    },
    {
      "epoch": 0.5136065348079483,
      "grad_norm": 0.44165536761283875,
      "learning_rate": 9.424103085541768e-06,
      "loss": 0.3201,
      "step": 11192
    },
    {
      "epoch": 0.5136524253132027,
      "grad_norm": 0.44161972403526306,
      "learning_rate": 9.423988839370879e-06,
      "loss": 0.3759,
      "step": 11193
    },
    {
      "epoch": 0.5136983158184572,
      "grad_norm": 0.42835134267807007,
      "learning_rate": 9.423874582561692e-06,
      "loss": 0.3324,
      "step": 11194
    },
    {
      "epoch": 0.5137442063237116,
      "grad_norm": 0.4730234444141388,
      "learning_rate": 9.423760315114485e-06,
      "loss": 0.41,
      "step": 11195
    },
    {
      "epoch": 0.513790096828966,
      "grad_norm": 0.43346846103668213,
      "learning_rate": 9.423646037029527e-06,
      "loss": 0.3518,
      "step": 11196
    },
    {
      "epoch": 0.5138359873342205,
      "grad_norm": 0.4523134231567383,
      "learning_rate": 9.4235317483071e-06,
      "loss": 0.3724,
      "step": 11197
    },
    {
      "epoch": 0.513881877839475,
      "grad_norm": 0.45653825998306274,
      "learning_rate": 9.423417448947476e-06,
      "loss": 0.4106,
      "step": 11198
    },
    {
      "epoch": 0.5139277683447295,
      "grad_norm": 0.5112180709838867,
      "learning_rate": 9.423303138950925e-06,
      "loss": 0.4898,
      "step": 11199
    },
    {
      "epoch": 0.5139736588499839,
      "grad_norm": 0.4864504337310791,
      "learning_rate": 9.42318881831773e-06,
      "loss": 0.443,
      "step": 11200
    },
    {
      "epoch": 0.5140195493552384,
      "grad_norm": 0.4419673681259155,
      "learning_rate": 9.42307448704816e-06,
      "loss": 0.3944,
      "step": 11201
    },
    {
      "epoch": 0.5140654398604929,
      "grad_norm": 0.47507017850875854,
      "learning_rate": 9.42296014514249e-06,
      "loss": 0.4619,
      "step": 11202
    },
    {
      "epoch": 0.5141113303657473,
      "grad_norm": 0.4882911741733551,
      "learning_rate": 9.422845792601e-06,
      "loss": 0.4918,
      "step": 11203
    },
    {
      "epoch": 0.5141572208710018,
      "grad_norm": 0.45332691073417664,
      "learning_rate": 9.42273142942396e-06,
      "loss": 0.3999,
      "step": 11204
    },
    {
      "epoch": 0.5142031113762563,
      "grad_norm": 0.4811588525772095,
      "learning_rate": 9.422617055611647e-06,
      "loss": 0.4413,
      "step": 11205
    },
    {
      "epoch": 0.5142490018815107,
      "grad_norm": 0.45333409309387207,
      "learning_rate": 9.422502671164337e-06,
      "loss": 0.3316,
      "step": 11206
    },
    {
      "epoch": 0.5142948923867652,
      "grad_norm": 0.4645090103149414,
      "learning_rate": 9.422388276082302e-06,
      "loss": 0.3916,
      "step": 11207
    },
    {
      "epoch": 0.5143407828920197,
      "grad_norm": 0.44360506534576416,
      "learning_rate": 9.42227387036582e-06,
      "loss": 0.3455,
      "step": 11208
    },
    {
      "epoch": 0.5143866733972741,
      "grad_norm": 0.4784619212150574,
      "learning_rate": 9.422159454015164e-06,
      "loss": 0.496,
      "step": 11209
    },
    {
      "epoch": 0.5144325639025286,
      "grad_norm": 0.4477009177207947,
      "learning_rate": 9.422045027030612e-06,
      "loss": 0.3971,
      "step": 11210
    },
    {
      "epoch": 0.5144784544077831,
      "grad_norm": 0.440959632396698,
      "learning_rate": 9.421930589412436e-06,
      "loss": 0.3597,
      "step": 11211
    },
    {
      "epoch": 0.5145243449130374,
      "grad_norm": 0.4549063742160797,
      "learning_rate": 9.421816141160913e-06,
      "loss": 0.3818,
      "step": 11212
    },
    {
      "epoch": 0.5145702354182919,
      "grad_norm": 0.5039348602294922,
      "learning_rate": 9.421701682276317e-06,
      "loss": 0.4979,
      "step": 11213
    },
    {
      "epoch": 0.5146161259235464,
      "grad_norm": 0.47268053889274597,
      "learning_rate": 9.421587212758924e-06,
      "loss": 0.3998,
      "step": 11214
    },
    {
      "epoch": 0.5146620164288008,
      "grad_norm": 0.45106935501098633,
      "learning_rate": 9.421472732609011e-06,
      "loss": 0.381,
      "step": 11215
    },
    {
      "epoch": 0.5147079069340553,
      "grad_norm": 0.44838130474090576,
      "learning_rate": 9.42135824182685e-06,
      "loss": 0.4023,
      "step": 11216
    },
    {
      "epoch": 0.5147537974393098,
      "grad_norm": 0.4636859595775604,
      "learning_rate": 9.421243740412719e-06,
      "loss": 0.4065,
      "step": 11217
    },
    {
      "epoch": 0.5147996879445643,
      "grad_norm": 0.4705791175365448,
      "learning_rate": 9.421129228366892e-06,
      "loss": 0.3972,
      "step": 11218
    },
    {
      "epoch": 0.5148455784498187,
      "grad_norm": 0.4705198109149933,
      "learning_rate": 9.421014705689644e-06,
      "loss": 0.3906,
      "step": 11219
    },
    {
      "epoch": 0.5148914689550732,
      "grad_norm": 0.4702683687210083,
      "learning_rate": 9.42090017238125e-06,
      "loss": 0.4814,
      "step": 11220
    },
    {
      "epoch": 0.5149373594603277,
      "grad_norm": 0.4810901880264282,
      "learning_rate": 9.420785628441987e-06,
      "loss": 0.5011,
      "step": 11221
    },
    {
      "epoch": 0.5149832499655821,
      "grad_norm": 0.4607618749141693,
      "learning_rate": 9.42067107387213e-06,
      "loss": 0.4343,
      "step": 11222
    },
    {
      "epoch": 0.5150291404708366,
      "grad_norm": 0.4946284890174866,
      "learning_rate": 9.420556508671954e-06,
      "loss": 0.4498,
      "step": 11223
    },
    {
      "epoch": 0.5150750309760911,
      "grad_norm": 0.46579962968826294,
      "learning_rate": 9.420441932841736e-06,
      "loss": 0.3796,
      "step": 11224
    },
    {
      "epoch": 0.5151209214813455,
      "grad_norm": 0.4398934245109558,
      "learning_rate": 9.42032734638175e-06,
      "loss": 0.3724,
      "step": 11225
    },
    {
      "epoch": 0.5151668119866,
      "grad_norm": 0.4489741921424866,
      "learning_rate": 9.42021274929227e-06,
      "loss": 0.4077,
      "step": 11226
    },
    {
      "epoch": 0.5152127024918545,
      "grad_norm": 0.44262582063674927,
      "learning_rate": 9.420098141573574e-06,
      "loss": 0.3703,
      "step": 11227
    },
    {
      "epoch": 0.5152585929971089,
      "grad_norm": 0.4338358938694,
      "learning_rate": 9.41998352322594e-06,
      "loss": 0.3489,
      "step": 11228
    },
    {
      "epoch": 0.5153044835023634,
      "grad_norm": 0.4774375557899475,
      "learning_rate": 9.419868894249637e-06,
      "loss": 0.4242,
      "step": 11229
    },
    {
      "epoch": 0.5153503740076179,
      "grad_norm": 0.5442906022071838,
      "learning_rate": 9.419754254644946e-06,
      "loss": 0.4632,
      "step": 11230
    },
    {
      "epoch": 0.5153962645128722,
      "grad_norm": 0.5080773234367371,
      "learning_rate": 9.41963960441214e-06,
      "loss": 0.458,
      "step": 11231
    },
    {
      "epoch": 0.5154421550181267,
      "grad_norm": 0.48662588000297546,
      "learning_rate": 9.419524943551497e-06,
      "loss": 0.4054,
      "step": 11232
    },
    {
      "epoch": 0.5154880455233812,
      "grad_norm": 0.4499012231826782,
      "learning_rate": 9.41941027206329e-06,
      "loss": 0.3873,
      "step": 11233
    },
    {
      "epoch": 0.5155339360286356,
      "grad_norm": 0.4747801721096039,
      "learning_rate": 9.419295589947797e-06,
      "loss": 0.4194,
      "step": 11234
    },
    {
      "epoch": 0.5155798265338901,
      "grad_norm": 0.4485040307044983,
      "learning_rate": 9.419180897205292e-06,
      "loss": 0.3695,
      "step": 11235
    },
    {
      "epoch": 0.5156257170391446,
      "grad_norm": 0.5051221251487732,
      "learning_rate": 9.419066193836051e-06,
      "loss": 0.5025,
      "step": 11236
    },
    {
      "epoch": 0.5156716075443991,
      "grad_norm": 0.45362600684165955,
      "learning_rate": 9.418951479840353e-06,
      "loss": 0.4341,
      "step": 11237
    },
    {
      "epoch": 0.5157174980496535,
      "grad_norm": 0.4738071858882904,
      "learning_rate": 9.41883675521847e-06,
      "loss": 0.3016,
      "step": 11238
    },
    {
      "epoch": 0.515763388554908,
      "grad_norm": 0.4678727388381958,
      "learning_rate": 9.418722019970679e-06,
      "loss": 0.3595,
      "step": 11239
    },
    {
      "epoch": 0.5158092790601625,
      "grad_norm": 0.45817965269088745,
      "learning_rate": 9.418607274097255e-06,
      "loss": 0.3979,
      "step": 11240
    },
    {
      "epoch": 0.5158551695654169,
      "grad_norm": 0.44150492548942566,
      "learning_rate": 9.418492517598476e-06,
      "loss": 0.3624,
      "step": 11241
    },
    {
      "epoch": 0.5159010600706714,
      "grad_norm": 0.42746803164482117,
      "learning_rate": 9.418377750474618e-06,
      "loss": 0.3503,
      "step": 11242
    },
    {
      "epoch": 0.5159469505759259,
      "grad_norm": 0.4773419499397278,
      "learning_rate": 9.418262972725956e-06,
      "loss": 0.4107,
      "step": 11243
    },
    {
      "epoch": 0.5159928410811803,
      "grad_norm": 0.4755682051181793,
      "learning_rate": 9.418148184352765e-06,
      "loss": 0.4418,
      "step": 11244
    },
    {
      "epoch": 0.5160387315864348,
      "grad_norm": 0.5008369088172913,
      "learning_rate": 9.418033385355322e-06,
      "loss": 0.4241,
      "step": 11245
    },
    {
      "epoch": 0.5160846220916893,
      "grad_norm": 0.4306611716747284,
      "learning_rate": 9.417918575733904e-06,
      "loss": 0.3425,
      "step": 11246
    },
    {
      "epoch": 0.5161305125969436,
      "grad_norm": 0.4905848205089569,
      "learning_rate": 9.417803755488785e-06,
      "loss": 0.4509,
      "step": 11247
    },
    {
      "epoch": 0.5161764031021981,
      "grad_norm": 0.4466972053050995,
      "learning_rate": 9.417688924620244e-06,
      "loss": 0.3305,
      "step": 11248
    },
    {
      "epoch": 0.5162222936074526,
      "grad_norm": 0.438510924577713,
      "learning_rate": 9.417574083128554e-06,
      "loss": 0.3923,
      "step": 11249
    },
    {
      "epoch": 0.516268184112707,
      "grad_norm": 0.4085960388183594,
      "learning_rate": 9.417459231013993e-06,
      "loss": 0.307,
      "step": 11250
    },
    {
      "epoch": 0.5163140746179615,
      "grad_norm": 0.5797719955444336,
      "learning_rate": 9.417344368276838e-06,
      "loss": 0.4863,
      "step": 11251
    },
    {
      "epoch": 0.516359965123216,
      "grad_norm": 0.4440360963344574,
      "learning_rate": 9.417229494917361e-06,
      "loss": 0.4086,
      "step": 11252
    },
    {
      "epoch": 0.5164058556284705,
      "grad_norm": 0.455028772354126,
      "learning_rate": 9.417114610935842e-06,
      "loss": 0.4288,
      "step": 11253
    },
    {
      "epoch": 0.5164517461337249,
      "grad_norm": 0.493502140045166,
      "learning_rate": 9.416999716332557e-06,
      "loss": 0.479,
      "step": 11254
    },
    {
      "epoch": 0.5164976366389794,
      "grad_norm": 0.43186095356941223,
      "learning_rate": 9.416884811107782e-06,
      "loss": 0.3474,
      "step": 11255
    },
    {
      "epoch": 0.5165435271442339,
      "grad_norm": 0.4689897894859314,
      "learning_rate": 9.416769895261793e-06,
      "loss": 0.4217,
      "step": 11256
    },
    {
      "epoch": 0.5165894176494883,
      "grad_norm": 0.4879116714000702,
      "learning_rate": 9.416654968794867e-06,
      "loss": 0.4469,
      "step": 11257
    },
    {
      "epoch": 0.5166353081547428,
      "grad_norm": 0.5197396278381348,
      "learning_rate": 9.416540031707279e-06,
      "loss": 0.5213,
      "step": 11258
    },
    {
      "epoch": 0.5166811986599973,
      "grad_norm": 0.42598289251327515,
      "learning_rate": 9.416425083999307e-06,
      "loss": 0.3433,
      "step": 11259
    },
    {
      "epoch": 0.5167270891652517,
      "grad_norm": 0.4675031900405884,
      "learning_rate": 9.416310125671225e-06,
      "loss": 0.3736,
      "step": 11260
    },
    {
      "epoch": 0.5167729796705062,
      "grad_norm": 0.4364950954914093,
      "learning_rate": 9.41619515672331e-06,
      "loss": 0.3398,
      "step": 11261
    },
    {
      "epoch": 0.5168188701757607,
      "grad_norm": 0.43757903575897217,
      "learning_rate": 9.416080177155843e-06,
      "loss": 0.3723,
      "step": 11262
    },
    {
      "epoch": 0.5168647606810151,
      "grad_norm": 0.45544275641441345,
      "learning_rate": 9.415965186969094e-06,
      "loss": 0.4007,
      "step": 11263
    },
    {
      "epoch": 0.5169106511862696,
      "grad_norm": 0.48667579889297485,
      "learning_rate": 9.415850186163345e-06,
      "loss": 0.4928,
      "step": 11264
    },
    {
      "epoch": 0.516956541691524,
      "grad_norm": 0.4592079520225525,
      "learning_rate": 9.415735174738866e-06,
      "loss": 0.4204,
      "step": 11265
    },
    {
      "epoch": 0.5170024321967784,
      "grad_norm": 0.4929644763469696,
      "learning_rate": 9.41562015269594e-06,
      "loss": 0.4273,
      "step": 11266
    },
    {
      "epoch": 0.5170483227020329,
      "grad_norm": 0.42213869094848633,
      "learning_rate": 9.415505120034842e-06,
      "loss": 0.3686,
      "step": 11267
    },
    {
      "epoch": 0.5170942132072874,
      "grad_norm": 0.39224839210510254,
      "learning_rate": 9.415390076755848e-06,
      "loss": 0.2905,
      "step": 11268
    },
    {
      "epoch": 0.5171401037125418,
      "grad_norm": 0.5034732818603516,
      "learning_rate": 9.415275022859233e-06,
      "loss": 0.5203,
      "step": 11269
    },
    {
      "epoch": 0.5171859942177963,
      "grad_norm": 0.5695082545280457,
      "learning_rate": 9.415159958345276e-06,
      "loss": 0.4646,
      "step": 11270
    },
    {
      "epoch": 0.5172318847230508,
      "grad_norm": 0.4319281578063965,
      "learning_rate": 9.415044883214252e-06,
      "loss": 0.3618,
      "step": 11271
    },
    {
      "epoch": 0.5172777752283053,
      "grad_norm": 0.46490177512168884,
      "learning_rate": 9.414929797466439e-06,
      "loss": 0.4124,
      "step": 11272
    },
    {
      "epoch": 0.5173236657335597,
      "grad_norm": 0.46130844950675964,
      "learning_rate": 9.414814701102113e-06,
      "loss": 0.4481,
      "step": 11273
    },
    {
      "epoch": 0.5173695562388142,
      "grad_norm": 0.44523581862449646,
      "learning_rate": 9.41469959412155e-06,
      "loss": 0.393,
      "step": 11274
    },
    {
      "epoch": 0.5174154467440687,
      "grad_norm": 0.44459274411201477,
      "learning_rate": 9.414584476525028e-06,
      "loss": 0.357,
      "step": 11275
    },
    {
      "epoch": 0.5174613372493231,
      "grad_norm": 0.5191458463668823,
      "learning_rate": 9.414469348312824e-06,
      "loss": 0.4802,
      "step": 11276
    },
    {
      "epoch": 0.5175072277545776,
      "grad_norm": 0.4215477705001831,
      "learning_rate": 9.414354209485215e-06,
      "loss": 0.3271,
      "step": 11277
    },
    {
      "epoch": 0.5175531182598321,
      "grad_norm": 0.46454212069511414,
      "learning_rate": 9.414239060042478e-06,
      "loss": 0.345,
      "step": 11278
    },
    {
      "epoch": 0.5175990087650865,
      "grad_norm": 0.47585931420326233,
      "learning_rate": 9.414123899984888e-06,
      "loss": 0.4178,
      "step": 11279
    },
    {
      "epoch": 0.517644899270341,
      "grad_norm": 0.4738839864730835,
      "learning_rate": 9.414008729312724e-06,
      "loss": 0.3712,
      "step": 11280
    },
    {
      "epoch": 0.5176907897755955,
      "grad_norm": 0.483776718378067,
      "learning_rate": 9.41389354802626e-06,
      "loss": 0.4149,
      "step": 11281
    },
    {
      "epoch": 0.5177366802808498,
      "grad_norm": 0.4590279161930084,
      "learning_rate": 9.413778356125778e-06,
      "loss": 0.4018,
      "step": 11282
    },
    {
      "epoch": 0.5177825707861043,
      "grad_norm": 0.5332219004631042,
      "learning_rate": 9.41366315361155e-06,
      "loss": 0.4948,
      "step": 11283
    },
    {
      "epoch": 0.5178284612913588,
      "grad_norm": 0.432334840297699,
      "learning_rate": 9.413547940483855e-06,
      "loss": 0.3738,
      "step": 11284
    },
    {
      "epoch": 0.5178743517966132,
      "grad_norm": 0.4356936514377594,
      "learning_rate": 9.413432716742972e-06,
      "loss": 0.3552,
      "step": 11285
    },
    {
      "epoch": 0.5179202423018677,
      "grad_norm": 0.4573189318180084,
      "learning_rate": 9.413317482389177e-06,
      "loss": 0.3378,
      "step": 11286
    },
    {
      "epoch": 0.5179661328071222,
      "grad_norm": 0.43666619062423706,
      "learning_rate": 9.413202237422743e-06,
      "loss": 0.3541,
      "step": 11287
    },
    {
      "epoch": 0.5180120233123767,
      "grad_norm": 0.4871177673339844,
      "learning_rate": 9.413086981843953e-06,
      "loss": 0.4226,
      "step": 11288
    },
    {
      "epoch": 0.5180579138176311,
      "grad_norm": 0.5462905764579773,
      "learning_rate": 9.41297171565308e-06,
      "loss": 0.4618,
      "step": 11289
    },
    {
      "epoch": 0.5181038043228856,
      "grad_norm": 0.4553225040435791,
      "learning_rate": 9.412856438850405e-06,
      "loss": 0.3837,
      "step": 11290
    },
    {
      "epoch": 0.5181496948281401,
      "grad_norm": 0.49372997879981995,
      "learning_rate": 9.4127411514362e-06,
      "loss": 0.4312,
      "step": 11291
    },
    {
      "epoch": 0.5181955853333945,
      "grad_norm": 0.43431931734085083,
      "learning_rate": 9.412625853410747e-06,
      "loss": 0.4,
      "step": 11292
    },
    {
      "epoch": 0.518241475838649,
      "grad_norm": 0.47358494997024536,
      "learning_rate": 9.41251054477432e-06,
      "loss": 0.3717,
      "step": 11293
    },
    {
      "epoch": 0.5182873663439035,
      "grad_norm": 0.4579281806945801,
      "learning_rate": 9.4123952255272e-06,
      "loss": 0.4159,
      "step": 11294
    },
    {
      "epoch": 0.5183332568491579,
      "grad_norm": 0.4474659562110901,
      "learning_rate": 9.412279895669661e-06,
      "loss": 0.4231,
      "step": 11295
    },
    {
      "epoch": 0.5183791473544124,
      "grad_norm": 0.5042991042137146,
      "learning_rate": 9.412164555201981e-06,
      "loss": 0.4906,
      "step": 11296
    },
    {
      "epoch": 0.5184250378596669,
      "grad_norm": 0.4518040120601654,
      "learning_rate": 9.412049204124438e-06,
      "loss": 0.3935,
      "step": 11297
    },
    {
      "epoch": 0.5184709283649213,
      "grad_norm": 0.48454368114471436,
      "learning_rate": 9.41193384243731e-06,
      "loss": 0.3942,
      "step": 11298
    },
    {
      "epoch": 0.5185168188701758,
      "grad_norm": 0.48193231225013733,
      "learning_rate": 9.411818470140873e-06,
      "loss": 0.4169,
      "step": 11299
    },
    {
      "epoch": 0.5185627093754303,
      "grad_norm": 0.48706933856010437,
      "learning_rate": 9.411703087235405e-06,
      "loss": 0.3714,
      "step": 11300
    },
    {
      "epoch": 0.5186085998806846,
      "grad_norm": 0.4524175226688385,
      "learning_rate": 9.411587693721182e-06,
      "loss": 0.4065,
      "step": 11301
    },
    {
      "epoch": 0.5186544903859391,
      "grad_norm": 0.42802923917770386,
      "learning_rate": 9.411472289598486e-06,
      "loss": 0.3655,
      "step": 11302
    },
    {
      "epoch": 0.5187003808911936,
      "grad_norm": 0.4512304663658142,
      "learning_rate": 9.41135687486759e-06,
      "loss": 0.4031,
      "step": 11303
    },
    {
      "epoch": 0.518746271396448,
      "grad_norm": 0.468533992767334,
      "learning_rate": 9.411241449528772e-06,
      "loss": 0.4353,
      "step": 11304
    },
    {
      "epoch": 0.5187921619017025,
      "grad_norm": 0.4598747491836548,
      "learning_rate": 9.411126013582312e-06,
      "loss": 0.4363,
      "step": 11305
    },
    {
      "epoch": 0.518838052406957,
      "grad_norm": 0.4329906404018402,
      "learning_rate": 9.411010567028484e-06,
      "loss": 0.3792,
      "step": 11306
    },
    {
      "epoch": 0.5188839429122115,
      "grad_norm": 0.47453564405441284,
      "learning_rate": 9.410895109867571e-06,
      "loss": 0.4805,
      "step": 11307
    },
    {
      "epoch": 0.5189298334174659,
      "grad_norm": 0.5077623724937439,
      "learning_rate": 9.410779642099845e-06,
      "loss": 0.5025,
      "step": 11308
    },
    {
      "epoch": 0.5189757239227204,
      "grad_norm": 0.47868186235427856,
      "learning_rate": 9.410664163725586e-06,
      "loss": 0.4849,
      "step": 11309
    },
    {
      "epoch": 0.5190216144279749,
      "grad_norm": 0.4690362811088562,
      "learning_rate": 9.410548674745072e-06,
      "loss": 0.4197,
      "step": 11310
    },
    {
      "epoch": 0.5190675049332293,
      "grad_norm": 0.45661434531211853,
      "learning_rate": 9.41043317515858e-06,
      "loss": 0.4265,
      "step": 11311
    },
    {
      "epoch": 0.5191133954384838,
      "grad_norm": 0.4610253870487213,
      "learning_rate": 9.41031766496639e-06,
      "loss": 0.4281,
      "step": 11312
    },
    {
      "epoch": 0.5191592859437383,
      "grad_norm": 0.4959578216075897,
      "learning_rate": 9.410202144168776e-06,
      "loss": 0.4603,
      "step": 11313
    },
    {
      "epoch": 0.5192051764489927,
      "grad_norm": 0.483648419380188,
      "learning_rate": 9.410086612766018e-06,
      "loss": 0.4091,
      "step": 11314
    },
    {
      "epoch": 0.5192510669542472,
      "grad_norm": 0.4500217139720917,
      "learning_rate": 9.409971070758395e-06,
      "loss": 0.3782,
      "step": 11315
    },
    {
      "epoch": 0.5192969574595017,
      "grad_norm": 0.46670421957969666,
      "learning_rate": 9.409855518146183e-06,
      "loss": 0.3608,
      "step": 11316
    },
    {
      "epoch": 0.519342847964756,
      "grad_norm": 0.4461580216884613,
      "learning_rate": 9.40973995492966e-06,
      "loss": 0.3545,
      "step": 11317
    },
    {
      "epoch": 0.5193887384700105,
      "grad_norm": 0.46640920639038086,
      "learning_rate": 9.409624381109105e-06,
      "loss": 0.4148,
      "step": 11318
    },
    {
      "epoch": 0.519434628975265,
      "grad_norm": 0.47005024552345276,
      "learning_rate": 9.409508796684793e-06,
      "loss": 0.4295,
      "step": 11319
    },
    {
      "epoch": 0.5194805194805194,
      "grad_norm": 0.5063243508338928,
      "learning_rate": 9.409393201657004e-06,
      "loss": 0.4065,
      "step": 11320
    },
    {
      "epoch": 0.5195264099857739,
      "grad_norm": 0.4501003921031952,
      "learning_rate": 9.409277596026018e-06,
      "loss": 0.4125,
      "step": 11321
    },
    {
      "epoch": 0.5195723004910284,
      "grad_norm": 0.4959845542907715,
      "learning_rate": 9.40916197979211e-06,
      "loss": 0.4505,
      "step": 11322
    },
    {
      "epoch": 0.5196181909962828,
      "grad_norm": 0.4186255633831024,
      "learning_rate": 9.40904635295556e-06,
      "loss": 0.3528,
      "step": 11323
    },
    {
      "epoch": 0.5196640815015373,
      "grad_norm": 0.44498705863952637,
      "learning_rate": 9.408930715516643e-06,
      "loss": 0.3825,
      "step": 11324
    },
    {
      "epoch": 0.5197099720067918,
      "grad_norm": 0.5048653483390808,
      "learning_rate": 9.40881506747564e-06,
      "loss": 0.4308,
      "step": 11325
    },
    {
      "epoch": 0.5197558625120463,
      "grad_norm": 0.46696797013282776,
      "learning_rate": 9.408699408832828e-06,
      "loss": 0.4046,
      "step": 11326
    },
    {
      "epoch": 0.5198017530173007,
      "grad_norm": 0.4412844777107239,
      "learning_rate": 9.408583739588487e-06,
      "loss": 0.3626,
      "step": 11327
    },
    {
      "epoch": 0.5198476435225552,
      "grad_norm": 0.428541362285614,
      "learning_rate": 9.408468059742893e-06,
      "loss": 0.3192,
      "step": 11328
    },
    {
      "epoch": 0.5198935340278097,
      "grad_norm": 0.4453152120113373,
      "learning_rate": 9.408352369296323e-06,
      "loss": 0.3238,
      "step": 11329
    },
    {
      "epoch": 0.5199394245330641,
      "grad_norm": 0.5214999914169312,
      "learning_rate": 9.408236668249059e-06,
      "loss": 0.5346,
      "step": 11330
    },
    {
      "epoch": 0.5199853150383186,
      "grad_norm": 0.42444345355033875,
      "learning_rate": 9.408120956601376e-06,
      "loss": 0.3515,
      "step": 11331
    },
    {
      "epoch": 0.5200312055435731,
      "grad_norm": 0.46152263879776,
      "learning_rate": 9.408005234353553e-06,
      "loss": 0.404,
      "step": 11332
    },
    {
      "epoch": 0.5200770960488275,
      "grad_norm": 0.5059174299240112,
      "learning_rate": 9.407889501505868e-06,
      "loss": 0.5288,
      "step": 11333
    },
    {
      "epoch": 0.520122986554082,
      "grad_norm": 0.4300854206085205,
      "learning_rate": 9.407773758058602e-06,
      "loss": 0.356,
      "step": 11334
    },
    {
      "epoch": 0.5201688770593365,
      "grad_norm": 0.4807341396808624,
      "learning_rate": 9.407658004012029e-06,
      "loss": 0.438,
      "step": 11335
    },
    {
      "epoch": 0.5202147675645908,
      "grad_norm": 0.47700512409210205,
      "learning_rate": 9.407542239366431e-06,
      "loss": 0.3492,
      "step": 11336
    },
    {
      "epoch": 0.5202606580698453,
      "grad_norm": 0.5245160460472107,
      "learning_rate": 9.407426464122085e-06,
      "loss": 0.4858,
      "step": 11337
    },
    {
      "epoch": 0.5203065485750998,
      "grad_norm": 0.45137128233909607,
      "learning_rate": 9.407310678279269e-06,
      "loss": 0.3663,
      "step": 11338
    },
    {
      "epoch": 0.5203524390803542,
      "grad_norm": 0.4665279984474182,
      "learning_rate": 9.407194881838262e-06,
      "loss": 0.4469,
      "step": 11339
    },
    {
      "epoch": 0.5203983295856087,
      "grad_norm": 0.4119986891746521,
      "learning_rate": 9.407079074799341e-06,
      "loss": 0.2969,
      "step": 11340
    },
    {
      "epoch": 0.5204442200908632,
      "grad_norm": 0.4704146981239319,
      "learning_rate": 9.406963257162788e-06,
      "loss": 0.422,
      "step": 11341
    },
    {
      "epoch": 0.5204901105961177,
      "grad_norm": 0.4676615297794342,
      "learning_rate": 9.406847428928878e-06,
      "loss": 0.4316,
      "step": 11342
    },
    {
      "epoch": 0.5205360011013721,
      "grad_norm": 0.4753597676753998,
      "learning_rate": 9.40673159009789e-06,
      "loss": 0.4498,
      "step": 11343
    },
    {
      "epoch": 0.5205818916066266,
      "grad_norm": 0.4642932713031769,
      "learning_rate": 9.406615740670104e-06,
      "loss": 0.4192,
      "step": 11344
    },
    {
      "epoch": 0.5206277821118811,
      "grad_norm": 0.4393087923526764,
      "learning_rate": 9.406499880645799e-06,
      "loss": 0.3927,
      "step": 11345
    },
    {
      "epoch": 0.5206736726171355,
      "grad_norm": 0.4112776219844818,
      "learning_rate": 9.406384010025251e-06,
      "loss": 0.313,
      "step": 11346
    },
    {
      "epoch": 0.52071956312239,
      "grad_norm": 0.43229940533638,
      "learning_rate": 9.406268128808742e-06,
      "loss": 0.3612,
      "step": 11347
    },
    {
      "epoch": 0.5207654536276445,
      "grad_norm": 0.4701545834541321,
      "learning_rate": 9.406152236996546e-06,
      "loss": 0.4393,
      "step": 11348
    },
    {
      "epoch": 0.5208113441328989,
      "grad_norm": 0.4098416566848755,
      "learning_rate": 9.406036334588946e-06,
      "loss": 0.336,
      "step": 11349
    },
    {
      "epoch": 0.5208572346381534,
      "grad_norm": 0.41647979617118835,
      "learning_rate": 9.40592042158622e-06,
      "loss": 0.322,
      "step": 11350
    },
    {
      "epoch": 0.5209031251434079,
      "grad_norm": 0.45846766233444214,
      "learning_rate": 9.405804497988645e-06,
      "loss": 0.4217,
      "step": 11351
    },
    {
      "epoch": 0.5209490156486623,
      "grad_norm": 0.5021017789840698,
      "learning_rate": 9.405688563796503e-06,
      "loss": 0.4517,
      "step": 11352
    },
    {
      "epoch": 0.5209949061539167,
      "grad_norm": 0.47045889496803284,
      "learning_rate": 9.405572619010065e-06,
      "loss": 0.4074,
      "step": 11353
    },
    {
      "epoch": 0.5210407966591712,
      "grad_norm": 0.4577224850654602,
      "learning_rate": 9.405456663629619e-06,
      "loss": 0.438,
      "step": 11354
    },
    {
      "epoch": 0.5210866871644256,
      "grad_norm": 0.5118107795715332,
      "learning_rate": 9.40534069765544e-06,
      "loss": 0.4618,
      "step": 11355
    },
    {
      "epoch": 0.5211325776696801,
      "grad_norm": 0.4609542191028595,
      "learning_rate": 9.405224721087808e-06,
      "loss": 0.382,
      "step": 11356
    },
    {
      "epoch": 0.5211784681749346,
      "grad_norm": 0.465101957321167,
      "learning_rate": 9.405108733926999e-06,
      "loss": 0.3927,
      "step": 11357
    },
    {
      "epoch": 0.521224358680189,
      "grad_norm": 0.4497661590576172,
      "learning_rate": 9.404992736173296e-06,
      "loss": 0.3517,
      "step": 11358
    },
    {
      "epoch": 0.5212702491854435,
      "grad_norm": 0.4259025752544403,
      "learning_rate": 9.404876727826973e-06,
      "loss": 0.3641,
      "step": 11359
    },
    {
      "epoch": 0.521316139690698,
      "grad_norm": 0.5044989585876465,
      "learning_rate": 9.404760708888313e-06,
      "loss": 0.4948,
      "step": 11360
    },
    {
      "epoch": 0.5213620301959525,
      "grad_norm": 0.43325573205947876,
      "learning_rate": 9.404644679357593e-06,
      "loss": 0.3806,
      "step": 11361
    },
    {
      "epoch": 0.5214079207012069,
      "grad_norm": 0.46886196732521057,
      "learning_rate": 9.404528639235094e-06,
      "loss": 0.4116,
      "step": 11362
    },
    {
      "epoch": 0.5214538112064614,
      "grad_norm": 0.42928507924079895,
      "learning_rate": 9.404412588521093e-06,
      "loss": 0.3297,
      "step": 11363
    },
    {
      "epoch": 0.5214997017117159,
      "grad_norm": 0.46476927399635315,
      "learning_rate": 9.404296527215871e-06,
      "loss": 0.44,
      "step": 11364
    },
    {
      "epoch": 0.5215455922169703,
      "grad_norm": 0.4753119945526123,
      "learning_rate": 9.404180455319704e-06,
      "loss": 0.416,
      "step": 11365
    },
    {
      "epoch": 0.5215914827222248,
      "grad_norm": 0.4441254436969757,
      "learning_rate": 9.404064372832873e-06,
      "loss": 0.3513,
      "step": 11366
    },
    {
      "epoch": 0.5216373732274793,
      "grad_norm": 0.42617788910865784,
      "learning_rate": 9.403948279755658e-06,
      "loss": 0.3654,
      "step": 11367
    },
    {
      "epoch": 0.5216832637327337,
      "grad_norm": 0.480732798576355,
      "learning_rate": 9.403832176088337e-06,
      "loss": 0.4372,
      "step": 11368
    },
    {
      "epoch": 0.5217291542379882,
      "grad_norm": 0.45252206921577454,
      "learning_rate": 9.40371606183119e-06,
      "loss": 0.3618,
      "step": 11369
    },
    {
      "epoch": 0.5217750447432427,
      "grad_norm": 0.48564642667770386,
      "learning_rate": 9.403599936984495e-06,
      "loss": 0.4535,
      "step": 11370
    },
    {
      "epoch": 0.521820935248497,
      "grad_norm": 0.4580940902233124,
      "learning_rate": 9.403483801548532e-06,
      "loss": 0.3796,
      "step": 11371
    },
    {
      "epoch": 0.5218668257537515,
      "grad_norm": 0.47691118717193604,
      "learning_rate": 9.40336765552358e-06,
      "loss": 0.4255,
      "step": 11372
    },
    {
      "epoch": 0.521912716259006,
      "grad_norm": 0.5334489941596985,
      "learning_rate": 9.40325149890992e-06,
      "loss": 0.4755,
      "step": 11373
    },
    {
      "epoch": 0.5219586067642604,
      "grad_norm": 0.45369982719421387,
      "learning_rate": 9.403135331707828e-06,
      "loss": 0.4912,
      "step": 11374
    },
    {
      "epoch": 0.5220044972695149,
      "grad_norm": 0.4181894063949585,
      "learning_rate": 9.403019153917587e-06,
      "loss": 0.3399,
      "step": 11375
    },
    {
      "epoch": 0.5220503877747694,
      "grad_norm": 0.439365953207016,
      "learning_rate": 9.402902965539473e-06,
      "loss": 0.4084,
      "step": 11376
    },
    {
      "epoch": 0.5220962782800239,
      "grad_norm": 0.4301898181438446,
      "learning_rate": 9.402786766573768e-06,
      "loss": 0.3335,
      "step": 11377
    },
    {
      "epoch": 0.5221421687852783,
      "grad_norm": 0.5846481323242188,
      "learning_rate": 9.40267055702075e-06,
      "loss": 0.5041,
      "step": 11378
    },
    {
      "epoch": 0.5221880592905328,
      "grad_norm": 0.48720747232437134,
      "learning_rate": 9.4025543368807e-06,
      "loss": 0.4601,
      "step": 11379
    },
    {
      "epoch": 0.5222339497957873,
      "grad_norm": 0.4171673655509949,
      "learning_rate": 9.402438106153895e-06,
      "loss": 0.3284,
      "step": 11380
    },
    {
      "epoch": 0.5222798403010417,
      "grad_norm": 0.49452322721481323,
      "learning_rate": 9.402321864840616e-06,
      "loss": 0.4907,
      "step": 11381
    },
    {
      "epoch": 0.5223257308062962,
      "grad_norm": 0.4706476032733917,
      "learning_rate": 9.402205612941142e-06,
      "loss": 0.3945,
      "step": 11382
    },
    {
      "epoch": 0.5223716213115507,
      "grad_norm": 0.442869633436203,
      "learning_rate": 9.402089350455753e-06,
      "loss": 0.3734,
      "step": 11383
    },
    {
      "epoch": 0.5224175118168051,
      "grad_norm": 0.4574196934700012,
      "learning_rate": 9.401973077384729e-06,
      "loss": 0.3933,
      "step": 11384
    },
    {
      "epoch": 0.5224634023220596,
      "grad_norm": 0.4647204875946045,
      "learning_rate": 9.401856793728348e-06,
      "loss": 0.4053,
      "step": 11385
    },
    {
      "epoch": 0.5225092928273141,
      "grad_norm": 0.46060729026794434,
      "learning_rate": 9.401740499486892e-06,
      "loss": 0.4149,
      "step": 11386
    },
    {
      "epoch": 0.5225551833325685,
      "grad_norm": 0.47708842158317566,
      "learning_rate": 9.401624194660638e-06,
      "loss": 0.4305,
      "step": 11387
    },
    {
      "epoch": 0.522601073837823,
      "grad_norm": 0.4464203715324402,
      "learning_rate": 9.401507879249868e-06,
      "loss": 0.3981,
      "step": 11388
    },
    {
      "epoch": 0.5226469643430774,
      "grad_norm": 0.5870711207389832,
      "learning_rate": 9.40139155325486e-06,
      "loss": 0.4062,
      "step": 11389
    },
    {
      "epoch": 0.5226928548483318,
      "grad_norm": 0.46573159098625183,
      "learning_rate": 9.401275216675896e-06,
      "loss": 0.4284,
      "step": 11390
    },
    {
      "epoch": 0.5227387453535863,
      "grad_norm": 0.46834370493888855,
      "learning_rate": 9.401158869513253e-06,
      "loss": 0.4096,
      "step": 11391
    },
    {
      "epoch": 0.5227846358588408,
      "grad_norm": 0.4613339900970459,
      "learning_rate": 9.40104251176721e-06,
      "loss": 0.4453,
      "step": 11392
    },
    {
      "epoch": 0.5228305263640952,
      "grad_norm": 0.4914063811302185,
      "learning_rate": 9.400926143438053e-06,
      "loss": 0.4452,
      "step": 11393
    },
    {
      "epoch": 0.5228764168693497,
      "grad_norm": 0.4860925078392029,
      "learning_rate": 9.400809764526056e-06,
      "loss": 0.3974,
      "step": 11394
    },
    {
      "epoch": 0.5229223073746042,
      "grad_norm": 0.4686044454574585,
      "learning_rate": 9.400693375031498e-06,
      "loss": 0.3828,
      "step": 11395
    },
    {
      "epoch": 0.5229681978798587,
      "grad_norm": 0.4933861494064331,
      "learning_rate": 9.400576974954664e-06,
      "loss": 0.3958,
      "step": 11396
    },
    {
      "epoch": 0.5230140883851131,
      "grad_norm": 0.46955105662345886,
      "learning_rate": 9.400460564295831e-06,
      "loss": 0.4582,
      "step": 11397
    },
    {
      "epoch": 0.5230599788903676,
      "grad_norm": 0.5272611975669861,
      "learning_rate": 9.400344143055278e-06,
      "loss": 0.3525,
      "step": 11398
    },
    {
      "epoch": 0.5231058693956221,
      "grad_norm": 0.4410994052886963,
      "learning_rate": 9.400227711233287e-06,
      "loss": 0.381,
      "step": 11399
    },
    {
      "epoch": 0.5231517599008765,
      "grad_norm": 0.4742319583892822,
      "learning_rate": 9.400111268830136e-06,
      "loss": 0.4431,
      "step": 11400
    },
    {
      "epoch": 0.523197650406131,
      "grad_norm": 0.44880059361457825,
      "learning_rate": 9.399994815846107e-06,
      "loss": 0.346,
      "step": 11401
    },
    {
      "epoch": 0.5232435409113855,
      "grad_norm": 0.48278918862342834,
      "learning_rate": 9.39987835228148e-06,
      "loss": 0.3819,
      "step": 11402
    },
    {
      "epoch": 0.5232894314166399,
      "grad_norm": 0.479733407497406,
      "learning_rate": 9.399761878136533e-06,
      "loss": 0.3756,
      "step": 11403
    },
    {
      "epoch": 0.5233353219218944,
      "grad_norm": 0.4563409984111786,
      "learning_rate": 9.399645393411547e-06,
      "loss": 0.4333,
      "step": 11404
    },
    {
      "epoch": 0.5233812124271489,
      "grad_norm": 0.4592234194278717,
      "learning_rate": 9.399528898106802e-06,
      "loss": 0.3642,
      "step": 11405
    },
    {
      "epoch": 0.5234271029324032,
      "grad_norm": 0.4508698284626007,
      "learning_rate": 9.39941239222258e-06,
      "loss": 0.4531,
      "step": 11406
    },
    {
      "epoch": 0.5234729934376577,
      "grad_norm": 0.4498033821582794,
      "learning_rate": 9.39929587575916e-06,
      "loss": 0.4034,
      "step": 11407
    },
    {
      "epoch": 0.5235188839429122,
      "grad_norm": 0.490349143743515,
      "learning_rate": 9.399179348716821e-06,
      "loss": 0.4193,
      "step": 11408
    },
    {
      "epoch": 0.5235647744481666,
      "grad_norm": 0.4533957839012146,
      "learning_rate": 9.399062811095844e-06,
      "loss": 0.4236,
      "step": 11409
    },
    {
      "epoch": 0.5236106649534211,
      "grad_norm": 0.42886388301849365,
      "learning_rate": 9.39894626289651e-06,
      "loss": 0.3874,
      "step": 11410
    },
    {
      "epoch": 0.5236565554586756,
      "grad_norm": 0.48032861948013306,
      "learning_rate": 9.398829704119097e-06,
      "loss": 0.3871,
      "step": 11411
    },
    {
      "epoch": 0.52370244596393,
      "grad_norm": 0.488559752702713,
      "learning_rate": 9.398713134763888e-06,
      "loss": 0.4298,
      "step": 11412
    },
    {
      "epoch": 0.5237483364691845,
      "grad_norm": 0.5009146928787231,
      "learning_rate": 9.398596554831164e-06,
      "loss": 0.4035,
      "step": 11413
    },
    {
      "epoch": 0.523794226974439,
      "grad_norm": 0.5082897543907166,
      "learning_rate": 9.3984799643212e-06,
      "loss": 0.4693,
      "step": 11414
    },
    {
      "epoch": 0.5238401174796935,
      "grad_norm": 0.47286292910575867,
      "learning_rate": 9.398363363234282e-06,
      "loss": 0.4326,
      "step": 11415
    },
    {
      "epoch": 0.5238860079849479,
      "grad_norm": 0.4564986526966095,
      "learning_rate": 9.398246751570688e-06,
      "loss": 0.4197,
      "step": 11416
    },
    {
      "epoch": 0.5239318984902024,
      "grad_norm": 0.4848954379558563,
      "learning_rate": 9.398130129330699e-06,
      "loss": 0.4419,
      "step": 11417
    },
    {
      "epoch": 0.5239777889954569,
      "grad_norm": 0.49837687611579895,
      "learning_rate": 9.398013496514592e-06,
      "loss": 0.4929,
      "step": 11418
    },
    {
      "epoch": 0.5240236795007113,
      "grad_norm": 0.705758810043335,
      "learning_rate": 9.397896853122654e-06,
      "loss": 0.3419,
      "step": 11419
    },
    {
      "epoch": 0.5240695700059658,
      "grad_norm": 0.4462769031524658,
      "learning_rate": 9.397780199155163e-06,
      "loss": 0.3325,
      "step": 11420
    },
    {
      "epoch": 0.5241154605112203,
      "grad_norm": 0.4101276099681854,
      "learning_rate": 9.397663534612395e-06,
      "loss": 0.3415,
      "step": 11421
    },
    {
      "epoch": 0.5241613510164747,
      "grad_norm": 0.49243876338005066,
      "learning_rate": 9.397546859494637e-06,
      "loss": 0.4201,
      "step": 11422
    },
    {
      "epoch": 0.5242072415217292,
      "grad_norm": 0.4809768497943878,
      "learning_rate": 9.397430173802165e-06,
      "loss": 0.4516,
      "step": 11423
    },
    {
      "epoch": 0.5242531320269836,
      "grad_norm": 0.46915292739868164,
      "learning_rate": 9.39731347753526e-06,
      "loss": 0.4209,
      "step": 11424
    },
    {
      "epoch": 0.524299022532238,
      "grad_norm": 0.5892186164855957,
      "learning_rate": 9.397196770694207e-06,
      "loss": 0.4467,
      "step": 11425
    },
    {
      "epoch": 0.5243449130374925,
      "grad_norm": 0.46034806966781616,
      "learning_rate": 9.397080053279284e-06,
      "loss": 0.4012,
      "step": 11426
    },
    {
      "epoch": 0.524390803542747,
      "grad_norm": 0.698807418346405,
      "learning_rate": 9.39696332529077e-06,
      "loss": 0.4257,
      "step": 11427
    },
    {
      "epoch": 0.5244366940480014,
      "grad_norm": 0.5127604603767395,
      "learning_rate": 9.396846586728945e-06,
      "loss": 0.4437,
      "step": 11428
    },
    {
      "epoch": 0.5244825845532559,
      "grad_norm": 0.4466533958911896,
      "learning_rate": 9.396729837594094e-06,
      "loss": 0.3849,
      "step": 11429
    },
    {
      "epoch": 0.5245284750585104,
      "grad_norm": 0.4703664183616638,
      "learning_rate": 9.396613077886495e-06,
      "loss": 0.4689,
      "step": 11430
    },
    {
      "epoch": 0.5245743655637649,
      "grad_norm": 0.47790801525115967,
      "learning_rate": 9.396496307606428e-06,
      "loss": 0.4715,
      "step": 11431
    },
    {
      "epoch": 0.5246202560690193,
      "grad_norm": 0.49904265999794006,
      "learning_rate": 9.396379526754177e-06,
      "loss": 0.4157,
      "step": 11432
    },
    {
      "epoch": 0.5246661465742738,
      "grad_norm": 0.4207460582256317,
      "learning_rate": 9.39626273533002e-06,
      "loss": 0.3297,
      "step": 11433
    },
    {
      "epoch": 0.5247120370795283,
      "grad_norm": 0.46035128831863403,
      "learning_rate": 9.396145933334237e-06,
      "loss": 0.3639,
      "step": 11434
    },
    {
      "epoch": 0.5247579275847827,
      "grad_norm": 0.45932647585868835,
      "learning_rate": 9.396029120767112e-06,
      "loss": 0.4179,
      "step": 11435
    },
    {
      "epoch": 0.5248038180900372,
      "grad_norm": 0.4972432851791382,
      "learning_rate": 9.395912297628925e-06,
      "loss": 0.4649,
      "step": 11436
    },
    {
      "epoch": 0.5248497085952917,
      "grad_norm": 0.45799997448921204,
      "learning_rate": 9.395795463919954e-06,
      "loss": 0.3257,
      "step": 11437
    },
    {
      "epoch": 0.5248955991005461,
      "grad_norm": 0.4304791986942291,
      "learning_rate": 9.395678619640485e-06,
      "loss": 0.3226,
      "step": 11438
    },
    {
      "epoch": 0.5249414896058006,
      "grad_norm": 0.4275456368923187,
      "learning_rate": 9.395561764790795e-06,
      "loss": 0.3379,
      "step": 11439
    },
    {
      "epoch": 0.5249873801110551,
      "grad_norm": 0.4632872939109802,
      "learning_rate": 9.395444899371166e-06,
      "loss": 0.4199,
      "step": 11440
    },
    {
      "epoch": 0.5250332706163094,
      "grad_norm": 0.47015640139579773,
      "learning_rate": 9.395328023381878e-06,
      "loss": 0.3496,
      "step": 11441
    },
    {
      "epoch": 0.5250791611215639,
      "grad_norm": 0.45110246539115906,
      "learning_rate": 9.395211136823215e-06,
      "loss": 0.4339,
      "step": 11442
    },
    {
      "epoch": 0.5251250516268184,
      "grad_norm": 0.4446107745170593,
      "learning_rate": 9.395094239695456e-06,
      "loss": 0.3649,
      "step": 11443
    },
    {
      "epoch": 0.5251709421320728,
      "grad_norm": 0.45945775508880615,
      "learning_rate": 9.394977331998883e-06,
      "loss": 0.3658,
      "step": 11444
    },
    {
      "epoch": 0.5252168326373273,
      "grad_norm": 0.42512863874435425,
      "learning_rate": 9.394860413733776e-06,
      "loss": 0.3323,
      "step": 11445
    },
    {
      "epoch": 0.5252627231425818,
      "grad_norm": 0.45659297704696655,
      "learning_rate": 9.394743484900417e-06,
      "loss": 0.4295,
      "step": 11446
    },
    {
      "epoch": 0.5253086136478362,
      "grad_norm": 0.42990642786026,
      "learning_rate": 9.394626545499087e-06,
      "loss": 0.366,
      "step": 11447
    },
    {
      "epoch": 0.5253545041530907,
      "grad_norm": 0.47807377576828003,
      "learning_rate": 9.394509595530067e-06,
      "loss": 0.4271,
      "step": 11448
    },
    {
      "epoch": 0.5254003946583452,
      "grad_norm": 0.4524679183959961,
      "learning_rate": 9.394392634993639e-06,
      "loss": 0.4002,
      "step": 11449
    },
    {
      "epoch": 0.5254462851635997,
      "grad_norm": 0.4388274848461151,
      "learning_rate": 9.394275663890083e-06,
      "loss": 0.3454,
      "step": 11450
    },
    {
      "epoch": 0.5254921756688541,
      "grad_norm": 0.4434483051300049,
      "learning_rate": 9.394158682219681e-06,
      "loss": 0.3799,
      "step": 11451
    },
    {
      "epoch": 0.5255380661741086,
      "grad_norm": 0.46120551228523254,
      "learning_rate": 9.394041689982714e-06,
      "loss": 0.3764,
      "step": 11452
    },
    {
      "epoch": 0.5255839566793631,
      "grad_norm": 0.48892879486083984,
      "learning_rate": 9.393924687179464e-06,
      "loss": 0.4386,
      "step": 11453
    },
    {
      "epoch": 0.5256298471846175,
      "grad_norm": 0.4669158160686493,
      "learning_rate": 9.39380767381021e-06,
      "loss": 0.382,
      "step": 11454
    },
    {
      "epoch": 0.525675737689872,
      "grad_norm": 0.5157306790351868,
      "learning_rate": 9.393690649875238e-06,
      "loss": 0.4459,
      "step": 11455
    },
    {
      "epoch": 0.5257216281951265,
      "grad_norm": 0.4693259596824646,
      "learning_rate": 9.393573615374825e-06,
      "loss": 0.4466,
      "step": 11456
    },
    {
      "epoch": 0.5257675187003809,
      "grad_norm": 0.4892548620700836,
      "learning_rate": 9.393456570309256e-06,
      "loss": 0.4453,
      "step": 11457
    },
    {
      "epoch": 0.5258134092056354,
      "grad_norm": 0.4841972887516022,
      "learning_rate": 9.393339514678808e-06,
      "loss": 0.4389,
      "step": 11458
    },
    {
      "epoch": 0.5258592997108898,
      "grad_norm": 0.46456146240234375,
      "learning_rate": 9.393222448483766e-06,
      "loss": 0.3919,
      "step": 11459
    },
    {
      "epoch": 0.5259051902161442,
      "grad_norm": 0.46003010869026184,
      "learning_rate": 9.393105371724412e-06,
      "loss": 0.4,
      "step": 11460
    },
    {
      "epoch": 0.5259510807213987,
      "grad_norm": 0.4810815453529358,
      "learning_rate": 9.392988284401023e-06,
      "loss": 0.4253,
      "step": 11461
    },
    {
      "epoch": 0.5259969712266532,
      "grad_norm": 0.456248939037323,
      "learning_rate": 9.392871186513885e-06,
      "loss": 0.3881,
      "step": 11462
    },
    {
      "epoch": 0.5260428617319076,
      "grad_norm": 0.4862266480922699,
      "learning_rate": 9.39275407806328e-06,
      "loss": 0.4147,
      "step": 11463
    },
    {
      "epoch": 0.5260887522371621,
      "grad_norm": 0.42009127140045166,
      "learning_rate": 9.392636959049486e-06,
      "loss": 0.331,
      "step": 11464
    },
    {
      "epoch": 0.5261346427424166,
      "grad_norm": 0.4570986032485962,
      "learning_rate": 9.392519829472788e-06,
      "loss": 0.4042,
      "step": 11465
    },
    {
      "epoch": 0.5261805332476711,
      "grad_norm": 0.42909303307533264,
      "learning_rate": 9.392402689333463e-06,
      "loss": 0.3171,
      "step": 11466
    },
    {
      "epoch": 0.5262264237529255,
      "grad_norm": 0.4438481628894806,
      "learning_rate": 9.3922855386318e-06,
      "loss": 0.3948,
      "step": 11467
    },
    {
      "epoch": 0.52627231425818,
      "grad_norm": 0.45915961265563965,
      "learning_rate": 9.392168377368074e-06,
      "loss": 0.3877,
      "step": 11468
    },
    {
      "epoch": 0.5263182047634345,
      "grad_norm": 0.42701494693756104,
      "learning_rate": 9.392051205542569e-06,
      "loss": 0.3343,
      "step": 11469
    },
    {
      "epoch": 0.5263640952686889,
      "grad_norm": 0.44611984491348267,
      "learning_rate": 9.391934023155569e-06,
      "loss": 0.3746,
      "step": 11470
    },
    {
      "epoch": 0.5264099857739434,
      "grad_norm": 0.4977155327796936,
      "learning_rate": 9.391816830207352e-06,
      "loss": 0.4444,
      "step": 11471
    },
    {
      "epoch": 0.5264558762791979,
      "grad_norm": 0.44357648491859436,
      "learning_rate": 9.391699626698203e-06,
      "loss": 0.3478,
      "step": 11472
    },
    {
      "epoch": 0.5265017667844523,
      "grad_norm": 0.5211843252182007,
      "learning_rate": 9.391582412628401e-06,
      "loss": 0.4295,
      "step": 11473
    },
    {
      "epoch": 0.5265476572897068,
      "grad_norm": 0.47220176458358765,
      "learning_rate": 9.39146518799823e-06,
      "loss": 0.4472,
      "step": 11474
    },
    {
      "epoch": 0.5265935477949613,
      "grad_norm": 0.4938763678073883,
      "learning_rate": 9.391347952807972e-06,
      "loss": 0.5066,
      "step": 11475
    },
    {
      "epoch": 0.5266394383002156,
      "grad_norm": 0.46761298179626465,
      "learning_rate": 9.391230707057906e-06,
      "loss": 0.4883,
      "step": 11476
    },
    {
      "epoch": 0.5266853288054701,
      "grad_norm": 0.4353277385234833,
      "learning_rate": 9.391113450748318e-06,
      "loss": 0.3962,
      "step": 11477
    },
    {
      "epoch": 0.5267312193107246,
      "grad_norm": 0.4347732365131378,
      "learning_rate": 9.390996183879488e-06,
      "loss": 0.3452,
      "step": 11478
    },
    {
      "epoch": 0.526777109815979,
      "grad_norm": 0.5078023672103882,
      "learning_rate": 9.390878906451697e-06,
      "loss": 0.5234,
      "step": 11479
    },
    {
      "epoch": 0.5268230003212335,
      "grad_norm": 0.4796895384788513,
      "learning_rate": 9.390761618465228e-06,
      "loss": 0.4465,
      "step": 11480
    },
    {
      "epoch": 0.526868890826488,
      "grad_norm": 0.5031535625457764,
      "learning_rate": 9.390644319920363e-06,
      "loss": 0.5324,
      "step": 11481
    },
    {
      "epoch": 0.5269147813317424,
      "grad_norm": 0.432140588760376,
      "learning_rate": 9.390527010817385e-06,
      "loss": 0.3336,
      "step": 11482
    },
    {
      "epoch": 0.5269606718369969,
      "grad_norm": 0.4209963083267212,
      "learning_rate": 9.390409691156576e-06,
      "loss": 0.3781,
      "step": 11483
    },
    {
      "epoch": 0.5270065623422514,
      "grad_norm": 0.4993205964565277,
      "learning_rate": 9.390292360938215e-06,
      "loss": 0.4563,
      "step": 11484
    },
    {
      "epoch": 0.5270524528475059,
      "grad_norm": 0.4465543329715729,
      "learning_rate": 9.390175020162588e-06,
      "loss": 0.4116,
      "step": 11485
    },
    {
      "epoch": 0.5270983433527603,
      "grad_norm": 0.40897610783576965,
      "learning_rate": 9.390057668829976e-06,
      "loss": 0.3078,
      "step": 11486
    },
    {
      "epoch": 0.5271442338580148,
      "grad_norm": 0.4544394314289093,
      "learning_rate": 9.38994030694066e-06,
      "loss": 0.463,
      "step": 11487
    },
    {
      "epoch": 0.5271901243632693,
      "grad_norm": 0.464700847864151,
      "learning_rate": 9.389822934494923e-06,
      "loss": 0.4466,
      "step": 11488
    },
    {
      "epoch": 0.5272360148685237,
      "grad_norm": 0.4608149826526642,
      "learning_rate": 9.389705551493048e-06,
      "loss": 0.39,
      "step": 11489
    },
    {
      "epoch": 0.5272819053737782,
      "grad_norm": 0.48113757371902466,
      "learning_rate": 9.389588157935317e-06,
      "loss": 0.4944,
      "step": 11490
    },
    {
      "epoch": 0.5273277958790327,
      "grad_norm": 0.45270052552223206,
      "learning_rate": 9.38947075382201e-06,
      "loss": 0.4081,
      "step": 11491
    },
    {
      "epoch": 0.527373686384287,
      "grad_norm": 0.4248492419719696,
      "learning_rate": 9.389353339153413e-06,
      "loss": 0.3493,
      "step": 11492
    },
    {
      "epoch": 0.5274195768895416,
      "grad_norm": 0.463619202375412,
      "learning_rate": 9.389235913929806e-06,
      "loss": 0.4243,
      "step": 11493
    },
    {
      "epoch": 0.527465467394796,
      "grad_norm": 0.5470955967903137,
      "learning_rate": 9.389118478151471e-06,
      "loss": 0.5999,
      "step": 11494
    },
    {
      "epoch": 0.5275113579000504,
      "grad_norm": 0.4415670335292816,
      "learning_rate": 9.389001031818691e-06,
      "loss": 0.385,
      "step": 11495
    },
    {
      "epoch": 0.5275572484053049,
      "grad_norm": 0.49321937561035156,
      "learning_rate": 9.38888357493175e-06,
      "loss": 0.4911,
      "step": 11496
    },
    {
      "epoch": 0.5276031389105594,
      "grad_norm": 0.4510864317417145,
      "learning_rate": 9.38876610749093e-06,
      "loss": 0.3403,
      "step": 11497
    },
    {
      "epoch": 0.5276490294158138,
      "grad_norm": 0.45155200362205505,
      "learning_rate": 9.38864862949651e-06,
      "loss": 0.3588,
      "step": 11498
    },
    {
      "epoch": 0.5276949199210683,
      "grad_norm": 0.44299718737602234,
      "learning_rate": 9.388531140948778e-06,
      "loss": 0.3903,
      "step": 11499
    },
    {
      "epoch": 0.5277408104263228,
      "grad_norm": 0.4556327760219574,
      "learning_rate": 9.388413641848012e-06,
      "loss": 0.3739,
      "step": 11500
    },
    {
      "epoch": 0.5277867009315772,
      "grad_norm": 0.4424279034137726,
      "learning_rate": 9.388296132194496e-06,
      "loss": 0.33,
      "step": 11501
    },
    {
      "epoch": 0.5278325914368317,
      "grad_norm": 0.44124263525009155,
      "learning_rate": 9.388178611988515e-06,
      "loss": 0.3677,
      "step": 11502
    },
    {
      "epoch": 0.5278784819420862,
      "grad_norm": 0.471493124961853,
      "learning_rate": 9.388061081230347e-06,
      "loss": 0.3666,
      "step": 11503
    },
    {
      "epoch": 0.5279243724473407,
      "grad_norm": 0.47322705388069153,
      "learning_rate": 9.387943539920278e-06,
      "loss": 0.4306,
      "step": 11504
    },
    {
      "epoch": 0.5279702629525951,
      "grad_norm": 0.4894491136074066,
      "learning_rate": 9.38782598805859e-06,
      "loss": 0.383,
      "step": 11505
    },
    {
      "epoch": 0.5280161534578496,
      "grad_norm": 0.4657978415489197,
      "learning_rate": 9.387708425645564e-06,
      "loss": 0.3862,
      "step": 11506
    },
    {
      "epoch": 0.5280620439631041,
      "grad_norm": 0.48629364371299744,
      "learning_rate": 9.387590852681485e-06,
      "loss": 0.4399,
      "step": 11507
    },
    {
      "epoch": 0.5281079344683585,
      "grad_norm": 0.4670019745826721,
      "learning_rate": 9.387473269166634e-06,
      "loss": 0.4486,
      "step": 11508
    },
    {
      "epoch": 0.528153824973613,
      "grad_norm": 0.4332893192768097,
      "learning_rate": 9.387355675101294e-06,
      "loss": 0.3464,
      "step": 11509
    },
    {
      "epoch": 0.5281997154788675,
      "grad_norm": 0.4641135334968567,
      "learning_rate": 9.387238070485751e-06,
      "loss": 0.3874,
      "step": 11510
    },
    {
      "epoch": 0.5282456059841218,
      "grad_norm": 0.496931254863739,
      "learning_rate": 9.387120455320284e-06,
      "loss": 0.472,
      "step": 11511
    },
    {
      "epoch": 0.5282914964893763,
      "grad_norm": 0.4468570053577423,
      "learning_rate": 9.387002829605175e-06,
      "loss": 0.3432,
      "step": 11512
    },
    {
      "epoch": 0.5283373869946308,
      "grad_norm": 0.4977463483810425,
      "learning_rate": 9.386885193340712e-06,
      "loss": 0.4429,
      "step": 11513
    },
    {
      "epoch": 0.5283832774998852,
      "grad_norm": 0.48747384548187256,
      "learning_rate": 9.386767546527172e-06,
      "loss": 0.5108,
      "step": 11514
    },
    {
      "epoch": 0.5284291680051397,
      "grad_norm": 0.45584872364997864,
      "learning_rate": 9.38664988916484e-06,
      "loss": 0.404,
      "step": 11515
    },
    {
      "epoch": 0.5284750585103942,
      "grad_norm": 0.4881237745285034,
      "learning_rate": 9.386532221254001e-06,
      "loss": 0.4138,
      "step": 11516
    },
    {
      "epoch": 0.5285209490156486,
      "grad_norm": 0.4662555754184723,
      "learning_rate": 9.386414542794937e-06,
      "loss": 0.4162,
      "step": 11517
    },
    {
      "epoch": 0.5285668395209031,
      "grad_norm": 0.4696439206600189,
      "learning_rate": 9.38629685378793e-06,
      "loss": 0.4537,
      "step": 11518
    },
    {
      "epoch": 0.5286127300261576,
      "grad_norm": 0.46920210123062134,
      "learning_rate": 9.386179154233264e-06,
      "loss": 0.4406,
      "step": 11519
    },
    {
      "epoch": 0.5286586205314121,
      "grad_norm": 0.47456490993499756,
      "learning_rate": 9.386061444131219e-06,
      "loss": 0.4727,
      "step": 11520
    },
    {
      "epoch": 0.5287045110366665,
      "grad_norm": 0.4964886009693146,
      "learning_rate": 9.385943723482083e-06,
      "loss": 0.4117,
      "step": 11521
    },
    {
      "epoch": 0.528750401541921,
      "grad_norm": 0.4609547555446625,
      "learning_rate": 9.385825992286136e-06,
      "loss": 0.3918,
      "step": 11522
    },
    {
      "epoch": 0.5287962920471755,
      "grad_norm": 0.4713539779186249,
      "learning_rate": 9.38570825054366e-06,
      "loss": 0.3993,
      "step": 11523
    },
    {
      "epoch": 0.5288421825524299,
      "grad_norm": 0.4443165957927704,
      "learning_rate": 9.385590498254942e-06,
      "loss": 0.3728,
      "step": 11524
    },
    {
      "epoch": 0.5288880730576844,
      "grad_norm": 0.5092284679412842,
      "learning_rate": 9.385472735420263e-06,
      "loss": 0.5394,
      "step": 11525
    },
    {
      "epoch": 0.5289339635629389,
      "grad_norm": 0.4156496524810791,
      "learning_rate": 9.385354962039904e-06,
      "loss": 0.3494,
      "step": 11526
    },
    {
      "epoch": 0.5289798540681933,
      "grad_norm": 0.48917466402053833,
      "learning_rate": 9.385237178114152e-06,
      "loss": 0.4481,
      "step": 11527
    },
    {
      "epoch": 0.5290257445734478,
      "grad_norm": 0.4595395028591156,
      "learning_rate": 9.385119383643288e-06,
      "loss": 0.3885,
      "step": 11528
    },
    {
      "epoch": 0.5290716350787023,
      "grad_norm": 0.4985535144805908,
      "learning_rate": 9.385001578627597e-06,
      "loss": 0.3819,
      "step": 11529
    },
    {
      "epoch": 0.5291175255839566,
      "grad_norm": 0.4437890946865082,
      "learning_rate": 9.384883763067361e-06,
      "loss": 0.373,
      "step": 11530
    },
    {
      "epoch": 0.5291634160892111,
      "grad_norm": 0.44906851649284363,
      "learning_rate": 9.384765936962862e-06,
      "loss": 0.3844,
      "step": 11531
    },
    {
      "epoch": 0.5292093065944656,
      "grad_norm": 0.4874633252620697,
      "learning_rate": 9.384648100314386e-06,
      "loss": 0.4687,
      "step": 11532
    },
    {
      "epoch": 0.52925519709972,
      "grad_norm": 0.4595133364200592,
      "learning_rate": 9.384530253122213e-06,
      "loss": 0.4055,
      "step": 11533
    },
    {
      "epoch": 0.5293010876049745,
      "grad_norm": 0.47851428389549255,
      "learning_rate": 9.38441239538663e-06,
      "loss": 0.4401,
      "step": 11534
    },
    {
      "epoch": 0.529346978110229,
      "grad_norm": 0.495633602142334,
      "learning_rate": 9.38429452710792e-06,
      "loss": 0.4195,
      "step": 11535
    },
    {
      "epoch": 0.5293928686154834,
      "grad_norm": 0.5485066175460815,
      "learning_rate": 9.384176648286365e-06,
      "loss": 0.5194,
      "step": 11536
    },
    {
      "epoch": 0.5294387591207379,
      "grad_norm": 0.4835604727268219,
      "learning_rate": 9.384058758922248e-06,
      "loss": 0.4097,
      "step": 11537
    },
    {
      "epoch": 0.5294846496259924,
      "grad_norm": 0.4388428032398224,
      "learning_rate": 9.383940859015854e-06,
      "loss": 0.3257,
      "step": 11538
    },
    {
      "epoch": 0.5295305401312469,
      "grad_norm": 0.46346622705459595,
      "learning_rate": 9.383822948567466e-06,
      "loss": 0.3945,
      "step": 11539
    },
    {
      "epoch": 0.5295764306365013,
      "grad_norm": 0.4459650218486786,
      "learning_rate": 9.383705027577368e-06,
      "loss": 0.3418,
      "step": 11540
    },
    {
      "epoch": 0.5296223211417558,
      "grad_norm": 0.49368637800216675,
      "learning_rate": 9.38358709604584e-06,
      "loss": 0.4209,
      "step": 11541
    },
    {
      "epoch": 0.5296682116470103,
      "grad_norm": 0.4806511402130127,
      "learning_rate": 9.38346915397317e-06,
      "loss": 0.3934,
      "step": 11542
    },
    {
      "epoch": 0.5297141021522647,
      "grad_norm": 0.4567352831363678,
      "learning_rate": 9.383351201359644e-06,
      "loss": 0.3847,
      "step": 11543
    },
    {
      "epoch": 0.5297599926575192,
      "grad_norm": 0.42276278138160706,
      "learning_rate": 9.383233238205537e-06,
      "loss": 0.3381,
      "step": 11544
    },
    {
      "epoch": 0.5298058831627737,
      "grad_norm": 0.4420825242996216,
      "learning_rate": 9.383115264511138e-06,
      "loss": 0.4147,
      "step": 11545
    },
    {
      "epoch": 0.529851773668028,
      "grad_norm": 0.5010154843330383,
      "learning_rate": 9.38299728027673e-06,
      "loss": 0.4365,
      "step": 11546
    },
    {
      "epoch": 0.5298976641732825,
      "grad_norm": 0.48686492443084717,
      "learning_rate": 9.3828792855026e-06,
      "loss": 0.4615,
      "step": 11547
    },
    {
      "epoch": 0.529943554678537,
      "grad_norm": 0.47311222553253174,
      "learning_rate": 9.382761280189025e-06,
      "loss": 0.3837,
      "step": 11548
    },
    {
      "epoch": 0.5299894451837914,
      "grad_norm": 0.4158177971839905,
      "learning_rate": 9.382643264336296e-06,
      "loss": 0.3249,
      "step": 11549
    },
    {
      "epoch": 0.5300353356890459,
      "grad_norm": 0.4302652180194855,
      "learning_rate": 9.382525237944691e-06,
      "loss": 0.3538,
      "step": 11550
    },
    {
      "epoch": 0.5300812261943004,
      "grad_norm": 0.41319578886032104,
      "learning_rate": 9.382407201014494e-06,
      "loss": 0.3415,
      "step": 11551
    },
    {
      "epoch": 0.5301271166995548,
      "grad_norm": 0.45538896322250366,
      "learning_rate": 9.382289153545994e-06,
      "loss": 0.3952,
      "step": 11552
    },
    {
      "epoch": 0.5301730072048093,
      "grad_norm": 0.4859313368797302,
      "learning_rate": 9.382171095539469e-06,
      "loss": 0.3714,
      "step": 11553
    },
    {
      "epoch": 0.5302188977100638,
      "grad_norm": 0.4542468786239624,
      "learning_rate": 9.382053026995209e-06,
      "loss": 0.4064,
      "step": 11554
    },
    {
      "epoch": 0.5302647882153183,
      "grad_norm": 0.43858861923217773,
      "learning_rate": 9.381934947913491e-06,
      "loss": 0.378,
      "step": 11555
    },
    {
      "epoch": 0.5303106787205727,
      "grad_norm": 0.45160600543022156,
      "learning_rate": 9.381816858294605e-06,
      "loss": 0.3273,
      "step": 11556
    },
    {
      "epoch": 0.5303565692258272,
      "grad_norm": 0.422331839799881,
      "learning_rate": 9.381698758138832e-06,
      "loss": 0.3369,
      "step": 11557
    },
    {
      "epoch": 0.5304024597310817,
      "grad_norm": 0.45455700159072876,
      "learning_rate": 9.381580647446454e-06,
      "loss": 0.4062,
      "step": 11558
    },
    {
      "epoch": 0.5304483502363361,
      "grad_norm": 0.4425019919872284,
      "learning_rate": 9.381462526217759e-06,
      "loss": 0.3686,
      "step": 11559
    },
    {
      "epoch": 0.5304942407415906,
      "grad_norm": 0.4182943105697632,
      "learning_rate": 9.38134439445303e-06,
      "loss": 0.3367,
      "step": 11560
    },
    {
      "epoch": 0.5305401312468451,
      "grad_norm": 0.46657779812812805,
      "learning_rate": 9.38122625215255e-06,
      "loss": 0.4064,
      "step": 11561
    },
    {
      "epoch": 0.5305860217520995,
      "grad_norm": 0.42145976424217224,
      "learning_rate": 9.381108099316605e-06,
      "loss": 0.3286,
      "step": 11562
    },
    {
      "epoch": 0.530631912257354,
      "grad_norm": 0.46870410442352295,
      "learning_rate": 9.380989935945474e-06,
      "loss": 0.4069,
      "step": 11563
    },
    {
      "epoch": 0.5306778027626085,
      "grad_norm": 0.47562527656555176,
      "learning_rate": 9.380871762039448e-06,
      "loss": 0.4392,
      "step": 11564
    },
    {
      "epoch": 0.5307236932678628,
      "grad_norm": 0.4466591477394104,
      "learning_rate": 9.380753577598807e-06,
      "loss": 0.3744,
      "step": 11565
    },
    {
      "epoch": 0.5307695837731173,
      "grad_norm": 0.5316721200942993,
      "learning_rate": 9.380635382623837e-06,
      "loss": 0.4571,
      "step": 11566
    },
    {
      "epoch": 0.5308154742783718,
      "grad_norm": 0.448160320520401,
      "learning_rate": 9.380517177114822e-06,
      "loss": 0.359,
      "step": 11567
    },
    {
      "epoch": 0.5308613647836262,
      "grad_norm": 0.48585203289985657,
      "learning_rate": 9.380398961072044e-06,
      "loss": 0.4666,
      "step": 11568
    },
    {
      "epoch": 0.5309072552888807,
      "grad_norm": 0.4485619068145752,
      "learning_rate": 9.38028073449579e-06,
      "loss": 0.3932,
      "step": 11569
    },
    {
      "epoch": 0.5309531457941352,
      "grad_norm": 0.7244715690612793,
      "learning_rate": 9.380162497386344e-06,
      "loss": 0.4252,
      "step": 11570
    },
    {
      "epoch": 0.5309990362993896,
      "grad_norm": 0.41888168454170227,
      "learning_rate": 9.380044249743989e-06,
      "loss": 0.3383,
      "step": 11571
    },
    {
      "epoch": 0.5310449268046441,
      "grad_norm": 0.6482976078987122,
      "learning_rate": 9.37992599156901e-06,
      "loss": 0.368,
      "step": 11572
    },
    {
      "epoch": 0.5310908173098986,
      "grad_norm": 0.4620508849620819,
      "learning_rate": 9.37980772286169e-06,
      "loss": 0.4066,
      "step": 11573
    },
    {
      "epoch": 0.5311367078151531,
      "grad_norm": 0.42144301533699036,
      "learning_rate": 9.379689443622316e-06,
      "loss": 0.3567,
      "step": 11574
    },
    {
      "epoch": 0.5311825983204075,
      "grad_norm": 0.42783674597740173,
      "learning_rate": 9.37957115385117e-06,
      "loss": 0.3331,
      "step": 11575
    },
    {
      "epoch": 0.531228488825662,
      "grad_norm": 0.44144707918167114,
      "learning_rate": 9.379452853548539e-06,
      "loss": 0.3111,
      "step": 11576
    },
    {
      "epoch": 0.5312743793309165,
      "grad_norm": 0.48687034845352173,
      "learning_rate": 9.379334542714707e-06,
      "loss": 0.5027,
      "step": 11577
    },
    {
      "epoch": 0.5313202698361709,
      "grad_norm": 0.4576660394668579,
      "learning_rate": 9.379216221349955e-06,
      "loss": 0.3857,
      "step": 11578
    },
    {
      "epoch": 0.5313661603414254,
      "grad_norm": 0.4898691773414612,
      "learning_rate": 9.379097889454571e-06,
      "loss": 0.473,
      "step": 11579
    },
    {
      "epoch": 0.5314120508466799,
      "grad_norm": 0.45174071192741394,
      "learning_rate": 9.37897954702884e-06,
      "loss": 0.352,
      "step": 11580
    },
    {
      "epoch": 0.5314579413519342,
      "grad_norm": 0.42047473788261414,
      "learning_rate": 9.378861194073044e-06,
      "loss": 0.3278,
      "step": 11581
    },
    {
      "epoch": 0.5315038318571887,
      "grad_norm": 0.49685177206993103,
      "learning_rate": 9.378742830587468e-06,
      "loss": 0.3997,
      "step": 11582
    },
    {
      "epoch": 0.5315497223624432,
      "grad_norm": 0.43215611577033997,
      "learning_rate": 9.3786244565724e-06,
      "loss": 0.3179,
      "step": 11583
    },
    {
      "epoch": 0.5315956128676976,
      "grad_norm": 0.4613479971885681,
      "learning_rate": 9.378506072028119e-06,
      "loss": 0.3661,
      "step": 11584
    },
    {
      "epoch": 0.5316415033729521,
      "grad_norm": 0.46476954221725464,
      "learning_rate": 9.378387676954916e-06,
      "loss": 0.4189,
      "step": 11585
    },
    {
      "epoch": 0.5316873938782066,
      "grad_norm": 0.48116689920425415,
      "learning_rate": 9.37826927135307e-06,
      "loss": 0.4661,
      "step": 11586
    },
    {
      "epoch": 0.531733284383461,
      "grad_norm": 0.4635394513607025,
      "learning_rate": 9.37815085522287e-06,
      "loss": 0.4361,
      "step": 11587
    },
    {
      "epoch": 0.5317791748887155,
      "grad_norm": 0.4625721275806427,
      "learning_rate": 9.378032428564597e-06,
      "loss": 0.44,
      "step": 11588
    },
    {
      "epoch": 0.53182506539397,
      "grad_norm": 0.4525059759616852,
      "learning_rate": 9.37791399137854e-06,
      "loss": 0.4199,
      "step": 11589
    },
    {
      "epoch": 0.5318709558992244,
      "grad_norm": 0.43967950344085693,
      "learning_rate": 9.37779554366498e-06,
      "loss": 0.3436,
      "step": 11590
    },
    {
      "epoch": 0.5319168464044789,
      "grad_norm": 0.5150948166847229,
      "learning_rate": 9.377677085424203e-06,
      "loss": 0.4618,
      "step": 11591
    },
    {
      "epoch": 0.5319627369097334,
      "grad_norm": 0.47768378257751465,
      "learning_rate": 9.377558616656494e-06,
      "loss": 0.4452,
      "step": 11592
    },
    {
      "epoch": 0.5320086274149879,
      "grad_norm": 0.4843105673789978,
      "learning_rate": 9.377440137362141e-06,
      "loss": 0.4901,
      "step": 11593
    },
    {
      "epoch": 0.5320545179202423,
      "grad_norm": 0.4529247581958771,
      "learning_rate": 9.377321647541425e-06,
      "loss": 0.4015,
      "step": 11594
    },
    {
      "epoch": 0.5321004084254968,
      "grad_norm": 0.45344090461730957,
      "learning_rate": 9.37720314719463e-06,
      "loss": 0.4497,
      "step": 11595
    },
    {
      "epoch": 0.5321462989307513,
      "grad_norm": 0.43733781576156616,
      "learning_rate": 9.377084636322045e-06,
      "loss": 0.3634,
      "step": 11596
    },
    {
      "epoch": 0.5321921894360057,
      "grad_norm": 0.4152829349040985,
      "learning_rate": 9.376966114923952e-06,
      "loss": 0.3617,
      "step": 11597
    },
    {
      "epoch": 0.5322380799412602,
      "grad_norm": 0.4740702211856842,
      "learning_rate": 9.376847583000636e-06,
      "loss": 0.4416,
      "step": 11598
    },
    {
      "epoch": 0.5322839704465147,
      "grad_norm": 0.5115492939949036,
      "learning_rate": 9.376729040552384e-06,
      "loss": 0.4452,
      "step": 11599
    },
    {
      "epoch": 0.532329860951769,
      "grad_norm": 0.4531647861003876,
      "learning_rate": 9.37661048757948e-06,
      "loss": 0.4305,
      "step": 11600
    },
    {
      "epoch": 0.5323757514570235,
      "grad_norm": 0.4669593870639801,
      "learning_rate": 9.376491924082208e-06,
      "loss": 0.4468,
      "step": 11601
    },
    {
      "epoch": 0.532421641962278,
      "grad_norm": 0.4381949305534363,
      "learning_rate": 9.376373350060855e-06,
      "loss": 0.3488,
      "step": 11602
    },
    {
      "epoch": 0.5324675324675324,
      "grad_norm": 0.4604457914829254,
      "learning_rate": 9.376254765515707e-06,
      "loss": 0.3846,
      "step": 11603
    },
    {
      "epoch": 0.5325134229727869,
      "grad_norm": 0.4993179142475128,
      "learning_rate": 9.376136170447046e-06,
      "loss": 0.3595,
      "step": 11604
    },
    {
      "epoch": 0.5325593134780414,
      "grad_norm": 0.4939204454421997,
      "learning_rate": 9.376017564855157e-06,
      "loss": 0.4765,
      "step": 11605
    },
    {
      "epoch": 0.5326052039832958,
      "grad_norm": 0.4435739517211914,
      "learning_rate": 9.375898948740329e-06,
      "loss": 0.3855,
      "step": 11606
    },
    {
      "epoch": 0.5326510944885503,
      "grad_norm": 0.49379995465278625,
      "learning_rate": 9.375780322102845e-06,
      "loss": 0.4484,
      "step": 11607
    },
    {
      "epoch": 0.5326969849938048,
      "grad_norm": 0.4521661698818207,
      "learning_rate": 9.37566168494299e-06,
      "loss": 0.3592,
      "step": 11608
    },
    {
      "epoch": 0.5327428754990593,
      "grad_norm": 0.4651557207107544,
      "learning_rate": 9.375543037261048e-06,
      "loss": 0.3696,
      "step": 11609
    },
    {
      "epoch": 0.5327887660043137,
      "grad_norm": 0.46903112530708313,
      "learning_rate": 9.375424379057307e-06,
      "loss": 0.4092,
      "step": 11610
    },
    {
      "epoch": 0.5328346565095682,
      "grad_norm": 0.4719286561012268,
      "learning_rate": 9.375305710332052e-06,
      "loss": 0.4762,
      "step": 11611
    },
    {
      "epoch": 0.5328805470148227,
      "grad_norm": 0.491604745388031,
      "learning_rate": 9.375187031085568e-06,
      "loss": 0.4428,
      "step": 11612
    },
    {
      "epoch": 0.5329264375200771,
      "grad_norm": 0.4755484163761139,
      "learning_rate": 9.37506834131814e-06,
      "loss": 0.4658,
      "step": 11613
    },
    {
      "epoch": 0.5329723280253316,
      "grad_norm": 0.43989041447639465,
      "learning_rate": 9.374949641030053e-06,
      "loss": 0.3405,
      "step": 11614
    },
    {
      "epoch": 0.5330182185305861,
      "grad_norm": 0.434423565864563,
      "learning_rate": 9.374830930221592e-06,
      "loss": 0.3818,
      "step": 11615
    },
    {
      "epoch": 0.5330641090358404,
      "grad_norm": 0.47206324338912964,
      "learning_rate": 9.374712208893044e-06,
      "loss": 0.4204,
      "step": 11616
    },
    {
      "epoch": 0.533109999541095,
      "grad_norm": 0.4455054700374603,
      "learning_rate": 9.374593477044694e-06,
      "loss": 0.3458,
      "step": 11617
    },
    {
      "epoch": 0.5331558900463494,
      "grad_norm": 0.45316562056541443,
      "learning_rate": 9.374474734676829e-06,
      "loss": 0.3834,
      "step": 11618
    },
    {
      "epoch": 0.5332017805516038,
      "grad_norm": 0.4805208444595337,
      "learning_rate": 9.37435598178973e-06,
      "loss": 0.4594,
      "step": 11619
    },
    {
      "epoch": 0.5332476710568583,
      "grad_norm": 0.44822531938552856,
      "learning_rate": 9.374237218383687e-06,
      "loss": 0.4097,
      "step": 11620
    },
    {
      "epoch": 0.5332935615621128,
      "grad_norm": 0.4362548291683197,
      "learning_rate": 9.374118444458984e-06,
      "loss": 0.3443,
      "step": 11621
    },
    {
      "epoch": 0.5333394520673672,
      "grad_norm": 0.5092342495918274,
      "learning_rate": 9.373999660015905e-06,
      "loss": 0.5684,
      "step": 11622
    },
    {
      "epoch": 0.5333853425726217,
      "grad_norm": 0.42923811078071594,
      "learning_rate": 9.37388086505474e-06,
      "loss": 0.3465,
      "step": 11623
    },
    {
      "epoch": 0.5334312330778762,
      "grad_norm": 0.44357872009277344,
      "learning_rate": 9.37376205957577e-06,
      "loss": 0.3662,
      "step": 11624
    },
    {
      "epoch": 0.5334771235831306,
      "grad_norm": 0.43685269355773926,
      "learning_rate": 9.373643243579283e-06,
      "loss": 0.39,
      "step": 11625
    },
    {
      "epoch": 0.5335230140883851,
      "grad_norm": 0.47921207547187805,
      "learning_rate": 9.373524417065566e-06,
      "loss": 0.4298,
      "step": 11626
    },
    {
      "epoch": 0.5335689045936396,
      "grad_norm": 0.48560789227485657,
      "learning_rate": 9.3734055800349e-06,
      "loss": 0.4327,
      "step": 11627
    },
    {
      "epoch": 0.5336147950988941,
      "grad_norm": 0.4798007011413574,
      "learning_rate": 9.373286732487576e-06,
      "loss": 0.3875,
      "step": 11628
    },
    {
      "epoch": 0.5336606856041485,
      "grad_norm": 0.4492725431919098,
      "learning_rate": 9.373167874423877e-06,
      "loss": 0.3537,
      "step": 11629
    },
    {
      "epoch": 0.533706576109403,
      "grad_norm": 0.4310928285121918,
      "learning_rate": 9.373049005844091e-06,
      "loss": 0.3474,
      "step": 11630
    },
    {
      "epoch": 0.5337524666146575,
      "grad_norm": 0.4532617926597595,
      "learning_rate": 9.372930126748501e-06,
      "loss": 0.3925,
      "step": 11631
    },
    {
      "epoch": 0.5337983571199119,
      "grad_norm": 0.4393414556980133,
      "learning_rate": 9.372811237137393e-06,
      "loss": 0.3754,
      "step": 11632
    },
    {
      "epoch": 0.5338442476251664,
      "grad_norm": 0.4893622398376465,
      "learning_rate": 9.372692337011056e-06,
      "loss": 0.4213,
      "step": 11633
    },
    {
      "epoch": 0.5338901381304209,
      "grad_norm": 0.46111804246902466,
      "learning_rate": 9.372573426369774e-06,
      "loss": 0.4179,
      "step": 11634
    },
    {
      "epoch": 0.5339360286356752,
      "grad_norm": 0.46457934379577637,
      "learning_rate": 9.372454505213831e-06,
      "loss": 0.3934,
      "step": 11635
    },
    {
      "epoch": 0.5339819191409297,
      "grad_norm": 0.46537649631500244,
      "learning_rate": 9.372335573543516e-06,
      "loss": 0.4486,
      "step": 11636
    },
    {
      "epoch": 0.5340278096461842,
      "grad_norm": 0.4430766701698303,
      "learning_rate": 9.372216631359115e-06,
      "loss": 0.3266,
      "step": 11637
    },
    {
      "epoch": 0.5340737001514386,
      "grad_norm": 0.44635188579559326,
      "learning_rate": 9.37209767866091e-06,
      "loss": 0.3726,
      "step": 11638
    },
    {
      "epoch": 0.5341195906566931,
      "grad_norm": 0.48563629388809204,
      "learning_rate": 9.371978715449192e-06,
      "loss": 0.4021,
      "step": 11639
    },
    {
      "epoch": 0.5341654811619476,
      "grad_norm": 0.4986581802368164,
      "learning_rate": 9.371859741724245e-06,
      "loss": 0.5052,
      "step": 11640
    },
    {
      "epoch": 0.534211371667202,
      "grad_norm": 0.4532596468925476,
      "learning_rate": 9.371740757486357e-06,
      "loss": 0.3847,
      "step": 11641
    },
    {
      "epoch": 0.5342572621724565,
      "grad_norm": 0.45764607191085815,
      "learning_rate": 9.37162176273581e-06,
      "loss": 0.4517,
      "step": 11642
    },
    {
      "epoch": 0.534303152677711,
      "grad_norm": 0.4384588301181793,
      "learning_rate": 9.371502757472891e-06,
      "loss": 0.3441,
      "step": 11643
    },
    {
      "epoch": 0.5343490431829655,
      "grad_norm": 0.4472794532775879,
      "learning_rate": 9.37138374169789e-06,
      "loss": 0.4206,
      "step": 11644
    },
    {
      "epoch": 0.5343949336882199,
      "grad_norm": 0.438943475484848,
      "learning_rate": 9.371264715411091e-06,
      "loss": 0.3478,
      "step": 11645
    },
    {
      "epoch": 0.5344408241934744,
      "grad_norm": 0.4645429849624634,
      "learning_rate": 9.371145678612777e-06,
      "loss": 0.4307,
      "step": 11646
    },
    {
      "epoch": 0.5344867146987289,
      "grad_norm": 0.4793485999107361,
      "learning_rate": 9.371026631303239e-06,
      "loss": 0.3829,
      "step": 11647
    },
    {
      "epoch": 0.5345326052039833,
      "grad_norm": 0.44711020588874817,
      "learning_rate": 9.370907573482762e-06,
      "loss": 0.4163,
      "step": 11648
    },
    {
      "epoch": 0.5345784957092378,
      "grad_norm": 0.4245031476020813,
      "learning_rate": 9.370788505151631e-06,
      "loss": 0.3568,
      "step": 11649
    },
    {
      "epoch": 0.5346243862144923,
      "grad_norm": 0.48311084508895874,
      "learning_rate": 9.370669426310134e-06,
      "loss": 0.4932,
      "step": 11650
    },
    {
      "epoch": 0.5346702767197467,
      "grad_norm": 0.46771153807640076,
      "learning_rate": 9.370550336958556e-06,
      "loss": 0.4558,
      "step": 11651
    },
    {
      "epoch": 0.5347161672250011,
      "grad_norm": 0.4367324113845825,
      "learning_rate": 9.370431237097184e-06,
      "loss": 0.3342,
      "step": 11652
    },
    {
      "epoch": 0.5347620577302556,
      "grad_norm": 0.4495399296283722,
      "learning_rate": 9.370312126726304e-06,
      "loss": 0.3618,
      "step": 11653
    },
    {
      "epoch": 0.53480794823551,
      "grad_norm": 0.41776782274246216,
      "learning_rate": 9.370193005846202e-06,
      "loss": 0.3108,
      "step": 11654
    },
    {
      "epoch": 0.5348538387407645,
      "grad_norm": 0.44117724895477295,
      "learning_rate": 9.370073874457167e-06,
      "loss": 0.3651,
      "step": 11655
    },
    {
      "epoch": 0.534899729246019,
      "grad_norm": 0.5484052896499634,
      "learning_rate": 9.369954732559481e-06,
      "loss": 0.3278,
      "step": 11656
    },
    {
      "epoch": 0.5349456197512734,
      "grad_norm": 0.4948674738407135,
      "learning_rate": 9.369835580153435e-06,
      "loss": 0.477,
      "step": 11657
    },
    {
      "epoch": 0.5349915102565279,
      "grad_norm": 0.4251708388328552,
      "learning_rate": 9.369716417239312e-06,
      "loss": 0.3038,
      "step": 11658
    },
    {
      "epoch": 0.5350374007617824,
      "grad_norm": 0.611009955406189,
      "learning_rate": 9.369597243817402e-06,
      "loss": 0.4037,
      "step": 11659
    },
    {
      "epoch": 0.5350832912670368,
      "grad_norm": 0.46289777755737305,
      "learning_rate": 9.369478059887988e-06,
      "loss": 0.398,
      "step": 11660
    },
    {
      "epoch": 0.5351291817722913,
      "grad_norm": 0.4309151768684387,
      "learning_rate": 9.369358865451359e-06,
      "loss": 0.3516,
      "step": 11661
    },
    {
      "epoch": 0.5351750722775458,
      "grad_norm": 0.4830739200115204,
      "learning_rate": 9.369239660507802e-06,
      "loss": 0.4298,
      "step": 11662
    },
    {
      "epoch": 0.5352209627828003,
      "grad_norm": 0.45701953768730164,
      "learning_rate": 9.369120445057602e-06,
      "loss": 0.3928,
      "step": 11663
    },
    {
      "epoch": 0.5352668532880547,
      "grad_norm": 0.501124382019043,
      "learning_rate": 9.369001219101046e-06,
      "loss": 0.4657,
      "step": 11664
    },
    {
      "epoch": 0.5353127437933092,
      "grad_norm": 0.45428067445755005,
      "learning_rate": 9.36888198263842e-06,
      "loss": 0.457,
      "step": 11665
    },
    {
      "epoch": 0.5353586342985637,
      "grad_norm": 0.416500985622406,
      "learning_rate": 9.368762735670012e-06,
      "loss": 0.3114,
      "step": 11666
    },
    {
      "epoch": 0.5354045248038181,
      "grad_norm": 0.48178282380104065,
      "learning_rate": 9.36864347819611e-06,
      "loss": 0.44,
      "step": 11667
    },
    {
      "epoch": 0.5354504153090726,
      "grad_norm": 0.4471374452114105,
      "learning_rate": 9.368524210216999e-06,
      "loss": 0.4024,
      "step": 11668
    },
    {
      "epoch": 0.5354963058143271,
      "grad_norm": 0.46065840125083923,
      "learning_rate": 9.368404931732963e-06,
      "loss": 0.4109,
      "step": 11669
    },
    {
      "epoch": 0.5355421963195814,
      "grad_norm": 0.42254650592803955,
      "learning_rate": 9.368285642744295e-06,
      "loss": 0.3375,
      "step": 11670
    },
    {
      "epoch": 0.5355880868248359,
      "grad_norm": 0.42463284730911255,
      "learning_rate": 9.368166343251277e-06,
      "loss": 0.3402,
      "step": 11671
    },
    {
      "epoch": 0.5356339773300904,
      "grad_norm": 0.4263712167739868,
      "learning_rate": 9.3680470332542e-06,
      "loss": 0.3509,
      "step": 11672
    },
    {
      "epoch": 0.5356798678353448,
      "grad_norm": 0.4777979552745819,
      "learning_rate": 9.367927712753347e-06,
      "loss": 0.3815,
      "step": 11673
    },
    {
      "epoch": 0.5357257583405993,
      "grad_norm": 0.4567091464996338,
      "learning_rate": 9.367808381749005e-06,
      "loss": 0.3422,
      "step": 11674
    },
    {
      "epoch": 0.5357716488458538,
      "grad_norm": 0.45422235131263733,
      "learning_rate": 9.367689040241465e-06,
      "loss": 0.4111,
      "step": 11675
    },
    {
      "epoch": 0.5358175393511082,
      "grad_norm": 0.4887014329433441,
      "learning_rate": 9.367569688231009e-06,
      "loss": 0.4326,
      "step": 11676
    },
    {
      "epoch": 0.5358634298563627,
      "grad_norm": 0.4648697078227997,
      "learning_rate": 9.367450325717927e-06,
      "loss": 0.3618,
      "step": 11677
    },
    {
      "epoch": 0.5359093203616172,
      "grad_norm": 0.42911064624786377,
      "learning_rate": 9.367330952702508e-06,
      "loss": 0.3263,
      "step": 11678
    },
    {
      "epoch": 0.5359552108668716,
      "grad_norm": 0.453212171792984,
      "learning_rate": 9.367211569185033e-06,
      "loss": 0.3818,
      "step": 11679
    },
    {
      "epoch": 0.5360011013721261,
      "grad_norm": 0.4420299828052521,
      "learning_rate": 9.367092175165792e-06,
      "loss": 0.3797,
      "step": 11680
    },
    {
      "epoch": 0.5360469918773806,
      "grad_norm": 0.48142096400260925,
      "learning_rate": 9.366972770645075e-06,
      "loss": 0.4953,
      "step": 11681
    },
    {
      "epoch": 0.5360928823826351,
      "grad_norm": 0.43701648712158203,
      "learning_rate": 9.366853355623166e-06,
      "loss": 0.4113,
      "step": 11682
    },
    {
      "epoch": 0.5361387728878895,
      "grad_norm": 0.46006593108177185,
      "learning_rate": 9.366733930100354e-06,
      "loss": 0.3981,
      "step": 11683
    },
    {
      "epoch": 0.536184663393144,
      "grad_norm": 0.459682822227478,
      "learning_rate": 9.366614494076923e-06,
      "loss": 0.3574,
      "step": 11684
    },
    {
      "epoch": 0.5362305538983985,
      "grad_norm": 0.42785394191741943,
      "learning_rate": 9.366495047553164e-06,
      "loss": 0.3232,
      "step": 11685
    },
    {
      "epoch": 0.5362764444036529,
      "grad_norm": 0.43644100427627563,
      "learning_rate": 9.36637559052936e-06,
      "loss": 0.3942,
      "step": 11686
    },
    {
      "epoch": 0.5363223349089073,
      "grad_norm": 0.45794540643692017,
      "learning_rate": 9.366256123005803e-06,
      "loss": 0.3359,
      "step": 11687
    },
    {
      "epoch": 0.5363682254141618,
      "grad_norm": 0.4445849359035492,
      "learning_rate": 9.366136644982777e-06,
      "loss": 0.3921,
      "step": 11688
    },
    {
      "epoch": 0.5364141159194162,
      "grad_norm": 0.4533580243587494,
      "learning_rate": 9.366017156460571e-06,
      "loss": 0.4263,
      "step": 11689
    },
    {
      "epoch": 0.5364600064246707,
      "grad_norm": 0.4564112424850464,
      "learning_rate": 9.365897657439474e-06,
      "loss": 0.3519,
      "step": 11690
    },
    {
      "epoch": 0.5365058969299252,
      "grad_norm": 0.5009704232215881,
      "learning_rate": 9.365778147919767e-06,
      "loss": 0.4842,
      "step": 11691
    },
    {
      "epoch": 0.5365517874351796,
      "grad_norm": 0.464406281709671,
      "learning_rate": 9.365658627901744e-06,
      "loss": 0.3908,
      "step": 11692
    },
    {
      "epoch": 0.5365976779404341,
      "grad_norm": 0.5183472037315369,
      "learning_rate": 9.365539097385687e-06,
      "loss": 0.5698,
      "step": 11693
    },
    {
      "epoch": 0.5366435684456886,
      "grad_norm": 0.4381266236305237,
      "learning_rate": 9.36541955637189e-06,
      "loss": 0.2997,
      "step": 11694
    },
    {
      "epoch": 0.536689458950943,
      "grad_norm": 0.4728393256664276,
      "learning_rate": 9.365300004860634e-06,
      "loss": 0.4102,
      "step": 11695
    },
    {
      "epoch": 0.5367353494561975,
      "grad_norm": 0.49420151114463806,
      "learning_rate": 9.365180442852211e-06,
      "loss": 0.4696,
      "step": 11696
    },
    {
      "epoch": 0.536781239961452,
      "grad_norm": 0.4743998646736145,
      "learning_rate": 9.365060870346906e-06,
      "loss": 0.4561,
      "step": 11697
    },
    {
      "epoch": 0.5368271304667065,
      "grad_norm": 0.4423024356365204,
      "learning_rate": 9.364941287345007e-06,
      "loss": 0.4031,
      "step": 11698
    },
    {
      "epoch": 0.5368730209719609,
      "grad_norm": 0.42873337864875793,
      "learning_rate": 9.364821693846801e-06,
      "loss": 0.3322,
      "step": 11699
    },
    {
      "epoch": 0.5369189114772154,
      "grad_norm": 0.44599631428718567,
      "learning_rate": 9.364702089852578e-06,
      "loss": 0.4245,
      "step": 11700
    },
    {
      "epoch": 0.5369648019824699,
      "grad_norm": 0.46549248695373535,
      "learning_rate": 9.364582475362623e-06,
      "loss": 0.4024,
      "step": 11701
    },
    {
      "epoch": 0.5370106924877243,
      "grad_norm": 0.4442719519138336,
      "learning_rate": 9.364462850377226e-06,
      "loss": 0.3646,
      "step": 11702
    },
    {
      "epoch": 0.5370565829929788,
      "grad_norm": 0.4664672315120697,
      "learning_rate": 9.364343214896672e-06,
      "loss": 0.3519,
      "step": 11703
    },
    {
      "epoch": 0.5371024734982333,
      "grad_norm": 0.4892199635505676,
      "learning_rate": 9.36422356892125e-06,
      "loss": 0.4619,
      "step": 11704
    },
    {
      "epoch": 0.5371483640034876,
      "grad_norm": 0.48552682995796204,
      "learning_rate": 9.364103912451247e-06,
      "loss": 0.4872,
      "step": 11705
    },
    {
      "epoch": 0.5371942545087421,
      "grad_norm": 0.5217673182487488,
      "learning_rate": 9.363984245486955e-06,
      "loss": 0.5127,
      "step": 11706
    },
    {
      "epoch": 0.5372401450139966,
      "grad_norm": 0.44143733382225037,
      "learning_rate": 9.363864568028655e-06,
      "loss": 0.4118,
      "step": 11707
    },
    {
      "epoch": 0.537286035519251,
      "grad_norm": 0.4720189571380615,
      "learning_rate": 9.363744880076638e-06,
      "loss": 0.4746,
      "step": 11708
    },
    {
      "epoch": 0.5373319260245055,
      "grad_norm": 0.4795798659324646,
      "learning_rate": 9.363625181631192e-06,
      "loss": 0.4008,
      "step": 11709
    },
    {
      "epoch": 0.53737781652976,
      "grad_norm": 0.477272093296051,
      "learning_rate": 9.363505472692606e-06,
      "loss": 0.3761,
      "step": 11710
    },
    {
      "epoch": 0.5374237070350144,
      "grad_norm": 0.47043347358703613,
      "learning_rate": 9.363385753261165e-06,
      "loss": 0.3995,
      "step": 11711
    },
    {
      "epoch": 0.5374695975402689,
      "grad_norm": 0.4559633433818817,
      "learning_rate": 9.363266023337158e-06,
      "loss": 0.3946,
      "step": 11712
    },
    {
      "epoch": 0.5375154880455234,
      "grad_norm": 0.49843499064445496,
      "learning_rate": 9.363146282920876e-06,
      "loss": 0.5041,
      "step": 11713
    },
    {
      "epoch": 0.5375613785507778,
      "grad_norm": 0.4743489921092987,
      "learning_rate": 9.363026532012603e-06,
      "loss": 0.436,
      "step": 11714
    },
    {
      "epoch": 0.5376072690560323,
      "grad_norm": 0.4795120060443878,
      "learning_rate": 9.362906770612628e-06,
      "loss": 0.4156,
      "step": 11715
    },
    {
      "epoch": 0.5376531595612868,
      "grad_norm": 0.4943495988845825,
      "learning_rate": 9.362786998721238e-06,
      "loss": 0.459,
      "step": 11716
    },
    {
      "epoch": 0.5376990500665413,
      "grad_norm": 0.44998136162757874,
      "learning_rate": 9.362667216338724e-06,
      "loss": 0.361,
      "step": 11717
    },
    {
      "epoch": 0.5377449405717957,
      "grad_norm": 0.4819561243057251,
      "learning_rate": 9.362547423465371e-06,
      "loss": 0.4879,
      "step": 11718
    },
    {
      "epoch": 0.5377908310770502,
      "grad_norm": 0.4418932795524597,
      "learning_rate": 9.362427620101468e-06,
      "loss": 0.3833,
      "step": 11719
    },
    {
      "epoch": 0.5378367215823047,
      "grad_norm": 0.46308183670043945,
      "learning_rate": 9.362307806247304e-06,
      "loss": 0.4314,
      "step": 11720
    },
    {
      "epoch": 0.537882612087559,
      "grad_norm": 0.5953032374382019,
      "learning_rate": 9.362187981903167e-06,
      "loss": 0.4444,
      "step": 11721
    },
    {
      "epoch": 0.5379285025928136,
      "grad_norm": 0.42617878317832947,
      "learning_rate": 9.362068147069343e-06,
      "loss": 0.3516,
      "step": 11722
    },
    {
      "epoch": 0.537974393098068,
      "grad_norm": 0.4837894141674042,
      "learning_rate": 9.361948301746125e-06,
      "loss": 0.4853,
      "step": 11723
    },
    {
      "epoch": 0.5380202836033224,
      "grad_norm": 0.47321876883506775,
      "learning_rate": 9.361828445933796e-06,
      "loss": 0.4159,
      "step": 11724
    },
    {
      "epoch": 0.5380661741085769,
      "grad_norm": 0.5161371231079102,
      "learning_rate": 9.361708579632646e-06,
      "loss": 0.3919,
      "step": 11725
    },
    {
      "epoch": 0.5381120646138314,
      "grad_norm": 0.572919487953186,
      "learning_rate": 9.361588702842963e-06,
      "loss": 0.53,
      "step": 11726
    },
    {
      "epoch": 0.5381579551190858,
      "grad_norm": 0.44206735491752625,
      "learning_rate": 9.361468815565037e-06,
      "loss": 0.342,
      "step": 11727
    },
    {
      "epoch": 0.5382038456243403,
      "grad_norm": 0.44542205333709717,
      "learning_rate": 9.361348917799154e-06,
      "loss": 0.3824,
      "step": 11728
    },
    {
      "epoch": 0.5382497361295948,
      "grad_norm": 0.47720059752464294,
      "learning_rate": 9.361229009545604e-06,
      "loss": 0.4126,
      "step": 11729
    },
    {
      "epoch": 0.5382956266348492,
      "grad_norm": 0.45526832342147827,
      "learning_rate": 9.361109090804673e-06,
      "loss": 0.3634,
      "step": 11730
    },
    {
      "epoch": 0.5383415171401037,
      "grad_norm": 0.47212526202201843,
      "learning_rate": 9.360989161576655e-06,
      "loss": 0.4119,
      "step": 11731
    },
    {
      "epoch": 0.5383874076453582,
      "grad_norm": 0.48589789867401123,
      "learning_rate": 9.360869221861831e-06,
      "loss": 0.4123,
      "step": 11732
    },
    {
      "epoch": 0.5384332981506127,
      "grad_norm": 0.4738694131374359,
      "learning_rate": 9.360749271660494e-06,
      "loss": 0.473,
      "step": 11733
    },
    {
      "epoch": 0.5384791886558671,
      "grad_norm": 0.46446332335472107,
      "learning_rate": 9.360629310972931e-06,
      "loss": 0.4203,
      "step": 11734
    },
    {
      "epoch": 0.5385250791611216,
      "grad_norm": 0.46781182289123535,
      "learning_rate": 9.360509339799431e-06,
      "loss": 0.4827,
      "step": 11735
    },
    {
      "epoch": 0.5385709696663761,
      "grad_norm": 0.4557516276836395,
      "learning_rate": 9.360389358140283e-06,
      "loss": 0.3776,
      "step": 11736
    },
    {
      "epoch": 0.5386168601716305,
      "grad_norm": 0.45764628052711487,
      "learning_rate": 9.360269365995773e-06,
      "loss": 0.4002,
      "step": 11737
    },
    {
      "epoch": 0.538662750676885,
      "grad_norm": 0.4630337357521057,
      "learning_rate": 9.360149363366194e-06,
      "loss": 0.3938,
      "step": 11738
    },
    {
      "epoch": 0.5387086411821395,
      "grad_norm": 0.4393448233604431,
      "learning_rate": 9.36002935025183e-06,
      "loss": 0.3385,
      "step": 11739
    },
    {
      "epoch": 0.5387545316873938,
      "grad_norm": 0.4994428753852844,
      "learning_rate": 9.35990932665297e-06,
      "loss": 0.4288,
      "step": 11740
    },
    {
      "epoch": 0.5388004221926483,
      "grad_norm": 0.49707406759262085,
      "learning_rate": 9.359789292569904e-06,
      "loss": 0.5072,
      "step": 11741
    },
    {
      "epoch": 0.5388463126979028,
      "grad_norm": 0.46496033668518066,
      "learning_rate": 9.359669248002925e-06,
      "loss": 0.3754,
      "step": 11742
    },
    {
      "epoch": 0.5388922032031572,
      "grad_norm": 0.4182380139827728,
      "learning_rate": 9.359549192952315e-06,
      "loss": 0.3434,
      "step": 11743
    },
    {
      "epoch": 0.5389380937084117,
      "grad_norm": 0.4334733188152313,
      "learning_rate": 9.359429127418365e-06,
      "loss": 0.4045,
      "step": 11744
    },
    {
      "epoch": 0.5389839842136662,
      "grad_norm": 0.46384796500205994,
      "learning_rate": 9.359309051401363e-06,
      "loss": 0.414,
      "step": 11745
    },
    {
      "epoch": 0.5390298747189206,
      "grad_norm": 0.43957823514938354,
      "learning_rate": 9.359188964901599e-06,
      "loss": 0.3752,
      "step": 11746
    },
    {
      "epoch": 0.5390757652241751,
      "grad_norm": 0.4827621877193451,
      "learning_rate": 9.359068867919363e-06,
      "loss": 0.4383,
      "step": 11747
    },
    {
      "epoch": 0.5391216557294296,
      "grad_norm": 0.49911588430404663,
      "learning_rate": 9.35894876045494e-06,
      "loss": 0.485,
      "step": 11748
    },
    {
      "epoch": 0.539167546234684,
      "grad_norm": 0.4338134229183197,
      "learning_rate": 9.358828642508622e-06,
      "loss": 0.3856,
      "step": 11749
    },
    {
      "epoch": 0.5392134367399385,
      "grad_norm": 0.5002368688583374,
      "learning_rate": 9.358708514080697e-06,
      "loss": 0.4655,
      "step": 11750
    },
    {
      "epoch": 0.539259327245193,
      "grad_norm": 0.4632793962955475,
      "learning_rate": 9.358588375171453e-06,
      "loss": 0.4596,
      "step": 11751
    },
    {
      "epoch": 0.5393052177504475,
      "grad_norm": 0.4487612843513489,
      "learning_rate": 9.358468225781178e-06,
      "loss": 0.3793,
      "step": 11752
    },
    {
      "epoch": 0.5393511082557019,
      "grad_norm": 0.4145508408546448,
      "learning_rate": 9.358348065910166e-06,
      "loss": 0.2976,
      "step": 11753
    },
    {
      "epoch": 0.5393969987609564,
      "grad_norm": 0.5098912119865417,
      "learning_rate": 9.358227895558701e-06,
      "loss": 0.4267,
      "step": 11754
    },
    {
      "epoch": 0.5394428892662109,
      "grad_norm": 0.4779066741466522,
      "learning_rate": 9.358107714727072e-06,
      "loss": 0.4517,
      "step": 11755
    },
    {
      "epoch": 0.5394887797714653,
      "grad_norm": 0.47584226727485657,
      "learning_rate": 9.35798752341557e-06,
      "loss": 0.4246,
      "step": 11756
    },
    {
      "epoch": 0.5395346702767198,
      "grad_norm": 0.7093757390975952,
      "learning_rate": 9.357867321624485e-06,
      "loss": 0.5258,
      "step": 11757
    },
    {
      "epoch": 0.5395805607819742,
      "grad_norm": 0.5028489828109741,
      "learning_rate": 9.357747109354102e-06,
      "loss": 0.4544,
      "step": 11758
    },
    {
      "epoch": 0.5396264512872286,
      "grad_norm": 0.4866946339607239,
      "learning_rate": 9.357626886604715e-06,
      "loss": 0.3878,
      "step": 11759
    },
    {
      "epoch": 0.5396723417924831,
      "grad_norm": 0.47552040219306946,
      "learning_rate": 9.357506653376609e-06,
      "loss": 0.439,
      "step": 11760
    },
    {
      "epoch": 0.5397182322977376,
      "grad_norm": 0.4580599367618561,
      "learning_rate": 9.357386409670076e-06,
      "loss": 0.4168,
      "step": 11761
    },
    {
      "epoch": 0.539764122802992,
      "grad_norm": 0.4799123704433441,
      "learning_rate": 9.357266155485404e-06,
      "loss": 0.4385,
      "step": 11762
    },
    {
      "epoch": 0.5398100133082465,
      "grad_norm": 0.4635271728038788,
      "learning_rate": 9.35714589082288e-06,
      "loss": 0.4093,
      "step": 11763
    },
    {
      "epoch": 0.539855903813501,
      "grad_norm": 0.4889325797557831,
      "learning_rate": 9.357025615682798e-06,
      "loss": 0.3931,
      "step": 11764
    },
    {
      "epoch": 0.5399017943187554,
      "grad_norm": 0.49372759461402893,
      "learning_rate": 9.356905330065442e-06,
      "loss": 0.3893,
      "step": 11765
    },
    {
      "epoch": 0.5399476848240099,
      "grad_norm": 0.468430757522583,
      "learning_rate": 9.356785033971106e-06,
      "loss": 0.3458,
      "step": 11766
    },
    {
      "epoch": 0.5399935753292644,
      "grad_norm": 0.46209850907325745,
      "learning_rate": 9.356664727400075e-06,
      "loss": 0.3918,
      "step": 11767
    },
    {
      "epoch": 0.5400394658345188,
      "grad_norm": 0.4556235373020172,
      "learning_rate": 9.356544410352641e-06,
      "loss": 0.3714,
      "step": 11768
    },
    {
      "epoch": 0.5400853563397733,
      "grad_norm": 0.4316776692867279,
      "learning_rate": 9.356424082829093e-06,
      "loss": 0.2993,
      "step": 11769
    },
    {
      "epoch": 0.5401312468450278,
      "grad_norm": 0.508688747882843,
      "learning_rate": 9.356303744829722e-06,
      "loss": 0.5229,
      "step": 11770
    },
    {
      "epoch": 0.5401771373502823,
      "grad_norm": 0.4347439110279083,
      "learning_rate": 9.356183396354813e-06,
      "loss": 0.3734,
      "step": 11771
    },
    {
      "epoch": 0.5402230278555367,
      "grad_norm": 0.46213340759277344,
      "learning_rate": 9.356063037404657e-06,
      "loss": 0.4129,
      "step": 11772
    },
    {
      "epoch": 0.5402689183607912,
      "grad_norm": 0.4284520447254181,
      "learning_rate": 9.355942667979547e-06,
      "loss": 0.3263,
      "step": 11773
    },
    {
      "epoch": 0.5403148088660457,
      "grad_norm": 0.5183601975440979,
      "learning_rate": 9.355822288079767e-06,
      "loss": 0.3736,
      "step": 11774
    },
    {
      "epoch": 0.5403606993713,
      "grad_norm": 0.45145323872566223,
      "learning_rate": 9.35570189770561e-06,
      "loss": 0.3761,
      "step": 11775
    },
    {
      "epoch": 0.5404065898765545,
      "grad_norm": 0.4858222007751465,
      "learning_rate": 9.355581496857364e-06,
      "loss": 0.4552,
      "step": 11776
    },
    {
      "epoch": 0.540452480381809,
      "grad_norm": 0.47047221660614014,
      "learning_rate": 9.35546108553532e-06,
      "loss": 0.4106,
      "step": 11777
    },
    {
      "epoch": 0.5404983708870634,
      "grad_norm": 0.45213961601257324,
      "learning_rate": 9.355340663739767e-06,
      "loss": 0.4236,
      "step": 11778
    },
    {
      "epoch": 0.5405442613923179,
      "grad_norm": 0.4616587460041046,
      "learning_rate": 9.355220231470994e-06,
      "loss": 0.4735,
      "step": 11779
    },
    {
      "epoch": 0.5405901518975724,
      "grad_norm": 0.47109919786453247,
      "learning_rate": 9.355099788729291e-06,
      "loss": 0.3975,
      "step": 11780
    },
    {
      "epoch": 0.5406360424028268,
      "grad_norm": 0.4649921655654907,
      "learning_rate": 9.354979335514948e-06,
      "loss": 0.4103,
      "step": 11781
    },
    {
      "epoch": 0.5406819329080813,
      "grad_norm": 0.4334821403026581,
      "learning_rate": 9.354858871828254e-06,
      "loss": 0.38,
      "step": 11782
    },
    {
      "epoch": 0.5407278234133358,
      "grad_norm": 0.413735955953598,
      "learning_rate": 9.354738397669498e-06,
      "loss": 0.2719,
      "step": 11783
    },
    {
      "epoch": 0.5407737139185902,
      "grad_norm": 0.4799814224243164,
      "learning_rate": 9.35461791303897e-06,
      "loss": 0.447,
      "step": 11784
    },
    {
      "epoch": 0.5408196044238447,
      "grad_norm": 0.5069194436073303,
      "learning_rate": 9.354497417936963e-06,
      "loss": 0.5104,
      "step": 11785
    },
    {
      "epoch": 0.5408654949290992,
      "grad_norm": 0.44394585490226746,
      "learning_rate": 9.35437691236376e-06,
      "loss": 0.3367,
      "step": 11786
    },
    {
      "epoch": 0.5409113854343537,
      "grad_norm": 0.4505567252635956,
      "learning_rate": 9.35425639631966e-06,
      "loss": 0.346,
      "step": 11787
    },
    {
      "epoch": 0.5409572759396081,
      "grad_norm": 0.4658847451210022,
      "learning_rate": 9.354135869804943e-06,
      "loss": 0.3795,
      "step": 11788
    },
    {
      "epoch": 0.5410031664448626,
      "grad_norm": 0.47817355394363403,
      "learning_rate": 9.354015332819906e-06,
      "loss": 0.4299,
      "step": 11789
    },
    {
      "epoch": 0.5410490569501171,
      "grad_norm": 0.46292224526405334,
      "learning_rate": 9.353894785364834e-06,
      "loss": 0.4355,
      "step": 11790
    },
    {
      "epoch": 0.5410949474553715,
      "grad_norm": 0.495517373085022,
      "learning_rate": 9.35377422744002e-06,
      "loss": 0.4362,
      "step": 11791
    },
    {
      "epoch": 0.541140837960626,
      "grad_norm": 0.48221826553344727,
      "learning_rate": 9.353653659045755e-06,
      "loss": 0.4937,
      "step": 11792
    },
    {
      "epoch": 0.5411867284658805,
      "grad_norm": 0.48616424202919006,
      "learning_rate": 9.353533080182324e-06,
      "loss": 0.4281,
      "step": 11793
    },
    {
      "epoch": 0.5412326189711348,
      "grad_norm": 0.46081969141960144,
      "learning_rate": 9.353412490850022e-06,
      "loss": 0.4162,
      "step": 11794
    },
    {
      "epoch": 0.5412785094763893,
      "grad_norm": 0.4284566640853882,
      "learning_rate": 9.353291891049137e-06,
      "loss": 0.3753,
      "step": 11795
    },
    {
      "epoch": 0.5413243999816438,
      "grad_norm": 0.4643787741661072,
      "learning_rate": 9.353171280779958e-06,
      "loss": 0.3701,
      "step": 11796
    },
    {
      "epoch": 0.5413702904868982,
      "grad_norm": 0.4975559413433075,
      "learning_rate": 9.353050660042777e-06,
      "loss": 0.487,
      "step": 11797
    },
    {
      "epoch": 0.5414161809921527,
      "grad_norm": 0.47529593110084534,
      "learning_rate": 9.352930028837884e-06,
      "loss": 0.422,
      "step": 11798
    },
    {
      "epoch": 0.5414620714974072,
      "grad_norm": 0.4581986367702484,
      "learning_rate": 9.352809387165564e-06,
      "loss": 0.4088,
      "step": 11799
    },
    {
      "epoch": 0.5415079620026616,
      "grad_norm": 0.4705793559551239,
      "learning_rate": 9.352688735026115e-06,
      "loss": 0.3395,
      "step": 11800
    },
    {
      "epoch": 0.5415538525079161,
      "grad_norm": 0.4848674237728119,
      "learning_rate": 9.352568072419823e-06,
      "loss": 0.4898,
      "step": 11801
    },
    {
      "epoch": 0.5415997430131706,
      "grad_norm": 0.4780259132385254,
      "learning_rate": 9.352447399346977e-06,
      "loss": 0.3892,
      "step": 11802
    },
    {
      "epoch": 0.541645633518425,
      "grad_norm": 0.4688941538333893,
      "learning_rate": 9.352326715807869e-06,
      "loss": 0.362,
      "step": 11803
    },
    {
      "epoch": 0.5416915240236795,
      "grad_norm": 0.47002193331718445,
      "learning_rate": 9.352206021802789e-06,
      "loss": 0.3776,
      "step": 11804
    },
    {
      "epoch": 0.541737414528934,
      "grad_norm": 0.5230007767677307,
      "learning_rate": 9.352085317332027e-06,
      "loss": 0.494,
      "step": 11805
    },
    {
      "epoch": 0.5417833050341885,
      "grad_norm": 0.47935017943382263,
      "learning_rate": 9.351964602395874e-06,
      "loss": 0.396,
      "step": 11806
    },
    {
      "epoch": 0.5418291955394429,
      "grad_norm": 0.42556673288345337,
      "learning_rate": 9.35184387699462e-06,
      "loss": 0.381,
      "step": 11807
    },
    {
      "epoch": 0.5418750860446974,
      "grad_norm": 0.4134579300880432,
      "learning_rate": 9.351723141128554e-06,
      "loss": 0.3652,
      "step": 11808
    },
    {
      "epoch": 0.5419209765499519,
      "grad_norm": 0.46036505699157715,
      "learning_rate": 9.351602394797968e-06,
      "loss": 0.392,
      "step": 11809
    },
    {
      "epoch": 0.5419668670552062,
      "grad_norm": 0.440643846988678,
      "learning_rate": 9.35148163800315e-06,
      "loss": 0.3571,
      "step": 11810
    },
    {
      "epoch": 0.5420127575604607,
      "grad_norm": 0.4625470042228699,
      "learning_rate": 9.351360870744393e-06,
      "loss": 0.4355,
      "step": 11811
    },
    {
      "epoch": 0.5420586480657152,
      "grad_norm": 0.46800902485847473,
      "learning_rate": 9.351240093021987e-06,
      "loss": 0.3927,
      "step": 11812
    },
    {
      "epoch": 0.5421045385709696,
      "grad_norm": 0.4468221366405487,
      "learning_rate": 9.351119304836222e-06,
      "loss": 0.3952,
      "step": 11813
    },
    {
      "epoch": 0.5421504290762241,
      "grad_norm": 0.4418472945690155,
      "learning_rate": 9.350998506187388e-06,
      "loss": 0.3619,
      "step": 11814
    },
    {
      "epoch": 0.5421963195814786,
      "grad_norm": 0.43051525950431824,
      "learning_rate": 9.350877697075775e-06,
      "loss": 0.3289,
      "step": 11815
    },
    {
      "epoch": 0.542242210086733,
      "grad_norm": 0.4173886477947235,
      "learning_rate": 9.350756877501676e-06,
      "loss": 0.3282,
      "step": 11816
    },
    {
      "epoch": 0.5422881005919875,
      "grad_norm": 0.46568477153778076,
      "learning_rate": 9.350636047465378e-06,
      "loss": 0.438,
      "step": 11817
    },
    {
      "epoch": 0.542333991097242,
      "grad_norm": 0.47144076228141785,
      "learning_rate": 9.350515206967175e-06,
      "loss": 0.3893,
      "step": 11818
    },
    {
      "epoch": 0.5423798816024964,
      "grad_norm": 0.44918006658554077,
      "learning_rate": 9.350394356007355e-06,
      "loss": 0.3491,
      "step": 11819
    },
    {
      "epoch": 0.5424257721077509,
      "grad_norm": 0.44110172986984253,
      "learning_rate": 9.35027349458621e-06,
      "loss": 0.3486,
      "step": 11820
    },
    {
      "epoch": 0.5424716626130054,
      "grad_norm": 0.4185820519924164,
      "learning_rate": 9.35015262270403e-06,
      "loss": 0.2931,
      "step": 11821
    },
    {
      "epoch": 0.5425175531182598,
      "grad_norm": 0.45126640796661377,
      "learning_rate": 9.350031740361104e-06,
      "loss": 0.4063,
      "step": 11822
    },
    {
      "epoch": 0.5425634436235143,
      "grad_norm": 0.43778812885284424,
      "learning_rate": 9.349910847557729e-06,
      "loss": 0.3596,
      "step": 11823
    },
    {
      "epoch": 0.5426093341287688,
      "grad_norm": 0.4214657247066498,
      "learning_rate": 9.349789944294188e-06,
      "loss": 0.3254,
      "step": 11824
    },
    {
      "epoch": 0.5426552246340233,
      "grad_norm": 0.48458001017570496,
      "learning_rate": 9.349669030570776e-06,
      "loss": 0.4779,
      "step": 11825
    },
    {
      "epoch": 0.5427011151392777,
      "grad_norm": 0.5321649312973022,
      "learning_rate": 9.349548106387783e-06,
      "loss": 0.3905,
      "step": 11826
    },
    {
      "epoch": 0.5427470056445322,
      "grad_norm": 0.4375259280204773,
      "learning_rate": 9.349427171745498e-06,
      "loss": 0.4041,
      "step": 11827
    },
    {
      "epoch": 0.5427928961497867,
      "grad_norm": 0.4763076901435852,
      "learning_rate": 9.349306226644214e-06,
      "loss": 0.4435,
      "step": 11828
    },
    {
      "epoch": 0.542838786655041,
      "grad_norm": 0.44987210631370544,
      "learning_rate": 9.349185271084222e-06,
      "loss": 0.386,
      "step": 11829
    },
    {
      "epoch": 0.5428846771602955,
      "grad_norm": 0.4584258198738098,
      "learning_rate": 9.349064305065812e-06,
      "loss": 0.4269,
      "step": 11830
    },
    {
      "epoch": 0.54293056766555,
      "grad_norm": 0.45892924070358276,
      "learning_rate": 9.348943328589275e-06,
      "loss": 0.4416,
      "step": 11831
    },
    {
      "epoch": 0.5429764581708044,
      "grad_norm": 0.4012521207332611,
      "learning_rate": 9.348822341654901e-06,
      "loss": 0.2997,
      "step": 11832
    },
    {
      "epoch": 0.5430223486760589,
      "grad_norm": 0.43315044045448303,
      "learning_rate": 9.348701344262982e-06,
      "loss": 0.3403,
      "step": 11833
    },
    {
      "epoch": 0.5430682391813134,
      "grad_norm": 0.45989835262298584,
      "learning_rate": 9.34858033641381e-06,
      "loss": 0.3515,
      "step": 11834
    },
    {
      "epoch": 0.5431141296865678,
      "grad_norm": 0.4613991975784302,
      "learning_rate": 9.348459318107675e-06,
      "loss": 0.4053,
      "step": 11835
    },
    {
      "epoch": 0.5431600201918223,
      "grad_norm": 0.4385504126548767,
      "learning_rate": 9.348338289344865e-06,
      "loss": 0.3647,
      "step": 11836
    },
    {
      "epoch": 0.5432059106970768,
      "grad_norm": 0.4765866994857788,
      "learning_rate": 9.348217250125676e-06,
      "loss": 0.473,
      "step": 11837
    },
    {
      "epoch": 0.5432518012023312,
      "grad_norm": 0.47670695185661316,
      "learning_rate": 9.348096200450396e-06,
      "loss": 0.4057,
      "step": 11838
    },
    {
      "epoch": 0.5432976917075857,
      "grad_norm": 0.43690064549446106,
      "learning_rate": 9.347975140319318e-06,
      "loss": 0.3628,
      "step": 11839
    },
    {
      "epoch": 0.5433435822128402,
      "grad_norm": 0.5245376229286194,
      "learning_rate": 9.347854069732731e-06,
      "loss": 0.5441,
      "step": 11840
    },
    {
      "epoch": 0.5433894727180947,
      "grad_norm": 0.457271933555603,
      "learning_rate": 9.347732988690927e-06,
      "loss": 0.3883,
      "step": 11841
    },
    {
      "epoch": 0.5434353632233491,
      "grad_norm": 0.4336269497871399,
      "learning_rate": 9.347611897194198e-06,
      "loss": 0.4003,
      "step": 11842
    },
    {
      "epoch": 0.5434812537286036,
      "grad_norm": 0.4405486285686493,
      "learning_rate": 9.347490795242835e-06,
      "loss": 0.3878,
      "step": 11843
    },
    {
      "epoch": 0.5435271442338581,
      "grad_norm": 0.4559611976146698,
      "learning_rate": 9.347369682837128e-06,
      "loss": 0.3894,
      "step": 11844
    },
    {
      "epoch": 0.5435730347391124,
      "grad_norm": 0.46457624435424805,
      "learning_rate": 9.347248559977368e-06,
      "loss": 0.4254,
      "step": 11845
    },
    {
      "epoch": 0.543618925244367,
      "grad_norm": 0.4827214479446411,
      "learning_rate": 9.34712742666385e-06,
      "loss": 0.3872,
      "step": 11846
    },
    {
      "epoch": 0.5436648157496214,
      "grad_norm": 0.4457905888557434,
      "learning_rate": 9.34700628289686e-06,
      "loss": 0.3979,
      "step": 11847
    },
    {
      "epoch": 0.5437107062548758,
      "grad_norm": 0.4217289686203003,
      "learning_rate": 9.346885128676693e-06,
      "loss": 0.3097,
      "step": 11848
    },
    {
      "epoch": 0.5437565967601303,
      "grad_norm": 0.45775216817855835,
      "learning_rate": 9.346763964003637e-06,
      "loss": 0.3797,
      "step": 11849
    },
    {
      "epoch": 0.5438024872653848,
      "grad_norm": 0.44266411662101746,
      "learning_rate": 9.346642788877989e-06,
      "loss": 0.3768,
      "step": 11850
    },
    {
      "epoch": 0.5438483777706392,
      "grad_norm": 0.4988372027873993,
      "learning_rate": 9.346521603300035e-06,
      "loss": 0.4875,
      "step": 11851
    },
    {
      "epoch": 0.5438942682758937,
      "grad_norm": 0.4533219337463379,
      "learning_rate": 9.346400407270069e-06,
      "loss": 0.3812,
      "step": 11852
    },
    {
      "epoch": 0.5439401587811482,
      "grad_norm": 0.43136364221572876,
      "learning_rate": 9.346279200788382e-06,
      "loss": 0.3439,
      "step": 11853
    },
    {
      "epoch": 0.5439860492864026,
      "grad_norm": 0.46980252861976624,
      "learning_rate": 9.346157983855265e-06,
      "loss": 0.4665,
      "step": 11854
    },
    {
      "epoch": 0.5440319397916571,
      "grad_norm": 0.4426124691963196,
      "learning_rate": 9.346036756471009e-06,
      "loss": 0.4032,
      "step": 11855
    },
    {
      "epoch": 0.5440778302969116,
      "grad_norm": 0.4331699013710022,
      "learning_rate": 9.345915518635906e-06,
      "loss": 0.339,
      "step": 11856
    },
    {
      "epoch": 0.544123720802166,
      "grad_norm": 0.5154035687446594,
      "learning_rate": 9.345794270350248e-06,
      "loss": 0.4104,
      "step": 11857
    },
    {
      "epoch": 0.5441696113074205,
      "grad_norm": 0.4645753502845764,
      "learning_rate": 9.345673011614327e-06,
      "loss": 0.3731,
      "step": 11858
    },
    {
      "epoch": 0.544215501812675,
      "grad_norm": 0.49769800901412964,
      "learning_rate": 9.345551742428434e-06,
      "loss": 0.4557,
      "step": 11859
    },
    {
      "epoch": 0.5442613923179295,
      "grad_norm": 0.44506654143333435,
      "learning_rate": 9.34543046279286e-06,
      "loss": 0.3917,
      "step": 11860
    },
    {
      "epoch": 0.5443072828231839,
      "grad_norm": 0.537725031375885,
      "learning_rate": 9.345309172707898e-06,
      "loss": 0.4401,
      "step": 11861
    },
    {
      "epoch": 0.5443531733284384,
      "grad_norm": 0.4936114251613617,
      "learning_rate": 9.345187872173838e-06,
      "loss": 0.4316,
      "step": 11862
    },
    {
      "epoch": 0.5443990638336929,
      "grad_norm": 0.47287797927856445,
      "learning_rate": 9.345066561190974e-06,
      "loss": 0.3715,
      "step": 11863
    },
    {
      "epoch": 0.5444449543389472,
      "grad_norm": 0.4486594796180725,
      "learning_rate": 9.344945239759595e-06,
      "loss": 0.397,
      "step": 11864
    },
    {
      "epoch": 0.5444908448442017,
      "grad_norm": 0.4418001174926758,
      "learning_rate": 9.344823907879995e-06,
      "loss": 0.3803,
      "step": 11865
    },
    {
      "epoch": 0.5445367353494562,
      "grad_norm": 0.4633978009223938,
      "learning_rate": 9.344702565552463e-06,
      "loss": 0.3903,
      "step": 11866
    },
    {
      "epoch": 0.5445826258547106,
      "grad_norm": 0.4715845286846161,
      "learning_rate": 9.344581212777292e-06,
      "loss": 0.4259,
      "step": 11867
    },
    {
      "epoch": 0.5446285163599651,
      "grad_norm": 0.4580090641975403,
      "learning_rate": 9.344459849554777e-06,
      "loss": 0.4025,
      "step": 11868
    },
    {
      "epoch": 0.5446744068652196,
      "grad_norm": 0.45308545231819153,
      "learning_rate": 9.344338475885206e-06,
      "loss": 0.4072,
      "step": 11869
    },
    {
      "epoch": 0.544720297370474,
      "grad_norm": 0.44486701488494873,
      "learning_rate": 9.344217091768872e-06,
      "loss": 0.3671,
      "step": 11870
    },
    {
      "epoch": 0.5447661878757285,
      "grad_norm": 0.4847283661365509,
      "learning_rate": 9.344095697206069e-06,
      "loss": 0.4654,
      "step": 11871
    },
    {
      "epoch": 0.544812078380983,
      "grad_norm": 0.5664860010147095,
      "learning_rate": 9.343974292197083e-06,
      "loss": 0.4344,
      "step": 11872
    },
    {
      "epoch": 0.5448579688862374,
      "grad_norm": 0.42176544666290283,
      "learning_rate": 9.343852876742212e-06,
      "loss": 0.3323,
      "step": 11873
    },
    {
      "epoch": 0.5449038593914919,
      "grad_norm": 0.4359741508960724,
      "learning_rate": 9.343731450841746e-06,
      "loss": 0.333,
      "step": 11874
    },
    {
      "epoch": 0.5449497498967464,
      "grad_norm": 0.4587786793708801,
      "learning_rate": 9.343610014495977e-06,
      "loss": 0.3576,
      "step": 11875
    },
    {
      "epoch": 0.5449956404020009,
      "grad_norm": 0.4939517080783844,
      "learning_rate": 9.343488567705197e-06,
      "loss": 0.4374,
      "step": 11876
    },
    {
      "epoch": 0.5450415309072553,
      "grad_norm": 0.5242145657539368,
      "learning_rate": 9.343367110469697e-06,
      "loss": 0.3855,
      "step": 11877
    },
    {
      "epoch": 0.5450874214125098,
      "grad_norm": 0.4620373249053955,
      "learning_rate": 9.343245642789768e-06,
      "loss": 0.4518,
      "step": 11878
    },
    {
      "epoch": 0.5451333119177643,
      "grad_norm": 0.49062758684158325,
      "learning_rate": 9.343124164665706e-06,
      "loss": 0.5207,
      "step": 11879
    },
    {
      "epoch": 0.5451792024230186,
      "grad_norm": 0.4475047290325165,
      "learning_rate": 9.343002676097803e-06,
      "loss": 0.3944,
      "step": 11880
    },
    {
      "epoch": 0.5452250929282731,
      "grad_norm": 0.49329304695129395,
      "learning_rate": 9.342881177086347e-06,
      "loss": 0.4507,
      "step": 11881
    },
    {
      "epoch": 0.5452709834335276,
      "grad_norm": 0.4308706820011139,
      "learning_rate": 9.342759667631633e-06,
      "loss": 0.3393,
      "step": 11882
    },
    {
      "epoch": 0.545316873938782,
      "grad_norm": 0.44557979702949524,
      "learning_rate": 9.34263814773395e-06,
      "loss": 0.4097,
      "step": 11883
    },
    {
      "epoch": 0.5453627644440365,
      "grad_norm": 0.43194103240966797,
      "learning_rate": 9.342516617393597e-06,
      "loss": 0.3929,
      "step": 11884
    },
    {
      "epoch": 0.545408654949291,
      "grad_norm": 0.4289681017398834,
      "learning_rate": 9.342395076610859e-06,
      "loss": 0.3572,
      "step": 11885
    },
    {
      "epoch": 0.5454545454545454,
      "grad_norm": 0.43574059009552,
      "learning_rate": 9.342273525386034e-06,
      "loss": 0.3628,
      "step": 11886
    },
    {
      "epoch": 0.5455004359597999,
      "grad_norm": 0.47124069929122925,
      "learning_rate": 9.342151963719411e-06,
      "loss": 0.3799,
      "step": 11887
    },
    {
      "epoch": 0.5455463264650544,
      "grad_norm": 0.4588020443916321,
      "learning_rate": 9.34203039161128e-06,
      "loss": 0.417,
      "step": 11888
    },
    {
      "epoch": 0.5455922169703088,
      "grad_norm": 0.44188037514686584,
      "learning_rate": 9.341908809061939e-06,
      "loss": 0.3678,
      "step": 11889
    },
    {
      "epoch": 0.5456381074755633,
      "grad_norm": 0.4796905815601349,
      "learning_rate": 9.341787216071678e-06,
      "loss": 0.4204,
      "step": 11890
    },
    {
      "epoch": 0.5456839979808178,
      "grad_norm": 0.46321603655815125,
      "learning_rate": 9.341665612640789e-06,
      "loss": 0.426,
      "step": 11891
    },
    {
      "epoch": 0.5457298884860722,
      "grad_norm": 0.4615887999534607,
      "learning_rate": 9.341543998769563e-06,
      "loss": 0.439,
      "step": 11892
    },
    {
      "epoch": 0.5457757789913267,
      "grad_norm": 0.436480313539505,
      "learning_rate": 9.341422374458294e-06,
      "loss": 0.3936,
      "step": 11893
    },
    {
      "epoch": 0.5458216694965812,
      "grad_norm": 0.44779476523399353,
      "learning_rate": 9.341300739707276e-06,
      "loss": 0.4189,
      "step": 11894
    },
    {
      "epoch": 0.5458675600018357,
      "grad_norm": 0.4934377372264862,
      "learning_rate": 9.341179094516799e-06,
      "loss": 0.4308,
      "step": 11895
    },
    {
      "epoch": 0.5459134505070901,
      "grad_norm": 0.43862295150756836,
      "learning_rate": 9.341057438887155e-06,
      "loss": 0.3299,
      "step": 11896
    },
    {
      "epoch": 0.5459593410123446,
      "grad_norm": 0.44630101323127747,
      "learning_rate": 9.34093577281864e-06,
      "loss": 0.3454,
      "step": 11897
    },
    {
      "epoch": 0.546005231517599,
      "grad_norm": 0.547847330570221,
      "learning_rate": 9.340814096311543e-06,
      "loss": 0.4226,
      "step": 11898
    },
    {
      "epoch": 0.5460511220228534,
      "grad_norm": 0.454950749874115,
      "learning_rate": 9.340692409366158e-06,
      "loss": 0.3915,
      "step": 11899
    },
    {
      "epoch": 0.5460970125281079,
      "grad_norm": 0.4577164053916931,
      "learning_rate": 9.34057071198278e-06,
      "loss": 0.3861,
      "step": 11900
    },
    {
      "epoch": 0.5461429030333624,
      "grad_norm": 0.48604515194892883,
      "learning_rate": 9.340449004161698e-06,
      "loss": 0.5,
      "step": 11901
    },
    {
      "epoch": 0.5461887935386168,
      "grad_norm": 0.47552889585494995,
      "learning_rate": 9.340327285903206e-06,
      "loss": 0.3318,
      "step": 11902
    },
    {
      "epoch": 0.5462346840438713,
      "grad_norm": 0.4860277771949768,
      "learning_rate": 9.340205557207595e-06,
      "loss": 0.4303,
      "step": 11903
    },
    {
      "epoch": 0.5462805745491258,
      "grad_norm": 0.435284823179245,
      "learning_rate": 9.340083818075163e-06,
      "loss": 0.3703,
      "step": 11904
    },
    {
      "epoch": 0.5463264650543802,
      "grad_norm": 0.4470125436782837,
      "learning_rate": 9.339962068506197e-06,
      "loss": 0.406,
      "step": 11905
    },
    {
      "epoch": 0.5463723555596347,
      "grad_norm": 0.5407623052597046,
      "learning_rate": 9.339840308500991e-06,
      "loss": 0.4444,
      "step": 11906
    },
    {
      "epoch": 0.5464182460648892,
      "grad_norm": 0.4669787287712097,
      "learning_rate": 9.33971853805984e-06,
      "loss": 0.4159,
      "step": 11907
    },
    {
      "epoch": 0.5464641365701436,
      "grad_norm": 0.39115869998931885,
      "learning_rate": 9.339596757183037e-06,
      "loss": 0.2823,
      "step": 11908
    },
    {
      "epoch": 0.5465100270753981,
      "grad_norm": 0.48859718441963196,
      "learning_rate": 9.339474965870872e-06,
      "loss": 0.4575,
      "step": 11909
    },
    {
      "epoch": 0.5465559175806526,
      "grad_norm": 0.4885725975036621,
      "learning_rate": 9.339353164123639e-06,
      "loss": 0.4367,
      "step": 11910
    },
    {
      "epoch": 0.546601808085907,
      "grad_norm": 0.5023432374000549,
      "learning_rate": 9.33923135194163e-06,
      "loss": 0.4342,
      "step": 11911
    },
    {
      "epoch": 0.5466476985911615,
      "grad_norm": 0.5017714500427246,
      "learning_rate": 9.33910952932514e-06,
      "loss": 0.5021,
      "step": 11912
    },
    {
      "epoch": 0.546693589096416,
      "grad_norm": 0.45476987957954407,
      "learning_rate": 9.338987696274463e-06,
      "loss": 0.3801,
      "step": 11913
    },
    {
      "epoch": 0.5467394796016705,
      "grad_norm": 0.44223013520240784,
      "learning_rate": 9.338865852789889e-06,
      "loss": 0.3816,
      "step": 11914
    },
    {
      "epoch": 0.5467853701069249,
      "grad_norm": 0.4331381022930145,
      "learning_rate": 9.338743998871711e-06,
      "loss": 0.3938,
      "step": 11915
    },
    {
      "epoch": 0.5468312606121793,
      "grad_norm": 0.4733550250530243,
      "learning_rate": 9.338622134520224e-06,
      "loss": 0.3526,
      "step": 11916
    },
    {
      "epoch": 0.5468771511174338,
      "grad_norm": 0.5198267102241516,
      "learning_rate": 9.338500259735718e-06,
      "loss": 0.4789,
      "step": 11917
    },
    {
      "epoch": 0.5469230416226882,
      "grad_norm": 0.47378867864608765,
      "learning_rate": 9.33837837451849e-06,
      "loss": 0.4302,
      "step": 11918
    },
    {
      "epoch": 0.5469689321279427,
      "grad_norm": 0.47132954001426697,
      "learning_rate": 9.338256478868833e-06,
      "loss": 0.388,
      "step": 11919
    },
    {
      "epoch": 0.5470148226331972,
      "grad_norm": 0.4417761564254761,
      "learning_rate": 9.338134572787036e-06,
      "loss": 0.331,
      "step": 11920
    },
    {
      "epoch": 0.5470607131384516,
      "grad_norm": 0.4476693272590637,
      "learning_rate": 9.338012656273395e-06,
      "loss": 0.3778,
      "step": 11921
    },
    {
      "epoch": 0.5471066036437061,
      "grad_norm": 0.4371757209300995,
      "learning_rate": 9.337890729328202e-06,
      "loss": 0.3213,
      "step": 11922
    },
    {
      "epoch": 0.5471524941489606,
      "grad_norm": 0.49356135725975037,
      "learning_rate": 9.337768791951751e-06,
      "loss": 0.4583,
      "step": 11923
    },
    {
      "epoch": 0.547198384654215,
      "grad_norm": 0.4758743941783905,
      "learning_rate": 9.337646844144337e-06,
      "loss": 0.4605,
      "step": 11924
    },
    {
      "epoch": 0.5472442751594695,
      "grad_norm": 0.45748040080070496,
      "learning_rate": 9.33752488590625e-06,
      "loss": 0.4023,
      "step": 11925
    },
    {
      "epoch": 0.547290165664724,
      "grad_norm": 0.4380097985267639,
      "learning_rate": 9.337402917237783e-06,
      "loss": 0.3501,
      "step": 11926
    },
    {
      "epoch": 0.5473360561699784,
      "grad_norm": 0.48652786016464233,
      "learning_rate": 9.337280938139233e-06,
      "loss": 0.4767,
      "step": 11927
    },
    {
      "epoch": 0.5473819466752329,
      "grad_norm": 0.476633220911026,
      "learning_rate": 9.337158948610891e-06,
      "loss": 0.4006,
      "step": 11928
    },
    {
      "epoch": 0.5474278371804874,
      "grad_norm": 0.4548337459564209,
      "learning_rate": 9.337036948653051e-06,
      "loss": 0.3516,
      "step": 11929
    },
    {
      "epoch": 0.5474737276857419,
      "grad_norm": 0.48502957820892334,
      "learning_rate": 9.336914938266004e-06,
      "loss": 0.4405,
      "step": 11930
    },
    {
      "epoch": 0.5475196181909963,
      "grad_norm": 0.4599230885505676,
      "learning_rate": 9.336792917450047e-06,
      "loss": 0.4292,
      "step": 11931
    },
    {
      "epoch": 0.5475655086962508,
      "grad_norm": 0.46210649609565735,
      "learning_rate": 9.336670886205471e-06,
      "loss": 0.4366,
      "step": 11932
    },
    {
      "epoch": 0.5476113992015053,
      "grad_norm": 0.4912729859352112,
      "learning_rate": 9.336548844532571e-06,
      "loss": 0.4015,
      "step": 11933
    },
    {
      "epoch": 0.5476572897067596,
      "grad_norm": 0.4451366364955902,
      "learning_rate": 9.336426792431639e-06,
      "loss": 0.3736,
      "step": 11934
    },
    {
      "epoch": 0.5477031802120141,
      "grad_norm": 0.4855445921421051,
      "learning_rate": 9.33630472990297e-06,
      "loss": 0.3833,
      "step": 11935
    },
    {
      "epoch": 0.5477490707172686,
      "grad_norm": 0.445516437292099,
      "learning_rate": 9.336182656946854e-06,
      "loss": 0.4041,
      "step": 11936
    },
    {
      "epoch": 0.547794961222523,
      "grad_norm": 0.523945152759552,
      "learning_rate": 9.33606057356359e-06,
      "loss": 0.4052,
      "step": 11937
    },
    {
      "epoch": 0.5478408517277775,
      "grad_norm": 0.4580520689487457,
      "learning_rate": 9.33593847975347e-06,
      "loss": 0.3829,
      "step": 11938
    },
    {
      "epoch": 0.547886742233032,
      "grad_norm": 0.41742679476737976,
      "learning_rate": 9.335816375516784e-06,
      "loss": 0.3285,
      "step": 11939
    },
    {
      "epoch": 0.5479326327382864,
      "grad_norm": 0.4721064269542694,
      "learning_rate": 9.335694260853828e-06,
      "loss": 0.394,
      "step": 11940
    },
    {
      "epoch": 0.5479785232435409,
      "grad_norm": 0.4114242494106293,
      "learning_rate": 9.335572135764896e-06,
      "loss": 0.3232,
      "step": 11941
    },
    {
      "epoch": 0.5480244137487954,
      "grad_norm": 0.4183559715747833,
      "learning_rate": 9.335450000250282e-06,
      "loss": 0.3183,
      "step": 11942
    },
    {
      "epoch": 0.5480703042540498,
      "grad_norm": 0.45973044633865356,
      "learning_rate": 9.335327854310279e-06,
      "loss": 0.3912,
      "step": 11943
    },
    {
      "epoch": 0.5481161947593043,
      "grad_norm": 0.4514411687850952,
      "learning_rate": 9.33520569794518e-06,
      "loss": 0.4181,
      "step": 11944
    },
    {
      "epoch": 0.5481620852645588,
      "grad_norm": 0.4932914078235626,
      "learning_rate": 9.335083531155281e-06,
      "loss": 0.4803,
      "step": 11945
    },
    {
      "epoch": 0.5482079757698132,
      "grad_norm": 0.4695689380168915,
      "learning_rate": 9.334961353940872e-06,
      "loss": 0.4479,
      "step": 11946
    },
    {
      "epoch": 0.5482538662750677,
      "grad_norm": 0.4880853295326233,
      "learning_rate": 9.334839166302252e-06,
      "loss": 0.4232,
      "step": 11947
    },
    {
      "epoch": 0.5482997567803222,
      "grad_norm": 0.45557525753974915,
      "learning_rate": 9.33471696823971e-06,
      "loss": 0.372,
      "step": 11948
    },
    {
      "epoch": 0.5483456472855767,
      "grad_norm": 0.46407681703567505,
      "learning_rate": 9.334594759753541e-06,
      "loss": 0.3999,
      "step": 11949
    },
    {
      "epoch": 0.548391537790831,
      "grad_norm": 0.48239511251449585,
      "learning_rate": 9.334472540844042e-06,
      "loss": 0.4207,
      "step": 11950
    },
    {
      "epoch": 0.5484374282960855,
      "grad_norm": 0.42568421363830566,
      "learning_rate": 9.334350311511503e-06,
      "loss": 0.3556,
      "step": 11951
    },
    {
      "epoch": 0.54848331880134,
      "grad_norm": 0.47539007663726807,
      "learning_rate": 9.33422807175622e-06,
      "loss": 0.4405,
      "step": 11952
    },
    {
      "epoch": 0.5485292093065944,
      "grad_norm": 0.47793328762054443,
      "learning_rate": 9.334105821578485e-06,
      "loss": 0.444,
      "step": 11953
    },
    {
      "epoch": 0.5485750998118489,
      "grad_norm": 0.4893568754196167,
      "learning_rate": 9.333983560978594e-06,
      "loss": 0.4586,
      "step": 11954
    },
    {
      "epoch": 0.5486209903171034,
      "grad_norm": 0.45939329266548157,
      "learning_rate": 9.333861289956842e-06,
      "loss": 0.3775,
      "step": 11955
    },
    {
      "epoch": 0.5486668808223578,
      "grad_norm": 0.4701369106769562,
      "learning_rate": 9.33373900851352e-06,
      "loss": 0.4121,
      "step": 11956
    },
    {
      "epoch": 0.5487127713276123,
      "grad_norm": 0.4366547465324402,
      "learning_rate": 9.333616716648924e-06,
      "loss": 0.3636,
      "step": 11957
    },
    {
      "epoch": 0.5487586618328668,
      "grad_norm": 0.4269086718559265,
      "learning_rate": 9.333494414363346e-06,
      "loss": 0.2984,
      "step": 11958
    },
    {
      "epoch": 0.5488045523381212,
      "grad_norm": 0.45478543639183044,
      "learning_rate": 9.333372101657084e-06,
      "loss": 0.4023,
      "step": 11959
    },
    {
      "epoch": 0.5488504428433757,
      "grad_norm": 0.46015894412994385,
      "learning_rate": 9.33324977853043e-06,
      "loss": 0.3679,
      "step": 11960
    },
    {
      "epoch": 0.5488963333486302,
      "grad_norm": 0.42885449528694153,
      "learning_rate": 9.333127444983676e-06,
      "loss": 0.375,
      "step": 11961
    },
    {
      "epoch": 0.5489422238538846,
      "grad_norm": 0.5015444159507751,
      "learning_rate": 9.333005101017118e-06,
      "loss": 0.4642,
      "step": 11962
    },
    {
      "epoch": 0.5489881143591391,
      "grad_norm": 0.4491008222103119,
      "learning_rate": 9.332882746631051e-06,
      "loss": 0.3625,
      "step": 11963
    },
    {
      "epoch": 0.5490340048643936,
      "grad_norm": 0.4310057759284973,
      "learning_rate": 9.332760381825768e-06,
      "loss": 0.4002,
      "step": 11964
    },
    {
      "epoch": 0.5490798953696481,
      "grad_norm": 0.46215924620628357,
      "learning_rate": 9.332638006601565e-06,
      "loss": 0.4374,
      "step": 11965
    },
    {
      "epoch": 0.5491257858749025,
      "grad_norm": 0.4195605218410492,
      "learning_rate": 9.332515620958732e-06,
      "loss": 0.3289,
      "step": 11966
    },
    {
      "epoch": 0.549171676380157,
      "grad_norm": 0.4543750286102295,
      "learning_rate": 9.332393224897571e-06,
      "loss": 0.3647,
      "step": 11967
    },
    {
      "epoch": 0.5492175668854115,
      "grad_norm": 0.4636421501636505,
      "learning_rate": 9.332270818418368e-06,
      "loss": 0.3641,
      "step": 11968
    },
    {
      "epoch": 0.5492634573906658,
      "grad_norm": 0.4815918207168579,
      "learning_rate": 9.332148401521423e-06,
      "loss": 0.4052,
      "step": 11969
    },
    {
      "epoch": 0.5493093478959203,
      "grad_norm": 0.4500230550765991,
      "learning_rate": 9.332025974207026e-06,
      "loss": 0.4557,
      "step": 11970
    },
    {
      "epoch": 0.5493552384011748,
      "grad_norm": 0.4289769232273102,
      "learning_rate": 9.331903536475476e-06,
      "loss": 0.342,
      "step": 11971
    },
    {
      "epoch": 0.5494011289064292,
      "grad_norm": 0.48944559693336487,
      "learning_rate": 9.331781088327065e-06,
      "loss": 0.4711,
      "step": 11972
    },
    {
      "epoch": 0.5494470194116837,
      "grad_norm": 0.6866310238838196,
      "learning_rate": 9.331658629762085e-06,
      "loss": 0.5128,
      "step": 11973
    },
    {
      "epoch": 0.5494929099169382,
      "grad_norm": 0.525269627571106,
      "learning_rate": 9.331536160780837e-06,
      "loss": 0.4636,
      "step": 11974
    },
    {
      "epoch": 0.5495388004221926,
      "grad_norm": 0.47854605317115784,
      "learning_rate": 9.33141368138361e-06,
      "loss": 0.4853,
      "step": 11975
    },
    {
      "epoch": 0.5495846909274471,
      "grad_norm": 0.47168368101119995,
      "learning_rate": 9.3312911915707e-06,
      "loss": 0.4211,
      "step": 11976
    },
    {
      "epoch": 0.5496305814327016,
      "grad_norm": 0.45862630009651184,
      "learning_rate": 9.3311686913424e-06,
      "loss": 0.4234,
      "step": 11977
    },
    {
      "epoch": 0.549676471937956,
      "grad_norm": 0.44495606422424316,
      "learning_rate": 9.331046180699008e-06,
      "loss": 0.3773,
      "step": 11978
    },
    {
      "epoch": 0.5497223624432105,
      "grad_norm": 0.4484730064868927,
      "learning_rate": 9.330923659640816e-06,
      "loss": 0.4324,
      "step": 11979
    },
    {
      "epoch": 0.549768252948465,
      "grad_norm": 0.4571169912815094,
      "learning_rate": 9.330801128168121e-06,
      "loss": 0.407,
      "step": 11980
    },
    {
      "epoch": 0.5498141434537194,
      "grad_norm": 0.4378121495246887,
      "learning_rate": 9.330678586281215e-06,
      "loss": 0.3449,
      "step": 11981
    },
    {
      "epoch": 0.5498600339589739,
      "grad_norm": 0.4404963254928589,
      "learning_rate": 9.330556033980395e-06,
      "loss": 0.367,
      "step": 11982
    },
    {
      "epoch": 0.5499059244642284,
      "grad_norm": 0.47941526770591736,
      "learning_rate": 9.330433471265954e-06,
      "loss": 0.4032,
      "step": 11983
    },
    {
      "epoch": 0.5499518149694829,
      "grad_norm": 0.45340219140052795,
      "learning_rate": 9.330310898138186e-06,
      "loss": 0.3782,
      "step": 11984
    },
    {
      "epoch": 0.5499977054747373,
      "grad_norm": 0.4625984728336334,
      "learning_rate": 9.330188314597388e-06,
      "loss": 0.3934,
      "step": 11985
    },
    {
      "epoch": 0.5500435959799918,
      "grad_norm": 0.42984768748283386,
      "learning_rate": 9.330065720643853e-06,
      "loss": 0.3826,
      "step": 11986
    },
    {
      "epoch": 0.5500894864852462,
      "grad_norm": 0.4522968530654907,
      "learning_rate": 9.329943116277877e-06,
      "loss": 0.4063,
      "step": 11987
    },
    {
      "epoch": 0.5501353769905006,
      "grad_norm": 0.46702197194099426,
      "learning_rate": 9.329820501499754e-06,
      "loss": 0.3816,
      "step": 11988
    },
    {
      "epoch": 0.5501812674957551,
      "grad_norm": 0.5477336049079895,
      "learning_rate": 9.32969787630978e-06,
      "loss": 0.466,
      "step": 11989
    },
    {
      "epoch": 0.5502271580010096,
      "grad_norm": 0.45845189690589905,
      "learning_rate": 9.329575240708249e-06,
      "loss": 0.4122,
      "step": 11990
    },
    {
      "epoch": 0.550273048506264,
      "grad_norm": 0.484897255897522,
      "learning_rate": 9.329452594695456e-06,
      "loss": 0.4681,
      "step": 11991
    },
    {
      "epoch": 0.5503189390115185,
      "grad_norm": 0.44475236535072327,
      "learning_rate": 9.329329938271696e-06,
      "loss": 0.3602,
      "step": 11992
    },
    {
      "epoch": 0.550364829516773,
      "grad_norm": 0.4254019856452942,
      "learning_rate": 9.329207271437264e-06,
      "loss": 0.3114,
      "step": 11993
    },
    {
      "epoch": 0.5504107200220274,
      "grad_norm": 0.41875210404396057,
      "learning_rate": 9.329084594192453e-06,
      "loss": 0.3442,
      "step": 11994
    },
    {
      "epoch": 0.5504566105272819,
      "grad_norm": 0.4428924024105072,
      "learning_rate": 9.32896190653756e-06,
      "loss": 0.3963,
      "step": 11995
    },
    {
      "epoch": 0.5505025010325364,
      "grad_norm": 0.43282920122146606,
      "learning_rate": 9.328839208472884e-06,
      "loss": 0.3964,
      "step": 11996
    },
    {
      "epoch": 0.5505483915377908,
      "grad_norm": 0.4443638026714325,
      "learning_rate": 9.328716499998712e-06,
      "loss": 0.4067,
      "step": 11997
    },
    {
      "epoch": 0.5505942820430453,
      "grad_norm": 0.46203410625457764,
      "learning_rate": 9.328593781115344e-06,
      "loss": 0.3466,
      "step": 11998
    },
    {
      "epoch": 0.5506401725482998,
      "grad_norm": 0.4323679208755493,
      "learning_rate": 9.328471051823074e-06,
      "loss": 0.3247,
      "step": 11999
    },
    {
      "epoch": 0.5506860630535542,
      "grad_norm": 0.4740195572376251,
      "learning_rate": 9.328348312122198e-06,
      "loss": 0.4242,
      "step": 12000
    },
    {
      "epoch": 0.5507319535588087,
      "grad_norm": 0.4352451264858246,
      "learning_rate": 9.32822556201301e-06,
      "loss": 0.4,
      "step": 12001
    },
    {
      "epoch": 0.5507778440640632,
      "grad_norm": 0.49147024750709534,
      "learning_rate": 9.328102801495804e-06,
      "loss": 0.4212,
      "step": 12002
    },
    {
      "epoch": 0.5508237345693177,
      "grad_norm": 0.4650729298591614,
      "learning_rate": 9.327980030570877e-06,
      "loss": 0.4151,
      "step": 12003
    },
    {
      "epoch": 0.550869625074572,
      "grad_norm": 0.45757368206977844,
      "learning_rate": 9.327857249238527e-06,
      "loss": 0.3818,
      "step": 12004
    },
    {
      "epoch": 0.5509155155798265,
      "grad_norm": 0.4687870740890503,
      "learning_rate": 9.327734457499044e-06,
      "loss": 0.3814,
      "step": 12005
    },
    {
      "epoch": 0.550961406085081,
      "grad_norm": 0.4887888729572296,
      "learning_rate": 9.327611655352726e-06,
      "loss": 0.4632,
      "step": 12006
    },
    {
      "epoch": 0.5510072965903354,
      "grad_norm": 0.49543240666389465,
      "learning_rate": 9.327488842799868e-06,
      "loss": 0.5111,
      "step": 12007
    },
    {
      "epoch": 0.5510531870955899,
      "grad_norm": 0.45357561111450195,
      "learning_rate": 9.327366019840764e-06,
      "loss": 0.4253,
      "step": 12008
    },
    {
      "epoch": 0.5510990776008444,
      "grad_norm": 0.46783414483070374,
      "learning_rate": 9.327243186475711e-06,
      "loss": 0.4283,
      "step": 12009
    },
    {
      "epoch": 0.5511449681060988,
      "grad_norm": 0.5238674283027649,
      "learning_rate": 9.327120342705004e-06,
      "loss": 0.4899,
      "step": 12010
    },
    {
      "epoch": 0.5511908586113533,
      "grad_norm": 0.45498740673065186,
      "learning_rate": 9.326997488528938e-06,
      "loss": 0.3767,
      "step": 12011
    },
    {
      "epoch": 0.5512367491166078,
      "grad_norm": 0.4747585654258728,
      "learning_rate": 9.326874623947809e-06,
      "loss": 0.466,
      "step": 12012
    },
    {
      "epoch": 0.5512826396218622,
      "grad_norm": 0.5111625790596008,
      "learning_rate": 9.326751748961914e-06,
      "loss": 0.4781,
      "step": 12013
    },
    {
      "epoch": 0.5513285301271167,
      "grad_norm": 0.4384748935699463,
      "learning_rate": 9.326628863571543e-06,
      "loss": 0.3585,
      "step": 12014
    },
    {
      "epoch": 0.5513744206323712,
      "grad_norm": 0.4508642554283142,
      "learning_rate": 9.326505967776997e-06,
      "loss": 0.4009,
      "step": 12015
    },
    {
      "epoch": 0.5514203111376256,
      "grad_norm": 0.4847089648246765,
      "learning_rate": 9.32638306157857e-06,
      "loss": 0.3967,
      "step": 12016
    },
    {
      "epoch": 0.5514662016428801,
      "grad_norm": 0.45215851068496704,
      "learning_rate": 9.326260144976555e-06,
      "loss": 0.4036,
      "step": 12017
    },
    {
      "epoch": 0.5515120921481346,
      "grad_norm": 0.49371060729026794,
      "learning_rate": 9.326137217971252e-06,
      "loss": 0.4417,
      "step": 12018
    },
    {
      "epoch": 0.5515579826533891,
      "grad_norm": 0.44972464442253113,
      "learning_rate": 9.326014280562954e-06,
      "loss": 0.389,
      "step": 12019
    },
    {
      "epoch": 0.5516038731586435,
      "grad_norm": 0.43199482560157776,
      "learning_rate": 9.325891332751957e-06,
      "loss": 0.36,
      "step": 12020
    },
    {
      "epoch": 0.551649763663898,
      "grad_norm": 0.4447985887527466,
      "learning_rate": 9.325768374538556e-06,
      "loss": 0.4016,
      "step": 12021
    },
    {
      "epoch": 0.5516956541691524,
      "grad_norm": 0.4255668520927429,
      "learning_rate": 9.325645405923048e-06,
      "loss": 0.3343,
      "step": 12022
    },
    {
      "epoch": 0.5517415446744068,
      "grad_norm": 0.43600738048553467,
      "learning_rate": 9.325522426905726e-06,
      "loss": 0.3156,
      "step": 12023
    },
    {
      "epoch": 0.5517874351796613,
      "grad_norm": 0.4471251368522644,
      "learning_rate": 9.32539943748689e-06,
      "loss": 0.4111,
      "step": 12024
    },
    {
      "epoch": 0.5518333256849158,
      "grad_norm": 0.44182848930358887,
      "learning_rate": 9.325276437666833e-06,
      "loss": 0.3506,
      "step": 12025
    },
    {
      "epoch": 0.5518792161901702,
      "grad_norm": 0.5259996652603149,
      "learning_rate": 9.325153427445853e-06,
      "loss": 0.4574,
      "step": 12026
    },
    {
      "epoch": 0.5519251066954247,
      "grad_norm": 0.42675042152404785,
      "learning_rate": 9.325030406824241e-06,
      "loss": 0.3743,
      "step": 12027
    },
    {
      "epoch": 0.5519709972006792,
      "grad_norm": 0.4457780718803406,
      "learning_rate": 9.324907375802297e-06,
      "loss": 0.37,
      "step": 12028
    },
    {
      "epoch": 0.5520168877059336,
      "grad_norm": 0.5031201243400574,
      "learning_rate": 9.324784334380315e-06,
      "loss": 0.432,
      "step": 12029
    },
    {
      "epoch": 0.5520627782111881,
      "grad_norm": 0.47958654165267944,
      "learning_rate": 9.324661282558594e-06,
      "loss": 0.4218,
      "step": 12030
    },
    {
      "epoch": 0.5521086687164426,
      "grad_norm": 0.45022791624069214,
      "learning_rate": 9.324538220337426e-06,
      "loss": 0.3839,
      "step": 12031
    },
    {
      "epoch": 0.552154559221697,
      "grad_norm": 0.4702366590499878,
      "learning_rate": 9.324415147717107e-06,
      "loss": 0.4302,
      "step": 12032
    },
    {
      "epoch": 0.5522004497269515,
      "grad_norm": 0.4438423216342926,
      "learning_rate": 9.324292064697938e-06,
      "loss": 0.3575,
      "step": 12033
    },
    {
      "epoch": 0.552246340232206,
      "grad_norm": 0.43346479535102844,
      "learning_rate": 9.324168971280208e-06,
      "loss": 0.3334,
      "step": 12034
    },
    {
      "epoch": 0.5522922307374604,
      "grad_norm": 0.4527474641799927,
      "learning_rate": 9.324045867464218e-06,
      "loss": 0.3864,
      "step": 12035
    },
    {
      "epoch": 0.5523381212427149,
      "grad_norm": 0.44111475348472595,
      "learning_rate": 9.323922753250261e-06,
      "loss": 0.3838,
      "step": 12036
    },
    {
      "epoch": 0.5523840117479694,
      "grad_norm": 0.45739734172821045,
      "learning_rate": 9.323799628638635e-06,
      "loss": 0.3895,
      "step": 12037
    },
    {
      "epoch": 0.5524299022532239,
      "grad_norm": 0.46910691261291504,
      "learning_rate": 9.323676493629637e-06,
      "loss": 0.405,
      "step": 12038
    },
    {
      "epoch": 0.5524757927584782,
      "grad_norm": 0.43242698907852173,
      "learning_rate": 9.32355334822356e-06,
      "loss": 0.3851,
      "step": 12039
    },
    {
      "epoch": 0.5525216832637327,
      "grad_norm": 0.4352954924106598,
      "learning_rate": 9.323430192420703e-06,
      "loss": 0.3579,
      "step": 12040
    },
    {
      "epoch": 0.5525675737689872,
      "grad_norm": 0.4331340789794922,
      "learning_rate": 9.32330702622136e-06,
      "loss": 0.3548,
      "step": 12041
    },
    {
      "epoch": 0.5526134642742416,
      "grad_norm": 0.44352322816848755,
      "learning_rate": 9.323183849625828e-06,
      "loss": 0.3477,
      "step": 12042
    },
    {
      "epoch": 0.5526593547794961,
      "grad_norm": 0.4381205439567566,
      "learning_rate": 9.323060662634403e-06,
      "loss": 0.392,
      "step": 12043
    },
    {
      "epoch": 0.5527052452847506,
      "grad_norm": 0.45821982622146606,
      "learning_rate": 9.322937465247384e-06,
      "loss": 0.445,
      "step": 12044
    },
    {
      "epoch": 0.552751135790005,
      "grad_norm": 0.43494147062301636,
      "learning_rate": 9.322814257465061e-06,
      "loss": 0.3214,
      "step": 12045
    },
    {
      "epoch": 0.5527970262952595,
      "grad_norm": 0.4614538848400116,
      "learning_rate": 9.322691039287736e-06,
      "loss": 0.4518,
      "step": 12046
    },
    {
      "epoch": 0.552842916800514,
      "grad_norm": 0.4773239493370056,
      "learning_rate": 9.322567810715704e-06,
      "loss": 0.4237,
      "step": 12047
    },
    {
      "epoch": 0.5528888073057684,
      "grad_norm": 0.44227465987205505,
      "learning_rate": 9.32244457174926e-06,
      "loss": 0.3585,
      "step": 12048
    },
    {
      "epoch": 0.5529346978110229,
      "grad_norm": 0.46889862418174744,
      "learning_rate": 9.322321322388701e-06,
      "loss": 0.4117,
      "step": 12049
    },
    {
      "epoch": 0.5529805883162774,
      "grad_norm": 0.4665222465991974,
      "learning_rate": 9.322198062634322e-06,
      "loss": 0.3821,
      "step": 12050
    },
    {
      "epoch": 0.5530264788215318,
      "grad_norm": 0.4897397458553314,
      "learning_rate": 9.322074792486421e-06,
      "loss": 0.4116,
      "step": 12051
    },
    {
      "epoch": 0.5530723693267863,
      "grad_norm": 0.4382251501083374,
      "learning_rate": 9.321951511945298e-06,
      "loss": 0.4295,
      "step": 12052
    },
    {
      "epoch": 0.5531182598320408,
      "grad_norm": 0.5055105090141296,
      "learning_rate": 9.321828221011243e-06,
      "loss": 0.469,
      "step": 12053
    },
    {
      "epoch": 0.5531641503372953,
      "grad_norm": 0.45462626218795776,
      "learning_rate": 9.321704919684554e-06,
      "loss": 0.3656,
      "step": 12054
    },
    {
      "epoch": 0.5532100408425497,
      "grad_norm": 0.4514736831188202,
      "learning_rate": 9.32158160796553e-06,
      "loss": 0.3895,
      "step": 12055
    },
    {
      "epoch": 0.5532559313478042,
      "grad_norm": 0.47061610221862793,
      "learning_rate": 9.321458285854465e-06,
      "loss": 0.4257,
      "step": 12056
    },
    {
      "epoch": 0.5533018218530587,
      "grad_norm": 0.4941685199737549,
      "learning_rate": 9.321334953351657e-06,
      "loss": 0.5324,
      "step": 12057
    },
    {
      "epoch": 0.553347712358313,
      "grad_norm": 0.4527307450771332,
      "learning_rate": 9.321211610457404e-06,
      "loss": 0.374,
      "step": 12058
    },
    {
      "epoch": 0.5533936028635675,
      "grad_norm": 0.4510420262813568,
      "learning_rate": 9.321088257172e-06,
      "loss": 0.3715,
      "step": 12059
    },
    {
      "epoch": 0.553439493368822,
      "grad_norm": 0.5284619927406311,
      "learning_rate": 9.320964893495741e-06,
      "loss": 0.4635,
      "step": 12060
    },
    {
      "epoch": 0.5534853838740764,
      "grad_norm": 0.4498647153377533,
      "learning_rate": 9.320841519428927e-06,
      "loss": 0.3746,
      "step": 12061
    },
    {
      "epoch": 0.5535312743793309,
      "grad_norm": 0.45350927114486694,
      "learning_rate": 9.320718134971852e-06,
      "loss": 0.4396,
      "step": 12062
    },
    {
      "epoch": 0.5535771648845854,
      "grad_norm": 0.44705232977867126,
      "learning_rate": 9.320594740124815e-06,
      "loss": 0.3685,
      "step": 12063
    },
    {
      "epoch": 0.5536230553898398,
      "grad_norm": 0.44742101430892944,
      "learning_rate": 9.320471334888108e-06,
      "loss": 0.3825,
      "step": 12064
    },
    {
      "epoch": 0.5536689458950943,
      "grad_norm": 0.47853606939315796,
      "learning_rate": 9.320347919262033e-06,
      "loss": 0.4122,
      "step": 12065
    },
    {
      "epoch": 0.5537148364003488,
      "grad_norm": 0.4651338458061218,
      "learning_rate": 9.320224493246885e-06,
      "loss": 0.3889,
      "step": 12066
    },
    {
      "epoch": 0.5537607269056032,
      "grad_norm": 0.4843061864376068,
      "learning_rate": 9.32010105684296e-06,
      "loss": 0.4366,
      "step": 12067
    },
    {
      "epoch": 0.5538066174108577,
      "grad_norm": 0.46956324577331543,
      "learning_rate": 9.319977610050556e-06,
      "loss": 0.3797,
      "step": 12068
    },
    {
      "epoch": 0.5538525079161122,
      "grad_norm": 0.4895978271961212,
      "learning_rate": 9.31985415286997e-06,
      "loss": 0.4515,
      "step": 12069
    },
    {
      "epoch": 0.5538983984213666,
      "grad_norm": 0.4545944929122925,
      "learning_rate": 9.319730685301498e-06,
      "loss": 0.3712,
      "step": 12070
    },
    {
      "epoch": 0.5539442889266211,
      "grad_norm": 0.4647114872932434,
      "learning_rate": 9.319607207345436e-06,
      "loss": 0.4646,
      "step": 12071
    },
    {
      "epoch": 0.5539901794318756,
      "grad_norm": 0.4699457883834839,
      "learning_rate": 9.319483719002082e-06,
      "loss": 0.4306,
      "step": 12072
    },
    {
      "epoch": 0.5540360699371301,
      "grad_norm": 0.48452994227409363,
      "learning_rate": 9.319360220271734e-06,
      "loss": 0.4109,
      "step": 12073
    },
    {
      "epoch": 0.5540819604423844,
      "grad_norm": 0.48863694071769714,
      "learning_rate": 9.319236711154687e-06,
      "loss": 0.4247,
      "step": 12074
    },
    {
      "epoch": 0.554127850947639,
      "grad_norm": 0.4846033751964569,
      "learning_rate": 9.319113191651239e-06,
      "loss": 0.4456,
      "step": 12075
    },
    {
      "epoch": 0.5541737414528934,
      "grad_norm": 0.46660470962524414,
      "learning_rate": 9.318989661761686e-06,
      "loss": 0.4015,
      "step": 12076
    },
    {
      "epoch": 0.5542196319581478,
      "grad_norm": 0.4736575186252594,
      "learning_rate": 9.318866121486327e-06,
      "loss": 0.3727,
      "step": 12077
    },
    {
      "epoch": 0.5542655224634023,
      "grad_norm": 0.43606826663017273,
      "learning_rate": 9.318742570825458e-06,
      "loss": 0.3333,
      "step": 12078
    },
    {
      "epoch": 0.5543114129686568,
      "grad_norm": 0.48082706332206726,
      "learning_rate": 9.318619009779378e-06,
      "loss": 0.4525,
      "step": 12079
    },
    {
      "epoch": 0.5543573034739112,
      "grad_norm": 0.5324326157569885,
      "learning_rate": 9.31849543834838e-06,
      "loss": 0.5368,
      "step": 12080
    },
    {
      "epoch": 0.5544031939791657,
      "grad_norm": 0.4841443598270416,
      "learning_rate": 9.318371856532763e-06,
      "loss": 0.4529,
      "step": 12081
    },
    {
      "epoch": 0.5544490844844202,
      "grad_norm": 0.48112794756889343,
      "learning_rate": 9.318248264332826e-06,
      "loss": 0.4459,
      "step": 12082
    },
    {
      "epoch": 0.5544949749896746,
      "grad_norm": 0.4488029181957245,
      "learning_rate": 9.318124661748865e-06,
      "loss": 0.361,
      "step": 12083
    },
    {
      "epoch": 0.5545408654949291,
      "grad_norm": 0.48441484570503235,
      "learning_rate": 9.318001048781175e-06,
      "loss": 0.4346,
      "step": 12084
    },
    {
      "epoch": 0.5545867560001836,
      "grad_norm": 0.492082804441452,
      "learning_rate": 9.317877425430058e-06,
      "loss": 0.4038,
      "step": 12085
    },
    {
      "epoch": 0.554632646505438,
      "grad_norm": 0.4680447280406952,
      "learning_rate": 9.317753791695806e-06,
      "loss": 0.4387,
      "step": 12086
    },
    {
      "epoch": 0.5546785370106925,
      "grad_norm": 0.46267813444137573,
      "learning_rate": 9.31763014757872e-06,
      "loss": 0.4013,
      "step": 12087
    },
    {
      "epoch": 0.554724427515947,
      "grad_norm": 0.4445572793483734,
      "learning_rate": 9.317506493079097e-06,
      "loss": 0.3491,
      "step": 12088
    },
    {
      "epoch": 0.5547703180212014,
      "grad_norm": 0.4425163269042969,
      "learning_rate": 9.317382828197232e-06,
      "loss": 0.3568,
      "step": 12089
    },
    {
      "epoch": 0.5548162085264559,
      "grad_norm": 1.0561143159866333,
      "learning_rate": 9.317259152933424e-06,
      "loss": 0.4236,
      "step": 12090
    },
    {
      "epoch": 0.5548620990317104,
      "grad_norm": 0.5127905607223511,
      "learning_rate": 9.317135467287972e-06,
      "loss": 0.5058,
      "step": 12091
    },
    {
      "epoch": 0.5549079895369649,
      "grad_norm": 0.49167007207870483,
      "learning_rate": 9.31701177126117e-06,
      "loss": 0.4713,
      "step": 12092
    },
    {
      "epoch": 0.5549538800422192,
      "grad_norm": 0.461435467004776,
      "learning_rate": 9.316888064853318e-06,
      "loss": 0.3655,
      "step": 12093
    },
    {
      "epoch": 0.5549997705474737,
      "grad_norm": 0.4498245418071747,
      "learning_rate": 9.316764348064713e-06,
      "loss": 0.393,
      "step": 12094
    },
    {
      "epoch": 0.5550456610527282,
      "grad_norm": 0.44873034954071045,
      "learning_rate": 9.316640620895651e-06,
      "loss": 0.3561,
      "step": 12095
    },
    {
      "epoch": 0.5550915515579826,
      "grad_norm": 0.43564528226852417,
      "learning_rate": 9.316516883346432e-06,
      "loss": 0.3666,
      "step": 12096
    },
    {
      "epoch": 0.5551374420632371,
      "grad_norm": 0.4924660623073578,
      "learning_rate": 9.316393135417351e-06,
      "loss": 0.4461,
      "step": 12097
    },
    {
      "epoch": 0.5551833325684916,
      "grad_norm": 0.9832836389541626,
      "learning_rate": 9.316269377108707e-06,
      "loss": 0.4736,
      "step": 12098
    },
    {
      "epoch": 0.555229223073746,
      "grad_norm": 0.5583959817886353,
      "learning_rate": 9.316145608420798e-06,
      "loss": 0.4646,
      "step": 12099
    },
    {
      "epoch": 0.5552751135790005,
      "grad_norm": 0.4562219977378845,
      "learning_rate": 9.31602182935392e-06,
      "loss": 0.388,
      "step": 12100
    },
    {
      "epoch": 0.555321004084255,
      "grad_norm": 0.5044378042221069,
      "learning_rate": 9.315898039908374e-06,
      "loss": 0.4816,
      "step": 12101
    },
    {
      "epoch": 0.5553668945895094,
      "grad_norm": 0.44881659746170044,
      "learning_rate": 9.315774240084452e-06,
      "loss": 0.376,
      "step": 12102
    },
    {
      "epoch": 0.5554127850947639,
      "grad_norm": 0.5203543901443481,
      "learning_rate": 9.315650429882458e-06,
      "loss": 0.4987,
      "step": 12103
    },
    {
      "epoch": 0.5554586756000184,
      "grad_norm": 0.4803381860256195,
      "learning_rate": 9.315526609302685e-06,
      "loss": 0.4262,
      "step": 12104
    },
    {
      "epoch": 0.5555045661052728,
      "grad_norm": 0.44015568494796753,
      "learning_rate": 9.315402778345432e-06,
      "loss": 0.3828,
      "step": 12105
    },
    {
      "epoch": 0.5555504566105273,
      "grad_norm": 0.5461061000823975,
      "learning_rate": 9.315278937011e-06,
      "loss": 0.4579,
      "step": 12106
    },
    {
      "epoch": 0.5555963471157818,
      "grad_norm": 0.48139676451683044,
      "learning_rate": 9.315155085299683e-06,
      "loss": 0.4675,
      "step": 12107
    },
    {
      "epoch": 0.5556422376210363,
      "grad_norm": 0.4233253300189972,
      "learning_rate": 9.315031223211779e-06,
      "loss": 0.3417,
      "step": 12108
    },
    {
      "epoch": 0.5556881281262906,
      "grad_norm": 0.4718100428581238,
      "learning_rate": 9.314907350747585e-06,
      "loss": 0.4456,
      "step": 12109
    },
    {
      "epoch": 0.5557340186315451,
      "grad_norm": 0.45583781599998474,
      "learning_rate": 9.314783467907405e-06,
      "loss": 0.3532,
      "step": 12110
    },
    {
      "epoch": 0.5557799091367996,
      "grad_norm": 0.45792990922927856,
      "learning_rate": 9.31465957469153e-06,
      "loss": 0.3881,
      "step": 12111
    },
    {
      "epoch": 0.555825799642054,
      "grad_norm": 0.4594878554344177,
      "learning_rate": 9.314535671100261e-06,
      "loss": 0.4063,
      "step": 12112
    },
    {
      "epoch": 0.5558716901473085,
      "grad_norm": 0.466871440410614,
      "learning_rate": 9.314411757133895e-06,
      "loss": 0.3855,
      "step": 12113
    },
    {
      "epoch": 0.555917580652563,
      "grad_norm": 0.4323359727859497,
      "learning_rate": 9.31428783279273e-06,
      "loss": 0.3189,
      "step": 12114
    },
    {
      "epoch": 0.5559634711578174,
      "grad_norm": 0.4370236098766327,
      "learning_rate": 9.314163898077067e-06,
      "loss": 0.3538,
      "step": 12115
    },
    {
      "epoch": 0.5560093616630719,
      "grad_norm": 0.4464944303035736,
      "learning_rate": 9.314039952987198e-06,
      "loss": 0.4118,
      "step": 12116
    },
    {
      "epoch": 0.5560552521683264,
      "grad_norm": 0.45803695917129517,
      "learning_rate": 9.313915997523426e-06,
      "loss": 0.3951,
      "step": 12117
    },
    {
      "epoch": 0.5561011426735808,
      "grad_norm": 0.8699121475219727,
      "learning_rate": 9.31379203168605e-06,
      "loss": 0.4238,
      "step": 12118
    },
    {
      "epoch": 0.5561470331788353,
      "grad_norm": 0.4901309609413147,
      "learning_rate": 9.313668055475362e-06,
      "loss": 0.4327,
      "step": 12119
    },
    {
      "epoch": 0.5561929236840898,
      "grad_norm": 0.46953085064888,
      "learning_rate": 9.313544068891665e-06,
      "loss": 0.4424,
      "step": 12120
    },
    {
      "epoch": 0.5562388141893442,
      "grad_norm": 0.4544709622859955,
      "learning_rate": 9.313420071935255e-06,
      "loss": 0.3977,
      "step": 12121
    },
    {
      "epoch": 0.5562847046945987,
      "grad_norm": 0.4466981291770935,
      "learning_rate": 9.313296064606432e-06,
      "loss": 0.4434,
      "step": 12122
    },
    {
      "epoch": 0.5563305951998532,
      "grad_norm": 0.5235185623168945,
      "learning_rate": 9.313172046905494e-06,
      "loss": 0.4996,
      "step": 12123
    },
    {
      "epoch": 0.5563764857051076,
      "grad_norm": 0.44973328709602356,
      "learning_rate": 9.313048018832736e-06,
      "loss": 0.3798,
      "step": 12124
    },
    {
      "epoch": 0.5564223762103621,
      "grad_norm": 0.4175124764442444,
      "learning_rate": 9.31292398038846e-06,
      "loss": 0.3282,
      "step": 12125
    },
    {
      "epoch": 0.5564682667156166,
      "grad_norm": 0.47038909792900085,
      "learning_rate": 9.312799931572965e-06,
      "loss": 0.414,
      "step": 12126
    },
    {
      "epoch": 0.556514157220871,
      "grad_norm": 0.45200538635253906,
      "learning_rate": 9.312675872386546e-06,
      "loss": 0.4343,
      "step": 12127
    },
    {
      "epoch": 0.5565600477261254,
      "grad_norm": 0.4696628451347351,
      "learning_rate": 9.312551802829502e-06,
      "loss": 0.4272,
      "step": 12128
    },
    {
      "epoch": 0.5566059382313799,
      "grad_norm": 0.4801579415798187,
      "learning_rate": 9.312427722902132e-06,
      "loss": 0.4942,
      "step": 12129
    },
    {
      "epoch": 0.5566518287366344,
      "grad_norm": 0.4296312928199768,
      "learning_rate": 9.312303632604734e-06,
      "loss": 0.3275,
      "step": 12130
    },
    {
      "epoch": 0.5566977192418888,
      "grad_norm": 0.48296621441841125,
      "learning_rate": 9.312179531937609e-06,
      "loss": 0.4251,
      "step": 12131
    },
    {
      "epoch": 0.5567436097471433,
      "grad_norm": 0.49556785821914673,
      "learning_rate": 9.31205542090105e-06,
      "loss": 0.3913,
      "step": 12132
    },
    {
      "epoch": 0.5567895002523978,
      "grad_norm": 0.4593547284603119,
      "learning_rate": 9.31193129949536e-06,
      "loss": 0.3791,
      "step": 12133
    },
    {
      "epoch": 0.5568353907576522,
      "grad_norm": 0.4720228910446167,
      "learning_rate": 9.311807167720838e-06,
      "loss": 0.4064,
      "step": 12134
    },
    {
      "epoch": 0.5568812812629067,
      "grad_norm": 0.4836314022541046,
      "learning_rate": 9.311683025577777e-06,
      "loss": 0.4393,
      "step": 12135
    },
    {
      "epoch": 0.5569271717681612,
      "grad_norm": 0.43410229682922363,
      "learning_rate": 9.311558873066482e-06,
      "loss": 0.3565,
      "step": 12136
    },
    {
      "epoch": 0.5569730622734156,
      "grad_norm": 0.46988412737846375,
      "learning_rate": 9.311434710187248e-06,
      "loss": 0.3803,
      "step": 12137
    },
    {
      "epoch": 0.5570189527786701,
      "grad_norm": 0.485321968793869,
      "learning_rate": 9.311310536940374e-06,
      "loss": 0.4171,
      "step": 12138
    },
    {
      "epoch": 0.5570648432839246,
      "grad_norm": 0.45082175731658936,
      "learning_rate": 9.31118635332616e-06,
      "loss": 0.3735,
      "step": 12139
    },
    {
      "epoch": 0.557110733789179,
      "grad_norm": 0.4475308060646057,
      "learning_rate": 9.311062159344903e-06,
      "loss": 0.3761,
      "step": 12140
    },
    {
      "epoch": 0.5571566242944335,
      "grad_norm": 0.4941766560077667,
      "learning_rate": 9.310937954996902e-06,
      "loss": 0.4191,
      "step": 12141
    },
    {
      "epoch": 0.557202514799688,
      "grad_norm": 0.4788995087146759,
      "learning_rate": 9.310813740282453e-06,
      "loss": 0.4459,
      "step": 12142
    },
    {
      "epoch": 0.5572484053049425,
      "grad_norm": 0.4992770552635193,
      "learning_rate": 9.310689515201861e-06,
      "loss": 0.4888,
      "step": 12143
    },
    {
      "epoch": 0.5572942958101968,
      "grad_norm": 0.4674097001552582,
      "learning_rate": 9.31056527975542e-06,
      "loss": 0.3727,
      "step": 12144
    },
    {
      "epoch": 0.5573401863154513,
      "grad_norm": 0.48826098442077637,
      "learning_rate": 9.310441033943431e-06,
      "loss": 0.4298,
      "step": 12145
    },
    {
      "epoch": 0.5573860768207058,
      "grad_norm": 0.5561777949333191,
      "learning_rate": 9.31031677776619e-06,
      "loss": 0.5093,
      "step": 12146
    },
    {
      "epoch": 0.5574319673259602,
      "grad_norm": 0.46843668818473816,
      "learning_rate": 9.310192511223997e-06,
      "loss": 0.4028,
      "step": 12147
    },
    {
      "epoch": 0.5574778578312147,
      "grad_norm": 0.4768430292606354,
      "learning_rate": 9.310068234317153e-06,
      "loss": 0.3792,
      "step": 12148
    },
    {
      "epoch": 0.5575237483364692,
      "grad_norm": 0.47033795714378357,
      "learning_rate": 9.309943947045956e-06,
      "loss": 0.3974,
      "step": 12149
    },
    {
      "epoch": 0.5575696388417236,
      "grad_norm": 0.4621320068836212,
      "learning_rate": 9.309819649410703e-06,
      "loss": 0.3748,
      "step": 12150
    },
    {
      "epoch": 0.5576155293469781,
      "grad_norm": 0.4729558527469635,
      "learning_rate": 9.309695341411693e-06,
      "loss": 0.353,
      "step": 12151
    },
    {
      "epoch": 0.5576614198522326,
      "grad_norm": 0.4460930824279785,
      "learning_rate": 9.309571023049227e-06,
      "loss": 0.3797,
      "step": 12152
    },
    {
      "epoch": 0.557707310357487,
      "grad_norm": 0.4192441403865814,
      "learning_rate": 9.309446694323603e-06,
      "loss": 0.3148,
      "step": 12153
    },
    {
      "epoch": 0.5577532008627415,
      "grad_norm": 0.4007530212402344,
      "learning_rate": 9.30932235523512e-06,
      "loss": 0.2906,
      "step": 12154
    },
    {
      "epoch": 0.557799091367996,
      "grad_norm": 0.479134202003479,
      "learning_rate": 9.309198005784075e-06,
      "loss": 0.4242,
      "step": 12155
    },
    {
      "epoch": 0.5578449818732504,
      "grad_norm": 0.4199838936328888,
      "learning_rate": 9.30907364597077e-06,
      "loss": 0.3325,
      "step": 12156
    },
    {
      "epoch": 0.5578908723785049,
      "grad_norm": 0.45139381289482117,
      "learning_rate": 9.308949275795502e-06,
      "loss": 0.3935,
      "step": 12157
    },
    {
      "epoch": 0.5579367628837594,
      "grad_norm": 0.4335850179195404,
      "learning_rate": 9.308824895258572e-06,
      "loss": 0.3441,
      "step": 12158
    },
    {
      "epoch": 0.5579826533890138,
      "grad_norm": 0.4658150374889374,
      "learning_rate": 9.308700504360277e-06,
      "loss": 0.3731,
      "step": 12159
    },
    {
      "epoch": 0.5580285438942683,
      "grad_norm": 0.4783141314983368,
      "learning_rate": 9.308576103100918e-06,
      "loss": 0.4304,
      "step": 12160
    },
    {
      "epoch": 0.5580744343995228,
      "grad_norm": 0.46721163392066956,
      "learning_rate": 9.308451691480795e-06,
      "loss": 0.4088,
      "step": 12161
    },
    {
      "epoch": 0.5581203249047773,
      "grad_norm": 0.43630751967430115,
      "learning_rate": 9.308327269500204e-06,
      "loss": 0.3263,
      "step": 12162
    },
    {
      "epoch": 0.5581662154100316,
      "grad_norm": 0.4469703435897827,
      "learning_rate": 9.308202837159444e-06,
      "loss": 0.3913,
      "step": 12163
    },
    {
      "epoch": 0.5582121059152861,
      "grad_norm": 0.5071561932563782,
      "learning_rate": 9.308078394458817e-06,
      "loss": 0.4468,
      "step": 12164
    },
    {
      "epoch": 0.5582579964205406,
      "grad_norm": 0.4541645646095276,
      "learning_rate": 9.307953941398622e-06,
      "loss": 0.3974,
      "step": 12165
    },
    {
      "epoch": 0.558303886925795,
      "grad_norm": 0.5984510183334351,
      "learning_rate": 9.307829477979158e-06,
      "loss": 0.4534,
      "step": 12166
    },
    {
      "epoch": 0.5583497774310495,
      "grad_norm": 0.5287955403327942,
      "learning_rate": 9.30770500420072e-06,
      "loss": 0.322,
      "step": 12167
    },
    {
      "epoch": 0.558395667936304,
      "grad_norm": 0.45505374670028687,
      "learning_rate": 9.307580520063616e-06,
      "loss": 0.4231,
      "step": 12168
    },
    {
      "epoch": 0.5584415584415584,
      "grad_norm": 0.4570363163948059,
      "learning_rate": 9.307456025568135e-06,
      "loss": 0.408,
      "step": 12169
    },
    {
      "epoch": 0.5584874489468129,
      "grad_norm": 0.4292166829109192,
      "learning_rate": 9.307331520714586e-06,
      "loss": 0.3278,
      "step": 12170
    },
    {
      "epoch": 0.5585333394520674,
      "grad_norm": 0.464189350605011,
      "learning_rate": 9.307207005503263e-06,
      "loss": 0.4442,
      "step": 12171
    },
    {
      "epoch": 0.5585792299573218,
      "grad_norm": 0.42920994758605957,
      "learning_rate": 9.307082479934467e-06,
      "loss": 0.3387,
      "step": 12172
    },
    {
      "epoch": 0.5586251204625763,
      "grad_norm": 0.47448137402534485,
      "learning_rate": 9.306957944008494e-06,
      "loss": 0.4054,
      "step": 12173
    },
    {
      "epoch": 0.5586710109678308,
      "grad_norm": 0.49340492486953735,
      "learning_rate": 9.30683339772565e-06,
      "loss": 0.4315,
      "step": 12174
    },
    {
      "epoch": 0.5587169014730852,
      "grad_norm": 0.5011741518974304,
      "learning_rate": 9.30670884108623e-06,
      "loss": 0.5097,
      "step": 12175
    },
    {
      "epoch": 0.5587627919783397,
      "grad_norm": 0.45290327072143555,
      "learning_rate": 9.306584274090534e-06,
      "loss": 0.3587,
      "step": 12176
    },
    {
      "epoch": 0.5588086824835942,
      "grad_norm": 0.4752029478549957,
      "learning_rate": 9.306459696738861e-06,
      "loss": 0.4603,
      "step": 12177
    },
    {
      "epoch": 0.5588545729888486,
      "grad_norm": 0.5530535578727722,
      "learning_rate": 9.306335109031514e-06,
      "loss": 0.322,
      "step": 12178
    },
    {
      "epoch": 0.558900463494103,
      "grad_norm": 0.5102943778038025,
      "learning_rate": 9.306210510968788e-06,
      "loss": 0.4541,
      "step": 12179
    },
    {
      "epoch": 0.5589463539993575,
      "grad_norm": 0.48672640323638916,
      "learning_rate": 9.306085902550986e-06,
      "loss": 0.4229,
      "step": 12180
    },
    {
      "epoch": 0.558992244504612,
      "grad_norm": 0.4599076509475708,
      "learning_rate": 9.305961283778406e-06,
      "loss": 0.4195,
      "step": 12181
    },
    {
      "epoch": 0.5590381350098664,
      "grad_norm": 0.48828449845314026,
      "learning_rate": 9.305836654651348e-06,
      "loss": 0.4713,
      "step": 12182
    },
    {
      "epoch": 0.5590840255151209,
      "grad_norm": 0.4966793656349182,
      "learning_rate": 9.305712015170112e-06,
      "loss": 0.4515,
      "step": 12183
    },
    {
      "epoch": 0.5591299160203754,
      "grad_norm": 0.46091508865356445,
      "learning_rate": 9.305587365334997e-06,
      "loss": 0.4148,
      "step": 12184
    },
    {
      "epoch": 0.5591758065256298,
      "grad_norm": 0.4421173930168152,
      "learning_rate": 9.305462705146306e-06,
      "loss": 0.3374,
      "step": 12185
    },
    {
      "epoch": 0.5592216970308843,
      "grad_norm": 0.44091668725013733,
      "learning_rate": 9.305338034604333e-06,
      "loss": 0.3312,
      "step": 12186
    },
    {
      "epoch": 0.5592675875361388,
      "grad_norm": 0.481839656829834,
      "learning_rate": 9.30521335370938e-06,
      "loss": 0.3931,
      "step": 12187
    },
    {
      "epoch": 0.5593134780413932,
      "grad_norm": 0.4522998034954071,
      "learning_rate": 9.305088662461751e-06,
      "loss": 0.3946,
      "step": 12188
    },
    {
      "epoch": 0.5593593685466477,
      "grad_norm": 0.4550017714500427,
      "learning_rate": 9.304963960861741e-06,
      "loss": 0.3488,
      "step": 12189
    },
    {
      "epoch": 0.5594052590519022,
      "grad_norm": 0.45765042304992676,
      "learning_rate": 9.304839248909652e-06,
      "loss": 0.3939,
      "step": 12190
    },
    {
      "epoch": 0.5594511495571566,
      "grad_norm": 0.46712562441825867,
      "learning_rate": 9.304714526605783e-06,
      "loss": 0.451,
      "step": 12191
    },
    {
      "epoch": 0.5594970400624111,
      "grad_norm": 0.46571141481399536,
      "learning_rate": 9.304589793950433e-06,
      "loss": 0.4242,
      "step": 12192
    },
    {
      "epoch": 0.5595429305676656,
      "grad_norm": 0.45479685068130493,
      "learning_rate": 9.304465050943904e-06,
      "loss": 0.3531,
      "step": 12193
    },
    {
      "epoch": 0.55958882107292,
      "grad_norm": 0.4918900728225708,
      "learning_rate": 9.304340297586494e-06,
      "loss": 0.4248,
      "step": 12194
    },
    {
      "epoch": 0.5596347115781745,
      "grad_norm": 0.4176943302154541,
      "learning_rate": 9.304215533878505e-06,
      "loss": 0.3563,
      "step": 12195
    },
    {
      "epoch": 0.559680602083429,
      "grad_norm": 0.44797900319099426,
      "learning_rate": 9.304090759820237e-06,
      "loss": 0.4209,
      "step": 12196
    },
    {
      "epoch": 0.5597264925886835,
      "grad_norm": 0.4970931112766266,
      "learning_rate": 9.303965975411987e-06,
      "loss": 0.5208,
      "step": 12197
    },
    {
      "epoch": 0.5597723830939378,
      "grad_norm": 0.43813297152519226,
      "learning_rate": 9.303841180654058e-06,
      "loss": 0.3997,
      "step": 12198
    },
    {
      "epoch": 0.5598182735991923,
      "grad_norm": 0.4533596634864807,
      "learning_rate": 9.303716375546749e-06,
      "loss": 0.4079,
      "step": 12199
    },
    {
      "epoch": 0.5598641641044468,
      "grad_norm": 0.4552463889122009,
      "learning_rate": 9.30359156009036e-06,
      "loss": 0.4388,
      "step": 12200
    },
    {
      "epoch": 0.5599100546097012,
      "grad_norm": 0.42153000831604004,
      "learning_rate": 9.303466734285195e-06,
      "loss": 0.3277,
      "step": 12201
    },
    {
      "epoch": 0.5599559451149557,
      "grad_norm": 0.4840945601463318,
      "learning_rate": 9.303341898131548e-06,
      "loss": 0.3911,
      "step": 12202
    },
    {
      "epoch": 0.5600018356202102,
      "grad_norm": 0.44774964451789856,
      "learning_rate": 9.303217051629722e-06,
      "loss": 0.3833,
      "step": 12203
    },
    {
      "epoch": 0.5600477261254646,
      "grad_norm": 0.4361903965473175,
      "learning_rate": 9.303092194780018e-06,
      "loss": 0.383,
      "step": 12204
    },
    {
      "epoch": 0.5600936166307191,
      "grad_norm": 0.44973692297935486,
      "learning_rate": 9.302967327582735e-06,
      "loss": 0.3897,
      "step": 12205
    },
    {
      "epoch": 0.5601395071359736,
      "grad_norm": 0.48008444905281067,
      "learning_rate": 9.302842450038173e-06,
      "loss": 0.4185,
      "step": 12206
    },
    {
      "epoch": 0.560185397641228,
      "grad_norm": 0.4797430634498596,
      "learning_rate": 9.302717562146632e-06,
      "loss": 0.4304,
      "step": 12207
    },
    {
      "epoch": 0.5602312881464825,
      "grad_norm": 0.43292778730392456,
      "learning_rate": 9.302592663908416e-06,
      "loss": 0.3752,
      "step": 12208
    },
    {
      "epoch": 0.560277178651737,
      "grad_norm": 0.4630085229873657,
      "learning_rate": 9.30246775532382e-06,
      "loss": 0.3733,
      "step": 12209
    },
    {
      "epoch": 0.5603230691569914,
      "grad_norm": 0.4516647160053253,
      "learning_rate": 9.302342836393148e-06,
      "loss": 0.3871,
      "step": 12210
    },
    {
      "epoch": 0.5603689596622459,
      "grad_norm": 0.4639822542667389,
      "learning_rate": 9.3022179071167e-06,
      "loss": 0.3882,
      "step": 12211
    },
    {
      "epoch": 0.5604148501675004,
      "grad_norm": 0.4666505753993988,
      "learning_rate": 9.302092967494776e-06,
      "loss": 0.444,
      "step": 12212
    },
    {
      "epoch": 0.5604607406727548,
      "grad_norm": 0.4380147457122803,
      "learning_rate": 9.301968017527676e-06,
      "loss": 0.3707,
      "step": 12213
    },
    {
      "epoch": 0.5605066311780093,
      "grad_norm": 0.43364036083221436,
      "learning_rate": 9.301843057215699e-06,
      "loss": 0.3832,
      "step": 12214
    },
    {
      "epoch": 0.5605525216832637,
      "grad_norm": 0.4586067497730255,
      "learning_rate": 9.301718086559148e-06,
      "loss": 0.3907,
      "step": 12215
    },
    {
      "epoch": 0.5605984121885182,
      "grad_norm": 0.4557298421859741,
      "learning_rate": 9.301593105558322e-06,
      "loss": 0.3968,
      "step": 12216
    },
    {
      "epoch": 0.5606443026937726,
      "grad_norm": 0.4413805603981018,
      "learning_rate": 9.301468114213523e-06,
      "loss": 0.3451,
      "step": 12217
    },
    {
      "epoch": 0.5606901931990271,
      "grad_norm": 0.47997286915779114,
      "learning_rate": 9.301343112525052e-06,
      "loss": 0.3989,
      "step": 12218
    },
    {
      "epoch": 0.5607360837042816,
      "grad_norm": 0.43419668078422546,
      "learning_rate": 9.301218100493206e-06,
      "loss": 0.3255,
      "step": 12219
    },
    {
      "epoch": 0.560781974209536,
      "grad_norm": 0.4807821214199066,
      "learning_rate": 9.30109307811829e-06,
      "loss": 0.4354,
      "step": 12220
    },
    {
      "epoch": 0.5608278647147905,
      "grad_norm": 0.6913928985595703,
      "learning_rate": 9.300968045400602e-06,
      "loss": 0.4538,
      "step": 12221
    },
    {
      "epoch": 0.560873755220045,
      "grad_norm": 0.4109992980957031,
      "learning_rate": 9.300843002340444e-06,
      "loss": 0.2966,
      "step": 12222
    },
    {
      "epoch": 0.5609196457252994,
      "grad_norm": 0.4806285798549652,
      "learning_rate": 9.300717948938114e-06,
      "loss": 0.4498,
      "step": 12223
    },
    {
      "epoch": 0.5609655362305539,
      "grad_norm": 0.4772484600543976,
      "learning_rate": 9.300592885193915e-06,
      "loss": 0.3953,
      "step": 12224
    },
    {
      "epoch": 0.5610114267358084,
      "grad_norm": 0.6011549830436707,
      "learning_rate": 9.300467811108149e-06,
      "loss": 0.3362,
      "step": 12225
    },
    {
      "epoch": 0.5610573172410628,
      "grad_norm": 0.4709915518760681,
      "learning_rate": 9.300342726681113e-06,
      "loss": 0.4224,
      "step": 12226
    },
    {
      "epoch": 0.5611032077463173,
      "grad_norm": 0.4223760962486267,
      "learning_rate": 9.30021763191311e-06,
      "loss": 0.3136,
      "step": 12227
    },
    {
      "epoch": 0.5611490982515718,
      "grad_norm": 0.4856641888618469,
      "learning_rate": 9.300092526804443e-06,
      "loss": 0.4319,
      "step": 12228
    },
    {
      "epoch": 0.5611949887568262,
      "grad_norm": 0.4588739573955536,
      "learning_rate": 9.29996741135541e-06,
      "loss": 0.4092,
      "step": 12229
    },
    {
      "epoch": 0.5612408792620807,
      "grad_norm": 0.4486434757709503,
      "learning_rate": 9.299842285566311e-06,
      "loss": 0.3413,
      "step": 12230
    },
    {
      "epoch": 0.5612867697673352,
      "grad_norm": 0.46499741077423096,
      "learning_rate": 9.29971714943745e-06,
      "loss": 0.4061,
      "step": 12231
    },
    {
      "epoch": 0.5613326602725897,
      "grad_norm": 0.47625085711479187,
      "learning_rate": 9.299592002969126e-06,
      "loss": 0.4565,
      "step": 12232
    },
    {
      "epoch": 0.561378550777844,
      "grad_norm": 0.45495492219924927,
      "learning_rate": 9.29946684616164e-06,
      "loss": 0.4314,
      "step": 12233
    },
    {
      "epoch": 0.5614244412830985,
      "grad_norm": 0.4289122223854065,
      "learning_rate": 9.299341679015293e-06,
      "loss": 0.3139,
      "step": 12234
    },
    {
      "epoch": 0.561470331788353,
      "grad_norm": 0.4708121120929718,
      "learning_rate": 9.299216501530386e-06,
      "loss": 0.355,
      "step": 12235
    },
    {
      "epoch": 0.5615162222936074,
      "grad_norm": 0.49518176913261414,
      "learning_rate": 9.29909131370722e-06,
      "loss": 0.5086,
      "step": 12236
    },
    {
      "epoch": 0.5615621127988619,
      "grad_norm": 0.45498061180114746,
      "learning_rate": 9.298966115546098e-06,
      "loss": 0.4002,
      "step": 12237
    },
    {
      "epoch": 0.5616080033041164,
      "grad_norm": 0.5243152379989624,
      "learning_rate": 9.298840907047318e-06,
      "loss": 0.5179,
      "step": 12238
    },
    {
      "epoch": 0.5616538938093708,
      "grad_norm": 0.44022777676582336,
      "learning_rate": 9.298715688211182e-06,
      "loss": 0.3185,
      "step": 12239
    },
    {
      "epoch": 0.5616997843146253,
      "grad_norm": 0.4458248019218445,
      "learning_rate": 9.298590459037993e-06,
      "loss": 0.3831,
      "step": 12240
    },
    {
      "epoch": 0.5617456748198798,
      "grad_norm": 0.4435526132583618,
      "learning_rate": 9.298465219528048e-06,
      "loss": 0.3782,
      "step": 12241
    },
    {
      "epoch": 0.5617915653251342,
      "grad_norm": 0.4742283225059509,
      "learning_rate": 9.298339969681653e-06,
      "loss": 0.4295,
      "step": 12242
    },
    {
      "epoch": 0.5618374558303887,
      "grad_norm": 0.4798366129398346,
      "learning_rate": 9.298214709499107e-06,
      "loss": 0.4263,
      "step": 12243
    },
    {
      "epoch": 0.5618833463356432,
      "grad_norm": 0.4598684310913086,
      "learning_rate": 9.29808943898071e-06,
      "loss": 0.425,
      "step": 12244
    },
    {
      "epoch": 0.5619292368408976,
      "grad_norm": 0.42091506719589233,
      "learning_rate": 9.297964158126765e-06,
      "loss": 0.3519,
      "step": 12245
    },
    {
      "epoch": 0.5619751273461521,
      "grad_norm": 0.45386961102485657,
      "learning_rate": 9.297838866937573e-06,
      "loss": 0.4419,
      "step": 12246
    },
    {
      "epoch": 0.5620210178514066,
      "grad_norm": 0.49147486686706543,
      "learning_rate": 9.297713565413433e-06,
      "loss": 0.4432,
      "step": 12247
    },
    {
      "epoch": 0.562066908356661,
      "grad_norm": 0.5244831442832947,
      "learning_rate": 9.29758825355465e-06,
      "loss": 0.4525,
      "step": 12248
    },
    {
      "epoch": 0.5621127988619155,
      "grad_norm": 0.4447228014469147,
      "learning_rate": 9.297462931361523e-06,
      "loss": 0.368,
      "step": 12249
    },
    {
      "epoch": 0.56215868936717,
      "grad_norm": 0.459503710269928,
      "learning_rate": 9.297337598834355e-06,
      "loss": 0.4391,
      "step": 12250
    },
    {
      "epoch": 0.5622045798724244,
      "grad_norm": 0.4367789924144745,
      "learning_rate": 9.297212255973446e-06,
      "loss": 0.3537,
      "step": 12251
    },
    {
      "epoch": 0.5622504703776788,
      "grad_norm": 0.47100570797920227,
      "learning_rate": 9.297086902779097e-06,
      "loss": 0.4811,
      "step": 12252
    },
    {
      "epoch": 0.5622963608829333,
      "grad_norm": 0.4529896378517151,
      "learning_rate": 9.29696153925161e-06,
      "loss": 0.3592,
      "step": 12253
    },
    {
      "epoch": 0.5623422513881878,
      "grad_norm": 0.4349445700645447,
      "learning_rate": 9.296836165391288e-06,
      "loss": 0.3587,
      "step": 12254
    },
    {
      "epoch": 0.5623881418934422,
      "grad_norm": 0.4545430839061737,
      "learning_rate": 9.29671078119843e-06,
      "loss": 0.414,
      "step": 12255
    },
    {
      "epoch": 0.5624340323986967,
      "grad_norm": 0.45265355706214905,
      "learning_rate": 9.29658538667334e-06,
      "loss": 0.3859,
      "step": 12256
    },
    {
      "epoch": 0.5624799229039512,
      "grad_norm": 0.47099509835243225,
      "learning_rate": 9.296459981816317e-06,
      "loss": 0.403,
      "step": 12257
    },
    {
      "epoch": 0.5625258134092056,
      "grad_norm": 0.4402739703655243,
      "learning_rate": 9.296334566627663e-06,
      "loss": 0.3782,
      "step": 12258
    },
    {
      "epoch": 0.5625717039144601,
      "grad_norm": 0.42116811871528625,
      "learning_rate": 9.296209141107682e-06,
      "loss": 0.3204,
      "step": 12259
    },
    {
      "epoch": 0.5626175944197146,
      "grad_norm": 0.4551568031311035,
      "learning_rate": 9.296083705256673e-06,
      "loss": 0.3607,
      "step": 12260
    },
    {
      "epoch": 0.562663484924969,
      "grad_norm": 0.4693467617034912,
      "learning_rate": 9.29595825907494e-06,
      "loss": 0.4515,
      "step": 12261
    },
    {
      "epoch": 0.5627093754302235,
      "grad_norm": 0.44797995686531067,
      "learning_rate": 9.295832802562782e-06,
      "loss": 0.3867,
      "step": 12262
    },
    {
      "epoch": 0.562755265935478,
      "grad_norm": 0.4529840052127838,
      "learning_rate": 9.295707335720501e-06,
      "loss": 0.421,
      "step": 12263
    },
    {
      "epoch": 0.5628011564407324,
      "grad_norm": 0.45345208048820496,
      "learning_rate": 9.295581858548402e-06,
      "loss": 0.3959,
      "step": 12264
    },
    {
      "epoch": 0.5628470469459869,
      "grad_norm": 0.43922895193099976,
      "learning_rate": 9.295456371046783e-06,
      "loss": 0.3661,
      "step": 12265
    },
    {
      "epoch": 0.5628929374512414,
      "grad_norm": 0.44701889157295227,
      "learning_rate": 9.295330873215946e-06,
      "loss": 0.3533,
      "step": 12266
    },
    {
      "epoch": 0.5629388279564957,
      "grad_norm": 0.45694538950920105,
      "learning_rate": 9.295205365056195e-06,
      "loss": 0.4229,
      "step": 12267
    },
    {
      "epoch": 0.5629847184617502,
      "grad_norm": 0.47272512316703796,
      "learning_rate": 9.295079846567832e-06,
      "loss": 0.4317,
      "step": 12268
    },
    {
      "epoch": 0.5630306089670047,
      "grad_norm": 0.47979989647865295,
      "learning_rate": 9.294954317751155e-06,
      "loss": 0.3983,
      "step": 12269
    },
    {
      "epoch": 0.5630764994722592,
      "grad_norm": 0.48282569646835327,
      "learning_rate": 9.29482877860647e-06,
      "loss": 0.4311,
      "step": 12270
    },
    {
      "epoch": 0.5631223899775136,
      "grad_norm": 0.5249196887016296,
      "learning_rate": 9.294703229134077e-06,
      "loss": 0.4931,
      "step": 12271
    },
    {
      "epoch": 0.5631682804827681,
      "grad_norm": 0.4999580681324005,
      "learning_rate": 9.294577669334277e-06,
      "loss": 0.4858,
      "step": 12272
    },
    {
      "epoch": 0.5632141709880226,
      "grad_norm": 0.4488961696624756,
      "learning_rate": 9.294452099207375e-06,
      "loss": 0.371,
      "step": 12273
    },
    {
      "epoch": 0.563260061493277,
      "grad_norm": 0.4934143126010895,
      "learning_rate": 9.29432651875367e-06,
      "loss": 0.4765,
      "step": 12274
    },
    {
      "epoch": 0.5633059519985315,
      "grad_norm": 0.44268858432769775,
      "learning_rate": 9.294200927973464e-06,
      "loss": 0.3678,
      "step": 12275
    },
    {
      "epoch": 0.563351842503786,
      "grad_norm": 0.4099865257740021,
      "learning_rate": 9.294075326867063e-06,
      "loss": 0.3204,
      "step": 12276
    },
    {
      "epoch": 0.5633977330090404,
      "grad_norm": 0.6284464001655579,
      "learning_rate": 9.293949715434764e-06,
      "loss": 0.4633,
      "step": 12277
    },
    {
      "epoch": 0.5634436235142949,
      "grad_norm": 0.47400614619255066,
      "learning_rate": 9.293824093676872e-06,
      "loss": 0.432,
      "step": 12278
    },
    {
      "epoch": 0.5634895140195494,
      "grad_norm": 0.4270019233226776,
      "learning_rate": 9.293698461593688e-06,
      "loss": 0.3407,
      "step": 12279
    },
    {
      "epoch": 0.5635354045248038,
      "grad_norm": 0.44250649213790894,
      "learning_rate": 9.293572819185513e-06,
      "loss": 0.3854,
      "step": 12280
    },
    {
      "epoch": 0.5635812950300583,
      "grad_norm": 0.42304226756095886,
      "learning_rate": 9.293447166452652e-06,
      "loss": 0.3455,
      "step": 12281
    },
    {
      "epoch": 0.5636271855353128,
      "grad_norm": 0.43560224771499634,
      "learning_rate": 9.293321503395405e-06,
      "loss": 0.338,
      "step": 12282
    },
    {
      "epoch": 0.5636730760405672,
      "grad_norm": 0.45362693071365356,
      "learning_rate": 9.293195830014076e-06,
      "loss": 0.3709,
      "step": 12283
    },
    {
      "epoch": 0.5637189665458217,
      "grad_norm": 0.41024479269981384,
      "learning_rate": 9.293070146308965e-06,
      "loss": 0.3211,
      "step": 12284
    },
    {
      "epoch": 0.5637648570510762,
      "grad_norm": 0.4536677300930023,
      "learning_rate": 9.292944452280375e-06,
      "loss": 0.4002,
      "step": 12285
    },
    {
      "epoch": 0.5638107475563306,
      "grad_norm": 0.44525158405303955,
      "learning_rate": 9.29281874792861e-06,
      "loss": 0.3607,
      "step": 12286
    },
    {
      "epoch": 0.563856638061585,
      "grad_norm": 0.6359711289405823,
      "learning_rate": 9.292693033253967e-06,
      "loss": 0.4739,
      "step": 12287
    },
    {
      "epoch": 0.5639025285668395,
      "grad_norm": 0.4852937161922455,
      "learning_rate": 9.292567308256756e-06,
      "loss": 0.4493,
      "step": 12288
    },
    {
      "epoch": 0.563948419072094,
      "grad_norm": 0.4483056962490082,
      "learning_rate": 9.292441572937275e-06,
      "loss": 0.3922,
      "step": 12289
    },
    {
      "epoch": 0.5639943095773484,
      "grad_norm": 0.47928833961486816,
      "learning_rate": 9.292315827295824e-06,
      "loss": 0.496,
      "step": 12290
    },
    {
      "epoch": 0.5640402000826029,
      "grad_norm": 0.442153662443161,
      "learning_rate": 9.29219007133271e-06,
      "loss": 0.3498,
      "step": 12291
    },
    {
      "epoch": 0.5640860905878574,
      "grad_norm": 0.47824329137802124,
      "learning_rate": 9.292064305048233e-06,
      "loss": 0.469,
      "step": 12292
    },
    {
      "epoch": 0.5641319810931118,
      "grad_norm": 0.47097247838974,
      "learning_rate": 9.291938528442697e-06,
      "loss": 0.4551,
      "step": 12293
    },
    {
      "epoch": 0.5641778715983663,
      "grad_norm": 0.49916717410087585,
      "learning_rate": 9.291812741516402e-06,
      "loss": 0.4219,
      "step": 12294
    },
    {
      "epoch": 0.5642237621036208,
      "grad_norm": 0.44558364152908325,
      "learning_rate": 9.291686944269652e-06,
      "loss": 0.355,
      "step": 12295
    },
    {
      "epoch": 0.5642696526088752,
      "grad_norm": 0.4427212178707123,
      "learning_rate": 9.291561136702752e-06,
      "loss": 0.3565,
      "step": 12296
    },
    {
      "epoch": 0.5643155431141297,
      "grad_norm": 0.45036619901657104,
      "learning_rate": 9.291435318816e-06,
      "loss": 0.3584,
      "step": 12297
    },
    {
      "epoch": 0.5643614336193842,
      "grad_norm": 0.4831230342388153,
      "learning_rate": 9.291309490609701e-06,
      "loss": 0.4038,
      "step": 12298
    },
    {
      "epoch": 0.5644073241246386,
      "grad_norm": 0.42990365624427795,
      "learning_rate": 9.291183652084154e-06,
      "loss": 0.3265,
      "step": 12299
    },
    {
      "epoch": 0.5644532146298931,
      "grad_norm": 0.4638040065765381,
      "learning_rate": 9.291057803239669e-06,
      "loss": 0.3719,
      "step": 12300
    },
    {
      "epoch": 0.5644991051351476,
      "grad_norm": 0.46280795335769653,
      "learning_rate": 9.290931944076542e-06,
      "loss": 0.387,
      "step": 12301
    },
    {
      "epoch": 0.564544995640402,
      "grad_norm": 0.464979350566864,
      "learning_rate": 9.290806074595079e-06,
      "loss": 0.4265,
      "step": 12302
    },
    {
      "epoch": 0.5645908861456564,
      "grad_norm": 0.4330803453922272,
      "learning_rate": 9.29068019479558e-06,
      "loss": 0.337,
      "step": 12303
    },
    {
      "epoch": 0.5646367766509109,
      "grad_norm": 0.45030567049980164,
      "learning_rate": 9.29055430467835e-06,
      "loss": 0.3736,
      "step": 12304
    },
    {
      "epoch": 0.5646826671561654,
      "grad_norm": 0.47264572978019714,
      "learning_rate": 9.290428404243692e-06,
      "loss": 0.4167,
      "step": 12305
    },
    {
      "epoch": 0.5647285576614198,
      "grad_norm": 0.5115957260131836,
      "learning_rate": 9.290302493491907e-06,
      "loss": 0.4501,
      "step": 12306
    },
    {
      "epoch": 0.5647744481666743,
      "grad_norm": 0.4238923490047455,
      "learning_rate": 9.290176572423298e-06,
      "loss": 0.3377,
      "step": 12307
    },
    {
      "epoch": 0.5648203386719288,
      "grad_norm": 0.49365487694740295,
      "learning_rate": 9.29005064103817e-06,
      "loss": 0.4826,
      "step": 12308
    },
    {
      "epoch": 0.5648662291771832,
      "grad_norm": 0.4579617977142334,
      "learning_rate": 9.289924699336823e-06,
      "loss": 0.3847,
      "step": 12309
    },
    {
      "epoch": 0.5649121196824377,
      "grad_norm": 0.47378009557724,
      "learning_rate": 9.289798747319561e-06,
      "loss": 0.4545,
      "step": 12310
    },
    {
      "epoch": 0.5649580101876922,
      "grad_norm": 0.4739331901073456,
      "learning_rate": 9.289672784986688e-06,
      "loss": 0.4167,
      "step": 12311
    },
    {
      "epoch": 0.5650039006929466,
      "grad_norm": 0.4547980725765228,
      "learning_rate": 9.289546812338504e-06,
      "loss": 0.3684,
      "step": 12312
    },
    {
      "epoch": 0.5650497911982011,
      "grad_norm": 0.45787227153778076,
      "learning_rate": 9.289420829375315e-06,
      "loss": 0.3574,
      "step": 12313
    },
    {
      "epoch": 0.5650956817034556,
      "grad_norm": 0.4667074382305145,
      "learning_rate": 9.289294836097423e-06,
      "loss": 0.4499,
      "step": 12314
    },
    {
      "epoch": 0.56514157220871,
      "grad_norm": 0.40683817863464355,
      "learning_rate": 9.28916883250513e-06,
      "loss": 0.3049,
      "step": 12315
    },
    {
      "epoch": 0.5651874627139645,
      "grad_norm": 0.43430161476135254,
      "learning_rate": 9.28904281859874e-06,
      "loss": 0.3436,
      "step": 12316
    },
    {
      "epoch": 0.565233353219219,
      "grad_norm": 0.7006710767745972,
      "learning_rate": 9.288916794378556e-06,
      "loss": 0.3966,
      "step": 12317
    },
    {
      "epoch": 0.5652792437244734,
      "grad_norm": 0.47191306948661804,
      "learning_rate": 9.288790759844879e-06,
      "loss": 0.3642,
      "step": 12318
    },
    {
      "epoch": 0.5653251342297279,
      "grad_norm": 0.44052189588546753,
      "learning_rate": 9.288664714998015e-06,
      "loss": 0.3523,
      "step": 12319
    },
    {
      "epoch": 0.5653710247349824,
      "grad_norm": 0.457026869058609,
      "learning_rate": 9.288538659838266e-06,
      "loss": 0.3781,
      "step": 12320
    },
    {
      "epoch": 0.5654169152402369,
      "grad_norm": 0.5153876543045044,
      "learning_rate": 9.288412594365935e-06,
      "loss": 0.5315,
      "step": 12321
    },
    {
      "epoch": 0.5654628057454912,
      "grad_norm": 0.4891815185546875,
      "learning_rate": 9.288286518581325e-06,
      "loss": 0.4521,
      "step": 12322
    },
    {
      "epoch": 0.5655086962507457,
      "grad_norm": 0.4297462999820709,
      "learning_rate": 9.288160432484739e-06,
      "loss": 0.3453,
      "step": 12323
    },
    {
      "epoch": 0.5655545867560002,
      "grad_norm": 0.46707066893577576,
      "learning_rate": 9.288034336076481e-06,
      "loss": 0.3803,
      "step": 12324
    },
    {
      "epoch": 0.5656004772612546,
      "grad_norm": 0.57640141248703,
      "learning_rate": 9.287908229356854e-06,
      "loss": 0.3316,
      "step": 12325
    },
    {
      "epoch": 0.5656463677665091,
      "grad_norm": 0.5167283415794373,
      "learning_rate": 9.28778211232616e-06,
      "loss": 0.4538,
      "step": 12326
    },
    {
      "epoch": 0.5656922582717636,
      "grad_norm": 0.4570002555847168,
      "learning_rate": 9.287655984984703e-06,
      "loss": 0.427,
      "step": 12327
    },
    {
      "epoch": 0.565738148777018,
      "grad_norm": 0.4519749879837036,
      "learning_rate": 9.287529847332788e-06,
      "loss": 0.3555,
      "step": 12328
    },
    {
      "epoch": 0.5657840392822725,
      "grad_norm": 0.43872126936912537,
      "learning_rate": 9.287403699370717e-06,
      "loss": 0.3468,
      "step": 12329
    },
    {
      "epoch": 0.565829929787527,
      "grad_norm": 0.4331133961677551,
      "learning_rate": 9.287277541098793e-06,
      "loss": 0.3627,
      "step": 12330
    },
    {
      "epoch": 0.5658758202927814,
      "grad_norm": 0.44898131489753723,
      "learning_rate": 9.287151372517319e-06,
      "loss": 0.3659,
      "step": 12331
    },
    {
      "epoch": 0.5659217107980359,
      "grad_norm": 0.4406052827835083,
      "learning_rate": 9.2870251936266e-06,
      "loss": 0.3557,
      "step": 12332
    },
    {
      "epoch": 0.5659676013032904,
      "grad_norm": 0.4835537374019623,
      "learning_rate": 9.286899004426936e-06,
      "loss": 0.4299,
      "step": 12333
    },
    {
      "epoch": 0.5660134918085448,
      "grad_norm": 0.43501392006874084,
      "learning_rate": 9.286772804918636e-06,
      "loss": 0.321,
      "step": 12334
    },
    {
      "epoch": 0.5660593823137993,
      "grad_norm": 0.4455309510231018,
      "learning_rate": 9.286646595101999e-06,
      "loss": 0.3968,
      "step": 12335
    },
    {
      "epoch": 0.5661052728190538,
      "grad_norm": 0.4751589596271515,
      "learning_rate": 9.28652037497733e-06,
      "loss": 0.3687,
      "step": 12336
    },
    {
      "epoch": 0.5661511633243081,
      "grad_norm": 0.4375549256801605,
      "learning_rate": 9.28639414454493e-06,
      "loss": 0.3732,
      "step": 12337
    },
    {
      "epoch": 0.5661970538295626,
      "grad_norm": 0.48499125242233276,
      "learning_rate": 9.286267903805108e-06,
      "loss": 0.4234,
      "step": 12338
    },
    {
      "epoch": 0.5662429443348171,
      "grad_norm": 0.4654761254787445,
      "learning_rate": 9.286141652758165e-06,
      "loss": 0.3718,
      "step": 12339
    },
    {
      "epoch": 0.5662888348400716,
      "grad_norm": 0.46959471702575684,
      "learning_rate": 9.2860153914044e-06,
      "loss": 0.4556,
      "step": 12340
    },
    {
      "epoch": 0.566334725345326,
      "grad_norm": 0.4355045258998871,
      "learning_rate": 9.285889119744125e-06,
      "loss": 0.3064,
      "step": 12341
    },
    {
      "epoch": 0.5663806158505805,
      "grad_norm": 0.4347067177295685,
      "learning_rate": 9.285762837777638e-06,
      "loss": 0.366,
      "step": 12342
    },
    {
      "epoch": 0.566426506355835,
      "grad_norm": 0.44626331329345703,
      "learning_rate": 9.285636545505244e-06,
      "loss": 0.3823,
      "step": 12343
    },
    {
      "epoch": 0.5664723968610894,
      "grad_norm": 0.5105581879615784,
      "learning_rate": 9.285510242927245e-06,
      "loss": 0.4178,
      "step": 12344
    },
    {
      "epoch": 0.5665182873663439,
      "grad_norm": 0.4818257987499237,
      "learning_rate": 9.28538393004395e-06,
      "loss": 0.4351,
      "step": 12345
    },
    {
      "epoch": 0.5665641778715984,
      "grad_norm": 0.4729830026626587,
      "learning_rate": 9.285257606855655e-06,
      "loss": 0.4205,
      "step": 12346
    },
    {
      "epoch": 0.5666100683768528,
      "grad_norm": 0.452766478061676,
      "learning_rate": 9.28513127336267e-06,
      "loss": 0.4502,
      "step": 12347
    },
    {
      "epoch": 0.5666559588821073,
      "grad_norm": 0.45837682485580444,
      "learning_rate": 9.285004929565296e-06,
      "loss": 0.3804,
      "step": 12348
    },
    {
      "epoch": 0.5667018493873618,
      "grad_norm": 0.4430794417858124,
      "learning_rate": 9.28487857546384e-06,
      "loss": 0.389,
      "step": 12349
    },
    {
      "epoch": 0.5667477398926162,
      "grad_norm": 0.4108556807041168,
      "learning_rate": 9.284752211058601e-06,
      "loss": 0.3445,
      "step": 12350
    },
    {
      "epoch": 0.5667936303978707,
      "grad_norm": 0.46997344493865967,
      "learning_rate": 9.284625836349886e-06,
      "loss": 0.3615,
      "step": 12351
    },
    {
      "epoch": 0.5668395209031252,
      "grad_norm": 0.4661126434803009,
      "learning_rate": 9.284499451337997e-06,
      "loss": 0.4088,
      "step": 12352
    },
    {
      "epoch": 0.5668854114083796,
      "grad_norm": 0.47118115425109863,
      "learning_rate": 9.284373056023238e-06,
      "loss": 0.4273,
      "step": 12353
    },
    {
      "epoch": 0.5669313019136341,
      "grad_norm": 0.42398905754089355,
      "learning_rate": 9.284246650405918e-06,
      "loss": 0.3474,
      "step": 12354
    },
    {
      "epoch": 0.5669771924188886,
      "grad_norm": 0.4324185252189636,
      "learning_rate": 9.284120234486334e-06,
      "loss": 0.3298,
      "step": 12355
    },
    {
      "epoch": 0.5670230829241429,
      "grad_norm": 0.47996610403060913,
      "learning_rate": 9.283993808264795e-06,
      "loss": 0.4485,
      "step": 12356
    },
    {
      "epoch": 0.5670689734293974,
      "grad_norm": 0.4737033545970917,
      "learning_rate": 9.283867371741602e-06,
      "loss": 0.3987,
      "step": 12357
    },
    {
      "epoch": 0.5671148639346519,
      "grad_norm": 0.4798785150051117,
      "learning_rate": 9.28374092491706e-06,
      "loss": 0.4083,
      "step": 12358
    },
    {
      "epoch": 0.5671607544399064,
      "grad_norm": 0.4558336138725281,
      "learning_rate": 9.283614467791472e-06,
      "loss": 0.3991,
      "step": 12359
    },
    {
      "epoch": 0.5672066449451608,
      "grad_norm": 0.44801995158195496,
      "learning_rate": 9.283488000365145e-06,
      "loss": 0.4273,
      "step": 12360
    },
    {
      "epoch": 0.5672525354504153,
      "grad_norm": 0.46096011996269226,
      "learning_rate": 9.283361522638378e-06,
      "loss": 0.3992,
      "step": 12361
    },
    {
      "epoch": 0.5672984259556698,
      "grad_norm": 0.42707687616348267,
      "learning_rate": 9.283235034611483e-06,
      "loss": 0.3199,
      "step": 12362
    },
    {
      "epoch": 0.5673443164609242,
      "grad_norm": 0.4474504292011261,
      "learning_rate": 9.283108536284754e-06,
      "loss": 0.4063,
      "step": 12363
    },
    {
      "epoch": 0.5673902069661787,
      "grad_norm": 0.43131572008132935,
      "learning_rate": 9.282982027658505e-06,
      "loss": 0.355,
      "step": 12364
    },
    {
      "epoch": 0.5674360974714332,
      "grad_norm": 0.4835481643676758,
      "learning_rate": 9.282855508733036e-06,
      "loss": 0.4254,
      "step": 12365
    },
    {
      "epoch": 0.5674819879766876,
      "grad_norm": 0.4040113389492035,
      "learning_rate": 9.28272897950865e-06,
      "loss": 0.3239,
      "step": 12366
    },
    {
      "epoch": 0.5675278784819421,
      "grad_norm": 0.45294812321662903,
      "learning_rate": 9.282602439985652e-06,
      "loss": 0.4037,
      "step": 12367
    },
    {
      "epoch": 0.5675737689871966,
      "grad_norm": 0.4249016344547272,
      "learning_rate": 9.282475890164347e-06,
      "loss": 0.3281,
      "step": 12368
    },
    {
      "epoch": 0.567619659492451,
      "grad_norm": 0.5042969584465027,
      "learning_rate": 9.28234933004504e-06,
      "loss": 0.4873,
      "step": 12369
    },
    {
      "epoch": 0.5676655499977055,
      "grad_norm": 0.42759576439857483,
      "learning_rate": 9.282222759628035e-06,
      "loss": 0.3611,
      "step": 12370
    },
    {
      "epoch": 0.56771144050296,
      "grad_norm": 0.450784832239151,
      "learning_rate": 9.282096178913633e-06,
      "loss": 0.3498,
      "step": 12371
    },
    {
      "epoch": 0.5677573310082143,
      "grad_norm": 0.5053176879882812,
      "learning_rate": 9.281969587902144e-06,
      "loss": 0.4637,
      "step": 12372
    },
    {
      "epoch": 0.5678032215134688,
      "grad_norm": 0.4744170010089874,
      "learning_rate": 9.281842986593868e-06,
      "loss": 0.4533,
      "step": 12373
    },
    {
      "epoch": 0.5678491120187233,
      "grad_norm": 0.463698148727417,
      "learning_rate": 9.28171637498911e-06,
      "loss": 0.3886,
      "step": 12374
    },
    {
      "epoch": 0.5678950025239778,
      "grad_norm": 0.4810331165790558,
      "learning_rate": 9.281589753088178e-06,
      "loss": 0.4165,
      "step": 12375
    },
    {
      "epoch": 0.5679408930292322,
      "grad_norm": 0.44187161326408386,
      "learning_rate": 9.281463120891373e-06,
      "loss": 0.3754,
      "step": 12376
    },
    {
      "epoch": 0.5679867835344867,
      "grad_norm": 0.4601793885231018,
      "learning_rate": 9.281336478398998e-06,
      "loss": 0.4035,
      "step": 12377
    },
    {
      "epoch": 0.5680326740397412,
      "grad_norm": 0.47379082441329956,
      "learning_rate": 9.281209825611364e-06,
      "loss": 0.4469,
      "step": 12378
    },
    {
      "epoch": 0.5680785645449956,
      "grad_norm": 0.46356967091560364,
      "learning_rate": 9.28108316252877e-06,
      "loss": 0.457,
      "step": 12379
    },
    {
      "epoch": 0.5681244550502501,
      "grad_norm": 0.4848254323005676,
      "learning_rate": 9.280956489151523e-06,
      "loss": 0.4114,
      "step": 12380
    },
    {
      "epoch": 0.5681703455555046,
      "grad_norm": 0.44569462537765503,
      "learning_rate": 9.280829805479925e-06,
      "loss": 0.3692,
      "step": 12381
    },
    {
      "epoch": 0.568216236060759,
      "grad_norm": 0.520349383354187,
      "learning_rate": 9.280703111514282e-06,
      "loss": 0.4715,
      "step": 12382
    },
    {
      "epoch": 0.5682621265660135,
      "grad_norm": 0.4963352382183075,
      "learning_rate": 9.2805764072549e-06,
      "loss": 0.4059,
      "step": 12383
    },
    {
      "epoch": 0.568308017071268,
      "grad_norm": 0.46492689847946167,
      "learning_rate": 9.280449692702083e-06,
      "loss": 0.4105,
      "step": 12384
    },
    {
      "epoch": 0.5683539075765224,
      "grad_norm": 0.4455277919769287,
      "learning_rate": 9.280322967856136e-06,
      "loss": 0.3325,
      "step": 12385
    },
    {
      "epoch": 0.5683997980817769,
      "grad_norm": 0.7298843860626221,
      "learning_rate": 9.280196232717363e-06,
      "loss": 0.4832,
      "step": 12386
    },
    {
      "epoch": 0.5684456885870314,
      "grad_norm": 0.451158732175827,
      "learning_rate": 9.280069487286067e-06,
      "loss": 0.3472,
      "step": 12387
    },
    {
      "epoch": 0.5684915790922858,
      "grad_norm": 0.4695546627044678,
      "learning_rate": 9.279942731562557e-06,
      "loss": 0.4215,
      "step": 12388
    },
    {
      "epoch": 0.5685374695975403,
      "grad_norm": 0.48479726910591125,
      "learning_rate": 9.279815965547134e-06,
      "loss": 0.4319,
      "step": 12389
    },
    {
      "epoch": 0.5685833601027948,
      "grad_norm": 0.510964035987854,
      "learning_rate": 9.279689189240105e-06,
      "loss": 0.4641,
      "step": 12390
    },
    {
      "epoch": 0.5686292506080491,
      "grad_norm": 0.4566155970096588,
      "learning_rate": 9.279562402641775e-06,
      "loss": 0.3817,
      "step": 12391
    },
    {
      "epoch": 0.5686751411133036,
      "grad_norm": 0.44954410195350647,
      "learning_rate": 9.279435605752446e-06,
      "loss": 0.3824,
      "step": 12392
    },
    {
      "epoch": 0.5687210316185581,
      "grad_norm": 0.45285600423812866,
      "learning_rate": 9.279308798572427e-06,
      "loss": 0.3757,
      "step": 12393
    },
    {
      "epoch": 0.5687669221238126,
      "grad_norm": 0.44981521368026733,
      "learning_rate": 9.27918198110202e-06,
      "loss": 0.3592,
      "step": 12394
    },
    {
      "epoch": 0.568812812629067,
      "grad_norm": 0.48487240076065063,
      "learning_rate": 9.279055153341531e-06,
      "loss": 0.5208,
      "step": 12395
    },
    {
      "epoch": 0.5688587031343215,
      "grad_norm": 0.45981818437576294,
      "learning_rate": 9.278928315291264e-06,
      "loss": 0.4691,
      "step": 12396
    },
    {
      "epoch": 0.568904593639576,
      "grad_norm": 0.5117197632789612,
      "learning_rate": 9.278801466951526e-06,
      "loss": 0.528,
      "step": 12397
    },
    {
      "epoch": 0.5689504841448304,
      "grad_norm": 0.4701516330242157,
      "learning_rate": 9.278674608322622e-06,
      "loss": 0.4579,
      "step": 12398
    },
    {
      "epoch": 0.5689963746500849,
      "grad_norm": 0.4414425194263458,
      "learning_rate": 9.278547739404854e-06,
      "loss": 0.3324,
      "step": 12399
    },
    {
      "epoch": 0.5690422651553394,
      "grad_norm": 0.4252696633338928,
      "learning_rate": 9.27842086019853e-06,
      "loss": 0.2948,
      "step": 12400
    },
    {
      "epoch": 0.5690881556605938,
      "grad_norm": 0.5155567526817322,
      "learning_rate": 9.278293970703954e-06,
      "loss": 0.5607,
      "step": 12401
    },
    {
      "epoch": 0.5691340461658483,
      "grad_norm": 0.47085556387901306,
      "learning_rate": 9.278167070921432e-06,
      "loss": 0.4165,
      "step": 12402
    },
    {
      "epoch": 0.5691799366711028,
      "grad_norm": 0.46698322892189026,
      "learning_rate": 9.278040160851266e-06,
      "loss": 0.4486,
      "step": 12403
    },
    {
      "epoch": 0.5692258271763572,
      "grad_norm": 0.4098964333534241,
      "learning_rate": 9.277913240493765e-06,
      "loss": 0.3355,
      "step": 12404
    },
    {
      "epoch": 0.5692717176816117,
      "grad_norm": 0.4780869781970978,
      "learning_rate": 9.277786309849232e-06,
      "loss": 0.4979,
      "step": 12405
    },
    {
      "epoch": 0.5693176081868662,
      "grad_norm": 0.4583963453769684,
      "learning_rate": 9.277659368917975e-06,
      "loss": 0.349,
      "step": 12406
    },
    {
      "epoch": 0.5693634986921206,
      "grad_norm": 0.4263805150985718,
      "learning_rate": 9.277532417700296e-06,
      "loss": 0.3427,
      "step": 12407
    },
    {
      "epoch": 0.569409389197375,
      "grad_norm": 0.6302890777587891,
      "learning_rate": 9.277405456196503e-06,
      "loss": 0.4112,
      "step": 12408
    },
    {
      "epoch": 0.5694552797026295,
      "grad_norm": 0.5228577256202698,
      "learning_rate": 9.277278484406897e-06,
      "loss": 0.5185,
      "step": 12409
    },
    {
      "epoch": 0.5695011702078839,
      "grad_norm": 0.4519951045513153,
      "learning_rate": 9.277151502331789e-06,
      "loss": 0.3903,
      "step": 12410
    },
    {
      "epoch": 0.5695470607131384,
      "grad_norm": 0.5260011553764343,
      "learning_rate": 9.27702450997148e-06,
      "loss": 0.5269,
      "step": 12411
    },
    {
      "epoch": 0.5695929512183929,
      "grad_norm": 0.43695566058158875,
      "learning_rate": 9.276897507326277e-06,
      "loss": 0.3653,
      "step": 12412
    },
    {
      "epoch": 0.5696388417236474,
      "grad_norm": 0.4952985644340515,
      "learning_rate": 9.276770494396486e-06,
      "loss": 0.4958,
      "step": 12413
    },
    {
      "epoch": 0.5696847322289018,
      "grad_norm": 0.4860233962535858,
      "learning_rate": 9.276643471182412e-06,
      "loss": 0.4614,
      "step": 12414
    },
    {
      "epoch": 0.5697306227341563,
      "grad_norm": 0.4407031536102295,
      "learning_rate": 9.27651643768436e-06,
      "loss": 0.4207,
      "step": 12415
    },
    {
      "epoch": 0.5697765132394108,
      "grad_norm": 0.5016089081764221,
      "learning_rate": 9.276389393902635e-06,
      "loss": 0.509,
      "step": 12416
    },
    {
      "epoch": 0.5698224037446652,
      "grad_norm": 0.4353282153606415,
      "learning_rate": 9.276262339837542e-06,
      "loss": 0.3723,
      "step": 12417
    },
    {
      "epoch": 0.5698682942499197,
      "grad_norm": 0.40254485607147217,
      "learning_rate": 9.27613527548939e-06,
      "loss": 0.2888,
      "step": 12418
    },
    {
      "epoch": 0.5699141847551742,
      "grad_norm": 0.42926833033561707,
      "learning_rate": 9.276008200858482e-06,
      "loss": 0.3797,
      "step": 12419
    },
    {
      "epoch": 0.5699600752604286,
      "grad_norm": 0.46133339405059814,
      "learning_rate": 9.275881115945123e-06,
      "loss": 0.3847,
      "step": 12420
    },
    {
      "epoch": 0.5700059657656831,
      "grad_norm": 0.42997005581855774,
      "learning_rate": 9.275754020749618e-06,
      "loss": 0.3794,
      "step": 12421
    },
    {
      "epoch": 0.5700518562709376,
      "grad_norm": 0.4597589373588562,
      "learning_rate": 9.275626915272277e-06,
      "loss": 0.4197,
      "step": 12422
    },
    {
      "epoch": 0.570097746776192,
      "grad_norm": 0.48366090655326843,
      "learning_rate": 9.275499799513401e-06,
      "loss": 0.4959,
      "step": 12423
    },
    {
      "epoch": 0.5701436372814465,
      "grad_norm": 0.47189608216285706,
      "learning_rate": 9.275372673473299e-06,
      "loss": 0.4332,
      "step": 12424
    },
    {
      "epoch": 0.570189527786701,
      "grad_norm": 0.4361773729324341,
      "learning_rate": 9.275245537152273e-06,
      "loss": 0.3878,
      "step": 12425
    },
    {
      "epoch": 0.5702354182919553,
      "grad_norm": 0.4524957537651062,
      "learning_rate": 9.275118390550632e-06,
      "loss": 0.4094,
      "step": 12426
    },
    {
      "epoch": 0.5702813087972098,
      "grad_norm": 0.4407818615436554,
      "learning_rate": 9.274991233668679e-06,
      "loss": 0.38,
      "step": 12427
    },
    {
      "epoch": 0.5703271993024643,
      "grad_norm": 0.4623456597328186,
      "learning_rate": 9.274864066506723e-06,
      "loss": 0.462,
      "step": 12428
    },
    {
      "epoch": 0.5703730898077188,
      "grad_norm": 0.42813539505004883,
      "learning_rate": 9.274736889065068e-06,
      "loss": 0.333,
      "step": 12429
    },
    {
      "epoch": 0.5704189803129732,
      "grad_norm": 0.45001348853111267,
      "learning_rate": 9.27460970134402e-06,
      "loss": 0.3637,
      "step": 12430
    },
    {
      "epoch": 0.5704648708182277,
      "grad_norm": 0.46588921546936035,
      "learning_rate": 9.274482503343884e-06,
      "loss": 0.3506,
      "step": 12431
    },
    {
      "epoch": 0.5705107613234822,
      "grad_norm": 0.4617343842983246,
      "learning_rate": 9.274355295064968e-06,
      "loss": 0.4483,
      "step": 12432
    },
    {
      "epoch": 0.5705566518287366,
      "grad_norm": 0.5144270658493042,
      "learning_rate": 9.274228076507575e-06,
      "loss": 0.434,
      "step": 12433
    },
    {
      "epoch": 0.5706025423339911,
      "grad_norm": 0.4424397051334381,
      "learning_rate": 9.274100847672011e-06,
      "loss": 0.3402,
      "step": 12434
    },
    {
      "epoch": 0.5706484328392456,
      "grad_norm": 0.4536787271499634,
      "learning_rate": 9.273973608558586e-06,
      "loss": 0.3404,
      "step": 12435
    },
    {
      "epoch": 0.5706943233445,
      "grad_norm": 0.5042724013328552,
      "learning_rate": 9.273846359167603e-06,
      "loss": 0.4661,
      "step": 12436
    },
    {
      "epoch": 0.5707402138497545,
      "grad_norm": 0.4644010365009308,
      "learning_rate": 9.273719099499367e-06,
      "loss": 0.4422,
      "step": 12437
    },
    {
      "epoch": 0.570786104355009,
      "grad_norm": 0.495913565158844,
      "learning_rate": 9.273591829554188e-06,
      "loss": 0.4226,
      "step": 12438
    },
    {
      "epoch": 0.5708319948602634,
      "grad_norm": 0.4370101988315582,
      "learning_rate": 9.273464549332368e-06,
      "loss": 0.3364,
      "step": 12439
    },
    {
      "epoch": 0.5708778853655179,
      "grad_norm": 0.46859753131866455,
      "learning_rate": 9.273337258834215e-06,
      "loss": 0.5125,
      "step": 12440
    },
    {
      "epoch": 0.5709237758707724,
      "grad_norm": 0.4720595180988312,
      "learning_rate": 9.273209958060033e-06,
      "loss": 0.4716,
      "step": 12441
    },
    {
      "epoch": 0.5709696663760268,
      "grad_norm": 0.473365843296051,
      "learning_rate": 9.27308264701013e-06,
      "loss": 0.4183,
      "step": 12442
    },
    {
      "epoch": 0.5710155568812812,
      "grad_norm": 0.4700353443622589,
      "learning_rate": 9.272955325684813e-06,
      "loss": 0.4227,
      "step": 12443
    },
    {
      "epoch": 0.5710614473865357,
      "grad_norm": 0.44280537962913513,
      "learning_rate": 9.272827994084387e-06,
      "loss": 0.3604,
      "step": 12444
    },
    {
      "epoch": 0.5711073378917901,
      "grad_norm": 0.4543342590332031,
      "learning_rate": 9.272700652209157e-06,
      "loss": 0.4131,
      "step": 12445
    },
    {
      "epoch": 0.5711532283970446,
      "grad_norm": 0.46248307824134827,
      "learning_rate": 9.272573300059432e-06,
      "loss": 0.44,
      "step": 12446
    },
    {
      "epoch": 0.5711991189022991,
      "grad_norm": 0.494863361120224,
      "learning_rate": 9.272445937635516e-06,
      "loss": 0.4419,
      "step": 12447
    },
    {
      "epoch": 0.5712450094075536,
      "grad_norm": 0.4952270984649658,
      "learning_rate": 9.272318564937716e-06,
      "loss": 0.5217,
      "step": 12448
    },
    {
      "epoch": 0.571290899912808,
      "grad_norm": 0.474663108587265,
      "learning_rate": 9.272191181966338e-06,
      "loss": 0.4384,
      "step": 12449
    },
    {
      "epoch": 0.5713367904180625,
      "grad_norm": 0.4027078449726105,
      "learning_rate": 9.272063788721687e-06,
      "loss": 0.3125,
      "step": 12450
    },
    {
      "epoch": 0.571382680923317,
      "grad_norm": 0.4733777940273285,
      "learning_rate": 9.271936385204074e-06,
      "loss": 0.4677,
      "step": 12451
    },
    {
      "epoch": 0.5714285714285714,
      "grad_norm": 0.4812890589237213,
      "learning_rate": 9.2718089714138e-06,
      "loss": 0.4389,
      "step": 12452
    },
    {
      "epoch": 0.5714744619338259,
      "grad_norm": 0.4387906491756439,
      "learning_rate": 9.271681547351175e-06,
      "loss": 0.3193,
      "step": 12453
    },
    {
      "epoch": 0.5715203524390804,
      "grad_norm": 0.4270033538341522,
      "learning_rate": 9.271554113016503e-06,
      "loss": 0.3529,
      "step": 12454
    },
    {
      "epoch": 0.5715662429443348,
      "grad_norm": 0.45658600330352783,
      "learning_rate": 9.271426668410092e-06,
      "loss": 0.3739,
      "step": 12455
    },
    {
      "epoch": 0.5716121334495893,
      "grad_norm": 0.45512616634368896,
      "learning_rate": 9.271299213532248e-06,
      "loss": 0.4447,
      "step": 12456
    },
    {
      "epoch": 0.5716580239548438,
      "grad_norm": 0.44947123527526855,
      "learning_rate": 9.271171748383278e-06,
      "loss": 0.4156,
      "step": 12457
    },
    {
      "epoch": 0.5717039144600982,
      "grad_norm": 0.4518815875053406,
      "learning_rate": 9.271044272963488e-06,
      "loss": 0.3759,
      "step": 12458
    },
    {
      "epoch": 0.5717498049653527,
      "grad_norm": 0.47350963950157166,
      "learning_rate": 9.270916787273184e-06,
      "loss": 0.4153,
      "step": 12459
    },
    {
      "epoch": 0.5717956954706072,
      "grad_norm": 0.455748975276947,
      "learning_rate": 9.270789291312672e-06,
      "loss": 0.3801,
      "step": 12460
    },
    {
      "epoch": 0.5718415859758615,
      "grad_norm": 0.4458254873752594,
      "learning_rate": 9.270661785082261e-06,
      "loss": 0.348,
      "step": 12461
    },
    {
      "epoch": 0.571887476481116,
      "grad_norm": 0.48263588547706604,
      "learning_rate": 9.270534268582257e-06,
      "loss": 0.4613,
      "step": 12462
    },
    {
      "epoch": 0.5719333669863705,
      "grad_norm": 0.43069809675216675,
      "learning_rate": 9.270406741812964e-06,
      "loss": 0.3351,
      "step": 12463
    },
    {
      "epoch": 0.571979257491625,
      "grad_norm": 0.4114397168159485,
      "learning_rate": 9.27027920477469e-06,
      "loss": 0.3238,
      "step": 12464
    },
    {
      "epoch": 0.5720251479968794,
      "grad_norm": 0.4413287043571472,
      "learning_rate": 9.270151657467745e-06,
      "loss": 0.3959,
      "step": 12465
    },
    {
      "epoch": 0.5720710385021339,
      "grad_norm": 0.45655664801597595,
      "learning_rate": 9.270024099892432e-06,
      "loss": 0.441,
      "step": 12466
    },
    {
      "epoch": 0.5721169290073884,
      "grad_norm": 0.46156468987464905,
      "learning_rate": 9.269896532049058e-06,
      "loss": 0.4258,
      "step": 12467
    },
    {
      "epoch": 0.5721628195126428,
      "grad_norm": 0.4382556676864624,
      "learning_rate": 9.26976895393793e-06,
      "loss": 0.4042,
      "step": 12468
    },
    {
      "epoch": 0.5722087100178973,
      "grad_norm": 0.456933856010437,
      "learning_rate": 9.269641365559356e-06,
      "loss": 0.3622,
      "step": 12469
    },
    {
      "epoch": 0.5722546005231518,
      "grad_norm": 0.5138749480247498,
      "learning_rate": 9.269513766913643e-06,
      "loss": 0.4866,
      "step": 12470
    },
    {
      "epoch": 0.5723004910284062,
      "grad_norm": 0.4266880750656128,
      "learning_rate": 9.269386158001094e-06,
      "loss": 0.3173,
      "step": 12471
    },
    {
      "epoch": 0.5723463815336607,
      "grad_norm": 0.41814228892326355,
      "learning_rate": 9.269258538822021e-06,
      "loss": 0.309,
      "step": 12472
    },
    {
      "epoch": 0.5723922720389152,
      "grad_norm": 0.4905936121940613,
      "learning_rate": 9.269130909376728e-06,
      "loss": 0.444,
      "step": 12473
    },
    {
      "epoch": 0.5724381625441696,
      "grad_norm": 0.4651744067668915,
      "learning_rate": 9.269003269665524e-06,
      "loss": 0.4218,
      "step": 12474
    },
    {
      "epoch": 0.5724840530494241,
      "grad_norm": 0.47762641310691833,
      "learning_rate": 9.268875619688713e-06,
      "loss": 0.416,
      "step": 12475
    },
    {
      "epoch": 0.5725299435546786,
      "grad_norm": 0.46365851163864136,
      "learning_rate": 9.268747959446605e-06,
      "loss": 0.3537,
      "step": 12476
    },
    {
      "epoch": 0.572575834059933,
      "grad_norm": 0.43689680099487305,
      "learning_rate": 9.268620288939502e-06,
      "loss": 0.3265,
      "step": 12477
    },
    {
      "epoch": 0.5726217245651875,
      "grad_norm": 0.492890328168869,
      "learning_rate": 9.268492608167717e-06,
      "loss": 0.4517,
      "step": 12478
    },
    {
      "epoch": 0.572667615070442,
      "grad_norm": 0.4098397195339203,
      "learning_rate": 9.268364917131555e-06,
      "loss": 0.3216,
      "step": 12479
    },
    {
      "epoch": 0.5727135055756963,
      "grad_norm": 0.43102848529815674,
      "learning_rate": 9.268237215831322e-06,
      "loss": 0.3347,
      "step": 12480
    },
    {
      "epoch": 0.5727593960809508,
      "grad_norm": 0.5144219994544983,
      "learning_rate": 9.268109504267324e-06,
      "loss": 0.438,
      "step": 12481
    },
    {
      "epoch": 0.5728052865862053,
      "grad_norm": 0.4524027705192566,
      "learning_rate": 9.267981782439871e-06,
      "loss": 0.3992,
      "step": 12482
    },
    {
      "epoch": 0.5728511770914598,
      "grad_norm": 0.43138790130615234,
      "learning_rate": 9.267854050349269e-06,
      "loss": 0.3514,
      "step": 12483
    },
    {
      "epoch": 0.5728970675967142,
      "grad_norm": 0.43252289295196533,
      "learning_rate": 9.267726307995824e-06,
      "loss": 0.3518,
      "step": 12484
    },
    {
      "epoch": 0.5729429581019687,
      "grad_norm": 0.4815121293067932,
      "learning_rate": 9.267598555379846e-06,
      "loss": 0.44,
      "step": 12485
    },
    {
      "epoch": 0.5729888486072232,
      "grad_norm": 0.45449697971343994,
      "learning_rate": 9.267470792501639e-06,
      "loss": 0.4081,
      "step": 12486
    },
    {
      "epoch": 0.5730347391124776,
      "grad_norm": 0.4681018888950348,
      "learning_rate": 9.26734301936151e-06,
      "loss": 0.4436,
      "step": 12487
    },
    {
      "epoch": 0.5730806296177321,
      "grad_norm": 0.45586466789245605,
      "learning_rate": 9.26721523595977e-06,
      "loss": 0.4667,
      "step": 12488
    },
    {
      "epoch": 0.5731265201229866,
      "grad_norm": 0.48727309703826904,
      "learning_rate": 9.267087442296725e-06,
      "loss": 0.5274,
      "step": 12489
    },
    {
      "epoch": 0.573172410628241,
      "grad_norm": 0.46886223554611206,
      "learning_rate": 9.266959638372678e-06,
      "loss": 0.4183,
      "step": 12490
    },
    {
      "epoch": 0.5732183011334955,
      "grad_norm": 0.4830036163330078,
      "learning_rate": 9.266831824187942e-06,
      "loss": 0.4417,
      "step": 12491
    },
    {
      "epoch": 0.57326419163875,
      "grad_norm": 0.43994027376174927,
      "learning_rate": 9.266703999742824e-06,
      "loss": 0.3357,
      "step": 12492
    },
    {
      "epoch": 0.5733100821440044,
      "grad_norm": 0.45810234546661377,
      "learning_rate": 9.266576165037627e-06,
      "loss": 0.3888,
      "step": 12493
    },
    {
      "epoch": 0.5733559726492589,
      "grad_norm": 0.5088083744049072,
      "learning_rate": 9.266448320072661e-06,
      "loss": 0.5074,
      "step": 12494
    },
    {
      "epoch": 0.5734018631545134,
      "grad_norm": 0.4527965486049652,
      "learning_rate": 9.266320464848233e-06,
      "loss": 0.3634,
      "step": 12495
    },
    {
      "epoch": 0.5734477536597677,
      "grad_norm": 0.4559270739555359,
      "learning_rate": 9.266192599364653e-06,
      "loss": 0.4186,
      "step": 12496
    },
    {
      "epoch": 0.5734936441650222,
      "grad_norm": 0.4693770706653595,
      "learning_rate": 9.266064723622225e-06,
      "loss": 0.4186,
      "step": 12497
    },
    {
      "epoch": 0.5735395346702767,
      "grad_norm": 0.4442390501499176,
      "learning_rate": 9.265936837621257e-06,
      "loss": 0.343,
      "step": 12498
    },
    {
      "epoch": 0.5735854251755311,
      "grad_norm": 0.4576526880264282,
      "learning_rate": 9.265808941362057e-06,
      "loss": 0.4029,
      "step": 12499
    },
    {
      "epoch": 0.5736313156807856,
      "grad_norm": 0.4814698398113251,
      "learning_rate": 9.265681034844935e-06,
      "loss": 0.3962,
      "step": 12500
    },
    {
      "epoch": 0.5736772061860401,
      "grad_norm": 0.45719659328460693,
      "learning_rate": 9.265553118070194e-06,
      "loss": 0.4066,
      "step": 12501
    },
    {
      "epoch": 0.5737230966912946,
      "grad_norm": 0.42948174476623535,
      "learning_rate": 9.265425191038145e-06,
      "loss": 0.332,
      "step": 12502
    },
    {
      "epoch": 0.573768987196549,
      "grad_norm": 0.5138431787490845,
      "learning_rate": 9.265297253749095e-06,
      "loss": 0.3634,
      "step": 12503
    },
    {
      "epoch": 0.5738148777018035,
      "grad_norm": 0.4744110107421875,
      "learning_rate": 9.26516930620335e-06,
      "loss": 0.3674,
      "step": 12504
    },
    {
      "epoch": 0.573860768207058,
      "grad_norm": 0.44775137305259705,
      "learning_rate": 9.26504134840122e-06,
      "loss": 0.3854,
      "step": 12505
    },
    {
      "epoch": 0.5739066587123124,
      "grad_norm": 0.4446086883544922,
      "learning_rate": 9.264913380343012e-06,
      "loss": 0.3962,
      "step": 12506
    },
    {
      "epoch": 0.5739525492175669,
      "grad_norm": 0.486645370721817,
      "learning_rate": 9.264785402029032e-06,
      "loss": 0.4352,
      "step": 12507
    },
    {
      "epoch": 0.5739984397228214,
      "grad_norm": 0.45591336488723755,
      "learning_rate": 9.26465741345959e-06,
      "loss": 0.3715,
      "step": 12508
    },
    {
      "epoch": 0.5740443302280758,
      "grad_norm": 0.43360835313796997,
      "learning_rate": 9.264529414634993e-06,
      "loss": 0.2985,
      "step": 12509
    },
    {
      "epoch": 0.5740902207333303,
      "grad_norm": 0.4525274634361267,
      "learning_rate": 9.26440140555555e-06,
      "loss": 0.3746,
      "step": 12510
    },
    {
      "epoch": 0.5741361112385848,
      "grad_norm": 0.4483006000518799,
      "learning_rate": 9.264273386221565e-06,
      "loss": 0.3833,
      "step": 12511
    },
    {
      "epoch": 0.5741820017438392,
      "grad_norm": 0.5403188467025757,
      "learning_rate": 9.264145356633349e-06,
      "loss": 0.444,
      "step": 12512
    },
    {
      "epoch": 0.5742278922490937,
      "grad_norm": 0.4745635390281677,
      "learning_rate": 9.26401731679121e-06,
      "loss": 0.4284,
      "step": 12513
    },
    {
      "epoch": 0.5742737827543481,
      "grad_norm": 0.4384280741214752,
      "learning_rate": 9.263889266695455e-06,
      "loss": 0.3585,
      "step": 12514
    },
    {
      "epoch": 0.5743196732596025,
      "grad_norm": 0.477106511592865,
      "learning_rate": 9.263761206346392e-06,
      "loss": 0.4046,
      "step": 12515
    },
    {
      "epoch": 0.574365563764857,
      "grad_norm": 0.4570589065551758,
      "learning_rate": 9.263633135744327e-06,
      "loss": 0.4352,
      "step": 12516
    },
    {
      "epoch": 0.5744114542701115,
      "grad_norm": 0.47551780939102173,
      "learning_rate": 9.263505054889572e-06,
      "loss": 0.3888,
      "step": 12517
    },
    {
      "epoch": 0.574457344775366,
      "grad_norm": 0.4811258316040039,
      "learning_rate": 9.263376963782434e-06,
      "loss": 0.4461,
      "step": 12518
    },
    {
      "epoch": 0.5745032352806204,
      "grad_norm": 0.42558398842811584,
      "learning_rate": 9.263248862423217e-06,
      "loss": 0.3786,
      "step": 12519
    },
    {
      "epoch": 0.5745491257858749,
      "grad_norm": 0.44696101546287537,
      "learning_rate": 9.263120750812236e-06,
      "loss": 0.426,
      "step": 12520
    },
    {
      "epoch": 0.5745950162911294,
      "grad_norm": 0.46655941009521484,
      "learning_rate": 9.262992628949792e-06,
      "loss": 0.437,
      "step": 12521
    },
    {
      "epoch": 0.5746409067963838,
      "grad_norm": 0.43725064396858215,
      "learning_rate": 9.262864496836197e-06,
      "loss": 0.3533,
      "step": 12522
    },
    {
      "epoch": 0.5746867973016383,
      "grad_norm": 0.42130446434020996,
      "learning_rate": 9.262736354471757e-06,
      "loss": 0.3299,
      "step": 12523
    },
    {
      "epoch": 0.5747326878068928,
      "grad_norm": 0.4209478497505188,
      "learning_rate": 9.262608201856783e-06,
      "loss": 0.3498,
      "step": 12524
    },
    {
      "epoch": 0.5747785783121472,
      "grad_norm": 0.4715080261230469,
      "learning_rate": 9.262480038991581e-06,
      "loss": 0.4646,
      "step": 12525
    },
    {
      "epoch": 0.5748244688174017,
      "grad_norm": 0.5360161662101746,
      "learning_rate": 9.26235186587646e-06,
      "loss": 0.3933,
      "step": 12526
    },
    {
      "epoch": 0.5748703593226562,
      "grad_norm": 0.42833325266838074,
      "learning_rate": 9.26222368251173e-06,
      "loss": 0.3197,
      "step": 12527
    },
    {
      "epoch": 0.5749162498279106,
      "grad_norm": 0.4850447475910187,
      "learning_rate": 9.262095488897694e-06,
      "loss": 0.4705,
      "step": 12528
    },
    {
      "epoch": 0.5749621403331651,
      "grad_norm": 0.4339088201522827,
      "learning_rate": 9.261967285034663e-06,
      "loss": 0.3984,
      "step": 12529
    },
    {
      "epoch": 0.5750080308384196,
      "grad_norm": 0.47556325793266296,
      "learning_rate": 9.26183907092295e-06,
      "loss": 0.4502,
      "step": 12530
    },
    {
      "epoch": 0.575053921343674,
      "grad_norm": 0.42540550231933594,
      "learning_rate": 9.261710846562855e-06,
      "loss": 0.3495,
      "step": 12531
    },
    {
      "epoch": 0.5750998118489284,
      "grad_norm": 0.4327041506767273,
      "learning_rate": 9.261582611954691e-06,
      "loss": 0.3361,
      "step": 12532
    },
    {
      "epoch": 0.5751457023541829,
      "grad_norm": 0.4750451147556305,
      "learning_rate": 9.261454367098767e-06,
      "loss": 0.4151,
      "step": 12533
    },
    {
      "epoch": 0.5751915928594373,
      "grad_norm": 0.47268441319465637,
      "learning_rate": 9.261326111995391e-06,
      "loss": 0.44,
      "step": 12534
    },
    {
      "epoch": 0.5752374833646918,
      "grad_norm": 0.44058722257614136,
      "learning_rate": 9.26119784664487e-06,
      "loss": 0.3601,
      "step": 12535
    },
    {
      "epoch": 0.5752833738699463,
      "grad_norm": 0.45171108841896057,
      "learning_rate": 9.261069571047513e-06,
      "loss": 0.3832,
      "step": 12536
    },
    {
      "epoch": 0.5753292643752008,
      "grad_norm": 0.47940173745155334,
      "learning_rate": 9.260941285203627e-06,
      "loss": 0.407,
      "step": 12537
    },
    {
      "epoch": 0.5753751548804552,
      "grad_norm": 0.42961037158966064,
      "learning_rate": 9.260812989113523e-06,
      "loss": 0.3125,
      "step": 12538
    },
    {
      "epoch": 0.5754210453857097,
      "grad_norm": 0.46799832582473755,
      "learning_rate": 9.26068468277751e-06,
      "loss": 0.4577,
      "step": 12539
    },
    {
      "epoch": 0.5754669358909642,
      "grad_norm": 0.4417872130870819,
      "learning_rate": 9.260556366195893e-06,
      "loss": 0.3986,
      "step": 12540
    },
    {
      "epoch": 0.5755128263962186,
      "grad_norm": 0.44156181812286377,
      "learning_rate": 9.260428039368982e-06,
      "loss": 0.4196,
      "step": 12541
    },
    {
      "epoch": 0.5755587169014731,
      "grad_norm": 0.4396057426929474,
      "learning_rate": 9.26029970229709e-06,
      "loss": 0.3349,
      "step": 12542
    },
    {
      "epoch": 0.5756046074067276,
      "grad_norm": 0.45526188611984253,
      "learning_rate": 9.260171354980518e-06,
      "loss": 0.3734,
      "step": 12543
    },
    {
      "epoch": 0.575650497911982,
      "grad_norm": 0.44120582938194275,
      "learning_rate": 9.26004299741958e-06,
      "loss": 0.341,
      "step": 12544
    },
    {
      "epoch": 0.5756963884172365,
      "grad_norm": 0.4317726194858551,
      "learning_rate": 9.259914629614583e-06,
      "loss": 0.335,
      "step": 12545
    },
    {
      "epoch": 0.575742278922491,
      "grad_norm": 0.49047619104385376,
      "learning_rate": 9.259786251565835e-06,
      "loss": 0.4674,
      "step": 12546
    },
    {
      "epoch": 0.5757881694277454,
      "grad_norm": 0.42871949076652527,
      "learning_rate": 9.259657863273648e-06,
      "loss": 0.3294,
      "step": 12547
    },
    {
      "epoch": 0.5758340599329999,
      "grad_norm": 0.4375298321247101,
      "learning_rate": 9.259529464738326e-06,
      "loss": 0.3966,
      "step": 12548
    },
    {
      "epoch": 0.5758799504382544,
      "grad_norm": 0.4428410232067108,
      "learning_rate": 9.25940105596018e-06,
      "loss": 0.3188,
      "step": 12549
    },
    {
      "epoch": 0.5759258409435087,
      "grad_norm": 0.46383482217788696,
      "learning_rate": 9.25927263693952e-06,
      "loss": 0.4116,
      "step": 12550
    },
    {
      "epoch": 0.5759717314487632,
      "grad_norm": 0.49060237407684326,
      "learning_rate": 9.25914420767665e-06,
      "loss": 0.4981,
      "step": 12551
    },
    {
      "epoch": 0.5760176219540177,
      "grad_norm": 0.4574492871761322,
      "learning_rate": 9.259015768171886e-06,
      "loss": 0.39,
      "step": 12552
    },
    {
      "epoch": 0.5760635124592722,
      "grad_norm": 0.40163856744766235,
      "learning_rate": 9.258887318425533e-06,
      "loss": 0.2927,
      "step": 12553
    },
    {
      "epoch": 0.5761094029645266,
      "grad_norm": 0.4383412301540375,
      "learning_rate": 9.258758858437899e-06,
      "loss": 0.378,
      "step": 12554
    },
    {
      "epoch": 0.5761552934697811,
      "grad_norm": 0.44665706157684326,
      "learning_rate": 9.258630388209296e-06,
      "loss": 0.3741,
      "step": 12555
    },
    {
      "epoch": 0.5762011839750356,
      "grad_norm": 0.4523334205150604,
      "learning_rate": 9.25850190774003e-06,
      "loss": 0.3845,
      "step": 12556
    },
    {
      "epoch": 0.57624707448029,
      "grad_norm": 0.5325130820274353,
      "learning_rate": 9.258373417030409e-06,
      "loss": 0.5737,
      "step": 12557
    },
    {
      "epoch": 0.5762929649855445,
      "grad_norm": 0.4551815390586853,
      "learning_rate": 9.258244916080745e-06,
      "loss": 0.4092,
      "step": 12558
    },
    {
      "epoch": 0.576338855490799,
      "grad_norm": 0.442103773355484,
      "learning_rate": 9.258116404891346e-06,
      "loss": 0.3817,
      "step": 12559
    },
    {
      "epoch": 0.5763847459960534,
      "grad_norm": 0.5180836915969849,
      "learning_rate": 9.257987883462523e-06,
      "loss": 0.3605,
      "step": 12560
    },
    {
      "epoch": 0.5764306365013079,
      "grad_norm": 0.4852864146232605,
      "learning_rate": 9.25785935179458e-06,
      "loss": 0.5063,
      "step": 12561
    },
    {
      "epoch": 0.5764765270065624,
      "grad_norm": 0.4747324287891388,
      "learning_rate": 9.25773080988783e-06,
      "loss": 0.4429,
      "step": 12562
    },
    {
      "epoch": 0.5765224175118168,
      "grad_norm": 0.44937577843666077,
      "learning_rate": 9.257602257742581e-06,
      "loss": 0.4234,
      "step": 12563
    },
    {
      "epoch": 0.5765683080170713,
      "grad_norm": 0.4900214672088623,
      "learning_rate": 9.257473695359145e-06,
      "loss": 0.4454,
      "step": 12564
    },
    {
      "epoch": 0.5766141985223258,
      "grad_norm": 0.455264151096344,
      "learning_rate": 9.257345122737827e-06,
      "loss": 0.3906,
      "step": 12565
    },
    {
      "epoch": 0.5766600890275801,
      "grad_norm": 0.43089550733566284,
      "learning_rate": 9.257216539878936e-06,
      "loss": 0.3673,
      "step": 12566
    },
    {
      "epoch": 0.5767059795328346,
      "grad_norm": 0.41970348358154297,
      "learning_rate": 9.257087946782784e-06,
      "loss": 0.3512,
      "step": 12567
    },
    {
      "epoch": 0.5767518700380891,
      "grad_norm": 0.5235967040061951,
      "learning_rate": 9.256959343449679e-06,
      "loss": 0.5105,
      "step": 12568
    },
    {
      "epoch": 0.5767977605433435,
      "grad_norm": 0.460002064704895,
      "learning_rate": 9.25683072987993e-06,
      "loss": 0.4157,
      "step": 12569
    },
    {
      "epoch": 0.576843651048598,
      "grad_norm": 0.4280596077442169,
      "learning_rate": 9.256702106073847e-06,
      "loss": 0.3994,
      "step": 12570
    },
    {
      "epoch": 0.5768895415538525,
      "grad_norm": 0.5014805197715759,
      "learning_rate": 9.256573472031737e-06,
      "loss": 0.4735,
      "step": 12571
    },
    {
      "epoch": 0.576935432059107,
      "grad_norm": 0.4504304528236389,
      "learning_rate": 9.256444827753914e-06,
      "loss": 0.376,
      "step": 12572
    },
    {
      "epoch": 0.5769813225643614,
      "grad_norm": 0.47284138202667236,
      "learning_rate": 9.256316173240683e-06,
      "loss": 0.4107,
      "step": 12573
    },
    {
      "epoch": 0.5770272130696159,
      "grad_norm": 0.4578664302825928,
      "learning_rate": 9.256187508492356e-06,
      "loss": 0.3668,
      "step": 12574
    },
    {
      "epoch": 0.5770731035748704,
      "grad_norm": 0.4515513777732849,
      "learning_rate": 9.25605883350924e-06,
      "loss": 0.3592,
      "step": 12575
    },
    {
      "epoch": 0.5771189940801248,
      "grad_norm": 0.48279306292533875,
      "learning_rate": 9.255930148291646e-06,
      "loss": 0.4758,
      "step": 12576
    },
    {
      "epoch": 0.5771648845853793,
      "grad_norm": 0.4338676929473877,
      "learning_rate": 9.255801452839885e-06,
      "loss": 0.3521,
      "step": 12577
    },
    {
      "epoch": 0.5772107750906338,
      "grad_norm": 0.43797609210014343,
      "learning_rate": 9.255672747154263e-06,
      "loss": 0.3461,
      "step": 12578
    },
    {
      "epoch": 0.5772566655958882,
      "grad_norm": 0.4690699577331543,
      "learning_rate": 9.25554403123509e-06,
      "loss": 0.4216,
      "step": 12579
    },
    {
      "epoch": 0.5773025561011427,
      "grad_norm": 0.447279155254364,
      "learning_rate": 9.255415305082681e-06,
      "loss": 0.4089,
      "step": 12580
    },
    {
      "epoch": 0.5773484466063972,
      "grad_norm": 0.45294392108917236,
      "learning_rate": 9.255286568697337e-06,
      "loss": 0.4177,
      "step": 12581
    },
    {
      "epoch": 0.5773943371116516,
      "grad_norm": 0.4197905361652374,
      "learning_rate": 9.255157822079373e-06,
      "loss": 0.3395,
      "step": 12582
    },
    {
      "epoch": 0.577440227616906,
      "grad_norm": 0.45428162813186646,
      "learning_rate": 9.255029065229097e-06,
      "loss": 0.4472,
      "step": 12583
    },
    {
      "epoch": 0.5774861181221606,
      "grad_norm": 0.44238606095314026,
      "learning_rate": 9.25490029814682e-06,
      "loss": 0.3714,
      "step": 12584
    },
    {
      "epoch": 0.5775320086274149,
      "grad_norm": 0.4275534451007843,
      "learning_rate": 9.25477152083285e-06,
      "loss": 0.3461,
      "step": 12585
    },
    {
      "epoch": 0.5775778991326694,
      "grad_norm": 0.45596247911453247,
      "learning_rate": 9.254642733287498e-06,
      "loss": 0.4239,
      "step": 12586
    },
    {
      "epoch": 0.5776237896379239,
      "grad_norm": 0.4034506678581238,
      "learning_rate": 9.254513935511073e-06,
      "loss": 0.2921,
      "step": 12587
    },
    {
      "epoch": 0.5776696801431783,
      "grad_norm": 0.46780630946159363,
      "learning_rate": 9.254385127503884e-06,
      "loss": 0.4069,
      "step": 12588
    },
    {
      "epoch": 0.5777155706484328,
      "grad_norm": 0.432293176651001,
      "learning_rate": 9.254256309266242e-06,
      "loss": 0.3323,
      "step": 12589
    },
    {
      "epoch": 0.5777614611536873,
      "grad_norm": 0.43386217951774597,
      "learning_rate": 9.254127480798456e-06,
      "loss": 0.3912,
      "step": 12590
    },
    {
      "epoch": 0.5778073516589418,
      "grad_norm": 0.4603815972805023,
      "learning_rate": 9.253998642100837e-06,
      "loss": 0.4116,
      "step": 12591
    },
    {
      "epoch": 0.5778532421641962,
      "grad_norm": 0.43377038836479187,
      "learning_rate": 9.253869793173693e-06,
      "loss": 0.3266,
      "step": 12592
    },
    {
      "epoch": 0.5778991326694507,
      "grad_norm": 0.42841583490371704,
      "learning_rate": 9.253740934017334e-06,
      "loss": 0.3766,
      "step": 12593
    },
    {
      "epoch": 0.5779450231747052,
      "grad_norm": 0.4495267868041992,
      "learning_rate": 9.253612064632073e-06,
      "loss": 0.3488,
      "step": 12594
    },
    {
      "epoch": 0.5779909136799596,
      "grad_norm": 0.4960354268550873,
      "learning_rate": 9.253483185018215e-06,
      "loss": 0.4806,
      "step": 12595
    },
    {
      "epoch": 0.5780368041852141,
      "grad_norm": 0.43426960706710815,
      "learning_rate": 9.253354295176074e-06,
      "loss": 0.3523,
      "step": 12596
    },
    {
      "epoch": 0.5780826946904686,
      "grad_norm": 0.47681307792663574,
      "learning_rate": 9.253225395105957e-06,
      "loss": 0.4644,
      "step": 12597
    },
    {
      "epoch": 0.578128585195723,
      "grad_norm": 0.47388502955436707,
      "learning_rate": 9.253096484808174e-06,
      "loss": 0.4434,
      "step": 12598
    },
    {
      "epoch": 0.5781744757009775,
      "grad_norm": 0.4419380724430084,
      "learning_rate": 9.25296756428304e-06,
      "loss": 0.4233,
      "step": 12599
    },
    {
      "epoch": 0.578220366206232,
      "grad_norm": 0.42498525977134705,
      "learning_rate": 9.252838633530859e-06,
      "loss": 0.364,
      "step": 12600
    },
    {
      "epoch": 0.5782662567114863,
      "grad_norm": 0.4250074028968811,
      "learning_rate": 9.252709692551944e-06,
      "loss": 0.3456,
      "step": 12601
    },
    {
      "epoch": 0.5783121472167408,
      "grad_norm": 0.44671469926834106,
      "learning_rate": 9.252580741346604e-06,
      "loss": 0.3926,
      "step": 12602
    },
    {
      "epoch": 0.5783580377219953,
      "grad_norm": 1.0606437921524048,
      "learning_rate": 9.25245177991515e-06,
      "loss": 0.416,
      "step": 12603
    },
    {
      "epoch": 0.5784039282272497,
      "grad_norm": 1.2155812978744507,
      "learning_rate": 9.252322808257891e-06,
      "loss": 0.4173,
      "step": 12604
    },
    {
      "epoch": 0.5784498187325042,
      "grad_norm": 0.45152273774147034,
      "learning_rate": 9.252193826375137e-06,
      "loss": 0.3432,
      "step": 12605
    },
    {
      "epoch": 0.5784957092377587,
      "grad_norm": 0.4285723567008972,
      "learning_rate": 9.2520648342672e-06,
      "loss": 0.3253,
      "step": 12606
    },
    {
      "epoch": 0.5785415997430132,
      "grad_norm": 0.47818103432655334,
      "learning_rate": 9.25193583193439e-06,
      "loss": 0.42,
      "step": 12607
    },
    {
      "epoch": 0.5785874902482676,
      "grad_norm": 0.47493234276771545,
      "learning_rate": 9.251806819377013e-06,
      "loss": 0.4245,
      "step": 12608
    },
    {
      "epoch": 0.5786333807535221,
      "grad_norm": 0.45272698998451233,
      "learning_rate": 9.251677796595386e-06,
      "loss": 0.3802,
      "step": 12609
    },
    {
      "epoch": 0.5786792712587766,
      "grad_norm": 0.48176708817481995,
      "learning_rate": 9.251548763589815e-06,
      "loss": 0.4999,
      "step": 12610
    },
    {
      "epoch": 0.578725161764031,
      "grad_norm": 0.44258081912994385,
      "learning_rate": 9.251419720360611e-06,
      "loss": 0.3405,
      "step": 12611
    },
    {
      "epoch": 0.5787710522692855,
      "grad_norm": 0.5007138252258301,
      "learning_rate": 9.251290666908082e-06,
      "loss": 0.5337,
      "step": 12612
    },
    {
      "epoch": 0.57881694277454,
      "grad_norm": 0.44774025678634644,
      "learning_rate": 9.251161603232544e-06,
      "loss": 0.4167,
      "step": 12613
    },
    {
      "epoch": 0.5788628332797944,
      "grad_norm": 0.4723808169364929,
      "learning_rate": 9.2510325293343e-06,
      "loss": 0.4288,
      "step": 12614
    },
    {
      "epoch": 0.5789087237850489,
      "grad_norm": 0.5099866986274719,
      "learning_rate": 9.250903445213668e-06,
      "loss": 0.394,
      "step": 12615
    },
    {
      "epoch": 0.5789546142903034,
      "grad_norm": 0.4534531235694885,
      "learning_rate": 9.250774350870954e-06,
      "loss": 0.3379,
      "step": 12616
    },
    {
      "epoch": 0.5790005047955578,
      "grad_norm": 0.4622512459754944,
      "learning_rate": 9.250645246306468e-06,
      "loss": 0.4141,
      "step": 12617
    },
    {
      "epoch": 0.5790463953008123,
      "grad_norm": 0.4934578537940979,
      "learning_rate": 9.250516131520525e-06,
      "loss": 0.4415,
      "step": 12618
    },
    {
      "epoch": 0.5790922858060668,
      "grad_norm": 0.4687318801879883,
      "learning_rate": 9.250387006513429e-06,
      "loss": 0.4715,
      "step": 12619
    },
    {
      "epoch": 0.5791381763113211,
      "grad_norm": 0.45984745025634766,
      "learning_rate": 9.250257871285495e-06,
      "loss": 0.4028,
      "step": 12620
    },
    {
      "epoch": 0.5791840668165756,
      "grad_norm": 0.4395900368690491,
      "learning_rate": 9.250128725837032e-06,
      "loss": 0.3503,
      "step": 12621
    },
    {
      "epoch": 0.5792299573218301,
      "grad_norm": 0.553170919418335,
      "learning_rate": 9.249999570168351e-06,
      "loss": 0.523,
      "step": 12622
    },
    {
      "epoch": 0.5792758478270845,
      "grad_norm": 0.44351068139076233,
      "learning_rate": 9.249870404279761e-06,
      "loss": 0.3423,
      "step": 12623
    },
    {
      "epoch": 0.579321738332339,
      "grad_norm": 0.4908098578453064,
      "learning_rate": 9.249741228171576e-06,
      "loss": 0.4551,
      "step": 12624
    },
    {
      "epoch": 0.5793676288375935,
      "grad_norm": 0.44533997774124146,
      "learning_rate": 9.249612041844103e-06,
      "loss": 0.3868,
      "step": 12625
    },
    {
      "epoch": 0.579413519342848,
      "grad_norm": 0.4501284956932068,
      "learning_rate": 9.249482845297655e-06,
      "loss": 0.4242,
      "step": 12626
    },
    {
      "epoch": 0.5794594098481024,
      "grad_norm": 0.48862409591674805,
      "learning_rate": 9.249353638532541e-06,
      "loss": 0.4189,
      "step": 12627
    },
    {
      "epoch": 0.5795053003533569,
      "grad_norm": 0.4677802324295044,
      "learning_rate": 9.249224421549074e-06,
      "loss": 0.4245,
      "step": 12628
    },
    {
      "epoch": 0.5795511908586114,
      "grad_norm": 0.4659879207611084,
      "learning_rate": 9.249095194347563e-06,
      "loss": 0.377,
      "step": 12629
    },
    {
      "epoch": 0.5795970813638658,
      "grad_norm": 0.434348464012146,
      "learning_rate": 9.24896595692832e-06,
      "loss": 0.3269,
      "step": 12630
    },
    {
      "epoch": 0.5796429718691203,
      "grad_norm": 0.47228026390075684,
      "learning_rate": 9.248836709291653e-06,
      "loss": 0.4207,
      "step": 12631
    },
    {
      "epoch": 0.5796888623743748,
      "grad_norm": 0.4434499144554138,
      "learning_rate": 9.248707451437875e-06,
      "loss": 0.3502,
      "step": 12632
    },
    {
      "epoch": 0.5797347528796292,
      "grad_norm": 0.41379526257514954,
      "learning_rate": 9.248578183367298e-06,
      "loss": 0.3284,
      "step": 12633
    },
    {
      "epoch": 0.5797806433848837,
      "grad_norm": 0.4771484434604645,
      "learning_rate": 9.24844890508023e-06,
      "loss": 0.4753,
      "step": 12634
    },
    {
      "epoch": 0.5798265338901382,
      "grad_norm": 0.4715487062931061,
      "learning_rate": 9.248319616576984e-06,
      "loss": 0.4184,
      "step": 12635
    },
    {
      "epoch": 0.5798724243953925,
      "grad_norm": 0.4768238067626953,
      "learning_rate": 9.24819031785787e-06,
      "loss": 0.4328,
      "step": 12636
    },
    {
      "epoch": 0.579918314900647,
      "grad_norm": 0.4236830472946167,
      "learning_rate": 9.2480610089232e-06,
      "loss": 0.371,
      "step": 12637
    },
    {
      "epoch": 0.5799642054059015,
      "grad_norm": 0.4577537775039673,
      "learning_rate": 9.247931689773281e-06,
      "loss": 0.3863,
      "step": 12638
    },
    {
      "epoch": 0.5800100959111559,
      "grad_norm": 0.43757155537605286,
      "learning_rate": 9.247802360408428e-06,
      "loss": 0.341,
      "step": 12639
    },
    {
      "epoch": 0.5800559864164104,
      "grad_norm": 0.42257827520370483,
      "learning_rate": 9.247673020828953e-06,
      "loss": 0.3469,
      "step": 12640
    },
    {
      "epoch": 0.5801018769216649,
      "grad_norm": 0.4678580164909363,
      "learning_rate": 9.247543671035164e-06,
      "loss": 0.3983,
      "step": 12641
    },
    {
      "epoch": 0.5801477674269194,
      "grad_norm": 0.49686145782470703,
      "learning_rate": 9.247414311027372e-06,
      "loss": 0.4309,
      "step": 12642
    },
    {
      "epoch": 0.5801936579321738,
      "grad_norm": 0.4756784737110138,
      "learning_rate": 9.24728494080589e-06,
      "loss": 0.3999,
      "step": 12643
    },
    {
      "epoch": 0.5802395484374283,
      "grad_norm": 0.4505109488964081,
      "learning_rate": 9.247155560371028e-06,
      "loss": 0.3575,
      "step": 12644
    },
    {
      "epoch": 0.5802854389426828,
      "grad_norm": 0.4537818729877472,
      "learning_rate": 9.247026169723098e-06,
      "loss": 0.3964,
      "step": 12645
    },
    {
      "epoch": 0.5803313294479372,
      "grad_norm": 0.4446287155151367,
      "learning_rate": 9.24689676886241e-06,
      "loss": 0.3825,
      "step": 12646
    },
    {
      "epoch": 0.5803772199531917,
      "grad_norm": 0.4738503396511078,
      "learning_rate": 9.246767357789274e-06,
      "loss": 0.3932,
      "step": 12647
    },
    {
      "epoch": 0.5804231104584462,
      "grad_norm": 0.4947744905948639,
      "learning_rate": 9.246637936504006e-06,
      "loss": 0.4513,
      "step": 12648
    },
    {
      "epoch": 0.5804690009637006,
      "grad_norm": 0.4701147973537445,
      "learning_rate": 9.246508505006912e-06,
      "loss": 0.3979,
      "step": 12649
    },
    {
      "epoch": 0.5805148914689551,
      "grad_norm": 0.5339137315750122,
      "learning_rate": 9.246379063298306e-06,
      "loss": 0.5207,
      "step": 12650
    },
    {
      "epoch": 0.5805607819742096,
      "grad_norm": 0.48089513182640076,
      "learning_rate": 9.246249611378497e-06,
      "loss": 0.4652,
      "step": 12651
    },
    {
      "epoch": 0.580606672479464,
      "grad_norm": 0.42181867361068726,
      "learning_rate": 9.246120149247799e-06,
      "loss": 0.3223,
      "step": 12652
    },
    {
      "epoch": 0.5806525629847185,
      "grad_norm": 0.5170144438743591,
      "learning_rate": 9.245990676906522e-06,
      "loss": 0.5261,
      "step": 12653
    },
    {
      "epoch": 0.580698453489973,
      "grad_norm": 0.5422033071517944,
      "learning_rate": 9.24586119435498e-06,
      "loss": 0.5302,
      "step": 12654
    },
    {
      "epoch": 0.5807443439952273,
      "grad_norm": 0.4499092996120453,
      "learning_rate": 9.245731701593477e-06,
      "loss": 0.4105,
      "step": 12655
    },
    {
      "epoch": 0.5807902345004818,
      "grad_norm": 0.43707072734832764,
      "learning_rate": 9.245602198622332e-06,
      "loss": 0.3927,
      "step": 12656
    },
    {
      "epoch": 0.5808361250057363,
      "grad_norm": 0.46806856989860535,
      "learning_rate": 9.245472685441854e-06,
      "loss": 0.4405,
      "step": 12657
    },
    {
      "epoch": 0.5808820155109907,
      "grad_norm": 0.47955837845802307,
      "learning_rate": 9.245343162052353e-06,
      "loss": 0.4913,
      "step": 12658
    },
    {
      "epoch": 0.5809279060162452,
      "grad_norm": 0.4481831192970276,
      "learning_rate": 9.245213628454143e-06,
      "loss": 0.4079,
      "step": 12659
    },
    {
      "epoch": 0.5809737965214997,
      "grad_norm": 0.500677764415741,
      "learning_rate": 9.245084084647533e-06,
      "loss": 0.4098,
      "step": 12660
    },
    {
      "epoch": 0.5810196870267542,
      "grad_norm": 0.6144744753837585,
      "learning_rate": 9.244954530632836e-06,
      "loss": 0.4519,
      "step": 12661
    },
    {
      "epoch": 0.5810655775320086,
      "grad_norm": 0.44593825936317444,
      "learning_rate": 9.244824966410363e-06,
      "loss": 0.3581,
      "step": 12662
    },
    {
      "epoch": 0.5811114680372631,
      "grad_norm": 0.4719686210155487,
      "learning_rate": 9.244695391980425e-06,
      "loss": 0.4126,
      "step": 12663
    },
    {
      "epoch": 0.5811573585425176,
      "grad_norm": 0.487368106842041,
      "learning_rate": 9.244565807343335e-06,
      "loss": 0.4748,
      "step": 12664
    },
    {
      "epoch": 0.581203249047772,
      "grad_norm": 0.4862326383590698,
      "learning_rate": 9.244436212499403e-06,
      "loss": 0.4145,
      "step": 12665
    },
    {
      "epoch": 0.5812491395530265,
      "grad_norm": 0.4453006982803345,
      "learning_rate": 9.244306607448943e-06,
      "loss": 0.3778,
      "step": 12666
    },
    {
      "epoch": 0.581295030058281,
      "grad_norm": 0.48194485902786255,
      "learning_rate": 9.244176992192265e-06,
      "loss": 0.4246,
      "step": 12667
    },
    {
      "epoch": 0.5813409205635354,
      "grad_norm": 0.4460377097129822,
      "learning_rate": 9.244047366729677e-06,
      "loss": 0.3984,
      "step": 12668
    },
    {
      "epoch": 0.5813868110687899,
      "grad_norm": 0.4344809353351593,
      "learning_rate": 9.243917731061499e-06,
      "loss": 0.3413,
      "step": 12669
    },
    {
      "epoch": 0.5814327015740444,
      "grad_norm": 0.4771958589553833,
      "learning_rate": 9.243788085188036e-06,
      "loss": 0.4356,
      "step": 12670
    },
    {
      "epoch": 0.5814785920792988,
      "grad_norm": 0.4335649013519287,
      "learning_rate": 9.243658429109604e-06,
      "loss": 0.4024,
      "step": 12671
    },
    {
      "epoch": 0.5815244825845532,
      "grad_norm": 0.4528760313987732,
      "learning_rate": 9.243528762826512e-06,
      "loss": 0.3225,
      "step": 12672
    },
    {
      "epoch": 0.5815703730898077,
      "grad_norm": 0.48238903284072876,
      "learning_rate": 9.243399086339074e-06,
      "loss": 0.4182,
      "step": 12673
    },
    {
      "epoch": 0.5816162635950621,
      "grad_norm": 0.46099919080734253,
      "learning_rate": 9.243269399647597e-06,
      "loss": 0.4123,
      "step": 12674
    },
    {
      "epoch": 0.5816621541003166,
      "grad_norm": 0.5078523755073547,
      "learning_rate": 9.243139702752399e-06,
      "loss": 0.5051,
      "step": 12675
    },
    {
      "epoch": 0.5817080446055711,
      "grad_norm": 0.47520917654037476,
      "learning_rate": 9.243009995653788e-06,
      "loss": 0.4133,
      "step": 12676
    },
    {
      "epoch": 0.5817539351108255,
      "grad_norm": 0.46814873814582825,
      "learning_rate": 9.242880278352078e-06,
      "loss": 0.4237,
      "step": 12677
    },
    {
      "epoch": 0.58179982561608,
      "grad_norm": 0.42546555399894714,
      "learning_rate": 9.24275055084758e-06,
      "loss": 0.3612,
      "step": 12678
    },
    {
      "epoch": 0.5818457161213345,
      "grad_norm": 0.4584414064884186,
      "learning_rate": 9.242620813140604e-06,
      "loss": 0.3781,
      "step": 12679
    },
    {
      "epoch": 0.581891606626589,
      "grad_norm": 0.477372407913208,
      "learning_rate": 9.242491065231465e-06,
      "loss": 0.4519,
      "step": 12680
    },
    {
      "epoch": 0.5819374971318434,
      "grad_norm": 0.46654728055000305,
      "learning_rate": 9.242361307120475e-06,
      "loss": 0.4435,
      "step": 12681
    },
    {
      "epoch": 0.5819833876370979,
      "grad_norm": 0.4543529152870178,
      "learning_rate": 9.242231538807945e-06,
      "loss": 0.3819,
      "step": 12682
    },
    {
      "epoch": 0.5820292781423524,
      "grad_norm": 0.47681164741516113,
      "learning_rate": 9.242101760294185e-06,
      "loss": 0.4223,
      "step": 12683
    },
    {
      "epoch": 0.5820751686476068,
      "grad_norm": 0.4439181387424469,
      "learning_rate": 9.241971971579511e-06,
      "loss": 0.3605,
      "step": 12684
    },
    {
      "epoch": 0.5821210591528613,
      "grad_norm": 0.446836918592453,
      "learning_rate": 9.241842172664232e-06,
      "loss": 0.3426,
      "step": 12685
    },
    {
      "epoch": 0.5821669496581158,
      "grad_norm": 1.3848943710327148,
      "learning_rate": 9.241712363548663e-06,
      "loss": 0.5195,
      "step": 12686
    },
    {
      "epoch": 0.5822128401633702,
      "grad_norm": 0.5124891400337219,
      "learning_rate": 9.241582544233113e-06,
      "loss": 0.4932,
      "step": 12687
    },
    {
      "epoch": 0.5822587306686247,
      "grad_norm": 0.430666983127594,
      "learning_rate": 9.241452714717895e-06,
      "loss": 0.3759,
      "step": 12688
    },
    {
      "epoch": 0.5823046211738792,
      "grad_norm": 0.49588513374328613,
      "learning_rate": 9.241322875003325e-06,
      "loss": 0.4262,
      "step": 12689
    },
    {
      "epoch": 0.5823505116791335,
      "grad_norm": 0.43820464611053467,
      "learning_rate": 9.24119302508971e-06,
      "loss": 0.359,
      "step": 12690
    },
    {
      "epoch": 0.582396402184388,
      "grad_norm": 0.4562884569168091,
      "learning_rate": 9.241063164977363e-06,
      "loss": 0.4009,
      "step": 12691
    },
    {
      "epoch": 0.5824422926896425,
      "grad_norm": 0.47813984751701355,
      "learning_rate": 9.2409332946666e-06,
      "loss": 0.434,
      "step": 12692
    },
    {
      "epoch": 0.5824881831948969,
      "grad_norm": 1.996543288230896,
      "learning_rate": 9.24080341415773e-06,
      "loss": 0.5269,
      "step": 12693
    },
    {
      "epoch": 0.5825340737001514,
      "grad_norm": 0.5161787867546082,
      "learning_rate": 9.240673523451066e-06,
      "loss": 0.4896,
      "step": 12694
    },
    {
      "epoch": 0.5825799642054059,
      "grad_norm": 0.4490312933921814,
      "learning_rate": 9.24054362254692e-06,
      "loss": 0.3964,
      "step": 12695
    },
    {
      "epoch": 0.5826258547106604,
      "grad_norm": 0.492544949054718,
      "learning_rate": 9.240413711445606e-06,
      "loss": 0.4414,
      "step": 12696
    },
    {
      "epoch": 0.5826717452159148,
      "grad_norm": 0.46340063214302063,
      "learning_rate": 9.240283790147435e-06,
      "loss": 0.3439,
      "step": 12697
    },
    {
      "epoch": 0.5827176357211693,
      "grad_norm": 0.5094013214111328,
      "learning_rate": 9.240153858652721e-06,
      "loss": 0.4496,
      "step": 12698
    },
    {
      "epoch": 0.5827635262264238,
      "grad_norm": 0.5043747425079346,
      "learning_rate": 9.240023916961774e-06,
      "loss": 0.3981,
      "step": 12699
    },
    {
      "epoch": 0.5828094167316782,
      "grad_norm": 0.4240081012248993,
      "learning_rate": 9.239893965074907e-06,
      "loss": 0.3532,
      "step": 12700
    },
    {
      "epoch": 0.5828553072369327,
      "grad_norm": 0.637977659702301,
      "learning_rate": 9.239764002992433e-06,
      "loss": 0.5507,
      "step": 12701
    },
    {
      "epoch": 0.5829011977421872,
      "grad_norm": 0.4460325241088867,
      "learning_rate": 9.239634030714667e-06,
      "loss": 0.3631,
      "step": 12702
    },
    {
      "epoch": 0.5829470882474416,
      "grad_norm": 0.5296462178230286,
      "learning_rate": 9.239504048241916e-06,
      "loss": 0.4795,
      "step": 12703
    },
    {
      "epoch": 0.5829929787526961,
      "grad_norm": 0.511194109916687,
      "learning_rate": 9.239374055574498e-06,
      "loss": 0.3773,
      "step": 12704
    },
    {
      "epoch": 0.5830388692579506,
      "grad_norm": 0.4857388436794281,
      "learning_rate": 9.239244052712724e-06,
      "loss": 0.3669,
      "step": 12705
    },
    {
      "epoch": 0.583084759763205,
      "grad_norm": 0.4807306230068207,
      "learning_rate": 9.239114039656903e-06,
      "loss": 0.4215,
      "step": 12706
    },
    {
      "epoch": 0.5831306502684594,
      "grad_norm": 0.4654237627983093,
      "learning_rate": 9.238984016407354e-06,
      "loss": 0.3478,
      "step": 12707
    },
    {
      "epoch": 0.583176540773714,
      "grad_norm": 0.4653848111629486,
      "learning_rate": 9.238853982964385e-06,
      "loss": 0.3929,
      "step": 12708
    },
    {
      "epoch": 0.5832224312789683,
      "grad_norm": 0.5456488728523254,
      "learning_rate": 9.23872393932831e-06,
      "loss": 0.3775,
      "step": 12709
    },
    {
      "epoch": 0.5832683217842228,
      "grad_norm": 0.5224929451942444,
      "learning_rate": 9.23859388549944e-06,
      "loss": 0.4646,
      "step": 12710
    },
    {
      "epoch": 0.5833142122894773,
      "grad_norm": 0.4654673933982849,
      "learning_rate": 9.238463821478092e-06,
      "loss": 0.4087,
      "step": 12711
    },
    {
      "epoch": 0.5833601027947317,
      "grad_norm": 0.4765751361846924,
      "learning_rate": 9.238333747264575e-06,
      "loss": 0.4423,
      "step": 12712
    },
    {
      "epoch": 0.5834059932999862,
      "grad_norm": 0.4661179184913635,
      "learning_rate": 9.238203662859201e-06,
      "loss": 0.3779,
      "step": 12713
    },
    {
      "epoch": 0.5834518838052407,
      "grad_norm": 0.5229474902153015,
      "learning_rate": 9.238073568262288e-06,
      "loss": 0.5761,
      "step": 12714
    },
    {
      "epoch": 0.5834977743104952,
      "grad_norm": 0.4295141398906708,
      "learning_rate": 9.237943463474146e-06,
      "loss": 0.3191,
      "step": 12715
    },
    {
      "epoch": 0.5835436648157496,
      "grad_norm": 0.44346287846565247,
      "learning_rate": 9.237813348495086e-06,
      "loss": 0.44,
      "step": 12716
    },
    {
      "epoch": 0.5835895553210041,
      "grad_norm": 0.5208945870399475,
      "learning_rate": 9.237683223325422e-06,
      "loss": 0.4486,
      "step": 12717
    },
    {
      "epoch": 0.5836354458262586,
      "grad_norm": 0.4688635468482971,
      "learning_rate": 9.237553087965467e-06,
      "loss": 0.3783,
      "step": 12718
    },
    {
      "epoch": 0.583681336331513,
      "grad_norm": 0.44729936122894287,
      "learning_rate": 9.237422942415536e-06,
      "loss": 0.3568,
      "step": 12719
    },
    {
      "epoch": 0.5837272268367675,
      "grad_norm": 0.4615308940410614,
      "learning_rate": 9.23729278667594e-06,
      "loss": 0.428,
      "step": 12720
    },
    {
      "epoch": 0.583773117342022,
      "grad_norm": 0.41692161560058594,
      "learning_rate": 9.23716262074699e-06,
      "loss": 0.3272,
      "step": 12721
    },
    {
      "epoch": 0.5838190078472764,
      "grad_norm": 0.47400984168052673,
      "learning_rate": 9.237032444629004e-06,
      "loss": 0.4147,
      "step": 12722
    },
    {
      "epoch": 0.5838648983525309,
      "grad_norm": 0.4694210886955261,
      "learning_rate": 9.23690225832229e-06,
      "loss": 0.4517,
      "step": 12723
    },
    {
      "epoch": 0.5839107888577854,
      "grad_norm": 0.4737335741519928,
      "learning_rate": 9.236772061827164e-06,
      "loss": 0.4174,
      "step": 12724
    },
    {
      "epoch": 0.5839566793630397,
      "grad_norm": 0.44994741678237915,
      "learning_rate": 9.236641855143938e-06,
      "loss": 0.3417,
      "step": 12725
    },
    {
      "epoch": 0.5840025698682942,
      "grad_norm": 0.42765989899635315,
      "learning_rate": 9.236511638272926e-06,
      "loss": 0.4105,
      "step": 12726
    },
    {
      "epoch": 0.5840484603735487,
      "grad_norm": 0.43511876463890076,
      "learning_rate": 9.236381411214441e-06,
      "loss": 0.3566,
      "step": 12727
    },
    {
      "epoch": 0.5840943508788031,
      "grad_norm": 0.5826748013496399,
      "learning_rate": 9.236251173968796e-06,
      "loss": 0.5251,
      "step": 12728
    },
    {
      "epoch": 0.5841402413840576,
      "grad_norm": 0.4832215905189514,
      "learning_rate": 9.236120926536303e-06,
      "loss": 0.4301,
      "step": 12729
    },
    {
      "epoch": 0.5841861318893121,
      "grad_norm": 0.4788094758987427,
      "learning_rate": 9.235990668917275e-06,
      "loss": 0.4141,
      "step": 12730
    },
    {
      "epoch": 0.5842320223945666,
      "grad_norm": 0.47492659091949463,
      "learning_rate": 9.235860401112028e-06,
      "loss": 0.3743,
      "step": 12731
    },
    {
      "epoch": 0.584277912899821,
      "grad_norm": 0.46850553154945374,
      "learning_rate": 9.235730123120875e-06,
      "loss": 0.4369,
      "step": 12732
    },
    {
      "epoch": 0.5843238034050755,
      "grad_norm": 0.49449992179870605,
      "learning_rate": 9.235599834944124e-06,
      "loss": 0.4657,
      "step": 12733
    },
    {
      "epoch": 0.58436969391033,
      "grad_norm": 0.46150949597358704,
      "learning_rate": 9.235469536582094e-06,
      "loss": 0.4304,
      "step": 12734
    },
    {
      "epoch": 0.5844155844155844,
      "grad_norm": 0.44881072640419006,
      "learning_rate": 9.2353392280351e-06,
      "loss": 0.3408,
      "step": 12735
    },
    {
      "epoch": 0.5844614749208389,
      "grad_norm": 0.4457770586013794,
      "learning_rate": 9.235208909303446e-06,
      "loss": 0.3854,
      "step": 12736
    },
    {
      "epoch": 0.5845073654260934,
      "grad_norm": 0.47274813055992126,
      "learning_rate": 9.235078580387454e-06,
      "loss": 0.3924,
      "step": 12737
    },
    {
      "epoch": 0.5845532559313478,
      "grad_norm": 0.4498476982116699,
      "learning_rate": 9.234948241287437e-06,
      "loss": 0.3077,
      "step": 12738
    },
    {
      "epoch": 0.5845991464366023,
      "grad_norm": 0.4387124478816986,
      "learning_rate": 9.234817892003702e-06,
      "loss": 0.3632,
      "step": 12739
    },
    {
      "epoch": 0.5846450369418568,
      "grad_norm": 0.44873833656311035,
      "learning_rate": 9.234687532536567e-06,
      "loss": 0.3143,
      "step": 12740
    },
    {
      "epoch": 0.5846909274471112,
      "grad_norm": 0.46687573194503784,
      "learning_rate": 9.234557162886347e-06,
      "loss": 0.3719,
      "step": 12741
    },
    {
      "epoch": 0.5847368179523657,
      "grad_norm": 0.47468191385269165,
      "learning_rate": 9.234426783053352e-06,
      "loss": 0.3728,
      "step": 12742
    },
    {
      "epoch": 0.5847827084576201,
      "grad_norm": 0.43557316064834595,
      "learning_rate": 9.234296393037899e-06,
      "loss": 0.3754,
      "step": 12743
    },
    {
      "epoch": 0.5848285989628745,
      "grad_norm": 0.4859871566295624,
      "learning_rate": 9.234165992840299e-06,
      "loss": 0.4656,
      "step": 12744
    },
    {
      "epoch": 0.584874489468129,
      "grad_norm": 0.45823734998703003,
      "learning_rate": 9.234035582460864e-06,
      "loss": 0.4359,
      "step": 12745
    },
    {
      "epoch": 0.5849203799733835,
      "grad_norm": 0.44153136014938354,
      "learning_rate": 9.233905161899913e-06,
      "loss": 0.363,
      "step": 12746
    },
    {
      "epoch": 0.5849662704786379,
      "grad_norm": 0.4379129409790039,
      "learning_rate": 9.233774731157753e-06,
      "loss": 0.3659,
      "step": 12747
    },
    {
      "epoch": 0.5850121609838924,
      "grad_norm": 0.4464224576950073,
      "learning_rate": 9.233644290234702e-06,
      "loss": 0.3756,
      "step": 12748
    },
    {
      "epoch": 0.5850580514891469,
      "grad_norm": 0.4254448115825653,
      "learning_rate": 9.233513839131075e-06,
      "loss": 0.3488,
      "step": 12749
    },
    {
      "epoch": 0.5851039419944014,
      "grad_norm": 0.4926731586456299,
      "learning_rate": 9.23338337784718e-06,
      "loss": 0.4671,
      "step": 12750
    },
    {
      "epoch": 0.5851498324996558,
      "grad_norm": 0.4746816158294678,
      "learning_rate": 9.233252906383335e-06,
      "loss": 0.4359,
      "step": 12751
    },
    {
      "epoch": 0.5851957230049103,
      "grad_norm": 0.46839094161987305,
      "learning_rate": 9.233122424739853e-06,
      "loss": 0.4016,
      "step": 12752
    },
    {
      "epoch": 0.5852416135101648,
      "grad_norm": 0.46539875864982605,
      "learning_rate": 9.232991932917048e-06,
      "loss": 0.4629,
      "step": 12753
    },
    {
      "epoch": 0.5852875040154192,
      "grad_norm": 0.4820875823497772,
      "learning_rate": 9.232861430915235e-06,
      "loss": 0.5143,
      "step": 12754
    },
    {
      "epoch": 0.5853333945206737,
      "grad_norm": 0.480930358171463,
      "learning_rate": 9.232730918734723e-06,
      "loss": 0.4125,
      "step": 12755
    },
    {
      "epoch": 0.5853792850259282,
      "grad_norm": 0.5174856781959534,
      "learning_rate": 9.23260039637583e-06,
      "loss": 0.4734,
      "step": 12756
    },
    {
      "epoch": 0.5854251755311826,
      "grad_norm": 0.464162677526474,
      "learning_rate": 9.23246986383887e-06,
      "loss": 0.3744,
      "step": 12757
    },
    {
      "epoch": 0.5854710660364371,
      "grad_norm": 0.454582542181015,
      "learning_rate": 9.232339321124156e-06,
      "loss": 0.3676,
      "step": 12758
    },
    {
      "epoch": 0.5855169565416916,
      "grad_norm": 0.4333985149860382,
      "learning_rate": 9.232208768232e-06,
      "loss": 0.385,
      "step": 12759
    },
    {
      "epoch": 0.585562847046946,
      "grad_norm": 0.421978622674942,
      "learning_rate": 9.232078205162719e-06,
      "loss": 0.3437,
      "step": 12760
    },
    {
      "epoch": 0.5856087375522004,
      "grad_norm": 0.4656358063220978,
      "learning_rate": 9.231947631916625e-06,
      "loss": 0.3763,
      "step": 12761
    },
    {
      "epoch": 0.5856546280574549,
      "grad_norm": 0.40220463275909424,
      "learning_rate": 9.231817048494032e-06,
      "loss": 0.3562,
      "step": 12762
    },
    {
      "epoch": 0.5857005185627093,
      "grad_norm": 0.46473538875579834,
      "learning_rate": 9.231686454895255e-06,
      "loss": 0.4125,
      "step": 12763
    },
    {
      "epoch": 0.5857464090679638,
      "grad_norm": 0.44042453169822693,
      "learning_rate": 9.231555851120608e-06,
      "loss": 0.3392,
      "step": 12764
    },
    {
      "epoch": 0.5857922995732183,
      "grad_norm": 0.4504871070384979,
      "learning_rate": 9.231425237170405e-06,
      "loss": 0.3142,
      "step": 12765
    },
    {
      "epoch": 0.5858381900784727,
      "grad_norm": 0.43226027488708496,
      "learning_rate": 9.23129461304496e-06,
      "loss": 0.3209,
      "step": 12766
    },
    {
      "epoch": 0.5858840805837272,
      "grad_norm": 0.4528903067111969,
      "learning_rate": 9.231163978744586e-06,
      "loss": 0.3629,
      "step": 12767
    },
    {
      "epoch": 0.5859299710889817,
      "grad_norm": 0.4520516097545624,
      "learning_rate": 9.231033334269598e-06,
      "loss": 0.4293,
      "step": 12768
    },
    {
      "epoch": 0.5859758615942362,
      "grad_norm": 0.5013335943222046,
      "learning_rate": 9.230902679620311e-06,
      "loss": 0.5089,
      "step": 12769
    },
    {
      "epoch": 0.5860217520994906,
      "grad_norm": 0.4365525245666504,
      "learning_rate": 9.230772014797039e-06,
      "loss": 0.409,
      "step": 12770
    },
    {
      "epoch": 0.5860676426047451,
      "grad_norm": 0.42906951904296875,
      "learning_rate": 9.230641339800092e-06,
      "loss": 0.3806,
      "step": 12771
    },
    {
      "epoch": 0.5861135331099996,
      "grad_norm": 0.43968191742897034,
      "learning_rate": 9.230510654629793e-06,
      "loss": 0.3212,
      "step": 12772
    },
    {
      "epoch": 0.586159423615254,
      "grad_norm": 0.4220793843269348,
      "learning_rate": 9.230379959286446e-06,
      "loss": 0.3821,
      "step": 12773
    },
    {
      "epoch": 0.5862053141205085,
      "grad_norm": 0.48120054602622986,
      "learning_rate": 9.230249253770374e-06,
      "loss": 0.4803,
      "step": 12774
    },
    {
      "epoch": 0.586251204625763,
      "grad_norm": 0.4396708607673645,
      "learning_rate": 9.230118538081885e-06,
      "loss": 0.3354,
      "step": 12775
    },
    {
      "epoch": 0.5862970951310174,
      "grad_norm": 0.4658530056476593,
      "learning_rate": 9.229987812221297e-06,
      "loss": 0.4721,
      "step": 12776
    },
    {
      "epoch": 0.5863429856362719,
      "grad_norm": 0.4603067636489868,
      "learning_rate": 9.229857076188923e-06,
      "loss": 0.4111,
      "step": 12777
    },
    {
      "epoch": 0.5863888761415263,
      "grad_norm": 0.4657119810581207,
      "learning_rate": 9.22972632998508e-06,
      "loss": 0.441,
      "step": 12778
    },
    {
      "epoch": 0.5864347666467807,
      "grad_norm": 0.4032839834690094,
      "learning_rate": 9.229595573610078e-06,
      "loss": 0.3053,
      "step": 12779
    },
    {
      "epoch": 0.5864806571520352,
      "grad_norm": 0.46165990829467773,
      "learning_rate": 9.229464807064234e-06,
      "loss": 0.3889,
      "step": 12780
    },
    {
      "epoch": 0.5865265476572897,
      "grad_norm": 0.45121240615844727,
      "learning_rate": 9.229334030347864e-06,
      "loss": 0.3873,
      "step": 12781
    },
    {
      "epoch": 0.5865724381625441,
      "grad_norm": 0.49502989649772644,
      "learning_rate": 9.229203243461277e-06,
      "loss": 0.4821,
      "step": 12782
    },
    {
      "epoch": 0.5866183286677986,
      "grad_norm": 0.4461996257305145,
      "learning_rate": 9.229072446404792e-06,
      "loss": 0.4068,
      "step": 12783
    },
    {
      "epoch": 0.5866642191730531,
      "grad_norm": 0.47412627935409546,
      "learning_rate": 9.228941639178724e-06,
      "loss": 0.4274,
      "step": 12784
    },
    {
      "epoch": 0.5867101096783076,
      "grad_norm": 0.443317174911499,
      "learning_rate": 9.228810821783384e-06,
      "loss": 0.379,
      "step": 12785
    },
    {
      "epoch": 0.586756000183562,
      "grad_norm": 0.4475384056568146,
      "learning_rate": 9.22867999421909e-06,
      "loss": 0.3795,
      "step": 12786
    },
    {
      "epoch": 0.5868018906888165,
      "grad_norm": 0.4661663770675659,
      "learning_rate": 9.228549156486154e-06,
      "loss": 0.3886,
      "step": 12787
    },
    {
      "epoch": 0.586847781194071,
      "grad_norm": 0.4729175269603729,
      "learning_rate": 9.228418308584893e-06,
      "loss": 0.4065,
      "step": 12788
    },
    {
      "epoch": 0.5868936716993254,
      "grad_norm": 0.44993671774864197,
      "learning_rate": 9.22828745051562e-06,
      "loss": 0.3758,
      "step": 12789
    },
    {
      "epoch": 0.5869395622045799,
      "grad_norm": 0.4828965663909912,
      "learning_rate": 9.228156582278651e-06,
      "loss": 0.4563,
      "step": 12790
    },
    {
      "epoch": 0.5869854527098344,
      "grad_norm": 0.42854830622673035,
      "learning_rate": 9.2280257038743e-06,
      "loss": 0.3696,
      "step": 12791
    },
    {
      "epoch": 0.5870313432150888,
      "grad_norm": 0.4355331361293793,
      "learning_rate": 9.22789481530288e-06,
      "loss": 0.3502,
      "step": 12792
    },
    {
      "epoch": 0.5870772337203433,
      "grad_norm": 0.40995606780052185,
      "learning_rate": 9.227763916564707e-06,
      "loss": 0.3095,
      "step": 12793
    },
    {
      "epoch": 0.5871231242255978,
      "grad_norm": 0.5143141746520996,
      "learning_rate": 9.227633007660097e-06,
      "loss": 0.4462,
      "step": 12794
    },
    {
      "epoch": 0.5871690147308521,
      "grad_norm": 0.4802948832511902,
      "learning_rate": 9.227502088589365e-06,
      "loss": 0.398,
      "step": 12795
    },
    {
      "epoch": 0.5872149052361066,
      "grad_norm": 0.48264527320861816,
      "learning_rate": 9.227371159352823e-06,
      "loss": 0.354,
      "step": 12796
    },
    {
      "epoch": 0.5872607957413611,
      "grad_norm": 0.5186548233032227,
      "learning_rate": 9.22724021995079e-06,
      "loss": 0.4432,
      "step": 12797
    },
    {
      "epoch": 0.5873066862466155,
      "grad_norm": 0.44044744968414307,
      "learning_rate": 9.227109270383576e-06,
      "loss": 0.3492,
      "step": 12798
    },
    {
      "epoch": 0.58735257675187,
      "grad_norm": 0.44781428575515747,
      "learning_rate": 9.2269783106515e-06,
      "loss": 0.3649,
      "step": 12799
    },
    {
      "epoch": 0.5873984672571245,
      "grad_norm": 0.44656169414520264,
      "learning_rate": 9.226847340754875e-06,
      "loss": 0.3589,
      "step": 12800
    },
    {
      "epoch": 0.5874443577623789,
      "grad_norm": 0.4392984211444855,
      "learning_rate": 9.226716360694015e-06,
      "loss": 0.3829,
      "step": 12801
    },
    {
      "epoch": 0.5874902482676334,
      "grad_norm": 0.4719136357307434,
      "learning_rate": 9.226585370469236e-06,
      "loss": 0.4101,
      "step": 12802
    },
    {
      "epoch": 0.5875361387728879,
      "grad_norm": 0.5066999793052673,
      "learning_rate": 9.226454370080854e-06,
      "loss": 0.518,
      "step": 12803
    },
    {
      "epoch": 0.5875820292781424,
      "grad_norm": 0.48161086440086365,
      "learning_rate": 9.226323359529184e-06,
      "loss": 0.4043,
      "step": 12804
    },
    {
      "epoch": 0.5876279197833968,
      "grad_norm": 0.45203524827957153,
      "learning_rate": 9.226192338814537e-06,
      "loss": 0.3777,
      "step": 12805
    },
    {
      "epoch": 0.5876738102886513,
      "grad_norm": 0.4613286852836609,
      "learning_rate": 9.226061307937235e-06,
      "loss": 0.4458,
      "step": 12806
    },
    {
      "epoch": 0.5877197007939058,
      "grad_norm": 0.4398617148399353,
      "learning_rate": 9.225930266897588e-06,
      "loss": 0.3449,
      "step": 12807
    },
    {
      "epoch": 0.5877655912991602,
      "grad_norm": 0.4350770115852356,
      "learning_rate": 9.225799215695911e-06,
      "loss": 0.4141,
      "step": 12808
    },
    {
      "epoch": 0.5878114818044147,
      "grad_norm": 0.4399188756942749,
      "learning_rate": 9.225668154332521e-06,
      "loss": 0.356,
      "step": 12809
    },
    {
      "epoch": 0.5878573723096692,
      "grad_norm": 0.4428427815437317,
      "learning_rate": 9.225537082807734e-06,
      "loss": 0.3676,
      "step": 12810
    },
    {
      "epoch": 0.5879032628149236,
      "grad_norm": 0.43981775641441345,
      "learning_rate": 9.225406001121864e-06,
      "loss": 0.3148,
      "step": 12811
    },
    {
      "epoch": 0.587949153320178,
      "grad_norm": 0.43860548734664917,
      "learning_rate": 9.225274909275226e-06,
      "loss": 0.4029,
      "step": 12812
    },
    {
      "epoch": 0.5879950438254326,
      "grad_norm": 0.4538743495941162,
      "learning_rate": 9.225143807268135e-06,
      "loss": 0.4003,
      "step": 12813
    },
    {
      "epoch": 0.5880409343306869,
      "grad_norm": 0.40248215198516846,
      "learning_rate": 9.225012695100907e-06,
      "loss": 0.3212,
      "step": 12814
    },
    {
      "epoch": 0.5880868248359414,
      "grad_norm": 0.4172745645046234,
      "learning_rate": 9.224881572773855e-06,
      "loss": 0.3243,
      "step": 12815
    },
    {
      "epoch": 0.5881327153411959,
      "grad_norm": 0.4479377567768097,
      "learning_rate": 9.224750440287299e-06,
      "loss": 0.3307,
      "step": 12816
    },
    {
      "epoch": 0.5881786058464503,
      "grad_norm": 0.4261455237865448,
      "learning_rate": 9.22461929764155e-06,
      "loss": 0.332,
      "step": 12817
    },
    {
      "epoch": 0.5882244963517048,
      "grad_norm": 0.461000919342041,
      "learning_rate": 9.224488144836925e-06,
      "loss": 0.4454,
      "step": 12818
    },
    {
      "epoch": 0.5882703868569593,
      "grad_norm": 0.4471290707588196,
      "learning_rate": 9.22435698187374e-06,
      "loss": 0.3871,
      "step": 12819
    },
    {
      "epoch": 0.5883162773622138,
      "grad_norm": 0.4553200304508209,
      "learning_rate": 9.22422580875231e-06,
      "loss": 0.3958,
      "step": 12820
    },
    {
      "epoch": 0.5883621678674682,
      "grad_norm": 0.4374959170818329,
      "learning_rate": 9.224094625472947e-06,
      "loss": 0.3654,
      "step": 12821
    },
    {
      "epoch": 0.5884080583727227,
      "grad_norm": 0.4218861758708954,
      "learning_rate": 9.223963432035974e-06,
      "loss": 0.3473,
      "step": 12822
    },
    {
      "epoch": 0.5884539488779772,
      "grad_norm": 0.47813570499420166,
      "learning_rate": 9.2238322284417e-06,
      "loss": 0.4665,
      "step": 12823
    },
    {
      "epoch": 0.5884998393832316,
      "grad_norm": 0.4410776197910309,
      "learning_rate": 9.223701014690442e-06,
      "loss": 0.3872,
      "step": 12824
    },
    {
      "epoch": 0.5885457298884861,
      "grad_norm": 0.45331352949142456,
      "learning_rate": 9.223569790782517e-06,
      "loss": 0.3533,
      "step": 12825
    },
    {
      "epoch": 0.5885916203937406,
      "grad_norm": 0.43798357248306274,
      "learning_rate": 9.223438556718238e-06,
      "loss": 0.3463,
      "step": 12826
    },
    {
      "epoch": 0.588637510898995,
      "grad_norm": 0.48944082856178284,
      "learning_rate": 9.223307312497925e-06,
      "loss": 0.5037,
      "step": 12827
    },
    {
      "epoch": 0.5886834014042495,
      "grad_norm": 0.48846033215522766,
      "learning_rate": 9.223176058121889e-06,
      "loss": 0.3967,
      "step": 12828
    },
    {
      "epoch": 0.588729291909504,
      "grad_norm": 0.4521346390247345,
      "learning_rate": 9.223044793590447e-06,
      "loss": 0.3699,
      "step": 12829
    },
    {
      "epoch": 0.5887751824147583,
      "grad_norm": 0.5086361169815063,
      "learning_rate": 9.222913518903918e-06,
      "loss": 0.4536,
      "step": 12830
    },
    {
      "epoch": 0.5888210729200128,
      "grad_norm": 0.4563990533351898,
      "learning_rate": 9.222782234062613e-06,
      "loss": 0.4157,
      "step": 12831
    },
    {
      "epoch": 0.5888669634252673,
      "grad_norm": 0.45894595980644226,
      "learning_rate": 9.222650939066849e-06,
      "loss": 0.4276,
      "step": 12832
    },
    {
      "epoch": 0.5889128539305217,
      "grad_norm": 0.483471155166626,
      "learning_rate": 9.222519633916943e-06,
      "loss": 0.5066,
      "step": 12833
    },
    {
      "epoch": 0.5889587444357762,
      "grad_norm": 0.43864163756370544,
      "learning_rate": 9.222388318613207e-06,
      "loss": 0.3811,
      "step": 12834
    },
    {
      "epoch": 0.5890046349410307,
      "grad_norm": 0.4597693681716919,
      "learning_rate": 9.222256993155962e-06,
      "loss": 0.4092,
      "step": 12835
    },
    {
      "epoch": 0.5890505254462851,
      "grad_norm": 0.45927566289901733,
      "learning_rate": 9.222125657545522e-06,
      "loss": 0.4452,
      "step": 12836
    },
    {
      "epoch": 0.5890964159515396,
      "grad_norm": 0.46506625413894653,
      "learning_rate": 9.221994311782204e-06,
      "loss": 0.4494,
      "step": 12837
    },
    {
      "epoch": 0.5891423064567941,
      "grad_norm": 0.4535025954246521,
      "learning_rate": 9.221862955866318e-06,
      "loss": 0.4237,
      "step": 12838
    },
    {
      "epoch": 0.5891881969620486,
      "grad_norm": 0.4407668113708496,
      "learning_rate": 9.221731589798186e-06,
      "loss": 0.4133,
      "step": 12839
    },
    {
      "epoch": 0.589234087467303,
      "grad_norm": 0.4473594129085541,
      "learning_rate": 9.221600213578124e-06,
      "loss": 0.3746,
      "step": 12840
    },
    {
      "epoch": 0.5892799779725575,
      "grad_norm": 0.4277879595756531,
      "learning_rate": 9.221468827206444e-06,
      "loss": 0.3623,
      "step": 12841
    },
    {
      "epoch": 0.589325868477812,
      "grad_norm": 0.47602543234825134,
      "learning_rate": 9.221337430683464e-06,
      "loss": 0.3443,
      "step": 12842
    },
    {
      "epoch": 0.5893717589830664,
      "grad_norm": 0.4712906777858734,
      "learning_rate": 9.221206024009498e-06,
      "loss": 0.4503,
      "step": 12843
    },
    {
      "epoch": 0.5894176494883209,
      "grad_norm": 0.4947783648967743,
      "learning_rate": 9.221074607184866e-06,
      "loss": 0.5037,
      "step": 12844
    },
    {
      "epoch": 0.5894635399935754,
      "grad_norm": 0.46926939487457275,
      "learning_rate": 9.22094318020988e-06,
      "loss": 0.4021,
      "step": 12845
    },
    {
      "epoch": 0.5895094304988298,
      "grad_norm": 0.44295534491539,
      "learning_rate": 9.22081174308486e-06,
      "loss": 0.3268,
      "step": 12846
    },
    {
      "epoch": 0.5895553210040843,
      "grad_norm": 0.46860864758491516,
      "learning_rate": 9.220680295810117e-06,
      "loss": 0.4382,
      "step": 12847
    },
    {
      "epoch": 0.5896012115093388,
      "grad_norm": 0.47284558415412903,
      "learning_rate": 9.220548838385973e-06,
      "loss": 0.4296,
      "step": 12848
    },
    {
      "epoch": 0.5896471020145931,
      "grad_norm": 0.5287516117095947,
      "learning_rate": 9.220417370812739e-06,
      "loss": 0.5469,
      "step": 12849
    },
    {
      "epoch": 0.5896929925198476,
      "grad_norm": 0.4415300190448761,
      "learning_rate": 9.220285893090734e-06,
      "loss": 0.3924,
      "step": 12850
    },
    {
      "epoch": 0.5897388830251021,
      "grad_norm": 0.4936564564704895,
      "learning_rate": 9.220154405220273e-06,
      "loss": 0.445,
      "step": 12851
    },
    {
      "epoch": 0.5897847735303565,
      "grad_norm": 0.4838492274284363,
      "learning_rate": 9.220022907201675e-06,
      "loss": 0.4188,
      "step": 12852
    },
    {
      "epoch": 0.589830664035611,
      "grad_norm": 0.44980093836784363,
      "learning_rate": 9.219891399035251e-06,
      "loss": 0.3771,
      "step": 12853
    },
    {
      "epoch": 0.5898765545408655,
      "grad_norm": 0.44502314925193787,
      "learning_rate": 9.219759880721322e-06,
      "loss": 0.3532,
      "step": 12854
    },
    {
      "epoch": 0.5899224450461199,
      "grad_norm": 0.4306795001029968,
      "learning_rate": 9.2196283522602e-06,
      "loss": 0.3516,
      "step": 12855
    },
    {
      "epoch": 0.5899683355513744,
      "grad_norm": 0.46965932846069336,
      "learning_rate": 9.219496813652205e-06,
      "loss": 0.42,
      "step": 12856
    },
    {
      "epoch": 0.5900142260566289,
      "grad_norm": 0.44421058893203735,
      "learning_rate": 9.219365264897651e-06,
      "loss": 0.3903,
      "step": 12857
    },
    {
      "epoch": 0.5900601165618834,
      "grad_norm": 0.4487733542919159,
      "learning_rate": 9.219233705996857e-06,
      "loss": 0.4001,
      "step": 12858
    },
    {
      "epoch": 0.5901060070671378,
      "grad_norm": 0.5017364621162415,
      "learning_rate": 9.219102136950135e-06,
      "loss": 0.5189,
      "step": 12859
    },
    {
      "epoch": 0.5901518975723923,
      "grad_norm": 0.44568100571632385,
      "learning_rate": 9.218970557757806e-06,
      "loss": 0.368,
      "step": 12860
    },
    {
      "epoch": 0.5901977880776468,
      "grad_norm": 0.48367759585380554,
      "learning_rate": 9.218838968420185e-06,
      "loss": 0.4322,
      "step": 12861
    },
    {
      "epoch": 0.5902436785829012,
      "grad_norm": 0.4829626679420471,
      "learning_rate": 9.218707368937585e-06,
      "loss": 0.4512,
      "step": 12862
    },
    {
      "epoch": 0.5902895690881557,
      "grad_norm": 0.44088661670684814,
      "learning_rate": 9.218575759310326e-06,
      "loss": 0.3828,
      "step": 12863
    },
    {
      "epoch": 0.5903354595934102,
      "grad_norm": 0.45712119340896606,
      "learning_rate": 9.218444139538725e-06,
      "loss": 0.3479,
      "step": 12864
    },
    {
      "epoch": 0.5903813500986645,
      "grad_norm": 0.41344553232192993,
      "learning_rate": 9.218312509623098e-06,
      "loss": 0.3004,
      "step": 12865
    },
    {
      "epoch": 0.590427240603919,
      "grad_norm": 0.4537729024887085,
      "learning_rate": 9.218180869563759e-06,
      "loss": 0.39,
      "step": 12866
    },
    {
      "epoch": 0.5904731311091735,
      "grad_norm": 0.44736534357070923,
      "learning_rate": 9.218049219361026e-06,
      "loss": 0.3991,
      "step": 12867
    },
    {
      "epoch": 0.5905190216144279,
      "grad_norm": 0.4564002752304077,
      "learning_rate": 9.217917559015216e-06,
      "loss": 0.378,
      "step": 12868
    },
    {
      "epoch": 0.5905649121196824,
      "grad_norm": 0.41807860136032104,
      "learning_rate": 9.217785888526648e-06,
      "loss": 0.3215,
      "step": 12869
    },
    {
      "epoch": 0.5906108026249369,
      "grad_norm": 0.4513213634490967,
      "learning_rate": 9.217654207895632e-06,
      "loss": 0.367,
      "step": 12870
    },
    {
      "epoch": 0.5906566931301913,
      "grad_norm": 0.49027618765830994,
      "learning_rate": 9.217522517122492e-06,
      "loss": 0.4537,
      "step": 12871
    },
    {
      "epoch": 0.5907025836354458,
      "grad_norm": 0.49609375,
      "learning_rate": 9.21739081620754e-06,
      "loss": 0.4781,
      "step": 12872
    },
    {
      "epoch": 0.5907484741407003,
      "grad_norm": 0.5928043723106384,
      "learning_rate": 9.217259105151095e-06,
      "loss": 0.4749,
      "step": 12873
    },
    {
      "epoch": 0.5907943646459548,
      "grad_norm": 0.4627956748008728,
      "learning_rate": 9.217127383953473e-06,
      "loss": 0.4436,
      "step": 12874
    },
    {
      "epoch": 0.5908402551512092,
      "grad_norm": 0.4158020317554474,
      "learning_rate": 9.216995652614989e-06,
      "loss": 0.3003,
      "step": 12875
    },
    {
      "epoch": 0.5908861456564637,
      "grad_norm": 0.46536651253700256,
      "learning_rate": 9.216863911135962e-06,
      "loss": 0.3831,
      "step": 12876
    },
    {
      "epoch": 0.5909320361617182,
      "grad_norm": 0.4562129080295563,
      "learning_rate": 9.216732159516708e-06,
      "loss": 0.4188,
      "step": 12877
    },
    {
      "epoch": 0.5909779266669726,
      "grad_norm": 0.4690381586551666,
      "learning_rate": 9.216600397757545e-06,
      "loss": 0.3794,
      "step": 12878
    },
    {
      "epoch": 0.5910238171722271,
      "grad_norm": 0.43046078085899353,
      "learning_rate": 9.216468625858789e-06,
      "loss": 0.3889,
      "step": 12879
    },
    {
      "epoch": 0.5910697076774816,
      "grad_norm": 0.45390385389328003,
      "learning_rate": 9.216336843820755e-06,
      "loss": 0.3209,
      "step": 12880
    },
    {
      "epoch": 0.591115598182736,
      "grad_norm": 0.4318632483482361,
      "learning_rate": 9.216205051643763e-06,
      "loss": 0.3556,
      "step": 12881
    },
    {
      "epoch": 0.5911614886879905,
      "grad_norm": 0.4421613812446594,
      "learning_rate": 9.216073249328128e-06,
      "loss": 0.435,
      "step": 12882
    },
    {
      "epoch": 0.591207379193245,
      "grad_norm": 0.493385910987854,
      "learning_rate": 9.215941436874168e-06,
      "loss": 0.4748,
      "step": 12883
    },
    {
      "epoch": 0.5912532696984993,
      "grad_norm": 0.49507957696914673,
      "learning_rate": 9.215809614282199e-06,
      "loss": 0.4049,
      "step": 12884
    },
    {
      "epoch": 0.5912991602037538,
      "grad_norm": 0.48176833987236023,
      "learning_rate": 9.21567778155254e-06,
      "loss": 0.4062,
      "step": 12885
    },
    {
      "epoch": 0.5913450507090083,
      "grad_norm": 0.44845959544181824,
      "learning_rate": 9.215545938685503e-06,
      "loss": 0.3786,
      "step": 12886
    },
    {
      "epoch": 0.5913909412142627,
      "grad_norm": 0.44788259267807007,
      "learning_rate": 9.215414085681411e-06,
      "loss": 0.3686,
      "step": 12887
    },
    {
      "epoch": 0.5914368317195172,
      "grad_norm": 0.4293821156024933,
      "learning_rate": 9.215282222540579e-06,
      "loss": 0.3231,
      "step": 12888
    },
    {
      "epoch": 0.5914827222247717,
      "grad_norm": 0.46858271956443787,
      "learning_rate": 9.215150349263322e-06,
      "loss": 0.3739,
      "step": 12889
    },
    {
      "epoch": 0.5915286127300261,
      "grad_norm": 0.4479474723339081,
      "learning_rate": 9.215018465849959e-06,
      "loss": 0.3709,
      "step": 12890
    },
    {
      "epoch": 0.5915745032352806,
      "grad_norm": 0.4279611110687256,
      "learning_rate": 9.214886572300808e-06,
      "loss": 0.3779,
      "step": 12891
    },
    {
      "epoch": 0.5916203937405351,
      "grad_norm": 0.4540843367576599,
      "learning_rate": 9.214754668616183e-06,
      "loss": 0.3693,
      "step": 12892
    },
    {
      "epoch": 0.5916662842457896,
      "grad_norm": 0.493689626455307,
      "learning_rate": 9.214622754796406e-06,
      "loss": 0.4196,
      "step": 12893
    },
    {
      "epoch": 0.591712174751044,
      "grad_norm": 0.4432911276817322,
      "learning_rate": 9.21449083084179e-06,
      "loss": 0.3919,
      "step": 12894
    },
    {
      "epoch": 0.5917580652562985,
      "grad_norm": 0.43998655676841736,
      "learning_rate": 9.214358896752653e-06,
      "loss": 0.367,
      "step": 12895
    },
    {
      "epoch": 0.591803955761553,
      "grad_norm": 0.4773479104042053,
      "learning_rate": 9.214226952529312e-06,
      "loss": 0.3971,
      "step": 12896
    },
    {
      "epoch": 0.5918498462668074,
      "grad_norm": 0.46761736273765564,
      "learning_rate": 9.214094998172088e-06,
      "loss": 0.3769,
      "step": 12897
    },
    {
      "epoch": 0.5918957367720619,
      "grad_norm": 0.4792700409889221,
      "learning_rate": 9.213963033681295e-06,
      "loss": 0.4446,
      "step": 12898
    },
    {
      "epoch": 0.5919416272773164,
      "grad_norm": 0.4793221056461334,
      "learning_rate": 9.21383105905725e-06,
      "loss": 0.4612,
      "step": 12899
    },
    {
      "epoch": 0.5919875177825707,
      "grad_norm": 0.4984211027622223,
      "learning_rate": 9.21369907430027e-06,
      "loss": 0.4945,
      "step": 12900
    },
    {
      "epoch": 0.5920334082878252,
      "grad_norm": 0.49336037039756775,
      "learning_rate": 9.213567079410676e-06,
      "loss": 0.4446,
      "step": 12901
    },
    {
      "epoch": 0.5920792987930797,
      "grad_norm": 0.47008034586906433,
      "learning_rate": 9.213435074388781e-06,
      "loss": 0.3767,
      "step": 12902
    },
    {
      "epoch": 0.5921251892983341,
      "grad_norm": 0.4513151943683624,
      "learning_rate": 9.213303059234907e-06,
      "loss": 0.4123,
      "step": 12903
    },
    {
      "epoch": 0.5921710798035886,
      "grad_norm": 0.5015353560447693,
      "learning_rate": 9.213171033949366e-06,
      "loss": 0.5016,
      "step": 12904
    },
    {
      "epoch": 0.5922169703088431,
      "grad_norm": 0.4718548357486725,
      "learning_rate": 9.213038998532481e-06,
      "loss": 0.4332,
      "step": 12905
    },
    {
      "epoch": 0.5922628608140975,
      "grad_norm": 0.40910062193870544,
      "learning_rate": 9.212906952984566e-06,
      "loss": 0.305,
      "step": 12906
    },
    {
      "epoch": 0.592308751319352,
      "grad_norm": 0.4839059114456177,
      "learning_rate": 9.212774897305938e-06,
      "loss": 0.4001,
      "step": 12907
    },
    {
      "epoch": 0.5923546418246065,
      "grad_norm": 0.47646403312683105,
      "learning_rate": 9.212642831496917e-06,
      "loss": 0.4662,
      "step": 12908
    },
    {
      "epoch": 0.592400532329861,
      "grad_norm": 0.46075674891471863,
      "learning_rate": 9.21251075555782e-06,
      "loss": 0.3963,
      "step": 12909
    },
    {
      "epoch": 0.5924464228351154,
      "grad_norm": 0.4128977656364441,
      "learning_rate": 9.212378669488965e-06,
      "loss": 0.3409,
      "step": 12910
    },
    {
      "epoch": 0.5924923133403699,
      "grad_norm": 0.43849748373031616,
      "learning_rate": 9.212246573290666e-06,
      "loss": 0.3695,
      "step": 12911
    },
    {
      "epoch": 0.5925382038456244,
      "grad_norm": 0.45872604846954346,
      "learning_rate": 9.212114466963246e-06,
      "loss": 0.4347,
      "step": 12912
    },
    {
      "epoch": 0.5925840943508788,
      "grad_norm": 0.50279301404953,
      "learning_rate": 9.211982350507018e-06,
      "loss": 0.5026,
      "step": 12913
    },
    {
      "epoch": 0.5926299848561333,
      "grad_norm": 0.44524189829826355,
      "learning_rate": 9.211850223922303e-06,
      "loss": 0.3756,
      "step": 12914
    },
    {
      "epoch": 0.5926758753613878,
      "grad_norm": 0.6861488223075867,
      "learning_rate": 9.211718087209417e-06,
      "loss": 0.4129,
      "step": 12915
    },
    {
      "epoch": 0.5927217658666422,
      "grad_norm": 0.5497481226921082,
      "learning_rate": 9.211585940368679e-06,
      "loss": 0.5359,
      "step": 12916
    },
    {
      "epoch": 0.5927676563718967,
      "grad_norm": 0.44153815507888794,
      "learning_rate": 9.211453783400406e-06,
      "loss": 0.3926,
      "step": 12917
    },
    {
      "epoch": 0.5928135468771512,
      "grad_norm": 0.4850890636444092,
      "learning_rate": 9.211321616304916e-06,
      "loss": 0.4052,
      "step": 12918
    },
    {
      "epoch": 0.5928594373824055,
      "grad_norm": 0.4923182725906372,
      "learning_rate": 9.211189439082526e-06,
      "loss": 0.4226,
      "step": 12919
    },
    {
      "epoch": 0.59290532788766,
      "grad_norm": 0.4720557928085327,
      "learning_rate": 9.211057251733553e-06,
      "loss": 0.4075,
      "step": 12920
    },
    {
      "epoch": 0.5929512183929145,
      "grad_norm": 0.5026051998138428,
      "learning_rate": 9.210925054258319e-06,
      "loss": 0.4457,
      "step": 12921
    },
    {
      "epoch": 0.5929971088981689,
      "grad_norm": 0.4773178696632385,
      "learning_rate": 9.210792846657137e-06,
      "loss": 0.4231,
      "step": 12922
    },
    {
      "epoch": 0.5930429994034234,
      "grad_norm": 0.4374334514141083,
      "learning_rate": 9.210660628930329e-06,
      "loss": 0.4285,
      "step": 12923
    },
    {
      "epoch": 0.5930888899086779,
      "grad_norm": 0.4798688590526581,
      "learning_rate": 9.210528401078211e-06,
      "loss": 0.452,
      "step": 12924
    },
    {
      "epoch": 0.5931347804139323,
      "grad_norm": 0.46923092007637024,
      "learning_rate": 9.210396163101102e-06,
      "loss": 0.4267,
      "step": 12925
    },
    {
      "epoch": 0.5931806709191868,
      "grad_norm": 0.4547783434391022,
      "learning_rate": 9.210263914999316e-06,
      "loss": 0.4137,
      "step": 12926
    },
    {
      "epoch": 0.5932265614244413,
      "grad_norm": 0.38375768065452576,
      "learning_rate": 9.210131656773176e-06,
      "loss": 0.3198,
      "step": 12927
    },
    {
      "epoch": 0.5932724519296958,
      "grad_norm": 0.4026845395565033,
      "learning_rate": 9.209999388422996e-06,
      "loss": 0.2886,
      "step": 12928
    },
    {
      "epoch": 0.5933183424349502,
      "grad_norm": 0.43749377131462097,
      "learning_rate": 9.2098671099491e-06,
      "loss": 0.3612,
      "step": 12929
    },
    {
      "epoch": 0.5933642329402047,
      "grad_norm": 0.4625486433506012,
      "learning_rate": 9.209734821351798e-06,
      "loss": 0.4003,
      "step": 12930
    },
    {
      "epoch": 0.5934101234454592,
      "grad_norm": 0.4336709976196289,
      "learning_rate": 9.209602522631415e-06,
      "loss": 0.3352,
      "step": 12931
    },
    {
      "epoch": 0.5934560139507136,
      "grad_norm": 0.4388015866279602,
      "learning_rate": 9.209470213788268e-06,
      "loss": 0.3827,
      "step": 12932
    },
    {
      "epoch": 0.5935019044559681,
      "grad_norm": 0.4855652451515198,
      "learning_rate": 9.209337894822671e-06,
      "loss": 0.432,
      "step": 12933
    },
    {
      "epoch": 0.5935477949612226,
      "grad_norm": 0.4404178559780121,
      "learning_rate": 9.209205565734946e-06,
      "loss": 0.3702,
      "step": 12934
    },
    {
      "epoch": 0.593593685466477,
      "grad_norm": 0.4284880459308624,
      "learning_rate": 9.209073226525408e-06,
      "loss": 0.3864,
      "step": 12935
    },
    {
      "epoch": 0.5936395759717314,
      "grad_norm": 0.441139280796051,
      "learning_rate": 9.208940877194381e-06,
      "loss": 0.3281,
      "step": 12936
    },
    {
      "epoch": 0.593685466476986,
      "grad_norm": 0.4923996329307556,
      "learning_rate": 9.208808517742176e-06,
      "loss": 0.5051,
      "step": 12937
    },
    {
      "epoch": 0.5937313569822403,
      "grad_norm": 0.44874125719070435,
      "learning_rate": 9.208676148169118e-06,
      "loss": 0.4281,
      "step": 12938
    },
    {
      "epoch": 0.5937772474874948,
      "grad_norm": 0.4644026756286621,
      "learning_rate": 9.20854376847552e-06,
      "loss": 0.4334,
      "step": 12939
    },
    {
      "epoch": 0.5938231379927493,
      "grad_norm": 0.4614405035972595,
      "learning_rate": 9.208411378661703e-06,
      "loss": 0.4126,
      "step": 12940
    },
    {
      "epoch": 0.5938690284980037,
      "grad_norm": 0.44731923937797546,
      "learning_rate": 9.208278978727984e-06,
      "loss": 0.3638,
      "step": 12941
    },
    {
      "epoch": 0.5939149190032582,
      "grad_norm": 0.47226008772850037,
      "learning_rate": 9.208146568674685e-06,
      "loss": 0.386,
      "step": 12942
    },
    {
      "epoch": 0.5939608095085127,
      "grad_norm": 0.41131842136383057,
      "learning_rate": 9.20801414850212e-06,
      "loss": 0.3158,
      "step": 12943
    },
    {
      "epoch": 0.5940067000137671,
      "grad_norm": 0.45567071437835693,
      "learning_rate": 9.207881718210608e-06,
      "loss": 0.414,
      "step": 12944
    },
    {
      "epoch": 0.5940525905190216,
      "grad_norm": 0.43952903151512146,
      "learning_rate": 9.20774927780047e-06,
      "loss": 0.4043,
      "step": 12945
    },
    {
      "epoch": 0.5940984810242761,
      "grad_norm": 0.4675942063331604,
      "learning_rate": 9.207616827272023e-06,
      "loss": 0.4278,
      "step": 12946
    },
    {
      "epoch": 0.5941443715295306,
      "grad_norm": 0.4753773808479309,
      "learning_rate": 9.207484366625584e-06,
      "loss": 0.4173,
      "step": 12947
    },
    {
      "epoch": 0.594190262034785,
      "grad_norm": 0.4498985707759857,
      "learning_rate": 9.207351895861473e-06,
      "loss": 0.3729,
      "step": 12948
    },
    {
      "epoch": 0.5942361525400395,
      "grad_norm": 0.43134593963623047,
      "learning_rate": 9.20721941498001e-06,
      "loss": 0.3277,
      "step": 12949
    },
    {
      "epoch": 0.594282043045294,
      "grad_norm": 0.46087270975112915,
      "learning_rate": 9.207086923981512e-06,
      "loss": 0.4323,
      "step": 12950
    },
    {
      "epoch": 0.5943279335505484,
      "grad_norm": 0.4703740179538727,
      "learning_rate": 9.206954422866297e-06,
      "loss": 0.421,
      "step": 12951
    },
    {
      "epoch": 0.5943738240558029,
      "grad_norm": 0.4733833968639374,
      "learning_rate": 9.206821911634685e-06,
      "loss": 0.4598,
      "step": 12952
    },
    {
      "epoch": 0.5944197145610574,
      "grad_norm": 0.46428442001342773,
      "learning_rate": 9.206689390286994e-06,
      "loss": 0.429,
      "step": 12953
    },
    {
      "epoch": 0.5944656050663117,
      "grad_norm": 0.46772095561027527,
      "learning_rate": 9.206556858823543e-06,
      "loss": 0.4016,
      "step": 12954
    },
    {
      "epoch": 0.5945114955715662,
      "grad_norm": 0.4766250252723694,
      "learning_rate": 9.20642431724465e-06,
      "loss": 0.4514,
      "step": 12955
    },
    {
      "epoch": 0.5945573860768207,
      "grad_norm": 0.43740561604499817,
      "learning_rate": 9.206291765550633e-06,
      "loss": 0.331,
      "step": 12956
    },
    {
      "epoch": 0.5946032765820751,
      "grad_norm": 0.44999006390571594,
      "learning_rate": 9.206159203741813e-06,
      "loss": 0.3664,
      "step": 12957
    },
    {
      "epoch": 0.5946491670873296,
      "grad_norm": 0.42002201080322266,
      "learning_rate": 9.206026631818508e-06,
      "loss": 0.3327,
      "step": 12958
    },
    {
      "epoch": 0.5946950575925841,
      "grad_norm": 0.5026551485061646,
      "learning_rate": 9.205894049781034e-06,
      "loss": 0.4749,
      "step": 12959
    },
    {
      "epoch": 0.5947409480978385,
      "grad_norm": 0.500328540802002,
      "learning_rate": 9.205761457629717e-06,
      "loss": 0.4272,
      "step": 12960
    },
    {
      "epoch": 0.594786838603093,
      "grad_norm": 0.48113298416137695,
      "learning_rate": 9.205628855364867e-06,
      "loss": 0.4308,
      "step": 12961
    },
    {
      "epoch": 0.5948327291083475,
      "grad_norm": 0.4434349536895752,
      "learning_rate": 9.205496242986806e-06,
      "loss": 0.3412,
      "step": 12962
    },
    {
      "epoch": 0.594878619613602,
      "grad_norm": 0.47223326563835144,
      "learning_rate": 9.205363620495858e-06,
      "loss": 0.3953,
      "step": 12963
    },
    {
      "epoch": 0.5949245101188564,
      "grad_norm": 0.4478762447834015,
      "learning_rate": 9.205230987892336e-06,
      "loss": 0.3718,
      "step": 12964
    },
    {
      "epoch": 0.5949704006241109,
      "grad_norm": 0.49147820472717285,
      "learning_rate": 9.20509834517656e-06,
      "loss": 0.4259,
      "step": 12965
    },
    {
      "epoch": 0.5950162911293654,
      "grad_norm": 0.4482985734939575,
      "learning_rate": 9.20496569234885e-06,
      "loss": 0.4404,
      "step": 12966
    },
    {
      "epoch": 0.5950621816346198,
      "grad_norm": 0.4371623992919922,
      "learning_rate": 9.204833029409523e-06,
      "loss": 0.3938,
      "step": 12967
    },
    {
      "epoch": 0.5951080721398743,
      "grad_norm": 0.45835113525390625,
      "learning_rate": 9.204700356358902e-06,
      "loss": 0.4275,
      "step": 12968
    },
    {
      "epoch": 0.5951539626451288,
      "grad_norm": 0.43346869945526123,
      "learning_rate": 9.204567673197303e-06,
      "loss": 0.3334,
      "step": 12969
    },
    {
      "epoch": 0.5951998531503832,
      "grad_norm": 0.4674853980541229,
      "learning_rate": 9.204434979925045e-06,
      "loss": 0.4065,
      "step": 12970
    },
    {
      "epoch": 0.5952457436556376,
      "grad_norm": 0.46480056643486023,
      "learning_rate": 9.204302276542449e-06,
      "loss": 0.4211,
      "step": 12971
    },
    {
      "epoch": 0.5952916341608921,
      "grad_norm": 0.444549560546875,
      "learning_rate": 9.204169563049833e-06,
      "loss": 0.3695,
      "step": 12972
    },
    {
      "epoch": 0.5953375246661465,
      "grad_norm": 0.4327852427959442,
      "learning_rate": 9.204036839447514e-06,
      "loss": 0.3494,
      "step": 12973
    },
    {
      "epoch": 0.595383415171401,
      "grad_norm": 0.49107569456100464,
      "learning_rate": 9.203904105735817e-06,
      "loss": 0.4681,
      "step": 12974
    },
    {
      "epoch": 0.5954293056766555,
      "grad_norm": 0.49489814043045044,
      "learning_rate": 9.203771361915054e-06,
      "loss": 0.5006,
      "step": 12975
    },
    {
      "epoch": 0.5954751961819099,
      "grad_norm": 0.4802511930465698,
      "learning_rate": 9.203638607985548e-06,
      "loss": 0.4732,
      "step": 12976
    },
    {
      "epoch": 0.5955210866871644,
      "grad_norm": 0.4593989849090576,
      "learning_rate": 9.203505843947619e-06,
      "loss": 0.3759,
      "step": 12977
    },
    {
      "epoch": 0.5955669771924189,
      "grad_norm": 0.5229728817939758,
      "learning_rate": 9.203373069801585e-06,
      "loss": 0.501,
      "step": 12978
    },
    {
      "epoch": 0.5956128676976733,
      "grad_norm": 0.49846121668815613,
      "learning_rate": 9.203240285547766e-06,
      "loss": 0.4222,
      "step": 12979
    },
    {
      "epoch": 0.5956587582029278,
      "grad_norm": 0.4297657012939453,
      "learning_rate": 9.203107491186479e-06,
      "loss": 0.3556,
      "step": 12980
    },
    {
      "epoch": 0.5957046487081823,
      "grad_norm": 0.49471333622932434,
      "learning_rate": 9.202974686718047e-06,
      "loss": 0.462,
      "step": 12981
    },
    {
      "epoch": 0.5957505392134368,
      "grad_norm": 0.4635196328163147,
      "learning_rate": 9.202841872142787e-06,
      "loss": 0.4241,
      "step": 12982
    },
    {
      "epoch": 0.5957964297186912,
      "grad_norm": 0.44151097536087036,
      "learning_rate": 9.202709047461017e-06,
      "loss": 0.3949,
      "step": 12983
    },
    {
      "epoch": 0.5958423202239457,
      "grad_norm": 0.45605647563934326,
      "learning_rate": 9.202576212673058e-06,
      "loss": 0.4089,
      "step": 12984
    },
    {
      "epoch": 0.5958882107292002,
      "grad_norm": 0.5137408971786499,
      "learning_rate": 9.202443367779232e-06,
      "loss": 0.5196,
      "step": 12985
    },
    {
      "epoch": 0.5959341012344546,
      "grad_norm": 0.4613582193851471,
      "learning_rate": 9.202310512779854e-06,
      "loss": 0.444,
      "step": 12986
    },
    {
      "epoch": 0.5959799917397091,
      "grad_norm": 0.43255722522735596,
      "learning_rate": 9.202177647675246e-06,
      "loss": 0.308,
      "step": 12987
    },
    {
      "epoch": 0.5960258822449636,
      "grad_norm": 0.4602659344673157,
      "learning_rate": 9.202044772465728e-06,
      "loss": 0.4659,
      "step": 12988
    },
    {
      "epoch": 0.5960717727502179,
      "grad_norm": 0.4723518490791321,
      "learning_rate": 9.201911887151618e-06,
      "loss": 0.3968,
      "step": 12989
    },
    {
      "epoch": 0.5961176632554724,
      "grad_norm": 0.49993953108787537,
      "learning_rate": 9.201778991733236e-06,
      "loss": 0.4746,
      "step": 12990
    },
    {
      "epoch": 0.5961635537607269,
      "grad_norm": 0.4536040127277374,
      "learning_rate": 9.2016460862109e-06,
      "loss": 0.3766,
      "step": 12991
    },
    {
      "epoch": 0.5962094442659813,
      "grad_norm": 0.45119616389274597,
      "learning_rate": 9.201513170584933e-06,
      "loss": 0.4346,
      "step": 12992
    },
    {
      "epoch": 0.5962553347712358,
      "grad_norm": 0.4271732270717621,
      "learning_rate": 9.201380244855653e-06,
      "loss": 0.3435,
      "step": 12993
    },
    {
      "epoch": 0.5963012252764903,
      "grad_norm": 0.5502074956893921,
      "learning_rate": 9.20124730902338e-06,
      "loss": 0.41,
      "step": 12994
    },
    {
      "epoch": 0.5963471157817447,
      "grad_norm": 0.4600518047809601,
      "learning_rate": 9.20111436308843e-06,
      "loss": 0.4339,
      "step": 12995
    },
    {
      "epoch": 0.5963930062869992,
      "grad_norm": 0.5103567242622375,
      "learning_rate": 9.20098140705113e-06,
      "loss": 0.4785,
      "step": 12996
    },
    {
      "epoch": 0.5964388967922537,
      "grad_norm": 0.49002790451049805,
      "learning_rate": 9.200848440911794e-06,
      "loss": 0.4409,
      "step": 12997
    },
    {
      "epoch": 0.5964847872975082,
      "grad_norm": 0.4779488742351532,
      "learning_rate": 9.200715464670743e-06,
      "loss": 0.4432,
      "step": 12998
    },
    {
      "epoch": 0.5965306778027626,
      "grad_norm": 0.4579794108867645,
      "learning_rate": 9.200582478328298e-06,
      "loss": 0.4017,
      "step": 12999
    },
    {
      "epoch": 0.5965765683080171,
      "grad_norm": 0.4982486367225647,
      "learning_rate": 9.200449481884775e-06,
      "loss": 0.4772,
      "step": 13000
    },
    {
      "epoch": 0.5966224588132716,
      "grad_norm": 0.47153058648109436,
      "learning_rate": 9.2003164753405e-06,
      "loss": 0.4126,
      "step": 13001
    },
    {
      "epoch": 0.596668349318526,
      "grad_norm": 0.4579368531703949,
      "learning_rate": 9.200183458695789e-06,
      "loss": 0.3991,
      "step": 13002
    },
    {
      "epoch": 0.5967142398237805,
      "grad_norm": 0.43567442893981934,
      "learning_rate": 9.20005043195096e-06,
      "loss": 0.3833,
      "step": 13003
    },
    {
      "epoch": 0.596760130329035,
      "grad_norm": 0.47357070446014404,
      "learning_rate": 9.199917395106338e-06,
      "loss": 0.4708,
      "step": 13004
    },
    {
      "epoch": 0.5968060208342894,
      "grad_norm": 0.4473586082458496,
      "learning_rate": 9.199784348162238e-06,
      "loss": 0.3603,
      "step": 13005
    },
    {
      "epoch": 0.5968519113395439,
      "grad_norm": 0.46122750639915466,
      "learning_rate": 9.199651291118983e-06,
      "loss": 0.4032,
      "step": 13006
    },
    {
      "epoch": 0.5968978018447983,
      "grad_norm": 0.48835256695747375,
      "learning_rate": 9.199518223976894e-06,
      "loss": 0.3967,
      "step": 13007
    },
    {
      "epoch": 0.5969436923500527,
      "grad_norm": 0.45220568776130676,
      "learning_rate": 9.199385146736286e-06,
      "loss": 0.4204,
      "step": 13008
    },
    {
      "epoch": 0.5969895828553072,
      "grad_norm": 0.42608579993247986,
      "learning_rate": 9.199252059397483e-06,
      "loss": 0.3677,
      "step": 13009
    },
    {
      "epoch": 0.5970354733605617,
      "grad_norm": 0.4402492046356201,
      "learning_rate": 9.199118961960806e-06,
      "loss": 0.3741,
      "step": 13010
    },
    {
      "epoch": 0.5970813638658161,
      "grad_norm": 0.44287845492362976,
      "learning_rate": 9.198985854426573e-06,
      "loss": 0.3666,
      "step": 13011
    },
    {
      "epoch": 0.5971272543710706,
      "grad_norm": 0.46072304248809814,
      "learning_rate": 9.198852736795102e-06,
      "loss": 0.4031,
      "step": 13012
    },
    {
      "epoch": 0.5971731448763251,
      "grad_norm": 0.45226195454597473,
      "learning_rate": 9.198719609066716e-06,
      "loss": 0.4338,
      "step": 13013
    },
    {
      "epoch": 0.5972190353815795,
      "grad_norm": 0.4359162747859955,
      "learning_rate": 9.198586471241736e-06,
      "loss": 0.44,
      "step": 13014
    },
    {
      "epoch": 0.597264925886834,
      "grad_norm": 0.4613572955131531,
      "learning_rate": 9.198453323320479e-06,
      "loss": 0.4341,
      "step": 13015
    },
    {
      "epoch": 0.5973108163920885,
      "grad_norm": 0.4477507174015045,
      "learning_rate": 9.198320165303266e-06,
      "loss": 0.4075,
      "step": 13016
    },
    {
      "epoch": 0.597356706897343,
      "grad_norm": 0.46776801347732544,
      "learning_rate": 9.198186997190421e-06,
      "loss": 0.4212,
      "step": 13017
    },
    {
      "epoch": 0.5974025974025974,
      "grad_norm": 0.4757107198238373,
      "learning_rate": 9.19805381898226e-06,
      "loss": 0.5101,
      "step": 13018
    },
    {
      "epoch": 0.5974484879078519,
      "grad_norm": 0.40856650471687317,
      "learning_rate": 9.197920630679103e-06,
      "loss": 0.3366,
      "step": 13019
    },
    {
      "epoch": 0.5974943784131064,
      "grad_norm": 0.4342365264892578,
      "learning_rate": 9.197787432281274e-06,
      "loss": 0.3665,
      "step": 13020
    },
    {
      "epoch": 0.5975402689183608,
      "grad_norm": 0.43029963970184326,
      "learning_rate": 9.19765422378909e-06,
      "loss": 0.3734,
      "step": 13021
    },
    {
      "epoch": 0.5975861594236153,
      "grad_norm": 0.4541354775428772,
      "learning_rate": 9.197521005202871e-06,
      "loss": 0.4506,
      "step": 13022
    },
    {
      "epoch": 0.5976320499288698,
      "grad_norm": 0.426876962184906,
      "learning_rate": 9.197387776522942e-06,
      "loss": 0.3396,
      "step": 13023
    },
    {
      "epoch": 0.5976779404341241,
      "grad_norm": 0.44111695885658264,
      "learning_rate": 9.197254537749619e-06,
      "loss": 0.3989,
      "step": 13024
    },
    {
      "epoch": 0.5977238309393786,
      "grad_norm": 0.5065500736236572,
      "learning_rate": 9.197121288883222e-06,
      "loss": 0.4454,
      "step": 13025
    },
    {
      "epoch": 0.5977697214446331,
      "grad_norm": 0.44995230436325073,
      "learning_rate": 9.196988029924074e-06,
      "loss": 0.4107,
      "step": 13026
    },
    {
      "epoch": 0.5978156119498875,
      "grad_norm": 0.5156726241111755,
      "learning_rate": 9.196854760872494e-06,
      "loss": 0.4785,
      "step": 13027
    },
    {
      "epoch": 0.597861502455142,
      "grad_norm": 0.5044330358505249,
      "learning_rate": 9.196721481728803e-06,
      "loss": 0.477,
      "step": 13028
    },
    {
      "epoch": 0.5979073929603965,
      "grad_norm": 0.4946177005767822,
      "learning_rate": 9.196588192493321e-06,
      "loss": 0.4441,
      "step": 13029
    },
    {
      "epoch": 0.5979532834656509,
      "grad_norm": 0.5054530501365662,
      "learning_rate": 9.19645489316637e-06,
      "loss": 0.5216,
      "step": 13030
    },
    {
      "epoch": 0.5979991739709054,
      "grad_norm": 0.4467655122280121,
      "learning_rate": 9.19632158374827e-06,
      "loss": 0.3715,
      "step": 13031
    },
    {
      "epoch": 0.5980450644761599,
      "grad_norm": 0.44318538904190063,
      "learning_rate": 9.196188264239338e-06,
      "loss": 0.324,
      "step": 13032
    },
    {
      "epoch": 0.5980909549814143,
      "grad_norm": 0.48898983001708984,
      "learning_rate": 9.1960549346399e-06,
      "loss": 0.395,
      "step": 13033
    },
    {
      "epoch": 0.5981368454866688,
      "grad_norm": 0.45287731289863586,
      "learning_rate": 9.195921594950274e-06,
      "loss": 0.3545,
      "step": 13034
    },
    {
      "epoch": 0.5981827359919233,
      "grad_norm": 0.460599809885025,
      "learning_rate": 9.195788245170781e-06,
      "loss": 0.4213,
      "step": 13035
    },
    {
      "epoch": 0.5982286264971778,
      "grad_norm": 0.5110242962837219,
      "learning_rate": 9.19565488530174e-06,
      "loss": 0.4765,
      "step": 13036
    },
    {
      "epoch": 0.5982745170024322,
      "grad_norm": 0.4277670085430145,
      "learning_rate": 9.195521515343472e-06,
      "loss": 0.3109,
      "step": 13037
    },
    {
      "epoch": 0.5983204075076867,
      "grad_norm": 0.46109262108802795,
      "learning_rate": 9.195388135296302e-06,
      "loss": 0.3857,
      "step": 13038
    },
    {
      "epoch": 0.5983662980129412,
      "grad_norm": 0.768392026424408,
      "learning_rate": 9.195254745160546e-06,
      "loss": 0.4387,
      "step": 13039
    },
    {
      "epoch": 0.5984121885181956,
      "grad_norm": 0.5008258819580078,
      "learning_rate": 9.195121344936528e-06,
      "loss": 0.5084,
      "step": 13040
    },
    {
      "epoch": 0.59845807902345,
      "grad_norm": 0.45953384041786194,
      "learning_rate": 9.194987934624567e-06,
      "loss": 0.418,
      "step": 13041
    },
    {
      "epoch": 0.5985039695287045,
      "grad_norm": 0.477068156003952,
      "learning_rate": 9.194854514224983e-06,
      "loss": 0.3709,
      "step": 13042
    },
    {
      "epoch": 0.5985498600339589,
      "grad_norm": 0.5379118323326111,
      "learning_rate": 9.194721083738097e-06,
      "loss": 0.3925,
      "step": 13043
    },
    {
      "epoch": 0.5985957505392134,
      "grad_norm": 0.45522406697273254,
      "learning_rate": 9.194587643164232e-06,
      "loss": 0.3671,
      "step": 13044
    },
    {
      "epoch": 0.5986416410444679,
      "grad_norm": 0.46020111441612244,
      "learning_rate": 9.194454192503708e-06,
      "loss": 0.433,
      "step": 13045
    },
    {
      "epoch": 0.5986875315497223,
      "grad_norm": 0.4651593863964081,
      "learning_rate": 9.194320731756844e-06,
      "loss": 0.4034,
      "step": 13046
    },
    {
      "epoch": 0.5987334220549768,
      "grad_norm": 0.532458484172821,
      "learning_rate": 9.194187260923963e-06,
      "loss": 0.4954,
      "step": 13047
    },
    {
      "epoch": 0.5987793125602313,
      "grad_norm": 0.4655955135822296,
      "learning_rate": 9.194053780005386e-06,
      "loss": 0.3916,
      "step": 13048
    },
    {
      "epoch": 0.5988252030654857,
      "grad_norm": 0.46037203073501587,
      "learning_rate": 9.193920289001433e-06,
      "loss": 0.3749,
      "step": 13049
    },
    {
      "epoch": 0.5988710935707402,
      "grad_norm": 0.42698484659194946,
      "learning_rate": 9.193786787912426e-06,
      "loss": 0.3004,
      "step": 13050
    },
    {
      "epoch": 0.5989169840759947,
      "grad_norm": 0.479028582572937,
      "learning_rate": 9.193653276738684e-06,
      "loss": 0.4179,
      "step": 13051
    },
    {
      "epoch": 0.5989628745812492,
      "grad_norm": 0.42838889360427856,
      "learning_rate": 9.193519755480528e-06,
      "loss": 0.3277,
      "step": 13052
    },
    {
      "epoch": 0.5990087650865036,
      "grad_norm": 0.4751914143562317,
      "learning_rate": 9.193386224138285e-06,
      "loss": 0.4125,
      "step": 13053
    },
    {
      "epoch": 0.5990546555917581,
      "grad_norm": 0.4815027117729187,
      "learning_rate": 9.193252682712268e-06,
      "loss": 0.4112,
      "step": 13054
    },
    {
      "epoch": 0.5991005460970126,
      "grad_norm": 0.458638072013855,
      "learning_rate": 9.193119131202804e-06,
      "loss": 0.3683,
      "step": 13055
    },
    {
      "epoch": 0.599146436602267,
      "grad_norm": 0.4553982615470886,
      "learning_rate": 9.19298556961021e-06,
      "loss": 0.3615,
      "step": 13056
    },
    {
      "epoch": 0.5991923271075215,
      "grad_norm": 0.48458924889564514,
      "learning_rate": 9.19285199793481e-06,
      "loss": 0.4219,
      "step": 13057
    },
    {
      "epoch": 0.599238217612776,
      "grad_norm": 0.4356871545314789,
      "learning_rate": 9.192718416176923e-06,
      "loss": 0.3365,
      "step": 13058
    },
    {
      "epoch": 0.5992841081180303,
      "grad_norm": 0.4829314351081848,
      "learning_rate": 9.192584824336873e-06,
      "loss": 0.4822,
      "step": 13059
    },
    {
      "epoch": 0.5993299986232848,
      "grad_norm": 0.44935545325279236,
      "learning_rate": 9.19245122241498e-06,
      "loss": 0.4036,
      "step": 13060
    },
    {
      "epoch": 0.5993758891285393,
      "grad_norm": 0.4538896083831787,
      "learning_rate": 9.192317610411563e-06,
      "loss": 0.4322,
      "step": 13061
    },
    {
      "epoch": 0.5994217796337937,
      "grad_norm": 0.4422077536582947,
      "learning_rate": 9.192183988326945e-06,
      "loss": 0.3865,
      "step": 13062
    },
    {
      "epoch": 0.5994676701390482,
      "grad_norm": 0.4704171121120453,
      "learning_rate": 9.19205035616145e-06,
      "loss": 0.4327,
      "step": 13063
    },
    {
      "epoch": 0.5995135606443027,
      "grad_norm": 8.319628715515137,
      "learning_rate": 9.191916713915397e-06,
      "loss": 0.3921,
      "step": 13064
    },
    {
      "epoch": 0.5995594511495571,
      "grad_norm": 0.4551405906677246,
      "learning_rate": 9.191783061589106e-06,
      "loss": 0.3639,
      "step": 13065
    },
    {
      "epoch": 0.5996053416548116,
      "grad_norm": 0.45395177602767944,
      "learning_rate": 9.1916493991829e-06,
      "loss": 0.3551,
      "step": 13066
    },
    {
      "epoch": 0.5996512321600661,
      "grad_norm": 0.5292371511459351,
      "learning_rate": 9.1915157266971e-06,
      "loss": 0.3898,
      "step": 13067
    },
    {
      "epoch": 0.5996971226653205,
      "grad_norm": 0.49894002079963684,
      "learning_rate": 9.191382044132026e-06,
      "loss": 0.4119,
      "step": 13068
    },
    {
      "epoch": 0.599743013170575,
      "grad_norm": 0.44736912846565247,
      "learning_rate": 9.191248351488004e-06,
      "loss": 0.3723,
      "step": 13069
    },
    {
      "epoch": 0.5997889036758295,
      "grad_norm": 0.4497484862804413,
      "learning_rate": 9.19111464876535e-06,
      "loss": 0.3748,
      "step": 13070
    },
    {
      "epoch": 0.599834794181084,
      "grad_norm": 0.4900120794773102,
      "learning_rate": 9.19098093596439e-06,
      "loss": 0.3982,
      "step": 13071
    },
    {
      "epoch": 0.5998806846863384,
      "grad_norm": 0.44614365696907043,
      "learning_rate": 9.190847213085443e-06,
      "loss": 0.3762,
      "step": 13072
    },
    {
      "epoch": 0.5999265751915929,
      "grad_norm": 0.48492252826690674,
      "learning_rate": 9.190713480128833e-06,
      "loss": 0.3949,
      "step": 13073
    },
    {
      "epoch": 0.5999724656968474,
      "grad_norm": 0.42985138297080994,
      "learning_rate": 9.190579737094876e-06,
      "loss": 0.3741,
      "step": 13074
    },
    {
      "epoch": 0.6000183562021018,
      "grad_norm": 0.5253840088844299,
      "learning_rate": 9.1904459839839e-06,
      "loss": 0.5328,
      "step": 13075
    },
    {
      "epoch": 0.6000642467073563,
      "grad_norm": 0.5018855929374695,
      "learning_rate": 9.190312220796223e-06,
      "loss": 0.4682,
      "step": 13076
    },
    {
      "epoch": 0.6001101372126108,
      "grad_norm": 0.4584819972515106,
      "learning_rate": 9.190178447532167e-06,
      "loss": 0.3859,
      "step": 13077
    },
    {
      "epoch": 0.6001560277178651,
      "grad_norm": 0.4401921331882477,
      "learning_rate": 9.190044664192055e-06,
      "loss": 0.3887,
      "step": 13078
    },
    {
      "epoch": 0.6002019182231196,
      "grad_norm": 0.44682690501213074,
      "learning_rate": 9.189910870776206e-06,
      "loss": 0.3979,
      "step": 13079
    },
    {
      "epoch": 0.6002478087283741,
      "grad_norm": 0.46294525265693665,
      "learning_rate": 9.189777067284947e-06,
      "loss": 0.4099,
      "step": 13080
    },
    {
      "epoch": 0.6002936992336285,
      "grad_norm": 0.46135830879211426,
      "learning_rate": 9.189643253718594e-06,
      "loss": 0.4875,
      "step": 13081
    },
    {
      "epoch": 0.600339589738883,
      "grad_norm": 0.45244288444519043,
      "learning_rate": 9.189509430077473e-06,
      "loss": 0.4327,
      "step": 13082
    },
    {
      "epoch": 0.6003854802441375,
      "grad_norm": 0.4156610071659088,
      "learning_rate": 9.189375596361902e-06,
      "loss": 0.3175,
      "step": 13083
    },
    {
      "epoch": 0.6004313707493919,
      "grad_norm": 0.44481128454208374,
      "learning_rate": 9.189241752572206e-06,
      "loss": 0.3866,
      "step": 13084
    },
    {
      "epoch": 0.6004772612546464,
      "grad_norm": 0.516258716583252,
      "learning_rate": 9.189107898708707e-06,
      "loss": 0.3719,
      "step": 13085
    },
    {
      "epoch": 0.6005231517599009,
      "grad_norm": 0.49316585063934326,
      "learning_rate": 9.188974034771724e-06,
      "loss": 0.4779,
      "step": 13086
    },
    {
      "epoch": 0.6005690422651553,
      "grad_norm": 0.4243500530719757,
      "learning_rate": 9.188840160761583e-06,
      "loss": 0.4062,
      "step": 13087
    },
    {
      "epoch": 0.6006149327704098,
      "grad_norm": 0.46520867943763733,
      "learning_rate": 9.188706276678601e-06,
      "loss": 0.3863,
      "step": 13088
    },
    {
      "epoch": 0.6006608232756643,
      "grad_norm": 0.4622286260128021,
      "learning_rate": 9.188572382523102e-06,
      "loss": 0.405,
      "step": 13089
    },
    {
      "epoch": 0.6007067137809188,
      "grad_norm": 0.5205509662628174,
      "learning_rate": 9.18843847829541e-06,
      "loss": 0.4794,
      "step": 13090
    },
    {
      "epoch": 0.6007526042861732,
      "grad_norm": 0.4583280086517334,
      "learning_rate": 9.188304563995845e-06,
      "loss": 0.433,
      "step": 13091
    },
    {
      "epoch": 0.6007984947914277,
      "grad_norm": 0.43881797790527344,
      "learning_rate": 9.188170639624729e-06,
      "loss": 0.3604,
      "step": 13092
    },
    {
      "epoch": 0.6008443852966822,
      "grad_norm": 0.45393168926239014,
      "learning_rate": 9.188036705182384e-06,
      "loss": 0.382,
      "step": 13093
    },
    {
      "epoch": 0.6008902758019365,
      "grad_norm": 0.4754018485546112,
      "learning_rate": 9.187902760669134e-06,
      "loss": 0.4136,
      "step": 13094
    },
    {
      "epoch": 0.600936166307191,
      "grad_norm": 0.4563855826854706,
      "learning_rate": 9.187768806085297e-06,
      "loss": 0.4001,
      "step": 13095
    },
    {
      "epoch": 0.6009820568124455,
      "grad_norm": 0.5106279850006104,
      "learning_rate": 9.187634841431201e-06,
      "loss": 0.4664,
      "step": 13096
    },
    {
      "epoch": 0.6010279473176999,
      "grad_norm": 0.4618285596370697,
      "learning_rate": 9.187500866707162e-06,
      "loss": 0.4364,
      "step": 13097
    },
    {
      "epoch": 0.6010738378229544,
      "grad_norm": 0.4490296244621277,
      "learning_rate": 9.187366881913506e-06,
      "loss": 0.3727,
      "step": 13098
    },
    {
      "epoch": 0.6011197283282089,
      "grad_norm": 0.4416576027870178,
      "learning_rate": 9.187232887050556e-06,
      "loss": 0.371,
      "step": 13099
    },
    {
      "epoch": 0.6011656188334633,
      "grad_norm": 0.4476524889469147,
      "learning_rate": 9.18709888211863e-06,
      "loss": 0.4184,
      "step": 13100
    },
    {
      "epoch": 0.6012115093387178,
      "grad_norm": 0.4171536862850189,
      "learning_rate": 9.186964867118054e-06,
      "loss": 0.3235,
      "step": 13101
    },
    {
      "epoch": 0.6012573998439723,
      "grad_norm": 0.45746901631355286,
      "learning_rate": 9.186830842049148e-06,
      "loss": 0.3861,
      "step": 13102
    },
    {
      "epoch": 0.6013032903492267,
      "grad_norm": 0.46083250641822815,
      "learning_rate": 9.186696806912236e-06,
      "loss": 0.3978,
      "step": 13103
    },
    {
      "epoch": 0.6013491808544812,
      "grad_norm": 0.5189059972763062,
      "learning_rate": 9.18656276170764e-06,
      "loss": 0.47,
      "step": 13104
    },
    {
      "epoch": 0.6013950713597357,
      "grad_norm": 0.44035518169403076,
      "learning_rate": 9.186428706435682e-06,
      "loss": 0.364,
      "step": 13105
    },
    {
      "epoch": 0.6014409618649902,
      "grad_norm": 0.43447956442832947,
      "learning_rate": 9.186294641096682e-06,
      "loss": 0.4012,
      "step": 13106
    },
    {
      "epoch": 0.6014868523702446,
      "grad_norm": 0.47762805223464966,
      "learning_rate": 9.186160565690967e-06,
      "loss": 0.4402,
      "step": 13107
    },
    {
      "epoch": 0.6015327428754991,
      "grad_norm": 0.4448474645614624,
      "learning_rate": 9.186026480218854e-06,
      "loss": 0.4035,
      "step": 13108
    },
    {
      "epoch": 0.6015786333807536,
      "grad_norm": 0.46709585189819336,
      "learning_rate": 9.185892384680671e-06,
      "loss": 0.4616,
      "step": 13109
    },
    {
      "epoch": 0.601624523886008,
      "grad_norm": 0.3941563367843628,
      "learning_rate": 9.185758279076737e-06,
      "loss": 0.3021,
      "step": 13110
    },
    {
      "epoch": 0.6016704143912625,
      "grad_norm": 0.48306068778038025,
      "learning_rate": 9.185624163407377e-06,
      "loss": 0.4861,
      "step": 13111
    },
    {
      "epoch": 0.601716304896517,
      "grad_norm": 0.42975151538848877,
      "learning_rate": 9.18549003767291e-06,
      "loss": 0.3632,
      "step": 13112
    },
    {
      "epoch": 0.6017621954017713,
      "grad_norm": 0.44125816226005554,
      "learning_rate": 9.185355901873662e-06,
      "loss": 0.3278,
      "step": 13113
    },
    {
      "epoch": 0.6018080859070258,
      "grad_norm": 0.5092422962188721,
      "learning_rate": 9.185221756009954e-06,
      "loss": 0.527,
      "step": 13114
    },
    {
      "epoch": 0.6018539764122803,
      "grad_norm": 0.4492291808128357,
      "learning_rate": 9.185087600082107e-06,
      "loss": 0.3573,
      "step": 13115
    },
    {
      "epoch": 0.6018998669175347,
      "grad_norm": 0.4542471766471863,
      "learning_rate": 9.184953434090446e-06,
      "loss": 0.3562,
      "step": 13116
    },
    {
      "epoch": 0.6019457574227892,
      "grad_norm": 0.4844908118247986,
      "learning_rate": 9.184819258035293e-06,
      "loss": 0.4132,
      "step": 13117
    },
    {
      "epoch": 0.6019916479280437,
      "grad_norm": 0.44149231910705566,
      "learning_rate": 9.18468507191697e-06,
      "loss": 0.3448,
      "step": 13118
    },
    {
      "epoch": 0.6020375384332981,
      "grad_norm": 0.46837496757507324,
      "learning_rate": 9.1845508757358e-06,
      "loss": 0.4653,
      "step": 13119
    },
    {
      "epoch": 0.6020834289385526,
      "grad_norm": 0.4606911838054657,
      "learning_rate": 9.184416669492107e-06,
      "loss": 0.3354,
      "step": 13120
    },
    {
      "epoch": 0.6021293194438071,
      "grad_norm": 0.43521472811698914,
      "learning_rate": 9.18428245318621e-06,
      "loss": 0.3792,
      "step": 13121
    },
    {
      "epoch": 0.6021752099490615,
      "grad_norm": 0.4705839157104492,
      "learning_rate": 9.184148226818437e-06,
      "loss": 0.4403,
      "step": 13122
    },
    {
      "epoch": 0.602221100454316,
      "grad_norm": 0.4833974838256836,
      "learning_rate": 9.184013990389107e-06,
      "loss": 0.3946,
      "step": 13123
    },
    {
      "epoch": 0.6022669909595705,
      "grad_norm": 0.45704784989356995,
      "learning_rate": 9.183879743898544e-06,
      "loss": 0.411,
      "step": 13124
    },
    {
      "epoch": 0.602312881464825,
      "grad_norm": 0.4900606870651245,
      "learning_rate": 9.18374548734707e-06,
      "loss": 0.4437,
      "step": 13125
    },
    {
      "epoch": 0.6023587719700794,
      "grad_norm": 0.3757486045360565,
      "learning_rate": 9.183611220735009e-06,
      "loss": 0.2631,
      "step": 13126
    },
    {
      "epoch": 0.6024046624753339,
      "grad_norm": 0.4267931878566742,
      "learning_rate": 9.183476944062684e-06,
      "loss": 0.3302,
      "step": 13127
    },
    {
      "epoch": 0.6024505529805884,
      "grad_norm": 0.46986663341522217,
      "learning_rate": 9.183342657330416e-06,
      "loss": 0.471,
      "step": 13128
    },
    {
      "epoch": 0.6024964434858427,
      "grad_norm": 0.4305763244628906,
      "learning_rate": 9.183208360538531e-06,
      "loss": 0.3432,
      "step": 13129
    },
    {
      "epoch": 0.6025423339910972,
      "grad_norm": 0.9164122939109802,
      "learning_rate": 9.183074053687348e-06,
      "loss": 0.441,
      "step": 13130
    },
    {
      "epoch": 0.6025882244963517,
      "grad_norm": 0.4516807794570923,
      "learning_rate": 9.182939736777194e-06,
      "loss": 0.3938,
      "step": 13131
    },
    {
      "epoch": 0.6026341150016061,
      "grad_norm": 0.45371124148368835,
      "learning_rate": 9.18280540980839e-06,
      "loss": 0.4188,
      "step": 13132
    },
    {
      "epoch": 0.6026800055068606,
      "grad_norm": 0.4521670341491699,
      "learning_rate": 9.182671072781257e-06,
      "loss": 0.4084,
      "step": 13133
    },
    {
      "epoch": 0.6027258960121151,
      "grad_norm": 0.4752746820449829,
      "learning_rate": 9.182536725696122e-06,
      "loss": 0.4118,
      "step": 13134
    },
    {
      "epoch": 0.6027717865173695,
      "grad_norm": 0.4611586332321167,
      "learning_rate": 9.182402368553308e-06,
      "loss": 0.4334,
      "step": 13135
    },
    {
      "epoch": 0.602817677022624,
      "grad_norm": 0.49664604663848877,
      "learning_rate": 9.182268001353133e-06,
      "loss": 0.5023,
      "step": 13136
    },
    {
      "epoch": 0.6028635675278785,
      "grad_norm": 0.5093265771865845,
      "learning_rate": 9.182133624095925e-06,
      "loss": 0.4809,
      "step": 13137
    },
    {
      "epoch": 0.6029094580331329,
      "grad_norm": 0.4916189908981323,
      "learning_rate": 9.181999236782004e-06,
      "loss": 0.4513,
      "step": 13138
    },
    {
      "epoch": 0.6029553485383874,
      "grad_norm": 0.4603444039821625,
      "learning_rate": 9.181864839411696e-06,
      "loss": 0.4259,
      "step": 13139
    },
    {
      "epoch": 0.6030012390436419,
      "grad_norm": 0.4792085587978363,
      "learning_rate": 9.181730431985323e-06,
      "loss": 0.4014,
      "step": 13140
    },
    {
      "epoch": 0.6030471295488964,
      "grad_norm": 0.47428810596466064,
      "learning_rate": 9.18159601450321e-06,
      "loss": 0.3854,
      "step": 13141
    },
    {
      "epoch": 0.6030930200541508,
      "grad_norm": 0.4688205420970917,
      "learning_rate": 9.181461586965674e-06,
      "loss": 0.3986,
      "step": 13142
    },
    {
      "epoch": 0.6031389105594053,
      "grad_norm": 0.4682396352291107,
      "learning_rate": 9.181327149373046e-06,
      "loss": 0.3962,
      "step": 13143
    },
    {
      "epoch": 0.6031848010646598,
      "grad_norm": 0.46324437856674194,
      "learning_rate": 9.181192701725644e-06,
      "loss": 0.3916,
      "step": 13144
    },
    {
      "epoch": 0.6032306915699142,
      "grad_norm": 0.45583561062812805,
      "learning_rate": 9.181058244023795e-06,
      "loss": 0.3465,
      "step": 13145
    },
    {
      "epoch": 0.6032765820751687,
      "grad_norm": 0.4615073502063751,
      "learning_rate": 9.180923776267819e-06,
      "loss": 0.3766,
      "step": 13146
    },
    {
      "epoch": 0.6033224725804232,
      "grad_norm": 0.422166645526886,
      "learning_rate": 9.180789298458041e-06,
      "loss": 0.3244,
      "step": 13147
    },
    {
      "epoch": 0.6033683630856775,
      "grad_norm": 0.49139878153800964,
      "learning_rate": 9.180654810594785e-06,
      "loss": 0.3963,
      "step": 13148
    },
    {
      "epoch": 0.603414253590932,
      "grad_norm": 0.47618353366851807,
      "learning_rate": 9.180520312678373e-06,
      "loss": 0.4462,
      "step": 13149
    },
    {
      "epoch": 0.6034601440961865,
      "grad_norm": 0.4901646375656128,
      "learning_rate": 9.180385804709128e-06,
      "loss": 0.4548,
      "step": 13150
    },
    {
      "epoch": 0.6035060346014409,
      "grad_norm": 0.448930025100708,
      "learning_rate": 9.180251286687376e-06,
      "loss": 0.3536,
      "step": 13151
    },
    {
      "epoch": 0.6035519251066954,
      "grad_norm": 0.48926153779029846,
      "learning_rate": 9.180116758613439e-06,
      "loss": 0.3892,
      "step": 13152
    },
    {
      "epoch": 0.6035978156119499,
      "grad_norm": 0.4527420103549957,
      "learning_rate": 9.179982220487639e-06,
      "loss": 0.4013,
      "step": 13153
    },
    {
      "epoch": 0.6036437061172043,
      "grad_norm": 0.4529235064983368,
      "learning_rate": 9.179847672310303e-06,
      "loss": 0.3397,
      "step": 13154
    },
    {
      "epoch": 0.6036895966224588,
      "grad_norm": 0.49350103735923767,
      "learning_rate": 9.179713114081752e-06,
      "loss": 0.4067,
      "step": 13155
    },
    {
      "epoch": 0.6037354871277133,
      "grad_norm": 0.44399493932724,
      "learning_rate": 9.17957854580231e-06,
      "loss": 0.3685,
      "step": 13156
    },
    {
      "epoch": 0.6037813776329677,
      "grad_norm": 0.8354093432426453,
      "learning_rate": 9.1794439674723e-06,
      "loss": 0.4493,
      "step": 13157
    },
    {
      "epoch": 0.6038272681382222,
      "grad_norm": 0.501746416091919,
      "learning_rate": 9.179309379092047e-06,
      "loss": 0.4754,
      "step": 13158
    },
    {
      "epoch": 0.6038731586434767,
      "grad_norm": 0.443144291639328,
      "learning_rate": 9.179174780661874e-06,
      "loss": 0.3863,
      "step": 13159
    },
    {
      "epoch": 0.6039190491487312,
      "grad_norm": 0.5173042416572571,
      "learning_rate": 9.179040172182104e-06,
      "loss": 0.4403,
      "step": 13160
    },
    {
      "epoch": 0.6039649396539856,
      "grad_norm": 0.455334335565567,
      "learning_rate": 9.178905553653062e-06,
      "loss": 0.3632,
      "step": 13161
    },
    {
      "epoch": 0.6040108301592401,
      "grad_norm": 0.4414371848106384,
      "learning_rate": 9.17877092507507e-06,
      "loss": 0.344,
      "step": 13162
    },
    {
      "epoch": 0.6040567206644946,
      "grad_norm": 0.4432876706123352,
      "learning_rate": 9.178636286448454e-06,
      "loss": 0.3586,
      "step": 13163
    },
    {
      "epoch": 0.604102611169749,
      "grad_norm": 0.4676642119884491,
      "learning_rate": 9.178501637773535e-06,
      "loss": 0.4242,
      "step": 13164
    },
    {
      "epoch": 0.6041485016750034,
      "grad_norm": 0.4625203311443329,
      "learning_rate": 9.17836697905064e-06,
      "loss": 0.4118,
      "step": 13165
    },
    {
      "epoch": 0.6041943921802579,
      "grad_norm": 0.430313378572464,
      "learning_rate": 9.17823231028009e-06,
      "loss": 0.3686,
      "step": 13166
    },
    {
      "epoch": 0.6042402826855123,
      "grad_norm": 0.46006646752357483,
      "learning_rate": 9.178097631462213e-06,
      "loss": 0.3957,
      "step": 13167
    },
    {
      "epoch": 0.6042861731907668,
      "grad_norm": 0.4284728765487671,
      "learning_rate": 9.177962942597327e-06,
      "loss": 0.3454,
      "step": 13168
    },
    {
      "epoch": 0.6043320636960213,
      "grad_norm": 0.4706713855266571,
      "learning_rate": 9.177828243685758e-06,
      "loss": 0.4013,
      "step": 13169
    },
    {
      "epoch": 0.6043779542012757,
      "grad_norm": 0.4403558671474457,
      "learning_rate": 9.177693534727832e-06,
      "loss": 0.3451,
      "step": 13170
    },
    {
      "epoch": 0.6044238447065302,
      "grad_norm": 0.48263901472091675,
      "learning_rate": 9.177558815723872e-06,
      "loss": 0.47,
      "step": 13171
    },
    {
      "epoch": 0.6044697352117847,
      "grad_norm": 0.45973673462867737,
      "learning_rate": 9.1774240866742e-06,
      "loss": 0.4174,
      "step": 13172
    },
    {
      "epoch": 0.6045156257170391,
      "grad_norm": 0.4744017720222473,
      "learning_rate": 9.177289347579144e-06,
      "loss": 0.3564,
      "step": 13173
    },
    {
      "epoch": 0.6045615162222936,
      "grad_norm": 0.45266127586364746,
      "learning_rate": 9.177154598439021e-06,
      "loss": 0.358,
      "step": 13174
    },
    {
      "epoch": 0.6046074067275481,
      "grad_norm": 0.4903585910797119,
      "learning_rate": 9.177019839254165e-06,
      "loss": 0.4423,
      "step": 13175
    },
    {
      "epoch": 0.6046532972328025,
      "grad_norm": 0.45294323563575745,
      "learning_rate": 9.17688507002489e-06,
      "loss": 0.3936,
      "step": 13176
    },
    {
      "epoch": 0.604699187738057,
      "grad_norm": 0.49111440777778625,
      "learning_rate": 9.176750290751528e-06,
      "loss": 0.4366,
      "step": 13177
    },
    {
      "epoch": 0.6047450782433115,
      "grad_norm": 0.5553954243659973,
      "learning_rate": 9.176615501434398e-06,
      "loss": 0.4649,
      "step": 13178
    },
    {
      "epoch": 0.604790968748566,
      "grad_norm": 0.4900134801864624,
      "learning_rate": 9.176480702073827e-06,
      "loss": 0.5209,
      "step": 13179
    },
    {
      "epoch": 0.6048368592538204,
      "grad_norm": 0.4586549401283264,
      "learning_rate": 9.176345892670137e-06,
      "loss": 0.4335,
      "step": 13180
    },
    {
      "epoch": 0.6048827497590749,
      "grad_norm": 0.44659167528152466,
      "learning_rate": 9.176211073223652e-06,
      "loss": 0.3732,
      "step": 13181
    },
    {
      "epoch": 0.6049286402643294,
      "grad_norm": 0.45426857471466064,
      "learning_rate": 9.176076243734702e-06,
      "loss": 0.3362,
      "step": 13182
    },
    {
      "epoch": 0.6049745307695837,
      "grad_norm": 0.44608205556869507,
      "learning_rate": 9.175941404203603e-06,
      "loss": 0.3709,
      "step": 13183
    },
    {
      "epoch": 0.6050204212748382,
      "grad_norm": 0.5027086138725281,
      "learning_rate": 9.175806554630685e-06,
      "loss": 0.4305,
      "step": 13184
    },
    {
      "epoch": 0.6050663117800927,
      "grad_norm": 0.47224509716033936,
      "learning_rate": 9.175671695016268e-06,
      "loss": 0.4154,
      "step": 13185
    },
    {
      "epoch": 0.6051122022853471,
      "grad_norm": 0.49410581588745117,
      "learning_rate": 9.175536825360678e-06,
      "loss": 0.5186,
      "step": 13186
    },
    {
      "epoch": 0.6051580927906016,
      "grad_norm": 0.46911725401878357,
      "learning_rate": 9.175401945664242e-06,
      "loss": 0.3999,
      "step": 13187
    },
    {
      "epoch": 0.6052039832958561,
      "grad_norm": 0.4600519835948944,
      "learning_rate": 9.17526705592728e-06,
      "loss": 0.3712,
      "step": 13188
    },
    {
      "epoch": 0.6052498738011105,
      "grad_norm": 0.46647652983665466,
      "learning_rate": 9.17513215615012e-06,
      "loss": 0.4224,
      "step": 13189
    },
    {
      "epoch": 0.605295764306365,
      "grad_norm": 0.44988003373146057,
      "learning_rate": 9.174997246333084e-06,
      "loss": 0.3621,
      "step": 13190
    },
    {
      "epoch": 0.6053416548116195,
      "grad_norm": 0.44794827699661255,
      "learning_rate": 9.174862326476498e-06,
      "loss": 0.442,
      "step": 13191
    },
    {
      "epoch": 0.6053875453168739,
      "grad_norm": 0.5060514807701111,
      "learning_rate": 9.174727396580685e-06,
      "loss": 0.5335,
      "step": 13192
    },
    {
      "epoch": 0.6054334358221284,
      "grad_norm": 0.43283554911613464,
      "learning_rate": 9.17459245664597e-06,
      "loss": 0.3723,
      "step": 13193
    },
    {
      "epoch": 0.6054793263273829,
      "grad_norm": 0.45293816924095154,
      "learning_rate": 9.174457506672679e-06,
      "loss": 0.4109,
      "step": 13194
    },
    {
      "epoch": 0.6055252168326374,
      "grad_norm": 0.4975714683532715,
      "learning_rate": 9.174322546661135e-06,
      "loss": 0.5223,
      "step": 13195
    },
    {
      "epoch": 0.6055711073378918,
      "grad_norm": 0.4433005750179291,
      "learning_rate": 9.174187576611662e-06,
      "loss": 0.3349,
      "step": 13196
    },
    {
      "epoch": 0.6056169978431463,
      "grad_norm": 0.4572184383869171,
      "learning_rate": 9.174052596524586e-06,
      "loss": 0.4109,
      "step": 13197
    },
    {
      "epoch": 0.6056628883484008,
      "grad_norm": 0.4908665120601654,
      "learning_rate": 9.17391760640023e-06,
      "loss": 0.423,
      "step": 13198
    },
    {
      "epoch": 0.6057087788536551,
      "grad_norm": 0.5083826184272766,
      "learning_rate": 9.17378260623892e-06,
      "loss": 0.4312,
      "step": 13199
    },
    {
      "epoch": 0.6057546693589096,
      "grad_norm": 0.46048974990844727,
      "learning_rate": 9.17364759604098e-06,
      "loss": 0.413,
      "step": 13200
    },
    {
      "epoch": 0.6058005598641641,
      "grad_norm": 0.45665237307548523,
      "learning_rate": 9.173512575806735e-06,
      "loss": 0.3945,
      "step": 13201
    },
    {
      "epoch": 0.6058464503694185,
      "grad_norm": 0.4536088705062866,
      "learning_rate": 9.17337754553651e-06,
      "loss": 0.4003,
      "step": 13202
    },
    {
      "epoch": 0.605892340874673,
      "grad_norm": 0.4466998279094696,
      "learning_rate": 9.173242505230628e-06,
      "loss": 0.3578,
      "step": 13203
    },
    {
      "epoch": 0.6059382313799275,
      "grad_norm": 0.4961372911930084,
      "learning_rate": 9.173107454889415e-06,
      "loss": 0.423,
      "step": 13204
    },
    {
      "epoch": 0.6059841218851819,
      "grad_norm": 0.5069820284843445,
      "learning_rate": 9.172972394513195e-06,
      "loss": 0.3985,
      "step": 13205
    },
    {
      "epoch": 0.6060300123904364,
      "grad_norm": 0.4713674783706665,
      "learning_rate": 9.172837324102294e-06,
      "loss": 0.3787,
      "step": 13206
    },
    {
      "epoch": 0.6060759028956909,
      "grad_norm": 0.4359363913536072,
      "learning_rate": 9.172702243657038e-06,
      "loss": 0.3892,
      "step": 13207
    },
    {
      "epoch": 0.6061217934009453,
      "grad_norm": 0.44410932064056396,
      "learning_rate": 9.172567153177749e-06,
      "loss": 0.3682,
      "step": 13208
    },
    {
      "epoch": 0.6061676839061998,
      "grad_norm": 0.4700978696346283,
      "learning_rate": 9.172432052664753e-06,
      "loss": 0.4116,
      "step": 13209
    },
    {
      "epoch": 0.6062135744114543,
      "grad_norm": 0.5533850789070129,
      "learning_rate": 9.172296942118373e-06,
      "loss": 0.4899,
      "step": 13210
    },
    {
      "epoch": 0.6062594649167087,
      "grad_norm": 0.543442964553833,
      "learning_rate": 9.172161821538938e-06,
      "loss": 0.4912,
      "step": 13211
    },
    {
      "epoch": 0.6063053554219632,
      "grad_norm": 0.43508365750312805,
      "learning_rate": 9.172026690926768e-06,
      "loss": 0.3445,
      "step": 13212
    },
    {
      "epoch": 0.6063512459272177,
      "grad_norm": 0.4867077171802521,
      "learning_rate": 9.171891550282193e-06,
      "loss": 0.452,
      "step": 13213
    },
    {
      "epoch": 0.6063971364324722,
      "grad_norm": 0.4436841607093811,
      "learning_rate": 9.171756399605535e-06,
      "loss": 0.3598,
      "step": 13214
    },
    {
      "epoch": 0.6064430269377266,
      "grad_norm": 0.45380640029907227,
      "learning_rate": 9.171621238897118e-06,
      "loss": 0.3598,
      "step": 13215
    },
    {
      "epoch": 0.6064889174429811,
      "grad_norm": 0.46638739109039307,
      "learning_rate": 9.17148606815727e-06,
      "loss": 0.4026,
      "step": 13216
    },
    {
      "epoch": 0.6065348079482356,
      "grad_norm": 0.45554307103157043,
      "learning_rate": 9.171350887386315e-06,
      "loss": 0.4218,
      "step": 13217
    },
    {
      "epoch": 0.6065806984534899,
      "grad_norm": 0.4145069122314453,
      "learning_rate": 9.171215696584577e-06,
      "loss": 0.3313,
      "step": 13218
    },
    {
      "epoch": 0.6066265889587444,
      "grad_norm": 0.48824459314346313,
      "learning_rate": 9.17108049575238e-06,
      "loss": 0.4952,
      "step": 13219
    },
    {
      "epoch": 0.6066724794639989,
      "grad_norm": 0.4915045201778412,
      "learning_rate": 9.170945284890054e-06,
      "loss": 0.4724,
      "step": 13220
    },
    {
      "epoch": 0.6067183699692533,
      "grad_norm": 0.4418303668498993,
      "learning_rate": 9.17081006399792e-06,
      "loss": 0.3506,
      "step": 13221
    },
    {
      "epoch": 0.6067642604745078,
      "grad_norm": 0.6353857517242432,
      "learning_rate": 9.170674833076301e-06,
      "loss": 0.325,
      "step": 13222
    },
    {
      "epoch": 0.6068101509797623,
      "grad_norm": 0.4405895471572876,
      "learning_rate": 9.170539592125529e-06,
      "loss": 0.3936,
      "step": 13223
    },
    {
      "epoch": 0.6068560414850167,
      "grad_norm": 0.47357064485549927,
      "learning_rate": 9.170404341145925e-06,
      "loss": 0.4479,
      "step": 13224
    },
    {
      "epoch": 0.6069019319902712,
      "grad_norm": 0.4581132233142853,
      "learning_rate": 9.170269080137814e-06,
      "loss": 0.367,
      "step": 13225
    },
    {
      "epoch": 0.6069478224955257,
      "grad_norm": 0.5103758573532104,
      "learning_rate": 9.170133809101522e-06,
      "loss": 0.3908,
      "step": 13226
    },
    {
      "epoch": 0.6069937130007801,
      "grad_norm": 0.48814964294433594,
      "learning_rate": 9.169998528037375e-06,
      "loss": 0.3826,
      "step": 13227
    },
    {
      "epoch": 0.6070396035060346,
      "grad_norm": 0.45264437794685364,
      "learning_rate": 9.169863236945697e-06,
      "loss": 0.3538,
      "step": 13228
    },
    {
      "epoch": 0.6070854940112891,
      "grad_norm": 0.47573983669281006,
      "learning_rate": 9.169727935826813e-06,
      "loss": 0.3834,
      "step": 13229
    },
    {
      "epoch": 0.6071313845165436,
      "grad_norm": 0.4457363486289978,
      "learning_rate": 9.169592624681052e-06,
      "loss": 0.4592,
      "step": 13230
    },
    {
      "epoch": 0.607177275021798,
      "grad_norm": 0.44521382451057434,
      "learning_rate": 9.169457303508734e-06,
      "loss": 0.4084,
      "step": 13231
    },
    {
      "epoch": 0.6072231655270525,
      "grad_norm": 0.44730156660079956,
      "learning_rate": 9.169321972310189e-06,
      "loss": 0.3678,
      "step": 13232
    },
    {
      "epoch": 0.607269056032307,
      "grad_norm": 0.4619680643081665,
      "learning_rate": 9.169186631085738e-06,
      "loss": 0.4317,
      "step": 13233
    },
    {
      "epoch": 0.6073149465375614,
      "grad_norm": 0.47980552911758423,
      "learning_rate": 9.169051279835712e-06,
      "loss": 0.4293,
      "step": 13234
    },
    {
      "epoch": 0.6073608370428158,
      "grad_norm": 0.4269371032714844,
      "learning_rate": 9.168915918560433e-06,
      "loss": 0.3021,
      "step": 13235
    },
    {
      "epoch": 0.6074067275480703,
      "grad_norm": 0.44181379675865173,
      "learning_rate": 9.168780547260225e-06,
      "loss": 0.3778,
      "step": 13236
    },
    {
      "epoch": 0.6074526180533247,
      "grad_norm": 0.4818260967731476,
      "learning_rate": 9.168645165935417e-06,
      "loss": 0.387,
      "step": 13237
    },
    {
      "epoch": 0.6074985085585792,
      "grad_norm": 0.45843955874443054,
      "learning_rate": 9.168509774586331e-06,
      "loss": 0.4406,
      "step": 13238
    },
    {
      "epoch": 0.6075443990638337,
      "grad_norm": 0.4779668152332306,
      "learning_rate": 9.168374373213298e-06,
      "loss": 0.4188,
      "step": 13239
    },
    {
      "epoch": 0.6075902895690881,
      "grad_norm": 0.45344918966293335,
      "learning_rate": 9.168238961816637e-06,
      "loss": 0.3955,
      "step": 13240
    },
    {
      "epoch": 0.6076361800743426,
      "grad_norm": 0.43827101588249207,
      "learning_rate": 9.168103540396681e-06,
      "loss": 0.3971,
      "step": 13241
    },
    {
      "epoch": 0.6076820705795971,
      "grad_norm": 0.464828759431839,
      "learning_rate": 9.167968108953748e-06,
      "loss": 0.3808,
      "step": 13242
    },
    {
      "epoch": 0.6077279610848515,
      "grad_norm": 0.4554233253002167,
      "learning_rate": 9.167832667488168e-06,
      "loss": 0.3729,
      "step": 13243
    },
    {
      "epoch": 0.607773851590106,
      "grad_norm": 0.453374445438385,
      "learning_rate": 9.167697216000265e-06,
      "loss": 0.3919,
      "step": 13244
    },
    {
      "epoch": 0.6078197420953605,
      "grad_norm": 0.46702778339385986,
      "learning_rate": 9.167561754490366e-06,
      "loss": 0.4137,
      "step": 13245
    },
    {
      "epoch": 0.6078656326006149,
      "grad_norm": 0.4790914058685303,
      "learning_rate": 9.167426282958797e-06,
      "loss": 0.4441,
      "step": 13246
    },
    {
      "epoch": 0.6079115231058694,
      "grad_norm": 0.4316692650318146,
      "learning_rate": 9.167290801405883e-06,
      "loss": 0.3765,
      "step": 13247
    },
    {
      "epoch": 0.6079574136111239,
      "grad_norm": 0.45797622203826904,
      "learning_rate": 9.16715530983195e-06,
      "loss": 0.4203,
      "step": 13248
    },
    {
      "epoch": 0.6080033041163784,
      "grad_norm": 0.442089319229126,
      "learning_rate": 9.167019808237323e-06,
      "loss": 0.361,
      "step": 13249
    },
    {
      "epoch": 0.6080491946216328,
      "grad_norm": 0.4688459634780884,
      "learning_rate": 9.166884296622329e-06,
      "loss": 0.3473,
      "step": 13250
    },
    {
      "epoch": 0.6080950851268873,
      "grad_norm": 0.4200957119464874,
      "learning_rate": 9.166748774987292e-06,
      "loss": 0.3012,
      "step": 13251
    },
    {
      "epoch": 0.6081409756321418,
      "grad_norm": 0.43397995829582214,
      "learning_rate": 9.16661324333254e-06,
      "loss": 0.3807,
      "step": 13252
    },
    {
      "epoch": 0.6081868661373961,
      "grad_norm": 0.5297274589538574,
      "learning_rate": 9.1664777016584e-06,
      "loss": 0.5295,
      "step": 13253
    },
    {
      "epoch": 0.6082327566426506,
      "grad_norm": 0.4836972653865814,
      "learning_rate": 9.166342149965195e-06,
      "loss": 0.3667,
      "step": 13254
    },
    {
      "epoch": 0.6082786471479051,
      "grad_norm": 0.4567781388759613,
      "learning_rate": 9.16620658825325e-06,
      "loss": 0.3751,
      "step": 13255
    },
    {
      "epoch": 0.6083245376531595,
      "grad_norm": 0.4763796329498291,
      "learning_rate": 9.166071016522895e-06,
      "loss": 0.4774,
      "step": 13256
    },
    {
      "epoch": 0.608370428158414,
      "grad_norm": 0.4118577241897583,
      "learning_rate": 9.165935434774455e-06,
      "loss": 0.3197,
      "step": 13257
    },
    {
      "epoch": 0.6084163186636685,
      "grad_norm": 0.43617308139801025,
      "learning_rate": 9.165799843008253e-06,
      "loss": 0.3347,
      "step": 13258
    },
    {
      "epoch": 0.6084622091689229,
      "grad_norm": 0.44588226079940796,
      "learning_rate": 9.165664241224618e-06,
      "loss": 0.4007,
      "step": 13259
    },
    {
      "epoch": 0.6085080996741774,
      "grad_norm": 0.5056838393211365,
      "learning_rate": 9.165528629423877e-06,
      "loss": 0.5304,
      "step": 13260
    },
    {
      "epoch": 0.6085539901794319,
      "grad_norm": 0.47182509303092957,
      "learning_rate": 9.165393007606352e-06,
      "loss": 0.4269,
      "step": 13261
    },
    {
      "epoch": 0.6085998806846863,
      "grad_norm": 0.5026556849479675,
      "learning_rate": 9.165257375772371e-06,
      "loss": 0.5562,
      "step": 13262
    },
    {
      "epoch": 0.6086457711899408,
      "grad_norm": 0.47351646423339844,
      "learning_rate": 9.165121733922262e-06,
      "loss": 0.4186,
      "step": 13263
    },
    {
      "epoch": 0.6086916616951953,
      "grad_norm": 0.4371229112148285,
      "learning_rate": 9.16498608205635e-06,
      "loss": 0.3282,
      "step": 13264
    },
    {
      "epoch": 0.6087375522004497,
      "grad_norm": 0.5177859663963318,
      "learning_rate": 9.16485042017496e-06,
      "loss": 0.5101,
      "step": 13265
    },
    {
      "epoch": 0.6087834427057042,
      "grad_norm": 0.46313774585723877,
      "learning_rate": 9.16471474827842e-06,
      "loss": 0.4103,
      "step": 13266
    },
    {
      "epoch": 0.6088293332109587,
      "grad_norm": 0.46497783064842224,
      "learning_rate": 9.164579066367054e-06,
      "loss": 0.4165,
      "step": 13267
    },
    {
      "epoch": 0.6088752237162132,
      "grad_norm": 0.4599396884441376,
      "learning_rate": 9.164443374441191e-06,
      "loss": 0.4143,
      "step": 13268
    },
    {
      "epoch": 0.6089211142214676,
      "grad_norm": 0.45004889369010925,
      "learning_rate": 9.164307672501157e-06,
      "loss": 0.3463,
      "step": 13269
    },
    {
      "epoch": 0.608967004726722,
      "grad_norm": 0.4713543951511383,
      "learning_rate": 9.164171960547275e-06,
      "loss": 0.4281,
      "step": 13270
    },
    {
      "epoch": 0.6090128952319765,
      "grad_norm": 0.511316180229187,
      "learning_rate": 9.164036238579876e-06,
      "loss": 0.4624,
      "step": 13271
    },
    {
      "epoch": 0.6090587857372309,
      "grad_norm": 0.4661910831928253,
      "learning_rate": 9.163900506599283e-06,
      "loss": 0.4194,
      "step": 13272
    },
    {
      "epoch": 0.6091046762424854,
      "grad_norm": 0.4375600218772888,
      "learning_rate": 9.163764764605823e-06,
      "loss": 0.3624,
      "step": 13273
    },
    {
      "epoch": 0.6091505667477399,
      "grad_norm": 0.4491141438484192,
      "learning_rate": 9.163629012599824e-06,
      "loss": 0.3493,
      "step": 13274
    },
    {
      "epoch": 0.6091964572529943,
      "grad_norm": 0.4904569983482361,
      "learning_rate": 9.16349325058161e-06,
      "loss": 0.4841,
      "step": 13275
    },
    {
      "epoch": 0.6092423477582488,
      "grad_norm": 0.46743354201316833,
      "learning_rate": 9.16335747855151e-06,
      "loss": 0.3889,
      "step": 13276
    },
    {
      "epoch": 0.6092882382635033,
      "grad_norm": 0.41804802417755127,
      "learning_rate": 9.163221696509847e-06,
      "loss": 0.3099,
      "step": 13277
    },
    {
      "epoch": 0.6093341287687577,
      "grad_norm": 0.43893468379974365,
      "learning_rate": 9.163085904456952e-06,
      "loss": 0.3489,
      "step": 13278
    },
    {
      "epoch": 0.6093800192740122,
      "grad_norm": 0.46362921595573425,
      "learning_rate": 9.16295010239315e-06,
      "loss": 0.4929,
      "step": 13279
    },
    {
      "epoch": 0.6094259097792667,
      "grad_norm": 0.44740214943885803,
      "learning_rate": 9.162814290318764e-06,
      "loss": 0.389,
      "step": 13280
    },
    {
      "epoch": 0.6094718002845211,
      "grad_norm": 0.45767709612846375,
      "learning_rate": 9.162678468234126e-06,
      "loss": 0.4196,
      "step": 13281
    },
    {
      "epoch": 0.6095176907897756,
      "grad_norm": 0.46190154552459717,
      "learning_rate": 9.162542636139557e-06,
      "loss": 0.3874,
      "step": 13282
    },
    {
      "epoch": 0.6095635812950301,
      "grad_norm": 0.4538164734840393,
      "learning_rate": 9.16240679403539e-06,
      "loss": 0.3929,
      "step": 13283
    },
    {
      "epoch": 0.6096094718002846,
      "grad_norm": 0.45440441370010376,
      "learning_rate": 9.162270941921947e-06,
      "loss": 0.421,
      "step": 13284
    },
    {
      "epoch": 0.609655362305539,
      "grad_norm": 0.4806460738182068,
      "learning_rate": 9.162135079799557e-06,
      "loss": 0.4569,
      "step": 13285
    },
    {
      "epoch": 0.6097012528107935,
      "grad_norm": 0.4554196894168854,
      "learning_rate": 9.161999207668544e-06,
      "loss": 0.4114,
      "step": 13286
    },
    {
      "epoch": 0.609747143316048,
      "grad_norm": 0.47221019864082336,
      "learning_rate": 9.161863325529237e-06,
      "loss": 0.4237,
      "step": 13287
    },
    {
      "epoch": 0.6097930338213023,
      "grad_norm": 0.44815418124198914,
      "learning_rate": 9.161727433381962e-06,
      "loss": 0.3815,
      "step": 13288
    },
    {
      "epoch": 0.6098389243265568,
      "grad_norm": 0.472644180059433,
      "learning_rate": 9.161591531227048e-06,
      "loss": 0.4549,
      "step": 13289
    },
    {
      "epoch": 0.6098848148318113,
      "grad_norm": 0.45040163397789,
      "learning_rate": 9.161455619064818e-06,
      "loss": 0.3823,
      "step": 13290
    },
    {
      "epoch": 0.6099307053370657,
      "grad_norm": 0.45927244424819946,
      "learning_rate": 9.1613196968956e-06,
      "loss": 0.3916,
      "step": 13291
    },
    {
      "epoch": 0.6099765958423202,
      "grad_norm": 0.45920559763908386,
      "learning_rate": 9.161183764719724e-06,
      "loss": 0.366,
      "step": 13292
    },
    {
      "epoch": 0.6100224863475747,
      "grad_norm": 0.4457528591156006,
      "learning_rate": 9.161047822537513e-06,
      "loss": 0.376,
      "step": 13293
    },
    {
      "epoch": 0.6100683768528291,
      "grad_norm": 0.43101900815963745,
      "learning_rate": 9.160911870349293e-06,
      "loss": 0.3429,
      "step": 13294
    },
    {
      "epoch": 0.6101142673580836,
      "grad_norm": 0.5050458908081055,
      "learning_rate": 9.160775908155397e-06,
      "loss": 0.4963,
      "step": 13295
    },
    {
      "epoch": 0.6101601578633381,
      "grad_norm": 0.5203952193260193,
      "learning_rate": 9.160639935956147e-06,
      "loss": 0.4884,
      "step": 13296
    },
    {
      "epoch": 0.6102060483685925,
      "grad_norm": 0.4877369999885559,
      "learning_rate": 9.16050395375187e-06,
      "loss": 0.4454,
      "step": 13297
    },
    {
      "epoch": 0.610251938873847,
      "grad_norm": 0.4694271683692932,
      "learning_rate": 9.160367961542894e-06,
      "loss": 0.4539,
      "step": 13298
    },
    {
      "epoch": 0.6102978293791015,
      "grad_norm": 0.4723474383354187,
      "learning_rate": 9.160231959329547e-06,
      "loss": 0.4457,
      "step": 13299
    },
    {
      "epoch": 0.6103437198843559,
      "grad_norm": 0.49191516637802124,
      "learning_rate": 9.160095947112154e-06,
      "loss": 0.4472,
      "step": 13300
    },
    {
      "epoch": 0.6103896103896104,
      "grad_norm": 0.46903061866760254,
      "learning_rate": 9.159959924891045e-06,
      "loss": 0.4144,
      "step": 13301
    },
    {
      "epoch": 0.6104355008948649,
      "grad_norm": 0.4640098512172699,
      "learning_rate": 9.159823892666544e-06,
      "loss": 0.4087,
      "step": 13302
    },
    {
      "epoch": 0.6104813914001194,
      "grad_norm": 0.4533601999282837,
      "learning_rate": 9.15968785043898e-06,
      "loss": 0.3645,
      "step": 13303
    },
    {
      "epoch": 0.6105272819053738,
      "grad_norm": 0.47778424620628357,
      "learning_rate": 9.159551798208678e-06,
      "loss": 0.4073,
      "step": 13304
    },
    {
      "epoch": 0.6105731724106283,
      "grad_norm": 0.4762365520000458,
      "learning_rate": 9.159415735975967e-06,
      "loss": 0.4052,
      "step": 13305
    },
    {
      "epoch": 0.6106190629158827,
      "grad_norm": 0.4551175534725189,
      "learning_rate": 9.159279663741177e-06,
      "loss": 0.3907,
      "step": 13306
    },
    {
      "epoch": 0.6106649534211371,
      "grad_norm": 0.4675368666648865,
      "learning_rate": 9.15914358150463e-06,
      "loss": 0.3595,
      "step": 13307
    },
    {
      "epoch": 0.6107108439263916,
      "grad_norm": 0.409780889749527,
      "learning_rate": 9.159007489266655e-06,
      "loss": 0.3363,
      "step": 13308
    },
    {
      "epoch": 0.6107567344316461,
      "grad_norm": 0.4369770884513855,
      "learning_rate": 9.158871387027579e-06,
      "loss": 0.3007,
      "step": 13309
    },
    {
      "epoch": 0.6108026249369005,
      "grad_norm": 0.4592125415802002,
      "learning_rate": 9.15873527478773e-06,
      "loss": 0.3907,
      "step": 13310
    },
    {
      "epoch": 0.610848515442155,
      "grad_norm": 0.44635066390037537,
      "learning_rate": 9.158599152547437e-06,
      "loss": 0.4421,
      "step": 13311
    },
    {
      "epoch": 0.6108944059474095,
      "grad_norm": 0.4271751940250397,
      "learning_rate": 9.158463020307025e-06,
      "loss": 0.3018,
      "step": 13312
    },
    {
      "epoch": 0.6109402964526639,
      "grad_norm": 0.4735080599784851,
      "learning_rate": 9.158326878066821e-06,
      "loss": 0.4225,
      "step": 13313
    },
    {
      "epoch": 0.6109861869579184,
      "grad_norm": 0.4223187267780304,
      "learning_rate": 9.158190725827155e-06,
      "loss": 0.3714,
      "step": 13314
    },
    {
      "epoch": 0.6110320774631729,
      "grad_norm": 0.45918720960617065,
      "learning_rate": 9.158054563588349e-06,
      "loss": 0.3681,
      "step": 13315
    },
    {
      "epoch": 0.6110779679684273,
      "grad_norm": 0.459389865398407,
      "learning_rate": 9.157918391350738e-06,
      "loss": 0.3977,
      "step": 13316
    },
    {
      "epoch": 0.6111238584736818,
      "grad_norm": 0.4682282507419586,
      "learning_rate": 9.157782209114643e-06,
      "loss": 0.4298,
      "step": 13317
    },
    {
      "epoch": 0.6111697489789363,
      "grad_norm": 0.5041065812110901,
      "learning_rate": 9.157646016880395e-06,
      "loss": 0.4551,
      "step": 13318
    },
    {
      "epoch": 0.6112156394841908,
      "grad_norm": 0.47502198815345764,
      "learning_rate": 9.157509814648323e-06,
      "loss": 0.3647,
      "step": 13319
    },
    {
      "epoch": 0.6112615299894452,
      "grad_norm": 0.47289445996284485,
      "learning_rate": 9.157373602418749e-06,
      "loss": 0.4238,
      "step": 13320
    },
    {
      "epoch": 0.6113074204946997,
      "grad_norm": 0.42670324444770813,
      "learning_rate": 9.157237380192003e-06,
      "loss": 0.3684,
      "step": 13321
    },
    {
      "epoch": 0.6113533109999542,
      "grad_norm": 0.46294015645980835,
      "learning_rate": 9.157101147968414e-06,
      "loss": 0.4177,
      "step": 13322
    },
    {
      "epoch": 0.6113992015052085,
      "grad_norm": 0.47807419300079346,
      "learning_rate": 9.15696490574831e-06,
      "loss": 0.3844,
      "step": 13323
    },
    {
      "epoch": 0.611445092010463,
      "grad_norm": 0.4557643532752991,
      "learning_rate": 9.156828653532018e-06,
      "loss": 0.3651,
      "step": 13324
    },
    {
      "epoch": 0.6114909825157175,
      "grad_norm": 0.4858211278915405,
      "learning_rate": 9.156692391319862e-06,
      "loss": 0.3605,
      "step": 13325
    },
    {
      "epoch": 0.6115368730209719,
      "grad_norm": 0.47030797600746155,
      "learning_rate": 9.156556119112175e-06,
      "loss": 0.447,
      "step": 13326
    },
    {
      "epoch": 0.6115827635262264,
      "grad_norm": 0.47900688648223877,
      "learning_rate": 9.15641983690928e-06,
      "loss": 0.4221,
      "step": 13327
    },
    {
      "epoch": 0.6116286540314809,
      "grad_norm": 0.43420517444610596,
      "learning_rate": 9.156283544711509e-06,
      "loss": 0.3492,
      "step": 13328
    },
    {
      "epoch": 0.6116745445367353,
      "grad_norm": 0.4537833333015442,
      "learning_rate": 9.156147242519188e-06,
      "loss": 0.3679,
      "step": 13329
    },
    {
      "epoch": 0.6117204350419898,
      "grad_norm": 0.44130414724349976,
      "learning_rate": 9.156010930332643e-06,
      "loss": 0.3723,
      "step": 13330
    },
    {
      "epoch": 0.6117663255472443,
      "grad_norm": 0.47028690576553345,
      "learning_rate": 9.155874608152206e-06,
      "loss": 0.4246,
      "step": 13331
    },
    {
      "epoch": 0.6118122160524987,
      "grad_norm": 0.423086553812027,
      "learning_rate": 9.155738275978201e-06,
      "loss": 0.3828,
      "step": 13332
    },
    {
      "epoch": 0.6118581065577532,
      "grad_norm": 0.4757053554058075,
      "learning_rate": 9.155601933810956e-06,
      "loss": 0.3893,
      "step": 13333
    },
    {
      "epoch": 0.6119039970630077,
      "grad_norm": 0.49114707112312317,
      "learning_rate": 9.155465581650801e-06,
      "loss": 0.3914,
      "step": 13334
    },
    {
      "epoch": 0.6119498875682621,
      "grad_norm": 0.46154144406318665,
      "learning_rate": 9.155329219498063e-06,
      "loss": 0.3805,
      "step": 13335
    },
    {
      "epoch": 0.6119957780735166,
      "grad_norm": 0.48199644684791565,
      "learning_rate": 9.155192847353068e-06,
      "loss": 0.4478,
      "step": 13336
    },
    {
      "epoch": 0.6120416685787711,
      "grad_norm": 0.4245755672454834,
      "learning_rate": 9.155056465216147e-06,
      "loss": 0.301,
      "step": 13337
    },
    {
      "epoch": 0.6120875590840256,
      "grad_norm": 0.4448850750923157,
      "learning_rate": 9.154920073087626e-06,
      "loss": 0.438,
      "step": 13338
    },
    {
      "epoch": 0.61213344958928,
      "grad_norm": 0.44625014066696167,
      "learning_rate": 9.154783670967835e-06,
      "loss": 0.4024,
      "step": 13339
    },
    {
      "epoch": 0.6121793400945345,
      "grad_norm": 0.4629194140434265,
      "learning_rate": 9.1546472588571e-06,
      "loss": 0.4416,
      "step": 13340
    },
    {
      "epoch": 0.612225230599789,
      "grad_norm": 0.46314817667007446,
      "learning_rate": 9.154510836755748e-06,
      "loss": 0.3864,
      "step": 13341
    },
    {
      "epoch": 0.6122711211050433,
      "grad_norm": 0.49812954664230347,
      "learning_rate": 9.15437440466411e-06,
      "loss": 0.4162,
      "step": 13342
    },
    {
      "epoch": 0.6123170116102978,
      "grad_norm": 0.44824784994125366,
      "learning_rate": 9.154237962582511e-06,
      "loss": 0.378,
      "step": 13343
    },
    {
      "epoch": 0.6123629021155523,
      "grad_norm": 0.4450624883174896,
      "learning_rate": 9.154101510511283e-06,
      "loss": 0.3625,
      "step": 13344
    },
    {
      "epoch": 0.6124087926208067,
      "grad_norm": 0.4278448224067688,
      "learning_rate": 9.15396504845075e-06,
      "loss": 0.3262,
      "step": 13345
    },
    {
      "epoch": 0.6124546831260612,
      "grad_norm": 0.47586536407470703,
      "learning_rate": 9.153828576401245e-06,
      "loss": 0.469,
      "step": 13346
    },
    {
      "epoch": 0.6125005736313157,
      "grad_norm": 0.4475921094417572,
      "learning_rate": 9.153692094363092e-06,
      "loss": 0.3417,
      "step": 13347
    },
    {
      "epoch": 0.6125464641365701,
      "grad_norm": 0.4293804168701172,
      "learning_rate": 9.15355560233662e-06,
      "loss": 0.3636,
      "step": 13348
    },
    {
      "epoch": 0.6125923546418246,
      "grad_norm": 0.4287230372428894,
      "learning_rate": 9.153419100322157e-06,
      "loss": 0.3868,
      "step": 13349
    },
    {
      "epoch": 0.6126382451470791,
      "grad_norm": 0.4459531903266907,
      "learning_rate": 9.153282588320032e-06,
      "loss": 0.3788,
      "step": 13350
    },
    {
      "epoch": 0.6126841356523335,
      "grad_norm": 0.4465651512145996,
      "learning_rate": 9.153146066330574e-06,
      "loss": 0.3785,
      "step": 13351
    },
    {
      "epoch": 0.612730026157588,
      "grad_norm": 0.43893948197364807,
      "learning_rate": 9.15300953435411e-06,
      "loss": 0.3819,
      "step": 13352
    },
    {
      "epoch": 0.6127759166628425,
      "grad_norm": 0.4302557408809662,
      "learning_rate": 9.152872992390969e-06,
      "loss": 0.389,
      "step": 13353
    },
    {
      "epoch": 0.6128218071680969,
      "grad_norm": 0.466801255941391,
      "learning_rate": 9.15273644044148e-06,
      "loss": 0.44,
      "step": 13354
    },
    {
      "epoch": 0.6128676976733514,
      "grad_norm": 0.45544758439064026,
      "learning_rate": 9.152599878505968e-06,
      "loss": 0.3866,
      "step": 13355
    },
    {
      "epoch": 0.6129135881786059,
      "grad_norm": 0.4682729244232178,
      "learning_rate": 9.152463306584766e-06,
      "loss": 0.3874,
      "step": 13356
    },
    {
      "epoch": 0.6129594786838604,
      "grad_norm": 0.4749312102794647,
      "learning_rate": 9.1523267246782e-06,
      "loss": 0.4494,
      "step": 13357
    },
    {
      "epoch": 0.6130053691891147,
      "grad_norm": 0.5078727602958679,
      "learning_rate": 9.152190132786598e-06,
      "loss": 0.5024,
      "step": 13358
    },
    {
      "epoch": 0.6130512596943692,
      "grad_norm": 0.4542948305606842,
      "learning_rate": 9.15205353091029e-06,
      "loss": 0.3657,
      "step": 13359
    },
    {
      "epoch": 0.6130971501996237,
      "grad_norm": 0.44105008244514465,
      "learning_rate": 9.151916919049604e-06,
      "loss": 0.3682,
      "step": 13360
    },
    {
      "epoch": 0.6131430407048781,
      "grad_norm": 0.4381992220878601,
      "learning_rate": 9.151780297204867e-06,
      "loss": 0.3544,
      "step": 13361
    },
    {
      "epoch": 0.6131889312101326,
      "grad_norm": 0.4764685332775116,
      "learning_rate": 9.151643665376408e-06,
      "loss": 0.4242,
      "step": 13362
    },
    {
      "epoch": 0.6132348217153871,
      "grad_norm": 0.4655154049396515,
      "learning_rate": 9.15150702356456e-06,
      "loss": 0.3931,
      "step": 13363
    },
    {
      "epoch": 0.6132807122206415,
      "grad_norm": 0.49365729093551636,
      "learning_rate": 9.151370371769645e-06,
      "loss": 0.4438,
      "step": 13364
    },
    {
      "epoch": 0.613326602725896,
      "grad_norm": 0.4606526494026184,
      "learning_rate": 9.151233709991995e-06,
      "loss": 0.4109,
      "step": 13365
    },
    {
      "epoch": 0.6133724932311505,
      "grad_norm": 0.4721617102622986,
      "learning_rate": 9.151097038231938e-06,
      "loss": 0.4438,
      "step": 13366
    },
    {
      "epoch": 0.6134183837364049,
      "grad_norm": 0.4713938236236572,
      "learning_rate": 9.150960356489803e-06,
      "loss": 0.3668,
      "step": 13367
    },
    {
      "epoch": 0.6134642742416594,
      "grad_norm": 0.4764348268508911,
      "learning_rate": 9.150823664765918e-06,
      "loss": 0.3837,
      "step": 13368
    },
    {
      "epoch": 0.6135101647469139,
      "grad_norm": 0.4997861683368683,
      "learning_rate": 9.150686963060612e-06,
      "loss": 0.4279,
      "step": 13369
    },
    {
      "epoch": 0.6135560552521683,
      "grad_norm": 0.49250558018684387,
      "learning_rate": 9.150550251374213e-06,
      "loss": 0.4223,
      "step": 13370
    },
    {
      "epoch": 0.6136019457574228,
      "grad_norm": 0.4279305636882782,
      "learning_rate": 9.150413529707053e-06,
      "loss": 0.3603,
      "step": 13371
    },
    {
      "epoch": 0.6136478362626773,
      "grad_norm": 0.49269410967826843,
      "learning_rate": 9.150276798059456e-06,
      "loss": 0.4214,
      "step": 13372
    },
    {
      "epoch": 0.6136937267679318,
      "grad_norm": 0.44942909479141235,
      "learning_rate": 9.150140056431756e-06,
      "loss": 0.3561,
      "step": 13373
    },
    {
      "epoch": 0.6137396172731862,
      "grad_norm": 0.44858014583587646,
      "learning_rate": 9.150003304824278e-06,
      "loss": 0.3724,
      "step": 13374
    },
    {
      "epoch": 0.6137855077784407,
      "grad_norm": 0.4494982063770294,
      "learning_rate": 9.14986654323735e-06,
      "loss": 0.4124,
      "step": 13375
    },
    {
      "epoch": 0.6138313982836952,
      "grad_norm": 0.47216880321502686,
      "learning_rate": 9.149729771671304e-06,
      "loss": 0.3985,
      "step": 13376
    },
    {
      "epoch": 0.6138772887889495,
      "grad_norm": 0.4310190677642822,
      "learning_rate": 9.149592990126468e-06,
      "loss": 0.3818,
      "step": 13377
    },
    {
      "epoch": 0.613923179294204,
      "grad_norm": 0.446488618850708,
      "learning_rate": 9.149456198603169e-06,
      "loss": 0.3549,
      "step": 13378
    },
    {
      "epoch": 0.6139690697994585,
      "grad_norm": 0.4528619050979614,
      "learning_rate": 9.149319397101739e-06,
      "loss": 0.3739,
      "step": 13379
    },
    {
      "epoch": 0.6140149603047129,
      "grad_norm": 0.4326205253601074,
      "learning_rate": 9.149182585622504e-06,
      "loss": 0.3846,
      "step": 13380
    },
    {
      "epoch": 0.6140608508099674,
      "grad_norm": 0.4209747314453125,
      "learning_rate": 9.149045764165796e-06,
      "loss": 0.3718,
      "step": 13381
    },
    {
      "epoch": 0.6141067413152219,
      "grad_norm": 0.43557941913604736,
      "learning_rate": 9.148908932731941e-06,
      "loss": 0.3653,
      "step": 13382
    },
    {
      "epoch": 0.6141526318204763,
      "grad_norm": 0.5107849836349487,
      "learning_rate": 9.14877209132127e-06,
      "loss": 0.4576,
      "step": 13383
    },
    {
      "epoch": 0.6141985223257308,
      "grad_norm": 0.5057132840156555,
      "learning_rate": 9.148635239934112e-06,
      "loss": 0.4478,
      "step": 13384
    },
    {
      "epoch": 0.6142444128309853,
      "grad_norm": 0.4581402838230133,
      "learning_rate": 9.148498378570796e-06,
      "loss": 0.4345,
      "step": 13385
    },
    {
      "epoch": 0.6142903033362397,
      "grad_norm": 0.4547920227050781,
      "learning_rate": 9.148361507231649e-06,
      "loss": 0.3676,
      "step": 13386
    },
    {
      "epoch": 0.6143361938414942,
      "grad_norm": 0.4942670166492462,
      "learning_rate": 9.148224625917003e-06,
      "loss": 0.5,
      "step": 13387
    },
    {
      "epoch": 0.6143820843467487,
      "grad_norm": 0.4619365930557251,
      "learning_rate": 9.148087734627185e-06,
      "loss": 0.4184,
      "step": 13388
    },
    {
      "epoch": 0.6144279748520031,
      "grad_norm": 0.45077359676361084,
      "learning_rate": 9.147950833362526e-06,
      "loss": 0.4329,
      "step": 13389
    },
    {
      "epoch": 0.6144738653572576,
      "grad_norm": 0.46725034713745117,
      "learning_rate": 9.147813922123354e-06,
      "loss": 0.4685,
      "step": 13390
    },
    {
      "epoch": 0.6145197558625121,
      "grad_norm": 0.4426793158054352,
      "learning_rate": 9.14767700091e-06,
      "loss": 0.4055,
      "step": 13391
    },
    {
      "epoch": 0.6145656463677666,
      "grad_norm": 0.477081835269928,
      "learning_rate": 9.14754006972279e-06,
      "loss": 0.4882,
      "step": 13392
    },
    {
      "epoch": 0.614611536873021,
      "grad_norm": 0.44980233907699585,
      "learning_rate": 9.147403128562056e-06,
      "loss": 0.3851,
      "step": 13393
    },
    {
      "epoch": 0.6146574273782754,
      "grad_norm": 0.45756906270980835,
      "learning_rate": 9.147266177428127e-06,
      "loss": 0.4074,
      "step": 13394
    },
    {
      "epoch": 0.6147033178835299,
      "grad_norm": 0.45969247817993164,
      "learning_rate": 9.14712921632133e-06,
      "loss": 0.4428,
      "step": 13395
    },
    {
      "epoch": 0.6147492083887843,
      "grad_norm": 0.4489195644855499,
      "learning_rate": 9.146992245241999e-06,
      "loss": 0.411,
      "step": 13396
    },
    {
      "epoch": 0.6147950988940388,
      "grad_norm": 0.41751763224601746,
      "learning_rate": 9.146855264190458e-06,
      "loss": 0.3479,
      "step": 13397
    },
    {
      "epoch": 0.6148409893992933,
      "grad_norm": 0.49162569642066956,
      "learning_rate": 9.146718273167039e-06,
      "loss": 0.4783,
      "step": 13398
    },
    {
      "epoch": 0.6148868799045477,
      "grad_norm": 0.5050580501556396,
      "learning_rate": 9.146581272172072e-06,
      "loss": 0.4207,
      "step": 13399
    },
    {
      "epoch": 0.6149327704098022,
      "grad_norm": 0.47160983085632324,
      "learning_rate": 9.146444261205884e-06,
      "loss": 0.4591,
      "step": 13400
    },
    {
      "epoch": 0.6149786609150567,
      "grad_norm": 0.45013126730918884,
      "learning_rate": 9.146307240268808e-06,
      "loss": 0.3408,
      "step": 13401
    },
    {
      "epoch": 0.6150245514203111,
      "grad_norm": 0.46886682510375977,
      "learning_rate": 9.146170209361171e-06,
      "loss": 0.3841,
      "step": 13402
    },
    {
      "epoch": 0.6150704419255656,
      "grad_norm": 0.4279814064502716,
      "learning_rate": 9.146033168483305e-06,
      "loss": 0.339,
      "step": 13403
    },
    {
      "epoch": 0.6151163324308201,
      "grad_norm": 0.46263134479522705,
      "learning_rate": 9.145896117635537e-06,
      "loss": 0.4046,
      "step": 13404
    },
    {
      "epoch": 0.6151622229360745,
      "grad_norm": 0.43564674258232117,
      "learning_rate": 9.145759056818195e-06,
      "loss": 0.3551,
      "step": 13405
    },
    {
      "epoch": 0.615208113441329,
      "grad_norm": 0.48240020871162415,
      "learning_rate": 9.145621986031615e-06,
      "loss": 0.3635,
      "step": 13406
    },
    {
      "epoch": 0.6152540039465835,
      "grad_norm": 0.445943146944046,
      "learning_rate": 9.145484905276119e-06,
      "loss": 0.3919,
      "step": 13407
    },
    {
      "epoch": 0.615299894451838,
      "grad_norm": 0.4761308431625366,
      "learning_rate": 9.145347814552041e-06,
      "loss": 0.4406,
      "step": 13408
    },
    {
      "epoch": 0.6153457849570924,
      "grad_norm": 0.4697396755218506,
      "learning_rate": 9.14521071385971e-06,
      "loss": 0.4571,
      "step": 13409
    },
    {
      "epoch": 0.6153916754623469,
      "grad_norm": 0.4474799931049347,
      "learning_rate": 9.145073603199456e-06,
      "loss": 0.3616,
      "step": 13410
    },
    {
      "epoch": 0.6154375659676014,
      "grad_norm": 0.4395371973514557,
      "learning_rate": 9.144936482571607e-06,
      "loss": 0.3648,
      "step": 13411
    },
    {
      "epoch": 0.6154834564728557,
      "grad_norm": 0.432668000459671,
      "learning_rate": 9.144799351976496e-06,
      "loss": 0.3309,
      "step": 13412
    },
    {
      "epoch": 0.6155293469781102,
      "grad_norm": 0.5442088842391968,
      "learning_rate": 9.144662211414448e-06,
      "loss": 0.519,
      "step": 13413
    },
    {
      "epoch": 0.6155752374833647,
      "grad_norm": 0.4398118257522583,
      "learning_rate": 9.144525060885798e-06,
      "loss": 0.3592,
      "step": 13414
    },
    {
      "epoch": 0.6156211279886191,
      "grad_norm": 0.48946818709373474,
      "learning_rate": 9.144387900390872e-06,
      "loss": 0.4304,
      "step": 13415
    },
    {
      "epoch": 0.6156670184938736,
      "grad_norm": 0.4541175067424774,
      "learning_rate": 9.144250729930002e-06,
      "loss": 0.3945,
      "step": 13416
    },
    {
      "epoch": 0.6157129089991281,
      "grad_norm": 0.505452573299408,
      "learning_rate": 9.144113549503516e-06,
      "loss": 0.4431,
      "step": 13417
    },
    {
      "epoch": 0.6157587995043825,
      "grad_norm": 0.4383891522884369,
      "learning_rate": 9.143976359111744e-06,
      "loss": 0.4418,
      "step": 13418
    },
    {
      "epoch": 0.615804690009637,
      "grad_norm": 0.4019473195075989,
      "learning_rate": 9.143839158755018e-06,
      "loss": 0.2872,
      "step": 13419
    },
    {
      "epoch": 0.6158505805148915,
      "grad_norm": 0.47392404079437256,
      "learning_rate": 9.143701948433668e-06,
      "loss": 0.4154,
      "step": 13420
    },
    {
      "epoch": 0.6158964710201459,
      "grad_norm": 0.49308133125305176,
      "learning_rate": 9.14356472814802e-06,
      "loss": 0.4972,
      "step": 13421
    },
    {
      "epoch": 0.6159423615254004,
      "grad_norm": 0.5038841366767883,
      "learning_rate": 9.143427497898408e-06,
      "loss": 0.4542,
      "step": 13422
    },
    {
      "epoch": 0.6159882520306549,
      "grad_norm": 0.4757876694202423,
      "learning_rate": 9.14329025768516e-06,
      "loss": 0.433,
      "step": 13423
    },
    {
      "epoch": 0.6160341425359093,
      "grad_norm": 0.44435879588127136,
      "learning_rate": 9.143153007508607e-06,
      "loss": 0.35,
      "step": 13424
    },
    {
      "epoch": 0.6160800330411638,
      "grad_norm": 0.4811384677886963,
      "learning_rate": 9.14301574736908e-06,
      "loss": 0.4689,
      "step": 13425
    },
    {
      "epoch": 0.6161259235464183,
      "grad_norm": 0.4709923267364502,
      "learning_rate": 9.142878477266906e-06,
      "loss": 0.4225,
      "step": 13426
    },
    {
      "epoch": 0.6161718140516728,
      "grad_norm": 0.4372086524963379,
      "learning_rate": 9.142741197202416e-06,
      "loss": 0.3423,
      "step": 13427
    },
    {
      "epoch": 0.6162177045569271,
      "grad_norm": 0.4705607295036316,
      "learning_rate": 9.142603907175942e-06,
      "loss": 0.3895,
      "step": 13428
    },
    {
      "epoch": 0.6162635950621816,
      "grad_norm": 0.41223546862602234,
      "learning_rate": 9.142466607187812e-06,
      "loss": 0.3212,
      "step": 13429
    },
    {
      "epoch": 0.6163094855674361,
      "grad_norm": 0.4744665324687958,
      "learning_rate": 9.142329297238359e-06,
      "loss": 0.4495,
      "step": 13430
    },
    {
      "epoch": 0.6163553760726905,
      "grad_norm": 0.45777949690818787,
      "learning_rate": 9.14219197732791e-06,
      "loss": 0.4456,
      "step": 13431
    },
    {
      "epoch": 0.616401266577945,
      "grad_norm": 0.4812173843383789,
      "learning_rate": 9.142054647456797e-06,
      "loss": 0.4251,
      "step": 13432
    },
    {
      "epoch": 0.6164471570831995,
      "grad_norm": 0.46591857075691223,
      "learning_rate": 9.14191730762535e-06,
      "loss": 0.4481,
      "step": 13433
    },
    {
      "epoch": 0.6164930475884539,
      "grad_norm": 0.40645402669906616,
      "learning_rate": 9.1417799578339e-06,
      "loss": 0.3316,
      "step": 13434
    },
    {
      "epoch": 0.6165389380937084,
      "grad_norm": 0.4581991136074066,
      "learning_rate": 9.141642598082775e-06,
      "loss": 0.3956,
      "step": 13435
    },
    {
      "epoch": 0.6165848285989629,
      "grad_norm": 0.46937936544418335,
      "learning_rate": 9.141505228372308e-06,
      "loss": 0.4868,
      "step": 13436
    },
    {
      "epoch": 0.6166307191042173,
      "grad_norm": 0.44141504168510437,
      "learning_rate": 9.141367848702827e-06,
      "loss": 0.3926,
      "step": 13437
    },
    {
      "epoch": 0.6166766096094718,
      "grad_norm": 0.46020522713661194,
      "learning_rate": 9.141230459074663e-06,
      "loss": 0.4207,
      "step": 13438
    },
    {
      "epoch": 0.6167225001147263,
      "grad_norm": 0.46896132826805115,
      "learning_rate": 9.141093059488148e-06,
      "loss": 0.4405,
      "step": 13439
    },
    {
      "epoch": 0.6167683906199807,
      "grad_norm": 0.46983203291893005,
      "learning_rate": 9.14095564994361e-06,
      "loss": 0.4496,
      "step": 13440
    },
    {
      "epoch": 0.6168142811252352,
      "grad_norm": 0.480518639087677,
      "learning_rate": 9.140818230441381e-06,
      "loss": 0.4331,
      "step": 13441
    },
    {
      "epoch": 0.6168601716304897,
      "grad_norm": 0.49814900755882263,
      "learning_rate": 9.140680800981791e-06,
      "loss": 0.4712,
      "step": 13442
    },
    {
      "epoch": 0.6169060621357441,
      "grad_norm": 0.44378945231437683,
      "learning_rate": 9.140543361565171e-06,
      "loss": 0.3667,
      "step": 13443
    },
    {
      "epoch": 0.6169519526409986,
      "grad_norm": 0.41307932138442993,
      "learning_rate": 9.14040591219185e-06,
      "loss": 0.3258,
      "step": 13444
    },
    {
      "epoch": 0.616997843146253,
      "grad_norm": 0.44544902443885803,
      "learning_rate": 9.140268452862159e-06,
      "loss": 0.371,
      "step": 13445
    },
    {
      "epoch": 0.6170437336515076,
      "grad_norm": 0.4458356201648712,
      "learning_rate": 9.140130983576432e-06,
      "loss": 0.3614,
      "step": 13446
    },
    {
      "epoch": 0.6170896241567619,
      "grad_norm": 0.5368770360946655,
      "learning_rate": 9.139993504334995e-06,
      "loss": 0.3311,
      "step": 13447
    },
    {
      "epoch": 0.6171355146620164,
      "grad_norm": 0.4459013044834137,
      "learning_rate": 9.13985601513818e-06,
      "loss": 0.3881,
      "step": 13448
    },
    {
      "epoch": 0.6171814051672709,
      "grad_norm": 0.4980817139148712,
      "learning_rate": 9.139718515986318e-06,
      "loss": 0.451,
      "step": 13449
    },
    {
      "epoch": 0.6172272956725253,
      "grad_norm": 0.44086116552352905,
      "learning_rate": 9.13958100687974e-06,
      "loss": 0.3691,
      "step": 13450
    },
    {
      "epoch": 0.6172731861777798,
      "grad_norm": 0.4244758188724518,
      "learning_rate": 9.139443487818775e-06,
      "loss": 0.328,
      "step": 13451
    },
    {
      "epoch": 0.6173190766830343,
      "grad_norm": 0.4575919508934021,
      "learning_rate": 9.139305958803755e-06,
      "loss": 0.4326,
      "step": 13452
    },
    {
      "epoch": 0.6173649671882887,
      "grad_norm": 0.4396938979625702,
      "learning_rate": 9.139168419835012e-06,
      "loss": 0.3601,
      "step": 13453
    },
    {
      "epoch": 0.6174108576935432,
      "grad_norm": 0.4686066508293152,
      "learning_rate": 9.139030870912875e-06,
      "loss": 0.4289,
      "step": 13454
    },
    {
      "epoch": 0.6174567481987977,
      "grad_norm": 0.4795283377170563,
      "learning_rate": 9.138893312037674e-06,
      "loss": 0.4225,
      "step": 13455
    },
    {
      "epoch": 0.6175026387040521,
      "grad_norm": 0.4413226842880249,
      "learning_rate": 9.13875574320974e-06,
      "loss": 0.3551,
      "step": 13456
    },
    {
      "epoch": 0.6175485292093066,
      "grad_norm": 0.43298423290252686,
      "learning_rate": 9.138618164429407e-06,
      "loss": 0.3673,
      "step": 13457
    },
    {
      "epoch": 0.6175944197145611,
      "grad_norm": 0.4431157112121582,
      "learning_rate": 9.138480575697002e-06,
      "loss": 0.2969,
      "step": 13458
    },
    {
      "epoch": 0.6176403102198155,
      "grad_norm": 0.44631248712539673,
      "learning_rate": 9.138342977012859e-06,
      "loss": 0.3797,
      "step": 13459
    },
    {
      "epoch": 0.61768620072507,
      "grad_norm": 0.39273110032081604,
      "learning_rate": 9.138205368377306e-06,
      "loss": 0.2929,
      "step": 13460
    },
    {
      "epoch": 0.6177320912303245,
      "grad_norm": 0.46222198009490967,
      "learning_rate": 9.138067749790674e-06,
      "loss": 0.372,
      "step": 13461
    },
    {
      "epoch": 0.617777981735579,
      "grad_norm": 0.4541102945804596,
      "learning_rate": 9.137930121253296e-06,
      "loss": 0.376,
      "step": 13462
    },
    {
      "epoch": 0.6178238722408333,
      "grad_norm": 0.49308663606643677,
      "learning_rate": 9.137792482765504e-06,
      "loss": 0.4419,
      "step": 13463
    },
    {
      "epoch": 0.6178697627460878,
      "grad_norm": 0.4454323351383209,
      "learning_rate": 9.137654834327626e-06,
      "loss": 0.416,
      "step": 13464
    },
    {
      "epoch": 0.6179156532513423,
      "grad_norm": 0.48245516419410706,
      "learning_rate": 9.137517175939992e-06,
      "loss": 0.4506,
      "step": 13465
    },
    {
      "epoch": 0.6179615437565967,
      "grad_norm": 0.44801098108291626,
      "learning_rate": 9.137379507602938e-06,
      "loss": 0.4178,
      "step": 13466
    },
    {
      "epoch": 0.6180074342618512,
      "grad_norm": 0.4346368610858917,
      "learning_rate": 9.137241829316789e-06,
      "loss": 0.3134,
      "step": 13467
    },
    {
      "epoch": 0.6180533247671057,
      "grad_norm": 0.44964346289634705,
      "learning_rate": 9.13710414108188e-06,
      "loss": 0.3548,
      "step": 13468
    },
    {
      "epoch": 0.6180992152723601,
      "grad_norm": 0.44593289494514465,
      "learning_rate": 9.136966442898544e-06,
      "loss": 0.3955,
      "step": 13469
    },
    {
      "epoch": 0.6181451057776146,
      "grad_norm": 0.45217379927635193,
      "learning_rate": 9.136828734767104e-06,
      "loss": 0.3361,
      "step": 13470
    },
    {
      "epoch": 0.6181909962828691,
      "grad_norm": 0.4572787880897522,
      "learning_rate": 9.136691016687901e-06,
      "loss": 0.3448,
      "step": 13471
    },
    {
      "epoch": 0.6182368867881235,
      "grad_norm": 0.4399280846118927,
      "learning_rate": 9.13655328866126e-06,
      "loss": 0.3578,
      "step": 13472
    },
    {
      "epoch": 0.618282777293378,
      "grad_norm": 0.41310587525367737,
      "learning_rate": 9.136415550687514e-06,
      "loss": 0.3396,
      "step": 13473
    },
    {
      "epoch": 0.6183286677986325,
      "grad_norm": 0.4077087640762329,
      "learning_rate": 9.136277802766995e-06,
      "loss": 0.3127,
      "step": 13474
    },
    {
      "epoch": 0.6183745583038869,
      "grad_norm": 0.51629638671875,
      "learning_rate": 9.136140044900032e-06,
      "loss": 0.4296,
      "step": 13475
    },
    {
      "epoch": 0.6184204488091414,
      "grad_norm": 0.45166507363319397,
      "learning_rate": 9.13600227708696e-06,
      "loss": 0.3942,
      "step": 13476
    },
    {
      "epoch": 0.6184663393143959,
      "grad_norm": 0.45083045959472656,
      "learning_rate": 9.135864499328106e-06,
      "loss": 0.394,
      "step": 13477
    },
    {
      "epoch": 0.6185122298196503,
      "grad_norm": 0.4511651396751404,
      "learning_rate": 9.135726711623803e-06,
      "loss": 0.3556,
      "step": 13478
    },
    {
      "epoch": 0.6185581203249048,
      "grad_norm": 0.4588184356689453,
      "learning_rate": 9.135588913974383e-06,
      "loss": 0.4425,
      "step": 13479
    },
    {
      "epoch": 0.6186040108301593,
      "grad_norm": 0.5027110576629639,
      "learning_rate": 9.135451106380176e-06,
      "loss": 0.4816,
      "step": 13480
    },
    {
      "epoch": 0.6186499013354138,
      "grad_norm": 0.4566195011138916,
      "learning_rate": 9.135313288841516e-06,
      "loss": 0.4289,
      "step": 13481
    },
    {
      "epoch": 0.6186957918406681,
      "grad_norm": 0.4354590177536011,
      "learning_rate": 9.135175461358731e-06,
      "loss": 0.3499,
      "step": 13482
    },
    {
      "epoch": 0.6187416823459226,
      "grad_norm": 0.485817551612854,
      "learning_rate": 9.135037623932154e-06,
      "loss": 0.4534,
      "step": 13483
    },
    {
      "epoch": 0.6187875728511771,
      "grad_norm": 0.4408957064151764,
      "learning_rate": 9.134899776562119e-06,
      "loss": 0.3638,
      "step": 13484
    },
    {
      "epoch": 0.6188334633564315,
      "grad_norm": 0.45121341943740845,
      "learning_rate": 9.134761919248953e-06,
      "loss": 0.3676,
      "step": 13485
    },
    {
      "epoch": 0.618879353861686,
      "grad_norm": 0.5158239006996155,
      "learning_rate": 9.134624051992991e-06,
      "loss": 0.5162,
      "step": 13486
    },
    {
      "epoch": 0.6189252443669405,
      "grad_norm": 0.48907721042633057,
      "learning_rate": 9.134486174794561e-06,
      "loss": 0.4788,
      "step": 13487
    },
    {
      "epoch": 0.6189711348721949,
      "grad_norm": 0.456527441740036,
      "learning_rate": 9.134348287653998e-06,
      "loss": 0.317,
      "step": 13488
    },
    {
      "epoch": 0.6190170253774494,
      "grad_norm": 0.5023139715194702,
      "learning_rate": 9.134210390571632e-06,
      "loss": 0.4736,
      "step": 13489
    },
    {
      "epoch": 0.6190629158827039,
      "grad_norm": 0.48126810789108276,
      "learning_rate": 9.134072483547794e-06,
      "loss": 0.4275,
      "step": 13490
    },
    {
      "epoch": 0.6191088063879583,
      "grad_norm": 0.49952611327171326,
      "learning_rate": 9.133934566582819e-06,
      "loss": 0.4731,
      "step": 13491
    },
    {
      "epoch": 0.6191546968932128,
      "grad_norm": 0.44755539298057556,
      "learning_rate": 9.133796639677033e-06,
      "loss": 0.3962,
      "step": 13492
    },
    {
      "epoch": 0.6192005873984673,
      "grad_norm": 0.46768897771835327,
      "learning_rate": 9.133658702830772e-06,
      "loss": 0.4181,
      "step": 13493
    },
    {
      "epoch": 0.6192464779037217,
      "grad_norm": 0.48543643951416016,
      "learning_rate": 9.133520756044368e-06,
      "loss": 0.4425,
      "step": 13494
    },
    {
      "epoch": 0.6192923684089762,
      "grad_norm": 0.44651708006858826,
      "learning_rate": 9.133382799318149e-06,
      "loss": 0.3988,
      "step": 13495
    },
    {
      "epoch": 0.6193382589142307,
      "grad_norm": 0.46467939019203186,
      "learning_rate": 9.133244832652448e-06,
      "loss": 0.4005,
      "step": 13496
    },
    {
      "epoch": 0.6193841494194852,
      "grad_norm": 0.4582372307777405,
      "learning_rate": 9.1331068560476e-06,
      "loss": 0.369,
      "step": 13497
    },
    {
      "epoch": 0.6194300399247396,
      "grad_norm": 0.4708572030067444,
      "learning_rate": 9.132968869503933e-06,
      "loss": 0.4574,
      "step": 13498
    },
    {
      "epoch": 0.619475930429994,
      "grad_norm": 0.43059664964675903,
      "learning_rate": 9.132830873021782e-06,
      "loss": 0.3366,
      "step": 13499
    },
    {
      "epoch": 0.6195218209352485,
      "grad_norm": 0.4699653387069702,
      "learning_rate": 9.132692866601476e-06,
      "loss": 0.4237,
      "step": 13500
    },
    {
      "epoch": 0.6195677114405029,
      "grad_norm": 0.4363558888435364,
      "learning_rate": 9.132554850243347e-06,
      "loss": 0.3363,
      "step": 13501
    },
    {
      "epoch": 0.6196136019457574,
      "grad_norm": 0.4825230836868286,
      "learning_rate": 9.13241682394773e-06,
      "loss": 0.3787,
      "step": 13502
    },
    {
      "epoch": 0.6196594924510119,
      "grad_norm": 0.48228612542152405,
      "learning_rate": 9.132278787714953e-06,
      "loss": 0.4631,
      "step": 13503
    },
    {
      "epoch": 0.6197053829562663,
      "grad_norm": 0.4433227777481079,
      "learning_rate": 9.13214074154535e-06,
      "loss": 0.3874,
      "step": 13504
    },
    {
      "epoch": 0.6197512734615208,
      "grad_norm": 0.45059722661972046,
      "learning_rate": 9.132002685439255e-06,
      "loss": 0.3824,
      "step": 13505
    },
    {
      "epoch": 0.6197971639667753,
      "grad_norm": 0.4203972816467285,
      "learning_rate": 9.131864619396994e-06,
      "loss": 0.3441,
      "step": 13506
    },
    {
      "epoch": 0.6198430544720297,
      "grad_norm": 0.45209014415740967,
      "learning_rate": 9.131726543418904e-06,
      "loss": 0.368,
      "step": 13507
    },
    {
      "epoch": 0.6198889449772842,
      "grad_norm": 0.47339892387390137,
      "learning_rate": 9.131588457505315e-06,
      "loss": 0.4602,
      "step": 13508
    },
    {
      "epoch": 0.6199348354825387,
      "grad_norm": 0.440996378660202,
      "learning_rate": 9.131450361656562e-06,
      "loss": 0.3785,
      "step": 13509
    },
    {
      "epoch": 0.6199807259877931,
      "grad_norm": 0.47609198093414307,
      "learning_rate": 9.131312255872973e-06,
      "loss": 0.4478,
      "step": 13510
    },
    {
      "epoch": 0.6200266164930476,
      "grad_norm": 0.47027817368507385,
      "learning_rate": 9.131174140154883e-06,
      "loss": 0.4089,
      "step": 13511
    },
    {
      "epoch": 0.6200725069983021,
      "grad_norm": 0.47962456941604614,
      "learning_rate": 9.131036014502622e-06,
      "loss": 0.477,
      "step": 13512
    },
    {
      "epoch": 0.6201183975035565,
      "grad_norm": 0.45942842960357666,
      "learning_rate": 9.130897878916524e-06,
      "loss": 0.3986,
      "step": 13513
    },
    {
      "epoch": 0.620164288008811,
      "grad_norm": 0.4236631989479065,
      "learning_rate": 9.130759733396918e-06,
      "loss": 0.3908,
      "step": 13514
    },
    {
      "epoch": 0.6202101785140655,
      "grad_norm": 0.4336707293987274,
      "learning_rate": 9.130621577944144e-06,
      "loss": 0.3623,
      "step": 13515
    },
    {
      "epoch": 0.62025606901932,
      "grad_norm": 0.43616944551467896,
      "learning_rate": 9.130483412558524e-06,
      "loss": 0.3456,
      "step": 13516
    },
    {
      "epoch": 0.6203019595245743,
      "grad_norm": 0.44043534994125366,
      "learning_rate": 9.130345237240396e-06,
      "loss": 0.4436,
      "step": 13517
    },
    {
      "epoch": 0.6203478500298288,
      "grad_norm": 0.43045178055763245,
      "learning_rate": 9.130207051990091e-06,
      "loss": 0.3828,
      "step": 13518
    },
    {
      "epoch": 0.6203937405350833,
      "grad_norm": 0.4150160253047943,
      "learning_rate": 9.130068856807943e-06,
      "loss": 0.3233,
      "step": 13519
    },
    {
      "epoch": 0.6204396310403377,
      "grad_norm": 0.4437844753265381,
      "learning_rate": 9.12993065169428e-06,
      "loss": 0.3986,
      "step": 13520
    },
    {
      "epoch": 0.6204855215455922,
      "grad_norm": 0.4456350803375244,
      "learning_rate": 9.12979243664944e-06,
      "loss": 0.4179,
      "step": 13521
    },
    {
      "epoch": 0.6205314120508467,
      "grad_norm": 0.4221799969673157,
      "learning_rate": 9.129654211673753e-06,
      "loss": 0.3547,
      "step": 13522
    },
    {
      "epoch": 0.6205773025561011,
      "grad_norm": 0.39722132682800293,
      "learning_rate": 9.129515976767548e-06,
      "loss": 0.2889,
      "step": 13523
    },
    {
      "epoch": 0.6206231930613556,
      "grad_norm": 0.46047040820121765,
      "learning_rate": 9.129377731931164e-06,
      "loss": 0.3442,
      "step": 13524
    },
    {
      "epoch": 0.6206690835666101,
      "grad_norm": 0.4540072977542877,
      "learning_rate": 9.129239477164927e-06,
      "loss": 0.3938,
      "step": 13525
    },
    {
      "epoch": 0.6207149740718645,
      "grad_norm": 0.4674703776836395,
      "learning_rate": 9.129101212469173e-06,
      "loss": 0.4126,
      "step": 13526
    },
    {
      "epoch": 0.620760864577119,
      "grad_norm": 0.4764498174190521,
      "learning_rate": 9.128962937844235e-06,
      "loss": 0.4466,
      "step": 13527
    },
    {
      "epoch": 0.6208067550823735,
      "grad_norm": 0.4278314411640167,
      "learning_rate": 9.128824653290443e-06,
      "loss": 0.3777,
      "step": 13528
    },
    {
      "epoch": 0.6208526455876279,
      "grad_norm": 0.48098185658454895,
      "learning_rate": 9.128686358808131e-06,
      "loss": 0.4257,
      "step": 13529
    },
    {
      "epoch": 0.6208985360928824,
      "grad_norm": 0.4629579484462738,
      "learning_rate": 9.128548054397632e-06,
      "loss": 0.3991,
      "step": 13530
    },
    {
      "epoch": 0.6209444265981369,
      "grad_norm": 0.4307628571987152,
      "learning_rate": 9.128409740059276e-06,
      "loss": 0.3492,
      "step": 13531
    },
    {
      "epoch": 0.6209903171033913,
      "grad_norm": 0.4825531244277954,
      "learning_rate": 9.1282714157934e-06,
      "loss": 0.4488,
      "step": 13532
    },
    {
      "epoch": 0.6210362076086458,
      "grad_norm": 0.42392441630363464,
      "learning_rate": 9.128133081600332e-06,
      "loss": 0.3785,
      "step": 13533
    },
    {
      "epoch": 0.6210820981139002,
      "grad_norm": 0.48532333970069885,
      "learning_rate": 9.12799473748041e-06,
      "loss": 0.4709,
      "step": 13534
    },
    {
      "epoch": 0.6211279886191547,
      "grad_norm": 0.4688650667667389,
      "learning_rate": 9.12785638343396e-06,
      "loss": 0.4506,
      "step": 13535
    },
    {
      "epoch": 0.6211738791244091,
      "grad_norm": 0.4706922471523285,
      "learning_rate": 9.12771801946132e-06,
      "loss": 0.46,
      "step": 13536
    },
    {
      "epoch": 0.6212197696296636,
      "grad_norm": 0.4879792332649231,
      "learning_rate": 9.127579645562821e-06,
      "loss": 0.5071,
      "step": 13537
    },
    {
      "epoch": 0.6212656601349181,
      "grad_norm": 0.4436523914337158,
      "learning_rate": 9.127441261738794e-06,
      "loss": 0.3738,
      "step": 13538
    },
    {
      "epoch": 0.6213115506401725,
      "grad_norm": 0.47871869802474976,
      "learning_rate": 9.127302867989576e-06,
      "loss": 0.4666,
      "step": 13539
    },
    {
      "epoch": 0.621357441145427,
      "grad_norm": 0.4736569821834564,
      "learning_rate": 9.127164464315494e-06,
      "loss": 0.4248,
      "step": 13540
    },
    {
      "epoch": 0.6214033316506815,
      "grad_norm": 0.4707769751548767,
      "learning_rate": 9.127026050716887e-06,
      "loss": 0.4659,
      "step": 13541
    },
    {
      "epoch": 0.6214492221559359,
      "grad_norm": 0.49210190773010254,
      "learning_rate": 9.126887627194084e-06,
      "loss": 0.443,
      "step": 13542
    },
    {
      "epoch": 0.6214951126611904,
      "grad_norm": 0.47874125838279724,
      "learning_rate": 9.126749193747419e-06,
      "loss": 0.4074,
      "step": 13543
    },
    {
      "epoch": 0.6215410031664449,
      "grad_norm": 0.4750124514102936,
      "learning_rate": 9.126610750377222e-06,
      "loss": 0.4295,
      "step": 13544
    },
    {
      "epoch": 0.6215868936716993,
      "grad_norm": 0.48303088545799255,
      "learning_rate": 9.126472297083833e-06,
      "loss": 0.4154,
      "step": 13545
    },
    {
      "epoch": 0.6216327841769538,
      "grad_norm": 0.48089855909347534,
      "learning_rate": 9.126333833867578e-06,
      "loss": 0.4946,
      "step": 13546
    },
    {
      "epoch": 0.6216786746822083,
      "grad_norm": 0.44583579897880554,
      "learning_rate": 9.126195360728791e-06,
      "loss": 0.3699,
      "step": 13547
    },
    {
      "epoch": 0.6217245651874627,
      "grad_norm": 0.46115127205848694,
      "learning_rate": 9.126056877667808e-06,
      "loss": 0.37,
      "step": 13548
    },
    {
      "epoch": 0.6217704556927172,
      "grad_norm": 0.4346621036529541,
      "learning_rate": 9.12591838468496e-06,
      "loss": 0.3509,
      "step": 13549
    },
    {
      "epoch": 0.6218163461979717,
      "grad_norm": 0.43880918622016907,
      "learning_rate": 9.12577988178058e-06,
      "loss": 0.3731,
      "step": 13550
    },
    {
      "epoch": 0.6218622367032262,
      "grad_norm": 0.4512544870376587,
      "learning_rate": 9.125641368955004e-06,
      "loss": 0.3658,
      "step": 13551
    },
    {
      "epoch": 0.6219081272084805,
      "grad_norm": 0.4697294533252716,
      "learning_rate": 9.12550284620856e-06,
      "loss": 0.4098,
      "step": 13552
    },
    {
      "epoch": 0.621954017713735,
      "grad_norm": 0.4583834111690521,
      "learning_rate": 9.125364313541586e-06,
      "loss": 0.4131,
      "step": 13553
    },
    {
      "epoch": 0.6219999082189895,
      "grad_norm": 0.43589186668395996,
      "learning_rate": 9.125225770954412e-06,
      "loss": 0.3181,
      "step": 13554
    },
    {
      "epoch": 0.6220457987242439,
      "grad_norm": 0.4776705801486969,
      "learning_rate": 9.12508721844737e-06,
      "loss": 0.4709,
      "step": 13555
    },
    {
      "epoch": 0.6220916892294984,
      "grad_norm": 0.44285672903060913,
      "learning_rate": 9.124948656020798e-06,
      "loss": 0.3752,
      "step": 13556
    },
    {
      "epoch": 0.6221375797347529,
      "grad_norm": 0.48146137595176697,
      "learning_rate": 9.124810083675024e-06,
      "loss": 0.3763,
      "step": 13557
    },
    {
      "epoch": 0.6221834702400073,
      "grad_norm": 0.42901986837387085,
      "learning_rate": 9.124671501410384e-06,
      "loss": 0.3293,
      "step": 13558
    },
    {
      "epoch": 0.6222293607452618,
      "grad_norm": 0.4728184640407562,
      "learning_rate": 9.124532909227214e-06,
      "loss": 0.3793,
      "step": 13559
    },
    {
      "epoch": 0.6222752512505163,
      "grad_norm": 0.4332300126552582,
      "learning_rate": 9.12439430712584e-06,
      "loss": 0.324,
      "step": 13560
    },
    {
      "epoch": 0.6223211417557707,
      "grad_norm": 0.4822970926761627,
      "learning_rate": 9.1242556951066e-06,
      "loss": 0.466,
      "step": 13561
    },
    {
      "epoch": 0.6223670322610252,
      "grad_norm": 0.4643024206161499,
      "learning_rate": 9.124117073169828e-06,
      "loss": 0.3782,
      "step": 13562
    },
    {
      "epoch": 0.6224129227662797,
      "grad_norm": 0.4630125164985657,
      "learning_rate": 9.123978441315855e-06,
      "loss": 0.3873,
      "step": 13563
    },
    {
      "epoch": 0.6224588132715341,
      "grad_norm": 0.4361162781715393,
      "learning_rate": 9.123839799545016e-06,
      "loss": 0.3889,
      "step": 13564
    },
    {
      "epoch": 0.6225047037767886,
      "grad_norm": 0.45569661259651184,
      "learning_rate": 9.123701147857644e-06,
      "loss": 0.3633,
      "step": 13565
    },
    {
      "epoch": 0.6225505942820431,
      "grad_norm": 0.48102834820747375,
      "learning_rate": 9.123562486254072e-06,
      "loss": 0.464,
      "step": 13566
    },
    {
      "epoch": 0.6225964847872975,
      "grad_norm": 0.4715704917907715,
      "learning_rate": 9.123423814734635e-06,
      "loss": 0.473,
      "step": 13567
    },
    {
      "epoch": 0.622642375292552,
      "grad_norm": 0.47909998893737793,
      "learning_rate": 9.123285133299664e-06,
      "loss": 0.4112,
      "step": 13568
    },
    {
      "epoch": 0.6226882657978065,
      "grad_norm": 0.4800023138523102,
      "learning_rate": 9.123146441949493e-06,
      "loss": 0.3939,
      "step": 13569
    },
    {
      "epoch": 0.622734156303061,
      "grad_norm": 0.4740374982357025,
      "learning_rate": 9.123007740684457e-06,
      "loss": 0.4785,
      "step": 13570
    },
    {
      "epoch": 0.6227800468083153,
      "grad_norm": 0.4531978964805603,
      "learning_rate": 9.122869029504887e-06,
      "loss": 0.4513,
      "step": 13571
    },
    {
      "epoch": 0.6228259373135698,
      "grad_norm": 0.4631929099559784,
      "learning_rate": 9.12273030841112e-06,
      "loss": 0.4951,
      "step": 13572
    },
    {
      "epoch": 0.6228718278188243,
      "grad_norm": 0.4883246123790741,
      "learning_rate": 9.122591577403488e-06,
      "loss": 0.4653,
      "step": 13573
    },
    {
      "epoch": 0.6229177183240787,
      "grad_norm": 0.4810955226421356,
      "learning_rate": 9.122452836482323e-06,
      "loss": 0.4904,
      "step": 13574
    },
    {
      "epoch": 0.6229636088293332,
      "grad_norm": 0.4530024826526642,
      "learning_rate": 9.12231408564796e-06,
      "loss": 0.4192,
      "step": 13575
    },
    {
      "epoch": 0.6230094993345877,
      "grad_norm": 0.4662466049194336,
      "learning_rate": 9.122175324900733e-06,
      "loss": 0.4455,
      "step": 13576
    },
    {
      "epoch": 0.6230553898398421,
      "grad_norm": 0.4660511016845703,
      "learning_rate": 9.122036554240976e-06,
      "loss": 0.4019,
      "step": 13577
    },
    {
      "epoch": 0.6231012803450966,
      "grad_norm": 0.5148074626922607,
      "learning_rate": 9.121897773669021e-06,
      "loss": 0.5202,
      "step": 13578
    },
    {
      "epoch": 0.6231471708503511,
      "grad_norm": 0.47557902336120605,
      "learning_rate": 9.121758983185204e-06,
      "loss": 0.3713,
      "step": 13579
    },
    {
      "epoch": 0.6231930613556055,
      "grad_norm": 0.5172755122184753,
      "learning_rate": 9.121620182789856e-06,
      "loss": 0.5078,
      "step": 13580
    },
    {
      "epoch": 0.62323895186086,
      "grad_norm": 0.4523959159851074,
      "learning_rate": 9.121481372483313e-06,
      "loss": 0.3695,
      "step": 13581
    },
    {
      "epoch": 0.6232848423661145,
      "grad_norm": 0.47998732328414917,
      "learning_rate": 9.121342552265909e-06,
      "loss": 0.467,
      "step": 13582
    },
    {
      "epoch": 0.6233307328713689,
      "grad_norm": 0.4421985447406769,
      "learning_rate": 9.121203722137976e-06,
      "loss": 0.3577,
      "step": 13583
    },
    {
      "epoch": 0.6233766233766234,
      "grad_norm": 0.48794248700141907,
      "learning_rate": 9.12106488209985e-06,
      "loss": 0.4482,
      "step": 13584
    },
    {
      "epoch": 0.6234225138818779,
      "grad_norm": 0.4165756106376648,
      "learning_rate": 9.120926032151862e-06,
      "loss": 0.3055,
      "step": 13585
    },
    {
      "epoch": 0.6234684043871324,
      "grad_norm": 0.4510299861431122,
      "learning_rate": 9.120787172294347e-06,
      "loss": 0.3997,
      "step": 13586
    },
    {
      "epoch": 0.6235142948923867,
      "grad_norm": 0.4689398407936096,
      "learning_rate": 9.120648302527641e-06,
      "loss": 0.423,
      "step": 13587
    },
    {
      "epoch": 0.6235601853976412,
      "grad_norm": 0.4426478445529938,
      "learning_rate": 9.120509422852076e-06,
      "loss": 0.366,
      "step": 13588
    },
    {
      "epoch": 0.6236060759028957,
      "grad_norm": 0.443268746137619,
      "learning_rate": 9.120370533267986e-06,
      "loss": 0.361,
      "step": 13589
    },
    {
      "epoch": 0.6236519664081501,
      "grad_norm": 0.42122969031333923,
      "learning_rate": 9.120231633775705e-06,
      "loss": 0.3699,
      "step": 13590
    },
    {
      "epoch": 0.6236978569134046,
      "grad_norm": 0.4692622423171997,
      "learning_rate": 9.120092724375567e-06,
      "loss": 0.3662,
      "step": 13591
    },
    {
      "epoch": 0.6237437474186591,
      "grad_norm": 0.440532386302948,
      "learning_rate": 9.119953805067907e-06,
      "loss": 0.3686,
      "step": 13592
    },
    {
      "epoch": 0.6237896379239135,
      "grad_norm": 0.45228350162506104,
      "learning_rate": 9.119814875853059e-06,
      "loss": 0.3998,
      "step": 13593
    },
    {
      "epoch": 0.623835528429168,
      "grad_norm": 0.4641745686531067,
      "learning_rate": 9.119675936731356e-06,
      "loss": 0.4433,
      "step": 13594
    },
    {
      "epoch": 0.6238814189344225,
      "grad_norm": 0.42702949047088623,
      "learning_rate": 9.119536987703132e-06,
      "loss": 0.3702,
      "step": 13595
    },
    {
      "epoch": 0.6239273094396769,
      "grad_norm": 0.4658576250076294,
      "learning_rate": 9.119398028768723e-06,
      "loss": 0.3977,
      "step": 13596
    },
    {
      "epoch": 0.6239731999449314,
      "grad_norm": 0.4700034558773041,
      "learning_rate": 9.11925905992846e-06,
      "loss": 0.4631,
      "step": 13597
    },
    {
      "epoch": 0.6240190904501859,
      "grad_norm": 0.4417860209941864,
      "learning_rate": 9.11912008118268e-06,
      "loss": 0.3801,
      "step": 13598
    },
    {
      "epoch": 0.6240649809554403,
      "grad_norm": 0.4522622227668762,
      "learning_rate": 9.118981092531717e-06,
      "loss": 0.4095,
      "step": 13599
    },
    {
      "epoch": 0.6241108714606948,
      "grad_norm": 0.4747748374938965,
      "learning_rate": 9.118842093975903e-06,
      "loss": 0.4609,
      "step": 13600
    },
    {
      "epoch": 0.6241567619659493,
      "grad_norm": 0.46969231963157654,
      "learning_rate": 9.118703085515575e-06,
      "loss": 0.4466,
      "step": 13601
    },
    {
      "epoch": 0.6242026524712037,
      "grad_norm": 0.4503541886806488,
      "learning_rate": 9.118564067151065e-06,
      "loss": 0.3771,
      "step": 13602
    },
    {
      "epoch": 0.6242485429764582,
      "grad_norm": 0.4588680565357208,
      "learning_rate": 9.11842503888271e-06,
      "loss": 0.4259,
      "step": 13603
    },
    {
      "epoch": 0.6242944334817127,
      "grad_norm": 0.45701706409454346,
      "learning_rate": 9.118286000710839e-06,
      "loss": 0.4672,
      "step": 13604
    },
    {
      "epoch": 0.6243403239869671,
      "grad_norm": 0.4905075430870056,
      "learning_rate": 9.118146952635793e-06,
      "loss": 0.4112,
      "step": 13605
    },
    {
      "epoch": 0.6243862144922215,
      "grad_norm": 0.4455539584159851,
      "learning_rate": 9.118007894657904e-06,
      "loss": 0.3967,
      "step": 13606
    },
    {
      "epoch": 0.624432104997476,
      "grad_norm": 0.4949384033679962,
      "learning_rate": 9.117868826777503e-06,
      "loss": 0.5338,
      "step": 13607
    },
    {
      "epoch": 0.6244779955027305,
      "grad_norm": 0.49051153659820557,
      "learning_rate": 9.117729748994928e-06,
      "loss": 0.4619,
      "step": 13608
    },
    {
      "epoch": 0.6245238860079849,
      "grad_norm": 0.46464529633522034,
      "learning_rate": 9.117590661310514e-06,
      "loss": 0.382,
      "step": 13609
    },
    {
      "epoch": 0.6245697765132394,
      "grad_norm": 0.4725222885608673,
      "learning_rate": 9.117451563724592e-06,
      "loss": 0.4114,
      "step": 13610
    },
    {
      "epoch": 0.6246156670184939,
      "grad_norm": 0.5443709492683411,
      "learning_rate": 9.1173124562375e-06,
      "loss": 0.5746,
      "step": 13611
    },
    {
      "epoch": 0.6246615575237483,
      "grad_norm": 0.42529138922691345,
      "learning_rate": 9.11717333884957e-06,
      "loss": 0.3011,
      "step": 13612
    },
    {
      "epoch": 0.6247074480290028,
      "grad_norm": 0.44743481278419495,
      "learning_rate": 9.117034211561138e-06,
      "loss": 0.3896,
      "step": 13613
    },
    {
      "epoch": 0.6247533385342573,
      "grad_norm": 0.43031972646713257,
      "learning_rate": 9.116895074372537e-06,
      "loss": 0.3718,
      "step": 13614
    },
    {
      "epoch": 0.6247992290395117,
      "grad_norm": 0.47878044843673706,
      "learning_rate": 9.116755927284104e-06,
      "loss": 0.4014,
      "step": 13615
    },
    {
      "epoch": 0.6248451195447662,
      "grad_norm": 0.4515795111656189,
      "learning_rate": 9.116616770296172e-06,
      "loss": 0.3964,
      "step": 13616
    },
    {
      "epoch": 0.6248910100500207,
      "grad_norm": 0.4825457036495209,
      "learning_rate": 9.116477603409075e-06,
      "loss": 0.4308,
      "step": 13617
    },
    {
      "epoch": 0.6249369005552751,
      "grad_norm": 0.4621758759021759,
      "learning_rate": 9.11633842662315e-06,
      "loss": 0.3712,
      "step": 13618
    },
    {
      "epoch": 0.6249827910605296,
      "grad_norm": 0.4357132315635681,
      "learning_rate": 9.11619923993873e-06,
      "loss": 0.3843,
      "step": 13619
    },
    {
      "epoch": 0.6250286815657841,
      "grad_norm": 0.4798409938812256,
      "learning_rate": 9.11606004335615e-06,
      "loss": 0.4475,
      "step": 13620
    },
    {
      "epoch": 0.6250745720710384,
      "grad_norm": 0.47069260478019714,
      "learning_rate": 9.115920836875745e-06,
      "loss": 0.4638,
      "step": 13621
    },
    {
      "epoch": 0.625120462576293,
      "grad_norm": 0.45995283126831055,
      "learning_rate": 9.115781620497848e-06,
      "loss": 0.4307,
      "step": 13622
    },
    {
      "epoch": 0.6251663530815474,
      "grad_norm": 0.4853038787841797,
      "learning_rate": 9.115642394222794e-06,
      "loss": 0.325,
      "step": 13623
    },
    {
      "epoch": 0.6252122435868019,
      "grad_norm": 0.45901229977607727,
      "learning_rate": 9.115503158050923e-06,
      "loss": 0.4347,
      "step": 13624
    },
    {
      "epoch": 0.6252581340920563,
      "grad_norm": 0.45606330037117004,
      "learning_rate": 9.115363911982563e-06,
      "loss": 0.3845,
      "step": 13625
    },
    {
      "epoch": 0.6253040245973108,
      "grad_norm": 0.492565393447876,
      "learning_rate": 9.115224656018052e-06,
      "loss": 0.4871,
      "step": 13626
    },
    {
      "epoch": 0.6253499151025653,
      "grad_norm": 0.4361816942691803,
      "learning_rate": 9.115085390157725e-06,
      "loss": 0.3388,
      "step": 13627
    },
    {
      "epoch": 0.6253958056078197,
      "grad_norm": 0.4343664348125458,
      "learning_rate": 9.114946114401915e-06,
      "loss": 0.31,
      "step": 13628
    },
    {
      "epoch": 0.6254416961130742,
      "grad_norm": 0.44521623849868774,
      "learning_rate": 9.11480682875096e-06,
      "loss": 0.3835,
      "step": 13629
    },
    {
      "epoch": 0.6254875866183287,
      "grad_norm": 0.489960253238678,
      "learning_rate": 9.114667533205193e-06,
      "loss": 0.477,
      "step": 13630
    },
    {
      "epoch": 0.6255334771235831,
      "grad_norm": 0.4271267354488373,
      "learning_rate": 9.11452822776495e-06,
      "loss": 0.3262,
      "step": 13631
    },
    {
      "epoch": 0.6255793676288376,
      "grad_norm": 0.44223764538764954,
      "learning_rate": 9.114388912430563e-06,
      "loss": 0.3759,
      "step": 13632
    },
    {
      "epoch": 0.6256252581340921,
      "grad_norm": 0.42774978280067444,
      "learning_rate": 9.114249587202371e-06,
      "loss": 0.3884,
      "step": 13633
    },
    {
      "epoch": 0.6256711486393465,
      "grad_norm": 0.44865235686302185,
      "learning_rate": 9.114110252080705e-06,
      "loss": 0.423,
      "step": 13634
    },
    {
      "epoch": 0.625717039144601,
      "grad_norm": 0.4593746066093445,
      "learning_rate": 9.113970907065905e-06,
      "loss": 0.4284,
      "step": 13635
    },
    {
      "epoch": 0.6257629296498555,
      "grad_norm": 0.48485514521598816,
      "learning_rate": 9.113831552158302e-06,
      "loss": 0.4491,
      "step": 13636
    },
    {
      "epoch": 0.6258088201551099,
      "grad_norm": 0.46986663341522217,
      "learning_rate": 9.113692187358233e-06,
      "loss": 0.3846,
      "step": 13637
    },
    {
      "epoch": 0.6258547106603644,
      "grad_norm": 0.4344359040260315,
      "learning_rate": 9.113552812666033e-06,
      "loss": 0.3783,
      "step": 13638
    },
    {
      "epoch": 0.6259006011656189,
      "grad_norm": 0.4404149651527405,
      "learning_rate": 9.113413428082036e-06,
      "loss": 0.3892,
      "step": 13639
    },
    {
      "epoch": 0.6259464916708734,
      "grad_norm": 0.47666043043136597,
      "learning_rate": 9.113274033606578e-06,
      "loss": 0.3968,
      "step": 13640
    },
    {
      "epoch": 0.6259923821761277,
      "grad_norm": 0.46210432052612305,
      "learning_rate": 9.113134629239994e-06,
      "loss": 0.4457,
      "step": 13641
    },
    {
      "epoch": 0.6260382726813822,
      "grad_norm": 0.4827441871166229,
      "learning_rate": 9.112995214982619e-06,
      "loss": 0.3991,
      "step": 13642
    },
    {
      "epoch": 0.6260841631866367,
      "grad_norm": 0.4562370479106903,
      "learning_rate": 9.112855790834789e-06,
      "loss": 0.3974,
      "step": 13643
    },
    {
      "epoch": 0.6261300536918911,
      "grad_norm": 0.5029146075248718,
      "learning_rate": 9.112716356796839e-06,
      "loss": 0.4656,
      "step": 13644
    },
    {
      "epoch": 0.6261759441971456,
      "grad_norm": 0.47383129596710205,
      "learning_rate": 9.112576912869104e-06,
      "loss": 0.3909,
      "step": 13645
    },
    {
      "epoch": 0.6262218347024001,
      "grad_norm": 0.4470995366573334,
      "learning_rate": 9.112437459051919e-06,
      "loss": 0.4201,
      "step": 13646
    },
    {
      "epoch": 0.6262677252076545,
      "grad_norm": 0.42845162749290466,
      "learning_rate": 9.11229799534562e-06,
      "loss": 0.3235,
      "step": 13647
    },
    {
      "epoch": 0.626313615712909,
      "grad_norm": 0.42569148540496826,
      "learning_rate": 9.112158521750544e-06,
      "loss": 0.3395,
      "step": 13648
    },
    {
      "epoch": 0.6263595062181635,
      "grad_norm": 0.4630505442619324,
      "learning_rate": 9.112019038267023e-06,
      "loss": 0.3906,
      "step": 13649
    },
    {
      "epoch": 0.6264053967234179,
      "grad_norm": 0.4603320062160492,
      "learning_rate": 9.111879544895395e-06,
      "loss": 0.386,
      "step": 13650
    },
    {
      "epoch": 0.6264512872286724,
      "grad_norm": 0.4445750117301941,
      "learning_rate": 9.111740041635993e-06,
      "loss": 0.3796,
      "step": 13651
    },
    {
      "epoch": 0.6264971777339269,
      "grad_norm": 0.470172256231308,
      "learning_rate": 9.111600528489154e-06,
      "loss": 0.4074,
      "step": 13652
    },
    {
      "epoch": 0.6265430682391813,
      "grad_norm": 0.4349626898765564,
      "learning_rate": 9.111461005455215e-06,
      "loss": 0.3776,
      "step": 13653
    },
    {
      "epoch": 0.6265889587444358,
      "grad_norm": 0.4975203275680542,
      "learning_rate": 9.111321472534509e-06,
      "loss": 0.4588,
      "step": 13654
    },
    {
      "epoch": 0.6266348492496903,
      "grad_norm": 0.4368135631084442,
      "learning_rate": 9.111181929727371e-06,
      "loss": 0.318,
      "step": 13655
    },
    {
      "epoch": 0.6266807397549446,
      "grad_norm": 0.4769822061061859,
      "learning_rate": 9.111042377034141e-06,
      "loss": 0.4914,
      "step": 13656
    },
    {
      "epoch": 0.6267266302601991,
      "grad_norm": 0.4669618010520935,
      "learning_rate": 9.110902814455149e-06,
      "loss": 0.4441,
      "step": 13657
    },
    {
      "epoch": 0.6267725207654536,
      "grad_norm": 0.4871855676174164,
      "learning_rate": 9.110763241990735e-06,
      "loss": 0.4918,
      "step": 13658
    },
    {
      "epoch": 0.6268184112707081,
      "grad_norm": 0.4395117461681366,
      "learning_rate": 9.110623659641231e-06,
      "loss": 0.4073,
      "step": 13659
    },
    {
      "epoch": 0.6268643017759625,
      "grad_norm": 0.46548977494239807,
      "learning_rate": 9.110484067406975e-06,
      "loss": 0.4305,
      "step": 13660
    },
    {
      "epoch": 0.626910192281217,
      "grad_norm": 0.523999810218811,
      "learning_rate": 9.110344465288303e-06,
      "loss": 0.5316,
      "step": 13661
    },
    {
      "epoch": 0.6269560827864715,
      "grad_norm": 0.4806918501853943,
      "learning_rate": 9.110204853285549e-06,
      "loss": 0.3853,
      "step": 13662
    },
    {
      "epoch": 0.6270019732917259,
      "grad_norm": 0.44492775201797485,
      "learning_rate": 9.11006523139905e-06,
      "loss": 0.3527,
      "step": 13663
    },
    {
      "epoch": 0.6270478637969804,
      "grad_norm": 0.462098091840744,
      "learning_rate": 9.10992559962914e-06,
      "loss": 0.3896,
      "step": 13664
    },
    {
      "epoch": 0.6270937543022349,
      "grad_norm": 0.4390881657600403,
      "learning_rate": 9.109785957976158e-06,
      "loss": 0.3566,
      "step": 13665
    },
    {
      "epoch": 0.6271396448074893,
      "grad_norm": 0.4432145655155182,
      "learning_rate": 9.109646306440438e-06,
      "loss": 0.3377,
      "step": 13666
    },
    {
      "epoch": 0.6271855353127438,
      "grad_norm": 0.4953557550907135,
      "learning_rate": 9.109506645022314e-06,
      "loss": 0.4393,
      "step": 13667
    },
    {
      "epoch": 0.6272314258179983,
      "grad_norm": 0.5088863968849182,
      "learning_rate": 9.109366973722125e-06,
      "loss": 0.4847,
      "step": 13668
    },
    {
      "epoch": 0.6272773163232527,
      "grad_norm": 0.5103203654289246,
      "learning_rate": 9.109227292540204e-06,
      "loss": 0.4544,
      "step": 13669
    },
    {
      "epoch": 0.6273232068285072,
      "grad_norm": 0.46356844902038574,
      "learning_rate": 9.10908760147689e-06,
      "loss": 0.4299,
      "step": 13670
    },
    {
      "epoch": 0.6273690973337617,
      "grad_norm": 0.4585416615009308,
      "learning_rate": 9.108947900532514e-06,
      "loss": 0.3872,
      "step": 13671
    },
    {
      "epoch": 0.6274149878390161,
      "grad_norm": 0.42603009939193726,
      "learning_rate": 9.108808189707418e-06,
      "loss": 0.3522,
      "step": 13672
    },
    {
      "epoch": 0.6274608783442706,
      "grad_norm": 0.45882779359817505,
      "learning_rate": 9.108668469001933e-06,
      "loss": 0.4076,
      "step": 13673
    },
    {
      "epoch": 0.627506768849525,
      "grad_norm": 0.46660116314888,
      "learning_rate": 9.108528738416399e-06,
      "loss": 0.3818,
      "step": 13674
    },
    {
      "epoch": 0.6275526593547794,
      "grad_norm": 0.4421790838241577,
      "learning_rate": 9.10838899795115e-06,
      "loss": 0.3738,
      "step": 13675
    },
    {
      "epoch": 0.6275985498600339,
      "grad_norm": 0.4388272762298584,
      "learning_rate": 9.10824924760652e-06,
      "loss": 0.3667,
      "step": 13676
    },
    {
      "epoch": 0.6276444403652884,
      "grad_norm": 0.4716217815876007,
      "learning_rate": 9.10810948738285e-06,
      "loss": 0.4028,
      "step": 13677
    },
    {
      "epoch": 0.6276903308705429,
      "grad_norm": 0.4507223963737488,
      "learning_rate": 9.107969717280471e-06,
      "loss": 0.391,
      "step": 13678
    },
    {
      "epoch": 0.6277362213757973,
      "grad_norm": 0.47650957107543945,
      "learning_rate": 9.107829937299724e-06,
      "loss": 0.3869,
      "step": 13679
    },
    {
      "epoch": 0.6277821118810518,
      "grad_norm": 0.45618265867233276,
      "learning_rate": 9.10769014744094e-06,
      "loss": 0.4075,
      "step": 13680
    },
    {
      "epoch": 0.6278280023863063,
      "grad_norm": 0.5932179093360901,
      "learning_rate": 9.107550347704457e-06,
      "loss": 0.4486,
      "step": 13681
    },
    {
      "epoch": 0.6278738928915607,
      "grad_norm": 0.4464779496192932,
      "learning_rate": 9.107410538090615e-06,
      "loss": 0.4153,
      "step": 13682
    },
    {
      "epoch": 0.6279197833968152,
      "grad_norm": 0.4385484457015991,
      "learning_rate": 9.107270718599743e-06,
      "loss": 0.3811,
      "step": 13683
    },
    {
      "epoch": 0.6279656739020697,
      "grad_norm": 0.47455912828445435,
      "learning_rate": 9.107130889232185e-06,
      "loss": 0.415,
      "step": 13684
    },
    {
      "epoch": 0.6280115644073241,
      "grad_norm": 0.4071928858757019,
      "learning_rate": 9.106991049988271e-06,
      "loss": 0.3222,
      "step": 13685
    },
    {
      "epoch": 0.6280574549125786,
      "grad_norm": 0.46810925006866455,
      "learning_rate": 9.106851200868341e-06,
      "loss": 0.3986,
      "step": 13686
    },
    {
      "epoch": 0.6281033454178331,
      "grad_norm": 0.45160913467407227,
      "learning_rate": 9.106711341872728e-06,
      "loss": 0.4065,
      "step": 13687
    },
    {
      "epoch": 0.6281492359230875,
      "grad_norm": 0.46360260248184204,
      "learning_rate": 9.106571473001772e-06,
      "loss": 0.3917,
      "step": 13688
    },
    {
      "epoch": 0.628195126428342,
      "grad_norm": 0.4638102650642395,
      "learning_rate": 9.10643159425581e-06,
      "loss": 0.4669,
      "step": 13689
    },
    {
      "epoch": 0.6282410169335965,
      "grad_norm": 0.43756335973739624,
      "learning_rate": 9.106291705635172e-06,
      "loss": 0.3737,
      "step": 13690
    },
    {
      "epoch": 0.6282869074388509,
      "grad_norm": 0.4349897503852844,
      "learning_rate": 9.1061518071402e-06,
      "loss": 0.3504,
      "step": 13691
    },
    {
      "epoch": 0.6283327979441053,
      "grad_norm": 0.46866628527641296,
      "learning_rate": 9.10601189877123e-06,
      "loss": 0.4293,
      "step": 13692
    },
    {
      "epoch": 0.6283786884493598,
      "grad_norm": 0.4611052870750427,
      "learning_rate": 9.105871980528595e-06,
      "loss": 0.397,
      "step": 13693
    },
    {
      "epoch": 0.6284245789546143,
      "grad_norm": 0.4438941478729248,
      "learning_rate": 9.105732052412634e-06,
      "loss": 0.3831,
      "step": 13694
    },
    {
      "epoch": 0.6284704694598687,
      "grad_norm": 0.4622237980365753,
      "learning_rate": 9.105592114423685e-06,
      "loss": 0.4501,
      "step": 13695
    },
    {
      "epoch": 0.6285163599651232,
      "grad_norm": 0.4574432373046875,
      "learning_rate": 9.105452166562082e-06,
      "loss": 0.4012,
      "step": 13696
    },
    {
      "epoch": 0.6285622504703777,
      "grad_norm": 0.45375698804855347,
      "learning_rate": 9.105312208828162e-06,
      "loss": 0.3427,
      "step": 13697
    },
    {
      "epoch": 0.6286081409756321,
      "grad_norm": 0.41117915511131287,
      "learning_rate": 9.105172241222262e-06,
      "loss": 0.3411,
      "step": 13698
    },
    {
      "epoch": 0.6286540314808866,
      "grad_norm": 0.4615662097930908,
      "learning_rate": 9.105032263744717e-06,
      "loss": 0.4107,
      "step": 13699
    },
    {
      "epoch": 0.6286999219861411,
      "grad_norm": 0.42452019453048706,
      "learning_rate": 9.104892276395866e-06,
      "loss": 0.363,
      "step": 13700
    },
    {
      "epoch": 0.6287458124913955,
      "grad_norm": 0.418771892786026,
      "learning_rate": 9.104752279176045e-06,
      "loss": 0.3387,
      "step": 13701
    },
    {
      "epoch": 0.62879170299665,
      "grad_norm": 0.47144344449043274,
      "learning_rate": 9.10461227208559e-06,
      "loss": 0.4406,
      "step": 13702
    },
    {
      "epoch": 0.6288375935019045,
      "grad_norm": 0.499202162027359,
      "learning_rate": 9.104472255124839e-06,
      "loss": 0.3083,
      "step": 13703
    },
    {
      "epoch": 0.6288834840071589,
      "grad_norm": 0.40753117203712463,
      "learning_rate": 9.104332228294127e-06,
      "loss": 0.3427,
      "step": 13704
    },
    {
      "epoch": 0.6289293745124134,
      "grad_norm": 0.4543326795101166,
      "learning_rate": 9.10419219159379e-06,
      "loss": 0.3791,
      "step": 13705
    },
    {
      "epoch": 0.6289752650176679,
      "grad_norm": 0.4498511552810669,
      "learning_rate": 9.104052145024169e-06,
      "loss": 0.3764,
      "step": 13706
    },
    {
      "epoch": 0.6290211555229223,
      "grad_norm": 0.4355706572532654,
      "learning_rate": 9.103912088585596e-06,
      "loss": 0.3275,
      "step": 13707
    },
    {
      "epoch": 0.6290670460281768,
      "grad_norm": 0.4331609904766083,
      "learning_rate": 9.103772022278409e-06,
      "loss": 0.3975,
      "step": 13708
    },
    {
      "epoch": 0.6291129365334313,
      "grad_norm": 0.4652096629142761,
      "learning_rate": 9.103631946102946e-06,
      "loss": 0.4657,
      "step": 13709
    },
    {
      "epoch": 0.6291588270386856,
      "grad_norm": 0.4701501727104187,
      "learning_rate": 9.103491860059544e-06,
      "loss": 0.3938,
      "step": 13710
    },
    {
      "epoch": 0.6292047175439401,
      "grad_norm": 0.49782222509384155,
      "learning_rate": 9.103351764148538e-06,
      "loss": 0.4854,
      "step": 13711
    },
    {
      "epoch": 0.6292506080491946,
      "grad_norm": 0.4514663517475128,
      "learning_rate": 9.103211658370268e-06,
      "loss": 0.3872,
      "step": 13712
    },
    {
      "epoch": 0.6292964985544491,
      "grad_norm": 0.44818544387817383,
      "learning_rate": 9.103071542725067e-06,
      "loss": 0.3653,
      "step": 13713
    },
    {
      "epoch": 0.6293423890597035,
      "grad_norm": 0.46137815713882446,
      "learning_rate": 9.102931417213275e-06,
      "loss": 0.3352,
      "step": 13714
    },
    {
      "epoch": 0.629388279564958,
      "grad_norm": 0.5046994686126709,
      "learning_rate": 9.102791281835227e-06,
      "loss": 0.4767,
      "step": 13715
    },
    {
      "epoch": 0.6294341700702125,
      "grad_norm": 0.45495080947875977,
      "learning_rate": 9.10265113659126e-06,
      "loss": 0.3999,
      "step": 13716
    },
    {
      "epoch": 0.6294800605754669,
      "grad_norm": 0.4386732876300812,
      "learning_rate": 9.102510981481715e-06,
      "loss": 0.3868,
      "step": 13717
    },
    {
      "epoch": 0.6295259510807214,
      "grad_norm": 0.4759403467178345,
      "learning_rate": 9.102370816506922e-06,
      "loss": 0.3955,
      "step": 13718
    },
    {
      "epoch": 0.6295718415859759,
      "grad_norm": 0.436422735452652,
      "learning_rate": 9.102230641667225e-06,
      "loss": 0.3701,
      "step": 13719
    },
    {
      "epoch": 0.6296177320912303,
      "grad_norm": 0.4523036777973175,
      "learning_rate": 9.102090456962957e-06,
      "loss": 0.439,
      "step": 13720
    },
    {
      "epoch": 0.6296636225964848,
      "grad_norm": 0.4791155457496643,
      "learning_rate": 9.101950262394456e-06,
      "loss": 0.4185,
      "step": 13721
    },
    {
      "epoch": 0.6297095131017393,
      "grad_norm": 0.43001556396484375,
      "learning_rate": 9.10181005796206e-06,
      "loss": 0.3468,
      "step": 13722
    },
    {
      "epoch": 0.6297554036069937,
      "grad_norm": 0.4744299054145813,
      "learning_rate": 9.101669843666105e-06,
      "loss": 0.4486,
      "step": 13723
    },
    {
      "epoch": 0.6298012941122482,
      "grad_norm": 0.4610910415649414,
      "learning_rate": 9.101529619506928e-06,
      "loss": 0.3687,
      "step": 13724
    },
    {
      "epoch": 0.6298471846175027,
      "grad_norm": 0.43942463397979736,
      "learning_rate": 9.101389385484866e-06,
      "loss": 0.3915,
      "step": 13725
    },
    {
      "epoch": 0.629893075122757,
      "grad_norm": 0.470424622297287,
      "learning_rate": 9.101249141600259e-06,
      "loss": 0.4098,
      "step": 13726
    },
    {
      "epoch": 0.6299389656280115,
      "grad_norm": 0.4710494875907898,
      "learning_rate": 9.10110888785344e-06,
      "loss": 0.4275,
      "step": 13727
    },
    {
      "epoch": 0.629984856133266,
      "grad_norm": 0.4294699430465698,
      "learning_rate": 9.10096862424475e-06,
      "loss": 0.3216,
      "step": 13728
    },
    {
      "epoch": 0.6300307466385205,
      "grad_norm": 0.44323498010635376,
      "learning_rate": 9.100828350774525e-06,
      "loss": 0.3515,
      "step": 13729
    },
    {
      "epoch": 0.6300766371437749,
      "grad_norm": 0.4185742139816284,
      "learning_rate": 9.100688067443102e-06,
      "loss": 0.3879,
      "step": 13730
    },
    {
      "epoch": 0.6301225276490294,
      "grad_norm": 0.4734041094779968,
      "learning_rate": 9.100547774250817e-06,
      "loss": 0.4252,
      "step": 13731
    },
    {
      "epoch": 0.6301684181542839,
      "grad_norm": 0.45422691106796265,
      "learning_rate": 9.100407471198009e-06,
      "loss": 0.4046,
      "step": 13732
    },
    {
      "epoch": 0.6302143086595383,
      "grad_norm": 0.44218188524246216,
      "learning_rate": 9.100267158285016e-06,
      "loss": 0.3592,
      "step": 13733
    },
    {
      "epoch": 0.6302601991647928,
      "grad_norm": 0.42819520831108093,
      "learning_rate": 9.100126835512175e-06,
      "loss": 0.3182,
      "step": 13734
    },
    {
      "epoch": 0.6303060896700473,
      "grad_norm": 0.4627501368522644,
      "learning_rate": 9.099986502879823e-06,
      "loss": 0.3996,
      "step": 13735
    },
    {
      "epoch": 0.6303519801753017,
      "grad_norm": 0.4762738347053528,
      "learning_rate": 9.099846160388297e-06,
      "loss": 0.4549,
      "step": 13736
    },
    {
      "epoch": 0.6303978706805562,
      "grad_norm": 0.5006212592124939,
      "learning_rate": 9.099705808037935e-06,
      "loss": 0.4958,
      "step": 13737
    },
    {
      "epoch": 0.6304437611858107,
      "grad_norm": 0.4639914929866791,
      "learning_rate": 9.099565445829072e-06,
      "loss": 0.4508,
      "step": 13738
    },
    {
      "epoch": 0.6304896516910651,
      "grad_norm": 0.48809918761253357,
      "learning_rate": 9.099425073762051e-06,
      "loss": 0.4518,
      "step": 13739
    },
    {
      "epoch": 0.6305355421963196,
      "grad_norm": 0.44706717133522034,
      "learning_rate": 9.099284691837206e-06,
      "loss": 0.3364,
      "step": 13740
    },
    {
      "epoch": 0.6305814327015741,
      "grad_norm": 0.4834180176258087,
      "learning_rate": 9.099144300054875e-06,
      "loss": 0.4764,
      "step": 13741
    },
    {
      "epoch": 0.6306273232068285,
      "grad_norm": 0.4418417513370514,
      "learning_rate": 9.099003898415394e-06,
      "loss": 0.3603,
      "step": 13742
    },
    {
      "epoch": 0.630673213712083,
      "grad_norm": 0.4699194133281708,
      "learning_rate": 9.098863486919105e-06,
      "loss": 0.4584,
      "step": 13743
    },
    {
      "epoch": 0.6307191042173375,
      "grad_norm": 0.4323606789112091,
      "learning_rate": 9.098723065566341e-06,
      "loss": 0.3384,
      "step": 13744
    },
    {
      "epoch": 0.6307649947225918,
      "grad_norm": 0.4393993616104126,
      "learning_rate": 9.098582634357442e-06,
      "loss": 0.368,
      "step": 13745
    },
    {
      "epoch": 0.6308108852278463,
      "grad_norm": 0.5486024618148804,
      "learning_rate": 9.098442193292747e-06,
      "loss": 0.476,
      "step": 13746
    },
    {
      "epoch": 0.6308567757331008,
      "grad_norm": 0.44537246227264404,
      "learning_rate": 9.09830174237259e-06,
      "loss": 0.4006,
      "step": 13747
    },
    {
      "epoch": 0.6309026662383553,
      "grad_norm": 0.4700680077075958,
      "learning_rate": 9.098161281597312e-06,
      "loss": 0.4597,
      "step": 13748
    },
    {
      "epoch": 0.6309485567436097,
      "grad_norm": 0.44967445731163025,
      "learning_rate": 9.09802081096725e-06,
      "loss": 0.3649,
      "step": 13749
    },
    {
      "epoch": 0.6309944472488642,
      "grad_norm": 0.45598292350769043,
      "learning_rate": 9.09788033048274e-06,
      "loss": 0.405,
      "step": 13750
    },
    {
      "epoch": 0.6310403377541187,
      "grad_norm": 0.48644453287124634,
      "learning_rate": 9.09773984014412e-06,
      "loss": 0.4496,
      "step": 13751
    },
    {
      "epoch": 0.6310862282593731,
      "grad_norm": 0.47058364748954773,
      "learning_rate": 9.09759933995173e-06,
      "loss": 0.4834,
      "step": 13752
    },
    {
      "epoch": 0.6311321187646276,
      "grad_norm": 0.4332495629787445,
      "learning_rate": 9.097458829905908e-06,
      "loss": 0.343,
      "step": 13753
    },
    {
      "epoch": 0.6311780092698821,
      "grad_norm": 0.4840836226940155,
      "learning_rate": 9.09731831000699e-06,
      "loss": 0.4337,
      "step": 13754
    },
    {
      "epoch": 0.6312238997751365,
      "grad_norm": 0.5089073181152344,
      "learning_rate": 9.097177780255315e-06,
      "loss": 0.4504,
      "step": 13755
    },
    {
      "epoch": 0.631269790280391,
      "grad_norm": 0.4627860188484192,
      "learning_rate": 9.097037240651219e-06,
      "loss": 0.3985,
      "step": 13756
    },
    {
      "epoch": 0.6313156807856455,
      "grad_norm": 0.4629143476486206,
      "learning_rate": 9.096896691195045e-06,
      "loss": 0.3851,
      "step": 13757
    },
    {
      "epoch": 0.6313615712908999,
      "grad_norm": 0.44181910157203674,
      "learning_rate": 9.096756131887123e-06,
      "loss": 0.3344,
      "step": 13758
    },
    {
      "epoch": 0.6314074617961544,
      "grad_norm": 0.5098085403442383,
      "learning_rate": 9.096615562727797e-06,
      "loss": 0.4219,
      "step": 13759
    },
    {
      "epoch": 0.6314533523014089,
      "grad_norm": 0.49601680040359497,
      "learning_rate": 9.096474983717404e-06,
      "loss": 0.4059,
      "step": 13760
    },
    {
      "epoch": 0.6314992428066633,
      "grad_norm": 0.44838351011276245,
      "learning_rate": 9.096334394856283e-06,
      "loss": 0.396,
      "step": 13761
    },
    {
      "epoch": 0.6315451333119178,
      "grad_norm": 0.4441055655479431,
      "learning_rate": 9.09619379614477e-06,
      "loss": 0.3732,
      "step": 13762
    },
    {
      "epoch": 0.6315910238171722,
      "grad_norm": 0.4844288229942322,
      "learning_rate": 9.096053187583202e-06,
      "loss": 0.4626,
      "step": 13763
    },
    {
      "epoch": 0.6316369143224266,
      "grad_norm": 0.4508432447910309,
      "learning_rate": 9.09591256917192e-06,
      "loss": 0.3457,
      "step": 13764
    },
    {
      "epoch": 0.6316828048276811,
      "grad_norm": 0.44080203771591187,
      "learning_rate": 9.095771940911261e-06,
      "loss": 0.3562,
      "step": 13765
    },
    {
      "epoch": 0.6317286953329356,
      "grad_norm": 0.4298355281352997,
      "learning_rate": 9.095631302801563e-06,
      "loss": 0.3687,
      "step": 13766
    },
    {
      "epoch": 0.6317745858381901,
      "grad_norm": 0.4905388057231903,
      "learning_rate": 9.095490654843164e-06,
      "loss": 0.4773,
      "step": 13767
    },
    {
      "epoch": 0.6318204763434445,
      "grad_norm": 0.4200710952281952,
      "learning_rate": 9.095349997036402e-06,
      "loss": 0.3635,
      "step": 13768
    },
    {
      "epoch": 0.631866366848699,
      "grad_norm": 0.4724201560020447,
      "learning_rate": 9.095209329381616e-06,
      "loss": 0.3926,
      "step": 13769
    },
    {
      "epoch": 0.6319122573539535,
      "grad_norm": 0.4468879699707031,
      "learning_rate": 9.095068651879146e-06,
      "loss": 0.3713,
      "step": 13770
    },
    {
      "epoch": 0.6319581478592079,
      "grad_norm": 0.4829399883747101,
      "learning_rate": 9.094927964529328e-06,
      "loss": 0.3525,
      "step": 13771
    },
    {
      "epoch": 0.6320040383644624,
      "grad_norm": 0.46195679903030396,
      "learning_rate": 9.0947872673325e-06,
      "loss": 0.3963,
      "step": 13772
    },
    {
      "epoch": 0.6320499288697169,
      "grad_norm": 0.4676263928413391,
      "learning_rate": 9.094646560289e-06,
      "loss": 0.4127,
      "step": 13773
    },
    {
      "epoch": 0.6320958193749713,
      "grad_norm": 0.45218780636787415,
      "learning_rate": 9.094505843399167e-06,
      "loss": 0.4086,
      "step": 13774
    },
    {
      "epoch": 0.6321417098802258,
      "grad_norm": 0.4736347794532776,
      "learning_rate": 9.09436511666334e-06,
      "loss": 0.3823,
      "step": 13775
    },
    {
      "epoch": 0.6321876003854803,
      "grad_norm": 0.4676770567893982,
      "learning_rate": 9.094224380081858e-06,
      "loss": 0.3827,
      "step": 13776
    },
    {
      "epoch": 0.6322334908907347,
      "grad_norm": 0.454408198595047,
      "learning_rate": 9.094083633655058e-06,
      "loss": 0.337,
      "step": 13777
    },
    {
      "epoch": 0.6322793813959892,
      "grad_norm": 0.45019590854644775,
      "learning_rate": 9.093942877383279e-06,
      "loss": 0.393,
      "step": 13778
    },
    {
      "epoch": 0.6323252719012437,
      "grad_norm": 0.4695059359073639,
      "learning_rate": 9.093802111266861e-06,
      "loss": 0.4311,
      "step": 13779
    },
    {
      "epoch": 0.632371162406498,
      "grad_norm": 0.46594831347465515,
      "learning_rate": 9.093661335306138e-06,
      "loss": 0.392,
      "step": 13780
    },
    {
      "epoch": 0.6324170529117525,
      "grad_norm": 0.390536367893219,
      "learning_rate": 9.093520549501454e-06,
      "loss": 0.2994,
      "step": 13781
    },
    {
      "epoch": 0.632462943417007,
      "grad_norm": 0.49860137701034546,
      "learning_rate": 9.093379753853144e-06,
      "loss": 0.5128,
      "step": 13782
    },
    {
      "epoch": 0.6325088339222615,
      "grad_norm": 0.4420652687549591,
      "learning_rate": 9.093238948361547e-06,
      "loss": 0.3341,
      "step": 13783
    },
    {
      "epoch": 0.6325547244275159,
      "grad_norm": 0.4485955834388733,
      "learning_rate": 9.093098133027004e-06,
      "loss": 0.3357,
      "step": 13784
    },
    {
      "epoch": 0.6326006149327704,
      "grad_norm": 0.46887052059173584,
      "learning_rate": 9.09295730784985e-06,
      "loss": 0.4137,
      "step": 13785
    },
    {
      "epoch": 0.6326465054380249,
      "grad_norm": 0.4433824419975281,
      "learning_rate": 9.092816472830426e-06,
      "loss": 0.3636,
      "step": 13786
    },
    {
      "epoch": 0.6326923959432793,
      "grad_norm": 0.4383932650089264,
      "learning_rate": 9.092675627969072e-06,
      "loss": 0.3971,
      "step": 13787
    },
    {
      "epoch": 0.6327382864485338,
      "grad_norm": 0.4448809027671814,
      "learning_rate": 9.092534773266123e-06,
      "loss": 0.3344,
      "step": 13788
    },
    {
      "epoch": 0.6327841769537883,
      "grad_norm": 0.4861891269683838,
      "learning_rate": 9.092393908721918e-06,
      "loss": 0.4417,
      "step": 13789
    },
    {
      "epoch": 0.6328300674590427,
      "grad_norm": 0.5005356669425964,
      "learning_rate": 9.0922530343368e-06,
      "loss": 0.4339,
      "step": 13790
    },
    {
      "epoch": 0.6328759579642972,
      "grad_norm": 0.46084117889404297,
      "learning_rate": 9.092112150111104e-06,
      "loss": 0.3971,
      "step": 13791
    },
    {
      "epoch": 0.6329218484695517,
      "grad_norm": 0.4766274392604828,
      "learning_rate": 9.091971256045171e-06,
      "loss": 0.4974,
      "step": 13792
    },
    {
      "epoch": 0.6329677389748061,
      "grad_norm": 0.4248136579990387,
      "learning_rate": 9.091830352139337e-06,
      "loss": 0.3507,
      "step": 13793
    },
    {
      "epoch": 0.6330136294800606,
      "grad_norm": 0.4457894563674927,
      "learning_rate": 9.091689438393943e-06,
      "loss": 0.3404,
      "step": 13794
    },
    {
      "epoch": 0.6330595199853151,
      "grad_norm": 0.46522021293640137,
      "learning_rate": 9.091548514809326e-06,
      "loss": 0.3995,
      "step": 13795
    },
    {
      "epoch": 0.6331054104905695,
      "grad_norm": 0.4518631398677826,
      "learning_rate": 9.091407581385827e-06,
      "loss": 0.3947,
      "step": 13796
    },
    {
      "epoch": 0.633151300995824,
      "grad_norm": 0.4420549273490906,
      "learning_rate": 9.091266638123783e-06,
      "loss": 0.354,
      "step": 13797
    },
    {
      "epoch": 0.6331971915010784,
      "grad_norm": 0.49655985832214355,
      "learning_rate": 9.091125685023536e-06,
      "loss": 0.4533,
      "step": 13798
    },
    {
      "epoch": 0.6332430820063328,
      "grad_norm": 0.45235541462898254,
      "learning_rate": 9.090984722085423e-06,
      "loss": 0.4258,
      "step": 13799
    },
    {
      "epoch": 0.6332889725115873,
      "grad_norm": 0.5090921521186829,
      "learning_rate": 9.09084374930978e-06,
      "loss": 0.5052,
      "step": 13800
    },
    {
      "epoch": 0.6333348630168418,
      "grad_norm": 0.38741999864578247,
      "learning_rate": 9.090702766696952e-06,
      "loss": 0.2875,
      "step": 13801
    },
    {
      "epoch": 0.6333807535220963,
      "grad_norm": 0.482356458902359,
      "learning_rate": 9.090561774247275e-06,
      "loss": 0.4361,
      "step": 13802
    },
    {
      "epoch": 0.6334266440273507,
      "grad_norm": 0.4722704291343689,
      "learning_rate": 9.090420771961086e-06,
      "loss": 0.4196,
      "step": 13803
    },
    {
      "epoch": 0.6334725345326052,
      "grad_norm": 0.47964537143707275,
      "learning_rate": 9.090279759838728e-06,
      "loss": 0.4708,
      "step": 13804
    },
    {
      "epoch": 0.6335184250378597,
      "grad_norm": 0.4740954637527466,
      "learning_rate": 9.090138737880536e-06,
      "loss": 0.4126,
      "step": 13805
    },
    {
      "epoch": 0.6335643155431141,
      "grad_norm": 0.533951997756958,
      "learning_rate": 9.089997706086852e-06,
      "loss": 0.4267,
      "step": 13806
    },
    {
      "epoch": 0.6336102060483686,
      "grad_norm": 0.4479615092277527,
      "learning_rate": 9.089856664458015e-06,
      "loss": 0.4248,
      "step": 13807
    },
    {
      "epoch": 0.6336560965536231,
      "grad_norm": 0.42520639300346375,
      "learning_rate": 9.089715612994362e-06,
      "loss": 0.3249,
      "step": 13808
    },
    {
      "epoch": 0.6337019870588775,
      "grad_norm": 0.47059741616249084,
      "learning_rate": 9.089574551696236e-06,
      "loss": 0.4684,
      "step": 13809
    },
    {
      "epoch": 0.633747877564132,
      "grad_norm": 0.44987934827804565,
      "learning_rate": 9.089433480563972e-06,
      "loss": 0.3834,
      "step": 13810
    },
    {
      "epoch": 0.6337937680693865,
      "grad_norm": 0.49606969952583313,
      "learning_rate": 9.08929239959791e-06,
      "loss": 0.4808,
      "step": 13811
    },
    {
      "epoch": 0.6338396585746409,
      "grad_norm": 0.40498825907707214,
      "learning_rate": 9.089151308798394e-06,
      "loss": 0.3397,
      "step": 13812
    },
    {
      "epoch": 0.6338855490798954,
      "grad_norm": 0.4510362446308136,
      "learning_rate": 9.089010208165758e-06,
      "loss": 0.3977,
      "step": 13813
    },
    {
      "epoch": 0.6339314395851499,
      "grad_norm": 0.4799993336200714,
      "learning_rate": 9.088869097700342e-06,
      "loss": 0.4164,
      "step": 13814
    },
    {
      "epoch": 0.6339773300904042,
      "grad_norm": 0.46112826466560364,
      "learning_rate": 9.088727977402487e-06,
      "loss": 0.3934,
      "step": 13815
    },
    {
      "epoch": 0.6340232205956587,
      "grad_norm": 0.44037288427352905,
      "learning_rate": 9.088586847272532e-06,
      "loss": 0.398,
      "step": 13816
    },
    {
      "epoch": 0.6340691111009132,
      "grad_norm": 0.45116597414016724,
      "learning_rate": 9.088445707310816e-06,
      "loss": 0.4015,
      "step": 13817
    },
    {
      "epoch": 0.6341150016061677,
      "grad_norm": 0.44511812925338745,
      "learning_rate": 9.088304557517677e-06,
      "loss": 0.337,
      "step": 13818
    },
    {
      "epoch": 0.6341608921114221,
      "grad_norm": 0.4653298258781433,
      "learning_rate": 9.088163397893459e-06,
      "loss": 0.4567,
      "step": 13819
    },
    {
      "epoch": 0.6342067826166766,
      "grad_norm": 0.44936805963516235,
      "learning_rate": 9.088022228438496e-06,
      "loss": 0.3789,
      "step": 13820
    },
    {
      "epoch": 0.6342526731219311,
      "grad_norm": 0.45591941475868225,
      "learning_rate": 9.087881049153128e-06,
      "loss": 0.4454,
      "step": 13821
    },
    {
      "epoch": 0.6342985636271855,
      "grad_norm": 0.45965129137039185,
      "learning_rate": 9.0877398600377e-06,
      "loss": 0.3848,
      "step": 13822
    },
    {
      "epoch": 0.63434445413244,
      "grad_norm": 0.42764368653297424,
      "learning_rate": 9.087598661092546e-06,
      "loss": 0.3248,
      "step": 13823
    },
    {
      "epoch": 0.6343903446376945,
      "grad_norm": 0.43337202072143555,
      "learning_rate": 9.087457452318008e-06,
      "loss": 0.3662,
      "step": 13824
    },
    {
      "epoch": 0.6344362351429489,
      "grad_norm": 0.42204010486602783,
      "learning_rate": 9.087316233714422e-06,
      "loss": 0.3336,
      "step": 13825
    },
    {
      "epoch": 0.6344821256482034,
      "grad_norm": 0.46364161372184753,
      "learning_rate": 9.087175005282135e-06,
      "loss": 0.4186,
      "step": 13826
    },
    {
      "epoch": 0.6345280161534579,
      "grad_norm": 0.5562419891357422,
      "learning_rate": 9.08703376702148e-06,
      "loss": 0.4086,
      "step": 13827
    },
    {
      "epoch": 0.6345739066587123,
      "grad_norm": 0.4431101977825165,
      "learning_rate": 9.086892518932798e-06,
      "loss": 0.3519,
      "step": 13828
    },
    {
      "epoch": 0.6346197971639668,
      "grad_norm": 0.42469343543052673,
      "learning_rate": 9.086751261016428e-06,
      "loss": 0.3766,
      "step": 13829
    },
    {
      "epoch": 0.6346656876692213,
      "grad_norm": 0.47859928011894226,
      "learning_rate": 9.086609993272715e-06,
      "loss": 0.4104,
      "step": 13830
    },
    {
      "epoch": 0.6347115781744757,
      "grad_norm": 0.47052422165870667,
      "learning_rate": 9.08646871570199e-06,
      "loss": 0.4489,
      "step": 13831
    },
    {
      "epoch": 0.6347574686797302,
      "grad_norm": 0.4955495297908783,
      "learning_rate": 9.086327428304602e-06,
      "loss": 0.5019,
      "step": 13832
    },
    {
      "epoch": 0.6348033591849847,
      "grad_norm": 0.46023106575012207,
      "learning_rate": 9.086186131080884e-06,
      "loss": 0.4016,
      "step": 13833
    },
    {
      "epoch": 0.634849249690239,
      "grad_norm": 0.4287262260913849,
      "learning_rate": 9.086044824031177e-06,
      "loss": 0.3529,
      "step": 13834
    },
    {
      "epoch": 0.6348951401954935,
      "grad_norm": 0.45648735761642456,
      "learning_rate": 9.085903507155824e-06,
      "loss": 0.4002,
      "step": 13835
    },
    {
      "epoch": 0.634941030700748,
      "grad_norm": 0.45051008462905884,
      "learning_rate": 9.085762180455162e-06,
      "loss": 0.3551,
      "step": 13836
    },
    {
      "epoch": 0.6349869212060025,
      "grad_norm": 0.42927584052085876,
      "learning_rate": 9.08562084392953e-06,
      "loss": 0.4034,
      "step": 13837
    },
    {
      "epoch": 0.6350328117112569,
      "grad_norm": 0.49887391924858093,
      "learning_rate": 9.085479497579272e-06,
      "loss": 0.4655,
      "step": 13838
    },
    {
      "epoch": 0.6350787022165114,
      "grad_norm": 0.444959431886673,
      "learning_rate": 9.085338141404722e-06,
      "loss": 0.3696,
      "step": 13839
    },
    {
      "epoch": 0.6351245927217659,
      "grad_norm": 0.4477539658546448,
      "learning_rate": 9.085196775406227e-06,
      "loss": 0.4385,
      "step": 13840
    },
    {
      "epoch": 0.6351704832270203,
      "grad_norm": 0.4772639572620392,
      "learning_rate": 9.08505539958412e-06,
      "loss": 0.4742,
      "step": 13841
    },
    {
      "epoch": 0.6352163737322748,
      "grad_norm": 0.4190462529659271,
      "learning_rate": 9.084914013938745e-06,
      "loss": 0.3382,
      "step": 13842
    },
    {
      "epoch": 0.6352622642375293,
      "grad_norm": 0.44994276762008667,
      "learning_rate": 9.08477261847044e-06,
      "loss": 0.3774,
      "step": 13843
    },
    {
      "epoch": 0.6353081547427837,
      "grad_norm": 0.45749905705451965,
      "learning_rate": 9.084631213179547e-06,
      "loss": 0.3602,
      "step": 13844
    },
    {
      "epoch": 0.6353540452480382,
      "grad_norm": 0.4623963236808777,
      "learning_rate": 9.084489798066405e-06,
      "loss": 0.3818,
      "step": 13845
    },
    {
      "epoch": 0.6353999357532927,
      "grad_norm": 0.4434499740600586,
      "learning_rate": 9.084348373131356e-06,
      "loss": 0.3486,
      "step": 13846
    },
    {
      "epoch": 0.6354458262585471,
      "grad_norm": 0.4792173504829407,
      "learning_rate": 9.084206938374736e-06,
      "loss": 0.4209,
      "step": 13847
    },
    {
      "epoch": 0.6354917167638016,
      "grad_norm": 0.47200843691825867,
      "learning_rate": 9.084065493796887e-06,
      "loss": 0.3741,
      "step": 13848
    },
    {
      "epoch": 0.6355376072690561,
      "grad_norm": 0.5126557350158691,
      "learning_rate": 9.08392403939815e-06,
      "loss": 0.4596,
      "step": 13849
    },
    {
      "epoch": 0.6355834977743104,
      "grad_norm": 0.5037282705307007,
      "learning_rate": 9.083782575178865e-06,
      "loss": 0.4655,
      "step": 13850
    },
    {
      "epoch": 0.6356293882795649,
      "grad_norm": 0.46655210852622986,
      "learning_rate": 9.083641101139372e-06,
      "loss": 0.4056,
      "step": 13851
    },
    {
      "epoch": 0.6356752787848194,
      "grad_norm": 0.4462193548679352,
      "learning_rate": 9.083499617280011e-06,
      "loss": 0.3545,
      "step": 13852
    },
    {
      "epoch": 0.6357211692900738,
      "grad_norm": 0.43894606828689575,
      "learning_rate": 9.083358123601123e-06,
      "loss": 0.4147,
      "step": 13853
    },
    {
      "epoch": 0.6357670597953283,
      "grad_norm": 0.44434648752212524,
      "learning_rate": 9.083216620103046e-06,
      "loss": 0.3847,
      "step": 13854
    },
    {
      "epoch": 0.6358129503005828,
      "grad_norm": 0.4428238868713379,
      "learning_rate": 9.08307510678612e-06,
      "loss": 0.3579,
      "step": 13855
    },
    {
      "epoch": 0.6358588408058373,
      "grad_norm": 0.441191703081131,
      "learning_rate": 9.082933583650691e-06,
      "loss": 0.3968,
      "step": 13856
    },
    {
      "epoch": 0.6359047313110917,
      "grad_norm": 0.4150821566581726,
      "learning_rate": 9.082792050697092e-06,
      "loss": 0.3311,
      "step": 13857
    },
    {
      "epoch": 0.6359506218163462,
      "grad_norm": 0.45145383477211,
      "learning_rate": 9.082650507925669e-06,
      "loss": 0.3701,
      "step": 13858
    },
    {
      "epoch": 0.6359965123216007,
      "grad_norm": 0.47092947363853455,
      "learning_rate": 9.082508955336759e-06,
      "loss": 0.3701,
      "step": 13859
    },
    {
      "epoch": 0.6360424028268551,
      "grad_norm": 0.5156068205833435,
      "learning_rate": 9.082367392930704e-06,
      "loss": 0.4711,
      "step": 13860
    },
    {
      "epoch": 0.6360882933321096,
      "grad_norm": 0.47675833106040955,
      "learning_rate": 9.082225820707844e-06,
      "loss": 0.4824,
      "step": 13861
    },
    {
      "epoch": 0.6361341838373641,
      "grad_norm": 0.42623695731163025,
      "learning_rate": 9.082084238668518e-06,
      "loss": 0.338,
      "step": 13862
    },
    {
      "epoch": 0.6361800743426185,
      "grad_norm": 0.5204612016677856,
      "learning_rate": 9.081942646813068e-06,
      "loss": 0.518,
      "step": 13863
    },
    {
      "epoch": 0.636225964847873,
      "grad_norm": 0.48516273498535156,
      "learning_rate": 9.081801045141834e-06,
      "loss": 0.4623,
      "step": 13864
    },
    {
      "epoch": 0.6362718553531275,
      "grad_norm": 0.4353639483451843,
      "learning_rate": 9.081659433655157e-06,
      "loss": 0.3804,
      "step": 13865
    },
    {
      "epoch": 0.6363177458583819,
      "grad_norm": 0.4441174566745758,
      "learning_rate": 9.081517812353378e-06,
      "loss": 0.4093,
      "step": 13866
    },
    {
      "epoch": 0.6363636363636364,
      "grad_norm": 0.4391174912452698,
      "learning_rate": 9.081376181236838e-06,
      "loss": 0.355,
      "step": 13867
    },
    {
      "epoch": 0.6364095268688909,
      "grad_norm": 0.4572235643863678,
      "learning_rate": 9.081234540305875e-06,
      "loss": 0.3853,
      "step": 13868
    },
    {
      "epoch": 0.6364554173741452,
      "grad_norm": 0.501128077507019,
      "learning_rate": 9.08109288956083e-06,
      "loss": 0.5306,
      "step": 13869
    },
    {
      "epoch": 0.6365013078793997,
      "grad_norm": 0.4364604949951172,
      "learning_rate": 9.080951229002046e-06,
      "loss": 0.3854,
      "step": 13870
    },
    {
      "epoch": 0.6365471983846542,
      "grad_norm": 0.4568452835083008,
      "learning_rate": 9.080809558629861e-06,
      "loss": 0.4229,
      "step": 13871
    },
    {
      "epoch": 0.6365930888899087,
      "grad_norm": 0.4498665928840637,
      "learning_rate": 9.080667878444618e-06,
      "loss": 0.401,
      "step": 13872
    },
    {
      "epoch": 0.6366389793951631,
      "grad_norm": 0.46098843216896057,
      "learning_rate": 9.080526188446657e-06,
      "loss": 0.4135,
      "step": 13873
    },
    {
      "epoch": 0.6366848699004176,
      "grad_norm": 0.44554442167282104,
      "learning_rate": 9.080384488636317e-06,
      "loss": 0.3927,
      "step": 13874
    },
    {
      "epoch": 0.6367307604056721,
      "grad_norm": 0.45249930024147034,
      "learning_rate": 9.080242779013941e-06,
      "loss": 0.37,
      "step": 13875
    },
    {
      "epoch": 0.6367766509109265,
      "grad_norm": 0.4642948806285858,
      "learning_rate": 9.08010105957987e-06,
      "loss": 0.3913,
      "step": 13876
    },
    {
      "epoch": 0.636822541416181,
      "grad_norm": 0.46328550577163696,
      "learning_rate": 9.079959330334442e-06,
      "loss": 0.3425,
      "step": 13877
    },
    {
      "epoch": 0.6368684319214355,
      "grad_norm": 0.4690941572189331,
      "learning_rate": 9.079817591278e-06,
      "loss": 0.4086,
      "step": 13878
    },
    {
      "epoch": 0.6369143224266899,
      "grad_norm": 0.46301230788230896,
      "learning_rate": 9.079675842410884e-06,
      "loss": 0.4015,
      "step": 13879
    },
    {
      "epoch": 0.6369602129319444,
      "grad_norm": 0.4607619345188141,
      "learning_rate": 9.079534083733435e-06,
      "loss": 0.4079,
      "step": 13880
    },
    {
      "epoch": 0.6370061034371989,
      "grad_norm": 0.4334520995616913,
      "learning_rate": 9.079392315245994e-06,
      "loss": 0.4022,
      "step": 13881
    },
    {
      "epoch": 0.6370519939424533,
      "grad_norm": 0.45250433683395386,
      "learning_rate": 9.079250536948905e-06,
      "loss": 0.3221,
      "step": 13882
    },
    {
      "epoch": 0.6370978844477078,
      "grad_norm": 0.4378756880760193,
      "learning_rate": 9.079108748842502e-06,
      "loss": 0.3635,
      "step": 13883
    },
    {
      "epoch": 0.6371437749529623,
      "grad_norm": 0.43111056089401245,
      "learning_rate": 9.078966950927131e-06,
      "loss": 0.3332,
      "step": 13884
    },
    {
      "epoch": 0.6371896654582166,
      "grad_norm": 0.46981289982795715,
      "learning_rate": 9.07882514320313e-06,
      "loss": 0.4495,
      "step": 13885
    },
    {
      "epoch": 0.6372355559634711,
      "grad_norm": 0.45617085695266724,
      "learning_rate": 9.078683325670846e-06,
      "loss": 0.3945,
      "step": 13886
    },
    {
      "epoch": 0.6372814464687256,
      "grad_norm": 0.435811847448349,
      "learning_rate": 9.078541498330613e-06,
      "loss": 0.3651,
      "step": 13887
    },
    {
      "epoch": 0.63732733697398,
      "grad_norm": 0.4522098898887634,
      "learning_rate": 9.078399661182777e-06,
      "loss": 0.3725,
      "step": 13888
    },
    {
      "epoch": 0.6373732274792345,
      "grad_norm": 0.44129475951194763,
      "learning_rate": 9.078257814227676e-06,
      "loss": 0.3686,
      "step": 13889
    },
    {
      "epoch": 0.637419117984489,
      "grad_norm": 0.44493645429611206,
      "learning_rate": 9.078115957465652e-06,
      "loss": 0.3638,
      "step": 13890
    },
    {
      "epoch": 0.6374650084897435,
      "grad_norm": 0.4450368583202362,
      "learning_rate": 9.077974090897046e-06,
      "loss": 0.3894,
      "step": 13891
    },
    {
      "epoch": 0.6375108989949979,
      "grad_norm": 0.4202248752117157,
      "learning_rate": 9.077832214522199e-06,
      "loss": 0.3365,
      "step": 13892
    },
    {
      "epoch": 0.6375567895002524,
      "grad_norm": 0.4527968764305115,
      "learning_rate": 9.077690328341453e-06,
      "loss": 0.3608,
      "step": 13893
    },
    {
      "epoch": 0.6376026800055069,
      "grad_norm": 0.45292168855667114,
      "learning_rate": 9.077548432355148e-06,
      "loss": 0.4014,
      "step": 13894
    },
    {
      "epoch": 0.6376485705107613,
      "grad_norm": 0.42058977484703064,
      "learning_rate": 9.077406526563627e-06,
      "loss": 0.3322,
      "step": 13895
    },
    {
      "epoch": 0.6376944610160158,
      "grad_norm": 0.41648879647254944,
      "learning_rate": 9.07726461096723e-06,
      "loss": 0.3272,
      "step": 13896
    },
    {
      "epoch": 0.6377403515212703,
      "grad_norm": 0.49815255403518677,
      "learning_rate": 9.077122685566297e-06,
      "loss": 0.4135,
      "step": 13897
    },
    {
      "epoch": 0.6377862420265247,
      "grad_norm": 0.49946486949920654,
      "learning_rate": 9.076980750361174e-06,
      "loss": 0.4479,
      "step": 13898
    },
    {
      "epoch": 0.6378321325317792,
      "grad_norm": 0.4121783375740051,
      "learning_rate": 9.076838805352196e-06,
      "loss": 0.3214,
      "step": 13899
    },
    {
      "epoch": 0.6378780230370337,
      "grad_norm": 0.45106053352355957,
      "learning_rate": 9.076696850539707e-06,
      "loss": 0.393,
      "step": 13900
    },
    {
      "epoch": 0.6379239135422881,
      "grad_norm": 0.5474677681922913,
      "learning_rate": 9.076554885924051e-06,
      "loss": 0.5361,
      "step": 13901
    },
    {
      "epoch": 0.6379698040475426,
      "grad_norm": 0.5133782029151917,
      "learning_rate": 9.076412911505566e-06,
      "loss": 0.4563,
      "step": 13902
    },
    {
      "epoch": 0.638015694552797,
      "grad_norm": 0.43608003854751587,
      "learning_rate": 9.076270927284594e-06,
      "loss": 0.3838,
      "step": 13903
    },
    {
      "epoch": 0.6380615850580514,
      "grad_norm": 0.4868790805339813,
      "learning_rate": 9.076128933261478e-06,
      "loss": 0.4459,
      "step": 13904
    },
    {
      "epoch": 0.6381074755633059,
      "grad_norm": 0.4699779152870178,
      "learning_rate": 9.075986929436557e-06,
      "loss": 0.3611,
      "step": 13905
    },
    {
      "epoch": 0.6381533660685604,
      "grad_norm": 0.4602355360984802,
      "learning_rate": 9.075844915810175e-06,
      "loss": 0.3695,
      "step": 13906
    },
    {
      "epoch": 0.6381992565738149,
      "grad_norm": 0.46012914180755615,
      "learning_rate": 9.075702892382673e-06,
      "loss": 0.4332,
      "step": 13907
    },
    {
      "epoch": 0.6382451470790693,
      "grad_norm": 0.4306870698928833,
      "learning_rate": 9.07556085915439e-06,
      "loss": 0.3671,
      "step": 13908
    },
    {
      "epoch": 0.6382910375843238,
      "grad_norm": 0.4583587348461151,
      "learning_rate": 9.075418816125673e-06,
      "loss": 0.394,
      "step": 13909
    },
    {
      "epoch": 0.6383369280895783,
      "grad_norm": 0.4546503722667694,
      "learning_rate": 9.075276763296856e-06,
      "loss": 0.4076,
      "step": 13910
    },
    {
      "epoch": 0.6383828185948327,
      "grad_norm": 0.4136744439601898,
      "learning_rate": 9.075134700668285e-06,
      "loss": 0.3242,
      "step": 13911
    },
    {
      "epoch": 0.6384287091000872,
      "grad_norm": 0.45430830121040344,
      "learning_rate": 9.074992628240302e-06,
      "loss": 0.4446,
      "step": 13912
    },
    {
      "epoch": 0.6384745996053417,
      "grad_norm": 0.4768562614917755,
      "learning_rate": 9.074850546013249e-06,
      "loss": 0.4077,
      "step": 13913
    },
    {
      "epoch": 0.6385204901105961,
      "grad_norm": 0.43960222601890564,
      "learning_rate": 9.074708453987466e-06,
      "loss": 0.3533,
      "step": 13914
    },
    {
      "epoch": 0.6385663806158506,
      "grad_norm": 0.4887329339981079,
      "learning_rate": 9.074566352163294e-06,
      "loss": 0.4536,
      "step": 13915
    },
    {
      "epoch": 0.6386122711211051,
      "grad_norm": 0.4777091145515442,
      "learning_rate": 9.074424240541077e-06,
      "loss": 0.3993,
      "step": 13916
    },
    {
      "epoch": 0.6386581616263595,
      "grad_norm": 0.44843190908432007,
      "learning_rate": 9.074282119121155e-06,
      "loss": 0.4163,
      "step": 13917
    },
    {
      "epoch": 0.638704052131614,
      "grad_norm": 0.4648585617542267,
      "learning_rate": 9.074139987903871e-06,
      "loss": 0.4279,
      "step": 13918
    },
    {
      "epoch": 0.6387499426368685,
      "grad_norm": 0.4809342622756958,
      "learning_rate": 9.073997846889563e-06,
      "loss": 0.4374,
      "step": 13919
    },
    {
      "epoch": 0.6387958331421228,
      "grad_norm": 0.42647239565849304,
      "learning_rate": 9.07385569607858e-06,
      "loss": 0.3586,
      "step": 13920
    },
    {
      "epoch": 0.6388417236473773,
      "grad_norm": 0.43390122056007385,
      "learning_rate": 9.07371353547126e-06,
      "loss": 0.3712,
      "step": 13921
    },
    {
      "epoch": 0.6388876141526318,
      "grad_norm": 0.4528542459011078,
      "learning_rate": 9.073571365067942e-06,
      "loss": 0.4075,
      "step": 13922
    },
    {
      "epoch": 0.6389335046578862,
      "grad_norm": 0.4658913016319275,
      "learning_rate": 9.073429184868972e-06,
      "loss": 0.4395,
      "step": 13923
    },
    {
      "epoch": 0.6389793951631407,
      "grad_norm": 0.4689754247665405,
      "learning_rate": 9.073286994874691e-06,
      "loss": 0.367,
      "step": 13924
    },
    {
      "epoch": 0.6390252856683952,
      "grad_norm": 0.4604182243347168,
      "learning_rate": 9.073144795085438e-06,
      "loss": 0.4074,
      "step": 13925
    },
    {
      "epoch": 0.6390711761736497,
      "grad_norm": 0.4281062185764313,
      "learning_rate": 9.073002585501558e-06,
      "loss": 0.3907,
      "step": 13926
    },
    {
      "epoch": 0.6391170666789041,
      "grad_norm": 0.4650721251964569,
      "learning_rate": 9.072860366123394e-06,
      "loss": 0.3575,
      "step": 13927
    },
    {
      "epoch": 0.6391629571841586,
      "grad_norm": 0.42986929416656494,
      "learning_rate": 9.072718136951287e-06,
      "loss": 0.3582,
      "step": 13928
    },
    {
      "epoch": 0.6392088476894131,
      "grad_norm": 0.4436732530593872,
      "learning_rate": 9.072575897985576e-06,
      "loss": 0.331,
      "step": 13929
    },
    {
      "epoch": 0.6392547381946675,
      "grad_norm": 0.4522177278995514,
      "learning_rate": 9.072433649226603e-06,
      "loss": 0.3716,
      "step": 13930
    },
    {
      "epoch": 0.639300628699922,
      "grad_norm": 0.45224958658218384,
      "learning_rate": 9.072291390674717e-06,
      "loss": 0.4476,
      "step": 13931
    },
    {
      "epoch": 0.6393465192051765,
      "grad_norm": 0.4480723738670349,
      "learning_rate": 9.072149122330253e-06,
      "loss": 0.369,
      "step": 13932
    },
    {
      "epoch": 0.6393924097104309,
      "grad_norm": 0.48439711332321167,
      "learning_rate": 9.072006844193558e-06,
      "loss": 0.3625,
      "step": 13933
    },
    {
      "epoch": 0.6394383002156854,
      "grad_norm": 0.45903655886650085,
      "learning_rate": 9.071864556264968e-06,
      "loss": 0.376,
      "step": 13934
    },
    {
      "epoch": 0.6394841907209399,
      "grad_norm": 0.4892946183681488,
      "learning_rate": 9.071722258544832e-06,
      "loss": 0.3771,
      "step": 13935
    },
    {
      "epoch": 0.6395300812261943,
      "grad_norm": 0.5190251469612122,
      "learning_rate": 9.071579951033486e-06,
      "loss": 0.5222,
      "step": 13936
    },
    {
      "epoch": 0.6395759717314488,
      "grad_norm": 0.48577019572257996,
      "learning_rate": 9.071437633731278e-06,
      "loss": 0.3985,
      "step": 13937
    },
    {
      "epoch": 0.6396218622367033,
      "grad_norm": 0.5002421140670776,
      "learning_rate": 9.071295306638546e-06,
      "loss": 0.3798,
      "step": 13938
    },
    {
      "epoch": 0.6396677527419576,
      "grad_norm": 0.488899827003479,
      "learning_rate": 9.071152969755635e-06,
      "loss": 0.4324,
      "step": 13939
    },
    {
      "epoch": 0.6397136432472121,
      "grad_norm": 0.43541520833969116,
      "learning_rate": 9.071010623082885e-06,
      "loss": 0.3299,
      "step": 13940
    },
    {
      "epoch": 0.6397595337524666,
      "grad_norm": 0.4609094560146332,
      "learning_rate": 9.070868266620641e-06,
      "loss": 0.4445,
      "step": 13941
    },
    {
      "epoch": 0.639805424257721,
      "grad_norm": 0.48501887917518616,
      "learning_rate": 9.070725900369241e-06,
      "loss": 0.4542,
      "step": 13942
    },
    {
      "epoch": 0.6398513147629755,
      "grad_norm": 0.42530471086502075,
      "learning_rate": 9.070583524329032e-06,
      "loss": 0.3468,
      "step": 13943
    },
    {
      "epoch": 0.63989720526823,
      "grad_norm": 0.4728492498397827,
      "learning_rate": 9.070441138500354e-06,
      "loss": 0.4661,
      "step": 13944
    },
    {
      "epoch": 0.6399430957734845,
      "grad_norm": 0.4799095690250397,
      "learning_rate": 9.070298742883548e-06,
      "loss": 0.4524,
      "step": 13945
    },
    {
      "epoch": 0.6399889862787389,
      "grad_norm": 0.47106826305389404,
      "learning_rate": 9.07015633747896e-06,
      "loss": 0.3504,
      "step": 13946
    },
    {
      "epoch": 0.6400348767839934,
      "grad_norm": 0.43228691816329956,
      "learning_rate": 9.070013922286929e-06,
      "loss": 0.3619,
      "step": 13947
    },
    {
      "epoch": 0.6400807672892479,
      "grad_norm": 0.47381365299224854,
      "learning_rate": 9.0698714973078e-06,
      "loss": 0.4266,
      "step": 13948
    },
    {
      "epoch": 0.6401266577945023,
      "grad_norm": 0.49155116081237793,
      "learning_rate": 9.069729062541914e-06,
      "loss": 0.4674,
      "step": 13949
    },
    {
      "epoch": 0.6401725482997568,
      "grad_norm": 0.45604097843170166,
      "learning_rate": 9.069586617989614e-06,
      "loss": 0.3445,
      "step": 13950
    },
    {
      "epoch": 0.6402184388050113,
      "grad_norm": 0.4782208502292633,
      "learning_rate": 9.069444163651245e-06,
      "loss": 0.4567,
      "step": 13951
    },
    {
      "epoch": 0.6402643293102657,
      "grad_norm": 0.426633358001709,
      "learning_rate": 9.069301699527144e-06,
      "loss": 0.3235,
      "step": 13952
    },
    {
      "epoch": 0.6403102198155202,
      "grad_norm": 0.463579922914505,
      "learning_rate": 9.069159225617658e-06,
      "loss": 0.4393,
      "step": 13953
    },
    {
      "epoch": 0.6403561103207747,
      "grad_norm": 0.4582008123397827,
      "learning_rate": 9.069016741923129e-06,
      "loss": 0.3854,
      "step": 13954
    },
    {
      "epoch": 0.640402000826029,
      "grad_norm": 0.4605084955692291,
      "learning_rate": 9.068874248443899e-06,
      "loss": 0.3833,
      "step": 13955
    },
    {
      "epoch": 0.6404478913312835,
      "grad_norm": 0.48323988914489746,
      "learning_rate": 9.06873174518031e-06,
      "loss": 0.4414,
      "step": 13956
    },
    {
      "epoch": 0.640493781836538,
      "grad_norm": 0.4243892729282379,
      "learning_rate": 9.068589232132704e-06,
      "loss": 0.3501,
      "step": 13957
    },
    {
      "epoch": 0.6405396723417924,
      "grad_norm": 0.44702792167663574,
      "learning_rate": 9.068446709301428e-06,
      "loss": 0.4225,
      "step": 13958
    },
    {
      "epoch": 0.6405855628470469,
      "grad_norm": 0.426170289516449,
      "learning_rate": 9.06830417668682e-06,
      "loss": 0.3419,
      "step": 13959
    },
    {
      "epoch": 0.6406314533523014,
      "grad_norm": 0.53554368019104,
      "learning_rate": 9.068161634289224e-06,
      "loss": 0.5226,
      "step": 13960
    },
    {
      "epoch": 0.6406773438575559,
      "grad_norm": 0.49567604064941406,
      "learning_rate": 9.068019082108982e-06,
      "loss": 0.4261,
      "step": 13961
    },
    {
      "epoch": 0.6407232343628103,
      "grad_norm": 0.4362735152244568,
      "learning_rate": 9.06787652014644e-06,
      "loss": 0.3433,
      "step": 13962
    },
    {
      "epoch": 0.6407691248680648,
      "grad_norm": 0.473021000623703,
      "learning_rate": 9.067733948401939e-06,
      "loss": 0.4611,
      "step": 13963
    },
    {
      "epoch": 0.6408150153733193,
      "grad_norm": 0.4598488211631775,
      "learning_rate": 9.06759136687582e-06,
      "loss": 0.3457,
      "step": 13964
    },
    {
      "epoch": 0.6408609058785737,
      "grad_norm": 0.4515002965927124,
      "learning_rate": 9.067448775568429e-06,
      "loss": 0.322,
      "step": 13965
    },
    {
      "epoch": 0.6409067963838282,
      "grad_norm": 0.4563615322113037,
      "learning_rate": 9.067306174480109e-06,
      "loss": 0.4141,
      "step": 13966
    },
    {
      "epoch": 0.6409526868890827,
      "grad_norm": 0.48475298285484314,
      "learning_rate": 9.067163563611197e-06,
      "loss": 0.4492,
      "step": 13967
    },
    {
      "epoch": 0.6409985773943371,
      "grad_norm": 0.4568878710269928,
      "learning_rate": 9.067020942962044e-06,
      "loss": 0.4575,
      "step": 13968
    },
    {
      "epoch": 0.6410444678995916,
      "grad_norm": 0.48156845569610596,
      "learning_rate": 9.066878312532987e-06,
      "loss": 0.4488,
      "step": 13969
    },
    {
      "epoch": 0.6410903584048461,
      "grad_norm": 0.44092920422554016,
      "learning_rate": 9.066735672324372e-06,
      "loss": 0.3715,
      "step": 13970
    },
    {
      "epoch": 0.6411362489101005,
      "grad_norm": 0.5198221206665039,
      "learning_rate": 9.06659302233654e-06,
      "loss": 0.4106,
      "step": 13971
    },
    {
      "epoch": 0.641182139415355,
      "grad_norm": 0.46925362944602966,
      "learning_rate": 9.06645036256984e-06,
      "loss": 0.3715,
      "step": 13972
    },
    {
      "epoch": 0.6412280299206095,
      "grad_norm": 0.45931297540664673,
      "learning_rate": 9.066307693024606e-06,
      "loss": 0.3282,
      "step": 13973
    },
    {
      "epoch": 0.6412739204258638,
      "grad_norm": 0.4393952786922455,
      "learning_rate": 9.066165013701185e-06,
      "loss": 0.322,
      "step": 13974
    },
    {
      "epoch": 0.6413198109311183,
      "grad_norm": 0.4711958169937134,
      "learning_rate": 9.066022324599921e-06,
      "loss": 0.3803,
      "step": 13975
    },
    {
      "epoch": 0.6413657014363728,
      "grad_norm": 0.4949358105659485,
      "learning_rate": 9.065879625721156e-06,
      "loss": 0.4167,
      "step": 13976
    },
    {
      "epoch": 0.6414115919416272,
      "grad_norm": 0.46667972207069397,
      "learning_rate": 9.065736917065237e-06,
      "loss": 0.4286,
      "step": 13977
    },
    {
      "epoch": 0.6414574824468817,
      "grad_norm": 0.4384036064147949,
      "learning_rate": 9.065594198632501e-06,
      "loss": 0.3634,
      "step": 13978
    },
    {
      "epoch": 0.6415033729521362,
      "grad_norm": 0.4735531508922577,
      "learning_rate": 9.065451470423294e-06,
      "loss": 0.4101,
      "step": 13979
    },
    {
      "epoch": 0.6415492634573907,
      "grad_norm": 0.48377203941345215,
      "learning_rate": 9.065308732437961e-06,
      "loss": 0.3827,
      "step": 13980
    },
    {
      "epoch": 0.6415951539626451,
      "grad_norm": 0.44137996435165405,
      "learning_rate": 9.065165984676843e-06,
      "loss": 0.3439,
      "step": 13981
    },
    {
      "epoch": 0.6416410444678996,
      "grad_norm": 0.4500725269317627,
      "learning_rate": 9.065023227140284e-06,
      "loss": 0.4202,
      "step": 13982
    },
    {
      "epoch": 0.6416869349731541,
      "grad_norm": 0.47660550475120544,
      "learning_rate": 9.064880459828626e-06,
      "loss": 0.3925,
      "step": 13983
    },
    {
      "epoch": 0.6417328254784085,
      "grad_norm": 0.43130001425743103,
      "learning_rate": 9.064737682742214e-06,
      "loss": 0.3211,
      "step": 13984
    },
    {
      "epoch": 0.641778715983663,
      "grad_norm": 0.46992093324661255,
      "learning_rate": 9.064594895881391e-06,
      "loss": 0.417,
      "step": 13985
    },
    {
      "epoch": 0.6418246064889175,
      "grad_norm": 0.4720174968242645,
      "learning_rate": 9.0644520992465e-06,
      "loss": 0.5006,
      "step": 13986
    },
    {
      "epoch": 0.6418704969941719,
      "grad_norm": 0.4534390866756439,
      "learning_rate": 9.064309292837884e-06,
      "loss": 0.3778,
      "step": 13987
    },
    {
      "epoch": 0.6419163874994264,
      "grad_norm": 0.4395870566368103,
      "learning_rate": 9.064166476655888e-06,
      "loss": 0.3524,
      "step": 13988
    },
    {
      "epoch": 0.6419622780046809,
      "grad_norm": 0.47124695777893066,
      "learning_rate": 9.064023650700852e-06,
      "loss": 0.4288,
      "step": 13989
    },
    {
      "epoch": 0.6420081685099353,
      "grad_norm": 0.43165233731269836,
      "learning_rate": 9.063880814973125e-06,
      "loss": 0.3405,
      "step": 13990
    },
    {
      "epoch": 0.6420540590151897,
      "grad_norm": 0.47706490755081177,
      "learning_rate": 9.063737969473046e-06,
      "loss": 0.3841,
      "step": 13991
    },
    {
      "epoch": 0.6420999495204442,
      "grad_norm": 0.4723082184791565,
      "learning_rate": 9.063595114200958e-06,
      "loss": 0.383,
      "step": 13992
    },
    {
      "epoch": 0.6421458400256986,
      "grad_norm": 0.5112971663475037,
      "learning_rate": 9.063452249157207e-06,
      "loss": 0.5565,
      "step": 13993
    },
    {
      "epoch": 0.6421917305309531,
      "grad_norm": 0.43778523802757263,
      "learning_rate": 9.063309374342138e-06,
      "loss": 0.3497,
      "step": 13994
    },
    {
      "epoch": 0.6422376210362076,
      "grad_norm": 0.41746553778648376,
      "learning_rate": 9.06316648975609e-06,
      "loss": 0.3253,
      "step": 13995
    },
    {
      "epoch": 0.6422835115414621,
      "grad_norm": 0.4504404067993164,
      "learning_rate": 9.063023595399409e-06,
      "loss": 0.3874,
      "step": 13996
    },
    {
      "epoch": 0.6423294020467165,
      "grad_norm": 0.454401433467865,
      "learning_rate": 9.06288069127244e-06,
      "loss": 0.3804,
      "step": 13997
    },
    {
      "epoch": 0.642375292551971,
      "grad_norm": 0.44376546144485474,
      "learning_rate": 9.062737777375523e-06,
      "loss": 0.3351,
      "step": 13998
    },
    {
      "epoch": 0.6424211830572255,
      "grad_norm": 0.4739118218421936,
      "learning_rate": 9.062594853709006e-06,
      "loss": 0.4415,
      "step": 13999
    },
    {
      "epoch": 0.6424670735624799,
      "grad_norm": 0.4341551959514618,
      "learning_rate": 9.06245192027323e-06,
      "loss": 0.3624,
      "step": 14000
    },
    {
      "epoch": 0.6425129640677344,
      "grad_norm": 0.45949438214302063,
      "learning_rate": 9.062308977068539e-06,
      "loss": 0.4062,
      "step": 14001
    },
    {
      "epoch": 0.6425588545729889,
      "grad_norm": 0.4945415258407593,
      "learning_rate": 9.062166024095278e-06,
      "loss": 0.4361,
      "step": 14002
    },
    {
      "epoch": 0.6426047450782433,
      "grad_norm": 0.46946659684181213,
      "learning_rate": 9.062023061353788e-06,
      "loss": 0.428,
      "step": 14003
    },
    {
      "epoch": 0.6426506355834978,
      "grad_norm": 0.47292548418045044,
      "learning_rate": 9.061880088844418e-06,
      "loss": 0.4204,
      "step": 14004
    },
    {
      "epoch": 0.6426965260887523,
      "grad_norm": 0.4691673815250397,
      "learning_rate": 9.061737106567505e-06,
      "loss": 0.4077,
      "step": 14005
    },
    {
      "epoch": 0.6427424165940067,
      "grad_norm": 0.4639378488063812,
      "learning_rate": 9.061594114523395e-06,
      "loss": 0.3308,
      "step": 14006
    },
    {
      "epoch": 0.6427883070992612,
      "grad_norm": 0.5107695460319519,
      "learning_rate": 9.061451112712435e-06,
      "loss": 0.47,
      "step": 14007
    },
    {
      "epoch": 0.6428341976045157,
      "grad_norm": 0.4573849141597748,
      "learning_rate": 9.06130810113497e-06,
      "loss": 0.3584,
      "step": 14008
    },
    {
      "epoch": 0.64288008810977,
      "grad_norm": 0.415199339389801,
      "learning_rate": 9.061165079791337e-06,
      "loss": 0.3335,
      "step": 14009
    },
    {
      "epoch": 0.6429259786150245,
      "grad_norm": 0.45390504598617554,
      "learning_rate": 9.061022048681884e-06,
      "loss": 0.4357,
      "step": 14010
    },
    {
      "epoch": 0.642971869120279,
      "grad_norm": 0.45877087116241455,
      "learning_rate": 9.060879007806956e-06,
      "loss": 0.3778,
      "step": 14011
    },
    {
      "epoch": 0.6430177596255334,
      "grad_norm": 0.5054169297218323,
      "learning_rate": 9.060735957166894e-06,
      "loss": 0.433,
      "step": 14012
    },
    {
      "epoch": 0.6430636501307879,
      "grad_norm": 0.4727879762649536,
      "learning_rate": 9.060592896762044e-06,
      "loss": 0.4213,
      "step": 14013
    },
    {
      "epoch": 0.6431095406360424,
      "grad_norm": 0.43782058358192444,
      "learning_rate": 9.06044982659275e-06,
      "loss": 0.331,
      "step": 14014
    },
    {
      "epoch": 0.6431554311412969,
      "grad_norm": 0.4261011481285095,
      "learning_rate": 9.060306746659357e-06,
      "loss": 0.347,
      "step": 14015
    },
    {
      "epoch": 0.6432013216465513,
      "grad_norm": 0.4498082995414734,
      "learning_rate": 9.060163656962207e-06,
      "loss": 0.3799,
      "step": 14016
    },
    {
      "epoch": 0.6432472121518058,
      "grad_norm": 0.49998435378074646,
      "learning_rate": 9.060020557501644e-06,
      "loss": 0.4703,
      "step": 14017
    },
    {
      "epoch": 0.6432931026570603,
      "grad_norm": 0.43560680747032166,
      "learning_rate": 9.059877448278013e-06,
      "loss": 0.342,
      "step": 14018
    },
    {
      "epoch": 0.6433389931623147,
      "grad_norm": 0.49742043018341064,
      "learning_rate": 9.059734329291658e-06,
      "loss": 0.4222,
      "step": 14019
    },
    {
      "epoch": 0.6433848836675692,
      "grad_norm": 0.5405114889144897,
      "learning_rate": 9.059591200542923e-06,
      "loss": 0.523,
      "step": 14020
    },
    {
      "epoch": 0.6434307741728237,
      "grad_norm": 0.5079758167266846,
      "learning_rate": 9.059448062032153e-06,
      "loss": 0.4544,
      "step": 14021
    },
    {
      "epoch": 0.6434766646780781,
      "grad_norm": 0.4347251057624817,
      "learning_rate": 9.059304913759693e-06,
      "loss": 0.3731,
      "step": 14022
    },
    {
      "epoch": 0.6435225551833326,
      "grad_norm": 0.45435693860054016,
      "learning_rate": 9.059161755725883e-06,
      "loss": 0.4245,
      "step": 14023
    },
    {
      "epoch": 0.6435684456885871,
      "grad_norm": 0.49252989888191223,
      "learning_rate": 9.059018587931073e-06,
      "loss": 0.4438,
      "step": 14024
    },
    {
      "epoch": 0.6436143361938415,
      "grad_norm": 0.5004079341888428,
      "learning_rate": 9.058875410375602e-06,
      "loss": 0.4749,
      "step": 14025
    },
    {
      "epoch": 0.643660226699096,
      "grad_norm": 0.47762390971183777,
      "learning_rate": 9.058732223059817e-06,
      "loss": 0.4307,
      "step": 14026
    },
    {
      "epoch": 0.6437061172043504,
      "grad_norm": 0.47857779264450073,
      "learning_rate": 9.058589025984064e-06,
      "loss": 0.3567,
      "step": 14027
    },
    {
      "epoch": 0.6437520077096048,
      "grad_norm": 0.43675971031188965,
      "learning_rate": 9.058445819148683e-06,
      "loss": 0.3597,
      "step": 14028
    },
    {
      "epoch": 0.6437978982148593,
      "grad_norm": 0.4836585819721222,
      "learning_rate": 9.05830260255402e-06,
      "loss": 0.5072,
      "step": 14029
    },
    {
      "epoch": 0.6438437887201138,
      "grad_norm": 0.47460660338401794,
      "learning_rate": 9.058159376200424e-06,
      "loss": 0.4011,
      "step": 14030
    },
    {
      "epoch": 0.6438896792253682,
      "grad_norm": 0.4512506127357483,
      "learning_rate": 9.058016140088232e-06,
      "loss": 0.4144,
      "step": 14031
    },
    {
      "epoch": 0.6439355697306227,
      "grad_norm": 0.4772023558616638,
      "learning_rate": 9.057872894217794e-06,
      "loss": 0.429,
      "step": 14032
    },
    {
      "epoch": 0.6439814602358772,
      "grad_norm": 0.4694622755050659,
      "learning_rate": 9.057729638589451e-06,
      "loss": 0.3801,
      "step": 14033
    },
    {
      "epoch": 0.6440273507411317,
      "grad_norm": 0.45099690556526184,
      "learning_rate": 9.057586373203548e-06,
      "loss": 0.4225,
      "step": 14034
    },
    {
      "epoch": 0.6440732412463861,
      "grad_norm": 0.460796058177948,
      "learning_rate": 9.057443098060432e-06,
      "loss": 0.3942,
      "step": 14035
    },
    {
      "epoch": 0.6441191317516406,
      "grad_norm": 0.45044317841529846,
      "learning_rate": 9.057299813160446e-06,
      "loss": 0.3983,
      "step": 14036
    },
    {
      "epoch": 0.6441650222568951,
      "grad_norm": 0.44518229365348816,
      "learning_rate": 9.057156518503934e-06,
      "loss": 0.3815,
      "step": 14037
    },
    {
      "epoch": 0.6442109127621495,
      "grad_norm": 0.45808717608451843,
      "learning_rate": 9.057013214091242e-06,
      "loss": 0.4432,
      "step": 14038
    },
    {
      "epoch": 0.644256803267404,
      "grad_norm": 0.4451320469379425,
      "learning_rate": 9.056869899922713e-06,
      "loss": 0.3604,
      "step": 14039
    },
    {
      "epoch": 0.6443026937726585,
      "grad_norm": 0.4737377464771271,
      "learning_rate": 9.056726575998691e-06,
      "loss": 0.3932,
      "step": 14040
    },
    {
      "epoch": 0.6443485842779129,
      "grad_norm": 0.4431821405887604,
      "learning_rate": 9.056583242319524e-06,
      "loss": 0.3491,
      "step": 14041
    },
    {
      "epoch": 0.6443944747831674,
      "grad_norm": 0.48313918709754944,
      "learning_rate": 9.056439898885553e-06,
      "loss": 0.43,
      "step": 14042
    },
    {
      "epoch": 0.6444403652884219,
      "grad_norm": 0.4375656843185425,
      "learning_rate": 9.056296545697125e-06,
      "loss": 0.4175,
      "step": 14043
    },
    {
      "epoch": 0.6444862557936762,
      "grad_norm": 0.4607250690460205,
      "learning_rate": 9.056153182754583e-06,
      "loss": 0.3792,
      "step": 14044
    },
    {
      "epoch": 0.6445321462989307,
      "grad_norm": 0.5055986642837524,
      "learning_rate": 9.056009810058274e-06,
      "loss": 0.4739,
      "step": 14045
    },
    {
      "epoch": 0.6445780368041852,
      "grad_norm": 0.43453145027160645,
      "learning_rate": 9.055866427608542e-06,
      "loss": 0.3315,
      "step": 14046
    },
    {
      "epoch": 0.6446239273094396,
      "grad_norm": 0.46851763129234314,
      "learning_rate": 9.05572303540573e-06,
      "loss": 0.3929,
      "step": 14047
    },
    {
      "epoch": 0.6446698178146941,
      "grad_norm": 0.5469868779182434,
      "learning_rate": 9.055579633450184e-06,
      "loss": 0.4833,
      "step": 14048
    },
    {
      "epoch": 0.6447157083199486,
      "grad_norm": 0.40964213013648987,
      "learning_rate": 9.05543622174225e-06,
      "loss": 0.2965,
      "step": 14049
    },
    {
      "epoch": 0.6447615988252031,
      "grad_norm": 0.4448835849761963,
      "learning_rate": 9.05529280028227e-06,
      "loss": 0.3719,
      "step": 14050
    },
    {
      "epoch": 0.6448074893304575,
      "grad_norm": 0.4760794937610626,
      "learning_rate": 9.055149369070593e-06,
      "loss": 0.3891,
      "step": 14051
    },
    {
      "epoch": 0.644853379835712,
      "grad_norm": 0.44954636693000793,
      "learning_rate": 9.05500592810756e-06,
      "loss": 0.3798,
      "step": 14052
    },
    {
      "epoch": 0.6448992703409665,
      "grad_norm": 0.44240760803222656,
      "learning_rate": 9.05486247739352e-06,
      "loss": 0.3628,
      "step": 14053
    },
    {
      "epoch": 0.6449451608462209,
      "grad_norm": 0.4264172315597534,
      "learning_rate": 9.054719016928813e-06,
      "loss": 0.378,
      "step": 14054
    },
    {
      "epoch": 0.6449910513514754,
      "grad_norm": 0.4289540946483612,
      "learning_rate": 9.054575546713787e-06,
      "loss": 0.361,
      "step": 14055
    },
    {
      "epoch": 0.6450369418567299,
      "grad_norm": 0.4617115557193756,
      "learning_rate": 9.054432066748787e-06,
      "loss": 0.4149,
      "step": 14056
    },
    {
      "epoch": 0.6450828323619843,
      "grad_norm": 0.4650711119174957,
      "learning_rate": 9.054288577034158e-06,
      "loss": 0.4174,
      "step": 14057
    },
    {
      "epoch": 0.6451287228672388,
      "grad_norm": 0.4586659371852875,
      "learning_rate": 9.054145077570242e-06,
      "loss": 0.3998,
      "step": 14058
    },
    {
      "epoch": 0.6451746133724933,
      "grad_norm": 0.47312062978744507,
      "learning_rate": 9.05400156835739e-06,
      "loss": 0.4246,
      "step": 14059
    },
    {
      "epoch": 0.6452205038777477,
      "grad_norm": 0.43735408782958984,
      "learning_rate": 9.053858049395941e-06,
      "loss": 0.3348,
      "step": 14060
    },
    {
      "epoch": 0.6452663943830022,
      "grad_norm": 0.46725115180015564,
      "learning_rate": 9.053714520686244e-06,
      "loss": 0.493,
      "step": 14061
    },
    {
      "epoch": 0.6453122848882566,
      "grad_norm": 0.43562230467796326,
      "learning_rate": 9.053570982228644e-06,
      "loss": 0.3694,
      "step": 14062
    },
    {
      "epoch": 0.645358175393511,
      "grad_norm": 0.48411309719085693,
      "learning_rate": 9.053427434023483e-06,
      "loss": 0.4414,
      "step": 14063
    },
    {
      "epoch": 0.6454040658987655,
      "grad_norm": 0.42066919803619385,
      "learning_rate": 9.053283876071109e-06,
      "loss": 0.3306,
      "step": 14064
    },
    {
      "epoch": 0.64544995640402,
      "grad_norm": 0.43161800503730774,
      "learning_rate": 9.053140308371866e-06,
      "loss": 0.3478,
      "step": 14065
    },
    {
      "epoch": 0.6454958469092744,
      "grad_norm": 0.49468934535980225,
      "learning_rate": 9.0529967309261e-06,
      "loss": 0.5155,
      "step": 14066
    },
    {
      "epoch": 0.6455417374145289,
      "grad_norm": 0.4748426079750061,
      "learning_rate": 9.052853143734157e-06,
      "loss": 0.429,
      "step": 14067
    },
    {
      "epoch": 0.6455876279197834,
      "grad_norm": 0.4822480380535126,
      "learning_rate": 9.052709546796381e-06,
      "loss": 0.4432,
      "step": 14068
    },
    {
      "epoch": 0.6456335184250379,
      "grad_norm": 0.47248879075050354,
      "learning_rate": 9.052565940113118e-06,
      "loss": 0.4557,
      "step": 14069
    },
    {
      "epoch": 0.6456794089302923,
      "grad_norm": 0.45363280177116394,
      "learning_rate": 9.052422323684713e-06,
      "loss": 0.4107,
      "step": 14070
    },
    {
      "epoch": 0.6457252994355468,
      "grad_norm": 0.48576921224594116,
      "learning_rate": 9.05227869751151e-06,
      "loss": 0.4458,
      "step": 14071
    },
    {
      "epoch": 0.6457711899408013,
      "grad_norm": 0.5302210450172424,
      "learning_rate": 9.052135061593856e-06,
      "loss": 0.5498,
      "step": 14072
    },
    {
      "epoch": 0.6458170804460557,
      "grad_norm": 0.41996899247169495,
      "learning_rate": 9.051991415932098e-06,
      "loss": 0.321,
      "step": 14073
    },
    {
      "epoch": 0.6458629709513102,
      "grad_norm": 0.45224234461784363,
      "learning_rate": 9.051847760526577e-06,
      "loss": 0.4136,
      "step": 14074
    },
    {
      "epoch": 0.6459088614565647,
      "grad_norm": 0.4406568109989166,
      "learning_rate": 9.051704095377642e-06,
      "loss": 0.3732,
      "step": 14075
    },
    {
      "epoch": 0.6459547519618191,
      "grad_norm": 0.45294687151908875,
      "learning_rate": 9.051560420485638e-06,
      "loss": 0.4073,
      "step": 14076
    },
    {
      "epoch": 0.6460006424670736,
      "grad_norm": 0.44497764110565186,
      "learning_rate": 9.051416735850908e-06,
      "loss": 0.3907,
      "step": 14077
    },
    {
      "epoch": 0.6460465329723281,
      "grad_norm": 0.4505990445613861,
      "learning_rate": 9.051273041473801e-06,
      "loss": 0.3679,
      "step": 14078
    },
    {
      "epoch": 0.6460924234775824,
      "grad_norm": 0.4884462356567383,
      "learning_rate": 9.05112933735466e-06,
      "loss": 0.4363,
      "step": 14079
    },
    {
      "epoch": 0.6461383139828369,
      "grad_norm": 0.42937272787094116,
      "learning_rate": 9.050985623493833e-06,
      "loss": 0.3557,
      "step": 14080
    },
    {
      "epoch": 0.6461842044880914,
      "grad_norm": 0.4646994173526764,
      "learning_rate": 9.050841899891663e-06,
      "loss": 0.4127,
      "step": 14081
    },
    {
      "epoch": 0.6462300949933458,
      "grad_norm": 0.4247244596481323,
      "learning_rate": 9.050698166548498e-06,
      "loss": 0.3251,
      "step": 14082
    },
    {
      "epoch": 0.6462759854986003,
      "grad_norm": 0.5226595401763916,
      "learning_rate": 9.05055442346468e-06,
      "loss": 0.4812,
      "step": 14083
    },
    {
      "epoch": 0.6463218760038548,
      "grad_norm": 0.519533097743988,
      "learning_rate": 9.050410670640558e-06,
      "loss": 0.5737,
      "step": 14084
    },
    {
      "epoch": 0.6463677665091093,
      "grad_norm": 0.466232031583786,
      "learning_rate": 9.050266908076476e-06,
      "loss": 0.3895,
      "step": 14085
    },
    {
      "epoch": 0.6464136570143637,
      "grad_norm": 0.4415220618247986,
      "learning_rate": 9.050123135772784e-06,
      "loss": 0.37,
      "step": 14086
    },
    {
      "epoch": 0.6464595475196182,
      "grad_norm": 0.4179003834724426,
      "learning_rate": 9.04997935372982e-06,
      "loss": 0.3231,
      "step": 14087
    },
    {
      "epoch": 0.6465054380248727,
      "grad_norm": 0.4091072380542755,
      "learning_rate": 9.049835561947936e-06,
      "loss": 0.3461,
      "step": 14088
    },
    {
      "epoch": 0.6465513285301271,
      "grad_norm": 0.4746333360671997,
      "learning_rate": 9.049691760427475e-06,
      "loss": 0.4101,
      "step": 14089
    },
    {
      "epoch": 0.6465972190353816,
      "grad_norm": 0.507542073726654,
      "learning_rate": 9.049547949168782e-06,
      "loss": 0.5176,
      "step": 14090
    },
    {
      "epoch": 0.6466431095406361,
      "grad_norm": 0.46834030747413635,
      "learning_rate": 9.049404128172205e-06,
      "loss": 0.4444,
      "step": 14091
    },
    {
      "epoch": 0.6466890000458905,
      "grad_norm": 0.43442246317863464,
      "learning_rate": 9.049260297438092e-06,
      "loss": 0.3612,
      "step": 14092
    },
    {
      "epoch": 0.646734890551145,
      "grad_norm": 0.44404488801956177,
      "learning_rate": 9.049116456966782e-06,
      "loss": 0.3586,
      "step": 14093
    },
    {
      "epoch": 0.6467807810563995,
      "grad_norm": 0.46931061148643494,
      "learning_rate": 9.048972606758627e-06,
      "loss": 0.4024,
      "step": 14094
    },
    {
      "epoch": 0.6468266715616539,
      "grad_norm": 0.46310216188430786,
      "learning_rate": 9.048828746813971e-06,
      "loss": 0.3673,
      "step": 14095
    },
    {
      "epoch": 0.6468725620669084,
      "grad_norm": 0.4568098187446594,
      "learning_rate": 9.048684877133157e-06,
      "loss": 0.4299,
      "step": 14096
    },
    {
      "epoch": 0.6469184525721628,
      "grad_norm": 1.5224976539611816,
      "learning_rate": 9.048540997716537e-06,
      "loss": 0.3141,
      "step": 14097
    },
    {
      "epoch": 0.6469643430774172,
      "grad_norm": 0.44508805871009827,
      "learning_rate": 9.048397108564453e-06,
      "loss": 0.3744,
      "step": 14098
    },
    {
      "epoch": 0.6470102335826717,
      "grad_norm": 0.482736736536026,
      "learning_rate": 9.04825320967725e-06,
      "loss": 0.3953,
      "step": 14099
    },
    {
      "epoch": 0.6470561240879262,
      "grad_norm": 0.4863525927066803,
      "learning_rate": 9.048109301055278e-06,
      "loss": 0.4456,
      "step": 14100
    },
    {
      "epoch": 0.6471020145931806,
      "grad_norm": 0.43306997418403625,
      "learning_rate": 9.047965382698879e-06,
      "loss": 0.3105,
      "step": 14101
    },
    {
      "epoch": 0.6471479050984351,
      "grad_norm": 0.4912335276603699,
      "learning_rate": 9.047821454608401e-06,
      "loss": 0.4687,
      "step": 14102
    },
    {
      "epoch": 0.6471937956036896,
      "grad_norm": 0.41020840406417847,
      "learning_rate": 9.04767751678419e-06,
      "loss": 0.2855,
      "step": 14103
    },
    {
      "epoch": 0.6472396861089441,
      "grad_norm": 0.4550251364707947,
      "learning_rate": 9.047533569226593e-06,
      "loss": 0.3624,
      "step": 14104
    },
    {
      "epoch": 0.6472855766141985,
      "grad_norm": 0.42754465341567993,
      "learning_rate": 9.047389611935955e-06,
      "loss": 0.3495,
      "step": 14105
    },
    {
      "epoch": 0.647331467119453,
      "grad_norm": 0.4354294538497925,
      "learning_rate": 9.04724564491262e-06,
      "loss": 0.3514,
      "step": 14106
    },
    {
      "epoch": 0.6473773576247075,
      "grad_norm": 0.4663691520690918,
      "learning_rate": 9.047101668156939e-06,
      "loss": 0.4118,
      "step": 14107
    },
    {
      "epoch": 0.6474232481299619,
      "grad_norm": 0.4161934554576874,
      "learning_rate": 9.046957681669257e-06,
      "loss": 0.2873,
      "step": 14108
    },
    {
      "epoch": 0.6474691386352164,
      "grad_norm": 0.4455099403858185,
      "learning_rate": 9.046813685449915e-06,
      "loss": 0.3925,
      "step": 14109
    },
    {
      "epoch": 0.6475150291404709,
      "grad_norm": 0.44092604517936707,
      "learning_rate": 9.046669679499267e-06,
      "loss": 0.406,
      "step": 14110
    },
    {
      "epoch": 0.6475609196457253,
      "grad_norm": 0.432089626789093,
      "learning_rate": 9.046525663817653e-06,
      "loss": 0.4165,
      "step": 14111
    },
    {
      "epoch": 0.6476068101509798,
      "grad_norm": 0.4164719879627228,
      "learning_rate": 9.046381638405424e-06,
      "loss": 0.3049,
      "step": 14112
    },
    {
      "epoch": 0.6476527006562343,
      "grad_norm": 0.5399343371391296,
      "learning_rate": 9.046237603262924e-06,
      "loss": 0.5593,
      "step": 14113
    },
    {
      "epoch": 0.6476985911614886,
      "grad_norm": 0.542177677154541,
      "learning_rate": 9.0460935583905e-06,
      "loss": 0.4102,
      "step": 14114
    },
    {
      "epoch": 0.6477444816667431,
      "grad_norm": 0.4234831631183624,
      "learning_rate": 9.045949503788495e-06,
      "loss": 0.3363,
      "step": 14115
    },
    {
      "epoch": 0.6477903721719976,
      "grad_norm": 0.441610723733902,
      "learning_rate": 9.04580543945726e-06,
      "loss": 0.375,
      "step": 14116
    },
    {
      "epoch": 0.647836262677252,
      "grad_norm": 0.47733840346336365,
      "learning_rate": 9.04566136539714e-06,
      "loss": 0.4484,
      "step": 14117
    },
    {
      "epoch": 0.6478821531825065,
      "grad_norm": 0.44860878586769104,
      "learning_rate": 9.045517281608482e-06,
      "loss": 0.4018,
      "step": 14118
    },
    {
      "epoch": 0.647928043687761,
      "grad_norm": 0.46530213952064514,
      "learning_rate": 9.04537318809163e-06,
      "loss": 0.396,
      "step": 14119
    },
    {
      "epoch": 0.6479739341930154,
      "grad_norm": 0.5208595991134644,
      "learning_rate": 9.045229084846933e-06,
      "loss": 0.4631,
      "step": 14120
    },
    {
      "epoch": 0.6480198246982699,
      "grad_norm": 0.49520808458328247,
      "learning_rate": 9.045084971874738e-06,
      "loss": 0.4434,
      "step": 14121
    },
    {
      "epoch": 0.6480657152035244,
      "grad_norm": 0.4780859351158142,
      "learning_rate": 9.04494084917539e-06,
      "loss": 0.4364,
      "step": 14122
    },
    {
      "epoch": 0.6481116057087789,
      "grad_norm": 0.4597601294517517,
      "learning_rate": 9.044796716749235e-06,
      "loss": 0.3539,
      "step": 14123
    },
    {
      "epoch": 0.6481574962140333,
      "grad_norm": 0.4641176462173462,
      "learning_rate": 9.044652574596621e-06,
      "loss": 0.457,
      "step": 14124
    },
    {
      "epoch": 0.6482033867192878,
      "grad_norm": 0.48095694184303284,
      "learning_rate": 9.044508422717892e-06,
      "loss": 0.3798,
      "step": 14125
    },
    {
      "epoch": 0.6482492772245423,
      "grad_norm": 0.4546697735786438,
      "learning_rate": 9.044364261113399e-06,
      "loss": 0.3612,
      "step": 14126
    },
    {
      "epoch": 0.6482951677297967,
      "grad_norm": 0.4892033636569977,
      "learning_rate": 9.044220089783487e-06,
      "loss": 0.4157,
      "step": 14127
    },
    {
      "epoch": 0.6483410582350512,
      "grad_norm": 0.46192535758018494,
      "learning_rate": 9.0440759087285e-06,
      "loss": 0.3593,
      "step": 14128
    },
    {
      "epoch": 0.6483869487403057,
      "grad_norm": 0.44785141944885254,
      "learning_rate": 9.043931717948788e-06,
      "loss": 0.3933,
      "step": 14129
    },
    {
      "epoch": 0.64843283924556,
      "grad_norm": 0.4695050120353699,
      "learning_rate": 9.043787517444697e-06,
      "loss": 0.431,
      "step": 14130
    },
    {
      "epoch": 0.6484787297508146,
      "grad_norm": 0.4779984652996063,
      "learning_rate": 9.043643307216572e-06,
      "loss": 0.4186,
      "step": 14131
    },
    {
      "epoch": 0.648524620256069,
      "grad_norm": 0.4555525779724121,
      "learning_rate": 9.043499087264763e-06,
      "loss": 0.3876,
      "step": 14132
    },
    {
      "epoch": 0.6485705107613234,
      "grad_norm": 0.4493536651134491,
      "learning_rate": 9.043354857589612e-06,
      "loss": 0.4263,
      "step": 14133
    },
    {
      "epoch": 0.6486164012665779,
      "grad_norm": 0.44953417778015137,
      "learning_rate": 9.04321061819147e-06,
      "loss": 0.419,
      "step": 14134
    },
    {
      "epoch": 0.6486622917718324,
      "grad_norm": 0.4317088723182678,
      "learning_rate": 9.043066369070683e-06,
      "loss": 0.3206,
      "step": 14135
    },
    {
      "epoch": 0.6487081822770868,
      "grad_norm": 0.4712981581687927,
      "learning_rate": 9.042922110227597e-06,
      "loss": 0.4451,
      "step": 14136
    },
    {
      "epoch": 0.6487540727823413,
      "grad_norm": 0.45407041907310486,
      "learning_rate": 9.04277784166256e-06,
      "loss": 0.4773,
      "step": 14137
    },
    {
      "epoch": 0.6487999632875958,
      "grad_norm": 0.4809816777706146,
      "learning_rate": 9.042633563375917e-06,
      "loss": 0.4239,
      "step": 14138
    },
    {
      "epoch": 0.6488458537928503,
      "grad_norm": 0.49848106503486633,
      "learning_rate": 9.042489275368018e-06,
      "loss": 0.5347,
      "step": 14139
    },
    {
      "epoch": 0.6488917442981047,
      "grad_norm": 0.4500460922718048,
      "learning_rate": 9.042344977639206e-06,
      "loss": 0.3855,
      "step": 14140
    },
    {
      "epoch": 0.6489376348033592,
      "grad_norm": 0.47755295038223267,
      "learning_rate": 9.042200670189832e-06,
      "loss": 0.4204,
      "step": 14141
    },
    {
      "epoch": 0.6489835253086137,
      "grad_norm": 0.469647616147995,
      "learning_rate": 9.04205635302024e-06,
      "loss": 0.4155,
      "step": 14142
    },
    {
      "epoch": 0.6490294158138681,
      "grad_norm": 0.4649830162525177,
      "learning_rate": 9.041912026130778e-06,
      "loss": 0.4485,
      "step": 14143
    },
    {
      "epoch": 0.6490753063191226,
      "grad_norm": 0.4809988737106323,
      "learning_rate": 9.041767689521794e-06,
      "loss": 0.4988,
      "step": 14144
    },
    {
      "epoch": 0.6491211968243771,
      "grad_norm": 0.4577782154083252,
      "learning_rate": 9.041623343193633e-06,
      "loss": 0.3877,
      "step": 14145
    },
    {
      "epoch": 0.6491670873296315,
      "grad_norm": 0.4883314371109009,
      "learning_rate": 9.041478987146645e-06,
      "loss": 0.4486,
      "step": 14146
    },
    {
      "epoch": 0.649212977834886,
      "grad_norm": 0.4496147334575653,
      "learning_rate": 9.041334621381175e-06,
      "loss": 0.4687,
      "step": 14147
    },
    {
      "epoch": 0.6492588683401405,
      "grad_norm": 0.45364218950271606,
      "learning_rate": 9.041190245897571e-06,
      "loss": 0.4014,
      "step": 14148
    },
    {
      "epoch": 0.6493047588453948,
      "grad_norm": 0.4536188244819641,
      "learning_rate": 9.04104586069618e-06,
      "loss": 0.4487,
      "step": 14149
    },
    {
      "epoch": 0.6493506493506493,
      "grad_norm": 0.4537698030471802,
      "learning_rate": 9.040901465777347e-06,
      "loss": 0.393,
      "step": 14150
    },
    {
      "epoch": 0.6493965398559038,
      "grad_norm": 0.47897160053253174,
      "learning_rate": 9.040757061141424e-06,
      "loss": 0.4634,
      "step": 14151
    },
    {
      "epoch": 0.6494424303611582,
      "grad_norm": 0.46400317549705505,
      "learning_rate": 9.040612646788755e-06,
      "loss": 0.4032,
      "step": 14152
    },
    {
      "epoch": 0.6494883208664127,
      "grad_norm": 0.4907035231590271,
      "learning_rate": 9.040468222719687e-06,
      "loss": 0.4523,
      "step": 14153
    },
    {
      "epoch": 0.6495342113716672,
      "grad_norm": 0.4830639958381653,
      "learning_rate": 9.040323788934568e-06,
      "loss": 0.4138,
      "step": 14154
    },
    {
      "epoch": 0.6495801018769216,
      "grad_norm": 0.4401586055755615,
      "learning_rate": 9.040179345433745e-06,
      "loss": 0.361,
      "step": 14155
    },
    {
      "epoch": 0.6496259923821761,
      "grad_norm": 0.4571276605129242,
      "learning_rate": 9.040034892217569e-06,
      "loss": 0.3553,
      "step": 14156
    },
    {
      "epoch": 0.6496718828874306,
      "grad_norm": 0.494339257478714,
      "learning_rate": 9.039890429286381e-06,
      "loss": 0.4193,
      "step": 14157
    },
    {
      "epoch": 0.6497177733926851,
      "grad_norm": 0.4599703550338745,
      "learning_rate": 9.039745956640533e-06,
      "loss": 0.4102,
      "step": 14158
    },
    {
      "epoch": 0.6497636638979395,
      "grad_norm": 0.48626193404197693,
      "learning_rate": 9.039601474280371e-06,
      "loss": 0.4348,
      "step": 14159
    },
    {
      "epoch": 0.649809554403194,
      "grad_norm": 0.45206359028816223,
      "learning_rate": 9.039456982206241e-06,
      "loss": 0.4062,
      "step": 14160
    },
    {
      "epoch": 0.6498554449084485,
      "grad_norm": 0.44994422793388367,
      "learning_rate": 9.039312480418493e-06,
      "loss": 0.3898,
      "step": 14161
    },
    {
      "epoch": 0.6499013354137029,
      "grad_norm": 0.41774889826774597,
      "learning_rate": 9.039167968917473e-06,
      "loss": 0.3291,
      "step": 14162
    },
    {
      "epoch": 0.6499472259189574,
      "grad_norm": 0.4634527266025543,
      "learning_rate": 9.039023447703529e-06,
      "loss": 0.3544,
      "step": 14163
    },
    {
      "epoch": 0.6499931164242119,
      "grad_norm": 0.49590712785720825,
      "learning_rate": 9.038878916777007e-06,
      "loss": 0.3944,
      "step": 14164
    },
    {
      "epoch": 0.6500390069294663,
      "grad_norm": 0.4279087781906128,
      "learning_rate": 9.038734376138258e-06,
      "loss": 0.2985,
      "step": 14165
    },
    {
      "epoch": 0.6500848974347208,
      "grad_norm": 0.464444637298584,
      "learning_rate": 9.038589825787626e-06,
      "loss": 0.41,
      "step": 14166
    },
    {
      "epoch": 0.6501307879399753,
      "grad_norm": 0.4504774510860443,
      "learning_rate": 9.038445265725462e-06,
      "loss": 0.3904,
      "step": 14167
    },
    {
      "epoch": 0.6501766784452296,
      "grad_norm": 0.4301224946975708,
      "learning_rate": 9.03830069595211e-06,
      "loss": 0.332,
      "step": 14168
    },
    {
      "epoch": 0.6502225689504841,
      "grad_norm": 0.4455925226211548,
      "learning_rate": 9.03815611646792e-06,
      "loss": 0.3554,
      "step": 14169
    },
    {
      "epoch": 0.6502684594557386,
      "grad_norm": 0.4768845736980438,
      "learning_rate": 9.038011527273238e-06,
      "loss": 0.4724,
      "step": 14170
    },
    {
      "epoch": 0.650314349960993,
      "grad_norm": 0.4593009054660797,
      "learning_rate": 9.037866928368414e-06,
      "loss": 0.3981,
      "step": 14171
    },
    {
      "epoch": 0.6503602404662475,
      "grad_norm": 0.4494643211364746,
      "learning_rate": 9.037722319753794e-06,
      "loss": 0.4357,
      "step": 14172
    },
    {
      "epoch": 0.650406130971502,
      "grad_norm": 0.42832937836647034,
      "learning_rate": 9.037577701429725e-06,
      "loss": 0.3538,
      "step": 14173
    },
    {
      "epoch": 0.6504520214767565,
      "grad_norm": 0.430900514125824,
      "learning_rate": 9.037433073396558e-06,
      "loss": 0.3572,
      "step": 14174
    },
    {
      "epoch": 0.6504979119820109,
      "grad_norm": 0.4725225269794464,
      "learning_rate": 9.037288435654638e-06,
      "loss": 0.456,
      "step": 14175
    },
    {
      "epoch": 0.6505438024872654,
      "grad_norm": 0.4682311713695526,
      "learning_rate": 9.037143788204313e-06,
      "loss": 0.3594,
      "step": 14176
    },
    {
      "epoch": 0.6505896929925199,
      "grad_norm": 0.4947977662086487,
      "learning_rate": 9.036999131045932e-06,
      "loss": 0.506,
      "step": 14177
    },
    {
      "epoch": 0.6506355834977743,
      "grad_norm": 0.5047365427017212,
      "learning_rate": 9.036854464179841e-06,
      "loss": 0.4631,
      "step": 14178
    },
    {
      "epoch": 0.6506814740030288,
      "grad_norm": 0.47522905468940735,
      "learning_rate": 9.03670978760639e-06,
      "loss": 0.3945,
      "step": 14179
    },
    {
      "epoch": 0.6507273645082833,
      "grad_norm": 0.4783630669116974,
      "learning_rate": 9.036565101325925e-06,
      "loss": 0.4521,
      "step": 14180
    },
    {
      "epoch": 0.6507732550135377,
      "grad_norm": 0.45084160566329956,
      "learning_rate": 9.036420405338798e-06,
      "loss": 0.4154,
      "step": 14181
    },
    {
      "epoch": 0.6508191455187922,
      "grad_norm": 0.45110151171684265,
      "learning_rate": 9.036275699645349e-06,
      "loss": 0.3603,
      "step": 14182
    },
    {
      "epoch": 0.6508650360240467,
      "grad_norm": 0.4372854232788086,
      "learning_rate": 9.036130984245934e-06,
      "loss": 0.3491,
      "step": 14183
    },
    {
      "epoch": 0.650910926529301,
      "grad_norm": 0.4539983570575714,
      "learning_rate": 9.035986259140897e-06,
      "loss": 0.3706,
      "step": 14184
    },
    {
      "epoch": 0.6509568170345555,
      "grad_norm": 0.45694440603256226,
      "learning_rate": 9.035841524330589e-06,
      "loss": 0.3792,
      "step": 14185
    },
    {
      "epoch": 0.65100270753981,
      "grad_norm": 0.41765594482421875,
      "learning_rate": 9.035696779815353e-06,
      "loss": 0.3209,
      "step": 14186
    },
    {
      "epoch": 0.6510485980450644,
      "grad_norm": 0.41461434960365295,
      "learning_rate": 9.03555202559554e-06,
      "loss": 0.3301,
      "step": 14187
    },
    {
      "epoch": 0.6510944885503189,
      "grad_norm": 0.5036013126373291,
      "learning_rate": 9.035407261671499e-06,
      "loss": 0.465,
      "step": 14188
    },
    {
      "epoch": 0.6511403790555734,
      "grad_norm": 0.46942099928855896,
      "learning_rate": 9.035262488043578e-06,
      "loss": 0.4167,
      "step": 14189
    },
    {
      "epoch": 0.6511862695608278,
      "grad_norm": 0.44426435232162476,
      "learning_rate": 9.035117704712123e-06,
      "loss": 0.393,
      "step": 14190
    },
    {
      "epoch": 0.6512321600660823,
      "grad_norm": 0.44255638122558594,
      "learning_rate": 9.034972911677484e-06,
      "loss": 0.3944,
      "step": 14191
    },
    {
      "epoch": 0.6512780505713368,
      "grad_norm": 0.44644495844841003,
      "learning_rate": 9.034828108940008e-06,
      "loss": 0.3591,
      "step": 14192
    },
    {
      "epoch": 0.6513239410765913,
      "grad_norm": 0.4471990168094635,
      "learning_rate": 9.034683296500045e-06,
      "loss": 0.4238,
      "step": 14193
    },
    {
      "epoch": 0.6513698315818457,
      "grad_norm": 0.4770403504371643,
      "learning_rate": 9.034538474357941e-06,
      "loss": 0.4435,
      "step": 14194
    },
    {
      "epoch": 0.6514157220871002,
      "grad_norm": 0.49885910749435425,
      "learning_rate": 9.034393642514046e-06,
      "loss": 0.4334,
      "step": 14195
    },
    {
      "epoch": 0.6514616125923547,
      "grad_norm": 0.4737358093261719,
      "learning_rate": 9.034248800968707e-06,
      "loss": 0.396,
      "step": 14196
    },
    {
      "epoch": 0.6515075030976091,
      "grad_norm": 0.4722176194190979,
      "learning_rate": 9.034103949722272e-06,
      "loss": 0.4733,
      "step": 14197
    },
    {
      "epoch": 0.6515533936028636,
      "grad_norm": 0.4813441336154938,
      "learning_rate": 9.033959088775092e-06,
      "loss": 0.4093,
      "step": 14198
    },
    {
      "epoch": 0.6515992841081181,
      "grad_norm": 0.4499693512916565,
      "learning_rate": 9.033814218127513e-06,
      "loss": 0.3775,
      "step": 14199
    },
    {
      "epoch": 0.6516451746133725,
      "grad_norm": 0.45358219742774963,
      "learning_rate": 9.033669337779885e-06,
      "loss": 0.3782,
      "step": 14200
    },
    {
      "epoch": 0.651691065118627,
      "grad_norm": 0.45872077345848083,
      "learning_rate": 9.033524447732554e-06,
      "loss": 0.3659,
      "step": 14201
    },
    {
      "epoch": 0.6517369556238815,
      "grad_norm": 0.4243559241294861,
      "learning_rate": 9.03337954798587e-06,
      "loss": 0.3089,
      "step": 14202
    },
    {
      "epoch": 0.6517828461291358,
      "grad_norm": 0.4494980573654175,
      "learning_rate": 9.03323463854018e-06,
      "loss": 0.3517,
      "step": 14203
    },
    {
      "epoch": 0.6518287366343903,
      "grad_norm": 0.42859968543052673,
      "learning_rate": 9.033089719395833e-06,
      "loss": 0.3393,
      "step": 14204
    },
    {
      "epoch": 0.6518746271396448,
      "grad_norm": 0.4387156367301941,
      "learning_rate": 9.032944790553182e-06,
      "loss": 0.4012,
      "step": 14205
    },
    {
      "epoch": 0.6519205176448992,
      "grad_norm": 0.4646153151988983,
      "learning_rate": 9.032799852012567e-06,
      "loss": 0.3992,
      "step": 14206
    },
    {
      "epoch": 0.6519664081501537,
      "grad_norm": 0.4445086419582367,
      "learning_rate": 9.032654903774344e-06,
      "loss": 0.4378,
      "step": 14207
    },
    {
      "epoch": 0.6520122986554082,
      "grad_norm": 0.4946781396865845,
      "learning_rate": 9.03250994583886e-06,
      "loss": 0.4068,
      "step": 14208
    },
    {
      "epoch": 0.6520581891606626,
      "grad_norm": 0.440120130777359,
      "learning_rate": 9.03236497820646e-06,
      "loss": 0.3387,
      "step": 14209
    },
    {
      "epoch": 0.6521040796659171,
      "grad_norm": 0.49724826216697693,
      "learning_rate": 9.032220000877493e-06,
      "loss": 0.4319,
      "step": 14210
    },
    {
      "epoch": 0.6521499701711716,
      "grad_norm": 0.4096190929412842,
      "learning_rate": 9.032075013852314e-06,
      "loss": 0.3478,
      "step": 14211
    },
    {
      "epoch": 0.6521958606764261,
      "grad_norm": 0.4368962347507477,
      "learning_rate": 9.031930017131265e-06,
      "loss": 0.3788,
      "step": 14212
    },
    {
      "epoch": 0.6522417511816805,
      "grad_norm": 0.462880939245224,
      "learning_rate": 9.031785010714695e-06,
      "loss": 0.4816,
      "step": 14213
    },
    {
      "epoch": 0.652287641686935,
      "grad_norm": 0.42750978469848633,
      "learning_rate": 9.031639994602956e-06,
      "loss": 0.3207,
      "step": 14214
    },
    {
      "epoch": 0.6523335321921895,
      "grad_norm": 0.49696362018585205,
      "learning_rate": 9.031494968796397e-06,
      "loss": 0.4518,
      "step": 14215
    },
    {
      "epoch": 0.6523794226974439,
      "grad_norm": 0.40314921736717224,
      "learning_rate": 9.031349933295364e-06,
      "loss": 0.2979,
      "step": 14216
    },
    {
      "epoch": 0.6524253132026984,
      "grad_norm": 0.49822306632995605,
      "learning_rate": 9.031204888100205e-06,
      "loss": 0.3275,
      "step": 14217
    },
    {
      "epoch": 0.6524712037079529,
      "grad_norm": 0.4009248614311218,
      "learning_rate": 9.031059833211272e-06,
      "loss": 0.3344,
      "step": 14218
    },
    {
      "epoch": 0.6525170942132072,
      "grad_norm": 0.43659937381744385,
      "learning_rate": 9.030914768628913e-06,
      "loss": 0.3628,
      "step": 14219
    },
    {
      "epoch": 0.6525629847184617,
      "grad_norm": 0.4257620573043823,
      "learning_rate": 9.030769694353473e-06,
      "loss": 0.3328,
      "step": 14220
    },
    {
      "epoch": 0.6526088752237162,
      "grad_norm": 0.4416256844997406,
      "learning_rate": 9.030624610385308e-06,
      "loss": 0.3649,
      "step": 14221
    },
    {
      "epoch": 0.6526547657289706,
      "grad_norm": 0.4330078661441803,
      "learning_rate": 9.030479516724762e-06,
      "loss": 0.3821,
      "step": 14222
    },
    {
      "epoch": 0.6527006562342251,
      "grad_norm": 0.4581303298473358,
      "learning_rate": 9.030334413372184e-06,
      "loss": 0.3833,
      "step": 14223
    },
    {
      "epoch": 0.6527465467394796,
      "grad_norm": 0.4504668116569519,
      "learning_rate": 9.030189300327922e-06,
      "loss": 0.4051,
      "step": 14224
    },
    {
      "epoch": 0.652792437244734,
      "grad_norm": 0.42233186960220337,
      "learning_rate": 9.03004417759233e-06,
      "loss": 0.339,
      "step": 14225
    },
    {
      "epoch": 0.6528383277499885,
      "grad_norm": 0.4453276991844177,
      "learning_rate": 9.02989904516575e-06,
      "loss": 0.3743,
      "step": 14226
    },
    {
      "epoch": 0.652884218255243,
      "grad_norm": 0.4239007830619812,
      "learning_rate": 9.029753903048538e-06,
      "loss": 0.2985,
      "step": 14227
    },
    {
      "epoch": 0.6529301087604975,
      "grad_norm": 0.4869670569896698,
      "learning_rate": 9.029608751241038e-06,
      "loss": 0.3967,
      "step": 14228
    },
    {
      "epoch": 0.6529759992657519,
      "grad_norm": 0.46322375535964966,
      "learning_rate": 9.0294635897436e-06,
      "loss": 0.3948,
      "step": 14229
    },
    {
      "epoch": 0.6530218897710064,
      "grad_norm": 0.4836317300796509,
      "learning_rate": 9.029318418556577e-06,
      "loss": 0.376,
      "step": 14230
    },
    {
      "epoch": 0.6530677802762609,
      "grad_norm": 0.4076985716819763,
      "learning_rate": 9.029173237680312e-06,
      "loss": 0.3342,
      "step": 14231
    },
    {
      "epoch": 0.6531136707815153,
      "grad_norm": 0.4360843002796173,
      "learning_rate": 9.029028047115158e-06,
      "loss": 0.3499,
      "step": 14232
    },
    {
      "epoch": 0.6531595612867698,
      "grad_norm": 0.5220640301704407,
      "learning_rate": 9.028882846861464e-06,
      "loss": 0.4185,
      "step": 14233
    },
    {
      "epoch": 0.6532054517920243,
      "grad_norm": 0.4931797683238983,
      "learning_rate": 9.028737636919577e-06,
      "loss": 0.4511,
      "step": 14234
    },
    {
      "epoch": 0.6532513422972787,
      "grad_norm": 0.4669604003429413,
      "learning_rate": 9.028592417289848e-06,
      "loss": 0.3986,
      "step": 14235
    },
    {
      "epoch": 0.6532972328025332,
      "grad_norm": 0.4486613869667053,
      "learning_rate": 9.028447187972627e-06,
      "loss": 0.3956,
      "step": 14236
    },
    {
      "epoch": 0.6533431233077877,
      "grad_norm": 0.4503069519996643,
      "learning_rate": 9.028301948968259e-06,
      "loss": 0.3652,
      "step": 14237
    },
    {
      "epoch": 0.653389013813042,
      "grad_norm": 0.43136584758758545,
      "learning_rate": 9.028156700277097e-06,
      "loss": 0.3808,
      "step": 14238
    },
    {
      "epoch": 0.6534349043182965,
      "grad_norm": 0.4658714234828949,
      "learning_rate": 9.028011441899491e-06,
      "loss": 0.4422,
      "step": 14239
    },
    {
      "epoch": 0.653480794823551,
      "grad_norm": 0.45481759309768677,
      "learning_rate": 9.027866173835788e-06,
      "loss": 0.4494,
      "step": 14240
    },
    {
      "epoch": 0.6535266853288054,
      "grad_norm": 0.4836934506893158,
      "learning_rate": 9.027720896086338e-06,
      "loss": 0.3856,
      "step": 14241
    },
    {
      "epoch": 0.6535725758340599,
      "grad_norm": 0.430143803358078,
      "learning_rate": 9.02757560865149e-06,
      "loss": 0.3788,
      "step": 14242
    },
    {
      "epoch": 0.6536184663393144,
      "grad_norm": 0.47186678647994995,
      "learning_rate": 9.027430311531594e-06,
      "loss": 0.4509,
      "step": 14243
    },
    {
      "epoch": 0.6536643568445688,
      "grad_norm": 0.4348425567150116,
      "learning_rate": 9.027285004727e-06,
      "loss": 0.3482,
      "step": 14244
    },
    {
      "epoch": 0.6537102473498233,
      "grad_norm": 0.4389311671257019,
      "learning_rate": 9.027139688238056e-06,
      "loss": 0.3724,
      "step": 14245
    },
    {
      "epoch": 0.6537561378550778,
      "grad_norm": 0.43273451924324036,
      "learning_rate": 9.026994362065113e-06,
      "loss": 0.3212,
      "step": 14246
    },
    {
      "epoch": 0.6538020283603323,
      "grad_norm": 0.41638949513435364,
      "learning_rate": 9.026849026208518e-06,
      "loss": 0.346,
      "step": 14247
    },
    {
      "epoch": 0.6538479188655867,
      "grad_norm": 0.4424901306629181,
      "learning_rate": 9.026703680668623e-06,
      "loss": 0.3425,
      "step": 14248
    },
    {
      "epoch": 0.6538938093708412,
      "grad_norm": 0.4880313277244568,
      "learning_rate": 9.026558325445775e-06,
      "loss": 0.3837,
      "step": 14249
    },
    {
      "epoch": 0.6539396998760957,
      "grad_norm": 0.4541361927986145,
      "learning_rate": 9.026412960540327e-06,
      "loss": 0.4208,
      "step": 14250
    },
    {
      "epoch": 0.6539855903813501,
      "grad_norm": 0.45776498317718506,
      "learning_rate": 9.026267585952626e-06,
      "loss": 0.3715,
      "step": 14251
    },
    {
      "epoch": 0.6540314808866046,
      "grad_norm": 0.43914008140563965,
      "learning_rate": 9.02612220168302e-06,
      "loss": 0.3789,
      "step": 14252
    },
    {
      "epoch": 0.6540773713918591,
      "grad_norm": 0.4991929531097412,
      "learning_rate": 9.025976807731864e-06,
      "loss": 0.4508,
      "step": 14253
    },
    {
      "epoch": 0.6541232618971135,
      "grad_norm": 0.49503421783447266,
      "learning_rate": 9.025831404099503e-06,
      "loss": 0.408,
      "step": 14254
    },
    {
      "epoch": 0.654169152402368,
      "grad_norm": 0.4243558347225189,
      "learning_rate": 9.025685990786288e-06,
      "loss": 0.3195,
      "step": 14255
    },
    {
      "epoch": 0.6542150429076224,
      "grad_norm": 0.49006035923957825,
      "learning_rate": 9.02554056779257e-06,
      "loss": 0.4205,
      "step": 14256
    },
    {
      "epoch": 0.6542609334128768,
      "grad_norm": 0.4685148000717163,
      "learning_rate": 9.025395135118695e-06,
      "loss": 0.4653,
      "step": 14257
    },
    {
      "epoch": 0.6543068239181313,
      "grad_norm": 0.4466828405857086,
      "learning_rate": 9.025249692765017e-06,
      "loss": 0.353,
      "step": 14258
    },
    {
      "epoch": 0.6543527144233858,
      "grad_norm": 0.5397123098373413,
      "learning_rate": 9.025104240731883e-06,
      "loss": 0.5319,
      "step": 14259
    },
    {
      "epoch": 0.6543986049286402,
      "grad_norm": 0.45067355036735535,
      "learning_rate": 9.024958779019645e-06,
      "loss": 0.3727,
      "step": 14260
    },
    {
      "epoch": 0.6544444954338947,
      "grad_norm": 0.4255097508430481,
      "learning_rate": 9.02481330762865e-06,
      "loss": 0.3411,
      "step": 14261
    },
    {
      "epoch": 0.6544903859391492,
      "grad_norm": 0.4505007565021515,
      "learning_rate": 9.02466782655925e-06,
      "loss": 0.4574,
      "step": 14262
    },
    {
      "epoch": 0.6545362764444036,
      "grad_norm": 0.4589614272117615,
      "learning_rate": 9.024522335811794e-06,
      "loss": 0.4258,
      "step": 14263
    },
    {
      "epoch": 0.6545821669496581,
      "grad_norm": 0.4743839502334595,
      "learning_rate": 9.024376835386631e-06,
      "loss": 0.4652,
      "step": 14264
    },
    {
      "epoch": 0.6546280574549126,
      "grad_norm": 0.45333167910575867,
      "learning_rate": 9.024231325284113e-06,
      "loss": 0.3775,
      "step": 14265
    },
    {
      "epoch": 0.6546739479601671,
      "grad_norm": 0.4392283260822296,
      "learning_rate": 9.024085805504588e-06,
      "loss": 0.3831,
      "step": 14266
    },
    {
      "epoch": 0.6547198384654215,
      "grad_norm": 0.44775786995887756,
      "learning_rate": 9.023940276048408e-06,
      "loss": 0.3266,
      "step": 14267
    },
    {
      "epoch": 0.654765728970676,
      "grad_norm": 0.4305972456932068,
      "learning_rate": 9.02379473691592e-06,
      "loss": 0.3965,
      "step": 14268
    },
    {
      "epoch": 0.6548116194759305,
      "grad_norm": 0.436907559633255,
      "learning_rate": 9.023649188107478e-06,
      "loss": 0.3977,
      "step": 14269
    },
    {
      "epoch": 0.6548575099811849,
      "grad_norm": 0.4303225874900818,
      "learning_rate": 9.023503629623426e-06,
      "loss": 0.3797,
      "step": 14270
    },
    {
      "epoch": 0.6549034004864394,
      "grad_norm": 0.42467400431632996,
      "learning_rate": 9.02335806146412e-06,
      "loss": 0.3632,
      "step": 14271
    },
    {
      "epoch": 0.6549492909916939,
      "grad_norm": 0.6963770389556885,
      "learning_rate": 9.023212483629907e-06,
      "loss": 0.402,
      "step": 14272
    },
    {
      "epoch": 0.6549951814969482,
      "grad_norm": 0.45656120777130127,
      "learning_rate": 9.02306689612114e-06,
      "loss": 0.3755,
      "step": 14273
    },
    {
      "epoch": 0.6550410720022027,
      "grad_norm": 0.44843944907188416,
      "learning_rate": 9.022921298938163e-06,
      "loss": 0.4065,
      "step": 14274
    },
    {
      "epoch": 0.6550869625074572,
      "grad_norm": 0.45828837156295776,
      "learning_rate": 9.022775692081332e-06,
      "loss": 0.38,
      "step": 14275
    },
    {
      "epoch": 0.6551328530127116,
      "grad_norm": 0.45324090123176575,
      "learning_rate": 9.022630075550995e-06,
      "loss": 0.4004,
      "step": 14276
    },
    {
      "epoch": 0.6551787435179661,
      "grad_norm": 0.45647162199020386,
      "learning_rate": 9.022484449347502e-06,
      "loss": 0.4304,
      "step": 14277
    },
    {
      "epoch": 0.6552246340232206,
      "grad_norm": 0.45964911580085754,
      "learning_rate": 9.022338813471204e-06,
      "loss": 0.3928,
      "step": 14278
    },
    {
      "epoch": 0.655270524528475,
      "grad_norm": 0.46865180134773254,
      "learning_rate": 9.02219316792245e-06,
      "loss": 0.4365,
      "step": 14279
    },
    {
      "epoch": 0.6553164150337295,
      "grad_norm": 0.459452360868454,
      "learning_rate": 9.02204751270159e-06,
      "loss": 0.3792,
      "step": 14280
    },
    {
      "epoch": 0.655362305538984,
      "grad_norm": 0.5320595502853394,
      "learning_rate": 9.021901847808975e-06,
      "loss": 0.4703,
      "step": 14281
    },
    {
      "epoch": 0.6554081960442385,
      "grad_norm": 0.4537582993507385,
      "learning_rate": 9.021756173244958e-06,
      "loss": 0.3559,
      "step": 14282
    },
    {
      "epoch": 0.6554540865494929,
      "grad_norm": 0.45568475127220154,
      "learning_rate": 9.021610489009885e-06,
      "loss": 0.3764,
      "step": 14283
    },
    {
      "epoch": 0.6554999770547474,
      "grad_norm": 0.42102038860321045,
      "learning_rate": 9.021464795104108e-06,
      "loss": 0.3606,
      "step": 14284
    },
    {
      "epoch": 0.6555458675600019,
      "grad_norm": 0.45229196548461914,
      "learning_rate": 9.021319091527978e-06,
      "loss": 0.3937,
      "step": 14285
    },
    {
      "epoch": 0.6555917580652563,
      "grad_norm": 0.45309579372406006,
      "learning_rate": 9.021173378281845e-06,
      "loss": 0.3704,
      "step": 14286
    },
    {
      "epoch": 0.6556376485705108,
      "grad_norm": 0.47985339164733887,
      "learning_rate": 9.021027655366059e-06,
      "loss": 0.4844,
      "step": 14287
    },
    {
      "epoch": 0.6556835390757653,
      "grad_norm": 0.510222852230072,
      "learning_rate": 9.02088192278097e-06,
      "loss": 0.5047,
      "step": 14288
    },
    {
      "epoch": 0.6557294295810197,
      "grad_norm": 0.4962892532348633,
      "learning_rate": 9.020736180526928e-06,
      "loss": 0.4492,
      "step": 14289
    },
    {
      "epoch": 0.6557753200862741,
      "grad_norm": 0.4696367084980011,
      "learning_rate": 9.020590428604286e-06,
      "loss": 0.4638,
      "step": 14290
    },
    {
      "epoch": 0.6558212105915286,
      "grad_norm": 0.45266345143318176,
      "learning_rate": 9.020444667013392e-06,
      "loss": 0.3963,
      "step": 14291
    },
    {
      "epoch": 0.655867101096783,
      "grad_norm": 0.4869053363800049,
      "learning_rate": 9.0202988957546e-06,
      "loss": 0.4932,
      "step": 14292
    },
    {
      "epoch": 0.6559129916020375,
      "grad_norm": 0.44247767329216003,
      "learning_rate": 9.020153114828254e-06,
      "loss": 0.3328,
      "step": 14293
    },
    {
      "epoch": 0.655958882107292,
      "grad_norm": 0.4148915708065033,
      "learning_rate": 9.020007324234713e-06,
      "loss": 0.3048,
      "step": 14294
    },
    {
      "epoch": 0.6560047726125464,
      "grad_norm": 0.4607759416103363,
      "learning_rate": 9.01986152397432e-06,
      "loss": 0.4011,
      "step": 14295
    },
    {
      "epoch": 0.6560506631178009,
      "grad_norm": 0.4817531704902649,
      "learning_rate": 9.019715714047428e-06,
      "loss": 0.4129,
      "step": 14296
    },
    {
      "epoch": 0.6560965536230554,
      "grad_norm": 0.4889727234840393,
      "learning_rate": 9.01956989445439e-06,
      "loss": 0.4222,
      "step": 14297
    },
    {
      "epoch": 0.6561424441283098,
      "grad_norm": 0.4200547933578491,
      "learning_rate": 9.019424065195556e-06,
      "loss": 0.3608,
      "step": 14298
    },
    {
      "epoch": 0.6561883346335643,
      "grad_norm": 0.44878315925598145,
      "learning_rate": 9.019278226271272e-06,
      "loss": 0.3678,
      "step": 14299
    },
    {
      "epoch": 0.6562342251388188,
      "grad_norm": 0.48445549607276917,
      "learning_rate": 9.019132377681897e-06,
      "loss": 0.4833,
      "step": 14300
    },
    {
      "epoch": 0.6562801156440733,
      "grad_norm": 0.4433364272117615,
      "learning_rate": 9.018986519427776e-06,
      "loss": 0.3626,
      "step": 14301
    },
    {
      "epoch": 0.6563260061493277,
      "grad_norm": 0.4375712275505066,
      "learning_rate": 9.018840651509258e-06,
      "loss": 0.3851,
      "step": 14302
    },
    {
      "epoch": 0.6563718966545822,
      "grad_norm": 0.4861016273498535,
      "learning_rate": 9.0186947739267e-06,
      "loss": 0.416,
      "step": 14303
    },
    {
      "epoch": 0.6564177871598367,
      "grad_norm": 0.42424270510673523,
      "learning_rate": 9.018548886680447e-06,
      "loss": 0.3436,
      "step": 14304
    },
    {
      "epoch": 0.6564636776650911,
      "grad_norm": 0.4334239363670349,
      "learning_rate": 9.018402989770854e-06,
      "loss": 0.3579,
      "step": 14305
    },
    {
      "epoch": 0.6565095681703456,
      "grad_norm": 0.43622320890426636,
      "learning_rate": 9.018257083198267e-06,
      "loss": 0.3446,
      "step": 14306
    },
    {
      "epoch": 0.6565554586756001,
      "grad_norm": 0.4864494204521179,
      "learning_rate": 9.018111166963044e-06,
      "loss": 0.4219,
      "step": 14307
    },
    {
      "epoch": 0.6566013491808544,
      "grad_norm": 0.43667083978652954,
      "learning_rate": 9.01796524106553e-06,
      "loss": 0.3525,
      "step": 14308
    },
    {
      "epoch": 0.6566472396861089,
      "grad_norm": 0.45527803897857666,
      "learning_rate": 9.017819305506077e-06,
      "loss": 0.4047,
      "step": 14309
    },
    {
      "epoch": 0.6566931301913634,
      "grad_norm": 0.47500160336494446,
      "learning_rate": 9.017673360285037e-06,
      "loss": 0.4044,
      "step": 14310
    },
    {
      "epoch": 0.6567390206966178,
      "grad_norm": 0.45390433073043823,
      "learning_rate": 9.01752740540276e-06,
      "loss": 0.4165,
      "step": 14311
    },
    {
      "epoch": 0.6567849112018723,
      "grad_norm": 0.46602702140808105,
      "learning_rate": 9.017381440859598e-06,
      "loss": 0.4384,
      "step": 14312
    },
    {
      "epoch": 0.6568308017071268,
      "grad_norm": 0.4664005637168884,
      "learning_rate": 9.017235466655902e-06,
      "loss": 0.4154,
      "step": 14313
    },
    {
      "epoch": 0.6568766922123812,
      "grad_norm": 0.47409966588020325,
      "learning_rate": 9.017089482792023e-06,
      "loss": 0.4471,
      "step": 14314
    },
    {
      "epoch": 0.6569225827176357,
      "grad_norm": 0.45713138580322266,
      "learning_rate": 9.016943489268312e-06,
      "loss": 0.4411,
      "step": 14315
    },
    {
      "epoch": 0.6569684732228902,
      "grad_norm": 0.4602508544921875,
      "learning_rate": 9.016797486085119e-06,
      "loss": 0.4133,
      "step": 14316
    },
    {
      "epoch": 0.6570143637281447,
      "grad_norm": 0.46990537643432617,
      "learning_rate": 9.016651473242795e-06,
      "loss": 0.4098,
      "step": 14317
    },
    {
      "epoch": 0.6570602542333991,
      "grad_norm": 0.4914743900299072,
      "learning_rate": 9.016505450741693e-06,
      "loss": 0.449,
      "step": 14318
    },
    {
      "epoch": 0.6571061447386536,
      "grad_norm": 0.44506338238716125,
      "learning_rate": 9.016359418582161e-06,
      "loss": 0.378,
      "step": 14319
    },
    {
      "epoch": 0.6571520352439081,
      "grad_norm": 0.4237197935581207,
      "learning_rate": 9.016213376764554e-06,
      "loss": 0.3439,
      "step": 14320
    },
    {
      "epoch": 0.6571979257491625,
      "grad_norm": 0.4724003076553345,
      "learning_rate": 9.016067325289222e-06,
      "loss": 0.4165,
      "step": 14321
    },
    {
      "epoch": 0.657243816254417,
      "grad_norm": 0.47174176573753357,
      "learning_rate": 9.015921264156513e-06,
      "loss": 0.4175,
      "step": 14322
    },
    {
      "epoch": 0.6572897067596715,
      "grad_norm": 0.45713236927986145,
      "learning_rate": 9.015775193366784e-06,
      "loss": 0.3784,
      "step": 14323
    },
    {
      "epoch": 0.6573355972649259,
      "grad_norm": 0.43167969584465027,
      "learning_rate": 9.015629112920381e-06,
      "loss": 0.349,
      "step": 14324
    },
    {
      "epoch": 0.6573814877701804,
      "grad_norm": 0.4784649908542633,
      "learning_rate": 9.015483022817659e-06,
      "loss": 0.3854,
      "step": 14325
    },
    {
      "epoch": 0.6574273782754348,
      "grad_norm": 0.4564891755580902,
      "learning_rate": 9.015336923058965e-06,
      "loss": 0.406,
      "step": 14326
    },
    {
      "epoch": 0.6574732687806892,
      "grad_norm": 0.4788930118083954,
      "learning_rate": 9.015190813644654e-06,
      "loss": 0.4721,
      "step": 14327
    },
    {
      "epoch": 0.6575191592859437,
      "grad_norm": 0.4814184308052063,
      "learning_rate": 9.015044694575078e-06,
      "loss": 0.3603,
      "step": 14328
    },
    {
      "epoch": 0.6575650497911982,
      "grad_norm": 0.45181649923324585,
      "learning_rate": 9.014898565850584e-06,
      "loss": 0.3708,
      "step": 14329
    },
    {
      "epoch": 0.6576109402964526,
      "grad_norm": 0.497953861951828,
      "learning_rate": 9.014752427471529e-06,
      "loss": 0.4345,
      "step": 14330
    },
    {
      "epoch": 0.6576568308017071,
      "grad_norm": 0.47690749168395996,
      "learning_rate": 9.014606279438261e-06,
      "loss": 0.4239,
      "step": 14331
    },
    {
      "epoch": 0.6577027213069616,
      "grad_norm": 0.4662839472293854,
      "learning_rate": 9.01446012175113e-06,
      "loss": 0.3978,
      "step": 14332
    },
    {
      "epoch": 0.657748611812216,
      "grad_norm": 0.42566877603530884,
      "learning_rate": 9.014313954410492e-06,
      "loss": 0.3447,
      "step": 14333
    },
    {
      "epoch": 0.6577945023174705,
      "grad_norm": 0.44314879179000854,
      "learning_rate": 9.014167777416694e-06,
      "loss": 0.3501,
      "step": 14334
    },
    {
      "epoch": 0.657840392822725,
      "grad_norm": 0.4898779094219208,
      "learning_rate": 9.01402159077009e-06,
      "loss": 0.4714,
      "step": 14335
    },
    {
      "epoch": 0.6578862833279795,
      "grad_norm": 0.45057713985443115,
      "learning_rate": 9.01387539447103e-06,
      "loss": 0.3914,
      "step": 14336
    },
    {
      "epoch": 0.6579321738332339,
      "grad_norm": 0.41558974981307983,
      "learning_rate": 9.013729188519867e-06,
      "loss": 0.3315,
      "step": 14337
    },
    {
      "epoch": 0.6579780643384884,
      "grad_norm": 0.4637351632118225,
      "learning_rate": 9.013582972916951e-06,
      "loss": 0.4349,
      "step": 14338
    },
    {
      "epoch": 0.6580239548437429,
      "grad_norm": 0.4306354522705078,
      "learning_rate": 9.013436747662638e-06,
      "loss": 0.3611,
      "step": 14339
    },
    {
      "epoch": 0.6580698453489973,
      "grad_norm": 0.4567810297012329,
      "learning_rate": 9.013290512757273e-06,
      "loss": 0.4682,
      "step": 14340
    },
    {
      "epoch": 0.6581157358542518,
      "grad_norm": 0.45350244641304016,
      "learning_rate": 9.013144268201213e-06,
      "loss": 0.3851,
      "step": 14341
    },
    {
      "epoch": 0.6581616263595063,
      "grad_norm": 0.47463229298591614,
      "learning_rate": 9.012998013994807e-06,
      "loss": 0.4284,
      "step": 14342
    },
    {
      "epoch": 0.6582075168647606,
      "grad_norm": 0.44029948115348816,
      "learning_rate": 9.012851750138408e-06,
      "loss": 0.3701,
      "step": 14343
    },
    {
      "epoch": 0.6582534073700151,
      "grad_norm": 0.4572148323059082,
      "learning_rate": 9.012705476632364e-06,
      "loss": 0.3879,
      "step": 14344
    },
    {
      "epoch": 0.6582992978752696,
      "grad_norm": 0.43075481057167053,
      "learning_rate": 9.012559193477033e-06,
      "loss": 0.3397,
      "step": 14345
    },
    {
      "epoch": 0.658345188380524,
      "grad_norm": 0.47975191473960876,
      "learning_rate": 9.012412900672762e-06,
      "loss": 0.4115,
      "step": 14346
    },
    {
      "epoch": 0.6583910788857785,
      "grad_norm": 0.4551202356815338,
      "learning_rate": 9.012266598219906e-06,
      "loss": 0.3805,
      "step": 14347
    },
    {
      "epoch": 0.658436969391033,
      "grad_norm": 0.48552870750427246,
      "learning_rate": 9.012120286118813e-06,
      "loss": 0.4382,
      "step": 14348
    },
    {
      "epoch": 0.6584828598962874,
      "grad_norm": 0.4749126136302948,
      "learning_rate": 9.011973964369839e-06,
      "loss": 0.4695,
      "step": 14349
    },
    {
      "epoch": 0.6585287504015419,
      "grad_norm": 0.4661877453327179,
      "learning_rate": 9.011827632973332e-06,
      "loss": 0.3642,
      "step": 14350
    },
    {
      "epoch": 0.6585746409067964,
      "grad_norm": 0.4630640745162964,
      "learning_rate": 9.011681291929646e-06,
      "loss": 0.4108,
      "step": 14351
    },
    {
      "epoch": 0.6586205314120508,
      "grad_norm": 0.4235508441925049,
      "learning_rate": 9.011534941239135e-06,
      "loss": 0.3507,
      "step": 14352
    },
    {
      "epoch": 0.6586664219173053,
      "grad_norm": 0.43487074971199036,
      "learning_rate": 9.011388580902145e-06,
      "loss": 0.3318,
      "step": 14353
    },
    {
      "epoch": 0.6587123124225598,
      "grad_norm": 0.4923548400402069,
      "learning_rate": 9.011242210919033e-06,
      "loss": 0.5123,
      "step": 14354
    },
    {
      "epoch": 0.6587582029278143,
      "grad_norm": 0.4181765615940094,
      "learning_rate": 9.011095831290151e-06,
      "loss": 0.3352,
      "step": 14355
    },
    {
      "epoch": 0.6588040934330687,
      "grad_norm": 0.4527325928211212,
      "learning_rate": 9.010949442015849e-06,
      "loss": 0.4278,
      "step": 14356
    },
    {
      "epoch": 0.6588499839383232,
      "grad_norm": 0.47471654415130615,
      "learning_rate": 9.010803043096477e-06,
      "loss": 0.411,
      "step": 14357
    },
    {
      "epoch": 0.6588958744435777,
      "grad_norm": 0.47008365392684937,
      "learning_rate": 9.01065663453239e-06,
      "loss": 0.4118,
      "step": 14358
    },
    {
      "epoch": 0.658941764948832,
      "grad_norm": 0.4287841320037842,
      "learning_rate": 9.010510216323943e-06,
      "loss": 0.3199,
      "step": 14359
    },
    {
      "epoch": 0.6589876554540866,
      "grad_norm": 0.47300583124160767,
      "learning_rate": 9.010363788471481e-06,
      "loss": 0.4267,
      "step": 14360
    },
    {
      "epoch": 0.659033545959341,
      "grad_norm": 0.45789220929145813,
      "learning_rate": 9.010217350975361e-06,
      "loss": 0.373,
      "step": 14361
    },
    {
      "epoch": 0.6590794364645954,
      "grad_norm": 0.450854629278183,
      "learning_rate": 9.010070903835934e-06,
      "loss": 0.4102,
      "step": 14362
    },
    {
      "epoch": 0.6591253269698499,
      "grad_norm": 0.43615835905075073,
      "learning_rate": 9.009924447053553e-06,
      "loss": 0.3417,
      "step": 14363
    },
    {
      "epoch": 0.6591712174751044,
      "grad_norm": 0.4559480845928192,
      "learning_rate": 9.009777980628568e-06,
      "loss": 0.4037,
      "step": 14364
    },
    {
      "epoch": 0.6592171079803588,
      "grad_norm": 0.45068272948265076,
      "learning_rate": 9.009631504561334e-06,
      "loss": 0.3763,
      "step": 14365
    },
    {
      "epoch": 0.6592629984856133,
      "grad_norm": 0.4611908793449402,
      "learning_rate": 9.0094850188522e-06,
      "loss": 0.4163,
      "step": 14366
    },
    {
      "epoch": 0.6593088889908678,
      "grad_norm": 0.45859256386756897,
      "learning_rate": 9.009338523501521e-06,
      "loss": 0.3631,
      "step": 14367
    },
    {
      "epoch": 0.6593547794961222,
      "grad_norm": 0.47314950823783875,
      "learning_rate": 9.009192018509647e-06,
      "loss": 0.4006,
      "step": 14368
    },
    {
      "epoch": 0.6594006700013767,
      "grad_norm": 0.48570716381073,
      "learning_rate": 9.009045503876932e-06,
      "loss": 0.4335,
      "step": 14369
    },
    {
      "epoch": 0.6594465605066312,
      "grad_norm": 0.49148470163345337,
      "learning_rate": 9.008898979603728e-06,
      "loss": 0.4471,
      "step": 14370
    },
    {
      "epoch": 0.6594924510118857,
      "grad_norm": 0.46235740184783936,
      "learning_rate": 9.008752445690388e-06,
      "loss": 0.3753,
      "step": 14371
    },
    {
      "epoch": 0.6595383415171401,
      "grad_norm": 0.49642324447631836,
      "learning_rate": 9.008605902137263e-06,
      "loss": 0.511,
      "step": 14372
    },
    {
      "epoch": 0.6595842320223946,
      "grad_norm": 0.4772595167160034,
      "learning_rate": 9.008459348944706e-06,
      "loss": 0.4478,
      "step": 14373
    },
    {
      "epoch": 0.6596301225276491,
      "grad_norm": 0.447195440530777,
      "learning_rate": 9.00831278611307e-06,
      "loss": 0.3771,
      "step": 14374
    },
    {
      "epoch": 0.6596760130329035,
      "grad_norm": 0.48210033774375916,
      "learning_rate": 9.008166213642705e-06,
      "loss": 0.3939,
      "step": 14375
    },
    {
      "epoch": 0.659721903538158,
      "grad_norm": 0.49801814556121826,
      "learning_rate": 9.008019631533967e-06,
      "loss": 0.4349,
      "step": 14376
    },
    {
      "epoch": 0.6597677940434125,
      "grad_norm": 0.513995885848999,
      "learning_rate": 9.007873039787206e-06,
      "loss": 0.4538,
      "step": 14377
    },
    {
      "epoch": 0.6598136845486668,
      "grad_norm": 0.46197929978370667,
      "learning_rate": 9.007726438402776e-06,
      "loss": 0.3815,
      "step": 14378
    },
    {
      "epoch": 0.6598595750539213,
      "grad_norm": 0.47664257884025574,
      "learning_rate": 9.007579827381028e-06,
      "loss": 0.4159,
      "step": 14379
    },
    {
      "epoch": 0.6599054655591758,
      "grad_norm": 0.4619807004928589,
      "learning_rate": 9.007433206722316e-06,
      "loss": 0.3946,
      "step": 14380
    },
    {
      "epoch": 0.6599513560644302,
      "grad_norm": 0.4921075403690338,
      "learning_rate": 9.00728657642699e-06,
      "loss": 0.473,
      "step": 14381
    },
    {
      "epoch": 0.6599972465696847,
      "grad_norm": 0.4531581997871399,
      "learning_rate": 9.007139936495406e-06,
      "loss": 0.3653,
      "step": 14382
    },
    {
      "epoch": 0.6600431370749392,
      "grad_norm": 0.5287548899650574,
      "learning_rate": 9.006993286927916e-06,
      "loss": 0.4458,
      "step": 14383
    },
    {
      "epoch": 0.6600890275801936,
      "grad_norm": 0.4318279027938843,
      "learning_rate": 9.00684662772487e-06,
      "loss": 0.3566,
      "step": 14384
    },
    {
      "epoch": 0.6601349180854481,
      "grad_norm": 0.5398879647254944,
      "learning_rate": 9.006699958886625e-06,
      "loss": 0.3729,
      "step": 14385
    },
    {
      "epoch": 0.6601808085907026,
      "grad_norm": 0.4714904725551605,
      "learning_rate": 9.006553280413528e-06,
      "loss": 0.488,
      "step": 14386
    },
    {
      "epoch": 0.660226699095957,
      "grad_norm": 0.4445558190345764,
      "learning_rate": 9.006406592305936e-06,
      "loss": 0.3829,
      "step": 14387
    },
    {
      "epoch": 0.6602725896012115,
      "grad_norm": 0.5022744536399841,
      "learning_rate": 9.006259894564203e-06,
      "loss": 0.4582,
      "step": 14388
    },
    {
      "epoch": 0.660318480106466,
      "grad_norm": 0.4796127676963806,
      "learning_rate": 9.006113187188676e-06,
      "loss": 0.3501,
      "step": 14389
    },
    {
      "epoch": 0.6603643706117205,
      "grad_norm": 0.4823330044746399,
      "learning_rate": 9.005966470179712e-06,
      "loss": 0.4156,
      "step": 14390
    },
    {
      "epoch": 0.6604102611169749,
      "grad_norm": 0.4508378803730011,
      "learning_rate": 9.005819743537666e-06,
      "loss": 0.3661,
      "step": 14391
    },
    {
      "epoch": 0.6604561516222294,
      "grad_norm": 0.44636479020118713,
      "learning_rate": 9.005673007262883e-06,
      "loss": 0.3417,
      "step": 14392
    },
    {
      "epoch": 0.6605020421274839,
      "grad_norm": 0.4510963559150696,
      "learning_rate": 9.005526261355724e-06,
      "loss": 0.3894,
      "step": 14393
    },
    {
      "epoch": 0.6605479326327383,
      "grad_norm": 0.4635971188545227,
      "learning_rate": 9.005379505816537e-06,
      "loss": 0.4203,
      "step": 14394
    },
    {
      "epoch": 0.6605938231379928,
      "grad_norm": 0.4718793034553528,
      "learning_rate": 9.005232740645677e-06,
      "loss": 0.361,
      "step": 14395
    },
    {
      "epoch": 0.6606397136432473,
      "grad_norm": 0.46480438113212585,
      "learning_rate": 9.005085965843496e-06,
      "loss": 0.4049,
      "step": 14396
    },
    {
      "epoch": 0.6606856041485016,
      "grad_norm": 0.41840851306915283,
      "learning_rate": 9.004939181410348e-06,
      "loss": 0.3314,
      "step": 14397
    },
    {
      "epoch": 0.6607314946537561,
      "grad_norm": 0.4678163528442383,
      "learning_rate": 9.004792387346583e-06,
      "loss": 0.4545,
      "step": 14398
    },
    {
      "epoch": 0.6607773851590106,
      "grad_norm": 0.4513968229293823,
      "learning_rate": 9.004645583652558e-06,
      "loss": 0.3961,
      "step": 14399
    },
    {
      "epoch": 0.660823275664265,
      "grad_norm": 0.47312647104263306,
      "learning_rate": 9.004498770328626e-06,
      "loss": 0.4601,
      "step": 14400
    },
    {
      "epoch": 0.6608691661695195,
      "grad_norm": 0.4664294719696045,
      "learning_rate": 9.004351947375136e-06,
      "loss": 0.4182,
      "step": 14401
    },
    {
      "epoch": 0.660915056674774,
      "grad_norm": 0.46030306816101074,
      "learning_rate": 9.004205114792444e-06,
      "loss": 0.4505,
      "step": 14402
    },
    {
      "epoch": 0.6609609471800284,
      "grad_norm": 0.45685482025146484,
      "learning_rate": 9.004058272580902e-06,
      "loss": 0.3583,
      "step": 14403
    },
    {
      "epoch": 0.6610068376852829,
      "grad_norm": 0.49850285053253174,
      "learning_rate": 9.003911420740866e-06,
      "loss": 0.4697,
      "step": 14404
    },
    {
      "epoch": 0.6610527281905374,
      "grad_norm": 0.4729253351688385,
      "learning_rate": 9.003764559272684e-06,
      "loss": 0.4314,
      "step": 14405
    },
    {
      "epoch": 0.6610986186957919,
      "grad_norm": 0.4743448793888092,
      "learning_rate": 9.003617688176712e-06,
      "loss": 0.4117,
      "step": 14406
    },
    {
      "epoch": 0.6611445092010463,
      "grad_norm": 0.44755828380584717,
      "learning_rate": 9.003470807453304e-06,
      "loss": 0.3448,
      "step": 14407
    },
    {
      "epoch": 0.6611903997063008,
      "grad_norm": 0.42651161551475525,
      "learning_rate": 9.003323917102811e-06,
      "loss": 0.3361,
      "step": 14408
    },
    {
      "epoch": 0.6612362902115553,
      "grad_norm": 0.41352325677871704,
      "learning_rate": 9.003177017125589e-06,
      "loss": 0.3552,
      "step": 14409
    },
    {
      "epoch": 0.6612821807168097,
      "grad_norm": 0.5120527744293213,
      "learning_rate": 9.003030107521988e-06,
      "loss": 0.4965,
      "step": 14410
    },
    {
      "epoch": 0.6613280712220642,
      "grad_norm": 0.5948606729507446,
      "learning_rate": 9.002883188292365e-06,
      "loss": 0.4096,
      "step": 14411
    },
    {
      "epoch": 0.6613739617273187,
      "grad_norm": 0.5069320797920227,
      "learning_rate": 9.00273625943707e-06,
      "loss": 0.4442,
      "step": 14412
    },
    {
      "epoch": 0.661419852232573,
      "grad_norm": 0.44752973318099976,
      "learning_rate": 9.002589320956458e-06,
      "loss": 0.3978,
      "step": 14413
    },
    {
      "epoch": 0.6614657427378275,
      "grad_norm": 0.4588022530078888,
      "learning_rate": 9.002442372850881e-06,
      "loss": 0.4436,
      "step": 14414
    },
    {
      "epoch": 0.661511633243082,
      "grad_norm": 0.5019850134849548,
      "learning_rate": 9.002295415120695e-06,
      "loss": 0.4936,
      "step": 14415
    },
    {
      "epoch": 0.6615575237483364,
      "grad_norm": 0.49573180079460144,
      "learning_rate": 9.00214844776625e-06,
      "loss": 0.5295,
      "step": 14416
    },
    {
      "epoch": 0.6616034142535909,
      "grad_norm": 0.5096704959869385,
      "learning_rate": 9.002001470787902e-06,
      "loss": 0.5112,
      "step": 14417
    },
    {
      "epoch": 0.6616493047588454,
      "grad_norm": 0.4680679142475128,
      "learning_rate": 9.001854484186005e-06,
      "loss": 0.3913,
      "step": 14418
    },
    {
      "epoch": 0.6616951952640998,
      "grad_norm": 0.4610459506511688,
      "learning_rate": 9.001707487960908e-06,
      "loss": 0.3951,
      "step": 14419
    },
    {
      "epoch": 0.6617410857693543,
      "grad_norm": 0.4382835626602173,
      "learning_rate": 9.00156048211297e-06,
      "loss": 0.3489,
      "step": 14420
    },
    {
      "epoch": 0.6617869762746088,
      "grad_norm": 0.4403751492500305,
      "learning_rate": 9.00141346664254e-06,
      "loss": 0.3629,
      "step": 14421
    },
    {
      "epoch": 0.6618328667798632,
      "grad_norm": 0.48926788568496704,
      "learning_rate": 9.001266441549975e-06,
      "loss": 0.474,
      "step": 14422
    },
    {
      "epoch": 0.6618787572851177,
      "grad_norm": 0.49508532881736755,
      "learning_rate": 9.001119406835626e-06,
      "loss": 0.4787,
      "step": 14423
    },
    {
      "epoch": 0.6619246477903722,
      "grad_norm": 0.4106306731700897,
      "learning_rate": 9.000972362499848e-06,
      "loss": 0.3086,
      "step": 14424
    },
    {
      "epoch": 0.6619705382956267,
      "grad_norm": 0.43695786595344543,
      "learning_rate": 9.000825308542995e-06,
      "loss": 0.3671,
      "step": 14425
    },
    {
      "epoch": 0.6620164288008811,
      "grad_norm": 0.46513378620147705,
      "learning_rate": 9.000678244965418e-06,
      "loss": 0.441,
      "step": 14426
    },
    {
      "epoch": 0.6620623193061356,
      "grad_norm": 0.47764840722084045,
      "learning_rate": 9.000531171767473e-06,
      "loss": 0.4182,
      "step": 14427
    },
    {
      "epoch": 0.6621082098113901,
      "grad_norm": 0.4981585144996643,
      "learning_rate": 9.000384088949514e-06,
      "loss": 0.4649,
      "step": 14428
    },
    {
      "epoch": 0.6621541003166445,
      "grad_norm": 0.4998481869697571,
      "learning_rate": 9.000236996511895e-06,
      "loss": 0.4607,
      "step": 14429
    },
    {
      "epoch": 0.662199990821899,
      "grad_norm": 0.41319283843040466,
      "learning_rate": 9.000089894454966e-06,
      "loss": 0.3318,
      "step": 14430
    },
    {
      "epoch": 0.6622458813271535,
      "grad_norm": 0.449195921421051,
      "learning_rate": 8.999942782779086e-06,
      "loss": 0.3836,
      "step": 14431
    },
    {
      "epoch": 0.6622917718324078,
      "grad_norm": 0.4686277508735657,
      "learning_rate": 8.999795661484603e-06,
      "loss": 0.4201,
      "step": 14432
    },
    {
      "epoch": 0.6623376623376623,
      "grad_norm": 0.5053922533988953,
      "learning_rate": 8.999648530571874e-06,
      "loss": 0.422,
      "step": 14433
    },
    {
      "epoch": 0.6623835528429168,
      "grad_norm": 0.4889240264892578,
      "learning_rate": 8.999501390041254e-06,
      "loss": 0.4721,
      "step": 14434
    },
    {
      "epoch": 0.6624294433481712,
      "grad_norm": 0.461119145154953,
      "learning_rate": 8.999354239893095e-06,
      "loss": 0.3945,
      "step": 14435
    },
    {
      "epoch": 0.6624753338534257,
      "grad_norm": 0.46428972482681274,
      "learning_rate": 8.999207080127752e-06,
      "loss": 0.456,
      "step": 14436
    },
    {
      "epoch": 0.6625212243586802,
      "grad_norm": 0.4733748137950897,
      "learning_rate": 8.999059910745577e-06,
      "loss": 0.4364,
      "step": 14437
    },
    {
      "epoch": 0.6625671148639346,
      "grad_norm": 0.45619314908981323,
      "learning_rate": 8.998912731746927e-06,
      "loss": 0.3941,
      "step": 14438
    },
    {
      "epoch": 0.6626130053691891,
      "grad_norm": 0.4507034718990326,
      "learning_rate": 8.998765543132153e-06,
      "loss": 0.4018,
      "step": 14439
    },
    {
      "epoch": 0.6626588958744436,
      "grad_norm": 0.4502747654914856,
      "learning_rate": 8.998618344901609e-06,
      "loss": 0.4111,
      "step": 14440
    },
    {
      "epoch": 0.662704786379698,
      "grad_norm": 0.4471191167831421,
      "learning_rate": 8.99847113705565e-06,
      "loss": 0.3817,
      "step": 14441
    },
    {
      "epoch": 0.6627506768849525,
      "grad_norm": 0.45347607135772705,
      "learning_rate": 8.99832391959463e-06,
      "loss": 0.3988,
      "step": 14442
    },
    {
      "epoch": 0.662796567390207,
      "grad_norm": 0.4385649561882019,
      "learning_rate": 8.998176692518904e-06,
      "loss": 0.3471,
      "step": 14443
    },
    {
      "epoch": 0.6628424578954615,
      "grad_norm": 0.43591105937957764,
      "learning_rate": 8.998029455828822e-06,
      "loss": 0.3484,
      "step": 14444
    },
    {
      "epoch": 0.6628883484007159,
      "grad_norm": 0.46470215916633606,
      "learning_rate": 8.997882209524745e-06,
      "loss": 0.447,
      "step": 14445
    },
    {
      "epoch": 0.6629342389059704,
      "grad_norm": 0.44905996322631836,
      "learning_rate": 8.99773495360702e-06,
      "loss": 0.4161,
      "step": 14446
    },
    {
      "epoch": 0.6629801294112249,
      "grad_norm": 0.4456089735031128,
      "learning_rate": 8.997587688076006e-06,
      "loss": 0.3882,
      "step": 14447
    },
    {
      "epoch": 0.6630260199164792,
      "grad_norm": 0.4242366850376129,
      "learning_rate": 8.997440412932054e-06,
      "loss": 0.3331,
      "step": 14448
    },
    {
      "epoch": 0.6630719104217337,
      "grad_norm": 0.44686028361320496,
      "learning_rate": 8.997293128175521e-06,
      "loss": 0.4227,
      "step": 14449
    },
    {
      "epoch": 0.6631178009269882,
      "grad_norm": 0.45099058747291565,
      "learning_rate": 8.997145833806759e-06,
      "loss": 0.3418,
      "step": 14450
    },
    {
      "epoch": 0.6631636914322426,
      "grad_norm": 0.4072268307209015,
      "learning_rate": 8.996998529826122e-06,
      "loss": 0.2773,
      "step": 14451
    },
    {
      "epoch": 0.6632095819374971,
      "grad_norm": 0.44425255060195923,
      "learning_rate": 8.996851216233967e-06,
      "loss": 0.3559,
      "step": 14452
    },
    {
      "epoch": 0.6632554724427516,
      "grad_norm": 0.46948766708374023,
      "learning_rate": 8.996703893030642e-06,
      "loss": 0.4116,
      "step": 14453
    },
    {
      "epoch": 0.663301362948006,
      "grad_norm": 0.9936320185661316,
      "learning_rate": 8.99655656021651e-06,
      "loss": 0.4005,
      "step": 14454
    },
    {
      "epoch": 0.6633472534532605,
      "grad_norm": 0.45868706703186035,
      "learning_rate": 8.996409217791919e-06,
      "loss": 0.412,
      "step": 14455
    },
    {
      "epoch": 0.663393143958515,
      "grad_norm": 0.4456869661808014,
      "learning_rate": 8.996261865757224e-06,
      "loss": 0.3545,
      "step": 14456
    },
    {
      "epoch": 0.6634390344637694,
      "grad_norm": 0.4670179784297943,
      "learning_rate": 8.996114504112784e-06,
      "loss": 0.3862,
      "step": 14457
    },
    {
      "epoch": 0.6634849249690239,
      "grad_norm": 0.47607988119125366,
      "learning_rate": 8.995967132858945e-06,
      "loss": 0.45,
      "step": 14458
    },
    {
      "epoch": 0.6635308154742784,
      "grad_norm": 0.4479120373725891,
      "learning_rate": 8.995819751996069e-06,
      "loss": 0.3858,
      "step": 14459
    },
    {
      "epoch": 0.6635767059795329,
      "grad_norm": 0.5265428423881531,
      "learning_rate": 8.995672361524508e-06,
      "loss": 0.5329,
      "step": 14460
    },
    {
      "epoch": 0.6636225964847873,
      "grad_norm": 0.4702661335468292,
      "learning_rate": 8.995524961444614e-06,
      "loss": 0.4157,
      "step": 14461
    },
    {
      "epoch": 0.6636684869900418,
      "grad_norm": 0.4336968958377838,
      "learning_rate": 8.995377551756746e-06,
      "loss": 0.3322,
      "step": 14462
    },
    {
      "epoch": 0.6637143774952963,
      "grad_norm": 0.4466639459133148,
      "learning_rate": 8.995230132461254e-06,
      "loss": 0.3634,
      "step": 14463
    },
    {
      "epoch": 0.6637602680005507,
      "grad_norm": 0.4759974479675293,
      "learning_rate": 8.995082703558496e-06,
      "loss": 0.4997,
      "step": 14464
    },
    {
      "epoch": 0.6638061585058052,
      "grad_norm": 0.44881126284599304,
      "learning_rate": 8.994935265048823e-06,
      "loss": 0.4193,
      "step": 14465
    },
    {
      "epoch": 0.6638520490110597,
      "grad_norm": 0.45830437541007996,
      "learning_rate": 8.994787816932595e-06,
      "loss": 0.4268,
      "step": 14466
    },
    {
      "epoch": 0.663897939516314,
      "grad_norm": 0.4368213415145874,
      "learning_rate": 8.99464035921016e-06,
      "loss": 0.4081,
      "step": 14467
    },
    {
      "epoch": 0.6639438300215685,
      "grad_norm": 0.5081778764724731,
      "learning_rate": 8.994492891881876e-06,
      "loss": 0.3771,
      "step": 14468
    },
    {
      "epoch": 0.663989720526823,
      "grad_norm": 0.46657243371009827,
      "learning_rate": 8.994345414948098e-06,
      "loss": 0.3276,
      "step": 14469
    },
    {
      "epoch": 0.6640356110320774,
      "grad_norm": 0.433438777923584,
      "learning_rate": 8.99419792840918e-06,
      "loss": 0.3954,
      "step": 14470
    },
    {
      "epoch": 0.6640815015373319,
      "grad_norm": 0.42769020795822144,
      "learning_rate": 8.994050432265476e-06,
      "loss": 0.3323,
      "step": 14471
    },
    {
      "epoch": 0.6641273920425864,
      "grad_norm": 0.49134954810142517,
      "learning_rate": 8.99390292651734e-06,
      "loss": 0.5142,
      "step": 14472
    },
    {
      "epoch": 0.6641732825478408,
      "grad_norm": 0.458547979593277,
      "learning_rate": 8.993755411165131e-06,
      "loss": 0.4146,
      "step": 14473
    },
    {
      "epoch": 0.6642191730530953,
      "grad_norm": 0.5217342376708984,
      "learning_rate": 8.9936078862092e-06,
      "loss": 0.406,
      "step": 14474
    },
    {
      "epoch": 0.6642650635583498,
      "grad_norm": 0.4476647675037384,
      "learning_rate": 8.993460351649902e-06,
      "loss": 0.385,
      "step": 14475
    },
    {
      "epoch": 0.6643109540636042,
      "grad_norm": 0.4613754153251648,
      "learning_rate": 8.993312807487591e-06,
      "loss": 0.3942,
      "step": 14476
    },
    {
      "epoch": 0.6643568445688587,
      "grad_norm": 0.46307024359703064,
      "learning_rate": 8.993165253722624e-06,
      "loss": 0.3989,
      "step": 14477
    },
    {
      "epoch": 0.6644027350741132,
      "grad_norm": 0.4667181968688965,
      "learning_rate": 8.993017690355355e-06,
      "loss": 0.3628,
      "step": 14478
    },
    {
      "epoch": 0.6644486255793677,
      "grad_norm": 0.4860840439796448,
      "learning_rate": 8.992870117386138e-06,
      "loss": 0.3664,
      "step": 14479
    },
    {
      "epoch": 0.6644945160846221,
      "grad_norm": 0.42400819063186646,
      "learning_rate": 8.99272253481533e-06,
      "loss": 0.3631,
      "step": 14480
    },
    {
      "epoch": 0.6645404065898766,
      "grad_norm": 0.4680634140968323,
      "learning_rate": 8.992574942643283e-06,
      "loss": 0.3683,
      "step": 14481
    },
    {
      "epoch": 0.6645862970951311,
      "grad_norm": 0.46209123730659485,
      "learning_rate": 8.992427340870355e-06,
      "loss": 0.428,
      "step": 14482
    },
    {
      "epoch": 0.6646321876003854,
      "grad_norm": 0.4881763458251953,
      "learning_rate": 8.992279729496898e-06,
      "loss": 0.4533,
      "step": 14483
    },
    {
      "epoch": 0.66467807810564,
      "grad_norm": 0.43561673164367676,
      "learning_rate": 8.992132108523268e-06,
      "loss": 0.3243,
      "step": 14484
    },
    {
      "epoch": 0.6647239686108944,
      "grad_norm": 0.474380761384964,
      "learning_rate": 8.99198447794982e-06,
      "loss": 0.4367,
      "step": 14485
    },
    {
      "epoch": 0.6647698591161488,
      "grad_norm": 0.44446250796318054,
      "learning_rate": 8.991836837776911e-06,
      "loss": 0.4074,
      "step": 14486
    },
    {
      "epoch": 0.6648157496214033,
      "grad_norm": 0.4481843113899231,
      "learning_rate": 8.991689188004892e-06,
      "loss": 0.3712,
      "step": 14487
    },
    {
      "epoch": 0.6648616401266578,
      "grad_norm": 0.44498389959335327,
      "learning_rate": 8.991541528634122e-06,
      "loss": 0.4235,
      "step": 14488
    },
    {
      "epoch": 0.6649075306319122,
      "grad_norm": 0.48705291748046875,
      "learning_rate": 8.991393859664954e-06,
      "loss": 0.4217,
      "step": 14489
    },
    {
      "epoch": 0.6649534211371667,
      "grad_norm": 0.46249619126319885,
      "learning_rate": 8.991246181097743e-06,
      "loss": 0.3389,
      "step": 14490
    },
    {
      "epoch": 0.6649993116424212,
      "grad_norm": 0.4388614296913147,
      "learning_rate": 8.991098492932845e-06,
      "loss": 0.4142,
      "step": 14491
    },
    {
      "epoch": 0.6650452021476756,
      "grad_norm": 0.41345667839050293,
      "learning_rate": 8.990950795170616e-06,
      "loss": 0.3253,
      "step": 14492
    },
    {
      "epoch": 0.6650910926529301,
      "grad_norm": 0.4584944546222687,
      "learning_rate": 8.990803087811408e-06,
      "loss": 0.3939,
      "step": 14493
    },
    {
      "epoch": 0.6651369831581846,
      "grad_norm": 0.5011733174324036,
      "learning_rate": 8.990655370855579e-06,
      "loss": 0.4773,
      "step": 14494
    },
    {
      "epoch": 0.6651828736634391,
      "grad_norm": 0.45549070835113525,
      "learning_rate": 8.990507644303483e-06,
      "loss": 0.4169,
      "step": 14495
    },
    {
      "epoch": 0.6652287641686935,
      "grad_norm": 0.4474542438983917,
      "learning_rate": 8.990359908155476e-06,
      "loss": 0.3782,
      "step": 14496
    },
    {
      "epoch": 0.665274654673948,
      "grad_norm": 0.4395194351673126,
      "learning_rate": 8.990212162411914e-06,
      "loss": 0.3451,
      "step": 14497
    },
    {
      "epoch": 0.6653205451792025,
      "grad_norm": 0.4599972367286682,
      "learning_rate": 8.990064407073149e-06,
      "loss": 0.3913,
      "step": 14498
    },
    {
      "epoch": 0.6653664356844569,
      "grad_norm": 0.4522850811481476,
      "learning_rate": 8.989916642139539e-06,
      "loss": 0.3441,
      "step": 14499
    },
    {
      "epoch": 0.6654123261897114,
      "grad_norm": 0.4209553897380829,
      "learning_rate": 8.989768867611439e-06,
      "loss": 0.3208,
      "step": 14500
    },
    {
      "epoch": 0.6654582166949659,
      "grad_norm": 0.42573878169059753,
      "learning_rate": 8.989621083489202e-06,
      "loss": 0.3388,
      "step": 14501
    },
    {
      "epoch": 0.6655041072002202,
      "grad_norm": 0.469412237405777,
      "learning_rate": 8.989473289773187e-06,
      "loss": 0.4739,
      "step": 14502
    },
    {
      "epoch": 0.6655499977054747,
      "grad_norm": 0.4056329131126404,
      "learning_rate": 8.98932548646375e-06,
      "loss": 0.2768,
      "step": 14503
    },
    {
      "epoch": 0.6655958882107292,
      "grad_norm": 0.47525283694267273,
      "learning_rate": 8.98917767356124e-06,
      "loss": 0.4509,
      "step": 14504
    },
    {
      "epoch": 0.6656417787159836,
      "grad_norm": 0.4655442237854004,
      "learning_rate": 8.989029851066021e-06,
      "loss": 0.4029,
      "step": 14505
    },
    {
      "epoch": 0.6656876692212381,
      "grad_norm": 0.4710235595703125,
      "learning_rate": 8.988882018978443e-06,
      "loss": 0.3992,
      "step": 14506
    },
    {
      "epoch": 0.6657335597264926,
      "grad_norm": 0.518703818321228,
      "learning_rate": 8.98873417729886e-06,
      "loss": 0.4532,
      "step": 14507
    },
    {
      "epoch": 0.665779450231747,
      "grad_norm": 0.47108733654022217,
      "learning_rate": 8.988586326027632e-06,
      "loss": 0.4316,
      "step": 14508
    },
    {
      "epoch": 0.6658253407370015,
      "grad_norm": 0.4415122866630554,
      "learning_rate": 8.988438465165113e-06,
      "loss": 0.4086,
      "step": 14509
    },
    {
      "epoch": 0.665871231242256,
      "grad_norm": 0.4599289000034332,
      "learning_rate": 8.988290594711659e-06,
      "loss": 0.3222,
      "step": 14510
    },
    {
      "epoch": 0.6659171217475104,
      "grad_norm": 0.4541757106781006,
      "learning_rate": 8.988142714667623e-06,
      "loss": 0.4082,
      "step": 14511
    },
    {
      "epoch": 0.6659630122527649,
      "grad_norm": 0.44427844882011414,
      "learning_rate": 8.987994825033363e-06,
      "loss": 0.361,
      "step": 14512
    },
    {
      "epoch": 0.6660089027580194,
      "grad_norm": 0.42946553230285645,
      "learning_rate": 8.987846925809235e-06,
      "loss": 0.3422,
      "step": 14513
    },
    {
      "epoch": 0.6660547932632739,
      "grad_norm": 0.56458979845047,
      "learning_rate": 8.987699016995592e-06,
      "loss": 0.5755,
      "step": 14514
    },
    {
      "epoch": 0.6661006837685283,
      "grad_norm": 0.4725033640861511,
      "learning_rate": 8.987551098592795e-06,
      "loss": 0.4118,
      "step": 14515
    },
    {
      "epoch": 0.6661465742737828,
      "grad_norm": 0.46170443296432495,
      "learning_rate": 8.987403170601193e-06,
      "loss": 0.3907,
      "step": 14516
    },
    {
      "epoch": 0.6661924647790373,
      "grad_norm": 0.4450153410434723,
      "learning_rate": 8.987255233021145e-06,
      "loss": 0.3392,
      "step": 14517
    },
    {
      "epoch": 0.6662383552842917,
      "grad_norm": 0.4926857650279999,
      "learning_rate": 8.987107285853008e-06,
      "loss": 0.5061,
      "step": 14518
    },
    {
      "epoch": 0.6662842457895461,
      "grad_norm": 0.44132015109062195,
      "learning_rate": 8.986959329097135e-06,
      "loss": 0.369,
      "step": 14519
    },
    {
      "epoch": 0.6663301362948006,
      "grad_norm": 0.46381857991218567,
      "learning_rate": 8.986811362753884e-06,
      "loss": 0.4105,
      "step": 14520
    },
    {
      "epoch": 0.666376026800055,
      "grad_norm": 0.450851708650589,
      "learning_rate": 8.98666338682361e-06,
      "loss": 0.3662,
      "step": 14521
    },
    {
      "epoch": 0.6664219173053095,
      "grad_norm": 0.9633417725563049,
      "learning_rate": 8.986515401306669e-06,
      "loss": 0.4472,
      "step": 14522
    },
    {
      "epoch": 0.666467807810564,
      "grad_norm": 0.4649474620819092,
      "learning_rate": 8.986367406203416e-06,
      "loss": 0.3965,
      "step": 14523
    },
    {
      "epoch": 0.6665136983158184,
      "grad_norm": 0.4647963345050812,
      "learning_rate": 8.986219401514207e-06,
      "loss": 0.4213,
      "step": 14524
    },
    {
      "epoch": 0.6665595888210729,
      "grad_norm": 0.46133482456207275,
      "learning_rate": 8.986071387239399e-06,
      "loss": 0.3496,
      "step": 14525
    },
    {
      "epoch": 0.6666054793263274,
      "grad_norm": 0.44894808530807495,
      "learning_rate": 8.985923363379348e-06,
      "loss": 0.3457,
      "step": 14526
    },
    {
      "epoch": 0.6666513698315818,
      "grad_norm": 0.46966975927352905,
      "learning_rate": 8.985775329934408e-06,
      "loss": 0.4401,
      "step": 14527
    },
    {
      "epoch": 0.6666972603368363,
      "grad_norm": 0.4736202657222748,
      "learning_rate": 8.985627286904936e-06,
      "loss": 0.4353,
      "step": 14528
    },
    {
      "epoch": 0.6667431508420908,
      "grad_norm": 0.48000848293304443,
      "learning_rate": 8.98547923429129e-06,
      "loss": 0.4629,
      "step": 14529
    },
    {
      "epoch": 0.6667890413473452,
      "grad_norm": 0.39614972472190857,
      "learning_rate": 8.985331172093823e-06,
      "loss": 0.2836,
      "step": 14530
    },
    {
      "epoch": 0.6668349318525997,
      "grad_norm": 0.5184483528137207,
      "learning_rate": 8.985183100312892e-06,
      "loss": 0.4443,
      "step": 14531
    },
    {
      "epoch": 0.6668808223578542,
      "grad_norm": 0.502870500087738,
      "learning_rate": 8.985035018948856e-06,
      "loss": 0.4438,
      "step": 14532
    },
    {
      "epoch": 0.6669267128631087,
      "grad_norm": 0.45267996191978455,
      "learning_rate": 8.984886928002066e-06,
      "loss": 0.4024,
      "step": 14533
    },
    {
      "epoch": 0.6669726033683631,
      "grad_norm": 0.4268811345100403,
      "learning_rate": 8.98473882747288e-06,
      "loss": 0.3448,
      "step": 14534
    },
    {
      "epoch": 0.6670184938736176,
      "grad_norm": 0.4367494583129883,
      "learning_rate": 8.984590717361657e-06,
      "loss": 0.3023,
      "step": 14535
    },
    {
      "epoch": 0.667064384378872,
      "grad_norm": 0.5010473728179932,
      "learning_rate": 8.98444259766875e-06,
      "loss": 0.4966,
      "step": 14536
    },
    {
      "epoch": 0.6671102748841264,
      "grad_norm": 0.4206502437591553,
      "learning_rate": 8.984294468394517e-06,
      "loss": 0.3492,
      "step": 14537
    },
    {
      "epoch": 0.6671561653893809,
      "grad_norm": 0.4631282389163971,
      "learning_rate": 8.984146329539312e-06,
      "loss": 0.4074,
      "step": 14538
    },
    {
      "epoch": 0.6672020558946354,
      "grad_norm": 0.4614452123641968,
      "learning_rate": 8.983998181103493e-06,
      "loss": 0.4561,
      "step": 14539
    },
    {
      "epoch": 0.6672479463998898,
      "grad_norm": 0.4451823830604553,
      "learning_rate": 8.983850023087416e-06,
      "loss": 0.4328,
      "step": 14540
    },
    {
      "epoch": 0.6672938369051443,
      "grad_norm": 0.4404507875442505,
      "learning_rate": 8.983701855491436e-06,
      "loss": 0.3616,
      "step": 14541
    },
    {
      "epoch": 0.6673397274103988,
      "grad_norm": 0.4492647349834442,
      "learning_rate": 8.98355367831591e-06,
      "loss": 0.3417,
      "step": 14542
    },
    {
      "epoch": 0.6673856179156532,
      "grad_norm": 0.4315362870693207,
      "learning_rate": 8.983405491561197e-06,
      "loss": 0.3581,
      "step": 14543
    },
    {
      "epoch": 0.6674315084209077,
      "grad_norm": 0.4772603511810303,
      "learning_rate": 8.98325729522765e-06,
      "loss": 0.3932,
      "step": 14544
    },
    {
      "epoch": 0.6674773989261622,
      "grad_norm": 0.5019184350967407,
      "learning_rate": 8.983109089315625e-06,
      "loss": 0.4983,
      "step": 14545
    },
    {
      "epoch": 0.6675232894314166,
      "grad_norm": 0.49283310770988464,
      "learning_rate": 8.982960873825481e-06,
      "loss": 0.4789,
      "step": 14546
    },
    {
      "epoch": 0.6675691799366711,
      "grad_norm": 0.4469640851020813,
      "learning_rate": 8.982812648757573e-06,
      "loss": 0.365,
      "step": 14547
    },
    {
      "epoch": 0.6676150704419256,
      "grad_norm": 0.5307384729385376,
      "learning_rate": 8.982664414112258e-06,
      "loss": 0.5309,
      "step": 14548
    },
    {
      "epoch": 0.6676609609471801,
      "grad_norm": 0.4557315707206726,
      "learning_rate": 8.98251616988989e-06,
      "loss": 0.3406,
      "step": 14549
    },
    {
      "epoch": 0.6677068514524345,
      "grad_norm": 0.42902955412864685,
      "learning_rate": 8.98236791609083e-06,
      "loss": 0.3635,
      "step": 14550
    },
    {
      "epoch": 0.667752741957689,
      "grad_norm": 0.45770910382270813,
      "learning_rate": 8.982219652715431e-06,
      "loss": 0.3486,
      "step": 14551
    },
    {
      "epoch": 0.6677986324629435,
      "grad_norm": 0.3997988998889923,
      "learning_rate": 8.98207137976405e-06,
      "loss": 0.2828,
      "step": 14552
    },
    {
      "epoch": 0.6678445229681979,
      "grad_norm": 0.39941275119781494,
      "learning_rate": 8.981923097237047e-06,
      "loss": 0.2977,
      "step": 14553
    },
    {
      "epoch": 0.6678904134734523,
      "grad_norm": 0.4932500422000885,
      "learning_rate": 8.981774805134772e-06,
      "loss": 0.4292,
      "step": 14554
    },
    {
      "epoch": 0.6679363039787068,
      "grad_norm": 0.4331819415092468,
      "learning_rate": 8.981626503457586e-06,
      "loss": 0.3473,
      "step": 14555
    },
    {
      "epoch": 0.6679821944839612,
      "grad_norm": 0.49689358472824097,
      "learning_rate": 8.981478192205845e-06,
      "loss": 0.4922,
      "step": 14556
    },
    {
      "epoch": 0.6680280849892157,
      "grad_norm": 0.5731780529022217,
      "learning_rate": 8.981329871379907e-06,
      "loss": 0.4214,
      "step": 14557
    },
    {
      "epoch": 0.6680739754944702,
      "grad_norm": 0.443897545337677,
      "learning_rate": 8.981181540980126e-06,
      "loss": 0.4044,
      "step": 14558
    },
    {
      "epoch": 0.6681198659997246,
      "grad_norm": 0.4272734820842743,
      "learning_rate": 8.98103320100686e-06,
      "loss": 0.3732,
      "step": 14559
    },
    {
      "epoch": 0.6681657565049791,
      "grad_norm": 0.4343867301940918,
      "learning_rate": 8.980884851460467e-06,
      "loss": 0.3385,
      "step": 14560
    },
    {
      "epoch": 0.6682116470102336,
      "grad_norm": 0.46007993817329407,
      "learning_rate": 8.980736492341301e-06,
      "loss": 0.4155,
      "step": 14561
    },
    {
      "epoch": 0.668257537515488,
      "grad_norm": 0.4299311935901642,
      "learning_rate": 8.980588123649721e-06,
      "loss": 0.3426,
      "step": 14562
    },
    {
      "epoch": 0.6683034280207425,
      "grad_norm": 0.4787943661212921,
      "learning_rate": 8.980439745386081e-06,
      "loss": 0.4455,
      "step": 14563
    },
    {
      "epoch": 0.668349318525997,
      "grad_norm": 0.46967437863349915,
      "learning_rate": 8.980291357550742e-06,
      "loss": 0.4884,
      "step": 14564
    },
    {
      "epoch": 0.6683952090312514,
      "grad_norm": 0.44886350631713867,
      "learning_rate": 8.980142960144057e-06,
      "loss": 0.3801,
      "step": 14565
    },
    {
      "epoch": 0.6684410995365059,
      "grad_norm": 0.4726877808570862,
      "learning_rate": 8.979994553166385e-06,
      "loss": 0.3976,
      "step": 14566
    },
    {
      "epoch": 0.6684869900417604,
      "grad_norm": 0.4741431474685669,
      "learning_rate": 8.97984613661808e-06,
      "loss": 0.4569,
      "step": 14567
    },
    {
      "epoch": 0.6685328805470149,
      "grad_norm": 0.45089778304100037,
      "learning_rate": 8.979697710499505e-06,
      "loss": 0.409,
      "step": 14568
    },
    {
      "epoch": 0.6685787710522693,
      "grad_norm": 0.4743574857711792,
      "learning_rate": 8.97954927481101e-06,
      "loss": 0.4622,
      "step": 14569
    },
    {
      "epoch": 0.6686246615575238,
      "grad_norm": 0.48954150080680847,
      "learning_rate": 8.979400829552956e-06,
      "loss": 0.4292,
      "step": 14570
    },
    {
      "epoch": 0.6686705520627783,
      "grad_norm": 0.44962358474731445,
      "learning_rate": 8.9792523747257e-06,
      "loss": 0.4257,
      "step": 14571
    },
    {
      "epoch": 0.6687164425680326,
      "grad_norm": 0.4855929911136627,
      "learning_rate": 8.979103910329597e-06,
      "loss": 0.4557,
      "step": 14572
    },
    {
      "epoch": 0.6687623330732871,
      "grad_norm": 0.4931475520133972,
      "learning_rate": 8.978955436365005e-06,
      "loss": 0.4891,
      "step": 14573
    },
    {
      "epoch": 0.6688082235785416,
      "grad_norm": 0.4573855996131897,
      "learning_rate": 8.97880695283228e-06,
      "loss": 0.3818,
      "step": 14574
    },
    {
      "epoch": 0.668854114083796,
      "grad_norm": 0.45578575134277344,
      "learning_rate": 8.978658459731782e-06,
      "loss": 0.3665,
      "step": 14575
    },
    {
      "epoch": 0.6689000045890505,
      "grad_norm": 0.4740530252456665,
      "learning_rate": 8.978509957063864e-06,
      "loss": 0.4676,
      "step": 14576
    },
    {
      "epoch": 0.668945895094305,
      "grad_norm": 0.44990482926368713,
      "learning_rate": 8.978361444828885e-06,
      "loss": 0.4267,
      "step": 14577
    },
    {
      "epoch": 0.6689917855995594,
      "grad_norm": 0.46761006116867065,
      "learning_rate": 8.978212923027204e-06,
      "loss": 0.48,
      "step": 14578
    },
    {
      "epoch": 0.6690376761048139,
      "grad_norm": 0.4664131999015808,
      "learning_rate": 8.978064391659176e-06,
      "loss": 0.3967,
      "step": 14579
    },
    {
      "epoch": 0.6690835666100684,
      "grad_norm": 0.5295001268386841,
      "learning_rate": 8.977915850725159e-06,
      "loss": 0.5084,
      "step": 14580
    },
    {
      "epoch": 0.6691294571153228,
      "grad_norm": 0.46649545431137085,
      "learning_rate": 8.97776730022551e-06,
      "loss": 0.4218,
      "step": 14581
    },
    {
      "epoch": 0.6691753476205773,
      "grad_norm": 0.46385177969932556,
      "learning_rate": 8.977618740160586e-06,
      "loss": 0.3492,
      "step": 14582
    },
    {
      "epoch": 0.6692212381258318,
      "grad_norm": 0.4378233551979065,
      "learning_rate": 8.977470170530742e-06,
      "loss": 0.3492,
      "step": 14583
    },
    {
      "epoch": 0.6692671286310863,
      "grad_norm": 0.45596447587013245,
      "learning_rate": 8.97732159133634e-06,
      "loss": 0.3564,
      "step": 14584
    },
    {
      "epoch": 0.6693130191363407,
      "grad_norm": 0.46202313899993896,
      "learning_rate": 8.977173002577735e-06,
      "loss": 0.4165,
      "step": 14585
    },
    {
      "epoch": 0.6693589096415952,
      "grad_norm": 0.4557839632034302,
      "learning_rate": 8.977024404255283e-06,
      "loss": 0.4258,
      "step": 14586
    },
    {
      "epoch": 0.6694048001468497,
      "grad_norm": 0.49688103795051575,
      "learning_rate": 8.976875796369342e-06,
      "loss": 0.3961,
      "step": 14587
    },
    {
      "epoch": 0.669450690652104,
      "grad_norm": 0.455834299325943,
      "learning_rate": 8.976727178920271e-06,
      "loss": 0.3753,
      "step": 14588
    },
    {
      "epoch": 0.6694965811573586,
      "grad_norm": 0.5413268804550171,
      "learning_rate": 8.976578551908426e-06,
      "loss": 0.4244,
      "step": 14589
    },
    {
      "epoch": 0.669542471662613,
      "grad_norm": 0.43210649490356445,
      "learning_rate": 8.976429915334164e-06,
      "loss": 0.3519,
      "step": 14590
    },
    {
      "epoch": 0.6695883621678674,
      "grad_norm": 0.43613702058792114,
      "learning_rate": 8.976281269197844e-06,
      "loss": 0.3621,
      "step": 14591
    },
    {
      "epoch": 0.6696342526731219,
      "grad_norm": 0.44548624753952026,
      "learning_rate": 8.976132613499822e-06,
      "loss": 0.3904,
      "step": 14592
    },
    {
      "epoch": 0.6696801431783764,
      "grad_norm": 0.5107896327972412,
      "learning_rate": 8.975983948240457e-06,
      "loss": 0.5,
      "step": 14593
    },
    {
      "epoch": 0.6697260336836308,
      "grad_norm": 0.44990479946136475,
      "learning_rate": 8.975835273420103e-06,
      "loss": 0.4299,
      "step": 14594
    },
    {
      "epoch": 0.6697719241888853,
      "grad_norm": 0.44652318954467773,
      "learning_rate": 8.975686589039123e-06,
      "loss": 0.4021,
      "step": 14595
    },
    {
      "epoch": 0.6698178146941398,
      "grad_norm": 0.40969118475914,
      "learning_rate": 8.97553789509787e-06,
      "loss": 0.2973,
      "step": 14596
    },
    {
      "epoch": 0.6698637051993942,
      "grad_norm": 0.4366760849952698,
      "learning_rate": 8.975389191596703e-06,
      "loss": 0.3393,
      "step": 14597
    },
    {
      "epoch": 0.6699095957046487,
      "grad_norm": 0.49897250533103943,
      "learning_rate": 8.975240478535981e-06,
      "loss": 0.5323,
      "step": 14598
    },
    {
      "epoch": 0.6699554862099032,
      "grad_norm": 0.428006112575531,
      "learning_rate": 8.975091755916058e-06,
      "loss": 0.3204,
      "step": 14599
    },
    {
      "epoch": 0.6700013767151576,
      "grad_norm": 0.48574578762054443,
      "learning_rate": 8.974943023737294e-06,
      "loss": 0.4228,
      "step": 14600
    },
    {
      "epoch": 0.6700472672204121,
      "grad_norm": 0.4550846517086029,
      "learning_rate": 8.974794282000048e-06,
      "loss": 0.3444,
      "step": 14601
    },
    {
      "epoch": 0.6700931577256666,
      "grad_norm": 0.49023714661598206,
      "learning_rate": 8.974645530704675e-06,
      "loss": 0.4368,
      "step": 14602
    },
    {
      "epoch": 0.6701390482309211,
      "grad_norm": 0.512380838394165,
      "learning_rate": 8.974496769851535e-06,
      "loss": 0.475,
      "step": 14603
    },
    {
      "epoch": 0.6701849387361755,
      "grad_norm": 0.4899825155735016,
      "learning_rate": 8.974347999440985e-06,
      "loss": 0.4847,
      "step": 14604
    },
    {
      "epoch": 0.67023082924143,
      "grad_norm": 0.45175325870513916,
      "learning_rate": 8.974199219473382e-06,
      "loss": 0.4281,
      "step": 14605
    },
    {
      "epoch": 0.6702767197466845,
      "grad_norm": 0.48382988572120667,
      "learning_rate": 8.974050429949083e-06,
      "loss": 0.4152,
      "step": 14606
    },
    {
      "epoch": 0.6703226102519388,
      "grad_norm": 0.4746871292591095,
      "learning_rate": 8.973901630868448e-06,
      "loss": 0.4255,
      "step": 14607
    },
    {
      "epoch": 0.6703685007571933,
      "grad_norm": 0.46630939841270447,
      "learning_rate": 8.973752822231833e-06,
      "loss": 0.3901,
      "step": 14608
    },
    {
      "epoch": 0.6704143912624478,
      "grad_norm": 0.46052226424217224,
      "learning_rate": 8.973604004039598e-06,
      "loss": 0.4168,
      "step": 14609
    },
    {
      "epoch": 0.6704602817677022,
      "grad_norm": 0.5103359222412109,
      "learning_rate": 8.973455176292098e-06,
      "loss": 0.473,
      "step": 14610
    },
    {
      "epoch": 0.6705061722729567,
      "grad_norm": 0.47642409801483154,
      "learning_rate": 8.973306338989693e-06,
      "loss": 0.4881,
      "step": 14611
    },
    {
      "epoch": 0.6705520627782112,
      "grad_norm": 0.4452926218509674,
      "learning_rate": 8.973157492132742e-06,
      "loss": 0.394,
      "step": 14612
    },
    {
      "epoch": 0.6705979532834656,
      "grad_norm": 0.47272583842277527,
      "learning_rate": 8.973008635721599e-06,
      "loss": 0.4618,
      "step": 14613
    },
    {
      "epoch": 0.6706438437887201,
      "grad_norm": 0.46520477533340454,
      "learning_rate": 8.972859769756625e-06,
      "loss": 0.3628,
      "step": 14614
    },
    {
      "epoch": 0.6706897342939746,
      "grad_norm": 0.450047105550766,
      "learning_rate": 8.972710894238177e-06,
      "loss": 0.383,
      "step": 14615
    },
    {
      "epoch": 0.670735624799229,
      "grad_norm": 0.4566286504268646,
      "learning_rate": 8.972562009166613e-06,
      "loss": 0.435,
      "step": 14616
    },
    {
      "epoch": 0.6707815153044835,
      "grad_norm": 0.4192664325237274,
      "learning_rate": 8.972413114542291e-06,
      "loss": 0.3369,
      "step": 14617
    },
    {
      "epoch": 0.670827405809738,
      "grad_norm": 0.44099557399749756,
      "learning_rate": 8.97226421036557e-06,
      "loss": 0.3721,
      "step": 14618
    },
    {
      "epoch": 0.6708732963149924,
      "grad_norm": 0.43638014793395996,
      "learning_rate": 8.972115296636806e-06,
      "loss": 0.3469,
      "step": 14619
    },
    {
      "epoch": 0.6709191868202469,
      "grad_norm": 0.43593382835388184,
      "learning_rate": 8.971966373356358e-06,
      "loss": 0.3774,
      "step": 14620
    },
    {
      "epoch": 0.6709650773255014,
      "grad_norm": 0.44241613149642944,
      "learning_rate": 8.971817440524586e-06,
      "loss": 0.3465,
      "step": 14621
    },
    {
      "epoch": 0.6710109678307559,
      "grad_norm": 0.43346571922302246,
      "learning_rate": 8.971668498141845e-06,
      "loss": 0.3822,
      "step": 14622
    },
    {
      "epoch": 0.6710568583360103,
      "grad_norm": 0.4866408407688141,
      "learning_rate": 8.971519546208497e-06,
      "loss": 0.4261,
      "step": 14623
    },
    {
      "epoch": 0.6711027488412648,
      "grad_norm": 0.46682170033454895,
      "learning_rate": 8.971370584724897e-06,
      "loss": 0.3573,
      "step": 14624
    },
    {
      "epoch": 0.6711486393465192,
      "grad_norm": 0.4426473379135132,
      "learning_rate": 8.971221613691403e-06,
      "loss": 0.3665,
      "step": 14625
    },
    {
      "epoch": 0.6711945298517736,
      "grad_norm": 0.46930718421936035,
      "learning_rate": 8.971072633108376e-06,
      "loss": 0.4059,
      "step": 14626
    },
    {
      "epoch": 0.6712404203570281,
      "grad_norm": 0.4628923237323761,
      "learning_rate": 8.97092364297617e-06,
      "loss": 0.3733,
      "step": 14627
    },
    {
      "epoch": 0.6712863108622826,
      "grad_norm": 0.516466498374939,
      "learning_rate": 8.970774643295148e-06,
      "loss": 0.4747,
      "step": 14628
    },
    {
      "epoch": 0.671332201367537,
      "grad_norm": 0.4853317141532898,
      "learning_rate": 8.970625634065666e-06,
      "loss": 0.4156,
      "step": 14629
    },
    {
      "epoch": 0.6713780918727915,
      "grad_norm": 0.502302348613739,
      "learning_rate": 8.970476615288083e-06,
      "loss": 0.4347,
      "step": 14630
    },
    {
      "epoch": 0.671423982378046,
      "grad_norm": 0.4493391513824463,
      "learning_rate": 8.970327586962757e-06,
      "loss": 0.3599,
      "step": 14631
    },
    {
      "epoch": 0.6714698728833004,
      "grad_norm": 0.420227974653244,
      "learning_rate": 8.970178549090046e-06,
      "loss": 0.326,
      "step": 14632
    },
    {
      "epoch": 0.6715157633885549,
      "grad_norm": 0.41016077995300293,
      "learning_rate": 8.970029501670308e-06,
      "loss": 0.313,
      "step": 14633
    },
    {
      "epoch": 0.6715616538938094,
      "grad_norm": 0.4221818447113037,
      "learning_rate": 8.9698804447039e-06,
      "loss": 0.3294,
      "step": 14634
    },
    {
      "epoch": 0.6716075443990638,
      "grad_norm": 0.48555177450180054,
      "learning_rate": 8.969731378191186e-06,
      "loss": 0.4609,
      "step": 14635
    },
    {
      "epoch": 0.6716534349043183,
      "grad_norm": 0.4398956894874573,
      "learning_rate": 8.969582302132521e-06,
      "loss": 0.3247,
      "step": 14636
    },
    {
      "epoch": 0.6716993254095728,
      "grad_norm": 0.47701141238212585,
      "learning_rate": 8.96943321652826e-06,
      "loss": 0.3832,
      "step": 14637
    },
    {
      "epoch": 0.6717452159148273,
      "grad_norm": 0.4711799919605255,
      "learning_rate": 8.969284121378768e-06,
      "loss": 0.3713,
      "step": 14638
    },
    {
      "epoch": 0.6717911064200817,
      "grad_norm": 0.4529390335083008,
      "learning_rate": 8.969135016684399e-06,
      "loss": 0.3703,
      "step": 14639
    },
    {
      "epoch": 0.6718369969253362,
      "grad_norm": 0.4878658652305603,
      "learning_rate": 8.968985902445513e-06,
      "loss": 0.4743,
      "step": 14640
    },
    {
      "epoch": 0.6718828874305907,
      "grad_norm": 0.533019483089447,
      "learning_rate": 8.968836778662468e-06,
      "loss": 0.4382,
      "step": 14641
    },
    {
      "epoch": 0.671928777935845,
      "grad_norm": 0.4211060404777527,
      "learning_rate": 8.968687645335624e-06,
      "loss": 0.3163,
      "step": 14642
    },
    {
      "epoch": 0.6719746684410995,
      "grad_norm": 0.46758297085762024,
      "learning_rate": 8.96853850246534e-06,
      "loss": 0.4251,
      "step": 14643
    },
    {
      "epoch": 0.672020558946354,
      "grad_norm": 0.42853522300720215,
      "learning_rate": 8.968389350051971e-06,
      "loss": 0.3129,
      "step": 14644
    },
    {
      "epoch": 0.6720664494516084,
      "grad_norm": 0.45601895451545715,
      "learning_rate": 8.96824018809588e-06,
      "loss": 0.3979,
      "step": 14645
    },
    {
      "epoch": 0.6721123399568629,
      "grad_norm": 0.5159990787506104,
      "learning_rate": 8.968091016597423e-06,
      "loss": 0.4433,
      "step": 14646
    },
    {
      "epoch": 0.6721582304621174,
      "grad_norm": 0.46587973833084106,
      "learning_rate": 8.967941835556958e-06,
      "loss": 0.3804,
      "step": 14647
    },
    {
      "epoch": 0.6722041209673718,
      "grad_norm": 0.44248902797698975,
      "learning_rate": 8.967792644974846e-06,
      "loss": 0.3919,
      "step": 14648
    },
    {
      "epoch": 0.6722500114726263,
      "grad_norm": 0.44894465804100037,
      "learning_rate": 8.967643444851446e-06,
      "loss": 0.3343,
      "step": 14649
    },
    {
      "epoch": 0.6722959019778808,
      "grad_norm": 0.47158700227737427,
      "learning_rate": 8.967494235187116e-06,
      "loss": 0.4089,
      "step": 14650
    },
    {
      "epoch": 0.6723417924831352,
      "grad_norm": 0.5258386135101318,
      "learning_rate": 8.967345015982213e-06,
      "loss": 0.5154,
      "step": 14651
    },
    {
      "epoch": 0.6723876829883897,
      "grad_norm": 0.4748987555503845,
      "learning_rate": 8.9671957872371e-06,
      "loss": 0.4823,
      "step": 14652
    },
    {
      "epoch": 0.6724335734936442,
      "grad_norm": 0.5245237350463867,
      "learning_rate": 8.967046548952129e-06,
      "loss": 0.4588,
      "step": 14653
    },
    {
      "epoch": 0.6724794639988986,
      "grad_norm": 0.46021100878715515,
      "learning_rate": 8.966897301127666e-06,
      "loss": 0.3796,
      "step": 14654
    },
    {
      "epoch": 0.6725253545041531,
      "grad_norm": 0.43166399002075195,
      "learning_rate": 8.966748043764068e-06,
      "loss": 0.3554,
      "step": 14655
    },
    {
      "epoch": 0.6725712450094076,
      "grad_norm": 0.471636563539505,
      "learning_rate": 8.96659877686169e-06,
      "loss": 0.4221,
      "step": 14656
    },
    {
      "epoch": 0.6726171355146621,
      "grad_norm": 0.4382438659667969,
      "learning_rate": 8.966449500420896e-06,
      "loss": 0.3705,
      "step": 14657
    },
    {
      "epoch": 0.6726630260199165,
      "grad_norm": 0.46581512689590454,
      "learning_rate": 8.966300214442041e-06,
      "loss": 0.405,
      "step": 14658
    },
    {
      "epoch": 0.672708916525171,
      "grad_norm": 0.43589815497398376,
      "learning_rate": 8.966150918925488e-06,
      "loss": 0.3347,
      "step": 14659
    },
    {
      "epoch": 0.6727548070304255,
      "grad_norm": 0.4498421549797058,
      "learning_rate": 8.966001613871594e-06,
      "loss": 0.3869,
      "step": 14660
    },
    {
      "epoch": 0.6728006975356798,
      "grad_norm": 0.47100743651390076,
      "learning_rate": 8.965852299280715e-06,
      "loss": 0.4378,
      "step": 14661
    },
    {
      "epoch": 0.6728465880409343,
      "grad_norm": 0.4656829833984375,
      "learning_rate": 8.965702975153215e-06,
      "loss": 0.379,
      "step": 14662
    },
    {
      "epoch": 0.6728924785461888,
      "grad_norm": 0.4608096182346344,
      "learning_rate": 8.96555364148945e-06,
      "loss": 0.4239,
      "step": 14663
    },
    {
      "epoch": 0.6729383690514432,
      "grad_norm": 0.4436311721801758,
      "learning_rate": 8.96540429828978e-06,
      "loss": 0.3442,
      "step": 14664
    },
    {
      "epoch": 0.6729842595566977,
      "grad_norm": 0.4427100419998169,
      "learning_rate": 8.965254945554564e-06,
      "loss": 0.3455,
      "step": 14665
    },
    {
      "epoch": 0.6730301500619522,
      "grad_norm": 0.5839512348175049,
      "learning_rate": 8.965105583284162e-06,
      "loss": 0.5596,
      "step": 14666
    },
    {
      "epoch": 0.6730760405672066,
      "grad_norm": 0.4420351982116699,
      "learning_rate": 8.964956211478932e-06,
      "loss": 0.3753,
      "step": 14667
    },
    {
      "epoch": 0.6731219310724611,
      "grad_norm": 0.4297736585140228,
      "learning_rate": 8.964806830139235e-06,
      "loss": 0.3771,
      "step": 14668
    },
    {
      "epoch": 0.6731678215777156,
      "grad_norm": 0.42432901263237,
      "learning_rate": 8.964657439265428e-06,
      "loss": 0.3356,
      "step": 14669
    },
    {
      "epoch": 0.67321371208297,
      "grad_norm": 0.42087340354919434,
      "learning_rate": 8.96450803885787e-06,
      "loss": 0.3067,
      "step": 14670
    },
    {
      "epoch": 0.6732596025882245,
      "grad_norm": 0.4212096333503723,
      "learning_rate": 8.964358628916922e-06,
      "loss": 0.316,
      "step": 14671
    },
    {
      "epoch": 0.673305493093479,
      "grad_norm": 0.4495421051979065,
      "learning_rate": 8.964209209442944e-06,
      "loss": 0.3763,
      "step": 14672
    },
    {
      "epoch": 0.6733513835987335,
      "grad_norm": 0.47295546531677246,
      "learning_rate": 8.964059780436294e-06,
      "loss": 0.4679,
      "step": 14673
    },
    {
      "epoch": 0.6733972741039879,
      "grad_norm": 0.4856405556201935,
      "learning_rate": 8.96391034189733e-06,
      "loss": 0.4599,
      "step": 14674
    },
    {
      "epoch": 0.6734431646092424,
      "grad_norm": 0.46058133244514465,
      "learning_rate": 8.963760893826413e-06,
      "loss": 0.385,
      "step": 14675
    },
    {
      "epoch": 0.6734890551144969,
      "grad_norm": 0.43373391032218933,
      "learning_rate": 8.963611436223903e-06,
      "loss": 0.3625,
      "step": 14676
    },
    {
      "epoch": 0.6735349456197512,
      "grad_norm": 0.487519234418869,
      "learning_rate": 8.963461969090158e-06,
      "loss": 0.413,
      "step": 14677
    },
    {
      "epoch": 0.6735808361250057,
      "grad_norm": 0.5002069473266602,
      "learning_rate": 8.963312492425535e-06,
      "loss": 0.4941,
      "step": 14678
    },
    {
      "epoch": 0.6736267266302602,
      "grad_norm": 0.4721301794052124,
      "learning_rate": 8.9631630062304e-06,
      "loss": 0.4304,
      "step": 14679
    },
    {
      "epoch": 0.6736726171355146,
      "grad_norm": 0.5168606638908386,
      "learning_rate": 8.963013510505108e-06,
      "loss": 0.4571,
      "step": 14680
    },
    {
      "epoch": 0.6737185076407691,
      "grad_norm": 0.46825718879699707,
      "learning_rate": 8.962864005250019e-06,
      "loss": 0.4,
      "step": 14681
    },
    {
      "epoch": 0.6737643981460236,
      "grad_norm": 0.46966734528541565,
      "learning_rate": 8.962714490465492e-06,
      "loss": 0.4101,
      "step": 14682
    },
    {
      "epoch": 0.673810288651278,
      "grad_norm": 0.4863331913948059,
      "learning_rate": 8.962564966151888e-06,
      "loss": 0.3914,
      "step": 14683
    },
    {
      "epoch": 0.6738561791565325,
      "grad_norm": 0.4328271746635437,
      "learning_rate": 8.962415432309565e-06,
      "loss": 0.3375,
      "step": 14684
    },
    {
      "epoch": 0.673902069661787,
      "grad_norm": 0.4547343850135803,
      "learning_rate": 8.962265888938886e-06,
      "loss": 0.3539,
      "step": 14685
    },
    {
      "epoch": 0.6739479601670414,
      "grad_norm": 0.47346803545951843,
      "learning_rate": 8.962116336040205e-06,
      "loss": 0.3967,
      "step": 14686
    },
    {
      "epoch": 0.6739938506722959,
      "grad_norm": 0.42227375507354736,
      "learning_rate": 8.961966773613887e-06,
      "loss": 0.3423,
      "step": 14687
    },
    {
      "epoch": 0.6740397411775504,
      "grad_norm": 0.4545140564441681,
      "learning_rate": 8.961817201660287e-06,
      "loss": 0.383,
      "step": 14688
    },
    {
      "epoch": 0.6740856316828048,
      "grad_norm": 0.5202023386955261,
      "learning_rate": 8.961667620179769e-06,
      "loss": 0.5039,
      "step": 14689
    },
    {
      "epoch": 0.6741315221880593,
      "grad_norm": 0.4110974371433258,
      "learning_rate": 8.96151802917269e-06,
      "loss": 0.3307,
      "step": 14690
    },
    {
      "epoch": 0.6741774126933138,
      "grad_norm": 0.4705328047275543,
      "learning_rate": 8.961368428639411e-06,
      "loss": 0.4119,
      "step": 14691
    },
    {
      "epoch": 0.6742233031985683,
      "grad_norm": 0.6943110227584839,
      "learning_rate": 8.961218818580291e-06,
      "loss": 0.4051,
      "step": 14692
    },
    {
      "epoch": 0.6742691937038227,
      "grad_norm": 0.4538700580596924,
      "learning_rate": 8.961069198995688e-06,
      "loss": 0.3729,
      "step": 14693
    },
    {
      "epoch": 0.6743150842090772,
      "grad_norm": 0.4574923515319824,
      "learning_rate": 8.960919569885967e-06,
      "loss": 0.3912,
      "step": 14694
    },
    {
      "epoch": 0.6743609747143317,
      "grad_norm": 0.44742801785469055,
      "learning_rate": 8.960769931251483e-06,
      "loss": 0.4014,
      "step": 14695
    },
    {
      "epoch": 0.674406865219586,
      "grad_norm": 0.42899659276008606,
      "learning_rate": 8.960620283092597e-06,
      "loss": 0.3326,
      "step": 14696
    },
    {
      "epoch": 0.6744527557248405,
      "grad_norm": 0.4652882516384125,
      "learning_rate": 8.96047062540967e-06,
      "loss": 0.4231,
      "step": 14697
    },
    {
      "epoch": 0.674498646230095,
      "grad_norm": 0.4359254539012909,
      "learning_rate": 8.96032095820306e-06,
      "loss": 0.3176,
      "step": 14698
    },
    {
      "epoch": 0.6745445367353494,
      "grad_norm": 0.4278014600276947,
      "learning_rate": 8.96017128147313e-06,
      "loss": 0.3306,
      "step": 14699
    },
    {
      "epoch": 0.6745904272406039,
      "grad_norm": 0.4722307622432709,
      "learning_rate": 8.960021595220236e-06,
      "loss": 0.4185,
      "step": 14700
    },
    {
      "epoch": 0.6746363177458584,
      "grad_norm": 0.4911159574985504,
      "learning_rate": 8.95987189944474e-06,
      "loss": 0.4744,
      "step": 14701
    },
    {
      "epoch": 0.6746822082511128,
      "grad_norm": 0.4785071909427643,
      "learning_rate": 8.959722194147002e-06,
      "loss": 0.4276,
      "step": 14702
    },
    {
      "epoch": 0.6747280987563673,
      "grad_norm": 0.43101930618286133,
      "learning_rate": 8.959572479327383e-06,
      "loss": 0.3434,
      "step": 14703
    },
    {
      "epoch": 0.6747739892616218,
      "grad_norm": 0.4710412323474884,
      "learning_rate": 8.95942275498624e-06,
      "loss": 0.4143,
      "step": 14704
    },
    {
      "epoch": 0.6748198797668762,
      "grad_norm": 0.45896440744400024,
      "learning_rate": 8.959273021123936e-06,
      "loss": 0.4157,
      "step": 14705
    },
    {
      "epoch": 0.6748657702721307,
      "grad_norm": 0.4247373044490814,
      "learning_rate": 8.959123277740831e-06,
      "loss": 0.3343,
      "step": 14706
    },
    {
      "epoch": 0.6749116607773852,
      "grad_norm": 0.4538171887397766,
      "learning_rate": 8.958973524837282e-06,
      "loss": 0.4136,
      "step": 14707
    },
    {
      "epoch": 0.6749575512826396,
      "grad_norm": 0.46192261576652527,
      "learning_rate": 8.958823762413652e-06,
      "loss": 0.3514,
      "step": 14708
    },
    {
      "epoch": 0.6750034417878941,
      "grad_norm": 0.4493251442909241,
      "learning_rate": 8.958673990470302e-06,
      "loss": 0.3473,
      "step": 14709
    },
    {
      "epoch": 0.6750493322931486,
      "grad_norm": 0.441125750541687,
      "learning_rate": 8.958524209007587e-06,
      "loss": 0.3468,
      "step": 14710
    },
    {
      "epoch": 0.6750952227984031,
      "grad_norm": 0.43881431221961975,
      "learning_rate": 8.958374418025873e-06,
      "loss": 0.3499,
      "step": 14711
    },
    {
      "epoch": 0.6751411133036574,
      "grad_norm": 0.42891258001327515,
      "learning_rate": 8.958224617525517e-06,
      "loss": 0.3406,
      "step": 14712
    },
    {
      "epoch": 0.675187003808912,
      "grad_norm": 0.48588812351226807,
      "learning_rate": 8.95807480750688e-06,
      "loss": 0.4883,
      "step": 14713
    },
    {
      "epoch": 0.6752328943141664,
      "grad_norm": 0.525397777557373,
      "learning_rate": 8.957924987970324e-06,
      "loss": 0.4651,
      "step": 14714
    },
    {
      "epoch": 0.6752787848194208,
      "grad_norm": 0.4215129613876343,
      "learning_rate": 8.957775158916206e-06,
      "loss": 0.3095,
      "step": 14715
    },
    {
      "epoch": 0.6753246753246753,
      "grad_norm": 0.45480912923812866,
      "learning_rate": 8.957625320344886e-06,
      "loss": 0.416,
      "step": 14716
    },
    {
      "epoch": 0.6753705658299298,
      "grad_norm": 0.43501749634742737,
      "learning_rate": 8.95747547225673e-06,
      "loss": 0.3321,
      "step": 14717
    },
    {
      "epoch": 0.6754164563351842,
      "grad_norm": 0.43271464109420776,
      "learning_rate": 8.957325614652092e-06,
      "loss": 0.2907,
      "step": 14718
    },
    {
      "epoch": 0.6754623468404387,
      "grad_norm": 0.46629834175109863,
      "learning_rate": 8.957175747531336e-06,
      "loss": 0.3797,
      "step": 14719
    },
    {
      "epoch": 0.6755082373456932,
      "grad_norm": 0.4651343524456024,
      "learning_rate": 8.957025870894822e-06,
      "loss": 0.4487,
      "step": 14720
    },
    {
      "epoch": 0.6755541278509476,
      "grad_norm": 0.48341336846351624,
      "learning_rate": 8.956875984742907e-06,
      "loss": 0.4806,
      "step": 14721
    },
    {
      "epoch": 0.6756000183562021,
      "grad_norm": 0.726714551448822,
      "learning_rate": 8.956726089075956e-06,
      "loss": 0.5004,
      "step": 14722
    },
    {
      "epoch": 0.6756459088614566,
      "grad_norm": 0.46652036905288696,
      "learning_rate": 8.956576183894328e-06,
      "loss": 0.4476,
      "step": 14723
    },
    {
      "epoch": 0.675691799366711,
      "grad_norm": 0.42625826597213745,
      "learning_rate": 8.956426269198381e-06,
      "loss": 0.3535,
      "step": 14724
    },
    {
      "epoch": 0.6757376898719655,
      "grad_norm": 0.4924218952655792,
      "learning_rate": 8.95627634498848e-06,
      "loss": 0.484,
      "step": 14725
    },
    {
      "epoch": 0.67578358037722,
      "grad_norm": 0.40904316306114197,
      "learning_rate": 8.95612641126498e-06,
      "loss": 0.3107,
      "step": 14726
    },
    {
      "epoch": 0.6758294708824745,
      "grad_norm": 0.4393841624259949,
      "learning_rate": 8.955976468028246e-06,
      "loss": 0.373,
      "step": 14727
    },
    {
      "epoch": 0.6758753613877289,
      "grad_norm": 0.46549513936042786,
      "learning_rate": 8.955826515278639e-06,
      "loss": 0.4033,
      "step": 14728
    },
    {
      "epoch": 0.6759212518929834,
      "grad_norm": 0.43072983622550964,
      "learning_rate": 8.955676553016514e-06,
      "loss": 0.3792,
      "step": 14729
    },
    {
      "epoch": 0.6759671423982379,
      "grad_norm": 0.5095604658126831,
      "learning_rate": 8.955526581242239e-06,
      "loss": 0.4312,
      "step": 14730
    },
    {
      "epoch": 0.6760130329034922,
      "grad_norm": 0.5377375483512878,
      "learning_rate": 8.955376599956169e-06,
      "loss": 0.5254,
      "step": 14731
    },
    {
      "epoch": 0.6760589234087467,
      "grad_norm": 0.48529115319252014,
      "learning_rate": 8.955226609158667e-06,
      "loss": 0.4435,
      "step": 14732
    },
    {
      "epoch": 0.6761048139140012,
      "grad_norm": 0.4673723578453064,
      "learning_rate": 8.955076608850092e-06,
      "loss": 0.418,
      "step": 14733
    },
    {
      "epoch": 0.6761507044192556,
      "grad_norm": 0.46938684582710266,
      "learning_rate": 8.954926599030808e-06,
      "loss": 0.4093,
      "step": 14734
    },
    {
      "epoch": 0.6761965949245101,
      "grad_norm": 0.45263391733169556,
      "learning_rate": 8.954776579701174e-06,
      "loss": 0.3938,
      "step": 14735
    },
    {
      "epoch": 0.6762424854297646,
      "grad_norm": 0.49526724219322205,
      "learning_rate": 8.95462655086155e-06,
      "loss": 0.3636,
      "step": 14736
    },
    {
      "epoch": 0.676288375935019,
      "grad_norm": 0.5028539299964905,
      "learning_rate": 8.954476512512296e-06,
      "loss": 0.2912,
      "step": 14737
    },
    {
      "epoch": 0.6763342664402735,
      "grad_norm": 0.4028368294239044,
      "learning_rate": 8.954326464653775e-06,
      "loss": 0.2976,
      "step": 14738
    },
    {
      "epoch": 0.676380156945528,
      "grad_norm": 0.4950895309448242,
      "learning_rate": 8.954176407286347e-06,
      "loss": 0.451,
      "step": 14739
    },
    {
      "epoch": 0.6764260474507824,
      "grad_norm": 0.46924006938934326,
      "learning_rate": 8.954026340410372e-06,
      "loss": 0.4481,
      "step": 14740
    },
    {
      "epoch": 0.6764719379560369,
      "grad_norm": 0.4596254229545593,
      "learning_rate": 8.953876264026213e-06,
      "loss": 0.3429,
      "step": 14741
    },
    {
      "epoch": 0.6765178284612914,
      "grad_norm": 0.4995757043361664,
      "learning_rate": 8.953726178134227e-06,
      "loss": 0.4562,
      "step": 14742
    },
    {
      "epoch": 0.6765637189665458,
      "grad_norm": 0.45612114667892456,
      "learning_rate": 8.95357608273478e-06,
      "loss": 0.4297,
      "step": 14743
    },
    {
      "epoch": 0.6766096094718003,
      "grad_norm": 0.6454023718833923,
      "learning_rate": 8.953425977828229e-06,
      "loss": 0.5747,
      "step": 14744
    },
    {
      "epoch": 0.6766554999770548,
      "grad_norm": 0.5305061936378479,
      "learning_rate": 8.953275863414937e-06,
      "loss": 0.5749,
      "step": 14745
    },
    {
      "epoch": 0.6767013904823093,
      "grad_norm": 0.4578821659088135,
      "learning_rate": 8.953125739495263e-06,
      "loss": 0.3847,
      "step": 14746
    },
    {
      "epoch": 0.6767472809875636,
      "grad_norm": 0.4212392270565033,
      "learning_rate": 8.952975606069569e-06,
      "loss": 0.3283,
      "step": 14747
    },
    {
      "epoch": 0.6767931714928181,
      "grad_norm": 0.47139298915863037,
      "learning_rate": 8.952825463138217e-06,
      "loss": 0.3851,
      "step": 14748
    },
    {
      "epoch": 0.6768390619980726,
      "grad_norm": 0.42642828822135925,
      "learning_rate": 8.952675310701568e-06,
      "loss": 0.3367,
      "step": 14749
    },
    {
      "epoch": 0.676884952503327,
      "grad_norm": 0.3850325345993042,
      "learning_rate": 8.952525148759983e-06,
      "loss": 0.2918,
      "step": 14750
    },
    {
      "epoch": 0.6769308430085815,
      "grad_norm": 0.4391419589519501,
      "learning_rate": 8.95237497731382e-06,
      "loss": 0.332,
      "step": 14751
    },
    {
      "epoch": 0.676976733513836,
      "grad_norm": 0.4256323575973511,
      "learning_rate": 8.952224796363444e-06,
      "loss": 0.3299,
      "step": 14752
    },
    {
      "epoch": 0.6770226240190904,
      "grad_norm": 0.39353954792022705,
      "learning_rate": 8.952074605909216e-06,
      "loss": 0.2775,
      "step": 14753
    },
    {
      "epoch": 0.6770685145243449,
      "grad_norm": 0.44493716955184937,
      "learning_rate": 8.951924405951494e-06,
      "loss": 0.4082,
      "step": 14754
    },
    {
      "epoch": 0.6771144050295994,
      "grad_norm": 0.4622228741645813,
      "learning_rate": 8.951774196490641e-06,
      "loss": 0.3933,
      "step": 14755
    },
    {
      "epoch": 0.6771602955348538,
      "grad_norm": 0.4285729229450226,
      "learning_rate": 8.95162397752702e-06,
      "loss": 0.3694,
      "step": 14756
    },
    {
      "epoch": 0.6772061860401083,
      "grad_norm": 0.4257205128669739,
      "learning_rate": 8.951473749060988e-06,
      "loss": 0.3408,
      "step": 14757
    },
    {
      "epoch": 0.6772520765453628,
      "grad_norm": 0.4245329797267914,
      "learning_rate": 8.95132351109291e-06,
      "loss": 0.3418,
      "step": 14758
    },
    {
      "epoch": 0.6772979670506172,
      "grad_norm": 0.413326621055603,
      "learning_rate": 8.951173263623148e-06,
      "loss": 0.3548,
      "step": 14759
    },
    {
      "epoch": 0.6773438575558717,
      "grad_norm": 0.4849330186843872,
      "learning_rate": 8.951023006652059e-06,
      "loss": 0.4878,
      "step": 14760
    },
    {
      "epoch": 0.6773897480611262,
      "grad_norm": 0.48579591512680054,
      "learning_rate": 8.950872740180006e-06,
      "loss": 0.4245,
      "step": 14761
    },
    {
      "epoch": 0.6774356385663807,
      "grad_norm": 0.43916749954223633,
      "learning_rate": 8.950722464207353e-06,
      "loss": 0.3923,
      "step": 14762
    },
    {
      "epoch": 0.6774815290716351,
      "grad_norm": 0.48171040415763855,
      "learning_rate": 8.950572178734457e-06,
      "loss": 0.3637,
      "step": 14763
    },
    {
      "epoch": 0.6775274195768896,
      "grad_norm": 0.461260050535202,
      "learning_rate": 8.950421883761684e-06,
      "loss": 0.4024,
      "step": 14764
    },
    {
      "epoch": 0.677573310082144,
      "grad_norm": 0.48985669016838074,
      "learning_rate": 8.950271579289394e-06,
      "loss": 0.4547,
      "step": 14765
    },
    {
      "epoch": 0.6776192005873984,
      "grad_norm": 0.4428180158138275,
      "learning_rate": 8.950121265317946e-06,
      "loss": 0.3599,
      "step": 14766
    },
    {
      "epoch": 0.6776650910926529,
      "grad_norm": 0.49926432967185974,
      "learning_rate": 8.949970941847701e-06,
      "loss": 0.5125,
      "step": 14767
    },
    {
      "epoch": 0.6777109815979074,
      "grad_norm": 0.4786827564239502,
      "learning_rate": 8.949820608879026e-06,
      "loss": 0.4975,
      "step": 14768
    },
    {
      "epoch": 0.6777568721031618,
      "grad_norm": 0.47190189361572266,
      "learning_rate": 8.949670266412278e-06,
      "loss": 0.4733,
      "step": 14769
    },
    {
      "epoch": 0.6778027626084163,
      "grad_norm": 0.46229228377342224,
      "learning_rate": 8.949519914447818e-06,
      "loss": 0.4407,
      "step": 14770
    },
    {
      "epoch": 0.6778486531136708,
      "grad_norm": 0.4745892882347107,
      "learning_rate": 8.949369552986011e-06,
      "loss": 0.4293,
      "step": 14771
    },
    {
      "epoch": 0.6778945436189252,
      "grad_norm": 0.4559294581413269,
      "learning_rate": 8.949219182027215e-06,
      "loss": 0.3471,
      "step": 14772
    },
    {
      "epoch": 0.6779404341241797,
      "grad_norm": 0.5512930154800415,
      "learning_rate": 8.949068801571794e-06,
      "loss": 0.332,
      "step": 14773
    },
    {
      "epoch": 0.6779863246294342,
      "grad_norm": 0.4335198998451233,
      "learning_rate": 8.94891841162011e-06,
      "loss": 0.3852,
      "step": 14774
    },
    {
      "epoch": 0.6780322151346886,
      "grad_norm": 0.4324401319026947,
      "learning_rate": 8.948768012172523e-06,
      "loss": 0.3789,
      "step": 14775
    },
    {
      "epoch": 0.6780781056399431,
      "grad_norm": 0.4959743916988373,
      "learning_rate": 8.948617603229395e-06,
      "loss": 0.4826,
      "step": 14776
    },
    {
      "epoch": 0.6781239961451976,
      "grad_norm": 0.45083341002464294,
      "learning_rate": 8.948467184791087e-06,
      "loss": 0.4073,
      "step": 14777
    },
    {
      "epoch": 0.678169886650452,
      "grad_norm": 0.4678332805633545,
      "learning_rate": 8.948316756857963e-06,
      "loss": 0.3927,
      "step": 14778
    },
    {
      "epoch": 0.6782157771557065,
      "grad_norm": 0.4929785430431366,
      "learning_rate": 8.948166319430382e-06,
      "loss": 0.439,
      "step": 14779
    },
    {
      "epoch": 0.678261667660961,
      "grad_norm": 0.40612107515335083,
      "learning_rate": 8.948015872508709e-06,
      "loss": 0.329,
      "step": 14780
    },
    {
      "epoch": 0.6783075581662155,
      "grad_norm": 0.48002389073371887,
      "learning_rate": 8.947865416093303e-06,
      "loss": 0.4819,
      "step": 14781
    },
    {
      "epoch": 0.6783534486714698,
      "grad_norm": 0.44353726506233215,
      "learning_rate": 8.947714950184525e-06,
      "loss": 0.3833,
      "step": 14782
    },
    {
      "epoch": 0.6783993391767243,
      "grad_norm": 0.4811346232891083,
      "learning_rate": 8.94756447478274e-06,
      "loss": 0.3768,
      "step": 14783
    },
    {
      "epoch": 0.6784452296819788,
      "grad_norm": 0.4238256812095642,
      "learning_rate": 8.947413989888307e-06,
      "loss": 0.3652,
      "step": 14784
    },
    {
      "epoch": 0.6784911201872332,
      "grad_norm": 0.4736774265766144,
      "learning_rate": 8.947263495501591e-06,
      "loss": 0.3939,
      "step": 14785
    },
    {
      "epoch": 0.6785370106924877,
      "grad_norm": 0.4651627838611603,
      "learning_rate": 8.947112991622951e-06,
      "loss": 0.3982,
      "step": 14786
    },
    {
      "epoch": 0.6785829011977422,
      "grad_norm": 0.4665450155735016,
      "learning_rate": 8.946962478252751e-06,
      "loss": 0.4067,
      "step": 14787
    },
    {
      "epoch": 0.6786287917029966,
      "grad_norm": 0.44031134247779846,
      "learning_rate": 8.94681195539135e-06,
      "loss": 0.3311,
      "step": 14788
    },
    {
      "epoch": 0.6786746822082511,
      "grad_norm": 0.4378812611103058,
      "learning_rate": 8.946661423039113e-06,
      "loss": 0.3972,
      "step": 14789
    },
    {
      "epoch": 0.6787205727135056,
      "grad_norm": 0.44837936758995056,
      "learning_rate": 8.946510881196402e-06,
      "loss": 0.3429,
      "step": 14790
    },
    {
      "epoch": 0.67876646321876,
      "grad_norm": 0.4941911995410919,
      "learning_rate": 8.946360329863576e-06,
      "loss": 0.4892,
      "step": 14791
    },
    {
      "epoch": 0.6788123537240145,
      "grad_norm": 0.4322866201400757,
      "learning_rate": 8.946209769041002e-06,
      "loss": 0.3849,
      "step": 14792
    },
    {
      "epoch": 0.678858244229269,
      "grad_norm": 0.44300970435142517,
      "learning_rate": 8.946059198729036e-06,
      "loss": 0.3974,
      "step": 14793
    },
    {
      "epoch": 0.6789041347345234,
      "grad_norm": 0.44444549083709717,
      "learning_rate": 8.945908618928044e-06,
      "loss": 0.374,
      "step": 14794
    },
    {
      "epoch": 0.6789500252397779,
      "grad_norm": 0.45047274231910706,
      "learning_rate": 8.945758029638388e-06,
      "loss": 0.3879,
      "step": 14795
    },
    {
      "epoch": 0.6789959157450324,
      "grad_norm": 0.4511168301105499,
      "learning_rate": 8.945607430860428e-06,
      "loss": 0.3884,
      "step": 14796
    },
    {
      "epoch": 0.6790418062502868,
      "grad_norm": 0.4733699858188629,
      "learning_rate": 8.945456822594528e-06,
      "loss": 0.4326,
      "step": 14797
    },
    {
      "epoch": 0.6790876967555413,
      "grad_norm": 0.45085573196411133,
      "learning_rate": 8.945306204841048e-06,
      "loss": 0.3735,
      "step": 14798
    },
    {
      "epoch": 0.6791335872607958,
      "grad_norm": 0.4270944595336914,
      "learning_rate": 8.945155577600353e-06,
      "loss": 0.3446,
      "step": 14799
    },
    {
      "epoch": 0.6791794777660503,
      "grad_norm": 0.4654461741447449,
      "learning_rate": 8.945004940872803e-06,
      "loss": 0.4252,
      "step": 14800
    },
    {
      "epoch": 0.6792253682713046,
      "grad_norm": 0.479457288980484,
      "learning_rate": 8.944854294658763e-06,
      "loss": 0.4698,
      "step": 14801
    },
    {
      "epoch": 0.6792712587765591,
      "grad_norm": 0.4618765413761139,
      "learning_rate": 8.944703638958591e-06,
      "loss": 0.392,
      "step": 14802
    },
    {
      "epoch": 0.6793171492818136,
      "grad_norm": 0.45287376642227173,
      "learning_rate": 8.944552973772654e-06,
      "loss": 0.4296,
      "step": 14803
    },
    {
      "epoch": 0.679363039787068,
      "grad_norm": 0.47194233536720276,
      "learning_rate": 8.944402299101312e-06,
      "loss": 0.4388,
      "step": 14804
    },
    {
      "epoch": 0.6794089302923225,
      "grad_norm": 0.44376716017723083,
      "learning_rate": 8.944251614944926e-06,
      "loss": 0.3598,
      "step": 14805
    },
    {
      "epoch": 0.679454820797577,
      "grad_norm": 0.45693904161453247,
      "learning_rate": 8.944100921303862e-06,
      "loss": 0.3911,
      "step": 14806
    },
    {
      "epoch": 0.6795007113028314,
      "grad_norm": 0.4648887813091278,
      "learning_rate": 8.943950218178478e-06,
      "loss": 0.435,
      "step": 14807
    },
    {
      "epoch": 0.6795466018080859,
      "grad_norm": 0.46242383122444153,
      "learning_rate": 8.94379950556914e-06,
      "loss": 0.409,
      "step": 14808
    },
    {
      "epoch": 0.6795924923133404,
      "grad_norm": 0.46837151050567627,
      "learning_rate": 8.943648783476206e-06,
      "loss": 0.395,
      "step": 14809
    },
    {
      "epoch": 0.6796383828185948,
      "grad_norm": 0.4884977340698242,
      "learning_rate": 8.943498051900044e-06,
      "loss": 0.4566,
      "step": 14810
    },
    {
      "epoch": 0.6796842733238493,
      "grad_norm": 0.431994765996933,
      "learning_rate": 8.943347310841013e-06,
      "loss": 0.3216,
      "step": 14811
    },
    {
      "epoch": 0.6797301638291038,
      "grad_norm": 0.45315325260162354,
      "learning_rate": 8.943196560299476e-06,
      "loss": 0.3843,
      "step": 14812
    },
    {
      "epoch": 0.6797760543343582,
      "grad_norm": 0.4566023647785187,
      "learning_rate": 8.943045800275795e-06,
      "loss": 0.3861,
      "step": 14813
    },
    {
      "epoch": 0.6798219448396127,
      "grad_norm": 0.44639286398887634,
      "learning_rate": 8.942895030770335e-06,
      "loss": 0.4079,
      "step": 14814
    },
    {
      "epoch": 0.6798678353448672,
      "grad_norm": 0.4505566954612732,
      "learning_rate": 8.942744251783456e-06,
      "loss": 0.367,
      "step": 14815
    },
    {
      "epoch": 0.6799137258501217,
      "grad_norm": 0.4297548830509186,
      "learning_rate": 8.942593463315522e-06,
      "loss": 0.3101,
      "step": 14816
    },
    {
      "epoch": 0.679959616355376,
      "grad_norm": 0.44771164655685425,
      "learning_rate": 8.942442665366895e-06,
      "loss": 0.4195,
      "step": 14817
    },
    {
      "epoch": 0.6800055068606305,
      "grad_norm": 0.48178502917289734,
      "learning_rate": 8.942291857937937e-06,
      "loss": 0.4598,
      "step": 14818
    },
    {
      "epoch": 0.680051397365885,
      "grad_norm": 0.468656450510025,
      "learning_rate": 8.942141041029011e-06,
      "loss": 0.3862,
      "step": 14819
    },
    {
      "epoch": 0.6800972878711394,
      "grad_norm": 0.47881656885147095,
      "learning_rate": 8.941990214640481e-06,
      "loss": 0.4548,
      "step": 14820
    },
    {
      "epoch": 0.6801431783763939,
      "grad_norm": 0.5204012393951416,
      "learning_rate": 8.941839378772708e-06,
      "loss": 0.4471,
      "step": 14821
    },
    {
      "epoch": 0.6801890688816484,
      "grad_norm": 0.43512874841690063,
      "learning_rate": 8.941688533426056e-06,
      "loss": 0.3805,
      "step": 14822
    },
    {
      "epoch": 0.6802349593869028,
      "grad_norm": 0.44788533449172974,
      "learning_rate": 8.941537678600886e-06,
      "loss": 0.3776,
      "step": 14823
    },
    {
      "epoch": 0.6802808498921573,
      "grad_norm": 0.4891255497932434,
      "learning_rate": 8.941386814297564e-06,
      "loss": 0.3898,
      "step": 14824
    },
    {
      "epoch": 0.6803267403974118,
      "grad_norm": 0.44314494729042053,
      "learning_rate": 8.941235940516449e-06,
      "loss": 0.344,
      "step": 14825
    },
    {
      "epoch": 0.6803726309026662,
      "grad_norm": 0.4424200654029846,
      "learning_rate": 8.941085057257905e-06,
      "loss": 0.3804,
      "step": 14826
    },
    {
      "epoch": 0.6804185214079207,
      "grad_norm": 0.4754483997821808,
      "learning_rate": 8.940934164522297e-06,
      "loss": 0.4385,
      "step": 14827
    },
    {
      "epoch": 0.6804644119131752,
      "grad_norm": 0.47143200039863586,
      "learning_rate": 8.940783262309984e-06,
      "loss": 0.4318,
      "step": 14828
    },
    {
      "epoch": 0.6805103024184296,
      "grad_norm": 0.414588987827301,
      "learning_rate": 8.940632350621333e-06,
      "loss": 0.3723,
      "step": 14829
    },
    {
      "epoch": 0.6805561929236841,
      "grad_norm": 0.46653369069099426,
      "learning_rate": 8.940481429456704e-06,
      "loss": 0.4689,
      "step": 14830
    },
    {
      "epoch": 0.6806020834289386,
      "grad_norm": 0.4801202714443207,
      "learning_rate": 8.940330498816461e-06,
      "loss": 0.3994,
      "step": 14831
    },
    {
      "epoch": 0.680647973934193,
      "grad_norm": 0.45494574308395386,
      "learning_rate": 8.940179558700967e-06,
      "loss": 0.4564,
      "step": 14832
    },
    {
      "epoch": 0.6806938644394475,
      "grad_norm": 0.4337141811847687,
      "learning_rate": 8.940028609110585e-06,
      "loss": 0.3746,
      "step": 14833
    },
    {
      "epoch": 0.680739754944702,
      "grad_norm": 0.461557000875473,
      "learning_rate": 8.939877650045677e-06,
      "loss": 0.4024,
      "step": 14834
    },
    {
      "epoch": 0.6807856454499565,
      "grad_norm": 0.4299245774745941,
      "learning_rate": 8.939726681506607e-06,
      "loss": 0.3367,
      "step": 14835
    },
    {
      "epoch": 0.6808315359552108,
      "grad_norm": 0.4379175007343292,
      "learning_rate": 8.939575703493738e-06,
      "loss": 0.4301,
      "step": 14836
    },
    {
      "epoch": 0.6808774264604653,
      "grad_norm": 0.4237874448299408,
      "learning_rate": 8.93942471600743e-06,
      "loss": 0.3268,
      "step": 14837
    },
    {
      "epoch": 0.6809233169657198,
      "grad_norm": 0.460771769285202,
      "learning_rate": 8.939273719048052e-06,
      "loss": 0.3951,
      "step": 14838
    },
    {
      "epoch": 0.6809692074709742,
      "grad_norm": 0.4490431547164917,
      "learning_rate": 8.939122712615961e-06,
      "loss": 0.4312,
      "step": 14839
    },
    {
      "epoch": 0.6810150979762287,
      "grad_norm": 0.4115942716598511,
      "learning_rate": 8.938971696711527e-06,
      "loss": 0.3175,
      "step": 14840
    },
    {
      "epoch": 0.6810609884814832,
      "grad_norm": 0.4233342111110687,
      "learning_rate": 8.938820671335107e-06,
      "loss": 0.3126,
      "step": 14841
    },
    {
      "epoch": 0.6811068789867376,
      "grad_norm": 0.493661105632782,
      "learning_rate": 8.938669636487067e-06,
      "loss": 0.4859,
      "step": 14842
    },
    {
      "epoch": 0.6811527694919921,
      "grad_norm": 0.4322417378425598,
      "learning_rate": 8.938518592167769e-06,
      "loss": 0.38,
      "step": 14843
    },
    {
      "epoch": 0.6811986599972466,
      "grad_norm": 0.4188777506351471,
      "learning_rate": 8.938367538377577e-06,
      "loss": 0.3445,
      "step": 14844
    },
    {
      "epoch": 0.681244550502501,
      "grad_norm": 0.43958789110183716,
      "learning_rate": 8.938216475116856e-06,
      "loss": 0.3813,
      "step": 14845
    },
    {
      "epoch": 0.6812904410077555,
      "grad_norm": 0.4574650228023529,
      "learning_rate": 8.938065402385965e-06,
      "loss": 0.374,
      "step": 14846
    },
    {
      "epoch": 0.68133633151301,
      "grad_norm": 0.4355171322822571,
      "learning_rate": 8.93791432018527e-06,
      "loss": 0.3578,
      "step": 14847
    },
    {
      "epoch": 0.6813822220182644,
      "grad_norm": 0.45174604654312134,
      "learning_rate": 8.937763228515135e-06,
      "loss": 0.4306,
      "step": 14848
    },
    {
      "epoch": 0.6814281125235189,
      "grad_norm": 0.42888110876083374,
      "learning_rate": 8.93761212737592e-06,
      "loss": 0.3575,
      "step": 14849
    },
    {
      "epoch": 0.6814740030287734,
      "grad_norm": 0.4579628109931946,
      "learning_rate": 8.937461016767993e-06,
      "loss": 0.4177,
      "step": 14850
    },
    {
      "epoch": 0.6815198935340278,
      "grad_norm": 0.4721222519874573,
      "learning_rate": 8.937309896691714e-06,
      "loss": 0.3787,
      "step": 14851
    },
    {
      "epoch": 0.6815657840392823,
      "grad_norm": 0.4567517638206482,
      "learning_rate": 8.937158767147447e-06,
      "loss": 0.4229,
      "step": 14852
    },
    {
      "epoch": 0.6816116745445367,
      "grad_norm": 0.4372602105140686,
      "learning_rate": 8.937007628135556e-06,
      "loss": 0.4225,
      "step": 14853
    },
    {
      "epoch": 0.6816575650497912,
      "grad_norm": 0.47924935817718506,
      "learning_rate": 8.936856479656403e-06,
      "loss": 0.417,
      "step": 14854
    },
    {
      "epoch": 0.6817034555550456,
      "grad_norm": 0.4316025376319885,
      "learning_rate": 8.936705321710356e-06,
      "loss": 0.3549,
      "step": 14855
    },
    {
      "epoch": 0.6817493460603001,
      "grad_norm": 0.44067510962486267,
      "learning_rate": 8.936554154297772e-06,
      "loss": 0.4328,
      "step": 14856
    },
    {
      "epoch": 0.6817952365655546,
      "grad_norm": 0.41196808218955994,
      "learning_rate": 8.93640297741902e-06,
      "loss": 0.3613,
      "step": 14857
    },
    {
      "epoch": 0.681841127070809,
      "grad_norm": 0.4127485454082489,
      "learning_rate": 8.93625179107446e-06,
      "loss": 0.3671,
      "step": 14858
    },
    {
      "epoch": 0.6818870175760635,
      "grad_norm": 0.469026654958725,
      "learning_rate": 8.936100595264456e-06,
      "loss": 0.4294,
      "step": 14859
    },
    {
      "epoch": 0.681932908081318,
      "grad_norm": 0.46003252267837524,
      "learning_rate": 8.935949389989374e-06,
      "loss": 0.3821,
      "step": 14860
    },
    {
      "epoch": 0.6819787985865724,
      "grad_norm": 0.699417769908905,
      "learning_rate": 8.935798175249575e-06,
      "loss": 0.3418,
      "step": 14861
    },
    {
      "epoch": 0.6820246890918269,
      "grad_norm": 0.45537805557250977,
      "learning_rate": 8.935646951045422e-06,
      "loss": 0.4173,
      "step": 14862
    },
    {
      "epoch": 0.6820705795970814,
      "grad_norm": 0.4469729959964752,
      "learning_rate": 8.935495717377283e-06,
      "loss": 0.3634,
      "step": 14863
    },
    {
      "epoch": 0.6821164701023358,
      "grad_norm": 0.48699885606765747,
      "learning_rate": 8.935344474245517e-06,
      "loss": 0.4698,
      "step": 14864
    },
    {
      "epoch": 0.6821623606075903,
      "grad_norm": 0.45305201411247253,
      "learning_rate": 8.93519322165049e-06,
      "loss": 0.377,
      "step": 14865
    },
    {
      "epoch": 0.6822082511128448,
      "grad_norm": 0.40661904215812683,
      "learning_rate": 8.935041959592566e-06,
      "loss": 0.2866,
      "step": 14866
    },
    {
      "epoch": 0.6822541416180992,
      "grad_norm": 0.460922509431839,
      "learning_rate": 8.934890688072108e-06,
      "loss": 0.4332,
      "step": 14867
    },
    {
      "epoch": 0.6823000321233537,
      "grad_norm": 0.4418080449104309,
      "learning_rate": 8.934739407089479e-06,
      "loss": 0.3841,
      "step": 14868
    },
    {
      "epoch": 0.6823459226286082,
      "grad_norm": 0.4592224657535553,
      "learning_rate": 8.934588116645044e-06,
      "loss": 0.3989,
      "step": 14869
    },
    {
      "epoch": 0.6823918131338627,
      "grad_norm": 0.46631962060928345,
      "learning_rate": 8.934436816739167e-06,
      "loss": 0.4323,
      "step": 14870
    },
    {
      "epoch": 0.682437703639117,
      "grad_norm": 0.44841068983078003,
      "learning_rate": 8.93428550737221e-06,
      "loss": 0.3376,
      "step": 14871
    },
    {
      "epoch": 0.6824835941443715,
      "grad_norm": 0.5000798106193542,
      "learning_rate": 8.93413418854454e-06,
      "loss": 0.5115,
      "step": 14872
    },
    {
      "epoch": 0.682529484649626,
      "grad_norm": 0.4701462388038635,
      "learning_rate": 8.933982860256516e-06,
      "loss": 0.4426,
      "step": 14873
    },
    {
      "epoch": 0.6825753751548804,
      "grad_norm": 0.4299839735031128,
      "learning_rate": 8.933831522508507e-06,
      "loss": 0.3602,
      "step": 14874
    },
    {
      "epoch": 0.6826212656601349,
      "grad_norm": 0.41019177436828613,
      "learning_rate": 8.933680175300872e-06,
      "loss": 0.2949,
      "step": 14875
    },
    {
      "epoch": 0.6826671561653894,
      "grad_norm": 0.4400378465652466,
      "learning_rate": 8.93352881863398e-06,
      "loss": 0.4081,
      "step": 14876
    },
    {
      "epoch": 0.6827130466706438,
      "grad_norm": 0.42967677116394043,
      "learning_rate": 8.933377452508192e-06,
      "loss": 0.3443,
      "step": 14877
    },
    {
      "epoch": 0.6827589371758983,
      "grad_norm": 0.44078895449638367,
      "learning_rate": 8.933226076923874e-06,
      "loss": 0.3476,
      "step": 14878
    },
    {
      "epoch": 0.6828048276811528,
      "grad_norm": 0.461739182472229,
      "learning_rate": 8.933074691881387e-06,
      "loss": 0.4514,
      "step": 14879
    },
    {
      "epoch": 0.6828507181864072,
      "grad_norm": 0.4512290954589844,
      "learning_rate": 8.932923297381098e-06,
      "loss": 0.3866,
      "step": 14880
    },
    {
      "epoch": 0.6828966086916617,
      "grad_norm": 0.4742060601711273,
      "learning_rate": 8.932771893423369e-06,
      "loss": 0.4279,
      "step": 14881
    },
    {
      "epoch": 0.6829424991969162,
      "grad_norm": 0.45089131593704224,
      "learning_rate": 8.932620480008565e-06,
      "loss": 0.386,
      "step": 14882
    },
    {
      "epoch": 0.6829883897021706,
      "grad_norm": 0.4825097322463989,
      "learning_rate": 8.93246905713705e-06,
      "loss": 0.4175,
      "step": 14883
    },
    {
      "epoch": 0.6830342802074251,
      "grad_norm": 0.4531537890434265,
      "learning_rate": 8.932317624809187e-06,
      "loss": 0.3994,
      "step": 14884
    },
    {
      "epoch": 0.6830801707126796,
      "grad_norm": 0.4647573232650757,
      "learning_rate": 8.932166183025342e-06,
      "loss": 0.3952,
      "step": 14885
    },
    {
      "epoch": 0.683126061217934,
      "grad_norm": 0.47387707233428955,
      "learning_rate": 8.932014731785878e-06,
      "loss": 0.4392,
      "step": 14886
    },
    {
      "epoch": 0.6831719517231885,
      "grad_norm": 0.47839128971099854,
      "learning_rate": 8.93186327109116e-06,
      "loss": 0.4101,
      "step": 14887
    },
    {
      "epoch": 0.683217842228443,
      "grad_norm": 0.4589564800262451,
      "learning_rate": 8.93171180094155e-06,
      "loss": 0.4254,
      "step": 14888
    },
    {
      "epoch": 0.6832637327336974,
      "grad_norm": 0.47041061520576477,
      "learning_rate": 8.931560321337419e-06,
      "loss": 0.4403,
      "step": 14889
    },
    {
      "epoch": 0.6833096232389518,
      "grad_norm": 0.4205566942691803,
      "learning_rate": 8.931408832279122e-06,
      "loss": 0.3572,
      "step": 14890
    },
    {
      "epoch": 0.6833555137442063,
      "grad_norm": 0.44884923100471497,
      "learning_rate": 8.931257333767029e-06,
      "loss": 0.4118,
      "step": 14891
    },
    {
      "epoch": 0.6834014042494608,
      "grad_norm": 0.468823105096817,
      "learning_rate": 8.931105825801501e-06,
      "loss": 0.4288,
      "step": 14892
    },
    {
      "epoch": 0.6834472947547152,
      "grad_norm": 0.5058838725090027,
      "learning_rate": 8.930954308382906e-06,
      "loss": 0.3961,
      "step": 14893
    },
    {
      "epoch": 0.6834931852599697,
      "grad_norm": 0.4750557243824005,
      "learning_rate": 8.930802781511606e-06,
      "loss": 0.4471,
      "step": 14894
    },
    {
      "epoch": 0.6835390757652242,
      "grad_norm": 0.44421112537384033,
      "learning_rate": 8.930651245187966e-06,
      "loss": 0.3701,
      "step": 14895
    },
    {
      "epoch": 0.6835849662704786,
      "grad_norm": 0.44765132665634155,
      "learning_rate": 8.930499699412351e-06,
      "loss": 0.3522,
      "step": 14896
    },
    {
      "epoch": 0.6836308567757331,
      "grad_norm": 0.4314253628253937,
      "learning_rate": 8.930348144185124e-06,
      "loss": 0.3348,
      "step": 14897
    },
    {
      "epoch": 0.6836767472809876,
      "grad_norm": 0.46028751134872437,
      "learning_rate": 8.93019657950665e-06,
      "loss": 0.4128,
      "step": 14898
    },
    {
      "epoch": 0.683722637786242,
      "grad_norm": 0.4367113411426544,
      "learning_rate": 8.930045005377293e-06,
      "loss": 0.3779,
      "step": 14899
    },
    {
      "epoch": 0.6837685282914965,
      "grad_norm": 0.39101332426071167,
      "learning_rate": 8.929893421797418e-06,
      "loss": 0.2956,
      "step": 14900
    },
    {
      "epoch": 0.683814418796751,
      "grad_norm": 0.457587867975235,
      "learning_rate": 8.92974182876739e-06,
      "loss": 0.3595,
      "step": 14901
    },
    {
      "epoch": 0.6838603093020054,
      "grad_norm": 0.43467581272125244,
      "learning_rate": 8.929590226287575e-06,
      "loss": 0.3868,
      "step": 14902
    },
    {
      "epoch": 0.6839061998072599,
      "grad_norm": 0.458671510219574,
      "learning_rate": 8.929438614358335e-06,
      "loss": 0.4591,
      "step": 14903
    },
    {
      "epoch": 0.6839520903125144,
      "grad_norm": 0.43546923995018005,
      "learning_rate": 8.929286992980035e-06,
      "loss": 0.3952,
      "step": 14904
    },
    {
      "epoch": 0.6839979808177689,
      "grad_norm": 0.44588080048561096,
      "learning_rate": 8.92913536215304e-06,
      "loss": 0.4154,
      "step": 14905
    },
    {
      "epoch": 0.6840438713230232,
      "grad_norm": 0.44860562682151794,
      "learning_rate": 8.928983721877713e-06,
      "loss": 0.3873,
      "step": 14906
    },
    {
      "epoch": 0.6840897618282777,
      "grad_norm": 0.44021889567375183,
      "learning_rate": 8.928832072154421e-06,
      "loss": 0.3916,
      "step": 14907
    },
    {
      "epoch": 0.6841356523335322,
      "grad_norm": 0.448893278837204,
      "learning_rate": 8.928680412983527e-06,
      "loss": 0.4228,
      "step": 14908
    },
    {
      "epoch": 0.6841815428387866,
      "grad_norm": 0.4396386444568634,
      "learning_rate": 8.9285287443654e-06,
      "loss": 0.3549,
      "step": 14909
    },
    {
      "epoch": 0.6842274333440411,
      "grad_norm": 0.4707038998603821,
      "learning_rate": 8.928377066300397e-06,
      "loss": 0.4569,
      "step": 14910
    },
    {
      "epoch": 0.6842733238492956,
      "grad_norm": 0.43428173661231995,
      "learning_rate": 8.92822537878889e-06,
      "loss": 0.3333,
      "step": 14911
    },
    {
      "epoch": 0.68431921435455,
      "grad_norm": 0.46840330958366394,
      "learning_rate": 8.928073681831239e-06,
      "loss": 0.4366,
      "step": 14912
    },
    {
      "epoch": 0.6843651048598045,
      "grad_norm": 0.47723808884620667,
      "learning_rate": 8.927921975427813e-06,
      "loss": 0.4582,
      "step": 14913
    },
    {
      "epoch": 0.684410995365059,
      "grad_norm": 0.44206565618515015,
      "learning_rate": 8.927770259578972e-06,
      "loss": 0.3392,
      "step": 14914
    },
    {
      "epoch": 0.6844568858703134,
      "grad_norm": 0.40152665972709656,
      "learning_rate": 8.927618534285083e-06,
      "loss": 0.3133,
      "step": 14915
    },
    {
      "epoch": 0.6845027763755679,
      "grad_norm": 0.44860610365867615,
      "learning_rate": 8.92746679954651e-06,
      "loss": 0.4498,
      "step": 14916
    },
    {
      "epoch": 0.6845486668808224,
      "grad_norm": 0.4252551794052124,
      "learning_rate": 8.927315055363623e-06,
      "loss": 0.3075,
      "step": 14917
    },
    {
      "epoch": 0.6845945573860768,
      "grad_norm": 0.4503668546676636,
      "learning_rate": 8.92716330173678e-06,
      "loss": 0.3748,
      "step": 14918
    },
    {
      "epoch": 0.6846404478913313,
      "grad_norm": 0.4311141073703766,
      "learning_rate": 8.927011538666349e-06,
      "loss": 0.391,
      "step": 14919
    },
    {
      "epoch": 0.6846863383965858,
      "grad_norm": 0.4734487533569336,
      "learning_rate": 8.926859766152694e-06,
      "loss": 0.386,
      "step": 14920
    },
    {
      "epoch": 0.6847322289018402,
      "grad_norm": 0.4565974473953247,
      "learning_rate": 8.926707984196182e-06,
      "loss": 0.4107,
      "step": 14921
    },
    {
      "epoch": 0.6847781194070947,
      "grad_norm": 0.47143927216529846,
      "learning_rate": 8.926556192797175e-06,
      "loss": 0.4595,
      "step": 14922
    },
    {
      "epoch": 0.6848240099123492,
      "grad_norm": 0.45793333649635315,
      "learning_rate": 8.926404391956042e-06,
      "loss": 0.3985,
      "step": 14923
    },
    {
      "epoch": 0.6848699004176036,
      "grad_norm": 0.4369822144508362,
      "learning_rate": 8.926252581673144e-06,
      "loss": 0.3685,
      "step": 14924
    },
    {
      "epoch": 0.684915790922858,
      "grad_norm": 0.4429752230644226,
      "learning_rate": 8.926100761948848e-06,
      "loss": 0.385,
      "step": 14925
    },
    {
      "epoch": 0.6849616814281125,
      "grad_norm": 0.4601396918296814,
      "learning_rate": 8.925948932783519e-06,
      "loss": 0.4679,
      "step": 14926
    },
    {
      "epoch": 0.685007571933367,
      "grad_norm": 0.4582616984844208,
      "learning_rate": 8.925797094177521e-06,
      "loss": 0.4112,
      "step": 14927
    },
    {
      "epoch": 0.6850534624386214,
      "grad_norm": 0.4293476641178131,
      "learning_rate": 8.925645246131221e-06,
      "loss": 0.3497,
      "step": 14928
    },
    {
      "epoch": 0.6850993529438759,
      "grad_norm": 0.5002524852752686,
      "learning_rate": 8.925493388644983e-06,
      "loss": 0.5079,
      "step": 14929
    },
    {
      "epoch": 0.6851452434491304,
      "grad_norm": 0.4156542718410492,
      "learning_rate": 8.925341521719173e-06,
      "loss": 0.3523,
      "step": 14930
    },
    {
      "epoch": 0.6851911339543848,
      "grad_norm": 0.4862723648548126,
      "learning_rate": 8.925189645354154e-06,
      "loss": 0.3732,
      "step": 14931
    },
    {
      "epoch": 0.6852370244596393,
      "grad_norm": 0.447369247674942,
      "learning_rate": 8.925037759550293e-06,
      "loss": 0.3512,
      "step": 14932
    },
    {
      "epoch": 0.6852829149648938,
      "grad_norm": 0.41253650188446045,
      "learning_rate": 8.924885864307954e-06,
      "loss": 0.3021,
      "step": 14933
    },
    {
      "epoch": 0.6853288054701482,
      "grad_norm": 0.5070491433143616,
      "learning_rate": 8.924733959627505e-06,
      "loss": 0.5105,
      "step": 14934
    },
    {
      "epoch": 0.6853746959754027,
      "grad_norm": 0.5056631565093994,
      "learning_rate": 8.924582045509308e-06,
      "loss": 0.4951,
      "step": 14935
    },
    {
      "epoch": 0.6854205864806572,
      "grad_norm": 0.47183579206466675,
      "learning_rate": 8.924430121953733e-06,
      "loss": 0.4581,
      "step": 14936
    },
    {
      "epoch": 0.6854664769859116,
      "grad_norm": 0.4818021357059479,
      "learning_rate": 8.92427818896114e-06,
      "loss": 0.4623,
      "step": 14937
    },
    {
      "epoch": 0.6855123674911661,
      "grad_norm": 0.4583207368850708,
      "learning_rate": 8.924126246531896e-06,
      "loss": 0.4314,
      "step": 14938
    },
    {
      "epoch": 0.6855582579964206,
      "grad_norm": 0.5014256238937378,
      "learning_rate": 8.923974294666367e-06,
      "loss": 0.4633,
      "step": 14939
    },
    {
      "epoch": 0.685604148501675,
      "grad_norm": 0.4391101002693176,
      "learning_rate": 8.923822333364918e-06,
      "loss": 0.3682,
      "step": 14940
    },
    {
      "epoch": 0.6856500390069294,
      "grad_norm": 0.44086316227912903,
      "learning_rate": 8.923670362627917e-06,
      "loss": 0.3728,
      "step": 14941
    },
    {
      "epoch": 0.6856959295121839,
      "grad_norm": 0.4357971251010895,
      "learning_rate": 8.923518382455724e-06,
      "loss": 0.384,
      "step": 14942
    },
    {
      "epoch": 0.6857418200174384,
      "grad_norm": 0.45085909962654114,
      "learning_rate": 8.92336639284871e-06,
      "loss": 0.38,
      "step": 14943
    },
    {
      "epoch": 0.6857877105226928,
      "grad_norm": 0.464751660823822,
      "learning_rate": 8.923214393807236e-06,
      "loss": 0.4683,
      "step": 14944
    },
    {
      "epoch": 0.6858336010279473,
      "grad_norm": 0.4326508045196533,
      "learning_rate": 8.923062385331671e-06,
      "loss": 0.4027,
      "step": 14945
    },
    {
      "epoch": 0.6858794915332018,
      "grad_norm": 0.44074225425720215,
      "learning_rate": 8.922910367422379e-06,
      "loss": 0.3505,
      "step": 14946
    },
    {
      "epoch": 0.6859253820384562,
      "grad_norm": 0.4517102539539337,
      "learning_rate": 8.922758340079724e-06,
      "loss": 0.391,
      "step": 14947
    },
    {
      "epoch": 0.6859712725437107,
      "grad_norm": 0.4954013228416443,
      "learning_rate": 8.922606303304075e-06,
      "loss": 0.4792,
      "step": 14948
    },
    {
      "epoch": 0.6860171630489652,
      "grad_norm": 0.4592500925064087,
      "learning_rate": 8.922454257095796e-06,
      "loss": 0.4164,
      "step": 14949
    },
    {
      "epoch": 0.6860630535542196,
      "grad_norm": 0.4353947639465332,
      "learning_rate": 8.922302201455253e-06,
      "loss": 0.3573,
      "step": 14950
    },
    {
      "epoch": 0.6861089440594741,
      "grad_norm": 0.44320446252822876,
      "learning_rate": 8.92215013638281e-06,
      "loss": 0.3776,
      "step": 14951
    },
    {
      "epoch": 0.6861548345647286,
      "grad_norm": 0.4774834215641022,
      "learning_rate": 8.921998061878833e-06,
      "loss": 0.364,
      "step": 14952
    },
    {
      "epoch": 0.686200725069983,
      "grad_norm": 0.46389034390449524,
      "learning_rate": 8.921845977943691e-06,
      "loss": 0.4105,
      "step": 14953
    },
    {
      "epoch": 0.6862466155752375,
      "grad_norm": 0.4633306860923767,
      "learning_rate": 8.921693884577744e-06,
      "loss": 0.3347,
      "step": 14954
    },
    {
      "epoch": 0.686292506080492,
      "grad_norm": 0.4615379273891449,
      "learning_rate": 8.921541781781362e-06,
      "loss": 0.3844,
      "step": 14955
    },
    {
      "epoch": 0.6863383965857464,
      "grad_norm": 0.4374758303165436,
      "learning_rate": 8.921389669554912e-06,
      "loss": 0.4007,
      "step": 14956
    },
    {
      "epoch": 0.6863842870910009,
      "grad_norm": 0.44353294372558594,
      "learning_rate": 8.921237547898755e-06,
      "loss": 0.3565,
      "step": 14957
    },
    {
      "epoch": 0.6864301775962554,
      "grad_norm": 0.4245252311229706,
      "learning_rate": 8.92108541681326e-06,
      "loss": 0.3584,
      "step": 14958
    },
    {
      "epoch": 0.6864760681015099,
      "grad_norm": 0.43870529532432556,
      "learning_rate": 8.920933276298791e-06,
      "loss": 0.3811,
      "step": 14959
    },
    {
      "epoch": 0.6865219586067642,
      "grad_norm": 0.4415648579597473,
      "learning_rate": 8.920781126355717e-06,
      "loss": 0.3679,
      "step": 14960
    },
    {
      "epoch": 0.6865678491120187,
      "grad_norm": 0.4761502146720886,
      "learning_rate": 8.920628966984402e-06,
      "loss": 0.4011,
      "step": 14961
    },
    {
      "epoch": 0.6866137396172732,
      "grad_norm": 0.4760671854019165,
      "learning_rate": 8.920476798185209e-06,
      "loss": 0.4554,
      "step": 14962
    },
    {
      "epoch": 0.6866596301225276,
      "grad_norm": 0.4473685622215271,
      "learning_rate": 8.920324619958509e-06,
      "loss": 0.4023,
      "step": 14963
    },
    {
      "epoch": 0.6867055206277821,
      "grad_norm": 0.4060588479042053,
      "learning_rate": 8.920172432304664e-06,
      "loss": 0.2982,
      "step": 14964
    },
    {
      "epoch": 0.6867514111330366,
      "grad_norm": 0.4596617519855499,
      "learning_rate": 8.920020235224044e-06,
      "loss": 0.3659,
      "step": 14965
    },
    {
      "epoch": 0.686797301638291,
      "grad_norm": 0.45741286873817444,
      "learning_rate": 8.919868028717009e-06,
      "loss": 0.4374,
      "step": 14966
    },
    {
      "epoch": 0.6868431921435455,
      "grad_norm": 0.41568976640701294,
      "learning_rate": 8.91971581278393e-06,
      "loss": 0.2884,
      "step": 14967
    },
    {
      "epoch": 0.6868890826488,
      "grad_norm": 0.484148770570755,
      "learning_rate": 8.919563587425173e-06,
      "loss": 0.4862,
      "step": 14968
    },
    {
      "epoch": 0.6869349731540544,
      "grad_norm": 0.5416460633277893,
      "learning_rate": 8.919411352641102e-06,
      "loss": 0.442,
      "step": 14969
    },
    {
      "epoch": 0.6869808636593089,
      "grad_norm": 0.4361247718334198,
      "learning_rate": 8.919259108432083e-06,
      "loss": 0.3608,
      "step": 14970
    },
    {
      "epoch": 0.6870267541645634,
      "grad_norm": 0.4695515036582947,
      "learning_rate": 8.919106854798483e-06,
      "loss": 0.4105,
      "step": 14971
    },
    {
      "epoch": 0.6870726446698178,
      "grad_norm": 0.42950406670570374,
      "learning_rate": 8.918954591740667e-06,
      "loss": 0.3842,
      "step": 14972
    },
    {
      "epoch": 0.6871185351750723,
      "grad_norm": 0.48853814601898193,
      "learning_rate": 8.918802319259004e-06,
      "loss": 0.4699,
      "step": 14973
    },
    {
      "epoch": 0.6871644256803268,
      "grad_norm": 0.46227094531059265,
      "learning_rate": 8.918650037353858e-06,
      "loss": 0.4438,
      "step": 14974
    },
    {
      "epoch": 0.6872103161855811,
      "grad_norm": 0.42016327381134033,
      "learning_rate": 8.918497746025594e-06,
      "loss": 0.3694,
      "step": 14975
    },
    {
      "epoch": 0.6872562066908356,
      "grad_norm": 0.45480069518089294,
      "learning_rate": 8.91834544527458e-06,
      "loss": 0.3921,
      "step": 14976
    },
    {
      "epoch": 0.6873020971960901,
      "grad_norm": 0.47756290435791016,
      "learning_rate": 8.918193135101184e-06,
      "loss": 0.4539,
      "step": 14977
    },
    {
      "epoch": 0.6873479877013446,
      "grad_norm": 0.4812740087509155,
      "learning_rate": 8.918040815505767e-06,
      "loss": 0.4032,
      "step": 14978
    },
    {
      "epoch": 0.687393878206599,
      "grad_norm": 0.46387481689453125,
      "learning_rate": 8.9178884864887e-06,
      "loss": 0.4336,
      "step": 14979
    },
    {
      "epoch": 0.6874397687118535,
      "grad_norm": 0.4229836165904999,
      "learning_rate": 8.917736148050347e-06,
      "loss": 0.3388,
      "step": 14980
    },
    {
      "epoch": 0.687485659217108,
      "grad_norm": 0.42387086153030396,
      "learning_rate": 8.917583800191076e-06,
      "loss": 0.3531,
      "step": 14981
    },
    {
      "epoch": 0.6875315497223624,
      "grad_norm": 0.46882203221321106,
      "learning_rate": 8.917431442911251e-06,
      "loss": 0.4454,
      "step": 14982
    },
    {
      "epoch": 0.6875774402276169,
      "grad_norm": 0.49281471967697144,
      "learning_rate": 8.917279076211242e-06,
      "loss": 0.4341,
      "step": 14983
    },
    {
      "epoch": 0.6876233307328714,
      "grad_norm": 0.46577614545822144,
      "learning_rate": 8.917126700091412e-06,
      "loss": 0.3798,
      "step": 14984
    },
    {
      "epoch": 0.6876692212381258,
      "grad_norm": 0.4871450960636139,
      "learning_rate": 8.916974314552128e-06,
      "loss": 0.4174,
      "step": 14985
    },
    {
      "epoch": 0.6877151117433803,
      "grad_norm": 0.501464307308197,
      "learning_rate": 8.916821919593757e-06,
      "loss": 0.5031,
      "step": 14986
    },
    {
      "epoch": 0.6877610022486348,
      "grad_norm": 0.5052502751350403,
      "learning_rate": 8.916669515216664e-06,
      "loss": 0.385,
      "step": 14987
    },
    {
      "epoch": 0.6878068927538892,
      "grad_norm": 0.423660010099411,
      "learning_rate": 8.916517101421221e-06,
      "loss": 0.3372,
      "step": 14988
    },
    {
      "epoch": 0.6878527832591437,
      "grad_norm": 0.4601968824863434,
      "learning_rate": 8.916364678207788e-06,
      "loss": 0.3893,
      "step": 14989
    },
    {
      "epoch": 0.6878986737643982,
      "grad_norm": 0.4666600823402405,
      "learning_rate": 8.916212245576734e-06,
      "loss": 0.3893,
      "step": 14990
    },
    {
      "epoch": 0.6879445642696526,
      "grad_norm": 0.4620806872844696,
      "learning_rate": 8.916059803528426e-06,
      "loss": 0.4488,
      "step": 14991
    },
    {
      "epoch": 0.6879904547749071,
      "grad_norm": 0.42871516942977905,
      "learning_rate": 8.91590735206323e-06,
      "loss": 0.3543,
      "step": 14992
    },
    {
      "epoch": 0.6880363452801616,
      "grad_norm": 0.43291613459587097,
      "learning_rate": 8.915754891181512e-06,
      "loss": 0.3422,
      "step": 14993
    },
    {
      "epoch": 0.688082235785416,
      "grad_norm": 0.46761688590049744,
      "learning_rate": 8.91560242088364e-06,
      "loss": 0.3907,
      "step": 14994
    },
    {
      "epoch": 0.6881281262906704,
      "grad_norm": 0.513192892074585,
      "learning_rate": 8.91544994116998e-06,
      "loss": 0.4788,
      "step": 14995
    },
    {
      "epoch": 0.6881740167959249,
      "grad_norm": 0.4781024158000946,
      "learning_rate": 8.915297452040899e-06,
      "loss": 0.4213,
      "step": 14996
    },
    {
      "epoch": 0.6882199073011794,
      "grad_norm": 0.4775366485118866,
      "learning_rate": 8.915144953496763e-06,
      "loss": 0.4836,
      "step": 14997
    },
    {
      "epoch": 0.6882657978064338,
      "grad_norm": 0.45348578691482544,
      "learning_rate": 8.91499244553794e-06,
      "loss": 0.3806,
      "step": 14998
    },
    {
      "epoch": 0.6883116883116883,
      "grad_norm": 0.4858158230781555,
      "learning_rate": 8.914839928164793e-06,
      "loss": 0.391,
      "step": 14999
    },
    {
      "epoch": 0.6883575788169428,
      "grad_norm": 0.4520188868045807,
      "learning_rate": 8.914687401377693e-06,
      "loss": 0.3567,
      "step": 15000
    },
    {
      "epoch": 0.6884034693221972,
      "grad_norm": 0.4315658211708069,
      "learning_rate": 8.914534865177005e-06,
      "loss": 0.3568,
      "step": 15001
    },
    {
      "epoch": 0.6884493598274517,
      "grad_norm": 0.4763556718826294,
      "learning_rate": 8.914382319563098e-06,
      "loss": 0.4319,
      "step": 15002
    },
    {
      "epoch": 0.6884952503327062,
      "grad_norm": 0.4684775173664093,
      "learning_rate": 8.914229764536335e-06,
      "loss": 0.4401,
      "step": 15003
    },
    {
      "epoch": 0.6885411408379606,
      "grad_norm": 0.4878898561000824,
      "learning_rate": 8.914077200097085e-06,
      "loss": 0.478,
      "step": 15004
    },
    {
      "epoch": 0.6885870313432151,
      "grad_norm": 0.4497095048427582,
      "learning_rate": 8.913924626245715e-06,
      "loss": 0.3742,
      "step": 15005
    },
    {
      "epoch": 0.6886329218484696,
      "grad_norm": 0.4629368782043457,
      "learning_rate": 8.913772042982593e-06,
      "loss": 0.4273,
      "step": 15006
    },
    {
      "epoch": 0.688678812353724,
      "grad_norm": 0.45213782787323,
      "learning_rate": 8.913619450308083e-06,
      "loss": 0.3808,
      "step": 15007
    },
    {
      "epoch": 0.6887247028589785,
      "grad_norm": 0.468328595161438,
      "learning_rate": 8.913466848222553e-06,
      "loss": 0.3621,
      "step": 15008
    },
    {
      "epoch": 0.688770593364233,
      "grad_norm": 0.48691269755363464,
      "learning_rate": 8.91331423672637e-06,
      "loss": 0.4772,
      "step": 15009
    },
    {
      "epoch": 0.6888164838694874,
      "grad_norm": 0.4540717601776123,
      "learning_rate": 8.913161615819903e-06,
      "loss": 0.3953,
      "step": 15010
    },
    {
      "epoch": 0.6888623743747418,
      "grad_norm": 0.507427453994751,
      "learning_rate": 8.913008985503516e-06,
      "loss": 0.4699,
      "step": 15011
    },
    {
      "epoch": 0.6889082648799963,
      "grad_norm": 0.515504777431488,
      "learning_rate": 8.912856345777579e-06,
      "loss": 0.5051,
      "step": 15012
    },
    {
      "epoch": 0.6889541553852508,
      "grad_norm": 0.4142671525478363,
      "learning_rate": 8.912703696642456e-06,
      "loss": 0.317,
      "step": 15013
    },
    {
      "epoch": 0.6890000458905052,
      "grad_norm": 0.43825942277908325,
      "learning_rate": 8.912551038098516e-06,
      "loss": 0.3791,
      "step": 15014
    },
    {
      "epoch": 0.6890459363957597,
      "grad_norm": 0.48192837834358215,
      "learning_rate": 8.912398370146124e-06,
      "loss": 0.3631,
      "step": 15015
    },
    {
      "epoch": 0.6890918269010142,
      "grad_norm": 0.4621369242668152,
      "learning_rate": 8.912245692785651e-06,
      "loss": 0.4348,
      "step": 15016
    },
    {
      "epoch": 0.6891377174062686,
      "grad_norm": 0.4731169641017914,
      "learning_rate": 8.91209300601746e-06,
      "loss": 0.4126,
      "step": 15017
    },
    {
      "epoch": 0.6891836079115231,
      "grad_norm": 0.4565713405609131,
      "learning_rate": 8.911940309841923e-06,
      "loss": 0.3899,
      "step": 15018
    },
    {
      "epoch": 0.6892294984167776,
      "grad_norm": 0.45031362771987915,
      "learning_rate": 8.9117876042594e-06,
      "loss": 0.3908,
      "step": 15019
    },
    {
      "epoch": 0.689275388922032,
      "grad_norm": 0.46312087774276733,
      "learning_rate": 8.911634889270265e-06,
      "loss": 0.403,
      "step": 15020
    },
    {
      "epoch": 0.6893212794272865,
      "grad_norm": 0.4490703046321869,
      "learning_rate": 8.911482164874884e-06,
      "loss": 0.3989,
      "step": 15021
    },
    {
      "epoch": 0.689367169932541,
      "grad_norm": 0.4504191279411316,
      "learning_rate": 8.911329431073618e-06,
      "loss": 0.4131,
      "step": 15022
    },
    {
      "epoch": 0.6894130604377954,
      "grad_norm": 0.48080718517303467,
      "learning_rate": 8.911176687866844e-06,
      "loss": 0.418,
      "step": 15023
    },
    {
      "epoch": 0.6894589509430499,
      "grad_norm": 0.4744383692741394,
      "learning_rate": 8.911023935254921e-06,
      "loss": 0.4024,
      "step": 15024
    },
    {
      "epoch": 0.6895048414483044,
      "grad_norm": 0.42912599444389343,
      "learning_rate": 8.910871173238223e-06,
      "loss": 0.3842,
      "step": 15025
    },
    {
      "epoch": 0.6895507319535588,
      "grad_norm": 0.4116972088813782,
      "learning_rate": 8.910718401817113e-06,
      "loss": 0.3377,
      "step": 15026
    },
    {
      "epoch": 0.6895966224588133,
      "grad_norm": 0.4719600975513458,
      "learning_rate": 8.910565620991958e-06,
      "loss": 0.4582,
      "step": 15027
    },
    {
      "epoch": 0.6896425129640678,
      "grad_norm": 0.45434215664863586,
      "learning_rate": 8.910412830763128e-06,
      "loss": 0.4046,
      "step": 15028
    },
    {
      "epoch": 0.6896884034693221,
      "grad_norm": 0.4496857523918152,
      "learning_rate": 8.91026003113099e-06,
      "loss": 0.3962,
      "step": 15029
    },
    {
      "epoch": 0.6897342939745766,
      "grad_norm": 0.4435851275920868,
      "learning_rate": 8.91010722209591e-06,
      "loss": 0.4155,
      "step": 15030
    },
    {
      "epoch": 0.6897801844798311,
      "grad_norm": 0.44755688309669495,
      "learning_rate": 8.909954403658256e-06,
      "loss": 0.3505,
      "step": 15031
    },
    {
      "epoch": 0.6898260749850856,
      "grad_norm": 0.5074527859687805,
      "learning_rate": 8.909801575818397e-06,
      "loss": 0.5101,
      "step": 15032
    },
    {
      "epoch": 0.68987196549034,
      "grad_norm": 0.47744691371917725,
      "learning_rate": 8.909648738576697e-06,
      "loss": 0.4434,
      "step": 15033
    },
    {
      "epoch": 0.6899178559955945,
      "grad_norm": 0.6416549682617188,
      "learning_rate": 8.909495891933526e-06,
      "loss": 0.3874,
      "step": 15034
    },
    {
      "epoch": 0.689963746500849,
      "grad_norm": 0.43551865220069885,
      "learning_rate": 8.909343035889252e-06,
      "loss": 0.4028,
      "step": 15035
    },
    {
      "epoch": 0.6900096370061034,
      "grad_norm": 0.4496057331562042,
      "learning_rate": 8.909190170444242e-06,
      "loss": 0.3852,
      "step": 15036
    },
    {
      "epoch": 0.6900555275113579,
      "grad_norm": 0.4212131202220917,
      "learning_rate": 8.909037295598864e-06,
      "loss": 0.3642,
      "step": 15037
    },
    {
      "epoch": 0.6901014180166124,
      "grad_norm": 0.4716685116291046,
      "learning_rate": 8.908884411353485e-06,
      "loss": 0.4415,
      "step": 15038
    },
    {
      "epoch": 0.6901473085218668,
      "grad_norm": 0.4818684160709381,
      "learning_rate": 8.908731517708472e-06,
      "loss": 0.4558,
      "step": 15039
    },
    {
      "epoch": 0.6901931990271213,
      "grad_norm": 0.4738227427005768,
      "learning_rate": 8.908578614664194e-06,
      "loss": 0.4122,
      "step": 15040
    },
    {
      "epoch": 0.6902390895323758,
      "grad_norm": 0.47754526138305664,
      "learning_rate": 8.908425702221016e-06,
      "loss": 0.4358,
      "step": 15041
    },
    {
      "epoch": 0.6902849800376302,
      "grad_norm": 0.4478150010108948,
      "learning_rate": 8.90827278037931e-06,
      "loss": 0.4292,
      "step": 15042
    },
    {
      "epoch": 0.6903308705428847,
      "grad_norm": 0.4927135407924652,
      "learning_rate": 8.908119849139441e-06,
      "loss": 0.4963,
      "step": 15043
    },
    {
      "epoch": 0.6903767610481392,
      "grad_norm": 0.469346284866333,
      "learning_rate": 8.907966908501775e-06,
      "loss": 0.4111,
      "step": 15044
    },
    {
      "epoch": 0.6904226515533936,
      "grad_norm": 0.45989370346069336,
      "learning_rate": 8.907813958466684e-06,
      "loss": 0.4432,
      "step": 15045
    },
    {
      "epoch": 0.690468542058648,
      "grad_norm": 0.44060951471328735,
      "learning_rate": 8.907660999034534e-06,
      "loss": 0.343,
      "step": 15046
    },
    {
      "epoch": 0.6905144325639025,
      "grad_norm": 0.45614075660705566,
      "learning_rate": 8.907508030205692e-06,
      "loss": 0.4079,
      "step": 15047
    },
    {
      "epoch": 0.690560323069157,
      "grad_norm": 0.42114952206611633,
      "learning_rate": 8.907355051980525e-06,
      "loss": 0.341,
      "step": 15048
    },
    {
      "epoch": 0.6906062135744114,
      "grad_norm": 0.42861485481262207,
      "learning_rate": 8.907202064359405e-06,
      "loss": 0.3322,
      "step": 15049
    },
    {
      "epoch": 0.6906521040796659,
      "grad_norm": 0.43844321370124817,
      "learning_rate": 8.907049067342697e-06,
      "loss": 0.3592,
      "step": 15050
    },
    {
      "epoch": 0.6906979945849204,
      "grad_norm": 0.40976181626319885,
      "learning_rate": 8.906896060930768e-06,
      "loss": 0.3518,
      "step": 15051
    },
    {
      "epoch": 0.6907438850901748,
      "grad_norm": 0.4560471177101135,
      "learning_rate": 8.906743045123986e-06,
      "loss": 0.4374,
      "step": 15052
    },
    {
      "epoch": 0.6907897755954293,
      "grad_norm": 0.4475359320640564,
      "learning_rate": 8.906590019922721e-06,
      "loss": 0.3923,
      "step": 15053
    },
    {
      "epoch": 0.6908356661006838,
      "grad_norm": 0.47006359696388245,
      "learning_rate": 8.90643698532734e-06,
      "loss": 0.3878,
      "step": 15054
    },
    {
      "epoch": 0.6908815566059382,
      "grad_norm": 0.4970708191394806,
      "learning_rate": 8.906283941338212e-06,
      "loss": 0.4285,
      "step": 15055
    },
    {
      "epoch": 0.6909274471111927,
      "grad_norm": 0.42144298553466797,
      "learning_rate": 8.906130887955702e-06,
      "loss": 0.3216,
      "step": 15056
    },
    {
      "epoch": 0.6909733376164472,
      "grad_norm": 0.4323214590549469,
      "learning_rate": 8.905977825180181e-06,
      "loss": 0.3316,
      "step": 15057
    },
    {
      "epoch": 0.6910192281217016,
      "grad_norm": 0.4509257674217224,
      "learning_rate": 8.905824753012015e-06,
      "loss": 0.3824,
      "step": 15058
    },
    {
      "epoch": 0.6910651186269561,
      "grad_norm": 0.4705160856246948,
      "learning_rate": 8.905671671451577e-06,
      "loss": 0.425,
      "step": 15059
    },
    {
      "epoch": 0.6911110091322106,
      "grad_norm": 0.45417124032974243,
      "learning_rate": 8.905518580499228e-06,
      "loss": 0.3772,
      "step": 15060
    },
    {
      "epoch": 0.691156899637465,
      "grad_norm": 0.4337405264377594,
      "learning_rate": 8.90536548015534e-06,
      "loss": 0.3679,
      "step": 15061
    },
    {
      "epoch": 0.6912027901427195,
      "grad_norm": 0.47078076004981995,
      "learning_rate": 8.90521237042028e-06,
      "loss": 0.4144,
      "step": 15062
    },
    {
      "epoch": 0.691248680647974,
      "grad_norm": 0.47007036209106445,
      "learning_rate": 8.905059251294416e-06,
      "loss": 0.3477,
      "step": 15063
    },
    {
      "epoch": 0.6912945711532283,
      "grad_norm": 0.4606531858444214,
      "learning_rate": 8.90490612277812e-06,
      "loss": 0.3665,
      "step": 15064
    },
    {
      "epoch": 0.6913404616584828,
      "grad_norm": 0.4860847294330597,
      "learning_rate": 8.904752984871756e-06,
      "loss": 0.4713,
      "step": 15065
    },
    {
      "epoch": 0.6913863521637373,
      "grad_norm": 0.44397976994514465,
      "learning_rate": 8.904599837575692e-06,
      "loss": 0.3735,
      "step": 15066
    },
    {
      "epoch": 0.6914322426689918,
      "grad_norm": 0.42582622170448303,
      "learning_rate": 8.904446680890299e-06,
      "loss": 0.3328,
      "step": 15067
    },
    {
      "epoch": 0.6914781331742462,
      "grad_norm": 0.41647234559059143,
      "learning_rate": 8.904293514815944e-06,
      "loss": 0.3264,
      "step": 15068
    },
    {
      "epoch": 0.6915240236795007,
      "grad_norm": 0.4586029052734375,
      "learning_rate": 8.904140339352994e-06,
      "loss": 0.4058,
      "step": 15069
    },
    {
      "epoch": 0.6915699141847552,
      "grad_norm": 0.43757638335227966,
      "learning_rate": 8.903987154501821e-06,
      "loss": 0.3785,
      "step": 15070
    },
    {
      "epoch": 0.6916158046900096,
      "grad_norm": 0.44139403104782104,
      "learning_rate": 8.903833960262788e-06,
      "loss": 0.3863,
      "step": 15071
    },
    {
      "epoch": 0.6916616951952641,
      "grad_norm": 0.430138498544693,
      "learning_rate": 8.90368075663627e-06,
      "loss": 0.3448,
      "step": 15072
    },
    {
      "epoch": 0.6917075857005186,
      "grad_norm": 0.46243876218795776,
      "learning_rate": 8.90352754362263e-06,
      "loss": 0.4137,
      "step": 15073
    },
    {
      "epoch": 0.691753476205773,
      "grad_norm": 0.451144278049469,
      "learning_rate": 8.903374321222238e-06,
      "loss": 0.3687,
      "step": 15074
    },
    {
      "epoch": 0.6917993667110275,
      "grad_norm": 0.44124680757522583,
      "learning_rate": 8.903221089435463e-06,
      "loss": 0.4067,
      "step": 15075
    },
    {
      "epoch": 0.691845257216282,
      "grad_norm": 0.5448135137557983,
      "learning_rate": 8.903067848262672e-06,
      "loss": 0.4468,
      "step": 15076
    },
    {
      "epoch": 0.6918911477215364,
      "grad_norm": 0.4734453856945038,
      "learning_rate": 8.902914597704236e-06,
      "loss": 0.4131,
      "step": 15077
    },
    {
      "epoch": 0.6919370382267909,
      "grad_norm": 0.43498197197914124,
      "learning_rate": 8.90276133776052e-06,
      "loss": 0.3443,
      "step": 15078
    },
    {
      "epoch": 0.6919829287320454,
      "grad_norm": 0.48123863339424133,
      "learning_rate": 8.902608068431899e-06,
      "loss": 0.4618,
      "step": 15079
    },
    {
      "epoch": 0.6920288192372998,
      "grad_norm": 0.5311402082443237,
      "learning_rate": 8.902454789718735e-06,
      "loss": 0.6245,
      "step": 15080
    },
    {
      "epoch": 0.6920747097425543,
      "grad_norm": 0.47681382298469543,
      "learning_rate": 8.9023015016214e-06,
      "loss": 0.4363,
      "step": 15081
    },
    {
      "epoch": 0.6921206002478087,
      "grad_norm": 0.4866882264614105,
      "learning_rate": 8.902148204140258e-06,
      "loss": 0.4313,
      "step": 15082
    },
    {
      "epoch": 0.6921664907530632,
      "grad_norm": 0.4718891382217407,
      "learning_rate": 8.901994897275685e-06,
      "loss": 0.4181,
      "step": 15083
    },
    {
      "epoch": 0.6922123812583176,
      "grad_norm": 0.4618433117866516,
      "learning_rate": 8.901841581028043e-06,
      "loss": 0.4029,
      "step": 15084
    },
    {
      "epoch": 0.6922582717635721,
      "grad_norm": 0.44824185967445374,
      "learning_rate": 8.901688255397705e-06,
      "loss": 0.3769,
      "step": 15085
    },
    {
      "epoch": 0.6923041622688266,
      "grad_norm": 0.4765690565109253,
      "learning_rate": 8.901534920385037e-06,
      "loss": 0.4352,
      "step": 15086
    },
    {
      "epoch": 0.692350052774081,
      "grad_norm": 0.4359038472175598,
      "learning_rate": 8.90138157599041e-06,
      "loss": 0.3808,
      "step": 15087
    },
    {
      "epoch": 0.6923959432793355,
      "grad_norm": 0.43580570816993713,
      "learning_rate": 8.901228222214191e-06,
      "loss": 0.3701,
      "step": 15088
    },
    {
      "epoch": 0.69244183378459,
      "grad_norm": 0.4450821876525879,
      "learning_rate": 8.90107485905675e-06,
      "loss": 0.3581,
      "step": 15089
    },
    {
      "epoch": 0.6924877242898444,
      "grad_norm": 0.46013516187667847,
      "learning_rate": 8.900921486518456e-06,
      "loss": 0.3828,
      "step": 15090
    },
    {
      "epoch": 0.6925336147950989,
      "grad_norm": 0.4831649661064148,
      "learning_rate": 8.900768104599676e-06,
      "loss": 0.4514,
      "step": 15091
    },
    {
      "epoch": 0.6925795053003534,
      "grad_norm": 0.46518245339393616,
      "learning_rate": 8.900614713300779e-06,
      "loss": 0.4314,
      "step": 15092
    },
    {
      "epoch": 0.6926253958056078,
      "grad_norm": 0.4164680242538452,
      "learning_rate": 8.900461312622135e-06,
      "loss": 0.3519,
      "step": 15093
    },
    {
      "epoch": 0.6926712863108623,
      "grad_norm": 0.4460037052631378,
      "learning_rate": 8.900307902564113e-06,
      "loss": 0.3734,
      "step": 15094
    },
    {
      "epoch": 0.6927171768161168,
      "grad_norm": 0.4843788743019104,
      "learning_rate": 8.900154483127082e-06,
      "loss": 0.4899,
      "step": 15095
    },
    {
      "epoch": 0.6927630673213712,
      "grad_norm": 0.48087796568870544,
      "learning_rate": 8.90000105431141e-06,
      "loss": 0.454,
      "step": 15096
    },
    {
      "epoch": 0.6928089578266257,
      "grad_norm": 0.49372223019599915,
      "learning_rate": 8.899847616117466e-06,
      "loss": 0.4397,
      "step": 15097
    },
    {
      "epoch": 0.6928548483318802,
      "grad_norm": 0.4934097230434418,
      "learning_rate": 8.89969416854562e-06,
      "loss": 0.4248,
      "step": 15098
    },
    {
      "epoch": 0.6929007388371345,
      "grad_norm": 0.45487314462661743,
      "learning_rate": 8.89954071159624e-06,
      "loss": 0.3673,
      "step": 15099
    },
    {
      "epoch": 0.692946629342389,
      "grad_norm": 0.5339693427085876,
      "learning_rate": 8.899387245269695e-06,
      "loss": 0.4942,
      "step": 15100
    },
    {
      "epoch": 0.6929925198476435,
      "grad_norm": 0.4568933844566345,
      "learning_rate": 8.899233769566353e-06,
      "loss": 0.4216,
      "step": 15101
    },
    {
      "epoch": 0.693038410352898,
      "grad_norm": 0.45888862013816833,
      "learning_rate": 8.899080284486586e-06,
      "loss": 0.3464,
      "step": 15102
    },
    {
      "epoch": 0.6930843008581524,
      "grad_norm": 0.4929792582988739,
      "learning_rate": 8.898926790030762e-06,
      "loss": 0.4545,
      "step": 15103
    },
    {
      "epoch": 0.6931301913634069,
      "grad_norm": 0.4200250506401062,
      "learning_rate": 8.89877328619925e-06,
      "loss": 0.3314,
      "step": 15104
    },
    {
      "epoch": 0.6931760818686614,
      "grad_norm": 0.49718424677848816,
      "learning_rate": 8.898619772992417e-06,
      "loss": 0.4436,
      "step": 15105
    },
    {
      "epoch": 0.6932219723739158,
      "grad_norm": 0.4373233914375305,
      "learning_rate": 8.898466250410634e-06,
      "loss": 0.3507,
      "step": 15106
    },
    {
      "epoch": 0.6932678628791703,
      "grad_norm": 0.4857986569404602,
      "learning_rate": 8.898312718454271e-06,
      "loss": 0.3894,
      "step": 15107
    },
    {
      "epoch": 0.6933137533844248,
      "grad_norm": 0.45529282093048096,
      "learning_rate": 8.898159177123697e-06,
      "loss": 0.3844,
      "step": 15108
    },
    {
      "epoch": 0.6933596438896792,
      "grad_norm": 0.46354246139526367,
      "learning_rate": 8.898005626419278e-06,
      "loss": 0.3848,
      "step": 15109
    },
    {
      "epoch": 0.6934055343949337,
      "grad_norm": 0.4467262029647827,
      "learning_rate": 8.897852066341388e-06,
      "loss": 0.3708,
      "step": 15110
    },
    {
      "epoch": 0.6934514249001882,
      "grad_norm": 0.47388315200805664,
      "learning_rate": 8.897698496890392e-06,
      "loss": 0.46,
      "step": 15111
    },
    {
      "epoch": 0.6934973154054426,
      "grad_norm": 0.43150198459625244,
      "learning_rate": 8.897544918066663e-06,
      "loss": 0.3362,
      "step": 15112
    },
    {
      "epoch": 0.6935432059106971,
      "grad_norm": 0.4536038935184479,
      "learning_rate": 8.897391329870569e-06,
      "loss": 0.368,
      "step": 15113
    },
    {
      "epoch": 0.6935890964159516,
      "grad_norm": 0.4295598864555359,
      "learning_rate": 8.897237732302477e-06,
      "loss": 0.3591,
      "step": 15114
    },
    {
      "epoch": 0.693634986921206,
      "grad_norm": 0.47307121753692627,
      "learning_rate": 8.897084125362758e-06,
      "loss": 0.4564,
      "step": 15115
    },
    {
      "epoch": 0.6936808774264605,
      "grad_norm": 0.44129276275634766,
      "learning_rate": 8.896930509051783e-06,
      "loss": 0.3645,
      "step": 15116
    },
    {
      "epoch": 0.693726767931715,
      "grad_norm": 0.48488351702690125,
      "learning_rate": 8.896776883369919e-06,
      "loss": 0.4329,
      "step": 15117
    },
    {
      "epoch": 0.6937726584369693,
      "grad_norm": 0.4221949279308319,
      "learning_rate": 8.896623248317538e-06,
      "loss": 0.3324,
      "step": 15118
    },
    {
      "epoch": 0.6938185489422238,
      "grad_norm": 0.4085926115512848,
      "learning_rate": 8.896469603895007e-06,
      "loss": 0.3358,
      "step": 15119
    },
    {
      "epoch": 0.6938644394474783,
      "grad_norm": 0.4322345554828644,
      "learning_rate": 8.896315950102697e-06,
      "loss": 0.3938,
      "step": 15120
    },
    {
      "epoch": 0.6939103299527328,
      "grad_norm": 0.4339832663536072,
      "learning_rate": 8.896162286940975e-06,
      "loss": 0.3829,
      "step": 15121
    },
    {
      "epoch": 0.6939562204579872,
      "grad_norm": 0.4838652014732361,
      "learning_rate": 8.896008614410215e-06,
      "loss": 0.4948,
      "step": 15122
    },
    {
      "epoch": 0.6940021109632417,
      "grad_norm": 0.4479416608810425,
      "learning_rate": 8.895854932510782e-06,
      "loss": 0.3512,
      "step": 15123
    },
    {
      "epoch": 0.6940480014684962,
      "grad_norm": 0.46026772260665894,
      "learning_rate": 8.895701241243048e-06,
      "loss": 0.4037,
      "step": 15124
    },
    {
      "epoch": 0.6940938919737506,
      "grad_norm": 0.4251273572444916,
      "learning_rate": 8.895547540607383e-06,
      "loss": 0.333,
      "step": 15125
    },
    {
      "epoch": 0.6941397824790051,
      "grad_norm": 0.4681026041507721,
      "learning_rate": 8.895393830604152e-06,
      "loss": 0.4454,
      "step": 15126
    },
    {
      "epoch": 0.6941856729842596,
      "grad_norm": 0.4373570382595062,
      "learning_rate": 8.895240111233734e-06,
      "loss": 0.4031,
      "step": 15127
    },
    {
      "epoch": 0.694231563489514,
      "grad_norm": 0.4509614408016205,
      "learning_rate": 8.89508638249649e-06,
      "loss": 0.4005,
      "step": 15128
    },
    {
      "epoch": 0.6942774539947685,
      "grad_norm": 0.4581938683986664,
      "learning_rate": 8.89493264439279e-06,
      "loss": 0.4054,
      "step": 15129
    },
    {
      "epoch": 0.694323344500023,
      "grad_norm": 0.4731428921222687,
      "learning_rate": 8.894778896923011e-06,
      "loss": 0.4397,
      "step": 15130
    },
    {
      "epoch": 0.6943692350052774,
      "grad_norm": 0.4233674108982086,
      "learning_rate": 8.894625140087514e-06,
      "loss": 0.3535,
      "step": 15131
    },
    {
      "epoch": 0.6944151255105319,
      "grad_norm": 0.4344710409641266,
      "learning_rate": 8.894471373886675e-06,
      "loss": 0.3771,
      "step": 15132
    },
    {
      "epoch": 0.6944610160157864,
      "grad_norm": 0.45964351296424866,
      "learning_rate": 8.89431759832086e-06,
      "loss": 0.4419,
      "step": 15133
    },
    {
      "epoch": 0.6945069065210407,
      "grad_norm": 0.4150189459323883,
      "learning_rate": 8.894163813390441e-06,
      "loss": 0.335,
      "step": 15134
    },
    {
      "epoch": 0.6945527970262952,
      "grad_norm": 0.4308984577655792,
      "learning_rate": 8.89401001909579e-06,
      "loss": 0.3139,
      "step": 15135
    },
    {
      "epoch": 0.6945986875315497,
      "grad_norm": 0.48209941387176514,
      "learning_rate": 8.89385621543727e-06,
      "loss": 0.4233,
      "step": 15136
    },
    {
      "epoch": 0.6946445780368042,
      "grad_norm": 0.4352678954601288,
      "learning_rate": 8.893702402415255e-06,
      "loss": 0.3792,
      "step": 15137
    },
    {
      "epoch": 0.6946904685420586,
      "grad_norm": 0.4760005474090576,
      "learning_rate": 8.893548580030116e-06,
      "loss": 0.4458,
      "step": 15138
    },
    {
      "epoch": 0.6947363590473131,
      "grad_norm": 0.46707314252853394,
      "learning_rate": 8.89339474828222e-06,
      "loss": 0.4116,
      "step": 15139
    },
    {
      "epoch": 0.6947822495525676,
      "grad_norm": 0.44792258739471436,
      "learning_rate": 8.89324090717194e-06,
      "loss": 0.4397,
      "step": 15140
    },
    {
      "epoch": 0.694828140057822,
      "grad_norm": 0.4552158713340759,
      "learning_rate": 8.893087056699643e-06,
      "loss": 0.403,
      "step": 15141
    },
    {
      "epoch": 0.6948740305630765,
      "grad_norm": 0.4578076899051666,
      "learning_rate": 8.892933196865701e-06,
      "loss": 0.3688,
      "step": 15142
    },
    {
      "epoch": 0.694919921068331,
      "grad_norm": 0.45841383934020996,
      "learning_rate": 8.892779327670484e-06,
      "loss": 0.3899,
      "step": 15143
    },
    {
      "epoch": 0.6949658115735854,
      "grad_norm": 0.43301504850387573,
      "learning_rate": 8.89262544911436e-06,
      "loss": 0.3611,
      "step": 15144
    },
    {
      "epoch": 0.6950117020788399,
      "grad_norm": 0.4400251805782318,
      "learning_rate": 8.892471561197702e-06,
      "loss": 0.3719,
      "step": 15145
    },
    {
      "epoch": 0.6950575925840944,
      "grad_norm": 0.4428766965866089,
      "learning_rate": 8.892317663920878e-06,
      "loss": 0.4023,
      "step": 15146
    },
    {
      "epoch": 0.6951034830893488,
      "grad_norm": 0.4433557689189911,
      "learning_rate": 8.892163757284257e-06,
      "loss": 0.4037,
      "step": 15147
    },
    {
      "epoch": 0.6951493735946033,
      "grad_norm": 0.44974005222320557,
      "learning_rate": 8.892009841288212e-06,
      "loss": 0.359,
      "step": 15148
    },
    {
      "epoch": 0.6951952640998578,
      "grad_norm": 0.45068299770355225,
      "learning_rate": 8.891855915933112e-06,
      "loss": 0.4228,
      "step": 15149
    },
    {
      "epoch": 0.6952411546051122,
      "grad_norm": 0.4278578758239746,
      "learning_rate": 8.891701981219325e-06,
      "loss": 0.3954,
      "step": 15150
    },
    {
      "epoch": 0.6952870451103667,
      "grad_norm": 0.4306012690067291,
      "learning_rate": 8.891548037147224e-06,
      "loss": 0.3567,
      "step": 15151
    },
    {
      "epoch": 0.6953329356156212,
      "grad_norm": 0.4072541892528534,
      "learning_rate": 8.891394083717177e-06,
      "loss": 0.3252,
      "step": 15152
    },
    {
      "epoch": 0.6953788261208755,
      "grad_norm": 0.44118452072143555,
      "learning_rate": 8.891240120929556e-06,
      "loss": 0.3963,
      "step": 15153
    },
    {
      "epoch": 0.69542471662613,
      "grad_norm": 0.4333735704421997,
      "learning_rate": 8.891086148784732e-06,
      "loss": 0.3523,
      "step": 15154
    },
    {
      "epoch": 0.6954706071313845,
      "grad_norm": 0.4640464782714844,
      "learning_rate": 8.890932167283074e-06,
      "loss": 0.4167,
      "step": 15155
    },
    {
      "epoch": 0.695516497636639,
      "grad_norm": 0.46337980031967163,
      "learning_rate": 8.89077817642495e-06,
      "loss": 0.436,
      "step": 15156
    },
    {
      "epoch": 0.6955623881418934,
      "grad_norm": 0.47425082325935364,
      "learning_rate": 8.890624176210734e-06,
      "loss": 0.436,
      "step": 15157
    },
    {
      "epoch": 0.6956082786471479,
      "grad_norm": 0.46500587463378906,
      "learning_rate": 8.890470166640795e-06,
      "loss": 0.4316,
      "step": 15158
    },
    {
      "epoch": 0.6956541691524024,
      "grad_norm": 0.48668068647384644,
      "learning_rate": 8.890316147715501e-06,
      "loss": 0.4771,
      "step": 15159
    },
    {
      "epoch": 0.6957000596576568,
      "grad_norm": 0.4826662838459015,
      "learning_rate": 8.890162119435226e-06,
      "loss": 0.4572,
      "step": 15160
    },
    {
      "epoch": 0.6957459501629113,
      "grad_norm": 0.4761209487915039,
      "learning_rate": 8.890008081800338e-06,
      "loss": 0.4245,
      "step": 15161
    },
    {
      "epoch": 0.6957918406681658,
      "grad_norm": 0.5205764770507812,
      "learning_rate": 8.889854034811211e-06,
      "loss": 0.533,
      "step": 15162
    },
    {
      "epoch": 0.6958377311734202,
      "grad_norm": 0.436120867729187,
      "learning_rate": 8.889699978468211e-06,
      "loss": 0.3431,
      "step": 15163
    },
    {
      "epoch": 0.6958836216786747,
      "grad_norm": 0.4808082580566406,
      "learning_rate": 8.88954591277171e-06,
      "loss": 0.4243,
      "step": 15164
    },
    {
      "epoch": 0.6959295121839292,
      "grad_norm": 0.41901248693466187,
      "learning_rate": 8.889391837722079e-06,
      "loss": 0.3097,
      "step": 15165
    },
    {
      "epoch": 0.6959754026891836,
      "grad_norm": 0.4621783494949341,
      "learning_rate": 8.889237753319688e-06,
      "loss": 0.4003,
      "step": 15166
    },
    {
      "epoch": 0.6960212931944381,
      "grad_norm": 0.4483255445957184,
      "learning_rate": 8.889083659564906e-06,
      "loss": 0.3621,
      "step": 15167
    },
    {
      "epoch": 0.6960671836996926,
      "grad_norm": 0.4626440703868866,
      "learning_rate": 8.888929556458108e-06,
      "loss": 0.3907,
      "step": 15168
    },
    {
      "epoch": 0.696113074204947,
      "grad_norm": 0.4225757420063019,
      "learning_rate": 8.88877544399966e-06,
      "loss": 0.3679,
      "step": 15169
    },
    {
      "epoch": 0.6961589647102014,
      "grad_norm": 0.4677918553352356,
      "learning_rate": 8.888621322189934e-06,
      "loss": 0.4134,
      "step": 15170
    },
    {
      "epoch": 0.6962048552154559,
      "grad_norm": 0.409432977437973,
      "learning_rate": 8.888467191029304e-06,
      "loss": 0.3283,
      "step": 15171
    },
    {
      "epoch": 0.6962507457207104,
      "grad_norm": 0.4502343237400055,
      "learning_rate": 8.888313050518134e-06,
      "loss": 0.4071,
      "step": 15172
    },
    {
      "epoch": 0.6962966362259648,
      "grad_norm": 0.45514896512031555,
      "learning_rate": 8.8881589006568e-06,
      "loss": 0.3861,
      "step": 15173
    },
    {
      "epoch": 0.6963425267312193,
      "grad_norm": 0.40824124217033386,
      "learning_rate": 8.88800474144567e-06,
      "loss": 0.3174,
      "step": 15174
    },
    {
      "epoch": 0.6963884172364738,
      "grad_norm": 0.4617963135242462,
      "learning_rate": 8.887850572885118e-06,
      "loss": 0.4366,
      "step": 15175
    },
    {
      "epoch": 0.6964343077417282,
      "grad_norm": 0.4800441861152649,
      "learning_rate": 8.88769639497551e-06,
      "loss": 0.4868,
      "step": 15176
    },
    {
      "epoch": 0.6964801982469827,
      "grad_norm": 0.4768453538417816,
      "learning_rate": 8.887542207717221e-06,
      "loss": 0.5377,
      "step": 15177
    },
    {
      "epoch": 0.6965260887522372,
      "grad_norm": 0.4700711965560913,
      "learning_rate": 8.887388011110619e-06,
      "loss": 0.3582,
      "step": 15178
    },
    {
      "epoch": 0.6965719792574916,
      "grad_norm": 0.45769718289375305,
      "learning_rate": 8.887233805156076e-06,
      "loss": 0.4358,
      "step": 15179
    },
    {
      "epoch": 0.6966178697627461,
      "grad_norm": 0.39754197001457214,
      "learning_rate": 8.887079589853962e-06,
      "loss": 0.2722,
      "step": 15180
    },
    {
      "epoch": 0.6966637602680006,
      "grad_norm": 0.4777667224407196,
      "learning_rate": 8.886925365204648e-06,
      "loss": 0.3749,
      "step": 15181
    },
    {
      "epoch": 0.696709650773255,
      "grad_norm": 0.49719342589378357,
      "learning_rate": 8.886771131208507e-06,
      "loss": 0.4603,
      "step": 15182
    },
    {
      "epoch": 0.6967555412785095,
      "grad_norm": 0.4033268988132477,
      "learning_rate": 8.886616887865906e-06,
      "loss": 0.271,
      "step": 15183
    },
    {
      "epoch": 0.696801431783764,
      "grad_norm": 0.4623813033103943,
      "learning_rate": 8.88646263517722e-06,
      "loss": 0.4315,
      "step": 15184
    },
    {
      "epoch": 0.6968473222890184,
      "grad_norm": 0.4295229911804199,
      "learning_rate": 8.886308373142816e-06,
      "loss": 0.3273,
      "step": 15185
    },
    {
      "epoch": 0.6968932127942729,
      "grad_norm": 0.45323947072029114,
      "learning_rate": 8.886154101763069e-06,
      "loss": 0.3642,
      "step": 15186
    },
    {
      "epoch": 0.6969391032995274,
      "grad_norm": 0.46795228123664856,
      "learning_rate": 8.885999821038346e-06,
      "loss": 0.4505,
      "step": 15187
    },
    {
      "epoch": 0.6969849938047817,
      "grad_norm": 0.4585321247577667,
      "learning_rate": 8.885845530969022e-06,
      "loss": 0.3842,
      "step": 15188
    },
    {
      "epoch": 0.6970308843100362,
      "grad_norm": 0.45521658658981323,
      "learning_rate": 8.885691231555464e-06,
      "loss": 0.3854,
      "step": 15189
    },
    {
      "epoch": 0.6970767748152907,
      "grad_norm": 0.4429137110710144,
      "learning_rate": 8.885536922798045e-06,
      "loss": 0.4124,
      "step": 15190
    },
    {
      "epoch": 0.6971226653205452,
      "grad_norm": 0.4948519468307495,
      "learning_rate": 8.885382604697137e-06,
      "loss": 0.5098,
      "step": 15191
    },
    {
      "epoch": 0.6971685558257996,
      "grad_norm": 0.42685559391975403,
      "learning_rate": 8.88522827725311e-06,
      "loss": 0.3336,
      "step": 15192
    },
    {
      "epoch": 0.6972144463310541,
      "grad_norm": 0.44985103607177734,
      "learning_rate": 8.885073940466336e-06,
      "loss": 0.327,
      "step": 15193
    },
    {
      "epoch": 0.6972603368363086,
      "grad_norm": 0.47658848762512207,
      "learning_rate": 8.884919594337183e-06,
      "loss": 0.4124,
      "step": 15194
    },
    {
      "epoch": 0.697306227341563,
      "grad_norm": 0.4722121059894562,
      "learning_rate": 8.884765238866026e-06,
      "loss": 0.4603,
      "step": 15195
    },
    {
      "epoch": 0.6973521178468175,
      "grad_norm": 0.5297197103500366,
      "learning_rate": 8.884610874053234e-06,
      "loss": 0.4366,
      "step": 15196
    },
    {
      "epoch": 0.697398008352072,
      "grad_norm": 0.4262726604938507,
      "learning_rate": 8.884456499899181e-06,
      "loss": 0.3539,
      "step": 15197
    },
    {
      "epoch": 0.6974438988573264,
      "grad_norm": 0.4734591841697693,
      "learning_rate": 8.884302116404232e-06,
      "loss": 0.3998,
      "step": 15198
    },
    {
      "epoch": 0.6974897893625809,
      "grad_norm": 0.4541209936141968,
      "learning_rate": 8.884147723568766e-06,
      "loss": 0.3887,
      "step": 15199
    },
    {
      "epoch": 0.6975356798678354,
      "grad_norm": 0.4840013384819031,
      "learning_rate": 8.88399332139315e-06,
      "loss": 0.4349,
      "step": 15200
    },
    {
      "epoch": 0.6975815703730898,
      "grad_norm": 0.4816952049732208,
      "learning_rate": 8.883838909877756e-06,
      "loss": 0.4119,
      "step": 15201
    },
    {
      "epoch": 0.6976274608783443,
      "grad_norm": 0.40447670221328735,
      "learning_rate": 8.883684489022953e-06,
      "loss": 0.3008,
      "step": 15202
    },
    {
      "epoch": 0.6976733513835988,
      "grad_norm": 0.43527907133102417,
      "learning_rate": 8.883530058829117e-06,
      "loss": 0.3787,
      "step": 15203
    },
    {
      "epoch": 0.6977192418888531,
      "grad_norm": 0.4083555340766907,
      "learning_rate": 8.883375619296615e-06,
      "loss": 0.3129,
      "step": 15204
    },
    {
      "epoch": 0.6977651323941076,
      "grad_norm": 0.43514540791511536,
      "learning_rate": 8.883221170425822e-06,
      "loss": 0.3388,
      "step": 15205
    },
    {
      "epoch": 0.6978110228993621,
      "grad_norm": 0.5058029890060425,
      "learning_rate": 8.883066712217105e-06,
      "loss": 0.4727,
      "step": 15206
    },
    {
      "epoch": 0.6978569134046165,
      "grad_norm": 0.4844582974910736,
      "learning_rate": 8.88291224467084e-06,
      "loss": 0.4164,
      "step": 15207
    },
    {
      "epoch": 0.697902803909871,
      "grad_norm": 0.44629886746406555,
      "learning_rate": 8.882757767787397e-06,
      "loss": 0.3801,
      "step": 15208
    },
    {
      "epoch": 0.6979486944151255,
      "grad_norm": 0.44696423411369324,
      "learning_rate": 8.882603281567147e-06,
      "loss": 0.3628,
      "step": 15209
    },
    {
      "epoch": 0.69799458492038,
      "grad_norm": 0.4221518337726593,
      "learning_rate": 8.88244878601046e-06,
      "loss": 0.3442,
      "step": 15210
    },
    {
      "epoch": 0.6980404754256344,
      "grad_norm": 0.454863041639328,
      "learning_rate": 8.882294281117711e-06,
      "loss": 0.432,
      "step": 15211
    },
    {
      "epoch": 0.6980863659308889,
      "grad_norm": 0.4666585326194763,
      "learning_rate": 8.882139766889269e-06,
      "loss": 0.4061,
      "step": 15212
    },
    {
      "epoch": 0.6981322564361434,
      "grad_norm": 0.5056252479553223,
      "learning_rate": 8.881985243325504e-06,
      "loss": 0.4975,
      "step": 15213
    },
    {
      "epoch": 0.6981781469413978,
      "grad_norm": 0.46398425102233887,
      "learning_rate": 8.881830710426793e-06,
      "loss": 0.4198,
      "step": 15214
    },
    {
      "epoch": 0.6982240374466523,
      "grad_norm": 0.45258086919784546,
      "learning_rate": 8.8816761681935e-06,
      "loss": 0.3783,
      "step": 15215
    },
    {
      "epoch": 0.6982699279519068,
      "grad_norm": 0.46092015504837036,
      "learning_rate": 8.881521616626005e-06,
      "loss": 0.3925,
      "step": 15216
    },
    {
      "epoch": 0.6983158184571612,
      "grad_norm": 0.4350115954875946,
      "learning_rate": 8.881367055724674e-06,
      "loss": 0.3644,
      "step": 15217
    },
    {
      "epoch": 0.6983617089624157,
      "grad_norm": 0.4020456075668335,
      "learning_rate": 8.88121248548988e-06,
      "loss": 0.3162,
      "step": 15218
    },
    {
      "epoch": 0.6984075994676702,
      "grad_norm": 0.44675400853157043,
      "learning_rate": 8.881057905921994e-06,
      "loss": 0.4137,
      "step": 15219
    },
    {
      "epoch": 0.6984534899729246,
      "grad_norm": 0.4497157037258148,
      "learning_rate": 8.88090331702139e-06,
      "loss": 0.3478,
      "step": 15220
    },
    {
      "epoch": 0.698499380478179,
      "grad_norm": 0.5763722062110901,
      "learning_rate": 8.88074871878844e-06,
      "loss": 0.3754,
      "step": 15221
    },
    {
      "epoch": 0.6985452709834336,
      "grad_norm": 0.4855414628982544,
      "learning_rate": 8.880594111223511e-06,
      "loss": 0.3975,
      "step": 15222
    },
    {
      "epoch": 0.6985911614886879,
      "grad_norm": 0.5418866276741028,
      "learning_rate": 8.880439494326978e-06,
      "loss": 0.3314,
      "step": 15223
    },
    {
      "epoch": 0.6986370519939424,
      "grad_norm": 0.49025633931159973,
      "learning_rate": 8.880284868099216e-06,
      "loss": 0.5026,
      "step": 15224
    },
    {
      "epoch": 0.6986829424991969,
      "grad_norm": 0.5110155940055847,
      "learning_rate": 8.88013023254059e-06,
      "loss": 0.4984,
      "step": 15225
    },
    {
      "epoch": 0.6987288330044514,
      "grad_norm": 0.48928919434547424,
      "learning_rate": 8.879975587651477e-06,
      "loss": 0.4981,
      "step": 15226
    },
    {
      "epoch": 0.6987747235097058,
      "grad_norm": 0.4417930543422699,
      "learning_rate": 8.879820933432247e-06,
      "loss": 0.3299,
      "step": 15227
    },
    {
      "epoch": 0.6988206140149603,
      "grad_norm": 0.4514995813369751,
      "learning_rate": 8.879666269883272e-06,
      "loss": 0.3651,
      "step": 15228
    },
    {
      "epoch": 0.6988665045202148,
      "grad_norm": 0.4827274680137634,
      "learning_rate": 8.879511597004924e-06,
      "loss": 0.4159,
      "step": 15229
    },
    {
      "epoch": 0.6989123950254692,
      "grad_norm": 0.46608299016952515,
      "learning_rate": 8.879356914797577e-06,
      "loss": 0.4027,
      "step": 15230
    },
    {
      "epoch": 0.6989582855307237,
      "grad_norm": 0.4592164158821106,
      "learning_rate": 8.879202223261598e-06,
      "loss": 0.3725,
      "step": 15231
    },
    {
      "epoch": 0.6990041760359782,
      "grad_norm": 0.4692539870738983,
      "learning_rate": 8.879047522397365e-06,
      "loss": 0.4422,
      "step": 15232
    },
    {
      "epoch": 0.6990500665412326,
      "grad_norm": 0.4496704041957855,
      "learning_rate": 8.878892812205245e-06,
      "loss": 0.3631,
      "step": 15233
    },
    {
      "epoch": 0.6990959570464871,
      "grad_norm": 0.4508979022502899,
      "learning_rate": 8.878738092685612e-06,
      "loss": 0.415,
      "step": 15234
    },
    {
      "epoch": 0.6991418475517416,
      "grad_norm": 0.40440821647644043,
      "learning_rate": 8.878583363838837e-06,
      "loss": 0.3145,
      "step": 15235
    },
    {
      "epoch": 0.699187738056996,
      "grad_norm": 0.48993462324142456,
      "learning_rate": 8.878428625665296e-06,
      "loss": 0.4803,
      "step": 15236
    },
    {
      "epoch": 0.6992336285622505,
      "grad_norm": 0.45309996604919434,
      "learning_rate": 8.878273878165356e-06,
      "loss": 0.3798,
      "step": 15237
    },
    {
      "epoch": 0.699279519067505,
      "grad_norm": 0.43323734402656555,
      "learning_rate": 8.878119121339392e-06,
      "loss": 0.3625,
      "step": 15238
    },
    {
      "epoch": 0.6993254095727593,
      "grad_norm": 0.45553144812583923,
      "learning_rate": 8.877964355187774e-06,
      "loss": 0.3906,
      "step": 15239
    },
    {
      "epoch": 0.6993713000780138,
      "grad_norm": 0.43476083874702454,
      "learning_rate": 8.877809579710878e-06,
      "loss": 0.3584,
      "step": 15240
    },
    {
      "epoch": 0.6994171905832683,
      "grad_norm": 0.4173594117164612,
      "learning_rate": 8.877654794909075e-06,
      "loss": 0.3464,
      "step": 15241
    },
    {
      "epoch": 0.6994630810885227,
      "grad_norm": 0.4591556787490845,
      "learning_rate": 8.877500000782732e-06,
      "loss": 0.4078,
      "step": 15242
    },
    {
      "epoch": 0.6995089715937772,
      "grad_norm": 0.4686826169490814,
      "learning_rate": 8.877345197332228e-06,
      "loss": 0.4042,
      "step": 15243
    },
    {
      "epoch": 0.6995548620990317,
      "grad_norm": 0.44457918405532837,
      "learning_rate": 8.877190384557931e-06,
      "loss": 0.3833,
      "step": 15244
    },
    {
      "epoch": 0.6996007526042862,
      "grad_norm": 0.4244305491447449,
      "learning_rate": 8.877035562460217e-06,
      "loss": 0.3153,
      "step": 15245
    },
    {
      "epoch": 0.6996466431095406,
      "grad_norm": 0.45465418696403503,
      "learning_rate": 8.876880731039454e-06,
      "loss": 0.3434,
      "step": 15246
    },
    {
      "epoch": 0.6996925336147951,
      "grad_norm": 0.48485392332077026,
      "learning_rate": 8.876725890296017e-06,
      "loss": 0.395,
      "step": 15247
    },
    {
      "epoch": 0.6997384241200496,
      "grad_norm": 0.42503198981285095,
      "learning_rate": 8.876571040230277e-06,
      "loss": 0.3792,
      "step": 15248
    },
    {
      "epoch": 0.699784314625304,
      "grad_norm": 0.45768308639526367,
      "learning_rate": 8.876416180842609e-06,
      "loss": 0.4183,
      "step": 15249
    },
    {
      "epoch": 0.6998302051305585,
      "grad_norm": 0.4520810842514038,
      "learning_rate": 8.876261312133383e-06,
      "loss": 0.423,
      "step": 15250
    },
    {
      "epoch": 0.699876095635813,
      "grad_norm": 0.4430385231971741,
      "learning_rate": 8.876106434102972e-06,
      "loss": 0.3566,
      "step": 15251
    },
    {
      "epoch": 0.6999219861410674,
      "grad_norm": 0.4753483235836029,
      "learning_rate": 8.875951546751746e-06,
      "loss": 0.4526,
      "step": 15252
    },
    {
      "epoch": 0.6999678766463219,
      "grad_norm": 0.41026216745376587,
      "learning_rate": 8.875796650080082e-06,
      "loss": 0.3403,
      "step": 15253
    },
    {
      "epoch": 0.7000137671515764,
      "grad_norm": 0.47939690947532654,
      "learning_rate": 8.87564174408835e-06,
      "loss": 0.4367,
      "step": 15254
    },
    {
      "epoch": 0.7000596576568308,
      "grad_norm": 0.5063282251358032,
      "learning_rate": 8.875486828776922e-06,
      "loss": 0.4948,
      "step": 15255
    },
    {
      "epoch": 0.7001055481620853,
      "grad_norm": 0.4945315420627594,
      "learning_rate": 8.875331904146171e-06,
      "loss": 0.4229,
      "step": 15256
    },
    {
      "epoch": 0.7001514386673398,
      "grad_norm": 0.442889541387558,
      "learning_rate": 8.875176970196473e-06,
      "loss": 0.3531,
      "step": 15257
    },
    {
      "epoch": 0.7001973291725941,
      "grad_norm": 0.4211094081401825,
      "learning_rate": 8.875022026928195e-06,
      "loss": 0.3241,
      "step": 15258
    },
    {
      "epoch": 0.7002432196778486,
      "grad_norm": 0.44450193643569946,
      "learning_rate": 8.874867074341712e-06,
      "loss": 0.3698,
      "step": 15259
    },
    {
      "epoch": 0.7002891101831031,
      "grad_norm": 0.470569372177124,
      "learning_rate": 8.874712112437397e-06,
      "loss": 0.4297,
      "step": 15260
    },
    {
      "epoch": 0.7003350006883576,
      "grad_norm": 0.4490543305873871,
      "learning_rate": 8.874557141215622e-06,
      "loss": 0.3704,
      "step": 15261
    },
    {
      "epoch": 0.700380891193612,
      "grad_norm": 0.445046991109848,
      "learning_rate": 8.874402160676761e-06,
      "loss": 0.4095,
      "step": 15262
    },
    {
      "epoch": 0.7004267816988665,
      "grad_norm": 0.46994394063949585,
      "learning_rate": 8.874247170821184e-06,
      "loss": 0.4145,
      "step": 15263
    },
    {
      "epoch": 0.700472672204121,
      "grad_norm": 0.4204995632171631,
      "learning_rate": 8.874092171649267e-06,
      "loss": 0.3051,
      "step": 15264
    },
    {
      "epoch": 0.7005185627093754,
      "grad_norm": 0.44294098019599915,
      "learning_rate": 8.87393716316138e-06,
      "loss": 0.3617,
      "step": 15265
    },
    {
      "epoch": 0.7005644532146299,
      "grad_norm": 0.4946003556251526,
      "learning_rate": 8.873782145357896e-06,
      "loss": 0.4141,
      "step": 15266
    },
    {
      "epoch": 0.7006103437198844,
      "grad_norm": 0.4453360438346863,
      "learning_rate": 8.873627118239189e-06,
      "loss": 0.4096,
      "step": 15267
    },
    {
      "epoch": 0.7006562342251388,
      "grad_norm": 0.4466598331928253,
      "learning_rate": 8.873472081805631e-06,
      "loss": 0.4037,
      "step": 15268
    },
    {
      "epoch": 0.7007021247303933,
      "grad_norm": 0.5217832326889038,
      "learning_rate": 8.873317036057598e-06,
      "loss": 0.5215,
      "step": 15269
    },
    {
      "epoch": 0.7007480152356478,
      "grad_norm": 0.45922544598579407,
      "learning_rate": 8.873161980995457e-06,
      "loss": 0.399,
      "step": 15270
    },
    {
      "epoch": 0.7007939057409022,
      "grad_norm": 0.4928055703639984,
      "learning_rate": 8.873006916619585e-06,
      "loss": 0.388,
      "step": 15271
    },
    {
      "epoch": 0.7008397962461567,
      "grad_norm": 0.4843306839466095,
      "learning_rate": 8.872851842930354e-06,
      "loss": 0.4444,
      "step": 15272
    },
    {
      "epoch": 0.7008856867514112,
      "grad_norm": 0.4655596911907196,
      "learning_rate": 8.872696759928134e-06,
      "loss": 0.3961,
      "step": 15273
    },
    {
      "epoch": 0.7009315772566656,
      "grad_norm": 0.4409341812133789,
      "learning_rate": 8.872541667613304e-06,
      "loss": 0.362,
      "step": 15274
    },
    {
      "epoch": 0.70097746776192,
      "grad_norm": 0.4653756320476532,
      "learning_rate": 8.872386565986231e-06,
      "loss": 0.4159,
      "step": 15275
    },
    {
      "epoch": 0.7010233582671745,
      "grad_norm": 0.4453827142715454,
      "learning_rate": 8.872231455047293e-06,
      "loss": 0.367,
      "step": 15276
    },
    {
      "epoch": 0.7010692487724289,
      "grad_norm": 0.5066048502922058,
      "learning_rate": 8.872076334796859e-06,
      "loss": 0.5074,
      "step": 15277
    },
    {
      "epoch": 0.7011151392776834,
      "grad_norm": 0.4510207176208496,
      "learning_rate": 8.871921205235304e-06,
      "loss": 0.4183,
      "step": 15278
    },
    {
      "epoch": 0.7011610297829379,
      "grad_norm": 0.4590039551258087,
      "learning_rate": 8.871766066363e-06,
      "loss": 0.4264,
      "step": 15279
    },
    {
      "epoch": 0.7012069202881924,
      "grad_norm": 0.4871574938297272,
      "learning_rate": 8.87161091818032e-06,
      "loss": 0.4082,
      "step": 15280
    },
    {
      "epoch": 0.7012528107934468,
      "grad_norm": 0.447847843170166,
      "learning_rate": 8.871455760687637e-06,
      "loss": 0.3883,
      "step": 15281
    },
    {
      "epoch": 0.7012987012987013,
      "grad_norm": 0.44927653670310974,
      "learning_rate": 8.871300593885327e-06,
      "loss": 0.416,
      "step": 15282
    },
    {
      "epoch": 0.7013445918039558,
      "grad_norm": 0.4417235553264618,
      "learning_rate": 8.871145417773758e-06,
      "loss": 0.3578,
      "step": 15283
    },
    {
      "epoch": 0.7013904823092102,
      "grad_norm": 0.4543023407459259,
      "learning_rate": 8.870990232353309e-06,
      "loss": 0.4264,
      "step": 15284
    },
    {
      "epoch": 0.7014363728144647,
      "grad_norm": 0.47682252526283264,
      "learning_rate": 8.87083503762435e-06,
      "loss": 0.4168,
      "step": 15285
    },
    {
      "epoch": 0.7014822633197192,
      "grad_norm": 0.49393871426582336,
      "learning_rate": 8.870679833587253e-06,
      "loss": 0.3984,
      "step": 15286
    },
    {
      "epoch": 0.7015281538249736,
      "grad_norm": 0.4497290253639221,
      "learning_rate": 8.870524620242393e-06,
      "loss": 0.3731,
      "step": 15287
    },
    {
      "epoch": 0.7015740443302281,
      "grad_norm": 0.5179337859153748,
      "learning_rate": 8.870369397590144e-06,
      "loss": 0.5057,
      "step": 15288
    },
    {
      "epoch": 0.7016199348354826,
      "grad_norm": 0.45535415410995483,
      "learning_rate": 8.870214165630878e-06,
      "loss": 0.3953,
      "step": 15289
    },
    {
      "epoch": 0.701665825340737,
      "grad_norm": 0.4394325613975525,
      "learning_rate": 8.870058924364967e-06,
      "loss": 0.3673,
      "step": 15290
    },
    {
      "epoch": 0.7017117158459915,
      "grad_norm": 0.4603269100189209,
      "learning_rate": 8.869903673792786e-06,
      "loss": 0.4307,
      "step": 15291
    },
    {
      "epoch": 0.701757606351246,
      "grad_norm": 0.43877696990966797,
      "learning_rate": 8.869748413914707e-06,
      "loss": 0.39,
      "step": 15292
    },
    {
      "epoch": 0.7018034968565003,
      "grad_norm": 0.45589444041252136,
      "learning_rate": 8.869593144731107e-06,
      "loss": 0.3915,
      "step": 15293
    },
    {
      "epoch": 0.7018493873617548,
      "grad_norm": 0.48724251985549927,
      "learning_rate": 8.869437866242355e-06,
      "loss": 0.5122,
      "step": 15294
    },
    {
      "epoch": 0.7018952778670093,
      "grad_norm": 0.43473827838897705,
      "learning_rate": 8.869282578448828e-06,
      "loss": 0.3901,
      "step": 15295
    },
    {
      "epoch": 0.7019411683722637,
      "grad_norm": 0.49126118421554565,
      "learning_rate": 8.869127281350896e-06,
      "loss": 0.5008,
      "step": 15296
    },
    {
      "epoch": 0.7019870588775182,
      "grad_norm": 0.4684297442436218,
      "learning_rate": 8.868971974948932e-06,
      "loss": 0.3945,
      "step": 15297
    },
    {
      "epoch": 0.7020329493827727,
      "grad_norm": 0.46284162998199463,
      "learning_rate": 8.868816659243316e-06,
      "loss": 0.3986,
      "step": 15298
    },
    {
      "epoch": 0.7020788398880272,
      "grad_norm": 0.47493064403533936,
      "learning_rate": 8.868661334234414e-06,
      "loss": 0.4379,
      "step": 15299
    },
    {
      "epoch": 0.7021247303932816,
      "grad_norm": 0.4944457411766052,
      "learning_rate": 8.868505999922603e-06,
      "loss": 0.444,
      "step": 15300
    },
    {
      "epoch": 0.7021706208985361,
      "grad_norm": 0.4436239004135132,
      "learning_rate": 8.868350656308256e-06,
      "loss": 0.3766,
      "step": 15301
    },
    {
      "epoch": 0.7022165114037906,
      "grad_norm": 0.4351639151573181,
      "learning_rate": 8.868195303391747e-06,
      "loss": 0.3513,
      "step": 15302
    },
    {
      "epoch": 0.702262401909045,
      "grad_norm": 0.4514792263507843,
      "learning_rate": 8.868039941173448e-06,
      "loss": 0.3806,
      "step": 15303
    },
    {
      "epoch": 0.7023082924142995,
      "grad_norm": 0.47317075729370117,
      "learning_rate": 8.867884569653735e-06,
      "loss": 0.4121,
      "step": 15304
    },
    {
      "epoch": 0.702354182919554,
      "grad_norm": 0.6164528131484985,
      "learning_rate": 8.867729188832981e-06,
      "loss": 0.4767,
      "step": 15305
    },
    {
      "epoch": 0.7024000734248084,
      "grad_norm": 0.42011305689811707,
      "learning_rate": 8.867573798711558e-06,
      "loss": 0.3357,
      "step": 15306
    },
    {
      "epoch": 0.7024459639300629,
      "grad_norm": 0.4813268780708313,
      "learning_rate": 8.86741839928984e-06,
      "loss": 0.4818,
      "step": 15307
    },
    {
      "epoch": 0.7024918544353174,
      "grad_norm": 0.47319892048835754,
      "learning_rate": 8.867262990568201e-06,
      "loss": 0.4657,
      "step": 15308
    },
    {
      "epoch": 0.7025377449405718,
      "grad_norm": 0.47796881198883057,
      "learning_rate": 8.867107572547017e-06,
      "loss": 0.4507,
      "step": 15309
    },
    {
      "epoch": 0.7025836354458262,
      "grad_norm": 0.45318812131881714,
      "learning_rate": 8.866952145226659e-06,
      "loss": 0.3631,
      "step": 15310
    },
    {
      "epoch": 0.7026295259510807,
      "grad_norm": 0.4498036503791809,
      "learning_rate": 8.8667967086075e-06,
      "loss": 0.3838,
      "step": 15311
    },
    {
      "epoch": 0.7026754164563351,
      "grad_norm": 0.4448075592517853,
      "learning_rate": 8.866641262689918e-06,
      "loss": 0.3628,
      "step": 15312
    },
    {
      "epoch": 0.7027213069615896,
      "grad_norm": 0.4660280644893646,
      "learning_rate": 8.866485807474282e-06,
      "loss": 0.3896,
      "step": 15313
    },
    {
      "epoch": 0.7027671974668441,
      "grad_norm": 0.41872724890708923,
      "learning_rate": 8.86633034296097e-06,
      "loss": 0.3136,
      "step": 15314
    },
    {
      "epoch": 0.7028130879720986,
      "grad_norm": 0.4663838744163513,
      "learning_rate": 8.866174869150352e-06,
      "loss": 0.3866,
      "step": 15315
    },
    {
      "epoch": 0.702858978477353,
      "grad_norm": 0.43932923674583435,
      "learning_rate": 8.866019386042804e-06,
      "loss": 0.3628,
      "step": 15316
    },
    {
      "epoch": 0.7029048689826075,
      "grad_norm": 0.47448521852493286,
      "learning_rate": 8.865863893638699e-06,
      "loss": 0.4381,
      "step": 15317
    },
    {
      "epoch": 0.702950759487862,
      "grad_norm": 0.4701744318008423,
      "learning_rate": 8.865708391938412e-06,
      "loss": 0.4104,
      "step": 15318
    },
    {
      "epoch": 0.7029966499931164,
      "grad_norm": 0.4675033390522003,
      "learning_rate": 8.865552880942316e-06,
      "loss": 0.4181,
      "step": 15319
    },
    {
      "epoch": 0.7030425404983709,
      "grad_norm": 0.4824400842189789,
      "learning_rate": 8.865397360650786e-06,
      "loss": 0.4482,
      "step": 15320
    },
    {
      "epoch": 0.7030884310036254,
      "grad_norm": 0.40212276577949524,
      "learning_rate": 8.865241831064196e-06,
      "loss": 0.3151,
      "step": 15321
    },
    {
      "epoch": 0.7031343215088798,
      "grad_norm": 0.47795942425727844,
      "learning_rate": 8.865086292182917e-06,
      "loss": 0.3506,
      "step": 15322
    },
    {
      "epoch": 0.7031802120141343,
      "grad_norm": 0.4723162353038788,
      "learning_rate": 8.864930744007328e-06,
      "loss": 0.4514,
      "step": 15323
    },
    {
      "epoch": 0.7032261025193888,
      "grad_norm": 0.46392154693603516,
      "learning_rate": 8.864775186537797e-06,
      "loss": 0.4163,
      "step": 15324
    },
    {
      "epoch": 0.7032719930246432,
      "grad_norm": 0.44807443022727966,
      "learning_rate": 8.864619619774706e-06,
      "loss": 0.4158,
      "step": 15325
    },
    {
      "epoch": 0.7033178835298977,
      "grad_norm": 0.4523496627807617,
      "learning_rate": 8.864464043718421e-06,
      "loss": 0.3223,
      "step": 15326
    },
    {
      "epoch": 0.7033637740351522,
      "grad_norm": 0.48893532156944275,
      "learning_rate": 8.864308458369319e-06,
      "loss": 0.4379,
      "step": 15327
    },
    {
      "epoch": 0.7034096645404065,
      "grad_norm": 0.42681941390037537,
      "learning_rate": 8.864152863727778e-06,
      "loss": 0.3116,
      "step": 15328
    },
    {
      "epoch": 0.703455555045661,
      "grad_norm": 0.4386947453022003,
      "learning_rate": 8.863997259794167e-06,
      "loss": 0.3552,
      "step": 15329
    },
    {
      "epoch": 0.7035014455509155,
      "grad_norm": 0.450403094291687,
      "learning_rate": 8.863841646568863e-06,
      "loss": 0.4357,
      "step": 15330
    },
    {
      "epoch": 0.7035473360561699,
      "grad_norm": 0.45087432861328125,
      "learning_rate": 8.863686024052239e-06,
      "loss": 0.3724,
      "step": 15331
    },
    {
      "epoch": 0.7035932265614244,
      "grad_norm": 0.4423559606075287,
      "learning_rate": 8.863530392244667e-06,
      "loss": 0.3465,
      "step": 15332
    },
    {
      "epoch": 0.7036391170666789,
      "grad_norm": 0.4362533688545227,
      "learning_rate": 8.863374751146527e-06,
      "loss": 0.3275,
      "step": 15333
    },
    {
      "epoch": 0.7036850075719334,
      "grad_norm": 0.46797019243240356,
      "learning_rate": 8.86321910075819e-06,
      "loss": 0.4128,
      "step": 15334
    },
    {
      "epoch": 0.7037308980771878,
      "grad_norm": 0.5074214935302734,
      "learning_rate": 8.863063441080029e-06,
      "loss": 0.5412,
      "step": 15335
    },
    {
      "epoch": 0.7037767885824423,
      "grad_norm": 0.45028167963027954,
      "learning_rate": 8.86290777211242e-06,
      "loss": 0.3533,
      "step": 15336
    },
    {
      "epoch": 0.7038226790876968,
      "grad_norm": 0.44906750321388245,
      "learning_rate": 8.862752093855736e-06,
      "loss": 0.4064,
      "step": 15337
    },
    {
      "epoch": 0.7038685695929512,
      "grad_norm": 0.4328353703022003,
      "learning_rate": 8.862596406310352e-06,
      "loss": 0.3423,
      "step": 15338
    },
    {
      "epoch": 0.7039144600982057,
      "grad_norm": 0.48786914348602295,
      "learning_rate": 8.862440709476646e-06,
      "loss": 0.4428,
      "step": 15339
    },
    {
      "epoch": 0.7039603506034602,
      "grad_norm": 2.4023451805114746,
      "learning_rate": 8.862285003354986e-06,
      "loss": 0.4632,
      "step": 15340
    },
    {
      "epoch": 0.7040062411087146,
      "grad_norm": 0.4542851448059082,
      "learning_rate": 8.86212928794575e-06,
      "loss": 0.425,
      "step": 15341
    },
    {
      "epoch": 0.7040521316139691,
      "grad_norm": 0.4616072177886963,
      "learning_rate": 8.861973563249314e-06,
      "loss": 0.3674,
      "step": 15342
    },
    {
      "epoch": 0.7040980221192236,
      "grad_norm": 0.4486668109893799,
      "learning_rate": 8.86181782926605e-06,
      "loss": 0.3823,
      "step": 15343
    },
    {
      "epoch": 0.704143912624478,
      "grad_norm": 0.45665648579597473,
      "learning_rate": 8.861662085996332e-06,
      "loss": 0.3836,
      "step": 15344
    },
    {
      "epoch": 0.7041898031297325,
      "grad_norm": 0.47215524315834045,
      "learning_rate": 8.861506333440535e-06,
      "loss": 0.44,
      "step": 15345
    },
    {
      "epoch": 0.704235693634987,
      "grad_norm": 0.4670223891735077,
      "learning_rate": 8.861350571599035e-06,
      "loss": 0.416,
      "step": 15346
    },
    {
      "epoch": 0.7042815841402413,
      "grad_norm": 0.4645777642726898,
      "learning_rate": 8.861194800472206e-06,
      "loss": 0.4039,
      "step": 15347
    },
    {
      "epoch": 0.7043274746454958,
      "grad_norm": 0.48237645626068115,
      "learning_rate": 8.86103902006042e-06,
      "loss": 0.4796,
      "step": 15348
    },
    {
      "epoch": 0.7043733651507503,
      "grad_norm": 0.4395120441913605,
      "learning_rate": 8.860883230364054e-06,
      "loss": 0.3401,
      "step": 15349
    },
    {
      "epoch": 0.7044192556560048,
      "grad_norm": 0.4802260100841522,
      "learning_rate": 8.860727431383484e-06,
      "loss": 0.4507,
      "step": 15350
    },
    {
      "epoch": 0.7044651461612592,
      "grad_norm": 0.40397658944129944,
      "learning_rate": 8.860571623119083e-06,
      "loss": 0.3237,
      "step": 15351
    },
    {
      "epoch": 0.7045110366665137,
      "grad_norm": 0.4285014271736145,
      "learning_rate": 8.860415805571225e-06,
      "loss": 0.3334,
      "step": 15352
    },
    {
      "epoch": 0.7045569271717682,
      "grad_norm": 0.5113673210144043,
      "learning_rate": 8.860259978740286e-06,
      "loss": 0.5131,
      "step": 15353
    },
    {
      "epoch": 0.7046028176770226,
      "grad_norm": 0.5180298089981079,
      "learning_rate": 8.86010414262664e-06,
      "loss": 0.4846,
      "step": 15354
    },
    {
      "epoch": 0.7046487081822771,
      "grad_norm": 0.4271198809146881,
      "learning_rate": 8.85994829723066e-06,
      "loss": 0.3729,
      "step": 15355
    },
    {
      "epoch": 0.7046945986875316,
      "grad_norm": 0.4483506977558136,
      "learning_rate": 8.859792442552723e-06,
      "loss": 0.3834,
      "step": 15356
    },
    {
      "epoch": 0.704740489192786,
      "grad_norm": 0.4692017734050751,
      "learning_rate": 8.859636578593204e-06,
      "loss": 0.4107,
      "step": 15357
    },
    {
      "epoch": 0.7047863796980405,
      "grad_norm": 0.48315146565437317,
      "learning_rate": 8.85948070535248e-06,
      "loss": 0.4653,
      "step": 15358
    },
    {
      "epoch": 0.704832270203295,
      "grad_norm": 0.4643354117870331,
      "learning_rate": 8.85932482283092e-06,
      "loss": 0.4135,
      "step": 15359
    },
    {
      "epoch": 0.7048781607085494,
      "grad_norm": 0.5070493221282959,
      "learning_rate": 8.8591689310289e-06,
      "loss": 0.4661,
      "step": 15360
    },
    {
      "epoch": 0.7049240512138039,
      "grad_norm": 0.4862259328365326,
      "learning_rate": 8.8590130299468e-06,
      "loss": 0.4639,
      "step": 15361
    },
    {
      "epoch": 0.7049699417190584,
      "grad_norm": 0.44670534133911133,
      "learning_rate": 8.85885711958499e-06,
      "loss": 0.4108,
      "step": 15362
    },
    {
      "epoch": 0.7050158322243127,
      "grad_norm": 0.47248533368110657,
      "learning_rate": 8.858701199943848e-06,
      "loss": 0.4209,
      "step": 15363
    },
    {
      "epoch": 0.7050617227295672,
      "grad_norm": 0.4051218628883362,
      "learning_rate": 8.858545271023746e-06,
      "loss": 0.3029,
      "step": 15364
    },
    {
      "epoch": 0.7051076132348217,
      "grad_norm": 0.44173547625541687,
      "learning_rate": 8.858389332825062e-06,
      "loss": 0.4628,
      "step": 15365
    },
    {
      "epoch": 0.7051535037400761,
      "grad_norm": 0.42290088534355164,
      "learning_rate": 8.858233385348167e-06,
      "loss": 0.3289,
      "step": 15366
    },
    {
      "epoch": 0.7051993942453306,
      "grad_norm": 0.48536011576652527,
      "learning_rate": 8.85807742859344e-06,
      "loss": 0.4264,
      "step": 15367
    },
    {
      "epoch": 0.7052452847505851,
      "grad_norm": 0.4500810503959656,
      "learning_rate": 8.857921462561255e-06,
      "loss": 0.3948,
      "step": 15368
    },
    {
      "epoch": 0.7052911752558396,
      "grad_norm": 0.43791645765304565,
      "learning_rate": 8.857765487251986e-06,
      "loss": 0.3233,
      "step": 15369
    },
    {
      "epoch": 0.705337065761094,
      "grad_norm": 0.4729611873626709,
      "learning_rate": 8.857609502666006e-06,
      "loss": 0.3852,
      "step": 15370
    },
    {
      "epoch": 0.7053829562663485,
      "grad_norm": 0.41330844163894653,
      "learning_rate": 8.857453508803695e-06,
      "loss": 0.3089,
      "step": 15371
    },
    {
      "epoch": 0.705428846771603,
      "grad_norm": 0.4496743977069855,
      "learning_rate": 8.857297505665426e-06,
      "loss": 0.4099,
      "step": 15372
    },
    {
      "epoch": 0.7054747372768574,
      "grad_norm": 0.438028484582901,
      "learning_rate": 8.857141493251574e-06,
      "loss": 0.3495,
      "step": 15373
    },
    {
      "epoch": 0.7055206277821119,
      "grad_norm": 0.44070401787757874,
      "learning_rate": 8.856985471562514e-06,
      "loss": 0.3634,
      "step": 15374
    },
    {
      "epoch": 0.7055665182873664,
      "grad_norm": 0.46683695912361145,
      "learning_rate": 8.85682944059862e-06,
      "loss": 0.402,
      "step": 15375
    },
    {
      "epoch": 0.7056124087926208,
      "grad_norm": 0.4212355315685272,
      "learning_rate": 8.856673400360269e-06,
      "loss": 0.3403,
      "step": 15376
    },
    {
      "epoch": 0.7056582992978753,
      "grad_norm": 0.46170690655708313,
      "learning_rate": 8.856517350847837e-06,
      "loss": 0.4548,
      "step": 15377
    },
    {
      "epoch": 0.7057041898031298,
      "grad_norm": 0.43015649914741516,
      "learning_rate": 8.856361292061696e-06,
      "loss": 0.3252,
      "step": 15378
    },
    {
      "epoch": 0.7057500803083842,
      "grad_norm": 0.4924994707107544,
      "learning_rate": 8.856205224002224e-06,
      "loss": 0.448,
      "step": 15379
    },
    {
      "epoch": 0.7057959708136387,
      "grad_norm": 0.44125211238861084,
      "learning_rate": 8.856049146669796e-06,
      "loss": 0.3615,
      "step": 15380
    },
    {
      "epoch": 0.7058418613188931,
      "grad_norm": 0.45056939125061035,
      "learning_rate": 8.855893060064787e-06,
      "loss": 0.4104,
      "step": 15381
    },
    {
      "epoch": 0.7058877518241475,
      "grad_norm": 0.42887240648269653,
      "learning_rate": 8.85573696418757e-06,
      "loss": 0.3692,
      "step": 15382
    },
    {
      "epoch": 0.705933642329402,
      "grad_norm": 0.4376792013645172,
      "learning_rate": 8.855580859038524e-06,
      "loss": 0.3633,
      "step": 15383
    },
    {
      "epoch": 0.7059795328346565,
      "grad_norm": 0.4779284596443176,
      "learning_rate": 8.855424744618023e-06,
      "loss": 0.4299,
      "step": 15384
    },
    {
      "epoch": 0.7060254233399109,
      "grad_norm": 0.4681902825832367,
      "learning_rate": 8.855268620926442e-06,
      "loss": 0.379,
      "step": 15385
    },
    {
      "epoch": 0.7060713138451654,
      "grad_norm": 0.4625047743320465,
      "learning_rate": 8.855112487964157e-06,
      "loss": 0.4148,
      "step": 15386
    },
    {
      "epoch": 0.7061172043504199,
      "grad_norm": 0.4290919303894043,
      "learning_rate": 8.854956345731543e-06,
      "loss": 0.3344,
      "step": 15387
    },
    {
      "epoch": 0.7061630948556744,
      "grad_norm": 0.4543091952800751,
      "learning_rate": 8.854800194228977e-06,
      "loss": 0.3843,
      "step": 15388
    },
    {
      "epoch": 0.7062089853609288,
      "grad_norm": 0.4304720461368561,
      "learning_rate": 8.854644033456831e-06,
      "loss": 0.3215,
      "step": 15389
    },
    {
      "epoch": 0.7062548758661833,
      "grad_norm": 0.47646355628967285,
      "learning_rate": 8.854487863415483e-06,
      "loss": 0.4136,
      "step": 15390
    },
    {
      "epoch": 0.7063007663714378,
      "grad_norm": 0.42578306794166565,
      "learning_rate": 8.854331684105308e-06,
      "loss": 0.3387,
      "step": 15391
    },
    {
      "epoch": 0.7063466568766922,
      "grad_norm": 0.4847331643104553,
      "learning_rate": 8.854175495526684e-06,
      "loss": 0.4895,
      "step": 15392
    },
    {
      "epoch": 0.7063925473819467,
      "grad_norm": 0.524849534034729,
      "learning_rate": 8.854019297679982e-06,
      "loss": 0.4883,
      "step": 15393
    },
    {
      "epoch": 0.7064384378872012,
      "grad_norm": 0.54474276304245,
      "learning_rate": 8.853863090565581e-06,
      "loss": 0.5256,
      "step": 15394
    },
    {
      "epoch": 0.7064843283924556,
      "grad_norm": 0.4386679232120514,
      "learning_rate": 8.853706874183854e-06,
      "loss": 0.3499,
      "step": 15395
    },
    {
      "epoch": 0.7065302188977101,
      "grad_norm": 0.438931941986084,
      "learning_rate": 8.85355064853518e-06,
      "loss": 0.3724,
      "step": 15396
    },
    {
      "epoch": 0.7065761094029646,
      "grad_norm": 0.4804537892341614,
      "learning_rate": 8.853394413619933e-06,
      "loss": 0.4568,
      "step": 15397
    },
    {
      "epoch": 0.706621999908219,
      "grad_norm": 0.421892374753952,
      "learning_rate": 8.853238169438488e-06,
      "loss": 0.3617,
      "step": 15398
    },
    {
      "epoch": 0.7066678904134734,
      "grad_norm": 0.44802194833755493,
      "learning_rate": 8.853081915991222e-06,
      "loss": 0.4069,
      "step": 15399
    },
    {
      "epoch": 0.7067137809187279,
      "grad_norm": 0.44459229707717896,
      "learning_rate": 8.852925653278509e-06,
      "loss": 0.3931,
      "step": 15400
    },
    {
      "epoch": 0.7067596714239823,
      "grad_norm": 0.4145488440990448,
      "learning_rate": 8.852769381300727e-06,
      "loss": 0.3125,
      "step": 15401
    },
    {
      "epoch": 0.7068055619292368,
      "grad_norm": 0.44795331358909607,
      "learning_rate": 8.85261310005825e-06,
      "loss": 0.3436,
      "step": 15402
    },
    {
      "epoch": 0.7068514524344913,
      "grad_norm": 0.5064924955368042,
      "learning_rate": 8.852456809551453e-06,
      "loss": 0.3545,
      "step": 15403
    },
    {
      "epoch": 0.7068973429397458,
      "grad_norm": 0.43935829401016235,
      "learning_rate": 8.852300509780716e-06,
      "loss": 0.3849,
      "step": 15404
    },
    {
      "epoch": 0.7069432334450002,
      "grad_norm": 0.44526517391204834,
      "learning_rate": 8.85214420074641e-06,
      "loss": 0.3471,
      "step": 15405
    },
    {
      "epoch": 0.7069891239502547,
      "grad_norm": 0.4927416741847992,
      "learning_rate": 8.851987882448913e-06,
      "loss": 0.4469,
      "step": 15406
    },
    {
      "epoch": 0.7070350144555092,
      "grad_norm": 0.6479256749153137,
      "learning_rate": 8.8518315548886e-06,
      "loss": 0.3492,
      "step": 15407
    },
    {
      "epoch": 0.7070809049607636,
      "grad_norm": 0.4275621771812439,
      "learning_rate": 8.851675218065851e-06,
      "loss": 0.3228,
      "step": 15408
    },
    {
      "epoch": 0.7071267954660181,
      "grad_norm": 0.4764132797718048,
      "learning_rate": 8.851518871981038e-06,
      "loss": 0.4261,
      "step": 15409
    },
    {
      "epoch": 0.7071726859712726,
      "grad_norm": 0.477669894695282,
      "learning_rate": 8.851362516634537e-06,
      "loss": 0.391,
      "step": 15410
    },
    {
      "epoch": 0.707218576476527,
      "grad_norm": 0.46213603019714355,
      "learning_rate": 8.851206152026724e-06,
      "loss": 0.3907,
      "step": 15411
    },
    {
      "epoch": 0.7072644669817815,
      "grad_norm": 0.47017014026641846,
      "learning_rate": 8.851049778157975e-06,
      "loss": 0.4335,
      "step": 15412
    },
    {
      "epoch": 0.707310357487036,
      "grad_norm": 0.4846242070198059,
      "learning_rate": 8.850893395028668e-06,
      "loss": 0.4037,
      "step": 15413
    },
    {
      "epoch": 0.7073562479922904,
      "grad_norm": 0.5150488615036011,
      "learning_rate": 8.850737002639178e-06,
      "loss": 0.419,
      "step": 15414
    },
    {
      "epoch": 0.7074021384975449,
      "grad_norm": 0.4290221333503723,
      "learning_rate": 8.85058060098988e-06,
      "loss": 0.3586,
      "step": 15415
    },
    {
      "epoch": 0.7074480290027994,
      "grad_norm": 0.48115789890289307,
      "learning_rate": 8.85042419008115e-06,
      "loss": 0.4246,
      "step": 15416
    },
    {
      "epoch": 0.7074939195080537,
      "grad_norm": 0.48476967215538025,
      "learning_rate": 8.850267769913368e-06,
      "loss": 0.5144,
      "step": 15417
    },
    {
      "epoch": 0.7075398100133082,
      "grad_norm": 0.4595085680484772,
      "learning_rate": 8.850111340486904e-06,
      "loss": 0.436,
      "step": 15418
    },
    {
      "epoch": 0.7075857005185627,
      "grad_norm": 0.4457300007343292,
      "learning_rate": 8.84995490180214e-06,
      "loss": 0.3629,
      "step": 15419
    },
    {
      "epoch": 0.7076315910238171,
      "grad_norm": 0.4694974720478058,
      "learning_rate": 8.849798453859448e-06,
      "loss": 0.4544,
      "step": 15420
    },
    {
      "epoch": 0.7076774815290716,
      "grad_norm": 0.4014597237110138,
      "learning_rate": 8.849641996659206e-06,
      "loss": 0.3443,
      "step": 15421
    },
    {
      "epoch": 0.7077233720343261,
      "grad_norm": 0.46210533380508423,
      "learning_rate": 8.84948553020179e-06,
      "loss": 0.369,
      "step": 15422
    },
    {
      "epoch": 0.7077692625395806,
      "grad_norm": 0.4576537609100342,
      "learning_rate": 8.849329054487576e-06,
      "loss": 0.3984,
      "step": 15423
    },
    {
      "epoch": 0.707815153044835,
      "grad_norm": 0.4509550929069519,
      "learning_rate": 8.849172569516942e-06,
      "loss": 0.3924,
      "step": 15424
    },
    {
      "epoch": 0.7078610435500895,
      "grad_norm": 0.4774022102355957,
      "learning_rate": 8.849016075290262e-06,
      "loss": 0.4329,
      "step": 15425
    },
    {
      "epoch": 0.707906934055344,
      "grad_norm": 0.4913226068019867,
      "learning_rate": 8.848859571807913e-06,
      "loss": 0.4757,
      "step": 15426
    },
    {
      "epoch": 0.7079528245605984,
      "grad_norm": 0.5216846466064453,
      "learning_rate": 8.848703059070269e-06,
      "loss": 0.5284,
      "step": 15427
    },
    {
      "epoch": 0.7079987150658529,
      "grad_norm": 0.4755108952522278,
      "learning_rate": 8.848546537077713e-06,
      "loss": 0.4076,
      "step": 15428
    },
    {
      "epoch": 0.7080446055711074,
      "grad_norm": 0.45100149512290955,
      "learning_rate": 8.848390005830615e-06,
      "loss": 0.3951,
      "step": 15429
    },
    {
      "epoch": 0.7080904960763618,
      "grad_norm": 0.45336994528770447,
      "learning_rate": 8.848233465329354e-06,
      "loss": 0.4049,
      "step": 15430
    },
    {
      "epoch": 0.7081363865816163,
      "grad_norm": 0.4541419744491577,
      "learning_rate": 8.848076915574306e-06,
      "loss": 0.3455,
      "step": 15431
    },
    {
      "epoch": 0.7081822770868708,
      "grad_norm": 0.4507831633090973,
      "learning_rate": 8.847920356565846e-06,
      "loss": 0.375,
      "step": 15432
    },
    {
      "epoch": 0.7082281675921251,
      "grad_norm": 0.4275922477245331,
      "learning_rate": 8.847763788304356e-06,
      "loss": 0.3775,
      "step": 15433
    },
    {
      "epoch": 0.7082740580973796,
      "grad_norm": 0.49520406126976013,
      "learning_rate": 8.847607210790204e-06,
      "loss": 0.4534,
      "step": 15434
    },
    {
      "epoch": 0.7083199486026341,
      "grad_norm": 0.42178016901016235,
      "learning_rate": 8.847450624023773e-06,
      "loss": 0.3304,
      "step": 15435
    },
    {
      "epoch": 0.7083658391078885,
      "grad_norm": 0.5048931837081909,
      "learning_rate": 8.847294028005438e-06,
      "loss": 0.4421,
      "step": 15436
    },
    {
      "epoch": 0.708411729613143,
      "grad_norm": 0.4656357169151306,
      "learning_rate": 8.847137422735575e-06,
      "loss": 0.3858,
      "step": 15437
    },
    {
      "epoch": 0.7084576201183975,
      "grad_norm": 0.4552110731601715,
      "learning_rate": 8.846980808214561e-06,
      "loss": 0.4303,
      "step": 15438
    },
    {
      "epoch": 0.7085035106236519,
      "grad_norm": 0.5013412833213806,
      "learning_rate": 8.84682418444277e-06,
      "loss": 0.4973,
      "step": 15439
    },
    {
      "epoch": 0.7085494011289064,
      "grad_norm": 0.4904690086841583,
      "learning_rate": 8.846667551420582e-06,
      "loss": 0.4064,
      "step": 15440
    },
    {
      "epoch": 0.7085952916341609,
      "grad_norm": 0.46359390020370483,
      "learning_rate": 8.846510909148373e-06,
      "loss": 0.4401,
      "step": 15441
    },
    {
      "epoch": 0.7086411821394154,
      "grad_norm": 0.4749346673488617,
      "learning_rate": 8.84635425762652e-06,
      "loss": 0.3759,
      "step": 15442
    },
    {
      "epoch": 0.7086870726446698,
      "grad_norm": 0.4849734604358673,
      "learning_rate": 8.846197596855399e-06,
      "loss": 0.4385,
      "step": 15443
    },
    {
      "epoch": 0.7087329631499243,
      "grad_norm": 0.4457416236400604,
      "learning_rate": 8.846040926835385e-06,
      "loss": 0.3309,
      "step": 15444
    },
    {
      "epoch": 0.7087788536551788,
      "grad_norm": 0.42112627625465393,
      "learning_rate": 8.845884247566859e-06,
      "loss": 0.3268,
      "step": 15445
    },
    {
      "epoch": 0.7088247441604332,
      "grad_norm": 0.4570595920085907,
      "learning_rate": 8.845727559050193e-06,
      "loss": 0.4305,
      "step": 15446
    },
    {
      "epoch": 0.7088706346656877,
      "grad_norm": 0.4492507576942444,
      "learning_rate": 8.845570861285766e-06,
      "loss": 0.3891,
      "step": 15447
    },
    {
      "epoch": 0.7089165251709422,
      "grad_norm": 0.5062021017074585,
      "learning_rate": 8.845414154273955e-06,
      "loss": 0.3823,
      "step": 15448
    },
    {
      "epoch": 0.7089624156761966,
      "grad_norm": 0.43542948365211487,
      "learning_rate": 8.845257438015137e-06,
      "loss": 0.3505,
      "step": 15449
    },
    {
      "epoch": 0.709008306181451,
      "grad_norm": 0.45576897263526917,
      "learning_rate": 8.84510071250969e-06,
      "loss": 0.352,
      "step": 15450
    },
    {
      "epoch": 0.7090541966867056,
      "grad_norm": 0.43054863810539246,
      "learning_rate": 8.844943977757986e-06,
      "loss": 0.3805,
      "step": 15451
    },
    {
      "epoch": 0.7091000871919599,
      "grad_norm": 0.4283328354358673,
      "learning_rate": 8.844787233760409e-06,
      "loss": 0.3652,
      "step": 15452
    },
    {
      "epoch": 0.7091459776972144,
      "grad_norm": 0.4831283390522003,
      "learning_rate": 8.84463048051733e-06,
      "loss": 0.4783,
      "step": 15453
    },
    {
      "epoch": 0.7091918682024689,
      "grad_norm": 0.432017058134079,
      "learning_rate": 8.844473718029127e-06,
      "loss": 0.396,
      "step": 15454
    },
    {
      "epoch": 0.7092377587077233,
      "grad_norm": 0.4756677746772766,
      "learning_rate": 8.844316946296179e-06,
      "loss": 0.5007,
      "step": 15455
    },
    {
      "epoch": 0.7092836492129778,
      "grad_norm": 0.45042964816093445,
      "learning_rate": 8.844160165318862e-06,
      "loss": 0.3903,
      "step": 15456
    },
    {
      "epoch": 0.7093295397182323,
      "grad_norm": 0.4317377209663391,
      "learning_rate": 8.844003375097553e-06,
      "loss": 0.3587,
      "step": 15457
    },
    {
      "epoch": 0.7093754302234868,
      "grad_norm": 0.48564499616622925,
      "learning_rate": 8.843846575632631e-06,
      "loss": 0.4416,
      "step": 15458
    },
    {
      "epoch": 0.7094213207287412,
      "grad_norm": 0.4756935238838196,
      "learning_rate": 8.843689766924469e-06,
      "loss": 0.4378,
      "step": 15459
    },
    {
      "epoch": 0.7094672112339957,
      "grad_norm": 0.4673527777194977,
      "learning_rate": 8.843532948973447e-06,
      "loss": 0.3964,
      "step": 15460
    },
    {
      "epoch": 0.7095131017392502,
      "grad_norm": 0.44505345821380615,
      "learning_rate": 8.843376121779941e-06,
      "loss": 0.3783,
      "step": 15461
    },
    {
      "epoch": 0.7095589922445046,
      "grad_norm": 0.43367916345596313,
      "learning_rate": 8.843219285344328e-06,
      "loss": 0.3309,
      "step": 15462
    },
    {
      "epoch": 0.7096048827497591,
      "grad_norm": 0.43835362792015076,
      "learning_rate": 8.843062439666986e-06,
      "loss": 0.3278,
      "step": 15463
    },
    {
      "epoch": 0.7096507732550136,
      "grad_norm": 0.5083480477333069,
      "learning_rate": 8.842905584748292e-06,
      "loss": 0.5513,
      "step": 15464
    },
    {
      "epoch": 0.709696663760268,
      "grad_norm": 0.44464245438575745,
      "learning_rate": 8.842748720588623e-06,
      "loss": 0.382,
      "step": 15465
    },
    {
      "epoch": 0.7097425542655225,
      "grad_norm": 0.4823538064956665,
      "learning_rate": 8.842591847188355e-06,
      "loss": 0.4262,
      "step": 15466
    },
    {
      "epoch": 0.709788444770777,
      "grad_norm": 0.44182926416397095,
      "learning_rate": 8.842434964547867e-06,
      "loss": 0.3419,
      "step": 15467
    },
    {
      "epoch": 0.7098343352760313,
      "grad_norm": 0.4673125743865967,
      "learning_rate": 8.842278072667536e-06,
      "loss": 0.4209,
      "step": 15468
    },
    {
      "epoch": 0.7098802257812858,
      "grad_norm": 0.4260706901550293,
      "learning_rate": 8.84212117154774e-06,
      "loss": 0.3617,
      "step": 15469
    },
    {
      "epoch": 0.7099261162865403,
      "grad_norm": 0.47984281182289124,
      "learning_rate": 8.841964261188853e-06,
      "loss": 0.4356,
      "step": 15470
    },
    {
      "epoch": 0.7099720067917947,
      "grad_norm": 0.5031965374946594,
      "learning_rate": 8.841807341591255e-06,
      "loss": 0.4856,
      "step": 15471
    },
    {
      "epoch": 0.7100178972970492,
      "grad_norm": 0.46491318941116333,
      "learning_rate": 8.841650412755325e-06,
      "loss": 0.4304,
      "step": 15472
    },
    {
      "epoch": 0.7100637878023037,
      "grad_norm": 0.44671010971069336,
      "learning_rate": 8.841493474681437e-06,
      "loss": 0.4109,
      "step": 15473
    },
    {
      "epoch": 0.7101096783075581,
      "grad_norm": 0.5169432759284973,
      "learning_rate": 8.841336527369968e-06,
      "loss": 0.454,
      "step": 15474
    },
    {
      "epoch": 0.7101555688128126,
      "grad_norm": 0.46499308943748474,
      "learning_rate": 8.841179570821299e-06,
      "loss": 0.4107,
      "step": 15475
    },
    {
      "epoch": 0.7102014593180671,
      "grad_norm": 0.46452900767326355,
      "learning_rate": 8.841022605035804e-06,
      "loss": 0.4144,
      "step": 15476
    },
    {
      "epoch": 0.7102473498233216,
      "grad_norm": 0.4622470736503601,
      "learning_rate": 8.840865630013861e-06,
      "loss": 0.406,
      "step": 15477
    },
    {
      "epoch": 0.710293240328576,
      "grad_norm": 0.435015469789505,
      "learning_rate": 8.84070864575585e-06,
      "loss": 0.3799,
      "step": 15478
    },
    {
      "epoch": 0.7103391308338305,
      "grad_norm": 0.4552239775657654,
      "learning_rate": 8.840551652262148e-06,
      "loss": 0.3863,
      "step": 15479
    },
    {
      "epoch": 0.710385021339085,
      "grad_norm": 0.4081590175628662,
      "learning_rate": 8.84039464953313e-06,
      "loss": 0.3369,
      "step": 15480
    },
    {
      "epoch": 0.7104309118443394,
      "grad_norm": 0.44889116287231445,
      "learning_rate": 8.840237637569175e-06,
      "loss": 0.4123,
      "step": 15481
    },
    {
      "epoch": 0.7104768023495939,
      "grad_norm": 0.5037751793861389,
      "learning_rate": 8.84008061637066e-06,
      "loss": 0.4937,
      "step": 15482
    },
    {
      "epoch": 0.7105226928548484,
      "grad_norm": 0.4261554479598999,
      "learning_rate": 8.839923585937965e-06,
      "loss": 0.3322,
      "step": 15483
    },
    {
      "epoch": 0.7105685833601028,
      "grad_norm": 0.39779770374298096,
      "learning_rate": 8.839766546271465e-06,
      "loss": 0.2995,
      "step": 15484
    },
    {
      "epoch": 0.7106144738653573,
      "grad_norm": 0.4634862542152405,
      "learning_rate": 8.839609497371537e-06,
      "loss": 0.4458,
      "step": 15485
    },
    {
      "epoch": 0.7106603643706118,
      "grad_norm": 0.47490790486335754,
      "learning_rate": 8.83945243923856e-06,
      "loss": 0.4558,
      "step": 15486
    },
    {
      "epoch": 0.7107062548758661,
      "grad_norm": 0.4840882122516632,
      "learning_rate": 8.839295371872912e-06,
      "loss": 0.4899,
      "step": 15487
    },
    {
      "epoch": 0.7107521453811206,
      "grad_norm": 0.45656970143318176,
      "learning_rate": 8.839138295274971e-06,
      "loss": 0.3998,
      "step": 15488
    },
    {
      "epoch": 0.7107980358863751,
      "grad_norm": 0.4725770950317383,
      "learning_rate": 8.838981209445114e-06,
      "loss": 0.443,
      "step": 15489
    },
    {
      "epoch": 0.7108439263916295,
      "grad_norm": 0.46444380283355713,
      "learning_rate": 8.838824114383719e-06,
      "loss": 0.3618,
      "step": 15490
    },
    {
      "epoch": 0.710889816896884,
      "grad_norm": 0.4527285695075989,
      "learning_rate": 8.838667010091161e-06,
      "loss": 0.3725,
      "step": 15491
    },
    {
      "epoch": 0.7109357074021385,
      "grad_norm": 0.4533938765525818,
      "learning_rate": 8.838509896567823e-06,
      "loss": 0.393,
      "step": 15492
    },
    {
      "epoch": 0.710981597907393,
      "grad_norm": 0.46753445267677307,
      "learning_rate": 8.838352773814081e-06,
      "loss": 0.423,
      "step": 15493
    },
    {
      "epoch": 0.7110274884126474,
      "grad_norm": 0.43710485100746155,
      "learning_rate": 8.83819564183031e-06,
      "loss": 0.4377,
      "step": 15494
    },
    {
      "epoch": 0.7110733789179019,
      "grad_norm": 0.48153793811798096,
      "learning_rate": 8.83803850061689e-06,
      "loss": 0.4319,
      "step": 15495
    },
    {
      "epoch": 0.7111192694231564,
      "grad_norm": 0.48703303933143616,
      "learning_rate": 8.8378813501742e-06,
      "loss": 0.4008,
      "step": 15496
    },
    {
      "epoch": 0.7111651599284108,
      "grad_norm": 0.5200701355934143,
      "learning_rate": 8.837724190502615e-06,
      "loss": 0.502,
      "step": 15497
    },
    {
      "epoch": 0.7112110504336653,
      "grad_norm": 0.4861406087875366,
      "learning_rate": 8.837567021602516e-06,
      "loss": 0.431,
      "step": 15498
    },
    {
      "epoch": 0.7112569409389198,
      "grad_norm": 0.4543859362602234,
      "learning_rate": 8.837409843474278e-06,
      "loss": 0.3904,
      "step": 15499
    },
    {
      "epoch": 0.7113028314441742,
      "grad_norm": 0.4633701741695404,
      "learning_rate": 8.837252656118283e-06,
      "loss": 0.4255,
      "step": 15500
    },
    {
      "epoch": 0.7113487219494287,
      "grad_norm": 0.4609166085720062,
      "learning_rate": 8.837095459534903e-06,
      "loss": 0.4612,
      "step": 15501
    },
    {
      "epoch": 0.7113946124546832,
      "grad_norm": 0.4380700886249542,
      "learning_rate": 8.836938253724522e-06,
      "loss": 0.365,
      "step": 15502
    },
    {
      "epoch": 0.7114405029599375,
      "grad_norm": 0.48445984721183777,
      "learning_rate": 8.836781038687514e-06,
      "loss": 0.4468,
      "step": 15503
    },
    {
      "epoch": 0.711486393465192,
      "grad_norm": 0.4344775974750519,
      "learning_rate": 8.836623814424258e-06,
      "loss": 0.3525,
      "step": 15504
    },
    {
      "epoch": 0.7115322839704465,
      "grad_norm": 0.4633212089538574,
      "learning_rate": 8.836466580935134e-06,
      "loss": 0.4431,
      "step": 15505
    },
    {
      "epoch": 0.7115781744757009,
      "grad_norm": 0.469025194644928,
      "learning_rate": 8.836309338220518e-06,
      "loss": 0.4019,
      "step": 15506
    },
    {
      "epoch": 0.7116240649809554,
      "grad_norm": 0.45229193568229675,
      "learning_rate": 8.83615208628079e-06,
      "loss": 0.4172,
      "step": 15507
    },
    {
      "epoch": 0.7116699554862099,
      "grad_norm": 0.4528132975101471,
      "learning_rate": 8.835994825116327e-06,
      "loss": 0.3682,
      "step": 15508
    },
    {
      "epoch": 0.7117158459914643,
      "grad_norm": 0.4499376714229584,
      "learning_rate": 8.835837554727507e-06,
      "loss": 0.3638,
      "step": 15509
    },
    {
      "epoch": 0.7117617364967188,
      "grad_norm": 0.46299105882644653,
      "learning_rate": 8.835680275114708e-06,
      "loss": 0.4122,
      "step": 15510
    },
    {
      "epoch": 0.7118076270019733,
      "grad_norm": 0.5443454384803772,
      "learning_rate": 8.835522986278308e-06,
      "loss": 0.4729,
      "step": 15511
    },
    {
      "epoch": 0.7118535175072278,
      "grad_norm": 0.5224723815917969,
      "learning_rate": 8.835365688218684e-06,
      "loss": 0.545,
      "step": 15512
    },
    {
      "epoch": 0.7118994080124822,
      "grad_norm": 0.4485406279563904,
      "learning_rate": 8.83520838093622e-06,
      "loss": 0.4107,
      "step": 15513
    },
    {
      "epoch": 0.7119452985177367,
      "grad_norm": 0.4891459047794342,
      "learning_rate": 8.835051064431288e-06,
      "loss": 0.4447,
      "step": 15514
    },
    {
      "epoch": 0.7119911890229912,
      "grad_norm": 0.41155001521110535,
      "learning_rate": 8.834893738704269e-06,
      "loss": 0.2997,
      "step": 15515
    },
    {
      "epoch": 0.7120370795282456,
      "grad_norm": 0.45614489912986755,
      "learning_rate": 8.83473640375554e-06,
      "loss": 0.4056,
      "step": 15516
    },
    {
      "epoch": 0.7120829700335001,
      "grad_norm": 0.46351298689842224,
      "learning_rate": 8.83457905958548e-06,
      "loss": 0.4325,
      "step": 15517
    },
    {
      "epoch": 0.7121288605387546,
      "grad_norm": 0.4697418212890625,
      "learning_rate": 8.834421706194469e-06,
      "loss": 0.4376,
      "step": 15518
    },
    {
      "epoch": 0.712174751044009,
      "grad_norm": 0.4287526309490204,
      "learning_rate": 8.834264343582883e-06,
      "loss": 0.3818,
      "step": 15519
    },
    {
      "epoch": 0.7122206415492635,
      "grad_norm": 0.4640028774738312,
      "learning_rate": 8.834106971751102e-06,
      "loss": 0.4379,
      "step": 15520
    },
    {
      "epoch": 0.712266532054518,
      "grad_norm": 0.5065779685974121,
      "learning_rate": 8.833949590699504e-06,
      "loss": 0.4895,
      "step": 15521
    },
    {
      "epoch": 0.7123124225597723,
      "grad_norm": 0.8899006247520447,
      "learning_rate": 8.833792200428468e-06,
      "loss": 0.4139,
      "step": 15522
    },
    {
      "epoch": 0.7123583130650268,
      "grad_norm": 0.4375467896461487,
      "learning_rate": 8.833634800938369e-06,
      "loss": 0.3691,
      "step": 15523
    },
    {
      "epoch": 0.7124042035702813,
      "grad_norm": 0.4735585153102875,
      "learning_rate": 8.83347739222959e-06,
      "loss": 0.4622,
      "step": 15524
    },
    {
      "epoch": 0.7124500940755357,
      "grad_norm": 0.41544556617736816,
      "learning_rate": 8.833319974302508e-06,
      "loss": 0.306,
      "step": 15525
    },
    {
      "epoch": 0.7124959845807902,
      "grad_norm": 0.44986918568611145,
      "learning_rate": 8.833162547157502e-06,
      "loss": 0.3501,
      "step": 15526
    },
    {
      "epoch": 0.7125418750860447,
      "grad_norm": 0.4355300962924957,
      "learning_rate": 8.833005110794947e-06,
      "loss": 0.3818,
      "step": 15527
    },
    {
      "epoch": 0.7125877655912991,
      "grad_norm": 0.46408483386039734,
      "learning_rate": 8.832847665215228e-06,
      "loss": 0.4484,
      "step": 15528
    },
    {
      "epoch": 0.7126336560965536,
      "grad_norm": 0.4280332028865814,
      "learning_rate": 8.832690210418717e-06,
      "loss": 0.3776,
      "step": 15529
    },
    {
      "epoch": 0.7126795466018081,
      "grad_norm": 0.47469472885131836,
      "learning_rate": 8.832532746405797e-06,
      "loss": 0.4085,
      "step": 15530
    },
    {
      "epoch": 0.7127254371070626,
      "grad_norm": 0.509865939617157,
      "learning_rate": 8.832375273176845e-06,
      "loss": 0.5008,
      "step": 15531
    },
    {
      "epoch": 0.712771327612317,
      "grad_norm": 0.46818438172340393,
      "learning_rate": 8.83221779073224e-06,
      "loss": 0.4143,
      "step": 15532
    },
    {
      "epoch": 0.7128172181175715,
      "grad_norm": 0.4492754638195038,
      "learning_rate": 8.83206029907236e-06,
      "loss": 0.4014,
      "step": 15533
    },
    {
      "epoch": 0.712863108622826,
      "grad_norm": 0.46445271372795105,
      "learning_rate": 8.831902798197588e-06,
      "loss": 0.3813,
      "step": 15534
    },
    {
      "epoch": 0.7129089991280804,
      "grad_norm": 0.47076019644737244,
      "learning_rate": 8.831745288108296e-06,
      "loss": 0.3953,
      "step": 15535
    },
    {
      "epoch": 0.7129548896333349,
      "grad_norm": 0.44588086009025574,
      "learning_rate": 8.831587768804865e-06,
      "loss": 0.3871,
      "step": 15536
    },
    {
      "epoch": 0.7130007801385894,
      "grad_norm": 0.48752206563949585,
      "learning_rate": 8.831430240287677e-06,
      "loss": 0.4733,
      "step": 15537
    },
    {
      "epoch": 0.7130466706438437,
      "grad_norm": 0.48644009232521057,
      "learning_rate": 8.831272702557107e-06,
      "loss": 0.4379,
      "step": 15538
    },
    {
      "epoch": 0.7130925611490982,
      "grad_norm": 0.4382927119731903,
      "learning_rate": 8.831115155613537e-06,
      "loss": 0.3885,
      "step": 15539
    },
    {
      "epoch": 0.7131384516543527,
      "grad_norm": 0.4415043890476227,
      "learning_rate": 8.830957599457341e-06,
      "loss": 0.3226,
      "step": 15540
    },
    {
      "epoch": 0.7131843421596071,
      "grad_norm": 0.48629793524742126,
      "learning_rate": 8.830800034088904e-06,
      "loss": 0.4203,
      "step": 15541
    },
    {
      "epoch": 0.7132302326648616,
      "grad_norm": 0.4292840361595154,
      "learning_rate": 8.830642459508602e-06,
      "loss": 0.3802,
      "step": 15542
    },
    {
      "epoch": 0.7132761231701161,
      "grad_norm": 0.43133819103240967,
      "learning_rate": 8.830484875716813e-06,
      "loss": 0.378,
      "step": 15543
    },
    {
      "epoch": 0.7133220136753705,
      "grad_norm": 0.4392184913158417,
      "learning_rate": 8.830327282713917e-06,
      "loss": 0.3893,
      "step": 15544
    },
    {
      "epoch": 0.713367904180625,
      "grad_norm": 0.42994388937950134,
      "learning_rate": 8.83016968050029e-06,
      "loss": 0.3524,
      "step": 15545
    },
    {
      "epoch": 0.7134137946858795,
      "grad_norm": 0.39495158195495605,
      "learning_rate": 8.830012069076317e-06,
      "loss": 0.3075,
      "step": 15546
    },
    {
      "epoch": 0.713459685191134,
      "grad_norm": 0.427035391330719,
      "learning_rate": 8.829854448442374e-06,
      "loss": 0.3647,
      "step": 15547
    },
    {
      "epoch": 0.7135055756963884,
      "grad_norm": 0.43526822328567505,
      "learning_rate": 8.82969681859884e-06,
      "loss": 0.3589,
      "step": 15548
    },
    {
      "epoch": 0.7135514662016429,
      "grad_norm": 0.4792308211326599,
      "learning_rate": 8.829539179546091e-06,
      "loss": 0.3816,
      "step": 15549
    },
    {
      "epoch": 0.7135973567068974,
      "grad_norm": 0.440013587474823,
      "learning_rate": 8.829381531284511e-06,
      "loss": 0.3521,
      "step": 15550
    },
    {
      "epoch": 0.7136432472121518,
      "grad_norm": 0.4347813129425049,
      "learning_rate": 8.829223873814477e-06,
      "loss": 0.3415,
      "step": 15551
    },
    {
      "epoch": 0.7136891377174063,
      "grad_norm": 0.42824214696884155,
      "learning_rate": 8.829066207136368e-06,
      "loss": 0.3796,
      "step": 15552
    },
    {
      "epoch": 0.7137350282226608,
      "grad_norm": 0.46256357431411743,
      "learning_rate": 8.828908531250563e-06,
      "loss": 0.4141,
      "step": 15553
    },
    {
      "epoch": 0.7137809187279152,
      "grad_norm": 0.4591212570667267,
      "learning_rate": 8.828750846157441e-06,
      "loss": 0.4207,
      "step": 15554
    },
    {
      "epoch": 0.7138268092331697,
      "grad_norm": 0.43453794717788696,
      "learning_rate": 8.828593151857383e-06,
      "loss": 0.3921,
      "step": 15555
    },
    {
      "epoch": 0.7138726997384242,
      "grad_norm": 0.4260489046573639,
      "learning_rate": 8.828435448350766e-06,
      "loss": 0.3783,
      "step": 15556
    },
    {
      "epoch": 0.7139185902436785,
      "grad_norm": 0.4535537660121918,
      "learning_rate": 8.82827773563797e-06,
      "loss": 0.4047,
      "step": 15557
    },
    {
      "epoch": 0.713964480748933,
      "grad_norm": 0.49402597546577454,
      "learning_rate": 8.828120013719373e-06,
      "loss": 0.4856,
      "step": 15558
    },
    {
      "epoch": 0.7140103712541875,
      "grad_norm": 0.4657919704914093,
      "learning_rate": 8.827962282595357e-06,
      "loss": 0.4457,
      "step": 15559
    },
    {
      "epoch": 0.7140562617594419,
      "grad_norm": 0.4483382999897003,
      "learning_rate": 8.827804542266301e-06,
      "loss": 0.4082,
      "step": 15560
    },
    {
      "epoch": 0.7141021522646964,
      "grad_norm": 0.4551346004009247,
      "learning_rate": 8.82764679273258e-06,
      "loss": 0.4579,
      "step": 15561
    },
    {
      "epoch": 0.7141480427699509,
      "grad_norm": 0.43995821475982666,
      "learning_rate": 8.82748903399458e-06,
      "loss": 0.3646,
      "step": 15562
    },
    {
      "epoch": 0.7141939332752053,
      "grad_norm": 0.516647219657898,
      "learning_rate": 8.827331266052675e-06,
      "loss": 0.4977,
      "step": 15563
    },
    {
      "epoch": 0.7142398237804598,
      "grad_norm": 0.45906272530555725,
      "learning_rate": 8.827173488907246e-06,
      "loss": 0.4255,
      "step": 15564
    },
    {
      "epoch": 0.7142857142857143,
      "grad_norm": 0.5009387731552124,
      "learning_rate": 8.827015702558672e-06,
      "loss": 0.4686,
      "step": 15565
    },
    {
      "epoch": 0.7143316047909688,
      "grad_norm": 0.4899574816226959,
      "learning_rate": 8.826857907007336e-06,
      "loss": 0.4398,
      "step": 15566
    },
    {
      "epoch": 0.7143774952962232,
      "grad_norm": 0.4522970914840698,
      "learning_rate": 8.826700102253612e-06,
      "loss": 0.3935,
      "step": 15567
    },
    {
      "epoch": 0.7144233858014777,
      "grad_norm": 0.46897608041763306,
      "learning_rate": 8.826542288297881e-06,
      "loss": 0.4551,
      "step": 15568
    },
    {
      "epoch": 0.7144692763067322,
      "grad_norm": 0.4823964536190033,
      "learning_rate": 8.826384465140527e-06,
      "loss": 0.4483,
      "step": 15569
    },
    {
      "epoch": 0.7145151668119866,
      "grad_norm": 0.49233704805374146,
      "learning_rate": 8.826226632781923e-06,
      "loss": 0.4402,
      "step": 15570
    },
    {
      "epoch": 0.7145610573172411,
      "grad_norm": 0.4153207540512085,
      "learning_rate": 8.826068791222454e-06,
      "loss": 0.3539,
      "step": 15571
    },
    {
      "epoch": 0.7146069478224956,
      "grad_norm": 0.46038228273391724,
      "learning_rate": 8.825910940462494e-06,
      "loss": 0.4195,
      "step": 15572
    },
    {
      "epoch": 0.71465283832775,
      "grad_norm": 0.42529672384262085,
      "learning_rate": 8.825753080502429e-06,
      "loss": 0.3627,
      "step": 15573
    },
    {
      "epoch": 0.7146987288330044,
      "grad_norm": 0.45616424083709717,
      "learning_rate": 8.825595211342632e-06,
      "loss": 0.3882,
      "step": 15574
    },
    {
      "epoch": 0.714744619338259,
      "grad_norm": 0.44404497742652893,
      "learning_rate": 8.825437332983488e-06,
      "loss": 0.3591,
      "step": 15575
    },
    {
      "epoch": 0.7147905098435133,
      "grad_norm": 0.4738733172416687,
      "learning_rate": 8.825279445425375e-06,
      "loss": 0.4096,
      "step": 15576
    },
    {
      "epoch": 0.7148364003487678,
      "grad_norm": 0.45659273862838745,
      "learning_rate": 8.825121548668672e-06,
      "loss": 0.4061,
      "step": 15577
    },
    {
      "epoch": 0.7148822908540223,
      "grad_norm": 0.4600314795970917,
      "learning_rate": 8.824963642713757e-06,
      "loss": 0.3981,
      "step": 15578
    },
    {
      "epoch": 0.7149281813592767,
      "grad_norm": 0.46980634331703186,
      "learning_rate": 8.824805727561013e-06,
      "loss": 0.4687,
      "step": 15579
    },
    {
      "epoch": 0.7149740718645312,
      "grad_norm": 0.4423784911632538,
      "learning_rate": 8.824647803210819e-06,
      "loss": 0.4163,
      "step": 15580
    },
    {
      "epoch": 0.7150199623697857,
      "grad_norm": 0.4743698239326477,
      "learning_rate": 8.824489869663553e-06,
      "loss": 0.4638,
      "step": 15581
    },
    {
      "epoch": 0.7150658528750402,
      "grad_norm": 0.4626905024051666,
      "learning_rate": 8.824331926919595e-06,
      "loss": 0.3833,
      "step": 15582
    },
    {
      "epoch": 0.7151117433802946,
      "grad_norm": 0.436654657125473,
      "learning_rate": 8.82417397497933e-06,
      "loss": 0.3337,
      "step": 15583
    },
    {
      "epoch": 0.7151576338855491,
      "grad_norm": 0.4399971067905426,
      "learning_rate": 8.824016013843128e-06,
      "loss": 0.3837,
      "step": 15584
    },
    {
      "epoch": 0.7152035243908036,
      "grad_norm": 0.5217958092689514,
      "learning_rate": 8.823858043511378e-06,
      "loss": 0.4122,
      "step": 15585
    },
    {
      "epoch": 0.715249414896058,
      "grad_norm": 0.4876897931098938,
      "learning_rate": 8.823700063984455e-06,
      "loss": 0.4012,
      "step": 15586
    },
    {
      "epoch": 0.7152953054013125,
      "grad_norm": 0.5061797499656677,
      "learning_rate": 8.823542075262741e-06,
      "loss": 0.4912,
      "step": 15587
    },
    {
      "epoch": 0.715341195906567,
      "grad_norm": 0.5132040977478027,
      "learning_rate": 8.823384077346614e-06,
      "loss": 0.4995,
      "step": 15588
    },
    {
      "epoch": 0.7153870864118214,
      "grad_norm": 0.4687791168689728,
      "learning_rate": 8.823226070236456e-06,
      "loss": 0.3787,
      "step": 15589
    },
    {
      "epoch": 0.7154329769170759,
      "grad_norm": 0.4527410864830017,
      "learning_rate": 8.823068053932645e-06,
      "loss": 0.3877,
      "step": 15590
    },
    {
      "epoch": 0.7154788674223304,
      "grad_norm": 0.4342135787010193,
      "learning_rate": 8.822910028435562e-06,
      "loss": 0.3906,
      "step": 15591
    },
    {
      "epoch": 0.7155247579275847,
      "grad_norm": 0.43960365653038025,
      "learning_rate": 8.822751993745587e-06,
      "loss": 0.3434,
      "step": 15592
    },
    {
      "epoch": 0.7155706484328392,
      "grad_norm": 0.4276764988899231,
      "learning_rate": 8.822593949863101e-06,
      "loss": 0.3492,
      "step": 15593
    },
    {
      "epoch": 0.7156165389380937,
      "grad_norm": 0.44562748074531555,
      "learning_rate": 8.822435896788482e-06,
      "loss": 0.3591,
      "step": 15594
    },
    {
      "epoch": 0.7156624294433481,
      "grad_norm": 0.4125620126724243,
      "learning_rate": 8.82227783452211e-06,
      "loss": 0.3187,
      "step": 15595
    },
    {
      "epoch": 0.7157083199486026,
      "grad_norm": 0.4754997789859772,
      "learning_rate": 8.822119763064369e-06,
      "loss": 0.4474,
      "step": 15596
    },
    {
      "epoch": 0.7157542104538571,
      "grad_norm": 0.4303065240383148,
      "learning_rate": 8.821961682415634e-06,
      "loss": 0.3504,
      "step": 15597
    },
    {
      "epoch": 0.7158001009591115,
      "grad_norm": 0.46645036339759827,
      "learning_rate": 8.821803592576288e-06,
      "loss": 0.421,
      "step": 15598
    },
    {
      "epoch": 0.715845991464366,
      "grad_norm": 0.48790207505226135,
      "learning_rate": 8.821645493546712e-06,
      "loss": 0.4389,
      "step": 15599
    },
    {
      "epoch": 0.7158918819696205,
      "grad_norm": 0.4692898988723755,
      "learning_rate": 8.821487385327283e-06,
      "loss": 0.3723,
      "step": 15600
    },
    {
      "epoch": 0.715937772474875,
      "grad_norm": 0.49176734685897827,
      "learning_rate": 8.821329267918382e-06,
      "loss": 0.4284,
      "step": 15601
    },
    {
      "epoch": 0.7159836629801294,
      "grad_norm": 0.43893522024154663,
      "learning_rate": 8.821171141320392e-06,
      "loss": 0.342,
      "step": 15602
    },
    {
      "epoch": 0.7160295534853839,
      "grad_norm": 0.45134127140045166,
      "learning_rate": 8.821013005533691e-06,
      "loss": 0.3989,
      "step": 15603
    },
    {
      "epoch": 0.7160754439906384,
      "grad_norm": 0.4622332751750946,
      "learning_rate": 8.820854860558659e-06,
      "loss": 0.3613,
      "step": 15604
    },
    {
      "epoch": 0.7161213344958928,
      "grad_norm": 0.42012733221054077,
      "learning_rate": 8.820696706395677e-06,
      "loss": 0.3348,
      "step": 15605
    },
    {
      "epoch": 0.7161672250011473,
      "grad_norm": 0.459369957447052,
      "learning_rate": 8.820538543045127e-06,
      "loss": 0.3951,
      "step": 15606
    },
    {
      "epoch": 0.7162131155064018,
      "grad_norm": 0.49629658460617065,
      "learning_rate": 8.820380370507386e-06,
      "loss": 0.5011,
      "step": 15607
    },
    {
      "epoch": 0.7162590060116562,
      "grad_norm": 0.48579883575439453,
      "learning_rate": 8.820222188782837e-06,
      "loss": 0.4652,
      "step": 15608
    },
    {
      "epoch": 0.7163048965169106,
      "grad_norm": 0.500974178314209,
      "learning_rate": 8.820063997871857e-06,
      "loss": 0.4412,
      "step": 15609
    },
    {
      "epoch": 0.7163507870221651,
      "grad_norm": 0.45358914136886597,
      "learning_rate": 8.81990579777483e-06,
      "loss": 0.4202,
      "step": 15610
    },
    {
      "epoch": 0.7163966775274195,
      "grad_norm": 0.47065556049346924,
      "learning_rate": 8.819747588492135e-06,
      "loss": 0.3794,
      "step": 15611
    },
    {
      "epoch": 0.716442568032674,
      "grad_norm": 0.45115119218826294,
      "learning_rate": 8.819589370024155e-06,
      "loss": 0.4398,
      "step": 15612
    },
    {
      "epoch": 0.7164884585379285,
      "grad_norm": 0.4697340726852417,
      "learning_rate": 8.819431142371267e-06,
      "loss": 0.4316,
      "step": 15613
    },
    {
      "epoch": 0.7165343490431829,
      "grad_norm": 0.45078858733177185,
      "learning_rate": 8.81927290553385e-06,
      "loss": 0.4139,
      "step": 15614
    },
    {
      "epoch": 0.7165802395484374,
      "grad_norm": 0.47174978256225586,
      "learning_rate": 8.819114659512289e-06,
      "loss": 0.5085,
      "step": 15615
    },
    {
      "epoch": 0.7166261300536919,
      "grad_norm": 0.43557462096214294,
      "learning_rate": 8.818956404306962e-06,
      "loss": 0.4072,
      "step": 15616
    },
    {
      "epoch": 0.7166720205589463,
      "grad_norm": 0.42182135581970215,
      "learning_rate": 8.818798139918253e-06,
      "loss": 0.3414,
      "step": 15617
    },
    {
      "epoch": 0.7167179110642008,
      "grad_norm": 0.47500914335250854,
      "learning_rate": 8.818639866346534e-06,
      "loss": 0.3834,
      "step": 15618
    },
    {
      "epoch": 0.7167638015694553,
      "grad_norm": 0.48851120471954346,
      "learning_rate": 8.818481583592196e-06,
      "loss": 0.4616,
      "step": 15619
    },
    {
      "epoch": 0.7168096920747098,
      "grad_norm": 0.4447363317012787,
      "learning_rate": 8.818323291655613e-06,
      "loss": 0.4035,
      "step": 15620
    },
    {
      "epoch": 0.7168555825799642,
      "grad_norm": 0.4497104287147522,
      "learning_rate": 8.818164990537169e-06,
      "loss": 0.4008,
      "step": 15621
    },
    {
      "epoch": 0.7169014730852187,
      "grad_norm": 0.4267825782299042,
      "learning_rate": 8.818006680237241e-06,
      "loss": 0.303,
      "step": 15622
    },
    {
      "epoch": 0.7169473635904732,
      "grad_norm": 0.41800087690353394,
      "learning_rate": 8.817848360756213e-06,
      "loss": 0.3896,
      "step": 15623
    },
    {
      "epoch": 0.7169932540957276,
      "grad_norm": 0.4382498264312744,
      "learning_rate": 8.817690032094466e-06,
      "loss": 0.3299,
      "step": 15624
    },
    {
      "epoch": 0.7170391446009821,
      "grad_norm": 0.45760563015937805,
      "learning_rate": 8.817531694252378e-06,
      "loss": 0.4024,
      "step": 15625
    },
    {
      "epoch": 0.7170850351062366,
      "grad_norm": 0.43356430530548096,
      "learning_rate": 8.81737334723033e-06,
      "loss": 0.3394,
      "step": 15626
    },
    {
      "epoch": 0.7171309256114909,
      "grad_norm": 0.5737821459770203,
      "learning_rate": 8.817214991028707e-06,
      "loss": 0.3363,
      "step": 15627
    },
    {
      "epoch": 0.7171768161167454,
      "grad_norm": 0.4786701798439026,
      "learning_rate": 8.817056625647885e-06,
      "loss": 0.398,
      "step": 15628
    },
    {
      "epoch": 0.7172227066219999,
      "grad_norm": 0.45972543954849243,
      "learning_rate": 8.816898251088246e-06,
      "loss": 0.3669,
      "step": 15629
    },
    {
      "epoch": 0.7172685971272543,
      "grad_norm": 0.46570178866386414,
      "learning_rate": 8.816739867350172e-06,
      "loss": 0.3675,
      "step": 15630
    },
    {
      "epoch": 0.7173144876325088,
      "grad_norm": 0.5328686237335205,
      "learning_rate": 8.816581474434042e-06,
      "loss": 0.4888,
      "step": 15631
    },
    {
      "epoch": 0.7173603781377633,
      "grad_norm": 0.4507195055484772,
      "learning_rate": 8.816423072340241e-06,
      "loss": 0.3662,
      "step": 15632
    },
    {
      "epoch": 0.7174062686430177,
      "grad_norm": 0.4924786388874054,
      "learning_rate": 8.816264661069144e-06,
      "loss": 0.4201,
      "step": 15633
    },
    {
      "epoch": 0.7174521591482722,
      "grad_norm": 0.474643349647522,
      "learning_rate": 8.816106240621137e-06,
      "loss": 0.4369,
      "step": 15634
    },
    {
      "epoch": 0.7174980496535267,
      "grad_norm": 0.4365168511867523,
      "learning_rate": 8.815947810996596e-06,
      "loss": 0.3771,
      "step": 15635
    },
    {
      "epoch": 0.7175439401587812,
      "grad_norm": 0.5028250813484192,
      "learning_rate": 8.815789372195909e-06,
      "loss": 0.4004,
      "step": 15636
    },
    {
      "epoch": 0.7175898306640356,
      "grad_norm": 0.4864856004714966,
      "learning_rate": 8.81563092421945e-06,
      "loss": 0.4997,
      "step": 15637
    },
    {
      "epoch": 0.7176357211692901,
      "grad_norm": 0.4270686209201813,
      "learning_rate": 8.815472467067603e-06,
      "loss": 0.3344,
      "step": 15638
    },
    {
      "epoch": 0.7176816116745446,
      "grad_norm": 0.48633384704589844,
      "learning_rate": 8.81531400074075e-06,
      "loss": 0.4453,
      "step": 15639
    },
    {
      "epoch": 0.717727502179799,
      "grad_norm": 0.46596699953079224,
      "learning_rate": 8.815155525239271e-06,
      "loss": 0.4321,
      "step": 15640
    },
    {
      "epoch": 0.7177733926850535,
      "grad_norm": 0.44546231627464294,
      "learning_rate": 8.814997040563547e-06,
      "loss": 0.3905,
      "step": 15641
    },
    {
      "epoch": 0.717819283190308,
      "grad_norm": 0.42674729228019714,
      "learning_rate": 8.814838546713958e-06,
      "loss": 0.3726,
      "step": 15642
    },
    {
      "epoch": 0.7178651736955624,
      "grad_norm": 0.47995319962501526,
      "learning_rate": 8.814680043690887e-06,
      "loss": 0.4005,
      "step": 15643
    },
    {
      "epoch": 0.7179110642008169,
      "grad_norm": 0.40687865018844604,
      "learning_rate": 8.814521531494715e-06,
      "loss": 0.3193,
      "step": 15644
    },
    {
      "epoch": 0.7179569547060713,
      "grad_norm": 0.4590286612510681,
      "learning_rate": 8.814363010125822e-06,
      "loss": 0.4412,
      "step": 15645
    },
    {
      "epoch": 0.7180028452113257,
      "grad_norm": 0.42670056223869324,
      "learning_rate": 8.81420447958459e-06,
      "loss": 0.3724,
      "step": 15646
    },
    {
      "epoch": 0.7180487357165802,
      "grad_norm": 0.7275289297103882,
      "learning_rate": 8.8140459398714e-06,
      "loss": 0.3582,
      "step": 15647
    },
    {
      "epoch": 0.7180946262218347,
      "grad_norm": 0.4692992568016052,
      "learning_rate": 8.813887390986632e-06,
      "loss": 0.3931,
      "step": 15648
    },
    {
      "epoch": 0.7181405167270891,
      "grad_norm": 0.4102296531200409,
      "learning_rate": 8.81372883293067e-06,
      "loss": 0.3119,
      "step": 15649
    },
    {
      "epoch": 0.7181864072323436,
      "grad_norm": 0.5070441365242004,
      "learning_rate": 8.813570265703894e-06,
      "loss": 0.4509,
      "step": 15650
    },
    {
      "epoch": 0.7182322977375981,
      "grad_norm": 0.4424387514591217,
      "learning_rate": 8.813411689306684e-06,
      "loss": 0.3823,
      "step": 15651
    },
    {
      "epoch": 0.7182781882428525,
      "grad_norm": 0.47151249647140503,
      "learning_rate": 8.813253103739422e-06,
      "loss": 0.3514,
      "step": 15652
    },
    {
      "epoch": 0.718324078748107,
      "grad_norm": 0.43100136518478394,
      "learning_rate": 8.813094509002492e-06,
      "loss": 0.3487,
      "step": 15653
    },
    {
      "epoch": 0.7183699692533615,
      "grad_norm": 0.4997897744178772,
      "learning_rate": 8.812935905096272e-06,
      "loss": 0.4515,
      "step": 15654
    },
    {
      "epoch": 0.718415859758616,
      "grad_norm": 0.44973400235176086,
      "learning_rate": 8.812777292021144e-06,
      "loss": 0.3713,
      "step": 15655
    },
    {
      "epoch": 0.7184617502638704,
      "grad_norm": 0.4724183678627014,
      "learning_rate": 8.812618669777492e-06,
      "loss": 0.3864,
      "step": 15656
    },
    {
      "epoch": 0.7185076407691249,
      "grad_norm": 0.48242223262786865,
      "learning_rate": 8.812460038365693e-06,
      "loss": 0.4544,
      "step": 15657
    },
    {
      "epoch": 0.7185535312743794,
      "grad_norm": 0.46911078691482544,
      "learning_rate": 8.812301397786131e-06,
      "loss": 0.454,
      "step": 15658
    },
    {
      "epoch": 0.7185994217796338,
      "grad_norm": 0.40428584814071655,
      "learning_rate": 8.812142748039187e-06,
      "loss": 0.3066,
      "step": 15659
    },
    {
      "epoch": 0.7186453122848883,
      "grad_norm": 0.4288439154624939,
      "learning_rate": 8.811984089125245e-06,
      "loss": 0.3393,
      "step": 15660
    },
    {
      "epoch": 0.7186912027901428,
      "grad_norm": 0.4545484185218811,
      "learning_rate": 8.811825421044683e-06,
      "loss": 0.3824,
      "step": 15661
    },
    {
      "epoch": 0.7187370932953971,
      "grad_norm": 0.4439581036567688,
      "learning_rate": 8.811666743797884e-06,
      "loss": 0.3568,
      "step": 15662
    },
    {
      "epoch": 0.7187829838006516,
      "grad_norm": 0.44223231077194214,
      "learning_rate": 8.811508057385229e-06,
      "loss": 0.357,
      "step": 15663
    },
    {
      "epoch": 0.7188288743059061,
      "grad_norm": 0.4481886327266693,
      "learning_rate": 8.811349361807102e-06,
      "loss": 0.3503,
      "step": 15664
    },
    {
      "epoch": 0.7188747648111605,
      "grad_norm": 0.4579429626464844,
      "learning_rate": 8.811190657063881e-06,
      "loss": 0.4495,
      "step": 15665
    },
    {
      "epoch": 0.718920655316415,
      "grad_norm": 0.4518534243106842,
      "learning_rate": 8.81103194315595e-06,
      "loss": 0.3189,
      "step": 15666
    },
    {
      "epoch": 0.7189665458216695,
      "grad_norm": 0.44195178151130676,
      "learning_rate": 8.810873220083688e-06,
      "loss": 0.347,
      "step": 15667
    },
    {
      "epoch": 0.7190124363269239,
      "grad_norm": 0.4850746989250183,
      "learning_rate": 8.810714487847483e-06,
      "loss": 0.4549,
      "step": 15668
    },
    {
      "epoch": 0.7190583268321784,
      "grad_norm": 0.46074604988098145,
      "learning_rate": 8.810555746447709e-06,
      "loss": 0.4036,
      "step": 15669
    },
    {
      "epoch": 0.7191042173374329,
      "grad_norm": 0.46106013655662537,
      "learning_rate": 8.810396995884751e-06,
      "loss": 0.3964,
      "step": 15670
    },
    {
      "epoch": 0.7191501078426874,
      "grad_norm": 0.45544198155403137,
      "learning_rate": 8.810238236158992e-06,
      "loss": 0.3777,
      "step": 15671
    },
    {
      "epoch": 0.7191959983479418,
      "grad_norm": 0.5385237336158752,
      "learning_rate": 8.810079467270812e-06,
      "loss": 0.47,
      "step": 15672
    },
    {
      "epoch": 0.7192418888531963,
      "grad_norm": 0.4805903434753418,
      "learning_rate": 8.809920689220594e-06,
      "loss": 0.4525,
      "step": 15673
    },
    {
      "epoch": 0.7192877793584508,
      "grad_norm": 0.45054641366004944,
      "learning_rate": 8.809761902008718e-06,
      "loss": 0.4088,
      "step": 15674
    },
    {
      "epoch": 0.7193336698637052,
      "grad_norm": 0.4502006471157074,
      "learning_rate": 8.809603105635569e-06,
      "loss": 0.3836,
      "step": 15675
    },
    {
      "epoch": 0.7193795603689597,
      "grad_norm": 0.47026869654655457,
      "learning_rate": 8.809444300101525e-06,
      "loss": 0.4403,
      "step": 15676
    },
    {
      "epoch": 0.7194254508742142,
      "grad_norm": 0.444906622171402,
      "learning_rate": 8.80928548540697e-06,
      "loss": 0.3636,
      "step": 15677
    },
    {
      "epoch": 0.7194713413794686,
      "grad_norm": 0.46094810962677,
      "learning_rate": 8.809126661552286e-06,
      "loss": 0.4027,
      "step": 15678
    },
    {
      "epoch": 0.719517231884723,
      "grad_norm": 0.4577856659889221,
      "learning_rate": 8.808967828537855e-06,
      "loss": 0.4096,
      "step": 15679
    },
    {
      "epoch": 0.7195631223899775,
      "grad_norm": 0.4108022451400757,
      "learning_rate": 8.808808986364056e-06,
      "loss": 0.3196,
      "step": 15680
    },
    {
      "epoch": 0.7196090128952319,
      "grad_norm": 0.4910483658313751,
      "learning_rate": 8.808650135031278e-06,
      "loss": 0.4724,
      "step": 15681
    },
    {
      "epoch": 0.7196549034004864,
      "grad_norm": 0.4647468030452728,
      "learning_rate": 8.808491274539896e-06,
      "loss": 0.4193,
      "step": 15682
    },
    {
      "epoch": 0.7197007939057409,
      "grad_norm": 0.4430696368217468,
      "learning_rate": 8.808332404890295e-06,
      "loss": 0.3394,
      "step": 15683
    },
    {
      "epoch": 0.7197466844109953,
      "grad_norm": 0.4280908703804016,
      "learning_rate": 8.808173526082856e-06,
      "loss": 0.3852,
      "step": 15684
    },
    {
      "epoch": 0.7197925749162498,
      "grad_norm": 0.4214204251766205,
      "learning_rate": 8.80801463811796e-06,
      "loss": 0.3362,
      "step": 15685
    },
    {
      "epoch": 0.7198384654215043,
      "grad_norm": 0.4595947861671448,
      "learning_rate": 8.807855740995994e-06,
      "loss": 0.3998,
      "step": 15686
    },
    {
      "epoch": 0.7198843559267587,
      "grad_norm": 0.45970043540000916,
      "learning_rate": 8.807696834717335e-06,
      "loss": 0.4393,
      "step": 15687
    },
    {
      "epoch": 0.7199302464320132,
      "grad_norm": 0.45524463057518005,
      "learning_rate": 8.807537919282365e-06,
      "loss": 0.3673,
      "step": 15688
    },
    {
      "epoch": 0.7199761369372677,
      "grad_norm": 0.4703691601753235,
      "learning_rate": 8.80737899469147e-06,
      "loss": 0.4272,
      "step": 15689
    },
    {
      "epoch": 0.7200220274425222,
      "grad_norm": 0.44374561309814453,
      "learning_rate": 8.80722006094503e-06,
      "loss": 0.4016,
      "step": 15690
    },
    {
      "epoch": 0.7200679179477766,
      "grad_norm": 0.46652352809906006,
      "learning_rate": 8.807061118043427e-06,
      "loss": 0.4595,
      "step": 15691
    },
    {
      "epoch": 0.7201138084530311,
      "grad_norm": 0.47145673632621765,
      "learning_rate": 8.806902165987045e-06,
      "loss": 0.4457,
      "step": 15692
    },
    {
      "epoch": 0.7201596989582856,
      "grad_norm": 0.45192602276802063,
      "learning_rate": 8.806743204776264e-06,
      "loss": 0.4388,
      "step": 15693
    },
    {
      "epoch": 0.72020558946354,
      "grad_norm": 0.42427805066108704,
      "learning_rate": 8.806584234411466e-06,
      "loss": 0.3501,
      "step": 15694
    },
    {
      "epoch": 0.7202514799687945,
      "grad_norm": 0.4368148446083069,
      "learning_rate": 8.806425254893037e-06,
      "loss": 0.3617,
      "step": 15695
    },
    {
      "epoch": 0.720297370474049,
      "grad_norm": 0.44040966033935547,
      "learning_rate": 8.806266266221354e-06,
      "loss": 0.3735,
      "step": 15696
    },
    {
      "epoch": 0.7203432609793033,
      "grad_norm": 0.4907754361629486,
      "learning_rate": 8.806107268396802e-06,
      "loss": 0.4621,
      "step": 15697
    },
    {
      "epoch": 0.7203891514845578,
      "grad_norm": 0.5049310922622681,
      "learning_rate": 8.805948261419763e-06,
      "loss": 0.5043,
      "step": 15698
    },
    {
      "epoch": 0.7204350419898123,
      "grad_norm": 0.4873790740966797,
      "learning_rate": 8.80578924529062e-06,
      "loss": 0.3838,
      "step": 15699
    },
    {
      "epoch": 0.7204809324950667,
      "grad_norm": 0.45794540643692017,
      "learning_rate": 8.805630220009757e-06,
      "loss": 0.3345,
      "step": 15700
    },
    {
      "epoch": 0.7205268230003212,
      "grad_norm": 0.44353345036506653,
      "learning_rate": 8.805471185577553e-06,
      "loss": 0.3902,
      "step": 15701
    },
    {
      "epoch": 0.7205727135055757,
      "grad_norm": 0.5043193101882935,
      "learning_rate": 8.80531214199439e-06,
      "loss": 0.4802,
      "step": 15702
    },
    {
      "epoch": 0.7206186040108301,
      "grad_norm": 0.46371254324913025,
      "learning_rate": 8.805153089260655e-06,
      "loss": 0.4082,
      "step": 15703
    },
    {
      "epoch": 0.7206644945160846,
      "grad_norm": 0.4520680606365204,
      "learning_rate": 8.804994027376726e-06,
      "loss": 0.3905,
      "step": 15704
    },
    {
      "epoch": 0.7207103850213391,
      "grad_norm": 0.4368646740913391,
      "learning_rate": 8.804834956342988e-06,
      "loss": 0.3622,
      "step": 15705
    },
    {
      "epoch": 0.7207562755265935,
      "grad_norm": 0.4347103238105774,
      "learning_rate": 8.804675876159824e-06,
      "loss": 0.3431,
      "step": 15706
    },
    {
      "epoch": 0.720802166031848,
      "grad_norm": 0.47889891266822815,
      "learning_rate": 8.804516786827613e-06,
      "loss": 0.5021,
      "step": 15707
    },
    {
      "epoch": 0.7208480565371025,
      "grad_norm": 0.4814665615558624,
      "learning_rate": 8.804357688346741e-06,
      "loss": 0.441,
      "step": 15708
    },
    {
      "epoch": 0.720893947042357,
      "grad_norm": 0.45779088139533997,
      "learning_rate": 8.80419858071759e-06,
      "loss": 0.3927,
      "step": 15709
    },
    {
      "epoch": 0.7209398375476114,
      "grad_norm": 0.4778338670730591,
      "learning_rate": 8.80403946394054e-06,
      "loss": 0.4164,
      "step": 15710
    },
    {
      "epoch": 0.7209857280528659,
      "grad_norm": 0.4592438340187073,
      "learning_rate": 8.80388033801598e-06,
      "loss": 0.376,
      "step": 15711
    },
    {
      "epoch": 0.7210316185581204,
      "grad_norm": 0.4740516245365143,
      "learning_rate": 8.803721202944284e-06,
      "loss": 0.3791,
      "step": 15712
    },
    {
      "epoch": 0.7210775090633748,
      "grad_norm": 0.4968891441822052,
      "learning_rate": 8.80356205872584e-06,
      "loss": 0.4507,
      "step": 15713
    },
    {
      "epoch": 0.7211233995686293,
      "grad_norm": 0.46413061022758484,
      "learning_rate": 8.803402905361032e-06,
      "loss": 0.438,
      "step": 15714
    },
    {
      "epoch": 0.7211692900738838,
      "grad_norm": 0.4207325875759125,
      "learning_rate": 8.803243742850238e-06,
      "loss": 0.3337,
      "step": 15715
    },
    {
      "epoch": 0.7212151805791381,
      "grad_norm": 0.48167723417282104,
      "learning_rate": 8.803084571193842e-06,
      "loss": 0.4237,
      "step": 15716
    },
    {
      "epoch": 0.7212610710843926,
      "grad_norm": 0.4395751357078552,
      "learning_rate": 8.80292539039223e-06,
      "loss": 0.3688,
      "step": 15717
    },
    {
      "epoch": 0.7213069615896471,
      "grad_norm": 0.5004598498344421,
      "learning_rate": 8.802766200445783e-06,
      "loss": 0.45,
      "step": 15718
    },
    {
      "epoch": 0.7213528520949015,
      "grad_norm": 0.4956144094467163,
      "learning_rate": 8.802607001354882e-06,
      "loss": 0.4351,
      "step": 15719
    },
    {
      "epoch": 0.721398742600156,
      "grad_norm": 0.47222280502319336,
      "learning_rate": 8.802447793119913e-06,
      "loss": 0.4149,
      "step": 15720
    },
    {
      "epoch": 0.7214446331054105,
      "grad_norm": 0.4475928843021393,
      "learning_rate": 8.802288575741256e-06,
      "loss": 0.4144,
      "step": 15721
    },
    {
      "epoch": 0.7214905236106649,
      "grad_norm": 0.4386114478111267,
      "learning_rate": 8.802129349219295e-06,
      "loss": 0.3823,
      "step": 15722
    },
    {
      "epoch": 0.7215364141159194,
      "grad_norm": 0.4891201853752136,
      "learning_rate": 8.801970113554414e-06,
      "loss": 0.416,
      "step": 15723
    },
    {
      "epoch": 0.7215823046211739,
      "grad_norm": 0.445738285779953,
      "learning_rate": 8.801810868746993e-06,
      "loss": 0.3818,
      "step": 15724
    },
    {
      "epoch": 0.7216281951264284,
      "grad_norm": 0.49361205101013184,
      "learning_rate": 8.801651614797418e-06,
      "loss": 0.4799,
      "step": 15725
    },
    {
      "epoch": 0.7216740856316828,
      "grad_norm": 0.45539504289627075,
      "learning_rate": 8.80149235170607e-06,
      "loss": 0.3808,
      "step": 15726
    },
    {
      "epoch": 0.7217199761369373,
      "grad_norm": 0.46922767162323,
      "learning_rate": 8.801333079473332e-06,
      "loss": 0.3942,
      "step": 15727
    },
    {
      "epoch": 0.7217658666421918,
      "grad_norm": 0.43232956528663635,
      "learning_rate": 8.801173798099589e-06,
      "loss": 0.3545,
      "step": 15728
    },
    {
      "epoch": 0.7218117571474462,
      "grad_norm": 0.4611961543560028,
      "learning_rate": 8.801014507585223e-06,
      "loss": 0.4056,
      "step": 15729
    },
    {
      "epoch": 0.7218576476527007,
      "grad_norm": 0.4680825471878052,
      "learning_rate": 8.800855207930616e-06,
      "loss": 0.3757,
      "step": 15730
    },
    {
      "epoch": 0.7219035381579552,
      "grad_norm": 0.4400262236595154,
      "learning_rate": 8.800695899136153e-06,
      "loss": 0.3925,
      "step": 15731
    },
    {
      "epoch": 0.7219494286632095,
      "grad_norm": 0.4886402189731598,
      "learning_rate": 8.800536581202214e-06,
      "loss": 0.484,
      "step": 15732
    },
    {
      "epoch": 0.721995319168464,
      "grad_norm": 0.4436909854412079,
      "learning_rate": 8.800377254129185e-06,
      "loss": 0.3534,
      "step": 15733
    },
    {
      "epoch": 0.7220412096737185,
      "grad_norm": 0.440734326839447,
      "learning_rate": 8.800217917917449e-06,
      "loss": 0.4411,
      "step": 15734
    },
    {
      "epoch": 0.7220871001789729,
      "grad_norm": 0.45247772336006165,
      "learning_rate": 8.800058572567387e-06,
      "loss": 0.4142,
      "step": 15735
    },
    {
      "epoch": 0.7221329906842274,
      "grad_norm": 0.47396689653396606,
      "learning_rate": 8.799899218079384e-06,
      "loss": 0.3968,
      "step": 15736
    },
    {
      "epoch": 0.7221788811894819,
      "grad_norm": 0.42098331451416016,
      "learning_rate": 8.799739854453821e-06,
      "loss": 0.3176,
      "step": 15737
    },
    {
      "epoch": 0.7222247716947363,
      "grad_norm": 0.44899746775627136,
      "learning_rate": 8.799580481691084e-06,
      "loss": 0.4095,
      "step": 15738
    },
    {
      "epoch": 0.7222706621999908,
      "grad_norm": 0.45648160576820374,
      "learning_rate": 8.799421099791557e-06,
      "loss": 0.412,
      "step": 15739
    },
    {
      "epoch": 0.7223165527052453,
      "grad_norm": 0.4719427824020386,
      "learning_rate": 8.79926170875562e-06,
      "loss": 0.4832,
      "step": 15740
    },
    {
      "epoch": 0.7223624432104997,
      "grad_norm": 0.4499124586582184,
      "learning_rate": 8.799102308583656e-06,
      "loss": 0.3783,
      "step": 15741
    },
    {
      "epoch": 0.7224083337157542,
      "grad_norm": 0.4321404993534088,
      "learning_rate": 8.798942899276052e-06,
      "loss": 0.3812,
      "step": 15742
    },
    {
      "epoch": 0.7224542242210087,
      "grad_norm": 0.47287285327911377,
      "learning_rate": 8.798783480833189e-06,
      "loss": 0.442,
      "step": 15743
    },
    {
      "epoch": 0.7225001147262632,
      "grad_norm": 0.4085655212402344,
      "learning_rate": 8.798624053255451e-06,
      "loss": 0.2896,
      "step": 15744
    },
    {
      "epoch": 0.7225460052315176,
      "grad_norm": 0.4424719512462616,
      "learning_rate": 8.79846461654322e-06,
      "loss": 0.3453,
      "step": 15745
    },
    {
      "epoch": 0.7225918957367721,
      "grad_norm": 0.4693467617034912,
      "learning_rate": 8.79830517069688e-06,
      "loss": 0.4388,
      "step": 15746
    },
    {
      "epoch": 0.7226377862420266,
      "grad_norm": 0.4066821336746216,
      "learning_rate": 8.798145715716817e-06,
      "loss": 0.3207,
      "step": 15747
    },
    {
      "epoch": 0.722683676747281,
      "grad_norm": 0.4815947115421295,
      "learning_rate": 8.79798625160341e-06,
      "loss": 0.4453,
      "step": 15748
    },
    {
      "epoch": 0.7227295672525355,
      "grad_norm": 0.3973434865474701,
      "learning_rate": 8.797826778357046e-06,
      "loss": 0.2875,
      "step": 15749
    },
    {
      "epoch": 0.72277545775779,
      "grad_norm": 0.4871484339237213,
      "learning_rate": 8.797667295978108e-06,
      "loss": 0.4225,
      "step": 15750
    },
    {
      "epoch": 0.7228213482630443,
      "grad_norm": 0.4572736620903015,
      "learning_rate": 8.797507804466978e-06,
      "loss": 0.4126,
      "step": 15751
    },
    {
      "epoch": 0.7228672387682988,
      "grad_norm": 0.4817626178264618,
      "learning_rate": 8.797348303824038e-06,
      "loss": 0.491,
      "step": 15752
    },
    {
      "epoch": 0.7229131292735533,
      "grad_norm": 0.4603731036186218,
      "learning_rate": 8.797188794049676e-06,
      "loss": 0.4229,
      "step": 15753
    },
    {
      "epoch": 0.7229590197788077,
      "grad_norm": 0.4694717228412628,
      "learning_rate": 8.797029275144274e-06,
      "loss": 0.4637,
      "step": 15754
    },
    {
      "epoch": 0.7230049102840622,
      "grad_norm": 0.42534172534942627,
      "learning_rate": 8.796869747108214e-06,
      "loss": 0.3375,
      "step": 15755
    },
    {
      "epoch": 0.7230508007893167,
      "grad_norm": 0.4359208941459656,
      "learning_rate": 8.79671020994188e-06,
      "loss": 0.342,
      "step": 15756
    },
    {
      "epoch": 0.7230966912945711,
      "grad_norm": 0.48801472783088684,
      "learning_rate": 8.796550663645657e-06,
      "loss": 0.4459,
      "step": 15757
    },
    {
      "epoch": 0.7231425817998256,
      "grad_norm": 0.4650600850582123,
      "learning_rate": 8.796391108219928e-06,
      "loss": 0.457,
      "step": 15758
    },
    {
      "epoch": 0.7231884723050801,
      "grad_norm": 0.46825850009918213,
      "learning_rate": 8.796231543665077e-06,
      "loss": 0.3878,
      "step": 15759
    },
    {
      "epoch": 0.7232343628103346,
      "grad_norm": 0.4077177047729492,
      "learning_rate": 8.796071969981487e-06,
      "loss": 0.3022,
      "step": 15760
    },
    {
      "epoch": 0.723280253315589,
      "grad_norm": 0.486931711435318,
      "learning_rate": 8.79591238716954e-06,
      "loss": 0.3872,
      "step": 15761
    },
    {
      "epoch": 0.7233261438208435,
      "grad_norm": 0.4552956223487854,
      "learning_rate": 8.795752795229624e-06,
      "loss": 0.3953,
      "step": 15762
    },
    {
      "epoch": 0.723372034326098,
      "grad_norm": 0.4604277014732361,
      "learning_rate": 8.79559319416212e-06,
      "loss": 0.4052,
      "step": 15763
    },
    {
      "epoch": 0.7234179248313524,
      "grad_norm": 0.41771620512008667,
      "learning_rate": 8.795433583967412e-06,
      "loss": 0.3204,
      "step": 15764
    },
    {
      "epoch": 0.7234638153366069,
      "grad_norm": 0.435382217168808,
      "learning_rate": 8.795273964645886e-06,
      "loss": 0.3855,
      "step": 15765
    },
    {
      "epoch": 0.7235097058418614,
      "grad_norm": 0.43797045946121216,
      "learning_rate": 8.79511433619792e-06,
      "loss": 0.321,
      "step": 15766
    },
    {
      "epoch": 0.7235555963471157,
      "grad_norm": 0.466752290725708,
      "learning_rate": 8.794954698623905e-06,
      "loss": 0.375,
      "step": 15767
    },
    {
      "epoch": 0.7236014868523702,
      "grad_norm": 0.42924177646636963,
      "learning_rate": 8.794795051924221e-06,
      "loss": 0.308,
      "step": 15768
    },
    {
      "epoch": 0.7236473773576247,
      "grad_norm": 0.46656152606010437,
      "learning_rate": 8.794635396099252e-06,
      "loss": 0.3575,
      "step": 15769
    },
    {
      "epoch": 0.7236932678628791,
      "grad_norm": 0.43610289692878723,
      "learning_rate": 8.794475731149384e-06,
      "loss": 0.3506,
      "step": 15770
    },
    {
      "epoch": 0.7237391583681336,
      "grad_norm": 0.44478684663772583,
      "learning_rate": 8.794316057074996e-06,
      "loss": 0.3658,
      "step": 15771
    },
    {
      "epoch": 0.7237850488733881,
      "grad_norm": 0.4541526734828949,
      "learning_rate": 8.79415637387648e-06,
      "loss": 0.3795,
      "step": 15772
    },
    {
      "epoch": 0.7238309393786425,
      "grad_norm": 0.4828837513923645,
      "learning_rate": 8.79399668155421e-06,
      "loss": 0.4326,
      "step": 15773
    },
    {
      "epoch": 0.723876829883897,
      "grad_norm": 0.4287390410900116,
      "learning_rate": 8.79383698010858e-06,
      "loss": 0.3495,
      "step": 15774
    },
    {
      "epoch": 0.7239227203891515,
      "grad_norm": 0.4825621247291565,
      "learning_rate": 8.793677269539967e-06,
      "loss": 0.4722,
      "step": 15775
    },
    {
      "epoch": 0.7239686108944059,
      "grad_norm": 0.44831010699272156,
      "learning_rate": 8.793517549848757e-06,
      "loss": 0.3835,
      "step": 15776
    },
    {
      "epoch": 0.7240145013996604,
      "grad_norm": 0.4726196527481079,
      "learning_rate": 8.793357821035338e-06,
      "loss": 0.407,
      "step": 15777
    },
    {
      "epoch": 0.7240603919049149,
      "grad_norm": 0.5705484747886658,
      "learning_rate": 8.793198083100089e-06,
      "loss": 0.4894,
      "step": 15778
    },
    {
      "epoch": 0.7241062824101694,
      "grad_norm": 0.4445796608924866,
      "learning_rate": 8.793038336043394e-06,
      "loss": 0.3307,
      "step": 15779
    },
    {
      "epoch": 0.7241521729154238,
      "grad_norm": 0.4393491744995117,
      "learning_rate": 8.79287857986564e-06,
      "loss": 0.4233,
      "step": 15780
    },
    {
      "epoch": 0.7241980634206783,
      "grad_norm": 0.6517716646194458,
      "learning_rate": 8.792718814567212e-06,
      "loss": 0.3973,
      "step": 15781
    },
    {
      "epoch": 0.7242439539259328,
      "grad_norm": 0.4433182179927826,
      "learning_rate": 8.79255904014849e-06,
      "loss": 0.3467,
      "step": 15782
    },
    {
      "epoch": 0.7242898444311872,
      "grad_norm": 0.48077258467674255,
      "learning_rate": 8.792399256609861e-06,
      "loss": 0.423,
      "step": 15783
    },
    {
      "epoch": 0.7243357349364417,
      "grad_norm": 0.4420410096645355,
      "learning_rate": 8.792239463951708e-06,
      "loss": 0.361,
      "step": 15784
    },
    {
      "epoch": 0.7243816254416962,
      "grad_norm": 0.39362818002700806,
      "learning_rate": 8.792079662174418e-06,
      "loss": 0.2834,
      "step": 15785
    },
    {
      "epoch": 0.7244275159469505,
      "grad_norm": 0.506283700466156,
      "learning_rate": 8.791919851278371e-06,
      "loss": 0.482,
      "step": 15786
    },
    {
      "epoch": 0.724473406452205,
      "grad_norm": 0.4326857030391693,
      "learning_rate": 8.791760031263954e-06,
      "loss": 0.3553,
      "step": 15787
    },
    {
      "epoch": 0.7245192969574595,
      "grad_norm": 0.4446062445640564,
      "learning_rate": 8.791600202131552e-06,
      "loss": 0.387,
      "step": 15788
    },
    {
      "epoch": 0.7245651874627139,
      "grad_norm": 0.46657004952430725,
      "learning_rate": 8.791440363881547e-06,
      "loss": 0.4079,
      "step": 15789
    },
    {
      "epoch": 0.7246110779679684,
      "grad_norm": 0.44427162408828735,
      "learning_rate": 8.791280516514325e-06,
      "loss": 0.3856,
      "step": 15790
    },
    {
      "epoch": 0.7246569684732229,
      "grad_norm": 0.4696272313594818,
      "learning_rate": 8.79112066003027e-06,
      "loss": 0.4756,
      "step": 15791
    },
    {
      "epoch": 0.7247028589784773,
      "grad_norm": 0.4774477481842041,
      "learning_rate": 8.790960794429765e-06,
      "loss": 0.4132,
      "step": 15792
    },
    {
      "epoch": 0.7247487494837318,
      "grad_norm": 0.4282197654247284,
      "learning_rate": 8.790800919713198e-06,
      "loss": 0.3524,
      "step": 15793
    },
    {
      "epoch": 0.7247946399889863,
      "grad_norm": 0.41371479630470276,
      "learning_rate": 8.79064103588095e-06,
      "loss": 0.3509,
      "step": 15794
    },
    {
      "epoch": 0.7248405304942407,
      "grad_norm": 0.5138885378837585,
      "learning_rate": 8.790481142933406e-06,
      "loss": 0.4376,
      "step": 15795
    },
    {
      "epoch": 0.7248864209994952,
      "grad_norm": 0.4904124438762665,
      "learning_rate": 8.790321240870954e-06,
      "loss": 0.3749,
      "step": 15796
    },
    {
      "epoch": 0.7249323115047497,
      "grad_norm": 0.47268912196159363,
      "learning_rate": 8.790161329693974e-06,
      "loss": 0.5133,
      "step": 15797
    },
    {
      "epoch": 0.7249782020100042,
      "grad_norm": 0.48660027980804443,
      "learning_rate": 8.790001409402851e-06,
      "loss": 0.3938,
      "step": 15798
    },
    {
      "epoch": 0.7250240925152586,
      "grad_norm": 0.46966612339019775,
      "learning_rate": 8.789841479997971e-06,
      "loss": 0.4424,
      "step": 15799
    },
    {
      "epoch": 0.7250699830205131,
      "grad_norm": 0.4524163007736206,
      "learning_rate": 8.78968154147972e-06,
      "loss": 0.3697,
      "step": 15800
    },
    {
      "epoch": 0.7251158735257676,
      "grad_norm": 0.43152281641960144,
      "learning_rate": 8.78952159384848e-06,
      "loss": 0.3336,
      "step": 15801
    },
    {
      "epoch": 0.725161764031022,
      "grad_norm": 0.48487457633018494,
      "learning_rate": 8.789361637104636e-06,
      "loss": 0.4048,
      "step": 15802
    },
    {
      "epoch": 0.7252076545362764,
      "grad_norm": 0.44444575905799866,
      "learning_rate": 8.789201671248572e-06,
      "loss": 0.3474,
      "step": 15803
    },
    {
      "epoch": 0.725253545041531,
      "grad_norm": 0.43307289481163025,
      "learning_rate": 8.789041696280678e-06,
      "loss": 0.3929,
      "step": 15804
    },
    {
      "epoch": 0.7252994355467853,
      "grad_norm": 0.45821964740753174,
      "learning_rate": 8.78888171220133e-06,
      "loss": 0.4303,
      "step": 15805
    },
    {
      "epoch": 0.7253453260520398,
      "grad_norm": 0.4718412458896637,
      "learning_rate": 8.788721719010922e-06,
      "loss": 0.4431,
      "step": 15806
    },
    {
      "epoch": 0.7253912165572943,
      "grad_norm": 0.43296968936920166,
      "learning_rate": 8.788561716709831e-06,
      "loss": 0.3778,
      "step": 15807
    },
    {
      "epoch": 0.7254371070625487,
      "grad_norm": 0.47573763132095337,
      "learning_rate": 8.788401705298445e-06,
      "loss": 0.4566,
      "step": 15808
    },
    {
      "epoch": 0.7254829975678032,
      "grad_norm": 0.44513794779777527,
      "learning_rate": 8.78824168477715e-06,
      "loss": 0.3362,
      "step": 15809
    },
    {
      "epoch": 0.7255288880730577,
      "grad_norm": 0.40864741802215576,
      "learning_rate": 8.788081655146328e-06,
      "loss": 0.3464,
      "step": 15810
    },
    {
      "epoch": 0.7255747785783121,
      "grad_norm": 0.4739340841770172,
      "learning_rate": 8.787921616406366e-06,
      "loss": 0.42,
      "step": 15811
    },
    {
      "epoch": 0.7256206690835666,
      "grad_norm": 0.45560669898986816,
      "learning_rate": 8.787761568557646e-06,
      "loss": 0.3421,
      "step": 15812
    },
    {
      "epoch": 0.7256665595888211,
      "grad_norm": 0.5029919147491455,
      "learning_rate": 8.787601511600556e-06,
      "loss": 0.4782,
      "step": 15813
    },
    {
      "epoch": 0.7257124500940756,
      "grad_norm": 0.4636812210083008,
      "learning_rate": 8.78744144553548e-06,
      "loss": 0.4033,
      "step": 15814
    },
    {
      "epoch": 0.72575834059933,
      "grad_norm": 0.45345252752304077,
      "learning_rate": 8.787281370362804e-06,
      "loss": 0.3715,
      "step": 15815
    },
    {
      "epoch": 0.7258042311045845,
      "grad_norm": 0.42035001516342163,
      "learning_rate": 8.78712128608291e-06,
      "loss": 0.3218,
      "step": 15816
    },
    {
      "epoch": 0.725850121609839,
      "grad_norm": 0.43998706340789795,
      "learning_rate": 8.786961192696185e-06,
      "loss": 0.3663,
      "step": 15817
    },
    {
      "epoch": 0.7258960121150934,
      "grad_norm": 0.4895240068435669,
      "learning_rate": 8.786801090203012e-06,
      "loss": 0.4713,
      "step": 15818
    },
    {
      "epoch": 0.7259419026203479,
      "grad_norm": 0.47725537419319153,
      "learning_rate": 8.78664097860378e-06,
      "loss": 0.4439,
      "step": 15819
    },
    {
      "epoch": 0.7259877931256024,
      "grad_norm": 0.45937857031822205,
      "learning_rate": 8.78648085789887e-06,
      "loss": 0.3434,
      "step": 15820
    },
    {
      "epoch": 0.7260336836308567,
      "grad_norm": 0.4288168251514435,
      "learning_rate": 8.786320728088668e-06,
      "loss": 0.3238,
      "step": 15821
    },
    {
      "epoch": 0.7260795741361112,
      "grad_norm": 0.512618899345398,
      "learning_rate": 8.786160589173563e-06,
      "loss": 0.486,
      "step": 15822
    },
    {
      "epoch": 0.7261254646413657,
      "grad_norm": 0.4348469078540802,
      "learning_rate": 8.786000441153934e-06,
      "loss": 0.3983,
      "step": 15823
    },
    {
      "epoch": 0.7261713551466201,
      "grad_norm": 0.47717514634132385,
      "learning_rate": 8.785840284030169e-06,
      "loss": 0.4099,
      "step": 15824
    },
    {
      "epoch": 0.7262172456518746,
      "grad_norm": 0.4558356702327728,
      "learning_rate": 8.785680117802654e-06,
      "loss": 0.4238,
      "step": 15825
    },
    {
      "epoch": 0.7262631361571291,
      "grad_norm": 0.4055575728416443,
      "learning_rate": 8.785519942471772e-06,
      "loss": 0.3053,
      "step": 15826
    },
    {
      "epoch": 0.7263090266623835,
      "grad_norm": 0.5978828072547913,
      "learning_rate": 8.785359758037908e-06,
      "loss": 0.4399,
      "step": 15827
    },
    {
      "epoch": 0.726354917167638,
      "grad_norm": 0.44162386655807495,
      "learning_rate": 8.785199564501451e-06,
      "loss": 0.3508,
      "step": 15828
    },
    {
      "epoch": 0.7264008076728925,
      "grad_norm": 0.4567866623401642,
      "learning_rate": 8.785039361862783e-06,
      "loss": 0.4367,
      "step": 15829
    },
    {
      "epoch": 0.7264466981781469,
      "grad_norm": 0.4178774058818817,
      "learning_rate": 8.784879150122289e-06,
      "loss": 0.3169,
      "step": 15830
    },
    {
      "epoch": 0.7264925886834014,
      "grad_norm": 0.4412555396556854,
      "learning_rate": 8.784718929280356e-06,
      "loss": 0.4174,
      "step": 15831
    },
    {
      "epoch": 0.7265384791886559,
      "grad_norm": 0.4471972584724426,
      "learning_rate": 8.784558699337368e-06,
      "loss": 0.3945,
      "step": 15832
    },
    {
      "epoch": 0.7265843696939104,
      "grad_norm": 0.4341907501220703,
      "learning_rate": 8.784398460293712e-06,
      "loss": 0.3805,
      "step": 15833
    },
    {
      "epoch": 0.7266302601991648,
      "grad_norm": 0.479985773563385,
      "learning_rate": 8.784238212149771e-06,
      "loss": 0.3843,
      "step": 15834
    },
    {
      "epoch": 0.7266761507044193,
      "grad_norm": 0.48687973618507385,
      "learning_rate": 8.784077954905932e-06,
      "loss": 0.5055,
      "step": 15835
    },
    {
      "epoch": 0.7267220412096738,
      "grad_norm": 0.47694897651672363,
      "learning_rate": 8.78391768856258e-06,
      "loss": 0.469,
      "step": 15836
    },
    {
      "epoch": 0.7267679317149282,
      "grad_norm": 0.45083072781562805,
      "learning_rate": 8.7837574131201e-06,
      "loss": 0.4119,
      "step": 15837
    },
    {
      "epoch": 0.7268138222201826,
      "grad_norm": 0.4373519718647003,
      "learning_rate": 8.783597128578877e-06,
      "loss": 0.3641,
      "step": 15838
    },
    {
      "epoch": 0.7268597127254371,
      "grad_norm": 0.4509945511817932,
      "learning_rate": 8.783436834939297e-06,
      "loss": 0.3495,
      "step": 15839
    },
    {
      "epoch": 0.7269056032306915,
      "grad_norm": 0.47613340616226196,
      "learning_rate": 8.783276532201745e-06,
      "loss": 0.403,
      "step": 15840
    },
    {
      "epoch": 0.726951493735946,
      "grad_norm": 6.271466255187988,
      "learning_rate": 8.783116220366608e-06,
      "loss": 0.5028,
      "step": 15841
    },
    {
      "epoch": 0.7269973842412005,
      "grad_norm": 0.39961832761764526,
      "learning_rate": 8.782955899434272e-06,
      "loss": 0.2846,
      "step": 15842
    },
    {
      "epoch": 0.7270432747464549,
      "grad_norm": 0.4274345934391022,
      "learning_rate": 8.782795569405118e-06,
      "loss": 0.348,
      "step": 15843
    },
    {
      "epoch": 0.7270891652517094,
      "grad_norm": 0.4682891368865967,
      "learning_rate": 8.782635230279536e-06,
      "loss": 0.3889,
      "step": 15844
    },
    {
      "epoch": 0.7271350557569639,
      "grad_norm": 0.5083214640617371,
      "learning_rate": 8.78247488205791e-06,
      "loss": 0.4117,
      "step": 15845
    },
    {
      "epoch": 0.7271809462622183,
      "grad_norm": 0.4408183991909027,
      "learning_rate": 8.782314524740625e-06,
      "loss": 0.3956,
      "step": 15846
    },
    {
      "epoch": 0.7272268367674728,
      "grad_norm": 0.4753551483154297,
      "learning_rate": 8.782154158328069e-06,
      "loss": 0.4564,
      "step": 15847
    },
    {
      "epoch": 0.7272727272727273,
      "grad_norm": 0.4275411367416382,
      "learning_rate": 8.781993782820624e-06,
      "loss": 0.3444,
      "step": 15848
    },
    {
      "epoch": 0.7273186177779818,
      "grad_norm": 0.44078466296195984,
      "learning_rate": 8.781833398218679e-06,
      "loss": 0.3627,
      "step": 15849
    },
    {
      "epoch": 0.7273645082832362,
      "grad_norm": 0.469561904668808,
      "learning_rate": 8.781673004522618e-06,
      "loss": 0.4105,
      "step": 15850
    },
    {
      "epoch": 0.7274103987884907,
      "grad_norm": 0.5032742023468018,
      "learning_rate": 8.781512601732827e-06,
      "loss": 0.477,
      "step": 15851
    },
    {
      "epoch": 0.7274562892937452,
      "grad_norm": 0.5570603013038635,
      "learning_rate": 8.78135218984969e-06,
      "loss": 0.5461,
      "step": 15852
    },
    {
      "epoch": 0.7275021797989996,
      "grad_norm": 0.4768106937408447,
      "learning_rate": 8.781191768873596e-06,
      "loss": 0.4157,
      "step": 15853
    },
    {
      "epoch": 0.7275480703042541,
      "grad_norm": 0.4715902805328369,
      "learning_rate": 8.781031338804928e-06,
      "loss": 0.3883,
      "step": 15854
    },
    {
      "epoch": 0.7275939608095086,
      "grad_norm": 0.495225727558136,
      "learning_rate": 8.780870899644075e-06,
      "loss": 0.5223,
      "step": 15855
    },
    {
      "epoch": 0.7276398513147629,
      "grad_norm": 0.47783002257347107,
      "learning_rate": 8.78071045139142e-06,
      "loss": 0.4299,
      "step": 15856
    },
    {
      "epoch": 0.7276857418200174,
      "grad_norm": 0.4998358488082886,
      "learning_rate": 8.780549994047349e-06,
      "loss": 0.466,
      "step": 15857
    },
    {
      "epoch": 0.7277316323252719,
      "grad_norm": 0.46260157227516174,
      "learning_rate": 8.780389527612249e-06,
      "loss": 0.4372,
      "step": 15858
    },
    {
      "epoch": 0.7277775228305263,
      "grad_norm": 0.6398410797119141,
      "learning_rate": 8.780229052086504e-06,
      "loss": 0.4804,
      "step": 15859
    },
    {
      "epoch": 0.7278234133357808,
      "grad_norm": 0.4729681611061096,
      "learning_rate": 8.780068567470504e-06,
      "loss": 0.3669,
      "step": 15860
    },
    {
      "epoch": 0.7278693038410353,
      "grad_norm": 0.44764643907546997,
      "learning_rate": 8.779908073764629e-06,
      "loss": 0.3719,
      "step": 15861
    },
    {
      "epoch": 0.7279151943462897,
      "grad_norm": 0.48311564326286316,
      "learning_rate": 8.779747570969269e-06,
      "loss": 0.4057,
      "step": 15862
    },
    {
      "epoch": 0.7279610848515442,
      "grad_norm": 0.47539085149765015,
      "learning_rate": 8.77958705908481e-06,
      "loss": 0.4121,
      "step": 15863
    },
    {
      "epoch": 0.7280069753567987,
      "grad_norm": 0.45124635100364685,
      "learning_rate": 8.779426538111636e-06,
      "loss": 0.4226,
      "step": 15864
    },
    {
      "epoch": 0.7280528658620531,
      "grad_norm": 0.48039737343788147,
      "learning_rate": 8.779266008050134e-06,
      "loss": 0.4472,
      "step": 15865
    },
    {
      "epoch": 0.7280987563673076,
      "grad_norm": 0.4407140612602234,
      "learning_rate": 8.779105468900691e-06,
      "loss": 0.3745,
      "step": 15866
    },
    {
      "epoch": 0.7281446468725621,
      "grad_norm": 0.4737776815891266,
      "learning_rate": 8.778944920663692e-06,
      "loss": 0.4633,
      "step": 15867
    },
    {
      "epoch": 0.7281905373778166,
      "grad_norm": 0.4528125822544098,
      "learning_rate": 8.778784363339522e-06,
      "loss": 0.3636,
      "step": 15868
    },
    {
      "epoch": 0.728236427883071,
      "grad_norm": 0.4774063527584076,
      "learning_rate": 8.778623796928569e-06,
      "loss": 0.4024,
      "step": 15869
    },
    {
      "epoch": 0.7282823183883255,
      "grad_norm": 0.4615359902381897,
      "learning_rate": 8.778463221431219e-06,
      "loss": 0.4051,
      "step": 15870
    },
    {
      "epoch": 0.72832820889358,
      "grad_norm": 0.5139020681381226,
      "learning_rate": 8.778302636847855e-06,
      "loss": 0.4715,
      "step": 15871
    },
    {
      "epoch": 0.7283740993988344,
      "grad_norm": 0.44127950072288513,
      "learning_rate": 8.778142043178868e-06,
      "loss": 0.3549,
      "step": 15872
    },
    {
      "epoch": 0.7284199899040888,
      "grad_norm": 0.4397776126861572,
      "learning_rate": 8.777981440424641e-06,
      "loss": 0.3408,
      "step": 15873
    },
    {
      "epoch": 0.7284658804093433,
      "grad_norm": 0.4101635813713074,
      "learning_rate": 8.777820828585561e-06,
      "loss": 0.3016,
      "step": 15874
    },
    {
      "epoch": 0.7285117709145977,
      "grad_norm": 0.48236075043678284,
      "learning_rate": 8.777660207662014e-06,
      "loss": 0.446,
      "step": 15875
    },
    {
      "epoch": 0.7285576614198522,
      "grad_norm": 0.46139639616012573,
      "learning_rate": 8.777499577654387e-06,
      "loss": 0.4006,
      "step": 15876
    },
    {
      "epoch": 0.7286035519251067,
      "grad_norm": 0.41257408261299133,
      "learning_rate": 8.777338938563065e-06,
      "loss": 0.2957,
      "step": 15877
    },
    {
      "epoch": 0.7286494424303611,
      "grad_norm": 0.49086636304855347,
      "learning_rate": 8.777178290388437e-06,
      "loss": 0.4718,
      "step": 15878
    },
    {
      "epoch": 0.7286953329356156,
      "grad_norm": 0.47208768129348755,
      "learning_rate": 8.777017633130884e-06,
      "loss": 0.4255,
      "step": 15879
    },
    {
      "epoch": 0.7287412234408701,
      "grad_norm": 0.4711315631866455,
      "learning_rate": 8.776856966790799e-06,
      "loss": 0.4034,
      "step": 15880
    },
    {
      "epoch": 0.7287871139461245,
      "grad_norm": 0.4615982174873352,
      "learning_rate": 8.776696291368562e-06,
      "loss": 0.3992,
      "step": 15881
    },
    {
      "epoch": 0.728833004451379,
      "grad_norm": 0.4622534513473511,
      "learning_rate": 8.776535606864565e-06,
      "loss": 0.4563,
      "step": 15882
    },
    {
      "epoch": 0.7288788949566335,
      "grad_norm": 0.43618839979171753,
      "learning_rate": 8.77637491327919e-06,
      "loss": 0.4097,
      "step": 15883
    },
    {
      "epoch": 0.7289247854618879,
      "grad_norm": 0.4568190276622772,
      "learning_rate": 8.776214210612825e-06,
      "loss": 0.4119,
      "step": 15884
    },
    {
      "epoch": 0.7289706759671424,
      "grad_norm": 0.4401451647281647,
      "learning_rate": 8.776053498865858e-06,
      "loss": 0.3797,
      "step": 15885
    },
    {
      "epoch": 0.7290165664723969,
      "grad_norm": 0.4336271584033966,
      "learning_rate": 8.775892778038675e-06,
      "loss": 0.3034,
      "step": 15886
    },
    {
      "epoch": 0.7290624569776514,
      "grad_norm": 0.4753570556640625,
      "learning_rate": 8.77573204813166e-06,
      "loss": 0.427,
      "step": 15887
    },
    {
      "epoch": 0.7291083474829058,
      "grad_norm": 0.4271291196346283,
      "learning_rate": 8.775571309145201e-06,
      "loss": 0.3838,
      "step": 15888
    },
    {
      "epoch": 0.7291542379881603,
      "grad_norm": 0.4823809862136841,
      "learning_rate": 8.775410561079684e-06,
      "loss": 0.3684,
      "step": 15889
    },
    {
      "epoch": 0.7292001284934148,
      "grad_norm": 0.4168423116207123,
      "learning_rate": 8.775249803935497e-06,
      "loss": 0.3519,
      "step": 15890
    },
    {
      "epoch": 0.7292460189986691,
      "grad_norm": 0.4381709694862366,
      "learning_rate": 8.775089037713026e-06,
      "loss": 0.3568,
      "step": 15891
    },
    {
      "epoch": 0.7292919095039236,
      "grad_norm": 0.45209264755249023,
      "learning_rate": 8.774928262412657e-06,
      "loss": 0.3538,
      "step": 15892
    },
    {
      "epoch": 0.7293378000091781,
      "grad_norm": 0.5010458827018738,
      "learning_rate": 8.774767478034776e-06,
      "loss": 0.4856,
      "step": 15893
    },
    {
      "epoch": 0.7293836905144325,
      "grad_norm": 0.48654666543006897,
      "learning_rate": 8.774606684579772e-06,
      "loss": 0.4921,
      "step": 15894
    },
    {
      "epoch": 0.729429581019687,
      "grad_norm": 0.479793906211853,
      "learning_rate": 8.77444588204803e-06,
      "loss": 0.4094,
      "step": 15895
    },
    {
      "epoch": 0.7294754715249415,
      "grad_norm": 0.4602452218532562,
      "learning_rate": 8.774285070439936e-06,
      "loss": 0.3709,
      "step": 15896
    },
    {
      "epoch": 0.7295213620301959,
      "grad_norm": 0.4258638918399811,
      "learning_rate": 8.77412424975588e-06,
      "loss": 0.3343,
      "step": 15897
    },
    {
      "epoch": 0.7295672525354504,
      "grad_norm": 0.5356239080429077,
      "learning_rate": 8.773963419996243e-06,
      "loss": 0.5135,
      "step": 15898
    },
    {
      "epoch": 0.7296131430407049,
      "grad_norm": 0.4787670373916626,
      "learning_rate": 8.773802581161417e-06,
      "loss": 0.4878,
      "step": 15899
    },
    {
      "epoch": 0.7296590335459593,
      "grad_norm": 0.4582371711730957,
      "learning_rate": 8.773641733251786e-06,
      "loss": 0.3828,
      "step": 15900
    },
    {
      "epoch": 0.7297049240512138,
      "grad_norm": 0.42731910943984985,
      "learning_rate": 8.773480876267739e-06,
      "loss": 0.3633,
      "step": 15901
    },
    {
      "epoch": 0.7297508145564683,
      "grad_norm": 0.4013974368572235,
      "learning_rate": 8.77332001020966e-06,
      "loss": 0.3134,
      "step": 15902
    },
    {
      "epoch": 0.7297967050617228,
      "grad_norm": 0.46906599402427673,
      "learning_rate": 8.773159135077937e-06,
      "loss": 0.3933,
      "step": 15903
    },
    {
      "epoch": 0.7298425955669772,
      "grad_norm": 0.45351725816726685,
      "learning_rate": 8.77299825087296e-06,
      "loss": 0.3672,
      "step": 15904
    },
    {
      "epoch": 0.7298884860722317,
      "grad_norm": 0.4581536054611206,
      "learning_rate": 8.77283735759511e-06,
      "loss": 0.3884,
      "step": 15905
    },
    {
      "epoch": 0.7299343765774862,
      "grad_norm": 0.48480382561683655,
      "learning_rate": 8.772676455244777e-06,
      "loss": 0.3946,
      "step": 15906
    },
    {
      "epoch": 0.7299802670827406,
      "grad_norm": 0.46109631657600403,
      "learning_rate": 8.77251554382235e-06,
      "loss": 0.4245,
      "step": 15907
    },
    {
      "epoch": 0.730026157587995,
      "grad_norm": 0.4518281817436218,
      "learning_rate": 8.772354623328212e-06,
      "loss": 0.3814,
      "step": 15908
    },
    {
      "epoch": 0.7300720480932495,
      "grad_norm": 0.4148745834827423,
      "learning_rate": 8.772193693762752e-06,
      "loss": 0.3318,
      "step": 15909
    },
    {
      "epoch": 0.7301179385985039,
      "grad_norm": 0.4644750654697418,
      "learning_rate": 8.772032755126357e-06,
      "loss": 0.417,
      "step": 15910
    },
    {
      "epoch": 0.7301638291037584,
      "grad_norm": 0.48413005471229553,
      "learning_rate": 8.771871807419412e-06,
      "loss": 0.3517,
      "step": 15911
    },
    {
      "epoch": 0.7302097196090129,
      "grad_norm": 0.5228267908096313,
      "learning_rate": 8.771710850642307e-06,
      "loss": 0.4318,
      "step": 15912
    },
    {
      "epoch": 0.7302556101142673,
      "grad_norm": 0.44682028889656067,
      "learning_rate": 8.771549884795429e-06,
      "loss": 0.385,
      "step": 15913
    },
    {
      "epoch": 0.7303015006195218,
      "grad_norm": 0.4255032539367676,
      "learning_rate": 8.771388909879164e-06,
      "loss": 0.3339,
      "step": 15914
    },
    {
      "epoch": 0.7303473911247763,
      "grad_norm": 0.4471794068813324,
      "learning_rate": 8.771227925893897e-06,
      "loss": 0.3399,
      "step": 15915
    },
    {
      "epoch": 0.7303932816300307,
      "grad_norm": 0.46839335560798645,
      "learning_rate": 8.771066932840017e-06,
      "loss": 0.3697,
      "step": 15916
    },
    {
      "epoch": 0.7304391721352852,
      "grad_norm": 0.4998243451118469,
      "learning_rate": 8.770905930717912e-06,
      "loss": 0.3839,
      "step": 15917
    },
    {
      "epoch": 0.7304850626405397,
      "grad_norm": 0.44538918137550354,
      "learning_rate": 8.770744919527969e-06,
      "loss": 0.3545,
      "step": 15918
    },
    {
      "epoch": 0.7305309531457941,
      "grad_norm": 0.4177456200122833,
      "learning_rate": 8.770583899270574e-06,
      "loss": 0.3235,
      "step": 15919
    },
    {
      "epoch": 0.7305768436510486,
      "grad_norm": 0.45691508054733276,
      "learning_rate": 8.770422869946115e-06,
      "loss": 0.3793,
      "step": 15920
    },
    {
      "epoch": 0.7306227341563031,
      "grad_norm": 0.4530298411846161,
      "learning_rate": 8.77026183155498e-06,
      "loss": 0.3397,
      "step": 15921
    },
    {
      "epoch": 0.7306686246615576,
      "grad_norm": 0.4441429078578949,
      "learning_rate": 8.770100784097553e-06,
      "loss": 0.3753,
      "step": 15922
    },
    {
      "epoch": 0.730714515166812,
      "grad_norm": 0.43963178992271423,
      "learning_rate": 8.769939727574224e-06,
      "loss": 0.3572,
      "step": 15923
    },
    {
      "epoch": 0.7307604056720665,
      "grad_norm": 0.46432948112487793,
      "learning_rate": 8.769778661985381e-06,
      "loss": 0.3434,
      "step": 15924
    },
    {
      "epoch": 0.730806296177321,
      "grad_norm": 0.5117664337158203,
      "learning_rate": 8.769617587331409e-06,
      "loss": 0.3587,
      "step": 15925
    },
    {
      "epoch": 0.7308521866825753,
      "grad_norm": 0.442678838968277,
      "learning_rate": 8.769456503612698e-06,
      "loss": 0.3704,
      "step": 15926
    },
    {
      "epoch": 0.7308980771878298,
      "grad_norm": 0.4432217478752136,
      "learning_rate": 8.769295410829632e-06,
      "loss": 0.4025,
      "step": 15927
    },
    {
      "epoch": 0.7309439676930843,
      "grad_norm": 0.47961968183517456,
      "learning_rate": 8.769134308982601e-06,
      "loss": 0.4367,
      "step": 15928
    },
    {
      "epoch": 0.7309898581983387,
      "grad_norm": 0.4575701951980591,
      "learning_rate": 8.76897319807199e-06,
      "loss": 0.4375,
      "step": 15929
    },
    {
      "epoch": 0.7310357487035932,
      "grad_norm": 0.44129225611686707,
      "learning_rate": 8.76881207809819e-06,
      "loss": 0.3409,
      "step": 15930
    },
    {
      "epoch": 0.7310816392088477,
      "grad_norm": 0.4386305809020996,
      "learning_rate": 8.768650949061585e-06,
      "loss": 0.3593,
      "step": 15931
    },
    {
      "epoch": 0.7311275297141021,
      "grad_norm": 0.4412127435207367,
      "learning_rate": 8.768489810962567e-06,
      "loss": 0.3497,
      "step": 15932
    },
    {
      "epoch": 0.7311734202193566,
      "grad_norm": 0.4620092213153839,
      "learning_rate": 8.768328663801517e-06,
      "loss": 0.4123,
      "step": 15933
    },
    {
      "epoch": 0.7312193107246111,
      "grad_norm": 0.4103982746601105,
      "learning_rate": 8.768167507578827e-06,
      "loss": 0.3206,
      "step": 15934
    },
    {
      "epoch": 0.7312652012298655,
      "grad_norm": 0.4471447169780731,
      "learning_rate": 8.768006342294884e-06,
      "loss": 0.3564,
      "step": 15935
    },
    {
      "epoch": 0.73131109173512,
      "grad_norm": 0.49769216775894165,
      "learning_rate": 8.767845167950073e-06,
      "loss": 0.4991,
      "step": 15936
    },
    {
      "epoch": 0.7313569822403745,
      "grad_norm": 0.4922749400138855,
      "learning_rate": 8.767683984544786e-06,
      "loss": 0.5009,
      "step": 15937
    },
    {
      "epoch": 0.731402872745629,
      "grad_norm": 0.49424469470977783,
      "learning_rate": 8.767522792079405e-06,
      "loss": 0.4887,
      "step": 15938
    },
    {
      "epoch": 0.7314487632508834,
      "grad_norm": 0.46387025713920593,
      "learning_rate": 8.767361590554324e-06,
      "loss": 0.4332,
      "step": 15939
    },
    {
      "epoch": 0.7314946537561379,
      "grad_norm": 0.4701308310031891,
      "learning_rate": 8.767200379969925e-06,
      "loss": 0.4268,
      "step": 15940
    },
    {
      "epoch": 0.7315405442613924,
      "grad_norm": 0.4916800856590271,
      "learning_rate": 8.7670391603266e-06,
      "loss": 0.452,
      "step": 15941
    },
    {
      "epoch": 0.7315864347666468,
      "grad_norm": 0.4440118968486786,
      "learning_rate": 8.766877931624733e-06,
      "loss": 0.3528,
      "step": 15942
    },
    {
      "epoch": 0.7316323252719013,
      "grad_norm": 0.4619618356227875,
      "learning_rate": 8.766716693864714e-06,
      "loss": 0.4314,
      "step": 15943
    },
    {
      "epoch": 0.7316782157771557,
      "grad_norm": 0.45432421565055847,
      "learning_rate": 8.766555447046932e-06,
      "loss": 0.3938,
      "step": 15944
    },
    {
      "epoch": 0.7317241062824101,
      "grad_norm": 0.4804248809814453,
      "learning_rate": 8.766394191171772e-06,
      "loss": 0.4084,
      "step": 15945
    },
    {
      "epoch": 0.7317699967876646,
      "grad_norm": 0.4608505070209503,
      "learning_rate": 8.76623292623962e-06,
      "loss": 0.4376,
      "step": 15946
    },
    {
      "epoch": 0.7318158872929191,
      "grad_norm": 0.46721744537353516,
      "learning_rate": 8.76607165225087e-06,
      "loss": 0.4387,
      "step": 15947
    },
    {
      "epoch": 0.7318617777981735,
      "grad_norm": 0.4271980822086334,
      "learning_rate": 8.765910369205904e-06,
      "loss": 0.3398,
      "step": 15948
    },
    {
      "epoch": 0.731907668303428,
      "grad_norm": 0.4862414002418518,
      "learning_rate": 8.765749077105112e-06,
      "loss": 0.4595,
      "step": 15949
    },
    {
      "epoch": 0.7319535588086825,
      "grad_norm": 0.44620829820632935,
      "learning_rate": 8.765587775948884e-06,
      "loss": 0.3916,
      "step": 15950
    },
    {
      "epoch": 0.7319994493139369,
      "grad_norm": 0.4894838035106659,
      "learning_rate": 8.765426465737605e-06,
      "loss": 0.5279,
      "step": 15951
    },
    {
      "epoch": 0.7320453398191914,
      "grad_norm": 0.4907549023628235,
      "learning_rate": 8.765265146471663e-06,
      "loss": 0.4735,
      "step": 15952
    },
    {
      "epoch": 0.7320912303244459,
      "grad_norm": 0.48567628860473633,
      "learning_rate": 8.765103818151448e-06,
      "loss": 0.4721,
      "step": 15953
    },
    {
      "epoch": 0.7321371208297003,
      "grad_norm": 0.4452628493309021,
      "learning_rate": 8.764942480777346e-06,
      "loss": 0.3599,
      "step": 15954
    },
    {
      "epoch": 0.7321830113349548,
      "grad_norm": 0.47635406255722046,
      "learning_rate": 8.764781134349745e-06,
      "loss": 0.4738,
      "step": 15955
    },
    {
      "epoch": 0.7322289018402093,
      "grad_norm": 0.514101505279541,
      "learning_rate": 8.764619778869034e-06,
      "loss": 0.4896,
      "step": 15956
    },
    {
      "epoch": 0.7322747923454638,
      "grad_norm": 0.4813731014728546,
      "learning_rate": 8.7644584143356e-06,
      "loss": 0.4239,
      "step": 15957
    },
    {
      "epoch": 0.7323206828507182,
      "grad_norm": 0.5118345022201538,
      "learning_rate": 8.764297040749832e-06,
      "loss": 0.5033,
      "step": 15958
    },
    {
      "epoch": 0.7323665733559727,
      "grad_norm": 0.525173544883728,
      "learning_rate": 8.764135658112118e-06,
      "loss": 0.4572,
      "step": 15959
    },
    {
      "epoch": 0.7324124638612272,
      "grad_norm": 0.48939767479896545,
      "learning_rate": 8.763974266422846e-06,
      "loss": 0.4634,
      "step": 15960
    },
    {
      "epoch": 0.7324583543664815,
      "grad_norm": 0.44155463576316833,
      "learning_rate": 8.763812865682402e-06,
      "loss": 0.3747,
      "step": 15961
    },
    {
      "epoch": 0.732504244871736,
      "grad_norm": 0.4541780352592468,
      "learning_rate": 8.763651455891177e-06,
      "loss": 0.3457,
      "step": 15962
    },
    {
      "epoch": 0.7325501353769905,
      "grad_norm": 0.43352413177490234,
      "learning_rate": 8.763490037049561e-06,
      "loss": 0.3809,
      "step": 15963
    },
    {
      "epoch": 0.7325960258822449,
      "grad_norm": 0.40558886528015137,
      "learning_rate": 8.763328609157935e-06,
      "loss": 0.3297,
      "step": 15964
    },
    {
      "epoch": 0.7326419163874994,
      "grad_norm": 0.5159944891929626,
      "learning_rate": 8.763167172216693e-06,
      "loss": 0.4175,
      "step": 15965
    },
    {
      "epoch": 0.7326878068927539,
      "grad_norm": 0.4289466142654419,
      "learning_rate": 8.763005726226221e-06,
      "loss": 0.3589,
      "step": 15966
    },
    {
      "epoch": 0.7327336973980083,
      "grad_norm": 0.45825204253196716,
      "learning_rate": 8.76284427118691e-06,
      "loss": 0.3848,
      "step": 15967
    },
    {
      "epoch": 0.7327795879032628,
      "grad_norm": 0.45539066195487976,
      "learning_rate": 8.762682807099143e-06,
      "loss": 0.4597,
      "step": 15968
    },
    {
      "epoch": 0.7328254784085173,
      "grad_norm": 0.47126147150993347,
      "learning_rate": 8.762521333963314e-06,
      "loss": 0.4245,
      "step": 15969
    },
    {
      "epoch": 0.7328713689137717,
      "grad_norm": 0.4594539999961853,
      "learning_rate": 8.762359851779807e-06,
      "loss": 0.406,
      "step": 15970
    },
    {
      "epoch": 0.7329172594190262,
      "grad_norm": 0.507514238357544,
      "learning_rate": 8.762198360549013e-06,
      "loss": 0.4746,
      "step": 15971
    },
    {
      "epoch": 0.7329631499242807,
      "grad_norm": 0.48064494132995605,
      "learning_rate": 8.76203686027132e-06,
      "loss": 0.4464,
      "step": 15972
    },
    {
      "epoch": 0.7330090404295351,
      "grad_norm": 0.4547726809978485,
      "learning_rate": 8.761875350947114e-06,
      "loss": 0.3689,
      "step": 15973
    },
    {
      "epoch": 0.7330549309347896,
      "grad_norm": 0.5376349091529846,
      "learning_rate": 8.761713832576785e-06,
      "loss": 0.3879,
      "step": 15974
    },
    {
      "epoch": 0.7331008214400441,
      "grad_norm": 0.45568010210990906,
      "learning_rate": 8.761552305160724e-06,
      "loss": 0.4075,
      "step": 15975
    },
    {
      "epoch": 0.7331467119452986,
      "grad_norm": 0.7805156111717224,
      "learning_rate": 8.761390768699315e-06,
      "loss": 0.4608,
      "step": 15976
    },
    {
      "epoch": 0.733192602450553,
      "grad_norm": 0.45124515891075134,
      "learning_rate": 8.761229223192949e-06,
      "loss": 0.4038,
      "step": 15977
    },
    {
      "epoch": 0.7332384929558075,
      "grad_norm": 0.520440936088562,
      "learning_rate": 8.761067668642013e-06,
      "loss": 0.495,
      "step": 15978
    },
    {
      "epoch": 0.733284383461062,
      "grad_norm": 0.4846421778202057,
      "learning_rate": 8.760906105046896e-06,
      "loss": 0.4409,
      "step": 15979
    },
    {
      "epoch": 0.7333302739663163,
      "grad_norm": 0.448649525642395,
      "learning_rate": 8.760744532407988e-06,
      "loss": 0.4349,
      "step": 15980
    },
    {
      "epoch": 0.7333761644715708,
      "grad_norm": 0.5006638765335083,
      "learning_rate": 8.760582950725675e-06,
      "loss": 0.4606,
      "step": 15981
    },
    {
      "epoch": 0.7334220549768253,
      "grad_norm": 0.47614434361457825,
      "learning_rate": 8.760421360000349e-06,
      "loss": 0.4111,
      "step": 15982
    },
    {
      "epoch": 0.7334679454820797,
      "grad_norm": 0.46055370569229126,
      "learning_rate": 8.760259760232395e-06,
      "loss": 0.4169,
      "step": 15983
    },
    {
      "epoch": 0.7335138359873342,
      "grad_norm": 0.45865610241889954,
      "learning_rate": 8.760098151422202e-06,
      "loss": 0.3945,
      "step": 15984
    },
    {
      "epoch": 0.7335597264925887,
      "grad_norm": 0.4412473142147064,
      "learning_rate": 8.75993653357016e-06,
      "loss": 0.3454,
      "step": 15985
    },
    {
      "epoch": 0.7336056169978431,
      "grad_norm": 0.4638570249080658,
      "learning_rate": 8.75977490667666e-06,
      "loss": 0.4087,
      "step": 15986
    },
    {
      "epoch": 0.7336515075030976,
      "grad_norm": 0.42857828736305237,
      "learning_rate": 8.759613270742085e-06,
      "loss": 0.3946,
      "step": 15987
    },
    {
      "epoch": 0.7336973980083521,
      "grad_norm": 0.49333256483078003,
      "learning_rate": 8.759451625766827e-06,
      "loss": 0.5215,
      "step": 15988
    },
    {
      "epoch": 0.7337432885136065,
      "grad_norm": 0.4647095501422882,
      "learning_rate": 8.759289971751275e-06,
      "loss": 0.3684,
      "step": 15989
    },
    {
      "epoch": 0.733789179018861,
      "grad_norm": 0.44152405858039856,
      "learning_rate": 8.759128308695818e-06,
      "loss": 0.3806,
      "step": 15990
    },
    {
      "epoch": 0.7338350695241155,
      "grad_norm": 0.45702317357063293,
      "learning_rate": 8.758966636600843e-06,
      "loss": 0.4124,
      "step": 15991
    },
    {
      "epoch": 0.73388096002937,
      "grad_norm": 0.4066959619522095,
      "learning_rate": 8.75880495546674e-06,
      "loss": 0.3076,
      "step": 15992
    },
    {
      "epoch": 0.7339268505346244,
      "grad_norm": 0.6690467000007629,
      "learning_rate": 8.758643265293896e-06,
      "loss": 0.3237,
      "step": 15993
    },
    {
      "epoch": 0.7339727410398789,
      "grad_norm": 0.4665399491786957,
      "learning_rate": 8.758481566082702e-06,
      "loss": 0.4458,
      "step": 15994
    },
    {
      "epoch": 0.7340186315451334,
      "grad_norm": 0.4557625949382782,
      "learning_rate": 8.758319857833546e-06,
      "loss": 0.3295,
      "step": 15995
    },
    {
      "epoch": 0.7340645220503877,
      "grad_norm": 0.44812726974487305,
      "learning_rate": 8.758158140546817e-06,
      "loss": 0.3598,
      "step": 15996
    },
    {
      "epoch": 0.7341104125556422,
      "grad_norm": 0.3999248743057251,
      "learning_rate": 8.757996414222905e-06,
      "loss": 0.3116,
      "step": 15997
    },
    {
      "epoch": 0.7341563030608967,
      "grad_norm": 0.45262986421585083,
      "learning_rate": 8.757834678862197e-06,
      "loss": 0.3819,
      "step": 15998
    },
    {
      "epoch": 0.7342021935661511,
      "grad_norm": 0.4417167603969574,
      "learning_rate": 8.757672934465082e-06,
      "loss": 0.3857,
      "step": 15999
    },
    {
      "epoch": 0.7342480840714056,
      "grad_norm": 0.40012234449386597,
      "learning_rate": 8.75751118103195e-06,
      "loss": 0.291,
      "step": 16000
    },
    {
      "epoch": 0.7342939745766601,
      "grad_norm": 0.4767579436302185,
      "learning_rate": 8.757349418563188e-06,
      "loss": 0.394,
      "step": 16001
    },
    {
      "epoch": 0.7343398650819145,
      "grad_norm": 0.48150455951690674,
      "learning_rate": 8.75718764705919e-06,
      "loss": 0.4305,
      "step": 16002
    },
    {
      "epoch": 0.734385755587169,
      "grad_norm": 0.5056874752044678,
      "learning_rate": 8.757025866520337e-06,
      "loss": 0.4422,
      "step": 16003
    },
    {
      "epoch": 0.7344316460924235,
      "grad_norm": 0.47848936915397644,
      "learning_rate": 8.756864076947025e-06,
      "loss": 0.4248,
      "step": 16004
    },
    {
      "epoch": 0.7344775365976779,
      "grad_norm": 0.44424909353256226,
      "learning_rate": 8.75670227833964e-06,
      "loss": 0.4041,
      "step": 16005
    },
    {
      "epoch": 0.7345234271029324,
      "grad_norm": 0.4556829035282135,
      "learning_rate": 8.756540470698573e-06,
      "loss": 0.3615,
      "step": 16006
    },
    {
      "epoch": 0.7345693176081869,
      "grad_norm": 0.4247104525566101,
      "learning_rate": 8.75637865402421e-06,
      "loss": 0.3395,
      "step": 16007
    },
    {
      "epoch": 0.7346152081134413,
      "grad_norm": 0.496927410364151,
      "learning_rate": 8.756216828316944e-06,
      "loss": 0.4665,
      "step": 16008
    },
    {
      "epoch": 0.7346610986186958,
      "grad_norm": 0.4343481659889221,
      "learning_rate": 8.75605499357716e-06,
      "loss": 0.3832,
      "step": 16009
    },
    {
      "epoch": 0.7347069891239503,
      "grad_norm": 0.46922731399536133,
      "learning_rate": 8.75589314980525e-06,
      "loss": 0.3707,
      "step": 16010
    },
    {
      "epoch": 0.7347528796292048,
      "grad_norm": 0.41550156474113464,
      "learning_rate": 8.755731297001601e-06,
      "loss": 0.3228,
      "step": 16011
    },
    {
      "epoch": 0.7347987701344592,
      "grad_norm": 0.4693114459514618,
      "learning_rate": 8.755569435166607e-06,
      "loss": 0.4068,
      "step": 16012
    },
    {
      "epoch": 0.7348446606397137,
      "grad_norm": 0.4681084156036377,
      "learning_rate": 8.755407564300651e-06,
      "loss": 0.4597,
      "step": 16013
    },
    {
      "epoch": 0.7348905511449682,
      "grad_norm": 0.48601141571998596,
      "learning_rate": 8.755245684404126e-06,
      "loss": 0.4388,
      "step": 16014
    },
    {
      "epoch": 0.7349364416502225,
      "grad_norm": 0.6528716683387756,
      "learning_rate": 8.75508379547742e-06,
      "loss": 0.4107,
      "step": 16015
    },
    {
      "epoch": 0.734982332155477,
      "grad_norm": 0.46836912631988525,
      "learning_rate": 8.754921897520923e-06,
      "loss": 0.4546,
      "step": 16016
    },
    {
      "epoch": 0.7350282226607315,
      "grad_norm": 0.5222553610801697,
      "learning_rate": 8.754759990535022e-06,
      "loss": 0.4365,
      "step": 16017
    },
    {
      "epoch": 0.7350741131659859,
      "grad_norm": 0.46116402745246887,
      "learning_rate": 8.75459807452011e-06,
      "loss": 0.4407,
      "step": 16018
    },
    {
      "epoch": 0.7351200036712404,
      "grad_norm": 0.6612066030502319,
      "learning_rate": 8.754436149476576e-06,
      "loss": 0.4789,
      "step": 16019
    },
    {
      "epoch": 0.7351658941764949,
      "grad_norm": 0.4208511412143707,
      "learning_rate": 8.754274215404806e-06,
      "loss": 0.3641,
      "step": 16020
    },
    {
      "epoch": 0.7352117846817493,
      "grad_norm": 0.4309150278568268,
      "learning_rate": 8.754112272305191e-06,
      "loss": 0.3518,
      "step": 16021
    },
    {
      "epoch": 0.7352576751870038,
      "grad_norm": 0.46029770374298096,
      "learning_rate": 8.753950320178121e-06,
      "loss": 0.3844,
      "step": 16022
    },
    {
      "epoch": 0.7353035656922583,
      "grad_norm": 0.45581042766571045,
      "learning_rate": 8.753788359023988e-06,
      "loss": 0.4,
      "step": 16023
    },
    {
      "epoch": 0.7353494561975127,
      "grad_norm": 0.4901447296142578,
      "learning_rate": 8.753626388843175e-06,
      "loss": 0.4308,
      "step": 16024
    },
    {
      "epoch": 0.7353953467027672,
      "grad_norm": 0.45787447690963745,
      "learning_rate": 8.753464409636077e-06,
      "loss": 0.3838,
      "step": 16025
    },
    {
      "epoch": 0.7354412372080217,
      "grad_norm": 0.46150150895118713,
      "learning_rate": 8.753302421403081e-06,
      "loss": 0.4225,
      "step": 16026
    },
    {
      "epoch": 0.7354871277132762,
      "grad_norm": 0.4517897367477417,
      "learning_rate": 8.75314042414458e-06,
      "loss": 0.4191,
      "step": 16027
    },
    {
      "epoch": 0.7355330182185306,
      "grad_norm": 0.507862389087677,
      "learning_rate": 8.752978417860957e-06,
      "loss": 0.4706,
      "step": 16028
    },
    {
      "epoch": 0.7355789087237851,
      "grad_norm": 0.4780707061290741,
      "learning_rate": 8.752816402552607e-06,
      "loss": 0.4458,
      "step": 16029
    },
    {
      "epoch": 0.7356247992290396,
      "grad_norm": 0.46732306480407715,
      "learning_rate": 8.752654378219917e-06,
      "loss": 0.4289,
      "step": 16030
    },
    {
      "epoch": 0.735670689734294,
      "grad_norm": 0.4724268615245819,
      "learning_rate": 8.752492344863279e-06,
      "loss": 0.4087,
      "step": 16031
    },
    {
      "epoch": 0.7357165802395484,
      "grad_norm": 0.4329822063446045,
      "learning_rate": 8.752330302483082e-06,
      "loss": 0.3865,
      "step": 16032
    },
    {
      "epoch": 0.7357624707448029,
      "grad_norm": 0.4339953362941742,
      "learning_rate": 8.752168251079713e-06,
      "loss": 0.3849,
      "step": 16033
    },
    {
      "epoch": 0.7358083612500573,
      "grad_norm": 0.47654518485069275,
      "learning_rate": 8.752006190653564e-06,
      "loss": 0.4491,
      "step": 16034
    },
    {
      "epoch": 0.7358542517553118,
      "grad_norm": 0.4512265622615814,
      "learning_rate": 8.751844121205024e-06,
      "loss": 0.3672,
      "step": 16035
    },
    {
      "epoch": 0.7359001422605663,
      "grad_norm": 0.42702749371528625,
      "learning_rate": 8.751682042734485e-06,
      "loss": 0.3908,
      "step": 16036
    },
    {
      "epoch": 0.7359460327658207,
      "grad_norm": 0.4433613121509552,
      "learning_rate": 8.751519955242334e-06,
      "loss": 0.3575,
      "step": 16037
    },
    {
      "epoch": 0.7359919232710752,
      "grad_norm": 0.4446146786212921,
      "learning_rate": 8.751357858728959e-06,
      "loss": 0.3793,
      "step": 16038
    },
    {
      "epoch": 0.7360378137763297,
      "grad_norm": 0.5112154483795166,
      "learning_rate": 8.751195753194754e-06,
      "loss": 0.3998,
      "step": 16039
    },
    {
      "epoch": 0.7360837042815841,
      "grad_norm": 0.46949493885040283,
      "learning_rate": 8.751033638640108e-06,
      "loss": 0.4367,
      "step": 16040
    },
    {
      "epoch": 0.7361295947868386,
      "grad_norm": 0.4684097468852997,
      "learning_rate": 8.75087151506541e-06,
      "loss": 0.4225,
      "step": 16041
    },
    {
      "epoch": 0.7361754852920931,
      "grad_norm": 0.44512760639190674,
      "learning_rate": 8.750709382471048e-06,
      "loss": 0.3536,
      "step": 16042
    },
    {
      "epoch": 0.7362213757973475,
      "grad_norm": 0.46184590458869934,
      "learning_rate": 8.750547240857416e-06,
      "loss": 0.4236,
      "step": 16043
    },
    {
      "epoch": 0.736267266302602,
      "grad_norm": 0.4564439058303833,
      "learning_rate": 8.7503850902249e-06,
      "loss": 0.3485,
      "step": 16044
    },
    {
      "epoch": 0.7363131568078565,
      "grad_norm": 0.4779980480670929,
      "learning_rate": 8.75022293057389e-06,
      "loss": 0.4504,
      "step": 16045
    },
    {
      "epoch": 0.736359047313111,
      "grad_norm": 0.48121780157089233,
      "learning_rate": 8.75006076190478e-06,
      "loss": 0.4323,
      "step": 16046
    },
    {
      "epoch": 0.7364049378183654,
      "grad_norm": 0.4444717466831207,
      "learning_rate": 8.749898584217957e-06,
      "loss": 0.3553,
      "step": 16047
    },
    {
      "epoch": 0.7364508283236199,
      "grad_norm": 0.46610522270202637,
      "learning_rate": 8.749736397513811e-06,
      "loss": 0.4079,
      "step": 16048
    },
    {
      "epoch": 0.7364967188288744,
      "grad_norm": 0.4422728419303894,
      "learning_rate": 8.749574201792734e-06,
      "loss": 0.3882,
      "step": 16049
    },
    {
      "epoch": 0.7365426093341287,
      "grad_norm": 0.43803870677948,
      "learning_rate": 8.749411997055111e-06,
      "loss": 0.3349,
      "step": 16050
    },
    {
      "epoch": 0.7365884998393832,
      "grad_norm": 0.462265282869339,
      "learning_rate": 8.74924978330134e-06,
      "loss": 0.3599,
      "step": 16051
    },
    {
      "epoch": 0.7366343903446377,
      "grad_norm": 0.5062516927719116,
      "learning_rate": 8.749087560531803e-06,
      "loss": 0.4379,
      "step": 16052
    },
    {
      "epoch": 0.7366802808498921,
      "grad_norm": 0.44809144735336304,
      "learning_rate": 8.748925328746894e-06,
      "loss": 0.3878,
      "step": 16053
    },
    {
      "epoch": 0.7367261713551466,
      "grad_norm": 0.4590931832790375,
      "learning_rate": 8.748763087947003e-06,
      "loss": 0.4043,
      "step": 16054
    },
    {
      "epoch": 0.7367720618604011,
      "grad_norm": 0.42366498708724976,
      "learning_rate": 8.748600838132521e-06,
      "loss": 0.3375,
      "step": 16055
    },
    {
      "epoch": 0.7368179523656555,
      "grad_norm": 0.41984736919403076,
      "learning_rate": 8.748438579303837e-06,
      "loss": 0.3973,
      "step": 16056
    },
    {
      "epoch": 0.73686384287091,
      "grad_norm": 0.426851361989975,
      "learning_rate": 8.748276311461341e-06,
      "loss": 0.3411,
      "step": 16057
    },
    {
      "epoch": 0.7369097333761645,
      "grad_norm": 0.44430434703826904,
      "learning_rate": 8.748114034605423e-06,
      "loss": 0.4268,
      "step": 16058
    },
    {
      "epoch": 0.7369556238814189,
      "grad_norm": 0.4714086651802063,
      "learning_rate": 8.747951748736474e-06,
      "loss": 0.4242,
      "step": 16059
    },
    {
      "epoch": 0.7370015143866734,
      "grad_norm": 0.4476342797279358,
      "learning_rate": 8.747789453854884e-06,
      "loss": 0.3771,
      "step": 16060
    },
    {
      "epoch": 0.7370474048919279,
      "grad_norm": 0.4290663003921509,
      "learning_rate": 8.747627149961042e-06,
      "loss": 0.3553,
      "step": 16061
    },
    {
      "epoch": 0.7370932953971823,
      "grad_norm": 0.40016189217567444,
      "learning_rate": 8.747464837055342e-06,
      "loss": 0.3311,
      "step": 16062
    },
    {
      "epoch": 0.7371391859024368,
      "grad_norm": 0.45907601714134216,
      "learning_rate": 8.74730251513817e-06,
      "loss": 0.4162,
      "step": 16063
    },
    {
      "epoch": 0.7371850764076913,
      "grad_norm": 0.47077348828315735,
      "learning_rate": 8.747140184209917e-06,
      "loss": 0.4257,
      "step": 16064
    },
    {
      "epoch": 0.7372309669129458,
      "grad_norm": 0.44760432839393616,
      "learning_rate": 8.746977844270976e-06,
      "loss": 0.3215,
      "step": 16065
    },
    {
      "epoch": 0.7372768574182001,
      "grad_norm": 0.4860057830810547,
      "learning_rate": 8.746815495321737e-06,
      "loss": 0.4057,
      "step": 16066
    },
    {
      "epoch": 0.7373227479234546,
      "grad_norm": 0.4765504002571106,
      "learning_rate": 8.746653137362588e-06,
      "loss": 0.5047,
      "step": 16067
    },
    {
      "epoch": 0.7373686384287091,
      "grad_norm": 0.5118085741996765,
      "learning_rate": 8.74649077039392e-06,
      "loss": 0.4715,
      "step": 16068
    },
    {
      "epoch": 0.7374145289339635,
      "grad_norm": 0.4291047155857086,
      "learning_rate": 8.746328394416124e-06,
      "loss": 0.3812,
      "step": 16069
    },
    {
      "epoch": 0.737460419439218,
      "grad_norm": 0.4587209224700928,
      "learning_rate": 8.746166009429592e-06,
      "loss": 0.4194,
      "step": 16070
    },
    {
      "epoch": 0.7375063099444725,
      "grad_norm": 0.4775572121143341,
      "learning_rate": 8.746003615434712e-06,
      "loss": 0.4702,
      "step": 16071
    },
    {
      "epoch": 0.7375522004497269,
      "grad_norm": 0.4398815333843231,
      "learning_rate": 8.745841212431878e-06,
      "loss": 0.3559,
      "step": 16072
    },
    {
      "epoch": 0.7375980909549814,
      "grad_norm": 0.466252863407135,
      "learning_rate": 8.745678800421476e-06,
      "loss": 0.4202,
      "step": 16073
    },
    {
      "epoch": 0.7376439814602359,
      "grad_norm": 0.4556097388267517,
      "learning_rate": 8.745516379403897e-06,
      "loss": 0.3742,
      "step": 16074
    },
    {
      "epoch": 0.7376898719654903,
      "grad_norm": 0.5115895867347717,
      "learning_rate": 8.745353949379536e-06,
      "loss": 0.498,
      "step": 16075
    },
    {
      "epoch": 0.7377357624707448,
      "grad_norm": 0.4719484746456146,
      "learning_rate": 8.74519151034878e-06,
      "loss": 0.3418,
      "step": 16076
    },
    {
      "epoch": 0.7377816529759993,
      "grad_norm": 0.42186978459358215,
      "learning_rate": 8.74502906231202e-06,
      "loss": 0.3525,
      "step": 16077
    },
    {
      "epoch": 0.7378275434812537,
      "grad_norm": 0.5289642214775085,
      "learning_rate": 8.744866605269648e-06,
      "loss": 0.4206,
      "step": 16078
    },
    {
      "epoch": 0.7378734339865082,
      "grad_norm": 0.4700157046318054,
      "learning_rate": 8.744704139222054e-06,
      "loss": 0.4882,
      "step": 16079
    },
    {
      "epoch": 0.7379193244917627,
      "grad_norm": 0.5120685696601868,
      "learning_rate": 8.744541664169626e-06,
      "loss": 0.3681,
      "step": 16080
    },
    {
      "epoch": 0.7379652149970172,
      "grad_norm": 0.4274618923664093,
      "learning_rate": 8.744379180112759e-06,
      "loss": 0.3345,
      "step": 16081
    },
    {
      "epoch": 0.7380111055022716,
      "grad_norm": 0.46479174494743347,
      "learning_rate": 8.744216687051841e-06,
      "loss": 0.4017,
      "step": 16082
    },
    {
      "epoch": 0.7380569960075261,
      "grad_norm": 0.4495057761669159,
      "learning_rate": 8.744054184987265e-06,
      "loss": 0.3548,
      "step": 16083
    },
    {
      "epoch": 0.7381028865127806,
      "grad_norm": 0.4770502746105194,
      "learning_rate": 8.74389167391942e-06,
      "loss": 0.3745,
      "step": 16084
    },
    {
      "epoch": 0.7381487770180349,
      "grad_norm": 0.4586268365383148,
      "learning_rate": 8.743729153848697e-06,
      "loss": 0.3839,
      "step": 16085
    },
    {
      "epoch": 0.7381946675232894,
      "grad_norm": 0.45045191049575806,
      "learning_rate": 8.743566624775485e-06,
      "loss": 0.3582,
      "step": 16086
    },
    {
      "epoch": 0.7382405580285439,
      "grad_norm": 0.4877607226371765,
      "learning_rate": 8.743404086700178e-06,
      "loss": 0.4244,
      "step": 16087
    },
    {
      "epoch": 0.7382864485337983,
      "grad_norm": 0.44610029458999634,
      "learning_rate": 8.743241539623166e-06,
      "loss": 0.3395,
      "step": 16088
    },
    {
      "epoch": 0.7383323390390528,
      "grad_norm": 0.455594003200531,
      "learning_rate": 8.743078983544839e-06,
      "loss": 0.3781,
      "step": 16089
    },
    {
      "epoch": 0.7383782295443073,
      "grad_norm": 0.483226478099823,
      "learning_rate": 8.74291641846559e-06,
      "loss": 0.4468,
      "step": 16090
    },
    {
      "epoch": 0.7384241200495617,
      "grad_norm": 0.48671430349349976,
      "learning_rate": 8.742753844385807e-06,
      "loss": 0.3759,
      "step": 16091
    },
    {
      "epoch": 0.7384700105548162,
      "grad_norm": 0.4697237014770508,
      "learning_rate": 8.742591261305882e-06,
      "loss": 0.3863,
      "step": 16092
    },
    {
      "epoch": 0.7385159010600707,
      "grad_norm": 0.43328678607940674,
      "learning_rate": 8.742428669226206e-06,
      "loss": 0.3554,
      "step": 16093
    },
    {
      "epoch": 0.7385617915653251,
      "grad_norm": 0.41437193751335144,
      "learning_rate": 8.742266068147171e-06,
      "loss": 0.3302,
      "step": 16094
    },
    {
      "epoch": 0.7386076820705796,
      "grad_norm": 0.5397595763206482,
      "learning_rate": 8.742103458069167e-06,
      "loss": 0.5184,
      "step": 16095
    },
    {
      "epoch": 0.7386535725758341,
      "grad_norm": 0.44679442048072815,
      "learning_rate": 8.741940838992585e-06,
      "loss": 0.3775,
      "step": 16096
    },
    {
      "epoch": 0.7386994630810885,
      "grad_norm": 0.44663742184638977,
      "learning_rate": 8.741778210917817e-06,
      "loss": 0.3961,
      "step": 16097
    },
    {
      "epoch": 0.738745353586343,
      "grad_norm": 0.48417043685913086,
      "learning_rate": 8.741615573845252e-06,
      "loss": 0.455,
      "step": 16098
    },
    {
      "epoch": 0.7387912440915975,
      "grad_norm": 0.43472760915756226,
      "learning_rate": 8.741452927775283e-06,
      "loss": 0.361,
      "step": 16099
    },
    {
      "epoch": 0.738837134596852,
      "grad_norm": 0.4686301052570343,
      "learning_rate": 8.741290272708302e-06,
      "loss": 0.3647,
      "step": 16100
    },
    {
      "epoch": 0.7388830251021063,
      "grad_norm": 0.44507551193237305,
      "learning_rate": 8.741127608644697e-06,
      "loss": 0.3847,
      "step": 16101
    },
    {
      "epoch": 0.7389289156073608,
      "grad_norm": 0.46205881237983704,
      "learning_rate": 8.740964935584861e-06,
      "loss": 0.3729,
      "step": 16102
    },
    {
      "epoch": 0.7389748061126153,
      "grad_norm": 0.4986153542995453,
      "learning_rate": 8.740802253529185e-06,
      "loss": 0.4484,
      "step": 16103
    },
    {
      "epoch": 0.7390206966178697,
      "grad_norm": 0.4620817303657532,
      "learning_rate": 8.740639562478062e-06,
      "loss": 0.3903,
      "step": 16104
    },
    {
      "epoch": 0.7390665871231242,
      "grad_norm": 0.4672529995441437,
      "learning_rate": 8.740476862431879e-06,
      "loss": 0.4594,
      "step": 16105
    },
    {
      "epoch": 0.7391124776283787,
      "grad_norm": 0.46298748254776,
      "learning_rate": 8.740314153391031e-06,
      "loss": 0.3803,
      "step": 16106
    },
    {
      "epoch": 0.7391583681336331,
      "grad_norm": 0.45846372842788696,
      "learning_rate": 8.740151435355906e-06,
      "loss": 0.3394,
      "step": 16107
    },
    {
      "epoch": 0.7392042586388876,
      "grad_norm": 0.4562525451183319,
      "learning_rate": 8.739988708326899e-06,
      "loss": 0.3802,
      "step": 16108
    },
    {
      "epoch": 0.7392501491441421,
      "grad_norm": 0.4561963379383087,
      "learning_rate": 8.739825972304401e-06,
      "loss": 0.4144,
      "step": 16109
    },
    {
      "epoch": 0.7392960396493965,
      "grad_norm": 0.4169153571128845,
      "learning_rate": 8.7396632272888e-06,
      "loss": 0.3103,
      "step": 16110
    },
    {
      "epoch": 0.739341930154651,
      "grad_norm": 0.48917412757873535,
      "learning_rate": 8.73950047328049e-06,
      "loss": 0.4745,
      "step": 16111
    },
    {
      "epoch": 0.7393878206599055,
      "grad_norm": 0.5736055374145508,
      "learning_rate": 8.73933771027986e-06,
      "loss": 0.6125,
      "step": 16112
    },
    {
      "epoch": 0.7394337111651599,
      "grad_norm": 0.4355540871620178,
      "learning_rate": 8.739174938287303e-06,
      "loss": 0.3656,
      "step": 16113
    },
    {
      "epoch": 0.7394796016704144,
      "grad_norm": 0.4507342278957367,
      "learning_rate": 8.739012157303212e-06,
      "loss": 0.398,
      "step": 16114
    },
    {
      "epoch": 0.7395254921756689,
      "grad_norm": 0.43851613998413086,
      "learning_rate": 8.738849367327976e-06,
      "loss": 0.3798,
      "step": 16115
    },
    {
      "epoch": 0.7395713826809233,
      "grad_norm": 0.4429139792919159,
      "learning_rate": 8.738686568361987e-06,
      "loss": 0.3877,
      "step": 16116
    },
    {
      "epoch": 0.7396172731861778,
      "grad_norm": 0.4806617796421051,
      "learning_rate": 8.738523760405637e-06,
      "loss": 0.4209,
      "step": 16117
    },
    {
      "epoch": 0.7396631636914323,
      "grad_norm": 0.45465174317359924,
      "learning_rate": 8.738360943459318e-06,
      "loss": 0.3526,
      "step": 16118
    },
    {
      "epoch": 0.7397090541966868,
      "grad_norm": 0.4482746124267578,
      "learning_rate": 8.73819811752342e-06,
      "loss": 0.3685,
      "step": 16119
    },
    {
      "epoch": 0.7397549447019411,
      "grad_norm": 0.45512130856513977,
      "learning_rate": 8.738035282598336e-06,
      "loss": 0.3541,
      "step": 16120
    },
    {
      "epoch": 0.7398008352071956,
      "grad_norm": 0.47207358479499817,
      "learning_rate": 8.737872438684457e-06,
      "loss": 0.4498,
      "step": 16121
    },
    {
      "epoch": 0.7398467257124501,
      "grad_norm": 0.46567198634147644,
      "learning_rate": 8.737709585782173e-06,
      "loss": 0.3997,
      "step": 16122
    },
    {
      "epoch": 0.7398926162177045,
      "grad_norm": 0.42531922459602356,
      "learning_rate": 8.737546723891878e-06,
      "loss": 0.3223,
      "step": 16123
    },
    {
      "epoch": 0.739938506722959,
      "grad_norm": 0.45814183354377747,
      "learning_rate": 8.737383853013963e-06,
      "loss": 0.4003,
      "step": 16124
    },
    {
      "epoch": 0.7399843972282135,
      "grad_norm": 0.42679598927497864,
      "learning_rate": 8.73722097314882e-06,
      "loss": 0.3884,
      "step": 16125
    },
    {
      "epoch": 0.7400302877334679,
      "grad_norm": 0.4293581247329712,
      "learning_rate": 8.737058084296839e-06,
      "loss": 0.3569,
      "step": 16126
    },
    {
      "epoch": 0.7400761782387224,
      "grad_norm": 0.4700793921947479,
      "learning_rate": 8.736895186458414e-06,
      "loss": 0.4256,
      "step": 16127
    },
    {
      "epoch": 0.7401220687439769,
      "grad_norm": 0.48024290800094604,
      "learning_rate": 8.736732279633933e-06,
      "loss": 0.3761,
      "step": 16128
    },
    {
      "epoch": 0.7401679592492313,
      "grad_norm": 0.4698987901210785,
      "learning_rate": 8.736569363823792e-06,
      "loss": 0.3934,
      "step": 16129
    },
    {
      "epoch": 0.7402138497544858,
      "grad_norm": 0.5660913586616516,
      "learning_rate": 8.73640643902838e-06,
      "loss": 0.4613,
      "step": 16130
    },
    {
      "epoch": 0.7402597402597403,
      "grad_norm": 0.6585381031036377,
      "learning_rate": 8.736243505248091e-06,
      "loss": 0.4257,
      "step": 16131
    },
    {
      "epoch": 0.7403056307649947,
      "grad_norm": 0.49519163370132446,
      "learning_rate": 8.736080562483316e-06,
      "loss": 0.5133,
      "step": 16132
    },
    {
      "epoch": 0.7403515212702492,
      "grad_norm": 0.46089187264442444,
      "learning_rate": 8.735917610734445e-06,
      "loss": 0.4457,
      "step": 16133
    },
    {
      "epoch": 0.7403974117755037,
      "grad_norm": 0.4055814743041992,
      "learning_rate": 8.73575465000187e-06,
      "loss": 0.3195,
      "step": 16134
    },
    {
      "epoch": 0.7404433022807582,
      "grad_norm": 0.4679926335811615,
      "learning_rate": 8.735591680285986e-06,
      "loss": 0.3813,
      "step": 16135
    },
    {
      "epoch": 0.7404891927860126,
      "grad_norm": 0.45333167910575867,
      "learning_rate": 8.735428701587183e-06,
      "loss": 0.3841,
      "step": 16136
    },
    {
      "epoch": 0.740535083291267,
      "grad_norm": 0.4486124813556671,
      "learning_rate": 8.735265713905853e-06,
      "loss": 0.4347,
      "step": 16137
    },
    {
      "epoch": 0.7405809737965215,
      "grad_norm": 0.47174447774887085,
      "learning_rate": 8.735102717242388e-06,
      "loss": 0.4276,
      "step": 16138
    },
    {
      "epoch": 0.7406268643017759,
      "grad_norm": 0.41880178451538086,
      "learning_rate": 8.734939711597178e-06,
      "loss": 0.3558,
      "step": 16139
    },
    {
      "epoch": 0.7406727548070304,
      "grad_norm": 0.4770233631134033,
      "learning_rate": 8.73477669697062e-06,
      "loss": 0.4692,
      "step": 16140
    },
    {
      "epoch": 0.7407186453122849,
      "grad_norm": 0.4251320958137512,
      "learning_rate": 8.734613673363099e-06,
      "loss": 0.3254,
      "step": 16141
    },
    {
      "epoch": 0.7407645358175393,
      "grad_norm": 0.5523183345794678,
      "learning_rate": 8.734450640775014e-06,
      "loss": 0.4772,
      "step": 16142
    },
    {
      "epoch": 0.7408104263227938,
      "grad_norm": 0.4386793076992035,
      "learning_rate": 8.734287599206751e-06,
      "loss": 0.3902,
      "step": 16143
    },
    {
      "epoch": 0.7408563168280483,
      "grad_norm": 0.46539005637168884,
      "learning_rate": 8.734124548658708e-06,
      "loss": 0.4615,
      "step": 16144
    },
    {
      "epoch": 0.7409022073333027,
      "grad_norm": 0.4543258547782898,
      "learning_rate": 8.73396148913127e-06,
      "loss": 0.3897,
      "step": 16145
    },
    {
      "epoch": 0.7409480978385572,
      "grad_norm": 0.5004029273986816,
      "learning_rate": 8.733798420624837e-06,
      "loss": 0.5044,
      "step": 16146
    },
    {
      "epoch": 0.7409939883438117,
      "grad_norm": 0.4698163568973541,
      "learning_rate": 8.733635343139796e-06,
      "loss": 0.4278,
      "step": 16147
    },
    {
      "epoch": 0.7410398788490661,
      "grad_norm": 0.4536164104938507,
      "learning_rate": 8.73347225667654e-06,
      "loss": 0.396,
      "step": 16148
    },
    {
      "epoch": 0.7410857693543206,
      "grad_norm": 0.4138358235359192,
      "learning_rate": 8.733309161235462e-06,
      "loss": 0.3147,
      "step": 16149
    },
    {
      "epoch": 0.7411316598595751,
      "grad_norm": 0.4568508565425873,
      "learning_rate": 8.733146056816953e-06,
      "loss": 0.3945,
      "step": 16150
    },
    {
      "epoch": 0.7411775503648295,
      "grad_norm": 0.44541874527931213,
      "learning_rate": 8.732982943421405e-06,
      "loss": 0.3942,
      "step": 16151
    },
    {
      "epoch": 0.741223440870084,
      "grad_norm": 0.4574776887893677,
      "learning_rate": 8.732819821049213e-06,
      "loss": 0.4135,
      "step": 16152
    },
    {
      "epoch": 0.7412693313753385,
      "grad_norm": 0.46046558022499084,
      "learning_rate": 8.732656689700767e-06,
      "loss": 0.4032,
      "step": 16153
    },
    {
      "epoch": 0.741315221880593,
      "grad_norm": 0.45076408982276917,
      "learning_rate": 8.73249354937646e-06,
      "loss": 0.3837,
      "step": 16154
    },
    {
      "epoch": 0.7413611123858473,
      "grad_norm": 0.4659837782382965,
      "learning_rate": 8.732330400076683e-06,
      "loss": 0.4083,
      "step": 16155
    },
    {
      "epoch": 0.7414070028911018,
      "grad_norm": 0.4350017011165619,
      "learning_rate": 8.73216724180183e-06,
      "loss": 0.3409,
      "step": 16156
    },
    {
      "epoch": 0.7414528933963563,
      "grad_norm": 0.469978928565979,
      "learning_rate": 8.732004074552294e-06,
      "loss": 0.358,
      "step": 16157
    },
    {
      "epoch": 0.7414987839016107,
      "grad_norm": 0.4696197211742401,
      "learning_rate": 8.731840898328462e-06,
      "loss": 0.429,
      "step": 16158
    },
    {
      "epoch": 0.7415446744068652,
      "grad_norm": 0.46041393280029297,
      "learning_rate": 8.731677713130735e-06,
      "loss": 0.4356,
      "step": 16159
    },
    {
      "epoch": 0.7415905649121197,
      "grad_norm": 0.4846954643726349,
      "learning_rate": 8.731514518959498e-06,
      "loss": 0.4385,
      "step": 16160
    },
    {
      "epoch": 0.7416364554173741,
      "grad_norm": 0.46422895789146423,
      "learning_rate": 8.731351315815149e-06,
      "loss": 0.4,
      "step": 16161
    },
    {
      "epoch": 0.7416823459226286,
      "grad_norm": 0.4660011827945709,
      "learning_rate": 8.731188103698076e-06,
      "loss": 0.4094,
      "step": 16162
    },
    {
      "epoch": 0.7417282364278831,
      "grad_norm": 0.4654594659805298,
      "learning_rate": 8.731024882608672e-06,
      "loss": 0.4115,
      "step": 16163
    },
    {
      "epoch": 0.7417741269331375,
      "grad_norm": 0.479792058467865,
      "learning_rate": 8.730861652547331e-06,
      "loss": 0.4584,
      "step": 16164
    },
    {
      "epoch": 0.741820017438392,
      "grad_norm": 0.47134044766426086,
      "learning_rate": 8.730698413514446e-06,
      "loss": 0.445,
      "step": 16165
    },
    {
      "epoch": 0.7418659079436465,
      "grad_norm": 0.44977229833602905,
      "learning_rate": 8.73053516551041e-06,
      "loss": 0.421,
      "step": 16166
    },
    {
      "epoch": 0.7419117984489009,
      "grad_norm": 0.47854113578796387,
      "learning_rate": 8.730371908535612e-06,
      "loss": 0.4239,
      "step": 16167
    },
    {
      "epoch": 0.7419576889541554,
      "grad_norm": 0.45471468567848206,
      "learning_rate": 8.73020864259045e-06,
      "loss": 0.3717,
      "step": 16168
    },
    {
      "epoch": 0.7420035794594099,
      "grad_norm": 0.4277383089065552,
      "learning_rate": 8.73004536767531e-06,
      "loss": 0.3291,
      "step": 16169
    },
    {
      "epoch": 0.7420494699646644,
      "grad_norm": 0.4229739010334015,
      "learning_rate": 8.729882083790591e-06,
      "loss": 0.3653,
      "step": 16170
    },
    {
      "epoch": 0.7420953604699188,
      "grad_norm": 0.4585472345352173,
      "learning_rate": 8.72971879093668e-06,
      "loss": 0.4243,
      "step": 16171
    },
    {
      "epoch": 0.7421412509751733,
      "grad_norm": 0.45656928420066833,
      "learning_rate": 8.729555489113975e-06,
      "loss": 0.3805,
      "step": 16172
    },
    {
      "epoch": 0.7421871414804277,
      "grad_norm": 0.4621952772140503,
      "learning_rate": 8.729392178322864e-06,
      "loss": 0.4046,
      "step": 16173
    },
    {
      "epoch": 0.7422330319856821,
      "grad_norm": 0.47696802020072937,
      "learning_rate": 8.729228858563745e-06,
      "loss": 0.4154,
      "step": 16174
    },
    {
      "epoch": 0.7422789224909366,
      "grad_norm": 0.4633540213108063,
      "learning_rate": 8.729065529837005e-06,
      "loss": 0.4577,
      "step": 16175
    },
    {
      "epoch": 0.7423248129961911,
      "grad_norm": 0.4880008101463318,
      "learning_rate": 8.72890219214304e-06,
      "loss": 0.4325,
      "step": 16176
    },
    {
      "epoch": 0.7423707035014455,
      "grad_norm": 0.4501653015613556,
      "learning_rate": 8.728738845482242e-06,
      "loss": 0.426,
      "step": 16177
    },
    {
      "epoch": 0.7424165940067,
      "grad_norm": 0.4904680550098419,
      "learning_rate": 8.728575489855003e-06,
      "loss": 0.4366,
      "step": 16178
    },
    {
      "epoch": 0.7424624845119545,
      "grad_norm": 0.464344322681427,
      "learning_rate": 8.728412125261718e-06,
      "loss": 0.4261,
      "step": 16179
    },
    {
      "epoch": 0.7425083750172089,
      "grad_norm": 0.4450833201408386,
      "learning_rate": 8.728248751702777e-06,
      "loss": 0.353,
      "step": 16180
    },
    {
      "epoch": 0.7425542655224634,
      "grad_norm": 0.4504692554473877,
      "learning_rate": 8.728085369178576e-06,
      "loss": 0.3482,
      "step": 16181
    },
    {
      "epoch": 0.7426001560277179,
      "grad_norm": 0.49562129378318787,
      "learning_rate": 8.727921977689506e-06,
      "loss": 0.4734,
      "step": 16182
    },
    {
      "epoch": 0.7426460465329723,
      "grad_norm": 0.4686272442340851,
      "learning_rate": 8.72775857723596e-06,
      "loss": 0.4348,
      "step": 16183
    },
    {
      "epoch": 0.7426919370382268,
      "grad_norm": 0.46457576751708984,
      "learning_rate": 8.727595167818332e-06,
      "loss": 0.4272,
      "step": 16184
    },
    {
      "epoch": 0.7427378275434813,
      "grad_norm": 0.46697044372558594,
      "learning_rate": 8.727431749437013e-06,
      "loss": 0.4274,
      "step": 16185
    },
    {
      "epoch": 0.7427837180487357,
      "grad_norm": 0.46541154384613037,
      "learning_rate": 8.727268322092397e-06,
      "loss": 0.3902,
      "step": 16186
    },
    {
      "epoch": 0.7428296085539902,
      "grad_norm": 0.4154607951641083,
      "learning_rate": 8.727104885784878e-06,
      "loss": 0.3416,
      "step": 16187
    },
    {
      "epoch": 0.7428754990592447,
      "grad_norm": 0.4436182379722595,
      "learning_rate": 8.726941440514846e-06,
      "loss": 0.3524,
      "step": 16188
    },
    {
      "epoch": 0.7429213895644992,
      "grad_norm": 0.4693727195262909,
      "learning_rate": 8.726777986282699e-06,
      "loss": 0.462,
      "step": 16189
    },
    {
      "epoch": 0.7429672800697535,
      "grad_norm": 0.4393177330493927,
      "learning_rate": 8.726614523088826e-06,
      "loss": 0.371,
      "step": 16190
    },
    {
      "epoch": 0.743013170575008,
      "grad_norm": 0.4943798780441284,
      "learning_rate": 8.72645105093362e-06,
      "loss": 0.4609,
      "step": 16191
    },
    {
      "epoch": 0.7430590610802625,
      "grad_norm": 0.46387019753456116,
      "learning_rate": 8.726287569817477e-06,
      "loss": 0.3802,
      "step": 16192
    },
    {
      "epoch": 0.7431049515855169,
      "grad_norm": 0.46085554361343384,
      "learning_rate": 8.726124079740788e-06,
      "loss": 0.3758,
      "step": 16193
    },
    {
      "epoch": 0.7431508420907714,
      "grad_norm": 0.44855400919914246,
      "learning_rate": 8.725960580703945e-06,
      "loss": 0.3517,
      "step": 16194
    },
    {
      "epoch": 0.7431967325960259,
      "grad_norm": 0.47725751996040344,
      "learning_rate": 8.725797072707345e-06,
      "loss": 0.4232,
      "step": 16195
    },
    {
      "epoch": 0.7432426231012803,
      "grad_norm": 0.4839957654476166,
      "learning_rate": 8.725633555751376e-06,
      "loss": 0.44,
      "step": 16196
    },
    {
      "epoch": 0.7432885136065348,
      "grad_norm": 0.45585089921951294,
      "learning_rate": 8.725470029836438e-06,
      "loss": 0.3766,
      "step": 16197
    },
    {
      "epoch": 0.7433344041117893,
      "grad_norm": 0.4755672514438629,
      "learning_rate": 8.725306494962917e-06,
      "loss": 0.4036,
      "step": 16198
    },
    {
      "epoch": 0.7433802946170437,
      "grad_norm": 0.49170634150505066,
      "learning_rate": 8.725142951131212e-06,
      "loss": 0.4461,
      "step": 16199
    },
    {
      "epoch": 0.7434261851222982,
      "grad_norm": 0.47405627369880676,
      "learning_rate": 8.724979398341714e-06,
      "loss": 0.4342,
      "step": 16200
    },
    {
      "epoch": 0.7434720756275527,
      "grad_norm": 0.48027193546295166,
      "learning_rate": 8.724815836594814e-06,
      "loss": 0.4459,
      "step": 16201
    },
    {
      "epoch": 0.7435179661328071,
      "grad_norm": 0.42188361287117004,
      "learning_rate": 8.72465226589091e-06,
      "loss": 0.3066,
      "step": 16202
    },
    {
      "epoch": 0.7435638566380616,
      "grad_norm": 0.4514961242675781,
      "learning_rate": 8.72448868623039e-06,
      "loss": 0.3437,
      "step": 16203
    },
    {
      "epoch": 0.7436097471433161,
      "grad_norm": 0.4493747353553772,
      "learning_rate": 8.724325097613651e-06,
      "loss": 0.4069,
      "step": 16204
    },
    {
      "epoch": 0.7436556376485705,
      "grad_norm": 0.5089772939682007,
      "learning_rate": 8.724161500041088e-06,
      "loss": 0.4212,
      "step": 16205
    },
    {
      "epoch": 0.743701528153825,
      "grad_norm": 0.4710423946380615,
      "learning_rate": 8.723997893513089e-06,
      "loss": 0.4122,
      "step": 16206
    },
    {
      "epoch": 0.7437474186590795,
      "grad_norm": 0.4489966332912445,
      "learning_rate": 8.723834278030053e-06,
      "loss": 0.3714,
      "step": 16207
    },
    {
      "epoch": 0.743793309164334,
      "grad_norm": 0.4418530762195587,
      "learning_rate": 8.72367065359237e-06,
      "loss": 0.383,
      "step": 16208
    },
    {
      "epoch": 0.7438391996695883,
      "grad_norm": 0.463349312543869,
      "learning_rate": 8.723507020200433e-06,
      "loss": 0.4312,
      "step": 16209
    },
    {
      "epoch": 0.7438850901748428,
      "grad_norm": 0.5727741718292236,
      "learning_rate": 8.723343377854636e-06,
      "loss": 0.4575,
      "step": 16210
    },
    {
      "epoch": 0.7439309806800973,
      "grad_norm": 0.46262651681900024,
      "learning_rate": 8.723179726555376e-06,
      "loss": 0.4151,
      "step": 16211
    },
    {
      "epoch": 0.7439768711853517,
      "grad_norm": 0.44559091329574585,
      "learning_rate": 8.723016066303042e-06,
      "loss": 0.3787,
      "step": 16212
    },
    {
      "epoch": 0.7440227616906062,
      "grad_norm": 0.4506240785121918,
      "learning_rate": 8.722852397098029e-06,
      "loss": 0.4261,
      "step": 16213
    },
    {
      "epoch": 0.7440686521958607,
      "grad_norm": 0.4291321635246277,
      "learning_rate": 8.722688718940733e-06,
      "loss": 0.3241,
      "step": 16214
    },
    {
      "epoch": 0.7441145427011151,
      "grad_norm": 0.4535481035709381,
      "learning_rate": 8.722525031831544e-06,
      "loss": 0.4105,
      "step": 16215
    },
    {
      "epoch": 0.7441604332063696,
      "grad_norm": 0.4677073359489441,
      "learning_rate": 8.722361335770858e-06,
      "loss": 0.4491,
      "step": 16216
    },
    {
      "epoch": 0.7442063237116241,
      "grad_norm": 0.4451531171798706,
      "learning_rate": 8.722197630759069e-06,
      "loss": 0.4068,
      "step": 16217
    },
    {
      "epoch": 0.7442522142168785,
      "grad_norm": 0.4810009300708771,
      "learning_rate": 8.722033916796567e-06,
      "loss": 0.47,
      "step": 16218
    },
    {
      "epoch": 0.744298104722133,
      "grad_norm": 0.4522962272167206,
      "learning_rate": 8.721870193883748e-06,
      "loss": 0.3713,
      "step": 16219
    },
    {
      "epoch": 0.7443439952273875,
      "grad_norm": 0.4376360774040222,
      "learning_rate": 8.721706462021007e-06,
      "loss": 0.3839,
      "step": 16220
    },
    {
      "epoch": 0.7443898857326419,
      "grad_norm": 0.4575299024581909,
      "learning_rate": 8.721542721208738e-06,
      "loss": 0.396,
      "step": 16221
    },
    {
      "epoch": 0.7444357762378964,
      "grad_norm": 0.43664589524269104,
      "learning_rate": 8.721378971447331e-06,
      "loss": 0.3538,
      "step": 16222
    },
    {
      "epoch": 0.7444816667431509,
      "grad_norm": 0.440625935792923,
      "learning_rate": 8.721215212737183e-06,
      "loss": 0.3999,
      "step": 16223
    },
    {
      "epoch": 0.7445275572484054,
      "grad_norm": 0.4351752996444702,
      "learning_rate": 8.721051445078686e-06,
      "loss": 0.3299,
      "step": 16224
    },
    {
      "epoch": 0.7445734477536597,
      "grad_norm": 0.5112953782081604,
      "learning_rate": 8.720887668472235e-06,
      "loss": 0.4671,
      "step": 16225
    },
    {
      "epoch": 0.7446193382589142,
      "grad_norm": 0.4620542824268341,
      "learning_rate": 8.720723882918225e-06,
      "loss": 0.4266,
      "step": 16226
    },
    {
      "epoch": 0.7446652287641687,
      "grad_norm": 0.4738379120826721,
      "learning_rate": 8.720560088417048e-06,
      "loss": 0.3883,
      "step": 16227
    },
    {
      "epoch": 0.7447111192694231,
      "grad_norm": 0.46384352445602417,
      "learning_rate": 8.720396284969096e-06,
      "loss": 0.4194,
      "step": 16228
    },
    {
      "epoch": 0.7447570097746776,
      "grad_norm": 0.4519800543785095,
      "learning_rate": 8.720232472574768e-06,
      "loss": 0.4543,
      "step": 16229
    },
    {
      "epoch": 0.7448029002799321,
      "grad_norm": 0.5369689464569092,
      "learning_rate": 8.720068651234452e-06,
      "loss": 0.4525,
      "step": 16230
    },
    {
      "epoch": 0.7448487907851865,
      "grad_norm": 0.4495983421802521,
      "learning_rate": 8.719904820948548e-06,
      "loss": 0.425,
      "step": 16231
    },
    {
      "epoch": 0.744894681290441,
      "grad_norm": 0.4389208257198334,
      "learning_rate": 8.719740981717445e-06,
      "loss": 0.3543,
      "step": 16232
    },
    {
      "epoch": 0.7449405717956955,
      "grad_norm": 0.4325524568557739,
      "learning_rate": 8.71957713354154e-06,
      "loss": 0.341,
      "step": 16233
    },
    {
      "epoch": 0.7449864623009499,
      "grad_norm": 0.44216424226760864,
      "learning_rate": 8.719413276421225e-06,
      "loss": 0.3566,
      "step": 16234
    },
    {
      "epoch": 0.7450323528062044,
      "grad_norm": 0.45559072494506836,
      "learning_rate": 8.719249410356895e-06,
      "loss": 0.3588,
      "step": 16235
    },
    {
      "epoch": 0.7450782433114589,
      "grad_norm": 0.4964854121208191,
      "learning_rate": 8.719085535348946e-06,
      "loss": 0.5206,
      "step": 16236
    },
    {
      "epoch": 0.7451241338167133,
      "grad_norm": 0.49076318740844727,
      "learning_rate": 8.718921651397767e-06,
      "loss": 0.4464,
      "step": 16237
    },
    {
      "epoch": 0.7451700243219678,
      "grad_norm": 0.47047311067581177,
      "learning_rate": 8.718757758503758e-06,
      "loss": 0.4493,
      "step": 16238
    },
    {
      "epoch": 0.7452159148272223,
      "grad_norm": 0.4499049186706543,
      "learning_rate": 8.718593856667307e-06,
      "loss": 0.3685,
      "step": 16239
    },
    {
      "epoch": 0.7452618053324767,
      "grad_norm": 0.4402592182159424,
      "learning_rate": 8.718429945888813e-06,
      "loss": 0.3348,
      "step": 16240
    },
    {
      "epoch": 0.7453076958377312,
      "grad_norm": 0.4658980965614319,
      "learning_rate": 8.71826602616867e-06,
      "loss": 0.4115,
      "step": 16241
    },
    {
      "epoch": 0.7453535863429857,
      "grad_norm": 0.420261949300766,
      "learning_rate": 8.71810209750727e-06,
      "loss": 0.3359,
      "step": 16242
    },
    {
      "epoch": 0.7453994768482402,
      "grad_norm": 0.4881352186203003,
      "learning_rate": 8.717938159905006e-06,
      "loss": 0.4537,
      "step": 16243
    },
    {
      "epoch": 0.7454453673534945,
      "grad_norm": 0.41938018798828125,
      "learning_rate": 8.717774213362277e-06,
      "loss": 0.3201,
      "step": 16244
    },
    {
      "epoch": 0.745491257858749,
      "grad_norm": 0.44861623644828796,
      "learning_rate": 8.717610257879472e-06,
      "loss": 0.3196,
      "step": 16245
    },
    {
      "epoch": 0.7455371483640035,
      "grad_norm": 0.4728897213935852,
      "learning_rate": 8.717446293456989e-06,
      "loss": 0.4528,
      "step": 16246
    },
    {
      "epoch": 0.7455830388692579,
      "grad_norm": 0.4771125316619873,
      "learning_rate": 8.71728232009522e-06,
      "loss": 0.4281,
      "step": 16247
    },
    {
      "epoch": 0.7456289293745124,
      "grad_norm": 0.47516292333602905,
      "learning_rate": 8.717118337794562e-06,
      "loss": 0.4029,
      "step": 16248
    },
    {
      "epoch": 0.7456748198797669,
      "grad_norm": 0.48476868867874146,
      "learning_rate": 8.716954346555406e-06,
      "loss": 0.4898,
      "step": 16249
    },
    {
      "epoch": 0.7457207103850213,
      "grad_norm": 0.4975925385951996,
      "learning_rate": 8.716790346378147e-06,
      "loss": 0.4811,
      "step": 16250
    },
    {
      "epoch": 0.7457666008902758,
      "grad_norm": 0.4206581115722656,
      "learning_rate": 8.716626337263183e-06,
      "loss": 0.3017,
      "step": 16251
    },
    {
      "epoch": 0.7458124913955303,
      "grad_norm": 0.4390853941440582,
      "learning_rate": 8.716462319210903e-06,
      "loss": 0.3789,
      "step": 16252
    },
    {
      "epoch": 0.7458583819007847,
      "grad_norm": 0.4813790023326874,
      "learning_rate": 8.716298292221706e-06,
      "loss": 0.3625,
      "step": 16253
    },
    {
      "epoch": 0.7459042724060392,
      "grad_norm": 0.47483494877815247,
      "learning_rate": 8.716134256295982e-06,
      "loss": 0.405,
      "step": 16254
    },
    {
      "epoch": 0.7459501629112937,
      "grad_norm": 0.448069304227829,
      "learning_rate": 8.715970211434132e-06,
      "loss": 0.3949,
      "step": 16255
    },
    {
      "epoch": 0.7459960534165481,
      "grad_norm": 0.4797340929508209,
      "learning_rate": 8.715806157636543e-06,
      "loss": 0.4413,
      "step": 16256
    },
    {
      "epoch": 0.7460419439218026,
      "grad_norm": 0.457722544670105,
      "learning_rate": 8.715642094903614e-06,
      "loss": 0.3786,
      "step": 16257
    },
    {
      "epoch": 0.7460878344270571,
      "grad_norm": 0.427398145198822,
      "learning_rate": 8.71547802323574e-06,
      "loss": 0.3218,
      "step": 16258
    },
    {
      "epoch": 0.7461337249323116,
      "grad_norm": 0.418654203414917,
      "learning_rate": 8.715313942633312e-06,
      "loss": 0.3191,
      "step": 16259
    },
    {
      "epoch": 0.746179615437566,
      "grad_norm": 0.4768824577331543,
      "learning_rate": 8.715149853096727e-06,
      "loss": 0.4881,
      "step": 16260
    },
    {
      "epoch": 0.7462255059428204,
      "grad_norm": 0.4630188047885895,
      "learning_rate": 8.71498575462638e-06,
      "loss": 0.3974,
      "step": 16261
    },
    {
      "epoch": 0.7462713964480749,
      "grad_norm": 0.4186602532863617,
      "learning_rate": 8.714821647222662e-06,
      "loss": 0.3233,
      "step": 16262
    },
    {
      "epoch": 0.7463172869533293,
      "grad_norm": 0.4300456643104553,
      "learning_rate": 8.714657530885972e-06,
      "loss": 0.3746,
      "step": 16263
    },
    {
      "epoch": 0.7463631774585838,
      "grad_norm": 0.46721217036247253,
      "learning_rate": 8.714493405616704e-06,
      "loss": 0.406,
      "step": 16264
    },
    {
      "epoch": 0.7464090679638383,
      "grad_norm": 0.48147737979888916,
      "learning_rate": 8.714329271415252e-06,
      "loss": 0.3742,
      "step": 16265
    },
    {
      "epoch": 0.7464549584690927,
      "grad_norm": 0.46314895153045654,
      "learning_rate": 8.714165128282008e-06,
      "loss": 0.4183,
      "step": 16266
    },
    {
      "epoch": 0.7465008489743472,
      "grad_norm": 0.4475615620613098,
      "learning_rate": 8.714000976217372e-06,
      "loss": 0.367,
      "step": 16267
    },
    {
      "epoch": 0.7465467394796017,
      "grad_norm": 0.4105927050113678,
      "learning_rate": 8.713836815221733e-06,
      "loss": 0.2798,
      "step": 16268
    },
    {
      "epoch": 0.7465926299848561,
      "grad_norm": 0.42869147658348083,
      "learning_rate": 8.71367264529549e-06,
      "loss": 0.336,
      "step": 16269
    },
    {
      "epoch": 0.7466385204901106,
      "grad_norm": 0.4547214210033417,
      "learning_rate": 8.713508466439037e-06,
      "loss": 0.393,
      "step": 16270
    },
    {
      "epoch": 0.7466844109953651,
      "grad_norm": 0.4529729187488556,
      "learning_rate": 8.713344278652766e-06,
      "loss": 0.3674,
      "step": 16271
    },
    {
      "epoch": 0.7467303015006195,
      "grad_norm": 0.49468713998794556,
      "learning_rate": 8.713180081937077e-06,
      "loss": 0.4506,
      "step": 16272
    },
    {
      "epoch": 0.746776192005874,
      "grad_norm": 0.4592164158821106,
      "learning_rate": 8.71301587629236e-06,
      "loss": 0.4144,
      "step": 16273
    },
    {
      "epoch": 0.7468220825111285,
      "grad_norm": 0.417090505361557,
      "learning_rate": 8.712851661719011e-06,
      "loss": 0.3323,
      "step": 16274
    },
    {
      "epoch": 0.7468679730163829,
      "grad_norm": 0.4867130219936371,
      "learning_rate": 8.712687438217427e-06,
      "loss": 0.4587,
      "step": 16275
    },
    {
      "epoch": 0.7469138635216374,
      "grad_norm": 0.47335928678512573,
      "learning_rate": 8.712523205788001e-06,
      "loss": 0.409,
      "step": 16276
    },
    {
      "epoch": 0.7469597540268919,
      "grad_norm": 0.4299606382846832,
      "learning_rate": 8.71235896443113e-06,
      "loss": 0.3584,
      "step": 16277
    },
    {
      "epoch": 0.7470056445321464,
      "grad_norm": 0.45408281683921814,
      "learning_rate": 8.712194714147204e-06,
      "loss": 0.3485,
      "step": 16278
    },
    {
      "epoch": 0.7470515350374007,
      "grad_norm": 0.4508644938468933,
      "learning_rate": 8.712030454936621e-06,
      "loss": 0.3595,
      "step": 16279
    },
    {
      "epoch": 0.7470974255426552,
      "grad_norm": 0.458680659532547,
      "learning_rate": 8.71186618679978e-06,
      "loss": 0.3975,
      "step": 16280
    },
    {
      "epoch": 0.7471433160479097,
      "grad_norm": 0.5161157250404358,
      "learning_rate": 8.711701909737071e-06,
      "loss": 0.5008,
      "step": 16281
    },
    {
      "epoch": 0.7471892065531641,
      "grad_norm": 0.47313717007637024,
      "learning_rate": 8.71153762374889e-06,
      "loss": 0.44,
      "step": 16282
    },
    {
      "epoch": 0.7472350970584186,
      "grad_norm": 0.44445687532424927,
      "learning_rate": 8.711373328835633e-06,
      "loss": 0.4042,
      "step": 16283
    },
    {
      "epoch": 0.7472809875636731,
      "grad_norm": 0.4926909804344177,
      "learning_rate": 8.711209024997694e-06,
      "loss": 0.4408,
      "step": 16284
    },
    {
      "epoch": 0.7473268780689275,
      "grad_norm": 0.4849625527858734,
      "learning_rate": 8.711044712235467e-06,
      "loss": 0.435,
      "step": 16285
    },
    {
      "epoch": 0.747372768574182,
      "grad_norm": 0.4572621285915375,
      "learning_rate": 8.71088039054935e-06,
      "loss": 0.3374,
      "step": 16286
    },
    {
      "epoch": 0.7474186590794365,
      "grad_norm": 0.4671629071235657,
      "learning_rate": 8.710716059939736e-06,
      "loss": 0.3991,
      "step": 16287
    },
    {
      "epoch": 0.7474645495846909,
      "grad_norm": 0.47363194823265076,
      "learning_rate": 8.710551720407024e-06,
      "loss": 0.405,
      "step": 16288
    },
    {
      "epoch": 0.7475104400899454,
      "grad_norm": 0.45135387778282166,
      "learning_rate": 8.710387371951603e-06,
      "loss": 0.3688,
      "step": 16289
    },
    {
      "epoch": 0.7475563305951999,
      "grad_norm": 0.44756782054901123,
      "learning_rate": 8.710223014573872e-06,
      "loss": 0.3438,
      "step": 16290
    },
    {
      "epoch": 0.7476022211004543,
      "grad_norm": 0.44525274634361267,
      "learning_rate": 8.710058648274228e-06,
      "loss": 0.3487,
      "step": 16291
    },
    {
      "epoch": 0.7476481116057088,
      "grad_norm": 0.4298510253429413,
      "learning_rate": 8.70989427305306e-06,
      "loss": 0.3668,
      "step": 16292
    },
    {
      "epoch": 0.7476940021109633,
      "grad_norm": 0.5232069492340088,
      "learning_rate": 8.709729888910771e-06,
      "loss": 0.5645,
      "step": 16293
    },
    {
      "epoch": 0.7477398926162176,
      "grad_norm": 0.46848854422569275,
      "learning_rate": 8.70956549584775e-06,
      "loss": 0.4005,
      "step": 16294
    },
    {
      "epoch": 0.7477857831214721,
      "grad_norm": 0.5575319528579712,
      "learning_rate": 8.709401093864396e-06,
      "loss": 0.3737,
      "step": 16295
    },
    {
      "epoch": 0.7478316736267266,
      "grad_norm": 0.43608811497688293,
      "learning_rate": 8.709236682961101e-06,
      "loss": 0.3516,
      "step": 16296
    },
    {
      "epoch": 0.7478775641319811,
      "grad_norm": 0.4218136668205261,
      "learning_rate": 8.709072263138265e-06,
      "loss": 0.333,
      "step": 16297
    },
    {
      "epoch": 0.7479234546372355,
      "grad_norm": 0.4742417335510254,
      "learning_rate": 8.708907834396281e-06,
      "loss": 0.426,
      "step": 16298
    },
    {
      "epoch": 0.74796934514249,
      "grad_norm": 0.49672842025756836,
      "learning_rate": 8.708743396735541e-06,
      "loss": 0.4655,
      "step": 16299
    },
    {
      "epoch": 0.7480152356477445,
      "grad_norm": 0.42824864387512207,
      "learning_rate": 8.708578950156447e-06,
      "loss": 0.3524,
      "step": 16300
    },
    {
      "epoch": 0.7480611261529989,
      "grad_norm": 0.5222131609916687,
      "learning_rate": 8.70841449465939e-06,
      "loss": 0.5447,
      "step": 16301
    },
    {
      "epoch": 0.7481070166582534,
      "grad_norm": 0.4452823996543884,
      "learning_rate": 8.708250030244766e-06,
      "loss": 0.3818,
      "step": 16302
    },
    {
      "epoch": 0.7481529071635079,
      "grad_norm": 0.45923227071762085,
      "learning_rate": 8.708085556912971e-06,
      "loss": 0.4081,
      "step": 16303
    },
    {
      "epoch": 0.7481987976687623,
      "grad_norm": 0.44174447655677795,
      "learning_rate": 8.707921074664402e-06,
      "loss": 0.3811,
      "step": 16304
    },
    {
      "epoch": 0.7482446881740168,
      "grad_norm": 0.5227293372154236,
      "learning_rate": 8.707756583499452e-06,
      "loss": 0.5099,
      "step": 16305
    },
    {
      "epoch": 0.7482905786792713,
      "grad_norm": 0.4361780285835266,
      "learning_rate": 8.707592083418516e-06,
      "loss": 0.3473,
      "step": 16306
    },
    {
      "epoch": 0.7483364691845257,
      "grad_norm": 0.4863373041152954,
      "learning_rate": 8.707427574421994e-06,
      "loss": 0.4552,
      "step": 16307
    },
    {
      "epoch": 0.7483823596897802,
      "grad_norm": 0.4374549686908722,
      "learning_rate": 8.707263056510278e-06,
      "loss": 0.3911,
      "step": 16308
    },
    {
      "epoch": 0.7484282501950347,
      "grad_norm": 0.43347620964050293,
      "learning_rate": 8.707098529683764e-06,
      "loss": 0.3572,
      "step": 16309
    },
    {
      "epoch": 0.7484741407002891,
      "grad_norm": 0.4563251733779907,
      "learning_rate": 8.706933993942848e-06,
      "loss": 0.4066,
      "step": 16310
    },
    {
      "epoch": 0.7485200312055436,
      "grad_norm": 0.5133817195892334,
      "learning_rate": 8.706769449287923e-06,
      "loss": 0.5011,
      "step": 16311
    },
    {
      "epoch": 0.748565921710798,
      "grad_norm": 0.4653419554233551,
      "learning_rate": 8.706604895719391e-06,
      "loss": 0.4015,
      "step": 16312
    },
    {
      "epoch": 0.7486118122160526,
      "grad_norm": 0.5610264539718628,
      "learning_rate": 8.706440333237645e-06,
      "loss": 0.4136,
      "step": 16313
    },
    {
      "epoch": 0.7486577027213069,
      "grad_norm": 0.4334607720375061,
      "learning_rate": 8.706275761843078e-06,
      "loss": 0.3714,
      "step": 16314
    },
    {
      "epoch": 0.7487035932265614,
      "grad_norm": 0.447208970785141,
      "learning_rate": 8.706111181536087e-06,
      "loss": 0.4384,
      "step": 16315
    },
    {
      "epoch": 0.7487494837318159,
      "grad_norm": 0.4321472644805908,
      "learning_rate": 8.705946592317067e-06,
      "loss": 0.3503,
      "step": 16316
    },
    {
      "epoch": 0.7487953742370703,
      "grad_norm": 0.48500803112983704,
      "learning_rate": 8.705781994186418e-06,
      "loss": 0.4111,
      "step": 16317
    },
    {
      "epoch": 0.7488412647423248,
      "grad_norm": 0.5076285600662231,
      "learning_rate": 8.705617387144532e-06,
      "loss": 0.4968,
      "step": 16318
    },
    {
      "epoch": 0.7488871552475793,
      "grad_norm": 0.4363185167312622,
      "learning_rate": 8.705452771191804e-06,
      "loss": 0.3582,
      "step": 16319
    },
    {
      "epoch": 0.7489330457528337,
      "grad_norm": 0.4652787744998932,
      "learning_rate": 8.705288146328635e-06,
      "loss": 0.4665,
      "step": 16320
    },
    {
      "epoch": 0.7489789362580882,
      "grad_norm": 0.4444766938686371,
      "learning_rate": 8.705123512555414e-06,
      "loss": 0.362,
      "step": 16321
    },
    {
      "epoch": 0.7490248267633427,
      "grad_norm": 0.47886624932289124,
      "learning_rate": 8.704958869872542e-06,
      "loss": 0.3991,
      "step": 16322
    },
    {
      "epoch": 0.7490707172685971,
      "grad_norm": 0.4478308856487274,
      "learning_rate": 8.704794218280413e-06,
      "loss": 0.4066,
      "step": 16323
    },
    {
      "epoch": 0.7491166077738516,
      "grad_norm": 0.4371248781681061,
      "learning_rate": 8.704629557779423e-06,
      "loss": 0.3767,
      "step": 16324
    },
    {
      "epoch": 0.7491624982791061,
      "grad_norm": 0.46291181445121765,
      "learning_rate": 8.704464888369967e-06,
      "loss": 0.4233,
      "step": 16325
    },
    {
      "epoch": 0.7492083887843605,
      "grad_norm": 0.4726863503456116,
      "learning_rate": 8.704300210052444e-06,
      "loss": 0.4017,
      "step": 16326
    },
    {
      "epoch": 0.749254279289615,
      "grad_norm": 0.4388914108276367,
      "learning_rate": 8.704135522827247e-06,
      "loss": 0.4206,
      "step": 16327
    },
    {
      "epoch": 0.7493001697948695,
      "grad_norm": 0.4627699851989746,
      "learning_rate": 8.703970826694774e-06,
      "loss": 0.3713,
      "step": 16328
    },
    {
      "epoch": 0.7493460603001239,
      "grad_norm": 0.4611329436302185,
      "learning_rate": 8.70380612165542e-06,
      "loss": 0.3726,
      "step": 16329
    },
    {
      "epoch": 0.7493919508053783,
      "grad_norm": 0.4225649833679199,
      "learning_rate": 8.70364140770958e-06,
      "loss": 0.2842,
      "step": 16330
    },
    {
      "epoch": 0.7494378413106328,
      "grad_norm": 0.4449273943901062,
      "learning_rate": 8.703476684857653e-06,
      "loss": 0.4146,
      "step": 16331
    },
    {
      "epoch": 0.7494837318158873,
      "grad_norm": 0.46716463565826416,
      "learning_rate": 8.703311953100033e-06,
      "loss": 0.4352,
      "step": 16332
    },
    {
      "epoch": 0.7495296223211417,
      "grad_norm": 0.43684014678001404,
      "learning_rate": 8.703147212437115e-06,
      "loss": 0.3728,
      "step": 16333
    },
    {
      "epoch": 0.7495755128263962,
      "grad_norm": 0.49534955620765686,
      "learning_rate": 8.702982462869298e-06,
      "loss": 0.4512,
      "step": 16334
    },
    {
      "epoch": 0.7496214033316507,
      "grad_norm": 0.45564666390419006,
      "learning_rate": 8.702817704396977e-06,
      "loss": 0.394,
      "step": 16335
    },
    {
      "epoch": 0.7496672938369051,
      "grad_norm": 0.46841639280319214,
      "learning_rate": 8.702652937020548e-06,
      "loss": 0.4657,
      "step": 16336
    },
    {
      "epoch": 0.7497131843421596,
      "grad_norm": 0.4222855269908905,
      "learning_rate": 8.702488160740408e-06,
      "loss": 0.3589,
      "step": 16337
    },
    {
      "epoch": 0.7497590748474141,
      "grad_norm": 0.49153923988342285,
      "learning_rate": 8.70232337555695e-06,
      "loss": 0.4462,
      "step": 16338
    },
    {
      "epoch": 0.7498049653526685,
      "grad_norm": 0.4328003525733948,
      "learning_rate": 8.702158581470576e-06,
      "loss": 0.3967,
      "step": 16339
    },
    {
      "epoch": 0.749850855857923,
      "grad_norm": 0.4350324869155884,
      "learning_rate": 8.701993778481676e-06,
      "loss": 0.3751,
      "step": 16340
    },
    {
      "epoch": 0.7498967463631775,
      "grad_norm": 0.44061386585235596,
      "learning_rate": 8.701828966590652e-06,
      "loss": 0.3913,
      "step": 16341
    },
    {
      "epoch": 0.7499426368684319,
      "grad_norm": 0.4632883071899414,
      "learning_rate": 8.701664145797897e-06,
      "loss": 0.4319,
      "step": 16342
    },
    {
      "epoch": 0.7499885273736864,
      "grad_norm": 0.43110471963882446,
      "learning_rate": 8.701499316103807e-06,
      "loss": 0.3248,
      "step": 16343
    },
    {
      "epoch": 0.7500344178789409,
      "grad_norm": 0.4401867687702179,
      "learning_rate": 8.70133447750878e-06,
      "loss": 0.4011,
      "step": 16344
    },
    {
      "epoch": 0.7500803083841953,
      "grad_norm": 0.45379939675331116,
      "learning_rate": 8.70116963001321e-06,
      "loss": 0.4471,
      "step": 16345
    },
    {
      "epoch": 0.7501261988894498,
      "grad_norm": 0.47710928320884705,
      "learning_rate": 8.7010047736175e-06,
      "loss": 0.4331,
      "step": 16346
    },
    {
      "epoch": 0.7501720893947043,
      "grad_norm": 0.4136173129081726,
      "learning_rate": 8.700839908322037e-06,
      "loss": 0.2936,
      "step": 16347
    },
    {
      "epoch": 0.7502179798999588,
      "grad_norm": 0.45674893260002136,
      "learning_rate": 8.700675034127225e-06,
      "loss": 0.4143,
      "step": 16348
    },
    {
      "epoch": 0.7502638704052131,
      "grad_norm": 0.42345237731933594,
      "learning_rate": 8.700510151033454e-06,
      "loss": 0.3128,
      "step": 16349
    },
    {
      "epoch": 0.7503097609104676,
      "grad_norm": 0.4282125234603882,
      "learning_rate": 8.700345259041125e-06,
      "loss": 0.3288,
      "step": 16350
    },
    {
      "epoch": 0.7503556514157221,
      "grad_norm": 0.45034828782081604,
      "learning_rate": 8.700180358150637e-06,
      "loss": 0.4057,
      "step": 16351
    },
    {
      "epoch": 0.7504015419209765,
      "grad_norm": 0.4370351731777191,
      "learning_rate": 8.700015448362378e-06,
      "loss": 0.3321,
      "step": 16352
    },
    {
      "epoch": 0.750447432426231,
      "grad_norm": 0.4691845774650574,
      "learning_rate": 8.699850529676754e-06,
      "loss": 0.4051,
      "step": 16353
    },
    {
      "epoch": 0.7504933229314855,
      "grad_norm": 0.460775226354599,
      "learning_rate": 8.699685602094154e-06,
      "loss": 0.4517,
      "step": 16354
    },
    {
      "epoch": 0.7505392134367399,
      "grad_norm": 0.47734689712524414,
      "learning_rate": 8.69952066561498e-06,
      "loss": 0.4748,
      "step": 16355
    },
    {
      "epoch": 0.7505851039419944,
      "grad_norm": 0.4480847418308258,
      "learning_rate": 8.699355720239625e-06,
      "loss": 0.3702,
      "step": 16356
    },
    {
      "epoch": 0.7506309944472489,
      "grad_norm": 0.4404084086418152,
      "learning_rate": 8.699190765968486e-06,
      "loss": 0.3583,
      "step": 16357
    },
    {
      "epoch": 0.7506768849525033,
      "grad_norm": 0.48154380917549133,
      "learning_rate": 8.699025802801963e-06,
      "loss": 0.4294,
      "step": 16358
    },
    {
      "epoch": 0.7507227754577578,
      "grad_norm": 0.47850754857063293,
      "learning_rate": 8.69886083074045e-06,
      "loss": 0.4756,
      "step": 16359
    },
    {
      "epoch": 0.7507686659630123,
      "grad_norm": 0.41177526116371155,
      "learning_rate": 8.698695849784344e-06,
      "loss": 0.3374,
      "step": 16360
    },
    {
      "epoch": 0.7508145564682667,
      "grad_norm": 0.448499470949173,
      "learning_rate": 8.698530859934042e-06,
      "loss": 0.4106,
      "step": 16361
    },
    {
      "epoch": 0.7508604469735212,
      "grad_norm": 0.41752296686172485,
      "learning_rate": 8.69836586118994e-06,
      "loss": 0.3313,
      "step": 16362
    },
    {
      "epoch": 0.7509063374787757,
      "grad_norm": 0.43674740195274353,
      "learning_rate": 8.698200853552434e-06,
      "loss": 0.3934,
      "step": 16363
    },
    {
      "epoch": 0.75095222798403,
      "grad_norm": 0.44744357466697693,
      "learning_rate": 8.698035837021925e-06,
      "loss": 0.3776,
      "step": 16364
    },
    {
      "epoch": 0.7509981184892845,
      "grad_norm": 0.4480912685394287,
      "learning_rate": 8.697870811598806e-06,
      "loss": 0.3739,
      "step": 16365
    },
    {
      "epoch": 0.751044008994539,
      "grad_norm": 0.47732213139533997,
      "learning_rate": 8.697705777283474e-06,
      "loss": 0.3588,
      "step": 16366
    },
    {
      "epoch": 0.7510898994997935,
      "grad_norm": 0.43123915791511536,
      "learning_rate": 8.697540734076327e-06,
      "loss": 0.3273,
      "step": 16367
    },
    {
      "epoch": 0.7511357900050479,
      "grad_norm": 0.47396403551101685,
      "learning_rate": 8.69737568197776e-06,
      "loss": 0.3506,
      "step": 16368
    },
    {
      "epoch": 0.7511816805103024,
      "grad_norm": 0.47032302618026733,
      "learning_rate": 8.697210620988174e-06,
      "loss": 0.4085,
      "step": 16369
    },
    {
      "epoch": 0.7512275710155569,
      "grad_norm": 0.43720537424087524,
      "learning_rate": 8.697045551107962e-06,
      "loss": 0.3714,
      "step": 16370
    },
    {
      "epoch": 0.7512734615208113,
      "grad_norm": 0.46425676345825195,
      "learning_rate": 8.696880472337521e-06,
      "loss": 0.4333,
      "step": 16371
    },
    {
      "epoch": 0.7513193520260658,
      "grad_norm": 0.43903085589408875,
      "learning_rate": 8.696715384677252e-06,
      "loss": 0.3721,
      "step": 16372
    },
    {
      "epoch": 0.7513652425313203,
      "grad_norm": 0.4716162085533142,
      "learning_rate": 8.696550288127548e-06,
      "loss": 0.4093,
      "step": 16373
    },
    {
      "epoch": 0.7514111330365747,
      "grad_norm": 0.47652992606163025,
      "learning_rate": 8.696385182688807e-06,
      "loss": 0.4341,
      "step": 16374
    },
    {
      "epoch": 0.7514570235418292,
      "grad_norm": 0.45425400137901306,
      "learning_rate": 8.696220068361427e-06,
      "loss": 0.4149,
      "step": 16375
    },
    {
      "epoch": 0.7515029140470837,
      "grad_norm": 0.469321608543396,
      "learning_rate": 8.696054945145804e-06,
      "loss": 0.3955,
      "step": 16376
    },
    {
      "epoch": 0.7515488045523381,
      "grad_norm": 0.4663131833076477,
      "learning_rate": 8.695889813042334e-06,
      "loss": 0.436,
      "step": 16377
    },
    {
      "epoch": 0.7515946950575926,
      "grad_norm": 0.44780367612838745,
      "learning_rate": 8.695724672051417e-06,
      "loss": 0.4238,
      "step": 16378
    },
    {
      "epoch": 0.7516405855628471,
      "grad_norm": 0.424713134765625,
      "learning_rate": 8.695559522173447e-06,
      "loss": 0.29,
      "step": 16379
    },
    {
      "epoch": 0.7516864760681015,
      "grad_norm": 0.4987753629684448,
      "learning_rate": 8.695394363408825e-06,
      "loss": 0.4812,
      "step": 16380
    },
    {
      "epoch": 0.751732366573356,
      "grad_norm": 0.47493046522140503,
      "learning_rate": 8.695229195757944e-06,
      "loss": 0.4146,
      "step": 16381
    },
    {
      "epoch": 0.7517782570786105,
      "grad_norm": 0.41972991824150085,
      "learning_rate": 8.695064019221203e-06,
      "loss": 0.3408,
      "step": 16382
    },
    {
      "epoch": 0.7518241475838648,
      "grad_norm": 0.4698231816291809,
      "learning_rate": 8.694898833799001e-06,
      "loss": 0.4231,
      "step": 16383
    },
    {
      "epoch": 0.7518700380891193,
      "grad_norm": 0.47360682487487793,
      "learning_rate": 8.694733639491732e-06,
      "loss": 0.4395,
      "step": 16384
    },
    {
      "epoch": 0.7519159285943738,
      "grad_norm": 0.4774980843067169,
      "learning_rate": 8.694568436299795e-06,
      "loss": 0.4217,
      "step": 16385
    },
    {
      "epoch": 0.7519618190996283,
      "grad_norm": 0.4981301724910736,
      "learning_rate": 8.694403224223586e-06,
      "loss": 0.4273,
      "step": 16386
    },
    {
      "epoch": 0.7520077096048827,
      "grad_norm": 0.4048086702823639,
      "learning_rate": 8.694238003263504e-06,
      "loss": 0.2919,
      "step": 16387
    },
    {
      "epoch": 0.7520536001101372,
      "grad_norm": 0.48239797353744507,
      "learning_rate": 8.694072773419947e-06,
      "loss": 0.4837,
      "step": 16388
    },
    {
      "epoch": 0.7520994906153917,
      "grad_norm": 0.45203641057014465,
      "learning_rate": 8.69390753469331e-06,
      "loss": 0.3708,
      "step": 16389
    },
    {
      "epoch": 0.7521453811206461,
      "grad_norm": 0.47509756684303284,
      "learning_rate": 8.69374228708399e-06,
      "loss": 0.4623,
      "step": 16390
    },
    {
      "epoch": 0.7521912716259006,
      "grad_norm": 0.43781110644340515,
      "learning_rate": 8.693577030592387e-06,
      "loss": 0.3517,
      "step": 16391
    },
    {
      "epoch": 0.7522371621311551,
      "grad_norm": 0.49243250489234924,
      "learning_rate": 8.693411765218896e-06,
      "loss": 0.4689,
      "step": 16392
    },
    {
      "epoch": 0.7522830526364095,
      "grad_norm": 0.4801262617111206,
      "learning_rate": 8.693246490963916e-06,
      "loss": 0.4359,
      "step": 16393
    },
    {
      "epoch": 0.752328943141664,
      "grad_norm": 0.46314555406570435,
      "learning_rate": 8.693081207827843e-06,
      "loss": 0.3836,
      "step": 16394
    },
    {
      "epoch": 0.7523748336469185,
      "grad_norm": 0.4475667178630829,
      "learning_rate": 8.692915915811076e-06,
      "loss": 0.3643,
      "step": 16395
    },
    {
      "epoch": 0.7524207241521729,
      "grad_norm": 0.49406206607818604,
      "learning_rate": 8.692750614914011e-06,
      "loss": 0.5016,
      "step": 16396
    },
    {
      "epoch": 0.7524666146574274,
      "grad_norm": 0.47319164872169495,
      "learning_rate": 8.692585305137047e-06,
      "loss": 0.4051,
      "step": 16397
    },
    {
      "epoch": 0.7525125051626819,
      "grad_norm": 0.46962156891822815,
      "learning_rate": 8.69241998648058e-06,
      "loss": 0.4382,
      "step": 16398
    },
    {
      "epoch": 0.7525583956679363,
      "grad_norm": 0.4622677266597748,
      "learning_rate": 8.69225465894501e-06,
      "loss": 0.4107,
      "step": 16399
    },
    {
      "epoch": 0.7526042861731908,
      "grad_norm": 0.46496719121932983,
      "learning_rate": 8.692089322530732e-06,
      "loss": 0.3784,
      "step": 16400
    },
    {
      "epoch": 0.7526501766784452,
      "grad_norm": 0.4915473461151123,
      "learning_rate": 8.691923977238144e-06,
      "loss": 0.4607,
      "step": 16401
    },
    {
      "epoch": 0.7526960671836997,
      "grad_norm": 0.4346165359020233,
      "learning_rate": 8.691758623067643e-06,
      "loss": 0.4136,
      "step": 16402
    },
    {
      "epoch": 0.7527419576889541,
      "grad_norm": 0.45603784918785095,
      "learning_rate": 8.691593260019629e-06,
      "loss": 0.4152,
      "step": 16403
    },
    {
      "epoch": 0.7527878481942086,
      "grad_norm": 0.4316512644290924,
      "learning_rate": 8.691427888094499e-06,
      "loss": 0.3384,
      "step": 16404
    },
    {
      "epoch": 0.7528337386994631,
      "grad_norm": 0.47658947110176086,
      "learning_rate": 8.691262507292649e-06,
      "loss": 0.4681,
      "step": 16405
    },
    {
      "epoch": 0.7528796292047175,
      "grad_norm": 0.46447667479515076,
      "learning_rate": 8.691097117614478e-06,
      "loss": 0.4034,
      "step": 16406
    },
    {
      "epoch": 0.752925519709972,
      "grad_norm": 0.4404204785823822,
      "learning_rate": 8.690931719060382e-06,
      "loss": 0.3795,
      "step": 16407
    },
    {
      "epoch": 0.7529714102152265,
      "grad_norm": 0.4307515323162079,
      "learning_rate": 8.690766311630762e-06,
      "loss": 0.3734,
      "step": 16408
    },
    {
      "epoch": 0.7530173007204809,
      "grad_norm": 0.4438691735267639,
      "learning_rate": 8.690600895326013e-06,
      "loss": 0.3748,
      "step": 16409
    },
    {
      "epoch": 0.7530631912257354,
      "grad_norm": 0.41612228751182556,
      "learning_rate": 8.690435470146535e-06,
      "loss": 0.3,
      "step": 16410
    },
    {
      "epoch": 0.7531090817309899,
      "grad_norm": 0.45534875988960266,
      "learning_rate": 8.690270036092721e-06,
      "loss": 0.3793,
      "step": 16411
    },
    {
      "epoch": 0.7531549722362443,
      "grad_norm": 0.448817640542984,
      "learning_rate": 8.690104593164975e-06,
      "loss": 0.4143,
      "step": 16412
    },
    {
      "epoch": 0.7532008627414988,
      "grad_norm": 0.4827724099159241,
      "learning_rate": 8.689939141363693e-06,
      "loss": 0.4489,
      "step": 16413
    },
    {
      "epoch": 0.7532467532467533,
      "grad_norm": 0.4931320548057556,
      "learning_rate": 8.68977368068927e-06,
      "loss": 0.4687,
      "step": 16414
    },
    {
      "epoch": 0.7532926437520077,
      "grad_norm": 0.5032831430435181,
      "learning_rate": 8.689608211142106e-06,
      "loss": 0.5154,
      "step": 16415
    },
    {
      "epoch": 0.7533385342572622,
      "grad_norm": 0.4744589626789093,
      "learning_rate": 8.6894427327226e-06,
      "loss": 0.4999,
      "step": 16416
    },
    {
      "epoch": 0.7533844247625167,
      "grad_norm": 0.4592866897583008,
      "learning_rate": 8.689277245431148e-06,
      "loss": 0.3769,
      "step": 16417
    },
    {
      "epoch": 0.753430315267771,
      "grad_norm": 0.43515658378601074,
      "learning_rate": 8.689111749268147e-06,
      "loss": 0.3653,
      "step": 16418
    },
    {
      "epoch": 0.7534762057730255,
      "grad_norm": 0.45515766739845276,
      "learning_rate": 8.688946244233998e-06,
      "loss": 0.4512,
      "step": 16419
    },
    {
      "epoch": 0.75352209627828,
      "grad_norm": 0.4465077519416809,
      "learning_rate": 8.688780730329096e-06,
      "loss": 0.3767,
      "step": 16420
    },
    {
      "epoch": 0.7535679867835345,
      "grad_norm": 0.43124011158943176,
      "learning_rate": 8.688615207553843e-06,
      "loss": 0.3691,
      "step": 16421
    },
    {
      "epoch": 0.7536138772887889,
      "grad_norm": 0.4072433114051819,
      "learning_rate": 8.688449675908633e-06,
      "loss": 0.3323,
      "step": 16422
    },
    {
      "epoch": 0.7536597677940434,
      "grad_norm": 0.4665652811527252,
      "learning_rate": 8.688284135393867e-06,
      "loss": 0.454,
      "step": 16423
    },
    {
      "epoch": 0.7537056582992979,
      "grad_norm": 0.4780343770980835,
      "learning_rate": 8.688118586009941e-06,
      "loss": 0.3746,
      "step": 16424
    },
    {
      "epoch": 0.7537515488045523,
      "grad_norm": 0.4679764211177826,
      "learning_rate": 8.687953027757251e-06,
      "loss": 0.4435,
      "step": 16425
    },
    {
      "epoch": 0.7537974393098068,
      "grad_norm": 0.4295421838760376,
      "learning_rate": 8.687787460636203e-06,
      "loss": 0.3573,
      "step": 16426
    },
    {
      "epoch": 0.7538433298150613,
      "grad_norm": 0.45141521096229553,
      "learning_rate": 8.687621884647187e-06,
      "loss": 0.3828,
      "step": 16427
    },
    {
      "epoch": 0.7538892203203157,
      "grad_norm": 0.45784273743629456,
      "learning_rate": 8.687456299790604e-06,
      "loss": 0.3848,
      "step": 16428
    },
    {
      "epoch": 0.7539351108255702,
      "grad_norm": 0.45207175612449646,
      "learning_rate": 8.687290706066854e-06,
      "loss": 0.4452,
      "step": 16429
    },
    {
      "epoch": 0.7539810013308247,
      "grad_norm": 0.45299622416496277,
      "learning_rate": 8.687125103476332e-06,
      "loss": 0.4178,
      "step": 16430
    },
    {
      "epoch": 0.7540268918360791,
      "grad_norm": 0.42734596133232117,
      "learning_rate": 8.68695949201944e-06,
      "loss": 0.3321,
      "step": 16431
    },
    {
      "epoch": 0.7540727823413336,
      "grad_norm": 0.43607038259506226,
      "learning_rate": 8.686793871696574e-06,
      "loss": 0.3247,
      "step": 16432
    },
    {
      "epoch": 0.7541186728465881,
      "grad_norm": 0.520976185798645,
      "learning_rate": 8.686628242508129e-06,
      "loss": 0.4701,
      "step": 16433
    },
    {
      "epoch": 0.7541645633518425,
      "grad_norm": 0.4357532560825348,
      "learning_rate": 8.686462604454509e-06,
      "loss": 0.3318,
      "step": 16434
    },
    {
      "epoch": 0.754210453857097,
      "grad_norm": 0.4754031002521515,
      "learning_rate": 8.68629695753611e-06,
      "loss": 0.3997,
      "step": 16435
    },
    {
      "epoch": 0.7542563443623514,
      "grad_norm": 0.4282938241958618,
      "learning_rate": 8.686131301753329e-06,
      "loss": 0.3727,
      "step": 16436
    },
    {
      "epoch": 0.754302234867606,
      "grad_norm": 0.48741579055786133,
      "learning_rate": 8.685965637106567e-06,
      "loss": 0.4259,
      "step": 16437
    },
    {
      "epoch": 0.7543481253728603,
      "grad_norm": 0.45609360933303833,
      "learning_rate": 8.68579996359622e-06,
      "loss": 0.3947,
      "step": 16438
    },
    {
      "epoch": 0.7543940158781148,
      "grad_norm": 0.44795170426368713,
      "learning_rate": 8.685634281222687e-06,
      "loss": 0.3983,
      "step": 16439
    },
    {
      "epoch": 0.7544399063833693,
      "grad_norm": 0.5011052489280701,
      "learning_rate": 8.685468589986369e-06,
      "loss": 0.4525,
      "step": 16440
    },
    {
      "epoch": 0.7544857968886237,
      "grad_norm": 0.4526043236255646,
      "learning_rate": 8.685302889887661e-06,
      "loss": 0.3774,
      "step": 16441
    },
    {
      "epoch": 0.7545316873938782,
      "grad_norm": 0.47486117482185364,
      "learning_rate": 8.685137180926961e-06,
      "loss": 0.4035,
      "step": 16442
    },
    {
      "epoch": 0.7545775778991327,
      "grad_norm": 0.46526750922203064,
      "learning_rate": 8.68497146310467e-06,
      "loss": 0.416,
      "step": 16443
    },
    {
      "epoch": 0.7546234684043871,
      "grad_norm": 0.4377737045288086,
      "learning_rate": 8.684805736421187e-06,
      "loss": 0.3469,
      "step": 16444
    },
    {
      "epoch": 0.7546693589096416,
      "grad_norm": 0.449629545211792,
      "learning_rate": 8.684640000876909e-06,
      "loss": 0.4686,
      "step": 16445
    },
    {
      "epoch": 0.7547152494148961,
      "grad_norm": 0.45988085865974426,
      "learning_rate": 8.684474256472234e-06,
      "loss": 0.4318,
      "step": 16446
    },
    {
      "epoch": 0.7547611399201505,
      "grad_norm": 0.4764769375324249,
      "learning_rate": 8.684308503207562e-06,
      "loss": 0.4015,
      "step": 16447
    },
    {
      "epoch": 0.754807030425405,
      "grad_norm": 0.46452459692955017,
      "learning_rate": 8.68414274108329e-06,
      "loss": 0.4128,
      "step": 16448
    },
    {
      "epoch": 0.7548529209306595,
      "grad_norm": 0.43204861879348755,
      "learning_rate": 8.683976970099817e-06,
      "loss": 0.37,
      "step": 16449
    },
    {
      "epoch": 0.7548988114359139,
      "grad_norm": 0.4518873989582062,
      "learning_rate": 8.683811190257544e-06,
      "loss": 0.4096,
      "step": 16450
    },
    {
      "epoch": 0.7549447019411684,
      "grad_norm": 0.4920696020126343,
      "learning_rate": 8.683645401556866e-06,
      "loss": 0.4952,
      "step": 16451
    },
    {
      "epoch": 0.7549905924464229,
      "grad_norm": 0.4557972252368927,
      "learning_rate": 8.683479603998183e-06,
      "loss": 0.3712,
      "step": 16452
    },
    {
      "epoch": 0.7550364829516772,
      "grad_norm": 0.43759390711784363,
      "learning_rate": 8.683313797581894e-06,
      "loss": 0.3727,
      "step": 16453
    },
    {
      "epoch": 0.7550823734569317,
      "grad_norm": 0.47462791204452515,
      "learning_rate": 8.6831479823084e-06,
      "loss": 0.4151,
      "step": 16454
    },
    {
      "epoch": 0.7551282639621862,
      "grad_norm": 0.4747932553291321,
      "learning_rate": 8.682982158178096e-06,
      "loss": 0.442,
      "step": 16455
    },
    {
      "epoch": 0.7551741544674407,
      "grad_norm": 0.4594792127609253,
      "learning_rate": 8.682816325191383e-06,
      "loss": 0.4028,
      "step": 16456
    },
    {
      "epoch": 0.7552200449726951,
      "grad_norm": 0.48942697048187256,
      "learning_rate": 8.682650483348657e-06,
      "loss": 0.4626,
      "step": 16457
    },
    {
      "epoch": 0.7552659354779496,
      "grad_norm": 0.4576258361339569,
      "learning_rate": 8.682484632650321e-06,
      "loss": 0.358,
      "step": 16458
    },
    {
      "epoch": 0.7553118259832041,
      "grad_norm": 0.43294620513916016,
      "learning_rate": 8.682318773096772e-06,
      "loss": 0.3543,
      "step": 16459
    },
    {
      "epoch": 0.7553577164884585,
      "grad_norm": 0.5066655278205872,
      "learning_rate": 8.682152904688408e-06,
      "loss": 0.5148,
      "step": 16460
    },
    {
      "epoch": 0.755403606993713,
      "grad_norm": 0.45729944109916687,
      "learning_rate": 8.681987027425626e-06,
      "loss": 0.4365,
      "step": 16461
    },
    {
      "epoch": 0.7554494974989675,
      "grad_norm": 0.47522297501564026,
      "learning_rate": 8.68182114130883e-06,
      "loss": 0.4227,
      "step": 16462
    },
    {
      "epoch": 0.7554953880042219,
      "grad_norm": 0.5697293281555176,
      "learning_rate": 8.681655246338417e-06,
      "loss": 0.521,
      "step": 16463
    },
    {
      "epoch": 0.7555412785094764,
      "grad_norm": 0.5223415493965149,
      "learning_rate": 8.681489342514782e-06,
      "loss": 0.459,
      "step": 16464
    },
    {
      "epoch": 0.7555871690147309,
      "grad_norm": 0.4290444552898407,
      "learning_rate": 8.68132342983833e-06,
      "loss": 0.3523,
      "step": 16465
    },
    {
      "epoch": 0.7556330595199853,
      "grad_norm": 0.4455793797969818,
      "learning_rate": 8.681157508309454e-06,
      "loss": 0.3867,
      "step": 16466
    },
    {
      "epoch": 0.7556789500252398,
      "grad_norm": 0.48968544602394104,
      "learning_rate": 8.68099157792856e-06,
      "loss": 0.4686,
      "step": 16467
    },
    {
      "epoch": 0.7557248405304943,
      "grad_norm": 0.4723726212978363,
      "learning_rate": 8.68082563869604e-06,
      "loss": 0.4642,
      "step": 16468
    },
    {
      "epoch": 0.7557707310357487,
      "grad_norm": 0.45363059639930725,
      "learning_rate": 8.680659690612297e-06,
      "loss": 0.3969,
      "step": 16469
    },
    {
      "epoch": 0.7558166215410032,
      "grad_norm": 0.4921032786369324,
      "learning_rate": 8.680493733677728e-06,
      "loss": 0.4451,
      "step": 16470
    },
    {
      "epoch": 0.7558625120462577,
      "grad_norm": 0.47581279277801514,
      "learning_rate": 8.680327767892734e-06,
      "loss": 0.38,
      "step": 16471
    },
    {
      "epoch": 0.755908402551512,
      "grad_norm": 0.4593745172023773,
      "learning_rate": 8.680161793257712e-06,
      "loss": 0.3867,
      "step": 16472
    },
    {
      "epoch": 0.7559542930567665,
      "grad_norm": 0.4932304322719574,
      "learning_rate": 8.679995809773065e-06,
      "loss": 0.4229,
      "step": 16473
    },
    {
      "epoch": 0.756000183562021,
      "grad_norm": 0.4568394720554352,
      "learning_rate": 8.679829817439187e-06,
      "loss": 0.4038,
      "step": 16474
    },
    {
      "epoch": 0.7560460740672755,
      "grad_norm": 0.4508468508720398,
      "learning_rate": 8.679663816256481e-06,
      "loss": 0.3477,
      "step": 16475
    },
    {
      "epoch": 0.7560919645725299,
      "grad_norm": 0.42869001626968384,
      "learning_rate": 8.679497806225345e-06,
      "loss": 0.3324,
      "step": 16476
    },
    {
      "epoch": 0.7561378550777844,
      "grad_norm": 0.48927417397499084,
      "learning_rate": 8.679331787346178e-06,
      "loss": 0.4479,
      "step": 16477
    },
    {
      "epoch": 0.7561837455830389,
      "grad_norm": 0.45819932222366333,
      "learning_rate": 8.679165759619378e-06,
      "loss": 0.4732,
      "step": 16478
    },
    {
      "epoch": 0.7562296360882933,
      "grad_norm": 0.4924762547016144,
      "learning_rate": 8.678999723045347e-06,
      "loss": 0.4736,
      "step": 16479
    },
    {
      "epoch": 0.7562755265935478,
      "grad_norm": 0.45785123109817505,
      "learning_rate": 8.678833677624483e-06,
      "loss": 0.4352,
      "step": 16480
    },
    {
      "epoch": 0.7563214170988023,
      "grad_norm": 0.551338255405426,
      "learning_rate": 8.678667623357184e-06,
      "loss": 0.4952,
      "step": 16481
    },
    {
      "epoch": 0.7563673076040567,
      "grad_norm": 0.4435712993144989,
      "learning_rate": 8.67850156024385e-06,
      "loss": 0.4096,
      "step": 16482
    },
    {
      "epoch": 0.7564131981093112,
      "grad_norm": 0.4520529508590698,
      "learning_rate": 8.678335488284882e-06,
      "loss": 0.3379,
      "step": 16483
    },
    {
      "epoch": 0.7564590886145657,
      "grad_norm": 0.4418685734272003,
      "learning_rate": 8.67816940748068e-06,
      "loss": 0.3527,
      "step": 16484
    },
    {
      "epoch": 0.7565049791198201,
      "grad_norm": 0.45493197441101074,
      "learning_rate": 8.678003317831636e-06,
      "loss": 0.4279,
      "step": 16485
    },
    {
      "epoch": 0.7565508696250746,
      "grad_norm": 0.41695305705070496,
      "learning_rate": 8.677837219338158e-06,
      "loss": 0.3624,
      "step": 16486
    },
    {
      "epoch": 0.7565967601303291,
      "grad_norm": 0.45581063628196716,
      "learning_rate": 8.677671112000643e-06,
      "loss": 0.4423,
      "step": 16487
    },
    {
      "epoch": 0.7566426506355834,
      "grad_norm": 0.4257953464984894,
      "learning_rate": 8.677504995819489e-06,
      "loss": 0.3605,
      "step": 16488
    },
    {
      "epoch": 0.756688541140838,
      "grad_norm": 0.4474286735057831,
      "learning_rate": 8.677338870795096e-06,
      "loss": 0.3764,
      "step": 16489
    },
    {
      "epoch": 0.7567344316460924,
      "grad_norm": 0.4403073191642761,
      "learning_rate": 8.677172736927864e-06,
      "loss": 0.3728,
      "step": 16490
    },
    {
      "epoch": 0.7567803221513469,
      "grad_norm": 0.4952709972858429,
      "learning_rate": 8.677006594218192e-06,
      "loss": 0.4687,
      "step": 16491
    },
    {
      "epoch": 0.7568262126566013,
      "grad_norm": 0.4735431373119354,
      "learning_rate": 8.676840442666478e-06,
      "loss": 0.462,
      "step": 16492
    },
    {
      "epoch": 0.7568721031618558,
      "grad_norm": 0.4452124834060669,
      "learning_rate": 8.676674282273125e-06,
      "loss": 0.3573,
      "step": 16493
    },
    {
      "epoch": 0.7569179936671103,
      "grad_norm": 0.4764384925365448,
      "learning_rate": 8.67650811303853e-06,
      "loss": 0.4161,
      "step": 16494
    },
    {
      "epoch": 0.7569638841723647,
      "grad_norm": 0.47172823548316956,
      "learning_rate": 8.676341934963093e-06,
      "loss": 0.4291,
      "step": 16495
    },
    {
      "epoch": 0.7570097746776192,
      "grad_norm": 0.46439728140830994,
      "learning_rate": 8.676175748047215e-06,
      "loss": 0.4583,
      "step": 16496
    },
    {
      "epoch": 0.7570556651828737,
      "grad_norm": 0.4912375211715698,
      "learning_rate": 8.676009552291293e-06,
      "loss": 0.4791,
      "step": 16497
    },
    {
      "epoch": 0.7571015556881281,
      "grad_norm": 0.4739128053188324,
      "learning_rate": 8.67584334769573e-06,
      "loss": 0.3964,
      "step": 16498
    },
    {
      "epoch": 0.7571474461933826,
      "grad_norm": 0.44093504548072815,
      "learning_rate": 8.675677134260922e-06,
      "loss": 0.3673,
      "step": 16499
    },
    {
      "epoch": 0.7571933366986371,
      "grad_norm": 0.4649273753166199,
      "learning_rate": 8.675510911987273e-06,
      "loss": 0.4304,
      "step": 16500
    },
    {
      "epoch": 0.7572392272038915,
      "grad_norm": 0.4365490972995758,
      "learning_rate": 8.67534468087518e-06,
      "loss": 0.3799,
      "step": 16501
    },
    {
      "epoch": 0.757285117709146,
      "grad_norm": 0.4415822923183441,
      "learning_rate": 8.675178440925041e-06,
      "loss": 0.3886,
      "step": 16502
    },
    {
      "epoch": 0.7573310082144005,
      "grad_norm": 0.45539602637290955,
      "learning_rate": 8.67501219213726e-06,
      "loss": 0.3644,
      "step": 16503
    },
    {
      "epoch": 0.7573768987196549,
      "grad_norm": 0.501395583152771,
      "learning_rate": 8.674845934512233e-06,
      "loss": 0.4997,
      "step": 16504
    },
    {
      "epoch": 0.7574227892249094,
      "grad_norm": 0.43436360359191895,
      "learning_rate": 8.674679668050362e-06,
      "loss": 0.3489,
      "step": 16505
    },
    {
      "epoch": 0.7574686797301639,
      "grad_norm": 0.41186556220054626,
      "learning_rate": 8.674513392752044e-06,
      "loss": 0.323,
      "step": 16506
    },
    {
      "epoch": 0.7575145702354182,
      "grad_norm": 0.4714110195636749,
      "learning_rate": 8.674347108617682e-06,
      "loss": 0.4293,
      "step": 16507
    },
    {
      "epoch": 0.7575604607406727,
      "grad_norm": 0.47999030351638794,
      "learning_rate": 8.674180815647677e-06,
      "loss": 0.4871,
      "step": 16508
    },
    {
      "epoch": 0.7576063512459272,
      "grad_norm": 0.4613196849822998,
      "learning_rate": 8.674014513842426e-06,
      "loss": 0.4404,
      "step": 16509
    },
    {
      "epoch": 0.7576522417511817,
      "grad_norm": 0.4788079857826233,
      "learning_rate": 8.67384820320233e-06,
      "loss": 0.4202,
      "step": 16510
    },
    {
      "epoch": 0.7576981322564361,
      "grad_norm": 0.46879836916923523,
      "learning_rate": 8.673681883727786e-06,
      "loss": 0.4148,
      "step": 16511
    },
    {
      "epoch": 0.7577440227616906,
      "grad_norm": 0.46039465069770813,
      "learning_rate": 8.673515555419199e-06,
      "loss": 0.4514,
      "step": 16512
    },
    {
      "epoch": 0.7577899132669451,
      "grad_norm": 0.44031867384910583,
      "learning_rate": 8.673349218276965e-06,
      "loss": 0.4,
      "step": 16513
    },
    {
      "epoch": 0.7578358037721995,
      "grad_norm": 0.49343207478523254,
      "learning_rate": 8.673182872301486e-06,
      "loss": 0.4607,
      "step": 16514
    },
    {
      "epoch": 0.757881694277454,
      "grad_norm": 0.5610526204109192,
      "learning_rate": 8.673016517493162e-06,
      "loss": 0.3694,
      "step": 16515
    },
    {
      "epoch": 0.7579275847827085,
      "grad_norm": 0.735427975654602,
      "learning_rate": 8.672850153852392e-06,
      "loss": 0.5461,
      "step": 16516
    },
    {
      "epoch": 0.7579734752879629,
      "grad_norm": 0.45693618059158325,
      "learning_rate": 8.672683781379575e-06,
      "loss": 0.4801,
      "step": 16517
    },
    {
      "epoch": 0.7580193657932174,
      "grad_norm": 0.4393463432788849,
      "learning_rate": 8.672517400075115e-06,
      "loss": 0.3573,
      "step": 16518
    },
    {
      "epoch": 0.7580652562984719,
      "grad_norm": 0.43958809971809387,
      "learning_rate": 8.672351009939411e-06,
      "loss": 0.3648,
      "step": 16519
    },
    {
      "epoch": 0.7581111468037263,
      "grad_norm": 0.43055394291877747,
      "learning_rate": 8.672184610972858e-06,
      "loss": 0.3439,
      "step": 16520
    },
    {
      "epoch": 0.7581570373089808,
      "grad_norm": 0.4080805480480194,
      "learning_rate": 8.672018203175862e-06,
      "loss": 0.3749,
      "step": 16521
    },
    {
      "epoch": 0.7582029278142353,
      "grad_norm": 0.44646701216697693,
      "learning_rate": 8.671851786548822e-06,
      "loss": 0.3987,
      "step": 16522
    },
    {
      "epoch": 0.7582488183194896,
      "grad_norm": 0.42510348558425903,
      "learning_rate": 8.671685361092135e-06,
      "loss": 0.346,
      "step": 16523
    },
    {
      "epoch": 0.7582947088247441,
      "grad_norm": 0.4436143636703491,
      "learning_rate": 8.671518926806206e-06,
      "loss": 0.4066,
      "step": 16524
    },
    {
      "epoch": 0.7583405993299986,
      "grad_norm": 0.4393865168094635,
      "learning_rate": 8.67135248369143e-06,
      "loss": 0.3947,
      "step": 16525
    },
    {
      "epoch": 0.7583864898352531,
      "grad_norm": 0.4510743021965027,
      "learning_rate": 8.671186031748214e-06,
      "loss": 0.3864,
      "step": 16526
    },
    {
      "epoch": 0.7584323803405075,
      "grad_norm": 0.46614620089530945,
      "learning_rate": 8.67101957097695e-06,
      "loss": 0.4331,
      "step": 16527
    },
    {
      "epoch": 0.758478270845762,
      "grad_norm": 0.4818055033683777,
      "learning_rate": 8.670853101378045e-06,
      "loss": 0.458,
      "step": 16528
    },
    {
      "epoch": 0.7585241613510165,
      "grad_norm": 0.44497495889663696,
      "learning_rate": 8.670686622951895e-06,
      "loss": 0.4144,
      "step": 16529
    },
    {
      "epoch": 0.7585700518562709,
      "grad_norm": 0.48932763934135437,
      "learning_rate": 8.670520135698906e-06,
      "loss": 0.4206,
      "step": 16530
    },
    {
      "epoch": 0.7586159423615254,
      "grad_norm": 0.4336932599544525,
      "learning_rate": 8.670353639619472e-06,
      "loss": 0.3425,
      "step": 16531
    },
    {
      "epoch": 0.7586618328667799,
      "grad_norm": 0.45773589611053467,
      "learning_rate": 8.670187134713995e-06,
      "loss": 0.3691,
      "step": 16532
    },
    {
      "epoch": 0.7587077233720343,
      "grad_norm": 0.5234536528587341,
      "learning_rate": 8.670020620982876e-06,
      "loss": 0.5156,
      "step": 16533
    },
    {
      "epoch": 0.7587536138772888,
      "grad_norm": 0.43747788667678833,
      "learning_rate": 8.669854098426517e-06,
      "loss": 0.3189,
      "step": 16534
    },
    {
      "epoch": 0.7587995043825433,
      "grad_norm": 0.47076040506362915,
      "learning_rate": 8.669687567045318e-06,
      "loss": 0.3831,
      "step": 16535
    },
    {
      "epoch": 0.7588453948877977,
      "grad_norm": 0.4278438985347748,
      "learning_rate": 8.669521026839677e-06,
      "loss": 0.3404,
      "step": 16536
    },
    {
      "epoch": 0.7588912853930522,
      "grad_norm": 0.4545997977256775,
      "learning_rate": 8.669354477809998e-06,
      "loss": 0.3997,
      "step": 16537
    },
    {
      "epoch": 0.7589371758983067,
      "grad_norm": 0.44615286588668823,
      "learning_rate": 8.669187919956676e-06,
      "loss": 0.3808,
      "step": 16538
    },
    {
      "epoch": 0.7589830664035611,
      "grad_norm": 0.4327333867549896,
      "learning_rate": 8.669021353280117e-06,
      "loss": 0.3799,
      "step": 16539
    },
    {
      "epoch": 0.7590289569088156,
      "grad_norm": 0.42314204573631287,
      "learning_rate": 8.66885477778072e-06,
      "loss": 0.353,
      "step": 16540
    },
    {
      "epoch": 0.75907484741407,
      "grad_norm": 0.44395628571510315,
      "learning_rate": 8.668688193458886e-06,
      "loss": 0.3871,
      "step": 16541
    },
    {
      "epoch": 0.7591207379193244,
      "grad_norm": 0.45105060935020447,
      "learning_rate": 8.668521600315014e-06,
      "loss": 0.3851,
      "step": 16542
    },
    {
      "epoch": 0.7591666284245789,
      "grad_norm": 0.9237419962882996,
      "learning_rate": 8.668354998349504e-06,
      "loss": 0.3966,
      "step": 16543
    },
    {
      "epoch": 0.7592125189298334,
      "grad_norm": 0.44396528601646423,
      "learning_rate": 8.668188387562759e-06,
      "loss": 0.3863,
      "step": 16544
    },
    {
      "epoch": 0.7592584094350879,
      "grad_norm": 0.5016133189201355,
      "learning_rate": 8.66802176795518e-06,
      "loss": 0.4464,
      "step": 16545
    },
    {
      "epoch": 0.7593042999403423,
      "grad_norm": 0.5487783551216125,
      "learning_rate": 8.667855139527164e-06,
      "loss": 0.4224,
      "step": 16546
    },
    {
      "epoch": 0.7593501904455968,
      "grad_norm": 0.4307863116264343,
      "learning_rate": 8.667688502279115e-06,
      "loss": 0.3748,
      "step": 16547
    },
    {
      "epoch": 0.7593960809508513,
      "grad_norm": 0.4519132077693939,
      "learning_rate": 8.667521856211434e-06,
      "loss": 0.3574,
      "step": 16548
    },
    {
      "epoch": 0.7594419714561057,
      "grad_norm": 0.4867001175880432,
      "learning_rate": 8.667355201324518e-06,
      "loss": 0.3804,
      "step": 16549
    },
    {
      "epoch": 0.7594878619613602,
      "grad_norm": 0.4645751714706421,
      "learning_rate": 8.667188537618772e-06,
      "loss": 0.3625,
      "step": 16550
    },
    {
      "epoch": 0.7595337524666147,
      "grad_norm": 0.48748108744621277,
      "learning_rate": 8.667021865094594e-06,
      "loss": 0.4576,
      "step": 16551
    },
    {
      "epoch": 0.7595796429718691,
      "grad_norm": 0.4626893401145935,
      "learning_rate": 8.666855183752387e-06,
      "loss": 0.3804,
      "step": 16552
    },
    {
      "epoch": 0.7596255334771236,
      "grad_norm": 0.4826729893684387,
      "learning_rate": 8.66668849359255e-06,
      "loss": 0.5,
      "step": 16553
    },
    {
      "epoch": 0.7596714239823781,
      "grad_norm": 0.49610671401023865,
      "learning_rate": 8.666521794615484e-06,
      "loss": 0.5087,
      "step": 16554
    },
    {
      "epoch": 0.7597173144876325,
      "grad_norm": 0.4287147521972656,
      "learning_rate": 8.66635508682159e-06,
      "loss": 0.3909,
      "step": 16555
    },
    {
      "epoch": 0.759763204992887,
      "grad_norm": 0.5004259943962097,
      "learning_rate": 8.66618837021127e-06,
      "loss": 0.4298,
      "step": 16556
    },
    {
      "epoch": 0.7598090954981415,
      "grad_norm": 0.4613230228424072,
      "learning_rate": 8.666021644784922e-06,
      "loss": 0.4072,
      "step": 16557
    },
    {
      "epoch": 0.7598549860033958,
      "grad_norm": 0.43367594480514526,
      "learning_rate": 8.665854910542953e-06,
      "loss": 0.3635,
      "step": 16558
    },
    {
      "epoch": 0.7599008765086503,
      "grad_norm": 0.46060261130332947,
      "learning_rate": 8.665688167485755e-06,
      "loss": 0.4287,
      "step": 16559
    },
    {
      "epoch": 0.7599467670139048,
      "grad_norm": 0.4652125835418701,
      "learning_rate": 8.665521415613739e-06,
      "loss": 0.4072,
      "step": 16560
    },
    {
      "epoch": 0.7599926575191592,
      "grad_norm": 0.4747850000858307,
      "learning_rate": 8.665354654927297e-06,
      "loss": 0.4742,
      "step": 16561
    },
    {
      "epoch": 0.7600385480244137,
      "grad_norm": 0.43166792392730713,
      "learning_rate": 8.665187885426836e-06,
      "loss": 0.3164,
      "step": 16562
    },
    {
      "epoch": 0.7600844385296682,
      "grad_norm": 0.4606620967388153,
      "learning_rate": 8.665021107112752e-06,
      "loss": 0.4316,
      "step": 16563
    },
    {
      "epoch": 0.7601303290349227,
      "grad_norm": 0.45336636900901794,
      "learning_rate": 8.664854319985452e-06,
      "loss": 0.4017,
      "step": 16564
    },
    {
      "epoch": 0.7601762195401771,
      "grad_norm": 0.4771563410758972,
      "learning_rate": 8.664687524045333e-06,
      "loss": 0.428,
      "step": 16565
    },
    {
      "epoch": 0.7602221100454316,
      "grad_norm": 0.4496387243270874,
      "learning_rate": 8.664520719292797e-06,
      "loss": 0.3962,
      "step": 16566
    },
    {
      "epoch": 0.7602680005506861,
      "grad_norm": 0.4229981601238251,
      "learning_rate": 8.664353905728245e-06,
      "loss": 0.3603,
      "step": 16567
    },
    {
      "epoch": 0.7603138910559405,
      "grad_norm": 0.4515206217765808,
      "learning_rate": 8.664187083352081e-06,
      "loss": 0.3802,
      "step": 16568
    },
    {
      "epoch": 0.760359781561195,
      "grad_norm": 0.4783252477645874,
      "learning_rate": 8.6640202521647e-06,
      "loss": 0.5008,
      "step": 16569
    },
    {
      "epoch": 0.7604056720664495,
      "grad_norm": 0.39518433809280396,
      "learning_rate": 8.663853412166508e-06,
      "loss": 0.3421,
      "step": 16570
    },
    {
      "epoch": 0.7604515625717039,
      "grad_norm": 0.4148382246494293,
      "learning_rate": 8.663686563357905e-06,
      "loss": 0.3316,
      "step": 16571
    },
    {
      "epoch": 0.7604974530769584,
      "grad_norm": 0.46770212054252625,
      "learning_rate": 8.663519705739291e-06,
      "loss": 0.4911,
      "step": 16572
    },
    {
      "epoch": 0.7605433435822129,
      "grad_norm": 0.4535134434700012,
      "learning_rate": 8.66335283931107e-06,
      "loss": 0.3615,
      "step": 16573
    },
    {
      "epoch": 0.7605892340874673,
      "grad_norm": 0.4684820771217346,
      "learning_rate": 8.663185964073642e-06,
      "loss": 0.4815,
      "step": 16574
    },
    {
      "epoch": 0.7606351245927218,
      "grad_norm": 0.47132790088653564,
      "learning_rate": 8.663019080027405e-06,
      "loss": 0.4779,
      "step": 16575
    },
    {
      "epoch": 0.7606810150979763,
      "grad_norm": 0.42527490854263306,
      "learning_rate": 8.662852187172765e-06,
      "loss": 0.3648,
      "step": 16576
    },
    {
      "epoch": 0.7607269056032306,
      "grad_norm": 0.4820786714553833,
      "learning_rate": 8.662685285510122e-06,
      "loss": 0.5175,
      "step": 16577
    },
    {
      "epoch": 0.7607727961084851,
      "grad_norm": 0.4523543417453766,
      "learning_rate": 8.662518375039877e-06,
      "loss": 0.4059,
      "step": 16578
    },
    {
      "epoch": 0.7608186866137396,
      "grad_norm": 0.45798397064208984,
      "learning_rate": 8.66235145576243e-06,
      "loss": 0.3828,
      "step": 16579
    },
    {
      "epoch": 0.7608645771189941,
      "grad_norm": 0.4404222071170807,
      "learning_rate": 8.662184527678183e-06,
      "loss": 0.3581,
      "step": 16580
    },
    {
      "epoch": 0.7609104676242485,
      "grad_norm": 0.47726085782051086,
      "learning_rate": 8.66201759078754e-06,
      "loss": 0.3756,
      "step": 16581
    },
    {
      "epoch": 0.760956358129503,
      "grad_norm": 0.4335137605667114,
      "learning_rate": 8.6618506450909e-06,
      "loss": 0.3604,
      "step": 16582
    },
    {
      "epoch": 0.7610022486347575,
      "grad_norm": 0.4485705494880676,
      "learning_rate": 8.661683690588665e-06,
      "loss": 0.3982,
      "step": 16583
    },
    {
      "epoch": 0.7610481391400119,
      "grad_norm": 0.45291033387184143,
      "learning_rate": 8.661516727281235e-06,
      "loss": 0.3807,
      "step": 16584
    },
    {
      "epoch": 0.7610940296452664,
      "grad_norm": 0.4448527991771698,
      "learning_rate": 8.661349755169016e-06,
      "loss": 0.3799,
      "step": 16585
    },
    {
      "epoch": 0.7611399201505209,
      "grad_norm": 0.44245779514312744,
      "learning_rate": 8.661182774252402e-06,
      "loss": 0.3932,
      "step": 16586
    },
    {
      "epoch": 0.7611858106557753,
      "grad_norm": 0.4559962749481201,
      "learning_rate": 8.661015784531802e-06,
      "loss": 0.414,
      "step": 16587
    },
    {
      "epoch": 0.7612317011610298,
      "grad_norm": 0.4283487796783447,
      "learning_rate": 8.660848786007615e-06,
      "loss": 0.3268,
      "step": 16588
    },
    {
      "epoch": 0.7612775916662843,
      "grad_norm": 0.4997137188911438,
      "learning_rate": 8.660681778680239e-06,
      "loss": 0.4902,
      "step": 16589
    },
    {
      "epoch": 0.7613234821715387,
      "grad_norm": 0.5060127377510071,
      "learning_rate": 8.66051476255008e-06,
      "loss": 0.521,
      "step": 16590
    },
    {
      "epoch": 0.7613693726767932,
      "grad_norm": 0.43822968006134033,
      "learning_rate": 8.66034773761754e-06,
      "loss": 0.3736,
      "step": 16591
    },
    {
      "epoch": 0.7614152631820477,
      "grad_norm": 0.4695410132408142,
      "learning_rate": 8.660180703883016e-06,
      "loss": 0.3914,
      "step": 16592
    },
    {
      "epoch": 0.761461153687302,
      "grad_norm": 0.44119271636009216,
      "learning_rate": 8.660013661346915e-06,
      "loss": 0.3476,
      "step": 16593
    },
    {
      "epoch": 0.7615070441925565,
      "grad_norm": 0.4380929172039032,
      "learning_rate": 8.659846610009636e-06,
      "loss": 0.3806,
      "step": 16594
    },
    {
      "epoch": 0.761552934697811,
      "grad_norm": 0.4260578453540802,
      "learning_rate": 8.65967954987158e-06,
      "loss": 0.3251,
      "step": 16595
    },
    {
      "epoch": 0.7615988252030654,
      "grad_norm": 0.47210654616355896,
      "learning_rate": 8.65951248093315e-06,
      "loss": 0.4451,
      "step": 16596
    },
    {
      "epoch": 0.7616447157083199,
      "grad_norm": 0.4752641022205353,
      "learning_rate": 8.659345403194747e-06,
      "loss": 0.3797,
      "step": 16597
    },
    {
      "epoch": 0.7616906062135744,
      "grad_norm": 0.4741351902484894,
      "learning_rate": 8.659178316656776e-06,
      "loss": 0.4153,
      "step": 16598
    },
    {
      "epoch": 0.7617364967188289,
      "grad_norm": 0.46650680899620056,
      "learning_rate": 8.659011221319633e-06,
      "loss": 0.4135,
      "step": 16599
    },
    {
      "epoch": 0.7617823872240833,
      "grad_norm": 0.439491331577301,
      "learning_rate": 8.658844117183725e-06,
      "loss": 0.3495,
      "step": 16600
    },
    {
      "epoch": 0.7618282777293378,
      "grad_norm": 0.4569900929927826,
      "learning_rate": 8.65867700424945e-06,
      "loss": 0.4482,
      "step": 16601
    },
    {
      "epoch": 0.7618741682345923,
      "grad_norm": 0.4787359833717346,
      "learning_rate": 8.65850988251721e-06,
      "loss": 0.5269,
      "step": 16602
    },
    {
      "epoch": 0.7619200587398467,
      "grad_norm": 0.48187947273254395,
      "learning_rate": 8.658342751987412e-06,
      "loss": 0.4349,
      "step": 16603
    },
    {
      "epoch": 0.7619659492451012,
      "grad_norm": 0.45106056332588196,
      "learning_rate": 8.658175612660452e-06,
      "loss": 0.4194,
      "step": 16604
    },
    {
      "epoch": 0.7620118397503557,
      "grad_norm": 0.47324132919311523,
      "learning_rate": 8.658008464536735e-06,
      "loss": 0.4265,
      "step": 16605
    },
    {
      "epoch": 0.7620577302556101,
      "grad_norm": 0.49539393186569214,
      "learning_rate": 8.657841307616662e-06,
      "loss": 0.4765,
      "step": 16606
    },
    {
      "epoch": 0.7621036207608646,
      "grad_norm": 0.4333312511444092,
      "learning_rate": 8.657674141900635e-06,
      "loss": 0.3488,
      "step": 16607
    },
    {
      "epoch": 0.7621495112661191,
      "grad_norm": 0.4328710436820984,
      "learning_rate": 8.657506967389056e-06,
      "loss": 0.369,
      "step": 16608
    },
    {
      "epoch": 0.7621954017713735,
      "grad_norm": 0.44848158955574036,
      "learning_rate": 8.657339784082329e-06,
      "loss": 0.3697,
      "step": 16609
    },
    {
      "epoch": 0.762241292276628,
      "grad_norm": 0.463376522064209,
      "learning_rate": 8.65717259198085e-06,
      "loss": 0.3678,
      "step": 16610
    },
    {
      "epoch": 0.7622871827818825,
      "grad_norm": 0.4599843919277191,
      "learning_rate": 8.657005391085028e-06,
      "loss": 0.4161,
      "step": 16611
    },
    {
      "epoch": 0.7623330732871368,
      "grad_norm": 0.4671322703361511,
      "learning_rate": 8.656838181395262e-06,
      "loss": 0.4055,
      "step": 16612
    },
    {
      "epoch": 0.7623789637923913,
      "grad_norm": 0.4879436790943146,
      "learning_rate": 8.656670962911953e-06,
      "loss": 0.5166,
      "step": 16613
    },
    {
      "epoch": 0.7624248542976458,
      "grad_norm": 0.48430460691452026,
      "learning_rate": 8.656503735635507e-06,
      "loss": 0.4219,
      "step": 16614
    },
    {
      "epoch": 0.7624707448029003,
      "grad_norm": 0.45655158162117004,
      "learning_rate": 8.656336499566321e-06,
      "loss": 0.4095,
      "step": 16615
    },
    {
      "epoch": 0.7625166353081547,
      "grad_norm": 0.5079159736633301,
      "learning_rate": 8.6561692547048e-06,
      "loss": 0.4591,
      "step": 16616
    },
    {
      "epoch": 0.7625625258134092,
      "grad_norm": 0.4365522563457489,
      "learning_rate": 8.656002001051346e-06,
      "loss": 0.3645,
      "step": 16617
    },
    {
      "epoch": 0.7626084163186637,
      "grad_norm": 0.4415024220943451,
      "learning_rate": 8.655834738606363e-06,
      "loss": 0.3685,
      "step": 16618
    },
    {
      "epoch": 0.7626543068239181,
      "grad_norm": 0.4379120171070099,
      "learning_rate": 8.65566746737025e-06,
      "loss": 0.3622,
      "step": 16619
    },
    {
      "epoch": 0.7627001973291726,
      "grad_norm": 0.4383671283721924,
      "learning_rate": 8.65550018734341e-06,
      "loss": 0.3937,
      "step": 16620
    },
    {
      "epoch": 0.7627460878344271,
      "grad_norm": 0.4518575966358185,
      "learning_rate": 8.655332898526246e-06,
      "loss": 0.4059,
      "step": 16621
    },
    {
      "epoch": 0.7627919783396815,
      "grad_norm": 0.4578760862350464,
      "learning_rate": 8.65516560091916e-06,
      "loss": 0.4229,
      "step": 16622
    },
    {
      "epoch": 0.762837868844936,
      "grad_norm": 0.46214017271995544,
      "learning_rate": 8.654998294522553e-06,
      "loss": 0.4434,
      "step": 16623
    },
    {
      "epoch": 0.7628837593501905,
      "grad_norm": 0.4209538996219635,
      "learning_rate": 8.654830979336833e-06,
      "loss": 0.3403,
      "step": 16624
    },
    {
      "epoch": 0.7629296498554449,
      "grad_norm": 0.44051656126976013,
      "learning_rate": 8.654663655362394e-06,
      "loss": 0.3712,
      "step": 16625
    },
    {
      "epoch": 0.7629755403606994,
      "grad_norm": 0.4441680312156677,
      "learning_rate": 8.654496322599644e-06,
      "loss": 0.3734,
      "step": 16626
    },
    {
      "epoch": 0.7630214308659539,
      "grad_norm": 0.4254602789878845,
      "learning_rate": 8.654328981048983e-06,
      "loss": 0.3596,
      "step": 16627
    },
    {
      "epoch": 0.7630673213712083,
      "grad_norm": 0.44044029712677,
      "learning_rate": 8.654161630710813e-06,
      "loss": 0.3769,
      "step": 16628
    },
    {
      "epoch": 0.7631132118764627,
      "grad_norm": 0.46766823530197144,
      "learning_rate": 8.65399427158554e-06,
      "loss": 0.4027,
      "step": 16629
    },
    {
      "epoch": 0.7631591023817172,
      "grad_norm": 0.43795838952064514,
      "learning_rate": 8.653826903673564e-06,
      "loss": 0.3841,
      "step": 16630
    },
    {
      "epoch": 0.7632049928869716,
      "grad_norm": 0.6902680397033691,
      "learning_rate": 8.653659526975286e-06,
      "loss": 0.358,
      "step": 16631
    },
    {
      "epoch": 0.7632508833922261,
      "grad_norm": 0.4811858832836151,
      "learning_rate": 8.653492141491108e-06,
      "loss": 0.4587,
      "step": 16632
    },
    {
      "epoch": 0.7632967738974806,
      "grad_norm": 0.4566099941730499,
      "learning_rate": 8.653324747221439e-06,
      "loss": 0.4051,
      "step": 16633
    },
    {
      "epoch": 0.7633426644027351,
      "grad_norm": 0.46556583046913147,
      "learning_rate": 8.653157344166674e-06,
      "loss": 0.4038,
      "step": 16634
    },
    {
      "epoch": 0.7633885549079895,
      "grad_norm": 0.48172369599342346,
      "learning_rate": 8.652989932327218e-06,
      "loss": 0.435,
      "step": 16635
    },
    {
      "epoch": 0.763434445413244,
      "grad_norm": 0.43391183018684387,
      "learning_rate": 8.652822511703477e-06,
      "loss": 0.3315,
      "step": 16636
    },
    {
      "epoch": 0.7634803359184985,
      "grad_norm": 0.4354071319103241,
      "learning_rate": 8.652655082295849e-06,
      "loss": 0.3412,
      "step": 16637
    },
    {
      "epoch": 0.7635262264237529,
      "grad_norm": 0.4992811679840088,
      "learning_rate": 8.65248764410474e-06,
      "loss": 0.4329,
      "step": 16638
    },
    {
      "epoch": 0.7635721169290074,
      "grad_norm": 0.4186078608036041,
      "learning_rate": 8.652320197130549e-06,
      "loss": 0.3361,
      "step": 16639
    },
    {
      "epoch": 0.7636180074342619,
      "grad_norm": 0.43802008032798767,
      "learning_rate": 8.65215274137368e-06,
      "loss": 0.341,
      "step": 16640
    },
    {
      "epoch": 0.7636638979395163,
      "grad_norm": 0.4748854339122772,
      "learning_rate": 8.651985276834538e-06,
      "loss": 0.4741,
      "step": 16641
    },
    {
      "epoch": 0.7637097884447708,
      "grad_norm": 0.48368120193481445,
      "learning_rate": 8.651817803513523e-06,
      "loss": 0.5021,
      "step": 16642
    },
    {
      "epoch": 0.7637556789500253,
      "grad_norm": 0.45552507042884827,
      "learning_rate": 8.65165032141104e-06,
      "loss": 0.4288,
      "step": 16643
    },
    {
      "epoch": 0.7638015694552797,
      "grad_norm": 0.4552462697029114,
      "learning_rate": 8.65148283052749e-06,
      "loss": 0.4514,
      "step": 16644
    },
    {
      "epoch": 0.7638474599605342,
      "grad_norm": 0.4808780550956726,
      "learning_rate": 8.651315330863276e-06,
      "loss": 0.4657,
      "step": 16645
    },
    {
      "epoch": 0.7638933504657887,
      "grad_norm": 0.4388098120689392,
      "learning_rate": 8.651147822418801e-06,
      "loss": 0.3926,
      "step": 16646
    },
    {
      "epoch": 0.763939240971043,
      "grad_norm": 0.4399193823337555,
      "learning_rate": 8.650980305194468e-06,
      "loss": 0.345,
      "step": 16647
    },
    {
      "epoch": 0.7639851314762975,
      "grad_norm": 0.4355769753456116,
      "learning_rate": 8.650812779190678e-06,
      "loss": 0.3043,
      "step": 16648
    },
    {
      "epoch": 0.764031021981552,
      "grad_norm": 0.5149824619293213,
      "learning_rate": 8.650645244407838e-06,
      "loss": 0.4716,
      "step": 16649
    },
    {
      "epoch": 0.7640769124868064,
      "grad_norm": 0.4733157753944397,
      "learning_rate": 8.650477700846346e-06,
      "loss": 0.4006,
      "step": 16650
    },
    {
      "epoch": 0.7641228029920609,
      "grad_norm": 0.4078599810600281,
      "learning_rate": 8.650310148506609e-06,
      "loss": 0.2985,
      "step": 16651
    },
    {
      "epoch": 0.7641686934973154,
      "grad_norm": 0.477463036775589,
      "learning_rate": 8.65014258738903e-06,
      "loss": 0.4361,
      "step": 16652
    },
    {
      "epoch": 0.7642145840025699,
      "grad_norm": 0.44911378622055054,
      "learning_rate": 8.649975017494005e-06,
      "loss": 0.3763,
      "step": 16653
    },
    {
      "epoch": 0.7642604745078243,
      "grad_norm": 0.46447426080703735,
      "learning_rate": 8.649807438821946e-06,
      "loss": 0.4093,
      "step": 16654
    },
    {
      "epoch": 0.7643063650130788,
      "grad_norm": 0.4102897346019745,
      "learning_rate": 8.64963985137325e-06,
      "loss": 0.35,
      "step": 16655
    },
    {
      "epoch": 0.7643522555183333,
      "grad_norm": 0.4692215025424957,
      "learning_rate": 8.649472255148324e-06,
      "loss": 0.4715,
      "step": 16656
    },
    {
      "epoch": 0.7643981460235877,
      "grad_norm": 0.4415777325630188,
      "learning_rate": 8.649304650147567e-06,
      "loss": 0.401,
      "step": 16657
    },
    {
      "epoch": 0.7644440365288422,
      "grad_norm": 0.4136008322238922,
      "learning_rate": 8.649137036371386e-06,
      "loss": 0.355,
      "step": 16658
    },
    {
      "epoch": 0.7644899270340967,
      "grad_norm": 0.44934991002082825,
      "learning_rate": 8.64896941382018e-06,
      "loss": 0.3894,
      "step": 16659
    },
    {
      "epoch": 0.7645358175393511,
      "grad_norm": 0.45560553669929504,
      "learning_rate": 8.648801782494356e-06,
      "loss": 0.3785,
      "step": 16660
    },
    {
      "epoch": 0.7645817080446056,
      "grad_norm": 0.44760313630104065,
      "learning_rate": 8.648634142394314e-06,
      "loss": 0.3738,
      "step": 16661
    },
    {
      "epoch": 0.7646275985498601,
      "grad_norm": 0.4740528166294098,
      "learning_rate": 8.64846649352046e-06,
      "loss": 0.4697,
      "step": 16662
    },
    {
      "epoch": 0.7646734890551145,
      "grad_norm": 0.4445270597934723,
      "learning_rate": 8.648298835873194e-06,
      "loss": 0.4285,
      "step": 16663
    },
    {
      "epoch": 0.764719379560369,
      "grad_norm": 0.4388747811317444,
      "learning_rate": 8.648131169452923e-06,
      "loss": 0.3598,
      "step": 16664
    },
    {
      "epoch": 0.7647652700656234,
      "grad_norm": 0.4456874430179596,
      "learning_rate": 8.647963494260045e-06,
      "loss": 0.4431,
      "step": 16665
    },
    {
      "epoch": 0.7648111605708778,
      "grad_norm": 0.5008742809295654,
      "learning_rate": 8.647795810294968e-06,
      "loss": 0.5462,
      "step": 16666
    },
    {
      "epoch": 0.7648570510761323,
      "grad_norm": 0.4694848954677582,
      "learning_rate": 8.647628117558093e-06,
      "loss": 0.4428,
      "step": 16667
    },
    {
      "epoch": 0.7649029415813868,
      "grad_norm": 0.4653250277042389,
      "learning_rate": 8.647460416049823e-06,
      "loss": 0.4335,
      "step": 16668
    },
    {
      "epoch": 0.7649488320866413,
      "grad_norm": 0.4807059168815613,
      "learning_rate": 8.647292705770563e-06,
      "loss": 0.5,
      "step": 16669
    },
    {
      "epoch": 0.7649947225918957,
      "grad_norm": 0.48507270216941833,
      "learning_rate": 8.647124986720714e-06,
      "loss": 0.4982,
      "step": 16670
    },
    {
      "epoch": 0.7650406130971502,
      "grad_norm": 0.4438900947570801,
      "learning_rate": 8.646957258900682e-06,
      "loss": 0.4102,
      "step": 16671
    },
    {
      "epoch": 0.7650865036024047,
      "grad_norm": 0.41129499673843384,
      "learning_rate": 8.646789522310866e-06,
      "loss": 0.2967,
      "step": 16672
    },
    {
      "epoch": 0.7651323941076591,
      "grad_norm": 0.44413602352142334,
      "learning_rate": 8.646621776951674e-06,
      "loss": 0.3596,
      "step": 16673
    },
    {
      "epoch": 0.7651782846129136,
      "grad_norm": 0.4327761232852936,
      "learning_rate": 8.646454022823508e-06,
      "loss": 0.3422,
      "step": 16674
    },
    {
      "epoch": 0.7652241751181681,
      "grad_norm": 0.4554285705089569,
      "learning_rate": 8.64628625992677e-06,
      "loss": 0.4242,
      "step": 16675
    },
    {
      "epoch": 0.7652700656234225,
      "grad_norm": 0.47412264347076416,
      "learning_rate": 8.646118488261866e-06,
      "loss": 0.4382,
      "step": 16676
    },
    {
      "epoch": 0.765315956128677,
      "grad_norm": 0.4398435056209564,
      "learning_rate": 8.645950707829194e-06,
      "loss": 0.3474,
      "step": 16677
    },
    {
      "epoch": 0.7653618466339315,
      "grad_norm": 0.4326326847076416,
      "learning_rate": 8.645782918629165e-06,
      "loss": 0.3432,
      "step": 16678
    },
    {
      "epoch": 0.7654077371391859,
      "grad_norm": 0.4290851950645447,
      "learning_rate": 8.645615120662176e-06,
      "loss": 0.3891,
      "step": 16679
    },
    {
      "epoch": 0.7654536276444404,
      "grad_norm": 0.4184976816177368,
      "learning_rate": 8.645447313928636e-06,
      "loss": 0.3359,
      "step": 16680
    },
    {
      "epoch": 0.7654995181496949,
      "grad_norm": 0.4594894349575043,
      "learning_rate": 8.645279498428944e-06,
      "loss": 0.3823,
      "step": 16681
    },
    {
      "epoch": 0.7655454086549492,
      "grad_norm": 0.4451071321964264,
      "learning_rate": 8.645111674163506e-06,
      "loss": 0.3567,
      "step": 16682
    },
    {
      "epoch": 0.7655912991602037,
      "grad_norm": 0.462082177400589,
      "learning_rate": 8.644943841132724e-06,
      "loss": 0.413,
      "step": 16683
    },
    {
      "epoch": 0.7656371896654582,
      "grad_norm": 0.4698992073535919,
      "learning_rate": 8.644775999337004e-06,
      "loss": 0.4484,
      "step": 16684
    },
    {
      "epoch": 0.7656830801707126,
      "grad_norm": 0.4230175018310547,
      "learning_rate": 8.644608148776745e-06,
      "loss": 0.3427,
      "step": 16685
    },
    {
      "epoch": 0.7657289706759671,
      "grad_norm": 0.44336095452308655,
      "learning_rate": 8.644440289452357e-06,
      "loss": 0.3648,
      "step": 16686
    },
    {
      "epoch": 0.7657748611812216,
      "grad_norm": 0.4543813169002533,
      "learning_rate": 8.644272421364239e-06,
      "loss": 0.3752,
      "step": 16687
    },
    {
      "epoch": 0.7658207516864761,
      "grad_norm": 0.4402708411216736,
      "learning_rate": 8.644104544512796e-06,
      "loss": 0.3667,
      "step": 16688
    },
    {
      "epoch": 0.7658666421917305,
      "grad_norm": 0.410935640335083,
      "learning_rate": 8.643936658898431e-06,
      "loss": 0.2947,
      "step": 16689
    },
    {
      "epoch": 0.765912532696985,
      "grad_norm": 0.585806667804718,
      "learning_rate": 8.64376876452155e-06,
      "loss": 0.3698,
      "step": 16690
    },
    {
      "epoch": 0.7659584232022395,
      "grad_norm": 0.4716147482395172,
      "learning_rate": 8.643600861382554e-06,
      "loss": 0.375,
      "step": 16691
    },
    {
      "epoch": 0.7660043137074939,
      "grad_norm": 0.46725332736968994,
      "learning_rate": 8.643432949481847e-06,
      "loss": 0.3955,
      "step": 16692
    },
    {
      "epoch": 0.7660502042127484,
      "grad_norm": 0.4907747209072113,
      "learning_rate": 8.643265028819837e-06,
      "loss": 0.4524,
      "step": 16693
    },
    {
      "epoch": 0.7660960947180029,
      "grad_norm": 0.45562073588371277,
      "learning_rate": 8.643097099396921e-06,
      "loss": 0.4352,
      "step": 16694
    },
    {
      "epoch": 0.7661419852232573,
      "grad_norm": 0.4529406726360321,
      "learning_rate": 8.642929161213507e-06,
      "loss": 0.4104,
      "step": 16695
    },
    {
      "epoch": 0.7661878757285118,
      "grad_norm": 0.45797809958457947,
      "learning_rate": 8.642761214269999e-06,
      "loss": 0.3875,
      "step": 16696
    },
    {
      "epoch": 0.7662337662337663,
      "grad_norm": 0.4381345808506012,
      "learning_rate": 8.642593258566799e-06,
      "loss": 0.3322,
      "step": 16697
    },
    {
      "epoch": 0.7662796567390207,
      "grad_norm": 0.47425922751426697,
      "learning_rate": 8.642425294104312e-06,
      "loss": 0.4524,
      "step": 16698
    },
    {
      "epoch": 0.7663255472442752,
      "grad_norm": 0.4533551335334778,
      "learning_rate": 8.642257320882942e-06,
      "loss": 0.4062,
      "step": 16699
    },
    {
      "epoch": 0.7663714377495296,
      "grad_norm": 0.43710264563560486,
      "learning_rate": 8.642089338903093e-06,
      "loss": 0.3433,
      "step": 16700
    },
    {
      "epoch": 0.766417328254784,
      "grad_norm": 0.4641318917274475,
      "learning_rate": 8.64192134816517e-06,
      "loss": 0.3851,
      "step": 16701
    },
    {
      "epoch": 0.7664632187600385,
      "grad_norm": 0.5152125954627991,
      "learning_rate": 8.641753348669573e-06,
      "loss": 0.5483,
      "step": 16702
    },
    {
      "epoch": 0.766509109265293,
      "grad_norm": 0.4314947724342346,
      "learning_rate": 8.64158534041671e-06,
      "loss": 0.3106,
      "step": 16703
    },
    {
      "epoch": 0.7665549997705474,
      "grad_norm": 0.45960569381713867,
      "learning_rate": 8.641417323406983e-06,
      "loss": 0.4333,
      "step": 16704
    },
    {
      "epoch": 0.7666008902758019,
      "grad_norm": 0.5077401399612427,
      "learning_rate": 8.641249297640796e-06,
      "loss": 0.4126,
      "step": 16705
    },
    {
      "epoch": 0.7666467807810564,
      "grad_norm": 0.5035547614097595,
      "learning_rate": 8.641081263118556e-06,
      "loss": 0.4219,
      "step": 16706
    },
    {
      "epoch": 0.7666926712863109,
      "grad_norm": 0.6299911141395569,
      "learning_rate": 8.640913219840664e-06,
      "loss": 0.5065,
      "step": 16707
    },
    {
      "epoch": 0.7667385617915653,
      "grad_norm": 0.4525778293609619,
      "learning_rate": 8.640745167807523e-06,
      "loss": 0.4516,
      "step": 16708
    },
    {
      "epoch": 0.7667844522968198,
      "grad_norm": 0.4584752023220062,
      "learning_rate": 8.640577107019538e-06,
      "loss": 0.3727,
      "step": 16709
    },
    {
      "epoch": 0.7668303428020743,
      "grad_norm": 0.4778797924518585,
      "learning_rate": 8.640409037477118e-06,
      "loss": 0.4252,
      "step": 16710
    },
    {
      "epoch": 0.7668762333073287,
      "grad_norm": 0.5247454643249512,
      "learning_rate": 8.64024095918066e-06,
      "loss": 0.5022,
      "step": 16711
    },
    {
      "epoch": 0.7669221238125832,
      "grad_norm": 0.45136985182762146,
      "learning_rate": 8.640072872130573e-06,
      "loss": 0.4008,
      "step": 16712
    },
    {
      "epoch": 0.7669680143178377,
      "grad_norm": 0.4450985789299011,
      "learning_rate": 8.63990477632726e-06,
      "loss": 0.3848,
      "step": 16713
    },
    {
      "epoch": 0.7670139048230921,
      "grad_norm": 0.44062259793281555,
      "learning_rate": 8.639736671771122e-06,
      "loss": 0.3586,
      "step": 16714
    },
    {
      "epoch": 0.7670597953283466,
      "grad_norm": 0.44705069065093994,
      "learning_rate": 8.639568558462568e-06,
      "loss": 0.3808,
      "step": 16715
    },
    {
      "epoch": 0.7671056858336011,
      "grad_norm": 0.473104864358902,
      "learning_rate": 8.639400436402002e-06,
      "loss": 0.4683,
      "step": 16716
    },
    {
      "epoch": 0.7671515763388554,
      "grad_norm": 0.45324209332466125,
      "learning_rate": 8.639232305589824e-06,
      "loss": 0.3779,
      "step": 16717
    },
    {
      "epoch": 0.7671974668441099,
      "grad_norm": 0.46641144156455994,
      "learning_rate": 8.63906416602644e-06,
      "loss": 0.3826,
      "step": 16718
    },
    {
      "epoch": 0.7672433573493644,
      "grad_norm": 0.46970435976982117,
      "learning_rate": 8.638896017712257e-06,
      "loss": 0.4369,
      "step": 16719
    },
    {
      "epoch": 0.7672892478546188,
      "grad_norm": 0.4547556936740875,
      "learning_rate": 8.638727860647678e-06,
      "loss": 0.3491,
      "step": 16720
    },
    {
      "epoch": 0.7673351383598733,
      "grad_norm": 0.4516390860080719,
      "learning_rate": 8.638559694833105e-06,
      "loss": 0.3725,
      "step": 16721
    },
    {
      "epoch": 0.7673810288651278,
      "grad_norm": 0.46866875886917114,
      "learning_rate": 8.638391520268946e-06,
      "loss": 0.3657,
      "step": 16722
    },
    {
      "epoch": 0.7674269193703823,
      "grad_norm": 0.43218621611595154,
      "learning_rate": 8.638223336955604e-06,
      "loss": 0.3285,
      "step": 16723
    },
    {
      "epoch": 0.7674728098756367,
      "grad_norm": 0.44133177399635315,
      "learning_rate": 8.638055144893483e-06,
      "loss": 0.3942,
      "step": 16724
    },
    {
      "epoch": 0.7675187003808912,
      "grad_norm": 0.4587036073207855,
      "learning_rate": 8.637886944082986e-06,
      "loss": 0.4121,
      "step": 16725
    },
    {
      "epoch": 0.7675645908861457,
      "grad_norm": 0.4630714952945709,
      "learning_rate": 8.637718734524522e-06,
      "loss": 0.327,
      "step": 16726
    },
    {
      "epoch": 0.7676104813914001,
      "grad_norm": 0.45840924978256226,
      "learning_rate": 8.63755051621849e-06,
      "loss": 0.4179,
      "step": 16727
    },
    {
      "epoch": 0.7676563718966546,
      "grad_norm": 0.445102721452713,
      "learning_rate": 8.637382289165299e-06,
      "loss": 0.4043,
      "step": 16728
    },
    {
      "epoch": 0.7677022624019091,
      "grad_norm": 0.4620171785354614,
      "learning_rate": 8.63721405336535e-06,
      "loss": 0.4234,
      "step": 16729
    },
    {
      "epoch": 0.7677481529071635,
      "grad_norm": 0.46046048402786255,
      "learning_rate": 8.63704580881905e-06,
      "loss": 0.3923,
      "step": 16730
    },
    {
      "epoch": 0.767794043412418,
      "grad_norm": 0.47586095333099365,
      "learning_rate": 8.636877555526801e-06,
      "loss": 0.4441,
      "step": 16731
    },
    {
      "epoch": 0.7678399339176725,
      "grad_norm": 0.434968501329422,
      "learning_rate": 8.636709293489011e-06,
      "loss": 0.3277,
      "step": 16732
    },
    {
      "epoch": 0.7678858244229269,
      "grad_norm": 0.4700027108192444,
      "learning_rate": 8.636541022706082e-06,
      "loss": 0.4437,
      "step": 16733
    },
    {
      "epoch": 0.7679317149281814,
      "grad_norm": 0.44369786977767944,
      "learning_rate": 8.63637274317842e-06,
      "loss": 0.3821,
      "step": 16734
    },
    {
      "epoch": 0.7679776054334359,
      "grad_norm": 0.4425202012062073,
      "learning_rate": 8.63620445490643e-06,
      "loss": 0.4142,
      "step": 16735
    },
    {
      "epoch": 0.7680234959386902,
      "grad_norm": 0.4245244264602661,
      "learning_rate": 8.636036157890514e-06,
      "loss": 0.3616,
      "step": 16736
    },
    {
      "epoch": 0.7680693864439447,
      "grad_norm": 0.43719521164894104,
      "learning_rate": 8.635867852131079e-06,
      "loss": 0.364,
      "step": 16737
    },
    {
      "epoch": 0.7681152769491992,
      "grad_norm": 0.44025149941444397,
      "learning_rate": 8.63569953762853e-06,
      "loss": 0.3419,
      "step": 16738
    },
    {
      "epoch": 0.7681611674544536,
      "grad_norm": 0.42897433042526245,
      "learning_rate": 8.635531214383272e-06,
      "loss": 0.3867,
      "step": 16739
    },
    {
      "epoch": 0.7682070579597081,
      "grad_norm": 0.4870644807815552,
      "learning_rate": 8.635362882395709e-06,
      "loss": 0.4401,
      "step": 16740
    },
    {
      "epoch": 0.7682529484649626,
      "grad_norm": 0.4196176826953888,
      "learning_rate": 8.635194541666243e-06,
      "loss": 0.3293,
      "step": 16741
    },
    {
      "epoch": 0.7682988389702171,
      "grad_norm": 0.45129823684692383,
      "learning_rate": 8.635026192195283e-06,
      "loss": 0.4493,
      "step": 16742
    },
    {
      "epoch": 0.7683447294754715,
      "grad_norm": 0.44881734251976013,
      "learning_rate": 8.634857833983232e-06,
      "loss": 0.4094,
      "step": 16743
    },
    {
      "epoch": 0.768390619980726,
      "grad_norm": 0.47999757528305054,
      "learning_rate": 8.634689467030496e-06,
      "loss": 0.4797,
      "step": 16744
    },
    {
      "epoch": 0.7684365104859805,
      "grad_norm": 0.4394710659980774,
      "learning_rate": 8.634521091337478e-06,
      "loss": 0.3692,
      "step": 16745
    },
    {
      "epoch": 0.7684824009912349,
      "grad_norm": 0.45004144310951233,
      "learning_rate": 8.634352706904584e-06,
      "loss": 0.3982,
      "step": 16746
    },
    {
      "epoch": 0.7685282914964894,
      "grad_norm": 0.48920729756355286,
      "learning_rate": 8.634184313732219e-06,
      "loss": 0.4488,
      "step": 16747
    },
    {
      "epoch": 0.7685741820017439,
      "grad_norm": 0.47797054052352905,
      "learning_rate": 8.634015911820788e-06,
      "loss": 0.3946,
      "step": 16748
    },
    {
      "epoch": 0.7686200725069983,
      "grad_norm": 0.4358716309070587,
      "learning_rate": 8.633847501170696e-06,
      "loss": 0.3434,
      "step": 16749
    },
    {
      "epoch": 0.7686659630122528,
      "grad_norm": 0.5199787020683289,
      "learning_rate": 8.633679081782347e-06,
      "loss": 0.5065,
      "step": 16750
    },
    {
      "epoch": 0.7687118535175073,
      "grad_norm": 0.47950631380081177,
      "learning_rate": 8.633510653656146e-06,
      "loss": 0.4067,
      "step": 16751
    },
    {
      "epoch": 0.7687577440227616,
      "grad_norm": 0.4889020621776581,
      "learning_rate": 8.633342216792498e-06,
      "loss": 0.4974,
      "step": 16752
    },
    {
      "epoch": 0.7688036345280161,
      "grad_norm": 0.5132023692131042,
      "learning_rate": 8.633173771191811e-06,
      "loss": 0.5192,
      "step": 16753
    },
    {
      "epoch": 0.7688495250332706,
      "grad_norm": 0.45993226766586304,
      "learning_rate": 8.633005316854486e-06,
      "loss": 0.4084,
      "step": 16754
    },
    {
      "epoch": 0.768895415538525,
      "grad_norm": 0.4833483397960663,
      "learning_rate": 8.63283685378093e-06,
      "loss": 0.4479,
      "step": 16755
    },
    {
      "epoch": 0.7689413060437795,
      "grad_norm": 0.4572613537311554,
      "learning_rate": 8.63266838197155e-06,
      "loss": 0.3786,
      "step": 16756
    },
    {
      "epoch": 0.768987196549034,
      "grad_norm": 0.4406229853630066,
      "learning_rate": 8.632499901426747e-06,
      "loss": 0.3842,
      "step": 16757
    },
    {
      "epoch": 0.7690330870542885,
      "grad_norm": 0.3853614628314972,
      "learning_rate": 8.632331412146928e-06,
      "loss": 0.244,
      "step": 16758
    },
    {
      "epoch": 0.7690789775595429,
      "grad_norm": 0.4902091324329376,
      "learning_rate": 8.6321629141325e-06,
      "loss": 0.449,
      "step": 16759
    },
    {
      "epoch": 0.7691248680647974,
      "grad_norm": 0.45169204473495483,
      "learning_rate": 8.631994407383865e-06,
      "loss": 0.4294,
      "step": 16760
    },
    {
      "epoch": 0.7691707585700519,
      "grad_norm": 0.49083441495895386,
      "learning_rate": 8.63182589190143e-06,
      "loss": 0.4061,
      "step": 16761
    },
    {
      "epoch": 0.7692166490753063,
      "grad_norm": 0.4819958209991455,
      "learning_rate": 8.6316573676856e-06,
      "loss": 0.3858,
      "step": 16762
    },
    {
      "epoch": 0.7692625395805608,
      "grad_norm": 0.4892674684524536,
      "learning_rate": 8.63148883473678e-06,
      "loss": 0.5148,
      "step": 16763
    },
    {
      "epoch": 0.7693084300858153,
      "grad_norm": 0.46496593952178955,
      "learning_rate": 8.631320293055376e-06,
      "loss": 0.3916,
      "step": 16764
    },
    {
      "epoch": 0.7693543205910697,
      "grad_norm": 0.45585861802101135,
      "learning_rate": 8.631151742641794e-06,
      "loss": 0.404,
      "step": 16765
    },
    {
      "epoch": 0.7694002110963242,
      "grad_norm": 0.43657055497169495,
      "learning_rate": 8.630983183496436e-06,
      "loss": 0.4274,
      "step": 16766
    },
    {
      "epoch": 0.7694461016015787,
      "grad_norm": 0.5027413964271545,
      "learning_rate": 8.630814615619711e-06,
      "loss": 0.5115,
      "step": 16767
    },
    {
      "epoch": 0.7694919921068331,
      "grad_norm": 0.5198984742164612,
      "learning_rate": 8.630646039012021e-06,
      "loss": 0.5225,
      "step": 16768
    },
    {
      "epoch": 0.7695378826120876,
      "grad_norm": 0.45199620723724365,
      "learning_rate": 8.630477453673774e-06,
      "loss": 0.3944,
      "step": 16769
    },
    {
      "epoch": 0.769583773117342,
      "grad_norm": 0.4292090833187103,
      "learning_rate": 8.630308859605374e-06,
      "loss": 0.3655,
      "step": 16770
    },
    {
      "epoch": 0.7696296636225964,
      "grad_norm": 0.4415838420391083,
      "learning_rate": 8.630140256807228e-06,
      "loss": 0.3388,
      "step": 16771
    },
    {
      "epoch": 0.7696755541278509,
      "grad_norm": 0.4891478717327118,
      "learning_rate": 8.62997164527974e-06,
      "loss": 0.4582,
      "step": 16772
    },
    {
      "epoch": 0.7697214446331054,
      "grad_norm": 0.4409283399581909,
      "learning_rate": 8.629803025023315e-06,
      "loss": 0.3891,
      "step": 16773
    },
    {
      "epoch": 0.7697673351383598,
      "grad_norm": 0.458002507686615,
      "learning_rate": 8.629634396038361e-06,
      "loss": 0.3758,
      "step": 16774
    },
    {
      "epoch": 0.7698132256436143,
      "grad_norm": 0.46311643719673157,
      "learning_rate": 8.629465758325282e-06,
      "loss": 0.4111,
      "step": 16775
    },
    {
      "epoch": 0.7698591161488688,
      "grad_norm": 0.48786279559135437,
      "learning_rate": 8.629297111884481e-06,
      "loss": 0.4008,
      "step": 16776
    },
    {
      "epoch": 0.7699050066541233,
      "grad_norm": 0.4939804673194885,
      "learning_rate": 8.629128456716369e-06,
      "loss": 0.4607,
      "step": 16777
    },
    {
      "epoch": 0.7699508971593777,
      "grad_norm": 0.4647766053676605,
      "learning_rate": 8.628959792821346e-06,
      "loss": 0.4244,
      "step": 16778
    },
    {
      "epoch": 0.7699967876646322,
      "grad_norm": 0.4827609062194824,
      "learning_rate": 8.62879112019982e-06,
      "loss": 0.4225,
      "step": 16779
    },
    {
      "epoch": 0.7700426781698867,
      "grad_norm": 0.45400139689445496,
      "learning_rate": 8.628622438852197e-06,
      "loss": 0.3897,
      "step": 16780
    },
    {
      "epoch": 0.7700885686751411,
      "grad_norm": 0.44279295206069946,
      "learning_rate": 8.628453748778882e-06,
      "loss": 0.4062,
      "step": 16781
    },
    {
      "epoch": 0.7701344591803956,
      "grad_norm": 0.4537462592124939,
      "learning_rate": 8.628285049980283e-06,
      "loss": 0.397,
      "step": 16782
    },
    {
      "epoch": 0.7701803496856501,
      "grad_norm": 0.4813184142112732,
      "learning_rate": 8.628116342456801e-06,
      "loss": 0.4178,
      "step": 16783
    },
    {
      "epoch": 0.7702262401909045,
      "grad_norm": 0.517454206943512,
      "learning_rate": 8.627947626208845e-06,
      "loss": 0.5322,
      "step": 16784
    },
    {
      "epoch": 0.770272130696159,
      "grad_norm": 0.4143744111061096,
      "learning_rate": 8.62777890123682e-06,
      "loss": 0.3441,
      "step": 16785
    },
    {
      "epoch": 0.7703180212014135,
      "grad_norm": 0.4471566677093506,
      "learning_rate": 8.627610167541134e-06,
      "loss": 0.3458,
      "step": 16786
    },
    {
      "epoch": 0.7703639117066678,
      "grad_norm": 0.479196697473526,
      "learning_rate": 8.627441425122187e-06,
      "loss": 0.4291,
      "step": 16787
    },
    {
      "epoch": 0.7704098022119223,
      "grad_norm": 0.436245322227478,
      "learning_rate": 8.62727267398039e-06,
      "loss": 0.3422,
      "step": 16788
    },
    {
      "epoch": 0.7704556927171768,
      "grad_norm": 0.517333984375,
      "learning_rate": 8.627103914116146e-06,
      "loss": 0.4887,
      "step": 16789
    },
    {
      "epoch": 0.7705015832224312,
      "grad_norm": 0.46458911895751953,
      "learning_rate": 8.626935145529863e-06,
      "loss": 0.4494,
      "step": 16790
    },
    {
      "epoch": 0.7705474737276857,
      "grad_norm": 0.5149855017662048,
      "learning_rate": 8.626766368221946e-06,
      "loss": 0.4964,
      "step": 16791
    },
    {
      "epoch": 0.7705933642329402,
      "grad_norm": 0.42306336760520935,
      "learning_rate": 8.626597582192799e-06,
      "loss": 0.2988,
      "step": 16792
    },
    {
      "epoch": 0.7706392547381946,
      "grad_norm": 0.4497124254703522,
      "learning_rate": 8.62642878744283e-06,
      "loss": 0.4026,
      "step": 16793
    },
    {
      "epoch": 0.7706851452434491,
      "grad_norm": 0.44155851006507874,
      "learning_rate": 8.626259983972446e-06,
      "loss": 0.3805,
      "step": 16794
    },
    {
      "epoch": 0.7707310357487036,
      "grad_norm": 0.4343738853931427,
      "learning_rate": 8.626091171782048e-06,
      "loss": 0.3773,
      "step": 16795
    },
    {
      "epoch": 0.7707769262539581,
      "grad_norm": 0.4640238881111145,
      "learning_rate": 8.625922350872047e-06,
      "loss": 0.4028,
      "step": 16796
    },
    {
      "epoch": 0.7708228167592125,
      "grad_norm": 0.4515562653541565,
      "learning_rate": 8.625753521242848e-06,
      "loss": 0.3748,
      "step": 16797
    },
    {
      "epoch": 0.770868707264467,
      "grad_norm": 0.4636695086956024,
      "learning_rate": 8.625584682894854e-06,
      "loss": 0.4203,
      "step": 16798
    },
    {
      "epoch": 0.7709145977697215,
      "grad_norm": 0.4181848168373108,
      "learning_rate": 8.625415835828475e-06,
      "loss": 0.3057,
      "step": 16799
    },
    {
      "epoch": 0.7709604882749759,
      "grad_norm": 0.426975280046463,
      "learning_rate": 8.625246980044112e-06,
      "loss": 0.3615,
      "step": 16800
    },
    {
      "epoch": 0.7710063787802304,
      "grad_norm": 0.5417707562446594,
      "learning_rate": 8.625078115542178e-06,
      "loss": 0.4595,
      "step": 16801
    },
    {
      "epoch": 0.7710522692854849,
      "grad_norm": 0.4715006649494171,
      "learning_rate": 8.624909242323074e-06,
      "loss": 0.4089,
      "step": 16802
    },
    {
      "epoch": 0.7710981597907393,
      "grad_norm": 0.5248382687568665,
      "learning_rate": 8.624740360387206e-06,
      "loss": 0.524,
      "step": 16803
    },
    {
      "epoch": 0.7711440502959938,
      "grad_norm": 0.4586992561817169,
      "learning_rate": 8.624571469734983e-06,
      "loss": 0.3935,
      "step": 16804
    },
    {
      "epoch": 0.7711899408012483,
      "grad_norm": 0.4734436869621277,
      "learning_rate": 8.624402570366808e-06,
      "loss": 0.4287,
      "step": 16805
    },
    {
      "epoch": 0.7712358313065026,
      "grad_norm": 0.4462924897670746,
      "learning_rate": 8.624233662283091e-06,
      "loss": 0.3858,
      "step": 16806
    },
    {
      "epoch": 0.7712817218117571,
      "grad_norm": 0.4268195331096649,
      "learning_rate": 8.624064745484235e-06,
      "loss": 0.3541,
      "step": 16807
    },
    {
      "epoch": 0.7713276123170116,
      "grad_norm": 0.4548666179180145,
      "learning_rate": 8.623895819970646e-06,
      "loss": 0.3857,
      "step": 16808
    },
    {
      "epoch": 0.771373502822266,
      "grad_norm": 0.4665895402431488,
      "learning_rate": 8.623726885742732e-06,
      "loss": 0.381,
      "step": 16809
    },
    {
      "epoch": 0.7714193933275205,
      "grad_norm": 0.440584272146225,
      "learning_rate": 8.623557942800898e-06,
      "loss": 0.3502,
      "step": 16810
    },
    {
      "epoch": 0.771465283832775,
      "grad_norm": 0.4469936192035675,
      "learning_rate": 8.623388991145552e-06,
      "loss": 0.3486,
      "step": 16811
    },
    {
      "epoch": 0.7715111743380295,
      "grad_norm": 0.4688628911972046,
      "learning_rate": 8.623220030777097e-06,
      "loss": 0.4345,
      "step": 16812
    },
    {
      "epoch": 0.7715570648432839,
      "grad_norm": 0.4318545460700989,
      "learning_rate": 8.623051061695943e-06,
      "loss": 0.3582,
      "step": 16813
    },
    {
      "epoch": 0.7716029553485384,
      "grad_norm": 0.44012364745140076,
      "learning_rate": 8.622882083902495e-06,
      "loss": 0.401,
      "step": 16814
    },
    {
      "epoch": 0.7716488458537929,
      "grad_norm": 0.44759029150009155,
      "learning_rate": 8.622713097397158e-06,
      "loss": 0.4351,
      "step": 16815
    },
    {
      "epoch": 0.7716947363590473,
      "grad_norm": 0.49206414818763733,
      "learning_rate": 8.622544102180339e-06,
      "loss": 0.4573,
      "step": 16816
    },
    {
      "epoch": 0.7717406268643018,
      "grad_norm": 0.417838990688324,
      "learning_rate": 8.622375098252446e-06,
      "loss": 0.3379,
      "step": 16817
    },
    {
      "epoch": 0.7717865173695563,
      "grad_norm": 0.4458671808242798,
      "learning_rate": 8.622206085613884e-06,
      "loss": 0.3685,
      "step": 16818
    },
    {
      "epoch": 0.7718324078748107,
      "grad_norm": 0.44375380873680115,
      "learning_rate": 8.622037064265058e-06,
      "loss": 0.354,
      "step": 16819
    },
    {
      "epoch": 0.7718782983800652,
      "grad_norm": 0.5009838342666626,
      "learning_rate": 8.621868034206377e-06,
      "loss": 0.459,
      "step": 16820
    },
    {
      "epoch": 0.7719241888853197,
      "grad_norm": 0.42498329281806946,
      "learning_rate": 8.621698995438248e-06,
      "loss": 0.3224,
      "step": 16821
    },
    {
      "epoch": 0.771970079390574,
      "grad_norm": 0.4551143944263458,
      "learning_rate": 8.621529947961073e-06,
      "loss": 0.4056,
      "step": 16822
    },
    {
      "epoch": 0.7720159698958285,
      "grad_norm": 0.4636278748512268,
      "learning_rate": 8.621360891775265e-06,
      "loss": 0.4314,
      "step": 16823
    },
    {
      "epoch": 0.772061860401083,
      "grad_norm": 0.45101889967918396,
      "learning_rate": 8.621191826881224e-06,
      "loss": 0.3415,
      "step": 16824
    },
    {
      "epoch": 0.7721077509063374,
      "grad_norm": 0.4125545024871826,
      "learning_rate": 8.62102275327936e-06,
      "loss": 0.3254,
      "step": 16825
    },
    {
      "epoch": 0.7721536414115919,
      "grad_norm": 0.42588305473327637,
      "learning_rate": 8.62085367097008e-06,
      "loss": 0.3132,
      "step": 16826
    },
    {
      "epoch": 0.7721995319168464,
      "grad_norm": 0.46967780590057373,
      "learning_rate": 8.620684579953789e-06,
      "loss": 0.4171,
      "step": 16827
    },
    {
      "epoch": 0.7722454224221008,
      "grad_norm": 0.473286896944046,
      "learning_rate": 8.620515480230895e-06,
      "loss": 0.466,
      "step": 16828
    },
    {
      "epoch": 0.7722913129273553,
      "grad_norm": 0.44213396310806274,
      "learning_rate": 8.620346371801804e-06,
      "loss": 0.402,
      "step": 16829
    },
    {
      "epoch": 0.7723372034326098,
      "grad_norm": 0.43393293023109436,
      "learning_rate": 8.620177254666921e-06,
      "loss": 0.3829,
      "step": 16830
    },
    {
      "epoch": 0.7723830939378643,
      "grad_norm": 0.44293439388275146,
      "learning_rate": 8.620008128826656e-06,
      "loss": 0.3679,
      "step": 16831
    },
    {
      "epoch": 0.7724289844431187,
      "grad_norm": 0.44425949454307556,
      "learning_rate": 8.619838994281415e-06,
      "loss": 0.3191,
      "step": 16832
    },
    {
      "epoch": 0.7724748749483732,
      "grad_norm": 0.4396025538444519,
      "learning_rate": 8.619669851031602e-06,
      "loss": 0.3403,
      "step": 16833
    },
    {
      "epoch": 0.7725207654536277,
      "grad_norm": 0.571470320224762,
      "learning_rate": 8.619500699077626e-06,
      "loss": 0.3282,
      "step": 16834
    },
    {
      "epoch": 0.7725666559588821,
      "grad_norm": 0.4428199231624603,
      "learning_rate": 8.61933153841989e-06,
      "loss": 0.3644,
      "step": 16835
    },
    {
      "epoch": 0.7726125464641366,
      "grad_norm": 0.4916760325431824,
      "learning_rate": 8.619162369058808e-06,
      "loss": 0.4579,
      "step": 16836
    },
    {
      "epoch": 0.7726584369693911,
      "grad_norm": 0.5163108110427856,
      "learning_rate": 8.618993190994782e-06,
      "loss": 0.4467,
      "step": 16837
    },
    {
      "epoch": 0.7727043274746455,
      "grad_norm": 0.4558572471141815,
      "learning_rate": 8.61882400422822e-06,
      "loss": 0.4037,
      "step": 16838
    },
    {
      "epoch": 0.7727502179799,
      "grad_norm": 0.4770214557647705,
      "learning_rate": 8.618654808759525e-06,
      "loss": 0.4112,
      "step": 16839
    },
    {
      "epoch": 0.7727961084851545,
      "grad_norm": 0.46561184525489807,
      "learning_rate": 8.61848560458911e-06,
      "loss": 0.4434,
      "step": 16840
    },
    {
      "epoch": 0.7728419989904088,
      "grad_norm": 0.42374444007873535,
      "learning_rate": 8.618316391717378e-06,
      "loss": 0.3482,
      "step": 16841
    },
    {
      "epoch": 0.7728878894956633,
      "grad_norm": 0.468965619802475,
      "learning_rate": 8.618147170144738e-06,
      "loss": 0.4067,
      "step": 16842
    },
    {
      "epoch": 0.7729337800009178,
      "grad_norm": 0.4533752202987671,
      "learning_rate": 8.617977939871595e-06,
      "loss": 0.3618,
      "step": 16843
    },
    {
      "epoch": 0.7729796705061722,
      "grad_norm": 0.471626877784729,
      "learning_rate": 8.617808700898359e-06,
      "loss": 0.4275,
      "step": 16844
    },
    {
      "epoch": 0.7730255610114267,
      "grad_norm": 0.4311619997024536,
      "learning_rate": 8.617639453225433e-06,
      "loss": 0.3534,
      "step": 16845
    },
    {
      "epoch": 0.7730714515166812,
      "grad_norm": 0.4441860318183899,
      "learning_rate": 8.617470196853226e-06,
      "loss": 0.3883,
      "step": 16846
    },
    {
      "epoch": 0.7731173420219357,
      "grad_norm": 0.4784258008003235,
      "learning_rate": 8.617300931782146e-06,
      "loss": 0.4544,
      "step": 16847
    },
    {
      "epoch": 0.7731632325271901,
      "grad_norm": 0.4597739577293396,
      "learning_rate": 8.617131658012599e-06,
      "loss": 0.4052,
      "step": 16848
    },
    {
      "epoch": 0.7732091230324446,
      "grad_norm": 0.4413967430591583,
      "learning_rate": 8.61696237554499e-06,
      "loss": 0.3515,
      "step": 16849
    },
    {
      "epoch": 0.7732550135376991,
      "grad_norm": 0.4443657100200653,
      "learning_rate": 8.61679308437973e-06,
      "loss": 0.408,
      "step": 16850
    },
    {
      "epoch": 0.7733009040429535,
      "grad_norm": 0.47268804907798767,
      "learning_rate": 8.616623784517224e-06,
      "loss": 0.3758,
      "step": 16851
    },
    {
      "epoch": 0.773346794548208,
      "grad_norm": 0.4530348479747772,
      "learning_rate": 8.616454475957878e-06,
      "loss": 0.4328,
      "step": 16852
    },
    {
      "epoch": 0.7733926850534625,
      "grad_norm": 0.45229071378707886,
      "learning_rate": 8.6162851587021e-06,
      "loss": 0.386,
      "step": 16853
    },
    {
      "epoch": 0.7734385755587169,
      "grad_norm": 0.45669665932655334,
      "learning_rate": 8.6161158327503e-06,
      "loss": 0.3584,
      "step": 16854
    },
    {
      "epoch": 0.7734844660639714,
      "grad_norm": 0.45503878593444824,
      "learning_rate": 8.61594649810288e-06,
      "loss": 0.3845,
      "step": 16855
    },
    {
      "epoch": 0.7735303565692259,
      "grad_norm": 0.42145493626594543,
      "learning_rate": 8.615777154760252e-06,
      "loss": 0.3327,
      "step": 16856
    },
    {
      "epoch": 0.7735762470744802,
      "grad_norm": 0.40514248609542847,
      "learning_rate": 8.61560780272282e-06,
      "loss": 0.2942,
      "step": 16857
    },
    {
      "epoch": 0.7736221375797347,
      "grad_norm": 0.4242538809776306,
      "learning_rate": 8.615438441990992e-06,
      "loss": 0.365,
      "step": 16858
    },
    {
      "epoch": 0.7736680280849892,
      "grad_norm": 0.4237339496612549,
      "learning_rate": 8.615269072565176e-06,
      "loss": 0.3319,
      "step": 16859
    },
    {
      "epoch": 0.7737139185902436,
      "grad_norm": 0.4476374089717865,
      "learning_rate": 8.615099694445778e-06,
      "loss": 0.3486,
      "step": 16860
    },
    {
      "epoch": 0.7737598090954981,
      "grad_norm": 0.4669581353664398,
      "learning_rate": 8.61493030763321e-06,
      "loss": 0.4397,
      "step": 16861
    },
    {
      "epoch": 0.7738056996007526,
      "grad_norm": 0.49119532108306885,
      "learning_rate": 8.614760912127871e-06,
      "loss": 0.4058,
      "step": 16862
    },
    {
      "epoch": 0.773851590106007,
      "grad_norm": 0.4685514569282532,
      "learning_rate": 8.614591507930174e-06,
      "loss": 0.3299,
      "step": 16863
    },
    {
      "epoch": 0.7738974806112615,
      "grad_norm": 0.4511096179485321,
      "learning_rate": 8.614422095040527e-06,
      "loss": 0.3647,
      "step": 16864
    },
    {
      "epoch": 0.773943371116516,
      "grad_norm": 0.5055418014526367,
      "learning_rate": 8.614252673459335e-06,
      "loss": 0.52,
      "step": 16865
    },
    {
      "epoch": 0.7739892616217705,
      "grad_norm": 0.45167604088783264,
      "learning_rate": 8.614083243187004e-06,
      "loss": 0.3475,
      "step": 16866
    },
    {
      "epoch": 0.7740351521270249,
      "grad_norm": 0.46586674451828003,
      "learning_rate": 8.613913804223946e-06,
      "loss": 0.4013,
      "step": 16867
    },
    {
      "epoch": 0.7740810426322794,
      "grad_norm": 0.4665861129760742,
      "learning_rate": 8.613744356570563e-06,
      "loss": 0.3975,
      "step": 16868
    },
    {
      "epoch": 0.7741269331375339,
      "grad_norm": 0.46501991152763367,
      "learning_rate": 8.613574900227269e-06,
      "loss": 0.3893,
      "step": 16869
    },
    {
      "epoch": 0.7741728236427883,
      "grad_norm": 0.4572458565235138,
      "learning_rate": 8.613405435194466e-06,
      "loss": 0.4155,
      "step": 16870
    },
    {
      "epoch": 0.7742187141480428,
      "grad_norm": 0.5149495005607605,
      "learning_rate": 8.613235961472562e-06,
      "loss": 0.4377,
      "step": 16871
    },
    {
      "epoch": 0.7742646046532973,
      "grad_norm": 0.46111032366752625,
      "learning_rate": 8.61306647906197e-06,
      "loss": 0.432,
      "step": 16872
    },
    {
      "epoch": 0.7743104951585517,
      "grad_norm": 0.40280014276504517,
      "learning_rate": 8.612896987963088e-06,
      "loss": 0.2958,
      "step": 16873
    },
    {
      "epoch": 0.7743563856638062,
      "grad_norm": 0.4556792080402374,
      "learning_rate": 8.612727488176333e-06,
      "loss": 0.3978,
      "step": 16874
    },
    {
      "epoch": 0.7744022761690607,
      "grad_norm": 0.4502858817577362,
      "learning_rate": 8.612557979702107e-06,
      "loss": 0.3888,
      "step": 16875
    },
    {
      "epoch": 0.774448166674315,
      "grad_norm": 0.4924439489841461,
      "learning_rate": 8.61238846254082e-06,
      "loss": 0.4787,
      "step": 16876
    },
    {
      "epoch": 0.7744940571795695,
      "grad_norm": 0.4544217884540558,
      "learning_rate": 8.612218936692878e-06,
      "loss": 0.3752,
      "step": 16877
    },
    {
      "epoch": 0.774539947684824,
      "grad_norm": 0.4419492483139038,
      "learning_rate": 8.61204940215869e-06,
      "loss": 0.3947,
      "step": 16878
    },
    {
      "epoch": 0.7745858381900784,
      "grad_norm": 0.47296708822250366,
      "learning_rate": 8.611879858938662e-06,
      "loss": 0.3819,
      "step": 16879
    },
    {
      "epoch": 0.7746317286953329,
      "grad_norm": 0.49265536665916443,
      "learning_rate": 8.611710307033204e-06,
      "loss": 0.4409,
      "step": 16880
    },
    {
      "epoch": 0.7746776192005874,
      "grad_norm": 0.4374864995479584,
      "learning_rate": 8.611540746442723e-06,
      "loss": 0.3715,
      "step": 16881
    },
    {
      "epoch": 0.7747235097058418,
      "grad_norm": 0.4849449396133423,
      "learning_rate": 8.611371177167625e-06,
      "loss": 0.3959,
      "step": 16882
    },
    {
      "epoch": 0.7747694002110963,
      "grad_norm": 0.4384102523326874,
      "learning_rate": 8.611201599208321e-06,
      "loss": 0.3721,
      "step": 16883
    },
    {
      "epoch": 0.7748152907163508,
      "grad_norm": 0.5025368928909302,
      "learning_rate": 8.611032012565214e-06,
      "loss": 0.5033,
      "step": 16884
    },
    {
      "epoch": 0.7748611812216053,
      "grad_norm": 0.46315592527389526,
      "learning_rate": 8.610862417238716e-06,
      "loss": 0.4067,
      "step": 16885
    },
    {
      "epoch": 0.7749070717268597,
      "grad_norm": 0.4575701951980591,
      "learning_rate": 8.610692813229235e-06,
      "loss": 0.4217,
      "step": 16886
    },
    {
      "epoch": 0.7749529622321142,
      "grad_norm": 0.4692322909832001,
      "learning_rate": 8.610523200537178e-06,
      "loss": 0.4057,
      "step": 16887
    },
    {
      "epoch": 0.7749988527373687,
      "grad_norm": 0.45591580867767334,
      "learning_rate": 8.61035357916295e-06,
      "loss": 0.3993,
      "step": 16888
    },
    {
      "epoch": 0.7750447432426231,
      "grad_norm": 0.42353740334510803,
      "learning_rate": 8.610183949106961e-06,
      "loss": 0.3249,
      "step": 16889
    },
    {
      "epoch": 0.7750906337478776,
      "grad_norm": 0.4491284489631653,
      "learning_rate": 8.610014310369618e-06,
      "loss": 0.4171,
      "step": 16890
    },
    {
      "epoch": 0.7751365242531321,
      "grad_norm": 0.4470716714859009,
      "learning_rate": 8.609844662951333e-06,
      "loss": 0.3534,
      "step": 16891
    },
    {
      "epoch": 0.7751824147583865,
      "grad_norm": 0.46116113662719727,
      "learning_rate": 8.60967500685251e-06,
      "loss": 0.4358,
      "step": 16892
    },
    {
      "epoch": 0.775228305263641,
      "grad_norm": 0.45538708567619324,
      "learning_rate": 8.609505342073556e-06,
      "loss": 0.3734,
      "step": 16893
    },
    {
      "epoch": 0.7752741957688954,
      "grad_norm": 0.440536767244339,
      "learning_rate": 8.609335668614882e-06,
      "loss": 0.3909,
      "step": 16894
    },
    {
      "epoch": 0.7753200862741498,
      "grad_norm": 0.4551977515220642,
      "learning_rate": 8.609165986476898e-06,
      "loss": 0.3655,
      "step": 16895
    },
    {
      "epoch": 0.7753659767794043,
      "grad_norm": 0.46975958347320557,
      "learning_rate": 8.608996295660004e-06,
      "loss": 0.3925,
      "step": 16896
    },
    {
      "epoch": 0.7754118672846588,
      "grad_norm": 0.48826202750205994,
      "learning_rate": 8.608826596164615e-06,
      "loss": 0.4516,
      "step": 16897
    },
    {
      "epoch": 0.7754577577899132,
      "grad_norm": 0.49547696113586426,
      "learning_rate": 8.608656887991136e-06,
      "loss": 0.4593,
      "step": 16898
    },
    {
      "epoch": 0.7755036482951677,
      "grad_norm": 0.4274131655693054,
      "learning_rate": 8.608487171139978e-06,
      "loss": 0.3203,
      "step": 16899
    },
    {
      "epoch": 0.7755495388004222,
      "grad_norm": 0.43693792819976807,
      "learning_rate": 8.608317445611547e-06,
      "loss": 0.4103,
      "step": 16900
    },
    {
      "epoch": 0.7755954293056767,
      "grad_norm": 0.40566906332969666,
      "learning_rate": 8.60814771140625e-06,
      "loss": 0.3196,
      "step": 16901
    },
    {
      "epoch": 0.7756413198109311,
      "grad_norm": 0.43336349725723267,
      "learning_rate": 8.607977968524497e-06,
      "loss": 0.3718,
      "step": 16902
    },
    {
      "epoch": 0.7756872103161856,
      "grad_norm": 0.48081040382385254,
      "learning_rate": 8.607808216966696e-06,
      "loss": 0.4117,
      "step": 16903
    },
    {
      "epoch": 0.7757331008214401,
      "grad_norm": 0.4720076620578766,
      "learning_rate": 8.607638456733255e-06,
      "loss": 0.3469,
      "step": 16904
    },
    {
      "epoch": 0.7757789913266945,
      "grad_norm": 0.4629420042037964,
      "learning_rate": 8.607468687824581e-06,
      "loss": 0.4151,
      "step": 16905
    },
    {
      "epoch": 0.775824881831949,
      "grad_norm": 0.4876801371574402,
      "learning_rate": 8.607298910241084e-06,
      "loss": 0.4627,
      "step": 16906
    },
    {
      "epoch": 0.7758707723372035,
      "grad_norm": 0.4505068063735962,
      "learning_rate": 8.607129123983172e-06,
      "loss": 0.3839,
      "step": 16907
    },
    {
      "epoch": 0.7759166628424579,
      "grad_norm": 0.4997379183769226,
      "learning_rate": 8.606959329051253e-06,
      "loss": 0.5035,
      "step": 16908
    },
    {
      "epoch": 0.7759625533477124,
      "grad_norm": 0.44132763147354126,
      "learning_rate": 8.606789525445735e-06,
      "loss": 0.3963,
      "step": 16909
    },
    {
      "epoch": 0.7760084438529669,
      "grad_norm": 0.44596344232559204,
      "learning_rate": 8.606619713167025e-06,
      "loss": 0.3627,
      "step": 16910
    },
    {
      "epoch": 0.7760543343582212,
      "grad_norm": 0.45813503861427307,
      "learning_rate": 8.606449892215536e-06,
      "loss": 0.3769,
      "step": 16911
    },
    {
      "epoch": 0.7761002248634757,
      "grad_norm": 0.4365362524986267,
      "learning_rate": 8.60628006259167e-06,
      "loss": 0.3597,
      "step": 16912
    },
    {
      "epoch": 0.7761461153687302,
      "grad_norm": 0.417276531457901,
      "learning_rate": 8.606110224295842e-06,
      "loss": 0.324,
      "step": 16913
    },
    {
      "epoch": 0.7761920058739846,
      "grad_norm": 0.47401151061058044,
      "learning_rate": 8.605940377328454e-06,
      "loss": 0.4582,
      "step": 16914
    },
    {
      "epoch": 0.7762378963792391,
      "grad_norm": 0.4661887586116791,
      "learning_rate": 8.60577052168992e-06,
      "loss": 0.4247,
      "step": 16915
    },
    {
      "epoch": 0.7762837868844936,
      "grad_norm": 0.45259785652160645,
      "learning_rate": 8.605600657380643e-06,
      "loss": 0.3892,
      "step": 16916
    },
    {
      "epoch": 0.776329677389748,
      "grad_norm": 0.4371308386325836,
      "learning_rate": 8.605430784401037e-06,
      "loss": 0.3999,
      "step": 16917
    },
    {
      "epoch": 0.7763755678950025,
      "grad_norm": 0.4300791323184967,
      "learning_rate": 8.605260902751507e-06,
      "loss": 0.3667,
      "step": 16918
    },
    {
      "epoch": 0.776421458400257,
      "grad_norm": 0.45549625158309937,
      "learning_rate": 8.605091012432462e-06,
      "loss": 0.3963,
      "step": 16919
    },
    {
      "epoch": 0.7764673489055115,
      "grad_norm": 0.45245325565338135,
      "learning_rate": 8.60492111344431e-06,
      "loss": 0.3855,
      "step": 16920
    },
    {
      "epoch": 0.7765132394107659,
      "grad_norm": 0.4253554046154022,
      "learning_rate": 8.604751205787463e-06,
      "loss": 0.3163,
      "step": 16921
    },
    {
      "epoch": 0.7765591299160204,
      "grad_norm": 0.4317529499530792,
      "learning_rate": 8.604581289462324e-06,
      "loss": 0.3426,
      "step": 16922
    },
    {
      "epoch": 0.7766050204212749,
      "grad_norm": 0.4471127688884735,
      "learning_rate": 8.604411364469307e-06,
      "loss": 0.3502,
      "step": 16923
    },
    {
      "epoch": 0.7766509109265293,
      "grad_norm": 0.4521816372871399,
      "learning_rate": 8.604241430808818e-06,
      "loss": 0.4294,
      "step": 16924
    },
    {
      "epoch": 0.7766968014317838,
      "grad_norm": 0.44483277201652527,
      "learning_rate": 8.604071488481265e-06,
      "loss": 0.4056,
      "step": 16925
    },
    {
      "epoch": 0.7767426919370383,
      "grad_norm": 0.46801871061325073,
      "learning_rate": 8.603901537487059e-06,
      "loss": 0.3919,
      "step": 16926
    },
    {
      "epoch": 0.7767885824422927,
      "grad_norm": 0.4468994140625,
      "learning_rate": 8.603731577826605e-06,
      "loss": 0.3746,
      "step": 16927
    },
    {
      "epoch": 0.7768344729475471,
      "grad_norm": 0.47118332982063293,
      "learning_rate": 8.603561609500314e-06,
      "loss": 0.409,
      "step": 16928
    },
    {
      "epoch": 0.7768803634528016,
      "grad_norm": 0.4489784240722656,
      "learning_rate": 8.603391632508597e-06,
      "loss": 0.3757,
      "step": 16929
    },
    {
      "epoch": 0.776926253958056,
      "grad_norm": 0.44380053877830505,
      "learning_rate": 8.603221646851859e-06,
      "loss": 0.3841,
      "step": 16930
    },
    {
      "epoch": 0.7769721444633105,
      "grad_norm": 0.4650479257106781,
      "learning_rate": 8.60305165253051e-06,
      "loss": 0.4163,
      "step": 16931
    },
    {
      "epoch": 0.777018034968565,
      "grad_norm": 0.4571394920349121,
      "learning_rate": 8.602881649544959e-06,
      "loss": 0.4402,
      "step": 16932
    },
    {
      "epoch": 0.7770639254738194,
      "grad_norm": 0.4364791214466095,
      "learning_rate": 8.602711637895615e-06,
      "loss": 0.3638,
      "step": 16933
    },
    {
      "epoch": 0.7771098159790739,
      "grad_norm": 0.46674564480781555,
      "learning_rate": 8.602541617582887e-06,
      "loss": 0.4149,
      "step": 16934
    },
    {
      "epoch": 0.7771557064843284,
      "grad_norm": 0.4759582281112671,
      "learning_rate": 8.602371588607183e-06,
      "loss": 0.4254,
      "step": 16935
    },
    {
      "epoch": 0.7772015969895829,
      "grad_norm": 0.46773627400398254,
      "learning_rate": 8.602201550968911e-06,
      "loss": 0.4399,
      "step": 16936
    },
    {
      "epoch": 0.7772474874948373,
      "grad_norm": 0.45208248496055603,
      "learning_rate": 8.602031504668483e-06,
      "loss": 0.4296,
      "step": 16937
    },
    {
      "epoch": 0.7772933780000918,
      "grad_norm": 0.40974459052085876,
      "learning_rate": 8.601861449706305e-06,
      "loss": 0.3439,
      "step": 16938
    },
    {
      "epoch": 0.7773392685053463,
      "grad_norm": 0.48815661668777466,
      "learning_rate": 8.601691386082789e-06,
      "loss": 0.4397,
      "step": 16939
    },
    {
      "epoch": 0.7773851590106007,
      "grad_norm": 0.44761961698532104,
      "learning_rate": 8.60152131379834e-06,
      "loss": 0.4047,
      "step": 16940
    },
    {
      "epoch": 0.7774310495158552,
      "grad_norm": 0.4427563548088074,
      "learning_rate": 8.601351232853368e-06,
      "loss": 0.3595,
      "step": 16941
    },
    {
      "epoch": 0.7774769400211097,
      "grad_norm": 0.4863508939743042,
      "learning_rate": 8.601181143248285e-06,
      "loss": 0.4809,
      "step": 16942
    },
    {
      "epoch": 0.7775228305263641,
      "grad_norm": 0.41547828912734985,
      "learning_rate": 8.601011044983497e-06,
      "loss": 0.3279,
      "step": 16943
    },
    {
      "epoch": 0.7775687210316186,
      "grad_norm": 0.5034317374229431,
      "learning_rate": 8.600840938059414e-06,
      "loss": 0.329,
      "step": 16944
    },
    {
      "epoch": 0.7776146115368731,
      "grad_norm": 0.4456773102283478,
      "learning_rate": 8.600670822476444e-06,
      "loss": 0.3311,
      "step": 16945
    },
    {
      "epoch": 0.7776605020421274,
      "grad_norm": 0.4610023498535156,
      "learning_rate": 8.600500698235e-06,
      "loss": 0.3721,
      "step": 16946
    },
    {
      "epoch": 0.7777063925473819,
      "grad_norm": 0.47589805722236633,
      "learning_rate": 8.600330565335486e-06,
      "loss": 0.439,
      "step": 16947
    },
    {
      "epoch": 0.7777522830526364,
      "grad_norm": 0.4469650387763977,
      "learning_rate": 8.600160423778314e-06,
      "loss": 0.4025,
      "step": 16948
    },
    {
      "epoch": 0.7777981735578908,
      "grad_norm": 0.4813380241394043,
      "learning_rate": 8.599990273563891e-06,
      "loss": 0.4631,
      "step": 16949
    },
    {
      "epoch": 0.7778440640631453,
      "grad_norm": 0.4371643364429474,
      "learning_rate": 8.599820114692628e-06,
      "loss": 0.3442,
      "step": 16950
    },
    {
      "epoch": 0.7778899545683998,
      "grad_norm": 0.4149708151817322,
      "learning_rate": 8.599649947164934e-06,
      "loss": 0.3996,
      "step": 16951
    },
    {
      "epoch": 0.7779358450736542,
      "grad_norm": 0.44185352325439453,
      "learning_rate": 8.599479770981219e-06,
      "loss": 0.4062,
      "step": 16952
    },
    {
      "epoch": 0.7779817355789087,
      "grad_norm": 0.4264153242111206,
      "learning_rate": 8.59930958614189e-06,
      "loss": 0.3801,
      "step": 16953
    },
    {
      "epoch": 0.7780276260841632,
      "grad_norm": 0.47552672028541565,
      "learning_rate": 8.59913939264736e-06,
      "loss": 0.44,
      "step": 16954
    },
    {
      "epoch": 0.7780735165894177,
      "grad_norm": 0.4497244358062744,
      "learning_rate": 8.598969190498032e-06,
      "loss": 0.3831,
      "step": 16955
    },
    {
      "epoch": 0.7781194070946721,
      "grad_norm": 0.4605443775653839,
      "learning_rate": 8.59879897969432e-06,
      "loss": 0.3706,
      "step": 16956
    },
    {
      "epoch": 0.7781652975999266,
      "grad_norm": 0.4667157232761383,
      "learning_rate": 8.598628760236633e-06,
      "loss": 0.4186,
      "step": 16957
    },
    {
      "epoch": 0.7782111881051811,
      "grad_norm": 0.4445166289806366,
      "learning_rate": 8.59845853212538e-06,
      "loss": 0.3698,
      "step": 16958
    },
    {
      "epoch": 0.7782570786104355,
      "grad_norm": 0.46820008754730225,
      "learning_rate": 8.598288295360969e-06,
      "loss": 0.3991,
      "step": 16959
    },
    {
      "epoch": 0.77830296911569,
      "grad_norm": 0.45852819085121155,
      "learning_rate": 8.598118049943811e-06,
      "loss": 0.4201,
      "step": 16960
    },
    {
      "epoch": 0.7783488596209445,
      "grad_norm": 0.5182089805603027,
      "learning_rate": 8.597947795874313e-06,
      "loss": 0.5053,
      "step": 16961
    },
    {
      "epoch": 0.7783947501261989,
      "grad_norm": 0.4940803349018097,
      "learning_rate": 8.59777753315289e-06,
      "loss": 0.4411,
      "step": 16962
    },
    {
      "epoch": 0.7784406406314534,
      "grad_norm": 0.47086775302886963,
      "learning_rate": 8.597607261779943e-06,
      "loss": 0.3839,
      "step": 16963
    },
    {
      "epoch": 0.7784865311367078,
      "grad_norm": 0.4992276132106781,
      "learning_rate": 8.597436981755889e-06,
      "loss": 0.3903,
      "step": 16964
    },
    {
      "epoch": 0.7785324216419622,
      "grad_norm": 0.4815516471862793,
      "learning_rate": 8.597266693081131e-06,
      "loss": 0.4351,
      "step": 16965
    },
    {
      "epoch": 0.7785783121472167,
      "grad_norm": 0.4850395917892456,
      "learning_rate": 8.597096395756085e-06,
      "loss": 0.4152,
      "step": 16966
    },
    {
      "epoch": 0.7786242026524712,
      "grad_norm": 0.4815860986709595,
      "learning_rate": 8.596926089781157e-06,
      "loss": 0.4305,
      "step": 16967
    },
    {
      "epoch": 0.7786700931577256,
      "grad_norm": 0.4743412733078003,
      "learning_rate": 8.596755775156758e-06,
      "loss": 0.421,
      "step": 16968
    },
    {
      "epoch": 0.7787159836629801,
      "grad_norm": 0.4712136387825012,
      "learning_rate": 8.596585451883294e-06,
      "loss": 0.4196,
      "step": 16969
    },
    {
      "epoch": 0.7787618741682346,
      "grad_norm": 0.41681602597236633,
      "learning_rate": 8.596415119961179e-06,
      "loss": 0.3049,
      "step": 16970
    },
    {
      "epoch": 0.778807764673489,
      "grad_norm": 0.45696568489074707,
      "learning_rate": 8.59624477939082e-06,
      "loss": 0.3799,
      "step": 16971
    },
    {
      "epoch": 0.7788536551787435,
      "grad_norm": 0.47887057065963745,
      "learning_rate": 8.596074430172627e-06,
      "loss": 0.4131,
      "step": 16972
    },
    {
      "epoch": 0.778899545683998,
      "grad_norm": 0.4474768042564392,
      "learning_rate": 8.595904072307011e-06,
      "loss": 0.3473,
      "step": 16973
    },
    {
      "epoch": 0.7789454361892525,
      "grad_norm": 0.46849435567855835,
      "learning_rate": 8.595733705794381e-06,
      "loss": 0.4595,
      "step": 16974
    },
    {
      "epoch": 0.7789913266945069,
      "grad_norm": 0.4657447040081024,
      "learning_rate": 8.595563330635144e-06,
      "loss": 0.4284,
      "step": 16975
    },
    {
      "epoch": 0.7790372171997614,
      "grad_norm": 0.4583030045032501,
      "learning_rate": 8.595392946829712e-06,
      "loss": 0.4212,
      "step": 16976
    },
    {
      "epoch": 0.7790831077050159,
      "grad_norm": 0.5090739130973816,
      "learning_rate": 8.595222554378497e-06,
      "loss": 0.4935,
      "step": 16977
    },
    {
      "epoch": 0.7791289982102703,
      "grad_norm": 0.47014614939689636,
      "learning_rate": 8.595052153281905e-06,
      "loss": 0.4966,
      "step": 16978
    },
    {
      "epoch": 0.7791748887155248,
      "grad_norm": 0.4327756464481354,
      "learning_rate": 8.594881743540347e-06,
      "loss": 0.3579,
      "step": 16979
    },
    {
      "epoch": 0.7792207792207793,
      "grad_norm": 0.46908295154571533,
      "learning_rate": 8.594711325154235e-06,
      "loss": 0.4082,
      "step": 16980
    },
    {
      "epoch": 0.7792666697260336,
      "grad_norm": 0.47554317116737366,
      "learning_rate": 8.594540898123976e-06,
      "loss": 0.4799,
      "step": 16981
    },
    {
      "epoch": 0.7793125602312881,
      "grad_norm": 0.47663891315460205,
      "learning_rate": 8.594370462449979e-06,
      "loss": 0.4053,
      "step": 16982
    },
    {
      "epoch": 0.7793584507365426,
      "grad_norm": 0.4742930829524994,
      "learning_rate": 8.594200018132654e-06,
      "loss": 0.4108,
      "step": 16983
    },
    {
      "epoch": 0.779404341241797,
      "grad_norm": 0.4363305866718292,
      "learning_rate": 8.594029565172418e-06,
      "loss": 0.369,
      "step": 16984
    },
    {
      "epoch": 0.7794502317470515,
      "grad_norm": 0.4817432463169098,
      "learning_rate": 8.59385910356967e-06,
      "loss": 0.5153,
      "step": 16985
    },
    {
      "epoch": 0.779496122252306,
      "grad_norm": 0.4340348541736603,
      "learning_rate": 8.593688633324826e-06,
      "loss": 0.3834,
      "step": 16986
    },
    {
      "epoch": 0.7795420127575604,
      "grad_norm": 0.5035750865936279,
      "learning_rate": 8.593518154438297e-06,
      "loss": 0.5914,
      "step": 16987
    },
    {
      "epoch": 0.7795879032628149,
      "grad_norm": 0.4310015141963959,
      "learning_rate": 8.59334766691049e-06,
      "loss": 0.3538,
      "step": 16988
    },
    {
      "epoch": 0.7796337937680694,
      "grad_norm": 0.49534115195274353,
      "learning_rate": 8.593177170741815e-06,
      "loss": 0.3981,
      "step": 16989
    },
    {
      "epoch": 0.7796796842733239,
      "grad_norm": 0.45083001255989075,
      "learning_rate": 8.593006665932685e-06,
      "loss": 0.332,
      "step": 16990
    },
    {
      "epoch": 0.7797255747785783,
      "grad_norm": 0.4978991150856018,
      "learning_rate": 8.592836152483507e-06,
      "loss": 0.4671,
      "step": 16991
    },
    {
      "epoch": 0.7797714652838328,
      "grad_norm": 0.5678019523620605,
      "learning_rate": 8.592665630394692e-06,
      "loss": 0.5559,
      "step": 16992
    },
    {
      "epoch": 0.7798173557890873,
      "grad_norm": 0.43045276403427124,
      "learning_rate": 8.59249509966665e-06,
      "loss": 0.3558,
      "step": 16993
    },
    {
      "epoch": 0.7798632462943417,
      "grad_norm": 0.4593508243560791,
      "learning_rate": 8.59232456029979e-06,
      "loss": 0.4298,
      "step": 16994
    },
    {
      "epoch": 0.7799091367995962,
      "grad_norm": 0.48363614082336426,
      "learning_rate": 8.592154012294525e-06,
      "loss": 0.4232,
      "step": 16995
    },
    {
      "epoch": 0.7799550273048507,
      "grad_norm": 0.46471211314201355,
      "learning_rate": 8.591983455651262e-06,
      "loss": 0.3879,
      "step": 16996
    },
    {
      "epoch": 0.780000917810105,
      "grad_norm": 0.47445499897003174,
      "learning_rate": 8.591812890370413e-06,
      "loss": 0.3882,
      "step": 16997
    },
    {
      "epoch": 0.7800468083153596,
      "grad_norm": 0.47170308232307434,
      "learning_rate": 8.591642316452387e-06,
      "loss": 0.4082,
      "step": 16998
    },
    {
      "epoch": 0.780092698820614,
      "grad_norm": 0.46292388439178467,
      "learning_rate": 8.591471733897595e-06,
      "loss": 0.4217,
      "step": 16999
    },
    {
      "epoch": 0.7801385893258684,
      "grad_norm": 0.4319770038127899,
      "learning_rate": 8.591301142706449e-06,
      "loss": 0.3736,
      "step": 17000
    },
    {
      "epoch": 0.7801844798311229,
      "grad_norm": 0.4480537176132202,
      "learning_rate": 8.591130542879356e-06,
      "loss": 0.374,
      "step": 17001
    },
    {
      "epoch": 0.7802303703363774,
      "grad_norm": 0.44551634788513184,
      "learning_rate": 8.590959934416726e-06,
      "loss": 0.3389,
      "step": 17002
    },
    {
      "epoch": 0.7802762608416318,
      "grad_norm": 0.4868551790714264,
      "learning_rate": 8.590789317318972e-06,
      "loss": 0.4295,
      "step": 17003
    },
    {
      "epoch": 0.7803221513468863,
      "grad_norm": 0.44733572006225586,
      "learning_rate": 8.590618691586502e-06,
      "loss": 0.3556,
      "step": 17004
    },
    {
      "epoch": 0.7803680418521408,
      "grad_norm": 0.4398041367530823,
      "learning_rate": 8.590448057219728e-06,
      "loss": 0.3466,
      "step": 17005
    },
    {
      "epoch": 0.7804139323573952,
      "grad_norm": 0.4365110397338867,
      "learning_rate": 8.590277414219059e-06,
      "loss": 0.4271,
      "step": 17006
    },
    {
      "epoch": 0.7804598228626497,
      "grad_norm": 0.460732638835907,
      "learning_rate": 8.590106762584907e-06,
      "loss": 0.4107,
      "step": 17007
    },
    {
      "epoch": 0.7805057133679042,
      "grad_norm": 0.47732633352279663,
      "learning_rate": 8.589936102317683e-06,
      "loss": 0.4355,
      "step": 17008
    },
    {
      "epoch": 0.7805516038731587,
      "grad_norm": 0.46782296895980835,
      "learning_rate": 8.589765433417792e-06,
      "loss": 0.443,
      "step": 17009
    },
    {
      "epoch": 0.7805974943784131,
      "grad_norm": 0.432796835899353,
      "learning_rate": 8.589594755885648e-06,
      "loss": 0.3796,
      "step": 17010
    },
    {
      "epoch": 0.7806433848836676,
      "grad_norm": 0.4564862549304962,
      "learning_rate": 8.589424069721664e-06,
      "loss": 0.4201,
      "step": 17011
    },
    {
      "epoch": 0.7806892753889221,
      "grad_norm": 0.4354449212551117,
      "learning_rate": 8.589253374926248e-06,
      "loss": 0.3419,
      "step": 17012
    },
    {
      "epoch": 0.7807351658941765,
      "grad_norm": 0.4580015540122986,
      "learning_rate": 8.589082671499809e-06,
      "loss": 0.4061,
      "step": 17013
    },
    {
      "epoch": 0.780781056399431,
      "grad_norm": 0.4637317657470703,
      "learning_rate": 8.588911959442758e-06,
      "loss": 0.3456,
      "step": 17014
    },
    {
      "epoch": 0.7808269469046855,
      "grad_norm": 0.4457900822162628,
      "learning_rate": 8.58874123875551e-06,
      "loss": 0.3577,
      "step": 17015
    },
    {
      "epoch": 0.7808728374099398,
      "grad_norm": 0.4248909056186676,
      "learning_rate": 8.588570509438468e-06,
      "loss": 0.352,
      "step": 17016
    },
    {
      "epoch": 0.7809187279151943,
      "grad_norm": 0.4572480320930481,
      "learning_rate": 8.588399771492048e-06,
      "loss": 0.3969,
      "step": 17017
    },
    {
      "epoch": 0.7809646184204488,
      "grad_norm": 0.44316598773002625,
      "learning_rate": 8.58822902491666e-06,
      "loss": 0.3704,
      "step": 17018
    },
    {
      "epoch": 0.7810105089257032,
      "grad_norm": 0.45023679733276367,
      "learning_rate": 8.588058269712714e-06,
      "loss": 0.4063,
      "step": 17019
    },
    {
      "epoch": 0.7810563994309577,
      "grad_norm": 0.4691431224346161,
      "learning_rate": 8.587887505880618e-06,
      "loss": 0.4028,
      "step": 17020
    },
    {
      "epoch": 0.7811022899362122,
      "grad_norm": 0.4431575536727905,
      "learning_rate": 8.587716733420787e-06,
      "loss": 0.4269,
      "step": 17021
    },
    {
      "epoch": 0.7811481804414666,
      "grad_norm": 0.4495683014392853,
      "learning_rate": 8.587545952333629e-06,
      "loss": 0.3554,
      "step": 17022
    },
    {
      "epoch": 0.7811940709467211,
      "grad_norm": 0.4431593418121338,
      "learning_rate": 8.587375162619555e-06,
      "loss": 0.3478,
      "step": 17023
    },
    {
      "epoch": 0.7812399614519756,
      "grad_norm": 0.4422595202922821,
      "learning_rate": 8.587204364278976e-06,
      "loss": 0.376,
      "step": 17024
    },
    {
      "epoch": 0.7812858519572301,
      "grad_norm": 0.4178480803966522,
      "learning_rate": 8.587033557312303e-06,
      "loss": 0.3281,
      "step": 17025
    },
    {
      "epoch": 0.7813317424624845,
      "grad_norm": 0.4244552254676819,
      "learning_rate": 8.586862741719945e-06,
      "loss": 0.3638,
      "step": 17026
    },
    {
      "epoch": 0.781377632967739,
      "grad_norm": 0.45411446690559387,
      "learning_rate": 8.586691917502315e-06,
      "loss": 0.4002,
      "step": 17027
    },
    {
      "epoch": 0.7814235234729935,
      "grad_norm": 0.46251872181892395,
      "learning_rate": 8.586521084659824e-06,
      "loss": 0.3461,
      "step": 17028
    },
    {
      "epoch": 0.7814694139782479,
      "grad_norm": 0.4846585988998413,
      "learning_rate": 8.586350243192881e-06,
      "loss": 0.504,
      "step": 17029
    },
    {
      "epoch": 0.7815153044835024,
      "grad_norm": 0.49908068776130676,
      "learning_rate": 8.586179393101897e-06,
      "loss": 0.4969,
      "step": 17030
    },
    {
      "epoch": 0.7815611949887569,
      "grad_norm": 0.4757101535797119,
      "learning_rate": 8.586008534387285e-06,
      "loss": 0.3561,
      "step": 17031
    },
    {
      "epoch": 0.7816070854940113,
      "grad_norm": 0.42304080724716187,
      "learning_rate": 8.585837667049452e-06,
      "loss": 0.404,
      "step": 17032
    },
    {
      "epoch": 0.7816529759992658,
      "grad_norm": 0.4756632447242737,
      "learning_rate": 8.585666791088812e-06,
      "loss": 0.425,
      "step": 17033
    },
    {
      "epoch": 0.7816988665045203,
      "grad_norm": 0.43152543902397156,
      "learning_rate": 8.585495906505776e-06,
      "loss": 0.3655,
      "step": 17034
    },
    {
      "epoch": 0.7817447570097746,
      "grad_norm": 0.4368312656879425,
      "learning_rate": 8.585325013300754e-06,
      "loss": 0.3646,
      "step": 17035
    },
    {
      "epoch": 0.7817906475150291,
      "grad_norm": 0.484638512134552,
      "learning_rate": 8.585154111474155e-06,
      "loss": 0.3886,
      "step": 17036
    },
    {
      "epoch": 0.7818365380202836,
      "grad_norm": 0.45188310742378235,
      "learning_rate": 8.584983201026394e-06,
      "loss": 0.3834,
      "step": 17037
    },
    {
      "epoch": 0.781882428525538,
      "grad_norm": 0.44477757811546326,
      "learning_rate": 8.584812281957877e-06,
      "loss": 0.3963,
      "step": 17038
    },
    {
      "epoch": 0.7819283190307925,
      "grad_norm": 0.4587835371494293,
      "learning_rate": 8.584641354269021e-06,
      "loss": 0.4422,
      "step": 17039
    },
    {
      "epoch": 0.781974209536047,
      "grad_norm": 0.41530269384384155,
      "learning_rate": 8.584470417960232e-06,
      "loss": 0.3083,
      "step": 17040
    },
    {
      "epoch": 0.7820201000413014,
      "grad_norm": 0.4844847023487091,
      "learning_rate": 8.584299473031923e-06,
      "loss": 0.5143,
      "step": 17041
    },
    {
      "epoch": 0.7820659905465559,
      "grad_norm": 0.46292179822921753,
      "learning_rate": 8.584128519484507e-06,
      "loss": 0.3993,
      "step": 17042
    },
    {
      "epoch": 0.7821118810518104,
      "grad_norm": 0.4614207148551941,
      "learning_rate": 8.58395755731839e-06,
      "loss": 0.4493,
      "step": 17043
    },
    {
      "epoch": 0.7821577715570649,
      "grad_norm": 0.4714962840080261,
      "learning_rate": 8.58378658653399e-06,
      "loss": 0.4075,
      "step": 17044
    },
    {
      "epoch": 0.7822036620623193,
      "grad_norm": 0.44587165117263794,
      "learning_rate": 8.58361560713171e-06,
      "loss": 0.39,
      "step": 17045
    },
    {
      "epoch": 0.7822495525675738,
      "grad_norm": 0.440456360578537,
      "learning_rate": 8.583444619111968e-06,
      "loss": 0.3782,
      "step": 17046
    },
    {
      "epoch": 0.7822954430728283,
      "grad_norm": 0.44569656252861023,
      "learning_rate": 8.583273622475173e-06,
      "loss": 0.3722,
      "step": 17047
    },
    {
      "epoch": 0.7823413335780827,
      "grad_norm": 0.4473586082458496,
      "learning_rate": 8.583102617221734e-06,
      "loss": 0.3615,
      "step": 17048
    },
    {
      "epoch": 0.7823872240833372,
      "grad_norm": 0.5035783052444458,
      "learning_rate": 8.582931603352067e-06,
      "loss": 0.4535,
      "step": 17049
    },
    {
      "epoch": 0.7824331145885917,
      "grad_norm": 0.4415551424026489,
      "learning_rate": 8.582760580866576e-06,
      "loss": 0.3573,
      "step": 17050
    },
    {
      "epoch": 0.782479005093846,
      "grad_norm": 0.43495893478393555,
      "learning_rate": 8.58258954976568e-06,
      "loss": 0.3558,
      "step": 17051
    },
    {
      "epoch": 0.7825248955991005,
      "grad_norm": 0.4600459635257721,
      "learning_rate": 8.582418510049787e-06,
      "loss": 0.4106,
      "step": 17052
    },
    {
      "epoch": 0.782570786104355,
      "grad_norm": 0.45710983872413635,
      "learning_rate": 8.582247461719307e-06,
      "loss": 0.3976,
      "step": 17053
    },
    {
      "epoch": 0.7826166766096094,
      "grad_norm": 0.43762823939323425,
      "learning_rate": 8.582076404774651e-06,
      "loss": 0.4038,
      "step": 17054
    },
    {
      "epoch": 0.7826625671148639,
      "grad_norm": 0.4170437753200531,
      "learning_rate": 8.581905339216235e-06,
      "loss": 0.2982,
      "step": 17055
    },
    {
      "epoch": 0.7827084576201184,
      "grad_norm": 0.4715779721736908,
      "learning_rate": 8.581734265044465e-06,
      "loss": 0.4138,
      "step": 17056
    },
    {
      "epoch": 0.7827543481253728,
      "grad_norm": 0.464697003364563,
      "learning_rate": 8.581563182259755e-06,
      "loss": 0.4222,
      "step": 17057
    },
    {
      "epoch": 0.7828002386306273,
      "grad_norm": 0.42772409319877625,
      "learning_rate": 8.581392090862515e-06,
      "loss": 0.3022,
      "step": 17058
    },
    {
      "epoch": 0.7828461291358818,
      "grad_norm": 0.42809921503067017,
      "learning_rate": 8.581220990853158e-06,
      "loss": 0.3617,
      "step": 17059
    },
    {
      "epoch": 0.7828920196411362,
      "grad_norm": 0.44953227043151855,
      "learning_rate": 8.581049882232095e-06,
      "loss": 0.4155,
      "step": 17060
    },
    {
      "epoch": 0.7829379101463907,
      "grad_norm": 0.47323521971702576,
      "learning_rate": 8.580878764999739e-06,
      "loss": 0.4827,
      "step": 17061
    },
    {
      "epoch": 0.7829838006516452,
      "grad_norm": 0.4372754991054535,
      "learning_rate": 8.580707639156496e-06,
      "loss": 0.4088,
      "step": 17062
    },
    {
      "epoch": 0.7830296911568997,
      "grad_norm": 0.41560351848602295,
      "learning_rate": 8.580536504702783e-06,
      "loss": 0.3393,
      "step": 17063
    },
    {
      "epoch": 0.7830755816621541,
      "grad_norm": 0.4736132323741913,
      "learning_rate": 8.58036536163901e-06,
      "loss": 0.449,
      "step": 17064
    },
    {
      "epoch": 0.7831214721674086,
      "grad_norm": 0.5012199878692627,
      "learning_rate": 8.580194209965588e-06,
      "loss": 0.451,
      "step": 17065
    },
    {
      "epoch": 0.7831673626726631,
      "grad_norm": 0.46233174204826355,
      "learning_rate": 8.580023049682931e-06,
      "loss": 0.4665,
      "step": 17066
    },
    {
      "epoch": 0.7832132531779175,
      "grad_norm": 0.43320024013519287,
      "learning_rate": 8.579851880791445e-06,
      "loss": 0.3723,
      "step": 17067
    },
    {
      "epoch": 0.783259143683172,
      "grad_norm": 0.47294214367866516,
      "learning_rate": 8.579680703291548e-06,
      "loss": 0.4725,
      "step": 17068
    },
    {
      "epoch": 0.7833050341884265,
      "grad_norm": 0.46606338024139404,
      "learning_rate": 8.579509517183646e-06,
      "loss": 0.4087,
      "step": 17069
    },
    {
      "epoch": 0.7833509246936808,
      "grad_norm": 0.474692702293396,
      "learning_rate": 8.579338322468155e-06,
      "loss": 0.3951,
      "step": 17070
    },
    {
      "epoch": 0.7833968151989353,
      "grad_norm": 0.42577216029167175,
      "learning_rate": 8.579167119145484e-06,
      "loss": 0.3369,
      "step": 17071
    },
    {
      "epoch": 0.7834427057041898,
      "grad_norm": 0.45509079098701477,
      "learning_rate": 8.578995907216048e-06,
      "loss": 0.404,
      "step": 17072
    },
    {
      "epoch": 0.7834885962094442,
      "grad_norm": 0.4243028461933136,
      "learning_rate": 8.578824686680255e-06,
      "loss": 0.3444,
      "step": 17073
    },
    {
      "epoch": 0.7835344867146987,
      "grad_norm": 0.4326368570327759,
      "learning_rate": 8.578653457538517e-06,
      "loss": 0.3511,
      "step": 17074
    },
    {
      "epoch": 0.7835803772199532,
      "grad_norm": 0.44456160068511963,
      "learning_rate": 8.578482219791248e-06,
      "loss": 0.3929,
      "step": 17075
    },
    {
      "epoch": 0.7836262677252076,
      "grad_norm": 0.4653504490852356,
      "learning_rate": 8.578310973438858e-06,
      "loss": 0.324,
      "step": 17076
    },
    {
      "epoch": 0.7836721582304621,
      "grad_norm": 0.4456257224082947,
      "learning_rate": 8.578139718481759e-06,
      "loss": 0.3969,
      "step": 17077
    },
    {
      "epoch": 0.7837180487357166,
      "grad_norm": 0.47462353110313416,
      "learning_rate": 8.577968454920365e-06,
      "loss": 0.4681,
      "step": 17078
    },
    {
      "epoch": 0.7837639392409711,
      "grad_norm": 0.41800087690353394,
      "learning_rate": 8.577797182755086e-06,
      "loss": 0.3125,
      "step": 17079
    },
    {
      "epoch": 0.7838098297462255,
      "grad_norm": 0.461171954870224,
      "learning_rate": 8.577625901986333e-06,
      "loss": 0.3905,
      "step": 17080
    },
    {
      "epoch": 0.78385572025148,
      "grad_norm": 0.6888251304626465,
      "learning_rate": 8.577454612614519e-06,
      "loss": 0.4178,
      "step": 17081
    },
    {
      "epoch": 0.7839016107567345,
      "grad_norm": 0.4391978681087494,
      "learning_rate": 8.577283314640055e-06,
      "loss": 0.3701,
      "step": 17082
    },
    {
      "epoch": 0.7839475012619889,
      "grad_norm": 0.475168377161026,
      "learning_rate": 8.577112008063355e-06,
      "loss": 0.4198,
      "step": 17083
    },
    {
      "epoch": 0.7839933917672434,
      "grad_norm": 0.46289145946502686,
      "learning_rate": 8.57694069288483e-06,
      "loss": 0.4213,
      "step": 17084
    },
    {
      "epoch": 0.7840392822724979,
      "grad_norm": 0.501196563243866,
      "learning_rate": 8.576769369104889e-06,
      "loss": 0.4319,
      "step": 17085
    },
    {
      "epoch": 0.7840851727777522,
      "grad_norm": 0.42122530937194824,
      "learning_rate": 8.576598036723948e-06,
      "loss": 0.3415,
      "step": 17086
    },
    {
      "epoch": 0.7841310632830067,
      "grad_norm": 0.42527833580970764,
      "learning_rate": 8.57642669574242e-06,
      "loss": 0.3299,
      "step": 17087
    },
    {
      "epoch": 0.7841769537882612,
      "grad_norm": 0.47453293204307556,
      "learning_rate": 8.57625534616071e-06,
      "loss": 0.4425,
      "step": 17088
    },
    {
      "epoch": 0.7842228442935156,
      "grad_norm": 0.4598444402217865,
      "learning_rate": 8.576083987979237e-06,
      "loss": 0.4105,
      "step": 17089
    },
    {
      "epoch": 0.7842687347987701,
      "grad_norm": 0.4670698940753937,
      "learning_rate": 8.575912621198412e-06,
      "loss": 0.4127,
      "step": 17090
    },
    {
      "epoch": 0.7843146253040246,
      "grad_norm": 0.5091941952705383,
      "learning_rate": 8.575741245818643e-06,
      "loss": 0.4805,
      "step": 17091
    },
    {
      "epoch": 0.784360515809279,
      "grad_norm": 0.47951096296310425,
      "learning_rate": 8.575569861840347e-06,
      "loss": 0.3716,
      "step": 17092
    },
    {
      "epoch": 0.7844064063145335,
      "grad_norm": 0.44542521238327026,
      "learning_rate": 8.57539846926393e-06,
      "loss": 0.3869,
      "step": 17093
    },
    {
      "epoch": 0.784452296819788,
      "grad_norm": 0.4719654619693756,
      "learning_rate": 8.575227068089813e-06,
      "loss": 0.4124,
      "step": 17094
    },
    {
      "epoch": 0.7844981873250424,
      "grad_norm": 0.41698333621025085,
      "learning_rate": 8.575055658318402e-06,
      "loss": 0.3454,
      "step": 17095
    },
    {
      "epoch": 0.7845440778302969,
      "grad_norm": 0.4496590495109558,
      "learning_rate": 8.57488423995011e-06,
      "loss": 0.3882,
      "step": 17096
    },
    {
      "epoch": 0.7845899683355514,
      "grad_norm": 0.45034921169281006,
      "learning_rate": 8.574712812985349e-06,
      "loss": 0.3333,
      "step": 17097
    },
    {
      "epoch": 0.7846358588408059,
      "grad_norm": 0.43507716059684753,
      "learning_rate": 8.574541377424532e-06,
      "loss": 0.3565,
      "step": 17098
    },
    {
      "epoch": 0.7846817493460603,
      "grad_norm": 0.4196447432041168,
      "learning_rate": 8.574369933268071e-06,
      "loss": 0.3203,
      "step": 17099
    },
    {
      "epoch": 0.7847276398513148,
      "grad_norm": 0.4530881345272064,
      "learning_rate": 8.57419848051638e-06,
      "loss": 0.3783,
      "step": 17100
    },
    {
      "epoch": 0.7847735303565693,
      "grad_norm": 0.4364980161190033,
      "learning_rate": 8.574027019169868e-06,
      "loss": 0.3728,
      "step": 17101
    },
    {
      "epoch": 0.7848194208618237,
      "grad_norm": 0.5109471082687378,
      "learning_rate": 8.57385554922895e-06,
      "loss": 0.4933,
      "step": 17102
    },
    {
      "epoch": 0.7848653113670782,
      "grad_norm": 0.46648159623146057,
      "learning_rate": 8.573684070694037e-06,
      "loss": 0.385,
      "step": 17103
    },
    {
      "epoch": 0.7849112018723327,
      "grad_norm": 0.47312548756599426,
      "learning_rate": 8.573512583565542e-06,
      "loss": 0.4222,
      "step": 17104
    },
    {
      "epoch": 0.784957092377587,
      "grad_norm": 0.467281311750412,
      "learning_rate": 8.573341087843876e-06,
      "loss": 0.4192,
      "step": 17105
    },
    {
      "epoch": 0.7850029828828415,
      "grad_norm": 0.42873871326446533,
      "learning_rate": 8.573169583529454e-06,
      "loss": 0.3433,
      "step": 17106
    },
    {
      "epoch": 0.785048873388096,
      "grad_norm": 0.43758949637413025,
      "learning_rate": 8.572998070622685e-06,
      "loss": 0.3839,
      "step": 17107
    },
    {
      "epoch": 0.7850947638933504,
      "grad_norm": 0.45014718174934387,
      "learning_rate": 8.572826549123985e-06,
      "loss": 0.3918,
      "step": 17108
    },
    {
      "epoch": 0.7851406543986049,
      "grad_norm": 0.46529287099838257,
      "learning_rate": 8.572655019033764e-06,
      "loss": 0.4469,
      "step": 17109
    },
    {
      "epoch": 0.7851865449038594,
      "grad_norm": 0.45789286494255066,
      "learning_rate": 8.572483480352437e-06,
      "loss": 0.3508,
      "step": 17110
    },
    {
      "epoch": 0.7852324354091138,
      "grad_norm": 0.45020776987075806,
      "learning_rate": 8.572311933080413e-06,
      "loss": 0.4032,
      "step": 17111
    },
    {
      "epoch": 0.7852783259143683,
      "grad_norm": 0.43734240531921387,
      "learning_rate": 8.572140377218105e-06,
      "loss": 0.426,
      "step": 17112
    },
    {
      "epoch": 0.7853242164196228,
      "grad_norm": 0.502072274684906,
      "learning_rate": 8.571968812765929e-06,
      "loss": 0.4527,
      "step": 17113
    },
    {
      "epoch": 0.7853701069248773,
      "grad_norm": 0.4603514075279236,
      "learning_rate": 8.571797239724295e-06,
      "loss": 0.4174,
      "step": 17114
    },
    {
      "epoch": 0.7854159974301317,
      "grad_norm": 0.4474254548549652,
      "learning_rate": 8.571625658093615e-06,
      "loss": 0.4367,
      "step": 17115
    },
    {
      "epoch": 0.7854618879353862,
      "grad_norm": 0.4662390649318695,
      "learning_rate": 8.571454067874304e-06,
      "loss": 0.4441,
      "step": 17116
    },
    {
      "epoch": 0.7855077784406407,
      "grad_norm": 0.4569355547428131,
      "learning_rate": 8.571282469066771e-06,
      "loss": 0.4148,
      "step": 17117
    },
    {
      "epoch": 0.7855536689458951,
      "grad_norm": 0.4233655631542206,
      "learning_rate": 8.571110861671432e-06,
      "loss": 0.3438,
      "step": 17118
    },
    {
      "epoch": 0.7855995594511496,
      "grad_norm": 0.4323183000087738,
      "learning_rate": 8.570939245688699e-06,
      "loss": 0.3684,
      "step": 17119
    },
    {
      "epoch": 0.7856454499564041,
      "grad_norm": 0.48957934975624084,
      "learning_rate": 8.570767621118983e-06,
      "loss": 0.4345,
      "step": 17120
    },
    {
      "epoch": 0.7856913404616584,
      "grad_norm": 0.4283546507358551,
      "learning_rate": 8.570595987962698e-06,
      "loss": 0.3838,
      "step": 17121
    },
    {
      "epoch": 0.785737230966913,
      "grad_norm": 0.47225281596183777,
      "learning_rate": 8.570424346220258e-06,
      "loss": 0.411,
      "step": 17122
    },
    {
      "epoch": 0.7857831214721674,
      "grad_norm": 0.4663572609424591,
      "learning_rate": 8.570252695892073e-06,
      "loss": 0.4315,
      "step": 17123
    },
    {
      "epoch": 0.7858290119774218,
      "grad_norm": 0.49042749404907227,
      "learning_rate": 8.570081036978557e-06,
      "loss": 0.453,
      "step": 17124
    },
    {
      "epoch": 0.7858749024826763,
      "grad_norm": 0.4535020589828491,
      "learning_rate": 8.569909369480122e-06,
      "loss": 0.4077,
      "step": 17125
    },
    {
      "epoch": 0.7859207929879308,
      "grad_norm": 0.4530177414417267,
      "learning_rate": 8.569737693397184e-06,
      "loss": 0.3506,
      "step": 17126
    },
    {
      "epoch": 0.7859666834931852,
      "grad_norm": 0.45346471667289734,
      "learning_rate": 8.569566008730152e-06,
      "loss": 0.3929,
      "step": 17127
    },
    {
      "epoch": 0.7860125739984397,
      "grad_norm": 0.44365355372428894,
      "learning_rate": 8.56939431547944e-06,
      "loss": 0.3639,
      "step": 17128
    },
    {
      "epoch": 0.7860584645036942,
      "grad_norm": 0.44638627767562866,
      "learning_rate": 8.569222613645463e-06,
      "loss": 0.3677,
      "step": 17129
    },
    {
      "epoch": 0.7861043550089486,
      "grad_norm": 0.4293568432331085,
      "learning_rate": 8.56905090322863e-06,
      "loss": 0.3366,
      "step": 17130
    },
    {
      "epoch": 0.7861502455142031,
      "grad_norm": 0.44036248326301575,
      "learning_rate": 8.568879184229357e-06,
      "loss": 0.393,
      "step": 17131
    },
    {
      "epoch": 0.7861961360194576,
      "grad_norm": 0.43053802847862244,
      "learning_rate": 8.568707456648056e-06,
      "loss": 0.3833,
      "step": 17132
    },
    {
      "epoch": 0.7862420265247121,
      "grad_norm": 0.5189328789710999,
      "learning_rate": 8.568535720485141e-06,
      "loss": 0.5208,
      "step": 17133
    },
    {
      "epoch": 0.7862879170299665,
      "grad_norm": 0.4671753942966461,
      "learning_rate": 8.568363975741023e-06,
      "loss": 0.3971,
      "step": 17134
    },
    {
      "epoch": 0.786333807535221,
      "grad_norm": 0.4216882288455963,
      "learning_rate": 8.568192222416114e-06,
      "loss": 0.3568,
      "step": 17135
    },
    {
      "epoch": 0.7863796980404755,
      "grad_norm": 0.4572247862815857,
      "learning_rate": 8.56802046051083e-06,
      "loss": 0.4232,
      "step": 17136
    },
    {
      "epoch": 0.7864255885457299,
      "grad_norm": 0.4463280141353607,
      "learning_rate": 8.567848690025584e-06,
      "loss": 0.3988,
      "step": 17137
    },
    {
      "epoch": 0.7864714790509844,
      "grad_norm": 0.44375041127204895,
      "learning_rate": 8.567676910960787e-06,
      "loss": 0.3999,
      "step": 17138
    },
    {
      "epoch": 0.7865173695562389,
      "grad_norm": 0.4713706970214844,
      "learning_rate": 8.567505123316854e-06,
      "loss": 0.4051,
      "step": 17139
    },
    {
      "epoch": 0.7865632600614932,
      "grad_norm": 0.4290619194507599,
      "learning_rate": 8.567333327094196e-06,
      "loss": 0.3728,
      "step": 17140
    },
    {
      "epoch": 0.7866091505667477,
      "grad_norm": 0.49214574694633484,
      "learning_rate": 8.567161522293227e-06,
      "loss": 0.4228,
      "step": 17141
    },
    {
      "epoch": 0.7866550410720022,
      "grad_norm": 0.47395938634872437,
      "learning_rate": 8.566989708914362e-06,
      "loss": 0.3957,
      "step": 17142
    },
    {
      "epoch": 0.7867009315772566,
      "grad_norm": 0.4636525809764862,
      "learning_rate": 8.566817886958012e-06,
      "loss": 0.385,
      "step": 17143
    },
    {
      "epoch": 0.7867468220825111,
      "grad_norm": 0.4787483513355255,
      "learning_rate": 8.56664605642459e-06,
      "loss": 0.5046,
      "step": 17144
    },
    {
      "epoch": 0.7867927125877656,
      "grad_norm": 0.4723854959011078,
      "learning_rate": 8.56647421731451e-06,
      "loss": 0.5408,
      "step": 17145
    },
    {
      "epoch": 0.78683860309302,
      "grad_norm": 0.4569069743156433,
      "learning_rate": 8.566302369628185e-06,
      "loss": 0.4194,
      "step": 17146
    },
    {
      "epoch": 0.7868844935982745,
      "grad_norm": 0.49809855222702026,
      "learning_rate": 8.566130513366028e-06,
      "loss": 0.4522,
      "step": 17147
    },
    {
      "epoch": 0.786930384103529,
      "grad_norm": 0.43556126952171326,
      "learning_rate": 8.565958648528454e-06,
      "loss": 0.404,
      "step": 17148
    },
    {
      "epoch": 0.7869762746087834,
      "grad_norm": 0.44776713848114014,
      "learning_rate": 8.565786775115874e-06,
      "loss": 0.4036,
      "step": 17149
    },
    {
      "epoch": 0.7870221651140379,
      "grad_norm": 0.447168231010437,
      "learning_rate": 8.565614893128702e-06,
      "loss": 0.3733,
      "step": 17150
    },
    {
      "epoch": 0.7870680556192924,
      "grad_norm": 0.4839850068092346,
      "learning_rate": 8.56544300256735e-06,
      "loss": 0.475,
      "step": 17151
    },
    {
      "epoch": 0.7871139461245469,
      "grad_norm": 0.5121183395385742,
      "learning_rate": 8.565271103432235e-06,
      "loss": 0.4356,
      "step": 17152
    },
    {
      "epoch": 0.7871598366298013,
      "grad_norm": 0.45978716015815735,
      "learning_rate": 8.56509919572377e-06,
      "loss": 0.3872,
      "step": 17153
    },
    {
      "epoch": 0.7872057271350558,
      "grad_norm": 0.48227548599243164,
      "learning_rate": 8.564927279442363e-06,
      "loss": 0.4932,
      "step": 17154
    },
    {
      "epoch": 0.7872516176403103,
      "grad_norm": 0.436812162399292,
      "learning_rate": 8.564755354588434e-06,
      "loss": 0.3784,
      "step": 17155
    },
    {
      "epoch": 0.7872975081455647,
      "grad_norm": 0.42787501215934753,
      "learning_rate": 8.56458342116239e-06,
      "loss": 0.3763,
      "step": 17156
    },
    {
      "epoch": 0.7873433986508191,
      "grad_norm": 0.4604097604751587,
      "learning_rate": 8.564411479164652e-06,
      "loss": 0.4596,
      "step": 17157
    },
    {
      "epoch": 0.7873892891560736,
      "grad_norm": 0.48435500264167786,
      "learning_rate": 8.564239528595625e-06,
      "loss": 0.4402,
      "step": 17158
    },
    {
      "epoch": 0.787435179661328,
      "grad_norm": 0.5035337805747986,
      "learning_rate": 8.56406756945573e-06,
      "loss": 0.5083,
      "step": 17159
    },
    {
      "epoch": 0.7874810701665825,
      "grad_norm": 0.4315187335014343,
      "learning_rate": 8.563895601745376e-06,
      "loss": 0.4164,
      "step": 17160
    },
    {
      "epoch": 0.787526960671837,
      "grad_norm": 0.4794437885284424,
      "learning_rate": 8.56372362546498e-06,
      "loss": 0.4233,
      "step": 17161
    },
    {
      "epoch": 0.7875728511770914,
      "grad_norm": 0.46684151887893677,
      "learning_rate": 8.56355164061495e-06,
      "loss": 0.437,
      "step": 17162
    },
    {
      "epoch": 0.7876187416823459,
      "grad_norm": 0.4438982605934143,
      "learning_rate": 8.563379647195704e-06,
      "loss": 0.4235,
      "step": 17163
    },
    {
      "epoch": 0.7876646321876004,
      "grad_norm": 0.4240376055240631,
      "learning_rate": 8.563207645207656e-06,
      "loss": 0.334,
      "step": 17164
    },
    {
      "epoch": 0.7877105226928548,
      "grad_norm": 0.44225120544433594,
      "learning_rate": 8.563035634651218e-06,
      "loss": 0.3766,
      "step": 17165
    },
    {
      "epoch": 0.7877564131981093,
      "grad_norm": 0.49483808875083923,
      "learning_rate": 8.562863615526802e-06,
      "loss": 0.487,
      "step": 17166
    },
    {
      "epoch": 0.7878023037033638,
      "grad_norm": 0.42341458797454834,
      "learning_rate": 8.562691587834827e-06,
      "loss": 0.3419,
      "step": 17167
    },
    {
      "epoch": 0.7878481942086183,
      "grad_norm": 0.4706038534641266,
      "learning_rate": 8.5625195515757e-06,
      "loss": 0.4684,
      "step": 17168
    },
    {
      "epoch": 0.7878940847138727,
      "grad_norm": 0.4092176556587219,
      "learning_rate": 8.562347506749839e-06,
      "loss": 0.3221,
      "step": 17169
    },
    {
      "epoch": 0.7879399752191272,
      "grad_norm": 0.44596561789512634,
      "learning_rate": 8.562175453357655e-06,
      "loss": 0.3862,
      "step": 17170
    },
    {
      "epoch": 0.7879858657243817,
      "grad_norm": 0.4571721851825714,
      "learning_rate": 8.562003391399565e-06,
      "loss": 0.4149,
      "step": 17171
    },
    {
      "epoch": 0.7880317562296361,
      "grad_norm": 0.409482479095459,
      "learning_rate": 8.56183132087598e-06,
      "loss": 0.3049,
      "step": 17172
    },
    {
      "epoch": 0.7880776467348906,
      "grad_norm": 0.4265441596508026,
      "learning_rate": 8.561659241787318e-06,
      "loss": 0.3053,
      "step": 17173
    },
    {
      "epoch": 0.7881235372401451,
      "grad_norm": 0.4258749485015869,
      "learning_rate": 8.561487154133984e-06,
      "loss": 0.3523,
      "step": 17174
    },
    {
      "epoch": 0.7881694277453994,
      "grad_norm": 0.4452821612358093,
      "learning_rate": 8.561315057916401e-06,
      "loss": 0.376,
      "step": 17175
    },
    {
      "epoch": 0.7882153182506539,
      "grad_norm": 0.4856232702732086,
      "learning_rate": 8.56114295313498e-06,
      "loss": 0.4267,
      "step": 17176
    },
    {
      "epoch": 0.7882612087559084,
      "grad_norm": 0.43713143467903137,
      "learning_rate": 8.560970839790132e-06,
      "loss": 0.3845,
      "step": 17177
    },
    {
      "epoch": 0.7883070992611628,
      "grad_norm": 0.5090425610542297,
      "learning_rate": 8.560798717882273e-06,
      "loss": 0.3699,
      "step": 17178
    },
    {
      "epoch": 0.7883529897664173,
      "grad_norm": 0.43923458456993103,
      "learning_rate": 8.56062658741182e-06,
      "loss": 0.3135,
      "step": 17179
    },
    {
      "epoch": 0.7883988802716718,
      "grad_norm": 0.49745458364486694,
      "learning_rate": 8.56045444837918e-06,
      "loss": 0.4978,
      "step": 17180
    },
    {
      "epoch": 0.7884447707769262,
      "grad_norm": 0.4496169686317444,
      "learning_rate": 8.560282300784773e-06,
      "loss": 0.3551,
      "step": 17181
    },
    {
      "epoch": 0.7884906612821807,
      "grad_norm": 0.4361990988254547,
      "learning_rate": 8.560110144629011e-06,
      "loss": 0.3779,
      "step": 17182
    },
    {
      "epoch": 0.7885365517874352,
      "grad_norm": 0.42884111404418945,
      "learning_rate": 8.559937979912307e-06,
      "loss": 0.3674,
      "step": 17183
    },
    {
      "epoch": 0.7885824422926896,
      "grad_norm": 0.45398902893066406,
      "learning_rate": 8.559765806635074e-06,
      "loss": 0.3516,
      "step": 17184
    },
    {
      "epoch": 0.7886283327979441,
      "grad_norm": 0.4343807101249695,
      "learning_rate": 8.559593624797729e-06,
      "loss": 0.3432,
      "step": 17185
    },
    {
      "epoch": 0.7886742233031986,
      "grad_norm": 0.43676406145095825,
      "learning_rate": 8.559421434400685e-06,
      "loss": 0.361,
      "step": 17186
    },
    {
      "epoch": 0.7887201138084531,
      "grad_norm": 0.43893498182296753,
      "learning_rate": 8.559249235444356e-06,
      "loss": 0.3728,
      "step": 17187
    },
    {
      "epoch": 0.7887660043137075,
      "grad_norm": 0.4559212625026703,
      "learning_rate": 8.559077027929158e-06,
      "loss": 0.3611,
      "step": 17188
    },
    {
      "epoch": 0.788811894818962,
      "grad_norm": 0.4440266788005829,
      "learning_rate": 8.558904811855499e-06,
      "loss": 0.4048,
      "step": 17189
    },
    {
      "epoch": 0.7888577853242165,
      "grad_norm": 0.47478795051574707,
      "learning_rate": 8.5587325872238e-06,
      "loss": 0.4537,
      "step": 17190
    },
    {
      "epoch": 0.7889036758294709,
      "grad_norm": 0.42379775643348694,
      "learning_rate": 8.558560354034471e-06,
      "loss": 0.379,
      "step": 17191
    },
    {
      "epoch": 0.7889495663347253,
      "grad_norm": 0.4595082700252533,
      "learning_rate": 8.558388112287927e-06,
      "loss": 0.3799,
      "step": 17192
    },
    {
      "epoch": 0.7889954568399798,
      "grad_norm": 0.4567592144012451,
      "learning_rate": 8.558215861984583e-06,
      "loss": 0.3912,
      "step": 17193
    },
    {
      "epoch": 0.7890413473452342,
      "grad_norm": 0.44178637862205505,
      "learning_rate": 8.558043603124853e-06,
      "loss": 0.3992,
      "step": 17194
    },
    {
      "epoch": 0.7890872378504887,
      "grad_norm": 0.470222145318985,
      "learning_rate": 8.557871335709151e-06,
      "loss": 0.4012,
      "step": 17195
    },
    {
      "epoch": 0.7891331283557432,
      "grad_norm": 0.47202497720718384,
      "learning_rate": 8.557699059737893e-06,
      "loss": 0.3892,
      "step": 17196
    },
    {
      "epoch": 0.7891790188609976,
      "grad_norm": 0.4622460603713989,
      "learning_rate": 8.557526775211489e-06,
      "loss": 0.4201,
      "step": 17197
    },
    {
      "epoch": 0.7892249093662521,
      "grad_norm": 0.5011447072029114,
      "learning_rate": 8.557354482130357e-06,
      "loss": 0.4721,
      "step": 17198
    },
    {
      "epoch": 0.7892707998715066,
      "grad_norm": 0.4705965518951416,
      "learning_rate": 8.557182180494912e-06,
      "loss": 0.3908,
      "step": 17199
    },
    {
      "epoch": 0.789316690376761,
      "grad_norm": 0.420924574136734,
      "learning_rate": 8.557009870305563e-06,
      "loss": 0.3252,
      "step": 17200
    },
    {
      "epoch": 0.7893625808820155,
      "grad_norm": 0.43872472643852234,
      "learning_rate": 8.556837551562731e-06,
      "loss": 0.3416,
      "step": 17201
    },
    {
      "epoch": 0.78940847138727,
      "grad_norm": 0.4418359100818634,
      "learning_rate": 8.556665224266826e-06,
      "loss": 0.3739,
      "step": 17202
    },
    {
      "epoch": 0.7894543618925245,
      "grad_norm": 0.4496125876903534,
      "learning_rate": 8.556492888418264e-06,
      "loss": 0.3944,
      "step": 17203
    },
    {
      "epoch": 0.7895002523977789,
      "grad_norm": 0.40348702669143677,
      "learning_rate": 8.556320544017459e-06,
      "loss": 0.2703,
      "step": 17204
    },
    {
      "epoch": 0.7895461429030334,
      "grad_norm": 0.45193061232566833,
      "learning_rate": 8.556148191064824e-06,
      "loss": 0.3613,
      "step": 17205
    },
    {
      "epoch": 0.7895920334082879,
      "grad_norm": 0.41576990485191345,
      "learning_rate": 8.555975829560776e-06,
      "loss": 0.3328,
      "step": 17206
    },
    {
      "epoch": 0.7896379239135423,
      "grad_norm": 0.4641425311565399,
      "learning_rate": 8.55580345950573e-06,
      "loss": 0.382,
      "step": 17207
    },
    {
      "epoch": 0.7896838144187968,
      "grad_norm": 0.4670281410217285,
      "learning_rate": 8.555631080900096e-06,
      "loss": 0.4256,
      "step": 17208
    },
    {
      "epoch": 0.7897297049240513,
      "grad_norm": 0.47999438643455505,
      "learning_rate": 8.555458693744293e-06,
      "loss": 0.3747,
      "step": 17209
    },
    {
      "epoch": 0.7897755954293056,
      "grad_norm": 0.4544064700603485,
      "learning_rate": 8.555286298038735e-06,
      "loss": 0.3791,
      "step": 17210
    },
    {
      "epoch": 0.7898214859345601,
      "grad_norm": 0.4362332224845886,
      "learning_rate": 8.555113893783834e-06,
      "loss": 0.4168,
      "step": 17211
    },
    {
      "epoch": 0.7898673764398146,
      "grad_norm": 0.4537018835544586,
      "learning_rate": 8.554941480980008e-06,
      "loss": 0.4548,
      "step": 17212
    },
    {
      "epoch": 0.789913266945069,
      "grad_norm": 0.4793187975883484,
      "learning_rate": 8.554769059627667e-06,
      "loss": 0.4001,
      "step": 17213
    },
    {
      "epoch": 0.7899591574503235,
      "grad_norm": 0.4365011751651764,
      "learning_rate": 8.55459662972723e-06,
      "loss": 0.3787,
      "step": 17214
    },
    {
      "epoch": 0.790005047955578,
      "grad_norm": 0.45444753766059875,
      "learning_rate": 8.55442419127911e-06,
      "loss": 0.414,
      "step": 17215
    },
    {
      "epoch": 0.7900509384608324,
      "grad_norm": 0.45154058933258057,
      "learning_rate": 8.55425174428372e-06,
      "loss": 0.4287,
      "step": 17216
    },
    {
      "epoch": 0.7900968289660869,
      "grad_norm": 0.4343357980251312,
      "learning_rate": 8.554079288741478e-06,
      "loss": 0.3431,
      "step": 17217
    },
    {
      "epoch": 0.7901427194713414,
      "grad_norm": 0.4607161581516266,
      "learning_rate": 8.553906824652796e-06,
      "loss": 0.4175,
      "step": 17218
    },
    {
      "epoch": 0.7901886099765958,
      "grad_norm": 0.4593159556388855,
      "learning_rate": 8.553734352018091e-06,
      "loss": 0.4183,
      "step": 17219
    },
    {
      "epoch": 0.7902345004818503,
      "grad_norm": 0.43572238087654114,
      "learning_rate": 8.553561870837777e-06,
      "loss": 0.3746,
      "step": 17220
    },
    {
      "epoch": 0.7902803909871048,
      "grad_norm": 0.6195092797279358,
      "learning_rate": 8.553389381112265e-06,
      "loss": 0.3037,
      "step": 17221
    },
    {
      "epoch": 0.7903262814923593,
      "grad_norm": 0.46369999647140503,
      "learning_rate": 8.553216882841975e-06,
      "loss": 0.3833,
      "step": 17222
    },
    {
      "epoch": 0.7903721719976137,
      "grad_norm": 0.4104870557785034,
      "learning_rate": 8.55304437602732e-06,
      "loss": 0.3304,
      "step": 17223
    },
    {
      "epoch": 0.7904180625028682,
      "grad_norm": 0.4783789813518524,
      "learning_rate": 8.552871860668714e-06,
      "loss": 0.4509,
      "step": 17224
    },
    {
      "epoch": 0.7904639530081227,
      "grad_norm": 0.45948266983032227,
      "learning_rate": 8.552699336766574e-06,
      "loss": 0.3993,
      "step": 17225
    },
    {
      "epoch": 0.790509843513377,
      "grad_norm": 0.5095104575157166,
      "learning_rate": 8.552526804321314e-06,
      "loss": 0.5109,
      "step": 17226
    },
    {
      "epoch": 0.7905557340186316,
      "grad_norm": 0.4878319203853607,
      "learning_rate": 8.552354263333345e-06,
      "loss": 0.4764,
      "step": 17227
    },
    {
      "epoch": 0.790601624523886,
      "grad_norm": 0.4699997901916504,
      "learning_rate": 8.552181713803086e-06,
      "loss": 0.466,
      "step": 17228
    },
    {
      "epoch": 0.7906475150291404,
      "grad_norm": 0.47339287400245667,
      "learning_rate": 8.552009155730953e-06,
      "loss": 0.3921,
      "step": 17229
    },
    {
      "epoch": 0.7906934055343949,
      "grad_norm": 0.4630031883716583,
      "learning_rate": 8.551836589117358e-06,
      "loss": 0.3882,
      "step": 17230
    },
    {
      "epoch": 0.7907392960396494,
      "grad_norm": 0.4879360496997833,
      "learning_rate": 8.551664013962718e-06,
      "loss": 0.4707,
      "step": 17231
    },
    {
      "epoch": 0.7907851865449038,
      "grad_norm": 0.39771568775177,
      "learning_rate": 8.551491430267445e-06,
      "loss": 0.321,
      "step": 17232
    },
    {
      "epoch": 0.7908310770501583,
      "grad_norm": 0.4232565462589264,
      "learning_rate": 8.551318838031955e-06,
      "loss": 0.3502,
      "step": 17233
    },
    {
      "epoch": 0.7908769675554128,
      "grad_norm": 0.48249053955078125,
      "learning_rate": 8.551146237256667e-06,
      "loss": 0.5121,
      "step": 17234
    },
    {
      "epoch": 0.7909228580606672,
      "grad_norm": 0.4547002911567688,
      "learning_rate": 8.550973627941991e-06,
      "loss": 0.4512,
      "step": 17235
    },
    {
      "epoch": 0.7909687485659217,
      "grad_norm": 0.4193003177642822,
      "learning_rate": 8.550801010088346e-06,
      "loss": 0.3754,
      "step": 17236
    },
    {
      "epoch": 0.7910146390711762,
      "grad_norm": 0.4411562383174896,
      "learning_rate": 8.550628383696144e-06,
      "loss": 0.3801,
      "step": 17237
    },
    {
      "epoch": 0.7910605295764306,
      "grad_norm": 0.43179214000701904,
      "learning_rate": 8.550455748765801e-06,
      "loss": 0.3409,
      "step": 17238
    },
    {
      "epoch": 0.7911064200816851,
      "grad_norm": 0.47141706943511963,
      "learning_rate": 8.550283105297734e-06,
      "loss": 0.4333,
      "step": 17239
    },
    {
      "epoch": 0.7911523105869396,
      "grad_norm": 0.4539576470851898,
      "learning_rate": 8.550110453292355e-06,
      "loss": 0.432,
      "step": 17240
    },
    {
      "epoch": 0.7911982010921941,
      "grad_norm": 0.47078731656074524,
      "learning_rate": 8.549937792750082e-06,
      "loss": 0.4216,
      "step": 17241
    },
    {
      "epoch": 0.7912440915974485,
      "grad_norm": 0.4110214412212372,
      "learning_rate": 8.549765123671328e-06,
      "loss": 0.3031,
      "step": 17242
    },
    {
      "epoch": 0.791289982102703,
      "grad_norm": 0.4198560416698456,
      "learning_rate": 8.54959244605651e-06,
      "loss": 0.351,
      "step": 17243
    },
    {
      "epoch": 0.7913358726079575,
      "grad_norm": 0.4529282748699188,
      "learning_rate": 8.549419759906042e-06,
      "loss": 0.3712,
      "step": 17244
    },
    {
      "epoch": 0.7913817631132118,
      "grad_norm": 0.4488847851753235,
      "learning_rate": 8.54924706522034e-06,
      "loss": 0.3676,
      "step": 17245
    },
    {
      "epoch": 0.7914276536184663,
      "grad_norm": 0.46307677030563354,
      "learning_rate": 8.54907436199982e-06,
      "loss": 0.4308,
      "step": 17246
    },
    {
      "epoch": 0.7914735441237208,
      "grad_norm": 0.45445457100868225,
      "learning_rate": 8.548901650244894e-06,
      "loss": 0.4041,
      "step": 17247
    },
    {
      "epoch": 0.7915194346289752,
      "grad_norm": 0.45837122201919556,
      "learning_rate": 8.548728929955982e-06,
      "loss": 0.4286,
      "step": 17248
    },
    {
      "epoch": 0.7915653251342297,
      "grad_norm": 0.45341235399246216,
      "learning_rate": 8.548556201133495e-06,
      "loss": 0.3607,
      "step": 17249
    },
    {
      "epoch": 0.7916112156394842,
      "grad_norm": 0.4551328122615814,
      "learning_rate": 8.548383463777852e-06,
      "loss": 0.4024,
      "step": 17250
    },
    {
      "epoch": 0.7916571061447386,
      "grad_norm": 0.4732138514518738,
      "learning_rate": 8.548210717889465e-06,
      "loss": 0.4792,
      "step": 17251
    },
    {
      "epoch": 0.7917029966499931,
      "grad_norm": 0.4484398663043976,
      "learning_rate": 8.548037963468752e-06,
      "loss": 0.4293,
      "step": 17252
    },
    {
      "epoch": 0.7917488871552476,
      "grad_norm": 0.4463064968585968,
      "learning_rate": 8.547865200516128e-06,
      "loss": 0.3447,
      "step": 17253
    },
    {
      "epoch": 0.791794777660502,
      "grad_norm": 0.4638603627681732,
      "learning_rate": 8.547692429032006e-06,
      "loss": 0.4079,
      "step": 17254
    },
    {
      "epoch": 0.7918406681657565,
      "grad_norm": 0.43883687257766724,
      "learning_rate": 8.547519649016805e-06,
      "loss": 0.3863,
      "step": 17255
    },
    {
      "epoch": 0.791886558671011,
      "grad_norm": 0.5217450261116028,
      "learning_rate": 8.547346860470938e-06,
      "loss": 0.4902,
      "step": 17256
    },
    {
      "epoch": 0.7919324491762655,
      "grad_norm": 0.43585386872291565,
      "learning_rate": 8.547174063394821e-06,
      "loss": 0.357,
      "step": 17257
    },
    {
      "epoch": 0.7919783396815199,
      "grad_norm": 0.4599902927875519,
      "learning_rate": 8.547001257788871e-06,
      "loss": 0.4322,
      "step": 17258
    },
    {
      "epoch": 0.7920242301867744,
      "grad_norm": 0.49865201115608215,
      "learning_rate": 8.546828443653504e-06,
      "loss": 0.4451,
      "step": 17259
    },
    {
      "epoch": 0.7920701206920289,
      "grad_norm": 0.4622938632965088,
      "learning_rate": 8.54665562098913e-06,
      "loss": 0.4438,
      "step": 17260
    },
    {
      "epoch": 0.7921160111972833,
      "grad_norm": 0.4822182059288025,
      "learning_rate": 8.546482789796172e-06,
      "loss": 0.5275,
      "step": 17261
    },
    {
      "epoch": 0.7921619017025378,
      "grad_norm": 0.4941704571247101,
      "learning_rate": 8.54630995007504e-06,
      "loss": 0.4894,
      "step": 17262
    },
    {
      "epoch": 0.7922077922077922,
      "grad_norm": 0.42387962341308594,
      "learning_rate": 8.546137101826154e-06,
      "loss": 0.3621,
      "step": 17263
    },
    {
      "epoch": 0.7922536827130466,
      "grad_norm": 0.44690072536468506,
      "learning_rate": 8.545964245049926e-06,
      "loss": 0.3939,
      "step": 17264
    },
    {
      "epoch": 0.7922995732183011,
      "grad_norm": 0.49550363421440125,
      "learning_rate": 8.545791379746774e-06,
      "loss": 0.4635,
      "step": 17265
    },
    {
      "epoch": 0.7923454637235556,
      "grad_norm": 0.48905831575393677,
      "learning_rate": 8.545618505917112e-06,
      "loss": 0.4749,
      "step": 17266
    },
    {
      "epoch": 0.79239135422881,
      "grad_norm": 0.40370485186576843,
      "learning_rate": 8.545445623561357e-06,
      "loss": 0.3018,
      "step": 17267
    },
    {
      "epoch": 0.7924372447340645,
      "grad_norm": 0.4208553731441498,
      "learning_rate": 8.545272732679923e-06,
      "loss": 0.3212,
      "step": 17268
    },
    {
      "epoch": 0.792483135239319,
      "grad_norm": 0.4605855643749237,
      "learning_rate": 8.545099833273229e-06,
      "loss": 0.4316,
      "step": 17269
    },
    {
      "epoch": 0.7925290257445734,
      "grad_norm": 0.4115740954875946,
      "learning_rate": 8.544926925341687e-06,
      "loss": 0.3111,
      "step": 17270
    },
    {
      "epoch": 0.7925749162498279,
      "grad_norm": 0.4296366274356842,
      "learning_rate": 8.544754008885715e-06,
      "loss": 0.3674,
      "step": 17271
    },
    {
      "epoch": 0.7926208067550824,
      "grad_norm": 0.44176560640335083,
      "learning_rate": 8.54458108390573e-06,
      "loss": 0.4243,
      "step": 17272
    },
    {
      "epoch": 0.7926666972603368,
      "grad_norm": 0.41267749667167664,
      "learning_rate": 8.544408150402144e-06,
      "loss": 0.3149,
      "step": 17273
    },
    {
      "epoch": 0.7927125877655913,
      "grad_norm": 0.46088534593582153,
      "learning_rate": 8.544235208375375e-06,
      "loss": 0.4384,
      "step": 17274
    },
    {
      "epoch": 0.7927584782708458,
      "grad_norm": 0.4485809803009033,
      "learning_rate": 8.544062257825841e-06,
      "loss": 0.3614,
      "step": 17275
    },
    {
      "epoch": 0.7928043687761003,
      "grad_norm": 0.46488314867019653,
      "learning_rate": 8.543889298753953e-06,
      "loss": 0.4268,
      "step": 17276
    },
    {
      "epoch": 0.7928502592813547,
      "grad_norm": 0.48688897490501404,
      "learning_rate": 8.543716331160133e-06,
      "loss": 0.2753,
      "step": 17277
    },
    {
      "epoch": 0.7928961497866092,
      "grad_norm": 0.6692848205566406,
      "learning_rate": 8.543543355044791e-06,
      "loss": 0.5327,
      "step": 17278
    },
    {
      "epoch": 0.7929420402918637,
      "grad_norm": 0.48614612221717834,
      "learning_rate": 8.543370370408346e-06,
      "loss": 0.4461,
      "step": 17279
    },
    {
      "epoch": 0.792987930797118,
      "grad_norm": 0.4368221163749695,
      "learning_rate": 8.543197377251215e-06,
      "loss": 0.3705,
      "step": 17280
    },
    {
      "epoch": 0.7930338213023725,
      "grad_norm": 0.45697784423828125,
      "learning_rate": 8.543024375573813e-06,
      "loss": 0.3852,
      "step": 17281
    },
    {
      "epoch": 0.793079711807627,
      "grad_norm": 0.47184309363365173,
      "learning_rate": 8.542851365376552e-06,
      "loss": 0.4475,
      "step": 17282
    },
    {
      "epoch": 0.7931256023128814,
      "grad_norm": 0.4400814473628998,
      "learning_rate": 8.542678346659855e-06,
      "loss": 0.4017,
      "step": 17283
    },
    {
      "epoch": 0.7931714928181359,
      "grad_norm": 0.45345747470855713,
      "learning_rate": 8.542505319424134e-06,
      "loss": 0.364,
      "step": 17284
    },
    {
      "epoch": 0.7932173833233904,
      "grad_norm": 0.46244215965270996,
      "learning_rate": 8.542332283669805e-06,
      "loss": 0.3801,
      "step": 17285
    },
    {
      "epoch": 0.7932632738286448,
      "grad_norm": 0.4394710659980774,
      "learning_rate": 8.542159239397285e-06,
      "loss": 0.3414,
      "step": 17286
    },
    {
      "epoch": 0.7933091643338993,
      "grad_norm": 0.45818227529525757,
      "learning_rate": 8.54198618660699e-06,
      "loss": 0.4087,
      "step": 17287
    },
    {
      "epoch": 0.7933550548391538,
      "grad_norm": 0.4504507780075073,
      "learning_rate": 8.541813125299337e-06,
      "loss": 0.4135,
      "step": 17288
    },
    {
      "epoch": 0.7934009453444082,
      "grad_norm": 0.4910365641117096,
      "learning_rate": 8.54164005547474e-06,
      "loss": 0.4444,
      "step": 17289
    },
    {
      "epoch": 0.7934468358496627,
      "grad_norm": 0.4514201283454895,
      "learning_rate": 8.541466977133616e-06,
      "loss": 0.4317,
      "step": 17290
    },
    {
      "epoch": 0.7934927263549172,
      "grad_norm": 0.42900437116622925,
      "learning_rate": 8.541293890276383e-06,
      "loss": 0.3609,
      "step": 17291
    },
    {
      "epoch": 0.7935386168601716,
      "grad_norm": 0.46192467212677,
      "learning_rate": 8.541120794903456e-06,
      "loss": 0.4148,
      "step": 17292
    },
    {
      "epoch": 0.7935845073654261,
      "grad_norm": 0.45300108194351196,
      "learning_rate": 8.540947691015252e-06,
      "loss": 0.4043,
      "step": 17293
    },
    {
      "epoch": 0.7936303978706806,
      "grad_norm": 0.43007004261016846,
      "learning_rate": 8.540774578612182e-06,
      "loss": 0.3495,
      "step": 17294
    },
    {
      "epoch": 0.7936762883759351,
      "grad_norm": 0.44749870896339417,
      "learning_rate": 8.540601457694672e-06,
      "loss": 0.3853,
      "step": 17295
    },
    {
      "epoch": 0.7937221788811895,
      "grad_norm": 0.4807637929916382,
      "learning_rate": 8.54042832826313e-06,
      "loss": 0.4808,
      "step": 17296
    },
    {
      "epoch": 0.793768069386444,
      "grad_norm": 0.459590345621109,
      "learning_rate": 8.540255190317977e-06,
      "loss": 0.3991,
      "step": 17297
    },
    {
      "epoch": 0.7938139598916985,
      "grad_norm": 0.4139643907546997,
      "learning_rate": 8.540082043859626e-06,
      "loss": 0.3383,
      "step": 17298
    },
    {
      "epoch": 0.7938598503969528,
      "grad_norm": 0.4161096513271332,
      "learning_rate": 8.539908888888495e-06,
      "loss": 0.3658,
      "step": 17299
    },
    {
      "epoch": 0.7939057409022073,
      "grad_norm": 0.44474828243255615,
      "learning_rate": 8.539735725405e-06,
      "loss": 0.3876,
      "step": 17300
    },
    {
      "epoch": 0.7939516314074618,
      "grad_norm": 0.44705772399902344,
      "learning_rate": 8.539562553409559e-06,
      "loss": 0.3861,
      "step": 17301
    },
    {
      "epoch": 0.7939975219127162,
      "grad_norm": 0.45543137192726135,
      "learning_rate": 8.539389372902587e-06,
      "loss": 0.4253,
      "step": 17302
    },
    {
      "epoch": 0.7940434124179707,
      "grad_norm": 0.4713234007358551,
      "learning_rate": 8.5392161838845e-06,
      "loss": 0.4405,
      "step": 17303
    },
    {
      "epoch": 0.7940893029232252,
      "grad_norm": 0.42139607667922974,
      "learning_rate": 8.539042986355714e-06,
      "loss": 0.3238,
      "step": 17304
    },
    {
      "epoch": 0.7941351934284796,
      "grad_norm": 0.47023651003837585,
      "learning_rate": 8.53886978031665e-06,
      "loss": 0.4378,
      "step": 17305
    },
    {
      "epoch": 0.7941810839337341,
      "grad_norm": 0.41005706787109375,
      "learning_rate": 8.538696565767718e-06,
      "loss": 0.3412,
      "step": 17306
    },
    {
      "epoch": 0.7942269744389886,
      "grad_norm": 0.4828278124332428,
      "learning_rate": 8.538523342709338e-06,
      "loss": 0.4553,
      "step": 17307
    },
    {
      "epoch": 0.794272864944243,
      "grad_norm": 0.4378403425216675,
      "learning_rate": 8.538350111141927e-06,
      "loss": 0.3604,
      "step": 17308
    },
    {
      "epoch": 0.7943187554494975,
      "grad_norm": 0.4598485827445984,
      "learning_rate": 8.538176871065901e-06,
      "loss": 0.3663,
      "step": 17309
    },
    {
      "epoch": 0.794364645954752,
      "grad_norm": 0.44187384843826294,
      "learning_rate": 8.538003622481675e-06,
      "loss": 0.3966,
      "step": 17310
    },
    {
      "epoch": 0.7944105364600065,
      "grad_norm": 0.44187048077583313,
      "learning_rate": 8.537830365389668e-06,
      "loss": 0.3433,
      "step": 17311
    },
    {
      "epoch": 0.7944564269652609,
      "grad_norm": 0.428903192281723,
      "learning_rate": 8.537657099790295e-06,
      "loss": 0.3552,
      "step": 17312
    },
    {
      "epoch": 0.7945023174705154,
      "grad_norm": 0.43505802750587463,
      "learning_rate": 8.537483825683974e-06,
      "loss": 0.3478,
      "step": 17313
    },
    {
      "epoch": 0.7945482079757699,
      "grad_norm": 0.45521458983421326,
      "learning_rate": 8.53731054307112e-06,
      "loss": 0.4305,
      "step": 17314
    },
    {
      "epoch": 0.7945940984810242,
      "grad_norm": 0.4717352092266083,
      "learning_rate": 8.537137251952152e-06,
      "loss": 0.4433,
      "step": 17315
    },
    {
      "epoch": 0.7946399889862787,
      "grad_norm": 0.4372889995574951,
      "learning_rate": 8.536963952327484e-06,
      "loss": 0.4333,
      "step": 17316
    },
    {
      "epoch": 0.7946858794915332,
      "grad_norm": 0.510814905166626,
      "learning_rate": 8.536790644197532e-06,
      "loss": 0.4286,
      "step": 17317
    },
    {
      "epoch": 0.7947317699967876,
      "grad_norm": 0.41214531660079956,
      "learning_rate": 8.536617327562718e-06,
      "loss": 0.3132,
      "step": 17318
    },
    {
      "epoch": 0.7947776605020421,
      "grad_norm": 0.43785569071769714,
      "learning_rate": 8.536444002423453e-06,
      "loss": 0.3895,
      "step": 17319
    },
    {
      "epoch": 0.7948235510072966,
      "grad_norm": 0.5033149719238281,
      "learning_rate": 8.53627066878016e-06,
      "loss": 0.4852,
      "step": 17320
    },
    {
      "epoch": 0.794869441512551,
      "grad_norm": 0.4413720369338989,
      "learning_rate": 8.53609732663325e-06,
      "loss": 0.371,
      "step": 17321
    },
    {
      "epoch": 0.7949153320178055,
      "grad_norm": 0.442349910736084,
      "learning_rate": 8.535923975983141e-06,
      "loss": 0.3539,
      "step": 17322
    },
    {
      "epoch": 0.79496122252306,
      "grad_norm": 0.497665673494339,
      "learning_rate": 8.535750616830252e-06,
      "loss": 0.4871,
      "step": 17323
    },
    {
      "epoch": 0.7950071130283144,
      "grad_norm": 0.44707366824150085,
      "learning_rate": 8.535577249174998e-06,
      "loss": 0.3719,
      "step": 17324
    },
    {
      "epoch": 0.7950530035335689,
      "grad_norm": 0.5090526938438416,
      "learning_rate": 8.535403873017795e-06,
      "loss": 0.4827,
      "step": 17325
    },
    {
      "epoch": 0.7950988940388234,
      "grad_norm": 0.4506312906742096,
      "learning_rate": 8.535230488359065e-06,
      "loss": 0.3841,
      "step": 17326
    },
    {
      "epoch": 0.7951447845440778,
      "grad_norm": 0.42831799387931824,
      "learning_rate": 8.53505709519922e-06,
      "loss": 0.3492,
      "step": 17327
    },
    {
      "epoch": 0.7951906750493323,
      "grad_norm": 0.4776977598667145,
      "learning_rate": 8.534883693538678e-06,
      "loss": 0.3997,
      "step": 17328
    },
    {
      "epoch": 0.7952365655545868,
      "grad_norm": 0.47641322016716003,
      "learning_rate": 8.534710283377856e-06,
      "loss": 0.4173,
      "step": 17329
    },
    {
      "epoch": 0.7952824560598413,
      "grad_norm": 0.45216673612594604,
      "learning_rate": 8.534536864717172e-06,
      "loss": 0.4064,
      "step": 17330
    },
    {
      "epoch": 0.7953283465650957,
      "grad_norm": 0.45900672674179077,
      "learning_rate": 8.534363437557042e-06,
      "loss": 0.395,
      "step": 17331
    },
    {
      "epoch": 0.7953742370703502,
      "grad_norm": 0.4505752921104431,
      "learning_rate": 8.534190001897884e-06,
      "loss": 0.4086,
      "step": 17332
    },
    {
      "epoch": 0.7954201275756047,
      "grad_norm": 0.46673840284347534,
      "learning_rate": 8.534016557740112e-06,
      "loss": 0.3915,
      "step": 17333
    },
    {
      "epoch": 0.795466018080859,
      "grad_norm": 0.46488118171691895,
      "learning_rate": 8.53384310508415e-06,
      "loss": 0.4727,
      "step": 17334
    },
    {
      "epoch": 0.7955119085861135,
      "grad_norm": 0.4449230432510376,
      "learning_rate": 8.533669643930407e-06,
      "loss": 0.3842,
      "step": 17335
    },
    {
      "epoch": 0.795557799091368,
      "grad_norm": 0.41877156496047974,
      "learning_rate": 8.533496174279305e-06,
      "loss": 0.3396,
      "step": 17336
    },
    {
      "epoch": 0.7956036895966224,
      "grad_norm": 0.46254199743270874,
      "learning_rate": 8.533322696131262e-06,
      "loss": 0.3944,
      "step": 17337
    },
    {
      "epoch": 0.7956495801018769,
      "grad_norm": 0.4546830654144287,
      "learning_rate": 8.533149209486688e-06,
      "loss": 0.4357,
      "step": 17338
    },
    {
      "epoch": 0.7956954706071314,
      "grad_norm": 0.46741804480552673,
      "learning_rate": 8.53297571434601e-06,
      "loss": 0.429,
      "step": 17339
    },
    {
      "epoch": 0.7957413611123858,
      "grad_norm": 0.43567103147506714,
      "learning_rate": 8.532802210709638e-06,
      "loss": 0.4027,
      "step": 17340
    },
    {
      "epoch": 0.7957872516176403,
      "grad_norm": 0.4721885919570923,
      "learning_rate": 8.532628698577992e-06,
      "loss": 0.4921,
      "step": 17341
    },
    {
      "epoch": 0.7958331421228948,
      "grad_norm": 0.5093852281570435,
      "learning_rate": 8.53245517795149e-06,
      "loss": 0.5067,
      "step": 17342
    },
    {
      "epoch": 0.7958790326281492,
      "grad_norm": 0.4458777904510498,
      "learning_rate": 8.532281648830547e-06,
      "loss": 0.3388,
      "step": 17343
    },
    {
      "epoch": 0.7959249231334037,
      "grad_norm": 0.4870307743549347,
      "learning_rate": 8.532108111215583e-06,
      "loss": 0.5027,
      "step": 17344
    },
    {
      "epoch": 0.7959708136386582,
      "grad_norm": 0.42581629753112793,
      "learning_rate": 8.531934565107011e-06,
      "loss": 0.3147,
      "step": 17345
    },
    {
      "epoch": 0.7960167041439127,
      "grad_norm": 0.4919399321079254,
      "learning_rate": 8.531761010505253e-06,
      "loss": 0.4224,
      "step": 17346
    },
    {
      "epoch": 0.7960625946491671,
      "grad_norm": 0.42880934476852417,
      "learning_rate": 8.531587447410725e-06,
      "loss": 0.3363,
      "step": 17347
    },
    {
      "epoch": 0.7961084851544216,
      "grad_norm": 0.4705013036727905,
      "learning_rate": 8.531413875823843e-06,
      "loss": 0.4625,
      "step": 17348
    },
    {
      "epoch": 0.7961543756596761,
      "grad_norm": 0.419408917427063,
      "learning_rate": 8.531240295745025e-06,
      "loss": 0.3152,
      "step": 17349
    },
    {
      "epoch": 0.7962002661649304,
      "grad_norm": 0.49158501625061035,
      "learning_rate": 8.53106670717469e-06,
      "loss": 0.4014,
      "step": 17350
    },
    {
      "epoch": 0.796246156670185,
      "grad_norm": 0.47131234407424927,
      "learning_rate": 8.530893110113252e-06,
      "loss": 0.3938,
      "step": 17351
    },
    {
      "epoch": 0.7962920471754394,
      "grad_norm": 0.46139469742774963,
      "learning_rate": 8.53071950456113e-06,
      "loss": 0.3707,
      "step": 17352
    },
    {
      "epoch": 0.7963379376806938,
      "grad_norm": 0.4659257233142853,
      "learning_rate": 8.530545890518743e-06,
      "loss": 0.4573,
      "step": 17353
    },
    {
      "epoch": 0.7963838281859483,
      "grad_norm": 0.4700642228126526,
      "learning_rate": 8.530372267986509e-06,
      "loss": 0.4123,
      "step": 17354
    },
    {
      "epoch": 0.7964297186912028,
      "grad_norm": 0.4003869593143463,
      "learning_rate": 8.530198636964842e-06,
      "loss": 0.3231,
      "step": 17355
    },
    {
      "epoch": 0.7964756091964572,
      "grad_norm": 0.48110300302505493,
      "learning_rate": 8.530024997454161e-06,
      "loss": 0.4564,
      "step": 17356
    },
    {
      "epoch": 0.7965214997017117,
      "grad_norm": 0.49582207202911377,
      "learning_rate": 8.529851349454885e-06,
      "loss": 0.4915,
      "step": 17357
    },
    {
      "epoch": 0.7965673902069662,
      "grad_norm": 0.4540678858757019,
      "learning_rate": 8.52967769296743e-06,
      "loss": 0.3496,
      "step": 17358
    },
    {
      "epoch": 0.7966132807122206,
      "grad_norm": 0.522850513458252,
      "learning_rate": 8.529504027992214e-06,
      "loss": 0.415,
      "step": 17359
    },
    {
      "epoch": 0.7966591712174751,
      "grad_norm": 0.4113677740097046,
      "learning_rate": 8.529330354529657e-06,
      "loss": 0.2884,
      "step": 17360
    },
    {
      "epoch": 0.7967050617227296,
      "grad_norm": 0.4390864968299866,
      "learning_rate": 8.529156672580172e-06,
      "loss": 0.3632,
      "step": 17361
    },
    {
      "epoch": 0.796750952227984,
      "grad_norm": 0.47810623049736023,
      "learning_rate": 8.52898298214418e-06,
      "loss": 0.4182,
      "step": 17362
    },
    {
      "epoch": 0.7967968427332385,
      "grad_norm": 0.48694515228271484,
      "learning_rate": 8.528809283222097e-06,
      "loss": 0.4599,
      "step": 17363
    },
    {
      "epoch": 0.796842733238493,
      "grad_norm": 0.44697320461273193,
      "learning_rate": 8.528635575814342e-06,
      "loss": 0.3668,
      "step": 17364
    },
    {
      "epoch": 0.7968886237437475,
      "grad_norm": 0.4329720735549927,
      "learning_rate": 8.528461859921333e-06,
      "loss": 0.3807,
      "step": 17365
    },
    {
      "epoch": 0.7969345142490019,
      "grad_norm": 0.460861474275589,
      "learning_rate": 8.528288135543484e-06,
      "loss": 0.4023,
      "step": 17366
    },
    {
      "epoch": 0.7969804047542564,
      "grad_norm": 0.45921745896339417,
      "learning_rate": 8.528114402681219e-06,
      "loss": 0.4221,
      "step": 17367
    },
    {
      "epoch": 0.7970262952595109,
      "grad_norm": 0.5792778730392456,
      "learning_rate": 8.52794066133495e-06,
      "loss": 0.4293,
      "step": 17368
    },
    {
      "epoch": 0.7970721857647652,
      "grad_norm": 0.4442204535007477,
      "learning_rate": 8.527766911505097e-06,
      "loss": 0.3727,
      "step": 17369
    },
    {
      "epoch": 0.7971180762700197,
      "grad_norm": 0.4427334666252136,
      "learning_rate": 8.52759315319208e-06,
      "loss": 0.3318,
      "step": 17370
    },
    {
      "epoch": 0.7971639667752742,
      "grad_norm": 0.44364240765571594,
      "learning_rate": 8.527419386396313e-06,
      "loss": 0.4154,
      "step": 17371
    },
    {
      "epoch": 0.7972098572805286,
      "grad_norm": 0.45921948552131653,
      "learning_rate": 8.527245611118215e-06,
      "loss": 0.451,
      "step": 17372
    },
    {
      "epoch": 0.7972557477857831,
      "grad_norm": 0.475739985704422,
      "learning_rate": 8.527071827358207e-06,
      "loss": 0.4503,
      "step": 17373
    },
    {
      "epoch": 0.7973016382910376,
      "grad_norm": 0.4728918969631195,
      "learning_rate": 8.526898035116704e-06,
      "loss": 0.4576,
      "step": 17374
    },
    {
      "epoch": 0.797347528796292,
      "grad_norm": 0.4763958156108856,
      "learning_rate": 8.526724234394123e-06,
      "loss": 0.4421,
      "step": 17375
    },
    {
      "epoch": 0.7973934193015465,
      "grad_norm": 0.4438845217227936,
      "learning_rate": 8.526550425190886e-06,
      "loss": 0.3751,
      "step": 17376
    },
    {
      "epoch": 0.797439309806801,
      "grad_norm": 0.46401268243789673,
      "learning_rate": 8.526376607507405e-06,
      "loss": 0.4364,
      "step": 17377
    },
    {
      "epoch": 0.7974852003120554,
      "grad_norm": 0.5637767910957336,
      "learning_rate": 8.526202781344103e-06,
      "loss": 0.4913,
      "step": 17378
    },
    {
      "epoch": 0.7975310908173099,
      "grad_norm": 0.4584812819957733,
      "learning_rate": 8.526028946701394e-06,
      "loss": 0.4298,
      "step": 17379
    },
    {
      "epoch": 0.7975769813225644,
      "grad_norm": 0.463896244764328,
      "learning_rate": 8.5258551035797e-06,
      "loss": 0.3945,
      "step": 17380
    },
    {
      "epoch": 0.7976228718278188,
      "grad_norm": 0.45107871294021606,
      "learning_rate": 8.525681251979437e-06,
      "loss": 0.4025,
      "step": 17381
    },
    {
      "epoch": 0.7976687623330733,
      "grad_norm": 0.45406076312065125,
      "learning_rate": 8.525507391901024e-06,
      "loss": 0.4273,
      "step": 17382
    },
    {
      "epoch": 0.7977146528383278,
      "grad_norm": 0.4512777626514435,
      "learning_rate": 8.525333523344878e-06,
      "loss": 0.3852,
      "step": 17383
    },
    {
      "epoch": 0.7977605433435823,
      "grad_norm": 0.4962588846683502,
      "learning_rate": 8.525159646311416e-06,
      "loss": 0.4518,
      "step": 17384
    },
    {
      "epoch": 0.7978064338488366,
      "grad_norm": 0.43524184823036194,
      "learning_rate": 8.524985760801059e-06,
      "loss": 0.3747,
      "step": 17385
    },
    {
      "epoch": 0.7978523243540911,
      "grad_norm": 0.4741273522377014,
      "learning_rate": 8.524811866814225e-06,
      "loss": 0.4351,
      "step": 17386
    },
    {
      "epoch": 0.7978982148593456,
      "grad_norm": 0.4265655279159546,
      "learning_rate": 8.524637964351328e-06,
      "loss": 0.3562,
      "step": 17387
    },
    {
      "epoch": 0.7979441053646,
      "grad_norm": 0.4441221356391907,
      "learning_rate": 8.52446405341279e-06,
      "loss": 0.3676,
      "step": 17388
    },
    {
      "epoch": 0.7979899958698545,
      "grad_norm": 0.47150611877441406,
      "learning_rate": 8.524290133999032e-06,
      "loss": 0.4396,
      "step": 17389
    },
    {
      "epoch": 0.798035886375109,
      "grad_norm": 0.4297012984752655,
      "learning_rate": 8.524116206110464e-06,
      "loss": 0.3181,
      "step": 17390
    },
    {
      "epoch": 0.7980817768803634,
      "grad_norm": 0.48709699511528015,
      "learning_rate": 8.52394226974751e-06,
      "loss": 0.4649,
      "step": 17391
    },
    {
      "epoch": 0.7981276673856179,
      "grad_norm": 0.45016026496887207,
      "learning_rate": 8.523768324910589e-06,
      "loss": 0.3667,
      "step": 17392
    },
    {
      "epoch": 0.7981735578908724,
      "grad_norm": 0.46889638900756836,
      "learning_rate": 8.523594371600114e-06,
      "loss": 0.4093,
      "step": 17393
    },
    {
      "epoch": 0.7982194483961268,
      "grad_norm": 0.4164358973503113,
      "learning_rate": 8.52342040981651e-06,
      "loss": 0.2915,
      "step": 17394
    },
    {
      "epoch": 0.7982653389013813,
      "grad_norm": 0.45595481991767883,
      "learning_rate": 8.523246439560188e-06,
      "loss": 0.3906,
      "step": 17395
    },
    {
      "epoch": 0.7983112294066358,
      "grad_norm": 0.4730428457260132,
      "learning_rate": 8.523072460831575e-06,
      "loss": 0.3975,
      "step": 17396
    },
    {
      "epoch": 0.7983571199118902,
      "grad_norm": 0.5198157429695129,
      "learning_rate": 8.52289847363108e-06,
      "loss": 0.5012,
      "step": 17397
    },
    {
      "epoch": 0.7984030104171447,
      "grad_norm": 0.487239807844162,
      "learning_rate": 8.522724477959128e-06,
      "loss": 0.4577,
      "step": 17398
    },
    {
      "epoch": 0.7984489009223992,
      "grad_norm": 0.44806692004203796,
      "learning_rate": 8.522550473816136e-06,
      "loss": 0.3806,
      "step": 17399
    },
    {
      "epoch": 0.7984947914276537,
      "grad_norm": 0.5005094408988953,
      "learning_rate": 8.522376461202522e-06,
      "loss": 0.4376,
      "step": 17400
    },
    {
      "epoch": 0.7985406819329081,
      "grad_norm": 0.45714861154556274,
      "learning_rate": 8.522202440118704e-06,
      "loss": 0.4746,
      "step": 17401
    },
    {
      "epoch": 0.7985865724381626,
      "grad_norm": 0.45959559082984924,
      "learning_rate": 8.5220284105651e-06,
      "loss": 0.436,
      "step": 17402
    },
    {
      "epoch": 0.798632462943417,
      "grad_norm": 0.3999626934528351,
      "learning_rate": 8.52185437254213e-06,
      "loss": 0.3271,
      "step": 17403
    },
    {
      "epoch": 0.7986783534486714,
      "grad_norm": 0.44875892996788025,
      "learning_rate": 8.521680326050212e-06,
      "loss": 0.3785,
      "step": 17404
    },
    {
      "epoch": 0.7987242439539259,
      "grad_norm": 0.5110406279563904,
      "learning_rate": 8.521506271089764e-06,
      "loss": 0.4941,
      "step": 17405
    },
    {
      "epoch": 0.7987701344591804,
      "grad_norm": 0.4528273642063141,
      "learning_rate": 8.521332207661203e-06,
      "loss": 0.3241,
      "step": 17406
    },
    {
      "epoch": 0.7988160249644348,
      "grad_norm": 0.477131187915802,
      "learning_rate": 8.521158135764954e-06,
      "loss": 0.3722,
      "step": 17407
    },
    {
      "epoch": 0.7988619154696893,
      "grad_norm": 0.456575870513916,
      "learning_rate": 8.520984055401426e-06,
      "loss": 0.3487,
      "step": 17408
    },
    {
      "epoch": 0.7989078059749438,
      "grad_norm": 0.42182672023773193,
      "learning_rate": 8.520809966571044e-06,
      "loss": 0.3185,
      "step": 17409
    },
    {
      "epoch": 0.7989536964801982,
      "grad_norm": 0.42592522501945496,
      "learning_rate": 8.520635869274227e-06,
      "loss": 0.3615,
      "step": 17410
    },
    {
      "epoch": 0.7989995869854527,
      "grad_norm": 0.518481433391571,
      "learning_rate": 8.520461763511391e-06,
      "loss": 0.5343,
      "step": 17411
    },
    {
      "epoch": 0.7990454774907072,
      "grad_norm": 0.4880763590335846,
      "learning_rate": 8.520287649282955e-06,
      "loss": 0.4196,
      "step": 17412
    },
    {
      "epoch": 0.7990913679959616,
      "grad_norm": 0.4802713692188263,
      "learning_rate": 8.52011352658934e-06,
      "loss": 0.4797,
      "step": 17413
    },
    {
      "epoch": 0.7991372585012161,
      "grad_norm": 0.46200743317604065,
      "learning_rate": 8.519939395430961e-06,
      "loss": 0.4293,
      "step": 17414
    },
    {
      "epoch": 0.7991831490064706,
      "grad_norm": 0.46974611282348633,
      "learning_rate": 8.51976525580824e-06,
      "loss": 0.4038,
      "step": 17415
    },
    {
      "epoch": 0.799229039511725,
      "grad_norm": 0.44320976734161377,
      "learning_rate": 8.519591107721592e-06,
      "loss": 0.4185,
      "step": 17416
    },
    {
      "epoch": 0.7992749300169795,
      "grad_norm": 0.45917925238609314,
      "learning_rate": 8.519416951171439e-06,
      "loss": 0.4233,
      "step": 17417
    },
    {
      "epoch": 0.799320820522234,
      "grad_norm": 0.43788355588912964,
      "learning_rate": 8.519242786158202e-06,
      "loss": 0.3756,
      "step": 17418
    },
    {
      "epoch": 0.7993667110274885,
      "grad_norm": 0.48166245222091675,
      "learning_rate": 8.519068612682294e-06,
      "loss": 0.404,
      "step": 17419
    },
    {
      "epoch": 0.7994126015327429,
      "grad_norm": 0.4552880525588989,
      "learning_rate": 8.518894430744137e-06,
      "loss": 0.3964,
      "step": 17420
    },
    {
      "epoch": 0.7994584920379973,
      "grad_norm": 0.4498491585254669,
      "learning_rate": 8.518720240344149e-06,
      "loss": 0.3724,
      "step": 17421
    },
    {
      "epoch": 0.7995043825432518,
      "grad_norm": 0.44619670510292053,
      "learning_rate": 8.518546041482751e-06,
      "loss": 0.4167,
      "step": 17422
    },
    {
      "epoch": 0.7995502730485062,
      "grad_norm": 0.43638819456100464,
      "learning_rate": 8.51837183416036e-06,
      "loss": 0.3551,
      "step": 17423
    },
    {
      "epoch": 0.7995961635537607,
      "grad_norm": 0.42472153902053833,
      "learning_rate": 8.518197618377393e-06,
      "loss": 0.3617,
      "step": 17424
    },
    {
      "epoch": 0.7996420540590152,
      "grad_norm": 0.4819573760032654,
      "learning_rate": 8.518023394134275e-06,
      "loss": 0.4802,
      "step": 17425
    },
    {
      "epoch": 0.7996879445642696,
      "grad_norm": 0.4771646559238434,
      "learning_rate": 8.517849161431418e-06,
      "loss": 0.5027,
      "step": 17426
    },
    {
      "epoch": 0.7997338350695241,
      "grad_norm": 0.457241952419281,
      "learning_rate": 8.517674920269244e-06,
      "loss": 0.3518,
      "step": 17427
    },
    {
      "epoch": 0.7997797255747786,
      "grad_norm": 0.48232996463775635,
      "learning_rate": 8.517500670648173e-06,
      "loss": 0.4211,
      "step": 17428
    },
    {
      "epoch": 0.799825616080033,
      "grad_norm": 0.4540688097476959,
      "learning_rate": 8.517326412568624e-06,
      "loss": 0.4017,
      "step": 17429
    },
    {
      "epoch": 0.7998715065852875,
      "grad_norm": 0.45765626430511475,
      "learning_rate": 8.517152146031012e-06,
      "loss": 0.3539,
      "step": 17430
    },
    {
      "epoch": 0.799917397090542,
      "grad_norm": 0.4929465353488922,
      "learning_rate": 8.51697787103576e-06,
      "loss": 0.4819,
      "step": 17431
    },
    {
      "epoch": 0.7999632875957964,
      "grad_norm": 0.46364501118659973,
      "learning_rate": 8.516803587583289e-06,
      "loss": 0.4479,
      "step": 17432
    },
    {
      "epoch": 0.8000091781010509,
      "grad_norm": 0.5282514095306396,
      "learning_rate": 8.516629295674015e-06,
      "loss": 0.4615,
      "step": 17433
    },
    {
      "epoch": 0.8000550686063054,
      "grad_norm": 0.4677409827709198,
      "learning_rate": 8.516454995308355e-06,
      "loss": 0.349,
      "step": 17434
    },
    {
      "epoch": 0.8001009591115599,
      "grad_norm": 0.44146832823753357,
      "learning_rate": 8.516280686486731e-06,
      "loss": 0.3271,
      "step": 17435
    },
    {
      "epoch": 0.8001468496168143,
      "grad_norm": 0.49169692397117615,
      "learning_rate": 8.516106369209562e-06,
      "loss": 0.4626,
      "step": 17436
    },
    {
      "epoch": 0.8001927401220688,
      "grad_norm": 0.4435870945453644,
      "learning_rate": 8.515932043477268e-06,
      "loss": 0.4047,
      "step": 17437
    },
    {
      "epoch": 0.8002386306273233,
      "grad_norm": 0.4633375108242035,
      "learning_rate": 8.515757709290265e-06,
      "loss": 0.4493,
      "step": 17438
    },
    {
      "epoch": 0.8002845211325776,
      "grad_norm": 0.4477768540382385,
      "learning_rate": 8.515583366648976e-06,
      "loss": 0.4031,
      "step": 17439
    },
    {
      "epoch": 0.8003304116378321,
      "grad_norm": 0.4911952316761017,
      "learning_rate": 8.515409015553818e-06,
      "loss": 0.4519,
      "step": 17440
    },
    {
      "epoch": 0.8003763021430866,
      "grad_norm": 0.4222570061683655,
      "learning_rate": 8.51523465600521e-06,
      "loss": 0.3052,
      "step": 17441
    },
    {
      "epoch": 0.800422192648341,
      "grad_norm": 0.44558194279670715,
      "learning_rate": 8.515060288003574e-06,
      "loss": 0.3944,
      "step": 17442
    },
    {
      "epoch": 0.8004680831535955,
      "grad_norm": 0.4427376687526703,
      "learning_rate": 8.514885911549327e-06,
      "loss": 0.3586,
      "step": 17443
    },
    {
      "epoch": 0.80051397365885,
      "grad_norm": 0.4829748272895813,
      "learning_rate": 8.514711526642887e-06,
      "loss": 0.4635,
      "step": 17444
    },
    {
      "epoch": 0.8005598641641044,
      "grad_norm": 0.47320976853370667,
      "learning_rate": 8.514537133284677e-06,
      "loss": 0.4464,
      "step": 17445
    },
    {
      "epoch": 0.8006057546693589,
      "grad_norm": 0.4252266585826874,
      "learning_rate": 8.514362731475112e-06,
      "loss": 0.3229,
      "step": 17446
    },
    {
      "epoch": 0.8006516451746134,
      "grad_norm": 0.4535079598426819,
      "learning_rate": 8.514188321214617e-06,
      "loss": 0.4452,
      "step": 17447
    },
    {
      "epoch": 0.8006975356798678,
      "grad_norm": 0.43816477060317993,
      "learning_rate": 8.514013902503606e-06,
      "loss": 0.3239,
      "step": 17448
    },
    {
      "epoch": 0.8007434261851223,
      "grad_norm": 0.44645780324935913,
      "learning_rate": 8.513839475342501e-06,
      "loss": 0.3695,
      "step": 17449
    },
    {
      "epoch": 0.8007893166903768,
      "grad_norm": 0.46480223536491394,
      "learning_rate": 8.513665039731722e-06,
      "loss": 0.442,
      "step": 17450
    },
    {
      "epoch": 0.8008352071956312,
      "grad_norm": 0.44286370277404785,
      "learning_rate": 8.513490595671687e-06,
      "loss": 0.3446,
      "step": 17451
    },
    {
      "epoch": 0.8008810977008857,
      "grad_norm": 0.44705668091773987,
      "learning_rate": 8.513316143162816e-06,
      "loss": 0.3569,
      "step": 17452
    },
    {
      "epoch": 0.8009269882061402,
      "grad_norm": 0.44909295439720154,
      "learning_rate": 8.513141682205528e-06,
      "loss": 0.3953,
      "step": 17453
    },
    {
      "epoch": 0.8009728787113947,
      "grad_norm": 0.4579792320728302,
      "learning_rate": 8.512967212800243e-06,
      "loss": 0.4125,
      "step": 17454
    },
    {
      "epoch": 0.801018769216649,
      "grad_norm": 0.47918441891670227,
      "learning_rate": 8.51279273494738e-06,
      "loss": 0.4624,
      "step": 17455
    },
    {
      "epoch": 0.8010646597219035,
      "grad_norm": 0.4858976900577545,
      "learning_rate": 8.51261824864736e-06,
      "loss": 0.4731,
      "step": 17456
    },
    {
      "epoch": 0.801110550227158,
      "grad_norm": 0.4246160686016083,
      "learning_rate": 8.512443753900603e-06,
      "loss": 0.3131,
      "step": 17457
    },
    {
      "epoch": 0.8011564407324124,
      "grad_norm": 0.46327030658721924,
      "learning_rate": 8.512269250707526e-06,
      "loss": 0.4237,
      "step": 17458
    },
    {
      "epoch": 0.8012023312376669,
      "grad_norm": 0.4327425956726074,
      "learning_rate": 8.51209473906855e-06,
      "loss": 0.3592,
      "step": 17459
    },
    {
      "epoch": 0.8012482217429214,
      "grad_norm": 0.4413203299045563,
      "learning_rate": 8.511920218984094e-06,
      "loss": 0.3996,
      "step": 17460
    },
    {
      "epoch": 0.8012941122481758,
      "grad_norm": 0.4644427001476288,
      "learning_rate": 8.51174569045458e-06,
      "loss": 0.4113,
      "step": 17461
    },
    {
      "epoch": 0.8013400027534303,
      "grad_norm": 0.46664342284202576,
      "learning_rate": 8.511571153480424e-06,
      "loss": 0.3955,
      "step": 17462
    },
    {
      "epoch": 0.8013858932586848,
      "grad_norm": 0.4471636414527893,
      "learning_rate": 8.511396608062048e-06,
      "loss": 0.4296,
      "step": 17463
    },
    {
      "epoch": 0.8014317837639392,
      "grad_norm": 0.47241243720054626,
      "learning_rate": 8.511222054199872e-06,
      "loss": 0.4212,
      "step": 17464
    },
    {
      "epoch": 0.8014776742691937,
      "grad_norm": 0.4173448383808136,
      "learning_rate": 8.511047491894315e-06,
      "loss": 0.3196,
      "step": 17465
    },
    {
      "epoch": 0.8015235647744482,
      "grad_norm": 0.4807930588722229,
      "learning_rate": 8.510872921145798e-06,
      "loss": 0.4352,
      "step": 17466
    },
    {
      "epoch": 0.8015694552797026,
      "grad_norm": 0.5392604470252991,
      "learning_rate": 8.51069834195474e-06,
      "loss": 0.4899,
      "step": 17467
    },
    {
      "epoch": 0.8016153457849571,
      "grad_norm": 0.4549521803855896,
      "learning_rate": 8.510523754321558e-06,
      "loss": 0.3699,
      "step": 17468
    },
    {
      "epoch": 0.8016612362902116,
      "grad_norm": 0.4334564805030823,
      "learning_rate": 8.510349158246676e-06,
      "loss": 0.3577,
      "step": 17469
    },
    {
      "epoch": 0.801707126795466,
      "grad_norm": 0.43394628167152405,
      "learning_rate": 8.51017455373051e-06,
      "loss": 0.3687,
      "step": 17470
    },
    {
      "epoch": 0.8017530173007205,
      "grad_norm": 0.4640337824821472,
      "learning_rate": 8.509999940773484e-06,
      "loss": 0.4002,
      "step": 17471
    },
    {
      "epoch": 0.801798907805975,
      "grad_norm": 0.47587671875953674,
      "learning_rate": 8.509825319376017e-06,
      "loss": 0.4445,
      "step": 17472
    },
    {
      "epoch": 0.8018447983112295,
      "grad_norm": 0.44264906644821167,
      "learning_rate": 8.509650689538527e-06,
      "loss": 0.3228,
      "step": 17473
    },
    {
      "epoch": 0.8018906888164838,
      "grad_norm": 0.5028887391090393,
      "learning_rate": 8.509476051261433e-06,
      "loss": 0.4662,
      "step": 17474
    },
    {
      "epoch": 0.8019365793217383,
      "grad_norm": 0.47195491194725037,
      "learning_rate": 8.509301404545158e-06,
      "loss": 0.4313,
      "step": 17475
    },
    {
      "epoch": 0.8019824698269928,
      "grad_norm": 0.5014392137527466,
      "learning_rate": 8.50912674939012e-06,
      "loss": 0.3277,
      "step": 17476
    },
    {
      "epoch": 0.8020283603322472,
      "grad_norm": 0.4536079168319702,
      "learning_rate": 8.50895208579674e-06,
      "loss": 0.3755,
      "step": 17477
    },
    {
      "epoch": 0.8020742508375017,
      "grad_norm": 0.46835607290267944,
      "learning_rate": 8.508777413765438e-06,
      "loss": 0.4183,
      "step": 17478
    },
    {
      "epoch": 0.8021201413427562,
      "grad_norm": 0.42910054326057434,
      "learning_rate": 8.508602733296635e-06,
      "loss": 0.3474,
      "step": 17479
    },
    {
      "epoch": 0.8021660318480106,
      "grad_norm": 0.4519830346107483,
      "learning_rate": 8.508428044390748e-06,
      "loss": 0.3767,
      "step": 17480
    },
    {
      "epoch": 0.8022119223532651,
      "grad_norm": 0.44371291995048523,
      "learning_rate": 8.5082533470482e-06,
      "loss": 0.3803,
      "step": 17481
    },
    {
      "epoch": 0.8022578128585196,
      "grad_norm": 0.42579391598701477,
      "learning_rate": 8.508078641269409e-06,
      "loss": 0.3284,
      "step": 17482
    },
    {
      "epoch": 0.802303703363774,
      "grad_norm": 0.4859975278377533,
      "learning_rate": 8.507903927054798e-06,
      "loss": 0.489,
      "step": 17483
    },
    {
      "epoch": 0.8023495938690285,
      "grad_norm": 0.43391838669776917,
      "learning_rate": 8.507729204404782e-06,
      "loss": 0.4006,
      "step": 17484
    },
    {
      "epoch": 0.802395484374283,
      "grad_norm": 0.43761229515075684,
      "learning_rate": 8.507554473319785e-06,
      "loss": 0.3717,
      "step": 17485
    },
    {
      "epoch": 0.8024413748795374,
      "grad_norm": 0.4543003737926483,
      "learning_rate": 8.507379733800228e-06,
      "loss": 0.3859,
      "step": 17486
    },
    {
      "epoch": 0.8024872653847919,
      "grad_norm": 0.4755701720714569,
      "learning_rate": 8.50720498584653e-06,
      "loss": 0.3691,
      "step": 17487
    },
    {
      "epoch": 0.8025331558900464,
      "grad_norm": 0.4766960144042969,
      "learning_rate": 8.507030229459108e-06,
      "loss": 0.3994,
      "step": 17488
    },
    {
      "epoch": 0.8025790463953009,
      "grad_norm": 0.4752466678619385,
      "learning_rate": 8.506855464638388e-06,
      "loss": 0.4749,
      "step": 17489
    },
    {
      "epoch": 0.8026249369005553,
      "grad_norm": 0.40678051114082336,
      "learning_rate": 8.506680691384786e-06,
      "loss": 0.3033,
      "step": 17490
    },
    {
      "epoch": 0.8026708274058098,
      "grad_norm": 0.4832918643951416,
      "learning_rate": 8.506505909698724e-06,
      "loss": 0.4608,
      "step": 17491
    },
    {
      "epoch": 0.8027167179110642,
      "grad_norm": 0.48920491337776184,
      "learning_rate": 8.506331119580623e-06,
      "loss": 0.4157,
      "step": 17492
    },
    {
      "epoch": 0.8027626084163186,
      "grad_norm": 0.49876484274864197,
      "learning_rate": 8.506156321030903e-06,
      "loss": 0.5038,
      "step": 17493
    },
    {
      "epoch": 0.8028084989215731,
      "grad_norm": 0.457685649394989,
      "learning_rate": 8.505981514049982e-06,
      "loss": 0.4347,
      "step": 17494
    },
    {
      "epoch": 0.8028543894268276,
      "grad_norm": 0.4268772602081299,
      "learning_rate": 8.505806698638281e-06,
      "loss": 0.3287,
      "step": 17495
    },
    {
      "epoch": 0.802900279932082,
      "grad_norm": 0.42562031745910645,
      "learning_rate": 8.505631874796223e-06,
      "loss": 0.2905,
      "step": 17496
    },
    {
      "epoch": 0.8029461704373365,
      "grad_norm": 0.4237743318080902,
      "learning_rate": 8.505457042524226e-06,
      "loss": 0.3442,
      "step": 17497
    },
    {
      "epoch": 0.802992060942591,
      "grad_norm": 0.4356409013271332,
      "learning_rate": 8.50528220182271e-06,
      "loss": 0.3755,
      "step": 17498
    },
    {
      "epoch": 0.8030379514478454,
      "grad_norm": 0.4810279309749603,
      "learning_rate": 8.505107352692098e-06,
      "loss": 0.4423,
      "step": 17499
    },
    {
      "epoch": 0.8030838419530999,
      "grad_norm": 0.47642117738723755,
      "learning_rate": 8.50493249513281e-06,
      "loss": 0.4336,
      "step": 17500
    },
    {
      "epoch": 0.8031297324583544,
      "grad_norm": 0.4263390898704529,
      "learning_rate": 8.504757629145264e-06,
      "loss": 0.3394,
      "step": 17501
    },
    {
      "epoch": 0.8031756229636088,
      "grad_norm": 0.41579416394233704,
      "learning_rate": 8.504582754729882e-06,
      "loss": 0.3482,
      "step": 17502
    },
    {
      "epoch": 0.8032215134688633,
      "grad_norm": 0.5019678473472595,
      "learning_rate": 8.504407871887086e-06,
      "loss": 0.4498,
      "step": 17503
    },
    {
      "epoch": 0.8032674039741178,
      "grad_norm": 0.4784577190876007,
      "learning_rate": 8.504232980617293e-06,
      "loss": 0.4234,
      "step": 17504
    },
    {
      "epoch": 0.8033132944793722,
      "grad_norm": 0.4198947846889496,
      "learning_rate": 8.504058080920929e-06,
      "loss": 0.3387,
      "step": 17505
    },
    {
      "epoch": 0.8033591849846267,
      "grad_norm": 0.4566434621810913,
      "learning_rate": 8.503883172798408e-06,
      "loss": 0.3935,
      "step": 17506
    },
    {
      "epoch": 0.8034050754898812,
      "grad_norm": 0.4633612632751465,
      "learning_rate": 8.503708256250155e-06,
      "loss": 0.4872,
      "step": 17507
    },
    {
      "epoch": 0.8034509659951357,
      "grad_norm": 0.41579049825668335,
      "learning_rate": 8.50353333127659e-06,
      "loss": 0.346,
      "step": 17508
    },
    {
      "epoch": 0.80349685650039,
      "grad_norm": 0.44550225138664246,
      "learning_rate": 8.503358397878132e-06,
      "loss": 0.3658,
      "step": 17509
    },
    {
      "epoch": 0.8035427470056445,
      "grad_norm": 0.46103397011756897,
      "learning_rate": 8.503183456055202e-06,
      "loss": 0.4376,
      "step": 17510
    },
    {
      "epoch": 0.803588637510899,
      "grad_norm": 0.44786983728408813,
      "learning_rate": 8.503008505808223e-06,
      "loss": 0.3436,
      "step": 17511
    },
    {
      "epoch": 0.8036345280161534,
      "grad_norm": 0.41981229186058044,
      "learning_rate": 8.502833547137615e-06,
      "loss": 0.3341,
      "step": 17512
    },
    {
      "epoch": 0.8036804185214079,
      "grad_norm": 0.43122565746307373,
      "learning_rate": 8.502658580043796e-06,
      "loss": 0.3309,
      "step": 17513
    },
    {
      "epoch": 0.8037263090266624,
      "grad_norm": 0.5017602443695068,
      "learning_rate": 8.50248360452719e-06,
      "loss": 0.4463,
      "step": 17514
    },
    {
      "epoch": 0.8037721995319168,
      "grad_norm": 0.43788447976112366,
      "learning_rate": 8.502308620588215e-06,
      "loss": 0.3788,
      "step": 17515
    },
    {
      "epoch": 0.8038180900371713,
      "grad_norm": 0.4142209589481354,
      "learning_rate": 8.502133628227294e-06,
      "loss": 0.3728,
      "step": 17516
    },
    {
      "epoch": 0.8038639805424258,
      "grad_norm": 0.460007905960083,
      "learning_rate": 8.501958627444846e-06,
      "loss": 0.3887,
      "step": 17517
    },
    {
      "epoch": 0.8039098710476802,
      "grad_norm": 0.4433433711528778,
      "learning_rate": 8.501783618241292e-06,
      "loss": 0.3426,
      "step": 17518
    },
    {
      "epoch": 0.8039557615529347,
      "grad_norm": 0.4799129366874695,
      "learning_rate": 8.501608600617057e-06,
      "loss": 0.4972,
      "step": 17519
    },
    {
      "epoch": 0.8040016520581892,
      "grad_norm": 0.47678110003471375,
      "learning_rate": 8.501433574572555e-06,
      "loss": 0.4419,
      "step": 17520
    },
    {
      "epoch": 0.8040475425634436,
      "grad_norm": 0.4584716558456421,
      "learning_rate": 8.501258540108212e-06,
      "loss": 0.3936,
      "step": 17521
    },
    {
      "epoch": 0.8040934330686981,
      "grad_norm": 0.5016146898269653,
      "learning_rate": 8.501083497224446e-06,
      "loss": 0.5283,
      "step": 17522
    },
    {
      "epoch": 0.8041393235739526,
      "grad_norm": 0.4731155037879944,
      "learning_rate": 8.500908445921681e-06,
      "loss": 0.4581,
      "step": 17523
    },
    {
      "epoch": 0.8041852140792071,
      "grad_norm": 0.5109710693359375,
      "learning_rate": 8.500733386200336e-06,
      "loss": 0.4965,
      "step": 17524
    },
    {
      "epoch": 0.8042311045844615,
      "grad_norm": 0.5115070343017578,
      "learning_rate": 8.50055831806083e-06,
      "loss": 0.5215,
      "step": 17525
    },
    {
      "epoch": 0.804276995089716,
      "grad_norm": 0.4108628034591675,
      "learning_rate": 8.500383241503589e-06,
      "loss": 0.2982,
      "step": 17526
    },
    {
      "epoch": 0.8043228855949704,
      "grad_norm": 0.4349287748336792,
      "learning_rate": 8.500208156529028e-06,
      "loss": 0.3516,
      "step": 17527
    },
    {
      "epoch": 0.8043687761002248,
      "grad_norm": 0.4549620747566223,
      "learning_rate": 8.500033063137574e-06,
      "loss": 0.383,
      "step": 17528
    },
    {
      "epoch": 0.8044146666054793,
      "grad_norm": 0.4648036062717438,
      "learning_rate": 8.499857961329642e-06,
      "loss": 0.4123,
      "step": 17529
    },
    {
      "epoch": 0.8044605571107338,
      "grad_norm": 0.4485704004764557,
      "learning_rate": 8.49968285110566e-06,
      "loss": 0.3971,
      "step": 17530
    },
    {
      "epoch": 0.8045064476159882,
      "grad_norm": 0.46702978014945984,
      "learning_rate": 8.499507732466042e-06,
      "loss": 0.3784,
      "step": 17531
    },
    {
      "epoch": 0.8045523381212427,
      "grad_norm": 0.46648073196411133,
      "learning_rate": 8.499332605411213e-06,
      "loss": 0.4191,
      "step": 17532
    },
    {
      "epoch": 0.8045982286264972,
      "grad_norm": 0.4410386383533478,
      "learning_rate": 8.499157469941595e-06,
      "loss": 0.3697,
      "step": 17533
    },
    {
      "epoch": 0.8046441191317516,
      "grad_norm": 0.41777467727661133,
      "learning_rate": 8.498982326057605e-06,
      "loss": 0.3146,
      "step": 17534
    },
    {
      "epoch": 0.8046900096370061,
      "grad_norm": 0.4639178216457367,
      "learning_rate": 8.49880717375967e-06,
      "loss": 0.4536,
      "step": 17535
    },
    {
      "epoch": 0.8047359001422606,
      "grad_norm": 0.4542098343372345,
      "learning_rate": 8.498632013048207e-06,
      "loss": 0.4077,
      "step": 17536
    },
    {
      "epoch": 0.804781790647515,
      "grad_norm": 0.4633980095386505,
      "learning_rate": 8.498456843923636e-06,
      "loss": 0.4125,
      "step": 17537
    },
    {
      "epoch": 0.8048276811527695,
      "grad_norm": 0.4290273189544678,
      "learning_rate": 8.498281666386384e-06,
      "loss": 0.4065,
      "step": 17538
    },
    {
      "epoch": 0.804873571658024,
      "grad_norm": 0.41363418102264404,
      "learning_rate": 8.498106480436865e-06,
      "loss": 0.3374,
      "step": 17539
    },
    {
      "epoch": 0.8049194621632784,
      "grad_norm": 0.4671345353126526,
      "learning_rate": 8.497931286075507e-06,
      "loss": 0.4141,
      "step": 17540
    },
    {
      "epoch": 0.8049653526685329,
      "grad_norm": 0.48397475481033325,
      "learning_rate": 8.497756083302726e-06,
      "loss": 0.4576,
      "step": 17541
    },
    {
      "epoch": 0.8050112431737874,
      "grad_norm": 0.5039771795272827,
      "learning_rate": 8.497580872118946e-06,
      "loss": 0.4963,
      "step": 17542
    },
    {
      "epoch": 0.8050571336790419,
      "grad_norm": 0.4717659056186676,
      "learning_rate": 8.497405652524589e-06,
      "loss": 0.3314,
      "step": 17543
    },
    {
      "epoch": 0.8051030241842962,
      "grad_norm": 0.4212946891784668,
      "learning_rate": 8.497230424520075e-06,
      "loss": 0.3366,
      "step": 17544
    },
    {
      "epoch": 0.8051489146895507,
      "grad_norm": 0.4275185465812683,
      "learning_rate": 8.497055188105825e-06,
      "loss": 0.3431,
      "step": 17545
    },
    {
      "epoch": 0.8051948051948052,
      "grad_norm": 0.4715089499950409,
      "learning_rate": 8.49687994328226e-06,
      "loss": 0.4586,
      "step": 17546
    },
    {
      "epoch": 0.8052406957000596,
      "grad_norm": 0.45079877972602844,
      "learning_rate": 8.496704690049804e-06,
      "loss": 0.3992,
      "step": 17547
    },
    {
      "epoch": 0.8052865862053141,
      "grad_norm": 0.41887596249580383,
      "learning_rate": 8.496529428408877e-06,
      "loss": 0.3294,
      "step": 17548
    },
    {
      "epoch": 0.8053324767105686,
      "grad_norm": 0.4424496591091156,
      "learning_rate": 8.496354158359899e-06,
      "loss": 0.3757,
      "step": 17549
    },
    {
      "epoch": 0.805378367215823,
      "grad_norm": 0.42896175384521484,
      "learning_rate": 8.496178879903294e-06,
      "loss": 0.3755,
      "step": 17550
    },
    {
      "epoch": 0.8054242577210775,
      "grad_norm": 0.44552144408226013,
      "learning_rate": 8.49600359303948e-06,
      "loss": 0.4048,
      "step": 17551
    },
    {
      "epoch": 0.805470148226332,
      "grad_norm": 0.46317818760871887,
      "learning_rate": 8.495828297768884e-06,
      "loss": 0.3982,
      "step": 17552
    },
    {
      "epoch": 0.8055160387315864,
      "grad_norm": 0.48188456892967224,
      "learning_rate": 8.495652994091922e-06,
      "loss": 0.4211,
      "step": 17553
    },
    {
      "epoch": 0.8055619292368409,
      "grad_norm": 0.4468204975128174,
      "learning_rate": 8.495477682009019e-06,
      "loss": 0.3773,
      "step": 17554
    },
    {
      "epoch": 0.8056078197420954,
      "grad_norm": 0.426616370677948,
      "learning_rate": 8.495302361520594e-06,
      "loss": 0.3708,
      "step": 17555
    },
    {
      "epoch": 0.8056537102473498,
      "grad_norm": 0.4044811427593231,
      "learning_rate": 8.49512703262707e-06,
      "loss": 0.3229,
      "step": 17556
    },
    {
      "epoch": 0.8056996007526043,
      "grad_norm": 0.46874159574508667,
      "learning_rate": 8.49495169532887e-06,
      "loss": 0.4263,
      "step": 17557
    },
    {
      "epoch": 0.8057454912578588,
      "grad_norm": 0.4282771348953247,
      "learning_rate": 8.494776349626413e-06,
      "loss": 0.3676,
      "step": 17558
    },
    {
      "epoch": 0.8057913817631132,
      "grad_norm": 0.4589526951313019,
      "learning_rate": 8.494600995520122e-06,
      "loss": 0.3943,
      "step": 17559
    },
    {
      "epoch": 0.8058372722683677,
      "grad_norm": 0.44682082533836365,
      "learning_rate": 8.49442563301042e-06,
      "loss": 0.3813,
      "step": 17560
    },
    {
      "epoch": 0.8058831627736222,
      "grad_norm": 0.4717269241809845,
      "learning_rate": 8.494250262097726e-06,
      "loss": 0.3924,
      "step": 17561
    },
    {
      "epoch": 0.8059290532788767,
      "grad_norm": 0.4556524455547333,
      "learning_rate": 8.494074882782462e-06,
      "loss": 0.403,
      "step": 17562
    },
    {
      "epoch": 0.805974943784131,
      "grad_norm": 0.45353206992149353,
      "learning_rate": 8.493899495065053e-06,
      "loss": 0.4196,
      "step": 17563
    },
    {
      "epoch": 0.8060208342893855,
      "grad_norm": 0.416664719581604,
      "learning_rate": 8.493724098945917e-06,
      "loss": 0.3681,
      "step": 17564
    },
    {
      "epoch": 0.80606672479464,
      "grad_norm": 0.45668983459472656,
      "learning_rate": 8.493548694425478e-06,
      "loss": 0.3924,
      "step": 17565
    },
    {
      "epoch": 0.8061126152998944,
      "grad_norm": 0.429853230714798,
      "learning_rate": 8.493373281504156e-06,
      "loss": 0.3456,
      "step": 17566
    },
    {
      "epoch": 0.8061585058051489,
      "grad_norm": 0.4246322512626648,
      "learning_rate": 8.493197860182374e-06,
      "loss": 0.3459,
      "step": 17567
    },
    {
      "epoch": 0.8062043963104034,
      "grad_norm": 0.5401395559310913,
      "learning_rate": 8.493022430460553e-06,
      "loss": 0.5133,
      "step": 17568
    },
    {
      "epoch": 0.8062502868156578,
      "grad_norm": 0.5249103307723999,
      "learning_rate": 8.492846992339117e-06,
      "loss": 0.5732,
      "step": 17569
    },
    {
      "epoch": 0.8062961773209123,
      "grad_norm": 0.49460428953170776,
      "learning_rate": 8.492671545818485e-06,
      "loss": 0.5163,
      "step": 17570
    },
    {
      "epoch": 0.8063420678261668,
      "grad_norm": 0.4316990077495575,
      "learning_rate": 8.492496090899081e-06,
      "loss": 0.3835,
      "step": 17571
    },
    {
      "epoch": 0.8063879583314212,
      "grad_norm": 0.4703753888607025,
      "learning_rate": 8.492320627581325e-06,
      "loss": 0.4047,
      "step": 17572
    },
    {
      "epoch": 0.8064338488366757,
      "grad_norm": 0.4622040092945099,
      "learning_rate": 8.492145155865642e-06,
      "loss": 0.4116,
      "step": 17573
    },
    {
      "epoch": 0.8064797393419302,
      "grad_norm": 0.4789207875728607,
      "learning_rate": 8.491969675752451e-06,
      "loss": 0.4588,
      "step": 17574
    },
    {
      "epoch": 0.8065256298471846,
      "grad_norm": 0.46948862075805664,
      "learning_rate": 8.491794187242175e-06,
      "loss": 0.4119,
      "step": 17575
    },
    {
      "epoch": 0.8065715203524391,
      "grad_norm": 0.4332551062107086,
      "learning_rate": 8.491618690335238e-06,
      "loss": 0.4058,
      "step": 17576
    },
    {
      "epoch": 0.8066174108576936,
      "grad_norm": 0.48013734817504883,
      "learning_rate": 8.491443185032058e-06,
      "loss": 0.4626,
      "step": 17577
    },
    {
      "epoch": 0.8066633013629481,
      "grad_norm": 0.44204577803611755,
      "learning_rate": 8.491267671333058e-06,
      "loss": 0.4212,
      "step": 17578
    },
    {
      "epoch": 0.8067091918682024,
      "grad_norm": 0.4621376097202301,
      "learning_rate": 8.491092149238663e-06,
      "loss": 0.3937,
      "step": 17579
    },
    {
      "epoch": 0.8067550823734569,
      "grad_norm": 0.4336664080619812,
      "learning_rate": 8.490916618749292e-06,
      "loss": 0.4195,
      "step": 17580
    },
    {
      "epoch": 0.8068009728787114,
      "grad_norm": 0.5178913474082947,
      "learning_rate": 8.490741079865368e-06,
      "loss": 0.5187,
      "step": 17581
    },
    {
      "epoch": 0.8068468633839658,
      "grad_norm": 0.4744816720485687,
      "learning_rate": 8.490565532587316e-06,
      "loss": 0.4203,
      "step": 17582
    },
    {
      "epoch": 0.8068927538892203,
      "grad_norm": 0.440844863653183,
      "learning_rate": 8.490389976915554e-06,
      "loss": 0.3704,
      "step": 17583
    },
    {
      "epoch": 0.8069386443944748,
      "grad_norm": 0.4821132719516754,
      "learning_rate": 8.490214412850505e-06,
      "loss": 0.4639,
      "step": 17584
    },
    {
      "epoch": 0.8069845348997292,
      "grad_norm": 0.4527164101600647,
      "learning_rate": 8.49003884039259e-06,
      "loss": 0.3881,
      "step": 17585
    },
    {
      "epoch": 0.8070304254049837,
      "grad_norm": 0.44395074248313904,
      "learning_rate": 8.489863259542239e-06,
      "loss": 0.3855,
      "step": 17586
    },
    {
      "epoch": 0.8070763159102382,
      "grad_norm": 0.45803093910217285,
      "learning_rate": 8.489687670299864e-06,
      "loss": 0.4374,
      "step": 17587
    },
    {
      "epoch": 0.8071222064154926,
      "grad_norm": 0.4702085852622986,
      "learning_rate": 8.489512072665893e-06,
      "loss": 0.3696,
      "step": 17588
    },
    {
      "epoch": 0.8071680969207471,
      "grad_norm": 0.4370243549346924,
      "learning_rate": 8.489336466640746e-06,
      "loss": 0.3497,
      "step": 17589
    },
    {
      "epoch": 0.8072139874260016,
      "grad_norm": 0.43086567521095276,
      "learning_rate": 8.489160852224845e-06,
      "loss": 0.3518,
      "step": 17590
    },
    {
      "epoch": 0.807259877931256,
      "grad_norm": 0.4483693242073059,
      "learning_rate": 8.488985229418616e-06,
      "loss": 0.4225,
      "step": 17591
    },
    {
      "epoch": 0.8073057684365105,
      "grad_norm": 0.5005820989608765,
      "learning_rate": 8.488809598222476e-06,
      "loss": 0.4964,
      "step": 17592
    },
    {
      "epoch": 0.807351658941765,
      "grad_norm": 0.4495147168636322,
      "learning_rate": 8.488633958636852e-06,
      "loss": 0.4018,
      "step": 17593
    },
    {
      "epoch": 0.8073975494470194,
      "grad_norm": 0.5154932737350464,
      "learning_rate": 8.488458310662164e-06,
      "loss": 0.3851,
      "step": 17594
    },
    {
      "epoch": 0.8074434399522739,
      "grad_norm": 0.45882177352905273,
      "learning_rate": 8.488282654298834e-06,
      "loss": 0.4501,
      "step": 17595
    },
    {
      "epoch": 0.8074893304575284,
      "grad_norm": 0.4814545214176178,
      "learning_rate": 8.488106989547285e-06,
      "loss": 0.4558,
      "step": 17596
    },
    {
      "epoch": 0.8075352209627829,
      "grad_norm": 0.4598345160484314,
      "learning_rate": 8.48793131640794e-06,
      "loss": 0.4113,
      "step": 17597
    },
    {
      "epoch": 0.8075811114680372,
      "grad_norm": 0.4630196988582611,
      "learning_rate": 8.487755634881222e-06,
      "loss": 0.3791,
      "step": 17598
    },
    {
      "epoch": 0.8076270019732917,
      "grad_norm": 0.4725218415260315,
      "learning_rate": 8.487579944967551e-06,
      "loss": 0.4473,
      "step": 17599
    },
    {
      "epoch": 0.8076728924785462,
      "grad_norm": 0.4276121258735657,
      "learning_rate": 8.487404246667352e-06,
      "loss": 0.3427,
      "step": 17600
    },
    {
      "epoch": 0.8077187829838006,
      "grad_norm": 0.501977801322937,
      "learning_rate": 8.487228539981045e-06,
      "loss": 0.5395,
      "step": 17601
    },
    {
      "epoch": 0.8077646734890551,
      "grad_norm": 0.4523829519748688,
      "learning_rate": 8.487052824909056e-06,
      "loss": 0.4354,
      "step": 17602
    },
    {
      "epoch": 0.8078105639943096,
      "grad_norm": 0.46578773856163025,
      "learning_rate": 8.486877101451805e-06,
      "loss": 0.4089,
      "step": 17603
    },
    {
      "epoch": 0.807856454499564,
      "grad_norm": 0.4490325152873993,
      "learning_rate": 8.486701369609714e-06,
      "loss": 0.3935,
      "step": 17604
    },
    {
      "epoch": 0.8079023450048185,
      "grad_norm": 0.45202866196632385,
      "learning_rate": 8.486525629383206e-06,
      "loss": 0.3702,
      "step": 17605
    },
    {
      "epoch": 0.807948235510073,
      "grad_norm": 0.467622309923172,
      "learning_rate": 8.486349880772707e-06,
      "loss": 0.419,
      "step": 17606
    },
    {
      "epoch": 0.8079941260153274,
      "grad_norm": 0.4658639132976532,
      "learning_rate": 8.486174123778635e-06,
      "loss": 0.4304,
      "step": 17607
    },
    {
      "epoch": 0.8080400165205819,
      "grad_norm": 0.45953649282455444,
      "learning_rate": 8.485998358401415e-06,
      "loss": 0.3995,
      "step": 17608
    },
    {
      "epoch": 0.8080859070258364,
      "grad_norm": 0.42754751443862915,
      "learning_rate": 8.485822584641467e-06,
      "loss": 0.3353,
      "step": 17609
    },
    {
      "epoch": 0.8081317975310908,
      "grad_norm": 0.47774219512939453,
      "learning_rate": 8.485646802499219e-06,
      "loss": 0.4575,
      "step": 17610
    },
    {
      "epoch": 0.8081776880363453,
      "grad_norm": 0.4445607662200928,
      "learning_rate": 8.485471011975089e-06,
      "loss": 0.3858,
      "step": 17611
    },
    {
      "epoch": 0.8082235785415998,
      "grad_norm": 0.5216554999351501,
      "learning_rate": 8.485295213069503e-06,
      "loss": 0.5137,
      "step": 17612
    },
    {
      "epoch": 0.8082694690468543,
      "grad_norm": 0.6023508310317993,
      "learning_rate": 8.48511940578288e-06,
      "loss": 0.3842,
      "step": 17613
    },
    {
      "epoch": 0.8083153595521086,
      "grad_norm": 0.4393535554409027,
      "learning_rate": 8.484943590115644e-06,
      "loss": 0.3455,
      "step": 17614
    },
    {
      "epoch": 0.8083612500573631,
      "grad_norm": 0.43418657779693604,
      "learning_rate": 8.48476776606822e-06,
      "loss": 0.407,
      "step": 17615
    },
    {
      "epoch": 0.8084071405626176,
      "grad_norm": 0.4357938766479492,
      "learning_rate": 8.48459193364103e-06,
      "loss": 0.3664,
      "step": 17616
    },
    {
      "epoch": 0.808453031067872,
      "grad_norm": 0.4474564492702484,
      "learning_rate": 8.484416092834496e-06,
      "loss": 0.3923,
      "step": 17617
    },
    {
      "epoch": 0.8084989215731265,
      "grad_norm": 0.46380454301834106,
      "learning_rate": 8.48424024364904e-06,
      "loss": 0.417,
      "step": 17618
    },
    {
      "epoch": 0.808544812078381,
      "grad_norm": 0.4853716194629669,
      "learning_rate": 8.484064386085087e-06,
      "loss": 0.5074,
      "step": 17619
    },
    {
      "epoch": 0.8085907025836354,
      "grad_norm": 0.46699202060699463,
      "learning_rate": 8.483888520143057e-06,
      "loss": 0.4481,
      "step": 17620
    },
    {
      "epoch": 0.8086365930888899,
      "grad_norm": 0.5123815536499023,
      "learning_rate": 8.483712645823375e-06,
      "loss": 0.4693,
      "step": 17621
    },
    {
      "epoch": 0.8086824835941444,
      "grad_norm": 0.472379595041275,
      "learning_rate": 8.483536763126466e-06,
      "loss": 0.4485,
      "step": 17622
    },
    {
      "epoch": 0.8087283740993988,
      "grad_norm": 0.4936237037181854,
      "learning_rate": 8.483360872052748e-06,
      "loss": 0.4482,
      "step": 17623
    },
    {
      "epoch": 0.8087742646046533,
      "grad_norm": 0.4686533510684967,
      "learning_rate": 8.483184972602647e-06,
      "loss": 0.4409,
      "step": 17624
    },
    {
      "epoch": 0.8088201551099078,
      "grad_norm": 0.44081053137779236,
      "learning_rate": 8.483009064776587e-06,
      "loss": 0.3979,
      "step": 17625
    },
    {
      "epoch": 0.8088660456151622,
      "grad_norm": 0.5010126233100891,
      "learning_rate": 8.482833148574988e-06,
      "loss": 0.4262,
      "step": 17626
    },
    {
      "epoch": 0.8089119361204167,
      "grad_norm": 0.4746077358722687,
      "learning_rate": 8.482657223998275e-06,
      "loss": 0.4347,
      "step": 17627
    },
    {
      "epoch": 0.8089578266256712,
      "grad_norm": 0.46039292216300964,
      "learning_rate": 8.482481291046869e-06,
      "loss": 0.4191,
      "step": 17628
    },
    {
      "epoch": 0.8090037171309256,
      "grad_norm": 0.45977118611335754,
      "learning_rate": 8.482305349721197e-06,
      "loss": 0.4037,
      "step": 17629
    },
    {
      "epoch": 0.8090496076361801,
      "grad_norm": 0.4507550299167633,
      "learning_rate": 8.482129400021678e-06,
      "loss": 0.359,
      "step": 17630
    },
    {
      "epoch": 0.8090954981414346,
      "grad_norm": 0.4625616669654846,
      "learning_rate": 8.481953441948736e-06,
      "loss": 0.4245,
      "step": 17631
    },
    {
      "epoch": 0.809141388646689,
      "grad_norm": 0.4414457678794861,
      "learning_rate": 8.481777475502797e-06,
      "loss": 0.3174,
      "step": 17632
    },
    {
      "epoch": 0.8091872791519434,
      "grad_norm": 0.4520516097545624,
      "learning_rate": 8.48160150068428e-06,
      "loss": 0.3905,
      "step": 17633
    },
    {
      "epoch": 0.8092331696571979,
      "grad_norm": 0.4836430549621582,
      "learning_rate": 8.481425517493613e-06,
      "loss": 0.4242,
      "step": 17634
    },
    {
      "epoch": 0.8092790601624524,
      "grad_norm": 0.5088282227516174,
      "learning_rate": 8.481249525931214e-06,
      "loss": 0.4561,
      "step": 17635
    },
    {
      "epoch": 0.8093249506677068,
      "grad_norm": 0.4247996509075165,
      "learning_rate": 8.48107352599751e-06,
      "loss": 0.3605,
      "step": 17636
    },
    {
      "epoch": 0.8093708411729613,
      "grad_norm": 0.4483327865600586,
      "learning_rate": 8.480897517692923e-06,
      "loss": 0.4181,
      "step": 17637
    },
    {
      "epoch": 0.8094167316782158,
      "grad_norm": 0.44462546706199646,
      "learning_rate": 8.480721501017875e-06,
      "loss": 0.3928,
      "step": 17638
    },
    {
      "epoch": 0.8094626221834702,
      "grad_norm": 0.47163113951683044,
      "learning_rate": 8.48054547597279e-06,
      "loss": 0.4427,
      "step": 17639
    },
    {
      "epoch": 0.8095085126887247,
      "grad_norm": 0.46032679080963135,
      "learning_rate": 8.480369442558092e-06,
      "loss": 0.3676,
      "step": 17640
    },
    {
      "epoch": 0.8095544031939792,
      "grad_norm": 0.4542371928691864,
      "learning_rate": 8.480193400774205e-06,
      "loss": 0.384,
      "step": 17641
    },
    {
      "epoch": 0.8096002936992336,
      "grad_norm": 0.5070081353187561,
      "learning_rate": 8.48001735062155e-06,
      "loss": 0.413,
      "step": 17642
    },
    {
      "epoch": 0.8096461842044881,
      "grad_norm": 0.43948009610176086,
      "learning_rate": 8.479841292100552e-06,
      "loss": 0.3536,
      "step": 17643
    },
    {
      "epoch": 0.8096920747097426,
      "grad_norm": 0.4311222434043884,
      "learning_rate": 8.479665225211634e-06,
      "loss": 0.3184,
      "step": 17644
    },
    {
      "epoch": 0.809737965214997,
      "grad_norm": 0.5065967440605164,
      "learning_rate": 8.47948914995522e-06,
      "loss": 0.4688,
      "step": 17645
    },
    {
      "epoch": 0.8097838557202515,
      "grad_norm": 0.45486101508140564,
      "learning_rate": 8.479313066331732e-06,
      "loss": 0.3618,
      "step": 17646
    },
    {
      "epoch": 0.809829746225506,
      "grad_norm": 0.4497266113758087,
      "learning_rate": 8.479136974341594e-06,
      "loss": 0.384,
      "step": 17647
    },
    {
      "epoch": 0.8098756367307604,
      "grad_norm": 0.46375998854637146,
      "learning_rate": 8.478960873985231e-06,
      "loss": 0.4549,
      "step": 17648
    },
    {
      "epoch": 0.8099215272360148,
      "grad_norm": 0.4329161047935486,
      "learning_rate": 8.478784765263064e-06,
      "loss": 0.3388,
      "step": 17649
    },
    {
      "epoch": 0.8099674177412693,
      "grad_norm": 0.48870131373405457,
      "learning_rate": 8.478608648175518e-06,
      "loss": 0.511,
      "step": 17650
    },
    {
      "epoch": 0.8100133082465238,
      "grad_norm": 0.48089170455932617,
      "learning_rate": 8.478432522723015e-06,
      "loss": 0.4801,
      "step": 17651
    },
    {
      "epoch": 0.8100591987517782,
      "grad_norm": 0.45487064123153687,
      "learning_rate": 8.47825638890598e-06,
      "loss": 0.4623,
      "step": 17652
    },
    {
      "epoch": 0.8101050892570327,
      "grad_norm": 0.45531803369522095,
      "learning_rate": 8.478080246724837e-06,
      "loss": 0.3652,
      "step": 17653
    },
    {
      "epoch": 0.8101509797622872,
      "grad_norm": 0.44705572724342346,
      "learning_rate": 8.477904096180008e-06,
      "loss": 0.3457,
      "step": 17654
    },
    {
      "epoch": 0.8101968702675416,
      "grad_norm": 0.4420647919178009,
      "learning_rate": 8.477727937271918e-06,
      "loss": 0.374,
      "step": 17655
    },
    {
      "epoch": 0.8102427607727961,
      "grad_norm": 0.4088599383831024,
      "learning_rate": 8.477551770000989e-06,
      "loss": 0.3002,
      "step": 17656
    },
    {
      "epoch": 0.8102886512780506,
      "grad_norm": 0.447159081697464,
      "learning_rate": 8.477375594367645e-06,
      "loss": 0.372,
      "step": 17657
    },
    {
      "epoch": 0.810334541783305,
      "grad_norm": 0.5326886177062988,
      "learning_rate": 8.477199410372313e-06,
      "loss": 0.5348,
      "step": 17658
    },
    {
      "epoch": 0.8103804322885595,
      "grad_norm": 0.4563516676425934,
      "learning_rate": 8.477023218015411e-06,
      "loss": 0.4119,
      "step": 17659
    },
    {
      "epoch": 0.810426322793814,
      "grad_norm": 0.5115314722061157,
      "learning_rate": 8.476847017297367e-06,
      "loss": 0.364,
      "step": 17660
    },
    {
      "epoch": 0.8104722132990684,
      "grad_norm": 0.4515569806098938,
      "learning_rate": 8.476670808218602e-06,
      "loss": 0.3922,
      "step": 17661
    },
    {
      "epoch": 0.8105181038043229,
      "grad_norm": 0.48099735379219055,
      "learning_rate": 8.476494590779542e-06,
      "loss": 0.4781,
      "step": 17662
    },
    {
      "epoch": 0.8105639943095774,
      "grad_norm": 0.442349910736084,
      "learning_rate": 8.47631836498061e-06,
      "loss": 0.3612,
      "step": 17663
    },
    {
      "epoch": 0.8106098848148318,
      "grad_norm": 0.4784599542617798,
      "learning_rate": 8.476142130822228e-06,
      "loss": 0.3984,
      "step": 17664
    },
    {
      "epoch": 0.8106557753200863,
      "grad_norm": 0.40242910385131836,
      "learning_rate": 8.475965888304823e-06,
      "loss": 0.3235,
      "step": 17665
    },
    {
      "epoch": 0.8107016658253408,
      "grad_norm": 0.4843137562274933,
      "learning_rate": 8.475789637428816e-06,
      "loss": 0.4081,
      "step": 17666
    },
    {
      "epoch": 0.8107475563305953,
      "grad_norm": 0.42413055896759033,
      "learning_rate": 8.475613378194632e-06,
      "loss": 0.3606,
      "step": 17667
    },
    {
      "epoch": 0.8107934468358496,
      "grad_norm": 0.45493200421333313,
      "learning_rate": 8.475437110602696e-06,
      "loss": 0.4236,
      "step": 17668
    },
    {
      "epoch": 0.8108393373411041,
      "grad_norm": 0.4516465365886688,
      "learning_rate": 8.47526083465343e-06,
      "loss": 0.3909,
      "step": 17669
    },
    {
      "epoch": 0.8108852278463586,
      "grad_norm": 0.4620436728000641,
      "learning_rate": 8.475084550347256e-06,
      "loss": 0.3587,
      "step": 17670
    },
    {
      "epoch": 0.810931118351613,
      "grad_norm": 0.46484631299972534,
      "learning_rate": 8.474908257684605e-06,
      "loss": 0.4072,
      "step": 17671
    },
    {
      "epoch": 0.8109770088568675,
      "grad_norm": 0.4686445891857147,
      "learning_rate": 8.474731956665892e-06,
      "loss": 0.3783,
      "step": 17672
    },
    {
      "epoch": 0.811022899362122,
      "grad_norm": 0.46818459033966064,
      "learning_rate": 8.474555647291549e-06,
      "loss": 0.3592,
      "step": 17673
    },
    {
      "epoch": 0.8110687898673764,
      "grad_norm": 0.425287127494812,
      "learning_rate": 8.474379329561994e-06,
      "loss": 0.3205,
      "step": 17674
    },
    {
      "epoch": 0.8111146803726309,
      "grad_norm": 0.4470694363117218,
      "learning_rate": 8.474203003477652e-06,
      "loss": 0.3387,
      "step": 17675
    },
    {
      "epoch": 0.8111605708778854,
      "grad_norm": 0.42304760217666626,
      "learning_rate": 8.47402666903895e-06,
      "loss": 0.337,
      "step": 17676
    },
    {
      "epoch": 0.8112064613831398,
      "grad_norm": 0.46694934368133545,
      "learning_rate": 8.473850326246312e-06,
      "loss": 0.4201,
      "step": 17677
    },
    {
      "epoch": 0.8112523518883943,
      "grad_norm": 0.4868218004703522,
      "learning_rate": 8.473673975100159e-06,
      "loss": 0.4813,
      "step": 17678
    },
    {
      "epoch": 0.8112982423936488,
      "grad_norm": 0.4300846457481384,
      "learning_rate": 8.473497615600915e-06,
      "loss": 0.3997,
      "step": 17679
    },
    {
      "epoch": 0.8113441328989032,
      "grad_norm": 0.4250863492488861,
      "learning_rate": 8.473321247749007e-06,
      "loss": 0.3252,
      "step": 17680
    },
    {
      "epoch": 0.8113900234041577,
      "grad_norm": 0.886294960975647,
      "learning_rate": 8.473144871544857e-06,
      "loss": 0.3889,
      "step": 17681
    },
    {
      "epoch": 0.8114359139094122,
      "grad_norm": 0.437728613615036,
      "learning_rate": 8.47296848698889e-06,
      "loss": 0.3461,
      "step": 17682
    },
    {
      "epoch": 0.8114818044146666,
      "grad_norm": 0.47892123460769653,
      "learning_rate": 8.47279209408153e-06,
      "loss": 0.4935,
      "step": 17683
    },
    {
      "epoch": 0.811527694919921,
      "grad_norm": 0.48336225748062134,
      "learning_rate": 8.472615692823201e-06,
      "loss": 0.4275,
      "step": 17684
    },
    {
      "epoch": 0.8115735854251755,
      "grad_norm": 0.4525268077850342,
      "learning_rate": 8.472439283214326e-06,
      "loss": 0.4056,
      "step": 17685
    },
    {
      "epoch": 0.81161947593043,
      "grad_norm": 0.4483889937400818,
      "learning_rate": 8.472262865255333e-06,
      "loss": 0.3851,
      "step": 17686
    },
    {
      "epoch": 0.8116653664356844,
      "grad_norm": 0.4581266939640045,
      "learning_rate": 8.472086438946642e-06,
      "loss": 0.3602,
      "step": 17687
    },
    {
      "epoch": 0.8117112569409389,
      "grad_norm": 0.5165950059890747,
      "learning_rate": 8.471910004288679e-06,
      "loss": 0.44,
      "step": 17688
    },
    {
      "epoch": 0.8117571474461934,
      "grad_norm": 0.4431636333465576,
      "learning_rate": 8.471733561281867e-06,
      "loss": 0.3582,
      "step": 17689
    },
    {
      "epoch": 0.8118030379514478,
      "grad_norm": 0.4877323806285858,
      "learning_rate": 8.471557109926634e-06,
      "loss": 0.4592,
      "step": 17690
    },
    {
      "epoch": 0.8118489284567023,
      "grad_norm": 0.4613015055656433,
      "learning_rate": 8.4713806502234e-06,
      "loss": 0.4314,
      "step": 17691
    },
    {
      "epoch": 0.8118948189619568,
      "grad_norm": 0.5059880018234253,
      "learning_rate": 8.471204182172593e-06,
      "loss": 0.4786,
      "step": 17692
    },
    {
      "epoch": 0.8119407094672112,
      "grad_norm": 0.4817078709602356,
      "learning_rate": 8.471027705774634e-06,
      "loss": 0.4132,
      "step": 17693
    },
    {
      "epoch": 0.8119865999724657,
      "grad_norm": 0.46134814620018005,
      "learning_rate": 8.470851221029949e-06,
      "loss": 0.4168,
      "step": 17694
    },
    {
      "epoch": 0.8120324904777202,
      "grad_norm": 0.451264351606369,
      "learning_rate": 8.470674727938964e-06,
      "loss": 0.3973,
      "step": 17695
    },
    {
      "epoch": 0.8120783809829746,
      "grad_norm": 0.4588071405887604,
      "learning_rate": 8.4704982265021e-06,
      "loss": 0.3651,
      "step": 17696
    },
    {
      "epoch": 0.8121242714882291,
      "grad_norm": 0.42154666781425476,
      "learning_rate": 8.470321716719783e-06,
      "loss": 0.3349,
      "step": 17697
    },
    {
      "epoch": 0.8121701619934836,
      "grad_norm": 0.4712468087673187,
      "learning_rate": 8.470145198592436e-06,
      "loss": 0.4473,
      "step": 17698
    },
    {
      "epoch": 0.812216052498738,
      "grad_norm": 0.45039933919906616,
      "learning_rate": 8.469968672120487e-06,
      "loss": 0.4269,
      "step": 17699
    },
    {
      "epoch": 0.8122619430039925,
      "grad_norm": 0.4717012941837311,
      "learning_rate": 8.469792137304357e-06,
      "loss": 0.4735,
      "step": 17700
    },
    {
      "epoch": 0.812307833509247,
      "grad_norm": 0.4294164776802063,
      "learning_rate": 8.469615594144477e-06,
      "loss": 0.4119,
      "step": 17701
    },
    {
      "epoch": 0.8123537240145015,
      "grad_norm": 0.4132399559020996,
      "learning_rate": 8.46943904264126e-06,
      "loss": 0.3268,
      "step": 17702
    },
    {
      "epoch": 0.8123996145197558,
      "grad_norm": 0.5065289735794067,
      "learning_rate": 8.469262482795139e-06,
      "loss": 0.4573,
      "step": 17703
    },
    {
      "epoch": 0.8124455050250103,
      "grad_norm": 0.4541685879230499,
      "learning_rate": 8.46908591460654e-06,
      "loss": 0.4235,
      "step": 17704
    },
    {
      "epoch": 0.8124913955302648,
      "grad_norm": 0.45357418060302734,
      "learning_rate": 8.468909338075882e-06,
      "loss": 0.3866,
      "step": 17705
    },
    {
      "epoch": 0.8125372860355192,
      "grad_norm": 0.41954857110977173,
      "learning_rate": 8.46873275320359e-06,
      "loss": 0.3607,
      "step": 17706
    },
    {
      "epoch": 0.8125831765407737,
      "grad_norm": 0.44689586758613586,
      "learning_rate": 8.468556159990094e-06,
      "loss": 0.3671,
      "step": 17707
    },
    {
      "epoch": 0.8126290670460282,
      "grad_norm": 0.43343621492385864,
      "learning_rate": 8.468379558435813e-06,
      "loss": 0.3803,
      "step": 17708
    },
    {
      "epoch": 0.8126749575512826,
      "grad_norm": 0.4828508794307709,
      "learning_rate": 8.468202948541173e-06,
      "loss": 0.4271,
      "step": 17709
    },
    {
      "epoch": 0.8127208480565371,
      "grad_norm": 0.45076027512550354,
      "learning_rate": 8.468026330306602e-06,
      "loss": 0.3199,
      "step": 17710
    },
    {
      "epoch": 0.8127667385617916,
      "grad_norm": 0.4199490249156952,
      "learning_rate": 8.46784970373252e-06,
      "loss": 0.3491,
      "step": 17711
    },
    {
      "epoch": 0.812812629067046,
      "grad_norm": 0.4521474540233612,
      "learning_rate": 8.467673068819355e-06,
      "loss": 0.4007,
      "step": 17712
    },
    {
      "epoch": 0.8128585195723005,
      "grad_norm": 0.45946773886680603,
      "learning_rate": 8.467496425567529e-06,
      "loss": 0.3645,
      "step": 17713
    },
    {
      "epoch": 0.812904410077555,
      "grad_norm": 0.4599514305591583,
      "learning_rate": 8.46731977397747e-06,
      "loss": 0.4248,
      "step": 17714
    },
    {
      "epoch": 0.8129503005828094,
      "grad_norm": 0.4499518573284149,
      "learning_rate": 8.467143114049602e-06,
      "loss": 0.336,
      "step": 17715
    },
    {
      "epoch": 0.8129961910880639,
      "grad_norm": 0.4637841582298279,
      "learning_rate": 8.466966445784349e-06,
      "loss": 0.3748,
      "step": 17716
    },
    {
      "epoch": 0.8130420815933184,
      "grad_norm": 0.4588336944580078,
      "learning_rate": 8.466789769182134e-06,
      "loss": 0.3932,
      "step": 17717
    },
    {
      "epoch": 0.8130879720985728,
      "grad_norm": 0.4422934651374817,
      "learning_rate": 8.466613084243386e-06,
      "loss": 0.3789,
      "step": 17718
    },
    {
      "epoch": 0.8131338626038273,
      "grad_norm": 0.4200917184352875,
      "learning_rate": 8.466436390968525e-06,
      "loss": 0.3412,
      "step": 17719
    },
    {
      "epoch": 0.8131797531090817,
      "grad_norm": 0.4593302011489868,
      "learning_rate": 8.466259689357981e-06,
      "loss": 0.3764,
      "step": 17720
    },
    {
      "epoch": 0.8132256436143362,
      "grad_norm": 0.44620248675346375,
      "learning_rate": 8.466082979412175e-06,
      "loss": 0.3821,
      "step": 17721
    },
    {
      "epoch": 0.8132715341195906,
      "grad_norm": 0.451384037733078,
      "learning_rate": 8.465906261131533e-06,
      "loss": 0.4022,
      "step": 17722
    },
    {
      "epoch": 0.8133174246248451,
      "grad_norm": 0.48389217257499695,
      "learning_rate": 8.465729534516482e-06,
      "loss": 0.4304,
      "step": 17723
    },
    {
      "epoch": 0.8133633151300996,
      "grad_norm": 0.4236779808998108,
      "learning_rate": 8.465552799567445e-06,
      "loss": 0.2994,
      "step": 17724
    },
    {
      "epoch": 0.813409205635354,
      "grad_norm": 0.4794600307941437,
      "learning_rate": 8.465376056284846e-06,
      "loss": 0.482,
      "step": 17725
    },
    {
      "epoch": 0.8134550961406085,
      "grad_norm": 0.46359068155288696,
      "learning_rate": 8.465199304669112e-06,
      "loss": 0.3918,
      "step": 17726
    },
    {
      "epoch": 0.813500986645863,
      "grad_norm": 0.4240097403526306,
      "learning_rate": 8.465022544720667e-06,
      "loss": 0.3488,
      "step": 17727
    },
    {
      "epoch": 0.8135468771511174,
      "grad_norm": 0.44780975580215454,
      "learning_rate": 8.464845776439938e-06,
      "loss": 0.3851,
      "step": 17728
    },
    {
      "epoch": 0.8135927676563719,
      "grad_norm": 0.4641299545764923,
      "learning_rate": 8.464668999827345e-06,
      "loss": 0.3647,
      "step": 17729
    },
    {
      "epoch": 0.8136386581616264,
      "grad_norm": 0.4356493651866913,
      "learning_rate": 8.464492214883319e-06,
      "loss": 0.3396,
      "step": 17730
    },
    {
      "epoch": 0.8136845486668808,
      "grad_norm": 0.5075135827064514,
      "learning_rate": 8.464315421608284e-06,
      "loss": 0.4969,
      "step": 17731
    },
    {
      "epoch": 0.8137304391721353,
      "grad_norm": 0.4463634490966797,
      "learning_rate": 8.464138620002662e-06,
      "loss": 0.431,
      "step": 17732
    },
    {
      "epoch": 0.8137763296773898,
      "grad_norm": 0.48259538412094116,
      "learning_rate": 8.46396181006688e-06,
      "loss": 0.4567,
      "step": 17733
    },
    {
      "epoch": 0.8138222201826442,
      "grad_norm": 0.46133771538734436,
      "learning_rate": 8.463784991801362e-06,
      "loss": 0.409,
      "step": 17734
    },
    {
      "epoch": 0.8138681106878987,
      "grad_norm": 0.4328539967536926,
      "learning_rate": 8.463608165206536e-06,
      "loss": 0.3628,
      "step": 17735
    },
    {
      "epoch": 0.8139140011931532,
      "grad_norm": 0.4174555242061615,
      "learning_rate": 8.463431330282826e-06,
      "loss": 0.3275,
      "step": 17736
    },
    {
      "epoch": 0.8139598916984075,
      "grad_norm": 0.4667373597621918,
      "learning_rate": 8.463254487030655e-06,
      "loss": 0.3951,
      "step": 17737
    },
    {
      "epoch": 0.814005782203662,
      "grad_norm": 0.4213361442089081,
      "learning_rate": 8.463077635450452e-06,
      "loss": 0.3493,
      "step": 17738
    },
    {
      "epoch": 0.8140516727089165,
      "grad_norm": 0.5281664729118347,
      "learning_rate": 8.462900775542638e-06,
      "loss": 0.4382,
      "step": 17739
    },
    {
      "epoch": 0.814097563214171,
      "grad_norm": 0.46837037801742554,
      "learning_rate": 8.462723907307641e-06,
      "loss": 0.3871,
      "step": 17740
    },
    {
      "epoch": 0.8141434537194254,
      "grad_norm": 0.4668399691581726,
      "learning_rate": 8.462547030745887e-06,
      "loss": 0.4469,
      "step": 17741
    },
    {
      "epoch": 0.8141893442246799,
      "grad_norm": 0.4656510651111603,
      "learning_rate": 8.4623701458578e-06,
      "loss": 0.3728,
      "step": 17742
    },
    {
      "epoch": 0.8142352347299344,
      "grad_norm": 0.4243941903114319,
      "learning_rate": 8.462193252643805e-06,
      "loss": 0.3588,
      "step": 17743
    },
    {
      "epoch": 0.8142811252351888,
      "grad_norm": 0.4399278163909912,
      "learning_rate": 8.462016351104327e-06,
      "loss": 0.3382,
      "step": 17744
    },
    {
      "epoch": 0.8143270157404433,
      "grad_norm": 0.48134341835975647,
      "learning_rate": 8.461839441239794e-06,
      "loss": 0.4667,
      "step": 17745
    },
    {
      "epoch": 0.8143729062456978,
      "grad_norm": 0.42967841029167175,
      "learning_rate": 8.46166252305063e-06,
      "loss": 0.3647,
      "step": 17746
    },
    {
      "epoch": 0.8144187967509522,
      "grad_norm": 0.469441682100296,
      "learning_rate": 8.461485596537257e-06,
      "loss": 0.4075,
      "step": 17747
    },
    {
      "epoch": 0.8144646872562067,
      "grad_norm": 0.43080922961235046,
      "learning_rate": 8.461308661700107e-06,
      "loss": 0.3914,
      "step": 17748
    },
    {
      "epoch": 0.8145105777614612,
      "grad_norm": 0.444606751203537,
      "learning_rate": 8.461131718539602e-06,
      "loss": 0.3824,
      "step": 17749
    },
    {
      "epoch": 0.8145564682667156,
      "grad_norm": 0.4257248044013977,
      "learning_rate": 8.460954767056166e-06,
      "loss": 0.3498,
      "step": 17750
    },
    {
      "epoch": 0.8146023587719701,
      "grad_norm": 0.41601067781448364,
      "learning_rate": 8.460777807250226e-06,
      "loss": 0.3451,
      "step": 17751
    },
    {
      "epoch": 0.8146482492772246,
      "grad_norm": 0.4621587097644806,
      "learning_rate": 8.460600839122208e-06,
      "loss": 0.4252,
      "step": 17752
    },
    {
      "epoch": 0.814694139782479,
      "grad_norm": 0.4444040060043335,
      "learning_rate": 8.460423862672537e-06,
      "loss": 0.3813,
      "step": 17753
    },
    {
      "epoch": 0.8147400302877335,
      "grad_norm": 0.39249908924102783,
      "learning_rate": 8.46024687790164e-06,
      "loss": 0.2776,
      "step": 17754
    },
    {
      "epoch": 0.814785920792988,
      "grad_norm": 0.45111456513404846,
      "learning_rate": 8.46006988480994e-06,
      "loss": 0.3825,
      "step": 17755
    },
    {
      "epoch": 0.8148318112982424,
      "grad_norm": 0.4661438763141632,
      "learning_rate": 8.459892883397866e-06,
      "loss": 0.4689,
      "step": 17756
    },
    {
      "epoch": 0.8148777018034968,
      "grad_norm": 0.48406240344047546,
      "learning_rate": 8.459715873665839e-06,
      "loss": 0.4558,
      "step": 17757
    },
    {
      "epoch": 0.8149235923087513,
      "grad_norm": 0.4763455390930176,
      "learning_rate": 8.459538855614289e-06,
      "loss": 0.4329,
      "step": 17758
    },
    {
      "epoch": 0.8149694828140058,
      "grad_norm": 0.5002901554107666,
      "learning_rate": 8.45936182924364e-06,
      "loss": 0.4089,
      "step": 17759
    },
    {
      "epoch": 0.8150153733192602,
      "grad_norm": 0.46355754137039185,
      "learning_rate": 8.459184794554316e-06,
      "loss": 0.4187,
      "step": 17760
    },
    {
      "epoch": 0.8150612638245147,
      "grad_norm": 0.4148882031440735,
      "learning_rate": 8.459007751546746e-06,
      "loss": 0.3309,
      "step": 17761
    },
    {
      "epoch": 0.8151071543297692,
      "grad_norm": 0.4791349470615387,
      "learning_rate": 8.458830700221353e-06,
      "loss": 0.4112,
      "step": 17762
    },
    {
      "epoch": 0.8151530448350236,
      "grad_norm": 0.4429086744785309,
      "learning_rate": 8.458653640578564e-06,
      "loss": 0.3804,
      "step": 17763
    },
    {
      "epoch": 0.8151989353402781,
      "grad_norm": 0.48680782318115234,
      "learning_rate": 8.458476572618804e-06,
      "loss": 0.4349,
      "step": 17764
    },
    {
      "epoch": 0.8152448258455326,
      "grad_norm": 0.44916871190071106,
      "learning_rate": 8.4582994963425e-06,
      "loss": 0.3886,
      "step": 17765
    },
    {
      "epoch": 0.815290716350787,
      "grad_norm": 0.44826626777648926,
      "learning_rate": 8.458122411750078e-06,
      "loss": 0.3648,
      "step": 17766
    },
    {
      "epoch": 0.8153366068560415,
      "grad_norm": 0.4474397301673889,
      "learning_rate": 8.457945318841962e-06,
      "loss": 0.3615,
      "step": 17767
    },
    {
      "epoch": 0.815382497361296,
      "grad_norm": 0.4609276056289673,
      "learning_rate": 8.457768217618578e-06,
      "loss": 0.3807,
      "step": 17768
    },
    {
      "epoch": 0.8154283878665504,
      "grad_norm": 0.4670076370239258,
      "learning_rate": 8.457591108080354e-06,
      "loss": 0.4156,
      "step": 17769
    },
    {
      "epoch": 0.8154742783718049,
      "grad_norm": 0.4769991636276245,
      "learning_rate": 8.457413990227714e-06,
      "loss": 0.469,
      "step": 17770
    },
    {
      "epoch": 0.8155201688770594,
      "grad_norm": 0.384021133184433,
      "learning_rate": 8.457236864061085e-06,
      "loss": 0.2575,
      "step": 17771
    },
    {
      "epoch": 0.8155660593823137,
      "grad_norm": 0.4797494411468506,
      "learning_rate": 8.457059729580894e-06,
      "loss": 0.4431,
      "step": 17772
    },
    {
      "epoch": 0.8156119498875682,
      "grad_norm": 0.4846058487892151,
      "learning_rate": 8.456882586787562e-06,
      "loss": 0.4859,
      "step": 17773
    },
    {
      "epoch": 0.8156578403928227,
      "grad_norm": 0.48710381984710693,
      "learning_rate": 8.456705435681522e-06,
      "loss": 0.4823,
      "step": 17774
    },
    {
      "epoch": 0.8157037308980772,
      "grad_norm": 0.4556836187839508,
      "learning_rate": 8.456528276263193e-06,
      "loss": 0.4011,
      "step": 17775
    },
    {
      "epoch": 0.8157496214033316,
      "grad_norm": 0.6782156229019165,
      "learning_rate": 8.456351108533007e-06,
      "loss": 0.3793,
      "step": 17776
    },
    {
      "epoch": 0.8157955119085861,
      "grad_norm": 0.4809435307979584,
      "learning_rate": 8.456173932491387e-06,
      "loss": 0.3905,
      "step": 17777
    },
    {
      "epoch": 0.8158414024138406,
      "grad_norm": 0.42793768644332886,
      "learning_rate": 8.45599674813876e-06,
      "loss": 0.3663,
      "step": 17778
    },
    {
      "epoch": 0.815887292919095,
      "grad_norm": 0.45739156007766724,
      "learning_rate": 8.455819555475548e-06,
      "loss": 0.4303,
      "step": 17779
    },
    {
      "epoch": 0.8159331834243495,
      "grad_norm": 0.43945157527923584,
      "learning_rate": 8.455642354502185e-06,
      "loss": 0.3708,
      "step": 17780
    },
    {
      "epoch": 0.815979073929604,
      "grad_norm": 0.5045421123504639,
      "learning_rate": 8.455465145219091e-06,
      "loss": 0.379,
      "step": 17781
    },
    {
      "epoch": 0.8160249644348584,
      "grad_norm": 0.46192875504493713,
      "learning_rate": 8.455287927626694e-06,
      "loss": 0.4686,
      "step": 17782
    },
    {
      "epoch": 0.8160708549401129,
      "grad_norm": 0.48617199063301086,
      "learning_rate": 8.45511070172542e-06,
      "loss": 0.4485,
      "step": 17783
    },
    {
      "epoch": 0.8161167454453674,
      "grad_norm": 0.5151456594467163,
      "learning_rate": 8.454933467515695e-06,
      "loss": 0.404,
      "step": 17784
    },
    {
      "epoch": 0.8161626359506218,
      "grad_norm": 0.5029243230819702,
      "learning_rate": 8.454756224997945e-06,
      "loss": 0.4266,
      "step": 17785
    },
    {
      "epoch": 0.8162085264558763,
      "grad_norm": 0.5284098386764526,
      "learning_rate": 8.454578974172598e-06,
      "loss": 0.4798,
      "step": 17786
    },
    {
      "epoch": 0.8162544169611308,
      "grad_norm": 0.552433431148529,
      "learning_rate": 8.454401715040077e-06,
      "loss": 0.4519,
      "step": 17787
    },
    {
      "epoch": 0.8163003074663852,
      "grad_norm": 0.44040995836257935,
      "learning_rate": 8.454224447600812e-06,
      "loss": 0.3602,
      "step": 17788
    },
    {
      "epoch": 0.8163461979716397,
      "grad_norm": 0.4876677989959717,
      "learning_rate": 8.454047171855227e-06,
      "loss": 0.4315,
      "step": 17789
    },
    {
      "epoch": 0.8163920884768942,
      "grad_norm": 0.5032410025596619,
      "learning_rate": 8.453869887803747e-06,
      "loss": 0.5389,
      "step": 17790
    },
    {
      "epoch": 0.8164379789821486,
      "grad_norm": 0.4923698902130127,
      "learning_rate": 8.453692595446803e-06,
      "loss": 0.4377,
      "step": 17791
    },
    {
      "epoch": 0.816483869487403,
      "grad_norm": 0.44457751512527466,
      "learning_rate": 8.453515294784815e-06,
      "loss": 0.3908,
      "step": 17792
    },
    {
      "epoch": 0.8165297599926575,
      "grad_norm": 0.44208553433418274,
      "learning_rate": 8.453337985818215e-06,
      "loss": 0.3672,
      "step": 17793
    },
    {
      "epoch": 0.816575650497912,
      "grad_norm": 0.45864975452423096,
      "learning_rate": 8.453160668547426e-06,
      "loss": 0.4236,
      "step": 17794
    },
    {
      "epoch": 0.8166215410031664,
      "grad_norm": 0.5064997673034668,
      "learning_rate": 8.452983342972875e-06,
      "loss": 0.4952,
      "step": 17795
    },
    {
      "epoch": 0.8166674315084209,
      "grad_norm": 0.473934531211853,
      "learning_rate": 8.452806009094992e-06,
      "loss": 0.4459,
      "step": 17796
    },
    {
      "epoch": 0.8167133220136754,
      "grad_norm": 0.47915735840797424,
      "learning_rate": 8.452628666914197e-06,
      "loss": 0.4555,
      "step": 17797
    },
    {
      "epoch": 0.8167592125189298,
      "grad_norm": 0.41730380058288574,
      "learning_rate": 8.452451316430921e-06,
      "loss": 0.3159,
      "step": 17798
    },
    {
      "epoch": 0.8168051030241843,
      "grad_norm": 0.44993406534194946,
      "learning_rate": 8.452273957645589e-06,
      "loss": 0.4068,
      "step": 17799
    },
    {
      "epoch": 0.8168509935294388,
      "grad_norm": 0.4649324119091034,
      "learning_rate": 8.452096590558627e-06,
      "loss": 0.4428,
      "step": 17800
    },
    {
      "epoch": 0.8168968840346932,
      "grad_norm": 0.49509313702583313,
      "learning_rate": 8.451919215170463e-06,
      "loss": 0.4426,
      "step": 17801
    },
    {
      "epoch": 0.8169427745399477,
      "grad_norm": 0.4350048899650574,
      "learning_rate": 8.451741831481522e-06,
      "loss": 0.4111,
      "step": 17802
    },
    {
      "epoch": 0.8169886650452022,
      "grad_norm": 0.4675920307636261,
      "learning_rate": 8.451564439492232e-06,
      "loss": 0.4116,
      "step": 17803
    },
    {
      "epoch": 0.8170345555504566,
      "grad_norm": 0.49004170298576355,
      "learning_rate": 8.451387039203019e-06,
      "loss": 0.4487,
      "step": 17804
    },
    {
      "epoch": 0.8170804460557111,
      "grad_norm": 0.44811636209487915,
      "learning_rate": 8.451209630614309e-06,
      "loss": 0.3915,
      "step": 17805
    },
    {
      "epoch": 0.8171263365609656,
      "grad_norm": 0.4539348781108856,
      "learning_rate": 8.451032213726528e-06,
      "loss": 0.3384,
      "step": 17806
    },
    {
      "epoch": 0.81717222706622,
      "grad_norm": 0.4370141327381134,
      "learning_rate": 8.450854788540106e-06,
      "loss": 0.4094,
      "step": 17807
    },
    {
      "epoch": 0.8172181175714744,
      "grad_norm": 0.42220792174339294,
      "learning_rate": 8.450677355055465e-06,
      "loss": 0.367,
      "step": 17808
    },
    {
      "epoch": 0.8172640080767289,
      "grad_norm": 0.45550698041915894,
      "learning_rate": 8.450499913273037e-06,
      "loss": 0.3917,
      "step": 17809
    },
    {
      "epoch": 0.8173098985819834,
      "grad_norm": 0.40571096539497375,
      "learning_rate": 8.450322463193243e-06,
      "loss": 0.3088,
      "step": 17810
    },
    {
      "epoch": 0.8173557890872378,
      "grad_norm": 0.44377827644348145,
      "learning_rate": 8.450145004816513e-06,
      "loss": 0.3963,
      "step": 17811
    },
    {
      "epoch": 0.8174016795924923,
      "grad_norm": 0.4445282816886902,
      "learning_rate": 8.449967538143274e-06,
      "loss": 0.3661,
      "step": 17812
    },
    {
      "epoch": 0.8174475700977468,
      "grad_norm": 0.43142879009246826,
      "learning_rate": 8.449790063173953e-06,
      "loss": 0.3811,
      "step": 17813
    },
    {
      "epoch": 0.8174934606030012,
      "grad_norm": 0.43431249260902405,
      "learning_rate": 8.449612579908972e-06,
      "loss": 0.3608,
      "step": 17814
    },
    {
      "epoch": 0.8175393511082557,
      "grad_norm": 0.4378563463687897,
      "learning_rate": 8.449435088348765e-06,
      "loss": 0.3352,
      "step": 17815
    },
    {
      "epoch": 0.8175852416135102,
      "grad_norm": 0.41473838686943054,
      "learning_rate": 8.449257588493753e-06,
      "loss": 0.3297,
      "step": 17816
    },
    {
      "epoch": 0.8176311321187646,
      "grad_norm": 0.5034177899360657,
      "learning_rate": 8.449080080344367e-06,
      "loss": 0.4264,
      "step": 17817
    },
    {
      "epoch": 0.8176770226240191,
      "grad_norm": 0.4560909569263458,
      "learning_rate": 8.448902563901031e-06,
      "loss": 0.3817,
      "step": 17818
    },
    {
      "epoch": 0.8177229131292736,
      "grad_norm": 0.4419739842414856,
      "learning_rate": 8.448725039164174e-06,
      "loss": 0.3791,
      "step": 17819
    },
    {
      "epoch": 0.817768803634528,
      "grad_norm": 0.4512062966823578,
      "learning_rate": 8.44854750613422e-06,
      "loss": 0.3826,
      "step": 17820
    },
    {
      "epoch": 0.8178146941397825,
      "grad_norm": 0.4667685627937317,
      "learning_rate": 8.448369964811598e-06,
      "loss": 0.4273,
      "step": 17821
    },
    {
      "epoch": 0.817860584645037,
      "grad_norm": 0.4926930367946625,
      "learning_rate": 8.448192415196734e-06,
      "loss": 0.4432,
      "step": 17822
    },
    {
      "epoch": 0.8179064751502914,
      "grad_norm": 0.47055089473724365,
      "learning_rate": 8.448014857290057e-06,
      "loss": 0.4775,
      "step": 17823
    },
    {
      "epoch": 0.8179523656555459,
      "grad_norm": 0.4343009293079376,
      "learning_rate": 8.447837291091992e-06,
      "loss": 0.3395,
      "step": 17824
    },
    {
      "epoch": 0.8179982561608004,
      "grad_norm": 0.4633493423461914,
      "learning_rate": 8.447659716602965e-06,
      "loss": 0.3775,
      "step": 17825
    },
    {
      "epoch": 0.8180441466660547,
      "grad_norm": 0.4561808705329895,
      "learning_rate": 8.447482133823407e-06,
      "loss": 0.4117,
      "step": 17826
    },
    {
      "epoch": 0.8180900371713092,
      "grad_norm": 0.44060763716697693,
      "learning_rate": 8.447304542753741e-06,
      "loss": 0.3724,
      "step": 17827
    },
    {
      "epoch": 0.8181359276765637,
      "grad_norm": 0.4121582806110382,
      "learning_rate": 8.447126943394397e-06,
      "loss": 0.3328,
      "step": 17828
    },
    {
      "epoch": 0.8181818181818182,
      "grad_norm": 0.4680311977863312,
      "learning_rate": 8.4469493357458e-06,
      "loss": 0.3787,
      "step": 17829
    },
    {
      "epoch": 0.8182277086870726,
      "grad_norm": 0.6310790777206421,
      "learning_rate": 8.446771719808376e-06,
      "loss": 0.3654,
      "step": 17830
    },
    {
      "epoch": 0.8182735991923271,
      "grad_norm": 0.45473721623420715,
      "learning_rate": 8.446594095582555e-06,
      "loss": 0.372,
      "step": 17831
    },
    {
      "epoch": 0.8183194896975816,
      "grad_norm": 0.44879406690597534,
      "learning_rate": 8.446416463068763e-06,
      "loss": 0.3656,
      "step": 17832
    },
    {
      "epoch": 0.818365380202836,
      "grad_norm": 0.45837101340293884,
      "learning_rate": 8.446238822267428e-06,
      "loss": 0.436,
      "step": 17833
    },
    {
      "epoch": 0.8184112707080905,
      "grad_norm": 0.4236229360103607,
      "learning_rate": 8.446061173178975e-06,
      "loss": 0.3522,
      "step": 17834
    },
    {
      "epoch": 0.818457161213345,
      "grad_norm": 0.47889837622642517,
      "learning_rate": 8.445883515803834e-06,
      "loss": 0.4767,
      "step": 17835
    },
    {
      "epoch": 0.8185030517185994,
      "grad_norm": 0.4336996376514435,
      "learning_rate": 8.445705850142429e-06,
      "loss": 0.3264,
      "step": 17836
    },
    {
      "epoch": 0.8185489422238539,
      "grad_norm": 0.42929285764694214,
      "learning_rate": 8.44552817619519e-06,
      "loss": 0.3331,
      "step": 17837
    },
    {
      "epoch": 0.8185948327291084,
      "grad_norm": 0.4646855592727661,
      "learning_rate": 8.445350493962542e-06,
      "loss": 0.4369,
      "step": 17838
    },
    {
      "epoch": 0.8186407232343628,
      "grad_norm": 0.4273969531059265,
      "learning_rate": 8.445172803444915e-06,
      "loss": 0.3411,
      "step": 17839
    },
    {
      "epoch": 0.8186866137396173,
      "grad_norm": 0.4667750298976898,
      "learning_rate": 8.444995104642734e-06,
      "loss": 0.4047,
      "step": 17840
    },
    {
      "epoch": 0.8187325042448718,
      "grad_norm": 0.4668880105018616,
      "learning_rate": 8.444817397556426e-06,
      "loss": 0.3899,
      "step": 17841
    },
    {
      "epoch": 0.8187783947501261,
      "grad_norm": 0.4907797873020172,
      "learning_rate": 8.444639682186421e-06,
      "loss": 0.4785,
      "step": 17842
    },
    {
      "epoch": 0.8188242852553806,
      "grad_norm": 0.45590436458587646,
      "learning_rate": 8.444461958533146e-06,
      "loss": 0.3761,
      "step": 17843
    },
    {
      "epoch": 0.8188701757606351,
      "grad_norm": 0.42426902055740356,
      "learning_rate": 8.444284226597023e-06,
      "loss": 0.3575,
      "step": 17844
    },
    {
      "epoch": 0.8189160662658896,
      "grad_norm": 0.4448358416557312,
      "learning_rate": 8.444106486378486e-06,
      "loss": 0.4036,
      "step": 17845
    },
    {
      "epoch": 0.818961956771144,
      "grad_norm": 0.45628583431243896,
      "learning_rate": 8.44392873787796e-06,
      "loss": 0.406,
      "step": 17846
    },
    {
      "epoch": 0.8190078472763985,
      "grad_norm": 0.43884557485580444,
      "learning_rate": 8.443750981095873e-06,
      "loss": 0.3797,
      "step": 17847
    },
    {
      "epoch": 0.819053737781653,
      "grad_norm": 0.4737260639667511,
      "learning_rate": 8.44357321603265e-06,
      "loss": 0.4298,
      "step": 17848
    },
    {
      "epoch": 0.8190996282869074,
      "grad_norm": 0.43591901659965515,
      "learning_rate": 8.443395442688721e-06,
      "loss": 0.3548,
      "step": 17849
    },
    {
      "epoch": 0.8191455187921619,
      "grad_norm": 0.46873801946640015,
      "learning_rate": 8.443217661064513e-06,
      "loss": 0.4097,
      "step": 17850
    },
    {
      "epoch": 0.8191914092974164,
      "grad_norm": 0.4389280378818512,
      "learning_rate": 8.443039871160452e-06,
      "loss": 0.3992,
      "step": 17851
    },
    {
      "epoch": 0.8192372998026708,
      "grad_norm": 0.4239269495010376,
      "learning_rate": 8.442862072976968e-06,
      "loss": 0.3839,
      "step": 17852
    },
    {
      "epoch": 0.8192831903079253,
      "grad_norm": 0.44302040338516235,
      "learning_rate": 8.442684266514488e-06,
      "loss": 0.3586,
      "step": 17853
    },
    {
      "epoch": 0.8193290808131798,
      "grad_norm": 0.4616401493549347,
      "learning_rate": 8.442506451773437e-06,
      "loss": 0.4071,
      "step": 17854
    },
    {
      "epoch": 0.8193749713184342,
      "grad_norm": 0.4629646837711334,
      "learning_rate": 8.442328628754244e-06,
      "loss": 0.4824,
      "step": 17855
    },
    {
      "epoch": 0.8194208618236887,
      "grad_norm": 0.5071123242378235,
      "learning_rate": 8.44215079745734e-06,
      "loss": 0.464,
      "step": 17856
    },
    {
      "epoch": 0.8194667523289432,
      "grad_norm": 0.4645283818244934,
      "learning_rate": 8.44197295788315e-06,
      "loss": 0.4206,
      "step": 17857
    },
    {
      "epoch": 0.8195126428341976,
      "grad_norm": 0.4429694712162018,
      "learning_rate": 8.441795110032098e-06,
      "loss": 0.4576,
      "step": 17858
    },
    {
      "epoch": 0.8195585333394521,
      "grad_norm": 0.5408008098602295,
      "learning_rate": 8.441617253904615e-06,
      "loss": 0.4782,
      "step": 17859
    },
    {
      "epoch": 0.8196044238447066,
      "grad_norm": 0.4746640622615814,
      "learning_rate": 8.441439389501132e-06,
      "loss": 0.4357,
      "step": 17860
    },
    {
      "epoch": 0.8196503143499609,
      "grad_norm": 0.47243228554725647,
      "learning_rate": 8.44126151682207e-06,
      "loss": 0.4166,
      "step": 17861
    },
    {
      "epoch": 0.8196962048552154,
      "grad_norm": 0.45268604159355164,
      "learning_rate": 8.441083635867862e-06,
      "loss": 0.3731,
      "step": 17862
    },
    {
      "epoch": 0.8197420953604699,
      "grad_norm": 0.5436232089996338,
      "learning_rate": 8.440905746638934e-06,
      "loss": 0.517,
      "step": 17863
    },
    {
      "epoch": 0.8197879858657244,
      "grad_norm": 0.5798476338386536,
      "learning_rate": 8.440727849135714e-06,
      "loss": 0.5543,
      "step": 17864
    },
    {
      "epoch": 0.8198338763709788,
      "grad_norm": 0.4424605965614319,
      "learning_rate": 8.440549943358629e-06,
      "loss": 0.4041,
      "step": 17865
    },
    {
      "epoch": 0.8198797668762333,
      "grad_norm": 0.42778280377388,
      "learning_rate": 8.440372029308105e-06,
      "loss": 0.3294,
      "step": 17866
    },
    {
      "epoch": 0.8199256573814878,
      "grad_norm": 0.4639279544353485,
      "learning_rate": 8.440194106984576e-06,
      "loss": 0.4592,
      "step": 17867
    },
    {
      "epoch": 0.8199715478867422,
      "grad_norm": 0.42307767271995544,
      "learning_rate": 8.440016176388463e-06,
      "loss": 0.3687,
      "step": 17868
    },
    {
      "epoch": 0.8200174383919967,
      "grad_norm": 0.4594457447528839,
      "learning_rate": 8.439838237520197e-06,
      "loss": 0.4395,
      "step": 17869
    },
    {
      "epoch": 0.8200633288972512,
      "grad_norm": 0.47358983755111694,
      "learning_rate": 8.439660290380206e-06,
      "loss": 0.4527,
      "step": 17870
    },
    {
      "epoch": 0.8201092194025056,
      "grad_norm": 0.43438515067100525,
      "learning_rate": 8.43948233496892e-06,
      "loss": 0.3603,
      "step": 17871
    },
    {
      "epoch": 0.8201551099077601,
      "grad_norm": 0.4331245422363281,
      "learning_rate": 8.439304371286764e-06,
      "loss": 0.357,
      "step": 17872
    },
    {
      "epoch": 0.8202010004130146,
      "grad_norm": 0.4388468563556671,
      "learning_rate": 8.439126399334164e-06,
      "loss": 0.366,
      "step": 17873
    },
    {
      "epoch": 0.820246890918269,
      "grad_norm": 0.45032402873039246,
      "learning_rate": 8.438948419111553e-06,
      "loss": 0.3719,
      "step": 17874
    },
    {
      "epoch": 0.8202927814235235,
      "grad_norm": 0.4262039065361023,
      "learning_rate": 8.438770430619354e-06,
      "loss": 0.3166,
      "step": 17875
    },
    {
      "epoch": 0.820338671928778,
      "grad_norm": 0.4503183662891388,
      "learning_rate": 8.438592433857998e-06,
      "loss": 0.3877,
      "step": 17876
    },
    {
      "epoch": 0.8203845624340323,
      "grad_norm": 0.4305194914340973,
      "learning_rate": 8.438414428827913e-06,
      "loss": 0.362,
      "step": 17877
    },
    {
      "epoch": 0.8204304529392868,
      "grad_norm": 0.494642436504364,
      "learning_rate": 8.438236415529526e-06,
      "loss": 0.4674,
      "step": 17878
    },
    {
      "epoch": 0.8204763434445413,
      "grad_norm": 0.4419185519218445,
      "learning_rate": 8.438058393963265e-06,
      "loss": 0.359,
      "step": 17879
    },
    {
      "epoch": 0.8205222339497957,
      "grad_norm": 0.4418424069881439,
      "learning_rate": 8.43788036412956e-06,
      "loss": 0.3533,
      "step": 17880
    },
    {
      "epoch": 0.8205681244550502,
      "grad_norm": 0.44324707984924316,
      "learning_rate": 8.437702326028837e-06,
      "loss": 0.3828,
      "step": 17881
    },
    {
      "epoch": 0.8206140149603047,
      "grad_norm": 0.4626299738883972,
      "learning_rate": 8.437524279661527e-06,
      "loss": 0.4356,
      "step": 17882
    },
    {
      "epoch": 0.8206599054655592,
      "grad_norm": 0.45641717314720154,
      "learning_rate": 8.437346225028054e-06,
      "loss": 0.404,
      "step": 17883
    },
    {
      "epoch": 0.8207057959708136,
      "grad_norm": 0.4591040313243866,
      "learning_rate": 8.437168162128847e-06,
      "loss": 0.385,
      "step": 17884
    },
    {
      "epoch": 0.8207516864760681,
      "grad_norm": 0.4994543790817261,
      "learning_rate": 8.436990090964337e-06,
      "loss": 0.5062,
      "step": 17885
    },
    {
      "epoch": 0.8207975769813226,
      "grad_norm": 0.4303171634674072,
      "learning_rate": 8.436812011534949e-06,
      "loss": 0.3609,
      "step": 17886
    },
    {
      "epoch": 0.820843467486577,
      "grad_norm": 0.7444688081741333,
      "learning_rate": 8.436633923841116e-06,
      "loss": 0.4836,
      "step": 17887
    },
    {
      "epoch": 0.8208893579918315,
      "grad_norm": 0.41902852058410645,
      "learning_rate": 8.43645582788326e-06,
      "loss": 0.3326,
      "step": 17888
    },
    {
      "epoch": 0.820935248497086,
      "grad_norm": 0.4969289302825928,
      "learning_rate": 8.436277723661814e-06,
      "loss": 0.4586,
      "step": 17889
    },
    {
      "epoch": 0.8209811390023404,
      "grad_norm": 0.44333550333976746,
      "learning_rate": 8.436099611177203e-06,
      "loss": 0.4034,
      "step": 17890
    },
    {
      "epoch": 0.8210270295075949,
      "grad_norm": 0.46270135045051575,
      "learning_rate": 8.43592149042986e-06,
      "loss": 0.3446,
      "step": 17891
    },
    {
      "epoch": 0.8210729200128494,
      "grad_norm": 0.48731061816215515,
      "learning_rate": 8.435743361420206e-06,
      "loss": 0.4634,
      "step": 17892
    },
    {
      "epoch": 0.8211188105181038,
      "grad_norm": 0.45579615235328674,
      "learning_rate": 8.435565224148676e-06,
      "loss": 0.4197,
      "step": 17893
    },
    {
      "epoch": 0.8211647010233583,
      "grad_norm": 0.45256131887435913,
      "learning_rate": 8.435387078615696e-06,
      "loss": 0.4199,
      "step": 17894
    },
    {
      "epoch": 0.8212105915286128,
      "grad_norm": 0.44706952571868896,
      "learning_rate": 8.435208924821694e-06,
      "loss": 0.332,
      "step": 17895
    },
    {
      "epoch": 0.8212564820338671,
      "grad_norm": 0.4572947025299072,
      "learning_rate": 8.4350307627671e-06,
      "loss": 0.3887,
      "step": 17896
    },
    {
      "epoch": 0.8213023725391216,
      "grad_norm": 0.41217607259750366,
      "learning_rate": 8.434852592452338e-06,
      "loss": 0.3355,
      "step": 17897
    },
    {
      "epoch": 0.8213482630443761,
      "grad_norm": 0.4582231938838959,
      "learning_rate": 8.434674413877841e-06,
      "loss": 0.4259,
      "step": 17898
    },
    {
      "epoch": 0.8213941535496306,
      "grad_norm": 0.4440719485282898,
      "learning_rate": 8.434496227044038e-06,
      "loss": 0.4137,
      "step": 17899
    },
    {
      "epoch": 0.821440044054885,
      "grad_norm": 0.4457356035709381,
      "learning_rate": 8.434318031951354e-06,
      "loss": 0.4449,
      "step": 17900
    },
    {
      "epoch": 0.8214859345601395,
      "grad_norm": 0.508654773235321,
      "learning_rate": 8.434139828600219e-06,
      "loss": 0.4741,
      "step": 17901
    },
    {
      "epoch": 0.821531825065394,
      "grad_norm": 0.46433717012405396,
      "learning_rate": 8.433961616991061e-06,
      "loss": 0.4628,
      "step": 17902
    },
    {
      "epoch": 0.8215777155706484,
      "grad_norm": 0.50074303150177,
      "learning_rate": 8.43378339712431e-06,
      "loss": 0.3391,
      "step": 17903
    },
    {
      "epoch": 0.8216236060759029,
      "grad_norm": 0.4183187782764435,
      "learning_rate": 8.433605169000393e-06,
      "loss": 0.2966,
      "step": 17904
    },
    {
      "epoch": 0.8216694965811574,
      "grad_norm": 0.4308905601501465,
      "learning_rate": 8.43342693261974e-06,
      "loss": 0.3542,
      "step": 17905
    },
    {
      "epoch": 0.8217153870864118,
      "grad_norm": 0.5007399320602417,
      "learning_rate": 8.433248687982778e-06,
      "loss": 0.4733,
      "step": 17906
    },
    {
      "epoch": 0.8217612775916663,
      "grad_norm": 0.47217103838920593,
      "learning_rate": 8.433070435089937e-06,
      "loss": 0.4128,
      "step": 17907
    },
    {
      "epoch": 0.8218071680969208,
      "grad_norm": 0.4739191234111786,
      "learning_rate": 8.432892173941646e-06,
      "loss": 0.4132,
      "step": 17908
    },
    {
      "epoch": 0.8218530586021752,
      "grad_norm": 0.47100046277046204,
      "learning_rate": 8.432713904538331e-06,
      "loss": 0.4096,
      "step": 17909
    },
    {
      "epoch": 0.8218989491074297,
      "grad_norm": 0.45202040672302246,
      "learning_rate": 8.432535626880424e-06,
      "loss": 0.3803,
      "step": 17910
    },
    {
      "epoch": 0.8219448396126842,
      "grad_norm": 0.4423539638519287,
      "learning_rate": 8.43235734096835e-06,
      "loss": 0.3711,
      "step": 17911
    },
    {
      "epoch": 0.8219907301179386,
      "grad_norm": 0.4939267933368683,
      "learning_rate": 8.432179046802542e-06,
      "loss": 0.4421,
      "step": 17912
    },
    {
      "epoch": 0.822036620623193,
      "grad_norm": 0.418667197227478,
      "learning_rate": 8.432000744383426e-06,
      "loss": 0.3186,
      "step": 17913
    },
    {
      "epoch": 0.8220825111284475,
      "grad_norm": 0.4393908381462097,
      "learning_rate": 8.431822433711431e-06,
      "loss": 0.4131,
      "step": 17914
    },
    {
      "epoch": 0.8221284016337019,
      "grad_norm": 0.5270518660545349,
      "learning_rate": 8.431644114786987e-06,
      "loss": 0.5361,
      "step": 17915
    },
    {
      "epoch": 0.8221742921389564,
      "grad_norm": 0.4561271667480469,
      "learning_rate": 8.431465787610523e-06,
      "loss": 0.4399,
      "step": 17916
    },
    {
      "epoch": 0.8222201826442109,
      "grad_norm": 0.4458649754524231,
      "learning_rate": 8.431287452182463e-06,
      "loss": 0.4392,
      "step": 17917
    },
    {
      "epoch": 0.8222660731494654,
      "grad_norm": 0.4482864737510681,
      "learning_rate": 8.431109108503243e-06,
      "loss": 0.3997,
      "step": 17918
    },
    {
      "epoch": 0.8223119636547198,
      "grad_norm": 0.4181709587574005,
      "learning_rate": 8.430930756573287e-06,
      "loss": 0.3454,
      "step": 17919
    },
    {
      "epoch": 0.8223578541599743,
      "grad_norm": 0.4522830545902252,
      "learning_rate": 8.430752396393024e-06,
      "loss": 0.3992,
      "step": 17920
    },
    {
      "epoch": 0.8224037446652288,
      "grad_norm": 0.42259693145751953,
      "learning_rate": 8.430574027962888e-06,
      "loss": 0.3171,
      "step": 17921
    },
    {
      "epoch": 0.8224496351704832,
      "grad_norm": 0.4334827959537506,
      "learning_rate": 8.430395651283302e-06,
      "loss": 0.3478,
      "step": 17922
    },
    {
      "epoch": 0.8224955256757377,
      "grad_norm": 0.4645261764526367,
      "learning_rate": 8.430217266354697e-06,
      "loss": 0.3887,
      "step": 17923
    },
    {
      "epoch": 0.8225414161809922,
      "grad_norm": 0.4732913672924042,
      "learning_rate": 8.430038873177501e-06,
      "loss": 0.3864,
      "step": 17924
    },
    {
      "epoch": 0.8225873066862466,
      "grad_norm": 0.4776921570301056,
      "learning_rate": 8.429860471752146e-06,
      "loss": 0.3955,
      "step": 17925
    },
    {
      "epoch": 0.8226331971915011,
      "grad_norm": 0.5224975943565369,
      "learning_rate": 8.429682062079059e-06,
      "loss": 0.4604,
      "step": 17926
    },
    {
      "epoch": 0.8226790876967556,
      "grad_norm": 0.4713437855243683,
      "learning_rate": 8.429503644158668e-06,
      "loss": 0.4134,
      "step": 17927
    },
    {
      "epoch": 0.82272497820201,
      "grad_norm": 0.43453097343444824,
      "learning_rate": 8.429325217991404e-06,
      "loss": 0.3699,
      "step": 17928
    },
    {
      "epoch": 0.8227708687072645,
      "grad_norm": 0.4282715916633606,
      "learning_rate": 8.429146783577695e-06,
      "loss": 0.3633,
      "step": 17929
    },
    {
      "epoch": 0.822816759212519,
      "grad_norm": 0.4438944160938263,
      "learning_rate": 8.428968340917972e-06,
      "loss": 0.3589,
      "step": 17930
    },
    {
      "epoch": 0.8228626497177733,
      "grad_norm": 0.46914318203926086,
      "learning_rate": 8.428789890012659e-06,
      "loss": 0.4943,
      "step": 17931
    },
    {
      "epoch": 0.8229085402230278,
      "grad_norm": 0.48776641488075256,
      "learning_rate": 8.42861143086219e-06,
      "loss": 0.4223,
      "step": 17932
    },
    {
      "epoch": 0.8229544307282823,
      "grad_norm": 0.42592111229896545,
      "learning_rate": 8.428432963466992e-06,
      "loss": 0.3623,
      "step": 17933
    },
    {
      "epoch": 0.8230003212335368,
      "grad_norm": 0.4786435663700104,
      "learning_rate": 8.428254487827495e-06,
      "loss": 0.438,
      "step": 17934
    },
    {
      "epoch": 0.8230462117387912,
      "grad_norm": 0.4186241030693054,
      "learning_rate": 8.428076003944128e-06,
      "loss": 0.3482,
      "step": 17935
    },
    {
      "epoch": 0.8230921022440457,
      "grad_norm": 0.46027281880378723,
      "learning_rate": 8.427897511817322e-06,
      "loss": 0.4654,
      "step": 17936
    },
    {
      "epoch": 0.8231379927493002,
      "grad_norm": 0.4466524124145508,
      "learning_rate": 8.427719011447502e-06,
      "loss": 0.3654,
      "step": 17937
    },
    {
      "epoch": 0.8231838832545546,
      "grad_norm": 0.4481244683265686,
      "learning_rate": 8.4275405028351e-06,
      "loss": 0.4511,
      "step": 17938
    },
    {
      "epoch": 0.8232297737598091,
      "grad_norm": 0.4062945544719696,
      "learning_rate": 8.427361985980545e-06,
      "loss": 0.2924,
      "step": 17939
    },
    {
      "epoch": 0.8232756642650636,
      "grad_norm": 0.4455844461917877,
      "learning_rate": 8.427183460884267e-06,
      "loss": 0.4099,
      "step": 17940
    },
    {
      "epoch": 0.823321554770318,
      "grad_norm": 0.4624093770980835,
      "learning_rate": 8.427004927546693e-06,
      "loss": 0.3913,
      "step": 17941
    },
    {
      "epoch": 0.8233674452755725,
      "grad_norm": 0.4717477858066559,
      "learning_rate": 8.426826385968253e-06,
      "loss": 0.4078,
      "step": 17942
    },
    {
      "epoch": 0.823413335780827,
      "grad_norm": 0.43445292115211487,
      "learning_rate": 8.42664783614938e-06,
      "loss": 0.4065,
      "step": 17943
    },
    {
      "epoch": 0.8234592262860814,
      "grad_norm": 0.4906216561794281,
      "learning_rate": 8.426469278090498e-06,
      "loss": 0.4367,
      "step": 17944
    },
    {
      "epoch": 0.8235051167913359,
      "grad_norm": 0.4314855635166168,
      "learning_rate": 8.426290711792042e-06,
      "loss": 0.3329,
      "step": 17945
    },
    {
      "epoch": 0.8235510072965904,
      "grad_norm": 0.47723203897476196,
      "learning_rate": 8.426112137254435e-06,
      "loss": 0.4804,
      "step": 17946
    },
    {
      "epoch": 0.8235968978018448,
      "grad_norm": 0.43511807918548584,
      "learning_rate": 8.425933554478112e-06,
      "loss": 0.3578,
      "step": 17947
    },
    {
      "epoch": 0.8236427883070992,
      "grad_norm": 0.4902743399143219,
      "learning_rate": 8.425754963463497e-06,
      "loss": 0.388,
      "step": 17948
    },
    {
      "epoch": 0.8236886788123537,
      "grad_norm": 0.4480859041213989,
      "learning_rate": 8.425576364211025e-06,
      "loss": 0.3628,
      "step": 17949
    },
    {
      "epoch": 0.8237345693176081,
      "grad_norm": 0.45992904901504517,
      "learning_rate": 8.425397756721121e-06,
      "loss": 0.3921,
      "step": 17950
    },
    {
      "epoch": 0.8237804598228626,
      "grad_norm": 0.4630306661128998,
      "learning_rate": 8.425219140994218e-06,
      "loss": 0.3884,
      "step": 17951
    },
    {
      "epoch": 0.8238263503281171,
      "grad_norm": 0.4694446921348572,
      "learning_rate": 8.425040517030745e-06,
      "loss": 0.4498,
      "step": 17952
    },
    {
      "epoch": 0.8238722408333716,
      "grad_norm": 0.463346391916275,
      "learning_rate": 8.424861884831128e-06,
      "loss": 0.3904,
      "step": 17953
    },
    {
      "epoch": 0.823918131338626,
      "grad_norm": 0.46538862586021423,
      "learning_rate": 8.424683244395801e-06,
      "loss": 0.4092,
      "step": 17954
    },
    {
      "epoch": 0.8239640218438805,
      "grad_norm": 0.47268155217170715,
      "learning_rate": 8.42450459572519e-06,
      "loss": 0.4475,
      "step": 17955
    },
    {
      "epoch": 0.824009912349135,
      "grad_norm": 0.43658512830734253,
      "learning_rate": 8.424325938819728e-06,
      "loss": 0.4048,
      "step": 17956
    },
    {
      "epoch": 0.8240558028543894,
      "grad_norm": 0.4581782817840576,
      "learning_rate": 8.424147273679842e-06,
      "loss": 0.43,
      "step": 17957
    },
    {
      "epoch": 0.8241016933596439,
      "grad_norm": 0.47186845541000366,
      "learning_rate": 8.423968600305963e-06,
      "loss": 0.4317,
      "step": 17958
    },
    {
      "epoch": 0.8241475838648984,
      "grad_norm": 0.4590403139591217,
      "learning_rate": 8.42378991869852e-06,
      "loss": 0.4038,
      "step": 17959
    },
    {
      "epoch": 0.8241934743701528,
      "grad_norm": 0.48367607593536377,
      "learning_rate": 8.423611228857942e-06,
      "loss": 0.4672,
      "step": 17960
    },
    {
      "epoch": 0.8242393648754073,
      "grad_norm": 0.4666590392589569,
      "learning_rate": 8.423432530784659e-06,
      "loss": 0.4204,
      "step": 17961
    },
    {
      "epoch": 0.8242852553806618,
      "grad_norm": 0.41143351793289185,
      "learning_rate": 8.423253824479104e-06,
      "loss": 0.3132,
      "step": 17962
    },
    {
      "epoch": 0.8243311458859162,
      "grad_norm": 0.41674065589904785,
      "learning_rate": 8.4230751099417e-06,
      "loss": 0.3145,
      "step": 17963
    },
    {
      "epoch": 0.8243770363911707,
      "grad_norm": 0.44985729455947876,
      "learning_rate": 8.422896387172884e-06,
      "loss": 0.4251,
      "step": 17964
    },
    {
      "epoch": 0.8244229268964252,
      "grad_norm": 0.43448880314826965,
      "learning_rate": 8.422717656173083e-06,
      "loss": 0.3628,
      "step": 17965
    },
    {
      "epoch": 0.8244688174016795,
      "grad_norm": 0.4244597554206848,
      "learning_rate": 8.422538916942723e-06,
      "loss": 0.3223,
      "step": 17966
    },
    {
      "epoch": 0.824514707906934,
      "grad_norm": 0.42590245604515076,
      "learning_rate": 8.42236016948224e-06,
      "loss": 0.3684,
      "step": 17967
    },
    {
      "epoch": 0.8245605984121885,
      "grad_norm": 0.47466611862182617,
      "learning_rate": 8.42218141379206e-06,
      "loss": 0.4639,
      "step": 17968
    },
    {
      "epoch": 0.8246064889174429,
      "grad_norm": 0.5025848746299744,
      "learning_rate": 8.422002649872612e-06,
      "loss": 0.4968,
      "step": 17969
    },
    {
      "epoch": 0.8246523794226974,
      "grad_norm": 0.4497338831424713,
      "learning_rate": 8.42182387772433e-06,
      "loss": 0.391,
      "step": 17970
    },
    {
      "epoch": 0.8246982699279519,
      "grad_norm": 0.4293939769268036,
      "learning_rate": 8.42164509734764e-06,
      "loss": 0.366,
      "step": 17971
    },
    {
      "epoch": 0.8247441604332064,
      "grad_norm": 0.4464573264122009,
      "learning_rate": 8.421466308742973e-06,
      "loss": 0.3431,
      "step": 17972
    },
    {
      "epoch": 0.8247900509384608,
      "grad_norm": 0.47099459171295166,
      "learning_rate": 8.421287511910763e-06,
      "loss": 0.3905,
      "step": 17973
    },
    {
      "epoch": 0.8248359414437153,
      "grad_norm": 0.4275107681751251,
      "learning_rate": 8.421108706851432e-06,
      "loss": 0.3297,
      "step": 17974
    },
    {
      "epoch": 0.8248818319489698,
      "grad_norm": 0.43864890933036804,
      "learning_rate": 8.420929893565417e-06,
      "loss": 0.3892,
      "step": 17975
    },
    {
      "epoch": 0.8249277224542242,
      "grad_norm": 0.4424125850200653,
      "learning_rate": 8.420751072053146e-06,
      "loss": 0.3921,
      "step": 17976
    },
    {
      "epoch": 0.8249736129594787,
      "grad_norm": 0.5345363020896912,
      "learning_rate": 8.420572242315046e-06,
      "loss": 0.4595,
      "step": 17977
    },
    {
      "epoch": 0.8250195034647332,
      "grad_norm": 0.43983420729637146,
      "learning_rate": 8.42039340435155e-06,
      "loss": 0.3847,
      "step": 17978
    },
    {
      "epoch": 0.8250653939699876,
      "grad_norm": 0.4110962748527527,
      "learning_rate": 8.420214558163087e-06,
      "loss": 0.304,
      "step": 17979
    },
    {
      "epoch": 0.8251112844752421,
      "grad_norm": 0.4997156858444214,
      "learning_rate": 8.420035703750087e-06,
      "loss": 0.5184,
      "step": 17980
    },
    {
      "epoch": 0.8251571749804966,
      "grad_norm": 0.44162389636039734,
      "learning_rate": 8.419856841112982e-06,
      "loss": 0.3694,
      "step": 17981
    },
    {
      "epoch": 0.825203065485751,
      "grad_norm": 0.44219037890434265,
      "learning_rate": 8.419677970252201e-06,
      "loss": 0.4004,
      "step": 17982
    },
    {
      "epoch": 0.8252489559910055,
      "grad_norm": 0.4799107313156128,
      "learning_rate": 8.41949909116817e-06,
      "loss": 0.5016,
      "step": 17983
    },
    {
      "epoch": 0.82529484649626,
      "grad_norm": 0.434451162815094,
      "learning_rate": 8.419320203861327e-06,
      "loss": 0.3824,
      "step": 17984
    },
    {
      "epoch": 0.8253407370015143,
      "grad_norm": 0.4709138870239258,
      "learning_rate": 8.419141308332098e-06,
      "loss": 0.4162,
      "step": 17985
    },
    {
      "epoch": 0.8253866275067688,
      "grad_norm": 0.4469585418701172,
      "learning_rate": 8.418962404580911e-06,
      "loss": 0.4102,
      "step": 17986
    },
    {
      "epoch": 0.8254325180120233,
      "grad_norm": 0.43084341287612915,
      "learning_rate": 8.4187834926082e-06,
      "loss": 0.3168,
      "step": 17987
    },
    {
      "epoch": 0.8254784085172778,
      "grad_norm": 0.44466376304626465,
      "learning_rate": 8.418604572414393e-06,
      "loss": 0.4108,
      "step": 17988
    },
    {
      "epoch": 0.8255242990225322,
      "grad_norm": 0.4413444697856903,
      "learning_rate": 8.418425643999921e-06,
      "loss": 0.3613,
      "step": 17989
    },
    {
      "epoch": 0.8255701895277867,
      "grad_norm": 0.44197750091552734,
      "learning_rate": 8.418246707365215e-06,
      "loss": 0.3733,
      "step": 17990
    },
    {
      "epoch": 0.8256160800330412,
      "grad_norm": 0.4442468583583832,
      "learning_rate": 8.418067762510702e-06,
      "loss": 0.3969,
      "step": 17991
    },
    {
      "epoch": 0.8256619705382956,
      "grad_norm": 0.5202842950820923,
      "learning_rate": 8.417888809436817e-06,
      "loss": 0.4697,
      "step": 17992
    },
    {
      "epoch": 0.8257078610435501,
      "grad_norm": 0.4693809449672699,
      "learning_rate": 8.417709848143989e-06,
      "loss": 0.4368,
      "step": 17993
    },
    {
      "epoch": 0.8257537515488046,
      "grad_norm": 0.4337964355945587,
      "learning_rate": 8.417530878632645e-06,
      "loss": 0.3444,
      "step": 17994
    },
    {
      "epoch": 0.825799642054059,
      "grad_norm": 0.4276637136936188,
      "learning_rate": 8.41735190090322e-06,
      "loss": 0.3561,
      "step": 17995
    },
    {
      "epoch": 0.8258455325593135,
      "grad_norm": 0.46315792202949524,
      "learning_rate": 8.417172914956141e-06,
      "loss": 0.4518,
      "step": 17996
    },
    {
      "epoch": 0.825891423064568,
      "grad_norm": 0.4326402544975281,
      "learning_rate": 8.416993920791841e-06,
      "loss": 0.366,
      "step": 17997
    },
    {
      "epoch": 0.8259373135698224,
      "grad_norm": 0.4896598756313324,
      "learning_rate": 8.416814918410747e-06,
      "loss": 0.4922,
      "step": 17998
    },
    {
      "epoch": 0.8259832040750769,
      "grad_norm": 0.409915566444397,
      "learning_rate": 8.416635907813291e-06,
      "loss": 0.3404,
      "step": 17999
    },
    {
      "epoch": 0.8260290945803314,
      "grad_norm": 0.466506689786911,
      "learning_rate": 8.416456888999907e-06,
      "loss": 0.4754,
      "step": 18000
    },
    {
      "epoch": 0.8260749850855857,
      "grad_norm": 0.5669476985931396,
      "learning_rate": 8.41627786197102e-06,
      "loss": 0.4346,
      "step": 18001
    },
    {
      "epoch": 0.8261208755908402,
      "grad_norm": 0.48300784826278687,
      "learning_rate": 8.416098826727064e-06,
      "loss": 0.4153,
      "step": 18002
    },
    {
      "epoch": 0.8261667660960947,
      "grad_norm": 0.44610118865966797,
      "learning_rate": 8.415919783268468e-06,
      "loss": 0.3821,
      "step": 18003
    },
    {
      "epoch": 0.8262126566013491,
      "grad_norm": 0.43876445293426514,
      "learning_rate": 8.415740731595664e-06,
      "loss": 0.3887,
      "step": 18004
    },
    {
      "epoch": 0.8262585471066036,
      "grad_norm": 0.4780218303203583,
      "learning_rate": 8.41556167170908e-06,
      "loss": 0.4776,
      "step": 18005
    },
    {
      "epoch": 0.8263044376118581,
      "grad_norm": 0.49615898728370667,
      "learning_rate": 8.415382603609151e-06,
      "loss": 0.5029,
      "step": 18006
    },
    {
      "epoch": 0.8263503281171126,
      "grad_norm": 0.42977532744407654,
      "learning_rate": 8.415203527296302e-06,
      "loss": 0.3576,
      "step": 18007
    },
    {
      "epoch": 0.826396218622367,
      "grad_norm": 0.4524504244327545,
      "learning_rate": 8.415024442770966e-06,
      "loss": 0.4226,
      "step": 18008
    },
    {
      "epoch": 0.8264421091276215,
      "grad_norm": 0.4816759526729584,
      "learning_rate": 8.414845350033576e-06,
      "loss": 0.4386,
      "step": 18009
    },
    {
      "epoch": 0.826487999632876,
      "grad_norm": 0.4708247184753418,
      "learning_rate": 8.41466624908456e-06,
      "loss": 0.3641,
      "step": 18010
    },
    {
      "epoch": 0.8265338901381304,
      "grad_norm": 0.48955750465393066,
      "learning_rate": 8.41448713992435e-06,
      "loss": 0.426,
      "step": 18011
    },
    {
      "epoch": 0.8265797806433849,
      "grad_norm": 0.4376161992549896,
      "learning_rate": 8.414308022553374e-06,
      "loss": 0.3622,
      "step": 18012
    },
    {
      "epoch": 0.8266256711486394,
      "grad_norm": 0.46941858530044556,
      "learning_rate": 8.414128896972066e-06,
      "loss": 0.4396,
      "step": 18013
    },
    {
      "epoch": 0.8266715616538938,
      "grad_norm": 0.4503255784511566,
      "learning_rate": 8.413949763180858e-06,
      "loss": 0.435,
      "step": 18014
    },
    {
      "epoch": 0.8267174521591483,
      "grad_norm": 0.4422112703323364,
      "learning_rate": 8.413770621180176e-06,
      "loss": 0.3759,
      "step": 18015
    },
    {
      "epoch": 0.8267633426644028,
      "grad_norm": 0.47206711769104004,
      "learning_rate": 8.413591470970453e-06,
      "loss": 0.4762,
      "step": 18016
    },
    {
      "epoch": 0.8268092331696572,
      "grad_norm": 0.7578540444374084,
      "learning_rate": 8.413412312552118e-06,
      "loss": 0.4061,
      "step": 18017
    },
    {
      "epoch": 0.8268551236749117,
      "grad_norm": 0.48906344175338745,
      "learning_rate": 8.413233145925606e-06,
      "loss": 0.4378,
      "step": 18018
    },
    {
      "epoch": 0.8269010141801661,
      "grad_norm": 0.4059424102306366,
      "learning_rate": 8.413053971091347e-06,
      "loss": 0.3237,
      "step": 18019
    },
    {
      "epoch": 0.8269469046854205,
      "grad_norm": 0.4637203812599182,
      "learning_rate": 8.412874788049768e-06,
      "loss": 0.3884,
      "step": 18020
    },
    {
      "epoch": 0.826992795190675,
      "grad_norm": 0.41301819682121277,
      "learning_rate": 8.412695596801303e-06,
      "loss": 0.3487,
      "step": 18021
    },
    {
      "epoch": 0.8270386856959295,
      "grad_norm": 0.47388318181037903,
      "learning_rate": 8.412516397346383e-06,
      "loss": 0.4085,
      "step": 18022
    },
    {
      "epoch": 0.827084576201184,
      "grad_norm": 0.5158187747001648,
      "learning_rate": 8.412337189685436e-06,
      "loss": 0.5103,
      "step": 18023
    },
    {
      "epoch": 0.8271304667064384,
      "grad_norm": 0.5430427193641663,
      "learning_rate": 8.412157973818895e-06,
      "loss": 0.5405,
      "step": 18024
    },
    {
      "epoch": 0.8271763572116929,
      "grad_norm": 0.4692544639110565,
      "learning_rate": 8.411978749747193e-06,
      "loss": 0.3055,
      "step": 18025
    },
    {
      "epoch": 0.8272222477169474,
      "grad_norm": 0.40423405170440674,
      "learning_rate": 8.411799517470758e-06,
      "loss": 0.3459,
      "step": 18026
    },
    {
      "epoch": 0.8272681382222018,
      "grad_norm": 0.4462067484855652,
      "learning_rate": 8.411620276990022e-06,
      "loss": 0.4153,
      "step": 18027
    },
    {
      "epoch": 0.8273140287274563,
      "grad_norm": 0.4292373061180115,
      "learning_rate": 8.411441028305416e-06,
      "loss": 0.3682,
      "step": 18028
    },
    {
      "epoch": 0.8273599192327108,
      "grad_norm": 0.4671582281589508,
      "learning_rate": 8.411261771417371e-06,
      "loss": 0.4075,
      "step": 18029
    },
    {
      "epoch": 0.8274058097379652,
      "grad_norm": 0.4636654257774353,
      "learning_rate": 8.411082506326318e-06,
      "loss": 0.3751,
      "step": 18030
    },
    {
      "epoch": 0.8274517002432197,
      "grad_norm": 0.4226151704788208,
      "learning_rate": 8.410903233032688e-06,
      "loss": 0.3436,
      "step": 18031
    },
    {
      "epoch": 0.8274975907484742,
      "grad_norm": 0.44645023345947266,
      "learning_rate": 8.410723951536913e-06,
      "loss": 0.4009,
      "step": 18032
    },
    {
      "epoch": 0.8275434812537286,
      "grad_norm": 0.4608747065067291,
      "learning_rate": 8.410544661839421e-06,
      "loss": 0.3945,
      "step": 18033
    },
    {
      "epoch": 0.8275893717589831,
      "grad_norm": 0.48663607239723206,
      "learning_rate": 8.410365363940648e-06,
      "loss": 0.3995,
      "step": 18034
    },
    {
      "epoch": 0.8276352622642376,
      "grad_norm": 0.4439545273780823,
      "learning_rate": 8.410186057841022e-06,
      "loss": 0.3795,
      "step": 18035
    },
    {
      "epoch": 0.827681152769492,
      "grad_norm": 0.4712582230567932,
      "learning_rate": 8.410006743540975e-06,
      "loss": 0.445,
      "step": 18036
    },
    {
      "epoch": 0.8277270432747464,
      "grad_norm": 0.4870122969150543,
      "learning_rate": 8.409827421040938e-06,
      "loss": 0.4372,
      "step": 18037
    },
    {
      "epoch": 0.8277729337800009,
      "grad_norm": 0.436261922121048,
      "learning_rate": 8.409648090341342e-06,
      "loss": 0.3909,
      "step": 18038
    },
    {
      "epoch": 0.8278188242852553,
      "grad_norm": 0.45746394991874695,
      "learning_rate": 8.409468751442618e-06,
      "loss": 0.419,
      "step": 18039
    },
    {
      "epoch": 0.8278647147905098,
      "grad_norm": 0.4783158302307129,
      "learning_rate": 8.409289404345198e-06,
      "loss": 0.5015,
      "step": 18040
    },
    {
      "epoch": 0.8279106052957643,
      "grad_norm": 0.4209771752357483,
      "learning_rate": 8.409110049049514e-06,
      "loss": 0.3269,
      "step": 18041
    },
    {
      "epoch": 0.8279564958010188,
      "grad_norm": 0.44271835684776306,
      "learning_rate": 8.408930685555994e-06,
      "loss": 0.4013,
      "step": 18042
    },
    {
      "epoch": 0.8280023863062732,
      "grad_norm": 0.5674777626991272,
      "learning_rate": 8.408751313865073e-06,
      "loss": 0.341,
      "step": 18043
    },
    {
      "epoch": 0.8280482768115277,
      "grad_norm": 0.4137333631515503,
      "learning_rate": 8.408571933977181e-06,
      "loss": 0.3353,
      "step": 18044
    },
    {
      "epoch": 0.8280941673167822,
      "grad_norm": 0.4755222797393799,
      "learning_rate": 8.40839254589275e-06,
      "loss": 0.4725,
      "step": 18045
    },
    {
      "epoch": 0.8281400578220366,
      "grad_norm": 0.4928211271762848,
      "learning_rate": 8.40821314961221e-06,
      "loss": 0.497,
      "step": 18046
    },
    {
      "epoch": 0.8281859483272911,
      "grad_norm": 0.43681150674819946,
      "learning_rate": 8.408033745135993e-06,
      "loss": 0.395,
      "step": 18047
    },
    {
      "epoch": 0.8282318388325456,
      "grad_norm": 0.48329344391822815,
      "learning_rate": 8.407854332464528e-06,
      "loss": 0.4503,
      "step": 18048
    },
    {
      "epoch": 0.8282777293378,
      "grad_norm": 0.48742738366127014,
      "learning_rate": 8.407674911598254e-06,
      "loss": 0.5051,
      "step": 18049
    },
    {
      "epoch": 0.8283236198430545,
      "grad_norm": 0.47547316551208496,
      "learning_rate": 8.407495482537594e-06,
      "loss": 0.4974,
      "step": 18050
    },
    {
      "epoch": 0.828369510348309,
      "grad_norm": 0.4149698317050934,
      "learning_rate": 8.407316045282983e-06,
      "loss": 0.358,
      "step": 18051
    },
    {
      "epoch": 0.8284154008535634,
      "grad_norm": 0.42962345480918884,
      "learning_rate": 8.407136599834853e-06,
      "loss": 0.4039,
      "step": 18052
    },
    {
      "epoch": 0.8284612913588179,
      "grad_norm": 0.4373871684074402,
      "learning_rate": 8.406957146193635e-06,
      "loss": 0.3708,
      "step": 18053
    },
    {
      "epoch": 0.8285071818640724,
      "grad_norm": 0.4803074896335602,
      "learning_rate": 8.406777684359759e-06,
      "loss": 0.3612,
      "step": 18054
    },
    {
      "epoch": 0.8285530723693267,
      "grad_norm": 0.4561117887496948,
      "learning_rate": 8.406598214333659e-06,
      "loss": 0.4283,
      "step": 18055
    },
    {
      "epoch": 0.8285989628745812,
      "grad_norm": 0.47030138969421387,
      "learning_rate": 8.406418736115764e-06,
      "loss": 0.4109,
      "step": 18056
    },
    {
      "epoch": 0.8286448533798357,
      "grad_norm": 0.4206443727016449,
      "learning_rate": 8.406239249706508e-06,
      "loss": 0.3379,
      "step": 18057
    },
    {
      "epoch": 0.8286907438850901,
      "grad_norm": 0.47873127460479736,
      "learning_rate": 8.406059755106323e-06,
      "loss": 0.474,
      "step": 18058
    },
    {
      "epoch": 0.8287366343903446,
      "grad_norm": 0.5150394439697266,
      "learning_rate": 8.405880252315637e-06,
      "loss": 0.4727,
      "step": 18059
    },
    {
      "epoch": 0.8287825248955991,
      "grad_norm": 0.49425581097602844,
      "learning_rate": 8.405700741334885e-06,
      "loss": 0.4338,
      "step": 18060
    },
    {
      "epoch": 0.8288284154008536,
      "grad_norm": 0.409773051738739,
      "learning_rate": 8.4055212221645e-06,
      "loss": 0.3596,
      "step": 18061
    },
    {
      "epoch": 0.828874305906108,
      "grad_norm": 0.39452025294303894,
      "learning_rate": 8.405341694804907e-06,
      "loss": 0.2546,
      "step": 18062
    },
    {
      "epoch": 0.8289201964113625,
      "grad_norm": 0.44225457310676575,
      "learning_rate": 8.405162159256544e-06,
      "loss": 0.3571,
      "step": 18063
    },
    {
      "epoch": 0.828966086916617,
      "grad_norm": 0.45075926184654236,
      "learning_rate": 8.40498261551984e-06,
      "loss": 0.3445,
      "step": 18064
    },
    {
      "epoch": 0.8290119774218714,
      "grad_norm": 0.4690568447113037,
      "learning_rate": 8.404803063595228e-06,
      "loss": 0.4195,
      "step": 18065
    },
    {
      "epoch": 0.8290578679271259,
      "grad_norm": 0.44292670488357544,
      "learning_rate": 8.40462350348314e-06,
      "loss": 0.3491,
      "step": 18066
    },
    {
      "epoch": 0.8291037584323804,
      "grad_norm": 0.4242332875728607,
      "learning_rate": 8.404443935184006e-06,
      "loss": 0.3804,
      "step": 18067
    },
    {
      "epoch": 0.8291496489376348,
      "grad_norm": 0.435048907995224,
      "learning_rate": 8.404264358698258e-06,
      "loss": 0.3937,
      "step": 18068
    },
    {
      "epoch": 0.8291955394428893,
      "grad_norm": 0.5266005396842957,
      "learning_rate": 8.404084774026331e-06,
      "loss": 0.5278,
      "step": 18069
    },
    {
      "epoch": 0.8292414299481438,
      "grad_norm": 0.5077800750732422,
      "learning_rate": 8.403905181168653e-06,
      "loss": 0.5428,
      "step": 18070
    },
    {
      "epoch": 0.8292873204533981,
      "grad_norm": 0.4602810740470886,
      "learning_rate": 8.403725580125656e-06,
      "loss": 0.3739,
      "step": 18071
    },
    {
      "epoch": 0.8293332109586526,
      "grad_norm": 0.4061303436756134,
      "learning_rate": 8.403545970897775e-06,
      "loss": 0.2948,
      "step": 18072
    },
    {
      "epoch": 0.8293791014639071,
      "grad_norm": 0.45500680804252625,
      "learning_rate": 8.40336635348544e-06,
      "loss": 0.3849,
      "step": 18073
    },
    {
      "epoch": 0.8294249919691615,
      "grad_norm": 0.4685817062854767,
      "learning_rate": 8.403186727889084e-06,
      "loss": 0.4493,
      "step": 18074
    },
    {
      "epoch": 0.829470882474416,
      "grad_norm": 0.44068941473960876,
      "learning_rate": 8.403007094109136e-06,
      "loss": 0.348,
      "step": 18075
    },
    {
      "epoch": 0.8295167729796705,
      "grad_norm": 0.47200459241867065,
      "learning_rate": 8.40282745214603e-06,
      "loss": 0.4508,
      "step": 18076
    },
    {
      "epoch": 0.829562663484925,
      "grad_norm": 0.44364723563194275,
      "learning_rate": 8.402647802000198e-06,
      "loss": 0.3701,
      "step": 18077
    },
    {
      "epoch": 0.8296085539901794,
      "grad_norm": 0.4850562512874603,
      "learning_rate": 8.402468143672074e-06,
      "loss": 0.448,
      "step": 18078
    },
    {
      "epoch": 0.8296544444954339,
      "grad_norm": 0.4606315493583679,
      "learning_rate": 8.402288477162085e-06,
      "loss": 0.3653,
      "step": 18079
    },
    {
      "epoch": 0.8297003350006884,
      "grad_norm": 0.4254797697067261,
      "learning_rate": 8.40210880247067e-06,
      "loss": 0.3601,
      "step": 18080
    },
    {
      "epoch": 0.8297462255059428,
      "grad_norm": 0.43236732482910156,
      "learning_rate": 8.401929119598253e-06,
      "loss": 0.3325,
      "step": 18081
    },
    {
      "epoch": 0.8297921160111973,
      "grad_norm": 0.48773565888404846,
      "learning_rate": 8.401749428545273e-06,
      "loss": 0.486,
      "step": 18082
    },
    {
      "epoch": 0.8298380065164518,
      "grad_norm": 0.4719679057598114,
      "learning_rate": 8.401569729312159e-06,
      "loss": 0.464,
      "step": 18083
    },
    {
      "epoch": 0.8298838970217062,
      "grad_norm": 0.4486527144908905,
      "learning_rate": 8.401390021899342e-06,
      "loss": 0.4266,
      "step": 18084
    },
    {
      "epoch": 0.8299297875269607,
      "grad_norm": 0.49645915627479553,
      "learning_rate": 8.401210306307255e-06,
      "loss": 0.4334,
      "step": 18085
    },
    {
      "epoch": 0.8299756780322152,
      "grad_norm": 0.47308865189552307,
      "learning_rate": 8.401030582536332e-06,
      "loss": 0.4524,
      "step": 18086
    },
    {
      "epoch": 0.8300215685374696,
      "grad_norm": 0.43211984634399414,
      "learning_rate": 8.400850850587004e-06,
      "loss": 0.355,
      "step": 18087
    },
    {
      "epoch": 0.830067459042724,
      "grad_norm": 0.42160764336586,
      "learning_rate": 8.400671110459701e-06,
      "loss": 0.3562,
      "step": 18088
    },
    {
      "epoch": 0.8301133495479786,
      "grad_norm": 0.4502413868904114,
      "learning_rate": 8.40049136215486e-06,
      "loss": 0.3833,
      "step": 18089
    },
    {
      "epoch": 0.8301592400532329,
      "grad_norm": 0.7507365942001343,
      "learning_rate": 8.40031160567291e-06,
      "loss": 0.3645,
      "step": 18090
    },
    {
      "epoch": 0.8302051305584874,
      "grad_norm": 0.455634742975235,
      "learning_rate": 8.400131841014283e-06,
      "loss": 0.4178,
      "step": 18091
    },
    {
      "epoch": 0.8302510210637419,
      "grad_norm": 0.46778732538223267,
      "learning_rate": 8.399952068179411e-06,
      "loss": 0.4109,
      "step": 18092
    },
    {
      "epoch": 0.8302969115689963,
      "grad_norm": 0.4121411144733429,
      "learning_rate": 8.399772287168729e-06,
      "loss": 0.3349,
      "step": 18093
    },
    {
      "epoch": 0.8303428020742508,
      "grad_norm": 0.4600745439529419,
      "learning_rate": 8.399592497982667e-06,
      "loss": 0.3824,
      "step": 18094
    },
    {
      "epoch": 0.8303886925795053,
      "grad_norm": 0.45569780468940735,
      "learning_rate": 8.399412700621658e-06,
      "loss": 0.3557,
      "step": 18095
    },
    {
      "epoch": 0.8304345830847598,
      "grad_norm": 0.45500025153160095,
      "learning_rate": 8.399232895086134e-06,
      "loss": 0.4198,
      "step": 18096
    },
    {
      "epoch": 0.8304804735900142,
      "grad_norm": 0.438245952129364,
      "learning_rate": 8.399053081376529e-06,
      "loss": 0.3797,
      "step": 18097
    },
    {
      "epoch": 0.8305263640952687,
      "grad_norm": 0.40246495604515076,
      "learning_rate": 8.398873259493273e-06,
      "loss": 0.318,
      "step": 18098
    },
    {
      "epoch": 0.8305722546005232,
      "grad_norm": 0.44568273425102234,
      "learning_rate": 8.3986934294368e-06,
      "loss": 0.3898,
      "step": 18099
    },
    {
      "epoch": 0.8306181451057776,
      "grad_norm": 0.4664110839366913,
      "learning_rate": 8.398513591207542e-06,
      "loss": 0.4373,
      "step": 18100
    },
    {
      "epoch": 0.8306640356110321,
      "grad_norm": 0.4412544369697571,
      "learning_rate": 8.39833374480593e-06,
      "loss": 0.3627,
      "step": 18101
    },
    {
      "epoch": 0.8307099261162866,
      "grad_norm": 0.43946385383605957,
      "learning_rate": 8.398153890232398e-06,
      "loss": 0.4009,
      "step": 18102
    },
    {
      "epoch": 0.830755816621541,
      "grad_norm": 0.4306648373603821,
      "learning_rate": 8.397974027487382e-06,
      "loss": 0.3395,
      "step": 18103
    },
    {
      "epoch": 0.8308017071267955,
      "grad_norm": 0.45021679997444153,
      "learning_rate": 8.397794156571307e-06,
      "loss": 0.4637,
      "step": 18104
    },
    {
      "epoch": 0.83084759763205,
      "grad_norm": 0.46398481726646423,
      "learning_rate": 8.397614277484611e-06,
      "loss": 0.375,
      "step": 18105
    },
    {
      "epoch": 0.8308934881373043,
      "grad_norm": 0.4511909782886505,
      "learning_rate": 8.397434390227726e-06,
      "loss": 0.3994,
      "step": 18106
    },
    {
      "epoch": 0.8309393786425588,
      "grad_norm": 0.45857441425323486,
      "learning_rate": 8.397254494801082e-06,
      "loss": 0.3875,
      "step": 18107
    },
    {
      "epoch": 0.8309852691478133,
      "grad_norm": 0.44027769565582275,
      "learning_rate": 8.397074591205113e-06,
      "loss": 0.3892,
      "step": 18108
    },
    {
      "epoch": 0.8310311596530677,
      "grad_norm": 0.4730708599090576,
      "learning_rate": 8.396894679440253e-06,
      "loss": 0.439,
      "step": 18109
    },
    {
      "epoch": 0.8310770501583222,
      "grad_norm": 0.4575815498828888,
      "learning_rate": 8.396714759506931e-06,
      "loss": 0.3903,
      "step": 18110
    },
    {
      "epoch": 0.8311229406635767,
      "grad_norm": 0.45963162183761597,
      "learning_rate": 8.396534831405584e-06,
      "loss": 0.3984,
      "step": 18111
    },
    {
      "epoch": 0.8311688311688312,
      "grad_norm": 0.45422109961509705,
      "learning_rate": 8.396354895136642e-06,
      "loss": 0.4081,
      "step": 18112
    },
    {
      "epoch": 0.8312147216740856,
      "grad_norm": 0.4470139443874359,
      "learning_rate": 8.396174950700539e-06,
      "loss": 0.3702,
      "step": 18113
    },
    {
      "epoch": 0.8312606121793401,
      "grad_norm": 0.49207833409309387,
      "learning_rate": 8.395994998097707e-06,
      "loss": 0.48,
      "step": 18114
    },
    {
      "epoch": 0.8313065026845946,
      "grad_norm": 0.4725109040737152,
      "learning_rate": 8.395815037328579e-06,
      "loss": 0.4251,
      "step": 18115
    },
    {
      "epoch": 0.831352393189849,
      "grad_norm": 0.4591810703277588,
      "learning_rate": 8.395635068393586e-06,
      "loss": 0.3748,
      "step": 18116
    },
    {
      "epoch": 0.8313982836951035,
      "grad_norm": 0.5284903645515442,
      "learning_rate": 8.395455091293165e-06,
      "loss": 0.4397,
      "step": 18117
    },
    {
      "epoch": 0.831444174200358,
      "grad_norm": 0.4265201985836029,
      "learning_rate": 8.395275106027745e-06,
      "loss": 0.3229,
      "step": 18118
    },
    {
      "epoch": 0.8314900647056124,
      "grad_norm": 0.4584498405456543,
      "learning_rate": 8.395095112597758e-06,
      "loss": 0.4244,
      "step": 18119
    },
    {
      "epoch": 0.8315359552108669,
      "grad_norm": 0.46957993507385254,
      "learning_rate": 8.394915111003642e-06,
      "loss": 0.3933,
      "step": 18120
    },
    {
      "epoch": 0.8315818457161214,
      "grad_norm": 0.4478274881839752,
      "learning_rate": 8.394735101245823e-06,
      "loss": 0.3288,
      "step": 18121
    },
    {
      "epoch": 0.8316277362213758,
      "grad_norm": 0.4760933220386505,
      "learning_rate": 8.39455508332474e-06,
      "loss": 0.4662,
      "step": 18122
    },
    {
      "epoch": 0.8316736267266303,
      "grad_norm": 0.4454164206981659,
      "learning_rate": 8.394375057240825e-06,
      "loss": 0.3571,
      "step": 18123
    },
    {
      "epoch": 0.8317195172318848,
      "grad_norm": 0.46467339992523193,
      "learning_rate": 8.394195022994506e-06,
      "loss": 0.363,
      "step": 18124
    },
    {
      "epoch": 0.8317654077371391,
      "grad_norm": 0.49284371733665466,
      "learning_rate": 8.39401498058622e-06,
      "loss": 0.5076,
      "step": 18125
    },
    {
      "epoch": 0.8318112982423936,
      "grad_norm": 0.4246115982532501,
      "learning_rate": 8.3938349300164e-06,
      "loss": 0.3383,
      "step": 18126
    },
    {
      "epoch": 0.8318571887476481,
      "grad_norm": 0.5000753402709961,
      "learning_rate": 8.393654871285477e-06,
      "loss": 0.4716,
      "step": 18127
    },
    {
      "epoch": 0.8319030792529025,
      "grad_norm": 0.48261353373527527,
      "learning_rate": 8.393474804393887e-06,
      "loss": 0.3555,
      "step": 18128
    },
    {
      "epoch": 0.831948969758157,
      "grad_norm": 0.4638386368751526,
      "learning_rate": 8.393294729342058e-06,
      "loss": 0.3854,
      "step": 18129
    },
    {
      "epoch": 0.8319948602634115,
      "grad_norm": 0.49651774764060974,
      "learning_rate": 8.39311464613043e-06,
      "loss": 0.4523,
      "step": 18130
    },
    {
      "epoch": 0.832040750768666,
      "grad_norm": 0.42594966292381287,
      "learning_rate": 8.39293455475943e-06,
      "loss": 0.3154,
      "step": 18131
    },
    {
      "epoch": 0.8320866412739204,
      "grad_norm": 0.42838552594184875,
      "learning_rate": 8.392754455229493e-06,
      "loss": 0.3466,
      "step": 18132
    },
    {
      "epoch": 0.8321325317791749,
      "grad_norm": 0.47378867864608765,
      "learning_rate": 8.392574347541053e-06,
      "loss": 0.3814,
      "step": 18133
    },
    {
      "epoch": 0.8321784222844294,
      "grad_norm": 0.41699907183647156,
      "learning_rate": 8.392394231694543e-06,
      "loss": 0.3009,
      "step": 18134
    },
    {
      "epoch": 0.8322243127896838,
      "grad_norm": 0.45433488488197327,
      "learning_rate": 8.392214107690392e-06,
      "loss": 0.3768,
      "step": 18135
    },
    {
      "epoch": 0.8322702032949383,
      "grad_norm": 0.4418494701385498,
      "learning_rate": 8.392033975529042e-06,
      "loss": 0.3363,
      "step": 18136
    },
    {
      "epoch": 0.8323160938001928,
      "grad_norm": 0.42773592472076416,
      "learning_rate": 8.391853835210917e-06,
      "loss": 0.3528,
      "step": 18137
    },
    {
      "epoch": 0.8323619843054472,
      "grad_norm": 0.4696636199951172,
      "learning_rate": 8.391673686736455e-06,
      "loss": 0.457,
      "step": 18138
    },
    {
      "epoch": 0.8324078748107017,
      "grad_norm": 0.467499315738678,
      "learning_rate": 8.39149353010609e-06,
      "loss": 0.4812,
      "step": 18139
    },
    {
      "epoch": 0.8324537653159562,
      "grad_norm": 0.434304416179657,
      "learning_rate": 8.391313365320251e-06,
      "loss": 0.3549,
      "step": 18140
    },
    {
      "epoch": 0.8324996558212105,
      "grad_norm": 0.46710005402565,
      "learning_rate": 8.391133192379375e-06,
      "loss": 0.386,
      "step": 18141
    },
    {
      "epoch": 0.832545546326465,
      "grad_norm": 0.4291851222515106,
      "learning_rate": 8.390953011283893e-06,
      "loss": 0.3763,
      "step": 18142
    },
    {
      "epoch": 0.8325914368317195,
      "grad_norm": 0.45171165466308594,
      "learning_rate": 8.390772822034241e-06,
      "loss": 0.4121,
      "step": 18143
    },
    {
      "epoch": 0.8326373273369739,
      "grad_norm": 0.4355604946613312,
      "learning_rate": 8.390592624630849e-06,
      "loss": 0.3786,
      "step": 18144
    },
    {
      "epoch": 0.8326832178422284,
      "grad_norm": 0.4775046408176422,
      "learning_rate": 8.390412419074151e-06,
      "loss": 0.4161,
      "step": 18145
    },
    {
      "epoch": 0.8327291083474829,
      "grad_norm": 0.47253572940826416,
      "learning_rate": 8.390232205364584e-06,
      "loss": 0.4262,
      "step": 18146
    },
    {
      "epoch": 0.8327749988527373,
      "grad_norm": 0.43773478269577026,
      "learning_rate": 8.390051983502576e-06,
      "loss": 0.3803,
      "step": 18147
    },
    {
      "epoch": 0.8328208893579918,
      "grad_norm": 0.44730404019355774,
      "learning_rate": 8.389871753488565e-06,
      "loss": 0.4128,
      "step": 18148
    },
    {
      "epoch": 0.8328667798632463,
      "grad_norm": 0.46576371788978577,
      "learning_rate": 8.389691515322979e-06,
      "loss": 0.4009,
      "step": 18149
    },
    {
      "epoch": 0.8329126703685008,
      "grad_norm": 0.4602697193622589,
      "learning_rate": 8.389511269006258e-06,
      "loss": 0.4042,
      "step": 18150
    },
    {
      "epoch": 0.8329585608737552,
      "grad_norm": 0.4739201068878174,
      "learning_rate": 8.38933101453883e-06,
      "loss": 0.456,
      "step": 18151
    },
    {
      "epoch": 0.8330044513790097,
      "grad_norm": 0.44872918725013733,
      "learning_rate": 8.389150751921133e-06,
      "loss": 0.3798,
      "step": 18152
    },
    {
      "epoch": 0.8330503418842642,
      "grad_norm": 0.44699302315711975,
      "learning_rate": 8.388970481153595e-06,
      "loss": 0.3213,
      "step": 18153
    },
    {
      "epoch": 0.8330962323895186,
      "grad_norm": 0.48090022802352905,
      "learning_rate": 8.388790202236656e-06,
      "loss": 0.4092,
      "step": 18154
    },
    {
      "epoch": 0.8331421228947731,
      "grad_norm": 0.45971307158470154,
      "learning_rate": 8.388609915170745e-06,
      "loss": 0.3906,
      "step": 18155
    },
    {
      "epoch": 0.8331880134000276,
      "grad_norm": 0.655051589012146,
      "learning_rate": 8.388429619956296e-06,
      "loss": 0.3838,
      "step": 18156
    },
    {
      "epoch": 0.833233903905282,
      "grad_norm": 0.4284975230693817,
      "learning_rate": 8.388249316593742e-06,
      "loss": 0.3236,
      "step": 18157
    },
    {
      "epoch": 0.8332797944105365,
      "grad_norm": 0.45652008056640625,
      "learning_rate": 8.38806900508352e-06,
      "loss": 0.3793,
      "step": 18158
    },
    {
      "epoch": 0.833325684915791,
      "grad_norm": 0.46670088171958923,
      "learning_rate": 8.38788868542606e-06,
      "loss": 0.4473,
      "step": 18159
    },
    {
      "epoch": 0.8333715754210453,
      "grad_norm": 0.4363967776298523,
      "learning_rate": 8.387708357621796e-06,
      "loss": 0.3746,
      "step": 18160
    },
    {
      "epoch": 0.8334174659262998,
      "grad_norm": 0.48592546582221985,
      "learning_rate": 8.387528021671166e-06,
      "loss": 0.4717,
      "step": 18161
    },
    {
      "epoch": 0.8334633564315543,
      "grad_norm": 0.4747212529182434,
      "learning_rate": 8.387347677574598e-06,
      "loss": 0.4166,
      "step": 18162
    },
    {
      "epoch": 0.8335092469368087,
      "grad_norm": 0.463144987821579,
      "learning_rate": 8.387167325332527e-06,
      "loss": 0.4142,
      "step": 18163
    },
    {
      "epoch": 0.8335551374420632,
      "grad_norm": 0.4370797574520111,
      "learning_rate": 8.386986964945388e-06,
      "loss": 0.3815,
      "step": 18164
    },
    {
      "epoch": 0.8336010279473177,
      "grad_norm": 0.4630209803581238,
      "learning_rate": 8.386806596413615e-06,
      "loss": 0.4011,
      "step": 18165
    },
    {
      "epoch": 0.8336469184525722,
      "grad_norm": 0.41693389415740967,
      "learning_rate": 8.386626219737641e-06,
      "loss": 0.3209,
      "step": 18166
    },
    {
      "epoch": 0.8336928089578266,
      "grad_norm": 0.4283023178577423,
      "learning_rate": 8.3864458349179e-06,
      "loss": 0.3379,
      "step": 18167
    },
    {
      "epoch": 0.8337386994630811,
      "grad_norm": 0.47956883907318115,
      "learning_rate": 8.386265441954825e-06,
      "loss": 0.4359,
      "step": 18168
    },
    {
      "epoch": 0.8337845899683356,
      "grad_norm": 0.46075186133384705,
      "learning_rate": 8.38608504084885e-06,
      "loss": 0.4225,
      "step": 18169
    },
    {
      "epoch": 0.83383048047359,
      "grad_norm": 0.4661857783794403,
      "learning_rate": 8.38590463160041e-06,
      "loss": 0.4501,
      "step": 18170
    },
    {
      "epoch": 0.8338763709788445,
      "grad_norm": 0.44938957691192627,
      "learning_rate": 8.385724214209939e-06,
      "loss": 0.3756,
      "step": 18171
    },
    {
      "epoch": 0.833922261484099,
      "grad_norm": 0.45433881878852844,
      "learning_rate": 8.385543788677868e-06,
      "loss": 0.4328,
      "step": 18172
    },
    {
      "epoch": 0.8339681519893534,
      "grad_norm": 0.4779817461967468,
      "learning_rate": 8.385363355004634e-06,
      "loss": 0.4703,
      "step": 18173
    },
    {
      "epoch": 0.8340140424946079,
      "grad_norm": 0.47456952929496765,
      "learning_rate": 8.385182913190669e-06,
      "loss": 0.4505,
      "step": 18174
    },
    {
      "epoch": 0.8340599329998624,
      "grad_norm": 0.47255849838256836,
      "learning_rate": 8.385002463236407e-06,
      "loss": 0.4404,
      "step": 18175
    },
    {
      "epoch": 0.8341058235051168,
      "grad_norm": 0.4196443259716034,
      "learning_rate": 8.384822005142283e-06,
      "loss": 0.3262,
      "step": 18176
    },
    {
      "epoch": 0.8341517140103712,
      "grad_norm": 0.49156394600868225,
      "learning_rate": 8.38464153890873e-06,
      "loss": 0.4085,
      "step": 18177
    },
    {
      "epoch": 0.8341976045156257,
      "grad_norm": 0.4468965530395508,
      "learning_rate": 8.384461064536183e-06,
      "loss": 0.3537,
      "step": 18178
    },
    {
      "epoch": 0.8342434950208801,
      "grad_norm": 0.4583326578140259,
      "learning_rate": 8.384280582025075e-06,
      "loss": 0.4362,
      "step": 18179
    },
    {
      "epoch": 0.8342893855261346,
      "grad_norm": 0.4402487874031067,
      "learning_rate": 8.384100091375839e-06,
      "loss": 0.3745,
      "step": 18180
    },
    {
      "epoch": 0.8343352760313891,
      "grad_norm": 0.4688296616077423,
      "learning_rate": 8.383919592588912e-06,
      "loss": 0.4041,
      "step": 18181
    },
    {
      "epoch": 0.8343811665366435,
      "grad_norm": 0.4449317455291748,
      "learning_rate": 8.383739085664725e-06,
      "loss": 0.3708,
      "step": 18182
    },
    {
      "epoch": 0.834427057041898,
      "grad_norm": 0.4405859708786011,
      "learning_rate": 8.383558570603715e-06,
      "loss": 0.3814,
      "step": 18183
    },
    {
      "epoch": 0.8344729475471525,
      "grad_norm": 0.45990121364593506,
      "learning_rate": 8.383378047406313e-06,
      "loss": 0.4024,
      "step": 18184
    },
    {
      "epoch": 0.834518838052407,
      "grad_norm": 0.4787445068359375,
      "learning_rate": 8.383197516072956e-06,
      "loss": 0.4482,
      "step": 18185
    },
    {
      "epoch": 0.8345647285576614,
      "grad_norm": 0.4425095319747925,
      "learning_rate": 8.383016976604077e-06,
      "loss": 0.3798,
      "step": 18186
    },
    {
      "epoch": 0.8346106190629159,
      "grad_norm": 0.46628013253211975,
      "learning_rate": 8.382836429000108e-06,
      "loss": 0.4836,
      "step": 18187
    },
    {
      "epoch": 0.8346565095681704,
      "grad_norm": 0.4309116005897522,
      "learning_rate": 8.382655873261487e-06,
      "loss": 0.3399,
      "step": 18188
    },
    {
      "epoch": 0.8347024000734248,
      "grad_norm": 0.47421401739120483,
      "learning_rate": 8.382475309388645e-06,
      "loss": 0.4396,
      "step": 18189
    },
    {
      "epoch": 0.8347482905786793,
      "grad_norm": 0.4512055516242981,
      "learning_rate": 8.382294737382018e-06,
      "loss": 0.3535,
      "step": 18190
    },
    {
      "epoch": 0.8347941810839338,
      "grad_norm": 0.4429827928543091,
      "learning_rate": 8.382114157242038e-06,
      "loss": 0.4072,
      "step": 18191
    },
    {
      "epoch": 0.8348400715891882,
      "grad_norm": 0.5159306526184082,
      "learning_rate": 8.381933568969143e-06,
      "loss": 0.525,
      "step": 18192
    },
    {
      "epoch": 0.8348859620944427,
      "grad_norm": 0.45336470007896423,
      "learning_rate": 8.381752972563764e-06,
      "loss": 0.3956,
      "step": 18193
    },
    {
      "epoch": 0.8349318525996972,
      "grad_norm": 0.47070544958114624,
      "learning_rate": 8.381572368026337e-06,
      "loss": 0.4198,
      "step": 18194
    },
    {
      "epoch": 0.8349777431049515,
      "grad_norm": 0.47601374983787537,
      "learning_rate": 8.381391755357295e-06,
      "loss": 0.4548,
      "step": 18195
    },
    {
      "epoch": 0.835023633610206,
      "grad_norm": 0.4822837710380554,
      "learning_rate": 8.381211134557074e-06,
      "loss": 0.4666,
      "step": 18196
    },
    {
      "epoch": 0.8350695241154605,
      "grad_norm": 0.448219358921051,
      "learning_rate": 8.381030505626106e-06,
      "loss": 0.3418,
      "step": 18197
    },
    {
      "epoch": 0.8351154146207149,
      "grad_norm": 0.44391027092933655,
      "learning_rate": 8.380849868564828e-06,
      "loss": 0.3623,
      "step": 18198
    },
    {
      "epoch": 0.8351613051259694,
      "grad_norm": 0.48952609300613403,
      "learning_rate": 8.380669223373674e-06,
      "loss": 0.4899,
      "step": 18199
    },
    {
      "epoch": 0.8352071956312239,
      "grad_norm": 0.43809714913368225,
      "learning_rate": 8.380488570053076e-06,
      "loss": 0.3717,
      "step": 18200
    },
    {
      "epoch": 0.8352530861364784,
      "grad_norm": 0.44115597009658813,
      "learning_rate": 8.380307908603468e-06,
      "loss": 0.3665,
      "step": 18201
    },
    {
      "epoch": 0.8352989766417328,
      "grad_norm": 0.441062867641449,
      "learning_rate": 8.38012723902529e-06,
      "loss": 0.3884,
      "step": 18202
    },
    {
      "epoch": 0.8353448671469873,
      "grad_norm": 0.4904654920101166,
      "learning_rate": 8.37994656131897e-06,
      "loss": 0.451,
      "step": 18203
    },
    {
      "epoch": 0.8353907576522418,
      "grad_norm": 0.45128825306892395,
      "learning_rate": 8.379765875484946e-06,
      "loss": 0.4153,
      "step": 18204
    },
    {
      "epoch": 0.8354366481574962,
      "grad_norm": 0.4851352870464325,
      "learning_rate": 8.379585181523652e-06,
      "loss": 0.4445,
      "step": 18205
    },
    {
      "epoch": 0.8354825386627507,
      "grad_norm": 0.41595596075057983,
      "learning_rate": 8.379404479435523e-06,
      "loss": 0.3566,
      "step": 18206
    },
    {
      "epoch": 0.8355284291680052,
      "grad_norm": 0.45313572883605957,
      "learning_rate": 8.379223769220993e-06,
      "loss": 0.3559,
      "step": 18207
    },
    {
      "epoch": 0.8355743196732596,
      "grad_norm": 0.45733004808425903,
      "learning_rate": 8.379043050880495e-06,
      "loss": 0.4468,
      "step": 18208
    },
    {
      "epoch": 0.8356202101785141,
      "grad_norm": 0.4851774275302887,
      "learning_rate": 8.378862324414466e-06,
      "loss": 0.4502,
      "step": 18209
    },
    {
      "epoch": 0.8356661006837686,
      "grad_norm": 0.44199225306510925,
      "learning_rate": 8.37868158982334e-06,
      "loss": 0.3817,
      "step": 18210
    },
    {
      "epoch": 0.835711991189023,
      "grad_norm": 0.46941936016082764,
      "learning_rate": 8.378500847107548e-06,
      "loss": 0.4253,
      "step": 18211
    },
    {
      "epoch": 0.8357578816942774,
      "grad_norm": 0.43103986978530884,
      "learning_rate": 8.378320096267531e-06,
      "loss": 0.3524,
      "step": 18212
    },
    {
      "epoch": 0.835803772199532,
      "grad_norm": 0.45080307126045227,
      "learning_rate": 8.37813933730372e-06,
      "loss": 0.4403,
      "step": 18213
    },
    {
      "epoch": 0.8358496627047863,
      "grad_norm": 0.43782469630241394,
      "learning_rate": 8.37795857021655e-06,
      "loss": 0.3331,
      "step": 18214
    },
    {
      "epoch": 0.8358955532100408,
      "grad_norm": 0.4822864830493927,
      "learning_rate": 8.377777795006454e-06,
      "loss": 0.5019,
      "step": 18215
    },
    {
      "epoch": 0.8359414437152953,
      "grad_norm": 0.4819743037223816,
      "learning_rate": 8.37759701167387e-06,
      "loss": 0.4553,
      "step": 18216
    },
    {
      "epoch": 0.8359873342205497,
      "grad_norm": 0.4571407735347748,
      "learning_rate": 8.37741622021923e-06,
      "loss": 0.3739,
      "step": 18217
    },
    {
      "epoch": 0.8360332247258042,
      "grad_norm": 0.4282774329185486,
      "learning_rate": 8.377235420642971e-06,
      "loss": 0.3644,
      "step": 18218
    },
    {
      "epoch": 0.8360791152310587,
      "grad_norm": 0.4403384029865265,
      "learning_rate": 8.377054612945528e-06,
      "loss": 0.3684,
      "step": 18219
    },
    {
      "epoch": 0.8361250057363132,
      "grad_norm": 0.42528074979782104,
      "learning_rate": 8.376873797127332e-06,
      "loss": 0.3603,
      "step": 18220
    },
    {
      "epoch": 0.8361708962415676,
      "grad_norm": 0.4658953845500946,
      "learning_rate": 8.37669297318882e-06,
      "loss": 0.3937,
      "step": 18221
    },
    {
      "epoch": 0.8362167867468221,
      "grad_norm": 0.41958317160606384,
      "learning_rate": 8.376512141130429e-06,
      "loss": 0.307,
      "step": 18222
    },
    {
      "epoch": 0.8362626772520766,
      "grad_norm": 0.5113511085510254,
      "learning_rate": 8.376331300952591e-06,
      "loss": 0.4992,
      "step": 18223
    },
    {
      "epoch": 0.836308567757331,
      "grad_norm": 0.45085346698760986,
      "learning_rate": 8.376150452655743e-06,
      "loss": 0.3644,
      "step": 18224
    },
    {
      "epoch": 0.8363544582625855,
      "grad_norm": 0.4446466267108917,
      "learning_rate": 8.375969596240318e-06,
      "loss": 0.3469,
      "step": 18225
    },
    {
      "epoch": 0.83640034876784,
      "grad_norm": 0.4998534321784973,
      "learning_rate": 8.375788731706751e-06,
      "loss": 0.4714,
      "step": 18226
    },
    {
      "epoch": 0.8364462392730944,
      "grad_norm": 0.42451003193855286,
      "learning_rate": 8.375607859055478e-06,
      "loss": 0.3173,
      "step": 18227
    },
    {
      "epoch": 0.8364921297783489,
      "grad_norm": 0.44549790024757385,
      "learning_rate": 8.375426978286934e-06,
      "loss": 0.348,
      "step": 18228
    },
    {
      "epoch": 0.8365380202836034,
      "grad_norm": 0.4476400911808014,
      "learning_rate": 8.375246089401553e-06,
      "loss": 0.3919,
      "step": 18229
    },
    {
      "epoch": 0.8365839107888577,
      "grad_norm": 0.43966755270957947,
      "learning_rate": 8.37506519239977e-06,
      "loss": 0.3566,
      "step": 18230
    },
    {
      "epoch": 0.8366298012941122,
      "grad_norm": 0.48417019844055176,
      "learning_rate": 8.37488428728202e-06,
      "loss": 0.4397,
      "step": 18231
    },
    {
      "epoch": 0.8366756917993667,
      "grad_norm": 0.4825865924358368,
      "learning_rate": 8.374703374048738e-06,
      "loss": 0.452,
      "step": 18232
    },
    {
      "epoch": 0.8367215823046211,
      "grad_norm": 0.4816177189350128,
      "learning_rate": 8.374522452700361e-06,
      "loss": 0.4485,
      "step": 18233
    },
    {
      "epoch": 0.8367674728098756,
      "grad_norm": 0.43582573533058167,
      "learning_rate": 8.374341523237323e-06,
      "loss": 0.3637,
      "step": 18234
    },
    {
      "epoch": 0.8368133633151301,
      "grad_norm": 0.4633936285972595,
      "learning_rate": 8.374160585660058e-06,
      "loss": 0.4161,
      "step": 18235
    },
    {
      "epoch": 0.8368592538203845,
      "grad_norm": 0.4648289978504181,
      "learning_rate": 8.373979639969002e-06,
      "loss": 0.3864,
      "step": 18236
    },
    {
      "epoch": 0.836905144325639,
      "grad_norm": 0.45461776852607727,
      "learning_rate": 8.373798686164588e-06,
      "loss": 0.3718,
      "step": 18237
    },
    {
      "epoch": 0.8369510348308935,
      "grad_norm": 0.43692269921302795,
      "learning_rate": 8.373617724247253e-06,
      "loss": 0.3473,
      "step": 18238
    },
    {
      "epoch": 0.836996925336148,
      "grad_norm": 0.45504483580589294,
      "learning_rate": 8.373436754217434e-06,
      "loss": 0.4197,
      "step": 18239
    },
    {
      "epoch": 0.8370428158414024,
      "grad_norm": 0.47626927495002747,
      "learning_rate": 8.373255776075564e-06,
      "loss": 0.4046,
      "step": 18240
    },
    {
      "epoch": 0.8370887063466569,
      "grad_norm": 0.4373861849308014,
      "learning_rate": 8.373074789822079e-06,
      "loss": 0.389,
      "step": 18241
    },
    {
      "epoch": 0.8371345968519114,
      "grad_norm": 0.46969908475875854,
      "learning_rate": 8.372893795457413e-06,
      "loss": 0.404,
      "step": 18242
    },
    {
      "epoch": 0.8371804873571658,
      "grad_norm": 0.4690396189689636,
      "learning_rate": 8.372712792982001e-06,
      "loss": 0.3867,
      "step": 18243
    },
    {
      "epoch": 0.8372263778624203,
      "grad_norm": 0.4490091800689697,
      "learning_rate": 8.372531782396281e-06,
      "loss": 0.3749,
      "step": 18244
    },
    {
      "epoch": 0.8372722683676748,
      "grad_norm": 0.5162858366966248,
      "learning_rate": 8.372350763700685e-06,
      "loss": 0.4186,
      "step": 18245
    },
    {
      "epoch": 0.8373181588729292,
      "grad_norm": 0.531032919883728,
      "learning_rate": 8.37216973689565e-06,
      "loss": 0.5088,
      "step": 18246
    },
    {
      "epoch": 0.8373640493781837,
      "grad_norm": 0.38909170031547546,
      "learning_rate": 8.37198870198161e-06,
      "loss": 0.2762,
      "step": 18247
    },
    {
      "epoch": 0.8374099398834381,
      "grad_norm": 0.45522668957710266,
      "learning_rate": 8.371807658959004e-06,
      "loss": 0.4091,
      "step": 18248
    },
    {
      "epoch": 0.8374558303886925,
      "grad_norm": 0.45884183049201965,
      "learning_rate": 8.371626607828262e-06,
      "loss": 0.4561,
      "step": 18249
    },
    {
      "epoch": 0.837501720893947,
      "grad_norm": 0.4830606281757355,
      "learning_rate": 8.371445548589823e-06,
      "loss": 0.4357,
      "step": 18250
    },
    {
      "epoch": 0.8375476113992015,
      "grad_norm": 0.5501541495323181,
      "learning_rate": 8.371264481244124e-06,
      "loss": 0.458,
      "step": 18251
    },
    {
      "epoch": 0.8375935019044559,
      "grad_norm": 0.46663227677345276,
      "learning_rate": 8.371083405791595e-06,
      "loss": 0.4706,
      "step": 18252
    },
    {
      "epoch": 0.8376393924097104,
      "grad_norm": 0.4751712381839752,
      "learning_rate": 8.370902322232675e-06,
      "loss": 0.407,
      "step": 18253
    },
    {
      "epoch": 0.8376852829149649,
      "grad_norm": 0.4225381016731262,
      "learning_rate": 8.3707212305678e-06,
      "loss": 0.3707,
      "step": 18254
    },
    {
      "epoch": 0.8377311734202194,
      "grad_norm": 0.45665115118026733,
      "learning_rate": 8.370540130797401e-06,
      "loss": 0.4122,
      "step": 18255
    },
    {
      "epoch": 0.8377770639254738,
      "grad_norm": 0.44249656796455383,
      "learning_rate": 8.370359022921921e-06,
      "loss": 0.4172,
      "step": 18256
    },
    {
      "epoch": 0.8378229544307283,
      "grad_norm": 0.4695450961589813,
      "learning_rate": 8.37017790694179e-06,
      "loss": 0.3845,
      "step": 18257
    },
    {
      "epoch": 0.8378688449359828,
      "grad_norm": 0.44053804874420166,
      "learning_rate": 8.369996782857444e-06,
      "loss": 0.3528,
      "step": 18258
    },
    {
      "epoch": 0.8379147354412372,
      "grad_norm": 0.5228598713874817,
      "learning_rate": 8.36981565066932e-06,
      "loss": 0.4423,
      "step": 18259
    },
    {
      "epoch": 0.8379606259464917,
      "grad_norm": 0.47499021887779236,
      "learning_rate": 8.36963451037785e-06,
      "loss": 0.4314,
      "step": 18260
    },
    {
      "epoch": 0.8380065164517462,
      "grad_norm": 0.43316978216171265,
      "learning_rate": 8.369453361983477e-06,
      "loss": 0.3891,
      "step": 18261
    },
    {
      "epoch": 0.8380524069570006,
      "grad_norm": 0.40092164278030396,
      "learning_rate": 8.36927220548663e-06,
      "loss": 0.3224,
      "step": 18262
    },
    {
      "epoch": 0.8380982974622551,
      "grad_norm": 0.4555911719799042,
      "learning_rate": 8.369091040887745e-06,
      "loss": 0.3705,
      "step": 18263
    },
    {
      "epoch": 0.8381441879675096,
      "grad_norm": 0.4745902419090271,
      "learning_rate": 8.368909868187263e-06,
      "loss": 0.391,
      "step": 18264
    },
    {
      "epoch": 0.8381900784727639,
      "grad_norm": 0.48085346817970276,
      "learning_rate": 8.368728687385614e-06,
      "loss": 0.4961,
      "step": 18265
    },
    {
      "epoch": 0.8382359689780184,
      "grad_norm": 0.41196852922439575,
      "learning_rate": 8.368547498483235e-06,
      "loss": 0.2914,
      "step": 18266
    },
    {
      "epoch": 0.8382818594832729,
      "grad_norm": 0.48756515979766846,
      "learning_rate": 8.368366301480564e-06,
      "loss": 0.5019,
      "step": 18267
    },
    {
      "epoch": 0.8383277499885273,
      "grad_norm": 0.4667830765247345,
      "learning_rate": 8.368185096378034e-06,
      "loss": 0.4408,
      "step": 18268
    },
    {
      "epoch": 0.8383736404937818,
      "grad_norm": 0.41695255041122437,
      "learning_rate": 8.368003883176082e-06,
      "loss": 0.3197,
      "step": 18269
    },
    {
      "epoch": 0.8384195309990363,
      "grad_norm": 0.4801861345767975,
      "learning_rate": 8.367822661875145e-06,
      "loss": 0.3842,
      "step": 18270
    },
    {
      "epoch": 0.8384654215042907,
      "grad_norm": 0.4447426497936249,
      "learning_rate": 8.367641432475654e-06,
      "loss": 0.3564,
      "step": 18271
    },
    {
      "epoch": 0.8385113120095452,
      "grad_norm": 0.476460725069046,
      "learning_rate": 8.367460194978051e-06,
      "loss": 0.4585,
      "step": 18272
    },
    {
      "epoch": 0.8385572025147997,
      "grad_norm": 0.43056103587150574,
      "learning_rate": 8.36727894938277e-06,
      "loss": 0.4062,
      "step": 18273
    },
    {
      "epoch": 0.8386030930200542,
      "grad_norm": 0.4621809422969818,
      "learning_rate": 8.367097695690243e-06,
      "loss": 0.3807,
      "step": 18274
    },
    {
      "epoch": 0.8386489835253086,
      "grad_norm": 0.47685378789901733,
      "learning_rate": 8.366916433900911e-06,
      "loss": 0.3642,
      "step": 18275
    },
    {
      "epoch": 0.8386948740305631,
      "grad_norm": 0.4333129823207855,
      "learning_rate": 8.366735164015208e-06,
      "loss": 0.3853,
      "step": 18276
    },
    {
      "epoch": 0.8387407645358176,
      "grad_norm": 0.45602378249168396,
      "learning_rate": 8.366553886033567e-06,
      "loss": 0.3948,
      "step": 18277
    },
    {
      "epoch": 0.838786655041072,
      "grad_norm": 0.4518270194530487,
      "learning_rate": 8.366372599956427e-06,
      "loss": 0.4088,
      "step": 18278
    },
    {
      "epoch": 0.8388325455463265,
      "grad_norm": 0.44950994849205017,
      "learning_rate": 8.366191305784223e-06,
      "loss": 0.3594,
      "step": 18279
    },
    {
      "epoch": 0.838878436051581,
      "grad_norm": 0.48702433705329895,
      "learning_rate": 8.366010003517392e-06,
      "loss": 0.4785,
      "step": 18280
    },
    {
      "epoch": 0.8389243265568354,
      "grad_norm": 0.5133915543556213,
      "learning_rate": 8.365828693156372e-06,
      "loss": 0.4987,
      "step": 18281
    },
    {
      "epoch": 0.8389702170620899,
      "grad_norm": 0.4809609353542328,
      "learning_rate": 8.365647374701592e-06,
      "loss": 0.3996,
      "step": 18282
    },
    {
      "epoch": 0.8390161075673443,
      "grad_norm": 0.45083382725715637,
      "learning_rate": 8.365466048153496e-06,
      "loss": 0.3895,
      "step": 18283
    },
    {
      "epoch": 0.8390619980725987,
      "grad_norm": 0.4805561602115631,
      "learning_rate": 8.365284713512513e-06,
      "loss": 0.4041,
      "step": 18284
    },
    {
      "epoch": 0.8391078885778532,
      "grad_norm": 0.4367426037788391,
      "learning_rate": 8.365103370779085e-06,
      "loss": 0.3589,
      "step": 18285
    },
    {
      "epoch": 0.8391537790831077,
      "grad_norm": 0.4850913882255554,
      "learning_rate": 8.364922019953644e-06,
      "loss": 0.4533,
      "step": 18286
    },
    {
      "epoch": 0.8391996695883621,
      "grad_norm": 0.4381568431854248,
      "learning_rate": 8.364740661036627e-06,
      "loss": 0.3806,
      "step": 18287
    },
    {
      "epoch": 0.8392455600936166,
      "grad_norm": 0.47120267152786255,
      "learning_rate": 8.364559294028472e-06,
      "loss": 0.472,
      "step": 18288
    },
    {
      "epoch": 0.8392914505988711,
      "grad_norm": 0.42072293162345886,
      "learning_rate": 8.364377918929613e-06,
      "loss": 0.3556,
      "step": 18289
    },
    {
      "epoch": 0.8393373411041256,
      "grad_norm": 0.44752365350723267,
      "learning_rate": 8.364196535740488e-06,
      "loss": 0.3455,
      "step": 18290
    },
    {
      "epoch": 0.83938323160938,
      "grad_norm": 0.4691773056983948,
      "learning_rate": 8.364015144461531e-06,
      "loss": 0.428,
      "step": 18291
    },
    {
      "epoch": 0.8394291221146345,
      "grad_norm": 0.4422893226146698,
      "learning_rate": 8.36383374509318e-06,
      "loss": 0.3528,
      "step": 18292
    },
    {
      "epoch": 0.839475012619889,
      "grad_norm": 0.4254034757614136,
      "learning_rate": 8.36365233763587e-06,
      "loss": 0.3421,
      "step": 18293
    },
    {
      "epoch": 0.8395209031251434,
      "grad_norm": 0.4813319444656372,
      "learning_rate": 8.363470922090038e-06,
      "loss": 0.4386,
      "step": 18294
    },
    {
      "epoch": 0.8395667936303979,
      "grad_norm": 0.460416316986084,
      "learning_rate": 8.36328949845612e-06,
      "loss": 0.3825,
      "step": 18295
    },
    {
      "epoch": 0.8396126841356524,
      "grad_norm": 0.41950246691703796,
      "learning_rate": 8.363108066734553e-06,
      "loss": 0.3682,
      "step": 18296
    },
    {
      "epoch": 0.8396585746409068,
      "grad_norm": 0.46988046169281006,
      "learning_rate": 8.362926626925771e-06,
      "loss": 0.4225,
      "step": 18297
    },
    {
      "epoch": 0.8397044651461613,
      "grad_norm": 0.4681702256202698,
      "learning_rate": 8.362745179030212e-06,
      "loss": 0.4359,
      "step": 18298
    },
    {
      "epoch": 0.8397503556514158,
      "grad_norm": 0.4623136520385742,
      "learning_rate": 8.362563723048314e-06,
      "loss": 0.4619,
      "step": 18299
    },
    {
      "epoch": 0.8397962461566701,
      "grad_norm": 0.4576776921749115,
      "learning_rate": 8.362382258980509e-06,
      "loss": 0.4467,
      "step": 18300
    },
    {
      "epoch": 0.8398421366619246,
      "grad_norm": 0.4294199049472809,
      "learning_rate": 8.362200786827237e-06,
      "loss": 0.3744,
      "step": 18301
    },
    {
      "epoch": 0.8398880271671791,
      "grad_norm": 0.45716628432273865,
      "learning_rate": 8.362019306588933e-06,
      "loss": 0.4641,
      "step": 18302
    },
    {
      "epoch": 0.8399339176724335,
      "grad_norm": 0.4630660116672516,
      "learning_rate": 8.361837818266034e-06,
      "loss": 0.4396,
      "step": 18303
    },
    {
      "epoch": 0.839979808177688,
      "grad_norm": 0.42552071809768677,
      "learning_rate": 8.361656321858975e-06,
      "loss": 0.3565,
      "step": 18304
    },
    {
      "epoch": 0.8400256986829425,
      "grad_norm": 0.421857625246048,
      "learning_rate": 8.361474817368197e-06,
      "loss": 0.3349,
      "step": 18305
    },
    {
      "epoch": 0.8400715891881969,
      "grad_norm": 0.47285741567611694,
      "learning_rate": 8.36129330479413e-06,
      "loss": 0.4435,
      "step": 18306
    },
    {
      "epoch": 0.8401174796934514,
      "grad_norm": 0.44761720299720764,
      "learning_rate": 8.361111784137212e-06,
      "loss": 0.3719,
      "step": 18307
    },
    {
      "epoch": 0.8401633701987059,
      "grad_norm": 0.42355242371559143,
      "learning_rate": 8.360930255397883e-06,
      "loss": 0.3288,
      "step": 18308
    },
    {
      "epoch": 0.8402092607039604,
      "grad_norm": 0.4656539857387543,
      "learning_rate": 8.36074871857658e-06,
      "loss": 0.3869,
      "step": 18309
    },
    {
      "epoch": 0.8402551512092148,
      "grad_norm": 0.4362870454788208,
      "learning_rate": 8.360567173673732e-06,
      "loss": 0.3263,
      "step": 18310
    },
    {
      "epoch": 0.8403010417144693,
      "grad_norm": 0.4727589190006256,
      "learning_rate": 8.360385620689783e-06,
      "loss": 0.5223,
      "step": 18311
    },
    {
      "epoch": 0.8403469322197238,
      "grad_norm": 0.48550984263420105,
      "learning_rate": 8.360204059625168e-06,
      "loss": 0.4289,
      "step": 18312
    },
    {
      "epoch": 0.8403928227249782,
      "grad_norm": 0.7085279822349548,
      "learning_rate": 8.360022490480322e-06,
      "loss": 0.4521,
      "step": 18313
    },
    {
      "epoch": 0.8404387132302327,
      "grad_norm": 0.4335705637931824,
      "learning_rate": 8.35984091325568e-06,
      "loss": 0.3139,
      "step": 18314
    },
    {
      "epoch": 0.8404846037354872,
      "grad_norm": 0.41161349415779114,
      "learning_rate": 8.359659327951684e-06,
      "loss": 0.2958,
      "step": 18315
    },
    {
      "epoch": 0.8405304942407416,
      "grad_norm": 0.47334349155426025,
      "learning_rate": 8.359477734568767e-06,
      "loss": 0.4034,
      "step": 18316
    },
    {
      "epoch": 0.840576384745996,
      "grad_norm": 0.42495861649513245,
      "learning_rate": 8.359296133107366e-06,
      "loss": 0.3786,
      "step": 18317
    },
    {
      "epoch": 0.8406222752512506,
      "grad_norm": 0.414435476064682,
      "learning_rate": 8.35911452356792e-06,
      "loss": 0.3136,
      "step": 18318
    },
    {
      "epoch": 0.8406681657565049,
      "grad_norm": 0.4815065860748291,
      "learning_rate": 8.358932905950861e-06,
      "loss": 0.3652,
      "step": 18319
    },
    {
      "epoch": 0.8407140562617594,
      "grad_norm": 0.4765389561653137,
      "learning_rate": 8.358751280256629e-06,
      "loss": 0.4399,
      "step": 18320
    },
    {
      "epoch": 0.8407599467670139,
      "grad_norm": 0.45406317710876465,
      "learning_rate": 8.358569646485661e-06,
      "loss": 0.3658,
      "step": 18321
    },
    {
      "epoch": 0.8408058372722683,
      "grad_norm": 0.4345114231109619,
      "learning_rate": 8.358388004638391e-06,
      "loss": 0.3932,
      "step": 18322
    },
    {
      "epoch": 0.8408517277775228,
      "grad_norm": 0.43068015575408936,
      "learning_rate": 8.35820635471526e-06,
      "loss": 0.3707,
      "step": 18323
    },
    {
      "epoch": 0.8408976182827773,
      "grad_norm": 0.44780364632606506,
      "learning_rate": 8.358024696716703e-06,
      "loss": 0.4148,
      "step": 18324
    },
    {
      "epoch": 0.8409435087880317,
      "grad_norm": 0.42362454533576965,
      "learning_rate": 8.357843030643154e-06,
      "loss": 0.3397,
      "step": 18325
    },
    {
      "epoch": 0.8409893992932862,
      "grad_norm": 0.43837878108024597,
      "learning_rate": 8.357661356495056e-06,
      "loss": 0.3599,
      "step": 18326
    },
    {
      "epoch": 0.8410352897985407,
      "grad_norm": 0.46053776144981384,
      "learning_rate": 8.35747967427284e-06,
      "loss": 0.4272,
      "step": 18327
    },
    {
      "epoch": 0.8410811803037952,
      "grad_norm": 0.5229914784431458,
      "learning_rate": 8.357297983976944e-06,
      "loss": 0.5187,
      "step": 18328
    },
    {
      "epoch": 0.8411270708090496,
      "grad_norm": 0.41766518354415894,
      "learning_rate": 8.357116285607808e-06,
      "loss": 0.325,
      "step": 18329
    },
    {
      "epoch": 0.8411729613143041,
      "grad_norm": 0.48048317432403564,
      "learning_rate": 8.356934579165865e-06,
      "loss": 0.4417,
      "step": 18330
    },
    {
      "epoch": 0.8412188518195586,
      "grad_norm": 0.4603572189807892,
      "learning_rate": 8.356752864651554e-06,
      "loss": 0.4003,
      "step": 18331
    },
    {
      "epoch": 0.841264742324813,
      "grad_norm": 0.42645716667175293,
      "learning_rate": 8.356571142065312e-06,
      "loss": 0.3305,
      "step": 18332
    },
    {
      "epoch": 0.8413106328300675,
      "grad_norm": 0.43814700841903687,
      "learning_rate": 8.356389411407575e-06,
      "loss": 0.3518,
      "step": 18333
    },
    {
      "epoch": 0.841356523335322,
      "grad_norm": 0.47320249676704407,
      "learning_rate": 8.356207672678783e-06,
      "loss": 0.4403,
      "step": 18334
    },
    {
      "epoch": 0.8414024138405763,
      "grad_norm": 0.4550647437572479,
      "learning_rate": 8.35602592587937e-06,
      "loss": 0.4552,
      "step": 18335
    },
    {
      "epoch": 0.8414483043458308,
      "grad_norm": 0.4302375614643097,
      "learning_rate": 8.355844171009773e-06,
      "loss": 0.342,
      "step": 18336
    },
    {
      "epoch": 0.8414941948510853,
      "grad_norm": 0.4398466944694519,
      "learning_rate": 8.355662408070431e-06,
      "loss": 0.3827,
      "step": 18337
    },
    {
      "epoch": 0.8415400853563397,
      "grad_norm": 0.4282795786857605,
      "learning_rate": 8.355480637061779e-06,
      "loss": 0.3622,
      "step": 18338
    },
    {
      "epoch": 0.8415859758615942,
      "grad_norm": 0.492548406124115,
      "learning_rate": 8.355298857984256e-06,
      "loss": 0.4309,
      "step": 18339
    },
    {
      "epoch": 0.8416318663668487,
      "grad_norm": 0.4126216471195221,
      "learning_rate": 8.355117070838296e-06,
      "loss": 0.339,
      "step": 18340
    },
    {
      "epoch": 0.8416777568721031,
      "grad_norm": 0.4651237726211548,
      "learning_rate": 8.35493527562434e-06,
      "loss": 0.4344,
      "step": 18341
    },
    {
      "epoch": 0.8417236473773576,
      "grad_norm": 0.4612409174442291,
      "learning_rate": 8.354753472342825e-06,
      "loss": 0.4002,
      "step": 18342
    },
    {
      "epoch": 0.8417695378826121,
      "grad_norm": 0.49226418137550354,
      "learning_rate": 8.354571660994183e-06,
      "loss": 0.4589,
      "step": 18343
    },
    {
      "epoch": 0.8418154283878666,
      "grad_norm": 0.5086532831192017,
      "learning_rate": 8.354389841578859e-06,
      "loss": 0.4529,
      "step": 18344
    },
    {
      "epoch": 0.841861318893121,
      "grad_norm": 0.45290786027908325,
      "learning_rate": 8.354208014097283e-06,
      "loss": 0.4188,
      "step": 18345
    },
    {
      "epoch": 0.8419072093983755,
      "grad_norm": 0.4725872874259949,
      "learning_rate": 8.354026178549898e-06,
      "loss": 0.3949,
      "step": 18346
    },
    {
      "epoch": 0.84195309990363,
      "grad_norm": 0.4875481128692627,
      "learning_rate": 8.353844334937136e-06,
      "loss": 0.4106,
      "step": 18347
    },
    {
      "epoch": 0.8419989904088844,
      "grad_norm": 0.47561973333358765,
      "learning_rate": 8.353662483259437e-06,
      "loss": 0.451,
      "step": 18348
    },
    {
      "epoch": 0.8420448809141389,
      "grad_norm": 0.43495455384254456,
      "learning_rate": 8.35348062351724e-06,
      "loss": 0.3278,
      "step": 18349
    },
    {
      "epoch": 0.8420907714193934,
      "grad_norm": 0.4717242121696472,
      "learning_rate": 8.35329875571098e-06,
      "loss": 0.3881,
      "step": 18350
    },
    {
      "epoch": 0.8421366619246478,
      "grad_norm": 0.4546155631542206,
      "learning_rate": 8.353116879841095e-06,
      "loss": 0.3838,
      "step": 18351
    },
    {
      "epoch": 0.8421825524299023,
      "grad_norm": 0.4862527549266815,
      "learning_rate": 8.352934995908021e-06,
      "loss": 0.4725,
      "step": 18352
    },
    {
      "epoch": 0.8422284429351568,
      "grad_norm": 0.4699269235134125,
      "learning_rate": 8.352753103912198e-06,
      "loss": 0.4307,
      "step": 18353
    },
    {
      "epoch": 0.8422743334404111,
      "grad_norm": 0.44853541254997253,
      "learning_rate": 8.35257120385406e-06,
      "loss": 0.3502,
      "step": 18354
    },
    {
      "epoch": 0.8423202239456656,
      "grad_norm": 0.5033037066459656,
      "learning_rate": 8.352389295734049e-06,
      "loss": 0.4932,
      "step": 18355
    },
    {
      "epoch": 0.8423661144509201,
      "grad_norm": 0.4705391526222229,
      "learning_rate": 8.352207379552598e-06,
      "loss": 0.3691,
      "step": 18356
    },
    {
      "epoch": 0.8424120049561745,
      "grad_norm": 0.5278042554855347,
      "learning_rate": 8.352025455310146e-06,
      "loss": 0.6233,
      "step": 18357
    },
    {
      "epoch": 0.842457895461429,
      "grad_norm": 0.48265302181243896,
      "learning_rate": 8.351843523007132e-06,
      "loss": 0.4672,
      "step": 18358
    },
    {
      "epoch": 0.8425037859666835,
      "grad_norm": 0.7072678804397583,
      "learning_rate": 8.351661582643991e-06,
      "loss": 0.4261,
      "step": 18359
    },
    {
      "epoch": 0.8425496764719379,
      "grad_norm": 0.4573521018028259,
      "learning_rate": 8.351479634221163e-06,
      "loss": 0.4106,
      "step": 18360
    },
    {
      "epoch": 0.8425955669771924,
      "grad_norm": 0.46562162041664124,
      "learning_rate": 8.351297677739084e-06,
      "loss": 0.4555,
      "step": 18361
    },
    {
      "epoch": 0.8426414574824469,
      "grad_norm": 0.4400818943977356,
      "learning_rate": 8.35111571319819e-06,
      "loss": 0.3902,
      "step": 18362
    },
    {
      "epoch": 0.8426873479877014,
      "grad_norm": 0.4723038971424103,
      "learning_rate": 8.350933740598921e-06,
      "loss": 0.4621,
      "step": 18363
    },
    {
      "epoch": 0.8427332384929558,
      "grad_norm": 0.4322201609611511,
      "learning_rate": 8.350751759941715e-06,
      "loss": 0.3817,
      "step": 18364
    },
    {
      "epoch": 0.8427791289982103,
      "grad_norm": 0.43046241998672485,
      "learning_rate": 8.350569771227008e-06,
      "loss": 0.3625,
      "step": 18365
    },
    {
      "epoch": 0.8428250195034648,
      "grad_norm": 0.4378136098384857,
      "learning_rate": 8.35038777445524e-06,
      "loss": 0.3413,
      "step": 18366
    },
    {
      "epoch": 0.8428709100087192,
      "grad_norm": 0.42299291491508484,
      "learning_rate": 8.350205769626844e-06,
      "loss": 0.3738,
      "step": 18367
    },
    {
      "epoch": 0.8429168005139737,
      "grad_norm": 0.44240838289260864,
      "learning_rate": 8.35002375674226e-06,
      "loss": 0.3759,
      "step": 18368
    },
    {
      "epoch": 0.8429626910192282,
      "grad_norm": 0.45195770263671875,
      "learning_rate": 8.349841735801929e-06,
      "loss": 0.4503,
      "step": 18369
    },
    {
      "epoch": 0.8430085815244825,
      "grad_norm": 0.4279383718967438,
      "learning_rate": 8.349659706806284e-06,
      "loss": 0.3035,
      "step": 18370
    },
    {
      "epoch": 0.843054472029737,
      "grad_norm": 0.44820088148117065,
      "learning_rate": 8.349477669755766e-06,
      "loss": 0.3793,
      "step": 18371
    },
    {
      "epoch": 0.8431003625349915,
      "grad_norm": 0.45392870903015137,
      "learning_rate": 8.34929562465081e-06,
      "loss": 0.3796,
      "step": 18372
    },
    {
      "epoch": 0.8431462530402459,
      "grad_norm": 0.512766420841217,
      "learning_rate": 8.349113571491856e-06,
      "loss": 0.4709,
      "step": 18373
    },
    {
      "epoch": 0.8431921435455004,
      "grad_norm": 0.4132944643497467,
      "learning_rate": 8.34893151027934e-06,
      "loss": 0.3124,
      "step": 18374
    },
    {
      "epoch": 0.8432380340507549,
      "grad_norm": 0.44059717655181885,
      "learning_rate": 8.348749441013702e-06,
      "loss": 0.3556,
      "step": 18375
    },
    {
      "epoch": 0.8432839245560093,
      "grad_norm": 0.4390440881252289,
      "learning_rate": 8.348567363695376e-06,
      "loss": 0.3768,
      "step": 18376
    },
    {
      "epoch": 0.8433298150612638,
      "grad_norm": 0.4518889784812927,
      "learning_rate": 8.348385278324805e-06,
      "loss": 0.3843,
      "step": 18377
    },
    {
      "epoch": 0.8433757055665183,
      "grad_norm": 0.42900529503822327,
      "learning_rate": 8.348203184902423e-06,
      "loss": 0.3658,
      "step": 18378
    },
    {
      "epoch": 0.8434215960717728,
      "grad_norm": 0.4830401539802551,
      "learning_rate": 8.348021083428667e-06,
      "loss": 0.493,
      "step": 18379
    },
    {
      "epoch": 0.8434674865770272,
      "grad_norm": 0.4534807503223419,
      "learning_rate": 8.34783897390398e-06,
      "loss": 0.365,
      "step": 18380
    },
    {
      "epoch": 0.8435133770822817,
      "grad_norm": 0.4254912734031677,
      "learning_rate": 8.347656856328796e-06,
      "loss": 0.351,
      "step": 18381
    },
    {
      "epoch": 0.8435592675875362,
      "grad_norm": 0.4541734755039215,
      "learning_rate": 8.347474730703555e-06,
      "loss": 0.3866,
      "step": 18382
    },
    {
      "epoch": 0.8436051580927906,
      "grad_norm": 0.4775274097919464,
      "learning_rate": 8.347292597028691e-06,
      "loss": 0.4458,
      "step": 18383
    },
    {
      "epoch": 0.8436510485980451,
      "grad_norm": 0.4680865406990051,
      "learning_rate": 8.347110455304649e-06,
      "loss": 0.4412,
      "step": 18384
    },
    {
      "epoch": 0.8436969391032996,
      "grad_norm": 0.468293160200119,
      "learning_rate": 8.346928305531858e-06,
      "loss": 0.4562,
      "step": 18385
    },
    {
      "epoch": 0.843742829608554,
      "grad_norm": 0.47502392530441284,
      "learning_rate": 8.346746147710764e-06,
      "loss": 0.4234,
      "step": 18386
    },
    {
      "epoch": 0.8437887201138085,
      "grad_norm": 0.45605209469795227,
      "learning_rate": 8.3465639818418e-06,
      "loss": 0.3644,
      "step": 18387
    },
    {
      "epoch": 0.843834610619063,
      "grad_norm": 0.4365732669830322,
      "learning_rate": 8.346381807925407e-06,
      "loss": 0.4029,
      "step": 18388
    },
    {
      "epoch": 0.8438805011243173,
      "grad_norm": 0.44695624709129333,
      "learning_rate": 8.34619962596202e-06,
      "loss": 0.3661,
      "step": 18389
    },
    {
      "epoch": 0.8439263916295718,
      "grad_norm": 0.6114743947982788,
      "learning_rate": 8.34601743595208e-06,
      "loss": 0.4372,
      "step": 18390
    },
    {
      "epoch": 0.8439722821348263,
      "grad_norm": 0.45880502462387085,
      "learning_rate": 8.345835237896025e-06,
      "loss": 0.4248,
      "step": 18391
    },
    {
      "epoch": 0.8440181726400807,
      "grad_norm": 0.4660395085811615,
      "learning_rate": 8.345653031794292e-06,
      "loss": 0.4061,
      "step": 18392
    },
    {
      "epoch": 0.8440640631453352,
      "grad_norm": 0.4742988049983978,
      "learning_rate": 8.345470817647319e-06,
      "loss": 0.4584,
      "step": 18393
    },
    {
      "epoch": 0.8441099536505897,
      "grad_norm": 0.45204228162765503,
      "learning_rate": 8.345288595455546e-06,
      "loss": 0.3911,
      "step": 18394
    },
    {
      "epoch": 0.8441558441558441,
      "grad_norm": 0.44109320640563965,
      "learning_rate": 8.345106365219406e-06,
      "loss": 0.3983,
      "step": 18395
    },
    {
      "epoch": 0.8442017346610986,
      "grad_norm": 0.4372449517250061,
      "learning_rate": 8.344924126939343e-06,
      "loss": 0.3579,
      "step": 18396
    },
    {
      "epoch": 0.8442476251663531,
      "grad_norm": 0.47066545486450195,
      "learning_rate": 8.344741880615795e-06,
      "loss": 0.4207,
      "step": 18397
    },
    {
      "epoch": 0.8442935156716076,
      "grad_norm": 0.3910118043422699,
      "learning_rate": 8.344559626249196e-06,
      "loss": 0.2738,
      "step": 18398
    },
    {
      "epoch": 0.844339406176862,
      "grad_norm": 0.4371149241924286,
      "learning_rate": 8.344377363839987e-06,
      "loss": 0.3625,
      "step": 18399
    },
    {
      "epoch": 0.8443852966821165,
      "grad_norm": 0.5049681663513184,
      "learning_rate": 8.344195093388608e-06,
      "loss": 0.5397,
      "step": 18400
    },
    {
      "epoch": 0.844431187187371,
      "grad_norm": 0.4227767288684845,
      "learning_rate": 8.344012814895492e-06,
      "loss": 0.3677,
      "step": 18401
    },
    {
      "epoch": 0.8444770776926254,
      "grad_norm": 0.48698192834854126,
      "learning_rate": 8.343830528361084e-06,
      "loss": 0.4733,
      "step": 18402
    },
    {
      "epoch": 0.8445229681978799,
      "grad_norm": 0.4347972273826599,
      "learning_rate": 8.343648233785815e-06,
      "loss": 0.3526,
      "step": 18403
    },
    {
      "epoch": 0.8445688587031344,
      "grad_norm": 0.48539137840270996,
      "learning_rate": 8.34346593117013e-06,
      "loss": 0.4649,
      "step": 18404
    },
    {
      "epoch": 0.8446147492083887,
      "grad_norm": 0.4273861348628998,
      "learning_rate": 8.343283620514464e-06,
      "loss": 0.3588,
      "step": 18405
    },
    {
      "epoch": 0.8446606397136432,
      "grad_norm": 0.4330707788467407,
      "learning_rate": 8.343101301819256e-06,
      "loss": 0.3516,
      "step": 18406
    },
    {
      "epoch": 0.8447065302188977,
      "grad_norm": 0.6371825337409973,
      "learning_rate": 8.342918975084946e-06,
      "loss": 0.3855,
      "step": 18407
    },
    {
      "epoch": 0.8447524207241521,
      "grad_norm": 0.437532514333725,
      "learning_rate": 8.34273664031197e-06,
      "loss": 0.3413,
      "step": 18408
    },
    {
      "epoch": 0.8447983112294066,
      "grad_norm": 0.44700440764427185,
      "learning_rate": 8.342554297500766e-06,
      "loss": 0.3565,
      "step": 18409
    },
    {
      "epoch": 0.8448442017346611,
      "grad_norm": 0.4678964912891388,
      "learning_rate": 8.342371946651775e-06,
      "loss": 0.4224,
      "step": 18410
    },
    {
      "epoch": 0.8448900922399155,
      "grad_norm": 0.5274521708488464,
      "learning_rate": 8.342189587765435e-06,
      "loss": 0.5183,
      "step": 18411
    },
    {
      "epoch": 0.84493598274517,
      "grad_norm": 0.47910451889038086,
      "learning_rate": 8.342007220842182e-06,
      "loss": 0.4269,
      "step": 18412
    },
    {
      "epoch": 0.8449818732504245,
      "grad_norm": 0.44621360301971436,
      "learning_rate": 8.341824845882457e-06,
      "loss": 0.3836,
      "step": 18413
    },
    {
      "epoch": 0.8450277637556789,
      "grad_norm": 0.4184834361076355,
      "learning_rate": 8.341642462886699e-06,
      "loss": 0.319,
      "step": 18414
    },
    {
      "epoch": 0.8450736542609334,
      "grad_norm": 0.42441990971565247,
      "learning_rate": 8.341460071855345e-06,
      "loss": 0.3375,
      "step": 18415
    },
    {
      "epoch": 0.8451195447661879,
      "grad_norm": 0.4489527940750122,
      "learning_rate": 8.341277672788834e-06,
      "loss": 0.4049,
      "step": 18416
    },
    {
      "epoch": 0.8451654352714424,
      "grad_norm": 0.4490134119987488,
      "learning_rate": 8.341095265687605e-06,
      "loss": 0.4034,
      "step": 18417
    },
    {
      "epoch": 0.8452113257766968,
      "grad_norm": 0.44045600295066833,
      "learning_rate": 8.340912850552096e-06,
      "loss": 0.3914,
      "step": 18418
    },
    {
      "epoch": 0.8452572162819513,
      "grad_norm": 0.4616234004497528,
      "learning_rate": 8.340730427382745e-06,
      "loss": 0.4417,
      "step": 18419
    },
    {
      "epoch": 0.8453031067872058,
      "grad_norm": 0.4152618646621704,
      "learning_rate": 8.340547996179994e-06,
      "loss": 0.3694,
      "step": 18420
    },
    {
      "epoch": 0.8453489972924602,
      "grad_norm": 0.45368459820747375,
      "learning_rate": 8.340365556944278e-06,
      "loss": 0.4182,
      "step": 18421
    },
    {
      "epoch": 0.8453948877977147,
      "grad_norm": 0.4593093991279602,
      "learning_rate": 8.340183109676037e-06,
      "loss": 0.3833,
      "step": 18422
    },
    {
      "epoch": 0.8454407783029692,
      "grad_norm": 0.44983601570129395,
      "learning_rate": 8.340000654375711e-06,
      "loss": 0.3924,
      "step": 18423
    },
    {
      "epoch": 0.8454866688082235,
      "grad_norm": 0.46669793128967285,
      "learning_rate": 8.339818191043736e-06,
      "loss": 0.5078,
      "step": 18424
    },
    {
      "epoch": 0.845532559313478,
      "grad_norm": 0.4939838647842407,
      "learning_rate": 8.339635719680554e-06,
      "loss": 0.4607,
      "step": 18425
    },
    {
      "epoch": 0.8455784498187325,
      "grad_norm": 0.44249817728996277,
      "learning_rate": 8.3394532402866e-06,
      "loss": 0.3549,
      "step": 18426
    },
    {
      "epoch": 0.8456243403239869,
      "grad_norm": 0.46848225593566895,
      "learning_rate": 8.339270752862316e-06,
      "loss": 0.4172,
      "step": 18427
    },
    {
      "epoch": 0.8456702308292414,
      "grad_norm": 0.44798702001571655,
      "learning_rate": 8.33908825740814e-06,
      "loss": 0.3945,
      "step": 18428
    },
    {
      "epoch": 0.8457161213344959,
      "grad_norm": 0.4461829662322998,
      "learning_rate": 8.338905753924512e-06,
      "loss": 0.3975,
      "step": 18429
    },
    {
      "epoch": 0.8457620118397503,
      "grad_norm": 0.47843924164772034,
      "learning_rate": 8.338723242411866e-06,
      "loss": 0.4731,
      "step": 18430
    },
    {
      "epoch": 0.8458079023450048,
      "grad_norm": 0.44294023513793945,
      "learning_rate": 8.338540722870647e-06,
      "loss": 0.4116,
      "step": 18431
    },
    {
      "epoch": 0.8458537928502593,
      "grad_norm": 0.4541102349758148,
      "learning_rate": 8.338358195301291e-06,
      "loss": 0.3913,
      "step": 18432
    },
    {
      "epoch": 0.8458996833555138,
      "grad_norm": 0.43997722864151,
      "learning_rate": 8.338175659704236e-06,
      "loss": 0.4244,
      "step": 18433
    },
    {
      "epoch": 0.8459455738607682,
      "grad_norm": 0.43457984924316406,
      "learning_rate": 8.337993116079922e-06,
      "loss": 0.3292,
      "step": 18434
    },
    {
      "epoch": 0.8459914643660227,
      "grad_norm": 0.5279772877693176,
      "learning_rate": 8.337810564428789e-06,
      "loss": 0.4977,
      "step": 18435
    },
    {
      "epoch": 0.8460373548712772,
      "grad_norm": 0.46021613478660583,
      "learning_rate": 8.337628004751275e-06,
      "loss": 0.4089,
      "step": 18436
    },
    {
      "epoch": 0.8460832453765316,
      "grad_norm": 0.4456234872341156,
      "learning_rate": 8.337445437047819e-06,
      "loss": 0.4165,
      "step": 18437
    },
    {
      "epoch": 0.8461291358817861,
      "grad_norm": 0.44975969195365906,
      "learning_rate": 8.33726286131886e-06,
      "loss": 0.3651,
      "step": 18438
    },
    {
      "epoch": 0.8461750263870406,
      "grad_norm": 0.4139656722545624,
      "learning_rate": 8.337080277564837e-06,
      "loss": 0.3413,
      "step": 18439
    },
    {
      "epoch": 0.846220916892295,
      "grad_norm": 0.49429044127464294,
      "learning_rate": 8.336897685786189e-06,
      "loss": 0.4377,
      "step": 18440
    },
    {
      "epoch": 0.8462668073975494,
      "grad_norm": 0.4104306101799011,
      "learning_rate": 8.336715085983355e-06,
      "loss": 0.3182,
      "step": 18441
    },
    {
      "epoch": 0.846312697902804,
      "grad_norm": 0.4455753564834595,
      "learning_rate": 8.336532478156775e-06,
      "loss": 0.3796,
      "step": 18442
    },
    {
      "epoch": 0.8463585884080583,
      "grad_norm": 0.45576009154319763,
      "learning_rate": 8.336349862306887e-06,
      "loss": 0.4097,
      "step": 18443
    },
    {
      "epoch": 0.8464044789133128,
      "grad_norm": 0.4532299339771271,
      "learning_rate": 8.33616723843413e-06,
      "loss": 0.3906,
      "step": 18444
    },
    {
      "epoch": 0.8464503694185673,
      "grad_norm": 0.39093905687332153,
      "learning_rate": 8.335984606538946e-06,
      "loss": 0.2928,
      "step": 18445
    },
    {
      "epoch": 0.8464962599238217,
      "grad_norm": 0.4254453778266907,
      "learning_rate": 8.335801966621768e-06,
      "loss": 0.3199,
      "step": 18446
    },
    {
      "epoch": 0.8465421504290762,
      "grad_norm": 0.4364699125289917,
      "learning_rate": 8.335619318683042e-06,
      "loss": 0.363,
      "step": 18447
    },
    {
      "epoch": 0.8465880409343307,
      "grad_norm": 0.440714955329895,
      "learning_rate": 8.335436662723203e-06,
      "loss": 0.3435,
      "step": 18448
    },
    {
      "epoch": 0.8466339314395851,
      "grad_norm": 0.4626173675060272,
      "learning_rate": 8.335253998742692e-06,
      "loss": 0.4146,
      "step": 18449
    },
    {
      "epoch": 0.8466798219448396,
      "grad_norm": 0.4529416859149933,
      "learning_rate": 8.335071326741948e-06,
      "loss": 0.3727,
      "step": 18450
    },
    {
      "epoch": 0.8467257124500941,
      "grad_norm": 0.44712498784065247,
      "learning_rate": 8.334888646721411e-06,
      "loss": 0.4054,
      "step": 18451
    },
    {
      "epoch": 0.8467716029553486,
      "grad_norm": 0.45306190848350525,
      "learning_rate": 8.334705958681517e-06,
      "loss": 0.4309,
      "step": 18452
    },
    {
      "epoch": 0.846817493460603,
      "grad_norm": 0.44328179955482483,
      "learning_rate": 8.334523262622708e-06,
      "loss": 0.3426,
      "step": 18453
    },
    {
      "epoch": 0.8468633839658575,
      "grad_norm": 0.42018651962280273,
      "learning_rate": 8.334340558545425e-06,
      "loss": 0.3619,
      "step": 18454
    },
    {
      "epoch": 0.846909274471112,
      "grad_norm": 0.45521971583366394,
      "learning_rate": 8.334157846450104e-06,
      "loss": 0.3669,
      "step": 18455
    },
    {
      "epoch": 0.8469551649763664,
      "grad_norm": 0.4465028643608093,
      "learning_rate": 8.333975126337187e-06,
      "loss": 0.3927,
      "step": 18456
    },
    {
      "epoch": 0.8470010554816209,
      "grad_norm": 0.4658412039279938,
      "learning_rate": 8.33379239820711e-06,
      "loss": 0.3816,
      "step": 18457
    },
    {
      "epoch": 0.8470469459868754,
      "grad_norm": 0.4661136567592621,
      "learning_rate": 8.333609662060315e-06,
      "loss": 0.404,
      "step": 18458
    },
    {
      "epoch": 0.8470928364921297,
      "grad_norm": 0.46162107586860657,
      "learning_rate": 8.333426917897241e-06,
      "loss": 0.4418,
      "step": 18459
    },
    {
      "epoch": 0.8471387269973842,
      "grad_norm": 0.4501282572746277,
      "learning_rate": 8.333244165718329e-06,
      "loss": 0.3695,
      "step": 18460
    },
    {
      "epoch": 0.8471846175026387,
      "grad_norm": 0.4711464047431946,
      "learning_rate": 8.333061405524015e-06,
      "loss": 0.3887,
      "step": 18461
    },
    {
      "epoch": 0.8472305080078931,
      "grad_norm": 0.4515655040740967,
      "learning_rate": 8.33287863731474e-06,
      "loss": 0.3872,
      "step": 18462
    },
    {
      "epoch": 0.8472763985131476,
      "grad_norm": 0.4372484087944031,
      "learning_rate": 8.332695861090945e-06,
      "loss": 0.4313,
      "step": 18463
    },
    {
      "epoch": 0.8473222890184021,
      "grad_norm": 0.4591462314128876,
      "learning_rate": 8.332513076853065e-06,
      "loss": 0.3766,
      "step": 18464
    },
    {
      "epoch": 0.8473681795236565,
      "grad_norm": 0.42607882618904114,
      "learning_rate": 8.332330284601548e-06,
      "loss": 0.3273,
      "step": 18465
    },
    {
      "epoch": 0.847414070028911,
      "grad_norm": 0.44026491045951843,
      "learning_rate": 8.332147484336825e-06,
      "loss": 0.3971,
      "step": 18466
    },
    {
      "epoch": 0.8474599605341655,
      "grad_norm": 0.43532034754753113,
      "learning_rate": 8.331964676059339e-06,
      "loss": 0.3756,
      "step": 18467
    },
    {
      "epoch": 0.8475058510394199,
      "grad_norm": 0.43992215394973755,
      "learning_rate": 8.331781859769531e-06,
      "loss": 0.3814,
      "step": 18468
    },
    {
      "epoch": 0.8475517415446744,
      "grad_norm": 0.4336024522781372,
      "learning_rate": 8.331599035467838e-06,
      "loss": 0.4217,
      "step": 18469
    },
    {
      "epoch": 0.8475976320499289,
      "grad_norm": 0.44434642791748047,
      "learning_rate": 8.331416203154701e-06,
      "loss": 0.3816,
      "step": 18470
    },
    {
      "epoch": 0.8476435225551834,
      "grad_norm": 0.417959600687027,
      "learning_rate": 8.33123336283056e-06,
      "loss": 0.3413,
      "step": 18471
    },
    {
      "epoch": 0.8476894130604378,
      "grad_norm": 0.44532114267349243,
      "learning_rate": 8.331050514495854e-06,
      "loss": 0.391,
      "step": 18472
    },
    {
      "epoch": 0.8477353035656923,
      "grad_norm": 0.4096547067165375,
      "learning_rate": 8.330867658151022e-06,
      "loss": 0.3494,
      "step": 18473
    },
    {
      "epoch": 0.8477811940709468,
      "grad_norm": 0.45381149649620056,
      "learning_rate": 8.330684793796506e-06,
      "loss": 0.391,
      "step": 18474
    },
    {
      "epoch": 0.8478270845762012,
      "grad_norm": 0.44133439660072327,
      "learning_rate": 8.330501921432744e-06,
      "loss": 0.4022,
      "step": 18475
    },
    {
      "epoch": 0.8478729750814556,
      "grad_norm": 0.46249449253082275,
      "learning_rate": 8.330319041060175e-06,
      "loss": 0.3803,
      "step": 18476
    },
    {
      "epoch": 0.8479188655867101,
      "grad_norm": 0.43686965107917786,
      "learning_rate": 8.33013615267924e-06,
      "loss": 0.4109,
      "step": 18477
    },
    {
      "epoch": 0.8479647560919645,
      "grad_norm": 0.450529009103775,
      "learning_rate": 8.329953256290379e-06,
      "loss": 0.3488,
      "step": 18478
    },
    {
      "epoch": 0.848010646597219,
      "grad_norm": 0.48208731412887573,
      "learning_rate": 8.329770351894031e-06,
      "loss": 0.4748,
      "step": 18479
    },
    {
      "epoch": 0.8480565371024735,
      "grad_norm": 0.47667890787124634,
      "learning_rate": 8.329587439490636e-06,
      "loss": 0.5026,
      "step": 18480
    },
    {
      "epoch": 0.8481024276077279,
      "grad_norm": 0.4826364815235138,
      "learning_rate": 8.329404519080634e-06,
      "loss": 0.4538,
      "step": 18481
    },
    {
      "epoch": 0.8481483181129824,
      "grad_norm": 0.4959820806980133,
      "learning_rate": 8.329221590664467e-06,
      "loss": 0.4832,
      "step": 18482
    },
    {
      "epoch": 0.8481942086182369,
      "grad_norm": 0.4422644078731537,
      "learning_rate": 8.329038654242572e-06,
      "loss": 0.4061,
      "step": 18483
    },
    {
      "epoch": 0.8482400991234913,
      "grad_norm": 0.4402746856212616,
      "learning_rate": 8.328855709815388e-06,
      "loss": 0.4055,
      "step": 18484
    },
    {
      "epoch": 0.8482859896287458,
      "grad_norm": 0.44546085596084595,
      "learning_rate": 8.328672757383358e-06,
      "loss": 0.3857,
      "step": 18485
    },
    {
      "epoch": 0.8483318801340003,
      "grad_norm": 0.41366586089134216,
      "learning_rate": 8.32848979694692e-06,
      "loss": 0.3514,
      "step": 18486
    },
    {
      "epoch": 0.8483777706392548,
      "grad_norm": 0.4797897934913635,
      "learning_rate": 8.328306828506515e-06,
      "loss": 0.4058,
      "step": 18487
    },
    {
      "epoch": 0.8484236611445092,
      "grad_norm": 0.4312388002872467,
      "learning_rate": 8.328123852062582e-06,
      "loss": 0.3356,
      "step": 18488
    },
    {
      "epoch": 0.8484695516497637,
      "grad_norm": 0.443000465631485,
      "learning_rate": 8.327940867615561e-06,
      "loss": 0.4115,
      "step": 18489
    },
    {
      "epoch": 0.8485154421550182,
      "grad_norm": 0.47328194975852966,
      "learning_rate": 8.327757875165894e-06,
      "loss": 0.4985,
      "step": 18490
    },
    {
      "epoch": 0.8485613326602726,
      "grad_norm": 0.42449358105659485,
      "learning_rate": 8.32757487471402e-06,
      "loss": 0.3807,
      "step": 18491
    },
    {
      "epoch": 0.8486072231655271,
      "grad_norm": 0.4469325840473175,
      "learning_rate": 8.327391866260376e-06,
      "loss": 0.382,
      "step": 18492
    },
    {
      "epoch": 0.8486531136707816,
      "grad_norm": 0.47733375430107117,
      "learning_rate": 8.327208849805407e-06,
      "loss": 0.403,
      "step": 18493
    },
    {
      "epoch": 0.8486990041760359,
      "grad_norm": 0.5070577263832092,
      "learning_rate": 8.32702582534955e-06,
      "loss": 0.4304,
      "step": 18494
    },
    {
      "epoch": 0.8487448946812904,
      "grad_norm": 0.46794626116752625,
      "learning_rate": 8.326842792893245e-06,
      "loss": 0.4226,
      "step": 18495
    },
    {
      "epoch": 0.8487907851865449,
      "grad_norm": 0.4611136019229889,
      "learning_rate": 8.326659752436933e-06,
      "loss": 0.4079,
      "step": 18496
    },
    {
      "epoch": 0.8488366756917993,
      "grad_norm": 0.4353518784046173,
      "learning_rate": 8.326476703981056e-06,
      "loss": 0.3284,
      "step": 18497
    },
    {
      "epoch": 0.8488825661970538,
      "grad_norm": 0.4331098794937134,
      "learning_rate": 8.326293647526052e-06,
      "loss": 0.4093,
      "step": 18498
    },
    {
      "epoch": 0.8489284567023083,
      "grad_norm": 0.5203938484191895,
      "learning_rate": 8.326110583072362e-06,
      "loss": 0.4435,
      "step": 18499
    },
    {
      "epoch": 0.8489743472075627,
      "grad_norm": 0.4364112913608551,
      "learning_rate": 8.325927510620426e-06,
      "loss": 0.391,
      "step": 18500
    },
    {
      "epoch": 0.8490202377128172,
      "grad_norm": 0.45561668276786804,
      "learning_rate": 8.32574443017068e-06,
      "loss": 0.3826,
      "step": 18501
    },
    {
      "epoch": 0.8490661282180717,
      "grad_norm": 0.4086368680000305,
      "learning_rate": 8.325561341723574e-06,
      "loss": 0.3244,
      "step": 18502
    },
    {
      "epoch": 0.8491120187233261,
      "grad_norm": 0.4363986849784851,
      "learning_rate": 8.325378245279538e-06,
      "loss": 0.3655,
      "step": 18503
    },
    {
      "epoch": 0.8491579092285806,
      "grad_norm": 0.42900383472442627,
      "learning_rate": 8.32519514083902e-06,
      "loss": 0.3601,
      "step": 18504
    },
    {
      "epoch": 0.8492037997338351,
      "grad_norm": 0.468606173992157,
      "learning_rate": 8.325012028402455e-06,
      "loss": 0.4435,
      "step": 18505
    },
    {
      "epoch": 0.8492496902390896,
      "grad_norm": 0.4418804347515106,
      "learning_rate": 8.324828907970287e-06,
      "loss": 0.3598,
      "step": 18506
    },
    {
      "epoch": 0.849295580744344,
      "grad_norm": 0.47901755571365356,
      "learning_rate": 8.324645779542954e-06,
      "loss": 0.4437,
      "step": 18507
    },
    {
      "epoch": 0.8493414712495985,
      "grad_norm": 0.4587444067001343,
      "learning_rate": 8.324462643120898e-06,
      "loss": 0.4468,
      "step": 18508
    },
    {
      "epoch": 0.849387361754853,
      "grad_norm": 0.48112258315086365,
      "learning_rate": 8.324279498704558e-06,
      "loss": 0.4598,
      "step": 18509
    },
    {
      "epoch": 0.8494332522601074,
      "grad_norm": 0.4726746380329132,
      "learning_rate": 8.324096346294375e-06,
      "loss": 0.433,
      "step": 18510
    },
    {
      "epoch": 0.8494791427653618,
      "grad_norm": 0.43957963585853577,
      "learning_rate": 8.323913185890788e-06,
      "loss": 0.3793,
      "step": 18511
    },
    {
      "epoch": 0.8495250332706163,
      "grad_norm": 0.5012571215629578,
      "learning_rate": 8.32373001749424e-06,
      "loss": 0.4783,
      "step": 18512
    },
    {
      "epoch": 0.8495709237758707,
      "grad_norm": 0.4703497588634491,
      "learning_rate": 8.323546841105172e-06,
      "loss": 0.4067,
      "step": 18513
    },
    {
      "epoch": 0.8496168142811252,
      "grad_norm": 0.44172635674476624,
      "learning_rate": 8.32336365672402e-06,
      "loss": 0.3572,
      "step": 18514
    },
    {
      "epoch": 0.8496627047863797,
      "grad_norm": 0.452589750289917,
      "learning_rate": 8.323180464351228e-06,
      "loss": 0.4158,
      "step": 18515
    },
    {
      "epoch": 0.8497085952916341,
      "grad_norm": 0.4321780502796173,
      "learning_rate": 8.322997263987239e-06,
      "loss": 0.339,
      "step": 18516
    },
    {
      "epoch": 0.8497544857968886,
      "grad_norm": 0.4353664517402649,
      "learning_rate": 8.322814055632486e-06,
      "loss": 0.3732,
      "step": 18517
    },
    {
      "epoch": 0.8498003763021431,
      "grad_norm": 0.45058342814445496,
      "learning_rate": 8.322630839287416e-06,
      "loss": 0.397,
      "step": 18518
    },
    {
      "epoch": 0.8498462668073975,
      "grad_norm": 0.430913507938385,
      "learning_rate": 8.322447614952468e-06,
      "loss": 0.3457,
      "step": 18519
    },
    {
      "epoch": 0.849892157312652,
      "grad_norm": 0.46599411964416504,
      "learning_rate": 8.32226438262808e-06,
      "loss": 0.4194,
      "step": 18520
    },
    {
      "epoch": 0.8499380478179065,
      "grad_norm": 0.47300225496292114,
      "learning_rate": 8.322081142314696e-06,
      "loss": 0.4706,
      "step": 18521
    },
    {
      "epoch": 0.849983938323161,
      "grad_norm": 0.4319131374359131,
      "learning_rate": 8.321897894012756e-06,
      "loss": 0.3355,
      "step": 18522
    },
    {
      "epoch": 0.8500298288284154,
      "grad_norm": 0.46814319491386414,
      "learning_rate": 8.321714637722697e-06,
      "loss": 0.3992,
      "step": 18523
    },
    {
      "epoch": 0.8500757193336699,
      "grad_norm": 0.46923312544822693,
      "learning_rate": 8.321531373444966e-06,
      "loss": 0.3778,
      "step": 18524
    },
    {
      "epoch": 0.8501216098389244,
      "grad_norm": 0.4729360342025757,
      "learning_rate": 8.32134810118e-06,
      "loss": 0.4113,
      "step": 18525
    },
    {
      "epoch": 0.8501675003441788,
      "grad_norm": 0.4241567552089691,
      "learning_rate": 8.321164820928238e-06,
      "loss": 0.3441,
      "step": 18526
    },
    {
      "epoch": 0.8502133908494333,
      "grad_norm": 0.5104889869689941,
      "learning_rate": 8.320981532690125e-06,
      "loss": 0.5479,
      "step": 18527
    },
    {
      "epoch": 0.8502592813546878,
      "grad_norm": 0.41151878237724304,
      "learning_rate": 8.3207982364661e-06,
      "loss": 0.344,
      "step": 18528
    },
    {
      "epoch": 0.8503051718599421,
      "grad_norm": 0.44778236746788025,
      "learning_rate": 8.3206149322566e-06,
      "loss": 0.369,
      "step": 18529
    },
    {
      "epoch": 0.8503510623651966,
      "grad_norm": 0.42515379190444946,
      "learning_rate": 8.320431620062072e-06,
      "loss": 0.3764,
      "step": 18530
    },
    {
      "epoch": 0.8503969528704511,
      "grad_norm": 0.45258450508117676,
      "learning_rate": 8.320248299882952e-06,
      "loss": 0.4129,
      "step": 18531
    },
    {
      "epoch": 0.8504428433757055,
      "grad_norm": 0.4497392475605011,
      "learning_rate": 8.320064971719684e-06,
      "loss": 0.3608,
      "step": 18532
    },
    {
      "epoch": 0.85048873388096,
      "grad_norm": 0.4949532151222229,
      "learning_rate": 8.319881635572708e-06,
      "loss": 0.4794,
      "step": 18533
    },
    {
      "epoch": 0.8505346243862145,
      "grad_norm": 0.5146548748016357,
      "learning_rate": 8.319698291442462e-06,
      "loss": 0.3877,
      "step": 18534
    },
    {
      "epoch": 0.8505805148914689,
      "grad_norm": 0.4761143922805786,
      "learning_rate": 8.319514939329393e-06,
      "loss": 0.4111,
      "step": 18535
    },
    {
      "epoch": 0.8506264053967234,
      "grad_norm": 0.4928510785102844,
      "learning_rate": 8.319331579233937e-06,
      "loss": 0.5018,
      "step": 18536
    },
    {
      "epoch": 0.8506722959019779,
      "grad_norm": 0.49157974123954773,
      "learning_rate": 8.319148211156534e-06,
      "loss": 0.4549,
      "step": 18537
    },
    {
      "epoch": 0.8507181864072323,
      "grad_norm": 0.5313320159912109,
      "learning_rate": 8.318964835097629e-06,
      "loss": 0.4106,
      "step": 18538
    },
    {
      "epoch": 0.8507640769124868,
      "grad_norm": 0.45103299617767334,
      "learning_rate": 8.31878145105766e-06,
      "loss": 0.3407,
      "step": 18539
    },
    {
      "epoch": 0.8508099674177413,
      "grad_norm": 0.47341710329055786,
      "learning_rate": 8.31859805903707e-06,
      "loss": 0.4069,
      "step": 18540
    },
    {
      "epoch": 0.8508558579229958,
      "grad_norm": 0.5049566030502319,
      "learning_rate": 8.318414659036299e-06,
      "loss": 0.4744,
      "step": 18541
    },
    {
      "epoch": 0.8509017484282502,
      "grad_norm": 0.4703834354877472,
      "learning_rate": 8.318231251055788e-06,
      "loss": 0.3931,
      "step": 18542
    },
    {
      "epoch": 0.8509476389335047,
      "grad_norm": 0.4895935654640198,
      "learning_rate": 8.318047835095977e-06,
      "loss": 0.4637,
      "step": 18543
    },
    {
      "epoch": 0.8509935294387592,
      "grad_norm": 0.5321122407913208,
      "learning_rate": 8.317864411157311e-06,
      "loss": 0.4621,
      "step": 18544
    },
    {
      "epoch": 0.8510394199440136,
      "grad_norm": 0.47079581022262573,
      "learning_rate": 8.317680979240225e-06,
      "loss": 0.3934,
      "step": 18545
    },
    {
      "epoch": 0.851085310449268,
      "grad_norm": 0.44207099080085754,
      "learning_rate": 8.317497539345164e-06,
      "loss": 0.3761,
      "step": 18546
    },
    {
      "epoch": 0.8511312009545225,
      "grad_norm": 0.5199916958808899,
      "learning_rate": 8.317314091472571e-06,
      "loss": 0.4984,
      "step": 18547
    },
    {
      "epoch": 0.8511770914597769,
      "grad_norm": 0.4560734033584595,
      "learning_rate": 8.31713063562288e-06,
      "loss": 0.4275,
      "step": 18548
    },
    {
      "epoch": 0.8512229819650314,
      "grad_norm": 0.46544256806373596,
      "learning_rate": 8.31694717179654e-06,
      "loss": 0.4061,
      "step": 18549
    },
    {
      "epoch": 0.8512688724702859,
      "grad_norm": 0.4387304186820984,
      "learning_rate": 8.316763699993988e-06,
      "loss": 0.3374,
      "step": 18550
    },
    {
      "epoch": 0.8513147629755403,
      "grad_norm": 0.4940645694732666,
      "learning_rate": 8.316580220215667e-06,
      "loss": 0.4477,
      "step": 18551
    },
    {
      "epoch": 0.8513606534807948,
      "grad_norm": 0.4244164228439331,
      "learning_rate": 8.316396732462017e-06,
      "loss": 0.4088,
      "step": 18552
    },
    {
      "epoch": 0.8514065439860493,
      "grad_norm": 0.516417384147644,
      "learning_rate": 8.31621323673348e-06,
      "loss": 0.5139,
      "step": 18553
    },
    {
      "epoch": 0.8514524344913037,
      "grad_norm": 0.4516875743865967,
      "learning_rate": 8.316029733030496e-06,
      "loss": 0.3484,
      "step": 18554
    },
    {
      "epoch": 0.8514983249965582,
      "grad_norm": 0.4425870180130005,
      "learning_rate": 8.315846221353507e-06,
      "loss": 0.4015,
      "step": 18555
    },
    {
      "epoch": 0.8515442155018127,
      "grad_norm": 0.4379623532295227,
      "learning_rate": 8.315662701702953e-06,
      "loss": 0.3755,
      "step": 18556
    },
    {
      "epoch": 0.8515901060070671,
      "grad_norm": 0.45250391960144043,
      "learning_rate": 8.31547917407928e-06,
      "loss": 0.3962,
      "step": 18557
    },
    {
      "epoch": 0.8516359965123216,
      "grad_norm": 0.4658588767051697,
      "learning_rate": 8.315295638482923e-06,
      "loss": 0.4651,
      "step": 18558
    },
    {
      "epoch": 0.8516818870175761,
      "grad_norm": 0.47544774413108826,
      "learning_rate": 8.315112094914328e-06,
      "loss": 0.4591,
      "step": 18559
    },
    {
      "epoch": 0.8517277775228306,
      "grad_norm": 0.42730605602264404,
      "learning_rate": 8.314928543373933e-06,
      "loss": 0.3413,
      "step": 18560
    },
    {
      "epoch": 0.851773668028085,
      "grad_norm": 0.4330829381942749,
      "learning_rate": 8.314744983862184e-06,
      "loss": 0.3714,
      "step": 18561
    },
    {
      "epoch": 0.8518195585333395,
      "grad_norm": 0.4877643287181854,
      "learning_rate": 8.314561416379516e-06,
      "loss": 0.4765,
      "step": 18562
    },
    {
      "epoch": 0.851865449038594,
      "grad_norm": 0.4689941108226776,
      "learning_rate": 8.314377840926375e-06,
      "loss": 0.4251,
      "step": 18563
    },
    {
      "epoch": 0.8519113395438483,
      "grad_norm": 0.44928622245788574,
      "learning_rate": 8.314194257503203e-06,
      "loss": 0.3528,
      "step": 18564
    },
    {
      "epoch": 0.8519572300491028,
      "grad_norm": 0.45337674021720886,
      "learning_rate": 8.31401066611044e-06,
      "loss": 0.4386,
      "step": 18565
    },
    {
      "epoch": 0.8520031205543573,
      "grad_norm": 0.46326684951782227,
      "learning_rate": 8.313827066748526e-06,
      "loss": 0.4128,
      "step": 18566
    },
    {
      "epoch": 0.8520490110596117,
      "grad_norm": 0.43587666749954224,
      "learning_rate": 8.313643459417903e-06,
      "loss": 0.3621,
      "step": 18567
    },
    {
      "epoch": 0.8520949015648662,
      "grad_norm": 0.38895031809806824,
      "learning_rate": 8.313459844119014e-06,
      "loss": 0.2821,
      "step": 18568
    },
    {
      "epoch": 0.8521407920701207,
      "grad_norm": 0.4513740837574005,
      "learning_rate": 8.3132762208523e-06,
      "loss": 0.3668,
      "step": 18569
    },
    {
      "epoch": 0.8521866825753751,
      "grad_norm": 0.4800945222377777,
      "learning_rate": 8.313092589618203e-06,
      "loss": 0.427,
      "step": 18570
    },
    {
      "epoch": 0.8522325730806296,
      "grad_norm": 0.44802138209342957,
      "learning_rate": 8.312908950417162e-06,
      "loss": 0.3725,
      "step": 18571
    },
    {
      "epoch": 0.8522784635858841,
      "grad_norm": 0.48024633526802063,
      "learning_rate": 8.312725303249622e-06,
      "loss": 0.4676,
      "step": 18572
    },
    {
      "epoch": 0.8523243540911385,
      "grad_norm": 0.4557300806045532,
      "learning_rate": 8.312541648116024e-06,
      "loss": 0.4089,
      "step": 18573
    },
    {
      "epoch": 0.852370244596393,
      "grad_norm": 0.44319817423820496,
      "learning_rate": 8.312357985016807e-06,
      "loss": 0.4264,
      "step": 18574
    },
    {
      "epoch": 0.8524161351016475,
      "grad_norm": 0.4716143310070038,
      "learning_rate": 8.312174313952415e-06,
      "loss": 0.4204,
      "step": 18575
    },
    {
      "epoch": 0.852462025606902,
      "grad_norm": 0.4595557153224945,
      "learning_rate": 8.311990634923289e-06,
      "loss": 0.4356,
      "step": 18576
    },
    {
      "epoch": 0.8525079161121564,
      "grad_norm": 0.42150160670280457,
      "learning_rate": 8.311806947929872e-06,
      "loss": 0.3187,
      "step": 18577
    },
    {
      "epoch": 0.8525538066174109,
      "grad_norm": 0.45504552125930786,
      "learning_rate": 8.311623252972603e-06,
      "loss": 0.4048,
      "step": 18578
    },
    {
      "epoch": 0.8525996971226654,
      "grad_norm": 0.42402997612953186,
      "learning_rate": 8.311439550051927e-06,
      "loss": 0.3682,
      "step": 18579
    },
    {
      "epoch": 0.8526455876279198,
      "grad_norm": 0.4750461280345917,
      "learning_rate": 8.311255839168282e-06,
      "loss": 0.4533,
      "step": 18580
    },
    {
      "epoch": 0.8526914781331743,
      "grad_norm": 0.46617522835731506,
      "learning_rate": 8.311072120322114e-06,
      "loss": 0.4073,
      "step": 18581
    },
    {
      "epoch": 0.8527373686384287,
      "grad_norm": 0.4335267245769501,
      "learning_rate": 8.310888393513861e-06,
      "loss": 0.364,
      "step": 18582
    },
    {
      "epoch": 0.8527832591436831,
      "grad_norm": 0.48728227615356445,
      "learning_rate": 8.310704658743966e-06,
      "loss": 0.5208,
      "step": 18583
    },
    {
      "epoch": 0.8528291496489376,
      "grad_norm": 0.47027409076690674,
      "learning_rate": 8.310520916012872e-06,
      "loss": 0.4293,
      "step": 18584
    },
    {
      "epoch": 0.8528750401541921,
      "grad_norm": 0.46774375438690186,
      "learning_rate": 8.310337165321021e-06,
      "loss": 0.4487,
      "step": 18585
    },
    {
      "epoch": 0.8529209306594465,
      "grad_norm": 0.4528515338897705,
      "learning_rate": 8.310153406668853e-06,
      "loss": 0.3858,
      "step": 18586
    },
    {
      "epoch": 0.852966821164701,
      "grad_norm": 0.43130457401275635,
      "learning_rate": 8.30996964005681e-06,
      "loss": 0.3484,
      "step": 18587
    },
    {
      "epoch": 0.8530127116699555,
      "grad_norm": 0.4431041181087494,
      "learning_rate": 8.309785865485335e-06,
      "loss": 0.4093,
      "step": 18588
    },
    {
      "epoch": 0.8530586021752099,
      "grad_norm": 0.4667322635650635,
      "learning_rate": 8.30960208295487e-06,
      "loss": 0.4439,
      "step": 18589
    },
    {
      "epoch": 0.8531044926804644,
      "grad_norm": 0.4563106596469879,
      "learning_rate": 8.309418292465857e-06,
      "loss": 0.4261,
      "step": 18590
    },
    {
      "epoch": 0.8531503831857189,
      "grad_norm": 0.4588871896266937,
      "learning_rate": 8.309234494018735e-06,
      "loss": 0.3695,
      "step": 18591
    },
    {
      "epoch": 0.8531962736909733,
      "grad_norm": 0.4740724563598633,
      "learning_rate": 8.309050687613952e-06,
      "loss": 0.4199,
      "step": 18592
    },
    {
      "epoch": 0.8532421641962278,
      "grad_norm": 0.4747357964515686,
      "learning_rate": 8.308866873251944e-06,
      "loss": 0.43,
      "step": 18593
    },
    {
      "epoch": 0.8532880547014823,
      "grad_norm": 0.43254557251930237,
      "learning_rate": 8.308683050933156e-06,
      "loss": 0.3535,
      "step": 18594
    },
    {
      "epoch": 0.8533339452067368,
      "grad_norm": 0.4472940266132355,
      "learning_rate": 8.308499220658031e-06,
      "loss": 0.4224,
      "step": 18595
    },
    {
      "epoch": 0.8533798357119912,
      "grad_norm": 0.4438001811504364,
      "learning_rate": 8.308315382427006e-06,
      "loss": 0.3622,
      "step": 18596
    },
    {
      "epoch": 0.8534257262172457,
      "grad_norm": 0.47395575046539307,
      "learning_rate": 8.308131536240529e-06,
      "loss": 0.4788,
      "step": 18597
    },
    {
      "epoch": 0.8534716167225002,
      "grad_norm": 0.47048479318618774,
      "learning_rate": 8.30794768209904e-06,
      "loss": 0.4443,
      "step": 18598
    },
    {
      "epoch": 0.8535175072277545,
      "grad_norm": 0.44436895847320557,
      "learning_rate": 8.307763820002981e-06,
      "loss": 0.3924,
      "step": 18599
    },
    {
      "epoch": 0.853563397733009,
      "grad_norm": 0.4608636200428009,
      "learning_rate": 8.307579949952794e-06,
      "loss": 0.376,
      "step": 18600
    },
    {
      "epoch": 0.8536092882382635,
      "grad_norm": 0.5043268203735352,
      "learning_rate": 8.307396071948922e-06,
      "loss": 0.5206,
      "step": 18601
    },
    {
      "epoch": 0.8536551787435179,
      "grad_norm": 0.42557060718536377,
      "learning_rate": 8.307212185991803e-06,
      "loss": 0.3709,
      "step": 18602
    },
    {
      "epoch": 0.8537010692487724,
      "grad_norm": 0.4356067478656769,
      "learning_rate": 8.307028292081885e-06,
      "loss": 0.4129,
      "step": 18603
    },
    {
      "epoch": 0.8537469597540269,
      "grad_norm": 0.45736825466156006,
      "learning_rate": 8.306844390219607e-06,
      "loss": 0.4255,
      "step": 18604
    },
    {
      "epoch": 0.8537928502592813,
      "grad_norm": 0.46578824520111084,
      "learning_rate": 8.306660480405412e-06,
      "loss": 0.4371,
      "step": 18605
    },
    {
      "epoch": 0.8538387407645358,
      "grad_norm": 0.43310442566871643,
      "learning_rate": 8.306476562639743e-06,
      "loss": 0.4351,
      "step": 18606
    },
    {
      "epoch": 0.8538846312697903,
      "grad_norm": 0.46917885541915894,
      "learning_rate": 8.306292636923038e-06,
      "loss": 0.3807,
      "step": 18607
    },
    {
      "epoch": 0.8539305217750447,
      "grad_norm": 0.5084924101829529,
      "learning_rate": 8.306108703255747e-06,
      "loss": 0.4844,
      "step": 18608
    },
    {
      "epoch": 0.8539764122802992,
      "grad_norm": 0.45002466440200806,
      "learning_rate": 8.305924761638306e-06,
      "loss": 0.3673,
      "step": 18609
    },
    {
      "epoch": 0.8540223027855537,
      "grad_norm": 0.4343791604042053,
      "learning_rate": 8.30574081207116e-06,
      "loss": 0.3481,
      "step": 18610
    },
    {
      "epoch": 0.8540681932908082,
      "grad_norm": 0.41949135065078735,
      "learning_rate": 8.30555685455475e-06,
      "loss": 0.3172,
      "step": 18611
    },
    {
      "epoch": 0.8541140837960626,
      "grad_norm": 0.45560339093208313,
      "learning_rate": 8.30537288908952e-06,
      "loss": 0.3854,
      "step": 18612
    },
    {
      "epoch": 0.8541599743013171,
      "grad_norm": 0.44569453597068787,
      "learning_rate": 8.30518891567591e-06,
      "loss": 0.3651,
      "step": 18613
    },
    {
      "epoch": 0.8542058648065716,
      "grad_norm": 0.441384494304657,
      "learning_rate": 8.305004934314365e-06,
      "loss": 0.3679,
      "step": 18614
    },
    {
      "epoch": 0.854251755311826,
      "grad_norm": 0.4357912540435791,
      "learning_rate": 8.304820945005325e-06,
      "loss": 0.3469,
      "step": 18615
    },
    {
      "epoch": 0.8542976458170805,
      "grad_norm": 0.48267582058906555,
      "learning_rate": 8.304636947749234e-06,
      "loss": 0.434,
      "step": 18616
    },
    {
      "epoch": 0.854343536322335,
      "grad_norm": 0.4754229784011841,
      "learning_rate": 8.304452942546534e-06,
      "loss": 0.49,
      "step": 18617
    },
    {
      "epoch": 0.8543894268275893,
      "grad_norm": 0.4574068784713745,
      "learning_rate": 8.304268929397669e-06,
      "loss": 0.3908,
      "step": 18618
    },
    {
      "epoch": 0.8544353173328438,
      "grad_norm": 0.42670682072639465,
      "learning_rate": 8.304084908303079e-06,
      "loss": 0.3148,
      "step": 18619
    },
    {
      "epoch": 0.8544812078380983,
      "grad_norm": 0.44537562131881714,
      "learning_rate": 8.303900879263208e-06,
      "loss": 0.3758,
      "step": 18620
    },
    {
      "epoch": 0.8545270983433527,
      "grad_norm": 0.5011444091796875,
      "learning_rate": 8.303716842278499e-06,
      "loss": 0.508,
      "step": 18621
    },
    {
      "epoch": 0.8545729888486072,
      "grad_norm": 0.4539116621017456,
      "learning_rate": 8.303532797349391e-06,
      "loss": 0.3978,
      "step": 18622
    },
    {
      "epoch": 0.8546188793538617,
      "grad_norm": 0.4529321789741516,
      "learning_rate": 8.303348744476331e-06,
      "loss": 0.3833,
      "step": 18623
    },
    {
      "epoch": 0.8546647698591161,
      "grad_norm": 0.4328615069389343,
      "learning_rate": 8.30316468365976e-06,
      "loss": 0.3186,
      "step": 18624
    },
    {
      "epoch": 0.8547106603643706,
      "grad_norm": 0.4334472715854645,
      "learning_rate": 8.302980614900121e-06,
      "loss": 0.3518,
      "step": 18625
    },
    {
      "epoch": 0.8547565508696251,
      "grad_norm": 0.4645361602306366,
      "learning_rate": 8.302796538197855e-06,
      "loss": 0.4387,
      "step": 18626
    },
    {
      "epoch": 0.8548024413748795,
      "grad_norm": 0.4892256557941437,
      "learning_rate": 8.302612453553404e-06,
      "loss": 0.4383,
      "step": 18627
    },
    {
      "epoch": 0.854848331880134,
      "grad_norm": 0.4317033588886261,
      "learning_rate": 8.302428360967216e-06,
      "loss": 0.377,
      "step": 18628
    },
    {
      "epoch": 0.8548942223853885,
      "grad_norm": 0.4649083912372589,
      "learning_rate": 8.30224426043973e-06,
      "loss": 0.4483,
      "step": 18629
    },
    {
      "epoch": 0.854940112890643,
      "grad_norm": 0.47438356280326843,
      "learning_rate": 8.302060151971386e-06,
      "loss": 0.404,
      "step": 18630
    },
    {
      "epoch": 0.8549860033958974,
      "grad_norm": 0.47471895813941956,
      "learning_rate": 8.301876035562631e-06,
      "loss": 0.464,
      "step": 18631
    },
    {
      "epoch": 0.8550318939011519,
      "grad_norm": 0.4541177749633789,
      "learning_rate": 8.301691911213905e-06,
      "loss": 0.3931,
      "step": 18632
    },
    {
      "epoch": 0.8550777844064064,
      "grad_norm": 0.4797215163707733,
      "learning_rate": 8.301507778925653e-06,
      "loss": 0.4198,
      "step": 18633
    },
    {
      "epoch": 0.8551236749116607,
      "grad_norm": 0.46634167432785034,
      "learning_rate": 8.301323638698317e-06,
      "loss": 0.4109,
      "step": 18634
    },
    {
      "epoch": 0.8551695654169152,
      "grad_norm": 0.5000649094581604,
      "learning_rate": 8.301139490532339e-06,
      "loss": 0.465,
      "step": 18635
    },
    {
      "epoch": 0.8552154559221697,
      "grad_norm": 0.4616800546646118,
      "learning_rate": 8.300955334428163e-06,
      "loss": 0.3838,
      "step": 18636
    },
    {
      "epoch": 0.8552613464274241,
      "grad_norm": 0.5212560892105103,
      "learning_rate": 8.30077117038623e-06,
      "loss": 0.4669,
      "step": 18637
    },
    {
      "epoch": 0.8553072369326786,
      "grad_norm": 0.436909556388855,
      "learning_rate": 8.300586998406986e-06,
      "loss": 0.3449,
      "step": 18638
    },
    {
      "epoch": 0.8553531274379331,
      "grad_norm": 0.4428279995918274,
      "learning_rate": 8.30040281849087e-06,
      "loss": 0.3696,
      "step": 18639
    },
    {
      "epoch": 0.8553990179431875,
      "grad_norm": 0.44313177466392517,
      "learning_rate": 8.300218630638328e-06,
      "loss": 0.3716,
      "step": 18640
    },
    {
      "epoch": 0.855444908448442,
      "grad_norm": 0.4456782341003418,
      "learning_rate": 8.300034434849801e-06,
      "loss": 0.4077,
      "step": 18641
    },
    {
      "epoch": 0.8554907989536965,
      "grad_norm": 0.4661862254142761,
      "learning_rate": 8.299850231125733e-06,
      "loss": 0.4072,
      "step": 18642
    },
    {
      "epoch": 0.8555366894589509,
      "grad_norm": 0.42991846799850464,
      "learning_rate": 8.299666019466569e-06,
      "loss": 0.324,
      "step": 18643
    },
    {
      "epoch": 0.8555825799642054,
      "grad_norm": 0.44299641251564026,
      "learning_rate": 8.299481799872747e-06,
      "loss": 0.4042,
      "step": 18644
    },
    {
      "epoch": 0.8556284704694599,
      "grad_norm": 0.5018248558044434,
      "learning_rate": 8.299297572344711e-06,
      "loss": 0.51,
      "step": 18645
    },
    {
      "epoch": 0.8556743609747143,
      "grad_norm": 0.4239949882030487,
      "learning_rate": 8.29911333688291e-06,
      "loss": 0.3129,
      "step": 18646
    },
    {
      "epoch": 0.8557202514799688,
      "grad_norm": 0.4197768568992615,
      "learning_rate": 8.298929093487779e-06,
      "loss": 0.3262,
      "step": 18647
    },
    {
      "epoch": 0.8557661419852233,
      "grad_norm": 0.4195209741592407,
      "learning_rate": 8.298744842159766e-06,
      "loss": 0.3253,
      "step": 18648
    },
    {
      "epoch": 0.8558120324904778,
      "grad_norm": 0.4123789072036743,
      "learning_rate": 8.298560582899314e-06,
      "loss": 0.32,
      "step": 18649
    },
    {
      "epoch": 0.8558579229957322,
      "grad_norm": 0.4651382267475128,
      "learning_rate": 8.298376315706863e-06,
      "loss": 0.3669,
      "step": 18650
    },
    {
      "epoch": 0.8559038135009867,
      "grad_norm": 0.4356662631034851,
      "learning_rate": 8.298192040582859e-06,
      "loss": 0.369,
      "step": 18651
    },
    {
      "epoch": 0.8559497040062412,
      "grad_norm": 0.45351937413215637,
      "learning_rate": 8.298007757527745e-06,
      "loss": 0.4056,
      "step": 18652
    },
    {
      "epoch": 0.8559955945114955,
      "grad_norm": 0.44646474719047546,
      "learning_rate": 8.297823466541961e-06,
      "loss": 0.3953,
      "step": 18653
    },
    {
      "epoch": 0.85604148501675,
      "grad_norm": 0.45580729842185974,
      "learning_rate": 8.297639167625953e-06,
      "loss": 0.3711,
      "step": 18654
    },
    {
      "epoch": 0.8560873755220045,
      "grad_norm": 0.4668574333190918,
      "learning_rate": 8.297454860780164e-06,
      "loss": 0.3977,
      "step": 18655
    },
    {
      "epoch": 0.8561332660272589,
      "grad_norm": 0.46768882870674133,
      "learning_rate": 8.297270546005036e-06,
      "loss": 0.4483,
      "step": 18656
    },
    {
      "epoch": 0.8561791565325134,
      "grad_norm": 0.42442333698272705,
      "learning_rate": 8.297086223301015e-06,
      "loss": 0.3607,
      "step": 18657
    },
    {
      "epoch": 0.8562250470377679,
      "grad_norm": 0.449093222618103,
      "learning_rate": 8.296901892668541e-06,
      "loss": 0.4262,
      "step": 18658
    },
    {
      "epoch": 0.8562709375430223,
      "grad_norm": 0.46689721941947937,
      "learning_rate": 8.296717554108057e-06,
      "loss": 0.4007,
      "step": 18659
    },
    {
      "epoch": 0.8563168280482768,
      "grad_norm": 0.43982720375061035,
      "learning_rate": 8.29653320762001e-06,
      "loss": 0.4177,
      "step": 18660
    },
    {
      "epoch": 0.8563627185535313,
      "grad_norm": 0.47519272565841675,
      "learning_rate": 8.29634885320484e-06,
      "loss": 0.4488,
      "step": 18661
    },
    {
      "epoch": 0.8564086090587857,
      "grad_norm": 0.5117306113243103,
      "learning_rate": 8.296164490862991e-06,
      "loss": 0.4836,
      "step": 18662
    },
    {
      "epoch": 0.8564544995640402,
      "grad_norm": 0.44239315390586853,
      "learning_rate": 8.295980120594908e-06,
      "loss": 0.3719,
      "step": 18663
    },
    {
      "epoch": 0.8565003900692947,
      "grad_norm": 0.4526567757129669,
      "learning_rate": 8.295795742401031e-06,
      "loss": 0.403,
      "step": 18664
    },
    {
      "epoch": 0.8565462805745492,
      "grad_norm": 0.41223886609077454,
      "learning_rate": 8.295611356281808e-06,
      "loss": 0.3551,
      "step": 18665
    },
    {
      "epoch": 0.8565921710798036,
      "grad_norm": 0.4405532777309418,
      "learning_rate": 8.295426962237678e-06,
      "loss": 0.3365,
      "step": 18666
    },
    {
      "epoch": 0.8566380615850581,
      "grad_norm": 0.4507589042186737,
      "learning_rate": 8.295242560269087e-06,
      "loss": 0.3982,
      "step": 18667
    },
    {
      "epoch": 0.8566839520903126,
      "grad_norm": 0.41458186507225037,
      "learning_rate": 8.295058150376478e-06,
      "loss": 0.3309,
      "step": 18668
    },
    {
      "epoch": 0.856729842595567,
      "grad_norm": 0.5396983623504639,
      "learning_rate": 8.294873732560294e-06,
      "loss": 0.551,
      "step": 18669
    },
    {
      "epoch": 0.8567757331008214,
      "grad_norm": 0.45366430282592773,
      "learning_rate": 8.294689306820979e-06,
      "loss": 0.3921,
      "step": 18670
    },
    {
      "epoch": 0.8568216236060759,
      "grad_norm": 0.43828102946281433,
      "learning_rate": 8.294504873158973e-06,
      "loss": 0.3962,
      "step": 18671
    },
    {
      "epoch": 0.8568675141113303,
      "grad_norm": 0.47282692790031433,
      "learning_rate": 8.294320431574724e-06,
      "loss": 0.4356,
      "step": 18672
    },
    {
      "epoch": 0.8569134046165848,
      "grad_norm": 0.44989287853240967,
      "learning_rate": 8.294135982068676e-06,
      "loss": 0.4196,
      "step": 18673
    },
    {
      "epoch": 0.8569592951218393,
      "grad_norm": 0.48343321681022644,
      "learning_rate": 8.293951524641269e-06,
      "loss": 0.4344,
      "step": 18674
    },
    {
      "epoch": 0.8570051856270937,
      "grad_norm": 0.4631628096103668,
      "learning_rate": 8.29376705929295e-06,
      "loss": 0.3957,
      "step": 18675
    },
    {
      "epoch": 0.8570510761323482,
      "grad_norm": 0.4562375247478485,
      "learning_rate": 8.293582586024158e-06,
      "loss": 0.3857,
      "step": 18676
    },
    {
      "epoch": 0.8570969666376027,
      "grad_norm": 0.45833832025527954,
      "learning_rate": 8.293398104835342e-06,
      "loss": 0.3971,
      "step": 18677
    },
    {
      "epoch": 0.8571428571428571,
      "grad_norm": 0.44750750064849854,
      "learning_rate": 8.293213615726941e-06,
      "loss": 0.374,
      "step": 18678
    },
    {
      "epoch": 0.8571887476481116,
      "grad_norm": 0.4641525149345398,
      "learning_rate": 8.2930291186994e-06,
      "loss": 0.4355,
      "step": 18679
    },
    {
      "epoch": 0.8572346381533661,
      "grad_norm": 0.4336824119091034,
      "learning_rate": 8.292844613753165e-06,
      "loss": 0.3266,
      "step": 18680
    },
    {
      "epoch": 0.8572805286586205,
      "grad_norm": 0.49045807123184204,
      "learning_rate": 8.292660100888678e-06,
      "loss": 0.3933,
      "step": 18681
    },
    {
      "epoch": 0.857326419163875,
      "grad_norm": 0.44010084867477417,
      "learning_rate": 8.292475580106383e-06,
      "loss": 0.3997,
      "step": 18682
    },
    {
      "epoch": 0.8573723096691295,
      "grad_norm": 0.44646966457366943,
      "learning_rate": 8.292291051406722e-06,
      "loss": 0.3818,
      "step": 18683
    },
    {
      "epoch": 0.857418200174384,
      "grad_norm": 0.4187343716621399,
      "learning_rate": 8.292106514790141e-06,
      "loss": 0.3297,
      "step": 18684
    },
    {
      "epoch": 0.8574640906796384,
      "grad_norm": 0.4503140449523926,
      "learning_rate": 8.291921970257081e-06,
      "loss": 0.4082,
      "step": 18685
    },
    {
      "epoch": 0.8575099811848929,
      "grad_norm": 0.43613624572753906,
      "learning_rate": 8.29173741780799e-06,
      "loss": 0.3359,
      "step": 18686
    },
    {
      "epoch": 0.8575558716901474,
      "grad_norm": 0.47834959626197815,
      "learning_rate": 8.291552857443308e-06,
      "loss": 0.394,
      "step": 18687
    },
    {
      "epoch": 0.8576017621954017,
      "grad_norm": 0.49096235632896423,
      "learning_rate": 8.291368289163481e-06,
      "loss": 0.4281,
      "step": 18688
    },
    {
      "epoch": 0.8576476527006562,
      "grad_norm": 0.43833431601524353,
      "learning_rate": 8.291183712968953e-06,
      "loss": 0.3746,
      "step": 18689
    },
    {
      "epoch": 0.8576935432059107,
      "grad_norm": 0.44437000155448914,
      "learning_rate": 8.290999128860165e-06,
      "loss": 0.3657,
      "step": 18690
    },
    {
      "epoch": 0.8577394337111651,
      "grad_norm": 0.46153831481933594,
      "learning_rate": 8.290814536837564e-06,
      "loss": 0.4234,
      "step": 18691
    },
    {
      "epoch": 0.8577853242164196,
      "grad_norm": 0.4904228746891022,
      "learning_rate": 8.290629936901592e-06,
      "loss": 0.4044,
      "step": 18692
    },
    {
      "epoch": 0.8578312147216741,
      "grad_norm": 0.44085782766342163,
      "learning_rate": 8.290445329052694e-06,
      "loss": 0.3581,
      "step": 18693
    },
    {
      "epoch": 0.8578771052269285,
      "grad_norm": 0.5074067115783691,
      "learning_rate": 8.290260713291313e-06,
      "loss": 0.4661,
      "step": 18694
    },
    {
      "epoch": 0.857922995732183,
      "grad_norm": 0.45443975925445557,
      "learning_rate": 8.290076089617893e-06,
      "loss": 0.3883,
      "step": 18695
    },
    {
      "epoch": 0.8579688862374375,
      "grad_norm": 0.44784635305404663,
      "learning_rate": 8.289891458032879e-06,
      "loss": 0.3704,
      "step": 18696
    },
    {
      "epoch": 0.8580147767426919,
      "grad_norm": 0.45705053210258484,
      "learning_rate": 8.289706818536715e-06,
      "loss": 0.4182,
      "step": 18697
    },
    {
      "epoch": 0.8580606672479464,
      "grad_norm": 0.4935908615589142,
      "learning_rate": 8.289522171129843e-06,
      "loss": 0.45,
      "step": 18698
    },
    {
      "epoch": 0.8581065577532009,
      "grad_norm": 0.4599513113498688,
      "learning_rate": 8.28933751581271e-06,
      "loss": 0.4033,
      "step": 18699
    },
    {
      "epoch": 0.8581524482584554,
      "grad_norm": 0.47726890444755554,
      "learning_rate": 8.289152852585759e-06,
      "loss": 0.4131,
      "step": 18700
    },
    {
      "epoch": 0.8581983387637098,
      "grad_norm": 0.4352322518825531,
      "learning_rate": 8.28896818144943e-06,
      "loss": 0.3757,
      "step": 18701
    },
    {
      "epoch": 0.8582442292689643,
      "grad_norm": 0.42752277851104736,
      "learning_rate": 8.288783502404173e-06,
      "loss": 0.3671,
      "step": 18702
    },
    {
      "epoch": 0.8582901197742188,
      "grad_norm": 0.5073032379150391,
      "learning_rate": 8.28859881545043e-06,
      "loss": 0.506,
      "step": 18703
    },
    {
      "epoch": 0.8583360102794731,
      "grad_norm": 0.46708911657333374,
      "learning_rate": 8.288414120588644e-06,
      "loss": 0.4275,
      "step": 18704
    },
    {
      "epoch": 0.8583819007847276,
      "grad_norm": 0.45909813046455383,
      "learning_rate": 8.28822941781926e-06,
      "loss": 0.3875,
      "step": 18705
    },
    {
      "epoch": 0.8584277912899821,
      "grad_norm": 0.41881436109542847,
      "learning_rate": 8.288044707142721e-06,
      "loss": 0.32,
      "step": 18706
    },
    {
      "epoch": 0.8584736817952365,
      "grad_norm": 0.4687194526195526,
      "learning_rate": 8.287859988559473e-06,
      "loss": 0.3865,
      "step": 18707
    },
    {
      "epoch": 0.858519572300491,
      "grad_norm": 0.4324939548969269,
      "learning_rate": 8.28767526206996e-06,
      "loss": 0.4157,
      "step": 18708
    },
    {
      "epoch": 0.8585654628057455,
      "grad_norm": 0.4381049573421478,
      "learning_rate": 8.287490527674624e-06,
      "loss": 0.3792,
      "step": 18709
    },
    {
      "epoch": 0.8586113533109999,
      "grad_norm": 0.47735533118247986,
      "learning_rate": 8.287305785373912e-06,
      "loss": 0.4707,
      "step": 18710
    },
    {
      "epoch": 0.8586572438162544,
      "grad_norm": 0.4391564726829529,
      "learning_rate": 8.287121035168269e-06,
      "loss": 0.4021,
      "step": 18711
    },
    {
      "epoch": 0.8587031343215089,
      "grad_norm": 0.4485732913017273,
      "learning_rate": 8.286936277058135e-06,
      "loss": 0.4079,
      "step": 18712
    },
    {
      "epoch": 0.8587490248267633,
      "grad_norm": 0.5098724365234375,
      "learning_rate": 8.286751511043955e-06,
      "loss": 0.4337,
      "step": 18713
    },
    {
      "epoch": 0.8587949153320178,
      "grad_norm": 0.422309935092926,
      "learning_rate": 8.286566737126179e-06,
      "loss": 0.351,
      "step": 18714
    },
    {
      "epoch": 0.8588408058372723,
      "grad_norm": 0.4879932105541229,
      "learning_rate": 8.286381955305244e-06,
      "loss": 0.4442,
      "step": 18715
    },
    {
      "epoch": 0.8588866963425267,
      "grad_norm": 0.5038551688194275,
      "learning_rate": 8.2861971655816e-06,
      "loss": 0.4103,
      "step": 18716
    },
    {
      "epoch": 0.8589325868477812,
      "grad_norm": 0.47220665216445923,
      "learning_rate": 8.286012367955687e-06,
      "loss": 0.4086,
      "step": 18717
    },
    {
      "epoch": 0.8589784773530357,
      "grad_norm": 0.5163639187812805,
      "learning_rate": 8.285827562427953e-06,
      "loss": 0.5222,
      "step": 18718
    },
    {
      "epoch": 0.8590243678582902,
      "grad_norm": 0.4519052505493164,
      "learning_rate": 8.285642748998839e-06,
      "loss": 0.4086,
      "step": 18719
    },
    {
      "epoch": 0.8590702583635446,
      "grad_norm": 0.47476083040237427,
      "learning_rate": 8.285457927668791e-06,
      "loss": 0.4979,
      "step": 18720
    },
    {
      "epoch": 0.8591161488687991,
      "grad_norm": 0.43274304270744324,
      "learning_rate": 8.285273098438255e-06,
      "loss": 0.3378,
      "step": 18721
    },
    {
      "epoch": 0.8591620393740536,
      "grad_norm": 0.4863831400871277,
      "learning_rate": 8.285088261307674e-06,
      "loss": 0.4995,
      "step": 18722
    },
    {
      "epoch": 0.8592079298793079,
      "grad_norm": 0.3944115936756134,
      "learning_rate": 8.284903416277494e-06,
      "loss": 0.293,
      "step": 18723
    },
    {
      "epoch": 0.8592538203845624,
      "grad_norm": 0.47482049465179443,
      "learning_rate": 8.284718563348155e-06,
      "loss": 0.3915,
      "step": 18724
    },
    {
      "epoch": 0.8592997108898169,
      "grad_norm": 0.4530799984931946,
      "learning_rate": 8.284533702520106e-06,
      "loss": 0.4082,
      "step": 18725
    },
    {
      "epoch": 0.8593456013950713,
      "grad_norm": 0.4449470639228821,
      "learning_rate": 8.284348833793791e-06,
      "loss": 0.3999,
      "step": 18726
    },
    {
      "epoch": 0.8593914919003258,
      "grad_norm": 0.45700088143348694,
      "learning_rate": 8.284163957169651e-06,
      "loss": 0.3598,
      "step": 18727
    },
    {
      "epoch": 0.8594373824055803,
      "grad_norm": 0.43505656719207764,
      "learning_rate": 8.283979072648135e-06,
      "loss": 0.3794,
      "step": 18728
    },
    {
      "epoch": 0.8594832729108347,
      "grad_norm": 0.44833219051361084,
      "learning_rate": 8.283794180229686e-06,
      "loss": 0.374,
      "step": 18729
    },
    {
      "epoch": 0.8595291634160892,
      "grad_norm": 0.4922808110713959,
      "learning_rate": 8.283609279914749e-06,
      "loss": 0.2908,
      "step": 18730
    },
    {
      "epoch": 0.8595750539213437,
      "grad_norm": 0.4684850871562958,
      "learning_rate": 8.283424371703765e-06,
      "loss": 0.4366,
      "step": 18731
    },
    {
      "epoch": 0.8596209444265981,
      "grad_norm": 0.46290862560272217,
      "learning_rate": 8.283239455597185e-06,
      "loss": 0.4576,
      "step": 18732
    },
    {
      "epoch": 0.8596668349318526,
      "grad_norm": 0.4056648910045624,
      "learning_rate": 8.28305453159545e-06,
      "loss": 0.3366,
      "step": 18733
    },
    {
      "epoch": 0.8597127254371071,
      "grad_norm": 0.43588516116142273,
      "learning_rate": 8.282869599699003e-06,
      "loss": 0.3381,
      "step": 18734
    },
    {
      "epoch": 0.8597586159423615,
      "grad_norm": 0.4483650028705597,
      "learning_rate": 8.282684659908291e-06,
      "loss": 0.4338,
      "step": 18735
    },
    {
      "epoch": 0.859804506447616,
      "grad_norm": 0.47305282950401306,
      "learning_rate": 8.282499712223759e-06,
      "loss": 0.3745,
      "step": 18736
    },
    {
      "epoch": 0.8598503969528705,
      "grad_norm": 0.451948881149292,
      "learning_rate": 8.282314756645853e-06,
      "loss": 0.4039,
      "step": 18737
    },
    {
      "epoch": 0.859896287458125,
      "grad_norm": 0.44452202320098877,
      "learning_rate": 8.282129793175014e-06,
      "loss": 0.331,
      "step": 18738
    },
    {
      "epoch": 0.8599421779633794,
      "grad_norm": 0.44473472237586975,
      "learning_rate": 8.281944821811689e-06,
      "loss": 0.3594,
      "step": 18739
    },
    {
      "epoch": 0.8599880684686338,
      "grad_norm": 0.4818864166736603,
      "learning_rate": 8.281759842556321e-06,
      "loss": 0.4491,
      "step": 18740
    },
    {
      "epoch": 0.8600339589738883,
      "grad_norm": 0.4821096956729889,
      "learning_rate": 8.281574855409357e-06,
      "loss": 0.4181,
      "step": 18741
    },
    {
      "epoch": 0.8600798494791427,
      "grad_norm": 0.43771257996559143,
      "learning_rate": 8.281389860371243e-06,
      "loss": 0.3676,
      "step": 18742
    },
    {
      "epoch": 0.8601257399843972,
      "grad_norm": 0.4804643988609314,
      "learning_rate": 8.281204857442422e-06,
      "loss": 0.4131,
      "step": 18743
    },
    {
      "epoch": 0.8601716304896517,
      "grad_norm": 0.48040661215782166,
      "learning_rate": 8.281019846623337e-06,
      "loss": 0.5054,
      "step": 18744
    },
    {
      "epoch": 0.8602175209949061,
      "grad_norm": 0.44240328669548035,
      "learning_rate": 8.280834827914436e-06,
      "loss": 0.3443,
      "step": 18745
    },
    {
      "epoch": 0.8602634115001606,
      "grad_norm": 0.4366568624973297,
      "learning_rate": 8.280649801316162e-06,
      "loss": 0.3426,
      "step": 18746
    },
    {
      "epoch": 0.8603093020054151,
      "grad_norm": 0.41266071796417236,
      "learning_rate": 8.280464766828962e-06,
      "loss": 0.2916,
      "step": 18747
    },
    {
      "epoch": 0.8603551925106695,
      "grad_norm": 0.4484911859035492,
      "learning_rate": 8.280279724453279e-06,
      "loss": 0.4316,
      "step": 18748
    },
    {
      "epoch": 0.860401083015924,
      "grad_norm": 0.445197731256485,
      "learning_rate": 8.280094674189556e-06,
      "loss": 0.3644,
      "step": 18749
    },
    {
      "epoch": 0.8604469735211785,
      "grad_norm": 0.5216633677482605,
      "learning_rate": 8.279909616038245e-06,
      "loss": 0.4693,
      "step": 18750
    },
    {
      "epoch": 0.8604928640264329,
      "grad_norm": 0.485584557056427,
      "learning_rate": 8.279724549999784e-06,
      "loss": 0.4725,
      "step": 18751
    },
    {
      "epoch": 0.8605387545316874,
      "grad_norm": 0.4341680407524109,
      "learning_rate": 8.27953947607462e-06,
      "loss": 0.3859,
      "step": 18752
    },
    {
      "epoch": 0.8605846450369419,
      "grad_norm": 0.4843033254146576,
      "learning_rate": 8.2793543942632e-06,
      "loss": 0.4162,
      "step": 18753
    },
    {
      "epoch": 0.8606305355421964,
      "grad_norm": 0.46687307953834534,
      "learning_rate": 8.279169304565965e-06,
      "loss": 0.4097,
      "step": 18754
    },
    {
      "epoch": 0.8606764260474508,
      "grad_norm": 0.47676655650138855,
      "learning_rate": 8.278984206983366e-06,
      "loss": 0.4575,
      "step": 18755
    },
    {
      "epoch": 0.8607223165527053,
      "grad_norm": 0.44742581248283386,
      "learning_rate": 8.278799101515845e-06,
      "loss": 0.3877,
      "step": 18756
    },
    {
      "epoch": 0.8607682070579598,
      "grad_norm": 0.46821296215057373,
      "learning_rate": 8.278613988163845e-06,
      "loss": 0.4131,
      "step": 18757
    },
    {
      "epoch": 0.8608140975632141,
      "grad_norm": 0.4712432026863098,
      "learning_rate": 8.278428866927813e-06,
      "loss": 0.3864,
      "step": 18758
    },
    {
      "epoch": 0.8608599880684686,
      "grad_norm": 0.4531097114086151,
      "learning_rate": 8.278243737808195e-06,
      "loss": 0.3753,
      "step": 18759
    },
    {
      "epoch": 0.8609058785737231,
      "grad_norm": 0.46649661660194397,
      "learning_rate": 8.278058600805436e-06,
      "loss": 0.3167,
      "step": 18760
    },
    {
      "epoch": 0.8609517690789775,
      "grad_norm": 0.4456283152103424,
      "learning_rate": 8.27787345591998e-06,
      "loss": 0.3792,
      "step": 18761
    },
    {
      "epoch": 0.860997659584232,
      "grad_norm": 0.44161224365234375,
      "learning_rate": 8.277688303152274e-06,
      "loss": 0.4132,
      "step": 18762
    },
    {
      "epoch": 0.8610435500894865,
      "grad_norm": 0.42752066254615784,
      "learning_rate": 8.277503142502761e-06,
      "loss": 0.3963,
      "step": 18763
    },
    {
      "epoch": 0.8610894405947409,
      "grad_norm": 0.428093284368515,
      "learning_rate": 8.277317973971888e-06,
      "loss": 0.3516,
      "step": 18764
    },
    {
      "epoch": 0.8611353310999954,
      "grad_norm": 0.4360617399215698,
      "learning_rate": 8.277132797560099e-06,
      "loss": 0.4017,
      "step": 18765
    },
    {
      "epoch": 0.8611812216052499,
      "grad_norm": 0.4670145809650421,
      "learning_rate": 8.276947613267838e-06,
      "loss": 0.4073,
      "step": 18766
    },
    {
      "epoch": 0.8612271121105043,
      "grad_norm": 0.4451378285884857,
      "learning_rate": 8.276762421095556e-06,
      "loss": 0.4104,
      "step": 18767
    },
    {
      "epoch": 0.8612730026157588,
      "grad_norm": 0.4484672546386719,
      "learning_rate": 8.276577221043692e-06,
      "loss": 0.3867,
      "step": 18768
    },
    {
      "epoch": 0.8613188931210133,
      "grad_norm": 0.47523221373558044,
      "learning_rate": 8.276392013112696e-06,
      "loss": 0.4791,
      "step": 18769
    },
    {
      "epoch": 0.8613647836262677,
      "grad_norm": 0.4498392641544342,
      "learning_rate": 8.27620679730301e-06,
      "loss": 0.3865,
      "step": 18770
    },
    {
      "epoch": 0.8614106741315222,
      "grad_norm": 0.4520481526851654,
      "learning_rate": 8.27602157361508e-06,
      "loss": 0.4076,
      "step": 18771
    },
    {
      "epoch": 0.8614565646367767,
      "grad_norm": 0.49443531036376953,
      "learning_rate": 8.275836342049353e-06,
      "loss": 0.3854,
      "step": 18772
    },
    {
      "epoch": 0.8615024551420312,
      "grad_norm": 0.4288277328014374,
      "learning_rate": 8.275651102606273e-06,
      "loss": 0.3735,
      "step": 18773
    },
    {
      "epoch": 0.8615483456472856,
      "grad_norm": 0.41894081234931946,
      "learning_rate": 8.275465855286285e-06,
      "loss": 0.3472,
      "step": 18774
    },
    {
      "epoch": 0.86159423615254,
      "grad_norm": 0.42563942074775696,
      "learning_rate": 8.275280600089836e-06,
      "loss": 0.3352,
      "step": 18775
    },
    {
      "epoch": 0.8616401266577945,
      "grad_norm": 0.44500932097435,
      "learning_rate": 8.275095337017374e-06,
      "loss": 0.3777,
      "step": 18776
    },
    {
      "epoch": 0.8616860171630489,
      "grad_norm": 0.43110716342926025,
      "learning_rate": 8.274910066069337e-06,
      "loss": 0.398,
      "step": 18777
    },
    {
      "epoch": 0.8617319076683034,
      "grad_norm": 0.5619285106658936,
      "learning_rate": 8.274724787246177e-06,
      "loss": 0.369,
      "step": 18778
    },
    {
      "epoch": 0.8617777981735579,
      "grad_norm": 0.4262608587741852,
      "learning_rate": 8.274539500548337e-06,
      "loss": 0.3207,
      "step": 18779
    },
    {
      "epoch": 0.8618236886788123,
      "grad_norm": 0.5058470964431763,
      "learning_rate": 8.274354205976262e-06,
      "loss": 0.5169,
      "step": 18780
    },
    {
      "epoch": 0.8618695791840668,
      "grad_norm": 0.5004580616950989,
      "learning_rate": 8.274168903530403e-06,
      "loss": 0.465,
      "step": 18781
    },
    {
      "epoch": 0.8619154696893213,
      "grad_norm": 0.4529927968978882,
      "learning_rate": 8.273983593211194e-06,
      "loss": 0.3909,
      "step": 18782
    },
    {
      "epoch": 0.8619613601945757,
      "grad_norm": 0.45167145133018494,
      "learning_rate": 8.273798275019092e-06,
      "loss": 0.3974,
      "step": 18783
    },
    {
      "epoch": 0.8620072506998302,
      "grad_norm": 0.44238150119781494,
      "learning_rate": 8.273612948954539e-06,
      "loss": 0.3991,
      "step": 18784
    },
    {
      "epoch": 0.8620531412050847,
      "grad_norm": 0.4169408977031708,
      "learning_rate": 8.273427615017978e-06,
      "loss": 0.318,
      "step": 18785
    },
    {
      "epoch": 0.8620990317103391,
      "grad_norm": 0.42822885513305664,
      "learning_rate": 8.273242273209858e-06,
      "loss": 0.3789,
      "step": 18786
    },
    {
      "epoch": 0.8621449222155936,
      "grad_norm": 0.42024970054626465,
      "learning_rate": 8.273056923530624e-06,
      "loss": 0.3529,
      "step": 18787
    },
    {
      "epoch": 0.8621908127208481,
      "grad_norm": 0.4831533432006836,
      "learning_rate": 8.27287156598072e-06,
      "loss": 0.3879,
      "step": 18788
    },
    {
      "epoch": 0.8622367032261026,
      "grad_norm": 0.44380542635917664,
      "learning_rate": 8.272686200560591e-06,
      "loss": 0.3583,
      "step": 18789
    },
    {
      "epoch": 0.862282593731357,
      "grad_norm": 0.4034491181373596,
      "learning_rate": 8.272500827270687e-06,
      "loss": 0.3084,
      "step": 18790
    },
    {
      "epoch": 0.8623284842366115,
      "grad_norm": 0.4250091314315796,
      "learning_rate": 8.272315446111452e-06,
      "loss": 0.3122,
      "step": 18791
    },
    {
      "epoch": 0.862374374741866,
      "grad_norm": 0.4368078410625458,
      "learning_rate": 8.27213005708333e-06,
      "loss": 0.3774,
      "step": 18792
    },
    {
      "epoch": 0.8624202652471203,
      "grad_norm": 0.489898145198822,
      "learning_rate": 8.271944660186768e-06,
      "loss": 0.5368,
      "step": 18793
    },
    {
      "epoch": 0.8624661557523748,
      "grad_norm": 0.50797438621521,
      "learning_rate": 8.271759255422211e-06,
      "loss": 0.5204,
      "step": 18794
    },
    {
      "epoch": 0.8625120462576293,
      "grad_norm": 0.4738045632839203,
      "learning_rate": 8.271573842790107e-06,
      "loss": 0.3832,
      "step": 18795
    },
    {
      "epoch": 0.8625579367628837,
      "grad_norm": 0.4486936330795288,
      "learning_rate": 8.271388422290897e-06,
      "loss": 0.3959,
      "step": 18796
    },
    {
      "epoch": 0.8626038272681382,
      "grad_norm": 0.4754834771156311,
      "learning_rate": 8.271202993925035e-06,
      "loss": 0.3931,
      "step": 18797
    },
    {
      "epoch": 0.8626497177733927,
      "grad_norm": 0.4636716842651367,
      "learning_rate": 8.27101755769296e-06,
      "loss": 0.4171,
      "step": 18798
    },
    {
      "epoch": 0.8626956082786471,
      "grad_norm": 0.42925846576690674,
      "learning_rate": 8.270832113595119e-06,
      "loss": 0.3415,
      "step": 18799
    },
    {
      "epoch": 0.8627414987839016,
      "grad_norm": 0.47872334718704224,
      "learning_rate": 8.270646661631961e-06,
      "loss": 0.4016,
      "step": 18800
    },
    {
      "epoch": 0.8627873892891561,
      "grad_norm": 0.5090519785881042,
      "learning_rate": 8.27046120180393e-06,
      "loss": 0.4043,
      "step": 18801
    },
    {
      "epoch": 0.8628332797944105,
      "grad_norm": 0.45238083600997925,
      "learning_rate": 8.270275734111473e-06,
      "loss": 0.3654,
      "step": 18802
    },
    {
      "epoch": 0.862879170299665,
      "grad_norm": 0.43062824010849,
      "learning_rate": 8.270090258555032e-06,
      "loss": 0.3457,
      "step": 18803
    },
    {
      "epoch": 0.8629250608049195,
      "grad_norm": 0.4600858688354492,
      "learning_rate": 8.269904775135059e-06,
      "loss": 0.4174,
      "step": 18804
    },
    {
      "epoch": 0.8629709513101739,
      "grad_norm": 0.42829200625419617,
      "learning_rate": 8.269719283851993e-06,
      "loss": 0.383,
      "step": 18805
    },
    {
      "epoch": 0.8630168418154284,
      "grad_norm": 0.4466390907764435,
      "learning_rate": 8.269533784706287e-06,
      "loss": 0.3521,
      "step": 18806
    },
    {
      "epoch": 0.8630627323206829,
      "grad_norm": 0.4547543227672577,
      "learning_rate": 8.269348277698383e-06,
      "loss": 0.3734,
      "step": 18807
    },
    {
      "epoch": 0.8631086228259374,
      "grad_norm": 0.4274538457393646,
      "learning_rate": 8.269162762828731e-06,
      "loss": 0.3557,
      "step": 18808
    },
    {
      "epoch": 0.8631545133311918,
      "grad_norm": 0.46227550506591797,
      "learning_rate": 8.268977240097771e-06,
      "loss": 0.385,
      "step": 18809
    },
    {
      "epoch": 0.8632004038364463,
      "grad_norm": 0.4527345895767212,
      "learning_rate": 8.268791709505954e-06,
      "loss": 0.3543,
      "step": 18810
    },
    {
      "epoch": 0.8632462943417007,
      "grad_norm": 0.4236842095851898,
      "learning_rate": 8.268606171053723e-06,
      "loss": 0.3491,
      "step": 18811
    },
    {
      "epoch": 0.8632921848469551,
      "grad_norm": 0.4758046567440033,
      "learning_rate": 8.268420624741529e-06,
      "loss": 0.432,
      "step": 18812
    },
    {
      "epoch": 0.8633380753522096,
      "grad_norm": 0.4896450340747833,
      "learning_rate": 8.268235070569812e-06,
      "loss": 0.4555,
      "step": 18813
    },
    {
      "epoch": 0.8633839658574641,
      "grad_norm": 0.5041840672492981,
      "learning_rate": 8.268049508539022e-06,
      "loss": 0.5418,
      "step": 18814
    },
    {
      "epoch": 0.8634298563627185,
      "grad_norm": 0.46307623386383057,
      "learning_rate": 8.267863938649604e-06,
      "loss": 0.4701,
      "step": 18815
    },
    {
      "epoch": 0.863475746867973,
      "grad_norm": 0.46478649973869324,
      "learning_rate": 8.267678360902006e-06,
      "loss": 0.4203,
      "step": 18816
    },
    {
      "epoch": 0.8635216373732275,
      "grad_norm": 0.4625875651836395,
      "learning_rate": 8.267492775296671e-06,
      "loss": 0.36,
      "step": 18817
    },
    {
      "epoch": 0.8635675278784819,
      "grad_norm": 0.4322952330112457,
      "learning_rate": 8.26730718183405e-06,
      "loss": 0.3775,
      "step": 18818
    },
    {
      "epoch": 0.8636134183837364,
      "grad_norm": 0.46911996603012085,
      "learning_rate": 8.267121580514583e-06,
      "loss": 0.3921,
      "step": 18819
    },
    {
      "epoch": 0.8636593088889909,
      "grad_norm": 0.43232011795043945,
      "learning_rate": 8.266935971338721e-06,
      "loss": 0.3507,
      "step": 18820
    },
    {
      "epoch": 0.8637051993942453,
      "grad_norm": 0.4690300524234772,
      "learning_rate": 8.266750354306909e-06,
      "loss": 0.4397,
      "step": 18821
    },
    {
      "epoch": 0.8637510898994998,
      "grad_norm": 0.47257089614868164,
      "learning_rate": 8.266564729419595e-06,
      "loss": 0.4168,
      "step": 18822
    },
    {
      "epoch": 0.8637969804047543,
      "grad_norm": 0.4271184802055359,
      "learning_rate": 8.266379096677222e-06,
      "loss": 0.3878,
      "step": 18823
    },
    {
      "epoch": 0.8638428709100087,
      "grad_norm": 0.498144268989563,
      "learning_rate": 8.26619345608024e-06,
      "loss": 0.5204,
      "step": 18824
    },
    {
      "epoch": 0.8638887614152632,
      "grad_norm": 0.4803226888179779,
      "learning_rate": 8.26600780762909e-06,
      "loss": 0.396,
      "step": 18825
    },
    {
      "epoch": 0.8639346519205177,
      "grad_norm": 0.47004008293151855,
      "learning_rate": 8.265822151324225e-06,
      "loss": 0.445,
      "step": 18826
    },
    {
      "epoch": 0.8639805424257722,
      "grad_norm": 0.5159043073654175,
      "learning_rate": 8.26563648716609e-06,
      "loss": 0.4985,
      "step": 18827
    },
    {
      "epoch": 0.8640264329310265,
      "grad_norm": 0.4207015037536621,
      "learning_rate": 8.265450815155125e-06,
      "loss": 0.3383,
      "step": 18828
    },
    {
      "epoch": 0.864072323436281,
      "grad_norm": 0.4481675624847412,
      "learning_rate": 8.265265135291785e-06,
      "loss": 0.4124,
      "step": 18829
    },
    {
      "epoch": 0.8641182139415355,
      "grad_norm": 0.4736972153186798,
      "learning_rate": 8.265079447576511e-06,
      "loss": 0.503,
      "step": 18830
    },
    {
      "epoch": 0.8641641044467899,
      "grad_norm": 0.4788176715373993,
      "learning_rate": 8.264893752009752e-06,
      "loss": 0.4876,
      "step": 18831
    },
    {
      "epoch": 0.8642099949520444,
      "grad_norm": 0.45518481731414795,
      "learning_rate": 8.264708048591955e-06,
      "loss": 0.3471,
      "step": 18832
    },
    {
      "epoch": 0.8642558854572989,
      "grad_norm": 0.46216586232185364,
      "learning_rate": 8.264522337323565e-06,
      "loss": 0.3826,
      "step": 18833
    },
    {
      "epoch": 0.8643017759625533,
      "grad_norm": 0.45073795318603516,
      "learning_rate": 8.26433661820503e-06,
      "loss": 0.407,
      "step": 18834
    },
    {
      "epoch": 0.8643476664678078,
      "grad_norm": 0.4431460499763489,
      "learning_rate": 8.264150891236794e-06,
      "loss": 0.3706,
      "step": 18835
    },
    {
      "epoch": 0.8643935569730623,
      "grad_norm": 0.45016369223594666,
      "learning_rate": 8.263965156419304e-06,
      "loss": 0.3712,
      "step": 18836
    },
    {
      "epoch": 0.8644394474783167,
      "grad_norm": 0.4362359344959259,
      "learning_rate": 8.26377941375301e-06,
      "loss": 0.3516,
      "step": 18837
    },
    {
      "epoch": 0.8644853379835712,
      "grad_norm": 0.47872158885002136,
      "learning_rate": 8.263593663238359e-06,
      "loss": 0.4214,
      "step": 18838
    },
    {
      "epoch": 0.8645312284888257,
      "grad_norm": 0.4413442611694336,
      "learning_rate": 8.26340790487579e-06,
      "loss": 0.3548,
      "step": 18839
    },
    {
      "epoch": 0.8645771189940801,
      "grad_norm": 0.4541882574558258,
      "learning_rate": 8.263222138665759e-06,
      "loss": 0.3644,
      "step": 18840
    },
    {
      "epoch": 0.8646230094993346,
      "grad_norm": 0.4624963104724884,
      "learning_rate": 8.263036364608707e-06,
      "loss": 0.4304,
      "step": 18841
    },
    {
      "epoch": 0.8646689000045891,
      "grad_norm": 0.4434128403663635,
      "learning_rate": 8.262850582705083e-06,
      "loss": 0.4376,
      "step": 18842
    },
    {
      "epoch": 0.8647147905098436,
      "grad_norm": 0.48018375039100647,
      "learning_rate": 8.262664792955333e-06,
      "loss": 0.4827,
      "step": 18843
    },
    {
      "epoch": 0.864760681015098,
      "grad_norm": 0.44617676734924316,
      "learning_rate": 8.262478995359905e-06,
      "loss": 0.3349,
      "step": 18844
    },
    {
      "epoch": 0.8648065715203525,
      "grad_norm": 0.4706483781337738,
      "learning_rate": 8.262293189919242e-06,
      "loss": 0.441,
      "step": 18845
    },
    {
      "epoch": 0.864852462025607,
      "grad_norm": 0.44068026542663574,
      "learning_rate": 8.262107376633794e-06,
      "loss": 0.332,
      "step": 18846
    },
    {
      "epoch": 0.8648983525308613,
      "grad_norm": 0.46326667070388794,
      "learning_rate": 8.26192155550401e-06,
      "loss": 0.3601,
      "step": 18847
    },
    {
      "epoch": 0.8649442430361158,
      "grad_norm": 0.48375481367111206,
      "learning_rate": 8.261735726530332e-06,
      "loss": 0.4311,
      "step": 18848
    },
    {
      "epoch": 0.8649901335413703,
      "grad_norm": 0.4684853255748749,
      "learning_rate": 8.261549889713209e-06,
      "loss": 0.397,
      "step": 18849
    },
    {
      "epoch": 0.8650360240466247,
      "grad_norm": 0.42297425866127014,
      "learning_rate": 8.261364045053088e-06,
      "loss": 0.3174,
      "step": 18850
    },
    {
      "epoch": 0.8650819145518792,
      "grad_norm": 0.4757521450519562,
      "learning_rate": 8.261178192550416e-06,
      "loss": 0.4633,
      "step": 18851
    },
    {
      "epoch": 0.8651278050571337,
      "grad_norm": 0.46760454773902893,
      "learning_rate": 8.26099233220564e-06,
      "loss": 0.4217,
      "step": 18852
    },
    {
      "epoch": 0.8651736955623881,
      "grad_norm": 0.45929697155952454,
      "learning_rate": 8.260806464019206e-06,
      "loss": 0.3428,
      "step": 18853
    },
    {
      "epoch": 0.8652195860676426,
      "grad_norm": 0.430281400680542,
      "learning_rate": 8.260620587991561e-06,
      "loss": 0.315,
      "step": 18854
    },
    {
      "epoch": 0.8652654765728971,
      "grad_norm": 0.4463421702384949,
      "learning_rate": 8.260434704123155e-06,
      "loss": 0.4245,
      "step": 18855
    },
    {
      "epoch": 0.8653113670781515,
      "grad_norm": 0.47195881605148315,
      "learning_rate": 8.26024881241443e-06,
      "loss": 0.3876,
      "step": 18856
    },
    {
      "epoch": 0.865357257583406,
      "grad_norm": 0.4438212215900421,
      "learning_rate": 8.260062912865835e-06,
      "loss": 0.3988,
      "step": 18857
    },
    {
      "epoch": 0.8654031480886605,
      "grad_norm": 0.4748304784297943,
      "learning_rate": 8.25987700547782e-06,
      "loss": 0.3946,
      "step": 18858
    },
    {
      "epoch": 0.8654490385939149,
      "grad_norm": 0.43733811378479004,
      "learning_rate": 8.25969109025083e-06,
      "loss": 0.3932,
      "step": 18859
    },
    {
      "epoch": 0.8654949290991694,
      "grad_norm": 0.45298895239830017,
      "learning_rate": 8.25950516718531e-06,
      "loss": 0.4099,
      "step": 18860
    },
    {
      "epoch": 0.8655408196044239,
      "grad_norm": 0.4728960692882538,
      "learning_rate": 8.259319236281709e-06,
      "loss": 0.4913,
      "step": 18861
    },
    {
      "epoch": 0.8655867101096784,
      "grad_norm": 0.4645313620567322,
      "learning_rate": 8.259133297540474e-06,
      "loss": 0.378,
      "step": 18862
    },
    {
      "epoch": 0.8656326006149327,
      "grad_norm": 0.4668658375740051,
      "learning_rate": 8.258947350962054e-06,
      "loss": 0.3918,
      "step": 18863
    },
    {
      "epoch": 0.8656784911201872,
      "grad_norm": 0.4860525131225586,
      "learning_rate": 8.258761396546892e-06,
      "loss": 0.4637,
      "step": 18864
    },
    {
      "epoch": 0.8657243816254417,
      "grad_norm": 0.4890578091144562,
      "learning_rate": 8.258575434295438e-06,
      "loss": 0.51,
      "step": 18865
    },
    {
      "epoch": 0.8657702721306961,
      "grad_norm": 0.45143988728523254,
      "learning_rate": 8.258389464208138e-06,
      "loss": 0.4269,
      "step": 18866
    },
    {
      "epoch": 0.8658161626359506,
      "grad_norm": 0.414246141910553,
      "learning_rate": 8.258203486285442e-06,
      "loss": 0.3436,
      "step": 18867
    },
    {
      "epoch": 0.8658620531412051,
      "grad_norm": 0.47159600257873535,
      "learning_rate": 8.25801750052779e-06,
      "loss": 0.4173,
      "step": 18868
    },
    {
      "epoch": 0.8659079436464595,
      "grad_norm": 0.44948068261146545,
      "learning_rate": 8.257831506935638e-06,
      "loss": 0.4243,
      "step": 18869
    },
    {
      "epoch": 0.865953834151714,
      "grad_norm": 0.4541017413139343,
      "learning_rate": 8.257645505509429e-06,
      "loss": 0.3663,
      "step": 18870
    },
    {
      "epoch": 0.8659997246569685,
      "grad_norm": 0.39956530928611755,
      "learning_rate": 8.257459496249608e-06,
      "loss": 0.3021,
      "step": 18871
    },
    {
      "epoch": 0.8660456151622229,
      "grad_norm": 0.6714522838592529,
      "learning_rate": 8.257273479156629e-06,
      "loss": 0.396,
      "step": 18872
    },
    {
      "epoch": 0.8660915056674774,
      "grad_norm": 0.47061097621917725,
      "learning_rate": 8.257087454230934e-06,
      "loss": 0.4329,
      "step": 18873
    },
    {
      "epoch": 0.8661373961727319,
      "grad_norm": 0.47941428422927856,
      "learning_rate": 8.256901421472971e-06,
      "loss": 0.3428,
      "step": 18874
    },
    {
      "epoch": 0.8661832866779863,
      "grad_norm": 0.46274253726005554,
      "learning_rate": 8.256715380883189e-06,
      "loss": 0.3713,
      "step": 18875
    },
    {
      "epoch": 0.8662291771832408,
      "grad_norm": 0.4951189160346985,
      "learning_rate": 8.256529332462033e-06,
      "loss": 0.4903,
      "step": 18876
    },
    {
      "epoch": 0.8662750676884953,
      "grad_norm": 0.45886680483818054,
      "learning_rate": 8.256343276209953e-06,
      "loss": 0.3921,
      "step": 18877
    },
    {
      "epoch": 0.8663209581937498,
      "grad_norm": 0.49009016156196594,
      "learning_rate": 8.256157212127394e-06,
      "loss": 0.386,
      "step": 18878
    },
    {
      "epoch": 0.8663668486990042,
      "grad_norm": 0.46046024560928345,
      "learning_rate": 8.255971140214804e-06,
      "loss": 0.412,
      "step": 18879
    },
    {
      "epoch": 0.8664127392042587,
      "grad_norm": 0.4428119659423828,
      "learning_rate": 8.255785060472634e-06,
      "loss": 0.3836,
      "step": 18880
    },
    {
      "epoch": 0.8664586297095132,
      "grad_norm": 0.4646400511264801,
      "learning_rate": 8.255598972901327e-06,
      "loss": 0.423,
      "step": 18881
    },
    {
      "epoch": 0.8665045202147675,
      "grad_norm": 0.4498136341571808,
      "learning_rate": 8.25541287750133e-06,
      "loss": 0.3831,
      "step": 18882
    },
    {
      "epoch": 0.866550410720022,
      "grad_norm": 0.5011245012283325,
      "learning_rate": 8.255226774273095e-06,
      "loss": 0.4586,
      "step": 18883
    },
    {
      "epoch": 0.8665963012252765,
      "grad_norm": 0.45868098735809326,
      "learning_rate": 8.255040663217066e-06,
      "loss": 0.4041,
      "step": 18884
    },
    {
      "epoch": 0.8666421917305309,
      "grad_norm": 0.44030332565307617,
      "learning_rate": 8.254854544333691e-06,
      "loss": 0.3901,
      "step": 18885
    },
    {
      "epoch": 0.8666880822357854,
      "grad_norm": 0.4524097740650177,
      "learning_rate": 8.25466841762342e-06,
      "loss": 0.4439,
      "step": 18886
    },
    {
      "epoch": 0.8667339727410399,
      "grad_norm": 0.4806548058986664,
      "learning_rate": 8.254482283086696e-06,
      "loss": 0.4949,
      "step": 18887
    },
    {
      "epoch": 0.8667798632462943,
      "grad_norm": 0.42640072107315063,
      "learning_rate": 8.254296140723969e-06,
      "loss": 0.3111,
      "step": 18888
    },
    {
      "epoch": 0.8668257537515488,
      "grad_norm": 0.48186102509498596,
      "learning_rate": 8.254109990535691e-06,
      "loss": 0.4387,
      "step": 18889
    },
    {
      "epoch": 0.8668716442568033,
      "grad_norm": 0.4597827196121216,
      "learning_rate": 8.2539238325223e-06,
      "loss": 0.3959,
      "step": 18890
    },
    {
      "epoch": 0.8669175347620577,
      "grad_norm": 0.45699310302734375,
      "learning_rate": 8.253737666684254e-06,
      "loss": 0.3857,
      "step": 18891
    },
    {
      "epoch": 0.8669634252673122,
      "grad_norm": 0.5193095207214355,
      "learning_rate": 8.253551493021994e-06,
      "loss": 0.5535,
      "step": 18892
    },
    {
      "epoch": 0.8670093157725667,
      "grad_norm": 0.4379752576351166,
      "learning_rate": 8.253365311535967e-06,
      "loss": 0.3737,
      "step": 18893
    },
    {
      "epoch": 0.8670552062778211,
      "grad_norm": 0.4666605293750763,
      "learning_rate": 8.253179122226627e-06,
      "loss": 0.4324,
      "step": 18894
    },
    {
      "epoch": 0.8671010967830756,
      "grad_norm": 0.4511547088623047,
      "learning_rate": 8.252992925094418e-06,
      "loss": 0.3947,
      "step": 18895
    },
    {
      "epoch": 0.8671469872883301,
      "grad_norm": 0.44838061928749084,
      "learning_rate": 8.252806720139785e-06,
      "loss": 0.4125,
      "step": 18896
    },
    {
      "epoch": 0.8671928777935846,
      "grad_norm": 0.48228031396865845,
      "learning_rate": 8.25262050736318e-06,
      "loss": 0.4575,
      "step": 18897
    },
    {
      "epoch": 0.867238768298839,
      "grad_norm": 0.4544408917427063,
      "learning_rate": 8.252434286765048e-06,
      "loss": 0.4058,
      "step": 18898
    },
    {
      "epoch": 0.8672846588040934,
      "grad_norm": 0.43944641947746277,
      "learning_rate": 8.25224805834584e-06,
      "loss": 0.3845,
      "step": 18899
    },
    {
      "epoch": 0.8673305493093479,
      "grad_norm": 0.4750918447971344,
      "learning_rate": 8.252061822106001e-06,
      "loss": 0.4626,
      "step": 18900
    },
    {
      "epoch": 0.8673764398146023,
      "grad_norm": 0.48726651072502136,
      "learning_rate": 8.25187557804598e-06,
      "loss": 0.4475,
      "step": 18901
    },
    {
      "epoch": 0.8674223303198568,
      "grad_norm": 0.4713202118873596,
      "learning_rate": 8.251689326166224e-06,
      "loss": 0.514,
      "step": 18902
    },
    {
      "epoch": 0.8674682208251113,
      "grad_norm": 0.6841461658477783,
      "learning_rate": 8.25150306646718e-06,
      "loss": 0.393,
      "step": 18903
    },
    {
      "epoch": 0.8675141113303657,
      "grad_norm": 0.4626181125640869,
      "learning_rate": 8.2513167989493e-06,
      "loss": 0.3756,
      "step": 18904
    },
    {
      "epoch": 0.8675600018356202,
      "grad_norm": 0.46891921758651733,
      "learning_rate": 8.251130523613027e-06,
      "loss": 0.4165,
      "step": 18905
    },
    {
      "epoch": 0.8676058923408747,
      "grad_norm": 0.415382981300354,
      "learning_rate": 8.250944240458813e-06,
      "loss": 0.305,
      "step": 18906
    },
    {
      "epoch": 0.8676517828461291,
      "grad_norm": 0.4734896719455719,
      "learning_rate": 8.250757949487103e-06,
      "loss": 0.4161,
      "step": 18907
    },
    {
      "epoch": 0.8676976733513836,
      "grad_norm": 0.4632915258407593,
      "learning_rate": 8.250571650698346e-06,
      "loss": 0.3782,
      "step": 18908
    },
    {
      "epoch": 0.8677435638566381,
      "grad_norm": 0.4909222722053528,
      "learning_rate": 8.250385344092992e-06,
      "loss": 0.4448,
      "step": 18909
    },
    {
      "epoch": 0.8677894543618925,
      "grad_norm": 0.4747442603111267,
      "learning_rate": 8.250199029671485e-06,
      "loss": 0.4056,
      "step": 18910
    },
    {
      "epoch": 0.867835344867147,
      "grad_norm": 0.5308018326759338,
      "learning_rate": 8.250012707434276e-06,
      "loss": 0.5501,
      "step": 18911
    },
    {
      "epoch": 0.8678812353724015,
      "grad_norm": 0.44531968235969543,
      "learning_rate": 8.24982637738181e-06,
      "loss": 0.382,
      "step": 18912
    },
    {
      "epoch": 0.8679271258776559,
      "grad_norm": 0.4495735168457031,
      "learning_rate": 8.249640039514542e-06,
      "loss": 0.3871,
      "step": 18913
    },
    {
      "epoch": 0.8679730163829104,
      "grad_norm": 0.4664168059825897,
      "learning_rate": 8.249453693832911e-06,
      "loss": 0.3973,
      "step": 18914
    },
    {
      "epoch": 0.8680189068881649,
      "grad_norm": 0.4821634590625763,
      "learning_rate": 8.249267340337371e-06,
      "loss": 0.4222,
      "step": 18915
    },
    {
      "epoch": 0.8680647973934194,
      "grad_norm": 0.4313802123069763,
      "learning_rate": 8.249080979028369e-06,
      "loss": 0.3544,
      "step": 18916
    },
    {
      "epoch": 0.8681106878986737,
      "grad_norm": 0.44742894172668457,
      "learning_rate": 8.248894609906353e-06,
      "loss": 0.3964,
      "step": 18917
    },
    {
      "epoch": 0.8681565784039282,
      "grad_norm": 0.5019928812980652,
      "learning_rate": 8.24870823297177e-06,
      "loss": 0.473,
      "step": 18918
    },
    {
      "epoch": 0.8682024689091827,
      "grad_norm": 0.4228384792804718,
      "learning_rate": 8.248521848225068e-06,
      "loss": 0.3336,
      "step": 18919
    },
    {
      "epoch": 0.8682483594144371,
      "grad_norm": 0.4851352572441101,
      "learning_rate": 8.248335455666698e-06,
      "loss": 0.4783,
      "step": 18920
    },
    {
      "epoch": 0.8682942499196916,
      "grad_norm": 0.47568362951278687,
      "learning_rate": 8.248149055297105e-06,
      "loss": 0.3757,
      "step": 18921
    },
    {
      "epoch": 0.8683401404249461,
      "grad_norm": 0.43546515703201294,
      "learning_rate": 8.247962647116738e-06,
      "loss": 0.3672,
      "step": 18922
    },
    {
      "epoch": 0.8683860309302005,
      "grad_norm": 0.42523741722106934,
      "learning_rate": 8.247776231126049e-06,
      "loss": 0.4136,
      "step": 18923
    },
    {
      "epoch": 0.868431921435455,
      "grad_norm": 0.4796278774738312,
      "learning_rate": 8.247589807325481e-06,
      "loss": 0.4977,
      "step": 18924
    },
    {
      "epoch": 0.8684778119407095,
      "grad_norm": 0.45294222235679626,
      "learning_rate": 8.247403375715485e-06,
      "loss": 0.4213,
      "step": 18925
    },
    {
      "epoch": 0.8685237024459639,
      "grad_norm": 0.4636421203613281,
      "learning_rate": 8.24721693629651e-06,
      "loss": 0.4012,
      "step": 18926
    },
    {
      "epoch": 0.8685695929512184,
      "grad_norm": 0.48574891686439514,
      "learning_rate": 8.247030489069002e-06,
      "loss": 0.4308,
      "step": 18927
    },
    {
      "epoch": 0.8686154834564729,
      "grad_norm": 0.4446605145931244,
      "learning_rate": 8.24684403403341e-06,
      "loss": 0.3599,
      "step": 18928
    },
    {
      "epoch": 0.8686613739617273,
      "grad_norm": 0.45167192816734314,
      "learning_rate": 8.246657571190182e-06,
      "loss": 0.4395,
      "step": 18929
    },
    {
      "epoch": 0.8687072644669818,
      "grad_norm": 0.45047852396965027,
      "learning_rate": 8.246471100539769e-06,
      "loss": 0.4222,
      "step": 18930
    },
    {
      "epoch": 0.8687531549722363,
      "grad_norm": 0.4923040270805359,
      "learning_rate": 8.246284622082617e-06,
      "loss": 0.4653,
      "step": 18931
    },
    {
      "epoch": 0.8687990454774908,
      "grad_norm": 0.44747284054756165,
      "learning_rate": 8.246098135819173e-06,
      "loss": 0.4096,
      "step": 18932
    },
    {
      "epoch": 0.8688449359827451,
      "grad_norm": 0.4296838641166687,
      "learning_rate": 8.245911641749891e-06,
      "loss": 0.3386,
      "step": 18933
    },
    {
      "epoch": 0.8688908264879996,
      "grad_norm": 0.4538974463939667,
      "learning_rate": 8.245725139875214e-06,
      "loss": 0.3581,
      "step": 18934
    },
    {
      "epoch": 0.8689367169932541,
      "grad_norm": 0.4676780700683594,
      "learning_rate": 8.245538630195592e-06,
      "loss": 0.3865,
      "step": 18935
    },
    {
      "epoch": 0.8689826074985085,
      "grad_norm": 0.4392344653606415,
      "learning_rate": 8.245352112711475e-06,
      "loss": 0.3712,
      "step": 18936
    },
    {
      "epoch": 0.869028498003763,
      "grad_norm": 0.43207836151123047,
      "learning_rate": 8.245165587423312e-06,
      "loss": 0.3321,
      "step": 18937
    },
    {
      "epoch": 0.8690743885090175,
      "grad_norm": 1.0018877983093262,
      "learning_rate": 8.244979054331548e-06,
      "loss": 0.4496,
      "step": 18938
    },
    {
      "epoch": 0.8691202790142719,
      "grad_norm": 0.5198634266853333,
      "learning_rate": 8.244792513436633e-06,
      "loss": 0.4679,
      "step": 18939
    },
    {
      "epoch": 0.8691661695195264,
      "grad_norm": 0.4379604756832123,
      "learning_rate": 8.244605964739017e-06,
      "loss": 0.4103,
      "step": 18940
    },
    {
      "epoch": 0.8692120600247809,
      "grad_norm": 0.4432399868965149,
      "learning_rate": 8.244419408239149e-06,
      "loss": 0.3553,
      "step": 18941
    },
    {
      "epoch": 0.8692579505300353,
      "grad_norm": 0.48757341504096985,
      "learning_rate": 8.244232843937473e-06,
      "loss": 0.3878,
      "step": 18942
    },
    {
      "epoch": 0.8693038410352898,
      "grad_norm": 0.5038857460021973,
      "learning_rate": 8.244046271834443e-06,
      "loss": 0.4388,
      "step": 18943
    },
    {
      "epoch": 0.8693497315405443,
      "grad_norm": 0.412352591753006,
      "learning_rate": 8.243859691930505e-06,
      "loss": 0.3514,
      "step": 18944
    },
    {
      "epoch": 0.8693956220457987,
      "grad_norm": 0.46718764305114746,
      "learning_rate": 8.243673104226108e-06,
      "loss": 0.4264,
      "step": 18945
    },
    {
      "epoch": 0.8694415125510532,
      "grad_norm": 0.4639747440814972,
      "learning_rate": 8.243486508721703e-06,
      "loss": 0.3984,
      "step": 18946
    },
    {
      "epoch": 0.8694874030563077,
      "grad_norm": 0.43775689601898193,
      "learning_rate": 8.243299905417734e-06,
      "loss": 0.3452,
      "step": 18947
    },
    {
      "epoch": 0.8695332935615621,
      "grad_norm": 0.4658461809158325,
      "learning_rate": 8.243113294314653e-06,
      "loss": 0.4497,
      "step": 18948
    },
    {
      "epoch": 0.8695791840668166,
      "grad_norm": 0.41739100217819214,
      "learning_rate": 8.24292667541291e-06,
      "loss": 0.3434,
      "step": 18949
    },
    {
      "epoch": 0.869625074572071,
      "grad_norm": 0.43689823150634766,
      "learning_rate": 8.242740048712949e-06,
      "loss": 0.3791,
      "step": 18950
    },
    {
      "epoch": 0.8696709650773256,
      "grad_norm": 0.4136965572834015,
      "learning_rate": 8.242553414215225e-06,
      "loss": 0.3271,
      "step": 18951
    },
    {
      "epoch": 0.8697168555825799,
      "grad_norm": 0.4825584590435028,
      "learning_rate": 8.242366771920181e-06,
      "loss": 0.4237,
      "step": 18952
    },
    {
      "epoch": 0.8697627460878344,
      "grad_norm": 0.43105411529541016,
      "learning_rate": 8.242180121828267e-06,
      "loss": 0.3479,
      "step": 18953
    },
    {
      "epoch": 0.8698086365930889,
      "grad_norm": 0.4749305844306946,
      "learning_rate": 8.241993463939934e-06,
      "loss": 0.37,
      "step": 18954
    },
    {
      "epoch": 0.8698545270983433,
      "grad_norm": 0.5094777345657349,
      "learning_rate": 8.241806798255631e-06,
      "loss": 0.4555,
      "step": 18955
    },
    {
      "epoch": 0.8699004176035978,
      "grad_norm": 0.4483426809310913,
      "learning_rate": 8.241620124775806e-06,
      "loss": 0.3528,
      "step": 18956
    },
    {
      "epoch": 0.8699463081088523,
      "grad_norm": 0.45195645093917847,
      "learning_rate": 8.241433443500907e-06,
      "loss": 0.4251,
      "step": 18957
    },
    {
      "epoch": 0.8699921986141067,
      "grad_norm": 0.4376021921634674,
      "learning_rate": 8.241246754431383e-06,
      "loss": 0.3619,
      "step": 18958
    },
    {
      "epoch": 0.8700380891193612,
      "grad_norm": 0.45279720425605774,
      "learning_rate": 8.241060057567685e-06,
      "loss": 0.3452,
      "step": 18959
    },
    {
      "epoch": 0.8700839796246157,
      "grad_norm": 0.4642816483974457,
      "learning_rate": 8.240873352910259e-06,
      "loss": 0.3771,
      "step": 18960
    },
    {
      "epoch": 0.8701298701298701,
      "grad_norm": 0.45004868507385254,
      "learning_rate": 8.240686640459557e-06,
      "loss": 0.4384,
      "step": 18961
    },
    {
      "epoch": 0.8701757606351246,
      "grad_norm": 0.4612734615802765,
      "learning_rate": 8.240499920216025e-06,
      "loss": 0.4433,
      "step": 18962
    },
    {
      "epoch": 0.8702216511403791,
      "grad_norm": 0.4524123966693878,
      "learning_rate": 8.240313192180114e-06,
      "loss": 0.3948,
      "step": 18963
    },
    {
      "epoch": 0.8702675416456335,
      "grad_norm": 0.4905470907688141,
      "learning_rate": 8.240126456352272e-06,
      "loss": 0.5041,
      "step": 18964
    },
    {
      "epoch": 0.870313432150888,
      "grad_norm": 0.44299811124801636,
      "learning_rate": 8.23993971273295e-06,
      "loss": 0.376,
      "step": 18965
    },
    {
      "epoch": 0.8703593226561425,
      "grad_norm": 0.46464115381240845,
      "learning_rate": 8.239752961322594e-06,
      "loss": 0.4058,
      "step": 18966
    },
    {
      "epoch": 0.870405213161397,
      "grad_norm": 0.4572221040725708,
      "learning_rate": 8.239566202121653e-06,
      "loss": 0.4469,
      "step": 18967
    },
    {
      "epoch": 0.8704511036666513,
      "grad_norm": 0.4484426975250244,
      "learning_rate": 8.239379435130581e-06,
      "loss": 0.3954,
      "step": 18968
    },
    {
      "epoch": 0.8704969941719058,
      "grad_norm": 0.4404235780239105,
      "learning_rate": 8.239192660349822e-06,
      "loss": 0.3412,
      "step": 18969
    },
    {
      "epoch": 0.8705428846771603,
      "grad_norm": 0.4255707561969757,
      "learning_rate": 8.239005877779829e-06,
      "loss": 0.3297,
      "step": 18970
    },
    {
      "epoch": 0.8705887751824147,
      "grad_norm": 0.4300839900970459,
      "learning_rate": 8.238819087421047e-06,
      "loss": 0.3896,
      "step": 18971
    },
    {
      "epoch": 0.8706346656876692,
      "grad_norm": 0.4114404320716858,
      "learning_rate": 8.238632289273929e-06,
      "loss": 0.2998,
      "step": 18972
    },
    {
      "epoch": 0.8706805561929237,
      "grad_norm": 0.45493021607398987,
      "learning_rate": 8.23844548333892e-06,
      "loss": 0.3901,
      "step": 18973
    },
    {
      "epoch": 0.8707264466981781,
      "grad_norm": 0.42662930488586426,
      "learning_rate": 8.238258669616474e-06,
      "loss": 0.3739,
      "step": 18974
    },
    {
      "epoch": 0.8707723372034326,
      "grad_norm": 0.4271929860115051,
      "learning_rate": 8.238071848107036e-06,
      "loss": 0.3469,
      "step": 18975
    },
    {
      "epoch": 0.8708182277086871,
      "grad_norm": 0.46819308400154114,
      "learning_rate": 8.237885018811059e-06,
      "loss": 0.3612,
      "step": 18976
    },
    {
      "epoch": 0.8708641182139415,
      "grad_norm": 0.4325588643550873,
      "learning_rate": 8.23769818172899e-06,
      "loss": 0.3985,
      "step": 18977
    },
    {
      "epoch": 0.870910008719196,
      "grad_norm": 0.4454439580440521,
      "learning_rate": 8.237511336861277e-06,
      "loss": 0.4144,
      "step": 18978
    },
    {
      "epoch": 0.8709558992244505,
      "grad_norm": 0.42623424530029297,
      "learning_rate": 8.237324484208374e-06,
      "loss": 0.3741,
      "step": 18979
    },
    {
      "epoch": 0.8710017897297049,
      "grad_norm": 0.5444596409797668,
      "learning_rate": 8.237137623770726e-06,
      "loss": 0.5267,
      "step": 18980
    },
    {
      "epoch": 0.8710476802349594,
      "grad_norm": 0.38361847400665283,
      "learning_rate": 8.236950755548782e-06,
      "loss": 0.2846,
      "step": 18981
    },
    {
      "epoch": 0.8710935707402139,
      "grad_norm": 0.45108702778816223,
      "learning_rate": 8.236763879542994e-06,
      "loss": 0.3312,
      "step": 18982
    },
    {
      "epoch": 0.8711394612454683,
      "grad_norm": 0.437872052192688,
      "learning_rate": 8.23657699575381e-06,
      "loss": 0.3938,
      "step": 18983
    },
    {
      "epoch": 0.8711853517507228,
      "grad_norm": 0.4380604326725006,
      "learning_rate": 8.236390104181682e-06,
      "loss": 0.368,
      "step": 18984
    },
    {
      "epoch": 0.8712312422559773,
      "grad_norm": 0.49645957350730896,
      "learning_rate": 8.236203204827057e-06,
      "loss": 0.4459,
      "step": 18985
    },
    {
      "epoch": 0.8712771327612318,
      "grad_norm": 0.4470571279525757,
      "learning_rate": 8.236016297690384e-06,
      "loss": 0.3791,
      "step": 18986
    },
    {
      "epoch": 0.8713230232664861,
      "grad_norm": 0.4759295880794525,
      "learning_rate": 8.235829382772111e-06,
      "loss": 0.4757,
      "step": 18987
    },
    {
      "epoch": 0.8713689137717406,
      "grad_norm": 0.4687975347042084,
      "learning_rate": 8.235642460072692e-06,
      "loss": 0.433,
      "step": 18988
    },
    {
      "epoch": 0.8714148042769951,
      "grad_norm": 0.4125252962112427,
      "learning_rate": 8.235455529592573e-06,
      "loss": 0.3391,
      "step": 18989
    },
    {
      "epoch": 0.8714606947822495,
      "grad_norm": 0.4338414669036865,
      "learning_rate": 8.235268591332204e-06,
      "loss": 0.3897,
      "step": 18990
    },
    {
      "epoch": 0.871506585287504,
      "grad_norm": 0.4246147871017456,
      "learning_rate": 8.235081645292039e-06,
      "loss": 0.3868,
      "step": 18991
    },
    {
      "epoch": 0.8715524757927585,
      "grad_norm": 0.7082940340042114,
      "learning_rate": 8.234894691472519e-06,
      "loss": 0.4117,
      "step": 18992
    },
    {
      "epoch": 0.8715983662980129,
      "grad_norm": 0.48406529426574707,
      "learning_rate": 8.234707729874101e-06,
      "loss": 0.406,
      "step": 18993
    },
    {
      "epoch": 0.8716442568032674,
      "grad_norm": 0.43817996978759766,
      "learning_rate": 8.23452076049723e-06,
      "loss": 0.374,
      "step": 18994
    },
    {
      "epoch": 0.8716901473085219,
      "grad_norm": 0.46787071228027344,
      "learning_rate": 8.234333783342358e-06,
      "loss": 0.4087,
      "step": 18995
    },
    {
      "epoch": 0.8717360378137763,
      "grad_norm": 0.476106733083725,
      "learning_rate": 8.234146798409935e-06,
      "loss": 0.4314,
      "step": 18996
    },
    {
      "epoch": 0.8717819283190308,
      "grad_norm": 0.43047866225242615,
      "learning_rate": 8.233959805700408e-06,
      "loss": 0.3203,
      "step": 18997
    },
    {
      "epoch": 0.8718278188242853,
      "grad_norm": 0.4719417989253998,
      "learning_rate": 8.23377280521423e-06,
      "loss": 0.4279,
      "step": 18998
    },
    {
      "epoch": 0.8718737093295397,
      "grad_norm": 0.4607294797897339,
      "learning_rate": 8.233585796951849e-06,
      "loss": 0.4264,
      "step": 18999
    },
    {
      "epoch": 0.8719195998347942,
      "grad_norm": 0.4721177816390991,
      "learning_rate": 8.233398780913712e-06,
      "loss": 0.3925,
      "step": 19000
    },
    {
      "epoch": 0.8719654903400487,
      "grad_norm": 0.4804874360561371,
      "learning_rate": 8.233211757100274e-06,
      "loss": 0.422,
      "step": 19001
    },
    {
      "epoch": 0.872011380845303,
      "grad_norm": 0.45685338973999023,
      "learning_rate": 8.233024725511983e-06,
      "loss": 0.3845,
      "step": 19002
    },
    {
      "epoch": 0.8720572713505576,
      "grad_norm": 0.4628857672214508,
      "learning_rate": 8.232837686149286e-06,
      "loss": 0.399,
      "step": 19003
    },
    {
      "epoch": 0.872103161855812,
      "grad_norm": 0.45493608713150024,
      "learning_rate": 8.232650639012635e-06,
      "loss": 0.3675,
      "step": 19004
    },
    {
      "epoch": 0.8721490523610665,
      "grad_norm": 0.4375508725643158,
      "learning_rate": 8.232463584102479e-06,
      "loss": 0.3223,
      "step": 19005
    },
    {
      "epoch": 0.8721949428663209,
      "grad_norm": 0.4677862524986267,
      "learning_rate": 8.23227652141927e-06,
      "loss": 0.4167,
      "step": 19006
    },
    {
      "epoch": 0.8722408333715754,
      "grad_norm": 0.41010594367980957,
      "learning_rate": 8.232089450963453e-06,
      "loss": 0.332,
      "step": 19007
    },
    {
      "epoch": 0.8722867238768299,
      "grad_norm": 0.5010298490524292,
      "learning_rate": 8.231902372735485e-06,
      "loss": 0.5066,
      "step": 19008
    },
    {
      "epoch": 0.8723326143820843,
      "grad_norm": 0.42294782400131226,
      "learning_rate": 8.231715286735808e-06,
      "loss": 0.3387,
      "step": 19009
    },
    {
      "epoch": 0.8723785048873388,
      "grad_norm": 0.48107004165649414,
      "learning_rate": 8.231528192964879e-06,
      "loss": 0.4038,
      "step": 19010
    },
    {
      "epoch": 0.8724243953925933,
      "grad_norm": 0.4320259094238281,
      "learning_rate": 8.231341091423143e-06,
      "loss": 0.3129,
      "step": 19011
    },
    {
      "epoch": 0.8724702858978477,
      "grad_norm": 0.4658711552619934,
      "learning_rate": 8.231153982111051e-06,
      "loss": 0.3647,
      "step": 19012
    },
    {
      "epoch": 0.8725161764031022,
      "grad_norm": 0.6039180755615234,
      "learning_rate": 8.230966865029055e-06,
      "loss": 0.4519,
      "step": 19013
    },
    {
      "epoch": 0.8725620669083567,
      "grad_norm": 0.4518488347530365,
      "learning_rate": 8.230779740177601e-06,
      "loss": 0.4049,
      "step": 19014
    },
    {
      "epoch": 0.8726079574136111,
      "grad_norm": 0.4543609321117401,
      "learning_rate": 8.230592607557145e-06,
      "loss": 0.3625,
      "step": 19015
    },
    {
      "epoch": 0.8726538479188656,
      "grad_norm": 0.5034462809562683,
      "learning_rate": 8.23040546716813e-06,
      "loss": 0.4367,
      "step": 19016
    },
    {
      "epoch": 0.8726997384241201,
      "grad_norm": 0.45420554280281067,
      "learning_rate": 8.230218319011011e-06,
      "loss": 0.4361,
      "step": 19017
    },
    {
      "epoch": 0.8727456289293745,
      "grad_norm": 0.4854552149772644,
      "learning_rate": 8.230031163086235e-06,
      "loss": 0.4314,
      "step": 19018
    },
    {
      "epoch": 0.872791519434629,
      "grad_norm": 0.4491330087184906,
      "learning_rate": 8.229843999394255e-06,
      "loss": 0.3849,
      "step": 19019
    },
    {
      "epoch": 0.8728374099398835,
      "grad_norm": 0.45390549302101135,
      "learning_rate": 8.229656827935519e-06,
      "loss": 0.4149,
      "step": 19020
    },
    {
      "epoch": 0.872883300445138,
      "grad_norm": 0.41391807794570923,
      "learning_rate": 8.229469648710477e-06,
      "loss": 0.3386,
      "step": 19021
    },
    {
      "epoch": 0.8729291909503923,
      "grad_norm": 0.4536173939704895,
      "learning_rate": 8.22928246171958e-06,
      "loss": 0.4153,
      "step": 19022
    },
    {
      "epoch": 0.8729750814556468,
      "grad_norm": 0.47305217385292053,
      "learning_rate": 8.229095266963279e-06,
      "loss": 0.4441,
      "step": 19023
    },
    {
      "epoch": 0.8730209719609013,
      "grad_norm": 0.40642306208610535,
      "learning_rate": 8.228908064442023e-06,
      "loss": 0.2845,
      "step": 19024
    },
    {
      "epoch": 0.8730668624661557,
      "grad_norm": 0.4578503370285034,
      "learning_rate": 8.228720854156262e-06,
      "loss": 0.3653,
      "step": 19025
    },
    {
      "epoch": 0.8731127529714102,
      "grad_norm": 0.4243710935115814,
      "learning_rate": 8.228533636106445e-06,
      "loss": 0.3605,
      "step": 19026
    },
    {
      "epoch": 0.8731586434766647,
      "grad_norm": 0.4913235902786255,
      "learning_rate": 8.228346410293025e-06,
      "loss": 0.4864,
      "step": 19027
    },
    {
      "epoch": 0.8732045339819191,
      "grad_norm": 0.5014052391052246,
      "learning_rate": 8.22815917671645e-06,
      "loss": 0.514,
      "step": 19028
    },
    {
      "epoch": 0.8732504244871736,
      "grad_norm": 0.446957528591156,
      "learning_rate": 8.22797193537717e-06,
      "loss": 0.3933,
      "step": 19029
    },
    {
      "epoch": 0.8732963149924281,
      "grad_norm": 0.43559396266937256,
      "learning_rate": 8.227784686275639e-06,
      "loss": 0.3524,
      "step": 19030
    },
    {
      "epoch": 0.8733422054976825,
      "grad_norm": 0.47324392199516296,
      "learning_rate": 8.227597429412303e-06,
      "loss": 0.3665,
      "step": 19031
    },
    {
      "epoch": 0.873388096002937,
      "grad_norm": 0.45771029591560364,
      "learning_rate": 8.227410164787613e-06,
      "loss": 0.3457,
      "step": 19032
    },
    {
      "epoch": 0.8734339865081915,
      "grad_norm": 0.4326404333114624,
      "learning_rate": 8.227222892402022e-06,
      "loss": 0.3516,
      "step": 19033
    },
    {
      "epoch": 0.8734798770134459,
      "grad_norm": 0.49116235971450806,
      "learning_rate": 8.227035612255979e-06,
      "loss": 0.4225,
      "step": 19034
    },
    {
      "epoch": 0.8735257675187004,
      "grad_norm": 0.4289708435535431,
      "learning_rate": 8.22684832434993e-06,
      "loss": 0.3725,
      "step": 19035
    },
    {
      "epoch": 0.8735716580239549,
      "grad_norm": 0.47323527932167053,
      "learning_rate": 8.226661028684332e-06,
      "loss": 0.4856,
      "step": 19036
    },
    {
      "epoch": 0.8736175485292093,
      "grad_norm": 0.481403112411499,
      "learning_rate": 8.226473725259631e-06,
      "loss": 0.4341,
      "step": 19037
    },
    {
      "epoch": 0.8736634390344638,
      "grad_norm": 0.45308342576026917,
      "learning_rate": 8.22628641407628e-06,
      "loss": 0.4133,
      "step": 19038
    },
    {
      "epoch": 0.8737093295397182,
      "grad_norm": 0.4240788519382477,
      "learning_rate": 8.22609909513473e-06,
      "loss": 0.3597,
      "step": 19039
    },
    {
      "epoch": 0.8737552200449727,
      "grad_norm": 0.44183915853500366,
      "learning_rate": 8.225911768435429e-06,
      "loss": 0.379,
      "step": 19040
    },
    {
      "epoch": 0.8738011105502271,
      "grad_norm": 0.4424301087856293,
      "learning_rate": 8.225724433978829e-06,
      "loss": 0.3993,
      "step": 19041
    },
    {
      "epoch": 0.8738470010554816,
      "grad_norm": 0.41829046607017517,
      "learning_rate": 8.225537091765377e-06,
      "loss": 0.3602,
      "step": 19042
    },
    {
      "epoch": 0.8738928915607361,
      "grad_norm": 0.48969876766204834,
      "learning_rate": 8.225349741795527e-06,
      "loss": 0.4225,
      "step": 19043
    },
    {
      "epoch": 0.8739387820659905,
      "grad_norm": 0.47473007440567017,
      "learning_rate": 8.22516238406973e-06,
      "loss": 0.377,
      "step": 19044
    },
    {
      "epoch": 0.873984672571245,
      "grad_norm": 0.4479834735393524,
      "learning_rate": 8.224975018588436e-06,
      "loss": 0.3745,
      "step": 19045
    },
    {
      "epoch": 0.8740305630764995,
      "grad_norm": 0.4330675005912781,
      "learning_rate": 8.224787645352094e-06,
      "loss": 0.4078,
      "step": 19046
    },
    {
      "epoch": 0.8740764535817539,
      "grad_norm": 0.4831138551235199,
      "learning_rate": 8.224600264361155e-06,
      "loss": 0.4365,
      "step": 19047
    },
    {
      "epoch": 0.8741223440870084,
      "grad_norm": 0.4695335626602173,
      "learning_rate": 8.22441287561607e-06,
      "loss": 0.3889,
      "step": 19048
    },
    {
      "epoch": 0.8741682345922629,
      "grad_norm": 0.44154486060142517,
      "learning_rate": 8.22422547911729e-06,
      "loss": 0.3969,
      "step": 19049
    },
    {
      "epoch": 0.8742141250975173,
      "grad_norm": 0.43685922026634216,
      "learning_rate": 8.224038074865266e-06,
      "loss": 0.3652,
      "step": 19050
    },
    {
      "epoch": 0.8742600156027718,
      "grad_norm": 0.4427495002746582,
      "learning_rate": 8.223850662860448e-06,
      "loss": 0.3511,
      "step": 19051
    },
    {
      "epoch": 0.8743059061080263,
      "grad_norm": 0.4470098912715912,
      "learning_rate": 8.223663243103287e-06,
      "loss": 0.3903,
      "step": 19052
    },
    {
      "epoch": 0.8743517966132807,
      "grad_norm": 0.42210766673088074,
      "learning_rate": 8.22347581559423e-06,
      "loss": 0.3422,
      "step": 19053
    },
    {
      "epoch": 0.8743976871185352,
      "grad_norm": 0.4271153509616852,
      "learning_rate": 8.223288380333733e-06,
      "loss": 0.3378,
      "step": 19054
    },
    {
      "epoch": 0.8744435776237897,
      "grad_norm": 0.4593006372451782,
      "learning_rate": 8.223100937322246e-06,
      "loss": 0.4304,
      "step": 19055
    },
    {
      "epoch": 0.8744894681290442,
      "grad_norm": 0.5093809366226196,
      "learning_rate": 8.222913486560219e-06,
      "loss": 0.4774,
      "step": 19056
    },
    {
      "epoch": 0.8745353586342985,
      "grad_norm": 0.48551321029663086,
      "learning_rate": 8.2227260280481e-06,
      "loss": 0.3749,
      "step": 19057
    },
    {
      "epoch": 0.874581249139553,
      "grad_norm": 0.45765891671180725,
      "learning_rate": 8.222538561786344e-06,
      "loss": 0.363,
      "step": 19058
    },
    {
      "epoch": 0.8746271396448075,
      "grad_norm": 0.4293176233768463,
      "learning_rate": 8.222351087775397e-06,
      "loss": 0.3477,
      "step": 19059
    },
    {
      "epoch": 0.8746730301500619,
      "grad_norm": 0.4705008268356323,
      "learning_rate": 8.222163606015717e-06,
      "loss": 0.4266,
      "step": 19060
    },
    {
      "epoch": 0.8747189206553164,
      "grad_norm": 0.46184825897216797,
      "learning_rate": 8.221976116507747e-06,
      "loss": 0.4191,
      "step": 19061
    },
    {
      "epoch": 0.8747648111605709,
      "grad_norm": 0.4687638580799103,
      "learning_rate": 8.221788619251942e-06,
      "loss": 0.3593,
      "step": 19062
    },
    {
      "epoch": 0.8748107016658253,
      "grad_norm": 0.42995747923851013,
      "learning_rate": 8.221601114248751e-06,
      "loss": 0.3872,
      "step": 19063
    },
    {
      "epoch": 0.8748565921710798,
      "grad_norm": 0.45432549715042114,
      "learning_rate": 8.221413601498629e-06,
      "loss": 0.4013,
      "step": 19064
    },
    {
      "epoch": 0.8749024826763343,
      "grad_norm": 0.5138528943061829,
      "learning_rate": 8.221226081002023e-06,
      "loss": 0.5026,
      "step": 19065
    },
    {
      "epoch": 0.8749483731815887,
      "grad_norm": 0.4685092866420746,
      "learning_rate": 8.221038552759383e-06,
      "loss": 0.4332,
      "step": 19066
    },
    {
      "epoch": 0.8749942636868432,
      "grad_norm": 0.46465200185775757,
      "learning_rate": 8.220851016771165e-06,
      "loss": 0.4145,
      "step": 19067
    },
    {
      "epoch": 0.8750401541920977,
      "grad_norm": 0.4531886577606201,
      "learning_rate": 8.220663473037814e-06,
      "loss": 0.3997,
      "step": 19068
    },
    {
      "epoch": 0.8750860446973521,
      "grad_norm": 0.49405908584594727,
      "learning_rate": 8.220475921559785e-06,
      "loss": 0.4468,
      "step": 19069
    },
    {
      "epoch": 0.8751319352026066,
      "grad_norm": 0.42974936962127686,
      "learning_rate": 8.220288362337527e-06,
      "loss": 0.3707,
      "step": 19070
    },
    {
      "epoch": 0.8751778257078611,
      "grad_norm": 0.4380975067615509,
      "learning_rate": 8.220100795371492e-06,
      "loss": 0.3733,
      "step": 19071
    },
    {
      "epoch": 0.8752237162131155,
      "grad_norm": 0.4969840347766876,
      "learning_rate": 8.219913220662132e-06,
      "loss": 0.4724,
      "step": 19072
    },
    {
      "epoch": 0.87526960671837,
      "grad_norm": 0.4308072030544281,
      "learning_rate": 8.219725638209897e-06,
      "loss": 0.3414,
      "step": 19073
    },
    {
      "epoch": 0.8753154972236245,
      "grad_norm": 1.2691686153411865,
      "learning_rate": 8.219538048015237e-06,
      "loss": 0.5702,
      "step": 19074
    },
    {
      "epoch": 0.875361387728879,
      "grad_norm": 0.42908746004104614,
      "learning_rate": 8.219350450078606e-06,
      "loss": 0.3945,
      "step": 19075
    },
    {
      "epoch": 0.8754072782341333,
      "grad_norm": 0.4702817499637604,
      "learning_rate": 8.21916284440045e-06,
      "loss": 0.3811,
      "step": 19076
    },
    {
      "epoch": 0.8754531687393878,
      "grad_norm": 0.5116704106330872,
      "learning_rate": 8.218975230981225e-06,
      "loss": 0.4832,
      "step": 19077
    },
    {
      "epoch": 0.8754990592446423,
      "grad_norm": 0.5056906342506409,
      "learning_rate": 8.218787609821382e-06,
      "loss": 0.4586,
      "step": 19078
    },
    {
      "epoch": 0.8755449497498967,
      "grad_norm": 0.4328555762767792,
      "learning_rate": 8.218599980921368e-06,
      "loss": 0.3988,
      "step": 19079
    },
    {
      "epoch": 0.8755908402551512,
      "grad_norm": 0.4441078305244446,
      "learning_rate": 8.218412344281639e-06,
      "loss": 0.3223,
      "step": 19080
    },
    {
      "epoch": 0.8756367307604057,
      "grad_norm": 0.4605567157268524,
      "learning_rate": 8.218224699902644e-06,
      "loss": 0.4304,
      "step": 19081
    },
    {
      "epoch": 0.8756826212656601,
      "grad_norm": 0.45149603486061096,
      "learning_rate": 8.218037047784833e-06,
      "loss": 0.4199,
      "step": 19082
    },
    {
      "epoch": 0.8757285117709146,
      "grad_norm": 0.4639022946357727,
      "learning_rate": 8.217849387928659e-06,
      "loss": 0.3796,
      "step": 19083
    },
    {
      "epoch": 0.8757744022761691,
      "grad_norm": 0.48578768968582153,
      "learning_rate": 8.217661720334573e-06,
      "loss": 0.475,
      "step": 19084
    },
    {
      "epoch": 0.8758202927814235,
      "grad_norm": 0.45571503043174744,
      "learning_rate": 8.217474045003026e-06,
      "loss": 0.3777,
      "step": 19085
    },
    {
      "epoch": 0.875866183286678,
      "grad_norm": 0.46412986516952515,
      "learning_rate": 8.21728636193447e-06,
      "loss": 0.3963,
      "step": 19086
    },
    {
      "epoch": 0.8759120737919325,
      "grad_norm": 0.44513291120529175,
      "learning_rate": 8.217098671129355e-06,
      "loss": 0.4224,
      "step": 19087
    },
    {
      "epoch": 0.8759579642971869,
      "grad_norm": 0.4397667944431305,
      "learning_rate": 8.216910972588133e-06,
      "loss": 0.3378,
      "step": 19088
    },
    {
      "epoch": 0.8760038548024414,
      "grad_norm": 0.44363319873809814,
      "learning_rate": 8.216723266311255e-06,
      "loss": 0.368,
      "step": 19089
    },
    {
      "epoch": 0.8760497453076959,
      "grad_norm": 0.4495210349559784,
      "learning_rate": 8.216535552299174e-06,
      "loss": 0.3483,
      "step": 19090
    },
    {
      "epoch": 0.8760956358129502,
      "grad_norm": 0.46948179602622986,
      "learning_rate": 8.216347830552339e-06,
      "loss": 0.3974,
      "step": 19091
    },
    {
      "epoch": 0.8761415263182047,
      "grad_norm": 0.46009600162506104,
      "learning_rate": 8.216160101071203e-06,
      "loss": 0.4348,
      "step": 19092
    },
    {
      "epoch": 0.8761874168234592,
      "grad_norm": 0.4910718500614166,
      "learning_rate": 8.215972363856217e-06,
      "loss": 0.4031,
      "step": 19093
    },
    {
      "epoch": 0.8762333073287137,
      "grad_norm": 0.4709832966327667,
      "learning_rate": 8.215784618907831e-06,
      "loss": 0.4033,
      "step": 19094
    },
    {
      "epoch": 0.8762791978339681,
      "grad_norm": 0.48434245586395264,
      "learning_rate": 8.215596866226501e-06,
      "loss": 0.4424,
      "step": 19095
    },
    {
      "epoch": 0.8763250883392226,
      "grad_norm": 0.45961257815361023,
      "learning_rate": 8.215409105812674e-06,
      "loss": 0.3994,
      "step": 19096
    },
    {
      "epoch": 0.8763709788444771,
      "grad_norm": 0.4488162398338318,
      "learning_rate": 8.215221337666802e-06,
      "loss": 0.3801,
      "step": 19097
    },
    {
      "epoch": 0.8764168693497315,
      "grad_norm": 0.45952269434928894,
      "learning_rate": 8.21503356178934e-06,
      "loss": 0.41,
      "step": 19098
    },
    {
      "epoch": 0.876462759854986,
      "grad_norm": 0.45821550488471985,
      "learning_rate": 8.214845778180733e-06,
      "loss": 0.3548,
      "step": 19099
    },
    {
      "epoch": 0.8765086503602405,
      "grad_norm": 0.43187767267227173,
      "learning_rate": 8.21465798684144e-06,
      "loss": 0.3674,
      "step": 19100
    },
    {
      "epoch": 0.8765545408654949,
      "grad_norm": 0.4792787432670593,
      "learning_rate": 8.214470187771907e-06,
      "loss": 0.4454,
      "step": 19101
    },
    {
      "epoch": 0.8766004313707494,
      "grad_norm": 0.44913235306739807,
      "learning_rate": 8.21428238097259e-06,
      "loss": 0.3836,
      "step": 19102
    },
    {
      "epoch": 0.8766463218760039,
      "grad_norm": 0.4326745271682739,
      "learning_rate": 8.214094566443936e-06,
      "loss": 0.324,
      "step": 19103
    },
    {
      "epoch": 0.8766922123812583,
      "grad_norm": 0.43010279536247253,
      "learning_rate": 8.213906744186399e-06,
      "loss": 0.3622,
      "step": 19104
    },
    {
      "epoch": 0.8767381028865128,
      "grad_norm": 0.48416927456855774,
      "learning_rate": 8.213718914200432e-06,
      "loss": 0.4649,
      "step": 19105
    },
    {
      "epoch": 0.8767839933917673,
      "grad_norm": 0.4400700628757477,
      "learning_rate": 8.213531076486484e-06,
      "loss": 0.394,
      "step": 19106
    },
    {
      "epoch": 0.8768298838970217,
      "grad_norm": 0.4606837332248688,
      "learning_rate": 8.21334323104501e-06,
      "loss": 0.391,
      "step": 19107
    },
    {
      "epoch": 0.8768757744022762,
      "grad_norm": 0.47147029638290405,
      "learning_rate": 8.213155377876458e-06,
      "loss": 0.388,
      "step": 19108
    },
    {
      "epoch": 0.8769216649075307,
      "grad_norm": 0.44349220395088196,
      "learning_rate": 8.212967516981281e-06,
      "loss": 0.4602,
      "step": 19109
    },
    {
      "epoch": 0.8769675554127851,
      "grad_norm": 0.4863872528076172,
      "learning_rate": 8.212779648359932e-06,
      "loss": 0.4602,
      "step": 19110
    },
    {
      "epoch": 0.8770134459180395,
      "grad_norm": 0.4206690788269043,
      "learning_rate": 8.21259177201286e-06,
      "loss": 0.3361,
      "step": 19111
    },
    {
      "epoch": 0.877059336423294,
      "grad_norm": 0.4568292498588562,
      "learning_rate": 8.212403887940523e-06,
      "loss": 0.4317,
      "step": 19112
    },
    {
      "epoch": 0.8771052269285485,
      "grad_norm": 0.4856686592102051,
      "learning_rate": 8.212215996143365e-06,
      "loss": 0.3888,
      "step": 19113
    },
    {
      "epoch": 0.8771511174338029,
      "grad_norm": 0.44106027483940125,
      "learning_rate": 8.212028096621843e-06,
      "loss": 0.3697,
      "step": 19114
    },
    {
      "epoch": 0.8771970079390574,
      "grad_norm": 0.4441925585269928,
      "learning_rate": 8.211840189376407e-06,
      "loss": 0.3883,
      "step": 19115
    },
    {
      "epoch": 0.8772428984443119,
      "grad_norm": 0.43758425116539,
      "learning_rate": 8.21165227440751e-06,
      "loss": 0.4002,
      "step": 19116
    },
    {
      "epoch": 0.8772887889495663,
      "grad_norm": 0.42489108443260193,
      "learning_rate": 8.2114643517156e-06,
      "loss": 0.3662,
      "step": 19117
    },
    {
      "epoch": 0.8773346794548208,
      "grad_norm": 0.43275192379951477,
      "learning_rate": 8.211276421301133e-06,
      "loss": 0.3759,
      "step": 19118
    },
    {
      "epoch": 0.8773805699600753,
      "grad_norm": 0.4375339448451996,
      "learning_rate": 8.21108848316456e-06,
      "loss": 0.3837,
      "step": 19119
    },
    {
      "epoch": 0.8774264604653297,
      "grad_norm": 0.487178236246109,
      "learning_rate": 8.210900537306334e-06,
      "loss": 0.4211,
      "step": 19120
    },
    {
      "epoch": 0.8774723509705842,
      "grad_norm": 0.46321022510528564,
      "learning_rate": 8.210712583726903e-06,
      "loss": 0.4348,
      "step": 19121
    },
    {
      "epoch": 0.8775182414758387,
      "grad_norm": 0.49689555168151855,
      "learning_rate": 8.210524622426723e-06,
      "loss": 0.4191,
      "step": 19122
    },
    {
      "epoch": 0.8775641319810931,
      "grad_norm": 0.4719978868961334,
      "learning_rate": 8.210336653406245e-06,
      "loss": 0.3815,
      "step": 19123
    },
    {
      "epoch": 0.8776100224863476,
      "grad_norm": 0.42269986867904663,
      "learning_rate": 8.21014867666592e-06,
      "loss": 0.359,
      "step": 19124
    },
    {
      "epoch": 0.8776559129916021,
      "grad_norm": 0.47000640630722046,
      "learning_rate": 8.2099606922062e-06,
      "loss": 0.3486,
      "step": 19125
    },
    {
      "epoch": 0.8777018034968564,
      "grad_norm": 0.5031851530075073,
      "learning_rate": 8.209772700027539e-06,
      "loss": 0.5425,
      "step": 19126
    },
    {
      "epoch": 0.877747694002111,
      "grad_norm": 0.4548361599445343,
      "learning_rate": 8.209584700130386e-06,
      "loss": 0.4071,
      "step": 19127
    },
    {
      "epoch": 0.8777935845073654,
      "grad_norm": 0.4200451076030731,
      "learning_rate": 8.209396692515197e-06,
      "loss": 0.3156,
      "step": 19128
    },
    {
      "epoch": 0.8778394750126199,
      "grad_norm": 0.44227859377861023,
      "learning_rate": 8.20920867718242e-06,
      "loss": 0.3847,
      "step": 19129
    },
    {
      "epoch": 0.8778853655178743,
      "grad_norm": 0.46306532621383667,
      "learning_rate": 8.209020654132508e-06,
      "loss": 0.4217,
      "step": 19130
    },
    {
      "epoch": 0.8779312560231288,
      "grad_norm": 0.49015966057777405,
      "learning_rate": 8.208832623365916e-06,
      "loss": 0.4639,
      "step": 19131
    },
    {
      "epoch": 0.8779771465283833,
      "grad_norm": 0.46084481477737427,
      "learning_rate": 8.208644584883096e-06,
      "loss": 0.4153,
      "step": 19132
    },
    {
      "epoch": 0.8780230370336377,
      "grad_norm": 0.4666290581226349,
      "learning_rate": 8.208456538684495e-06,
      "loss": 0.3982,
      "step": 19133
    },
    {
      "epoch": 0.8780689275388922,
      "grad_norm": 0.4263615012168884,
      "learning_rate": 8.208268484770569e-06,
      "loss": 0.349,
      "step": 19134
    },
    {
      "epoch": 0.8781148180441467,
      "grad_norm": 0.4571050703525543,
      "learning_rate": 8.208080423141772e-06,
      "loss": 0.3969,
      "step": 19135
    },
    {
      "epoch": 0.8781607085494011,
      "grad_norm": 0.4740146994590759,
      "learning_rate": 8.207892353798553e-06,
      "loss": 0.4822,
      "step": 19136
    },
    {
      "epoch": 0.8782065990546556,
      "grad_norm": 0.4939623475074768,
      "learning_rate": 8.207704276741365e-06,
      "loss": 0.4584,
      "step": 19137
    },
    {
      "epoch": 0.8782524895599101,
      "grad_norm": 0.4671875834465027,
      "learning_rate": 8.207516191970662e-06,
      "loss": 0.479,
      "step": 19138
    },
    {
      "epoch": 0.8782983800651645,
      "grad_norm": 0.4533805847167969,
      "learning_rate": 8.207328099486893e-06,
      "loss": 0.3422,
      "step": 19139
    },
    {
      "epoch": 0.878344270570419,
      "grad_norm": 0.48585575819015503,
      "learning_rate": 8.207139999290512e-06,
      "loss": 0.4072,
      "step": 19140
    },
    {
      "epoch": 0.8783901610756735,
      "grad_norm": 0.4155079424381256,
      "learning_rate": 8.206951891381973e-06,
      "loss": 0.3136,
      "step": 19141
    },
    {
      "epoch": 0.8784360515809279,
      "grad_norm": 0.44877225160598755,
      "learning_rate": 8.206763775761726e-06,
      "loss": 0.3597,
      "step": 19142
    },
    {
      "epoch": 0.8784819420861824,
      "grad_norm": 0.45714521408081055,
      "learning_rate": 8.206575652430225e-06,
      "loss": 0.4447,
      "step": 19143
    },
    {
      "epoch": 0.8785278325914369,
      "grad_norm": 0.43482619524002075,
      "learning_rate": 8.206387521387921e-06,
      "loss": 0.3604,
      "step": 19144
    },
    {
      "epoch": 0.8785737230966912,
      "grad_norm": 0.44424766302108765,
      "learning_rate": 8.206199382635267e-06,
      "loss": 0.3941,
      "step": 19145
    },
    {
      "epoch": 0.8786196136019457,
      "grad_norm": 0.456699013710022,
      "learning_rate": 8.206011236172715e-06,
      "loss": 0.3533,
      "step": 19146
    },
    {
      "epoch": 0.8786655041072002,
      "grad_norm": 0.46785077452659607,
      "learning_rate": 8.205823082000719e-06,
      "loss": 0.3988,
      "step": 19147
    },
    {
      "epoch": 0.8787113946124547,
      "grad_norm": 0.42983683943748474,
      "learning_rate": 8.20563492011973e-06,
      "loss": 0.353,
      "step": 19148
    },
    {
      "epoch": 0.8787572851177091,
      "grad_norm": 0.42816194891929626,
      "learning_rate": 8.2054467505302e-06,
      "loss": 0.3347,
      "step": 19149
    },
    {
      "epoch": 0.8788031756229636,
      "grad_norm": 0.4148336350917816,
      "learning_rate": 8.205258573232582e-06,
      "loss": 0.3583,
      "step": 19150
    },
    {
      "epoch": 0.8788490661282181,
      "grad_norm": 0.46545499563217163,
      "learning_rate": 8.20507038822733e-06,
      "loss": 0.3883,
      "step": 19151
    },
    {
      "epoch": 0.8788949566334725,
      "grad_norm": 0.47809427976608276,
      "learning_rate": 8.204882195514893e-06,
      "loss": 0.3982,
      "step": 19152
    },
    {
      "epoch": 0.878940847138727,
      "grad_norm": 0.4478949308395386,
      "learning_rate": 8.20469399509573e-06,
      "loss": 0.4083,
      "step": 19153
    },
    {
      "epoch": 0.8789867376439815,
      "grad_norm": 0.43192213773727417,
      "learning_rate": 8.204505786970284e-06,
      "loss": 0.3591,
      "step": 19154
    },
    {
      "epoch": 0.8790326281492359,
      "grad_norm": 0.4654843807220459,
      "learning_rate": 8.204317571139017e-06,
      "loss": 0.4525,
      "step": 19155
    },
    {
      "epoch": 0.8790785186544904,
      "grad_norm": 0.4616166651248932,
      "learning_rate": 8.204129347602377e-06,
      "loss": 0.3945,
      "step": 19156
    },
    {
      "epoch": 0.8791244091597449,
      "grad_norm": 0.4762967526912689,
      "learning_rate": 8.203941116360816e-06,
      "loss": 0.4315,
      "step": 19157
    },
    {
      "epoch": 0.8791702996649993,
      "grad_norm": 0.4483889937400818,
      "learning_rate": 8.203752877414787e-06,
      "loss": 0.3611,
      "step": 19158
    },
    {
      "epoch": 0.8792161901702538,
      "grad_norm": 0.46035635471343994,
      "learning_rate": 8.203564630764745e-06,
      "loss": 0.429,
      "step": 19159
    },
    {
      "epoch": 0.8792620806755083,
      "grad_norm": 0.5128582119941711,
      "learning_rate": 8.203376376411141e-06,
      "loss": 0.4537,
      "step": 19160
    },
    {
      "epoch": 0.8793079711807626,
      "grad_norm": 0.4780116677284241,
      "learning_rate": 8.203188114354429e-06,
      "loss": 0.4266,
      "step": 19161
    },
    {
      "epoch": 0.8793538616860171,
      "grad_norm": 0.44106414914131165,
      "learning_rate": 8.202999844595058e-06,
      "loss": 0.3786,
      "step": 19162
    },
    {
      "epoch": 0.8793997521912716,
      "grad_norm": 0.4755333960056305,
      "learning_rate": 8.202811567133484e-06,
      "loss": 0.4437,
      "step": 19163
    },
    {
      "epoch": 0.8794456426965261,
      "grad_norm": 0.4286287724971771,
      "learning_rate": 8.20262328197016e-06,
      "loss": 0.3903,
      "step": 19164
    },
    {
      "epoch": 0.8794915332017805,
      "grad_norm": 0.4155310094356537,
      "learning_rate": 8.202434989105537e-06,
      "loss": 0.3348,
      "step": 19165
    },
    {
      "epoch": 0.879537423707035,
      "grad_norm": 0.43057844042778015,
      "learning_rate": 8.202246688540069e-06,
      "loss": 0.3603,
      "step": 19166
    },
    {
      "epoch": 0.8795833142122895,
      "grad_norm": 0.5113843679428101,
      "learning_rate": 8.202058380274207e-06,
      "loss": 0.4967,
      "step": 19167
    },
    {
      "epoch": 0.8796292047175439,
      "grad_norm": 0.42816320061683655,
      "learning_rate": 8.201870064308406e-06,
      "loss": 0.3787,
      "step": 19168
    },
    {
      "epoch": 0.8796750952227984,
      "grad_norm": 0.43738627433776855,
      "learning_rate": 8.20168174064312e-06,
      "loss": 0.3681,
      "step": 19169
    },
    {
      "epoch": 0.8797209857280529,
      "grad_norm": 0.488862007856369,
      "learning_rate": 8.201493409278799e-06,
      "loss": 0.446,
      "step": 19170
    },
    {
      "epoch": 0.8797668762333073,
      "grad_norm": 0.4588591158390045,
      "learning_rate": 8.201305070215895e-06,
      "loss": 0.385,
      "step": 19171
    },
    {
      "epoch": 0.8798127667385618,
      "grad_norm": 0.49046656489372253,
      "learning_rate": 8.201116723454864e-06,
      "loss": 0.4596,
      "step": 19172
    },
    {
      "epoch": 0.8798586572438163,
      "grad_norm": 0.42806512117385864,
      "learning_rate": 8.200928368996157e-06,
      "loss": 0.3537,
      "step": 19173
    },
    {
      "epoch": 0.8799045477490707,
      "grad_norm": 0.4339893162250519,
      "learning_rate": 8.200740006840228e-06,
      "loss": 0.3469,
      "step": 19174
    },
    {
      "epoch": 0.8799504382543252,
      "grad_norm": 0.4892531931400299,
      "learning_rate": 8.20055163698753e-06,
      "loss": 0.4743,
      "step": 19175
    },
    {
      "epoch": 0.8799963287595797,
      "grad_norm": 0.45419391989707947,
      "learning_rate": 8.200363259438513e-06,
      "loss": 0.3634,
      "step": 19176
    },
    {
      "epoch": 0.8800422192648341,
      "grad_norm": 0.4677983820438385,
      "learning_rate": 8.200174874193637e-06,
      "loss": 0.4083,
      "step": 19177
    },
    {
      "epoch": 0.8800881097700886,
      "grad_norm": 0.4166705310344696,
      "learning_rate": 8.199986481253346e-06,
      "loss": 0.3117,
      "step": 19178
    },
    {
      "epoch": 0.880134000275343,
      "grad_norm": 0.4390350878238678,
      "learning_rate": 8.199798080618097e-06,
      "loss": 0.3637,
      "step": 19179
    },
    {
      "epoch": 0.8801798907805974,
      "grad_norm": 0.46039873361587524,
      "learning_rate": 8.199609672288348e-06,
      "loss": 0.3936,
      "step": 19180
    },
    {
      "epoch": 0.8802257812858519,
      "grad_norm": 0.4135136604309082,
      "learning_rate": 8.199421256264543e-06,
      "loss": 0.3721,
      "step": 19181
    },
    {
      "epoch": 0.8802716717911064,
      "grad_norm": 0.4674646258354187,
      "learning_rate": 8.199232832547144e-06,
      "loss": 0.4342,
      "step": 19182
    },
    {
      "epoch": 0.8803175622963609,
      "grad_norm": 0.46967923641204834,
      "learning_rate": 8.199044401136597e-06,
      "loss": 0.4042,
      "step": 19183
    },
    {
      "epoch": 0.8803634528016153,
      "grad_norm": 0.4395766854286194,
      "learning_rate": 8.19885596203336e-06,
      "loss": 0.37,
      "step": 19184
    },
    {
      "epoch": 0.8804093433068698,
      "grad_norm": 0.426149845123291,
      "learning_rate": 8.19866751523788e-06,
      "loss": 0.3292,
      "step": 19185
    },
    {
      "epoch": 0.8804552338121243,
      "grad_norm": 0.43080008029937744,
      "learning_rate": 8.198479060750616e-06,
      "loss": 0.2938,
      "step": 19186
    },
    {
      "epoch": 0.8805011243173787,
      "grad_norm": 0.4458032250404358,
      "learning_rate": 8.19829059857202e-06,
      "loss": 0.3785,
      "step": 19187
    },
    {
      "epoch": 0.8805470148226332,
      "grad_norm": 0.5042188167572021,
      "learning_rate": 8.198102128702546e-06,
      "loss": 0.5322,
      "step": 19188
    },
    {
      "epoch": 0.8805929053278877,
      "grad_norm": 0.41657865047454834,
      "learning_rate": 8.197913651142643e-06,
      "loss": 0.3355,
      "step": 19189
    },
    {
      "epoch": 0.8806387958331421,
      "grad_norm": 0.4787415564060211,
      "learning_rate": 8.197725165892768e-06,
      "loss": 0.4486,
      "step": 19190
    },
    {
      "epoch": 0.8806846863383966,
      "grad_norm": 0.4416230618953705,
      "learning_rate": 8.197536672953374e-06,
      "loss": 0.3629,
      "step": 19191
    },
    {
      "epoch": 0.8807305768436511,
      "grad_norm": 0.42962655425071716,
      "learning_rate": 8.197348172324913e-06,
      "loss": 0.3845,
      "step": 19192
    },
    {
      "epoch": 0.8807764673489055,
      "grad_norm": 0.4953404366970062,
      "learning_rate": 8.197159664007838e-06,
      "loss": 0.4766,
      "step": 19193
    },
    {
      "epoch": 0.88082235785416,
      "grad_norm": 0.43361860513687134,
      "learning_rate": 8.196971148002604e-06,
      "loss": 0.3509,
      "step": 19194
    },
    {
      "epoch": 0.8808682483594145,
      "grad_norm": 0.43758490681648254,
      "learning_rate": 8.196782624309662e-06,
      "loss": 0.4059,
      "step": 19195
    },
    {
      "epoch": 0.8809141388646688,
      "grad_norm": 0.4441516101360321,
      "learning_rate": 8.19659409292947e-06,
      "loss": 0.4421,
      "step": 19196
    },
    {
      "epoch": 0.8809600293699233,
      "grad_norm": 0.49107858538627625,
      "learning_rate": 8.196405553862474e-06,
      "loss": 0.4542,
      "step": 19197
    },
    {
      "epoch": 0.8810059198751778,
      "grad_norm": 0.45045992732048035,
      "learning_rate": 8.196217007109133e-06,
      "loss": 0.3674,
      "step": 19198
    },
    {
      "epoch": 0.8810518103804323,
      "grad_norm": 0.41383570432662964,
      "learning_rate": 8.196028452669898e-06,
      "loss": 0.3509,
      "step": 19199
    },
    {
      "epoch": 0.8810977008856867,
      "grad_norm": 0.47147536277770996,
      "learning_rate": 8.195839890545225e-06,
      "loss": 0.4257,
      "step": 19200
    },
    {
      "epoch": 0.8811435913909412,
      "grad_norm": 0.43871334195137024,
      "learning_rate": 8.195651320735564e-06,
      "loss": 0.3667,
      "step": 19201
    },
    {
      "epoch": 0.8811894818961957,
      "grad_norm": 0.4312410056591034,
      "learning_rate": 8.195462743241373e-06,
      "loss": 0.3377,
      "step": 19202
    },
    {
      "epoch": 0.8812353724014501,
      "grad_norm": 0.4550964832305908,
      "learning_rate": 8.1952741580631e-06,
      "loss": 0.4604,
      "step": 19203
    },
    {
      "epoch": 0.8812812629067046,
      "grad_norm": 0.47329220175743103,
      "learning_rate": 8.195085565201202e-06,
      "loss": 0.4485,
      "step": 19204
    },
    {
      "epoch": 0.8813271534119591,
      "grad_norm": 0.43617936968803406,
      "learning_rate": 8.194896964656131e-06,
      "loss": 0.375,
      "step": 19205
    },
    {
      "epoch": 0.8813730439172135,
      "grad_norm": 0.44242623448371887,
      "learning_rate": 8.194708356428342e-06,
      "loss": 0.3909,
      "step": 19206
    },
    {
      "epoch": 0.881418934422468,
      "grad_norm": 0.45489150285720825,
      "learning_rate": 8.194519740518288e-06,
      "loss": 0.3533,
      "step": 19207
    },
    {
      "epoch": 0.8814648249277225,
      "grad_norm": 0.4922693073749542,
      "learning_rate": 8.194331116926422e-06,
      "loss": 0.4515,
      "step": 19208
    },
    {
      "epoch": 0.8815107154329769,
      "grad_norm": 0.47709330916404724,
      "learning_rate": 8.194142485653196e-06,
      "loss": 0.4943,
      "step": 19209
    },
    {
      "epoch": 0.8815566059382314,
      "grad_norm": 0.42918339371681213,
      "learning_rate": 8.193953846699067e-06,
      "loss": 0.3616,
      "step": 19210
    },
    {
      "epoch": 0.8816024964434859,
      "grad_norm": 0.43919846415519714,
      "learning_rate": 8.193765200064488e-06,
      "loss": 0.3529,
      "step": 19211
    },
    {
      "epoch": 0.8816483869487403,
      "grad_norm": 0.4629294276237488,
      "learning_rate": 8.19357654574991e-06,
      "loss": 0.3806,
      "step": 19212
    },
    {
      "epoch": 0.8816942774539948,
      "grad_norm": 0.4489581882953644,
      "learning_rate": 8.193387883755788e-06,
      "loss": 0.3347,
      "step": 19213
    },
    {
      "epoch": 0.8817401679592493,
      "grad_norm": 0.5061411261558533,
      "learning_rate": 8.193199214082579e-06,
      "loss": 0.4636,
      "step": 19214
    },
    {
      "epoch": 0.8817860584645036,
      "grad_norm": 0.46993300318717957,
      "learning_rate": 8.19301053673073e-06,
      "loss": 0.4459,
      "step": 19215
    },
    {
      "epoch": 0.8818319489697581,
      "grad_norm": 0.45036575198173523,
      "learning_rate": 8.1928218517007e-06,
      "loss": 0.4146,
      "step": 19216
    },
    {
      "epoch": 0.8818778394750126,
      "grad_norm": 0.47681519389152527,
      "learning_rate": 8.192633158992941e-06,
      "loss": 0.4183,
      "step": 19217
    },
    {
      "epoch": 0.8819237299802671,
      "grad_norm": 0.4516066312789917,
      "learning_rate": 8.192444458607907e-06,
      "loss": 0.3764,
      "step": 19218
    },
    {
      "epoch": 0.8819696204855215,
      "grad_norm": 0.4285201132297516,
      "learning_rate": 8.192255750546052e-06,
      "loss": 0.3258,
      "step": 19219
    },
    {
      "epoch": 0.882015510990776,
      "grad_norm": 0.47717711329460144,
      "learning_rate": 8.192067034807828e-06,
      "loss": 0.3776,
      "step": 19220
    },
    {
      "epoch": 0.8820614014960305,
      "grad_norm": 0.44581788778305054,
      "learning_rate": 8.191878311393693e-06,
      "loss": 0.3441,
      "step": 19221
    },
    {
      "epoch": 0.8821072920012849,
      "grad_norm": 0.4888061583042145,
      "learning_rate": 8.191689580304096e-06,
      "loss": 0.4792,
      "step": 19222
    },
    {
      "epoch": 0.8821531825065394,
      "grad_norm": 0.44188132882118225,
      "learning_rate": 8.191500841539492e-06,
      "loss": 0.3679,
      "step": 19223
    },
    {
      "epoch": 0.8821990730117939,
      "grad_norm": 0.4308532178401947,
      "learning_rate": 8.191312095100336e-06,
      "loss": 0.3444,
      "step": 19224
    },
    {
      "epoch": 0.8822449635170483,
      "grad_norm": 0.48385700583457947,
      "learning_rate": 8.191123340987084e-06,
      "loss": 0.4421,
      "step": 19225
    },
    {
      "epoch": 0.8822908540223028,
      "grad_norm": 0.4256320893764496,
      "learning_rate": 8.190934579200187e-06,
      "loss": 0.3103,
      "step": 19226
    },
    {
      "epoch": 0.8823367445275573,
      "grad_norm": 0.47393524646759033,
      "learning_rate": 8.190745809740097e-06,
      "loss": 0.398,
      "step": 19227
    },
    {
      "epoch": 0.8823826350328117,
      "grad_norm": 0.4953345060348511,
      "learning_rate": 8.190557032607272e-06,
      "loss": 0.4916,
      "step": 19228
    },
    {
      "epoch": 0.8824285255380662,
      "grad_norm": 0.5270404815673828,
      "learning_rate": 8.190368247802165e-06,
      "loss": 0.5274,
      "step": 19229
    },
    {
      "epoch": 0.8824744160433207,
      "grad_norm": 0.4783361852169037,
      "learning_rate": 8.190179455325228e-06,
      "loss": 0.3336,
      "step": 19230
    },
    {
      "epoch": 0.882520306548575,
      "grad_norm": 0.6263243556022644,
      "learning_rate": 8.189990655176916e-06,
      "loss": 0.3912,
      "step": 19231
    },
    {
      "epoch": 0.8825661970538295,
      "grad_norm": 0.4723162055015564,
      "learning_rate": 8.189801847357685e-06,
      "loss": 0.4188,
      "step": 19232
    },
    {
      "epoch": 0.882612087559084,
      "grad_norm": 0.47945454716682434,
      "learning_rate": 8.189613031867986e-06,
      "loss": 0.4809,
      "step": 19233
    },
    {
      "epoch": 0.8826579780643384,
      "grad_norm": 0.4349053204059601,
      "learning_rate": 8.189424208708275e-06,
      "loss": 0.3721,
      "step": 19234
    },
    {
      "epoch": 0.8827038685695929,
      "grad_norm": 0.46414634585380554,
      "learning_rate": 8.189235377879004e-06,
      "loss": 0.3714,
      "step": 19235
    },
    {
      "epoch": 0.8827497590748474,
      "grad_norm": 0.5051312446594238,
      "learning_rate": 8.18904653938063e-06,
      "loss": 0.4797,
      "step": 19236
    },
    {
      "epoch": 0.8827956495801019,
      "grad_norm": 0.4527251422405243,
      "learning_rate": 8.188857693213605e-06,
      "loss": 0.4054,
      "step": 19237
    },
    {
      "epoch": 0.8828415400853563,
      "grad_norm": 0.4478316009044647,
      "learning_rate": 8.188668839378383e-06,
      "loss": 0.3223,
      "step": 19238
    },
    {
      "epoch": 0.8828874305906108,
      "grad_norm": 0.40648671984672546,
      "learning_rate": 8.188479977875421e-06,
      "loss": 0.2897,
      "step": 19239
    },
    {
      "epoch": 0.8829333210958653,
      "grad_norm": 0.4584740996360779,
      "learning_rate": 8.188291108705169e-06,
      "loss": 0.3973,
      "step": 19240
    },
    {
      "epoch": 0.8829792116011197,
      "grad_norm": 0.4096563458442688,
      "learning_rate": 8.188102231868082e-06,
      "loss": 0.3115,
      "step": 19241
    },
    {
      "epoch": 0.8830251021063742,
      "grad_norm": 0.4632634222507477,
      "learning_rate": 8.187913347364617e-06,
      "loss": 0.3664,
      "step": 19242
    },
    {
      "epoch": 0.8830709926116287,
      "grad_norm": 0.44173023104667664,
      "learning_rate": 8.187724455195227e-06,
      "loss": 0.3757,
      "step": 19243
    },
    {
      "epoch": 0.8831168831168831,
      "grad_norm": 0.4560413956642151,
      "learning_rate": 8.187535555360365e-06,
      "loss": 0.403,
      "step": 19244
    },
    {
      "epoch": 0.8831627736221376,
      "grad_norm": 0.44466885924339294,
      "learning_rate": 8.187346647860486e-06,
      "loss": 0.3713,
      "step": 19245
    },
    {
      "epoch": 0.8832086641273921,
      "grad_norm": 0.4600166976451874,
      "learning_rate": 8.187157732696045e-06,
      "loss": 0.3456,
      "step": 19246
    },
    {
      "epoch": 0.8832545546326465,
      "grad_norm": 0.4309530258178711,
      "learning_rate": 8.186968809867495e-06,
      "loss": 0.4007,
      "step": 19247
    },
    {
      "epoch": 0.883300445137901,
      "grad_norm": 0.44089317321777344,
      "learning_rate": 8.186779879375291e-06,
      "loss": 0.3504,
      "step": 19248
    },
    {
      "epoch": 0.8833463356431555,
      "grad_norm": 0.4826483428478241,
      "learning_rate": 8.186590941219887e-06,
      "loss": 0.4684,
      "step": 19249
    },
    {
      "epoch": 0.8833922261484098,
      "grad_norm": 0.45807716250419617,
      "learning_rate": 8.186401995401738e-06,
      "loss": 0.4353,
      "step": 19250
    },
    {
      "epoch": 0.8834381166536643,
      "grad_norm": 0.5067712664604187,
      "learning_rate": 8.186213041921299e-06,
      "loss": 0.4752,
      "step": 19251
    },
    {
      "epoch": 0.8834840071589188,
      "grad_norm": 0.4552050530910492,
      "learning_rate": 8.18602408077902e-06,
      "loss": 0.4034,
      "step": 19252
    },
    {
      "epoch": 0.8835298976641733,
      "grad_norm": 0.45308005809783936,
      "learning_rate": 8.18583511197536e-06,
      "loss": 0.3748,
      "step": 19253
    },
    {
      "epoch": 0.8835757881694277,
      "grad_norm": 0.43959176540374756,
      "learning_rate": 8.185646135510771e-06,
      "loss": 0.381,
      "step": 19254
    },
    {
      "epoch": 0.8836216786746822,
      "grad_norm": 0.4747507870197296,
      "learning_rate": 8.18545715138571e-06,
      "loss": 0.4829,
      "step": 19255
    },
    {
      "epoch": 0.8836675691799367,
      "grad_norm": 0.4480806291103363,
      "learning_rate": 8.18526815960063e-06,
      "loss": 0.3869,
      "step": 19256
    },
    {
      "epoch": 0.8837134596851911,
      "grad_norm": 0.43594253063201904,
      "learning_rate": 8.185079160155987e-06,
      "loss": 0.3535,
      "step": 19257
    },
    {
      "epoch": 0.8837593501904456,
      "grad_norm": 0.5030718445777893,
      "learning_rate": 8.18489015305223e-06,
      "loss": 0.5101,
      "step": 19258
    },
    {
      "epoch": 0.8838052406957001,
      "grad_norm": 0.4753264784812927,
      "learning_rate": 8.184701138289822e-06,
      "loss": 0.4829,
      "step": 19259
    },
    {
      "epoch": 0.8838511312009545,
      "grad_norm": 0.4426402449607849,
      "learning_rate": 8.184512115869207e-06,
      "loss": 0.3486,
      "step": 19260
    },
    {
      "epoch": 0.883897021706209,
      "grad_norm": 0.47613900899887085,
      "learning_rate": 8.18432308579085e-06,
      "loss": 0.4292,
      "step": 19261
    },
    {
      "epoch": 0.8839429122114635,
      "grad_norm": 0.44666293263435364,
      "learning_rate": 8.1841340480552e-06,
      "loss": 0.3956,
      "step": 19262
    },
    {
      "epoch": 0.8839888027167179,
      "grad_norm": 0.42694327235221863,
      "learning_rate": 8.18394500266271e-06,
      "loss": 0.359,
      "step": 19263
    },
    {
      "epoch": 0.8840346932219724,
      "grad_norm": 0.46381524205207825,
      "learning_rate": 8.183755949613841e-06,
      "loss": 0.3994,
      "step": 19264
    },
    {
      "epoch": 0.8840805837272269,
      "grad_norm": 0.471210241317749,
      "learning_rate": 8.183566888909043e-06,
      "loss": 0.4526,
      "step": 19265
    },
    {
      "epoch": 0.8841264742324813,
      "grad_norm": 0.448869526386261,
      "learning_rate": 8.183377820548767e-06,
      "loss": 0.4297,
      "step": 19266
    },
    {
      "epoch": 0.8841723647377357,
      "grad_norm": 0.43165886402130127,
      "learning_rate": 8.183188744533477e-06,
      "loss": 0.3601,
      "step": 19267
    },
    {
      "epoch": 0.8842182552429902,
      "grad_norm": 0.4133877456188202,
      "learning_rate": 8.182999660863622e-06,
      "loss": 0.3198,
      "step": 19268
    },
    {
      "epoch": 0.8842641457482446,
      "grad_norm": 0.4343601167201996,
      "learning_rate": 8.182810569539656e-06,
      "loss": 0.3786,
      "step": 19269
    },
    {
      "epoch": 0.8843100362534991,
      "grad_norm": 0.45204848051071167,
      "learning_rate": 8.182621470562035e-06,
      "loss": 0.3613,
      "step": 19270
    },
    {
      "epoch": 0.8843559267587536,
      "grad_norm": 0.44887587428092957,
      "learning_rate": 8.182432363931215e-06,
      "loss": 0.3869,
      "step": 19271
    },
    {
      "epoch": 0.8844018172640081,
      "grad_norm": 0.4319516122341156,
      "learning_rate": 8.18224324964765e-06,
      "loss": 0.395,
      "step": 19272
    },
    {
      "epoch": 0.8844477077692625,
      "grad_norm": 0.4634430408477783,
      "learning_rate": 8.182054127711794e-06,
      "loss": 0.4075,
      "step": 19273
    },
    {
      "epoch": 0.884493598274517,
      "grad_norm": 0.4933769404888153,
      "learning_rate": 8.181864998124101e-06,
      "loss": 0.528,
      "step": 19274
    },
    {
      "epoch": 0.8845394887797715,
      "grad_norm": 0.47723472118377686,
      "learning_rate": 8.181675860885026e-06,
      "loss": 0.3863,
      "step": 19275
    },
    {
      "epoch": 0.8845853792850259,
      "grad_norm": 0.4558815062046051,
      "learning_rate": 8.181486715995027e-06,
      "loss": 0.3617,
      "step": 19276
    },
    {
      "epoch": 0.8846312697902804,
      "grad_norm": 0.4399058520793915,
      "learning_rate": 8.181297563454556e-06,
      "loss": 0.3429,
      "step": 19277
    },
    {
      "epoch": 0.8846771602955349,
      "grad_norm": 0.4488939642906189,
      "learning_rate": 8.181108403264067e-06,
      "loss": 0.3535,
      "step": 19278
    },
    {
      "epoch": 0.8847230508007893,
      "grad_norm": 0.4201013147830963,
      "learning_rate": 8.180919235424019e-06,
      "loss": 0.3529,
      "step": 19279
    },
    {
      "epoch": 0.8847689413060438,
      "grad_norm": 0.4374241530895233,
      "learning_rate": 8.180730059934862e-06,
      "loss": 0.3702,
      "step": 19280
    },
    {
      "epoch": 0.8848148318112983,
      "grad_norm": 0.4302213788032532,
      "learning_rate": 8.180540876797053e-06,
      "loss": 0.344,
      "step": 19281
    },
    {
      "epoch": 0.8848607223165527,
      "grad_norm": 0.42838823795318604,
      "learning_rate": 8.180351686011047e-06,
      "loss": 0.3085,
      "step": 19282
    },
    {
      "epoch": 0.8849066128218072,
      "grad_norm": 0.42228570580482483,
      "learning_rate": 8.1801624875773e-06,
      "loss": 0.3306,
      "step": 19283
    },
    {
      "epoch": 0.8849525033270617,
      "grad_norm": 0.44273045659065247,
      "learning_rate": 8.179973281496264e-06,
      "loss": 0.4385,
      "step": 19284
    },
    {
      "epoch": 0.884998393832316,
      "grad_norm": 0.5049851536750793,
      "learning_rate": 8.1797840677684e-06,
      "loss": 0.5346,
      "step": 19285
    },
    {
      "epoch": 0.8850442843375705,
      "grad_norm": 0.43154647946357727,
      "learning_rate": 8.179594846394155e-06,
      "loss": 0.3589,
      "step": 19286
    },
    {
      "epoch": 0.885090174842825,
      "grad_norm": 0.397867351770401,
      "learning_rate": 8.17940561737399e-06,
      "loss": 0.2926,
      "step": 19287
    },
    {
      "epoch": 0.8851360653480795,
      "grad_norm": 0.4596880376338959,
      "learning_rate": 8.179216380708357e-06,
      "loss": 0.4383,
      "step": 19288
    },
    {
      "epoch": 0.8851819558533339,
      "grad_norm": 0.47491440176963806,
      "learning_rate": 8.179027136397712e-06,
      "loss": 0.4078,
      "step": 19289
    },
    {
      "epoch": 0.8852278463585884,
      "grad_norm": 0.4638192057609558,
      "learning_rate": 8.17883788444251e-06,
      "loss": 0.4266,
      "step": 19290
    },
    {
      "epoch": 0.8852737368638429,
      "grad_norm": 0.43218016624450684,
      "learning_rate": 8.178648624843206e-06,
      "loss": 0.3714,
      "step": 19291
    },
    {
      "epoch": 0.8853196273690973,
      "grad_norm": 0.47718334197998047,
      "learning_rate": 8.178459357600255e-06,
      "loss": 0.413,
      "step": 19292
    },
    {
      "epoch": 0.8853655178743518,
      "grad_norm": 0.44189852476119995,
      "learning_rate": 8.178270082714113e-06,
      "loss": 0.3808,
      "step": 19293
    },
    {
      "epoch": 0.8854114083796063,
      "grad_norm": 0.4517497718334198,
      "learning_rate": 8.178080800185235e-06,
      "loss": 0.372,
      "step": 19294
    },
    {
      "epoch": 0.8854572988848607,
      "grad_norm": 0.48035383224487305,
      "learning_rate": 8.177891510014076e-06,
      "loss": 0.3876,
      "step": 19295
    },
    {
      "epoch": 0.8855031893901152,
      "grad_norm": 0.4711470603942871,
      "learning_rate": 8.17770221220109e-06,
      "loss": 0.4336,
      "step": 19296
    },
    {
      "epoch": 0.8855490798953697,
      "grad_norm": 0.4574022591114044,
      "learning_rate": 8.177512906746732e-06,
      "loss": 0.396,
      "step": 19297
    },
    {
      "epoch": 0.8855949704006241,
      "grad_norm": 0.41939762234687805,
      "learning_rate": 8.177323593651459e-06,
      "loss": 0.3193,
      "step": 19298
    },
    {
      "epoch": 0.8856408609058786,
      "grad_norm": 0.46054959297180176,
      "learning_rate": 8.177134272915728e-06,
      "loss": 0.431,
      "step": 19299
    },
    {
      "epoch": 0.8856867514111331,
      "grad_norm": 0.45045262575149536,
      "learning_rate": 8.176944944539989e-06,
      "loss": 0.3818,
      "step": 19300
    },
    {
      "epoch": 0.8857326419163875,
      "grad_norm": 0.45295828580856323,
      "learning_rate": 8.1767556085247e-06,
      "loss": 0.3919,
      "step": 19301
    },
    {
      "epoch": 0.885778532421642,
      "grad_norm": 0.45955708622932434,
      "learning_rate": 8.176566264870319e-06,
      "loss": 0.4432,
      "step": 19302
    },
    {
      "epoch": 0.8858244229268964,
      "grad_norm": 0.4525637924671173,
      "learning_rate": 8.176376913577297e-06,
      "loss": 0.3716,
      "step": 19303
    },
    {
      "epoch": 0.8858703134321508,
      "grad_norm": 0.4529217779636383,
      "learning_rate": 8.17618755464609e-06,
      "loss": 0.4023,
      "step": 19304
    },
    {
      "epoch": 0.8859162039374053,
      "grad_norm": 0.4236811101436615,
      "learning_rate": 8.175998188077156e-06,
      "loss": 0.3336,
      "step": 19305
    },
    {
      "epoch": 0.8859620944426598,
      "grad_norm": 0.431275874376297,
      "learning_rate": 8.175808813870947e-06,
      "loss": 0.3596,
      "step": 19306
    },
    {
      "epoch": 0.8860079849479143,
      "grad_norm": 0.45622673630714417,
      "learning_rate": 8.175619432027922e-06,
      "loss": 0.3928,
      "step": 19307
    },
    {
      "epoch": 0.8860538754531687,
      "grad_norm": 0.4416138231754303,
      "learning_rate": 8.175430042548533e-06,
      "loss": 0.3979,
      "step": 19308
    },
    {
      "epoch": 0.8860997659584232,
      "grad_norm": 0.44979560375213623,
      "learning_rate": 8.175240645433238e-06,
      "loss": 0.3815,
      "step": 19309
    },
    {
      "epoch": 0.8861456564636777,
      "grad_norm": 0.48101869225502014,
      "learning_rate": 8.175051240682493e-06,
      "loss": 0.4039,
      "step": 19310
    },
    {
      "epoch": 0.8861915469689321,
      "grad_norm": 0.44577956199645996,
      "learning_rate": 8.174861828296748e-06,
      "loss": 0.3847,
      "step": 19311
    },
    {
      "epoch": 0.8862374374741866,
      "grad_norm": 0.44522202014923096,
      "learning_rate": 8.174672408276464e-06,
      "loss": 0.3657,
      "step": 19312
    },
    {
      "epoch": 0.8862833279794411,
      "grad_norm": 0.4311541020870209,
      "learning_rate": 8.174482980622096e-06,
      "loss": 0.3283,
      "step": 19313
    },
    {
      "epoch": 0.8863292184846955,
      "grad_norm": 0.49103447794914246,
      "learning_rate": 8.174293545334097e-06,
      "loss": 0.4119,
      "step": 19314
    },
    {
      "epoch": 0.88637510898995,
      "grad_norm": 0.5373789072036743,
      "learning_rate": 8.174104102412924e-06,
      "loss": 0.3826,
      "step": 19315
    },
    {
      "epoch": 0.8864209994952045,
      "grad_norm": 0.45798683166503906,
      "learning_rate": 8.173914651859033e-06,
      "loss": 0.3872,
      "step": 19316
    },
    {
      "epoch": 0.8864668900004589,
      "grad_norm": 0.46065694093704224,
      "learning_rate": 8.173725193672878e-06,
      "loss": 0.4294,
      "step": 19317
    },
    {
      "epoch": 0.8865127805057134,
      "grad_norm": 0.42420560121536255,
      "learning_rate": 8.173535727854917e-06,
      "loss": 0.3112,
      "step": 19318
    },
    {
      "epoch": 0.8865586710109679,
      "grad_norm": 0.43694329261779785,
      "learning_rate": 8.173346254405603e-06,
      "loss": 0.3574,
      "step": 19319
    },
    {
      "epoch": 0.8866045615162222,
      "grad_norm": 0.42598527669906616,
      "learning_rate": 8.173156773325392e-06,
      "loss": 0.3653,
      "step": 19320
    },
    {
      "epoch": 0.8866504520214767,
      "grad_norm": 0.43212082982063293,
      "learning_rate": 8.172967284614743e-06,
      "loss": 0.3699,
      "step": 19321
    },
    {
      "epoch": 0.8866963425267312,
      "grad_norm": 0.4443436563014984,
      "learning_rate": 8.172777788274109e-06,
      "loss": 0.4238,
      "step": 19322
    },
    {
      "epoch": 0.8867422330319856,
      "grad_norm": 0.4680856168270111,
      "learning_rate": 8.172588284303943e-06,
      "loss": 0.4438,
      "step": 19323
    },
    {
      "epoch": 0.8867881235372401,
      "grad_norm": 0.47539687156677246,
      "learning_rate": 8.172398772704704e-06,
      "loss": 0.4193,
      "step": 19324
    },
    {
      "epoch": 0.8868340140424946,
      "grad_norm": 0.4608498513698578,
      "learning_rate": 8.172209253476848e-06,
      "loss": 0.3983,
      "step": 19325
    },
    {
      "epoch": 0.8868799045477491,
      "grad_norm": 0.4740413427352905,
      "learning_rate": 8.172019726620832e-06,
      "loss": 0.4472,
      "step": 19326
    },
    {
      "epoch": 0.8869257950530035,
      "grad_norm": 0.44513195753097534,
      "learning_rate": 8.171830192137107e-06,
      "loss": 0.4111,
      "step": 19327
    },
    {
      "epoch": 0.886971685558258,
      "grad_norm": 0.44581103324890137,
      "learning_rate": 8.17164065002613e-06,
      "loss": 0.3915,
      "step": 19328
    },
    {
      "epoch": 0.8870175760635125,
      "grad_norm": 0.4642084240913391,
      "learning_rate": 8.171451100288363e-06,
      "loss": 0.3842,
      "step": 19329
    },
    {
      "epoch": 0.8870634665687669,
      "grad_norm": 0.4672383666038513,
      "learning_rate": 8.171261542924254e-06,
      "loss": 0.4206,
      "step": 19330
    },
    {
      "epoch": 0.8871093570740214,
      "grad_norm": 0.4160286486148834,
      "learning_rate": 8.171071977934262e-06,
      "loss": 0.3557,
      "step": 19331
    },
    {
      "epoch": 0.8871552475792759,
      "grad_norm": 0.43047216534614563,
      "learning_rate": 8.170882405318842e-06,
      "loss": 0.3295,
      "step": 19332
    },
    {
      "epoch": 0.8872011380845303,
      "grad_norm": 0.45906203985214233,
      "learning_rate": 8.170692825078452e-06,
      "loss": 0.4172,
      "step": 19333
    },
    {
      "epoch": 0.8872470285897848,
      "grad_norm": 0.4445382356643677,
      "learning_rate": 8.170503237213546e-06,
      "loss": 0.3602,
      "step": 19334
    },
    {
      "epoch": 0.8872929190950393,
      "grad_norm": 0.41604000329971313,
      "learning_rate": 8.170313641724581e-06,
      "loss": 0.3324,
      "step": 19335
    },
    {
      "epoch": 0.8873388096002937,
      "grad_norm": 0.42382022738456726,
      "learning_rate": 8.170124038612011e-06,
      "loss": 0.3639,
      "step": 19336
    },
    {
      "epoch": 0.8873847001055482,
      "grad_norm": 0.41887417435646057,
      "learning_rate": 8.169934427876295e-06,
      "loss": 0.3163,
      "step": 19337
    },
    {
      "epoch": 0.8874305906108026,
      "grad_norm": 0.4523906707763672,
      "learning_rate": 8.169744809517886e-06,
      "loss": 0.3908,
      "step": 19338
    },
    {
      "epoch": 0.887476481116057,
      "grad_norm": 0.4558180272579193,
      "learning_rate": 8.169555183537241e-06,
      "loss": 0.4401,
      "step": 19339
    },
    {
      "epoch": 0.8875223716213115,
      "grad_norm": 0.426932156085968,
      "learning_rate": 8.169365549934816e-06,
      "loss": 0.337,
      "step": 19340
    },
    {
      "epoch": 0.887568262126566,
      "grad_norm": 0.47371241450309753,
      "learning_rate": 8.169175908711067e-06,
      "loss": 0.4917,
      "step": 19341
    },
    {
      "epoch": 0.8876141526318205,
      "grad_norm": 0.4252172112464905,
      "learning_rate": 8.168986259866451e-06,
      "loss": 0.3109,
      "step": 19342
    },
    {
      "epoch": 0.8876600431370749,
      "grad_norm": 0.44239091873168945,
      "learning_rate": 8.168796603401423e-06,
      "loss": 0.4288,
      "step": 19343
    },
    {
      "epoch": 0.8877059336423294,
      "grad_norm": 0.42298561334609985,
      "learning_rate": 8.16860693931644e-06,
      "loss": 0.3796,
      "step": 19344
    },
    {
      "epoch": 0.8877518241475839,
      "grad_norm": 0.46961742639541626,
      "learning_rate": 8.168417267611955e-06,
      "loss": 0.4316,
      "step": 19345
    },
    {
      "epoch": 0.8877977146528383,
      "grad_norm": 0.49516335129737854,
      "learning_rate": 8.168227588288429e-06,
      "loss": 0.5119,
      "step": 19346
    },
    {
      "epoch": 0.8878436051580928,
      "grad_norm": 0.6207625269889832,
      "learning_rate": 8.168037901346314e-06,
      "loss": 0.4027,
      "step": 19347
    },
    {
      "epoch": 0.8878894956633473,
      "grad_norm": 0.42306700348854065,
      "learning_rate": 8.167848206786068e-06,
      "loss": 0.3537,
      "step": 19348
    },
    {
      "epoch": 0.8879353861686017,
      "grad_norm": 0.4330935776233673,
      "learning_rate": 8.167658504608147e-06,
      "loss": 0.3604,
      "step": 19349
    },
    {
      "epoch": 0.8879812766738562,
      "grad_norm": 0.47012266516685486,
      "learning_rate": 8.167468794813007e-06,
      "loss": 0.4203,
      "step": 19350
    },
    {
      "epoch": 0.8880271671791107,
      "grad_norm": 0.4753623604774475,
      "learning_rate": 8.167279077401105e-06,
      "loss": 0.4585,
      "step": 19351
    },
    {
      "epoch": 0.8880730576843651,
      "grad_norm": 0.44926005601882935,
      "learning_rate": 8.167089352372896e-06,
      "loss": 0.4219,
      "step": 19352
    },
    {
      "epoch": 0.8881189481896196,
      "grad_norm": 0.4491520822048187,
      "learning_rate": 8.166899619728837e-06,
      "loss": 0.402,
      "step": 19353
    },
    {
      "epoch": 0.8881648386948741,
      "grad_norm": 0.45438364148139954,
      "learning_rate": 8.166709879469382e-06,
      "loss": 0.4053,
      "step": 19354
    },
    {
      "epoch": 0.8882107292001284,
      "grad_norm": 0.47404855489730835,
      "learning_rate": 8.16652013159499e-06,
      "loss": 0.4175,
      "step": 19355
    },
    {
      "epoch": 0.8882566197053829,
      "grad_norm": 0.4858325719833374,
      "learning_rate": 8.166330376106117e-06,
      "loss": 0.4552,
      "step": 19356
    },
    {
      "epoch": 0.8883025102106374,
      "grad_norm": 0.43599143624305725,
      "learning_rate": 8.16614061300322e-06,
      "loss": 0.3832,
      "step": 19357
    },
    {
      "epoch": 0.8883484007158918,
      "grad_norm": 0.4565216898918152,
      "learning_rate": 8.165950842286752e-06,
      "loss": 0.4016,
      "step": 19358
    },
    {
      "epoch": 0.8883942912211463,
      "grad_norm": 0.4756777584552765,
      "learning_rate": 8.165761063957173e-06,
      "loss": 0.4104,
      "step": 19359
    },
    {
      "epoch": 0.8884401817264008,
      "grad_norm": 0.43310877680778503,
      "learning_rate": 8.165571278014937e-06,
      "loss": 0.3608,
      "step": 19360
    },
    {
      "epoch": 0.8884860722316553,
      "grad_norm": 0.46654167771339417,
      "learning_rate": 8.1653814844605e-06,
      "loss": 0.3933,
      "step": 19361
    },
    {
      "epoch": 0.8885319627369097,
      "grad_norm": 0.4520196318626404,
      "learning_rate": 8.165191683294321e-06,
      "loss": 0.4361,
      "step": 19362
    },
    {
      "epoch": 0.8885778532421642,
      "grad_norm": 0.450872540473938,
      "learning_rate": 8.165001874516854e-06,
      "loss": 0.4019,
      "step": 19363
    },
    {
      "epoch": 0.8886237437474187,
      "grad_norm": 0.4551583528518677,
      "learning_rate": 8.164812058128557e-06,
      "loss": 0.4125,
      "step": 19364
    },
    {
      "epoch": 0.8886696342526731,
      "grad_norm": 0.4707236886024475,
      "learning_rate": 8.164622234129886e-06,
      "loss": 0.4166,
      "step": 19365
    },
    {
      "epoch": 0.8887155247579276,
      "grad_norm": 0.42759400606155396,
      "learning_rate": 8.164432402521299e-06,
      "loss": 0.3469,
      "step": 19366
    },
    {
      "epoch": 0.8887614152631821,
      "grad_norm": 0.4595560133457184,
      "learning_rate": 8.164242563303248e-06,
      "loss": 0.3966,
      "step": 19367
    },
    {
      "epoch": 0.8888073057684365,
      "grad_norm": 0.4359815716743469,
      "learning_rate": 8.164052716476194e-06,
      "loss": 0.3524,
      "step": 19368
    },
    {
      "epoch": 0.888853196273691,
      "grad_norm": 0.5280541181564331,
      "learning_rate": 8.163862862040592e-06,
      "loss": 0.504,
      "step": 19369
    },
    {
      "epoch": 0.8888990867789455,
      "grad_norm": 0.47906383872032166,
      "learning_rate": 8.163672999996897e-06,
      "loss": 0.4936,
      "step": 19370
    },
    {
      "epoch": 0.8889449772841999,
      "grad_norm": 0.47506964206695557,
      "learning_rate": 8.163483130345566e-06,
      "loss": 0.3906,
      "step": 19371
    },
    {
      "epoch": 0.8889908677894544,
      "grad_norm": 0.45775508880615234,
      "learning_rate": 8.163293253087059e-06,
      "loss": 0.3548,
      "step": 19372
    },
    {
      "epoch": 0.8890367582947089,
      "grad_norm": 0.4755198061466217,
      "learning_rate": 8.16310336822183e-06,
      "loss": 0.441,
      "step": 19373
    },
    {
      "epoch": 0.8890826487999632,
      "grad_norm": 0.5096787214279175,
      "learning_rate": 8.162913475750334e-06,
      "loss": 0.588,
      "step": 19374
    },
    {
      "epoch": 0.8891285393052177,
      "grad_norm": 0.4467904269695282,
      "learning_rate": 8.16272357567303e-06,
      "loss": 0.4172,
      "step": 19375
    },
    {
      "epoch": 0.8891744298104722,
      "grad_norm": 0.4519994854927063,
      "learning_rate": 8.162533667990373e-06,
      "loss": 0.4157,
      "step": 19376
    },
    {
      "epoch": 0.8892203203157267,
      "grad_norm": 0.4626026451587677,
      "learning_rate": 8.162343752702822e-06,
      "loss": 0.447,
      "step": 19377
    },
    {
      "epoch": 0.8892662108209811,
      "grad_norm": 0.46387556195259094,
      "learning_rate": 8.162153829810832e-06,
      "loss": 0.3745,
      "step": 19378
    },
    {
      "epoch": 0.8893121013262356,
      "grad_norm": 0.47166651487350464,
      "learning_rate": 8.161963899314861e-06,
      "loss": 0.389,
      "step": 19379
    },
    {
      "epoch": 0.8893579918314901,
      "grad_norm": 0.46069812774658203,
      "learning_rate": 8.161773961215362e-06,
      "loss": 0.4462,
      "step": 19380
    },
    {
      "epoch": 0.8894038823367445,
      "grad_norm": 0.45364218950271606,
      "learning_rate": 8.161584015512798e-06,
      "loss": 0.4048,
      "step": 19381
    },
    {
      "epoch": 0.889449772841999,
      "grad_norm": 0.48208382725715637,
      "learning_rate": 8.16139406220762e-06,
      "loss": 0.4026,
      "step": 19382
    },
    {
      "epoch": 0.8894956633472535,
      "grad_norm": 0.4571487307548523,
      "learning_rate": 8.161204101300289e-06,
      "loss": 0.3607,
      "step": 19383
    },
    {
      "epoch": 0.8895415538525079,
      "grad_norm": 0.4381783902645111,
      "learning_rate": 8.161014132791258e-06,
      "loss": 0.3745,
      "step": 19384
    },
    {
      "epoch": 0.8895874443577624,
      "grad_norm": 0.44830700755119324,
      "learning_rate": 8.160824156680987e-06,
      "loss": 0.4434,
      "step": 19385
    },
    {
      "epoch": 0.8896333348630169,
      "grad_norm": 0.44642579555511475,
      "learning_rate": 8.16063417296993e-06,
      "loss": 0.3977,
      "step": 19386
    },
    {
      "epoch": 0.8896792253682713,
      "grad_norm": 0.6056147217750549,
      "learning_rate": 8.160444181658546e-06,
      "loss": 0.3896,
      "step": 19387
    },
    {
      "epoch": 0.8897251158735258,
      "grad_norm": 0.4642273187637329,
      "learning_rate": 8.16025418274729e-06,
      "loss": 0.4185,
      "step": 19388
    },
    {
      "epoch": 0.8897710063787803,
      "grad_norm": 0.46919429302215576,
      "learning_rate": 8.160064176236623e-06,
      "loss": 0.377,
      "step": 19389
    },
    {
      "epoch": 0.8898168968840346,
      "grad_norm": 0.41915735602378845,
      "learning_rate": 8.159874162126997e-06,
      "loss": 0.3309,
      "step": 19390
    },
    {
      "epoch": 0.8898627873892891,
      "grad_norm": 0.4426107406616211,
      "learning_rate": 8.159684140418871e-06,
      "loss": 0.3477,
      "step": 19391
    },
    {
      "epoch": 0.8899086778945436,
      "grad_norm": 0.45022663474082947,
      "learning_rate": 8.159494111112702e-06,
      "loss": 0.3971,
      "step": 19392
    },
    {
      "epoch": 0.889954568399798,
      "grad_norm": 0.4152921736240387,
      "learning_rate": 8.159304074208947e-06,
      "loss": 0.3506,
      "step": 19393
    },
    {
      "epoch": 0.8900004589050525,
      "grad_norm": 0.46691566705703735,
      "learning_rate": 8.159114029708062e-06,
      "loss": 0.4117,
      "step": 19394
    },
    {
      "epoch": 0.890046349410307,
      "grad_norm": 0.47423386573791504,
      "learning_rate": 8.158923977610507e-06,
      "loss": 0.4652,
      "step": 19395
    },
    {
      "epoch": 0.8900922399155615,
      "grad_norm": 0.4628556966781616,
      "learning_rate": 8.158733917916734e-06,
      "loss": 0.3962,
      "step": 19396
    },
    {
      "epoch": 0.8901381304208159,
      "grad_norm": 0.4876999258995056,
      "learning_rate": 8.158543850627205e-06,
      "loss": 0.4913,
      "step": 19397
    },
    {
      "epoch": 0.8901840209260704,
      "grad_norm": 0.463866263628006,
      "learning_rate": 8.158353775742373e-06,
      "loss": 0.4127,
      "step": 19398
    },
    {
      "epoch": 0.8902299114313249,
      "grad_norm": 0.47970008850097656,
      "learning_rate": 8.1581636932627e-06,
      "loss": 0.4333,
      "step": 19399
    },
    {
      "epoch": 0.8902758019365793,
      "grad_norm": 0.4780748188495636,
      "learning_rate": 8.157973603188636e-06,
      "loss": 0.4085,
      "step": 19400
    },
    {
      "epoch": 0.8903216924418338,
      "grad_norm": 0.43783771991729736,
      "learning_rate": 8.157783505520646e-06,
      "loss": 0.4089,
      "step": 19401
    },
    {
      "epoch": 0.8903675829470883,
      "grad_norm": 0.4221642315387726,
      "learning_rate": 8.15759340025918e-06,
      "loss": 0.3503,
      "step": 19402
    },
    {
      "epoch": 0.8904134734523427,
      "grad_norm": 0.3947789669036865,
      "learning_rate": 8.157403287404702e-06,
      "loss": 0.2945,
      "step": 19403
    },
    {
      "epoch": 0.8904593639575972,
      "grad_norm": 0.45561426877975464,
      "learning_rate": 8.157213166957662e-06,
      "loss": 0.3947,
      "step": 19404
    },
    {
      "epoch": 0.8905052544628517,
      "grad_norm": 0.44601961970329285,
      "learning_rate": 8.157023038918524e-06,
      "loss": 0.3466,
      "step": 19405
    },
    {
      "epoch": 0.8905511449681061,
      "grad_norm": 0.4656178057193756,
      "learning_rate": 8.15683290328774e-06,
      "loss": 0.3487,
      "step": 19406
    },
    {
      "epoch": 0.8905970354733606,
      "grad_norm": 0.45121946930885315,
      "learning_rate": 8.15664276006577e-06,
      "loss": 0.4242,
      "step": 19407
    },
    {
      "epoch": 0.890642925978615,
      "grad_norm": 0.442094087600708,
      "learning_rate": 8.15645260925307e-06,
      "loss": 0.3695,
      "step": 19408
    },
    {
      "epoch": 0.8906888164838694,
      "grad_norm": 0.485878050327301,
      "learning_rate": 8.156262450850098e-06,
      "loss": 0.4875,
      "step": 19409
    },
    {
      "epoch": 0.8907347069891239,
      "grad_norm": 0.48015186190605164,
      "learning_rate": 8.156072284857309e-06,
      "loss": 0.4757,
      "step": 19410
    },
    {
      "epoch": 0.8907805974943784,
      "grad_norm": 0.4578854441642761,
      "learning_rate": 8.155882111275164e-06,
      "loss": 0.441,
      "step": 19411
    },
    {
      "epoch": 0.8908264879996328,
      "grad_norm": 0.4316173195838928,
      "learning_rate": 8.155691930104119e-06,
      "loss": 0.3557,
      "step": 19412
    },
    {
      "epoch": 0.8908723785048873,
      "grad_norm": 0.4704018235206604,
      "learning_rate": 8.155501741344631e-06,
      "loss": 0.3957,
      "step": 19413
    },
    {
      "epoch": 0.8909182690101418,
      "grad_norm": 0.46602052450180054,
      "learning_rate": 8.155311544997156e-06,
      "loss": 0.4051,
      "step": 19414
    },
    {
      "epoch": 0.8909641595153963,
      "grad_norm": 0.4631572663784027,
      "learning_rate": 8.155121341062153e-06,
      "loss": 0.4095,
      "step": 19415
    },
    {
      "epoch": 0.8910100500206507,
      "grad_norm": 0.48925384879112244,
      "learning_rate": 8.15493112954008e-06,
      "loss": 0.4705,
      "step": 19416
    },
    {
      "epoch": 0.8910559405259052,
      "grad_norm": 0.4450492560863495,
      "learning_rate": 8.154740910431391e-06,
      "loss": 0.3583,
      "step": 19417
    },
    {
      "epoch": 0.8911018310311597,
      "grad_norm": 0.7268593907356262,
      "learning_rate": 8.154550683736548e-06,
      "loss": 0.4314,
      "step": 19418
    },
    {
      "epoch": 0.8911477215364141,
      "grad_norm": 0.41752228140830994,
      "learning_rate": 8.154360449456005e-06,
      "loss": 0.3139,
      "step": 19419
    },
    {
      "epoch": 0.8911936120416686,
      "grad_norm": 0.4441676437854767,
      "learning_rate": 8.154170207590222e-06,
      "loss": 0.4163,
      "step": 19420
    },
    {
      "epoch": 0.8912395025469231,
      "grad_norm": 0.4454962909221649,
      "learning_rate": 8.153979958139653e-06,
      "loss": 0.3932,
      "step": 19421
    },
    {
      "epoch": 0.8912853930521775,
      "grad_norm": 0.3958258032798767,
      "learning_rate": 8.153789701104759e-06,
      "loss": 0.2881,
      "step": 19422
    },
    {
      "epoch": 0.891331283557432,
      "grad_norm": 0.4647879898548126,
      "learning_rate": 8.153599436485996e-06,
      "loss": 0.4728,
      "step": 19423
    },
    {
      "epoch": 0.8913771740626865,
      "grad_norm": 0.48763689398765564,
      "learning_rate": 8.153409164283822e-06,
      "loss": 0.4135,
      "step": 19424
    },
    {
      "epoch": 0.8914230645679408,
      "grad_norm": 0.4693635404109955,
      "learning_rate": 8.153218884498694e-06,
      "loss": 0.3899,
      "step": 19425
    },
    {
      "epoch": 0.8914689550731953,
      "grad_norm": 0.48482662439346313,
      "learning_rate": 8.15302859713107e-06,
      "loss": 0.494,
      "step": 19426
    },
    {
      "epoch": 0.8915148455784498,
      "grad_norm": 0.4161026179790497,
      "learning_rate": 8.152838302181408e-06,
      "loss": 0.3518,
      "step": 19427
    },
    {
      "epoch": 0.8915607360837042,
      "grad_norm": 0.42444315552711487,
      "learning_rate": 8.152647999650164e-06,
      "loss": 0.3662,
      "step": 19428
    },
    {
      "epoch": 0.8916066265889587,
      "grad_norm": 0.47405731678009033,
      "learning_rate": 8.152457689537795e-06,
      "loss": 0.4703,
      "step": 19429
    },
    {
      "epoch": 0.8916525170942132,
      "grad_norm": 0.44718196988105774,
      "learning_rate": 8.152267371844761e-06,
      "loss": 0.375,
      "step": 19430
    },
    {
      "epoch": 0.8916984075994677,
      "grad_norm": 0.5276437997817993,
      "learning_rate": 8.15207704657152e-06,
      "loss": 0.5015,
      "step": 19431
    },
    {
      "epoch": 0.8917442981047221,
      "grad_norm": 0.45904842019081116,
      "learning_rate": 8.151886713718529e-06,
      "loss": 0.4428,
      "step": 19432
    },
    {
      "epoch": 0.8917901886099766,
      "grad_norm": 0.47445428371429443,
      "learning_rate": 8.151696373286243e-06,
      "loss": 0.4712,
      "step": 19433
    },
    {
      "epoch": 0.8918360791152311,
      "grad_norm": 0.42241159081459045,
      "learning_rate": 8.151506025275124e-06,
      "loss": 0.3146,
      "step": 19434
    },
    {
      "epoch": 0.8918819696204855,
      "grad_norm": 0.45344409346580505,
      "learning_rate": 8.151315669685626e-06,
      "loss": 0.3824,
      "step": 19435
    },
    {
      "epoch": 0.89192786012574,
      "grad_norm": 0.4522955119609833,
      "learning_rate": 8.151125306518209e-06,
      "loss": 0.3765,
      "step": 19436
    },
    {
      "epoch": 0.8919737506309945,
      "grad_norm": 0.46909478306770325,
      "learning_rate": 8.150934935773331e-06,
      "loss": 0.4037,
      "step": 19437
    },
    {
      "epoch": 0.8920196411362489,
      "grad_norm": 0.4537869393825531,
      "learning_rate": 8.15074455745145e-06,
      "loss": 0.4173,
      "step": 19438
    },
    {
      "epoch": 0.8920655316415034,
      "grad_norm": 0.46868228912353516,
      "learning_rate": 8.15055417155302e-06,
      "loss": 0.3929,
      "step": 19439
    },
    {
      "epoch": 0.8921114221467579,
      "grad_norm": 0.48121026158332825,
      "learning_rate": 8.150363778078503e-06,
      "loss": 0.4305,
      "step": 19440
    },
    {
      "epoch": 0.8921573126520123,
      "grad_norm": 0.46466583013534546,
      "learning_rate": 8.150173377028354e-06,
      "loss": 0.3833,
      "step": 19441
    },
    {
      "epoch": 0.8922032031572668,
      "grad_norm": 0.48855188488960266,
      "learning_rate": 8.149982968403033e-06,
      "loss": 0.4156,
      "step": 19442
    },
    {
      "epoch": 0.8922490936625213,
      "grad_norm": 0.4826533794403076,
      "learning_rate": 8.149792552202997e-06,
      "loss": 0.4471,
      "step": 19443
    },
    {
      "epoch": 0.8922949841677756,
      "grad_norm": 0.4371528625488281,
      "learning_rate": 8.149602128428707e-06,
      "loss": 0.3623,
      "step": 19444
    },
    {
      "epoch": 0.8923408746730301,
      "grad_norm": 0.4289834797382355,
      "learning_rate": 8.149411697080613e-06,
      "loss": 0.3436,
      "step": 19445
    },
    {
      "epoch": 0.8923867651782846,
      "grad_norm": 0.6618348360061646,
      "learning_rate": 8.149221258159182e-06,
      "loss": 0.5212,
      "step": 19446
    },
    {
      "epoch": 0.892432655683539,
      "grad_norm": 0.42819881439208984,
      "learning_rate": 8.149030811664866e-06,
      "loss": 0.3639,
      "step": 19447
    },
    {
      "epoch": 0.8924785461887935,
      "grad_norm": 0.46663379669189453,
      "learning_rate": 8.148840357598123e-06,
      "loss": 0.3956,
      "step": 19448
    },
    {
      "epoch": 0.892524436694048,
      "grad_norm": 0.43926116824150085,
      "learning_rate": 8.148649895959414e-06,
      "loss": 0.3862,
      "step": 19449
    },
    {
      "epoch": 0.8925703271993025,
      "grad_norm": 0.41980671882629395,
      "learning_rate": 8.148459426749198e-06,
      "loss": 0.299,
      "step": 19450
    },
    {
      "epoch": 0.8926162177045569,
      "grad_norm": 0.40325164794921875,
      "learning_rate": 8.148268949967929e-06,
      "loss": 0.3075,
      "step": 19451
    },
    {
      "epoch": 0.8926621082098114,
      "grad_norm": 0.46803274750709534,
      "learning_rate": 8.148078465616067e-06,
      "loss": 0.3607,
      "step": 19452
    },
    {
      "epoch": 0.8927079987150659,
      "grad_norm": 0.5105648636817932,
      "learning_rate": 8.147887973694068e-06,
      "loss": 0.3665,
      "step": 19453
    },
    {
      "epoch": 0.8927538892203203,
      "grad_norm": 0.4938548803329468,
      "learning_rate": 8.147697474202395e-06,
      "loss": 0.4717,
      "step": 19454
    },
    {
      "epoch": 0.8927997797255748,
      "grad_norm": 0.41242504119873047,
      "learning_rate": 8.147506967141502e-06,
      "loss": 0.29,
      "step": 19455
    },
    {
      "epoch": 0.8928456702308293,
      "grad_norm": 0.4445120096206665,
      "learning_rate": 8.147316452511846e-06,
      "loss": 0.3549,
      "step": 19456
    },
    {
      "epoch": 0.8928915607360837,
      "grad_norm": 0.48165270686149597,
      "learning_rate": 8.147125930313891e-06,
      "loss": 0.4379,
      "step": 19457
    },
    {
      "epoch": 0.8929374512413382,
      "grad_norm": 0.45806530117988586,
      "learning_rate": 8.146935400548088e-06,
      "loss": 0.4343,
      "step": 19458
    },
    {
      "epoch": 0.8929833417465927,
      "grad_norm": 0.4479197561740875,
      "learning_rate": 8.1467448632149e-06,
      "loss": 0.414,
      "step": 19459
    },
    {
      "epoch": 0.893029232251847,
      "grad_norm": 0.43394848704338074,
      "learning_rate": 8.146554318314785e-06,
      "loss": 0.3716,
      "step": 19460
    },
    {
      "epoch": 0.8930751227571015,
      "grad_norm": 0.4871610403060913,
      "learning_rate": 8.146363765848198e-06,
      "loss": 0.4521,
      "step": 19461
    },
    {
      "epoch": 0.893121013262356,
      "grad_norm": 0.4443134069442749,
      "learning_rate": 8.146173205815599e-06,
      "loss": 0.3725,
      "step": 19462
    },
    {
      "epoch": 0.8931669037676104,
      "grad_norm": 0.5022876858711243,
      "learning_rate": 8.145982638217449e-06,
      "loss": 0.471,
      "step": 19463
    },
    {
      "epoch": 0.8932127942728649,
      "grad_norm": 0.4829522967338562,
      "learning_rate": 8.1457920630542e-06,
      "loss": 0.4495,
      "step": 19464
    },
    {
      "epoch": 0.8932586847781194,
      "grad_norm": 0.4767495393753052,
      "learning_rate": 8.145601480326318e-06,
      "loss": 0.4627,
      "step": 19465
    },
    {
      "epoch": 0.8933045752833739,
      "grad_norm": 0.4510904550552368,
      "learning_rate": 8.145410890034254e-06,
      "loss": 0.357,
      "step": 19466
    },
    {
      "epoch": 0.8933504657886283,
      "grad_norm": 0.4517460763454437,
      "learning_rate": 8.145220292178471e-06,
      "loss": 0.3537,
      "step": 19467
    },
    {
      "epoch": 0.8933963562938828,
      "grad_norm": 0.4477836489677429,
      "learning_rate": 8.145029686759425e-06,
      "loss": 0.4395,
      "step": 19468
    },
    {
      "epoch": 0.8934422467991373,
      "grad_norm": 0.5164333581924438,
      "learning_rate": 8.144839073777575e-06,
      "loss": 0.4942,
      "step": 19469
    },
    {
      "epoch": 0.8934881373043917,
      "grad_norm": 0.4846258759498596,
      "learning_rate": 8.14464845323338e-06,
      "loss": 0.4961,
      "step": 19470
    },
    {
      "epoch": 0.8935340278096462,
      "grad_norm": 0.4388260543346405,
      "learning_rate": 8.144457825127298e-06,
      "loss": 0.3844,
      "step": 19471
    },
    {
      "epoch": 0.8935799183149007,
      "grad_norm": 0.4482676088809967,
      "learning_rate": 8.144267189459789e-06,
      "loss": 0.3695,
      "step": 19472
    },
    {
      "epoch": 0.8936258088201551,
      "grad_norm": 0.41031697392463684,
      "learning_rate": 8.144076546231307e-06,
      "loss": 0.3397,
      "step": 19473
    },
    {
      "epoch": 0.8936716993254096,
      "grad_norm": 0.4591888189315796,
      "learning_rate": 8.143885895442316e-06,
      "loss": 0.4345,
      "step": 19474
    },
    {
      "epoch": 0.8937175898306641,
      "grad_norm": 0.41118794679641724,
      "learning_rate": 8.14369523709327e-06,
      "loss": 0.2958,
      "step": 19475
    },
    {
      "epoch": 0.8937634803359185,
      "grad_norm": 0.4338220953941345,
      "learning_rate": 8.143504571184628e-06,
      "loss": 0.3604,
      "step": 19476
    },
    {
      "epoch": 0.893809370841173,
      "grad_norm": 0.47428327798843384,
      "learning_rate": 8.143313897716852e-06,
      "loss": 0.4571,
      "step": 19477
    },
    {
      "epoch": 0.8938552613464275,
      "grad_norm": 0.43253785371780396,
      "learning_rate": 8.143123216690396e-06,
      "loss": 0.389,
      "step": 19478
    },
    {
      "epoch": 0.8939011518516818,
      "grad_norm": 0.4552631974220276,
      "learning_rate": 8.142932528105722e-06,
      "loss": 0.3858,
      "step": 19479
    },
    {
      "epoch": 0.8939470423569363,
      "grad_norm": 0.5024164319038391,
      "learning_rate": 8.142741831963287e-06,
      "loss": 0.4098,
      "step": 19480
    },
    {
      "epoch": 0.8939929328621908,
      "grad_norm": 0.4136143624782562,
      "learning_rate": 8.14255112826355e-06,
      "loss": 0.3277,
      "step": 19481
    },
    {
      "epoch": 0.8940388233674452,
      "grad_norm": 0.4403443932533264,
      "learning_rate": 8.142360417006969e-06,
      "loss": 0.3789,
      "step": 19482
    },
    {
      "epoch": 0.8940847138726997,
      "grad_norm": 0.45116573572158813,
      "learning_rate": 8.142169698194002e-06,
      "loss": 0.3877,
      "step": 19483
    },
    {
      "epoch": 0.8941306043779542,
      "grad_norm": 0.45773056149482727,
      "learning_rate": 8.141978971825111e-06,
      "loss": 0.424,
      "step": 19484
    },
    {
      "epoch": 0.8941764948832087,
      "grad_norm": 0.46291065216064453,
      "learning_rate": 8.14178823790075e-06,
      "loss": 0.4208,
      "step": 19485
    },
    {
      "epoch": 0.8942223853884631,
      "grad_norm": 0.42325177788734436,
      "learning_rate": 8.14159749642138e-06,
      "loss": 0.3411,
      "step": 19486
    },
    {
      "epoch": 0.8942682758937176,
      "grad_norm": 0.45321425795555115,
      "learning_rate": 8.14140674738746e-06,
      "loss": 0.3633,
      "step": 19487
    },
    {
      "epoch": 0.8943141663989721,
      "grad_norm": 0.4999252259731293,
      "learning_rate": 8.14121599079945e-06,
      "loss": 0.4538,
      "step": 19488
    },
    {
      "epoch": 0.8943600569042265,
      "grad_norm": 0.4953620135784149,
      "learning_rate": 8.141025226657804e-06,
      "loss": 0.4249,
      "step": 19489
    },
    {
      "epoch": 0.894405947409481,
      "grad_norm": 0.44975546002388,
      "learning_rate": 8.140834454962985e-06,
      "loss": 0.4221,
      "step": 19490
    },
    {
      "epoch": 0.8944518379147355,
      "grad_norm": 0.4184361398220062,
      "learning_rate": 8.140643675715452e-06,
      "loss": 0.3573,
      "step": 19491
    },
    {
      "epoch": 0.8944977284199899,
      "grad_norm": 0.43070849776268005,
      "learning_rate": 8.14045288891566e-06,
      "loss": 0.3436,
      "step": 19492
    },
    {
      "epoch": 0.8945436189252444,
      "grad_norm": 0.46900585293769836,
      "learning_rate": 8.14026209456407e-06,
      "loss": 0.4992,
      "step": 19493
    },
    {
      "epoch": 0.8945895094304989,
      "grad_norm": 0.4533962607383728,
      "learning_rate": 8.140071292661141e-06,
      "loss": 0.3825,
      "step": 19494
    },
    {
      "epoch": 0.8946353999357533,
      "grad_norm": 0.45486292243003845,
      "learning_rate": 8.139880483207333e-06,
      "loss": 0.4242,
      "step": 19495
    },
    {
      "epoch": 0.8946812904410077,
      "grad_norm": 0.5144616961479187,
      "learning_rate": 8.139689666203101e-06,
      "loss": 0.4739,
      "step": 19496
    },
    {
      "epoch": 0.8947271809462622,
      "grad_norm": 0.42338499426841736,
      "learning_rate": 8.139498841648908e-06,
      "loss": 0.3798,
      "step": 19497
    },
    {
      "epoch": 0.8947730714515166,
      "grad_norm": 0.41857266426086426,
      "learning_rate": 8.13930800954521e-06,
      "loss": 0.3643,
      "step": 19498
    },
    {
      "epoch": 0.8948189619567711,
      "grad_norm": 0.5519826412200928,
      "learning_rate": 8.139117169892469e-06,
      "loss": 0.4356,
      "step": 19499
    },
    {
      "epoch": 0.8948648524620256,
      "grad_norm": 0.41876888275146484,
      "learning_rate": 8.13892632269114e-06,
      "loss": 0.3612,
      "step": 19500
    },
    {
      "epoch": 0.89491074296728,
      "grad_norm": 0.48882347345352173,
      "learning_rate": 8.138735467941684e-06,
      "loss": 0.4796,
      "step": 19501
    },
    {
      "epoch": 0.8949566334725345,
      "grad_norm": 0.4193541705608368,
      "learning_rate": 8.13854460564456e-06,
      "loss": 0.34,
      "step": 19502
    },
    {
      "epoch": 0.895002523977789,
      "grad_norm": 0.46662354469299316,
      "learning_rate": 8.138353735800228e-06,
      "loss": 0.454,
      "step": 19503
    },
    {
      "epoch": 0.8950484144830435,
      "grad_norm": 0.4273943603038788,
      "learning_rate": 8.138162858409144e-06,
      "loss": 0.3875,
      "step": 19504
    },
    {
      "epoch": 0.8950943049882979,
      "grad_norm": 0.4244268238544464,
      "learning_rate": 8.137971973471769e-06,
      "loss": 0.3485,
      "step": 19505
    },
    {
      "epoch": 0.8951401954935524,
      "grad_norm": 0.4576314687728882,
      "learning_rate": 8.137781080988564e-06,
      "loss": 0.4461,
      "step": 19506
    },
    {
      "epoch": 0.8951860859988069,
      "grad_norm": 0.4545513689517975,
      "learning_rate": 8.137590180959984e-06,
      "loss": 0.3799,
      "step": 19507
    },
    {
      "epoch": 0.8952319765040613,
      "grad_norm": 0.4554949700832367,
      "learning_rate": 8.13739927338649e-06,
      "loss": 0.3752,
      "step": 19508
    },
    {
      "epoch": 0.8952778670093158,
      "grad_norm": 0.4584297835826874,
      "learning_rate": 8.13720835826854e-06,
      "loss": 0.4076,
      "step": 19509
    },
    {
      "epoch": 0.8953237575145703,
      "grad_norm": 0.3998614549636841,
      "learning_rate": 8.137017435606595e-06,
      "loss": 0.3302,
      "step": 19510
    },
    {
      "epoch": 0.8953696480198247,
      "grad_norm": 0.4565545916557312,
      "learning_rate": 8.136826505401115e-06,
      "loss": 0.4467,
      "step": 19511
    },
    {
      "epoch": 0.8954155385250792,
      "grad_norm": 0.44657012820243835,
      "learning_rate": 8.136635567652555e-06,
      "loss": 0.3759,
      "step": 19512
    },
    {
      "epoch": 0.8954614290303337,
      "grad_norm": 0.4878428280353546,
      "learning_rate": 8.136444622361377e-06,
      "loss": 0.4024,
      "step": 19513
    },
    {
      "epoch": 0.895507319535588,
      "grad_norm": 0.4634716510772705,
      "learning_rate": 8.136253669528039e-06,
      "loss": 0.4205,
      "step": 19514
    },
    {
      "epoch": 0.8955532100408425,
      "grad_norm": 0.4810328185558319,
      "learning_rate": 8.136062709153001e-06,
      "loss": 0.4497,
      "step": 19515
    },
    {
      "epoch": 0.895599100546097,
      "grad_norm": 0.45394232869148254,
      "learning_rate": 8.135871741236723e-06,
      "loss": 0.3941,
      "step": 19516
    },
    {
      "epoch": 0.8956449910513514,
      "grad_norm": 0.4583016633987427,
      "learning_rate": 8.135680765779663e-06,
      "loss": 0.4446,
      "step": 19517
    },
    {
      "epoch": 0.8956908815566059,
      "grad_norm": 0.454923540353775,
      "learning_rate": 8.135489782782278e-06,
      "loss": 0.357,
      "step": 19518
    },
    {
      "epoch": 0.8957367720618604,
      "grad_norm": 0.45031121373176575,
      "learning_rate": 8.135298792245033e-06,
      "loss": 0.4097,
      "step": 19519
    },
    {
      "epoch": 0.8957826625671149,
      "grad_norm": 0.4369613230228424,
      "learning_rate": 8.135107794168381e-06,
      "loss": 0.3681,
      "step": 19520
    },
    {
      "epoch": 0.8958285530723693,
      "grad_norm": 0.42453238368034363,
      "learning_rate": 8.134916788552784e-06,
      "loss": 0.3637,
      "step": 19521
    },
    {
      "epoch": 0.8958744435776238,
      "grad_norm": 0.5763978958129883,
      "learning_rate": 8.134725775398704e-06,
      "loss": 0.3798,
      "step": 19522
    },
    {
      "epoch": 0.8959203340828783,
      "grad_norm": 0.4394003450870514,
      "learning_rate": 8.134534754706598e-06,
      "loss": 0.334,
      "step": 19523
    },
    {
      "epoch": 0.8959662245881327,
      "grad_norm": 0.46039608120918274,
      "learning_rate": 8.134343726476923e-06,
      "loss": 0.467,
      "step": 19524
    },
    {
      "epoch": 0.8960121150933872,
      "grad_norm": 0.4259471297264099,
      "learning_rate": 8.134152690710142e-06,
      "loss": 0.3353,
      "step": 19525
    },
    {
      "epoch": 0.8960580055986417,
      "grad_norm": 0.4664720594882965,
      "learning_rate": 8.133961647406715e-06,
      "loss": 0.3832,
      "step": 19526
    },
    {
      "epoch": 0.8961038961038961,
      "grad_norm": 0.46415975689888,
      "learning_rate": 8.133770596567095e-06,
      "loss": 0.412,
      "step": 19527
    },
    {
      "epoch": 0.8961497866091506,
      "grad_norm": 0.4015985429286957,
      "learning_rate": 8.133579538191748e-06,
      "loss": 0.3376,
      "step": 19528
    },
    {
      "epoch": 0.8961956771144051,
      "grad_norm": 0.4777730107307434,
      "learning_rate": 8.13338847228113e-06,
      "loss": 0.3748,
      "step": 19529
    },
    {
      "epoch": 0.8962415676196595,
      "grad_norm": 0.4471748173236847,
      "learning_rate": 8.133197398835704e-06,
      "loss": 0.3893,
      "step": 19530
    },
    {
      "epoch": 0.896287458124914,
      "grad_norm": 0.4751160144805908,
      "learning_rate": 8.133006317855925e-06,
      "loss": 0.4672,
      "step": 19531
    },
    {
      "epoch": 0.8963333486301684,
      "grad_norm": 0.4724672734737396,
      "learning_rate": 8.132815229342255e-06,
      "loss": 0.4506,
      "step": 19532
    },
    {
      "epoch": 0.8963792391354228,
      "grad_norm": 0.4228498041629791,
      "learning_rate": 8.132624133295155e-06,
      "loss": 0.3631,
      "step": 19533
    },
    {
      "epoch": 0.8964251296406773,
      "grad_norm": 0.45519793033599854,
      "learning_rate": 8.132433029715082e-06,
      "loss": 0.4204,
      "step": 19534
    },
    {
      "epoch": 0.8964710201459318,
      "grad_norm": 0.5106019973754883,
      "learning_rate": 8.132241918602493e-06,
      "loss": 0.5202,
      "step": 19535
    },
    {
      "epoch": 0.8965169106511862,
      "grad_norm": 0.48189252614974976,
      "learning_rate": 8.132050799957854e-06,
      "loss": 0.4086,
      "step": 19536
    },
    {
      "epoch": 0.8965628011564407,
      "grad_norm": 0.44232305884361267,
      "learning_rate": 8.13185967378162e-06,
      "loss": 0.3583,
      "step": 19537
    },
    {
      "epoch": 0.8966086916616952,
      "grad_norm": 0.43216806650161743,
      "learning_rate": 8.131668540074253e-06,
      "loss": 0.3812,
      "step": 19538
    },
    {
      "epoch": 0.8966545821669497,
      "grad_norm": 0.45896944403648376,
      "learning_rate": 8.13147739883621e-06,
      "loss": 0.4333,
      "step": 19539
    },
    {
      "epoch": 0.8967004726722041,
      "grad_norm": 0.4305015206336975,
      "learning_rate": 8.131286250067953e-06,
      "loss": 0.3465,
      "step": 19540
    },
    {
      "epoch": 0.8967463631774586,
      "grad_norm": 0.43023139238357544,
      "learning_rate": 8.131095093769941e-06,
      "loss": 0.3835,
      "step": 19541
    },
    {
      "epoch": 0.8967922536827131,
      "grad_norm": 0.41526398062705994,
      "learning_rate": 8.130903929942633e-06,
      "loss": 0.3219,
      "step": 19542
    },
    {
      "epoch": 0.8968381441879675,
      "grad_norm": 0.43843820691108704,
      "learning_rate": 8.13071275858649e-06,
      "loss": 0.3577,
      "step": 19543
    },
    {
      "epoch": 0.896884034693222,
      "grad_norm": 0.41350796818733215,
      "learning_rate": 8.13052157970197e-06,
      "loss": 0.347,
      "step": 19544
    },
    {
      "epoch": 0.8969299251984765,
      "grad_norm": 0.48723945021629333,
      "learning_rate": 8.130330393289533e-06,
      "loss": 0.4487,
      "step": 19545
    },
    {
      "epoch": 0.8969758157037309,
      "grad_norm": 0.448485791683197,
      "learning_rate": 8.130139199349642e-06,
      "loss": 0.3683,
      "step": 19546
    },
    {
      "epoch": 0.8970217062089854,
      "grad_norm": 0.48506516218185425,
      "learning_rate": 8.129947997882753e-06,
      "loss": 0.4672,
      "step": 19547
    },
    {
      "epoch": 0.8970675967142399,
      "grad_norm": 0.4289757311344147,
      "learning_rate": 8.129756788889325e-06,
      "loss": 0.3664,
      "step": 19548
    },
    {
      "epoch": 0.8971134872194942,
      "grad_norm": 0.4899705648422241,
      "learning_rate": 8.129565572369822e-06,
      "loss": 0.4229,
      "step": 19549
    },
    {
      "epoch": 0.8971593777247487,
      "grad_norm": 0.4408937990665436,
      "learning_rate": 8.129374348324701e-06,
      "loss": 0.3513,
      "step": 19550
    },
    {
      "epoch": 0.8972052682300032,
      "grad_norm": 0.4421949088573456,
      "learning_rate": 8.129183116754422e-06,
      "loss": 0.4231,
      "step": 19551
    },
    {
      "epoch": 0.8972511587352576,
      "grad_norm": 0.4801378846168518,
      "learning_rate": 8.128991877659444e-06,
      "loss": 0.4851,
      "step": 19552
    },
    {
      "epoch": 0.8972970492405121,
      "grad_norm": 0.4513093829154968,
      "learning_rate": 8.12880063104023e-06,
      "loss": 0.3491,
      "step": 19553
    },
    {
      "epoch": 0.8973429397457666,
      "grad_norm": 0.45260581374168396,
      "learning_rate": 8.128609376897237e-06,
      "loss": 0.3993,
      "step": 19554
    },
    {
      "epoch": 0.8973888302510211,
      "grad_norm": 0.45366472005844116,
      "learning_rate": 8.128418115230927e-06,
      "loss": 0.3746,
      "step": 19555
    },
    {
      "epoch": 0.8974347207562755,
      "grad_norm": 0.46830397844314575,
      "learning_rate": 8.128226846041758e-06,
      "loss": 0.4631,
      "step": 19556
    },
    {
      "epoch": 0.89748061126153,
      "grad_norm": 0.47571054100990295,
      "learning_rate": 8.12803556933019e-06,
      "loss": 0.4127,
      "step": 19557
    },
    {
      "epoch": 0.8975265017667845,
      "grad_norm": 0.4808880090713501,
      "learning_rate": 8.127844285096685e-06,
      "loss": 0.4326,
      "step": 19558
    },
    {
      "epoch": 0.8975723922720389,
      "grad_norm": 0.4478267431259155,
      "learning_rate": 8.127652993341701e-06,
      "loss": 0.3591,
      "step": 19559
    },
    {
      "epoch": 0.8976182827772934,
      "grad_norm": 0.46266263723373413,
      "learning_rate": 8.127461694065698e-06,
      "loss": 0.4416,
      "step": 19560
    },
    {
      "epoch": 0.8976641732825479,
      "grad_norm": 0.4565480053424835,
      "learning_rate": 8.127270387269137e-06,
      "loss": 0.3679,
      "step": 19561
    },
    {
      "epoch": 0.8977100637878023,
      "grad_norm": 0.4885430634021759,
      "learning_rate": 8.12707907295248e-06,
      "loss": 0.4122,
      "step": 19562
    },
    {
      "epoch": 0.8977559542930568,
      "grad_norm": 0.46203354001045227,
      "learning_rate": 8.126887751116181e-06,
      "loss": 0.3917,
      "step": 19563
    },
    {
      "epoch": 0.8978018447983113,
      "grad_norm": 0.45533332228660583,
      "learning_rate": 8.126696421760705e-06,
      "loss": 0.4326,
      "step": 19564
    },
    {
      "epoch": 0.8978477353035657,
      "grad_norm": 0.47094833850860596,
      "learning_rate": 8.126505084886514e-06,
      "loss": 0.3773,
      "step": 19565
    },
    {
      "epoch": 0.8978936258088202,
      "grad_norm": 0.4857780337333679,
      "learning_rate": 8.12631374049406e-06,
      "loss": 0.4352,
      "step": 19566
    },
    {
      "epoch": 0.8979395163140746,
      "grad_norm": 0.4782284200191498,
      "learning_rate": 8.126122388583812e-06,
      "loss": 0.389,
      "step": 19567
    },
    {
      "epoch": 0.897985406819329,
      "grad_norm": 0.47144559025764465,
      "learning_rate": 8.125931029156224e-06,
      "loss": 0.4197,
      "step": 19568
    },
    {
      "epoch": 0.8980312973245835,
      "grad_norm": 0.4282037913799286,
      "learning_rate": 8.12573966221176e-06,
      "loss": 0.3579,
      "step": 19569
    },
    {
      "epoch": 0.898077187829838,
      "grad_norm": 0.48832759261131287,
      "learning_rate": 8.12554828775088e-06,
      "loss": 0.4056,
      "step": 19570
    },
    {
      "epoch": 0.8981230783350924,
      "grad_norm": 0.44976314902305603,
      "learning_rate": 8.12535690577404e-06,
      "loss": 0.4303,
      "step": 19571
    },
    {
      "epoch": 0.8981689688403469,
      "grad_norm": 0.43278777599334717,
      "learning_rate": 8.125165516281703e-06,
      "loss": 0.3617,
      "step": 19572
    },
    {
      "epoch": 0.8982148593456014,
      "grad_norm": 0.477529376745224,
      "learning_rate": 8.124974119274331e-06,
      "loss": 0.4876,
      "step": 19573
    },
    {
      "epoch": 0.8982607498508559,
      "grad_norm": 0.40531647205352783,
      "learning_rate": 8.124782714752382e-06,
      "loss": 0.3093,
      "step": 19574
    },
    {
      "epoch": 0.8983066403561103,
      "grad_norm": 0.4100872278213501,
      "learning_rate": 8.124591302716316e-06,
      "loss": 0.3193,
      "step": 19575
    },
    {
      "epoch": 0.8983525308613648,
      "grad_norm": 0.43445339798927307,
      "learning_rate": 8.124399883166595e-06,
      "loss": 0.3474,
      "step": 19576
    },
    {
      "epoch": 0.8983984213666193,
      "grad_norm": 0.39190801978111267,
      "learning_rate": 8.124208456103677e-06,
      "loss": 0.2875,
      "step": 19577
    },
    {
      "epoch": 0.8984443118718737,
      "grad_norm": 0.45114728808403015,
      "learning_rate": 8.124017021528025e-06,
      "loss": 0.4313,
      "step": 19578
    },
    {
      "epoch": 0.8984902023771282,
      "grad_norm": 0.4255850315093994,
      "learning_rate": 8.123825579440098e-06,
      "loss": 0.3683,
      "step": 19579
    },
    {
      "epoch": 0.8985360928823827,
      "grad_norm": 0.4089559018611908,
      "learning_rate": 8.123634129840357e-06,
      "loss": 0.3592,
      "step": 19580
    },
    {
      "epoch": 0.8985819833876371,
      "grad_norm": 0.44748973846435547,
      "learning_rate": 8.12344267272926e-06,
      "loss": 0.407,
      "step": 19581
    },
    {
      "epoch": 0.8986278738928916,
      "grad_norm": 0.4420090317726135,
      "learning_rate": 8.12325120810727e-06,
      "loss": 0.3911,
      "step": 19582
    },
    {
      "epoch": 0.8986737643981461,
      "grad_norm": 0.4208831191062927,
      "learning_rate": 8.123059735974847e-06,
      "loss": 0.3505,
      "step": 19583
    },
    {
      "epoch": 0.8987196549034004,
      "grad_norm": 0.4403655230998993,
      "learning_rate": 8.12286825633245e-06,
      "loss": 0.3507,
      "step": 19584
    },
    {
      "epoch": 0.8987655454086549,
      "grad_norm": 0.49103429913520813,
      "learning_rate": 8.122676769180542e-06,
      "loss": 0.4367,
      "step": 19585
    },
    {
      "epoch": 0.8988114359139094,
      "grad_norm": 0.4869801700115204,
      "learning_rate": 8.12248527451958e-06,
      "loss": 0.4308,
      "step": 19586
    },
    {
      "epoch": 0.8988573264191638,
      "grad_norm": 0.4368007183074951,
      "learning_rate": 8.122293772350028e-06,
      "loss": 0.3526,
      "step": 19587
    },
    {
      "epoch": 0.8989032169244183,
      "grad_norm": 0.4613476097583771,
      "learning_rate": 8.122102262672344e-06,
      "loss": 0.3677,
      "step": 19588
    },
    {
      "epoch": 0.8989491074296728,
      "grad_norm": 0.4299507439136505,
      "learning_rate": 8.12191074548699e-06,
      "loss": 0.3715,
      "step": 19589
    },
    {
      "epoch": 0.8989949979349272,
      "grad_norm": 0.4662078619003296,
      "learning_rate": 8.121719220794426e-06,
      "loss": 0.4227,
      "step": 19590
    },
    {
      "epoch": 0.8990408884401817,
      "grad_norm": 0.4886345863342285,
      "learning_rate": 8.121527688595112e-06,
      "loss": 0.4282,
      "step": 19591
    },
    {
      "epoch": 0.8990867789454362,
      "grad_norm": 0.45164838433265686,
      "learning_rate": 8.121336148889508e-06,
      "loss": 0.3907,
      "step": 19592
    },
    {
      "epoch": 0.8991326694506907,
      "grad_norm": 0.4625115990638733,
      "learning_rate": 8.121144601678077e-06,
      "loss": 0.4151,
      "step": 19593
    },
    {
      "epoch": 0.8991785599559451,
      "grad_norm": 0.4364461898803711,
      "learning_rate": 8.120953046961279e-06,
      "loss": 0.4488,
      "step": 19594
    },
    {
      "epoch": 0.8992244504611996,
      "grad_norm": 0.4251542091369629,
      "learning_rate": 8.120761484739571e-06,
      "loss": 0.3757,
      "step": 19595
    },
    {
      "epoch": 0.8992703409664541,
      "grad_norm": 0.48554930090904236,
      "learning_rate": 8.120569915013421e-06,
      "loss": 0.4133,
      "step": 19596
    },
    {
      "epoch": 0.8993162314717085,
      "grad_norm": 0.47698280215263367,
      "learning_rate": 8.120378337783283e-06,
      "loss": 0.4238,
      "step": 19597
    },
    {
      "epoch": 0.899362121976963,
      "grad_norm": 0.4423588514328003,
      "learning_rate": 8.12018675304962e-06,
      "loss": 0.3888,
      "step": 19598
    },
    {
      "epoch": 0.8994080124822175,
      "grad_norm": 0.4697827398777008,
      "learning_rate": 8.119995160812893e-06,
      "loss": 0.4517,
      "step": 19599
    },
    {
      "epoch": 0.8994539029874719,
      "grad_norm": 0.47125673294067383,
      "learning_rate": 8.11980356107356e-06,
      "loss": 0.3779,
      "step": 19600
    },
    {
      "epoch": 0.8994997934927264,
      "grad_norm": 0.4564676284790039,
      "learning_rate": 8.119611953832087e-06,
      "loss": 0.3723,
      "step": 19601
    },
    {
      "epoch": 0.8995456839979808,
      "grad_norm": 0.44823724031448364,
      "learning_rate": 8.11942033908893e-06,
      "loss": 0.3204,
      "step": 19602
    },
    {
      "epoch": 0.8995915745032352,
      "grad_norm": 0.45878884196281433,
      "learning_rate": 8.11922871684455e-06,
      "loss": 0.3732,
      "step": 19603
    },
    {
      "epoch": 0.8996374650084897,
      "grad_norm": 0.4678589403629303,
      "learning_rate": 8.119037087099415e-06,
      "loss": 0.4272,
      "step": 19604
    },
    {
      "epoch": 0.8996833555137442,
      "grad_norm": 0.4269801676273346,
      "learning_rate": 8.118845449853976e-06,
      "loss": 0.3402,
      "step": 19605
    },
    {
      "epoch": 0.8997292460189986,
      "grad_norm": 0.45910999178886414,
      "learning_rate": 8.118653805108698e-06,
      "loss": 0.404,
      "step": 19606
    },
    {
      "epoch": 0.8997751365242531,
      "grad_norm": 0.4273187220096588,
      "learning_rate": 8.118462152864044e-06,
      "loss": 0.3369,
      "step": 19607
    },
    {
      "epoch": 0.8998210270295076,
      "grad_norm": 0.45106983184814453,
      "learning_rate": 8.118270493120472e-06,
      "loss": 0.3719,
      "step": 19608
    },
    {
      "epoch": 0.8998669175347621,
      "grad_norm": 0.4412156939506531,
      "learning_rate": 8.118078825878441e-06,
      "loss": 0.3556,
      "step": 19609
    },
    {
      "epoch": 0.8999128080400165,
      "grad_norm": 0.43864861130714417,
      "learning_rate": 8.117887151138418e-06,
      "loss": 0.3601,
      "step": 19610
    },
    {
      "epoch": 0.899958698545271,
      "grad_norm": 0.46257439255714417,
      "learning_rate": 8.11769546890086e-06,
      "loss": 0.3951,
      "step": 19611
    },
    {
      "epoch": 0.9000045890505255,
      "grad_norm": 0.46688273549079895,
      "learning_rate": 8.117503779166228e-06,
      "loss": 0.4378,
      "step": 19612
    },
    {
      "epoch": 0.9000504795557799,
      "grad_norm": 0.489102303981781,
      "learning_rate": 8.117312081934982e-06,
      "loss": 0.4875,
      "step": 19613
    },
    {
      "epoch": 0.9000963700610344,
      "grad_norm": 0.44962283968925476,
      "learning_rate": 8.117120377207585e-06,
      "loss": 0.3868,
      "step": 19614
    },
    {
      "epoch": 0.9001422605662889,
      "grad_norm": 0.47287270426750183,
      "learning_rate": 8.116928664984498e-06,
      "loss": 0.39,
      "step": 19615
    },
    {
      "epoch": 0.9001881510715433,
      "grad_norm": 0.43893179297447205,
      "learning_rate": 8.116736945266179e-06,
      "loss": 0.4042,
      "step": 19616
    },
    {
      "epoch": 0.9002340415767978,
      "grad_norm": 0.4542989730834961,
      "learning_rate": 8.116545218053093e-06,
      "loss": 0.4311,
      "step": 19617
    },
    {
      "epoch": 0.9002799320820523,
      "grad_norm": 0.5006599426269531,
      "learning_rate": 8.116353483345701e-06,
      "loss": 0.487,
      "step": 19618
    },
    {
      "epoch": 0.9003258225873066,
      "grad_norm": 0.434628963470459,
      "learning_rate": 8.11616174114446e-06,
      "loss": 0.3361,
      "step": 19619
    },
    {
      "epoch": 0.9003717130925611,
      "grad_norm": 0.4790666103363037,
      "learning_rate": 8.115969991449835e-06,
      "loss": 0.5053,
      "step": 19620
    },
    {
      "epoch": 0.9004176035978156,
      "grad_norm": 0.4567667245864868,
      "learning_rate": 8.115778234262284e-06,
      "loss": 0.3786,
      "step": 19621
    },
    {
      "epoch": 0.90046349410307,
      "grad_norm": 0.41192808747291565,
      "learning_rate": 8.115586469582271e-06,
      "loss": 0.3136,
      "step": 19622
    },
    {
      "epoch": 0.9005093846083245,
      "grad_norm": 0.4622694253921509,
      "learning_rate": 8.115394697410254e-06,
      "loss": 0.4316,
      "step": 19623
    },
    {
      "epoch": 0.900555275113579,
      "grad_norm": 0.4549066424369812,
      "learning_rate": 8.115202917746699e-06,
      "loss": 0.417,
      "step": 19624
    },
    {
      "epoch": 0.9006011656188334,
      "grad_norm": 0.46292001008987427,
      "learning_rate": 8.115011130592061e-06,
      "loss": 0.4331,
      "step": 19625
    },
    {
      "epoch": 0.9006470561240879,
      "grad_norm": 0.39980098605155945,
      "learning_rate": 8.114819335946807e-06,
      "loss": 0.3193,
      "step": 19626
    },
    {
      "epoch": 0.9006929466293424,
      "grad_norm": 0.439362108707428,
      "learning_rate": 8.114627533811394e-06,
      "loss": 0.383,
      "step": 19627
    },
    {
      "epoch": 0.9007388371345969,
      "grad_norm": 0.4720875918865204,
      "learning_rate": 8.114435724186287e-06,
      "loss": 0.3825,
      "step": 19628
    },
    {
      "epoch": 0.9007847276398513,
      "grad_norm": 0.4588710367679596,
      "learning_rate": 8.114243907071942e-06,
      "loss": 0.3452,
      "step": 19629
    },
    {
      "epoch": 0.9008306181451058,
      "grad_norm": 0.4508844316005707,
      "learning_rate": 8.114052082468825e-06,
      "loss": 0.3733,
      "step": 19630
    },
    {
      "epoch": 0.9008765086503603,
      "grad_norm": 0.4787546992301941,
      "learning_rate": 8.113860250377395e-06,
      "loss": 0.4559,
      "step": 19631
    },
    {
      "epoch": 0.9009223991556147,
      "grad_norm": 0.4768097400665283,
      "learning_rate": 8.113668410798114e-06,
      "loss": 0.4449,
      "step": 19632
    },
    {
      "epoch": 0.9009682896608692,
      "grad_norm": 0.5088914036750793,
      "learning_rate": 8.113476563731444e-06,
      "loss": 0.5128,
      "step": 19633
    },
    {
      "epoch": 0.9010141801661237,
      "grad_norm": 0.42240893840789795,
      "learning_rate": 8.113284709177844e-06,
      "loss": 0.3298,
      "step": 19634
    },
    {
      "epoch": 0.901060070671378,
      "grad_norm": 0.48385047912597656,
      "learning_rate": 8.113092847137778e-06,
      "loss": 0.4638,
      "step": 19635
    },
    {
      "epoch": 0.9011059611766326,
      "grad_norm": 0.49349191784858704,
      "learning_rate": 8.112900977611706e-06,
      "loss": 0.4078,
      "step": 19636
    },
    {
      "epoch": 0.901151851681887,
      "grad_norm": 0.4590332508087158,
      "learning_rate": 8.11270910060009e-06,
      "loss": 0.3944,
      "step": 19637
    },
    {
      "epoch": 0.9011977421871414,
      "grad_norm": 0.4321778416633606,
      "learning_rate": 8.112517216103391e-06,
      "loss": 0.3755,
      "step": 19638
    },
    {
      "epoch": 0.9012436326923959,
      "grad_norm": 0.4985816180706024,
      "learning_rate": 8.112325324122069e-06,
      "loss": 0.5117,
      "step": 19639
    },
    {
      "epoch": 0.9012895231976504,
      "grad_norm": 0.4458997845649719,
      "learning_rate": 8.112133424656587e-06,
      "loss": 0.3632,
      "step": 19640
    },
    {
      "epoch": 0.9013354137029048,
      "grad_norm": 0.4296756386756897,
      "learning_rate": 8.111941517707406e-06,
      "loss": 0.3431,
      "step": 19641
    },
    {
      "epoch": 0.9013813042081593,
      "grad_norm": 0.455159455537796,
      "learning_rate": 8.11174960327499e-06,
      "loss": 0.4295,
      "step": 19642
    },
    {
      "epoch": 0.9014271947134138,
      "grad_norm": 0.4792745113372803,
      "learning_rate": 8.111557681359796e-06,
      "loss": 0.4772,
      "step": 19643
    },
    {
      "epoch": 0.9014730852186683,
      "grad_norm": 0.4505153298377991,
      "learning_rate": 8.11136575196229e-06,
      "loss": 0.4169,
      "step": 19644
    },
    {
      "epoch": 0.9015189757239227,
      "grad_norm": 0.42460453510284424,
      "learning_rate": 8.111173815082931e-06,
      "loss": 0.3971,
      "step": 19645
    },
    {
      "epoch": 0.9015648662291772,
      "grad_norm": 0.4281695485115051,
      "learning_rate": 8.110981870722179e-06,
      "loss": 0.3691,
      "step": 19646
    },
    {
      "epoch": 0.9016107567344317,
      "grad_norm": 0.4597682058811188,
      "learning_rate": 8.1107899188805e-06,
      "loss": 0.43,
      "step": 19647
    },
    {
      "epoch": 0.9016566472396861,
      "grad_norm": 0.47610172629356384,
      "learning_rate": 8.11059795955835e-06,
      "loss": 0.5058,
      "step": 19648
    },
    {
      "epoch": 0.9017025377449406,
      "grad_norm": 0.4642077386379242,
      "learning_rate": 8.110405992756197e-06,
      "loss": 0.4283,
      "step": 19649
    },
    {
      "epoch": 0.9017484282501951,
      "grad_norm": 0.4710111916065216,
      "learning_rate": 8.110214018474497e-06,
      "loss": 0.4772,
      "step": 19650
    },
    {
      "epoch": 0.9017943187554495,
      "grad_norm": 0.44823944568634033,
      "learning_rate": 8.110022036713715e-06,
      "loss": 0.4001,
      "step": 19651
    },
    {
      "epoch": 0.901840209260704,
      "grad_norm": 0.46762052178382874,
      "learning_rate": 8.109830047474311e-06,
      "loss": 0.4389,
      "step": 19652
    },
    {
      "epoch": 0.9018860997659585,
      "grad_norm": 0.4464554786682129,
      "learning_rate": 8.109638050756747e-06,
      "loss": 0.3994,
      "step": 19653
    },
    {
      "epoch": 0.9019319902712128,
      "grad_norm": 0.42765629291534424,
      "learning_rate": 8.109446046561486e-06,
      "loss": 0.354,
      "step": 19654
    },
    {
      "epoch": 0.9019778807764673,
      "grad_norm": 0.47394561767578125,
      "learning_rate": 8.109254034888987e-06,
      "loss": 0.4081,
      "step": 19655
    },
    {
      "epoch": 0.9020237712817218,
      "grad_norm": 0.4481166899204254,
      "learning_rate": 8.109062015739715e-06,
      "loss": 0.37,
      "step": 19656
    },
    {
      "epoch": 0.9020696617869762,
      "grad_norm": 0.5158683657646179,
      "learning_rate": 8.108869989114128e-06,
      "loss": 0.5874,
      "step": 19657
    },
    {
      "epoch": 0.9021155522922307,
      "grad_norm": 0.4390256404876709,
      "learning_rate": 8.108677955012692e-06,
      "loss": 0.3885,
      "step": 19658
    },
    {
      "epoch": 0.9021614427974852,
      "grad_norm": 0.4856985807418823,
      "learning_rate": 8.108485913435866e-06,
      "loss": 0.4557,
      "step": 19659
    },
    {
      "epoch": 0.9022073333027396,
      "grad_norm": 0.4250340163707733,
      "learning_rate": 8.108293864384111e-06,
      "loss": 0.3444,
      "step": 19660
    },
    {
      "epoch": 0.9022532238079941,
      "grad_norm": 0.46024569869041443,
      "learning_rate": 8.108101807857893e-06,
      "loss": 0.4347,
      "step": 19661
    },
    {
      "epoch": 0.9022991143132486,
      "grad_norm": 0.4689312279224396,
      "learning_rate": 8.107909743857668e-06,
      "loss": 0.4106,
      "step": 19662
    },
    {
      "epoch": 0.9023450048185031,
      "grad_norm": 0.47165659070014954,
      "learning_rate": 8.107717672383903e-06,
      "loss": 0.3943,
      "step": 19663
    },
    {
      "epoch": 0.9023908953237575,
      "grad_norm": 0.43660423159599304,
      "learning_rate": 8.107525593437058e-06,
      "loss": 0.3986,
      "step": 19664
    },
    {
      "epoch": 0.902436785829012,
      "grad_norm": 0.4649445116519928,
      "learning_rate": 8.107333507017595e-06,
      "loss": 0.411,
      "step": 19665
    },
    {
      "epoch": 0.9024826763342665,
      "grad_norm": 0.4339834749698639,
      "learning_rate": 8.107141413125974e-06,
      "loss": 0.3468,
      "step": 19666
    },
    {
      "epoch": 0.9025285668395209,
      "grad_norm": 0.4421994686126709,
      "learning_rate": 8.10694931176266e-06,
      "loss": 0.3054,
      "step": 19667
    },
    {
      "epoch": 0.9025744573447754,
      "grad_norm": 0.4310256838798523,
      "learning_rate": 8.106757202928112e-06,
      "loss": 0.3524,
      "step": 19668
    },
    {
      "epoch": 0.9026203478500299,
      "grad_norm": 0.49119946360588074,
      "learning_rate": 8.106565086622796e-06,
      "loss": 0.4681,
      "step": 19669
    },
    {
      "epoch": 0.9026662383552843,
      "grad_norm": 0.44626155495643616,
      "learning_rate": 8.106372962847168e-06,
      "loss": 0.4243,
      "step": 19670
    },
    {
      "epoch": 0.9027121288605388,
      "grad_norm": 0.45187458395957947,
      "learning_rate": 8.106180831601695e-06,
      "loss": 0.3848,
      "step": 19671
    },
    {
      "epoch": 0.9027580193657933,
      "grad_norm": 0.4735959470272064,
      "learning_rate": 8.105988692886838e-06,
      "loss": 0.4805,
      "step": 19672
    },
    {
      "epoch": 0.9028039098710476,
      "grad_norm": 0.4806026518344879,
      "learning_rate": 8.105796546703057e-06,
      "loss": 0.4554,
      "step": 19673
    },
    {
      "epoch": 0.9028498003763021,
      "grad_norm": 0.4555499255657196,
      "learning_rate": 8.105604393050816e-06,
      "loss": 0.3798,
      "step": 19674
    },
    {
      "epoch": 0.9028956908815566,
      "grad_norm": 0.4304138422012329,
      "learning_rate": 8.105412231930579e-06,
      "loss": 0.3734,
      "step": 19675
    },
    {
      "epoch": 0.902941581386811,
      "grad_norm": 0.463528037071228,
      "learning_rate": 8.105220063342802e-06,
      "loss": 0.3765,
      "step": 19676
    },
    {
      "epoch": 0.9029874718920655,
      "grad_norm": 0.43964168429374695,
      "learning_rate": 8.105027887287953e-06,
      "loss": 0.3585,
      "step": 19677
    },
    {
      "epoch": 0.90303336239732,
      "grad_norm": 0.4342395067214966,
      "learning_rate": 8.104835703766492e-06,
      "loss": 0.3945,
      "step": 19678
    },
    {
      "epoch": 0.9030792529025744,
      "grad_norm": 0.43120867013931274,
      "learning_rate": 8.10464351277888e-06,
      "loss": 0.3402,
      "step": 19679
    },
    {
      "epoch": 0.9031251434078289,
      "grad_norm": 0.46068307757377625,
      "learning_rate": 8.104451314325581e-06,
      "loss": 0.3732,
      "step": 19680
    },
    {
      "epoch": 0.9031710339130834,
      "grad_norm": 0.4437250792980194,
      "learning_rate": 8.104259108407055e-06,
      "loss": 0.3487,
      "step": 19681
    },
    {
      "epoch": 0.9032169244183379,
      "grad_norm": 0.4706919193267822,
      "learning_rate": 8.104066895023769e-06,
      "loss": 0.4619,
      "step": 19682
    },
    {
      "epoch": 0.9032628149235923,
      "grad_norm": 0.6024381518363953,
      "learning_rate": 8.103874674176178e-06,
      "loss": 0.4279,
      "step": 19683
    },
    {
      "epoch": 0.9033087054288468,
      "grad_norm": 0.5033702850341797,
      "learning_rate": 8.10368244586475e-06,
      "loss": 0.479,
      "step": 19684
    },
    {
      "epoch": 0.9033545959341013,
      "grad_norm": 0.39673084020614624,
      "learning_rate": 8.103490210089943e-06,
      "loss": 0.3073,
      "step": 19685
    },
    {
      "epoch": 0.9034004864393557,
      "grad_norm": 0.5670091509819031,
      "learning_rate": 8.103297966852225e-06,
      "loss": 0.3645,
      "step": 19686
    },
    {
      "epoch": 0.9034463769446102,
      "grad_norm": 0.46902140974998474,
      "learning_rate": 8.103105716152053e-06,
      "loss": 0.4012,
      "step": 19687
    },
    {
      "epoch": 0.9034922674498647,
      "grad_norm": 0.4556429386138916,
      "learning_rate": 8.10291345798989e-06,
      "loss": 0.4015,
      "step": 19688
    },
    {
      "epoch": 0.903538157955119,
      "grad_norm": 0.4349612891674042,
      "learning_rate": 8.102721192366202e-06,
      "loss": 0.3649,
      "step": 19689
    },
    {
      "epoch": 0.9035840484603735,
      "grad_norm": 0.44740423560142517,
      "learning_rate": 8.102528919281447e-06,
      "loss": 0.3565,
      "step": 19690
    },
    {
      "epoch": 0.903629938965628,
      "grad_norm": 0.4618123471736908,
      "learning_rate": 8.102336638736089e-06,
      "loss": 0.3929,
      "step": 19691
    },
    {
      "epoch": 0.9036758294708824,
      "grad_norm": 0.4255295395851135,
      "learning_rate": 8.102144350730592e-06,
      "loss": 0.3454,
      "step": 19692
    },
    {
      "epoch": 0.9037217199761369,
      "grad_norm": 0.46345847845077515,
      "learning_rate": 8.101952055265415e-06,
      "loss": 0.3731,
      "step": 19693
    },
    {
      "epoch": 0.9037676104813914,
      "grad_norm": 0.4830007255077362,
      "learning_rate": 8.101759752341024e-06,
      "loss": 0.4272,
      "step": 19694
    },
    {
      "epoch": 0.9038135009866458,
      "grad_norm": 0.4533284306526184,
      "learning_rate": 8.101567441957879e-06,
      "loss": 0.4166,
      "step": 19695
    },
    {
      "epoch": 0.9038593914919003,
      "grad_norm": 0.47592705488204956,
      "learning_rate": 8.101375124116442e-06,
      "loss": 0.4573,
      "step": 19696
    },
    {
      "epoch": 0.9039052819971548,
      "grad_norm": 0.4929142892360687,
      "learning_rate": 8.10118279881718e-06,
      "loss": 0.4403,
      "step": 19697
    },
    {
      "epoch": 0.9039511725024093,
      "grad_norm": 0.4358735680580139,
      "learning_rate": 8.100990466060548e-06,
      "loss": 0.3767,
      "step": 19698
    },
    {
      "epoch": 0.9039970630076637,
      "grad_norm": 0.4009090065956116,
      "learning_rate": 8.100798125847015e-06,
      "loss": 0.3252,
      "step": 19699
    },
    {
      "epoch": 0.9040429535129182,
      "grad_norm": 0.4724183976650238,
      "learning_rate": 8.100605778177041e-06,
      "loss": 0.4651,
      "step": 19700
    },
    {
      "epoch": 0.9040888440181727,
      "grad_norm": 0.6317563652992249,
      "learning_rate": 8.100413423051088e-06,
      "loss": 0.3671,
      "step": 19701
    },
    {
      "epoch": 0.9041347345234271,
      "grad_norm": 0.4557824432849884,
      "learning_rate": 8.10022106046962e-06,
      "loss": 0.4204,
      "step": 19702
    },
    {
      "epoch": 0.9041806250286816,
      "grad_norm": 0.4439411759376526,
      "learning_rate": 8.1000286904331e-06,
      "loss": 0.4014,
      "step": 19703
    },
    {
      "epoch": 0.9042265155339361,
      "grad_norm": 0.435846209526062,
      "learning_rate": 8.099836312941987e-06,
      "loss": 0.3511,
      "step": 19704
    },
    {
      "epoch": 0.9042724060391905,
      "grad_norm": 0.47710445523262024,
      "learning_rate": 8.099643927996746e-06,
      "loss": 0.4292,
      "step": 19705
    },
    {
      "epoch": 0.904318296544445,
      "grad_norm": 0.41192251443862915,
      "learning_rate": 8.099451535597842e-06,
      "loss": 0.3224,
      "step": 19706
    },
    {
      "epoch": 0.9043641870496995,
      "grad_norm": 0.48495781421661377,
      "learning_rate": 8.099259135745732e-06,
      "loss": 0.4691,
      "step": 19707
    },
    {
      "epoch": 0.9044100775549538,
      "grad_norm": 0.4241655468940735,
      "learning_rate": 8.099066728440885e-06,
      "loss": 0.3095,
      "step": 19708
    },
    {
      "epoch": 0.9044559680602083,
      "grad_norm": 0.4737190902233124,
      "learning_rate": 8.09887431368376e-06,
      "loss": 0.4292,
      "step": 19709
    },
    {
      "epoch": 0.9045018585654628,
      "grad_norm": 0.4402461051940918,
      "learning_rate": 8.098681891474818e-06,
      "loss": 0.3813,
      "step": 19710
    },
    {
      "epoch": 0.9045477490707172,
      "grad_norm": 0.4946078956127167,
      "learning_rate": 8.098489461814527e-06,
      "loss": 0.4747,
      "step": 19711
    },
    {
      "epoch": 0.9045936395759717,
      "grad_norm": 0.4547879695892334,
      "learning_rate": 8.098297024703345e-06,
      "loss": 0.3686,
      "step": 19712
    },
    {
      "epoch": 0.9046395300812262,
      "grad_norm": 0.4980894923210144,
      "learning_rate": 8.098104580141736e-06,
      "loss": 0.4627,
      "step": 19713
    },
    {
      "epoch": 0.9046854205864806,
      "grad_norm": 0.43748903274536133,
      "learning_rate": 8.097912128130166e-06,
      "loss": 0.3812,
      "step": 19714
    },
    {
      "epoch": 0.9047313110917351,
      "grad_norm": 0.42651379108428955,
      "learning_rate": 8.097719668669092e-06,
      "loss": 0.3572,
      "step": 19715
    },
    {
      "epoch": 0.9047772015969896,
      "grad_norm": 0.4586358964443207,
      "learning_rate": 8.097527201758981e-06,
      "loss": 0.4143,
      "step": 19716
    },
    {
      "epoch": 0.9048230921022441,
      "grad_norm": 0.45888128876686096,
      "learning_rate": 8.097334727400295e-06,
      "loss": 0.4015,
      "step": 19717
    },
    {
      "epoch": 0.9048689826074985,
      "grad_norm": 0.503143846988678,
      "learning_rate": 8.097142245593497e-06,
      "loss": 0.4646,
      "step": 19718
    },
    {
      "epoch": 0.904914873112753,
      "grad_norm": 0.47953394055366516,
      "learning_rate": 8.096949756339047e-06,
      "loss": 0.5367,
      "step": 19719
    },
    {
      "epoch": 0.9049607636180075,
      "grad_norm": 0.4579585790634155,
      "learning_rate": 8.096757259637414e-06,
      "loss": 0.4313,
      "step": 19720
    },
    {
      "epoch": 0.9050066541232619,
      "grad_norm": 0.4499635696411133,
      "learning_rate": 8.096564755489053e-06,
      "loss": 0.3513,
      "step": 19721
    },
    {
      "epoch": 0.9050525446285164,
      "grad_norm": 0.4733886420726776,
      "learning_rate": 8.096372243894434e-06,
      "loss": 0.4272,
      "step": 19722
    },
    {
      "epoch": 0.9050984351337709,
      "grad_norm": 0.42385441064834595,
      "learning_rate": 8.096179724854016e-06,
      "loss": 0.3411,
      "step": 19723
    },
    {
      "epoch": 0.9051443256390252,
      "grad_norm": 0.46012550592422485,
      "learning_rate": 8.095987198368263e-06,
      "loss": 0.3899,
      "step": 19724
    },
    {
      "epoch": 0.9051902161442797,
      "grad_norm": 0.43229904770851135,
      "learning_rate": 8.095794664437638e-06,
      "loss": 0.3621,
      "step": 19725
    },
    {
      "epoch": 0.9052361066495342,
      "grad_norm": 0.4131454825401306,
      "learning_rate": 8.095602123062604e-06,
      "loss": 0.337,
      "step": 19726
    },
    {
      "epoch": 0.9052819971547886,
      "grad_norm": 0.4104025363922119,
      "learning_rate": 8.095409574243625e-06,
      "loss": 0.3026,
      "step": 19727
    },
    {
      "epoch": 0.9053278876600431,
      "grad_norm": 0.44264572858810425,
      "learning_rate": 8.095217017981163e-06,
      "loss": 0.3369,
      "step": 19728
    },
    {
      "epoch": 0.9053737781652976,
      "grad_norm": 0.44883832335472107,
      "learning_rate": 8.095024454275679e-06,
      "loss": 0.4187,
      "step": 19729
    },
    {
      "epoch": 0.905419668670552,
      "grad_norm": 0.486637681722641,
      "learning_rate": 8.094831883127639e-06,
      "loss": 0.3784,
      "step": 19730
    },
    {
      "epoch": 0.9054655591758065,
      "grad_norm": 0.44340798258781433,
      "learning_rate": 8.094639304537503e-06,
      "loss": 0.3357,
      "step": 19731
    },
    {
      "epoch": 0.905511449681061,
      "grad_norm": 0.5192316174507141,
      "learning_rate": 8.094446718505739e-06,
      "loss": 0.5257,
      "step": 19732
    },
    {
      "epoch": 0.9055573401863154,
      "grad_norm": 0.4040854871273041,
      "learning_rate": 8.094254125032806e-06,
      "loss": 0.3076,
      "step": 19733
    },
    {
      "epoch": 0.9056032306915699,
      "grad_norm": 0.4588043689727783,
      "learning_rate": 8.094061524119169e-06,
      "loss": 0.386,
      "step": 19734
    },
    {
      "epoch": 0.9056491211968244,
      "grad_norm": 0.4462362825870514,
      "learning_rate": 8.09386891576529e-06,
      "loss": 0.4073,
      "step": 19735
    },
    {
      "epoch": 0.9056950117020789,
      "grad_norm": 0.42839282751083374,
      "learning_rate": 8.093676299971632e-06,
      "loss": 0.3176,
      "step": 19736
    },
    {
      "epoch": 0.9057409022073333,
      "grad_norm": 0.46844154596328735,
      "learning_rate": 8.09348367673866e-06,
      "loss": 0.423,
      "step": 19737
    },
    {
      "epoch": 0.9057867927125878,
      "grad_norm": 0.5201075077056885,
      "learning_rate": 8.093291046066835e-06,
      "loss": 0.5594,
      "step": 19738
    },
    {
      "epoch": 0.9058326832178423,
      "grad_norm": 0.4085085093975067,
      "learning_rate": 8.093098407956624e-06,
      "loss": 0.3497,
      "step": 19739
    },
    {
      "epoch": 0.9058785737230967,
      "grad_norm": 0.4580923914909363,
      "learning_rate": 8.092905762408485e-06,
      "loss": 0.4414,
      "step": 19740
    },
    {
      "epoch": 0.9059244642283512,
      "grad_norm": 0.491268515586853,
      "learning_rate": 8.092713109422884e-06,
      "loss": 0.4009,
      "step": 19741
    },
    {
      "epoch": 0.9059703547336057,
      "grad_norm": 0.5214706063270569,
      "learning_rate": 8.092520449000285e-06,
      "loss": 0.5347,
      "step": 19742
    },
    {
      "epoch": 0.90601624523886,
      "grad_norm": 0.4822435975074768,
      "learning_rate": 8.09232778114115e-06,
      "loss": 0.5074,
      "step": 19743
    },
    {
      "epoch": 0.9060621357441145,
      "grad_norm": 0.45583540201187134,
      "learning_rate": 8.092135105845943e-06,
      "loss": 0.3695,
      "step": 19744
    },
    {
      "epoch": 0.906108026249369,
      "grad_norm": 0.4919693171977997,
      "learning_rate": 8.091942423115127e-06,
      "loss": 0.4122,
      "step": 19745
    },
    {
      "epoch": 0.9061539167546234,
      "grad_norm": 0.4304092228412628,
      "learning_rate": 8.091749732949164e-06,
      "loss": 0.3745,
      "step": 19746
    },
    {
      "epoch": 0.9061998072598779,
      "grad_norm": 0.47973886132240295,
      "learning_rate": 8.09155703534852e-06,
      "loss": 0.4426,
      "step": 19747
    },
    {
      "epoch": 0.9062456977651324,
      "grad_norm": 0.4488719701766968,
      "learning_rate": 8.091364330313657e-06,
      "loss": 0.3689,
      "step": 19748
    },
    {
      "epoch": 0.9062915882703868,
      "grad_norm": 0.4439634382724762,
      "learning_rate": 8.091171617845038e-06,
      "loss": 0.372,
      "step": 19749
    },
    {
      "epoch": 0.9063374787756413,
      "grad_norm": 0.47419968247413635,
      "learning_rate": 8.090978897943126e-06,
      "loss": 0.4593,
      "step": 19750
    },
    {
      "epoch": 0.9063833692808958,
      "grad_norm": 0.4732223153114319,
      "learning_rate": 8.090786170608387e-06,
      "loss": 0.4224,
      "step": 19751
    },
    {
      "epoch": 0.9064292597861503,
      "grad_norm": 0.4581719934940338,
      "learning_rate": 8.090593435841282e-06,
      "loss": 0.4042,
      "step": 19752
    },
    {
      "epoch": 0.9064751502914047,
      "grad_norm": 0.43798547983169556,
      "learning_rate": 8.090400693642275e-06,
      "loss": 0.3868,
      "step": 19753
    },
    {
      "epoch": 0.9065210407966592,
      "grad_norm": 0.44240084290504456,
      "learning_rate": 8.09020794401183e-06,
      "loss": 0.4101,
      "step": 19754
    },
    {
      "epoch": 0.9065669313019137,
      "grad_norm": 0.4754796326160431,
      "learning_rate": 8.09001518695041e-06,
      "loss": 0.4768,
      "step": 19755
    },
    {
      "epoch": 0.9066128218071681,
      "grad_norm": 0.49641653895378113,
      "learning_rate": 8.089822422458477e-06,
      "loss": 0.37,
      "step": 19756
    },
    {
      "epoch": 0.9066587123124226,
      "grad_norm": 0.4078812897205353,
      "learning_rate": 8.0896296505365e-06,
      "loss": 0.3272,
      "step": 19757
    },
    {
      "epoch": 0.9067046028176771,
      "grad_norm": 0.49050173163414,
      "learning_rate": 8.089436871184936e-06,
      "loss": 0.3953,
      "step": 19758
    },
    {
      "epoch": 0.9067504933229315,
      "grad_norm": 0.4416808485984802,
      "learning_rate": 8.089244084404252e-06,
      "loss": 0.4268,
      "step": 19759
    },
    {
      "epoch": 0.906796383828186,
      "grad_norm": 0.4150858521461487,
      "learning_rate": 8.089051290194912e-06,
      "loss": 0.3428,
      "step": 19760
    },
    {
      "epoch": 0.9068422743334404,
      "grad_norm": 0.48082083463668823,
      "learning_rate": 8.088858488557377e-06,
      "loss": 0.3959,
      "step": 19761
    },
    {
      "epoch": 0.9068881648386948,
      "grad_norm": 0.4783626198768616,
      "learning_rate": 8.088665679492114e-06,
      "loss": 0.4319,
      "step": 19762
    },
    {
      "epoch": 0.9069340553439493,
      "grad_norm": 0.4647321105003357,
      "learning_rate": 8.088472862999583e-06,
      "loss": 0.358,
      "step": 19763
    },
    {
      "epoch": 0.9069799458492038,
      "grad_norm": 0.4621385931968689,
      "learning_rate": 8.088280039080249e-06,
      "loss": 0.335,
      "step": 19764
    },
    {
      "epoch": 0.9070258363544582,
      "grad_norm": 0.4645345211029053,
      "learning_rate": 8.088087207734579e-06,
      "loss": 0.3836,
      "step": 19765
    },
    {
      "epoch": 0.9070717268597127,
      "grad_norm": 0.4636897146701813,
      "learning_rate": 8.087894368963032e-06,
      "loss": 0.4543,
      "step": 19766
    },
    {
      "epoch": 0.9071176173649672,
      "grad_norm": 0.48653948307037354,
      "learning_rate": 8.087701522766073e-06,
      "loss": 0.4341,
      "step": 19767
    },
    {
      "epoch": 0.9071635078702216,
      "grad_norm": 0.4266863167285919,
      "learning_rate": 8.087508669144168e-06,
      "loss": 0.3898,
      "step": 19768
    },
    {
      "epoch": 0.9072093983754761,
      "grad_norm": 0.4340168535709381,
      "learning_rate": 8.08731580809778e-06,
      "loss": 0.3109,
      "step": 19769
    },
    {
      "epoch": 0.9072552888807306,
      "grad_norm": 0.49628832936286926,
      "learning_rate": 8.08712293962737e-06,
      "loss": 0.4759,
      "step": 19770
    },
    {
      "epoch": 0.9073011793859851,
      "grad_norm": 0.520392656326294,
      "learning_rate": 8.086930063733404e-06,
      "loss": 0.4677,
      "step": 19771
    },
    {
      "epoch": 0.9073470698912395,
      "grad_norm": 0.4720556437969208,
      "learning_rate": 8.086737180416344e-06,
      "loss": 0.3397,
      "step": 19772
    },
    {
      "epoch": 0.907392960396494,
      "grad_norm": 0.5229846835136414,
      "learning_rate": 8.086544289676657e-06,
      "loss": 0.5064,
      "step": 19773
    },
    {
      "epoch": 0.9074388509017485,
      "grad_norm": 0.43044430017471313,
      "learning_rate": 8.086351391514807e-06,
      "loss": 0.3096,
      "step": 19774
    },
    {
      "epoch": 0.9074847414070029,
      "grad_norm": 0.4077827036380768,
      "learning_rate": 8.086158485931252e-06,
      "loss": 0.3027,
      "step": 19775
    },
    {
      "epoch": 0.9075306319122574,
      "grad_norm": 0.46161678433418274,
      "learning_rate": 8.085965572926461e-06,
      "loss": 0.3893,
      "step": 19776
    },
    {
      "epoch": 0.9075765224175119,
      "grad_norm": 0.4817831814289093,
      "learning_rate": 8.085772652500899e-06,
      "loss": 0.4474,
      "step": 19777
    },
    {
      "epoch": 0.9076224129227662,
      "grad_norm": 0.4783119559288025,
      "learning_rate": 8.085579724655027e-06,
      "loss": 0.4144,
      "step": 19778
    },
    {
      "epoch": 0.9076683034280207,
      "grad_norm": 0.43341249227523804,
      "learning_rate": 8.085386789389306e-06,
      "loss": 0.3478,
      "step": 19779
    },
    {
      "epoch": 0.9077141939332752,
      "grad_norm": 0.5075531005859375,
      "learning_rate": 8.085193846704206e-06,
      "loss": 0.4774,
      "step": 19780
    },
    {
      "epoch": 0.9077600844385296,
      "grad_norm": 0.4320594370365143,
      "learning_rate": 8.08500089660019e-06,
      "loss": 0.3683,
      "step": 19781
    },
    {
      "epoch": 0.9078059749437841,
      "grad_norm": 0.42616933584213257,
      "learning_rate": 8.084807939077719e-06,
      "loss": 0.3979,
      "step": 19782
    },
    {
      "epoch": 0.9078518654490386,
      "grad_norm": 0.49974194169044495,
      "learning_rate": 8.08461497413726e-06,
      "loss": 0.4522,
      "step": 19783
    },
    {
      "epoch": 0.907897755954293,
      "grad_norm": 0.4830091893672943,
      "learning_rate": 8.084422001779273e-06,
      "loss": 0.431,
      "step": 19784
    },
    {
      "epoch": 0.9079436464595475,
      "grad_norm": 0.45190754532814026,
      "learning_rate": 8.084229022004226e-06,
      "loss": 0.3531,
      "step": 19785
    },
    {
      "epoch": 0.907989536964802,
      "grad_norm": 0.5041647553443909,
      "learning_rate": 8.08403603481258e-06,
      "loss": 0.4798,
      "step": 19786
    },
    {
      "epoch": 0.9080354274700565,
      "grad_norm": 0.42994552850723267,
      "learning_rate": 8.083843040204805e-06,
      "loss": 0.364,
      "step": 19787
    },
    {
      "epoch": 0.9080813179753109,
      "grad_norm": 0.45369604229927063,
      "learning_rate": 8.083650038181357e-06,
      "loss": 0.445,
      "step": 19788
    },
    {
      "epoch": 0.9081272084805654,
      "grad_norm": 0.4129980504512787,
      "learning_rate": 8.083457028742705e-06,
      "loss": 0.3227,
      "step": 19789
    },
    {
      "epoch": 0.9081730989858199,
      "grad_norm": 0.48503991961479187,
      "learning_rate": 8.083264011889311e-06,
      "loss": 0.4515,
      "step": 19790
    },
    {
      "epoch": 0.9082189894910743,
      "grad_norm": 0.47612816095352173,
      "learning_rate": 8.083070987621641e-06,
      "loss": 0.4655,
      "step": 19791
    },
    {
      "epoch": 0.9082648799963288,
      "grad_norm": 0.4916422963142395,
      "learning_rate": 8.082877955940158e-06,
      "loss": 0.4408,
      "step": 19792
    },
    {
      "epoch": 0.9083107705015833,
      "grad_norm": 0.43666934967041016,
      "learning_rate": 8.082684916845327e-06,
      "loss": 0.3652,
      "step": 19793
    },
    {
      "epoch": 0.9083566610068377,
      "grad_norm": 0.42117840051651,
      "learning_rate": 8.082491870337612e-06,
      "loss": 0.3495,
      "step": 19794
    },
    {
      "epoch": 0.9084025515120921,
      "grad_norm": 0.4729611873626709,
      "learning_rate": 8.082298816417477e-06,
      "loss": 0.4331,
      "step": 19795
    },
    {
      "epoch": 0.9084484420173466,
      "grad_norm": 0.4573438763618469,
      "learning_rate": 8.082105755085385e-06,
      "loss": 0.4016,
      "step": 19796
    },
    {
      "epoch": 0.908494332522601,
      "grad_norm": 0.4837673604488373,
      "learning_rate": 8.081912686341802e-06,
      "loss": 0.4623,
      "step": 19797
    },
    {
      "epoch": 0.9085402230278555,
      "grad_norm": 0.4852626323699951,
      "learning_rate": 8.081719610187192e-06,
      "loss": 0.5122,
      "step": 19798
    },
    {
      "epoch": 0.90858611353311,
      "grad_norm": 0.48174750804901123,
      "learning_rate": 8.081526526622018e-06,
      "loss": 0.4641,
      "step": 19799
    },
    {
      "epoch": 0.9086320040383644,
      "grad_norm": 0.4797123670578003,
      "learning_rate": 8.081333435646745e-06,
      "loss": 0.4096,
      "step": 19800
    },
    {
      "epoch": 0.9086778945436189,
      "grad_norm": 0.4647011458873749,
      "learning_rate": 8.081140337261839e-06,
      "loss": 0.3364,
      "step": 19801
    },
    {
      "epoch": 0.9087237850488734,
      "grad_norm": 0.45662909746170044,
      "learning_rate": 8.080947231467763e-06,
      "loss": 0.3653,
      "step": 19802
    },
    {
      "epoch": 0.9087696755541278,
      "grad_norm": 0.4086991846561432,
      "learning_rate": 8.080754118264981e-06,
      "loss": 0.3084,
      "step": 19803
    },
    {
      "epoch": 0.9088155660593823,
      "grad_norm": 0.4146935045719147,
      "learning_rate": 8.080560997653956e-06,
      "loss": 0.3149,
      "step": 19804
    },
    {
      "epoch": 0.9088614565646368,
      "grad_norm": 0.49385935068130493,
      "learning_rate": 8.080367869635157e-06,
      "loss": 0.4775,
      "step": 19805
    },
    {
      "epoch": 0.9089073470698913,
      "grad_norm": 0.4818452000617981,
      "learning_rate": 8.080174734209043e-06,
      "loss": 0.5147,
      "step": 19806
    },
    {
      "epoch": 0.9089532375751457,
      "grad_norm": 0.46864286065101624,
      "learning_rate": 8.079981591376081e-06,
      "loss": 0.4153,
      "step": 19807
    },
    {
      "epoch": 0.9089991280804002,
      "grad_norm": 0.472176194190979,
      "learning_rate": 8.079788441136737e-06,
      "loss": 0.4551,
      "step": 19808
    },
    {
      "epoch": 0.9090450185856547,
      "grad_norm": 0.45830628275871277,
      "learning_rate": 8.079595283491473e-06,
      "loss": 0.3701,
      "step": 19809
    },
    {
      "epoch": 0.9090909090909091,
      "grad_norm": 0.48785656690597534,
      "learning_rate": 8.079402118440754e-06,
      "loss": 0.4689,
      "step": 19810
    },
    {
      "epoch": 0.9091367995961636,
      "grad_norm": 0.45173221826553345,
      "learning_rate": 8.079208945985044e-06,
      "loss": 0.4162,
      "step": 19811
    },
    {
      "epoch": 0.9091826901014181,
      "grad_norm": 0.4538235664367676,
      "learning_rate": 8.079015766124808e-06,
      "loss": 0.402,
      "step": 19812
    },
    {
      "epoch": 0.9092285806066724,
      "grad_norm": 0.44791096448898315,
      "learning_rate": 8.078822578860515e-06,
      "loss": 0.365,
      "step": 19813
    },
    {
      "epoch": 0.9092744711119269,
      "grad_norm": 0.47733187675476074,
      "learning_rate": 8.078629384192622e-06,
      "loss": 0.4148,
      "step": 19814
    },
    {
      "epoch": 0.9093203616171814,
      "grad_norm": 0.49350598454475403,
      "learning_rate": 8.078436182121596e-06,
      "loss": 0.4302,
      "step": 19815
    },
    {
      "epoch": 0.9093662521224358,
      "grad_norm": 0.4441482722759247,
      "learning_rate": 8.078242972647904e-06,
      "loss": 0.422,
      "step": 19816
    },
    {
      "epoch": 0.9094121426276903,
      "grad_norm": 0.4430387020111084,
      "learning_rate": 8.07804975577201e-06,
      "loss": 0.3952,
      "step": 19817
    },
    {
      "epoch": 0.9094580331329448,
      "grad_norm": 0.4480283260345459,
      "learning_rate": 8.077856531494376e-06,
      "loss": 0.4257,
      "step": 19818
    },
    {
      "epoch": 0.9095039236381992,
      "grad_norm": 0.45085176825523376,
      "learning_rate": 8.07766329981547e-06,
      "loss": 0.3773,
      "step": 19819
    },
    {
      "epoch": 0.9095498141434537,
      "grad_norm": 0.46520161628723145,
      "learning_rate": 8.077470060735755e-06,
      "loss": 0.3381,
      "step": 19820
    },
    {
      "epoch": 0.9095957046487082,
      "grad_norm": 0.4408283233642578,
      "learning_rate": 8.077276814255695e-06,
      "loss": 0.3625,
      "step": 19821
    },
    {
      "epoch": 0.9096415951539626,
      "grad_norm": 0.4771498143672943,
      "learning_rate": 8.077083560375755e-06,
      "loss": 0.4321,
      "step": 19822
    },
    {
      "epoch": 0.9096874856592171,
      "grad_norm": 0.4559750258922577,
      "learning_rate": 8.076890299096402e-06,
      "loss": 0.4335,
      "step": 19823
    },
    {
      "epoch": 0.9097333761644716,
      "grad_norm": 0.42101922631263733,
      "learning_rate": 8.076697030418099e-06,
      "loss": 0.3129,
      "step": 19824
    },
    {
      "epoch": 0.9097792666697261,
      "grad_norm": 0.4436720311641693,
      "learning_rate": 8.07650375434131e-06,
      "loss": 0.3988,
      "step": 19825
    },
    {
      "epoch": 0.9098251571749805,
      "grad_norm": 0.4155568480491638,
      "learning_rate": 8.076310470866498e-06,
      "loss": 0.3239,
      "step": 19826
    },
    {
      "epoch": 0.909871047680235,
      "grad_norm": 0.4136298894882202,
      "learning_rate": 8.076117179994134e-06,
      "loss": 0.3226,
      "step": 19827
    },
    {
      "epoch": 0.9099169381854895,
      "grad_norm": 0.45082250237464905,
      "learning_rate": 8.075923881724677e-06,
      "loss": 0.3586,
      "step": 19828
    },
    {
      "epoch": 0.9099628286907439,
      "grad_norm": 0.44058364629745483,
      "learning_rate": 8.075730576058593e-06,
      "loss": 0.3983,
      "step": 19829
    },
    {
      "epoch": 0.9100087191959984,
      "grad_norm": 0.4453901946544647,
      "learning_rate": 8.07553726299635e-06,
      "loss": 0.3627,
      "step": 19830
    },
    {
      "epoch": 0.9100546097012528,
      "grad_norm": 0.4743308126926422,
      "learning_rate": 8.075343942538409e-06,
      "loss": 0.3877,
      "step": 19831
    },
    {
      "epoch": 0.9101005002065072,
      "grad_norm": 0.44347691535949707,
      "learning_rate": 8.075150614685237e-06,
      "loss": 0.4089,
      "step": 19832
    },
    {
      "epoch": 0.9101463907117617,
      "grad_norm": 0.4939873516559601,
      "learning_rate": 8.074957279437298e-06,
      "loss": 0.5108,
      "step": 19833
    },
    {
      "epoch": 0.9101922812170162,
      "grad_norm": 0.4718095064163208,
      "learning_rate": 8.074763936795056e-06,
      "loss": 0.4908,
      "step": 19834
    },
    {
      "epoch": 0.9102381717222706,
      "grad_norm": 0.49301615357398987,
      "learning_rate": 8.07457058675898e-06,
      "loss": 0.5109,
      "step": 19835
    },
    {
      "epoch": 0.9102840622275251,
      "grad_norm": 0.46436113119125366,
      "learning_rate": 8.074377229329531e-06,
      "loss": 0.455,
      "step": 19836
    },
    {
      "epoch": 0.9103299527327796,
      "grad_norm": 0.5124197006225586,
      "learning_rate": 8.074183864507174e-06,
      "loss": 0.4732,
      "step": 19837
    },
    {
      "epoch": 0.910375843238034,
      "grad_norm": 0.4848898649215698,
      "learning_rate": 8.073990492292376e-06,
      "loss": 0.4178,
      "step": 19838
    },
    {
      "epoch": 0.9104217337432885,
      "grad_norm": 0.505601704120636,
      "learning_rate": 8.073797112685601e-06,
      "loss": 0.4688,
      "step": 19839
    },
    {
      "epoch": 0.910467624248543,
      "grad_norm": 0.47832393646240234,
      "learning_rate": 8.073603725687314e-06,
      "loss": 0.4475,
      "step": 19840
    },
    {
      "epoch": 0.9105135147537975,
      "grad_norm": 0.44735515117645264,
      "learning_rate": 8.07341033129798e-06,
      "loss": 0.3819,
      "step": 19841
    },
    {
      "epoch": 0.9105594052590519,
      "grad_norm": 0.5028278827667236,
      "learning_rate": 8.073216929518064e-06,
      "loss": 0.4476,
      "step": 19842
    },
    {
      "epoch": 0.9106052957643064,
      "grad_norm": 0.43822920322418213,
      "learning_rate": 8.073023520348032e-06,
      "loss": 0.3756,
      "step": 19843
    },
    {
      "epoch": 0.9106511862695609,
      "grad_norm": 0.4420967400074005,
      "learning_rate": 8.072830103788347e-06,
      "loss": 0.3508,
      "step": 19844
    },
    {
      "epoch": 0.9106970767748153,
      "grad_norm": 0.4544496536254883,
      "learning_rate": 8.072636679839477e-06,
      "loss": 0.3701,
      "step": 19845
    },
    {
      "epoch": 0.9107429672800698,
      "grad_norm": 0.4804290533065796,
      "learning_rate": 8.072443248501884e-06,
      "loss": 0.3595,
      "step": 19846
    },
    {
      "epoch": 0.9107888577853243,
      "grad_norm": 0.4530576169490814,
      "learning_rate": 8.072249809776035e-06,
      "loss": 0.3991,
      "step": 19847
    },
    {
      "epoch": 0.9108347482905786,
      "grad_norm": 0.4644770920276642,
      "learning_rate": 8.072056363662395e-06,
      "loss": 0.3893,
      "step": 19848
    },
    {
      "epoch": 0.9108806387958331,
      "grad_norm": 0.48486125469207764,
      "learning_rate": 8.07186291016143e-06,
      "loss": 0.4514,
      "step": 19849
    },
    {
      "epoch": 0.9109265293010876,
      "grad_norm": 0.9036334753036499,
      "learning_rate": 8.071669449273603e-06,
      "loss": 0.5016,
      "step": 19850
    },
    {
      "epoch": 0.910972419806342,
      "grad_norm": 0.4751227796077728,
      "learning_rate": 8.071475980999379e-06,
      "loss": 0.4692,
      "step": 19851
    },
    {
      "epoch": 0.9110183103115965,
      "grad_norm": 0.4260842502117157,
      "learning_rate": 8.071282505339227e-06,
      "loss": 0.353,
      "step": 19852
    },
    {
      "epoch": 0.911064200816851,
      "grad_norm": 0.47310179471969604,
      "learning_rate": 8.07108902229361e-06,
      "loss": 0.416,
      "step": 19853
    },
    {
      "epoch": 0.9111100913221054,
      "grad_norm": 0.4531238377094269,
      "learning_rate": 8.070895531862993e-06,
      "loss": 0.4025,
      "step": 19854
    },
    {
      "epoch": 0.9111559818273599,
      "grad_norm": 0.4726232886314392,
      "learning_rate": 8.070702034047839e-06,
      "loss": 0.4631,
      "step": 19855
    },
    {
      "epoch": 0.9112018723326144,
      "grad_norm": 0.4613339900970459,
      "learning_rate": 8.070508528848619e-06,
      "loss": 0.3946,
      "step": 19856
    },
    {
      "epoch": 0.9112477628378688,
      "grad_norm": 0.4331399202346802,
      "learning_rate": 8.070315016265793e-06,
      "loss": 0.3683,
      "step": 19857
    },
    {
      "epoch": 0.9112936533431233,
      "grad_norm": 0.4676276743412018,
      "learning_rate": 8.07012149629983e-06,
      "loss": 0.3971,
      "step": 19858
    },
    {
      "epoch": 0.9113395438483778,
      "grad_norm": 0.4394001066684723,
      "learning_rate": 8.069927968951193e-06,
      "loss": 0.3748,
      "step": 19859
    },
    {
      "epoch": 0.9113854343536323,
      "grad_norm": 0.42299097776412964,
      "learning_rate": 8.069734434220347e-06,
      "loss": 0.3271,
      "step": 19860
    },
    {
      "epoch": 0.9114313248588867,
      "grad_norm": 0.4486300051212311,
      "learning_rate": 8.06954089210776e-06,
      "loss": 0.3649,
      "step": 19861
    },
    {
      "epoch": 0.9114772153641412,
      "grad_norm": 0.4315125048160553,
      "learning_rate": 8.069347342613896e-06,
      "loss": 0.3226,
      "step": 19862
    },
    {
      "epoch": 0.9115231058693957,
      "grad_norm": 0.41584667563438416,
      "learning_rate": 8.06915378573922e-06,
      "loss": 0.3526,
      "step": 19863
    },
    {
      "epoch": 0.91156899637465,
      "grad_norm": 0.4659893810749054,
      "learning_rate": 8.068960221484197e-06,
      "loss": 0.4397,
      "step": 19864
    },
    {
      "epoch": 0.9116148868799046,
      "grad_norm": 0.4490303099155426,
      "learning_rate": 8.068766649849293e-06,
      "loss": 0.4026,
      "step": 19865
    },
    {
      "epoch": 0.911660777385159,
      "grad_norm": 0.46143200993537903,
      "learning_rate": 8.068573070834975e-06,
      "loss": 0.3662,
      "step": 19866
    },
    {
      "epoch": 0.9117066678904134,
      "grad_norm": 0.4608106315135956,
      "learning_rate": 8.068379484441707e-06,
      "loss": 0.4089,
      "step": 19867
    },
    {
      "epoch": 0.9117525583956679,
      "grad_norm": 0.45863357186317444,
      "learning_rate": 8.068185890669954e-06,
      "loss": 0.4211,
      "step": 19868
    },
    {
      "epoch": 0.9117984489009224,
      "grad_norm": 0.4429805874824524,
      "learning_rate": 8.067992289520184e-06,
      "loss": 0.3909,
      "step": 19869
    },
    {
      "epoch": 0.9118443394061768,
      "grad_norm": 0.46386998891830444,
      "learning_rate": 8.06779868099286e-06,
      "loss": 0.4042,
      "step": 19870
    },
    {
      "epoch": 0.9118902299114313,
      "grad_norm": 0.46973997354507446,
      "learning_rate": 8.067605065088447e-06,
      "loss": 0.4609,
      "step": 19871
    },
    {
      "epoch": 0.9119361204166858,
      "grad_norm": 0.46516549587249756,
      "learning_rate": 8.067411441807413e-06,
      "loss": 0.4079,
      "step": 19872
    },
    {
      "epoch": 0.9119820109219402,
      "grad_norm": 0.4418089985847473,
      "learning_rate": 8.067217811150224e-06,
      "loss": 0.4042,
      "step": 19873
    },
    {
      "epoch": 0.9120279014271947,
      "grad_norm": 0.41043296456336975,
      "learning_rate": 8.067024173117342e-06,
      "loss": 0.3078,
      "step": 19874
    },
    {
      "epoch": 0.9120737919324492,
      "grad_norm": 0.4313291311264038,
      "learning_rate": 8.066830527709235e-06,
      "loss": 0.3174,
      "step": 19875
    },
    {
      "epoch": 0.9121196824377037,
      "grad_norm": 0.44088274240493774,
      "learning_rate": 8.06663687492637e-06,
      "loss": 0.3989,
      "step": 19876
    },
    {
      "epoch": 0.9121655729429581,
      "grad_norm": 0.4955364763736725,
      "learning_rate": 8.06644321476921e-06,
      "loss": 0.5142,
      "step": 19877
    },
    {
      "epoch": 0.9122114634482126,
      "grad_norm": 0.44110217690467834,
      "learning_rate": 8.066249547238222e-06,
      "loss": 0.3664,
      "step": 19878
    },
    {
      "epoch": 0.9122573539534671,
      "grad_norm": 0.43643081188201904,
      "learning_rate": 8.066055872333873e-06,
      "loss": 0.3236,
      "step": 19879
    },
    {
      "epoch": 0.9123032444587215,
      "grad_norm": 0.45125144720077515,
      "learning_rate": 8.065862190056625e-06,
      "loss": 0.3877,
      "step": 19880
    },
    {
      "epoch": 0.912349134963976,
      "grad_norm": 0.43269750475883484,
      "learning_rate": 8.065668500406949e-06,
      "loss": 0.3709,
      "step": 19881
    },
    {
      "epoch": 0.9123950254692305,
      "grad_norm": 0.4689232409000397,
      "learning_rate": 8.065474803385305e-06,
      "loss": 0.4402,
      "step": 19882
    },
    {
      "epoch": 0.9124409159744848,
      "grad_norm": 0.4378909468650818,
      "learning_rate": 8.065281098992161e-06,
      "loss": 0.3454,
      "step": 19883
    },
    {
      "epoch": 0.9124868064797393,
      "grad_norm": 0.4553365707397461,
      "learning_rate": 8.065087387227986e-06,
      "loss": 0.4899,
      "step": 19884
    },
    {
      "epoch": 0.9125326969849938,
      "grad_norm": 0.5806536078453064,
      "learning_rate": 8.064893668093243e-06,
      "loss": 0.493,
      "step": 19885
    },
    {
      "epoch": 0.9125785874902482,
      "grad_norm": 0.47003448009490967,
      "learning_rate": 8.064699941588397e-06,
      "loss": 0.4509,
      "step": 19886
    },
    {
      "epoch": 0.9126244779955027,
      "grad_norm": 0.44664162397384644,
      "learning_rate": 8.064506207713916e-06,
      "loss": 0.381,
      "step": 19887
    },
    {
      "epoch": 0.9126703685007572,
      "grad_norm": 0.42585647106170654,
      "learning_rate": 8.064312466470263e-06,
      "loss": 0.3048,
      "step": 19888
    },
    {
      "epoch": 0.9127162590060116,
      "grad_norm": 0.5074831247329712,
      "learning_rate": 8.064118717857906e-06,
      "loss": 0.5234,
      "step": 19889
    },
    {
      "epoch": 0.9127621495112661,
      "grad_norm": 0.43581974506378174,
      "learning_rate": 8.063924961877311e-06,
      "loss": 0.3638,
      "step": 19890
    },
    {
      "epoch": 0.9128080400165206,
      "grad_norm": 0.4487568140029907,
      "learning_rate": 8.063731198528943e-06,
      "loss": 0.393,
      "step": 19891
    },
    {
      "epoch": 0.912853930521775,
      "grad_norm": 0.4952116012573242,
      "learning_rate": 8.063537427813271e-06,
      "loss": 0.498,
      "step": 19892
    },
    {
      "epoch": 0.9128998210270295,
      "grad_norm": 0.5095416903495789,
      "learning_rate": 8.063343649730755e-06,
      "loss": 0.4865,
      "step": 19893
    },
    {
      "epoch": 0.912945711532284,
      "grad_norm": 0.39973193407058716,
      "learning_rate": 8.063149864281865e-06,
      "loss": 0.3132,
      "step": 19894
    },
    {
      "epoch": 0.9129916020375385,
      "grad_norm": 0.44961073994636536,
      "learning_rate": 8.062956071467068e-06,
      "loss": 0.3699,
      "step": 19895
    },
    {
      "epoch": 0.9130374925427929,
      "grad_norm": 0.4452573359012604,
      "learning_rate": 8.062762271286828e-06,
      "loss": 0.3553,
      "step": 19896
    },
    {
      "epoch": 0.9130833830480474,
      "grad_norm": 0.4196516275405884,
      "learning_rate": 8.06256846374161e-06,
      "loss": 0.3343,
      "step": 19897
    },
    {
      "epoch": 0.9131292735533019,
      "grad_norm": 0.40853917598724365,
      "learning_rate": 8.062374648831883e-06,
      "loss": 0.3333,
      "step": 19898
    },
    {
      "epoch": 0.9131751640585563,
      "grad_norm": 0.4579904079437256,
      "learning_rate": 8.062180826558109e-06,
      "loss": 0.37,
      "step": 19899
    },
    {
      "epoch": 0.9132210545638108,
      "grad_norm": 0.4494280219078064,
      "learning_rate": 8.061986996920759e-06,
      "loss": 0.4049,
      "step": 19900
    },
    {
      "epoch": 0.9132669450690653,
      "grad_norm": 0.4431005120277405,
      "learning_rate": 8.061793159920295e-06,
      "loss": 0.4046,
      "step": 19901
    },
    {
      "epoch": 0.9133128355743196,
      "grad_norm": 0.4471645951271057,
      "learning_rate": 8.061599315557186e-06,
      "loss": 0.4016,
      "step": 19902
    },
    {
      "epoch": 0.9133587260795741,
      "grad_norm": 0.46120721101760864,
      "learning_rate": 8.061405463831897e-06,
      "loss": 0.4149,
      "step": 19903
    },
    {
      "epoch": 0.9134046165848286,
      "grad_norm": 0.4943949282169342,
      "learning_rate": 8.061211604744894e-06,
      "loss": 0.5121,
      "step": 19904
    },
    {
      "epoch": 0.913450507090083,
      "grad_norm": 0.4496508240699768,
      "learning_rate": 8.061017738296642e-06,
      "loss": 0.4046,
      "step": 19905
    },
    {
      "epoch": 0.9134963975953375,
      "grad_norm": 0.45679551362991333,
      "learning_rate": 8.06082386448761e-06,
      "loss": 0.4141,
      "step": 19906
    },
    {
      "epoch": 0.913542288100592,
      "grad_norm": 0.4222595691680908,
      "learning_rate": 8.060629983318261e-06,
      "loss": 0.3812,
      "step": 19907
    },
    {
      "epoch": 0.9135881786058464,
      "grad_norm": 0.4915534257888794,
      "learning_rate": 8.060436094789063e-06,
      "loss": 0.4432,
      "step": 19908
    },
    {
      "epoch": 0.9136340691111009,
      "grad_norm": 0.4558557868003845,
      "learning_rate": 8.060242198900484e-06,
      "loss": 0.4186,
      "step": 19909
    },
    {
      "epoch": 0.9136799596163554,
      "grad_norm": 0.4360322058200836,
      "learning_rate": 8.060048295652986e-06,
      "loss": 0.343,
      "step": 19910
    },
    {
      "epoch": 0.9137258501216098,
      "grad_norm": 0.46147429943084717,
      "learning_rate": 8.059854385047039e-06,
      "loss": 0.3794,
      "step": 19911
    },
    {
      "epoch": 0.9137717406268643,
      "grad_norm": 0.47554031014442444,
      "learning_rate": 8.059660467083107e-06,
      "loss": 0.4283,
      "step": 19912
    },
    {
      "epoch": 0.9138176311321188,
      "grad_norm": 0.4555242955684662,
      "learning_rate": 8.05946654176166e-06,
      "loss": 0.4083,
      "step": 19913
    },
    {
      "epoch": 0.9138635216373733,
      "grad_norm": 0.45906832814216614,
      "learning_rate": 8.059272609083158e-06,
      "loss": 0.3667,
      "step": 19914
    },
    {
      "epoch": 0.9139094121426277,
      "grad_norm": 0.49573925137519836,
      "learning_rate": 8.059078669048074e-06,
      "loss": 0.4579,
      "step": 19915
    },
    {
      "epoch": 0.9139553026478822,
      "grad_norm": 0.4538806676864624,
      "learning_rate": 8.058884721656868e-06,
      "loss": 0.415,
      "step": 19916
    },
    {
      "epoch": 0.9140011931531367,
      "grad_norm": 0.4226219058036804,
      "learning_rate": 8.058690766910013e-06,
      "loss": 0.3171,
      "step": 19917
    },
    {
      "epoch": 0.914047083658391,
      "grad_norm": 0.6014626622200012,
      "learning_rate": 8.058496804807968e-06,
      "loss": 0.2844,
      "step": 19918
    },
    {
      "epoch": 0.9140929741636455,
      "grad_norm": 0.4240710139274597,
      "learning_rate": 8.058302835351208e-06,
      "loss": 0.37,
      "step": 19919
    },
    {
      "epoch": 0.9141388646689,
      "grad_norm": 0.5129101276397705,
      "learning_rate": 8.058108858540192e-06,
      "loss": 0.464,
      "step": 19920
    },
    {
      "epoch": 0.9141847551741544,
      "grad_norm": 0.4401739835739136,
      "learning_rate": 8.05791487437539e-06,
      "loss": 0.3532,
      "step": 19921
    },
    {
      "epoch": 0.9142306456794089,
      "grad_norm": 0.46502354741096497,
      "learning_rate": 8.057720882857268e-06,
      "loss": 0.4096,
      "step": 19922
    },
    {
      "epoch": 0.9142765361846634,
      "grad_norm": 0.4845057725906372,
      "learning_rate": 8.057526883986293e-06,
      "loss": 0.4383,
      "step": 19923
    },
    {
      "epoch": 0.9143224266899178,
      "grad_norm": 0.4627869129180908,
      "learning_rate": 8.05733287776293e-06,
      "loss": 0.3899,
      "step": 19924
    },
    {
      "epoch": 0.9143683171951723,
      "grad_norm": 0.41061171889305115,
      "learning_rate": 8.057138864187646e-06,
      "loss": 0.3357,
      "step": 19925
    },
    {
      "epoch": 0.9144142077004268,
      "grad_norm": 0.4645308554172516,
      "learning_rate": 8.056944843260909e-06,
      "loss": 0.3974,
      "step": 19926
    },
    {
      "epoch": 0.9144600982056812,
      "grad_norm": 0.48850756883621216,
      "learning_rate": 8.056750814983183e-06,
      "loss": 0.478,
      "step": 19927
    },
    {
      "epoch": 0.9145059887109357,
      "grad_norm": 0.44728296995162964,
      "learning_rate": 8.056556779354938e-06,
      "loss": 0.3874,
      "step": 19928
    },
    {
      "epoch": 0.9145518792161902,
      "grad_norm": 0.4454975128173828,
      "learning_rate": 8.056362736376637e-06,
      "loss": 0.3744,
      "step": 19929
    },
    {
      "epoch": 0.9145977697214447,
      "grad_norm": 0.5057328343391418,
      "learning_rate": 8.056168686048749e-06,
      "loss": 0.471,
      "step": 19930
    },
    {
      "epoch": 0.9146436602266991,
      "grad_norm": 0.4555968940258026,
      "learning_rate": 8.055974628371739e-06,
      "loss": 0.4284,
      "step": 19931
    },
    {
      "epoch": 0.9146895507319536,
      "grad_norm": 0.4612051844596863,
      "learning_rate": 8.055780563346076e-06,
      "loss": 0.3834,
      "step": 19932
    },
    {
      "epoch": 0.9147354412372081,
      "grad_norm": 0.4492006301879883,
      "learning_rate": 8.055586490972224e-06,
      "loss": 0.4029,
      "step": 19933
    },
    {
      "epoch": 0.9147813317424625,
      "grad_norm": 0.48880547285079956,
      "learning_rate": 8.055392411250651e-06,
      "loss": 0.3611,
      "step": 19934
    },
    {
      "epoch": 0.914827222247717,
      "grad_norm": 0.47262459993362427,
      "learning_rate": 8.055198324181823e-06,
      "loss": 0.4158,
      "step": 19935
    },
    {
      "epoch": 0.9148731127529715,
      "grad_norm": 0.44569703936576843,
      "learning_rate": 8.055004229766207e-06,
      "loss": 0.3799,
      "step": 19936
    },
    {
      "epoch": 0.9149190032582258,
      "grad_norm": 0.4228842258453369,
      "learning_rate": 8.054810128004273e-06,
      "loss": 0.3545,
      "step": 19937
    },
    {
      "epoch": 0.9149648937634803,
      "grad_norm": 0.42943331599235535,
      "learning_rate": 8.054616018896481e-06,
      "loss": 0.3312,
      "step": 19938
    },
    {
      "epoch": 0.9150107842687348,
      "grad_norm": 0.48538121581077576,
      "learning_rate": 8.054421902443304e-06,
      "loss": 0.4179,
      "step": 19939
    },
    {
      "epoch": 0.9150566747739892,
      "grad_norm": 0.4716055989265442,
      "learning_rate": 8.054227778645205e-06,
      "loss": 0.3715,
      "step": 19940
    },
    {
      "epoch": 0.9151025652792437,
      "grad_norm": 0.4691489040851593,
      "learning_rate": 8.054033647502654e-06,
      "loss": 0.3795,
      "step": 19941
    },
    {
      "epoch": 0.9151484557844982,
      "grad_norm": 0.4136064350605011,
      "learning_rate": 8.053839509016116e-06,
      "loss": 0.351,
      "step": 19942
    },
    {
      "epoch": 0.9151943462897526,
      "grad_norm": 0.4332267642021179,
      "learning_rate": 8.053645363186056e-06,
      "loss": 0.3902,
      "step": 19943
    },
    {
      "epoch": 0.9152402367950071,
      "grad_norm": 0.4606090486049652,
      "learning_rate": 8.053451210012942e-06,
      "loss": 0.3707,
      "step": 19944
    },
    {
      "epoch": 0.9152861273002616,
      "grad_norm": 0.47834905982017517,
      "learning_rate": 8.053257049497243e-06,
      "loss": 0.4658,
      "step": 19945
    },
    {
      "epoch": 0.915332017805516,
      "grad_norm": 0.4282083511352539,
      "learning_rate": 8.053062881639424e-06,
      "loss": 0.3798,
      "step": 19946
    },
    {
      "epoch": 0.9153779083107705,
      "grad_norm": 0.4588189423084259,
      "learning_rate": 8.052868706439953e-06,
      "loss": 0.4247,
      "step": 19947
    },
    {
      "epoch": 0.915423798816025,
      "grad_norm": 0.5361977815628052,
      "learning_rate": 8.052674523899296e-06,
      "loss": 0.3573,
      "step": 19948
    },
    {
      "epoch": 0.9154696893212795,
      "grad_norm": 0.46259623765945435,
      "learning_rate": 8.05248033401792e-06,
      "loss": 0.4022,
      "step": 19949
    },
    {
      "epoch": 0.9155155798265339,
      "grad_norm": 0.4493080675601959,
      "learning_rate": 8.052286136796293e-06,
      "loss": 0.4291,
      "step": 19950
    },
    {
      "epoch": 0.9155614703317884,
      "grad_norm": 0.49828290939331055,
      "learning_rate": 8.052091932234877e-06,
      "loss": 0.5224,
      "step": 19951
    },
    {
      "epoch": 0.9156073608370429,
      "grad_norm": 0.4824344515800476,
      "learning_rate": 8.05189772033415e-06,
      "loss": 0.4669,
      "step": 19952
    },
    {
      "epoch": 0.9156532513422972,
      "grad_norm": 0.40372446179389954,
      "learning_rate": 8.051703501094567e-06,
      "loss": 0.2576,
      "step": 19953
    },
    {
      "epoch": 0.9156991418475517,
      "grad_norm": 0.4589158594608307,
      "learning_rate": 8.051509274516602e-06,
      "loss": 0.4159,
      "step": 19954
    },
    {
      "epoch": 0.9157450323528062,
      "grad_norm": 0.46112895011901855,
      "learning_rate": 8.05131504060072e-06,
      "loss": 0.3726,
      "step": 19955
    },
    {
      "epoch": 0.9157909228580606,
      "grad_norm": 0.4675448536872864,
      "learning_rate": 8.051120799347389e-06,
      "loss": 0.3208,
      "step": 19956
    },
    {
      "epoch": 0.9158368133633151,
      "grad_norm": 0.4543447494506836,
      "learning_rate": 8.050926550757074e-06,
      "loss": 0.3947,
      "step": 19957
    },
    {
      "epoch": 0.9158827038685696,
      "grad_norm": 0.44077757000923157,
      "learning_rate": 8.050732294830246e-06,
      "loss": 0.3527,
      "step": 19958
    },
    {
      "epoch": 0.915928594373824,
      "grad_norm": 0.42064371705055237,
      "learning_rate": 8.050538031567367e-06,
      "loss": 0.3216,
      "step": 19959
    },
    {
      "epoch": 0.9159744848790785,
      "grad_norm": 0.492087721824646,
      "learning_rate": 8.050343760968908e-06,
      "loss": 0.4724,
      "step": 19960
    },
    {
      "epoch": 0.916020375384333,
      "grad_norm": 0.46470850706100464,
      "learning_rate": 8.050149483035336e-06,
      "loss": 0.4136,
      "step": 19961
    },
    {
      "epoch": 0.9160662658895874,
      "grad_norm": 0.44022080302238464,
      "learning_rate": 8.049955197767115e-06,
      "loss": 0.3909,
      "step": 19962
    },
    {
      "epoch": 0.9161121563948419,
      "grad_norm": 0.4655168056488037,
      "learning_rate": 8.049760905164716e-06,
      "loss": 0.428,
      "step": 19963
    },
    {
      "epoch": 0.9161580469000964,
      "grad_norm": 0.42304664850234985,
      "learning_rate": 8.049566605228606e-06,
      "loss": 0.3525,
      "step": 19964
    },
    {
      "epoch": 0.9162039374053509,
      "grad_norm": 0.4397325813770294,
      "learning_rate": 8.049372297959246e-06,
      "loss": 0.3929,
      "step": 19965
    },
    {
      "epoch": 0.9162498279106053,
      "grad_norm": 0.4821664094924927,
      "learning_rate": 8.049177983357113e-06,
      "loss": 0.3569,
      "step": 19966
    },
    {
      "epoch": 0.9162957184158598,
      "grad_norm": 0.45952117443084717,
      "learning_rate": 8.048983661422667e-06,
      "loss": 0.4112,
      "step": 19967
    },
    {
      "epoch": 0.9163416089211143,
      "grad_norm": 0.41459429264068604,
      "learning_rate": 8.048789332156377e-06,
      "loss": 0.3127,
      "step": 19968
    },
    {
      "epoch": 0.9163874994263687,
      "grad_norm": 0.4986949563026428,
      "learning_rate": 8.048594995558712e-06,
      "loss": 0.4403,
      "step": 19969
    },
    {
      "epoch": 0.9164333899316232,
      "grad_norm": 0.4376751482486725,
      "learning_rate": 8.048400651630136e-06,
      "loss": 0.3592,
      "step": 19970
    },
    {
      "epoch": 0.9164792804368777,
      "grad_norm": 0.4477252960205078,
      "learning_rate": 8.048206300371122e-06,
      "loss": 0.3843,
      "step": 19971
    },
    {
      "epoch": 0.916525170942132,
      "grad_norm": 0.44658413529396057,
      "learning_rate": 8.048011941782132e-06,
      "loss": 0.4374,
      "step": 19972
    },
    {
      "epoch": 0.9165710614473865,
      "grad_norm": 0.4536164402961731,
      "learning_rate": 8.047817575863634e-06,
      "loss": 0.4159,
      "step": 19973
    },
    {
      "epoch": 0.916616951952641,
      "grad_norm": 0.49663740396499634,
      "learning_rate": 8.0476232026161e-06,
      "loss": 0.4357,
      "step": 19974
    },
    {
      "epoch": 0.9166628424578954,
      "grad_norm": 0.48059847950935364,
      "learning_rate": 8.047428822039992e-06,
      "loss": 0.425,
      "step": 19975
    },
    {
      "epoch": 0.9167087329631499,
      "grad_norm": 0.41684722900390625,
      "learning_rate": 8.04723443413578e-06,
      "loss": 0.3373,
      "step": 19976
    },
    {
      "epoch": 0.9167546234684044,
      "grad_norm": 0.4607388973236084,
      "learning_rate": 8.04704003890393e-06,
      "loss": 0.4456,
      "step": 19977
    },
    {
      "epoch": 0.9168005139736588,
      "grad_norm": 0.49048784375190735,
      "learning_rate": 8.046845636344912e-06,
      "loss": 0.5197,
      "step": 19978
    },
    {
      "epoch": 0.9168464044789133,
      "grad_norm": 0.485093355178833,
      "learning_rate": 8.04665122645919e-06,
      "loss": 0.399,
      "step": 19979
    },
    {
      "epoch": 0.9168922949841678,
      "grad_norm": 0.46711257100105286,
      "learning_rate": 8.046456809247234e-06,
      "loss": 0.4088,
      "step": 19980
    },
    {
      "epoch": 0.9169381854894222,
      "grad_norm": 0.4614191949367523,
      "learning_rate": 8.04626238470951e-06,
      "loss": 0.4342,
      "step": 19981
    },
    {
      "epoch": 0.9169840759946767,
      "grad_norm": 0.47194069623947144,
      "learning_rate": 8.046067952846487e-06,
      "loss": 0.4618,
      "step": 19982
    },
    {
      "epoch": 0.9170299664999312,
      "grad_norm": 0.41760337352752686,
      "learning_rate": 8.045873513658634e-06,
      "loss": 0.3651,
      "step": 19983
    },
    {
      "epoch": 0.9170758570051857,
      "grad_norm": 0.4043579697608948,
      "learning_rate": 8.045679067146415e-06,
      "loss": 0.3126,
      "step": 19984
    },
    {
      "epoch": 0.9171217475104401,
      "grad_norm": 0.4702475070953369,
      "learning_rate": 8.0454846133103e-06,
      "loss": 0.4424,
      "step": 19985
    },
    {
      "epoch": 0.9171676380156946,
      "grad_norm": 0.5053181052207947,
      "learning_rate": 8.045290152150753e-06,
      "loss": 0.4882,
      "step": 19986
    },
    {
      "epoch": 0.9172135285209491,
      "grad_norm": 0.40354758501052856,
      "learning_rate": 8.045095683668247e-06,
      "loss": 0.3292,
      "step": 19987
    },
    {
      "epoch": 0.9172594190262034,
      "grad_norm": 0.45261818170547485,
      "learning_rate": 8.044901207863245e-06,
      "loss": 0.4013,
      "step": 19988
    },
    {
      "epoch": 0.917305309531458,
      "grad_norm": 0.4700707495212555,
      "learning_rate": 8.044706724736217e-06,
      "loss": 0.4063,
      "step": 19989
    },
    {
      "epoch": 0.9173512000367124,
      "grad_norm": 0.4582161605358124,
      "learning_rate": 8.044512234287631e-06,
      "loss": 0.4639,
      "step": 19990
    },
    {
      "epoch": 0.9173970905419668,
      "grad_norm": 0.42430198192596436,
      "learning_rate": 8.044317736517955e-06,
      "loss": 0.3375,
      "step": 19991
    },
    {
      "epoch": 0.9174429810472213,
      "grad_norm": 0.44287121295928955,
      "learning_rate": 8.044123231427654e-06,
      "loss": 0.3383,
      "step": 19992
    },
    {
      "epoch": 0.9174888715524758,
      "grad_norm": 0.4396236836910248,
      "learning_rate": 8.043928719017198e-06,
      "loss": 0.3898,
      "step": 19993
    },
    {
      "epoch": 0.9175347620577302,
      "grad_norm": 0.4521273374557495,
      "learning_rate": 8.043734199287054e-06,
      "loss": 0.3822,
      "step": 19994
    },
    {
      "epoch": 0.9175806525629847,
      "grad_norm": 0.4202002286911011,
      "learning_rate": 8.043539672237689e-06,
      "loss": 0.3467,
      "step": 19995
    },
    {
      "epoch": 0.9176265430682392,
      "grad_norm": 0.46127039194107056,
      "learning_rate": 8.043345137869572e-06,
      "loss": 0.4181,
      "step": 19996
    },
    {
      "epoch": 0.9176724335734936,
      "grad_norm": 0.484381765127182,
      "learning_rate": 8.043150596183172e-06,
      "loss": 0.4604,
      "step": 19997
    },
    {
      "epoch": 0.9177183240787481,
      "grad_norm": 0.4581066071987152,
      "learning_rate": 8.042956047178953e-06,
      "loss": 0.3909,
      "step": 19998
    },
    {
      "epoch": 0.9177642145840026,
      "grad_norm": 0.4827679991722107,
      "learning_rate": 8.042761490857387e-06,
      "loss": 0.4763,
      "step": 19999
    },
    {
      "epoch": 0.917810105089257,
      "grad_norm": 0.4396640658378601,
      "learning_rate": 8.04256692721894e-06,
      "loss": 0.3916,
      "step": 20000
    },
    {
      "epoch": 0.9178559955945115,
      "grad_norm": 0.4575493037700653,
      "learning_rate": 8.042372356264078e-06,
      "loss": 0.459,
      "step": 20001
    },
    {
      "epoch": 0.917901886099766,
      "grad_norm": 0.43947094678878784,
      "learning_rate": 8.042177777993272e-06,
      "loss": 0.3648,
      "step": 20002
    },
    {
      "epoch": 0.9179477766050205,
      "grad_norm": 0.46974506974220276,
      "learning_rate": 8.041983192406989e-06,
      "loss": 0.3986,
      "step": 20003
    },
    {
      "epoch": 0.9179936671102749,
      "grad_norm": 0.4219629466533661,
      "learning_rate": 8.041788599505695e-06,
      "loss": 0.3473,
      "step": 20004
    },
    {
      "epoch": 0.9180395576155294,
      "grad_norm": 0.4444812536239624,
      "learning_rate": 8.041593999289859e-06,
      "loss": 0.3948,
      "step": 20005
    },
    {
      "epoch": 0.9180854481207839,
      "grad_norm": 0.42322418093681335,
      "learning_rate": 8.041399391759953e-06,
      "loss": 0.3511,
      "step": 20006
    },
    {
      "epoch": 0.9181313386260382,
      "grad_norm": 0.46399614214897156,
      "learning_rate": 8.041204776916439e-06,
      "loss": 0.4547,
      "step": 20007
    },
    {
      "epoch": 0.9181772291312927,
      "grad_norm": 0.4384278953075409,
      "learning_rate": 8.041010154759786e-06,
      "loss": 0.396,
      "step": 20008
    },
    {
      "epoch": 0.9182231196365472,
      "grad_norm": 0.434222936630249,
      "learning_rate": 8.040815525290464e-06,
      "loss": 0.3772,
      "step": 20009
    },
    {
      "epoch": 0.9182690101418016,
      "grad_norm": 0.4341912865638733,
      "learning_rate": 8.04062088850894e-06,
      "loss": 0.3813,
      "step": 20010
    },
    {
      "epoch": 0.9183149006470561,
      "grad_norm": 0.4584401547908783,
      "learning_rate": 8.040426244415685e-06,
      "loss": 0.3765,
      "step": 20011
    },
    {
      "epoch": 0.9183607911523106,
      "grad_norm": 0.4477466940879822,
      "learning_rate": 8.040231593011162e-06,
      "loss": 0.354,
      "step": 20012
    },
    {
      "epoch": 0.918406681657565,
      "grad_norm": 0.4670342803001404,
      "learning_rate": 8.040036934295843e-06,
      "loss": 0.3703,
      "step": 20013
    },
    {
      "epoch": 0.9184525721628195,
      "grad_norm": 0.4566385746002197,
      "learning_rate": 8.039842268270195e-06,
      "loss": 0.3713,
      "step": 20014
    },
    {
      "epoch": 0.918498462668074,
      "grad_norm": 0.4833486080169678,
      "learning_rate": 8.039647594934684e-06,
      "loss": 0.4588,
      "step": 20015
    },
    {
      "epoch": 0.9185443531733284,
      "grad_norm": 0.4764838218688965,
      "learning_rate": 8.039452914289781e-06,
      "loss": 0.3828,
      "step": 20016
    },
    {
      "epoch": 0.9185902436785829,
      "grad_norm": 0.5860703587532043,
      "learning_rate": 8.039258226335951e-06,
      "loss": 0.3878,
      "step": 20017
    },
    {
      "epoch": 0.9186361341838374,
      "grad_norm": 0.4405742883682251,
      "learning_rate": 8.039063531073667e-06,
      "loss": 0.3742,
      "step": 20018
    },
    {
      "epoch": 0.9186820246890919,
      "grad_norm": 0.4318040609359741,
      "learning_rate": 8.038868828503393e-06,
      "loss": 0.3583,
      "step": 20019
    },
    {
      "epoch": 0.9187279151943463,
      "grad_norm": 0.44224053621292114,
      "learning_rate": 8.038674118625601e-06,
      "loss": 0.3587,
      "step": 20020
    },
    {
      "epoch": 0.9187738056996008,
      "grad_norm": 0.42527085542678833,
      "learning_rate": 8.038479401440752e-06,
      "loss": 0.353,
      "step": 20021
    },
    {
      "epoch": 0.9188196962048553,
      "grad_norm": 0.43495824933052063,
      "learning_rate": 8.038284676949321e-06,
      "loss": 0.3638,
      "step": 20022
    },
    {
      "epoch": 0.9188655867101096,
      "grad_norm": 0.48858779668807983,
      "learning_rate": 8.038089945151777e-06,
      "loss": 0.5264,
      "step": 20023
    },
    {
      "epoch": 0.9189114772153641,
      "grad_norm": 0.4363512396812439,
      "learning_rate": 8.037895206048582e-06,
      "loss": 0.3412,
      "step": 20024
    },
    {
      "epoch": 0.9189573677206186,
      "grad_norm": 0.4311063289642334,
      "learning_rate": 8.037700459640209e-06,
      "loss": 0.3787,
      "step": 20025
    },
    {
      "epoch": 0.919003258225873,
      "grad_norm": 0.47028255462646484,
      "learning_rate": 8.037505705927127e-06,
      "loss": 0.3232,
      "step": 20026
    },
    {
      "epoch": 0.9190491487311275,
      "grad_norm": 0.42899638414382935,
      "learning_rate": 8.037310944909799e-06,
      "loss": 0.4066,
      "step": 20027
    },
    {
      "epoch": 0.919095039236382,
      "grad_norm": 0.6811000108718872,
      "learning_rate": 8.037116176588699e-06,
      "loss": 0.4209,
      "step": 20028
    },
    {
      "epoch": 0.9191409297416364,
      "grad_norm": 0.42036280035972595,
      "learning_rate": 8.036921400964293e-06,
      "loss": 0.3289,
      "step": 20029
    },
    {
      "epoch": 0.9191868202468909,
      "grad_norm": 0.4487990438938141,
      "learning_rate": 8.03672661803705e-06,
      "loss": 0.364,
      "step": 20030
    },
    {
      "epoch": 0.9192327107521454,
      "grad_norm": 0.4839121699333191,
      "learning_rate": 8.036531827807437e-06,
      "loss": 0.4054,
      "step": 20031
    },
    {
      "epoch": 0.9192786012573998,
      "grad_norm": 0.450795441865921,
      "learning_rate": 8.036337030275923e-06,
      "loss": 0.3916,
      "step": 20032
    },
    {
      "epoch": 0.9193244917626543,
      "grad_norm": 0.4859609603881836,
      "learning_rate": 8.036142225442976e-06,
      "loss": 0.4448,
      "step": 20033
    },
    {
      "epoch": 0.9193703822679088,
      "grad_norm": 0.4316902458667755,
      "learning_rate": 8.035947413309068e-06,
      "loss": 0.3317,
      "step": 20034
    },
    {
      "epoch": 0.9194162727731632,
      "grad_norm": 0.5015361905097961,
      "learning_rate": 8.035752593874661e-06,
      "loss": 0.48,
      "step": 20035
    },
    {
      "epoch": 0.9194621632784177,
      "grad_norm": 0.5162017941474915,
      "learning_rate": 8.03555776714023e-06,
      "loss": 0.4559,
      "step": 20036
    },
    {
      "epoch": 0.9195080537836722,
      "grad_norm": 0.4470103681087494,
      "learning_rate": 8.03536293310624e-06,
      "loss": 0.399,
      "step": 20037
    },
    {
      "epoch": 0.9195539442889267,
      "grad_norm": 0.4689035415649414,
      "learning_rate": 8.03516809177316e-06,
      "loss": 0.4016,
      "step": 20038
    },
    {
      "epoch": 0.9195998347941811,
      "grad_norm": 0.4740046262741089,
      "learning_rate": 8.034973243141458e-06,
      "loss": 0.4962,
      "step": 20039
    },
    {
      "epoch": 0.9196457252994356,
      "grad_norm": 0.43549588322639465,
      "learning_rate": 8.034778387211604e-06,
      "loss": 0.3851,
      "step": 20040
    },
    {
      "epoch": 0.91969161580469,
      "grad_norm": 0.4624907076358795,
      "learning_rate": 8.034583523984066e-06,
      "loss": 0.3467,
      "step": 20041
    },
    {
      "epoch": 0.9197375063099444,
      "grad_norm": 0.4143589735031128,
      "learning_rate": 8.034388653459311e-06,
      "loss": 0.2863,
      "step": 20042
    },
    {
      "epoch": 0.9197833968151989,
      "grad_norm": 0.47427281737327576,
      "learning_rate": 8.03419377563781e-06,
      "loss": 0.4501,
      "step": 20043
    },
    {
      "epoch": 0.9198292873204534,
      "grad_norm": 0.43546831607818604,
      "learning_rate": 8.033998890520031e-06,
      "loss": 0.373,
      "step": 20044
    },
    {
      "epoch": 0.9198751778257078,
      "grad_norm": 0.48160433769226074,
      "learning_rate": 8.033803998106443e-06,
      "loss": 0.3408,
      "step": 20045
    },
    {
      "epoch": 0.9199210683309623,
      "grad_norm": 0.46631836891174316,
      "learning_rate": 8.03360909839751e-06,
      "loss": 0.4566,
      "step": 20046
    },
    {
      "epoch": 0.9199669588362168,
      "grad_norm": 0.43506160378456116,
      "learning_rate": 8.033414191393706e-06,
      "loss": 0.3533,
      "step": 20047
    },
    {
      "epoch": 0.9200128493414712,
      "grad_norm": 0.4559778869152069,
      "learning_rate": 8.0332192770955e-06,
      "loss": 0.3432,
      "step": 20048
    },
    {
      "epoch": 0.9200587398467257,
      "grad_norm": 0.4768766760826111,
      "learning_rate": 8.033024355503358e-06,
      "loss": 0.4827,
      "step": 20049
    },
    {
      "epoch": 0.9201046303519802,
      "grad_norm": 0.44587239623069763,
      "learning_rate": 8.03282942661775e-06,
      "loss": 0.3481,
      "step": 20050
    },
    {
      "epoch": 0.9201505208572346,
      "grad_norm": 0.44141775369644165,
      "learning_rate": 8.032634490439145e-06,
      "loss": 0.3577,
      "step": 20051
    },
    {
      "epoch": 0.9201964113624891,
      "grad_norm": 0.4313547909259796,
      "learning_rate": 8.032439546968012e-06,
      "loss": 0.3614,
      "step": 20052
    },
    {
      "epoch": 0.9202423018677436,
      "grad_norm": 0.47053706645965576,
      "learning_rate": 8.032244596204817e-06,
      "loss": 0.4278,
      "step": 20053
    },
    {
      "epoch": 0.9202881923729981,
      "grad_norm": 0.39661288261413574,
      "learning_rate": 8.032049638150031e-06,
      "loss": 0.2892,
      "step": 20054
    },
    {
      "epoch": 0.9203340828782525,
      "grad_norm": 0.4752148687839508,
      "learning_rate": 8.031854672804122e-06,
      "loss": 0.4296,
      "step": 20055
    },
    {
      "epoch": 0.920379973383507,
      "grad_norm": 0.4749104082584381,
      "learning_rate": 8.03165970016756e-06,
      "loss": 0.4455,
      "step": 20056
    },
    {
      "epoch": 0.9204258638887615,
      "grad_norm": 0.4639680087566376,
      "learning_rate": 8.031464720240813e-06,
      "loss": 0.3953,
      "step": 20057
    },
    {
      "epoch": 0.9204717543940159,
      "grad_norm": 0.4204411804676056,
      "learning_rate": 8.031269733024351e-06,
      "loss": 0.3213,
      "step": 20058
    },
    {
      "epoch": 0.9205176448992703,
      "grad_norm": 0.45318472385406494,
      "learning_rate": 8.031074738518642e-06,
      "loss": 0.4161,
      "step": 20059
    },
    {
      "epoch": 0.9205635354045248,
      "grad_norm": 0.4550301134586334,
      "learning_rate": 8.030879736724155e-06,
      "loss": 0.4025,
      "step": 20060
    },
    {
      "epoch": 0.9206094259097792,
      "grad_norm": 0.42499274015426636,
      "learning_rate": 8.030684727641358e-06,
      "loss": 0.3417,
      "step": 20061
    },
    {
      "epoch": 0.9206553164150337,
      "grad_norm": 0.4087652266025543,
      "learning_rate": 8.030489711270722e-06,
      "loss": 0.2857,
      "step": 20062
    },
    {
      "epoch": 0.9207012069202882,
      "grad_norm": 0.4118112325668335,
      "learning_rate": 8.030294687612712e-06,
      "loss": 0.3055,
      "step": 20063
    },
    {
      "epoch": 0.9207470974255426,
      "grad_norm": 0.48875170946121216,
      "learning_rate": 8.030099656667802e-06,
      "loss": 0.5009,
      "step": 20064
    },
    {
      "epoch": 0.9207929879307971,
      "grad_norm": 0.4298875033855438,
      "learning_rate": 8.029904618436458e-06,
      "loss": 0.3868,
      "step": 20065
    },
    {
      "epoch": 0.9208388784360516,
      "grad_norm": 0.42632368206977844,
      "learning_rate": 8.02970957291915e-06,
      "loss": 0.3502,
      "step": 20066
    },
    {
      "epoch": 0.920884768941306,
      "grad_norm": 0.4784495532512665,
      "learning_rate": 8.029514520116346e-06,
      "loss": 0.4702,
      "step": 20067
    },
    {
      "epoch": 0.9209306594465605,
      "grad_norm": 0.44892582297325134,
      "learning_rate": 8.029319460028516e-06,
      "loss": 0.3774,
      "step": 20068
    },
    {
      "epoch": 0.920976549951815,
      "grad_norm": 0.4614852964878082,
      "learning_rate": 8.029124392656129e-06,
      "loss": 0.4035,
      "step": 20069
    },
    {
      "epoch": 0.9210224404570694,
      "grad_norm": 0.44602301716804504,
      "learning_rate": 8.028929317999653e-06,
      "loss": 0.3819,
      "step": 20070
    },
    {
      "epoch": 0.9210683309623239,
      "grad_norm": 0.47170859575271606,
      "learning_rate": 8.028734236059559e-06,
      "loss": 0.4034,
      "step": 20071
    },
    {
      "epoch": 0.9211142214675784,
      "grad_norm": 0.44395312666893005,
      "learning_rate": 8.028539146836315e-06,
      "loss": 0.3904,
      "step": 20072
    },
    {
      "epoch": 0.9211601119728329,
      "grad_norm": 0.4048980176448822,
      "learning_rate": 8.02834405033039e-06,
      "loss": 0.2981,
      "step": 20073
    },
    {
      "epoch": 0.9212060024780873,
      "grad_norm": 0.43619513511657715,
      "learning_rate": 8.028148946542254e-06,
      "loss": 0.3579,
      "step": 20074
    },
    {
      "epoch": 0.9212518929833418,
      "grad_norm": 0.46999847888946533,
      "learning_rate": 8.027953835472374e-06,
      "loss": 0.4417,
      "step": 20075
    },
    {
      "epoch": 0.9212977834885963,
      "grad_norm": 0.48393091559410095,
      "learning_rate": 8.027758717121222e-06,
      "loss": 0.4378,
      "step": 20076
    },
    {
      "epoch": 0.9213436739938506,
      "grad_norm": 0.47350654006004333,
      "learning_rate": 8.027563591489265e-06,
      "loss": 0.4778,
      "step": 20077
    },
    {
      "epoch": 0.9213895644991051,
      "grad_norm": 0.48437902331352234,
      "learning_rate": 8.027368458576971e-06,
      "loss": 0.4554,
      "step": 20078
    },
    {
      "epoch": 0.9214354550043596,
      "grad_norm": 0.43889960646629333,
      "learning_rate": 8.027173318384816e-06,
      "loss": 0.3562,
      "step": 20079
    },
    {
      "epoch": 0.921481345509614,
      "grad_norm": 0.4920620322227478,
      "learning_rate": 8.026978170913262e-06,
      "loss": 0.3858,
      "step": 20080
    },
    {
      "epoch": 0.9215272360148685,
      "grad_norm": 0.41910985112190247,
      "learning_rate": 8.02678301616278e-06,
      "loss": 0.3628,
      "step": 20081
    },
    {
      "epoch": 0.921573126520123,
      "grad_norm": 0.4516465961933136,
      "learning_rate": 8.026587854133841e-06,
      "loss": 0.4541,
      "step": 20082
    },
    {
      "epoch": 0.9216190170253774,
      "grad_norm": 0.45636186003685,
      "learning_rate": 8.026392684826913e-06,
      "loss": 0.4223,
      "step": 20083
    },
    {
      "epoch": 0.9216649075306319,
      "grad_norm": 0.4509398937225342,
      "learning_rate": 8.026197508242467e-06,
      "loss": 0.4182,
      "step": 20084
    },
    {
      "epoch": 0.9217107980358864,
      "grad_norm": 0.4412829875946045,
      "learning_rate": 8.026002324380972e-06,
      "loss": 0.4368,
      "step": 20085
    },
    {
      "epoch": 0.9217566885411408,
      "grad_norm": 0.44978055357933044,
      "learning_rate": 8.025807133242893e-06,
      "loss": 0.3996,
      "step": 20086
    },
    {
      "epoch": 0.9218025790463953,
      "grad_norm": 0.47315067052841187,
      "learning_rate": 8.025611934828705e-06,
      "loss": 0.4247,
      "step": 20087
    },
    {
      "epoch": 0.9218484695516498,
      "grad_norm": 0.4475835859775543,
      "learning_rate": 8.025416729138876e-06,
      "loss": 0.4324,
      "step": 20088
    },
    {
      "epoch": 0.9218943600569042,
      "grad_norm": 0.4375230669975281,
      "learning_rate": 8.025221516173873e-06,
      "loss": 0.3693,
      "step": 20089
    },
    {
      "epoch": 0.9219402505621587,
      "grad_norm": 0.4591505825519562,
      "learning_rate": 8.025026295934169e-06,
      "loss": 0.4283,
      "step": 20090
    },
    {
      "epoch": 0.9219861410674132,
      "grad_norm": 0.4266013503074646,
      "learning_rate": 8.02483106842023e-06,
      "loss": 0.343,
      "step": 20091
    },
    {
      "epoch": 0.9220320315726677,
      "grad_norm": 0.4743383526802063,
      "learning_rate": 8.024635833632525e-06,
      "loss": 0.3979,
      "step": 20092
    },
    {
      "epoch": 0.922077922077922,
      "grad_norm": 0.45974451303482056,
      "learning_rate": 8.024440591571529e-06,
      "loss": 0.4193,
      "step": 20093
    },
    {
      "epoch": 0.9221238125831765,
      "grad_norm": 0.475330114364624,
      "learning_rate": 8.024245342237706e-06,
      "loss": 0.429,
      "step": 20094
    },
    {
      "epoch": 0.922169703088431,
      "grad_norm": 0.48697683215141296,
      "learning_rate": 8.024050085631527e-06,
      "loss": 0.4344,
      "step": 20095
    },
    {
      "epoch": 0.9222155935936854,
      "grad_norm": 0.4553377330303192,
      "learning_rate": 8.023854821753464e-06,
      "loss": 0.4091,
      "step": 20096
    },
    {
      "epoch": 0.9222614840989399,
      "grad_norm": 0.4588093161582947,
      "learning_rate": 8.023659550603983e-06,
      "loss": 0.3754,
      "step": 20097
    },
    {
      "epoch": 0.9223073746041944,
      "grad_norm": 0.444650799036026,
      "learning_rate": 8.023464272183555e-06,
      "loss": 0.3891,
      "step": 20098
    },
    {
      "epoch": 0.9223532651094488,
      "grad_norm": 0.4421999156475067,
      "learning_rate": 8.02326898649265e-06,
      "loss": 0.3887,
      "step": 20099
    },
    {
      "epoch": 0.9223991556147033,
      "grad_norm": 0.4136192202568054,
      "learning_rate": 8.023073693531738e-06,
      "loss": 0.381,
      "step": 20100
    },
    {
      "epoch": 0.9224450461199578,
      "grad_norm": 0.4877961575984955,
      "learning_rate": 8.022878393301286e-06,
      "loss": 0.5001,
      "step": 20101
    },
    {
      "epoch": 0.9224909366252122,
      "grad_norm": 0.4695016145706177,
      "learning_rate": 8.022683085801768e-06,
      "loss": 0.4151,
      "step": 20102
    },
    {
      "epoch": 0.9225368271304667,
      "grad_norm": 0.42179787158966064,
      "learning_rate": 8.022487771033649e-06,
      "loss": 0.3761,
      "step": 20103
    },
    {
      "epoch": 0.9225827176357212,
      "grad_norm": 0.4642117917537689,
      "learning_rate": 8.022292448997402e-06,
      "loss": 0.4799,
      "step": 20104
    },
    {
      "epoch": 0.9226286081409756,
      "grad_norm": 0.437956303358078,
      "learning_rate": 8.022097119693495e-06,
      "loss": 0.3855,
      "step": 20105
    },
    {
      "epoch": 0.9226744986462301,
      "grad_norm": 0.46346476674079895,
      "learning_rate": 8.021901783122398e-06,
      "loss": 0.4583,
      "step": 20106
    },
    {
      "epoch": 0.9227203891514846,
      "grad_norm": 0.4763093590736389,
      "learning_rate": 8.021706439284583e-06,
      "loss": 0.5004,
      "step": 20107
    },
    {
      "epoch": 0.9227662796567391,
      "grad_norm": 0.4424879848957062,
      "learning_rate": 8.021511088180513e-06,
      "loss": 0.3886,
      "step": 20108
    },
    {
      "epoch": 0.9228121701619935,
      "grad_norm": 0.47741714119911194,
      "learning_rate": 8.021315729810667e-06,
      "loss": 0.425,
      "step": 20109
    },
    {
      "epoch": 0.922858060667248,
      "grad_norm": 0.4704485833644867,
      "learning_rate": 8.02112036417551e-06,
      "loss": 0.4276,
      "step": 20110
    },
    {
      "epoch": 0.9229039511725025,
      "grad_norm": 0.46119987964630127,
      "learning_rate": 8.02092499127551e-06,
      "loss": 0.3763,
      "step": 20111
    },
    {
      "epoch": 0.9229498416777568,
      "grad_norm": 0.429625004529953,
      "learning_rate": 8.02072961111114e-06,
      "loss": 0.3968,
      "step": 20112
    },
    {
      "epoch": 0.9229957321830113,
      "grad_norm": 0.453208863735199,
      "learning_rate": 8.020534223682867e-06,
      "loss": 0.4158,
      "step": 20113
    },
    {
      "epoch": 0.9230416226882658,
      "grad_norm": 0.44942331314086914,
      "learning_rate": 8.020338828991163e-06,
      "loss": 0.4078,
      "step": 20114
    },
    {
      "epoch": 0.9230875131935202,
      "grad_norm": 5.099334239959717,
      "learning_rate": 8.0201434270365e-06,
      "loss": 0.3706,
      "step": 20115
    },
    {
      "epoch": 0.9231334036987747,
      "grad_norm": 0.4599052667617798,
      "learning_rate": 8.019948017819343e-06,
      "loss": 0.3558,
      "step": 20116
    },
    {
      "epoch": 0.9231792942040292,
      "grad_norm": 0.4213622510433197,
      "learning_rate": 8.019752601340163e-06,
      "loss": 0.3486,
      "step": 20117
    },
    {
      "epoch": 0.9232251847092836,
      "grad_norm": 0.45478516817092896,
      "learning_rate": 8.019557177599434e-06,
      "loss": 0.4009,
      "step": 20118
    },
    {
      "epoch": 0.9232710752145381,
      "grad_norm": 0.46097317337989807,
      "learning_rate": 8.01936174659762e-06,
      "loss": 0.4205,
      "step": 20119
    },
    {
      "epoch": 0.9233169657197926,
      "grad_norm": 0.4721134603023529,
      "learning_rate": 8.019166308335195e-06,
      "loss": 0.4693,
      "step": 20120
    },
    {
      "epoch": 0.923362856225047,
      "grad_norm": 0.44674164056777954,
      "learning_rate": 8.01897086281263e-06,
      "loss": 0.3164,
      "step": 20121
    },
    {
      "epoch": 0.9234087467303015,
      "grad_norm": 0.4272751808166504,
      "learning_rate": 8.018775410030391e-06,
      "loss": 0.3643,
      "step": 20122
    },
    {
      "epoch": 0.923454637235556,
      "grad_norm": 0.4608440399169922,
      "learning_rate": 8.018579949988949e-06,
      "loss": 0.4397,
      "step": 20123
    },
    {
      "epoch": 0.9235005277408104,
      "grad_norm": 0.46902430057525635,
      "learning_rate": 8.018384482688776e-06,
      "loss": 0.3754,
      "step": 20124
    },
    {
      "epoch": 0.9235464182460649,
      "grad_norm": 0.4426458775997162,
      "learning_rate": 8.018189008130341e-06,
      "loss": 0.3738,
      "step": 20125
    },
    {
      "epoch": 0.9235923087513194,
      "grad_norm": 0.9838173985481262,
      "learning_rate": 8.017993526314114e-06,
      "loss": 0.4594,
      "step": 20126
    },
    {
      "epoch": 0.9236381992565739,
      "grad_norm": 0.4765073359012604,
      "learning_rate": 8.017798037240567e-06,
      "loss": 0.4034,
      "step": 20127
    },
    {
      "epoch": 0.9236840897618283,
      "grad_norm": 0.44089826941490173,
      "learning_rate": 8.017602540910166e-06,
      "loss": 0.3262,
      "step": 20128
    },
    {
      "epoch": 0.9237299802670828,
      "grad_norm": 0.4565177261829376,
      "learning_rate": 8.017407037323383e-06,
      "loss": 0.4016,
      "step": 20129
    },
    {
      "epoch": 0.9237758707723372,
      "grad_norm": 0.4744439423084259,
      "learning_rate": 8.017211526480688e-06,
      "loss": 0.4537,
      "step": 20130
    },
    {
      "epoch": 0.9238217612775916,
      "grad_norm": 0.48319289088249207,
      "learning_rate": 8.017016008382553e-06,
      "loss": 0.4483,
      "step": 20131
    },
    {
      "epoch": 0.9238676517828461,
      "grad_norm": 0.4439082145690918,
      "learning_rate": 8.016820483029446e-06,
      "loss": 0.4143,
      "step": 20132
    },
    {
      "epoch": 0.9239135422881006,
      "grad_norm": 0.43708109855651855,
      "learning_rate": 8.016624950421839e-06,
      "loss": 0.3739,
      "step": 20133
    },
    {
      "epoch": 0.923959432793355,
      "grad_norm": 0.4252750277519226,
      "learning_rate": 8.0164294105602e-06,
      "loss": 0.3337,
      "step": 20134
    },
    {
      "epoch": 0.9240053232986095,
      "grad_norm": 0.4561387002468109,
      "learning_rate": 8.016233863444998e-06,
      "loss": 0.388,
      "step": 20135
    },
    {
      "epoch": 0.924051213803864,
      "grad_norm": 0.4649335741996765,
      "learning_rate": 8.01603830907671e-06,
      "loss": 0.4489,
      "step": 20136
    },
    {
      "epoch": 0.9240971043091184,
      "grad_norm": 0.4832724928855896,
      "learning_rate": 8.015842747455798e-06,
      "loss": 0.4688,
      "step": 20137
    },
    {
      "epoch": 0.9241429948143729,
      "grad_norm": 0.405512273311615,
      "learning_rate": 8.01564717858274e-06,
      "loss": 0.3192,
      "step": 20138
    },
    {
      "epoch": 0.9241888853196274,
      "grad_norm": 0.44366398453712463,
      "learning_rate": 8.015451602457997e-06,
      "loss": 0.3836,
      "step": 20139
    },
    {
      "epoch": 0.9242347758248818,
      "grad_norm": 0.6866410374641418,
      "learning_rate": 8.015256019082049e-06,
      "loss": 0.4445,
      "step": 20140
    },
    {
      "epoch": 0.9242806663301363,
      "grad_norm": 0.4202505052089691,
      "learning_rate": 8.015060428455361e-06,
      "loss": 0.3179,
      "step": 20141
    },
    {
      "epoch": 0.9243265568353908,
      "grad_norm": 0.45955929160118103,
      "learning_rate": 8.014864830578404e-06,
      "loss": 0.4358,
      "step": 20142
    },
    {
      "epoch": 0.9243724473406453,
      "grad_norm": 0.47378605604171753,
      "learning_rate": 8.014669225451647e-06,
      "loss": 0.3847,
      "step": 20143
    },
    {
      "epoch": 0.9244183378458997,
      "grad_norm": 0.4619116187095642,
      "learning_rate": 8.014473613075565e-06,
      "loss": 0.4137,
      "step": 20144
    },
    {
      "epoch": 0.9244642283511542,
      "grad_norm": 0.4496588706970215,
      "learning_rate": 8.014277993450623e-06,
      "loss": 0.3871,
      "step": 20145
    },
    {
      "epoch": 0.9245101188564087,
      "grad_norm": 0.5027875900268555,
      "learning_rate": 8.014082366577293e-06,
      "loss": 0.4595,
      "step": 20146
    },
    {
      "epoch": 0.924556009361663,
      "grad_norm": 0.44417211413383484,
      "learning_rate": 8.013886732456049e-06,
      "loss": 0.3703,
      "step": 20147
    },
    {
      "epoch": 0.9246018998669175,
      "grad_norm": 0.45157408714294434,
      "learning_rate": 8.013691091087356e-06,
      "loss": 0.4061,
      "step": 20148
    },
    {
      "epoch": 0.924647790372172,
      "grad_norm": 0.44583213329315186,
      "learning_rate": 8.013495442471689e-06,
      "loss": 0.3976,
      "step": 20149
    },
    {
      "epoch": 0.9246936808774264,
      "grad_norm": 0.44881442189216614,
      "learning_rate": 8.013299786609517e-06,
      "loss": 0.4059,
      "step": 20150
    },
    {
      "epoch": 0.9247395713826809,
      "grad_norm": 0.4406236708164215,
      "learning_rate": 8.013104123501307e-06,
      "loss": 0.3763,
      "step": 20151
    },
    {
      "epoch": 0.9247854618879354,
      "grad_norm": 0.5094813108444214,
      "learning_rate": 8.012908453147534e-06,
      "loss": 0.5005,
      "step": 20152
    },
    {
      "epoch": 0.9248313523931898,
      "grad_norm": 0.42629730701446533,
      "learning_rate": 8.012712775548669e-06,
      "loss": 0.3677,
      "step": 20153
    },
    {
      "epoch": 0.9248772428984443,
      "grad_norm": 0.444350004196167,
      "learning_rate": 8.012517090705179e-06,
      "loss": 0.3918,
      "step": 20154
    },
    {
      "epoch": 0.9249231334036988,
      "grad_norm": 0.45564523339271545,
      "learning_rate": 8.012321398617536e-06,
      "loss": 0.3831,
      "step": 20155
    },
    {
      "epoch": 0.9249690239089532,
      "grad_norm": 0.5198468565940857,
      "learning_rate": 8.01212569928621e-06,
      "loss": 0.401,
      "step": 20156
    },
    {
      "epoch": 0.9250149144142077,
      "grad_norm": 0.4857330620288849,
      "learning_rate": 8.011929992711674e-06,
      "loss": 0.4181,
      "step": 20157
    },
    {
      "epoch": 0.9250608049194622,
      "grad_norm": 0.4150199890136719,
      "learning_rate": 8.011734278894396e-06,
      "loss": 0.3555,
      "step": 20158
    },
    {
      "epoch": 0.9251066954247166,
      "grad_norm": 0.5214964747428894,
      "learning_rate": 8.011538557834849e-06,
      "loss": 0.4432,
      "step": 20159
    },
    {
      "epoch": 0.9251525859299711,
      "grad_norm": 0.45616814494132996,
      "learning_rate": 8.0113428295335e-06,
      "loss": 0.3556,
      "step": 20160
    },
    {
      "epoch": 0.9251984764352256,
      "grad_norm": 0.45148035883903503,
      "learning_rate": 8.011147093990823e-06,
      "loss": 0.3517,
      "step": 20161
    },
    {
      "epoch": 0.9252443669404801,
      "grad_norm": 0.4500608742237091,
      "learning_rate": 8.010951351207288e-06,
      "loss": 0.4106,
      "step": 20162
    },
    {
      "epoch": 0.9252902574457345,
      "grad_norm": 0.4324566125869751,
      "learning_rate": 8.010755601183364e-06,
      "loss": 0.3378,
      "step": 20163
    },
    {
      "epoch": 0.925336147950989,
      "grad_norm": 0.4405294358730316,
      "learning_rate": 8.010559843919526e-06,
      "loss": 0.3646,
      "step": 20164
    },
    {
      "epoch": 0.9253820384562434,
      "grad_norm": 0.39832887053489685,
      "learning_rate": 8.010364079416239e-06,
      "loss": 0.2831,
      "step": 20165
    },
    {
      "epoch": 0.9254279289614978,
      "grad_norm": 0.5034353137016296,
      "learning_rate": 8.010168307673977e-06,
      "loss": 0.4928,
      "step": 20166
    },
    {
      "epoch": 0.9254738194667523,
      "grad_norm": 0.4509574770927429,
      "learning_rate": 8.00997252869321e-06,
      "loss": 0.4148,
      "step": 20167
    },
    {
      "epoch": 0.9255197099720068,
      "grad_norm": 0.4546881914138794,
      "learning_rate": 8.00977674247441e-06,
      "loss": 0.4365,
      "step": 20168
    },
    {
      "epoch": 0.9255656004772612,
      "grad_norm": 0.519939124584198,
      "learning_rate": 8.009580949018046e-06,
      "loss": 0.4972,
      "step": 20169
    },
    {
      "epoch": 0.9256114909825157,
      "grad_norm": 0.47314193844795227,
      "learning_rate": 8.00938514832459e-06,
      "loss": 0.4006,
      "step": 20170
    },
    {
      "epoch": 0.9256573814877702,
      "grad_norm": 0.45718103647232056,
      "learning_rate": 8.009189340394513e-06,
      "loss": 0.3915,
      "step": 20171
    },
    {
      "epoch": 0.9257032719930246,
      "grad_norm": 0.43417084217071533,
      "learning_rate": 8.008993525228283e-06,
      "loss": 0.3884,
      "step": 20172
    },
    {
      "epoch": 0.9257491624982791,
      "grad_norm": 0.43725332617759705,
      "learning_rate": 8.008797702826376e-06,
      "loss": 0.3967,
      "step": 20173
    },
    {
      "epoch": 0.9257950530035336,
      "grad_norm": 0.48367947340011597,
      "learning_rate": 8.00860187318926e-06,
      "loss": 0.4433,
      "step": 20174
    },
    {
      "epoch": 0.925840943508788,
      "grad_norm": 0.441753089427948,
      "learning_rate": 8.008406036317404e-06,
      "loss": 0.3638,
      "step": 20175
    },
    {
      "epoch": 0.9258868340140425,
      "grad_norm": 0.4716617465019226,
      "learning_rate": 8.008210192211283e-06,
      "loss": 0.4681,
      "step": 20176
    },
    {
      "epoch": 0.925932724519297,
      "grad_norm": 0.48167508840560913,
      "learning_rate": 8.008014340871364e-06,
      "loss": 0.4604,
      "step": 20177
    },
    {
      "epoch": 0.9259786150245514,
      "grad_norm": 0.452449232339859,
      "learning_rate": 8.007818482298121e-06,
      "loss": 0.3927,
      "step": 20178
    },
    {
      "epoch": 0.9260245055298059,
      "grad_norm": 0.46801790595054626,
      "learning_rate": 8.007622616492024e-06,
      "loss": 0.3827,
      "step": 20179
    },
    {
      "epoch": 0.9260703960350604,
      "grad_norm": 0.3981826603412628,
      "learning_rate": 8.007426743453544e-06,
      "loss": 0.3064,
      "step": 20180
    },
    {
      "epoch": 0.9261162865403149,
      "grad_norm": 0.47127172350883484,
      "learning_rate": 8.007230863183152e-06,
      "loss": 0.3514,
      "step": 20181
    },
    {
      "epoch": 0.9261621770455692,
      "grad_norm": 0.4735325276851654,
      "learning_rate": 8.007034975681319e-06,
      "loss": 0.4181,
      "step": 20182
    },
    {
      "epoch": 0.9262080675508237,
      "grad_norm": 0.47084641456604004,
      "learning_rate": 8.006839080948514e-06,
      "loss": 0.3889,
      "step": 20183
    },
    {
      "epoch": 0.9262539580560782,
      "grad_norm": 0.6653023362159729,
      "learning_rate": 8.006643178985212e-06,
      "loss": 0.3866,
      "step": 20184
    },
    {
      "epoch": 0.9262998485613326,
      "grad_norm": 0.451484352350235,
      "learning_rate": 8.006447269791881e-06,
      "loss": 0.3882,
      "step": 20185
    },
    {
      "epoch": 0.9263457390665871,
      "grad_norm": 0.45510998368263245,
      "learning_rate": 8.006251353368994e-06,
      "loss": 0.3988,
      "step": 20186
    },
    {
      "epoch": 0.9263916295718416,
      "grad_norm": 0.42842280864715576,
      "learning_rate": 8.00605542971702e-06,
      "loss": 0.3843,
      "step": 20187
    },
    {
      "epoch": 0.926437520077096,
      "grad_norm": 0.5064033269882202,
      "learning_rate": 8.005859498836432e-06,
      "loss": 0.4596,
      "step": 20188
    },
    {
      "epoch": 0.9264834105823505,
      "grad_norm": 0.456572026014328,
      "learning_rate": 8.005663560727702e-06,
      "loss": 0.4091,
      "step": 20189
    },
    {
      "epoch": 0.926529301087605,
      "grad_norm": 0.4626701772212982,
      "learning_rate": 8.005467615391298e-06,
      "loss": 0.4468,
      "step": 20190
    },
    {
      "epoch": 0.9265751915928594,
      "grad_norm": 0.49238666892051697,
      "learning_rate": 8.005271662827694e-06,
      "loss": 0.4446,
      "step": 20191
    },
    {
      "epoch": 0.9266210820981139,
      "grad_norm": 0.4442947208881378,
      "learning_rate": 8.005075703037359e-06,
      "loss": 0.4082,
      "step": 20192
    },
    {
      "epoch": 0.9266669726033684,
      "grad_norm": 0.4686109721660614,
      "learning_rate": 8.004879736020766e-06,
      "loss": 0.3892,
      "step": 20193
    },
    {
      "epoch": 0.9267128631086228,
      "grad_norm": 0.440669983625412,
      "learning_rate": 8.004683761778386e-06,
      "loss": 0.3981,
      "step": 20194
    },
    {
      "epoch": 0.9267587536138773,
      "grad_norm": 0.43945297598838806,
      "learning_rate": 8.00448778031069e-06,
      "loss": 0.3492,
      "step": 20195
    },
    {
      "epoch": 0.9268046441191318,
      "grad_norm": 0.4125298857688904,
      "learning_rate": 8.004291791618149e-06,
      "loss": 0.3223,
      "step": 20196
    },
    {
      "epoch": 0.9268505346243863,
      "grad_norm": 0.4818645119667053,
      "learning_rate": 8.004095795701233e-06,
      "loss": 0.3943,
      "step": 20197
    },
    {
      "epoch": 0.9268964251296407,
      "grad_norm": 0.4749382436275482,
      "learning_rate": 8.003899792560416e-06,
      "loss": 0.4584,
      "step": 20198
    },
    {
      "epoch": 0.9269423156348952,
      "grad_norm": 0.44277647137641907,
      "learning_rate": 8.003703782196169e-06,
      "loss": 0.3488,
      "step": 20199
    },
    {
      "epoch": 0.9269882061401497,
      "grad_norm": 0.4344618022441864,
      "learning_rate": 8.00350776460896e-06,
      "loss": 0.3582,
      "step": 20200
    },
    {
      "epoch": 0.927034096645404,
      "grad_norm": 0.4796612560749054,
      "learning_rate": 8.003311739799264e-06,
      "loss": 0.4864,
      "step": 20201
    },
    {
      "epoch": 0.9270799871506585,
      "grad_norm": 0.45990604162216187,
      "learning_rate": 8.003115707767553e-06,
      "loss": 0.3976,
      "step": 20202
    },
    {
      "epoch": 0.927125877655913,
      "grad_norm": 0.4325892925262451,
      "learning_rate": 8.002919668514294e-06,
      "loss": 0.3676,
      "step": 20203
    },
    {
      "epoch": 0.9271717681611674,
      "grad_norm": 0.4816102981567383,
      "learning_rate": 8.002723622039961e-06,
      "loss": 0.4824,
      "step": 20204
    },
    {
      "epoch": 0.9272176586664219,
      "grad_norm": 0.4213895797729492,
      "learning_rate": 8.002527568345027e-06,
      "loss": 0.3878,
      "step": 20205
    },
    {
      "epoch": 0.9272635491716764,
      "grad_norm": 0.4386197030544281,
      "learning_rate": 8.00233150742996e-06,
      "loss": 0.3727,
      "step": 20206
    },
    {
      "epoch": 0.9273094396769308,
      "grad_norm": 0.6181452870368958,
      "learning_rate": 8.002135439295234e-06,
      "loss": 0.4254,
      "step": 20207
    },
    {
      "epoch": 0.9273553301821853,
      "grad_norm": 0.4586374759674072,
      "learning_rate": 8.00193936394132e-06,
      "loss": 0.3798,
      "step": 20208
    },
    {
      "epoch": 0.9274012206874398,
      "grad_norm": 0.4967198371887207,
      "learning_rate": 8.00174328136869e-06,
      "loss": 0.5202,
      "step": 20209
    },
    {
      "epoch": 0.9274471111926942,
      "grad_norm": 0.45364734530448914,
      "learning_rate": 8.001547191577814e-06,
      "loss": 0.3352,
      "step": 20210
    },
    {
      "epoch": 0.9274930016979487,
      "grad_norm": 0.48298773169517517,
      "learning_rate": 8.001351094569164e-06,
      "loss": 0.4848,
      "step": 20211
    },
    {
      "epoch": 0.9275388922032032,
      "grad_norm": 0.43446314334869385,
      "learning_rate": 8.001154990343212e-06,
      "loss": 0.3639,
      "step": 20212
    },
    {
      "epoch": 0.9275847827084576,
      "grad_norm": 0.44874873757362366,
      "learning_rate": 8.00095887890043e-06,
      "loss": 0.4134,
      "step": 20213
    },
    {
      "epoch": 0.9276306732137121,
      "grad_norm": 0.481990784406662,
      "learning_rate": 8.000762760241287e-06,
      "loss": 0.4814,
      "step": 20214
    },
    {
      "epoch": 0.9276765637189666,
      "grad_norm": 0.48851269483566284,
      "learning_rate": 8.000566634366259e-06,
      "loss": 0.4551,
      "step": 20215
    },
    {
      "epoch": 0.9277224542242211,
      "grad_norm": 0.46863317489624023,
      "learning_rate": 8.000370501275814e-06,
      "loss": 0.3918,
      "step": 20216
    },
    {
      "epoch": 0.9277683447294754,
      "grad_norm": 0.4697883725166321,
      "learning_rate": 8.000174360970425e-06,
      "loss": 0.3421,
      "step": 20217
    },
    {
      "epoch": 0.92781423523473,
      "grad_norm": 0.4470983147621155,
      "learning_rate": 7.999978213450566e-06,
      "loss": 0.4035,
      "step": 20218
    },
    {
      "epoch": 0.9278601257399844,
      "grad_norm": 0.4085080921649933,
      "learning_rate": 7.999782058716703e-06,
      "loss": 0.3045,
      "step": 20219
    },
    {
      "epoch": 0.9279060162452388,
      "grad_norm": 0.4527370035648346,
      "learning_rate": 7.999585896769314e-06,
      "loss": 0.3876,
      "step": 20220
    },
    {
      "epoch": 0.9279519067504933,
      "grad_norm": 0.45354482531547546,
      "learning_rate": 7.999389727608864e-06,
      "loss": 0.4025,
      "step": 20221
    },
    {
      "epoch": 0.9279977972557478,
      "grad_norm": 0.4277035892009735,
      "learning_rate": 7.99919355123583e-06,
      "loss": 0.3524,
      "step": 20222
    },
    {
      "epoch": 0.9280436877610022,
      "grad_norm": 0.4565936326980591,
      "learning_rate": 7.998997367650683e-06,
      "loss": 0.4241,
      "step": 20223
    },
    {
      "epoch": 0.9280895782662567,
      "grad_norm": 0.45384135842323303,
      "learning_rate": 7.998801176853893e-06,
      "loss": 0.412,
      "step": 20224
    },
    {
      "epoch": 0.9281354687715112,
      "grad_norm": 0.424320787191391,
      "learning_rate": 7.998604978845932e-06,
      "loss": 0.3433,
      "step": 20225
    },
    {
      "epoch": 0.9281813592767656,
      "grad_norm": 0.43937960267066956,
      "learning_rate": 7.998408773627273e-06,
      "loss": 0.3589,
      "step": 20226
    },
    {
      "epoch": 0.9282272497820201,
      "grad_norm": 0.4242733120918274,
      "learning_rate": 7.998212561198386e-06,
      "loss": 0.3556,
      "step": 20227
    },
    {
      "epoch": 0.9282731402872746,
      "grad_norm": 0.4264681339263916,
      "learning_rate": 7.998016341559745e-06,
      "loss": 0.3627,
      "step": 20228
    },
    {
      "epoch": 0.928319030792529,
      "grad_norm": 0.44101452827453613,
      "learning_rate": 7.997820114711822e-06,
      "loss": 0.3785,
      "step": 20229
    },
    {
      "epoch": 0.9283649212977835,
      "grad_norm": 0.46124914288520813,
      "learning_rate": 7.997623880655087e-06,
      "loss": 0.4366,
      "step": 20230
    },
    {
      "epoch": 0.928410811803038,
      "grad_norm": 0.42966219782829285,
      "learning_rate": 7.997427639390011e-06,
      "loss": 0.3627,
      "step": 20231
    },
    {
      "epoch": 0.9284567023082925,
      "grad_norm": 0.43765339255332947,
      "learning_rate": 7.99723139091707e-06,
      "loss": 0.3926,
      "step": 20232
    },
    {
      "epoch": 0.9285025928135469,
      "grad_norm": 0.42160043120384216,
      "learning_rate": 7.997035135236732e-06,
      "loss": 0.2973,
      "step": 20233
    },
    {
      "epoch": 0.9285484833188014,
      "grad_norm": 0.42998433113098145,
      "learning_rate": 7.996838872349471e-06,
      "loss": 0.3295,
      "step": 20234
    },
    {
      "epoch": 0.9285943738240559,
      "grad_norm": 0.4393029510974884,
      "learning_rate": 7.996642602255759e-06,
      "loss": 0.3439,
      "step": 20235
    },
    {
      "epoch": 0.9286402643293102,
      "grad_norm": 0.47461357712745667,
      "learning_rate": 7.996446324956067e-06,
      "loss": 0.3829,
      "step": 20236
    },
    {
      "epoch": 0.9286861548345647,
      "grad_norm": 0.47021517157554626,
      "learning_rate": 7.996250040450866e-06,
      "loss": 0.4406,
      "step": 20237
    },
    {
      "epoch": 0.9287320453398192,
      "grad_norm": 0.4942142367362976,
      "learning_rate": 7.996053748740631e-06,
      "loss": 0.4161,
      "step": 20238
    },
    {
      "epoch": 0.9287779358450736,
      "grad_norm": 0.446237176656723,
      "learning_rate": 7.995857449825832e-06,
      "loss": 0.3393,
      "step": 20239
    },
    {
      "epoch": 0.9288238263503281,
      "grad_norm": 0.421702116727829,
      "learning_rate": 7.995661143706941e-06,
      "loss": 0.3168,
      "step": 20240
    },
    {
      "epoch": 0.9288697168555826,
      "grad_norm": 0.4544486999511719,
      "learning_rate": 7.99546483038443e-06,
      "loss": 0.4166,
      "step": 20241
    },
    {
      "epoch": 0.928915607360837,
      "grad_norm": 0.5076224207878113,
      "learning_rate": 7.99526850985877e-06,
      "loss": 0.4473,
      "step": 20242
    },
    {
      "epoch": 0.9289614978660915,
      "grad_norm": 0.39172518253326416,
      "learning_rate": 7.99507218213044e-06,
      "loss": 0.3092,
      "step": 20243
    },
    {
      "epoch": 0.929007388371346,
      "grad_norm": 0.44923603534698486,
      "learning_rate": 7.994875847199901e-06,
      "loss": 0.3855,
      "step": 20244
    },
    {
      "epoch": 0.9290532788766004,
      "grad_norm": 0.42723509669303894,
      "learning_rate": 7.994679505067633e-06,
      "loss": 0.364,
      "step": 20245
    },
    {
      "epoch": 0.9290991693818549,
      "grad_norm": 0.4429986774921417,
      "learning_rate": 7.994483155734105e-06,
      "loss": 0.4011,
      "step": 20246
    },
    {
      "epoch": 0.9291450598871094,
      "grad_norm": 0.46733662486076355,
      "learning_rate": 7.994286799199791e-06,
      "loss": 0.3823,
      "step": 20247
    },
    {
      "epoch": 0.9291909503923638,
      "grad_norm": 0.46118026971817017,
      "learning_rate": 7.994090435465162e-06,
      "loss": 0.3961,
      "step": 20248
    },
    {
      "epoch": 0.9292368408976183,
      "grad_norm": 0.44163617491722107,
      "learning_rate": 7.99389406453069e-06,
      "loss": 0.3762,
      "step": 20249
    },
    {
      "epoch": 0.9292827314028728,
      "grad_norm": 0.51311856508255,
      "learning_rate": 7.993697686396848e-06,
      "loss": 0.5183,
      "step": 20250
    },
    {
      "epoch": 0.9293286219081273,
      "grad_norm": 0.4374142289161682,
      "learning_rate": 7.993501301064109e-06,
      "loss": 0.3621,
      "step": 20251
    },
    {
      "epoch": 0.9293745124133816,
      "grad_norm": 0.42916128039360046,
      "learning_rate": 7.993304908532944e-06,
      "loss": 0.3733,
      "step": 20252
    },
    {
      "epoch": 0.9294204029186361,
      "grad_norm": 0.48459580540657043,
      "learning_rate": 7.993108508803825e-06,
      "loss": 0.4256,
      "step": 20253
    },
    {
      "epoch": 0.9294662934238906,
      "grad_norm": 0.42442217469215393,
      "learning_rate": 7.992912101877224e-06,
      "loss": 0.3325,
      "step": 20254
    },
    {
      "epoch": 0.929512183929145,
      "grad_norm": 0.45944973826408386,
      "learning_rate": 7.992715687753614e-06,
      "loss": 0.4378,
      "step": 20255
    },
    {
      "epoch": 0.9295580744343995,
      "grad_norm": 0.4844738841056824,
      "learning_rate": 7.992519266433467e-06,
      "loss": 0.4322,
      "step": 20256
    },
    {
      "epoch": 0.929603964939654,
      "grad_norm": 0.4237200617790222,
      "learning_rate": 7.992322837917256e-06,
      "loss": 0.3583,
      "step": 20257
    },
    {
      "epoch": 0.9296498554449084,
      "grad_norm": 0.47028473019599915,
      "learning_rate": 7.992126402205454e-06,
      "loss": 0.3961,
      "step": 20258
    },
    {
      "epoch": 0.9296957459501629,
      "grad_norm": 0.440351665019989,
      "learning_rate": 7.991929959298533e-06,
      "loss": 0.3921,
      "step": 20259
    },
    {
      "epoch": 0.9297416364554174,
      "grad_norm": 0.4935568571090698,
      "learning_rate": 7.991733509196962e-06,
      "loss": 0.4865,
      "step": 20260
    },
    {
      "epoch": 0.9297875269606718,
      "grad_norm": 0.49787211418151855,
      "learning_rate": 7.991537051901216e-06,
      "loss": 0.4678,
      "step": 20261
    },
    {
      "epoch": 0.9298334174659263,
      "grad_norm": 0.47834399342536926,
      "learning_rate": 7.99134058741177e-06,
      "loss": 0.4534,
      "step": 20262
    },
    {
      "epoch": 0.9298793079711808,
      "grad_norm": 0.46421730518341064,
      "learning_rate": 7.991144115729094e-06,
      "loss": 0.374,
      "step": 20263
    },
    {
      "epoch": 0.9299251984764352,
      "grad_norm": 0.4456886947154999,
      "learning_rate": 7.990947636853658e-06,
      "loss": 0.3795,
      "step": 20264
    },
    {
      "epoch": 0.9299710889816897,
      "grad_norm": 0.4817242920398712,
      "learning_rate": 7.990751150785938e-06,
      "loss": 0.4159,
      "step": 20265
    },
    {
      "epoch": 0.9300169794869442,
      "grad_norm": 0.474458783864975,
      "learning_rate": 7.990554657526407e-06,
      "loss": 0.4355,
      "step": 20266
    },
    {
      "epoch": 0.9300628699921986,
      "grad_norm": 0.4144432842731476,
      "learning_rate": 7.990358157075535e-06,
      "loss": 0.3368,
      "step": 20267
    },
    {
      "epoch": 0.9301087604974531,
      "grad_norm": 0.47062501311302185,
      "learning_rate": 7.990161649433795e-06,
      "loss": 0.4567,
      "step": 20268
    },
    {
      "epoch": 0.9301546510027076,
      "grad_norm": 0.6802839040756226,
      "learning_rate": 7.989965134601659e-06,
      "loss": 0.4008,
      "step": 20269
    },
    {
      "epoch": 0.930200541507962,
      "grad_norm": 0.44600000977516174,
      "learning_rate": 7.989768612579603e-06,
      "loss": 0.3317,
      "step": 20270
    },
    {
      "epoch": 0.9302464320132164,
      "grad_norm": 0.5261929631233215,
      "learning_rate": 7.989572083368095e-06,
      "loss": 0.4635,
      "step": 20271
    },
    {
      "epoch": 0.9302923225184709,
      "grad_norm": 0.4651545584201813,
      "learning_rate": 7.98937554696761e-06,
      "loss": 0.4539,
      "step": 20272
    },
    {
      "epoch": 0.9303382130237254,
      "grad_norm": 0.48982152342796326,
      "learning_rate": 7.98917900337862e-06,
      "loss": 0.4352,
      "step": 20273
    },
    {
      "epoch": 0.9303841035289798,
      "grad_norm": 0.4288163185119629,
      "learning_rate": 7.9889824526016e-06,
      "loss": 0.3159,
      "step": 20274
    },
    {
      "epoch": 0.9304299940342343,
      "grad_norm": 0.49071741104125977,
      "learning_rate": 7.988785894637018e-06,
      "loss": 0.4307,
      "step": 20275
    },
    {
      "epoch": 0.9304758845394888,
      "grad_norm": 0.4224623739719391,
      "learning_rate": 7.98858932948535e-06,
      "loss": 0.3389,
      "step": 20276
    },
    {
      "epoch": 0.9305217750447432,
      "grad_norm": 0.4941238462924957,
      "learning_rate": 7.988392757147067e-06,
      "loss": 0.5115,
      "step": 20277
    },
    {
      "epoch": 0.9305676655499977,
      "grad_norm": 0.5270593762397766,
      "learning_rate": 7.988196177622645e-06,
      "loss": 0.4703,
      "step": 20278
    },
    {
      "epoch": 0.9306135560552522,
      "grad_norm": 0.4579271376132965,
      "learning_rate": 7.987999590912552e-06,
      "loss": 0.3839,
      "step": 20279
    },
    {
      "epoch": 0.9306594465605066,
      "grad_norm": 0.45771870017051697,
      "learning_rate": 7.987802997017265e-06,
      "loss": 0.4263,
      "step": 20280
    },
    {
      "epoch": 0.9307053370657611,
      "grad_norm": 0.46832096576690674,
      "learning_rate": 7.987606395937253e-06,
      "loss": 0.4243,
      "step": 20281
    },
    {
      "epoch": 0.9307512275710156,
      "grad_norm": 0.4530988335609436,
      "learning_rate": 7.987409787672992e-06,
      "loss": 0.4262,
      "step": 20282
    },
    {
      "epoch": 0.93079711807627,
      "grad_norm": 0.47719812393188477,
      "learning_rate": 7.987213172224952e-06,
      "loss": 0.4251,
      "step": 20283
    },
    {
      "epoch": 0.9308430085815245,
      "grad_norm": 0.4545637369155884,
      "learning_rate": 7.987016549593608e-06,
      "loss": 0.3854,
      "step": 20284
    },
    {
      "epoch": 0.930888899086779,
      "grad_norm": 0.4217701852321625,
      "learning_rate": 7.986819919779432e-06,
      "loss": 0.3313,
      "step": 20285
    },
    {
      "epoch": 0.9309347895920335,
      "grad_norm": 0.4428486227989197,
      "learning_rate": 7.986623282782897e-06,
      "loss": 0.4072,
      "step": 20286
    },
    {
      "epoch": 0.9309806800972878,
      "grad_norm": 0.5097830891609192,
      "learning_rate": 7.986426638604475e-06,
      "loss": 0.3917,
      "step": 20287
    },
    {
      "epoch": 0.9310265706025423,
      "grad_norm": 0.4565395414829254,
      "learning_rate": 7.98622998724464e-06,
      "loss": 0.3843,
      "step": 20288
    },
    {
      "epoch": 0.9310724611077968,
      "grad_norm": 0.4487934112548828,
      "learning_rate": 7.986033328703864e-06,
      "loss": 0.3677,
      "step": 20289
    },
    {
      "epoch": 0.9311183516130512,
      "grad_norm": 0.47644639015197754,
      "learning_rate": 7.98583666298262e-06,
      "loss": 0.3911,
      "step": 20290
    },
    {
      "epoch": 0.9311642421183057,
      "grad_norm": 0.4260871112346649,
      "learning_rate": 7.98563999008138e-06,
      "loss": 0.3829,
      "step": 20291
    },
    {
      "epoch": 0.9312101326235602,
      "grad_norm": 0.41332828998565674,
      "learning_rate": 7.98544331000062e-06,
      "loss": 0.3215,
      "step": 20292
    },
    {
      "epoch": 0.9312560231288146,
      "grad_norm": 0.42774856090545654,
      "learning_rate": 7.98524662274081e-06,
      "loss": 0.3751,
      "step": 20293
    },
    {
      "epoch": 0.9313019136340691,
      "grad_norm": 0.4162341356277466,
      "learning_rate": 7.985049928302426e-06,
      "loss": 0.3333,
      "step": 20294
    },
    {
      "epoch": 0.9313478041393236,
      "grad_norm": 0.43373873829841614,
      "learning_rate": 7.984853226685936e-06,
      "loss": 0.345,
      "step": 20295
    },
    {
      "epoch": 0.931393694644578,
      "grad_norm": 0.4557092785835266,
      "learning_rate": 7.984656517891819e-06,
      "loss": 0.3962,
      "step": 20296
    },
    {
      "epoch": 0.9314395851498325,
      "grad_norm": 0.45525923371315,
      "learning_rate": 7.984459801920542e-06,
      "loss": 0.3868,
      "step": 20297
    },
    {
      "epoch": 0.931485475655087,
      "grad_norm": 0.44930750131607056,
      "learning_rate": 7.984263078772583e-06,
      "loss": 0.3625,
      "step": 20298
    },
    {
      "epoch": 0.9315313661603414,
      "grad_norm": 0.4670160114765167,
      "learning_rate": 7.984066348448413e-06,
      "loss": 0.4444,
      "step": 20299
    },
    {
      "epoch": 0.9315772566655959,
      "grad_norm": 0.47160497307777405,
      "learning_rate": 7.983869610948505e-06,
      "loss": 0.3931,
      "step": 20300
    },
    {
      "epoch": 0.9316231471708504,
      "grad_norm": 0.4703989028930664,
      "learning_rate": 7.983672866273332e-06,
      "loss": 0.4068,
      "step": 20301
    },
    {
      "epoch": 0.9316690376761048,
      "grad_norm": 0.4948856830596924,
      "learning_rate": 7.98347611442337e-06,
      "loss": 0.3906,
      "step": 20302
    },
    {
      "epoch": 0.9317149281813593,
      "grad_norm": 0.4895839989185333,
      "learning_rate": 7.983279355399088e-06,
      "loss": 0.4743,
      "step": 20303
    },
    {
      "epoch": 0.9317608186866138,
      "grad_norm": 0.47057029604911804,
      "learning_rate": 7.983082589200959e-06,
      "loss": 0.3756,
      "step": 20304
    },
    {
      "epoch": 0.9318067091918683,
      "grad_norm": 0.4436322748661041,
      "learning_rate": 7.98288581582946e-06,
      "loss": 0.3742,
      "step": 20305
    },
    {
      "epoch": 0.9318525996971226,
      "grad_norm": 0.4512331485748291,
      "learning_rate": 7.98268903528506e-06,
      "loss": 0.3841,
      "step": 20306
    },
    {
      "epoch": 0.9318984902023771,
      "grad_norm": 0.4703245162963867,
      "learning_rate": 7.982492247568233e-06,
      "loss": 0.3756,
      "step": 20307
    },
    {
      "epoch": 0.9319443807076316,
      "grad_norm": 0.46462470293045044,
      "learning_rate": 7.982295452679458e-06,
      "loss": 0.4346,
      "step": 20308
    },
    {
      "epoch": 0.931990271212886,
      "grad_norm": 0.45305514335632324,
      "learning_rate": 7.9820986506192e-06,
      "loss": 0.3817,
      "step": 20309
    },
    {
      "epoch": 0.9320361617181405,
      "grad_norm": 0.4430617094039917,
      "learning_rate": 7.981901841387939e-06,
      "loss": 0.3965,
      "step": 20310
    },
    {
      "epoch": 0.932082052223395,
      "grad_norm": 0.43959179520606995,
      "learning_rate": 7.981705024986144e-06,
      "loss": 0.3835,
      "step": 20311
    },
    {
      "epoch": 0.9321279427286494,
      "grad_norm": 0.43895307183265686,
      "learning_rate": 7.981508201414287e-06,
      "loss": 0.353,
      "step": 20312
    },
    {
      "epoch": 0.9321738332339039,
      "grad_norm": 0.4520552158355713,
      "learning_rate": 7.981311370672847e-06,
      "loss": 0.3773,
      "step": 20313
    },
    {
      "epoch": 0.9322197237391584,
      "grad_norm": 0.7275726795196533,
      "learning_rate": 7.981114532762293e-06,
      "loss": 0.3479,
      "step": 20314
    },
    {
      "epoch": 0.9322656142444128,
      "grad_norm": 0.43414735794067383,
      "learning_rate": 7.980917687683099e-06,
      "loss": 0.3737,
      "step": 20315
    },
    {
      "epoch": 0.9323115047496673,
      "grad_norm": 0.4539187252521515,
      "learning_rate": 7.980720835435739e-06,
      "loss": 0.4287,
      "step": 20316
    },
    {
      "epoch": 0.9323573952549218,
      "grad_norm": 0.467336505651474,
      "learning_rate": 7.980523976020687e-06,
      "loss": 0.419,
      "step": 20317
    },
    {
      "epoch": 0.9324032857601762,
      "grad_norm": 0.46483537554740906,
      "learning_rate": 7.980327109438415e-06,
      "loss": 0.4167,
      "step": 20318
    },
    {
      "epoch": 0.9324491762654307,
      "grad_norm": 0.4978630244731903,
      "learning_rate": 7.980130235689396e-06,
      "loss": 0.4392,
      "step": 20319
    },
    {
      "epoch": 0.9324950667706852,
      "grad_norm": 0.4810475707054138,
      "learning_rate": 7.979933354774105e-06,
      "loss": 0.4129,
      "step": 20320
    },
    {
      "epoch": 0.9325409572759396,
      "grad_norm": 0.4841994345188141,
      "learning_rate": 7.979736466693015e-06,
      "loss": 0.4371,
      "step": 20321
    },
    {
      "epoch": 0.932586847781194,
      "grad_norm": 0.5075539946556091,
      "learning_rate": 7.979539571446599e-06,
      "loss": 0.4438,
      "step": 20322
    },
    {
      "epoch": 0.9326327382864485,
      "grad_norm": 0.4406054615974426,
      "learning_rate": 7.979342669035331e-06,
      "loss": 0.4068,
      "step": 20323
    },
    {
      "epoch": 0.932678628791703,
      "grad_norm": 0.4545266926288605,
      "learning_rate": 7.979145759459684e-06,
      "loss": 0.4097,
      "step": 20324
    },
    {
      "epoch": 0.9327245192969574,
      "grad_norm": 0.43244898319244385,
      "learning_rate": 7.97894884272013e-06,
      "loss": 0.3541,
      "step": 20325
    },
    {
      "epoch": 0.9327704098022119,
      "grad_norm": 0.44165661931037903,
      "learning_rate": 7.978751918817146e-06,
      "loss": 0.386,
      "step": 20326
    },
    {
      "epoch": 0.9328163003074664,
      "grad_norm": 0.4612163305282593,
      "learning_rate": 7.978554987751204e-06,
      "loss": 0.418,
      "step": 20327
    },
    {
      "epoch": 0.9328621908127208,
      "grad_norm": 0.4388722777366638,
      "learning_rate": 7.978358049522777e-06,
      "loss": 0.3154,
      "step": 20328
    },
    {
      "epoch": 0.9329080813179753,
      "grad_norm": 0.43616822361946106,
      "learning_rate": 7.978161104132338e-06,
      "loss": 0.3794,
      "step": 20329
    },
    {
      "epoch": 0.9329539718232298,
      "grad_norm": 0.4512796998023987,
      "learning_rate": 7.977964151580361e-06,
      "loss": 0.3774,
      "step": 20330
    },
    {
      "epoch": 0.9329998623284842,
      "grad_norm": 0.4513742923736572,
      "learning_rate": 7.977767191867323e-06,
      "loss": 0.4075,
      "step": 20331
    },
    {
      "epoch": 0.9330457528337387,
      "grad_norm": 0.4227915108203888,
      "learning_rate": 7.977570224993693e-06,
      "loss": 0.4004,
      "step": 20332
    },
    {
      "epoch": 0.9330916433389932,
      "grad_norm": 0.4252769649028778,
      "learning_rate": 7.977373250959947e-06,
      "loss": 0.3662,
      "step": 20333
    },
    {
      "epoch": 0.9331375338442476,
      "grad_norm": 0.44232627749443054,
      "learning_rate": 7.977176269766556e-06,
      "loss": 0.3664,
      "step": 20334
    },
    {
      "epoch": 0.9331834243495021,
      "grad_norm": 0.4657588303089142,
      "learning_rate": 7.976979281413997e-06,
      "loss": 0.4052,
      "step": 20335
    },
    {
      "epoch": 0.9332293148547566,
      "grad_norm": 0.4564688801765442,
      "learning_rate": 7.976782285902743e-06,
      "loss": 0.4025,
      "step": 20336
    },
    {
      "epoch": 0.933275205360011,
      "grad_norm": 0.41343116760253906,
      "learning_rate": 7.976585283233267e-06,
      "loss": 0.3549,
      "step": 20337
    },
    {
      "epoch": 0.9333210958652655,
      "grad_norm": 0.43240121006965637,
      "learning_rate": 7.976388273406041e-06,
      "loss": 0.362,
      "step": 20338
    },
    {
      "epoch": 0.93336698637052,
      "grad_norm": 0.4117271900177002,
      "learning_rate": 7.976191256421544e-06,
      "loss": 0.3331,
      "step": 20339
    },
    {
      "epoch": 0.9334128768757745,
      "grad_norm": 0.431355744600296,
      "learning_rate": 7.975994232280242e-06,
      "loss": 0.331,
      "step": 20340
    },
    {
      "epoch": 0.9334587673810288,
      "grad_norm": 0.4309585392475128,
      "learning_rate": 7.975797200982617e-06,
      "loss": 0.3847,
      "step": 20341
    },
    {
      "epoch": 0.9335046578862833,
      "grad_norm": 0.4521945118904114,
      "learning_rate": 7.975600162529138e-06,
      "loss": 0.3793,
      "step": 20342
    },
    {
      "epoch": 0.9335505483915378,
      "grad_norm": 0.42503949999809265,
      "learning_rate": 7.975403116920279e-06,
      "loss": 0.35,
      "step": 20343
    },
    {
      "epoch": 0.9335964388967922,
      "grad_norm": 0.4731177091598511,
      "learning_rate": 7.975206064156512e-06,
      "loss": 0.4491,
      "step": 20344
    },
    {
      "epoch": 0.9336423294020467,
      "grad_norm": 0.5135696530342102,
      "learning_rate": 7.97500900423832e-06,
      "loss": 0.3289,
      "step": 20345
    },
    {
      "epoch": 0.9336882199073012,
      "grad_norm": 0.42334139347076416,
      "learning_rate": 7.974811937166164e-06,
      "loss": 0.3227,
      "step": 20346
    },
    {
      "epoch": 0.9337341104125556,
      "grad_norm": 0.40354225039482117,
      "learning_rate": 7.974614862940526e-06,
      "loss": 0.2903,
      "step": 20347
    },
    {
      "epoch": 0.9337800009178101,
      "grad_norm": 0.4627375602722168,
      "learning_rate": 7.974417781561879e-06,
      "loss": 0.4055,
      "step": 20348
    },
    {
      "epoch": 0.9338258914230646,
      "grad_norm": 0.44367659091949463,
      "learning_rate": 7.974220693030696e-06,
      "loss": 0.3758,
      "step": 20349
    },
    {
      "epoch": 0.933871781928319,
      "grad_norm": 0.4567160904407501,
      "learning_rate": 7.974023597347451e-06,
      "loss": 0.3798,
      "step": 20350
    },
    {
      "epoch": 0.9339176724335735,
      "grad_norm": 0.451743483543396,
      "learning_rate": 7.973826494512618e-06,
      "loss": 0.4048,
      "step": 20351
    },
    {
      "epoch": 0.933963562938828,
      "grad_norm": 0.48589593172073364,
      "learning_rate": 7.973629384526668e-06,
      "loss": 0.38,
      "step": 20352
    },
    {
      "epoch": 0.9340094534440824,
      "grad_norm": 0.46221789717674255,
      "learning_rate": 7.973432267390082e-06,
      "loss": 0.4306,
      "step": 20353
    },
    {
      "epoch": 0.9340553439493369,
      "grad_norm": 0.5218206644058228,
      "learning_rate": 7.973235143103327e-06,
      "loss": 0.5291,
      "step": 20354
    },
    {
      "epoch": 0.9341012344545914,
      "grad_norm": 0.44337570667266846,
      "learning_rate": 7.97303801166688e-06,
      "loss": 0.4109,
      "step": 20355
    },
    {
      "epoch": 0.9341471249598458,
      "grad_norm": 0.41345709562301636,
      "learning_rate": 7.972840873081217e-06,
      "loss": 0.3393,
      "step": 20356
    },
    {
      "epoch": 0.9341930154651003,
      "grad_norm": 0.4655740559101105,
      "learning_rate": 7.972643727346809e-06,
      "loss": 0.4,
      "step": 20357
    },
    {
      "epoch": 0.9342389059703547,
      "grad_norm": 0.5042961835861206,
      "learning_rate": 7.97244657446413e-06,
      "loss": 0.4546,
      "step": 20358
    },
    {
      "epoch": 0.9342847964756092,
      "grad_norm": 0.43266043066978455,
      "learning_rate": 7.972249414433657e-06,
      "loss": 0.3475,
      "step": 20359
    },
    {
      "epoch": 0.9343306869808636,
      "grad_norm": 0.4629168212413788,
      "learning_rate": 7.97205224725586e-06,
      "loss": 0.3716,
      "step": 20360
    },
    {
      "epoch": 0.9343765774861181,
      "grad_norm": 0.4277596175670624,
      "learning_rate": 7.971855072931217e-06,
      "loss": 0.376,
      "step": 20361
    },
    {
      "epoch": 0.9344224679913726,
      "grad_norm": 0.4335121512413025,
      "learning_rate": 7.971657891460199e-06,
      "loss": 0.3687,
      "step": 20362
    },
    {
      "epoch": 0.934468358496627,
      "grad_norm": 0.4694441556930542,
      "learning_rate": 7.971460702843283e-06,
      "loss": 0.4486,
      "step": 20363
    },
    {
      "epoch": 0.9345142490018815,
      "grad_norm": 0.4622114300727844,
      "learning_rate": 7.97126350708094e-06,
      "loss": 0.4866,
      "step": 20364
    },
    {
      "epoch": 0.934560139507136,
      "grad_norm": 0.4361158609390259,
      "learning_rate": 7.971066304173649e-06,
      "loss": 0.3551,
      "step": 20365
    },
    {
      "epoch": 0.9346060300123904,
      "grad_norm": 0.47331929206848145,
      "learning_rate": 7.970869094121878e-06,
      "loss": 0.4246,
      "step": 20366
    },
    {
      "epoch": 0.9346519205176449,
      "grad_norm": 0.4339955151081085,
      "learning_rate": 7.970671876926108e-06,
      "loss": 0.3374,
      "step": 20367
    },
    {
      "epoch": 0.9346978110228994,
      "grad_norm": 0.4230981171131134,
      "learning_rate": 7.970474652586807e-06,
      "loss": 0.3337,
      "step": 20368
    },
    {
      "epoch": 0.9347437015281538,
      "grad_norm": 0.48214393854141235,
      "learning_rate": 7.970277421104452e-06,
      "loss": 0.4184,
      "step": 20369
    },
    {
      "epoch": 0.9347895920334083,
      "grad_norm": 0.4599495232105255,
      "learning_rate": 7.970080182479519e-06,
      "loss": 0.3888,
      "step": 20370
    },
    {
      "epoch": 0.9348354825386628,
      "grad_norm": 0.45671120285987854,
      "learning_rate": 7.969882936712478e-06,
      "loss": 0.3895,
      "step": 20371
    },
    {
      "epoch": 0.9348813730439172,
      "grad_norm": 0.4504025876522064,
      "learning_rate": 7.969685683803808e-06,
      "loss": 0.4116,
      "step": 20372
    },
    {
      "epoch": 0.9349272635491717,
      "grad_norm": 0.4363076388835907,
      "learning_rate": 7.969488423753981e-06,
      "loss": 0.3894,
      "step": 20373
    },
    {
      "epoch": 0.9349731540544262,
      "grad_norm": 0.43687349557876587,
      "learning_rate": 7.96929115656347e-06,
      "loss": 0.4082,
      "step": 20374
    },
    {
      "epoch": 0.9350190445596807,
      "grad_norm": 0.43681254982948303,
      "learning_rate": 7.969093882232752e-06,
      "loss": 0.3989,
      "step": 20375
    },
    {
      "epoch": 0.935064935064935,
      "grad_norm": 0.4616544842720032,
      "learning_rate": 7.968896600762302e-06,
      "loss": 0.4028,
      "step": 20376
    },
    {
      "epoch": 0.9351108255701895,
      "grad_norm": 0.43930792808532715,
      "learning_rate": 7.968699312152591e-06,
      "loss": 0.3782,
      "step": 20377
    },
    {
      "epoch": 0.935156716075444,
      "grad_norm": 0.44546934962272644,
      "learning_rate": 7.968502016404095e-06,
      "loss": 0.3732,
      "step": 20378
    },
    {
      "epoch": 0.9352026065806984,
      "grad_norm": 0.44198182225227356,
      "learning_rate": 7.968304713517287e-06,
      "loss": 0.3757,
      "step": 20379
    },
    {
      "epoch": 0.9352484970859529,
      "grad_norm": 0.4190891683101654,
      "learning_rate": 7.968107403492645e-06,
      "loss": 0.3177,
      "step": 20380
    },
    {
      "epoch": 0.9352943875912074,
      "grad_norm": 0.46070799231529236,
      "learning_rate": 7.967910086330641e-06,
      "loss": 0.3991,
      "step": 20381
    },
    {
      "epoch": 0.9353402780964618,
      "grad_norm": 0.42522281408309937,
      "learning_rate": 7.96771276203175e-06,
      "loss": 0.3237,
      "step": 20382
    },
    {
      "epoch": 0.9353861686017163,
      "grad_norm": 0.45431822538375854,
      "learning_rate": 7.967515430596447e-06,
      "loss": 0.3971,
      "step": 20383
    },
    {
      "epoch": 0.9354320591069708,
      "grad_norm": 0.44196534156799316,
      "learning_rate": 7.967318092025205e-06,
      "loss": 0.4057,
      "step": 20384
    },
    {
      "epoch": 0.9354779496122252,
      "grad_norm": 0.42814087867736816,
      "learning_rate": 7.967120746318499e-06,
      "loss": 0.3866,
      "step": 20385
    },
    {
      "epoch": 0.9355238401174797,
      "grad_norm": 0.4517696797847748,
      "learning_rate": 7.966923393476807e-06,
      "loss": 0.3823,
      "step": 20386
    },
    {
      "epoch": 0.9355697306227342,
      "grad_norm": 0.4592064917087555,
      "learning_rate": 7.966726033500598e-06,
      "loss": 0.3588,
      "step": 20387
    },
    {
      "epoch": 0.9356156211279886,
      "grad_norm": 0.4304770231246948,
      "learning_rate": 7.96652866639035e-06,
      "loss": 0.4023,
      "step": 20388
    },
    {
      "epoch": 0.9356615116332431,
      "grad_norm": 0.4599981904029846,
      "learning_rate": 7.966331292146536e-06,
      "loss": 0.4582,
      "step": 20389
    },
    {
      "epoch": 0.9357074021384976,
      "grad_norm": 0.42678970098495483,
      "learning_rate": 7.966133910769634e-06,
      "loss": 0.3494,
      "step": 20390
    },
    {
      "epoch": 0.935753292643752,
      "grad_norm": 0.4330138564109802,
      "learning_rate": 7.965936522260113e-06,
      "loss": 0.3165,
      "step": 20391
    },
    {
      "epoch": 0.9357991831490065,
      "grad_norm": 0.42991411685943604,
      "learning_rate": 7.965739126618451e-06,
      "loss": 0.3481,
      "step": 20392
    },
    {
      "epoch": 0.935845073654261,
      "grad_norm": 0.4437566101551056,
      "learning_rate": 7.965541723845125e-06,
      "loss": 0.4191,
      "step": 20393
    },
    {
      "epoch": 0.9358909641595154,
      "grad_norm": 0.49238455295562744,
      "learning_rate": 7.965344313940604e-06,
      "loss": 0.4834,
      "step": 20394
    },
    {
      "epoch": 0.9359368546647698,
      "grad_norm": 0.445453941822052,
      "learning_rate": 7.965146896905366e-06,
      "loss": 0.369,
      "step": 20395
    },
    {
      "epoch": 0.9359827451700243,
      "grad_norm": 0.5172765851020813,
      "learning_rate": 7.964949472739889e-06,
      "loss": 0.5219,
      "step": 20396
    },
    {
      "epoch": 0.9360286356752788,
      "grad_norm": 0.423674613237381,
      "learning_rate": 7.964752041444642e-06,
      "loss": 0.3265,
      "step": 20397
    },
    {
      "epoch": 0.9360745261805332,
      "grad_norm": 0.5061588287353516,
      "learning_rate": 7.9645546030201e-06,
      "loss": 0.4757,
      "step": 20398
    },
    {
      "epoch": 0.9361204166857877,
      "grad_norm": 0.47754406929016113,
      "learning_rate": 7.964357157466744e-06,
      "loss": 0.4947,
      "step": 20399
    },
    {
      "epoch": 0.9361663071910422,
      "grad_norm": 0.4733749330043793,
      "learning_rate": 7.964159704785041e-06,
      "loss": 0.4653,
      "step": 20400
    },
    {
      "epoch": 0.9362121976962966,
      "grad_norm": 0.456330806016922,
      "learning_rate": 7.963962244975472e-06,
      "loss": 0.3753,
      "step": 20401
    },
    {
      "epoch": 0.9362580882015511,
      "grad_norm": 0.4928845465183258,
      "learning_rate": 7.963764778038507e-06,
      "loss": 0.4028,
      "step": 20402
    },
    {
      "epoch": 0.9363039787068056,
      "grad_norm": 0.45592936873435974,
      "learning_rate": 7.963567303974625e-06,
      "loss": 0.3488,
      "step": 20403
    },
    {
      "epoch": 0.93634986921206,
      "grad_norm": 0.4464655816555023,
      "learning_rate": 7.963369822784299e-06,
      "loss": 0.3111,
      "step": 20404
    },
    {
      "epoch": 0.9363957597173145,
      "grad_norm": 0.4612474739551544,
      "learning_rate": 7.963172334468004e-06,
      "loss": 0.3762,
      "step": 20405
    },
    {
      "epoch": 0.936441650222569,
      "grad_norm": 0.4287279546260834,
      "learning_rate": 7.962974839026211e-06,
      "loss": 0.3699,
      "step": 20406
    },
    {
      "epoch": 0.9364875407278234,
      "grad_norm": 0.49074408411979675,
      "learning_rate": 7.962777336459405e-06,
      "loss": 0.4481,
      "step": 20407
    },
    {
      "epoch": 0.9365334312330779,
      "grad_norm": 0.4766649603843689,
      "learning_rate": 7.962579826768048e-06,
      "loss": 0.4713,
      "step": 20408
    },
    {
      "epoch": 0.9365793217383324,
      "grad_norm": 0.46244728565216064,
      "learning_rate": 7.962382309952626e-06,
      "loss": 0.4029,
      "step": 20409
    },
    {
      "epoch": 0.9366252122435867,
      "grad_norm": 0.43920326232910156,
      "learning_rate": 7.962184786013608e-06,
      "loss": 0.3638,
      "step": 20410
    },
    {
      "epoch": 0.9366711027488412,
      "grad_norm": 0.45780420303344727,
      "learning_rate": 7.961987254951471e-06,
      "loss": 0.3586,
      "step": 20411
    },
    {
      "epoch": 0.9367169932540957,
      "grad_norm": 0.45953822135925293,
      "learning_rate": 7.961789716766689e-06,
      "loss": 0.3774,
      "step": 20412
    },
    {
      "epoch": 0.9367628837593502,
      "grad_norm": 0.4686456322669983,
      "learning_rate": 7.961592171459738e-06,
      "loss": 0.4109,
      "step": 20413
    },
    {
      "epoch": 0.9368087742646046,
      "grad_norm": 0.45339253544807434,
      "learning_rate": 7.96139461903109e-06,
      "loss": 0.4045,
      "step": 20414
    },
    {
      "epoch": 0.9368546647698591,
      "grad_norm": 0.4976917803287506,
      "learning_rate": 7.961197059481228e-06,
      "loss": 0.5062,
      "step": 20415
    },
    {
      "epoch": 0.9369005552751136,
      "grad_norm": 0.4211575388908386,
      "learning_rate": 7.960999492810617e-06,
      "loss": 0.3706,
      "step": 20416
    },
    {
      "epoch": 0.936946445780368,
      "grad_norm": 0.48366421461105347,
      "learning_rate": 7.960801919019738e-06,
      "loss": 0.4443,
      "step": 20417
    },
    {
      "epoch": 0.9369923362856225,
      "grad_norm": 0.43069833517074585,
      "learning_rate": 7.960604338109066e-06,
      "loss": 0.3553,
      "step": 20418
    },
    {
      "epoch": 0.937038226790877,
      "grad_norm": 0.6587809920310974,
      "learning_rate": 7.960406750079073e-06,
      "loss": 0.4857,
      "step": 20419
    },
    {
      "epoch": 0.9370841172961314,
      "grad_norm": 0.44759494066238403,
      "learning_rate": 7.960209154930236e-06,
      "loss": 0.4094,
      "step": 20420
    },
    {
      "epoch": 0.9371300078013859,
      "grad_norm": 0.44655126333236694,
      "learning_rate": 7.960011552663032e-06,
      "loss": 0.3584,
      "step": 20421
    },
    {
      "epoch": 0.9371758983066404,
      "grad_norm": 0.4556111693382263,
      "learning_rate": 7.959813943277933e-06,
      "loss": 0.4029,
      "step": 20422
    },
    {
      "epoch": 0.9372217888118948,
      "grad_norm": 0.4245489239692688,
      "learning_rate": 7.959616326775417e-06,
      "loss": 0.3837,
      "step": 20423
    },
    {
      "epoch": 0.9372676793171493,
      "grad_norm": 0.44670405983924866,
      "learning_rate": 7.959418703155956e-06,
      "loss": 0.4045,
      "step": 20424
    },
    {
      "epoch": 0.9373135698224038,
      "grad_norm": 0.4947430491447449,
      "learning_rate": 7.959221072420029e-06,
      "loss": 0.4641,
      "step": 20425
    },
    {
      "epoch": 0.9373594603276582,
      "grad_norm": 0.45807018876075745,
      "learning_rate": 7.959023434568108e-06,
      "loss": 0.4043,
      "step": 20426
    },
    {
      "epoch": 0.9374053508329127,
      "grad_norm": 0.4390721619129181,
      "learning_rate": 7.95882578960067e-06,
      "loss": 0.3856,
      "step": 20427
    },
    {
      "epoch": 0.9374512413381672,
      "grad_norm": 0.44408053159713745,
      "learning_rate": 7.958628137518191e-06,
      "loss": 0.3847,
      "step": 20428
    },
    {
      "epoch": 0.9374971318434216,
      "grad_norm": 0.4389475882053375,
      "learning_rate": 7.958430478321144e-06,
      "loss": 0.3983,
      "step": 20429
    },
    {
      "epoch": 0.937543022348676,
      "grad_norm": 0.45934879779815674,
      "learning_rate": 7.958232812010005e-06,
      "loss": 0.424,
      "step": 20430
    },
    {
      "epoch": 0.9375889128539305,
      "grad_norm": 0.4600032567977905,
      "learning_rate": 7.958035138585249e-06,
      "loss": 0.4207,
      "step": 20431
    },
    {
      "epoch": 0.937634803359185,
      "grad_norm": 0.4573203921318054,
      "learning_rate": 7.957837458047354e-06,
      "loss": 0.3927,
      "step": 20432
    },
    {
      "epoch": 0.9376806938644394,
      "grad_norm": 0.4376120865345001,
      "learning_rate": 7.957639770396792e-06,
      "loss": 0.37,
      "step": 20433
    },
    {
      "epoch": 0.9377265843696939,
      "grad_norm": 0.4864038825035095,
      "learning_rate": 7.957442075634041e-06,
      "loss": 0.5003,
      "step": 20434
    },
    {
      "epoch": 0.9377724748749484,
      "grad_norm": 0.4329606592655182,
      "learning_rate": 7.957244373759574e-06,
      "loss": 0.3318,
      "step": 20435
    },
    {
      "epoch": 0.9378183653802028,
      "grad_norm": 0.4388364851474762,
      "learning_rate": 7.957046664773869e-06,
      "loss": 0.3154,
      "step": 20436
    },
    {
      "epoch": 0.9378642558854573,
      "grad_norm": 0.4855524003505707,
      "learning_rate": 7.956848948677398e-06,
      "loss": 0.4364,
      "step": 20437
    },
    {
      "epoch": 0.9379101463907118,
      "grad_norm": 0.47706377506256104,
      "learning_rate": 7.95665122547064e-06,
      "loss": 0.4316,
      "step": 20438
    },
    {
      "epoch": 0.9379560368959662,
      "grad_norm": 0.4241121709346771,
      "learning_rate": 7.956453495154069e-06,
      "loss": 0.3421,
      "step": 20439
    },
    {
      "epoch": 0.9380019274012207,
      "grad_norm": 0.48005160689353943,
      "learning_rate": 7.95625575772816e-06,
      "loss": 0.4906,
      "step": 20440
    },
    {
      "epoch": 0.9380478179064752,
      "grad_norm": 0.4619748592376709,
      "learning_rate": 7.956058013193389e-06,
      "loss": 0.3642,
      "step": 20441
    },
    {
      "epoch": 0.9380937084117296,
      "grad_norm": 0.47387444972991943,
      "learning_rate": 7.955860261550232e-06,
      "loss": 0.3681,
      "step": 20442
    },
    {
      "epoch": 0.9381395989169841,
      "grad_norm": 0.4567619562149048,
      "learning_rate": 7.955662502799163e-06,
      "loss": 0.3738,
      "step": 20443
    },
    {
      "epoch": 0.9381854894222386,
      "grad_norm": 0.4508633017539978,
      "learning_rate": 7.955464736940658e-06,
      "loss": 0.4289,
      "step": 20444
    },
    {
      "epoch": 0.938231379927493,
      "grad_norm": 0.5032143592834473,
      "learning_rate": 7.955266963975194e-06,
      "loss": 0.4889,
      "step": 20445
    },
    {
      "epoch": 0.9382772704327474,
      "grad_norm": 0.4434879422187805,
      "learning_rate": 7.955069183903246e-06,
      "loss": 0.3607,
      "step": 20446
    },
    {
      "epoch": 0.9383231609380019,
      "grad_norm": 0.4410766065120697,
      "learning_rate": 7.954871396725289e-06,
      "loss": 0.3437,
      "step": 20447
    },
    {
      "epoch": 0.9383690514432564,
      "grad_norm": 0.45540422201156616,
      "learning_rate": 7.954673602441799e-06,
      "loss": 0.4294,
      "step": 20448
    },
    {
      "epoch": 0.9384149419485108,
      "grad_norm": 0.44661667943000793,
      "learning_rate": 7.954475801053251e-06,
      "loss": 0.3385,
      "step": 20449
    },
    {
      "epoch": 0.9384608324537653,
      "grad_norm": 0.4406035542488098,
      "learning_rate": 7.954277992560123e-06,
      "loss": 0.3618,
      "step": 20450
    },
    {
      "epoch": 0.9385067229590198,
      "grad_norm": 0.4627743363380432,
      "learning_rate": 7.954080176962885e-06,
      "loss": 0.4299,
      "step": 20451
    },
    {
      "epoch": 0.9385526134642742,
      "grad_norm": 0.4619785249233246,
      "learning_rate": 7.953882354262019e-06,
      "loss": 0.4026,
      "step": 20452
    },
    {
      "epoch": 0.9385985039695287,
      "grad_norm": 0.4931475818157196,
      "learning_rate": 7.953684524457999e-06,
      "loss": 0.4647,
      "step": 20453
    },
    {
      "epoch": 0.9386443944747832,
      "grad_norm": 0.48674526810646057,
      "learning_rate": 7.953486687551298e-06,
      "loss": 0.4789,
      "step": 20454
    },
    {
      "epoch": 0.9386902849800376,
      "grad_norm": 0.4216320216655731,
      "learning_rate": 7.953288843542395e-06,
      "loss": 0.3546,
      "step": 20455
    },
    {
      "epoch": 0.9387361754852921,
      "grad_norm": 0.46260204911231995,
      "learning_rate": 7.953090992431765e-06,
      "loss": 0.3816,
      "step": 20456
    },
    {
      "epoch": 0.9387820659905466,
      "grad_norm": 0.4285966157913208,
      "learning_rate": 7.952893134219881e-06,
      "loss": 0.3851,
      "step": 20457
    },
    {
      "epoch": 0.938827956495801,
      "grad_norm": 0.4379470646381378,
      "learning_rate": 7.952695268907222e-06,
      "loss": 0.3848,
      "step": 20458
    },
    {
      "epoch": 0.9388738470010555,
      "grad_norm": 0.45765313506126404,
      "learning_rate": 7.952497396494263e-06,
      "loss": 0.3935,
      "step": 20459
    },
    {
      "epoch": 0.93891973750631,
      "grad_norm": 0.40586400032043457,
      "learning_rate": 7.95229951698148e-06,
      "loss": 0.3466,
      "step": 20460
    },
    {
      "epoch": 0.9389656280115644,
      "grad_norm": 0.44342708587646484,
      "learning_rate": 7.952101630369347e-06,
      "loss": 0.381,
      "step": 20461
    },
    {
      "epoch": 0.9390115185168189,
      "grad_norm": 0.4633656442165375,
      "learning_rate": 7.951903736658341e-06,
      "loss": 0.4039,
      "step": 20462
    },
    {
      "epoch": 0.9390574090220734,
      "grad_norm": 0.4547712504863739,
      "learning_rate": 7.95170583584894e-06,
      "loss": 0.371,
      "step": 20463
    },
    {
      "epoch": 0.9391032995273279,
      "grad_norm": 0.44663509726524353,
      "learning_rate": 7.951507927941618e-06,
      "loss": 0.3727,
      "step": 20464
    },
    {
      "epoch": 0.9391491900325822,
      "grad_norm": 0.44239839911460876,
      "learning_rate": 7.95131001293685e-06,
      "loss": 0.3964,
      "step": 20465
    },
    {
      "epoch": 0.9391950805378367,
      "grad_norm": 0.46243494749069214,
      "learning_rate": 7.951112090835114e-06,
      "loss": 0.4166,
      "step": 20466
    },
    {
      "epoch": 0.9392409710430912,
      "grad_norm": 0.4498955309391022,
      "learning_rate": 7.950914161636885e-06,
      "loss": 0.3986,
      "step": 20467
    },
    {
      "epoch": 0.9392868615483456,
      "grad_norm": 0.4769071638584137,
      "learning_rate": 7.950716225342637e-06,
      "loss": 0.4745,
      "step": 20468
    },
    {
      "epoch": 0.9393327520536001,
      "grad_norm": 0.5031524300575256,
      "learning_rate": 7.950518281952849e-06,
      "loss": 0.3978,
      "step": 20469
    },
    {
      "epoch": 0.9393786425588546,
      "grad_norm": 0.4308260977268219,
      "learning_rate": 7.950320331467996e-06,
      "loss": 0.3452,
      "step": 20470
    },
    {
      "epoch": 0.939424533064109,
      "grad_norm": 0.4867488145828247,
      "learning_rate": 7.950122373888553e-06,
      "loss": 0.4589,
      "step": 20471
    },
    {
      "epoch": 0.9394704235693635,
      "grad_norm": 0.45031046867370605,
      "learning_rate": 7.949924409214997e-06,
      "loss": 0.3789,
      "step": 20472
    },
    {
      "epoch": 0.939516314074618,
      "grad_norm": 0.4227551221847534,
      "learning_rate": 7.949726437447803e-06,
      "loss": 0.3841,
      "step": 20473
    },
    {
      "epoch": 0.9395622045798724,
      "grad_norm": 0.4574127793312073,
      "learning_rate": 7.949528458587448e-06,
      "loss": 0.4045,
      "step": 20474
    },
    {
      "epoch": 0.9396080950851269,
      "grad_norm": 0.4406052529811859,
      "learning_rate": 7.949330472634409e-06,
      "loss": 0.4271,
      "step": 20475
    },
    {
      "epoch": 0.9396539855903814,
      "grad_norm": 0.44044560194015503,
      "learning_rate": 7.949132479589161e-06,
      "loss": 0.3389,
      "step": 20476
    },
    {
      "epoch": 0.9396998760956358,
      "grad_norm": 0.4874519109725952,
      "learning_rate": 7.948934479452179e-06,
      "loss": 0.5168,
      "step": 20477
    },
    {
      "epoch": 0.9397457666008903,
      "grad_norm": 0.46071672439575195,
      "learning_rate": 7.948736472223942e-06,
      "loss": 0.332,
      "step": 20478
    },
    {
      "epoch": 0.9397916571061448,
      "grad_norm": 0.45353972911834717,
      "learning_rate": 7.948538457904923e-06,
      "loss": 0.371,
      "step": 20479
    },
    {
      "epoch": 0.9398375476113991,
      "grad_norm": 0.4417482912540436,
      "learning_rate": 7.9483404364956e-06,
      "loss": 0.3356,
      "step": 20480
    },
    {
      "epoch": 0.9398834381166536,
      "grad_norm": 0.4866461157798767,
      "learning_rate": 7.948142407996449e-06,
      "loss": 0.4622,
      "step": 20481
    },
    {
      "epoch": 0.9399293286219081,
      "grad_norm": 0.4350813925266266,
      "learning_rate": 7.947944372407945e-06,
      "loss": 0.3773,
      "step": 20482
    },
    {
      "epoch": 0.9399752191271626,
      "grad_norm": 0.4414445757865906,
      "learning_rate": 7.947746329730568e-06,
      "loss": 0.4034,
      "step": 20483
    },
    {
      "epoch": 0.940021109632417,
      "grad_norm": 0.44155555963516235,
      "learning_rate": 7.947548279964787e-06,
      "loss": 0.3983,
      "step": 20484
    },
    {
      "epoch": 0.9400670001376715,
      "grad_norm": 0.5072855353355408,
      "learning_rate": 7.947350223111085e-06,
      "loss": 0.5023,
      "step": 20485
    },
    {
      "epoch": 0.940112890642926,
      "grad_norm": 0.4436251223087311,
      "learning_rate": 7.947152159169938e-06,
      "loss": 0.3721,
      "step": 20486
    },
    {
      "epoch": 0.9401587811481804,
      "grad_norm": 0.43811696767807007,
      "learning_rate": 7.946954088141818e-06,
      "loss": 0.408,
      "step": 20487
    },
    {
      "epoch": 0.9402046716534349,
      "grad_norm": 0.4547506272792816,
      "learning_rate": 7.946756010027203e-06,
      "loss": 0.3765,
      "step": 20488
    },
    {
      "epoch": 0.9402505621586894,
      "grad_norm": 0.4463140666484833,
      "learning_rate": 7.946557924826571e-06,
      "loss": 0.3928,
      "step": 20489
    },
    {
      "epoch": 0.9402964526639438,
      "grad_norm": 0.5091535449028015,
      "learning_rate": 7.946359832540395e-06,
      "loss": 0.5123,
      "step": 20490
    },
    {
      "epoch": 0.9403423431691983,
      "grad_norm": 0.41807276010513306,
      "learning_rate": 7.946161733169156e-06,
      "loss": 0.3297,
      "step": 20491
    },
    {
      "epoch": 0.9403882336744528,
      "grad_norm": 0.4202379286289215,
      "learning_rate": 7.945963626713329e-06,
      "loss": 0.3396,
      "step": 20492
    },
    {
      "epoch": 0.9404341241797072,
      "grad_norm": 0.4405921399593353,
      "learning_rate": 7.945765513173385e-06,
      "loss": 0.408,
      "step": 20493
    },
    {
      "epoch": 0.9404800146849617,
      "grad_norm": 0.4647994041442871,
      "learning_rate": 7.945567392549807e-06,
      "loss": 0.4377,
      "step": 20494
    },
    {
      "epoch": 0.9405259051902162,
      "grad_norm": 0.4153446555137634,
      "learning_rate": 7.945369264843071e-06,
      "loss": 0.3302,
      "step": 20495
    },
    {
      "epoch": 0.9405717956954706,
      "grad_norm": 0.4453582465648651,
      "learning_rate": 7.945171130053648e-06,
      "loss": 0.3683,
      "step": 20496
    },
    {
      "epoch": 0.9406176862007251,
      "grad_norm": 0.44170495867729187,
      "learning_rate": 7.94497298818202e-06,
      "loss": 0.3588,
      "step": 20497
    },
    {
      "epoch": 0.9406635767059796,
      "grad_norm": 0.4771113991737366,
      "learning_rate": 7.944774839228661e-06,
      "loss": 0.4313,
      "step": 20498
    },
    {
      "epoch": 0.9407094672112339,
      "grad_norm": 0.432087242603302,
      "learning_rate": 7.944576683194047e-06,
      "loss": 0.3837,
      "step": 20499
    },
    {
      "epoch": 0.9407553577164884,
      "grad_norm": 0.44468769431114197,
      "learning_rate": 7.944378520078658e-06,
      "loss": 0.3901,
      "step": 20500
    },
    {
      "epoch": 0.9408012482217429,
      "grad_norm": 0.43146848678588867,
      "learning_rate": 7.944180349882966e-06,
      "loss": 0.4161,
      "step": 20501
    },
    {
      "epoch": 0.9408471387269974,
      "grad_norm": 0.46250441670417786,
      "learning_rate": 7.943982172607449e-06,
      "loss": 0.4459,
      "step": 20502
    },
    {
      "epoch": 0.9408930292322518,
      "grad_norm": 0.46426820755004883,
      "learning_rate": 7.943783988252584e-06,
      "loss": 0.3854,
      "step": 20503
    },
    {
      "epoch": 0.9409389197375063,
      "grad_norm": 0.45717164874076843,
      "learning_rate": 7.943585796818849e-06,
      "loss": 0.3806,
      "step": 20504
    },
    {
      "epoch": 0.9409848102427608,
      "grad_norm": 0.4335591197013855,
      "learning_rate": 7.943387598306718e-06,
      "loss": 0.3504,
      "step": 20505
    },
    {
      "epoch": 0.9410307007480152,
      "grad_norm": 0.4693076014518738,
      "learning_rate": 7.943189392716669e-06,
      "loss": 0.4584,
      "step": 20506
    },
    {
      "epoch": 0.9410765912532697,
      "grad_norm": 0.4661948084831238,
      "learning_rate": 7.94299118004918e-06,
      "loss": 0.4282,
      "step": 20507
    },
    {
      "epoch": 0.9411224817585242,
      "grad_norm": 0.45022183656692505,
      "learning_rate": 7.942792960304723e-06,
      "loss": 0.4266,
      "step": 20508
    },
    {
      "epoch": 0.9411683722637786,
      "grad_norm": 0.47297295928001404,
      "learning_rate": 7.942594733483779e-06,
      "loss": 0.3875,
      "step": 20509
    },
    {
      "epoch": 0.9412142627690331,
      "grad_norm": 0.446294367313385,
      "learning_rate": 7.942396499586825e-06,
      "loss": 0.3784,
      "step": 20510
    },
    {
      "epoch": 0.9412601532742876,
      "grad_norm": 0.41808173060417175,
      "learning_rate": 7.942198258614334e-06,
      "loss": 0.3508,
      "step": 20511
    },
    {
      "epoch": 0.941306043779542,
      "grad_norm": 0.4683513939380646,
      "learning_rate": 7.942000010566786e-06,
      "loss": 0.4148,
      "step": 20512
    },
    {
      "epoch": 0.9413519342847965,
      "grad_norm": 0.44048818945884705,
      "learning_rate": 7.941801755444655e-06,
      "loss": 0.3725,
      "step": 20513
    },
    {
      "epoch": 0.941397824790051,
      "grad_norm": 0.43999844789505005,
      "learning_rate": 7.941603493248422e-06,
      "loss": 0.3567,
      "step": 20514
    },
    {
      "epoch": 0.9414437152953054,
      "grad_norm": 0.4405522048473358,
      "learning_rate": 7.941405223978558e-06,
      "loss": 0.3886,
      "step": 20515
    },
    {
      "epoch": 0.9414896058005598,
      "grad_norm": 0.45946046710014343,
      "learning_rate": 7.941206947635544e-06,
      "loss": 0.4154,
      "step": 20516
    },
    {
      "epoch": 0.9415354963058143,
      "grad_norm": 0.4293038249015808,
      "learning_rate": 7.941008664219857e-06,
      "loss": 0.3398,
      "step": 20517
    },
    {
      "epoch": 0.9415813868110688,
      "grad_norm": 0.4158743619918823,
      "learning_rate": 7.94081037373197e-06,
      "loss": 0.3254,
      "step": 20518
    },
    {
      "epoch": 0.9416272773163232,
      "grad_norm": 0.48147132992744446,
      "learning_rate": 7.940612076172363e-06,
      "loss": 0.4805,
      "step": 20519
    },
    {
      "epoch": 0.9416731678215777,
      "grad_norm": 0.4803291857242584,
      "learning_rate": 7.940413771541514e-06,
      "loss": 0.3906,
      "step": 20520
    },
    {
      "epoch": 0.9417190583268322,
      "grad_norm": 0.48220449686050415,
      "learning_rate": 7.940215459839894e-06,
      "loss": 0.4324,
      "step": 20521
    },
    {
      "epoch": 0.9417649488320866,
      "grad_norm": 0.45679429173469543,
      "learning_rate": 7.940017141067986e-06,
      "loss": 0.4041,
      "step": 20522
    },
    {
      "epoch": 0.9418108393373411,
      "grad_norm": 0.4601984918117523,
      "learning_rate": 7.939818815226266e-06,
      "loss": 0.4016,
      "step": 20523
    },
    {
      "epoch": 0.9418567298425956,
      "grad_norm": 0.47521165013313293,
      "learning_rate": 7.939620482315206e-06,
      "loss": 0.4541,
      "step": 20524
    },
    {
      "epoch": 0.94190262034785,
      "grad_norm": 0.49009448289871216,
      "learning_rate": 7.939422142335288e-06,
      "loss": 0.4407,
      "step": 20525
    },
    {
      "epoch": 0.9419485108531045,
      "grad_norm": 0.46726834774017334,
      "learning_rate": 7.93922379528699e-06,
      "loss": 0.4268,
      "step": 20526
    },
    {
      "epoch": 0.941994401358359,
      "grad_norm": 0.4550146460533142,
      "learning_rate": 7.939025441170785e-06,
      "loss": 0.4198,
      "step": 20527
    },
    {
      "epoch": 0.9420402918636134,
      "grad_norm": 0.45934078097343445,
      "learning_rate": 7.93882707998715e-06,
      "loss": 0.4295,
      "step": 20528
    },
    {
      "epoch": 0.9420861823688679,
      "grad_norm": 0.45375436544418335,
      "learning_rate": 7.938628711736565e-06,
      "loss": 0.4052,
      "step": 20529
    },
    {
      "epoch": 0.9421320728741224,
      "grad_norm": 0.47653117775917053,
      "learning_rate": 7.938430336419505e-06,
      "loss": 0.3923,
      "step": 20530
    },
    {
      "epoch": 0.9421779633793768,
      "grad_norm": 0.4723985195159912,
      "learning_rate": 7.938231954036447e-06,
      "loss": 0.4428,
      "step": 20531
    },
    {
      "epoch": 0.9422238538846313,
      "grad_norm": 0.4493003785610199,
      "learning_rate": 7.938033564587867e-06,
      "loss": 0.4453,
      "step": 20532
    },
    {
      "epoch": 0.9422697443898858,
      "grad_norm": 0.4434211850166321,
      "learning_rate": 7.937835168074247e-06,
      "loss": 0.3989,
      "step": 20533
    },
    {
      "epoch": 0.9423156348951401,
      "grad_norm": 0.47145169973373413,
      "learning_rate": 7.93763676449606e-06,
      "loss": 0.4008,
      "step": 20534
    },
    {
      "epoch": 0.9423615254003946,
      "grad_norm": 0.44714632630348206,
      "learning_rate": 7.93743835385378e-06,
      "loss": 0.3402,
      "step": 20535
    },
    {
      "epoch": 0.9424074159056491,
      "grad_norm": 0.4586819112300873,
      "learning_rate": 7.937239936147892e-06,
      "loss": 0.3638,
      "step": 20536
    },
    {
      "epoch": 0.9424533064109036,
      "grad_norm": 0.44157424569129944,
      "learning_rate": 7.937041511378867e-06,
      "loss": 0.3735,
      "step": 20537
    },
    {
      "epoch": 0.942499196916158,
      "grad_norm": 0.45087864995002747,
      "learning_rate": 7.936843079547186e-06,
      "loss": 0.3601,
      "step": 20538
    },
    {
      "epoch": 0.9425450874214125,
      "grad_norm": 0.4548627436161041,
      "learning_rate": 7.936644640653322e-06,
      "loss": 0.4057,
      "step": 20539
    },
    {
      "epoch": 0.942590977926667,
      "grad_norm": 0.4633611738681793,
      "learning_rate": 7.936446194697755e-06,
      "loss": 0.3818,
      "step": 20540
    },
    {
      "epoch": 0.9426368684319214,
      "grad_norm": 0.48234230279922485,
      "learning_rate": 7.936247741680962e-06,
      "loss": 0.4655,
      "step": 20541
    },
    {
      "epoch": 0.9426827589371759,
      "grad_norm": 0.44105643033981323,
      "learning_rate": 7.936049281603419e-06,
      "loss": 0.374,
      "step": 20542
    },
    {
      "epoch": 0.9427286494424304,
      "grad_norm": 0.44646039605140686,
      "learning_rate": 7.935850814465606e-06,
      "loss": 0.3674,
      "step": 20543
    },
    {
      "epoch": 0.9427745399476848,
      "grad_norm": 0.4160648584365845,
      "learning_rate": 7.935652340267997e-06,
      "loss": 0.3087,
      "step": 20544
    },
    {
      "epoch": 0.9428204304529393,
      "grad_norm": 0.4639683663845062,
      "learning_rate": 7.93545385901107e-06,
      "loss": 0.4297,
      "step": 20545
    },
    {
      "epoch": 0.9428663209581938,
      "grad_norm": 0.4474770426750183,
      "learning_rate": 7.935255370695305e-06,
      "loss": 0.363,
      "step": 20546
    },
    {
      "epoch": 0.9429122114634482,
      "grad_norm": 0.43001678586006165,
      "learning_rate": 7.935056875321177e-06,
      "loss": 0.329,
      "step": 20547
    },
    {
      "epoch": 0.9429581019687027,
      "grad_norm": 0.444112628698349,
      "learning_rate": 7.934858372889163e-06,
      "loss": 0.3825,
      "step": 20548
    },
    {
      "epoch": 0.9430039924739572,
      "grad_norm": 0.4795422852039337,
      "learning_rate": 7.934659863399738e-06,
      "loss": 0.4728,
      "step": 20549
    },
    {
      "epoch": 0.9430498829792116,
      "grad_norm": 0.4481862783432007,
      "learning_rate": 7.934461346853387e-06,
      "loss": 0.4282,
      "step": 20550
    },
    {
      "epoch": 0.943095773484466,
      "grad_norm": 0.43370258808135986,
      "learning_rate": 7.934262823250579e-06,
      "loss": 0.3918,
      "step": 20551
    },
    {
      "epoch": 0.9431416639897205,
      "grad_norm": 0.45356547832489014,
      "learning_rate": 7.934064292591796e-06,
      "loss": 0.3973,
      "step": 20552
    },
    {
      "epoch": 0.943187554494975,
      "grad_norm": 0.4091116487979889,
      "learning_rate": 7.933865754877514e-06,
      "loss": 0.3087,
      "step": 20553
    },
    {
      "epoch": 0.9432334450002294,
      "grad_norm": 0.44692227244377136,
      "learning_rate": 7.933667210108213e-06,
      "loss": 0.3874,
      "step": 20554
    },
    {
      "epoch": 0.9432793355054839,
      "grad_norm": 0.47830453515052795,
      "learning_rate": 7.933468658284368e-06,
      "loss": 0.4373,
      "step": 20555
    },
    {
      "epoch": 0.9433252260107384,
      "grad_norm": 0.437059611082077,
      "learning_rate": 7.933270099406454e-06,
      "loss": 0.3545,
      "step": 20556
    },
    {
      "epoch": 0.9433711165159928,
      "grad_norm": 0.460025429725647,
      "learning_rate": 7.933071533474953e-06,
      "loss": 0.4303,
      "step": 20557
    },
    {
      "epoch": 0.9434170070212473,
      "grad_norm": 0.46681928634643555,
      "learning_rate": 7.93287296049034e-06,
      "loss": 0.3844,
      "step": 20558
    },
    {
      "epoch": 0.9434628975265018,
      "grad_norm": 0.45596134662628174,
      "learning_rate": 7.932674380453093e-06,
      "loss": 0.3945,
      "step": 20559
    },
    {
      "epoch": 0.9435087880317562,
      "grad_norm": 0.4022560119628906,
      "learning_rate": 7.93247579336369e-06,
      "loss": 0.2916,
      "step": 20560
    },
    {
      "epoch": 0.9435546785370107,
      "grad_norm": 0.4666641056537628,
      "learning_rate": 7.932277199222609e-06,
      "loss": 0.4167,
      "step": 20561
    },
    {
      "epoch": 0.9436005690422652,
      "grad_norm": 0.4485519528388977,
      "learning_rate": 7.932078598030328e-06,
      "loss": 0.3714,
      "step": 20562
    },
    {
      "epoch": 0.9436464595475196,
      "grad_norm": 0.4475579559803009,
      "learning_rate": 7.931879989787322e-06,
      "loss": 0.3691,
      "step": 20563
    },
    {
      "epoch": 0.9436923500527741,
      "grad_norm": 0.4419931471347809,
      "learning_rate": 7.931681374494068e-06,
      "loss": 0.371,
      "step": 20564
    },
    {
      "epoch": 0.9437382405580286,
      "grad_norm": 0.42998212575912476,
      "learning_rate": 7.931482752151047e-06,
      "loss": 0.3598,
      "step": 20565
    },
    {
      "epoch": 0.943784131063283,
      "grad_norm": 0.4144437313079834,
      "learning_rate": 7.931284122758736e-06,
      "loss": 0.3354,
      "step": 20566
    },
    {
      "epoch": 0.9438300215685375,
      "grad_norm": 0.4541586637496948,
      "learning_rate": 7.93108548631761e-06,
      "loss": 0.3829,
      "step": 20567
    },
    {
      "epoch": 0.943875912073792,
      "grad_norm": 0.41379663348197937,
      "learning_rate": 7.930886842828151e-06,
      "loss": 0.2998,
      "step": 20568
    },
    {
      "epoch": 0.9439218025790463,
      "grad_norm": 0.4231383800506592,
      "learning_rate": 7.930688192290832e-06,
      "loss": 0.3246,
      "step": 20569
    },
    {
      "epoch": 0.9439676930843008,
      "grad_norm": 0.5150914192199707,
      "learning_rate": 7.930489534706134e-06,
      "loss": 0.365,
      "step": 20570
    },
    {
      "epoch": 0.9440135835895553,
      "grad_norm": 0.4140473008155823,
      "learning_rate": 7.930290870074532e-06,
      "loss": 0.3285,
      "step": 20571
    },
    {
      "epoch": 0.9440594740948098,
      "grad_norm": 0.4788515865802765,
      "learning_rate": 7.930092198396508e-06,
      "loss": 0.4577,
      "step": 20572
    },
    {
      "epoch": 0.9441053646000642,
      "grad_norm": 0.4513852596282959,
      "learning_rate": 7.929893519672534e-06,
      "loss": 0.4023,
      "step": 20573
    },
    {
      "epoch": 0.9441512551053187,
      "grad_norm": 0.46589335799217224,
      "learning_rate": 7.929694833903093e-06,
      "loss": 0.3527,
      "step": 20574
    },
    {
      "epoch": 0.9441971456105732,
      "grad_norm": 0.44956740736961365,
      "learning_rate": 7.92949614108866e-06,
      "loss": 0.3999,
      "step": 20575
    },
    {
      "epoch": 0.9442430361158276,
      "grad_norm": 0.4463992118835449,
      "learning_rate": 7.929297441229714e-06,
      "loss": 0.3719,
      "step": 20576
    },
    {
      "epoch": 0.9442889266210821,
      "grad_norm": 0.46329715847969055,
      "learning_rate": 7.929098734326731e-06,
      "loss": 0.3625,
      "step": 20577
    },
    {
      "epoch": 0.9443348171263366,
      "grad_norm": 0.4653306007385254,
      "learning_rate": 7.92890002038019e-06,
      "loss": 0.416,
      "step": 20578
    },
    {
      "epoch": 0.944380707631591,
      "grad_norm": 0.43152478337287903,
      "learning_rate": 7.928701299390568e-06,
      "loss": 0.3312,
      "step": 20579
    },
    {
      "epoch": 0.9444265981368455,
      "grad_norm": 0.44317883253097534,
      "learning_rate": 7.928502571358347e-06,
      "loss": 0.3534,
      "step": 20580
    },
    {
      "epoch": 0.9444724886421,
      "grad_norm": 0.4298538565635681,
      "learning_rate": 7.928303836283999e-06,
      "loss": 0.3579,
      "step": 20581
    },
    {
      "epoch": 0.9445183791473544,
      "grad_norm": 0.4726695418357849,
      "learning_rate": 7.928105094168005e-06,
      "loss": 0.4102,
      "step": 20582
    },
    {
      "epoch": 0.9445642696526089,
      "grad_norm": 0.45329025387763977,
      "learning_rate": 7.927906345010843e-06,
      "loss": 0.3734,
      "step": 20583
    },
    {
      "epoch": 0.9446101601578634,
      "grad_norm": 0.4720298945903778,
      "learning_rate": 7.927707588812988e-06,
      "loss": 0.4127,
      "step": 20584
    },
    {
      "epoch": 0.9446560506631178,
      "grad_norm": 0.4826267957687378,
      "learning_rate": 7.927508825574923e-06,
      "loss": 0.4049,
      "step": 20585
    },
    {
      "epoch": 0.9447019411683723,
      "grad_norm": 0.47594770789146423,
      "learning_rate": 7.927310055297121e-06,
      "loss": 0.4457,
      "step": 20586
    },
    {
      "epoch": 0.9447478316736267,
      "grad_norm": 0.4583769142627716,
      "learning_rate": 7.927111277980065e-06,
      "loss": 0.4402,
      "step": 20587
    },
    {
      "epoch": 0.9447937221788811,
      "grad_norm": 0.5291416645050049,
      "learning_rate": 7.926912493624228e-06,
      "loss": 0.4824,
      "step": 20588
    },
    {
      "epoch": 0.9448396126841356,
      "grad_norm": 0.43763118982315063,
      "learning_rate": 7.92671370223009e-06,
      "loss": 0.3775,
      "step": 20589
    },
    {
      "epoch": 0.9448855031893901,
      "grad_norm": 0.4599716365337372,
      "learning_rate": 7.92651490379813e-06,
      "loss": 0.3965,
      "step": 20590
    },
    {
      "epoch": 0.9449313936946446,
      "grad_norm": 0.4727698564529419,
      "learning_rate": 7.926316098328825e-06,
      "loss": 0.3925,
      "step": 20591
    },
    {
      "epoch": 0.944977284199899,
      "grad_norm": 0.43567967414855957,
      "learning_rate": 7.926117285822654e-06,
      "loss": 0.3482,
      "step": 20592
    },
    {
      "epoch": 0.9450231747051535,
      "grad_norm": 0.4175252318382263,
      "learning_rate": 7.925918466280093e-06,
      "loss": 0.3107,
      "step": 20593
    },
    {
      "epoch": 0.945069065210408,
      "grad_norm": 0.45489469170570374,
      "learning_rate": 7.925719639701624e-06,
      "loss": 0.4015,
      "step": 20594
    },
    {
      "epoch": 0.9451149557156624,
      "grad_norm": 0.45946747064590454,
      "learning_rate": 7.92552080608772e-06,
      "loss": 0.4001,
      "step": 20595
    },
    {
      "epoch": 0.9451608462209169,
      "grad_norm": 0.4459153711795807,
      "learning_rate": 7.925321965438864e-06,
      "loss": 0.3661,
      "step": 20596
    },
    {
      "epoch": 0.9452067367261714,
      "grad_norm": 0.417134553194046,
      "learning_rate": 7.925123117755531e-06,
      "loss": 0.3228,
      "step": 20597
    },
    {
      "epoch": 0.9452526272314258,
      "grad_norm": 0.478728324174881,
      "learning_rate": 7.924924263038201e-06,
      "loss": 0.4073,
      "step": 20598
    },
    {
      "epoch": 0.9452985177366803,
      "grad_norm": 0.4658650755882263,
      "learning_rate": 7.92472540128735e-06,
      "loss": 0.43,
      "step": 20599
    },
    {
      "epoch": 0.9453444082419348,
      "grad_norm": 0.4409851133823395,
      "learning_rate": 7.924526532503459e-06,
      "loss": 0.3557,
      "step": 20600
    },
    {
      "epoch": 0.9453902987471892,
      "grad_norm": 0.4726598858833313,
      "learning_rate": 7.924327656687002e-06,
      "loss": 0.4445,
      "step": 20601
    },
    {
      "epoch": 0.9454361892524437,
      "grad_norm": 0.4191921353340149,
      "learning_rate": 7.924128773838463e-06,
      "loss": 0.3286,
      "step": 20602
    },
    {
      "epoch": 0.9454820797576982,
      "grad_norm": 0.5137768983840942,
      "learning_rate": 7.923929883958315e-06,
      "loss": 0.5433,
      "step": 20603
    },
    {
      "epoch": 0.9455279702629525,
      "grad_norm": 0.43485307693481445,
      "learning_rate": 7.923730987047038e-06,
      "loss": 0.3726,
      "step": 20604
    },
    {
      "epoch": 0.945573860768207,
      "grad_norm": 0.46116751432418823,
      "learning_rate": 7.923532083105112e-06,
      "loss": 0.413,
      "step": 20605
    },
    {
      "epoch": 0.9456197512734615,
      "grad_norm": 0.4280173182487488,
      "learning_rate": 7.923333172133013e-06,
      "loss": 0.3357,
      "step": 20606
    },
    {
      "epoch": 0.945665641778716,
      "grad_norm": 0.46372225880622864,
      "learning_rate": 7.923134254131221e-06,
      "loss": 0.3641,
      "step": 20607
    },
    {
      "epoch": 0.9457115322839704,
      "grad_norm": 0.4610538184642792,
      "learning_rate": 7.922935329100216e-06,
      "loss": 0.42,
      "step": 20608
    },
    {
      "epoch": 0.9457574227892249,
      "grad_norm": 0.45174291729927063,
      "learning_rate": 7.922736397040471e-06,
      "loss": 0.4167,
      "step": 20609
    },
    {
      "epoch": 0.9458033132944794,
      "grad_norm": 0.4547256827354431,
      "learning_rate": 7.922537457952468e-06,
      "loss": 0.361,
      "step": 20610
    },
    {
      "epoch": 0.9458492037997338,
      "grad_norm": 0.4482285678386688,
      "learning_rate": 7.922338511836686e-06,
      "loss": 0.4119,
      "step": 20611
    },
    {
      "epoch": 0.9458950943049883,
      "grad_norm": 0.4133509397506714,
      "learning_rate": 7.922139558693601e-06,
      "loss": 0.3624,
      "step": 20612
    },
    {
      "epoch": 0.9459409848102428,
      "grad_norm": 0.4788237512111664,
      "learning_rate": 7.921940598523695e-06,
      "loss": 0.475,
      "step": 20613
    },
    {
      "epoch": 0.9459868753154972,
      "grad_norm": 0.5016488432884216,
      "learning_rate": 7.92174163132744e-06,
      "loss": 0.4567,
      "step": 20614
    },
    {
      "epoch": 0.9460327658207517,
      "grad_norm": 0.41740864515304565,
      "learning_rate": 7.921542657105321e-06,
      "loss": 0.3528,
      "step": 20615
    },
    {
      "epoch": 0.9460786563260062,
      "grad_norm": 0.45040395855903625,
      "learning_rate": 7.921343675857813e-06,
      "loss": 0.4155,
      "step": 20616
    },
    {
      "epoch": 0.9461245468312606,
      "grad_norm": 0.4375624358654022,
      "learning_rate": 7.921144687585397e-06,
      "loss": 0.3534,
      "step": 20617
    },
    {
      "epoch": 0.9461704373365151,
      "grad_norm": 0.41773343086242676,
      "learning_rate": 7.92094569228855e-06,
      "loss": 0.3349,
      "step": 20618
    },
    {
      "epoch": 0.9462163278417696,
      "grad_norm": 0.45106643438339233,
      "learning_rate": 7.920746689967748e-06,
      "loss": 0.3813,
      "step": 20619
    },
    {
      "epoch": 0.946262218347024,
      "grad_norm": 0.4562336206436157,
      "learning_rate": 7.920547680623472e-06,
      "loss": 0.402,
      "step": 20620
    },
    {
      "epoch": 0.9463081088522785,
      "grad_norm": 0.45603933930397034,
      "learning_rate": 7.920348664256203e-06,
      "loss": 0.4117,
      "step": 20621
    },
    {
      "epoch": 0.946353999357533,
      "grad_norm": 0.42456790804862976,
      "learning_rate": 7.920149640866415e-06,
      "loss": 0.3332,
      "step": 20622
    },
    {
      "epoch": 0.9463998898627873,
      "grad_norm": 0.4077037274837494,
      "learning_rate": 7.91995061045459e-06,
      "loss": 0.3086,
      "step": 20623
    },
    {
      "epoch": 0.9464457803680418,
      "grad_norm": 0.45619267225265503,
      "learning_rate": 7.919751573021207e-06,
      "loss": 0.4484,
      "step": 20624
    },
    {
      "epoch": 0.9464916708732963,
      "grad_norm": 0.4805120527744293,
      "learning_rate": 7.919552528566741e-06,
      "loss": 0.484,
      "step": 20625
    },
    {
      "epoch": 0.9465375613785508,
      "grad_norm": 0.4972884953022003,
      "learning_rate": 7.91935347709167e-06,
      "loss": 0.4779,
      "step": 20626
    },
    {
      "epoch": 0.9465834518838052,
      "grad_norm": 0.46159547567367554,
      "learning_rate": 7.919154418596479e-06,
      "loss": 0.4523,
      "step": 20627
    },
    {
      "epoch": 0.9466293423890597,
      "grad_norm": 0.4503284990787506,
      "learning_rate": 7.918955353081644e-06,
      "loss": 0.391,
      "step": 20628
    },
    {
      "epoch": 0.9466752328943142,
      "grad_norm": 0.45481449365615845,
      "learning_rate": 7.91875628054764e-06,
      "loss": 0.3924,
      "step": 20629
    },
    {
      "epoch": 0.9467211233995686,
      "grad_norm": 0.40619122982025146,
      "learning_rate": 7.91855720099495e-06,
      "loss": 0.2942,
      "step": 20630
    },
    {
      "epoch": 0.9467670139048231,
      "grad_norm": 0.4413028955459595,
      "learning_rate": 7.918358114424048e-06,
      "loss": 0.3321,
      "step": 20631
    },
    {
      "epoch": 0.9468129044100776,
      "grad_norm": 0.45986267924308777,
      "learning_rate": 7.918159020835419e-06,
      "loss": 0.408,
      "step": 20632
    },
    {
      "epoch": 0.946858794915332,
      "grad_norm": 0.6788783073425293,
      "learning_rate": 7.917959920229537e-06,
      "loss": 0.4702,
      "step": 20633
    },
    {
      "epoch": 0.9469046854205865,
      "grad_norm": 0.44639572501182556,
      "learning_rate": 7.917760812606883e-06,
      "loss": 0.3881,
      "step": 20634
    },
    {
      "epoch": 0.946950575925841,
      "grad_norm": 0.43014928698539734,
      "learning_rate": 7.917561697967936e-06,
      "loss": 0.3422,
      "step": 20635
    },
    {
      "epoch": 0.9469964664310954,
      "grad_norm": 0.43558913469314575,
      "learning_rate": 7.917362576313171e-06,
      "loss": 0.3342,
      "step": 20636
    },
    {
      "epoch": 0.9470423569363499,
      "grad_norm": 0.49919962882995605,
      "learning_rate": 7.917163447643072e-06,
      "loss": 0.4866,
      "step": 20637
    },
    {
      "epoch": 0.9470882474416044,
      "grad_norm": 0.4474511444568634,
      "learning_rate": 7.916964311958115e-06,
      "loss": 0.4468,
      "step": 20638
    },
    {
      "epoch": 0.9471341379468587,
      "grad_norm": 0.442472904920578,
      "learning_rate": 7.91676516925878e-06,
      "loss": 0.3878,
      "step": 20639
    },
    {
      "epoch": 0.9471800284521132,
      "grad_norm": 0.4569697678089142,
      "learning_rate": 7.916566019545545e-06,
      "loss": 0.4273,
      "step": 20640
    },
    {
      "epoch": 0.9472259189573677,
      "grad_norm": 0.42616939544677734,
      "learning_rate": 7.916366862818886e-06,
      "loss": 0.3539,
      "step": 20641
    },
    {
      "epoch": 0.9472718094626222,
      "grad_norm": 0.49635106325149536,
      "learning_rate": 7.91616769907929e-06,
      "loss": 0.4673,
      "step": 20642
    },
    {
      "epoch": 0.9473176999678766,
      "grad_norm": 0.48511362075805664,
      "learning_rate": 7.915968528327229e-06,
      "loss": 0.4588,
      "step": 20643
    },
    {
      "epoch": 0.9473635904731311,
      "grad_norm": 0.4555867612361908,
      "learning_rate": 7.915769350563184e-06,
      "loss": 0.3828,
      "step": 20644
    },
    {
      "epoch": 0.9474094809783856,
      "grad_norm": 0.48230454325675964,
      "learning_rate": 7.915570165787637e-06,
      "loss": 0.3997,
      "step": 20645
    },
    {
      "epoch": 0.94745537148364,
      "grad_norm": 0.4838423430919647,
      "learning_rate": 7.91537097400106e-06,
      "loss": 0.4573,
      "step": 20646
    },
    {
      "epoch": 0.9475012619888945,
      "grad_norm": 0.4689597189426422,
      "learning_rate": 7.915171775203936e-06,
      "loss": 0.3862,
      "step": 20647
    },
    {
      "epoch": 0.947547152494149,
      "grad_norm": 0.47464242577552795,
      "learning_rate": 7.914972569396746e-06,
      "loss": 0.4179,
      "step": 20648
    },
    {
      "epoch": 0.9475930429994034,
      "grad_norm": 0.430608332157135,
      "learning_rate": 7.914773356579966e-06,
      "loss": 0.3228,
      "step": 20649
    },
    {
      "epoch": 0.9476389335046579,
      "grad_norm": 0.4151085317134857,
      "learning_rate": 7.914574136754075e-06,
      "loss": 0.313,
      "step": 20650
    },
    {
      "epoch": 0.9476848240099124,
      "grad_norm": 0.4168482720851898,
      "learning_rate": 7.914374909919554e-06,
      "loss": 0.3585,
      "step": 20651
    },
    {
      "epoch": 0.9477307145151668,
      "grad_norm": 0.5066651105880737,
      "learning_rate": 7.91417567607688e-06,
      "loss": 0.4477,
      "step": 20652
    },
    {
      "epoch": 0.9477766050204213,
      "grad_norm": 0.4801095128059387,
      "learning_rate": 7.913976435226537e-06,
      "loss": 0.4523,
      "step": 20653
    },
    {
      "epoch": 0.9478224955256758,
      "grad_norm": 0.42309999465942383,
      "learning_rate": 7.913777187368996e-06,
      "loss": 0.3095,
      "step": 20654
    },
    {
      "epoch": 0.9478683860309302,
      "grad_norm": 0.490983247756958,
      "learning_rate": 7.913577932504742e-06,
      "loss": 0.4194,
      "step": 20655
    },
    {
      "epoch": 0.9479142765361847,
      "grad_norm": 0.4292239546775818,
      "learning_rate": 7.913378670634254e-06,
      "loss": 0.3263,
      "step": 20656
    },
    {
      "epoch": 0.9479601670414392,
      "grad_norm": 0.49975183606147766,
      "learning_rate": 7.913179401758008e-06,
      "loss": 0.469,
      "step": 20657
    },
    {
      "epoch": 0.9480060575466935,
      "grad_norm": 0.4253979027271271,
      "learning_rate": 7.912980125876486e-06,
      "loss": 0.3346,
      "step": 20658
    },
    {
      "epoch": 0.948051948051948,
      "grad_norm": 0.4483835697174072,
      "learning_rate": 7.912780842990166e-06,
      "loss": 0.3816,
      "step": 20659
    },
    {
      "epoch": 0.9480978385572025,
      "grad_norm": 0.435971736907959,
      "learning_rate": 7.912581553099527e-06,
      "loss": 0.4119,
      "step": 20660
    },
    {
      "epoch": 0.948143729062457,
      "grad_norm": 0.4344462454319,
      "learning_rate": 7.912382256205049e-06,
      "loss": 0.3594,
      "step": 20661
    },
    {
      "epoch": 0.9481896195677114,
      "grad_norm": 0.4509414732456207,
      "learning_rate": 7.91218295230721e-06,
      "loss": 0.3616,
      "step": 20662
    },
    {
      "epoch": 0.9482355100729659,
      "grad_norm": 0.4775043725967407,
      "learning_rate": 7.91198364140649e-06,
      "loss": 0.4025,
      "step": 20663
    },
    {
      "epoch": 0.9482814005782204,
      "grad_norm": 0.483792781829834,
      "learning_rate": 7.91178432350337e-06,
      "loss": 0.4983,
      "step": 20664
    },
    {
      "epoch": 0.9483272910834748,
      "grad_norm": 0.5223774313926697,
      "learning_rate": 7.911584998598328e-06,
      "loss": 0.5122,
      "step": 20665
    },
    {
      "epoch": 0.9483731815887293,
      "grad_norm": 0.4326172173023224,
      "learning_rate": 7.91138566669184e-06,
      "loss": 0.3483,
      "step": 20666
    },
    {
      "epoch": 0.9484190720939838,
      "grad_norm": 0.4393482804298401,
      "learning_rate": 7.911186327784391e-06,
      "loss": 0.3757,
      "step": 20667
    },
    {
      "epoch": 0.9484649625992382,
      "grad_norm": 0.45733290910720825,
      "learning_rate": 7.910986981876457e-06,
      "loss": 0.3374,
      "step": 20668
    },
    {
      "epoch": 0.9485108531044927,
      "grad_norm": 0.4015486240386963,
      "learning_rate": 7.910787628968516e-06,
      "loss": 0.296,
      "step": 20669
    },
    {
      "epoch": 0.9485567436097472,
      "grad_norm": 0.4993155300617218,
      "learning_rate": 7.910588269061053e-06,
      "loss": 0.4599,
      "step": 20670
    },
    {
      "epoch": 0.9486026341150016,
      "grad_norm": 0.4317129850387573,
      "learning_rate": 7.910388902154541e-06,
      "loss": 0.3018,
      "step": 20671
    },
    {
      "epoch": 0.9486485246202561,
      "grad_norm": 0.48533621430397034,
      "learning_rate": 7.910189528249464e-06,
      "loss": 0.4709,
      "step": 20672
    },
    {
      "epoch": 0.9486944151255106,
      "grad_norm": 0.48693662881851196,
      "learning_rate": 7.909990147346299e-06,
      "loss": 0.457,
      "step": 20673
    },
    {
      "epoch": 0.948740305630765,
      "grad_norm": 0.4365951716899872,
      "learning_rate": 7.909790759445525e-06,
      "loss": 0.4013,
      "step": 20674
    },
    {
      "epoch": 0.9487861961360194,
      "grad_norm": 0.43023934960365295,
      "learning_rate": 7.909591364547624e-06,
      "loss": 0.3515,
      "step": 20675
    },
    {
      "epoch": 0.9488320866412739,
      "grad_norm": 0.497045636177063,
      "learning_rate": 7.909391962653075e-06,
      "loss": 0.4461,
      "step": 20676
    },
    {
      "epoch": 0.9488779771465283,
      "grad_norm": 0.5104749798774719,
      "learning_rate": 7.909192553762355e-06,
      "loss": 0.4884,
      "step": 20677
    },
    {
      "epoch": 0.9489238676517828,
      "grad_norm": 0.43133291602134705,
      "learning_rate": 7.908993137875946e-06,
      "loss": 0.3429,
      "step": 20678
    },
    {
      "epoch": 0.9489697581570373,
      "grad_norm": 0.46663662791252136,
      "learning_rate": 7.908793714994325e-06,
      "loss": 0.3907,
      "step": 20679
    },
    {
      "epoch": 0.9490156486622918,
      "grad_norm": 0.43567994236946106,
      "learning_rate": 7.908594285117974e-06,
      "loss": 0.3703,
      "step": 20680
    },
    {
      "epoch": 0.9490615391675462,
      "grad_norm": 0.43265992403030396,
      "learning_rate": 7.908394848247373e-06,
      "loss": 0.3345,
      "step": 20681
    },
    {
      "epoch": 0.9491074296728007,
      "grad_norm": 0.41491079330444336,
      "learning_rate": 7.908195404382998e-06,
      "loss": 0.3619,
      "step": 20682
    },
    {
      "epoch": 0.9491533201780552,
      "grad_norm": 0.47136786580085754,
      "learning_rate": 7.907995953525332e-06,
      "loss": 0.4477,
      "step": 20683
    },
    {
      "epoch": 0.9491992106833096,
      "grad_norm": 0.48946306109428406,
      "learning_rate": 7.907796495674856e-06,
      "loss": 0.4438,
      "step": 20684
    },
    {
      "epoch": 0.9492451011885641,
      "grad_norm": 0.5118829011917114,
      "learning_rate": 7.907597030832042e-06,
      "loss": 0.4825,
      "step": 20685
    },
    {
      "epoch": 0.9492909916938186,
      "grad_norm": 0.48146259784698486,
      "learning_rate": 7.907397558997378e-06,
      "loss": 0.4707,
      "step": 20686
    },
    {
      "epoch": 0.949336882199073,
      "grad_norm": 0.4402076005935669,
      "learning_rate": 7.90719808017134e-06,
      "loss": 0.3964,
      "step": 20687
    },
    {
      "epoch": 0.9493827727043275,
      "grad_norm": 0.4592094123363495,
      "learning_rate": 7.906998594354408e-06,
      "loss": 0.4473,
      "step": 20688
    },
    {
      "epoch": 0.949428663209582,
      "grad_norm": 0.46768084168434143,
      "learning_rate": 7.906799101547062e-06,
      "loss": 0.3994,
      "step": 20689
    },
    {
      "epoch": 0.9494745537148364,
      "grad_norm": 0.41256946325302124,
      "learning_rate": 7.906599601749782e-06,
      "loss": 0.3389,
      "step": 20690
    },
    {
      "epoch": 0.9495204442200909,
      "grad_norm": 0.4697292149066925,
      "learning_rate": 7.906400094963047e-06,
      "loss": 0.4382,
      "step": 20691
    },
    {
      "epoch": 0.9495663347253454,
      "grad_norm": 0.4967111647129059,
      "learning_rate": 7.906200581187338e-06,
      "loss": 0.5058,
      "step": 20692
    },
    {
      "epoch": 0.9496122252305997,
      "grad_norm": 0.4405536949634552,
      "learning_rate": 7.906001060423132e-06,
      "loss": 0.3582,
      "step": 20693
    },
    {
      "epoch": 0.9496581157358542,
      "grad_norm": 0.4394819438457489,
      "learning_rate": 7.905801532670912e-06,
      "loss": 0.4288,
      "step": 20694
    },
    {
      "epoch": 0.9497040062411087,
      "grad_norm": 0.9397099614143372,
      "learning_rate": 7.905601997931156e-06,
      "loss": 0.3832,
      "step": 20695
    },
    {
      "epoch": 0.9497498967463632,
      "grad_norm": 0.48197418451309204,
      "learning_rate": 7.905402456204343e-06,
      "loss": 0.434,
      "step": 20696
    },
    {
      "epoch": 0.9497957872516176,
      "grad_norm": 0.48722895979881287,
      "learning_rate": 7.905202907490956e-06,
      "loss": 0.4268,
      "step": 20697
    },
    {
      "epoch": 0.9498416777568721,
      "grad_norm": 0.440244197845459,
      "learning_rate": 7.905003351791472e-06,
      "loss": 0.3725,
      "step": 20698
    },
    {
      "epoch": 0.9498875682621266,
      "grad_norm": 0.41337862610816956,
      "learning_rate": 7.90480378910637e-06,
      "loss": 0.3324,
      "step": 20699
    },
    {
      "epoch": 0.949933458767381,
      "grad_norm": 0.5012292265892029,
      "learning_rate": 7.904604219436135e-06,
      "loss": 0.4619,
      "step": 20700
    },
    {
      "epoch": 0.9499793492726355,
      "grad_norm": 0.48096033930778503,
      "learning_rate": 7.904404642781243e-06,
      "loss": 0.4099,
      "step": 20701
    },
    {
      "epoch": 0.95002523977789,
      "grad_norm": 0.6364090442657471,
      "learning_rate": 7.904205059142173e-06,
      "loss": 0.3482,
      "step": 20702
    },
    {
      "epoch": 0.9500711302831444,
      "grad_norm": 0.4528335630893707,
      "learning_rate": 7.904005468519407e-06,
      "loss": 0.3697,
      "step": 20703
    },
    {
      "epoch": 0.9501170207883989,
      "grad_norm": 0.4561319947242737,
      "learning_rate": 7.903805870913424e-06,
      "loss": 0.3253,
      "step": 20704
    },
    {
      "epoch": 0.9501629112936534,
      "grad_norm": 0.4610041677951813,
      "learning_rate": 7.903606266324706e-06,
      "loss": 0.428,
      "step": 20705
    },
    {
      "epoch": 0.9502088017989078,
      "grad_norm": 0.47606360912323,
      "learning_rate": 7.90340665475373e-06,
      "loss": 0.4612,
      "step": 20706
    },
    {
      "epoch": 0.9502546923041623,
      "grad_norm": 0.41068035364151,
      "learning_rate": 7.903207036200977e-06,
      "loss": 0.3229,
      "step": 20707
    },
    {
      "epoch": 0.9503005828094168,
      "grad_norm": 0.5289531350135803,
      "learning_rate": 7.903007410666927e-06,
      "loss": 0.4248,
      "step": 20708
    },
    {
      "epoch": 0.9503464733146711,
      "grad_norm": 0.4537547826766968,
      "learning_rate": 7.902807778152061e-06,
      "loss": 0.4066,
      "step": 20709
    },
    {
      "epoch": 0.9503923638199256,
      "grad_norm": 0.4345776438713074,
      "learning_rate": 7.902608138656859e-06,
      "loss": 0.3457,
      "step": 20710
    },
    {
      "epoch": 0.9504382543251801,
      "grad_norm": 0.44283393025398254,
      "learning_rate": 7.902408492181798e-06,
      "loss": 0.3896,
      "step": 20711
    },
    {
      "epoch": 0.9504841448304345,
      "grad_norm": 0.4521791636943817,
      "learning_rate": 7.902208838727366e-06,
      "loss": 0.3989,
      "step": 20712
    },
    {
      "epoch": 0.950530035335689,
      "grad_norm": 0.454632967710495,
      "learning_rate": 7.902009178294033e-06,
      "loss": 0.3949,
      "step": 20713
    },
    {
      "epoch": 0.9505759258409435,
      "grad_norm": 0.41678258776664734,
      "learning_rate": 7.901809510882285e-06,
      "loss": 0.3321,
      "step": 20714
    },
    {
      "epoch": 0.950621816346198,
      "grad_norm": 0.4800679385662079,
      "learning_rate": 7.9016098364926e-06,
      "loss": 0.4598,
      "step": 20715
    },
    {
      "epoch": 0.9506677068514524,
      "grad_norm": 0.45400476455688477,
      "learning_rate": 7.901410155125462e-06,
      "loss": 0.38,
      "step": 20716
    },
    {
      "epoch": 0.9507135973567069,
      "grad_norm": 0.44501349329948425,
      "learning_rate": 7.901210466781347e-06,
      "loss": 0.3791,
      "step": 20717
    },
    {
      "epoch": 0.9507594878619614,
      "grad_norm": 0.43400585651397705,
      "learning_rate": 7.901010771460736e-06,
      "loss": 0.3862,
      "step": 20718
    },
    {
      "epoch": 0.9508053783672158,
      "grad_norm": 0.43263059854507446,
      "learning_rate": 7.90081106916411e-06,
      "loss": 0.3346,
      "step": 20719
    },
    {
      "epoch": 0.9508512688724703,
      "grad_norm": 0.4371837377548218,
      "learning_rate": 7.900611359891948e-06,
      "loss": 0.3623,
      "step": 20720
    },
    {
      "epoch": 0.9508971593777248,
      "grad_norm": 0.4345303773880005,
      "learning_rate": 7.900411643644732e-06,
      "loss": 0.367,
      "step": 20721
    },
    {
      "epoch": 0.9509430498829792,
      "grad_norm": 0.4456794857978821,
      "learning_rate": 7.90021192042294e-06,
      "loss": 0.3879,
      "step": 20722
    },
    {
      "epoch": 0.9509889403882337,
      "grad_norm": 0.4169251322746277,
      "learning_rate": 7.900012190227056e-06,
      "loss": 0.3399,
      "step": 20723
    },
    {
      "epoch": 0.9510348308934882,
      "grad_norm": 0.5187283754348755,
      "learning_rate": 7.899812453057555e-06,
      "loss": 0.409,
      "step": 20724
    },
    {
      "epoch": 0.9510807213987426,
      "grad_norm": 0.4520646929740906,
      "learning_rate": 7.899612708914924e-06,
      "loss": 0.3881,
      "step": 20725
    },
    {
      "epoch": 0.951126611903997,
      "grad_norm": 0.5836667418479919,
      "learning_rate": 7.899412957799638e-06,
      "loss": 0.372,
      "step": 20726
    },
    {
      "epoch": 0.9511725024092516,
      "grad_norm": 0.4666779935359955,
      "learning_rate": 7.899213199712177e-06,
      "loss": 0.359,
      "step": 20727
    },
    {
      "epoch": 0.9512183929145059,
      "grad_norm": 0.46841078996658325,
      "learning_rate": 7.899013434653025e-06,
      "loss": 0.4067,
      "step": 20728
    },
    {
      "epoch": 0.9512642834197604,
      "grad_norm": 0.4843532145023346,
      "learning_rate": 7.89881366262266e-06,
      "loss": 0.4617,
      "step": 20729
    },
    {
      "epoch": 0.9513101739250149,
      "grad_norm": 0.47399812936782837,
      "learning_rate": 7.898613883621563e-06,
      "loss": 0.4484,
      "step": 20730
    },
    {
      "epoch": 0.9513560644302694,
      "grad_norm": 0.46369126439094543,
      "learning_rate": 7.898414097650216e-06,
      "loss": 0.4273,
      "step": 20731
    },
    {
      "epoch": 0.9514019549355238,
      "grad_norm": 0.5030677318572998,
      "learning_rate": 7.898214304709098e-06,
      "loss": 0.5421,
      "step": 20732
    },
    {
      "epoch": 0.9514478454407783,
      "grad_norm": 0.4228552579879761,
      "learning_rate": 7.898014504798688e-06,
      "loss": 0.3583,
      "step": 20733
    },
    {
      "epoch": 0.9514937359460328,
      "grad_norm": 0.4302417039871216,
      "learning_rate": 7.897814697919467e-06,
      "loss": 0.3899,
      "step": 20734
    },
    {
      "epoch": 0.9515396264512872,
      "grad_norm": 0.43733251094818115,
      "learning_rate": 7.897614884071918e-06,
      "loss": 0.4149,
      "step": 20735
    },
    {
      "epoch": 0.9515855169565417,
      "grad_norm": 0.4229368567466736,
      "learning_rate": 7.897415063256516e-06,
      "loss": 0.3587,
      "step": 20736
    },
    {
      "epoch": 0.9516314074617962,
      "grad_norm": 0.4426785409450531,
      "learning_rate": 7.89721523547375e-06,
      "loss": 0.3554,
      "step": 20737
    },
    {
      "epoch": 0.9516772979670506,
      "grad_norm": 0.43730103969573975,
      "learning_rate": 7.897015400724094e-06,
      "loss": 0.403,
      "step": 20738
    },
    {
      "epoch": 0.9517231884723051,
      "grad_norm": 0.46226879954338074,
      "learning_rate": 7.89681555900803e-06,
      "loss": 0.3828,
      "step": 20739
    },
    {
      "epoch": 0.9517690789775596,
      "grad_norm": 0.49004727602005005,
      "learning_rate": 7.896615710326039e-06,
      "loss": 0.4405,
      "step": 20740
    },
    {
      "epoch": 0.951814969482814,
      "grad_norm": 0.41702762246131897,
      "learning_rate": 7.896415854678601e-06,
      "loss": 0.3253,
      "step": 20741
    },
    {
      "epoch": 0.9518608599880685,
      "grad_norm": 0.4566820561885834,
      "learning_rate": 7.896215992066198e-06,
      "loss": 0.4242,
      "step": 20742
    },
    {
      "epoch": 0.951906750493323,
      "grad_norm": 0.42468541860580444,
      "learning_rate": 7.896016122489308e-06,
      "loss": 0.3775,
      "step": 20743
    },
    {
      "epoch": 0.9519526409985773,
      "grad_norm": 1.297866940498352,
      "learning_rate": 7.895816245948415e-06,
      "loss": 0.4291,
      "step": 20744
    },
    {
      "epoch": 0.9519985315038318,
      "grad_norm": 0.43348339200019836,
      "learning_rate": 7.895616362443997e-06,
      "loss": 0.335,
      "step": 20745
    },
    {
      "epoch": 0.9520444220090863,
      "grad_norm": 0.49392303824424744,
      "learning_rate": 7.895416471976536e-06,
      "loss": 0.3662,
      "step": 20746
    },
    {
      "epoch": 0.9520903125143407,
      "grad_norm": 0.44551748037338257,
      "learning_rate": 7.895216574546511e-06,
      "loss": 0.3639,
      "step": 20747
    },
    {
      "epoch": 0.9521362030195952,
      "grad_norm": 0.521085798740387,
      "learning_rate": 7.895016670154404e-06,
      "loss": 0.4802,
      "step": 20748
    },
    {
      "epoch": 0.9521820935248497,
      "grad_norm": 0.4230537414550781,
      "learning_rate": 7.894816758800698e-06,
      "loss": 0.3342,
      "step": 20749
    },
    {
      "epoch": 0.9522279840301042,
      "grad_norm": 0.4655434191226959,
      "learning_rate": 7.89461684048587e-06,
      "loss": 0.4184,
      "step": 20750
    },
    {
      "epoch": 0.9522738745353586,
      "grad_norm": 0.4300721287727356,
      "learning_rate": 7.8944169152104e-06,
      "loss": 0.3741,
      "step": 20751
    },
    {
      "epoch": 0.9523197650406131,
      "grad_norm": 0.45191511511802673,
      "learning_rate": 7.894216982974775e-06,
      "loss": 0.4049,
      "step": 20752
    },
    {
      "epoch": 0.9523656555458676,
      "grad_norm": 0.46399304270744324,
      "learning_rate": 7.894017043779468e-06,
      "loss": 0.3554,
      "step": 20753
    },
    {
      "epoch": 0.952411546051122,
      "grad_norm": 0.49666401743888855,
      "learning_rate": 7.893817097624962e-06,
      "loss": 0.4969,
      "step": 20754
    },
    {
      "epoch": 0.9524574365563765,
      "grad_norm": 0.44455644488334656,
      "learning_rate": 7.893617144511742e-06,
      "loss": 0.3802,
      "step": 20755
    },
    {
      "epoch": 0.952503327061631,
      "grad_norm": 0.43826672434806824,
      "learning_rate": 7.893417184440287e-06,
      "loss": 0.3319,
      "step": 20756
    },
    {
      "epoch": 0.9525492175668854,
      "grad_norm": 0.46469804644584656,
      "learning_rate": 7.893217217411076e-06,
      "loss": 0.4246,
      "step": 20757
    },
    {
      "epoch": 0.9525951080721399,
      "grad_norm": 0.612488865852356,
      "learning_rate": 7.89301724342459e-06,
      "loss": 0.3574,
      "step": 20758
    },
    {
      "epoch": 0.9526409985773944,
      "grad_norm": 0.4565034806728363,
      "learning_rate": 7.89281726248131e-06,
      "loss": 0.3851,
      "step": 20759
    },
    {
      "epoch": 0.9526868890826488,
      "grad_norm": 0.4787023365497589,
      "learning_rate": 7.89261727458172e-06,
      "loss": 0.4298,
      "step": 20760
    },
    {
      "epoch": 0.9527327795879033,
      "grad_norm": 0.4315487742424011,
      "learning_rate": 7.892417279726296e-06,
      "loss": 0.3318,
      "step": 20761
    },
    {
      "epoch": 0.9527786700931578,
      "grad_norm": 0.4546329975128174,
      "learning_rate": 7.892217277915522e-06,
      "loss": 0.3912,
      "step": 20762
    },
    {
      "epoch": 0.9528245605984121,
      "grad_norm": 0.4488099217414856,
      "learning_rate": 7.892017269149879e-06,
      "loss": 0.4382,
      "step": 20763
    },
    {
      "epoch": 0.9528704511036666,
      "grad_norm": 0.47105345129966736,
      "learning_rate": 7.891817253429845e-06,
      "loss": 0.4058,
      "step": 20764
    },
    {
      "epoch": 0.9529163416089211,
      "grad_norm": 0.45920976996421814,
      "learning_rate": 7.891617230755906e-06,
      "loss": 0.4499,
      "step": 20765
    },
    {
      "epoch": 0.9529622321141755,
      "grad_norm": 0.44265061616897583,
      "learning_rate": 7.891417201128538e-06,
      "loss": 0.3386,
      "step": 20766
    },
    {
      "epoch": 0.95300812261943,
      "grad_norm": 0.42424607276916504,
      "learning_rate": 7.891217164548227e-06,
      "loss": 0.3487,
      "step": 20767
    },
    {
      "epoch": 0.9530540131246845,
      "grad_norm": 0.4415209889411926,
      "learning_rate": 7.891017121015447e-06,
      "loss": 0.3918,
      "step": 20768
    },
    {
      "epoch": 0.953099903629939,
      "grad_norm": 0.44983819127082825,
      "learning_rate": 7.890817070530686e-06,
      "loss": 0.4442,
      "step": 20769
    },
    {
      "epoch": 0.9531457941351934,
      "grad_norm": 0.46973806619644165,
      "learning_rate": 7.890617013094421e-06,
      "loss": 0.3998,
      "step": 20770
    },
    {
      "epoch": 0.9531916846404479,
      "grad_norm": 0.4453645944595337,
      "learning_rate": 7.890416948707135e-06,
      "loss": 0.3927,
      "step": 20771
    },
    {
      "epoch": 0.9532375751457024,
      "grad_norm": 0.4991135597229004,
      "learning_rate": 7.89021687736931e-06,
      "loss": 0.497,
      "step": 20772
    },
    {
      "epoch": 0.9532834656509568,
      "grad_norm": 0.5002502202987671,
      "learning_rate": 7.890016799081424e-06,
      "loss": 0.4351,
      "step": 20773
    },
    {
      "epoch": 0.9533293561562113,
      "grad_norm": 0.45086973905563354,
      "learning_rate": 7.88981671384396e-06,
      "loss": 0.3636,
      "step": 20774
    },
    {
      "epoch": 0.9533752466614658,
      "grad_norm": 0.44091665744781494,
      "learning_rate": 7.889616621657399e-06,
      "loss": 0.3162,
      "step": 20775
    },
    {
      "epoch": 0.9534211371667202,
      "grad_norm": 0.4561329782009125,
      "learning_rate": 7.889416522522223e-06,
      "loss": 0.4344,
      "step": 20776
    },
    {
      "epoch": 0.9534670276719747,
      "grad_norm": 0.424672394990921,
      "learning_rate": 7.889216416438911e-06,
      "loss": 0.3273,
      "step": 20777
    },
    {
      "epoch": 0.9535129181772292,
      "grad_norm": 0.4568372070789337,
      "learning_rate": 7.889016303407944e-06,
      "loss": 0.3587,
      "step": 20778
    },
    {
      "epoch": 0.9535588086824835,
      "grad_norm": 0.4501543343067169,
      "learning_rate": 7.888816183429806e-06,
      "loss": 0.3833,
      "step": 20779
    },
    {
      "epoch": 0.953604699187738,
      "grad_norm": 0.4728638231754303,
      "learning_rate": 7.888616056504978e-06,
      "loss": 0.4522,
      "step": 20780
    },
    {
      "epoch": 0.9536505896929925,
      "grad_norm": 0.4682685136795044,
      "learning_rate": 7.888415922633938e-06,
      "loss": 0.3971,
      "step": 20781
    },
    {
      "epoch": 0.9536964801982469,
      "grad_norm": 0.42580798268318176,
      "learning_rate": 7.88821578181717e-06,
      "loss": 0.4002,
      "step": 20782
    },
    {
      "epoch": 0.9537423707035014,
      "grad_norm": 0.45228299498558044,
      "learning_rate": 7.888015634055155e-06,
      "loss": 0.4241,
      "step": 20783
    },
    {
      "epoch": 0.9537882612087559,
      "grad_norm": 0.4737030565738678,
      "learning_rate": 7.887815479348375e-06,
      "loss": 0.4563,
      "step": 20784
    },
    {
      "epoch": 0.9538341517140104,
      "grad_norm": 0.4778626561164856,
      "learning_rate": 7.887615317697308e-06,
      "loss": 0.462,
      "step": 20785
    },
    {
      "epoch": 0.9538800422192648,
      "grad_norm": 0.45809170603752136,
      "learning_rate": 7.887415149102438e-06,
      "loss": 0.4037,
      "step": 20786
    },
    {
      "epoch": 0.9539259327245193,
      "grad_norm": 0.4994875490665436,
      "learning_rate": 7.887214973564248e-06,
      "loss": 0.422,
      "step": 20787
    },
    {
      "epoch": 0.9539718232297738,
      "grad_norm": 0.4918251633644104,
      "learning_rate": 7.887014791083213e-06,
      "loss": 0.5085,
      "step": 20788
    },
    {
      "epoch": 0.9540177137350282,
      "grad_norm": 0.5089061260223389,
      "learning_rate": 7.886814601659822e-06,
      "loss": 0.4307,
      "step": 20789
    },
    {
      "epoch": 0.9540636042402827,
      "grad_norm": 0.4239089787006378,
      "learning_rate": 7.886614405294552e-06,
      "loss": 0.3622,
      "step": 20790
    },
    {
      "epoch": 0.9541094947455372,
      "grad_norm": 0.5597901344299316,
      "learning_rate": 7.886414201987886e-06,
      "loss": 0.5794,
      "step": 20791
    },
    {
      "epoch": 0.9541553852507916,
      "grad_norm": 0.45320701599121094,
      "learning_rate": 7.886213991740305e-06,
      "loss": 0.4071,
      "step": 20792
    },
    {
      "epoch": 0.9542012757560461,
      "grad_norm": 0.474230021238327,
      "learning_rate": 7.88601377455229e-06,
      "loss": 0.4685,
      "step": 20793
    },
    {
      "epoch": 0.9542471662613006,
      "grad_norm": 0.4227583110332489,
      "learning_rate": 7.885813550424322e-06,
      "loss": 0.3669,
      "step": 20794
    },
    {
      "epoch": 0.954293056766555,
      "grad_norm": 0.46631181240081787,
      "learning_rate": 7.885613319356885e-06,
      "loss": 0.4262,
      "step": 20795
    },
    {
      "epoch": 0.9543389472718095,
      "grad_norm": 0.45406773686408997,
      "learning_rate": 7.885413081350455e-06,
      "loss": 0.4159,
      "step": 20796
    },
    {
      "epoch": 0.954384837777064,
      "grad_norm": 0.45712265372276306,
      "learning_rate": 7.88521283640552e-06,
      "loss": 0.4107,
      "step": 20797
    },
    {
      "epoch": 0.9544307282823183,
      "grad_norm": 0.7764745354652405,
      "learning_rate": 7.885012584522557e-06,
      "loss": 0.4635,
      "step": 20798
    },
    {
      "epoch": 0.9544766187875728,
      "grad_norm": 0.44126757979393005,
      "learning_rate": 7.884812325702051e-06,
      "loss": 0.4059,
      "step": 20799
    },
    {
      "epoch": 0.9545225092928273,
      "grad_norm": 0.449358195066452,
      "learning_rate": 7.884612059944481e-06,
      "loss": 0.3043,
      "step": 20800
    },
    {
      "epoch": 0.9545683997980817,
      "grad_norm": 0.437394380569458,
      "learning_rate": 7.88441178725033e-06,
      "loss": 0.4103,
      "step": 20801
    },
    {
      "epoch": 0.9546142903033362,
      "grad_norm": 0.4676348865032196,
      "learning_rate": 7.884211507620079e-06,
      "loss": 0.3948,
      "step": 20802
    },
    {
      "epoch": 0.9546601808085907,
      "grad_norm": 0.4307887852191925,
      "learning_rate": 7.884011221054208e-06,
      "loss": 0.3514,
      "step": 20803
    },
    {
      "epoch": 0.9547060713138452,
      "grad_norm": 0.5165539979934692,
      "learning_rate": 7.883810927553203e-06,
      "loss": 0.4866,
      "step": 20804
    },
    {
      "epoch": 0.9547519618190996,
      "grad_norm": 0.44806474447250366,
      "learning_rate": 7.88361062711754e-06,
      "loss": 0.3778,
      "step": 20805
    },
    {
      "epoch": 0.9547978523243541,
      "grad_norm": 0.47774726152420044,
      "learning_rate": 7.883410319747706e-06,
      "loss": 0.4258,
      "step": 20806
    },
    {
      "epoch": 0.9548437428296086,
      "grad_norm": 0.49291083216667175,
      "learning_rate": 7.883210005444179e-06,
      "loss": 0.4453,
      "step": 20807
    },
    {
      "epoch": 0.954889633334863,
      "grad_norm": 0.4454309940338135,
      "learning_rate": 7.883009684207441e-06,
      "loss": 0.3518,
      "step": 20808
    },
    {
      "epoch": 0.9549355238401175,
      "grad_norm": 0.46012288331985474,
      "learning_rate": 7.882809356037977e-06,
      "loss": 0.4144,
      "step": 20809
    },
    {
      "epoch": 0.954981414345372,
      "grad_norm": 0.4394964873790741,
      "learning_rate": 7.882609020936265e-06,
      "loss": 0.3166,
      "step": 20810
    },
    {
      "epoch": 0.9550273048506264,
      "grad_norm": 0.43290844559669495,
      "learning_rate": 7.882408678902789e-06,
      "loss": 0.3412,
      "step": 20811
    },
    {
      "epoch": 0.9550731953558809,
      "grad_norm": 0.4824354648590088,
      "learning_rate": 7.882208329938028e-06,
      "loss": 0.4184,
      "step": 20812
    },
    {
      "epoch": 0.9551190858611354,
      "grad_norm": 0.45939159393310547,
      "learning_rate": 7.882007974042466e-06,
      "loss": 0.3937,
      "step": 20813
    },
    {
      "epoch": 0.9551649763663898,
      "grad_norm": 0.4506071209907532,
      "learning_rate": 7.881807611216585e-06,
      "loss": 0.4092,
      "step": 20814
    },
    {
      "epoch": 0.9552108668716442,
      "grad_norm": 0.48458293080329895,
      "learning_rate": 7.881607241460865e-06,
      "loss": 0.4812,
      "step": 20815
    },
    {
      "epoch": 0.9552567573768987,
      "grad_norm": 0.4451203942298889,
      "learning_rate": 7.88140686477579e-06,
      "loss": 0.4007,
      "step": 20816
    },
    {
      "epoch": 0.9553026478821531,
      "grad_norm": 0.44466251134872437,
      "learning_rate": 7.881206481161843e-06,
      "loss": 0.3636,
      "step": 20817
    },
    {
      "epoch": 0.9553485383874076,
      "grad_norm": 0.4523621201515198,
      "learning_rate": 7.8810060906195e-06,
      "loss": 0.3544,
      "step": 20818
    },
    {
      "epoch": 0.9553944288926621,
      "grad_norm": 0.438016414642334,
      "learning_rate": 7.880805693149249e-06,
      "loss": 0.3771,
      "step": 20819
    },
    {
      "epoch": 0.9554403193979166,
      "grad_norm": 0.4882332384586334,
      "learning_rate": 7.88060528875157e-06,
      "loss": 0.4456,
      "step": 20820
    },
    {
      "epoch": 0.955486209903171,
      "grad_norm": 0.4875311255455017,
      "learning_rate": 7.880404877426943e-06,
      "loss": 0.4042,
      "step": 20821
    },
    {
      "epoch": 0.9555321004084255,
      "grad_norm": 0.46924158930778503,
      "learning_rate": 7.880204459175851e-06,
      "loss": 0.4049,
      "step": 20822
    },
    {
      "epoch": 0.95557799091368,
      "grad_norm": 0.4193507432937622,
      "learning_rate": 7.880004033998778e-06,
      "loss": 0.3298,
      "step": 20823
    },
    {
      "epoch": 0.9556238814189344,
      "grad_norm": 0.5255727171897888,
      "learning_rate": 7.879803601896203e-06,
      "loss": 0.4959,
      "step": 20824
    },
    {
      "epoch": 0.9556697719241889,
      "grad_norm": 0.44380924105644226,
      "learning_rate": 7.87960316286861e-06,
      "loss": 0.3631,
      "step": 20825
    },
    {
      "epoch": 0.9557156624294434,
      "grad_norm": 0.4121224880218506,
      "learning_rate": 7.87940271691648e-06,
      "loss": 0.3359,
      "step": 20826
    },
    {
      "epoch": 0.9557615529346978,
      "grad_norm": 0.48192572593688965,
      "learning_rate": 7.879202264040295e-06,
      "loss": 0.4581,
      "step": 20827
    },
    {
      "epoch": 0.9558074434399523,
      "grad_norm": 0.4743208587169647,
      "learning_rate": 7.879001804240538e-06,
      "loss": 0.4369,
      "step": 20828
    },
    {
      "epoch": 0.9558533339452068,
      "grad_norm": 0.48124873638153076,
      "learning_rate": 7.878801337517687e-06,
      "loss": 0.4854,
      "step": 20829
    },
    {
      "epoch": 0.9558992244504612,
      "grad_norm": 0.46403101086616516,
      "learning_rate": 7.87860086387223e-06,
      "loss": 0.3861,
      "step": 20830
    },
    {
      "epoch": 0.9559451149557157,
      "grad_norm": 0.4555166959762573,
      "learning_rate": 7.878400383304647e-06,
      "loss": 0.3918,
      "step": 20831
    },
    {
      "epoch": 0.9559910054609702,
      "grad_norm": 0.43670785427093506,
      "learning_rate": 7.878199895815419e-06,
      "loss": 0.4028,
      "step": 20832
    },
    {
      "epoch": 0.9560368959662245,
      "grad_norm": 0.4959140419960022,
      "learning_rate": 7.877999401405028e-06,
      "loss": 0.459,
      "step": 20833
    },
    {
      "epoch": 0.956082786471479,
      "grad_norm": 0.4522519111633301,
      "learning_rate": 7.877798900073958e-06,
      "loss": 0.3892,
      "step": 20834
    },
    {
      "epoch": 0.9561286769767335,
      "grad_norm": 0.43764665722846985,
      "learning_rate": 7.87759839182269e-06,
      "loss": 0.3898,
      "step": 20835
    },
    {
      "epoch": 0.9561745674819879,
      "grad_norm": 0.46248313784599304,
      "learning_rate": 7.877397876651704e-06,
      "loss": 0.4378,
      "step": 20836
    },
    {
      "epoch": 0.9562204579872424,
      "grad_norm": 0.45338696241378784,
      "learning_rate": 7.877197354561485e-06,
      "loss": 0.3937,
      "step": 20837
    },
    {
      "epoch": 0.9562663484924969,
      "grad_norm": 0.4152873754501343,
      "learning_rate": 7.876996825552515e-06,
      "loss": 0.3622,
      "step": 20838
    },
    {
      "epoch": 0.9563122389977514,
      "grad_norm": 0.4421502947807312,
      "learning_rate": 7.876796289625276e-06,
      "loss": 0.3506,
      "step": 20839
    },
    {
      "epoch": 0.9563581295030058,
      "grad_norm": 0.4926114082336426,
      "learning_rate": 7.876595746780248e-06,
      "loss": 0.4249,
      "step": 20840
    },
    {
      "epoch": 0.9564040200082603,
      "grad_norm": 0.4404122233390808,
      "learning_rate": 7.876395197017918e-06,
      "loss": 0.3656,
      "step": 20841
    },
    {
      "epoch": 0.9564499105135148,
      "grad_norm": 0.44246649742126465,
      "learning_rate": 7.876194640338764e-06,
      "loss": 0.348,
      "step": 20842
    },
    {
      "epoch": 0.9564958010187692,
      "grad_norm": 0.4617062211036682,
      "learning_rate": 7.87599407674327e-06,
      "loss": 0.3688,
      "step": 20843
    },
    {
      "epoch": 0.9565416915240237,
      "grad_norm": 0.5105952024459839,
      "learning_rate": 7.875793506231917e-06,
      "loss": 0.4995,
      "step": 20844
    },
    {
      "epoch": 0.9565875820292782,
      "grad_norm": 0.4528931677341461,
      "learning_rate": 7.875592928805189e-06,
      "loss": 0.3884,
      "step": 20845
    },
    {
      "epoch": 0.9566334725345326,
      "grad_norm": 0.43256860971450806,
      "learning_rate": 7.875392344463567e-06,
      "loss": 0.3758,
      "step": 20846
    },
    {
      "epoch": 0.9566793630397871,
      "grad_norm": 0.4371575117111206,
      "learning_rate": 7.875191753207534e-06,
      "loss": 0.3407,
      "step": 20847
    },
    {
      "epoch": 0.9567252535450416,
      "grad_norm": 0.4334792196750641,
      "learning_rate": 7.874991155037572e-06,
      "loss": 0.3409,
      "step": 20848
    },
    {
      "epoch": 0.956771144050296,
      "grad_norm": 0.48748162388801575,
      "learning_rate": 7.874790549954165e-06,
      "loss": 0.4891,
      "step": 20849
    },
    {
      "epoch": 0.9568170345555504,
      "grad_norm": 0.46133026480674744,
      "learning_rate": 7.874589937957793e-06,
      "loss": 0.4194,
      "step": 20850
    },
    {
      "epoch": 0.956862925060805,
      "grad_norm": 0.43453001976013184,
      "learning_rate": 7.874389319048938e-06,
      "loss": 0.3542,
      "step": 20851
    },
    {
      "epoch": 0.9569088155660593,
      "grad_norm": 0.4716585576534271,
      "learning_rate": 7.874188693228087e-06,
      "loss": 0.4297,
      "step": 20852
    },
    {
      "epoch": 0.9569547060713138,
      "grad_norm": 0.44578486680984497,
      "learning_rate": 7.873988060495717e-06,
      "loss": 0.3966,
      "step": 20853
    },
    {
      "epoch": 0.9570005965765683,
      "grad_norm": 0.4477722644805908,
      "learning_rate": 7.873787420852315e-06,
      "loss": 0.4122,
      "step": 20854
    },
    {
      "epoch": 0.9570464870818227,
      "grad_norm": 0.43393674492836,
      "learning_rate": 7.873586774298359e-06,
      "loss": 0.3834,
      "step": 20855
    },
    {
      "epoch": 0.9570923775870772,
      "grad_norm": 0.47393274307250977,
      "learning_rate": 7.873386120834334e-06,
      "loss": 0.3582,
      "step": 20856
    },
    {
      "epoch": 0.9571382680923317,
      "grad_norm": 0.4155539572238922,
      "learning_rate": 7.873185460460724e-06,
      "loss": 0.3437,
      "step": 20857
    },
    {
      "epoch": 0.9571841585975862,
      "grad_norm": 0.47553786635398865,
      "learning_rate": 7.872984793178007e-06,
      "loss": 0.4073,
      "step": 20858
    },
    {
      "epoch": 0.9572300491028406,
      "grad_norm": 0.4622083604335785,
      "learning_rate": 7.87278411898667e-06,
      "loss": 0.3941,
      "step": 20859
    },
    {
      "epoch": 0.9572759396080951,
      "grad_norm": 0.406045138835907,
      "learning_rate": 7.872583437887194e-06,
      "loss": 0.3177,
      "step": 20860
    },
    {
      "epoch": 0.9573218301133496,
      "grad_norm": 0.47201257944107056,
      "learning_rate": 7.87238274988006e-06,
      "loss": 0.4067,
      "step": 20861
    },
    {
      "epoch": 0.957367720618604,
      "grad_norm": 0.4298066794872284,
      "learning_rate": 7.872182054965755e-06,
      "loss": 0.3962,
      "step": 20862
    },
    {
      "epoch": 0.9574136111238585,
      "grad_norm": 0.468861848115921,
      "learning_rate": 7.871981353144756e-06,
      "loss": 0.4244,
      "step": 20863
    },
    {
      "epoch": 0.957459501629113,
      "grad_norm": 0.5078380107879639,
      "learning_rate": 7.871780644417549e-06,
      "loss": 0.4262,
      "step": 20864
    },
    {
      "epoch": 0.9575053921343674,
      "grad_norm": 0.4903586506843567,
      "learning_rate": 7.871579928784617e-06,
      "loss": 0.4362,
      "step": 20865
    },
    {
      "epoch": 0.9575512826396219,
      "grad_norm": 0.47607383131980896,
      "learning_rate": 7.87137920624644e-06,
      "loss": 0.4865,
      "step": 20866
    },
    {
      "epoch": 0.9575971731448764,
      "grad_norm": 0.4929946959018707,
      "learning_rate": 7.871178476803503e-06,
      "loss": 0.4602,
      "step": 20867
    },
    {
      "epoch": 0.9576430636501307,
      "grad_norm": 0.4569588303565979,
      "learning_rate": 7.870977740456289e-06,
      "loss": 0.3949,
      "step": 20868
    },
    {
      "epoch": 0.9576889541553852,
      "grad_norm": 0.43647006154060364,
      "learning_rate": 7.870776997205275e-06,
      "loss": 0.3603,
      "step": 20869
    },
    {
      "epoch": 0.9577348446606397,
      "grad_norm": 0.44699007272720337,
      "learning_rate": 7.870576247050954e-06,
      "loss": 0.4012,
      "step": 20870
    },
    {
      "epoch": 0.9577807351658941,
      "grad_norm": 0.45056474208831787,
      "learning_rate": 7.8703754899938e-06,
      "loss": 0.4085,
      "step": 20871
    },
    {
      "epoch": 0.9578266256711486,
      "grad_norm": 0.4675581753253937,
      "learning_rate": 7.870174726034301e-06,
      "loss": 0.4308,
      "step": 20872
    },
    {
      "epoch": 0.9578725161764031,
      "grad_norm": 0.4479956030845642,
      "learning_rate": 7.869973955172936e-06,
      "loss": 0.3491,
      "step": 20873
    },
    {
      "epoch": 0.9579184066816576,
      "grad_norm": 0.45006901025772095,
      "learning_rate": 7.86977317741019e-06,
      "loss": 0.3839,
      "step": 20874
    },
    {
      "epoch": 0.957964297186912,
      "grad_norm": 0.4226112961769104,
      "learning_rate": 7.869572392746543e-06,
      "loss": 0.3572,
      "step": 20875
    },
    {
      "epoch": 0.9580101876921665,
      "grad_norm": 0.4571353495121002,
      "learning_rate": 7.869371601182483e-06,
      "loss": 0.4363,
      "step": 20876
    },
    {
      "epoch": 0.958056078197421,
      "grad_norm": 0.4391780495643616,
      "learning_rate": 7.86917080271849e-06,
      "loss": 0.3488,
      "step": 20877
    },
    {
      "epoch": 0.9581019687026754,
      "grad_norm": 0.43637895584106445,
      "learning_rate": 7.868969997355045e-06,
      "loss": 0.4066,
      "step": 20878
    },
    {
      "epoch": 0.9581478592079299,
      "grad_norm": 0.479347288608551,
      "learning_rate": 7.868769185092634e-06,
      "loss": 0.4566,
      "step": 20879
    },
    {
      "epoch": 0.9581937497131844,
      "grad_norm": 0.4303530752658844,
      "learning_rate": 7.868568365931737e-06,
      "loss": 0.2995,
      "step": 20880
    },
    {
      "epoch": 0.9582396402184388,
      "grad_norm": 0.45962753891944885,
      "learning_rate": 7.86836753987284e-06,
      "loss": 0.4237,
      "step": 20881
    },
    {
      "epoch": 0.9582855307236933,
      "grad_norm": 0.4313022494316101,
      "learning_rate": 7.868166706916424e-06,
      "loss": 0.3594,
      "step": 20882
    },
    {
      "epoch": 0.9583314212289478,
      "grad_norm": 0.42465588450431824,
      "learning_rate": 7.867965867062972e-06,
      "loss": 0.3177,
      "step": 20883
    },
    {
      "epoch": 0.9583773117342022,
      "grad_norm": 0.4715488851070404,
      "learning_rate": 7.867765020312967e-06,
      "loss": 0.4126,
      "step": 20884
    },
    {
      "epoch": 0.9584232022394567,
      "grad_norm": 0.42767980694770813,
      "learning_rate": 7.867564166666892e-06,
      "loss": 0.3868,
      "step": 20885
    },
    {
      "epoch": 0.9584690927447111,
      "grad_norm": 0.44922980666160583,
      "learning_rate": 7.867363306125231e-06,
      "loss": 0.3999,
      "step": 20886
    },
    {
      "epoch": 0.9585149832499655,
      "grad_norm": 0.44685885310173035,
      "learning_rate": 7.867162438688464e-06,
      "loss": 0.3643,
      "step": 20887
    },
    {
      "epoch": 0.95856087375522,
      "grad_norm": 0.4425986111164093,
      "learning_rate": 7.866961564357081e-06,
      "loss": 0.4006,
      "step": 20888
    },
    {
      "epoch": 0.9586067642604745,
      "grad_norm": 0.45024463534355164,
      "learning_rate": 7.866760683131557e-06,
      "loss": 0.3878,
      "step": 20889
    },
    {
      "epoch": 0.9586526547657289,
      "grad_norm": 0.4349338710308075,
      "learning_rate": 7.866559795012378e-06,
      "loss": 0.3493,
      "step": 20890
    },
    {
      "epoch": 0.9586985452709834,
      "grad_norm": 0.42824018001556396,
      "learning_rate": 7.866358900000028e-06,
      "loss": 0.3583,
      "step": 20891
    },
    {
      "epoch": 0.9587444357762379,
      "grad_norm": 0.468671590089798,
      "learning_rate": 7.86615799809499e-06,
      "loss": 0.4025,
      "step": 20892
    },
    {
      "epoch": 0.9587903262814924,
      "grad_norm": 0.44965383410453796,
      "learning_rate": 7.865957089297747e-06,
      "loss": 0.371,
      "step": 20893
    },
    {
      "epoch": 0.9588362167867468,
      "grad_norm": 0.4732389450073242,
      "learning_rate": 7.865756173608779e-06,
      "loss": 0.4197,
      "step": 20894
    },
    {
      "epoch": 0.9588821072920013,
      "grad_norm": 0.4478726387023926,
      "learning_rate": 7.865555251028573e-06,
      "loss": 0.34,
      "step": 20895
    },
    {
      "epoch": 0.9589279977972558,
      "grad_norm": 0.41316279768943787,
      "learning_rate": 7.865354321557614e-06,
      "loss": 0.3472,
      "step": 20896
    },
    {
      "epoch": 0.9589738883025102,
      "grad_norm": 0.4628801941871643,
      "learning_rate": 7.86515338519638e-06,
      "loss": 0.4314,
      "step": 20897
    },
    {
      "epoch": 0.9590197788077647,
      "grad_norm": 0.4545949697494507,
      "learning_rate": 7.864952441945355e-06,
      "loss": 0.3971,
      "step": 20898
    },
    {
      "epoch": 0.9590656693130192,
      "grad_norm": 0.4226926565170288,
      "learning_rate": 7.864751491805026e-06,
      "loss": 0.352,
      "step": 20899
    },
    {
      "epoch": 0.9591115598182736,
      "grad_norm": 0.4720945656299591,
      "learning_rate": 7.864550534775872e-06,
      "loss": 0.476,
      "step": 20900
    },
    {
      "epoch": 0.9591574503235281,
      "grad_norm": 0.6004335284233093,
      "learning_rate": 7.864349570858378e-06,
      "loss": 0.346,
      "step": 20901
    },
    {
      "epoch": 0.9592033408287826,
      "grad_norm": 0.4516894817352295,
      "learning_rate": 7.86414860005303e-06,
      "loss": 0.4036,
      "step": 20902
    },
    {
      "epoch": 0.959249231334037,
      "grad_norm": 0.43953096866607666,
      "learning_rate": 7.863947622360307e-06,
      "loss": 0.3593,
      "step": 20903
    },
    {
      "epoch": 0.9592951218392914,
      "grad_norm": 0.45459914207458496,
      "learning_rate": 7.863746637780691e-06,
      "loss": 0.4401,
      "step": 20904
    },
    {
      "epoch": 0.9593410123445459,
      "grad_norm": 0.6540465354919434,
      "learning_rate": 7.863545646314672e-06,
      "loss": 0.4282,
      "step": 20905
    },
    {
      "epoch": 0.9593869028498003,
      "grad_norm": 0.4416063725948334,
      "learning_rate": 7.863344647962729e-06,
      "loss": 0.3214,
      "step": 20906
    },
    {
      "epoch": 0.9594327933550548,
      "grad_norm": 0.44792690873146057,
      "learning_rate": 7.863143642725343e-06,
      "loss": 0.3353,
      "step": 20907
    },
    {
      "epoch": 0.9594786838603093,
      "grad_norm": 0.4199868142604828,
      "learning_rate": 7.862942630603004e-06,
      "loss": 0.3419,
      "step": 20908
    },
    {
      "epoch": 0.9595245743655637,
      "grad_norm": 0.41641104221343994,
      "learning_rate": 7.86274161159619e-06,
      "loss": 0.3035,
      "step": 20909
    },
    {
      "epoch": 0.9595704648708182,
      "grad_norm": 0.4729260206222534,
      "learning_rate": 7.862540585705384e-06,
      "loss": 0.4102,
      "step": 20910
    },
    {
      "epoch": 0.9596163553760727,
      "grad_norm": 0.45794007182121277,
      "learning_rate": 7.862339552931073e-06,
      "loss": 0.3912,
      "step": 20911
    },
    {
      "epoch": 0.9596622458813272,
      "grad_norm": 0.4144395589828491,
      "learning_rate": 7.862138513273739e-06,
      "loss": 0.3077,
      "step": 20912
    },
    {
      "epoch": 0.9597081363865816,
      "grad_norm": 0.4522329270839691,
      "learning_rate": 7.861937466733866e-06,
      "loss": 0.3491,
      "step": 20913
    },
    {
      "epoch": 0.9597540268918361,
      "grad_norm": 0.4080905020236969,
      "learning_rate": 7.861736413311935e-06,
      "loss": 0.3622,
      "step": 20914
    },
    {
      "epoch": 0.9597999173970906,
      "grad_norm": 0.4383505582809448,
      "learning_rate": 7.861535353008431e-06,
      "loss": 0.3965,
      "step": 20915
    },
    {
      "epoch": 0.959845807902345,
      "grad_norm": 0.4504050612449646,
      "learning_rate": 7.861334285823838e-06,
      "loss": 0.3965,
      "step": 20916
    },
    {
      "epoch": 0.9598916984075995,
      "grad_norm": 0.48084089159965515,
      "learning_rate": 7.86113321175864e-06,
      "loss": 0.4578,
      "step": 20917
    },
    {
      "epoch": 0.959937588912854,
      "grad_norm": 0.49057522416114807,
      "learning_rate": 7.86093213081332e-06,
      "loss": 0.4462,
      "step": 20918
    },
    {
      "epoch": 0.9599834794181084,
      "grad_norm": 0.45509740710258484,
      "learning_rate": 7.86073104298836e-06,
      "loss": 0.4448,
      "step": 20919
    },
    {
      "epoch": 0.9600293699233629,
      "grad_norm": 0.4939304292201996,
      "learning_rate": 7.860529948284243e-06,
      "loss": 0.5069,
      "step": 20920
    },
    {
      "epoch": 0.9600752604286173,
      "grad_norm": 0.4295014441013336,
      "learning_rate": 7.860328846701457e-06,
      "loss": 0.3644,
      "step": 20921
    },
    {
      "epoch": 0.9601211509338717,
      "grad_norm": 0.4555025100708008,
      "learning_rate": 7.860127738240481e-06,
      "loss": 0.4255,
      "step": 20922
    },
    {
      "epoch": 0.9601670414391262,
      "grad_norm": 0.4333624243736267,
      "learning_rate": 7.859926622901804e-06,
      "loss": 0.3702,
      "step": 20923
    },
    {
      "epoch": 0.9602129319443807,
      "grad_norm": 0.4415133595466614,
      "learning_rate": 7.859725500685904e-06,
      "loss": 0.3791,
      "step": 20924
    },
    {
      "epoch": 0.9602588224496351,
      "grad_norm": 0.4623734951019287,
      "learning_rate": 7.859524371593266e-06,
      "loss": 0.4285,
      "step": 20925
    },
    {
      "epoch": 0.9603047129548896,
      "grad_norm": 0.5063101649284363,
      "learning_rate": 7.859323235624373e-06,
      "loss": 0.4574,
      "step": 20926
    },
    {
      "epoch": 0.9603506034601441,
      "grad_norm": 0.45520374178886414,
      "learning_rate": 7.859122092779712e-06,
      "loss": 0.4017,
      "step": 20927
    },
    {
      "epoch": 0.9603964939653986,
      "grad_norm": 0.43664202094078064,
      "learning_rate": 7.858920943059765e-06,
      "loss": 0.3731,
      "step": 20928
    },
    {
      "epoch": 0.960442384470653,
      "grad_norm": 0.45038917660713196,
      "learning_rate": 7.858719786465015e-06,
      "loss": 0.3485,
      "step": 20929
    },
    {
      "epoch": 0.9604882749759075,
      "grad_norm": 0.46419450640678406,
      "learning_rate": 7.858518622995945e-06,
      "loss": 0.3368,
      "step": 20930
    },
    {
      "epoch": 0.960534165481162,
      "grad_norm": 0.46753352880477905,
      "learning_rate": 7.858317452653042e-06,
      "loss": 0.3796,
      "step": 20931
    },
    {
      "epoch": 0.9605800559864164,
      "grad_norm": 0.4680713415145874,
      "learning_rate": 7.858116275436785e-06,
      "loss": 0.4106,
      "step": 20932
    },
    {
      "epoch": 0.9606259464916709,
      "grad_norm": 0.46388867497444153,
      "learning_rate": 7.857915091347663e-06,
      "loss": 0.3925,
      "step": 20933
    },
    {
      "epoch": 0.9606718369969254,
      "grad_norm": 0.4984924793243408,
      "learning_rate": 7.857713900386157e-06,
      "loss": 0.4396,
      "step": 20934
    },
    {
      "epoch": 0.9607177275021798,
      "grad_norm": 0.43912291526794434,
      "learning_rate": 7.85751270255275e-06,
      "loss": 0.3252,
      "step": 20935
    },
    {
      "epoch": 0.9607636180074343,
      "grad_norm": 0.48950526118278503,
      "learning_rate": 7.857311497847926e-06,
      "loss": 0.4279,
      "step": 20936
    },
    {
      "epoch": 0.9608095085126888,
      "grad_norm": 0.4625595211982727,
      "learning_rate": 7.857110286272171e-06,
      "loss": 0.3978,
      "step": 20937
    },
    {
      "epoch": 0.9608553990179431,
      "grad_norm": 0.457743763923645,
      "learning_rate": 7.856909067825967e-06,
      "loss": 0.3734,
      "step": 20938
    },
    {
      "epoch": 0.9609012895231976,
      "grad_norm": 0.44777795672416687,
      "learning_rate": 7.8567078425098e-06,
      "loss": 0.3738,
      "step": 20939
    },
    {
      "epoch": 0.9609471800284521,
      "grad_norm": 0.43666982650756836,
      "learning_rate": 7.856506610324149e-06,
      "loss": 0.3561,
      "step": 20940
    },
    {
      "epoch": 0.9609930705337065,
      "grad_norm": 0.43774130940437317,
      "learning_rate": 7.856305371269504e-06,
      "loss": 0.3882,
      "step": 20941
    },
    {
      "epoch": 0.961038961038961,
      "grad_norm": 0.4220998287200928,
      "learning_rate": 7.856104125346345e-06,
      "loss": 0.3022,
      "step": 20942
    },
    {
      "epoch": 0.9610848515442155,
      "grad_norm": 0.4717370569705963,
      "learning_rate": 7.855902872555155e-06,
      "loss": 0.3858,
      "step": 20943
    },
    {
      "epoch": 0.9611307420494699,
      "grad_norm": 0.4620283842086792,
      "learning_rate": 7.855701612896423e-06,
      "loss": 0.4077,
      "step": 20944
    },
    {
      "epoch": 0.9611766325547244,
      "grad_norm": 0.4665840268135071,
      "learning_rate": 7.855500346370628e-06,
      "loss": 0.3592,
      "step": 20945
    },
    {
      "epoch": 0.9612225230599789,
      "grad_norm": 0.4513375163078308,
      "learning_rate": 7.855299072978258e-06,
      "loss": 0.3693,
      "step": 20946
    },
    {
      "epoch": 0.9612684135652334,
      "grad_norm": 0.42944517731666565,
      "learning_rate": 7.855097792719794e-06,
      "loss": 0.3375,
      "step": 20947
    },
    {
      "epoch": 0.9613143040704878,
      "grad_norm": 0.45802319049835205,
      "learning_rate": 7.854896505595719e-06,
      "loss": 0.419,
      "step": 20948
    },
    {
      "epoch": 0.9613601945757423,
      "grad_norm": 0.44568687677383423,
      "learning_rate": 7.85469521160652e-06,
      "loss": 0.3452,
      "step": 20949
    },
    {
      "epoch": 0.9614060850809968,
      "grad_norm": 0.43978944420814514,
      "learning_rate": 7.854493910752681e-06,
      "loss": 0.3525,
      "step": 20950
    },
    {
      "epoch": 0.9614519755862512,
      "grad_norm": 0.44539856910705566,
      "learning_rate": 7.854292603034685e-06,
      "loss": 0.3518,
      "step": 20951
    },
    {
      "epoch": 0.9614978660915057,
      "grad_norm": 0.43859735131263733,
      "learning_rate": 7.854091288453015e-06,
      "loss": 0.3841,
      "step": 20952
    },
    {
      "epoch": 0.9615437565967602,
      "grad_norm": 0.4652654528617859,
      "learning_rate": 7.853889967008157e-06,
      "loss": 0.4054,
      "step": 20953
    },
    {
      "epoch": 0.9615896471020146,
      "grad_norm": 0.4313921630382538,
      "learning_rate": 7.853688638700594e-06,
      "loss": 0.369,
      "step": 20954
    },
    {
      "epoch": 0.961635537607269,
      "grad_norm": 0.4485391676425934,
      "learning_rate": 7.85348730353081e-06,
      "loss": 0.3667,
      "step": 20955
    },
    {
      "epoch": 0.9616814281125236,
      "grad_norm": 0.46054553985595703,
      "learning_rate": 7.853285961499291e-06,
      "loss": 0.4359,
      "step": 20956
    },
    {
      "epoch": 0.9617273186177779,
      "grad_norm": 0.4694073498249054,
      "learning_rate": 7.853084612606519e-06,
      "loss": 0.3659,
      "step": 20957
    },
    {
      "epoch": 0.9617732091230324,
      "grad_norm": 0.47876498103141785,
      "learning_rate": 7.85288325685298e-06,
      "loss": 0.4051,
      "step": 20958
    },
    {
      "epoch": 0.9618190996282869,
      "grad_norm": 0.44158291816711426,
      "learning_rate": 7.852681894239155e-06,
      "loss": 0.395,
      "step": 20959
    },
    {
      "epoch": 0.9618649901335413,
      "grad_norm": 0.4468650817871094,
      "learning_rate": 7.85248052476553e-06,
      "loss": 0.3955,
      "step": 20960
    },
    {
      "epoch": 0.9619108806387958,
      "grad_norm": 0.42960789799690247,
      "learning_rate": 7.852279148432594e-06,
      "loss": 0.3786,
      "step": 20961
    },
    {
      "epoch": 0.9619567711440503,
      "grad_norm": 0.4395516812801361,
      "learning_rate": 7.852077765240823e-06,
      "loss": 0.375,
      "step": 20962
    },
    {
      "epoch": 0.9620026616493048,
      "grad_norm": 0.45655351877212524,
      "learning_rate": 7.851876375190706e-06,
      "loss": 0.3913,
      "step": 20963
    },
    {
      "epoch": 0.9620485521545592,
      "grad_norm": 0.44486233592033386,
      "learning_rate": 7.851674978282725e-06,
      "loss": 0.3391,
      "step": 20964
    },
    {
      "epoch": 0.9620944426598137,
      "grad_norm": 0.44822797179222107,
      "learning_rate": 7.851473574517368e-06,
      "loss": 0.381,
      "step": 20965
    },
    {
      "epoch": 0.9621403331650682,
      "grad_norm": 0.44449833035469055,
      "learning_rate": 7.851272163895116e-06,
      "loss": 0.4055,
      "step": 20966
    },
    {
      "epoch": 0.9621862236703226,
      "grad_norm": 0.43741005659103394,
      "learning_rate": 7.851070746416454e-06,
      "loss": 0.3562,
      "step": 20967
    },
    {
      "epoch": 0.9622321141755771,
      "grad_norm": 0.4560190439224243,
      "learning_rate": 7.850869322081867e-06,
      "loss": 0.3746,
      "step": 20968
    },
    {
      "epoch": 0.9622780046808316,
      "grad_norm": 0.43159958720207214,
      "learning_rate": 7.85066789089184e-06,
      "loss": 0.3675,
      "step": 20969
    },
    {
      "epoch": 0.962323895186086,
      "grad_norm": 0.44962483644485474,
      "learning_rate": 7.850466452846855e-06,
      "loss": 0.4299,
      "step": 20970
    },
    {
      "epoch": 0.9623697856913405,
      "grad_norm": 0.3931387960910797,
      "learning_rate": 7.8502650079474e-06,
      "loss": 0.2903,
      "step": 20971
    },
    {
      "epoch": 0.962415676196595,
      "grad_norm": 0.43479329347610474,
      "learning_rate": 7.850063556193954e-06,
      "loss": 0.3648,
      "step": 20972
    },
    {
      "epoch": 0.9624615667018493,
      "grad_norm": 0.46950218081474304,
      "learning_rate": 7.849862097587007e-06,
      "loss": 0.4388,
      "step": 20973
    },
    {
      "epoch": 0.9625074572071038,
      "grad_norm": 0.44674310088157654,
      "learning_rate": 7.84966063212704e-06,
      "loss": 0.3559,
      "step": 20974
    },
    {
      "epoch": 0.9625533477123583,
      "grad_norm": 0.4506414234638214,
      "learning_rate": 7.84945915981454e-06,
      "loss": 0.3176,
      "step": 20975
    },
    {
      "epoch": 0.9625992382176127,
      "grad_norm": 0.43529245257377625,
      "learning_rate": 7.84925768064999e-06,
      "loss": 0.3358,
      "step": 20976
    },
    {
      "epoch": 0.9626451287228672,
      "grad_norm": 0.4200156629085541,
      "learning_rate": 7.849056194633872e-06,
      "loss": 0.3879,
      "step": 20977
    },
    {
      "epoch": 0.9626910192281217,
      "grad_norm": 0.47084841132164,
      "learning_rate": 7.848854701766674e-06,
      "loss": 0.4465,
      "step": 20978
    },
    {
      "epoch": 0.9627369097333761,
      "grad_norm": 0.45223331451416016,
      "learning_rate": 7.84865320204888e-06,
      "loss": 0.4225,
      "step": 20979
    },
    {
      "epoch": 0.9627828002386306,
      "grad_norm": 0.48021236062049866,
      "learning_rate": 7.848451695480974e-06,
      "loss": 0.4146,
      "step": 20980
    },
    {
      "epoch": 0.9628286907438851,
      "grad_norm": 0.4375455379486084,
      "learning_rate": 7.848250182063441e-06,
      "loss": 0.365,
      "step": 20981
    },
    {
      "epoch": 0.9628745812491396,
      "grad_norm": 0.4971928298473358,
      "learning_rate": 7.848048661796765e-06,
      "loss": 0.4547,
      "step": 20982
    },
    {
      "epoch": 0.962920471754394,
      "grad_norm": 0.46941232681274414,
      "learning_rate": 7.847847134681432e-06,
      "loss": 0.4001,
      "step": 20983
    },
    {
      "epoch": 0.9629663622596485,
      "grad_norm": 0.46193164587020874,
      "learning_rate": 7.847645600717923e-06,
      "loss": 0.4513,
      "step": 20984
    },
    {
      "epoch": 0.963012252764903,
      "grad_norm": 0.43983450531959534,
      "learning_rate": 7.847444059906728e-06,
      "loss": 0.366,
      "step": 20985
    },
    {
      "epoch": 0.9630581432701574,
      "grad_norm": 0.4574683904647827,
      "learning_rate": 7.847242512248326e-06,
      "loss": 0.3674,
      "step": 20986
    },
    {
      "epoch": 0.9631040337754119,
      "grad_norm": 0.43494752049446106,
      "learning_rate": 7.847040957743207e-06,
      "loss": 0.3434,
      "step": 20987
    },
    {
      "epoch": 0.9631499242806664,
      "grad_norm": 0.40936917066574097,
      "learning_rate": 7.846839396391852e-06,
      "loss": 0.3169,
      "step": 20988
    },
    {
      "epoch": 0.9631958147859208,
      "grad_norm": 0.43817195296287537,
      "learning_rate": 7.846637828194746e-06,
      "loss": 0.3793,
      "step": 20989
    },
    {
      "epoch": 0.9632417052911753,
      "grad_norm": 0.4614676535129547,
      "learning_rate": 7.846436253152375e-06,
      "loss": 0.4012,
      "step": 20990
    },
    {
      "epoch": 0.9632875957964298,
      "grad_norm": 0.46629729866981506,
      "learning_rate": 7.846234671265223e-06,
      "loss": 0.4019,
      "step": 20991
    },
    {
      "epoch": 0.9633334863016841,
      "grad_norm": 0.4600580632686615,
      "learning_rate": 7.846033082533775e-06,
      "loss": 0.4099,
      "step": 20992
    },
    {
      "epoch": 0.9633793768069386,
      "grad_norm": 0.4633762836456299,
      "learning_rate": 7.845831486958515e-06,
      "loss": 0.415,
      "step": 20993
    },
    {
      "epoch": 0.9634252673121931,
      "grad_norm": 0.5003665089607239,
      "learning_rate": 7.84562988453993e-06,
      "loss": 0.4958,
      "step": 20994
    },
    {
      "epoch": 0.9634711578174475,
      "grad_norm": 0.422722727060318,
      "learning_rate": 7.845428275278502e-06,
      "loss": 0.3444,
      "step": 20995
    },
    {
      "epoch": 0.963517048322702,
      "grad_norm": 0.48283904790878296,
      "learning_rate": 7.845226659174718e-06,
      "loss": 0.4693,
      "step": 20996
    },
    {
      "epoch": 0.9635629388279565,
      "grad_norm": 0.49916785955429077,
      "learning_rate": 7.84502503622906e-06,
      "loss": 0.4765,
      "step": 20997
    },
    {
      "epoch": 0.9636088293332109,
      "grad_norm": 0.44016388058662415,
      "learning_rate": 7.844823406442015e-06,
      "loss": 0.4169,
      "step": 20998
    },
    {
      "epoch": 0.9636547198384654,
      "grad_norm": 0.4360785484313965,
      "learning_rate": 7.84462176981407e-06,
      "loss": 0.3507,
      "step": 20999
    },
    {
      "epoch": 0.9637006103437199,
      "grad_norm": 0.42157822847366333,
      "learning_rate": 7.844420126345705e-06,
      "loss": 0.3367,
      "step": 21000
    },
    {
      "epoch": 0.9637465008489744,
      "grad_norm": 0.46879708766937256,
      "learning_rate": 7.844218476037408e-06,
      "loss": 0.4685,
      "step": 21001
    },
    {
      "epoch": 0.9637923913542288,
      "grad_norm": 0.41719454526901245,
      "learning_rate": 7.844016818889665e-06,
      "loss": 0.3618,
      "step": 21002
    },
    {
      "epoch": 0.9638382818594833,
      "grad_norm": 0.4576605260372162,
      "learning_rate": 7.843815154902956e-06,
      "loss": 0.413,
      "step": 21003
    },
    {
      "epoch": 0.9638841723647378,
      "grad_norm": 0.46193671226501465,
      "learning_rate": 7.843613484077772e-06,
      "loss": 0.3368,
      "step": 21004
    },
    {
      "epoch": 0.9639300628699922,
      "grad_norm": 0.5043448209762573,
      "learning_rate": 7.843411806414592e-06,
      "loss": 0.4139,
      "step": 21005
    },
    {
      "epoch": 0.9639759533752467,
      "grad_norm": 0.451667845249176,
      "learning_rate": 7.843210121913907e-06,
      "loss": 0.4207,
      "step": 21006
    },
    {
      "epoch": 0.9640218438805012,
      "grad_norm": 0.4618699252605438,
      "learning_rate": 7.843008430576198e-06,
      "loss": 0.4121,
      "step": 21007
    },
    {
      "epoch": 0.9640677343857555,
      "grad_norm": 0.4833938777446747,
      "learning_rate": 7.84280673240195e-06,
      "loss": 0.4302,
      "step": 21008
    },
    {
      "epoch": 0.96411362489101,
      "grad_norm": 0.4793543815612793,
      "learning_rate": 7.84260502739165e-06,
      "loss": 0.374,
      "step": 21009
    },
    {
      "epoch": 0.9641595153962645,
      "grad_norm": 0.44299066066741943,
      "learning_rate": 7.842403315545781e-06,
      "loss": 0.3669,
      "step": 21010
    },
    {
      "epoch": 0.9642054059015189,
      "grad_norm": 0.45445573329925537,
      "learning_rate": 7.84220159686483e-06,
      "loss": 0.3856,
      "step": 21011
    },
    {
      "epoch": 0.9642512964067734,
      "grad_norm": 0.438190758228302,
      "learning_rate": 7.841999871349282e-06,
      "loss": 0.3919,
      "step": 21012
    },
    {
      "epoch": 0.9642971869120279,
      "grad_norm": 0.4598987102508545,
      "learning_rate": 7.84179813899962e-06,
      "loss": 0.4253,
      "step": 21013
    },
    {
      "epoch": 0.9643430774172823,
      "grad_norm": 0.45900970697402954,
      "learning_rate": 7.841596399816331e-06,
      "loss": 0.4082,
      "step": 21014
    },
    {
      "epoch": 0.9643889679225368,
      "grad_norm": 0.4682950973510742,
      "learning_rate": 7.8413946537999e-06,
      "loss": 0.3842,
      "step": 21015
    },
    {
      "epoch": 0.9644348584277913,
      "grad_norm": 0.4462374150753021,
      "learning_rate": 7.84119290095081e-06,
      "loss": 0.3508,
      "step": 21016
    },
    {
      "epoch": 0.9644807489330458,
      "grad_norm": 0.4466807246208191,
      "learning_rate": 7.840991141269548e-06,
      "loss": 0.3515,
      "step": 21017
    },
    {
      "epoch": 0.9645266394383002,
      "grad_norm": 0.4427233040332794,
      "learning_rate": 7.840789374756601e-06,
      "loss": 0.3656,
      "step": 21018
    },
    {
      "epoch": 0.9645725299435547,
      "grad_norm": 0.44723081588745117,
      "learning_rate": 7.840587601412451e-06,
      "loss": 0.3239,
      "step": 21019
    },
    {
      "epoch": 0.9646184204488092,
      "grad_norm": 0.46869444847106934,
      "learning_rate": 7.840385821237583e-06,
      "loss": 0.4478,
      "step": 21020
    },
    {
      "epoch": 0.9646643109540636,
      "grad_norm": 0.46258923411369324,
      "learning_rate": 7.840184034232485e-06,
      "loss": 0.4799,
      "step": 21021
    },
    {
      "epoch": 0.9647102014593181,
      "grad_norm": 0.46032845973968506,
      "learning_rate": 7.83998224039764e-06,
      "loss": 0.3959,
      "step": 21022
    },
    {
      "epoch": 0.9647560919645726,
      "grad_norm": 0.43300849199295044,
      "learning_rate": 7.839780439733535e-06,
      "loss": 0.3844,
      "step": 21023
    },
    {
      "epoch": 0.964801982469827,
      "grad_norm": 0.46978551149368286,
      "learning_rate": 7.839578632240654e-06,
      "loss": 0.437,
      "step": 21024
    },
    {
      "epoch": 0.9648478729750815,
      "grad_norm": 0.4499961733818054,
      "learning_rate": 7.839376817919482e-06,
      "loss": 0.4128,
      "step": 21025
    },
    {
      "epoch": 0.964893763480336,
      "grad_norm": 0.5136227607727051,
      "learning_rate": 7.839174996770505e-06,
      "loss": 0.5116,
      "step": 21026
    },
    {
      "epoch": 0.9649396539855903,
      "grad_norm": 0.4140956997871399,
      "learning_rate": 7.838973168794208e-06,
      "loss": 0.3553,
      "step": 21027
    },
    {
      "epoch": 0.9649855444908448,
      "grad_norm": 0.47263965010643005,
      "learning_rate": 7.838771333991077e-06,
      "loss": 0.4575,
      "step": 21028
    },
    {
      "epoch": 0.9650314349960993,
      "grad_norm": 0.4347728192806244,
      "learning_rate": 7.838569492361597e-06,
      "loss": 0.4071,
      "step": 21029
    },
    {
      "epoch": 0.9650773255013537,
      "grad_norm": 0.4320104718208313,
      "learning_rate": 7.838367643906253e-06,
      "loss": 0.3186,
      "step": 21030
    },
    {
      "epoch": 0.9651232160066082,
      "grad_norm": 0.42781323194503784,
      "learning_rate": 7.838165788625531e-06,
      "loss": 0.3421,
      "step": 21031
    },
    {
      "epoch": 0.9651691065118627,
      "grad_norm": 0.45310214161872864,
      "learning_rate": 7.837963926519916e-06,
      "loss": 0.3606,
      "step": 21032
    },
    {
      "epoch": 0.9652149970171171,
      "grad_norm": 0.44083476066589355,
      "learning_rate": 7.837762057589893e-06,
      "loss": 0.3582,
      "step": 21033
    },
    {
      "epoch": 0.9652608875223716,
      "grad_norm": 0.4155220687389374,
      "learning_rate": 7.837560181835947e-06,
      "loss": 0.3218,
      "step": 21034
    },
    {
      "epoch": 0.9653067780276261,
      "grad_norm": 0.4668925106525421,
      "learning_rate": 7.837358299258565e-06,
      "loss": 0.3452,
      "step": 21035
    },
    {
      "epoch": 0.9653526685328806,
      "grad_norm": 0.4866597056388855,
      "learning_rate": 7.837156409858233e-06,
      "loss": 0.4141,
      "step": 21036
    },
    {
      "epoch": 0.965398559038135,
      "grad_norm": 0.4267602264881134,
      "learning_rate": 7.836954513635435e-06,
      "loss": 0.3619,
      "step": 21037
    },
    {
      "epoch": 0.9654444495433895,
      "grad_norm": 0.4759226441383362,
      "learning_rate": 7.836752610590657e-06,
      "loss": 0.4767,
      "step": 21038
    },
    {
      "epoch": 0.965490340048644,
      "grad_norm": 0.42197442054748535,
      "learning_rate": 7.836550700724383e-06,
      "loss": 0.3296,
      "step": 21039
    },
    {
      "epoch": 0.9655362305538984,
      "grad_norm": 0.44642555713653564,
      "learning_rate": 7.8363487840371e-06,
      "loss": 0.4062,
      "step": 21040
    },
    {
      "epoch": 0.9655821210591529,
      "grad_norm": 0.571806013584137,
      "learning_rate": 7.836146860529293e-06,
      "loss": 0.4227,
      "step": 21041
    },
    {
      "epoch": 0.9656280115644074,
      "grad_norm": 0.4606640934944153,
      "learning_rate": 7.83594493020145e-06,
      "loss": 0.3871,
      "step": 21042
    },
    {
      "epoch": 0.9656739020696617,
      "grad_norm": 0.4591871201992035,
      "learning_rate": 7.835742993054051e-06,
      "loss": 0.3578,
      "step": 21043
    },
    {
      "epoch": 0.9657197925749162,
      "grad_norm": 0.4183909296989441,
      "learning_rate": 7.835541049087589e-06,
      "loss": 0.3201,
      "step": 21044
    },
    {
      "epoch": 0.9657656830801707,
      "grad_norm": 0.49177372455596924,
      "learning_rate": 7.835339098302544e-06,
      "loss": 0.4626,
      "step": 21045
    },
    {
      "epoch": 0.9658115735854251,
      "grad_norm": 0.5221524238586426,
      "learning_rate": 7.835137140699403e-06,
      "loss": 0.4181,
      "step": 21046
    },
    {
      "epoch": 0.9658574640906796,
      "grad_norm": 0.45898792147636414,
      "learning_rate": 7.83493517627865e-06,
      "loss": 0.394,
      "step": 21047
    },
    {
      "epoch": 0.9659033545959341,
      "grad_norm": 0.4413659870624542,
      "learning_rate": 7.834733205040776e-06,
      "loss": 0.3642,
      "step": 21048
    },
    {
      "epoch": 0.9659492451011885,
      "grad_norm": 0.4048059284687042,
      "learning_rate": 7.834531226986262e-06,
      "loss": 0.3191,
      "step": 21049
    },
    {
      "epoch": 0.965995135606443,
      "grad_norm": 0.486472487449646,
      "learning_rate": 7.834329242115594e-06,
      "loss": 0.4802,
      "step": 21050
    },
    {
      "epoch": 0.9660410261116975,
      "grad_norm": 0.4253324270248413,
      "learning_rate": 7.83412725042926e-06,
      "loss": 0.3773,
      "step": 21051
    },
    {
      "epoch": 0.966086916616952,
      "grad_norm": 0.45402684807777405,
      "learning_rate": 7.833925251927746e-06,
      "loss": 0.3646,
      "step": 21052
    },
    {
      "epoch": 0.9661328071222064,
      "grad_norm": 0.46813470125198364,
      "learning_rate": 7.833723246611534e-06,
      "loss": 0.4363,
      "step": 21053
    },
    {
      "epoch": 0.9661786976274609,
      "grad_norm": 0.48408305644989014,
      "learning_rate": 7.833521234481112e-06,
      "loss": 0.3983,
      "step": 21054
    },
    {
      "epoch": 0.9662245881327154,
      "grad_norm": 0.42531606554985046,
      "learning_rate": 7.833319215536965e-06,
      "loss": 0.3695,
      "step": 21055
    },
    {
      "epoch": 0.9662704786379698,
      "grad_norm": 0.4530765414237976,
      "learning_rate": 7.833117189779583e-06,
      "loss": 0.4031,
      "step": 21056
    },
    {
      "epoch": 0.9663163691432243,
      "grad_norm": 0.4948331117630005,
      "learning_rate": 7.832915157209444e-06,
      "loss": 0.5189,
      "step": 21057
    },
    {
      "epoch": 0.9663622596484788,
      "grad_norm": 0.4198783040046692,
      "learning_rate": 7.832713117827041e-06,
      "loss": 0.3443,
      "step": 21058
    },
    {
      "epoch": 0.9664081501537332,
      "grad_norm": 0.4507180154323578,
      "learning_rate": 7.832511071632856e-06,
      "loss": 0.4008,
      "step": 21059
    },
    {
      "epoch": 0.9664540406589877,
      "grad_norm": 0.47342008352279663,
      "learning_rate": 7.832309018627375e-06,
      "loss": 0.4691,
      "step": 21060
    },
    {
      "epoch": 0.9664999311642422,
      "grad_norm": 0.4678534269332886,
      "learning_rate": 7.832106958811087e-06,
      "loss": 0.392,
      "step": 21061
    },
    {
      "epoch": 0.9665458216694965,
      "grad_norm": 0.4501871168613434,
      "learning_rate": 7.831904892184473e-06,
      "loss": 0.417,
      "step": 21062
    },
    {
      "epoch": 0.966591712174751,
      "grad_norm": 0.4435878396034241,
      "learning_rate": 7.831702818748024e-06,
      "loss": 0.3537,
      "step": 21063
    },
    {
      "epoch": 0.9666376026800055,
      "grad_norm": 0.42602279782295227,
      "learning_rate": 7.83150073850222e-06,
      "loss": 0.375,
      "step": 21064
    },
    {
      "epoch": 0.9666834931852599,
      "grad_norm": 0.44667181372642517,
      "learning_rate": 7.831298651447553e-06,
      "loss": 0.4209,
      "step": 21065
    },
    {
      "epoch": 0.9667293836905144,
      "grad_norm": 0.427642822265625,
      "learning_rate": 7.831096557584506e-06,
      "loss": 0.3737,
      "step": 21066
    },
    {
      "epoch": 0.9667752741957689,
      "grad_norm": 0.4266265332698822,
      "learning_rate": 7.830894456913565e-06,
      "loss": 0.3259,
      "step": 21067
    },
    {
      "epoch": 0.9668211647010233,
      "grad_norm": 0.4106382429599762,
      "learning_rate": 7.830692349435217e-06,
      "loss": 0.3306,
      "step": 21068
    },
    {
      "epoch": 0.9668670552062778,
      "grad_norm": 0.43538612127304077,
      "learning_rate": 7.830490235149946e-06,
      "loss": 0.3844,
      "step": 21069
    },
    {
      "epoch": 0.9669129457115323,
      "grad_norm": 0.504745364189148,
      "learning_rate": 7.83028811405824e-06,
      "loss": 0.5082,
      "step": 21070
    },
    {
      "epoch": 0.9669588362167868,
      "grad_norm": 0.438167542219162,
      "learning_rate": 7.830085986160584e-06,
      "loss": 0.3666,
      "step": 21071
    },
    {
      "epoch": 0.9670047267220412,
      "grad_norm": 0.47544726729393005,
      "learning_rate": 7.829883851457466e-06,
      "loss": 0.4385,
      "step": 21072
    },
    {
      "epoch": 0.9670506172272957,
      "grad_norm": 0.47371727228164673,
      "learning_rate": 7.829681709949367e-06,
      "loss": 0.4712,
      "step": 21073
    },
    {
      "epoch": 0.9670965077325502,
      "grad_norm": 0.43370920419692993,
      "learning_rate": 7.82947956163678e-06,
      "loss": 0.3615,
      "step": 21074
    },
    {
      "epoch": 0.9671423982378046,
      "grad_norm": 0.410230815410614,
      "learning_rate": 7.829277406520187e-06,
      "loss": 0.3019,
      "step": 21075
    },
    {
      "epoch": 0.9671882887430591,
      "grad_norm": 0.5153003334999084,
      "learning_rate": 7.829075244600073e-06,
      "loss": 0.4857,
      "step": 21076
    },
    {
      "epoch": 0.9672341792483136,
      "grad_norm": 0.48740634322166443,
      "learning_rate": 7.828873075876929e-06,
      "loss": 0.4853,
      "step": 21077
    },
    {
      "epoch": 0.967280069753568,
      "grad_norm": 0.4406638443470001,
      "learning_rate": 7.828670900351237e-06,
      "loss": 0.3795,
      "step": 21078
    },
    {
      "epoch": 0.9673259602588224,
      "grad_norm": 0.45234760642051697,
      "learning_rate": 7.828468718023482e-06,
      "loss": 0.4253,
      "step": 21079
    },
    {
      "epoch": 0.967371850764077,
      "grad_norm": 0.44228604435920715,
      "learning_rate": 7.828266528894155e-06,
      "loss": 0.374,
      "step": 21080
    },
    {
      "epoch": 0.9674177412693313,
      "grad_norm": 0.44766926765441895,
      "learning_rate": 7.828064332963737e-06,
      "loss": 0.3742,
      "step": 21081
    },
    {
      "epoch": 0.9674636317745858,
      "grad_norm": 0.46373316645622253,
      "learning_rate": 7.82786213023272e-06,
      "loss": 0.4184,
      "step": 21082
    },
    {
      "epoch": 0.9675095222798403,
      "grad_norm": 0.442690908908844,
      "learning_rate": 7.827659920701585e-06,
      "loss": 0.4045,
      "step": 21083
    },
    {
      "epoch": 0.9675554127850947,
      "grad_norm": 0.44209998846054077,
      "learning_rate": 7.82745770437082e-06,
      "loss": 0.3812,
      "step": 21084
    },
    {
      "epoch": 0.9676013032903492,
      "grad_norm": 0.47506585717201233,
      "learning_rate": 7.827255481240912e-06,
      "loss": 0.4477,
      "step": 21085
    },
    {
      "epoch": 0.9676471937956037,
      "grad_norm": 0.4498728811740875,
      "learning_rate": 7.827053251312348e-06,
      "loss": 0.348,
      "step": 21086
    },
    {
      "epoch": 0.9676930843008581,
      "grad_norm": 0.43462246656417847,
      "learning_rate": 7.826851014585613e-06,
      "loss": 0.3296,
      "step": 21087
    },
    {
      "epoch": 0.9677389748061126,
      "grad_norm": 0.4444442391395569,
      "learning_rate": 7.826648771061192e-06,
      "loss": 0.4297,
      "step": 21088
    },
    {
      "epoch": 0.9677848653113671,
      "grad_norm": 0.4629170596599579,
      "learning_rate": 7.826446520739575e-06,
      "loss": 0.3847,
      "step": 21089
    },
    {
      "epoch": 0.9678307558166216,
      "grad_norm": 0.47325995564460754,
      "learning_rate": 7.826244263621243e-06,
      "loss": 0.376,
      "step": 21090
    },
    {
      "epoch": 0.967876646321876,
      "grad_norm": 0.3935380280017853,
      "learning_rate": 7.826041999706689e-06,
      "loss": 0.2765,
      "step": 21091
    },
    {
      "epoch": 0.9679225368271305,
      "grad_norm": 0.4389955401420593,
      "learning_rate": 7.825839728996394e-06,
      "loss": 0.3521,
      "step": 21092
    },
    {
      "epoch": 0.967968427332385,
      "grad_norm": 0.4877501130104065,
      "learning_rate": 7.825637451490846e-06,
      "loss": 0.4147,
      "step": 21093
    },
    {
      "epoch": 0.9680143178376394,
      "grad_norm": 0.47158002853393555,
      "learning_rate": 7.825435167190533e-06,
      "loss": 0.4302,
      "step": 21094
    },
    {
      "epoch": 0.9680602083428939,
      "grad_norm": 0.4394701421260834,
      "learning_rate": 7.825232876095939e-06,
      "loss": 0.4122,
      "step": 21095
    },
    {
      "epoch": 0.9681060988481484,
      "grad_norm": 0.4515072703361511,
      "learning_rate": 7.825030578207552e-06,
      "loss": 0.3639,
      "step": 21096
    },
    {
      "epoch": 0.9681519893534027,
      "grad_norm": 0.4458642899990082,
      "learning_rate": 7.824828273525858e-06,
      "loss": 0.4163,
      "step": 21097
    },
    {
      "epoch": 0.9681978798586572,
      "grad_norm": 0.44289305806159973,
      "learning_rate": 7.824625962051344e-06,
      "loss": 0.3678,
      "step": 21098
    },
    {
      "epoch": 0.9682437703639117,
      "grad_norm": 0.5296807289123535,
      "learning_rate": 7.824423643784495e-06,
      "loss": 0.5281,
      "step": 21099
    },
    {
      "epoch": 0.9682896608691661,
      "grad_norm": 0.45049849152565,
      "learning_rate": 7.8242213187258e-06,
      "loss": 0.375,
      "step": 21100
    },
    {
      "epoch": 0.9683355513744206,
      "grad_norm": 0.43462711572647095,
      "learning_rate": 7.824018986875743e-06,
      "loss": 0.3342,
      "step": 21101
    },
    {
      "epoch": 0.9683814418796751,
      "grad_norm": 0.46062541007995605,
      "learning_rate": 7.823816648234813e-06,
      "loss": 0.3784,
      "step": 21102
    },
    {
      "epoch": 0.9684273323849295,
      "grad_norm": 0.453469455242157,
      "learning_rate": 7.823614302803493e-06,
      "loss": 0.4211,
      "step": 21103
    },
    {
      "epoch": 0.968473222890184,
      "grad_norm": 0.43442052602767944,
      "learning_rate": 7.823411950582273e-06,
      "loss": 0.3566,
      "step": 21104
    },
    {
      "epoch": 0.9685191133954385,
      "grad_norm": 0.4289034605026245,
      "learning_rate": 7.823209591571638e-06,
      "loss": 0.3398,
      "step": 21105
    },
    {
      "epoch": 0.968565003900693,
      "grad_norm": 0.43402358889579773,
      "learning_rate": 7.823007225772077e-06,
      "loss": 0.3721,
      "step": 21106
    },
    {
      "epoch": 0.9686108944059474,
      "grad_norm": 0.4767535328865051,
      "learning_rate": 7.822804853184072e-06,
      "loss": 0.4797,
      "step": 21107
    },
    {
      "epoch": 0.9686567849112019,
      "grad_norm": 0.4255218505859375,
      "learning_rate": 7.822602473808114e-06,
      "loss": 0.3872,
      "step": 21108
    },
    {
      "epoch": 0.9687026754164564,
      "grad_norm": 0.4691382348537445,
      "learning_rate": 7.822400087644686e-06,
      "loss": 0.3923,
      "step": 21109
    },
    {
      "epoch": 0.9687485659217108,
      "grad_norm": 0.42823004722595215,
      "learning_rate": 7.822197694694278e-06,
      "loss": 0.3972,
      "step": 21110
    },
    {
      "epoch": 0.9687944564269653,
      "grad_norm": 0.4656471014022827,
      "learning_rate": 7.821995294957374e-06,
      "loss": 0.3782,
      "step": 21111
    },
    {
      "epoch": 0.9688403469322198,
      "grad_norm": 0.4245080351829529,
      "learning_rate": 7.821792888434463e-06,
      "loss": 0.3502,
      "step": 21112
    },
    {
      "epoch": 0.9688862374374742,
      "grad_norm": 0.41375812888145447,
      "learning_rate": 7.821590475126031e-06,
      "loss": 0.297,
      "step": 21113
    },
    {
      "epoch": 0.9689321279427286,
      "grad_norm": 0.5158706903457642,
      "learning_rate": 7.821388055032564e-06,
      "loss": 0.4769,
      "step": 21114
    },
    {
      "epoch": 0.9689780184479831,
      "grad_norm": 0.4364645779132843,
      "learning_rate": 7.821185628154548e-06,
      "loss": 0.3315,
      "step": 21115
    },
    {
      "epoch": 0.9690239089532375,
      "grad_norm": 0.6473841667175293,
      "learning_rate": 7.820983194492473e-06,
      "loss": 0.35,
      "step": 21116
    },
    {
      "epoch": 0.969069799458492,
      "grad_norm": 0.4310930371284485,
      "learning_rate": 7.820780754046822e-06,
      "loss": 0.3448,
      "step": 21117
    },
    {
      "epoch": 0.9691156899637465,
      "grad_norm": 0.4209231734275818,
      "learning_rate": 7.820578306818086e-06,
      "loss": 0.3445,
      "step": 21118
    },
    {
      "epoch": 0.9691615804690009,
      "grad_norm": 0.44525182247161865,
      "learning_rate": 7.820375852806748e-06,
      "loss": 0.3936,
      "step": 21119
    },
    {
      "epoch": 0.9692074709742554,
      "grad_norm": 0.4654008150100708,
      "learning_rate": 7.820173392013295e-06,
      "loss": 0.4431,
      "step": 21120
    },
    {
      "epoch": 0.9692533614795099,
      "grad_norm": 0.4261218309402466,
      "learning_rate": 7.819970924438216e-06,
      "loss": 0.3627,
      "step": 21121
    },
    {
      "epoch": 0.9692992519847643,
      "grad_norm": 0.4572601318359375,
      "learning_rate": 7.819768450081997e-06,
      "loss": 0.3545,
      "step": 21122
    },
    {
      "epoch": 0.9693451424900188,
      "grad_norm": 0.43420615792274475,
      "learning_rate": 7.819565968945123e-06,
      "loss": 0.3455,
      "step": 21123
    },
    {
      "epoch": 0.9693910329952733,
      "grad_norm": 0.4477049708366394,
      "learning_rate": 7.819363481028087e-06,
      "loss": 0.3986,
      "step": 21124
    },
    {
      "epoch": 0.9694369235005278,
      "grad_norm": 0.4257064461708069,
      "learning_rate": 7.819160986331368e-06,
      "loss": 0.3416,
      "step": 21125
    },
    {
      "epoch": 0.9694828140057822,
      "grad_norm": 0.4889322519302368,
      "learning_rate": 7.818958484855456e-06,
      "loss": 0.4409,
      "step": 21126
    },
    {
      "epoch": 0.9695287045110367,
      "grad_norm": 0.45275017619132996,
      "learning_rate": 7.818755976600842e-06,
      "loss": 0.3871,
      "step": 21127
    },
    {
      "epoch": 0.9695745950162912,
      "grad_norm": 0.4485742747783661,
      "learning_rate": 7.818553461568009e-06,
      "loss": 0.4102,
      "step": 21128
    },
    {
      "epoch": 0.9696204855215456,
      "grad_norm": 0.4430753290653229,
      "learning_rate": 7.818350939757442e-06,
      "loss": 0.4007,
      "step": 21129
    },
    {
      "epoch": 0.9696663760268001,
      "grad_norm": 0.4442596137523651,
      "learning_rate": 7.81814841116963e-06,
      "loss": 0.3871,
      "step": 21130
    },
    {
      "epoch": 0.9697122665320546,
      "grad_norm": 0.4851895868778229,
      "learning_rate": 7.817945875805062e-06,
      "loss": 0.4121,
      "step": 21131
    },
    {
      "epoch": 0.9697581570373089,
      "grad_norm": 0.44421815872192383,
      "learning_rate": 7.817743333664226e-06,
      "loss": 0.3984,
      "step": 21132
    },
    {
      "epoch": 0.9698040475425634,
      "grad_norm": 0.4512902498245239,
      "learning_rate": 7.817540784747603e-06,
      "loss": 0.4218,
      "step": 21133
    },
    {
      "epoch": 0.9698499380478179,
      "grad_norm": 0.4349035918712616,
      "learning_rate": 7.817338229055686e-06,
      "loss": 0.3482,
      "step": 21134
    },
    {
      "epoch": 0.9698958285530723,
      "grad_norm": 0.44838371872901917,
      "learning_rate": 7.817135666588957e-06,
      "loss": 0.3546,
      "step": 21135
    },
    {
      "epoch": 0.9699417190583268,
      "grad_norm": 0.47474604845046997,
      "learning_rate": 7.81693309734791e-06,
      "loss": 0.4221,
      "step": 21136
    },
    {
      "epoch": 0.9699876095635813,
      "grad_norm": 0.45858582854270935,
      "learning_rate": 7.816730521333024e-06,
      "loss": 0.3988,
      "step": 21137
    },
    {
      "epoch": 0.9700335000688357,
      "grad_norm": 0.4085386395454407,
      "learning_rate": 7.81652793854479e-06,
      "loss": 0.3196,
      "step": 21138
    },
    {
      "epoch": 0.9700793905740902,
      "grad_norm": 0.5764844417572021,
      "learning_rate": 7.816325348983698e-06,
      "loss": 0.3993,
      "step": 21139
    },
    {
      "epoch": 0.9701252810793447,
      "grad_norm": 0.4346778988838196,
      "learning_rate": 7.816122752650232e-06,
      "loss": 0.3432,
      "step": 21140
    },
    {
      "epoch": 0.9701711715845992,
      "grad_norm": 0.5452931523323059,
      "learning_rate": 7.815920149544877e-06,
      "loss": 0.4214,
      "step": 21141
    },
    {
      "epoch": 0.9702170620898536,
      "grad_norm": 0.4635372757911682,
      "learning_rate": 7.815717539668127e-06,
      "loss": 0.4394,
      "step": 21142
    },
    {
      "epoch": 0.9702629525951081,
      "grad_norm": 0.47558003664016724,
      "learning_rate": 7.815514923020463e-06,
      "loss": 0.4838,
      "step": 21143
    },
    {
      "epoch": 0.9703088431003626,
      "grad_norm": 0.4808276295661926,
      "learning_rate": 7.815312299602371e-06,
      "loss": 0.4896,
      "step": 21144
    },
    {
      "epoch": 0.970354733605617,
      "grad_norm": 0.418119341135025,
      "learning_rate": 7.815109669414345e-06,
      "loss": 0.3234,
      "step": 21145
    },
    {
      "epoch": 0.9704006241108715,
      "grad_norm": 0.4798365831375122,
      "learning_rate": 7.814907032456869e-06,
      "loss": 0.4698,
      "step": 21146
    },
    {
      "epoch": 0.970446514616126,
      "grad_norm": 0.4244768023490906,
      "learning_rate": 7.814704388730427e-06,
      "loss": 0.3539,
      "step": 21147
    },
    {
      "epoch": 0.9704924051213804,
      "grad_norm": 0.43037888407707214,
      "learning_rate": 7.814501738235513e-06,
      "loss": 0.3373,
      "step": 21148
    },
    {
      "epoch": 0.9705382956266349,
      "grad_norm": 0.4320886731147766,
      "learning_rate": 7.814299080972609e-06,
      "loss": 0.3163,
      "step": 21149
    },
    {
      "epoch": 0.9705841861318893,
      "grad_norm": 0.43833795189857483,
      "learning_rate": 7.814096416942203e-06,
      "loss": 0.3364,
      "step": 21150
    },
    {
      "epoch": 0.9706300766371437,
      "grad_norm": 0.4364391267299652,
      "learning_rate": 7.813893746144783e-06,
      "loss": 0.376,
      "step": 21151
    },
    {
      "epoch": 0.9706759671423982,
      "grad_norm": 0.4638986885547638,
      "learning_rate": 7.813691068580837e-06,
      "loss": 0.3938,
      "step": 21152
    },
    {
      "epoch": 0.9707218576476527,
      "grad_norm": 0.4367411434650421,
      "learning_rate": 7.813488384250853e-06,
      "loss": 0.3725,
      "step": 21153
    },
    {
      "epoch": 0.9707677481529071,
      "grad_norm": 0.5896845459938049,
      "learning_rate": 7.813285693155319e-06,
      "loss": 0.3863,
      "step": 21154
    },
    {
      "epoch": 0.9708136386581616,
      "grad_norm": 0.4394785165786743,
      "learning_rate": 7.813082995294718e-06,
      "loss": 0.386,
      "step": 21155
    },
    {
      "epoch": 0.9708595291634161,
      "grad_norm": 0.47244152426719666,
      "learning_rate": 7.812880290669542e-06,
      "loss": 0.4755,
      "step": 21156
    },
    {
      "epoch": 0.9709054196686705,
      "grad_norm": 0.4761359989643097,
      "learning_rate": 7.812677579280274e-06,
      "loss": 0.394,
      "step": 21157
    },
    {
      "epoch": 0.970951310173925,
      "grad_norm": 0.5197685956954956,
      "learning_rate": 7.812474861127405e-06,
      "loss": 0.5072,
      "step": 21158
    },
    {
      "epoch": 0.9709972006791795,
      "grad_norm": 0.45970508456230164,
      "learning_rate": 7.812272136211426e-06,
      "loss": 0.3974,
      "step": 21159
    },
    {
      "epoch": 0.971043091184434,
      "grad_norm": 0.487955778837204,
      "learning_rate": 7.812069404532815e-06,
      "loss": 0.4804,
      "step": 21160
    },
    {
      "epoch": 0.9710889816896884,
      "grad_norm": 0.4408271610736847,
      "learning_rate": 7.811866666092065e-06,
      "loss": 0.3387,
      "step": 21161
    },
    {
      "epoch": 0.9711348721949429,
      "grad_norm": 0.4879717528820038,
      "learning_rate": 7.811663920889665e-06,
      "loss": 0.5086,
      "step": 21162
    },
    {
      "epoch": 0.9711807627001974,
      "grad_norm": 0.44969722628593445,
      "learning_rate": 7.8114611689261e-06,
      "loss": 0.3655,
      "step": 21163
    },
    {
      "epoch": 0.9712266532054518,
      "grad_norm": 0.46268007159233093,
      "learning_rate": 7.811258410201859e-06,
      "loss": 0.3541,
      "step": 21164
    },
    {
      "epoch": 0.9712725437107063,
      "grad_norm": 0.39891159534454346,
      "learning_rate": 7.811055644717426e-06,
      "loss": 0.2671,
      "step": 21165
    },
    {
      "epoch": 0.9713184342159608,
      "grad_norm": 0.5018169283866882,
      "learning_rate": 7.810852872473294e-06,
      "loss": 0.4653,
      "step": 21166
    },
    {
      "epoch": 0.9713643247212151,
      "grad_norm": 0.4044756293296814,
      "learning_rate": 7.810650093469948e-06,
      "loss": 0.291,
      "step": 21167
    },
    {
      "epoch": 0.9714102152264696,
      "grad_norm": 0.4241901934146881,
      "learning_rate": 7.810447307707875e-06,
      "loss": 0.2995,
      "step": 21168
    },
    {
      "epoch": 0.9714561057317241,
      "grad_norm": 0.45387449860572815,
      "learning_rate": 7.810244515187564e-06,
      "loss": 0.4064,
      "step": 21169
    },
    {
      "epoch": 0.9715019962369785,
      "grad_norm": 0.44848009943962097,
      "learning_rate": 7.8100417159095e-06,
      "loss": 0.4305,
      "step": 21170
    },
    {
      "epoch": 0.971547886742233,
      "grad_norm": 0.43236395716667175,
      "learning_rate": 7.809838909874174e-06,
      "loss": 0.3945,
      "step": 21171
    },
    {
      "epoch": 0.9715937772474875,
      "grad_norm": 0.47296348214149475,
      "learning_rate": 7.809636097082072e-06,
      "loss": 0.3897,
      "step": 21172
    },
    {
      "epoch": 0.9716396677527419,
      "grad_norm": 0.45015496015548706,
      "learning_rate": 7.809433277533681e-06,
      "loss": 0.3663,
      "step": 21173
    },
    {
      "epoch": 0.9716855582579964,
      "grad_norm": 0.4243199825286865,
      "learning_rate": 7.809230451229491e-06,
      "loss": 0.3466,
      "step": 21174
    },
    {
      "epoch": 0.9717314487632509,
      "grad_norm": 0.46948209404945374,
      "learning_rate": 7.809027618169986e-06,
      "loss": 0.4214,
      "step": 21175
    },
    {
      "epoch": 0.9717773392685053,
      "grad_norm": 0.421957790851593,
      "learning_rate": 7.80882477835566e-06,
      "loss": 0.3309,
      "step": 21176
    },
    {
      "epoch": 0.9718232297737598,
      "grad_norm": 0.42971280217170715,
      "learning_rate": 7.808621931786992e-06,
      "loss": 0.352,
      "step": 21177
    },
    {
      "epoch": 0.9718691202790143,
      "grad_norm": 0.4489957094192505,
      "learning_rate": 7.808419078464479e-06,
      "loss": 0.4202,
      "step": 21178
    },
    {
      "epoch": 0.9719150107842688,
      "grad_norm": 0.42000627517700195,
      "learning_rate": 7.808216218388602e-06,
      "loss": 0.3301,
      "step": 21179
    },
    {
      "epoch": 0.9719609012895232,
      "grad_norm": 0.43955910205841064,
      "learning_rate": 7.808013351559853e-06,
      "loss": 0.3903,
      "step": 21180
    },
    {
      "epoch": 0.9720067917947777,
      "grad_norm": 0.4894885718822479,
      "learning_rate": 7.807810477978716e-06,
      "loss": 0.4783,
      "step": 21181
    },
    {
      "epoch": 0.9720526823000322,
      "grad_norm": 0.4848614037036896,
      "learning_rate": 7.807607597645684e-06,
      "loss": 0.3969,
      "step": 21182
    },
    {
      "epoch": 0.9720985728052866,
      "grad_norm": 0.4755922853946686,
      "learning_rate": 7.80740471056124e-06,
      "loss": 0.333,
      "step": 21183
    },
    {
      "epoch": 0.972144463310541,
      "grad_norm": 0.4566933214664459,
      "learning_rate": 7.807201816725873e-06,
      "loss": 0.3954,
      "step": 21184
    },
    {
      "epoch": 0.9721903538157955,
      "grad_norm": 0.41520270705223083,
      "learning_rate": 7.806998916140075e-06,
      "loss": 0.3111,
      "step": 21185
    },
    {
      "epoch": 0.9722362443210499,
      "grad_norm": 0.45734426379203796,
      "learning_rate": 7.806796008804327e-06,
      "loss": 0.3766,
      "step": 21186
    },
    {
      "epoch": 0.9722821348263044,
      "grad_norm": 0.4437125623226166,
      "learning_rate": 7.806593094719122e-06,
      "loss": 0.3519,
      "step": 21187
    },
    {
      "epoch": 0.9723280253315589,
      "grad_norm": 0.4472242295742035,
      "learning_rate": 7.806390173884944e-06,
      "loss": 0.3862,
      "step": 21188
    },
    {
      "epoch": 0.9723739158368133,
      "grad_norm": 0.4509631395339966,
      "learning_rate": 7.806187246302286e-06,
      "loss": 0.3667,
      "step": 21189
    },
    {
      "epoch": 0.9724198063420678,
      "grad_norm": 0.44084715843200684,
      "learning_rate": 7.805984311971632e-06,
      "loss": 0.3625,
      "step": 21190
    },
    {
      "epoch": 0.9724656968473223,
      "grad_norm": 0.48754870891571045,
      "learning_rate": 7.805781370893471e-06,
      "loss": 0.4501,
      "step": 21191
    },
    {
      "epoch": 0.9725115873525767,
      "grad_norm": 0.44010522961616516,
      "learning_rate": 7.805578423068293e-06,
      "loss": 0.3676,
      "step": 21192
    },
    {
      "epoch": 0.9725574778578312,
      "grad_norm": 0.4474983513355255,
      "learning_rate": 7.805375468496583e-06,
      "loss": 0.4333,
      "step": 21193
    },
    {
      "epoch": 0.9726033683630857,
      "grad_norm": 0.43101781606674194,
      "learning_rate": 7.80517250717883e-06,
      "loss": 0.3148,
      "step": 21194
    },
    {
      "epoch": 0.9726492588683402,
      "grad_norm": 0.4258679151535034,
      "learning_rate": 7.804969539115525e-06,
      "loss": 0.3564,
      "step": 21195
    },
    {
      "epoch": 0.9726951493735946,
      "grad_norm": 0.4388556480407715,
      "learning_rate": 7.804766564307152e-06,
      "loss": 0.3925,
      "step": 21196
    },
    {
      "epoch": 0.9727410398788491,
      "grad_norm": 0.45231202244758606,
      "learning_rate": 7.8045635827542e-06,
      "loss": 0.3976,
      "step": 21197
    },
    {
      "epoch": 0.9727869303841036,
      "grad_norm": 0.47787415981292725,
      "learning_rate": 7.804360594457158e-06,
      "loss": 0.4625,
      "step": 21198
    },
    {
      "epoch": 0.972832820889358,
      "grad_norm": 0.45963776111602783,
      "learning_rate": 7.804157599416516e-06,
      "loss": 0.3989,
      "step": 21199
    },
    {
      "epoch": 0.9728787113946125,
      "grad_norm": 0.4579993784427643,
      "learning_rate": 7.803954597632757e-06,
      "loss": 0.4139,
      "step": 21200
    },
    {
      "epoch": 0.972924601899867,
      "grad_norm": 0.44423502683639526,
      "learning_rate": 7.803751589106374e-06,
      "loss": 0.345,
      "step": 21201
    },
    {
      "epoch": 0.9729704924051213,
      "grad_norm": 0.38857337832450867,
      "learning_rate": 7.803548573837851e-06,
      "loss": 0.3109,
      "step": 21202
    },
    {
      "epoch": 0.9730163829103758,
      "grad_norm": 0.4532497823238373,
      "learning_rate": 7.803345551827681e-06,
      "loss": 0.4056,
      "step": 21203
    },
    {
      "epoch": 0.9730622734156303,
      "grad_norm": 0.4809609353542328,
      "learning_rate": 7.803142523076348e-06,
      "loss": 0.4587,
      "step": 21204
    },
    {
      "epoch": 0.9731081639208847,
      "grad_norm": 0.40073466300964355,
      "learning_rate": 7.802939487584343e-06,
      "loss": 0.3098,
      "step": 21205
    },
    {
      "epoch": 0.9731540544261392,
      "grad_norm": 0.4464949369430542,
      "learning_rate": 7.802736445352152e-06,
      "loss": 0.3664,
      "step": 21206
    },
    {
      "epoch": 0.9731999449313937,
      "grad_norm": 0.4136262536048889,
      "learning_rate": 7.802533396380268e-06,
      "loss": 0.3525,
      "step": 21207
    },
    {
      "epoch": 0.9732458354366481,
      "grad_norm": 0.44977107644081116,
      "learning_rate": 7.80233034066917e-06,
      "loss": 0.3864,
      "step": 21208
    },
    {
      "epoch": 0.9732917259419026,
      "grad_norm": 0.44609615206718445,
      "learning_rate": 7.802127278219356e-06,
      "loss": 0.3833,
      "step": 21209
    },
    {
      "epoch": 0.9733376164471571,
      "grad_norm": 0.47108766436576843,
      "learning_rate": 7.80192420903131e-06,
      "loss": 0.4076,
      "step": 21210
    },
    {
      "epoch": 0.9733835069524115,
      "grad_norm": 0.46633467078208923,
      "learning_rate": 7.80172113310552e-06,
      "loss": 0.3995,
      "step": 21211
    },
    {
      "epoch": 0.973429397457666,
      "grad_norm": 0.4224914312362671,
      "learning_rate": 7.801518050442473e-06,
      "loss": 0.3394,
      "step": 21212
    },
    {
      "epoch": 0.9734752879629205,
      "grad_norm": 0.44341763854026794,
      "learning_rate": 7.801314961042661e-06,
      "loss": 0.3569,
      "step": 21213
    },
    {
      "epoch": 0.973521178468175,
      "grad_norm": 0.45019346475601196,
      "learning_rate": 7.801111864906571e-06,
      "loss": 0.3801,
      "step": 21214
    },
    {
      "epoch": 0.9735670689734294,
      "grad_norm": 0.44327664375305176,
      "learning_rate": 7.80090876203469e-06,
      "loss": 0.3898,
      "step": 21215
    },
    {
      "epoch": 0.9736129594786839,
      "grad_norm": 0.45129847526550293,
      "learning_rate": 7.800705652427509e-06,
      "loss": 0.4004,
      "step": 21216
    },
    {
      "epoch": 0.9736588499839384,
      "grad_norm": 0.4433116018772125,
      "learning_rate": 7.800502536085514e-06,
      "loss": 0.3998,
      "step": 21217
    },
    {
      "epoch": 0.9737047404891928,
      "grad_norm": 0.4202738404273987,
      "learning_rate": 7.800299413009195e-06,
      "loss": 0.3308,
      "step": 21218
    },
    {
      "epoch": 0.9737506309944473,
      "grad_norm": 0.44121769070625305,
      "learning_rate": 7.80009628319904e-06,
      "loss": 0.3602,
      "step": 21219
    },
    {
      "epoch": 0.9737965214997018,
      "grad_norm": 0.4185062646865845,
      "learning_rate": 7.799893146655536e-06,
      "loss": 0.3001,
      "step": 21220
    },
    {
      "epoch": 0.9738424120049561,
      "grad_norm": 0.46005144715309143,
      "learning_rate": 7.799690003379172e-06,
      "loss": 0.388,
      "step": 21221
    },
    {
      "epoch": 0.9738883025102106,
      "grad_norm": 0.44479671120643616,
      "learning_rate": 7.799486853370436e-06,
      "loss": 0.3703,
      "step": 21222
    },
    {
      "epoch": 0.9739341930154651,
      "grad_norm": 0.5487539172172546,
      "learning_rate": 7.79928369662982e-06,
      "loss": 0.3823,
      "step": 21223
    },
    {
      "epoch": 0.9739800835207195,
      "grad_norm": 0.42862531542778015,
      "learning_rate": 7.79908053315781e-06,
      "loss": 0.4007,
      "step": 21224
    },
    {
      "epoch": 0.974025974025974,
      "grad_norm": 0.45887845754623413,
      "learning_rate": 7.798877362954895e-06,
      "loss": 0.3923,
      "step": 21225
    },
    {
      "epoch": 0.9740718645312285,
      "grad_norm": 0.4348030090332031,
      "learning_rate": 7.798674186021563e-06,
      "loss": 0.3599,
      "step": 21226
    },
    {
      "epoch": 0.9741177550364829,
      "grad_norm": 0.45813387632369995,
      "learning_rate": 7.798471002358303e-06,
      "loss": 0.4275,
      "step": 21227
    },
    {
      "epoch": 0.9741636455417374,
      "grad_norm": 0.4671647548675537,
      "learning_rate": 7.798267811965603e-06,
      "loss": 0.3307,
      "step": 21228
    },
    {
      "epoch": 0.9742095360469919,
      "grad_norm": 0.4675450026988983,
      "learning_rate": 7.798064614843953e-06,
      "loss": 0.3933,
      "step": 21229
    },
    {
      "epoch": 0.9742554265522464,
      "grad_norm": 0.4739743173122406,
      "learning_rate": 7.797861410993839e-06,
      "loss": 0.4181,
      "step": 21230
    },
    {
      "epoch": 0.9743013170575008,
      "grad_norm": 0.4238833785057068,
      "learning_rate": 7.797658200415752e-06,
      "loss": 0.3638,
      "step": 21231
    },
    {
      "epoch": 0.9743472075627553,
      "grad_norm": 0.46204468607902527,
      "learning_rate": 7.79745498311018e-06,
      "loss": 0.4128,
      "step": 21232
    },
    {
      "epoch": 0.9743930980680098,
      "grad_norm": 0.5130534768104553,
      "learning_rate": 7.797251759077612e-06,
      "loss": 0.5088,
      "step": 21233
    },
    {
      "epoch": 0.9744389885732642,
      "grad_norm": 0.43905895948410034,
      "learning_rate": 7.797048528318537e-06,
      "loss": 0.3916,
      "step": 21234
    },
    {
      "epoch": 0.9744848790785187,
      "grad_norm": 0.4126501679420471,
      "learning_rate": 7.79684529083344e-06,
      "loss": 0.324,
      "step": 21235
    },
    {
      "epoch": 0.9745307695837732,
      "grad_norm": 0.44259434938430786,
      "learning_rate": 7.796642046622817e-06,
      "loss": 0.3917,
      "step": 21236
    },
    {
      "epoch": 0.9745766600890275,
      "grad_norm": 0.4427030086517334,
      "learning_rate": 7.79643879568715e-06,
      "loss": 0.3769,
      "step": 21237
    },
    {
      "epoch": 0.974622550594282,
      "grad_norm": 0.44074147939682007,
      "learning_rate": 7.796235538026931e-06,
      "loss": 0.389,
      "step": 21238
    },
    {
      "epoch": 0.9746684410995365,
      "grad_norm": 0.48109227418899536,
      "learning_rate": 7.796032273642649e-06,
      "loss": 0.4461,
      "step": 21239
    },
    {
      "epoch": 0.9747143316047909,
      "grad_norm": 0.4920327663421631,
      "learning_rate": 7.795829002534789e-06,
      "loss": 0.476,
      "step": 21240
    },
    {
      "epoch": 0.9747602221100454,
      "grad_norm": 0.4298250675201416,
      "learning_rate": 7.795625724703846e-06,
      "loss": 0.3411,
      "step": 21241
    },
    {
      "epoch": 0.9748061126152999,
      "grad_norm": 0.5079853534698486,
      "learning_rate": 7.795422440150302e-06,
      "loss": 0.4579,
      "step": 21242
    },
    {
      "epoch": 0.9748520031205543,
      "grad_norm": 0.4972611963748932,
      "learning_rate": 7.795219148874651e-06,
      "loss": 0.4889,
      "step": 21243
    },
    {
      "epoch": 0.9748978936258088,
      "grad_norm": 0.44802266359329224,
      "learning_rate": 7.795015850877381e-06,
      "loss": 0.3993,
      "step": 21244
    },
    {
      "epoch": 0.9749437841310633,
      "grad_norm": 0.4328779876232147,
      "learning_rate": 7.79481254615898e-06,
      "loss": 0.3291,
      "step": 21245
    },
    {
      "epoch": 0.9749896746363177,
      "grad_norm": 0.456595242023468,
      "learning_rate": 7.794609234719936e-06,
      "loss": 0.3895,
      "step": 21246
    },
    {
      "epoch": 0.9750355651415722,
      "grad_norm": 0.46625617146492004,
      "learning_rate": 7.794405916560738e-06,
      "loss": 0.3816,
      "step": 21247
    },
    {
      "epoch": 0.9750814556468267,
      "grad_norm": 0.5095096230506897,
      "learning_rate": 7.794202591681878e-06,
      "loss": 0.4663,
      "step": 21248
    },
    {
      "epoch": 0.9751273461520812,
      "grad_norm": 0.4543043076992035,
      "learning_rate": 7.793999260083842e-06,
      "loss": 0.3773,
      "step": 21249
    },
    {
      "epoch": 0.9751732366573356,
      "grad_norm": 0.41282159090042114,
      "learning_rate": 7.793795921767118e-06,
      "loss": 0.3197,
      "step": 21250
    },
    {
      "epoch": 0.9752191271625901,
      "grad_norm": 0.4473336935043335,
      "learning_rate": 7.793592576732197e-06,
      "loss": 0.3975,
      "step": 21251
    },
    {
      "epoch": 0.9752650176678446,
      "grad_norm": 0.4673854410648346,
      "learning_rate": 7.793389224979568e-06,
      "loss": 0.4064,
      "step": 21252
    },
    {
      "epoch": 0.975310908173099,
      "grad_norm": 0.4323561489582062,
      "learning_rate": 7.79318586650972e-06,
      "loss": 0.3272,
      "step": 21253
    },
    {
      "epoch": 0.9753567986783535,
      "grad_norm": 0.43160274624824524,
      "learning_rate": 7.79298250132314e-06,
      "loss": 0.3748,
      "step": 21254
    },
    {
      "epoch": 0.975402689183608,
      "grad_norm": 0.4504204988479614,
      "learning_rate": 7.79277912942032e-06,
      "loss": 0.3937,
      "step": 21255
    },
    {
      "epoch": 0.9754485796888623,
      "grad_norm": 0.466084361076355,
      "learning_rate": 7.792575750801748e-06,
      "loss": 0.4584,
      "step": 21256
    },
    {
      "epoch": 0.9754944701941168,
      "grad_norm": 0.44342315196990967,
      "learning_rate": 7.792372365467912e-06,
      "loss": 0.354,
      "step": 21257
    },
    {
      "epoch": 0.9755403606993713,
      "grad_norm": 0.4270223379135132,
      "learning_rate": 7.792168973419301e-06,
      "loss": 0.3491,
      "step": 21258
    },
    {
      "epoch": 0.9755862512046257,
      "grad_norm": 0.46173208951950073,
      "learning_rate": 7.791965574656406e-06,
      "loss": 0.361,
      "step": 21259
    },
    {
      "epoch": 0.9756321417098802,
      "grad_norm": 0.5064407587051392,
      "learning_rate": 7.791762169179715e-06,
      "loss": 0.425,
      "step": 21260
    },
    {
      "epoch": 0.9756780322151347,
      "grad_norm": 0.42244037985801697,
      "learning_rate": 7.791558756989716e-06,
      "loss": 0.3744,
      "step": 21261
    },
    {
      "epoch": 0.9757239227203891,
      "grad_norm": 0.5055184364318848,
      "learning_rate": 7.7913553380869e-06,
      "loss": 0.4271,
      "step": 21262
    },
    {
      "epoch": 0.9757698132256436,
      "grad_norm": 0.40691080689430237,
      "learning_rate": 7.791151912471755e-06,
      "loss": 0.3124,
      "step": 21263
    },
    {
      "epoch": 0.9758157037308981,
      "grad_norm": 0.41500139236450195,
      "learning_rate": 7.790948480144771e-06,
      "loss": 0.3222,
      "step": 21264
    },
    {
      "epoch": 0.9758615942361525,
      "grad_norm": 0.4799208641052246,
      "learning_rate": 7.790745041106437e-06,
      "loss": 0.4426,
      "step": 21265
    },
    {
      "epoch": 0.975907484741407,
      "grad_norm": 0.4495140314102173,
      "learning_rate": 7.79054159535724e-06,
      "loss": 0.3828,
      "step": 21266
    },
    {
      "epoch": 0.9759533752466615,
      "grad_norm": 0.40395477414131165,
      "learning_rate": 7.790338142897675e-06,
      "loss": 0.3175,
      "step": 21267
    },
    {
      "epoch": 0.975999265751916,
      "grad_norm": 0.4843035936355591,
      "learning_rate": 7.790134683728224e-06,
      "loss": 0.4215,
      "step": 21268
    },
    {
      "epoch": 0.9760451562571704,
      "grad_norm": 0.45031601190567017,
      "learning_rate": 7.78993121784938e-06,
      "loss": 0.3719,
      "step": 21269
    },
    {
      "epoch": 0.9760910467624249,
      "grad_norm": 0.4547281265258789,
      "learning_rate": 7.789727745261635e-06,
      "loss": 0.4158,
      "step": 21270
    },
    {
      "epoch": 0.9761369372676794,
      "grad_norm": 0.4601418375968933,
      "learning_rate": 7.789524265965472e-06,
      "loss": 0.396,
      "step": 21271
    },
    {
      "epoch": 0.9761828277729337,
      "grad_norm": 0.4114876985549927,
      "learning_rate": 7.789320779961385e-06,
      "loss": 0.317,
      "step": 21272
    },
    {
      "epoch": 0.9762287182781882,
      "grad_norm": 0.43526938557624817,
      "learning_rate": 7.789117287249861e-06,
      "loss": 0.3713,
      "step": 21273
    },
    {
      "epoch": 0.9762746087834427,
      "grad_norm": 0.46151474118232727,
      "learning_rate": 7.78891378783139e-06,
      "loss": 0.4576,
      "step": 21274
    },
    {
      "epoch": 0.9763204992886971,
      "grad_norm": 0.4807446599006653,
      "learning_rate": 7.788710281706463e-06,
      "loss": 0.4393,
      "step": 21275
    },
    {
      "epoch": 0.9763663897939516,
      "grad_norm": 0.450620174407959,
      "learning_rate": 7.788506768875567e-06,
      "loss": 0.3955,
      "step": 21276
    },
    {
      "epoch": 0.9764122802992061,
      "grad_norm": 0.47831541299819946,
      "learning_rate": 7.78830324933919e-06,
      "loss": 0.473,
      "step": 21277
    },
    {
      "epoch": 0.9764581708044605,
      "grad_norm": 0.49323272705078125,
      "learning_rate": 7.788099723097829e-06,
      "loss": 0.4659,
      "step": 21278
    },
    {
      "epoch": 0.976504061309715,
      "grad_norm": 0.46170616149902344,
      "learning_rate": 7.787896190151964e-06,
      "loss": 0.437,
      "step": 21279
    },
    {
      "epoch": 0.9765499518149695,
      "grad_norm": 0.4136543273925781,
      "learning_rate": 7.78769265050209e-06,
      "loss": 0.3737,
      "step": 21280
    },
    {
      "epoch": 0.9765958423202239,
      "grad_norm": 0.4596693813800812,
      "learning_rate": 7.787489104148696e-06,
      "loss": 0.374,
      "step": 21281
    },
    {
      "epoch": 0.9766417328254784,
      "grad_norm": 0.4372900426387787,
      "learning_rate": 7.787285551092268e-06,
      "loss": 0.3675,
      "step": 21282
    },
    {
      "epoch": 0.9766876233307329,
      "grad_norm": 0.48668229579925537,
      "learning_rate": 7.787081991333299e-06,
      "loss": 0.4074,
      "step": 21283
    },
    {
      "epoch": 0.9767335138359874,
      "grad_norm": 0.3620638847351074,
      "learning_rate": 7.786878424872277e-06,
      "loss": 0.2345,
      "step": 21284
    },
    {
      "epoch": 0.9767794043412418,
      "grad_norm": 0.4575035274028778,
      "learning_rate": 7.786674851709692e-06,
      "loss": 0.4152,
      "step": 21285
    },
    {
      "epoch": 0.9768252948464963,
      "grad_norm": 0.49209102988243103,
      "learning_rate": 7.786471271846034e-06,
      "loss": 0.4525,
      "step": 21286
    },
    {
      "epoch": 0.9768711853517508,
      "grad_norm": 0.4940110743045807,
      "learning_rate": 7.786267685281792e-06,
      "loss": 0.3959,
      "step": 21287
    },
    {
      "epoch": 0.9769170758570052,
      "grad_norm": 0.4221439063549042,
      "learning_rate": 7.786064092017456e-06,
      "loss": 0.3466,
      "step": 21288
    },
    {
      "epoch": 0.9769629663622597,
      "grad_norm": 0.4160853326320648,
      "learning_rate": 7.785860492053516e-06,
      "loss": 0.316,
      "step": 21289
    },
    {
      "epoch": 0.9770088568675142,
      "grad_norm": 0.5121729373931885,
      "learning_rate": 7.785656885390459e-06,
      "loss": 0.575,
      "step": 21290
    },
    {
      "epoch": 0.9770547473727685,
      "grad_norm": 0.45629557967185974,
      "learning_rate": 7.785453272028776e-06,
      "loss": 0.4339,
      "step": 21291
    },
    {
      "epoch": 0.977100637878023,
      "grad_norm": 0.4879673421382904,
      "learning_rate": 7.78524965196896e-06,
      "loss": 0.3919,
      "step": 21292
    },
    {
      "epoch": 0.9771465283832775,
      "grad_norm": 0.43076443672180176,
      "learning_rate": 7.785046025211495e-06,
      "loss": 0.3607,
      "step": 21293
    },
    {
      "epoch": 0.9771924188885319,
      "grad_norm": 0.4592038094997406,
      "learning_rate": 7.784842391756872e-06,
      "loss": 0.3992,
      "step": 21294
    },
    {
      "epoch": 0.9772383093937864,
      "grad_norm": 0.5081127285957336,
      "learning_rate": 7.784638751605586e-06,
      "loss": 0.4755,
      "step": 21295
    },
    {
      "epoch": 0.9772841998990409,
      "grad_norm": 0.4801236093044281,
      "learning_rate": 7.78443510475812e-06,
      "loss": 0.3923,
      "step": 21296
    },
    {
      "epoch": 0.9773300904042953,
      "grad_norm": 0.4494534730911255,
      "learning_rate": 7.784231451214965e-06,
      "loss": 0.3865,
      "step": 21297
    },
    {
      "epoch": 0.9773759809095498,
      "grad_norm": 0.4855068325996399,
      "learning_rate": 7.784027790976616e-06,
      "loss": 0.4493,
      "step": 21298
    },
    {
      "epoch": 0.9774218714148043,
      "grad_norm": 0.4874282479286194,
      "learning_rate": 7.783824124043556e-06,
      "loss": 0.4962,
      "step": 21299
    },
    {
      "epoch": 0.9774677619200587,
      "grad_norm": 0.4335455000400543,
      "learning_rate": 7.783620450416277e-06,
      "loss": 0.3479,
      "step": 21300
    },
    {
      "epoch": 0.9775136524253132,
      "grad_norm": 0.44533759355545044,
      "learning_rate": 7.783416770095272e-06,
      "loss": 0.3868,
      "step": 21301
    },
    {
      "epoch": 0.9775595429305677,
      "grad_norm": 0.4257276654243469,
      "learning_rate": 7.783213083081028e-06,
      "loss": 0.3363,
      "step": 21302
    },
    {
      "epoch": 0.9776054334358222,
      "grad_norm": 0.4454062581062317,
      "learning_rate": 7.783009389374034e-06,
      "loss": 0.3675,
      "step": 21303
    },
    {
      "epoch": 0.9776513239410766,
      "grad_norm": 0.5463020205497742,
      "learning_rate": 7.78280568897478e-06,
      "loss": 0.4869,
      "step": 21304
    },
    {
      "epoch": 0.9776972144463311,
      "grad_norm": 0.44738584756851196,
      "learning_rate": 7.782601981883755e-06,
      "loss": 0.3246,
      "step": 21305
    },
    {
      "epoch": 0.9777431049515856,
      "grad_norm": 0.4820520579814911,
      "learning_rate": 7.782398268101454e-06,
      "loss": 0.4739,
      "step": 21306
    },
    {
      "epoch": 0.97778899545684,
      "grad_norm": 0.46542009711265564,
      "learning_rate": 7.78219454762836e-06,
      "loss": 0.4137,
      "step": 21307
    },
    {
      "epoch": 0.9778348859620944,
      "grad_norm": 0.4879249930381775,
      "learning_rate": 7.78199082046497e-06,
      "loss": 0.4728,
      "step": 21308
    },
    {
      "epoch": 0.9778807764673489,
      "grad_norm": 0.47857338190078735,
      "learning_rate": 7.781787086611768e-06,
      "loss": 0.422,
      "step": 21309
    },
    {
      "epoch": 0.9779266669726033,
      "grad_norm": 0.43803471326828003,
      "learning_rate": 7.781583346069246e-06,
      "loss": 0.3979,
      "step": 21310
    },
    {
      "epoch": 0.9779725574778578,
      "grad_norm": 0.4761965572834015,
      "learning_rate": 7.781379598837893e-06,
      "loss": 0.4081,
      "step": 21311
    },
    {
      "epoch": 0.9780184479831123,
      "grad_norm": 0.4983188211917877,
      "learning_rate": 7.781175844918203e-06,
      "loss": 0.4873,
      "step": 21312
    },
    {
      "epoch": 0.9780643384883667,
      "grad_norm": 0.47174206376075745,
      "learning_rate": 7.78097208431066e-06,
      "loss": 0.4497,
      "step": 21313
    },
    {
      "epoch": 0.9781102289936212,
      "grad_norm": 0.4350973069667816,
      "learning_rate": 7.780768317015758e-06,
      "loss": 0.4037,
      "step": 21314
    },
    {
      "epoch": 0.9781561194988757,
      "grad_norm": 0.46594348549842834,
      "learning_rate": 7.780564543033986e-06,
      "loss": 0.4302,
      "step": 21315
    },
    {
      "epoch": 0.9782020100041301,
      "grad_norm": 0.42700326442718506,
      "learning_rate": 7.780360762365833e-06,
      "loss": 0.3752,
      "step": 21316
    },
    {
      "epoch": 0.9782479005093846,
      "grad_norm": 0.43529656529426575,
      "learning_rate": 7.78015697501179e-06,
      "loss": 0.3707,
      "step": 21317
    },
    {
      "epoch": 0.9782937910146391,
      "grad_norm": 0.45831698179244995,
      "learning_rate": 7.779953180972346e-06,
      "loss": 0.4508,
      "step": 21318
    },
    {
      "epoch": 0.9783396815198936,
      "grad_norm": 0.5671592950820923,
      "learning_rate": 7.779749380247995e-06,
      "loss": 0.3461,
      "step": 21319
    },
    {
      "epoch": 0.978385572025148,
      "grad_norm": 0.4470706880092621,
      "learning_rate": 7.779545572839223e-06,
      "loss": 0.3537,
      "step": 21320
    },
    {
      "epoch": 0.9784314625304025,
      "grad_norm": 0.45512986183166504,
      "learning_rate": 7.77934175874652e-06,
      "loss": 0.4098,
      "step": 21321
    },
    {
      "epoch": 0.978477353035657,
      "grad_norm": 0.4430018961429596,
      "learning_rate": 7.779137937970375e-06,
      "loss": 0.3799,
      "step": 21322
    },
    {
      "epoch": 0.9785232435409114,
      "grad_norm": 0.4226023256778717,
      "learning_rate": 7.778934110511285e-06,
      "loss": 0.344,
      "step": 21323
    },
    {
      "epoch": 0.9785691340461659,
      "grad_norm": 0.4300136864185333,
      "learning_rate": 7.778730276369733e-06,
      "loss": 0.3708,
      "step": 21324
    },
    {
      "epoch": 0.9786150245514204,
      "grad_norm": 0.43972426652908325,
      "learning_rate": 7.778526435546211e-06,
      "loss": 0.3872,
      "step": 21325
    },
    {
      "epoch": 0.9786609150566747,
      "grad_norm": 0.45692771673202515,
      "learning_rate": 7.778322588041212e-06,
      "loss": 0.3705,
      "step": 21326
    },
    {
      "epoch": 0.9787068055619292,
      "grad_norm": 0.4309283494949341,
      "learning_rate": 7.778118733855224e-06,
      "loss": 0.369,
      "step": 21327
    },
    {
      "epoch": 0.9787526960671837,
      "grad_norm": 0.4978989064693451,
      "learning_rate": 7.777914872988736e-06,
      "loss": 0.5149,
      "step": 21328
    },
    {
      "epoch": 0.9787985865724381,
      "grad_norm": 0.4511111080646515,
      "learning_rate": 7.77771100544224e-06,
      "loss": 0.3991,
      "step": 21329
    },
    {
      "epoch": 0.9788444770776926,
      "grad_norm": 0.4558192193508148,
      "learning_rate": 7.777507131216226e-06,
      "loss": 0.4615,
      "step": 21330
    },
    {
      "epoch": 0.9788903675829471,
      "grad_norm": 0.4370228350162506,
      "learning_rate": 7.777303250311183e-06,
      "loss": 0.3271,
      "step": 21331
    },
    {
      "epoch": 0.9789362580882015,
      "grad_norm": 0.4747064411640167,
      "learning_rate": 7.777099362727602e-06,
      "loss": 0.4273,
      "step": 21332
    },
    {
      "epoch": 0.978982148593456,
      "grad_norm": 0.4303992986679077,
      "learning_rate": 7.776895468465974e-06,
      "loss": 0.344,
      "step": 21333
    },
    {
      "epoch": 0.9790280390987105,
      "grad_norm": 0.4882585406303406,
      "learning_rate": 7.776691567526791e-06,
      "loss": 0.3492,
      "step": 21334
    },
    {
      "epoch": 0.9790739296039649,
      "grad_norm": 0.402586966753006,
      "learning_rate": 7.776487659910537e-06,
      "loss": 0.3432,
      "step": 21335
    },
    {
      "epoch": 0.9791198201092194,
      "grad_norm": 0.4301648437976837,
      "learning_rate": 7.77628374561771e-06,
      "loss": 0.3524,
      "step": 21336
    },
    {
      "epoch": 0.9791657106144739,
      "grad_norm": 0.44376108050346375,
      "learning_rate": 7.776079824648795e-06,
      "loss": 0.3596,
      "step": 21337
    },
    {
      "epoch": 0.9792116011197284,
      "grad_norm": 0.44621798396110535,
      "learning_rate": 7.775875897004284e-06,
      "loss": 0.4547,
      "step": 21338
    },
    {
      "epoch": 0.9792574916249828,
      "grad_norm": 0.44577354192733765,
      "learning_rate": 7.775671962684667e-06,
      "loss": 0.3892,
      "step": 21339
    },
    {
      "epoch": 0.9793033821302373,
      "grad_norm": 0.468477725982666,
      "learning_rate": 7.775468021690438e-06,
      "loss": 0.4118,
      "step": 21340
    },
    {
      "epoch": 0.9793492726354918,
      "grad_norm": 0.44831007719039917,
      "learning_rate": 7.775264074022082e-06,
      "loss": 0.3343,
      "step": 21341
    },
    {
      "epoch": 0.9793951631407461,
      "grad_norm": 0.4985017478466034,
      "learning_rate": 7.775060119680092e-06,
      "loss": 0.4381,
      "step": 21342
    },
    {
      "epoch": 0.9794410536460006,
      "grad_norm": 0.5300086140632629,
      "learning_rate": 7.774856158664958e-06,
      "loss": 0.4295,
      "step": 21343
    },
    {
      "epoch": 0.9794869441512551,
      "grad_norm": 0.47846633195877075,
      "learning_rate": 7.774652190977172e-06,
      "loss": 0.4111,
      "step": 21344
    },
    {
      "epoch": 0.9795328346565095,
      "grad_norm": 0.44218578934669495,
      "learning_rate": 7.774448216617224e-06,
      "loss": 0.3795,
      "step": 21345
    },
    {
      "epoch": 0.979578725161764,
      "grad_norm": 0.4750552773475647,
      "learning_rate": 7.7742442355856e-06,
      "loss": 0.4566,
      "step": 21346
    },
    {
      "epoch": 0.9796246156670185,
      "grad_norm": 0.4614761471748352,
      "learning_rate": 7.774040247882798e-06,
      "loss": 0.3875,
      "step": 21347
    },
    {
      "epoch": 0.9796705061722729,
      "grad_norm": 0.4379121959209442,
      "learning_rate": 7.773836253509302e-06,
      "loss": 0.3631,
      "step": 21348
    },
    {
      "epoch": 0.9797163966775274,
      "grad_norm": 0.44099926948547363,
      "learning_rate": 7.773632252465606e-06,
      "loss": 0.3951,
      "step": 21349
    },
    {
      "epoch": 0.9797622871827819,
      "grad_norm": 0.46316298842430115,
      "learning_rate": 7.7734282447522e-06,
      "loss": 0.4105,
      "step": 21350
    },
    {
      "epoch": 0.9798081776880363,
      "grad_norm": 0.4682682454586029,
      "learning_rate": 7.773224230369576e-06,
      "loss": 0.4671,
      "step": 21351
    },
    {
      "epoch": 0.9798540681932908,
      "grad_norm": 0.4232224225997925,
      "learning_rate": 7.77302020931822e-06,
      "loss": 0.3117,
      "step": 21352
    },
    {
      "epoch": 0.9798999586985453,
      "grad_norm": 0.4490441381931305,
      "learning_rate": 7.772816181598627e-06,
      "loss": 0.3763,
      "step": 21353
    },
    {
      "epoch": 0.9799458492037997,
      "grad_norm": 0.471253365278244,
      "learning_rate": 7.772612147211287e-06,
      "loss": 0.4326,
      "step": 21354
    },
    {
      "epoch": 0.9799917397090542,
      "grad_norm": 0.43836891651153564,
      "learning_rate": 7.77240810615669e-06,
      "loss": 0.336,
      "step": 21355
    },
    {
      "epoch": 0.9800376302143087,
      "grad_norm": 0.47028762102127075,
      "learning_rate": 7.772204058435325e-06,
      "loss": 0.419,
      "step": 21356
    },
    {
      "epoch": 0.9800835207195632,
      "grad_norm": 0.4634118974208832,
      "learning_rate": 7.772000004047684e-06,
      "loss": 0.445,
      "step": 21357
    },
    {
      "epoch": 0.9801294112248176,
      "grad_norm": 0.4604529142379761,
      "learning_rate": 7.771795942994259e-06,
      "loss": 0.4378,
      "step": 21358
    },
    {
      "epoch": 0.9801753017300721,
      "grad_norm": 0.4447031319141388,
      "learning_rate": 7.771591875275539e-06,
      "loss": 0.372,
      "step": 21359
    },
    {
      "epoch": 0.9802211922353266,
      "grad_norm": 0.46683269739151,
      "learning_rate": 7.771387800892016e-06,
      "loss": 0.4355,
      "step": 21360
    },
    {
      "epoch": 0.9802670827405809,
      "grad_norm": 0.43625301122665405,
      "learning_rate": 7.77118371984418e-06,
      "loss": 0.351,
      "step": 21361
    },
    {
      "epoch": 0.9803129732458354,
      "grad_norm": 0.44543755054473877,
      "learning_rate": 7.77097963213252e-06,
      "loss": 0.3932,
      "step": 21362
    },
    {
      "epoch": 0.9803588637510899,
      "grad_norm": 0.5127671360969543,
      "learning_rate": 7.770775537757532e-06,
      "loss": 0.4604,
      "step": 21363
    },
    {
      "epoch": 0.9804047542563443,
      "grad_norm": 0.43850618600845337,
      "learning_rate": 7.770571436719698e-06,
      "loss": 0.3897,
      "step": 21364
    },
    {
      "epoch": 0.9804506447615988,
      "grad_norm": 0.43809282779693604,
      "learning_rate": 7.770367329019519e-06,
      "loss": 0.3509,
      "step": 21365
    },
    {
      "epoch": 0.9804965352668533,
      "grad_norm": 0.4692409932613373,
      "learning_rate": 7.770163214657477e-06,
      "loss": 0.4612,
      "step": 21366
    },
    {
      "epoch": 0.9805424257721077,
      "grad_norm": 0.44617512822151184,
      "learning_rate": 7.76995909363407e-06,
      "loss": 0.3635,
      "step": 21367
    },
    {
      "epoch": 0.9805883162773622,
      "grad_norm": 0.42949068546295166,
      "learning_rate": 7.769754965949785e-06,
      "loss": 0.3733,
      "step": 21368
    },
    {
      "epoch": 0.9806342067826167,
      "grad_norm": 0.47484028339385986,
      "learning_rate": 7.769550831605112e-06,
      "loss": 0.4098,
      "step": 21369
    },
    {
      "epoch": 0.9806800972878711,
      "grad_norm": 0.5011220574378967,
      "learning_rate": 7.769346690600546e-06,
      "loss": 0.4773,
      "step": 21370
    },
    {
      "epoch": 0.9807259877931256,
      "grad_norm": 0.43971386551856995,
      "learning_rate": 7.769142542936572e-06,
      "loss": 0.3832,
      "step": 21371
    },
    {
      "epoch": 0.9807718782983801,
      "grad_norm": 0.442545086145401,
      "learning_rate": 7.768938388613687e-06,
      "loss": 0.3766,
      "step": 21372
    },
    {
      "epoch": 0.9808177688036346,
      "grad_norm": 0.4600960910320282,
      "learning_rate": 7.768734227632377e-06,
      "loss": 0.3953,
      "step": 21373
    },
    {
      "epoch": 0.980863659308889,
      "grad_norm": 0.4301464855670929,
      "learning_rate": 7.768530059993135e-06,
      "loss": 0.3698,
      "step": 21374
    },
    {
      "epoch": 0.9809095498141435,
      "grad_norm": 0.45661431550979614,
      "learning_rate": 7.768325885696454e-06,
      "loss": 0.4081,
      "step": 21375
    },
    {
      "epoch": 0.980955440319398,
      "grad_norm": 0.4460228383541107,
      "learning_rate": 7.768121704742822e-06,
      "loss": 0.3747,
      "step": 21376
    },
    {
      "epoch": 0.9810013308246524,
      "grad_norm": 0.5190106630325317,
      "learning_rate": 7.76791751713273e-06,
      "loss": 0.4475,
      "step": 21377
    },
    {
      "epoch": 0.9810472213299068,
      "grad_norm": 0.4498862326145172,
      "learning_rate": 7.76771332286667e-06,
      "loss": 0.3375,
      "step": 21378
    },
    {
      "epoch": 0.9810931118351613,
      "grad_norm": 0.4673357903957367,
      "learning_rate": 7.767509121945134e-06,
      "loss": 0.3915,
      "step": 21379
    },
    {
      "epoch": 0.9811390023404157,
      "grad_norm": 0.4895716607570648,
      "learning_rate": 7.767304914368613e-06,
      "loss": 0.5354,
      "step": 21380
    },
    {
      "epoch": 0.9811848928456702,
      "grad_norm": 0.43378815054893494,
      "learning_rate": 7.767100700137594e-06,
      "loss": 0.325,
      "step": 21381
    },
    {
      "epoch": 0.9812307833509247,
      "grad_norm": 0.4778186082839966,
      "learning_rate": 7.766896479252574e-06,
      "loss": 0.4569,
      "step": 21382
    },
    {
      "epoch": 0.9812766738561791,
      "grad_norm": 0.46774858236312866,
      "learning_rate": 7.766692251714039e-06,
      "loss": 0.3482,
      "step": 21383
    },
    {
      "epoch": 0.9813225643614336,
      "grad_norm": 0.4294929802417755,
      "learning_rate": 7.766488017522484e-06,
      "loss": 0.3832,
      "step": 21384
    },
    {
      "epoch": 0.9813684548666881,
      "grad_norm": 0.43981316685676575,
      "learning_rate": 7.766283776678398e-06,
      "loss": 0.3742,
      "step": 21385
    },
    {
      "epoch": 0.9814143453719425,
      "grad_norm": 0.45869478583335876,
      "learning_rate": 7.766079529182273e-06,
      "loss": 0.4192,
      "step": 21386
    },
    {
      "epoch": 0.981460235877197,
      "grad_norm": 0.4419727921485901,
      "learning_rate": 7.765875275034598e-06,
      "loss": 0.3874,
      "step": 21387
    },
    {
      "epoch": 0.9815061263824515,
      "grad_norm": 0.43578439950942993,
      "learning_rate": 7.765671014235867e-06,
      "loss": 0.3907,
      "step": 21388
    },
    {
      "epoch": 0.9815520168877059,
      "grad_norm": 0.45674291253089905,
      "learning_rate": 7.76546674678657e-06,
      "loss": 0.3853,
      "step": 21389
    },
    {
      "epoch": 0.9815979073929604,
      "grad_norm": 0.4652651250362396,
      "learning_rate": 7.765262472687198e-06,
      "loss": 0.4219,
      "step": 21390
    },
    {
      "epoch": 0.9816437978982149,
      "grad_norm": 0.4809570014476776,
      "learning_rate": 7.765058191938243e-06,
      "loss": 0.4897,
      "step": 21391
    },
    {
      "epoch": 0.9816896884034694,
      "grad_norm": 0.4326867163181305,
      "learning_rate": 7.764853904540197e-06,
      "loss": 0.4157,
      "step": 21392
    },
    {
      "epoch": 0.9817355789087238,
      "grad_norm": 0.43645909428596497,
      "learning_rate": 7.764649610493548e-06,
      "loss": 0.3618,
      "step": 21393
    },
    {
      "epoch": 0.9817814694139783,
      "grad_norm": 0.42483335733413696,
      "learning_rate": 7.76444530979879e-06,
      "loss": 0.319,
      "step": 21394
    },
    {
      "epoch": 0.9818273599192328,
      "grad_norm": 0.4879879951477051,
      "learning_rate": 7.764241002456414e-06,
      "loss": 0.4356,
      "step": 21395
    },
    {
      "epoch": 0.9818732504244871,
      "grad_norm": 0.4646546542644501,
      "learning_rate": 7.76403668846691e-06,
      "loss": 0.4412,
      "step": 21396
    },
    {
      "epoch": 0.9819191409297416,
      "grad_norm": 0.43390581011772156,
      "learning_rate": 7.76383236783077e-06,
      "loss": 0.3747,
      "step": 21397
    },
    {
      "epoch": 0.9819650314349961,
      "grad_norm": 0.46593526005744934,
      "learning_rate": 7.763628040548485e-06,
      "loss": 0.3867,
      "step": 21398
    },
    {
      "epoch": 0.9820109219402505,
      "grad_norm": 0.44699764251708984,
      "learning_rate": 7.763423706620547e-06,
      "loss": 0.3927,
      "step": 21399
    },
    {
      "epoch": 0.982056812445505,
      "grad_norm": 0.4481673240661621,
      "learning_rate": 7.763219366047449e-06,
      "loss": 0.3931,
      "step": 21400
    },
    {
      "epoch": 0.9821027029507595,
      "grad_norm": 0.46811288595199585,
      "learning_rate": 7.763015018829679e-06,
      "loss": 0.3982,
      "step": 21401
    },
    {
      "epoch": 0.9821485934560139,
      "grad_norm": 0.5100700855255127,
      "learning_rate": 7.76281066496773e-06,
      "loss": 0.448,
      "step": 21402
    },
    {
      "epoch": 0.9821944839612684,
      "grad_norm": 0.4577378034591675,
      "learning_rate": 7.762606304462092e-06,
      "loss": 0.4182,
      "step": 21403
    },
    {
      "epoch": 0.9822403744665229,
      "grad_norm": 0.4335230588912964,
      "learning_rate": 7.76240193731326e-06,
      "loss": 0.3404,
      "step": 21404
    },
    {
      "epoch": 0.9822862649717773,
      "grad_norm": 0.5428340435028076,
      "learning_rate": 7.762197563521723e-06,
      "loss": 0.4815,
      "step": 21405
    },
    {
      "epoch": 0.9823321554770318,
      "grad_norm": 0.4262878894805908,
      "learning_rate": 7.761993183087973e-06,
      "loss": 0.3217,
      "step": 21406
    },
    {
      "epoch": 0.9823780459822863,
      "grad_norm": 0.4616797864437103,
      "learning_rate": 7.761788796012498e-06,
      "loss": 0.3745,
      "step": 21407
    },
    {
      "epoch": 0.9824239364875408,
      "grad_norm": 0.49222397804260254,
      "learning_rate": 7.761584402295796e-06,
      "loss": 0.4363,
      "step": 21408
    },
    {
      "epoch": 0.9824698269927952,
      "grad_norm": 0.46838656067848206,
      "learning_rate": 7.761380001938354e-06,
      "loss": 0.458,
      "step": 21409
    },
    {
      "epoch": 0.9825157174980497,
      "grad_norm": 0.44366154074668884,
      "learning_rate": 7.761175594940663e-06,
      "loss": 0.3787,
      "step": 21410
    },
    {
      "epoch": 0.9825616080033042,
      "grad_norm": 0.4549409747123718,
      "learning_rate": 7.760971181303218e-06,
      "loss": 0.4404,
      "step": 21411
    },
    {
      "epoch": 0.9826074985085586,
      "grad_norm": 0.43890735507011414,
      "learning_rate": 7.760766761026508e-06,
      "loss": 0.3638,
      "step": 21412
    },
    {
      "epoch": 0.982653389013813,
      "grad_norm": 0.4766365885734558,
      "learning_rate": 7.760562334111026e-06,
      "loss": 0.4074,
      "step": 21413
    },
    {
      "epoch": 0.9826992795190675,
      "grad_norm": 0.47251564264297485,
      "learning_rate": 7.760357900557261e-06,
      "loss": 0.4745,
      "step": 21414
    },
    {
      "epoch": 0.9827451700243219,
      "grad_norm": 0.41670334339141846,
      "learning_rate": 7.760153460365709e-06,
      "loss": 0.3315,
      "step": 21415
    },
    {
      "epoch": 0.9827910605295764,
      "grad_norm": 0.40815308690071106,
      "learning_rate": 7.759949013536858e-06,
      "loss": 0.32,
      "step": 21416
    },
    {
      "epoch": 0.9828369510348309,
      "grad_norm": 0.4602589011192322,
      "learning_rate": 7.7597445600712e-06,
      "loss": 0.4036,
      "step": 21417
    },
    {
      "epoch": 0.9828828415400853,
      "grad_norm": 0.4398094713687897,
      "learning_rate": 7.759540099969227e-06,
      "loss": 0.36,
      "step": 21418
    },
    {
      "epoch": 0.9829287320453398,
      "grad_norm": 0.4365086257457733,
      "learning_rate": 7.759335633231433e-06,
      "loss": 0.3931,
      "step": 21419
    },
    {
      "epoch": 0.9829746225505943,
      "grad_norm": 0.508642852306366,
      "learning_rate": 7.759131159858307e-06,
      "loss": 0.4806,
      "step": 21420
    },
    {
      "epoch": 0.9830205130558487,
      "grad_norm": 0.4520360827445984,
      "learning_rate": 7.75892667985034e-06,
      "loss": 0.3607,
      "step": 21421
    },
    {
      "epoch": 0.9830664035611032,
      "grad_norm": 0.4412452280521393,
      "learning_rate": 7.758722193208026e-06,
      "loss": 0.3806,
      "step": 21422
    },
    {
      "epoch": 0.9831122940663577,
      "grad_norm": 0.4269466698169708,
      "learning_rate": 7.758517699931856e-06,
      "loss": 0.3095,
      "step": 21423
    },
    {
      "epoch": 0.9831581845716121,
      "grad_norm": 0.47109657526016235,
      "learning_rate": 7.75831320002232e-06,
      "loss": 0.4583,
      "step": 21424
    },
    {
      "epoch": 0.9832040750768666,
      "grad_norm": 0.4524524211883545,
      "learning_rate": 7.758108693479914e-06,
      "loss": 0.4248,
      "step": 21425
    },
    {
      "epoch": 0.9832499655821211,
      "grad_norm": 0.43599799275398254,
      "learning_rate": 7.757904180305125e-06,
      "loss": 0.3326,
      "step": 21426
    },
    {
      "epoch": 0.9832958560873756,
      "grad_norm": 0.41575875878334045,
      "learning_rate": 7.757699660498448e-06,
      "loss": 0.3457,
      "step": 21427
    },
    {
      "epoch": 0.98334174659263,
      "grad_norm": 0.707442045211792,
      "learning_rate": 7.757495134060373e-06,
      "loss": 0.4842,
      "step": 21428
    },
    {
      "epoch": 0.9833876370978845,
      "grad_norm": 0.4459569454193115,
      "learning_rate": 7.757290600991391e-06,
      "loss": 0.3772,
      "step": 21429
    },
    {
      "epoch": 0.983433527603139,
      "grad_norm": 0.45639532804489136,
      "learning_rate": 7.757086061292e-06,
      "loss": 0.4159,
      "step": 21430
    },
    {
      "epoch": 0.9834794181083933,
      "grad_norm": 0.4375492334365845,
      "learning_rate": 7.756881514962682e-06,
      "loss": 0.3439,
      "step": 21431
    },
    {
      "epoch": 0.9835253086136478,
      "grad_norm": 0.41172361373901367,
      "learning_rate": 7.756676962003937e-06,
      "loss": 0.2978,
      "step": 21432
    },
    {
      "epoch": 0.9835711991189023,
      "grad_norm": 0.4495747685432434,
      "learning_rate": 7.756472402416254e-06,
      "loss": 0.4454,
      "step": 21433
    },
    {
      "epoch": 0.9836170896241567,
      "grad_norm": 0.4470408260822296,
      "learning_rate": 7.756267836200125e-06,
      "loss": 0.4074,
      "step": 21434
    },
    {
      "epoch": 0.9836629801294112,
      "grad_norm": 0.44430989027023315,
      "learning_rate": 7.756063263356039e-06,
      "loss": 0.4113,
      "step": 21435
    },
    {
      "epoch": 0.9837088706346657,
      "grad_norm": 0.4442964196205139,
      "learning_rate": 7.755858683884493e-06,
      "loss": 0.3907,
      "step": 21436
    },
    {
      "epoch": 0.9837547611399201,
      "grad_norm": 0.48289164900779724,
      "learning_rate": 7.755654097785976e-06,
      "loss": 0.4799,
      "step": 21437
    },
    {
      "epoch": 0.9838006516451746,
      "grad_norm": 0.4302290976047516,
      "learning_rate": 7.755449505060982e-06,
      "loss": 0.3702,
      "step": 21438
    },
    {
      "epoch": 0.9838465421504291,
      "grad_norm": 0.46389952301979065,
      "learning_rate": 7.75524490571e-06,
      "loss": 0.389,
      "step": 21439
    },
    {
      "epoch": 0.9838924326556835,
      "grad_norm": 0.5733098387718201,
      "learning_rate": 7.755040299733522e-06,
      "loss": 0.5442,
      "step": 21440
    },
    {
      "epoch": 0.983938323160938,
      "grad_norm": 0.4466260075569153,
      "learning_rate": 7.754835687132044e-06,
      "loss": 0.3657,
      "step": 21441
    },
    {
      "epoch": 0.9839842136661925,
      "grad_norm": 0.4355860650539398,
      "learning_rate": 7.754631067906055e-06,
      "loss": 0.3725,
      "step": 21442
    },
    {
      "epoch": 0.9840301041714469,
      "grad_norm": 0.4843028485774994,
      "learning_rate": 7.754426442056049e-06,
      "loss": 0.4774,
      "step": 21443
    },
    {
      "epoch": 0.9840759946767014,
      "grad_norm": 0.4177616834640503,
      "learning_rate": 7.754221809582514e-06,
      "loss": 0.3237,
      "step": 21444
    },
    {
      "epoch": 0.9841218851819559,
      "grad_norm": 0.4778716266155243,
      "learning_rate": 7.754017170485949e-06,
      "loss": 0.3935,
      "step": 21445
    },
    {
      "epoch": 0.9841677756872104,
      "grad_norm": 0.4714200496673584,
      "learning_rate": 7.753812524766838e-06,
      "loss": 0.4357,
      "step": 21446
    },
    {
      "epoch": 0.9842136661924648,
      "grad_norm": 0.4577895402908325,
      "learning_rate": 7.753607872425678e-06,
      "loss": 0.4082,
      "step": 21447
    },
    {
      "epoch": 0.9842595566977193,
      "grad_norm": 0.4341447949409485,
      "learning_rate": 7.75340321346296e-06,
      "loss": 0.3996,
      "step": 21448
    },
    {
      "epoch": 0.9843054472029737,
      "grad_norm": 0.4190983474254608,
      "learning_rate": 7.753198547879177e-06,
      "loss": 0.3264,
      "step": 21449
    },
    {
      "epoch": 0.9843513377082281,
      "grad_norm": 0.43608057498931885,
      "learning_rate": 7.75299387567482e-06,
      "loss": 0.3553,
      "step": 21450
    },
    {
      "epoch": 0.9843972282134826,
      "grad_norm": 0.4212685823440552,
      "learning_rate": 7.75278919685038e-06,
      "loss": 0.3445,
      "step": 21451
    },
    {
      "epoch": 0.9844431187187371,
      "grad_norm": 0.43053552508354187,
      "learning_rate": 7.752584511406354e-06,
      "loss": 0.3289,
      "step": 21452
    },
    {
      "epoch": 0.9844890092239915,
      "grad_norm": 0.46430766582489014,
      "learning_rate": 7.75237981934323e-06,
      "loss": 0.4367,
      "step": 21453
    },
    {
      "epoch": 0.984534899729246,
      "grad_norm": 0.44362694025039673,
      "learning_rate": 7.7521751206615e-06,
      "loss": 0.3871,
      "step": 21454
    },
    {
      "epoch": 0.9845807902345005,
      "grad_norm": 0.461493581533432,
      "learning_rate": 7.751970415361658e-06,
      "loss": 0.4497,
      "step": 21455
    },
    {
      "epoch": 0.9846266807397549,
      "grad_norm": 0.4793708622455597,
      "learning_rate": 7.751765703444197e-06,
      "loss": 0.4265,
      "step": 21456
    },
    {
      "epoch": 0.9846725712450094,
      "grad_norm": 0.49860772490501404,
      "learning_rate": 7.751560984909605e-06,
      "loss": 0.485,
      "step": 21457
    },
    {
      "epoch": 0.9847184617502639,
      "grad_norm": 0.4721015989780426,
      "learning_rate": 7.751356259758379e-06,
      "loss": 0.3945,
      "step": 21458
    },
    {
      "epoch": 0.9847643522555183,
      "grad_norm": 0.46541470289230347,
      "learning_rate": 7.75115152799101e-06,
      "loss": 0.4013,
      "step": 21459
    },
    {
      "epoch": 0.9848102427607728,
      "grad_norm": 0.44064638018608093,
      "learning_rate": 7.75094678960799e-06,
      "loss": 0.3588,
      "step": 21460
    },
    {
      "epoch": 0.9848561332660273,
      "grad_norm": 0.440988689661026,
      "learning_rate": 7.750742044609808e-06,
      "loss": 0.3489,
      "step": 21461
    },
    {
      "epoch": 0.9849020237712818,
      "grad_norm": 0.43416938185691833,
      "learning_rate": 7.750537292996964e-06,
      "loss": 0.3679,
      "step": 21462
    },
    {
      "epoch": 0.9849479142765362,
      "grad_norm": 0.45450904965400696,
      "learning_rate": 7.750332534769945e-06,
      "loss": 0.38,
      "step": 21463
    },
    {
      "epoch": 0.9849938047817907,
      "grad_norm": 0.4444591701030731,
      "learning_rate": 7.750127769929243e-06,
      "loss": 0.351,
      "step": 21464
    },
    {
      "epoch": 0.9850396952870452,
      "grad_norm": 0.4505695402622223,
      "learning_rate": 7.749922998475352e-06,
      "loss": 0.3247,
      "step": 21465
    },
    {
      "epoch": 0.9850855857922995,
      "grad_norm": 0.44429805874824524,
      "learning_rate": 7.749718220408765e-06,
      "loss": 0.3575,
      "step": 21466
    },
    {
      "epoch": 0.985131476297554,
      "grad_norm": 0.49736854434013367,
      "learning_rate": 7.749513435729972e-06,
      "loss": 0.5173,
      "step": 21467
    },
    {
      "epoch": 0.9851773668028085,
      "grad_norm": 0.4590285122394562,
      "learning_rate": 7.749308644439467e-06,
      "loss": 0.4094,
      "step": 21468
    },
    {
      "epoch": 0.9852232573080629,
      "grad_norm": 0.48910629749298096,
      "learning_rate": 7.749103846537743e-06,
      "loss": 0.4398,
      "step": 21469
    },
    {
      "epoch": 0.9852691478133174,
      "grad_norm": 0.47019514441490173,
      "learning_rate": 7.748899042025293e-06,
      "loss": 0.4568,
      "step": 21470
    },
    {
      "epoch": 0.9853150383185719,
      "grad_norm": 0.44276532530784607,
      "learning_rate": 7.748694230902607e-06,
      "loss": 0.3539,
      "step": 21471
    },
    {
      "epoch": 0.9853609288238263,
      "grad_norm": 0.4677266478538513,
      "learning_rate": 7.748489413170178e-06,
      "loss": 0.3014,
      "step": 21472
    },
    {
      "epoch": 0.9854068193290808,
      "grad_norm": 0.4319685697555542,
      "learning_rate": 7.7482845888285e-06,
      "loss": 0.3548,
      "step": 21473
    },
    {
      "epoch": 0.9854527098343353,
      "grad_norm": 0.4747561812400818,
      "learning_rate": 7.748079757878066e-06,
      "loss": 0.4033,
      "step": 21474
    },
    {
      "epoch": 0.9854986003395897,
      "grad_norm": 0.41377580165863037,
      "learning_rate": 7.747874920319367e-06,
      "loss": 0.3112,
      "step": 21475
    },
    {
      "epoch": 0.9855444908448442,
      "grad_norm": 0.4243742525577545,
      "learning_rate": 7.747670076152897e-06,
      "loss": 0.331,
      "step": 21476
    },
    {
      "epoch": 0.9855903813500987,
      "grad_norm": 0.483267217874527,
      "learning_rate": 7.747465225379148e-06,
      "loss": 0.4284,
      "step": 21477
    },
    {
      "epoch": 0.9856362718553531,
      "grad_norm": 0.44622886180877686,
      "learning_rate": 7.74726036799861e-06,
      "loss": 0.3535,
      "step": 21478
    },
    {
      "epoch": 0.9856821623606076,
      "grad_norm": 0.4148627817630768,
      "learning_rate": 7.747055504011777e-06,
      "loss": 0.2955,
      "step": 21479
    },
    {
      "epoch": 0.9857280528658621,
      "grad_norm": 0.5059966444969177,
      "learning_rate": 7.746850633419144e-06,
      "loss": 0.4539,
      "step": 21480
    },
    {
      "epoch": 0.9857739433711166,
      "grad_norm": 0.45976540446281433,
      "learning_rate": 7.746645756221204e-06,
      "loss": 0.4188,
      "step": 21481
    },
    {
      "epoch": 0.985819833876371,
      "grad_norm": 0.42554357647895813,
      "learning_rate": 7.746440872418447e-06,
      "loss": 0.3676,
      "step": 21482
    },
    {
      "epoch": 0.9858657243816255,
      "grad_norm": 0.45818156003952026,
      "learning_rate": 7.746235982011365e-06,
      "loss": 0.3794,
      "step": 21483
    },
    {
      "epoch": 0.98591161488688,
      "grad_norm": 0.4491177201271057,
      "learning_rate": 7.746031085000454e-06,
      "loss": 0.3697,
      "step": 21484
    },
    {
      "epoch": 0.9859575053921343,
      "grad_norm": 0.4614400267601013,
      "learning_rate": 7.745826181386204e-06,
      "loss": 0.3853,
      "step": 21485
    },
    {
      "epoch": 0.9860033958973888,
      "grad_norm": 0.49437251687049866,
      "learning_rate": 7.745621271169107e-06,
      "loss": 0.4828,
      "step": 21486
    },
    {
      "epoch": 0.9860492864026433,
      "grad_norm": 0.5032725930213928,
      "learning_rate": 7.74541635434966e-06,
      "loss": 0.4222,
      "step": 21487
    },
    {
      "epoch": 0.9860951769078977,
      "grad_norm": 0.45090243220329285,
      "learning_rate": 7.745211430928352e-06,
      "loss": 0.3947,
      "step": 21488
    },
    {
      "epoch": 0.9861410674131522,
      "grad_norm": 0.47474828362464905,
      "learning_rate": 7.745006500905677e-06,
      "loss": 0.4096,
      "step": 21489
    },
    {
      "epoch": 0.9861869579184067,
      "grad_norm": 0.4694094955921173,
      "learning_rate": 7.744801564282128e-06,
      "loss": 0.425,
      "step": 21490
    },
    {
      "epoch": 0.9862328484236611,
      "grad_norm": 0.48593080043792725,
      "learning_rate": 7.744596621058199e-06,
      "loss": 0.4749,
      "step": 21491
    },
    {
      "epoch": 0.9862787389289156,
      "grad_norm": 0.3990936577320099,
      "learning_rate": 7.744391671234376e-06,
      "loss": 0.3063,
      "step": 21492
    },
    {
      "epoch": 0.9863246294341701,
      "grad_norm": 0.4184863269329071,
      "learning_rate": 7.744186714811163e-06,
      "loss": 0.3824,
      "step": 21493
    },
    {
      "epoch": 0.9863705199394245,
      "grad_norm": 0.4958454966545105,
      "learning_rate": 7.743981751789045e-06,
      "loss": 0.4769,
      "step": 21494
    },
    {
      "epoch": 0.986416410444679,
      "grad_norm": 0.43260058760643005,
      "learning_rate": 7.743776782168517e-06,
      "loss": 0.3175,
      "step": 21495
    },
    {
      "epoch": 0.9864623009499335,
      "grad_norm": 0.48893362283706665,
      "learning_rate": 7.743571805950073e-06,
      "loss": 0.4516,
      "step": 21496
    },
    {
      "epoch": 0.9865081914551879,
      "grad_norm": 0.4828796088695526,
      "learning_rate": 7.743366823134202e-06,
      "loss": 0.3794,
      "step": 21497
    },
    {
      "epoch": 0.9865540819604424,
      "grad_norm": 0.46408811211586,
      "learning_rate": 7.743161833721402e-06,
      "loss": 0.4557,
      "step": 21498
    },
    {
      "epoch": 0.9865999724656969,
      "grad_norm": 0.4445958435535431,
      "learning_rate": 7.742956837712161e-06,
      "loss": 0.328,
      "step": 21499
    },
    {
      "epoch": 0.9866458629709514,
      "grad_norm": 0.43978598713874817,
      "learning_rate": 7.742751835106977e-06,
      "loss": 0.3646,
      "step": 21500
    },
    {
      "epoch": 0.9866917534762057,
      "grad_norm": 0.4201546609401703,
      "learning_rate": 7.742546825906338e-06,
      "loss": 0.3474,
      "step": 21501
    },
    {
      "epoch": 0.9867376439814602,
      "grad_norm": 0.4824674427509308,
      "learning_rate": 7.742341810110741e-06,
      "loss": 0.4708,
      "step": 21502
    },
    {
      "epoch": 0.9867835344867147,
      "grad_norm": 0.4747612476348877,
      "learning_rate": 7.742136787720678e-06,
      "loss": 0.4049,
      "step": 21503
    },
    {
      "epoch": 0.9868294249919691,
      "grad_norm": 0.7261636853218079,
      "learning_rate": 7.74193175873664e-06,
      "loss": 0.4409,
      "step": 21504
    },
    {
      "epoch": 0.9868753154972236,
      "grad_norm": 0.44836804270744324,
      "learning_rate": 7.741726723159122e-06,
      "loss": 0.4114,
      "step": 21505
    },
    {
      "epoch": 0.9869212060024781,
      "grad_norm": 0.45542311668395996,
      "learning_rate": 7.741521680988617e-06,
      "loss": 0.3968,
      "step": 21506
    },
    {
      "epoch": 0.9869670965077325,
      "grad_norm": 0.4229353666305542,
      "learning_rate": 7.74131663222562e-06,
      "loss": 0.3489,
      "step": 21507
    },
    {
      "epoch": 0.987012987012987,
      "grad_norm": 0.4770808517932892,
      "learning_rate": 7.741111576870618e-06,
      "loss": 0.3872,
      "step": 21508
    },
    {
      "epoch": 0.9870588775182415,
      "grad_norm": 0.432889848947525,
      "learning_rate": 7.740906514924109e-06,
      "loss": 0.3563,
      "step": 21509
    },
    {
      "epoch": 0.9871047680234959,
      "grad_norm": 0.47834312915802,
      "learning_rate": 7.740701446386585e-06,
      "loss": 0.4296,
      "step": 21510
    },
    {
      "epoch": 0.9871506585287504,
      "grad_norm": 0.4557517170906067,
      "learning_rate": 7.740496371258538e-06,
      "loss": 0.4572,
      "step": 21511
    },
    {
      "epoch": 0.9871965490340049,
      "grad_norm": 0.4850899875164032,
      "learning_rate": 7.740291289540464e-06,
      "loss": 0.4527,
      "step": 21512
    },
    {
      "epoch": 0.9872424395392593,
      "grad_norm": 0.47328728437423706,
      "learning_rate": 7.740086201232852e-06,
      "loss": 0.4522,
      "step": 21513
    },
    {
      "epoch": 0.9872883300445138,
      "grad_norm": 0.45360296964645386,
      "learning_rate": 7.7398811063362e-06,
      "loss": 0.3582,
      "step": 21514
    },
    {
      "epoch": 0.9873342205497683,
      "grad_norm": 0.4558408558368683,
      "learning_rate": 7.739676004850997e-06,
      "loss": 0.4076,
      "step": 21515
    },
    {
      "epoch": 0.9873801110550228,
      "grad_norm": 0.4704846739768982,
      "learning_rate": 7.73947089677774e-06,
      "loss": 0.3849,
      "step": 21516
    },
    {
      "epoch": 0.9874260015602772,
      "grad_norm": 0.4336792230606079,
      "learning_rate": 7.739265782116916e-06,
      "loss": 0.3325,
      "step": 21517
    },
    {
      "epoch": 0.9874718920655317,
      "grad_norm": 0.4466922879219055,
      "learning_rate": 7.739060660869026e-06,
      "loss": 0.3458,
      "step": 21518
    },
    {
      "epoch": 0.9875177825707862,
      "grad_norm": 0.4665732681751251,
      "learning_rate": 7.738855533034557e-06,
      "loss": 0.3661,
      "step": 21519
    },
    {
      "epoch": 0.9875636730760405,
      "grad_norm": 0.721197783946991,
      "learning_rate": 7.738650398614007e-06,
      "loss": 0.3462,
      "step": 21520
    },
    {
      "epoch": 0.987609563581295,
      "grad_norm": 0.4455837905406952,
      "learning_rate": 7.738445257607868e-06,
      "loss": 0.4257,
      "step": 21521
    },
    {
      "epoch": 0.9876554540865495,
      "grad_norm": 0.4395209550857544,
      "learning_rate": 7.738240110016628e-06,
      "loss": 0.3597,
      "step": 21522
    },
    {
      "epoch": 0.9877013445918039,
      "grad_norm": 0.44374555349349976,
      "learning_rate": 7.738034955840788e-06,
      "loss": 0.3592,
      "step": 21523
    },
    {
      "epoch": 0.9877472350970584,
      "grad_norm": 0.49391478300094604,
      "learning_rate": 7.737829795080837e-06,
      "loss": 0.4479,
      "step": 21524
    },
    {
      "epoch": 0.9877931256023129,
      "grad_norm": 0.4366222620010376,
      "learning_rate": 7.737624627737272e-06,
      "loss": 0.3704,
      "step": 21525
    },
    {
      "epoch": 0.9878390161075673,
      "grad_norm": 0.47419601678848267,
      "learning_rate": 7.73741945381058e-06,
      "loss": 0.4302,
      "step": 21526
    },
    {
      "epoch": 0.9878849066128218,
      "grad_norm": 0.44391629099845886,
      "learning_rate": 7.737214273301261e-06,
      "loss": 0.3577,
      "step": 21527
    },
    {
      "epoch": 0.9879307971180763,
      "grad_norm": 0.4660033583641052,
      "learning_rate": 7.737009086209804e-06,
      "loss": 0.4041,
      "step": 21528
    },
    {
      "epoch": 0.9879766876233307,
      "grad_norm": 0.511532187461853,
      "learning_rate": 7.736803892536706e-06,
      "loss": 0.5074,
      "step": 21529
    },
    {
      "epoch": 0.9880225781285852,
      "grad_norm": 0.4907664954662323,
      "learning_rate": 7.736598692282455e-06,
      "loss": 0.4506,
      "step": 21530
    },
    {
      "epoch": 0.9880684686338397,
      "grad_norm": 0.43260014057159424,
      "learning_rate": 7.73639348544755e-06,
      "loss": 0.3777,
      "step": 21531
    },
    {
      "epoch": 0.9881143591390941,
      "grad_norm": 0.42581993341445923,
      "learning_rate": 7.736188272032482e-06,
      "loss": 0.3603,
      "step": 21532
    },
    {
      "epoch": 0.9881602496443486,
      "grad_norm": 0.41736534237861633,
      "learning_rate": 7.735983052037746e-06,
      "loss": 0.3488,
      "step": 21533
    },
    {
      "epoch": 0.9882061401496031,
      "grad_norm": 0.4813328683376312,
      "learning_rate": 7.735777825463833e-06,
      "loss": 0.3712,
      "step": 21534
    },
    {
      "epoch": 0.9882520306548576,
      "grad_norm": 0.5098647475242615,
      "learning_rate": 7.735572592311239e-06,
      "loss": 0.4511,
      "step": 21535
    },
    {
      "epoch": 0.988297921160112,
      "grad_norm": 0.4447896480560303,
      "learning_rate": 7.735367352580455e-06,
      "loss": 0.3805,
      "step": 21536
    },
    {
      "epoch": 0.9883438116653664,
      "grad_norm": 0.42714807391166687,
      "learning_rate": 7.735162106271973e-06,
      "loss": 0.3678,
      "step": 21537
    },
    {
      "epoch": 0.9883897021706209,
      "grad_norm": 0.4847141206264496,
      "learning_rate": 7.734956853386295e-06,
      "loss": 0.4429,
      "step": 21538
    },
    {
      "epoch": 0.9884355926758753,
      "grad_norm": 0.4179786443710327,
      "learning_rate": 7.734751593923906e-06,
      "loss": 0.3308,
      "step": 21539
    },
    {
      "epoch": 0.9884814831811298,
      "grad_norm": 0.45779165625572205,
      "learning_rate": 7.734546327885301e-06,
      "loss": 0.4118,
      "step": 21540
    },
    {
      "epoch": 0.9885273736863843,
      "grad_norm": 0.45173683762550354,
      "learning_rate": 7.734341055270979e-06,
      "loss": 0.3948,
      "step": 21541
    },
    {
      "epoch": 0.9885732641916387,
      "grad_norm": 0.4324975907802582,
      "learning_rate": 7.734135776081426e-06,
      "loss": 0.3483,
      "step": 21542
    },
    {
      "epoch": 0.9886191546968932,
      "grad_norm": 0.4419863224029541,
      "learning_rate": 7.733930490317144e-06,
      "loss": 0.3175,
      "step": 21543
    },
    {
      "epoch": 0.9886650452021477,
      "grad_norm": 0.8894885182380676,
      "learning_rate": 7.733725197978617e-06,
      "loss": 0.4889,
      "step": 21544
    },
    {
      "epoch": 0.9887109357074021,
      "grad_norm": 0.45484885573387146,
      "learning_rate": 7.733519899066346e-06,
      "loss": 0.4199,
      "step": 21545
    },
    {
      "epoch": 0.9887568262126566,
      "grad_norm": 0.47910836338996887,
      "learning_rate": 7.733314593580822e-06,
      "loss": 0.4079,
      "step": 21546
    },
    {
      "epoch": 0.9888027167179111,
      "grad_norm": 0.5267349481582642,
      "learning_rate": 7.733109281522539e-06,
      "loss": 0.469,
      "step": 21547
    },
    {
      "epoch": 0.9888486072231655,
      "grad_norm": 0.47669488191604614,
      "learning_rate": 7.73290396289199e-06,
      "loss": 0.4661,
      "step": 21548
    },
    {
      "epoch": 0.98889449772842,
      "grad_norm": 0.49703556299209595,
      "learning_rate": 7.732698637689672e-06,
      "loss": 0.4459,
      "step": 21549
    },
    {
      "epoch": 0.9889403882336745,
      "grad_norm": 0.443128764629364,
      "learning_rate": 7.732493305916074e-06,
      "loss": 0.3163,
      "step": 21550
    },
    {
      "epoch": 0.988986278738929,
      "grad_norm": 0.4509771466255188,
      "learning_rate": 7.732287967571692e-06,
      "loss": 0.35,
      "step": 21551
    },
    {
      "epoch": 0.9890321692441834,
      "grad_norm": 0.4455101788043976,
      "learning_rate": 7.732082622657022e-06,
      "loss": 0.398,
      "step": 21552
    },
    {
      "epoch": 0.9890780597494379,
      "grad_norm": 0.45555704832077026,
      "learning_rate": 7.731877271172555e-06,
      "loss": 0.3506,
      "step": 21553
    },
    {
      "epoch": 0.9891239502546924,
      "grad_norm": 0.6252763867378235,
      "learning_rate": 7.731671913118783e-06,
      "loss": 0.3779,
      "step": 21554
    },
    {
      "epoch": 0.9891698407599467,
      "grad_norm": 0.5046229362487793,
      "learning_rate": 7.731466548496204e-06,
      "loss": 0.3319,
      "step": 21555
    },
    {
      "epoch": 0.9892157312652012,
      "grad_norm": 0.4692378044128418,
      "learning_rate": 7.731261177305309e-06,
      "loss": 0.4181,
      "step": 21556
    },
    {
      "epoch": 0.9892616217704557,
      "grad_norm": 0.46070632338523865,
      "learning_rate": 7.731055799546594e-06,
      "loss": 0.3578,
      "step": 21557
    },
    {
      "epoch": 0.9893075122757101,
      "grad_norm": 0.506351113319397,
      "learning_rate": 7.73085041522055e-06,
      "loss": 0.4795,
      "step": 21558
    },
    {
      "epoch": 0.9893534027809646,
      "grad_norm": 0.46631091833114624,
      "learning_rate": 7.730645024327674e-06,
      "loss": 0.3966,
      "step": 21559
    },
    {
      "epoch": 0.9893992932862191,
      "grad_norm": 0.45831891894340515,
      "learning_rate": 7.73043962686846e-06,
      "loss": 0.3728,
      "step": 21560
    },
    {
      "epoch": 0.9894451837914735,
      "grad_norm": 0.4572312831878662,
      "learning_rate": 7.730234222843398e-06,
      "loss": 0.3211,
      "step": 21561
    },
    {
      "epoch": 0.989491074296728,
      "grad_norm": 0.46948060393333435,
      "learning_rate": 7.730028812252985e-06,
      "loss": 0.4537,
      "step": 21562
    },
    {
      "epoch": 0.9895369648019825,
      "grad_norm": 0.4926060438156128,
      "learning_rate": 7.729823395097715e-06,
      "loss": 0.4461,
      "step": 21563
    },
    {
      "epoch": 0.9895828553072369,
      "grad_norm": 0.4430238902568817,
      "learning_rate": 7.729617971378081e-06,
      "loss": 0.3096,
      "step": 21564
    },
    {
      "epoch": 0.9896287458124914,
      "grad_norm": 0.44670823216438293,
      "learning_rate": 7.729412541094576e-06,
      "loss": 0.3575,
      "step": 21565
    },
    {
      "epoch": 0.9896746363177459,
      "grad_norm": 0.4326602518558502,
      "learning_rate": 7.729207104247699e-06,
      "loss": 0.3602,
      "step": 21566
    },
    {
      "epoch": 0.9897205268230003,
      "grad_norm": 0.4646248519420624,
      "learning_rate": 7.729001660837936e-06,
      "loss": 0.388,
      "step": 21567
    },
    {
      "epoch": 0.9897664173282548,
      "grad_norm": 0.48971086740493774,
      "learning_rate": 7.728796210865788e-06,
      "loss": 0.4093,
      "step": 21568
    },
    {
      "epoch": 0.9898123078335093,
      "grad_norm": 0.4400186538696289,
      "learning_rate": 7.728590754331744e-06,
      "loss": 0.3357,
      "step": 21569
    },
    {
      "epoch": 0.9898581983387638,
      "grad_norm": 0.44209638237953186,
      "learning_rate": 7.728385291236302e-06,
      "loss": 0.3822,
      "step": 21570
    },
    {
      "epoch": 0.9899040888440181,
      "grad_norm": 0.4958653450012207,
      "learning_rate": 7.728179821579954e-06,
      "loss": 0.3935,
      "step": 21571
    },
    {
      "epoch": 0.9899499793492726,
      "grad_norm": 0.4243580996990204,
      "learning_rate": 7.727974345363197e-06,
      "loss": 0.3558,
      "step": 21572
    },
    {
      "epoch": 0.9899958698545271,
      "grad_norm": 0.4566342234611511,
      "learning_rate": 7.727768862586519e-06,
      "loss": 0.4048,
      "step": 21573
    },
    {
      "epoch": 0.9900417603597815,
      "grad_norm": 0.4585874676704407,
      "learning_rate": 7.727563373250419e-06,
      "loss": 0.3752,
      "step": 21574
    },
    {
      "epoch": 0.990087650865036,
      "grad_norm": 0.4663224518299103,
      "learning_rate": 7.72735787735539e-06,
      "loss": 0.3834,
      "step": 21575
    },
    {
      "epoch": 0.9901335413702905,
      "grad_norm": 0.44143539667129517,
      "learning_rate": 7.727152374901927e-06,
      "loss": 0.3782,
      "step": 21576
    },
    {
      "epoch": 0.9901794318755449,
      "grad_norm": 0.43273141980171204,
      "learning_rate": 7.726946865890523e-06,
      "loss": 0.3531,
      "step": 21577
    },
    {
      "epoch": 0.9902253223807994,
      "grad_norm": 0.43265241384506226,
      "learning_rate": 7.726741350321672e-06,
      "loss": 0.3315,
      "step": 21578
    },
    {
      "epoch": 0.9902712128860539,
      "grad_norm": 0.42842116951942444,
      "learning_rate": 7.726535828195867e-06,
      "loss": 0.3578,
      "step": 21579
    },
    {
      "epoch": 0.9903171033913083,
      "grad_norm": 0.43500638008117676,
      "learning_rate": 7.726330299513608e-06,
      "loss": 0.3605,
      "step": 21580
    },
    {
      "epoch": 0.9903629938965628,
      "grad_norm": 0.4316575229167938,
      "learning_rate": 7.726124764275381e-06,
      "loss": 0.3555,
      "step": 21581
    },
    {
      "epoch": 0.9904088844018173,
      "grad_norm": 0.47060224413871765,
      "learning_rate": 7.725919222481686e-06,
      "loss": 0.4559,
      "step": 21582
    },
    {
      "epoch": 0.9904547749070717,
      "grad_norm": 0.43209201097488403,
      "learning_rate": 7.725713674133014e-06,
      "loss": 0.3684,
      "step": 21583
    },
    {
      "epoch": 0.9905006654123262,
      "grad_norm": 0.46089810132980347,
      "learning_rate": 7.725508119229863e-06,
      "loss": 0.4388,
      "step": 21584
    },
    {
      "epoch": 0.9905465559175807,
      "grad_norm": 0.39437952637672424,
      "learning_rate": 7.725302557772725e-06,
      "loss": 0.2912,
      "step": 21585
    },
    {
      "epoch": 0.9905924464228351,
      "grad_norm": 0.4594399631023407,
      "learning_rate": 7.725096989762094e-06,
      "loss": 0.4131,
      "step": 21586
    },
    {
      "epoch": 0.9906383369280896,
      "grad_norm": 0.5201370120048523,
      "learning_rate": 7.724891415198466e-06,
      "loss": 0.3625,
      "step": 21587
    },
    {
      "epoch": 0.9906842274333441,
      "grad_norm": 0.4138689935207367,
      "learning_rate": 7.724685834082332e-06,
      "loss": 0.3017,
      "step": 21588
    },
    {
      "epoch": 0.9907301179385986,
      "grad_norm": 0.44101497530937195,
      "learning_rate": 7.72448024641419e-06,
      "loss": 0.3345,
      "step": 21589
    },
    {
      "epoch": 0.9907760084438529,
      "grad_norm": 0.4594927728176117,
      "learning_rate": 7.724274652194531e-06,
      "loss": 0.4434,
      "step": 21590
    },
    {
      "epoch": 0.9908218989491074,
      "grad_norm": 0.4490070343017578,
      "learning_rate": 7.724069051423853e-06,
      "loss": 0.4037,
      "step": 21591
    },
    {
      "epoch": 0.9908677894543619,
      "grad_norm": 0.4688624143600464,
      "learning_rate": 7.723863444102648e-06,
      "loss": 0.4123,
      "step": 21592
    },
    {
      "epoch": 0.9909136799596163,
      "grad_norm": 0.44659847021102905,
      "learning_rate": 7.723657830231411e-06,
      "loss": 0.3959,
      "step": 21593
    },
    {
      "epoch": 0.9909595704648708,
      "grad_norm": 0.44296732544898987,
      "learning_rate": 7.723452209810637e-06,
      "loss": 0.3793,
      "step": 21594
    },
    {
      "epoch": 0.9910054609701253,
      "grad_norm": 0.46276307106018066,
      "learning_rate": 7.72324658284082e-06,
      "loss": 0.4222,
      "step": 21595
    },
    {
      "epoch": 0.9910513514753797,
      "grad_norm": 0.387199342250824,
      "learning_rate": 7.723040949322455e-06,
      "loss": 0.2598,
      "step": 21596
    },
    {
      "epoch": 0.9910972419806342,
      "grad_norm": 0.47061440348625183,
      "learning_rate": 7.722835309256036e-06,
      "loss": 0.3929,
      "step": 21597
    },
    {
      "epoch": 0.9911431324858887,
      "grad_norm": 0.4351453483104706,
      "learning_rate": 7.722629662642056e-06,
      "loss": 0.3659,
      "step": 21598
    },
    {
      "epoch": 0.9911890229911431,
      "grad_norm": 0.4130133092403412,
      "learning_rate": 7.72242400948101e-06,
      "loss": 0.3239,
      "step": 21599
    },
    {
      "epoch": 0.9912349134963976,
      "grad_norm": 0.41191717982292175,
      "learning_rate": 7.722218349773395e-06,
      "loss": 0.296,
      "step": 21600
    },
    {
      "epoch": 0.9912808040016521,
      "grad_norm": 0.44369596242904663,
      "learning_rate": 7.722012683519706e-06,
      "loss": 0.3722,
      "step": 21601
    },
    {
      "epoch": 0.9913266945069065,
      "grad_norm": 0.49208709597587585,
      "learning_rate": 7.721807010720432e-06,
      "loss": 0.417,
      "step": 21602
    },
    {
      "epoch": 0.991372585012161,
      "grad_norm": 0.39060091972351074,
      "learning_rate": 7.721601331376074e-06,
      "loss": 0.2928,
      "step": 21603
    },
    {
      "epoch": 0.9914184755174155,
      "grad_norm": 0.4794404208660126,
      "learning_rate": 7.721395645487122e-06,
      "loss": 0.4176,
      "step": 21604
    },
    {
      "epoch": 0.99146436602267,
      "grad_norm": 0.5412167906761169,
      "learning_rate": 7.721189953054074e-06,
      "loss": 0.4544,
      "step": 21605
    },
    {
      "epoch": 0.9915102565279243,
      "grad_norm": 0.4557233452796936,
      "learning_rate": 7.720984254077421e-06,
      "loss": 0.421,
      "step": 21606
    },
    {
      "epoch": 0.9915561470331788,
      "grad_norm": 0.4553992748260498,
      "learning_rate": 7.720778548557661e-06,
      "loss": 0.4069,
      "step": 21607
    },
    {
      "epoch": 0.9916020375384333,
      "grad_norm": 0.4262290298938751,
      "learning_rate": 7.720572836495287e-06,
      "loss": 0.3762,
      "step": 21608
    },
    {
      "epoch": 0.9916479280436877,
      "grad_norm": 0.45328226685523987,
      "learning_rate": 7.720367117890794e-06,
      "loss": 0.4046,
      "step": 21609
    },
    {
      "epoch": 0.9916938185489422,
      "grad_norm": 0.4489854574203491,
      "learning_rate": 7.720161392744677e-06,
      "loss": 0.4014,
      "step": 21610
    },
    {
      "epoch": 0.9917397090541967,
      "grad_norm": 0.4762423634529114,
      "learning_rate": 7.71995566105743e-06,
      "loss": 0.4318,
      "step": 21611
    },
    {
      "epoch": 0.9917855995594511,
      "grad_norm": 0.4956006705760956,
      "learning_rate": 7.719749922829547e-06,
      "loss": 0.4896,
      "step": 21612
    },
    {
      "epoch": 0.9918314900647056,
      "grad_norm": 0.44897791743278503,
      "learning_rate": 7.719544178061526e-06,
      "loss": 0.4264,
      "step": 21613
    },
    {
      "epoch": 0.9918773805699601,
      "grad_norm": 0.42185404896736145,
      "learning_rate": 7.719338426753859e-06,
      "loss": 0.3064,
      "step": 21614
    },
    {
      "epoch": 0.9919232710752145,
      "grad_norm": 0.4731590449810028,
      "learning_rate": 7.719132668907039e-06,
      "loss": 0.3953,
      "step": 21615
    },
    {
      "epoch": 0.991969161580469,
      "grad_norm": 0.5093813538551331,
      "learning_rate": 7.718926904521566e-06,
      "loss": 0.4764,
      "step": 21616
    },
    {
      "epoch": 0.9920150520857235,
      "grad_norm": 0.430996298789978,
      "learning_rate": 7.718721133597931e-06,
      "loss": 0.365,
      "step": 21617
    },
    {
      "epoch": 0.9920609425909779,
      "grad_norm": 0.5503207445144653,
      "learning_rate": 7.718515356136629e-06,
      "loss": 0.4755,
      "step": 21618
    },
    {
      "epoch": 0.9921068330962324,
      "grad_norm": 0.5031556487083435,
      "learning_rate": 7.718309572138157e-06,
      "loss": 0.4633,
      "step": 21619
    },
    {
      "epoch": 0.9921527236014869,
      "grad_norm": 0.4741862714290619,
      "learning_rate": 7.718103781603006e-06,
      "loss": 0.469,
      "step": 21620
    },
    {
      "epoch": 0.9921986141067413,
      "grad_norm": 0.4880513548851013,
      "learning_rate": 7.717897984531675e-06,
      "loss": 0.4062,
      "step": 21621
    },
    {
      "epoch": 0.9922445046119958,
      "grad_norm": 0.466846764087677,
      "learning_rate": 7.717692180924656e-06,
      "loss": 0.4579,
      "step": 21622
    },
    {
      "epoch": 0.9922903951172503,
      "grad_norm": 0.4392637014389038,
      "learning_rate": 7.717486370782447e-06,
      "loss": 0.3605,
      "step": 21623
    },
    {
      "epoch": 0.9923362856225048,
      "grad_norm": 0.4225007891654968,
      "learning_rate": 7.717280554105539e-06,
      "loss": 0.3875,
      "step": 21624
    },
    {
      "epoch": 0.9923821761277591,
      "grad_norm": 0.5165314674377441,
      "learning_rate": 7.71707473089443e-06,
      "loss": 0.5394,
      "step": 21625
    },
    {
      "epoch": 0.9924280666330136,
      "grad_norm": 0.47258275747299194,
      "learning_rate": 7.716868901149613e-06,
      "loss": 0.413,
      "step": 21626
    },
    {
      "epoch": 0.9924739571382681,
      "grad_norm": 0.45230138301849365,
      "learning_rate": 7.716663064871584e-06,
      "loss": 0.3946,
      "step": 21627
    },
    {
      "epoch": 0.9925198476435225,
      "grad_norm": 0.43344438076019287,
      "learning_rate": 7.71645722206084e-06,
      "loss": 0.343,
      "step": 21628
    },
    {
      "epoch": 0.992565738148777,
      "grad_norm": 0.42270031571388245,
      "learning_rate": 7.716251372717872e-06,
      "loss": 0.3445,
      "step": 21629
    },
    {
      "epoch": 0.9926116286540315,
      "grad_norm": 0.42478659749031067,
      "learning_rate": 7.716045516843175e-06,
      "loss": 0.2986,
      "step": 21630
    },
    {
      "epoch": 0.9926575191592859,
      "grad_norm": 0.4442451596260071,
      "learning_rate": 7.715839654437247e-06,
      "loss": 0.37,
      "step": 21631
    },
    {
      "epoch": 0.9927034096645404,
      "grad_norm": 0.4787028431892395,
      "learning_rate": 7.715633785500582e-06,
      "loss": 0.4284,
      "step": 21632
    },
    {
      "epoch": 0.9927493001697949,
      "grad_norm": 0.4370770752429962,
      "learning_rate": 7.715427910033675e-06,
      "loss": 0.3681,
      "step": 21633
    },
    {
      "epoch": 0.9927951906750493,
      "grad_norm": 0.44050177931785583,
      "learning_rate": 7.71522202803702e-06,
      "loss": 0.3469,
      "step": 21634
    },
    {
      "epoch": 0.9928410811803038,
      "grad_norm": 0.4654043912887573,
      "learning_rate": 7.715016139511114e-06,
      "loss": 0.3574,
      "step": 21635
    },
    {
      "epoch": 0.9928869716855583,
      "grad_norm": 0.47407266497612,
      "learning_rate": 7.71481024445645e-06,
      "loss": 0.4247,
      "step": 21636
    },
    {
      "epoch": 0.9929328621908127,
      "grad_norm": 0.4280468225479126,
      "learning_rate": 7.714604342873526e-06,
      "loss": 0.3264,
      "step": 21637
    },
    {
      "epoch": 0.9929787526960672,
      "grad_norm": 0.4646967947483063,
      "learning_rate": 7.714398434762833e-06,
      "loss": 0.4172,
      "step": 21638
    },
    {
      "epoch": 0.9930246432013217,
      "grad_norm": 0.4821687936782837,
      "learning_rate": 7.71419252012487e-06,
      "loss": 0.4618,
      "step": 21639
    },
    {
      "epoch": 0.9930705337065762,
      "grad_norm": 0.4772447347640991,
      "learning_rate": 7.71398659896013e-06,
      "loss": 0.4414,
      "step": 21640
    },
    {
      "epoch": 0.9931164242118306,
      "grad_norm": 0.4389433264732361,
      "learning_rate": 7.713780671269109e-06,
      "loss": 0.296,
      "step": 21641
    },
    {
      "epoch": 0.993162314717085,
      "grad_norm": 0.4953770041465759,
      "learning_rate": 7.713574737052303e-06,
      "loss": 0.4707,
      "step": 21642
    },
    {
      "epoch": 0.9932082052223395,
      "grad_norm": 0.41828301548957825,
      "learning_rate": 7.713368796310206e-06,
      "loss": 0.3651,
      "step": 21643
    },
    {
      "epoch": 0.9932540957275939,
      "grad_norm": 0.471941739320755,
      "learning_rate": 7.71316284904331e-06,
      "loss": 0.4027,
      "step": 21644
    },
    {
      "epoch": 0.9932999862328484,
      "grad_norm": 0.4663860499858856,
      "learning_rate": 7.712956895252117e-06,
      "loss": 0.4798,
      "step": 21645
    },
    {
      "epoch": 0.9933458767381029,
      "grad_norm": 0.4490600526332855,
      "learning_rate": 7.712750934937118e-06,
      "loss": 0.4023,
      "step": 21646
    },
    {
      "epoch": 0.9933917672433573,
      "grad_norm": 0.47742733359336853,
      "learning_rate": 7.712544968098809e-06,
      "loss": 0.4088,
      "step": 21647
    },
    {
      "epoch": 0.9934376577486118,
      "grad_norm": 0.4789804518222809,
      "learning_rate": 7.712338994737687e-06,
      "loss": 0.4748,
      "step": 21648
    },
    {
      "epoch": 0.9934835482538663,
      "grad_norm": 0.41957616806030273,
      "learning_rate": 7.712133014854245e-06,
      "loss": 0.3423,
      "step": 21649
    },
    {
      "epoch": 0.9935294387591207,
      "grad_norm": 0.4999156594276428,
      "learning_rate": 7.711927028448977e-06,
      "loss": 0.4646,
      "step": 21650
    },
    {
      "epoch": 0.9935753292643752,
      "grad_norm": 0.4107596278190613,
      "learning_rate": 7.711721035522383e-06,
      "loss": 0.3235,
      "step": 21651
    },
    {
      "epoch": 0.9936212197696297,
      "grad_norm": 0.3989337384700775,
      "learning_rate": 7.711515036074955e-06,
      "loss": 0.3136,
      "step": 21652
    },
    {
      "epoch": 0.9936671102748841,
      "grad_norm": 0.4458697736263275,
      "learning_rate": 7.711309030107188e-06,
      "loss": 0.3512,
      "step": 21653
    },
    {
      "epoch": 0.9937130007801386,
      "grad_norm": 0.4967235326766968,
      "learning_rate": 7.71110301761958e-06,
      "loss": 0.3707,
      "step": 21654
    },
    {
      "epoch": 0.9937588912853931,
      "grad_norm": 0.5026987195014954,
      "learning_rate": 7.710896998612626e-06,
      "loss": 0.5037,
      "step": 21655
    },
    {
      "epoch": 0.9938047817906475,
      "grad_norm": 0.4506838023662567,
      "learning_rate": 7.710690973086817e-06,
      "loss": 0.3704,
      "step": 21656
    },
    {
      "epoch": 0.993850672295902,
      "grad_norm": 0.4644213318824768,
      "learning_rate": 7.710484941042654e-06,
      "loss": 0.4294,
      "step": 21657
    },
    {
      "epoch": 0.9938965628011565,
      "grad_norm": 0.456689715385437,
      "learning_rate": 7.710278902480628e-06,
      "loss": 0.4328,
      "step": 21658
    },
    {
      "epoch": 0.993942453306411,
      "grad_norm": 0.4821111559867859,
      "learning_rate": 7.710072857401238e-06,
      "loss": 0.3946,
      "step": 21659
    },
    {
      "epoch": 0.9939883438116653,
      "grad_norm": 0.4732898473739624,
      "learning_rate": 7.70986680580498e-06,
      "loss": 0.4369,
      "step": 21660
    },
    {
      "epoch": 0.9940342343169198,
      "grad_norm": 0.4998220205307007,
      "learning_rate": 7.709660747692344e-06,
      "loss": 0.4792,
      "step": 21661
    },
    {
      "epoch": 0.9940801248221743,
      "grad_norm": 0.4651615619659424,
      "learning_rate": 7.709454683063832e-06,
      "loss": 0.4424,
      "step": 21662
    },
    {
      "epoch": 0.9941260153274287,
      "grad_norm": 0.4596205949783325,
      "learning_rate": 7.709248611919935e-06,
      "loss": 0.4252,
      "step": 21663
    },
    {
      "epoch": 0.9941719058326832,
      "grad_norm": 0.45824873447418213,
      "learning_rate": 7.709042534261151e-06,
      "loss": 0.3765,
      "step": 21664
    },
    {
      "epoch": 0.9942177963379377,
      "grad_norm": 0.44609710574150085,
      "learning_rate": 7.708836450087975e-06,
      "loss": 0.4265,
      "step": 21665
    },
    {
      "epoch": 0.9942636868431921,
      "grad_norm": 0.436423122882843,
      "learning_rate": 7.7086303594009e-06,
      "loss": 0.3078,
      "step": 21666
    },
    {
      "epoch": 0.9943095773484466,
      "grad_norm": 0.46959188580513,
      "learning_rate": 7.708424262200426e-06,
      "loss": 0.439,
      "step": 21667
    },
    {
      "epoch": 0.9943554678537011,
      "grad_norm": 0.43696752190589905,
      "learning_rate": 7.708218158487047e-06,
      "loss": 0.3468,
      "step": 21668
    },
    {
      "epoch": 0.9944013583589555,
      "grad_norm": 0.4352855086326599,
      "learning_rate": 7.708012048261255e-06,
      "loss": 0.3332,
      "step": 21669
    },
    {
      "epoch": 0.99444724886421,
      "grad_norm": 0.4379773437976837,
      "learning_rate": 7.707805931523552e-06,
      "loss": 0.3393,
      "step": 21670
    },
    {
      "epoch": 0.9944931393694645,
      "grad_norm": 0.4358983337879181,
      "learning_rate": 7.707599808274428e-06,
      "loss": 0.3656,
      "step": 21671
    },
    {
      "epoch": 0.9945390298747189,
      "grad_norm": 0.41573044657707214,
      "learning_rate": 7.707393678514383e-06,
      "loss": 0.3611,
      "step": 21672
    },
    {
      "epoch": 0.9945849203799734,
      "grad_norm": 0.434416264295578,
      "learning_rate": 7.707187542243909e-06,
      "loss": 0.3348,
      "step": 21673
    },
    {
      "epoch": 0.9946308108852279,
      "grad_norm": 0.5044904947280884,
      "learning_rate": 7.706981399463505e-06,
      "loss": 0.4103,
      "step": 21674
    },
    {
      "epoch": 0.9946767013904823,
      "grad_norm": 0.46483391523361206,
      "learning_rate": 7.706775250173663e-06,
      "loss": 0.4008,
      "step": 21675
    },
    {
      "epoch": 0.9947225918957368,
      "grad_norm": 0.40895169973373413,
      "learning_rate": 7.706569094374883e-06,
      "loss": 0.3169,
      "step": 21676
    },
    {
      "epoch": 0.9947684824009912,
      "grad_norm": 0.4912654459476471,
      "learning_rate": 7.706362932067656e-06,
      "loss": 0.4795,
      "step": 21677
    },
    {
      "epoch": 0.9948143729062457,
      "grad_norm": 0.44605332612991333,
      "learning_rate": 7.706156763252484e-06,
      "loss": 0.3415,
      "step": 21678
    },
    {
      "epoch": 0.9948602634115001,
      "grad_norm": 0.4418478012084961,
      "learning_rate": 7.705950587929855e-06,
      "loss": 0.3813,
      "step": 21679
    },
    {
      "epoch": 0.9949061539167546,
      "grad_norm": 0.5477805137634277,
      "learning_rate": 7.705744406100272e-06,
      "loss": 0.4568,
      "step": 21680
    },
    {
      "epoch": 0.9949520444220091,
      "grad_norm": 0.47867539525032043,
      "learning_rate": 7.705538217764226e-06,
      "loss": 0.4426,
      "step": 21681
    },
    {
      "epoch": 0.9949979349272635,
      "grad_norm": 0.4280222952365875,
      "learning_rate": 7.705332022922216e-06,
      "loss": 0.308,
      "step": 21682
    },
    {
      "epoch": 0.995043825432518,
      "grad_norm": 0.45183125138282776,
      "learning_rate": 7.705125821574734e-06,
      "loss": 0.3823,
      "step": 21683
    },
    {
      "epoch": 0.9950897159377725,
      "grad_norm": 0.4185204803943634,
      "learning_rate": 7.704919613722279e-06,
      "loss": 0.3081,
      "step": 21684
    },
    {
      "epoch": 0.9951356064430269,
      "grad_norm": 0.43980467319488525,
      "learning_rate": 7.704713399365347e-06,
      "loss": 0.3804,
      "step": 21685
    },
    {
      "epoch": 0.9951814969482814,
      "grad_norm": 0.44416001439094543,
      "learning_rate": 7.704507178504433e-06,
      "loss": 0.4303,
      "step": 21686
    },
    {
      "epoch": 0.9952273874535359,
      "grad_norm": 0.42041924595832825,
      "learning_rate": 7.70430095114003e-06,
      "loss": 0.3332,
      "step": 21687
    },
    {
      "epoch": 0.9952732779587903,
      "grad_norm": 0.9331541061401367,
      "learning_rate": 7.704094717272641e-06,
      "loss": 0.3972,
      "step": 21688
    },
    {
      "epoch": 0.9953191684640448,
      "grad_norm": 0.47611746191978455,
      "learning_rate": 7.703888476902754e-06,
      "loss": 0.4071,
      "step": 21689
    },
    {
      "epoch": 0.9953650589692993,
      "grad_norm": 0.4737643599510193,
      "learning_rate": 7.703682230030872e-06,
      "loss": 0.461,
      "step": 21690
    },
    {
      "epoch": 0.9954109494745537,
      "grad_norm": 0.427826464176178,
      "learning_rate": 7.703475976657484e-06,
      "loss": 0.3398,
      "step": 21691
    },
    {
      "epoch": 0.9954568399798082,
      "grad_norm": 0.4618673622608185,
      "learning_rate": 7.703269716783092e-06,
      "loss": 0.4115,
      "step": 21692
    },
    {
      "epoch": 0.9955027304850627,
      "grad_norm": 0.46752244234085083,
      "learning_rate": 7.703063450408189e-06,
      "loss": 0.4221,
      "step": 21693
    },
    {
      "epoch": 0.9955486209903172,
      "grad_norm": 0.42925283312797546,
      "learning_rate": 7.702857177533269e-06,
      "loss": 0.3596,
      "step": 21694
    },
    {
      "epoch": 0.9955945114955715,
      "grad_norm": 0.45623883605003357,
      "learning_rate": 7.702650898158834e-06,
      "loss": 0.384,
      "step": 21695
    },
    {
      "epoch": 0.995640402000826,
      "grad_norm": 0.4588542580604553,
      "learning_rate": 7.702444612285375e-06,
      "loss": 0.3704,
      "step": 21696
    },
    {
      "epoch": 0.9956862925060805,
      "grad_norm": 0.4644734859466553,
      "learning_rate": 7.702238319913392e-06,
      "loss": 0.4068,
      "step": 21697
    },
    {
      "epoch": 0.9957321830113349,
      "grad_norm": 0.42213189601898193,
      "learning_rate": 7.702032021043374e-06,
      "loss": 0.3388,
      "step": 21698
    },
    {
      "epoch": 0.9957780735165894,
      "grad_norm": 0.4811408519744873,
      "learning_rate": 7.701825715675826e-06,
      "loss": 0.4573,
      "step": 21699
    },
    {
      "epoch": 0.9958239640218439,
      "grad_norm": 0.44757407903671265,
      "learning_rate": 7.701619403811238e-06,
      "loss": 0.4053,
      "step": 21700
    },
    {
      "epoch": 0.9958698545270983,
      "grad_norm": 0.4473450481891632,
      "learning_rate": 7.701413085450106e-06,
      "loss": 0.3848,
      "step": 21701
    },
    {
      "epoch": 0.9959157450323528,
      "grad_norm": 0.4346800148487091,
      "learning_rate": 7.701206760592932e-06,
      "loss": 0.3787,
      "step": 21702
    },
    {
      "epoch": 0.9959616355376073,
      "grad_norm": 0.5073715448379517,
      "learning_rate": 7.701000429240207e-06,
      "loss": 0.3921,
      "step": 21703
    },
    {
      "epoch": 0.9960075260428617,
      "grad_norm": 0.5634883046150208,
      "learning_rate": 7.700794091392427e-06,
      "loss": 0.409,
      "step": 21704
    },
    {
      "epoch": 0.9960534165481162,
      "grad_norm": 0.42938217520713806,
      "learning_rate": 7.700587747050091e-06,
      "loss": 0.323,
      "step": 21705
    },
    {
      "epoch": 0.9960993070533707,
      "grad_norm": 0.47685715556144714,
      "learning_rate": 7.700381396213693e-06,
      "loss": 0.3688,
      "step": 21706
    },
    {
      "epoch": 0.9961451975586251,
      "grad_norm": 0.4331689178943634,
      "learning_rate": 7.70017503888373e-06,
      "loss": 0.3075,
      "step": 21707
    },
    {
      "epoch": 0.9961910880638796,
      "grad_norm": 0.4987686574459076,
      "learning_rate": 7.699968675060698e-06,
      "loss": 0.4481,
      "step": 21708
    },
    {
      "epoch": 0.9962369785691341,
      "grad_norm": 0.4822427034378052,
      "learning_rate": 7.699762304745095e-06,
      "loss": 0.4288,
      "step": 21709
    },
    {
      "epoch": 0.9962828690743885,
      "grad_norm": 0.47968757152557373,
      "learning_rate": 7.699555927937416e-06,
      "loss": 0.4152,
      "step": 21710
    },
    {
      "epoch": 0.996328759579643,
      "grad_norm": 0.4196031391620636,
      "learning_rate": 7.699349544638155e-06,
      "loss": 0.3659,
      "step": 21711
    },
    {
      "epoch": 0.9963746500848975,
      "grad_norm": 0.4223105013370514,
      "learning_rate": 7.69914315484781e-06,
      "loss": 0.3705,
      "step": 21712
    },
    {
      "epoch": 0.996420540590152,
      "grad_norm": 0.46879667043685913,
      "learning_rate": 7.69893675856688e-06,
      "loss": 0.4012,
      "step": 21713
    },
    {
      "epoch": 0.9964664310954063,
      "grad_norm": 0.4587671458721161,
      "learning_rate": 7.698730355795858e-06,
      "loss": 0.3977,
      "step": 21714
    },
    {
      "epoch": 0.9965123216006608,
      "grad_norm": 0.4594936966896057,
      "learning_rate": 7.698523946535241e-06,
      "loss": 0.381,
      "step": 21715
    },
    {
      "epoch": 0.9965582121059153,
      "grad_norm": 0.4323621094226837,
      "learning_rate": 7.698317530785524e-06,
      "loss": 0.3582,
      "step": 21716
    },
    {
      "epoch": 0.9966041026111697,
      "grad_norm": 0.43070316314697266,
      "learning_rate": 7.698111108547208e-06,
      "loss": 0.3657,
      "step": 21717
    },
    {
      "epoch": 0.9966499931164242,
      "grad_norm": 0.4744878113269806,
      "learning_rate": 7.697904679820784e-06,
      "loss": 0.4306,
      "step": 21718
    },
    {
      "epoch": 0.9966958836216787,
      "grad_norm": 0.5006880164146423,
      "learning_rate": 7.697698244606751e-06,
      "loss": 0.5074,
      "step": 21719
    },
    {
      "epoch": 0.9967417741269331,
      "grad_norm": 0.44790035486221313,
      "learning_rate": 7.697491802905606e-06,
      "loss": 0.3931,
      "step": 21720
    },
    {
      "epoch": 0.9967876646321876,
      "grad_norm": 0.44855648279190063,
      "learning_rate": 7.697285354717844e-06,
      "loss": 0.3575,
      "step": 21721
    },
    {
      "epoch": 0.9968335551374421,
      "grad_norm": 0.44482141733169556,
      "learning_rate": 7.697078900043963e-06,
      "loss": 0.4326,
      "step": 21722
    },
    {
      "epoch": 0.9968794456426965,
      "grad_norm": 0.4320143759250641,
      "learning_rate": 7.696872438884458e-06,
      "loss": 0.3614,
      "step": 21723
    },
    {
      "epoch": 0.996925336147951,
      "grad_norm": 0.5439617037773132,
      "learning_rate": 7.696665971239825e-06,
      "loss": 0.4861,
      "step": 21724
    },
    {
      "epoch": 0.9969712266532055,
      "grad_norm": 0.46905937790870667,
      "learning_rate": 7.696459497110561e-06,
      "loss": 0.4173,
      "step": 21725
    },
    {
      "epoch": 0.9970171171584599,
      "grad_norm": 0.44150659441947937,
      "learning_rate": 7.696253016497165e-06,
      "loss": 0.3674,
      "step": 21726
    },
    {
      "epoch": 0.9970630076637144,
      "grad_norm": 0.4883023202419281,
      "learning_rate": 7.69604652940013e-06,
      "loss": 0.4249,
      "step": 21727
    },
    {
      "epoch": 0.9971088981689689,
      "grad_norm": 0.47890734672546387,
      "learning_rate": 7.695840035819956e-06,
      "loss": 0.4406,
      "step": 21728
    },
    {
      "epoch": 0.9971547886742234,
      "grad_norm": 0.5204247832298279,
      "learning_rate": 7.695633535757135e-06,
      "loss": 0.4554,
      "step": 21729
    },
    {
      "epoch": 0.9972006791794777,
      "grad_norm": 0.5043074488639832,
      "learning_rate": 7.695427029212165e-06,
      "loss": 0.5133,
      "step": 21730
    },
    {
      "epoch": 0.9972465696847322,
      "grad_norm": 0.44892171025276184,
      "learning_rate": 7.695220516185548e-06,
      "loss": 0.422,
      "step": 21731
    },
    {
      "epoch": 0.9972924601899867,
      "grad_norm": 0.44195860624313354,
      "learning_rate": 7.695013996677774e-06,
      "loss": 0.4242,
      "step": 21732
    },
    {
      "epoch": 0.9973383506952411,
      "grad_norm": 0.48590004444122314,
      "learning_rate": 7.694807470689341e-06,
      "loss": 0.4726,
      "step": 21733
    },
    {
      "epoch": 0.9973842412004956,
      "grad_norm": 0.4636612832546234,
      "learning_rate": 7.694600938220748e-06,
      "loss": 0.3646,
      "step": 21734
    },
    {
      "epoch": 0.9974301317057501,
      "grad_norm": 0.48089244961738586,
      "learning_rate": 7.694394399272489e-06,
      "loss": 0.4993,
      "step": 21735
    },
    {
      "epoch": 0.9974760222110045,
      "grad_norm": 0.4314274787902832,
      "learning_rate": 7.694187853845063e-06,
      "loss": 0.3645,
      "step": 21736
    },
    {
      "epoch": 0.997521912716259,
      "grad_norm": 0.6634402871131897,
      "learning_rate": 7.693981301938966e-06,
      "loss": 0.3954,
      "step": 21737
    },
    {
      "epoch": 0.9975678032215135,
      "grad_norm": 0.47214239835739136,
      "learning_rate": 7.693774743554691e-06,
      "loss": 0.3893,
      "step": 21738
    },
    {
      "epoch": 0.9976136937267679,
      "grad_norm": 0.46950823068618774,
      "learning_rate": 7.693568178692742e-06,
      "loss": 0.422,
      "step": 21739
    },
    {
      "epoch": 0.9976595842320224,
      "grad_norm": 0.46035197377204895,
      "learning_rate": 7.69336160735361e-06,
      "loss": 0.3882,
      "step": 21740
    },
    {
      "epoch": 0.9977054747372769,
      "grad_norm": 0.474260151386261,
      "learning_rate": 7.693155029537794e-06,
      "loss": 0.4222,
      "step": 21741
    },
    {
      "epoch": 0.9977513652425313,
      "grad_norm": 0.45628631114959717,
      "learning_rate": 7.692948445245788e-06,
      "loss": 0.3784,
      "step": 21742
    },
    {
      "epoch": 0.9977972557477858,
      "grad_norm": 0.41917580366134644,
      "learning_rate": 7.692741854478091e-06,
      "loss": 0.3801,
      "step": 21743
    },
    {
      "epoch": 0.9978431462530403,
      "grad_norm": 0.4488079845905304,
      "learning_rate": 7.692535257235202e-06,
      "loss": 0.4226,
      "step": 21744
    },
    {
      "epoch": 0.9978890367582947,
      "grad_norm": 0.39841407537460327,
      "learning_rate": 7.692328653517614e-06,
      "loss": 0.2944,
      "step": 21745
    },
    {
      "epoch": 0.9979349272635492,
      "grad_norm": 0.4876246452331543,
      "learning_rate": 7.692122043325824e-06,
      "loss": 0.4891,
      "step": 21746
    },
    {
      "epoch": 0.9979808177688037,
      "grad_norm": 0.46109312772750854,
      "learning_rate": 7.691915426660334e-06,
      "loss": 0.3978,
      "step": 21747
    },
    {
      "epoch": 0.9980267082740581,
      "grad_norm": 0.5009638071060181,
      "learning_rate": 7.691708803521635e-06,
      "loss": 0.4096,
      "step": 21748
    },
    {
      "epoch": 0.9980725987793125,
      "grad_norm": 0.4462440609931946,
      "learning_rate": 7.691502173910226e-06,
      "loss": 0.3927,
      "step": 21749
    },
    {
      "epoch": 0.998118489284567,
      "grad_norm": 0.4259359836578369,
      "learning_rate": 7.691295537826604e-06,
      "loss": 0.3148,
      "step": 21750
    },
    {
      "epoch": 0.9981643797898215,
      "grad_norm": 0.46323710680007935,
      "learning_rate": 7.691088895271264e-06,
      "loss": 0.4048,
      "step": 21751
    },
    {
      "epoch": 0.9982102702950759,
      "grad_norm": 0.4062822461128235,
      "learning_rate": 7.690882246244706e-06,
      "loss": 0.3231,
      "step": 21752
    },
    {
      "epoch": 0.9982561608003304,
      "grad_norm": 0.45418989658355713,
      "learning_rate": 7.690675590747426e-06,
      "loss": 0.3882,
      "step": 21753
    },
    {
      "epoch": 0.9983020513055849,
      "grad_norm": 0.48705634474754333,
      "learning_rate": 7.69046892877992e-06,
      "loss": 0.4736,
      "step": 21754
    },
    {
      "epoch": 0.9983479418108393,
      "grad_norm": 0.462112694978714,
      "learning_rate": 7.690262260342682e-06,
      "loss": 0.366,
      "step": 21755
    },
    {
      "epoch": 0.9983938323160938,
      "grad_norm": 0.4815134108066559,
      "learning_rate": 7.690055585436216e-06,
      "loss": 0.4584,
      "step": 21756
    },
    {
      "epoch": 0.9984397228213483,
      "grad_norm": 0.4438479542732239,
      "learning_rate": 7.689848904061014e-06,
      "loss": 0.4083,
      "step": 21757
    },
    {
      "epoch": 0.9984856133266027,
      "grad_norm": 0.4772718548774719,
      "learning_rate": 7.689642216217577e-06,
      "loss": 0.4697,
      "step": 21758
    },
    {
      "epoch": 0.9985315038318572,
      "grad_norm": 0.45480525493621826,
      "learning_rate": 7.689435521906396e-06,
      "loss": 0.3954,
      "step": 21759
    },
    {
      "epoch": 0.9985773943371117,
      "grad_norm": 0.430342435836792,
      "learning_rate": 7.689228821127973e-06,
      "loss": 0.3789,
      "step": 21760
    },
    {
      "epoch": 0.9986232848423661,
      "grad_norm": 0.47365885972976685,
      "learning_rate": 7.689022113882803e-06,
      "loss": 0.4528,
      "step": 21761
    },
    {
      "epoch": 0.9986691753476206,
      "grad_norm": 0.46474674344062805,
      "learning_rate": 7.688815400171383e-06,
      "loss": 0.4082,
      "step": 21762
    },
    {
      "epoch": 0.9987150658528751,
      "grad_norm": 0.5705292224884033,
      "learning_rate": 7.688608679994211e-06,
      "loss": 0.4184,
      "step": 21763
    },
    {
      "epoch": 0.9987609563581294,
      "grad_norm": 0.46210139989852905,
      "learning_rate": 7.688401953351784e-06,
      "loss": 0.407,
      "step": 21764
    },
    {
      "epoch": 0.998806846863384,
      "grad_norm": 0.456602543592453,
      "learning_rate": 7.688195220244599e-06,
      "loss": 0.3848,
      "step": 21765
    },
    {
      "epoch": 0.9988527373686384,
      "grad_norm": 0.45093870162963867,
      "learning_rate": 7.687988480673151e-06,
      "loss": 0.3776,
      "step": 21766
    },
    {
      "epoch": 0.9988986278738929,
      "grad_norm": 0.446113646030426,
      "learning_rate": 7.687781734637942e-06,
      "loss": 0.3999,
      "step": 21767
    },
    {
      "epoch": 0.9989445183791473,
      "grad_norm": 0.47175708413124084,
      "learning_rate": 7.687574982139464e-06,
      "loss": 0.4173,
      "step": 21768
    },
    {
      "epoch": 0.9989904088844018,
      "grad_norm": 0.4484696090221405,
      "learning_rate": 7.687368223178217e-06,
      "loss": 0.3678,
      "step": 21769
    },
    {
      "epoch": 0.9990362993896563,
      "grad_norm": 0.4323655366897583,
      "learning_rate": 7.687161457754699e-06,
      "loss": 0.3768,
      "step": 21770
    },
    {
      "epoch": 0.9990821898949107,
      "grad_norm": 0.4674985408782959,
      "learning_rate": 7.686954685869403e-06,
      "loss": 0.4292,
      "step": 21771
    },
    {
      "epoch": 0.9991280804001652,
      "grad_norm": 0.4495973289012909,
      "learning_rate": 7.686747907522832e-06,
      "loss": 0.4136,
      "step": 21772
    },
    {
      "epoch": 0.9991739709054197,
      "grad_norm": 0.4574570953845978,
      "learning_rate": 7.68654112271548e-06,
      "loss": 0.3973,
      "step": 21773
    },
    {
      "epoch": 0.9992198614106741,
      "grad_norm": 0.4700450003147125,
      "learning_rate": 7.686334331447842e-06,
      "loss": 0.3816,
      "step": 21774
    },
    {
      "epoch": 0.9992657519159286,
      "grad_norm": 0.46827003359794617,
      "learning_rate": 7.68612753372042e-06,
      "loss": 0.437,
      "step": 21775
    },
    {
      "epoch": 0.9993116424211831,
      "grad_norm": 0.4525164067745209,
      "learning_rate": 7.685920729533708e-06,
      "loss": 0.3627,
      "step": 21776
    },
    {
      "epoch": 0.9993575329264375,
      "grad_norm": 0.4704478085041046,
      "learning_rate": 7.685713918888203e-06,
      "loss": 0.3883,
      "step": 21777
    },
    {
      "epoch": 0.999403423431692,
      "grad_norm": 0.5542486310005188,
      "learning_rate": 7.685507101784406e-06,
      "loss": 0.5781,
      "step": 21778
    },
    {
      "epoch": 0.9994493139369465,
      "grad_norm": 0.4475703835487366,
      "learning_rate": 7.685300278222809e-06,
      "loss": 0.3699,
      "step": 21779
    },
    {
      "epoch": 0.9994952044422009,
      "grad_norm": 0.42029353976249695,
      "learning_rate": 7.685093448203916e-06,
      "loss": 0.3214,
      "step": 21780
    },
    {
      "epoch": 0.9995410949474554,
      "grad_norm": 0.48289719223976135,
      "learning_rate": 7.68488661172822e-06,
      "loss": 0.4012,
      "step": 21781
    },
    {
      "epoch": 0.9995869854527099,
      "grad_norm": 0.5010626316070557,
      "learning_rate": 7.684679768796218e-06,
      "loss": 0.486,
      "step": 21782
    },
    {
      "epoch": 0.9996328759579644,
      "grad_norm": 0.4751811623573303,
      "learning_rate": 7.684472919408408e-06,
      "loss": 0.4817,
      "step": 21783
    },
    {
      "epoch": 0.9996787664632187,
      "grad_norm": 0.428971529006958,
      "learning_rate": 7.684266063565288e-06,
      "loss": 0.3478,
      "step": 21784
    },
    {
      "epoch": 0.9997246569684732,
      "grad_norm": 0.5563684105873108,
      "learning_rate": 7.684059201267357e-06,
      "loss": 0.478,
      "step": 21785
    },
    {
      "epoch": 0.9997705474737277,
      "grad_norm": 0.4718800485134125,
      "learning_rate": 7.683852332515108e-06,
      "loss": 0.4161,
      "step": 21786
    },
    {
      "epoch": 0.9998164379789821,
      "grad_norm": 0.4454321563243866,
      "learning_rate": 7.683645457309043e-06,
      "loss": 0.3814,
      "step": 21787
    },
    {
      "epoch": 0.9998623284842366,
      "grad_norm": 0.5012182593345642,
      "learning_rate": 7.683438575649658e-06,
      "loss": 0.5173,
      "step": 21788
    },
    {
      "epoch": 0.9999082189894911,
      "grad_norm": 0.5023024678230286,
      "learning_rate": 7.683231687537449e-06,
      "loss": 0.4572,
      "step": 21789
    },
    {
      "epoch": 0.9999541094947455,
      "grad_norm": 0.40421074628829956,
      "learning_rate": 7.683024792972916e-06,
      "loss": 0.3074,
      "step": 21790
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.4496239125728607,
      "learning_rate": 7.682817891956553e-06,
      "loss": 0.3598,
      "step": 21791
    },
    {
      "epoch": 1.0000458905052545,
      "grad_norm": 0.42824703454971313,
      "learning_rate": 7.682610984488863e-06,
      "loss": 0.2908,
      "step": 21792
    },
    {
      "epoch": 1.000091781010509,
      "grad_norm": 0.47233837842941284,
      "learning_rate": 7.682404070570337e-06,
      "loss": 0.352,
      "step": 21793
    },
    {
      "epoch": 1.0001376715157635,
      "grad_norm": 0.509062647819519,
      "learning_rate": 7.682197150201475e-06,
      "loss": 0.3996,
      "step": 21794
    },
    {
      "epoch": 1.0001835620210178,
      "grad_norm": 0.4523948132991791,
      "learning_rate": 7.68199022338278e-06,
      "loss": 0.3485,
      "step": 21795
    },
    {
      "epoch": 1.0002294525262723,
      "grad_norm": 0.47760623693466187,
      "learning_rate": 7.68178329011474e-06,
      "loss": 0.3841,
      "step": 21796
    },
    {
      "epoch": 1.0002753430315268,
      "grad_norm": 0.44862669706344604,
      "learning_rate": 7.681576350397861e-06,
      "loss": 0.3152,
      "step": 21797
    },
    {
      "epoch": 1.0003212335367813,
      "grad_norm": 0.45125558972358704,
      "learning_rate": 7.681369404232638e-06,
      "loss": 0.3284,
      "step": 21798
    },
    {
      "epoch": 1.0003671240420358,
      "grad_norm": 0.457284539937973,
      "learning_rate": 7.681162451619564e-06,
      "loss": 0.3306,
      "step": 21799
    },
    {
      "epoch": 1.0004130145472903,
      "grad_norm": 0.48717260360717773,
      "learning_rate": 7.680955492559144e-06,
      "loss": 0.3977,
      "step": 21800
    },
    {
      "epoch": 1.0004589050525445,
      "grad_norm": 0.4457889795303345,
      "learning_rate": 7.68074852705187e-06,
      "loss": 0.3105,
      "step": 21801
    },
    {
      "epoch": 1.000504795557799,
      "grad_norm": 0.4630339443683624,
      "learning_rate": 7.680541555098244e-06,
      "loss": 0.2963,
      "step": 21802
    },
    {
      "epoch": 1.0005506860630535,
      "grad_norm": 0.5100385546684265,
      "learning_rate": 7.68033457669876e-06,
      "loss": 0.3438,
      "step": 21803
    },
    {
      "epoch": 1.000596576568308,
      "grad_norm": 0.5151930451393127,
      "learning_rate": 7.680127591853916e-06,
      "loss": 0.4002,
      "step": 21804
    },
    {
      "epoch": 1.0006424670735625,
      "grad_norm": 0.4654126465320587,
      "learning_rate": 7.679920600564212e-06,
      "loss": 0.3522,
      "step": 21805
    },
    {
      "epoch": 1.000688357578817,
      "grad_norm": 0.48202213644981384,
      "learning_rate": 7.679713602830146e-06,
      "loss": 0.3588,
      "step": 21806
    },
    {
      "epoch": 1.0007342480840713,
      "grad_norm": 0.47520971298217773,
      "learning_rate": 7.679506598652214e-06,
      "loss": 0.389,
      "step": 21807
    },
    {
      "epoch": 1.0007801385893258,
      "grad_norm": 0.44335997104644775,
      "learning_rate": 7.679299588030914e-06,
      "loss": 0.2935,
      "step": 21808
    },
    {
      "epoch": 1.0008260290945803,
      "grad_norm": 0.4476031959056854,
      "learning_rate": 7.679092570966745e-06,
      "loss": 0.2821,
      "step": 21809
    },
    {
      "epoch": 1.0008719195998348,
      "grad_norm": 0.4472014009952545,
      "learning_rate": 7.678885547460202e-06,
      "loss": 0.3017,
      "step": 21810
    },
    {
      "epoch": 1.0009178101050893,
      "grad_norm": 0.45240941643714905,
      "learning_rate": 7.678678517511786e-06,
      "loss": 0.3034,
      "step": 21811
    },
    {
      "epoch": 1.0009637006103438,
      "grad_norm": 0.4844726026058197,
      "learning_rate": 7.678471481121994e-06,
      "loss": 0.3801,
      "step": 21812
    },
    {
      "epoch": 1.0010095911155983,
      "grad_norm": 0.4395175576210022,
      "learning_rate": 7.678264438291322e-06,
      "loss": 0.3341,
      "step": 21813
    },
    {
      "epoch": 1.0010554816208526,
      "grad_norm": 0.480497270822525,
      "learning_rate": 7.67805738902027e-06,
      "loss": 0.3875,
      "step": 21814
    },
    {
      "epoch": 1.001101372126107,
      "grad_norm": 0.724362313747406,
      "learning_rate": 7.677850333309336e-06,
      "loss": 0.3151,
      "step": 21815
    },
    {
      "epoch": 1.0011472626313616,
      "grad_norm": 0.436196893453598,
      "learning_rate": 7.677643271159016e-06,
      "loss": 0.2892,
      "step": 21816
    },
    {
      "epoch": 1.001193153136616,
      "grad_norm": 0.4740312695503235,
      "learning_rate": 7.677436202569809e-06,
      "loss": 0.3879,
      "step": 21817
    },
    {
      "epoch": 1.0012390436418706,
      "grad_norm": 0.431088924407959,
      "learning_rate": 7.677229127542214e-06,
      "loss": 0.2956,
      "step": 21818
    },
    {
      "epoch": 1.001284934147125,
      "grad_norm": 0.5393555164337158,
      "learning_rate": 7.677022046076727e-06,
      "loss": 0.449,
      "step": 21819
    },
    {
      "epoch": 1.0013308246523793,
      "grad_norm": 0.44642388820648193,
      "learning_rate": 7.676814958173847e-06,
      "loss": 0.302,
      "step": 21820
    },
    {
      "epoch": 1.0013767151576338,
      "grad_norm": 0.4677377939224243,
      "learning_rate": 7.676607863834072e-06,
      "loss": 0.349,
      "step": 21821
    },
    {
      "epoch": 1.0014226056628883,
      "grad_norm": 0.5177029371261597,
      "learning_rate": 7.676400763057901e-06,
      "loss": 0.3671,
      "step": 21822
    },
    {
      "epoch": 1.0014684961681428,
      "grad_norm": 0.503667414188385,
      "learning_rate": 7.676193655845829e-06,
      "loss": 0.435,
      "step": 21823
    },
    {
      "epoch": 1.0015143866733973,
      "grad_norm": 0.46695178747177124,
      "learning_rate": 7.675986542198357e-06,
      "loss": 0.3465,
      "step": 21824
    },
    {
      "epoch": 1.0015602771786518,
      "grad_norm": 0.46708744764328003,
      "learning_rate": 7.67577942211598e-06,
      "loss": 0.358,
      "step": 21825
    },
    {
      "epoch": 1.001606167683906,
      "grad_norm": 0.4863606095314026,
      "learning_rate": 7.675572295599201e-06,
      "loss": 0.407,
      "step": 21826
    },
    {
      "epoch": 1.0016520581891606,
      "grad_norm": 0.44961273670196533,
      "learning_rate": 7.675365162648514e-06,
      "loss": 0.3472,
      "step": 21827
    },
    {
      "epoch": 1.001697948694415,
      "grad_norm": 0.47937706112861633,
      "learning_rate": 7.675158023264418e-06,
      "loss": 0.4133,
      "step": 21828
    },
    {
      "epoch": 1.0017438391996696,
      "grad_norm": 0.46879705786705017,
      "learning_rate": 7.674950877447412e-06,
      "loss": 0.3555,
      "step": 21829
    },
    {
      "epoch": 1.001789729704924,
      "grad_norm": 0.4375566840171814,
      "learning_rate": 7.67474372519799e-06,
      "loss": 0.3088,
      "step": 21830
    },
    {
      "epoch": 1.0018356202101786,
      "grad_norm": 0.4670071601867676,
      "learning_rate": 7.674536566516658e-06,
      "loss": 0.3923,
      "step": 21831
    },
    {
      "epoch": 1.001881510715433,
      "grad_norm": 0.5112287402153015,
      "learning_rate": 7.674329401403906e-06,
      "loss": 0.4083,
      "step": 21832
    },
    {
      "epoch": 1.0019274012206874,
      "grad_norm": 0.4465145170688629,
      "learning_rate": 7.674122229860237e-06,
      "loss": 0.327,
      "step": 21833
    },
    {
      "epoch": 1.0019732917259419,
      "grad_norm": 0.4728211760520935,
      "learning_rate": 7.673915051886149e-06,
      "loss": 0.3607,
      "step": 21834
    },
    {
      "epoch": 1.0020191822311963,
      "grad_norm": 0.4644700288772583,
      "learning_rate": 7.67370786748214e-06,
      "loss": 0.3703,
      "step": 21835
    },
    {
      "epoch": 1.0020650727364508,
      "grad_norm": 0.4294247329235077,
      "learning_rate": 7.673500676648706e-06,
      "loss": 0.3106,
      "step": 21836
    },
    {
      "epoch": 1.0021109632417053,
      "grad_norm": 0.46702876687049866,
      "learning_rate": 7.673293479386346e-06,
      "loss": 0.383,
      "step": 21837
    },
    {
      "epoch": 1.0021568537469598,
      "grad_norm": 0.4914146363735199,
      "learning_rate": 7.67308627569556e-06,
      "loss": 0.3599,
      "step": 21838
    },
    {
      "epoch": 1.0022027442522141,
      "grad_norm": 0.4601387083530426,
      "learning_rate": 7.672879065576843e-06,
      "loss": 0.3416,
      "step": 21839
    },
    {
      "epoch": 1.0022486347574686,
      "grad_norm": 0.547964870929718,
      "learning_rate": 7.672671849030699e-06,
      "loss": 0.3251,
      "step": 21840
    },
    {
      "epoch": 1.0022945252627231,
      "grad_norm": 0.49092257022857666,
      "learning_rate": 7.672464626057621e-06,
      "loss": 0.3638,
      "step": 21841
    },
    {
      "epoch": 1.0023404157679776,
      "grad_norm": 0.48189881443977356,
      "learning_rate": 7.672257396658108e-06,
      "loss": 0.3993,
      "step": 21842
    },
    {
      "epoch": 1.002386306273232,
      "grad_norm": 0.5036380290985107,
      "learning_rate": 7.67205016083266e-06,
      "loss": 0.4284,
      "step": 21843
    },
    {
      "epoch": 1.0024321967784866,
      "grad_norm": 0.49139124155044556,
      "learning_rate": 7.671842918581776e-06,
      "loss": 0.3726,
      "step": 21844
    },
    {
      "epoch": 1.002478087283741,
      "grad_norm": 0.4567021429538727,
      "learning_rate": 7.671635669905953e-06,
      "loss": 0.3115,
      "step": 21845
    },
    {
      "epoch": 1.0025239777889954,
      "grad_norm": 0.4727135896682739,
      "learning_rate": 7.671428414805689e-06,
      "loss": 0.31,
      "step": 21846
    },
    {
      "epoch": 1.0025698682942499,
      "grad_norm": 0.44256845116615295,
      "learning_rate": 7.67122115328148e-06,
      "loss": 0.2916,
      "step": 21847
    },
    {
      "epoch": 1.0026157587995044,
      "grad_norm": 0.5284841060638428,
      "learning_rate": 7.67101388533383e-06,
      "loss": 0.4183,
      "step": 21848
    },
    {
      "epoch": 1.0026616493047589,
      "grad_norm": 0.4531751573085785,
      "learning_rate": 7.670806610963234e-06,
      "loss": 0.3172,
      "step": 21849
    },
    {
      "epoch": 1.0027075398100134,
      "grad_norm": 0.47695717215538025,
      "learning_rate": 7.67059933017019e-06,
      "loss": 0.3648,
      "step": 21850
    },
    {
      "epoch": 1.0027534303152679,
      "grad_norm": 0.46809929609298706,
      "learning_rate": 7.670392042955199e-06,
      "loss": 0.3632,
      "step": 21851
    },
    {
      "epoch": 1.0027993208205221,
      "grad_norm": 0.4728301167488098,
      "learning_rate": 7.670184749318758e-06,
      "loss": 0.3123,
      "step": 21852
    },
    {
      "epoch": 1.0028452113257766,
      "grad_norm": 0.461937814950943,
      "learning_rate": 7.669977449261366e-06,
      "loss": 0.3438,
      "step": 21853
    },
    {
      "epoch": 1.0028911018310311,
      "grad_norm": 0.5011401772499084,
      "learning_rate": 7.66977014278352e-06,
      "loss": 0.4198,
      "step": 21854
    },
    {
      "epoch": 1.0029369923362856,
      "grad_norm": 0.5030772686004639,
      "learning_rate": 7.669562829885717e-06,
      "loss": 0.4005,
      "step": 21855
    },
    {
      "epoch": 1.0029828828415401,
      "grad_norm": 0.5199286341667175,
      "learning_rate": 7.669355510568462e-06,
      "loss": 0.3862,
      "step": 21856
    },
    {
      "epoch": 1.0030287733467946,
      "grad_norm": 0.4653865098953247,
      "learning_rate": 7.669148184832247e-06,
      "loss": 0.3161,
      "step": 21857
    },
    {
      "epoch": 1.003074663852049,
      "grad_norm": 0.47198596596717834,
      "learning_rate": 7.668940852677573e-06,
      "loss": 0.3896,
      "step": 21858
    },
    {
      "epoch": 1.0031205543573034,
      "grad_norm": 0.5021759271621704,
      "learning_rate": 7.668733514104938e-06,
      "loss": 0.4297,
      "step": 21859
    },
    {
      "epoch": 1.003166444862558,
      "grad_norm": 0.5020312666893005,
      "learning_rate": 7.668526169114843e-06,
      "loss": 0.3891,
      "step": 21860
    },
    {
      "epoch": 1.0032123353678124,
      "grad_norm": 0.46755024790763855,
      "learning_rate": 7.668318817707786e-06,
      "loss": 0.3875,
      "step": 21861
    },
    {
      "epoch": 1.003258225873067,
      "grad_norm": 0.4605557322502136,
      "learning_rate": 7.668111459884262e-06,
      "loss": 0.3456,
      "step": 21862
    },
    {
      "epoch": 1.0033041163783214,
      "grad_norm": 0.44801077246665955,
      "learning_rate": 7.667904095644773e-06,
      "loss": 0.3866,
      "step": 21863
    },
    {
      "epoch": 1.003350006883576,
      "grad_norm": 0.48443830013275146,
      "learning_rate": 7.667696724989815e-06,
      "loss": 0.2886,
      "step": 21864
    },
    {
      "epoch": 1.0033958973888302,
      "grad_norm": 0.4684072732925415,
      "learning_rate": 7.66748934791989e-06,
      "loss": 0.3328,
      "step": 21865
    },
    {
      "epoch": 1.0034417878940847,
      "grad_norm": 0.4466891884803772,
      "learning_rate": 7.667281964435495e-06,
      "loss": 0.3197,
      "step": 21866
    },
    {
      "epoch": 1.0034876783993392,
      "grad_norm": 0.469329297542572,
      "learning_rate": 7.667074574537127e-06,
      "loss": 0.3854,
      "step": 21867
    },
    {
      "epoch": 1.0035335689045937,
      "grad_norm": 0.4909505546092987,
      "learning_rate": 7.666867178225287e-06,
      "loss": 0.3736,
      "step": 21868
    },
    {
      "epoch": 1.0035794594098482,
      "grad_norm": 0.41680142283439636,
      "learning_rate": 7.666659775500476e-06,
      "loss": 0.275,
      "step": 21869
    },
    {
      "epoch": 1.0036253499151027,
      "grad_norm": 0.45755481719970703,
      "learning_rate": 7.666452366363187e-06,
      "loss": 0.3215,
      "step": 21870
    },
    {
      "epoch": 1.003671240420357,
      "grad_norm": 0.47610172629356384,
      "learning_rate": 7.666244950813923e-06,
      "loss": 0.4085,
      "step": 21871
    },
    {
      "epoch": 1.0037171309256114,
      "grad_norm": 0.46707797050476074,
      "learning_rate": 7.66603752885318e-06,
      "loss": 0.3401,
      "step": 21872
    },
    {
      "epoch": 1.003763021430866,
      "grad_norm": 0.4567943513393402,
      "learning_rate": 7.66583010048146e-06,
      "loss": 0.3355,
      "step": 21873
    },
    {
      "epoch": 1.0038089119361204,
      "grad_norm": 0.4661584794521332,
      "learning_rate": 7.665622665699256e-06,
      "loss": 0.3766,
      "step": 21874
    },
    {
      "epoch": 1.003854802441375,
      "grad_norm": 0.44932228326797485,
      "learning_rate": 7.665415224507076e-06,
      "loss": 0.2956,
      "step": 21875
    },
    {
      "epoch": 1.0039006929466294,
      "grad_norm": 0.46109265089035034,
      "learning_rate": 7.66520777690541e-06,
      "loss": 0.336,
      "step": 21876
    },
    {
      "epoch": 1.0039465834518837,
      "grad_norm": 0.4688689112663269,
      "learning_rate": 7.665000322894762e-06,
      "loss": 0.3868,
      "step": 21877
    },
    {
      "epoch": 1.0039924739571382,
      "grad_norm": 0.42947739362716675,
      "learning_rate": 7.66479286247563e-06,
      "loss": 0.3205,
      "step": 21878
    },
    {
      "epoch": 1.0040383644623927,
      "grad_norm": 0.46897774934768677,
      "learning_rate": 7.664585395648509e-06,
      "loss": 0.3143,
      "step": 21879
    },
    {
      "epoch": 1.0040842549676472,
      "grad_norm": 0.45936164259910583,
      "learning_rate": 7.664377922413905e-06,
      "loss": 0.3573,
      "step": 21880
    },
    {
      "epoch": 1.0041301454729017,
      "grad_norm": 0.511625349521637,
      "learning_rate": 7.664170442772309e-06,
      "loss": 0.4068,
      "step": 21881
    },
    {
      "epoch": 1.0041760359781562,
      "grad_norm": 0.4670616388320923,
      "learning_rate": 7.663962956724226e-06,
      "loss": 0.3756,
      "step": 21882
    },
    {
      "epoch": 1.0042219264834107,
      "grad_norm": 0.4762481451034546,
      "learning_rate": 7.663755464270156e-06,
      "loss": 0.3653,
      "step": 21883
    },
    {
      "epoch": 1.004267816988665,
      "grad_norm": 0.49966031312942505,
      "learning_rate": 7.663547965410591e-06,
      "loss": 0.3737,
      "step": 21884
    },
    {
      "epoch": 1.0043137074939195,
      "grad_norm": 0.4603038430213928,
      "learning_rate": 7.663340460146033e-06,
      "loss": 0.3465,
      "step": 21885
    },
    {
      "epoch": 1.004359597999174,
      "grad_norm": 0.48679646849632263,
      "learning_rate": 7.663132948476985e-06,
      "loss": 0.3836,
      "step": 21886
    },
    {
      "epoch": 1.0044054885044285,
      "grad_norm": 0.47332707047462463,
      "learning_rate": 7.66292543040394e-06,
      "loss": 0.3929,
      "step": 21887
    },
    {
      "epoch": 1.004451379009683,
      "grad_norm": 0.48933321237564087,
      "learning_rate": 7.6627179059274e-06,
      "loss": 0.4085,
      "step": 21888
    },
    {
      "epoch": 1.0044972695149375,
      "grad_norm": 0.4848455786705017,
      "learning_rate": 7.662510375047867e-06,
      "loss": 0.3665,
      "step": 21889
    },
    {
      "epoch": 1.0045431600201917,
      "grad_norm": 0.4562510848045349,
      "learning_rate": 7.662302837765834e-06,
      "loss": 0.3861,
      "step": 21890
    },
    {
      "epoch": 1.0045890505254462,
      "grad_norm": 0.49389615654945374,
      "learning_rate": 7.662095294081805e-06,
      "loss": 0.3625,
      "step": 21891
    },
    {
      "epoch": 1.0046349410307007,
      "grad_norm": 0.46587130427360535,
      "learning_rate": 7.661887743996275e-06,
      "loss": 0.3115,
      "step": 21892
    },
    {
      "epoch": 1.0046808315359552,
      "grad_norm": 0.4521610140800476,
      "learning_rate": 7.661680187509744e-06,
      "loss": 0.348,
      "step": 21893
    },
    {
      "epoch": 1.0047267220412097,
      "grad_norm": 0.46794214844703674,
      "learning_rate": 7.661472624622715e-06,
      "loss": 0.3497,
      "step": 21894
    },
    {
      "epoch": 1.0047726125464642,
      "grad_norm": 0.47309887409210205,
      "learning_rate": 7.661265055335683e-06,
      "loss": 0.3306,
      "step": 21895
    },
    {
      "epoch": 1.0048185030517185,
      "grad_norm": 0.4100973308086395,
      "learning_rate": 7.661057479649148e-06,
      "loss": 0.2625,
      "step": 21896
    },
    {
      "epoch": 1.004864393556973,
      "grad_norm": 0.5007053017616272,
      "learning_rate": 7.660849897563611e-06,
      "loss": 0.4348,
      "step": 21897
    },
    {
      "epoch": 1.0049102840622275,
      "grad_norm": 0.43340030312538147,
      "learning_rate": 7.660642309079569e-06,
      "loss": 0.3068,
      "step": 21898
    },
    {
      "epoch": 1.004956174567482,
      "grad_norm": 0.51333087682724,
      "learning_rate": 7.660434714197522e-06,
      "loss": 0.4471,
      "step": 21899
    },
    {
      "epoch": 1.0050020650727365,
      "grad_norm": 0.43661823868751526,
      "learning_rate": 7.66022711291797e-06,
      "loss": 0.3365,
      "step": 21900
    },
    {
      "epoch": 1.005047955577991,
      "grad_norm": 0.44621703028678894,
      "learning_rate": 7.660019505241409e-06,
      "loss": 0.333,
      "step": 21901
    },
    {
      "epoch": 1.0050938460832455,
      "grad_norm": 0.46153634786605835,
      "learning_rate": 7.659811891168344e-06,
      "loss": 0.3652,
      "step": 21902
    },
    {
      "epoch": 1.0051397365884998,
      "grad_norm": 0.4785188138484955,
      "learning_rate": 7.659604270699268e-06,
      "loss": 0.3722,
      "step": 21903
    },
    {
      "epoch": 1.0051856270937543,
      "grad_norm": 0.44658544659614563,
      "learning_rate": 7.659396643834684e-06,
      "loss": 0.3468,
      "step": 21904
    },
    {
      "epoch": 1.0052315175990088,
      "grad_norm": 0.4938446581363678,
      "learning_rate": 7.65918901057509e-06,
      "loss": 0.3762,
      "step": 21905
    },
    {
      "epoch": 1.0052774081042632,
      "grad_norm": 0.4482554793357849,
      "learning_rate": 7.658981370920986e-06,
      "loss": 0.3194,
      "step": 21906
    },
    {
      "epoch": 1.0053232986095177,
      "grad_norm": 0.4671158790588379,
      "learning_rate": 7.658773724872873e-06,
      "loss": 0.3418,
      "step": 21907
    },
    {
      "epoch": 1.0053691891147722,
      "grad_norm": 0.4733157455921173,
      "learning_rate": 7.658566072431246e-06,
      "loss": 0.3091,
      "step": 21908
    },
    {
      "epoch": 1.0054150796200265,
      "grad_norm": 0.4278928339481354,
      "learning_rate": 7.658358413596608e-06,
      "loss": 0.3066,
      "step": 21909
    },
    {
      "epoch": 1.005460970125281,
      "grad_norm": 0.45247340202331543,
      "learning_rate": 7.658150748369456e-06,
      "loss": 0.3476,
      "step": 21910
    },
    {
      "epoch": 1.0055068606305355,
      "grad_norm": 0.494045615196228,
      "learning_rate": 7.65794307675029e-06,
      "loss": 0.3504,
      "step": 21911
    },
    {
      "epoch": 1.00555275113579,
      "grad_norm": 0.46854332089424133,
      "learning_rate": 7.65773539873961e-06,
      "loss": 0.3548,
      "step": 21912
    },
    {
      "epoch": 1.0055986416410445,
      "grad_norm": 0.5320063233375549,
      "learning_rate": 7.657527714337916e-06,
      "loss": 0.4251,
      "step": 21913
    },
    {
      "epoch": 1.005644532146299,
      "grad_norm": 0.4261787235736847,
      "learning_rate": 7.657320023545705e-06,
      "loss": 0.2969,
      "step": 21914
    },
    {
      "epoch": 1.0056904226515533,
      "grad_norm": 0.4555548131465912,
      "learning_rate": 7.65711232636348e-06,
      "loss": 0.3174,
      "step": 21915
    },
    {
      "epoch": 1.0057363131568078,
      "grad_norm": 0.4771327078342438,
      "learning_rate": 7.656904622791737e-06,
      "loss": 0.3838,
      "step": 21916
    },
    {
      "epoch": 1.0057822036620623,
      "grad_norm": 0.4637584984302521,
      "learning_rate": 7.656696912830978e-06,
      "loss": 0.3391,
      "step": 21917
    },
    {
      "epoch": 1.0058280941673168,
      "grad_norm": 0.4800531268119812,
      "learning_rate": 7.656489196481701e-06,
      "loss": 0.412,
      "step": 21918
    },
    {
      "epoch": 1.0058739846725713,
      "grad_norm": 0.4903578758239746,
      "learning_rate": 7.656281473744405e-06,
      "loss": 0.4026,
      "step": 21919
    },
    {
      "epoch": 1.0059198751778258,
      "grad_norm": 0.5070230960845947,
      "learning_rate": 7.656073744619592e-06,
      "loss": 0.4128,
      "step": 21920
    },
    {
      "epoch": 1.0059657656830803,
      "grad_norm": 0.46659526228904724,
      "learning_rate": 7.65586600910776e-06,
      "loss": 0.3424,
      "step": 21921
    },
    {
      "epoch": 1.0060116561883345,
      "grad_norm": 0.42623820900917053,
      "learning_rate": 7.655658267209408e-06,
      "loss": 0.3171,
      "step": 21922
    },
    {
      "epoch": 1.006057546693589,
      "grad_norm": 0.4843745827674866,
      "learning_rate": 7.655450518925035e-06,
      "loss": 0.3687,
      "step": 21923
    },
    {
      "epoch": 1.0061034371988435,
      "grad_norm": 0.5140113830566406,
      "learning_rate": 7.655242764255142e-06,
      "loss": 0.4488,
      "step": 21924
    },
    {
      "epoch": 1.006149327704098,
      "grad_norm": 0.49371007084846497,
      "learning_rate": 7.655035003200229e-06,
      "loss": 0.3789,
      "step": 21925
    },
    {
      "epoch": 1.0061952182093525,
      "grad_norm": 0.4540064036846161,
      "learning_rate": 7.654827235760793e-06,
      "loss": 0.3261,
      "step": 21926
    },
    {
      "epoch": 1.006241108714607,
      "grad_norm": 0.4493793249130249,
      "learning_rate": 7.654619461937337e-06,
      "loss": 0.3361,
      "step": 21927
    },
    {
      "epoch": 1.0062869992198613,
      "grad_norm": 0.48265212774276733,
      "learning_rate": 7.654411681730359e-06,
      "loss": 0.4193,
      "step": 21928
    },
    {
      "epoch": 1.0063328897251158,
      "grad_norm": 0.49837997555732727,
      "learning_rate": 7.654203895140358e-06,
      "loss": 0.4,
      "step": 21929
    },
    {
      "epoch": 1.0063787802303703,
      "grad_norm": 0.45213600993156433,
      "learning_rate": 7.653996102167834e-06,
      "loss": 0.344,
      "step": 21930
    },
    {
      "epoch": 1.0064246707356248,
      "grad_norm": 0.488018661737442,
      "learning_rate": 7.65378830281329e-06,
      "loss": 0.4059,
      "step": 21931
    },
    {
      "epoch": 1.0064705612408793,
      "grad_norm": 0.4590126574039459,
      "learning_rate": 7.65358049707722e-06,
      "loss": 0.3626,
      "step": 21932
    },
    {
      "epoch": 1.0065164517461338,
      "grad_norm": 0.48965781927108765,
      "learning_rate": 7.653372684960125e-06,
      "loss": 0.3723,
      "step": 21933
    },
    {
      "epoch": 1.0065623422513883,
      "grad_norm": 0.5480174422264099,
      "learning_rate": 7.653164866462507e-06,
      "loss": 0.3568,
      "step": 21934
    },
    {
      "epoch": 1.0066082327566426,
      "grad_norm": 0.4429236948490143,
      "learning_rate": 7.652957041584868e-06,
      "loss": 0.4014,
      "step": 21935
    },
    {
      "epoch": 1.006654123261897,
      "grad_norm": 0.4574386477470398,
      "learning_rate": 7.652749210327701e-06,
      "loss": 0.3494,
      "step": 21936
    },
    {
      "epoch": 1.0067000137671516,
      "grad_norm": 0.4678099453449249,
      "learning_rate": 7.65254137269151e-06,
      "loss": 0.3413,
      "step": 21937
    },
    {
      "epoch": 1.006745904272406,
      "grad_norm": 0.4735844135284424,
      "learning_rate": 7.652333528676796e-06,
      "loss": 0.3703,
      "step": 21938
    },
    {
      "epoch": 1.0067917947776606,
      "grad_norm": 0.5432665348052979,
      "learning_rate": 7.652125678284058e-06,
      "loss": 0.4741,
      "step": 21939
    },
    {
      "epoch": 1.006837685282915,
      "grad_norm": 0.47513774037361145,
      "learning_rate": 7.65191782151379e-06,
      "loss": 0.3884,
      "step": 21940
    },
    {
      "epoch": 1.0068835757881693,
      "grad_norm": 0.45072948932647705,
      "learning_rate": 7.651709958366498e-06,
      "loss": 0.3198,
      "step": 21941
    },
    {
      "epoch": 1.0069294662934238,
      "grad_norm": 0.45452946424484253,
      "learning_rate": 7.651502088842682e-06,
      "loss": 0.3392,
      "step": 21942
    },
    {
      "epoch": 1.0069753567986783,
      "grad_norm": 0.4512861967086792,
      "learning_rate": 7.651294212942842e-06,
      "loss": 0.3279,
      "step": 21943
    },
    {
      "epoch": 1.0070212473039328,
      "grad_norm": 0.42003974318504333,
      "learning_rate": 7.651086330667473e-06,
      "loss": 0.329,
      "step": 21944
    },
    {
      "epoch": 1.0070671378091873,
      "grad_norm": 0.5066702961921692,
      "learning_rate": 7.65087844201708e-06,
      "loss": 0.4268,
      "step": 21945
    },
    {
      "epoch": 1.0071130283144418,
      "grad_norm": 0.430603563785553,
      "learning_rate": 7.65067054699216e-06,
      "loss": 0.2816,
      "step": 21946
    },
    {
      "epoch": 1.007158918819696,
      "grad_norm": 0.4638647139072418,
      "learning_rate": 7.650462645593213e-06,
      "loss": 0.37,
      "step": 21947
    },
    {
      "epoch": 1.0072048093249506,
      "grad_norm": 0.4421878159046173,
      "learning_rate": 7.650254737820742e-06,
      "loss": 0.3056,
      "step": 21948
    },
    {
      "epoch": 1.007250699830205,
      "grad_norm": 0.4775584936141968,
      "learning_rate": 7.650046823675243e-06,
      "loss": 0.4191,
      "step": 21949
    },
    {
      "epoch": 1.0072965903354596,
      "grad_norm": 0.48896414041519165,
      "learning_rate": 7.649838903157218e-06,
      "loss": 0.3634,
      "step": 21950
    },
    {
      "epoch": 1.007342480840714,
      "grad_norm": 0.4620627760887146,
      "learning_rate": 7.649630976267169e-06,
      "loss": 0.3515,
      "step": 21951
    },
    {
      "epoch": 1.0073883713459686,
      "grad_norm": 0.44375136494636536,
      "learning_rate": 7.64942304300559e-06,
      "loss": 0.314,
      "step": 21952
    },
    {
      "epoch": 1.007434261851223,
      "grad_norm": 0.44201672077178955,
      "learning_rate": 7.649215103372987e-06,
      "loss": 0.3086,
      "step": 21953
    },
    {
      "epoch": 1.0074801523564774,
      "grad_norm": 0.45628949999809265,
      "learning_rate": 7.649007157369858e-06,
      "loss": 0.3484,
      "step": 21954
    },
    {
      "epoch": 1.0075260428617319,
      "grad_norm": 0.45369455218315125,
      "learning_rate": 7.648799204996702e-06,
      "loss": 0.3219,
      "step": 21955
    },
    {
      "epoch": 1.0075719333669864,
      "grad_norm": 0.4913589060306549,
      "learning_rate": 7.648591246254023e-06,
      "loss": 0.4055,
      "step": 21956
    },
    {
      "epoch": 1.0076178238722409,
      "grad_norm": 0.42942869663238525,
      "learning_rate": 7.648383281142314e-06,
      "loss": 0.2915,
      "step": 21957
    },
    {
      "epoch": 1.0076637143774954,
      "grad_norm": 0.4813835918903351,
      "learning_rate": 7.64817530966208e-06,
      "loss": 0.3754,
      "step": 21958
    },
    {
      "epoch": 1.0077096048827499,
      "grad_norm": 0.5294176936149597,
      "learning_rate": 7.64796733181382e-06,
      "loss": 0.4681,
      "step": 21959
    },
    {
      "epoch": 1.0077554953880041,
      "grad_norm": 0.46044209599494934,
      "learning_rate": 7.647759347598036e-06,
      "loss": 0.3876,
      "step": 21960
    },
    {
      "epoch": 1.0078013858932586,
      "grad_norm": 0.48537254333496094,
      "learning_rate": 7.647551357015224e-06,
      "loss": 0.3688,
      "step": 21961
    },
    {
      "epoch": 1.0078472763985131,
      "grad_norm": 0.44438159465789795,
      "learning_rate": 7.64734336006589e-06,
      "loss": 0.3267,
      "step": 21962
    },
    {
      "epoch": 1.0078931669037676,
      "grad_norm": 0.4591151177883148,
      "learning_rate": 7.647135356750527e-06,
      "loss": 0.3632,
      "step": 21963
    },
    {
      "epoch": 1.0079390574090221,
      "grad_norm": 0.4508124589920044,
      "learning_rate": 7.646927347069641e-06,
      "loss": 0.3289,
      "step": 21964
    },
    {
      "epoch": 1.0079849479142766,
      "grad_norm": 0.42664802074432373,
      "learning_rate": 7.64671933102373e-06,
      "loss": 0.2926,
      "step": 21965
    },
    {
      "epoch": 1.008030838419531,
      "grad_norm": 0.5398070216178894,
      "learning_rate": 7.646511308613294e-06,
      "loss": 0.5411,
      "step": 21966
    },
    {
      "epoch": 1.0080767289247854,
      "grad_norm": 0.47213688492774963,
      "learning_rate": 7.646303279838832e-06,
      "loss": 0.3601,
      "step": 21967
    },
    {
      "epoch": 1.00812261943004,
      "grad_norm": 0.4734417498111725,
      "learning_rate": 7.646095244700848e-06,
      "loss": 0.3695,
      "step": 21968
    },
    {
      "epoch": 1.0081685099352944,
      "grad_norm": 0.4657507538795471,
      "learning_rate": 7.645887203199838e-06,
      "loss": 0.3911,
      "step": 21969
    },
    {
      "epoch": 1.0082144004405489,
      "grad_norm": 0.4774101674556732,
      "learning_rate": 7.645679155336308e-06,
      "loss": 0.3264,
      "step": 21970
    },
    {
      "epoch": 1.0082602909458034,
      "grad_norm": 0.4606582522392273,
      "learning_rate": 7.645471101110752e-06,
      "loss": 0.3165,
      "step": 21971
    },
    {
      "epoch": 1.0083061814510579,
      "grad_norm": 0.4135957360267639,
      "learning_rate": 7.645263040523672e-06,
      "loss": 0.2728,
      "step": 21972
    },
    {
      "epoch": 1.0083520719563122,
      "grad_norm": 0.4791407585144043,
      "learning_rate": 7.645054973575571e-06,
      "loss": 0.3807,
      "step": 21973
    },
    {
      "epoch": 1.0083979624615667,
      "grad_norm": 0.4825863242149353,
      "learning_rate": 7.644846900266946e-06,
      "loss": 0.3287,
      "step": 21974
    },
    {
      "epoch": 1.0084438529668212,
      "grad_norm": 0.4641238749027252,
      "learning_rate": 7.644638820598303e-06,
      "loss": 0.3536,
      "step": 21975
    },
    {
      "epoch": 1.0084897434720757,
      "grad_norm": 0.4588776230812073,
      "learning_rate": 7.644430734570134e-06,
      "loss": 0.3181,
      "step": 21976
    },
    {
      "epoch": 1.0085356339773301,
      "grad_norm": 0.476427286863327,
      "learning_rate": 7.644222642182946e-06,
      "loss": 0.3512,
      "step": 21977
    },
    {
      "epoch": 1.0085815244825846,
      "grad_norm": 0.4788590669631958,
      "learning_rate": 7.644014543437235e-06,
      "loss": 0.3801,
      "step": 21978
    },
    {
      "epoch": 1.008627414987839,
      "grad_norm": 0.5241381525993347,
      "learning_rate": 7.643806438333505e-06,
      "loss": 0.4341,
      "step": 21979
    },
    {
      "epoch": 1.0086733054930934,
      "grad_norm": 0.45544761419296265,
      "learning_rate": 7.643598326872254e-06,
      "loss": 0.3528,
      "step": 21980
    },
    {
      "epoch": 1.008719195998348,
      "grad_norm": 0.45860937237739563,
      "learning_rate": 7.643390209053984e-06,
      "loss": 0.3573,
      "step": 21981
    },
    {
      "epoch": 1.0087650865036024,
      "grad_norm": 0.5526385307312012,
      "learning_rate": 7.643182084879195e-06,
      "loss": 0.4955,
      "step": 21982
    },
    {
      "epoch": 1.008810977008857,
      "grad_norm": 0.4804658591747284,
      "learning_rate": 7.642973954348387e-06,
      "loss": 0.3936,
      "step": 21983
    },
    {
      "epoch": 1.0088568675141114,
      "grad_norm": 0.4508894979953766,
      "learning_rate": 7.642765817462061e-06,
      "loss": 0.3517,
      "step": 21984
    },
    {
      "epoch": 1.0089027580193657,
      "grad_norm": 0.4676394462585449,
      "learning_rate": 7.642557674220717e-06,
      "loss": 0.3206,
      "step": 21985
    },
    {
      "epoch": 1.0089486485246202,
      "grad_norm": 0.4926576316356659,
      "learning_rate": 7.642349524624857e-06,
      "loss": 0.4203,
      "step": 21986
    },
    {
      "epoch": 1.0089945390298747,
      "grad_norm": 0.4182214140892029,
      "learning_rate": 7.64214136867498e-06,
      "loss": 0.2693,
      "step": 21987
    },
    {
      "epoch": 1.0090404295351292,
      "grad_norm": 0.4686330556869507,
      "learning_rate": 7.641933206371586e-06,
      "loss": 0.3717,
      "step": 21988
    },
    {
      "epoch": 1.0090863200403837,
      "grad_norm": 0.46840035915374756,
      "learning_rate": 7.641725037715175e-06,
      "loss": 0.3866,
      "step": 21989
    },
    {
      "epoch": 1.0091322105456382,
      "grad_norm": 0.4498589336872101,
      "learning_rate": 7.64151686270625e-06,
      "loss": 0.3249,
      "step": 21990
    },
    {
      "epoch": 1.0091781010508927,
      "grad_norm": 0.4583662450313568,
      "learning_rate": 7.641308681345311e-06,
      "loss": 0.3168,
      "step": 21991
    },
    {
      "epoch": 1.009223991556147,
      "grad_norm": 0.47554442286491394,
      "learning_rate": 7.64110049363286e-06,
      "loss": 0.4262,
      "step": 21992
    },
    {
      "epoch": 1.0092698820614014,
      "grad_norm": 0.4735945165157318,
      "learning_rate": 7.640892299569395e-06,
      "loss": 0.3788,
      "step": 21993
    },
    {
      "epoch": 1.009315772566656,
      "grad_norm": 0.4630342125892639,
      "learning_rate": 7.640684099155416e-06,
      "loss": 0.3648,
      "step": 21994
    },
    {
      "epoch": 1.0093616630719104,
      "grad_norm": 0.4587978422641754,
      "learning_rate": 7.640475892391425e-06,
      "loss": 0.3358,
      "step": 21995
    },
    {
      "epoch": 1.009407553577165,
      "grad_norm": 0.4704074561595917,
      "learning_rate": 7.640267679277926e-06,
      "loss": 0.3771,
      "step": 21996
    },
    {
      "epoch": 1.0094534440824194,
      "grad_norm": 0.4564983546733856,
      "learning_rate": 7.640059459815414e-06,
      "loss": 0.3881,
      "step": 21997
    },
    {
      "epoch": 1.0094993345876737,
      "grad_norm": 0.5114113092422485,
      "learning_rate": 7.639851234004392e-06,
      "loss": 0.424,
      "step": 21998
    },
    {
      "epoch": 1.0095452250929282,
      "grad_norm": 0.48488306999206543,
      "learning_rate": 7.639643001845363e-06,
      "loss": 0.3837,
      "step": 21999
    },
    {
      "epoch": 1.0095911155981827,
      "grad_norm": 0.5178372859954834,
      "learning_rate": 7.639434763338824e-06,
      "loss": 0.44,
      "step": 22000
    },
    {
      "epoch": 1.0096370061034372,
      "grad_norm": 0.42639902234077454,
      "learning_rate": 7.639226518485278e-06,
      "loss": 0.2889,
      "step": 22001
    },
    {
      "epoch": 1.0096828966086917,
      "grad_norm": 0.4664183557033539,
      "learning_rate": 7.639018267285224e-06,
      "loss": 0.3595,
      "step": 22002
    },
    {
      "epoch": 1.0097287871139462,
      "grad_norm": 0.5068734288215637,
      "learning_rate": 7.638810009739166e-06,
      "loss": 0.388,
      "step": 22003
    },
    {
      "epoch": 1.0097746776192005,
      "grad_norm": 0.4730684459209442,
      "learning_rate": 7.638601745847602e-06,
      "loss": 0.319,
      "step": 22004
    },
    {
      "epoch": 1.009820568124455,
      "grad_norm": 0.5206175446510315,
      "learning_rate": 7.638393475611032e-06,
      "loss": 0.3928,
      "step": 22005
    },
    {
      "epoch": 1.0098664586297095,
      "grad_norm": 0.474385529756546,
      "learning_rate": 7.638185199029959e-06,
      "loss": 0.3293,
      "step": 22006
    },
    {
      "epoch": 1.009912349134964,
      "grad_norm": 0.47156351804733276,
      "learning_rate": 7.637976916104886e-06,
      "loss": 0.3725,
      "step": 22007
    },
    {
      "epoch": 1.0099582396402185,
      "grad_norm": 0.44430676102638245,
      "learning_rate": 7.637768626836308e-06,
      "loss": 0.3003,
      "step": 22008
    },
    {
      "epoch": 1.010004130145473,
      "grad_norm": 0.4669210612773895,
      "learning_rate": 7.63756033122473e-06,
      "loss": 0.3869,
      "step": 22009
    },
    {
      "epoch": 1.0100500206507275,
      "grad_norm": 0.4672679901123047,
      "learning_rate": 7.637352029270653e-06,
      "loss": 0.3233,
      "step": 22010
    },
    {
      "epoch": 1.0100959111559817,
      "grad_norm": 0.4440314769744873,
      "learning_rate": 7.637143720974574e-06,
      "loss": 0.3169,
      "step": 22011
    },
    {
      "epoch": 1.0101418016612362,
      "grad_norm": 0.45189571380615234,
      "learning_rate": 7.636935406336998e-06,
      "loss": 0.2825,
      "step": 22012
    },
    {
      "epoch": 1.0101876921664907,
      "grad_norm": 0.4757099151611328,
      "learning_rate": 7.636727085358424e-06,
      "loss": 0.3569,
      "step": 22013
    },
    {
      "epoch": 1.0102335826717452,
      "grad_norm": 0.44854632019996643,
      "learning_rate": 7.636518758039356e-06,
      "loss": 0.3316,
      "step": 22014
    },
    {
      "epoch": 1.0102794731769997,
      "grad_norm": 0.4765455722808838,
      "learning_rate": 7.63631042438029e-06,
      "loss": 0.3827,
      "step": 22015
    },
    {
      "epoch": 1.0103253636822542,
      "grad_norm": 0.441520631313324,
      "learning_rate": 7.63610208438173e-06,
      "loss": 0.3495,
      "step": 22016
    },
    {
      "epoch": 1.0103712541875085,
      "grad_norm": 0.4542381465435028,
      "learning_rate": 7.635893738044176e-06,
      "loss": 0.3717,
      "step": 22017
    },
    {
      "epoch": 1.010417144692763,
      "grad_norm": 0.45801815390586853,
      "learning_rate": 7.635685385368132e-06,
      "loss": 0.3461,
      "step": 22018
    },
    {
      "epoch": 1.0104630351980175,
      "grad_norm": 0.46151214838027954,
      "learning_rate": 7.635477026354094e-06,
      "loss": 0.3385,
      "step": 22019
    },
    {
      "epoch": 1.010508925703272,
      "grad_norm": 0.4726400077342987,
      "learning_rate": 7.635268661002566e-06,
      "loss": 0.3682,
      "step": 22020
    },
    {
      "epoch": 1.0105548162085265,
      "grad_norm": 0.488136351108551,
      "learning_rate": 7.635060289314049e-06,
      "loss": 0.3808,
      "step": 22021
    },
    {
      "epoch": 1.010600706713781,
      "grad_norm": 0.4829421937465668,
      "learning_rate": 7.634851911289042e-06,
      "loss": 0.4273,
      "step": 22022
    },
    {
      "epoch": 1.0106465972190355,
      "grad_norm": 0.4190484583377838,
      "learning_rate": 7.63464352692805e-06,
      "loss": 0.2723,
      "step": 22023
    },
    {
      "epoch": 1.0106924877242898,
      "grad_norm": 0.47775572538375854,
      "learning_rate": 7.63443513623157e-06,
      "loss": 0.3372,
      "step": 22024
    },
    {
      "epoch": 1.0107383782295443,
      "grad_norm": 0.4634079933166504,
      "learning_rate": 7.634226739200108e-06,
      "loss": 0.3329,
      "step": 22025
    },
    {
      "epoch": 1.0107842687347988,
      "grad_norm": 0.459500253200531,
      "learning_rate": 7.634018335834158e-06,
      "loss": 0.3685,
      "step": 22026
    },
    {
      "epoch": 1.0108301592400533,
      "grad_norm": 0.44724348187446594,
      "learning_rate": 7.63380992613423e-06,
      "loss": 0.2979,
      "step": 22027
    },
    {
      "epoch": 1.0108760497453078,
      "grad_norm": 0.4745018184185028,
      "learning_rate": 7.633601510100817e-06,
      "loss": 0.3483,
      "step": 22028
    },
    {
      "epoch": 1.0109219402505623,
      "grad_norm": 0.5010080337524414,
      "learning_rate": 7.633393087734422e-06,
      "loss": 0.4189,
      "step": 22029
    },
    {
      "epoch": 1.0109678307558165,
      "grad_norm": 0.634324848651886,
      "learning_rate": 7.633184659035551e-06,
      "loss": 0.4415,
      "step": 22030
    },
    {
      "epoch": 1.011013721261071,
      "grad_norm": 0.45068231225013733,
      "learning_rate": 7.6329762240047e-06,
      "loss": 0.3038,
      "step": 22031
    },
    {
      "epoch": 1.0110596117663255,
      "grad_norm": 0.4984112083911896,
      "learning_rate": 7.632767782642374e-06,
      "loss": 0.4408,
      "step": 22032
    },
    {
      "epoch": 1.01110550227158,
      "grad_norm": 0.4544280171394348,
      "learning_rate": 7.632559334949072e-06,
      "loss": 0.3284,
      "step": 22033
    },
    {
      "epoch": 1.0111513927768345,
      "grad_norm": 0.4905652105808258,
      "learning_rate": 7.632350880925295e-06,
      "loss": 0.4332,
      "step": 22034
    },
    {
      "epoch": 1.011197283282089,
      "grad_norm": 0.46233922243118286,
      "learning_rate": 7.632142420571544e-06,
      "loss": 0.3227,
      "step": 22035
    },
    {
      "epoch": 1.0112431737873433,
      "grad_norm": 0.4680502712726593,
      "learning_rate": 7.631933953888324e-06,
      "loss": 0.3367,
      "step": 22036
    },
    {
      "epoch": 1.0112890642925978,
      "grad_norm": 0.47527310252189636,
      "learning_rate": 7.63172548087613e-06,
      "loss": 0.3461,
      "step": 22037
    },
    {
      "epoch": 1.0113349547978523,
      "grad_norm": 0.5120348334312439,
      "learning_rate": 7.63151700153547e-06,
      "loss": 0.4237,
      "step": 22038
    },
    {
      "epoch": 1.0113808453031068,
      "grad_norm": 0.44648757576942444,
      "learning_rate": 7.63130851586684e-06,
      "loss": 0.3164,
      "step": 22039
    },
    {
      "epoch": 1.0114267358083613,
      "grad_norm": 0.46843788027763367,
      "learning_rate": 7.631100023870743e-06,
      "loss": 0.3391,
      "step": 22040
    },
    {
      "epoch": 1.0114726263136158,
      "grad_norm": 0.4264986217021942,
      "learning_rate": 7.630891525547684e-06,
      "loss": 0.2909,
      "step": 22041
    },
    {
      "epoch": 1.0115185168188703,
      "grad_norm": 0.4607801139354706,
      "learning_rate": 7.63068302089816e-06,
      "loss": 0.3275,
      "step": 22042
    },
    {
      "epoch": 1.0115644073241246,
      "grad_norm": 0.4534933567047119,
      "learning_rate": 7.630474509922672e-06,
      "loss": 0.3536,
      "step": 22043
    },
    {
      "epoch": 1.011610297829379,
      "grad_norm": 0.4521688222885132,
      "learning_rate": 7.630265992621724e-06,
      "loss": 0.3343,
      "step": 22044
    },
    {
      "epoch": 1.0116561883346336,
      "grad_norm": 0.5108500123023987,
      "learning_rate": 7.630057468995815e-06,
      "loss": 0.3754,
      "step": 22045
    },
    {
      "epoch": 1.011702078839888,
      "grad_norm": 0.46493223309516907,
      "learning_rate": 7.62984893904545e-06,
      "loss": 0.3568,
      "step": 22046
    },
    {
      "epoch": 1.0117479693451426,
      "grad_norm": 0.47341030836105347,
      "learning_rate": 7.629640402771127e-06,
      "loss": 0.3782,
      "step": 22047
    },
    {
      "epoch": 1.011793859850397,
      "grad_norm": 0.4748906195163727,
      "learning_rate": 7.62943186017335e-06,
      "loss": 0.365,
      "step": 22048
    },
    {
      "epoch": 1.0118397503556513,
      "grad_norm": 0.4835397005081177,
      "learning_rate": 7.629223311252618e-06,
      "loss": 0.3636,
      "step": 22049
    },
    {
      "epoch": 1.0118856408609058,
      "grad_norm": 0.4603281617164612,
      "learning_rate": 7.629014756009434e-06,
      "loss": 0.3733,
      "step": 22050
    },
    {
      "epoch": 1.0119315313661603,
      "grad_norm": 0.44926467537879944,
      "learning_rate": 7.628806194444299e-06,
      "loss": 0.3839,
      "step": 22051
    },
    {
      "epoch": 1.0119774218714148,
      "grad_norm": 0.5048958659172058,
      "learning_rate": 7.628597626557717e-06,
      "loss": 0.382,
      "step": 22052
    },
    {
      "epoch": 1.0120233123766693,
      "grad_norm": 0.4764832854270935,
      "learning_rate": 7.628389052350185e-06,
      "loss": 0.3755,
      "step": 22053
    },
    {
      "epoch": 1.0120692028819238,
      "grad_norm": 0.4480188488960266,
      "learning_rate": 7.6281804718222065e-06,
      "loss": 0.3114,
      "step": 22054
    },
    {
      "epoch": 1.012115093387178,
      "grad_norm": 0.4723823666572571,
      "learning_rate": 7.627971884974286e-06,
      "loss": 0.3541,
      "step": 22055
    },
    {
      "epoch": 1.0121609838924326,
      "grad_norm": 0.4636901319026947,
      "learning_rate": 7.62776329180692e-06,
      "loss": 0.3137,
      "step": 22056
    },
    {
      "epoch": 1.012206874397687,
      "grad_norm": 0.4176749587059021,
      "learning_rate": 7.627554692320613e-06,
      "loss": 0.2731,
      "step": 22057
    },
    {
      "epoch": 1.0122527649029416,
      "grad_norm": 0.47738325595855713,
      "learning_rate": 7.627346086515868e-06,
      "loss": 0.3945,
      "step": 22058
    },
    {
      "epoch": 1.012298655408196,
      "grad_norm": 0.48821792006492615,
      "learning_rate": 7.627137474393183e-06,
      "loss": 0.4184,
      "step": 22059
    },
    {
      "epoch": 1.0123445459134506,
      "grad_norm": 0.49111419916152954,
      "learning_rate": 7.626928855953062e-06,
      "loss": 0.402,
      "step": 22060
    },
    {
      "epoch": 1.012390436418705,
      "grad_norm": 0.5020433068275452,
      "learning_rate": 7.626720231196006e-06,
      "loss": 0.4581,
      "step": 22061
    },
    {
      "epoch": 1.0124363269239594,
      "grad_norm": 0.49074047803878784,
      "learning_rate": 7.626511600122517e-06,
      "loss": 0.4346,
      "step": 22062
    },
    {
      "epoch": 1.0124822174292138,
      "grad_norm": 0.476683646440506,
      "learning_rate": 7.626302962733097e-06,
      "loss": 0.3367,
      "step": 22063
    },
    {
      "epoch": 1.0125281079344683,
      "grad_norm": 0.4653279185295105,
      "learning_rate": 7.626094319028248e-06,
      "loss": 0.3068,
      "step": 22064
    },
    {
      "epoch": 1.0125739984397228,
      "grad_norm": 0.4887130856513977,
      "learning_rate": 7.62588566900847e-06,
      "loss": 0.3564,
      "step": 22065
    },
    {
      "epoch": 1.0126198889449773,
      "grad_norm": 0.5384306907653809,
      "learning_rate": 7.625677012674264e-06,
      "loss": 0.4069,
      "step": 22066
    },
    {
      "epoch": 1.0126657794502318,
      "grad_norm": 0.43069005012512207,
      "learning_rate": 7.625468350026138e-06,
      "loss": 0.2997,
      "step": 22067
    },
    {
      "epoch": 1.0127116699554861,
      "grad_norm": 0.4755381643772125,
      "learning_rate": 7.625259681064585e-06,
      "loss": 0.3479,
      "step": 22068
    },
    {
      "epoch": 1.0127575604607406,
      "grad_norm": 0.47719621658325195,
      "learning_rate": 7.625051005790114e-06,
      "loss": 0.3845,
      "step": 22069
    },
    {
      "epoch": 1.0128034509659951,
      "grad_norm": 0.456089049577713,
      "learning_rate": 7.624842324203221e-06,
      "loss": 0.3396,
      "step": 22070
    },
    {
      "epoch": 1.0128493414712496,
      "grad_norm": 0.4463846683502197,
      "learning_rate": 7.624633636304411e-06,
      "loss": 0.3491,
      "step": 22071
    },
    {
      "epoch": 1.012895231976504,
      "grad_norm": 0.4942513108253479,
      "learning_rate": 7.624424942094188e-06,
      "loss": 0.3959,
      "step": 22072
    },
    {
      "epoch": 1.0129411224817586,
      "grad_norm": 0.5073105692863464,
      "learning_rate": 7.624216241573048e-06,
      "loss": 0.3661,
      "step": 22073
    },
    {
      "epoch": 1.0129870129870129,
      "grad_norm": 0.45593196153640747,
      "learning_rate": 7.624007534741499e-06,
      "loss": 0.3714,
      "step": 22074
    },
    {
      "epoch": 1.0130329034922674,
      "grad_norm": 0.45498326420783997,
      "learning_rate": 7.6237988216000394e-06,
      "loss": 0.3553,
      "step": 22075
    },
    {
      "epoch": 1.0130787939975219,
      "grad_norm": 0.4758533835411072,
      "learning_rate": 7.62359010214917e-06,
      "loss": 0.3803,
      "step": 22076
    },
    {
      "epoch": 1.0131246845027764,
      "grad_norm": 0.49344056844711304,
      "learning_rate": 7.623381376389396e-06,
      "loss": 0.4449,
      "step": 22077
    },
    {
      "epoch": 1.0131705750080309,
      "grad_norm": 0.42184823751449585,
      "learning_rate": 7.623172644321217e-06,
      "loss": 0.2739,
      "step": 22078
    },
    {
      "epoch": 1.0132164655132854,
      "grad_norm": 0.4517514407634735,
      "learning_rate": 7.6229639059451375e-06,
      "loss": 0.2948,
      "step": 22079
    },
    {
      "epoch": 1.0132623560185399,
      "grad_norm": 0.4632375240325928,
      "learning_rate": 7.622755161261655e-06,
      "loss": 0.4019,
      "step": 22080
    },
    {
      "epoch": 1.0133082465237941,
      "grad_norm": 0.38979053497314453,
      "learning_rate": 7.6225464102712766e-06,
      "loss": 0.2445,
      "step": 22081
    },
    {
      "epoch": 1.0133541370290486,
      "grad_norm": 0.50668865442276,
      "learning_rate": 7.6223376529745e-06,
      "loss": 0.4549,
      "step": 22082
    },
    {
      "epoch": 1.0134000275343031,
      "grad_norm": 0.477242648601532,
      "learning_rate": 7.622128889371829e-06,
      "loss": 0.3815,
      "step": 22083
    },
    {
      "epoch": 1.0134459180395576,
      "grad_norm": 0.4867419898509979,
      "learning_rate": 7.621920119463766e-06,
      "loss": 0.3795,
      "step": 22084
    },
    {
      "epoch": 1.0134918085448121,
      "grad_norm": 0.4102017879486084,
      "learning_rate": 7.621711343250813e-06,
      "loss": 0.2787,
      "step": 22085
    },
    {
      "epoch": 1.0135376990500666,
      "grad_norm": 0.4611830413341522,
      "learning_rate": 7.621502560733471e-06,
      "loss": 0.3307,
      "step": 22086
    },
    {
      "epoch": 1.013583589555321,
      "grad_norm": 0.4710094630718231,
      "learning_rate": 7.621293771912244e-06,
      "loss": 0.3803,
      "step": 22087
    },
    {
      "epoch": 1.0136294800605754,
      "grad_norm": 0.4376713037490845,
      "learning_rate": 7.621084976787632e-06,
      "loss": 0.3315,
      "step": 22088
    },
    {
      "epoch": 1.01367537056583,
      "grad_norm": 0.49071240425109863,
      "learning_rate": 7.620876175360137e-06,
      "loss": 0.3488,
      "step": 22089
    },
    {
      "epoch": 1.0137212610710844,
      "grad_norm": 0.46005964279174805,
      "learning_rate": 7.620667367630263e-06,
      "loss": 0.3383,
      "step": 22090
    },
    {
      "epoch": 1.013767151576339,
      "grad_norm": 0.44709059596061707,
      "learning_rate": 7.620458553598511e-06,
      "loss": 0.3598,
      "step": 22091
    },
    {
      "epoch": 1.0138130420815934,
      "grad_norm": 0.48072659969329834,
      "learning_rate": 7.620249733265383e-06,
      "loss": 0.3431,
      "step": 22092
    },
    {
      "epoch": 1.0138589325868477,
      "grad_norm": 0.4581223428249359,
      "learning_rate": 7.620040906631381e-06,
      "loss": 0.3581,
      "step": 22093
    },
    {
      "epoch": 1.0139048230921022,
      "grad_norm": 0.4809369146823883,
      "learning_rate": 7.6198320736970065e-06,
      "loss": 0.35,
      "step": 22094
    },
    {
      "epoch": 1.0139507135973567,
      "grad_norm": 0.5036215782165527,
      "learning_rate": 7.6196232344627655e-06,
      "loss": 0.4093,
      "step": 22095
    },
    {
      "epoch": 1.0139966041026112,
      "grad_norm": 0.5208719968795776,
      "learning_rate": 7.619414388929156e-06,
      "loss": 0.3769,
      "step": 22096
    },
    {
      "epoch": 1.0140424946078657,
      "grad_norm": 0.4754723906517029,
      "learning_rate": 7.619205537096681e-06,
      "loss": 0.3392,
      "step": 22097
    },
    {
      "epoch": 1.0140883851131202,
      "grad_norm": 0.4784787893295288,
      "learning_rate": 7.618996678965844e-06,
      "loss": 0.3603,
      "step": 22098
    },
    {
      "epoch": 1.0141342756183747,
      "grad_norm": 0.404583215713501,
      "learning_rate": 7.6187878145371475e-06,
      "loss": 0.2388,
      "step": 22099
    },
    {
      "epoch": 1.014180166123629,
      "grad_norm": 0.48957884311676025,
      "learning_rate": 7.618578943811092e-06,
      "loss": 0.4256,
      "step": 22100
    },
    {
      "epoch": 1.0142260566288834,
      "grad_norm": 0.46243757009506226,
      "learning_rate": 7.618370066788181e-06,
      "loss": 0.3476,
      "step": 22101
    },
    {
      "epoch": 1.014271947134138,
      "grad_norm": 0.47779539227485657,
      "learning_rate": 7.618161183468915e-06,
      "loss": 0.3823,
      "step": 22102
    },
    {
      "epoch": 1.0143178376393924,
      "grad_norm": 0.4467626214027405,
      "learning_rate": 7.617952293853798e-06,
      "loss": 0.3037,
      "step": 22103
    },
    {
      "epoch": 1.014363728144647,
      "grad_norm": 0.46788108348846436,
      "learning_rate": 7.617743397943333e-06,
      "loss": 0.3359,
      "step": 22104
    },
    {
      "epoch": 1.0144096186499014,
      "grad_norm": 0.46843335032463074,
      "learning_rate": 7.61753449573802e-06,
      "loss": 0.4208,
      "step": 22105
    },
    {
      "epoch": 1.0144555091551557,
      "grad_norm": 0.47303012013435364,
      "learning_rate": 7.617325587238365e-06,
      "loss": 0.3788,
      "step": 22106
    },
    {
      "epoch": 1.0145013996604102,
      "grad_norm": 0.5298566222190857,
      "learning_rate": 7.617116672444867e-06,
      "loss": 0.4689,
      "step": 22107
    },
    {
      "epoch": 1.0145472901656647,
      "grad_norm": 0.4463147819042206,
      "learning_rate": 7.616907751358027e-06,
      "loss": 0.3538,
      "step": 22108
    },
    {
      "epoch": 1.0145931806709192,
      "grad_norm": 0.4658949375152588,
      "learning_rate": 7.616698823978353e-06,
      "loss": 0.3709,
      "step": 22109
    },
    {
      "epoch": 1.0146390711761737,
      "grad_norm": 0.4489476978778839,
      "learning_rate": 7.616489890306341e-06,
      "loss": 0.3157,
      "step": 22110
    },
    {
      "epoch": 1.0146849616814282,
      "grad_norm": 0.4422892928123474,
      "learning_rate": 7.616280950342499e-06,
      "loss": 0.2934,
      "step": 22111
    },
    {
      "epoch": 1.0147308521866827,
      "grad_norm": 0.470743864774704,
      "learning_rate": 7.616072004087327e-06,
      "loss": 0.3352,
      "step": 22112
    },
    {
      "epoch": 1.014776742691937,
      "grad_norm": 0.493049681186676,
      "learning_rate": 7.615863051541325e-06,
      "loss": 0.4096,
      "step": 22113
    },
    {
      "epoch": 1.0148226331971915,
      "grad_norm": 0.44023841619491577,
      "learning_rate": 7.6156540927050005e-06,
      "loss": 0.2967,
      "step": 22114
    },
    {
      "epoch": 1.014868523702446,
      "grad_norm": 0.48853105306625366,
      "learning_rate": 7.615445127578853e-06,
      "loss": 0.3974,
      "step": 22115
    },
    {
      "epoch": 1.0149144142077005,
      "grad_norm": 0.4452958405017853,
      "learning_rate": 7.615236156163385e-06,
      "loss": 0.3108,
      "step": 22116
    },
    {
      "epoch": 1.014960304712955,
      "grad_norm": 0.5077585577964783,
      "learning_rate": 7.6150271784591e-06,
      "loss": 0.4559,
      "step": 22117
    },
    {
      "epoch": 1.0150061952182095,
      "grad_norm": 0.4653472602367401,
      "learning_rate": 7.6148181944665e-06,
      "loss": 0.3645,
      "step": 22118
    },
    {
      "epoch": 1.0150520857234637,
      "grad_norm": 0.4486215114593506,
      "learning_rate": 7.614609204186088e-06,
      "loss": 0.3269,
      "step": 22119
    },
    {
      "epoch": 1.0150979762287182,
      "grad_norm": 0.4825180470943451,
      "learning_rate": 7.6144002076183644e-06,
      "loss": 0.3551,
      "step": 22120
    },
    {
      "epoch": 1.0151438667339727,
      "grad_norm": 0.45526447892189026,
      "learning_rate": 7.614191204763834e-06,
      "loss": 0.3062,
      "step": 22121
    },
    {
      "epoch": 1.0151897572392272,
      "grad_norm": 0.4712159335613251,
      "learning_rate": 7.613982195623e-06,
      "loss": 0.3644,
      "step": 22122
    },
    {
      "epoch": 1.0152356477444817,
      "grad_norm": 0.45140278339385986,
      "learning_rate": 7.613773180196362e-06,
      "loss": 0.3637,
      "step": 22123
    },
    {
      "epoch": 1.0152815382497362,
      "grad_norm": 0.4519461989402771,
      "learning_rate": 7.613564158484426e-06,
      "loss": 0.3644,
      "step": 22124
    },
    {
      "epoch": 1.0153274287549905,
      "grad_norm": 0.4816216826438904,
      "learning_rate": 7.6133551304876905e-06,
      "loss": 0.3745,
      "step": 22125
    },
    {
      "epoch": 1.015373319260245,
      "grad_norm": 0.5087295174598694,
      "learning_rate": 7.613146096206663e-06,
      "loss": 0.4226,
      "step": 22126
    },
    {
      "epoch": 1.0154192097654995,
      "grad_norm": 0.4526520073413849,
      "learning_rate": 7.6129370556418424e-06,
      "loss": 0.3111,
      "step": 22127
    },
    {
      "epoch": 1.015465100270754,
      "grad_norm": 0.42599615454673767,
      "learning_rate": 7.612728008793735e-06,
      "loss": 0.2775,
      "step": 22128
    },
    {
      "epoch": 1.0155109907760085,
      "grad_norm": 0.4877137839794159,
      "learning_rate": 7.612518955662841e-06,
      "loss": 0.3461,
      "step": 22129
    },
    {
      "epoch": 1.015556881281263,
      "grad_norm": 0.47664228081703186,
      "learning_rate": 7.612309896249661e-06,
      "loss": 0.3419,
      "step": 22130
    },
    {
      "epoch": 1.0156027717865175,
      "grad_norm": 0.4753614664077759,
      "learning_rate": 7.612100830554703e-06,
      "loss": 0.3501,
      "step": 22131
    },
    {
      "epoch": 1.0156486622917718,
      "grad_norm": 0.5116774439811707,
      "learning_rate": 7.611891758578465e-06,
      "loss": 0.4237,
      "step": 22132
    },
    {
      "epoch": 1.0156945527970263,
      "grad_norm": 0.4641915261745453,
      "learning_rate": 7.6116826803214526e-06,
      "loss": 0.3205,
      "step": 22133
    },
    {
      "epoch": 1.0157404433022807,
      "grad_norm": 0.494960218667984,
      "learning_rate": 7.611473595784169e-06,
      "loss": 0.3826,
      "step": 22134
    },
    {
      "epoch": 1.0157863338075352,
      "grad_norm": 0.4413314461708069,
      "learning_rate": 7.611264504967113e-06,
      "loss": 0.2815,
      "step": 22135
    },
    {
      "epoch": 1.0158322243127897,
      "grad_norm": 0.48673513531684875,
      "learning_rate": 7.611055407870791e-06,
      "loss": 0.4043,
      "step": 22136
    },
    {
      "epoch": 1.0158781148180442,
      "grad_norm": 0.44349953532218933,
      "learning_rate": 7.610846304495705e-06,
      "loss": 0.3262,
      "step": 22137
    },
    {
      "epoch": 1.0159240053232985,
      "grad_norm": 0.4706258475780487,
      "learning_rate": 7.610637194842356e-06,
      "loss": 0.3706,
      "step": 22138
    },
    {
      "epoch": 1.015969895828553,
      "grad_norm": 0.4683552384376526,
      "learning_rate": 7.610428078911252e-06,
      "loss": 0.4074,
      "step": 22139
    },
    {
      "epoch": 1.0160157863338075,
      "grad_norm": 0.40438202023506165,
      "learning_rate": 7.610218956702889e-06,
      "loss": 0.2679,
      "step": 22140
    },
    {
      "epoch": 1.016061676839062,
      "grad_norm": 0.42577850818634033,
      "learning_rate": 7.610009828217775e-06,
      "loss": 0.306,
      "step": 22141
    },
    {
      "epoch": 1.0161075673443165,
      "grad_norm": 0.4815780818462372,
      "learning_rate": 7.60980069345641e-06,
      "loss": 0.3597,
      "step": 22142
    },
    {
      "epoch": 1.016153457849571,
      "grad_norm": 0.4786836802959442,
      "learning_rate": 7.609591552419299e-06,
      "loss": 0.3653,
      "step": 22143
    },
    {
      "epoch": 1.0161993483548253,
      "grad_norm": 0.453630268573761,
      "learning_rate": 7.609382405106945e-06,
      "loss": 0.3442,
      "step": 22144
    },
    {
      "epoch": 1.0162452388600798,
      "grad_norm": 0.4582040309906006,
      "learning_rate": 7.6091732515198465e-06,
      "loss": 0.3685,
      "step": 22145
    },
    {
      "epoch": 1.0162911293653343,
      "grad_norm": 0.45196786522865295,
      "learning_rate": 7.608964091658513e-06,
      "loss": 0.3028,
      "step": 22146
    },
    {
      "epoch": 1.0163370198705888,
      "grad_norm": 0.488829106092453,
      "learning_rate": 7.608754925523442e-06,
      "loss": 0.3768,
      "step": 22147
    },
    {
      "epoch": 1.0163829103758433,
      "grad_norm": 0.4882957935333252,
      "learning_rate": 7.608545753115139e-06,
      "loss": 0.3666,
      "step": 22148
    },
    {
      "epoch": 1.0164288008810978,
      "grad_norm": 0.5124367475509644,
      "learning_rate": 7.608336574434107e-06,
      "loss": 0.3818,
      "step": 22149
    },
    {
      "epoch": 1.0164746913863523,
      "grad_norm": 0.46642589569091797,
      "learning_rate": 7.608127389480851e-06,
      "loss": 0.3997,
      "step": 22150
    },
    {
      "epoch": 1.0165205818916065,
      "grad_norm": 0.5109196305274963,
      "learning_rate": 7.607918198255871e-06,
      "loss": 0.386,
      "step": 22151
    },
    {
      "epoch": 1.016566472396861,
      "grad_norm": 0.473605215549469,
      "learning_rate": 7.607709000759669e-06,
      "loss": 0.3946,
      "step": 22152
    },
    {
      "epoch": 1.0166123629021155,
      "grad_norm": 0.47982141375541687,
      "learning_rate": 7.60749979699275e-06,
      "loss": 0.3781,
      "step": 22153
    },
    {
      "epoch": 1.01665825340737,
      "grad_norm": 0.44807589054107666,
      "learning_rate": 7.607290586955618e-06,
      "loss": 0.3205,
      "step": 22154
    },
    {
      "epoch": 1.0167041439126245,
      "grad_norm": 0.47111403942108154,
      "learning_rate": 7.607081370648774e-06,
      "loss": 0.3551,
      "step": 22155
    },
    {
      "epoch": 1.016750034417879,
      "grad_norm": 0.4593442678451538,
      "learning_rate": 7.606872148072724e-06,
      "loss": 0.3392,
      "step": 22156
    },
    {
      "epoch": 1.0167959249231333,
      "grad_norm": 0.5090833306312561,
      "learning_rate": 7.606662919227967e-06,
      "loss": 0.3956,
      "step": 22157
    },
    {
      "epoch": 1.0168418154283878,
      "grad_norm": 0.4910643994808197,
      "learning_rate": 7.606453684115009e-06,
      "loss": 0.3442,
      "step": 22158
    },
    {
      "epoch": 1.0168877059336423,
      "grad_norm": 0.44392526149749756,
      "learning_rate": 7.606244442734353e-06,
      "loss": 0.3184,
      "step": 22159
    },
    {
      "epoch": 1.0169335964388968,
      "grad_norm": 0.42901718616485596,
      "learning_rate": 7.606035195086502e-06,
      "loss": 0.319,
      "step": 22160
    },
    {
      "epoch": 1.0169794869441513,
      "grad_norm": 0.4865281879901886,
      "learning_rate": 7.605825941171958e-06,
      "loss": 0.4078,
      "step": 22161
    },
    {
      "epoch": 1.0170253774494058,
      "grad_norm": 0.4496835470199585,
      "learning_rate": 7.605616680991226e-06,
      "loss": 0.3339,
      "step": 22162
    },
    {
      "epoch": 1.01707126795466,
      "grad_norm": 0.4560079574584961,
      "learning_rate": 7.605407414544807e-06,
      "loss": 0.3647,
      "step": 22163
    },
    {
      "epoch": 1.0171171584599146,
      "grad_norm": 0.4802283048629761,
      "learning_rate": 7.605198141833206e-06,
      "loss": 0.3636,
      "step": 22164
    },
    {
      "epoch": 1.017163048965169,
      "grad_norm": 0.47219231724739075,
      "learning_rate": 7.604988862856927e-06,
      "loss": 0.4009,
      "step": 22165
    },
    {
      "epoch": 1.0172089394704236,
      "grad_norm": 0.5257166028022766,
      "learning_rate": 7.604779577616471e-06,
      "loss": 0.4435,
      "step": 22166
    },
    {
      "epoch": 1.017254829975678,
      "grad_norm": 0.4396454989910126,
      "learning_rate": 7.604570286112342e-06,
      "loss": 0.3027,
      "step": 22167
    },
    {
      "epoch": 1.0173007204809326,
      "grad_norm": 0.4743387699127197,
      "learning_rate": 7.604360988345043e-06,
      "loss": 0.3754,
      "step": 22168
    },
    {
      "epoch": 1.017346610986187,
      "grad_norm": 0.49622318148612976,
      "learning_rate": 7.604151684315077e-06,
      "loss": 0.37,
      "step": 22169
    },
    {
      "epoch": 1.0173925014914413,
      "grad_norm": 0.4840548038482666,
      "learning_rate": 7.60394237402295e-06,
      "loss": 0.3448,
      "step": 22170
    },
    {
      "epoch": 1.0174383919966958,
      "grad_norm": 0.5096202492713928,
      "learning_rate": 7.603733057469163e-06,
      "loss": 0.4565,
      "step": 22171
    },
    {
      "epoch": 1.0174842825019503,
      "grad_norm": 0.4843490719795227,
      "learning_rate": 7.60352373465422e-06,
      "loss": 0.3535,
      "step": 22172
    },
    {
      "epoch": 1.0175301730072048,
      "grad_norm": 0.4585023820400238,
      "learning_rate": 7.603314405578624e-06,
      "loss": 0.342,
      "step": 22173
    },
    {
      "epoch": 1.0175760635124593,
      "grad_norm": 0.4567861258983612,
      "learning_rate": 7.6031050702428774e-06,
      "loss": 0.3399,
      "step": 22174
    },
    {
      "epoch": 1.0176219540177138,
      "grad_norm": 0.5014970302581787,
      "learning_rate": 7.602895728647486e-06,
      "loss": 0.4148,
      "step": 22175
    },
    {
      "epoch": 1.017667844522968,
      "grad_norm": 0.454193115234375,
      "learning_rate": 7.602686380792952e-06,
      "loss": 0.3307,
      "step": 22176
    },
    {
      "epoch": 1.0177137350282226,
      "grad_norm": 0.46800950169563293,
      "learning_rate": 7.602477026679778e-06,
      "loss": 0.3302,
      "step": 22177
    },
    {
      "epoch": 1.017759625533477,
      "grad_norm": 0.4763777256011963,
      "learning_rate": 7.602267666308467e-06,
      "loss": 0.3201,
      "step": 22178
    },
    {
      "epoch": 1.0178055160387316,
      "grad_norm": 0.49628642201423645,
      "learning_rate": 7.602058299679525e-06,
      "loss": 0.4226,
      "step": 22179
    },
    {
      "epoch": 1.017851406543986,
      "grad_norm": 0.4462299048900604,
      "learning_rate": 7.6018489267934535e-06,
      "loss": 0.3075,
      "step": 22180
    },
    {
      "epoch": 1.0178972970492406,
      "grad_norm": 0.4279725253582001,
      "learning_rate": 7.6016395476507565e-06,
      "loss": 0.3319,
      "step": 22181
    },
    {
      "epoch": 1.0179431875544949,
      "grad_norm": 0.47382137179374695,
      "learning_rate": 7.601430162251937e-06,
      "loss": 0.4164,
      "step": 22182
    },
    {
      "epoch": 1.0179890780597494,
      "grad_norm": 0.4706905484199524,
      "learning_rate": 7.6012207705975e-06,
      "loss": 0.3965,
      "step": 22183
    },
    {
      "epoch": 1.0180349685650039,
      "grad_norm": 0.4728710353374481,
      "learning_rate": 7.601011372687947e-06,
      "loss": 0.3989,
      "step": 22184
    },
    {
      "epoch": 1.0180808590702584,
      "grad_norm": 0.4665845036506653,
      "learning_rate": 7.6008019685237845e-06,
      "loss": 0.3428,
      "step": 22185
    },
    {
      "epoch": 1.0181267495755129,
      "grad_norm": 0.49987250566482544,
      "learning_rate": 7.600592558105513e-06,
      "loss": 0.3552,
      "step": 22186
    },
    {
      "epoch": 1.0181726400807674,
      "grad_norm": 0.4747629761695862,
      "learning_rate": 7.600383141433636e-06,
      "loss": 0.3458,
      "step": 22187
    },
    {
      "epoch": 1.0182185305860219,
      "grad_norm": 0.4755016565322876,
      "learning_rate": 7.600173718508661e-06,
      "loss": 0.3747,
      "step": 22188
    },
    {
      "epoch": 1.0182644210912761,
      "grad_norm": 0.4548684358596802,
      "learning_rate": 7.599964289331085e-06,
      "loss": 0.3263,
      "step": 22189
    },
    {
      "epoch": 1.0183103115965306,
      "grad_norm": 0.48045143485069275,
      "learning_rate": 7.599754853901418e-06,
      "loss": 0.3708,
      "step": 22190
    },
    {
      "epoch": 1.0183562021017851,
      "grad_norm": 0.4917643964290619,
      "learning_rate": 7.5995454122201615e-06,
      "loss": 0.3801,
      "step": 22191
    },
    {
      "epoch": 1.0184020926070396,
      "grad_norm": 0.4784308671951294,
      "learning_rate": 7.599335964287817e-06,
      "loss": 0.3704,
      "step": 22192
    },
    {
      "epoch": 1.0184479831122941,
      "grad_norm": 0.5150562524795532,
      "learning_rate": 7.59912651010489e-06,
      "loss": 0.3754,
      "step": 22193
    },
    {
      "epoch": 1.0184938736175486,
      "grad_norm": 0.47991785407066345,
      "learning_rate": 7.598917049671886e-06,
      "loss": 0.3675,
      "step": 22194
    },
    {
      "epoch": 1.018539764122803,
      "grad_norm": 0.45261263847351074,
      "learning_rate": 7.598707582989305e-06,
      "loss": 0.3431,
      "step": 22195
    },
    {
      "epoch": 1.0185856546280574,
      "grad_norm": 0.45501869916915894,
      "learning_rate": 7.598498110057653e-06,
      "loss": 0.3046,
      "step": 22196
    },
    {
      "epoch": 1.0186315451333119,
      "grad_norm": 0.5878657698631287,
      "learning_rate": 7.598288630877433e-06,
      "loss": 0.4706,
      "step": 22197
    },
    {
      "epoch": 1.0186774356385664,
      "grad_norm": 0.4284994602203369,
      "learning_rate": 7.59807914544915e-06,
      "loss": 0.2728,
      "step": 22198
    },
    {
      "epoch": 1.0187233261438209,
      "grad_norm": 0.48287561535835266,
      "learning_rate": 7.597869653773305e-06,
      "loss": 0.3283,
      "step": 22199
    },
    {
      "epoch": 1.0187692166490754,
      "grad_norm": 0.5041671395301819,
      "learning_rate": 7.597660155850404e-06,
      "loss": 0.422,
      "step": 22200
    },
    {
      "epoch": 1.0188151071543299,
      "grad_norm": 0.4774300754070282,
      "learning_rate": 7.59745065168095e-06,
      "loss": 0.3642,
      "step": 22201
    },
    {
      "epoch": 1.0188609976595842,
      "grad_norm": 0.4760906398296356,
      "learning_rate": 7.597241141265448e-06,
      "loss": 0.3888,
      "step": 22202
    },
    {
      "epoch": 1.0189068881648387,
      "grad_norm": 0.471473753452301,
      "learning_rate": 7.5970316246043985e-06,
      "loss": 0.4215,
      "step": 22203
    },
    {
      "epoch": 1.0189527786700932,
      "grad_norm": 0.45234355330467224,
      "learning_rate": 7.596822101698309e-06,
      "loss": 0.333,
      "step": 22204
    },
    {
      "epoch": 1.0189986691753476,
      "grad_norm": 0.49176210165023804,
      "learning_rate": 7.5966125725476836e-06,
      "loss": 0.4209,
      "step": 22205
    },
    {
      "epoch": 1.0190445596806021,
      "grad_norm": 0.5233871340751648,
      "learning_rate": 7.596403037153022e-06,
      "loss": 0.4668,
      "step": 22206
    },
    {
      "epoch": 1.0190904501858566,
      "grad_norm": 0.4806680977344513,
      "learning_rate": 7.596193495514831e-06,
      "loss": 0.3374,
      "step": 22207
    },
    {
      "epoch": 1.019136340691111,
      "grad_norm": 0.4347662031650543,
      "learning_rate": 7.595983947633614e-06,
      "loss": 0.3319,
      "step": 22208
    },
    {
      "epoch": 1.0191822311963654,
      "grad_norm": 0.44121137261390686,
      "learning_rate": 7.595774393509876e-06,
      "loss": 0.3411,
      "step": 22209
    },
    {
      "epoch": 1.01922812170162,
      "grad_norm": 0.447936087846756,
      "learning_rate": 7.595564833144119e-06,
      "loss": 0.3346,
      "step": 22210
    },
    {
      "epoch": 1.0192740122068744,
      "grad_norm": 0.4689260721206665,
      "learning_rate": 7.595355266536848e-06,
      "loss": 0.3576,
      "step": 22211
    },
    {
      "epoch": 1.019319902712129,
      "grad_norm": 0.5022258758544922,
      "learning_rate": 7.595145693688566e-06,
      "loss": 0.3895,
      "step": 22212
    },
    {
      "epoch": 1.0193657932173834,
      "grad_norm": 0.4731845557689667,
      "learning_rate": 7.594936114599778e-06,
      "loss": 0.3454,
      "step": 22213
    },
    {
      "epoch": 1.0194116837226377,
      "grad_norm": 0.4816189706325531,
      "learning_rate": 7.594726529270988e-06,
      "loss": 0.3797,
      "step": 22214
    },
    {
      "epoch": 1.0194575742278922,
      "grad_norm": 0.47558578848838806,
      "learning_rate": 7.594516937702701e-06,
      "loss": 0.3788,
      "step": 22215
    },
    {
      "epoch": 1.0195034647331467,
      "grad_norm": 0.7088468074798584,
      "learning_rate": 7.594307339895417e-06,
      "loss": 0.3268,
      "step": 22216
    },
    {
      "epoch": 1.0195493552384012,
      "grad_norm": 0.5633447170257568,
      "learning_rate": 7.594097735849644e-06,
      "loss": 0.4885,
      "step": 22217
    },
    {
      "epoch": 1.0195952457436557,
      "grad_norm": 0.4830838143825531,
      "learning_rate": 7.593888125565884e-06,
      "loss": 0.3362,
      "step": 22218
    },
    {
      "epoch": 1.0196411362489102,
      "grad_norm": 0.47700080275535583,
      "learning_rate": 7.593678509044642e-06,
      "loss": 0.3747,
      "step": 22219
    },
    {
      "epoch": 1.0196870267541647,
      "grad_norm": 0.4795612692832947,
      "learning_rate": 7.5934688862864236e-06,
      "loss": 0.3506,
      "step": 22220
    },
    {
      "epoch": 1.019732917259419,
      "grad_norm": 0.4981966018676758,
      "learning_rate": 7.593259257291727e-06,
      "loss": 0.4436,
      "step": 22221
    },
    {
      "epoch": 1.0197788077646734,
      "grad_norm": 0.5193784236907959,
      "learning_rate": 7.593049622061064e-06,
      "loss": 0.4309,
      "step": 22222
    },
    {
      "epoch": 1.019824698269928,
      "grad_norm": 0.458997905254364,
      "learning_rate": 7.592839980594934e-06,
      "loss": 0.3287,
      "step": 22223
    },
    {
      "epoch": 1.0198705887751824,
      "grad_norm": 0.4828920066356659,
      "learning_rate": 7.592630332893841e-06,
      "loss": 0.3525,
      "step": 22224
    },
    {
      "epoch": 1.019916479280437,
      "grad_norm": 0.4917052686214447,
      "learning_rate": 7.592420678958292e-06,
      "loss": 0.4105,
      "step": 22225
    },
    {
      "epoch": 1.0199623697856914,
      "grad_norm": 0.47202107310295105,
      "learning_rate": 7.592211018788789e-06,
      "loss": 0.3447,
      "step": 22226
    },
    {
      "epoch": 1.0200082602909457,
      "grad_norm": 0.4994295835494995,
      "learning_rate": 7.592001352385836e-06,
      "loss": 0.4273,
      "step": 22227
    },
    {
      "epoch": 1.0200541507962002,
      "grad_norm": 0.47072696685791016,
      "learning_rate": 7.591791679749938e-06,
      "loss": 0.3622,
      "step": 22228
    },
    {
      "epoch": 1.0201000413014547,
      "grad_norm": 0.46992817521095276,
      "learning_rate": 7.591582000881599e-06,
      "loss": 0.3538,
      "step": 22229
    },
    {
      "epoch": 1.0201459318067092,
      "grad_norm": 0.44903314113616943,
      "learning_rate": 7.591372315781324e-06,
      "loss": 0.3373,
      "step": 22230
    },
    {
      "epoch": 1.0201918223119637,
      "grad_norm": 0.4431670904159546,
      "learning_rate": 7.5911626244496155e-06,
      "loss": 0.3132,
      "step": 22231
    },
    {
      "epoch": 1.0202377128172182,
      "grad_norm": 0.463265061378479,
      "learning_rate": 7.5909529268869786e-06,
      "loss": 0.3407,
      "step": 22232
    },
    {
      "epoch": 1.0202836033224725,
      "grad_norm": 0.45582839846611023,
      "learning_rate": 7.590743223093918e-06,
      "loss": 0.3596,
      "step": 22233
    },
    {
      "epoch": 1.020329493827727,
      "grad_norm": 0.48671987652778625,
      "learning_rate": 7.590533513070937e-06,
      "loss": 0.3776,
      "step": 22234
    },
    {
      "epoch": 1.0203753843329815,
      "grad_norm": 0.49230876564979553,
      "learning_rate": 7.590323796818542e-06,
      "loss": 0.4038,
      "step": 22235
    },
    {
      "epoch": 1.020421274838236,
      "grad_norm": 0.4757263660430908,
      "learning_rate": 7.590114074337234e-06,
      "loss": 0.3632,
      "step": 22236
    },
    {
      "epoch": 1.0204671653434905,
      "grad_norm": 0.46610596776008606,
      "learning_rate": 7.589904345627521e-06,
      "loss": 0.3761,
      "step": 22237
    },
    {
      "epoch": 1.020513055848745,
      "grad_norm": 0.4523918926715851,
      "learning_rate": 7.589694610689904e-06,
      "loss": 0.3525,
      "step": 22238
    },
    {
      "epoch": 1.0205589463539995,
      "grad_norm": 0.48547810316085815,
      "learning_rate": 7.58948486952489e-06,
      "loss": 0.4132,
      "step": 22239
    },
    {
      "epoch": 1.0206048368592537,
      "grad_norm": 0.44026634097099304,
      "learning_rate": 7.58927512213298e-06,
      "loss": 0.3454,
      "step": 22240
    },
    {
      "epoch": 1.0206507273645082,
      "grad_norm": 0.46190929412841797,
      "learning_rate": 7.589065368514682e-06,
      "loss": 0.3051,
      "step": 22241
    },
    {
      "epoch": 1.0206966178697627,
      "grad_norm": 0.4953574240207672,
      "learning_rate": 7.5888556086705005e-06,
      "loss": 0.3709,
      "step": 22242
    },
    {
      "epoch": 1.0207425083750172,
      "grad_norm": 0.46077170968055725,
      "learning_rate": 7.588645842600936e-06,
      "loss": 0.3588,
      "step": 22243
    },
    {
      "epoch": 1.0207883988802717,
      "grad_norm": 0.5069426894187927,
      "learning_rate": 7.588436070306495e-06,
      "loss": 0.4364,
      "step": 22244
    },
    {
      "epoch": 1.0208342893855262,
      "grad_norm": 0.4488179683685303,
      "learning_rate": 7.588226291787683e-06,
      "loss": 0.3428,
      "step": 22245
    },
    {
      "epoch": 1.0208801798907805,
      "grad_norm": 0.4825003147125244,
      "learning_rate": 7.5880165070450044e-06,
      "loss": 0.356,
      "step": 22246
    },
    {
      "epoch": 1.020926070396035,
      "grad_norm": 0.432256817817688,
      "learning_rate": 7.587806716078963e-06,
      "loss": 0.2962,
      "step": 22247
    },
    {
      "epoch": 1.0209719609012895,
      "grad_norm": 0.4861150085926056,
      "learning_rate": 7.587596918890062e-06,
      "loss": 0.3505,
      "step": 22248
    },
    {
      "epoch": 1.021017851406544,
      "grad_norm": 0.4912828505039215,
      "learning_rate": 7.587387115478807e-06,
      "loss": 0.4064,
      "step": 22249
    },
    {
      "epoch": 1.0210637419117985,
      "grad_norm": 0.4781213104724884,
      "learning_rate": 7.587177305845703e-06,
      "loss": 0.3889,
      "step": 22250
    },
    {
      "epoch": 1.021109632417053,
      "grad_norm": 0.44183406233787537,
      "learning_rate": 7.586967489991255e-06,
      "loss": 0.2849,
      "step": 22251
    },
    {
      "epoch": 1.0211555229223073,
      "grad_norm": 0.45234328508377075,
      "learning_rate": 7.586757667915965e-06,
      "loss": 0.3247,
      "step": 22252
    },
    {
      "epoch": 1.0212014134275618,
      "grad_norm": 0.4553004503250122,
      "learning_rate": 7.586547839620342e-06,
      "loss": 0.3461,
      "step": 22253
    },
    {
      "epoch": 1.0212473039328163,
      "grad_norm": 0.4535216689109802,
      "learning_rate": 7.586338005104885e-06,
      "loss": 0.3283,
      "step": 22254
    },
    {
      "epoch": 1.0212931944380708,
      "grad_norm": 0.4335271120071411,
      "learning_rate": 7.586128164370101e-06,
      "loss": 0.3036,
      "step": 22255
    },
    {
      "epoch": 1.0213390849433253,
      "grad_norm": 0.4839659631252289,
      "learning_rate": 7.585918317416496e-06,
      "loss": 0.3481,
      "step": 22256
    },
    {
      "epoch": 1.0213849754485798,
      "grad_norm": 0.48041415214538574,
      "learning_rate": 7.585708464244574e-06,
      "loss": 0.3667,
      "step": 22257
    },
    {
      "epoch": 1.0214308659538343,
      "grad_norm": 0.45650431513786316,
      "learning_rate": 7.585498604854839e-06,
      "loss": 0.3721,
      "step": 22258
    },
    {
      "epoch": 1.0214767564590885,
      "grad_norm": 0.4898580014705658,
      "learning_rate": 7.585288739247797e-06,
      "loss": 0.3974,
      "step": 22259
    },
    {
      "epoch": 1.021522646964343,
      "grad_norm": 0.4334639310836792,
      "learning_rate": 7.5850788674239494e-06,
      "loss": 0.321,
      "step": 22260
    },
    {
      "epoch": 1.0215685374695975,
      "grad_norm": 0.531714916229248,
      "learning_rate": 7.584868989383804e-06,
      "loss": 0.4067,
      "step": 22261
    },
    {
      "epoch": 1.021614427974852,
      "grad_norm": 0.4242710769176483,
      "learning_rate": 7.584659105127865e-06,
      "loss": 0.2899,
      "step": 22262
    },
    {
      "epoch": 1.0216603184801065,
      "grad_norm": 0.42120546102523804,
      "learning_rate": 7.584449214656635e-06,
      "loss": 0.2983,
      "step": 22263
    },
    {
      "epoch": 1.021706208985361,
      "grad_norm": 0.4590929448604584,
      "learning_rate": 7.584239317970623e-06,
      "loss": 0.3597,
      "step": 22264
    },
    {
      "epoch": 1.0217520994906153,
      "grad_norm": 0.45939624309539795,
      "learning_rate": 7.5840294150703285e-06,
      "loss": 0.3646,
      "step": 22265
    },
    {
      "epoch": 1.0217979899958698,
      "grad_norm": 0.43377286195755005,
      "learning_rate": 7.5838195059562604e-06,
      "loss": 0.2648,
      "step": 22266
    },
    {
      "epoch": 1.0218438805011243,
      "grad_norm": 0.46497035026550293,
      "learning_rate": 7.583609590628922e-06,
      "loss": 0.3401,
      "step": 22267
    },
    {
      "epoch": 1.0218897710063788,
      "grad_norm": 0.45689284801483154,
      "learning_rate": 7.583399669088815e-06,
      "loss": 0.3344,
      "step": 22268
    },
    {
      "epoch": 1.0219356615116333,
      "grad_norm": 0.46664556860923767,
      "learning_rate": 7.58318974133645e-06,
      "loss": 0.3344,
      "step": 22269
    },
    {
      "epoch": 1.0219815520168878,
      "grad_norm": 0.4780256152153015,
      "learning_rate": 7.58297980737233e-06,
      "loss": 0.3253,
      "step": 22270
    },
    {
      "epoch": 1.022027442522142,
      "grad_norm": 0.46115368604660034,
      "learning_rate": 7.582769867196955e-06,
      "loss": 0.3519,
      "step": 22271
    },
    {
      "epoch": 1.0220733330273966,
      "grad_norm": 0.4536869525909424,
      "learning_rate": 7.582559920810837e-06,
      "loss": 0.3057,
      "step": 22272
    },
    {
      "epoch": 1.022119223532651,
      "grad_norm": 0.4997319281101227,
      "learning_rate": 7.582349968214475e-06,
      "loss": 0.4833,
      "step": 22273
    },
    {
      "epoch": 1.0221651140379056,
      "grad_norm": 0.4599466323852539,
      "learning_rate": 7.582140009408378e-06,
      "loss": 0.3722,
      "step": 22274
    },
    {
      "epoch": 1.02221100454316,
      "grad_norm": 0.4423424005508423,
      "learning_rate": 7.581930044393049e-06,
      "loss": 0.3337,
      "step": 22275
    },
    {
      "epoch": 1.0222568950484145,
      "grad_norm": 0.45354461669921875,
      "learning_rate": 7.5817200731689924e-06,
      "loss": 0.3306,
      "step": 22276
    },
    {
      "epoch": 1.022302785553669,
      "grad_norm": 0.4293654263019562,
      "learning_rate": 7.581510095736715e-06,
      "loss": 0.3303,
      "step": 22277
    },
    {
      "epoch": 1.0223486760589233,
      "grad_norm": 0.4467923939228058,
      "learning_rate": 7.581300112096719e-06,
      "loss": 0.3317,
      "step": 22278
    },
    {
      "epoch": 1.0223945665641778,
      "grad_norm": 0.42587533593177795,
      "learning_rate": 7.581090122249512e-06,
      "loss": 0.2877,
      "step": 22279
    },
    {
      "epoch": 1.0224404570694323,
      "grad_norm": 0.46021565794944763,
      "learning_rate": 7.5808801261955975e-06,
      "loss": 0.3167,
      "step": 22280
    },
    {
      "epoch": 1.0224863475746868,
      "grad_norm": 0.5043947100639343,
      "learning_rate": 7.580670123935482e-06,
      "loss": 0.4324,
      "step": 22281
    },
    {
      "epoch": 1.0225322380799413,
      "grad_norm": 0.45424848794937134,
      "learning_rate": 7.580460115469668e-06,
      "loss": 0.3529,
      "step": 22282
    },
    {
      "epoch": 1.0225781285851958,
      "grad_norm": 0.4748769998550415,
      "learning_rate": 7.5802501007986605e-06,
      "loss": 0.3872,
      "step": 22283
    },
    {
      "epoch": 1.02262401909045,
      "grad_norm": 0.5129307508468628,
      "learning_rate": 7.580040079922968e-06,
      "loss": 0.4047,
      "step": 22284
    },
    {
      "epoch": 1.0226699095957046,
      "grad_norm": 0.45659831166267395,
      "learning_rate": 7.579830052843093e-06,
      "loss": 0.3274,
      "step": 22285
    },
    {
      "epoch": 1.022715800100959,
      "grad_norm": 0.4234340488910675,
      "learning_rate": 7.579620019559542e-06,
      "loss": 0.2857,
      "step": 22286
    },
    {
      "epoch": 1.0227616906062136,
      "grad_norm": 0.4579523801803589,
      "learning_rate": 7.579409980072817e-06,
      "loss": 0.3701,
      "step": 22287
    },
    {
      "epoch": 1.022807581111468,
      "grad_norm": 0.46205469965934753,
      "learning_rate": 7.579199934383426e-06,
      "loss": 0.3143,
      "step": 22288
    },
    {
      "epoch": 1.0228534716167226,
      "grad_norm": 0.47239184379577637,
      "learning_rate": 7.578989882491871e-06,
      "loss": 0.3603,
      "step": 22289
    },
    {
      "epoch": 1.022899362121977,
      "grad_norm": 0.4593476951122284,
      "learning_rate": 7.5787798243986635e-06,
      "loss": 0.3271,
      "step": 22290
    },
    {
      "epoch": 1.0229452526272313,
      "grad_norm": 0.5322347283363342,
      "learning_rate": 7.578569760104302e-06,
      "loss": 0.4463,
      "step": 22291
    },
    {
      "epoch": 1.0229911431324858,
      "grad_norm": 0.45614251494407654,
      "learning_rate": 7.578359689609294e-06,
      "loss": 0.3366,
      "step": 22292
    },
    {
      "epoch": 1.0230370336377403,
      "grad_norm": 0.4433298408985138,
      "learning_rate": 7.578149612914145e-06,
      "loss": 0.3109,
      "step": 22293
    },
    {
      "epoch": 1.0230829241429948,
      "grad_norm": 0.4425281882286072,
      "learning_rate": 7.5779395300193606e-06,
      "loss": 0.2965,
      "step": 22294
    },
    {
      "epoch": 1.0231288146482493,
      "grad_norm": 0.5146587491035461,
      "learning_rate": 7.577729440925443e-06,
      "loss": 0.3712,
      "step": 22295
    },
    {
      "epoch": 1.0231747051535038,
      "grad_norm": 0.5087910294532776,
      "learning_rate": 7.577519345632902e-06,
      "loss": 0.4746,
      "step": 22296
    },
    {
      "epoch": 1.0232205956587581,
      "grad_norm": 0.48190248012542725,
      "learning_rate": 7.577309244142239e-06,
      "loss": 0.3913,
      "step": 22297
    },
    {
      "epoch": 1.0232664861640126,
      "grad_norm": 0.42928194999694824,
      "learning_rate": 7.577099136453962e-06,
      "loss": 0.3275,
      "step": 22298
    },
    {
      "epoch": 1.023312376669267,
      "grad_norm": 0.44376417994499207,
      "learning_rate": 7.576889022568575e-06,
      "loss": 0.3069,
      "step": 22299
    },
    {
      "epoch": 1.0233582671745216,
      "grad_norm": 0.46500691771507263,
      "learning_rate": 7.57667890248658e-06,
      "loss": 0.4016,
      "step": 22300
    },
    {
      "epoch": 1.023404157679776,
      "grad_norm": 0.45745527744293213,
      "learning_rate": 7.57646877620849e-06,
      "loss": 0.3742,
      "step": 22301
    },
    {
      "epoch": 1.0234500481850306,
      "grad_norm": 0.41581276059150696,
      "learning_rate": 7.576258643734804e-06,
      "loss": 0.2956,
      "step": 22302
    },
    {
      "epoch": 1.0234959386902849,
      "grad_norm": 0.46407440304756165,
      "learning_rate": 7.576048505066028e-06,
      "loss": 0.3286,
      "step": 22303
    },
    {
      "epoch": 1.0235418291955394,
      "grad_norm": 0.46160003542900085,
      "learning_rate": 7.575838360202669e-06,
      "loss": 0.3865,
      "step": 22304
    },
    {
      "epoch": 1.0235877197007939,
      "grad_norm": 0.46395233273506165,
      "learning_rate": 7.5756282091452315e-06,
      "loss": 0.3657,
      "step": 22305
    },
    {
      "epoch": 1.0236336102060484,
      "grad_norm": 0.470740407705307,
      "learning_rate": 7.575418051894221e-06,
      "loss": 0.3761,
      "step": 22306
    },
    {
      "epoch": 1.0236795007113029,
      "grad_norm": 0.4814998209476471,
      "learning_rate": 7.575207888450144e-06,
      "loss": 0.3611,
      "step": 22307
    },
    {
      "epoch": 1.0237253912165574,
      "grad_norm": 0.45941299200057983,
      "learning_rate": 7.574997718813504e-06,
      "loss": 0.3401,
      "step": 22308
    },
    {
      "epoch": 1.0237712817218119,
      "grad_norm": 0.47474756836891174,
      "learning_rate": 7.574787542984807e-06,
      "loss": 0.3656,
      "step": 22309
    },
    {
      "epoch": 1.0238171722270661,
      "grad_norm": 0.4592050015926361,
      "learning_rate": 7.574577360964559e-06,
      "loss": 0.3506,
      "step": 22310
    },
    {
      "epoch": 1.0238630627323206,
      "grad_norm": 0.4951978623867035,
      "learning_rate": 7.574367172753265e-06,
      "loss": 0.4082,
      "step": 22311
    },
    {
      "epoch": 1.0239089532375751,
      "grad_norm": 0.4819735288619995,
      "learning_rate": 7.57415697835143e-06,
      "loss": 0.3606,
      "step": 22312
    },
    {
      "epoch": 1.0239548437428296,
      "grad_norm": 0.5496799945831299,
      "learning_rate": 7.573946777759561e-06,
      "loss": 0.4771,
      "step": 22313
    },
    {
      "epoch": 1.0240007342480841,
      "grad_norm": 0.5177422761917114,
      "learning_rate": 7.573736570978161e-06,
      "loss": 0.414,
      "step": 22314
    },
    {
      "epoch": 1.0240466247533386,
      "grad_norm": 0.477649986743927,
      "learning_rate": 7.5735263580077365e-06,
      "loss": 0.4393,
      "step": 22315
    },
    {
      "epoch": 1.024092515258593,
      "grad_norm": 0.4308584928512573,
      "learning_rate": 7.573316138848794e-06,
      "loss": 0.2752,
      "step": 22316
    },
    {
      "epoch": 1.0241384057638474,
      "grad_norm": 0.45932307839393616,
      "learning_rate": 7.573105913501839e-06,
      "loss": 0.3971,
      "step": 22317
    },
    {
      "epoch": 1.024184296269102,
      "grad_norm": 0.46062222123146057,
      "learning_rate": 7.572895681967376e-06,
      "loss": 0.3178,
      "step": 22318
    },
    {
      "epoch": 1.0242301867743564,
      "grad_norm": 0.4273023009300232,
      "learning_rate": 7.572685444245911e-06,
      "loss": 0.3436,
      "step": 22319
    },
    {
      "epoch": 1.024276077279611,
      "grad_norm": 0.4688393175601959,
      "learning_rate": 7.57247520033795e-06,
      "loss": 0.4284,
      "step": 22320
    },
    {
      "epoch": 1.0243219677848654,
      "grad_norm": 0.44319379329681396,
      "learning_rate": 7.572264950243997e-06,
      "loss": 0.3157,
      "step": 22321
    },
    {
      "epoch": 1.0243678582901197,
      "grad_norm": 0.45848655700683594,
      "learning_rate": 7.5720546939645585e-06,
      "loss": 0.3344,
      "step": 22322
    },
    {
      "epoch": 1.0244137487953742,
      "grad_norm": 0.4423292875289917,
      "learning_rate": 7.571844431500142e-06,
      "loss": 0.2761,
      "step": 22323
    },
    {
      "epoch": 1.0244596393006287,
      "grad_norm": 0.44142836332321167,
      "learning_rate": 7.571634162851249e-06,
      "loss": 0.3337,
      "step": 22324
    },
    {
      "epoch": 1.0245055298058832,
      "grad_norm": 0.4918602705001831,
      "learning_rate": 7.571423888018388e-06,
      "loss": 0.409,
      "step": 22325
    },
    {
      "epoch": 1.0245514203111377,
      "grad_norm": 0.4588117003440857,
      "learning_rate": 7.571213607002064e-06,
      "loss": 0.3437,
      "step": 22326
    },
    {
      "epoch": 1.0245973108163922,
      "grad_norm": 0.44028419256210327,
      "learning_rate": 7.571003319802785e-06,
      "loss": 0.2856,
      "step": 22327
    },
    {
      "epoch": 1.0246432013216467,
      "grad_norm": 0.49826285243034363,
      "learning_rate": 7.570793026421052e-06,
      "loss": 0.4433,
      "step": 22328
    },
    {
      "epoch": 1.024689091826901,
      "grad_norm": 0.43680697679519653,
      "learning_rate": 7.570582726857374e-06,
      "loss": 0.3065,
      "step": 22329
    },
    {
      "epoch": 1.0247349823321554,
      "grad_norm": 0.4948607385158539,
      "learning_rate": 7.570372421112255e-06,
      "loss": 0.3827,
      "step": 22330
    },
    {
      "epoch": 1.02478087283741,
      "grad_norm": 0.462436318397522,
      "learning_rate": 7.570162109186203e-06,
      "loss": 0.3735,
      "step": 22331
    },
    {
      "epoch": 1.0248267633426644,
      "grad_norm": 0.5094931125640869,
      "learning_rate": 7.5699517910797216e-06,
      "loss": 0.4005,
      "step": 22332
    },
    {
      "epoch": 1.024872653847919,
      "grad_norm": 0.45026957988739014,
      "learning_rate": 7.569741466793317e-06,
      "loss": 0.2676,
      "step": 22333
    },
    {
      "epoch": 1.0249185443531734,
      "grad_norm": 0.5112370252609253,
      "learning_rate": 7.569531136327497e-06,
      "loss": 0.4344,
      "step": 22334
    },
    {
      "epoch": 1.0249644348584277,
      "grad_norm": 0.4346585273742676,
      "learning_rate": 7.569320799682765e-06,
      "loss": 0.328,
      "step": 22335
    },
    {
      "epoch": 1.0250103253636822,
      "grad_norm": 0.43592745065689087,
      "learning_rate": 7.569110456859625e-06,
      "loss": 0.3012,
      "step": 22336
    },
    {
      "epoch": 1.0250562158689367,
      "grad_norm": 0.4990173280239105,
      "learning_rate": 7.568900107858586e-06,
      "loss": 0.3874,
      "step": 22337
    },
    {
      "epoch": 1.0251021063741912,
      "grad_norm": 0.48806604743003845,
      "learning_rate": 7.568689752680154e-06,
      "loss": 0.4017,
      "step": 22338
    },
    {
      "epoch": 1.0251479968794457,
      "grad_norm": 0.5158664584159851,
      "learning_rate": 7.568479391324832e-06,
      "loss": 0.5076,
      "step": 22339
    },
    {
      "epoch": 1.0251938873847002,
      "grad_norm": 0.47117242217063904,
      "learning_rate": 7.568269023793131e-06,
      "loss": 0.3217,
      "step": 22340
    },
    {
      "epoch": 1.0252397778899545,
      "grad_norm": 0.5159891247749329,
      "learning_rate": 7.568058650085552e-06,
      "loss": 0.4878,
      "step": 22341
    },
    {
      "epoch": 1.025285668395209,
      "grad_norm": 0.47598549723625183,
      "learning_rate": 7.5678482702026016e-06,
      "loss": 0.3868,
      "step": 22342
    },
    {
      "epoch": 1.0253315589004635,
      "grad_norm": 0.4446183741092682,
      "learning_rate": 7.5676378841447875e-06,
      "loss": 0.3575,
      "step": 22343
    },
    {
      "epoch": 1.025377449405718,
      "grad_norm": 0.44090571999549866,
      "learning_rate": 7.567427491912613e-06,
      "loss": 0.2941,
      "step": 22344
    },
    {
      "epoch": 1.0254233399109725,
      "grad_norm": 0.48144838213920593,
      "learning_rate": 7.567217093506588e-06,
      "loss": 0.384,
      "step": 22345
    },
    {
      "epoch": 1.025469230416227,
      "grad_norm": 0.4567750096321106,
      "learning_rate": 7.567006688927217e-06,
      "loss": 0.2881,
      "step": 22346
    },
    {
      "epoch": 1.0255151209214814,
      "grad_norm": 0.511216938495636,
      "learning_rate": 7.566796278175002e-06,
      "loss": 0.4982,
      "step": 22347
    },
    {
      "epoch": 1.0255610114267357,
      "grad_norm": 0.46285387873649597,
      "learning_rate": 7.566585861250453e-06,
      "loss": 0.3172,
      "step": 22348
    },
    {
      "epoch": 1.0256069019319902,
      "grad_norm": 0.43159592151641846,
      "learning_rate": 7.566375438154075e-06,
      "loss": 0.2929,
      "step": 22349
    },
    {
      "epoch": 1.0256527924372447,
      "grad_norm": 0.48541539907455444,
      "learning_rate": 7.566165008886375e-06,
      "loss": 0.4359,
      "step": 22350
    },
    {
      "epoch": 1.0256986829424992,
      "grad_norm": 0.44930824637413025,
      "learning_rate": 7.565954573447857e-06,
      "loss": 0.3247,
      "step": 22351
    },
    {
      "epoch": 1.0257445734477537,
      "grad_norm": 0.4637567698955536,
      "learning_rate": 7.565744131839029e-06,
      "loss": 0.3545,
      "step": 22352
    },
    {
      "epoch": 1.0257904639530082,
      "grad_norm": 0.4864486753940582,
      "learning_rate": 7.565533684060396e-06,
      "loss": 0.3706,
      "step": 22353
    },
    {
      "epoch": 1.0258363544582625,
      "grad_norm": 0.47738155722618103,
      "learning_rate": 7.565323230112462e-06,
      "loss": 0.3973,
      "step": 22354
    },
    {
      "epoch": 1.025882244963517,
      "grad_norm": 0.4547047019004822,
      "learning_rate": 7.565112769995737e-06,
      "loss": 0.3622,
      "step": 22355
    },
    {
      "epoch": 1.0259281354687715,
      "grad_norm": 0.48292824625968933,
      "learning_rate": 7.5649023037107265e-06,
      "loss": 0.4028,
      "step": 22356
    },
    {
      "epoch": 1.025974025974026,
      "grad_norm": 0.4564628601074219,
      "learning_rate": 7.5646918312579345e-06,
      "loss": 0.3505,
      "step": 22357
    },
    {
      "epoch": 1.0260199164792805,
      "grad_norm": 0.4875357151031494,
      "learning_rate": 7.564481352637868e-06,
      "loss": 0.3746,
      "step": 22358
    },
    {
      "epoch": 1.026065806984535,
      "grad_norm": 0.484312504529953,
      "learning_rate": 7.5642708678510335e-06,
      "loss": 0.3452,
      "step": 22359
    },
    {
      "epoch": 1.0261116974897893,
      "grad_norm": 0.5036768913269043,
      "learning_rate": 7.564060376897935e-06,
      "loss": 0.3694,
      "step": 22360
    },
    {
      "epoch": 1.0261575879950438,
      "grad_norm": 0.48736995458602905,
      "learning_rate": 7.563849879779083e-06,
      "loss": 0.369,
      "step": 22361
    },
    {
      "epoch": 1.0262034785002982,
      "grad_norm": 0.4783438444137573,
      "learning_rate": 7.56363937649498e-06,
      "loss": 0.3026,
      "step": 22362
    },
    {
      "epoch": 1.0262493690055527,
      "grad_norm": 0.522567629814148,
      "learning_rate": 7.563428867046135e-06,
      "loss": 0.4368,
      "step": 22363
    },
    {
      "epoch": 1.0262952595108072,
      "grad_norm": 0.4245157539844513,
      "learning_rate": 7.563218351433051e-06,
      "loss": 0.2918,
      "step": 22364
    },
    {
      "epoch": 1.0263411500160617,
      "grad_norm": 0.4428459405899048,
      "learning_rate": 7.563007829656235e-06,
      "loss": 0.3266,
      "step": 22365
    },
    {
      "epoch": 1.0263870405213162,
      "grad_norm": 0.4249100089073181,
      "learning_rate": 7.562797301716196e-06,
      "loss": 0.2836,
      "step": 22366
    },
    {
      "epoch": 1.0264329310265705,
      "grad_norm": 0.4268397092819214,
      "learning_rate": 7.5625867676134375e-06,
      "loss": 0.3194,
      "step": 22367
    },
    {
      "epoch": 1.026478821531825,
      "grad_norm": 0.4480585753917694,
      "learning_rate": 7.562376227348467e-06,
      "loss": 0.3298,
      "step": 22368
    },
    {
      "epoch": 1.0265247120370795,
      "grad_norm": 0.46138545870780945,
      "learning_rate": 7.5621656809217915e-06,
      "loss": 0.3741,
      "step": 22369
    },
    {
      "epoch": 1.026570602542334,
      "grad_norm": 0.46135807037353516,
      "learning_rate": 7.5619551283339136e-06,
      "loss": 0.3364,
      "step": 22370
    },
    {
      "epoch": 1.0266164930475885,
      "grad_norm": 0.4899735748767853,
      "learning_rate": 7.561744569585343e-06,
      "loss": 0.3498,
      "step": 22371
    },
    {
      "epoch": 1.026662383552843,
      "grad_norm": 0.5253533124923706,
      "learning_rate": 7.561534004676585e-06,
      "loss": 0.4429,
      "step": 22372
    },
    {
      "epoch": 1.0267082740580973,
      "grad_norm": 0.44150033593177795,
      "learning_rate": 7.561323433608147e-06,
      "loss": 0.3043,
      "step": 22373
    },
    {
      "epoch": 1.0267541645633518,
      "grad_norm": 0.4299364387989044,
      "learning_rate": 7.561112856380535e-06,
      "loss": 0.3034,
      "step": 22374
    },
    {
      "epoch": 1.0268000550686063,
      "grad_norm": 0.45328256487846375,
      "learning_rate": 7.5609022729942546e-06,
      "loss": 0.3018,
      "step": 22375
    },
    {
      "epoch": 1.0268459455738608,
      "grad_norm": 0.4577149450778961,
      "learning_rate": 7.560691683449811e-06,
      "loss": 0.3207,
      "step": 22376
    },
    {
      "epoch": 1.0268918360791153,
      "grad_norm": 0.49876320362091064,
      "learning_rate": 7.5604810877477145e-06,
      "loss": 0.3512,
      "step": 22377
    },
    {
      "epoch": 1.0269377265843698,
      "grad_norm": 0.4864700734615326,
      "learning_rate": 7.560270485888468e-06,
      "loss": 0.3865,
      "step": 22378
    },
    {
      "epoch": 1.0269836170896243,
      "grad_norm": 0.4928778409957886,
      "learning_rate": 7.560059877872578e-06,
      "loss": 0.3884,
      "step": 22379
    },
    {
      "epoch": 1.0270295075948785,
      "grad_norm": 0.48024070262908936,
      "learning_rate": 7.5598492637005535e-06,
      "loss": 0.3936,
      "step": 22380
    },
    {
      "epoch": 1.027075398100133,
      "grad_norm": 0.5223364233970642,
      "learning_rate": 7.559638643372897e-06,
      "loss": 0.4064,
      "step": 22381
    },
    {
      "epoch": 1.0271212886053875,
      "grad_norm": 0.4410637319087982,
      "learning_rate": 7.559428016890119e-06,
      "loss": 0.3369,
      "step": 22382
    },
    {
      "epoch": 1.027167179110642,
      "grad_norm": 0.44987592101097107,
      "learning_rate": 7.559217384252725e-06,
      "loss": 0.3394,
      "step": 22383
    },
    {
      "epoch": 1.0272130696158965,
      "grad_norm": 0.4607445299625397,
      "learning_rate": 7.5590067454612215e-06,
      "loss": 0.2943,
      "step": 22384
    },
    {
      "epoch": 1.027258960121151,
      "grad_norm": 0.48052191734313965,
      "learning_rate": 7.558796100516113e-06,
      "loss": 0.4004,
      "step": 22385
    },
    {
      "epoch": 1.0273048506264053,
      "grad_norm": 0.45145803689956665,
      "learning_rate": 7.558585449417908e-06,
      "loss": 0.3537,
      "step": 22386
    },
    {
      "epoch": 1.0273507411316598,
      "grad_norm": 0.47235146164894104,
      "learning_rate": 7.5583747921671115e-06,
      "loss": 0.3792,
      "step": 22387
    },
    {
      "epoch": 1.0273966316369143,
      "grad_norm": 0.44858264923095703,
      "learning_rate": 7.5581641287642315e-06,
      "loss": 0.3211,
      "step": 22388
    },
    {
      "epoch": 1.0274425221421688,
      "grad_norm": 0.4441016614437103,
      "learning_rate": 7.557953459209775e-06,
      "loss": 0.3091,
      "step": 22389
    },
    {
      "epoch": 1.0274884126474233,
      "grad_norm": 0.4486677944660187,
      "learning_rate": 7.5577427835042476e-06,
      "loss": 0.3193,
      "step": 22390
    },
    {
      "epoch": 1.0275343031526778,
      "grad_norm": 0.437423437833786,
      "learning_rate": 7.557532101648155e-06,
      "loss": 0.3192,
      "step": 22391
    },
    {
      "epoch": 1.027580193657932,
      "grad_norm": 0.4621531367301941,
      "learning_rate": 7.557321413642006e-06,
      "loss": 0.3558,
      "step": 22392
    },
    {
      "epoch": 1.0276260841631866,
      "grad_norm": 0.44279295206069946,
      "learning_rate": 7.557110719486305e-06,
      "loss": 0.278,
      "step": 22393
    },
    {
      "epoch": 1.027671974668441,
      "grad_norm": 0.4348197877407074,
      "learning_rate": 7.556900019181561e-06,
      "loss": 0.307,
      "step": 22394
    },
    {
      "epoch": 1.0277178651736956,
      "grad_norm": 0.4530927836894989,
      "learning_rate": 7.556689312728279e-06,
      "loss": 0.3569,
      "step": 22395
    },
    {
      "epoch": 1.02776375567895,
      "grad_norm": 0.4591359794139862,
      "learning_rate": 7.556478600126965e-06,
      "loss": 0.3062,
      "step": 22396
    },
    {
      "epoch": 1.0278096461842046,
      "grad_norm": 0.4886361360549927,
      "learning_rate": 7.556267881378129e-06,
      "loss": 0.4112,
      "step": 22397
    },
    {
      "epoch": 1.027855536689459,
      "grad_norm": 0.4790240228176117,
      "learning_rate": 7.556057156482273e-06,
      "loss": 0.3501,
      "step": 22398
    },
    {
      "epoch": 1.0279014271947133,
      "grad_norm": 0.4777625799179077,
      "learning_rate": 7.555846425439908e-06,
      "loss": 0.3785,
      "step": 22399
    },
    {
      "epoch": 1.0279473176999678,
      "grad_norm": 0.4528575539588928,
      "learning_rate": 7.555635688251539e-06,
      "loss": 0.327,
      "step": 22400
    },
    {
      "epoch": 1.0279932082052223,
      "grad_norm": 0.43054521083831787,
      "learning_rate": 7.555424944917673e-06,
      "loss": 0.2655,
      "step": 22401
    },
    {
      "epoch": 1.0280390987104768,
      "grad_norm": 0.47178617119789124,
      "learning_rate": 7.555214195438815e-06,
      "loss": 0.3343,
      "step": 22402
    },
    {
      "epoch": 1.0280849892157313,
      "grad_norm": 0.44585275650024414,
      "learning_rate": 7.555003439815473e-06,
      "loss": 0.3348,
      "step": 22403
    },
    {
      "epoch": 1.0281308797209858,
      "grad_norm": 0.5001834630966187,
      "learning_rate": 7.554792678048156e-06,
      "loss": 0.3791,
      "step": 22404
    },
    {
      "epoch": 1.02817677022624,
      "grad_norm": 0.47962501645088196,
      "learning_rate": 7.554581910137369e-06,
      "loss": 0.3706,
      "step": 22405
    },
    {
      "epoch": 1.0282226607314946,
      "grad_norm": 0.46621567010879517,
      "learning_rate": 7.554371136083618e-06,
      "loss": 0.2978,
      "step": 22406
    },
    {
      "epoch": 1.028268551236749,
      "grad_norm": 0.46204936504364014,
      "learning_rate": 7.55416035588741e-06,
      "loss": 0.3623,
      "step": 22407
    },
    {
      "epoch": 1.0283144417420036,
      "grad_norm": 0.4533815085887909,
      "learning_rate": 7.553949569549253e-06,
      "loss": 0.362,
      "step": 22408
    },
    {
      "epoch": 1.028360332247258,
      "grad_norm": 0.4397274851799011,
      "learning_rate": 7.553738777069654e-06,
      "loss": 0.3278,
      "step": 22409
    },
    {
      "epoch": 1.0284062227525126,
      "grad_norm": 0.4385591745376587,
      "learning_rate": 7.5535279784491174e-06,
      "loss": 0.3379,
      "step": 22410
    },
    {
      "epoch": 1.0284521132577669,
      "grad_norm": 0.4620034098625183,
      "learning_rate": 7.553317173688155e-06,
      "loss": 0.3337,
      "step": 22411
    },
    {
      "epoch": 1.0284980037630214,
      "grad_norm": 0.4451151490211487,
      "learning_rate": 7.553106362787268e-06,
      "loss": 0.3408,
      "step": 22412
    },
    {
      "epoch": 1.0285438942682759,
      "grad_norm": 0.4579235315322876,
      "learning_rate": 7.552895545746966e-06,
      "loss": 0.3705,
      "step": 22413
    },
    {
      "epoch": 1.0285897847735304,
      "grad_norm": 0.45619964599609375,
      "learning_rate": 7.552684722567757e-06,
      "loss": 0.3306,
      "step": 22414
    },
    {
      "epoch": 1.0286356752787849,
      "grad_norm": 0.5159657597541809,
      "learning_rate": 7.552473893250146e-06,
      "loss": 0.3547,
      "step": 22415
    },
    {
      "epoch": 1.0286815657840394,
      "grad_norm": 0.46768614649772644,
      "learning_rate": 7.552263057794642e-06,
      "loss": 0.355,
      "step": 22416
    },
    {
      "epoch": 1.0287274562892939,
      "grad_norm": 0.4567314684391022,
      "learning_rate": 7.552052216201751e-06,
      "loss": 0.341,
      "step": 22417
    },
    {
      "epoch": 1.0287733467945481,
      "grad_norm": 0.49946001172065735,
      "learning_rate": 7.551841368471978e-06,
      "loss": 0.4013,
      "step": 22418
    },
    {
      "epoch": 1.0288192372998026,
      "grad_norm": 0.4695288836956024,
      "learning_rate": 7.551630514605832e-06,
      "loss": 0.3745,
      "step": 22419
    },
    {
      "epoch": 1.0288651278050571,
      "grad_norm": 0.4795888364315033,
      "learning_rate": 7.5514196546038224e-06,
      "loss": 0.3683,
      "step": 22420
    },
    {
      "epoch": 1.0289110183103116,
      "grad_norm": 0.48820048570632935,
      "learning_rate": 7.551208788466453e-06,
      "loss": 0.3745,
      "step": 22421
    },
    {
      "epoch": 1.0289569088155661,
      "grad_norm": 0.505426824092865,
      "learning_rate": 7.5509979161942304e-06,
      "loss": 0.386,
      "step": 22422
    },
    {
      "epoch": 1.0290027993208206,
      "grad_norm": 0.48134180903434753,
      "learning_rate": 7.550787037787664e-06,
      "loss": 0.3212,
      "step": 22423
    },
    {
      "epoch": 1.029048689826075,
      "grad_norm": 0.46590206027030945,
      "learning_rate": 7.550576153247259e-06,
      "loss": 0.4346,
      "step": 22424
    },
    {
      "epoch": 1.0290945803313294,
      "grad_norm": 0.4820304811000824,
      "learning_rate": 7.550365262573523e-06,
      "loss": 0.3483,
      "step": 22425
    },
    {
      "epoch": 1.0291404708365839,
      "grad_norm": 0.45988038182258606,
      "learning_rate": 7.550154365766964e-06,
      "loss": 0.3355,
      "step": 22426
    },
    {
      "epoch": 1.0291863613418384,
      "grad_norm": 0.44796645641326904,
      "learning_rate": 7.54994346282809e-06,
      "loss": 0.2933,
      "step": 22427
    },
    {
      "epoch": 1.0292322518470929,
      "grad_norm": 0.48457062244415283,
      "learning_rate": 7.549732553757405e-06,
      "loss": 0.3485,
      "step": 22428
    },
    {
      "epoch": 1.0292781423523474,
      "grad_norm": 0.4248168170452118,
      "learning_rate": 7.549521638555419e-06,
      "loss": 0.3095,
      "step": 22429
    },
    {
      "epoch": 1.0293240328576017,
      "grad_norm": 0.4504879117012024,
      "learning_rate": 7.549310717222636e-06,
      "loss": 0.3436,
      "step": 22430
    },
    {
      "epoch": 1.0293699233628562,
      "grad_norm": 0.5392913818359375,
      "learning_rate": 7.549099789759569e-06,
      "loss": 0.4611,
      "step": 22431
    },
    {
      "epoch": 1.0294158138681107,
      "grad_norm": 0.42346471548080444,
      "learning_rate": 7.548888856166718e-06,
      "loss": 0.2717,
      "step": 22432
    },
    {
      "epoch": 1.0294617043733651,
      "grad_norm": 0.49569517374038696,
      "learning_rate": 7.548677916444596e-06,
      "loss": 0.4101,
      "step": 22433
    },
    {
      "epoch": 1.0295075948786196,
      "grad_norm": 0.4461812674999237,
      "learning_rate": 7.548466970593707e-06,
      "loss": 0.3189,
      "step": 22434
    },
    {
      "epoch": 1.0295534853838741,
      "grad_norm": 0.4729243218898773,
      "learning_rate": 7.548256018614559e-06,
      "loss": 0.3769,
      "step": 22435
    },
    {
      "epoch": 1.0295993758891286,
      "grad_norm": 0.47125616669654846,
      "learning_rate": 7.548045060507659e-06,
      "loss": 0.4048,
      "step": 22436
    },
    {
      "epoch": 1.029645266394383,
      "grad_norm": 0.4643147587776184,
      "learning_rate": 7.547834096273516e-06,
      "loss": 0.3922,
      "step": 22437
    },
    {
      "epoch": 1.0296911568996374,
      "grad_norm": 0.502774715423584,
      "learning_rate": 7.547623125912637e-06,
      "loss": 0.4128,
      "step": 22438
    },
    {
      "epoch": 1.029737047404892,
      "grad_norm": 0.49359238147735596,
      "learning_rate": 7.5474121494255255e-06,
      "loss": 0.3545,
      "step": 22439
    },
    {
      "epoch": 1.0297829379101464,
      "grad_norm": 0.46166208386421204,
      "learning_rate": 7.547201166812694e-06,
      "loss": 0.3654,
      "step": 22440
    },
    {
      "epoch": 1.029828828415401,
      "grad_norm": 0.4602765738964081,
      "learning_rate": 7.546990178074648e-06,
      "loss": 0.3383,
      "step": 22441
    },
    {
      "epoch": 1.0298747189206554,
      "grad_norm": 0.4695824980735779,
      "learning_rate": 7.546779183211893e-06,
      "loss": 0.3435,
      "step": 22442
    },
    {
      "epoch": 1.0299206094259097,
      "grad_norm": 0.5052186250686646,
      "learning_rate": 7.5465681822249405e-06,
      "loss": 0.367,
      "step": 22443
    },
    {
      "epoch": 1.0299664999311642,
      "grad_norm": 0.4670850336551666,
      "learning_rate": 7.5463571751142915e-06,
      "loss": 0.3171,
      "step": 22444
    },
    {
      "epoch": 1.0300123904364187,
      "grad_norm": 0.48633328080177307,
      "learning_rate": 7.54614616188046e-06,
      "loss": 0.3527,
      "step": 22445
    },
    {
      "epoch": 1.0300582809416732,
      "grad_norm": 0.48769477009773254,
      "learning_rate": 7.545935142523948e-06,
      "loss": 0.4101,
      "step": 22446
    },
    {
      "epoch": 1.0301041714469277,
      "grad_norm": 0.5118875503540039,
      "learning_rate": 7.545724117045268e-06,
      "loss": 0.4122,
      "step": 22447
    },
    {
      "epoch": 1.0301500619521822,
      "grad_norm": 0.45789262652397156,
      "learning_rate": 7.545513085444924e-06,
      "loss": 0.3717,
      "step": 22448
    },
    {
      "epoch": 1.0301959524574364,
      "grad_norm": 0.4542714059352875,
      "learning_rate": 7.545302047723424e-06,
      "loss": 0.3339,
      "step": 22449
    },
    {
      "epoch": 1.030241842962691,
      "grad_norm": 0.4558868408203125,
      "learning_rate": 7.545091003881276e-06,
      "loss": 0.3077,
      "step": 22450
    },
    {
      "epoch": 1.0302877334679454,
      "grad_norm": 0.5119561553001404,
      "learning_rate": 7.544879953918988e-06,
      "loss": 0.395,
      "step": 22451
    },
    {
      "epoch": 1.0303336239732,
      "grad_norm": 1.0092087984085083,
      "learning_rate": 7.544668897837066e-06,
      "loss": 0.321,
      "step": 22452
    },
    {
      "epoch": 1.0303795144784544,
      "grad_norm": 0.5007705092430115,
      "learning_rate": 7.544457835636018e-06,
      "loss": 0.4672,
      "step": 22453
    },
    {
      "epoch": 1.030425404983709,
      "grad_norm": 0.4739861786365509,
      "learning_rate": 7.5442467673163546e-06,
      "loss": 0.3628,
      "step": 22454
    },
    {
      "epoch": 1.0304712954889634,
      "grad_norm": 0.45855095982551575,
      "learning_rate": 7.5440356928785786e-06,
      "loss": 0.3249,
      "step": 22455
    },
    {
      "epoch": 1.0305171859942177,
      "grad_norm": 0.4487762153148651,
      "learning_rate": 7.543824612323199e-06,
      "loss": 0.3154,
      "step": 22456
    },
    {
      "epoch": 1.0305630764994722,
      "grad_norm": 0.4607127606868744,
      "learning_rate": 7.543613525650726e-06,
      "loss": 0.3682,
      "step": 22457
    },
    {
      "epoch": 1.0306089670047267,
      "grad_norm": 0.45102253556251526,
      "learning_rate": 7.543402432861664e-06,
      "loss": 0.3564,
      "step": 22458
    },
    {
      "epoch": 1.0306548575099812,
      "grad_norm": 0.4379313588142395,
      "learning_rate": 7.543191333956524e-06,
      "loss": 0.3144,
      "step": 22459
    },
    {
      "epoch": 1.0307007480152357,
      "grad_norm": 0.4747353196144104,
      "learning_rate": 7.542980228935809e-06,
      "loss": 0.4,
      "step": 22460
    },
    {
      "epoch": 1.0307466385204902,
      "grad_norm": 0.4937050938606262,
      "learning_rate": 7.54276911780003e-06,
      "loss": 0.3691,
      "step": 22461
    },
    {
      "epoch": 1.0307925290257445,
      "grad_norm": 0.474519819021225,
      "learning_rate": 7.542558000549694e-06,
      "loss": 0.3733,
      "step": 22462
    },
    {
      "epoch": 1.030838419530999,
      "grad_norm": 0.49929216504096985,
      "learning_rate": 7.542346877185307e-06,
      "loss": 0.386,
      "step": 22463
    },
    {
      "epoch": 1.0308843100362535,
      "grad_norm": 0.4621559679508209,
      "learning_rate": 7.54213574770738e-06,
      "loss": 0.3374,
      "step": 22464
    },
    {
      "epoch": 1.030930200541508,
      "grad_norm": 0.4684692323207855,
      "learning_rate": 7.541924612116418e-06,
      "loss": 0.3907,
      "step": 22465
    },
    {
      "epoch": 1.0309760910467625,
      "grad_norm": 0.47210416197776794,
      "learning_rate": 7.54171347041293e-06,
      "loss": 0.3683,
      "step": 22466
    },
    {
      "epoch": 1.031021981552017,
      "grad_norm": 0.45705515146255493,
      "learning_rate": 7.541502322597423e-06,
      "loss": 0.3523,
      "step": 22467
    },
    {
      "epoch": 1.0310678720572715,
      "grad_norm": 0.4867582321166992,
      "learning_rate": 7.541291168670405e-06,
      "loss": 0.3563,
      "step": 22468
    },
    {
      "epoch": 1.0311137625625257,
      "grad_norm": 0.4607963562011719,
      "learning_rate": 7.541080008632383e-06,
      "loss": 0.3873,
      "step": 22469
    },
    {
      "epoch": 1.0311596530677802,
      "grad_norm": 0.44386789202690125,
      "learning_rate": 7.540868842483867e-06,
      "loss": 0.3271,
      "step": 22470
    },
    {
      "epoch": 1.0312055435730347,
      "grad_norm": 0.47724539041519165,
      "learning_rate": 7.540657670225363e-06,
      "loss": 0.3566,
      "step": 22471
    },
    {
      "epoch": 1.0312514340782892,
      "grad_norm": 0.4534601867198944,
      "learning_rate": 7.540446491857379e-06,
      "loss": 0.3138,
      "step": 22472
    },
    {
      "epoch": 1.0312973245835437,
      "grad_norm": 0.44626832008361816,
      "learning_rate": 7.540235307380422e-06,
      "loss": 0.3465,
      "step": 22473
    },
    {
      "epoch": 1.0313432150887982,
      "grad_norm": 0.49037685990333557,
      "learning_rate": 7.540024116795002e-06,
      "loss": 0.3913,
      "step": 22474
    },
    {
      "epoch": 1.0313891055940525,
      "grad_norm": 0.43536630272865295,
      "learning_rate": 7.539812920101626e-06,
      "loss": 0.3111,
      "step": 22475
    },
    {
      "epoch": 1.031434996099307,
      "grad_norm": 0.4773887097835541,
      "learning_rate": 7.539601717300802e-06,
      "loss": 0.3564,
      "step": 22476
    },
    {
      "epoch": 1.0314808866045615,
      "grad_norm": 0.41378337144851685,
      "learning_rate": 7.539390508393035e-06,
      "loss": 0.3069,
      "step": 22477
    },
    {
      "epoch": 1.031526777109816,
      "grad_norm": 0.4721636176109314,
      "learning_rate": 7.539179293378835e-06,
      "loss": 0.3578,
      "step": 22478
    },
    {
      "epoch": 1.0315726676150705,
      "grad_norm": 0.44223642349243164,
      "learning_rate": 7.538968072258713e-06,
      "loss": 0.3562,
      "step": 22479
    },
    {
      "epoch": 1.031618558120325,
      "grad_norm": 0.45466575026512146,
      "learning_rate": 7.538756845033173e-06,
      "loss": 0.3438,
      "step": 22480
    },
    {
      "epoch": 1.0316644486255793,
      "grad_norm": 0.44065895676612854,
      "learning_rate": 7.538545611702723e-06,
      "loss": 0.3637,
      "step": 22481
    },
    {
      "epoch": 1.0317103391308338,
      "grad_norm": 0.4750717580318451,
      "learning_rate": 7.538334372267875e-06,
      "loss": 0.3592,
      "step": 22482
    },
    {
      "epoch": 1.0317562296360883,
      "grad_norm": 0.6003902554512024,
      "learning_rate": 7.538123126729131e-06,
      "loss": 0.2882,
      "step": 22483
    },
    {
      "epoch": 1.0318021201413428,
      "grad_norm": 0.49109312891960144,
      "learning_rate": 7.5379118750870016e-06,
      "loss": 0.4042,
      "step": 22484
    },
    {
      "epoch": 1.0318480106465973,
      "grad_norm": 0.48089855909347534,
      "learning_rate": 7.537700617341996e-06,
      "loss": 0.3814,
      "step": 22485
    },
    {
      "epoch": 1.0318939011518518,
      "grad_norm": 0.4734475314617157,
      "learning_rate": 7.537489353494622e-06,
      "loss": 0.3347,
      "step": 22486
    },
    {
      "epoch": 1.031939791657106,
      "grad_norm": 0.4818091094493866,
      "learning_rate": 7.537278083545388e-06,
      "loss": 0.331,
      "step": 22487
    },
    {
      "epoch": 1.0319856821623605,
      "grad_norm": 0.47763901948928833,
      "learning_rate": 7.537066807494799e-06,
      "loss": 0.3867,
      "step": 22488
    },
    {
      "epoch": 1.032031572667615,
      "grad_norm": 0.45094922184944153,
      "learning_rate": 7.536855525343365e-06,
      "loss": 0.35,
      "step": 22489
    },
    {
      "epoch": 1.0320774631728695,
      "grad_norm": 0.4774303436279297,
      "learning_rate": 7.536644237091595e-06,
      "loss": 0.3475,
      "step": 22490
    },
    {
      "epoch": 1.032123353678124,
      "grad_norm": 0.4614737629890442,
      "learning_rate": 7.536432942739995e-06,
      "loss": 0.334,
      "step": 22491
    },
    {
      "epoch": 1.0321692441833785,
      "grad_norm": 0.44377779960632324,
      "learning_rate": 7.5362216422890766e-06,
      "loss": 0.3142,
      "step": 22492
    },
    {
      "epoch": 1.032215134688633,
      "grad_norm": 0.454022079706192,
      "learning_rate": 7.536010335739346e-06,
      "loss": 0.3599,
      "step": 22493
    },
    {
      "epoch": 1.0322610251938873,
      "grad_norm": 0.4996393322944641,
      "learning_rate": 7.535799023091308e-06,
      "loss": 0.418,
      "step": 22494
    },
    {
      "epoch": 1.0323069156991418,
      "grad_norm": 0.48204198479652405,
      "learning_rate": 7.535587704345474e-06,
      "loss": 0.4166,
      "step": 22495
    },
    {
      "epoch": 1.0323528062043963,
      "grad_norm": 0.4685451090335846,
      "learning_rate": 7.535376379502353e-06,
      "loss": 0.3339,
      "step": 22496
    },
    {
      "epoch": 1.0323986967096508,
      "grad_norm": 0.4433494806289673,
      "learning_rate": 7.535165048562452e-06,
      "loss": 0.3454,
      "step": 22497
    },
    {
      "epoch": 1.0324445872149053,
      "grad_norm": 0.4790070056915283,
      "learning_rate": 7.534953711526279e-06,
      "loss": 0.371,
      "step": 22498
    },
    {
      "epoch": 1.0324904777201598,
      "grad_norm": 0.516436755657196,
      "learning_rate": 7.5347423683943425e-06,
      "loss": 0.3939,
      "step": 22499
    },
    {
      "epoch": 1.032536368225414,
      "grad_norm": 0.44650667905807495,
      "learning_rate": 7.53453101916715e-06,
      "loss": 0.3283,
      "step": 22500
    },
    {
      "epoch": 1.0325822587306686,
      "grad_norm": 0.4918571412563324,
      "learning_rate": 7.534319663845209e-06,
      "loss": 0.4,
      "step": 22501
    },
    {
      "epoch": 1.032628149235923,
      "grad_norm": 0.4394564628601074,
      "learning_rate": 7.534108302429032e-06,
      "loss": 0.3027,
      "step": 22502
    },
    {
      "epoch": 1.0326740397411776,
      "grad_norm": 0.4654005765914917,
      "learning_rate": 7.533896934919122e-06,
      "loss": 0.3092,
      "step": 22503
    },
    {
      "epoch": 1.032719930246432,
      "grad_norm": 0.4264509677886963,
      "learning_rate": 7.533685561315991e-06,
      "loss": 0.2842,
      "step": 22504
    },
    {
      "epoch": 1.0327658207516865,
      "grad_norm": 0.43899038434028625,
      "learning_rate": 7.533474181620146e-06,
      "loss": 0.3363,
      "step": 22505
    },
    {
      "epoch": 1.032811711256941,
      "grad_norm": 0.4411478340625763,
      "learning_rate": 7.533262795832095e-06,
      "loss": 0.3095,
      "step": 22506
    },
    {
      "epoch": 1.0328576017621953,
      "grad_norm": 0.5276771187782288,
      "learning_rate": 7.533051403952345e-06,
      "loss": 0.4097,
      "step": 22507
    },
    {
      "epoch": 1.0329034922674498,
      "grad_norm": 0.501353919506073,
      "learning_rate": 7.532840005981407e-06,
      "loss": 0.3623,
      "step": 22508
    },
    {
      "epoch": 1.0329493827727043,
      "grad_norm": 0.4690876603126526,
      "learning_rate": 7.532628601919789e-06,
      "loss": 0.3719,
      "step": 22509
    },
    {
      "epoch": 1.0329952732779588,
      "grad_norm": 0.4469498097896576,
      "learning_rate": 7.5324171917679974e-06,
      "loss": 0.3235,
      "step": 22510
    },
    {
      "epoch": 1.0330411637832133,
      "grad_norm": 0.45745769143104553,
      "learning_rate": 7.532205775526542e-06,
      "loss": 0.3924,
      "step": 22511
    },
    {
      "epoch": 1.0330870542884678,
      "grad_norm": 0.4838915765285492,
      "learning_rate": 7.531994353195931e-06,
      "loss": 0.3445,
      "step": 22512
    },
    {
      "epoch": 1.033132944793722,
      "grad_norm": 0.4959193170070648,
      "learning_rate": 7.531782924776672e-06,
      "loss": 0.3854,
      "step": 22513
    },
    {
      "epoch": 1.0331788352989766,
      "grad_norm": 0.49049198627471924,
      "learning_rate": 7.531571490269276e-06,
      "loss": 0.3288,
      "step": 22514
    },
    {
      "epoch": 1.033224725804231,
      "grad_norm": 0.4525477886199951,
      "learning_rate": 7.531360049674248e-06,
      "loss": 0.3098,
      "step": 22515
    },
    {
      "epoch": 1.0332706163094856,
      "grad_norm": 0.4442325532436371,
      "learning_rate": 7.531148602992098e-06,
      "loss": 0.287,
      "step": 22516
    },
    {
      "epoch": 1.03331650681474,
      "grad_norm": 0.4822787344455719,
      "learning_rate": 7.530937150223334e-06,
      "loss": 0.412,
      "step": 22517
    },
    {
      "epoch": 1.0333623973199946,
      "grad_norm": 0.5001789331436157,
      "learning_rate": 7.530725691368466e-06,
      "loss": 0.3966,
      "step": 22518
    },
    {
      "epoch": 1.0334082878252489,
      "grad_norm": 0.4512426257133484,
      "learning_rate": 7.5305142264280005e-06,
      "loss": 0.3156,
      "step": 22519
    },
    {
      "epoch": 1.0334541783305033,
      "grad_norm": 0.4955967962741852,
      "learning_rate": 7.530302755402449e-06,
      "loss": 0.4016,
      "step": 22520
    },
    {
      "epoch": 1.0335000688357578,
      "grad_norm": 0.48659852147102356,
      "learning_rate": 7.530091278292315e-06,
      "loss": 0.404,
      "step": 22521
    },
    {
      "epoch": 1.0335459593410123,
      "grad_norm": 0.4869542717933655,
      "learning_rate": 7.529879795098113e-06,
      "loss": 0.3817,
      "step": 22522
    },
    {
      "epoch": 1.0335918498462668,
      "grad_norm": 0.510161817073822,
      "learning_rate": 7.529668305820346e-06,
      "loss": 0.4427,
      "step": 22523
    },
    {
      "epoch": 1.0336377403515213,
      "grad_norm": 0.4754716455936432,
      "learning_rate": 7.5294568104595276e-06,
      "loss": 0.3747,
      "step": 22524
    },
    {
      "epoch": 1.0336836308567758,
      "grad_norm": 0.5018302798271179,
      "learning_rate": 7.529245309016162e-06,
      "loss": 0.4186,
      "step": 22525
    },
    {
      "epoch": 1.0337295213620301,
      "grad_norm": 0.48589888215065,
      "learning_rate": 7.529033801490759e-06,
      "loss": 0.4025,
      "step": 22526
    },
    {
      "epoch": 1.0337754118672846,
      "grad_norm": 0.4735371768474579,
      "learning_rate": 7.5288222878838305e-06,
      "loss": 0.3753,
      "step": 22527
    },
    {
      "epoch": 1.033821302372539,
      "grad_norm": 0.46141767501831055,
      "learning_rate": 7.528610768195881e-06,
      "loss": 0.3529,
      "step": 22528
    },
    {
      "epoch": 1.0338671928777936,
      "grad_norm": 0.42864179611206055,
      "learning_rate": 7.52839924242742e-06,
      "loss": 0.3034,
      "step": 22529
    },
    {
      "epoch": 1.033913083383048,
      "grad_norm": 0.4565361738204956,
      "learning_rate": 7.528187710578958e-06,
      "loss": 0.313,
      "step": 22530
    },
    {
      "epoch": 1.0339589738883026,
      "grad_norm": 0.5064765810966492,
      "learning_rate": 7.527976172651002e-06,
      "loss": 0.3511,
      "step": 22531
    },
    {
      "epoch": 1.0340048643935569,
      "grad_norm": 0.4575657546520233,
      "learning_rate": 7.527764628644061e-06,
      "loss": 0.3425,
      "step": 22532
    },
    {
      "epoch": 1.0340507548988114,
      "grad_norm": 0.4638778567314148,
      "learning_rate": 7.5275530785586445e-06,
      "loss": 0.3382,
      "step": 22533
    },
    {
      "epoch": 1.0340966454040659,
      "grad_norm": 0.4642266035079956,
      "learning_rate": 7.52734152239526e-06,
      "loss": 0.3906,
      "step": 22534
    },
    {
      "epoch": 1.0341425359093204,
      "grad_norm": 0.47281980514526367,
      "learning_rate": 7.527129960154416e-06,
      "loss": 0.3753,
      "step": 22535
    },
    {
      "epoch": 1.0341884264145749,
      "grad_norm": 0.4999763071537018,
      "learning_rate": 7.526918391836624e-06,
      "loss": 0.4138,
      "step": 22536
    },
    {
      "epoch": 1.0342343169198294,
      "grad_norm": 0.457609087228775,
      "learning_rate": 7.526706817442388e-06,
      "loss": 0.3828,
      "step": 22537
    },
    {
      "epoch": 1.0342802074250836,
      "grad_norm": 0.4578341543674469,
      "learning_rate": 7.526495236972222e-06,
      "loss": 0.3289,
      "step": 22538
    },
    {
      "epoch": 1.0343260979303381,
      "grad_norm": 0.5044934749603271,
      "learning_rate": 7.526283650426631e-06,
      "loss": 0.4268,
      "step": 22539
    },
    {
      "epoch": 1.0343719884355926,
      "grad_norm": 0.46051138639450073,
      "learning_rate": 7.526072057806127e-06,
      "loss": 0.3223,
      "step": 22540
    },
    {
      "epoch": 1.0344178789408471,
      "grad_norm": 0.49362966418266296,
      "learning_rate": 7.525860459111215e-06,
      "loss": 0.3809,
      "step": 22541
    },
    {
      "epoch": 1.0344637694461016,
      "grad_norm": 0.4496925473213196,
      "learning_rate": 7.525648854342406e-06,
      "loss": 0.3493,
      "step": 22542
    },
    {
      "epoch": 1.0345096599513561,
      "grad_norm": 0.4411328434944153,
      "learning_rate": 7.525437243500209e-06,
      "loss": 0.3199,
      "step": 22543
    },
    {
      "epoch": 1.0345555504566106,
      "grad_norm": 0.4550195634365082,
      "learning_rate": 7.525225626585133e-06,
      "loss": 0.3437,
      "step": 22544
    },
    {
      "epoch": 1.034601440961865,
      "grad_norm": 0.44751280546188354,
      "learning_rate": 7.525014003597686e-06,
      "loss": 0.3359,
      "step": 22545
    },
    {
      "epoch": 1.0346473314671194,
      "grad_norm": 0.4736498296260834,
      "learning_rate": 7.524802374538378e-06,
      "loss": 0.4137,
      "step": 22546
    },
    {
      "epoch": 1.034693221972374,
      "grad_norm": 0.4899050295352936,
      "learning_rate": 7.524590739407715e-06,
      "loss": 0.3337,
      "step": 22547
    },
    {
      "epoch": 1.0347391124776284,
      "grad_norm": 0.47566068172454834,
      "learning_rate": 7.524379098206209e-06,
      "loss": 0.368,
      "step": 22548
    },
    {
      "epoch": 1.034785002982883,
      "grad_norm": 0.496096134185791,
      "learning_rate": 7.524167450934369e-06,
      "loss": 0.3804,
      "step": 22549
    },
    {
      "epoch": 1.0348308934881374,
      "grad_norm": 0.4961836636066437,
      "learning_rate": 7.523955797592703e-06,
      "loss": 0.3503,
      "step": 22550
    },
    {
      "epoch": 1.0348767839933917,
      "grad_norm": 0.4667801260948181,
      "learning_rate": 7.5237441381817185e-06,
      "loss": 0.3427,
      "step": 22551
    },
    {
      "epoch": 1.0349226744986462,
      "grad_norm": 0.522922933101654,
      "learning_rate": 7.523532472701926e-06,
      "loss": 0.4887,
      "step": 22552
    },
    {
      "epoch": 1.0349685650039007,
      "grad_norm": 0.4789958894252777,
      "learning_rate": 7.5233208011538364e-06,
      "loss": 0.3663,
      "step": 22553
    },
    {
      "epoch": 1.0350144555091552,
      "grad_norm": 0.4837660491466522,
      "learning_rate": 7.523109123537955e-06,
      "loss": 0.4513,
      "step": 22554
    },
    {
      "epoch": 1.0350603460144097,
      "grad_norm": 0.47886380553245544,
      "learning_rate": 7.52289743985479e-06,
      "loss": 0.3436,
      "step": 22555
    },
    {
      "epoch": 1.0351062365196642,
      "grad_norm": 0.5638324618339539,
      "learning_rate": 7.522685750104857e-06,
      "loss": 0.3796,
      "step": 22556
    },
    {
      "epoch": 1.0351521270249187,
      "grad_norm": 0.5084211230278015,
      "learning_rate": 7.52247405428866e-06,
      "loss": 0.3719,
      "step": 22557
    },
    {
      "epoch": 1.035198017530173,
      "grad_norm": 0.4882853627204895,
      "learning_rate": 7.5222623524067086e-06,
      "loss": 0.4071,
      "step": 22558
    },
    {
      "epoch": 1.0352439080354274,
      "grad_norm": 0.5179264545440674,
      "learning_rate": 7.5220506444595125e-06,
      "loss": 0.352,
      "step": 22559
    },
    {
      "epoch": 1.035289798540682,
      "grad_norm": 0.5166230797767639,
      "learning_rate": 7.521838930447579e-06,
      "loss": 0.3289,
      "step": 22560
    },
    {
      "epoch": 1.0353356890459364,
      "grad_norm": 0.4785296320915222,
      "learning_rate": 7.521627210371421e-06,
      "loss": 0.292,
      "step": 22561
    },
    {
      "epoch": 1.035381579551191,
      "grad_norm": 0.46852248907089233,
      "learning_rate": 7.521415484231544e-06,
      "loss": 0.3439,
      "step": 22562
    },
    {
      "epoch": 1.0354274700564454,
      "grad_norm": 0.45344555377960205,
      "learning_rate": 7.521203752028459e-06,
      "loss": 0.3172,
      "step": 22563
    },
    {
      "epoch": 1.0354733605616997,
      "grad_norm": 0.48477643728256226,
      "learning_rate": 7.5209920137626755e-06,
      "loss": 0.3688,
      "step": 22564
    },
    {
      "epoch": 1.0355192510669542,
      "grad_norm": 0.4915164113044739,
      "learning_rate": 7.5207802694347e-06,
      "loss": 0.3235,
      "step": 22565
    },
    {
      "epoch": 1.0355651415722087,
      "grad_norm": 0.5388010740280151,
      "learning_rate": 7.520568519045044e-06,
      "loss": 0.3831,
      "step": 22566
    },
    {
      "epoch": 1.0356110320774632,
      "grad_norm": 0.6704277396202087,
      "learning_rate": 7.520356762594218e-06,
      "loss": 0.2776,
      "step": 22567
    },
    {
      "epoch": 1.0356569225827177,
      "grad_norm": 0.48292919993400574,
      "learning_rate": 7.520145000082728e-06,
      "loss": 0.3918,
      "step": 22568
    },
    {
      "epoch": 1.0357028130879722,
      "grad_norm": 0.4321908950805664,
      "learning_rate": 7.519933231511085e-06,
      "loss": 0.3141,
      "step": 22569
    },
    {
      "epoch": 1.0357487035932265,
      "grad_norm": 0.4793359935283661,
      "learning_rate": 7.519721456879798e-06,
      "loss": 0.4296,
      "step": 22570
    },
    {
      "epoch": 1.035794594098481,
      "grad_norm": 0.4791930317878723,
      "learning_rate": 7.519509676189375e-06,
      "loss": 0.3902,
      "step": 22571
    },
    {
      "epoch": 1.0358404846037355,
      "grad_norm": 0.4553140103816986,
      "learning_rate": 7.519297889440327e-06,
      "loss": 0.3165,
      "step": 22572
    },
    {
      "epoch": 1.03588637510899,
      "grad_norm": 0.4684256315231323,
      "learning_rate": 7.519086096633163e-06,
      "loss": 0.4146,
      "step": 22573
    },
    {
      "epoch": 1.0359322656142445,
      "grad_norm": 0.47407516837120056,
      "learning_rate": 7.518874297768392e-06,
      "loss": 0.309,
      "step": 22574
    },
    {
      "epoch": 1.035978156119499,
      "grad_norm": 0.46104568243026733,
      "learning_rate": 7.518662492846524e-06,
      "loss": 0.3538,
      "step": 22575
    },
    {
      "epoch": 1.0360240466247532,
      "grad_norm": 0.455872118473053,
      "learning_rate": 7.518450681868067e-06,
      "loss": 0.348,
      "step": 22576
    },
    {
      "epoch": 1.0360699371300077,
      "grad_norm": 0.4080817997455597,
      "learning_rate": 7.51823886483353e-06,
      "loss": 0.2732,
      "step": 22577
    },
    {
      "epoch": 1.0361158276352622,
      "grad_norm": 0.4523342549800873,
      "learning_rate": 7.5180270417434256e-06,
      "loss": 0.371,
      "step": 22578
    },
    {
      "epoch": 1.0361617181405167,
      "grad_norm": 0.5079509019851685,
      "learning_rate": 7.5178152125982585e-06,
      "loss": 0.3771,
      "step": 22579
    },
    {
      "epoch": 1.0362076086457712,
      "grad_norm": 0.5014461278915405,
      "learning_rate": 7.517603377398541e-06,
      "loss": 0.4585,
      "step": 22580
    },
    {
      "epoch": 1.0362534991510257,
      "grad_norm": 0.4383416473865509,
      "learning_rate": 7.517391536144783e-06,
      "loss": 0.3451,
      "step": 22581
    },
    {
      "epoch": 1.0362993896562802,
      "grad_norm": 0.492683082818985,
      "learning_rate": 7.517179688837493e-06,
      "loss": 0.3875,
      "step": 22582
    },
    {
      "epoch": 1.0363452801615345,
      "grad_norm": 0.4581277668476105,
      "learning_rate": 7.516967835477178e-06,
      "loss": 0.323,
      "step": 22583
    },
    {
      "epoch": 1.036391170666789,
      "grad_norm": 0.4390082359313965,
      "learning_rate": 7.516755976064352e-06,
      "loss": 0.3261,
      "step": 22584
    },
    {
      "epoch": 1.0364370611720435,
      "grad_norm": 0.48436179757118225,
      "learning_rate": 7.5165441105995215e-06,
      "loss": 0.3806,
      "step": 22585
    },
    {
      "epoch": 1.036482951677298,
      "grad_norm": 0.4714542329311371,
      "learning_rate": 7.516332239083196e-06,
      "loss": 0.3279,
      "step": 22586
    },
    {
      "epoch": 1.0365288421825525,
      "grad_norm": 0.4567766785621643,
      "learning_rate": 7.516120361515886e-06,
      "loss": 0.3457,
      "step": 22587
    },
    {
      "epoch": 1.036574732687807,
      "grad_norm": 0.4897473156452179,
      "learning_rate": 7.515908477898102e-06,
      "loss": 0.3952,
      "step": 22588
    },
    {
      "epoch": 1.0366206231930613,
      "grad_norm": 0.4591223895549774,
      "learning_rate": 7.51569658823035e-06,
      "loss": 0.3352,
      "step": 22589
    },
    {
      "epoch": 1.0366665136983158,
      "grad_norm": 0.425460547208786,
      "learning_rate": 7.515484692513144e-06,
      "loss": 0.2697,
      "step": 22590
    },
    {
      "epoch": 1.0367124042035702,
      "grad_norm": 0.4897994101047516,
      "learning_rate": 7.5152727907469895e-06,
      "loss": 0.3967,
      "step": 22591
    },
    {
      "epoch": 1.0367582947088247,
      "grad_norm": 0.45869019627571106,
      "learning_rate": 7.515060882932399e-06,
      "loss": 0.3638,
      "step": 22592
    },
    {
      "epoch": 1.0368041852140792,
      "grad_norm": 0.46208545565605164,
      "learning_rate": 7.514848969069879e-06,
      "loss": 0.3789,
      "step": 22593
    },
    {
      "epoch": 1.0368500757193337,
      "grad_norm": 0.45192182064056396,
      "learning_rate": 7.514637049159941e-06,
      "loss": 0.301,
      "step": 22594
    },
    {
      "epoch": 1.0368959662245882,
      "grad_norm": 0.476268470287323,
      "learning_rate": 7.5144251232030975e-06,
      "loss": 0.3269,
      "step": 22595
    },
    {
      "epoch": 1.0369418567298425,
      "grad_norm": 0.4789285361766815,
      "learning_rate": 7.5142131911998524e-06,
      "loss": 0.2842,
      "step": 22596
    },
    {
      "epoch": 1.036987747235097,
      "grad_norm": 0.533838152885437,
      "learning_rate": 7.514001253150719e-06,
      "loss": 0.3816,
      "step": 22597
    },
    {
      "epoch": 1.0370336377403515,
      "grad_norm": 0.4740827679634094,
      "learning_rate": 7.513789309056205e-06,
      "loss": 0.3739,
      "step": 22598
    },
    {
      "epoch": 1.037079528245606,
      "grad_norm": 0.4522237479686737,
      "learning_rate": 7.513577358916822e-06,
      "loss": 0.3388,
      "step": 22599
    },
    {
      "epoch": 1.0371254187508605,
      "grad_norm": 0.4588858187198639,
      "learning_rate": 7.5133654027330795e-06,
      "loss": 0.3045,
      "step": 22600
    },
    {
      "epoch": 1.037171309256115,
      "grad_norm": 0.5120354294776917,
      "learning_rate": 7.513153440505486e-06,
      "loss": 0.4203,
      "step": 22601
    },
    {
      "epoch": 1.0372171997613693,
      "grad_norm": 0.4492606818675995,
      "learning_rate": 7.512941472234552e-06,
      "loss": 0.3582,
      "step": 22602
    },
    {
      "epoch": 1.0372630902666238,
      "grad_norm": 0.5001031756401062,
      "learning_rate": 7.512729497920786e-06,
      "loss": 0.4383,
      "step": 22603
    },
    {
      "epoch": 1.0373089807718783,
      "grad_norm": 0.4281148910522461,
      "learning_rate": 7.512517517564698e-06,
      "loss": 0.301,
      "step": 22604
    },
    {
      "epoch": 1.0373548712771328,
      "grad_norm": 0.46850812435150146,
      "learning_rate": 7.512305531166799e-06,
      "loss": 0.3786,
      "step": 22605
    },
    {
      "epoch": 1.0374007617823873,
      "grad_norm": 0.44978564977645874,
      "learning_rate": 7.5120935387275984e-06,
      "loss": 0.3337,
      "step": 22606
    },
    {
      "epoch": 1.0374466522876418,
      "grad_norm": 0.4837632179260254,
      "learning_rate": 7.511881540247607e-06,
      "loss": 0.4107,
      "step": 22607
    },
    {
      "epoch": 1.037492542792896,
      "grad_norm": 0.7123207449913025,
      "learning_rate": 7.511669535727331e-06,
      "loss": 0.4368,
      "step": 22608
    },
    {
      "epoch": 1.0375384332981505,
      "grad_norm": 0.4978242516517639,
      "learning_rate": 7.511457525167283e-06,
      "loss": 0.3347,
      "step": 22609
    },
    {
      "epoch": 1.037584323803405,
      "grad_norm": 0.4432408809661865,
      "learning_rate": 7.5112455085679725e-06,
      "loss": 0.3154,
      "step": 22610
    },
    {
      "epoch": 1.0376302143086595,
      "grad_norm": 0.5042496919631958,
      "learning_rate": 7.511033485929908e-06,
      "loss": 0.3367,
      "step": 22611
    },
    {
      "epoch": 1.037676104813914,
      "grad_norm": 0.5159973502159119,
      "learning_rate": 7.510821457253603e-06,
      "loss": 0.4577,
      "step": 22612
    },
    {
      "epoch": 1.0377219953191685,
      "grad_norm": 0.4588887393474579,
      "learning_rate": 7.510609422539563e-06,
      "loss": 0.283,
      "step": 22613
    },
    {
      "epoch": 1.037767885824423,
      "grad_norm": 0.4738307297229767,
      "learning_rate": 7.5103973817883e-06,
      "loss": 0.3621,
      "step": 22614
    },
    {
      "epoch": 1.0378137763296773,
      "grad_norm": 0.4923470914363861,
      "learning_rate": 7.510185335000324e-06,
      "loss": 0.3967,
      "step": 22615
    },
    {
      "epoch": 1.0378596668349318,
      "grad_norm": 0.44362521171569824,
      "learning_rate": 7.509973282176144e-06,
      "loss": 0.3437,
      "step": 22616
    },
    {
      "epoch": 1.0379055573401863,
      "grad_norm": 0.4708520770072937,
      "learning_rate": 7.509761223316272e-06,
      "loss": 0.37,
      "step": 22617
    },
    {
      "epoch": 1.0379514478454408,
      "grad_norm": 0.42257457971572876,
      "learning_rate": 7.509549158421215e-06,
      "loss": 0.3076,
      "step": 22618
    },
    {
      "epoch": 1.0379973383506953,
      "grad_norm": 0.44826650619506836,
      "learning_rate": 7.509337087491484e-06,
      "loss": 0.3493,
      "step": 22619
    },
    {
      "epoch": 1.0380432288559498,
      "grad_norm": 0.5177218317985535,
      "learning_rate": 7.509125010527589e-06,
      "loss": 0.4741,
      "step": 22620
    },
    {
      "epoch": 1.038089119361204,
      "grad_norm": 0.45172804594039917,
      "learning_rate": 7.508912927530041e-06,
      "loss": 0.3258,
      "step": 22621
    },
    {
      "epoch": 1.0381350098664586,
      "grad_norm": 0.47864997386932373,
      "learning_rate": 7.50870083849935e-06,
      "loss": 0.366,
      "step": 22622
    },
    {
      "epoch": 1.038180900371713,
      "grad_norm": 0.4468690752983093,
      "learning_rate": 7.508488743436025e-06,
      "loss": 0.3334,
      "step": 22623
    },
    {
      "epoch": 1.0382267908769676,
      "grad_norm": 0.4550061821937561,
      "learning_rate": 7.508276642340575e-06,
      "loss": 0.3672,
      "step": 22624
    },
    {
      "epoch": 1.038272681382222,
      "grad_norm": 0.4632030129432678,
      "learning_rate": 7.5080645352135115e-06,
      "loss": 0.3787,
      "step": 22625
    },
    {
      "epoch": 1.0383185718874766,
      "grad_norm": 0.5070819854736328,
      "learning_rate": 7.507852422055346e-06,
      "loss": 0.4104,
      "step": 22626
    },
    {
      "epoch": 1.0383644623927308,
      "grad_norm": 0.43967822194099426,
      "learning_rate": 7.507640302866586e-06,
      "loss": 0.3264,
      "step": 22627
    },
    {
      "epoch": 1.0384103528979853,
      "grad_norm": 0.452687531709671,
      "learning_rate": 7.507428177647742e-06,
      "loss": 0.2952,
      "step": 22628
    },
    {
      "epoch": 1.0384562434032398,
      "grad_norm": 0.5061430335044861,
      "learning_rate": 7.507216046399326e-06,
      "loss": 0.4171,
      "step": 22629
    },
    {
      "epoch": 1.0385021339084943,
      "grad_norm": 0.4669508635997772,
      "learning_rate": 7.507003909121844e-06,
      "loss": 0.353,
      "step": 22630
    },
    {
      "epoch": 1.0385480244137488,
      "grad_norm": 0.46823179721832275,
      "learning_rate": 7.506791765815811e-06,
      "loss": 0.3541,
      "step": 22631
    },
    {
      "epoch": 1.0385939149190033,
      "grad_norm": 0.4741915762424469,
      "learning_rate": 7.506579616481734e-06,
      "loss": 0.3441,
      "step": 22632
    },
    {
      "epoch": 1.0386398054242578,
      "grad_norm": 0.4528752863407135,
      "learning_rate": 7.506367461120124e-06,
      "loss": 0.3231,
      "step": 22633
    },
    {
      "epoch": 1.038685695929512,
      "grad_norm": 0.4942726492881775,
      "learning_rate": 7.5061552997314925e-06,
      "loss": 0.3764,
      "step": 22634
    },
    {
      "epoch": 1.0387315864347666,
      "grad_norm": 0.5069977641105652,
      "learning_rate": 7.505943132316349e-06,
      "loss": 0.4211,
      "step": 22635
    },
    {
      "epoch": 1.038777476940021,
      "grad_norm": 0.4757709205150604,
      "learning_rate": 7.505730958875202e-06,
      "loss": 0.3672,
      "step": 22636
    },
    {
      "epoch": 1.0388233674452756,
      "grad_norm": 0.4170849025249481,
      "learning_rate": 7.50551877940856e-06,
      "loss": 0.2808,
      "step": 22637
    },
    {
      "epoch": 1.03886925795053,
      "grad_norm": 0.4478960633277893,
      "learning_rate": 7.505306593916941e-06,
      "loss": 0.31,
      "step": 22638
    },
    {
      "epoch": 1.0389151484557846,
      "grad_norm": 0.48747068643569946,
      "learning_rate": 7.5050944024008475e-06,
      "loss": 0.4127,
      "step": 22639
    },
    {
      "epoch": 1.0389610389610389,
      "grad_norm": 0.4746107757091522,
      "learning_rate": 7.504882204860795e-06,
      "loss": 0.3455,
      "step": 22640
    },
    {
      "epoch": 1.0390069294662934,
      "grad_norm": 0.43102437257766724,
      "learning_rate": 7.504670001297289e-06,
      "loss": 0.2999,
      "step": 22641
    },
    {
      "epoch": 1.0390528199715479,
      "grad_norm": 0.5030134320259094,
      "learning_rate": 7.504457791710842e-06,
      "loss": 0.4436,
      "step": 22642
    },
    {
      "epoch": 1.0390987104768024,
      "grad_norm": 0.5042721629142761,
      "learning_rate": 7.504245576101966e-06,
      "loss": 0.387,
      "step": 22643
    },
    {
      "epoch": 1.0391446009820569,
      "grad_norm": 0.4811035692691803,
      "learning_rate": 7.504033354471168e-06,
      "loss": 0.3678,
      "step": 22644
    },
    {
      "epoch": 1.0391904914873114,
      "grad_norm": 0.4951101243495941,
      "learning_rate": 7.503821126818962e-06,
      "loss": 0.3788,
      "step": 22645
    },
    {
      "epoch": 1.0392363819925658,
      "grad_norm": 0.5301887392997742,
      "learning_rate": 7.503608893145856e-06,
      "loss": 0.4489,
      "step": 22646
    },
    {
      "epoch": 1.0392822724978201,
      "grad_norm": 0.4571898579597473,
      "learning_rate": 7.503396653452359e-06,
      "loss": 0.3303,
      "step": 22647
    },
    {
      "epoch": 1.0393281630030746,
      "grad_norm": 0.4381439685821533,
      "learning_rate": 7.503184407738985e-06,
      "loss": 0.336,
      "step": 22648
    },
    {
      "epoch": 1.0393740535083291,
      "grad_norm": 0.5590527653694153,
      "learning_rate": 7.502972156006243e-06,
      "loss": 0.4194,
      "step": 22649
    },
    {
      "epoch": 1.0394199440135836,
      "grad_norm": 0.4503445327281952,
      "learning_rate": 7.502759898254641e-06,
      "loss": 0.3431,
      "step": 22650
    },
    {
      "epoch": 1.0394658345188381,
      "grad_norm": 0.46255195140838623,
      "learning_rate": 7.502547634484693e-06,
      "loss": 0.3629,
      "step": 22651
    },
    {
      "epoch": 1.0395117250240926,
      "grad_norm": 0.4917968809604645,
      "learning_rate": 7.502335364696907e-06,
      "loss": 0.3681,
      "step": 22652
    },
    {
      "epoch": 1.039557615529347,
      "grad_norm": 0.479777455329895,
      "learning_rate": 7.502123088891794e-06,
      "loss": 0.3602,
      "step": 22653
    },
    {
      "epoch": 1.0396035060346014,
      "grad_norm": 0.45955249667167664,
      "learning_rate": 7.501910807069864e-06,
      "loss": 0.3248,
      "step": 22654
    },
    {
      "epoch": 1.0396493965398559,
      "grad_norm": 0.44344815611839294,
      "learning_rate": 7.50169851923163e-06,
      "loss": 0.3435,
      "step": 22655
    },
    {
      "epoch": 1.0396952870451104,
      "grad_norm": 0.4244428873062134,
      "learning_rate": 7.5014862253776e-06,
      "loss": 0.2685,
      "step": 22656
    },
    {
      "epoch": 1.0397411775503649,
      "grad_norm": 0.4281753897666931,
      "learning_rate": 7.501273925508286e-06,
      "loss": 0.29,
      "step": 22657
    },
    {
      "epoch": 1.0397870680556194,
      "grad_norm": 0.5082271099090576,
      "learning_rate": 7.5010616196241945e-06,
      "loss": 0.3853,
      "step": 22658
    },
    {
      "epoch": 1.0398329585608737,
      "grad_norm": 0.5000761151313782,
      "learning_rate": 7.500849307725842e-06,
      "loss": 0.3952,
      "step": 22659
    },
    {
      "epoch": 1.0398788490661282,
      "grad_norm": 0.46886980533599854,
      "learning_rate": 7.5006369898137344e-06,
      "loss": 0.3295,
      "step": 22660
    },
    {
      "epoch": 1.0399247395713827,
      "grad_norm": 0.45340049266815186,
      "learning_rate": 7.500424665888386e-06,
      "loss": 0.3238,
      "step": 22661
    },
    {
      "epoch": 1.0399706300766371,
      "grad_norm": 0.4961346685886383,
      "learning_rate": 7.500212335950303e-06,
      "loss": 0.3405,
      "step": 22662
    },
    {
      "epoch": 1.0400165205818916,
      "grad_norm": 0.4630239009857178,
      "learning_rate": 7.500000000000001e-06,
      "loss": 0.3059,
      "step": 22663
    },
    {
      "epoch": 1.0400624110871461,
      "grad_norm": 0.4659516513347626,
      "learning_rate": 7.4997876580379865e-06,
      "loss": 0.3297,
      "step": 22664
    },
    {
      "epoch": 1.0401083015924004,
      "grad_norm": 0.44990962743759155,
      "learning_rate": 7.4995753100647715e-06,
      "loss": 0.2981,
      "step": 22665
    },
    {
      "epoch": 1.040154192097655,
      "grad_norm": 0.519379734992981,
      "learning_rate": 7.4993629560808675e-06,
      "loss": 0.4338,
      "step": 22666
    },
    {
      "epoch": 1.0402000826029094,
      "grad_norm": 0.44374313950538635,
      "learning_rate": 7.499150596086785e-06,
      "loss": 0.2961,
      "step": 22667
    },
    {
      "epoch": 1.040245973108164,
      "grad_norm": 0.4624730050563812,
      "learning_rate": 7.498938230083032e-06,
      "loss": 0.342,
      "step": 22668
    },
    {
      "epoch": 1.0402918636134184,
      "grad_norm": 0.4602864682674408,
      "learning_rate": 7.498725858070124e-06,
      "loss": 0.3594,
      "step": 22669
    },
    {
      "epoch": 1.040337754118673,
      "grad_norm": 0.4900275468826294,
      "learning_rate": 7.4985134800485655e-06,
      "loss": 0.4081,
      "step": 22670
    },
    {
      "epoch": 1.0403836446239274,
      "grad_norm": 0.4650912880897522,
      "learning_rate": 7.498301096018874e-06,
      "loss": 0.3618,
      "step": 22671
    },
    {
      "epoch": 1.0404295351291817,
      "grad_norm": 0.5254382491111755,
      "learning_rate": 7.498088705981556e-06,
      "loss": 0.3991,
      "step": 22672
    },
    {
      "epoch": 1.0404754256344362,
      "grad_norm": 0.45846128463745117,
      "learning_rate": 7.497876309937122e-06,
      "loss": 0.3639,
      "step": 22673
    },
    {
      "epoch": 1.0405213161396907,
      "grad_norm": 0.4317995309829712,
      "learning_rate": 7.497663907886084e-06,
      "loss": 0.3293,
      "step": 22674
    },
    {
      "epoch": 1.0405672066449452,
      "grad_norm": 0.47075486183166504,
      "learning_rate": 7.497451499828952e-06,
      "loss": 0.3318,
      "step": 22675
    },
    {
      "epoch": 1.0406130971501997,
      "grad_norm": 0.4526347219944,
      "learning_rate": 7.497239085766238e-06,
      "loss": 0.3544,
      "step": 22676
    },
    {
      "epoch": 1.0406589876554542,
      "grad_norm": 0.4874648153781891,
      "learning_rate": 7.497026665698453e-06,
      "loss": 0.3455,
      "step": 22677
    },
    {
      "epoch": 1.0407048781607084,
      "grad_norm": 0.4680350124835968,
      "learning_rate": 7.496814239626107e-06,
      "loss": 0.4068,
      "step": 22678
    },
    {
      "epoch": 1.040750768665963,
      "grad_norm": 0.46096980571746826,
      "learning_rate": 7.496601807549709e-06,
      "loss": 0.3517,
      "step": 22679
    },
    {
      "epoch": 1.0407966591712174,
      "grad_norm": 0.4481469988822937,
      "learning_rate": 7.4963893694697744e-06,
      "loss": 0.3305,
      "step": 22680
    },
    {
      "epoch": 1.040842549676472,
      "grad_norm": 0.4577075242996216,
      "learning_rate": 7.496176925386809e-06,
      "loss": 0.3569,
      "step": 22681
    },
    {
      "epoch": 1.0408884401817264,
      "grad_norm": 0.47864237427711487,
      "learning_rate": 7.495964475301328e-06,
      "loss": 0.3418,
      "step": 22682
    },
    {
      "epoch": 1.040934330686981,
      "grad_norm": 0.4881422519683838,
      "learning_rate": 7.495752019213838e-06,
      "loss": 0.3744,
      "step": 22683
    },
    {
      "epoch": 1.0409802211922354,
      "grad_norm": 0.46769267320632935,
      "learning_rate": 7.495539557124851e-06,
      "loss": 0.3547,
      "step": 22684
    },
    {
      "epoch": 1.0410261116974897,
      "grad_norm": 0.4983559548854828,
      "learning_rate": 7.49532708903488e-06,
      "loss": 0.4137,
      "step": 22685
    },
    {
      "epoch": 1.0410720022027442,
      "grad_norm": 0.43223729729652405,
      "learning_rate": 7.495114614944438e-06,
      "loss": 0.2843,
      "step": 22686
    },
    {
      "epoch": 1.0411178927079987,
      "grad_norm": 0.45847636461257935,
      "learning_rate": 7.49490213485403e-06,
      "loss": 0.3344,
      "step": 22687
    },
    {
      "epoch": 1.0411637832132532,
      "grad_norm": 0.46122318506240845,
      "learning_rate": 7.4946896487641696e-06,
      "loss": 0.3316,
      "step": 22688
    },
    {
      "epoch": 1.0412096737185077,
      "grad_norm": 0.48886874318122864,
      "learning_rate": 7.4944771566753696e-06,
      "loss": 0.4359,
      "step": 22689
    },
    {
      "epoch": 1.0412555642237622,
      "grad_norm": 0.46576988697052,
      "learning_rate": 7.494264658588137e-06,
      "loss": 0.398,
      "step": 22690
    },
    {
      "epoch": 1.0413014547290165,
      "grad_norm": 0.4574759304523468,
      "learning_rate": 7.4940521545029875e-06,
      "loss": 0.3496,
      "step": 22691
    },
    {
      "epoch": 1.041347345234271,
      "grad_norm": 0.4674251973628998,
      "learning_rate": 7.493839644420428e-06,
      "loss": 0.3761,
      "step": 22692
    },
    {
      "epoch": 1.0413932357395255,
      "grad_norm": 0.47321823239326477,
      "learning_rate": 7.493627128340972e-06,
      "loss": 0.3713,
      "step": 22693
    },
    {
      "epoch": 1.04143912624478,
      "grad_norm": 0.48859670758247375,
      "learning_rate": 7.493414606265131e-06,
      "loss": 0.4428,
      "step": 22694
    },
    {
      "epoch": 1.0414850167500345,
      "grad_norm": 0.4617638885974884,
      "learning_rate": 7.493202078193412e-06,
      "loss": 0.3666,
      "step": 22695
    },
    {
      "epoch": 1.041530907255289,
      "grad_norm": 0.4848746657371521,
      "learning_rate": 7.49298954412633e-06,
      "loss": 0.3572,
      "step": 22696
    },
    {
      "epoch": 1.0415767977605432,
      "grad_norm": 0.4204821288585663,
      "learning_rate": 7.492777004064395e-06,
      "loss": 0.2638,
      "step": 22697
    },
    {
      "epoch": 1.0416226882657977,
      "grad_norm": 0.43197301030158997,
      "learning_rate": 7.492564458008118e-06,
      "loss": 0.3001,
      "step": 22698
    },
    {
      "epoch": 1.0416685787710522,
      "grad_norm": 0.4667450189590454,
      "learning_rate": 7.492351905958012e-06,
      "loss": 0.3788,
      "step": 22699
    },
    {
      "epoch": 1.0417144692763067,
      "grad_norm": 0.4375399947166443,
      "learning_rate": 7.492139347914584e-06,
      "loss": 0.2847,
      "step": 22700
    },
    {
      "epoch": 1.0417603597815612,
      "grad_norm": 0.4769582152366638,
      "learning_rate": 7.491926783878348e-06,
      "loss": 0.3806,
      "step": 22701
    },
    {
      "epoch": 1.0418062502868157,
      "grad_norm": 0.5065696835517883,
      "learning_rate": 7.491714213849814e-06,
      "loss": 0.4085,
      "step": 22702
    },
    {
      "epoch": 1.0418521407920702,
      "grad_norm": 0.5016975998878479,
      "learning_rate": 7.491501637829496e-06,
      "loss": 0.279,
      "step": 22703
    },
    {
      "epoch": 1.0418980312973245,
      "grad_norm": 0.4557572603225708,
      "learning_rate": 7.4912890558179005e-06,
      "loss": 0.3287,
      "step": 22704
    },
    {
      "epoch": 1.041943921802579,
      "grad_norm": 0.444677472114563,
      "learning_rate": 7.491076467815543e-06,
      "loss": 0.3626,
      "step": 22705
    },
    {
      "epoch": 1.0419898123078335,
      "grad_norm": 0.5197951197624207,
      "learning_rate": 7.490863873822931e-06,
      "loss": 0.4402,
      "step": 22706
    },
    {
      "epoch": 1.042035702813088,
      "grad_norm": 0.49283885955810547,
      "learning_rate": 7.490651273840578e-06,
      "loss": 0.4022,
      "step": 22707
    },
    {
      "epoch": 1.0420815933183425,
      "grad_norm": 0.4817564785480499,
      "learning_rate": 7.490438667868996e-06,
      "loss": 0.3448,
      "step": 22708
    },
    {
      "epoch": 1.042127483823597,
      "grad_norm": 0.45132896304130554,
      "learning_rate": 7.490226055908694e-06,
      "loss": 0.3499,
      "step": 22709
    },
    {
      "epoch": 1.0421733743288513,
      "grad_norm": 0.5125287771224976,
      "learning_rate": 7.4900134379601845e-06,
      "loss": 0.4471,
      "step": 22710
    },
    {
      "epoch": 1.0422192648341058,
      "grad_norm": 0.4781976044178009,
      "learning_rate": 7.489800814023978e-06,
      "loss": 0.3621,
      "step": 22711
    },
    {
      "epoch": 1.0422651553393603,
      "grad_norm": 0.46670758724212646,
      "learning_rate": 7.489588184100587e-06,
      "loss": 0.382,
      "step": 22712
    },
    {
      "epoch": 1.0423110458446148,
      "grad_norm": 0.4502623975276947,
      "learning_rate": 7.48937554819052e-06,
      "loss": 0.325,
      "step": 22713
    },
    {
      "epoch": 1.0423569363498693,
      "grad_norm": 0.46638572216033936,
      "learning_rate": 7.489162906294294e-06,
      "loss": 0.2733,
      "step": 22714
    },
    {
      "epoch": 1.0424028268551238,
      "grad_norm": 0.47697123885154724,
      "learning_rate": 7.488950258412415e-06,
      "loss": 0.3554,
      "step": 22715
    },
    {
      "epoch": 1.042448717360378,
      "grad_norm": 0.48570483922958374,
      "learning_rate": 7.4887376045453965e-06,
      "loss": 0.3729,
      "step": 22716
    },
    {
      "epoch": 1.0424946078656325,
      "grad_norm": 0.4322909116744995,
      "learning_rate": 7.488524944693749e-06,
      "loss": 0.3461,
      "step": 22717
    },
    {
      "epoch": 1.042540498370887,
      "grad_norm": 0.47378742694854736,
      "learning_rate": 7.488312278857984e-06,
      "loss": 0.3248,
      "step": 22718
    },
    {
      "epoch": 1.0425863888761415,
      "grad_norm": 0.4562162458896637,
      "learning_rate": 7.488099607038615e-06,
      "loss": 0.3533,
      "step": 22719
    },
    {
      "epoch": 1.042632279381396,
      "grad_norm": 0.47150278091430664,
      "learning_rate": 7.48788692923615e-06,
      "loss": 0.3804,
      "step": 22720
    },
    {
      "epoch": 1.0426781698866505,
      "grad_norm": 0.5105905532836914,
      "learning_rate": 7.487674245451104e-06,
      "loss": 0.3983,
      "step": 22721
    },
    {
      "epoch": 1.042724060391905,
      "grad_norm": 0.5189118981361389,
      "learning_rate": 7.487461555683986e-06,
      "loss": 0.4123,
      "step": 22722
    },
    {
      "epoch": 1.0427699508971593,
      "grad_norm": 0.4761057198047638,
      "learning_rate": 7.487248859935307e-06,
      "loss": 0.364,
      "step": 22723
    },
    {
      "epoch": 1.0428158414024138,
      "grad_norm": 0.4696687161922455,
      "learning_rate": 7.487036158205578e-06,
      "loss": 0.3585,
      "step": 22724
    },
    {
      "epoch": 1.0428617319076683,
      "grad_norm": 0.45252132415771484,
      "learning_rate": 7.486823450495315e-06,
      "loss": 0.3239,
      "step": 22725
    },
    {
      "epoch": 1.0429076224129228,
      "grad_norm": 0.4622134566307068,
      "learning_rate": 7.4866107368050265e-06,
      "loss": 0.3533,
      "step": 22726
    },
    {
      "epoch": 1.0429535129181773,
      "grad_norm": 0.432308554649353,
      "learning_rate": 7.486398017135223e-06,
      "loss": 0.2685,
      "step": 22727
    },
    {
      "epoch": 1.0429994034234318,
      "grad_norm": 0.465487003326416,
      "learning_rate": 7.486185291486416e-06,
      "loss": 0.3889,
      "step": 22728
    },
    {
      "epoch": 1.043045293928686,
      "grad_norm": 0.4786968529224396,
      "learning_rate": 7.485972559859119e-06,
      "loss": 0.3988,
      "step": 22729
    },
    {
      "epoch": 1.0430911844339406,
      "grad_norm": 0.5063127279281616,
      "learning_rate": 7.485759822253842e-06,
      "loss": 0.4489,
      "step": 22730
    },
    {
      "epoch": 1.043137074939195,
      "grad_norm": 0.44440793991088867,
      "learning_rate": 7.485547078671098e-06,
      "loss": 0.3213,
      "step": 22731
    },
    {
      "epoch": 1.0431829654444496,
      "grad_norm": 0.4996184706687927,
      "learning_rate": 7.485334329111398e-06,
      "loss": 0.4267,
      "step": 22732
    },
    {
      "epoch": 1.043228855949704,
      "grad_norm": 0.47585466504096985,
      "learning_rate": 7.485121573575252e-06,
      "loss": 0.3567,
      "step": 22733
    },
    {
      "epoch": 1.0432747464549585,
      "grad_norm": 0.42398712038993835,
      "learning_rate": 7.484908812063174e-06,
      "loss": 0.3105,
      "step": 22734
    },
    {
      "epoch": 1.043320636960213,
      "grad_norm": 0.4860897362232208,
      "learning_rate": 7.484696044575674e-06,
      "loss": 0.3428,
      "step": 22735
    },
    {
      "epoch": 1.0433665274654673,
      "grad_norm": 0.46490931510925293,
      "learning_rate": 7.484483271113265e-06,
      "loss": 0.3166,
      "step": 22736
    },
    {
      "epoch": 1.0434124179707218,
      "grad_norm": 0.5142837166786194,
      "learning_rate": 7.484270491676458e-06,
      "loss": 0.4525,
      "step": 22737
    },
    {
      "epoch": 1.0434583084759763,
      "grad_norm": 0.4415644705295563,
      "learning_rate": 7.484057706265764e-06,
      "loss": 0.3148,
      "step": 22738
    },
    {
      "epoch": 1.0435041989812308,
      "grad_norm": 0.4523180425167084,
      "learning_rate": 7.483844914881695e-06,
      "loss": 0.3414,
      "step": 22739
    },
    {
      "epoch": 1.0435500894864853,
      "grad_norm": 0.4948563277721405,
      "learning_rate": 7.483632117524762e-06,
      "loss": 0.3602,
      "step": 22740
    },
    {
      "epoch": 1.0435959799917398,
      "grad_norm": 0.48705506324768066,
      "learning_rate": 7.483419314195478e-06,
      "loss": 0.3965,
      "step": 22741
    },
    {
      "epoch": 1.043641870496994,
      "grad_norm": 0.4426090717315674,
      "learning_rate": 7.4832065048943555e-06,
      "loss": 0.2924,
      "step": 22742
    },
    {
      "epoch": 1.0436877610022486,
      "grad_norm": 0.43279650807380676,
      "learning_rate": 7.482993689621905e-06,
      "loss": 0.2956,
      "step": 22743
    },
    {
      "epoch": 1.043733651507503,
      "grad_norm": 0.5082806944847107,
      "learning_rate": 7.4827808683786375e-06,
      "loss": 0.4397,
      "step": 22744
    },
    {
      "epoch": 1.0437795420127576,
      "grad_norm": 0.4536983072757721,
      "learning_rate": 7.482568041165066e-06,
      "loss": 0.3218,
      "step": 22745
    },
    {
      "epoch": 1.043825432518012,
      "grad_norm": 0.5104808211326599,
      "learning_rate": 7.482355207981702e-06,
      "loss": 0.4038,
      "step": 22746
    },
    {
      "epoch": 1.0438713230232666,
      "grad_norm": 0.4574693739414215,
      "learning_rate": 7.4821423688290575e-06,
      "loss": 0.3663,
      "step": 22747
    },
    {
      "epoch": 1.0439172135285208,
      "grad_norm": 0.4624350666999817,
      "learning_rate": 7.481929523707643e-06,
      "loss": 0.3818,
      "step": 22748
    },
    {
      "epoch": 1.0439631040337753,
      "grad_norm": 0.45005643367767334,
      "learning_rate": 7.481716672617971e-06,
      "loss": 0.3147,
      "step": 22749
    },
    {
      "epoch": 1.0440089945390298,
      "grad_norm": 0.44013574719429016,
      "learning_rate": 7.481503815560555e-06,
      "loss": 0.2867,
      "step": 22750
    },
    {
      "epoch": 1.0440548850442843,
      "grad_norm": 0.47101548314094543,
      "learning_rate": 7.481290952535904e-06,
      "loss": 0.3487,
      "step": 22751
    },
    {
      "epoch": 1.0441007755495388,
      "grad_norm": 0.4666503369808197,
      "learning_rate": 7.4810780835445325e-06,
      "loss": 0.3492,
      "step": 22752
    },
    {
      "epoch": 1.0441466660547933,
      "grad_norm": 0.4673019051551819,
      "learning_rate": 7.480865208586951e-06,
      "loss": 0.3742,
      "step": 22753
    },
    {
      "epoch": 1.0441925565600476,
      "grad_norm": 0.44033849239349365,
      "learning_rate": 7.480652327663673e-06,
      "loss": 0.3144,
      "step": 22754
    },
    {
      "epoch": 1.0442384470653021,
      "grad_norm": 0.4528561532497406,
      "learning_rate": 7.480439440775207e-06,
      "loss": 0.349,
      "step": 22755
    },
    {
      "epoch": 1.0442843375705566,
      "grad_norm": 0.4471134841442108,
      "learning_rate": 7.480226547922068e-06,
      "loss": 0.3679,
      "step": 22756
    },
    {
      "epoch": 1.044330228075811,
      "grad_norm": 0.47342759370803833,
      "learning_rate": 7.480013649104767e-06,
      "loss": 0.3297,
      "step": 22757
    },
    {
      "epoch": 1.0443761185810656,
      "grad_norm": 0.457973450422287,
      "learning_rate": 7.479800744323815e-06,
      "loss": 0.3169,
      "step": 22758
    },
    {
      "epoch": 1.04442200908632,
      "grad_norm": 0.48210006952285767,
      "learning_rate": 7.479587833579727e-06,
      "loss": 0.3944,
      "step": 22759
    },
    {
      "epoch": 1.0444678995915746,
      "grad_norm": 0.5029793381690979,
      "learning_rate": 7.47937491687301e-06,
      "loss": 0.3751,
      "step": 22760
    },
    {
      "epoch": 1.0445137900968289,
      "grad_norm": 0.47599369287490845,
      "learning_rate": 7.47916199420418e-06,
      "loss": 0.3626,
      "step": 22761
    },
    {
      "epoch": 1.0445596806020834,
      "grad_norm": 0.4540698528289795,
      "learning_rate": 7.478949065573749e-06,
      "loss": 0.3531,
      "step": 22762
    },
    {
      "epoch": 1.0446055711073379,
      "grad_norm": 0.46855083107948303,
      "learning_rate": 7.478736130982226e-06,
      "loss": 0.4054,
      "step": 22763
    },
    {
      "epoch": 1.0446514616125924,
      "grad_norm": 0.4986293911933899,
      "learning_rate": 7.478523190430127e-06,
      "loss": 0.3902,
      "step": 22764
    },
    {
      "epoch": 1.0446973521178469,
      "grad_norm": 0.519072413444519,
      "learning_rate": 7.478310243917961e-06,
      "loss": 0.4757,
      "step": 22765
    },
    {
      "epoch": 1.0447432426231014,
      "grad_norm": 0.4353163540363312,
      "learning_rate": 7.47809729144624e-06,
      "loss": 0.298,
      "step": 22766
    },
    {
      "epoch": 1.0447891331283556,
      "grad_norm": 0.4433232843875885,
      "learning_rate": 7.477884333015477e-06,
      "loss": 0.3048,
      "step": 22767
    },
    {
      "epoch": 1.0448350236336101,
      "grad_norm": 0.44295734167099,
      "learning_rate": 7.477671368626186e-06,
      "loss": 0.3256,
      "step": 22768
    },
    {
      "epoch": 1.0448809141388646,
      "grad_norm": 0.5128352046012878,
      "learning_rate": 7.477458398278876e-06,
      "loss": 0.4521,
      "step": 22769
    },
    {
      "epoch": 1.0449268046441191,
      "grad_norm": 0.4496910870075226,
      "learning_rate": 7.477245421974062e-06,
      "loss": 0.3466,
      "step": 22770
    },
    {
      "epoch": 1.0449726951493736,
      "grad_norm": 0.47334206104278564,
      "learning_rate": 7.477032439712253e-06,
      "loss": 0.3779,
      "step": 22771
    },
    {
      "epoch": 1.0450185856546281,
      "grad_norm": 0.4903627634048462,
      "learning_rate": 7.476819451493963e-06,
      "loss": 0.3692,
      "step": 22772
    },
    {
      "epoch": 1.0450644761598826,
      "grad_norm": 0.4296421706676483,
      "learning_rate": 7.476606457319705e-06,
      "loss": 0.2867,
      "step": 22773
    },
    {
      "epoch": 1.045110366665137,
      "grad_norm": 0.48531919717788696,
      "learning_rate": 7.476393457189989e-06,
      "loss": 0.3669,
      "step": 22774
    },
    {
      "epoch": 1.0451562571703914,
      "grad_norm": 0.4488718807697296,
      "learning_rate": 7.4761804511053286e-06,
      "loss": 0.3384,
      "step": 22775
    },
    {
      "epoch": 1.045202147675646,
      "grad_norm": 0.4254857897758484,
      "learning_rate": 7.4759674390662365e-06,
      "loss": 0.2967,
      "step": 22776
    },
    {
      "epoch": 1.0452480381809004,
      "grad_norm": 0.47261327505111694,
      "learning_rate": 7.475754421073223e-06,
      "loss": 0.3181,
      "step": 22777
    },
    {
      "epoch": 1.045293928686155,
      "grad_norm": 0.4443094730377197,
      "learning_rate": 7.475541397126801e-06,
      "loss": 0.3342,
      "step": 22778
    },
    {
      "epoch": 1.0453398191914094,
      "grad_norm": 0.5133997201919556,
      "learning_rate": 7.475328367227485e-06,
      "loss": 0.3832,
      "step": 22779
    },
    {
      "epoch": 1.0453857096966637,
      "grad_norm": 0.45386335253715515,
      "learning_rate": 7.475115331375785e-06,
      "loss": 0.3253,
      "step": 22780
    },
    {
      "epoch": 1.0454316002019182,
      "grad_norm": 0.4405195713043213,
      "learning_rate": 7.474902289572213e-06,
      "loss": 0.3202,
      "step": 22781
    },
    {
      "epoch": 1.0454774907071727,
      "grad_norm": 0.4836391806602478,
      "learning_rate": 7.4746892418172835e-06,
      "loss": 0.4019,
      "step": 22782
    },
    {
      "epoch": 1.0455233812124272,
      "grad_norm": 0.4710160493850708,
      "learning_rate": 7.474476188111506e-06,
      "loss": 0.3867,
      "step": 22783
    },
    {
      "epoch": 1.0455692717176817,
      "grad_norm": 0.44884902238845825,
      "learning_rate": 7.4742631284553935e-06,
      "loss": 0.2971,
      "step": 22784
    },
    {
      "epoch": 1.0456151622229362,
      "grad_norm": 0.46134212613105774,
      "learning_rate": 7.47405006284946e-06,
      "loss": 0.3039,
      "step": 22785
    },
    {
      "epoch": 1.0456610527281904,
      "grad_norm": 0.4284680187702179,
      "learning_rate": 7.473836991294219e-06,
      "loss": 0.3339,
      "step": 22786
    },
    {
      "epoch": 1.045706943233445,
      "grad_norm": 0.43046730756759644,
      "learning_rate": 7.473623913790179e-06,
      "loss": 0.2781,
      "step": 22787
    },
    {
      "epoch": 1.0457528337386994,
      "grad_norm": 0.455280601978302,
      "learning_rate": 7.4734108303378535e-06,
      "loss": 0.3528,
      "step": 22788
    },
    {
      "epoch": 1.045798724243954,
      "grad_norm": 0.4981512427330017,
      "learning_rate": 7.473197740937757e-06,
      "loss": 0.3264,
      "step": 22789
    },
    {
      "epoch": 1.0458446147492084,
      "grad_norm": 0.4475570023059845,
      "learning_rate": 7.4729846455904e-06,
      "loss": 0.3071,
      "step": 22790
    },
    {
      "epoch": 1.045890505254463,
      "grad_norm": 0.4726376235485077,
      "learning_rate": 7.472771544296295e-06,
      "loss": 0.319,
      "step": 22791
    },
    {
      "epoch": 1.0459363957597174,
      "grad_norm": 0.47917553782463074,
      "learning_rate": 7.472558437055954e-06,
      "loss": 0.4013,
      "step": 22792
    },
    {
      "epoch": 1.0459822862649717,
      "grad_norm": 0.48621171712875366,
      "learning_rate": 7.472345323869892e-06,
      "loss": 0.3543,
      "step": 22793
    },
    {
      "epoch": 1.0460281767702262,
      "grad_norm": 0.45518121123313904,
      "learning_rate": 7.472132204738619e-06,
      "loss": 0.3161,
      "step": 22794
    },
    {
      "epoch": 1.0460740672754807,
      "grad_norm": 0.44455569982528687,
      "learning_rate": 7.471919079662648e-06,
      "loss": 0.3172,
      "step": 22795
    },
    {
      "epoch": 1.0461199577807352,
      "grad_norm": 0.4999527633190155,
      "learning_rate": 7.471705948642492e-06,
      "loss": 0.4392,
      "step": 22796
    },
    {
      "epoch": 1.0461658482859897,
      "grad_norm": 0.4634413719177246,
      "learning_rate": 7.471492811678665e-06,
      "loss": 0.3432,
      "step": 22797
    },
    {
      "epoch": 1.0462117387912442,
      "grad_norm": 0.4599563777446747,
      "learning_rate": 7.471279668771676e-06,
      "loss": 0.3406,
      "step": 22798
    },
    {
      "epoch": 1.0462576292964985,
      "grad_norm": 0.4741176664829254,
      "learning_rate": 7.471066519922039e-06,
      "loss": 0.3271,
      "step": 22799
    },
    {
      "epoch": 1.046303519801753,
      "grad_norm": 0.4608007073402405,
      "learning_rate": 7.470853365130267e-06,
      "loss": 0.3471,
      "step": 22800
    },
    {
      "epoch": 1.0463494103070075,
      "grad_norm": 0.4550667107105255,
      "learning_rate": 7.470640204396873e-06,
      "loss": 0.3344,
      "step": 22801
    },
    {
      "epoch": 1.046395300812262,
      "grad_norm": 0.5201148390769958,
      "learning_rate": 7.470427037722371e-06,
      "loss": 0.4416,
      "step": 22802
    },
    {
      "epoch": 1.0464411913175165,
      "grad_norm": 0.44585612416267395,
      "learning_rate": 7.470213865107269e-06,
      "loss": 0.3161,
      "step": 22803
    },
    {
      "epoch": 1.046487081822771,
      "grad_norm": 0.4689854383468628,
      "learning_rate": 7.470000686552083e-06,
      "loss": 0.3622,
      "step": 22804
    },
    {
      "epoch": 1.0465329723280252,
      "grad_norm": 0.4431607723236084,
      "learning_rate": 7.469787502057325e-06,
      "loss": 0.3027,
      "step": 22805
    },
    {
      "epoch": 1.0465788628332797,
      "grad_norm": 0.4926261901855469,
      "learning_rate": 7.469574311623506e-06,
      "loss": 0.3525,
      "step": 22806
    },
    {
      "epoch": 1.0466247533385342,
      "grad_norm": 0.46701228618621826,
      "learning_rate": 7.469361115251143e-06,
      "loss": 0.3846,
      "step": 22807
    },
    {
      "epoch": 1.0466706438437887,
      "grad_norm": 0.49617844820022583,
      "learning_rate": 7.469147912940745e-06,
      "loss": 0.351,
      "step": 22808
    },
    {
      "epoch": 1.0467165343490432,
      "grad_norm": 0.47484090924263,
      "learning_rate": 7.468934704692825e-06,
      "loss": 0.368,
      "step": 22809
    },
    {
      "epoch": 1.0467624248542977,
      "grad_norm": 0.4885796904563904,
      "learning_rate": 7.468721490507896e-06,
      "loss": 0.4141,
      "step": 22810
    },
    {
      "epoch": 1.0468083153595522,
      "grad_norm": 0.43781065940856934,
      "learning_rate": 7.468508270386472e-06,
      "loss": 0.2946,
      "step": 22811
    },
    {
      "epoch": 1.0468542058648065,
      "grad_norm": 0.47285762429237366,
      "learning_rate": 7.468295044329063e-06,
      "loss": 0.3534,
      "step": 22812
    },
    {
      "epoch": 1.046900096370061,
      "grad_norm": 0.4861748218536377,
      "learning_rate": 7.468081812336187e-06,
      "loss": 0.4159,
      "step": 22813
    },
    {
      "epoch": 1.0469459868753155,
      "grad_norm": 0.46343794465065,
      "learning_rate": 7.467868574408349e-06,
      "loss": 0.3294,
      "step": 22814
    },
    {
      "epoch": 1.04699187738057,
      "grad_norm": 0.4343664348125458,
      "learning_rate": 7.467655330546068e-06,
      "loss": 0.2997,
      "step": 22815
    },
    {
      "epoch": 1.0470377678858245,
      "grad_norm": 0.46754932403564453,
      "learning_rate": 7.467442080749854e-06,
      "loss": 0.339,
      "step": 22816
    },
    {
      "epoch": 1.047083658391079,
      "grad_norm": 0.4328851103782654,
      "learning_rate": 7.4672288250202205e-06,
      "loss": 0.3047,
      "step": 22817
    },
    {
      "epoch": 1.0471295488963333,
      "grad_norm": 0.48496606945991516,
      "learning_rate": 7.467015563357681e-06,
      "loss": 0.3726,
      "step": 22818
    },
    {
      "epoch": 1.0471754394015877,
      "grad_norm": 0.43926775455474854,
      "learning_rate": 7.466802295762749e-06,
      "loss": 0.3174,
      "step": 22819
    },
    {
      "epoch": 1.0472213299068422,
      "grad_norm": 0.5028486847877502,
      "learning_rate": 7.466589022235934e-06,
      "loss": 0.4347,
      "step": 22820
    },
    {
      "epoch": 1.0472672204120967,
      "grad_norm": 0.4662001132965088,
      "learning_rate": 7.466375742777752e-06,
      "loss": 0.3813,
      "step": 22821
    },
    {
      "epoch": 1.0473131109173512,
      "grad_norm": 0.46108874678611755,
      "learning_rate": 7.466162457388713e-06,
      "loss": 0.3097,
      "step": 22822
    },
    {
      "epoch": 1.0473590014226057,
      "grad_norm": 0.48175281286239624,
      "learning_rate": 7.465949166069333e-06,
      "loss": 0.4245,
      "step": 22823
    },
    {
      "epoch": 1.0474048919278602,
      "grad_norm": 0.44968870282173157,
      "learning_rate": 7.465735868820124e-06,
      "loss": 0.3365,
      "step": 22824
    },
    {
      "epoch": 1.0474507824331145,
      "grad_norm": 0.4968424439430237,
      "learning_rate": 7.465522565641597e-06,
      "loss": 0.3581,
      "step": 22825
    },
    {
      "epoch": 1.047496672938369,
      "grad_norm": 0.4804345965385437,
      "learning_rate": 7.465309256534267e-06,
      "loss": 0.3417,
      "step": 22826
    },
    {
      "epoch": 1.0475425634436235,
      "grad_norm": 0.42635828256607056,
      "learning_rate": 7.465095941498647e-06,
      "loss": 0.3148,
      "step": 22827
    },
    {
      "epoch": 1.047588453948878,
      "grad_norm": 0.4219377040863037,
      "learning_rate": 7.464882620535247e-06,
      "loss": 0.2871,
      "step": 22828
    },
    {
      "epoch": 1.0476343444541325,
      "grad_norm": 0.46893975138664246,
      "learning_rate": 7.4646692936445855e-06,
      "loss": 0.3269,
      "step": 22829
    },
    {
      "epoch": 1.047680234959387,
      "grad_norm": 0.4664924740791321,
      "learning_rate": 7.464455960827171e-06,
      "loss": 0.3929,
      "step": 22830
    },
    {
      "epoch": 1.0477261254646413,
      "grad_norm": 0.46971023082733154,
      "learning_rate": 7.464242622083516e-06,
      "loss": 0.3883,
      "step": 22831
    },
    {
      "epoch": 1.0477720159698958,
      "grad_norm": 0.45625361800193787,
      "learning_rate": 7.464029277414137e-06,
      "loss": 0.3164,
      "step": 22832
    },
    {
      "epoch": 1.0478179064751503,
      "grad_norm": 0.4419002830982208,
      "learning_rate": 7.463815926819545e-06,
      "loss": 0.3336,
      "step": 22833
    },
    {
      "epoch": 1.0478637969804048,
      "grad_norm": 0.42522820830345154,
      "learning_rate": 7.463602570300253e-06,
      "loss": 0.2759,
      "step": 22834
    },
    {
      "epoch": 1.0479096874856593,
      "grad_norm": 0.44821813702583313,
      "learning_rate": 7.463389207856776e-06,
      "loss": 0.346,
      "step": 22835
    },
    {
      "epoch": 1.0479555779909138,
      "grad_norm": 0.48051679134368896,
      "learning_rate": 7.463175839489622e-06,
      "loss": 0.376,
      "step": 22836
    },
    {
      "epoch": 1.048001468496168,
      "grad_norm": 0.4657687544822693,
      "learning_rate": 7.46296246519931e-06,
      "loss": 0.364,
      "step": 22837
    },
    {
      "epoch": 1.0480473590014225,
      "grad_norm": 0.5403136610984802,
      "learning_rate": 7.462749084986351e-06,
      "loss": 0.3918,
      "step": 22838
    },
    {
      "epoch": 1.048093249506677,
      "grad_norm": 0.4405789077281952,
      "learning_rate": 7.462535698851255e-06,
      "loss": 0.2859,
      "step": 22839
    },
    {
      "epoch": 1.0481391400119315,
      "grad_norm": 0.48423317074775696,
      "learning_rate": 7.4623223067945405e-06,
      "loss": 0.4092,
      "step": 22840
    },
    {
      "epoch": 1.048185030517186,
      "grad_norm": 0.5156185030937195,
      "learning_rate": 7.4621089088167174e-06,
      "loss": 0.4887,
      "step": 22841
    },
    {
      "epoch": 1.0482309210224405,
      "grad_norm": 0.49452975392341614,
      "learning_rate": 7.461895504918299e-06,
      "loss": 0.4065,
      "step": 22842
    },
    {
      "epoch": 1.0482768115276948,
      "grad_norm": 0.5345056653022766,
      "learning_rate": 7.461682095099798e-06,
      "loss": 0.4241,
      "step": 22843
    },
    {
      "epoch": 1.0483227020329493,
      "grad_norm": 0.46285098791122437,
      "learning_rate": 7.4614686793617305e-06,
      "loss": 0.3623,
      "step": 22844
    },
    {
      "epoch": 1.0483685925382038,
      "grad_norm": 0.4668427109718323,
      "learning_rate": 7.461255257704606e-06,
      "loss": 0.3379,
      "step": 22845
    },
    {
      "epoch": 1.0484144830434583,
      "grad_norm": 0.4942461848258972,
      "learning_rate": 7.461041830128941e-06,
      "loss": 0.373,
      "step": 22846
    },
    {
      "epoch": 1.0484603735487128,
      "grad_norm": 0.48151376843452454,
      "learning_rate": 7.460828396635247e-06,
      "loss": 0.3652,
      "step": 22847
    },
    {
      "epoch": 1.0485062640539673,
      "grad_norm": 0.4418684244155884,
      "learning_rate": 7.460614957224035e-06,
      "loss": 0.3245,
      "step": 22848
    },
    {
      "epoch": 1.0485521545592218,
      "grad_norm": 0.4535737633705139,
      "learning_rate": 7.460401511895823e-06,
      "loss": 0.3424,
      "step": 22849
    },
    {
      "epoch": 1.048598045064476,
      "grad_norm": 0.4610288143157959,
      "learning_rate": 7.4601880606511215e-06,
      "loss": 0.3143,
      "step": 22850
    },
    {
      "epoch": 1.0486439355697306,
      "grad_norm": 0.45131716132164,
      "learning_rate": 7.459974603490443e-06,
      "loss": 0.3322,
      "step": 22851
    },
    {
      "epoch": 1.048689826074985,
      "grad_norm": 0.5182852149009705,
      "learning_rate": 7.4597611404143035e-06,
      "loss": 0.3932,
      "step": 22852
    },
    {
      "epoch": 1.0487357165802396,
      "grad_norm": 0.4973585307598114,
      "learning_rate": 7.459547671423214e-06,
      "loss": 0.3939,
      "step": 22853
    },
    {
      "epoch": 1.048781607085494,
      "grad_norm": 0.48866260051727295,
      "learning_rate": 7.459334196517688e-06,
      "loss": 0.3632,
      "step": 22854
    },
    {
      "epoch": 1.0488274975907486,
      "grad_norm": 0.45171910524368286,
      "learning_rate": 7.459120715698241e-06,
      "loss": 0.3212,
      "step": 22855
    },
    {
      "epoch": 1.0488733880960028,
      "grad_norm": 0.4812033474445343,
      "learning_rate": 7.458907228965383e-06,
      "loss": 0.3551,
      "step": 22856
    },
    {
      "epoch": 1.0489192786012573,
      "grad_norm": 0.4883175790309906,
      "learning_rate": 7.458693736319631e-06,
      "loss": 0.3391,
      "step": 22857
    },
    {
      "epoch": 1.0489651691065118,
      "grad_norm": 0.4464930593967438,
      "learning_rate": 7.458480237761495e-06,
      "loss": 0.3321,
      "step": 22858
    },
    {
      "epoch": 1.0490110596117663,
      "grad_norm": 0.498491108417511,
      "learning_rate": 7.45826673329149e-06,
      "loss": 0.4228,
      "step": 22859
    },
    {
      "epoch": 1.0490569501170208,
      "grad_norm": 0.4749586582183838,
      "learning_rate": 7.45805322291013e-06,
      "loss": 0.3776,
      "step": 22860
    },
    {
      "epoch": 1.0491028406222753,
      "grad_norm": 0.47845861315727234,
      "learning_rate": 7.457839706617929e-06,
      "loss": 0.332,
      "step": 22861
    },
    {
      "epoch": 1.0491487311275298,
      "grad_norm": 0.5348946452140808,
      "learning_rate": 7.457626184415397e-06,
      "loss": 0.3735,
      "step": 22862
    },
    {
      "epoch": 1.049194621632784,
      "grad_norm": 0.448983758687973,
      "learning_rate": 7.45741265630305e-06,
      "loss": 0.3239,
      "step": 22863
    },
    {
      "epoch": 1.0492405121380386,
      "grad_norm": 0.47729814052581787,
      "learning_rate": 7.4571991222814025e-06,
      "loss": 0.3937,
      "step": 22864
    },
    {
      "epoch": 1.049286402643293,
      "grad_norm": 0.46043023467063904,
      "learning_rate": 7.456985582350965e-06,
      "loss": 0.3472,
      "step": 22865
    },
    {
      "epoch": 1.0493322931485476,
      "grad_norm": 0.5012228488922119,
      "learning_rate": 7.4567720365122535e-06,
      "loss": 0.414,
      "step": 22866
    },
    {
      "epoch": 1.049378183653802,
      "grad_norm": 0.5149646401405334,
      "learning_rate": 7.456558484765781e-06,
      "loss": 0.4113,
      "step": 22867
    },
    {
      "epoch": 1.0494240741590566,
      "grad_norm": 0.524141788482666,
      "learning_rate": 7.45634492711206e-06,
      "loss": 0.4496,
      "step": 22868
    },
    {
      "epoch": 1.0494699646643109,
      "grad_norm": 0.4521539807319641,
      "learning_rate": 7.4561313635516065e-06,
      "loss": 0.3337,
      "step": 22869
    },
    {
      "epoch": 1.0495158551695654,
      "grad_norm": 0.45072779059410095,
      "learning_rate": 7.45591779408493e-06,
      "loss": 0.3046,
      "step": 22870
    },
    {
      "epoch": 1.0495617456748199,
      "grad_norm": 0.4728637933731079,
      "learning_rate": 7.4557042187125464e-06,
      "loss": 0.3981,
      "step": 22871
    },
    {
      "epoch": 1.0496076361800744,
      "grad_norm": 0.4781298339366913,
      "learning_rate": 7.45549063743497e-06,
      "loss": 0.3435,
      "step": 22872
    },
    {
      "epoch": 1.0496535266853289,
      "grad_norm": 0.5465521812438965,
      "learning_rate": 7.455277050252714e-06,
      "loss": 0.4372,
      "step": 22873
    },
    {
      "epoch": 1.0496994171905834,
      "grad_norm": 0.4891458749771118,
      "learning_rate": 7.45506345716629e-06,
      "loss": 0.3884,
      "step": 22874
    },
    {
      "epoch": 1.0497453076958376,
      "grad_norm": 0.45702677965164185,
      "learning_rate": 7.4548498581762155e-06,
      "loss": 0.319,
      "step": 22875
    },
    {
      "epoch": 1.0497911982010921,
      "grad_norm": 0.5109661817550659,
      "learning_rate": 7.4546362532829995e-06,
      "loss": 0.4011,
      "step": 22876
    },
    {
      "epoch": 1.0498370887063466,
      "grad_norm": 0.524146556854248,
      "learning_rate": 7.454422642487159e-06,
      "loss": 0.4392,
      "step": 22877
    },
    {
      "epoch": 1.0498829792116011,
      "grad_norm": 0.47927236557006836,
      "learning_rate": 7.4542090257892074e-06,
      "loss": 0.4014,
      "step": 22878
    },
    {
      "epoch": 1.0499288697168556,
      "grad_norm": 0.45382750034332275,
      "learning_rate": 7.453995403189658e-06,
      "loss": 0.3632,
      "step": 22879
    },
    {
      "epoch": 1.0499747602221101,
      "grad_norm": 0.47037824988365173,
      "learning_rate": 7.453781774689021e-06,
      "loss": 0.3641,
      "step": 22880
    },
    {
      "epoch": 1.0500206507273646,
      "grad_norm": 0.44643646478652954,
      "learning_rate": 7.453568140287817e-06,
      "loss": 0.3389,
      "step": 22881
    },
    {
      "epoch": 1.0500665412326189,
      "grad_norm": 0.4922311007976532,
      "learning_rate": 7.4533544999865535e-06,
      "loss": 0.3982,
      "step": 22882
    },
    {
      "epoch": 1.0501124317378734,
      "grad_norm": 0.4400462806224823,
      "learning_rate": 7.453140853785749e-06,
      "loss": 0.3512,
      "step": 22883
    },
    {
      "epoch": 1.0501583222431279,
      "grad_norm": 0.5003165006637573,
      "learning_rate": 7.452927201685913e-06,
      "loss": 0.339,
      "step": 22884
    },
    {
      "epoch": 1.0502042127483824,
      "grad_norm": 0.4642479717731476,
      "learning_rate": 7.4527135436875615e-06,
      "loss": 0.382,
      "step": 22885
    },
    {
      "epoch": 1.0502501032536369,
      "grad_norm": 0.4973624646663666,
      "learning_rate": 7.4524998797912096e-06,
      "loss": 0.4472,
      "step": 22886
    },
    {
      "epoch": 1.0502959937588914,
      "grad_norm": 0.48698046803474426,
      "learning_rate": 7.452286209997368e-06,
      "loss": 0.3988,
      "step": 22887
    },
    {
      "epoch": 1.0503418842641457,
      "grad_norm": 0.44886457920074463,
      "learning_rate": 7.452072534306551e-06,
      "loss": 0.3123,
      "step": 22888
    },
    {
      "epoch": 1.0503877747694002,
      "grad_norm": 0.4723467230796814,
      "learning_rate": 7.451858852719276e-06,
      "loss": 0.3452,
      "step": 22889
    },
    {
      "epoch": 1.0504336652746546,
      "grad_norm": 0.4727070927619934,
      "learning_rate": 7.451645165236055e-06,
      "loss": 0.3534,
      "step": 22890
    },
    {
      "epoch": 1.0504795557799091,
      "grad_norm": 0.4845750629901886,
      "learning_rate": 7.451431471857398e-06,
      "loss": 0.4114,
      "step": 22891
    },
    {
      "epoch": 1.0505254462851636,
      "grad_norm": 0.43944084644317627,
      "learning_rate": 7.451217772583823e-06,
      "loss": 0.3232,
      "step": 22892
    },
    {
      "epoch": 1.0505713367904181,
      "grad_norm": 0.4886188507080078,
      "learning_rate": 7.451004067415843e-06,
      "loss": 0.4007,
      "step": 22893
    },
    {
      "epoch": 1.0506172272956724,
      "grad_norm": 0.5098027586936951,
      "learning_rate": 7.450790356353972e-06,
      "loss": 0.4342,
      "step": 22894
    },
    {
      "epoch": 1.050663117800927,
      "grad_norm": 0.44509491324424744,
      "learning_rate": 7.450576639398725e-06,
      "loss": 0.3484,
      "step": 22895
    },
    {
      "epoch": 1.0507090083061814,
      "grad_norm": 0.5053408145904541,
      "learning_rate": 7.450362916550612e-06,
      "loss": 0.4497,
      "step": 22896
    },
    {
      "epoch": 1.050754898811436,
      "grad_norm": 0.5024242997169495,
      "learning_rate": 7.4501491878101504e-06,
      "loss": 0.4456,
      "step": 22897
    },
    {
      "epoch": 1.0508007893166904,
      "grad_norm": 0.4739820659160614,
      "learning_rate": 7.449935453177854e-06,
      "loss": 0.372,
      "step": 22898
    },
    {
      "epoch": 1.050846679821945,
      "grad_norm": 0.47009769082069397,
      "learning_rate": 7.4497217126542355e-06,
      "loss": 0.367,
      "step": 22899
    },
    {
      "epoch": 1.0508925703271994,
      "grad_norm": 0.48233696818351746,
      "learning_rate": 7.449507966239811e-06,
      "loss": 0.3713,
      "step": 22900
    },
    {
      "epoch": 1.0509384608324537,
      "grad_norm": 0.47898244857788086,
      "learning_rate": 7.449294213935091e-06,
      "loss": 0.3429,
      "step": 22901
    },
    {
      "epoch": 1.0509843513377082,
      "grad_norm": 0.45203670859336853,
      "learning_rate": 7.449080455740592e-06,
      "loss": 0.3269,
      "step": 22902
    },
    {
      "epoch": 1.0510302418429627,
      "grad_norm": 0.47224658727645874,
      "learning_rate": 7.448866691656828e-06,
      "loss": 0.3593,
      "step": 22903
    },
    {
      "epoch": 1.0510761323482172,
      "grad_norm": 0.4399934709072113,
      "learning_rate": 7.448652921684312e-06,
      "loss": 0.3358,
      "step": 22904
    },
    {
      "epoch": 1.0511220228534717,
      "grad_norm": 0.574245274066925,
      "learning_rate": 7.448439145823558e-06,
      "loss": 0.4313,
      "step": 22905
    },
    {
      "epoch": 1.0511679133587262,
      "grad_norm": 0.4937978684902191,
      "learning_rate": 7.448225364075081e-06,
      "loss": 0.4129,
      "step": 22906
    },
    {
      "epoch": 1.0512138038639804,
      "grad_norm": 0.4915756583213806,
      "learning_rate": 7.448011576439396e-06,
      "loss": 0.355,
      "step": 22907
    },
    {
      "epoch": 1.051259694369235,
      "grad_norm": 0.4830446243286133,
      "learning_rate": 7.447797782917013e-06,
      "loss": 0.3597,
      "step": 22908
    },
    {
      "epoch": 1.0513055848744894,
      "grad_norm": 0.5387779474258423,
      "learning_rate": 7.447583983508451e-06,
      "loss": 0.4403,
      "step": 22909
    },
    {
      "epoch": 1.051351475379744,
      "grad_norm": 0.5020200610160828,
      "learning_rate": 7.4473701782142215e-06,
      "loss": 0.3967,
      "step": 22910
    },
    {
      "epoch": 1.0513973658849984,
      "grad_norm": 0.5463634729385376,
      "learning_rate": 7.447156367034839e-06,
      "loss": 0.3956,
      "step": 22911
    },
    {
      "epoch": 1.051443256390253,
      "grad_norm": 0.49161043763160706,
      "learning_rate": 7.446942549970818e-06,
      "loss": 0.4165,
      "step": 22912
    },
    {
      "epoch": 1.0514891468955074,
      "grad_norm": 0.467472106218338,
      "learning_rate": 7.446728727022672e-06,
      "loss": 0.306,
      "step": 22913
    },
    {
      "epoch": 1.0515350374007617,
      "grad_norm": 0.4945918917655945,
      "learning_rate": 7.4465148981909155e-06,
      "loss": 0.4093,
      "step": 22914
    },
    {
      "epoch": 1.0515809279060162,
      "grad_norm": 0.48945873975753784,
      "learning_rate": 7.446301063476063e-06,
      "loss": 0.3656,
      "step": 22915
    },
    {
      "epoch": 1.0516268184112707,
      "grad_norm": 0.4703664481639862,
      "learning_rate": 7.446087222878631e-06,
      "loss": 0.3401,
      "step": 22916
    },
    {
      "epoch": 1.0516727089165252,
      "grad_norm": 0.4353506863117218,
      "learning_rate": 7.445873376399129e-06,
      "loss": 0.3216,
      "step": 22917
    },
    {
      "epoch": 1.0517185994217797,
      "grad_norm": 0.4693823456764221,
      "learning_rate": 7.445659524038072e-06,
      "loss": 0.3282,
      "step": 22918
    },
    {
      "epoch": 1.0517644899270342,
      "grad_norm": 0.46539637446403503,
      "learning_rate": 7.445445665795977e-06,
      "loss": 0.3949,
      "step": 22919
    },
    {
      "epoch": 1.0518103804322885,
      "grad_norm": 0.44187384843826294,
      "learning_rate": 7.445231801673358e-06,
      "loss": 0.3067,
      "step": 22920
    },
    {
      "epoch": 1.051856270937543,
      "grad_norm": 0.4271581470966339,
      "learning_rate": 7.445017931670726e-06,
      "loss": 0.2993,
      "step": 22921
    },
    {
      "epoch": 1.0519021614427975,
      "grad_norm": 0.45700979232788086,
      "learning_rate": 7.444804055788601e-06,
      "loss": 0.3276,
      "step": 22922
    },
    {
      "epoch": 1.051948051948052,
      "grad_norm": 0.4775296747684479,
      "learning_rate": 7.444590174027492e-06,
      "loss": 0.3543,
      "step": 22923
    },
    {
      "epoch": 1.0519939424533065,
      "grad_norm": 0.4667254388332367,
      "learning_rate": 7.444376286387914e-06,
      "loss": 0.333,
      "step": 22924
    },
    {
      "epoch": 1.052039832958561,
      "grad_norm": 0.4507078230381012,
      "learning_rate": 7.444162392870383e-06,
      "loss": 0.3373,
      "step": 22925
    },
    {
      "epoch": 1.0520857234638152,
      "grad_norm": 0.4883470833301544,
      "learning_rate": 7.443948493475414e-06,
      "loss": 0.3574,
      "step": 22926
    },
    {
      "epoch": 1.0521316139690697,
      "grad_norm": 0.46691417694091797,
      "learning_rate": 7.443734588203519e-06,
      "loss": 0.3878,
      "step": 22927
    },
    {
      "epoch": 1.0521775044743242,
      "grad_norm": 0.461717814207077,
      "learning_rate": 7.443520677055214e-06,
      "loss": 0.3162,
      "step": 22928
    },
    {
      "epoch": 1.0522233949795787,
      "grad_norm": 0.43372756242752075,
      "learning_rate": 7.443306760031014e-06,
      "loss": 0.3279,
      "step": 22929
    },
    {
      "epoch": 1.0522692854848332,
      "grad_norm": 0.5015509724617004,
      "learning_rate": 7.443092837131432e-06,
      "loss": 0.3504,
      "step": 22930
    },
    {
      "epoch": 1.0523151759900877,
      "grad_norm": 0.4529511332511902,
      "learning_rate": 7.442878908356981e-06,
      "loss": 0.3419,
      "step": 22931
    },
    {
      "epoch": 1.052361066495342,
      "grad_norm": 0.4755386412143707,
      "learning_rate": 7.442664973708178e-06,
      "loss": 0.4048,
      "step": 22932
    },
    {
      "epoch": 1.0524069570005965,
      "grad_norm": 0.48132649064064026,
      "learning_rate": 7.442451033185538e-06,
      "loss": 0.392,
      "step": 22933
    },
    {
      "epoch": 1.052452847505851,
      "grad_norm": 0.47903576493263245,
      "learning_rate": 7.442237086789574e-06,
      "loss": 0.341,
      "step": 22934
    },
    {
      "epoch": 1.0524987380111055,
      "grad_norm": 0.4602435529232025,
      "learning_rate": 7.442023134520799e-06,
      "loss": 0.3279,
      "step": 22935
    },
    {
      "epoch": 1.05254462851636,
      "grad_norm": 0.4349225163459778,
      "learning_rate": 7.44180917637973e-06,
      "loss": 0.2945,
      "step": 22936
    },
    {
      "epoch": 1.0525905190216145,
      "grad_norm": 0.4564686119556427,
      "learning_rate": 7.441595212366881e-06,
      "loss": 0.3155,
      "step": 22937
    },
    {
      "epoch": 1.052636409526869,
      "grad_norm": 0.4449274241924286,
      "learning_rate": 7.4413812424827655e-06,
      "loss": 0.3212,
      "step": 22938
    },
    {
      "epoch": 1.0526823000321233,
      "grad_norm": 0.48479411005973816,
      "learning_rate": 7.441167266727897e-06,
      "loss": 0.3549,
      "step": 22939
    },
    {
      "epoch": 1.0527281905373778,
      "grad_norm": 0.45903050899505615,
      "learning_rate": 7.440953285102795e-06,
      "loss": 0.3488,
      "step": 22940
    },
    {
      "epoch": 1.0527740810426323,
      "grad_norm": 0.4260769188404083,
      "learning_rate": 7.440739297607968e-06,
      "loss": 0.2984,
      "step": 22941
    },
    {
      "epoch": 1.0528199715478868,
      "grad_norm": 0.486358642578125,
      "learning_rate": 7.440525304243935e-06,
      "loss": 0.3592,
      "step": 22942
    },
    {
      "epoch": 1.0528658620531413,
      "grad_norm": 0.5319312810897827,
      "learning_rate": 7.440311305011208e-06,
      "loss": 0.4014,
      "step": 22943
    },
    {
      "epoch": 1.0529117525583958,
      "grad_norm": 0.45271381735801697,
      "learning_rate": 7.4400972999103035e-06,
      "loss": 0.3488,
      "step": 22944
    },
    {
      "epoch": 1.05295764306365,
      "grad_norm": 0.44694584608078003,
      "learning_rate": 7.439883288941735e-06,
      "loss": 0.3234,
      "step": 22945
    },
    {
      "epoch": 1.0530035335689045,
      "grad_norm": 0.4485241174697876,
      "learning_rate": 7.439669272106017e-06,
      "loss": 0.316,
      "step": 22946
    },
    {
      "epoch": 1.053049424074159,
      "grad_norm": 0.49566006660461426,
      "learning_rate": 7.439455249403663e-06,
      "loss": 0.3893,
      "step": 22947
    },
    {
      "epoch": 1.0530953145794135,
      "grad_norm": 0.4916944205760956,
      "learning_rate": 7.439241220835191e-06,
      "loss": 0.3627,
      "step": 22948
    },
    {
      "epoch": 1.053141205084668,
      "grad_norm": 0.4673306345939636,
      "learning_rate": 7.439027186401113e-06,
      "loss": 0.347,
      "step": 22949
    },
    {
      "epoch": 1.0531870955899225,
      "grad_norm": 0.4862653911113739,
      "learning_rate": 7.4388131461019445e-06,
      "loss": 0.4043,
      "step": 22950
    },
    {
      "epoch": 1.053232986095177,
      "grad_norm": 0.47215336561203003,
      "learning_rate": 7.4385990999381996e-06,
      "loss": 0.3541,
      "step": 22951
    },
    {
      "epoch": 1.0532788766004313,
      "grad_norm": 0.4164862334728241,
      "learning_rate": 7.438385047910394e-06,
      "loss": 0.2744,
      "step": 22952
    },
    {
      "epoch": 1.0533247671056858,
      "grad_norm": 0.467967689037323,
      "learning_rate": 7.43817099001904e-06,
      "loss": 0.3867,
      "step": 22953
    },
    {
      "epoch": 1.0533706576109403,
      "grad_norm": 0.47614115476608276,
      "learning_rate": 7.4379569262646576e-06,
      "loss": 0.3649,
      "step": 22954
    },
    {
      "epoch": 1.0534165481161948,
      "grad_norm": 0.45890775322914124,
      "learning_rate": 7.437742856647756e-06,
      "loss": 0.3223,
      "step": 22955
    },
    {
      "epoch": 1.0534624386214493,
      "grad_norm": 0.533683717250824,
      "learning_rate": 7.437528781168852e-06,
      "loss": 0.4444,
      "step": 22956
    },
    {
      "epoch": 1.0535083291267038,
      "grad_norm": 0.4898475706577301,
      "learning_rate": 7.437314699828463e-06,
      "loss": 0.4274,
      "step": 22957
    },
    {
      "epoch": 1.053554219631958,
      "grad_norm": 0.4400314688682556,
      "learning_rate": 7.4371006126271e-06,
      "loss": 0.3304,
      "step": 22958
    },
    {
      "epoch": 1.0536001101372126,
      "grad_norm": 0.43901070952415466,
      "learning_rate": 7.436886519565278e-06,
      "loss": 0.3398,
      "step": 22959
    },
    {
      "epoch": 1.053646000642467,
      "grad_norm": 0.46425217390060425,
      "learning_rate": 7.4366724206435144e-06,
      "loss": 0.3782,
      "step": 22960
    },
    {
      "epoch": 1.0536918911477215,
      "grad_norm": 0.48289090394973755,
      "learning_rate": 7.436458315862323e-06,
      "loss": 0.3424,
      "step": 22961
    },
    {
      "epoch": 1.053737781652976,
      "grad_norm": 0.5225856304168701,
      "learning_rate": 7.436244205222216e-06,
      "loss": 0.4233,
      "step": 22962
    },
    {
      "epoch": 1.0537836721582305,
      "grad_norm": 0.47003406286239624,
      "learning_rate": 7.436030088723714e-06,
      "loss": 0.3803,
      "step": 22963
    },
    {
      "epoch": 1.0538295626634848,
      "grad_norm": 0.4456920325756073,
      "learning_rate": 7.435815966367326e-06,
      "loss": 0.3411,
      "step": 22964
    },
    {
      "epoch": 1.0538754531687393,
      "grad_norm": 0.4918728768825531,
      "learning_rate": 7.43560183815357e-06,
      "loss": 0.3988,
      "step": 22965
    },
    {
      "epoch": 1.0539213436739938,
      "grad_norm": 0.48221755027770996,
      "learning_rate": 7.43538770408296e-06,
      "loss": 0.318,
      "step": 22966
    },
    {
      "epoch": 1.0539672341792483,
      "grad_norm": 0.5157448649406433,
      "learning_rate": 7.435173564156012e-06,
      "loss": 0.4322,
      "step": 22967
    },
    {
      "epoch": 1.0540131246845028,
      "grad_norm": 0.44441649317741394,
      "learning_rate": 7.43495941837324e-06,
      "loss": 0.3144,
      "step": 22968
    },
    {
      "epoch": 1.0540590151897573,
      "grad_norm": 0.5442938804626465,
      "learning_rate": 7.4347452667351575e-06,
      "loss": 0.401,
      "step": 22969
    },
    {
      "epoch": 1.0541049056950118,
      "grad_norm": 0.4727689027786255,
      "learning_rate": 7.434531109242283e-06,
      "loss": 0.3208,
      "step": 22970
    },
    {
      "epoch": 1.054150796200266,
      "grad_norm": 0.4457760751247406,
      "learning_rate": 7.434316945895129e-06,
      "loss": 0.2937,
      "step": 22971
    },
    {
      "epoch": 1.0541966867055206,
      "grad_norm": 0.5181297063827515,
      "learning_rate": 7.434102776694211e-06,
      "loss": 0.4461,
      "step": 22972
    },
    {
      "epoch": 1.054242577210775,
      "grad_norm": 0.430737167596817,
      "learning_rate": 7.433888601640043e-06,
      "loss": 0.2814,
      "step": 22973
    },
    {
      "epoch": 1.0542884677160296,
      "grad_norm": 0.49610990285873413,
      "learning_rate": 7.433674420733142e-06,
      "loss": 0.3536,
      "step": 22974
    },
    {
      "epoch": 1.054334358221284,
      "grad_norm": 0.48436692357063293,
      "learning_rate": 7.433460233974023e-06,
      "loss": 0.3551,
      "step": 22975
    },
    {
      "epoch": 1.0543802487265386,
      "grad_norm": 0.4693945050239563,
      "learning_rate": 7.433246041363199e-06,
      "loss": 0.363,
      "step": 22976
    },
    {
      "epoch": 1.0544261392317928,
      "grad_norm": 0.4755348861217499,
      "learning_rate": 7.433031842901187e-06,
      "loss": 0.3393,
      "step": 22977
    },
    {
      "epoch": 1.0544720297370473,
      "grad_norm": 0.47202160954475403,
      "learning_rate": 7.4328176385885e-06,
      "loss": 0.3716,
      "step": 22978
    },
    {
      "epoch": 1.0545179202423018,
      "grad_norm": 0.4674060642719269,
      "learning_rate": 7.432603428425655e-06,
      "loss": 0.3943,
      "step": 22979
    },
    {
      "epoch": 1.0545638107475563,
      "grad_norm": 0.4633082449436188,
      "learning_rate": 7.432389212413167e-06,
      "loss": 0.3885,
      "step": 22980
    },
    {
      "epoch": 1.0546097012528108,
      "grad_norm": 0.47608405351638794,
      "learning_rate": 7.4321749905515505e-06,
      "loss": 0.3582,
      "step": 22981
    },
    {
      "epoch": 1.0546555917580653,
      "grad_norm": 0.48729392886161804,
      "learning_rate": 7.431960762841321e-06,
      "loss": 0.3607,
      "step": 22982
    },
    {
      "epoch": 1.0547014822633196,
      "grad_norm": 0.5162984728813171,
      "learning_rate": 7.431746529282993e-06,
      "loss": 0.3812,
      "step": 22983
    },
    {
      "epoch": 1.054747372768574,
      "grad_norm": 0.458340585231781,
      "learning_rate": 7.431532289877081e-06,
      "loss": 0.3422,
      "step": 22984
    },
    {
      "epoch": 1.0547932632738286,
      "grad_norm": 0.45092713832855225,
      "learning_rate": 7.4313180446241025e-06,
      "loss": 0.2996,
      "step": 22985
    },
    {
      "epoch": 1.054839153779083,
      "grad_norm": 0.4842176139354706,
      "learning_rate": 7.431103793524571e-06,
      "loss": 0.3624,
      "step": 22986
    },
    {
      "epoch": 1.0548850442843376,
      "grad_norm": 0.4518870413303375,
      "learning_rate": 7.430889536579003e-06,
      "loss": 0.3331,
      "step": 22987
    },
    {
      "epoch": 1.054930934789592,
      "grad_norm": 0.4805007576942444,
      "learning_rate": 7.430675273787913e-06,
      "loss": 0.4237,
      "step": 22988
    },
    {
      "epoch": 1.0549768252948466,
      "grad_norm": 0.46103641390800476,
      "learning_rate": 7.430461005151815e-06,
      "loss": 0.3646,
      "step": 22989
    },
    {
      "epoch": 1.0550227158001009,
      "grad_norm": 0.4092772901058197,
      "learning_rate": 7.430246730671226e-06,
      "loss": 0.2423,
      "step": 22990
    },
    {
      "epoch": 1.0550686063053554,
      "grad_norm": 0.4759734869003296,
      "learning_rate": 7.430032450346661e-06,
      "loss": 0.3269,
      "step": 22991
    },
    {
      "epoch": 1.0551144968106099,
      "grad_norm": 0.4694686532020569,
      "learning_rate": 7.429818164178634e-06,
      "loss": 0.3288,
      "step": 22992
    },
    {
      "epoch": 1.0551603873158644,
      "grad_norm": 0.5112667083740234,
      "learning_rate": 7.429603872167661e-06,
      "loss": 0.5059,
      "step": 22993
    },
    {
      "epoch": 1.0552062778211189,
      "grad_norm": 0.46050316095352173,
      "learning_rate": 7.429389574314259e-06,
      "loss": 0.3812,
      "step": 22994
    },
    {
      "epoch": 1.0552521683263734,
      "grad_norm": 0.48391494154930115,
      "learning_rate": 7.42917527061894e-06,
      "loss": 0.3247,
      "step": 22995
    },
    {
      "epoch": 1.0552980588316276,
      "grad_norm": 0.5052779316902161,
      "learning_rate": 7.428960961082221e-06,
      "loss": 0.4036,
      "step": 22996
    },
    {
      "epoch": 1.0553439493368821,
      "grad_norm": 0.48068252205848694,
      "learning_rate": 7.428746645704619e-06,
      "loss": 0.4004,
      "step": 22997
    },
    {
      "epoch": 1.0553898398421366,
      "grad_norm": 0.4690626859664917,
      "learning_rate": 7.4285323244866484e-06,
      "loss": 0.3736,
      "step": 22998
    },
    {
      "epoch": 1.0554357303473911,
      "grad_norm": 0.4160265326499939,
      "learning_rate": 7.428317997428822e-06,
      "loss": 0.3011,
      "step": 22999
    },
    {
      "epoch": 1.0554816208526456,
      "grad_norm": 0.4719209671020508,
      "learning_rate": 7.428103664531659e-06,
      "loss": 0.338,
      "step": 23000
    },
    {
      "epoch": 1.0555275113579001,
      "grad_norm": 0.4483465552330017,
      "learning_rate": 7.427889325795671e-06,
      "loss": 0.3047,
      "step": 23001
    },
    {
      "epoch": 1.0555734018631546,
      "grad_norm": 0.45481348037719727,
      "learning_rate": 7.427674981221377e-06,
      "loss": 0.3257,
      "step": 23002
    },
    {
      "epoch": 1.055619292368409,
      "grad_norm": 0.4806005656719208,
      "learning_rate": 7.427460630809291e-06,
      "loss": 0.3714,
      "step": 23003
    },
    {
      "epoch": 1.0556651828736634,
      "grad_norm": 0.4332531690597534,
      "learning_rate": 7.4272462745599275e-06,
      "loss": 0.2944,
      "step": 23004
    },
    {
      "epoch": 1.055711073378918,
      "grad_norm": 0.4874333143234253,
      "learning_rate": 7.427031912473803e-06,
      "loss": 0.3985,
      "step": 23005
    },
    {
      "epoch": 1.0557569638841724,
      "grad_norm": 0.4397573471069336,
      "learning_rate": 7.426817544551433e-06,
      "loss": 0.2927,
      "step": 23006
    },
    {
      "epoch": 1.055802854389427,
      "grad_norm": 0.42347070574760437,
      "learning_rate": 7.4266031707933315e-06,
      "loss": 0.3109,
      "step": 23007
    },
    {
      "epoch": 1.0558487448946814,
      "grad_norm": 0.4873827397823334,
      "learning_rate": 7.426388791200017e-06,
      "loss": 0.4254,
      "step": 23008
    },
    {
      "epoch": 1.0558946353999357,
      "grad_norm": 0.45702075958251953,
      "learning_rate": 7.426174405772002e-06,
      "loss": 0.3647,
      "step": 23009
    },
    {
      "epoch": 1.0559405259051902,
      "grad_norm": 0.41772589087486267,
      "learning_rate": 7.4259600145098034e-06,
      "loss": 0.2839,
      "step": 23010
    },
    {
      "epoch": 1.0559864164104447,
      "grad_norm": 0.48398715257644653,
      "learning_rate": 7.425745617413938e-06,
      "loss": 0.2997,
      "step": 23011
    },
    {
      "epoch": 1.0560323069156992,
      "grad_norm": 0.6297457218170166,
      "learning_rate": 7.425531214484918e-06,
      "loss": 0.3963,
      "step": 23012
    },
    {
      "epoch": 1.0560781974209537,
      "grad_norm": 0.4255613684654236,
      "learning_rate": 7.425316805723262e-06,
      "loss": 0.3037,
      "step": 23013
    },
    {
      "epoch": 1.0561240879262082,
      "grad_norm": 0.46519380807876587,
      "learning_rate": 7.425102391129485e-06,
      "loss": 0.3696,
      "step": 23014
    },
    {
      "epoch": 1.0561699784314624,
      "grad_norm": 0.4854913055896759,
      "learning_rate": 7.424887970704101e-06,
      "loss": 0.3303,
      "step": 23015
    },
    {
      "epoch": 1.056215868936717,
      "grad_norm": 0.48908936977386475,
      "learning_rate": 7.424673544447628e-06,
      "loss": 0.3966,
      "step": 23016
    },
    {
      "epoch": 1.0562617594419714,
      "grad_norm": 0.4920559227466583,
      "learning_rate": 7.424459112360579e-06,
      "loss": 0.3564,
      "step": 23017
    },
    {
      "epoch": 1.056307649947226,
      "grad_norm": 0.464352011680603,
      "learning_rate": 7.424244674443472e-06,
      "loss": 0.3172,
      "step": 23018
    },
    {
      "epoch": 1.0563535404524804,
      "grad_norm": 0.44081026315689087,
      "learning_rate": 7.424030230696821e-06,
      "loss": 0.3118,
      "step": 23019
    },
    {
      "epoch": 1.056399430957735,
      "grad_norm": 0.4807784855365753,
      "learning_rate": 7.423815781121144e-06,
      "loss": 0.3553,
      "step": 23020
    },
    {
      "epoch": 1.0564453214629892,
      "grad_norm": 0.44177550077438354,
      "learning_rate": 7.423601325716952e-06,
      "loss": 0.292,
      "step": 23021
    },
    {
      "epoch": 1.0564912119682437,
      "grad_norm": 0.462263822555542,
      "learning_rate": 7.423386864484768e-06,
      "loss": 0.3295,
      "step": 23022
    },
    {
      "epoch": 1.0565371024734982,
      "grad_norm": 0.5075477957725525,
      "learning_rate": 7.423172397425099e-06,
      "loss": 0.3814,
      "step": 23023
    },
    {
      "epoch": 1.0565829929787527,
      "grad_norm": 0.45924413204193115,
      "learning_rate": 7.422957924538467e-06,
      "loss": 0.3774,
      "step": 23024
    },
    {
      "epoch": 1.0566288834840072,
      "grad_norm": 0.4755311608314514,
      "learning_rate": 7.422743445825387e-06,
      "loss": 0.4136,
      "step": 23025
    },
    {
      "epoch": 1.0566747739892617,
      "grad_norm": 0.4683804214000702,
      "learning_rate": 7.4225289612863725e-06,
      "loss": 0.4206,
      "step": 23026
    },
    {
      "epoch": 1.0567206644945162,
      "grad_norm": 0.48861759901046753,
      "learning_rate": 7.42231447092194e-06,
      "loss": 0.4142,
      "step": 23027
    },
    {
      "epoch": 1.0567665549997705,
      "grad_norm": 0.47811856865882874,
      "learning_rate": 7.422099974732606e-06,
      "loss": 0.4095,
      "step": 23028
    },
    {
      "epoch": 1.056812445505025,
      "grad_norm": 0.4819469153881073,
      "learning_rate": 7.421885472718886e-06,
      "loss": 0.3831,
      "step": 23029
    },
    {
      "epoch": 1.0568583360102795,
      "grad_norm": 0.48023471236228943,
      "learning_rate": 7.421670964881297e-06,
      "loss": 0.3944,
      "step": 23030
    },
    {
      "epoch": 1.056904226515534,
      "grad_norm": 0.45764973759651184,
      "learning_rate": 7.421456451220353e-06,
      "loss": 0.3176,
      "step": 23031
    },
    {
      "epoch": 1.0569501170207884,
      "grad_norm": 0.43197572231292725,
      "learning_rate": 7.421241931736569e-06,
      "loss": 0.3015,
      "step": 23032
    },
    {
      "epoch": 1.056996007526043,
      "grad_norm": 0.4915072023868561,
      "learning_rate": 7.421027406430464e-06,
      "loss": 0.4018,
      "step": 23033
    },
    {
      "epoch": 1.0570418980312972,
      "grad_norm": 0.4470650553703308,
      "learning_rate": 7.420812875302551e-06,
      "loss": 0.3619,
      "step": 23034
    },
    {
      "epoch": 1.0570877885365517,
      "grad_norm": 0.4715549647808075,
      "learning_rate": 7.420598338353348e-06,
      "loss": 0.3094,
      "step": 23035
    },
    {
      "epoch": 1.0571336790418062,
      "grad_norm": 0.4616329073905945,
      "learning_rate": 7.42038379558337e-06,
      "loss": 0.3611,
      "step": 23036
    },
    {
      "epoch": 1.0571795695470607,
      "grad_norm": 0.47509992122650146,
      "learning_rate": 7.420169246993133e-06,
      "loss": 0.3676,
      "step": 23037
    },
    {
      "epoch": 1.0572254600523152,
      "grad_norm": 0.4342285990715027,
      "learning_rate": 7.419954692583151e-06,
      "loss": 0.3196,
      "step": 23038
    },
    {
      "epoch": 1.0572713505575697,
      "grad_norm": 0.4652702212333679,
      "learning_rate": 7.419740132353944e-06,
      "loss": 0.3333,
      "step": 23039
    },
    {
      "epoch": 1.0573172410628242,
      "grad_norm": 0.4997953772544861,
      "learning_rate": 7.419525566306024e-06,
      "loss": 0.3756,
      "step": 23040
    },
    {
      "epoch": 1.0573631315680785,
      "grad_norm": 0.4575677216053009,
      "learning_rate": 7.41931099443991e-06,
      "loss": 0.3302,
      "step": 23041
    },
    {
      "epoch": 1.057409022073333,
      "grad_norm": 0.5354112982749939,
      "learning_rate": 7.419096416756116e-06,
      "loss": 0.464,
      "step": 23042
    },
    {
      "epoch": 1.0574549125785875,
      "grad_norm": 0.4812207818031311,
      "learning_rate": 7.418881833255158e-06,
      "loss": 0.4137,
      "step": 23043
    },
    {
      "epoch": 1.057500803083842,
      "grad_norm": 0.48856064677238464,
      "learning_rate": 7.4186672439375525e-06,
      "loss": 0.4105,
      "step": 23044
    },
    {
      "epoch": 1.0575466935890965,
      "grad_norm": 0.47054198384284973,
      "learning_rate": 7.418452648803818e-06,
      "loss": 0.3897,
      "step": 23045
    },
    {
      "epoch": 1.057592584094351,
      "grad_norm": 0.4522615373134613,
      "learning_rate": 7.418238047854466e-06,
      "loss": 0.3388,
      "step": 23046
    },
    {
      "epoch": 1.0576384745996052,
      "grad_norm": 0.4636993408203125,
      "learning_rate": 7.418023441090016e-06,
      "loss": 0.3613,
      "step": 23047
    },
    {
      "epoch": 1.0576843651048597,
      "grad_norm": 0.5913541913032532,
      "learning_rate": 7.417808828510982e-06,
      "loss": 0.3344,
      "step": 23048
    },
    {
      "epoch": 1.0577302556101142,
      "grad_norm": 0.46286001801490784,
      "learning_rate": 7.41759421011788e-06,
      "loss": 0.3555,
      "step": 23049
    },
    {
      "epoch": 1.0577761461153687,
      "grad_norm": 0.47120001912117004,
      "learning_rate": 7.417379585911228e-06,
      "loss": 0.373,
      "step": 23050
    },
    {
      "epoch": 1.0578220366206232,
      "grad_norm": 0.49049481749534607,
      "learning_rate": 7.41716495589154e-06,
      "loss": 0.4402,
      "step": 23051
    },
    {
      "epoch": 1.0578679271258777,
      "grad_norm": 0.46740633249282837,
      "learning_rate": 7.416950320059335e-06,
      "loss": 0.3571,
      "step": 23052
    },
    {
      "epoch": 1.057913817631132,
      "grad_norm": 0.49441757798194885,
      "learning_rate": 7.416735678415127e-06,
      "loss": 0.3559,
      "step": 23053
    },
    {
      "epoch": 1.0579597081363865,
      "grad_norm": 0.46950918436050415,
      "learning_rate": 7.41652103095943e-06,
      "loss": 0.3521,
      "step": 23054
    },
    {
      "epoch": 1.058005598641641,
      "grad_norm": 0.46083974838256836,
      "learning_rate": 7.416306377692765e-06,
      "loss": 0.3477,
      "step": 23055
    },
    {
      "epoch": 1.0580514891468955,
      "grad_norm": 0.49763017892837524,
      "learning_rate": 7.416091718615645e-06,
      "loss": 0.3629,
      "step": 23056
    },
    {
      "epoch": 1.05809737965215,
      "grad_norm": 0.45558443665504456,
      "learning_rate": 7.4158770537285865e-06,
      "loss": 0.3224,
      "step": 23057
    },
    {
      "epoch": 1.0581432701574045,
      "grad_norm": 0.47926440834999084,
      "learning_rate": 7.415662383032107e-06,
      "loss": 0.3268,
      "step": 23058
    },
    {
      "epoch": 1.058189160662659,
      "grad_norm": 0.4860941767692566,
      "learning_rate": 7.415447706526722e-06,
      "loss": 0.3613,
      "step": 23059
    },
    {
      "epoch": 1.0582350511679133,
      "grad_norm": 0.5255950689315796,
      "learning_rate": 7.415233024212947e-06,
      "loss": 0.3531,
      "step": 23060
    },
    {
      "epoch": 1.0582809416731678,
      "grad_norm": 0.5197268724441528,
      "learning_rate": 7.415018336091299e-06,
      "loss": 0.4072,
      "step": 23061
    },
    {
      "epoch": 1.0583268321784223,
      "grad_norm": 0.4712773859500885,
      "learning_rate": 7.414803642162294e-06,
      "loss": 0.3674,
      "step": 23062
    },
    {
      "epoch": 1.0583727226836768,
      "grad_norm": 0.4689558744430542,
      "learning_rate": 7.414588942426448e-06,
      "loss": 0.3308,
      "step": 23063
    },
    {
      "epoch": 1.0584186131889313,
      "grad_norm": 0.47313669323921204,
      "learning_rate": 7.414374236884279e-06,
      "loss": 0.3418,
      "step": 23064
    },
    {
      "epoch": 1.0584645036941858,
      "grad_norm": 0.4789111912250519,
      "learning_rate": 7.414159525536301e-06,
      "loss": 0.3872,
      "step": 23065
    },
    {
      "epoch": 1.05851039419944,
      "grad_norm": 0.4722191393375397,
      "learning_rate": 7.413944808383031e-06,
      "loss": 0.4332,
      "step": 23066
    },
    {
      "epoch": 1.0585562847046945,
      "grad_norm": 0.48246580362319946,
      "learning_rate": 7.413730085424986e-06,
      "loss": 0.3737,
      "step": 23067
    },
    {
      "epoch": 1.058602175209949,
      "grad_norm": 0.4831930994987488,
      "learning_rate": 7.413515356662682e-06,
      "loss": 0.3683,
      "step": 23068
    },
    {
      "epoch": 1.0586480657152035,
      "grad_norm": 0.4684121906757355,
      "learning_rate": 7.413300622096635e-06,
      "loss": 0.3153,
      "step": 23069
    },
    {
      "epoch": 1.058693956220458,
      "grad_norm": 0.45734912157058716,
      "learning_rate": 7.413085881727362e-06,
      "loss": 0.2861,
      "step": 23070
    },
    {
      "epoch": 1.0587398467257125,
      "grad_norm": 0.45429527759552,
      "learning_rate": 7.41287113555538e-06,
      "loss": 0.3648,
      "step": 23071
    },
    {
      "epoch": 1.0587857372309668,
      "grad_norm": 0.4772087633609772,
      "learning_rate": 7.412656383581202e-06,
      "loss": 0.413,
      "step": 23072
    },
    {
      "epoch": 1.0588316277362213,
      "grad_norm": 0.45778122544288635,
      "learning_rate": 7.412441625805349e-06,
      "loss": 0.3621,
      "step": 23073
    },
    {
      "epoch": 1.0588775182414758,
      "grad_norm": 0.4943199157714844,
      "learning_rate": 7.412226862228334e-06,
      "loss": 0.3859,
      "step": 23074
    },
    {
      "epoch": 1.0589234087467303,
      "grad_norm": 0.46181315183639526,
      "learning_rate": 7.412012092850676e-06,
      "loss": 0.389,
      "step": 23075
    },
    {
      "epoch": 1.0589692992519848,
      "grad_norm": 0.4671902656555176,
      "learning_rate": 7.411797317672888e-06,
      "loss": 0.3655,
      "step": 23076
    },
    {
      "epoch": 1.0590151897572393,
      "grad_norm": 0.5070584416389465,
      "learning_rate": 7.411582536695491e-06,
      "loss": 0.3524,
      "step": 23077
    },
    {
      "epoch": 1.0590610802624938,
      "grad_norm": 0.45965513586997986,
      "learning_rate": 7.411367749918997e-06,
      "loss": 0.3852,
      "step": 23078
    },
    {
      "epoch": 1.059106970767748,
      "grad_norm": 0.47812527418136597,
      "learning_rate": 7.4111529573439265e-06,
      "loss": 0.3473,
      "step": 23079
    },
    {
      "epoch": 1.0591528612730026,
      "grad_norm": 0.4694713056087494,
      "learning_rate": 7.410938158970794e-06,
      "loss": 0.3476,
      "step": 23080
    },
    {
      "epoch": 1.059198751778257,
      "grad_norm": 0.4693313539028168,
      "learning_rate": 7.410723354800116e-06,
      "loss": 0.3695,
      "step": 23081
    },
    {
      "epoch": 1.0592446422835116,
      "grad_norm": 0.48752087354660034,
      "learning_rate": 7.410508544832407e-06,
      "loss": 0.3993,
      "step": 23082
    },
    {
      "epoch": 1.059290532788766,
      "grad_norm": 0.4743998050689697,
      "learning_rate": 7.4102937290681875e-06,
      "loss": 0.3366,
      "step": 23083
    },
    {
      "epoch": 1.0593364232940206,
      "grad_norm": 0.46278688311576843,
      "learning_rate": 7.410078907507973e-06,
      "loss": 0.394,
      "step": 23084
    },
    {
      "epoch": 1.0593823137992748,
      "grad_norm": 0.48054927587509155,
      "learning_rate": 7.409864080152279e-06,
      "loss": 0.3576,
      "step": 23085
    },
    {
      "epoch": 1.0594282043045293,
      "grad_norm": 0.46398183703422546,
      "learning_rate": 7.409649247001621e-06,
      "loss": 0.3442,
      "step": 23086
    },
    {
      "epoch": 1.0594740948097838,
      "grad_norm": 0.4538950026035309,
      "learning_rate": 7.409434408056519e-06,
      "loss": 0.3553,
      "step": 23087
    },
    {
      "epoch": 1.0595199853150383,
      "grad_norm": 0.5015366077423096,
      "learning_rate": 7.409219563317487e-06,
      "loss": 0.4608,
      "step": 23088
    },
    {
      "epoch": 1.0595658758202928,
      "grad_norm": 0.46929410099983215,
      "learning_rate": 7.409004712785041e-06,
      "loss": 0.3317,
      "step": 23089
    },
    {
      "epoch": 1.0596117663255473,
      "grad_norm": 0.4957880973815918,
      "learning_rate": 7.408789856459701e-06,
      "loss": 0.3769,
      "step": 23090
    },
    {
      "epoch": 1.0596576568308018,
      "grad_norm": 0.4621545374393463,
      "learning_rate": 7.4085749943419805e-06,
      "loss": 0.3403,
      "step": 23091
    },
    {
      "epoch": 1.059703547336056,
      "grad_norm": 0.45739486813545227,
      "learning_rate": 7.408360126432396e-06,
      "loss": 0.3399,
      "step": 23092
    },
    {
      "epoch": 1.0597494378413106,
      "grad_norm": 0.4303765296936035,
      "learning_rate": 7.408145252731469e-06,
      "loss": 0.309,
      "step": 23093
    },
    {
      "epoch": 1.059795328346565,
      "grad_norm": 0.48599541187286377,
      "learning_rate": 7.40793037323971e-06,
      "loss": 0.409,
      "step": 23094
    },
    {
      "epoch": 1.0598412188518196,
      "grad_norm": 0.4751538634300232,
      "learning_rate": 7.4077154879576396e-06,
      "loss": 0.3821,
      "step": 23095
    },
    {
      "epoch": 1.059887109357074,
      "grad_norm": 0.48826760053634644,
      "learning_rate": 7.407500596885773e-06,
      "loss": 0.3787,
      "step": 23096
    },
    {
      "epoch": 1.0599329998623286,
      "grad_norm": 0.4446692168712616,
      "learning_rate": 7.407285700024627e-06,
      "loss": 0.3352,
      "step": 23097
    },
    {
      "epoch": 1.0599788903675829,
      "grad_norm": 0.46527180075645447,
      "learning_rate": 7.407070797374719e-06,
      "loss": 0.3557,
      "step": 23098
    },
    {
      "epoch": 1.0600247808728374,
      "grad_norm": 0.4475729465484619,
      "learning_rate": 7.406855888936565e-06,
      "loss": 0.3325,
      "step": 23099
    },
    {
      "epoch": 1.0600706713780919,
      "grad_norm": 0.4595482349395752,
      "learning_rate": 7.406640974710682e-06,
      "loss": 0.3718,
      "step": 23100
    },
    {
      "epoch": 1.0601165618833464,
      "grad_norm": 0.430745005607605,
      "learning_rate": 7.406426054697587e-06,
      "loss": 0.2876,
      "step": 23101
    },
    {
      "epoch": 1.0601624523886009,
      "grad_norm": 0.4403243958950043,
      "learning_rate": 7.406211128897798e-06,
      "loss": 0.3052,
      "step": 23102
    },
    {
      "epoch": 1.0602083428938553,
      "grad_norm": 0.47933998703956604,
      "learning_rate": 7.40599619731183e-06,
      "loss": 0.4203,
      "step": 23103
    },
    {
      "epoch": 1.0602542333991096,
      "grad_norm": 0.5219547748565674,
      "learning_rate": 7.405781259940201e-06,
      "loss": 0.4483,
      "step": 23104
    },
    {
      "epoch": 1.0603001239043641,
      "grad_norm": 0.41271719336509705,
      "learning_rate": 7.405566316783426e-06,
      "loss": 0.2698,
      "step": 23105
    },
    {
      "epoch": 1.0603460144096186,
      "grad_norm": 0.410192608833313,
      "learning_rate": 7.405351367842024e-06,
      "loss": 0.2912,
      "step": 23106
    },
    {
      "epoch": 1.0603919049148731,
      "grad_norm": 0.465892493724823,
      "learning_rate": 7.405136413116511e-06,
      "loss": 0.371,
      "step": 23107
    },
    {
      "epoch": 1.0604377954201276,
      "grad_norm": 0.4614354968070984,
      "learning_rate": 7.404921452607405e-06,
      "loss": 0.3421,
      "step": 23108
    },
    {
      "epoch": 1.0604836859253821,
      "grad_norm": 0.4555666446685791,
      "learning_rate": 7.40470648631522e-06,
      "loss": 0.3047,
      "step": 23109
    },
    {
      "epoch": 1.0605295764306364,
      "grad_norm": 0.5120908617973328,
      "learning_rate": 7.404491514240477e-06,
      "loss": 0.4328,
      "step": 23110
    },
    {
      "epoch": 1.0605754669358909,
      "grad_norm": 0.46374353766441345,
      "learning_rate": 7.404276536383689e-06,
      "loss": 0.3572,
      "step": 23111
    },
    {
      "epoch": 1.0606213574411454,
      "grad_norm": 0.5516995191574097,
      "learning_rate": 7.4040615527453766e-06,
      "loss": 0.5163,
      "step": 23112
    },
    {
      "epoch": 1.0606672479463999,
      "grad_norm": 0.5185773968696594,
      "learning_rate": 7.403846563326054e-06,
      "loss": 0.4452,
      "step": 23113
    },
    {
      "epoch": 1.0607131384516544,
      "grad_norm": 0.48388969898223877,
      "learning_rate": 7.403631568126238e-06,
      "loss": 0.3541,
      "step": 23114
    },
    {
      "epoch": 1.0607590289569089,
      "grad_norm": 0.47464025020599365,
      "learning_rate": 7.4034165671464475e-06,
      "loss": 0.3576,
      "step": 23115
    },
    {
      "epoch": 1.0608049194621634,
      "grad_norm": 0.44310787320137024,
      "learning_rate": 7.4032015603871985e-06,
      "loss": 0.3125,
      "step": 23116
    },
    {
      "epoch": 1.0608508099674177,
      "grad_norm": 0.4825877845287323,
      "learning_rate": 7.40298654784901e-06,
      "loss": 0.3718,
      "step": 23117
    },
    {
      "epoch": 1.0608967004726721,
      "grad_norm": 0.45042884349823,
      "learning_rate": 7.402771529532396e-06,
      "loss": 0.3329,
      "step": 23118
    },
    {
      "epoch": 1.0609425909779266,
      "grad_norm": 0.4516732394695282,
      "learning_rate": 7.402556505437873e-06,
      "loss": 0.3462,
      "step": 23119
    },
    {
      "epoch": 1.0609884814831811,
      "grad_norm": 0.47098278999328613,
      "learning_rate": 7.402341475565962e-06,
      "loss": 0.3352,
      "step": 23120
    },
    {
      "epoch": 1.0610343719884356,
      "grad_norm": 0.4459187984466553,
      "learning_rate": 7.402126439917176e-06,
      "loss": 0.3664,
      "step": 23121
    },
    {
      "epoch": 1.0610802624936901,
      "grad_norm": 0.47348111867904663,
      "learning_rate": 7.401911398492037e-06,
      "loss": 0.4467,
      "step": 23122
    },
    {
      "epoch": 1.0611261529989444,
      "grad_norm": 0.4263926148414612,
      "learning_rate": 7.401696351291056e-06,
      "loss": 0.3143,
      "step": 23123
    },
    {
      "epoch": 1.061172043504199,
      "grad_norm": 0.48775508999824524,
      "learning_rate": 7.4014812983147565e-06,
      "loss": 0.4097,
      "step": 23124
    },
    {
      "epoch": 1.0612179340094534,
      "grad_norm": 0.4734116792678833,
      "learning_rate": 7.401266239563649e-06,
      "loss": 0.3436,
      "step": 23125
    },
    {
      "epoch": 1.061263824514708,
      "grad_norm": 0.44242510199546814,
      "learning_rate": 7.401051175038256e-06,
      "loss": 0.3596,
      "step": 23126
    },
    {
      "epoch": 1.0613097150199624,
      "grad_norm": 0.4346720576286316,
      "learning_rate": 7.400836104739093e-06,
      "loss": 0.284,
      "step": 23127
    },
    {
      "epoch": 1.061355605525217,
      "grad_norm": 0.49245285987854004,
      "learning_rate": 7.400621028666677e-06,
      "loss": 0.4474,
      "step": 23128
    },
    {
      "epoch": 1.0614014960304714,
      "grad_norm": 0.4828285276889801,
      "learning_rate": 7.400405946821524e-06,
      "loss": 0.3699,
      "step": 23129
    },
    {
      "epoch": 1.0614473865357257,
      "grad_norm": 0.48777469992637634,
      "learning_rate": 7.400190859204153e-06,
      "loss": 0.3516,
      "step": 23130
    },
    {
      "epoch": 1.0614932770409802,
      "grad_norm": 0.4609070420265198,
      "learning_rate": 7.39997576581508e-06,
      "loss": 0.3779,
      "step": 23131
    },
    {
      "epoch": 1.0615391675462347,
      "grad_norm": 0.5027076601982117,
      "learning_rate": 7.399760666654823e-06,
      "loss": 0.4028,
      "step": 23132
    },
    {
      "epoch": 1.0615850580514892,
      "grad_norm": 0.48082324862480164,
      "learning_rate": 7.3995455617238985e-06,
      "loss": 0.3607,
      "step": 23133
    },
    {
      "epoch": 1.0616309485567437,
      "grad_norm": 0.4866553843021393,
      "learning_rate": 7.399330451022826e-06,
      "loss": 0.3816,
      "step": 23134
    },
    {
      "epoch": 1.0616768390619982,
      "grad_norm": 0.45592764019966125,
      "learning_rate": 7.399115334552121e-06,
      "loss": 0.3465,
      "step": 23135
    },
    {
      "epoch": 1.0617227295672524,
      "grad_norm": 0.43276214599609375,
      "learning_rate": 7.398900212312298e-06,
      "loss": 0.2933,
      "step": 23136
    },
    {
      "epoch": 1.061768620072507,
      "grad_norm": 0.4474048316478729,
      "learning_rate": 7.398685084303878e-06,
      "loss": 0.2897,
      "step": 23137
    },
    {
      "epoch": 1.0618145105777614,
      "grad_norm": 0.4103580415248871,
      "learning_rate": 7.398469950527378e-06,
      "loss": 0.2656,
      "step": 23138
    },
    {
      "epoch": 1.061860401083016,
      "grad_norm": 0.44662556052207947,
      "learning_rate": 7.3982548109833165e-06,
      "loss": 0.3555,
      "step": 23139
    },
    {
      "epoch": 1.0619062915882704,
      "grad_norm": 0.4791778028011322,
      "learning_rate": 7.3980396656722066e-06,
      "loss": 0.3506,
      "step": 23140
    },
    {
      "epoch": 1.061952182093525,
      "grad_norm": 0.5053661465644836,
      "learning_rate": 7.397824514594569e-06,
      "loss": 0.3681,
      "step": 23141
    },
    {
      "epoch": 1.0619980725987792,
      "grad_norm": 0.460355281829834,
      "learning_rate": 7.39760935775092e-06,
      "loss": 0.3924,
      "step": 23142
    },
    {
      "epoch": 1.0620439631040337,
      "grad_norm": 0.4585028290748596,
      "learning_rate": 7.397394195141776e-06,
      "loss": 0.3829,
      "step": 23143
    },
    {
      "epoch": 1.0620898536092882,
      "grad_norm": 0.48816028237342834,
      "learning_rate": 7.397179026767658e-06,
      "loss": 0.4127,
      "step": 23144
    },
    {
      "epoch": 1.0621357441145427,
      "grad_norm": 0.47319790720939636,
      "learning_rate": 7.39696385262908e-06,
      "loss": 0.3609,
      "step": 23145
    },
    {
      "epoch": 1.0621816346197972,
      "grad_norm": 0.47193384170532227,
      "learning_rate": 7.396748672726561e-06,
      "loss": 0.3332,
      "step": 23146
    },
    {
      "epoch": 1.0622275251250517,
      "grad_norm": 0.4428778290748596,
      "learning_rate": 7.396533487060617e-06,
      "loss": 0.3268,
      "step": 23147
    },
    {
      "epoch": 1.0622734156303062,
      "grad_norm": 0.5135975480079651,
      "learning_rate": 7.396318295631766e-06,
      "loss": 0.4228,
      "step": 23148
    },
    {
      "epoch": 1.0623193061355605,
      "grad_norm": 0.46807822585105896,
      "learning_rate": 7.396103098440527e-06,
      "loss": 0.3524,
      "step": 23149
    },
    {
      "epoch": 1.062365196640815,
      "grad_norm": 0.48073747754096985,
      "learning_rate": 7.395887895487417e-06,
      "loss": 0.366,
      "step": 23150
    },
    {
      "epoch": 1.0624110871460695,
      "grad_norm": 0.48691093921661377,
      "learning_rate": 7.39567268677295e-06,
      "loss": 0.3823,
      "step": 23151
    },
    {
      "epoch": 1.062456977651324,
      "grad_norm": 0.4414513111114502,
      "learning_rate": 7.395457472297649e-06,
      "loss": 0.3341,
      "step": 23152
    },
    {
      "epoch": 1.0625028681565785,
      "grad_norm": 0.4797389507293701,
      "learning_rate": 7.395242252062027e-06,
      "loss": 0.3589,
      "step": 23153
    },
    {
      "epoch": 1.062548758661833,
      "grad_norm": 0.4618520140647888,
      "learning_rate": 7.395027026066604e-06,
      "loss": 0.3427,
      "step": 23154
    },
    {
      "epoch": 1.0625946491670872,
      "grad_norm": 0.4796909689903259,
      "learning_rate": 7.3948117943118954e-06,
      "loss": 0.3807,
      "step": 23155
    },
    {
      "epoch": 1.0626405396723417,
      "grad_norm": 0.4829982817173004,
      "learning_rate": 7.3945965567984235e-06,
      "loss": 0.4056,
      "step": 23156
    },
    {
      "epoch": 1.0626864301775962,
      "grad_norm": 0.42288678884506226,
      "learning_rate": 7.3943813135267e-06,
      "loss": 0.2691,
      "step": 23157
    },
    {
      "epoch": 1.0627323206828507,
      "grad_norm": 0.45509395003318787,
      "learning_rate": 7.394166064497246e-06,
      "loss": 0.3733,
      "step": 23158
    },
    {
      "epoch": 1.0627782111881052,
      "grad_norm": 0.42252933979034424,
      "learning_rate": 7.393950809710578e-06,
      "loss": 0.2908,
      "step": 23159
    },
    {
      "epoch": 1.0628241016933597,
      "grad_norm": 0.476764440536499,
      "learning_rate": 7.393735549167215e-06,
      "loss": 0.371,
      "step": 23160
    },
    {
      "epoch": 1.062869992198614,
      "grad_norm": 0.47859328985214233,
      "learning_rate": 7.393520282867672e-06,
      "loss": 0.3996,
      "step": 23161
    },
    {
      "epoch": 1.0629158827038685,
      "grad_norm": 0.45215800404548645,
      "learning_rate": 7.393305010812468e-06,
      "loss": 0.3298,
      "step": 23162
    },
    {
      "epoch": 1.062961773209123,
      "grad_norm": 0.46105122566223145,
      "learning_rate": 7.393089733002122e-06,
      "loss": 0.3205,
      "step": 23163
    },
    {
      "epoch": 1.0630076637143775,
      "grad_norm": 0.47521859407424927,
      "learning_rate": 7.392874449437151e-06,
      "loss": 0.4086,
      "step": 23164
    },
    {
      "epoch": 1.063053554219632,
      "grad_norm": 0.4428737163543701,
      "learning_rate": 7.39265916011807e-06,
      "loss": 0.3338,
      "step": 23165
    },
    {
      "epoch": 1.0630994447248865,
      "grad_norm": 0.42021840810775757,
      "learning_rate": 7.3924438650454e-06,
      "loss": 0.335,
      "step": 23166
    },
    {
      "epoch": 1.063145335230141,
      "grad_norm": 0.47668012976646423,
      "learning_rate": 7.392228564219658e-06,
      "loss": 0.3737,
      "step": 23167
    },
    {
      "epoch": 1.0631912257353953,
      "grad_norm": 0.46835047006607056,
      "learning_rate": 7.3920132576413605e-06,
      "loss": 0.3409,
      "step": 23168
    },
    {
      "epoch": 1.0632371162406498,
      "grad_norm": 0.44017091393470764,
      "learning_rate": 7.391797945311027e-06,
      "loss": 0.3242,
      "step": 23169
    },
    {
      "epoch": 1.0632830067459043,
      "grad_norm": 0.43188610672950745,
      "learning_rate": 7.391582627229173e-06,
      "loss": 0.2766,
      "step": 23170
    },
    {
      "epoch": 1.0633288972511588,
      "grad_norm": 0.4693223834037781,
      "learning_rate": 7.391367303396319e-06,
      "loss": 0.3401,
      "step": 23171
    },
    {
      "epoch": 1.0633747877564133,
      "grad_norm": 0.466511070728302,
      "learning_rate": 7.391151973812981e-06,
      "loss": 0.3639,
      "step": 23172
    },
    {
      "epoch": 1.0634206782616678,
      "grad_norm": 0.4652596414089203,
      "learning_rate": 7.390936638479676e-06,
      "loss": 0.3741,
      "step": 23173
    },
    {
      "epoch": 1.063466568766922,
      "grad_norm": 0.4702387750148773,
      "learning_rate": 7.390721297396924e-06,
      "loss": 0.357,
      "step": 23174
    },
    {
      "epoch": 1.0635124592721765,
      "grad_norm": 0.4831272065639496,
      "learning_rate": 7.3905059505652424e-06,
      "loss": 0.3535,
      "step": 23175
    },
    {
      "epoch": 1.063558349777431,
      "grad_norm": 0.49053022265434265,
      "learning_rate": 7.390290597985147e-06,
      "loss": 0.4237,
      "step": 23176
    },
    {
      "epoch": 1.0636042402826855,
      "grad_norm": 0.4525579810142517,
      "learning_rate": 7.390075239657157e-06,
      "loss": 0.3793,
      "step": 23177
    },
    {
      "epoch": 1.06365013078794,
      "grad_norm": 0.4613303542137146,
      "learning_rate": 7.389859875581793e-06,
      "loss": 0.3543,
      "step": 23178
    },
    {
      "epoch": 1.0636960212931945,
      "grad_norm": 0.4703523516654968,
      "learning_rate": 7.389644505759568e-06,
      "loss": 0.3624,
      "step": 23179
    },
    {
      "epoch": 1.063741911798449,
      "grad_norm": 0.5201200842857361,
      "learning_rate": 7.389429130191003e-06,
      "loss": 0.4358,
      "step": 23180
    },
    {
      "epoch": 1.0637878023037033,
      "grad_norm": 0.4796191453933716,
      "learning_rate": 7.389213748876614e-06,
      "loss": 0.368,
      "step": 23181
    },
    {
      "epoch": 1.0638336928089578,
      "grad_norm": 0.4687592089176178,
      "learning_rate": 7.388998361816921e-06,
      "loss": 0.389,
      "step": 23182
    },
    {
      "epoch": 1.0638795833142123,
      "grad_norm": 0.4758653938770294,
      "learning_rate": 7.388782969012441e-06,
      "loss": 0.3567,
      "step": 23183
    },
    {
      "epoch": 1.0639254738194668,
      "grad_norm": 0.42405059933662415,
      "learning_rate": 7.38856757046369e-06,
      "loss": 0.2867,
      "step": 23184
    },
    {
      "epoch": 1.0639713643247213,
      "grad_norm": 0.4148804247379303,
      "learning_rate": 7.38835216617119e-06,
      "loss": 0.2969,
      "step": 23185
    },
    {
      "epoch": 1.0640172548299758,
      "grad_norm": 0.462131142616272,
      "learning_rate": 7.388136756135457e-06,
      "loss": 0.3495,
      "step": 23186
    },
    {
      "epoch": 1.06406314533523,
      "grad_norm": 0.4592922329902649,
      "learning_rate": 7.387921340357007e-06,
      "loss": 0.3682,
      "step": 23187
    },
    {
      "epoch": 1.0641090358404846,
      "grad_norm": 0.47476041316986084,
      "learning_rate": 7.387705918836361e-06,
      "loss": 0.3651,
      "step": 23188
    },
    {
      "epoch": 1.064154926345739,
      "grad_norm": 0.4562109112739563,
      "learning_rate": 7.387490491574036e-06,
      "loss": 0.3206,
      "step": 23189
    },
    {
      "epoch": 1.0642008168509935,
      "grad_norm": 0.41860273480415344,
      "learning_rate": 7.387275058570549e-06,
      "loss": 0.3276,
      "step": 23190
    },
    {
      "epoch": 1.064246707356248,
      "grad_norm": 0.513583779335022,
      "learning_rate": 7.38705961982642e-06,
      "loss": 0.3918,
      "step": 23191
    },
    {
      "epoch": 1.0642925978615025,
      "grad_norm": 0.49761083722114563,
      "learning_rate": 7.386844175342165e-06,
      "loss": 0.3871,
      "step": 23192
    },
    {
      "epoch": 1.0643384883667568,
      "grad_norm": 0.48394280672073364,
      "learning_rate": 7.386628725118305e-06,
      "loss": 0.3469,
      "step": 23193
    },
    {
      "epoch": 1.0643843788720113,
      "grad_norm": 0.4566066861152649,
      "learning_rate": 7.386413269155353e-06,
      "loss": 0.3136,
      "step": 23194
    },
    {
      "epoch": 1.0644302693772658,
      "grad_norm": 0.4547611176967621,
      "learning_rate": 7.386197807453832e-06,
      "loss": 0.3344,
      "step": 23195
    },
    {
      "epoch": 1.0644761598825203,
      "grad_norm": 0.46940499544143677,
      "learning_rate": 7.385982340014258e-06,
      "loss": 0.3571,
      "step": 23196
    },
    {
      "epoch": 1.0645220503877748,
      "grad_norm": 0.42680832743644714,
      "learning_rate": 7.385766866837151e-06,
      "loss": 0.3315,
      "step": 23197
    },
    {
      "epoch": 1.0645679408930293,
      "grad_norm": 0.4713243246078491,
      "learning_rate": 7.385551387923025e-06,
      "loss": 0.3538,
      "step": 23198
    },
    {
      "epoch": 1.0646138313982836,
      "grad_norm": 0.46376854181289673,
      "learning_rate": 7.385335903272404e-06,
      "loss": 0.3255,
      "step": 23199
    },
    {
      "epoch": 1.064659721903538,
      "grad_norm": 0.4650602340698242,
      "learning_rate": 7.385120412885801e-06,
      "loss": 0.3971,
      "step": 23200
    },
    {
      "epoch": 1.0647056124087926,
      "grad_norm": 0.5004786252975464,
      "learning_rate": 7.384904916763736e-06,
      "loss": 0.3874,
      "step": 23201
    },
    {
      "epoch": 1.064751502914047,
      "grad_norm": 0.471303790807724,
      "learning_rate": 7.384689414906727e-06,
      "loss": 0.3913,
      "step": 23202
    },
    {
      "epoch": 1.0647973934193016,
      "grad_norm": 0.49919578433036804,
      "learning_rate": 7.384473907315295e-06,
      "loss": 0.3896,
      "step": 23203
    },
    {
      "epoch": 1.064843283924556,
      "grad_norm": 0.4801212251186371,
      "learning_rate": 7.384258393989954e-06,
      "loss": 0.4366,
      "step": 23204
    },
    {
      "epoch": 1.0648891744298106,
      "grad_norm": 0.4881910979747772,
      "learning_rate": 7.384042874931224e-06,
      "loss": 0.4146,
      "step": 23205
    },
    {
      "epoch": 1.0649350649350648,
      "grad_norm": 0.4771069586277008,
      "learning_rate": 7.383827350139624e-06,
      "loss": 0.35,
      "step": 23206
    },
    {
      "epoch": 1.0649809554403193,
      "grad_norm": 0.4326435327529907,
      "learning_rate": 7.383611819615671e-06,
      "loss": 0.2743,
      "step": 23207
    },
    {
      "epoch": 1.0650268459455738,
      "grad_norm": 0.49072134494781494,
      "learning_rate": 7.383396283359883e-06,
      "loss": 0.4059,
      "step": 23208
    },
    {
      "epoch": 1.0650727364508283,
      "grad_norm": 0.4681280255317688,
      "learning_rate": 7.383180741372781e-06,
      "loss": 0.3581,
      "step": 23209
    },
    {
      "epoch": 1.0651186269560828,
      "grad_norm": 0.42797401547431946,
      "learning_rate": 7.382965193654881e-06,
      "loss": 0.3102,
      "step": 23210
    },
    {
      "epoch": 1.0651645174613373,
      "grad_norm": 0.46155017614364624,
      "learning_rate": 7.3827496402067024e-06,
      "loss": 0.3426,
      "step": 23211
    },
    {
      "epoch": 1.0652104079665916,
      "grad_norm": 0.46186041831970215,
      "learning_rate": 7.3825340810287605e-06,
      "loss": 0.3193,
      "step": 23212
    },
    {
      "epoch": 1.065256298471846,
      "grad_norm": 0.4316633343696594,
      "learning_rate": 7.382318516121578e-06,
      "loss": 0.3385,
      "step": 23213
    },
    {
      "epoch": 1.0653021889771006,
      "grad_norm": 0.469899445772171,
      "learning_rate": 7.382102945485671e-06,
      "loss": 0.3812,
      "step": 23214
    },
    {
      "epoch": 1.065348079482355,
      "grad_norm": 0.44728755950927734,
      "learning_rate": 7.381887369121558e-06,
      "loss": 0.3416,
      "step": 23215
    },
    {
      "epoch": 1.0653939699876096,
      "grad_norm": 0.46301066875457764,
      "learning_rate": 7.381671787029758e-06,
      "loss": 0.3351,
      "step": 23216
    },
    {
      "epoch": 1.065439860492864,
      "grad_norm": 0.49562713503837585,
      "learning_rate": 7.381456199210789e-06,
      "loss": 0.2947,
      "step": 23217
    },
    {
      "epoch": 1.0654857509981186,
      "grad_norm": 0.4708433747291565,
      "learning_rate": 7.381240605665169e-06,
      "loss": 0.388,
      "step": 23218
    },
    {
      "epoch": 1.0655316415033729,
      "grad_norm": 0.4784267842769623,
      "learning_rate": 7.381025006393417e-06,
      "loss": 0.3432,
      "step": 23219
    },
    {
      "epoch": 1.0655775320086274,
      "grad_norm": 0.4662189185619354,
      "learning_rate": 7.380809401396052e-06,
      "loss": 0.3451,
      "step": 23220
    },
    {
      "epoch": 1.0656234225138819,
      "grad_norm": 0.5102803111076355,
      "learning_rate": 7.380593790673592e-06,
      "loss": 0.4064,
      "step": 23221
    },
    {
      "epoch": 1.0656693130191364,
      "grad_norm": 0.46173325181007385,
      "learning_rate": 7.380378174226554e-06,
      "loss": 0.2801,
      "step": 23222
    },
    {
      "epoch": 1.0657152035243909,
      "grad_norm": 0.4564485251903534,
      "learning_rate": 7.380162552055458e-06,
      "loss": 0.3569,
      "step": 23223
    },
    {
      "epoch": 1.0657610940296454,
      "grad_norm": 0.44951969385147095,
      "learning_rate": 7.3799469241608226e-06,
      "loss": 0.346,
      "step": 23224
    },
    {
      "epoch": 1.0658069845348996,
      "grad_norm": 0.4897228181362152,
      "learning_rate": 7.3797312905431654e-06,
      "loss": 0.3954,
      "step": 23225
    },
    {
      "epoch": 1.0658528750401541,
      "grad_norm": 0.4838710427284241,
      "learning_rate": 7.379515651203008e-06,
      "loss": 0.3318,
      "step": 23226
    },
    {
      "epoch": 1.0658987655454086,
      "grad_norm": 0.4566514194011688,
      "learning_rate": 7.379300006140863e-06,
      "loss": 0.3592,
      "step": 23227
    },
    {
      "epoch": 1.0659446560506631,
      "grad_norm": 0.5168399810791016,
      "learning_rate": 7.379084355357254e-06,
      "loss": 0.4536,
      "step": 23228
    },
    {
      "epoch": 1.0659905465559176,
      "grad_norm": 0.47856512665748596,
      "learning_rate": 7.378868698852698e-06,
      "loss": 0.3697,
      "step": 23229
    },
    {
      "epoch": 1.0660364370611721,
      "grad_norm": 0.5143166780471802,
      "learning_rate": 7.378653036627713e-06,
      "loss": 0.2195,
      "step": 23230
    },
    {
      "epoch": 1.0660823275664266,
      "grad_norm": 0.4776109755039215,
      "learning_rate": 7.378437368682819e-06,
      "loss": 0.3867,
      "step": 23231
    },
    {
      "epoch": 1.066128218071681,
      "grad_norm": 0.45903000235557556,
      "learning_rate": 7.3782216950185335e-06,
      "loss": 0.3569,
      "step": 23232
    },
    {
      "epoch": 1.0661741085769354,
      "grad_norm": 0.4664532244205475,
      "learning_rate": 7.378006015635375e-06,
      "loss": 0.3261,
      "step": 23233
    },
    {
      "epoch": 1.06621999908219,
      "grad_norm": 0.4476882517337799,
      "learning_rate": 7.377790330533864e-06,
      "loss": 0.3163,
      "step": 23234
    },
    {
      "epoch": 1.0662658895874444,
      "grad_norm": 0.4685955345630646,
      "learning_rate": 7.377574639714515e-06,
      "loss": 0.3792,
      "step": 23235
    },
    {
      "epoch": 1.066311780092699,
      "grad_norm": 0.4703274965286255,
      "learning_rate": 7.377358943177851e-06,
      "loss": 0.3535,
      "step": 23236
    },
    {
      "epoch": 1.0663576705979532,
      "grad_norm": 0.4951554238796234,
      "learning_rate": 7.377143240924392e-06,
      "loss": 0.3788,
      "step": 23237
    },
    {
      "epoch": 1.0664035611032077,
      "grad_norm": 0.5084136724472046,
      "learning_rate": 7.376927532954649e-06,
      "loss": 0.3986,
      "step": 23238
    },
    {
      "epoch": 1.0664494516084622,
      "grad_norm": 0.512022078037262,
      "learning_rate": 7.376711819269147e-06,
      "loss": 0.4653,
      "step": 23239
    },
    {
      "epoch": 1.0664953421137167,
      "grad_norm": 0.4714144766330719,
      "learning_rate": 7.376496099868404e-06,
      "loss": 0.371,
      "step": 23240
    },
    {
      "epoch": 1.0665412326189712,
      "grad_norm": 0.4414292275905609,
      "learning_rate": 7.376280374752936e-06,
      "loss": 0.2868,
      "step": 23241
    },
    {
      "epoch": 1.0665871231242257,
      "grad_norm": 0.4290190041065216,
      "learning_rate": 7.376064643923266e-06,
      "loss": 0.2913,
      "step": 23242
    },
    {
      "epoch": 1.0666330136294802,
      "grad_norm": 0.4632873833179474,
      "learning_rate": 7.37584890737991e-06,
      "loss": 0.3501,
      "step": 23243
    },
    {
      "epoch": 1.0666789041347344,
      "grad_norm": 0.4805276691913605,
      "learning_rate": 7.375633165123386e-06,
      "loss": 0.3605,
      "step": 23244
    },
    {
      "epoch": 1.066724794639989,
      "grad_norm": 0.4731072783470154,
      "learning_rate": 7.375417417154216e-06,
      "loss": 0.3668,
      "step": 23245
    },
    {
      "epoch": 1.0667706851452434,
      "grad_norm": 0.5391817688941956,
      "learning_rate": 7.375201663472915e-06,
      "loss": 0.4932,
      "step": 23246
    },
    {
      "epoch": 1.066816575650498,
      "grad_norm": 0.43861159682273865,
      "learning_rate": 7.374985904080003e-06,
      "loss": 0.3121,
      "step": 23247
    },
    {
      "epoch": 1.0668624661557524,
      "grad_norm": 0.4650154411792755,
      "learning_rate": 7.374770138976002e-06,
      "loss": 0.3502,
      "step": 23248
    },
    {
      "epoch": 1.066908356661007,
      "grad_norm": 0.4767824411392212,
      "learning_rate": 7.374554368161426e-06,
      "loss": 0.3588,
      "step": 23249
    },
    {
      "epoch": 1.0669542471662612,
      "grad_norm": 0.435228556394577,
      "learning_rate": 7.3743385916367974e-06,
      "loss": 0.2959,
      "step": 23250
    },
    {
      "epoch": 1.0670001376715157,
      "grad_norm": 0.44154050946235657,
      "learning_rate": 7.374122809402634e-06,
      "loss": 0.3001,
      "step": 23251
    },
    {
      "epoch": 1.0670460281767702,
      "grad_norm": 0.4635991156101227,
      "learning_rate": 7.373907021459454e-06,
      "loss": 0.344,
      "step": 23252
    },
    {
      "epoch": 1.0670919186820247,
      "grad_norm": 0.44885897636413574,
      "learning_rate": 7.373691227807778e-06,
      "loss": 0.3506,
      "step": 23253
    },
    {
      "epoch": 1.0671378091872792,
      "grad_norm": 0.4561697840690613,
      "learning_rate": 7.373475428448123e-06,
      "loss": 0.3225,
      "step": 23254
    },
    {
      "epoch": 1.0671836996925337,
      "grad_norm": 0.4934261739253998,
      "learning_rate": 7.373259623381007e-06,
      "loss": 0.3946,
      "step": 23255
    },
    {
      "epoch": 1.0672295901977882,
      "grad_norm": 0.5005345940589905,
      "learning_rate": 7.373043812606951e-06,
      "loss": 0.3925,
      "step": 23256
    },
    {
      "epoch": 1.0672754807030425,
      "grad_norm": 0.47204315662384033,
      "learning_rate": 7.372827996126476e-06,
      "loss": 0.3793,
      "step": 23257
    },
    {
      "epoch": 1.067321371208297,
      "grad_norm": 0.4954288601875305,
      "learning_rate": 7.372612173940097e-06,
      "loss": 0.3716,
      "step": 23258
    },
    {
      "epoch": 1.0673672617135515,
      "grad_norm": 0.5067251324653625,
      "learning_rate": 7.372396346048335e-06,
      "loss": 0.3958,
      "step": 23259
    },
    {
      "epoch": 1.067413152218806,
      "grad_norm": 0.48203274607658386,
      "learning_rate": 7.372180512451709e-06,
      "loss": 0.2547,
      "step": 23260
    },
    {
      "epoch": 1.0674590427240604,
      "grad_norm": 0.4858340919017792,
      "learning_rate": 7.371964673150735e-06,
      "loss": 0.3537,
      "step": 23261
    },
    {
      "epoch": 1.067504933229315,
      "grad_norm": 0.47714370489120483,
      "learning_rate": 7.371748828145937e-06,
      "loss": 0.3513,
      "step": 23262
    },
    {
      "epoch": 1.0675508237345692,
      "grad_norm": 0.4616807699203491,
      "learning_rate": 7.371532977437831e-06,
      "loss": 0.3211,
      "step": 23263
    },
    {
      "epoch": 1.0675967142398237,
      "grad_norm": 0.44778233766555786,
      "learning_rate": 7.371317121026937e-06,
      "loss": 0.3084,
      "step": 23264
    },
    {
      "epoch": 1.0676426047450782,
      "grad_norm": 0.4856511652469635,
      "learning_rate": 7.371101258913775e-06,
      "loss": 0.3442,
      "step": 23265
    },
    {
      "epoch": 1.0676884952503327,
      "grad_norm": 0.49679791927337646,
      "learning_rate": 7.37088539109886e-06,
      "loss": 0.398,
      "step": 23266
    },
    {
      "epoch": 1.0677343857555872,
      "grad_norm": 0.4765186011791229,
      "learning_rate": 7.370669517582714e-06,
      "loss": 0.357,
      "step": 23267
    },
    {
      "epoch": 1.0677802762608417,
      "grad_norm": 0.5197485089302063,
      "learning_rate": 7.370453638365858e-06,
      "loss": 0.3834,
      "step": 23268
    },
    {
      "epoch": 1.0678261667660962,
      "grad_norm": 0.5028704404830933,
      "learning_rate": 7.370237753448808e-06,
      "loss": 0.4068,
      "step": 23269
    },
    {
      "epoch": 1.0678720572713505,
      "grad_norm": 0.44178977608680725,
      "learning_rate": 7.370021862832084e-06,
      "loss": 0.2785,
      "step": 23270
    },
    {
      "epoch": 1.067917947776605,
      "grad_norm": 0.44496190547943115,
      "learning_rate": 7.369805966516207e-06,
      "loss": 0.3172,
      "step": 23271
    },
    {
      "epoch": 1.0679638382818595,
      "grad_norm": 0.4807088375091553,
      "learning_rate": 7.369590064501693e-06,
      "loss": 0.4275,
      "step": 23272
    },
    {
      "epoch": 1.068009728787114,
      "grad_norm": 0.4490721821784973,
      "learning_rate": 7.369374156789063e-06,
      "loss": 0.3447,
      "step": 23273
    },
    {
      "epoch": 1.0680556192923685,
      "grad_norm": 0.48562443256378174,
      "learning_rate": 7.369158243378836e-06,
      "loss": 0.3905,
      "step": 23274
    },
    {
      "epoch": 1.068101509797623,
      "grad_norm": 0.4537728428840637,
      "learning_rate": 7.368942324271532e-06,
      "loss": 0.3391,
      "step": 23275
    },
    {
      "epoch": 1.0681474003028772,
      "grad_norm": 0.48548364639282227,
      "learning_rate": 7.368726399467668e-06,
      "loss": 0.3217,
      "step": 23276
    },
    {
      "epoch": 1.0681932908081317,
      "grad_norm": 0.4656364321708679,
      "learning_rate": 7.368510468967767e-06,
      "loss": 0.3615,
      "step": 23277
    },
    {
      "epoch": 1.0682391813133862,
      "grad_norm": 0.4781460464000702,
      "learning_rate": 7.368294532772344e-06,
      "loss": 0.3479,
      "step": 23278
    },
    {
      "epoch": 1.0682850718186407,
      "grad_norm": 0.4770740568637848,
      "learning_rate": 7.36807859088192e-06,
      "loss": 0.382,
      "step": 23279
    },
    {
      "epoch": 1.0683309623238952,
      "grad_norm": 0.5299501419067383,
      "learning_rate": 7.367862643297015e-06,
      "loss": 0.4021,
      "step": 23280
    },
    {
      "epoch": 1.0683768528291497,
      "grad_norm": 0.4940410554409027,
      "learning_rate": 7.367646690018146e-06,
      "loss": 0.3752,
      "step": 23281
    },
    {
      "epoch": 1.068422743334404,
      "grad_norm": 0.4970751106739044,
      "learning_rate": 7.367430731045837e-06,
      "loss": 0.405,
      "step": 23282
    },
    {
      "epoch": 1.0684686338396585,
      "grad_norm": 0.6587731242179871,
      "learning_rate": 7.3672147663806015e-06,
      "loss": 0.4599,
      "step": 23283
    },
    {
      "epoch": 1.068514524344913,
      "grad_norm": 0.46962442994117737,
      "learning_rate": 7.366998796022962e-06,
      "loss": 0.3366,
      "step": 23284
    },
    {
      "epoch": 1.0685604148501675,
      "grad_norm": 0.4870610535144806,
      "learning_rate": 7.366782819973439e-06,
      "loss": 0.3511,
      "step": 23285
    },
    {
      "epoch": 1.068606305355422,
      "grad_norm": 0.4577733278274536,
      "learning_rate": 7.36656683823255e-06,
      "loss": 0.3258,
      "step": 23286
    },
    {
      "epoch": 1.0686521958606765,
      "grad_norm": 0.4981312155723572,
      "learning_rate": 7.3663508508008134e-06,
      "loss": 0.4167,
      "step": 23287
    },
    {
      "epoch": 1.0686980863659308,
      "grad_norm": 0.48805591464042664,
      "learning_rate": 7.366134857678751e-06,
      "loss": 0.3602,
      "step": 23288
    },
    {
      "epoch": 1.0687439768711853,
      "grad_norm": 0.46588394045829773,
      "learning_rate": 7.36591885886688e-06,
      "loss": 0.3851,
      "step": 23289
    },
    {
      "epoch": 1.0687898673764398,
      "grad_norm": 0.4560999274253845,
      "learning_rate": 7.365702854365721e-06,
      "loss": 0.3214,
      "step": 23290
    },
    {
      "epoch": 1.0688357578816943,
      "grad_norm": 0.521303117275238,
      "learning_rate": 7.365486844175794e-06,
      "loss": 0.398,
      "step": 23291
    },
    {
      "epoch": 1.0688816483869488,
      "grad_norm": 0.46699458360671997,
      "learning_rate": 7.365270828297618e-06,
      "loss": 0.3693,
      "step": 23292
    },
    {
      "epoch": 1.0689275388922033,
      "grad_norm": 0.43613696098327637,
      "learning_rate": 7.365054806731712e-06,
      "loss": 0.287,
      "step": 23293
    },
    {
      "epoch": 1.0689734293974578,
      "grad_norm": 0.47623035311698914,
      "learning_rate": 7.364838779478594e-06,
      "loss": 0.369,
      "step": 23294
    },
    {
      "epoch": 1.069019319902712,
      "grad_norm": 0.4262254536151886,
      "learning_rate": 7.3646227465387865e-06,
      "loss": 0.3035,
      "step": 23295
    },
    {
      "epoch": 1.0690652104079665,
      "grad_norm": 0.48829540610313416,
      "learning_rate": 7.364406707912808e-06,
      "loss": 0.4084,
      "step": 23296
    },
    {
      "epoch": 1.069111100913221,
      "grad_norm": 0.4771776795387268,
      "learning_rate": 7.364190663601178e-06,
      "loss": 0.371,
      "step": 23297
    },
    {
      "epoch": 1.0691569914184755,
      "grad_norm": 0.44669926166534424,
      "learning_rate": 7.363974613604413e-06,
      "loss": 0.28,
      "step": 23298
    },
    {
      "epoch": 1.06920288192373,
      "grad_norm": 0.5774834156036377,
      "learning_rate": 7.363758557923037e-06,
      "loss": 0.2672,
      "step": 23299
    },
    {
      "epoch": 1.0692487724289845,
      "grad_norm": 0.48455867171287537,
      "learning_rate": 7.363542496557568e-06,
      "loss": 0.4179,
      "step": 23300
    },
    {
      "epoch": 1.0692946629342388,
      "grad_norm": 0.44213566184043884,
      "learning_rate": 7.363326429508524e-06,
      "loss": 0.3204,
      "step": 23301
    },
    {
      "epoch": 1.0693405534394933,
      "grad_norm": 0.49709513783454895,
      "learning_rate": 7.3631103567764266e-06,
      "loss": 0.3919,
      "step": 23302
    },
    {
      "epoch": 1.0693864439447478,
      "grad_norm": 0.5035470128059387,
      "learning_rate": 7.362894278361795e-06,
      "loss": 0.4034,
      "step": 23303
    },
    {
      "epoch": 1.0694323344500023,
      "grad_norm": 0.46188509464263916,
      "learning_rate": 7.362678194265147e-06,
      "loss": 0.2875,
      "step": 23304
    },
    {
      "epoch": 1.0694782249552568,
      "grad_norm": 0.5251035094261169,
      "learning_rate": 7.362462104487005e-06,
      "loss": 0.4345,
      "step": 23305
    },
    {
      "epoch": 1.0695241154605113,
      "grad_norm": 0.48968422412872314,
      "learning_rate": 7.362246009027887e-06,
      "loss": 0.3836,
      "step": 23306
    },
    {
      "epoch": 1.0695700059657658,
      "grad_norm": 0.46700963377952576,
      "learning_rate": 7.362029907888313e-06,
      "loss": 0.3548,
      "step": 23307
    },
    {
      "epoch": 1.06961589647102,
      "grad_norm": 0.49318668246269226,
      "learning_rate": 7.361813801068802e-06,
      "loss": 0.3353,
      "step": 23308
    },
    {
      "epoch": 1.0696617869762746,
      "grad_norm": 0.4985136091709137,
      "learning_rate": 7.3615976885698745e-06,
      "loss": 0.367,
      "step": 23309
    },
    {
      "epoch": 1.069707677481529,
      "grad_norm": 0.4702286422252655,
      "learning_rate": 7.36138157039205e-06,
      "loss": 0.3426,
      "step": 23310
    },
    {
      "epoch": 1.0697535679867836,
      "grad_norm": 0.45572763681411743,
      "learning_rate": 7.361165446535848e-06,
      "loss": 0.3397,
      "step": 23311
    },
    {
      "epoch": 1.069799458492038,
      "grad_norm": 0.443445086479187,
      "learning_rate": 7.360949317001787e-06,
      "loss": 0.3118,
      "step": 23312
    },
    {
      "epoch": 1.0698453489972926,
      "grad_norm": 0.48508909344673157,
      "learning_rate": 7.360733181790391e-06,
      "loss": 0.4216,
      "step": 23313
    },
    {
      "epoch": 1.0698912395025468,
      "grad_norm": 0.4612192213535309,
      "learning_rate": 7.360517040902176e-06,
      "loss": 0.3493,
      "step": 23314
    },
    {
      "epoch": 1.0699371300078013,
      "grad_norm": 0.47229692339897156,
      "learning_rate": 7.36030089433766e-06,
      "loss": 0.3696,
      "step": 23315
    },
    {
      "epoch": 1.0699830205130558,
      "grad_norm": 0.4459943175315857,
      "learning_rate": 7.360084742097367e-06,
      "loss": 0.3022,
      "step": 23316
    },
    {
      "epoch": 1.0700289110183103,
      "grad_norm": 0.451636403799057,
      "learning_rate": 7.359868584181815e-06,
      "loss": 0.3013,
      "step": 23317
    },
    {
      "epoch": 1.0700748015235648,
      "grad_norm": 0.49632880091667175,
      "learning_rate": 7.3596524205915236e-06,
      "loss": 0.3557,
      "step": 23318
    },
    {
      "epoch": 1.0701206920288193,
      "grad_norm": 0.48757219314575195,
      "learning_rate": 7.359436251327014e-06,
      "loss": 0.408,
      "step": 23319
    },
    {
      "epoch": 1.0701665825340738,
      "grad_norm": 0.4749176800251007,
      "learning_rate": 7.3592200763888035e-06,
      "loss": 0.3829,
      "step": 23320
    },
    {
      "epoch": 1.070212473039328,
      "grad_norm": 0.4757743179798126,
      "learning_rate": 7.359003895777412e-06,
      "loss": 0.3764,
      "step": 23321
    },
    {
      "epoch": 1.0702583635445826,
      "grad_norm": 0.47482824325561523,
      "learning_rate": 7.358787709493364e-06,
      "loss": 0.352,
      "step": 23322
    },
    {
      "epoch": 1.070304254049837,
      "grad_norm": 0.4894445836544037,
      "learning_rate": 7.358571517537174e-06,
      "loss": 0.4084,
      "step": 23323
    },
    {
      "epoch": 1.0703501445550916,
      "grad_norm": 0.44325003027915955,
      "learning_rate": 7.358355319909365e-06,
      "loss": 0.3496,
      "step": 23324
    },
    {
      "epoch": 1.070396035060346,
      "grad_norm": 0.4937640428543091,
      "learning_rate": 7.358139116610454e-06,
      "loss": 0.3694,
      "step": 23325
    },
    {
      "epoch": 1.0704419255656004,
      "grad_norm": 0.4893208146095276,
      "learning_rate": 7.357922907640964e-06,
      "loss": 0.3918,
      "step": 23326
    },
    {
      "epoch": 1.0704878160708549,
      "grad_norm": 0.4667434096336365,
      "learning_rate": 7.357706693001414e-06,
      "loss": 0.342,
      "step": 23327
    },
    {
      "epoch": 1.0705337065761094,
      "grad_norm": 0.4757764935493469,
      "learning_rate": 7.357490472692323e-06,
      "loss": 0.3533,
      "step": 23328
    },
    {
      "epoch": 1.0705795970813639,
      "grad_norm": 0.4753258228302002,
      "learning_rate": 7.357274246714211e-06,
      "loss": 0.3798,
      "step": 23329
    },
    {
      "epoch": 1.0706254875866184,
      "grad_norm": 0.43887829780578613,
      "learning_rate": 7.357058015067599e-06,
      "loss": 0.3017,
      "step": 23330
    },
    {
      "epoch": 1.0706713780918728,
      "grad_norm": 0.49059784412384033,
      "learning_rate": 7.356841777753005e-06,
      "loss": 0.3649,
      "step": 23331
    },
    {
      "epoch": 1.0707172685971273,
      "grad_norm": 0.4883278012275696,
      "learning_rate": 7.356625534770951e-06,
      "loss": 0.3915,
      "step": 23332
    },
    {
      "epoch": 1.0707631591023816,
      "grad_norm": 0.46442916989326477,
      "learning_rate": 7.356409286121957e-06,
      "loss": 0.3725,
      "step": 23333
    },
    {
      "epoch": 1.0708090496076361,
      "grad_norm": 0.47397908568382263,
      "learning_rate": 7.356193031806542e-06,
      "loss": 0.3654,
      "step": 23334
    },
    {
      "epoch": 1.0708549401128906,
      "grad_norm": 0.490227073431015,
      "learning_rate": 7.355976771825227e-06,
      "loss": 0.334,
      "step": 23335
    },
    {
      "epoch": 1.0709008306181451,
      "grad_norm": 0.47585251927375793,
      "learning_rate": 7.355760506178532e-06,
      "loss": 0.3221,
      "step": 23336
    },
    {
      "epoch": 1.0709467211233996,
      "grad_norm": 0.45301082730293274,
      "learning_rate": 7.355544234866975e-06,
      "loss": 0.3593,
      "step": 23337
    },
    {
      "epoch": 1.0709926116286541,
      "grad_norm": 0.4856165945529938,
      "learning_rate": 7.355327957891077e-06,
      "loss": 0.374,
      "step": 23338
    },
    {
      "epoch": 1.0710385021339084,
      "grad_norm": 0.5137026906013489,
      "learning_rate": 7.355111675251361e-06,
      "loss": 0.4345,
      "step": 23339
    },
    {
      "epoch": 1.0710843926391629,
      "grad_norm": 0.4870001971721649,
      "learning_rate": 7.354895386948343e-06,
      "loss": 0.4087,
      "step": 23340
    },
    {
      "epoch": 1.0711302831444174,
      "grad_norm": 0.41522011160850525,
      "learning_rate": 7.354679092982545e-06,
      "loss": 0.2561,
      "step": 23341
    },
    {
      "epoch": 1.0711761736496719,
      "grad_norm": 0.4955638647079468,
      "learning_rate": 7.354462793354488e-06,
      "loss": 0.3879,
      "step": 23342
    },
    {
      "epoch": 1.0712220641549264,
      "grad_norm": 0.44853171706199646,
      "learning_rate": 7.3542464880646895e-06,
      "loss": 0.343,
      "step": 23343
    },
    {
      "epoch": 1.0712679546601809,
      "grad_norm": 0.4849448800086975,
      "learning_rate": 7.3540301771136726e-06,
      "loss": 0.3392,
      "step": 23344
    },
    {
      "epoch": 1.0713138451654354,
      "grad_norm": 0.49431079626083374,
      "learning_rate": 7.353813860501955e-06,
      "loss": 0.3212,
      "step": 23345
    },
    {
      "epoch": 1.0713597356706897,
      "grad_norm": 0.4501889646053314,
      "learning_rate": 7.353597538230059e-06,
      "loss": 0.3338,
      "step": 23346
    },
    {
      "epoch": 1.0714056261759441,
      "grad_norm": 0.46526914834976196,
      "learning_rate": 7.353381210298503e-06,
      "loss": 0.3298,
      "step": 23347
    },
    {
      "epoch": 1.0714515166811986,
      "grad_norm": 0.47608691453933716,
      "learning_rate": 7.353164876707808e-06,
      "loss": 0.3106,
      "step": 23348
    },
    {
      "epoch": 1.0714974071864531,
      "grad_norm": 0.46031177043914795,
      "learning_rate": 7.352948537458493e-06,
      "loss": 0.3112,
      "step": 23349
    },
    {
      "epoch": 1.0715432976917076,
      "grad_norm": 0.539963960647583,
      "learning_rate": 7.352732192551081e-06,
      "loss": 0.3646,
      "step": 23350
    },
    {
      "epoch": 1.0715891881969621,
      "grad_norm": 0.5081068873405457,
      "learning_rate": 7.352515841986091e-06,
      "loss": 0.3758,
      "step": 23351
    },
    {
      "epoch": 1.0716350787022164,
      "grad_norm": 0.4645172357559204,
      "learning_rate": 7.35229948576404e-06,
      "loss": 0.3316,
      "step": 23352
    },
    {
      "epoch": 1.071680969207471,
      "grad_norm": 0.5112524628639221,
      "learning_rate": 7.352083123885455e-06,
      "loss": 0.4177,
      "step": 23353
    },
    {
      "epoch": 1.0717268597127254,
      "grad_norm": 0.4954569339752197,
      "learning_rate": 7.3518667563508495e-06,
      "loss": 0.3525,
      "step": 23354
    },
    {
      "epoch": 1.07177275021798,
      "grad_norm": 0.4343908429145813,
      "learning_rate": 7.351650383160748e-06,
      "loss": 0.3266,
      "step": 23355
    },
    {
      "epoch": 1.0718186407232344,
      "grad_norm": 0.47046366333961487,
      "learning_rate": 7.351434004315669e-06,
      "loss": 0.3291,
      "step": 23356
    },
    {
      "epoch": 1.071864531228489,
      "grad_norm": 0.5239343047142029,
      "learning_rate": 7.3512176198161345e-06,
      "loss": 0.387,
      "step": 23357
    },
    {
      "epoch": 1.0719104217337434,
      "grad_norm": 0.48323768377304077,
      "learning_rate": 7.351001229662664e-06,
      "loss": 0.3725,
      "step": 23358
    },
    {
      "epoch": 1.0719563122389977,
      "grad_norm": 0.5232014656066895,
      "learning_rate": 7.3507848338557755e-06,
      "loss": 0.4212,
      "step": 23359
    },
    {
      "epoch": 1.0720022027442522,
      "grad_norm": 0.42359331250190735,
      "learning_rate": 7.350568432395992e-06,
      "loss": 0.3072,
      "step": 23360
    },
    {
      "epoch": 1.0720480932495067,
      "grad_norm": 0.4872820973396301,
      "learning_rate": 7.350352025283833e-06,
      "loss": 0.3852,
      "step": 23361
    },
    {
      "epoch": 1.0720939837547612,
      "grad_norm": 0.5258241295814514,
      "learning_rate": 7.350135612519822e-06,
      "loss": 0.4357,
      "step": 23362
    },
    {
      "epoch": 1.0721398742600157,
      "grad_norm": 0.5280765891075134,
      "learning_rate": 7.3499191941044736e-06,
      "loss": 0.3844,
      "step": 23363
    },
    {
      "epoch": 1.0721857647652702,
      "grad_norm": 0.4724825620651245,
      "learning_rate": 7.349702770038312e-06,
      "loss": 0.3237,
      "step": 23364
    },
    {
      "epoch": 1.0722316552705244,
      "grad_norm": 0.5144622921943665,
      "learning_rate": 7.349486340321856e-06,
      "loss": 0.4142,
      "step": 23365
    },
    {
      "epoch": 1.072277545775779,
      "grad_norm": 0.4855806231498718,
      "learning_rate": 7.349269904955627e-06,
      "loss": 0.3988,
      "step": 23366
    },
    {
      "epoch": 1.0723234362810334,
      "grad_norm": 0.4775330126285553,
      "learning_rate": 7.349053463940146e-06,
      "loss": 0.3406,
      "step": 23367
    },
    {
      "epoch": 1.072369326786288,
      "grad_norm": 0.4438757300376892,
      "learning_rate": 7.3488370172759334e-06,
      "loss": 0.2928,
      "step": 23368
    },
    {
      "epoch": 1.0724152172915424,
      "grad_norm": 0.44960495829582214,
      "learning_rate": 7.348620564963509e-06,
      "loss": 0.3377,
      "step": 23369
    },
    {
      "epoch": 1.072461107796797,
      "grad_norm": 0.5064049363136292,
      "learning_rate": 7.348404107003393e-06,
      "loss": 0.4473,
      "step": 23370
    },
    {
      "epoch": 1.0725069983020512,
      "grad_norm": 0.4997260570526123,
      "learning_rate": 7.348187643396105e-06,
      "loss": 0.4296,
      "step": 23371
    },
    {
      "epoch": 1.0725528888073057,
      "grad_norm": 0.5003837943077087,
      "learning_rate": 7.34797117414217e-06,
      "loss": 0.3817,
      "step": 23372
    },
    {
      "epoch": 1.0725987793125602,
      "grad_norm": 0.4416740834712982,
      "learning_rate": 7.3477546992421035e-06,
      "loss": 0.2908,
      "step": 23373
    },
    {
      "epoch": 1.0726446698178147,
      "grad_norm": 0.4635377824306488,
      "learning_rate": 7.347538218696427e-06,
      "loss": 0.3774,
      "step": 23374
    },
    {
      "epoch": 1.0726905603230692,
      "grad_norm": 0.507420539855957,
      "learning_rate": 7.347321732505663e-06,
      "loss": 0.4512,
      "step": 23375
    },
    {
      "epoch": 1.0727364508283237,
      "grad_norm": 0.46507561206817627,
      "learning_rate": 7.34710524067033e-06,
      "loss": 0.3801,
      "step": 23376
    },
    {
      "epoch": 1.072782341333578,
      "grad_norm": 0.4488803744316101,
      "learning_rate": 7.346888743190951e-06,
      "loss": 0.3262,
      "step": 23377
    },
    {
      "epoch": 1.0728282318388325,
      "grad_norm": 0.500082790851593,
      "learning_rate": 7.346672240068044e-06,
      "loss": 0.4205,
      "step": 23378
    },
    {
      "epoch": 1.072874122344087,
      "grad_norm": 0.4862593412399292,
      "learning_rate": 7.346455731302134e-06,
      "loss": 0.4223,
      "step": 23379
    },
    {
      "epoch": 1.0729200128493415,
      "grad_norm": 0.4844341576099396,
      "learning_rate": 7.346239216893735e-06,
      "loss": 0.409,
      "step": 23380
    },
    {
      "epoch": 1.072965903354596,
      "grad_norm": 0.46021410822868347,
      "learning_rate": 7.346022696843373e-06,
      "loss": 0.3262,
      "step": 23381
    },
    {
      "epoch": 1.0730117938598505,
      "grad_norm": 0.46575766801834106,
      "learning_rate": 7.345806171151565e-06,
      "loss": 0.3295,
      "step": 23382
    },
    {
      "epoch": 1.073057684365105,
      "grad_norm": 0.4837310016155243,
      "learning_rate": 7.345589639818835e-06,
      "loss": 0.3943,
      "step": 23383
    },
    {
      "epoch": 1.0731035748703592,
      "grad_norm": 0.45006975531578064,
      "learning_rate": 7.345373102845703e-06,
      "loss": 0.2927,
      "step": 23384
    },
    {
      "epoch": 1.0731494653756137,
      "grad_norm": 0.4840408265590668,
      "learning_rate": 7.345156560232687e-06,
      "loss": 0.3894,
      "step": 23385
    },
    {
      "epoch": 1.0731953558808682,
      "grad_norm": 0.4357624053955078,
      "learning_rate": 7.34494001198031e-06,
      "loss": 0.3197,
      "step": 23386
    },
    {
      "epoch": 1.0732412463861227,
      "grad_norm": 0.48195356130599976,
      "learning_rate": 7.344723458089094e-06,
      "loss": 0.3612,
      "step": 23387
    },
    {
      "epoch": 1.0732871368913772,
      "grad_norm": 0.48933321237564087,
      "learning_rate": 7.344506898559556e-06,
      "loss": 0.3396,
      "step": 23388
    },
    {
      "epoch": 1.0733330273966317,
      "grad_norm": 0.4602680802345276,
      "learning_rate": 7.34429033339222e-06,
      "loss": 0.3529,
      "step": 23389
    },
    {
      "epoch": 1.073378917901886,
      "grad_norm": 0.46789172291755676,
      "learning_rate": 7.344073762587605e-06,
      "loss": 0.3431,
      "step": 23390
    },
    {
      "epoch": 1.0734248084071405,
      "grad_norm": 0.4876914918422699,
      "learning_rate": 7.343857186146234e-06,
      "loss": 0.3635,
      "step": 23391
    },
    {
      "epoch": 1.073470698912395,
      "grad_norm": 0.5024553537368774,
      "learning_rate": 7.343640604068625e-06,
      "loss": 0.3824,
      "step": 23392
    },
    {
      "epoch": 1.0735165894176495,
      "grad_norm": 0.48838523030281067,
      "learning_rate": 7.3434240163552996e-06,
      "loss": 0.3726,
      "step": 23393
    },
    {
      "epoch": 1.073562479922904,
      "grad_norm": 0.45308491587638855,
      "learning_rate": 7.3432074230067794e-06,
      "loss": 0.3031,
      "step": 23394
    },
    {
      "epoch": 1.0736083704281585,
      "grad_norm": 0.5249916911125183,
      "learning_rate": 7.342990824023586e-06,
      "loss": 0.4444,
      "step": 23395
    },
    {
      "epoch": 1.073654260933413,
      "grad_norm": 0.5153510570526123,
      "learning_rate": 7.342774219406238e-06,
      "loss": 0.4147,
      "step": 23396
    },
    {
      "epoch": 1.0737001514386673,
      "grad_norm": 0.41731610894203186,
      "learning_rate": 7.3425576091552566e-06,
      "loss": 0.3057,
      "step": 23397
    },
    {
      "epoch": 1.0737460419439218,
      "grad_norm": 0.48786604404449463,
      "learning_rate": 7.342340993271164e-06,
      "loss": 0.3932,
      "step": 23398
    },
    {
      "epoch": 1.0737919324491763,
      "grad_norm": 0.49588921666145325,
      "learning_rate": 7.342124371754481e-06,
      "loss": 0.4131,
      "step": 23399
    },
    {
      "epoch": 1.0738378229544308,
      "grad_norm": 0.4938793480396271,
      "learning_rate": 7.341907744605729e-06,
      "loss": 0.3637,
      "step": 23400
    },
    {
      "epoch": 1.0738837134596853,
      "grad_norm": 0.4338071346282959,
      "learning_rate": 7.341691111825428e-06,
      "loss": 0.3108,
      "step": 23401
    },
    {
      "epoch": 1.0739296039649397,
      "grad_norm": 0.4732199013233185,
      "learning_rate": 7.341474473414097e-06,
      "loss": 0.3732,
      "step": 23402
    },
    {
      "epoch": 1.073975494470194,
      "grad_norm": 0.4547850489616394,
      "learning_rate": 7.341257829372259e-06,
      "loss": 0.3491,
      "step": 23403
    },
    {
      "epoch": 1.0740213849754485,
      "grad_norm": 0.435747891664505,
      "learning_rate": 7.341041179700436e-06,
      "loss": 0.3368,
      "step": 23404
    },
    {
      "epoch": 1.074067275480703,
      "grad_norm": 0.4430619180202484,
      "learning_rate": 7.340824524399148e-06,
      "loss": 0.3187,
      "step": 23405
    },
    {
      "epoch": 1.0741131659859575,
      "grad_norm": 0.4889989197254181,
      "learning_rate": 7.340607863468916e-06,
      "loss": 0.3176,
      "step": 23406
    },
    {
      "epoch": 1.074159056491212,
      "grad_norm": 0.7272564172744751,
      "learning_rate": 7.3403911969102605e-06,
      "loss": 0.4506,
      "step": 23407
    },
    {
      "epoch": 1.0742049469964665,
      "grad_norm": 0.44565701484680176,
      "learning_rate": 7.340174524723702e-06,
      "loss": 0.3339,
      "step": 23408
    },
    {
      "epoch": 1.0742508375017208,
      "grad_norm": 0.4596360921859741,
      "learning_rate": 7.339957846909763e-06,
      "loss": 0.3508,
      "step": 23409
    },
    {
      "epoch": 1.0742967280069753,
      "grad_norm": 0.47095972299575806,
      "learning_rate": 7.339741163468963e-06,
      "loss": 0.3483,
      "step": 23410
    },
    {
      "epoch": 1.0743426185122298,
      "grad_norm": 0.4944157898426056,
      "learning_rate": 7.339524474401824e-06,
      "loss": 0.4121,
      "step": 23411
    },
    {
      "epoch": 1.0743885090174843,
      "grad_norm": 0.47528985142707825,
      "learning_rate": 7.33930777970887e-06,
      "loss": 0.3897,
      "step": 23412
    },
    {
      "epoch": 1.0744343995227388,
      "grad_norm": 0.4616160988807678,
      "learning_rate": 7.339091079390615e-06,
      "loss": 0.3756,
      "step": 23413
    },
    {
      "epoch": 1.0744802900279933,
      "grad_norm": 0.487370103597641,
      "learning_rate": 7.338874373447587e-06,
      "loss": 0.3881,
      "step": 23414
    },
    {
      "epoch": 1.0745261805332476,
      "grad_norm": 0.4908181130886078,
      "learning_rate": 7.338657661880303e-06,
      "loss": 0.343,
      "step": 23415
    },
    {
      "epoch": 1.074572071038502,
      "grad_norm": 0.4783722758293152,
      "learning_rate": 7.338440944689286e-06,
      "loss": 0.3586,
      "step": 23416
    },
    {
      "epoch": 1.0746179615437566,
      "grad_norm": 0.4488353431224823,
      "learning_rate": 7.338224221875055e-06,
      "loss": 0.3149,
      "step": 23417
    },
    {
      "epoch": 1.074663852049011,
      "grad_norm": 0.5053941607475281,
      "learning_rate": 7.338007493438135e-06,
      "loss": 0.3496,
      "step": 23418
    },
    {
      "epoch": 1.0747097425542655,
      "grad_norm": 0.45824456214904785,
      "learning_rate": 7.337790759379043e-06,
      "loss": 0.3635,
      "step": 23419
    },
    {
      "epoch": 1.07475563305952,
      "grad_norm": 0.48228734731674194,
      "learning_rate": 7.337574019698304e-06,
      "loss": 0.3798,
      "step": 23420
    },
    {
      "epoch": 1.0748015235647745,
      "grad_norm": 0.41317692399024963,
      "learning_rate": 7.337357274396435e-06,
      "loss": 0.294,
      "step": 23421
    },
    {
      "epoch": 1.0748474140700288,
      "grad_norm": 0.43934279680252075,
      "learning_rate": 7.3371405234739625e-06,
      "loss": 0.3138,
      "step": 23422
    },
    {
      "epoch": 1.0748933045752833,
      "grad_norm": 0.47413378953933716,
      "learning_rate": 7.336923766931401e-06,
      "loss": 0.3472,
      "step": 23423
    },
    {
      "epoch": 1.0749391950805378,
      "grad_norm": 0.45065662264823914,
      "learning_rate": 7.336707004769278e-06,
      "loss": 0.364,
      "step": 23424
    },
    {
      "epoch": 1.0749850855857923,
      "grad_norm": 0.4456998109817505,
      "learning_rate": 7.336490236988111e-06,
      "loss": 0.3365,
      "step": 23425
    },
    {
      "epoch": 1.0750309760910468,
      "grad_norm": 0.46001824736595154,
      "learning_rate": 7.336273463588424e-06,
      "loss": 0.3756,
      "step": 23426
    },
    {
      "epoch": 1.0750768665963013,
      "grad_norm": 0.4887087941169739,
      "learning_rate": 7.336056684570735e-06,
      "loss": 0.4226,
      "step": 23427
    },
    {
      "epoch": 1.0751227571015556,
      "grad_norm": 0.44517433643341064,
      "learning_rate": 7.335839899935566e-06,
      "loss": 0.3725,
      "step": 23428
    },
    {
      "epoch": 1.07516864760681,
      "grad_norm": 0.4728129208087921,
      "learning_rate": 7.335623109683442e-06,
      "loss": 0.2938,
      "step": 23429
    },
    {
      "epoch": 1.0752145381120646,
      "grad_norm": 0.49400508403778076,
      "learning_rate": 7.335406313814879e-06,
      "loss": 0.3927,
      "step": 23430
    },
    {
      "epoch": 1.075260428617319,
      "grad_norm": 0.46701493859291077,
      "learning_rate": 7.335189512330402e-06,
      "loss": 0.3265,
      "step": 23431
    },
    {
      "epoch": 1.0753063191225736,
      "grad_norm": 0.4558035433292389,
      "learning_rate": 7.334972705230532e-06,
      "loss": 0.3212,
      "step": 23432
    },
    {
      "epoch": 1.075352209627828,
      "grad_norm": 0.5196183919906616,
      "learning_rate": 7.334755892515789e-06,
      "loss": 0.4139,
      "step": 23433
    },
    {
      "epoch": 1.0753981001330826,
      "grad_norm": 0.5299758315086365,
      "learning_rate": 7.334539074186694e-06,
      "loss": 0.4258,
      "step": 23434
    },
    {
      "epoch": 1.0754439906383368,
      "grad_norm": 0.4955865442752838,
      "learning_rate": 7.33432225024377e-06,
      "loss": 0.3623,
      "step": 23435
    },
    {
      "epoch": 1.0754898811435913,
      "grad_norm": 0.4471342861652374,
      "learning_rate": 7.334105420687539e-06,
      "loss": 0.3618,
      "step": 23436
    },
    {
      "epoch": 1.0755357716488458,
      "grad_norm": 0.4782399833202362,
      "learning_rate": 7.333888585518519e-06,
      "loss": 0.3779,
      "step": 23437
    },
    {
      "epoch": 1.0755816621541003,
      "grad_norm": 0.4569876194000244,
      "learning_rate": 7.333671744737235e-06,
      "loss": 0.3417,
      "step": 23438
    },
    {
      "epoch": 1.0756275526593548,
      "grad_norm": 0.5006552338600159,
      "learning_rate": 7.333454898344207e-06,
      "loss": 0.4064,
      "step": 23439
    },
    {
      "epoch": 1.0756734431646093,
      "grad_norm": 0.45773035287857056,
      "learning_rate": 7.3332380463399555e-06,
      "loss": 0.3109,
      "step": 23440
    },
    {
      "epoch": 1.0757193336698636,
      "grad_norm": 0.535605788230896,
      "learning_rate": 7.333021188725003e-06,
      "loss": 0.4409,
      "step": 23441
    },
    {
      "epoch": 1.075765224175118,
      "grad_norm": 0.4890125095844269,
      "learning_rate": 7.33280432549987e-06,
      "loss": 0.3332,
      "step": 23442
    },
    {
      "epoch": 1.0758111146803726,
      "grad_norm": 0.439662367105484,
      "learning_rate": 7.332587456665081e-06,
      "loss": 0.3415,
      "step": 23443
    },
    {
      "epoch": 1.075857005185627,
      "grad_norm": 0.4420439898967743,
      "learning_rate": 7.332370582221156e-06,
      "loss": 0.3129,
      "step": 23444
    },
    {
      "epoch": 1.0759028956908816,
      "grad_norm": 0.46956875920295715,
      "learning_rate": 7.332153702168613e-06,
      "loss": 0.3504,
      "step": 23445
    },
    {
      "epoch": 1.075948786196136,
      "grad_norm": 0.48991408944129944,
      "learning_rate": 7.331936816507977e-06,
      "loss": 0.3784,
      "step": 23446
    },
    {
      "epoch": 1.0759946767013906,
      "grad_norm": 0.4563630521297455,
      "learning_rate": 7.331719925239769e-06,
      "loss": 0.3503,
      "step": 23447
    },
    {
      "epoch": 1.0760405672066449,
      "grad_norm": 0.4389912486076355,
      "learning_rate": 7.331503028364512e-06,
      "loss": 0.267,
      "step": 23448
    },
    {
      "epoch": 1.0760864577118994,
      "grad_norm": 0.45491331815719604,
      "learning_rate": 7.331286125882726e-06,
      "loss": 0.3094,
      "step": 23449
    },
    {
      "epoch": 1.0761323482171539,
      "grad_norm": 0.4311503469944,
      "learning_rate": 7.331069217794931e-06,
      "loss": 0.2999,
      "step": 23450
    },
    {
      "epoch": 1.0761782387224084,
      "grad_norm": 0.4511682689189911,
      "learning_rate": 7.330852304101651e-06,
      "loss": 0.3648,
      "step": 23451
    },
    {
      "epoch": 1.0762241292276629,
      "grad_norm": 0.48167118430137634,
      "learning_rate": 7.3306353848034076e-06,
      "loss": 0.403,
      "step": 23452
    },
    {
      "epoch": 1.0762700197329174,
      "grad_norm": 0.500676155090332,
      "learning_rate": 7.33041845990072e-06,
      "loss": 0.4133,
      "step": 23453
    },
    {
      "epoch": 1.0763159102381716,
      "grad_norm": 0.4766101539134979,
      "learning_rate": 7.330201529394113e-06,
      "loss": 0.4234,
      "step": 23454
    },
    {
      "epoch": 1.0763618007434261,
      "grad_norm": 0.5137669444084167,
      "learning_rate": 7.329984593284106e-06,
      "loss": 0.4474,
      "step": 23455
    },
    {
      "epoch": 1.0764076912486806,
      "grad_norm": 0.5414932370185852,
      "learning_rate": 7.329767651571221e-06,
      "loss": 0.4451,
      "step": 23456
    },
    {
      "epoch": 1.0764535817539351,
      "grad_norm": 0.4387247562408447,
      "learning_rate": 7.329550704255981e-06,
      "loss": 0.3385,
      "step": 23457
    },
    {
      "epoch": 1.0764994722591896,
      "grad_norm": 0.48682117462158203,
      "learning_rate": 7.329333751338906e-06,
      "loss": 0.3513,
      "step": 23458
    },
    {
      "epoch": 1.0765453627644441,
      "grad_norm": 0.5059272050857544,
      "learning_rate": 7.329116792820519e-06,
      "loss": 0.3711,
      "step": 23459
    },
    {
      "epoch": 1.0765912532696984,
      "grad_norm": 0.514149010181427,
      "learning_rate": 7.3288998287013425e-06,
      "loss": 0.424,
      "step": 23460
    },
    {
      "epoch": 1.076637143774953,
      "grad_norm": 0.5497069954872131,
      "learning_rate": 7.328682858981895e-06,
      "loss": 0.4766,
      "step": 23461
    },
    {
      "epoch": 1.0766830342802074,
      "grad_norm": 0.4689859449863434,
      "learning_rate": 7.328465883662701e-06,
      "loss": 0.3535,
      "step": 23462
    },
    {
      "epoch": 1.076728924785462,
      "grad_norm": 0.46419641375541687,
      "learning_rate": 7.328248902744281e-06,
      "loss": 0.3472,
      "step": 23463
    },
    {
      "epoch": 1.0767748152907164,
      "grad_norm": 0.5596047043800354,
      "learning_rate": 7.3280319162271574e-06,
      "loss": 0.4311,
      "step": 23464
    },
    {
      "epoch": 1.076820705795971,
      "grad_norm": 0.48338043689727783,
      "learning_rate": 7.327814924111852e-06,
      "loss": 0.3364,
      "step": 23465
    },
    {
      "epoch": 1.0768665963012252,
      "grad_norm": 0.4785359501838684,
      "learning_rate": 7.327597926398886e-06,
      "loss": 0.3199,
      "step": 23466
    },
    {
      "epoch": 1.0769124868064797,
      "grad_norm": 0.5092898607254028,
      "learning_rate": 7.327380923088782e-06,
      "loss": 0.4056,
      "step": 23467
    },
    {
      "epoch": 1.0769583773117342,
      "grad_norm": 0.4456356167793274,
      "learning_rate": 7.3271639141820605e-06,
      "loss": 0.3173,
      "step": 23468
    },
    {
      "epoch": 1.0770042678169887,
      "grad_norm": 0.5203625559806824,
      "learning_rate": 7.326946899679246e-06,
      "loss": 0.4053,
      "step": 23469
    },
    {
      "epoch": 1.0770501583222432,
      "grad_norm": 0.41401246190071106,
      "learning_rate": 7.326729879580856e-06,
      "loss": 0.3092,
      "step": 23470
    },
    {
      "epoch": 1.0770960488274977,
      "grad_norm": 0.5333976149559021,
      "learning_rate": 7.326512853887418e-06,
      "loss": 0.4498,
      "step": 23471
    },
    {
      "epoch": 1.0771419393327522,
      "grad_norm": 0.5234068036079407,
      "learning_rate": 7.326295822599449e-06,
      "loss": 0.3757,
      "step": 23472
    },
    {
      "epoch": 1.0771878298380064,
      "grad_norm": 0.4695976972579956,
      "learning_rate": 7.326078785717473e-06,
      "loss": 0.3348,
      "step": 23473
    },
    {
      "epoch": 1.077233720343261,
      "grad_norm": 0.4300691783428192,
      "learning_rate": 7.325861743242013e-06,
      "loss": 0.2506,
      "step": 23474
    },
    {
      "epoch": 1.0772796108485154,
      "grad_norm": 0.4487209916114807,
      "learning_rate": 7.325644695173587e-06,
      "loss": 0.3454,
      "step": 23475
    },
    {
      "epoch": 1.07732550135377,
      "grad_norm": 0.491038978099823,
      "learning_rate": 7.325427641512721e-06,
      "loss": 0.4187,
      "step": 23476
    },
    {
      "epoch": 1.0773713918590244,
      "grad_norm": 0.41606199741363525,
      "learning_rate": 7.325210582259936e-06,
      "loss": 0.2914,
      "step": 23477
    },
    {
      "epoch": 1.077417282364279,
      "grad_norm": 0.4685066044330597,
      "learning_rate": 7.324993517415753e-06,
      "loss": 0.32,
      "step": 23478
    },
    {
      "epoch": 1.0774631728695332,
      "grad_norm": 0.4569603204727173,
      "learning_rate": 7.324776446980694e-06,
      "loss": 0.3398,
      "step": 23479
    },
    {
      "epoch": 1.0775090633747877,
      "grad_norm": 0.4687044024467468,
      "learning_rate": 7.324559370955282e-06,
      "loss": 0.3435,
      "step": 23480
    },
    {
      "epoch": 1.0775549538800422,
      "grad_norm": 0.4422742426395416,
      "learning_rate": 7.324342289340036e-06,
      "loss": 0.3446,
      "step": 23481
    },
    {
      "epoch": 1.0776008443852967,
      "grad_norm": 0.4986693859100342,
      "learning_rate": 7.324125202135483e-06,
      "loss": 0.3877,
      "step": 23482
    },
    {
      "epoch": 1.0776467348905512,
      "grad_norm": 0.4634506404399872,
      "learning_rate": 7.323908109342143e-06,
      "loss": 0.352,
      "step": 23483
    },
    {
      "epoch": 1.0776926253958057,
      "grad_norm": 0.4681623578071594,
      "learning_rate": 7.3236910109605354e-06,
      "loss": 0.3214,
      "step": 23484
    },
    {
      "epoch": 1.0777385159010602,
      "grad_norm": 0.5375463962554932,
      "learning_rate": 7.323473906991184e-06,
      "loss": 0.3676,
      "step": 23485
    },
    {
      "epoch": 1.0777844064063145,
      "grad_norm": 0.4827241599559784,
      "learning_rate": 7.323256797434612e-06,
      "loss": 0.3741,
      "step": 23486
    },
    {
      "epoch": 1.077830296911569,
      "grad_norm": 0.6304171681404114,
      "learning_rate": 7.323039682291342e-06,
      "loss": 0.386,
      "step": 23487
    },
    {
      "epoch": 1.0778761874168235,
      "grad_norm": 0.45733290910720825,
      "learning_rate": 7.322822561561893e-06,
      "loss": 0.3448,
      "step": 23488
    },
    {
      "epoch": 1.077922077922078,
      "grad_norm": 0.4650786221027374,
      "learning_rate": 7.32260543524679e-06,
      "loss": 0.3259,
      "step": 23489
    },
    {
      "epoch": 1.0779679684273324,
      "grad_norm": 0.48441633582115173,
      "learning_rate": 7.3223883033465525e-06,
      "loss": 0.3737,
      "step": 23490
    },
    {
      "epoch": 1.078013858932587,
      "grad_norm": 0.5224981307983398,
      "learning_rate": 7.3221711658617055e-06,
      "loss": 0.4214,
      "step": 23491
    },
    {
      "epoch": 1.0780597494378412,
      "grad_norm": 0.44984105229377747,
      "learning_rate": 7.3219540227927684e-06,
      "loss": 0.2963,
      "step": 23492
    },
    {
      "epoch": 1.0781056399430957,
      "grad_norm": 0.44990360736846924,
      "learning_rate": 7.321736874140266e-06,
      "loss": 0.3261,
      "step": 23493
    },
    {
      "epoch": 1.0781515304483502,
      "grad_norm": 0.5042288899421692,
      "learning_rate": 7.3215197199047195e-06,
      "loss": 0.3794,
      "step": 23494
    },
    {
      "epoch": 1.0781974209536047,
      "grad_norm": 0.47461003065109253,
      "learning_rate": 7.3213025600866494e-06,
      "loss": 0.3524,
      "step": 23495
    },
    {
      "epoch": 1.0782433114588592,
      "grad_norm": 0.4568961560726166,
      "learning_rate": 7.321085394686579e-06,
      "loss": 0.3355,
      "step": 23496
    },
    {
      "epoch": 1.0782892019641137,
      "grad_norm": 0.4486820101737976,
      "learning_rate": 7.320868223705032e-06,
      "loss": 0.3247,
      "step": 23497
    },
    {
      "epoch": 1.078335092469368,
      "grad_norm": 0.5103617906570435,
      "learning_rate": 7.3206510471425294e-06,
      "loss": 0.4005,
      "step": 23498
    },
    {
      "epoch": 1.0783809829746225,
      "grad_norm": 0.44462844729423523,
      "learning_rate": 7.320433864999592e-06,
      "loss": 0.3251,
      "step": 23499
    },
    {
      "epoch": 1.078426873479877,
      "grad_norm": 0.4477587342262268,
      "learning_rate": 7.320216677276746e-06,
      "loss": 0.3073,
      "step": 23500
    },
    {
      "epoch": 1.0784727639851315,
      "grad_norm": 0.5054042935371399,
      "learning_rate": 7.319999483974509e-06,
      "loss": 0.4346,
      "step": 23501
    },
    {
      "epoch": 1.078518654490386,
      "grad_norm": 0.472153902053833,
      "learning_rate": 7.3197822850934055e-06,
      "loss": 0.3799,
      "step": 23502
    },
    {
      "epoch": 1.0785645449956405,
      "grad_norm": 0.4857749342918396,
      "learning_rate": 7.319565080633959e-06,
      "loss": 0.4292,
      "step": 23503
    },
    {
      "epoch": 1.0786104355008947,
      "grad_norm": 0.4642336964607239,
      "learning_rate": 7.31934787059669e-06,
      "loss": 0.3438,
      "step": 23504
    },
    {
      "epoch": 1.0786563260061492,
      "grad_norm": 0.46415457129478455,
      "learning_rate": 7.319130654982121e-06,
      "loss": 0.3545,
      "step": 23505
    },
    {
      "epoch": 1.0787022165114037,
      "grad_norm": 0.49333053827285767,
      "learning_rate": 7.318913433790775e-06,
      "loss": 0.4277,
      "step": 23506
    },
    {
      "epoch": 1.0787481070166582,
      "grad_norm": 0.4162762463092804,
      "learning_rate": 7.318696207023175e-06,
      "loss": 0.2788,
      "step": 23507
    },
    {
      "epoch": 1.0787939975219127,
      "grad_norm": 0.431846559047699,
      "learning_rate": 7.318478974679841e-06,
      "loss": 0.287,
      "step": 23508
    },
    {
      "epoch": 1.0788398880271672,
      "grad_norm": 0.4641336500644684,
      "learning_rate": 7.318261736761299e-06,
      "loss": 0.3394,
      "step": 23509
    },
    {
      "epoch": 1.0788857785324217,
      "grad_norm": 0.46669909358024597,
      "learning_rate": 7.318044493268066e-06,
      "loss": 0.3956,
      "step": 23510
    },
    {
      "epoch": 1.078931669037676,
      "grad_norm": 0.473905473947525,
      "learning_rate": 7.31782724420067e-06,
      "loss": 0.3526,
      "step": 23511
    },
    {
      "epoch": 1.0789775595429305,
      "grad_norm": 0.4364517331123352,
      "learning_rate": 7.31760998955963e-06,
      "loss": 0.3281,
      "step": 23512
    },
    {
      "epoch": 1.079023450048185,
      "grad_norm": 0.4526953101158142,
      "learning_rate": 7.317392729345468e-06,
      "loss": 0.3407,
      "step": 23513
    },
    {
      "epoch": 1.0790693405534395,
      "grad_norm": 0.4453414976596832,
      "learning_rate": 7.31717546355871e-06,
      "loss": 0.272,
      "step": 23514
    },
    {
      "epoch": 1.079115231058694,
      "grad_norm": 0.4721308946609497,
      "learning_rate": 7.3169581921998764e-06,
      "loss": 0.3851,
      "step": 23515
    },
    {
      "epoch": 1.0791611215639485,
      "grad_norm": 0.4728175401687622,
      "learning_rate": 7.316740915269489e-06,
      "loss": 0.3426,
      "step": 23516
    },
    {
      "epoch": 1.0792070120692028,
      "grad_norm": 0.4878484606742859,
      "learning_rate": 7.316523632768071e-06,
      "loss": 0.3817,
      "step": 23517
    },
    {
      "epoch": 1.0792529025744573,
      "grad_norm": 0.45411980152130127,
      "learning_rate": 7.316306344696145e-06,
      "loss": 0.3714,
      "step": 23518
    },
    {
      "epoch": 1.0792987930797118,
      "grad_norm": 0.5303005576133728,
      "learning_rate": 7.3160890510542325e-06,
      "loss": 0.408,
      "step": 23519
    },
    {
      "epoch": 1.0793446835849663,
      "grad_norm": 0.46261075139045715,
      "learning_rate": 7.315871751842858e-06,
      "loss": 0.3733,
      "step": 23520
    },
    {
      "epoch": 1.0793905740902208,
      "grad_norm": 0.5062291026115417,
      "learning_rate": 7.315654447062541e-06,
      "loss": 0.3927,
      "step": 23521
    },
    {
      "epoch": 1.0794364645954753,
      "grad_norm": 0.4767845571041107,
      "learning_rate": 7.315437136713808e-06,
      "loss": 0.3823,
      "step": 23522
    },
    {
      "epoch": 1.0794823551007298,
      "grad_norm": 0.450222373008728,
      "learning_rate": 7.315219820797177e-06,
      "loss": 0.3543,
      "step": 23523
    },
    {
      "epoch": 1.079528245605984,
      "grad_norm": 0.46025407314300537,
      "learning_rate": 7.315002499313175e-06,
      "loss": 0.3548,
      "step": 23524
    },
    {
      "epoch": 1.0795741361112385,
      "grad_norm": 0.448490172624588,
      "learning_rate": 7.314785172262322e-06,
      "loss": 0.3702,
      "step": 23525
    },
    {
      "epoch": 1.079620026616493,
      "grad_norm": 0.48036590218544006,
      "learning_rate": 7.314567839645141e-06,
      "loss": 0.4345,
      "step": 23526
    },
    {
      "epoch": 1.0796659171217475,
      "grad_norm": 0.48849359154701233,
      "learning_rate": 7.314350501462155e-06,
      "loss": 0.3666,
      "step": 23527
    },
    {
      "epoch": 1.079711807627002,
      "grad_norm": 0.45617103576660156,
      "learning_rate": 7.314133157713888e-06,
      "loss": 0.3784,
      "step": 23528
    },
    {
      "epoch": 1.0797576981322565,
      "grad_norm": 0.474863737821579,
      "learning_rate": 7.3139158084008585e-06,
      "loss": 0.2838,
      "step": 23529
    },
    {
      "epoch": 1.0798035886375108,
      "grad_norm": 0.4458044171333313,
      "learning_rate": 7.3136984535235944e-06,
      "loss": 0.2934,
      "step": 23530
    },
    {
      "epoch": 1.0798494791427653,
      "grad_norm": 0.4668932259082794,
      "learning_rate": 7.313481093082614e-06,
      "loss": 0.3329,
      "step": 23531
    },
    {
      "epoch": 1.0798953696480198,
      "grad_norm": 0.450481653213501,
      "learning_rate": 7.313263727078441e-06,
      "loss": 0.3177,
      "step": 23532
    },
    {
      "epoch": 1.0799412601532743,
      "grad_norm": 0.5127546191215515,
      "learning_rate": 7.3130463555116e-06,
      "loss": 0.3904,
      "step": 23533
    },
    {
      "epoch": 1.0799871506585288,
      "grad_norm": 0.5017886757850647,
      "learning_rate": 7.312828978382614e-06,
      "loss": 0.4091,
      "step": 23534
    },
    {
      "epoch": 1.0800330411637833,
      "grad_norm": 0.4631175696849823,
      "learning_rate": 7.312611595692e-06,
      "loss": 0.3542,
      "step": 23535
    },
    {
      "epoch": 1.0800789316690378,
      "grad_norm": 0.517154335975647,
      "learning_rate": 7.31239420744029e-06,
      "loss": 0.4125,
      "step": 23536
    },
    {
      "epoch": 1.080124822174292,
      "grad_norm": 1.5738763809204102,
      "learning_rate": 7.312176813627999e-06,
      "loss": 0.2992,
      "step": 23537
    },
    {
      "epoch": 1.0801707126795466,
      "grad_norm": 0.47340983152389526,
      "learning_rate": 7.311959414255652e-06,
      "loss": 0.3911,
      "step": 23538
    },
    {
      "epoch": 1.080216603184801,
      "grad_norm": 0.503952145576477,
      "learning_rate": 7.311742009323774e-06,
      "loss": 0.3455,
      "step": 23539
    },
    {
      "epoch": 1.0802624936900556,
      "grad_norm": 0.478572815656662,
      "learning_rate": 7.311524598832886e-06,
      "loss": 0.4146,
      "step": 23540
    },
    {
      "epoch": 1.08030838419531,
      "grad_norm": 0.4561272859573364,
      "learning_rate": 7.31130718278351e-06,
      "loss": 0.3475,
      "step": 23541
    },
    {
      "epoch": 1.0803542747005646,
      "grad_norm": 0.5038200616836548,
      "learning_rate": 7.311089761176171e-06,
      "loss": 0.3937,
      "step": 23542
    },
    {
      "epoch": 1.0804001652058188,
      "grad_norm": 0.4809851050376892,
      "learning_rate": 7.310872334011389e-06,
      "loss": 0.3744,
      "step": 23543
    },
    {
      "epoch": 1.0804460557110733,
      "grad_norm": 0.4719792306423187,
      "learning_rate": 7.310654901289688e-06,
      "loss": 0.3198,
      "step": 23544
    },
    {
      "epoch": 1.0804919462163278,
      "grad_norm": 0.4645216464996338,
      "learning_rate": 7.310437463011593e-06,
      "loss": 0.3731,
      "step": 23545
    },
    {
      "epoch": 1.0805378367215823,
      "grad_norm": 0.4579533338546753,
      "learning_rate": 7.310220019177623e-06,
      "loss": 0.3148,
      "step": 23546
    },
    {
      "epoch": 1.0805837272268368,
      "grad_norm": 0.4964037239551544,
      "learning_rate": 7.310002569788304e-06,
      "loss": 0.39,
      "step": 23547
    },
    {
      "epoch": 1.0806296177320913,
      "grad_norm": 0.5044741630554199,
      "learning_rate": 7.309785114844159e-06,
      "loss": 0.4585,
      "step": 23548
    },
    {
      "epoch": 1.0806755082373456,
      "grad_norm": 0.5061672329902649,
      "learning_rate": 7.309567654345708e-06,
      "loss": 0.3963,
      "step": 23549
    },
    {
      "epoch": 1.0807213987426,
      "grad_norm": 0.474551796913147,
      "learning_rate": 7.309350188293476e-06,
      "loss": 0.4297,
      "step": 23550
    },
    {
      "epoch": 1.0807672892478546,
      "grad_norm": 0.47430896759033203,
      "learning_rate": 7.309132716687987e-06,
      "loss": 0.3718,
      "step": 23551
    },
    {
      "epoch": 1.080813179753109,
      "grad_norm": 0.4745233356952667,
      "learning_rate": 7.308915239529761e-06,
      "loss": 0.3382,
      "step": 23552
    },
    {
      "epoch": 1.0808590702583636,
      "grad_norm": 0.47661593556404114,
      "learning_rate": 7.308697756819323e-06,
      "loss": 0.4067,
      "step": 23553
    },
    {
      "epoch": 1.080904960763618,
      "grad_norm": 0.45257216691970825,
      "learning_rate": 7.3084802685571965e-06,
      "loss": 0.3406,
      "step": 23554
    },
    {
      "epoch": 1.0809508512688724,
      "grad_norm": 0.4612959921360016,
      "learning_rate": 7.308262774743903e-06,
      "loss": 0.3741,
      "step": 23555
    },
    {
      "epoch": 1.0809967417741269,
      "grad_norm": 0.4456912875175476,
      "learning_rate": 7.308045275379966e-06,
      "loss": 0.3317,
      "step": 23556
    },
    {
      "epoch": 1.0810426322793814,
      "grad_norm": 0.4832839369773865,
      "learning_rate": 7.307827770465907e-06,
      "loss": 0.4407,
      "step": 23557
    },
    {
      "epoch": 1.0810885227846359,
      "grad_norm": 0.4708153307437897,
      "learning_rate": 7.307610260002254e-06,
      "loss": 0.3425,
      "step": 23558
    },
    {
      "epoch": 1.0811344132898904,
      "grad_norm": 0.4880259931087494,
      "learning_rate": 7.307392743989524e-06,
      "loss": 0.4463,
      "step": 23559
    },
    {
      "epoch": 1.0811803037951448,
      "grad_norm": 0.4488755464553833,
      "learning_rate": 7.307175222428243e-06,
      "loss": 0.3593,
      "step": 23560
    },
    {
      "epoch": 1.0812261943003993,
      "grad_norm": 0.47707399725914,
      "learning_rate": 7.306957695318933e-06,
      "loss": 0.3803,
      "step": 23561
    },
    {
      "epoch": 1.0812720848056536,
      "grad_norm": 0.4680439829826355,
      "learning_rate": 7.30674016266212e-06,
      "loss": 0.3612,
      "step": 23562
    },
    {
      "epoch": 1.0813179753109081,
      "grad_norm": 0.46313175559043884,
      "learning_rate": 7.306522624458325e-06,
      "loss": 0.3499,
      "step": 23563
    },
    {
      "epoch": 1.0813638658161626,
      "grad_norm": 0.49966713786125183,
      "learning_rate": 7.306305080708069e-06,
      "loss": 0.3349,
      "step": 23564
    },
    {
      "epoch": 1.0814097563214171,
      "grad_norm": 0.47009003162384033,
      "learning_rate": 7.306087531411878e-06,
      "loss": 0.3159,
      "step": 23565
    },
    {
      "epoch": 1.0814556468266716,
      "grad_norm": 0.4476785659790039,
      "learning_rate": 7.305869976570274e-06,
      "loss": 0.3382,
      "step": 23566
    },
    {
      "epoch": 1.081501537331926,
      "grad_norm": 0.4336528778076172,
      "learning_rate": 7.30565241618378e-06,
      "loss": 0.2819,
      "step": 23567
    },
    {
      "epoch": 1.0815474278371804,
      "grad_norm": 0.49274057149887085,
      "learning_rate": 7.305434850252922e-06,
      "loss": 0.3566,
      "step": 23568
    },
    {
      "epoch": 1.0815933183424349,
      "grad_norm": 0.45834195613861084,
      "learning_rate": 7.305217278778219e-06,
      "loss": 0.3699,
      "step": 23569
    },
    {
      "epoch": 1.0816392088476894,
      "grad_norm": 0.4983423352241516,
      "learning_rate": 7.304999701760196e-06,
      "loss": 0.4491,
      "step": 23570
    },
    {
      "epoch": 1.0816850993529439,
      "grad_norm": 0.46036314964294434,
      "learning_rate": 7.304782119199377e-06,
      "loss": 0.3869,
      "step": 23571
    },
    {
      "epoch": 1.0817309898581984,
      "grad_norm": 0.4685608744621277,
      "learning_rate": 7.304564531096283e-06,
      "loss": 0.4066,
      "step": 23572
    },
    {
      "epoch": 1.0817768803634529,
      "grad_norm": 0.43725624680519104,
      "learning_rate": 7.304346937451441e-06,
      "loss": 0.3169,
      "step": 23573
    },
    {
      "epoch": 1.0818227708687074,
      "grad_norm": 0.49379169940948486,
      "learning_rate": 7.30412933826537e-06,
      "loss": 0.3927,
      "step": 23574
    },
    {
      "epoch": 1.0818686613739616,
      "grad_norm": 0.44622644782066345,
      "learning_rate": 7.303911733538596e-06,
      "loss": 0.3102,
      "step": 23575
    },
    {
      "epoch": 1.0819145518792161,
      "grad_norm": 0.4554527997970581,
      "learning_rate": 7.303694123271641e-06,
      "loss": 0.3332,
      "step": 23576
    },
    {
      "epoch": 1.0819604423844706,
      "grad_norm": 0.4792640507221222,
      "learning_rate": 7.303476507465028e-06,
      "loss": 0.406,
      "step": 23577
    },
    {
      "epoch": 1.0820063328897251,
      "grad_norm": 0.5107975602149963,
      "learning_rate": 7.303258886119281e-06,
      "loss": 0.3778,
      "step": 23578
    },
    {
      "epoch": 1.0820522233949796,
      "grad_norm": 0.48638445138931274,
      "learning_rate": 7.303041259234925e-06,
      "loss": 0.4464,
      "step": 23579
    },
    {
      "epoch": 1.0820981139002341,
      "grad_norm": 0.48854225873947144,
      "learning_rate": 7.3028236268124805e-06,
      "loss": 0.427,
      "step": 23580
    },
    {
      "epoch": 1.0821440044054884,
      "grad_norm": 0.4725310504436493,
      "learning_rate": 7.302605988852471e-06,
      "loss": 0.3896,
      "step": 23581
    },
    {
      "epoch": 1.082189894910743,
      "grad_norm": 0.44566768407821655,
      "learning_rate": 7.302388345355424e-06,
      "loss": 0.3479,
      "step": 23582
    },
    {
      "epoch": 1.0822357854159974,
      "grad_norm": 0.47477608919143677,
      "learning_rate": 7.302170696321856e-06,
      "loss": 0.3574,
      "step": 23583
    },
    {
      "epoch": 1.082281675921252,
      "grad_norm": 0.4395079016685486,
      "learning_rate": 7.3019530417522965e-06,
      "loss": 0.336,
      "step": 23584
    },
    {
      "epoch": 1.0823275664265064,
      "grad_norm": 0.5206356048583984,
      "learning_rate": 7.3017353816472655e-06,
      "loss": 0.4842,
      "step": 23585
    },
    {
      "epoch": 1.082373456931761,
      "grad_norm": 0.46876588463783264,
      "learning_rate": 7.301517716007287e-06,
      "loss": 0.3225,
      "step": 23586
    },
    {
      "epoch": 1.0824193474370152,
      "grad_norm": 0.48180779814720154,
      "learning_rate": 7.301300044832886e-06,
      "loss": 0.3751,
      "step": 23587
    },
    {
      "epoch": 1.0824652379422697,
      "grad_norm": 0.4468536078929901,
      "learning_rate": 7.301082368124583e-06,
      "loss": 0.3188,
      "step": 23588
    },
    {
      "epoch": 1.0825111284475242,
      "grad_norm": 0.4918350577354431,
      "learning_rate": 7.3008646858829035e-06,
      "loss": 0.367,
      "step": 23589
    },
    {
      "epoch": 1.0825570189527787,
      "grad_norm": 0.44985541701316833,
      "learning_rate": 7.3006469981083716e-06,
      "loss": 0.3198,
      "step": 23590
    },
    {
      "epoch": 1.0826029094580332,
      "grad_norm": 0.41953182220458984,
      "learning_rate": 7.300429304801511e-06,
      "loss": 0.2652,
      "step": 23591
    },
    {
      "epoch": 1.0826487999632877,
      "grad_norm": 0.4844287037849426,
      "learning_rate": 7.3002116059628415e-06,
      "loss": 0.3564,
      "step": 23592
    },
    {
      "epoch": 1.082694690468542,
      "grad_norm": 0.49515190720558167,
      "learning_rate": 7.29999390159289e-06,
      "loss": 0.3691,
      "step": 23593
    },
    {
      "epoch": 1.0827405809737964,
      "grad_norm": 0.44641926884651184,
      "learning_rate": 7.299776191692178e-06,
      "loss": 0.3513,
      "step": 23594
    },
    {
      "epoch": 1.082786471479051,
      "grad_norm": 0.5262213945388794,
      "learning_rate": 7.299558476261232e-06,
      "loss": 0.4242,
      "step": 23595
    },
    {
      "epoch": 1.0828323619843054,
      "grad_norm": 0.4499502182006836,
      "learning_rate": 7.2993407553005725e-06,
      "loss": 0.3392,
      "step": 23596
    },
    {
      "epoch": 1.08287825248956,
      "grad_norm": 0.47274500131607056,
      "learning_rate": 7.299123028810724e-06,
      "loss": 0.3556,
      "step": 23597
    },
    {
      "epoch": 1.0829241429948144,
      "grad_norm": 0.48076656460762024,
      "learning_rate": 7.29890529679221e-06,
      "loss": 0.3323,
      "step": 23598
    },
    {
      "epoch": 1.082970033500069,
      "grad_norm": 0.46518760919570923,
      "learning_rate": 7.298687559245556e-06,
      "loss": 0.3637,
      "step": 23599
    },
    {
      "epoch": 1.0830159240053232,
      "grad_norm": 0.43218323588371277,
      "learning_rate": 7.298469816171282e-06,
      "loss": 0.3253,
      "step": 23600
    },
    {
      "epoch": 1.0830618145105777,
      "grad_norm": 0.4853629469871521,
      "learning_rate": 7.298252067569915e-06,
      "loss": 0.3984,
      "step": 23601
    },
    {
      "epoch": 1.0831077050158322,
      "grad_norm": 0.4271202087402344,
      "learning_rate": 7.298034313441977e-06,
      "loss": 0.3062,
      "step": 23602
    },
    {
      "epoch": 1.0831535955210867,
      "grad_norm": 0.4597027003765106,
      "learning_rate": 7.2978165537879906e-06,
      "loss": 0.3511,
      "step": 23603
    },
    {
      "epoch": 1.0831994860263412,
      "grad_norm": 0.4460557997226715,
      "learning_rate": 7.297598788608482e-06,
      "loss": 0.3364,
      "step": 23604
    },
    {
      "epoch": 1.0832453765315957,
      "grad_norm": 0.6168267726898193,
      "learning_rate": 7.297381017903972e-06,
      "loss": 0.3702,
      "step": 23605
    },
    {
      "epoch": 1.08329126703685,
      "grad_norm": 0.44970476627349854,
      "learning_rate": 7.297163241674986e-06,
      "loss": 0.3154,
      "step": 23606
    },
    {
      "epoch": 1.0833371575421045,
      "grad_norm": 0.4559820294380188,
      "learning_rate": 7.296945459922049e-06,
      "loss": 0.3322,
      "step": 23607
    },
    {
      "epoch": 1.083383048047359,
      "grad_norm": 0.4744173288345337,
      "learning_rate": 7.29672767264568e-06,
      "loss": 0.3614,
      "step": 23608
    },
    {
      "epoch": 1.0834289385526135,
      "grad_norm": 0.4712355136871338,
      "learning_rate": 7.296509879846408e-06,
      "loss": 0.3527,
      "step": 23609
    },
    {
      "epoch": 1.083474829057868,
      "grad_norm": 0.465330570936203,
      "learning_rate": 7.296292081524755e-06,
      "loss": 0.3605,
      "step": 23610
    },
    {
      "epoch": 1.0835207195631225,
      "grad_norm": 0.4512239098548889,
      "learning_rate": 7.296074277681243e-06,
      "loss": 0.3715,
      "step": 23611
    },
    {
      "epoch": 1.083566610068377,
      "grad_norm": 0.4709427058696747,
      "learning_rate": 7.295856468316398e-06,
      "loss": 0.3847,
      "step": 23612
    },
    {
      "epoch": 1.0836125005736312,
      "grad_norm": 0.48046132922172546,
      "learning_rate": 7.295638653430743e-06,
      "loss": 0.3941,
      "step": 23613
    },
    {
      "epoch": 1.0836583910788857,
      "grad_norm": 0.4573792517185211,
      "learning_rate": 7.2954208330247996e-06,
      "loss": 0.3504,
      "step": 23614
    },
    {
      "epoch": 1.0837042815841402,
      "grad_norm": 0.43712276220321655,
      "learning_rate": 7.295203007099094e-06,
      "loss": 0.2852,
      "step": 23615
    },
    {
      "epoch": 1.0837501720893947,
      "grad_norm": 0.4590231478214264,
      "learning_rate": 7.294985175654151e-06,
      "loss": 0.3622,
      "step": 23616
    },
    {
      "epoch": 1.0837960625946492,
      "grad_norm": 0.4467950463294983,
      "learning_rate": 7.2947673386904935e-06,
      "loss": 0.3175,
      "step": 23617
    },
    {
      "epoch": 1.0838419530999037,
      "grad_norm": 0.4304211437702179,
      "learning_rate": 7.294549496208643e-06,
      "loss": 0.2923,
      "step": 23618
    },
    {
      "epoch": 1.083887843605158,
      "grad_norm": 0.5015869736671448,
      "learning_rate": 7.294331648209127e-06,
      "loss": 0.4172,
      "step": 23619
    },
    {
      "epoch": 1.0839337341104125,
      "grad_norm": 0.4822033941745758,
      "learning_rate": 7.2941137946924656e-06,
      "loss": 0.3966,
      "step": 23620
    },
    {
      "epoch": 1.083979624615667,
      "grad_norm": 0.49326276779174805,
      "learning_rate": 7.2938959356591855e-06,
      "loss": 0.3851,
      "step": 23621
    },
    {
      "epoch": 1.0840255151209215,
      "grad_norm": 0.4699822664260864,
      "learning_rate": 7.293678071109808e-06,
      "loss": 0.3709,
      "step": 23622
    },
    {
      "epoch": 1.084071405626176,
      "grad_norm": 0.4431056082248688,
      "learning_rate": 7.293460201044862e-06,
      "loss": 0.2958,
      "step": 23623
    },
    {
      "epoch": 1.0841172961314305,
      "grad_norm": 0.4783771336078644,
      "learning_rate": 7.293242325464865e-06,
      "loss": 0.3942,
      "step": 23624
    },
    {
      "epoch": 1.084163186636685,
      "grad_norm": 0.45762214064598083,
      "learning_rate": 7.293024444370346e-06,
      "loss": 0.3509,
      "step": 23625
    },
    {
      "epoch": 1.0842090771419393,
      "grad_norm": 0.5078889727592468,
      "learning_rate": 7.292806557761825e-06,
      "loss": 0.4282,
      "step": 23626
    },
    {
      "epoch": 1.0842549676471938,
      "grad_norm": 0.45577922463417053,
      "learning_rate": 7.292588665639829e-06,
      "loss": 0.3277,
      "step": 23627
    },
    {
      "epoch": 1.0843008581524483,
      "grad_norm": 0.45995786786079407,
      "learning_rate": 7.292370768004882e-06,
      "loss": 0.3244,
      "step": 23628
    },
    {
      "epoch": 1.0843467486577028,
      "grad_norm": 0.4844330847263336,
      "learning_rate": 7.292152864857504e-06,
      "loss": 0.3503,
      "step": 23629
    },
    {
      "epoch": 1.0843926391629573,
      "grad_norm": 0.47811073064804077,
      "learning_rate": 7.291934956198224e-06,
      "loss": 0.3549,
      "step": 23630
    },
    {
      "epoch": 1.0844385296682117,
      "grad_norm": 0.497405469417572,
      "learning_rate": 7.291717042027563e-06,
      "loss": 0.4296,
      "step": 23631
    },
    {
      "epoch": 1.084484420173466,
      "grad_norm": 0.4589798152446747,
      "learning_rate": 7.2914991223460455e-06,
      "loss": 0.3177,
      "step": 23632
    },
    {
      "epoch": 1.0845303106787205,
      "grad_norm": 0.4962727427482605,
      "learning_rate": 7.291281197154197e-06,
      "loss": 0.4368,
      "step": 23633
    },
    {
      "epoch": 1.084576201183975,
      "grad_norm": 0.46458354592323303,
      "learning_rate": 7.291063266452539e-06,
      "loss": 0.3327,
      "step": 23634
    },
    {
      "epoch": 1.0846220916892295,
      "grad_norm": 0.4918009638786316,
      "learning_rate": 7.290845330241597e-06,
      "loss": 0.3906,
      "step": 23635
    },
    {
      "epoch": 1.084667982194484,
      "grad_norm": 0.4659987986087799,
      "learning_rate": 7.290627388521896e-06,
      "loss": 0.3326,
      "step": 23636
    },
    {
      "epoch": 1.0847138726997385,
      "grad_norm": 0.4501245319843292,
      "learning_rate": 7.290409441293957e-06,
      "loss": 0.3463,
      "step": 23637
    },
    {
      "epoch": 1.0847597632049928,
      "grad_norm": 0.4552854001522064,
      "learning_rate": 7.290191488558309e-06,
      "loss": 0.379,
      "step": 23638
    },
    {
      "epoch": 1.0848056537102473,
      "grad_norm": 0.4765038788318634,
      "learning_rate": 7.289973530315473e-06,
      "loss": 0.4026,
      "step": 23639
    },
    {
      "epoch": 1.0848515442155018,
      "grad_norm": 0.4558756649494171,
      "learning_rate": 7.289755566565971e-06,
      "loss": 0.3766,
      "step": 23640
    },
    {
      "epoch": 1.0848974347207563,
      "grad_norm": 0.45331135392189026,
      "learning_rate": 7.289537597310332e-06,
      "loss": 0.3151,
      "step": 23641
    },
    {
      "epoch": 1.0849433252260108,
      "grad_norm": 0.45359963178634644,
      "learning_rate": 7.289319622549076e-06,
      "loss": 0.3141,
      "step": 23642
    },
    {
      "epoch": 1.0849892157312653,
      "grad_norm": 0.43193748593330383,
      "learning_rate": 7.289101642282729e-06,
      "loss": 0.312,
      "step": 23643
    },
    {
      "epoch": 1.0850351062365196,
      "grad_norm": 0.4798784554004669,
      "learning_rate": 7.288883656511816e-06,
      "loss": 0.4035,
      "step": 23644
    },
    {
      "epoch": 1.085080996741774,
      "grad_norm": 0.4726826846599579,
      "learning_rate": 7.2886656652368605e-06,
      "loss": 0.3881,
      "step": 23645
    },
    {
      "epoch": 1.0851268872470285,
      "grad_norm": 0.456220805644989,
      "learning_rate": 7.288447668458385e-06,
      "loss": 0.3845,
      "step": 23646
    },
    {
      "epoch": 1.085172777752283,
      "grad_norm": 0.4930247366428375,
      "learning_rate": 7.288229666176916e-06,
      "loss": 0.3831,
      "step": 23647
    },
    {
      "epoch": 1.0852186682575375,
      "grad_norm": 0.4811192750930786,
      "learning_rate": 7.288011658392975e-06,
      "loss": 0.3751,
      "step": 23648
    },
    {
      "epoch": 1.085264558762792,
      "grad_norm": 0.4941202998161316,
      "learning_rate": 7.28779364510709e-06,
      "loss": 0.4068,
      "step": 23649
    },
    {
      "epoch": 1.0853104492680465,
      "grad_norm": 0.46955496072769165,
      "learning_rate": 7.2875756263197835e-06,
      "loss": 0.3869,
      "step": 23650
    },
    {
      "epoch": 1.0853563397733008,
      "grad_norm": 0.5762729644775391,
      "learning_rate": 7.287357602031579e-06,
      "loss": 0.3655,
      "step": 23651
    },
    {
      "epoch": 1.0854022302785553,
      "grad_norm": 0.4939965307712555,
      "learning_rate": 7.287139572243e-06,
      "loss": 0.412,
      "step": 23652
    },
    {
      "epoch": 1.0854481207838098,
      "grad_norm": 0.4429567754268646,
      "learning_rate": 7.286921536954575e-06,
      "loss": 0.2944,
      "step": 23653
    },
    {
      "epoch": 1.0854940112890643,
      "grad_norm": 0.4839816689491272,
      "learning_rate": 7.286703496166822e-06,
      "loss": 0.429,
      "step": 23654
    },
    {
      "epoch": 1.0855399017943188,
      "grad_norm": 0.48793327808380127,
      "learning_rate": 7.286485449880273e-06,
      "loss": 0.3894,
      "step": 23655
    },
    {
      "epoch": 1.0855857922995733,
      "grad_norm": 0.46330705285072327,
      "learning_rate": 7.286267398095446e-06,
      "loss": 0.3664,
      "step": 23656
    },
    {
      "epoch": 1.0856316828048276,
      "grad_norm": 0.43843090534210205,
      "learning_rate": 7.286049340812866e-06,
      "loss": 0.3111,
      "step": 23657
    },
    {
      "epoch": 1.085677573310082,
      "grad_norm": 0.5005151033401489,
      "learning_rate": 7.285831278033062e-06,
      "loss": 0.429,
      "step": 23658
    },
    {
      "epoch": 1.0857234638153366,
      "grad_norm": 0.4948265850543976,
      "learning_rate": 7.285613209756552e-06,
      "loss": 0.3821,
      "step": 23659
    },
    {
      "epoch": 1.085769354320591,
      "grad_norm": 0.46440795063972473,
      "learning_rate": 7.285395135983865e-06,
      "loss": 0.3672,
      "step": 23660
    },
    {
      "epoch": 1.0858152448258456,
      "grad_norm": 0.44109541177749634,
      "learning_rate": 7.285177056715526e-06,
      "loss": 0.3195,
      "step": 23661
    },
    {
      "epoch": 1.0858611353311,
      "grad_norm": 0.4612303376197815,
      "learning_rate": 7.284958971952054e-06,
      "loss": 0.3183,
      "step": 23662
    },
    {
      "epoch": 1.0859070258363546,
      "grad_norm": 0.457851380109787,
      "learning_rate": 7.284740881693978e-06,
      "loss": 0.3376,
      "step": 23663
    },
    {
      "epoch": 1.0859529163416088,
      "grad_norm": 0.46595296263694763,
      "learning_rate": 7.284522785941822e-06,
      "loss": 0.3179,
      "step": 23664
    },
    {
      "epoch": 1.0859988068468633,
      "grad_norm": 0.4398355484008789,
      "learning_rate": 7.284304684696108e-06,
      "loss": 0.3219,
      "step": 23665
    },
    {
      "epoch": 1.0860446973521178,
      "grad_norm": 0.47172078490257263,
      "learning_rate": 7.284086577957364e-06,
      "loss": 0.3497,
      "step": 23666
    },
    {
      "epoch": 1.0860905878573723,
      "grad_norm": 0.4895589351654053,
      "learning_rate": 7.283868465726114e-06,
      "loss": 0.4177,
      "step": 23667
    },
    {
      "epoch": 1.0861364783626268,
      "grad_norm": 0.4705275893211365,
      "learning_rate": 7.2836503480028775e-06,
      "loss": 0.3734,
      "step": 23668
    },
    {
      "epoch": 1.0861823688678813,
      "grad_norm": 0.44050320982933044,
      "learning_rate": 7.283432224788185e-06,
      "loss": 0.2919,
      "step": 23669
    },
    {
      "epoch": 1.0862282593731356,
      "grad_norm": 0.475134938955307,
      "learning_rate": 7.283214096082558e-06,
      "loss": 0.3604,
      "step": 23670
    },
    {
      "epoch": 1.08627414987839,
      "grad_norm": 0.467201828956604,
      "learning_rate": 7.282995961886521e-06,
      "loss": 0.3648,
      "step": 23671
    },
    {
      "epoch": 1.0863200403836446,
      "grad_norm": 0.5183555483818054,
      "learning_rate": 7.282777822200602e-06,
      "loss": 0.4177,
      "step": 23672
    },
    {
      "epoch": 1.086365930888899,
      "grad_norm": 0.5371918678283691,
      "learning_rate": 7.28255967702532e-06,
      "loss": 0.3827,
      "step": 23673
    },
    {
      "epoch": 1.0864118213941536,
      "grad_norm": 0.4957188367843628,
      "learning_rate": 7.282341526361203e-06,
      "loss": 0.4142,
      "step": 23674
    },
    {
      "epoch": 1.086457711899408,
      "grad_norm": 0.428519606590271,
      "learning_rate": 7.282123370208776e-06,
      "loss": 0.2829,
      "step": 23675
    },
    {
      "epoch": 1.0865036024046624,
      "grad_norm": 0.498566597700119,
      "learning_rate": 7.281905208568561e-06,
      "loss": 0.418,
      "step": 23676
    },
    {
      "epoch": 1.0865494929099169,
      "grad_norm": 0.46495139598846436,
      "learning_rate": 7.281687041441085e-06,
      "loss": 0.345,
      "step": 23677
    },
    {
      "epoch": 1.0865953834151714,
      "grad_norm": 0.4814172387123108,
      "learning_rate": 7.281468868826872e-06,
      "loss": 0.4107,
      "step": 23678
    },
    {
      "epoch": 1.0866412739204259,
      "grad_norm": 0.443947434425354,
      "learning_rate": 7.281250690726446e-06,
      "loss": 0.3053,
      "step": 23679
    },
    {
      "epoch": 1.0866871644256804,
      "grad_norm": 0.45778733491897583,
      "learning_rate": 7.281032507140331e-06,
      "loss": 0.3658,
      "step": 23680
    },
    {
      "epoch": 1.0867330549309349,
      "grad_norm": 0.42524299025535583,
      "learning_rate": 7.280814318069054e-06,
      "loss": 0.2841,
      "step": 23681
    },
    {
      "epoch": 1.0867789454361891,
      "grad_norm": 0.46094146370887756,
      "learning_rate": 7.280596123513138e-06,
      "loss": 0.3757,
      "step": 23682
    },
    {
      "epoch": 1.0868248359414436,
      "grad_norm": 0.4793056547641754,
      "learning_rate": 7.280377923473109e-06,
      "loss": 0.3803,
      "step": 23683
    },
    {
      "epoch": 1.0868707264466981,
      "grad_norm": 0.45266926288604736,
      "learning_rate": 7.2801597179494906e-06,
      "loss": 0.3299,
      "step": 23684
    },
    {
      "epoch": 1.0869166169519526,
      "grad_norm": 0.44952377676963806,
      "learning_rate": 7.279941506942807e-06,
      "loss": 0.3225,
      "step": 23685
    },
    {
      "epoch": 1.0869625074572071,
      "grad_norm": 0.44711464643478394,
      "learning_rate": 7.279723290453584e-06,
      "loss": 0.3025,
      "step": 23686
    },
    {
      "epoch": 1.0870083979624616,
      "grad_norm": 0.5084816813468933,
      "learning_rate": 7.279505068482345e-06,
      "loss": 0.3985,
      "step": 23687
    },
    {
      "epoch": 1.0870542884677161,
      "grad_norm": 0.48148325085639954,
      "learning_rate": 7.279286841029618e-06,
      "loss": 0.367,
      "step": 23688
    },
    {
      "epoch": 1.0871001789729704,
      "grad_norm": 0.4846136271953583,
      "learning_rate": 7.279068608095925e-06,
      "loss": 0.3465,
      "step": 23689
    },
    {
      "epoch": 1.087146069478225,
      "grad_norm": 0.47388023138046265,
      "learning_rate": 7.278850369681789e-06,
      "loss": 0.3586,
      "step": 23690
    },
    {
      "epoch": 1.0871919599834794,
      "grad_norm": 0.48141899704933167,
      "learning_rate": 7.278632125787739e-06,
      "loss": 0.3425,
      "step": 23691
    },
    {
      "epoch": 1.087237850488734,
      "grad_norm": 0.43066883087158203,
      "learning_rate": 7.278413876414298e-06,
      "loss": 0.3051,
      "step": 23692
    },
    {
      "epoch": 1.0872837409939884,
      "grad_norm": 0.5115679502487183,
      "learning_rate": 7.278195621561989e-06,
      "loss": 0.4489,
      "step": 23693
    },
    {
      "epoch": 1.0873296314992429,
      "grad_norm": 0.48040705919265747,
      "learning_rate": 7.277977361231341e-06,
      "loss": 0.418,
      "step": 23694
    },
    {
      "epoch": 1.0873755220044972,
      "grad_norm": 0.46274641156196594,
      "learning_rate": 7.277759095422875e-06,
      "loss": 0.3186,
      "step": 23695
    },
    {
      "epoch": 1.0874214125097517,
      "grad_norm": 0.4759560227394104,
      "learning_rate": 7.277540824137117e-06,
      "loss": 0.4094,
      "step": 23696
    },
    {
      "epoch": 1.0874673030150062,
      "grad_norm": 0.4669491946697235,
      "learning_rate": 7.277322547374592e-06,
      "loss": 0.3088,
      "step": 23697
    },
    {
      "epoch": 1.0875131935202607,
      "grad_norm": 0.4498862028121948,
      "learning_rate": 7.277104265135826e-06,
      "loss": 0.307,
      "step": 23698
    },
    {
      "epoch": 1.0875590840255152,
      "grad_norm": 0.4459482729434967,
      "learning_rate": 7.276885977421343e-06,
      "loss": 0.3271,
      "step": 23699
    },
    {
      "epoch": 1.0876049745307697,
      "grad_norm": 0.4891681969165802,
      "learning_rate": 7.276667684231666e-06,
      "loss": 0.3656,
      "step": 23700
    },
    {
      "epoch": 1.0876508650360242,
      "grad_norm": 0.487946093082428,
      "learning_rate": 7.276449385567323e-06,
      "loss": 0.3566,
      "step": 23701
    },
    {
      "epoch": 1.0876967555412784,
      "grad_norm": 0.45046499371528625,
      "learning_rate": 7.276231081428839e-06,
      "loss": 0.3816,
      "step": 23702
    },
    {
      "epoch": 1.087742646046533,
      "grad_norm": 0.5621416568756104,
      "learning_rate": 7.276012771816736e-06,
      "loss": 0.3173,
      "step": 23703
    },
    {
      "epoch": 1.0877885365517874,
      "grad_norm": 0.48943835496902466,
      "learning_rate": 7.27579445673154e-06,
      "loss": 0.3753,
      "step": 23704
    },
    {
      "epoch": 1.087834427057042,
      "grad_norm": 0.4863723814487457,
      "learning_rate": 7.275576136173779e-06,
      "loss": 0.3636,
      "step": 23705
    },
    {
      "epoch": 1.0878803175622964,
      "grad_norm": 0.4484808146953583,
      "learning_rate": 7.2753578101439735e-06,
      "loss": 0.3311,
      "step": 23706
    },
    {
      "epoch": 1.087926208067551,
      "grad_norm": 0.4434199929237366,
      "learning_rate": 7.275139478642651e-06,
      "loss": 0.3326,
      "step": 23707
    },
    {
      "epoch": 1.0879720985728052,
      "grad_norm": 0.45532912015914917,
      "learning_rate": 7.274921141670336e-06,
      "loss": 0.3286,
      "step": 23708
    },
    {
      "epoch": 1.0880179890780597,
      "grad_norm": 0.46044889092445374,
      "learning_rate": 7.274702799227554e-06,
      "loss": 0.3478,
      "step": 23709
    },
    {
      "epoch": 1.0880638795833142,
      "grad_norm": 0.4920402467250824,
      "learning_rate": 7.27448445131483e-06,
      "loss": 0.3991,
      "step": 23710
    },
    {
      "epoch": 1.0881097700885687,
      "grad_norm": 0.534299373626709,
      "learning_rate": 7.2742660979326884e-06,
      "loss": 0.4813,
      "step": 23711
    },
    {
      "epoch": 1.0881556605938232,
      "grad_norm": 0.4399348199367523,
      "learning_rate": 7.2740477390816555e-06,
      "loss": 0.314,
      "step": 23712
    },
    {
      "epoch": 1.0882015510990777,
      "grad_norm": 0.4977019727230072,
      "learning_rate": 7.273829374762253e-06,
      "loss": 0.4242,
      "step": 23713
    },
    {
      "epoch": 1.0882474416043322,
      "grad_norm": 0.48229271173477173,
      "learning_rate": 7.273611004975011e-06,
      "loss": 0.3987,
      "step": 23714
    },
    {
      "epoch": 1.0882933321095865,
      "grad_norm": 0.49072936177253723,
      "learning_rate": 7.2733926297204514e-06,
      "loss": 0.4131,
      "step": 23715
    },
    {
      "epoch": 1.088339222614841,
      "grad_norm": 0.4612120985984802,
      "learning_rate": 7.2731742489991e-06,
      "loss": 0.3648,
      "step": 23716
    },
    {
      "epoch": 1.0883851131200954,
      "grad_norm": 0.4244402050971985,
      "learning_rate": 7.272955862811482e-06,
      "loss": 0.2625,
      "step": 23717
    },
    {
      "epoch": 1.08843100362535,
      "grad_norm": 0.49267107248306274,
      "learning_rate": 7.272737471158122e-06,
      "loss": 0.3844,
      "step": 23718
    },
    {
      "epoch": 1.0884768941306044,
      "grad_norm": 0.4852755069732666,
      "learning_rate": 7.272519074039546e-06,
      "loss": 0.3436,
      "step": 23719
    },
    {
      "epoch": 1.088522784635859,
      "grad_norm": 0.5186461806297302,
      "learning_rate": 7.272300671456279e-06,
      "loss": 0.4182,
      "step": 23720
    },
    {
      "epoch": 1.0885686751411132,
      "grad_norm": 0.4445938766002655,
      "learning_rate": 7.272082263408846e-06,
      "loss": 0.3078,
      "step": 23721
    },
    {
      "epoch": 1.0886145656463677,
      "grad_norm": 0.5094393491744995,
      "learning_rate": 7.271863849897772e-06,
      "loss": 0.4325,
      "step": 23722
    },
    {
      "epoch": 1.0886604561516222,
      "grad_norm": 0.47436264157295227,
      "learning_rate": 7.271645430923584e-06,
      "loss": 0.3225,
      "step": 23723
    },
    {
      "epoch": 1.0887063466568767,
      "grad_norm": 0.4687618315219879,
      "learning_rate": 7.271427006486803e-06,
      "loss": 0.3578,
      "step": 23724
    },
    {
      "epoch": 1.0887522371621312,
      "grad_norm": 0.4699172377586365,
      "learning_rate": 7.271208576587958e-06,
      "loss": 0.3092,
      "step": 23725
    },
    {
      "epoch": 1.0887981276673857,
      "grad_norm": 0.4778865873813629,
      "learning_rate": 7.270990141227574e-06,
      "loss": 0.3962,
      "step": 23726
    },
    {
      "epoch": 1.08884401817264,
      "grad_norm": 0.4780392348766327,
      "learning_rate": 7.270771700406175e-06,
      "loss": 0.3446,
      "step": 23727
    },
    {
      "epoch": 1.0888899086778945,
      "grad_norm": 0.4364818334579468,
      "learning_rate": 7.270553254124286e-06,
      "loss": 0.3145,
      "step": 23728
    },
    {
      "epoch": 1.088935799183149,
      "grad_norm": 0.44355857372283936,
      "learning_rate": 7.270334802382435e-06,
      "loss": 0.2938,
      "step": 23729
    },
    {
      "epoch": 1.0889816896884035,
      "grad_norm": 0.4932951033115387,
      "learning_rate": 7.270116345181143e-06,
      "loss": 0.4324,
      "step": 23730
    },
    {
      "epoch": 1.089027580193658,
      "grad_norm": 0.5153430104255676,
      "learning_rate": 7.26989788252094e-06,
      "loss": 0.4017,
      "step": 23731
    },
    {
      "epoch": 1.0890734706989125,
      "grad_norm": 0.5497186183929443,
      "learning_rate": 7.2696794144023475e-06,
      "loss": 0.4318,
      "step": 23732
    },
    {
      "epoch": 1.0891193612041667,
      "grad_norm": 0.4481191337108612,
      "learning_rate": 7.269460940825892e-06,
      "loss": 0.3195,
      "step": 23733
    },
    {
      "epoch": 1.0891652517094212,
      "grad_norm": 0.47311103343963623,
      "learning_rate": 7.269242461792099e-06,
      "loss": 0.3995,
      "step": 23734
    },
    {
      "epoch": 1.0892111422146757,
      "grad_norm": 0.5088847875595093,
      "learning_rate": 7.269023977301495e-06,
      "loss": 0.4427,
      "step": 23735
    },
    {
      "epoch": 1.0892570327199302,
      "grad_norm": 0.4518539011478424,
      "learning_rate": 7.268805487354604e-06,
      "loss": 0.3399,
      "step": 23736
    },
    {
      "epoch": 1.0893029232251847,
      "grad_norm": 0.4472644627094269,
      "learning_rate": 7.2685869919519525e-06,
      "loss": 0.3013,
      "step": 23737
    },
    {
      "epoch": 1.0893488137304392,
      "grad_norm": 0.5100380778312683,
      "learning_rate": 7.2683684910940654e-06,
      "loss": 0.3898,
      "step": 23738
    },
    {
      "epoch": 1.0893947042356937,
      "grad_norm": 0.5007476806640625,
      "learning_rate": 7.2681499847814675e-06,
      "loss": 0.3935,
      "step": 23739
    },
    {
      "epoch": 1.089440594740948,
      "grad_norm": 0.4192773103713989,
      "learning_rate": 7.267931473014685e-06,
      "loss": 0.2906,
      "step": 23740
    },
    {
      "epoch": 1.0894864852462025,
      "grad_norm": 0.48261842131614685,
      "learning_rate": 7.267712955794242e-06,
      "loss": 0.3887,
      "step": 23741
    },
    {
      "epoch": 1.089532375751457,
      "grad_norm": 0.42712298035621643,
      "learning_rate": 7.2674944331206665e-06,
      "loss": 0.296,
      "step": 23742
    },
    {
      "epoch": 1.0895782662567115,
      "grad_norm": 0.4437287151813507,
      "learning_rate": 7.267275904994482e-06,
      "loss": 0.3352,
      "step": 23743
    },
    {
      "epoch": 1.089624156761966,
      "grad_norm": 0.4409063458442688,
      "learning_rate": 7.267057371416214e-06,
      "loss": 0.33,
      "step": 23744
    },
    {
      "epoch": 1.0896700472672205,
      "grad_norm": 0.48689156770706177,
      "learning_rate": 7.266838832386388e-06,
      "loss": 0.4358,
      "step": 23745
    },
    {
      "epoch": 1.0897159377724748,
      "grad_norm": 0.4629826545715332,
      "learning_rate": 7.266620287905533e-06,
      "loss": 0.3295,
      "step": 23746
    },
    {
      "epoch": 1.0897618282777293,
      "grad_norm": 0.5226610898971558,
      "learning_rate": 7.266401737974168e-06,
      "loss": 0.4519,
      "step": 23747
    },
    {
      "epoch": 1.0898077187829838,
      "grad_norm": 0.5054270029067993,
      "learning_rate": 7.266183182592824e-06,
      "loss": 0.3746,
      "step": 23748
    },
    {
      "epoch": 1.0898536092882383,
      "grad_norm": 0.43981775641441345,
      "learning_rate": 7.265964621762025e-06,
      "loss": 0.3076,
      "step": 23749
    },
    {
      "epoch": 1.0898994997934928,
      "grad_norm": 0.4779706597328186,
      "learning_rate": 7.265746055482294e-06,
      "loss": 0.3563,
      "step": 23750
    },
    {
      "epoch": 1.0899453902987473,
      "grad_norm": 0.4893731474876404,
      "learning_rate": 7.2655274837541626e-06,
      "loss": 0.3602,
      "step": 23751
    },
    {
      "epoch": 1.0899912808040018,
      "grad_norm": 0.44082602858543396,
      "learning_rate": 7.265308906578149e-06,
      "loss": 0.3323,
      "step": 23752
    },
    {
      "epoch": 1.090037171309256,
      "grad_norm": 0.47685953974723816,
      "learning_rate": 7.265090323954785e-06,
      "loss": 0.3449,
      "step": 23753
    },
    {
      "epoch": 1.0900830618145105,
      "grad_norm": 0.4497973918914795,
      "learning_rate": 7.264871735884592e-06,
      "loss": 0.3354,
      "step": 23754
    },
    {
      "epoch": 1.090128952319765,
      "grad_norm": 0.45801833271980286,
      "learning_rate": 7.264653142368097e-06,
      "loss": 0.3396,
      "step": 23755
    },
    {
      "epoch": 1.0901748428250195,
      "grad_norm": 0.47226232290267944,
      "learning_rate": 7.264434543405827e-06,
      "loss": 0.3227,
      "step": 23756
    },
    {
      "epoch": 1.090220733330274,
      "grad_norm": 0.48458725214004517,
      "learning_rate": 7.264215938998307e-06,
      "loss": 0.3654,
      "step": 23757
    },
    {
      "epoch": 1.0902666238355285,
      "grad_norm": 0.45109155774116516,
      "learning_rate": 7.263997329146061e-06,
      "loss": 0.3319,
      "step": 23758
    },
    {
      "epoch": 1.0903125143407828,
      "grad_norm": 0.45144978165626526,
      "learning_rate": 7.263778713849616e-06,
      "loss": 0.3547,
      "step": 23759
    },
    {
      "epoch": 1.0903584048460373,
      "grad_norm": 0.4296868145465851,
      "learning_rate": 7.2635600931094995e-06,
      "loss": 0.3279,
      "step": 23760
    },
    {
      "epoch": 1.0904042953512918,
      "grad_norm": 0.423123836517334,
      "learning_rate": 7.263341466926234e-06,
      "loss": 0.2995,
      "step": 23761
    },
    {
      "epoch": 1.0904501858565463,
      "grad_norm": 0.4677504301071167,
      "learning_rate": 7.263122835300346e-06,
      "loss": 0.3272,
      "step": 23762
    },
    {
      "epoch": 1.0904960763618008,
      "grad_norm": 0.4793601930141449,
      "learning_rate": 7.262904198232364e-06,
      "loss": 0.3691,
      "step": 23763
    },
    {
      "epoch": 1.0905419668670553,
      "grad_norm": 0.47027599811553955,
      "learning_rate": 7.262685555722809e-06,
      "loss": 0.3846,
      "step": 23764
    },
    {
      "epoch": 1.0905878573723096,
      "grad_norm": 0.43046247959136963,
      "learning_rate": 7.26246690777221e-06,
      "loss": 0.2846,
      "step": 23765
    },
    {
      "epoch": 1.090633747877564,
      "grad_norm": 0.47335097193717957,
      "learning_rate": 7.262248254381093e-06,
      "loss": 0.3671,
      "step": 23766
    },
    {
      "epoch": 1.0906796383828186,
      "grad_norm": 0.4892774224281311,
      "learning_rate": 7.262029595549982e-06,
      "loss": 0.3849,
      "step": 23767
    },
    {
      "epoch": 1.090725528888073,
      "grad_norm": 0.42006880044937134,
      "learning_rate": 7.261810931279405e-06,
      "loss": 0.2898,
      "step": 23768
    },
    {
      "epoch": 1.0907714193933276,
      "grad_norm": 0.4683209955692291,
      "learning_rate": 7.261592261569885e-06,
      "loss": 0.3281,
      "step": 23769
    },
    {
      "epoch": 1.090817309898582,
      "grad_norm": 0.4926886260509491,
      "learning_rate": 7.261373586421951e-06,
      "loss": 0.427,
      "step": 23770
    },
    {
      "epoch": 1.0908632004038363,
      "grad_norm": 0.5026347041130066,
      "learning_rate": 7.261154905836128e-06,
      "loss": 0.449,
      "step": 23771
    },
    {
      "epoch": 1.0909090909090908,
      "grad_norm": 0.4456912577152252,
      "learning_rate": 7.260936219812939e-06,
      "loss": 0.329,
      "step": 23772
    },
    {
      "epoch": 1.0909549814143453,
      "grad_norm": 0.44129589200019836,
      "learning_rate": 7.260717528352912e-06,
      "loss": 0.3557,
      "step": 23773
    },
    {
      "epoch": 1.0910008719195998,
      "grad_norm": 0.5035292506217957,
      "learning_rate": 7.260498831456573e-06,
      "loss": 0.3313,
      "step": 23774
    },
    {
      "epoch": 1.0910467624248543,
      "grad_norm": 0.4490951597690582,
      "learning_rate": 7.260280129124449e-06,
      "loss": 0.349,
      "step": 23775
    },
    {
      "epoch": 1.0910926529301088,
      "grad_norm": 0.43476301431655884,
      "learning_rate": 7.260061421357063e-06,
      "loss": 0.3292,
      "step": 23776
    },
    {
      "epoch": 1.0911385434353633,
      "grad_norm": 0.4856126308441162,
      "learning_rate": 7.259842708154944e-06,
      "loss": 0.4148,
      "step": 23777
    },
    {
      "epoch": 1.0911844339406176,
      "grad_norm": 0.46921876072883606,
      "learning_rate": 7.259623989518616e-06,
      "loss": 0.3376,
      "step": 23778
    },
    {
      "epoch": 1.091230324445872,
      "grad_norm": 0.4979473948478699,
      "learning_rate": 7.259405265448605e-06,
      "loss": 0.4038,
      "step": 23779
    },
    {
      "epoch": 1.0912762149511266,
      "grad_norm": 0.4819836914539337,
      "learning_rate": 7.259186535945437e-06,
      "loss": 0.342,
      "step": 23780
    },
    {
      "epoch": 1.091322105456381,
      "grad_norm": 0.4810265898704529,
      "learning_rate": 7.25896780100964e-06,
      "loss": 0.4267,
      "step": 23781
    },
    {
      "epoch": 1.0913679959616356,
      "grad_norm": 0.45266208052635193,
      "learning_rate": 7.258749060641738e-06,
      "loss": 0.374,
      "step": 23782
    },
    {
      "epoch": 1.09141388646689,
      "grad_norm": 0.4391326606273651,
      "learning_rate": 7.258530314842257e-06,
      "loss": 0.3189,
      "step": 23783
    },
    {
      "epoch": 1.0914597769721444,
      "grad_norm": 0.5090668201446533,
      "learning_rate": 7.258311563611723e-06,
      "loss": 0.3757,
      "step": 23784
    },
    {
      "epoch": 1.0915056674773989,
      "grad_norm": 0.4570576548576355,
      "learning_rate": 7.258092806950663e-06,
      "loss": 0.3208,
      "step": 23785
    },
    {
      "epoch": 1.0915515579826534,
      "grad_norm": 0.48916321992874146,
      "learning_rate": 7.257874044859603e-06,
      "loss": 0.3917,
      "step": 23786
    },
    {
      "epoch": 1.0915974484879079,
      "grad_norm": 0.45773088932037354,
      "learning_rate": 7.2576552773390676e-06,
      "loss": 0.3229,
      "step": 23787
    },
    {
      "epoch": 1.0916433389931623,
      "grad_norm": 0.4675646424293518,
      "learning_rate": 7.257436504389583e-06,
      "loss": 0.36,
      "step": 23788
    },
    {
      "epoch": 1.0916892294984168,
      "grad_norm": 0.4666242003440857,
      "learning_rate": 7.257217726011677e-06,
      "loss": 0.3572,
      "step": 23789
    },
    {
      "epoch": 1.0917351200036713,
      "grad_norm": 0.4719341993331909,
      "learning_rate": 7.2569989422058755e-06,
      "loss": 0.421,
      "step": 23790
    },
    {
      "epoch": 1.0917810105089256,
      "grad_norm": 0.4655524790287018,
      "learning_rate": 7.2567801529727035e-06,
      "loss": 0.3228,
      "step": 23791
    },
    {
      "epoch": 1.0918269010141801,
      "grad_norm": 0.4659241735935211,
      "learning_rate": 7.256561358312687e-06,
      "loss": 0.3502,
      "step": 23792
    },
    {
      "epoch": 1.0918727915194346,
      "grad_norm": 0.4599721431732178,
      "learning_rate": 7.2563425582263525e-06,
      "loss": 0.4093,
      "step": 23793
    },
    {
      "epoch": 1.0919186820246891,
      "grad_norm": 0.460589200258255,
      "learning_rate": 7.256123752714227e-06,
      "loss": 0.3039,
      "step": 23794
    },
    {
      "epoch": 1.0919645725299436,
      "grad_norm": 0.44122937321662903,
      "learning_rate": 7.2559049417768345e-06,
      "loss": 0.3044,
      "step": 23795
    },
    {
      "epoch": 1.092010463035198,
      "grad_norm": 0.4588642418384552,
      "learning_rate": 7.255686125414704e-06,
      "loss": 0.3128,
      "step": 23796
    },
    {
      "epoch": 1.0920563535404524,
      "grad_norm": 0.49759557843208313,
      "learning_rate": 7.255467303628361e-06,
      "loss": 0.4116,
      "step": 23797
    },
    {
      "epoch": 1.0921022440457069,
      "grad_norm": 0.5044850707054138,
      "learning_rate": 7.25524847641833e-06,
      "loss": 0.373,
      "step": 23798
    },
    {
      "epoch": 1.0921481345509614,
      "grad_norm": 0.4305585026741028,
      "learning_rate": 7.255029643785136e-06,
      "loss": 0.3315,
      "step": 23799
    },
    {
      "epoch": 1.0921940250562159,
      "grad_norm": 0.4639173150062561,
      "learning_rate": 7.254810805729311e-06,
      "loss": 0.3628,
      "step": 23800
    },
    {
      "epoch": 1.0922399155614704,
      "grad_norm": 0.4546792507171631,
      "learning_rate": 7.254591962251376e-06,
      "loss": 0.3294,
      "step": 23801
    },
    {
      "epoch": 1.0922858060667249,
      "grad_norm": 0.47488027811050415,
      "learning_rate": 7.254373113351859e-06,
      "loss": 0.368,
      "step": 23802
    },
    {
      "epoch": 1.0923316965719794,
      "grad_norm": 0.46031689643859863,
      "learning_rate": 7.254154259031288e-06,
      "loss": 0.3739,
      "step": 23803
    },
    {
      "epoch": 1.0923775870772336,
      "grad_norm": 0.5144749879837036,
      "learning_rate": 7.253935399290184e-06,
      "loss": 0.388,
      "step": 23804
    },
    {
      "epoch": 1.0924234775824881,
      "grad_norm": 0.4697451591491699,
      "learning_rate": 7.253716534129078e-06,
      "loss": 0.3372,
      "step": 23805
    },
    {
      "epoch": 1.0924693680877426,
      "grad_norm": 0.4673503637313843,
      "learning_rate": 7.253497663548495e-06,
      "loss": 0.3132,
      "step": 23806
    },
    {
      "epoch": 1.0925152585929971,
      "grad_norm": 0.4911215007305145,
      "learning_rate": 7.253278787548962e-06,
      "loss": 0.4051,
      "step": 23807
    },
    {
      "epoch": 1.0925611490982516,
      "grad_norm": 0.45835134387016296,
      "learning_rate": 7.253059906131006e-06,
      "loss": 0.3212,
      "step": 23808
    },
    {
      "epoch": 1.0926070396035061,
      "grad_norm": 0.4836196005344391,
      "learning_rate": 7.2528410192951496e-06,
      "loss": 0.2525,
      "step": 23809
    },
    {
      "epoch": 1.0926529301087604,
      "grad_norm": 0.452505886554718,
      "learning_rate": 7.2526221270419215e-06,
      "loss": 0.2971,
      "step": 23810
    },
    {
      "epoch": 1.092698820614015,
      "grad_norm": 0.4454563558101654,
      "learning_rate": 7.252403229371851e-06,
      "loss": 0.3525,
      "step": 23811
    },
    {
      "epoch": 1.0927447111192694,
      "grad_norm": 0.4527369439601898,
      "learning_rate": 7.2521843262854585e-06,
      "loss": 0.3502,
      "step": 23812
    },
    {
      "epoch": 1.092790601624524,
      "grad_norm": 0.4554269313812256,
      "learning_rate": 7.2519654177832755e-06,
      "loss": 0.3629,
      "step": 23813
    },
    {
      "epoch": 1.0928364921297784,
      "grad_norm": 0.5047026872634888,
      "learning_rate": 7.251746503865825e-06,
      "loss": 0.4132,
      "step": 23814
    },
    {
      "epoch": 1.092882382635033,
      "grad_norm": 0.4550182521343231,
      "learning_rate": 7.251527584533635e-06,
      "loss": 0.3077,
      "step": 23815
    },
    {
      "epoch": 1.0929282731402872,
      "grad_norm": 0.4915163815021515,
      "learning_rate": 7.251308659787233e-06,
      "loss": 0.3832,
      "step": 23816
    },
    {
      "epoch": 1.0929741636455417,
      "grad_norm": 0.4778660535812378,
      "learning_rate": 7.2510897296271434e-06,
      "loss": 0.4038,
      "step": 23817
    },
    {
      "epoch": 1.0930200541507962,
      "grad_norm": 0.4644222557544708,
      "learning_rate": 7.2508707940538925e-06,
      "loss": 0.3737,
      "step": 23818
    },
    {
      "epoch": 1.0930659446560507,
      "grad_norm": 0.45537397265434265,
      "learning_rate": 7.25065185306801e-06,
      "loss": 0.3557,
      "step": 23819
    },
    {
      "epoch": 1.0931118351613052,
      "grad_norm": 0.4812740683555603,
      "learning_rate": 7.250432906670019e-06,
      "loss": 0.4309,
      "step": 23820
    },
    {
      "epoch": 1.0931577256665597,
      "grad_norm": 0.4736979603767395,
      "learning_rate": 7.250213954860446e-06,
      "loss": 0.3299,
      "step": 23821
    },
    {
      "epoch": 1.093203616171814,
      "grad_norm": 0.47063735127449036,
      "learning_rate": 7.2499949976398195e-06,
      "loss": 0.3503,
      "step": 23822
    },
    {
      "epoch": 1.0932495066770684,
      "grad_norm": 0.459395170211792,
      "learning_rate": 7.249776035008664e-06,
      "loss": 0.3651,
      "step": 23823
    },
    {
      "epoch": 1.093295397182323,
      "grad_norm": 0.4948347508907318,
      "learning_rate": 7.24955706696751e-06,
      "loss": 0.4022,
      "step": 23824
    },
    {
      "epoch": 1.0933412876875774,
      "grad_norm": 0.4732950031757355,
      "learning_rate": 7.24933809351688e-06,
      "loss": 0.3573,
      "step": 23825
    },
    {
      "epoch": 1.093387178192832,
      "grad_norm": 0.4327717423439026,
      "learning_rate": 7.2491191146573e-06,
      "loss": 0.2647,
      "step": 23826
    },
    {
      "epoch": 1.0934330686980864,
      "grad_norm": 0.46840932965278625,
      "learning_rate": 7.2489001303893e-06,
      "loss": 0.3658,
      "step": 23827
    },
    {
      "epoch": 1.093478959203341,
      "grad_norm": 0.4735131859779358,
      "learning_rate": 7.248681140713405e-06,
      "loss": 0.3346,
      "step": 23828
    },
    {
      "epoch": 1.0935248497085952,
      "grad_norm": 0.463206022977829,
      "learning_rate": 7.24846214563014e-06,
      "loss": 0.3254,
      "step": 23829
    },
    {
      "epoch": 1.0935707402138497,
      "grad_norm": 0.520291268825531,
      "learning_rate": 7.248243145140036e-06,
      "loss": 0.4152,
      "step": 23830
    },
    {
      "epoch": 1.0936166307191042,
      "grad_norm": 0.4621307849884033,
      "learning_rate": 7.2480241392436155e-06,
      "loss": 0.364,
      "step": 23831
    },
    {
      "epoch": 1.0936625212243587,
      "grad_norm": 0.4838724434375763,
      "learning_rate": 7.247805127941405e-06,
      "loss": 0.388,
      "step": 23832
    },
    {
      "epoch": 1.0937084117296132,
      "grad_norm": 0.46205249428749084,
      "learning_rate": 7.247586111233935e-06,
      "loss": 0.3222,
      "step": 23833
    },
    {
      "epoch": 1.0937543022348677,
      "grad_norm": 0.5231717824935913,
      "learning_rate": 7.247367089121729e-06,
      "loss": 0.4484,
      "step": 23834
    },
    {
      "epoch": 1.093800192740122,
      "grad_norm": 0.4860301911830902,
      "learning_rate": 7.247148061605314e-06,
      "loss": 0.3566,
      "step": 23835
    },
    {
      "epoch": 1.0938460832453765,
      "grad_norm": 0.4353000521659851,
      "learning_rate": 7.2469290286852165e-06,
      "loss": 0.2877,
      "step": 23836
    },
    {
      "epoch": 1.093891973750631,
      "grad_norm": 0.5021859407424927,
      "learning_rate": 7.246709990361964e-06,
      "loss": 0.4575,
      "step": 23837
    },
    {
      "epoch": 1.0939378642558855,
      "grad_norm": 0.43979597091674805,
      "learning_rate": 7.246490946636084e-06,
      "loss": 0.3647,
      "step": 23838
    },
    {
      "epoch": 1.09398375476114,
      "grad_norm": 0.5027796626091003,
      "learning_rate": 7.246271897508102e-06,
      "loss": 0.3874,
      "step": 23839
    },
    {
      "epoch": 1.0940296452663945,
      "grad_norm": 0.46337106823921204,
      "learning_rate": 7.246052842978544e-06,
      "loss": 0.3631,
      "step": 23840
    },
    {
      "epoch": 1.094075535771649,
      "grad_norm": 0.5020548105239868,
      "learning_rate": 7.24583378304794e-06,
      "loss": 0.4274,
      "step": 23841
    },
    {
      "epoch": 1.0941214262769032,
      "grad_norm": 0.5021292567253113,
      "learning_rate": 7.245614717716812e-06,
      "loss": 0.3626,
      "step": 23842
    },
    {
      "epoch": 1.0941673167821577,
      "grad_norm": 0.47919702529907227,
      "learning_rate": 7.245395646985691e-06,
      "loss": 0.387,
      "step": 23843
    },
    {
      "epoch": 1.0942132072874122,
      "grad_norm": 0.45636335015296936,
      "learning_rate": 7.245176570855102e-06,
      "loss": 0.3361,
      "step": 23844
    },
    {
      "epoch": 1.0942590977926667,
      "grad_norm": 0.4448567032814026,
      "learning_rate": 7.24495748932557e-06,
      "loss": 0.3358,
      "step": 23845
    },
    {
      "epoch": 1.0943049882979212,
      "grad_norm": 0.48710906505584717,
      "learning_rate": 7.2447384023976265e-06,
      "loss": 0.4065,
      "step": 23846
    },
    {
      "epoch": 1.0943508788031757,
      "grad_norm": 0.4803224503993988,
      "learning_rate": 7.2445193100717935e-06,
      "loss": 0.3668,
      "step": 23847
    },
    {
      "epoch": 1.09439676930843,
      "grad_norm": 0.4602110981941223,
      "learning_rate": 7.244300212348603e-06,
      "loss": 0.3648,
      "step": 23848
    },
    {
      "epoch": 1.0944426598136845,
      "grad_norm": 0.4637419283390045,
      "learning_rate": 7.244081109228576e-06,
      "loss": 0.3273,
      "step": 23849
    },
    {
      "epoch": 1.094488550318939,
      "grad_norm": 0.4870504140853882,
      "learning_rate": 7.2438620007122425e-06,
      "loss": 0.3853,
      "step": 23850
    },
    {
      "epoch": 1.0945344408241935,
      "grad_norm": 0.47301194071769714,
      "learning_rate": 7.24364288680013e-06,
      "loss": 0.3163,
      "step": 23851
    },
    {
      "epoch": 1.094580331329448,
      "grad_norm": 0.4585762619972229,
      "learning_rate": 7.2434237674927634e-06,
      "loss": 0.34,
      "step": 23852
    },
    {
      "epoch": 1.0946262218347025,
      "grad_norm": 0.4245307445526123,
      "learning_rate": 7.243204642790672e-06,
      "loss": 0.3212,
      "step": 23853
    },
    {
      "epoch": 1.0946721123399568,
      "grad_norm": 0.45641079545021057,
      "learning_rate": 7.2429855126943806e-06,
      "loss": 0.3276,
      "step": 23854
    },
    {
      "epoch": 1.0947180028452113,
      "grad_norm": 0.47232240438461304,
      "learning_rate": 7.242766377204415e-06,
      "loss": 0.3814,
      "step": 23855
    },
    {
      "epoch": 1.0947638933504658,
      "grad_norm": 0.46342048048973083,
      "learning_rate": 7.242547236321308e-06,
      "loss": 0.3675,
      "step": 23856
    },
    {
      "epoch": 1.0948097838557203,
      "grad_norm": 0.4842508137226105,
      "learning_rate": 7.242328090045581e-06,
      "loss": 0.3649,
      "step": 23857
    },
    {
      "epoch": 1.0948556743609748,
      "grad_norm": 0.4826911687850952,
      "learning_rate": 7.242108938377762e-06,
      "loss": 0.4135,
      "step": 23858
    },
    {
      "epoch": 1.0949015648662292,
      "grad_norm": 0.5023464560508728,
      "learning_rate": 7.241889781318379e-06,
      "loss": 0.3668,
      "step": 23859
    },
    {
      "epoch": 1.0949474553714835,
      "grad_norm": 0.4372011125087738,
      "learning_rate": 7.2416706188679585e-06,
      "loss": 0.3262,
      "step": 23860
    },
    {
      "epoch": 1.094993345876738,
      "grad_norm": 0.47171303629875183,
      "learning_rate": 7.241451451027027e-06,
      "loss": 0.3631,
      "step": 23861
    },
    {
      "epoch": 1.0950392363819925,
      "grad_norm": 0.5472835898399353,
      "learning_rate": 7.241232277796113e-06,
      "loss": 0.3197,
      "step": 23862
    },
    {
      "epoch": 1.095085126887247,
      "grad_norm": 0.4706948399543762,
      "learning_rate": 7.241013099175744e-06,
      "loss": 0.3305,
      "step": 23863
    },
    {
      "epoch": 1.0951310173925015,
      "grad_norm": 0.44000208377838135,
      "learning_rate": 7.240793915166444e-06,
      "loss": 0.336,
      "step": 23864
    },
    {
      "epoch": 1.095176907897756,
      "grad_norm": 0.4843046963214874,
      "learning_rate": 7.240574725768743e-06,
      "loss": 0.3277,
      "step": 23865
    },
    {
      "epoch": 1.0952227984030105,
      "grad_norm": 0.483981728553772,
      "learning_rate": 7.240355530983165e-06,
      "loss": 0.388,
      "step": 23866
    },
    {
      "epoch": 1.0952686889082648,
      "grad_norm": 0.43154647946357727,
      "learning_rate": 7.240136330810241e-06,
      "loss": 0.3198,
      "step": 23867
    },
    {
      "epoch": 1.0953145794135193,
      "grad_norm": 0.4725511372089386,
      "learning_rate": 7.2399171252504965e-06,
      "loss": 0.352,
      "step": 23868
    },
    {
      "epoch": 1.0953604699187738,
      "grad_norm": 0.4869677424430847,
      "learning_rate": 7.239697914304457e-06,
      "loss": 0.3394,
      "step": 23869
    },
    {
      "epoch": 1.0954063604240283,
      "grad_norm": 0.4616750180721283,
      "learning_rate": 7.2394786979726505e-06,
      "loss": 0.336,
      "step": 23870
    },
    {
      "epoch": 1.0954522509292828,
      "grad_norm": 0.45213887095451355,
      "learning_rate": 7.239259476255604e-06,
      "loss": 0.3414,
      "step": 23871
    },
    {
      "epoch": 1.0954981414345373,
      "grad_norm": 0.4849543571472168,
      "learning_rate": 7.239040249153847e-06,
      "loss": 0.3476,
      "step": 23872
    },
    {
      "epoch": 1.0955440319397916,
      "grad_norm": 0.49670228362083435,
      "learning_rate": 7.238821016667905e-06,
      "loss": 0.4076,
      "step": 23873
    },
    {
      "epoch": 1.095589922445046,
      "grad_norm": 0.4579293131828308,
      "learning_rate": 7.238601778798304e-06,
      "loss": 0.3249,
      "step": 23874
    },
    {
      "epoch": 1.0956358129503005,
      "grad_norm": 0.46338701248168945,
      "learning_rate": 7.238382535545572e-06,
      "loss": 0.3791,
      "step": 23875
    },
    {
      "epoch": 1.095681703455555,
      "grad_norm": 0.4487346112728119,
      "learning_rate": 7.238163286910238e-06,
      "loss": 0.3363,
      "step": 23876
    },
    {
      "epoch": 1.0957275939608095,
      "grad_norm": 0.46870821714401245,
      "learning_rate": 7.2379440328928255e-06,
      "loss": 0.4302,
      "step": 23877
    },
    {
      "epoch": 1.095773484466064,
      "grad_norm": 0.4493914246559143,
      "learning_rate": 7.237724773493866e-06,
      "loss": 0.341,
      "step": 23878
    },
    {
      "epoch": 1.0958193749713185,
      "grad_norm": 0.48510655760765076,
      "learning_rate": 7.237505508713884e-06,
      "loss": 0.3465,
      "step": 23879
    },
    {
      "epoch": 1.0958652654765728,
      "grad_norm": 0.4675643742084503,
      "learning_rate": 7.237286238553407e-06,
      "loss": 0.3639,
      "step": 23880
    },
    {
      "epoch": 1.0959111559818273,
      "grad_norm": 0.508488655090332,
      "learning_rate": 7.2370669630129615e-06,
      "loss": 0.413,
      "step": 23881
    },
    {
      "epoch": 1.0959570464870818,
      "grad_norm": 0.47606125473976135,
      "learning_rate": 7.236847682093078e-06,
      "loss": 0.3574,
      "step": 23882
    },
    {
      "epoch": 1.0960029369923363,
      "grad_norm": 0.4884919822216034,
      "learning_rate": 7.236628395794281e-06,
      "loss": 0.4183,
      "step": 23883
    },
    {
      "epoch": 1.0960488274975908,
      "grad_norm": 0.4317469894886017,
      "learning_rate": 7.236409104117099e-06,
      "loss": 0.3248,
      "step": 23884
    },
    {
      "epoch": 1.0960947180028453,
      "grad_norm": 0.460892915725708,
      "learning_rate": 7.23618980706206e-06,
      "loss": 0.3244,
      "step": 23885
    },
    {
      "epoch": 1.0961406085080996,
      "grad_norm": 0.3895472586154938,
      "learning_rate": 7.235970504629688e-06,
      "loss": 0.237,
      "step": 23886
    },
    {
      "epoch": 1.096186499013354,
      "grad_norm": 0.46065688133239746,
      "learning_rate": 7.235751196820513e-06,
      "loss": 0.343,
      "step": 23887
    },
    {
      "epoch": 1.0962323895186086,
      "grad_norm": 0.49066001176834106,
      "learning_rate": 7.235531883635063e-06,
      "loss": 0.4162,
      "step": 23888
    },
    {
      "epoch": 1.096278280023863,
      "grad_norm": 0.4778464436531067,
      "learning_rate": 7.235312565073864e-06,
      "loss": 0.3794,
      "step": 23889
    },
    {
      "epoch": 1.0963241705291176,
      "grad_norm": 0.5221874117851257,
      "learning_rate": 7.2350932411374446e-06,
      "loss": 0.434,
      "step": 23890
    },
    {
      "epoch": 1.096370061034372,
      "grad_norm": 0.4347390830516815,
      "learning_rate": 7.23487391182633e-06,
      "loss": 0.3194,
      "step": 23891
    },
    {
      "epoch": 1.0964159515396266,
      "grad_norm": 0.4671892821788788,
      "learning_rate": 7.234654577141048e-06,
      "loss": 0.3189,
      "step": 23892
    },
    {
      "epoch": 1.0964618420448808,
      "grad_norm": 0.449482262134552,
      "learning_rate": 7.2344352370821295e-06,
      "loss": 0.3193,
      "step": 23893
    },
    {
      "epoch": 1.0965077325501353,
      "grad_norm": 0.4491939842700958,
      "learning_rate": 7.234215891650098e-06,
      "loss": 0.3341,
      "step": 23894
    },
    {
      "epoch": 1.0965536230553898,
      "grad_norm": 0.4929482340812683,
      "learning_rate": 7.233996540845482e-06,
      "loss": 0.3637,
      "step": 23895
    },
    {
      "epoch": 1.0965995135606443,
      "grad_norm": 0.4477548599243164,
      "learning_rate": 7.2337771846688125e-06,
      "loss": 0.3241,
      "step": 23896
    },
    {
      "epoch": 1.0966454040658988,
      "grad_norm": 0.44800859689712524,
      "learning_rate": 7.23355782312061e-06,
      "loss": 0.3233,
      "step": 23897
    },
    {
      "epoch": 1.0966912945711533,
      "grad_norm": 0.4924427568912506,
      "learning_rate": 7.233338456201409e-06,
      "loss": 0.3756,
      "step": 23898
    },
    {
      "epoch": 1.0967371850764076,
      "grad_norm": 0.44594806432724,
      "learning_rate": 7.233119083911731e-06,
      "loss": 0.3206,
      "step": 23899
    },
    {
      "epoch": 1.096783075581662,
      "grad_norm": 0.4848455786705017,
      "learning_rate": 7.232899706252108e-06,
      "loss": 0.3748,
      "step": 23900
    },
    {
      "epoch": 1.0968289660869166,
      "grad_norm": 0.42995554208755493,
      "learning_rate": 7.232680323223067e-06,
      "loss": 0.3188,
      "step": 23901
    },
    {
      "epoch": 1.096874856592171,
      "grad_norm": 0.46562460064888,
      "learning_rate": 7.232460934825132e-06,
      "loss": 0.4028,
      "step": 23902
    },
    {
      "epoch": 1.0969207470974256,
      "grad_norm": 0.5087507367134094,
      "learning_rate": 7.232241541058834e-06,
      "loss": 0.4107,
      "step": 23903
    },
    {
      "epoch": 1.09696663760268,
      "grad_norm": 0.4918316900730133,
      "learning_rate": 7.232022141924701e-06,
      "loss": 0.4457,
      "step": 23904
    },
    {
      "epoch": 1.0970125281079344,
      "grad_norm": 0.5147467851638794,
      "learning_rate": 7.231802737423257e-06,
      "loss": 0.2945,
      "step": 23905
    },
    {
      "epoch": 1.0970584186131889,
      "grad_norm": 0.4690674841403961,
      "learning_rate": 7.231583327555033e-06,
      "loss": 0.3866,
      "step": 23906
    },
    {
      "epoch": 1.0971043091184434,
      "grad_norm": 0.5054628252983093,
      "learning_rate": 7.231363912320556e-06,
      "loss": 0.4408,
      "step": 23907
    },
    {
      "epoch": 1.0971501996236979,
      "grad_norm": 0.473814994096756,
      "learning_rate": 7.231144491720351e-06,
      "loss": 0.3847,
      "step": 23908
    },
    {
      "epoch": 1.0971960901289524,
      "grad_norm": 0.4512443542480469,
      "learning_rate": 7.230925065754949e-06,
      "loss": 0.3435,
      "step": 23909
    },
    {
      "epoch": 1.0972419806342069,
      "grad_norm": 0.5123233795166016,
      "learning_rate": 7.230705634424876e-06,
      "loss": 0.418,
      "step": 23910
    },
    {
      "epoch": 1.0972878711394611,
      "grad_norm": 0.4995267689228058,
      "learning_rate": 7.230486197730662e-06,
      "loss": 0.4127,
      "step": 23911
    },
    {
      "epoch": 1.0973337616447156,
      "grad_norm": 0.5109555721282959,
      "learning_rate": 7.23026675567283e-06,
      "loss": 0.4773,
      "step": 23912
    },
    {
      "epoch": 1.0973796521499701,
      "grad_norm": 0.4698633551597595,
      "learning_rate": 7.2300473082519105e-06,
      "loss": 0.3604,
      "step": 23913
    },
    {
      "epoch": 1.0974255426552246,
      "grad_norm": 0.4330309331417084,
      "learning_rate": 7.229827855468432e-06,
      "loss": 0.2692,
      "step": 23914
    },
    {
      "epoch": 1.0974714331604791,
      "grad_norm": 0.45150139927864075,
      "learning_rate": 7.229608397322921e-06,
      "loss": 0.2744,
      "step": 23915
    },
    {
      "epoch": 1.0975173236657336,
      "grad_norm": 0.507512092590332,
      "learning_rate": 7.2293889338159055e-06,
      "loss": 0.4556,
      "step": 23916
    },
    {
      "epoch": 1.0975632141709881,
      "grad_norm": 0.4656270742416382,
      "learning_rate": 7.229169464947914e-06,
      "loss": 0.3476,
      "step": 23917
    },
    {
      "epoch": 1.0976091046762424,
      "grad_norm": 0.4692942798137665,
      "learning_rate": 7.228949990719473e-06,
      "loss": 0.3322,
      "step": 23918
    },
    {
      "epoch": 1.097654995181497,
      "grad_norm": 0.46237239241600037,
      "learning_rate": 7.22873051113111e-06,
      "loss": 0.3607,
      "step": 23919
    },
    {
      "epoch": 1.0977008856867514,
      "grad_norm": 0.489328533411026,
      "learning_rate": 7.228511026183353e-06,
      "loss": 0.3563,
      "step": 23920
    },
    {
      "epoch": 1.097746776192006,
      "grad_norm": 0.4600256383419037,
      "learning_rate": 7.228291535876733e-06,
      "loss": 0.3421,
      "step": 23921
    },
    {
      "epoch": 1.0977926666972604,
      "grad_norm": 0.45237934589385986,
      "learning_rate": 7.228072040211773e-06,
      "loss": 0.2897,
      "step": 23922
    },
    {
      "epoch": 1.0978385572025149,
      "grad_norm": 0.4462050199508667,
      "learning_rate": 7.227852539189002e-06,
      "loss": 0.3305,
      "step": 23923
    },
    {
      "epoch": 1.0978844477077692,
      "grad_norm": 0.4860582649707794,
      "learning_rate": 7.227633032808952e-06,
      "loss": 0.3417,
      "step": 23924
    },
    {
      "epoch": 1.0979303382130237,
      "grad_norm": 0.44280707836151123,
      "learning_rate": 7.227413521072144e-06,
      "loss": 0.3151,
      "step": 23925
    },
    {
      "epoch": 1.0979762287182782,
      "grad_norm": 0.4673399031162262,
      "learning_rate": 7.227194003979111e-06,
      "loss": 0.3369,
      "step": 23926
    },
    {
      "epoch": 1.0980221192235327,
      "grad_norm": 0.48334190249443054,
      "learning_rate": 7.22697448153038e-06,
      "loss": 0.3884,
      "step": 23927
    },
    {
      "epoch": 1.0980680097287872,
      "grad_norm": 0.4762553572654724,
      "learning_rate": 7.226754953726478e-06,
      "loss": 0.3628,
      "step": 23928
    },
    {
      "epoch": 1.0981139002340417,
      "grad_norm": 0.4945092797279358,
      "learning_rate": 7.2265354205679325e-06,
      "loss": 0.3687,
      "step": 23929
    },
    {
      "epoch": 1.0981597907392961,
      "grad_norm": 0.45898112654685974,
      "learning_rate": 7.2263158820552735e-06,
      "loss": 0.3327,
      "step": 23930
    },
    {
      "epoch": 1.0982056812445504,
      "grad_norm": 0.444606214761734,
      "learning_rate": 7.226096338189026e-06,
      "loss": 0.3047,
      "step": 23931
    },
    {
      "epoch": 1.098251571749805,
      "grad_norm": 0.46046361327171326,
      "learning_rate": 7.22587678896972e-06,
      "loss": 0.3556,
      "step": 23932
    },
    {
      "epoch": 1.0982974622550594,
      "grad_norm": 0.44336238503456116,
      "learning_rate": 7.2256572343978825e-06,
      "loss": 0.3142,
      "step": 23933
    },
    {
      "epoch": 1.098343352760314,
      "grad_norm": 0.45828118920326233,
      "learning_rate": 7.2254376744740415e-06,
      "loss": 0.3582,
      "step": 23934
    },
    {
      "epoch": 1.0983892432655684,
      "grad_norm": 0.530708909034729,
      "learning_rate": 7.225218109198726e-06,
      "loss": 0.3805,
      "step": 23935
    },
    {
      "epoch": 1.098435133770823,
      "grad_norm": 0.436158686876297,
      "learning_rate": 7.224998538572463e-06,
      "loss": 0.3044,
      "step": 23936
    },
    {
      "epoch": 1.0984810242760772,
      "grad_norm": 0.46997302770614624,
      "learning_rate": 7.22477896259578e-06,
      "loss": 0.4233,
      "step": 23937
    },
    {
      "epoch": 1.0985269147813317,
      "grad_norm": 0.44957563281059265,
      "learning_rate": 7.224559381269208e-06,
      "loss": 0.3281,
      "step": 23938
    },
    {
      "epoch": 1.0985728052865862,
      "grad_norm": 0.4778108596801758,
      "learning_rate": 7.22433979459327e-06,
      "loss": 0.3716,
      "step": 23939
    },
    {
      "epoch": 1.0986186957918407,
      "grad_norm": 0.4355666935443878,
      "learning_rate": 7.224120202568498e-06,
      "loss": 0.3059,
      "step": 23940
    },
    {
      "epoch": 1.0986645862970952,
      "grad_norm": 0.46271151304244995,
      "learning_rate": 7.22390060519542e-06,
      "loss": 0.3929,
      "step": 23941
    },
    {
      "epoch": 1.0987104768023497,
      "grad_norm": 0.45417970418930054,
      "learning_rate": 7.223681002474561e-06,
      "loss": 0.3389,
      "step": 23942
    },
    {
      "epoch": 1.098756367307604,
      "grad_norm": 0.4929611384868622,
      "learning_rate": 7.223461394406452e-06,
      "loss": 0.4088,
      "step": 23943
    },
    {
      "epoch": 1.0988022578128585,
      "grad_norm": 0.4923354387283325,
      "learning_rate": 7.223241780991621e-06,
      "loss": 0.4291,
      "step": 23944
    },
    {
      "epoch": 1.098848148318113,
      "grad_norm": 0.47711026668548584,
      "learning_rate": 7.223022162230593e-06,
      "loss": 0.4143,
      "step": 23945
    },
    {
      "epoch": 1.0988940388233674,
      "grad_norm": 0.5049922466278076,
      "learning_rate": 7.222802538123898e-06,
      "loss": 0.4545,
      "step": 23946
    },
    {
      "epoch": 1.098939929328622,
      "grad_norm": 0.5047472715377808,
      "learning_rate": 7.222582908672067e-06,
      "loss": 0.4117,
      "step": 23947
    },
    {
      "epoch": 1.0989858198338764,
      "grad_norm": 0.4697808623313904,
      "learning_rate": 7.222363273875624e-06,
      "loss": 0.4552,
      "step": 23948
    },
    {
      "epoch": 1.0990317103391307,
      "grad_norm": 0.4586635231971741,
      "learning_rate": 7.222143633735099e-06,
      "loss": 0.3585,
      "step": 23949
    },
    {
      "epoch": 1.0990776008443852,
      "grad_norm": 0.412824422121048,
      "learning_rate": 7.22192398825102e-06,
      "loss": 0.2696,
      "step": 23950
    },
    {
      "epoch": 1.0991234913496397,
      "grad_norm": 0.4264359772205353,
      "learning_rate": 7.221704337423916e-06,
      "loss": 0.289,
      "step": 23951
    },
    {
      "epoch": 1.0991693818548942,
      "grad_norm": 0.489128053188324,
      "learning_rate": 7.221484681254314e-06,
      "loss": 0.4592,
      "step": 23952
    },
    {
      "epoch": 1.0992152723601487,
      "grad_norm": 0.46833640336990356,
      "learning_rate": 7.22126501974274e-06,
      "loss": 0.3883,
      "step": 23953
    },
    {
      "epoch": 1.0992611628654032,
      "grad_norm": 0.4626694321632385,
      "learning_rate": 7.221045352889729e-06,
      "loss": 0.36,
      "step": 23954
    },
    {
      "epoch": 1.0993070533706577,
      "grad_norm": 0.47982850670814514,
      "learning_rate": 7.220825680695802e-06,
      "loss": 0.3565,
      "step": 23955
    },
    {
      "epoch": 1.099352943875912,
      "grad_norm": 0.4920386075973511,
      "learning_rate": 7.22060600316149e-06,
      "loss": 0.3658,
      "step": 23956
    },
    {
      "epoch": 1.0993988343811665,
      "grad_norm": 0.47356560826301575,
      "learning_rate": 7.220386320287321e-06,
      "loss": 0.3536,
      "step": 23957
    },
    {
      "epoch": 1.099444724886421,
      "grad_norm": 0.48194918036460876,
      "learning_rate": 7.220166632073825e-06,
      "loss": 0.3514,
      "step": 23958
    },
    {
      "epoch": 1.0994906153916755,
      "grad_norm": 0.48215362429618835,
      "learning_rate": 7.219946938521528e-06,
      "loss": 0.3164,
      "step": 23959
    },
    {
      "epoch": 1.09953650589693,
      "grad_norm": 0.4408113360404968,
      "learning_rate": 7.219727239630961e-06,
      "loss": 0.3021,
      "step": 23960
    },
    {
      "epoch": 1.0995823964021845,
      "grad_norm": 0.464308500289917,
      "learning_rate": 7.219507535402649e-06,
      "loss": 0.3921,
      "step": 23961
    },
    {
      "epoch": 1.0996282869074387,
      "grad_norm": 0.4719148576259613,
      "learning_rate": 7.2192878258371225e-06,
      "loss": 0.3345,
      "step": 23962
    },
    {
      "epoch": 1.0996741774126932,
      "grad_norm": 0.48751190304756165,
      "learning_rate": 7.219068110934909e-06,
      "loss": 0.393,
      "step": 23963
    },
    {
      "epoch": 1.0997200679179477,
      "grad_norm": 0.4642772972583771,
      "learning_rate": 7.218848390696537e-06,
      "loss": 0.3316,
      "step": 23964
    },
    {
      "epoch": 1.0997659584232022,
      "grad_norm": 0.482030987739563,
      "learning_rate": 7.218628665122534e-06,
      "loss": 0.3998,
      "step": 23965
    },
    {
      "epoch": 1.0998118489284567,
      "grad_norm": 0.4981866180896759,
      "learning_rate": 7.218408934213431e-06,
      "loss": 0.3633,
      "step": 23966
    },
    {
      "epoch": 1.0998577394337112,
      "grad_norm": 0.5212001800537109,
      "learning_rate": 7.218189197969754e-06,
      "loss": 0.3884,
      "step": 23967
    },
    {
      "epoch": 1.0999036299389657,
      "grad_norm": 0.4873100817203522,
      "learning_rate": 7.217969456392031e-06,
      "loss": 0.3613,
      "step": 23968
    },
    {
      "epoch": 1.09994952044422,
      "grad_norm": 0.455390989780426,
      "learning_rate": 7.217749709480792e-06,
      "loss": 0.2966,
      "step": 23969
    },
    {
      "epoch": 1.0999954109494745,
      "grad_norm": 0.44289064407348633,
      "learning_rate": 7.217529957236566e-06,
      "loss": 0.3024,
      "step": 23970
    },
    {
      "epoch": 1.100041301454729,
      "grad_norm": 0.49131977558135986,
      "learning_rate": 7.217310199659879e-06,
      "loss": 0.3602,
      "step": 23971
    },
    {
      "epoch": 1.1000871919599835,
      "grad_norm": 0.4802387058734894,
      "learning_rate": 7.217090436751263e-06,
      "loss": 0.3948,
      "step": 23972
    },
    {
      "epoch": 1.100133082465238,
      "grad_norm": 0.47183507680892944,
      "learning_rate": 7.216870668511242e-06,
      "loss": 0.3968,
      "step": 23973
    },
    {
      "epoch": 1.1001789729704925,
      "grad_norm": 0.47388336062431335,
      "learning_rate": 7.216650894940347e-06,
      "loss": 0.3255,
      "step": 23974
    },
    {
      "epoch": 1.1002248634757468,
      "grad_norm": 0.479500412940979,
      "learning_rate": 7.2164311160391065e-06,
      "loss": 0.3953,
      "step": 23975
    },
    {
      "epoch": 1.1002707539810013,
      "grad_norm": 0.4891059696674347,
      "learning_rate": 7.216211331808049e-06,
      "loss": 0.44,
      "step": 23976
    },
    {
      "epoch": 1.1003166444862558,
      "grad_norm": 0.4971572160720825,
      "learning_rate": 7.215991542247702e-06,
      "loss": 0.4193,
      "step": 23977
    },
    {
      "epoch": 1.1003625349915103,
      "grad_norm": 0.45128193497657776,
      "learning_rate": 7.215771747358595e-06,
      "loss": 0.3447,
      "step": 23978
    },
    {
      "epoch": 1.1004084254967648,
      "grad_norm": 0.478458046913147,
      "learning_rate": 7.2155519471412574e-06,
      "loss": 0.3461,
      "step": 23979
    },
    {
      "epoch": 1.1004543160020193,
      "grad_norm": 0.4737812578678131,
      "learning_rate": 7.215332141596215e-06,
      "loss": 0.3627,
      "step": 23980
    },
    {
      "epoch": 1.1005002065072738,
      "grad_norm": 0.4808630049228668,
      "learning_rate": 7.215112330723999e-06,
      "loss": 0.4416,
      "step": 23981
    },
    {
      "epoch": 1.100546097012528,
      "grad_norm": 0.513425350189209,
      "learning_rate": 7.214892514525136e-06,
      "loss": 0.3953,
      "step": 23982
    },
    {
      "epoch": 1.1005919875177825,
      "grad_norm": 0.452442467212677,
      "learning_rate": 7.214672693000157e-06,
      "loss": 0.3492,
      "step": 23983
    },
    {
      "epoch": 1.100637878023037,
      "grad_norm": 0.47945713996887207,
      "learning_rate": 7.214452866149589e-06,
      "loss": 0.3568,
      "step": 23984
    },
    {
      "epoch": 1.1006837685282915,
      "grad_norm": 0.5168142914772034,
      "learning_rate": 7.2142330339739585e-06,
      "loss": 0.4286,
      "step": 23985
    },
    {
      "epoch": 1.100729659033546,
      "grad_norm": 0.45413991808891296,
      "learning_rate": 7.214013196473798e-06,
      "loss": 0.3273,
      "step": 23986
    },
    {
      "epoch": 1.1007755495388005,
      "grad_norm": 0.4300079345703125,
      "learning_rate": 7.2137933536496355e-06,
      "loss": 0.3478,
      "step": 23987
    },
    {
      "epoch": 1.1008214400440548,
      "grad_norm": 0.48418566584587097,
      "learning_rate": 7.213573505501996e-06,
      "loss": 0.3968,
      "step": 23988
    },
    {
      "epoch": 1.1008673305493093,
      "grad_norm": 0.4967102110385895,
      "learning_rate": 7.213353652031414e-06,
      "loss": 0.3387,
      "step": 23989
    },
    {
      "epoch": 1.1009132210545638,
      "grad_norm": 0.4554750621318817,
      "learning_rate": 7.213133793238412e-06,
      "loss": 0.3062,
      "step": 23990
    },
    {
      "epoch": 1.1009591115598183,
      "grad_norm": 0.5580289959907532,
      "learning_rate": 7.212913929123523e-06,
      "loss": 0.3075,
      "step": 23991
    },
    {
      "epoch": 1.1010050020650728,
      "grad_norm": 0.5773463845252991,
      "learning_rate": 7.212694059687275e-06,
      "loss": 0.3915,
      "step": 23992
    },
    {
      "epoch": 1.1010508925703273,
      "grad_norm": 0.46127811074256897,
      "learning_rate": 7.212474184930196e-06,
      "loss": 0.3437,
      "step": 23993
    },
    {
      "epoch": 1.1010967830755816,
      "grad_norm": 0.4302867650985718,
      "learning_rate": 7.212254304852815e-06,
      "loss": 0.3106,
      "step": 23994
    },
    {
      "epoch": 1.101142673580836,
      "grad_norm": 0.5103839635848999,
      "learning_rate": 7.21203441945566e-06,
      "loss": 0.4396,
      "step": 23995
    },
    {
      "epoch": 1.1011885640860906,
      "grad_norm": 0.5239191651344299,
      "learning_rate": 7.21181452873926e-06,
      "loss": 0.4235,
      "step": 23996
    },
    {
      "epoch": 1.101234454591345,
      "grad_norm": 0.4506818652153015,
      "learning_rate": 7.211594632704145e-06,
      "loss": 0.3131,
      "step": 23997
    },
    {
      "epoch": 1.1012803450965996,
      "grad_norm": 0.4750039875507355,
      "learning_rate": 7.211374731350843e-06,
      "loss": 0.3837,
      "step": 23998
    },
    {
      "epoch": 1.101326235601854,
      "grad_norm": 0.48315873742103577,
      "learning_rate": 7.211154824679882e-06,
      "loss": 0.3505,
      "step": 23999
    },
    {
      "epoch": 1.1013721261071083,
      "grad_norm": 0.4598916172981262,
      "learning_rate": 7.210934912691792e-06,
      "loss": 0.3284,
      "step": 24000
    },
    {
      "epoch": 1.1014180166123628,
      "grad_norm": 0.5089893341064453,
      "learning_rate": 7.2107149953871005e-06,
      "loss": 0.3594,
      "step": 24001
    },
    {
      "epoch": 1.1014639071176173,
      "grad_norm": 0.4536747932434082,
      "learning_rate": 7.210495072766337e-06,
      "loss": 0.3558,
      "step": 24002
    },
    {
      "epoch": 1.1015097976228718,
      "grad_norm": 0.4666988253593445,
      "learning_rate": 7.210275144830032e-06,
      "loss": 0.3635,
      "step": 24003
    },
    {
      "epoch": 1.1015556881281263,
      "grad_norm": 0.4292888939380646,
      "learning_rate": 7.2100552115787125e-06,
      "loss": 0.2877,
      "step": 24004
    },
    {
      "epoch": 1.1016015786333808,
      "grad_norm": 0.421105295419693,
      "learning_rate": 7.2098352730129074e-06,
      "loss": 0.3073,
      "step": 24005
    },
    {
      "epoch": 1.1016474691386353,
      "grad_norm": 0.45122382044792175,
      "learning_rate": 7.209615329133146e-06,
      "loss": 0.3216,
      "step": 24006
    },
    {
      "epoch": 1.1016933596438896,
      "grad_norm": 0.5267409086227417,
      "learning_rate": 7.209395379939956e-06,
      "loss": 0.4142,
      "step": 24007
    },
    {
      "epoch": 1.101739250149144,
      "grad_norm": 0.4449407160282135,
      "learning_rate": 7.2091754254338685e-06,
      "loss": 0.3203,
      "step": 24008
    },
    {
      "epoch": 1.1017851406543986,
      "grad_norm": 0.46412962675094604,
      "learning_rate": 7.208955465615412e-06,
      "loss": 0.3563,
      "step": 24009
    },
    {
      "epoch": 1.101831031159653,
      "grad_norm": 0.48806676268577576,
      "learning_rate": 7.208735500485113e-06,
      "loss": 0.4209,
      "step": 24010
    },
    {
      "epoch": 1.1018769216649076,
      "grad_norm": 0.42918795347213745,
      "learning_rate": 7.2085155300435035e-06,
      "loss": 0.306,
      "step": 24011
    },
    {
      "epoch": 1.101922812170162,
      "grad_norm": 0.44106441736221313,
      "learning_rate": 7.2082955542911115e-06,
      "loss": 0.3096,
      "step": 24012
    },
    {
      "epoch": 1.1019687026754164,
      "grad_norm": 0.4940788745880127,
      "learning_rate": 7.208075573228465e-06,
      "loss": 0.4145,
      "step": 24013
    },
    {
      "epoch": 1.1020145931806709,
      "grad_norm": 0.5082111954689026,
      "learning_rate": 7.207855586856094e-06,
      "loss": 0.3559,
      "step": 24014
    },
    {
      "epoch": 1.1020604836859254,
      "grad_norm": 0.4832407236099243,
      "learning_rate": 7.207635595174527e-06,
      "loss": 0.372,
      "step": 24015
    },
    {
      "epoch": 1.1021063741911798,
      "grad_norm": 0.5201005339622498,
      "learning_rate": 7.207415598184293e-06,
      "loss": 0.3522,
      "step": 24016
    },
    {
      "epoch": 1.1021522646964343,
      "grad_norm": 0.47041216492652893,
      "learning_rate": 7.207195595885921e-06,
      "loss": 0.3167,
      "step": 24017
    },
    {
      "epoch": 1.1021981552016888,
      "grad_norm": 0.46871718764305115,
      "learning_rate": 7.20697558827994e-06,
      "loss": 0.3449,
      "step": 24018
    },
    {
      "epoch": 1.1022440457069433,
      "grad_norm": 0.4304443895816803,
      "learning_rate": 7.206755575366879e-06,
      "loss": 0.3048,
      "step": 24019
    },
    {
      "epoch": 1.1022899362121976,
      "grad_norm": 0.5066967606544495,
      "learning_rate": 7.20653555714727e-06,
      "loss": 0.2805,
      "step": 24020
    },
    {
      "epoch": 1.1023358267174521,
      "grad_norm": 0.4770541191101074,
      "learning_rate": 7.206315533621636e-06,
      "loss": 0.3632,
      "step": 24021
    },
    {
      "epoch": 1.1023817172227066,
      "grad_norm": 0.5333412885665894,
      "learning_rate": 7.20609550479051e-06,
      "loss": 0.422,
      "step": 24022
    },
    {
      "epoch": 1.1024276077279611,
      "grad_norm": 0.448955237865448,
      "learning_rate": 7.205875470654422e-06,
      "loss": 0.2849,
      "step": 24023
    },
    {
      "epoch": 1.1024734982332156,
      "grad_norm": 0.507507860660553,
      "learning_rate": 7.205655431213898e-06,
      "loss": 0.3605,
      "step": 24024
    },
    {
      "epoch": 1.10251938873847,
      "grad_norm": 0.4831008315086365,
      "learning_rate": 7.2054353864694705e-06,
      "loss": 0.4347,
      "step": 24025
    },
    {
      "epoch": 1.1025652792437244,
      "grad_norm": 0.43569931387901306,
      "learning_rate": 7.205215336421667e-06,
      "loss": 0.3089,
      "step": 24026
    },
    {
      "epoch": 1.1026111697489789,
      "grad_norm": 0.5779900550842285,
      "learning_rate": 7.204995281071016e-06,
      "loss": 0.322,
      "step": 24027
    },
    {
      "epoch": 1.1026570602542334,
      "grad_norm": 0.473202645778656,
      "learning_rate": 7.204775220418046e-06,
      "loss": 0.3754,
      "step": 24028
    },
    {
      "epoch": 1.1027029507594879,
      "grad_norm": 0.4988468885421753,
      "learning_rate": 7.204555154463288e-06,
      "loss": 0.3866,
      "step": 24029
    },
    {
      "epoch": 1.1027488412647424,
      "grad_norm": 0.4621218144893646,
      "learning_rate": 7.2043350832072725e-06,
      "loss": 0.2992,
      "step": 24030
    },
    {
      "epoch": 1.1027947317699969,
      "grad_norm": 0.4547552466392517,
      "learning_rate": 7.204115006650526e-06,
      "loss": 0.3356,
      "step": 24031
    },
    {
      "epoch": 1.1028406222752511,
      "grad_norm": 0.5353094935417175,
      "learning_rate": 7.203894924793577e-06,
      "loss": 0.4615,
      "step": 24032
    },
    {
      "epoch": 1.1028865127805056,
      "grad_norm": 0.47319296002388,
      "learning_rate": 7.203674837636958e-06,
      "loss": 0.3257,
      "step": 24033
    },
    {
      "epoch": 1.1029324032857601,
      "grad_norm": 0.49430686235427856,
      "learning_rate": 7.203454745181197e-06,
      "loss": 0.4166,
      "step": 24034
    },
    {
      "epoch": 1.1029782937910146,
      "grad_norm": 0.5281789898872375,
      "learning_rate": 7.20323464742682e-06,
      "loss": 0.4042,
      "step": 24035
    },
    {
      "epoch": 1.1030241842962691,
      "grad_norm": 0.4187924563884735,
      "learning_rate": 7.203014544374361e-06,
      "loss": 0.2874,
      "step": 24036
    },
    {
      "epoch": 1.1030700748015236,
      "grad_norm": 0.48098692297935486,
      "learning_rate": 7.2027944360243476e-06,
      "loss": 0.3484,
      "step": 24037
    },
    {
      "epoch": 1.103115965306778,
      "grad_norm": 0.5104495286941528,
      "learning_rate": 7.202574322377308e-06,
      "loss": 0.372,
      "step": 24038
    },
    {
      "epoch": 1.1031618558120324,
      "grad_norm": 0.4671279489994049,
      "learning_rate": 7.202354203433772e-06,
      "loss": 0.3136,
      "step": 24039
    },
    {
      "epoch": 1.103207746317287,
      "grad_norm": 0.48503419756889343,
      "learning_rate": 7.202134079194269e-06,
      "loss": 0.4063,
      "step": 24040
    },
    {
      "epoch": 1.1032536368225414,
      "grad_norm": 0.4414476454257965,
      "learning_rate": 7.20191394965933e-06,
      "loss": 0.3167,
      "step": 24041
    },
    {
      "epoch": 1.103299527327796,
      "grad_norm": 0.49026811122894287,
      "learning_rate": 7.201693814829482e-06,
      "loss": 0.3833,
      "step": 24042
    },
    {
      "epoch": 1.1033454178330504,
      "grad_norm": 0.4644249379634857,
      "learning_rate": 7.201473674705257e-06,
      "loss": 0.3232,
      "step": 24043
    },
    {
      "epoch": 1.103391308338305,
      "grad_norm": 0.46489375829696655,
      "learning_rate": 7.20125352928718e-06,
      "loss": 0.3554,
      "step": 24044
    },
    {
      "epoch": 1.1034371988435592,
      "grad_norm": 0.47339317202568054,
      "learning_rate": 7.201033378575785e-06,
      "loss": 0.367,
      "step": 24045
    },
    {
      "epoch": 1.1034830893488137,
      "grad_norm": 0.43301594257354736,
      "learning_rate": 7.200813222571598e-06,
      "loss": 0.3114,
      "step": 24046
    },
    {
      "epoch": 1.1035289798540682,
      "grad_norm": 0.47108039259910583,
      "learning_rate": 7.20059306127515e-06,
      "loss": 0.3621,
      "step": 24047
    },
    {
      "epoch": 1.1035748703593227,
      "grad_norm": 0.4704408049583435,
      "learning_rate": 7.200372894686971e-06,
      "loss": 0.3212,
      "step": 24048
    },
    {
      "epoch": 1.1036207608645772,
      "grad_norm": 0.46565011143684387,
      "learning_rate": 7.200152722807589e-06,
      "loss": 0.3576,
      "step": 24049
    },
    {
      "epoch": 1.1036666513698317,
      "grad_norm": 0.4595431983470917,
      "learning_rate": 7.199932545637534e-06,
      "loss": 0.3365,
      "step": 24050
    },
    {
      "epoch": 1.103712541875086,
      "grad_norm": 0.45626500248908997,
      "learning_rate": 7.199712363177336e-06,
      "loss": 0.3408,
      "step": 24051
    },
    {
      "epoch": 1.1037584323803404,
      "grad_norm": 0.45966145396232605,
      "learning_rate": 7.1994921754275225e-06,
      "loss": 0.3691,
      "step": 24052
    },
    {
      "epoch": 1.103804322885595,
      "grad_norm": 0.48447123169898987,
      "learning_rate": 7.199271982388627e-06,
      "loss": 0.3685,
      "step": 24053
    },
    {
      "epoch": 1.1038502133908494,
      "grad_norm": 0.5024548768997192,
      "learning_rate": 7.199051784061175e-06,
      "loss": 0.4377,
      "step": 24054
    },
    {
      "epoch": 1.103896103896104,
      "grad_norm": 0.5267302393913269,
      "learning_rate": 7.198831580445698e-06,
      "loss": 0.4309,
      "step": 24055
    },
    {
      "epoch": 1.1039419944013584,
      "grad_norm": 0.42920824885368347,
      "learning_rate": 7.198611371542723e-06,
      "loss": 0.2545,
      "step": 24056
    },
    {
      "epoch": 1.103987884906613,
      "grad_norm": 0.424019992351532,
      "learning_rate": 7.1983911573527845e-06,
      "loss": 0.3093,
      "step": 24057
    },
    {
      "epoch": 1.1040337754118672,
      "grad_norm": 0.5126121044158936,
      "learning_rate": 7.198170937876409e-06,
      "loss": 0.4309,
      "step": 24058
    },
    {
      "epoch": 1.1040796659171217,
      "grad_norm": 0.4792507290840149,
      "learning_rate": 7.197950713114124e-06,
      "loss": 0.3912,
      "step": 24059
    },
    {
      "epoch": 1.1041255564223762,
      "grad_norm": 0.4704703688621521,
      "learning_rate": 7.197730483066463e-06,
      "loss": 0.3457,
      "step": 24060
    },
    {
      "epoch": 1.1041714469276307,
      "grad_norm": 0.4509371221065521,
      "learning_rate": 7.197510247733952e-06,
      "loss": 0.3532,
      "step": 24061
    },
    {
      "epoch": 1.1042173374328852,
      "grad_norm": 0.45793417096138,
      "learning_rate": 7.197290007117124e-06,
      "loss": 0.3332,
      "step": 24062
    },
    {
      "epoch": 1.1042632279381397,
      "grad_norm": 0.465079665184021,
      "learning_rate": 7.197069761216507e-06,
      "loss": 0.359,
      "step": 24063
    },
    {
      "epoch": 1.104309118443394,
      "grad_norm": 0.44251570105552673,
      "learning_rate": 7.19684951003263e-06,
      "loss": 0.3008,
      "step": 24064
    },
    {
      "epoch": 1.1043550089486485,
      "grad_norm": 0.5197700262069702,
      "learning_rate": 7.196629253566025e-06,
      "loss": 0.469,
      "step": 24065
    },
    {
      "epoch": 1.104400899453903,
      "grad_norm": 0.521238386631012,
      "learning_rate": 7.196408991817216e-06,
      "loss": 0.3781,
      "step": 24066
    },
    {
      "epoch": 1.1044467899591575,
      "grad_norm": 0.45557019114494324,
      "learning_rate": 7.1961887247867394e-06,
      "loss": 0.3316,
      "step": 24067
    },
    {
      "epoch": 1.104492680464412,
      "grad_norm": 0.5225608348846436,
      "learning_rate": 7.195968452475122e-06,
      "loss": 0.357,
      "step": 24068
    },
    {
      "epoch": 1.1045385709696665,
      "grad_norm": 0.6290162801742554,
      "learning_rate": 7.195748174882893e-06,
      "loss": 0.4414,
      "step": 24069
    },
    {
      "epoch": 1.104584461474921,
      "grad_norm": 0.44933953881263733,
      "learning_rate": 7.195527892010583e-06,
      "loss": 0.3821,
      "step": 24070
    },
    {
      "epoch": 1.1046303519801752,
      "grad_norm": 0.45774587988853455,
      "learning_rate": 7.195307603858721e-06,
      "loss": 0.3474,
      "step": 24071
    },
    {
      "epoch": 1.1046762424854297,
      "grad_norm": 0.46265432238578796,
      "learning_rate": 7.195087310427837e-06,
      "loss": 0.3529,
      "step": 24072
    },
    {
      "epoch": 1.1047221329906842,
      "grad_norm": 0.485806405544281,
      "learning_rate": 7.19486701171846e-06,
      "loss": 0.3329,
      "step": 24073
    },
    {
      "epoch": 1.1047680234959387,
      "grad_norm": 0.4814291000366211,
      "learning_rate": 7.194646707731122e-06,
      "loss": 0.3969,
      "step": 24074
    },
    {
      "epoch": 1.1048139140011932,
      "grad_norm": 0.42980897426605225,
      "learning_rate": 7.194426398466352e-06,
      "loss": 0.3065,
      "step": 24075
    },
    {
      "epoch": 1.1048598045064477,
      "grad_norm": 0.4809776246547699,
      "learning_rate": 7.194206083924678e-06,
      "loss": 0.3386,
      "step": 24076
    },
    {
      "epoch": 1.104905695011702,
      "grad_norm": 0.46797633171081543,
      "learning_rate": 7.19398576410663e-06,
      "loss": 0.358,
      "step": 24077
    },
    {
      "epoch": 1.1049515855169565,
      "grad_norm": 0.47154927253723145,
      "learning_rate": 7.1937654390127395e-06,
      "loss": 0.3865,
      "step": 24078
    },
    {
      "epoch": 1.104997476022211,
      "grad_norm": 0.4976094961166382,
      "learning_rate": 7.193545108643536e-06,
      "loss": 0.3465,
      "step": 24079
    },
    {
      "epoch": 1.1050433665274655,
      "grad_norm": 0.44048821926116943,
      "learning_rate": 7.19332477299955e-06,
      "loss": 0.3046,
      "step": 24080
    },
    {
      "epoch": 1.10508925703272,
      "grad_norm": 0.5054158568382263,
      "learning_rate": 7.193104432081307e-06,
      "loss": 0.4221,
      "step": 24081
    },
    {
      "epoch": 1.1051351475379745,
      "grad_norm": 0.4484591782093048,
      "learning_rate": 7.192884085889342e-06,
      "loss": 0.292,
      "step": 24082
    },
    {
      "epoch": 1.1051810380432288,
      "grad_norm": 0.4832470715045929,
      "learning_rate": 7.192663734424181e-06,
      "loss": 0.4032,
      "step": 24083
    },
    {
      "epoch": 1.1052269285484833,
      "grad_norm": 0.4845431447029114,
      "learning_rate": 7.192443377686358e-06,
      "loss": 0.385,
      "step": 24084
    },
    {
      "epoch": 1.1052728190537378,
      "grad_norm": 0.4488259255886078,
      "learning_rate": 7.1922230156763995e-06,
      "loss": 0.3151,
      "step": 24085
    },
    {
      "epoch": 1.1053187095589923,
      "grad_norm": 0.45792967081069946,
      "learning_rate": 7.192002648394838e-06,
      "loss": 0.367,
      "step": 24086
    },
    {
      "epoch": 1.1053646000642467,
      "grad_norm": 0.46047285199165344,
      "learning_rate": 7.1917822758421994e-06,
      "loss": 0.3316,
      "step": 24087
    },
    {
      "epoch": 1.1054104905695012,
      "grad_norm": 0.45247751474380493,
      "learning_rate": 7.191561898019018e-06,
      "loss": 0.3654,
      "step": 24088
    },
    {
      "epoch": 1.1054563810747555,
      "grad_norm": 0.47605806589126587,
      "learning_rate": 7.191341514925821e-06,
      "loss": 0.3939,
      "step": 24089
    },
    {
      "epoch": 1.10550227158001,
      "grad_norm": 0.44261467456817627,
      "learning_rate": 7.191121126563141e-06,
      "loss": 0.3633,
      "step": 24090
    },
    {
      "epoch": 1.1055481620852645,
      "grad_norm": 0.45486992597579956,
      "learning_rate": 7.190900732931505e-06,
      "loss": 0.3028,
      "step": 24091
    },
    {
      "epoch": 1.105594052590519,
      "grad_norm": 0.47403666377067566,
      "learning_rate": 7.190680334031443e-06,
      "loss": 0.3305,
      "step": 24092
    },
    {
      "epoch": 1.1056399430957735,
      "grad_norm": 0.4804691970348358,
      "learning_rate": 7.190459929863487e-06,
      "loss": 0.387,
      "step": 24093
    },
    {
      "epoch": 1.105685833601028,
      "grad_norm": 0.49163907766342163,
      "learning_rate": 7.190239520428167e-06,
      "loss": 0.4266,
      "step": 24094
    },
    {
      "epoch": 1.1057317241062825,
      "grad_norm": 0.44634315371513367,
      "learning_rate": 7.190019105726011e-06,
      "loss": 0.2993,
      "step": 24095
    },
    {
      "epoch": 1.1057776146115368,
      "grad_norm": 0.4560367166996002,
      "learning_rate": 7.189798685757552e-06,
      "loss": 0.3242,
      "step": 24096
    },
    {
      "epoch": 1.1058235051167913,
      "grad_norm": 0.4892953932285309,
      "learning_rate": 7.189578260523317e-06,
      "loss": 0.3797,
      "step": 24097
    },
    {
      "epoch": 1.1058693956220458,
      "grad_norm": 0.44532591104507446,
      "learning_rate": 7.189357830023838e-06,
      "loss": 0.345,
      "step": 24098
    },
    {
      "epoch": 1.1059152861273003,
      "grad_norm": 0.44525471329689026,
      "learning_rate": 7.189137394259645e-06,
      "loss": 0.3343,
      "step": 24099
    },
    {
      "epoch": 1.1059611766325548,
      "grad_norm": 0.46108290553092957,
      "learning_rate": 7.1889169532312644e-06,
      "loss": 0.3472,
      "step": 24100
    },
    {
      "epoch": 1.1060070671378093,
      "grad_norm": 0.4928763806819916,
      "learning_rate": 7.188696506939233e-06,
      "loss": 0.3852,
      "step": 24101
    },
    {
      "epoch": 1.1060529576430636,
      "grad_norm": 0.4712491035461426,
      "learning_rate": 7.188476055384075e-06,
      "loss": 0.3908,
      "step": 24102
    },
    {
      "epoch": 1.106098848148318,
      "grad_norm": 0.46864330768585205,
      "learning_rate": 7.188255598566322e-06,
      "loss": 0.3329,
      "step": 24103
    },
    {
      "epoch": 1.1061447386535725,
      "grad_norm": 0.5347345471382141,
      "learning_rate": 7.188035136486508e-06,
      "loss": 0.3856,
      "step": 24104
    },
    {
      "epoch": 1.106190629158827,
      "grad_norm": 0.49270766973495483,
      "learning_rate": 7.1878146691451575e-06,
      "loss": 0.4031,
      "step": 24105
    },
    {
      "epoch": 1.1062365196640815,
      "grad_norm": 0.48200806975364685,
      "learning_rate": 7.187594196542804e-06,
      "loss": 0.355,
      "step": 24106
    },
    {
      "epoch": 1.106282410169336,
      "grad_norm": 0.45261549949645996,
      "learning_rate": 7.1873737186799775e-06,
      "loss": 0.3193,
      "step": 24107
    },
    {
      "epoch": 1.1063283006745905,
      "grad_norm": 0.4746640920639038,
      "learning_rate": 7.187153235557208e-06,
      "loss": 0.4225,
      "step": 24108
    },
    {
      "epoch": 1.1063741911798448,
      "grad_norm": 0.5818424224853516,
      "learning_rate": 7.186932747175024e-06,
      "loss": 0.4012,
      "step": 24109
    },
    {
      "epoch": 1.1064200816850993,
      "grad_norm": 0.5333906412124634,
      "learning_rate": 7.1867122535339565e-06,
      "loss": 0.4098,
      "step": 24110
    },
    {
      "epoch": 1.1064659721903538,
      "grad_norm": 0.4457211196422577,
      "learning_rate": 7.186491754634537e-06,
      "loss": 0.2907,
      "step": 24111
    },
    {
      "epoch": 1.1065118626956083,
      "grad_norm": 0.4414770305156708,
      "learning_rate": 7.186271250477296e-06,
      "loss": 0.3271,
      "step": 24112
    },
    {
      "epoch": 1.1065577532008628,
      "grad_norm": 0.5181454420089722,
      "learning_rate": 7.186050741062762e-06,
      "loss": 0.4496,
      "step": 24113
    },
    {
      "epoch": 1.1066036437061173,
      "grad_norm": 0.5419386625289917,
      "learning_rate": 7.185830226391465e-06,
      "loss": 0.4375,
      "step": 24114
    },
    {
      "epoch": 1.1066495342113716,
      "grad_norm": 0.4356323480606079,
      "learning_rate": 7.185609706463937e-06,
      "loss": 0.3061,
      "step": 24115
    },
    {
      "epoch": 1.106695424716626,
      "grad_norm": 0.4951098561286926,
      "learning_rate": 7.185389181280706e-06,
      "loss": 0.3879,
      "step": 24116
    },
    {
      "epoch": 1.1067413152218806,
      "grad_norm": 0.43494734168052673,
      "learning_rate": 7.185168650842305e-06,
      "loss": 0.3128,
      "step": 24117
    },
    {
      "epoch": 1.106787205727135,
      "grad_norm": 0.44640690088272095,
      "learning_rate": 7.184948115149264e-06,
      "loss": 0.2986,
      "step": 24118
    },
    {
      "epoch": 1.1068330962323896,
      "grad_norm": 0.4308014512062073,
      "learning_rate": 7.184727574202111e-06,
      "loss": 0.2916,
      "step": 24119
    },
    {
      "epoch": 1.106878986737644,
      "grad_norm": 0.5008946657180786,
      "learning_rate": 7.184507028001377e-06,
      "loss": 0.4116,
      "step": 24120
    },
    {
      "epoch": 1.1069248772428983,
      "grad_norm": 0.5626621842384338,
      "learning_rate": 7.184286476547594e-06,
      "loss": 0.4505,
      "step": 24121
    },
    {
      "epoch": 1.1069707677481528,
      "grad_norm": 0.505445659160614,
      "learning_rate": 7.184065919841292e-06,
      "loss": 0.4631,
      "step": 24122
    },
    {
      "epoch": 1.1070166582534073,
      "grad_norm": 0.46523237228393555,
      "learning_rate": 7.183845357883e-06,
      "loss": 0.3401,
      "step": 24123
    },
    {
      "epoch": 1.1070625487586618,
      "grad_norm": 0.46505486965179443,
      "learning_rate": 7.183624790673249e-06,
      "loss": 0.3589,
      "step": 24124
    },
    {
      "epoch": 1.1071084392639163,
      "grad_norm": 0.48532193899154663,
      "learning_rate": 7.183404218212571e-06,
      "loss": 0.432,
      "step": 24125
    },
    {
      "epoch": 1.1071543297691708,
      "grad_norm": 0.4821535050868988,
      "learning_rate": 7.1831836405014935e-06,
      "loss": 0.3712,
      "step": 24126
    },
    {
      "epoch": 1.107200220274425,
      "grad_norm": 0.4672161638736725,
      "learning_rate": 7.182963057540548e-06,
      "loss": 0.3475,
      "step": 24127
    },
    {
      "epoch": 1.1072461107796796,
      "grad_norm": 0.4045127332210541,
      "learning_rate": 7.182742469330266e-06,
      "loss": 0.305,
      "step": 24128
    },
    {
      "epoch": 1.107292001284934,
      "grad_norm": 0.46311289072036743,
      "learning_rate": 7.182521875871178e-06,
      "loss": 0.3443,
      "step": 24129
    },
    {
      "epoch": 1.1073378917901886,
      "grad_norm": 0.44986966252326965,
      "learning_rate": 7.182301277163814e-06,
      "loss": 0.3521,
      "step": 24130
    },
    {
      "epoch": 1.107383782295443,
      "grad_norm": 0.4785291850566864,
      "learning_rate": 7.182080673208702e-06,
      "loss": 0.3792,
      "step": 24131
    },
    {
      "epoch": 1.1074296728006976,
      "grad_norm": 0.43488848209381104,
      "learning_rate": 7.181860064006376e-06,
      "loss": 0.3072,
      "step": 24132
    },
    {
      "epoch": 1.107475563305952,
      "grad_norm": 0.4507552981376648,
      "learning_rate": 7.181639449557365e-06,
      "loss": 0.3403,
      "step": 24133
    },
    {
      "epoch": 1.1075214538112064,
      "grad_norm": 0.46561646461486816,
      "learning_rate": 7.1814188298622e-06,
      "loss": 0.3187,
      "step": 24134
    },
    {
      "epoch": 1.1075673443164609,
      "grad_norm": 0.5131621956825256,
      "learning_rate": 7.1811982049214116e-06,
      "loss": 0.4175,
      "step": 24135
    },
    {
      "epoch": 1.1076132348217154,
      "grad_norm": 0.450484037399292,
      "learning_rate": 7.180977574735528e-06,
      "loss": 0.3298,
      "step": 24136
    },
    {
      "epoch": 1.1076591253269699,
      "grad_norm": 0.4876645803451538,
      "learning_rate": 7.180756939305083e-06,
      "loss": 0.3874,
      "step": 24137
    },
    {
      "epoch": 1.1077050158322244,
      "grad_norm": 0.4379865825176239,
      "learning_rate": 7.1805362986306045e-06,
      "loss": 0.3272,
      "step": 24138
    },
    {
      "epoch": 1.1077509063374789,
      "grad_norm": 0.45753467082977295,
      "learning_rate": 7.180315652712628e-06,
      "loss": 0.2913,
      "step": 24139
    },
    {
      "epoch": 1.1077967968427331,
      "grad_norm": 0.43448999524116516,
      "learning_rate": 7.180095001551678e-06,
      "loss": 0.3257,
      "step": 24140
    },
    {
      "epoch": 1.1078426873479876,
      "grad_norm": 0.43923598527908325,
      "learning_rate": 7.179874345148287e-06,
      "loss": 0.3149,
      "step": 24141
    },
    {
      "epoch": 1.1078885778532421,
      "grad_norm": 0.45992422103881836,
      "learning_rate": 7.1796536835029875e-06,
      "loss": 0.2937,
      "step": 24142
    },
    {
      "epoch": 1.1079344683584966,
      "grad_norm": 0.47843483090400696,
      "learning_rate": 7.179433016616308e-06,
      "loss": 0.352,
      "step": 24143
    },
    {
      "epoch": 1.1079803588637511,
      "grad_norm": 0.4803861677646637,
      "learning_rate": 7.179212344488779e-06,
      "loss": 0.3607,
      "step": 24144
    },
    {
      "epoch": 1.1080262493690056,
      "grad_norm": 0.4707183241844177,
      "learning_rate": 7.178991667120935e-06,
      "loss": 0.3673,
      "step": 24145
    },
    {
      "epoch": 1.1080721398742601,
      "grad_norm": 0.46289199590682983,
      "learning_rate": 7.1787709845133015e-06,
      "loss": 0.3816,
      "step": 24146
    },
    {
      "epoch": 1.1081180303795144,
      "grad_norm": 0.48825475573539734,
      "learning_rate": 7.178550296666411e-06,
      "loss": 0.3672,
      "step": 24147
    },
    {
      "epoch": 1.108163920884769,
      "grad_norm": 0.44593343138694763,
      "learning_rate": 7.1783296035807955e-06,
      "loss": 0.3036,
      "step": 24148
    },
    {
      "epoch": 1.1082098113900234,
      "grad_norm": 0.45376941561698914,
      "learning_rate": 7.178108905256984e-06,
      "loss": 0.334,
      "step": 24149
    },
    {
      "epoch": 1.108255701895278,
      "grad_norm": 0.4502517580986023,
      "learning_rate": 7.17788820169551e-06,
      "loss": 0.3352,
      "step": 24150
    },
    {
      "epoch": 1.1083015924005324,
      "grad_norm": 0.4482055604457855,
      "learning_rate": 7.177667492896901e-06,
      "loss": 0.3619,
      "step": 24151
    },
    {
      "epoch": 1.1083474829057869,
      "grad_norm": 0.5047576427459717,
      "learning_rate": 7.1774467788616885e-06,
      "loss": 0.4286,
      "step": 24152
    },
    {
      "epoch": 1.1083933734110412,
      "grad_norm": 0.45702528953552246,
      "learning_rate": 7.177226059590405e-06,
      "loss": 0.3513,
      "step": 24153
    },
    {
      "epoch": 1.1084392639162957,
      "grad_norm": 0.4475311040878296,
      "learning_rate": 7.177005335083579e-06,
      "loss": 0.3348,
      "step": 24154
    },
    {
      "epoch": 1.1084851544215502,
      "grad_norm": 0.46700724959373474,
      "learning_rate": 7.176784605341743e-06,
      "loss": 0.3786,
      "step": 24155
    },
    {
      "epoch": 1.1085310449268047,
      "grad_norm": 0.530242919921875,
      "learning_rate": 7.176563870365426e-06,
      "loss": 0.4173,
      "step": 24156
    },
    {
      "epoch": 1.1085769354320592,
      "grad_norm": 0.44964393973350525,
      "learning_rate": 7.17634313015516e-06,
      "loss": 0.2955,
      "step": 24157
    },
    {
      "epoch": 1.1086228259373136,
      "grad_norm": 0.5270382761955261,
      "learning_rate": 7.176122384711476e-06,
      "loss": 0.4128,
      "step": 24158
    },
    {
      "epoch": 1.1086687164425681,
      "grad_norm": 0.45300212502479553,
      "learning_rate": 7.175901634034905e-06,
      "loss": 0.3187,
      "step": 24159
    },
    {
      "epoch": 1.1087146069478224,
      "grad_norm": 0.45949307084083557,
      "learning_rate": 7.175680878125976e-06,
      "loss": 0.3442,
      "step": 24160
    },
    {
      "epoch": 1.108760497453077,
      "grad_norm": 0.4816446602344513,
      "learning_rate": 7.175460116985222e-06,
      "loss": 0.4225,
      "step": 24161
    },
    {
      "epoch": 1.1088063879583314,
      "grad_norm": 0.4883665144443512,
      "learning_rate": 7.175239350613173e-06,
      "loss": 0.3851,
      "step": 24162
    },
    {
      "epoch": 1.108852278463586,
      "grad_norm": 0.4621060788631439,
      "learning_rate": 7.17501857901036e-06,
      "loss": 0.3851,
      "step": 24163
    },
    {
      "epoch": 1.1088981689688404,
      "grad_norm": 0.4945509433746338,
      "learning_rate": 7.174797802177313e-06,
      "loss": 0.4239,
      "step": 24164
    },
    {
      "epoch": 1.108944059474095,
      "grad_norm": 0.45855578780174255,
      "learning_rate": 7.1745770201145646e-06,
      "loss": 0.3343,
      "step": 24165
    },
    {
      "epoch": 1.1089899499793492,
      "grad_norm": 0.4669126570224762,
      "learning_rate": 7.174356232822644e-06,
      "loss": 0.3528,
      "step": 24166
    },
    {
      "epoch": 1.1090358404846037,
      "grad_norm": 0.43554508686065674,
      "learning_rate": 7.1741354403020844e-06,
      "loss": 0.3546,
      "step": 24167
    },
    {
      "epoch": 1.1090817309898582,
      "grad_norm": 0.4729241132736206,
      "learning_rate": 7.1739146425534145e-06,
      "loss": 0.3945,
      "step": 24168
    },
    {
      "epoch": 1.1091276214951127,
      "grad_norm": 0.4337855577468872,
      "learning_rate": 7.173693839577165e-06,
      "loss": 0.31,
      "step": 24169
    },
    {
      "epoch": 1.1091735120003672,
      "grad_norm": 0.4468729794025421,
      "learning_rate": 7.173473031373869e-06,
      "loss": 0.3173,
      "step": 24170
    },
    {
      "epoch": 1.1092194025056217,
      "grad_norm": 0.46156302094459534,
      "learning_rate": 7.173252217944056e-06,
      "loss": 0.3144,
      "step": 24171
    },
    {
      "epoch": 1.109265293010876,
      "grad_norm": 0.48598507046699524,
      "learning_rate": 7.173031399288257e-06,
      "loss": 0.3981,
      "step": 24172
    },
    {
      "epoch": 1.1093111835161305,
      "grad_norm": 0.45429477095603943,
      "learning_rate": 7.172810575407003e-06,
      "loss": 0.3085,
      "step": 24173
    },
    {
      "epoch": 1.109357074021385,
      "grad_norm": 0.4479900300502777,
      "learning_rate": 7.172589746300827e-06,
      "loss": 0.3182,
      "step": 24174
    },
    {
      "epoch": 1.1094029645266394,
      "grad_norm": 0.4394559860229492,
      "learning_rate": 7.172368911970256e-06,
      "loss": 0.2955,
      "step": 24175
    },
    {
      "epoch": 1.109448855031894,
      "grad_norm": 0.4659448266029358,
      "learning_rate": 7.172148072415824e-06,
      "loss": 0.391,
      "step": 24176
    },
    {
      "epoch": 1.1094947455371484,
      "grad_norm": 0.42809024453163147,
      "learning_rate": 7.171927227638063e-06,
      "loss": 0.3169,
      "step": 24177
    },
    {
      "epoch": 1.1095406360424027,
      "grad_norm": 0.45912107825279236,
      "learning_rate": 7.171706377637502e-06,
      "loss": 0.3837,
      "step": 24178
    },
    {
      "epoch": 1.1095865265476572,
      "grad_norm": 0.493915319442749,
      "learning_rate": 7.171485522414671e-06,
      "loss": 0.3564,
      "step": 24179
    },
    {
      "epoch": 1.1096324170529117,
      "grad_norm": 0.44918495416641235,
      "learning_rate": 7.171264661970103e-06,
      "loss": 0.3425,
      "step": 24180
    },
    {
      "epoch": 1.1096783075581662,
      "grad_norm": 0.445166677236557,
      "learning_rate": 7.171043796304331e-06,
      "loss": 0.3278,
      "step": 24181
    },
    {
      "epoch": 1.1097241980634207,
      "grad_norm": 0.44932299852371216,
      "learning_rate": 7.170822925417882e-06,
      "loss": 0.3697,
      "step": 24182
    },
    {
      "epoch": 1.1097700885686752,
      "grad_norm": 0.4315730929374695,
      "learning_rate": 7.170602049311289e-06,
      "loss": 0.2813,
      "step": 24183
    },
    {
      "epoch": 1.1098159790739297,
      "grad_norm": 0.4613051116466522,
      "learning_rate": 7.170381167985084e-06,
      "loss": 0.361,
      "step": 24184
    },
    {
      "epoch": 1.109861869579184,
      "grad_norm": 0.49513137340545654,
      "learning_rate": 7.170160281439796e-06,
      "loss": 0.3918,
      "step": 24185
    },
    {
      "epoch": 1.1099077600844385,
      "grad_norm": 0.46481776237487793,
      "learning_rate": 7.169939389675957e-06,
      "loss": 0.3536,
      "step": 24186
    },
    {
      "epoch": 1.109953650589693,
      "grad_norm": 0.4210161864757538,
      "learning_rate": 7.169718492694101e-06,
      "loss": 0.272,
      "step": 24187
    },
    {
      "epoch": 1.1099995410949475,
      "grad_norm": 0.47596031427383423,
      "learning_rate": 7.169497590494756e-06,
      "loss": 0.4252,
      "step": 24188
    },
    {
      "epoch": 1.110045431600202,
      "grad_norm": 0.4196113348007202,
      "learning_rate": 7.169276683078453e-06,
      "loss": 0.2669,
      "step": 24189
    },
    {
      "epoch": 1.1100913221054565,
      "grad_norm": 0.49251946806907654,
      "learning_rate": 7.169055770445726e-06,
      "loss": 0.4089,
      "step": 24190
    },
    {
      "epoch": 1.1101372126107107,
      "grad_norm": 0.4743382930755615,
      "learning_rate": 7.168834852597102e-06,
      "loss": 0.389,
      "step": 24191
    },
    {
      "epoch": 1.1101831031159652,
      "grad_norm": 0.5145764350891113,
      "learning_rate": 7.168613929533117e-06,
      "loss": 0.4567,
      "step": 24192
    },
    {
      "epoch": 1.1102289936212197,
      "grad_norm": 0.4943943917751312,
      "learning_rate": 7.168393001254298e-06,
      "loss": 0.396,
      "step": 24193
    },
    {
      "epoch": 1.1102748841264742,
      "grad_norm": 0.43554216623306274,
      "learning_rate": 7.16817206776118e-06,
      "loss": 0.3004,
      "step": 24194
    },
    {
      "epoch": 1.1103207746317287,
      "grad_norm": 0.48390504717826843,
      "learning_rate": 7.167951129054291e-06,
      "loss": 0.3369,
      "step": 24195
    },
    {
      "epoch": 1.1103666651369832,
      "grad_norm": 0.4849126935005188,
      "learning_rate": 7.1677301851341655e-06,
      "loss": 0.3966,
      "step": 24196
    },
    {
      "epoch": 1.1104125556422377,
      "grad_norm": 0.48217445611953735,
      "learning_rate": 7.167509236001331e-06,
      "loss": 0.3979,
      "step": 24197
    },
    {
      "epoch": 1.110458446147492,
      "grad_norm": 0.4590473175048828,
      "learning_rate": 7.167288281656322e-06,
      "loss": 0.344,
      "step": 24198
    },
    {
      "epoch": 1.1105043366527465,
      "grad_norm": 0.45626410841941833,
      "learning_rate": 7.167067322099669e-06,
      "loss": 0.3582,
      "step": 24199
    },
    {
      "epoch": 1.110550227158001,
      "grad_norm": 0.44711849093437195,
      "learning_rate": 7.1668463573319015e-06,
      "loss": 0.3363,
      "step": 24200
    },
    {
      "epoch": 1.1105961176632555,
      "grad_norm": 0.4840066134929657,
      "learning_rate": 7.166625387353554e-06,
      "loss": 0.3515,
      "step": 24201
    },
    {
      "epoch": 1.11064200816851,
      "grad_norm": 0.4763987064361572,
      "learning_rate": 7.166404412165155e-06,
      "loss": 0.3869,
      "step": 24202
    },
    {
      "epoch": 1.1106878986737645,
      "grad_norm": 0.4718271791934967,
      "learning_rate": 7.166183431767237e-06,
      "loss": 0.3605,
      "step": 24203
    },
    {
      "epoch": 1.1107337891790188,
      "grad_norm": 0.42942309379577637,
      "learning_rate": 7.165962446160332e-06,
      "loss": 0.3222,
      "step": 24204
    },
    {
      "epoch": 1.1107796796842733,
      "grad_norm": 0.4857894480228424,
      "learning_rate": 7.165741455344974e-06,
      "loss": 0.3952,
      "step": 24205
    },
    {
      "epoch": 1.1108255701895278,
      "grad_norm": 0.5267897248268127,
      "learning_rate": 7.165520459321686e-06,
      "loss": 0.395,
      "step": 24206
    },
    {
      "epoch": 1.1108714606947823,
      "grad_norm": 0.4688046872615814,
      "learning_rate": 7.165299458091009e-06,
      "loss": 0.3728,
      "step": 24207
    },
    {
      "epoch": 1.1109173512000368,
      "grad_norm": 0.46491506695747375,
      "learning_rate": 7.165078451653467e-06,
      "loss": 0.365,
      "step": 24208
    },
    {
      "epoch": 1.1109632417052913,
      "grad_norm": 0.4369814693927765,
      "learning_rate": 7.164857440009597e-06,
      "loss": 0.3479,
      "step": 24209
    },
    {
      "epoch": 1.1110091322105455,
      "grad_norm": 0.4969015121459961,
      "learning_rate": 7.164636423159927e-06,
      "loss": 0.3753,
      "step": 24210
    },
    {
      "epoch": 1.1110550227158,
      "grad_norm": 0.4583773910999298,
      "learning_rate": 7.16441540110499e-06,
      "loss": 0.3717,
      "step": 24211
    },
    {
      "epoch": 1.1111009132210545,
      "grad_norm": 0.44166794419288635,
      "learning_rate": 7.164194373845318e-06,
      "loss": 0.2825,
      "step": 24212
    },
    {
      "epoch": 1.111146803726309,
      "grad_norm": 0.5372514724731445,
      "learning_rate": 7.163973341381439e-06,
      "loss": 0.4226,
      "step": 24213
    },
    {
      "epoch": 1.1111926942315635,
      "grad_norm": 0.4493069052696228,
      "learning_rate": 7.163752303713889e-06,
      "loss": 0.288,
      "step": 24214
    },
    {
      "epoch": 1.111238584736818,
      "grad_norm": 0.47101372480392456,
      "learning_rate": 7.163531260843197e-06,
      "loss": 0.3759,
      "step": 24215
    },
    {
      "epoch": 1.1112844752420723,
      "grad_norm": 0.45408353209495544,
      "learning_rate": 7.163310212769896e-06,
      "loss": 0.3162,
      "step": 24216
    },
    {
      "epoch": 1.1113303657473268,
      "grad_norm": 0.46379348635673523,
      "learning_rate": 7.163089159494515e-06,
      "loss": 0.3485,
      "step": 24217
    },
    {
      "epoch": 1.1113762562525813,
      "grad_norm": 0.46426066756248474,
      "learning_rate": 7.162868101017589e-06,
      "loss": 0.3356,
      "step": 24218
    },
    {
      "epoch": 1.1114221467578358,
      "grad_norm": 0.43641340732574463,
      "learning_rate": 7.162647037339646e-06,
      "loss": 0.3084,
      "step": 24219
    },
    {
      "epoch": 1.1114680372630903,
      "grad_norm": 0.4933439791202545,
      "learning_rate": 7.162425968461221e-06,
      "loss": 0.3932,
      "step": 24220
    },
    {
      "epoch": 1.1115139277683448,
      "grad_norm": 0.4686475694179535,
      "learning_rate": 7.1622048943828445e-06,
      "loss": 0.341,
      "step": 24221
    },
    {
      "epoch": 1.1115598182735993,
      "grad_norm": 0.4840691387653351,
      "learning_rate": 7.161983815105046e-06,
      "loss": 0.3609,
      "step": 24222
    },
    {
      "epoch": 1.1116057087788536,
      "grad_norm": 0.47255459427833557,
      "learning_rate": 7.1617627306283586e-06,
      "loss": 0.4029,
      "step": 24223
    },
    {
      "epoch": 1.111651599284108,
      "grad_norm": 0.5708515644073486,
      "learning_rate": 7.1615416409533155e-06,
      "loss": 0.4554,
      "step": 24224
    },
    {
      "epoch": 1.1116974897893626,
      "grad_norm": 0.471022367477417,
      "learning_rate": 7.161320546080445e-06,
      "loss": 0.3157,
      "step": 24225
    },
    {
      "epoch": 1.111743380294617,
      "grad_norm": 0.6310541033744812,
      "learning_rate": 7.1610994460102845e-06,
      "loss": 0.4361,
      "step": 24226
    },
    {
      "epoch": 1.1117892707998716,
      "grad_norm": 0.4477960765361786,
      "learning_rate": 7.160878340743359e-06,
      "loss": 0.3076,
      "step": 24227
    },
    {
      "epoch": 1.111835161305126,
      "grad_norm": 0.5310072898864746,
      "learning_rate": 7.1606572302802036e-06,
      "loss": 0.367,
      "step": 24228
    },
    {
      "epoch": 1.1118810518103803,
      "grad_norm": 0.5096598863601685,
      "learning_rate": 7.16043611462135e-06,
      "loss": 0.4059,
      "step": 24229
    },
    {
      "epoch": 1.1119269423156348,
      "grad_norm": 0.47221890091896057,
      "learning_rate": 7.160214993767328e-06,
      "loss": 0.3598,
      "step": 24230
    },
    {
      "epoch": 1.1119728328208893,
      "grad_norm": 0.5022576451301575,
      "learning_rate": 7.159993867718671e-06,
      "loss": 0.4224,
      "step": 24231
    },
    {
      "epoch": 1.1120187233261438,
      "grad_norm": 0.49132218956947327,
      "learning_rate": 7.159772736475913e-06,
      "loss": 0.4014,
      "step": 24232
    },
    {
      "epoch": 1.1120646138313983,
      "grad_norm": 0.5183481574058533,
      "learning_rate": 7.159551600039581e-06,
      "loss": 0.3684,
      "step": 24233
    },
    {
      "epoch": 1.1121105043366528,
      "grad_norm": 0.47301411628723145,
      "learning_rate": 7.159330458410207e-06,
      "loss": 0.3381,
      "step": 24234
    },
    {
      "epoch": 1.1121563948419073,
      "grad_norm": 0.43489307165145874,
      "learning_rate": 7.159109311588328e-06,
      "loss": 0.3124,
      "step": 24235
    },
    {
      "epoch": 1.1122022853471616,
      "grad_norm": 0.4650524854660034,
      "learning_rate": 7.158888159574472e-06,
      "loss": 0.3674,
      "step": 24236
    },
    {
      "epoch": 1.112248175852416,
      "grad_norm": 0.47108009457588196,
      "learning_rate": 7.158667002369171e-06,
      "loss": 0.3966,
      "step": 24237
    },
    {
      "epoch": 1.1122940663576706,
      "grad_norm": 0.41447097063064575,
      "learning_rate": 7.158445839972957e-06,
      "loss": 0.2928,
      "step": 24238
    },
    {
      "epoch": 1.112339956862925,
      "grad_norm": 0.5000222325325012,
      "learning_rate": 7.158224672386362e-06,
      "loss": 0.3827,
      "step": 24239
    },
    {
      "epoch": 1.1123858473681796,
      "grad_norm": 0.47842738032341003,
      "learning_rate": 7.158003499609918e-06,
      "loss": 0.3371,
      "step": 24240
    },
    {
      "epoch": 1.112431737873434,
      "grad_norm": 0.5362964272499084,
      "learning_rate": 7.157782321644157e-06,
      "loss": 0.3184,
      "step": 24241
    },
    {
      "epoch": 1.1124776283786884,
      "grad_norm": 0.5120749473571777,
      "learning_rate": 7.157561138489608e-06,
      "loss": 0.3852,
      "step": 24242
    },
    {
      "epoch": 1.1125235188839429,
      "grad_norm": 0.4918997883796692,
      "learning_rate": 7.15733995014681e-06,
      "loss": 0.4268,
      "step": 24243
    },
    {
      "epoch": 1.1125694093891974,
      "grad_norm": 0.49919113516807556,
      "learning_rate": 7.157118756616287e-06,
      "loss": 0.401,
      "step": 24244
    },
    {
      "epoch": 1.1126152998944518,
      "grad_norm": 0.48568886518478394,
      "learning_rate": 7.1568975578985744e-06,
      "loss": 0.4067,
      "step": 24245
    },
    {
      "epoch": 1.1126611903997063,
      "grad_norm": 0.4275054633617401,
      "learning_rate": 7.1566763539942055e-06,
      "loss": 0.2913,
      "step": 24246
    },
    {
      "epoch": 1.1127070809049608,
      "grad_norm": 0.49005022644996643,
      "learning_rate": 7.156455144903709e-06,
      "loss": 0.3789,
      "step": 24247
    },
    {
      "epoch": 1.1127529714102153,
      "grad_norm": 0.4599388837814331,
      "learning_rate": 7.156233930627619e-06,
      "loss": 0.3154,
      "step": 24248
    },
    {
      "epoch": 1.1127988619154696,
      "grad_norm": 0.4920576214790344,
      "learning_rate": 7.156012711166467e-06,
      "loss": 0.4163,
      "step": 24249
    },
    {
      "epoch": 1.1128447524207241,
      "grad_norm": 0.5525829792022705,
      "learning_rate": 7.155791486520785e-06,
      "loss": 0.3983,
      "step": 24250
    },
    {
      "epoch": 1.1128906429259786,
      "grad_norm": 0.4622592329978943,
      "learning_rate": 7.155570256691105e-06,
      "loss": 0.3388,
      "step": 24251
    },
    {
      "epoch": 1.112936533431233,
      "grad_norm": 0.47792983055114746,
      "learning_rate": 7.155349021677957e-06,
      "loss": 0.3851,
      "step": 24252
    },
    {
      "epoch": 1.1129824239364876,
      "grad_norm": 0.5089048147201538,
      "learning_rate": 7.155127781481876e-06,
      "loss": 0.4108,
      "step": 24253
    },
    {
      "epoch": 1.113028314441742,
      "grad_norm": 0.4869705140590668,
      "learning_rate": 7.154906536103395e-06,
      "loss": 0.4004,
      "step": 24254
    },
    {
      "epoch": 1.1130742049469964,
      "grad_norm": 0.4836026728153229,
      "learning_rate": 7.154685285543042e-06,
      "loss": 0.376,
      "step": 24255
    },
    {
      "epoch": 1.1131200954522509,
      "grad_norm": 0.5027388334274292,
      "learning_rate": 7.154464029801351e-06,
      "loss": 0.384,
      "step": 24256
    },
    {
      "epoch": 1.1131659859575054,
      "grad_norm": 0.47725412249565125,
      "learning_rate": 7.1542427688788545e-06,
      "loss": 0.3186,
      "step": 24257
    },
    {
      "epoch": 1.1132118764627599,
      "grad_norm": 0.4454929232597351,
      "learning_rate": 7.154021502776083e-06,
      "loss": 0.3131,
      "step": 24258
    },
    {
      "epoch": 1.1132577669680144,
      "grad_norm": 0.47490495443344116,
      "learning_rate": 7.153800231493571e-06,
      "loss": 0.3788,
      "step": 24259
    },
    {
      "epoch": 1.1133036574732689,
      "grad_norm": 0.46975889801979065,
      "learning_rate": 7.153578955031849e-06,
      "loss": 0.3333,
      "step": 24260
    },
    {
      "epoch": 1.1133495479785231,
      "grad_norm": 0.4466855227947235,
      "learning_rate": 7.153357673391448e-06,
      "loss": 0.3233,
      "step": 24261
    },
    {
      "epoch": 1.1133954384837776,
      "grad_norm": 0.46465012431144714,
      "learning_rate": 7.153136386572901e-06,
      "loss": 0.3881,
      "step": 24262
    },
    {
      "epoch": 1.1134413289890321,
      "grad_norm": 0.46011868119239807,
      "learning_rate": 7.152915094576743e-06,
      "loss": 0.3487,
      "step": 24263
    },
    {
      "epoch": 1.1134872194942866,
      "grad_norm": 0.4497684836387634,
      "learning_rate": 7.1526937974035015e-06,
      "loss": 0.2926,
      "step": 24264
    },
    {
      "epoch": 1.1135331099995411,
      "grad_norm": 0.5106773376464844,
      "learning_rate": 7.1524724950537125e-06,
      "loss": 0.376,
      "step": 24265
    },
    {
      "epoch": 1.1135790005047956,
      "grad_norm": 0.4786224961280823,
      "learning_rate": 7.152251187527906e-06,
      "loss": 0.3773,
      "step": 24266
    },
    {
      "epoch": 1.11362489101005,
      "grad_norm": 0.45936697721481323,
      "learning_rate": 7.1520298748266136e-06,
      "loss": 0.3348,
      "step": 24267
    },
    {
      "epoch": 1.1136707815153044,
      "grad_norm": 0.46732693910598755,
      "learning_rate": 7.151808556950369e-06,
      "loss": 0.3195,
      "step": 24268
    },
    {
      "epoch": 1.113716672020559,
      "grad_norm": 0.4996951222419739,
      "learning_rate": 7.151587233899705e-06,
      "loss": 0.3799,
      "step": 24269
    },
    {
      "epoch": 1.1137625625258134,
      "grad_norm": 0.43901556730270386,
      "learning_rate": 7.1513659056751515e-06,
      "loss": 0.3169,
      "step": 24270
    },
    {
      "epoch": 1.113808453031068,
      "grad_norm": 0.4447793960571289,
      "learning_rate": 7.151144572277242e-06,
      "loss": 0.3566,
      "step": 24271
    },
    {
      "epoch": 1.1138543435363224,
      "grad_norm": 0.5076403021812439,
      "learning_rate": 7.1509232337065105e-06,
      "loss": 0.396,
      "step": 24272
    },
    {
      "epoch": 1.113900234041577,
      "grad_norm": 0.49268388748168945,
      "learning_rate": 7.150701889963485e-06,
      "loss": 0.372,
      "step": 24273
    },
    {
      "epoch": 1.1139461245468312,
      "grad_norm": 0.463837593793869,
      "learning_rate": 7.150480541048702e-06,
      "loss": 0.3761,
      "step": 24274
    },
    {
      "epoch": 1.1139920150520857,
      "grad_norm": 0.43185147643089294,
      "learning_rate": 7.150259186962691e-06,
      "loss": 0.2797,
      "step": 24275
    },
    {
      "epoch": 1.1140379055573402,
      "grad_norm": 0.43756139278411865,
      "learning_rate": 7.150037827705987e-06,
      "loss": 0.2771,
      "step": 24276
    },
    {
      "epoch": 1.1140837960625947,
      "grad_norm": 0.4787544906139374,
      "learning_rate": 7.14981646327912e-06,
      "loss": 0.3822,
      "step": 24277
    },
    {
      "epoch": 1.1141296865678492,
      "grad_norm": 0.5242376327514648,
      "learning_rate": 7.1495950936826205e-06,
      "loss": 0.4725,
      "step": 24278
    },
    {
      "epoch": 1.1141755770731037,
      "grad_norm": 0.472313791513443,
      "learning_rate": 7.149373718917025e-06,
      "loss": 0.3547,
      "step": 24279
    },
    {
      "epoch": 1.114221467578358,
      "grad_norm": 0.5025116801261902,
      "learning_rate": 7.149152338982865e-06,
      "loss": 0.386,
      "step": 24280
    },
    {
      "epoch": 1.1142673580836124,
      "grad_norm": 0.46358856558799744,
      "learning_rate": 7.1489309538806705e-06,
      "loss": 0.306,
      "step": 24281
    },
    {
      "epoch": 1.114313248588867,
      "grad_norm": 0.482504665851593,
      "learning_rate": 7.148709563610976e-06,
      "loss": 0.3413,
      "step": 24282
    },
    {
      "epoch": 1.1143591390941214,
      "grad_norm": 0.45809096097946167,
      "learning_rate": 7.148488168174313e-06,
      "loss": 0.3472,
      "step": 24283
    },
    {
      "epoch": 1.114405029599376,
      "grad_norm": 0.5431184768676758,
      "learning_rate": 7.148266767571214e-06,
      "loss": 0.3429,
      "step": 24284
    },
    {
      "epoch": 1.1144509201046304,
      "grad_norm": 0.48125937581062317,
      "learning_rate": 7.14804536180221e-06,
      "loss": 0.3708,
      "step": 24285
    },
    {
      "epoch": 1.114496810609885,
      "grad_norm": 0.49789756536483765,
      "learning_rate": 7.147823950867837e-06,
      "loss": 0.4154,
      "step": 24286
    },
    {
      "epoch": 1.1145427011151392,
      "grad_norm": 0.49578002095222473,
      "learning_rate": 7.147602534768625e-06,
      "loss": 0.3758,
      "step": 24287
    },
    {
      "epoch": 1.1145885916203937,
      "grad_norm": 0.4344961643218994,
      "learning_rate": 7.147381113505105e-06,
      "loss": 0.3056,
      "step": 24288
    },
    {
      "epoch": 1.1146344821256482,
      "grad_norm": 0.5006224513053894,
      "learning_rate": 7.1471596870778135e-06,
      "loss": 0.4223,
      "step": 24289
    },
    {
      "epoch": 1.1146803726309027,
      "grad_norm": 0.4563644826412201,
      "learning_rate": 7.146938255487279e-06,
      "loss": 0.3149,
      "step": 24290
    },
    {
      "epoch": 1.1147262631361572,
      "grad_norm": 0.4680328667163849,
      "learning_rate": 7.1467168187340366e-06,
      "loss": 0.3596,
      "step": 24291
    },
    {
      "epoch": 1.1147721536414117,
      "grad_norm": 0.47042617201805115,
      "learning_rate": 7.146495376818618e-06,
      "loss": 0.3773,
      "step": 24292
    },
    {
      "epoch": 1.114818044146666,
      "grad_norm": 0.5124953389167786,
      "learning_rate": 7.146273929741554e-06,
      "loss": 0.4377,
      "step": 24293
    },
    {
      "epoch": 1.1148639346519205,
      "grad_norm": 0.5205598473548889,
      "learning_rate": 7.14605247750338e-06,
      "loss": 0.3475,
      "step": 24294
    },
    {
      "epoch": 1.114909825157175,
      "grad_norm": 0.443511962890625,
      "learning_rate": 7.145831020104625e-06,
      "loss": 0.3064,
      "step": 24295
    },
    {
      "epoch": 1.1149557156624295,
      "grad_norm": 0.4829126000404358,
      "learning_rate": 7.145609557545825e-06,
      "loss": 0.3631,
      "step": 24296
    },
    {
      "epoch": 1.115001606167684,
      "grad_norm": 0.4706784784793854,
      "learning_rate": 7.145388089827512e-06,
      "loss": 0.3377,
      "step": 24297
    },
    {
      "epoch": 1.1150474966729385,
      "grad_norm": 0.4684758484363556,
      "learning_rate": 7.145166616950217e-06,
      "loss": 0.3349,
      "step": 24298
    },
    {
      "epoch": 1.1150933871781927,
      "grad_norm": 0.44308769702911377,
      "learning_rate": 7.144945138914473e-06,
      "loss": 0.331,
      "step": 24299
    },
    {
      "epoch": 1.1151392776834472,
      "grad_norm": 0.45921099185943604,
      "learning_rate": 7.1447236557208135e-06,
      "loss": 0.3639,
      "step": 24300
    },
    {
      "epoch": 1.1151851681887017,
      "grad_norm": 0.5030645728111267,
      "learning_rate": 7.14450216736977e-06,
      "loss": 0.4407,
      "step": 24301
    },
    {
      "epoch": 1.1152310586939562,
      "grad_norm": 0.46139273047447205,
      "learning_rate": 7.144280673861876e-06,
      "loss": 0.3448,
      "step": 24302
    },
    {
      "epoch": 1.1152769491992107,
      "grad_norm": 0.4381040632724762,
      "learning_rate": 7.144059175197664e-06,
      "loss": 0.3139,
      "step": 24303
    },
    {
      "epoch": 1.1153228397044652,
      "grad_norm": 0.4558829367160797,
      "learning_rate": 7.143837671377665e-06,
      "loss": 0.3681,
      "step": 24304
    },
    {
      "epoch": 1.1153687302097195,
      "grad_norm": 0.4585329592227936,
      "learning_rate": 7.143616162402414e-06,
      "loss": 0.3888,
      "step": 24305
    },
    {
      "epoch": 1.115414620714974,
      "grad_norm": 0.5145308375358582,
      "learning_rate": 7.143394648272442e-06,
      "loss": 0.4057,
      "step": 24306
    },
    {
      "epoch": 1.1154605112202285,
      "grad_norm": 0.4452243149280548,
      "learning_rate": 7.143173128988283e-06,
      "loss": 0.313,
      "step": 24307
    },
    {
      "epoch": 1.115506401725483,
      "grad_norm": 0.5108509063720703,
      "learning_rate": 7.14295160455047e-06,
      "loss": 0.4256,
      "step": 24308
    },
    {
      "epoch": 1.1155522922307375,
      "grad_norm": 0.46300122141838074,
      "learning_rate": 7.142730074959534e-06,
      "loss": 0.3608,
      "step": 24309
    },
    {
      "epoch": 1.115598182735992,
      "grad_norm": 0.503735363483429,
      "learning_rate": 7.142508540216008e-06,
      "loss": 0.4172,
      "step": 24310
    },
    {
      "epoch": 1.1156440732412465,
      "grad_norm": 0.46704715490341187,
      "learning_rate": 7.142287000320425e-06,
      "loss": 0.3852,
      "step": 24311
    },
    {
      "epoch": 1.1156899637465008,
      "grad_norm": 0.452558308839798,
      "learning_rate": 7.142065455273318e-06,
      "loss": 0.3751,
      "step": 24312
    },
    {
      "epoch": 1.1157358542517553,
      "grad_norm": 0.4894804060459137,
      "learning_rate": 7.141843905075221e-06,
      "loss": 0.4063,
      "step": 24313
    },
    {
      "epoch": 1.1157817447570098,
      "grad_norm": 0.4186522662639618,
      "learning_rate": 7.1416223497266645e-06,
      "loss": 0.2778,
      "step": 24314
    },
    {
      "epoch": 1.1158276352622643,
      "grad_norm": 0.482189416885376,
      "learning_rate": 7.1414007892281815e-06,
      "loss": 0.342,
      "step": 24315
    },
    {
      "epoch": 1.1158735257675187,
      "grad_norm": 0.4641772508621216,
      "learning_rate": 7.141179223580305e-06,
      "loss": 0.2932,
      "step": 24316
    },
    {
      "epoch": 1.1159194162727732,
      "grad_norm": 0.5172176361083984,
      "learning_rate": 7.140957652783569e-06,
      "loss": 0.4675,
      "step": 24317
    },
    {
      "epoch": 1.1159653067780275,
      "grad_norm": 0.49548348784446716,
      "learning_rate": 7.140736076838505e-06,
      "loss": 0.363,
      "step": 24318
    },
    {
      "epoch": 1.116011197283282,
      "grad_norm": 0.4969908893108368,
      "learning_rate": 7.140514495745648e-06,
      "loss": 0.4397,
      "step": 24319
    },
    {
      "epoch": 1.1160570877885365,
      "grad_norm": 0.5102058053016663,
      "learning_rate": 7.140292909505528e-06,
      "loss": 0.4168,
      "step": 24320
    },
    {
      "epoch": 1.116102978293791,
      "grad_norm": 0.4525607228279114,
      "learning_rate": 7.140071318118679e-06,
      "loss": 0.3186,
      "step": 24321
    },
    {
      "epoch": 1.1161488687990455,
      "grad_norm": 0.46271592378616333,
      "learning_rate": 7.1398497215856336e-06,
      "loss": 0.3682,
      "step": 24322
    },
    {
      "epoch": 1.1161947593043,
      "grad_norm": 0.45411646366119385,
      "learning_rate": 7.139628119906926e-06,
      "loss": 0.3483,
      "step": 24323
    },
    {
      "epoch": 1.1162406498095545,
      "grad_norm": 0.4574289619922638,
      "learning_rate": 7.139406513083088e-06,
      "loss": 0.3757,
      "step": 24324
    },
    {
      "epoch": 1.1162865403148088,
      "grad_norm": 0.49333423376083374,
      "learning_rate": 7.139184901114653e-06,
      "loss": 0.4425,
      "step": 24325
    },
    {
      "epoch": 1.1163324308200633,
      "grad_norm": 0.4710083305835724,
      "learning_rate": 7.1389632840021514e-06,
      "loss": 0.3792,
      "step": 24326
    },
    {
      "epoch": 1.1163783213253178,
      "grad_norm": 0.48885902762413025,
      "learning_rate": 7.138741661746119e-06,
      "loss": 0.3394,
      "step": 24327
    },
    {
      "epoch": 1.1164242118305723,
      "grad_norm": 0.4619206488132477,
      "learning_rate": 7.1385200343470895e-06,
      "loss": 0.3134,
      "step": 24328
    },
    {
      "epoch": 1.1164701023358268,
      "grad_norm": 0.4482620060443878,
      "learning_rate": 7.138298401805592e-06,
      "loss": 0.32,
      "step": 24329
    },
    {
      "epoch": 1.1165159928410813,
      "grad_norm": 0.45085129141807556,
      "learning_rate": 7.138076764122164e-06,
      "loss": 0.327,
      "step": 24330
    },
    {
      "epoch": 1.1165618833463355,
      "grad_norm": 0.4841673970222473,
      "learning_rate": 7.137855121297335e-06,
      "loss": 0.3978,
      "step": 24331
    },
    {
      "epoch": 1.11660777385159,
      "grad_norm": 0.469740092754364,
      "learning_rate": 7.137633473331638e-06,
      "loss": 0.3839,
      "step": 24332
    },
    {
      "epoch": 1.1166536643568445,
      "grad_norm": 0.4611456096172333,
      "learning_rate": 7.137411820225608e-06,
      "loss": 0.359,
      "step": 24333
    },
    {
      "epoch": 1.116699554862099,
      "grad_norm": 0.48861223459243774,
      "learning_rate": 7.137190161979778e-06,
      "loss": 0.4263,
      "step": 24334
    },
    {
      "epoch": 1.1167454453673535,
      "grad_norm": 0.4440300166606903,
      "learning_rate": 7.136968498594679e-06,
      "loss": 0.3547,
      "step": 24335
    },
    {
      "epoch": 1.116791335872608,
      "grad_norm": 0.4237881600856781,
      "learning_rate": 7.136746830070846e-06,
      "loss": 0.2821,
      "step": 24336
    },
    {
      "epoch": 1.1168372263778625,
      "grad_norm": 0.5539296865463257,
      "learning_rate": 7.1365251564088115e-06,
      "loss": 0.5153,
      "step": 24337
    },
    {
      "epoch": 1.1168831168831168,
      "grad_norm": 0.5078336596488953,
      "learning_rate": 7.136303477609108e-06,
      "loss": 0.472,
      "step": 24338
    },
    {
      "epoch": 1.1169290073883713,
      "grad_norm": 0.4568607211112976,
      "learning_rate": 7.136081793672267e-06,
      "loss": 0.3557,
      "step": 24339
    },
    {
      "epoch": 1.1169748978936258,
      "grad_norm": 0.47729426622390747,
      "learning_rate": 7.135860104598826e-06,
      "loss": 0.3574,
      "step": 24340
    },
    {
      "epoch": 1.1170207883988803,
      "grad_norm": 0.5007487535476685,
      "learning_rate": 7.135638410389315e-06,
      "loss": 0.4103,
      "step": 24341
    },
    {
      "epoch": 1.1170666789041348,
      "grad_norm": 0.47860297560691833,
      "learning_rate": 7.135416711044267e-06,
      "loss": 0.4086,
      "step": 24342
    },
    {
      "epoch": 1.1171125694093893,
      "grad_norm": 0.45741936564445496,
      "learning_rate": 7.135195006564217e-06,
      "loss": 0.3155,
      "step": 24343
    },
    {
      "epoch": 1.1171584599146436,
      "grad_norm": 0.4515902101993561,
      "learning_rate": 7.134973296949695e-06,
      "loss": 0.3021,
      "step": 24344
    },
    {
      "epoch": 1.117204350419898,
      "grad_norm": 0.46438005566596985,
      "learning_rate": 7.134751582201238e-06,
      "loss": 0.3303,
      "step": 24345
    },
    {
      "epoch": 1.1172502409251526,
      "grad_norm": 0.5039751529693604,
      "learning_rate": 7.134529862319377e-06,
      "loss": 0.4346,
      "step": 24346
    },
    {
      "epoch": 1.117296131430407,
      "grad_norm": 0.5088385939598083,
      "learning_rate": 7.1343081373046445e-06,
      "loss": 0.3336,
      "step": 24347
    },
    {
      "epoch": 1.1173420219356616,
      "grad_norm": 0.47086963057518005,
      "learning_rate": 7.134086407157575e-06,
      "loss": 0.358,
      "step": 24348
    },
    {
      "epoch": 1.117387912440916,
      "grad_norm": 0.4846627712249756,
      "learning_rate": 7.1338646718787e-06,
      "loss": 0.3544,
      "step": 24349
    },
    {
      "epoch": 1.1174338029461703,
      "grad_norm": 0.48292118310928345,
      "learning_rate": 7.133642931468554e-06,
      "loss": 0.3753,
      "step": 24350
    },
    {
      "epoch": 1.1174796934514248,
      "grad_norm": 0.45573264360427856,
      "learning_rate": 7.133421185927672e-06,
      "loss": 0.3364,
      "step": 24351
    },
    {
      "epoch": 1.1175255839566793,
      "grad_norm": 0.4344509541988373,
      "learning_rate": 7.1331994352565836e-06,
      "loss": 0.3167,
      "step": 24352
    },
    {
      "epoch": 1.1175714744619338,
      "grad_norm": 0.45220839977264404,
      "learning_rate": 7.132977679455823e-06,
      "loss": 0.36,
      "step": 24353
    },
    {
      "epoch": 1.1176173649671883,
      "grad_norm": 0.46268683671951294,
      "learning_rate": 7.132755918525928e-06,
      "loss": 0.3731,
      "step": 24354
    },
    {
      "epoch": 1.1176632554724428,
      "grad_norm": 0.4575693905353546,
      "learning_rate": 7.132534152467425e-06,
      "loss": 0.3202,
      "step": 24355
    },
    {
      "epoch": 1.117709145977697,
      "grad_norm": 0.49020811915397644,
      "learning_rate": 7.132312381280852e-06,
      "loss": 0.3676,
      "step": 24356
    },
    {
      "epoch": 1.1177550364829516,
      "grad_norm": 0.5937674641609192,
      "learning_rate": 7.132090604966741e-06,
      "loss": 0.2857,
      "step": 24357
    },
    {
      "epoch": 1.117800926988206,
      "grad_norm": 0.44291162490844727,
      "learning_rate": 7.1318688235256224e-06,
      "loss": 0.3306,
      "step": 24358
    },
    {
      "epoch": 1.1178468174934606,
      "grad_norm": 0.4942046105861664,
      "learning_rate": 7.1316470369580345e-06,
      "loss": 0.4171,
      "step": 24359
    },
    {
      "epoch": 1.117892707998715,
      "grad_norm": 0.45690515637397766,
      "learning_rate": 7.131425245264507e-06,
      "loss": 0.2946,
      "step": 24360
    },
    {
      "epoch": 1.1179385985039696,
      "grad_norm": 0.43732553720474243,
      "learning_rate": 7.131203448445575e-06,
      "loss": 0.2946,
      "step": 24361
    },
    {
      "epoch": 1.117984489009224,
      "grad_norm": 0.47183582186698914,
      "learning_rate": 7.130981646501771e-06,
      "loss": 0.3474,
      "step": 24362
    },
    {
      "epoch": 1.1180303795144784,
      "grad_norm": 0.45972776412963867,
      "learning_rate": 7.130759839433629e-06,
      "loss": 0.305,
      "step": 24363
    },
    {
      "epoch": 1.1180762700197329,
      "grad_norm": 0.49848994612693787,
      "learning_rate": 7.130538027241683e-06,
      "loss": 0.4,
      "step": 24364
    },
    {
      "epoch": 1.1181221605249874,
      "grad_norm": 0.4721606969833374,
      "learning_rate": 7.130316209926464e-06,
      "loss": 0.4084,
      "step": 24365
    },
    {
      "epoch": 1.1181680510302419,
      "grad_norm": 0.43907833099365234,
      "learning_rate": 7.130094387488507e-06,
      "loss": 0.2847,
      "step": 24366
    },
    {
      "epoch": 1.1182139415354964,
      "grad_norm": 0.5151420831680298,
      "learning_rate": 7.129872559928347e-06,
      "loss": 0.4115,
      "step": 24367
    },
    {
      "epoch": 1.1182598320407509,
      "grad_norm": 0.5098459720611572,
      "learning_rate": 7.129650727246514e-06,
      "loss": 0.4528,
      "step": 24368
    },
    {
      "epoch": 1.1183057225460051,
      "grad_norm": 0.4762936532497406,
      "learning_rate": 7.129428889443544e-06,
      "loss": 0.3523,
      "step": 24369
    },
    {
      "epoch": 1.1183516130512596,
      "grad_norm": 0.45573607087135315,
      "learning_rate": 7.129207046519968e-06,
      "loss": 0.3107,
      "step": 24370
    },
    {
      "epoch": 1.1183975035565141,
      "grad_norm": 0.46376508474349976,
      "learning_rate": 7.128985198476323e-06,
      "loss": 0.3826,
      "step": 24371
    },
    {
      "epoch": 1.1184433940617686,
      "grad_norm": 0.5174378752708435,
      "learning_rate": 7.1287633453131385e-06,
      "loss": 0.4485,
      "step": 24372
    },
    {
      "epoch": 1.1184892845670231,
      "grad_norm": 0.448304146528244,
      "learning_rate": 7.128541487030952e-06,
      "loss": 0.3437,
      "step": 24373
    },
    {
      "epoch": 1.1185351750722776,
      "grad_norm": 0.43810001015663147,
      "learning_rate": 7.1283196236302956e-06,
      "loss": 0.3357,
      "step": 24374
    },
    {
      "epoch": 1.1185810655775321,
      "grad_norm": 0.48438560962677,
      "learning_rate": 7.128097755111699e-06,
      "loss": 0.3615,
      "step": 24375
    },
    {
      "epoch": 1.1186269560827864,
      "grad_norm": 0.4682760238647461,
      "learning_rate": 7.127875881475702e-06,
      "loss": 0.3787,
      "step": 24376
    },
    {
      "epoch": 1.118672846588041,
      "grad_norm": 0.46575334668159485,
      "learning_rate": 7.127654002722833e-06,
      "loss": 0.3825,
      "step": 24377
    },
    {
      "epoch": 1.1187187370932954,
      "grad_norm": 0.46889790892601013,
      "learning_rate": 7.127432118853629e-06,
      "loss": 0.3352,
      "step": 24378
    },
    {
      "epoch": 1.1187646275985499,
      "grad_norm": 0.4734900891780853,
      "learning_rate": 7.127210229868622e-06,
      "loss": 0.3238,
      "step": 24379
    },
    {
      "epoch": 1.1188105181038044,
      "grad_norm": 0.44081026315689087,
      "learning_rate": 7.1269883357683455e-06,
      "loss": 0.3224,
      "step": 24380
    },
    {
      "epoch": 1.1188564086090589,
      "grad_norm": 0.4423227906227112,
      "learning_rate": 7.126766436553332e-06,
      "loss": 0.3409,
      "step": 24381
    },
    {
      "epoch": 1.1189022991143132,
      "grad_norm": 0.5010138750076294,
      "learning_rate": 7.126544532224119e-06,
      "loss": 0.4052,
      "step": 24382
    },
    {
      "epoch": 1.1189481896195677,
      "grad_norm": 0.4685904085636139,
      "learning_rate": 7.126322622781235e-06,
      "loss": 0.3854,
      "step": 24383
    },
    {
      "epoch": 1.1189940801248222,
      "grad_norm": 0.4451734721660614,
      "learning_rate": 7.126100708225218e-06,
      "loss": 0.3215,
      "step": 24384
    },
    {
      "epoch": 1.1190399706300767,
      "grad_norm": 0.43488505482673645,
      "learning_rate": 7.1258787885566e-06,
      "loss": 0.3116,
      "step": 24385
    },
    {
      "epoch": 1.1190858611353312,
      "grad_norm": 0.4435853660106659,
      "learning_rate": 7.125656863775914e-06,
      "loss": 0.3023,
      "step": 24386
    },
    {
      "epoch": 1.1191317516405856,
      "grad_norm": 0.45220503211021423,
      "learning_rate": 7.1254349338836925e-06,
      "loss": 0.3249,
      "step": 24387
    },
    {
      "epoch": 1.11917764214584,
      "grad_norm": 0.4884524643421173,
      "learning_rate": 7.125212998880473e-06,
      "loss": 0.3938,
      "step": 24388
    },
    {
      "epoch": 1.1192235326510944,
      "grad_norm": 0.5248652696609497,
      "learning_rate": 7.124991058766786e-06,
      "loss": 0.4274,
      "step": 24389
    },
    {
      "epoch": 1.119269423156349,
      "grad_norm": 0.47318708896636963,
      "learning_rate": 7.1247691135431675e-06,
      "loss": 0.3922,
      "step": 24390
    },
    {
      "epoch": 1.1193153136616034,
      "grad_norm": 0.45581868290901184,
      "learning_rate": 7.124547163210148e-06,
      "loss": 0.3338,
      "step": 24391
    },
    {
      "epoch": 1.119361204166858,
      "grad_norm": 0.44147130846977234,
      "learning_rate": 7.124325207768263e-06,
      "loss": 0.3129,
      "step": 24392
    },
    {
      "epoch": 1.1194070946721124,
      "grad_norm": 0.4626491367816925,
      "learning_rate": 7.124103247218048e-06,
      "loss": 0.3162,
      "step": 24393
    },
    {
      "epoch": 1.1194529851773667,
      "grad_norm": 0.5089107155799866,
      "learning_rate": 7.1238812815600335e-06,
      "loss": 0.4024,
      "step": 24394
    },
    {
      "epoch": 1.1194988756826212,
      "grad_norm": 0.47981545329093933,
      "learning_rate": 7.123659310794756e-06,
      "loss": 0.4022,
      "step": 24395
    },
    {
      "epoch": 1.1195447661878757,
      "grad_norm": 0.42519521713256836,
      "learning_rate": 7.123437334922748e-06,
      "loss": 0.3044,
      "step": 24396
    },
    {
      "epoch": 1.1195906566931302,
      "grad_norm": 0.4342162013053894,
      "learning_rate": 7.1232153539445425e-06,
      "loss": 0.3335,
      "step": 24397
    },
    {
      "epoch": 1.1196365471983847,
      "grad_norm": 0.47132354974746704,
      "learning_rate": 7.122993367860674e-06,
      "loss": 0.3401,
      "step": 24398
    },
    {
      "epoch": 1.1196824377036392,
      "grad_norm": 0.4608250558376312,
      "learning_rate": 7.122771376671678e-06,
      "loss": 0.34,
      "step": 24399
    },
    {
      "epoch": 1.1197283282088937,
      "grad_norm": 0.4674767851829529,
      "learning_rate": 7.122549380378087e-06,
      "loss": 0.3435,
      "step": 24400
    },
    {
      "epoch": 1.119774218714148,
      "grad_norm": 0.4207720458507538,
      "learning_rate": 7.1223273789804326e-06,
      "loss": 0.271,
      "step": 24401
    },
    {
      "epoch": 1.1198201092194024,
      "grad_norm": 0.5110670328140259,
      "learning_rate": 7.1221053724792525e-06,
      "loss": 0.4602,
      "step": 24402
    },
    {
      "epoch": 1.119865999724657,
      "grad_norm": 0.48111262917518616,
      "learning_rate": 7.121883360875077e-06,
      "loss": 0.3799,
      "step": 24403
    },
    {
      "epoch": 1.1199118902299114,
      "grad_norm": 0.4518736004829407,
      "learning_rate": 7.121661344168442e-06,
      "loss": 0.3064,
      "step": 24404
    },
    {
      "epoch": 1.119957780735166,
      "grad_norm": 0.47767311334609985,
      "learning_rate": 7.1214393223598825e-06,
      "loss": 0.3916,
      "step": 24405
    },
    {
      "epoch": 1.1200036712404204,
      "grad_norm": 0.4766079783439636,
      "learning_rate": 7.121217295449931e-06,
      "loss": 0.3904,
      "step": 24406
    },
    {
      "epoch": 1.1200495617456747,
      "grad_norm": 0.5239993929862976,
      "learning_rate": 7.120995263439121e-06,
      "loss": 0.4783,
      "step": 24407
    },
    {
      "epoch": 1.1200954522509292,
      "grad_norm": 0.5094127058982849,
      "learning_rate": 7.120773226327986e-06,
      "loss": 0.4139,
      "step": 24408
    },
    {
      "epoch": 1.1201413427561837,
      "grad_norm": 0.4709087014198303,
      "learning_rate": 7.120551184117061e-06,
      "loss": 0.4434,
      "step": 24409
    },
    {
      "epoch": 1.1201872332614382,
      "grad_norm": 0.5592716336250305,
      "learning_rate": 7.120329136806879e-06,
      "loss": 0.3763,
      "step": 24410
    },
    {
      "epoch": 1.1202331237666927,
      "grad_norm": 0.4713756740093231,
      "learning_rate": 7.1201070843979756e-06,
      "loss": 0.3531,
      "step": 24411
    },
    {
      "epoch": 1.1202790142719472,
      "grad_norm": 0.45278388261795044,
      "learning_rate": 7.119885026890884e-06,
      "loss": 0.3546,
      "step": 24412
    },
    {
      "epoch": 1.1203249047772017,
      "grad_norm": 0.46855875849723816,
      "learning_rate": 7.1196629642861384e-06,
      "loss": 0.3116,
      "step": 24413
    },
    {
      "epoch": 1.120370795282456,
      "grad_norm": 0.4318397343158722,
      "learning_rate": 7.119440896584271e-06,
      "loss": 0.2751,
      "step": 24414
    },
    {
      "epoch": 1.1204166857877105,
      "grad_norm": 0.47066086530685425,
      "learning_rate": 7.119218823785816e-06,
      "loss": 0.3455,
      "step": 24415
    },
    {
      "epoch": 1.120462576292965,
      "grad_norm": 0.4986349642276764,
      "learning_rate": 7.118996745891311e-06,
      "loss": 0.3862,
      "step": 24416
    },
    {
      "epoch": 1.1205084667982195,
      "grad_norm": 0.4356958270072937,
      "learning_rate": 7.118774662901288e-06,
      "loss": 0.3364,
      "step": 24417
    },
    {
      "epoch": 1.120554357303474,
      "grad_norm": 0.5212225914001465,
      "learning_rate": 7.118552574816279e-06,
      "loss": 0.3989,
      "step": 24418
    },
    {
      "epoch": 1.1206002478087285,
      "grad_norm": 0.49545150995254517,
      "learning_rate": 7.1183304816368206e-06,
      "loss": 0.4239,
      "step": 24419
    },
    {
      "epoch": 1.1206461383139827,
      "grad_norm": 0.4666737914085388,
      "learning_rate": 7.118108383363444e-06,
      "loss": 0.3239,
      "step": 24420
    },
    {
      "epoch": 1.1206920288192372,
      "grad_norm": 0.46655821800231934,
      "learning_rate": 7.1178862799966875e-06,
      "loss": 0.3498,
      "step": 24421
    },
    {
      "epoch": 1.1207379193244917,
      "grad_norm": 0.443051278591156,
      "learning_rate": 7.1176641715370835e-06,
      "loss": 0.3387,
      "step": 24422
    },
    {
      "epoch": 1.1207838098297462,
      "grad_norm": 0.4540196657180786,
      "learning_rate": 7.117442057985163e-06,
      "loss": 0.3317,
      "step": 24423
    },
    {
      "epoch": 1.1208297003350007,
      "grad_norm": 0.5015744566917419,
      "learning_rate": 7.117219939341464e-06,
      "loss": 0.376,
      "step": 24424
    },
    {
      "epoch": 1.1208755908402552,
      "grad_norm": 0.4612747132778168,
      "learning_rate": 7.116997815606519e-06,
      "loss": 0.3307,
      "step": 24425
    },
    {
      "epoch": 1.1209214813455097,
      "grad_norm": 0.47574570775032043,
      "learning_rate": 7.116775686780861e-06,
      "loss": 0.3971,
      "step": 24426
    },
    {
      "epoch": 1.120967371850764,
      "grad_norm": 0.693388819694519,
      "learning_rate": 7.116553552865029e-06,
      "loss": 0.4142,
      "step": 24427
    },
    {
      "epoch": 1.1210132623560185,
      "grad_norm": 0.4336928129196167,
      "learning_rate": 7.1163314138595515e-06,
      "loss": 0.3286,
      "step": 24428
    },
    {
      "epoch": 1.121059152861273,
      "grad_norm": 0.4630883038043976,
      "learning_rate": 7.116109269764965e-06,
      "loss": 0.3681,
      "step": 24429
    },
    {
      "epoch": 1.1211050433665275,
      "grad_norm": 0.4876170754432678,
      "learning_rate": 7.115887120581804e-06,
      "loss": 0.3894,
      "step": 24430
    },
    {
      "epoch": 1.121150933871782,
      "grad_norm": 0.4595867693424225,
      "learning_rate": 7.1156649663106016e-06,
      "loss": 0.3778,
      "step": 24431
    },
    {
      "epoch": 1.1211968243770365,
      "grad_norm": 0.4024995267391205,
      "learning_rate": 7.115442806951892e-06,
      "loss": 0.2678,
      "step": 24432
    },
    {
      "epoch": 1.1212427148822908,
      "grad_norm": 0.4695373475551605,
      "learning_rate": 7.115220642506213e-06,
      "loss": 0.3899,
      "step": 24433
    },
    {
      "epoch": 1.1212886053875453,
      "grad_norm": 0.4568944275379181,
      "learning_rate": 7.114998472974093e-06,
      "loss": 0.3738,
      "step": 24434
    },
    {
      "epoch": 1.1213344958927998,
      "grad_norm": 0.478090763092041,
      "learning_rate": 7.11477629835607e-06,
      "loss": 0.3855,
      "step": 24435
    },
    {
      "epoch": 1.1213803863980543,
      "grad_norm": 0.4668237864971161,
      "learning_rate": 7.114554118652679e-06,
      "loss": 0.3565,
      "step": 24436
    },
    {
      "epoch": 1.1214262769033088,
      "grad_norm": 0.46233442425727844,
      "learning_rate": 7.11433193386445e-06,
      "loss": 0.3366,
      "step": 24437
    },
    {
      "epoch": 1.1214721674085633,
      "grad_norm": 0.4658839702606201,
      "learning_rate": 7.114109743991922e-06,
      "loss": 0.342,
      "step": 24438
    },
    {
      "epoch": 1.1215180579138175,
      "grad_norm": 0.48771369457244873,
      "learning_rate": 7.113887549035627e-06,
      "loss": 0.4359,
      "step": 24439
    },
    {
      "epoch": 1.121563948419072,
      "grad_norm": 0.4754769206047058,
      "learning_rate": 7.113665348996099e-06,
      "loss": 0.3885,
      "step": 24440
    },
    {
      "epoch": 1.1216098389243265,
      "grad_norm": 0.4918484687805176,
      "learning_rate": 7.113443143873874e-06,
      "loss": 0.3671,
      "step": 24441
    },
    {
      "epoch": 1.121655729429581,
      "grad_norm": 0.45737361907958984,
      "learning_rate": 7.113220933669485e-06,
      "loss": 0.3332,
      "step": 24442
    },
    {
      "epoch": 1.1217016199348355,
      "grad_norm": 0.4461697041988373,
      "learning_rate": 7.112998718383465e-06,
      "loss": 0.309,
      "step": 24443
    },
    {
      "epoch": 1.12174751044009,
      "grad_norm": 0.5030305981636047,
      "learning_rate": 7.112776498016352e-06,
      "loss": 0.4396,
      "step": 24444
    },
    {
      "epoch": 1.1217934009453443,
      "grad_norm": 0.47860968112945557,
      "learning_rate": 7.112554272568677e-06,
      "loss": 0.3351,
      "step": 24445
    },
    {
      "epoch": 1.1218392914505988,
      "grad_norm": 0.44691017270088196,
      "learning_rate": 7.112332042040976e-06,
      "loss": 0.315,
      "step": 24446
    },
    {
      "epoch": 1.1218851819558533,
      "grad_norm": 0.506725549697876,
      "learning_rate": 7.112109806433784e-06,
      "loss": 0.3941,
      "step": 24447
    },
    {
      "epoch": 1.1219310724611078,
      "grad_norm": 0.47101643681526184,
      "learning_rate": 7.111887565747633e-06,
      "loss": 0.3797,
      "step": 24448
    },
    {
      "epoch": 1.1219769629663623,
      "grad_norm": 0.4220737814903259,
      "learning_rate": 7.11166531998306e-06,
      "loss": 0.3103,
      "step": 24449
    },
    {
      "epoch": 1.1220228534716168,
      "grad_norm": 0.4636615216732025,
      "learning_rate": 7.111443069140598e-06,
      "loss": 0.3334,
      "step": 24450
    },
    {
      "epoch": 1.1220687439768713,
      "grad_norm": 0.43551406264305115,
      "learning_rate": 7.1112208132207815e-06,
      "loss": 0.2961,
      "step": 24451
    },
    {
      "epoch": 1.1221146344821256,
      "grad_norm": 0.4257749021053314,
      "learning_rate": 7.110998552224145e-06,
      "loss": 0.3068,
      "step": 24452
    },
    {
      "epoch": 1.12216052498738,
      "grad_norm": 0.4876939356327057,
      "learning_rate": 7.1107762861512245e-06,
      "loss": 0.3572,
      "step": 24453
    },
    {
      "epoch": 1.1222064154926346,
      "grad_norm": 0.5125496983528137,
      "learning_rate": 7.110554015002552e-06,
      "loss": 0.4043,
      "step": 24454
    },
    {
      "epoch": 1.122252305997889,
      "grad_norm": 0.4877931773662567,
      "learning_rate": 7.110331738778666e-06,
      "loss": 0.4135,
      "step": 24455
    },
    {
      "epoch": 1.1222981965031436,
      "grad_norm": 0.46114933490753174,
      "learning_rate": 7.110109457480095e-06,
      "loss": 0.3856,
      "step": 24456
    },
    {
      "epoch": 1.122344087008398,
      "grad_norm": 0.4857105016708374,
      "learning_rate": 7.109887171107377e-06,
      "loss": 0.3721,
      "step": 24457
    },
    {
      "epoch": 1.1223899775136523,
      "grad_norm": 0.485323429107666,
      "learning_rate": 7.109664879661048e-06,
      "loss": 0.3704,
      "step": 24458
    },
    {
      "epoch": 1.1224358680189068,
      "grad_norm": 0.4756920039653778,
      "learning_rate": 7.109442583141639e-06,
      "loss": 0.4091,
      "step": 24459
    },
    {
      "epoch": 1.1224817585241613,
      "grad_norm": 0.6531346440315247,
      "learning_rate": 7.109220281549687e-06,
      "loss": 0.4007,
      "step": 24460
    },
    {
      "epoch": 1.1225276490294158,
      "grad_norm": 0.5017656683921814,
      "learning_rate": 7.108997974885728e-06,
      "loss": 0.347,
      "step": 24461
    },
    {
      "epoch": 1.1225735395346703,
      "grad_norm": 0.45709168910980225,
      "learning_rate": 7.1087756631502915e-06,
      "loss": 0.3509,
      "step": 24462
    },
    {
      "epoch": 1.1226194300399248,
      "grad_norm": 0.4704689383506775,
      "learning_rate": 7.108553346343915e-06,
      "loss": 0.4163,
      "step": 24463
    },
    {
      "epoch": 1.1226653205451793,
      "grad_norm": 0.4220016896724701,
      "learning_rate": 7.1083310244671345e-06,
      "loss": 0.2975,
      "step": 24464
    },
    {
      "epoch": 1.1227112110504336,
      "grad_norm": 0.4469843804836273,
      "learning_rate": 7.1081086975204825e-06,
      "loss": 0.3218,
      "step": 24465
    },
    {
      "epoch": 1.122757101555688,
      "grad_norm": 0.4760843813419342,
      "learning_rate": 7.107886365504496e-06,
      "loss": 0.3728,
      "step": 24466
    },
    {
      "epoch": 1.1228029920609426,
      "grad_norm": 0.454483300447464,
      "learning_rate": 7.1076640284197075e-06,
      "loss": 0.3237,
      "step": 24467
    },
    {
      "epoch": 1.122848882566197,
      "grad_norm": 0.4691222012042999,
      "learning_rate": 7.107441686266652e-06,
      "loss": 0.3445,
      "step": 24468
    },
    {
      "epoch": 1.1228947730714516,
      "grad_norm": 0.47898927330970764,
      "learning_rate": 7.107219339045863e-06,
      "loss": 0.3548,
      "step": 24469
    },
    {
      "epoch": 1.122940663576706,
      "grad_norm": 0.4752456247806549,
      "learning_rate": 7.106996986757878e-06,
      "loss": 0.3619,
      "step": 24470
    },
    {
      "epoch": 1.1229865540819604,
      "grad_norm": 0.4784412980079651,
      "learning_rate": 7.1067746294032305e-06,
      "loss": 0.4044,
      "step": 24471
    },
    {
      "epoch": 1.1230324445872149,
      "grad_norm": 0.48600858449935913,
      "learning_rate": 7.106552266982455e-06,
      "loss": 0.3615,
      "step": 24472
    },
    {
      "epoch": 1.1230783350924693,
      "grad_norm": 0.4752776324748993,
      "learning_rate": 7.106329899496085e-06,
      "loss": 0.3459,
      "step": 24473
    },
    {
      "epoch": 1.1231242255977238,
      "grad_norm": 0.4536878764629364,
      "learning_rate": 7.106107526944657e-06,
      "loss": 0.3047,
      "step": 24474
    },
    {
      "epoch": 1.1231701161029783,
      "grad_norm": 0.4537631869316101,
      "learning_rate": 7.105885149328704e-06,
      "loss": 0.3378,
      "step": 24475
    },
    {
      "epoch": 1.1232160066082328,
      "grad_norm": 0.4891679883003235,
      "learning_rate": 7.105662766648763e-06,
      "loss": 0.3959,
      "step": 24476
    },
    {
      "epoch": 1.1232618971134871,
      "grad_norm": 0.5229505300521851,
      "learning_rate": 7.1054403789053684e-06,
      "loss": 0.3802,
      "step": 24477
    },
    {
      "epoch": 1.1233077876187416,
      "grad_norm": 0.5272250175476074,
      "learning_rate": 7.105217986099054e-06,
      "loss": 0.4326,
      "step": 24478
    },
    {
      "epoch": 1.1233536781239961,
      "grad_norm": 0.44651180505752563,
      "learning_rate": 7.1049955882303526e-06,
      "loss": 0.3184,
      "step": 24479
    },
    {
      "epoch": 1.1233995686292506,
      "grad_norm": 0.5361708402633667,
      "learning_rate": 7.104773185299803e-06,
      "loss": 0.4607,
      "step": 24480
    },
    {
      "epoch": 1.123445459134505,
      "grad_norm": 0.4506639838218689,
      "learning_rate": 7.104550777307937e-06,
      "loss": 0.3138,
      "step": 24481
    },
    {
      "epoch": 1.1234913496397596,
      "grad_norm": 0.47576722502708435,
      "learning_rate": 7.104328364255292e-06,
      "loss": 0.4276,
      "step": 24482
    },
    {
      "epoch": 1.1235372401450139,
      "grad_norm": 0.4305223524570465,
      "learning_rate": 7.1041059461424e-06,
      "loss": 0.3027,
      "step": 24483
    },
    {
      "epoch": 1.1235831306502684,
      "grad_norm": 0.4397425353527069,
      "learning_rate": 7.1038835229697985e-06,
      "loss": 0.3345,
      "step": 24484
    },
    {
      "epoch": 1.1236290211555229,
      "grad_norm": 0.4513666331768036,
      "learning_rate": 7.103661094738021e-06,
      "loss": 0.3533,
      "step": 24485
    },
    {
      "epoch": 1.1236749116607774,
      "grad_norm": 0.4771495759487152,
      "learning_rate": 7.103438661447601e-06,
      "loss": 0.3408,
      "step": 24486
    },
    {
      "epoch": 1.1237208021660319,
      "grad_norm": 0.5015978217124939,
      "learning_rate": 7.103216223099077e-06,
      "loss": 0.4135,
      "step": 24487
    },
    {
      "epoch": 1.1237666926712864,
      "grad_norm": 0.5056594610214233,
      "learning_rate": 7.1029937796929805e-06,
      "loss": 0.367,
      "step": 24488
    },
    {
      "epoch": 1.1238125831765409,
      "grad_norm": 0.44038277864456177,
      "learning_rate": 7.102771331229849e-06,
      "loss": 0.3394,
      "step": 24489
    },
    {
      "epoch": 1.1238584736817951,
      "grad_norm": 0.4544081687927246,
      "learning_rate": 7.102548877710215e-06,
      "loss": 0.3394,
      "step": 24490
    },
    {
      "epoch": 1.1239043641870496,
      "grad_norm": 0.5266464948654175,
      "learning_rate": 7.102326419134613e-06,
      "loss": 0.4399,
      "step": 24491
    },
    {
      "epoch": 1.1239502546923041,
      "grad_norm": 0.49854058027267456,
      "learning_rate": 7.102103955503582e-06,
      "loss": 0.4403,
      "step": 24492
    },
    {
      "epoch": 1.1239961451975586,
      "grad_norm": 0.4582069218158722,
      "learning_rate": 7.101881486817654e-06,
      "loss": 0.3308,
      "step": 24493
    },
    {
      "epoch": 1.1240420357028131,
      "grad_norm": 0.5077621340751648,
      "learning_rate": 7.101659013077363e-06,
      "loss": 0.4576,
      "step": 24494
    },
    {
      "epoch": 1.1240879262080676,
      "grad_norm": 0.47128525376319885,
      "learning_rate": 7.101436534283248e-06,
      "loss": 0.3936,
      "step": 24495
    },
    {
      "epoch": 1.124133816713322,
      "grad_norm": 0.45865651965141296,
      "learning_rate": 7.101214050435838e-06,
      "loss": 0.3334,
      "step": 24496
    },
    {
      "epoch": 1.1241797072185764,
      "grad_norm": 0.4583466649055481,
      "learning_rate": 7.100991561535673e-06,
      "loss": 0.3158,
      "step": 24497
    },
    {
      "epoch": 1.124225597723831,
      "grad_norm": 0.44251590967178345,
      "learning_rate": 7.100769067583288e-06,
      "loss": 0.3146,
      "step": 24498
    },
    {
      "epoch": 1.1242714882290854,
      "grad_norm": 0.4577972888946533,
      "learning_rate": 7.100546568579214e-06,
      "loss": 0.3522,
      "step": 24499
    },
    {
      "epoch": 1.12431737873434,
      "grad_norm": 0.46744784712791443,
      "learning_rate": 7.100324064523989e-06,
      "loss": 0.337,
      "step": 24500
    },
    {
      "epoch": 1.1243632692395944,
      "grad_norm": 0.46433404088020325,
      "learning_rate": 7.100101555418149e-06,
      "loss": 0.3773,
      "step": 24501
    },
    {
      "epoch": 1.124409159744849,
      "grad_norm": 0.4964696168899536,
      "learning_rate": 7.099879041262225e-06,
      "loss": 0.4166,
      "step": 24502
    },
    {
      "epoch": 1.1244550502501032,
      "grad_norm": 0.4846893846988678,
      "learning_rate": 7.099656522056757e-06,
      "loss": 0.3867,
      "step": 24503
    },
    {
      "epoch": 1.1245009407553577,
      "grad_norm": 0.46418121457099915,
      "learning_rate": 7.099433997802277e-06,
      "loss": 0.3493,
      "step": 24504
    },
    {
      "epoch": 1.1245468312606122,
      "grad_norm": 0.4548535943031311,
      "learning_rate": 7.099211468499321e-06,
      "loss": 0.3682,
      "step": 24505
    },
    {
      "epoch": 1.1245927217658667,
      "grad_norm": 0.46428796648979187,
      "learning_rate": 7.098988934148423e-06,
      "loss": 0.3751,
      "step": 24506
    },
    {
      "epoch": 1.1246386122711212,
      "grad_norm": 0.49877917766571045,
      "learning_rate": 7.098766394750119e-06,
      "loss": 0.432,
      "step": 24507
    },
    {
      "epoch": 1.1246845027763757,
      "grad_norm": 0.4951443076133728,
      "learning_rate": 7.098543850304944e-06,
      "loss": 0.3796,
      "step": 24508
    },
    {
      "epoch": 1.12473039328163,
      "grad_norm": 0.42637306451797485,
      "learning_rate": 7.0983213008134354e-06,
      "loss": 0.3174,
      "step": 24509
    },
    {
      "epoch": 1.1247762837868844,
      "grad_norm": 0.42890873551368713,
      "learning_rate": 7.098098746276125e-06,
      "loss": 0.3249,
      "step": 24510
    },
    {
      "epoch": 1.124822174292139,
      "grad_norm": 0.4537610411643982,
      "learning_rate": 7.09787618669355e-06,
      "loss": 0.3228,
      "step": 24511
    },
    {
      "epoch": 1.1248680647973934,
      "grad_norm": 0.46539074182510376,
      "learning_rate": 7.097653622066244e-06,
      "loss": 0.3955,
      "step": 24512
    },
    {
      "epoch": 1.124913955302648,
      "grad_norm": 0.45689210295677185,
      "learning_rate": 7.097431052394743e-06,
      "loss": 0.3165,
      "step": 24513
    },
    {
      "epoch": 1.1249598458079024,
      "grad_norm": 0.46390965580940247,
      "learning_rate": 7.0972084776795836e-06,
      "loss": 0.3422,
      "step": 24514
    },
    {
      "epoch": 1.125005736313157,
      "grad_norm": 0.4720083475112915,
      "learning_rate": 7.096985897921299e-06,
      "loss": 0.3683,
      "step": 24515
    },
    {
      "epoch": 1.1250516268184112,
      "grad_norm": 0.4342659115791321,
      "learning_rate": 7.096763313120424e-06,
      "loss": 0.311,
      "step": 24516
    },
    {
      "epoch": 1.1250975173236657,
      "grad_norm": 0.4513651430606842,
      "learning_rate": 7.096540723277497e-06,
      "loss": 0.312,
      "step": 24517
    },
    {
      "epoch": 1.1251434078289202,
      "grad_norm": 0.4567718505859375,
      "learning_rate": 7.096318128393049e-06,
      "loss": 0.3592,
      "step": 24518
    },
    {
      "epoch": 1.1251892983341747,
      "grad_norm": 0.48970913887023926,
      "learning_rate": 7.096095528467618e-06,
      "loss": 0.3802,
      "step": 24519
    },
    {
      "epoch": 1.1252351888394292,
      "grad_norm": 0.4593092203140259,
      "learning_rate": 7.095872923501741e-06,
      "loss": 0.3491,
      "step": 24520
    },
    {
      "epoch": 1.1252810793446835,
      "grad_norm": 0.481396347284317,
      "learning_rate": 7.09565031349595e-06,
      "loss": 0.4155,
      "step": 24521
    },
    {
      "epoch": 1.125326969849938,
      "grad_norm": 0.4844643175601959,
      "learning_rate": 7.09542769845078e-06,
      "loss": 0.3895,
      "step": 24522
    },
    {
      "epoch": 1.1253728603551925,
      "grad_norm": 0.4977850914001465,
      "learning_rate": 7.095205078366769e-06,
      "loss": 0.4252,
      "step": 24523
    },
    {
      "epoch": 1.125418750860447,
      "grad_norm": 0.4896126687526703,
      "learning_rate": 7.09498245324445e-06,
      "loss": 0.4079,
      "step": 24524
    },
    {
      "epoch": 1.1254646413657015,
      "grad_norm": 0.4772425889968872,
      "learning_rate": 7.094759823084361e-06,
      "loss": 0.3756,
      "step": 24525
    },
    {
      "epoch": 1.125510531870956,
      "grad_norm": 0.4161599278450012,
      "learning_rate": 7.094537187887035e-06,
      "loss": 0.2733,
      "step": 24526
    },
    {
      "epoch": 1.1255564223762105,
      "grad_norm": 0.43875566124916077,
      "learning_rate": 7.094314547653008e-06,
      "loss": 0.3373,
      "step": 24527
    },
    {
      "epoch": 1.1256023128814647,
      "grad_norm": 0.4744507372379303,
      "learning_rate": 7.094091902382815e-06,
      "loss": 0.3851,
      "step": 24528
    },
    {
      "epoch": 1.1256482033867192,
      "grad_norm": 0.47918373346328735,
      "learning_rate": 7.0938692520769926e-06,
      "loss": 0.3404,
      "step": 24529
    },
    {
      "epoch": 1.1256940938919737,
      "grad_norm": 0.4705836772918701,
      "learning_rate": 7.093646596736074e-06,
      "loss": 0.3283,
      "step": 24530
    },
    {
      "epoch": 1.1257399843972282,
      "grad_norm": 0.503592848777771,
      "learning_rate": 7.093423936360598e-06,
      "loss": 0.3549,
      "step": 24531
    },
    {
      "epoch": 1.1257858749024827,
      "grad_norm": 0.4714903235435486,
      "learning_rate": 7.093201270951099e-06,
      "loss": 0.3303,
      "step": 24532
    },
    {
      "epoch": 1.1258317654077372,
      "grad_norm": 0.4885201156139374,
      "learning_rate": 7.092978600508109e-06,
      "loss": 0.3575,
      "step": 24533
    },
    {
      "epoch": 1.1258776559129915,
      "grad_norm": 0.5011963844299316,
      "learning_rate": 7.092755925032166e-06,
      "loss": 0.3664,
      "step": 24534
    },
    {
      "epoch": 1.125923546418246,
      "grad_norm": 0.5127184987068176,
      "learning_rate": 7.092533244523808e-06,
      "loss": 0.4267,
      "step": 24535
    },
    {
      "epoch": 1.1259694369235005,
      "grad_norm": 0.48105233907699585,
      "learning_rate": 7.092310558983567e-06,
      "loss": 0.3442,
      "step": 24536
    },
    {
      "epoch": 1.126015327428755,
      "grad_norm": 0.48005300760269165,
      "learning_rate": 7.09208786841198e-06,
      "loss": 0.3795,
      "step": 24537
    },
    {
      "epoch": 1.1260612179340095,
      "grad_norm": 0.4687623977661133,
      "learning_rate": 7.09186517280958e-06,
      "loss": 0.4175,
      "step": 24538
    },
    {
      "epoch": 1.126107108439264,
      "grad_norm": 0.466586709022522,
      "learning_rate": 7.091642472176905e-06,
      "loss": 0.3601,
      "step": 24539
    },
    {
      "epoch": 1.1261529989445185,
      "grad_norm": 0.45807895064353943,
      "learning_rate": 7.0914197665144915e-06,
      "loss": 0.3879,
      "step": 24540
    },
    {
      "epoch": 1.1261988894497728,
      "grad_norm": 0.46288588643074036,
      "learning_rate": 7.091197055822871e-06,
      "loss": 0.3515,
      "step": 24541
    },
    {
      "epoch": 1.1262447799550273,
      "grad_norm": 0.4602619409561157,
      "learning_rate": 7.090974340102584e-06,
      "loss": 0.3451,
      "step": 24542
    },
    {
      "epoch": 1.1262906704602818,
      "grad_norm": 0.4624866545200348,
      "learning_rate": 7.090751619354164e-06,
      "loss": 0.3427,
      "step": 24543
    },
    {
      "epoch": 1.1263365609655362,
      "grad_norm": 0.42305925488471985,
      "learning_rate": 7.090528893578144e-06,
      "loss": 0.2961,
      "step": 24544
    },
    {
      "epoch": 1.1263824514707907,
      "grad_norm": 0.4329456686973572,
      "learning_rate": 7.090306162775062e-06,
      "loss": 0.2941,
      "step": 24545
    },
    {
      "epoch": 1.1264283419760452,
      "grad_norm": 0.4587375819683075,
      "learning_rate": 7.090083426945455e-06,
      "loss": 0.3478,
      "step": 24546
    },
    {
      "epoch": 1.1264742324812995,
      "grad_norm": 0.42938199639320374,
      "learning_rate": 7.089860686089855e-06,
      "loss": 0.2873,
      "step": 24547
    },
    {
      "epoch": 1.126520122986554,
      "grad_norm": 0.516125500202179,
      "learning_rate": 7.089637940208801e-06,
      "loss": 0.4387,
      "step": 24548
    },
    {
      "epoch": 1.1265660134918085,
      "grad_norm": 0.48361027240753174,
      "learning_rate": 7.089415189302829e-06,
      "loss": 0.4261,
      "step": 24549
    },
    {
      "epoch": 1.126611903997063,
      "grad_norm": 0.546575665473938,
      "learning_rate": 7.089192433372469e-06,
      "loss": 0.5349,
      "step": 24550
    },
    {
      "epoch": 1.1266577945023175,
      "grad_norm": 0.46081236004829407,
      "learning_rate": 7.088969672418262e-06,
      "loss": 0.3643,
      "step": 24551
    },
    {
      "epoch": 1.126703685007572,
      "grad_norm": 0.44108253717422485,
      "learning_rate": 7.088746906440744e-06,
      "loss": 0.3072,
      "step": 24552
    },
    {
      "epoch": 1.1267495755128265,
      "grad_norm": 0.46580585837364197,
      "learning_rate": 7.088524135440448e-06,
      "loss": 0.3513,
      "step": 24553
    },
    {
      "epoch": 1.1267954660180808,
      "grad_norm": 0.4716245234012604,
      "learning_rate": 7.088301359417911e-06,
      "loss": 0.3543,
      "step": 24554
    },
    {
      "epoch": 1.1268413565233353,
      "grad_norm": 0.5105167627334595,
      "learning_rate": 7.088078578373667e-06,
      "loss": 0.4201,
      "step": 24555
    },
    {
      "epoch": 1.1268872470285898,
      "grad_norm": 0.4353860914707184,
      "learning_rate": 7.087855792308252e-06,
      "loss": 0.3422,
      "step": 24556
    },
    {
      "epoch": 1.1269331375338443,
      "grad_norm": 0.44406914710998535,
      "learning_rate": 7.087633001222206e-06,
      "loss": 0.3169,
      "step": 24557
    },
    {
      "epoch": 1.1269790280390988,
      "grad_norm": 0.4783133864402771,
      "learning_rate": 7.08741020511606e-06,
      "loss": 0.3842,
      "step": 24558
    },
    {
      "epoch": 1.127024918544353,
      "grad_norm": 0.4782160222530365,
      "learning_rate": 7.087187403990351e-06,
      "loss": 0.3632,
      "step": 24559
    },
    {
      "epoch": 1.1270708090496075,
      "grad_norm": 0.4648149907588959,
      "learning_rate": 7.086964597845616e-06,
      "loss": 0.3127,
      "step": 24560
    },
    {
      "epoch": 1.127116699554862,
      "grad_norm": 0.5468202829360962,
      "learning_rate": 7.0867417866823875e-06,
      "loss": 0.4589,
      "step": 24561
    },
    {
      "epoch": 1.1271625900601165,
      "grad_norm": 0.46742403507232666,
      "learning_rate": 7.086518970501205e-06,
      "loss": 0.3526,
      "step": 24562
    },
    {
      "epoch": 1.127208480565371,
      "grad_norm": 0.46444547176361084,
      "learning_rate": 7.086296149302604e-06,
      "loss": 0.3873,
      "step": 24563
    },
    {
      "epoch": 1.1272543710706255,
      "grad_norm": 0.4539959132671356,
      "learning_rate": 7.086073323087118e-06,
      "loss": 0.3539,
      "step": 24564
    },
    {
      "epoch": 1.12730026157588,
      "grad_norm": 0.45937666296958923,
      "learning_rate": 7.085850491855284e-06,
      "loss": 0.3192,
      "step": 24565
    },
    {
      "epoch": 1.1273461520811345,
      "grad_norm": 0.46128958463668823,
      "learning_rate": 7.085627655607639e-06,
      "loss": 0.3579,
      "step": 24566
    },
    {
      "epoch": 1.1273920425863888,
      "grad_norm": 0.4815141558647156,
      "learning_rate": 7.085404814344716e-06,
      "loss": 0.3983,
      "step": 24567
    },
    {
      "epoch": 1.1274379330916433,
      "grad_norm": 0.46070626378059387,
      "learning_rate": 7.085181968067054e-06,
      "loss": 0.3242,
      "step": 24568
    },
    {
      "epoch": 1.1274838235968978,
      "grad_norm": 0.4765242040157318,
      "learning_rate": 7.084959116775187e-06,
      "loss": 0.3443,
      "step": 24569
    },
    {
      "epoch": 1.1275297141021523,
      "grad_norm": 0.47750386595726013,
      "learning_rate": 7.084736260469651e-06,
      "loss": 0.3533,
      "step": 24570
    },
    {
      "epoch": 1.1275756046074068,
      "grad_norm": 0.4724601209163666,
      "learning_rate": 7.084513399150983e-06,
      "loss": 0.3244,
      "step": 24571
    },
    {
      "epoch": 1.127621495112661,
      "grad_norm": 0.44304022192955017,
      "learning_rate": 7.084290532819718e-06,
      "loss": 0.3051,
      "step": 24572
    },
    {
      "epoch": 1.1276673856179156,
      "grad_norm": 0.4654863774776459,
      "learning_rate": 7.08406766147639e-06,
      "loss": 0.3769,
      "step": 24573
    },
    {
      "epoch": 1.12771327612317,
      "grad_norm": 0.4668884873390198,
      "learning_rate": 7.08384478512154e-06,
      "loss": 0.3638,
      "step": 24574
    },
    {
      "epoch": 1.1277591666284246,
      "grad_norm": 0.4853515326976776,
      "learning_rate": 7.0836219037557e-06,
      "loss": 0.3994,
      "step": 24575
    },
    {
      "epoch": 1.127805057133679,
      "grad_norm": 0.46285679936408997,
      "learning_rate": 7.083399017379406e-06,
      "loss": 0.3254,
      "step": 24576
    },
    {
      "epoch": 1.1278509476389336,
      "grad_norm": 0.4762583076953888,
      "learning_rate": 7.083176125993196e-06,
      "loss": 0.4098,
      "step": 24577
    },
    {
      "epoch": 1.127896838144188,
      "grad_norm": 0.4493516683578491,
      "learning_rate": 7.082953229597604e-06,
      "loss": 0.3506,
      "step": 24578
    },
    {
      "epoch": 1.1279427286494423,
      "grad_norm": 0.45163342356681824,
      "learning_rate": 7.082730328193168e-06,
      "loss": 0.336,
      "step": 24579
    },
    {
      "epoch": 1.1279886191546968,
      "grad_norm": 0.4574577808380127,
      "learning_rate": 7.082507421780423e-06,
      "loss": 0.3451,
      "step": 24580
    },
    {
      "epoch": 1.1280345096599513,
      "grad_norm": 0.47335854172706604,
      "learning_rate": 7.082284510359903e-06,
      "loss": 0.4202,
      "step": 24581
    },
    {
      "epoch": 1.1280804001652058,
      "grad_norm": 0.5153486728668213,
      "learning_rate": 7.082061593932148e-06,
      "loss": 0.3198,
      "step": 24582
    },
    {
      "epoch": 1.1281262906704603,
      "grad_norm": 0.4294804632663727,
      "learning_rate": 7.081838672497691e-06,
      "loss": 0.2976,
      "step": 24583
    },
    {
      "epoch": 1.1281721811757148,
      "grad_norm": 0.4901289939880371,
      "learning_rate": 7.081615746057069e-06,
      "loss": 0.3994,
      "step": 24584
    },
    {
      "epoch": 1.128218071680969,
      "grad_norm": 0.48187538981437683,
      "learning_rate": 7.081392814610819e-06,
      "loss": 0.3546,
      "step": 24585
    },
    {
      "epoch": 1.1282639621862236,
      "grad_norm": 0.43343669176101685,
      "learning_rate": 7.081169878159475e-06,
      "loss": 0.3408,
      "step": 24586
    },
    {
      "epoch": 1.128309852691478,
      "grad_norm": 0.47424763441085815,
      "learning_rate": 7.080946936703575e-06,
      "loss": 0.3516,
      "step": 24587
    },
    {
      "epoch": 1.1283557431967326,
      "grad_norm": 0.4683694541454315,
      "learning_rate": 7.080723990243654e-06,
      "loss": 0.2996,
      "step": 24588
    },
    {
      "epoch": 1.128401633701987,
      "grad_norm": 0.4467673897743225,
      "learning_rate": 7.080501038780249e-06,
      "loss": 0.2991,
      "step": 24589
    },
    {
      "epoch": 1.1284475242072416,
      "grad_norm": 0.4705336093902588,
      "learning_rate": 7.080278082313895e-06,
      "loss": 0.3772,
      "step": 24590
    },
    {
      "epoch": 1.128493414712496,
      "grad_norm": 0.4445994198322296,
      "learning_rate": 7.08005512084513e-06,
      "loss": 0.2934,
      "step": 24591
    },
    {
      "epoch": 1.1285393052177504,
      "grad_norm": 0.46373361349105835,
      "learning_rate": 7.079832154374488e-06,
      "loss": 0.335,
      "step": 24592
    },
    {
      "epoch": 1.1285851957230049,
      "grad_norm": 0.4306647479534149,
      "learning_rate": 7.079609182902505e-06,
      "loss": 0.3341,
      "step": 24593
    },
    {
      "epoch": 1.1286310862282594,
      "grad_norm": 0.49501678347587585,
      "learning_rate": 7.079386206429719e-06,
      "loss": 0.3928,
      "step": 24594
    },
    {
      "epoch": 1.1286769767335139,
      "grad_norm": 0.4977308213710785,
      "learning_rate": 7.079163224956666e-06,
      "loss": 0.4154,
      "step": 24595
    },
    {
      "epoch": 1.1287228672387684,
      "grad_norm": 0.46642938256263733,
      "learning_rate": 7.078940238483881e-06,
      "loss": 0.3263,
      "step": 24596
    },
    {
      "epoch": 1.1287687577440229,
      "grad_norm": 0.4599094092845917,
      "learning_rate": 7.078717247011903e-06,
      "loss": 0.3362,
      "step": 24597
    },
    {
      "epoch": 1.1288146482492771,
      "grad_norm": 0.4833031892776489,
      "learning_rate": 7.0784942505412636e-06,
      "loss": 0.3624,
      "step": 24598
    },
    {
      "epoch": 1.1288605387545316,
      "grad_norm": 0.5002352595329285,
      "learning_rate": 7.078271249072502e-06,
      "loss": 0.3685,
      "step": 24599
    },
    {
      "epoch": 1.1289064292597861,
      "grad_norm": 0.4528098702430725,
      "learning_rate": 7.078048242606154e-06,
      "loss": 0.3352,
      "step": 24600
    },
    {
      "epoch": 1.1289523197650406,
      "grad_norm": 0.49634674191474915,
      "learning_rate": 7.077825231142756e-06,
      "loss": 0.417,
      "step": 24601
    },
    {
      "epoch": 1.1289982102702951,
      "grad_norm": 0.4813739061355591,
      "learning_rate": 7.0776022146828445e-06,
      "loss": 0.3136,
      "step": 24602
    },
    {
      "epoch": 1.1290441007755496,
      "grad_norm": 0.46177199482917786,
      "learning_rate": 7.077379193226955e-06,
      "loss": 0.3851,
      "step": 24603
    },
    {
      "epoch": 1.1290899912808041,
      "grad_norm": 0.4620695412158966,
      "learning_rate": 7.077156166775624e-06,
      "loss": 0.3662,
      "step": 24604
    },
    {
      "epoch": 1.1291358817860584,
      "grad_norm": 0.46620461344718933,
      "learning_rate": 7.076933135329388e-06,
      "loss": 0.3399,
      "step": 24605
    },
    {
      "epoch": 1.129181772291313,
      "grad_norm": 0.47636425495147705,
      "learning_rate": 7.0767100988887836e-06,
      "loss": 0.3644,
      "step": 24606
    },
    {
      "epoch": 1.1292276627965674,
      "grad_norm": 0.493269145488739,
      "learning_rate": 7.076487057454347e-06,
      "loss": 0.4014,
      "step": 24607
    },
    {
      "epoch": 1.1292735533018219,
      "grad_norm": 0.47368744015693665,
      "learning_rate": 7.076264011026615e-06,
      "loss": 0.3567,
      "step": 24608
    },
    {
      "epoch": 1.1293194438070764,
      "grad_norm": 0.4601665735244751,
      "learning_rate": 7.07604095960612e-06,
      "loss": 0.3628,
      "step": 24609
    },
    {
      "epoch": 1.1293653343123307,
      "grad_norm": 0.4633275866508484,
      "learning_rate": 7.075817903193405e-06,
      "loss": 0.3639,
      "step": 24610
    },
    {
      "epoch": 1.1294112248175852,
      "grad_norm": 0.46169790625572205,
      "learning_rate": 7.075594841789003e-06,
      "loss": 0.3656,
      "step": 24611
    },
    {
      "epoch": 1.1294571153228397,
      "grad_norm": 0.5021532773971558,
      "learning_rate": 7.07537177539345e-06,
      "loss": 0.4629,
      "step": 24612
    },
    {
      "epoch": 1.1295030058280942,
      "grad_norm": 0.42406079173088074,
      "learning_rate": 7.075148704007282e-06,
      "loss": 0.2943,
      "step": 24613
    },
    {
      "epoch": 1.1295488963333487,
      "grad_norm": 0.46010372042655945,
      "learning_rate": 7.0749256276310376e-06,
      "loss": 0.3297,
      "step": 24614
    },
    {
      "epoch": 1.1295947868386031,
      "grad_norm": 0.5121824741363525,
      "learning_rate": 7.074702546265251e-06,
      "loss": 0.445,
      "step": 24615
    },
    {
      "epoch": 1.1296406773438576,
      "grad_norm": 0.49617666006088257,
      "learning_rate": 7.0744794599104595e-06,
      "loss": 0.3647,
      "step": 24616
    },
    {
      "epoch": 1.129686567849112,
      "grad_norm": 0.4901573956012726,
      "learning_rate": 7.074256368567201e-06,
      "loss": 0.3855,
      "step": 24617
    },
    {
      "epoch": 1.1297324583543664,
      "grad_norm": 0.4987618923187256,
      "learning_rate": 7.07403327223601e-06,
      "loss": 0.4014,
      "step": 24618
    },
    {
      "epoch": 1.129778348859621,
      "grad_norm": 0.46302151679992676,
      "learning_rate": 7.073810170917422e-06,
      "loss": 0.3687,
      "step": 24619
    },
    {
      "epoch": 1.1298242393648754,
      "grad_norm": 0.45192191004753113,
      "learning_rate": 7.073587064611976e-06,
      "loss": 0.3411,
      "step": 24620
    },
    {
      "epoch": 1.12987012987013,
      "grad_norm": 0.45732104778289795,
      "learning_rate": 7.073363953320207e-06,
      "loss": 0.2982,
      "step": 24621
    },
    {
      "epoch": 1.1299160203753844,
      "grad_norm": 0.47261545062065125,
      "learning_rate": 7.073140837042654e-06,
      "loss": 0.3621,
      "step": 24622
    },
    {
      "epoch": 1.1299619108806387,
      "grad_norm": 0.42634132504463196,
      "learning_rate": 7.07291771577985e-06,
      "loss": 0.3076,
      "step": 24623
    },
    {
      "epoch": 1.1300078013858932,
      "grad_norm": 0.4517354965209961,
      "learning_rate": 7.072694589532334e-06,
      "loss": 0.3536,
      "step": 24624
    },
    {
      "epoch": 1.1300536918911477,
      "grad_norm": 0.4304409623146057,
      "learning_rate": 7.072471458300642e-06,
      "loss": 0.3022,
      "step": 24625
    },
    {
      "epoch": 1.1300995823964022,
      "grad_norm": 0.43409183621406555,
      "learning_rate": 7.072248322085308e-06,
      "loss": 0.3677,
      "step": 24626
    },
    {
      "epoch": 1.1301454729016567,
      "grad_norm": 0.4629223942756653,
      "learning_rate": 7.072025180886872e-06,
      "loss": 0.3385,
      "step": 24627
    },
    {
      "epoch": 1.1301913634069112,
      "grad_norm": 0.47877368330955505,
      "learning_rate": 7.07180203470587e-06,
      "loss": 0.4395,
      "step": 24628
    },
    {
      "epoch": 1.1302372539121657,
      "grad_norm": 0.4613826274871826,
      "learning_rate": 7.071578883542838e-06,
      "loss": 0.3599,
      "step": 24629
    },
    {
      "epoch": 1.13028314441742,
      "grad_norm": 0.47733768820762634,
      "learning_rate": 7.071355727398311e-06,
      "loss": 0.3733,
      "step": 24630
    },
    {
      "epoch": 1.1303290349226744,
      "grad_norm": 0.48830389976501465,
      "learning_rate": 7.071132566272829e-06,
      "loss": 0.4233,
      "step": 24631
    },
    {
      "epoch": 1.130374925427929,
      "grad_norm": 0.4276810884475708,
      "learning_rate": 7.070909400166926e-06,
      "loss": 0.2697,
      "step": 24632
    },
    {
      "epoch": 1.1304208159331834,
      "grad_norm": 0.49892228841781616,
      "learning_rate": 7.07068622908114e-06,
      "loss": 0.4182,
      "step": 24633
    },
    {
      "epoch": 1.130466706438438,
      "grad_norm": 0.508402407169342,
      "learning_rate": 7.070463053016008e-06,
      "loss": 0.3765,
      "step": 24634
    },
    {
      "epoch": 1.1305125969436924,
      "grad_norm": 0.49524861574172974,
      "learning_rate": 7.070239871972064e-06,
      "loss": 0.4445,
      "step": 24635
    },
    {
      "epoch": 1.1305584874489467,
      "grad_norm": 0.45710158348083496,
      "learning_rate": 7.070016685949848e-06,
      "loss": 0.3204,
      "step": 24636
    },
    {
      "epoch": 1.1306043779542012,
      "grad_norm": 0.5117238163948059,
      "learning_rate": 7.069793494949894e-06,
      "loss": 0.3658,
      "step": 24637
    },
    {
      "epoch": 1.1306502684594557,
      "grad_norm": 0.4672723710536957,
      "learning_rate": 7.069570298972739e-06,
      "loss": 0.3298,
      "step": 24638
    },
    {
      "epoch": 1.1306961589647102,
      "grad_norm": 0.494704008102417,
      "learning_rate": 7.0693470980189225e-06,
      "loss": 0.4084,
      "step": 24639
    },
    {
      "epoch": 1.1307420494699647,
      "grad_norm": 0.464688241481781,
      "learning_rate": 7.06912389208898e-06,
      "loss": 0.3543,
      "step": 24640
    },
    {
      "epoch": 1.1307879399752192,
      "grad_norm": 0.4941105246543884,
      "learning_rate": 7.068900681183445e-06,
      "loss": 0.3867,
      "step": 24641
    },
    {
      "epoch": 1.1308338304804737,
      "grad_norm": 0.4670950472354889,
      "learning_rate": 7.068677465302858e-06,
      "loss": 0.3907,
      "step": 24642
    },
    {
      "epoch": 1.130879720985728,
      "grad_norm": 0.4825858473777771,
      "learning_rate": 7.068454244447754e-06,
      "loss": 0.3553,
      "step": 24643
    },
    {
      "epoch": 1.1309256114909825,
      "grad_norm": 0.44604653120040894,
      "learning_rate": 7.068231018618671e-06,
      "loss": 0.3232,
      "step": 24644
    },
    {
      "epoch": 1.130971501996237,
      "grad_norm": 0.4585740566253662,
      "learning_rate": 7.068007787816147e-06,
      "loss": 0.3514,
      "step": 24645
    },
    {
      "epoch": 1.1310173925014915,
      "grad_norm": 0.4931480586528778,
      "learning_rate": 7.067784552040715e-06,
      "loss": 0.3383,
      "step": 24646
    },
    {
      "epoch": 1.131063283006746,
      "grad_norm": 0.4740438759326935,
      "learning_rate": 7.067561311292913e-06,
      "loss": 0.3639,
      "step": 24647
    },
    {
      "epoch": 1.1311091735120002,
      "grad_norm": 0.44726133346557617,
      "learning_rate": 7.067338065573279e-06,
      "loss": 0.2799,
      "step": 24648
    },
    {
      "epoch": 1.1311550640172547,
      "grad_norm": 0.46154269576072693,
      "learning_rate": 7.06711481488235e-06,
      "loss": 0.3221,
      "step": 24649
    },
    {
      "epoch": 1.1312009545225092,
      "grad_norm": 0.5094354152679443,
      "learning_rate": 7.0668915592206635e-06,
      "loss": 0.4388,
      "step": 24650
    },
    {
      "epoch": 1.1312468450277637,
      "grad_norm": 0.46852555871009827,
      "learning_rate": 7.0666682985887536e-06,
      "loss": 0.3584,
      "step": 24651
    },
    {
      "epoch": 1.1312927355330182,
      "grad_norm": 0.46889567375183105,
      "learning_rate": 7.066445032987159e-06,
      "loss": 0.3898,
      "step": 24652
    },
    {
      "epoch": 1.1313386260382727,
      "grad_norm": 0.522243857383728,
      "learning_rate": 7.066221762416417e-06,
      "loss": 0.4256,
      "step": 24653
    },
    {
      "epoch": 1.1313845165435272,
      "grad_norm": 0.4601024389266968,
      "learning_rate": 7.0659984868770615e-06,
      "loss": 0.3743,
      "step": 24654
    },
    {
      "epoch": 1.1314304070487817,
      "grad_norm": 0.47413766384124756,
      "learning_rate": 7.065775206369633e-06,
      "loss": 0.3633,
      "step": 24655
    },
    {
      "epoch": 1.131476297554036,
      "grad_norm": 0.4264897406101227,
      "learning_rate": 7.065551920894669e-06,
      "loss": 0.2879,
      "step": 24656
    },
    {
      "epoch": 1.1315221880592905,
      "grad_norm": 0.46694231033325195,
      "learning_rate": 7.065328630452702e-06,
      "loss": 0.345,
      "step": 24657
    },
    {
      "epoch": 1.131568078564545,
      "grad_norm": 0.47167566418647766,
      "learning_rate": 7.065105335044271e-06,
      "loss": 0.3482,
      "step": 24658
    },
    {
      "epoch": 1.1316139690697995,
      "grad_norm": 0.4649851620197296,
      "learning_rate": 7.0648820346699156e-06,
      "loss": 0.3561,
      "step": 24659
    },
    {
      "epoch": 1.131659859575054,
      "grad_norm": 0.5117760896682739,
      "learning_rate": 7.064658729330169e-06,
      "loss": 0.4228,
      "step": 24660
    },
    {
      "epoch": 1.1317057500803083,
      "grad_norm": 0.46339282393455505,
      "learning_rate": 7.064435419025571e-06,
      "loss": 0.367,
      "step": 24661
    },
    {
      "epoch": 1.1317516405855628,
      "grad_norm": 0.48824307322502136,
      "learning_rate": 7.064212103756658e-06,
      "loss": 0.4471,
      "step": 24662
    },
    {
      "epoch": 1.1317975310908173,
      "grad_norm": 0.4682967960834503,
      "learning_rate": 7.063988783523965e-06,
      "loss": 0.3439,
      "step": 24663
    },
    {
      "epoch": 1.1318434215960718,
      "grad_norm": 0.45112791657447815,
      "learning_rate": 7.06376545832803e-06,
      "loss": 0.3571,
      "step": 24664
    },
    {
      "epoch": 1.1318893121013263,
      "grad_norm": 0.4737400412559509,
      "learning_rate": 7.063542128169392e-06,
      "loss": 0.3604,
      "step": 24665
    },
    {
      "epoch": 1.1319352026065808,
      "grad_norm": 0.46167582273483276,
      "learning_rate": 7.063318793048585e-06,
      "loss": 0.3604,
      "step": 24666
    },
    {
      "epoch": 1.1319810931118353,
      "grad_norm": 0.47838419675827026,
      "learning_rate": 7.063095452966149e-06,
      "loss": 0.3796,
      "step": 24667
    },
    {
      "epoch": 1.1320269836170895,
      "grad_norm": 0.44395148754119873,
      "learning_rate": 7.062872107922618e-06,
      "loss": 0.3372,
      "step": 24668
    },
    {
      "epoch": 1.132072874122344,
      "grad_norm": 0.4389018714427948,
      "learning_rate": 7.062648757918531e-06,
      "loss": 0.3084,
      "step": 24669
    },
    {
      "epoch": 1.1321187646275985,
      "grad_norm": 0.4572465121746063,
      "learning_rate": 7.062425402954425e-06,
      "loss": 0.3364,
      "step": 24670
    },
    {
      "epoch": 1.132164655132853,
      "grad_norm": 0.48669055104255676,
      "learning_rate": 7.062202043030837e-06,
      "loss": 0.3843,
      "step": 24671
    },
    {
      "epoch": 1.1322105456381075,
      "grad_norm": 0.46957796812057495,
      "learning_rate": 7.061978678148304e-06,
      "loss": 0.3954,
      "step": 24672
    },
    {
      "epoch": 1.132256436143362,
      "grad_norm": 0.47031453251838684,
      "learning_rate": 7.061755308307363e-06,
      "loss": 0.3608,
      "step": 24673
    },
    {
      "epoch": 1.1323023266486163,
      "grad_norm": 0.4731173813343048,
      "learning_rate": 7.061531933508551e-06,
      "loss": 0.3793,
      "step": 24674
    },
    {
      "epoch": 1.1323482171538708,
      "grad_norm": 0.5126436948776245,
      "learning_rate": 7.061308553752404e-06,
      "loss": 0.4407,
      "step": 24675
    },
    {
      "epoch": 1.1323941076591253,
      "grad_norm": 0.5006262063980103,
      "learning_rate": 7.061085169039463e-06,
      "loss": 0.4251,
      "step": 24676
    },
    {
      "epoch": 1.1324399981643798,
      "grad_norm": 0.47339576482772827,
      "learning_rate": 7.06086177937026e-06,
      "loss": 0.405,
      "step": 24677
    },
    {
      "epoch": 1.1324858886696343,
      "grad_norm": 0.46169745922088623,
      "learning_rate": 7.0606383847453374e-06,
      "loss": 0.332,
      "step": 24678
    },
    {
      "epoch": 1.1325317791748888,
      "grad_norm": 0.49350640177726746,
      "learning_rate": 7.06041498516523e-06,
      "loss": 0.3898,
      "step": 24679
    },
    {
      "epoch": 1.1325776696801433,
      "grad_norm": 0.4572957456111908,
      "learning_rate": 7.060191580630472e-06,
      "loss": 0.3527,
      "step": 24680
    },
    {
      "epoch": 1.1326235601853976,
      "grad_norm": 0.5221545100212097,
      "learning_rate": 7.0599681711416044e-06,
      "loss": 0.4074,
      "step": 24681
    },
    {
      "epoch": 1.132669450690652,
      "grad_norm": 0.5155816674232483,
      "learning_rate": 7.059744756699165e-06,
      "loss": 0.4505,
      "step": 24682
    },
    {
      "epoch": 1.1327153411959066,
      "grad_norm": 0.463096022605896,
      "learning_rate": 7.05952133730369e-06,
      "loss": 0.36,
      "step": 24683
    },
    {
      "epoch": 1.132761231701161,
      "grad_norm": 0.4870770573616028,
      "learning_rate": 7.059297912955715e-06,
      "loss": 0.3825,
      "step": 24684
    },
    {
      "epoch": 1.1328071222064156,
      "grad_norm": 0.42900702357292175,
      "learning_rate": 7.059074483655778e-06,
      "loss": 0.254,
      "step": 24685
    },
    {
      "epoch": 1.13285301271167,
      "grad_norm": 0.4253750443458557,
      "learning_rate": 7.058851049404416e-06,
      "loss": 0.262,
      "step": 24686
    },
    {
      "epoch": 1.1328989032169243,
      "grad_norm": 0.43896132707595825,
      "learning_rate": 7.058627610202169e-06,
      "loss": 0.3172,
      "step": 24687
    },
    {
      "epoch": 1.1329447937221788,
      "grad_norm": 0.5336152911186218,
      "learning_rate": 7.05840416604957e-06,
      "loss": 0.471,
      "step": 24688
    },
    {
      "epoch": 1.1329906842274333,
      "grad_norm": 0.479544460773468,
      "learning_rate": 7.05818071694716e-06,
      "loss": 0.366,
      "step": 24689
    },
    {
      "epoch": 1.1330365747326878,
      "grad_norm": 0.4719175398349762,
      "learning_rate": 7.057957262895476e-06,
      "loss": 0.3912,
      "step": 24690
    },
    {
      "epoch": 1.1330824652379423,
      "grad_norm": 0.4595673978328705,
      "learning_rate": 7.057733803895054e-06,
      "loss": 0.3402,
      "step": 24691
    },
    {
      "epoch": 1.1331283557431968,
      "grad_norm": 0.4903862476348877,
      "learning_rate": 7.05751033994643e-06,
      "loss": 0.396,
      "step": 24692
    },
    {
      "epoch": 1.1331742462484513,
      "grad_norm": 0.4468924403190613,
      "learning_rate": 7.057286871050144e-06,
      "loss": 0.3191,
      "step": 24693
    },
    {
      "epoch": 1.1332201367537056,
      "grad_norm": 0.4685289263725281,
      "learning_rate": 7.0570633972067335e-06,
      "loss": 0.3379,
      "step": 24694
    },
    {
      "epoch": 1.13326602725896,
      "grad_norm": 0.5066692233085632,
      "learning_rate": 7.056839918416733e-06,
      "loss": 0.4151,
      "step": 24695
    },
    {
      "epoch": 1.1333119177642146,
      "grad_norm": 0.48665502667427063,
      "learning_rate": 7.056616434680682e-06,
      "loss": 0.3849,
      "step": 24696
    },
    {
      "epoch": 1.133357808269469,
      "grad_norm": 0.4585694968700409,
      "learning_rate": 7.056392945999118e-06,
      "loss": 0.3672,
      "step": 24697
    },
    {
      "epoch": 1.1334036987747236,
      "grad_norm": 0.4581230878829956,
      "learning_rate": 7.056169452372577e-06,
      "loss": 0.366,
      "step": 24698
    },
    {
      "epoch": 1.1334495892799779,
      "grad_norm": 0.4281544089317322,
      "learning_rate": 7.055945953801599e-06,
      "loss": 0.3177,
      "step": 24699
    },
    {
      "epoch": 1.1334954797852324,
      "grad_norm": 0.48600640892982483,
      "learning_rate": 7.05572245028672e-06,
      "loss": 0.371,
      "step": 24700
    },
    {
      "epoch": 1.1335413702904868,
      "grad_norm": 0.4628463685512543,
      "learning_rate": 7.055498941828476e-06,
      "loss": 0.3471,
      "step": 24701
    },
    {
      "epoch": 1.1335872607957413,
      "grad_norm": 0.45399749279022217,
      "learning_rate": 7.055275428427406e-06,
      "loss": 0.323,
      "step": 24702
    },
    {
      "epoch": 1.1336331513009958,
      "grad_norm": 0.48110607266426086,
      "learning_rate": 7.055051910084047e-06,
      "loss": 0.4187,
      "step": 24703
    },
    {
      "epoch": 1.1336790418062503,
      "grad_norm": 0.4542180001735687,
      "learning_rate": 7.054828386798938e-06,
      "loss": 0.3465,
      "step": 24704
    },
    {
      "epoch": 1.1337249323115048,
      "grad_norm": 0.47487661242485046,
      "learning_rate": 7.0546048585726136e-06,
      "loss": 0.3824,
      "step": 24705
    },
    {
      "epoch": 1.1337708228167591,
      "grad_norm": 0.4624318480491638,
      "learning_rate": 7.054381325405614e-06,
      "loss": 0.3635,
      "step": 24706
    },
    {
      "epoch": 1.1338167133220136,
      "grad_norm": 0.44230806827545166,
      "learning_rate": 7.054157787298475e-06,
      "loss": 0.3511,
      "step": 24707
    },
    {
      "epoch": 1.1338626038272681,
      "grad_norm": 0.43595731258392334,
      "learning_rate": 7.0539342442517344e-06,
      "loss": 0.3402,
      "step": 24708
    },
    {
      "epoch": 1.1339084943325226,
      "grad_norm": 0.44333600997924805,
      "learning_rate": 7.053710696265932e-06,
      "loss": 0.313,
      "step": 24709
    },
    {
      "epoch": 1.133954384837777,
      "grad_norm": 0.5078650116920471,
      "learning_rate": 7.0534871433416016e-06,
      "loss": 0.3975,
      "step": 24710
    },
    {
      "epoch": 1.1340002753430316,
      "grad_norm": 0.4935639202594757,
      "learning_rate": 7.053263585479285e-06,
      "loss": 0.3834,
      "step": 24711
    },
    {
      "epoch": 1.1340461658482859,
      "grad_norm": 0.4503193795681,
      "learning_rate": 7.053040022679516e-06,
      "loss": 0.3036,
      "step": 24712
    },
    {
      "epoch": 1.1340920563535404,
      "grad_norm": 0.44749170541763306,
      "learning_rate": 7.052816454942834e-06,
      "loss": 0.3425,
      "step": 24713
    },
    {
      "epoch": 1.1341379468587949,
      "grad_norm": 0.47759994864463806,
      "learning_rate": 7.052592882269775e-06,
      "loss": 0.35,
      "step": 24714
    },
    {
      "epoch": 1.1341838373640494,
      "grad_norm": 0.4530743956565857,
      "learning_rate": 7.05236930466088e-06,
      "loss": 0.3216,
      "step": 24715
    },
    {
      "epoch": 1.1342297278693039,
      "grad_norm": 0.45324352383613586,
      "learning_rate": 7.052145722116684e-06,
      "loss": 0.3327,
      "step": 24716
    },
    {
      "epoch": 1.1342756183745584,
      "grad_norm": 0.4964286983013153,
      "learning_rate": 7.051922134637725e-06,
      "loss": 0.404,
      "step": 24717
    },
    {
      "epoch": 1.1343215088798129,
      "grad_norm": 0.4560413658618927,
      "learning_rate": 7.0516985422245406e-06,
      "loss": 0.3425,
      "step": 24718
    },
    {
      "epoch": 1.1343673993850671,
      "grad_norm": 0.4483110308647156,
      "learning_rate": 7.05147494487767e-06,
      "loss": 0.361,
      "step": 24719
    },
    {
      "epoch": 1.1344132898903216,
      "grad_norm": 0.4693838357925415,
      "learning_rate": 7.051251342597647e-06,
      "loss": 0.3028,
      "step": 24720
    },
    {
      "epoch": 1.1344591803955761,
      "grad_norm": 0.47053030133247375,
      "learning_rate": 7.051027735385014e-06,
      "loss": 0.348,
      "step": 24721
    },
    {
      "epoch": 1.1345050709008306,
      "grad_norm": 0.4874568283557892,
      "learning_rate": 7.050804123240307e-06,
      "loss": 0.3846,
      "step": 24722
    },
    {
      "epoch": 1.1345509614060851,
      "grad_norm": 0.535033106803894,
      "learning_rate": 7.050580506164062e-06,
      "loss": 0.4384,
      "step": 24723
    },
    {
      "epoch": 1.1345968519113396,
      "grad_norm": 0.4514215588569641,
      "learning_rate": 7.050356884156819e-06,
      "loss": 0.3333,
      "step": 24724
    },
    {
      "epoch": 1.134642742416594,
      "grad_norm": 0.4870937764644623,
      "learning_rate": 7.050133257219114e-06,
      "loss": 0.3837,
      "step": 24725
    },
    {
      "epoch": 1.1346886329218484,
      "grad_norm": 0.47671422362327576,
      "learning_rate": 7.049909625351486e-06,
      "loss": 0.3535,
      "step": 24726
    },
    {
      "epoch": 1.134734523427103,
      "grad_norm": 0.4436316192150116,
      "learning_rate": 7.049685988554473e-06,
      "loss": 0.3342,
      "step": 24727
    },
    {
      "epoch": 1.1347804139323574,
      "grad_norm": 0.4829087555408478,
      "learning_rate": 7.04946234682861e-06,
      "loss": 0.4447,
      "step": 24728
    },
    {
      "epoch": 1.134826304437612,
      "grad_norm": 0.43359872698783875,
      "learning_rate": 7.049238700174439e-06,
      "loss": 0.3305,
      "step": 24729
    },
    {
      "epoch": 1.1348721949428664,
      "grad_norm": 0.45679810643196106,
      "learning_rate": 7.049015048592495e-06,
      "loss": 0.363,
      "step": 24730
    },
    {
      "epoch": 1.134918085448121,
      "grad_norm": 0.43866032361984253,
      "learning_rate": 7.048791392083316e-06,
      "loss": 0.2955,
      "step": 24731
    },
    {
      "epoch": 1.1349639759533752,
      "grad_norm": 0.4772982597351074,
      "learning_rate": 7.048567730647441e-06,
      "loss": 0.348,
      "step": 24732
    },
    {
      "epoch": 1.1350098664586297,
      "grad_norm": 0.45870426297187805,
      "learning_rate": 7.0483440642854074e-06,
      "loss": 0.3746,
      "step": 24733
    },
    {
      "epoch": 1.1350557569638842,
      "grad_norm": 0.4722480773925781,
      "learning_rate": 7.048120392997751e-06,
      "loss": 0.3447,
      "step": 24734
    },
    {
      "epoch": 1.1351016474691387,
      "grad_norm": 0.4523755609989166,
      "learning_rate": 7.047896716785014e-06,
      "loss": 0.3116,
      "step": 24735
    },
    {
      "epoch": 1.1351475379743932,
      "grad_norm": 0.46340692043304443,
      "learning_rate": 7.0476730356477295e-06,
      "loss": 0.3528,
      "step": 24736
    },
    {
      "epoch": 1.1351934284796474,
      "grad_norm": 0.4610845446586609,
      "learning_rate": 7.04744934958644e-06,
      "loss": 0.3462,
      "step": 24737
    },
    {
      "epoch": 1.135239318984902,
      "grad_norm": 0.4838887155056,
      "learning_rate": 7.0472256586016794e-06,
      "loss": 0.4202,
      "step": 24738
    },
    {
      "epoch": 1.1352852094901564,
      "grad_norm": 0.4883697032928467,
      "learning_rate": 7.0470019626939865e-06,
      "loss": 0.3929,
      "step": 24739
    },
    {
      "epoch": 1.135331099995411,
      "grad_norm": 0.4319809377193451,
      "learning_rate": 7.046778261863901e-06,
      "loss": 0.284,
      "step": 24740
    },
    {
      "epoch": 1.1353769905006654,
      "grad_norm": 0.4681074321269989,
      "learning_rate": 7.0465545561119595e-06,
      "loss": 0.3536,
      "step": 24741
    },
    {
      "epoch": 1.13542288100592,
      "grad_norm": 0.4900055527687073,
      "learning_rate": 7.0463308454387e-06,
      "loss": 0.3611,
      "step": 24742
    },
    {
      "epoch": 1.1354687715111744,
      "grad_norm": 0.4769522249698639,
      "learning_rate": 7.046107129844661e-06,
      "loss": 0.36,
      "step": 24743
    },
    {
      "epoch": 1.135514662016429,
      "grad_norm": 0.4656541347503662,
      "learning_rate": 7.045883409330381e-06,
      "loss": 0.3462,
      "step": 24744
    },
    {
      "epoch": 1.1355605525216832,
      "grad_norm": 0.45617586374282837,
      "learning_rate": 7.045659683896395e-06,
      "loss": 0.3236,
      "step": 24745
    },
    {
      "epoch": 1.1356064430269377,
      "grad_norm": 0.4944656193256378,
      "learning_rate": 7.045435953543245e-06,
      "loss": 0.3934,
      "step": 24746
    },
    {
      "epoch": 1.1356523335321922,
      "grad_norm": 0.48666101694107056,
      "learning_rate": 7.045212218271466e-06,
      "loss": 0.3925,
      "step": 24747
    },
    {
      "epoch": 1.1356982240374467,
      "grad_norm": 0.47320905327796936,
      "learning_rate": 7.044988478081598e-06,
      "loss": 0.3339,
      "step": 24748
    },
    {
      "epoch": 1.1357441145427012,
      "grad_norm": 0.48678824305534363,
      "learning_rate": 7.044764732974179e-06,
      "loss": 0.3719,
      "step": 24749
    },
    {
      "epoch": 1.1357900050479555,
      "grad_norm": 0.454212486743927,
      "learning_rate": 7.044540982949743e-06,
      "loss": 0.3236,
      "step": 24750
    },
    {
      "epoch": 1.13583589555321,
      "grad_norm": 0.46318498253822327,
      "learning_rate": 7.044317228008833e-06,
      "loss": 0.3389,
      "step": 24751
    },
    {
      "epoch": 1.1358817860584645,
      "grad_norm": 0.4354458749294281,
      "learning_rate": 7.044093468151985e-06,
      "loss": 0.2534,
      "step": 24752
    },
    {
      "epoch": 1.135927676563719,
      "grad_norm": 0.46282947063446045,
      "learning_rate": 7.043869703379738e-06,
      "loss": 0.3546,
      "step": 24753
    },
    {
      "epoch": 1.1359735670689735,
      "grad_norm": 0.48586320877075195,
      "learning_rate": 7.0436459336926284e-06,
      "loss": 0.3845,
      "step": 24754
    },
    {
      "epoch": 1.136019457574228,
      "grad_norm": 0.4965175986289978,
      "learning_rate": 7.0434221590911964e-06,
      "loss": 0.4018,
      "step": 24755
    },
    {
      "epoch": 1.1360653480794825,
      "grad_norm": 0.4964290261268616,
      "learning_rate": 7.043198379575978e-06,
      "loss": 0.4209,
      "step": 24756
    },
    {
      "epoch": 1.1361112385847367,
      "grad_norm": 0.4509190022945404,
      "learning_rate": 7.042974595147513e-06,
      "loss": 0.3208,
      "step": 24757
    },
    {
      "epoch": 1.1361571290899912,
      "grad_norm": 0.4863429367542267,
      "learning_rate": 7.042750805806339e-06,
      "loss": 0.3863,
      "step": 24758
    },
    {
      "epoch": 1.1362030195952457,
      "grad_norm": 0.45994833111763,
      "learning_rate": 7.0425270115529945e-06,
      "loss": 0.3342,
      "step": 24759
    },
    {
      "epoch": 1.1362489101005002,
      "grad_norm": 0.45173588395118713,
      "learning_rate": 7.042303212388015e-06,
      "loss": 0.2975,
      "step": 24760
    },
    {
      "epoch": 1.1362948006057547,
      "grad_norm": 0.48274773359298706,
      "learning_rate": 7.042079408311943e-06,
      "loss": 0.3846,
      "step": 24761
    },
    {
      "epoch": 1.1363406911110092,
      "grad_norm": 0.4808579385280609,
      "learning_rate": 7.041855599325314e-06,
      "loss": 0.4071,
      "step": 24762
    },
    {
      "epoch": 1.1363865816162635,
      "grad_norm": 0.4286321997642517,
      "learning_rate": 7.0416317854286665e-06,
      "loss": 0.2862,
      "step": 24763
    },
    {
      "epoch": 1.136432472121518,
      "grad_norm": 0.5289644598960876,
      "learning_rate": 7.041407966622539e-06,
      "loss": 0.4298,
      "step": 24764
    },
    {
      "epoch": 1.1364783626267725,
      "grad_norm": 0.45408350229263306,
      "learning_rate": 7.04118414290747e-06,
      "loss": 0.3187,
      "step": 24765
    },
    {
      "epoch": 1.136524253132027,
      "grad_norm": 0.5313528180122375,
      "learning_rate": 7.040960314283998e-06,
      "loss": 0.4638,
      "step": 24766
    },
    {
      "epoch": 1.1365701436372815,
      "grad_norm": 0.4300351142883301,
      "learning_rate": 7.040736480752659e-06,
      "loss": 0.2891,
      "step": 24767
    },
    {
      "epoch": 1.136616034142536,
      "grad_norm": 0.4096066653728485,
      "learning_rate": 7.040512642313992e-06,
      "loss": 0.264,
      "step": 24768
    },
    {
      "epoch": 1.1366619246477905,
      "grad_norm": 0.4362323582172394,
      "learning_rate": 7.040288798968538e-06,
      "loss": 0.2998,
      "step": 24769
    },
    {
      "epoch": 1.1367078151530448,
      "grad_norm": 0.4990677833557129,
      "learning_rate": 7.040064950716834e-06,
      "loss": 0.3832,
      "step": 24770
    },
    {
      "epoch": 1.1367537056582993,
      "grad_norm": 0.4813346862792969,
      "learning_rate": 7.0398410975594155e-06,
      "loss": 0.3519,
      "step": 24771
    },
    {
      "epoch": 1.1367995961635537,
      "grad_norm": 0.4750055968761444,
      "learning_rate": 7.0396172394968245e-06,
      "loss": 0.354,
      "step": 24772
    },
    {
      "epoch": 1.1368454866688082,
      "grad_norm": 0.4687548875808716,
      "learning_rate": 7.039393376529596e-06,
      "loss": 0.3448,
      "step": 24773
    },
    {
      "epoch": 1.1368913771740627,
      "grad_norm": 0.470760315656662,
      "learning_rate": 7.039169508658271e-06,
      "loss": 0.3718,
      "step": 24774
    },
    {
      "epoch": 1.1369372676793172,
      "grad_norm": 0.5034664273262024,
      "learning_rate": 7.038945635883388e-06,
      "loss": 0.4164,
      "step": 24775
    },
    {
      "epoch": 1.1369831581845715,
      "grad_norm": 0.469431072473526,
      "learning_rate": 7.038721758205485e-06,
      "loss": 0.3913,
      "step": 24776
    },
    {
      "epoch": 1.137029048689826,
      "grad_norm": 0.4928151071071625,
      "learning_rate": 7.038497875625096e-06,
      "loss": 0.4353,
      "step": 24777
    },
    {
      "epoch": 1.1370749391950805,
      "grad_norm": 0.4459112584590912,
      "learning_rate": 7.038273988142767e-06,
      "loss": 0.2846,
      "step": 24778
    },
    {
      "epoch": 1.137120829700335,
      "grad_norm": 0.4745567739009857,
      "learning_rate": 7.038050095759029e-06,
      "loss": 0.3683,
      "step": 24779
    },
    {
      "epoch": 1.1371667202055895,
      "grad_norm": 0.4736587107181549,
      "learning_rate": 7.037826198474426e-06,
      "loss": 0.3141,
      "step": 24780
    },
    {
      "epoch": 1.137212610710844,
      "grad_norm": 0.5149346590042114,
      "learning_rate": 7.037602296289493e-06,
      "loss": 0.498,
      "step": 24781
    },
    {
      "epoch": 1.1372585012160985,
      "grad_norm": 0.495312362909317,
      "learning_rate": 7.037378389204769e-06,
      "loss": 0.3728,
      "step": 24782
    },
    {
      "epoch": 1.1373043917213528,
      "grad_norm": 0.4811500310897827,
      "learning_rate": 7.037154477220795e-06,
      "loss": 0.373,
      "step": 24783
    },
    {
      "epoch": 1.1373502822266073,
      "grad_norm": 0.4920310378074646,
      "learning_rate": 7.036930560338105e-06,
      "loss": 0.4206,
      "step": 24784
    },
    {
      "epoch": 1.1373961727318618,
      "grad_norm": 0.4230092167854309,
      "learning_rate": 7.0367066385572414e-06,
      "loss": 0.3127,
      "step": 24785
    },
    {
      "epoch": 1.1374420632371163,
      "grad_norm": 0.44914710521698,
      "learning_rate": 7.036482711878742e-06,
      "loss": 0.3539,
      "step": 24786
    },
    {
      "epoch": 1.1374879537423708,
      "grad_norm": 0.468374103307724,
      "learning_rate": 7.036258780303144e-06,
      "loss": 0.3465,
      "step": 24787
    },
    {
      "epoch": 1.137533844247625,
      "grad_norm": 0.5037892460823059,
      "learning_rate": 7.0360348438309846e-06,
      "loss": 0.4077,
      "step": 24788
    },
    {
      "epoch": 1.1375797347528795,
      "grad_norm": 0.49517032504081726,
      "learning_rate": 7.035810902462805e-06,
      "loss": 0.4562,
      "step": 24789
    },
    {
      "epoch": 1.137625625258134,
      "grad_norm": 0.4721580743789673,
      "learning_rate": 7.035586956199142e-06,
      "loss": 0.4031,
      "step": 24790
    },
    {
      "epoch": 1.1376715157633885,
      "grad_norm": 0.4820847511291504,
      "learning_rate": 7.035363005040536e-06,
      "loss": 0.3499,
      "step": 24791
    },
    {
      "epoch": 1.137717406268643,
      "grad_norm": 0.4424426257610321,
      "learning_rate": 7.035139048987525e-06,
      "loss": 0.3095,
      "step": 24792
    },
    {
      "epoch": 1.1377632967738975,
      "grad_norm": 0.47277045249938965,
      "learning_rate": 7.034915088040645e-06,
      "loss": 0.3207,
      "step": 24793
    },
    {
      "epoch": 1.137809187279152,
      "grad_norm": 0.4357203245162964,
      "learning_rate": 7.034691122200436e-06,
      "loss": 0.3053,
      "step": 24794
    },
    {
      "epoch": 1.1378550777844063,
      "grad_norm": 0.47035279870033264,
      "learning_rate": 7.034467151467439e-06,
      "loss": 0.3321,
      "step": 24795
    },
    {
      "epoch": 1.1379009682896608,
      "grad_norm": 0.4522012770175934,
      "learning_rate": 7.0342431758421896e-06,
      "loss": 0.3442,
      "step": 24796
    },
    {
      "epoch": 1.1379468587949153,
      "grad_norm": 0.46848559379577637,
      "learning_rate": 7.034019195325228e-06,
      "loss": 0.3866,
      "step": 24797
    },
    {
      "epoch": 1.1379927493001698,
      "grad_norm": 0.4865160882472992,
      "learning_rate": 7.033795209917093e-06,
      "loss": 0.3506,
      "step": 24798
    },
    {
      "epoch": 1.1380386398054243,
      "grad_norm": 0.5198656916618347,
      "learning_rate": 7.03357121961832e-06,
      "loss": 0.3868,
      "step": 24799
    },
    {
      "epoch": 1.1380845303106788,
      "grad_norm": 0.44505733251571655,
      "learning_rate": 7.033347224429452e-06,
      "loss": 0.3138,
      "step": 24800
    },
    {
      "epoch": 1.138130420815933,
      "grad_norm": 0.5344245433807373,
      "learning_rate": 7.033123224351024e-06,
      "loss": 0.4998,
      "step": 24801
    },
    {
      "epoch": 1.1381763113211876,
      "grad_norm": 0.48603877425193787,
      "learning_rate": 7.032899219383578e-06,
      "loss": 0.4286,
      "step": 24802
    },
    {
      "epoch": 1.138222201826442,
      "grad_norm": 0.4751037061214447,
      "learning_rate": 7.0326752095276505e-06,
      "loss": 0.3749,
      "step": 24803
    },
    {
      "epoch": 1.1382680923316966,
      "grad_norm": 0.46218571066856384,
      "learning_rate": 7.0324511947837805e-06,
      "loss": 0.3536,
      "step": 24804
    },
    {
      "epoch": 1.138313982836951,
      "grad_norm": 0.4770491421222687,
      "learning_rate": 7.032227175152506e-06,
      "loss": 0.4235,
      "step": 24805
    },
    {
      "epoch": 1.1383598733422056,
      "grad_norm": 0.47392261028289795,
      "learning_rate": 7.032003150634368e-06,
      "loss": 0.4171,
      "step": 24806
    },
    {
      "epoch": 1.13840576384746,
      "grad_norm": 0.5051649808883667,
      "learning_rate": 7.031779121229901e-06,
      "loss": 0.4046,
      "step": 24807
    },
    {
      "epoch": 1.1384516543527143,
      "grad_norm": 0.4590856432914734,
      "learning_rate": 7.03155508693965e-06,
      "loss": 0.34,
      "step": 24808
    },
    {
      "epoch": 1.1384975448579688,
      "grad_norm": 0.47395479679107666,
      "learning_rate": 7.031331047764149e-06,
      "loss": 0.3798,
      "step": 24809
    },
    {
      "epoch": 1.1385434353632233,
      "grad_norm": 0.45980486273765564,
      "learning_rate": 7.0311070037039365e-06,
      "loss": 0.3073,
      "step": 24810
    },
    {
      "epoch": 1.1385893258684778,
      "grad_norm": 0.4823946952819824,
      "learning_rate": 7.030882954759554e-06,
      "loss": 0.3593,
      "step": 24811
    },
    {
      "epoch": 1.1386352163737323,
      "grad_norm": 0.47585824131965637,
      "learning_rate": 7.030658900931539e-06,
      "loss": 0.3643,
      "step": 24812
    },
    {
      "epoch": 1.1386811068789868,
      "grad_norm": 0.45333290100097656,
      "learning_rate": 7.030434842220431e-06,
      "loss": 0.3292,
      "step": 24813
    },
    {
      "epoch": 1.138726997384241,
      "grad_norm": 0.4830933213233948,
      "learning_rate": 7.030210778626767e-06,
      "loss": 0.3837,
      "step": 24814
    },
    {
      "epoch": 1.1387728878894956,
      "grad_norm": 0.4814111888408661,
      "learning_rate": 7.029986710151085e-06,
      "loss": 0.364,
      "step": 24815
    },
    {
      "epoch": 1.13881877839475,
      "grad_norm": 0.542103111743927,
      "learning_rate": 7.029762636793928e-06,
      "loss": 0.3024,
      "step": 24816
    },
    {
      "epoch": 1.1388646689000046,
      "grad_norm": 0.4428088665008545,
      "learning_rate": 7.029538558555832e-06,
      "loss": 0.3153,
      "step": 24817
    },
    {
      "epoch": 1.138910559405259,
      "grad_norm": 0.49106183648109436,
      "learning_rate": 7.029314475437336e-06,
      "loss": 0.3727,
      "step": 24818
    },
    {
      "epoch": 1.1389564499105136,
      "grad_norm": 0.48322033882141113,
      "learning_rate": 7.029090387438979e-06,
      "loss": 0.409,
      "step": 24819
    },
    {
      "epoch": 1.139002340415768,
      "grad_norm": 0.5014519691467285,
      "learning_rate": 7.028866294561302e-06,
      "loss": 0.3839,
      "step": 24820
    },
    {
      "epoch": 1.1390482309210224,
      "grad_norm": 0.46592456102371216,
      "learning_rate": 7.028642196804839e-06,
      "loss": 0.3538,
      "step": 24821
    },
    {
      "epoch": 1.1390941214262769,
      "grad_norm": 0.4376320540904999,
      "learning_rate": 7.028418094170133e-06,
      "loss": 0.3161,
      "step": 24822
    },
    {
      "epoch": 1.1391400119315314,
      "grad_norm": 0.505426824092865,
      "learning_rate": 7.028193986657722e-06,
      "loss": 0.4151,
      "step": 24823
    },
    {
      "epoch": 1.1391859024367859,
      "grad_norm": 0.4779411554336548,
      "learning_rate": 7.0279698742681445e-06,
      "loss": 0.4402,
      "step": 24824
    },
    {
      "epoch": 1.1392317929420404,
      "grad_norm": 0.4424952566623688,
      "learning_rate": 7.027745757001939e-06,
      "loss": 0.3333,
      "step": 24825
    },
    {
      "epoch": 1.1392776834472946,
      "grad_norm": 0.4343445301055908,
      "learning_rate": 7.0275216348596465e-06,
      "loss": 0.3028,
      "step": 24826
    },
    {
      "epoch": 1.1393235739525491,
      "grad_norm": 0.45684632658958435,
      "learning_rate": 7.027297507841802e-06,
      "loss": 0.3089,
      "step": 24827
    },
    {
      "epoch": 1.1393694644578036,
      "grad_norm": 0.45610225200653076,
      "learning_rate": 7.027073375948948e-06,
      "loss": 0.3414,
      "step": 24828
    },
    {
      "epoch": 1.1394153549630581,
      "grad_norm": 0.47095775604248047,
      "learning_rate": 7.026849239181622e-06,
      "loss": 0.3725,
      "step": 24829
    },
    {
      "epoch": 1.1394612454683126,
      "grad_norm": 0.4666273891925812,
      "learning_rate": 7.0266250975403635e-06,
      "loss": 0.3772,
      "step": 24830
    },
    {
      "epoch": 1.1395071359735671,
      "grad_norm": 0.4947664141654968,
      "learning_rate": 7.0264009510257125e-06,
      "loss": 0.371,
      "step": 24831
    },
    {
      "epoch": 1.1395530264788216,
      "grad_norm": 0.48034748435020447,
      "learning_rate": 7.0261767996382046e-06,
      "loss": 0.3733,
      "step": 24832
    },
    {
      "epoch": 1.1395989169840761,
      "grad_norm": 0.4821909964084625,
      "learning_rate": 7.025952643378382e-06,
      "loss": 0.3318,
      "step": 24833
    },
    {
      "epoch": 1.1396448074893304,
      "grad_norm": 0.4427422881126404,
      "learning_rate": 7.025728482246783e-06,
      "loss": 0.3586,
      "step": 24834
    },
    {
      "epoch": 1.139690697994585,
      "grad_norm": 0.4751167893409729,
      "learning_rate": 7.025504316243946e-06,
      "loss": 0.3847,
      "step": 24835
    },
    {
      "epoch": 1.1397365884998394,
      "grad_norm": 0.4683513939380646,
      "learning_rate": 7.02528014537041e-06,
      "loss": 0.3577,
      "step": 24836
    },
    {
      "epoch": 1.1397824790050939,
      "grad_norm": 0.45081812143325806,
      "learning_rate": 7.025055969626715e-06,
      "loss": 0.3861,
      "step": 24837
    },
    {
      "epoch": 1.1398283695103484,
      "grad_norm": 0.4853639304637909,
      "learning_rate": 7.0248317890133984e-06,
      "loss": 0.4402,
      "step": 24838
    },
    {
      "epoch": 1.1398742600156027,
      "grad_norm": 0.4979836344718933,
      "learning_rate": 7.024607603531002e-06,
      "loss": 0.4082,
      "step": 24839
    },
    {
      "epoch": 1.1399201505208572,
      "grad_norm": 0.5283799767494202,
      "learning_rate": 7.024383413180062e-06,
      "loss": 0.4083,
      "step": 24840
    },
    {
      "epoch": 1.1399660410261117,
      "grad_norm": 0.4710626006126404,
      "learning_rate": 7.024159217961121e-06,
      "loss": 0.4031,
      "step": 24841
    },
    {
      "epoch": 1.1400119315313662,
      "grad_norm": 0.47772860527038574,
      "learning_rate": 7.023935017874713e-06,
      "loss": 0.3334,
      "step": 24842
    },
    {
      "epoch": 1.1400578220366206,
      "grad_norm": 0.4598221480846405,
      "learning_rate": 7.023710812921383e-06,
      "loss": 0.3629,
      "step": 24843
    },
    {
      "epoch": 1.1401037125418751,
      "grad_norm": 0.45816609263420105,
      "learning_rate": 7.023486603101664e-06,
      "loss": 0.3277,
      "step": 24844
    },
    {
      "epoch": 1.1401496030471296,
      "grad_norm": 0.55083167552948,
      "learning_rate": 7.023262388416099e-06,
      "loss": 0.4322,
      "step": 24845
    },
    {
      "epoch": 1.140195493552384,
      "grad_norm": 0.446175754070282,
      "learning_rate": 7.02303816886523e-06,
      "loss": 0.355,
      "step": 24846
    },
    {
      "epoch": 1.1402413840576384,
      "grad_norm": 0.46936631202697754,
      "learning_rate": 7.022813944449589e-06,
      "loss": 0.3307,
      "step": 24847
    },
    {
      "epoch": 1.140287274562893,
      "grad_norm": 0.47443658113479614,
      "learning_rate": 7.022589715169721e-06,
      "loss": 0.2998,
      "step": 24848
    },
    {
      "epoch": 1.1403331650681474,
      "grad_norm": 0.46371009945869446,
      "learning_rate": 7.022365481026161e-06,
      "loss": 0.347,
      "step": 24849
    },
    {
      "epoch": 1.140379055573402,
      "grad_norm": 0.46668490767478943,
      "learning_rate": 7.022141242019452e-06,
      "loss": 0.3118,
      "step": 24850
    },
    {
      "epoch": 1.1404249460786564,
      "grad_norm": 0.48154309391975403,
      "learning_rate": 7.02191699815013e-06,
      "loss": 0.3807,
      "step": 24851
    },
    {
      "epoch": 1.1404708365839107,
      "grad_norm": 0.489553302526474,
      "learning_rate": 7.021692749418738e-06,
      "loss": 0.3605,
      "step": 24852
    },
    {
      "epoch": 1.1405167270891652,
      "grad_norm": 0.5164687633514404,
      "learning_rate": 7.021468495825811e-06,
      "loss": 0.425,
      "step": 24853
    },
    {
      "epoch": 1.1405626175944197,
      "grad_norm": 0.470962792634964,
      "learning_rate": 7.021244237371892e-06,
      "loss": 0.3923,
      "step": 24854
    },
    {
      "epoch": 1.1406085080996742,
      "grad_norm": 0.4653065800666809,
      "learning_rate": 7.021019974057516e-06,
      "loss": 0.3119,
      "step": 24855
    },
    {
      "epoch": 1.1406543986049287,
      "grad_norm": 0.480281800031662,
      "learning_rate": 7.0207957058832256e-06,
      "loss": 0.3661,
      "step": 24856
    },
    {
      "epoch": 1.1407002891101832,
      "grad_norm": 0.44190484285354614,
      "learning_rate": 7.020571432849561e-06,
      "loss": 0.3474,
      "step": 24857
    },
    {
      "epoch": 1.1407461796154377,
      "grad_norm": 0.48670390248298645,
      "learning_rate": 7.020347154957059e-06,
      "loss": 0.3492,
      "step": 24858
    },
    {
      "epoch": 1.140792070120692,
      "grad_norm": 0.4663967788219452,
      "learning_rate": 7.020122872206258e-06,
      "loss": 0.3249,
      "step": 24859
    },
    {
      "epoch": 1.1408379606259464,
      "grad_norm": 0.4506220817565918,
      "learning_rate": 7.019898584597702e-06,
      "loss": 0.3557,
      "step": 24860
    },
    {
      "epoch": 1.140883851131201,
      "grad_norm": 0.44643524289131165,
      "learning_rate": 7.0196742921319265e-06,
      "loss": 0.3222,
      "step": 24861
    },
    {
      "epoch": 1.1409297416364554,
      "grad_norm": 0.4787890613079071,
      "learning_rate": 7.01944999480947e-06,
      "loss": 0.3974,
      "step": 24862
    },
    {
      "epoch": 1.14097563214171,
      "grad_norm": 0.4575692117214203,
      "learning_rate": 7.019225692630877e-06,
      "loss": 0.3303,
      "step": 24863
    },
    {
      "epoch": 1.1410215226469644,
      "grad_norm": 0.4653702676296234,
      "learning_rate": 7.01900138559668e-06,
      "loss": 0.3572,
      "step": 24864
    },
    {
      "epoch": 1.1410674131522187,
      "grad_norm": 0.47481268644332886,
      "learning_rate": 7.018777073707423e-06,
      "loss": 0.3566,
      "step": 24865
    },
    {
      "epoch": 1.1411133036574732,
      "grad_norm": 0.48292267322540283,
      "learning_rate": 7.018552756963645e-06,
      "loss": 0.3628,
      "step": 24866
    },
    {
      "epoch": 1.1411591941627277,
      "grad_norm": 0.4298136830329895,
      "learning_rate": 7.018328435365883e-06,
      "loss": 0.3155,
      "step": 24867
    },
    {
      "epoch": 1.1412050846679822,
      "grad_norm": 0.4496147930622101,
      "learning_rate": 7.018104108914682e-06,
      "loss": 0.3108,
      "step": 24868
    },
    {
      "epoch": 1.1412509751732367,
      "grad_norm": 0.5021294355392456,
      "learning_rate": 7.0178797776105735e-06,
      "loss": 0.3899,
      "step": 24869
    },
    {
      "epoch": 1.1412968656784912,
      "grad_norm": 0.46272557973861694,
      "learning_rate": 7.017655441454102e-06,
      "loss": 0.3379,
      "step": 24870
    },
    {
      "epoch": 1.1413427561837457,
      "grad_norm": 0.418933242559433,
      "learning_rate": 7.0174311004458075e-06,
      "loss": 0.271,
      "step": 24871
    },
    {
      "epoch": 1.141388646689,
      "grad_norm": 0.49746739864349365,
      "learning_rate": 7.017206754586226e-06,
      "loss": 0.3803,
      "step": 24872
    },
    {
      "epoch": 1.1414345371942545,
      "grad_norm": 0.4602879285812378,
      "learning_rate": 7.0169824038759e-06,
      "loss": 0.3506,
      "step": 24873
    },
    {
      "epoch": 1.141480427699509,
      "grad_norm": 0.449402779340744,
      "learning_rate": 7.016758048315369e-06,
      "loss": 0.3208,
      "step": 24874
    },
    {
      "epoch": 1.1415263182047635,
      "grad_norm": 0.461357444524765,
      "learning_rate": 7.016533687905169e-06,
      "loss": 0.3365,
      "step": 24875
    },
    {
      "epoch": 1.141572208710018,
      "grad_norm": 0.4784964919090271,
      "learning_rate": 7.016309322645842e-06,
      "loss": 0.4173,
      "step": 24876
    },
    {
      "epoch": 1.1416180992152722,
      "grad_norm": 0.5157486200332642,
      "learning_rate": 7.016084952537928e-06,
      "loss": 0.4099,
      "step": 24877
    },
    {
      "epoch": 1.1416639897205267,
      "grad_norm": 0.49279165267944336,
      "learning_rate": 7.015860577581966e-06,
      "loss": 0.3913,
      "step": 24878
    },
    {
      "epoch": 1.1417098802257812,
      "grad_norm": 0.45893847942352295,
      "learning_rate": 7.015636197778497e-06,
      "loss": 0.375,
      "step": 24879
    },
    {
      "epoch": 1.1417557707310357,
      "grad_norm": 0.4883574843406677,
      "learning_rate": 7.015411813128058e-06,
      "loss": 0.3809,
      "step": 24880
    },
    {
      "epoch": 1.1418016612362902,
      "grad_norm": 0.5022452473640442,
      "learning_rate": 7.015187423631189e-06,
      "loss": 0.3878,
      "step": 24881
    },
    {
      "epoch": 1.1418475517415447,
      "grad_norm": 0.4454035758972168,
      "learning_rate": 7.014963029288432e-06,
      "loss": 0.3534,
      "step": 24882
    },
    {
      "epoch": 1.1418934422467992,
      "grad_norm": 0.43624651432037354,
      "learning_rate": 7.014738630100323e-06,
      "loss": 0.2991,
      "step": 24883
    },
    {
      "epoch": 1.1419393327520535,
      "grad_norm": 0.4754253923892975,
      "learning_rate": 7.014514226067405e-06,
      "loss": 0.3754,
      "step": 24884
    },
    {
      "epoch": 1.141985223257308,
      "grad_norm": 0.47495603561401367,
      "learning_rate": 7.014289817190215e-06,
      "loss": 0.3664,
      "step": 24885
    },
    {
      "epoch": 1.1420311137625625,
      "grad_norm": 0.5194360017776489,
      "learning_rate": 7.014065403469294e-06,
      "loss": 0.4036,
      "step": 24886
    },
    {
      "epoch": 1.142077004267817,
      "grad_norm": 0.4147861897945404,
      "learning_rate": 7.01384098490518e-06,
      "loss": 0.2863,
      "step": 24887
    },
    {
      "epoch": 1.1421228947730715,
      "grad_norm": 0.5046252608299255,
      "learning_rate": 7.0136165614984165e-06,
      "loss": 0.4223,
      "step": 24888
    },
    {
      "epoch": 1.142168785278326,
      "grad_norm": 0.5072726607322693,
      "learning_rate": 7.013392133249539e-06,
      "loss": 0.3913,
      "step": 24889
    },
    {
      "epoch": 1.1422146757835803,
      "grad_norm": 0.4464978873729706,
      "learning_rate": 7.013167700159089e-06,
      "loss": 0.3661,
      "step": 24890
    },
    {
      "epoch": 1.1422605662888348,
      "grad_norm": 0.469239205121994,
      "learning_rate": 7.012943262227607e-06,
      "loss": 0.2996,
      "step": 24891
    },
    {
      "epoch": 1.1423064567940893,
      "grad_norm": 0.4954543113708496,
      "learning_rate": 7.012718819455632e-06,
      "loss": 0.3338,
      "step": 24892
    },
    {
      "epoch": 1.1423523472993438,
      "grad_norm": 0.492285817861557,
      "learning_rate": 7.012494371843701e-06,
      "loss": 0.373,
      "step": 24893
    },
    {
      "epoch": 1.1423982378045983,
      "grad_norm": 0.47986897826194763,
      "learning_rate": 7.01226991939236e-06,
      "loss": 0.4025,
      "step": 24894
    },
    {
      "epoch": 1.1424441283098528,
      "grad_norm": 0.47332343459129333,
      "learning_rate": 7.012045462102143e-06,
      "loss": 0.3426,
      "step": 24895
    },
    {
      "epoch": 1.1424900188151073,
      "grad_norm": 0.5081636905670166,
      "learning_rate": 7.011820999973591e-06,
      "loss": 0.4973,
      "step": 24896
    },
    {
      "epoch": 1.1425359093203615,
      "grad_norm": 0.46340858936309814,
      "learning_rate": 7.0115965330072455e-06,
      "loss": 0.3381,
      "step": 24897
    },
    {
      "epoch": 1.142581799825616,
      "grad_norm": 0.4857732951641083,
      "learning_rate": 7.011372061203644e-06,
      "loss": 0.401,
      "step": 24898
    },
    {
      "epoch": 1.1426276903308705,
      "grad_norm": 0.47230997681617737,
      "learning_rate": 7.011147584563328e-06,
      "loss": 0.3749,
      "step": 24899
    },
    {
      "epoch": 1.142673580836125,
      "grad_norm": 0.4550023674964905,
      "learning_rate": 7.010923103086837e-06,
      "loss": 0.3487,
      "step": 24900
    },
    {
      "epoch": 1.1427194713413795,
      "grad_norm": 0.4740363657474518,
      "learning_rate": 7.0106986167747104e-06,
      "loss": 0.4004,
      "step": 24901
    },
    {
      "epoch": 1.142765361846634,
      "grad_norm": 0.4842958450317383,
      "learning_rate": 7.010474125627489e-06,
      "loss": 0.3825,
      "step": 24902
    },
    {
      "epoch": 1.1428112523518883,
      "grad_norm": 0.48954150080680847,
      "learning_rate": 7.010249629645712e-06,
      "loss": 0.4297,
      "step": 24903
    },
    {
      "epoch": 1.1428571428571428,
      "grad_norm": 0.45172399282455444,
      "learning_rate": 7.010025128829917e-06,
      "loss": 0.3575,
      "step": 24904
    },
    {
      "epoch": 1.1429030333623973,
      "grad_norm": 0.4919714629650116,
      "learning_rate": 7.009800623180647e-06,
      "loss": 0.3651,
      "step": 24905
    },
    {
      "epoch": 1.1429489238676518,
      "grad_norm": 0.4649743437767029,
      "learning_rate": 7.009576112698442e-06,
      "loss": 0.3968,
      "step": 24906
    },
    {
      "epoch": 1.1429948143729063,
      "grad_norm": 0.46726515889167786,
      "learning_rate": 7.00935159738384e-06,
      "loss": 0.3666,
      "step": 24907
    },
    {
      "epoch": 1.1430407048781608,
      "grad_norm": 0.48515409231185913,
      "learning_rate": 7.009127077237381e-06,
      "loss": 0.3499,
      "step": 24908
    },
    {
      "epoch": 1.1430865953834153,
      "grad_norm": 0.5135096311569214,
      "learning_rate": 7.008902552259605e-06,
      "loss": 0.4144,
      "step": 24909
    },
    {
      "epoch": 1.1431324858886696,
      "grad_norm": 0.4409202039241791,
      "learning_rate": 7.008678022451053e-06,
      "loss": 0.3166,
      "step": 24910
    },
    {
      "epoch": 1.143178376393924,
      "grad_norm": 0.4671071469783783,
      "learning_rate": 7.008453487812266e-06,
      "loss": 0.3499,
      "step": 24911
    },
    {
      "epoch": 1.1432242668991786,
      "grad_norm": 0.47289708256721497,
      "learning_rate": 7.008228948343781e-06,
      "loss": 0.3298,
      "step": 24912
    },
    {
      "epoch": 1.143270157404433,
      "grad_norm": 0.4455314576625824,
      "learning_rate": 7.0080044040461395e-06,
      "loss": 0.3027,
      "step": 24913
    },
    {
      "epoch": 1.1433160479096875,
      "grad_norm": 0.47602707147598267,
      "learning_rate": 7.00777985491988e-06,
      "loss": 0.3941,
      "step": 24914
    },
    {
      "epoch": 1.1433619384149418,
      "grad_norm": 0.45630142092704773,
      "learning_rate": 7.0075553009655435e-06,
      "loss": 0.3272,
      "step": 24915
    },
    {
      "epoch": 1.1434078289201963,
      "grad_norm": 0.5175597667694092,
      "learning_rate": 7.007330742183671e-06,
      "loss": 0.3946,
      "step": 24916
    },
    {
      "epoch": 1.1434537194254508,
      "grad_norm": 0.4405609965324402,
      "learning_rate": 7.007106178574803e-06,
      "loss": 0.3392,
      "step": 24917
    },
    {
      "epoch": 1.1434996099307053,
      "grad_norm": 0.4584062099456787,
      "learning_rate": 7.006881610139476e-06,
      "loss": 0.2909,
      "step": 24918
    },
    {
      "epoch": 1.1435455004359598,
      "grad_norm": 0.4774465262889862,
      "learning_rate": 7.006657036878232e-06,
      "loss": 0.3676,
      "step": 24919
    },
    {
      "epoch": 1.1435913909412143,
      "grad_norm": 0.4936419129371643,
      "learning_rate": 7.006432458791612e-06,
      "loss": 0.3405,
      "step": 24920
    },
    {
      "epoch": 1.1436372814464688,
      "grad_norm": 0.4631178379058838,
      "learning_rate": 7.006207875880154e-06,
      "loss": 0.3645,
      "step": 24921
    },
    {
      "epoch": 1.1436831719517233,
      "grad_norm": 0.5628679394721985,
      "learning_rate": 7.005983288144401e-06,
      "loss": 0.3817,
      "step": 24922
    },
    {
      "epoch": 1.1437290624569776,
      "grad_norm": 0.49904584884643555,
      "learning_rate": 7.005758695584891e-06,
      "loss": 0.4479,
      "step": 24923
    },
    {
      "epoch": 1.143774952962232,
      "grad_norm": 0.4463474154472351,
      "learning_rate": 7.005534098202164e-06,
      "loss": 0.3298,
      "step": 24924
    },
    {
      "epoch": 1.1438208434674866,
      "grad_norm": 0.44511690735816956,
      "learning_rate": 7.00530949599676e-06,
      "loss": 0.3208,
      "step": 24925
    },
    {
      "epoch": 1.143866733972741,
      "grad_norm": 0.44681280851364136,
      "learning_rate": 7.00508488896922e-06,
      "loss": 0.3198,
      "step": 24926
    },
    {
      "epoch": 1.1439126244779956,
      "grad_norm": 0.47121742367744446,
      "learning_rate": 7.004860277120085e-06,
      "loss": 0.357,
      "step": 24927
    },
    {
      "epoch": 1.1439585149832499,
      "grad_norm": 0.448816180229187,
      "learning_rate": 7.004635660449892e-06,
      "loss": 0.2996,
      "step": 24928
    },
    {
      "epoch": 1.1440044054885044,
      "grad_norm": 0.4731786847114563,
      "learning_rate": 7.0044110389591825e-06,
      "loss": 0.2944,
      "step": 24929
    },
    {
      "epoch": 1.1440502959937588,
      "grad_norm": 0.5082247853279114,
      "learning_rate": 7.0041864126484995e-06,
      "loss": 0.4319,
      "step": 24930
    },
    {
      "epoch": 1.1440961864990133,
      "grad_norm": 0.44497427344322205,
      "learning_rate": 7.003961781518378e-06,
      "loss": 0.329,
      "step": 24931
    },
    {
      "epoch": 1.1441420770042678,
      "grad_norm": 0.4509630501270294,
      "learning_rate": 7.003737145569361e-06,
      "loss": 0.3339,
      "step": 24932
    },
    {
      "epoch": 1.1441879675095223,
      "grad_norm": 0.4785159230232239,
      "learning_rate": 7.003512504801991e-06,
      "loss": 0.4123,
      "step": 24933
    },
    {
      "epoch": 1.1442338580147768,
      "grad_norm": 0.469690203666687,
      "learning_rate": 7.0032878592168045e-06,
      "loss": 0.3286,
      "step": 24934
    },
    {
      "epoch": 1.1442797485200311,
      "grad_norm": 0.4968698024749756,
      "learning_rate": 7.003063208814343e-06,
      "loss": 0.3971,
      "step": 24935
    },
    {
      "epoch": 1.1443256390252856,
      "grad_norm": 0.421076238155365,
      "learning_rate": 7.002838553595146e-06,
      "loss": 0.2735,
      "step": 24936
    },
    {
      "epoch": 1.14437152953054,
      "grad_norm": 0.4518403112888336,
      "learning_rate": 7.002613893559754e-06,
      "loss": 0.3418,
      "step": 24937
    },
    {
      "epoch": 1.1444174200357946,
      "grad_norm": 0.4615568518638611,
      "learning_rate": 7.0023892287087095e-06,
      "loss": 0.3314,
      "step": 24938
    },
    {
      "epoch": 1.144463310541049,
      "grad_norm": 0.47610920667648315,
      "learning_rate": 7.00216455904255e-06,
      "loss": 0.3876,
      "step": 24939
    },
    {
      "epoch": 1.1445092010463036,
      "grad_norm": 0.49797719717025757,
      "learning_rate": 7.001939884561816e-06,
      "loss": 0.3535,
      "step": 24940
    },
    {
      "epoch": 1.1445550915515579,
      "grad_norm": 0.46331605315208435,
      "learning_rate": 7.001715205267048e-06,
      "loss": 0.3386,
      "step": 24941
    },
    {
      "epoch": 1.1446009820568124,
      "grad_norm": 0.444021999835968,
      "learning_rate": 7.001490521158788e-06,
      "loss": 0.3526,
      "step": 24942
    },
    {
      "epoch": 1.1446468725620669,
      "grad_norm": 0.4544728398323059,
      "learning_rate": 7.001265832237574e-06,
      "loss": 0.3222,
      "step": 24943
    },
    {
      "epoch": 1.1446927630673214,
      "grad_norm": 0.45844194293022156,
      "learning_rate": 7.001041138503948e-06,
      "loss": 0.3893,
      "step": 24944
    },
    {
      "epoch": 1.1447386535725759,
      "grad_norm": 0.4786856174468994,
      "learning_rate": 7.000816439958451e-06,
      "loss": 0.4069,
      "step": 24945
    },
    {
      "epoch": 1.1447845440778304,
      "grad_norm": 0.4699844419956207,
      "learning_rate": 7.00059173660162e-06,
      "loss": 0.3317,
      "step": 24946
    },
    {
      "epoch": 1.1448304345830849,
      "grad_norm": 0.47999995946884155,
      "learning_rate": 7.000367028433998e-06,
      "loss": 0.3695,
      "step": 24947
    },
    {
      "epoch": 1.1448763250883391,
      "grad_norm": 0.42146944999694824,
      "learning_rate": 7.000142315456123e-06,
      "loss": 0.3195,
      "step": 24948
    },
    {
      "epoch": 1.1449222155935936,
      "grad_norm": 0.508046567440033,
      "learning_rate": 6.999917597668538e-06,
      "loss": 0.3757,
      "step": 24949
    },
    {
      "epoch": 1.1449681060988481,
      "grad_norm": 0.4362163543701172,
      "learning_rate": 6.999692875071784e-06,
      "loss": 0.3348,
      "step": 24950
    },
    {
      "epoch": 1.1450139966041026,
      "grad_norm": 0.45824134349823,
      "learning_rate": 6.9994681476663975e-06,
      "loss": 0.3332,
      "step": 24951
    },
    {
      "epoch": 1.1450598871093571,
      "grad_norm": 0.42417895793914795,
      "learning_rate": 6.999243415452922e-06,
      "loss": 0.324,
      "step": 24952
    },
    {
      "epoch": 1.1451057776146116,
      "grad_norm": 0.44994741678237915,
      "learning_rate": 6.999018678431897e-06,
      "loss": 0.3086,
      "step": 24953
    },
    {
      "epoch": 1.145151668119866,
      "grad_norm": 0.47160306572914124,
      "learning_rate": 6.998793936603862e-06,
      "loss": 0.3572,
      "step": 24954
    },
    {
      "epoch": 1.1451975586251204,
      "grad_norm": 0.45894333720207214,
      "learning_rate": 6.998569189969361e-06,
      "loss": 0.3524,
      "step": 24955
    },
    {
      "epoch": 1.145243449130375,
      "grad_norm": 0.49134561419487,
      "learning_rate": 6.9983444385289314e-06,
      "loss": 0.4148,
      "step": 24956
    },
    {
      "epoch": 1.1452893396356294,
      "grad_norm": 0.44758114218711853,
      "learning_rate": 6.998119682283112e-06,
      "loss": 0.2971,
      "step": 24957
    },
    {
      "epoch": 1.145335230140884,
      "grad_norm": 0.43807709217071533,
      "learning_rate": 6.997894921232446e-06,
      "loss": 0.3042,
      "step": 24958
    },
    {
      "epoch": 1.1453811206461384,
      "grad_norm": 0.4467284381389618,
      "learning_rate": 6.997670155377474e-06,
      "loss": 0.3209,
      "step": 24959
    },
    {
      "epoch": 1.145427011151393,
      "grad_norm": 0.46406763792037964,
      "learning_rate": 6.997445384718736e-06,
      "loss": 0.3707,
      "step": 24960
    },
    {
      "epoch": 1.1454729016566472,
      "grad_norm": 0.4993608295917511,
      "learning_rate": 6.997220609256773e-06,
      "loss": 0.4195,
      "step": 24961
    },
    {
      "epoch": 1.1455187921619017,
      "grad_norm": 0.46961164474487305,
      "learning_rate": 6.996995828992123e-06,
      "loss": 0.3864,
      "step": 24962
    },
    {
      "epoch": 1.1455646826671562,
      "grad_norm": 0.4230894446372986,
      "learning_rate": 6.996771043925328e-06,
      "loss": 0.2755,
      "step": 24963
    },
    {
      "epoch": 1.1456105731724107,
      "grad_norm": 0.5033126473426819,
      "learning_rate": 6.996546254056931e-06,
      "loss": 0.4441,
      "step": 24964
    },
    {
      "epoch": 1.1456564636776652,
      "grad_norm": 0.4961814284324646,
      "learning_rate": 6.996321459387469e-06,
      "loss": 0.3651,
      "step": 24965
    },
    {
      "epoch": 1.1457023541829194,
      "grad_norm": 0.46157148480415344,
      "learning_rate": 6.996096659917484e-06,
      "loss": 0.3268,
      "step": 24966
    },
    {
      "epoch": 1.145748244688174,
      "grad_norm": 0.5087007880210876,
      "learning_rate": 6.995871855647517e-06,
      "loss": 0.4159,
      "step": 24967
    },
    {
      "epoch": 1.1457941351934284,
      "grad_norm": 0.4776315987110138,
      "learning_rate": 6.995647046578107e-06,
      "loss": 0.3599,
      "step": 24968
    },
    {
      "epoch": 1.145840025698683,
      "grad_norm": 0.4742911458015442,
      "learning_rate": 6.995422232709795e-06,
      "loss": 0.3703,
      "step": 24969
    },
    {
      "epoch": 1.1458859162039374,
      "grad_norm": 0.4834994971752167,
      "learning_rate": 6.995197414043125e-06,
      "loss": 0.409,
      "step": 24970
    },
    {
      "epoch": 1.145931806709192,
      "grad_norm": 0.4417472779750824,
      "learning_rate": 6.9949725905786334e-06,
      "loss": 0.3539,
      "step": 24971
    },
    {
      "epoch": 1.1459776972144464,
      "grad_norm": 0.4900830388069153,
      "learning_rate": 6.994747762316863e-06,
      "loss": 0.4409,
      "step": 24972
    },
    {
      "epoch": 1.1460235877197007,
      "grad_norm": 0.46553850173950195,
      "learning_rate": 6.994522929258354e-06,
      "loss": 0.3394,
      "step": 24973
    },
    {
      "epoch": 1.1460694782249552,
      "grad_norm": 0.48493367433547974,
      "learning_rate": 6.994298091403645e-06,
      "loss": 0.4149,
      "step": 24974
    },
    {
      "epoch": 1.1461153687302097,
      "grad_norm": 0.4602962136268616,
      "learning_rate": 6.994073248753279e-06,
      "loss": 0.343,
      "step": 24975
    },
    {
      "epoch": 1.1461612592354642,
      "grad_norm": 0.4861312210559845,
      "learning_rate": 6.993848401307798e-06,
      "loss": 0.3397,
      "step": 24976
    },
    {
      "epoch": 1.1462071497407187,
      "grad_norm": 0.479353129863739,
      "learning_rate": 6.99362354906774e-06,
      "loss": 0.3608,
      "step": 24977
    },
    {
      "epoch": 1.1462530402459732,
      "grad_norm": 0.47855016589164734,
      "learning_rate": 6.9933986920336464e-06,
      "loss": 0.3514,
      "step": 24978
    },
    {
      "epoch": 1.1462989307512275,
      "grad_norm": 0.418137788772583,
      "learning_rate": 6.993173830206057e-06,
      "loss": 0.2798,
      "step": 24979
    },
    {
      "epoch": 1.146344821256482,
      "grad_norm": 0.47327977418899536,
      "learning_rate": 6.992948963585515e-06,
      "loss": 0.3463,
      "step": 24980
    },
    {
      "epoch": 1.1463907117617365,
      "grad_norm": 0.4292834401130676,
      "learning_rate": 6.992724092172559e-06,
      "loss": 0.313,
      "step": 24981
    },
    {
      "epoch": 1.146436602266991,
      "grad_norm": 0.45037105679512024,
      "learning_rate": 6.9924992159677315e-06,
      "loss": 0.334,
      "step": 24982
    },
    {
      "epoch": 1.1464824927722455,
      "grad_norm": 0.45711550116539,
      "learning_rate": 6.992274334971572e-06,
      "loss": 0.3896,
      "step": 24983
    },
    {
      "epoch": 1.1465283832775,
      "grad_norm": 0.5297359228134155,
      "learning_rate": 6.992049449184621e-06,
      "loss": 0.3148,
      "step": 24984
    },
    {
      "epoch": 1.1465742737827544,
      "grad_norm": 0.4432131350040436,
      "learning_rate": 6.991824558607418e-06,
      "loss": 0.2908,
      "step": 24985
    },
    {
      "epoch": 1.1466201642880087,
      "grad_norm": 0.4860258996486664,
      "learning_rate": 6.9915996632405085e-06,
      "loss": 0.3926,
      "step": 24986
    },
    {
      "epoch": 1.1466660547932632,
      "grad_norm": 0.48344576358795166,
      "learning_rate": 6.9913747630844286e-06,
      "loss": 0.4093,
      "step": 24987
    },
    {
      "epoch": 1.1467119452985177,
      "grad_norm": 0.4971316456794739,
      "learning_rate": 6.991149858139723e-06,
      "loss": 0.3855,
      "step": 24988
    },
    {
      "epoch": 1.1467578358037722,
      "grad_norm": 0.4725916385650635,
      "learning_rate": 6.990924948406929e-06,
      "loss": 0.3957,
      "step": 24989
    },
    {
      "epoch": 1.1468037263090267,
      "grad_norm": 0.48780810832977295,
      "learning_rate": 6.990700033886588e-06,
      "loss": 0.3666,
      "step": 24990
    },
    {
      "epoch": 1.1468496168142812,
      "grad_norm": 0.46235063672065735,
      "learning_rate": 6.990475114579243e-06,
      "loss": 0.3314,
      "step": 24991
    },
    {
      "epoch": 1.1468955073195355,
      "grad_norm": 0.4588218033313751,
      "learning_rate": 6.990250190485433e-06,
      "loss": 0.3068,
      "step": 24992
    },
    {
      "epoch": 1.14694139782479,
      "grad_norm": 0.4766181707382202,
      "learning_rate": 6.9900252616057e-06,
      "loss": 0.3881,
      "step": 24993
    },
    {
      "epoch": 1.1469872883300445,
      "grad_norm": 0.47128942608833313,
      "learning_rate": 6.989800327940583e-06,
      "loss": 0.3417,
      "step": 24994
    },
    {
      "epoch": 1.147033178835299,
      "grad_norm": 0.4698542654514313,
      "learning_rate": 6.9895753894906256e-06,
      "loss": 0.3778,
      "step": 24995
    },
    {
      "epoch": 1.1470790693405535,
      "grad_norm": 0.45415112376213074,
      "learning_rate": 6.989350446256367e-06,
      "loss": 0.3225,
      "step": 24996
    },
    {
      "epoch": 1.147124959845808,
      "grad_norm": 0.4701600968837738,
      "learning_rate": 6.989125498238347e-06,
      "loss": 0.358,
      "step": 24997
    },
    {
      "epoch": 1.1471708503510625,
      "grad_norm": 0.43711742758750916,
      "learning_rate": 6.9889005454371095e-06,
      "loss": 0.3277,
      "step": 24998
    },
    {
      "epoch": 1.1472167408563168,
      "grad_norm": 0.47080016136169434,
      "learning_rate": 6.988675587853195e-06,
      "loss": 0.3526,
      "step": 24999
    },
    {
      "epoch": 1.1472626313615713,
      "grad_norm": 0.47006282210350037,
      "learning_rate": 6.9884506254871405e-06,
      "loss": 0.3544,
      "step": 25000
    },
    {
      "epoch": 1.1473085218668257,
      "grad_norm": 0.4615339934825897,
      "learning_rate": 6.9882256583394925e-06,
      "loss": 0.3519,
      "step": 25001
    },
    {
      "epoch": 1.1473544123720802,
      "grad_norm": 0.4525710344314575,
      "learning_rate": 6.988000686410787e-06,
      "loss": 0.3378,
      "step": 25002
    },
    {
      "epoch": 1.1474003028773347,
      "grad_norm": 0.4831273853778839,
      "learning_rate": 6.987775709701569e-06,
      "loss": 0.3861,
      "step": 25003
    },
    {
      "epoch": 1.147446193382589,
      "grad_norm": 0.47208690643310547,
      "learning_rate": 6.987550728212378e-06,
      "loss": 0.3589,
      "step": 25004
    },
    {
      "epoch": 1.1474920838878435,
      "grad_norm": 0.44566628336906433,
      "learning_rate": 6.987325741943753e-06,
      "loss": 0.3318,
      "step": 25005
    },
    {
      "epoch": 1.147537974393098,
      "grad_norm": 0.47661104798316956,
      "learning_rate": 6.987100750896237e-06,
      "loss": 0.3793,
      "step": 25006
    },
    {
      "epoch": 1.1475838648983525,
      "grad_norm": 0.4605101943016052,
      "learning_rate": 6.9868757550703726e-06,
      "loss": 0.3574,
      "step": 25007
    },
    {
      "epoch": 1.147629755403607,
      "grad_norm": 0.45379412174224854,
      "learning_rate": 6.986650754466698e-06,
      "loss": 0.3385,
      "step": 25008
    },
    {
      "epoch": 1.1476756459088615,
      "grad_norm": 0.45908984541893005,
      "learning_rate": 6.986425749085756e-06,
      "loss": 0.3595,
      "step": 25009
    },
    {
      "epoch": 1.147721536414116,
      "grad_norm": 0.47903504967689514,
      "learning_rate": 6.9862007389280864e-06,
      "loss": 0.386,
      "step": 25010
    },
    {
      "epoch": 1.1477674269193705,
      "grad_norm": 0.45941847562789917,
      "learning_rate": 6.985975723994229e-06,
      "loss": 0.3144,
      "step": 25011
    },
    {
      "epoch": 1.1478133174246248,
      "grad_norm": 0.49382543563842773,
      "learning_rate": 6.985750704284731e-06,
      "loss": 0.3635,
      "step": 25012
    },
    {
      "epoch": 1.1478592079298793,
      "grad_norm": 0.47839128971099854,
      "learning_rate": 6.985525679800126e-06,
      "loss": 0.2912,
      "step": 25013
    },
    {
      "epoch": 1.1479050984351338,
      "grad_norm": 0.4532592296600342,
      "learning_rate": 6.985300650540958e-06,
      "loss": 0.3685,
      "step": 25014
    },
    {
      "epoch": 1.1479509889403883,
      "grad_norm": 0.499133437871933,
      "learning_rate": 6.985075616507772e-06,
      "loss": 0.3948,
      "step": 25015
    },
    {
      "epoch": 1.1479968794456428,
      "grad_norm": 0.4837343096733093,
      "learning_rate": 6.984850577701103e-06,
      "loss": 0.3801,
      "step": 25016
    },
    {
      "epoch": 1.148042769950897,
      "grad_norm": 0.45721644163131714,
      "learning_rate": 6.984625534121494e-06,
      "loss": 0.3208,
      "step": 25017
    },
    {
      "epoch": 1.1480886604561515,
      "grad_norm": 0.4535588026046753,
      "learning_rate": 6.984400485769489e-06,
      "loss": 0.3555,
      "step": 25018
    },
    {
      "epoch": 1.148134550961406,
      "grad_norm": 0.47468113899230957,
      "learning_rate": 6.984175432645626e-06,
      "loss": 0.3572,
      "step": 25019
    },
    {
      "epoch": 1.1481804414666605,
      "grad_norm": 0.45623454451560974,
      "learning_rate": 6.983950374750448e-06,
      "loss": 0.317,
      "step": 25020
    },
    {
      "epoch": 1.148226331971915,
      "grad_norm": 0.4824194014072418,
      "learning_rate": 6.983725312084496e-06,
      "loss": 0.4157,
      "step": 25021
    },
    {
      "epoch": 1.1482722224771695,
      "grad_norm": 0.4603768289089203,
      "learning_rate": 6.9835002446483095e-06,
      "loss": 0.3386,
      "step": 25022
    },
    {
      "epoch": 1.148318112982424,
      "grad_norm": 0.4851486086845398,
      "learning_rate": 6.9832751724424306e-06,
      "loss": 0.4264,
      "step": 25023
    },
    {
      "epoch": 1.1483640034876783,
      "grad_norm": 0.5331316590309143,
      "learning_rate": 6.983050095467402e-06,
      "loss": 0.4294,
      "step": 25024
    },
    {
      "epoch": 1.1484098939929328,
      "grad_norm": 0.48154881596565247,
      "learning_rate": 6.982825013723763e-06,
      "loss": 0.3316,
      "step": 25025
    },
    {
      "epoch": 1.1484557844981873,
      "grad_norm": 0.46427908539772034,
      "learning_rate": 6.982599927212057e-06,
      "loss": 0.3834,
      "step": 25026
    },
    {
      "epoch": 1.1485016750034418,
      "grad_norm": 0.4564216732978821,
      "learning_rate": 6.982374835932823e-06,
      "loss": 0.3711,
      "step": 25027
    },
    {
      "epoch": 1.1485475655086963,
      "grad_norm": 0.47576141357421875,
      "learning_rate": 6.982149739886603e-06,
      "loss": 0.3755,
      "step": 25028
    },
    {
      "epoch": 1.1485934560139508,
      "grad_norm": 0.4606434106826782,
      "learning_rate": 6.9819246390739385e-06,
      "loss": 0.3609,
      "step": 25029
    },
    {
      "epoch": 1.148639346519205,
      "grad_norm": 0.44380488991737366,
      "learning_rate": 6.98169953349537e-06,
      "loss": 0.3267,
      "step": 25030
    },
    {
      "epoch": 1.1486852370244596,
      "grad_norm": 0.47449520230293274,
      "learning_rate": 6.981474423151441e-06,
      "loss": 0.362,
      "step": 25031
    },
    {
      "epoch": 1.148731127529714,
      "grad_norm": 0.4447198808193207,
      "learning_rate": 6.98124930804269e-06,
      "loss": 0.3052,
      "step": 25032
    },
    {
      "epoch": 1.1487770180349686,
      "grad_norm": 0.4412217140197754,
      "learning_rate": 6.9810241881696605e-06,
      "loss": 0.3239,
      "step": 25033
    },
    {
      "epoch": 1.148822908540223,
      "grad_norm": 0.4593147933483124,
      "learning_rate": 6.980799063532893e-06,
      "loss": 0.3502,
      "step": 25034
    },
    {
      "epoch": 1.1488687990454776,
      "grad_norm": 0.49950239062309265,
      "learning_rate": 6.980573934132929e-06,
      "loss": 0.4086,
      "step": 25035
    },
    {
      "epoch": 1.148914689550732,
      "grad_norm": 0.45279550552368164,
      "learning_rate": 6.980348799970309e-06,
      "loss": 0.3437,
      "step": 25036
    },
    {
      "epoch": 1.1489605800559863,
      "grad_norm": 0.49908551573753357,
      "learning_rate": 6.980123661045577e-06,
      "loss": 0.4254,
      "step": 25037
    },
    {
      "epoch": 1.1490064705612408,
      "grad_norm": 0.49343904852867126,
      "learning_rate": 6.979898517359272e-06,
      "loss": 0.4195,
      "step": 25038
    },
    {
      "epoch": 1.1490523610664953,
      "grad_norm": 0.4655086100101471,
      "learning_rate": 6.9796733689119345e-06,
      "loss": 0.3628,
      "step": 25039
    },
    {
      "epoch": 1.1490982515717498,
      "grad_norm": 0.4357227385044098,
      "learning_rate": 6.979448215704108e-06,
      "loss": 0.289,
      "step": 25040
    },
    {
      "epoch": 1.1491441420770043,
      "grad_norm": 0.43845608830451965,
      "learning_rate": 6.979223057736333e-06,
      "loss": 0.3178,
      "step": 25041
    },
    {
      "epoch": 1.1491900325822588,
      "grad_norm": 0.5134596228599548,
      "learning_rate": 6.978997895009152e-06,
      "loss": 0.4084,
      "step": 25042
    },
    {
      "epoch": 1.149235923087513,
      "grad_norm": 0.46119585633277893,
      "learning_rate": 6.978772727523106e-06,
      "loss": 0.3248,
      "step": 25043
    },
    {
      "epoch": 1.1492818135927676,
      "grad_norm": 0.5257668495178223,
      "learning_rate": 6.978547555278735e-06,
      "loss": 0.4427,
      "step": 25044
    },
    {
      "epoch": 1.149327704098022,
      "grad_norm": 0.4643856883049011,
      "learning_rate": 6.978322378276581e-06,
      "loss": 0.3563,
      "step": 25045
    },
    {
      "epoch": 1.1493735946032766,
      "grad_norm": 0.47052228450775146,
      "learning_rate": 6.978097196517189e-06,
      "loss": 0.3626,
      "step": 25046
    },
    {
      "epoch": 1.149419485108531,
      "grad_norm": 0.4688790440559387,
      "learning_rate": 6.977872010001094e-06,
      "loss": 0.3564,
      "step": 25047
    },
    {
      "epoch": 1.1494653756137856,
      "grad_norm": 0.5200122594833374,
      "learning_rate": 6.977646818728842e-06,
      "loss": 0.3544,
      "step": 25048
    },
    {
      "epoch": 1.14951126611904,
      "grad_norm": 0.4934723973274231,
      "learning_rate": 6.977421622700976e-06,
      "loss": 0.4402,
      "step": 25049
    },
    {
      "epoch": 1.1495571566242944,
      "grad_norm": 0.46207094192504883,
      "learning_rate": 6.9771964219180324e-06,
      "loss": 0.3619,
      "step": 25050
    },
    {
      "epoch": 1.1496030471295489,
      "grad_norm": 0.4683164954185486,
      "learning_rate": 6.976971216380556e-06,
      "loss": 0.3335,
      "step": 25051
    },
    {
      "epoch": 1.1496489376348034,
      "grad_norm": 0.45859313011169434,
      "learning_rate": 6.976746006089088e-06,
      "loss": 0.3262,
      "step": 25052
    },
    {
      "epoch": 1.1496948281400579,
      "grad_norm": 0.44494688510894775,
      "learning_rate": 6.97652079104417e-06,
      "loss": 0.3541,
      "step": 25053
    },
    {
      "epoch": 1.1497407186453124,
      "grad_norm": 0.4629078805446625,
      "learning_rate": 6.976295571246343e-06,
      "loss": 0.3477,
      "step": 25054
    },
    {
      "epoch": 1.1497866091505666,
      "grad_norm": 0.48030275106430054,
      "learning_rate": 6.97607034669615e-06,
      "loss": 0.3157,
      "step": 25055
    },
    {
      "epoch": 1.1498324996558211,
      "grad_norm": 0.4573032259941101,
      "learning_rate": 6.97584511739413e-06,
      "loss": 0.3494,
      "step": 25056
    },
    {
      "epoch": 1.1498783901610756,
      "grad_norm": 0.4814562201499939,
      "learning_rate": 6.9756198833408266e-06,
      "loss": 0.3862,
      "step": 25057
    },
    {
      "epoch": 1.1499242806663301,
      "grad_norm": 0.4575633704662323,
      "learning_rate": 6.975394644536781e-06,
      "loss": 0.3757,
      "step": 25058
    },
    {
      "epoch": 1.1499701711715846,
      "grad_norm": 0.5048182010650635,
      "learning_rate": 6.975169400982536e-06,
      "loss": 0.4492,
      "step": 25059
    },
    {
      "epoch": 1.1500160616768391,
      "grad_norm": 0.5243398547172546,
      "learning_rate": 6.97494415267863e-06,
      "loss": 0.3533,
      "step": 25060
    },
    {
      "epoch": 1.1500619521820936,
      "grad_norm": 0.45034703612327576,
      "learning_rate": 6.974718899625606e-06,
      "loss": 0.313,
      "step": 25061
    },
    {
      "epoch": 1.150107842687348,
      "grad_norm": 0.4944525957107544,
      "learning_rate": 6.974493641824007e-06,
      "loss": 0.3524,
      "step": 25062
    },
    {
      "epoch": 1.1501537331926024,
      "grad_norm": 0.4832689166069031,
      "learning_rate": 6.974268379274376e-06,
      "loss": 0.3611,
      "step": 25063
    },
    {
      "epoch": 1.1501996236978569,
      "grad_norm": 0.4869072437286377,
      "learning_rate": 6.974043111977251e-06,
      "loss": 0.3756,
      "step": 25064
    },
    {
      "epoch": 1.1502455142031114,
      "grad_norm": 0.49842196702957153,
      "learning_rate": 6.9738178399331746e-06,
      "loss": 0.4054,
      "step": 25065
    },
    {
      "epoch": 1.1502914047083659,
      "grad_norm": 0.4746134877204895,
      "learning_rate": 6.97359256314269e-06,
      "loss": 0.3668,
      "step": 25066
    },
    {
      "epoch": 1.1503372952136204,
      "grad_norm": 0.4971030056476593,
      "learning_rate": 6.9733672816063394e-06,
      "loss": 0.3943,
      "step": 25067
    },
    {
      "epoch": 1.1503831857188747,
      "grad_norm": 0.4551909267902374,
      "learning_rate": 6.973141995324661e-06,
      "loss": 0.3214,
      "step": 25068
    },
    {
      "epoch": 1.1504290762241292,
      "grad_norm": 0.5224339962005615,
      "learning_rate": 6.972916704298201e-06,
      "loss": 0.4281,
      "step": 25069
    },
    {
      "epoch": 1.1504749667293837,
      "grad_norm": 0.5089649558067322,
      "learning_rate": 6.972691408527498e-06,
      "loss": 0.3664,
      "step": 25070
    },
    {
      "epoch": 1.1505208572346382,
      "grad_norm": 0.4609854221343994,
      "learning_rate": 6.972466108013094e-06,
      "loss": 0.3539,
      "step": 25071
    },
    {
      "epoch": 1.1505667477398926,
      "grad_norm": 0.46647322177886963,
      "learning_rate": 6.972240802755533e-06,
      "loss": 0.4073,
      "step": 25072
    },
    {
      "epoch": 1.1506126382451471,
      "grad_norm": 0.511405885219574,
      "learning_rate": 6.972015492755354e-06,
      "loss": 0.3882,
      "step": 25073
    },
    {
      "epoch": 1.1506585287504016,
      "grad_norm": 0.4898398816585541,
      "learning_rate": 6.9717901780131015e-06,
      "loss": 0.4171,
      "step": 25074
    },
    {
      "epoch": 1.150704419255656,
      "grad_norm": 0.46157148480415344,
      "learning_rate": 6.9715648585293165e-06,
      "loss": 0.3196,
      "step": 25075
    },
    {
      "epoch": 1.1507503097609104,
      "grad_norm": 0.43727874755859375,
      "learning_rate": 6.971339534304537e-06,
      "loss": 0.2954,
      "step": 25076
    },
    {
      "epoch": 1.150796200266165,
      "grad_norm": 0.4599415063858032,
      "learning_rate": 6.971114205339311e-06,
      "loss": 0.3868,
      "step": 25077
    },
    {
      "epoch": 1.1508420907714194,
      "grad_norm": 0.4168073534965515,
      "learning_rate": 6.970888871634176e-06,
      "loss": 0.3008,
      "step": 25078
    },
    {
      "epoch": 1.150887981276674,
      "grad_norm": 0.526592493057251,
      "learning_rate": 6.970663533189676e-06,
      "loss": 0.5096,
      "step": 25079
    },
    {
      "epoch": 1.1509338717819284,
      "grad_norm": 0.43107932806015015,
      "learning_rate": 6.970438190006352e-06,
      "loss": 0.3221,
      "step": 25080
    },
    {
      "epoch": 1.1509797622871827,
      "grad_norm": 0.4706384837627411,
      "learning_rate": 6.970212842084746e-06,
      "loss": 0.3845,
      "step": 25081
    },
    {
      "epoch": 1.1510256527924372,
      "grad_norm": 1.004956841468811,
      "learning_rate": 6.9699874894253994e-06,
      "loss": 0.4594,
      "step": 25082
    },
    {
      "epoch": 1.1510715432976917,
      "grad_norm": 0.4478180408477783,
      "learning_rate": 6.969762132028856e-06,
      "loss": 0.3324,
      "step": 25083
    },
    {
      "epoch": 1.1511174338029462,
      "grad_norm": 0.4622480869293213,
      "learning_rate": 6.969536769895655e-06,
      "loss": 0.342,
      "step": 25084
    },
    {
      "epoch": 1.1511633243082007,
      "grad_norm": 0.4555993676185608,
      "learning_rate": 6.96931140302634e-06,
      "loss": 0.3336,
      "step": 25085
    },
    {
      "epoch": 1.1512092148134552,
      "grad_norm": 0.5121057033538818,
      "learning_rate": 6.969086031421454e-06,
      "loss": 0.4062,
      "step": 25086
    },
    {
      "epoch": 1.1512551053187097,
      "grad_norm": 0.4331211745738983,
      "learning_rate": 6.968860655081535e-06,
      "loss": 0.3127,
      "step": 25087
    },
    {
      "epoch": 1.151300995823964,
      "grad_norm": 0.5448230504989624,
      "learning_rate": 6.968635274007128e-06,
      "loss": 0.4088,
      "step": 25088
    },
    {
      "epoch": 1.1513468863292184,
      "grad_norm": 0.4776642620563507,
      "learning_rate": 6.968409888198777e-06,
      "loss": 0.384,
      "step": 25089
    },
    {
      "epoch": 1.151392776834473,
      "grad_norm": 0.44804856181144714,
      "learning_rate": 6.968184497657019e-06,
      "loss": 0.2944,
      "step": 25090
    },
    {
      "epoch": 1.1514386673397274,
      "grad_norm": 0.5468553900718689,
      "learning_rate": 6.9679591023824e-06,
      "loss": 0.46,
      "step": 25091
    },
    {
      "epoch": 1.151484557844982,
      "grad_norm": 0.4704417884349823,
      "learning_rate": 6.96773370237546e-06,
      "loss": 0.3806,
      "step": 25092
    },
    {
      "epoch": 1.1515304483502362,
      "grad_norm": 0.4762265086174011,
      "learning_rate": 6.967508297636741e-06,
      "loss": 0.3407,
      "step": 25093
    },
    {
      "epoch": 1.1515763388554907,
      "grad_norm": 0.4593409597873688,
      "learning_rate": 6.967282888166785e-06,
      "loss": 0.3661,
      "step": 25094
    },
    {
      "epoch": 1.1516222293607452,
      "grad_norm": 0.44153115153312683,
      "learning_rate": 6.967057473966136e-06,
      "loss": 0.3346,
      "step": 25095
    },
    {
      "epoch": 1.1516681198659997,
      "grad_norm": 0.4575146436691284,
      "learning_rate": 6.966832055035334e-06,
      "loss": 0.3252,
      "step": 25096
    },
    {
      "epoch": 1.1517140103712542,
      "grad_norm": 0.42585891485214233,
      "learning_rate": 6.966606631374922e-06,
      "loss": 0.3197,
      "step": 25097
    },
    {
      "epoch": 1.1517599008765087,
      "grad_norm": 0.4820810556411743,
      "learning_rate": 6.966381202985441e-06,
      "loss": 0.3356,
      "step": 25098
    },
    {
      "epoch": 1.1518057913817632,
      "grad_norm": 0.4558377265930176,
      "learning_rate": 6.966155769867433e-06,
      "loss": 0.3596,
      "step": 25099
    },
    {
      "epoch": 1.1518516818870177,
      "grad_norm": 0.5104401111602783,
      "learning_rate": 6.965930332021442e-06,
      "loss": 0.4248,
      "step": 25100
    },
    {
      "epoch": 1.151897572392272,
      "grad_norm": 0.46312642097473145,
      "learning_rate": 6.965704889448009e-06,
      "loss": 0.3531,
      "step": 25101
    },
    {
      "epoch": 1.1519434628975265,
      "grad_norm": 0.49762675166130066,
      "learning_rate": 6.965479442147677e-06,
      "loss": 0.4242,
      "step": 25102
    },
    {
      "epoch": 1.151989353402781,
      "grad_norm": 0.4090297520160675,
      "learning_rate": 6.965253990120986e-06,
      "loss": 0.2829,
      "step": 25103
    },
    {
      "epoch": 1.1520352439080355,
      "grad_norm": 0.47023677825927734,
      "learning_rate": 6.96502853336848e-06,
      "loss": 0.3319,
      "step": 25104
    },
    {
      "epoch": 1.15208113441329,
      "grad_norm": 0.5168111324310303,
      "learning_rate": 6.9648030718907e-06,
      "loss": 0.4479,
      "step": 25105
    },
    {
      "epoch": 1.1521270249185442,
      "grad_norm": 0.5148356556892395,
      "learning_rate": 6.964577605688189e-06,
      "loss": 0.4161,
      "step": 25106
    },
    {
      "epoch": 1.1521729154237987,
      "grad_norm": 0.4504427909851074,
      "learning_rate": 6.964352134761489e-06,
      "loss": 0.3489,
      "step": 25107
    },
    {
      "epoch": 1.1522188059290532,
      "grad_norm": 0.4421009123325348,
      "learning_rate": 6.964126659111144e-06,
      "loss": 0.3144,
      "step": 25108
    },
    {
      "epoch": 1.1522646964343077,
      "grad_norm": 0.4721224308013916,
      "learning_rate": 6.9639011787376905e-06,
      "loss": 0.3455,
      "step": 25109
    },
    {
      "epoch": 1.1523105869395622,
      "grad_norm": 0.4540870785713196,
      "learning_rate": 6.963675693641676e-06,
      "loss": 0.3221,
      "step": 25110
    },
    {
      "epoch": 1.1523564774448167,
      "grad_norm": 0.4359225630760193,
      "learning_rate": 6.9634502038236415e-06,
      "loss": 0.2943,
      "step": 25111
    },
    {
      "epoch": 1.1524023679500712,
      "grad_norm": 0.4352863132953644,
      "learning_rate": 6.963224709284129e-06,
      "loss": 0.2913,
      "step": 25112
    },
    {
      "epoch": 1.1524482584553255,
      "grad_norm": 0.48331311345100403,
      "learning_rate": 6.962999210023681e-06,
      "loss": 0.3788,
      "step": 25113
    },
    {
      "epoch": 1.15249414896058,
      "grad_norm": 0.4596453607082367,
      "learning_rate": 6.96277370604284e-06,
      "loss": 0.3547,
      "step": 25114
    },
    {
      "epoch": 1.1525400394658345,
      "grad_norm": 0.4663882255554199,
      "learning_rate": 6.962548197342146e-06,
      "loss": 0.4118,
      "step": 25115
    },
    {
      "epoch": 1.152585929971089,
      "grad_norm": 0.4594615399837494,
      "learning_rate": 6.962322683922143e-06,
      "loss": 0.3384,
      "step": 25116
    },
    {
      "epoch": 1.1526318204763435,
      "grad_norm": 0.46238380670547485,
      "learning_rate": 6.962097165783375e-06,
      "loss": 0.3311,
      "step": 25117
    },
    {
      "epoch": 1.152677710981598,
      "grad_norm": 0.44720688462257385,
      "learning_rate": 6.961871642926382e-06,
      "loss": 0.3011,
      "step": 25118
    },
    {
      "epoch": 1.1527236014868523,
      "grad_norm": 0.46242305636405945,
      "learning_rate": 6.961646115351706e-06,
      "loss": 0.3516,
      "step": 25119
    },
    {
      "epoch": 1.1527694919921068,
      "grad_norm": 0.4658792316913605,
      "learning_rate": 6.961420583059891e-06,
      "loss": 0.4176,
      "step": 25120
    },
    {
      "epoch": 1.1528153824973613,
      "grad_norm": 0.4210544228553772,
      "learning_rate": 6.961195046051478e-06,
      "loss": 0.293,
      "step": 25121
    },
    {
      "epoch": 1.1528612730026158,
      "grad_norm": 0.4550395905971527,
      "learning_rate": 6.96096950432701e-06,
      "loss": 0.3636,
      "step": 25122
    },
    {
      "epoch": 1.1529071635078703,
      "grad_norm": 0.4997739791870117,
      "learning_rate": 6.960743957887029e-06,
      "loss": 0.4454,
      "step": 25123
    },
    {
      "epoch": 1.1529530540131248,
      "grad_norm": 0.41521745920181274,
      "learning_rate": 6.9605184067320785e-06,
      "loss": 0.2538,
      "step": 25124
    },
    {
      "epoch": 1.1529989445183793,
      "grad_norm": 0.4375258982181549,
      "learning_rate": 6.9602928508627e-06,
      "loss": 0.3575,
      "step": 25125
    },
    {
      "epoch": 1.1530448350236335,
      "grad_norm": 0.4523445963859558,
      "learning_rate": 6.9600672902794355e-06,
      "loss": 0.3506,
      "step": 25126
    },
    {
      "epoch": 1.153090725528888,
      "grad_norm": 0.4959537386894226,
      "learning_rate": 6.959841724982827e-06,
      "loss": 0.3849,
      "step": 25127
    },
    {
      "epoch": 1.1531366160341425,
      "grad_norm": 0.44030407071113586,
      "learning_rate": 6.959616154973419e-06,
      "loss": 0.3315,
      "step": 25128
    },
    {
      "epoch": 1.153182506539397,
      "grad_norm": 0.4650985300540924,
      "learning_rate": 6.9593905802517525e-06,
      "loss": 0.3946,
      "step": 25129
    },
    {
      "epoch": 1.1532283970446515,
      "grad_norm": 0.5176205039024353,
      "learning_rate": 6.959165000818369e-06,
      "loss": 0.3847,
      "step": 25130
    },
    {
      "epoch": 1.153274287549906,
      "grad_norm": 0.4602415859699249,
      "learning_rate": 6.958939416673812e-06,
      "loss": 0.329,
      "step": 25131
    },
    {
      "epoch": 1.1533201780551603,
      "grad_norm": 0.4734410047531128,
      "learning_rate": 6.958713827818624e-06,
      "loss": 0.3744,
      "step": 25132
    },
    {
      "epoch": 1.1533660685604148,
      "grad_norm": 0.48495838046073914,
      "learning_rate": 6.958488234253347e-06,
      "loss": 0.4061,
      "step": 25133
    },
    {
      "epoch": 1.1534119590656693,
      "grad_norm": 0.4616854190826416,
      "learning_rate": 6.958262635978526e-06,
      "loss": 0.348,
      "step": 25134
    },
    {
      "epoch": 1.1534578495709238,
      "grad_norm": 0.4614546298980713,
      "learning_rate": 6.958037032994701e-06,
      "loss": 0.3626,
      "step": 25135
    },
    {
      "epoch": 1.1535037400761783,
      "grad_norm": 0.46961313486099243,
      "learning_rate": 6.9578114253024144e-06,
      "loss": 0.3573,
      "step": 25136
    },
    {
      "epoch": 1.1535496305814328,
      "grad_norm": 0.5064706206321716,
      "learning_rate": 6.9575858129022085e-06,
      "loss": 0.3466,
      "step": 25137
    },
    {
      "epoch": 1.1535955210866873,
      "grad_norm": 0.490616112947464,
      "learning_rate": 6.957360195794627e-06,
      "loss": 0.4145,
      "step": 25138
    },
    {
      "epoch": 1.1536414115919416,
      "grad_norm": 0.44774553179740906,
      "learning_rate": 6.957134573980212e-06,
      "loss": 0.342,
      "step": 25139
    },
    {
      "epoch": 1.153687302097196,
      "grad_norm": 0.470467209815979,
      "learning_rate": 6.956908947459507e-06,
      "loss": 0.3837,
      "step": 25140
    },
    {
      "epoch": 1.1537331926024506,
      "grad_norm": 0.4633384048938751,
      "learning_rate": 6.956683316233052e-06,
      "loss": 0.3311,
      "step": 25141
    },
    {
      "epoch": 1.153779083107705,
      "grad_norm": 0.47774162888526917,
      "learning_rate": 6.956457680301391e-06,
      "loss": 0.3326,
      "step": 25142
    },
    {
      "epoch": 1.1538249736129595,
      "grad_norm": 0.4819404184818268,
      "learning_rate": 6.9562320396650685e-06,
      "loss": 0.3999,
      "step": 25143
    },
    {
      "epoch": 1.1538708641182138,
      "grad_norm": 0.47395989298820496,
      "learning_rate": 6.956006394324624e-06,
      "loss": 0.3531,
      "step": 25144
    },
    {
      "epoch": 1.1539167546234683,
      "grad_norm": 0.4633278250694275,
      "learning_rate": 6.9557807442806025e-06,
      "loss": 0.3429,
      "step": 25145
    },
    {
      "epoch": 1.1539626451287228,
      "grad_norm": 0.49769124388694763,
      "learning_rate": 6.955555089533545e-06,
      "loss": 0.3978,
      "step": 25146
    },
    {
      "epoch": 1.1540085356339773,
      "grad_norm": 0.5108813643455505,
      "learning_rate": 6.955329430083995e-06,
      "loss": 0.4341,
      "step": 25147
    },
    {
      "epoch": 1.1540544261392318,
      "grad_norm": 0.44003668427467346,
      "learning_rate": 6.955103765932494e-06,
      "loss": 0.3034,
      "step": 25148
    },
    {
      "epoch": 1.1541003166444863,
      "grad_norm": 0.4526883065700531,
      "learning_rate": 6.954878097079584e-06,
      "loss": 0.2991,
      "step": 25149
    },
    {
      "epoch": 1.1541462071497408,
      "grad_norm": 0.46105608344078064,
      "learning_rate": 6.954652423525813e-06,
      "loss": 0.3586,
      "step": 25150
    },
    {
      "epoch": 1.154192097654995,
      "grad_norm": 0.44052544236183167,
      "learning_rate": 6.954426745271719e-06,
      "loss": 0.3502,
      "step": 25151
    },
    {
      "epoch": 1.1542379881602496,
      "grad_norm": 0.46082603931427,
      "learning_rate": 6.954201062317843e-06,
      "loss": 0.3472,
      "step": 25152
    },
    {
      "epoch": 1.154283878665504,
      "grad_norm": 0.4669497013092041,
      "learning_rate": 6.9539753746647325e-06,
      "loss": 0.3293,
      "step": 25153
    },
    {
      "epoch": 1.1543297691707586,
      "grad_norm": 0.431634783744812,
      "learning_rate": 6.953749682312927e-06,
      "loss": 0.2974,
      "step": 25154
    },
    {
      "epoch": 1.154375659676013,
      "grad_norm": 0.5211135745048523,
      "learning_rate": 6.9535239852629695e-06,
      "loss": 0.4514,
      "step": 25155
    },
    {
      "epoch": 1.1544215501812676,
      "grad_norm": 0.4230610728263855,
      "learning_rate": 6.953298283515404e-06,
      "loss": 0.2917,
      "step": 25156
    },
    {
      "epoch": 1.1544674406865219,
      "grad_norm": 0.4881574809551239,
      "learning_rate": 6.953072577070774e-06,
      "loss": 0.3732,
      "step": 25157
    },
    {
      "epoch": 1.1545133311917763,
      "grad_norm": 0.4339147210121155,
      "learning_rate": 6.9528468659296185e-06,
      "loss": 0.3174,
      "step": 25158
    },
    {
      "epoch": 1.1545592216970308,
      "grad_norm": 0.45540401339530945,
      "learning_rate": 6.952621150092484e-06,
      "loss": 0.3283,
      "step": 25159
    },
    {
      "epoch": 1.1546051122022853,
      "grad_norm": 0.48818036913871765,
      "learning_rate": 6.952395429559912e-06,
      "loss": 0.3536,
      "step": 25160
    },
    {
      "epoch": 1.1546510027075398,
      "grad_norm": 0.43416881561279297,
      "learning_rate": 6.9521697043324455e-06,
      "loss": 0.3232,
      "step": 25161
    },
    {
      "epoch": 1.1546968932127943,
      "grad_norm": 0.4796547591686249,
      "learning_rate": 6.951943974410626e-06,
      "loss": 0.3891,
      "step": 25162
    },
    {
      "epoch": 1.1547427837180488,
      "grad_norm": 0.46452245116233826,
      "learning_rate": 6.951718239794998e-06,
      "loss": 0.3416,
      "step": 25163
    },
    {
      "epoch": 1.1547886742233031,
      "grad_norm": 0.46585923433303833,
      "learning_rate": 6.951492500486101e-06,
      "loss": 0.3264,
      "step": 25164
    },
    {
      "epoch": 1.1548345647285576,
      "grad_norm": 0.41096577048301697,
      "learning_rate": 6.951266756484484e-06,
      "loss": 0.2852,
      "step": 25165
    },
    {
      "epoch": 1.154880455233812,
      "grad_norm": 0.47545772790908813,
      "learning_rate": 6.951041007790684e-06,
      "loss": 0.3939,
      "step": 25166
    },
    {
      "epoch": 1.1549263457390666,
      "grad_norm": 0.45712143182754517,
      "learning_rate": 6.950815254405248e-06,
      "loss": 0.3219,
      "step": 25167
    },
    {
      "epoch": 1.154972236244321,
      "grad_norm": 0.4984951317310333,
      "learning_rate": 6.9505894963287166e-06,
      "loss": 0.3738,
      "step": 25168
    },
    {
      "epoch": 1.1550181267495756,
      "grad_norm": 0.4408794045448303,
      "learning_rate": 6.950363733561631e-06,
      "loss": 0.3205,
      "step": 25169
    },
    {
      "epoch": 1.1550640172548299,
      "grad_norm": 0.4682091772556305,
      "learning_rate": 6.950137966104538e-06,
      "loss": 0.3211,
      "step": 25170
    },
    {
      "epoch": 1.1551099077600844,
      "grad_norm": 0.44546735286712646,
      "learning_rate": 6.94991219395798e-06,
      "loss": 0.3006,
      "step": 25171
    },
    {
      "epoch": 1.1551557982653389,
      "grad_norm": 0.5079686045646667,
      "learning_rate": 6.949686417122496e-06,
      "loss": 0.3882,
      "step": 25172
    },
    {
      "epoch": 1.1552016887705934,
      "grad_norm": 0.4540318548679352,
      "learning_rate": 6.949460635598633e-06,
      "loss": 0.3156,
      "step": 25173
    },
    {
      "epoch": 1.1552475792758479,
      "grad_norm": 0.4677923917770386,
      "learning_rate": 6.949234849386932e-06,
      "loss": 0.348,
      "step": 25174
    },
    {
      "epoch": 1.1552934697811024,
      "grad_norm": 0.44999071955680847,
      "learning_rate": 6.949009058487935e-06,
      "loss": 0.3397,
      "step": 25175
    },
    {
      "epoch": 1.1553393602863569,
      "grad_norm": 0.48444098234176636,
      "learning_rate": 6.948783262902189e-06,
      "loss": 0.3786,
      "step": 25176
    },
    {
      "epoch": 1.1553852507916111,
      "grad_norm": 0.4763903319835663,
      "learning_rate": 6.948557462630232e-06,
      "loss": 0.3723,
      "step": 25177
    },
    {
      "epoch": 1.1554311412968656,
      "grad_norm": 0.5264382362365723,
      "learning_rate": 6.948331657672611e-06,
      "loss": 0.3931,
      "step": 25178
    },
    {
      "epoch": 1.1554770318021201,
      "grad_norm": 0.4536794424057007,
      "learning_rate": 6.948105848029867e-06,
      "loss": 0.3503,
      "step": 25179
    },
    {
      "epoch": 1.1555229223073746,
      "grad_norm": 0.4981917142868042,
      "learning_rate": 6.947880033702542e-06,
      "loss": 0.415,
      "step": 25180
    },
    {
      "epoch": 1.1555688128126291,
      "grad_norm": 0.4961695969104767,
      "learning_rate": 6.947654214691181e-06,
      "loss": 0.4655,
      "step": 25181
    },
    {
      "epoch": 1.1556147033178834,
      "grad_norm": 0.49622824788093567,
      "learning_rate": 6.947428390996327e-06,
      "loss": 0.3966,
      "step": 25182
    },
    {
      "epoch": 1.155660593823138,
      "grad_norm": 0.4609910845756531,
      "learning_rate": 6.947202562618521e-06,
      "loss": 0.3658,
      "step": 25183
    },
    {
      "epoch": 1.1557064843283924,
      "grad_norm": 0.45820268988609314,
      "learning_rate": 6.946976729558308e-06,
      "loss": 0.3258,
      "step": 25184
    },
    {
      "epoch": 1.155752374833647,
      "grad_norm": 0.5128822326660156,
      "learning_rate": 6.946750891816232e-06,
      "loss": 0.4379,
      "step": 25185
    },
    {
      "epoch": 1.1557982653389014,
      "grad_norm": 0.46970242261886597,
      "learning_rate": 6.946525049392832e-06,
      "loss": 0.3597,
      "step": 25186
    },
    {
      "epoch": 1.155844155844156,
      "grad_norm": 0.44096696376800537,
      "learning_rate": 6.946299202288653e-06,
      "loss": 0.2919,
      "step": 25187
    },
    {
      "epoch": 1.1558900463494104,
      "grad_norm": 0.44802722334861755,
      "learning_rate": 6.9460733505042425e-06,
      "loss": 0.3339,
      "step": 25188
    },
    {
      "epoch": 1.155935936854665,
      "grad_norm": 0.4844057261943817,
      "learning_rate": 6.945847494040138e-06,
      "loss": 0.409,
      "step": 25189
    },
    {
      "epoch": 1.1559818273599192,
      "grad_norm": 0.4761686325073242,
      "learning_rate": 6.945621632896884e-06,
      "loss": 0.3674,
      "step": 25190
    },
    {
      "epoch": 1.1560277178651737,
      "grad_norm": 0.46879076957702637,
      "learning_rate": 6.9453957670750234e-06,
      "loss": 0.3138,
      "step": 25191
    },
    {
      "epoch": 1.1560736083704282,
      "grad_norm": 0.4658149778842926,
      "learning_rate": 6.9451698965751e-06,
      "loss": 0.3429,
      "step": 25192
    },
    {
      "epoch": 1.1561194988756827,
      "grad_norm": 0.45785707235336304,
      "learning_rate": 6.944944021397658e-06,
      "loss": 0.3426,
      "step": 25193
    },
    {
      "epoch": 1.1561653893809372,
      "grad_norm": 0.5196152925491333,
      "learning_rate": 6.94471814154324e-06,
      "loss": 0.4596,
      "step": 25194
    },
    {
      "epoch": 1.1562112798861914,
      "grad_norm": 0.4676278233528137,
      "learning_rate": 6.944492257012387e-06,
      "loss": 0.3429,
      "step": 25195
    },
    {
      "epoch": 1.156257170391446,
      "grad_norm": 0.47010675072669983,
      "learning_rate": 6.9442663678056455e-06,
      "loss": 0.3805,
      "step": 25196
    },
    {
      "epoch": 1.1563030608967004,
      "grad_norm": 0.4876898527145386,
      "learning_rate": 6.944040473923554e-06,
      "loss": 0.3749,
      "step": 25197
    },
    {
      "epoch": 1.156348951401955,
      "grad_norm": 0.5293440818786621,
      "learning_rate": 6.943814575366661e-06,
      "loss": 0.4463,
      "step": 25198
    },
    {
      "epoch": 1.1563948419072094,
      "grad_norm": 0.4519314467906952,
      "learning_rate": 6.943588672135509e-06,
      "loss": 0.3017,
      "step": 25199
    },
    {
      "epoch": 1.156440732412464,
      "grad_norm": 0.4078337848186493,
      "learning_rate": 6.943362764230636e-06,
      "loss": 0.2876,
      "step": 25200
    },
    {
      "epoch": 1.1564866229177184,
      "grad_norm": 0.5074722766876221,
      "learning_rate": 6.943136851652591e-06,
      "loss": 0.3904,
      "step": 25201
    },
    {
      "epoch": 1.1565325134229727,
      "grad_norm": 0.4733673632144928,
      "learning_rate": 6.9429109344019156e-06,
      "loss": 0.3445,
      "step": 25202
    },
    {
      "epoch": 1.1565784039282272,
      "grad_norm": 0.49511849880218506,
      "learning_rate": 6.942685012479151e-06,
      "loss": 0.429,
      "step": 25203
    },
    {
      "epoch": 1.1566242944334817,
      "grad_norm": 0.46987661719322205,
      "learning_rate": 6.9424590858848425e-06,
      "loss": 0.3306,
      "step": 25204
    },
    {
      "epoch": 1.1566701849387362,
      "grad_norm": 0.44555899500846863,
      "learning_rate": 6.942233154619535e-06,
      "loss": 0.3304,
      "step": 25205
    },
    {
      "epoch": 1.1567160754439907,
      "grad_norm": 0.4216049313545227,
      "learning_rate": 6.9420072186837685e-06,
      "loss": 0.3068,
      "step": 25206
    },
    {
      "epoch": 1.1567619659492452,
      "grad_norm": 0.48061463236808777,
      "learning_rate": 6.9417812780780865e-06,
      "loss": 0.371,
      "step": 25207
    },
    {
      "epoch": 1.1568078564544995,
      "grad_norm": 0.46951696276664734,
      "learning_rate": 6.941555332803034e-06,
      "loss": 0.4055,
      "step": 25208
    },
    {
      "epoch": 1.156853746959754,
      "grad_norm": 0.4714525640010834,
      "learning_rate": 6.941329382859153e-06,
      "loss": 0.3785,
      "step": 25209
    },
    {
      "epoch": 1.1568996374650085,
      "grad_norm": 0.4900856614112854,
      "learning_rate": 6.941103428246989e-06,
      "loss": 0.3723,
      "step": 25210
    },
    {
      "epoch": 1.156945527970263,
      "grad_norm": 0.5066500306129456,
      "learning_rate": 6.9408774689670845e-06,
      "loss": 0.4304,
      "step": 25211
    },
    {
      "epoch": 1.1569914184755175,
      "grad_norm": 0.4639904201030731,
      "learning_rate": 6.94065150501998e-06,
      "loss": 0.3456,
      "step": 25212
    },
    {
      "epoch": 1.157037308980772,
      "grad_norm": 0.46823763847351074,
      "learning_rate": 6.9404255364062226e-06,
      "loss": 0.3194,
      "step": 25213
    },
    {
      "epoch": 1.1570831994860264,
      "grad_norm": 0.47651904821395874,
      "learning_rate": 6.940199563126353e-06,
      "loss": 0.384,
      "step": 25214
    },
    {
      "epoch": 1.1571290899912807,
      "grad_norm": 0.4486108422279358,
      "learning_rate": 6.939973585180916e-06,
      "loss": 0.3337,
      "step": 25215
    },
    {
      "epoch": 1.1571749804965352,
      "grad_norm": 0.45768940448760986,
      "learning_rate": 6.939747602570457e-06,
      "loss": 0.3306,
      "step": 25216
    },
    {
      "epoch": 1.1572208710017897,
      "grad_norm": 0.510412871837616,
      "learning_rate": 6.939521615295515e-06,
      "loss": 0.3889,
      "step": 25217
    },
    {
      "epoch": 1.1572667615070442,
      "grad_norm": 0.4649767577648163,
      "learning_rate": 6.9392956233566344e-06,
      "loss": 0.3591,
      "step": 25218
    },
    {
      "epoch": 1.1573126520122987,
      "grad_norm": 0.4374452829360962,
      "learning_rate": 6.939069626754362e-06,
      "loss": 0.3173,
      "step": 25219
    },
    {
      "epoch": 1.1573585425175532,
      "grad_norm": 0.45500296354293823,
      "learning_rate": 6.938843625489239e-06,
      "loss": 0.3459,
      "step": 25220
    },
    {
      "epoch": 1.1574044330228075,
      "grad_norm": 0.47296443581581116,
      "learning_rate": 6.9386176195618094e-06,
      "loss": 0.4088,
      "step": 25221
    },
    {
      "epoch": 1.157450323528062,
      "grad_norm": 0.441616028547287,
      "learning_rate": 6.938391608972615e-06,
      "loss": 0.3063,
      "step": 25222
    },
    {
      "epoch": 1.1574962140333165,
      "grad_norm": 0.5002726912498474,
      "learning_rate": 6.938165593722201e-06,
      "loss": 0.3773,
      "step": 25223
    },
    {
      "epoch": 1.157542104538571,
      "grad_norm": 0.4404740035533905,
      "learning_rate": 6.937939573811111e-06,
      "loss": 0.3323,
      "step": 25224
    },
    {
      "epoch": 1.1575879950438255,
      "grad_norm": 0.46229061484336853,
      "learning_rate": 6.937713549239887e-06,
      "loss": 0.3706,
      "step": 25225
    },
    {
      "epoch": 1.15763388554908,
      "grad_norm": 0.47997400164604187,
      "learning_rate": 6.937487520009074e-06,
      "loss": 0.4016,
      "step": 25226
    },
    {
      "epoch": 1.1576797760543345,
      "grad_norm": 0.4615573287010193,
      "learning_rate": 6.937261486119216e-06,
      "loss": 0.349,
      "step": 25227
    },
    {
      "epoch": 1.1577256665595888,
      "grad_norm": 0.47035470604896545,
      "learning_rate": 6.937035447570855e-06,
      "loss": 0.3743,
      "step": 25228
    },
    {
      "epoch": 1.1577715570648432,
      "grad_norm": 0.46883466839790344,
      "learning_rate": 6.9368094043645325e-06,
      "loss": 0.3554,
      "step": 25229
    },
    {
      "epoch": 1.1578174475700977,
      "grad_norm": 0.4617813229560852,
      "learning_rate": 6.936583356500799e-06,
      "loss": 0.3241,
      "step": 25230
    },
    {
      "epoch": 1.1578633380753522,
      "grad_norm": 0.4416572153568268,
      "learning_rate": 6.93635730398019e-06,
      "loss": 0.3189,
      "step": 25231
    },
    {
      "epoch": 1.1579092285806067,
      "grad_norm": 0.4927392899990082,
      "learning_rate": 6.9361312468032545e-06,
      "loss": 0.4103,
      "step": 25232
    },
    {
      "epoch": 1.157955119085861,
      "grad_norm": 0.48185041546821594,
      "learning_rate": 6.935905184970535e-06,
      "loss": 0.3948,
      "step": 25233
    },
    {
      "epoch": 1.1580010095911155,
      "grad_norm": 0.505353569984436,
      "learning_rate": 6.935679118482574e-06,
      "loss": 0.4383,
      "step": 25234
    },
    {
      "epoch": 1.15804690009637,
      "grad_norm": 0.4556822180747986,
      "learning_rate": 6.935453047339915e-06,
      "loss": 0.3129,
      "step": 25235
    },
    {
      "epoch": 1.1580927906016245,
      "grad_norm": 0.4684167504310608,
      "learning_rate": 6.935226971543103e-06,
      "loss": 0.3553,
      "step": 25236
    },
    {
      "epoch": 1.158138681106879,
      "grad_norm": 0.48484474420547485,
      "learning_rate": 6.93500089109268e-06,
      "loss": 0.4001,
      "step": 25237
    },
    {
      "epoch": 1.1581845716121335,
      "grad_norm": 0.41930991411209106,
      "learning_rate": 6.934774805989191e-06,
      "loss": 0.2786,
      "step": 25238
    },
    {
      "epoch": 1.158230462117388,
      "grad_norm": 0.43269678950309753,
      "learning_rate": 6.93454871623318e-06,
      "loss": 0.3242,
      "step": 25239
    },
    {
      "epoch": 1.1582763526226423,
      "grad_norm": 0.5172913670539856,
      "learning_rate": 6.934322621825189e-06,
      "loss": 0.3978,
      "step": 25240
    },
    {
      "epoch": 1.1583222431278968,
      "grad_norm": 0.45003658533096313,
      "learning_rate": 6.934096522765763e-06,
      "loss": 0.3097,
      "step": 25241
    },
    {
      "epoch": 1.1583681336331513,
      "grad_norm": 0.4628738760948181,
      "learning_rate": 6.933870419055445e-06,
      "loss": 0.3356,
      "step": 25242
    },
    {
      "epoch": 1.1584140241384058,
      "grad_norm": 0.5006917119026184,
      "learning_rate": 6.9336443106947805e-06,
      "loss": 0.3966,
      "step": 25243
    },
    {
      "epoch": 1.1584599146436603,
      "grad_norm": 0.43102824687957764,
      "learning_rate": 6.9334181976843105e-06,
      "loss": 0.2692,
      "step": 25244
    },
    {
      "epoch": 1.1585058051489148,
      "grad_norm": 0.46626317501068115,
      "learning_rate": 6.933192080024579e-06,
      "loss": 0.3498,
      "step": 25245
    },
    {
      "epoch": 1.158551695654169,
      "grad_norm": 0.48062729835510254,
      "learning_rate": 6.932965957716131e-06,
      "loss": 0.4201,
      "step": 25246
    },
    {
      "epoch": 1.1585975861594235,
      "grad_norm": 0.48249563574790955,
      "learning_rate": 6.93273983075951e-06,
      "loss": 0.3944,
      "step": 25247
    },
    {
      "epoch": 1.158643476664678,
      "grad_norm": 0.49768292903900146,
      "learning_rate": 6.932513699155261e-06,
      "loss": 0.4218,
      "step": 25248
    },
    {
      "epoch": 1.1586893671699325,
      "grad_norm": 0.4939209520816803,
      "learning_rate": 6.932287562903926e-06,
      "loss": 0.4072,
      "step": 25249
    },
    {
      "epoch": 1.158735257675187,
      "grad_norm": 0.48353150486946106,
      "learning_rate": 6.932061422006049e-06,
      "loss": 0.3673,
      "step": 25250
    },
    {
      "epoch": 1.1587811481804415,
      "grad_norm": 0.5781121253967285,
      "learning_rate": 6.931835276462173e-06,
      "loss": 0.4524,
      "step": 25251
    },
    {
      "epoch": 1.158827038685696,
      "grad_norm": 0.4806913733482361,
      "learning_rate": 6.931609126272844e-06,
      "loss": 0.4372,
      "step": 25252
    },
    {
      "epoch": 1.1588729291909503,
      "grad_norm": 0.47583240270614624,
      "learning_rate": 6.931382971438605e-06,
      "loss": 0.325,
      "step": 25253
    },
    {
      "epoch": 1.1589188196962048,
      "grad_norm": 0.45305535197257996,
      "learning_rate": 6.931156811959999e-06,
      "loss": 0.3454,
      "step": 25254
    },
    {
      "epoch": 1.1589647102014593,
      "grad_norm": 0.468100905418396,
      "learning_rate": 6.930930647837571e-06,
      "loss": 0.4075,
      "step": 25255
    },
    {
      "epoch": 1.1590106007067138,
      "grad_norm": 0.4892304241657257,
      "learning_rate": 6.9307044790718625e-06,
      "loss": 0.4038,
      "step": 25256
    },
    {
      "epoch": 1.1590564912119683,
      "grad_norm": 0.4849899709224701,
      "learning_rate": 6.9304783056634194e-06,
      "loss": 0.4407,
      "step": 25257
    },
    {
      "epoch": 1.1591023817172228,
      "grad_norm": 0.46210673451423645,
      "learning_rate": 6.930252127612787e-06,
      "loss": 0.3722,
      "step": 25258
    },
    {
      "epoch": 1.159148272222477,
      "grad_norm": 0.5115764141082764,
      "learning_rate": 6.930025944920506e-06,
      "loss": 0.4995,
      "step": 25259
    },
    {
      "epoch": 1.1591941627277316,
      "grad_norm": 0.44128477573394775,
      "learning_rate": 6.9297997575871226e-06,
      "loss": 0.2958,
      "step": 25260
    },
    {
      "epoch": 1.159240053232986,
      "grad_norm": 0.4857756197452545,
      "learning_rate": 6.929573565613179e-06,
      "loss": 0.4311,
      "step": 25261
    },
    {
      "epoch": 1.1592859437382406,
      "grad_norm": 0.4294176995754242,
      "learning_rate": 6.929347368999221e-06,
      "loss": 0.2754,
      "step": 25262
    },
    {
      "epoch": 1.159331834243495,
      "grad_norm": 0.46627533435821533,
      "learning_rate": 6.929121167745789e-06,
      "loss": 0.3151,
      "step": 25263
    },
    {
      "epoch": 1.1593777247487496,
      "grad_norm": 0.477245956659317,
      "learning_rate": 6.928894961853433e-06,
      "loss": 0.3455,
      "step": 25264
    },
    {
      "epoch": 1.159423615254004,
      "grad_norm": 0.47511130571365356,
      "learning_rate": 6.928668751322692e-06,
      "loss": 0.3531,
      "step": 25265
    },
    {
      "epoch": 1.1594695057592583,
      "grad_norm": 0.4946886897087097,
      "learning_rate": 6.92844253615411e-06,
      "loss": 0.4103,
      "step": 25266
    },
    {
      "epoch": 1.1595153962645128,
      "grad_norm": 0.45668497681617737,
      "learning_rate": 6.928216316348234e-06,
      "loss": 0.3086,
      "step": 25267
    },
    {
      "epoch": 1.1595612867697673,
      "grad_norm": 0.5014060139656067,
      "learning_rate": 6.9279900919056054e-06,
      "loss": 0.4122,
      "step": 25268
    },
    {
      "epoch": 1.1596071772750218,
      "grad_norm": 0.47885796427726746,
      "learning_rate": 6.927763862826768e-06,
      "loss": 0.3637,
      "step": 25269
    },
    {
      "epoch": 1.1596530677802763,
      "grad_norm": 0.4194374084472656,
      "learning_rate": 6.92753762911227e-06,
      "loss": 0.2952,
      "step": 25270
    },
    {
      "epoch": 1.1596989582855306,
      "grad_norm": 0.4008837342262268,
      "learning_rate": 6.927311390762651e-06,
      "loss": 0.2415,
      "step": 25271
    },
    {
      "epoch": 1.159744848790785,
      "grad_norm": 0.5092943906784058,
      "learning_rate": 6.927085147778456e-06,
      "loss": 0.3767,
      "step": 25272
    },
    {
      "epoch": 1.1597907392960396,
      "grad_norm": 0.4472348392009735,
      "learning_rate": 6.926858900160229e-06,
      "loss": 0.2944,
      "step": 25273
    },
    {
      "epoch": 1.159836629801294,
      "grad_norm": 0.5028942227363586,
      "learning_rate": 6.926632647908516e-06,
      "loss": 0.4052,
      "step": 25274
    },
    {
      "epoch": 1.1598825203065486,
      "grad_norm": 0.434877872467041,
      "learning_rate": 6.926406391023858e-06,
      "loss": 0.3144,
      "step": 25275
    },
    {
      "epoch": 1.159928410811803,
      "grad_norm": 0.517604649066925,
      "learning_rate": 6.926180129506801e-06,
      "loss": 0.4569,
      "step": 25276
    },
    {
      "epoch": 1.1599743013170576,
      "grad_norm": 0.5182663202285767,
      "learning_rate": 6.925953863357888e-06,
      "loss": 0.4224,
      "step": 25277
    },
    {
      "epoch": 1.160020191822312,
      "grad_norm": 0.46685880422592163,
      "learning_rate": 6.925727592577666e-06,
      "loss": 0.347,
      "step": 25278
    },
    {
      "epoch": 1.1600660823275664,
      "grad_norm": 0.5070213675498962,
      "learning_rate": 6.925501317166675e-06,
      "loss": 0.4703,
      "step": 25279
    },
    {
      "epoch": 1.1601119728328209,
      "grad_norm": 0.45030733942985535,
      "learning_rate": 6.92527503712546e-06,
      "loss": 0.3131,
      "step": 25280
    },
    {
      "epoch": 1.1601578633380754,
      "grad_norm": 0.46435922384262085,
      "learning_rate": 6.925048752454569e-06,
      "loss": 0.3737,
      "step": 25281
    },
    {
      "epoch": 1.1602037538433299,
      "grad_norm": 0.4952225387096405,
      "learning_rate": 6.924822463154544e-06,
      "loss": 0.4117,
      "step": 25282
    },
    {
      "epoch": 1.1602496443485844,
      "grad_norm": 0.4554086923599243,
      "learning_rate": 6.924596169225925e-06,
      "loss": 0.3335,
      "step": 25283
    },
    {
      "epoch": 1.1602955348538386,
      "grad_norm": 0.514150083065033,
      "learning_rate": 6.924369870669262e-06,
      "loss": 0.3974,
      "step": 25284
    },
    {
      "epoch": 1.1603414253590931,
      "grad_norm": 0.4414784610271454,
      "learning_rate": 6.9241435674850954e-06,
      "loss": 0.339,
      "step": 25285
    },
    {
      "epoch": 1.1603873158643476,
      "grad_norm": 0.458553671836853,
      "learning_rate": 6.923917259673971e-06,
      "loss": 0.3528,
      "step": 25286
    },
    {
      "epoch": 1.1604332063696021,
      "grad_norm": 0.5030497908592224,
      "learning_rate": 6.923690947236435e-06,
      "loss": 0.3953,
      "step": 25287
    },
    {
      "epoch": 1.1604790968748566,
      "grad_norm": 0.46241697669029236,
      "learning_rate": 6.9234646301730265e-06,
      "loss": 0.3075,
      "step": 25288
    },
    {
      "epoch": 1.1605249873801111,
      "grad_norm": 0.47066304087638855,
      "learning_rate": 6.923238308484295e-06,
      "loss": 0.3541,
      "step": 25289
    },
    {
      "epoch": 1.1605708778853656,
      "grad_norm": 0.5196420550346375,
      "learning_rate": 6.92301198217078e-06,
      "loss": 0.444,
      "step": 25290
    },
    {
      "epoch": 1.16061676839062,
      "grad_norm": 0.46147480607032776,
      "learning_rate": 6.922785651233029e-06,
      "loss": 0.3269,
      "step": 25291
    },
    {
      "epoch": 1.1606626588958744,
      "grad_norm": 0.4913525879383087,
      "learning_rate": 6.922559315671586e-06,
      "loss": 0.3687,
      "step": 25292
    },
    {
      "epoch": 1.1607085494011289,
      "grad_norm": 0.497078001499176,
      "learning_rate": 6.922332975486994e-06,
      "loss": 0.3894,
      "step": 25293
    },
    {
      "epoch": 1.1607544399063834,
      "grad_norm": 0.5111665725708008,
      "learning_rate": 6.922106630679798e-06,
      "loss": 0.4513,
      "step": 25294
    },
    {
      "epoch": 1.1608003304116379,
      "grad_norm": 0.44512027502059937,
      "learning_rate": 6.921880281250543e-06,
      "loss": 0.3359,
      "step": 25295
    },
    {
      "epoch": 1.1608462209168924,
      "grad_norm": 0.46477943658828735,
      "learning_rate": 6.921653927199771e-06,
      "loss": 0.3589,
      "step": 25296
    },
    {
      "epoch": 1.1608921114221467,
      "grad_norm": 0.4413382411003113,
      "learning_rate": 6.921427568528029e-06,
      "loss": 0.3229,
      "step": 25297
    },
    {
      "epoch": 1.1609380019274012,
      "grad_norm": 0.432475209236145,
      "learning_rate": 6.9212012052358605e-06,
      "loss": 0.3076,
      "step": 25298
    },
    {
      "epoch": 1.1609838924326557,
      "grad_norm": 0.4439021050930023,
      "learning_rate": 6.9209748373238076e-06,
      "loss": 0.2889,
      "step": 25299
    },
    {
      "epoch": 1.1610297829379101,
      "grad_norm": 0.5060281753540039,
      "learning_rate": 6.920748464792417e-06,
      "loss": 0.4224,
      "step": 25300
    },
    {
      "epoch": 1.1610756734431646,
      "grad_norm": 0.46297165751457214,
      "learning_rate": 6.920522087642234e-06,
      "loss": 0.3559,
      "step": 25301
    },
    {
      "epoch": 1.1611215639484191,
      "grad_norm": 0.45897209644317627,
      "learning_rate": 6.9202957058738e-06,
      "loss": 0.2871,
      "step": 25302
    },
    {
      "epoch": 1.1611674544536736,
      "grad_norm": 0.44980618357658386,
      "learning_rate": 6.920069319487661e-06,
      "loss": 0.3623,
      "step": 25303
    },
    {
      "epoch": 1.161213344958928,
      "grad_norm": 0.46277329325675964,
      "learning_rate": 6.919842928484362e-06,
      "loss": 0.3662,
      "step": 25304
    },
    {
      "epoch": 1.1612592354641824,
      "grad_norm": 0.4254900813102722,
      "learning_rate": 6.919616532864446e-06,
      "loss": 0.2728,
      "step": 25305
    },
    {
      "epoch": 1.161305125969437,
      "grad_norm": 0.49500778317451477,
      "learning_rate": 6.91939013262846e-06,
      "loss": 0.4009,
      "step": 25306
    },
    {
      "epoch": 1.1613510164746914,
      "grad_norm": 0.48209160566329956,
      "learning_rate": 6.919163727776944e-06,
      "loss": 0.3866,
      "step": 25307
    },
    {
      "epoch": 1.161396906979946,
      "grad_norm": 0.44400402903556824,
      "learning_rate": 6.918937318310447e-06,
      "loss": 0.3085,
      "step": 25308
    },
    {
      "epoch": 1.1614427974852004,
      "grad_norm": 0.4763474762439728,
      "learning_rate": 6.9187109042295104e-06,
      "loss": 0.3417,
      "step": 25309
    },
    {
      "epoch": 1.1614886879904547,
      "grad_norm": 0.4443283677101135,
      "learning_rate": 6.918484485534679e-06,
      "loss": 0.298,
      "step": 25310
    },
    {
      "epoch": 1.1615345784957092,
      "grad_norm": 0.45248904824256897,
      "learning_rate": 6.918258062226498e-06,
      "loss": 0.3769,
      "step": 25311
    },
    {
      "epoch": 1.1615804690009637,
      "grad_norm": 0.4916403293609619,
      "learning_rate": 6.918031634305513e-06,
      "loss": 0.3548,
      "step": 25312
    },
    {
      "epoch": 1.1616263595062182,
      "grad_norm": 0.4852418005466461,
      "learning_rate": 6.917805201772265e-06,
      "loss": 0.3904,
      "step": 25313
    },
    {
      "epoch": 1.1616722500114727,
      "grad_norm": 0.4769500195980072,
      "learning_rate": 6.917578764627304e-06,
      "loss": 0.3941,
      "step": 25314
    },
    {
      "epoch": 1.1617181405167272,
      "grad_norm": 0.48406538367271423,
      "learning_rate": 6.917352322871169e-06,
      "loss": 0.3713,
      "step": 25315
    },
    {
      "epoch": 1.1617640310219817,
      "grad_norm": 0.44756704568862915,
      "learning_rate": 6.917125876504407e-06,
      "loss": 0.3288,
      "step": 25316
    },
    {
      "epoch": 1.161809921527236,
      "grad_norm": 0.5144450664520264,
      "learning_rate": 6.916899425527562e-06,
      "loss": 0.4499,
      "step": 25317
    },
    {
      "epoch": 1.1618558120324904,
      "grad_norm": 0.4784705340862274,
      "learning_rate": 6.9166729699411805e-06,
      "loss": 0.3915,
      "step": 25318
    },
    {
      "epoch": 1.161901702537745,
      "grad_norm": 0.45593640208244324,
      "learning_rate": 6.9164465097458045e-06,
      "loss": 0.3221,
      "step": 25319
    },
    {
      "epoch": 1.1619475930429994,
      "grad_norm": 0.4609876275062561,
      "learning_rate": 6.916220044941979e-06,
      "loss": 0.3682,
      "step": 25320
    },
    {
      "epoch": 1.161993483548254,
      "grad_norm": 0.47072482109069824,
      "learning_rate": 6.915993575530251e-06,
      "loss": 0.382,
      "step": 25321
    },
    {
      "epoch": 1.1620393740535082,
      "grad_norm": 0.4588839113712311,
      "learning_rate": 6.9157671015111606e-06,
      "loss": 0.3661,
      "step": 25322
    },
    {
      "epoch": 1.1620852645587627,
      "grad_norm": 0.4348493814468384,
      "learning_rate": 6.915540622885257e-06,
      "loss": 0.2828,
      "step": 25323
    },
    {
      "epoch": 1.1621311550640172,
      "grad_norm": 0.43835312128067017,
      "learning_rate": 6.915314139653081e-06,
      "loss": 0.3583,
      "step": 25324
    },
    {
      "epoch": 1.1621770455692717,
      "grad_norm": 0.4591067135334015,
      "learning_rate": 6.91508765181518e-06,
      "loss": 0.363,
      "step": 25325
    },
    {
      "epoch": 1.1622229360745262,
      "grad_norm": 0.4887416660785675,
      "learning_rate": 6.914861159372098e-06,
      "loss": 0.4255,
      "step": 25326
    },
    {
      "epoch": 1.1622688265797807,
      "grad_norm": 0.4880152940750122,
      "learning_rate": 6.9146346623243785e-06,
      "loss": 0.41,
      "step": 25327
    },
    {
      "epoch": 1.1623147170850352,
      "grad_norm": 0.4598153531551361,
      "learning_rate": 6.914408160672566e-06,
      "loss": 0.3955,
      "step": 25328
    },
    {
      "epoch": 1.1623606075902895,
      "grad_norm": 0.4725790321826935,
      "learning_rate": 6.914181654417208e-06,
      "loss": 0.3399,
      "step": 25329
    },
    {
      "epoch": 1.162406498095544,
      "grad_norm": 0.4594027101993561,
      "learning_rate": 6.913955143558847e-06,
      "loss": 0.388,
      "step": 25330
    },
    {
      "epoch": 1.1624523886007985,
      "grad_norm": 0.4589981734752655,
      "learning_rate": 6.913728628098027e-06,
      "loss": 0.3236,
      "step": 25331
    },
    {
      "epoch": 1.162498279106053,
      "grad_norm": 0.45130136609077454,
      "learning_rate": 6.913502108035295e-06,
      "loss": 0.3166,
      "step": 25332
    },
    {
      "epoch": 1.1625441696113075,
      "grad_norm": 0.5136481523513794,
      "learning_rate": 6.913275583371193e-06,
      "loss": 0.4793,
      "step": 25333
    },
    {
      "epoch": 1.162590060116562,
      "grad_norm": 0.4754766821861267,
      "learning_rate": 6.913049054106267e-06,
      "loss": 0.3851,
      "step": 25334
    },
    {
      "epoch": 1.1626359506218162,
      "grad_norm": 0.47055861353874207,
      "learning_rate": 6.912822520241063e-06,
      "loss": 0.338,
      "step": 25335
    },
    {
      "epoch": 1.1626818411270707,
      "grad_norm": 0.48005107045173645,
      "learning_rate": 6.912595981776124e-06,
      "loss": 0.3637,
      "step": 25336
    },
    {
      "epoch": 1.1627277316323252,
      "grad_norm": 0.4906352162361145,
      "learning_rate": 6.912369438711994e-06,
      "loss": 0.3415,
      "step": 25337
    },
    {
      "epoch": 1.1627736221375797,
      "grad_norm": 0.45833200216293335,
      "learning_rate": 6.91214289104922e-06,
      "loss": 0.3221,
      "step": 25338
    },
    {
      "epoch": 1.1628195126428342,
      "grad_norm": 0.4579266607761383,
      "learning_rate": 6.911916338788346e-06,
      "loss": 0.3306,
      "step": 25339
    },
    {
      "epoch": 1.1628654031480887,
      "grad_norm": 0.4561145305633545,
      "learning_rate": 6.911689781929916e-06,
      "loss": 0.3723,
      "step": 25340
    },
    {
      "epoch": 1.1629112936533432,
      "grad_norm": 0.4649745225906372,
      "learning_rate": 6.911463220474477e-06,
      "loss": 0.3454,
      "step": 25341
    },
    {
      "epoch": 1.1629571841585975,
      "grad_norm": 0.5232505798339844,
      "learning_rate": 6.9112366544225705e-06,
      "loss": 0.4065,
      "step": 25342
    },
    {
      "epoch": 1.163003074663852,
      "grad_norm": 0.4681083858013153,
      "learning_rate": 6.911010083774745e-06,
      "loss": 0.3952,
      "step": 25343
    },
    {
      "epoch": 1.1630489651691065,
      "grad_norm": 0.5087646245956421,
      "learning_rate": 6.910783508531542e-06,
      "loss": 0.4567,
      "step": 25344
    },
    {
      "epoch": 1.163094855674361,
      "grad_norm": 0.46805962920188904,
      "learning_rate": 6.910556928693507e-06,
      "loss": 0.3326,
      "step": 25345
    },
    {
      "epoch": 1.1631407461796155,
      "grad_norm": 0.4606439471244812,
      "learning_rate": 6.9103303442611865e-06,
      "loss": 0.3613,
      "step": 25346
    },
    {
      "epoch": 1.16318663668487,
      "grad_norm": 0.42853936553001404,
      "learning_rate": 6.910103755235125e-06,
      "loss": 0.2664,
      "step": 25347
    },
    {
      "epoch": 1.1632325271901243,
      "grad_norm": 0.4997710883617401,
      "learning_rate": 6.909877161615865e-06,
      "loss": 0.3501,
      "step": 25348
    },
    {
      "epoch": 1.1632784176953788,
      "grad_norm": 0.5081593990325928,
      "learning_rate": 6.9096505634039555e-06,
      "loss": 0.3867,
      "step": 25349
    },
    {
      "epoch": 1.1633243082006333,
      "grad_norm": 0.48145991563796997,
      "learning_rate": 6.909423960599938e-06,
      "loss": 0.4209,
      "step": 25350
    },
    {
      "epoch": 1.1633701987058878,
      "grad_norm": 0.5045231580734253,
      "learning_rate": 6.9091973532043584e-06,
      "loss": 0.4229,
      "step": 25351
    },
    {
      "epoch": 1.1634160892111423,
      "grad_norm": 0.48374122381210327,
      "learning_rate": 6.908970741217763e-06,
      "loss": 0.3364,
      "step": 25352
    },
    {
      "epoch": 1.1634619797163968,
      "grad_norm": 0.46208885312080383,
      "learning_rate": 6.908744124640694e-06,
      "loss": 0.3114,
      "step": 25353
    },
    {
      "epoch": 1.1635078702216513,
      "grad_norm": 0.49200767278671265,
      "learning_rate": 6.908517503473699e-06,
      "loss": 0.3584,
      "step": 25354
    },
    {
      "epoch": 1.1635537607269055,
      "grad_norm": 0.5330519080162048,
      "learning_rate": 6.908290877717319e-06,
      "loss": 0.4082,
      "step": 25355
    },
    {
      "epoch": 1.16359965123216,
      "grad_norm": 0.4337038993835449,
      "learning_rate": 6.908064247372104e-06,
      "loss": 0.3094,
      "step": 25356
    },
    {
      "epoch": 1.1636455417374145,
      "grad_norm": 0.4693990647792816,
      "learning_rate": 6.907837612438597e-06,
      "loss": 0.4038,
      "step": 25357
    },
    {
      "epoch": 1.163691432242669,
      "grad_norm": 0.4843118190765381,
      "learning_rate": 6.907610972917343e-06,
      "loss": 0.4056,
      "step": 25358
    },
    {
      "epoch": 1.1637373227479235,
      "grad_norm": 0.47962963581085205,
      "learning_rate": 6.9073843288088835e-06,
      "loss": 0.3479,
      "step": 25359
    },
    {
      "epoch": 1.1637832132531778,
      "grad_norm": 0.49168434739112854,
      "learning_rate": 6.9071576801137696e-06,
      "loss": 0.3839,
      "step": 25360
    },
    {
      "epoch": 1.1638291037584323,
      "grad_norm": 0.4488421678543091,
      "learning_rate": 6.9069310268325415e-06,
      "loss": 0.341,
      "step": 25361
    },
    {
      "epoch": 1.1638749942636868,
      "grad_norm": 0.5133140087127686,
      "learning_rate": 6.906704368965749e-06,
      "loss": 0.4209,
      "step": 25362
    },
    {
      "epoch": 1.1639208847689413,
      "grad_norm": 0.44302132725715637,
      "learning_rate": 6.906477706513932e-06,
      "loss": 0.3569,
      "step": 25363
    },
    {
      "epoch": 1.1639667752741958,
      "grad_norm": 0.46309328079223633,
      "learning_rate": 6.906251039477639e-06,
      "loss": 0.3753,
      "step": 25364
    },
    {
      "epoch": 1.1640126657794503,
      "grad_norm": 0.5443177223205566,
      "learning_rate": 6.906024367857411e-06,
      "loss": 0.4908,
      "step": 25365
    },
    {
      "epoch": 1.1640585562847048,
      "grad_norm": 0.5212720036506653,
      "learning_rate": 6.9057976916537995e-06,
      "loss": 0.4408,
      "step": 25366
    },
    {
      "epoch": 1.1641044467899593,
      "grad_norm": 0.4701431095600128,
      "learning_rate": 6.905571010867345e-06,
      "loss": 0.3286,
      "step": 25367
    },
    {
      "epoch": 1.1641503372952136,
      "grad_norm": 0.508514940738678,
      "learning_rate": 6.905344325498593e-06,
      "loss": 0.4153,
      "step": 25368
    },
    {
      "epoch": 1.164196227800468,
      "grad_norm": 0.46630316972732544,
      "learning_rate": 6.9051176355480895e-06,
      "loss": 0.3447,
      "step": 25369
    },
    {
      "epoch": 1.1642421183057226,
      "grad_norm": 0.4890194535255432,
      "learning_rate": 6.904890941016379e-06,
      "loss": 0.3874,
      "step": 25370
    },
    {
      "epoch": 1.164288008810977,
      "grad_norm": 0.4448188543319702,
      "learning_rate": 6.904664241904008e-06,
      "loss": 0.3095,
      "step": 25371
    },
    {
      "epoch": 1.1643338993162315,
      "grad_norm": 0.45271459221839905,
      "learning_rate": 6.904437538211519e-06,
      "loss": 0.3151,
      "step": 25372
    },
    {
      "epoch": 1.1643797898214858,
      "grad_norm": 0.48518118262290955,
      "learning_rate": 6.9042108299394604e-06,
      "loss": 0.3309,
      "step": 25373
    },
    {
      "epoch": 1.1644256803267403,
      "grad_norm": 0.47770512104034424,
      "learning_rate": 6.903984117088376e-06,
      "loss": 0.3442,
      "step": 25374
    },
    {
      "epoch": 1.1644715708319948,
      "grad_norm": 0.5800608992576599,
      "learning_rate": 6.9037573996588095e-06,
      "loss": 0.3897,
      "step": 25375
    },
    {
      "epoch": 1.1645174613372493,
      "grad_norm": 0.4707186222076416,
      "learning_rate": 6.903530677651307e-06,
      "loss": 0.342,
      "step": 25376
    },
    {
      "epoch": 1.1645633518425038,
      "grad_norm": 0.42263123393058777,
      "learning_rate": 6.903303951066416e-06,
      "loss": 0.3032,
      "step": 25377
    },
    {
      "epoch": 1.1646092423477583,
      "grad_norm": 0.4945104718208313,
      "learning_rate": 6.903077219904677e-06,
      "loss": 0.3525,
      "step": 25378
    },
    {
      "epoch": 1.1646551328530128,
      "grad_norm": 0.4968191087245941,
      "learning_rate": 6.90285048416664e-06,
      "loss": 0.3627,
      "step": 25379
    },
    {
      "epoch": 1.164701023358267,
      "grad_norm": 0.47359541058540344,
      "learning_rate": 6.902623743852848e-06,
      "loss": 0.3624,
      "step": 25380
    },
    {
      "epoch": 1.1647469138635216,
      "grad_norm": 0.4627218246459961,
      "learning_rate": 6.902396998963845e-06,
      "loss": 0.3408,
      "step": 25381
    },
    {
      "epoch": 1.164792804368776,
      "grad_norm": 0.46943625807762146,
      "learning_rate": 6.902170249500178e-06,
      "loss": 0.3878,
      "step": 25382
    },
    {
      "epoch": 1.1648386948740306,
      "grad_norm": 0.45839858055114746,
      "learning_rate": 6.901943495462392e-06,
      "loss": 0.3514,
      "step": 25383
    },
    {
      "epoch": 1.164884585379285,
      "grad_norm": 0.46193328499794006,
      "learning_rate": 6.901716736851033e-06,
      "loss": 0.3098,
      "step": 25384
    },
    {
      "epoch": 1.1649304758845396,
      "grad_norm": 0.4588625133037567,
      "learning_rate": 6.9014899736666444e-06,
      "loss": 0.3299,
      "step": 25385
    },
    {
      "epoch": 1.1649763663897938,
      "grad_norm": 0.45867034792900085,
      "learning_rate": 6.901263205909774e-06,
      "loss": 0.3076,
      "step": 25386
    },
    {
      "epoch": 1.1650222568950483,
      "grad_norm": 0.4793674051761627,
      "learning_rate": 6.901036433580965e-06,
      "loss": 0.3713,
      "step": 25387
    },
    {
      "epoch": 1.1650681474003028,
      "grad_norm": 0.4481409192085266,
      "learning_rate": 6.9008096566807624e-06,
      "loss": 0.3374,
      "step": 25388
    },
    {
      "epoch": 1.1651140379055573,
      "grad_norm": 0.506051242351532,
      "learning_rate": 6.900582875209713e-06,
      "loss": 0.4426,
      "step": 25389
    },
    {
      "epoch": 1.1651599284108118,
      "grad_norm": 0.43087759613990784,
      "learning_rate": 6.900356089168362e-06,
      "loss": 0.2899,
      "step": 25390
    },
    {
      "epoch": 1.1652058189160663,
      "grad_norm": 0.46279680728912354,
      "learning_rate": 6.900129298557254e-06,
      "loss": 0.3211,
      "step": 25391
    },
    {
      "epoch": 1.1652517094213208,
      "grad_norm": 0.4506016671657562,
      "learning_rate": 6.899902503376935e-06,
      "loss": 0.2686,
      "step": 25392
    },
    {
      "epoch": 1.1652975999265751,
      "grad_norm": 0.48673027753829956,
      "learning_rate": 6.899675703627949e-06,
      "loss": 0.439,
      "step": 25393
    },
    {
      "epoch": 1.1653434904318296,
      "grad_norm": 0.4714841842651367,
      "learning_rate": 6.899448899310844e-06,
      "loss": 0.3418,
      "step": 25394
    },
    {
      "epoch": 1.165389380937084,
      "grad_norm": 0.5234691500663757,
      "learning_rate": 6.8992220904261645e-06,
      "loss": 0.4416,
      "step": 25395
    },
    {
      "epoch": 1.1654352714423386,
      "grad_norm": 0.4563394784927368,
      "learning_rate": 6.8989952769744535e-06,
      "loss": 0.2863,
      "step": 25396
    },
    {
      "epoch": 1.165481161947593,
      "grad_norm": 0.453146368265152,
      "learning_rate": 6.898768458956258e-06,
      "loss": 0.283,
      "step": 25397
    },
    {
      "epoch": 1.1655270524528476,
      "grad_norm": 0.4867042303085327,
      "learning_rate": 6.898541636372124e-06,
      "loss": 0.3822,
      "step": 25398
    },
    {
      "epoch": 1.1655729429581019,
      "grad_norm": 0.49862805008888245,
      "learning_rate": 6.898314809222597e-06,
      "loss": 0.4114,
      "step": 25399
    },
    {
      "epoch": 1.1656188334633564,
      "grad_norm": 0.42709067463874817,
      "learning_rate": 6.898087977508223e-06,
      "loss": 0.3014,
      "step": 25400
    },
    {
      "epoch": 1.1656647239686109,
      "grad_norm": 0.4268510341644287,
      "learning_rate": 6.897861141229546e-06,
      "loss": 0.2818,
      "step": 25401
    },
    {
      "epoch": 1.1657106144738654,
      "grad_norm": 0.4229263365268707,
      "learning_rate": 6.897634300387111e-06,
      "loss": 0.2771,
      "step": 25402
    },
    {
      "epoch": 1.1657565049791199,
      "grad_norm": 0.4434099793434143,
      "learning_rate": 6.897407454981465e-06,
      "loss": 0.2923,
      "step": 25403
    },
    {
      "epoch": 1.1658023954843744,
      "grad_norm": 0.5200785398483276,
      "learning_rate": 6.8971806050131515e-06,
      "loss": 0.3991,
      "step": 25404
    },
    {
      "epoch": 1.1658482859896289,
      "grad_norm": 0.4354478418827057,
      "learning_rate": 6.89695375048272e-06,
      "loss": 0.2845,
      "step": 25405
    },
    {
      "epoch": 1.1658941764948831,
      "grad_norm": 0.4833483397960663,
      "learning_rate": 6.896726891390712e-06,
      "loss": 0.3548,
      "step": 25406
    },
    {
      "epoch": 1.1659400670001376,
      "grad_norm": 0.4884061813354492,
      "learning_rate": 6.896500027737674e-06,
      "loss": 0.3761,
      "step": 25407
    },
    {
      "epoch": 1.1659859575053921,
      "grad_norm": 0.5036185383796692,
      "learning_rate": 6.896273159524154e-06,
      "loss": 0.4164,
      "step": 25408
    },
    {
      "epoch": 1.1660318480106466,
      "grad_norm": 0.4726350009441376,
      "learning_rate": 6.896046286750692e-06,
      "loss": 0.3286,
      "step": 25409
    },
    {
      "epoch": 1.1660777385159011,
      "grad_norm": 0.4565579295158386,
      "learning_rate": 6.89581940941784e-06,
      "loss": 0.3271,
      "step": 25410
    },
    {
      "epoch": 1.1661236290211554,
      "grad_norm": 0.4399621784687042,
      "learning_rate": 6.895592527526141e-06,
      "loss": 0.3112,
      "step": 25411
    },
    {
      "epoch": 1.16616951952641,
      "grad_norm": 0.5009722709655762,
      "learning_rate": 6.895365641076139e-06,
      "loss": 0.3912,
      "step": 25412
    },
    {
      "epoch": 1.1662154100316644,
      "grad_norm": 0.46689409017562866,
      "learning_rate": 6.89513875006838e-06,
      "loss": 0.3777,
      "step": 25413
    },
    {
      "epoch": 1.166261300536919,
      "grad_norm": 0.4010829031467438,
      "learning_rate": 6.894911854503412e-06,
      "loss": 0.2553,
      "step": 25414
    },
    {
      "epoch": 1.1663071910421734,
      "grad_norm": 0.523944616317749,
      "learning_rate": 6.894684954381779e-06,
      "loss": 0.473,
      "step": 25415
    },
    {
      "epoch": 1.166353081547428,
      "grad_norm": 0.46361881494522095,
      "learning_rate": 6.8944580497040246e-06,
      "loss": 0.3912,
      "step": 25416
    },
    {
      "epoch": 1.1663989720526824,
      "grad_norm": 0.4904778003692627,
      "learning_rate": 6.8942311404706995e-06,
      "loss": 0.3859,
      "step": 25417
    },
    {
      "epoch": 1.1664448625579367,
      "grad_norm": 0.46722307801246643,
      "learning_rate": 6.894004226682344e-06,
      "loss": 0.3558,
      "step": 25418
    },
    {
      "epoch": 1.1664907530631912,
      "grad_norm": 0.4635440707206726,
      "learning_rate": 6.893777308339506e-06,
      "loss": 0.3694,
      "step": 25419
    },
    {
      "epoch": 1.1665366435684457,
      "grad_norm": 0.45665106177330017,
      "learning_rate": 6.893550385442733e-06,
      "loss": 0.3747,
      "step": 25420
    },
    {
      "epoch": 1.1665825340737002,
      "grad_norm": 0.4173048138618469,
      "learning_rate": 6.893323457992566e-06,
      "loss": 0.273,
      "step": 25421
    },
    {
      "epoch": 1.1666284245789547,
      "grad_norm": 0.45921358466148376,
      "learning_rate": 6.893096525989557e-06,
      "loss": 0.3133,
      "step": 25422
    },
    {
      "epoch": 1.1666743150842092,
      "grad_norm": 0.4952046573162079,
      "learning_rate": 6.892869589434248e-06,
      "loss": 0.4131,
      "step": 25423
    },
    {
      "epoch": 1.1667202055894634,
      "grad_norm": 0.4500468969345093,
      "learning_rate": 6.892642648327184e-06,
      "loss": 0.3685,
      "step": 25424
    },
    {
      "epoch": 1.166766096094718,
      "grad_norm": 0.4682065546512604,
      "learning_rate": 6.892415702668912e-06,
      "loss": 0.3819,
      "step": 25425
    },
    {
      "epoch": 1.1668119865999724,
      "grad_norm": 0.47604915499687195,
      "learning_rate": 6.892188752459976e-06,
      "loss": 0.3593,
      "step": 25426
    },
    {
      "epoch": 1.166857877105227,
      "grad_norm": 0.45105987787246704,
      "learning_rate": 6.8919617977009245e-06,
      "loss": 0.3524,
      "step": 25427
    },
    {
      "epoch": 1.1669037676104814,
      "grad_norm": 0.43191421031951904,
      "learning_rate": 6.891734838392303e-06,
      "loss": 0.2924,
      "step": 25428
    },
    {
      "epoch": 1.166949658115736,
      "grad_norm": 0.5031294226646423,
      "learning_rate": 6.891507874534655e-06,
      "loss": 0.3758,
      "step": 25429
    },
    {
      "epoch": 1.1669955486209904,
      "grad_norm": 0.5148729681968689,
      "learning_rate": 6.891280906128528e-06,
      "loss": 0.4353,
      "step": 25430
    },
    {
      "epoch": 1.1670414391262447,
      "grad_norm": 0.45822030305862427,
      "learning_rate": 6.891053933174467e-06,
      "loss": 0.3123,
      "step": 25431
    },
    {
      "epoch": 1.1670873296314992,
      "grad_norm": 0.48069366812705994,
      "learning_rate": 6.890826955673019e-06,
      "loss": 0.423,
      "step": 25432
    },
    {
      "epoch": 1.1671332201367537,
      "grad_norm": 0.43992629647254944,
      "learning_rate": 6.890599973624729e-06,
      "loss": 0.3065,
      "step": 25433
    },
    {
      "epoch": 1.1671791106420082,
      "grad_norm": 0.47621071338653564,
      "learning_rate": 6.890372987030141e-06,
      "loss": 0.3418,
      "step": 25434
    },
    {
      "epoch": 1.1672250011472627,
      "grad_norm": 0.4598264694213867,
      "learning_rate": 6.890145995889804e-06,
      "loss": 0.3401,
      "step": 25435
    },
    {
      "epoch": 1.1672708916525172,
      "grad_norm": 0.4815293550491333,
      "learning_rate": 6.889919000204263e-06,
      "loss": 0.3944,
      "step": 25436
    },
    {
      "epoch": 1.1673167821577715,
      "grad_norm": 0.4852719008922577,
      "learning_rate": 6.889691999974062e-06,
      "loss": 0.377,
      "step": 25437
    },
    {
      "epoch": 1.167362672663026,
      "grad_norm": 0.44850438833236694,
      "learning_rate": 6.889464995199747e-06,
      "loss": 0.3293,
      "step": 25438
    },
    {
      "epoch": 1.1674085631682805,
      "grad_norm": 0.490010529756546,
      "learning_rate": 6.889237985881868e-06,
      "loss": 0.3993,
      "step": 25439
    },
    {
      "epoch": 1.167454453673535,
      "grad_norm": 0.4726997911930084,
      "learning_rate": 6.889010972020967e-06,
      "loss": 0.3986,
      "step": 25440
    },
    {
      "epoch": 1.1675003441787895,
      "grad_norm": 0.532572865486145,
      "learning_rate": 6.888783953617588e-06,
      "loss": 0.359,
      "step": 25441
    },
    {
      "epoch": 1.167546234684044,
      "grad_norm": 0.4604973793029785,
      "learning_rate": 6.888556930672283e-06,
      "loss": 0.3485,
      "step": 25442
    },
    {
      "epoch": 1.1675921251892984,
      "grad_norm": 0.4495506286621094,
      "learning_rate": 6.888329903185593e-06,
      "loss": 0.2871,
      "step": 25443
    },
    {
      "epoch": 1.1676380156945527,
      "grad_norm": 0.45247775316238403,
      "learning_rate": 6.888102871158067e-06,
      "loss": 0.3337,
      "step": 25444
    },
    {
      "epoch": 1.1676839061998072,
      "grad_norm": 0.4888607859611511,
      "learning_rate": 6.887875834590249e-06,
      "loss": 0.3697,
      "step": 25445
    },
    {
      "epoch": 1.1677297967050617,
      "grad_norm": 0.46442824602127075,
      "learning_rate": 6.8876487934826845e-06,
      "loss": 0.3537,
      "step": 25446
    },
    {
      "epoch": 1.1677756872103162,
      "grad_norm": 0.44359052181243896,
      "learning_rate": 6.88742174783592e-06,
      "loss": 0.2998,
      "step": 25447
    },
    {
      "epoch": 1.1678215777155707,
      "grad_norm": 0.47937047481536865,
      "learning_rate": 6.887194697650503e-06,
      "loss": 0.3772,
      "step": 25448
    },
    {
      "epoch": 1.167867468220825,
      "grad_norm": 0.4643213450908661,
      "learning_rate": 6.886967642926978e-06,
      "loss": 0.347,
      "step": 25449
    },
    {
      "epoch": 1.1679133587260795,
      "grad_norm": 0.48892465233802795,
      "learning_rate": 6.8867405836658916e-06,
      "loss": 0.3745,
      "step": 25450
    },
    {
      "epoch": 1.167959249231334,
      "grad_norm": 0.4457699954509735,
      "learning_rate": 6.88651351986779e-06,
      "loss": 0.3598,
      "step": 25451
    },
    {
      "epoch": 1.1680051397365885,
      "grad_norm": 0.44912588596343994,
      "learning_rate": 6.886286451533219e-06,
      "loss": 0.3116,
      "step": 25452
    },
    {
      "epoch": 1.168051030241843,
      "grad_norm": 0.4273374676704407,
      "learning_rate": 6.8860593786627235e-06,
      "loss": 0.3425,
      "step": 25453
    },
    {
      "epoch": 1.1680969207470975,
      "grad_norm": 0.49525442719459534,
      "learning_rate": 6.8858323012568496e-06,
      "loss": 0.4119,
      "step": 25454
    },
    {
      "epoch": 1.168142811252352,
      "grad_norm": 0.463507741689682,
      "learning_rate": 6.885605219316146e-06,
      "loss": 0.4111,
      "step": 25455
    },
    {
      "epoch": 1.1681887017576065,
      "grad_norm": 0.5294719934463501,
      "learning_rate": 6.885378132841157e-06,
      "loss": 0.4825,
      "step": 25456
    },
    {
      "epoch": 1.1682345922628607,
      "grad_norm": 0.45793911814689636,
      "learning_rate": 6.885151041832427e-06,
      "loss": 0.3503,
      "step": 25457
    },
    {
      "epoch": 1.1682804827681152,
      "grad_norm": 0.4352853000164032,
      "learning_rate": 6.8849239462905045e-06,
      "loss": 0.3153,
      "step": 25458
    },
    {
      "epoch": 1.1683263732733697,
      "grad_norm": 0.47256094217300415,
      "learning_rate": 6.884696846215935e-06,
      "loss": 0.4032,
      "step": 25459
    },
    {
      "epoch": 1.1683722637786242,
      "grad_norm": 0.4533510208129883,
      "learning_rate": 6.884469741609264e-06,
      "loss": 0.3164,
      "step": 25460
    },
    {
      "epoch": 1.1684181542838787,
      "grad_norm": 0.45984402298927307,
      "learning_rate": 6.884242632471039e-06,
      "loss": 0.3677,
      "step": 25461
    },
    {
      "epoch": 1.168464044789133,
      "grad_norm": 0.5017122030258179,
      "learning_rate": 6.884015518801805e-06,
      "loss": 0.387,
      "step": 25462
    },
    {
      "epoch": 1.1685099352943875,
      "grad_norm": 0.42840832471847534,
      "learning_rate": 6.883788400602107e-06,
      "loss": 0.2877,
      "step": 25463
    },
    {
      "epoch": 1.168555825799642,
      "grad_norm": 0.4503520131111145,
      "learning_rate": 6.883561277872493e-06,
      "loss": 0.3155,
      "step": 25464
    },
    {
      "epoch": 1.1686017163048965,
      "grad_norm": 0.4706966280937195,
      "learning_rate": 6.883334150613509e-06,
      "loss": 0.3575,
      "step": 25465
    },
    {
      "epoch": 1.168647606810151,
      "grad_norm": 0.44303762912750244,
      "learning_rate": 6.883107018825701e-06,
      "loss": 0.3313,
      "step": 25466
    },
    {
      "epoch": 1.1686934973154055,
      "grad_norm": 0.4703931510448456,
      "learning_rate": 6.882879882509614e-06,
      "loss": 0.3647,
      "step": 25467
    },
    {
      "epoch": 1.16873938782066,
      "grad_norm": 0.4338523745536804,
      "learning_rate": 6.8826527416657964e-06,
      "loss": 0.3085,
      "step": 25468
    },
    {
      "epoch": 1.1687852783259143,
      "grad_norm": 0.46583548188209534,
      "learning_rate": 6.882425596294792e-06,
      "loss": 0.3921,
      "step": 25469
    },
    {
      "epoch": 1.1688311688311688,
      "grad_norm": 0.4471648931503296,
      "learning_rate": 6.882198446397148e-06,
      "loss": 0.3341,
      "step": 25470
    },
    {
      "epoch": 1.1688770593364233,
      "grad_norm": 0.4528176784515381,
      "learning_rate": 6.881971291973412e-06,
      "loss": 0.3289,
      "step": 25471
    },
    {
      "epoch": 1.1689229498416778,
      "grad_norm": 0.46092653274536133,
      "learning_rate": 6.881744133024128e-06,
      "loss": 0.3467,
      "step": 25472
    },
    {
      "epoch": 1.1689688403469323,
      "grad_norm": 0.48597103357315063,
      "learning_rate": 6.881516969549844e-06,
      "loss": 0.4126,
      "step": 25473
    },
    {
      "epoch": 1.1690147308521868,
      "grad_norm": 0.47000548243522644,
      "learning_rate": 6.881289801551105e-06,
      "loss": 0.3514,
      "step": 25474
    },
    {
      "epoch": 1.169060621357441,
      "grad_norm": 0.4788336157798767,
      "learning_rate": 6.881062629028458e-06,
      "loss": 0.3649,
      "step": 25475
    },
    {
      "epoch": 1.1691065118626955,
      "grad_norm": 0.4651496410369873,
      "learning_rate": 6.880835451982448e-06,
      "loss": 0.3849,
      "step": 25476
    },
    {
      "epoch": 1.16915240236795,
      "grad_norm": 0.4732339084148407,
      "learning_rate": 6.880608270413624e-06,
      "loss": 0.3713,
      "step": 25477
    },
    {
      "epoch": 1.1691982928732045,
      "grad_norm": 0.4777868390083313,
      "learning_rate": 6.88038108432253e-06,
      "loss": 0.3983,
      "step": 25478
    },
    {
      "epoch": 1.169244183378459,
      "grad_norm": 0.4273369014263153,
      "learning_rate": 6.880153893709714e-06,
      "loss": 0.2879,
      "step": 25479
    },
    {
      "epoch": 1.1692900738837135,
      "grad_norm": 0.4664499759674072,
      "learning_rate": 6.879926698575718e-06,
      "loss": 0.3304,
      "step": 25480
    },
    {
      "epoch": 1.169335964388968,
      "grad_norm": 0.47480228543281555,
      "learning_rate": 6.879699498921094e-06,
      "loss": 0.3704,
      "step": 25481
    },
    {
      "epoch": 1.1693818548942223,
      "grad_norm": 0.4617302715778351,
      "learning_rate": 6.879472294746386e-06,
      "loss": 0.3677,
      "step": 25482
    },
    {
      "epoch": 1.1694277453994768,
      "grad_norm": 0.4861379563808441,
      "learning_rate": 6.879245086052141e-06,
      "loss": 0.3729,
      "step": 25483
    },
    {
      "epoch": 1.1694736359047313,
      "grad_norm": 0.4335375726222992,
      "learning_rate": 6.879017872838904e-06,
      "loss": 0.3147,
      "step": 25484
    },
    {
      "epoch": 1.1695195264099858,
      "grad_norm": 0.4581329822540283,
      "learning_rate": 6.8787906551072214e-06,
      "loss": 0.3758,
      "step": 25485
    },
    {
      "epoch": 1.1695654169152403,
      "grad_norm": 0.4525098204612732,
      "learning_rate": 6.87856343285764e-06,
      "loss": 0.3256,
      "step": 25486
    },
    {
      "epoch": 1.1696113074204948,
      "grad_norm": 0.4844927191734314,
      "learning_rate": 6.878336206090708e-06,
      "loss": 0.3723,
      "step": 25487
    },
    {
      "epoch": 1.169657197925749,
      "grad_norm": 0.48377251625061035,
      "learning_rate": 6.87810897480697e-06,
      "loss": 0.3942,
      "step": 25488
    },
    {
      "epoch": 1.1697030884310036,
      "grad_norm": 0.496649831533432,
      "learning_rate": 6.877881739006971e-06,
      "loss": 0.437,
      "step": 25489
    },
    {
      "epoch": 1.169748978936258,
      "grad_norm": 0.4764748513698578,
      "learning_rate": 6.8776544986912595e-06,
      "loss": 0.4128,
      "step": 25490
    },
    {
      "epoch": 1.1697948694415126,
      "grad_norm": 0.45460137724876404,
      "learning_rate": 6.877427253860382e-06,
      "loss": 0.3368,
      "step": 25491
    },
    {
      "epoch": 1.169840759946767,
      "grad_norm": 0.49834832549095154,
      "learning_rate": 6.877200004514885e-06,
      "loss": 0.3702,
      "step": 25492
    },
    {
      "epoch": 1.1698866504520216,
      "grad_norm": 0.49267178773880005,
      "learning_rate": 6.876972750655314e-06,
      "loss": 0.4545,
      "step": 25493
    },
    {
      "epoch": 1.169932540957276,
      "grad_norm": 0.467572420835495,
      "learning_rate": 6.876745492282216e-06,
      "loss": 0.3415,
      "step": 25494
    },
    {
      "epoch": 1.1699784314625303,
      "grad_norm": 0.40683358907699585,
      "learning_rate": 6.876518229396137e-06,
      "loss": 0.2571,
      "step": 25495
    },
    {
      "epoch": 1.1700243219677848,
      "grad_norm": 0.5418713092803955,
      "learning_rate": 6.876290961997625e-06,
      "loss": 0.5379,
      "step": 25496
    },
    {
      "epoch": 1.1700702124730393,
      "grad_norm": 0.4679352343082428,
      "learning_rate": 6.876063690087224e-06,
      "loss": 0.3703,
      "step": 25497
    },
    {
      "epoch": 1.1701161029782938,
      "grad_norm": 0.49951857328414917,
      "learning_rate": 6.8758364136654834e-06,
      "loss": 0.3856,
      "step": 25498
    },
    {
      "epoch": 1.1701619934835483,
      "grad_norm": 0.46701744198799133,
      "learning_rate": 6.875609132732947e-06,
      "loss": 0.3543,
      "step": 25499
    },
    {
      "epoch": 1.1702078839888026,
      "grad_norm": 0.4533466398715973,
      "learning_rate": 6.875381847290163e-06,
      "loss": 0.3401,
      "step": 25500
    },
    {
      "epoch": 1.170253774494057,
      "grad_norm": 0.47816556692123413,
      "learning_rate": 6.875154557337677e-06,
      "loss": 0.3493,
      "step": 25501
    },
    {
      "epoch": 1.1702996649993116,
      "grad_norm": 0.4744378328323364,
      "learning_rate": 6.874927262876038e-06,
      "loss": 0.3789,
      "step": 25502
    },
    {
      "epoch": 1.170345555504566,
      "grad_norm": 0.4572046995162964,
      "learning_rate": 6.874699963905789e-06,
      "loss": 0.3496,
      "step": 25503
    },
    {
      "epoch": 1.1703914460098206,
      "grad_norm": 0.5027403235435486,
      "learning_rate": 6.874472660427478e-06,
      "loss": 0.3166,
      "step": 25504
    },
    {
      "epoch": 1.170437336515075,
      "grad_norm": 0.47575706243515015,
      "learning_rate": 6.874245352441653e-06,
      "loss": 0.3914,
      "step": 25505
    },
    {
      "epoch": 1.1704832270203296,
      "grad_norm": 0.4718310832977295,
      "learning_rate": 6.874018039948859e-06,
      "loss": 0.3845,
      "step": 25506
    },
    {
      "epoch": 1.1705291175255839,
      "grad_norm": 0.4763953983783722,
      "learning_rate": 6.873790722949643e-06,
      "loss": 0.3839,
      "step": 25507
    },
    {
      "epoch": 1.1705750080308384,
      "grad_norm": 0.4770369529724121,
      "learning_rate": 6.873563401444552e-06,
      "loss": 0.3491,
      "step": 25508
    },
    {
      "epoch": 1.1706208985360929,
      "grad_norm": 0.47129103541374207,
      "learning_rate": 6.873336075434132e-06,
      "loss": 0.3365,
      "step": 25509
    },
    {
      "epoch": 1.1706667890413474,
      "grad_norm": 0.44112104177474976,
      "learning_rate": 6.873108744918931e-06,
      "loss": 0.2996,
      "step": 25510
    },
    {
      "epoch": 1.1707126795466019,
      "grad_norm": 0.45545729994773865,
      "learning_rate": 6.872881409899492e-06,
      "loss": 0.2885,
      "step": 25511
    },
    {
      "epoch": 1.1707585700518564,
      "grad_norm": 0.48032909631729126,
      "learning_rate": 6.8726540703763654e-06,
      "loss": 0.3838,
      "step": 25512
    },
    {
      "epoch": 1.1708044605571106,
      "grad_norm": 0.4595769941806793,
      "learning_rate": 6.872426726350098e-06,
      "loss": 0.3441,
      "step": 25513
    },
    {
      "epoch": 1.1708503510623651,
      "grad_norm": 0.5449520349502563,
      "learning_rate": 6.872199377821235e-06,
      "loss": 0.3887,
      "step": 25514
    },
    {
      "epoch": 1.1708962415676196,
      "grad_norm": 0.4461211562156677,
      "learning_rate": 6.871972024790323e-06,
      "loss": 0.2934,
      "step": 25515
    },
    {
      "epoch": 1.1709421320728741,
      "grad_norm": 0.46130767464637756,
      "learning_rate": 6.871744667257909e-06,
      "loss": 0.3278,
      "step": 25516
    },
    {
      "epoch": 1.1709880225781286,
      "grad_norm": 0.4761263132095337,
      "learning_rate": 6.87151730522454e-06,
      "loss": 0.3662,
      "step": 25517
    },
    {
      "epoch": 1.1710339130833831,
      "grad_norm": 0.42998868227005005,
      "learning_rate": 6.8712899386907625e-06,
      "loss": 0.2858,
      "step": 25518
    },
    {
      "epoch": 1.1710798035886376,
      "grad_norm": 0.4436224102973938,
      "learning_rate": 6.871062567657123e-06,
      "loss": 0.2914,
      "step": 25519
    },
    {
      "epoch": 1.171125694093892,
      "grad_norm": 0.5102339386940002,
      "learning_rate": 6.87083519212417e-06,
      "loss": 0.4416,
      "step": 25520
    },
    {
      "epoch": 1.1711715845991464,
      "grad_norm": 0.5048193335533142,
      "learning_rate": 6.870607812092448e-06,
      "loss": 0.385,
      "step": 25521
    },
    {
      "epoch": 1.1712174751044009,
      "grad_norm": 0.45603710412979126,
      "learning_rate": 6.870380427562503e-06,
      "loss": 0.3494,
      "step": 25522
    },
    {
      "epoch": 1.1712633656096554,
      "grad_norm": 0.46542298793792725,
      "learning_rate": 6.870153038534884e-06,
      "loss": 0.3968,
      "step": 25523
    },
    {
      "epoch": 1.1713092561149099,
      "grad_norm": 0.4728873670101166,
      "learning_rate": 6.869925645010138e-06,
      "loss": 0.3807,
      "step": 25524
    },
    {
      "epoch": 1.1713551466201644,
      "grad_norm": 0.4072057604789734,
      "learning_rate": 6.8696982469888104e-06,
      "loss": 0.2677,
      "step": 25525
    },
    {
      "epoch": 1.1714010371254187,
      "grad_norm": 0.479343056678772,
      "learning_rate": 6.86947084447145e-06,
      "loss": 0.3752,
      "step": 25526
    },
    {
      "epoch": 1.1714469276306732,
      "grad_norm": 0.4866633713245392,
      "learning_rate": 6.8692434374586016e-06,
      "loss": 0.4162,
      "step": 25527
    },
    {
      "epoch": 1.1714928181359276,
      "grad_norm": 0.4854663014411926,
      "learning_rate": 6.869016025950812e-06,
      "loss": 0.3819,
      "step": 25528
    },
    {
      "epoch": 1.1715387086411821,
      "grad_norm": 0.44497814774513245,
      "learning_rate": 6.868788609948629e-06,
      "loss": 0.3335,
      "step": 25529
    },
    {
      "epoch": 1.1715845991464366,
      "grad_norm": 0.4877415597438812,
      "learning_rate": 6.8685611894526e-06,
      "loss": 0.3773,
      "step": 25530
    },
    {
      "epoch": 1.1716304896516911,
      "grad_norm": 0.4903859794139862,
      "learning_rate": 6.8683337644632695e-06,
      "loss": 0.3153,
      "step": 25531
    },
    {
      "epoch": 1.1716763801569456,
      "grad_norm": 0.4834446609020233,
      "learning_rate": 6.868106334981186e-06,
      "loss": 0.3352,
      "step": 25532
    },
    {
      "epoch": 1.1717222706622,
      "grad_norm": 0.4454856812953949,
      "learning_rate": 6.867878901006898e-06,
      "loss": 0.3213,
      "step": 25533
    },
    {
      "epoch": 1.1717681611674544,
      "grad_norm": 0.4969598948955536,
      "learning_rate": 6.86765146254095e-06,
      "loss": 0.3812,
      "step": 25534
    },
    {
      "epoch": 1.171814051672709,
      "grad_norm": 0.4507230222225189,
      "learning_rate": 6.867424019583889e-06,
      "loss": 0.3424,
      "step": 25535
    },
    {
      "epoch": 1.1718599421779634,
      "grad_norm": 0.4771530330181122,
      "learning_rate": 6.867196572136263e-06,
      "loss": 0.3642,
      "step": 25536
    },
    {
      "epoch": 1.171905832683218,
      "grad_norm": 0.46808308362960815,
      "learning_rate": 6.8669691201986185e-06,
      "loss": 0.3594,
      "step": 25537
    },
    {
      "epoch": 1.1719517231884722,
      "grad_norm": 0.4721062183380127,
      "learning_rate": 6.866741663771505e-06,
      "loss": 0.3545,
      "step": 25538
    },
    {
      "epoch": 1.1719976136937267,
      "grad_norm": 0.4789985120296478,
      "learning_rate": 6.866514202855463e-06,
      "loss": 0.3668,
      "step": 25539
    },
    {
      "epoch": 1.1720435041989812,
      "grad_norm": 0.457113116979599,
      "learning_rate": 6.8662867374510445e-06,
      "loss": 0.3594,
      "step": 25540
    },
    {
      "epoch": 1.1720893947042357,
      "grad_norm": 0.4772973954677582,
      "learning_rate": 6.866059267558797e-06,
      "loss": 0.3169,
      "step": 25541
    },
    {
      "epoch": 1.1721352852094902,
      "grad_norm": 0.46762266755104065,
      "learning_rate": 6.865831793179265e-06,
      "loss": 0.3948,
      "step": 25542
    },
    {
      "epoch": 1.1721811757147447,
      "grad_norm": 0.4718179404735565,
      "learning_rate": 6.865604314312996e-06,
      "loss": 0.3292,
      "step": 25543
    },
    {
      "epoch": 1.1722270662199992,
      "grad_norm": 0.47368523478507996,
      "learning_rate": 6.865376830960538e-06,
      "loss": 0.3759,
      "step": 25544
    },
    {
      "epoch": 1.1722729567252537,
      "grad_norm": 0.4707144498825073,
      "learning_rate": 6.865149343122437e-06,
      "loss": 0.3933,
      "step": 25545
    },
    {
      "epoch": 1.172318847230508,
      "grad_norm": 0.47251012921333313,
      "learning_rate": 6.8649218507992395e-06,
      "loss": 0.3768,
      "step": 25546
    },
    {
      "epoch": 1.1723647377357624,
      "grad_norm": 0.44325485825538635,
      "learning_rate": 6.864694353991496e-06,
      "loss": 0.3352,
      "step": 25547
    },
    {
      "epoch": 1.172410628241017,
      "grad_norm": 0.4542827904224396,
      "learning_rate": 6.8644668526997506e-06,
      "loss": 0.341,
      "step": 25548
    },
    {
      "epoch": 1.1724565187462714,
      "grad_norm": 0.4626162350177765,
      "learning_rate": 6.86423934692455e-06,
      "loss": 0.3746,
      "step": 25549
    },
    {
      "epoch": 1.172502409251526,
      "grad_norm": 0.47882887721061707,
      "learning_rate": 6.864011836666442e-06,
      "loss": 0.3779,
      "step": 25550
    },
    {
      "epoch": 1.1725482997567802,
      "grad_norm": 0.47287145256996155,
      "learning_rate": 6.8637843219259735e-06,
      "loss": 0.3969,
      "step": 25551
    },
    {
      "epoch": 1.1725941902620347,
      "grad_norm": 0.4821719825267792,
      "learning_rate": 6.8635568027036926e-06,
      "loss": 0.3075,
      "step": 25552
    },
    {
      "epoch": 1.1726400807672892,
      "grad_norm": 0.42865365743637085,
      "learning_rate": 6.863329279000146e-06,
      "loss": 0.2649,
      "step": 25553
    },
    {
      "epoch": 1.1726859712725437,
      "grad_norm": 0.4683023989200592,
      "learning_rate": 6.86310175081588e-06,
      "loss": 0.3813,
      "step": 25554
    },
    {
      "epoch": 1.1727318617777982,
      "grad_norm": 0.5178685188293457,
      "learning_rate": 6.862874218151443e-06,
      "loss": 0.4604,
      "step": 25555
    },
    {
      "epoch": 1.1727777522830527,
      "grad_norm": 0.5211215615272522,
      "learning_rate": 6.8626466810073815e-06,
      "loss": 0.4368,
      "step": 25556
    },
    {
      "epoch": 1.1728236427883072,
      "grad_norm": 0.4917925000190735,
      "learning_rate": 6.862419139384241e-06,
      "loss": 0.3835,
      "step": 25557
    },
    {
      "epoch": 1.1728695332935615,
      "grad_norm": 0.456134557723999,
      "learning_rate": 6.862191593282572e-06,
      "loss": 0.324,
      "step": 25558
    },
    {
      "epoch": 1.172915423798816,
      "grad_norm": 0.4897567927837372,
      "learning_rate": 6.86196404270292e-06,
      "loss": 0.3592,
      "step": 25559
    },
    {
      "epoch": 1.1729613143040705,
      "grad_norm": 0.4345453977584839,
      "learning_rate": 6.86173648764583e-06,
      "loss": 0.2746,
      "step": 25560
    },
    {
      "epoch": 1.173007204809325,
      "grad_norm": 0.48432499170303345,
      "learning_rate": 6.861508928111853e-06,
      "loss": 0.3551,
      "step": 25561
    },
    {
      "epoch": 1.1730530953145795,
      "grad_norm": 0.44443199038505554,
      "learning_rate": 6.861281364101534e-06,
      "loss": 0.3203,
      "step": 25562
    },
    {
      "epoch": 1.173098985819834,
      "grad_norm": 0.4816974699497223,
      "learning_rate": 6.861053795615422e-06,
      "loss": 0.3677,
      "step": 25563
    },
    {
      "epoch": 1.1731448763250882,
      "grad_norm": 0.4888845384120941,
      "learning_rate": 6.860826222654062e-06,
      "loss": 0.383,
      "step": 25564
    },
    {
      "epoch": 1.1731907668303427,
      "grad_norm": 0.4559001922607422,
      "learning_rate": 6.860598645218002e-06,
      "loss": 0.3036,
      "step": 25565
    },
    {
      "epoch": 1.1732366573355972,
      "grad_norm": 0.5008277297019958,
      "learning_rate": 6.86037106330779e-06,
      "loss": 0.4219,
      "step": 25566
    },
    {
      "epoch": 1.1732825478408517,
      "grad_norm": 0.4746609926223755,
      "learning_rate": 6.860143476923973e-06,
      "loss": 0.3543,
      "step": 25567
    },
    {
      "epoch": 1.1733284383461062,
      "grad_norm": 0.46074244379997253,
      "learning_rate": 6.8599158860670964e-06,
      "loss": 0.3508,
      "step": 25568
    },
    {
      "epoch": 1.1733743288513607,
      "grad_norm": 0.4928177297115326,
      "learning_rate": 6.859688290737711e-06,
      "loss": 0.4059,
      "step": 25569
    },
    {
      "epoch": 1.1734202193566152,
      "grad_norm": 0.5102314949035645,
      "learning_rate": 6.8594606909363615e-06,
      "loss": 0.408,
      "step": 25570
    },
    {
      "epoch": 1.1734661098618695,
      "grad_norm": 0.46034544706344604,
      "learning_rate": 6.859233086663595e-06,
      "loss": 0.3238,
      "step": 25571
    },
    {
      "epoch": 1.173512000367124,
      "grad_norm": 0.4639015197753906,
      "learning_rate": 6.859005477919961e-06,
      "loss": 0.354,
      "step": 25572
    },
    {
      "epoch": 1.1735578908723785,
      "grad_norm": 0.48340776562690735,
      "learning_rate": 6.858777864706004e-06,
      "loss": 0.331,
      "step": 25573
    },
    {
      "epoch": 1.173603781377633,
      "grad_norm": 0.5085130333900452,
      "learning_rate": 6.858550247022276e-06,
      "loss": 0.3755,
      "step": 25574
    },
    {
      "epoch": 1.1736496718828875,
      "grad_norm": 0.489210307598114,
      "learning_rate": 6.858322624869318e-06,
      "loss": 0.345,
      "step": 25575
    },
    {
      "epoch": 1.173695562388142,
      "grad_norm": 0.4802702069282532,
      "learning_rate": 6.858094998247682e-06,
      "loss": 0.383,
      "step": 25576
    },
    {
      "epoch": 1.1737414528933963,
      "grad_norm": 0.4607854187488556,
      "learning_rate": 6.857867367157913e-06,
      "loss": 0.3596,
      "step": 25577
    },
    {
      "epoch": 1.1737873433986508,
      "grad_norm": 0.48864156007766724,
      "learning_rate": 6.85763973160056e-06,
      "loss": 0.3358,
      "step": 25578
    },
    {
      "epoch": 1.1738332339039053,
      "grad_norm": 0.47249406576156616,
      "learning_rate": 6.85741209157617e-06,
      "loss": 0.328,
      "step": 25579
    },
    {
      "epoch": 1.1738791244091598,
      "grad_norm": 0.46810007095336914,
      "learning_rate": 6.85718444708529e-06,
      "loss": 0.3267,
      "step": 25580
    },
    {
      "epoch": 1.1739250149144143,
      "grad_norm": 0.4852651059627533,
      "learning_rate": 6.856956798128468e-06,
      "loss": 0.3885,
      "step": 25581
    },
    {
      "epoch": 1.1739709054196688,
      "grad_norm": 0.49484983086586,
      "learning_rate": 6.85672914470625e-06,
      "loss": 0.321,
      "step": 25582
    },
    {
      "epoch": 1.1740167959249233,
      "grad_norm": 0.492034912109375,
      "learning_rate": 6.856501486819185e-06,
      "loss": 0.3744,
      "step": 25583
    },
    {
      "epoch": 1.1740626864301775,
      "grad_norm": 0.4525633156299591,
      "learning_rate": 6.8562738244678195e-06,
      "loss": 0.3681,
      "step": 25584
    },
    {
      "epoch": 1.174108576935432,
      "grad_norm": 0.5134267807006836,
      "learning_rate": 6.856046157652703e-06,
      "loss": 0.422,
      "step": 25585
    },
    {
      "epoch": 1.1741544674406865,
      "grad_norm": 0.4716026782989502,
      "learning_rate": 6.855818486374379e-06,
      "loss": 0.3903,
      "step": 25586
    },
    {
      "epoch": 1.174200357945941,
      "grad_norm": 0.46050697565078735,
      "learning_rate": 6.855590810633398e-06,
      "loss": 0.342,
      "step": 25587
    },
    {
      "epoch": 1.1742462484511955,
      "grad_norm": 0.42134371399879456,
      "learning_rate": 6.855363130430306e-06,
      "loss": 0.2799,
      "step": 25588
    },
    {
      "epoch": 1.1742921389564498,
      "grad_norm": 0.5046181082725525,
      "learning_rate": 6.855135445765653e-06,
      "loss": 0.449,
      "step": 25589
    },
    {
      "epoch": 1.1743380294617043,
      "grad_norm": 0.5020381808280945,
      "learning_rate": 6.8549077566399825e-06,
      "loss": 0.4496,
      "step": 25590
    },
    {
      "epoch": 1.1743839199669588,
      "grad_norm": 0.42247965931892395,
      "learning_rate": 6.854680063053847e-06,
      "loss": 0.2492,
      "step": 25591
    },
    {
      "epoch": 1.1744298104722133,
      "grad_norm": 0.47944438457489014,
      "learning_rate": 6.85445236500779e-06,
      "loss": 0.4218,
      "step": 25592
    },
    {
      "epoch": 1.1744757009774678,
      "grad_norm": 0.435881108045578,
      "learning_rate": 6.8542246625023586e-06,
      "loss": 0.3162,
      "step": 25593
    },
    {
      "epoch": 1.1745215914827223,
      "grad_norm": 0.4483477473258972,
      "learning_rate": 6.853996955538104e-06,
      "loss": 0.3175,
      "step": 25594
    },
    {
      "epoch": 1.1745674819879768,
      "grad_norm": 0.4456421732902527,
      "learning_rate": 6.853769244115571e-06,
      "loss": 0.3153,
      "step": 25595
    },
    {
      "epoch": 1.174613372493231,
      "grad_norm": 0.44892561435699463,
      "learning_rate": 6.85354152823531e-06,
      "loss": 0.3115,
      "step": 25596
    },
    {
      "epoch": 1.1746592629984856,
      "grad_norm": 0.4607338607311249,
      "learning_rate": 6.853313807897864e-06,
      "loss": 0.3251,
      "step": 25597
    },
    {
      "epoch": 1.17470515350374,
      "grad_norm": 0.46220046281814575,
      "learning_rate": 6.853086083103786e-06,
      "loss": 0.3717,
      "step": 25598
    },
    {
      "epoch": 1.1747510440089945,
      "grad_norm": 0.4513967037200928,
      "learning_rate": 6.852858353853619e-06,
      "loss": 0.3286,
      "step": 25599
    },
    {
      "epoch": 1.174796934514249,
      "grad_norm": 0.4569966197013855,
      "learning_rate": 6.852630620147912e-06,
      "loss": 0.3589,
      "step": 25600
    },
    {
      "epoch": 1.1748428250195035,
      "grad_norm": 0.45557907223701477,
      "learning_rate": 6.852402881987213e-06,
      "loss": 0.3084,
      "step": 25601
    },
    {
      "epoch": 1.1748887155247578,
      "grad_norm": 0.48335859179496765,
      "learning_rate": 6.852175139372072e-06,
      "loss": 0.3435,
      "step": 25602
    },
    {
      "epoch": 1.1749346060300123,
      "grad_norm": 0.45509496331214905,
      "learning_rate": 6.851947392303033e-06,
      "loss": 0.3279,
      "step": 25603
    },
    {
      "epoch": 1.1749804965352668,
      "grad_norm": 0.5008500218391418,
      "learning_rate": 6.851719640780644e-06,
      "loss": 0.428,
      "step": 25604
    },
    {
      "epoch": 1.1750263870405213,
      "grad_norm": 0.48473989963531494,
      "learning_rate": 6.851491884805455e-06,
      "loss": 0.3611,
      "step": 25605
    },
    {
      "epoch": 1.1750722775457758,
      "grad_norm": 0.4900658428668976,
      "learning_rate": 6.851264124378012e-06,
      "loss": 0.3597,
      "step": 25606
    },
    {
      "epoch": 1.1751181680510303,
      "grad_norm": 0.42883649468421936,
      "learning_rate": 6.851036359498863e-06,
      "loss": 0.2933,
      "step": 25607
    },
    {
      "epoch": 1.1751640585562848,
      "grad_norm": 0.4408652186393738,
      "learning_rate": 6.850808590168555e-06,
      "loss": 0.3056,
      "step": 25608
    },
    {
      "epoch": 1.175209949061539,
      "grad_norm": 0.45024874806404114,
      "learning_rate": 6.850580816387638e-06,
      "loss": 0.3378,
      "step": 25609
    },
    {
      "epoch": 1.1752558395667936,
      "grad_norm": 0.47296851873397827,
      "learning_rate": 6.850353038156656e-06,
      "loss": 0.3958,
      "step": 25610
    },
    {
      "epoch": 1.175301730072048,
      "grad_norm": 0.46378618478775024,
      "learning_rate": 6.85012525547616e-06,
      "loss": 0.3969,
      "step": 25611
    },
    {
      "epoch": 1.1753476205773026,
      "grad_norm": 0.48663464188575745,
      "learning_rate": 6.849897468346698e-06,
      "loss": 0.3766,
      "step": 25612
    },
    {
      "epoch": 1.175393511082557,
      "grad_norm": 0.4591962397098541,
      "learning_rate": 6.849669676768816e-06,
      "loss": 0.3668,
      "step": 25613
    },
    {
      "epoch": 1.1754394015878116,
      "grad_norm": 0.4636283814907074,
      "learning_rate": 6.849441880743061e-06,
      "loss": 0.3545,
      "step": 25614
    },
    {
      "epoch": 1.1754852920930658,
      "grad_norm": 0.4776391088962555,
      "learning_rate": 6.849214080269982e-06,
      "loss": 0.3636,
      "step": 25615
    },
    {
      "epoch": 1.1755311825983203,
      "grad_norm": 0.6632169485092163,
      "learning_rate": 6.8489862753501264e-06,
      "loss": 0.4116,
      "step": 25616
    },
    {
      "epoch": 1.1755770731035748,
      "grad_norm": 0.4847368597984314,
      "learning_rate": 6.848758465984045e-06,
      "loss": 0.3627,
      "step": 25617
    },
    {
      "epoch": 1.1756229636088293,
      "grad_norm": 0.5001359581947327,
      "learning_rate": 6.848530652172282e-06,
      "loss": 0.389,
      "step": 25618
    },
    {
      "epoch": 1.1756688541140838,
      "grad_norm": 0.545688807964325,
      "learning_rate": 6.848302833915385e-06,
      "loss": 0.4821,
      "step": 25619
    },
    {
      "epoch": 1.1757147446193383,
      "grad_norm": 0.4666125774383545,
      "learning_rate": 6.8480750112139036e-06,
      "loss": 0.3225,
      "step": 25620
    },
    {
      "epoch": 1.1757606351245928,
      "grad_norm": 0.4411241114139557,
      "learning_rate": 6.8478471840683835e-06,
      "loss": 0.3399,
      "step": 25621
    },
    {
      "epoch": 1.175806525629847,
      "grad_norm": 0.5585257411003113,
      "learning_rate": 6.847619352479375e-06,
      "loss": 0.4158,
      "step": 25622
    },
    {
      "epoch": 1.1758524161351016,
      "grad_norm": 0.48042866587638855,
      "learning_rate": 6.8473915164474265e-06,
      "loss": 0.3272,
      "step": 25623
    },
    {
      "epoch": 1.175898306640356,
      "grad_norm": 0.45462745428085327,
      "learning_rate": 6.847163675973084e-06,
      "loss": 0.3389,
      "step": 25624
    },
    {
      "epoch": 1.1759441971456106,
      "grad_norm": 0.46928679943084717,
      "learning_rate": 6.846935831056896e-06,
      "loss": 0.3435,
      "step": 25625
    },
    {
      "epoch": 1.175990087650865,
      "grad_norm": 0.4700981080532074,
      "learning_rate": 6.8467079816994096e-06,
      "loss": 0.3285,
      "step": 25626
    },
    {
      "epoch": 1.1760359781561194,
      "grad_norm": 0.5453307032585144,
      "learning_rate": 6.846480127901173e-06,
      "loss": 0.4557,
      "step": 25627
    },
    {
      "epoch": 1.1760818686613739,
      "grad_norm": 0.48562899231910706,
      "learning_rate": 6.846252269662734e-06,
      "loss": 0.3247,
      "step": 25628
    },
    {
      "epoch": 1.1761277591666284,
      "grad_norm": 0.6901135444641113,
      "learning_rate": 6.846024406984644e-06,
      "loss": 0.4127,
      "step": 25629
    },
    {
      "epoch": 1.1761736496718829,
      "grad_norm": 0.47133857011795044,
      "learning_rate": 6.845796539867446e-06,
      "loss": 0.3205,
      "step": 25630
    },
    {
      "epoch": 1.1762195401771374,
      "grad_norm": 0.5129086375236511,
      "learning_rate": 6.845568668311688e-06,
      "loss": 0.417,
      "step": 25631
    },
    {
      "epoch": 1.1762654306823919,
      "grad_norm": 0.46819639205932617,
      "learning_rate": 6.845340792317923e-06,
      "loss": 0.3525,
      "step": 25632
    },
    {
      "epoch": 1.1763113211876464,
      "grad_norm": 0.48720142245292664,
      "learning_rate": 6.845112911886693e-06,
      "loss": 0.3302,
      "step": 25633
    },
    {
      "epoch": 1.1763572116929009,
      "grad_norm": 0.45668333768844604,
      "learning_rate": 6.84488502701855e-06,
      "loss": 0.324,
      "step": 25634
    },
    {
      "epoch": 1.1764031021981551,
      "grad_norm": 0.4703284502029419,
      "learning_rate": 6.844657137714042e-06,
      "loss": 0.3449,
      "step": 25635
    },
    {
      "epoch": 1.1764489927034096,
      "grad_norm": 0.4563477039337158,
      "learning_rate": 6.844429243973714e-06,
      "loss": 0.3197,
      "step": 25636
    },
    {
      "epoch": 1.1764948832086641,
      "grad_norm": 0.48183709383010864,
      "learning_rate": 6.844201345798117e-06,
      "loss": 0.3322,
      "step": 25637
    },
    {
      "epoch": 1.1765407737139186,
      "grad_norm": 0.44904422760009766,
      "learning_rate": 6.843973443187797e-06,
      "loss": 0.3087,
      "step": 25638
    },
    {
      "epoch": 1.1765866642191731,
      "grad_norm": 0.4431930482387543,
      "learning_rate": 6.843745536143302e-06,
      "loss": 0.2934,
      "step": 25639
    },
    {
      "epoch": 1.1766325547244274,
      "grad_norm": 0.4855479598045349,
      "learning_rate": 6.843517624665184e-06,
      "loss": 0.3854,
      "step": 25640
    },
    {
      "epoch": 1.176678445229682,
      "grad_norm": 0.48847299814224243,
      "learning_rate": 6.843289708753985e-06,
      "loss": 0.367,
      "step": 25641
    },
    {
      "epoch": 1.1767243357349364,
      "grad_norm": 0.4562525451183319,
      "learning_rate": 6.843061788410257e-06,
      "loss": 0.3174,
      "step": 25642
    },
    {
      "epoch": 1.176770226240191,
      "grad_norm": 0.4768557846546173,
      "learning_rate": 6.842833863634547e-06,
      "loss": 0.3949,
      "step": 25643
    },
    {
      "epoch": 1.1768161167454454,
      "grad_norm": 0.5186331272125244,
      "learning_rate": 6.842605934427402e-06,
      "loss": 0.4444,
      "step": 25644
    },
    {
      "epoch": 1.1768620072507,
      "grad_norm": 0.4745580554008484,
      "learning_rate": 6.8423780007893734e-06,
      "loss": 0.4202,
      "step": 25645
    },
    {
      "epoch": 1.1769078977559544,
      "grad_norm": 0.48131880164146423,
      "learning_rate": 6.842150062721007e-06,
      "loss": 0.3853,
      "step": 25646
    },
    {
      "epoch": 1.1769537882612087,
      "grad_norm": 0.457661509513855,
      "learning_rate": 6.8419221202228496e-06,
      "loss": 0.3322,
      "step": 25647
    },
    {
      "epoch": 1.1769996787664632,
      "grad_norm": 0.4485005736351013,
      "learning_rate": 6.841694173295452e-06,
      "loss": 0.2969,
      "step": 25648
    },
    {
      "epoch": 1.1770455692717177,
      "grad_norm": 0.4739619195461273,
      "learning_rate": 6.84146622193936e-06,
      "loss": 0.3797,
      "step": 25649
    },
    {
      "epoch": 1.1770914597769722,
      "grad_norm": 0.4598637819290161,
      "learning_rate": 6.8412382661551234e-06,
      "loss": 0.3463,
      "step": 25650
    },
    {
      "epoch": 1.1771373502822267,
      "grad_norm": 0.4538874924182892,
      "learning_rate": 6.841010305943292e-06,
      "loss": 0.3219,
      "step": 25651
    },
    {
      "epoch": 1.1771832407874812,
      "grad_norm": 0.4760102331638336,
      "learning_rate": 6.840782341304408e-06,
      "loss": 0.3449,
      "step": 25652
    },
    {
      "epoch": 1.1772291312927354,
      "grad_norm": 0.4779253900051117,
      "learning_rate": 6.840554372239025e-06,
      "loss": 0.3958,
      "step": 25653
    },
    {
      "epoch": 1.17727502179799,
      "grad_norm": 0.49629145860671997,
      "learning_rate": 6.840326398747691e-06,
      "loss": 0.384,
      "step": 25654
    },
    {
      "epoch": 1.1773209123032444,
      "grad_norm": 0.4855031669139862,
      "learning_rate": 6.840098420830952e-06,
      "loss": 0.3518,
      "step": 25655
    },
    {
      "epoch": 1.177366802808499,
      "grad_norm": 0.4801255464553833,
      "learning_rate": 6.839870438489356e-06,
      "loss": 0.3739,
      "step": 25656
    },
    {
      "epoch": 1.1774126933137534,
      "grad_norm": 0.45477619767189026,
      "learning_rate": 6.839642451723455e-06,
      "loss": 0.33,
      "step": 25657
    },
    {
      "epoch": 1.177458583819008,
      "grad_norm": 0.4763350784778595,
      "learning_rate": 6.839414460533792e-06,
      "loss": 0.3619,
      "step": 25658
    },
    {
      "epoch": 1.1775044743242624,
      "grad_norm": 0.46363896131515503,
      "learning_rate": 6.839186464920917e-06,
      "loss": 0.3178,
      "step": 25659
    },
    {
      "epoch": 1.1775503648295167,
      "grad_norm": 0.4309453070163727,
      "learning_rate": 6.838958464885382e-06,
      "loss": 0.2715,
      "step": 25660
    },
    {
      "epoch": 1.1775962553347712,
      "grad_norm": 0.45009729266166687,
      "learning_rate": 6.838730460427729e-06,
      "loss": 0.2688,
      "step": 25661
    },
    {
      "epoch": 1.1776421458400257,
      "grad_norm": 0.48578375577926636,
      "learning_rate": 6.838502451548512e-06,
      "loss": 0.3934,
      "step": 25662
    },
    {
      "epoch": 1.1776880363452802,
      "grad_norm": 0.4656767249107361,
      "learning_rate": 6.838274438248278e-06,
      "loss": 0.3317,
      "step": 25663
    },
    {
      "epoch": 1.1777339268505347,
      "grad_norm": 0.5049875974655151,
      "learning_rate": 6.838046420527571e-06,
      "loss": 0.3576,
      "step": 25664
    },
    {
      "epoch": 1.1777798173557892,
      "grad_norm": 0.45613908767700195,
      "learning_rate": 6.837818398386944e-06,
      "loss": 0.3219,
      "step": 25665
    },
    {
      "epoch": 1.1778257078610435,
      "grad_norm": 0.4677617847919464,
      "learning_rate": 6.837590371826943e-06,
      "loss": 0.3392,
      "step": 25666
    },
    {
      "epoch": 1.177871598366298,
      "grad_norm": 0.4620170593261719,
      "learning_rate": 6.837362340848119e-06,
      "loss": 0.323,
      "step": 25667
    },
    {
      "epoch": 1.1779174888715525,
      "grad_norm": 0.45046743750572205,
      "learning_rate": 6.8371343054510175e-06,
      "loss": 0.3301,
      "step": 25668
    },
    {
      "epoch": 1.177963379376807,
      "grad_norm": 0.5230370163917542,
      "learning_rate": 6.836906265636188e-06,
      "loss": 0.4577,
      "step": 25669
    },
    {
      "epoch": 1.1780092698820614,
      "grad_norm": 0.4484158754348755,
      "learning_rate": 6.836678221404176e-06,
      "loss": 0.2887,
      "step": 25670
    },
    {
      "epoch": 1.178055160387316,
      "grad_norm": 0.43846216797828674,
      "learning_rate": 6.836450172755535e-06,
      "loss": 0.3055,
      "step": 25671
    },
    {
      "epoch": 1.1781010508925704,
      "grad_norm": 0.47358906269073486,
      "learning_rate": 6.836222119690811e-06,
      "loss": 0.3138,
      "step": 25672
    },
    {
      "epoch": 1.1781469413978247,
      "grad_norm": 0.48473405838012695,
      "learning_rate": 6.835994062210552e-06,
      "loss": 0.3767,
      "step": 25673
    },
    {
      "epoch": 1.1781928319030792,
      "grad_norm": 0.47872573137283325,
      "learning_rate": 6.835766000315307e-06,
      "loss": 0.3694,
      "step": 25674
    },
    {
      "epoch": 1.1782387224083337,
      "grad_norm": 0.5203568339347839,
      "learning_rate": 6.835537934005624e-06,
      "loss": 0.4418,
      "step": 25675
    },
    {
      "epoch": 1.1782846129135882,
      "grad_norm": 0.6885707378387451,
      "learning_rate": 6.8353098632820504e-06,
      "loss": 0.3993,
      "step": 25676
    },
    {
      "epoch": 1.1783305034188427,
      "grad_norm": 0.4671296775341034,
      "learning_rate": 6.835081788145137e-06,
      "loss": 0.362,
      "step": 25677
    },
    {
      "epoch": 1.178376393924097,
      "grad_norm": 0.47926944494247437,
      "learning_rate": 6.834853708595433e-06,
      "loss": 0.4084,
      "step": 25678
    },
    {
      "epoch": 1.1784222844293515,
      "grad_norm": 0.47985947132110596,
      "learning_rate": 6.834625624633481e-06,
      "loss": 0.3555,
      "step": 25679
    },
    {
      "epoch": 1.178468174934606,
      "grad_norm": 0.4723998010158539,
      "learning_rate": 6.8343975362598355e-06,
      "loss": 0.3382,
      "step": 25680
    },
    {
      "epoch": 1.1785140654398605,
      "grad_norm": 0.482205331325531,
      "learning_rate": 6.834169443475043e-06,
      "loss": 0.2954,
      "step": 25681
    },
    {
      "epoch": 1.178559955945115,
      "grad_norm": 0.4411805272102356,
      "learning_rate": 6.833941346279651e-06,
      "loss": 0.2729,
      "step": 25682
    },
    {
      "epoch": 1.1786058464503695,
      "grad_norm": 0.44354721903800964,
      "learning_rate": 6.833713244674209e-06,
      "loss": 0.3232,
      "step": 25683
    },
    {
      "epoch": 1.178651736955624,
      "grad_norm": 0.4456596076488495,
      "learning_rate": 6.833485138659266e-06,
      "loss": 0.3318,
      "step": 25684
    },
    {
      "epoch": 1.1786976274608783,
      "grad_norm": 0.49992361664772034,
      "learning_rate": 6.833257028235371e-06,
      "loss": 0.3896,
      "step": 25685
    },
    {
      "epoch": 1.1787435179661327,
      "grad_norm": 0.42232975363731384,
      "learning_rate": 6.833028913403069e-06,
      "loss": 0.2892,
      "step": 25686
    },
    {
      "epoch": 1.1787894084713872,
      "grad_norm": 0.5516404509544373,
      "learning_rate": 6.832800794162912e-06,
      "loss": 0.4031,
      "step": 25687
    },
    {
      "epoch": 1.1788352989766417,
      "grad_norm": 0.4509449899196625,
      "learning_rate": 6.832572670515448e-06,
      "loss": 0.2954,
      "step": 25688
    },
    {
      "epoch": 1.1788811894818962,
      "grad_norm": 0.4559166431427002,
      "learning_rate": 6.8323445424612245e-06,
      "loss": 0.3141,
      "step": 25689
    },
    {
      "epoch": 1.1789270799871507,
      "grad_norm": 0.4875146448612213,
      "learning_rate": 6.8321164100007905e-06,
      "loss": 0.3361,
      "step": 25690
    },
    {
      "epoch": 1.178972970492405,
      "grad_norm": 0.46398815512657166,
      "learning_rate": 6.831888273134696e-06,
      "loss": 0.3456,
      "step": 25691
    },
    {
      "epoch": 1.1790188609976595,
      "grad_norm": 0.4595908224582672,
      "learning_rate": 6.831660131863487e-06,
      "loss": 0.3805,
      "step": 25692
    },
    {
      "epoch": 1.179064751502914,
      "grad_norm": 0.4863419532775879,
      "learning_rate": 6.831431986187713e-06,
      "loss": 0.3577,
      "step": 25693
    },
    {
      "epoch": 1.1791106420081685,
      "grad_norm": 0.4636259078979492,
      "learning_rate": 6.8312038361079246e-06,
      "loss": 0.3366,
      "step": 25694
    },
    {
      "epoch": 1.179156532513423,
      "grad_norm": 0.5204973816871643,
      "learning_rate": 6.8309756816246695e-06,
      "loss": 0.3578,
      "step": 25695
    },
    {
      "epoch": 1.1792024230186775,
      "grad_norm": 0.46602097153663635,
      "learning_rate": 6.830747522738495e-06,
      "loss": 0.3456,
      "step": 25696
    },
    {
      "epoch": 1.179248313523932,
      "grad_norm": 0.4656119644641876,
      "learning_rate": 6.8305193594499495e-06,
      "loss": 0.3534,
      "step": 25697
    },
    {
      "epoch": 1.1792942040291863,
      "grad_norm": 0.47305208444595337,
      "learning_rate": 6.830291191759582e-06,
      "loss": 0.3873,
      "step": 25698
    },
    {
      "epoch": 1.1793400945344408,
      "grad_norm": 0.5081974864006042,
      "learning_rate": 6.8300630196679455e-06,
      "loss": 0.3972,
      "step": 25699
    },
    {
      "epoch": 1.1793859850396953,
      "grad_norm": 0.45130231976509094,
      "learning_rate": 6.829834843175582e-06,
      "loss": 0.3215,
      "step": 25700
    },
    {
      "epoch": 1.1794318755449498,
      "grad_norm": 0.4616207480430603,
      "learning_rate": 6.829606662283043e-06,
      "loss": 0.3293,
      "step": 25701
    },
    {
      "epoch": 1.1794777660502043,
      "grad_norm": 0.4813922941684723,
      "learning_rate": 6.82937847699088e-06,
      "loss": 0.3318,
      "step": 25702
    },
    {
      "epoch": 1.1795236565554588,
      "grad_norm": 0.5179945230484009,
      "learning_rate": 6.829150287299637e-06,
      "loss": 0.4318,
      "step": 25703
    },
    {
      "epoch": 1.179569547060713,
      "grad_norm": 0.49133649468421936,
      "learning_rate": 6.828922093209865e-06,
      "loss": 0.3757,
      "step": 25704
    },
    {
      "epoch": 1.1796154375659675,
      "grad_norm": 0.473413348197937,
      "learning_rate": 6.828693894722113e-06,
      "loss": 0.3488,
      "step": 25705
    },
    {
      "epoch": 1.179661328071222,
      "grad_norm": 0.41615843772888184,
      "learning_rate": 6.82846569183693e-06,
      "loss": 0.2711,
      "step": 25706
    },
    {
      "epoch": 1.1797072185764765,
      "grad_norm": 0.46621787548065186,
      "learning_rate": 6.828237484554863e-06,
      "loss": 0.3223,
      "step": 25707
    },
    {
      "epoch": 1.179753109081731,
      "grad_norm": 0.44677236676216125,
      "learning_rate": 6.828009272876462e-06,
      "loss": 0.2948,
      "step": 25708
    },
    {
      "epoch": 1.1797989995869855,
      "grad_norm": 0.4850485920906067,
      "learning_rate": 6.827781056802275e-06,
      "loss": 0.3629,
      "step": 25709
    },
    {
      "epoch": 1.17984489009224,
      "grad_norm": 0.47587043046951294,
      "learning_rate": 6.8275528363328535e-06,
      "loss": 0.3564,
      "step": 25710
    },
    {
      "epoch": 1.1798907805974943,
      "grad_norm": 0.4721478819847107,
      "learning_rate": 6.827324611468743e-06,
      "loss": 0.3629,
      "step": 25711
    },
    {
      "epoch": 1.1799366711027488,
      "grad_norm": 0.485524982213974,
      "learning_rate": 6.827096382210493e-06,
      "loss": 0.3975,
      "step": 25712
    },
    {
      "epoch": 1.1799825616080033,
      "grad_norm": 0.4567892253398895,
      "learning_rate": 6.826868148558654e-06,
      "loss": 0.3295,
      "step": 25713
    },
    {
      "epoch": 1.1800284521132578,
      "grad_norm": 0.4610452651977539,
      "learning_rate": 6.826639910513772e-06,
      "loss": 0.3221,
      "step": 25714
    },
    {
      "epoch": 1.1800743426185123,
      "grad_norm": 0.45461663603782654,
      "learning_rate": 6.826411668076398e-06,
      "loss": 0.294,
      "step": 25715
    },
    {
      "epoch": 1.1801202331237666,
      "grad_norm": 0.43115073442459106,
      "learning_rate": 6.826183421247081e-06,
      "loss": 0.3046,
      "step": 25716
    },
    {
      "epoch": 1.180166123629021,
      "grad_norm": 0.44472235441207886,
      "learning_rate": 6.82595517002637e-06,
      "loss": 0.3308,
      "step": 25717
    },
    {
      "epoch": 1.1802120141342756,
      "grad_norm": 0.5108226537704468,
      "learning_rate": 6.825726914414812e-06,
      "loss": 0.4526,
      "step": 25718
    },
    {
      "epoch": 1.18025790463953,
      "grad_norm": 0.44894856214523315,
      "learning_rate": 6.825498654412958e-06,
      "loss": 0.3203,
      "step": 25719
    },
    {
      "epoch": 1.1803037951447846,
      "grad_norm": 0.4757935702800751,
      "learning_rate": 6.825270390021355e-06,
      "loss": 0.3813,
      "step": 25720
    },
    {
      "epoch": 1.180349685650039,
      "grad_norm": 0.4827066659927368,
      "learning_rate": 6.825042121240553e-06,
      "loss": 0.3366,
      "step": 25721
    },
    {
      "epoch": 1.1803955761552936,
      "grad_norm": 0.4591805338859558,
      "learning_rate": 6.824813848071101e-06,
      "loss": 0.344,
      "step": 25722
    },
    {
      "epoch": 1.180441466660548,
      "grad_norm": 0.49678781628608704,
      "learning_rate": 6.8245855705135466e-06,
      "loss": 0.4316,
      "step": 25723
    },
    {
      "epoch": 1.1804873571658023,
      "grad_norm": 0.4512389898300171,
      "learning_rate": 6.82435728856844e-06,
      "loss": 0.297,
      "step": 25724
    },
    {
      "epoch": 1.1805332476710568,
      "grad_norm": 0.4565200209617615,
      "learning_rate": 6.82412900223633e-06,
      "loss": 0.3211,
      "step": 25725
    },
    {
      "epoch": 1.1805791381763113,
      "grad_norm": 0.4762606620788574,
      "learning_rate": 6.8239007115177656e-06,
      "loss": 0.3737,
      "step": 25726
    },
    {
      "epoch": 1.1806250286815658,
      "grad_norm": 0.45365360379219055,
      "learning_rate": 6.8236724164132965e-06,
      "loss": 0.3543,
      "step": 25727
    },
    {
      "epoch": 1.1806709191868203,
      "grad_norm": 0.489239901304245,
      "learning_rate": 6.82344411692347e-06,
      "loss": 0.3743,
      "step": 25728
    },
    {
      "epoch": 1.1807168096920746,
      "grad_norm": 0.4461171627044678,
      "learning_rate": 6.8232158130488355e-06,
      "loss": 0.3059,
      "step": 25729
    },
    {
      "epoch": 1.180762700197329,
      "grad_norm": 0.47868844866752625,
      "learning_rate": 6.822987504789944e-06,
      "loss": 0.3467,
      "step": 25730
    },
    {
      "epoch": 1.1808085907025836,
      "grad_norm": 0.4418332874774933,
      "learning_rate": 6.822759192147341e-06,
      "loss": 0.3282,
      "step": 25731
    },
    {
      "epoch": 1.180854481207838,
      "grad_norm": 0.42974305152893066,
      "learning_rate": 6.8225308751215795e-06,
      "loss": 0.3079,
      "step": 25732
    },
    {
      "epoch": 1.1809003717130926,
      "grad_norm": 0.43337303400039673,
      "learning_rate": 6.8223025537132044e-06,
      "loss": 0.3025,
      "step": 25733
    },
    {
      "epoch": 1.180946262218347,
      "grad_norm": 0.4633924961090088,
      "learning_rate": 6.822074227922768e-06,
      "loss": 0.3528,
      "step": 25734
    },
    {
      "epoch": 1.1809921527236016,
      "grad_norm": 0.521598219871521,
      "learning_rate": 6.821845897750816e-06,
      "loss": 0.4649,
      "step": 25735
    },
    {
      "epoch": 1.1810380432288559,
      "grad_norm": 0.47039860486984253,
      "learning_rate": 6.821617563197902e-06,
      "loss": 0.3609,
      "step": 25736
    },
    {
      "epoch": 1.1810839337341104,
      "grad_norm": 0.48617884516716003,
      "learning_rate": 6.821389224264572e-06,
      "loss": 0.3588,
      "step": 25737
    },
    {
      "epoch": 1.1811298242393649,
      "grad_norm": 0.4291553497314453,
      "learning_rate": 6.821160880951376e-06,
      "loss": 0.2924,
      "step": 25738
    },
    {
      "epoch": 1.1811757147446194,
      "grad_norm": 0.4874695837497711,
      "learning_rate": 6.820932533258864e-06,
      "loss": 0.364,
      "step": 25739
    },
    {
      "epoch": 1.1812216052498739,
      "grad_norm": 0.4542138874530792,
      "learning_rate": 6.820704181187582e-06,
      "loss": 0.3632,
      "step": 25740
    },
    {
      "epoch": 1.1812674957551283,
      "grad_norm": 0.4938436448574066,
      "learning_rate": 6.8204758247380805e-06,
      "loss": 0.388,
      "step": 25741
    },
    {
      "epoch": 1.1813133862603826,
      "grad_norm": 0.432159960269928,
      "learning_rate": 6.8202474639109105e-06,
      "loss": 0.3214,
      "step": 25742
    },
    {
      "epoch": 1.1813592767656371,
      "grad_norm": 0.456684410572052,
      "learning_rate": 6.820019098706621e-06,
      "loss": 0.303,
      "step": 25743
    },
    {
      "epoch": 1.1814051672708916,
      "grad_norm": 0.45351463556289673,
      "learning_rate": 6.819790729125757e-06,
      "loss": 0.3644,
      "step": 25744
    },
    {
      "epoch": 1.1814510577761461,
      "grad_norm": 0.487506240606308,
      "learning_rate": 6.819562355168874e-06,
      "loss": 0.3601,
      "step": 25745
    },
    {
      "epoch": 1.1814969482814006,
      "grad_norm": 0.43712329864501953,
      "learning_rate": 6.819333976836515e-06,
      "loss": 0.3209,
      "step": 25746
    },
    {
      "epoch": 1.1815428387866551,
      "grad_norm": 0.5129201412200928,
      "learning_rate": 6.819105594129234e-06,
      "loss": 0.4145,
      "step": 25747
    },
    {
      "epoch": 1.1815887292919096,
      "grad_norm": 0.454344779253006,
      "learning_rate": 6.818877207047576e-06,
      "loss": 0.3187,
      "step": 25748
    },
    {
      "epoch": 1.1816346197971639,
      "grad_norm": 0.48502662777900696,
      "learning_rate": 6.818648815592095e-06,
      "loss": 0.3607,
      "step": 25749
    },
    {
      "epoch": 1.1816805103024184,
      "grad_norm": 0.4417001008987427,
      "learning_rate": 6.818420419763336e-06,
      "loss": 0.2897,
      "step": 25750
    },
    {
      "epoch": 1.1817264008076729,
      "grad_norm": 0.4868485927581787,
      "learning_rate": 6.818192019561849e-06,
      "loss": 0.3796,
      "step": 25751
    },
    {
      "epoch": 1.1817722913129274,
      "grad_norm": 0.4825170934200287,
      "learning_rate": 6.817963614988185e-06,
      "loss": 0.3756,
      "step": 25752
    },
    {
      "epoch": 1.1818181818181819,
      "grad_norm": 0.5274156332015991,
      "learning_rate": 6.817735206042893e-06,
      "loss": 0.4017,
      "step": 25753
    },
    {
      "epoch": 1.1818640723234364,
      "grad_norm": 0.5339662432670593,
      "learning_rate": 6.817506792726522e-06,
      "loss": 0.4089,
      "step": 25754
    },
    {
      "epoch": 1.1819099628286907,
      "grad_norm": 0.46307602524757385,
      "learning_rate": 6.817278375039617e-06,
      "loss": 0.3374,
      "step": 25755
    },
    {
      "epoch": 1.1819558533339452,
      "grad_norm": 0.48695841431617737,
      "learning_rate": 6.817049952982735e-06,
      "loss": 0.3594,
      "step": 25756
    },
    {
      "epoch": 1.1820017438391996,
      "grad_norm": 0.43565067648887634,
      "learning_rate": 6.816821526556419e-06,
      "loss": 0.2882,
      "step": 25757
    },
    {
      "epoch": 1.1820476343444541,
      "grad_norm": 0.41802576184272766,
      "learning_rate": 6.816593095761221e-06,
      "loss": 0.2853,
      "step": 25758
    },
    {
      "epoch": 1.1820935248497086,
      "grad_norm": 0.47147998213768005,
      "learning_rate": 6.816364660597692e-06,
      "loss": 0.356,
      "step": 25759
    },
    {
      "epoch": 1.1821394153549631,
      "grad_norm": 0.46147987246513367,
      "learning_rate": 6.816136221066378e-06,
      "loss": 0.3492,
      "step": 25760
    },
    {
      "epoch": 1.1821853058602176,
      "grad_norm": 0.4783500134944916,
      "learning_rate": 6.815907777167829e-06,
      "loss": 0.3575,
      "step": 25761
    },
    {
      "epoch": 1.182231196365472,
      "grad_norm": 0.44525212049484253,
      "learning_rate": 6.815679328902596e-06,
      "loss": 0.3525,
      "step": 25762
    },
    {
      "epoch": 1.1822770868707264,
      "grad_norm": 0.45950812101364136,
      "learning_rate": 6.815450876271225e-06,
      "loss": 0.3464,
      "step": 25763
    },
    {
      "epoch": 1.182322977375981,
      "grad_norm": 0.4603084623813629,
      "learning_rate": 6.8152224192742715e-06,
      "loss": 0.3368,
      "step": 25764
    },
    {
      "epoch": 1.1823688678812354,
      "grad_norm": 0.47860515117645264,
      "learning_rate": 6.814993957912279e-06,
      "loss": 0.3523,
      "step": 25765
    },
    {
      "epoch": 1.18241475838649,
      "grad_norm": 0.506956160068512,
      "learning_rate": 6.814765492185798e-06,
      "loss": 0.3872,
      "step": 25766
    },
    {
      "epoch": 1.1824606488917442,
      "grad_norm": 0.4673523008823395,
      "learning_rate": 6.81453702209538e-06,
      "loss": 0.36,
      "step": 25767
    },
    {
      "epoch": 1.1825065393969987,
      "grad_norm": 0.4840168356895447,
      "learning_rate": 6.8143085476415715e-06,
      "loss": 0.3727,
      "step": 25768
    },
    {
      "epoch": 1.1825524299022532,
      "grad_norm": 0.45054978132247925,
      "learning_rate": 6.814080068824924e-06,
      "loss": 0.283,
      "step": 25769
    },
    {
      "epoch": 1.1825983204075077,
      "grad_norm": 0.47714748978614807,
      "learning_rate": 6.813851585645987e-06,
      "loss": 0.4091,
      "step": 25770
    },
    {
      "epoch": 1.1826442109127622,
      "grad_norm": 0.4814198315143585,
      "learning_rate": 6.81362309810531e-06,
      "loss": 0.3678,
      "step": 25771
    },
    {
      "epoch": 1.1826901014180167,
      "grad_norm": 0.42910128831863403,
      "learning_rate": 6.81339460620344e-06,
      "loss": 0.3321,
      "step": 25772
    },
    {
      "epoch": 1.1827359919232712,
      "grad_norm": 0.4507873058319092,
      "learning_rate": 6.813166109940931e-06,
      "loss": 0.3279,
      "step": 25773
    },
    {
      "epoch": 1.1827818824285254,
      "grad_norm": 0.46349766850471497,
      "learning_rate": 6.812937609318328e-06,
      "loss": 0.3388,
      "step": 25774
    },
    {
      "epoch": 1.18282777293378,
      "grad_norm": 0.4771181046962738,
      "learning_rate": 6.812709104336182e-06,
      "loss": 0.4019,
      "step": 25775
    },
    {
      "epoch": 1.1828736634390344,
      "grad_norm": 0.45871517062187195,
      "learning_rate": 6.8124805949950444e-06,
      "loss": 0.3452,
      "step": 25776
    },
    {
      "epoch": 1.182919553944289,
      "grad_norm": 0.3916628062725067,
      "learning_rate": 6.8122520812954605e-06,
      "loss": 0.252,
      "step": 25777
    },
    {
      "epoch": 1.1829654444495434,
      "grad_norm": 0.48789238929748535,
      "learning_rate": 6.812023563237985e-06,
      "loss": 0.3824,
      "step": 25778
    },
    {
      "epoch": 1.183011334954798,
      "grad_norm": 0.47765928506851196,
      "learning_rate": 6.811795040823162e-06,
      "loss": 0.3755,
      "step": 25779
    },
    {
      "epoch": 1.1830572254600522,
      "grad_norm": 0.4620749056339264,
      "learning_rate": 6.811566514051545e-06,
      "loss": 0.3238,
      "step": 25780
    },
    {
      "epoch": 1.1831031159653067,
      "grad_norm": 0.432862251996994,
      "learning_rate": 6.811337982923683e-06,
      "loss": 0.29,
      "step": 25781
    },
    {
      "epoch": 1.1831490064705612,
      "grad_norm": 0.4892668128013611,
      "learning_rate": 6.811109447440125e-06,
      "loss": 0.3921,
      "step": 25782
    },
    {
      "epoch": 1.1831948969758157,
      "grad_norm": 0.448934406042099,
      "learning_rate": 6.8108809076014186e-06,
      "loss": 0.3325,
      "step": 25783
    },
    {
      "epoch": 1.1832407874810702,
      "grad_norm": 0.4568212926387787,
      "learning_rate": 6.8106523634081165e-06,
      "loss": 0.3572,
      "step": 25784
    },
    {
      "epoch": 1.1832866779863247,
      "grad_norm": 0.475515216588974,
      "learning_rate": 6.810423814860766e-06,
      "loss": 0.3484,
      "step": 25785
    },
    {
      "epoch": 1.1833325684915792,
      "grad_norm": 0.46559572219848633,
      "learning_rate": 6.810195261959919e-06,
      "loss": 0.4005,
      "step": 25786
    },
    {
      "epoch": 1.1833784589968335,
      "grad_norm": 0.4708154499530792,
      "learning_rate": 6.8099667047061235e-06,
      "loss": 0.3635,
      "step": 25787
    },
    {
      "epoch": 1.183424349502088,
      "grad_norm": 0.45441746711730957,
      "learning_rate": 6.809738143099928e-06,
      "loss": 0.337,
      "step": 25788
    },
    {
      "epoch": 1.1834702400073425,
      "grad_norm": 0.44106560945510864,
      "learning_rate": 6.809509577141884e-06,
      "loss": 0.3033,
      "step": 25789
    },
    {
      "epoch": 1.183516130512597,
      "grad_norm": 0.4885255694389343,
      "learning_rate": 6.809281006832541e-06,
      "loss": 0.3744,
      "step": 25790
    },
    {
      "epoch": 1.1835620210178515,
      "grad_norm": 0.5032009482383728,
      "learning_rate": 6.809052432172447e-06,
      "loss": 0.3978,
      "step": 25791
    },
    {
      "epoch": 1.183607911523106,
      "grad_norm": 0.4950259327888489,
      "learning_rate": 6.808823853162154e-06,
      "loss": 0.4124,
      "step": 25792
    },
    {
      "epoch": 1.1836538020283602,
      "grad_norm": 0.440635085105896,
      "learning_rate": 6.808595269802211e-06,
      "loss": 0.2972,
      "step": 25793
    },
    {
      "epoch": 1.1836996925336147,
      "grad_norm": 0.48225557804107666,
      "learning_rate": 6.808366682093166e-06,
      "loss": 0.3979,
      "step": 25794
    },
    {
      "epoch": 1.1837455830388692,
      "grad_norm": 0.45347875356674194,
      "learning_rate": 6.808138090035571e-06,
      "loss": 0.3325,
      "step": 25795
    },
    {
      "epoch": 1.1837914735441237,
      "grad_norm": 0.46068161725997925,
      "learning_rate": 6.807909493629972e-06,
      "loss": 0.3854,
      "step": 25796
    },
    {
      "epoch": 1.1838373640493782,
      "grad_norm": 0.47206705808639526,
      "learning_rate": 6.8076808928769245e-06,
      "loss": 0.3776,
      "step": 25797
    },
    {
      "epoch": 1.1838832545546327,
      "grad_norm": 0.4943772554397583,
      "learning_rate": 6.807452287776975e-06,
      "loss": 0.4392,
      "step": 25798
    },
    {
      "epoch": 1.1839291450598872,
      "grad_norm": 0.45922866463661194,
      "learning_rate": 6.80722367833067e-06,
      "loss": 0.3847,
      "step": 25799
    },
    {
      "epoch": 1.1839750355651415,
      "grad_norm": 0.46701470017433167,
      "learning_rate": 6.806995064538564e-06,
      "loss": 0.3494,
      "step": 25800
    },
    {
      "epoch": 1.184020926070396,
      "grad_norm": 0.6164257526397705,
      "learning_rate": 6.806766446401206e-06,
      "loss": 0.3562,
      "step": 25801
    },
    {
      "epoch": 1.1840668165756505,
      "grad_norm": 0.46305176615715027,
      "learning_rate": 6.806537823919144e-06,
      "loss": 0.3665,
      "step": 25802
    },
    {
      "epoch": 1.184112707080905,
      "grad_norm": 0.4965791404247284,
      "learning_rate": 6.8063091970929294e-06,
      "loss": 0.3783,
      "step": 25803
    },
    {
      "epoch": 1.1841585975861595,
      "grad_norm": 0.495858371257782,
      "learning_rate": 6.806080565923112e-06,
      "loss": 0.4197,
      "step": 25804
    },
    {
      "epoch": 1.1842044880914138,
      "grad_norm": 0.5072774887084961,
      "learning_rate": 6.805851930410239e-06,
      "loss": 0.3992,
      "step": 25805
    },
    {
      "epoch": 1.1842503785966683,
      "grad_norm": 0.4725784957408905,
      "learning_rate": 6.805623290554862e-06,
      "loss": 0.3286,
      "step": 25806
    },
    {
      "epoch": 1.1842962691019228,
      "grad_norm": 0.5396092534065247,
      "learning_rate": 6.805394646357533e-06,
      "loss": 0.4218,
      "step": 25807
    },
    {
      "epoch": 1.1843421596071773,
      "grad_norm": 0.44107815623283386,
      "learning_rate": 6.805165997818798e-06,
      "loss": 0.3224,
      "step": 25808
    },
    {
      "epoch": 1.1843880501124318,
      "grad_norm": 0.452438086271286,
      "learning_rate": 6.8049373449392086e-06,
      "loss": 0.3299,
      "step": 25809
    },
    {
      "epoch": 1.1844339406176863,
      "grad_norm": 0.4493481516838074,
      "learning_rate": 6.804708687719315e-06,
      "loss": 0.351,
      "step": 25810
    },
    {
      "epoch": 1.1844798311229408,
      "grad_norm": 0.4752193093299866,
      "learning_rate": 6.804480026159666e-06,
      "loss": 0.3159,
      "step": 25811
    },
    {
      "epoch": 1.1845257216281952,
      "grad_norm": 0.48184528946876526,
      "learning_rate": 6.804251360260813e-06,
      "loss": 0.3466,
      "step": 25812
    },
    {
      "epoch": 1.1845716121334495,
      "grad_norm": 0.4440290331840515,
      "learning_rate": 6.804022690023304e-06,
      "loss": 0.302,
      "step": 25813
    },
    {
      "epoch": 1.184617502638704,
      "grad_norm": 0.5067896246910095,
      "learning_rate": 6.80379401544769e-06,
      "loss": 0.4836,
      "step": 25814
    },
    {
      "epoch": 1.1846633931439585,
      "grad_norm": 0.45428988337516785,
      "learning_rate": 6.803565336534521e-06,
      "loss": 0.3232,
      "step": 25815
    },
    {
      "epoch": 1.184709283649213,
      "grad_norm": 0.47001200914382935,
      "learning_rate": 6.803336653284347e-06,
      "loss": 0.3567,
      "step": 25816
    },
    {
      "epoch": 1.1847551741544675,
      "grad_norm": 0.4891206622123718,
      "learning_rate": 6.803107965697715e-06,
      "loss": 0.3957,
      "step": 25817
    },
    {
      "epoch": 1.1848010646597218,
      "grad_norm": 0.47962456941604614,
      "learning_rate": 6.80287927377518e-06,
      "loss": 0.4058,
      "step": 25818
    },
    {
      "epoch": 1.1848469551649763,
      "grad_norm": 0.45609140396118164,
      "learning_rate": 6.802650577517289e-06,
      "loss": 0.3462,
      "step": 25819
    },
    {
      "epoch": 1.1848928456702308,
      "grad_norm": 0.44039976596832275,
      "learning_rate": 6.802421876924592e-06,
      "loss": 0.3102,
      "step": 25820
    },
    {
      "epoch": 1.1849387361754853,
      "grad_norm": 0.4810160994529724,
      "learning_rate": 6.802193171997639e-06,
      "loss": 0.3894,
      "step": 25821
    },
    {
      "epoch": 1.1849846266807398,
      "grad_norm": 0.42905473709106445,
      "learning_rate": 6.801964462736981e-06,
      "loss": 0.303,
      "step": 25822
    },
    {
      "epoch": 1.1850305171859943,
      "grad_norm": 0.5115122199058533,
      "learning_rate": 6.801735749143166e-06,
      "loss": 0.3651,
      "step": 25823
    },
    {
      "epoch": 1.1850764076912488,
      "grad_norm": 0.4448230564594269,
      "learning_rate": 6.801507031216746e-06,
      "loss": 0.3349,
      "step": 25824
    },
    {
      "epoch": 1.185122298196503,
      "grad_norm": 0.46673721075057983,
      "learning_rate": 6.8012783089582705e-06,
      "loss": 0.3442,
      "step": 25825
    },
    {
      "epoch": 1.1851681887017576,
      "grad_norm": 0.5651987195014954,
      "learning_rate": 6.801049582368288e-06,
      "loss": 0.4485,
      "step": 25826
    },
    {
      "epoch": 1.185214079207012,
      "grad_norm": 0.44050121307373047,
      "learning_rate": 6.800820851447351e-06,
      "loss": 0.3132,
      "step": 25827
    },
    {
      "epoch": 1.1852599697122665,
      "grad_norm": 0.432381272315979,
      "learning_rate": 6.800592116196006e-06,
      "loss": 0.3016,
      "step": 25828
    },
    {
      "epoch": 1.185305860217521,
      "grad_norm": 0.4215467870235443,
      "learning_rate": 6.800363376614808e-06,
      "loss": 0.2961,
      "step": 25829
    },
    {
      "epoch": 1.1853517507227755,
      "grad_norm": 0.5149332880973816,
      "learning_rate": 6.800134632704304e-06,
      "loss": 0.4266,
      "step": 25830
    },
    {
      "epoch": 1.1853976412280298,
      "grad_norm": 0.4385400712490082,
      "learning_rate": 6.799905884465042e-06,
      "loss": 0.3129,
      "step": 25831
    },
    {
      "epoch": 1.1854435317332843,
      "grad_norm": 0.4783274829387665,
      "learning_rate": 6.799677131897576e-06,
      "loss": 0.3715,
      "step": 25832
    },
    {
      "epoch": 1.1854894222385388,
      "grad_norm": 0.48721322417259216,
      "learning_rate": 6.799448375002454e-06,
      "loss": 0.3374,
      "step": 25833
    },
    {
      "epoch": 1.1855353127437933,
      "grad_norm": 0.44686073064804077,
      "learning_rate": 6.7992196137802266e-06,
      "loss": 0.2944,
      "step": 25834
    },
    {
      "epoch": 1.1855812032490478,
      "grad_norm": 0.45809438824653625,
      "learning_rate": 6.7989908482314444e-06,
      "loss": 0.3261,
      "step": 25835
    },
    {
      "epoch": 1.1856270937543023,
      "grad_norm": 0.41723141074180603,
      "learning_rate": 6.7987620783566575e-06,
      "loss": 0.2819,
      "step": 25836
    },
    {
      "epoch": 1.1856729842595568,
      "grad_norm": 0.43353667855262756,
      "learning_rate": 6.798533304156414e-06,
      "loss": 0.3106,
      "step": 25837
    },
    {
      "epoch": 1.185718874764811,
      "grad_norm": 0.45132872462272644,
      "learning_rate": 6.798304525631267e-06,
      "loss": 0.308,
      "step": 25838
    },
    {
      "epoch": 1.1857647652700656,
      "grad_norm": 0.45703521370887756,
      "learning_rate": 6.7980757427817645e-06,
      "loss": 0.3345,
      "step": 25839
    },
    {
      "epoch": 1.18581065577532,
      "grad_norm": 0.4652842581272125,
      "learning_rate": 6.797846955608456e-06,
      "loss": 0.3342,
      "step": 25840
    },
    {
      "epoch": 1.1858565462805746,
      "grad_norm": 0.4615703225135803,
      "learning_rate": 6.797618164111897e-06,
      "loss": 0.3714,
      "step": 25841
    },
    {
      "epoch": 1.185902436785829,
      "grad_norm": 0.43365955352783203,
      "learning_rate": 6.7973893682926294e-06,
      "loss": 0.3054,
      "step": 25842
    },
    {
      "epoch": 1.1859483272910836,
      "grad_norm": 0.4654073119163513,
      "learning_rate": 6.797160568151209e-06,
      "loss": 0.3332,
      "step": 25843
    },
    {
      "epoch": 1.1859942177963378,
      "grad_norm": 0.49784529209136963,
      "learning_rate": 6.796931763688185e-06,
      "loss": 0.3803,
      "step": 25844
    },
    {
      "epoch": 1.1860401083015923,
      "grad_norm": 0.42667368054389954,
      "learning_rate": 6.796702954904106e-06,
      "loss": 0.2666,
      "step": 25845
    },
    {
      "epoch": 1.1860859988068468,
      "grad_norm": 0.4812326729297638,
      "learning_rate": 6.796474141799525e-06,
      "loss": 0.405,
      "step": 25846
    },
    {
      "epoch": 1.1861318893121013,
      "grad_norm": 0.4847441613674164,
      "learning_rate": 6.796245324374991e-06,
      "loss": 0.3538,
      "step": 25847
    },
    {
      "epoch": 1.1861777798173558,
      "grad_norm": 0.46031486988067627,
      "learning_rate": 6.796016502631052e-06,
      "loss": 0.3846,
      "step": 25848
    },
    {
      "epoch": 1.1862236703226103,
      "grad_norm": 0.515594482421875,
      "learning_rate": 6.795787676568262e-06,
      "loss": 0.4836,
      "step": 25849
    },
    {
      "epoch": 1.1862695608278648,
      "grad_norm": 0.46542873978614807,
      "learning_rate": 6.795558846187168e-06,
      "loss": 0.3779,
      "step": 25850
    },
    {
      "epoch": 1.186315451333119,
      "grad_norm": 0.4470653235912323,
      "learning_rate": 6.795330011488322e-06,
      "loss": 0.344,
      "step": 25851
    },
    {
      "epoch": 1.1863613418383736,
      "grad_norm": 0.4862048923969269,
      "learning_rate": 6.795101172472276e-06,
      "loss": 0.4211,
      "step": 25852
    },
    {
      "epoch": 1.186407232343628,
      "grad_norm": 0.4666740894317627,
      "learning_rate": 6.794872329139576e-06,
      "loss": 0.3307,
      "step": 25853
    },
    {
      "epoch": 1.1864531228488826,
      "grad_norm": 0.46470269560813904,
      "learning_rate": 6.794643481490775e-06,
      "loss": 0.3489,
      "step": 25854
    },
    {
      "epoch": 1.186499013354137,
      "grad_norm": 0.472949355840683,
      "learning_rate": 6.794414629526423e-06,
      "loss": 0.3504,
      "step": 25855
    },
    {
      "epoch": 1.1865449038593914,
      "grad_norm": 0.43333902955055237,
      "learning_rate": 6.7941857732470686e-06,
      "loss": 0.315,
      "step": 25856
    },
    {
      "epoch": 1.1865907943646459,
      "grad_norm": 0.47295641899108887,
      "learning_rate": 6.793956912653266e-06,
      "loss": 0.3656,
      "step": 25857
    },
    {
      "epoch": 1.1866366848699004,
      "grad_norm": 0.47792354226112366,
      "learning_rate": 6.793728047745562e-06,
      "loss": 0.3869,
      "step": 25858
    },
    {
      "epoch": 1.1866825753751549,
      "grad_norm": 0.4636421203613281,
      "learning_rate": 6.793499178524509e-06,
      "loss": 0.3698,
      "step": 25859
    },
    {
      "epoch": 1.1867284658804094,
      "grad_norm": 0.4990634322166443,
      "learning_rate": 6.7932703049906555e-06,
      "loss": 0.3863,
      "step": 25860
    },
    {
      "epoch": 1.1867743563856639,
      "grad_norm": 0.5338382124900818,
      "learning_rate": 6.793041427144553e-06,
      "loss": 0.4114,
      "step": 25861
    },
    {
      "epoch": 1.1868202468909184,
      "grad_norm": 0.4765791893005371,
      "learning_rate": 6.792812544986752e-06,
      "loss": 0.3713,
      "step": 25862
    },
    {
      "epoch": 1.1868661373961726,
      "grad_norm": 0.48965150117874146,
      "learning_rate": 6.792583658517804e-06,
      "loss": 0.3929,
      "step": 25863
    },
    {
      "epoch": 1.1869120279014271,
      "grad_norm": 0.4491797685623169,
      "learning_rate": 6.7923547677382564e-06,
      "loss": 0.3446,
      "step": 25864
    },
    {
      "epoch": 1.1869579184066816,
      "grad_norm": 0.4675693213939667,
      "learning_rate": 6.792125872648661e-06,
      "loss": 0.3815,
      "step": 25865
    },
    {
      "epoch": 1.1870038089119361,
      "grad_norm": 0.4605647623538971,
      "learning_rate": 6.79189697324957e-06,
      "loss": 0.3513,
      "step": 25866
    },
    {
      "epoch": 1.1870496994171906,
      "grad_norm": 0.46836647391319275,
      "learning_rate": 6.791668069541531e-06,
      "loss": 0.3782,
      "step": 25867
    },
    {
      "epoch": 1.1870955899224451,
      "grad_norm": 0.47099512815475464,
      "learning_rate": 6.791439161525096e-06,
      "loss": 0.3867,
      "step": 25868
    },
    {
      "epoch": 1.1871414804276994,
      "grad_norm": 0.4759407639503479,
      "learning_rate": 6.791210249200817e-06,
      "loss": 0.412,
      "step": 25869
    },
    {
      "epoch": 1.187187370932954,
      "grad_norm": 0.4703865945339203,
      "learning_rate": 6.7909813325692395e-06,
      "loss": 0.347,
      "step": 25870
    },
    {
      "epoch": 1.1872332614382084,
      "grad_norm": 0.48088058829307556,
      "learning_rate": 6.790752411630919e-06,
      "loss": 0.4032,
      "step": 25871
    },
    {
      "epoch": 1.187279151943463,
      "grad_norm": 0.42564520239830017,
      "learning_rate": 6.790523486386404e-06,
      "loss": 0.264,
      "step": 25872
    },
    {
      "epoch": 1.1873250424487174,
      "grad_norm": 0.46695011854171753,
      "learning_rate": 6.7902945568362445e-06,
      "loss": 0.3181,
      "step": 25873
    },
    {
      "epoch": 1.187370932953972,
      "grad_norm": 0.48359858989715576,
      "learning_rate": 6.7900656229809925e-06,
      "loss": 0.3698,
      "step": 25874
    },
    {
      "epoch": 1.1874168234592264,
      "grad_norm": 0.514275074005127,
      "learning_rate": 6.789836684821198e-06,
      "loss": 0.4334,
      "step": 25875
    },
    {
      "epoch": 1.1874627139644807,
      "grad_norm": 0.5356282591819763,
      "learning_rate": 6.78960774235741e-06,
      "loss": 0.5047,
      "step": 25876
    },
    {
      "epoch": 1.1875086044697352,
      "grad_norm": 0.45261743664741516,
      "learning_rate": 6.789378795590181e-06,
      "loss": 0.3068,
      "step": 25877
    },
    {
      "epoch": 1.1875544949749897,
      "grad_norm": 0.48713764548301697,
      "learning_rate": 6.7891498445200595e-06,
      "loss": 0.3605,
      "step": 25878
    },
    {
      "epoch": 1.1876003854802442,
      "grad_norm": 0.5010494589805603,
      "learning_rate": 6.788920889147599e-06,
      "loss": 0.3639,
      "step": 25879
    },
    {
      "epoch": 1.1876462759854987,
      "grad_norm": 0.4508076012134552,
      "learning_rate": 6.788691929473348e-06,
      "loss": 0.3434,
      "step": 25880
    },
    {
      "epoch": 1.1876921664907532,
      "grad_norm": 0.4413585662841797,
      "learning_rate": 6.788462965497856e-06,
      "loss": 0.2992,
      "step": 25881
    },
    {
      "epoch": 1.1877380569960074,
      "grad_norm": 0.47807082533836365,
      "learning_rate": 6.788233997221676e-06,
      "loss": 0.372,
      "step": 25882
    },
    {
      "epoch": 1.187783947501262,
      "grad_norm": 0.47678834199905396,
      "learning_rate": 6.788005024645359e-06,
      "loss": 0.3926,
      "step": 25883
    },
    {
      "epoch": 1.1878298380065164,
      "grad_norm": 0.4768449366092682,
      "learning_rate": 6.787776047769452e-06,
      "loss": 0.3801,
      "step": 25884
    },
    {
      "epoch": 1.187875728511771,
      "grad_norm": 0.42654556035995483,
      "learning_rate": 6.787547066594509e-06,
      "loss": 0.2997,
      "step": 25885
    },
    {
      "epoch": 1.1879216190170254,
      "grad_norm": 0.48210984468460083,
      "learning_rate": 6.787318081121079e-06,
      "loss": 0.3324,
      "step": 25886
    },
    {
      "epoch": 1.18796750952228,
      "grad_norm": 0.43425706028938293,
      "learning_rate": 6.787089091349714e-06,
      "loss": 0.3248,
      "step": 25887
    },
    {
      "epoch": 1.1880134000275344,
      "grad_norm": 0.46285316348075867,
      "learning_rate": 6.7868600972809614e-06,
      "loss": 0.3718,
      "step": 25888
    },
    {
      "epoch": 1.1880592905327887,
      "grad_norm": 0.4728703796863556,
      "learning_rate": 6.786631098915376e-06,
      "loss": 0.3602,
      "step": 25889
    },
    {
      "epoch": 1.1881051810380432,
      "grad_norm": 0.46048831939697266,
      "learning_rate": 6.786402096253507e-06,
      "loss": 0.335,
      "step": 25890
    },
    {
      "epoch": 1.1881510715432977,
      "grad_norm": 0.4514266550540924,
      "learning_rate": 6.786173089295905e-06,
      "loss": 0.3114,
      "step": 25891
    },
    {
      "epoch": 1.1881969620485522,
      "grad_norm": 0.4883325695991516,
      "learning_rate": 6.78594407804312e-06,
      "loss": 0.3729,
      "step": 25892
    },
    {
      "epoch": 1.1882428525538067,
      "grad_norm": 0.4710811376571655,
      "learning_rate": 6.785715062495701e-06,
      "loss": 0.3855,
      "step": 25893
    },
    {
      "epoch": 1.188288743059061,
      "grad_norm": 0.4730255603790283,
      "learning_rate": 6.785486042654204e-06,
      "loss": 0.407,
      "step": 25894
    },
    {
      "epoch": 1.1883346335643155,
      "grad_norm": 0.4469076097011566,
      "learning_rate": 6.785257018519174e-06,
      "loss": 0.3189,
      "step": 25895
    },
    {
      "epoch": 1.18838052406957,
      "grad_norm": 0.4753378927707672,
      "learning_rate": 6.7850279900911665e-06,
      "loss": 0.4024,
      "step": 25896
    },
    {
      "epoch": 1.1884264145748245,
      "grad_norm": 0.4426412880420685,
      "learning_rate": 6.78479895737073e-06,
      "loss": 0.3401,
      "step": 25897
    },
    {
      "epoch": 1.188472305080079,
      "grad_norm": 0.46274593472480774,
      "learning_rate": 6.784569920358415e-06,
      "loss": 0.3505,
      "step": 25898
    },
    {
      "epoch": 1.1885181955853334,
      "grad_norm": 0.45692750811576843,
      "learning_rate": 6.78434087905477e-06,
      "loss": 0.339,
      "step": 25899
    },
    {
      "epoch": 1.188564086090588,
      "grad_norm": 0.4778342843055725,
      "learning_rate": 6.784111833460351e-06,
      "loss": 0.3796,
      "step": 25900
    },
    {
      "epoch": 1.1886099765958424,
      "grad_norm": 0.45717284083366394,
      "learning_rate": 6.783882783575707e-06,
      "loss": 0.303,
      "step": 25901
    },
    {
      "epoch": 1.1886558671010967,
      "grad_norm": 0.4876936078071594,
      "learning_rate": 6.783653729401386e-06,
      "loss": 0.3831,
      "step": 25902
    },
    {
      "epoch": 1.1887017576063512,
      "grad_norm": 0.4490300714969635,
      "learning_rate": 6.783424670937942e-06,
      "loss": 0.3035,
      "step": 25903
    },
    {
      "epoch": 1.1887476481116057,
      "grad_norm": 0.4157750904560089,
      "learning_rate": 6.783195608185923e-06,
      "loss": 0.2729,
      "step": 25904
    },
    {
      "epoch": 1.1887935386168602,
      "grad_norm": 0.5280031561851501,
      "learning_rate": 6.782966541145881e-06,
      "loss": 0.4031,
      "step": 25905
    },
    {
      "epoch": 1.1888394291221147,
      "grad_norm": 0.4866751730442047,
      "learning_rate": 6.782737469818369e-06,
      "loss": 0.388,
      "step": 25906
    },
    {
      "epoch": 1.188885319627369,
      "grad_norm": 0.4589093029499054,
      "learning_rate": 6.782508394203935e-06,
      "loss": 0.3296,
      "step": 25907
    },
    {
      "epoch": 1.1889312101326235,
      "grad_norm": 0.5291438102722168,
      "learning_rate": 6.782279314303131e-06,
      "loss": 0.4462,
      "step": 25908
    },
    {
      "epoch": 1.188977100637878,
      "grad_norm": 0.4655928611755371,
      "learning_rate": 6.7820502301165085e-06,
      "loss": 0.3521,
      "step": 25909
    },
    {
      "epoch": 1.1890229911431325,
      "grad_norm": 0.5223348736763,
      "learning_rate": 6.781821141644617e-06,
      "loss": 0.4315,
      "step": 25910
    },
    {
      "epoch": 1.189068881648387,
      "grad_norm": 0.45392537117004395,
      "learning_rate": 6.7815920488880085e-06,
      "loss": 0.3262,
      "step": 25911
    },
    {
      "epoch": 1.1891147721536415,
      "grad_norm": 0.6376566886901855,
      "learning_rate": 6.781362951847234e-06,
      "loss": 0.3814,
      "step": 25912
    },
    {
      "epoch": 1.189160662658896,
      "grad_norm": 0.4888342320919037,
      "learning_rate": 6.781133850522841e-06,
      "loss": 0.3679,
      "step": 25913
    },
    {
      "epoch": 1.1892065531641502,
      "grad_norm": 0.4981851577758789,
      "learning_rate": 6.780904744915387e-06,
      "loss": 0.4325,
      "step": 25914
    },
    {
      "epoch": 1.1892524436694047,
      "grad_norm": 0.46175143122673035,
      "learning_rate": 6.780675635025415e-06,
      "loss": 0.3626,
      "step": 25915
    },
    {
      "epoch": 1.1892983341746592,
      "grad_norm": 0.4537806808948517,
      "learning_rate": 6.780446520853482e-06,
      "loss": 0.2844,
      "step": 25916
    },
    {
      "epoch": 1.1893442246799137,
      "grad_norm": 0.46106258034706116,
      "learning_rate": 6.780217402400138e-06,
      "loss": 0.3602,
      "step": 25917
    },
    {
      "epoch": 1.1893901151851682,
      "grad_norm": 0.4781891703605652,
      "learning_rate": 6.779988279665933e-06,
      "loss": 0.3986,
      "step": 25918
    },
    {
      "epoch": 1.1894360056904227,
      "grad_norm": 0.5925687551498413,
      "learning_rate": 6.779759152651417e-06,
      "loss": 0.3336,
      "step": 25919
    },
    {
      "epoch": 1.189481896195677,
      "grad_norm": 0.45777544379234314,
      "learning_rate": 6.7795300213571426e-06,
      "loss": 0.3273,
      "step": 25920
    },
    {
      "epoch": 1.1895277867009315,
      "grad_norm": 0.4867023527622223,
      "learning_rate": 6.77930088578366e-06,
      "loss": 0.3918,
      "step": 25921
    },
    {
      "epoch": 1.189573677206186,
      "grad_norm": 0.5215076208114624,
      "learning_rate": 6.779071745931521e-06,
      "loss": 0.4414,
      "step": 25922
    },
    {
      "epoch": 1.1896195677114405,
      "grad_norm": 0.4816834330558777,
      "learning_rate": 6.778842601801275e-06,
      "loss": 0.3425,
      "step": 25923
    },
    {
      "epoch": 1.189665458216695,
      "grad_norm": 0.4663441777229309,
      "learning_rate": 6.778613453393472e-06,
      "loss": 0.3286,
      "step": 25924
    },
    {
      "epoch": 1.1897113487219495,
      "grad_norm": 0.4867030084133148,
      "learning_rate": 6.778384300708667e-06,
      "loss": 0.3435,
      "step": 25925
    },
    {
      "epoch": 1.189757239227204,
      "grad_norm": 0.4495197534561157,
      "learning_rate": 6.778155143747409e-06,
      "loss": 0.3137,
      "step": 25926
    },
    {
      "epoch": 1.1898031297324583,
      "grad_norm": 0.5067280530929565,
      "learning_rate": 6.77792598251025e-06,
      "loss": 0.3661,
      "step": 25927
    },
    {
      "epoch": 1.1898490202377128,
      "grad_norm": 0.44413623213768005,
      "learning_rate": 6.7776968169977386e-06,
      "loss": 0.3446,
      "step": 25928
    },
    {
      "epoch": 1.1898949107429673,
      "grad_norm": 0.46694329380989075,
      "learning_rate": 6.777467647210429e-06,
      "loss": 0.3936,
      "step": 25929
    },
    {
      "epoch": 1.1899408012482218,
      "grad_norm": 0.4870131313800812,
      "learning_rate": 6.77723847314887e-06,
      "loss": 0.3717,
      "step": 25930
    },
    {
      "epoch": 1.1899866917534763,
      "grad_norm": 0.4979531764984131,
      "learning_rate": 6.777009294813612e-06,
      "loss": 0.3763,
      "step": 25931
    },
    {
      "epoch": 1.1900325822587308,
      "grad_norm": 0.4934622347354889,
      "learning_rate": 6.776780112205209e-06,
      "loss": 0.4006,
      "step": 25932
    },
    {
      "epoch": 1.190078472763985,
      "grad_norm": 0.4944249093532562,
      "learning_rate": 6.77655092532421e-06,
      "loss": 0.3762,
      "step": 25933
    },
    {
      "epoch": 1.1901243632692395,
      "grad_norm": 0.49912315607070923,
      "learning_rate": 6.7763217341711685e-06,
      "loss": 0.3655,
      "step": 25934
    },
    {
      "epoch": 1.190170253774494,
      "grad_norm": 0.4628129005432129,
      "learning_rate": 6.776092538746632e-06,
      "loss": 0.4001,
      "step": 25935
    },
    {
      "epoch": 1.1902161442797485,
      "grad_norm": 0.4517259895801544,
      "learning_rate": 6.775863339051153e-06,
      "loss": 0.3198,
      "step": 25936
    },
    {
      "epoch": 1.190262034785003,
      "grad_norm": 0.5350093841552734,
      "learning_rate": 6.775634135085284e-06,
      "loss": 0.4095,
      "step": 25937
    },
    {
      "epoch": 1.1903079252902575,
      "grad_norm": 0.4601711928844452,
      "learning_rate": 6.775404926849575e-06,
      "loss": 0.3165,
      "step": 25938
    },
    {
      "epoch": 1.190353815795512,
      "grad_norm": 0.48809200525283813,
      "learning_rate": 6.775175714344579e-06,
      "loss": 0.3554,
      "step": 25939
    },
    {
      "epoch": 1.1903997063007663,
      "grad_norm": 0.47363296151161194,
      "learning_rate": 6.774946497570845e-06,
      "loss": 0.3667,
      "step": 25940
    },
    {
      "epoch": 1.1904455968060208,
      "grad_norm": 0.41922226548194885,
      "learning_rate": 6.774717276528923e-06,
      "loss": 0.2897,
      "step": 25941
    },
    {
      "epoch": 1.1904914873112753,
      "grad_norm": 0.47930824756622314,
      "learning_rate": 6.774488051219368e-06,
      "loss": 0.3724,
      "step": 25942
    },
    {
      "epoch": 1.1905373778165298,
      "grad_norm": 0.4527236819267273,
      "learning_rate": 6.774258821642727e-06,
      "loss": 0.3541,
      "step": 25943
    },
    {
      "epoch": 1.1905832683217843,
      "grad_norm": 0.47368934750556946,
      "learning_rate": 6.7740295877995565e-06,
      "loss": 0.3549,
      "step": 25944
    },
    {
      "epoch": 1.1906291588270386,
      "grad_norm": 0.4141766130924225,
      "learning_rate": 6.773800349690404e-06,
      "loss": 0.2789,
      "step": 25945
    },
    {
      "epoch": 1.190675049332293,
      "grad_norm": 0.48650825023651123,
      "learning_rate": 6.77357110731582e-06,
      "loss": 0.3666,
      "step": 25946
    },
    {
      "epoch": 1.1907209398375476,
      "grad_norm": 0.4485582113265991,
      "learning_rate": 6.7733418606763565e-06,
      "loss": 0.3516,
      "step": 25947
    },
    {
      "epoch": 1.190766830342802,
      "grad_norm": 0.5679374933242798,
      "learning_rate": 6.7731126097725685e-06,
      "loss": 0.3928,
      "step": 25948
    },
    {
      "epoch": 1.1908127208480566,
      "grad_norm": 0.4647097885608673,
      "learning_rate": 6.772883354605002e-06,
      "loss": 0.2742,
      "step": 25949
    },
    {
      "epoch": 1.190858611353311,
      "grad_norm": 0.49706825613975525,
      "learning_rate": 6.772654095174212e-06,
      "loss": 0.3798,
      "step": 25950
    },
    {
      "epoch": 1.1909045018585656,
      "grad_norm": 0.46696269512176514,
      "learning_rate": 6.7724248314807485e-06,
      "loss": 0.3478,
      "step": 25951
    },
    {
      "epoch": 1.1909503923638198,
      "grad_norm": 0.4528662860393524,
      "learning_rate": 6.7721955635251615e-06,
      "loss": 0.3582,
      "step": 25952
    },
    {
      "epoch": 1.1909962828690743,
      "grad_norm": 0.44142401218414307,
      "learning_rate": 6.771966291308003e-06,
      "loss": 0.3061,
      "step": 25953
    },
    {
      "epoch": 1.1910421733743288,
      "grad_norm": 0.5052987337112427,
      "learning_rate": 6.771737014829827e-06,
      "loss": 0.4165,
      "step": 25954
    },
    {
      "epoch": 1.1910880638795833,
      "grad_norm": 0.5155503153800964,
      "learning_rate": 6.771507734091182e-06,
      "loss": 0.3781,
      "step": 25955
    },
    {
      "epoch": 1.1911339543848378,
      "grad_norm": 0.47700414061546326,
      "learning_rate": 6.771278449092619e-06,
      "loss": 0.3653,
      "step": 25956
    },
    {
      "epoch": 1.1911798448900923,
      "grad_norm": 0.4596863389015198,
      "learning_rate": 6.77104915983469e-06,
      "loss": 0.3451,
      "step": 25957
    },
    {
      "epoch": 1.1912257353953466,
      "grad_norm": 0.49229907989501953,
      "learning_rate": 6.770819866317948e-06,
      "loss": 0.3955,
      "step": 25958
    },
    {
      "epoch": 1.191271625900601,
      "grad_norm": 0.48000815510749817,
      "learning_rate": 6.770590568542943e-06,
      "loss": 0.4009,
      "step": 25959
    },
    {
      "epoch": 1.1913175164058556,
      "grad_norm": 0.4557708501815796,
      "learning_rate": 6.770361266510226e-06,
      "loss": 0.3479,
      "step": 25960
    },
    {
      "epoch": 1.19136340691111,
      "grad_norm": 0.4382645785808563,
      "learning_rate": 6.770131960220349e-06,
      "loss": 0.3123,
      "step": 25961
    },
    {
      "epoch": 1.1914092974163646,
      "grad_norm": 0.4661819040775299,
      "learning_rate": 6.769902649673864e-06,
      "loss": 0.3936,
      "step": 25962
    },
    {
      "epoch": 1.191455187921619,
      "grad_norm": 0.4923458397388458,
      "learning_rate": 6.7696733348713206e-06,
      "loss": 0.4192,
      "step": 25963
    },
    {
      "epoch": 1.1915010784268736,
      "grad_norm": 0.4868899881839752,
      "learning_rate": 6.769444015813271e-06,
      "loss": 0.3863,
      "step": 25964
    },
    {
      "epoch": 1.1915469689321279,
      "grad_norm": 0.4725935161113739,
      "learning_rate": 6.7692146925002685e-06,
      "loss": 0.414,
      "step": 25965
    },
    {
      "epoch": 1.1915928594373824,
      "grad_norm": 0.48157671093940735,
      "learning_rate": 6.7689853649328625e-06,
      "loss": 0.4284,
      "step": 25966
    },
    {
      "epoch": 1.1916387499426369,
      "grad_norm": 0.441914439201355,
      "learning_rate": 6.7687560331116054e-06,
      "loss": 0.3121,
      "step": 25967
    },
    {
      "epoch": 1.1916846404478914,
      "grad_norm": 0.4885229170322418,
      "learning_rate": 6.768526697037048e-06,
      "loss": 0.3997,
      "step": 25968
    },
    {
      "epoch": 1.1917305309531459,
      "grad_norm": 0.44718679785728455,
      "learning_rate": 6.768297356709741e-06,
      "loss": 0.285,
      "step": 25969
    },
    {
      "epoch": 1.1917764214584003,
      "grad_norm": 0.48640960454940796,
      "learning_rate": 6.768068012130238e-06,
      "loss": 0.3566,
      "step": 25970
    },
    {
      "epoch": 1.1918223119636546,
      "grad_norm": 0.49467092752456665,
      "learning_rate": 6.767838663299089e-06,
      "loss": 0.4079,
      "step": 25971
    },
    {
      "epoch": 1.1918682024689091,
      "grad_norm": 0.49577146768569946,
      "learning_rate": 6.7676093102168475e-06,
      "loss": 0.4203,
      "step": 25972
    },
    {
      "epoch": 1.1919140929741636,
      "grad_norm": 0.4572511911392212,
      "learning_rate": 6.767379952884061e-06,
      "loss": 0.3519,
      "step": 25973
    },
    {
      "epoch": 1.1919599834794181,
      "grad_norm": 0.49097150564193726,
      "learning_rate": 6.767150591301286e-06,
      "loss": 0.3689,
      "step": 25974
    },
    {
      "epoch": 1.1920058739846726,
      "grad_norm": 0.44907674193382263,
      "learning_rate": 6.766921225469069e-06,
      "loss": 0.3174,
      "step": 25975
    },
    {
      "epoch": 1.1920517644899271,
      "grad_norm": 0.4941725432872772,
      "learning_rate": 6.766691855387965e-06,
      "loss": 0.4432,
      "step": 25976
    },
    {
      "epoch": 1.1920976549951816,
      "grad_norm": 0.46839800477027893,
      "learning_rate": 6.766462481058527e-06,
      "loss": 0.3289,
      "step": 25977
    },
    {
      "epoch": 1.1921435455004359,
      "grad_norm": 0.44031113386154175,
      "learning_rate": 6.766233102481301e-06,
      "loss": 0.3108,
      "step": 25978
    },
    {
      "epoch": 1.1921894360056904,
      "grad_norm": 0.47919028997421265,
      "learning_rate": 6.766003719656844e-06,
      "loss": 0.3907,
      "step": 25979
    },
    {
      "epoch": 1.1922353265109449,
      "grad_norm": 0.4434223771095276,
      "learning_rate": 6.765774332585704e-06,
      "loss": 0.3292,
      "step": 25980
    },
    {
      "epoch": 1.1922812170161994,
      "grad_norm": 0.457846462726593,
      "learning_rate": 6.765544941268433e-06,
      "loss": 0.3299,
      "step": 25981
    },
    {
      "epoch": 1.1923271075214539,
      "grad_norm": 0.42508599162101746,
      "learning_rate": 6.765315545705586e-06,
      "loss": 0.2625,
      "step": 25982
    },
    {
      "epoch": 1.1923729980267082,
      "grad_norm": 0.5122770667076111,
      "learning_rate": 6.7650861458977125e-06,
      "loss": 0.3031,
      "step": 25983
    },
    {
      "epoch": 1.1924188885319627,
      "grad_norm": 0.4588368535041809,
      "learning_rate": 6.764856741845362e-06,
      "loss": 0.3276,
      "step": 25984
    },
    {
      "epoch": 1.1924647790372171,
      "grad_norm": 0.4710131585597992,
      "learning_rate": 6.7646273335490895e-06,
      "loss": 0.3565,
      "step": 25985
    },
    {
      "epoch": 1.1925106695424716,
      "grad_norm": 0.4992418587207794,
      "learning_rate": 6.764397921009443e-06,
      "loss": 0.3728,
      "step": 25986
    },
    {
      "epoch": 1.1925565600477261,
      "grad_norm": 0.5704121589660645,
      "learning_rate": 6.764168504226979e-06,
      "loss": 0.3592,
      "step": 25987
    },
    {
      "epoch": 1.1926024505529806,
      "grad_norm": 0.5057510733604431,
      "learning_rate": 6.763939083202247e-06,
      "loss": 0.4246,
      "step": 25988
    },
    {
      "epoch": 1.1926483410582351,
      "grad_norm": 0.5328572392463684,
      "learning_rate": 6.763709657935795e-06,
      "loss": 0.381,
      "step": 25989
    },
    {
      "epoch": 1.1926942315634896,
      "grad_norm": 0.4945255517959595,
      "learning_rate": 6.763480228428179e-06,
      "loss": 0.4128,
      "step": 25990
    },
    {
      "epoch": 1.192740122068744,
      "grad_norm": 0.5088258385658264,
      "learning_rate": 6.763250794679951e-06,
      "loss": 0.408,
      "step": 25991
    },
    {
      "epoch": 1.1927860125739984,
      "grad_norm": 0.45027729868888855,
      "learning_rate": 6.76302135669166e-06,
      "loss": 0.3003,
      "step": 25992
    },
    {
      "epoch": 1.192831903079253,
      "grad_norm": 0.5133593678474426,
      "learning_rate": 6.76279191446386e-06,
      "loss": 0.4048,
      "step": 25993
    },
    {
      "epoch": 1.1928777935845074,
      "grad_norm": 0.42958834767341614,
      "learning_rate": 6.762562467997102e-06,
      "loss": 0.3216,
      "step": 25994
    },
    {
      "epoch": 1.192923684089762,
      "grad_norm": 0.48135435581207275,
      "learning_rate": 6.762333017291937e-06,
      "loss": 0.4292,
      "step": 25995
    },
    {
      "epoch": 1.1929695745950162,
      "grad_norm": 0.4664795994758606,
      "learning_rate": 6.7621035623489175e-06,
      "loss": 0.3474,
      "step": 25996
    },
    {
      "epoch": 1.1930154651002707,
      "grad_norm": 0.5002181529998779,
      "learning_rate": 6.7618741031685945e-06,
      "loss": 0.3412,
      "step": 25997
    },
    {
      "epoch": 1.1930613556055252,
      "grad_norm": 0.4589065909385681,
      "learning_rate": 6.761644639751521e-06,
      "loss": 0.3443,
      "step": 25998
    },
    {
      "epoch": 1.1931072461107797,
      "grad_norm": 0.5084084272384644,
      "learning_rate": 6.761415172098248e-06,
      "loss": 0.391,
      "step": 25999
    },
    {
      "epoch": 1.1931531366160342,
      "grad_norm": 0.462502658367157,
      "learning_rate": 6.761185700209327e-06,
      "loss": 0.3925,
      "step": 26000
    },
    {
      "epoch": 1.1931990271212887,
      "grad_norm": 0.504973292350769,
      "learning_rate": 6.7609562240853095e-06,
      "loss": 0.415,
      "step": 26001
    },
    {
      "epoch": 1.1932449176265432,
      "grad_norm": 0.4426838159561157,
      "learning_rate": 6.76072674372675e-06,
      "loss": 0.3282,
      "step": 26002
    },
    {
      "epoch": 1.1932908081317974,
      "grad_norm": 0.45195814967155457,
      "learning_rate": 6.7604972591341965e-06,
      "loss": 0.3168,
      "step": 26003
    },
    {
      "epoch": 1.193336698637052,
      "grad_norm": 0.5188601016998291,
      "learning_rate": 6.760267770308204e-06,
      "loss": 0.3947,
      "step": 26004
    },
    {
      "epoch": 1.1933825891423064,
      "grad_norm": 0.4773806631565094,
      "learning_rate": 6.760038277249324e-06,
      "loss": 0.4064,
      "step": 26005
    },
    {
      "epoch": 1.193428479647561,
      "grad_norm": 0.4679964780807495,
      "learning_rate": 6.759808779958105e-06,
      "loss": 0.3622,
      "step": 26006
    },
    {
      "epoch": 1.1934743701528154,
      "grad_norm": 0.4745357632637024,
      "learning_rate": 6.759579278435102e-06,
      "loss": 0.3311,
      "step": 26007
    },
    {
      "epoch": 1.19352026065807,
      "grad_norm": 0.45525866746902466,
      "learning_rate": 6.759349772680867e-06,
      "loss": 0.3717,
      "step": 26008
    },
    {
      "epoch": 1.1935661511633242,
      "grad_norm": 0.44895413517951965,
      "learning_rate": 6.7591202626959505e-06,
      "loss": 0.2865,
      "step": 26009
    },
    {
      "epoch": 1.1936120416685787,
      "grad_norm": 0.46721509099006653,
      "learning_rate": 6.758890748480906e-06,
      "loss": 0.3572,
      "step": 26010
    },
    {
      "epoch": 1.1936579321738332,
      "grad_norm": 0.49448081851005554,
      "learning_rate": 6.758661230036283e-06,
      "loss": 0.4002,
      "step": 26011
    },
    {
      "epoch": 1.1937038226790877,
      "grad_norm": 0.4414003789424896,
      "learning_rate": 6.758431707362635e-06,
      "loss": 0.3598,
      "step": 26012
    },
    {
      "epoch": 1.1937497131843422,
      "grad_norm": 0.46360763907432556,
      "learning_rate": 6.758202180460515e-06,
      "loss": 0.3176,
      "step": 26013
    },
    {
      "epoch": 1.1937956036895967,
      "grad_norm": 0.5055378079414368,
      "learning_rate": 6.757972649330471e-06,
      "loss": 0.3908,
      "step": 26014
    },
    {
      "epoch": 1.1938414941948512,
      "grad_norm": 0.4641452729701996,
      "learning_rate": 6.757743113973061e-06,
      "loss": 0.3369,
      "step": 26015
    },
    {
      "epoch": 1.1938873847001055,
      "grad_norm": 0.5657639503479004,
      "learning_rate": 6.757513574388832e-06,
      "loss": 0.3803,
      "step": 26016
    },
    {
      "epoch": 1.19393327520536,
      "grad_norm": 0.4505504071712494,
      "learning_rate": 6.7572840305783375e-06,
      "loss": 0.2996,
      "step": 26017
    },
    {
      "epoch": 1.1939791657106145,
      "grad_norm": 0.4757719337940216,
      "learning_rate": 6.757054482542128e-06,
      "loss": 0.3505,
      "step": 26018
    },
    {
      "epoch": 1.194025056215869,
      "grad_norm": 0.5091188549995422,
      "learning_rate": 6.756824930280759e-06,
      "loss": 0.4429,
      "step": 26019
    },
    {
      "epoch": 1.1940709467211235,
      "grad_norm": 0.44546669721603394,
      "learning_rate": 6.756595373794781e-06,
      "loss": 0.3109,
      "step": 26020
    },
    {
      "epoch": 1.194116837226378,
      "grad_norm": 0.43127334117889404,
      "learning_rate": 6.756365813084743e-06,
      "loss": 0.3169,
      "step": 26021
    },
    {
      "epoch": 1.1941627277316322,
      "grad_norm": 0.4896860718727112,
      "learning_rate": 6.756136248151202e-06,
      "loss": 0.3917,
      "step": 26022
    },
    {
      "epoch": 1.1942086182368867,
      "grad_norm": 0.49661147594451904,
      "learning_rate": 6.755906678994706e-06,
      "loss": 0.4013,
      "step": 26023
    },
    {
      "epoch": 1.1942545087421412,
      "grad_norm": 0.511570155620575,
      "learning_rate": 6.75567710561581e-06,
      "loss": 0.38,
      "step": 26024
    },
    {
      "epoch": 1.1943003992473957,
      "grad_norm": 0.43668752908706665,
      "learning_rate": 6.755447528015062e-06,
      "loss": 0.2987,
      "step": 26025
    },
    {
      "epoch": 1.1943462897526502,
      "grad_norm": 0.4372887909412384,
      "learning_rate": 6.75521794619302e-06,
      "loss": 0.3072,
      "step": 26026
    },
    {
      "epoch": 1.1943921802579047,
      "grad_norm": 0.4848904013633728,
      "learning_rate": 6.754988360150232e-06,
      "loss": 0.3867,
      "step": 26027
    },
    {
      "epoch": 1.1944380707631592,
      "grad_norm": 0.4502513110637665,
      "learning_rate": 6.75475876988725e-06,
      "loss": 0.3262,
      "step": 26028
    },
    {
      "epoch": 1.1944839612684135,
      "grad_norm": 0.45882129669189453,
      "learning_rate": 6.754529175404626e-06,
      "loss": 0.3628,
      "step": 26029
    },
    {
      "epoch": 1.194529851773668,
      "grad_norm": 0.47526127099990845,
      "learning_rate": 6.754299576702915e-06,
      "loss": 0.38,
      "step": 26030
    },
    {
      "epoch": 1.1945757422789225,
      "grad_norm": 0.4699185788631439,
      "learning_rate": 6.7540699737826666e-06,
      "loss": 0.3601,
      "step": 26031
    },
    {
      "epoch": 1.194621632784177,
      "grad_norm": 0.4853813946247101,
      "learning_rate": 6.753840366644433e-06,
      "loss": 0.3536,
      "step": 26032
    },
    {
      "epoch": 1.1946675232894315,
      "grad_norm": 0.536471962928772,
      "learning_rate": 6.753610755288769e-06,
      "loss": 0.4271,
      "step": 26033
    },
    {
      "epoch": 1.1947134137946858,
      "grad_norm": 0.4430462718009949,
      "learning_rate": 6.753381139716222e-06,
      "loss": 0.338,
      "step": 26034
    },
    {
      "epoch": 1.1947593042999403,
      "grad_norm": 0.4737832546234131,
      "learning_rate": 6.753151519927347e-06,
      "loss": 0.3228,
      "step": 26035
    },
    {
      "epoch": 1.1948051948051948,
      "grad_norm": 0.49070560932159424,
      "learning_rate": 6.752921895922697e-06,
      "loss": 0.4437,
      "step": 26036
    },
    {
      "epoch": 1.1948510853104493,
      "grad_norm": 0.529140055179596,
      "learning_rate": 6.752692267702823e-06,
      "loss": 0.468,
      "step": 26037
    },
    {
      "epoch": 1.1948969758157038,
      "grad_norm": 0.4984104633331299,
      "learning_rate": 6.752462635268275e-06,
      "loss": 0.3946,
      "step": 26038
    },
    {
      "epoch": 1.1949428663209583,
      "grad_norm": 0.4992067813873291,
      "learning_rate": 6.752232998619611e-06,
      "loss": 0.4241,
      "step": 26039
    },
    {
      "epoch": 1.1949887568262128,
      "grad_norm": 0.4722709059715271,
      "learning_rate": 6.752003357757377e-06,
      "loss": 0.3157,
      "step": 26040
    },
    {
      "epoch": 1.195034647331467,
      "grad_norm": 0.45488083362579346,
      "learning_rate": 6.751773712682129e-06,
      "loss": 0.3291,
      "step": 26041
    },
    {
      "epoch": 1.1950805378367215,
      "grad_norm": 0.44097647070884705,
      "learning_rate": 6.75154406339442e-06,
      "loss": 0.3208,
      "step": 26042
    },
    {
      "epoch": 1.195126428341976,
      "grad_norm": 0.4716513156890869,
      "learning_rate": 6.751314409894797e-06,
      "loss": 0.352,
      "step": 26043
    },
    {
      "epoch": 1.1951723188472305,
      "grad_norm": 0.5054554343223572,
      "learning_rate": 6.751084752183817e-06,
      "loss": 0.4051,
      "step": 26044
    },
    {
      "epoch": 1.195218209352485,
      "grad_norm": 0.5112261772155762,
      "learning_rate": 6.750855090262031e-06,
      "loss": 0.4042,
      "step": 26045
    },
    {
      "epoch": 1.1952640998577395,
      "grad_norm": 0.44552770256996155,
      "learning_rate": 6.750625424129989e-06,
      "loss": 0.324,
      "step": 26046
    },
    {
      "epoch": 1.1953099903629938,
      "grad_norm": 0.4912937581539154,
      "learning_rate": 6.750395753788248e-06,
      "loss": 0.4464,
      "step": 26047
    },
    {
      "epoch": 1.1953558808682483,
      "grad_norm": 0.4276401698589325,
      "learning_rate": 6.750166079237357e-06,
      "loss": 0.2715,
      "step": 26048
    },
    {
      "epoch": 1.1954017713735028,
      "grad_norm": 0.4514125883579254,
      "learning_rate": 6.749936400477869e-06,
      "loss": 0.3211,
      "step": 26049
    },
    {
      "epoch": 1.1954476618787573,
      "grad_norm": 0.48245614767074585,
      "learning_rate": 6.749706717510336e-06,
      "loss": 0.3617,
      "step": 26050
    },
    {
      "epoch": 1.1954935523840118,
      "grad_norm": 0.44183382391929626,
      "learning_rate": 6.749477030335309e-06,
      "loss": 0.3156,
      "step": 26051
    },
    {
      "epoch": 1.1955394428892663,
      "grad_norm": 0.4564642012119293,
      "learning_rate": 6.749247338953344e-06,
      "loss": 0.3091,
      "step": 26052
    },
    {
      "epoch": 1.1955853333945208,
      "grad_norm": 0.453042209148407,
      "learning_rate": 6.749017643364992e-06,
      "loss": 0.348,
      "step": 26053
    },
    {
      "epoch": 1.195631223899775,
      "grad_norm": 0.4675394296646118,
      "learning_rate": 6.7487879435708024e-06,
      "loss": 0.3484,
      "step": 26054
    },
    {
      "epoch": 1.1956771144050296,
      "grad_norm": 0.4503262937068939,
      "learning_rate": 6.74855823957133e-06,
      "loss": 0.3312,
      "step": 26055
    },
    {
      "epoch": 1.195723004910284,
      "grad_norm": 0.4758007824420929,
      "learning_rate": 6.748328531367128e-06,
      "loss": 0.3971,
      "step": 26056
    },
    {
      "epoch": 1.1957688954155385,
      "grad_norm": 0.4598787724971771,
      "learning_rate": 6.748098818958747e-06,
      "loss": 0.3515,
      "step": 26057
    },
    {
      "epoch": 1.195814785920793,
      "grad_norm": 0.424710750579834,
      "learning_rate": 6.7478691023467416e-06,
      "loss": 0.2848,
      "step": 26058
    },
    {
      "epoch": 1.1958606764260475,
      "grad_norm": 0.5040894746780396,
      "learning_rate": 6.7476393815316606e-06,
      "loss": 0.3673,
      "step": 26059
    },
    {
      "epoch": 1.1959065669313018,
      "grad_norm": 0.48820754885673523,
      "learning_rate": 6.747409656514059e-06,
      "loss": 0.3732,
      "step": 26060
    },
    {
      "epoch": 1.1959524574365563,
      "grad_norm": 0.4702054560184479,
      "learning_rate": 6.74717992729449e-06,
      "loss": 0.3873,
      "step": 26061
    },
    {
      "epoch": 1.1959983479418108,
      "grad_norm": 0.4773827791213989,
      "learning_rate": 6.746950193873503e-06,
      "loss": 0.3799,
      "step": 26062
    },
    {
      "epoch": 1.1960442384470653,
      "grad_norm": 0.5489702820777893,
      "learning_rate": 6.746720456251652e-06,
      "loss": 0.4463,
      "step": 26063
    },
    {
      "epoch": 1.1960901289523198,
      "grad_norm": 0.45339131355285645,
      "learning_rate": 6.746490714429491e-06,
      "loss": 0.3342,
      "step": 26064
    },
    {
      "epoch": 1.1961360194575743,
      "grad_norm": 0.47626203298568726,
      "learning_rate": 6.746260968407569e-06,
      "loss": 0.3506,
      "step": 26065
    },
    {
      "epoch": 1.1961819099628288,
      "grad_norm": 0.47827503085136414,
      "learning_rate": 6.746031218186442e-06,
      "loss": 0.3896,
      "step": 26066
    },
    {
      "epoch": 1.196227800468083,
      "grad_norm": 0.4923810362815857,
      "learning_rate": 6.745801463766661e-06,
      "loss": 0.3848,
      "step": 26067
    },
    {
      "epoch": 1.1962736909733376,
      "grad_norm": 0.5078055262565613,
      "learning_rate": 6.7455717051487775e-06,
      "loss": 0.4216,
      "step": 26068
    },
    {
      "epoch": 1.196319581478592,
      "grad_norm": 0.45491594076156616,
      "learning_rate": 6.745341942333347e-06,
      "loss": 0.3477,
      "step": 26069
    },
    {
      "epoch": 1.1963654719838466,
      "grad_norm": 0.4678437113761902,
      "learning_rate": 6.7451121753209194e-06,
      "loss": 0.3441,
      "step": 26070
    },
    {
      "epoch": 1.196411362489101,
      "grad_norm": 0.4935905635356903,
      "learning_rate": 6.744882404112046e-06,
      "loss": 0.3586,
      "step": 26071
    },
    {
      "epoch": 1.1964572529943553,
      "grad_norm": 0.42168205976486206,
      "learning_rate": 6.744652628707282e-06,
      "loss": 0.2881,
      "step": 26072
    },
    {
      "epoch": 1.1965031434996098,
      "grad_norm": 0.4519284665584564,
      "learning_rate": 6.74442284910718e-06,
      "loss": 0.3729,
      "step": 26073
    },
    {
      "epoch": 1.1965490340048643,
      "grad_norm": 0.5226909518241882,
      "learning_rate": 6.74419306531229e-06,
      "loss": 0.3775,
      "step": 26074
    },
    {
      "epoch": 1.1965949245101188,
      "grad_norm": 0.4678846299648285,
      "learning_rate": 6.743963277323169e-06,
      "loss": 0.3218,
      "step": 26075
    },
    {
      "epoch": 1.1966408150153733,
      "grad_norm": 0.4406469762325287,
      "learning_rate": 6.743733485140363e-06,
      "loss": 0.2907,
      "step": 26076
    },
    {
      "epoch": 1.1966867055206278,
      "grad_norm": 0.5737265348434448,
      "learning_rate": 6.74350368876443e-06,
      "loss": 0.416,
      "step": 26077
    },
    {
      "epoch": 1.1967325960258823,
      "grad_norm": 0.4920317232608795,
      "learning_rate": 6.743273888195922e-06,
      "loss": 0.367,
      "step": 26078
    },
    {
      "epoch": 1.1967784865311368,
      "grad_norm": 0.4715070128440857,
      "learning_rate": 6.743044083435388e-06,
      "loss": 0.4005,
      "step": 26079
    },
    {
      "epoch": 1.196824377036391,
      "grad_norm": 0.432295024394989,
      "learning_rate": 6.742814274483385e-06,
      "loss": 0.3024,
      "step": 26080
    },
    {
      "epoch": 1.1968702675416456,
      "grad_norm": 0.49165648221969604,
      "learning_rate": 6.742584461340465e-06,
      "loss": 0.4204,
      "step": 26081
    },
    {
      "epoch": 1.1969161580469,
      "grad_norm": 0.450988233089447,
      "learning_rate": 6.742354644007176e-06,
      "loss": 0.3148,
      "step": 26082
    },
    {
      "epoch": 1.1969620485521546,
      "grad_norm": 0.4914509952068329,
      "learning_rate": 6.742124822484075e-06,
      "loss": 0.3889,
      "step": 26083
    },
    {
      "epoch": 1.197007939057409,
      "grad_norm": 0.46313297748565674,
      "learning_rate": 6.741894996771714e-06,
      "loss": 0.3699,
      "step": 26084
    },
    {
      "epoch": 1.1970538295626634,
      "grad_norm": 0.4792465567588806,
      "learning_rate": 6.741665166870646e-06,
      "loss": 0.3049,
      "step": 26085
    },
    {
      "epoch": 1.1970997200679179,
      "grad_norm": 0.4526558518409729,
      "learning_rate": 6.741435332781423e-06,
      "loss": 0.3691,
      "step": 26086
    },
    {
      "epoch": 1.1971456105731724,
      "grad_norm": 0.4668537676334381,
      "learning_rate": 6.741205494504597e-06,
      "loss": 0.3487,
      "step": 26087
    },
    {
      "epoch": 1.1971915010784269,
      "grad_norm": 0.4602303206920624,
      "learning_rate": 6.740975652040721e-06,
      "loss": 0.3887,
      "step": 26088
    },
    {
      "epoch": 1.1972373915836814,
      "grad_norm": 0.4525696933269501,
      "learning_rate": 6.740745805390349e-06,
      "loss": 0.3406,
      "step": 26089
    },
    {
      "epoch": 1.1972832820889359,
      "grad_norm": 0.4667637050151825,
      "learning_rate": 6.7405159545540325e-06,
      "loss": 0.3631,
      "step": 26090
    },
    {
      "epoch": 1.1973291725941904,
      "grad_norm": 0.4322887659072876,
      "learning_rate": 6.740286099532324e-06,
      "loss": 0.302,
      "step": 26091
    },
    {
      "epoch": 1.1973750630994446,
      "grad_norm": 0.427450031042099,
      "learning_rate": 6.740056240325776e-06,
      "loss": 0.306,
      "step": 26092
    },
    {
      "epoch": 1.1974209536046991,
      "grad_norm": 0.48566752672195435,
      "learning_rate": 6.739826376934943e-06,
      "loss": 0.4197,
      "step": 26093
    },
    {
      "epoch": 1.1974668441099536,
      "grad_norm": 0.46061134338378906,
      "learning_rate": 6.739596509360376e-06,
      "loss": 0.3351,
      "step": 26094
    },
    {
      "epoch": 1.1975127346152081,
      "grad_norm": 0.4719395935535431,
      "learning_rate": 6.739366637602629e-06,
      "loss": 0.3298,
      "step": 26095
    },
    {
      "epoch": 1.1975586251204626,
      "grad_norm": 0.46186017990112305,
      "learning_rate": 6.739136761662254e-06,
      "loss": 0.3901,
      "step": 26096
    },
    {
      "epoch": 1.1976045156257171,
      "grad_norm": 0.46859508752822876,
      "learning_rate": 6.738906881539804e-06,
      "loss": 0.3761,
      "step": 26097
    },
    {
      "epoch": 1.1976504061309714,
      "grad_norm": 0.46848782896995544,
      "learning_rate": 6.7386769972358324e-06,
      "loss": 0.3675,
      "step": 26098
    },
    {
      "epoch": 1.197696296636226,
      "grad_norm": 0.47379419207572937,
      "learning_rate": 6.7384471087508896e-06,
      "loss": 0.4077,
      "step": 26099
    },
    {
      "epoch": 1.1977421871414804,
      "grad_norm": 0.4640737473964691,
      "learning_rate": 6.738217216085532e-06,
      "loss": 0.3888,
      "step": 26100
    },
    {
      "epoch": 1.197788077646735,
      "grad_norm": 0.46600186824798584,
      "learning_rate": 6.73798731924031e-06,
      "loss": 0.3275,
      "step": 26101
    },
    {
      "epoch": 1.1978339681519894,
      "grad_norm": 0.453753799200058,
      "learning_rate": 6.737757418215777e-06,
      "loss": 0.3375,
      "step": 26102
    },
    {
      "epoch": 1.197879858657244,
      "grad_norm": 0.481778085231781,
      "learning_rate": 6.737527513012485e-06,
      "loss": 0.3627,
      "step": 26103
    },
    {
      "epoch": 1.1979257491624984,
      "grad_norm": 0.4675348997116089,
      "learning_rate": 6.737297603630989e-06,
      "loss": 0.3111,
      "step": 26104
    },
    {
      "epoch": 1.1979716396677527,
      "grad_norm": 0.45217376947402954,
      "learning_rate": 6.73706769007184e-06,
      "loss": 0.3208,
      "step": 26105
    },
    {
      "epoch": 1.1980175301730072,
      "grad_norm": 0.4904526472091675,
      "learning_rate": 6.736837772335592e-06,
      "loss": 0.3744,
      "step": 26106
    },
    {
      "epoch": 1.1980634206782617,
      "grad_norm": 0.47067174315452576,
      "learning_rate": 6.736607850422796e-06,
      "loss": 0.3475,
      "step": 26107
    },
    {
      "epoch": 1.1981093111835162,
      "grad_norm": 0.4618355631828308,
      "learning_rate": 6.7363779243340075e-06,
      "loss": 0.369,
      "step": 26108
    },
    {
      "epoch": 1.1981552016887707,
      "grad_norm": 0.47348490357398987,
      "learning_rate": 6.736147994069779e-06,
      "loss": 0.4052,
      "step": 26109
    },
    {
      "epoch": 1.1982010921940252,
      "grad_norm": 0.4952775239944458,
      "learning_rate": 6.73591805963066e-06,
      "loss": 0.3741,
      "step": 26110
    },
    {
      "epoch": 1.1982469826992794,
      "grad_norm": 0.43276041746139526,
      "learning_rate": 6.735688121017207e-06,
      "loss": 0.3345,
      "step": 26111
    },
    {
      "epoch": 1.198292873204534,
      "grad_norm": 0.47764596343040466,
      "learning_rate": 6.735458178229972e-06,
      "loss": 0.3609,
      "step": 26112
    },
    {
      "epoch": 1.1983387637097884,
      "grad_norm": 0.49317577481269836,
      "learning_rate": 6.735228231269508e-06,
      "loss": 0.4288,
      "step": 26113
    },
    {
      "epoch": 1.198384654215043,
      "grad_norm": 0.48359501361846924,
      "learning_rate": 6.734998280136367e-06,
      "loss": 0.3724,
      "step": 26114
    },
    {
      "epoch": 1.1984305447202974,
      "grad_norm": 0.48468223214149475,
      "learning_rate": 6.734768324831104e-06,
      "loss": 0.3476,
      "step": 26115
    },
    {
      "epoch": 1.198476435225552,
      "grad_norm": 0.5027400851249695,
      "learning_rate": 6.734538365354269e-06,
      "loss": 0.4159,
      "step": 26116
    },
    {
      "epoch": 1.1985223257308064,
      "grad_norm": 0.45357784628868103,
      "learning_rate": 6.734308401706417e-06,
      "loss": 0.3586,
      "step": 26117
    },
    {
      "epoch": 1.1985682162360607,
      "grad_norm": 0.47452646493911743,
      "learning_rate": 6.734078433888102e-06,
      "loss": 0.3431,
      "step": 26118
    },
    {
      "epoch": 1.1986141067413152,
      "grad_norm": 0.4654628336429596,
      "learning_rate": 6.733848461899875e-06,
      "loss": 0.3542,
      "step": 26119
    },
    {
      "epoch": 1.1986599972465697,
      "grad_norm": 0.5092141032218933,
      "learning_rate": 6.733618485742289e-06,
      "loss": 0.4245,
      "step": 26120
    },
    {
      "epoch": 1.1987058877518242,
      "grad_norm": 0.5760002732276917,
      "learning_rate": 6.733388505415899e-06,
      "loss": 0.447,
      "step": 26121
    },
    {
      "epoch": 1.1987517782570787,
      "grad_norm": 0.4981926679611206,
      "learning_rate": 6.7331585209212555e-06,
      "loss": 0.4144,
      "step": 26122
    },
    {
      "epoch": 1.198797668762333,
      "grad_norm": 0.4596744775772095,
      "learning_rate": 6.7329285322589135e-06,
      "loss": 0.2877,
      "step": 26123
    },
    {
      "epoch": 1.1988435592675875,
      "grad_norm": 0.4607294201850891,
      "learning_rate": 6.732698539429425e-06,
      "loss": 0.3885,
      "step": 26124
    },
    {
      "epoch": 1.198889449772842,
      "grad_norm": 0.4720578193664551,
      "learning_rate": 6.732468542433344e-06,
      "loss": 0.3805,
      "step": 26125
    },
    {
      "epoch": 1.1989353402780965,
      "grad_norm": 0.4957895576953888,
      "learning_rate": 6.732238541271222e-06,
      "loss": 0.3739,
      "step": 26126
    },
    {
      "epoch": 1.198981230783351,
      "grad_norm": 0.48680728673934937,
      "learning_rate": 6.732008535943613e-06,
      "loss": 0.409,
      "step": 26127
    },
    {
      "epoch": 1.1990271212886054,
      "grad_norm": 0.4790697395801544,
      "learning_rate": 6.73177852645107e-06,
      "loss": 0.3344,
      "step": 26128
    },
    {
      "epoch": 1.19907301179386,
      "grad_norm": 0.4455365538597107,
      "learning_rate": 6.731548512794147e-06,
      "loss": 0.3461,
      "step": 26129
    },
    {
      "epoch": 1.1991189022991142,
      "grad_norm": 0.4932865500450134,
      "learning_rate": 6.7313184949733965e-06,
      "loss": 0.3814,
      "step": 26130
    },
    {
      "epoch": 1.1991647928043687,
      "grad_norm": 0.47092384099960327,
      "learning_rate": 6.731088472989371e-06,
      "loss": 0.344,
      "step": 26131
    },
    {
      "epoch": 1.1992106833096232,
      "grad_norm": 0.4598398506641388,
      "learning_rate": 6.730858446842624e-06,
      "loss": 0.3704,
      "step": 26132
    },
    {
      "epoch": 1.1992565738148777,
      "grad_norm": 0.5634466409683228,
      "learning_rate": 6.730628416533709e-06,
      "loss": 0.5104,
      "step": 26133
    },
    {
      "epoch": 1.1993024643201322,
      "grad_norm": 0.48962754011154175,
      "learning_rate": 6.730398382063179e-06,
      "loss": 0.3753,
      "step": 26134
    },
    {
      "epoch": 1.1993483548253867,
      "grad_norm": 0.45252367854118347,
      "learning_rate": 6.730168343431587e-06,
      "loss": 0.3523,
      "step": 26135
    },
    {
      "epoch": 1.199394245330641,
      "grad_norm": 0.5442420244216919,
      "learning_rate": 6.729938300639485e-06,
      "loss": 0.3365,
      "step": 26136
    },
    {
      "epoch": 1.1994401358358955,
      "grad_norm": 0.46540549397468567,
      "learning_rate": 6.729708253687428e-06,
      "loss": 0.3252,
      "step": 26137
    },
    {
      "epoch": 1.19948602634115,
      "grad_norm": 0.46021541953086853,
      "learning_rate": 6.729478202575969e-06,
      "loss": 0.3588,
      "step": 26138
    },
    {
      "epoch": 1.1995319168464045,
      "grad_norm": 0.5093053579330444,
      "learning_rate": 6.72924814730566e-06,
      "loss": 0.3894,
      "step": 26139
    },
    {
      "epoch": 1.199577807351659,
      "grad_norm": 0.5090047121047974,
      "learning_rate": 6.729018087877055e-06,
      "loss": 0.3832,
      "step": 26140
    },
    {
      "epoch": 1.1996236978569135,
      "grad_norm": 0.47606897354125977,
      "learning_rate": 6.728788024290709e-06,
      "loss": 0.3783,
      "step": 26141
    },
    {
      "epoch": 1.199669588362168,
      "grad_norm": 0.5121620893478394,
      "learning_rate": 6.728557956547171e-06,
      "loss": 0.3331,
      "step": 26142
    },
    {
      "epoch": 1.1997154788674222,
      "grad_norm": 0.4909416735172272,
      "learning_rate": 6.728327884646999e-06,
      "loss": 0.3806,
      "step": 26143
    },
    {
      "epoch": 1.1997613693726767,
      "grad_norm": 0.5029606223106384,
      "learning_rate": 6.728097808590741e-06,
      "loss": 0.4086,
      "step": 26144
    },
    {
      "epoch": 1.1998072598779312,
      "grad_norm": 0.5282790660858154,
      "learning_rate": 6.727867728378956e-06,
      "loss": 0.3713,
      "step": 26145
    },
    {
      "epoch": 1.1998531503831857,
      "grad_norm": 0.4807416796684265,
      "learning_rate": 6.727637644012193e-06,
      "loss": 0.3749,
      "step": 26146
    },
    {
      "epoch": 1.1998990408884402,
      "grad_norm": 0.4605295658111572,
      "learning_rate": 6.727407555491007e-06,
      "loss": 0.3555,
      "step": 26147
    },
    {
      "epoch": 1.1999449313936947,
      "grad_norm": 0.5081165432929993,
      "learning_rate": 6.7271774628159506e-06,
      "loss": 0.325,
      "step": 26148
    },
    {
      "epoch": 1.199990821898949,
      "grad_norm": 0.4737563729286194,
      "learning_rate": 6.726947365987578e-06,
      "loss": 0.3631,
      "step": 26149
    },
    {
      "epoch": 1.2000367124042035,
      "grad_norm": 0.5027435421943665,
      "learning_rate": 6.7267172650064414e-06,
      "loss": 0.4103,
      "step": 26150
    },
    {
      "epoch": 1.200082602909458,
      "grad_norm": 0.4983493983745575,
      "learning_rate": 6.7264871598730966e-06,
      "loss": 0.4112,
      "step": 26151
    },
    {
      "epoch": 1.2001284934147125,
      "grad_norm": 0.4430161416530609,
      "learning_rate": 6.726257050588094e-06,
      "loss": 0.3158,
      "step": 26152
    },
    {
      "epoch": 1.200174383919967,
      "grad_norm": 0.48199135065078735,
      "learning_rate": 6.726026937151987e-06,
      "loss": 0.3883,
      "step": 26153
    },
    {
      "epoch": 1.2002202744252215,
      "grad_norm": 0.46625375747680664,
      "learning_rate": 6.72579681956533e-06,
      "loss": 0.3563,
      "step": 26154
    },
    {
      "epoch": 1.200266164930476,
      "grad_norm": 0.49773257970809937,
      "learning_rate": 6.725566697828678e-06,
      "loss": 0.398,
      "step": 26155
    },
    {
      "epoch": 1.2003120554357303,
      "grad_norm": 0.5204520225524902,
      "learning_rate": 6.725336571942582e-06,
      "loss": 0.3651,
      "step": 26156
    },
    {
      "epoch": 1.2003579459409848,
      "grad_norm": 0.46030673384666443,
      "learning_rate": 6.725106441907596e-06,
      "loss": 0.3302,
      "step": 26157
    },
    {
      "epoch": 1.2004038364462393,
      "grad_norm": 0.4558969736099243,
      "learning_rate": 6.724876307724273e-06,
      "loss": 0.3558,
      "step": 26158
    },
    {
      "epoch": 1.2004497269514938,
      "grad_norm": 0.49725109338760376,
      "learning_rate": 6.7246461693931665e-06,
      "loss": 0.4036,
      "step": 26159
    },
    {
      "epoch": 1.2004956174567483,
      "grad_norm": 0.4795331358909607,
      "learning_rate": 6.724416026914832e-06,
      "loss": 0.4136,
      "step": 26160
    },
    {
      "epoch": 1.2005415079620025,
      "grad_norm": 0.4314914047718048,
      "learning_rate": 6.7241858802898195e-06,
      "loss": 0.323,
      "step": 26161
    },
    {
      "epoch": 1.200587398467257,
      "grad_norm": 0.47409364581108093,
      "learning_rate": 6.723955729518685e-06,
      "loss": 0.4046,
      "step": 26162
    },
    {
      "epoch": 1.2006332889725115,
      "grad_norm": 0.45685744285583496,
      "learning_rate": 6.7237255746019816e-06,
      "loss": 0.3049,
      "step": 26163
    },
    {
      "epoch": 1.200679179477766,
      "grad_norm": 0.42924290895462036,
      "learning_rate": 6.7234954155402606e-06,
      "loss": 0.2898,
      "step": 26164
    },
    {
      "epoch": 1.2007250699830205,
      "grad_norm": 0.44786280393600464,
      "learning_rate": 6.723265252334077e-06,
      "loss": 0.3286,
      "step": 26165
    },
    {
      "epoch": 1.200770960488275,
      "grad_norm": 0.4342515170574188,
      "learning_rate": 6.723035084983987e-06,
      "loss": 0.3151,
      "step": 26166
    },
    {
      "epoch": 1.2008168509935295,
      "grad_norm": 0.4713689088821411,
      "learning_rate": 6.722804913490539e-06,
      "loss": 0.4153,
      "step": 26167
    },
    {
      "epoch": 1.200862741498784,
      "grad_norm": 0.4732412099838257,
      "learning_rate": 6.7225747378542894e-06,
      "loss": 0.3737,
      "step": 26168
    },
    {
      "epoch": 1.2009086320040383,
      "grad_norm": 0.4564620554447174,
      "learning_rate": 6.722344558075792e-06,
      "loss": 0.3229,
      "step": 26169
    },
    {
      "epoch": 1.2009545225092928,
      "grad_norm": 0.5115981101989746,
      "learning_rate": 6.7221143741555985e-06,
      "loss": 0.4327,
      "step": 26170
    },
    {
      "epoch": 1.2010004130145473,
      "grad_norm": 0.4855562746524811,
      "learning_rate": 6.7218841860942654e-06,
      "loss": 0.4292,
      "step": 26171
    },
    {
      "epoch": 1.2010463035198018,
      "grad_norm": 0.4979616403579712,
      "learning_rate": 6.7216539938923406e-06,
      "loss": 0.3647,
      "step": 26172
    },
    {
      "epoch": 1.2010921940250563,
      "grad_norm": 0.4586527943611145,
      "learning_rate": 6.721423797550385e-06,
      "loss": 0.3281,
      "step": 26173
    },
    {
      "epoch": 1.2011380845303106,
      "grad_norm": 0.40455126762390137,
      "learning_rate": 6.7211935970689464e-06,
      "loss": 0.255,
      "step": 26174
    },
    {
      "epoch": 1.201183975035565,
      "grad_norm": 0.4757690727710724,
      "learning_rate": 6.720963392448581e-06,
      "loss": 0.356,
      "step": 26175
    },
    {
      "epoch": 1.2012298655408196,
      "grad_norm": 0.45172712206840515,
      "learning_rate": 6.720733183689842e-06,
      "loss": 0.3068,
      "step": 26176
    },
    {
      "epoch": 1.201275756046074,
      "grad_norm": 0.48597273230552673,
      "learning_rate": 6.720502970793283e-06,
      "loss": 0.3982,
      "step": 26177
    },
    {
      "epoch": 1.2013216465513286,
      "grad_norm": 0.4706517457962036,
      "learning_rate": 6.720272753759457e-06,
      "loss": 0.356,
      "step": 26178
    },
    {
      "epoch": 1.201367537056583,
      "grad_norm": 0.5427975654602051,
      "learning_rate": 6.720042532588917e-06,
      "loss": 0.43,
      "step": 26179
    },
    {
      "epoch": 1.2014134275618376,
      "grad_norm": 0.4759492874145508,
      "learning_rate": 6.719812307282219e-06,
      "loss": 0.3326,
      "step": 26180
    },
    {
      "epoch": 1.2014593180670918,
      "grad_norm": 0.462455689907074,
      "learning_rate": 6.719582077839915e-06,
      "loss": 0.3886,
      "step": 26181
    },
    {
      "epoch": 1.2015052085723463,
      "grad_norm": 0.4499509632587433,
      "learning_rate": 6.719351844262557e-06,
      "loss": 0.3279,
      "step": 26182
    },
    {
      "epoch": 1.2015510990776008,
      "grad_norm": 0.47346004843711853,
      "learning_rate": 6.719121606550702e-06,
      "loss": 0.3869,
      "step": 26183
    },
    {
      "epoch": 1.2015969895828553,
      "grad_norm": 0.45562854409217834,
      "learning_rate": 6.718891364704902e-06,
      "loss": 0.3497,
      "step": 26184
    },
    {
      "epoch": 1.2016428800881098,
      "grad_norm": 0.4594947397708893,
      "learning_rate": 6.718661118725711e-06,
      "loss": 0.3634,
      "step": 26185
    },
    {
      "epoch": 1.2016887705933643,
      "grad_norm": 0.4732590615749359,
      "learning_rate": 6.7184308686136835e-06,
      "loss": 0.357,
      "step": 26186
    },
    {
      "epoch": 1.2017346610986186,
      "grad_norm": 0.43773582577705383,
      "learning_rate": 6.71820061436937e-06,
      "loss": 0.2915,
      "step": 26187
    },
    {
      "epoch": 1.201780551603873,
      "grad_norm": 0.47648435831069946,
      "learning_rate": 6.717970355993327e-06,
      "loss": 0.4162,
      "step": 26188
    },
    {
      "epoch": 1.2018264421091276,
      "grad_norm": 0.49706488847732544,
      "learning_rate": 6.717740093486109e-06,
      "loss": 0.4175,
      "step": 26189
    },
    {
      "epoch": 1.201872332614382,
      "grad_norm": 0.4869595766067505,
      "learning_rate": 6.7175098268482666e-06,
      "loss": 0.4158,
      "step": 26190
    },
    {
      "epoch": 1.2019182231196366,
      "grad_norm": 0.42260971665382385,
      "learning_rate": 6.717279556080355e-06,
      "loss": 0.2884,
      "step": 26191
    },
    {
      "epoch": 1.201964113624891,
      "grad_norm": 0.49824056029319763,
      "learning_rate": 6.717049281182928e-06,
      "loss": 0.3658,
      "step": 26192
    },
    {
      "epoch": 1.2020100041301456,
      "grad_norm": 0.5208796262741089,
      "learning_rate": 6.716819002156539e-06,
      "loss": 0.4803,
      "step": 26193
    },
    {
      "epoch": 1.2020558946353999,
      "grad_norm": 0.44157856702804565,
      "learning_rate": 6.7165887190017445e-06,
      "loss": 0.2997,
      "step": 26194
    },
    {
      "epoch": 1.2021017851406544,
      "grad_norm": 0.46510204672813416,
      "learning_rate": 6.716358431719095e-06,
      "loss": 0.4025,
      "step": 26195
    },
    {
      "epoch": 1.2021476756459089,
      "grad_norm": 0.5099049210548401,
      "learning_rate": 6.716128140309144e-06,
      "loss": 0.3442,
      "step": 26196
    },
    {
      "epoch": 1.2021935661511634,
      "grad_norm": 0.4632403552532196,
      "learning_rate": 6.715897844772446e-06,
      "loss": 0.3525,
      "step": 26197
    },
    {
      "epoch": 1.2022394566564178,
      "grad_norm": 0.4940589666366577,
      "learning_rate": 6.715667545109556e-06,
      "loss": 0.4055,
      "step": 26198
    },
    {
      "epoch": 1.2022853471616723,
      "grad_norm": 0.4864792227745056,
      "learning_rate": 6.715437241321026e-06,
      "loss": 0.3509,
      "step": 26199
    },
    {
      "epoch": 1.2023312376669266,
      "grad_norm": 0.4595732092857361,
      "learning_rate": 6.7152069334074145e-06,
      "loss": 0.3577,
      "step": 26200
    },
    {
      "epoch": 1.2023771281721811,
      "grad_norm": 0.43776094913482666,
      "learning_rate": 6.714976621369269e-06,
      "loss": 0.3118,
      "step": 26201
    },
    {
      "epoch": 1.2024230186774356,
      "grad_norm": 0.45918309688568115,
      "learning_rate": 6.714746305207146e-06,
      "loss": 0.3461,
      "step": 26202
    },
    {
      "epoch": 1.2024689091826901,
      "grad_norm": 0.5382105112075806,
      "learning_rate": 6.714515984921599e-06,
      "loss": 0.4213,
      "step": 26203
    },
    {
      "epoch": 1.2025147996879446,
      "grad_norm": 0.44428038597106934,
      "learning_rate": 6.714285660513182e-06,
      "loss": 0.3214,
      "step": 26204
    },
    {
      "epoch": 1.2025606901931991,
      "grad_norm": 0.49366116523742676,
      "learning_rate": 6.714055331982451e-06,
      "loss": 0.3924,
      "step": 26205
    },
    {
      "epoch": 1.2026065806984536,
      "grad_norm": 0.4439978301525116,
      "learning_rate": 6.713824999329957e-06,
      "loss": 0.3111,
      "step": 26206
    },
    {
      "epoch": 1.2026524712037079,
      "grad_norm": 0.4891974627971649,
      "learning_rate": 6.713594662556254e-06,
      "loss": 0.3966,
      "step": 26207
    },
    {
      "epoch": 1.2026983617089624,
      "grad_norm": 0.42595887184143066,
      "learning_rate": 6.713364321661898e-06,
      "loss": 0.3179,
      "step": 26208
    },
    {
      "epoch": 1.2027442522142169,
      "grad_norm": 0.4877955913543701,
      "learning_rate": 6.71313397664744e-06,
      "loss": 0.3652,
      "step": 26209
    },
    {
      "epoch": 1.2027901427194714,
      "grad_norm": 0.4850307106971741,
      "learning_rate": 6.712903627513436e-06,
      "loss": 0.4385,
      "step": 26210
    },
    {
      "epoch": 1.2028360332247259,
      "grad_norm": 0.46195217967033386,
      "learning_rate": 6.712673274260441e-06,
      "loss": 0.3334,
      "step": 26211
    },
    {
      "epoch": 1.2028819237299802,
      "grad_norm": 0.46414753794670105,
      "learning_rate": 6.712442916889005e-06,
      "loss": 0.3513,
      "step": 26212
    },
    {
      "epoch": 1.2029278142352346,
      "grad_norm": 0.5122057795524597,
      "learning_rate": 6.712212555399684e-06,
      "loss": 0.4358,
      "step": 26213
    },
    {
      "epoch": 1.2029737047404891,
      "grad_norm": 0.4761491119861603,
      "learning_rate": 6.711982189793034e-06,
      "loss": 0.3571,
      "step": 26214
    },
    {
      "epoch": 1.2030195952457436,
      "grad_norm": 0.5160741209983826,
      "learning_rate": 6.711751820069607e-06,
      "loss": 0.4231,
      "step": 26215
    },
    {
      "epoch": 1.2030654857509981,
      "grad_norm": 0.43938082456588745,
      "learning_rate": 6.711521446229956e-06,
      "loss": 0.3292,
      "step": 26216
    },
    {
      "epoch": 1.2031113762562526,
      "grad_norm": 0.5030285716056824,
      "learning_rate": 6.711291068274637e-06,
      "loss": 0.4002,
      "step": 26217
    },
    {
      "epoch": 1.2031572667615071,
      "grad_norm": 0.49793127179145813,
      "learning_rate": 6.711060686204203e-06,
      "loss": 0.4086,
      "step": 26218
    },
    {
      "epoch": 1.2032031572667614,
      "grad_norm": 0.4553736746311188,
      "learning_rate": 6.710830300019207e-06,
      "loss": 0.3506,
      "step": 26219
    },
    {
      "epoch": 1.203249047772016,
      "grad_norm": 0.4768955707550049,
      "learning_rate": 6.710599909720206e-06,
      "loss": 0.3294,
      "step": 26220
    },
    {
      "epoch": 1.2032949382772704,
      "grad_norm": 0.47765055298805237,
      "learning_rate": 6.71036951530775e-06,
      "loss": 0.328,
      "step": 26221
    },
    {
      "epoch": 1.203340828782525,
      "grad_norm": 0.49707701802253723,
      "learning_rate": 6.710139116782397e-06,
      "loss": 0.4255,
      "step": 26222
    },
    {
      "epoch": 1.2033867192877794,
      "grad_norm": 0.5067639946937561,
      "learning_rate": 6.709908714144697e-06,
      "loss": 0.3856,
      "step": 26223
    },
    {
      "epoch": 1.203432609793034,
      "grad_norm": 0.5027065277099609,
      "learning_rate": 6.7096783073952074e-06,
      "loss": 0.4194,
      "step": 26224
    },
    {
      "epoch": 1.2034785002982882,
      "grad_norm": 0.4665835499763489,
      "learning_rate": 6.709447896534483e-06,
      "loss": 0.3319,
      "step": 26225
    },
    {
      "epoch": 1.2035243908035427,
      "grad_norm": 0.4455587565898895,
      "learning_rate": 6.709217481563073e-06,
      "loss": 0.3343,
      "step": 26226
    },
    {
      "epoch": 1.2035702813087972,
      "grad_norm": 0.5324321985244751,
      "learning_rate": 6.708987062481536e-06,
      "loss": 0.4107,
      "step": 26227
    },
    {
      "epoch": 1.2036161718140517,
      "grad_norm": 0.4528292417526245,
      "learning_rate": 6.708756639290424e-06,
      "loss": 0.3384,
      "step": 26228
    },
    {
      "epoch": 1.2036620623193062,
      "grad_norm": 0.5095716714859009,
      "learning_rate": 6.7085262119902905e-06,
      "loss": 0.4019,
      "step": 26229
    },
    {
      "epoch": 1.2037079528245607,
      "grad_norm": 0.4291735887527466,
      "learning_rate": 6.7082957805816916e-06,
      "loss": 0.2982,
      "step": 26230
    },
    {
      "epoch": 1.2037538433298152,
      "grad_norm": 0.46168407797813416,
      "learning_rate": 6.708065345065181e-06,
      "loss": 0.3472,
      "step": 26231
    },
    {
      "epoch": 1.2037997338350694,
      "grad_norm": 0.5247793793678284,
      "learning_rate": 6.707834905441312e-06,
      "loss": 0.43,
      "step": 26232
    },
    {
      "epoch": 1.203845624340324,
      "grad_norm": 0.4251473844051361,
      "learning_rate": 6.707604461710639e-06,
      "loss": 0.3184,
      "step": 26233
    },
    {
      "epoch": 1.2038915148455784,
      "grad_norm": 0.4705866277217865,
      "learning_rate": 6.707374013873716e-06,
      "loss": 0.3601,
      "step": 26234
    },
    {
      "epoch": 1.203937405350833,
      "grad_norm": 0.46238499879837036,
      "learning_rate": 6.707143561931097e-06,
      "loss": 0.3592,
      "step": 26235
    },
    {
      "epoch": 1.2039832958560874,
      "grad_norm": 0.4436958134174347,
      "learning_rate": 6.706913105883338e-06,
      "loss": 0.2951,
      "step": 26236
    },
    {
      "epoch": 1.204029186361342,
      "grad_norm": 0.5135268568992615,
      "learning_rate": 6.70668264573099e-06,
      "loss": 0.3812,
      "step": 26237
    },
    {
      "epoch": 1.2040750768665962,
      "grad_norm": 0.40967366099357605,
      "learning_rate": 6.7064521814746105e-06,
      "loss": 0.2781,
      "step": 26238
    },
    {
      "epoch": 1.2041209673718507,
      "grad_norm": 0.4583433270454407,
      "learning_rate": 6.706221713114752e-06,
      "loss": 0.3389,
      "step": 26239
    },
    {
      "epoch": 1.2041668578771052,
      "grad_norm": 0.4914122521877289,
      "learning_rate": 6.7059912406519674e-06,
      "loss": 0.3995,
      "step": 26240
    },
    {
      "epoch": 1.2042127483823597,
      "grad_norm": 0.4742193818092346,
      "learning_rate": 6.705760764086813e-06,
      "loss": 0.3831,
      "step": 26241
    },
    {
      "epoch": 1.2042586388876142,
      "grad_norm": 0.47057563066482544,
      "learning_rate": 6.705530283419842e-06,
      "loss": 0.3893,
      "step": 26242
    },
    {
      "epoch": 1.2043045293928687,
      "grad_norm": 0.45723989605903625,
      "learning_rate": 6.705299798651609e-06,
      "loss": 0.3587,
      "step": 26243
    },
    {
      "epoch": 1.2043504198981232,
      "grad_norm": 0.4973413944244385,
      "learning_rate": 6.705069309782668e-06,
      "loss": 0.3728,
      "step": 26244
    },
    {
      "epoch": 1.2043963104033775,
      "grad_norm": 0.45604830980300903,
      "learning_rate": 6.704838816813575e-06,
      "loss": 0.3525,
      "step": 26245
    },
    {
      "epoch": 1.204442200908632,
      "grad_norm": 0.4676319360733032,
      "learning_rate": 6.704608319744881e-06,
      "loss": 0.3323,
      "step": 26246
    },
    {
      "epoch": 1.2044880914138865,
      "grad_norm": 0.46418237686157227,
      "learning_rate": 6.7043778185771415e-06,
      "loss": 0.3319,
      "step": 26247
    },
    {
      "epoch": 1.204533981919141,
      "grad_norm": 0.4311172664165497,
      "learning_rate": 6.7041473133109135e-06,
      "loss": 0.3214,
      "step": 26248
    },
    {
      "epoch": 1.2045798724243955,
      "grad_norm": 0.45412370562553406,
      "learning_rate": 6.703916803946749e-06,
      "loss": 0.3558,
      "step": 26249
    },
    {
      "epoch": 1.2046257629296497,
      "grad_norm": 0.4197354316711426,
      "learning_rate": 6.7036862904852e-06,
      "loss": 0.2994,
      "step": 26250
    },
    {
      "epoch": 1.2046716534349042,
      "grad_norm": 0.45043888688087463,
      "learning_rate": 6.703455772926825e-06,
      "loss": 0.2961,
      "step": 26251
    },
    {
      "epoch": 1.2047175439401587,
      "grad_norm": 0.4745086431503296,
      "learning_rate": 6.703225251272174e-06,
      "loss": 0.324,
      "step": 26252
    },
    {
      "epoch": 1.2047634344454132,
      "grad_norm": 0.44831332564353943,
      "learning_rate": 6.7029947255218064e-06,
      "loss": 0.3126,
      "step": 26253
    },
    {
      "epoch": 1.2048093249506677,
      "grad_norm": 0.4630235731601715,
      "learning_rate": 6.702764195676272e-06,
      "loss": 0.3543,
      "step": 26254
    },
    {
      "epoch": 1.2048552154559222,
      "grad_norm": 0.4879850149154663,
      "learning_rate": 6.70253366173613e-06,
      "loss": 0.3889,
      "step": 26255
    },
    {
      "epoch": 1.2049011059611767,
      "grad_norm": 0.423740416765213,
      "learning_rate": 6.702303123701931e-06,
      "loss": 0.29,
      "step": 26256
    },
    {
      "epoch": 1.2049469964664312,
      "grad_norm": 0.43236997723579407,
      "learning_rate": 6.702072581574228e-06,
      "loss": 0.3418,
      "step": 26257
    },
    {
      "epoch": 1.2049928869716855,
      "grad_norm": 0.4861438572406769,
      "learning_rate": 6.701842035353579e-06,
      "loss": 0.3615,
      "step": 26258
    },
    {
      "epoch": 1.20503877747694,
      "grad_norm": 0.44887417554855347,
      "learning_rate": 6.701611485040538e-06,
      "loss": 0.3003,
      "step": 26259
    },
    {
      "epoch": 1.2050846679821945,
      "grad_norm": 0.48098137974739075,
      "learning_rate": 6.701380930635658e-06,
      "loss": 0.3792,
      "step": 26260
    },
    {
      "epoch": 1.205130558487449,
      "grad_norm": 0.45623332262039185,
      "learning_rate": 6.701150372139493e-06,
      "loss": 0.3218,
      "step": 26261
    },
    {
      "epoch": 1.2051764489927035,
      "grad_norm": 0.4856277406215668,
      "learning_rate": 6.700919809552598e-06,
      "loss": 0.3228,
      "step": 26262
    },
    {
      "epoch": 1.2052223394979578,
      "grad_norm": 0.46467652916908264,
      "learning_rate": 6.700689242875529e-06,
      "loss": 0.3404,
      "step": 26263
    },
    {
      "epoch": 1.2052682300032123,
      "grad_norm": 0.43331435322761536,
      "learning_rate": 6.7004586721088374e-06,
      "loss": 0.2844,
      "step": 26264
    },
    {
      "epoch": 1.2053141205084668,
      "grad_norm": 0.46413710713386536,
      "learning_rate": 6.700228097253081e-06,
      "loss": 0.3791,
      "step": 26265
    },
    {
      "epoch": 1.2053600110137213,
      "grad_norm": 0.48609262704849243,
      "learning_rate": 6.699997518308813e-06,
      "loss": 0.4106,
      "step": 26266
    },
    {
      "epoch": 1.2054059015189758,
      "grad_norm": 0.4860956370830536,
      "learning_rate": 6.699766935276586e-06,
      "loss": 0.337,
      "step": 26267
    },
    {
      "epoch": 1.2054517920242303,
      "grad_norm": 0.4540322422981262,
      "learning_rate": 6.699536348156956e-06,
      "loss": 0.3252,
      "step": 26268
    },
    {
      "epoch": 1.2054976825294847,
      "grad_norm": 0.43985068798065186,
      "learning_rate": 6.699305756950478e-06,
      "loss": 0.3024,
      "step": 26269
    },
    {
      "epoch": 1.205543573034739,
      "grad_norm": 0.48652422428131104,
      "learning_rate": 6.699075161657707e-06,
      "loss": 0.3808,
      "step": 26270
    },
    {
      "epoch": 1.2055894635399935,
      "grad_norm": 0.4419287145137787,
      "learning_rate": 6.698844562279195e-06,
      "loss": 0.3619,
      "step": 26271
    },
    {
      "epoch": 1.205635354045248,
      "grad_norm": 0.5074459910392761,
      "learning_rate": 6.6986139588154985e-06,
      "loss": 0.3877,
      "step": 26272
    },
    {
      "epoch": 1.2056812445505025,
      "grad_norm": 0.5021203756332397,
      "learning_rate": 6.6983833512671715e-06,
      "loss": 0.3737,
      "step": 26273
    },
    {
      "epoch": 1.205727135055757,
      "grad_norm": 0.465021550655365,
      "learning_rate": 6.698152739634769e-06,
      "loss": 0.3671,
      "step": 26274
    },
    {
      "epoch": 1.2057730255610115,
      "grad_norm": 0.44124123454093933,
      "learning_rate": 6.697922123918843e-06,
      "loss": 0.3016,
      "step": 26275
    },
    {
      "epoch": 1.2058189160662658,
      "grad_norm": 0.4807187020778656,
      "learning_rate": 6.697691504119953e-06,
      "loss": 0.3704,
      "step": 26276
    },
    {
      "epoch": 1.2058648065715203,
      "grad_norm": 0.4972396790981293,
      "learning_rate": 6.69746088023865e-06,
      "loss": 0.4012,
      "step": 26277
    },
    {
      "epoch": 1.2059106970767748,
      "grad_norm": 0.4556688070297241,
      "learning_rate": 6.697230252275488e-06,
      "loss": 0.3282,
      "step": 26278
    },
    {
      "epoch": 1.2059565875820293,
      "grad_norm": 0.5092169046401978,
      "learning_rate": 6.696999620231024e-06,
      "loss": 0.3938,
      "step": 26279
    },
    {
      "epoch": 1.2060024780872838,
      "grad_norm": 0.5076898336410522,
      "learning_rate": 6.6967689841058115e-06,
      "loss": 0.4275,
      "step": 26280
    },
    {
      "epoch": 1.2060483685925383,
      "grad_norm": 0.43630626797676086,
      "learning_rate": 6.696538343900405e-06,
      "loss": 0.2756,
      "step": 26281
    },
    {
      "epoch": 1.2060942590977928,
      "grad_norm": 0.5017236471176147,
      "learning_rate": 6.69630769961536e-06,
      "loss": 0.3919,
      "step": 26282
    },
    {
      "epoch": 1.206140149603047,
      "grad_norm": 0.46831414103507996,
      "learning_rate": 6.696077051251228e-06,
      "loss": 0.4012,
      "step": 26283
    },
    {
      "epoch": 1.2061860401083015,
      "grad_norm": 0.48872166872024536,
      "learning_rate": 6.695846398808568e-06,
      "loss": 0.3529,
      "step": 26284
    },
    {
      "epoch": 1.206231930613556,
      "grad_norm": 0.44164133071899414,
      "learning_rate": 6.695615742287932e-06,
      "loss": 0.3428,
      "step": 26285
    },
    {
      "epoch": 1.2062778211188105,
      "grad_norm": 0.4228169918060303,
      "learning_rate": 6.695385081689874e-06,
      "loss": 0.2839,
      "step": 26286
    },
    {
      "epoch": 1.206323711624065,
      "grad_norm": 0.5638141632080078,
      "learning_rate": 6.695154417014953e-06,
      "loss": 0.4096,
      "step": 26287
    },
    {
      "epoch": 1.2063696021293195,
      "grad_norm": 0.4505161643028259,
      "learning_rate": 6.69492374826372e-06,
      "loss": 0.3216,
      "step": 26288
    },
    {
      "epoch": 1.2064154926345738,
      "grad_norm": 0.49491262435913086,
      "learning_rate": 6.694693075436728e-06,
      "loss": 0.3865,
      "step": 26289
    },
    {
      "epoch": 1.2064613831398283,
      "grad_norm": 0.4188593327999115,
      "learning_rate": 6.694462398534536e-06,
      "loss": 0.2587,
      "step": 26290
    },
    {
      "epoch": 1.2065072736450828,
      "grad_norm": 0.4225176274776459,
      "learning_rate": 6.694231717557696e-06,
      "loss": 0.3194,
      "step": 26291
    },
    {
      "epoch": 1.2065531641503373,
      "grad_norm": 0.475223571062088,
      "learning_rate": 6.694001032506763e-06,
      "loss": 0.3147,
      "step": 26292
    },
    {
      "epoch": 1.2065990546555918,
      "grad_norm": 0.4677279591560364,
      "learning_rate": 6.693770343382293e-06,
      "loss": 0.4191,
      "step": 26293
    },
    {
      "epoch": 1.2066449451608463,
      "grad_norm": 0.4817289113998413,
      "learning_rate": 6.6935396501848395e-06,
      "loss": 0.404,
      "step": 26294
    },
    {
      "epoch": 1.2066908356661008,
      "grad_norm": 0.44928985834121704,
      "learning_rate": 6.693308952914958e-06,
      "loss": 0.3348,
      "step": 26295
    },
    {
      "epoch": 1.206736726171355,
      "grad_norm": 0.47700172662734985,
      "learning_rate": 6.693078251573202e-06,
      "loss": 0.3802,
      "step": 26296
    },
    {
      "epoch": 1.2067826166766096,
      "grad_norm": 0.4978356659412384,
      "learning_rate": 6.692847546160127e-06,
      "loss": 0.39,
      "step": 26297
    },
    {
      "epoch": 1.206828507181864,
      "grad_norm": 0.4507928788661957,
      "learning_rate": 6.692616836676289e-06,
      "loss": 0.3445,
      "step": 26298
    },
    {
      "epoch": 1.2068743976871186,
      "grad_norm": 0.45175814628601074,
      "learning_rate": 6.692386123122243e-06,
      "loss": 0.3249,
      "step": 26299
    },
    {
      "epoch": 1.206920288192373,
      "grad_norm": 0.45762115716934204,
      "learning_rate": 6.692155405498539e-06,
      "loss": 0.3458,
      "step": 26300
    },
    {
      "epoch": 1.2069661786976273,
      "grad_norm": 0.4887836277484894,
      "learning_rate": 6.691924683805737e-06,
      "loss": 0.3541,
      "step": 26301
    },
    {
      "epoch": 1.2070120692028818,
      "grad_norm": 0.5671867728233337,
      "learning_rate": 6.6916939580443914e-06,
      "loss": 0.4129,
      "step": 26302
    },
    {
      "epoch": 1.2070579597081363,
      "grad_norm": 0.44942793250083923,
      "learning_rate": 6.691463228215055e-06,
      "loss": 0.3406,
      "step": 26303
    },
    {
      "epoch": 1.2071038502133908,
      "grad_norm": 0.46860775351524353,
      "learning_rate": 6.6912324943182836e-06,
      "loss": 0.3897,
      "step": 26304
    },
    {
      "epoch": 1.2071497407186453,
      "grad_norm": 0.4592318534851074,
      "learning_rate": 6.691001756354631e-06,
      "loss": 0.3756,
      "step": 26305
    },
    {
      "epoch": 1.2071956312238998,
      "grad_norm": 0.41784992814064026,
      "learning_rate": 6.690771014324652e-06,
      "loss": 0.2671,
      "step": 26306
    },
    {
      "epoch": 1.2072415217291543,
      "grad_norm": 0.49648523330688477,
      "learning_rate": 6.690540268228904e-06,
      "loss": 0.3559,
      "step": 26307
    },
    {
      "epoch": 1.2072874122344086,
      "grad_norm": 0.4858267605304718,
      "learning_rate": 6.690309518067939e-06,
      "loss": 0.3568,
      "step": 26308
    },
    {
      "epoch": 1.207333302739663,
      "grad_norm": 0.4511782228946686,
      "learning_rate": 6.690078763842315e-06,
      "loss": 0.321,
      "step": 26309
    },
    {
      "epoch": 1.2073791932449176,
      "grad_norm": 0.4388591945171356,
      "learning_rate": 6.689848005552584e-06,
      "loss": 0.3106,
      "step": 26310
    },
    {
      "epoch": 1.207425083750172,
      "grad_norm": 0.5721585750579834,
      "learning_rate": 6.6896172431993e-06,
      "loss": 0.3226,
      "step": 26311
    },
    {
      "epoch": 1.2074709742554266,
      "grad_norm": 0.4585697054862976,
      "learning_rate": 6.689386476783022e-06,
      "loss": 0.3681,
      "step": 26312
    },
    {
      "epoch": 1.207516864760681,
      "grad_norm": 0.5368145108222961,
      "learning_rate": 6.689155706304302e-06,
      "loss": 0.4756,
      "step": 26313
    },
    {
      "epoch": 1.2075627552659354,
      "grad_norm": 0.42462533712387085,
      "learning_rate": 6.688924931763696e-06,
      "loss": 0.2955,
      "step": 26314
    },
    {
      "epoch": 1.2076086457711899,
      "grad_norm": 0.44715964794158936,
      "learning_rate": 6.688694153161757e-06,
      "loss": 0.3368,
      "step": 26315
    },
    {
      "epoch": 1.2076545362764444,
      "grad_norm": 0.45989862084388733,
      "learning_rate": 6.688463370499044e-06,
      "loss": 0.354,
      "step": 26316
    },
    {
      "epoch": 1.2077004267816989,
      "grad_norm": 0.45057427883148193,
      "learning_rate": 6.688232583776108e-06,
      "loss": 0.3022,
      "step": 26317
    },
    {
      "epoch": 1.2077463172869534,
      "grad_norm": 0.4942217469215393,
      "learning_rate": 6.688001792993506e-06,
      "loss": 0.3757,
      "step": 26318
    },
    {
      "epoch": 1.2077922077922079,
      "grad_norm": 0.46702468395233154,
      "learning_rate": 6.687770998151792e-06,
      "loss": 0.3216,
      "step": 26319
    },
    {
      "epoch": 1.2078380982974624,
      "grad_norm": 0.5165749788284302,
      "learning_rate": 6.687540199251522e-06,
      "loss": 0.4367,
      "step": 26320
    },
    {
      "epoch": 1.2078839888027166,
      "grad_norm": 0.5066128969192505,
      "learning_rate": 6.687309396293251e-06,
      "loss": 0.3999,
      "step": 26321
    },
    {
      "epoch": 1.2079298793079711,
      "grad_norm": 0.5232498645782471,
      "learning_rate": 6.687078589277531e-06,
      "loss": 0.4476,
      "step": 26322
    },
    {
      "epoch": 1.2079757698132256,
      "grad_norm": 0.47634729743003845,
      "learning_rate": 6.686847778204921e-06,
      "loss": 0.3649,
      "step": 26323
    },
    {
      "epoch": 1.2080216603184801,
      "grad_norm": 0.4598873555660248,
      "learning_rate": 6.686616963075975e-06,
      "loss": 0.3733,
      "step": 26324
    },
    {
      "epoch": 1.2080675508237346,
      "grad_norm": 0.4656367003917694,
      "learning_rate": 6.686386143891248e-06,
      "loss": 0.3467,
      "step": 26325
    },
    {
      "epoch": 1.2081134413289891,
      "grad_norm": 0.45353811979293823,
      "learning_rate": 6.686155320651293e-06,
      "loss": 0.3194,
      "step": 26326
    },
    {
      "epoch": 1.2081593318342434,
      "grad_norm": 0.7006787657737732,
      "learning_rate": 6.6859244933566665e-06,
      "loss": 0.325,
      "step": 26327
    },
    {
      "epoch": 1.208205222339498,
      "grad_norm": 0.4548848271369934,
      "learning_rate": 6.6856936620079234e-06,
      "loss": 0.3721,
      "step": 26328
    },
    {
      "epoch": 1.2082511128447524,
      "grad_norm": 0.4923316538333893,
      "learning_rate": 6.6854628266056195e-06,
      "loss": 0.4009,
      "step": 26329
    },
    {
      "epoch": 1.208297003350007,
      "grad_norm": 0.4743734300136566,
      "learning_rate": 6.685231987150311e-06,
      "loss": 0.3556,
      "step": 26330
    },
    {
      "epoch": 1.2083428938552614,
      "grad_norm": 0.48585203289985657,
      "learning_rate": 6.685001143642551e-06,
      "loss": 0.3862,
      "step": 26331
    },
    {
      "epoch": 1.2083887843605159,
      "grad_norm": 0.4953347146511078,
      "learning_rate": 6.684770296082894e-06,
      "loss": 0.4365,
      "step": 26332
    },
    {
      "epoch": 1.2084346748657704,
      "grad_norm": 0.45943862199783325,
      "learning_rate": 6.684539444471897e-06,
      "loss": 0.3838,
      "step": 26333
    },
    {
      "epoch": 1.2084805653710247,
      "grad_norm": 0.4836080074310303,
      "learning_rate": 6.684308588810113e-06,
      "loss": 0.3811,
      "step": 26334
    },
    {
      "epoch": 1.2085264558762792,
      "grad_norm": 0.4816010296344757,
      "learning_rate": 6.684077729098099e-06,
      "loss": 0.3048,
      "step": 26335
    },
    {
      "epoch": 1.2085723463815337,
      "grad_norm": 0.47682085633277893,
      "learning_rate": 6.683846865336412e-06,
      "loss": 0.348,
      "step": 26336
    },
    {
      "epoch": 1.2086182368867882,
      "grad_norm": 0.445208340883255,
      "learning_rate": 6.683615997525602e-06,
      "loss": 0.2998,
      "step": 26337
    },
    {
      "epoch": 1.2086641273920427,
      "grad_norm": 0.4780639111995697,
      "learning_rate": 6.683385125666227e-06,
      "loss": 0.3817,
      "step": 26338
    },
    {
      "epoch": 1.208710017897297,
      "grad_norm": 0.45561856031417847,
      "learning_rate": 6.6831542497588435e-06,
      "loss": 0.3545,
      "step": 26339
    },
    {
      "epoch": 1.2087559084025514,
      "grad_norm": 0.5210508704185486,
      "learning_rate": 6.682923369804003e-06,
      "loss": 0.4559,
      "step": 26340
    },
    {
      "epoch": 1.208801798907806,
      "grad_norm": 0.48558399081230164,
      "learning_rate": 6.682692485802265e-06,
      "loss": 0.4038,
      "step": 26341
    },
    {
      "epoch": 1.2088476894130604,
      "grad_norm": 0.45005717873573303,
      "learning_rate": 6.6824615977541815e-06,
      "loss": 0.3425,
      "step": 26342
    },
    {
      "epoch": 1.208893579918315,
      "grad_norm": 0.4151028096675873,
      "learning_rate": 6.68223070566031e-06,
      "loss": 0.2643,
      "step": 26343
    },
    {
      "epoch": 1.2089394704235694,
      "grad_norm": 0.45483630895614624,
      "learning_rate": 6.681999809521203e-06,
      "loss": 0.3502,
      "step": 26344
    },
    {
      "epoch": 1.208985360928824,
      "grad_norm": 0.4719693958759308,
      "learning_rate": 6.681768909337418e-06,
      "loss": 0.3489,
      "step": 26345
    },
    {
      "epoch": 1.2090312514340784,
      "grad_norm": 0.5175364017486572,
      "learning_rate": 6.68153800510951e-06,
      "loss": 0.4042,
      "step": 26346
    },
    {
      "epoch": 1.2090771419393327,
      "grad_norm": 0.408557653427124,
      "learning_rate": 6.681307096838034e-06,
      "loss": 0.2672,
      "step": 26347
    },
    {
      "epoch": 1.2091230324445872,
      "grad_norm": 0.46874916553497314,
      "learning_rate": 6.681076184523545e-06,
      "loss": 0.383,
      "step": 26348
    },
    {
      "epoch": 1.2091689229498417,
      "grad_norm": 0.43633827567100525,
      "learning_rate": 6.680845268166597e-06,
      "loss": 0.3444,
      "step": 26349
    },
    {
      "epoch": 1.2092148134550962,
      "grad_norm": 0.5227893590927124,
      "learning_rate": 6.680614347767748e-06,
      "loss": 0.4102,
      "step": 26350
    },
    {
      "epoch": 1.2092607039603507,
      "grad_norm": 0.4707088768482208,
      "learning_rate": 6.680383423327551e-06,
      "loss": 0.3518,
      "step": 26351
    },
    {
      "epoch": 1.209306594465605,
      "grad_norm": 0.46726468205451965,
      "learning_rate": 6.6801524948465635e-06,
      "loss": 0.3454,
      "step": 26352
    },
    {
      "epoch": 1.2093524849708595,
      "grad_norm": 0.48585352301597595,
      "learning_rate": 6.679921562325339e-06,
      "loss": 0.3918,
      "step": 26353
    },
    {
      "epoch": 1.209398375476114,
      "grad_norm": 0.5078784227371216,
      "learning_rate": 6.679690625764432e-06,
      "loss": 0.438,
      "step": 26354
    },
    {
      "epoch": 1.2094442659813684,
      "grad_norm": 0.46684470772743225,
      "learning_rate": 6.679459685164401e-06,
      "loss": 0.3509,
      "step": 26355
    },
    {
      "epoch": 1.209490156486623,
      "grad_norm": 0.4665757417678833,
      "learning_rate": 6.679228740525798e-06,
      "loss": 0.4004,
      "step": 26356
    },
    {
      "epoch": 1.2095360469918774,
      "grad_norm": 0.463304340839386,
      "learning_rate": 6.678997791849181e-06,
      "loss": 0.3455,
      "step": 26357
    },
    {
      "epoch": 1.209581937497132,
      "grad_norm": 0.4518081247806549,
      "learning_rate": 6.678766839135104e-06,
      "loss": 0.3337,
      "step": 26358
    },
    {
      "epoch": 1.2096278280023862,
      "grad_norm": 0.4891918897628784,
      "learning_rate": 6.678535882384122e-06,
      "loss": 0.3666,
      "step": 26359
    },
    {
      "epoch": 1.2096737185076407,
      "grad_norm": 0.4654916822910309,
      "learning_rate": 6.678304921596791e-06,
      "loss": 0.3643,
      "step": 26360
    },
    {
      "epoch": 1.2097196090128952,
      "grad_norm": 0.47407469153404236,
      "learning_rate": 6.678073956773668e-06,
      "loss": 0.3233,
      "step": 26361
    },
    {
      "epoch": 1.2097654995181497,
      "grad_norm": 0.49482840299606323,
      "learning_rate": 6.677842987915304e-06,
      "loss": 0.3442,
      "step": 26362
    },
    {
      "epoch": 1.2098113900234042,
      "grad_norm": 0.4636666476726532,
      "learning_rate": 6.677612015022259e-06,
      "loss": 0.3388,
      "step": 26363
    },
    {
      "epoch": 1.2098572805286587,
      "grad_norm": 0.45837557315826416,
      "learning_rate": 6.677381038095087e-06,
      "loss": 0.3071,
      "step": 26364
    },
    {
      "epoch": 1.209903171033913,
      "grad_norm": 0.4642752408981323,
      "learning_rate": 6.677150057134341e-06,
      "loss": 0.3054,
      "step": 26365
    },
    {
      "epoch": 1.2099490615391675,
      "grad_norm": 0.4554455876350403,
      "learning_rate": 6.67691907214058e-06,
      "loss": 0.319,
      "step": 26366
    },
    {
      "epoch": 1.209994952044422,
      "grad_norm": 0.4928278625011444,
      "learning_rate": 6.676688083114358e-06,
      "loss": 0.4069,
      "step": 26367
    },
    {
      "epoch": 1.2100408425496765,
      "grad_norm": 0.5136226415634155,
      "learning_rate": 6.67645709005623e-06,
      "loss": 0.3769,
      "step": 26368
    },
    {
      "epoch": 1.210086733054931,
      "grad_norm": 0.48957252502441406,
      "learning_rate": 6.676226092966751e-06,
      "loss": 0.4138,
      "step": 26369
    },
    {
      "epoch": 1.2101326235601855,
      "grad_norm": 0.4844408333301544,
      "learning_rate": 6.6759950918464775e-06,
      "loss": 0.3755,
      "step": 26370
    },
    {
      "epoch": 1.21017851406544,
      "grad_norm": 0.45277372002601624,
      "learning_rate": 6.675764086695965e-06,
      "loss": 0.3772,
      "step": 26371
    },
    {
      "epoch": 1.2102244045706942,
      "grad_norm": 0.4847615361213684,
      "learning_rate": 6.67553307751577e-06,
      "loss": 0.3897,
      "step": 26372
    },
    {
      "epoch": 1.2102702950759487,
      "grad_norm": 0.442344069480896,
      "learning_rate": 6.675302064306445e-06,
      "loss": 0.3397,
      "step": 26373
    },
    {
      "epoch": 1.2103161855812032,
      "grad_norm": 0.549284040927887,
      "learning_rate": 6.675071047068549e-06,
      "loss": 0.4431,
      "step": 26374
    },
    {
      "epoch": 1.2103620760864577,
      "grad_norm": 0.4563557803630829,
      "learning_rate": 6.674840025802635e-06,
      "loss": 0.322,
      "step": 26375
    },
    {
      "epoch": 1.2104079665917122,
      "grad_norm": 0.456889808177948,
      "learning_rate": 6.674609000509258e-06,
      "loss": 0.3434,
      "step": 26376
    },
    {
      "epoch": 1.2104538570969667,
      "grad_norm": 0.49003541469573975,
      "learning_rate": 6.674377971188975e-06,
      "loss": 0.3741,
      "step": 26377
    },
    {
      "epoch": 1.210499747602221,
      "grad_norm": 0.4436534345149994,
      "learning_rate": 6.674146937842343e-06,
      "loss": 0.3431,
      "step": 26378
    },
    {
      "epoch": 1.2105456381074755,
      "grad_norm": 0.4744948446750641,
      "learning_rate": 6.673915900469916e-06,
      "loss": 0.3959,
      "step": 26379
    },
    {
      "epoch": 1.21059152861273,
      "grad_norm": 0.4869026243686676,
      "learning_rate": 6.673684859072248e-06,
      "loss": 0.4364,
      "step": 26380
    },
    {
      "epoch": 1.2106374191179845,
      "grad_norm": 0.4801062047481537,
      "learning_rate": 6.6734538136498985e-06,
      "loss": 0.3892,
      "step": 26381
    },
    {
      "epoch": 1.210683309623239,
      "grad_norm": 0.4716678261756897,
      "learning_rate": 6.673222764203418e-06,
      "loss": 0.3588,
      "step": 26382
    },
    {
      "epoch": 1.2107292001284935,
      "grad_norm": 0.4320415258407593,
      "learning_rate": 6.672991710733367e-06,
      "loss": 0.2798,
      "step": 26383
    },
    {
      "epoch": 1.210775090633748,
      "grad_norm": 0.45236414670944214,
      "learning_rate": 6.6727606532402975e-06,
      "loss": 0.3026,
      "step": 26384
    },
    {
      "epoch": 1.2108209811390023,
      "grad_norm": 0.48730623722076416,
      "learning_rate": 6.6725295917247675e-06,
      "loss": 0.3428,
      "step": 26385
    },
    {
      "epoch": 1.2108668716442568,
      "grad_norm": 0.4927803874015808,
      "learning_rate": 6.672298526187332e-06,
      "loss": 0.3779,
      "step": 26386
    },
    {
      "epoch": 1.2109127621495113,
      "grad_norm": 0.4604911804199219,
      "learning_rate": 6.672067456628545e-06,
      "loss": 0.3406,
      "step": 26387
    },
    {
      "epoch": 1.2109586526547658,
      "grad_norm": 0.47935548424720764,
      "learning_rate": 6.671836383048964e-06,
      "loss": 0.3807,
      "step": 26388
    },
    {
      "epoch": 1.2110045431600203,
      "grad_norm": 0.5358847379684448,
      "learning_rate": 6.671605305449145e-06,
      "loss": 0.4129,
      "step": 26389
    },
    {
      "epoch": 1.2110504336652745,
      "grad_norm": 0.42999929189682007,
      "learning_rate": 6.671374223829642e-06,
      "loss": 0.2877,
      "step": 26390
    },
    {
      "epoch": 1.211096324170529,
      "grad_norm": 0.4419626295566559,
      "learning_rate": 6.671143138191012e-06,
      "loss": 0.3006,
      "step": 26391
    },
    {
      "epoch": 1.2111422146757835,
      "grad_norm": 0.4754124581813812,
      "learning_rate": 6.67091204853381e-06,
      "loss": 0.3742,
      "step": 26392
    },
    {
      "epoch": 1.211188105181038,
      "grad_norm": 0.5523090362548828,
      "learning_rate": 6.67068095485859e-06,
      "loss": 0.5309,
      "step": 26393
    },
    {
      "epoch": 1.2112339956862925,
      "grad_norm": 0.5166401863098145,
      "learning_rate": 6.670449857165911e-06,
      "loss": 0.3643,
      "step": 26394
    },
    {
      "epoch": 1.211279886191547,
      "grad_norm": 0.4745306074619293,
      "learning_rate": 6.670218755456329e-06,
      "loss": 0.3361,
      "step": 26395
    },
    {
      "epoch": 1.2113257766968015,
      "grad_norm": 0.5197217464447021,
      "learning_rate": 6.6699876497303974e-06,
      "loss": 0.429,
      "step": 26396
    },
    {
      "epoch": 1.2113716672020558,
      "grad_norm": 0.4642806947231293,
      "learning_rate": 6.669756539988671e-06,
      "loss": 0.3348,
      "step": 26397
    },
    {
      "epoch": 1.2114175577073103,
      "grad_norm": 0.5018004775047302,
      "learning_rate": 6.6695254262317075e-06,
      "loss": 0.4091,
      "step": 26398
    },
    {
      "epoch": 1.2114634482125648,
      "grad_norm": 0.4532126188278198,
      "learning_rate": 6.6692943084600625e-06,
      "loss": 0.3072,
      "step": 26399
    },
    {
      "epoch": 1.2115093387178193,
      "grad_norm": 0.4658781886100769,
      "learning_rate": 6.669063186674293e-06,
      "loss": 0.3126,
      "step": 26400
    },
    {
      "epoch": 1.2115552292230738,
      "grad_norm": 0.43039998412132263,
      "learning_rate": 6.668832060874952e-06,
      "loss": 0.3163,
      "step": 26401
    },
    {
      "epoch": 1.2116011197283283,
      "grad_norm": 0.5052329897880554,
      "learning_rate": 6.668600931062597e-06,
      "loss": 0.4467,
      "step": 26402
    },
    {
      "epoch": 1.2116470102335826,
      "grad_norm": 0.44631054997444153,
      "learning_rate": 6.6683697972377825e-06,
      "loss": 0.3454,
      "step": 26403
    },
    {
      "epoch": 1.211692900738837,
      "grad_norm": 0.4222322702407837,
      "learning_rate": 6.6681386594010645e-06,
      "loss": 0.2758,
      "step": 26404
    },
    {
      "epoch": 1.2117387912440916,
      "grad_norm": 0.46659547090530396,
      "learning_rate": 6.667907517553001e-06,
      "loss": 0.3681,
      "step": 26405
    },
    {
      "epoch": 1.211784681749346,
      "grad_norm": 0.44925379753112793,
      "learning_rate": 6.6676763716941465e-06,
      "loss": 0.3212,
      "step": 26406
    },
    {
      "epoch": 1.2118305722546006,
      "grad_norm": 0.5020961761474609,
      "learning_rate": 6.667445221825056e-06,
      "loss": 0.3701,
      "step": 26407
    },
    {
      "epoch": 1.211876462759855,
      "grad_norm": 0.4451366662979126,
      "learning_rate": 6.667214067946286e-06,
      "loss": 0.3001,
      "step": 26408
    },
    {
      "epoch": 1.2119223532651096,
      "grad_norm": 0.46006685495376587,
      "learning_rate": 6.666982910058393e-06,
      "loss": 0.3403,
      "step": 26409
    },
    {
      "epoch": 1.2119682437703638,
      "grad_norm": 0.45795106887817383,
      "learning_rate": 6.666751748161932e-06,
      "loss": 0.3149,
      "step": 26410
    },
    {
      "epoch": 1.2120141342756183,
      "grad_norm": 0.464626282453537,
      "learning_rate": 6.666520582257457e-06,
      "loss": 0.3928,
      "step": 26411
    },
    {
      "epoch": 1.2120600247808728,
      "grad_norm": 0.4776332378387451,
      "learning_rate": 6.66628941234553e-06,
      "loss": 0.3455,
      "step": 26412
    },
    {
      "epoch": 1.2121059152861273,
      "grad_norm": 0.40015533566474915,
      "learning_rate": 6.6660582384267e-06,
      "loss": 0.2512,
      "step": 26413
    },
    {
      "epoch": 1.2121518057913818,
      "grad_norm": 0.4789672791957855,
      "learning_rate": 6.665827060501527e-06,
      "loss": 0.3265,
      "step": 26414
    },
    {
      "epoch": 1.2121976962966363,
      "grad_norm": 0.4617554843425751,
      "learning_rate": 6.665595878570565e-06,
      "loss": 0.3525,
      "step": 26415
    },
    {
      "epoch": 1.2122435868018906,
      "grad_norm": 0.4716511070728302,
      "learning_rate": 6.665364692634369e-06,
      "loss": 0.3519,
      "step": 26416
    },
    {
      "epoch": 1.212289477307145,
      "grad_norm": 0.44423145055770874,
      "learning_rate": 6.6651335026934995e-06,
      "loss": 0.3406,
      "step": 26417
    },
    {
      "epoch": 1.2123353678123996,
      "grad_norm": 0.47908568382263184,
      "learning_rate": 6.6649023087485084e-06,
      "loss": 0.4083,
      "step": 26418
    },
    {
      "epoch": 1.212381258317654,
      "grad_norm": 0.5334530472755432,
      "learning_rate": 6.664671110799952e-06,
      "loss": 0.4623,
      "step": 26419
    },
    {
      "epoch": 1.2124271488229086,
      "grad_norm": 0.4641796350479126,
      "learning_rate": 6.664439908848388e-06,
      "loss": 0.3281,
      "step": 26420
    },
    {
      "epoch": 1.212473039328163,
      "grad_norm": 0.4517514109611511,
      "learning_rate": 6.66420870289437e-06,
      "loss": 0.3612,
      "step": 26421
    },
    {
      "epoch": 1.2125189298334176,
      "grad_norm": 0.46616753935813904,
      "learning_rate": 6.663977492938456e-06,
      "loss": 0.2958,
      "step": 26422
    },
    {
      "epoch": 1.2125648203386719,
      "grad_norm": 0.45648351311683655,
      "learning_rate": 6.663746278981203e-06,
      "loss": 0.3633,
      "step": 26423
    },
    {
      "epoch": 1.2126107108439264,
      "grad_norm": 0.48374369740486145,
      "learning_rate": 6.663515061023162e-06,
      "loss": 0.3769,
      "step": 26424
    },
    {
      "epoch": 1.2126566013491809,
      "grad_norm": 0.5126987099647522,
      "learning_rate": 6.6632838390648935e-06,
      "loss": 0.3928,
      "step": 26425
    },
    {
      "epoch": 1.2127024918544353,
      "grad_norm": 0.4751146733760834,
      "learning_rate": 6.663052613106954e-06,
      "loss": 0.4092,
      "step": 26426
    },
    {
      "epoch": 1.2127483823596898,
      "grad_norm": 0.45421040058135986,
      "learning_rate": 6.662821383149895e-06,
      "loss": 0.3283,
      "step": 26427
    },
    {
      "epoch": 1.2127942728649441,
      "grad_norm": 0.4889911711215973,
      "learning_rate": 6.662590149194278e-06,
      "loss": 0.3985,
      "step": 26428
    },
    {
      "epoch": 1.2128401633701986,
      "grad_norm": 0.5051060318946838,
      "learning_rate": 6.662358911240656e-06,
      "loss": 0.447,
      "step": 26429
    },
    {
      "epoch": 1.2128860538754531,
      "grad_norm": 0.4766351580619812,
      "learning_rate": 6.662127669289584e-06,
      "loss": 0.3786,
      "step": 26430
    },
    {
      "epoch": 1.2129319443807076,
      "grad_norm": 0.452815443277359,
      "learning_rate": 6.66189642334162e-06,
      "loss": 0.3179,
      "step": 26431
    },
    {
      "epoch": 1.2129778348859621,
      "grad_norm": 0.4612380862236023,
      "learning_rate": 6.66166517339732e-06,
      "loss": 0.3611,
      "step": 26432
    },
    {
      "epoch": 1.2130237253912166,
      "grad_norm": 0.451398104429245,
      "learning_rate": 6.66143391945724e-06,
      "loss": 0.3378,
      "step": 26433
    },
    {
      "epoch": 1.213069615896471,
      "grad_norm": 0.4431014358997345,
      "learning_rate": 6.6612026615219375e-06,
      "loss": 0.3556,
      "step": 26434
    },
    {
      "epoch": 1.2131155064017256,
      "grad_norm": 0.4857947528362274,
      "learning_rate": 6.660971399591965e-06,
      "loss": 0.3406,
      "step": 26435
    },
    {
      "epoch": 1.2131613969069799,
      "grad_norm": 0.45997658371925354,
      "learning_rate": 6.660740133667879e-06,
      "loss": 0.3117,
      "step": 26436
    },
    {
      "epoch": 1.2132072874122344,
      "grad_norm": 0.48422735929489136,
      "learning_rate": 6.660508863750239e-06,
      "loss": 0.3749,
      "step": 26437
    },
    {
      "epoch": 1.2132531779174889,
      "grad_norm": 0.6273134350776672,
      "learning_rate": 6.660277589839598e-06,
      "loss": 0.3967,
      "step": 26438
    },
    {
      "epoch": 1.2132990684227434,
      "grad_norm": 0.4786001741886139,
      "learning_rate": 6.660046311936514e-06,
      "loss": 0.3251,
      "step": 26439
    },
    {
      "epoch": 1.2133449589279979,
      "grad_norm": 0.5093516111373901,
      "learning_rate": 6.6598150300415434e-06,
      "loss": 0.3985,
      "step": 26440
    },
    {
      "epoch": 1.2133908494332521,
      "grad_norm": 0.44140133261680603,
      "learning_rate": 6.65958374415524e-06,
      "loss": 0.2994,
      "step": 26441
    },
    {
      "epoch": 1.2134367399385066,
      "grad_norm": 0.4560340642929077,
      "learning_rate": 6.659352454278163e-06,
      "loss": 0.3328,
      "step": 26442
    },
    {
      "epoch": 1.2134826304437611,
      "grad_norm": 0.4609656035900116,
      "learning_rate": 6.659121160410866e-06,
      "loss": 0.3109,
      "step": 26443
    },
    {
      "epoch": 1.2135285209490156,
      "grad_norm": 0.4403704106807709,
      "learning_rate": 6.658889862553907e-06,
      "loss": 0.2964,
      "step": 26444
    },
    {
      "epoch": 1.2135744114542701,
      "grad_norm": 0.4830181896686554,
      "learning_rate": 6.65865856070784e-06,
      "loss": 0.3805,
      "step": 26445
    },
    {
      "epoch": 1.2136203019595246,
      "grad_norm": 0.4448684751987457,
      "learning_rate": 6.658427254873225e-06,
      "loss": 0.3411,
      "step": 26446
    },
    {
      "epoch": 1.2136661924647791,
      "grad_norm": 0.5083604454994202,
      "learning_rate": 6.658195945050614e-06,
      "loss": 0.3782,
      "step": 26447
    },
    {
      "epoch": 1.2137120829700334,
      "grad_norm": 0.4937281608581543,
      "learning_rate": 6.6579646312405656e-06,
      "loss": 0.422,
      "step": 26448
    },
    {
      "epoch": 1.213757973475288,
      "grad_norm": 0.4181835353374481,
      "learning_rate": 6.657733313443635e-06,
      "loss": 0.2713,
      "step": 26449
    },
    {
      "epoch": 1.2138038639805424,
      "grad_norm": 0.47744104266166687,
      "learning_rate": 6.65750199166038e-06,
      "loss": 0.3566,
      "step": 26450
    },
    {
      "epoch": 1.213849754485797,
      "grad_norm": 0.49983495473861694,
      "learning_rate": 6.657270665891357e-06,
      "loss": 0.3754,
      "step": 26451
    },
    {
      "epoch": 1.2138956449910514,
      "grad_norm": 0.4469482898712158,
      "learning_rate": 6.657039336137119e-06,
      "loss": 0.3372,
      "step": 26452
    },
    {
      "epoch": 1.213941535496306,
      "grad_norm": 0.44555771350860596,
      "learning_rate": 6.656808002398224e-06,
      "loss": 0.339,
      "step": 26453
    },
    {
      "epoch": 1.2139874260015602,
      "grad_norm": 0.42917266488075256,
      "learning_rate": 6.656576664675229e-06,
      "loss": 0.3027,
      "step": 26454
    },
    {
      "epoch": 1.2140333165068147,
      "grad_norm": 0.47455111145973206,
      "learning_rate": 6.656345322968691e-06,
      "loss": 0.353,
      "step": 26455
    },
    {
      "epoch": 1.2140792070120692,
      "grad_norm": 0.4238443970680237,
      "learning_rate": 6.656113977279165e-06,
      "loss": 0.2895,
      "step": 26456
    },
    {
      "epoch": 1.2141250975173237,
      "grad_norm": 0.4935763478279114,
      "learning_rate": 6.655882627607207e-06,
      "loss": 0.4059,
      "step": 26457
    },
    {
      "epoch": 1.2141709880225782,
      "grad_norm": 0.43882501125335693,
      "learning_rate": 6.655651273953374e-06,
      "loss": 0.3077,
      "step": 26458
    },
    {
      "epoch": 1.2142168785278327,
      "grad_norm": 0.5318296551704407,
      "learning_rate": 6.655419916318222e-06,
      "loss": 0.4453,
      "step": 26459
    },
    {
      "epoch": 1.2142627690330872,
      "grad_norm": 0.4326549470424652,
      "learning_rate": 6.655188554702309e-06,
      "loss": 0.3145,
      "step": 26460
    },
    {
      "epoch": 1.2143086595383414,
      "grad_norm": 0.47563329339027405,
      "learning_rate": 6.654957189106189e-06,
      "loss": 0.3672,
      "step": 26461
    },
    {
      "epoch": 1.214354550043596,
      "grad_norm": 0.48194050788879395,
      "learning_rate": 6.6547258195304186e-06,
      "loss": 0.3967,
      "step": 26462
    },
    {
      "epoch": 1.2144004405488504,
      "grad_norm": 0.5419139266014099,
      "learning_rate": 6.6544944459755554e-06,
      "loss": 0.496,
      "step": 26463
    },
    {
      "epoch": 1.214446331054105,
      "grad_norm": 0.48219239711761475,
      "learning_rate": 6.654263068442155e-06,
      "loss": 0.363,
      "step": 26464
    },
    {
      "epoch": 1.2144922215593594,
      "grad_norm": 0.5022880434989929,
      "learning_rate": 6.654031686930775e-06,
      "loss": 0.3624,
      "step": 26465
    },
    {
      "epoch": 1.214538112064614,
      "grad_norm": 0.4882560670375824,
      "learning_rate": 6.65380030144197e-06,
      "loss": 0.3481,
      "step": 26466
    },
    {
      "epoch": 1.2145840025698682,
      "grad_norm": 0.5415197014808655,
      "learning_rate": 6.653568911976297e-06,
      "loss": 0.428,
      "step": 26467
    },
    {
      "epoch": 1.2146298930751227,
      "grad_norm": 0.478347510099411,
      "learning_rate": 6.653337518534314e-06,
      "loss": 0.3416,
      "step": 26468
    },
    {
      "epoch": 1.2146757835803772,
      "grad_norm": 0.49029624462127686,
      "learning_rate": 6.653106121116575e-06,
      "loss": 0.3324,
      "step": 26469
    },
    {
      "epoch": 1.2147216740856317,
      "grad_norm": 0.4428369998931885,
      "learning_rate": 6.652874719723637e-06,
      "loss": 0.2993,
      "step": 26470
    },
    {
      "epoch": 1.2147675645908862,
      "grad_norm": 0.4888448417186737,
      "learning_rate": 6.652643314356058e-06,
      "loss": 0.3229,
      "step": 26471
    },
    {
      "epoch": 1.2148134550961407,
      "grad_norm": 0.5043996572494507,
      "learning_rate": 6.652411905014393e-06,
      "loss": 0.4181,
      "step": 26472
    },
    {
      "epoch": 1.2148593456013952,
      "grad_norm": 0.4533562958240509,
      "learning_rate": 6.652180491699199e-06,
      "loss": 0.3288,
      "step": 26473
    },
    {
      "epoch": 1.2149052361066495,
      "grad_norm": 0.44803428649902344,
      "learning_rate": 6.651949074411032e-06,
      "loss": 0.3367,
      "step": 26474
    },
    {
      "epoch": 1.214951126611904,
      "grad_norm": 0.4290565252304077,
      "learning_rate": 6.651717653150448e-06,
      "loss": 0.272,
      "step": 26475
    },
    {
      "epoch": 1.2149970171171585,
      "grad_norm": 0.40704265236854553,
      "learning_rate": 6.651486227918005e-06,
      "loss": 0.2749,
      "step": 26476
    },
    {
      "epoch": 1.215042907622413,
      "grad_norm": 0.49190792441368103,
      "learning_rate": 6.6512547987142596e-06,
      "loss": 0.384,
      "step": 26477
    },
    {
      "epoch": 1.2150887981276675,
      "grad_norm": 0.7012590765953064,
      "learning_rate": 6.651023365539769e-06,
      "loss": 0.4232,
      "step": 26478
    },
    {
      "epoch": 1.2151346886329217,
      "grad_norm": 0.4812316596508026,
      "learning_rate": 6.650791928395085e-06,
      "loss": 0.4585,
      "step": 26479
    },
    {
      "epoch": 1.2151805791381762,
      "grad_norm": 0.43747609853744507,
      "learning_rate": 6.650560487280769e-06,
      "loss": 0.3326,
      "step": 26480
    },
    {
      "epoch": 1.2152264696434307,
      "grad_norm": 0.47373586893081665,
      "learning_rate": 6.650329042197375e-06,
      "loss": 0.3477,
      "step": 26481
    },
    {
      "epoch": 1.2152723601486852,
      "grad_norm": 0.4390137195587158,
      "learning_rate": 6.650097593145463e-06,
      "loss": 0.2851,
      "step": 26482
    },
    {
      "epoch": 1.2153182506539397,
      "grad_norm": 0.4655608534812927,
      "learning_rate": 6.6498661401255845e-06,
      "loss": 0.3319,
      "step": 26483
    },
    {
      "epoch": 1.2153641411591942,
      "grad_norm": 0.4914844334125519,
      "learning_rate": 6.6496346831383e-06,
      "loss": 0.3812,
      "step": 26484
    },
    {
      "epoch": 1.2154100316644487,
      "grad_norm": 0.5391208529472351,
      "learning_rate": 6.649403222184164e-06,
      "loss": 0.4293,
      "step": 26485
    },
    {
      "epoch": 1.215455922169703,
      "grad_norm": 0.4957500398159027,
      "learning_rate": 6.649171757263734e-06,
      "loss": 0.4034,
      "step": 26486
    },
    {
      "epoch": 1.2155018126749575,
      "grad_norm": 0.5317193865776062,
      "learning_rate": 6.648940288377565e-06,
      "loss": 0.3771,
      "step": 26487
    },
    {
      "epoch": 1.215547703180212,
      "grad_norm": 0.5028627514839172,
      "learning_rate": 6.648708815526218e-06,
      "loss": 0.4635,
      "step": 26488
    },
    {
      "epoch": 1.2155935936854665,
      "grad_norm": 0.44702839851379395,
      "learning_rate": 6.648477338710246e-06,
      "loss": 0.3613,
      "step": 26489
    },
    {
      "epoch": 1.215639484190721,
      "grad_norm": 0.47877851128578186,
      "learning_rate": 6.648245857930205e-06,
      "loss": 0.3658,
      "step": 26490
    },
    {
      "epoch": 1.2156853746959755,
      "grad_norm": 0.48605743050575256,
      "learning_rate": 6.648014373186653e-06,
      "loss": 0.3923,
      "step": 26491
    },
    {
      "epoch": 1.2157312652012298,
      "grad_norm": 0.4536574184894562,
      "learning_rate": 6.647782884480148e-06,
      "loss": 0.3288,
      "step": 26492
    },
    {
      "epoch": 1.2157771557064843,
      "grad_norm": 0.4504633843898773,
      "learning_rate": 6.647551391811243e-06,
      "loss": 0.2962,
      "step": 26493
    },
    {
      "epoch": 1.2158230462117388,
      "grad_norm": 0.5358149409294128,
      "learning_rate": 6.6473198951805e-06,
      "loss": 0.4343,
      "step": 26494
    },
    {
      "epoch": 1.2158689367169933,
      "grad_norm": 0.4692898094654083,
      "learning_rate": 6.647088394588469e-06,
      "loss": 0.3549,
      "step": 26495
    },
    {
      "epoch": 1.2159148272222478,
      "grad_norm": 0.4898407757282257,
      "learning_rate": 6.646856890035711e-06,
      "loss": 0.359,
      "step": 26496
    },
    {
      "epoch": 1.2159607177275022,
      "grad_norm": 0.484628289937973,
      "learning_rate": 6.646625381522784e-06,
      "loss": 0.3601,
      "step": 26497
    },
    {
      "epoch": 1.2160066082327567,
      "grad_norm": 0.4548763632774353,
      "learning_rate": 6.64639386905024e-06,
      "loss": 0.3018,
      "step": 26498
    },
    {
      "epoch": 1.216052498738011,
      "grad_norm": 0.4199632704257965,
      "learning_rate": 6.646162352618641e-06,
      "loss": 0.3319,
      "step": 26499
    },
    {
      "epoch": 1.2160983892432655,
      "grad_norm": 0.43360844254493713,
      "learning_rate": 6.645930832228541e-06,
      "loss": 0.3152,
      "step": 26500
    },
    {
      "epoch": 1.21614427974852,
      "grad_norm": 0.5200902223587036,
      "learning_rate": 6.645699307880496e-06,
      "loss": 0.4344,
      "step": 26501
    },
    {
      "epoch": 1.2161901702537745,
      "grad_norm": 0.46995633840560913,
      "learning_rate": 6.645467779575062e-06,
      "loss": 0.359,
      "step": 26502
    },
    {
      "epoch": 1.216236060759029,
      "grad_norm": 0.42737650871276855,
      "learning_rate": 6.6452362473127985e-06,
      "loss": 0.3075,
      "step": 26503
    },
    {
      "epoch": 1.2162819512642835,
      "grad_norm": 0.4868408143520355,
      "learning_rate": 6.645004711094262e-06,
      "loss": 0.3962,
      "step": 26504
    },
    {
      "epoch": 1.2163278417695378,
      "grad_norm": 0.48662471771240234,
      "learning_rate": 6.644773170920008e-06,
      "loss": 0.3435,
      "step": 26505
    },
    {
      "epoch": 1.2163737322747923,
      "grad_norm": 0.4803045988082886,
      "learning_rate": 6.644541626790592e-06,
      "loss": 0.4428,
      "step": 26506
    },
    {
      "epoch": 1.2164196227800468,
      "grad_norm": 0.4640991687774658,
      "learning_rate": 6.644310078706573e-06,
      "loss": 0.3423,
      "step": 26507
    },
    {
      "epoch": 1.2164655132853013,
      "grad_norm": 0.48399603366851807,
      "learning_rate": 6.64407852666851e-06,
      "loss": 0.4255,
      "step": 26508
    },
    {
      "epoch": 1.2165114037905558,
      "grad_norm": 0.4583907723426819,
      "learning_rate": 6.643846970676953e-06,
      "loss": 0.3853,
      "step": 26509
    },
    {
      "epoch": 1.2165572942958103,
      "grad_norm": 0.48331916332244873,
      "learning_rate": 6.643615410732465e-06,
      "loss": 0.4226,
      "step": 26510
    },
    {
      "epoch": 1.2166031848010648,
      "grad_norm": 0.5015382766723633,
      "learning_rate": 6.6433838468356e-06,
      "loss": 0.4226,
      "step": 26511
    },
    {
      "epoch": 1.216649075306319,
      "grad_norm": 0.4789227843284607,
      "learning_rate": 6.643152278986916e-06,
      "loss": 0.3551,
      "step": 26512
    },
    {
      "epoch": 1.2166949658115735,
      "grad_norm": 0.44056299328804016,
      "learning_rate": 6.642920707186968e-06,
      "loss": 0.3149,
      "step": 26513
    },
    {
      "epoch": 1.216740856316828,
      "grad_norm": 0.4695565104484558,
      "learning_rate": 6.642689131436317e-06,
      "loss": 0.4132,
      "step": 26514
    },
    {
      "epoch": 1.2167867468220825,
      "grad_norm": 0.4440353214740753,
      "learning_rate": 6.642457551735516e-06,
      "loss": 0.3034,
      "step": 26515
    },
    {
      "epoch": 1.216832637327337,
      "grad_norm": 0.46822142601013184,
      "learning_rate": 6.642225968085123e-06,
      "loss": 0.3553,
      "step": 26516
    },
    {
      "epoch": 1.2168785278325913,
      "grad_norm": 0.4997451901435852,
      "learning_rate": 6.641994380485693e-06,
      "loss": 0.441,
      "step": 26517
    },
    {
      "epoch": 1.2169244183378458,
      "grad_norm": 0.4446333944797516,
      "learning_rate": 6.641762788937786e-06,
      "loss": 0.3269,
      "step": 26518
    },
    {
      "epoch": 1.2169703088431003,
      "grad_norm": 0.45543307065963745,
      "learning_rate": 6.641531193441959e-06,
      "loss": 0.4177,
      "step": 26519
    },
    {
      "epoch": 1.2170161993483548,
      "grad_norm": 0.4097939431667328,
      "learning_rate": 6.641299593998765e-06,
      "loss": 0.2536,
      "step": 26520
    },
    {
      "epoch": 1.2170620898536093,
      "grad_norm": 0.5092282891273499,
      "learning_rate": 6.641067990608765e-06,
      "loss": 0.4529,
      "step": 26521
    },
    {
      "epoch": 1.2171079803588638,
      "grad_norm": 0.47216612100601196,
      "learning_rate": 6.6408363832725154e-06,
      "loss": 0.4249,
      "step": 26522
    },
    {
      "epoch": 1.2171538708641183,
      "grad_norm": 0.44188278913497925,
      "learning_rate": 6.640604771990571e-06,
      "loss": 0.3356,
      "step": 26523
    },
    {
      "epoch": 1.2171997613693728,
      "grad_norm": 0.43510234355926514,
      "learning_rate": 6.640373156763489e-06,
      "loss": 0.3104,
      "step": 26524
    },
    {
      "epoch": 1.217245651874627,
      "grad_norm": 0.4596518278121948,
      "learning_rate": 6.640141537591829e-06,
      "loss": 0.283,
      "step": 26525
    },
    {
      "epoch": 1.2172915423798816,
      "grad_norm": 0.4891955256462097,
      "learning_rate": 6.639909914476145e-06,
      "loss": 0.4092,
      "step": 26526
    },
    {
      "epoch": 1.217337432885136,
      "grad_norm": 0.5185647010803223,
      "learning_rate": 6.639678287416995e-06,
      "loss": 0.4304,
      "step": 26527
    },
    {
      "epoch": 1.2173833233903906,
      "grad_norm": 0.4624726474285126,
      "learning_rate": 6.639446656414937e-06,
      "loss": 0.3279,
      "step": 26528
    },
    {
      "epoch": 1.217429213895645,
      "grad_norm": 0.4289744794368744,
      "learning_rate": 6.639215021470526e-06,
      "loss": 0.2879,
      "step": 26529
    },
    {
      "epoch": 1.2174751044008993,
      "grad_norm": 0.4355056881904602,
      "learning_rate": 6.638983382584322e-06,
      "loss": 0.2924,
      "step": 26530
    },
    {
      "epoch": 1.2175209949061538,
      "grad_norm": 0.4906013011932373,
      "learning_rate": 6.63875173975688e-06,
      "loss": 0.4147,
      "step": 26531
    },
    {
      "epoch": 1.2175668854114083,
      "grad_norm": 0.4621824324131012,
      "learning_rate": 6.638520092988757e-06,
      "loss": 0.369,
      "step": 26532
    },
    {
      "epoch": 1.2176127759166628,
      "grad_norm": 0.4605434536933899,
      "learning_rate": 6.638288442280511e-06,
      "loss": 0.367,
      "step": 26533
    },
    {
      "epoch": 1.2176586664219173,
      "grad_norm": 0.4299205541610718,
      "learning_rate": 6.638056787632696e-06,
      "loss": 0.2991,
      "step": 26534
    },
    {
      "epoch": 1.2177045569271718,
      "grad_norm": 0.44541507959365845,
      "learning_rate": 6.637825129045873e-06,
      "loss": 0.3184,
      "step": 26535
    },
    {
      "epoch": 1.2177504474324263,
      "grad_norm": 0.45722100138664246,
      "learning_rate": 6.637593466520598e-06,
      "loss": 0.3374,
      "step": 26536
    },
    {
      "epoch": 1.2177963379376806,
      "grad_norm": 0.45860129594802856,
      "learning_rate": 6.637361800057427e-06,
      "loss": 0.3596,
      "step": 26537
    },
    {
      "epoch": 1.217842228442935,
      "grad_norm": 0.47997209429740906,
      "learning_rate": 6.637130129656917e-06,
      "loss": 0.3826,
      "step": 26538
    },
    {
      "epoch": 1.2178881189481896,
      "grad_norm": 0.48910272121429443,
      "learning_rate": 6.636898455319627e-06,
      "loss": 0.3881,
      "step": 26539
    },
    {
      "epoch": 1.217934009453444,
      "grad_norm": 0.49934619665145874,
      "learning_rate": 6.636666777046111e-06,
      "loss": 0.4209,
      "step": 26540
    },
    {
      "epoch": 1.2179798999586986,
      "grad_norm": 0.501197874546051,
      "learning_rate": 6.636435094836928e-06,
      "loss": 0.4366,
      "step": 26541
    },
    {
      "epoch": 1.218025790463953,
      "grad_norm": 0.4443577826023102,
      "learning_rate": 6.636203408692637e-06,
      "loss": 0.3429,
      "step": 26542
    },
    {
      "epoch": 1.2180716809692074,
      "grad_norm": 0.4614202678203583,
      "learning_rate": 6.635971718613794e-06,
      "loss": 0.323,
      "step": 26543
    },
    {
      "epoch": 1.2181175714744619,
      "grad_norm": 0.5034525394439697,
      "learning_rate": 6.6357400246009514e-06,
      "loss": 0.4231,
      "step": 26544
    },
    {
      "epoch": 1.2181634619797164,
      "grad_norm": 0.47992390394210815,
      "learning_rate": 6.635508326654674e-06,
      "loss": 0.3489,
      "step": 26545
    },
    {
      "epoch": 1.2182093524849709,
      "grad_norm": 0.4550305902957916,
      "learning_rate": 6.635276624775513e-06,
      "loss": 0.3266,
      "step": 26546
    },
    {
      "epoch": 1.2182552429902254,
      "grad_norm": 0.47528770565986633,
      "learning_rate": 6.63504491896403e-06,
      "loss": 0.3431,
      "step": 26547
    },
    {
      "epoch": 1.2183011334954799,
      "grad_norm": 0.4603108763694763,
      "learning_rate": 6.634813209220779e-06,
      "loss": 0.3525,
      "step": 26548
    },
    {
      "epoch": 1.2183470240007344,
      "grad_norm": 0.5018152594566345,
      "learning_rate": 6.634581495546317e-06,
      "loss": 0.4565,
      "step": 26549
    },
    {
      "epoch": 1.2183929145059886,
      "grad_norm": 0.4865402281284332,
      "learning_rate": 6.634349777941203e-06,
      "loss": 0.3871,
      "step": 26550
    },
    {
      "epoch": 1.2184388050112431,
      "grad_norm": 0.4930585026741028,
      "learning_rate": 6.634118056405994e-06,
      "loss": 0.4138,
      "step": 26551
    },
    {
      "epoch": 1.2184846955164976,
      "grad_norm": 0.5254998803138733,
      "learning_rate": 6.633886330941246e-06,
      "loss": 0.5326,
      "step": 26552
    },
    {
      "epoch": 1.2185305860217521,
      "grad_norm": 0.4398391842842102,
      "learning_rate": 6.633654601547519e-06,
      "loss": 0.3597,
      "step": 26553
    },
    {
      "epoch": 1.2185764765270066,
      "grad_norm": 0.4414946138858795,
      "learning_rate": 6.633422868225367e-06,
      "loss": 0.3108,
      "step": 26554
    },
    {
      "epoch": 1.2186223670322611,
      "grad_norm": 0.5112548470497131,
      "learning_rate": 6.6331911309753475e-06,
      "loss": 0.4108,
      "step": 26555
    },
    {
      "epoch": 1.2186682575375154,
      "grad_norm": 0.46837103366851807,
      "learning_rate": 6.63295938979802e-06,
      "loss": 0.357,
      "step": 26556
    },
    {
      "epoch": 1.21871414804277,
      "grad_norm": 0.44416484236717224,
      "learning_rate": 6.63272764469394e-06,
      "loss": 0.3515,
      "step": 26557
    },
    {
      "epoch": 1.2187600385480244,
      "grad_norm": 0.41245585680007935,
      "learning_rate": 6.632495895663667e-06,
      "loss": 0.2781,
      "step": 26558
    },
    {
      "epoch": 1.218805929053279,
      "grad_norm": 0.5223581790924072,
      "learning_rate": 6.632264142707755e-06,
      "loss": 0.4338,
      "step": 26559
    },
    {
      "epoch": 1.2188518195585334,
      "grad_norm": 0.5180321335792542,
      "learning_rate": 6.632032385826763e-06,
      "loss": 0.4397,
      "step": 26560
    },
    {
      "epoch": 1.2188977100637879,
      "grad_norm": 0.5114618539810181,
      "learning_rate": 6.631800625021248e-06,
      "loss": 0.4103,
      "step": 26561
    },
    {
      "epoch": 1.2189436005690424,
      "grad_norm": 0.48380526900291443,
      "learning_rate": 6.631568860291768e-06,
      "loss": 0.3766,
      "step": 26562
    },
    {
      "epoch": 1.2189894910742967,
      "grad_norm": 0.5175414681434631,
      "learning_rate": 6.631337091638878e-06,
      "loss": 0.3806,
      "step": 26563
    },
    {
      "epoch": 1.2190353815795512,
      "grad_norm": 0.4947982728481293,
      "learning_rate": 6.631105319063141e-06,
      "loss": 0.4123,
      "step": 26564
    },
    {
      "epoch": 1.2190812720848057,
      "grad_norm": 0.45331743359565735,
      "learning_rate": 6.6308735425651085e-06,
      "loss": 0.3472,
      "step": 26565
    },
    {
      "epoch": 1.2191271625900602,
      "grad_norm": 0.45677581429481506,
      "learning_rate": 6.630641762145339e-06,
      "loss": 0.3695,
      "step": 26566
    },
    {
      "epoch": 1.2191730530953147,
      "grad_norm": 0.4524634778499603,
      "learning_rate": 6.630409977804392e-06,
      "loss": 0.3537,
      "step": 26567
    },
    {
      "epoch": 1.219218943600569,
      "grad_norm": 0.47633957862854004,
      "learning_rate": 6.630178189542822e-06,
      "loss": 0.3775,
      "step": 26568
    },
    {
      "epoch": 1.2192648341058234,
      "grad_norm": 0.48561084270477295,
      "learning_rate": 6.629946397361189e-06,
      "loss": 0.4026,
      "step": 26569
    },
    {
      "epoch": 1.219310724611078,
      "grad_norm": 0.5073794722557068,
      "learning_rate": 6.62971460126005e-06,
      "loss": 0.4541,
      "step": 26570
    },
    {
      "epoch": 1.2193566151163324,
      "grad_norm": 0.49884694814682007,
      "learning_rate": 6.629482801239961e-06,
      "loss": 0.4371,
      "step": 26571
    },
    {
      "epoch": 1.219402505621587,
      "grad_norm": 0.4563567340373993,
      "learning_rate": 6.6292509973014796e-06,
      "loss": 0.3667,
      "step": 26572
    },
    {
      "epoch": 1.2194483961268414,
      "grad_norm": 0.4753012955188751,
      "learning_rate": 6.629019189445165e-06,
      "loss": 0.4142,
      "step": 26573
    },
    {
      "epoch": 1.219494286632096,
      "grad_norm": 0.4564734399318695,
      "learning_rate": 6.628787377671572e-06,
      "loss": 0.3011,
      "step": 26574
    },
    {
      "epoch": 1.2195401771373502,
      "grad_norm": 0.5007832646369934,
      "learning_rate": 6.62855556198126e-06,
      "loss": 0.4545,
      "step": 26575
    },
    {
      "epoch": 1.2195860676426047,
      "grad_norm": 0.4220879077911377,
      "learning_rate": 6.628323742374787e-06,
      "loss": 0.2826,
      "step": 26576
    },
    {
      "epoch": 1.2196319581478592,
      "grad_norm": 0.47722509503364563,
      "learning_rate": 6.628091918852707e-06,
      "loss": 0.3948,
      "step": 26577
    },
    {
      "epoch": 1.2196778486531137,
      "grad_norm": 0.46549999713897705,
      "learning_rate": 6.627860091415581e-06,
      "loss": 0.3289,
      "step": 26578
    },
    {
      "epoch": 1.2197237391583682,
      "grad_norm": 0.4445994198322296,
      "learning_rate": 6.627628260063965e-06,
      "loss": 0.3065,
      "step": 26579
    },
    {
      "epoch": 1.2197696296636227,
      "grad_norm": 0.47771313786506653,
      "learning_rate": 6.627396424798418e-06,
      "loss": 0.3604,
      "step": 26580
    },
    {
      "epoch": 1.219815520168877,
      "grad_norm": 0.4614464342594147,
      "learning_rate": 6.627164585619496e-06,
      "loss": 0.3547,
      "step": 26581
    },
    {
      "epoch": 1.2198614106741315,
      "grad_norm": 0.49790000915527344,
      "learning_rate": 6.626932742527755e-06,
      "loss": 0.4394,
      "step": 26582
    },
    {
      "epoch": 1.219907301179386,
      "grad_norm": 0.5032796859741211,
      "learning_rate": 6.626700895523754e-06,
      "loss": 0.4007,
      "step": 26583
    },
    {
      "epoch": 1.2199531916846404,
      "grad_norm": 0.48040270805358887,
      "learning_rate": 6.626469044608053e-06,
      "loss": 0.3841,
      "step": 26584
    },
    {
      "epoch": 1.219999082189895,
      "grad_norm": 0.45900288224220276,
      "learning_rate": 6.626237189781206e-06,
      "loss": 0.3341,
      "step": 26585
    },
    {
      "epoch": 1.2200449726951494,
      "grad_norm": 0.43939143419265747,
      "learning_rate": 6.626005331043772e-06,
      "loss": 0.3188,
      "step": 26586
    },
    {
      "epoch": 1.220090863200404,
      "grad_norm": 0.4664939343929291,
      "learning_rate": 6.625773468396309e-06,
      "loss": 0.334,
      "step": 26587
    },
    {
      "epoch": 1.2201367537056582,
      "grad_norm": 0.5194387435913086,
      "learning_rate": 6.625541601839374e-06,
      "loss": 0.4502,
      "step": 26588
    },
    {
      "epoch": 1.2201826442109127,
      "grad_norm": 0.5061895847320557,
      "learning_rate": 6.625309731373523e-06,
      "loss": 0.401,
      "step": 26589
    },
    {
      "epoch": 1.2202285347161672,
      "grad_norm": 0.4548039138317108,
      "learning_rate": 6.6250778569993155e-06,
      "loss": 0.2948,
      "step": 26590
    },
    {
      "epoch": 1.2202744252214217,
      "grad_norm": 0.46734148263931274,
      "learning_rate": 6.624845978717309e-06,
      "loss": 0.3494,
      "step": 26591
    },
    {
      "epoch": 1.2203203157266762,
      "grad_norm": 0.47018906474113464,
      "learning_rate": 6.624614096528061e-06,
      "loss": 0.3263,
      "step": 26592
    },
    {
      "epoch": 1.2203662062319307,
      "grad_norm": 0.49716323614120483,
      "learning_rate": 6.624382210432128e-06,
      "loss": 0.3946,
      "step": 26593
    },
    {
      "epoch": 1.220412096737185,
      "grad_norm": 0.48397278785705566,
      "learning_rate": 6.624150320430069e-06,
      "loss": 0.3766,
      "step": 26594
    },
    {
      "epoch": 1.2204579872424395,
      "grad_norm": 0.47527289390563965,
      "learning_rate": 6.623918426522441e-06,
      "loss": 0.3758,
      "step": 26595
    },
    {
      "epoch": 1.220503877747694,
      "grad_norm": 0.477151095867157,
      "learning_rate": 6.623686528709802e-06,
      "loss": 0.4356,
      "step": 26596
    },
    {
      "epoch": 1.2205497682529485,
      "grad_norm": 0.5443152785301208,
      "learning_rate": 6.623454626992709e-06,
      "loss": 0.3504,
      "step": 26597
    },
    {
      "epoch": 1.220595658758203,
      "grad_norm": 0.47887110710144043,
      "learning_rate": 6.623222721371721e-06,
      "loss": 0.4004,
      "step": 26598
    },
    {
      "epoch": 1.2206415492634575,
      "grad_norm": 0.46839427947998047,
      "learning_rate": 6.622990811847394e-06,
      "loss": 0.331,
      "step": 26599
    },
    {
      "epoch": 1.220687439768712,
      "grad_norm": 0.4455777406692505,
      "learning_rate": 6.622758898420285e-06,
      "loss": 0.3123,
      "step": 26600
    },
    {
      "epoch": 1.2207333302739662,
      "grad_norm": 0.4671574831008911,
      "learning_rate": 6.622526981090955e-06,
      "loss": 0.3371,
      "step": 26601
    },
    {
      "epoch": 1.2207792207792207,
      "grad_norm": 0.49463027715682983,
      "learning_rate": 6.622295059859959e-06,
      "loss": 0.4429,
      "step": 26602
    },
    {
      "epoch": 1.2208251112844752,
      "grad_norm": 0.44564488530158997,
      "learning_rate": 6.622063134727855e-06,
      "loss": 0.3624,
      "step": 26603
    },
    {
      "epoch": 1.2208710017897297,
      "grad_norm": 0.47399866580963135,
      "learning_rate": 6.621831205695202e-06,
      "loss": 0.3729,
      "step": 26604
    },
    {
      "epoch": 1.2209168922949842,
      "grad_norm": 0.47438278794288635,
      "learning_rate": 6.621599272762555e-06,
      "loss": 0.3911,
      "step": 26605
    },
    {
      "epoch": 1.2209627828002385,
      "grad_norm": 0.480461061000824,
      "learning_rate": 6.621367335930475e-06,
      "loss": 0.3966,
      "step": 26606
    },
    {
      "epoch": 1.221008673305493,
      "grad_norm": 0.48052120208740234,
      "learning_rate": 6.621135395199518e-06,
      "loss": 0.3406,
      "step": 26607
    },
    {
      "epoch": 1.2210545638107475,
      "grad_norm": 0.5039651393890381,
      "learning_rate": 6.620903450570242e-06,
      "loss": 0.4136,
      "step": 26608
    },
    {
      "epoch": 1.221100454316002,
      "grad_norm": 0.47502970695495605,
      "learning_rate": 6.620671502043205e-06,
      "loss": 0.4025,
      "step": 26609
    },
    {
      "epoch": 1.2211463448212565,
      "grad_norm": 0.4570407271385193,
      "learning_rate": 6.620439549618965e-06,
      "loss": 0.3472,
      "step": 26610
    },
    {
      "epoch": 1.221192235326511,
      "grad_norm": 0.4864865839481354,
      "learning_rate": 6.620207593298078e-06,
      "loss": 0.363,
      "step": 26611
    },
    {
      "epoch": 1.2212381258317655,
      "grad_norm": 0.44479483366012573,
      "learning_rate": 6.6199756330811036e-06,
      "loss": 0.3208,
      "step": 26612
    },
    {
      "epoch": 1.22128401633702,
      "grad_norm": 0.46428120136260986,
      "learning_rate": 6.619743668968599e-06,
      "loss": 0.3307,
      "step": 26613
    },
    {
      "epoch": 1.2213299068422743,
      "grad_norm": 0.42367973923683167,
      "learning_rate": 6.619511700961122e-06,
      "loss": 0.2835,
      "step": 26614
    },
    {
      "epoch": 1.2213757973475288,
      "grad_norm": 0.4806128144264221,
      "learning_rate": 6.619279729059231e-06,
      "loss": 0.3788,
      "step": 26615
    },
    {
      "epoch": 1.2214216878527833,
      "grad_norm": 0.44946739077568054,
      "learning_rate": 6.6190477532634835e-06,
      "loss": 0.3261,
      "step": 26616
    },
    {
      "epoch": 1.2214675783580378,
      "grad_norm": 0.5040525197982788,
      "learning_rate": 6.618815773574436e-06,
      "loss": 0.3887,
      "step": 26617
    },
    {
      "epoch": 1.2215134688632923,
      "grad_norm": 0.5048179030418396,
      "learning_rate": 6.618583789992649e-06,
      "loss": 0.3933,
      "step": 26618
    },
    {
      "epoch": 1.2215593593685465,
      "grad_norm": 0.5670266151428223,
      "learning_rate": 6.6183518025186775e-06,
      "loss": 0.4758,
      "step": 26619
    },
    {
      "epoch": 1.221605249873801,
      "grad_norm": 0.4258519411087036,
      "learning_rate": 6.618119811153081e-06,
      "loss": 0.3006,
      "step": 26620
    },
    {
      "epoch": 1.2216511403790555,
      "grad_norm": 0.4902563989162445,
      "learning_rate": 6.617887815896418e-06,
      "loss": 0.3195,
      "step": 26621
    },
    {
      "epoch": 1.22169703088431,
      "grad_norm": 0.46177878975868225,
      "learning_rate": 6.617655816749245e-06,
      "loss": 0.3574,
      "step": 26622
    },
    {
      "epoch": 1.2217429213895645,
      "grad_norm": 0.46349120140075684,
      "learning_rate": 6.617423813712119e-06,
      "loss": 0.3945,
      "step": 26623
    },
    {
      "epoch": 1.221788811894819,
      "grad_norm": 0.43407168984413147,
      "learning_rate": 6.617191806785601e-06,
      "loss": 0.2791,
      "step": 26624
    },
    {
      "epoch": 1.2218347024000735,
      "grad_norm": 0.45522263646125793,
      "learning_rate": 6.616959795970246e-06,
      "loss": 0.3159,
      "step": 26625
    },
    {
      "epoch": 1.2218805929053278,
      "grad_norm": 0.44813239574432373,
      "learning_rate": 6.616727781266612e-06,
      "loss": 0.3023,
      "step": 26626
    },
    {
      "epoch": 1.2219264834105823,
      "grad_norm": 0.48098501563072205,
      "learning_rate": 6.61649576267526e-06,
      "loss": 0.3839,
      "step": 26627
    },
    {
      "epoch": 1.2219723739158368,
      "grad_norm": 0.4383300542831421,
      "learning_rate": 6.616263740196744e-06,
      "loss": 0.3167,
      "step": 26628
    },
    {
      "epoch": 1.2220182644210913,
      "grad_norm": 0.4839453399181366,
      "learning_rate": 6.616031713831625e-06,
      "loss": 0.4099,
      "step": 26629
    },
    {
      "epoch": 1.2220641549263458,
      "grad_norm": 0.44403553009033203,
      "learning_rate": 6.6157996835804616e-06,
      "loss": 0.2893,
      "step": 26630
    },
    {
      "epoch": 1.2221100454316003,
      "grad_norm": 0.46382936835289,
      "learning_rate": 6.615567649443807e-06,
      "loss": 0.3223,
      "step": 26631
    },
    {
      "epoch": 1.2221559359368546,
      "grad_norm": 0.514137864112854,
      "learning_rate": 6.615335611422224e-06,
      "loss": 0.4226,
      "step": 26632
    },
    {
      "epoch": 1.222201826442109,
      "grad_norm": 0.4380362629890442,
      "learning_rate": 6.615103569516267e-06,
      "loss": 0.3039,
      "step": 26633
    },
    {
      "epoch": 1.2222477169473636,
      "grad_norm": 0.47946828603744507,
      "learning_rate": 6.614871523726496e-06,
      "loss": 0.3945,
      "step": 26634
    },
    {
      "epoch": 1.222293607452618,
      "grad_norm": 0.4449380040168762,
      "learning_rate": 6.614639474053471e-06,
      "loss": 0.3262,
      "step": 26635
    },
    {
      "epoch": 1.2223394979578726,
      "grad_norm": 0.45133545994758606,
      "learning_rate": 6.614407420497744e-06,
      "loss": 0.3285,
      "step": 26636
    },
    {
      "epoch": 1.222385388463127,
      "grad_norm": 0.4084988832473755,
      "learning_rate": 6.614175363059878e-06,
      "loss": 0.2693,
      "step": 26637
    },
    {
      "epoch": 1.2224312789683816,
      "grad_norm": 0.4470478892326355,
      "learning_rate": 6.613943301740431e-06,
      "loss": 0.3217,
      "step": 26638
    },
    {
      "epoch": 1.2224771694736358,
      "grad_norm": 0.44927704334259033,
      "learning_rate": 6.613711236539958e-06,
      "loss": 0.3392,
      "step": 26639
    },
    {
      "epoch": 1.2225230599788903,
      "grad_norm": 0.41971129179000854,
      "learning_rate": 6.6134791674590205e-06,
      "loss": 0.2846,
      "step": 26640
    },
    {
      "epoch": 1.2225689504841448,
      "grad_norm": 0.46104666590690613,
      "learning_rate": 6.613247094498174e-06,
      "loss": 0.3361,
      "step": 26641
    },
    {
      "epoch": 1.2226148409893993,
      "grad_norm": 0.5026975274085999,
      "learning_rate": 6.613015017657977e-06,
      "loss": 0.4027,
      "step": 26642
    },
    {
      "epoch": 1.2226607314946538,
      "grad_norm": 0.457438200712204,
      "learning_rate": 6.612782936938987e-06,
      "loss": 0.3322,
      "step": 26643
    },
    {
      "epoch": 1.222706621999908,
      "grad_norm": 0.5010858774185181,
      "learning_rate": 6.612550852341766e-06,
      "loss": 0.4125,
      "step": 26644
    },
    {
      "epoch": 1.2227525125051626,
      "grad_norm": 0.5267516374588013,
      "learning_rate": 6.612318763866866e-06,
      "loss": 0.4613,
      "step": 26645
    },
    {
      "epoch": 1.222798403010417,
      "grad_norm": 0.4591442346572876,
      "learning_rate": 6.612086671514851e-06,
      "loss": 0.3525,
      "step": 26646
    },
    {
      "epoch": 1.2228442935156716,
      "grad_norm": 0.4552706778049469,
      "learning_rate": 6.611854575286275e-06,
      "loss": 0.3012,
      "step": 26647
    },
    {
      "epoch": 1.222890184020926,
      "grad_norm": 0.5300567746162415,
      "learning_rate": 6.611622475181697e-06,
      "loss": 0.4432,
      "step": 26648
    },
    {
      "epoch": 1.2229360745261806,
      "grad_norm": 0.48310747742652893,
      "learning_rate": 6.611390371201677e-06,
      "loss": 0.377,
      "step": 26649
    },
    {
      "epoch": 1.222981965031435,
      "grad_norm": 0.4536762237548828,
      "learning_rate": 6.61115826334677e-06,
      "loss": 0.3634,
      "step": 26650
    },
    {
      "epoch": 1.2230278555366896,
      "grad_norm": 0.5072546005249023,
      "learning_rate": 6.610926151617537e-06,
      "loss": 0.4282,
      "step": 26651
    },
    {
      "epoch": 1.2230737460419439,
      "grad_norm": 0.46690991520881653,
      "learning_rate": 6.610694036014535e-06,
      "loss": 0.3988,
      "step": 26652
    },
    {
      "epoch": 1.2231196365471984,
      "grad_norm": 0.4949071705341339,
      "learning_rate": 6.6104619165383225e-06,
      "loss": 0.3899,
      "step": 26653
    },
    {
      "epoch": 1.2231655270524529,
      "grad_norm": 0.47922465205192566,
      "learning_rate": 6.610229793189456e-06,
      "loss": 0.3448,
      "step": 26654
    },
    {
      "epoch": 1.2232114175577073,
      "grad_norm": 0.4879293441772461,
      "learning_rate": 6.609997665968496e-06,
      "loss": 0.3738,
      "step": 26655
    },
    {
      "epoch": 1.2232573080629618,
      "grad_norm": 0.481689453125,
      "learning_rate": 6.609765534875999e-06,
      "loss": 0.4249,
      "step": 26656
    },
    {
      "epoch": 1.2233031985682161,
      "grad_norm": 0.4812041223049164,
      "learning_rate": 6.609533399912524e-06,
      "loss": 0.3779,
      "step": 26657
    },
    {
      "epoch": 1.2233490890734706,
      "grad_norm": 0.5270069241523743,
      "learning_rate": 6.6093012610786315e-06,
      "loss": 0.3324,
      "step": 26658
    },
    {
      "epoch": 1.2233949795787251,
      "grad_norm": 0.48506709933280945,
      "learning_rate": 6.609069118374875e-06,
      "loss": 0.3507,
      "step": 26659
    },
    {
      "epoch": 1.2234408700839796,
      "grad_norm": 0.47983241081237793,
      "learning_rate": 6.6088369718018155e-06,
      "loss": 0.3793,
      "step": 26660
    },
    {
      "epoch": 1.2234867605892341,
      "grad_norm": 0.47340551018714905,
      "learning_rate": 6.60860482136001e-06,
      "loss": 0.3265,
      "step": 26661
    },
    {
      "epoch": 1.2235326510944886,
      "grad_norm": 0.4626866579055786,
      "learning_rate": 6.60837266705002e-06,
      "loss": 0.3706,
      "step": 26662
    },
    {
      "epoch": 1.223578541599743,
      "grad_norm": 0.5242679715156555,
      "learning_rate": 6.6081405088724e-06,
      "loss": 0.3932,
      "step": 26663
    },
    {
      "epoch": 1.2236244321049974,
      "grad_norm": 0.500994086265564,
      "learning_rate": 6.607908346827709e-06,
      "loss": 0.3713,
      "step": 26664
    },
    {
      "epoch": 1.2236703226102519,
      "grad_norm": 0.44292595982551575,
      "learning_rate": 6.607676180916507e-06,
      "loss": 0.3095,
      "step": 26665
    },
    {
      "epoch": 1.2237162131155064,
      "grad_norm": 0.43451201915740967,
      "learning_rate": 6.60744401113935e-06,
      "loss": 0.3078,
      "step": 26666
    },
    {
      "epoch": 1.2237621036207609,
      "grad_norm": 0.49092504382133484,
      "learning_rate": 6.607211837496799e-06,
      "loss": 0.4224,
      "step": 26667
    },
    {
      "epoch": 1.2238079941260154,
      "grad_norm": 0.4264480173587799,
      "learning_rate": 6.6069796599894095e-06,
      "loss": 0.2876,
      "step": 26668
    },
    {
      "epoch": 1.2238538846312699,
      "grad_norm": 0.44917911291122437,
      "learning_rate": 6.606747478617742e-06,
      "loss": 0.3676,
      "step": 26669
    },
    {
      "epoch": 1.2238997751365241,
      "grad_norm": 0.4808778762817383,
      "learning_rate": 6.606515293382353e-06,
      "loss": 0.4156,
      "step": 26670
    },
    {
      "epoch": 1.2239456656417786,
      "grad_norm": 0.48629358410835266,
      "learning_rate": 6.606283104283801e-06,
      "loss": 0.3727,
      "step": 26671
    },
    {
      "epoch": 1.2239915561470331,
      "grad_norm": 0.47963541746139526,
      "learning_rate": 6.606050911322648e-06,
      "loss": 0.4065,
      "step": 26672
    },
    {
      "epoch": 1.2240374466522876,
      "grad_norm": 0.45682913064956665,
      "learning_rate": 6.605818714499448e-06,
      "loss": 0.3318,
      "step": 26673
    },
    {
      "epoch": 1.2240833371575421,
      "grad_norm": 0.47557318210601807,
      "learning_rate": 6.605586513814759e-06,
      "loss": 0.3235,
      "step": 26674
    },
    {
      "epoch": 1.2241292276627966,
      "grad_norm": 0.4608170986175537,
      "learning_rate": 6.605354309269144e-06,
      "loss": 0.3425,
      "step": 26675
    },
    {
      "epoch": 1.2241751181680511,
      "grad_norm": 0.5554015040397644,
      "learning_rate": 6.605122100863156e-06,
      "loss": 0.4969,
      "step": 26676
    },
    {
      "epoch": 1.2242210086733054,
      "grad_norm": 0.4447786211967468,
      "learning_rate": 6.604889888597357e-06,
      "loss": 0.326,
      "step": 26677
    },
    {
      "epoch": 1.22426689917856,
      "grad_norm": 0.5328484773635864,
      "learning_rate": 6.604657672472306e-06,
      "loss": 0.3906,
      "step": 26678
    },
    {
      "epoch": 1.2243127896838144,
      "grad_norm": 0.47341039776802063,
      "learning_rate": 6.604425452488558e-06,
      "loss": 0.3613,
      "step": 26679
    },
    {
      "epoch": 1.224358680189069,
      "grad_norm": 0.42413538694381714,
      "learning_rate": 6.604193228646674e-06,
      "loss": 0.2839,
      "step": 26680
    },
    {
      "epoch": 1.2244045706943234,
      "grad_norm": 0.4877067804336548,
      "learning_rate": 6.60396100094721e-06,
      "loss": 0.3349,
      "step": 26681
    },
    {
      "epoch": 1.224450461199578,
      "grad_norm": 0.45451751351356506,
      "learning_rate": 6.6037287693907275e-06,
      "loss": 0.3211,
      "step": 26682
    },
    {
      "epoch": 1.2244963517048322,
      "grad_norm": 0.4359849989414215,
      "learning_rate": 6.603496533977782e-06,
      "loss": 0.3061,
      "step": 26683
    },
    {
      "epoch": 1.2245422422100867,
      "grad_norm": 0.46876490116119385,
      "learning_rate": 6.6032642947089355e-06,
      "loss": 0.3669,
      "step": 26684
    },
    {
      "epoch": 1.2245881327153412,
      "grad_norm": 0.49478214979171753,
      "learning_rate": 6.603032051584744e-06,
      "loss": 0.4302,
      "step": 26685
    },
    {
      "epoch": 1.2246340232205957,
      "grad_norm": 0.4658558666706085,
      "learning_rate": 6.602799804605765e-06,
      "loss": 0.3306,
      "step": 26686
    },
    {
      "epoch": 1.2246799137258502,
      "grad_norm": 0.4269648492336273,
      "learning_rate": 6.602567553772559e-06,
      "loss": 0.3003,
      "step": 26687
    },
    {
      "epoch": 1.2247258042311047,
      "grad_norm": 0.45072057843208313,
      "learning_rate": 6.602335299085683e-06,
      "loss": 0.3522,
      "step": 26688
    },
    {
      "epoch": 1.2247716947363592,
      "grad_norm": 0.49658888578414917,
      "learning_rate": 6.6021030405456975e-06,
      "loss": 0.4318,
      "step": 26689
    },
    {
      "epoch": 1.2248175852416134,
      "grad_norm": 0.46530285477638245,
      "learning_rate": 6.60187077815316e-06,
      "loss": 0.3191,
      "step": 26690
    },
    {
      "epoch": 1.224863475746868,
      "grad_norm": 0.4730072319507599,
      "learning_rate": 6.6016385119086276e-06,
      "loss": 0.3655,
      "step": 26691
    },
    {
      "epoch": 1.2249093662521224,
      "grad_norm": 0.5185195803642273,
      "learning_rate": 6.601406241812661e-06,
      "loss": 0.398,
      "step": 26692
    },
    {
      "epoch": 1.224955256757377,
      "grad_norm": 0.49943414330482483,
      "learning_rate": 6.601173967865817e-06,
      "loss": 0.4051,
      "step": 26693
    },
    {
      "epoch": 1.2250011472626314,
      "grad_norm": 0.5599116683006287,
      "learning_rate": 6.600941690068656e-06,
      "loss": 0.3842,
      "step": 26694
    },
    {
      "epoch": 1.2250470377678857,
      "grad_norm": 0.4581001102924347,
      "learning_rate": 6.600709408421735e-06,
      "loss": 0.3438,
      "step": 26695
    },
    {
      "epoch": 1.2250929282731402,
      "grad_norm": 0.48061642050743103,
      "learning_rate": 6.600477122925613e-06,
      "loss": 0.3622,
      "step": 26696
    },
    {
      "epoch": 1.2251388187783947,
      "grad_norm": 0.4949869215488434,
      "learning_rate": 6.600244833580849e-06,
      "loss": 0.3871,
      "step": 26697
    },
    {
      "epoch": 1.2251847092836492,
      "grad_norm": 0.45928436517715454,
      "learning_rate": 6.600012540388001e-06,
      "loss": 0.3336,
      "step": 26698
    },
    {
      "epoch": 1.2252305997889037,
      "grad_norm": 0.48367682099342346,
      "learning_rate": 6.599780243347627e-06,
      "loss": 0.4157,
      "step": 26699
    },
    {
      "epoch": 1.2252764902941582,
      "grad_norm": 0.45616573095321655,
      "learning_rate": 6.599547942460287e-06,
      "loss": 0.3361,
      "step": 26700
    },
    {
      "epoch": 1.2253223807994127,
      "grad_norm": 0.4276859760284424,
      "learning_rate": 6.59931563772654e-06,
      "loss": 0.2667,
      "step": 26701
    },
    {
      "epoch": 1.2253682713046672,
      "grad_norm": 0.4788587689399719,
      "learning_rate": 6.5990833291469425e-06,
      "loss": 0.3476,
      "step": 26702
    },
    {
      "epoch": 1.2254141618099215,
      "grad_norm": 0.5263826847076416,
      "learning_rate": 6.598851016722054e-06,
      "loss": 0.3683,
      "step": 26703
    },
    {
      "epoch": 1.225460052315176,
      "grad_norm": 0.4685773253440857,
      "learning_rate": 6.598618700452435e-06,
      "loss": 0.3313,
      "step": 26704
    },
    {
      "epoch": 1.2255059428204305,
      "grad_norm": 0.45673033595085144,
      "learning_rate": 6.598386380338642e-06,
      "loss": 0.3552,
      "step": 26705
    },
    {
      "epoch": 1.225551833325685,
      "grad_norm": 0.44203996658325195,
      "learning_rate": 6.598154056381234e-06,
      "loss": 0.3027,
      "step": 26706
    },
    {
      "epoch": 1.2255977238309395,
      "grad_norm": 0.47695299983024597,
      "learning_rate": 6.59792172858077e-06,
      "loss": 0.3732,
      "step": 26707
    },
    {
      "epoch": 1.2256436143361937,
      "grad_norm": 0.46547001600265503,
      "learning_rate": 6.5976893969378065e-06,
      "loss": 0.3783,
      "step": 26708
    },
    {
      "epoch": 1.2256895048414482,
      "grad_norm": 0.47934696078300476,
      "learning_rate": 6.597457061452907e-06,
      "loss": 0.3194,
      "step": 26709
    },
    {
      "epoch": 1.2257353953467027,
      "grad_norm": 0.46296289563179016,
      "learning_rate": 6.597224722126626e-06,
      "loss": 0.3844,
      "step": 26710
    },
    {
      "epoch": 1.2257812858519572,
      "grad_norm": 0.44655686616897583,
      "learning_rate": 6.596992378959525e-06,
      "loss": 0.352,
      "step": 26711
    },
    {
      "epoch": 1.2258271763572117,
      "grad_norm": 0.4813513457775116,
      "learning_rate": 6.59676003195216e-06,
      "loss": 0.3595,
      "step": 26712
    },
    {
      "epoch": 1.2258730668624662,
      "grad_norm": 0.4577640891075134,
      "learning_rate": 6.596527681105092e-06,
      "loss": 0.3397,
      "step": 26713
    },
    {
      "epoch": 1.2259189573677207,
      "grad_norm": 0.4611766040325165,
      "learning_rate": 6.596295326418878e-06,
      "loss": 0.3521,
      "step": 26714
    },
    {
      "epoch": 1.225964847872975,
      "grad_norm": 0.4652535915374756,
      "learning_rate": 6.596062967894077e-06,
      "loss": 0.3547,
      "step": 26715
    },
    {
      "epoch": 1.2260107383782295,
      "grad_norm": 0.49030205607414246,
      "learning_rate": 6.59583060553125e-06,
      "loss": 0.418,
      "step": 26716
    },
    {
      "epoch": 1.226056628883484,
      "grad_norm": 0.5197466611862183,
      "learning_rate": 6.595598239330954e-06,
      "loss": 0.42,
      "step": 26717
    },
    {
      "epoch": 1.2261025193887385,
      "grad_norm": 0.48115867376327515,
      "learning_rate": 6.595365869293746e-06,
      "loss": 0.4039,
      "step": 26718
    },
    {
      "epoch": 1.226148409893993,
      "grad_norm": 0.4573347568511963,
      "learning_rate": 6.5951334954201875e-06,
      "loss": 0.3679,
      "step": 26719
    },
    {
      "epoch": 1.2261943003992475,
      "grad_norm": 0.49239447712898254,
      "learning_rate": 6.594901117710837e-06,
      "loss": 0.4232,
      "step": 26720
    },
    {
      "epoch": 1.2262401909045018,
      "grad_norm": 0.4402032196521759,
      "learning_rate": 6.594668736166252e-06,
      "loss": 0.2833,
      "step": 26721
    },
    {
      "epoch": 1.2262860814097563,
      "grad_norm": 0.4782503545284271,
      "learning_rate": 6.5944363507869926e-06,
      "loss": 0.3787,
      "step": 26722
    },
    {
      "epoch": 1.2263319719150108,
      "grad_norm": 0.47473153471946716,
      "learning_rate": 6.594203961573617e-06,
      "loss": 0.3923,
      "step": 26723
    },
    {
      "epoch": 1.2263778624202653,
      "grad_norm": 0.47355198860168457,
      "learning_rate": 6.593971568526683e-06,
      "loss": 0.3625,
      "step": 26724
    },
    {
      "epoch": 1.2264237529255198,
      "grad_norm": 0.44638296961784363,
      "learning_rate": 6.59373917164675e-06,
      "loss": 0.3068,
      "step": 26725
    },
    {
      "epoch": 1.2264696434307742,
      "grad_norm": 0.47800934314727783,
      "learning_rate": 6.593506770934379e-06,
      "loss": 0.3767,
      "step": 26726
    },
    {
      "epoch": 1.2265155339360287,
      "grad_norm": 0.4054155945777893,
      "learning_rate": 6.593274366390128e-06,
      "loss": 0.285,
      "step": 26727
    },
    {
      "epoch": 1.226561424441283,
      "grad_norm": 0.4662257134914398,
      "learning_rate": 6.593041958014553e-06,
      "loss": 0.3606,
      "step": 26728
    },
    {
      "epoch": 1.2266073149465375,
      "grad_norm": 0.5130449533462524,
      "learning_rate": 6.592809545808215e-06,
      "loss": 0.4042,
      "step": 26729
    },
    {
      "epoch": 1.226653205451792,
      "grad_norm": 0.4662942588329315,
      "learning_rate": 6.592577129771672e-06,
      "loss": 0.3165,
      "step": 26730
    },
    {
      "epoch": 1.2266990959570465,
      "grad_norm": 0.480329692363739,
      "learning_rate": 6.592344709905486e-06,
      "loss": 0.4103,
      "step": 26731
    },
    {
      "epoch": 1.226744986462301,
      "grad_norm": 0.45333853363990784,
      "learning_rate": 6.592112286210211e-06,
      "loss": 0.3518,
      "step": 26732
    },
    {
      "epoch": 1.2267908769675553,
      "grad_norm": 0.44146671891212463,
      "learning_rate": 6.591879858686411e-06,
      "loss": 0.3206,
      "step": 26733
    },
    {
      "epoch": 1.2268367674728098,
      "grad_norm": 0.4588165879249573,
      "learning_rate": 6.591647427334641e-06,
      "loss": 0.2941,
      "step": 26734
    },
    {
      "epoch": 1.2268826579780643,
      "grad_norm": 0.5139807462692261,
      "learning_rate": 6.5914149921554605e-06,
      "loss": 0.4425,
      "step": 26735
    },
    {
      "epoch": 1.2269285484833188,
      "grad_norm": 0.4632120728492737,
      "learning_rate": 6.59118255314943e-06,
      "loss": 0.374,
      "step": 26736
    },
    {
      "epoch": 1.2269744389885733,
      "grad_norm": 0.4953192174434662,
      "learning_rate": 6.590950110317108e-06,
      "loss": 0.3077,
      "step": 26737
    },
    {
      "epoch": 1.2270203294938278,
      "grad_norm": 0.4845247268676758,
      "learning_rate": 6.590717663659054e-06,
      "loss": 0.3685,
      "step": 26738
    },
    {
      "epoch": 1.2270662199990823,
      "grad_norm": 0.4843023121356964,
      "learning_rate": 6.590485213175823e-06,
      "loss": 0.3594,
      "step": 26739
    },
    {
      "epoch": 1.2271121105043368,
      "grad_norm": 0.48512545228004456,
      "learning_rate": 6.59025275886798e-06,
      "loss": 0.3811,
      "step": 26740
    },
    {
      "epoch": 1.227158001009591,
      "grad_norm": 0.49913138151168823,
      "learning_rate": 6.59002030073608e-06,
      "loss": 0.3861,
      "step": 26741
    },
    {
      "epoch": 1.2272038915148455,
      "grad_norm": 0.4929863214492798,
      "learning_rate": 6.589787838780682e-06,
      "loss": 0.4037,
      "step": 26742
    },
    {
      "epoch": 1.2272497820201,
      "grad_norm": 0.49018481373786926,
      "learning_rate": 6.589555373002347e-06,
      "loss": 0.4216,
      "step": 26743
    },
    {
      "epoch": 1.2272956725253545,
      "grad_norm": 0.45413222908973694,
      "learning_rate": 6.589322903401634e-06,
      "loss": 0.3564,
      "step": 26744
    },
    {
      "epoch": 1.227341563030609,
      "grad_norm": 0.48070651292800903,
      "learning_rate": 6.589090429979101e-06,
      "loss": 0.3251,
      "step": 26745
    },
    {
      "epoch": 1.2273874535358633,
      "grad_norm": 0.47393009066581726,
      "learning_rate": 6.588857952735305e-06,
      "loss": 0.3235,
      "step": 26746
    },
    {
      "epoch": 1.2274333440411178,
      "grad_norm": 0.48133018612861633,
      "learning_rate": 6.5886254716708066e-06,
      "loss": 0.3205,
      "step": 26747
    },
    {
      "epoch": 1.2274792345463723,
      "grad_norm": 0.44927507638931274,
      "learning_rate": 6.588392986786167e-06,
      "loss": 0.3194,
      "step": 26748
    },
    {
      "epoch": 1.2275251250516268,
      "grad_norm": 0.5140783190727234,
      "learning_rate": 6.588160498081945e-06,
      "loss": 0.3846,
      "step": 26749
    },
    {
      "epoch": 1.2275710155568813,
      "grad_norm": 0.4788554906845093,
      "learning_rate": 6.587928005558696e-06,
      "loss": 0.386,
      "step": 26750
    },
    {
      "epoch": 1.2276169060621358,
      "grad_norm": 0.47465676069259644,
      "learning_rate": 6.587695509216983e-06,
      "loss": 0.3351,
      "step": 26751
    },
    {
      "epoch": 1.2276627965673903,
      "grad_norm": 0.4933360815048218,
      "learning_rate": 6.587463009057361e-06,
      "loss": 0.3599,
      "step": 26752
    },
    {
      "epoch": 1.2277086870726446,
      "grad_norm": 0.43314623832702637,
      "learning_rate": 6.587230505080392e-06,
      "loss": 0.3161,
      "step": 26753
    },
    {
      "epoch": 1.227754577577899,
      "grad_norm": 0.48525261878967285,
      "learning_rate": 6.586997997286636e-06,
      "loss": 0.3938,
      "step": 26754
    },
    {
      "epoch": 1.2278004680831536,
      "grad_norm": 0.4501439929008484,
      "learning_rate": 6.586765485676651e-06,
      "loss": 0.3563,
      "step": 26755
    },
    {
      "epoch": 1.227846358588408,
      "grad_norm": 0.47431784868240356,
      "learning_rate": 6.586532970250994e-06,
      "loss": 0.327,
      "step": 26756
    },
    {
      "epoch": 1.2278922490936626,
      "grad_norm": 0.5014031529426575,
      "learning_rate": 6.586300451010227e-06,
      "loss": 0.3329,
      "step": 26757
    },
    {
      "epoch": 1.227938139598917,
      "grad_norm": 0.4664098918437958,
      "learning_rate": 6.586067927954907e-06,
      "loss": 0.3508,
      "step": 26758
    },
    {
      "epoch": 1.2279840301041713,
      "grad_norm": 0.5229552388191223,
      "learning_rate": 6.585835401085597e-06,
      "loss": 0.432,
      "step": 26759
    },
    {
      "epoch": 1.2280299206094258,
      "grad_norm": 0.46785643696784973,
      "learning_rate": 6.585602870402851e-06,
      "loss": 0.3618,
      "step": 26760
    },
    {
      "epoch": 1.2280758111146803,
      "grad_norm": 0.4325353801250458,
      "learning_rate": 6.585370335907231e-06,
      "loss": 0.2929,
      "step": 26761
    },
    {
      "epoch": 1.2281217016199348,
      "grad_norm": 0.5042179226875305,
      "learning_rate": 6.585137797599296e-06,
      "loss": 0.4254,
      "step": 26762
    },
    {
      "epoch": 1.2281675921251893,
      "grad_norm": 0.5022868514060974,
      "learning_rate": 6.584905255479603e-06,
      "loss": 0.4651,
      "step": 26763
    },
    {
      "epoch": 1.2282134826304438,
      "grad_norm": 0.4896521270275116,
      "learning_rate": 6.584672709548714e-06,
      "loss": 0.3747,
      "step": 26764
    },
    {
      "epoch": 1.2282593731356983,
      "grad_norm": 0.4874773621559143,
      "learning_rate": 6.584440159807188e-06,
      "loss": 0.4197,
      "step": 26765
    },
    {
      "epoch": 1.2283052636409526,
      "grad_norm": 0.4975726008415222,
      "learning_rate": 6.584207606255585e-06,
      "loss": 0.4746,
      "step": 26766
    },
    {
      "epoch": 1.228351154146207,
      "grad_norm": 0.46541446447372437,
      "learning_rate": 6.58397504889446e-06,
      "loss": 0.3522,
      "step": 26767
    },
    {
      "epoch": 1.2283970446514616,
      "grad_norm": 0.42959362268447876,
      "learning_rate": 6.583742487724377e-06,
      "loss": 0.3029,
      "step": 26768
    },
    {
      "epoch": 1.228442935156716,
      "grad_norm": 0.49454110860824585,
      "learning_rate": 6.583509922745892e-06,
      "loss": 0.3915,
      "step": 26769
    },
    {
      "epoch": 1.2284888256619706,
      "grad_norm": 0.462556928396225,
      "learning_rate": 6.583277353959565e-06,
      "loss": 0.4012,
      "step": 26770
    },
    {
      "epoch": 1.228534716167225,
      "grad_norm": 0.4470585584640503,
      "learning_rate": 6.583044781365957e-06,
      "loss": 0.3354,
      "step": 26771
    },
    {
      "epoch": 1.2285806066724794,
      "grad_norm": 0.4601151943206787,
      "learning_rate": 6.582812204965625e-06,
      "loss": 0.349,
      "step": 26772
    },
    {
      "epoch": 1.2286264971777339,
      "grad_norm": 0.508791983127594,
      "learning_rate": 6.582579624759129e-06,
      "loss": 0.3977,
      "step": 26773
    },
    {
      "epoch": 1.2286723876829884,
      "grad_norm": 0.47290509939193726,
      "learning_rate": 6.582347040747029e-06,
      "loss": 0.3758,
      "step": 26774
    },
    {
      "epoch": 1.2287182781882429,
      "grad_norm": 0.43615254759788513,
      "learning_rate": 6.582114452929883e-06,
      "loss": 0.3073,
      "step": 26775
    },
    {
      "epoch": 1.2287641686934974,
      "grad_norm": 0.500091016292572,
      "learning_rate": 6.581881861308253e-06,
      "loss": 0.367,
      "step": 26776
    },
    {
      "epoch": 1.2288100591987519,
      "grad_norm": 0.4326653778553009,
      "learning_rate": 6.581649265882695e-06,
      "loss": 0.3306,
      "step": 26777
    },
    {
      "epoch": 1.2288559497040064,
      "grad_norm": 0.4728100895881653,
      "learning_rate": 6.581416666653768e-06,
      "loss": 0.3392,
      "step": 26778
    },
    {
      "epoch": 1.2289018402092606,
      "grad_norm": 0.49063777923583984,
      "learning_rate": 6.581184063622036e-06,
      "loss": 0.3878,
      "step": 26779
    },
    {
      "epoch": 1.2289477307145151,
      "grad_norm": 0.5307078957557678,
      "learning_rate": 6.580951456788054e-06,
      "loss": 0.4702,
      "step": 26780
    },
    {
      "epoch": 1.2289936212197696,
      "grad_norm": 0.46190381050109863,
      "learning_rate": 6.580718846152383e-06,
      "loss": 0.313,
      "step": 26781
    },
    {
      "epoch": 1.2290395117250241,
      "grad_norm": 0.4525344967842102,
      "learning_rate": 6.580486231715582e-06,
      "loss": 0.3299,
      "step": 26782
    },
    {
      "epoch": 1.2290854022302786,
      "grad_norm": 0.4335699677467346,
      "learning_rate": 6.58025361347821e-06,
      "loss": 0.3196,
      "step": 26783
    },
    {
      "epoch": 1.229131292735533,
      "grad_norm": 0.5410346984863281,
      "learning_rate": 6.5800209914408266e-06,
      "loss": 0.4164,
      "step": 26784
    },
    {
      "epoch": 1.2291771832407874,
      "grad_norm": 0.47890084981918335,
      "learning_rate": 6.579788365603992e-06,
      "loss": 0.3683,
      "step": 26785
    },
    {
      "epoch": 1.229223073746042,
      "grad_norm": 0.4500812590122223,
      "learning_rate": 6.579555735968265e-06,
      "loss": 0.3211,
      "step": 26786
    },
    {
      "epoch": 1.2292689642512964,
      "grad_norm": 0.49350765347480774,
      "learning_rate": 6.579323102534205e-06,
      "loss": 0.405,
      "step": 26787
    },
    {
      "epoch": 1.229314854756551,
      "grad_norm": 0.503693699836731,
      "learning_rate": 6.579090465302371e-06,
      "loss": 0.4402,
      "step": 26788
    },
    {
      "epoch": 1.2293607452618054,
      "grad_norm": 0.4721743166446686,
      "learning_rate": 6.578857824273323e-06,
      "loss": 0.3828,
      "step": 26789
    },
    {
      "epoch": 1.2294066357670599,
      "grad_norm": 0.483810693025589,
      "learning_rate": 6.57862517944762e-06,
      "loss": 0.3364,
      "step": 26790
    },
    {
      "epoch": 1.2294525262723142,
      "grad_norm": 0.4622254967689514,
      "learning_rate": 6.578392530825823e-06,
      "loss": 0.35,
      "step": 26791
    },
    {
      "epoch": 1.2294984167775687,
      "grad_norm": 0.4170423448085785,
      "learning_rate": 6.578159878408489e-06,
      "loss": 0.2992,
      "step": 26792
    },
    {
      "epoch": 1.2295443072828232,
      "grad_norm": 0.49348220229148865,
      "learning_rate": 6.577927222196179e-06,
      "loss": 0.3829,
      "step": 26793
    },
    {
      "epoch": 1.2295901977880777,
      "grad_norm": 0.49790406227111816,
      "learning_rate": 6.577694562189452e-06,
      "loss": 0.3119,
      "step": 26794
    },
    {
      "epoch": 1.2296360882933322,
      "grad_norm": 0.4708254933357239,
      "learning_rate": 6.577461898388866e-06,
      "loss": 0.3073,
      "step": 26795
    },
    {
      "epoch": 1.2296819787985867,
      "grad_norm": 0.4812576472759247,
      "learning_rate": 6.5772292307949835e-06,
      "loss": 0.3589,
      "step": 26796
    },
    {
      "epoch": 1.229727869303841,
      "grad_norm": 0.4936792850494385,
      "learning_rate": 6.576996559408362e-06,
      "loss": 0.4122,
      "step": 26797
    },
    {
      "epoch": 1.2297737598090954,
      "grad_norm": 0.4703616797924042,
      "learning_rate": 6.576763884229562e-06,
      "loss": 0.3229,
      "step": 26798
    },
    {
      "epoch": 1.22981965031435,
      "grad_norm": 0.4486231207847595,
      "learning_rate": 6.576531205259142e-06,
      "loss": 0.3465,
      "step": 26799
    },
    {
      "epoch": 1.2298655408196044,
      "grad_norm": 0.49977901577949524,
      "learning_rate": 6.576298522497663e-06,
      "loss": 0.3883,
      "step": 26800
    },
    {
      "epoch": 1.229911431324859,
      "grad_norm": 0.4543660879135132,
      "learning_rate": 6.576065835945682e-06,
      "loss": 0.2982,
      "step": 26801
    },
    {
      "epoch": 1.2299573218301134,
      "grad_norm": 0.45072004199028015,
      "learning_rate": 6.575833145603761e-06,
      "loss": 0.3036,
      "step": 26802
    },
    {
      "epoch": 1.230003212335368,
      "grad_norm": 0.45425841212272644,
      "learning_rate": 6.575600451472459e-06,
      "loss": 0.3604,
      "step": 26803
    },
    {
      "epoch": 1.2300491028406222,
      "grad_norm": 0.4760380685329437,
      "learning_rate": 6.575367753552334e-06,
      "loss": 0.3756,
      "step": 26804
    },
    {
      "epoch": 1.2300949933458767,
      "grad_norm": 0.507774829864502,
      "learning_rate": 6.575135051843948e-06,
      "loss": 0.4101,
      "step": 26805
    },
    {
      "epoch": 1.2301408838511312,
      "grad_norm": 0.4474605917930603,
      "learning_rate": 6.574902346347858e-06,
      "loss": 0.3411,
      "step": 26806
    },
    {
      "epoch": 1.2301867743563857,
      "grad_norm": 0.4946867525577545,
      "learning_rate": 6.574669637064626e-06,
      "loss": 0.398,
      "step": 26807
    },
    {
      "epoch": 1.2302326648616402,
      "grad_norm": 0.4577609896659851,
      "learning_rate": 6.574436923994809e-06,
      "loss": 0.3508,
      "step": 26808
    },
    {
      "epoch": 1.2302785553668947,
      "grad_norm": 0.4675033390522003,
      "learning_rate": 6.57420420713897e-06,
      "loss": 0.3497,
      "step": 26809
    },
    {
      "epoch": 1.230324445872149,
      "grad_norm": 0.44263747334480286,
      "learning_rate": 6.573971486497667e-06,
      "loss": 0.2828,
      "step": 26810
    },
    {
      "epoch": 1.2303703363774035,
      "grad_norm": 0.4902964234352112,
      "learning_rate": 6.5737387620714574e-06,
      "loss": 0.3919,
      "step": 26811
    },
    {
      "epoch": 1.230416226882658,
      "grad_norm": 0.4896829128265381,
      "learning_rate": 6.573506033860904e-06,
      "loss": 0.4181,
      "step": 26812
    },
    {
      "epoch": 1.2304621173879124,
      "grad_norm": 0.46180468797683716,
      "learning_rate": 6.573273301866565e-06,
      "loss": 0.3498,
      "step": 26813
    },
    {
      "epoch": 1.230508007893167,
      "grad_norm": 0.4772176146507263,
      "learning_rate": 6.573040566089001e-06,
      "loss": 0.3521,
      "step": 26814
    },
    {
      "epoch": 1.2305538983984214,
      "grad_norm": 0.47646379470825195,
      "learning_rate": 6.572807826528769e-06,
      "loss": 0.3462,
      "step": 26815
    },
    {
      "epoch": 1.230599788903676,
      "grad_norm": 0.4549470543861389,
      "learning_rate": 6.572575083186432e-06,
      "loss": 0.3323,
      "step": 26816
    },
    {
      "epoch": 1.2306456794089302,
      "grad_norm": 0.5174025297164917,
      "learning_rate": 6.5723423360625475e-06,
      "loss": 0.4986,
      "step": 26817
    },
    {
      "epoch": 1.2306915699141847,
      "grad_norm": 0.4619024395942688,
      "learning_rate": 6.572109585157676e-06,
      "loss": 0.3496,
      "step": 26818
    },
    {
      "epoch": 1.2307374604194392,
      "grad_norm": 0.4593014419078827,
      "learning_rate": 6.5718768304723764e-06,
      "loss": 0.3146,
      "step": 26819
    },
    {
      "epoch": 1.2307833509246937,
      "grad_norm": 0.4824551045894623,
      "learning_rate": 6.571644072007211e-06,
      "loss": 0.3638,
      "step": 26820
    },
    {
      "epoch": 1.2308292414299482,
      "grad_norm": 0.469211608171463,
      "learning_rate": 6.571411309762736e-06,
      "loss": 0.3814,
      "step": 26821
    },
    {
      "epoch": 1.2308751319352025,
      "grad_norm": 0.47740665078163147,
      "learning_rate": 6.5711785437395145e-06,
      "loss": 0.3401,
      "step": 26822
    },
    {
      "epoch": 1.230921022440457,
      "grad_norm": 0.476235955953598,
      "learning_rate": 6.570945773938103e-06,
      "loss": 0.3494,
      "step": 26823
    },
    {
      "epoch": 1.2309669129457115,
      "grad_norm": 0.4675827622413635,
      "learning_rate": 6.5707130003590625e-06,
      "loss": 0.3653,
      "step": 26824
    },
    {
      "epoch": 1.231012803450966,
      "grad_norm": 0.45169585943222046,
      "learning_rate": 6.570480223002955e-06,
      "loss": 0.3657,
      "step": 26825
    },
    {
      "epoch": 1.2310586939562205,
      "grad_norm": 0.48313525319099426,
      "learning_rate": 6.570247441870336e-06,
      "loss": 0.4119,
      "step": 26826
    },
    {
      "epoch": 1.231104584461475,
      "grad_norm": 0.45709624886512756,
      "learning_rate": 6.57001465696177e-06,
      "loss": 0.3206,
      "step": 26827
    },
    {
      "epoch": 1.2311504749667295,
      "grad_norm": 0.45487353205680847,
      "learning_rate": 6.569781868277812e-06,
      "loss": 0.301,
      "step": 26828
    },
    {
      "epoch": 1.231196365471984,
      "grad_norm": 0.508780837059021,
      "learning_rate": 6.569549075819024e-06,
      "loss": 0.3975,
      "step": 26829
    },
    {
      "epoch": 1.2312422559772382,
      "grad_norm": 0.4841972589492798,
      "learning_rate": 6.569316279585967e-06,
      "loss": 0.3536,
      "step": 26830
    },
    {
      "epoch": 1.2312881464824927,
      "grad_norm": 0.47733545303344727,
      "learning_rate": 6.569083479579199e-06,
      "loss": 0.4017,
      "step": 26831
    },
    {
      "epoch": 1.2313340369877472,
      "grad_norm": 0.4763106405735016,
      "learning_rate": 6.568850675799281e-06,
      "loss": 0.374,
      "step": 26832
    },
    {
      "epoch": 1.2313799274930017,
      "grad_norm": 0.4054230749607086,
      "learning_rate": 6.568617868246771e-06,
      "loss": 0.2554,
      "step": 26833
    },
    {
      "epoch": 1.2314258179982562,
      "grad_norm": 0.48868876695632935,
      "learning_rate": 6.5683850569222305e-06,
      "loss": 0.3971,
      "step": 26834
    },
    {
      "epoch": 1.2314717085035105,
      "grad_norm": 0.4994022250175476,
      "learning_rate": 6.5681522418262204e-06,
      "loss": 0.4521,
      "step": 26835
    },
    {
      "epoch": 1.231517599008765,
      "grad_norm": 0.4670258164405823,
      "learning_rate": 6.567919422959299e-06,
      "loss": 0.3863,
      "step": 26836
    },
    {
      "epoch": 1.2315634895140195,
      "grad_norm": 0.4919496178627014,
      "learning_rate": 6.5676866003220245e-06,
      "loss": 0.4444,
      "step": 26837
    },
    {
      "epoch": 1.231609380019274,
      "grad_norm": 0.40840062499046326,
      "learning_rate": 6.567453773914959e-06,
      "loss": 0.2712,
      "step": 26838
    },
    {
      "epoch": 1.2316552705245285,
      "grad_norm": 0.5102701783180237,
      "learning_rate": 6.5672209437386625e-06,
      "loss": 0.4313,
      "step": 26839
    },
    {
      "epoch": 1.231701161029783,
      "grad_norm": 0.4413122534751892,
      "learning_rate": 6.566988109793693e-06,
      "loss": 0.3124,
      "step": 26840
    },
    {
      "epoch": 1.2317470515350375,
      "grad_norm": 0.43845438957214355,
      "learning_rate": 6.566755272080614e-06,
      "loss": 0.3198,
      "step": 26841
    },
    {
      "epoch": 1.2317929420402918,
      "grad_norm": 0.4456599950790405,
      "learning_rate": 6.566522430599982e-06,
      "loss": 0.2847,
      "step": 26842
    },
    {
      "epoch": 1.2318388325455463,
      "grad_norm": 0.7040122747421265,
      "learning_rate": 6.566289585352356e-06,
      "loss": 0.3266,
      "step": 26843
    },
    {
      "epoch": 1.2318847230508008,
      "grad_norm": 0.48630186915397644,
      "learning_rate": 6.5660567363383e-06,
      "loss": 0.3942,
      "step": 26844
    },
    {
      "epoch": 1.2319306135560553,
      "grad_norm": 0.5093220472335815,
      "learning_rate": 6.565823883558371e-06,
      "loss": 0.3951,
      "step": 26845
    },
    {
      "epoch": 1.2319765040613098,
      "grad_norm": 0.47909197211265564,
      "learning_rate": 6.56559102701313e-06,
      "loss": 0.2966,
      "step": 26846
    },
    {
      "epoch": 1.2320223945665643,
      "grad_norm": 0.5358510613441467,
      "learning_rate": 6.565358166703137e-06,
      "loss": 0.3791,
      "step": 26847
    },
    {
      "epoch": 1.2320682850718185,
      "grad_norm": 0.5173203945159912,
      "learning_rate": 6.565125302628952e-06,
      "loss": 0.3836,
      "step": 26848
    },
    {
      "epoch": 1.232114175577073,
      "grad_norm": 0.4838600754737854,
      "learning_rate": 6.564892434791134e-06,
      "loss": 0.3832,
      "step": 26849
    },
    {
      "epoch": 1.2321600660823275,
      "grad_norm": 0.45927414298057556,
      "learning_rate": 6.5646595631902436e-06,
      "loss": 0.2631,
      "step": 26850
    },
    {
      "epoch": 1.232205956587582,
      "grad_norm": 0.4793802797794342,
      "learning_rate": 6.564426687826841e-06,
      "loss": 0.3356,
      "step": 26851
    },
    {
      "epoch": 1.2322518470928365,
      "grad_norm": 0.4469888210296631,
      "learning_rate": 6.564193808701486e-06,
      "loss": 0.3038,
      "step": 26852
    },
    {
      "epoch": 1.232297737598091,
      "grad_norm": 0.4813072085380554,
      "learning_rate": 6.563960925814739e-06,
      "loss": 0.3151,
      "step": 26853
    },
    {
      "epoch": 1.2323436281033455,
      "grad_norm": 0.5374873280525208,
      "learning_rate": 6.563728039167158e-06,
      "loss": 0.4666,
      "step": 26854
    },
    {
      "epoch": 1.2323895186085998,
      "grad_norm": 0.4790475368499756,
      "learning_rate": 6.563495148759305e-06,
      "loss": 0.3576,
      "step": 26855
    },
    {
      "epoch": 1.2324354091138543,
      "grad_norm": 0.5061254501342773,
      "learning_rate": 6.56326225459174e-06,
      "loss": 0.3917,
      "step": 26856
    },
    {
      "epoch": 1.2324812996191088,
      "grad_norm": 0.47590959072113037,
      "learning_rate": 6.563029356665022e-06,
      "loss": 0.3691,
      "step": 26857
    },
    {
      "epoch": 1.2325271901243633,
      "grad_norm": 0.4405446946620941,
      "learning_rate": 6.562796454979714e-06,
      "loss": 0.3329,
      "step": 26858
    },
    {
      "epoch": 1.2325730806296178,
      "grad_norm": 0.4576973617076874,
      "learning_rate": 6.562563549536371e-06,
      "loss": 0.3423,
      "step": 26859
    },
    {
      "epoch": 1.2326189711348723,
      "grad_norm": 0.4621069133281708,
      "learning_rate": 6.562330640335557e-06,
      "loss": 0.3791,
      "step": 26860
    },
    {
      "epoch": 1.2326648616401266,
      "grad_norm": 0.42332491278648376,
      "learning_rate": 6.562097727377832e-06,
      "loss": 0.2838,
      "step": 26861
    },
    {
      "epoch": 1.232710752145381,
      "grad_norm": 0.4770330488681793,
      "learning_rate": 6.561864810663753e-06,
      "loss": 0.3204,
      "step": 26862
    },
    {
      "epoch": 1.2327566426506356,
      "grad_norm": 0.4412253201007843,
      "learning_rate": 6.5616318901938836e-06,
      "loss": 0.309,
      "step": 26863
    },
    {
      "epoch": 1.23280253315589,
      "grad_norm": 0.484821081161499,
      "learning_rate": 6.561398965968781e-06,
      "loss": 0.4124,
      "step": 26864
    },
    {
      "epoch": 1.2328484236611446,
      "grad_norm": 0.46991825103759766,
      "learning_rate": 6.561166037989009e-06,
      "loss": 0.3227,
      "step": 26865
    },
    {
      "epoch": 1.232894314166399,
      "grad_norm": 0.46430015563964844,
      "learning_rate": 6.560933106255122e-06,
      "loss": 0.3774,
      "step": 26866
    },
    {
      "epoch": 1.2329402046716536,
      "grad_norm": 0.46950358152389526,
      "learning_rate": 6.560700170767686e-06,
      "loss": 0.3203,
      "step": 26867
    },
    {
      "epoch": 1.2329860951769078,
      "grad_norm": 0.4655531942844391,
      "learning_rate": 6.560467231527258e-06,
      "loss": 0.3326,
      "step": 26868
    },
    {
      "epoch": 1.2330319856821623,
      "grad_norm": 0.5244982242584229,
      "learning_rate": 6.560234288534398e-06,
      "loss": 0.4344,
      "step": 26869
    },
    {
      "epoch": 1.2330778761874168,
      "grad_norm": 0.4458554685115814,
      "learning_rate": 6.560001341789669e-06,
      "loss": 0.3256,
      "step": 26870
    },
    {
      "epoch": 1.2331237666926713,
      "grad_norm": 0.45476922392845154,
      "learning_rate": 6.559768391293627e-06,
      "loss": 0.3349,
      "step": 26871
    },
    {
      "epoch": 1.2331696571979258,
      "grad_norm": 0.481824666261673,
      "learning_rate": 6.559535437046836e-06,
      "loss": 0.3337,
      "step": 26872
    },
    {
      "epoch": 1.23321554770318,
      "grad_norm": 0.45806920528411865,
      "learning_rate": 6.559302479049852e-06,
      "loss": 0.3283,
      "step": 26873
    },
    {
      "epoch": 1.2332614382084346,
      "grad_norm": 0.454267293214798,
      "learning_rate": 6.55906951730324e-06,
      "loss": 0.3487,
      "step": 26874
    },
    {
      "epoch": 1.233307328713689,
      "grad_norm": 0.45353686809539795,
      "learning_rate": 6.5588365518075594e-06,
      "loss": 0.3321,
      "step": 26875
    },
    {
      "epoch": 1.2333532192189436,
      "grad_norm": 0.4311416745185852,
      "learning_rate": 6.558603582563366e-06,
      "loss": 0.2918,
      "step": 26876
    },
    {
      "epoch": 1.233399109724198,
      "grad_norm": 0.48102760314941406,
      "learning_rate": 6.558370609571223e-06,
      "loss": 0.409,
      "step": 26877
    },
    {
      "epoch": 1.2334450002294526,
      "grad_norm": 0.45825618505477905,
      "learning_rate": 6.5581376328316916e-06,
      "loss": 0.3382,
      "step": 26878
    },
    {
      "epoch": 1.233490890734707,
      "grad_norm": 0.47809654474258423,
      "learning_rate": 6.557904652345329e-06,
      "loss": 0.3492,
      "step": 26879
    },
    {
      "epoch": 1.2335367812399614,
      "grad_norm": 0.48459911346435547,
      "learning_rate": 6.5576716681126995e-06,
      "loss": 0.3929,
      "step": 26880
    },
    {
      "epoch": 1.2335826717452159,
      "grad_norm": 0.4668811857700348,
      "learning_rate": 6.557438680134361e-06,
      "loss": 0.3612,
      "step": 26881
    },
    {
      "epoch": 1.2336285622504704,
      "grad_norm": 0.4894791543483734,
      "learning_rate": 6.557205688410873e-06,
      "loss": 0.3296,
      "step": 26882
    },
    {
      "epoch": 1.2336744527557248,
      "grad_norm": 0.45420366525650024,
      "learning_rate": 6.5569726929427965e-06,
      "loss": 0.3452,
      "step": 26883
    },
    {
      "epoch": 1.2337203432609793,
      "grad_norm": 0.4797327518463135,
      "learning_rate": 6.556739693730694e-06,
      "loss": 0.3514,
      "step": 26884
    },
    {
      "epoch": 1.2337662337662338,
      "grad_norm": 0.4625242352485657,
      "learning_rate": 6.556506690775124e-06,
      "loss": 0.3346,
      "step": 26885
    },
    {
      "epoch": 1.2338121242714881,
      "grad_norm": 0.49694740772247314,
      "learning_rate": 6.556273684076644e-06,
      "loss": 0.4041,
      "step": 26886
    },
    {
      "epoch": 1.2338580147767426,
      "grad_norm": 0.44436293840408325,
      "learning_rate": 6.556040673635819e-06,
      "loss": 0.2896,
      "step": 26887
    },
    {
      "epoch": 1.2339039052819971,
      "grad_norm": 0.4799072742462158,
      "learning_rate": 6.555807659453206e-06,
      "loss": 0.4117,
      "step": 26888
    },
    {
      "epoch": 1.2339497957872516,
      "grad_norm": 0.48397886753082275,
      "learning_rate": 6.555574641529368e-06,
      "loss": 0.3754,
      "step": 26889
    },
    {
      "epoch": 1.233995686292506,
      "grad_norm": 0.4468926191329956,
      "learning_rate": 6.555341619864863e-06,
      "loss": 0.3261,
      "step": 26890
    },
    {
      "epoch": 1.2340415767977606,
      "grad_norm": 0.4418638050556183,
      "learning_rate": 6.555108594460253e-06,
      "loss": 0.3159,
      "step": 26891
    },
    {
      "epoch": 1.234087467303015,
      "grad_norm": 0.47592806816101074,
      "learning_rate": 6.554875565316098e-06,
      "loss": 0.3464,
      "step": 26892
    },
    {
      "epoch": 1.2341333578082694,
      "grad_norm": 0.490142285823822,
      "learning_rate": 6.554642532432956e-06,
      "loss": 0.329,
      "step": 26893
    },
    {
      "epoch": 1.2341792483135239,
      "grad_norm": 0.4986588954925537,
      "learning_rate": 6.55440949581139e-06,
      "loss": 0.3877,
      "step": 26894
    },
    {
      "epoch": 1.2342251388187784,
      "grad_norm": 0.4800160527229309,
      "learning_rate": 6.554176455451961e-06,
      "loss": 0.3859,
      "step": 26895
    },
    {
      "epoch": 1.2342710293240329,
      "grad_norm": 0.4542986750602722,
      "learning_rate": 6.553943411355228e-06,
      "loss": 0.3504,
      "step": 26896
    },
    {
      "epoch": 1.2343169198292874,
      "grad_norm": 0.48610973358154297,
      "learning_rate": 6.55371036352175e-06,
      "loss": 0.3409,
      "step": 26897
    },
    {
      "epoch": 1.2343628103345419,
      "grad_norm": 0.44587013125419617,
      "learning_rate": 6.553477311952091e-06,
      "loss": 0.321,
      "step": 26898
    },
    {
      "epoch": 1.2344087008397961,
      "grad_norm": 0.4322940409183502,
      "learning_rate": 6.553244256646808e-06,
      "loss": 0.282,
      "step": 26899
    },
    {
      "epoch": 1.2344545913450506,
      "grad_norm": 0.5377055406570435,
      "learning_rate": 6.5530111976064625e-06,
      "loss": 0.4434,
      "step": 26900
    },
    {
      "epoch": 1.2345004818503051,
      "grad_norm": 0.46853041648864746,
      "learning_rate": 6.552778134831615e-06,
      "loss": 0.3261,
      "step": 26901
    },
    {
      "epoch": 1.2345463723555596,
      "grad_norm": 0.4879055619239807,
      "learning_rate": 6.552545068322829e-06,
      "loss": 0.3213,
      "step": 26902
    },
    {
      "epoch": 1.2345922628608141,
      "grad_norm": 0.45755067467689514,
      "learning_rate": 6.552311998080658e-06,
      "loss": 0.3143,
      "step": 26903
    },
    {
      "epoch": 1.2346381533660686,
      "grad_norm": 0.4872325360774994,
      "learning_rate": 6.55207892410567e-06,
      "loss": 0.4006,
      "step": 26904
    },
    {
      "epoch": 1.2346840438713231,
      "grad_norm": 0.47123241424560547,
      "learning_rate": 6.55184584639842e-06,
      "loss": 0.403,
      "step": 26905
    },
    {
      "epoch": 1.2347299343765774,
      "grad_norm": 0.4868556261062622,
      "learning_rate": 6.551612764959472e-06,
      "loss": 0.3382,
      "step": 26906
    },
    {
      "epoch": 1.234775824881832,
      "grad_norm": 0.4847387671470642,
      "learning_rate": 6.551379679789385e-06,
      "loss": 0.3544,
      "step": 26907
    },
    {
      "epoch": 1.2348217153870864,
      "grad_norm": 0.46393391489982605,
      "learning_rate": 6.551146590888718e-06,
      "loss": 0.3015,
      "step": 26908
    },
    {
      "epoch": 1.234867605892341,
      "grad_norm": 0.47640296816825867,
      "learning_rate": 6.550913498258033e-06,
      "loss": 0.2767,
      "step": 26909
    },
    {
      "epoch": 1.2349134963975954,
      "grad_norm": 0.4468924403190613,
      "learning_rate": 6.550680401897891e-06,
      "loss": 0.3234,
      "step": 26910
    },
    {
      "epoch": 1.2349593869028497,
      "grad_norm": 0.4720715880393982,
      "learning_rate": 6.5504473018088525e-06,
      "loss": 0.3577,
      "step": 26911
    },
    {
      "epoch": 1.2350052774081042,
      "grad_norm": 0.4553471505641937,
      "learning_rate": 6.550214197991477e-06,
      "loss": 0.3775,
      "step": 26912
    },
    {
      "epoch": 1.2350511679133587,
      "grad_norm": 0.4348546862602234,
      "learning_rate": 6.549981090446327e-06,
      "loss": 0.2837,
      "step": 26913
    },
    {
      "epoch": 1.2350970584186132,
      "grad_norm": 0.49493667483329773,
      "learning_rate": 6.549747979173959e-06,
      "loss": 0.3908,
      "step": 26914
    },
    {
      "epoch": 1.2351429489238677,
      "grad_norm": 0.4428330659866333,
      "learning_rate": 6.54951486417494e-06,
      "loss": 0.2846,
      "step": 26915
    },
    {
      "epoch": 1.2351888394291222,
      "grad_norm": 0.46119460463523865,
      "learning_rate": 6.549281745449822e-06,
      "loss": 0.2908,
      "step": 26916
    },
    {
      "epoch": 1.2352347299343767,
      "grad_norm": 0.4523824453353882,
      "learning_rate": 6.549048622999174e-06,
      "loss": 0.3318,
      "step": 26917
    },
    {
      "epoch": 1.2352806204396312,
      "grad_norm": 0.5092880725860596,
      "learning_rate": 6.5488154968235525e-06,
      "loss": 0.5022,
      "step": 26918
    },
    {
      "epoch": 1.2353265109448854,
      "grad_norm": 0.5115890502929688,
      "learning_rate": 6.5485823669235175e-06,
      "loss": 0.3908,
      "step": 26919
    },
    {
      "epoch": 1.23537240145014,
      "grad_norm": 0.4918225407600403,
      "learning_rate": 6.548349233299631e-06,
      "loss": 0.3776,
      "step": 26920
    },
    {
      "epoch": 1.2354182919553944,
      "grad_norm": 0.49637696146965027,
      "learning_rate": 6.5481160959524535e-06,
      "loss": 0.4294,
      "step": 26921
    },
    {
      "epoch": 1.235464182460649,
      "grad_norm": 0.4784561097621918,
      "learning_rate": 6.547882954882545e-06,
      "loss": 0.3609,
      "step": 26922
    },
    {
      "epoch": 1.2355100729659034,
      "grad_norm": 0.4702552855014801,
      "learning_rate": 6.547649810090467e-06,
      "loss": 0.3588,
      "step": 26923
    },
    {
      "epoch": 1.2355559634711577,
      "grad_norm": 0.44575244188308716,
      "learning_rate": 6.54741666157678e-06,
      "loss": 0.3504,
      "step": 26924
    },
    {
      "epoch": 1.2356018539764122,
      "grad_norm": 0.44645342230796814,
      "learning_rate": 6.547183509342043e-06,
      "loss": 0.3583,
      "step": 26925
    },
    {
      "epoch": 1.2356477444816667,
      "grad_norm": 0.49024203419685364,
      "learning_rate": 6.54695035338682e-06,
      "loss": 0.3396,
      "step": 26926
    },
    {
      "epoch": 1.2356936349869212,
      "grad_norm": 0.46850958466529846,
      "learning_rate": 6.546717193711667e-06,
      "loss": 0.4147,
      "step": 26927
    },
    {
      "epoch": 1.2357395254921757,
      "grad_norm": 0.5124899744987488,
      "learning_rate": 6.546484030317149e-06,
      "loss": 0.3334,
      "step": 26928
    },
    {
      "epoch": 1.2357854159974302,
      "grad_norm": 0.4713386297225952,
      "learning_rate": 6.546250863203825e-06,
      "loss": 0.4174,
      "step": 26929
    },
    {
      "epoch": 1.2358313065026847,
      "grad_norm": 0.44826072454452515,
      "learning_rate": 6.546017692372255e-06,
      "loss": 0.3338,
      "step": 26930
    },
    {
      "epoch": 1.235877197007939,
      "grad_norm": 0.43639466166496277,
      "learning_rate": 6.545784517823001e-06,
      "loss": 0.2978,
      "step": 26931
    },
    {
      "epoch": 1.2359230875131935,
      "grad_norm": 0.46823742985725403,
      "learning_rate": 6.545551339556623e-06,
      "loss": 0.3326,
      "step": 26932
    },
    {
      "epoch": 1.235968978018448,
      "grad_norm": 0.4965291917324066,
      "learning_rate": 6.545318157573681e-06,
      "loss": 0.3488,
      "step": 26933
    },
    {
      "epoch": 1.2360148685237025,
      "grad_norm": 0.487789124250412,
      "learning_rate": 6.545084971874738e-06,
      "loss": 0.4168,
      "step": 26934
    },
    {
      "epoch": 1.236060759028957,
      "grad_norm": 0.4685787260532379,
      "learning_rate": 6.544851782460353e-06,
      "loss": 0.3856,
      "step": 26935
    },
    {
      "epoch": 1.2361066495342115,
      "grad_norm": 0.4797455668449402,
      "learning_rate": 6.544618589331086e-06,
      "loss": 0.3865,
      "step": 26936
    },
    {
      "epoch": 1.2361525400394657,
      "grad_norm": 0.5031578540802002,
      "learning_rate": 6.544385392487499e-06,
      "loss": 0.392,
      "step": 26937
    },
    {
      "epoch": 1.2361984305447202,
      "grad_norm": 0.5607506036758423,
      "learning_rate": 6.544152191930153e-06,
      "loss": 0.4603,
      "step": 26938
    },
    {
      "epoch": 1.2362443210499747,
      "grad_norm": 0.4962524175643921,
      "learning_rate": 6.543918987659609e-06,
      "loss": 0.3546,
      "step": 26939
    },
    {
      "epoch": 1.2362902115552292,
      "grad_norm": 0.4590578079223633,
      "learning_rate": 6.543685779676426e-06,
      "loss": 0.3234,
      "step": 26940
    },
    {
      "epoch": 1.2363361020604837,
      "grad_norm": 0.4443036615848541,
      "learning_rate": 6.543452567981165e-06,
      "loss": 0.2957,
      "step": 26941
    },
    {
      "epoch": 1.2363819925657382,
      "grad_norm": 0.44369983673095703,
      "learning_rate": 6.543219352574389e-06,
      "loss": 0.3133,
      "step": 26942
    },
    {
      "epoch": 1.2364278830709927,
      "grad_norm": 0.48077884316444397,
      "learning_rate": 6.542986133456658e-06,
      "loss": 0.3768,
      "step": 26943
    },
    {
      "epoch": 1.236473773576247,
      "grad_norm": 0.4984472095966339,
      "learning_rate": 6.542752910628531e-06,
      "loss": 0.442,
      "step": 26944
    },
    {
      "epoch": 1.2365196640815015,
      "grad_norm": 0.47822272777557373,
      "learning_rate": 6.542519684090571e-06,
      "loss": 0.3725,
      "step": 26945
    },
    {
      "epoch": 1.236565554586756,
      "grad_norm": 0.47450053691864014,
      "learning_rate": 6.54228645384334e-06,
      "loss": 0.3869,
      "step": 26946
    },
    {
      "epoch": 1.2366114450920105,
      "grad_norm": 0.4498036801815033,
      "learning_rate": 6.5420532198873935e-06,
      "loss": 0.2921,
      "step": 26947
    },
    {
      "epoch": 1.236657335597265,
      "grad_norm": 0.5055592060089111,
      "learning_rate": 6.541819982223296e-06,
      "loss": 0.4103,
      "step": 26948
    },
    {
      "epoch": 1.2367032261025195,
      "grad_norm": 0.47839269042015076,
      "learning_rate": 6.541586740851609e-06,
      "loss": 0.3183,
      "step": 26949
    },
    {
      "epoch": 1.2367491166077738,
      "grad_norm": 0.4832196533679962,
      "learning_rate": 6.541353495772893e-06,
      "loss": 0.3915,
      "step": 26950
    },
    {
      "epoch": 1.2367950071130283,
      "grad_norm": 0.4669204652309418,
      "learning_rate": 6.541120246987708e-06,
      "loss": 0.3561,
      "step": 26951
    },
    {
      "epoch": 1.2368408976182828,
      "grad_norm": 0.49969688057899475,
      "learning_rate": 6.540886994496616e-06,
      "loss": 0.4243,
      "step": 26952
    },
    {
      "epoch": 1.2368867881235373,
      "grad_norm": 0.47471991181373596,
      "learning_rate": 6.5406537383001745e-06,
      "loss": 0.3639,
      "step": 26953
    },
    {
      "epoch": 1.2369326786287917,
      "grad_norm": 0.47805774211883545,
      "learning_rate": 6.5404204783989475e-06,
      "loss": 0.3665,
      "step": 26954
    },
    {
      "epoch": 1.2369785691340462,
      "grad_norm": 0.46443846821784973,
      "learning_rate": 6.540187214793497e-06,
      "loss": 0.3429,
      "step": 26955
    },
    {
      "epoch": 1.2370244596393007,
      "grad_norm": 0.48014798760414124,
      "learning_rate": 6.539953947484383e-06,
      "loss": 0.3773,
      "step": 26956
    },
    {
      "epoch": 1.237070350144555,
      "grad_norm": 0.500694990158081,
      "learning_rate": 6.539720676472165e-06,
      "loss": 0.4159,
      "step": 26957
    },
    {
      "epoch": 1.2371162406498095,
      "grad_norm": 0.4682350158691406,
      "learning_rate": 6.539487401757404e-06,
      "loss": 0.3489,
      "step": 26958
    },
    {
      "epoch": 1.237162131155064,
      "grad_norm": 0.4632737636566162,
      "learning_rate": 6.539254123340661e-06,
      "loss": 0.3415,
      "step": 26959
    },
    {
      "epoch": 1.2372080216603185,
      "grad_norm": 0.6620158553123474,
      "learning_rate": 6.539020841222499e-06,
      "loss": 0.3913,
      "step": 26960
    },
    {
      "epoch": 1.237253912165573,
      "grad_norm": 0.5854431390762329,
      "learning_rate": 6.538787555403477e-06,
      "loss": 0.3687,
      "step": 26961
    },
    {
      "epoch": 1.2372998026708273,
      "grad_norm": 1.7830449342727661,
      "learning_rate": 6.5385542658841575e-06,
      "loss": 0.3288,
      "step": 26962
    },
    {
      "epoch": 1.2373456931760818,
      "grad_norm": 0.4354385435581207,
      "learning_rate": 6.5383209726651e-06,
      "loss": 0.2974,
      "step": 26963
    },
    {
      "epoch": 1.2373915836813363,
      "grad_norm": 0.45981118083000183,
      "learning_rate": 6.538087675746865e-06,
      "loss": 0.3721,
      "step": 26964
    },
    {
      "epoch": 1.2374374741865908,
      "grad_norm": 0.4643864631652832,
      "learning_rate": 6.537854375130015e-06,
      "loss": 0.3472,
      "step": 26965
    },
    {
      "epoch": 1.2374833646918453,
      "grad_norm": 0.5077660083770752,
      "learning_rate": 6.537621070815112e-06,
      "loss": 0.3968,
      "step": 26966
    },
    {
      "epoch": 1.2375292551970998,
      "grad_norm": 0.4875354766845703,
      "learning_rate": 6.537387762802715e-06,
      "loss": 0.3605,
      "step": 26967
    },
    {
      "epoch": 1.2375751457023543,
      "grad_norm": 0.4593157172203064,
      "learning_rate": 6.537154451093385e-06,
      "loss": 0.3803,
      "step": 26968
    },
    {
      "epoch": 1.2376210362076085,
      "grad_norm": 0.5361263155937195,
      "learning_rate": 6.536921135687684e-06,
      "loss": 0.448,
      "step": 26969
    },
    {
      "epoch": 1.237666926712863,
      "grad_norm": 0.43269380927085876,
      "learning_rate": 6.5366878165861725e-06,
      "loss": 0.2993,
      "step": 26970
    },
    {
      "epoch": 1.2377128172181175,
      "grad_norm": 0.4363768994808197,
      "learning_rate": 6.536454493789413e-06,
      "loss": 0.3202,
      "step": 26971
    },
    {
      "epoch": 1.237758707723372,
      "grad_norm": 0.5158581733703613,
      "learning_rate": 6.536221167297965e-06,
      "loss": 0.4569,
      "step": 26972
    },
    {
      "epoch": 1.2378045982286265,
      "grad_norm": 0.4847043752670288,
      "learning_rate": 6.535987837112389e-06,
      "loss": 0.4479,
      "step": 26973
    },
    {
      "epoch": 1.237850488733881,
      "grad_norm": 0.4664457142353058,
      "learning_rate": 6.535754503233247e-06,
      "loss": 0.4061,
      "step": 26974
    },
    {
      "epoch": 1.2378963792391353,
      "grad_norm": 0.41358092427253723,
      "learning_rate": 6.535521165661101e-06,
      "loss": 0.301,
      "step": 26975
    },
    {
      "epoch": 1.2379422697443898,
      "grad_norm": 0.4339200556278229,
      "learning_rate": 6.535287824396509e-06,
      "loss": 0.3009,
      "step": 26976
    },
    {
      "epoch": 1.2379881602496443,
      "grad_norm": 0.5721866488456726,
      "learning_rate": 6.535054479440037e-06,
      "loss": 0.3713,
      "step": 26977
    },
    {
      "epoch": 1.2380340507548988,
      "grad_norm": 0.4904795289039612,
      "learning_rate": 6.5348211307922425e-06,
      "loss": 0.3775,
      "step": 26978
    },
    {
      "epoch": 1.2380799412601533,
      "grad_norm": 0.45070287585258484,
      "learning_rate": 6.534587778453687e-06,
      "loss": 0.341,
      "step": 26979
    },
    {
      "epoch": 1.2381258317654078,
      "grad_norm": 0.4584466516971588,
      "learning_rate": 6.534354422424933e-06,
      "loss": 0.3777,
      "step": 26980
    },
    {
      "epoch": 1.2381717222706623,
      "grad_norm": 0.47643256187438965,
      "learning_rate": 6.534121062706541e-06,
      "loss": 0.3443,
      "step": 26981
    },
    {
      "epoch": 1.2382176127759166,
      "grad_norm": 0.49365726113319397,
      "learning_rate": 6.533887699299072e-06,
      "loss": 0.3948,
      "step": 26982
    },
    {
      "epoch": 1.238263503281171,
      "grad_norm": 0.42879343032836914,
      "learning_rate": 6.533654332203088e-06,
      "loss": 0.2832,
      "step": 26983
    },
    {
      "epoch": 1.2383093937864256,
      "grad_norm": 0.44917890429496765,
      "learning_rate": 6.533420961419147e-06,
      "loss": 0.3251,
      "step": 26984
    },
    {
      "epoch": 1.23835528429168,
      "grad_norm": 0.49695175886154175,
      "learning_rate": 6.533187586947814e-06,
      "loss": 0.4568,
      "step": 26985
    },
    {
      "epoch": 1.2384011747969346,
      "grad_norm": 0.4405040144920349,
      "learning_rate": 6.532954208789648e-06,
      "loss": 0.3266,
      "step": 26986
    },
    {
      "epoch": 1.238447065302189,
      "grad_norm": 0.5252926349639893,
      "learning_rate": 6.532720826945212e-06,
      "loss": 0.47,
      "step": 26987
    },
    {
      "epoch": 1.2384929558074433,
      "grad_norm": 0.4611525535583496,
      "learning_rate": 6.532487441415065e-06,
      "loss": 0.3311,
      "step": 26988
    },
    {
      "epoch": 1.2385388463126978,
      "grad_norm": 0.47140541672706604,
      "learning_rate": 6.532254052199771e-06,
      "loss": 0.374,
      "step": 26989
    },
    {
      "epoch": 1.2385847368179523,
      "grad_norm": 0.45425155758857727,
      "learning_rate": 6.532020659299888e-06,
      "loss": 0.2876,
      "step": 26990
    },
    {
      "epoch": 1.2386306273232068,
      "grad_norm": 0.5095045566558838,
      "learning_rate": 6.53178726271598e-06,
      "loss": 0.4331,
      "step": 26991
    },
    {
      "epoch": 1.2386765178284613,
      "grad_norm": 0.46734675765037537,
      "learning_rate": 6.531553862448606e-06,
      "loss": 0.3577,
      "step": 26992
    },
    {
      "epoch": 1.2387224083337158,
      "grad_norm": 0.46144795417785645,
      "learning_rate": 6.531320458498329e-06,
      "loss": 0.345,
      "step": 26993
    },
    {
      "epoch": 1.2387682988389703,
      "grad_norm": 0.4448409974575043,
      "learning_rate": 6.5310870508657095e-06,
      "loss": 0.3145,
      "step": 26994
    },
    {
      "epoch": 1.2388141893442246,
      "grad_norm": 0.46644696593284607,
      "learning_rate": 6.530853639551309e-06,
      "loss": 0.3414,
      "step": 26995
    },
    {
      "epoch": 1.238860079849479,
      "grad_norm": 0.4957960844039917,
      "learning_rate": 6.530620224555689e-06,
      "loss": 0.382,
      "step": 26996
    },
    {
      "epoch": 1.2389059703547336,
      "grad_norm": 0.4218040108680725,
      "learning_rate": 6.53038680587941e-06,
      "loss": 0.3138,
      "step": 26997
    },
    {
      "epoch": 1.238951860859988,
      "grad_norm": 0.4730508327484131,
      "learning_rate": 6.530153383523032e-06,
      "loss": 0.3536,
      "step": 26998
    },
    {
      "epoch": 1.2389977513652426,
      "grad_norm": 0.49037477374076843,
      "learning_rate": 6.52991995748712e-06,
      "loss": 0.3787,
      "step": 26999
    },
    {
      "epoch": 1.2390436418704969,
      "grad_norm": 0.470256507396698,
      "learning_rate": 6.529686527772234e-06,
      "loss": 0.3712,
      "step": 27000
    },
    {
      "epoch": 1.2390895323757514,
      "grad_norm": 0.46006613969802856,
      "learning_rate": 6.529453094378934e-06,
      "loss": 0.3473,
      "step": 27001
    },
    {
      "epoch": 1.2391354228810059,
      "grad_norm": 0.46078142523765564,
      "learning_rate": 6.52921965730778e-06,
      "loss": 0.3201,
      "step": 27002
    },
    {
      "epoch": 1.2391813133862604,
      "grad_norm": 0.45331457257270813,
      "learning_rate": 6.528986216559339e-06,
      "loss": 0.3173,
      "step": 27003
    },
    {
      "epoch": 1.2392272038915149,
      "grad_norm": 0.5347026586532593,
      "learning_rate": 6.528752772134166e-06,
      "loss": 0.3577,
      "step": 27004
    },
    {
      "epoch": 1.2392730943967694,
      "grad_norm": 0.4585721492767334,
      "learning_rate": 6.528519324032827e-06,
      "loss": 0.3283,
      "step": 27005
    },
    {
      "epoch": 1.2393189849020239,
      "grad_norm": 0.4250403046607971,
      "learning_rate": 6.5282858722558796e-06,
      "loss": 0.2673,
      "step": 27006
    },
    {
      "epoch": 1.2393648754072784,
      "grad_norm": 0.5245912671089172,
      "learning_rate": 6.528052416803887e-06,
      "loss": 0.4066,
      "step": 27007
    },
    {
      "epoch": 1.2394107659125326,
      "grad_norm": 0.4414653778076172,
      "learning_rate": 6.527818957677412e-06,
      "loss": 0.3352,
      "step": 27008
    },
    {
      "epoch": 1.2394566564177871,
      "grad_norm": 0.5111692547798157,
      "learning_rate": 6.527585494877014e-06,
      "loss": 0.418,
      "step": 27009
    },
    {
      "epoch": 1.2395025469230416,
      "grad_norm": 0.47435471415519714,
      "learning_rate": 6.527352028403255e-06,
      "loss": 0.3396,
      "step": 27010
    },
    {
      "epoch": 1.2395484374282961,
      "grad_norm": 0.47256824374198914,
      "learning_rate": 6.527118558256696e-06,
      "loss": 0.3652,
      "step": 27011
    },
    {
      "epoch": 1.2395943279335506,
      "grad_norm": 0.5033325552940369,
      "learning_rate": 6.5268850844379e-06,
      "loss": 0.367,
      "step": 27012
    },
    {
      "epoch": 1.239640218438805,
      "grad_norm": 0.45205235481262207,
      "learning_rate": 6.526651606947425e-06,
      "loss": 0.364,
      "step": 27013
    },
    {
      "epoch": 1.2396861089440594,
      "grad_norm": 0.47104787826538086,
      "learning_rate": 6.526418125785837e-06,
      "loss": 0.3452,
      "step": 27014
    },
    {
      "epoch": 1.239731999449314,
      "grad_norm": 0.5172480344772339,
      "learning_rate": 6.526184640953694e-06,
      "loss": 0.3746,
      "step": 27015
    },
    {
      "epoch": 1.2397778899545684,
      "grad_norm": 0.4598272144794464,
      "learning_rate": 6.52595115245156e-06,
      "loss": 0.3468,
      "step": 27016
    },
    {
      "epoch": 1.2398237804598229,
      "grad_norm": 0.48532307147979736,
      "learning_rate": 6.525717660279993e-06,
      "loss": 0.4508,
      "step": 27017
    },
    {
      "epoch": 1.2398696709650774,
      "grad_norm": 0.5078039765357971,
      "learning_rate": 6.525484164439559e-06,
      "loss": 0.4335,
      "step": 27018
    },
    {
      "epoch": 1.2399155614703319,
      "grad_norm": 0.48845285177230835,
      "learning_rate": 6.525250664930814e-06,
      "loss": 0.3794,
      "step": 27019
    },
    {
      "epoch": 1.2399614519755862,
      "grad_norm": 0.47792676091194153,
      "learning_rate": 6.5250171617543244e-06,
      "loss": 0.3839,
      "step": 27020
    },
    {
      "epoch": 1.2400073424808407,
      "grad_norm": 0.4934951364994049,
      "learning_rate": 6.52478365491065e-06,
      "loss": 0.4532,
      "step": 27021
    },
    {
      "epoch": 1.2400532329860952,
      "grad_norm": 0.45053303241729736,
      "learning_rate": 6.524550144400352e-06,
      "loss": 0.3412,
      "step": 27022
    },
    {
      "epoch": 1.2400991234913497,
      "grad_norm": 0.495858371257782,
      "learning_rate": 6.524316630223993e-06,
      "loss": 0.4128,
      "step": 27023
    },
    {
      "epoch": 1.2401450139966042,
      "grad_norm": 0.4718649387359619,
      "learning_rate": 6.524083112382131e-06,
      "loss": 0.3944,
      "step": 27024
    },
    {
      "epoch": 1.2401909045018586,
      "grad_norm": 0.4177536368370056,
      "learning_rate": 6.523849590875332e-06,
      "loss": 0.2724,
      "step": 27025
    },
    {
      "epoch": 1.240236795007113,
      "grad_norm": 0.48354819416999817,
      "learning_rate": 6.523616065704156e-06,
      "loss": 0.3641,
      "step": 27026
    },
    {
      "epoch": 1.2402826855123674,
      "grad_norm": 0.4603208005428314,
      "learning_rate": 6.523382536869163e-06,
      "loss": 0.3502,
      "step": 27027
    },
    {
      "epoch": 1.240328576017622,
      "grad_norm": 0.47348088026046753,
      "learning_rate": 6.523149004370916e-06,
      "loss": 0.3881,
      "step": 27028
    },
    {
      "epoch": 1.2403744665228764,
      "grad_norm": 0.4721221923828125,
      "learning_rate": 6.522915468209976e-06,
      "loss": 0.3658,
      "step": 27029
    },
    {
      "epoch": 1.240420357028131,
      "grad_norm": 0.49434778094291687,
      "learning_rate": 6.522681928386905e-06,
      "loss": 0.3965,
      "step": 27030
    },
    {
      "epoch": 1.2404662475333854,
      "grad_norm": 0.4491208791732788,
      "learning_rate": 6.522448384902266e-06,
      "loss": 0.32,
      "step": 27031
    },
    {
      "epoch": 1.24051213803864,
      "grad_norm": 1.0284695625305176,
      "learning_rate": 6.5222148377566185e-06,
      "loss": 0.3332,
      "step": 27032
    },
    {
      "epoch": 1.2405580285438942,
      "grad_norm": 0.4934678077697754,
      "learning_rate": 6.521981286950524e-06,
      "loss": 0.3832,
      "step": 27033
    },
    {
      "epoch": 1.2406039190491487,
      "grad_norm": 0.4557664096355438,
      "learning_rate": 6.521747732484546e-06,
      "loss": 0.2853,
      "step": 27034
    },
    {
      "epoch": 1.2406498095544032,
      "grad_norm": 0.4769366383552551,
      "learning_rate": 6.521514174359244e-06,
      "loss": 0.3585,
      "step": 27035
    },
    {
      "epoch": 1.2406957000596577,
      "grad_norm": 0.5254905819892883,
      "learning_rate": 6.52128061257518e-06,
      "loss": 0.329,
      "step": 27036
    },
    {
      "epoch": 1.2407415905649122,
      "grad_norm": 0.4639548659324646,
      "learning_rate": 6.521047047132919e-06,
      "loss": 0.3544,
      "step": 27037
    },
    {
      "epoch": 1.2407874810701667,
      "grad_norm": 0.452947735786438,
      "learning_rate": 6.520813478033017e-06,
      "loss": 0.36,
      "step": 27038
    },
    {
      "epoch": 1.240833371575421,
      "grad_norm": 0.45113468170166016,
      "learning_rate": 6.520579905276041e-06,
      "loss": 0.34,
      "step": 27039
    },
    {
      "epoch": 1.2408792620806754,
      "grad_norm": 0.48438915610313416,
      "learning_rate": 6.520346328862548e-06,
      "loss": 0.3656,
      "step": 27040
    },
    {
      "epoch": 1.24092515258593,
      "grad_norm": 0.4963419437408447,
      "learning_rate": 6.520112748793101e-06,
      "loss": 0.3681,
      "step": 27041
    },
    {
      "epoch": 1.2409710430911844,
      "grad_norm": 0.49887871742248535,
      "learning_rate": 6.5198791650682655e-06,
      "loss": 0.3859,
      "step": 27042
    },
    {
      "epoch": 1.241016933596439,
      "grad_norm": 0.4632721543312073,
      "learning_rate": 6.5196455776886e-06,
      "loss": 0.3682,
      "step": 27043
    },
    {
      "epoch": 1.2410628241016934,
      "grad_norm": 0.4863838851451874,
      "learning_rate": 6.519411986654666e-06,
      "loss": 0.3986,
      "step": 27044
    },
    {
      "epoch": 1.241108714606948,
      "grad_norm": 0.48101967573165894,
      "learning_rate": 6.519178391967025e-06,
      "loss": 0.3614,
      "step": 27045
    },
    {
      "epoch": 1.2411546051122022,
      "grad_norm": 0.4778022766113281,
      "learning_rate": 6.51894479362624e-06,
      "loss": 0.3631,
      "step": 27046
    },
    {
      "epoch": 1.2412004956174567,
      "grad_norm": 0.476870596408844,
      "learning_rate": 6.518711191632871e-06,
      "loss": 0.3834,
      "step": 27047
    },
    {
      "epoch": 1.2412463861227112,
      "grad_norm": 0.4745154082775116,
      "learning_rate": 6.5184775859874835e-06,
      "loss": 0.2806,
      "step": 27048
    },
    {
      "epoch": 1.2412922766279657,
      "grad_norm": 0.5122151970863342,
      "learning_rate": 6.518243976690635e-06,
      "loss": 0.4812,
      "step": 27049
    },
    {
      "epoch": 1.2413381671332202,
      "grad_norm": 0.48808810114860535,
      "learning_rate": 6.518010363742888e-06,
      "loss": 0.3531,
      "step": 27050
    },
    {
      "epoch": 1.2413840576384745,
      "grad_norm": 0.4663884937763214,
      "learning_rate": 6.517776747144806e-06,
      "loss": 0.4014,
      "step": 27051
    },
    {
      "epoch": 1.241429948143729,
      "grad_norm": 0.453797847032547,
      "learning_rate": 6.51754312689695e-06,
      "loss": 0.3648,
      "step": 27052
    },
    {
      "epoch": 1.2414758386489835,
      "grad_norm": 0.44562968611717224,
      "learning_rate": 6.517309502999882e-06,
      "loss": 0.3443,
      "step": 27053
    },
    {
      "epoch": 1.241521729154238,
      "grad_norm": 0.4483977258205414,
      "learning_rate": 6.517075875454164e-06,
      "loss": 0.339,
      "step": 27054
    },
    {
      "epoch": 1.2415676196594925,
      "grad_norm": 0.5322286486625671,
      "learning_rate": 6.516842244260356e-06,
      "loss": 0.4193,
      "step": 27055
    },
    {
      "epoch": 1.241613510164747,
      "grad_norm": 0.4517311155796051,
      "learning_rate": 6.516608609419023e-06,
      "loss": 0.3207,
      "step": 27056
    },
    {
      "epoch": 1.2416594006700015,
      "grad_norm": 0.4524323642253876,
      "learning_rate": 6.516374970930724e-06,
      "loss": 0.3182,
      "step": 27057
    },
    {
      "epoch": 1.2417052911752557,
      "grad_norm": 0.44993674755096436,
      "learning_rate": 6.51614132879602e-06,
      "loss": 0.3157,
      "step": 27058
    },
    {
      "epoch": 1.2417511816805102,
      "grad_norm": 0.45945224165916443,
      "learning_rate": 6.515907683015478e-06,
      "loss": 0.3139,
      "step": 27059
    },
    {
      "epoch": 1.2417970721857647,
      "grad_norm": 0.49321249127388,
      "learning_rate": 6.515674033589654e-06,
      "loss": 0.3891,
      "step": 27060
    },
    {
      "epoch": 1.2418429626910192,
      "grad_norm": 0.5147197842597961,
      "learning_rate": 6.515440380519112e-06,
      "loss": 0.3847,
      "step": 27061
    },
    {
      "epoch": 1.2418888531962737,
      "grad_norm": 0.6308175921440125,
      "learning_rate": 6.515206723804415e-06,
      "loss": 0.2835,
      "step": 27062
    },
    {
      "epoch": 1.2419347437015282,
      "grad_norm": 0.5028957724571228,
      "learning_rate": 6.514973063446124e-06,
      "loss": 0.4323,
      "step": 27063
    },
    {
      "epoch": 1.2419806342067825,
      "grad_norm": 0.46063849329948425,
      "learning_rate": 6.5147393994448015e-06,
      "loss": 0.3438,
      "step": 27064
    },
    {
      "epoch": 1.242026524712037,
      "grad_norm": 0.4767475724220276,
      "learning_rate": 6.514505731801008e-06,
      "loss": 0.3757,
      "step": 27065
    },
    {
      "epoch": 1.2420724152172915,
      "grad_norm": 0.5024803280830383,
      "learning_rate": 6.514272060515306e-06,
      "loss": 0.4082,
      "step": 27066
    },
    {
      "epoch": 1.242118305722546,
      "grad_norm": 0.4939013421535492,
      "learning_rate": 6.514038385588258e-06,
      "loss": 0.4305,
      "step": 27067
    },
    {
      "epoch": 1.2421641962278005,
      "grad_norm": 0.48958125710487366,
      "learning_rate": 6.513804707020427e-06,
      "loss": 0.4225,
      "step": 27068
    },
    {
      "epoch": 1.242210086733055,
      "grad_norm": 0.4751177728176117,
      "learning_rate": 6.51357102481237e-06,
      "loss": 0.3562,
      "step": 27069
    },
    {
      "epoch": 1.2422559772383095,
      "grad_norm": 0.4508625268936157,
      "learning_rate": 6.513337338964657e-06,
      "loss": 0.279,
      "step": 27070
    },
    {
      "epoch": 1.2423018677435638,
      "grad_norm": 0.4461248517036438,
      "learning_rate": 6.513103649477843e-06,
      "loss": 0.3398,
      "step": 27071
    },
    {
      "epoch": 1.2423477582488183,
      "grad_norm": 0.5069220662117004,
      "learning_rate": 6.512869956352491e-06,
      "loss": 0.3689,
      "step": 27072
    },
    {
      "epoch": 1.2423936487540728,
      "grad_norm": 0.4731628894805908,
      "learning_rate": 6.512636259589166e-06,
      "loss": 0.427,
      "step": 27073
    },
    {
      "epoch": 1.2424395392593273,
      "grad_norm": 0.45678165555000305,
      "learning_rate": 6.512402559188427e-06,
      "loss": 0.3305,
      "step": 27074
    },
    {
      "epoch": 1.2424854297645818,
      "grad_norm": 0.4793204665184021,
      "learning_rate": 6.512168855150837e-06,
      "loss": 0.3611,
      "step": 27075
    },
    {
      "epoch": 1.2425313202698363,
      "grad_norm": 0.45435789227485657,
      "learning_rate": 6.511935147476961e-06,
      "loss": 0.3789,
      "step": 27076
    },
    {
      "epoch": 1.2425772107750905,
      "grad_norm": 0.48214176297187805,
      "learning_rate": 6.511701436167355e-06,
      "loss": 0.4038,
      "step": 27077
    },
    {
      "epoch": 1.242623101280345,
      "grad_norm": 0.454705148935318,
      "learning_rate": 6.511467721222584e-06,
      "loss": 0.3433,
      "step": 27078
    },
    {
      "epoch": 1.2426689917855995,
      "grad_norm": 0.45880579948425293,
      "learning_rate": 6.511234002643212e-06,
      "loss": 0.3627,
      "step": 27079
    },
    {
      "epoch": 1.242714882290854,
      "grad_norm": 0.4805181324481964,
      "learning_rate": 6.511000280429797e-06,
      "loss": 0.374,
      "step": 27080
    },
    {
      "epoch": 1.2427607727961085,
      "grad_norm": 0.5005503296852112,
      "learning_rate": 6.510766554582907e-06,
      "loss": 0.395,
      "step": 27081
    },
    {
      "epoch": 1.242806663301363,
      "grad_norm": 0.45770904421806335,
      "learning_rate": 6.510532825103099e-06,
      "loss": 0.3561,
      "step": 27082
    },
    {
      "epoch": 1.2428525538066175,
      "grad_norm": 0.4728145897388458,
      "learning_rate": 6.510299091990934e-06,
      "loss": 0.3565,
      "step": 27083
    },
    {
      "epoch": 1.2428984443118718,
      "grad_norm": 0.5317184329032898,
      "learning_rate": 6.510065355246977e-06,
      "loss": 0.4554,
      "step": 27084
    },
    {
      "epoch": 1.2429443348171263,
      "grad_norm": 0.47025206685066223,
      "learning_rate": 6.509831614871791e-06,
      "loss": 0.3775,
      "step": 27085
    },
    {
      "epoch": 1.2429902253223808,
      "grad_norm": 0.4879731237888336,
      "learning_rate": 6.509597870865936e-06,
      "loss": 0.4014,
      "step": 27086
    },
    {
      "epoch": 1.2430361158276353,
      "grad_norm": 0.4621317982673645,
      "learning_rate": 6.509364123229974e-06,
      "loss": 0.3109,
      "step": 27087
    },
    {
      "epoch": 1.2430820063328898,
      "grad_norm": 0.43527066707611084,
      "learning_rate": 6.5091303719644675e-06,
      "loss": 0.3271,
      "step": 27088
    },
    {
      "epoch": 1.243127896838144,
      "grad_norm": 0.49288082122802734,
      "learning_rate": 6.508896617069979e-06,
      "loss": 0.3941,
      "step": 27089
    },
    {
      "epoch": 1.2431737873433986,
      "grad_norm": 0.4846830368041992,
      "learning_rate": 6.50866285854707e-06,
      "loss": 0.3568,
      "step": 27090
    },
    {
      "epoch": 1.243219677848653,
      "grad_norm": 0.46757686138153076,
      "learning_rate": 6.508429096396303e-06,
      "loss": 0.3623,
      "step": 27091
    },
    {
      "epoch": 1.2432655683539076,
      "grad_norm": 0.49050650000572205,
      "learning_rate": 6.508195330618241e-06,
      "loss": 0.3479,
      "step": 27092
    },
    {
      "epoch": 1.243311458859162,
      "grad_norm": 0.5163153409957886,
      "learning_rate": 6.507961561213445e-06,
      "loss": 0.3729,
      "step": 27093
    },
    {
      "epoch": 1.2433573493644166,
      "grad_norm": 0.44548898935317993,
      "learning_rate": 6.507727788182476e-06,
      "loss": 0.3323,
      "step": 27094
    },
    {
      "epoch": 1.243403239869671,
      "grad_norm": 0.4674201011657715,
      "learning_rate": 6.5074940115258985e-06,
      "loss": 0.3293,
      "step": 27095
    },
    {
      "epoch": 1.2434491303749255,
      "grad_norm": 0.4328274130821228,
      "learning_rate": 6.507260231244275e-06,
      "loss": 0.2988,
      "step": 27096
    },
    {
      "epoch": 1.2434950208801798,
      "grad_norm": 0.4327007234096527,
      "learning_rate": 6.507026447338165e-06,
      "loss": 0.3085,
      "step": 27097
    },
    {
      "epoch": 1.2435409113854343,
      "grad_norm": 0.504535436630249,
      "learning_rate": 6.506792659808131e-06,
      "loss": 0.4225,
      "step": 27098
    },
    {
      "epoch": 1.2435868018906888,
      "grad_norm": 0.4561856985092163,
      "learning_rate": 6.506558868654738e-06,
      "loss": 0.3527,
      "step": 27099
    },
    {
      "epoch": 1.2436326923959433,
      "grad_norm": 0.4526936113834381,
      "learning_rate": 6.506325073878546e-06,
      "loss": 0.3482,
      "step": 27100
    },
    {
      "epoch": 1.2436785829011978,
      "grad_norm": 0.4887792468070984,
      "learning_rate": 6.506091275480115e-06,
      "loss": 0.3903,
      "step": 27101
    },
    {
      "epoch": 1.243724473406452,
      "grad_norm": 0.5030981302261353,
      "learning_rate": 6.5058574734600135e-06,
      "loss": 0.4261,
      "step": 27102
    },
    {
      "epoch": 1.2437703639117066,
      "grad_norm": 0.45109376311302185,
      "learning_rate": 6.505623667818799e-06,
      "loss": 0.362,
      "step": 27103
    },
    {
      "epoch": 1.243816254416961,
      "grad_norm": 0.4737619161605835,
      "learning_rate": 6.505389858557035e-06,
      "loss": 0.3107,
      "step": 27104
    },
    {
      "epoch": 1.2438621449222156,
      "grad_norm": 0.47073888778686523,
      "learning_rate": 6.505156045675282e-06,
      "loss": 0.3485,
      "step": 27105
    },
    {
      "epoch": 1.24390803542747,
      "grad_norm": 0.44162726402282715,
      "learning_rate": 6.5049222291741045e-06,
      "loss": 0.3237,
      "step": 27106
    },
    {
      "epoch": 1.2439539259327246,
      "grad_norm": 0.4589567482471466,
      "learning_rate": 6.504688409054065e-06,
      "loss": 0.3339,
      "step": 27107
    },
    {
      "epoch": 1.243999816437979,
      "grad_norm": 0.5723713636398315,
      "learning_rate": 6.504454585315725e-06,
      "loss": 0.3789,
      "step": 27108
    },
    {
      "epoch": 1.2440457069432334,
      "grad_norm": 0.48852717876434326,
      "learning_rate": 6.504220757959645e-06,
      "loss": 0.3488,
      "step": 27109
    },
    {
      "epoch": 1.2440915974484879,
      "grad_norm": 0.44218942523002625,
      "learning_rate": 6.5039869269863895e-06,
      "loss": 0.3132,
      "step": 27110
    },
    {
      "epoch": 1.2441374879537423,
      "grad_norm": 0.44037508964538574,
      "learning_rate": 6.503753092396519e-06,
      "loss": 0.2886,
      "step": 27111
    },
    {
      "epoch": 1.2441833784589968,
      "grad_norm": 0.5884010195732117,
      "learning_rate": 6.503519254190598e-06,
      "loss": 0.3991,
      "step": 27112
    },
    {
      "epoch": 1.2442292689642513,
      "grad_norm": 0.4810219407081604,
      "learning_rate": 6.503285412369187e-06,
      "loss": 0.3945,
      "step": 27113
    },
    {
      "epoch": 1.2442751594695058,
      "grad_norm": 0.45130655169487,
      "learning_rate": 6.503051566932851e-06,
      "loss": 0.3166,
      "step": 27114
    },
    {
      "epoch": 1.2443210499747601,
      "grad_norm": 0.5156421065330505,
      "learning_rate": 6.502817717882148e-06,
      "loss": 0.4503,
      "step": 27115
    },
    {
      "epoch": 1.2443669404800146,
      "grad_norm": 0.48779308795928955,
      "learning_rate": 6.502583865217644e-06,
      "loss": 0.415,
      "step": 27116
    },
    {
      "epoch": 1.2444128309852691,
      "grad_norm": 0.47189587354660034,
      "learning_rate": 6.5023500089399e-06,
      "loss": 0.3494,
      "step": 27117
    },
    {
      "epoch": 1.2444587214905236,
      "grad_norm": 0.5093399882316589,
      "learning_rate": 6.502116149049478e-06,
      "loss": 0.319,
      "step": 27118
    },
    {
      "epoch": 1.244504611995778,
      "grad_norm": 0.47336456179618835,
      "learning_rate": 6.501882285546941e-06,
      "loss": 0.3365,
      "step": 27119
    },
    {
      "epoch": 1.2445505025010326,
      "grad_norm": 0.4626924395561218,
      "learning_rate": 6.501648418432851e-06,
      "loss": 0.3972,
      "step": 27120
    },
    {
      "epoch": 1.244596393006287,
      "grad_norm": 0.5014014840126038,
      "learning_rate": 6.50141454770777e-06,
      "loss": 0.3588,
      "step": 27121
    },
    {
      "epoch": 1.2446422835115414,
      "grad_norm": 0.4969446361064911,
      "learning_rate": 6.501180673372262e-06,
      "loss": 0.3625,
      "step": 27122
    },
    {
      "epoch": 1.2446881740167959,
      "grad_norm": 0.4638162851333618,
      "learning_rate": 6.5009467954268855e-06,
      "loss": 0.3486,
      "step": 27123
    },
    {
      "epoch": 1.2447340645220504,
      "grad_norm": 0.5057275295257568,
      "learning_rate": 6.500712913872209e-06,
      "loss": 0.4504,
      "step": 27124
    },
    {
      "epoch": 1.2447799550273049,
      "grad_norm": 0.4455331861972809,
      "learning_rate": 6.500479028708791e-06,
      "loss": 0.3306,
      "step": 27125
    },
    {
      "epoch": 1.2448258455325594,
      "grad_norm": 0.4613574147224426,
      "learning_rate": 6.500245139937193e-06,
      "loss": 0.3458,
      "step": 27126
    },
    {
      "epoch": 1.2448717360378139,
      "grad_norm": 0.5263022780418396,
      "learning_rate": 6.50001124755798e-06,
      "loss": 0.409,
      "step": 27127
    },
    {
      "epoch": 1.2449176265430681,
      "grad_norm": 0.4352143108844757,
      "learning_rate": 6.499777351571713e-06,
      "loss": 0.3336,
      "step": 27128
    },
    {
      "epoch": 1.2449635170483226,
      "grad_norm": 0.476104736328125,
      "learning_rate": 6.499543451978956e-06,
      "loss": 0.357,
      "step": 27129
    },
    {
      "epoch": 1.2450094075535771,
      "grad_norm": 0.44013720750808716,
      "learning_rate": 6.499309548780269e-06,
      "loss": 0.312,
      "step": 27130
    },
    {
      "epoch": 1.2450552980588316,
      "grad_norm": 0.492827832698822,
      "learning_rate": 6.499075641976216e-06,
      "loss": 0.375,
      "step": 27131
    },
    {
      "epoch": 1.2451011885640861,
      "grad_norm": 0.4549662172794342,
      "learning_rate": 6.498841731567358e-06,
      "loss": 0.3835,
      "step": 27132
    },
    {
      "epoch": 1.2451470790693406,
      "grad_norm": 0.45174142718315125,
      "learning_rate": 6.49860781755426e-06,
      "loss": 0.3238,
      "step": 27133
    },
    {
      "epoch": 1.2451929695745951,
      "grad_norm": 0.46549710631370544,
      "learning_rate": 6.498373899937482e-06,
      "loss": 0.3863,
      "step": 27134
    },
    {
      "epoch": 1.2452388600798494,
      "grad_norm": 0.4798084795475006,
      "learning_rate": 6.498139978717591e-06,
      "loss": 0.3684,
      "step": 27135
    },
    {
      "epoch": 1.245284750585104,
      "grad_norm": 0.43202319741249084,
      "learning_rate": 6.4979060538951435e-06,
      "loss": 0.3268,
      "step": 27136
    },
    {
      "epoch": 1.2453306410903584,
      "grad_norm": 0.48386749625205994,
      "learning_rate": 6.497672125470704e-06,
      "loss": 0.3453,
      "step": 27137
    },
    {
      "epoch": 1.245376531595613,
      "grad_norm": 0.4435933232307434,
      "learning_rate": 6.497438193444837e-06,
      "loss": 0.2943,
      "step": 27138
    },
    {
      "epoch": 1.2454224221008674,
      "grad_norm": 0.4361483156681061,
      "learning_rate": 6.497204257818104e-06,
      "loss": 0.2846,
      "step": 27139
    },
    {
      "epoch": 1.2454683126061217,
      "grad_norm": 0.4878710210323334,
      "learning_rate": 6.496970318591067e-06,
      "loss": 0.3316,
      "step": 27140
    },
    {
      "epoch": 1.2455142031113762,
      "grad_norm": 0.45536360144615173,
      "learning_rate": 6.496736375764288e-06,
      "loss": 0.3738,
      "step": 27141
    },
    {
      "epoch": 1.2455600936166307,
      "grad_norm": 0.41184940934181213,
      "learning_rate": 6.496502429338332e-06,
      "loss": 0.2516,
      "step": 27142
    },
    {
      "epoch": 1.2456059841218852,
      "grad_norm": 0.48244231939315796,
      "learning_rate": 6.496268479313758e-06,
      "loss": 0.3632,
      "step": 27143
    },
    {
      "epoch": 1.2456518746271397,
      "grad_norm": 0.5049140453338623,
      "learning_rate": 6.496034525691131e-06,
      "loss": 0.4627,
      "step": 27144
    },
    {
      "epoch": 1.2456977651323942,
      "grad_norm": 0.5351001024246216,
      "learning_rate": 6.495800568471013e-06,
      "loss": 0.4518,
      "step": 27145
    },
    {
      "epoch": 1.2457436556376487,
      "grad_norm": 0.4875046908855438,
      "learning_rate": 6.495566607653969e-06,
      "loss": 0.3724,
      "step": 27146
    },
    {
      "epoch": 1.245789546142903,
      "grad_norm": 0.473929762840271,
      "learning_rate": 6.495332643240558e-06,
      "loss": 0.3876,
      "step": 27147
    },
    {
      "epoch": 1.2458354366481574,
      "grad_norm": 0.46286073327064514,
      "learning_rate": 6.495098675231343e-06,
      "loss": 0.3544,
      "step": 27148
    },
    {
      "epoch": 1.245881327153412,
      "grad_norm": 0.4387279152870178,
      "learning_rate": 6.494864703626887e-06,
      "loss": 0.2931,
      "step": 27149
    },
    {
      "epoch": 1.2459272176586664,
      "grad_norm": 0.4251987636089325,
      "learning_rate": 6.494630728427755e-06,
      "loss": 0.2825,
      "step": 27150
    },
    {
      "epoch": 1.245973108163921,
      "grad_norm": 0.48286065459251404,
      "learning_rate": 6.494396749634508e-06,
      "loss": 0.4215,
      "step": 27151
    },
    {
      "epoch": 1.2460189986691754,
      "grad_norm": 0.4984102249145508,
      "learning_rate": 6.494162767247708e-06,
      "loss": 0.4167,
      "step": 27152
    },
    {
      "epoch": 1.2460648891744297,
      "grad_norm": 0.4980386197566986,
      "learning_rate": 6.493928781267917e-06,
      "loss": 0.4143,
      "step": 27153
    },
    {
      "epoch": 1.2461107796796842,
      "grad_norm": 0.46676304936408997,
      "learning_rate": 6.493694791695699e-06,
      "loss": 0.3292,
      "step": 27154
    },
    {
      "epoch": 1.2461566701849387,
      "grad_norm": 0.45811837911605835,
      "learning_rate": 6.493460798531617e-06,
      "loss": 0.3526,
      "step": 27155
    },
    {
      "epoch": 1.2462025606901932,
      "grad_norm": 0.46026745438575745,
      "learning_rate": 6.493226801776234e-06,
      "loss": 0.3132,
      "step": 27156
    },
    {
      "epoch": 1.2462484511954477,
      "grad_norm": 0.4545271694660187,
      "learning_rate": 6.4929928014301114e-06,
      "loss": 0.327,
      "step": 27157
    },
    {
      "epoch": 1.2462943417007022,
      "grad_norm": 0.4925821125507355,
      "learning_rate": 6.492758797493812e-06,
      "loss": 0.401,
      "step": 27158
    },
    {
      "epoch": 1.2463402322059567,
      "grad_norm": 0.48520565032958984,
      "learning_rate": 6.492524789967898e-06,
      "loss": 0.406,
      "step": 27159
    },
    {
      "epoch": 1.246386122711211,
      "grad_norm": 0.45077237486839294,
      "learning_rate": 6.492290778852933e-06,
      "loss": 0.3173,
      "step": 27160
    },
    {
      "epoch": 1.2464320132164655,
      "grad_norm": 0.43259233236312866,
      "learning_rate": 6.492056764149481e-06,
      "loss": 0.2771,
      "step": 27161
    },
    {
      "epoch": 1.24647790372172,
      "grad_norm": 0.4657677412033081,
      "learning_rate": 6.491822745858103e-06,
      "loss": 0.3328,
      "step": 27162
    },
    {
      "epoch": 1.2465237942269745,
      "grad_norm": 0.45552247762680054,
      "learning_rate": 6.49158872397936e-06,
      "loss": 0.3116,
      "step": 27163
    },
    {
      "epoch": 1.246569684732229,
      "grad_norm": 0.49901121854782104,
      "learning_rate": 6.491354698513821e-06,
      "loss": 0.4331,
      "step": 27164
    },
    {
      "epoch": 1.2466155752374835,
      "grad_norm": 0.45761552453041077,
      "learning_rate": 6.491120669462041e-06,
      "loss": 0.3783,
      "step": 27165
    },
    {
      "epoch": 1.2466614657427377,
      "grad_norm": 0.5073399543762207,
      "learning_rate": 6.490886636824587e-06,
      "loss": 0.3487,
      "step": 27166
    },
    {
      "epoch": 1.2467073562479922,
      "grad_norm": 0.45310312509536743,
      "learning_rate": 6.490652600602023e-06,
      "loss": 0.3355,
      "step": 27167
    },
    {
      "epoch": 1.2467532467532467,
      "grad_norm": 0.4901655316352844,
      "learning_rate": 6.490418560794909e-06,
      "loss": 0.4591,
      "step": 27168
    },
    {
      "epoch": 1.2467991372585012,
      "grad_norm": 0.44069012999534607,
      "learning_rate": 6.490184517403809e-06,
      "loss": 0.3134,
      "step": 27169
    },
    {
      "epoch": 1.2468450277637557,
      "grad_norm": 0.44979336857795715,
      "learning_rate": 6.489950470429285e-06,
      "loss": 0.2566,
      "step": 27170
    },
    {
      "epoch": 1.2468909182690102,
      "grad_norm": 0.47115257382392883,
      "learning_rate": 6.4897164198718996e-06,
      "loss": 0.3662,
      "step": 27171
    },
    {
      "epoch": 1.2469368087742647,
      "grad_norm": 0.46368563175201416,
      "learning_rate": 6.489482365732218e-06,
      "loss": 0.3669,
      "step": 27172
    },
    {
      "epoch": 1.246982699279519,
      "grad_norm": 0.49674177169799805,
      "learning_rate": 6.4892483080108005e-06,
      "loss": 0.381,
      "step": 27173
    },
    {
      "epoch": 1.2470285897847735,
      "grad_norm": 0.46072840690612793,
      "learning_rate": 6.4890142467082104e-06,
      "loss": 0.383,
      "step": 27174
    },
    {
      "epoch": 1.247074480290028,
      "grad_norm": 0.47060084342956543,
      "learning_rate": 6.488780181825013e-06,
      "loss": 0.3264,
      "step": 27175
    },
    {
      "epoch": 1.2471203707952825,
      "grad_norm": 0.4531290531158447,
      "learning_rate": 6.488546113361767e-06,
      "loss": 0.3049,
      "step": 27176
    },
    {
      "epoch": 1.247166261300537,
      "grad_norm": 0.6784154176712036,
      "learning_rate": 6.488312041319037e-06,
      "loss": 0.3631,
      "step": 27177
    },
    {
      "epoch": 1.2472121518057913,
      "grad_norm": 0.43427348136901855,
      "learning_rate": 6.488077965697388e-06,
      "loss": 0.3193,
      "step": 27178
    },
    {
      "epoch": 1.2472580423110458,
      "grad_norm": 0.47857213020324707,
      "learning_rate": 6.487843886497381e-06,
      "loss": 0.3771,
      "step": 27179
    },
    {
      "epoch": 1.2473039328163003,
      "grad_norm": 0.4701474606990814,
      "learning_rate": 6.487609803719579e-06,
      "loss": 0.315,
      "step": 27180
    },
    {
      "epoch": 1.2473498233215548,
      "grad_norm": 0.47626417875289917,
      "learning_rate": 6.487375717364545e-06,
      "loss": 0.347,
      "step": 27181
    },
    {
      "epoch": 1.2473957138268092,
      "grad_norm": 0.519077479839325,
      "learning_rate": 6.48714162743284e-06,
      "loss": 0.4706,
      "step": 27182
    },
    {
      "epoch": 1.2474416043320637,
      "grad_norm": 0.5012746453285217,
      "learning_rate": 6.486907533925031e-06,
      "loss": 0.4132,
      "step": 27183
    },
    {
      "epoch": 1.2474874948373182,
      "grad_norm": 0.541907012462616,
      "learning_rate": 6.486673436841677e-06,
      "loss": 0.4564,
      "step": 27184
    },
    {
      "epoch": 1.2475333853425727,
      "grad_norm": 0.5083046555519104,
      "learning_rate": 6.4864393361833435e-06,
      "loss": 0.4422,
      "step": 27185
    },
    {
      "epoch": 1.247579275847827,
      "grad_norm": 0.4666883051395416,
      "learning_rate": 6.486205231950593e-06,
      "loss": 0.3647,
      "step": 27186
    },
    {
      "epoch": 1.2476251663530815,
      "grad_norm": 0.45006126165390015,
      "learning_rate": 6.485971124143987e-06,
      "loss": 0.3361,
      "step": 27187
    },
    {
      "epoch": 1.247671056858336,
      "grad_norm": 0.4726943075656891,
      "learning_rate": 6.4857370127640885e-06,
      "loss": 0.3291,
      "step": 27188
    },
    {
      "epoch": 1.2477169473635905,
      "grad_norm": 0.4755069613456726,
      "learning_rate": 6.485502897811464e-06,
      "loss": 0.3634,
      "step": 27189
    },
    {
      "epoch": 1.247762837868845,
      "grad_norm": 0.4654363691806793,
      "learning_rate": 6.485268779286673e-06,
      "loss": 0.4089,
      "step": 27190
    },
    {
      "epoch": 1.2478087283740993,
      "grad_norm": 0.47353288531303406,
      "learning_rate": 6.485034657190278e-06,
      "loss": 0.3729,
      "step": 27191
    },
    {
      "epoch": 1.2478546188793538,
      "grad_norm": 0.43786486983299255,
      "learning_rate": 6.484800531522844e-06,
      "loss": 0.2831,
      "step": 27192
    },
    {
      "epoch": 1.2479005093846083,
      "grad_norm": 0.49401047825813293,
      "learning_rate": 6.484566402284934e-06,
      "loss": 0.4025,
      "step": 27193
    },
    {
      "epoch": 1.2479463998898628,
      "grad_norm": 0.4683270752429962,
      "learning_rate": 6.484332269477111e-06,
      "loss": 0.3675,
      "step": 27194
    },
    {
      "epoch": 1.2479922903951173,
      "grad_norm": 0.43790438771247864,
      "learning_rate": 6.4840981330999365e-06,
      "loss": 0.3211,
      "step": 27195
    },
    {
      "epoch": 1.2480381809003718,
      "grad_norm": 0.4515499472618103,
      "learning_rate": 6.483863993153973e-06,
      "loss": 0.3473,
      "step": 27196
    },
    {
      "epoch": 1.2480840714056263,
      "grad_norm": 0.49788567423820496,
      "learning_rate": 6.483629849639785e-06,
      "loss": 0.4362,
      "step": 27197
    },
    {
      "epoch": 1.2481299619108805,
      "grad_norm": 0.4535635709762573,
      "learning_rate": 6.4833957025579376e-06,
      "loss": 0.3675,
      "step": 27198
    },
    {
      "epoch": 1.248175852416135,
      "grad_norm": 0.4510275423526764,
      "learning_rate": 6.48316155190899e-06,
      "loss": 0.3352,
      "step": 27199
    },
    {
      "epoch": 1.2482217429213895,
      "grad_norm": 0.43900376558303833,
      "learning_rate": 6.482927397693508e-06,
      "loss": 0.3238,
      "step": 27200
    },
    {
      "epoch": 1.248267633426644,
      "grad_norm": 0.5351582169532776,
      "learning_rate": 6.482693239912055e-06,
      "loss": 0.4256,
      "step": 27201
    },
    {
      "epoch": 1.2483135239318985,
      "grad_norm": 0.4647214710712433,
      "learning_rate": 6.482459078565189e-06,
      "loss": 0.3481,
      "step": 27202
    },
    {
      "epoch": 1.248359414437153,
      "grad_norm": 0.48952463269233704,
      "learning_rate": 6.48222491365348e-06,
      "loss": 0.4242,
      "step": 27203
    },
    {
      "epoch": 1.2484053049424073,
      "grad_norm": 0.44437965750694275,
      "learning_rate": 6.481990745177485e-06,
      "loss": 0.3148,
      "step": 27204
    },
    {
      "epoch": 1.2484511954476618,
      "grad_norm": 0.4677712917327881,
      "learning_rate": 6.481756573137773e-06,
      "loss": 0.3766,
      "step": 27205
    },
    {
      "epoch": 1.2484970859529163,
      "grad_norm": 0.4753119945526123,
      "learning_rate": 6.4815223975349025e-06,
      "loss": 0.3706,
      "step": 27206
    },
    {
      "epoch": 1.2485429764581708,
      "grad_norm": 0.4815599024295807,
      "learning_rate": 6.481288218369438e-06,
      "loss": 0.367,
      "step": 27207
    },
    {
      "epoch": 1.2485888669634253,
      "grad_norm": 0.5009384751319885,
      "learning_rate": 6.4810540356419424e-06,
      "loss": 0.3893,
      "step": 27208
    },
    {
      "epoch": 1.2486347574686798,
      "grad_norm": 0.4978771209716797,
      "learning_rate": 6.480819849352982e-06,
      "loss": 0.392,
      "step": 27209
    },
    {
      "epoch": 1.2486806479739343,
      "grad_norm": 0.4856208562850952,
      "learning_rate": 6.480585659503114e-06,
      "loss": 0.3929,
      "step": 27210
    },
    {
      "epoch": 1.2487265384791886,
      "grad_norm": 0.4629942774772644,
      "learning_rate": 6.4803514660929065e-06,
      "loss": 0.3576,
      "step": 27211
    },
    {
      "epoch": 1.248772428984443,
      "grad_norm": 0.48653069138526917,
      "learning_rate": 6.4801172691229204e-06,
      "loss": 0.3947,
      "step": 27212
    },
    {
      "epoch": 1.2488183194896976,
      "grad_norm": 0.43936172127723694,
      "learning_rate": 6.47988306859372e-06,
      "loss": 0.3342,
      "step": 27213
    },
    {
      "epoch": 1.248864209994952,
      "grad_norm": 0.49076971411705017,
      "learning_rate": 6.479648864505867e-06,
      "loss": 0.3826,
      "step": 27214
    },
    {
      "epoch": 1.2489101005002066,
      "grad_norm": 0.48805180191993713,
      "learning_rate": 6.479414656859926e-06,
      "loss": 0.3719,
      "step": 27215
    },
    {
      "epoch": 1.248955991005461,
      "grad_norm": 0.4522589445114136,
      "learning_rate": 6.47918044565646e-06,
      "loss": 0.3174,
      "step": 27216
    },
    {
      "epoch": 1.2490018815107153,
      "grad_norm": 0.4690479636192322,
      "learning_rate": 6.478946230896033e-06,
      "loss": 0.3743,
      "step": 27217
    },
    {
      "epoch": 1.2490477720159698,
      "grad_norm": 0.4970743954181671,
      "learning_rate": 6.478712012579205e-06,
      "loss": 0.378,
      "step": 27218
    },
    {
      "epoch": 1.2490936625212243,
      "grad_norm": 0.49913227558135986,
      "learning_rate": 6.478477790706541e-06,
      "loss": 0.4181,
      "step": 27219
    },
    {
      "epoch": 1.2491395530264788,
      "grad_norm": 0.4689364731311798,
      "learning_rate": 6.478243565278606e-06,
      "loss": 0.3248,
      "step": 27220
    },
    {
      "epoch": 1.2491854435317333,
      "grad_norm": 0.46845316886901855,
      "learning_rate": 6.4780093362959606e-06,
      "loss": 0.3372,
      "step": 27221
    },
    {
      "epoch": 1.2492313340369878,
      "grad_norm": 0.4573359489440918,
      "learning_rate": 6.477775103759169e-06,
      "loss": 0.3412,
      "step": 27222
    },
    {
      "epoch": 1.2492772245422423,
      "grad_norm": 0.4407694935798645,
      "learning_rate": 6.477540867668796e-06,
      "loss": 0.2921,
      "step": 27223
    },
    {
      "epoch": 1.2493231150474966,
      "grad_norm": 0.4281848669052124,
      "learning_rate": 6.477306628025404e-06,
      "loss": 0.2777,
      "step": 27224
    },
    {
      "epoch": 1.249369005552751,
      "grad_norm": 0.44407904148101807,
      "learning_rate": 6.4770723848295536e-06,
      "loss": 0.3662,
      "step": 27225
    },
    {
      "epoch": 1.2494148960580056,
      "grad_norm": 0.4712226688861847,
      "learning_rate": 6.476838138081811e-06,
      "loss": 0.4326,
      "step": 27226
    },
    {
      "epoch": 1.24946078656326,
      "grad_norm": 0.4651430547237396,
      "learning_rate": 6.476603887782741e-06,
      "loss": 0.3885,
      "step": 27227
    },
    {
      "epoch": 1.2495066770685146,
      "grad_norm": 0.42073217034339905,
      "learning_rate": 6.4763696339329016e-06,
      "loss": 0.3075,
      "step": 27228
    },
    {
      "epoch": 1.2495525675737689,
      "grad_norm": 0.4523237347602844,
      "learning_rate": 6.476135376532862e-06,
      "loss": 0.3176,
      "step": 27229
    },
    {
      "epoch": 1.2495984580790234,
      "grad_norm": 0.48597121238708496,
      "learning_rate": 6.4759011155831805e-06,
      "loss": 0.3587,
      "step": 27230
    },
    {
      "epoch": 1.2496443485842779,
      "grad_norm": 0.46277087926864624,
      "learning_rate": 6.475666851084422e-06,
      "loss": 0.3903,
      "step": 27231
    },
    {
      "epoch": 1.2496902390895324,
      "grad_norm": 0.4517253041267395,
      "learning_rate": 6.475432583037153e-06,
      "loss": 0.341,
      "step": 27232
    },
    {
      "epoch": 1.2497361295947869,
      "grad_norm": 0.5187800526618958,
      "learning_rate": 6.475198311441933e-06,
      "loss": 0.4078,
      "step": 27233
    },
    {
      "epoch": 1.2497820201000414,
      "grad_norm": 0.5033402442932129,
      "learning_rate": 6.474964036299328e-06,
      "loss": 0.4273,
      "step": 27234
    },
    {
      "epoch": 1.2498279106052959,
      "grad_norm": 0.45251283049583435,
      "learning_rate": 6.474729757609898e-06,
      "loss": 0.3037,
      "step": 27235
    },
    {
      "epoch": 1.2498738011105501,
      "grad_norm": 0.486616313457489,
      "learning_rate": 6.474495475374208e-06,
      "loss": 0.366,
      "step": 27236
    },
    {
      "epoch": 1.2499196916158046,
      "grad_norm": 0.49515849351882935,
      "learning_rate": 6.4742611895928245e-06,
      "loss": 0.4268,
      "step": 27237
    },
    {
      "epoch": 1.2499655821210591,
      "grad_norm": 0.4791102409362793,
      "learning_rate": 6.474026900266307e-06,
      "loss": 0.3547,
      "step": 27238
    },
    {
      "epoch": 1.2500114726263136,
      "grad_norm": 0.4404602646827698,
      "learning_rate": 6.4737926073952185e-06,
      "loss": 0.3096,
      "step": 27239
    },
    {
      "epoch": 1.2500573631315681,
      "grad_norm": 0.4520892798900604,
      "learning_rate": 6.473558310980126e-06,
      "loss": 0.3117,
      "step": 27240
    },
    {
      "epoch": 1.2501032536368226,
      "grad_norm": 0.5221748352050781,
      "learning_rate": 6.473324011021589e-06,
      "loss": 0.4096,
      "step": 27241
    },
    {
      "epoch": 1.250149144142077,
      "grad_norm": 0.4728541672229767,
      "learning_rate": 6.473089707520173e-06,
      "loss": 0.3596,
      "step": 27242
    },
    {
      "epoch": 1.2501950346473314,
      "grad_norm": 0.4331679344177246,
      "learning_rate": 6.472855400476442e-06,
      "loss": 0.2978,
      "step": 27243
    },
    {
      "epoch": 1.250240925152586,
      "grad_norm": 0.4619464576244354,
      "learning_rate": 6.4726210898909595e-06,
      "loss": 0.37,
      "step": 27244
    },
    {
      "epoch": 1.2502868156578404,
      "grad_norm": 0.47641655802726746,
      "learning_rate": 6.4723867757642855e-06,
      "loss": 0.3627,
      "step": 27245
    },
    {
      "epoch": 1.2503327061630949,
      "grad_norm": 0.49338141083717346,
      "learning_rate": 6.472152458096988e-06,
      "loss": 0.3702,
      "step": 27246
    },
    {
      "epoch": 1.2503785966683494,
      "grad_norm": 0.5107830166816711,
      "learning_rate": 6.471918136889628e-06,
      "loss": 0.3915,
      "step": 27247
    },
    {
      "epoch": 1.2504244871736039,
      "grad_norm": 0.5124556422233582,
      "learning_rate": 6.471683812142769e-06,
      "loss": 0.3591,
      "step": 27248
    },
    {
      "epoch": 1.2504703776788584,
      "grad_norm": 0.4866150915622711,
      "learning_rate": 6.471449483856977e-06,
      "loss": 0.3901,
      "step": 27249
    },
    {
      "epoch": 1.2505162681841127,
      "grad_norm": 0.4900915324687958,
      "learning_rate": 6.4712151520328106e-06,
      "loss": 0.3943,
      "step": 27250
    },
    {
      "epoch": 1.2505621586893672,
      "grad_norm": 0.48075392842292786,
      "learning_rate": 6.470980816670838e-06,
      "loss": 0.4632,
      "step": 27251
    },
    {
      "epoch": 1.2506080491946217,
      "grad_norm": 0.4823262393474579,
      "learning_rate": 6.470746477771619e-06,
      "loss": 0.3604,
      "step": 27252
    },
    {
      "epoch": 1.2506539396998761,
      "grad_norm": 0.4462185800075531,
      "learning_rate": 6.470512135335718e-06,
      "loss": 0.2931,
      "step": 27253
    },
    {
      "epoch": 1.2506998302051304,
      "grad_norm": 0.47803962230682373,
      "learning_rate": 6.470277789363703e-06,
      "loss": 0.3741,
      "step": 27254
    },
    {
      "epoch": 1.250745720710385,
      "grad_norm": 0.4516344666481018,
      "learning_rate": 6.470043439856132e-06,
      "loss": 0.3447,
      "step": 27255
    },
    {
      "epoch": 1.2507916112156394,
      "grad_norm": 0.4276409447193146,
      "learning_rate": 6.46980908681357e-06,
      "loss": 0.3131,
      "step": 27256
    },
    {
      "epoch": 1.250837501720894,
      "grad_norm": 0.4838365912437439,
      "learning_rate": 6.469574730236582e-06,
      "loss": 0.3954,
      "step": 27257
    },
    {
      "epoch": 1.2508833922261484,
      "grad_norm": 0.5313409566879272,
      "learning_rate": 6.46934037012573e-06,
      "loss": 0.4375,
      "step": 27258
    },
    {
      "epoch": 1.250929282731403,
      "grad_norm": 0.4585217535495758,
      "learning_rate": 6.4691060064815785e-06,
      "loss": 0.3827,
      "step": 27259
    },
    {
      "epoch": 1.2509751732366574,
      "grad_norm": 0.5102069973945618,
      "learning_rate": 6.4688716393046915e-06,
      "loss": 0.3667,
      "step": 27260
    },
    {
      "epoch": 1.251021063741912,
      "grad_norm": 0.4488201439380646,
      "learning_rate": 6.46863726859563e-06,
      "loss": 0.317,
      "step": 27261
    },
    {
      "epoch": 1.2510669542471662,
      "grad_norm": 0.5075010657310486,
      "learning_rate": 6.46840289435496e-06,
      "loss": 0.4359,
      "step": 27262
    },
    {
      "epoch": 1.2511128447524207,
      "grad_norm": 0.5182188749313354,
      "learning_rate": 6.468168516583246e-06,
      "loss": 0.4055,
      "step": 27263
    },
    {
      "epoch": 1.2511587352576752,
      "grad_norm": 0.4916212558746338,
      "learning_rate": 6.4679341352810474e-06,
      "loss": 0.4283,
      "step": 27264
    },
    {
      "epoch": 1.2512046257629297,
      "grad_norm": 0.5063515305519104,
      "learning_rate": 6.467699750448933e-06,
      "loss": 0.41,
      "step": 27265
    },
    {
      "epoch": 1.2512505162681842,
      "grad_norm": 0.4606711268424988,
      "learning_rate": 6.467465362087463e-06,
      "loss": 0.3375,
      "step": 27266
    },
    {
      "epoch": 1.2512964067734385,
      "grad_norm": 0.470992773771286,
      "learning_rate": 6.467230970197201e-06,
      "loss": 0.3258,
      "step": 27267
    },
    {
      "epoch": 1.251342297278693,
      "grad_norm": 0.44808486104011536,
      "learning_rate": 6.4669965747787135e-06,
      "loss": 0.3455,
      "step": 27268
    },
    {
      "epoch": 1.2513881877839474,
      "grad_norm": 0.47690799832344055,
      "learning_rate": 6.466762175832561e-06,
      "loss": 0.3703,
      "step": 27269
    },
    {
      "epoch": 1.251434078289202,
      "grad_norm": 0.4546023905277252,
      "learning_rate": 6.466527773359307e-06,
      "loss": 0.3986,
      "step": 27270
    },
    {
      "epoch": 1.2514799687944564,
      "grad_norm": 0.45347532629966736,
      "learning_rate": 6.466293367359519e-06,
      "loss": 0.3243,
      "step": 27271
    },
    {
      "epoch": 1.251525859299711,
      "grad_norm": 0.4647709131240845,
      "learning_rate": 6.466058957833757e-06,
      "loss": 0.374,
      "step": 27272
    },
    {
      "epoch": 1.2515717498049654,
      "grad_norm": 0.4561839699745178,
      "learning_rate": 6.4658245447825865e-06,
      "loss": 0.3298,
      "step": 27273
    },
    {
      "epoch": 1.25161764031022,
      "grad_norm": 0.5078872442245483,
      "learning_rate": 6.46559012820657e-06,
      "loss": 0.3837,
      "step": 27274
    },
    {
      "epoch": 1.2516635308154742,
      "grad_norm": 0.4874109625816345,
      "learning_rate": 6.465355708106272e-06,
      "loss": 0.3703,
      "step": 27275
    },
    {
      "epoch": 1.2517094213207287,
      "grad_norm": 0.46631520986557007,
      "learning_rate": 6.465121284482256e-06,
      "loss": 0.3529,
      "step": 27276
    },
    {
      "epoch": 1.2517553118259832,
      "grad_norm": 0.47768092155456543,
      "learning_rate": 6.464886857335087e-06,
      "loss": 0.3945,
      "step": 27277
    },
    {
      "epoch": 1.2518012023312377,
      "grad_norm": 0.49916690587997437,
      "learning_rate": 6.464652426665325e-06,
      "loss": 0.3714,
      "step": 27278
    },
    {
      "epoch": 1.2518470928364922,
      "grad_norm": 0.4545755684375763,
      "learning_rate": 6.4644179924735375e-06,
      "loss": 0.3271,
      "step": 27279
    },
    {
      "epoch": 1.2518929833417465,
      "grad_norm": 0.45780059695243835,
      "learning_rate": 6.464183554760287e-06,
      "loss": 0.3794,
      "step": 27280
    },
    {
      "epoch": 1.251938873847001,
      "grad_norm": 0.4386332929134369,
      "learning_rate": 6.463949113526137e-06,
      "loss": 0.3137,
      "step": 27281
    },
    {
      "epoch": 1.2519847643522555,
      "grad_norm": 0.49703049659729004,
      "learning_rate": 6.463714668771652e-06,
      "loss": 0.4014,
      "step": 27282
    },
    {
      "epoch": 1.25203065485751,
      "grad_norm": 0.4717913269996643,
      "learning_rate": 6.463480220497395e-06,
      "loss": 0.4001,
      "step": 27283
    },
    {
      "epoch": 1.2520765453627645,
      "grad_norm": 0.5007030367851257,
      "learning_rate": 6.4632457687039295e-06,
      "loss": 0.4419,
      "step": 27284
    },
    {
      "epoch": 1.252122435868019,
      "grad_norm": 0.4780285656452179,
      "learning_rate": 6.46301131339182e-06,
      "loss": 0.3943,
      "step": 27285
    },
    {
      "epoch": 1.2521683263732735,
      "grad_norm": 0.44789066910743713,
      "learning_rate": 6.46277685456163e-06,
      "loss": 0.3161,
      "step": 27286
    },
    {
      "epoch": 1.252214216878528,
      "grad_norm": 0.49652865529060364,
      "learning_rate": 6.462542392213923e-06,
      "loss": 0.4167,
      "step": 27287
    },
    {
      "epoch": 1.2522601073837822,
      "grad_norm": 0.42882874608039856,
      "learning_rate": 6.462307926349264e-06,
      "loss": 0.2817,
      "step": 27288
    },
    {
      "epoch": 1.2523059978890367,
      "grad_norm": 0.4951099157333374,
      "learning_rate": 6.462073456968214e-06,
      "loss": 0.3456,
      "step": 27289
    },
    {
      "epoch": 1.2523518883942912,
      "grad_norm": 0.4809695780277252,
      "learning_rate": 6.46183898407134e-06,
      "loss": 0.3839,
      "step": 27290
    },
    {
      "epoch": 1.2523977788995457,
      "grad_norm": 0.42414382100105286,
      "learning_rate": 6.461604507659205e-06,
      "loss": 0.2902,
      "step": 27291
    },
    {
      "epoch": 1.2524436694048,
      "grad_norm": 0.4763728976249695,
      "learning_rate": 6.461370027732372e-06,
      "loss": 0.3859,
      "step": 27292
    },
    {
      "epoch": 1.2524895599100545,
      "grad_norm": 0.5009507536888123,
      "learning_rate": 6.461135544291406e-06,
      "loss": 0.4565,
      "step": 27293
    },
    {
      "epoch": 1.252535450415309,
      "grad_norm": 0.5027029514312744,
      "learning_rate": 6.46090105733687e-06,
      "loss": 0.4065,
      "step": 27294
    },
    {
      "epoch": 1.2525813409205635,
      "grad_norm": 0.48650291562080383,
      "learning_rate": 6.460666566869327e-06,
      "loss": 0.4255,
      "step": 27295
    },
    {
      "epoch": 1.252627231425818,
      "grad_norm": 0.5015364289283752,
      "learning_rate": 6.4604320728893426e-06,
      "loss": 0.4235,
      "step": 27296
    },
    {
      "epoch": 1.2526731219310725,
      "grad_norm": 0.49396607279777527,
      "learning_rate": 6.46019757539748e-06,
      "loss": 0.3616,
      "step": 27297
    },
    {
      "epoch": 1.252719012436327,
      "grad_norm": 0.5200433731079102,
      "learning_rate": 6.459963074394303e-06,
      "loss": 0.3297,
      "step": 27298
    },
    {
      "epoch": 1.2527649029415815,
      "grad_norm": 0.4375634491443634,
      "learning_rate": 6.4597285698803755e-06,
      "loss": 0.3509,
      "step": 27299
    },
    {
      "epoch": 1.2528107934468358,
      "grad_norm": 0.47292789816856384,
      "learning_rate": 6.4594940618562615e-06,
      "loss": 0.3988,
      "step": 27300
    },
    {
      "epoch": 1.2528566839520903,
      "grad_norm": 0.4142512381076813,
      "learning_rate": 6.459259550322524e-06,
      "loss": 0.2741,
      "step": 27301
    },
    {
      "epoch": 1.2529025744573448,
      "grad_norm": 0.40675899386405945,
      "learning_rate": 6.45902503527973e-06,
      "loss": 0.2475,
      "step": 27302
    },
    {
      "epoch": 1.2529484649625993,
      "grad_norm": 0.5075702667236328,
      "learning_rate": 6.45879051672844e-06,
      "loss": 0.4184,
      "step": 27303
    },
    {
      "epoch": 1.2529943554678538,
      "grad_norm": 0.48809295892715454,
      "learning_rate": 6.458555994669219e-06,
      "loss": 0.3865,
      "step": 27304
    },
    {
      "epoch": 1.253040245973108,
      "grad_norm": 0.4407748281955719,
      "learning_rate": 6.458321469102632e-06,
      "loss": 0.3288,
      "step": 27305
    },
    {
      "epoch": 1.2530861364783625,
      "grad_norm": 0.46890732645988464,
      "learning_rate": 6.458086940029241e-06,
      "loss": 0.3237,
      "step": 27306
    },
    {
      "epoch": 1.253132026983617,
      "grad_norm": 0.48231950402259827,
      "learning_rate": 6.457852407449612e-06,
      "loss": 0.3485,
      "step": 27307
    },
    {
      "epoch": 1.2531779174888715,
      "grad_norm": 0.5226314663887024,
      "learning_rate": 6.457617871364308e-06,
      "loss": 0.3753,
      "step": 27308
    },
    {
      "epoch": 1.253223807994126,
      "grad_norm": 0.46313074231147766,
      "learning_rate": 6.457383331773893e-06,
      "loss": 0.3608,
      "step": 27309
    },
    {
      "epoch": 1.2532696984993805,
      "grad_norm": 0.4911595284938812,
      "learning_rate": 6.457148788678931e-06,
      "loss": 0.3746,
      "step": 27310
    },
    {
      "epoch": 1.253315589004635,
      "grad_norm": 0.45556026697158813,
      "learning_rate": 6.456914242079986e-06,
      "loss": 0.3074,
      "step": 27311
    },
    {
      "epoch": 1.2533614795098895,
      "grad_norm": 0.5266536474227905,
      "learning_rate": 6.4566796919776205e-06,
      "loss": 0.4278,
      "step": 27312
    },
    {
      "epoch": 1.2534073700151438,
      "grad_norm": 0.46113714575767517,
      "learning_rate": 6.456445138372401e-06,
      "loss": 0.3475,
      "step": 27313
    },
    {
      "epoch": 1.2534532605203983,
      "grad_norm": 0.4841267764568329,
      "learning_rate": 6.456210581264892e-06,
      "loss": 0.3841,
      "step": 27314
    },
    {
      "epoch": 1.2534991510256528,
      "grad_norm": 0.4584927558898926,
      "learning_rate": 6.455976020655656e-06,
      "loss": 0.327,
      "step": 27315
    },
    {
      "epoch": 1.2535450415309073,
      "grad_norm": 0.4986528158187866,
      "learning_rate": 6.455741456545257e-06,
      "loss": 0.4362,
      "step": 27316
    },
    {
      "epoch": 1.2535909320361618,
      "grad_norm": 0.4732993245124817,
      "learning_rate": 6.4555068889342586e-06,
      "loss": 0.3799,
      "step": 27317
    },
    {
      "epoch": 1.253636822541416,
      "grad_norm": 0.4676668345928192,
      "learning_rate": 6.455272317823225e-06,
      "loss": 0.3543,
      "step": 27318
    },
    {
      "epoch": 1.2536827130466706,
      "grad_norm": 0.46188101172447205,
      "learning_rate": 6.455037743212722e-06,
      "loss": 0.351,
      "step": 27319
    },
    {
      "epoch": 1.253728603551925,
      "grad_norm": 0.4650830030441284,
      "learning_rate": 6.454803165103312e-06,
      "loss": 0.3983,
      "step": 27320
    },
    {
      "epoch": 1.2537744940571796,
      "grad_norm": 0.4522091746330261,
      "learning_rate": 6.454568583495559e-06,
      "loss": 0.3641,
      "step": 27321
    },
    {
      "epoch": 1.253820384562434,
      "grad_norm": 0.4822550117969513,
      "learning_rate": 6.454333998390029e-06,
      "loss": 0.3818,
      "step": 27322
    },
    {
      "epoch": 1.2538662750676886,
      "grad_norm": 0.4960421621799469,
      "learning_rate": 6.454099409787284e-06,
      "loss": 0.3993,
      "step": 27323
    },
    {
      "epoch": 1.253912165572943,
      "grad_norm": 0.5017940998077393,
      "learning_rate": 6.453864817687887e-06,
      "loss": 0.461,
      "step": 27324
    },
    {
      "epoch": 1.2539580560781975,
      "grad_norm": 0.4477796256542206,
      "learning_rate": 6.453630222092406e-06,
      "loss": 0.3523,
      "step": 27325
    },
    {
      "epoch": 1.2540039465834518,
      "grad_norm": 0.4454113245010376,
      "learning_rate": 6.453395623001404e-06,
      "loss": 0.3267,
      "step": 27326
    },
    {
      "epoch": 1.2540498370887063,
      "grad_norm": 0.46828290820121765,
      "learning_rate": 6.453161020415442e-06,
      "loss": 0.4084,
      "step": 27327
    },
    {
      "epoch": 1.2540957275939608,
      "grad_norm": 0.46313586831092834,
      "learning_rate": 6.452926414335088e-06,
      "loss": 0.3653,
      "step": 27328
    },
    {
      "epoch": 1.2541416180992153,
      "grad_norm": 0.4159192740917206,
      "learning_rate": 6.452691804760902e-06,
      "loss": 0.2897,
      "step": 27329
    },
    {
      "epoch": 1.2541875086044698,
      "grad_norm": 0.44550374150276184,
      "learning_rate": 6.452457191693454e-06,
      "loss": 0.3429,
      "step": 27330
    },
    {
      "epoch": 1.254233399109724,
      "grad_norm": 0.43923720717430115,
      "learning_rate": 6.452222575133304e-06,
      "loss": 0.3058,
      "step": 27331
    },
    {
      "epoch": 1.2542792896149786,
      "grad_norm": 0.4404500424861908,
      "learning_rate": 6.451987955081016e-06,
      "loss": 0.3122,
      "step": 27332
    },
    {
      "epoch": 1.254325180120233,
      "grad_norm": 0.4310649037361145,
      "learning_rate": 6.451753331537156e-06,
      "loss": 0.3359,
      "step": 27333
    },
    {
      "epoch": 1.2543710706254876,
      "grad_norm": 0.416260689496994,
      "learning_rate": 6.451518704502286e-06,
      "loss": 0.2686,
      "step": 27334
    },
    {
      "epoch": 1.254416961130742,
      "grad_norm": 0.43878620862960815,
      "learning_rate": 6.4512840739769726e-06,
      "loss": 0.372,
      "step": 27335
    },
    {
      "epoch": 1.2544628516359966,
      "grad_norm": 0.4648166000843048,
      "learning_rate": 6.45104943996178e-06,
      "loss": 0.3414,
      "step": 27336
    },
    {
      "epoch": 1.254508742141251,
      "grad_norm": 0.4448631703853607,
      "learning_rate": 6.45081480245727e-06,
      "loss": 0.2514,
      "step": 27337
    },
    {
      "epoch": 1.2545546326465056,
      "grad_norm": 0.49841955304145813,
      "learning_rate": 6.450580161464008e-06,
      "loss": 0.385,
      "step": 27338
    },
    {
      "epoch": 1.2546005231517598,
      "grad_norm": 0.49078845977783203,
      "learning_rate": 6.450345516982561e-06,
      "loss": 0.4091,
      "step": 27339
    },
    {
      "epoch": 1.2546464136570143,
      "grad_norm": 0.46568289399147034,
      "learning_rate": 6.450110869013488e-06,
      "loss": 0.3583,
      "step": 27340
    },
    {
      "epoch": 1.2546923041622688,
      "grad_norm": 0.4634729325771332,
      "learning_rate": 6.449876217557358e-06,
      "loss": 0.3439,
      "step": 27341
    },
    {
      "epoch": 1.2547381946675233,
      "grad_norm": 0.5169396996498108,
      "learning_rate": 6.449641562614733e-06,
      "loss": 0.4693,
      "step": 27342
    },
    {
      "epoch": 1.2547840851727776,
      "grad_norm": 0.4925481677055359,
      "learning_rate": 6.449406904186177e-06,
      "loss": 0.3334,
      "step": 27343
    },
    {
      "epoch": 1.2548299756780321,
      "grad_norm": 0.4659465253353119,
      "learning_rate": 6.449172242272255e-06,
      "loss": 0.3493,
      "step": 27344
    },
    {
      "epoch": 1.2548758661832866,
      "grad_norm": 0.4478093087673187,
      "learning_rate": 6.448937576873531e-06,
      "loss": 0.3264,
      "step": 27345
    },
    {
      "epoch": 1.2549217566885411,
      "grad_norm": 0.4676917493343353,
      "learning_rate": 6.448702907990568e-06,
      "loss": 0.3635,
      "step": 27346
    },
    {
      "epoch": 1.2549676471937956,
      "grad_norm": 0.4481979012489319,
      "learning_rate": 6.448468235623935e-06,
      "loss": 0.3355,
      "step": 27347
    },
    {
      "epoch": 1.25501353769905,
      "grad_norm": 0.450545072555542,
      "learning_rate": 6.4482335597741905e-06,
      "loss": 0.3451,
      "step": 27348
    },
    {
      "epoch": 1.2550594282043046,
      "grad_norm": 0.45113304257392883,
      "learning_rate": 6.447998880441902e-06,
      "loss": 0.2902,
      "step": 27349
    },
    {
      "epoch": 1.255105318709559,
      "grad_norm": 0.4956468939781189,
      "learning_rate": 6.447764197627634e-06,
      "loss": 0.406,
      "step": 27350
    },
    {
      "epoch": 1.2551512092148134,
      "grad_norm": 0.47298917174339294,
      "learning_rate": 6.447529511331949e-06,
      "loss": 0.3796,
      "step": 27351
    },
    {
      "epoch": 1.2551970997200679,
      "grad_norm": 0.4574783444404602,
      "learning_rate": 6.447294821555413e-06,
      "loss": 0.3448,
      "step": 27352
    },
    {
      "epoch": 1.2552429902253224,
      "grad_norm": 0.4723958671092987,
      "learning_rate": 6.44706012829859e-06,
      "loss": 0.3888,
      "step": 27353
    },
    {
      "epoch": 1.2552888807305769,
      "grad_norm": 0.4733197093009949,
      "learning_rate": 6.446825431562043e-06,
      "loss": 0.3426,
      "step": 27354
    },
    {
      "epoch": 1.2553347712358314,
      "grad_norm": 0.4987959563732147,
      "learning_rate": 6.446590731346337e-06,
      "loss": 0.4193,
      "step": 27355
    },
    {
      "epoch": 1.2553806617410856,
      "grad_norm": 0.46568313241004944,
      "learning_rate": 6.446356027652039e-06,
      "loss": 0.3772,
      "step": 27356
    },
    {
      "epoch": 1.2554265522463401,
      "grad_norm": 0.44380950927734375,
      "learning_rate": 6.446121320479709e-06,
      "loss": 0.3134,
      "step": 27357
    },
    {
      "epoch": 1.2554724427515946,
      "grad_norm": 0.47735467553138733,
      "learning_rate": 6.445886609829916e-06,
      "loss": 0.4035,
      "step": 27358
    },
    {
      "epoch": 1.2555183332568491,
      "grad_norm": 0.44859904050827026,
      "learning_rate": 6.445651895703221e-06,
      "loss": 0.3243,
      "step": 27359
    },
    {
      "epoch": 1.2555642237621036,
      "grad_norm": 0.4424631893634796,
      "learning_rate": 6.445417178100188e-06,
      "loss": 0.3525,
      "step": 27360
    },
    {
      "epoch": 1.2556101142673581,
      "grad_norm": 0.48520800471305847,
      "learning_rate": 6.445182457021384e-06,
      "loss": 0.3882,
      "step": 27361
    },
    {
      "epoch": 1.2556560047726126,
      "grad_norm": 0.519506573677063,
      "learning_rate": 6.444947732467372e-06,
      "loss": 0.4342,
      "step": 27362
    },
    {
      "epoch": 1.2557018952778671,
      "grad_norm": 0.4862143099308014,
      "learning_rate": 6.444713004438719e-06,
      "loss": 0.3842,
      "step": 27363
    },
    {
      "epoch": 1.2557477857831214,
      "grad_norm": 0.47181349992752075,
      "learning_rate": 6.444478272935985e-06,
      "loss": 0.3706,
      "step": 27364
    },
    {
      "epoch": 1.255793676288376,
      "grad_norm": 0.46996310353279114,
      "learning_rate": 6.4442435379597365e-06,
      "loss": 0.3711,
      "step": 27365
    },
    {
      "epoch": 1.2558395667936304,
      "grad_norm": 0.46993768215179443,
      "learning_rate": 6.444008799510538e-06,
      "loss": 0.3296,
      "step": 27366
    },
    {
      "epoch": 1.255885457298885,
      "grad_norm": 0.4352472424507141,
      "learning_rate": 6.443774057588956e-06,
      "loss": 0.3062,
      "step": 27367
    },
    {
      "epoch": 1.2559313478041394,
      "grad_norm": 0.47730687260627747,
      "learning_rate": 6.44353931219555e-06,
      "loss": 0.3869,
      "step": 27368
    },
    {
      "epoch": 1.2559772383093937,
      "grad_norm": 0.46295374631881714,
      "learning_rate": 6.443304563330891e-06,
      "loss": 0.3439,
      "step": 27369
    },
    {
      "epoch": 1.2560231288146482,
      "grad_norm": 0.45740842819213867,
      "learning_rate": 6.44306981099554e-06,
      "loss": 0.3291,
      "step": 27370
    },
    {
      "epoch": 1.2560690193199027,
      "grad_norm": 0.4948623776435852,
      "learning_rate": 6.4428350551900574e-06,
      "loss": 0.3984,
      "step": 27371
    },
    {
      "epoch": 1.2561149098251572,
      "grad_norm": 0.45749011635780334,
      "learning_rate": 6.442600295915015e-06,
      "loss": 0.3415,
      "step": 27372
    },
    {
      "epoch": 1.2561608003304117,
      "grad_norm": 0.47931888699531555,
      "learning_rate": 6.442365533170973e-06,
      "loss": 0.3926,
      "step": 27373
    },
    {
      "epoch": 1.2562066908356662,
      "grad_norm": 0.48011156916618347,
      "learning_rate": 6.442130766958499e-06,
      "loss": 0.3607,
      "step": 27374
    },
    {
      "epoch": 1.2562525813409207,
      "grad_norm": 0.4658646285533905,
      "learning_rate": 6.441895997278154e-06,
      "loss": 0.3889,
      "step": 27375
    },
    {
      "epoch": 1.2562984718461752,
      "grad_norm": 0.4985107481479645,
      "learning_rate": 6.4416612241305045e-06,
      "loss": 0.4059,
      "step": 27376
    },
    {
      "epoch": 1.2563443623514294,
      "grad_norm": 0.479005366563797,
      "learning_rate": 6.441426447516114e-06,
      "loss": 0.3684,
      "step": 27377
    },
    {
      "epoch": 1.256390252856684,
      "grad_norm": 0.49608471989631653,
      "learning_rate": 6.4411916674355475e-06,
      "loss": 0.3488,
      "step": 27378
    },
    {
      "epoch": 1.2564361433619384,
      "grad_norm": 0.4559023678302765,
      "learning_rate": 6.440956883889372e-06,
      "loss": 0.3502,
      "step": 27379
    },
    {
      "epoch": 1.256482033867193,
      "grad_norm": 0.4452434480190277,
      "learning_rate": 6.440722096878149e-06,
      "loss": 0.299,
      "step": 27380
    },
    {
      "epoch": 1.2565279243724472,
      "grad_norm": 0.5096319317817688,
      "learning_rate": 6.440487306402446e-06,
      "loss": 0.3857,
      "step": 27381
    },
    {
      "epoch": 1.2565738148777017,
      "grad_norm": 0.48911380767822266,
      "learning_rate": 6.440252512462823e-06,
      "loss": 0.3912,
      "step": 27382
    },
    {
      "epoch": 1.2566197053829562,
      "grad_norm": 0.48096635937690735,
      "learning_rate": 6.4400177150598464e-06,
      "loss": 0.3833,
      "step": 27383
    },
    {
      "epoch": 1.2566655958882107,
      "grad_norm": 0.4651685357093811,
      "learning_rate": 6.4397829141940834e-06,
      "loss": 0.3839,
      "step": 27384
    },
    {
      "epoch": 1.2567114863934652,
      "grad_norm": 0.45169171690940857,
      "learning_rate": 6.439548109866097e-06,
      "loss": 0.3504,
      "step": 27385
    },
    {
      "epoch": 1.2567573768987197,
      "grad_norm": 0.5323764085769653,
      "learning_rate": 6.439313302076451e-06,
      "loss": 0.4764,
      "step": 27386
    },
    {
      "epoch": 1.2568032674039742,
      "grad_norm": 0.4629278779029846,
      "learning_rate": 6.439078490825712e-06,
      "loss": 0.3094,
      "step": 27387
    },
    {
      "epoch": 1.2568491579092287,
      "grad_norm": 0.5124077796936035,
      "learning_rate": 6.438843676114442e-06,
      "loss": 0.3843,
      "step": 27388
    },
    {
      "epoch": 1.256895048414483,
      "grad_norm": 0.44896194338798523,
      "learning_rate": 6.4386088579432075e-06,
      "loss": 0.3178,
      "step": 27389
    },
    {
      "epoch": 1.2569409389197375,
      "grad_norm": 0.4462103545665741,
      "learning_rate": 6.438374036312573e-06,
      "loss": 0.3131,
      "step": 27390
    },
    {
      "epoch": 1.256986829424992,
      "grad_norm": 0.5211013555526733,
      "learning_rate": 6.438139211223104e-06,
      "loss": 0.4195,
      "step": 27391
    },
    {
      "epoch": 1.2570327199302465,
      "grad_norm": 0.47750577330589294,
      "learning_rate": 6.437904382675362e-06,
      "loss": 0.3717,
      "step": 27392
    },
    {
      "epoch": 1.257078610435501,
      "grad_norm": 0.46769195795059204,
      "learning_rate": 6.437669550669915e-06,
      "loss": 0.3464,
      "step": 27393
    },
    {
      "epoch": 1.2571245009407552,
      "grad_norm": 0.4533455967903137,
      "learning_rate": 6.437434715207326e-06,
      "loss": 0.3603,
      "step": 27394
    },
    {
      "epoch": 1.2571703914460097,
      "grad_norm": 0.44893261790275574,
      "learning_rate": 6.43719987628816e-06,
      "loss": 0.3029,
      "step": 27395
    },
    {
      "epoch": 1.2572162819512642,
      "grad_norm": 0.48795512318611145,
      "learning_rate": 6.436965033912983e-06,
      "loss": 0.4208,
      "step": 27396
    },
    {
      "epoch": 1.2572621724565187,
      "grad_norm": 0.4330460727214813,
      "learning_rate": 6.436730188082357e-06,
      "loss": 0.3209,
      "step": 27397
    },
    {
      "epoch": 1.2573080629617732,
      "grad_norm": 0.4401741027832031,
      "learning_rate": 6.436495338796851e-06,
      "loss": 0.3352,
      "step": 27398
    },
    {
      "epoch": 1.2573539534670277,
      "grad_norm": 0.48083239793777466,
      "learning_rate": 6.4362604860570245e-06,
      "loss": 0.4377,
      "step": 27399
    },
    {
      "epoch": 1.2573998439722822,
      "grad_norm": 0.43463778495788574,
      "learning_rate": 6.436025629863445e-06,
      "loss": 0.3104,
      "step": 27400
    },
    {
      "epoch": 1.2574457344775367,
      "grad_norm": 0.4763415455818176,
      "learning_rate": 6.4357907702166774e-06,
      "loss": 0.3303,
      "step": 27401
    },
    {
      "epoch": 1.257491624982791,
      "grad_norm": 0.456005722284317,
      "learning_rate": 6.435555907117288e-06,
      "loss": 0.3485,
      "step": 27402
    },
    {
      "epoch": 1.2575375154880455,
      "grad_norm": 0.44495245814323425,
      "learning_rate": 6.435321040565838e-06,
      "loss": 0.3221,
      "step": 27403
    },
    {
      "epoch": 1.2575834059933,
      "grad_norm": 0.4608716666698456,
      "learning_rate": 6.435086170562893e-06,
      "loss": 0.3572,
      "step": 27404
    },
    {
      "epoch": 1.2576292964985545,
      "grad_norm": 0.4946003258228302,
      "learning_rate": 6.43485129710902e-06,
      "loss": 0.4525,
      "step": 27405
    },
    {
      "epoch": 1.257675187003809,
      "grad_norm": 0.4673561751842499,
      "learning_rate": 6.434616420204781e-06,
      "loss": 0.3377,
      "step": 27406
    },
    {
      "epoch": 1.2577210775090633,
      "grad_norm": 0.4342801570892334,
      "learning_rate": 6.4343815398507445e-06,
      "loss": 0.2992,
      "step": 27407
    },
    {
      "epoch": 1.2577669680143178,
      "grad_norm": 0.43486708402633667,
      "learning_rate": 6.4341466560474734e-06,
      "loss": 0.31,
      "step": 27408
    },
    {
      "epoch": 1.2578128585195723,
      "grad_norm": 0.4756821095943451,
      "learning_rate": 6.433911768795529e-06,
      "loss": 0.3776,
      "step": 27409
    },
    {
      "epoch": 1.2578587490248267,
      "grad_norm": 0.4886906147003174,
      "learning_rate": 6.433676878095483e-06,
      "loss": 0.3615,
      "step": 27410
    },
    {
      "epoch": 1.2579046395300812,
      "grad_norm": 0.4969492554664612,
      "learning_rate": 6.433441983947894e-06,
      "loss": 0.4425,
      "step": 27411
    },
    {
      "epoch": 1.2579505300353357,
      "grad_norm": 0.4656849503517151,
      "learning_rate": 6.433207086353331e-06,
      "loss": 0.3977,
      "step": 27412
    },
    {
      "epoch": 1.2579964205405902,
      "grad_norm": 0.4875065088272095,
      "learning_rate": 6.432972185312356e-06,
      "loss": 0.3805,
      "step": 27413
    },
    {
      "epoch": 1.2580423110458447,
      "grad_norm": 0.4738095998764038,
      "learning_rate": 6.432737280825536e-06,
      "loss": 0.3411,
      "step": 27414
    },
    {
      "epoch": 1.258088201551099,
      "grad_norm": 0.4672408103942871,
      "learning_rate": 6.432502372893435e-06,
      "loss": 0.3446,
      "step": 27415
    },
    {
      "epoch": 1.2581340920563535,
      "grad_norm": 0.5167033672332764,
      "learning_rate": 6.432267461516617e-06,
      "loss": 0.4079,
      "step": 27416
    },
    {
      "epoch": 1.258179982561608,
      "grad_norm": 0.49711379408836365,
      "learning_rate": 6.4320325466956475e-06,
      "loss": 0.3765,
      "step": 27417
    },
    {
      "epoch": 1.2582258730668625,
      "grad_norm": 0.4630419611930847,
      "learning_rate": 6.431797628431094e-06,
      "loss": 0.3144,
      "step": 27418
    },
    {
      "epoch": 1.258271763572117,
      "grad_norm": 0.5263121128082275,
      "learning_rate": 6.431562706723517e-06,
      "loss": 0.4252,
      "step": 27419
    },
    {
      "epoch": 1.2583176540773713,
      "grad_norm": 0.4480116367340088,
      "learning_rate": 6.431327781573484e-06,
      "loss": 0.3088,
      "step": 27420
    },
    {
      "epoch": 1.2583635445826258,
      "grad_norm": 0.4972078800201416,
      "learning_rate": 6.4310928529815605e-06,
      "loss": 0.3808,
      "step": 27421
    },
    {
      "epoch": 1.2584094350878803,
      "grad_norm": 0.5099966526031494,
      "learning_rate": 6.430857920948309e-06,
      "loss": 0.4092,
      "step": 27422
    },
    {
      "epoch": 1.2584553255931348,
      "grad_norm": 0.4731709659099579,
      "learning_rate": 6.4306229854742975e-06,
      "loss": 0.3377,
      "step": 27423
    },
    {
      "epoch": 1.2585012160983893,
      "grad_norm": 0.441536009311676,
      "learning_rate": 6.430388046560087e-06,
      "loss": 0.3295,
      "step": 27424
    },
    {
      "epoch": 1.2585471066036438,
      "grad_norm": 0.5139393210411072,
      "learning_rate": 6.430153104206246e-06,
      "loss": 0.3394,
      "step": 27425
    },
    {
      "epoch": 1.2585929971088983,
      "grad_norm": 0.4942586421966553,
      "learning_rate": 6.429918158413338e-06,
      "loss": 0.4194,
      "step": 27426
    },
    {
      "epoch": 1.2586388876141528,
      "grad_norm": 0.5035301446914673,
      "learning_rate": 6.429683209181928e-06,
      "loss": 0.3469,
      "step": 27427
    },
    {
      "epoch": 1.258684778119407,
      "grad_norm": 0.4992572069168091,
      "learning_rate": 6.429448256512581e-06,
      "loss": 0.3907,
      "step": 27428
    },
    {
      "epoch": 1.2587306686246615,
      "grad_norm": 0.45103514194488525,
      "learning_rate": 6.429213300405862e-06,
      "loss": 0.2883,
      "step": 27429
    },
    {
      "epoch": 1.258776559129916,
      "grad_norm": 0.5293766856193542,
      "learning_rate": 6.4289783408623375e-06,
      "loss": 0.3804,
      "step": 27430
    },
    {
      "epoch": 1.2588224496351705,
      "grad_norm": 0.4605061411857605,
      "learning_rate": 6.4287433778825694e-06,
      "loss": 0.3605,
      "step": 27431
    },
    {
      "epoch": 1.2588683401404248,
      "grad_norm": 0.527533769607544,
      "learning_rate": 6.428508411467125e-06,
      "loss": 0.4505,
      "step": 27432
    },
    {
      "epoch": 1.2589142306456793,
      "grad_norm": 0.45651015639305115,
      "learning_rate": 6.4282734416165685e-06,
      "loss": 0.3497,
      "step": 27433
    },
    {
      "epoch": 1.2589601211509338,
      "grad_norm": 0.468691885471344,
      "learning_rate": 6.4280384683314655e-06,
      "loss": 0.3984,
      "step": 27434
    },
    {
      "epoch": 1.2590060116561883,
      "grad_norm": 0.4796467125415802,
      "learning_rate": 6.427803491612381e-06,
      "loss": 0.4059,
      "step": 27435
    },
    {
      "epoch": 1.2590519021614428,
      "grad_norm": 0.48359155654907227,
      "learning_rate": 6.427568511459879e-06,
      "loss": 0.345,
      "step": 27436
    },
    {
      "epoch": 1.2590977926666973,
      "grad_norm": 0.5045320987701416,
      "learning_rate": 6.427333527874524e-06,
      "loss": 0.3704,
      "step": 27437
    },
    {
      "epoch": 1.2591436831719518,
      "grad_norm": 0.4661056101322174,
      "learning_rate": 6.427098540856884e-06,
      "loss": 0.3578,
      "step": 27438
    },
    {
      "epoch": 1.2591895736772063,
      "grad_norm": 0.4306385815143585,
      "learning_rate": 6.426863550407522e-06,
      "loss": 0.3231,
      "step": 27439
    },
    {
      "epoch": 1.2592354641824606,
      "grad_norm": 0.46582722663879395,
      "learning_rate": 6.426628556527005e-06,
      "loss": 0.3659,
      "step": 27440
    },
    {
      "epoch": 1.259281354687715,
      "grad_norm": 0.47344449162483215,
      "learning_rate": 6.426393559215895e-06,
      "loss": 0.3624,
      "step": 27441
    },
    {
      "epoch": 1.2593272451929696,
      "grad_norm": 0.4470025300979614,
      "learning_rate": 6.426158558474759e-06,
      "loss": 0.3232,
      "step": 27442
    },
    {
      "epoch": 1.259373135698224,
      "grad_norm": 0.4395503103733063,
      "learning_rate": 6.425923554304161e-06,
      "loss": 0.2822,
      "step": 27443
    },
    {
      "epoch": 1.2594190262034786,
      "grad_norm": 0.4978513717651367,
      "learning_rate": 6.425688546704668e-06,
      "loss": 0.4115,
      "step": 27444
    },
    {
      "epoch": 1.2594649167087328,
      "grad_norm": 0.47735491394996643,
      "learning_rate": 6.425453535676844e-06,
      "loss": 0.3859,
      "step": 27445
    },
    {
      "epoch": 1.2595108072139873,
      "grad_norm": 0.404693603515625,
      "learning_rate": 6.425218521221255e-06,
      "loss": 0.245,
      "step": 27446
    },
    {
      "epoch": 1.2595566977192418,
      "grad_norm": 0.45366817712783813,
      "learning_rate": 6.424983503338464e-06,
      "loss": 0.3298,
      "step": 27447
    },
    {
      "epoch": 1.2596025882244963,
      "grad_norm": 0.49920982122421265,
      "learning_rate": 6.424748482029036e-06,
      "loss": 0.4011,
      "step": 27448
    },
    {
      "epoch": 1.2596484787297508,
      "grad_norm": 0.4661264419555664,
      "learning_rate": 6.424513457293539e-06,
      "loss": 0.3911,
      "step": 27449
    },
    {
      "epoch": 1.2596943692350053,
      "grad_norm": 0.48897841572761536,
      "learning_rate": 6.424278429132537e-06,
      "loss": 0.4129,
      "step": 27450
    },
    {
      "epoch": 1.2597402597402598,
      "grad_norm": 0.46444171667099,
      "learning_rate": 6.424043397546595e-06,
      "loss": 0.3218,
      "step": 27451
    },
    {
      "epoch": 1.2597861502455143,
      "grad_norm": 0.5604879260063171,
      "learning_rate": 6.423808362536278e-06,
      "loss": 0.4183,
      "step": 27452
    },
    {
      "epoch": 1.2598320407507686,
      "grad_norm": 0.46222352981567383,
      "learning_rate": 6.4235733241021505e-06,
      "loss": 0.3919,
      "step": 27453
    },
    {
      "epoch": 1.259877931256023,
      "grad_norm": 0.4743099808692932,
      "learning_rate": 6.423338282244778e-06,
      "loss": 0.3528,
      "step": 27454
    },
    {
      "epoch": 1.2599238217612776,
      "grad_norm": 0.5122509002685547,
      "learning_rate": 6.423103236964727e-06,
      "loss": 0.396,
      "step": 27455
    },
    {
      "epoch": 1.259969712266532,
      "grad_norm": 0.4523339569568634,
      "learning_rate": 6.4228681882625634e-06,
      "loss": 0.3475,
      "step": 27456
    },
    {
      "epoch": 1.2600156027717866,
      "grad_norm": 0.4637714922428131,
      "learning_rate": 6.422633136138848e-06,
      "loss": 0.3559,
      "step": 27457
    },
    {
      "epoch": 1.2600614932770409,
      "grad_norm": 0.4727868139743805,
      "learning_rate": 6.422398080594151e-06,
      "loss": 0.3832,
      "step": 27458
    },
    {
      "epoch": 1.2601073837822954,
      "grad_norm": 0.47968143224716187,
      "learning_rate": 6.422163021629035e-06,
      "loss": 0.3862,
      "step": 27459
    },
    {
      "epoch": 1.2601532742875499,
      "grad_norm": 0.49586865305900574,
      "learning_rate": 6.4219279592440655e-06,
      "loss": 0.4087,
      "step": 27460
    },
    {
      "epoch": 1.2601991647928044,
      "grad_norm": 0.43487364053726196,
      "learning_rate": 6.4216928934398085e-06,
      "loss": 0.2956,
      "step": 27461
    },
    {
      "epoch": 1.2602450552980589,
      "grad_norm": 0.4320218861103058,
      "learning_rate": 6.421457824216828e-06,
      "loss": 0.3053,
      "step": 27462
    },
    {
      "epoch": 1.2602909458033134,
      "grad_norm": 0.44533586502075195,
      "learning_rate": 6.421222751575691e-06,
      "loss": 0.3244,
      "step": 27463
    },
    {
      "epoch": 1.2603368363085679,
      "grad_norm": 0.45658960938453674,
      "learning_rate": 6.420987675516961e-06,
      "loss": 0.3459,
      "step": 27464
    },
    {
      "epoch": 1.2603827268138224,
      "grad_norm": 0.4522017240524292,
      "learning_rate": 6.420752596041204e-06,
      "loss": 0.3183,
      "step": 27465
    },
    {
      "epoch": 1.2604286173190766,
      "grad_norm": 0.54719078540802,
      "learning_rate": 6.420517513148986e-06,
      "loss": 0.3792,
      "step": 27466
    },
    {
      "epoch": 1.2604745078243311,
      "grad_norm": 0.4125523865222931,
      "learning_rate": 6.420282426840872e-06,
      "loss": 0.2948,
      "step": 27467
    },
    {
      "epoch": 1.2605203983295856,
      "grad_norm": 0.46355879306793213,
      "learning_rate": 6.420047337117426e-06,
      "loss": 0.3354,
      "step": 27468
    },
    {
      "epoch": 1.2605662888348401,
      "grad_norm": 0.4678686261177063,
      "learning_rate": 6.419812243979215e-06,
      "loss": 0.3833,
      "step": 27469
    },
    {
      "epoch": 1.2606121793400944,
      "grad_norm": 0.5016003251075745,
      "learning_rate": 6.419577147426804e-06,
      "loss": 0.439,
      "step": 27470
    },
    {
      "epoch": 1.260658069845349,
      "grad_norm": 0.4770630896091461,
      "learning_rate": 6.419342047460757e-06,
      "loss": 0.3956,
      "step": 27471
    },
    {
      "epoch": 1.2607039603506034,
      "grad_norm": 0.42309603095054626,
      "learning_rate": 6.419106944081641e-06,
      "loss": 0.2696,
      "step": 27472
    },
    {
      "epoch": 1.260749850855858,
      "grad_norm": 0.4692918062210083,
      "learning_rate": 6.418871837290021e-06,
      "loss": 0.3546,
      "step": 27473
    },
    {
      "epoch": 1.2607957413611124,
      "grad_norm": 0.45896798372268677,
      "learning_rate": 6.4186367270864615e-06,
      "loss": 0.3419,
      "step": 27474
    },
    {
      "epoch": 1.2608416318663669,
      "grad_norm": 0.45902541279792786,
      "learning_rate": 6.418401613471529e-06,
      "loss": 0.3337,
      "step": 27475
    },
    {
      "epoch": 1.2608875223716214,
      "grad_norm": 0.4546019732952118,
      "learning_rate": 6.418166496445788e-06,
      "loss": 0.3661,
      "step": 27476
    },
    {
      "epoch": 1.2609334128768759,
      "grad_norm": 0.467183917760849,
      "learning_rate": 6.417931376009804e-06,
      "loss": 0.3613,
      "step": 27477
    },
    {
      "epoch": 1.2609793033821302,
      "grad_norm": 0.46778103709220886,
      "learning_rate": 6.417696252164143e-06,
      "loss": 0.3508,
      "step": 27478
    },
    {
      "epoch": 1.2610251938873847,
      "grad_norm": 0.4331210255622864,
      "learning_rate": 6.417461124909369e-06,
      "loss": 0.3027,
      "step": 27479
    },
    {
      "epoch": 1.2610710843926392,
      "grad_norm": 0.43157076835632324,
      "learning_rate": 6.417225994246049e-06,
      "loss": 0.2937,
      "step": 27480
    },
    {
      "epoch": 1.2611169748978937,
      "grad_norm": 0.5246948003768921,
      "learning_rate": 6.416990860174748e-06,
      "loss": 0.4274,
      "step": 27481
    },
    {
      "epoch": 1.2611628654031481,
      "grad_norm": 0.4471305310726166,
      "learning_rate": 6.416755722696028e-06,
      "loss": 0.3353,
      "step": 27482
    },
    {
      "epoch": 1.2612087559084024,
      "grad_norm": 0.4778052270412445,
      "learning_rate": 6.416520581810463e-06,
      "loss": 0.3411,
      "step": 27483
    },
    {
      "epoch": 1.261254646413657,
      "grad_norm": 0.4761069118976593,
      "learning_rate": 6.41628543751861e-06,
      "loss": 0.3694,
      "step": 27484
    },
    {
      "epoch": 1.2613005369189114,
      "grad_norm": 0.4455127418041229,
      "learning_rate": 6.416050289821037e-06,
      "loss": 0.3049,
      "step": 27485
    },
    {
      "epoch": 1.261346427424166,
      "grad_norm": 0.44119274616241455,
      "learning_rate": 6.415815138718312e-06,
      "loss": 0.2802,
      "step": 27486
    },
    {
      "epoch": 1.2613923179294204,
      "grad_norm": 0.4375648498535156,
      "learning_rate": 6.415579984210997e-06,
      "loss": 0.3163,
      "step": 27487
    },
    {
      "epoch": 1.261438208434675,
      "grad_norm": 0.4907718598842621,
      "learning_rate": 6.41534482629966e-06,
      "loss": 0.423,
      "step": 27488
    },
    {
      "epoch": 1.2614840989399294,
      "grad_norm": 0.4556412994861603,
      "learning_rate": 6.415109664984866e-06,
      "loss": 0.3281,
      "step": 27489
    },
    {
      "epoch": 1.261529989445184,
      "grad_norm": 0.45051687955856323,
      "learning_rate": 6.4148745002671775e-06,
      "loss": 0.3687,
      "step": 27490
    },
    {
      "epoch": 1.2615758799504382,
      "grad_norm": 0.4505678117275238,
      "learning_rate": 6.414639332147162e-06,
      "loss": 0.3196,
      "step": 27491
    },
    {
      "epoch": 1.2616217704556927,
      "grad_norm": 0.45384714007377625,
      "learning_rate": 6.414404160625388e-06,
      "loss": 0.3312,
      "step": 27492
    },
    {
      "epoch": 1.2616676609609472,
      "grad_norm": 0.44400930404663086,
      "learning_rate": 6.414168985702416e-06,
      "loss": 0.3492,
      "step": 27493
    },
    {
      "epoch": 1.2617135514662017,
      "grad_norm": 0.4834831953048706,
      "learning_rate": 6.413933807378816e-06,
      "loss": 0.3598,
      "step": 27494
    },
    {
      "epoch": 1.2617594419714562,
      "grad_norm": 0.45370545983314514,
      "learning_rate": 6.413698625655152e-06,
      "loss": 0.3321,
      "step": 27495
    },
    {
      "epoch": 1.2618053324767105,
      "grad_norm": 0.47711634635925293,
      "learning_rate": 6.413463440531986e-06,
      "loss": 0.3715,
      "step": 27496
    },
    {
      "epoch": 1.261851222981965,
      "grad_norm": 0.45871588587760925,
      "learning_rate": 6.4132282520098885e-06,
      "loss": 0.3501,
      "step": 27497
    },
    {
      "epoch": 1.2618971134872194,
      "grad_norm": 0.46482929587364197,
      "learning_rate": 6.412993060089423e-06,
      "loss": 0.3847,
      "step": 27498
    },
    {
      "epoch": 1.261943003992474,
      "grad_norm": 0.46674227714538574,
      "learning_rate": 6.412757864771155e-06,
      "loss": 0.3972,
      "step": 27499
    },
    {
      "epoch": 1.2619888944977284,
      "grad_norm": 0.45041075348854065,
      "learning_rate": 6.412522666055651e-06,
      "loss": 0.3488,
      "step": 27500
    },
    {
      "epoch": 1.262034785002983,
      "grad_norm": 0.45454224944114685,
      "learning_rate": 6.412287463943474e-06,
      "loss": 0.3364,
      "step": 27501
    },
    {
      "epoch": 1.2620806755082374,
      "grad_norm": 0.4382898807525635,
      "learning_rate": 6.4120522584351905e-06,
      "loss": 0.3233,
      "step": 27502
    },
    {
      "epoch": 1.262126566013492,
      "grad_norm": 0.49850529432296753,
      "learning_rate": 6.4118170495313705e-06,
      "loss": 0.3666,
      "step": 27503
    },
    {
      "epoch": 1.2621724565187462,
      "grad_norm": 0.4846073389053345,
      "learning_rate": 6.411581837232573e-06,
      "loss": 0.3796,
      "step": 27504
    },
    {
      "epoch": 1.2622183470240007,
      "grad_norm": 0.49917495250701904,
      "learning_rate": 6.41134662153937e-06,
      "loss": 0.415,
      "step": 27505
    },
    {
      "epoch": 1.2622642375292552,
      "grad_norm": 0.4647367596626282,
      "learning_rate": 6.4111114024523215e-06,
      "loss": 0.385,
      "step": 27506
    },
    {
      "epoch": 1.2623101280345097,
      "grad_norm": 0.4864891469478607,
      "learning_rate": 6.410876179971994e-06,
      "loss": 0.3972,
      "step": 27507
    },
    {
      "epoch": 1.2623560185397642,
      "grad_norm": 0.45369046926498413,
      "learning_rate": 6.4106409540989565e-06,
      "loss": 0.3603,
      "step": 27508
    },
    {
      "epoch": 1.2624019090450185,
      "grad_norm": 0.4304215908050537,
      "learning_rate": 6.410405724833772e-06,
      "loss": 0.3234,
      "step": 27509
    },
    {
      "epoch": 1.262447799550273,
      "grad_norm": 0.4529515504837036,
      "learning_rate": 6.410170492177009e-06,
      "loss": 0.3892,
      "step": 27510
    },
    {
      "epoch": 1.2624936900555275,
      "grad_norm": 0.4793500304222107,
      "learning_rate": 6.409935256129229e-06,
      "loss": 0.3516,
      "step": 27511
    },
    {
      "epoch": 1.262539580560782,
      "grad_norm": 0.45193198323249817,
      "learning_rate": 6.409700016691001e-06,
      "loss": 0.3521,
      "step": 27512
    },
    {
      "epoch": 1.2625854710660365,
      "grad_norm": 0.45834222435951233,
      "learning_rate": 6.409464773862887e-06,
      "loss": 0.3346,
      "step": 27513
    },
    {
      "epoch": 1.262631361571291,
      "grad_norm": 0.5123468041419983,
      "learning_rate": 6.409229527645457e-06,
      "loss": 0.424,
      "step": 27514
    },
    {
      "epoch": 1.2626772520765455,
      "grad_norm": 0.4853701889514923,
      "learning_rate": 6.408994278039273e-06,
      "loss": 0.3899,
      "step": 27515
    },
    {
      "epoch": 1.2627231425818,
      "grad_norm": 0.48521438241004944,
      "learning_rate": 6.408759025044905e-06,
      "loss": 0.3705,
      "step": 27516
    },
    {
      "epoch": 1.2627690330870542,
      "grad_norm": 0.44487500190734863,
      "learning_rate": 6.4085237686629155e-06,
      "loss": 0.3103,
      "step": 27517
    },
    {
      "epoch": 1.2628149235923087,
      "grad_norm": 0.5089526772499084,
      "learning_rate": 6.408288508893869e-06,
      "loss": 0.3754,
      "step": 27518
    },
    {
      "epoch": 1.2628608140975632,
      "grad_norm": 0.48424121737480164,
      "learning_rate": 6.408053245738335e-06,
      "loss": 0.3972,
      "step": 27519
    },
    {
      "epoch": 1.2629067046028177,
      "grad_norm": 0.5489816665649414,
      "learning_rate": 6.407817979196877e-06,
      "loss": 0.3357,
      "step": 27520
    },
    {
      "epoch": 1.262952595108072,
      "grad_norm": 0.4366982579231262,
      "learning_rate": 6.407582709270061e-06,
      "loss": 0.32,
      "step": 27521
    },
    {
      "epoch": 1.2629984856133265,
      "grad_norm": 0.45980849862098694,
      "learning_rate": 6.407347435958452e-06,
      "loss": 0.3309,
      "step": 27522
    },
    {
      "epoch": 1.263044376118581,
      "grad_norm": 0.4334490895271301,
      "learning_rate": 6.407112159262618e-06,
      "loss": 0.292,
      "step": 27523
    },
    {
      "epoch": 1.2630902666238355,
      "grad_norm": 0.4605666697025299,
      "learning_rate": 6.406876879183122e-06,
      "loss": 0.351,
      "step": 27524
    },
    {
      "epoch": 1.26313615712909,
      "grad_norm": 0.4492349624633789,
      "learning_rate": 6.406641595720531e-06,
      "loss": 0.3104,
      "step": 27525
    },
    {
      "epoch": 1.2631820476343445,
      "grad_norm": 0.4943992793560028,
      "learning_rate": 6.4064063088754126e-06,
      "loss": 0.3845,
      "step": 27526
    },
    {
      "epoch": 1.263227938139599,
      "grad_norm": 0.4596658945083618,
      "learning_rate": 6.4061710186483295e-06,
      "loss": 0.3239,
      "step": 27527
    },
    {
      "epoch": 1.2632738286448535,
      "grad_norm": 0.4748929738998413,
      "learning_rate": 6.40593572503985e-06,
      "loss": 0.3647,
      "step": 27528
    },
    {
      "epoch": 1.2633197191501078,
      "grad_norm": 0.4443962275981903,
      "learning_rate": 6.405700428050537e-06,
      "loss": 0.3043,
      "step": 27529
    },
    {
      "epoch": 1.2633656096553623,
      "grad_norm": 0.46790122985839844,
      "learning_rate": 6.405465127680959e-06,
      "loss": 0.3737,
      "step": 27530
    },
    {
      "epoch": 1.2634115001606168,
      "grad_norm": 0.4724853038787842,
      "learning_rate": 6.4052298239316824e-06,
      "loss": 0.4104,
      "step": 27531
    },
    {
      "epoch": 1.2634573906658713,
      "grad_norm": 0.4438176155090332,
      "learning_rate": 6.40499451680327e-06,
      "loss": 0.3637,
      "step": 27532
    },
    {
      "epoch": 1.2635032811711258,
      "grad_norm": 0.4821753203868866,
      "learning_rate": 6.404759206296289e-06,
      "loss": 0.3557,
      "step": 27533
    },
    {
      "epoch": 1.26354917167638,
      "grad_norm": 0.4461735785007477,
      "learning_rate": 6.4045238924113074e-06,
      "loss": 0.3095,
      "step": 27534
    },
    {
      "epoch": 1.2635950621816345,
      "grad_norm": 0.4672586917877197,
      "learning_rate": 6.404288575148887e-06,
      "loss": 0.3472,
      "step": 27535
    },
    {
      "epoch": 1.263640952686889,
      "grad_norm": 0.44491419196128845,
      "learning_rate": 6.404053254509596e-06,
      "loss": 0.3221,
      "step": 27536
    },
    {
      "epoch": 1.2636868431921435,
      "grad_norm": 0.4458046853542328,
      "learning_rate": 6.403817930494001e-06,
      "loss": 0.2938,
      "step": 27537
    },
    {
      "epoch": 1.263732733697398,
      "grad_norm": 0.4586712121963501,
      "learning_rate": 6.403582603102667e-06,
      "loss": 0.3624,
      "step": 27538
    },
    {
      "epoch": 1.2637786242026525,
      "grad_norm": 0.4363486170768738,
      "learning_rate": 6.403347272336159e-06,
      "loss": 0.3172,
      "step": 27539
    },
    {
      "epoch": 1.263824514707907,
      "grad_norm": 0.4918195903301239,
      "learning_rate": 6.403111938195044e-06,
      "loss": 0.391,
      "step": 27540
    },
    {
      "epoch": 1.2638704052131615,
      "grad_norm": 0.43595102429389954,
      "learning_rate": 6.402876600679888e-06,
      "loss": 0.3302,
      "step": 27541
    },
    {
      "epoch": 1.2639162957184158,
      "grad_norm": 0.545981228351593,
      "learning_rate": 6.402641259791256e-06,
      "loss": 0.4357,
      "step": 27542
    },
    {
      "epoch": 1.2639621862236703,
      "grad_norm": 0.4933203458786011,
      "learning_rate": 6.402405915529716e-06,
      "loss": 0.4018,
      "step": 27543
    },
    {
      "epoch": 1.2640080767289248,
      "grad_norm": 0.4720374047756195,
      "learning_rate": 6.4021705678958305e-06,
      "loss": 0.3914,
      "step": 27544
    },
    {
      "epoch": 1.2640539672341793,
      "grad_norm": 0.5160587430000305,
      "learning_rate": 6.401935216890168e-06,
      "loss": 0.4113,
      "step": 27545
    },
    {
      "epoch": 1.2640998577394338,
      "grad_norm": 0.49615901708602905,
      "learning_rate": 6.401699862513293e-06,
      "loss": 0.4063,
      "step": 27546
    },
    {
      "epoch": 1.264145748244688,
      "grad_norm": 0.46220389008522034,
      "learning_rate": 6.401464504765772e-06,
      "loss": 0.3632,
      "step": 27547
    },
    {
      "epoch": 1.2641916387499426,
      "grad_norm": 0.45945844054222107,
      "learning_rate": 6.4012291436481735e-06,
      "loss": 0.3449,
      "step": 27548
    },
    {
      "epoch": 1.264237529255197,
      "grad_norm": 0.4806166887283325,
      "learning_rate": 6.40099377916106e-06,
      "loss": 0.4271,
      "step": 27549
    },
    {
      "epoch": 1.2642834197604516,
      "grad_norm": 0.4782956540584564,
      "learning_rate": 6.400758411304997e-06,
      "loss": 0.3545,
      "step": 27550
    },
    {
      "epoch": 1.264329310265706,
      "grad_norm": 0.45929577946662903,
      "learning_rate": 6.400523040080554e-06,
      "loss": 0.3405,
      "step": 27551
    },
    {
      "epoch": 1.2643752007709606,
      "grad_norm": 0.5132370591163635,
      "learning_rate": 6.400287665488294e-06,
      "loss": 0.367,
      "step": 27552
    },
    {
      "epoch": 1.264421091276215,
      "grad_norm": 0.44626620411872864,
      "learning_rate": 6.400052287528785e-06,
      "loss": 0.2948,
      "step": 27553
    },
    {
      "epoch": 1.2644669817814695,
      "grad_norm": 0.4999266266822815,
      "learning_rate": 6.399816906202593e-06,
      "loss": 0.4058,
      "step": 27554
    },
    {
      "epoch": 1.2645128722867238,
      "grad_norm": 0.4727477729320526,
      "learning_rate": 6.3995815215102805e-06,
      "loss": 0.3277,
      "step": 27555
    },
    {
      "epoch": 1.2645587627919783,
      "grad_norm": 0.5031111240386963,
      "learning_rate": 6.3993461334524175e-06,
      "loss": 0.3776,
      "step": 27556
    },
    {
      "epoch": 1.2646046532972328,
      "grad_norm": 0.41710031032562256,
      "learning_rate": 6.3991107420295696e-06,
      "loss": 0.2401,
      "step": 27557
    },
    {
      "epoch": 1.2646505438024873,
      "grad_norm": 0.4364396035671234,
      "learning_rate": 6.3988753472423e-06,
      "loss": 0.3025,
      "step": 27558
    },
    {
      "epoch": 1.2646964343077416,
      "grad_norm": 0.4538706839084625,
      "learning_rate": 6.398639949091179e-06,
      "loss": 0.3334,
      "step": 27559
    },
    {
      "epoch": 1.264742324812996,
      "grad_norm": 0.47165536880493164,
      "learning_rate": 6.3984045475767685e-06,
      "loss": 0.3203,
      "step": 27560
    },
    {
      "epoch": 1.2647882153182506,
      "grad_norm": 0.4871077537536621,
      "learning_rate": 6.3981691426996374e-06,
      "loss": 0.2973,
      "step": 27561
    },
    {
      "epoch": 1.264834105823505,
      "grad_norm": 0.4624306559562683,
      "learning_rate": 6.397933734460351e-06,
      "loss": 0.3361,
      "step": 27562
    },
    {
      "epoch": 1.2648799963287596,
      "grad_norm": 0.47159844636917114,
      "learning_rate": 6.3976983228594735e-06,
      "loss": 0.3408,
      "step": 27563
    },
    {
      "epoch": 1.264925886834014,
      "grad_norm": 0.5319923162460327,
      "learning_rate": 6.397462907897575e-06,
      "loss": 0.4742,
      "step": 27564
    },
    {
      "epoch": 1.2649717773392686,
      "grad_norm": 0.48090559244155884,
      "learning_rate": 6.397227489575219e-06,
      "loss": 0.3832,
      "step": 27565
    },
    {
      "epoch": 1.265017667844523,
      "grad_norm": 0.5140282511711121,
      "learning_rate": 6.39699206789297e-06,
      "loss": 0.4137,
      "step": 27566
    },
    {
      "epoch": 1.2650635583497774,
      "grad_norm": 0.4645368754863739,
      "learning_rate": 6.396756642851396e-06,
      "loss": 0.3617,
      "step": 27567
    },
    {
      "epoch": 1.2651094488550318,
      "grad_norm": 0.4952695369720459,
      "learning_rate": 6.396521214451064e-06,
      "loss": 0.3922,
      "step": 27568
    },
    {
      "epoch": 1.2651553393602863,
      "grad_norm": 0.4704573154449463,
      "learning_rate": 6.39628578269254e-06,
      "loss": 0.3553,
      "step": 27569
    },
    {
      "epoch": 1.2652012298655408,
      "grad_norm": 0.42104077339172363,
      "learning_rate": 6.396050347576388e-06,
      "loss": 0.267,
      "step": 27570
    },
    {
      "epoch": 1.2652471203707953,
      "grad_norm": 0.4680958688259125,
      "learning_rate": 6.3958149091031775e-06,
      "loss": 0.3795,
      "step": 27571
    },
    {
      "epoch": 1.2652930108760496,
      "grad_norm": 0.47072237730026245,
      "learning_rate": 6.395579467273469e-06,
      "loss": 0.3296,
      "step": 27572
    },
    {
      "epoch": 1.2653389013813041,
      "grad_norm": 0.4713965654373169,
      "learning_rate": 6.395344022087835e-06,
      "loss": 0.3496,
      "step": 27573
    },
    {
      "epoch": 1.2653847918865586,
      "grad_norm": 0.45085129141807556,
      "learning_rate": 6.39510857354684e-06,
      "loss": 0.3299,
      "step": 27574
    },
    {
      "epoch": 1.265430682391813,
      "grad_norm": 0.5161347985267639,
      "learning_rate": 6.394873121651047e-06,
      "loss": 0.3893,
      "step": 27575
    },
    {
      "epoch": 1.2654765728970676,
      "grad_norm": 0.41738277673721313,
      "learning_rate": 6.394637666401025e-06,
      "loss": 0.2939,
      "step": 27576
    },
    {
      "epoch": 1.265522463402322,
      "grad_norm": 0.44335660338401794,
      "learning_rate": 6.394402207797341e-06,
      "loss": 0.2906,
      "step": 27577
    },
    {
      "epoch": 1.2655683539075766,
      "grad_norm": 0.5002272725105286,
      "learning_rate": 6.394166745840558e-06,
      "loss": 0.3658,
      "step": 27578
    },
    {
      "epoch": 1.265614244412831,
      "grad_norm": 0.47426778078079224,
      "learning_rate": 6.3939312805312445e-06,
      "loss": 0.3464,
      "step": 27579
    },
    {
      "epoch": 1.2656601349180854,
      "grad_norm": 0.47478267550468445,
      "learning_rate": 6.393695811869965e-06,
      "loss": 0.3656,
      "step": 27580
    },
    {
      "epoch": 1.2657060254233399,
      "grad_norm": 0.5770941376686096,
      "learning_rate": 6.393460339857289e-06,
      "loss": 0.5207,
      "step": 27581
    },
    {
      "epoch": 1.2657519159285944,
      "grad_norm": 0.4493562877178192,
      "learning_rate": 6.393224864493781e-06,
      "loss": 0.3333,
      "step": 27582
    },
    {
      "epoch": 1.2657978064338489,
      "grad_norm": 0.43955740332603455,
      "learning_rate": 6.392989385780005e-06,
      "loss": 0.3025,
      "step": 27583
    },
    {
      "epoch": 1.2658436969391034,
      "grad_norm": 0.4719174802303314,
      "learning_rate": 6.392753903716529e-06,
      "loss": 0.3893,
      "step": 27584
    },
    {
      "epoch": 1.2658895874443576,
      "grad_norm": 0.4536304771900177,
      "learning_rate": 6.392518418303921e-06,
      "loss": 0.3378,
      "step": 27585
    },
    {
      "epoch": 1.2659354779496121,
      "grad_norm": 0.46752119064331055,
      "learning_rate": 6.392282929542746e-06,
      "loss": 0.3754,
      "step": 27586
    },
    {
      "epoch": 1.2659813684548666,
      "grad_norm": 0.45089420676231384,
      "learning_rate": 6.392047437433568e-06,
      "loss": 0.2966,
      "step": 27587
    },
    {
      "epoch": 1.2660272589601211,
      "grad_norm": 0.48386380076408386,
      "learning_rate": 6.391811941976957e-06,
      "loss": 0.3946,
      "step": 27588
    },
    {
      "epoch": 1.2660731494653756,
      "grad_norm": 0.4909362494945526,
      "learning_rate": 6.391576443173476e-06,
      "loss": 0.3895,
      "step": 27589
    },
    {
      "epoch": 1.2661190399706301,
      "grad_norm": 0.45384734869003296,
      "learning_rate": 6.3913409410236935e-06,
      "loss": 0.3374,
      "step": 27590
    },
    {
      "epoch": 1.2661649304758846,
      "grad_norm": 0.4328356683254242,
      "learning_rate": 6.391105435528175e-06,
      "loss": 0.3194,
      "step": 27591
    },
    {
      "epoch": 1.2662108209811391,
      "grad_norm": 0.5189058780670166,
      "learning_rate": 6.390869926687489e-06,
      "loss": 0.4336,
      "step": 27592
    },
    {
      "epoch": 1.2662567114863934,
      "grad_norm": 0.4861741364002228,
      "learning_rate": 6.390634414502198e-06,
      "loss": 0.4137,
      "step": 27593
    },
    {
      "epoch": 1.266302601991648,
      "grad_norm": 0.4722899794578552,
      "learning_rate": 6.39039889897287e-06,
      "loss": 0.3337,
      "step": 27594
    },
    {
      "epoch": 1.2663484924969024,
      "grad_norm": 0.424702525138855,
      "learning_rate": 6.390163380100071e-06,
      "loss": 0.2705,
      "step": 27595
    },
    {
      "epoch": 1.266394383002157,
      "grad_norm": 0.4307391345500946,
      "learning_rate": 6.3899278578843695e-06,
      "loss": 0.293,
      "step": 27596
    },
    {
      "epoch": 1.2664402735074114,
      "grad_norm": 0.45151200890541077,
      "learning_rate": 6.389692332326329e-06,
      "loss": 0.3171,
      "step": 27597
    },
    {
      "epoch": 1.2664861640126657,
      "grad_norm": 0.504805326461792,
      "learning_rate": 6.389456803426517e-06,
      "loss": 0.4074,
      "step": 27598
    },
    {
      "epoch": 1.2665320545179202,
      "grad_norm": 0.5217872262001038,
      "learning_rate": 6.3892212711855006e-06,
      "loss": 0.4443,
      "step": 27599
    },
    {
      "epoch": 1.2665779450231747,
      "grad_norm": 0.42626968026161194,
      "learning_rate": 6.3889857356038445e-06,
      "loss": 0.3009,
      "step": 27600
    },
    {
      "epoch": 1.2666238355284292,
      "grad_norm": 0.47009164094924927,
      "learning_rate": 6.388750196682116e-06,
      "loss": 0.3473,
      "step": 27601
    },
    {
      "epoch": 1.2666697260336837,
      "grad_norm": 0.46404513716697693,
      "learning_rate": 6.388514654420883e-06,
      "loss": 0.3647,
      "step": 27602
    },
    {
      "epoch": 1.2667156165389382,
      "grad_norm": 0.4919244050979614,
      "learning_rate": 6.38827910882071e-06,
      "loss": 0.3633,
      "step": 27603
    },
    {
      "epoch": 1.2667615070441927,
      "grad_norm": 0.474376916885376,
      "learning_rate": 6.3880435598821634e-06,
      "loss": 0.3567,
      "step": 27604
    },
    {
      "epoch": 1.2668073975494472,
      "grad_norm": 0.488635778427124,
      "learning_rate": 6.38780800760581e-06,
      "loss": 0.4396,
      "step": 27605
    },
    {
      "epoch": 1.2668532880547014,
      "grad_norm": 0.5359803438186646,
      "learning_rate": 6.387572451992217e-06,
      "loss": 0.4788,
      "step": 27606
    },
    {
      "epoch": 1.266899178559956,
      "grad_norm": 0.5098769068717957,
      "learning_rate": 6.387336893041949e-06,
      "loss": 0.4241,
      "step": 27607
    },
    {
      "epoch": 1.2669450690652104,
      "grad_norm": 0.45125171542167664,
      "learning_rate": 6.387101330755576e-06,
      "loss": 0.3201,
      "step": 27608
    },
    {
      "epoch": 1.266990959570465,
      "grad_norm": 0.5050342679023743,
      "learning_rate": 6.386865765133659e-06,
      "loss": 0.3959,
      "step": 27609
    },
    {
      "epoch": 1.2670368500757192,
      "grad_norm": 0.47305357456207275,
      "learning_rate": 6.3866301961767705e-06,
      "loss": 0.4083,
      "step": 27610
    },
    {
      "epoch": 1.2670827405809737,
      "grad_norm": 0.438044011592865,
      "learning_rate": 6.386394623885472e-06,
      "loss": 0.3202,
      "step": 27611
    },
    {
      "epoch": 1.2671286310862282,
      "grad_norm": 0.46742984652519226,
      "learning_rate": 6.386159048260332e-06,
      "loss": 0.3689,
      "step": 27612
    },
    {
      "epoch": 1.2671745215914827,
      "grad_norm": 0.4527471363544464,
      "learning_rate": 6.385923469301918e-06,
      "loss": 0.3171,
      "step": 27613
    },
    {
      "epoch": 1.2672204120967372,
      "grad_norm": 0.4945203959941864,
      "learning_rate": 6.385687887010797e-06,
      "loss": 0.3693,
      "step": 27614
    },
    {
      "epoch": 1.2672663026019917,
      "grad_norm": 0.5503880381584167,
      "learning_rate": 6.38545230138753e-06,
      "loss": 0.396,
      "step": 27615
    },
    {
      "epoch": 1.2673121931072462,
      "grad_norm": 0.46545058488845825,
      "learning_rate": 6.385216712432691e-06,
      "loss": 0.3269,
      "step": 27616
    },
    {
      "epoch": 1.2673580836125007,
      "grad_norm": 0.4329378008842468,
      "learning_rate": 6.384981120146841e-06,
      "loss": 0.2852,
      "step": 27617
    },
    {
      "epoch": 1.267403974117755,
      "grad_norm": 0.4398607909679413,
      "learning_rate": 6.384745524530547e-06,
      "loss": 0.3364,
      "step": 27618
    },
    {
      "epoch": 1.2674498646230095,
      "grad_norm": 0.4571370780467987,
      "learning_rate": 6.3845099255843815e-06,
      "loss": 0.3314,
      "step": 27619
    },
    {
      "epoch": 1.267495755128264,
      "grad_norm": 0.4782978594303131,
      "learning_rate": 6.384274323308903e-06,
      "loss": 0.3351,
      "step": 27620
    },
    {
      "epoch": 1.2675416456335185,
      "grad_norm": 0.435498982667923,
      "learning_rate": 6.384038717704682e-06,
      "loss": 0.3132,
      "step": 27621
    },
    {
      "epoch": 1.267587536138773,
      "grad_norm": 0.460575133562088,
      "learning_rate": 6.383803108772286e-06,
      "loss": 0.3921,
      "step": 27622
    },
    {
      "epoch": 1.2676334266440272,
      "grad_norm": 0.47944214940071106,
      "learning_rate": 6.38356749651228e-06,
      "loss": 0.3586,
      "step": 27623
    },
    {
      "epoch": 1.2676793171492817,
      "grad_norm": 0.45531216263771057,
      "learning_rate": 6.383331880925231e-06,
      "loss": 0.3837,
      "step": 27624
    },
    {
      "epoch": 1.2677252076545362,
      "grad_norm": 0.4527263939380646,
      "learning_rate": 6.383096262011705e-06,
      "loss": 0.3522,
      "step": 27625
    },
    {
      "epoch": 1.2677710981597907,
      "grad_norm": 0.5104132890701294,
      "learning_rate": 6.382860639772268e-06,
      "loss": 0.3813,
      "step": 27626
    },
    {
      "epoch": 1.2678169886650452,
      "grad_norm": 0.46782177686691284,
      "learning_rate": 6.38262501420749e-06,
      "loss": 0.3899,
      "step": 27627
    },
    {
      "epoch": 1.2678628791702997,
      "grad_norm": 0.42873355746269226,
      "learning_rate": 6.382389385317933e-06,
      "loss": 0.2763,
      "step": 27628
    },
    {
      "epoch": 1.2679087696755542,
      "grad_norm": 0.45529672503471375,
      "learning_rate": 6.3821537531041655e-06,
      "loss": 0.3372,
      "step": 27629
    },
    {
      "epoch": 1.2679546601808087,
      "grad_norm": 0.528867781162262,
      "learning_rate": 6.381918117566758e-06,
      "loss": 0.4643,
      "step": 27630
    },
    {
      "epoch": 1.268000550686063,
      "grad_norm": 0.46760040521621704,
      "learning_rate": 6.38168247870627e-06,
      "loss": 0.3534,
      "step": 27631
    },
    {
      "epoch": 1.2680464411913175,
      "grad_norm": 0.4564807116985321,
      "learning_rate": 6.381446836523273e-06,
      "loss": 0.3144,
      "step": 27632
    },
    {
      "epoch": 1.268092331696572,
      "grad_norm": 0.529423177242279,
      "learning_rate": 6.381211191018333e-06,
      "loss": 0.4665,
      "step": 27633
    },
    {
      "epoch": 1.2681382222018265,
      "grad_norm": 0.49141743779182434,
      "learning_rate": 6.380975542192014e-06,
      "loss": 0.3926,
      "step": 27634
    },
    {
      "epoch": 1.268184112707081,
      "grad_norm": 0.46112850308418274,
      "learning_rate": 6.380739890044887e-06,
      "loss": 0.3651,
      "step": 27635
    },
    {
      "epoch": 1.2682300032123353,
      "grad_norm": 0.49295279383659363,
      "learning_rate": 6.380504234577515e-06,
      "loss": 0.4101,
      "step": 27636
    },
    {
      "epoch": 1.2682758937175898,
      "grad_norm": 0.45224055647850037,
      "learning_rate": 6.380268575790466e-06,
      "loss": 0.318,
      "step": 27637
    },
    {
      "epoch": 1.2683217842228443,
      "grad_norm": 0.44189614057540894,
      "learning_rate": 6.380032913684306e-06,
      "loss": 0.3637,
      "step": 27638
    },
    {
      "epoch": 1.2683676747280987,
      "grad_norm": 0.46932274103164673,
      "learning_rate": 6.379797248259603e-06,
      "loss": 0.3646,
      "step": 27639
    },
    {
      "epoch": 1.2684135652333532,
      "grad_norm": 0.44524338841438293,
      "learning_rate": 6.379561579516924e-06,
      "loss": 0.2983,
      "step": 27640
    },
    {
      "epoch": 1.2684594557386077,
      "grad_norm": 0.4686581492424011,
      "learning_rate": 6.379325907456835e-06,
      "loss": 0.3442,
      "step": 27641
    },
    {
      "epoch": 1.2685053462438622,
      "grad_norm": 0.4844357669353485,
      "learning_rate": 6.379090232079902e-06,
      "loss": 0.3847,
      "step": 27642
    },
    {
      "epoch": 1.2685512367491167,
      "grad_norm": 0.4317905306816101,
      "learning_rate": 6.378854553386691e-06,
      "loss": 0.2788,
      "step": 27643
    },
    {
      "epoch": 1.268597127254371,
      "grad_norm": 0.464546799659729,
      "learning_rate": 6.378618871377772e-06,
      "loss": 0.3323,
      "step": 27644
    },
    {
      "epoch": 1.2686430177596255,
      "grad_norm": 0.4505314826965332,
      "learning_rate": 6.378383186053708e-06,
      "loss": 0.3371,
      "step": 27645
    },
    {
      "epoch": 1.26868890826488,
      "grad_norm": 0.45072025060653687,
      "learning_rate": 6.378147497415068e-06,
      "loss": 0.3383,
      "step": 27646
    },
    {
      "epoch": 1.2687347987701345,
      "grad_norm": 0.458962619304657,
      "learning_rate": 6.37791180546242e-06,
      "loss": 0.3508,
      "step": 27647
    },
    {
      "epoch": 1.2687806892753888,
      "grad_norm": 0.4548460841178894,
      "learning_rate": 6.377676110196327e-06,
      "loss": 0.3543,
      "step": 27648
    },
    {
      "epoch": 1.2688265797806433,
      "grad_norm": 0.45942509174346924,
      "learning_rate": 6.377440411617357e-06,
      "loss": 0.3352,
      "step": 27649
    },
    {
      "epoch": 1.2688724702858978,
      "grad_norm": 0.5117129683494568,
      "learning_rate": 6.377204709726079e-06,
      "loss": 0.4137,
      "step": 27650
    },
    {
      "epoch": 1.2689183607911523,
      "grad_norm": 0.46068593859672546,
      "learning_rate": 6.376969004523058e-06,
      "loss": 0.3288,
      "step": 27651
    },
    {
      "epoch": 1.2689642512964068,
      "grad_norm": 0.487785667181015,
      "learning_rate": 6.376733296008862e-06,
      "loss": 0.3864,
      "step": 27652
    },
    {
      "epoch": 1.2690101418016613,
      "grad_norm": 0.4759564697742462,
      "learning_rate": 6.376497584184056e-06,
      "loss": 0.3675,
      "step": 27653
    },
    {
      "epoch": 1.2690560323069158,
      "grad_norm": 0.47933080792427063,
      "learning_rate": 6.3762618690492074e-06,
      "loss": 0.4023,
      "step": 27654
    },
    {
      "epoch": 1.2691019228121703,
      "grad_norm": 0.4949531853199005,
      "learning_rate": 6.376026150604884e-06,
      "loss": 0.4638,
      "step": 27655
    },
    {
      "epoch": 1.2691478133174245,
      "grad_norm": 0.49106329679489136,
      "learning_rate": 6.375790428851651e-06,
      "loss": 0.3574,
      "step": 27656
    },
    {
      "epoch": 1.269193703822679,
      "grad_norm": 0.49616047739982605,
      "learning_rate": 6.375554703790078e-06,
      "loss": 0.3837,
      "step": 27657
    },
    {
      "epoch": 1.2692395943279335,
      "grad_norm": 0.44554466009140015,
      "learning_rate": 6.375318975420728e-06,
      "loss": 0.3303,
      "step": 27658
    },
    {
      "epoch": 1.269285484833188,
      "grad_norm": 0.4941883683204651,
      "learning_rate": 6.375083243744172e-06,
      "loss": 0.3913,
      "step": 27659
    },
    {
      "epoch": 1.2693313753384425,
      "grad_norm": 0.45048707723617554,
      "learning_rate": 6.374847508760973e-06,
      "loss": 0.3259,
      "step": 27660
    },
    {
      "epoch": 1.2693772658436968,
      "grad_norm": 0.49106863141059875,
      "learning_rate": 6.374611770471702e-06,
      "loss": 0.3863,
      "step": 27661
    },
    {
      "epoch": 1.2694231563489513,
      "grad_norm": 0.4536474347114563,
      "learning_rate": 6.374376028876921e-06,
      "loss": 0.3374,
      "step": 27662
    },
    {
      "epoch": 1.2694690468542058,
      "grad_norm": 0.4945515990257263,
      "learning_rate": 6.3741402839772e-06,
      "loss": 0.3618,
      "step": 27663
    },
    {
      "epoch": 1.2695149373594603,
      "grad_norm": 0.47918277978897095,
      "learning_rate": 6.3739045357731066e-06,
      "loss": 0.3556,
      "step": 27664
    },
    {
      "epoch": 1.2695608278647148,
      "grad_norm": 0.4663005769252777,
      "learning_rate": 6.373668784265206e-06,
      "loss": 0.3711,
      "step": 27665
    },
    {
      "epoch": 1.2696067183699693,
      "grad_norm": 0.5003321170806885,
      "learning_rate": 6.373433029454064e-06,
      "loss": 0.3801,
      "step": 27666
    },
    {
      "epoch": 1.2696526088752238,
      "grad_norm": 0.482926607131958,
      "learning_rate": 6.37319727134025e-06,
      "loss": 0.4158,
      "step": 27667
    },
    {
      "epoch": 1.2696984993804783,
      "grad_norm": 0.5018066763877869,
      "learning_rate": 6.372961509924331e-06,
      "loss": 0.4695,
      "step": 27668
    },
    {
      "epoch": 1.2697443898857326,
      "grad_norm": 0.43777206540107727,
      "learning_rate": 6.372725745206871e-06,
      "loss": 0.2976,
      "step": 27669
    },
    {
      "epoch": 1.269790280390987,
      "grad_norm": 0.42043137550354004,
      "learning_rate": 6.372489977188441e-06,
      "loss": 0.2756,
      "step": 27670
    },
    {
      "epoch": 1.2698361708962416,
      "grad_norm": 0.45214179158210754,
      "learning_rate": 6.372254205869604e-06,
      "loss": 0.3646,
      "step": 27671
    },
    {
      "epoch": 1.269882061401496,
      "grad_norm": 0.6788287162780762,
      "learning_rate": 6.372018431250928e-06,
      "loss": 0.3146,
      "step": 27672
    },
    {
      "epoch": 1.2699279519067506,
      "grad_norm": 0.4498259723186493,
      "learning_rate": 6.371782653332983e-06,
      "loss": 0.3244,
      "step": 27673
    },
    {
      "epoch": 1.2699738424120048,
      "grad_norm": 0.5554794669151306,
      "learning_rate": 6.371546872116332e-06,
      "loss": 0.3083,
      "step": 27674
    },
    {
      "epoch": 1.2700197329172593,
      "grad_norm": 0.5111182928085327,
      "learning_rate": 6.371311087601546e-06,
      "loss": 0.4078,
      "step": 27675
    },
    {
      "epoch": 1.2700656234225138,
      "grad_norm": 0.4748038947582245,
      "learning_rate": 6.371075299789187e-06,
      "loss": 0.3292,
      "step": 27676
    },
    {
      "epoch": 1.2701115139277683,
      "grad_norm": 0.44243666529655457,
      "learning_rate": 6.370839508679825e-06,
      "loss": 0.2967,
      "step": 27677
    },
    {
      "epoch": 1.2701574044330228,
      "grad_norm": 0.4877657890319824,
      "learning_rate": 6.370603714274027e-06,
      "loss": 0.3723,
      "step": 27678
    },
    {
      "epoch": 1.2702032949382773,
      "grad_norm": 0.5037795305252075,
      "learning_rate": 6.370367916572361e-06,
      "loss": 0.3559,
      "step": 27679
    },
    {
      "epoch": 1.2702491854435318,
      "grad_norm": 0.4582667648792267,
      "learning_rate": 6.37013211557539e-06,
      "loss": 0.3292,
      "step": 27680
    },
    {
      "epoch": 1.2702950759487863,
      "grad_norm": 0.4378514289855957,
      "learning_rate": 6.369896311283686e-06,
      "loss": 0.2793,
      "step": 27681
    },
    {
      "epoch": 1.2703409664540406,
      "grad_norm": 0.45867490768432617,
      "learning_rate": 6.369660503697812e-06,
      "loss": 0.317,
      "step": 27682
    },
    {
      "epoch": 1.270386856959295,
      "grad_norm": 0.5073931217193604,
      "learning_rate": 6.369424692818338e-06,
      "loss": 0.4601,
      "step": 27683
    },
    {
      "epoch": 1.2704327474645496,
      "grad_norm": 0.4697217345237732,
      "learning_rate": 6.36918887864583e-06,
      "loss": 0.3317,
      "step": 27684
    },
    {
      "epoch": 1.270478637969804,
      "grad_norm": 0.4507666528224945,
      "learning_rate": 6.3689530611808545e-06,
      "loss": 0.3363,
      "step": 27685
    },
    {
      "epoch": 1.2705245284750586,
      "grad_norm": 0.5221207141876221,
      "learning_rate": 6.3687172404239785e-06,
      "loss": 0.443,
      "step": 27686
    },
    {
      "epoch": 1.2705704189803129,
      "grad_norm": 0.5138861536979675,
      "learning_rate": 6.3684814163757705e-06,
      "loss": 0.4166,
      "step": 27687
    },
    {
      "epoch": 1.2706163094855674,
      "grad_norm": 0.4725380837917328,
      "learning_rate": 6.368245589036795e-06,
      "loss": 0.3798,
      "step": 27688
    },
    {
      "epoch": 1.2706621999908219,
      "grad_norm": 0.4867097735404968,
      "learning_rate": 6.368009758407622e-06,
      "loss": 0.3819,
      "step": 27689
    },
    {
      "epoch": 1.2707080904960764,
      "grad_norm": 0.4538317322731018,
      "learning_rate": 6.367773924488817e-06,
      "loss": 0.348,
      "step": 27690
    },
    {
      "epoch": 1.2707539810013309,
      "grad_norm": 0.48461124300956726,
      "learning_rate": 6.367538087280947e-06,
      "loss": 0.396,
      "step": 27691
    },
    {
      "epoch": 1.2707998715065854,
      "grad_norm": 0.4916684925556183,
      "learning_rate": 6.36730224678458e-06,
      "loss": 0.3963,
      "step": 27692
    },
    {
      "epoch": 1.2708457620118399,
      "grad_norm": 0.4852026402950287,
      "learning_rate": 6.367066403000282e-06,
      "loss": 0.3604,
      "step": 27693
    },
    {
      "epoch": 1.2708916525170944,
      "grad_norm": 0.4776019752025604,
      "learning_rate": 6.366830555928621e-06,
      "loss": 0.3816,
      "step": 27694
    },
    {
      "epoch": 1.2709375430223486,
      "grad_norm": 0.4557274281978607,
      "learning_rate": 6.366594705570164e-06,
      "loss": 0.3461,
      "step": 27695
    },
    {
      "epoch": 1.2709834335276031,
      "grad_norm": 0.5726621150970459,
      "learning_rate": 6.366358851925478e-06,
      "loss": 0.3206,
      "step": 27696
    },
    {
      "epoch": 1.2710293240328576,
      "grad_norm": 0.4164447486400604,
      "learning_rate": 6.366122994995131e-06,
      "loss": 0.2849,
      "step": 27697
    },
    {
      "epoch": 1.2710752145381121,
      "grad_norm": 0.4856354296207428,
      "learning_rate": 6.365887134779689e-06,
      "loss": 0.3535,
      "step": 27698
    },
    {
      "epoch": 1.2711211050433664,
      "grad_norm": 0.4723111093044281,
      "learning_rate": 6.365651271279718e-06,
      "loss": 0.3109,
      "step": 27699
    },
    {
      "epoch": 1.271166995548621,
      "grad_norm": 0.468791663646698,
      "learning_rate": 6.365415404495788e-06,
      "loss": 0.3558,
      "step": 27700
    },
    {
      "epoch": 1.2712128860538754,
      "grad_norm": 0.501873791217804,
      "learning_rate": 6.365179534428466e-06,
      "loss": 0.4088,
      "step": 27701
    },
    {
      "epoch": 1.2712587765591299,
      "grad_norm": 0.45098021626472473,
      "learning_rate": 6.364943661078316e-06,
      "loss": 0.3059,
      "step": 27702
    },
    {
      "epoch": 1.2713046670643844,
      "grad_norm": 0.4567374289035797,
      "learning_rate": 6.364707784445908e-06,
      "loss": 0.2963,
      "step": 27703
    },
    {
      "epoch": 1.2713505575696389,
      "grad_norm": 0.46257248520851135,
      "learning_rate": 6.364471904531809e-06,
      "loss": 0.374,
      "step": 27704
    },
    {
      "epoch": 1.2713964480748934,
      "grad_norm": 0.4433690011501312,
      "learning_rate": 6.364236021336585e-06,
      "loss": 0.3029,
      "step": 27705
    },
    {
      "epoch": 1.2714423385801479,
      "grad_norm": 0.4488614499568939,
      "learning_rate": 6.364000134860803e-06,
      "loss": 0.2932,
      "step": 27706
    },
    {
      "epoch": 1.2714882290854022,
      "grad_norm": 0.4548358619213104,
      "learning_rate": 6.363764245105034e-06,
      "loss": 0.321,
      "step": 27707
    },
    {
      "epoch": 1.2715341195906567,
      "grad_norm": 0.46632638573646545,
      "learning_rate": 6.363528352069841e-06,
      "loss": 0.3196,
      "step": 27708
    },
    {
      "epoch": 1.2715800100959112,
      "grad_norm": 0.47722724080085754,
      "learning_rate": 6.363292455755793e-06,
      "loss": 0.3911,
      "step": 27709
    },
    {
      "epoch": 1.2716259006011656,
      "grad_norm": 0.4412669241428375,
      "learning_rate": 6.363056556163456e-06,
      "loss": 0.3152,
      "step": 27710
    },
    {
      "epoch": 1.2716717911064201,
      "grad_norm": 0.5177837014198303,
      "learning_rate": 6.362820653293399e-06,
      "loss": 0.4264,
      "step": 27711
    },
    {
      "epoch": 1.2717176816116744,
      "grad_norm": 0.46601539850234985,
      "learning_rate": 6.362584747146187e-06,
      "loss": 0.3251,
      "step": 27712
    },
    {
      "epoch": 1.271763572116929,
      "grad_norm": 0.4942941963672638,
      "learning_rate": 6.3623488377223904e-06,
      "loss": 0.403,
      "step": 27713
    },
    {
      "epoch": 1.2718094626221834,
      "grad_norm": 0.4795243442058563,
      "learning_rate": 6.362112925022573e-06,
      "loss": 0.3908,
      "step": 27714
    },
    {
      "epoch": 1.271855353127438,
      "grad_norm": 0.45312339067459106,
      "learning_rate": 6.361877009047306e-06,
      "loss": 0.3743,
      "step": 27715
    },
    {
      "epoch": 1.2719012436326924,
      "grad_norm": 0.5065512657165527,
      "learning_rate": 6.361641089797153e-06,
      "loss": 0.3953,
      "step": 27716
    },
    {
      "epoch": 1.271947134137947,
      "grad_norm": 0.4847389757633209,
      "learning_rate": 6.361405167272685e-06,
      "loss": 0.3872,
      "step": 27717
    },
    {
      "epoch": 1.2719930246432014,
      "grad_norm": 0.4697864055633545,
      "learning_rate": 6.361169241474465e-06,
      "loss": 0.3662,
      "step": 27718
    },
    {
      "epoch": 1.272038915148456,
      "grad_norm": 0.48790040612220764,
      "learning_rate": 6.360933312403063e-06,
      "loss": 0.3693,
      "step": 27719
    },
    {
      "epoch": 1.2720848056537102,
      "grad_norm": 0.459744393825531,
      "learning_rate": 6.360697380059047e-06,
      "loss": 0.3517,
      "step": 27720
    },
    {
      "epoch": 1.2721306961589647,
      "grad_norm": 0.4446001648902893,
      "learning_rate": 6.360461444442983e-06,
      "loss": 0.3219,
      "step": 27721
    },
    {
      "epoch": 1.2721765866642192,
      "grad_norm": 0.44017964601516724,
      "learning_rate": 6.360225505555439e-06,
      "loss": 0.3367,
      "step": 27722
    },
    {
      "epoch": 1.2722224771694737,
      "grad_norm": 0.5778737664222717,
      "learning_rate": 6.35998956339698e-06,
      "loss": 0.428,
      "step": 27723
    },
    {
      "epoch": 1.2722683676747282,
      "grad_norm": 0.4325180947780609,
      "learning_rate": 6.359753617968176e-06,
      "loss": 0.2938,
      "step": 27724
    },
    {
      "epoch": 1.2723142581799824,
      "grad_norm": 0.48611775040626526,
      "learning_rate": 6.359517669269595e-06,
      "loss": 0.405,
      "step": 27725
    },
    {
      "epoch": 1.272360148685237,
      "grad_norm": 0.497148722410202,
      "learning_rate": 6.359281717301802e-06,
      "loss": 0.3991,
      "step": 27726
    },
    {
      "epoch": 1.2724060391904914,
      "grad_norm": 0.4622134268283844,
      "learning_rate": 6.359045762065366e-06,
      "loss": 0.3067,
      "step": 27727
    },
    {
      "epoch": 1.272451929695746,
      "grad_norm": 0.4684372544288635,
      "learning_rate": 6.358809803560854e-06,
      "loss": 0.3542,
      "step": 27728
    },
    {
      "epoch": 1.2724978202010004,
      "grad_norm": 0.45311111211776733,
      "learning_rate": 6.358573841788833e-06,
      "loss": 0.3228,
      "step": 27729
    },
    {
      "epoch": 1.272543710706255,
      "grad_norm": 0.43367260694503784,
      "learning_rate": 6.358337876749869e-06,
      "loss": 0.3027,
      "step": 27730
    },
    {
      "epoch": 1.2725896012115094,
      "grad_norm": 0.491629034280777,
      "learning_rate": 6.358101908444533e-06,
      "loss": 0.4069,
      "step": 27731
    },
    {
      "epoch": 1.272635491716764,
      "grad_norm": 0.4487422704696655,
      "learning_rate": 6.357865936873391e-06,
      "loss": 0.3202,
      "step": 27732
    },
    {
      "epoch": 1.2726813822220182,
      "grad_norm": 0.44172459840774536,
      "learning_rate": 6.357629962037009e-06,
      "loss": 0.3243,
      "step": 27733
    },
    {
      "epoch": 1.2727272727272727,
      "grad_norm": 0.49263107776641846,
      "learning_rate": 6.3573939839359555e-06,
      "loss": 0.4038,
      "step": 27734
    },
    {
      "epoch": 1.2727731632325272,
      "grad_norm": 0.5221304297447205,
      "learning_rate": 6.357158002570798e-06,
      "loss": 0.4229,
      "step": 27735
    },
    {
      "epoch": 1.2728190537377817,
      "grad_norm": 0.46377697587013245,
      "learning_rate": 6.3569220179421046e-06,
      "loss": 0.3518,
      "step": 27736
    },
    {
      "epoch": 1.272864944243036,
      "grad_norm": 0.4802818298339844,
      "learning_rate": 6.35668603005044e-06,
      "loss": 0.4259,
      "step": 27737
    },
    {
      "epoch": 1.2729108347482905,
      "grad_norm": 0.45367375016212463,
      "learning_rate": 6.3564500388963754e-06,
      "loss": 0.3282,
      "step": 27738
    },
    {
      "epoch": 1.272956725253545,
      "grad_norm": 0.5213709473609924,
      "learning_rate": 6.356214044480477e-06,
      "loss": 0.3703,
      "step": 27739
    },
    {
      "epoch": 1.2730026157587995,
      "grad_norm": 0.4445798695087433,
      "learning_rate": 6.355978046803309e-06,
      "loss": 0.3346,
      "step": 27740
    },
    {
      "epoch": 1.273048506264054,
      "grad_norm": 0.4821888506412506,
      "learning_rate": 6.355742045865445e-06,
      "loss": 0.4088,
      "step": 27741
    },
    {
      "epoch": 1.2730943967693085,
      "grad_norm": 0.4570859372615814,
      "learning_rate": 6.355506041667447e-06,
      "loss": 0.3456,
      "step": 27742
    },
    {
      "epoch": 1.273140287274563,
      "grad_norm": 0.46337783336639404,
      "learning_rate": 6.355270034209886e-06,
      "loss": 0.3239,
      "step": 27743
    },
    {
      "epoch": 1.2731861777798175,
      "grad_norm": 0.47391360998153687,
      "learning_rate": 6.355034023493329e-06,
      "loss": 0.3342,
      "step": 27744
    },
    {
      "epoch": 1.2732320682850717,
      "grad_norm": 0.49589842557907104,
      "learning_rate": 6.354798009518341e-06,
      "loss": 0.3654,
      "step": 27745
    },
    {
      "epoch": 1.2732779587903262,
      "grad_norm": 0.5324549078941345,
      "learning_rate": 6.354561992285493e-06,
      "loss": 0.462,
      "step": 27746
    },
    {
      "epoch": 1.2733238492955807,
      "grad_norm": 0.448321670293808,
      "learning_rate": 6.35432597179535e-06,
      "loss": 0.3379,
      "step": 27747
    },
    {
      "epoch": 1.2733697398008352,
      "grad_norm": 0.45970746874809265,
      "learning_rate": 6.354089948048479e-06,
      "loss": 0.3228,
      "step": 27748
    },
    {
      "epoch": 1.2734156303060897,
      "grad_norm": 0.4916815757751465,
      "learning_rate": 6.353853921045452e-06,
      "loss": 0.4162,
      "step": 27749
    },
    {
      "epoch": 1.273461520811344,
      "grad_norm": 0.46182844042778015,
      "learning_rate": 6.3536178907868325e-06,
      "loss": 0.3783,
      "step": 27750
    },
    {
      "epoch": 1.2735074113165985,
      "grad_norm": 0.48527708649635315,
      "learning_rate": 6.353381857273189e-06,
      "loss": 0.4172,
      "step": 27751
    },
    {
      "epoch": 1.273553301821853,
      "grad_norm": 0.4575256109237671,
      "learning_rate": 6.353145820505089e-06,
      "loss": 0.3263,
      "step": 27752
    },
    {
      "epoch": 1.2735991923271075,
      "grad_norm": 0.4684695899486542,
      "learning_rate": 6.352909780483101e-06,
      "loss": 0.3489,
      "step": 27753
    },
    {
      "epoch": 1.273645082832362,
      "grad_norm": 0.4623975157737732,
      "learning_rate": 6.352673737207792e-06,
      "loss": 0.3885,
      "step": 27754
    },
    {
      "epoch": 1.2736909733376165,
      "grad_norm": 0.44689875841140747,
      "learning_rate": 6.352437690679729e-06,
      "loss": 0.3394,
      "step": 27755
    },
    {
      "epoch": 1.273736863842871,
      "grad_norm": 0.45736265182495117,
      "learning_rate": 6.35220164089948e-06,
      "loss": 0.3562,
      "step": 27756
    },
    {
      "epoch": 1.2737827543481255,
      "grad_norm": 0.4494754672050476,
      "learning_rate": 6.3519655878676134e-06,
      "loss": 0.3323,
      "step": 27757
    },
    {
      "epoch": 1.2738286448533798,
      "grad_norm": 0.45911118388175964,
      "learning_rate": 6.351729531584696e-06,
      "loss": 0.3249,
      "step": 27758
    },
    {
      "epoch": 1.2738745353586343,
      "grad_norm": 0.46890345215797424,
      "learning_rate": 6.351493472051296e-06,
      "loss": 0.3387,
      "step": 27759
    },
    {
      "epoch": 1.2739204258638888,
      "grad_norm": 0.4978278577327728,
      "learning_rate": 6.3512574092679805e-06,
      "loss": 0.3754,
      "step": 27760
    },
    {
      "epoch": 1.2739663163691433,
      "grad_norm": 0.4443426728248596,
      "learning_rate": 6.3510213432353175e-06,
      "loss": 0.3067,
      "step": 27761
    },
    {
      "epoch": 1.2740122068743978,
      "grad_norm": 0.4407588541507721,
      "learning_rate": 6.350785273953875e-06,
      "loss": 0.3352,
      "step": 27762
    },
    {
      "epoch": 1.274058097379652,
      "grad_norm": 0.4598753750324249,
      "learning_rate": 6.350549201424218e-06,
      "loss": 0.324,
      "step": 27763
    },
    {
      "epoch": 1.2741039878849065,
      "grad_norm": 0.45335572957992554,
      "learning_rate": 6.350313125646919e-06,
      "loss": 0.3249,
      "step": 27764
    },
    {
      "epoch": 1.274149878390161,
      "grad_norm": 0.5272057056427002,
      "learning_rate": 6.3500770466225425e-06,
      "loss": 0.4214,
      "step": 27765
    },
    {
      "epoch": 1.2741957688954155,
      "grad_norm": 0.49180492758750916,
      "learning_rate": 6.3498409643516564e-06,
      "loss": 0.3947,
      "step": 27766
    },
    {
      "epoch": 1.27424165940067,
      "grad_norm": 0.4625456631183624,
      "learning_rate": 6.349604878834828e-06,
      "loss": 0.3705,
      "step": 27767
    },
    {
      "epoch": 1.2742875499059245,
      "grad_norm": 0.5673028230667114,
      "learning_rate": 6.3493687900726265e-06,
      "loss": 0.4642,
      "step": 27768
    },
    {
      "epoch": 1.274333440411179,
      "grad_norm": 0.4585072994232178,
      "learning_rate": 6.34913269806562e-06,
      "loss": 0.344,
      "step": 27769
    },
    {
      "epoch": 1.2743793309164335,
      "grad_norm": 0.5007902383804321,
      "learning_rate": 6.3488966028143735e-06,
      "loss": 0.3926,
      "step": 27770
    },
    {
      "epoch": 1.2744252214216878,
      "grad_norm": 0.4929814338684082,
      "learning_rate": 6.348660504319457e-06,
      "loss": 0.4062,
      "step": 27771
    },
    {
      "epoch": 1.2744711119269423,
      "grad_norm": 0.44114091992378235,
      "learning_rate": 6.348424402581439e-06,
      "loss": 0.2567,
      "step": 27772
    },
    {
      "epoch": 1.2745170024321968,
      "grad_norm": 0.49472641944885254,
      "learning_rate": 6.348188297600885e-06,
      "loss": 0.4029,
      "step": 27773
    },
    {
      "epoch": 1.2745628929374513,
      "grad_norm": 0.4959041476249695,
      "learning_rate": 6.347952189378364e-06,
      "loss": 0.3903,
      "step": 27774
    },
    {
      "epoch": 1.2746087834427058,
      "grad_norm": 0.4321146309375763,
      "learning_rate": 6.347716077914441e-06,
      "loss": 0.3046,
      "step": 27775
    },
    {
      "epoch": 1.27465467394796,
      "grad_norm": 0.5117290616035461,
      "learning_rate": 6.347479963209688e-06,
      "loss": 0.4144,
      "step": 27776
    },
    {
      "epoch": 1.2747005644532146,
      "grad_norm": 0.47882482409477234,
      "learning_rate": 6.347243845264672e-06,
      "loss": 0.3656,
      "step": 27777
    },
    {
      "epoch": 1.274746454958469,
      "grad_norm": 0.51138836145401,
      "learning_rate": 6.347007724079959e-06,
      "loss": 0.3872,
      "step": 27778
    },
    {
      "epoch": 1.2747923454637236,
      "grad_norm": 0.48766204714775085,
      "learning_rate": 6.346771599656117e-06,
      "loss": 0.3752,
      "step": 27779
    },
    {
      "epoch": 1.274838235968978,
      "grad_norm": 0.48240339756011963,
      "learning_rate": 6.346535471993714e-06,
      "loss": 0.4053,
      "step": 27780
    },
    {
      "epoch": 1.2748841264742325,
      "grad_norm": 0.4583371579647064,
      "learning_rate": 6.346299341093318e-06,
      "loss": 0.3163,
      "step": 27781
    },
    {
      "epoch": 1.274930016979487,
      "grad_norm": 0.44486430287361145,
      "learning_rate": 6.346063206955499e-06,
      "loss": 0.3162,
      "step": 27782
    },
    {
      "epoch": 1.2749759074847415,
      "grad_norm": 0.4537929892539978,
      "learning_rate": 6.345827069580821e-06,
      "loss": 0.3073,
      "step": 27783
    },
    {
      "epoch": 1.2750217979899958,
      "grad_norm": 0.5201525688171387,
      "learning_rate": 6.3455909289698536e-06,
      "loss": 0.4283,
      "step": 27784
    },
    {
      "epoch": 1.2750676884952503,
      "grad_norm": 0.4402295649051666,
      "learning_rate": 6.345354785123165e-06,
      "loss": 0.3323,
      "step": 27785
    },
    {
      "epoch": 1.2751135790005048,
      "grad_norm": 0.47450944781303406,
      "learning_rate": 6.345118638041323e-06,
      "loss": 0.3929,
      "step": 27786
    },
    {
      "epoch": 1.2751594695057593,
      "grad_norm": 0.45171263813972473,
      "learning_rate": 6.344882487724897e-06,
      "loss": 0.3066,
      "step": 27787
    },
    {
      "epoch": 1.2752053600110136,
      "grad_norm": 0.41587045788764954,
      "learning_rate": 6.344646334174451e-06,
      "loss": 0.2839,
      "step": 27788
    },
    {
      "epoch": 1.275251250516268,
      "grad_norm": 0.48728400468826294,
      "learning_rate": 6.344410177390556e-06,
      "loss": 0.4047,
      "step": 27789
    },
    {
      "epoch": 1.2752971410215226,
      "grad_norm": 0.5393722057342529,
      "learning_rate": 6.344174017373776e-06,
      "loss": 0.4912,
      "step": 27790
    },
    {
      "epoch": 1.275343031526777,
      "grad_norm": 0.4480230212211609,
      "learning_rate": 6.343937854124686e-06,
      "loss": 0.3259,
      "step": 27791
    },
    {
      "epoch": 1.2753889220320316,
      "grad_norm": 0.4706697165966034,
      "learning_rate": 6.3437016876438465e-06,
      "loss": 0.3792,
      "step": 27792
    },
    {
      "epoch": 1.275434812537286,
      "grad_norm": 0.4190223515033722,
      "learning_rate": 6.343465517931829e-06,
      "loss": 0.3036,
      "step": 27793
    },
    {
      "epoch": 1.2754807030425406,
      "grad_norm": 0.45606130361557007,
      "learning_rate": 6.343229344989203e-06,
      "loss": 0.3293,
      "step": 27794
    },
    {
      "epoch": 1.275526593547795,
      "grad_norm": 0.47468486428260803,
      "learning_rate": 6.342993168816531e-06,
      "loss": 0.34,
      "step": 27795
    },
    {
      "epoch": 1.2755724840530493,
      "grad_norm": 0.46183130145072937,
      "learning_rate": 6.342756989414386e-06,
      "loss": 0.3538,
      "step": 27796
    },
    {
      "epoch": 1.2756183745583038,
      "grad_norm": 0.48262646794319153,
      "learning_rate": 6.3425208067833345e-06,
      "loss": 0.3488,
      "step": 27797
    },
    {
      "epoch": 1.2756642650635583,
      "grad_norm": 0.48504653573036194,
      "learning_rate": 6.342284620923945e-06,
      "loss": 0.4129,
      "step": 27798
    },
    {
      "epoch": 1.2757101555688128,
      "grad_norm": 0.41733768582344055,
      "learning_rate": 6.342048431836783e-06,
      "loss": 0.2754,
      "step": 27799
    },
    {
      "epoch": 1.2757560460740673,
      "grad_norm": 0.4881349503993988,
      "learning_rate": 6.3418122395224195e-06,
      "loss": 0.3698,
      "step": 27800
    },
    {
      "epoch": 1.2758019365793216,
      "grad_norm": 0.4576893448829651,
      "learning_rate": 6.341576043981419e-06,
      "loss": 0.3205,
      "step": 27801
    },
    {
      "epoch": 1.2758478270845761,
      "grad_norm": 0.45796912908554077,
      "learning_rate": 6.341339845214353e-06,
      "loss": 0.3611,
      "step": 27802
    },
    {
      "epoch": 1.2758937175898306,
      "grad_norm": 0.4663476049900055,
      "learning_rate": 6.341103643221788e-06,
      "loss": 0.3323,
      "step": 27803
    },
    {
      "epoch": 1.275939608095085,
      "grad_norm": 0.500990092754364,
      "learning_rate": 6.340867438004292e-06,
      "loss": 0.4587,
      "step": 27804
    },
    {
      "epoch": 1.2759854986003396,
      "grad_norm": 0.4402237832546234,
      "learning_rate": 6.340631229562433e-06,
      "loss": 0.3257,
      "step": 27805
    },
    {
      "epoch": 1.276031389105594,
      "grad_norm": 0.4308221638202667,
      "learning_rate": 6.340395017896779e-06,
      "loss": 0.2634,
      "step": 27806
    },
    {
      "epoch": 1.2760772796108486,
      "grad_norm": 0.46913942694664,
      "learning_rate": 6.3401588030078965e-06,
      "loss": 0.3688,
      "step": 27807
    },
    {
      "epoch": 1.276123170116103,
      "grad_norm": 0.46698325872421265,
      "learning_rate": 6.339922584896357e-06,
      "loss": 0.3226,
      "step": 27808
    },
    {
      "epoch": 1.2761690606213574,
      "grad_norm": 0.4292261600494385,
      "learning_rate": 6.3396863635627274e-06,
      "loss": 0.2728,
      "step": 27809
    },
    {
      "epoch": 1.2762149511266119,
      "grad_norm": 0.471407413482666,
      "learning_rate": 6.339450139007572e-06,
      "loss": 0.3625,
      "step": 27810
    },
    {
      "epoch": 1.2762608416318664,
      "grad_norm": 0.4777388274669647,
      "learning_rate": 6.339213911231463e-06,
      "loss": 0.3614,
      "step": 27811
    },
    {
      "epoch": 1.2763067321371209,
      "grad_norm": 0.4551113247871399,
      "learning_rate": 6.338977680234968e-06,
      "loss": 0.3365,
      "step": 27812
    },
    {
      "epoch": 1.2763526226423754,
      "grad_norm": 0.4405660629272461,
      "learning_rate": 6.338741446018652e-06,
      "loss": 0.2739,
      "step": 27813
    },
    {
      "epoch": 1.2763985131476296,
      "grad_norm": 0.5292825698852539,
      "learning_rate": 6.338505208583088e-06,
      "loss": 0.3245,
      "step": 27814
    },
    {
      "epoch": 1.2764444036528841,
      "grad_norm": 0.4380699098110199,
      "learning_rate": 6.338268967928841e-06,
      "loss": 0.3076,
      "step": 27815
    },
    {
      "epoch": 1.2764902941581386,
      "grad_norm": 0.4487307071685791,
      "learning_rate": 6.338032724056479e-06,
      "loss": 0.2934,
      "step": 27816
    },
    {
      "epoch": 1.2765361846633931,
      "grad_norm": 0.44275882840156555,
      "learning_rate": 6.337796476966571e-06,
      "loss": 0.3099,
      "step": 27817
    },
    {
      "epoch": 1.2765820751686476,
      "grad_norm": 0.48330292105674744,
      "learning_rate": 6.337560226659683e-06,
      "loss": 0.3393,
      "step": 27818
    },
    {
      "epoch": 1.2766279656739021,
      "grad_norm": 0.46951282024383545,
      "learning_rate": 6.337323973136386e-06,
      "loss": 0.3486,
      "step": 27819
    },
    {
      "epoch": 1.2766738561791566,
      "grad_norm": 0.47778233885765076,
      "learning_rate": 6.337087716397248e-06,
      "loss": 0.3767,
      "step": 27820
    },
    {
      "epoch": 1.2767197466844111,
      "grad_norm": 0.4995887279510498,
      "learning_rate": 6.336851456442834e-06,
      "loss": 0.3803,
      "step": 27821
    },
    {
      "epoch": 1.2767656371896654,
      "grad_norm": 0.4737626612186432,
      "learning_rate": 6.336615193273716e-06,
      "loss": 0.3694,
      "step": 27822
    },
    {
      "epoch": 1.27681152769492,
      "grad_norm": 0.4725928008556366,
      "learning_rate": 6.336378926890459e-06,
      "loss": 0.3573,
      "step": 27823
    },
    {
      "epoch": 1.2768574182001744,
      "grad_norm": 0.41937023401260376,
      "learning_rate": 6.336142657293633e-06,
      "loss": 0.2877,
      "step": 27824
    },
    {
      "epoch": 1.276903308705429,
      "grad_norm": 0.44668638706207275,
      "learning_rate": 6.335906384483805e-06,
      "loss": 0.3053,
      "step": 27825
    },
    {
      "epoch": 1.2769491992106832,
      "grad_norm": 0.4732566475868225,
      "learning_rate": 6.3356701084615454e-06,
      "loss": 0.3582,
      "step": 27826
    },
    {
      "epoch": 1.2769950897159377,
      "grad_norm": 0.45154693722724915,
      "learning_rate": 6.335433829227418e-06,
      "loss": 0.3078,
      "step": 27827
    },
    {
      "epoch": 1.2770409802211922,
      "grad_norm": 0.4531998932361603,
      "learning_rate": 6.335197546781997e-06,
      "loss": 0.3445,
      "step": 27828
    },
    {
      "epoch": 1.2770868707264467,
      "grad_norm": 0.459494948387146,
      "learning_rate": 6.334961261125845e-06,
      "loss": 0.3286,
      "step": 27829
    },
    {
      "epoch": 1.2771327612317012,
      "grad_norm": 0.5111603736877441,
      "learning_rate": 6.334724972259533e-06,
      "loss": 0.4654,
      "step": 27830
    },
    {
      "epoch": 1.2771786517369557,
      "grad_norm": 0.4479008913040161,
      "learning_rate": 6.3344886801836305e-06,
      "loss": 0.315,
      "step": 27831
    },
    {
      "epoch": 1.2772245422422102,
      "grad_norm": 0.4457554519176483,
      "learning_rate": 6.334252384898702e-06,
      "loss": 0.3266,
      "step": 27832
    },
    {
      "epoch": 1.2772704327474647,
      "grad_norm": 0.45665106177330017,
      "learning_rate": 6.334016086405317e-06,
      "loss": 0.3638,
      "step": 27833
    },
    {
      "epoch": 1.277316323252719,
      "grad_norm": 0.4667263925075531,
      "learning_rate": 6.333779784704045e-06,
      "loss": 0.378,
      "step": 27834
    },
    {
      "epoch": 1.2773622137579734,
      "grad_norm": 0.455411821603775,
      "learning_rate": 6.333543479795453e-06,
      "loss": 0.3221,
      "step": 27835
    },
    {
      "epoch": 1.277408104263228,
      "grad_norm": 0.5142090320587158,
      "learning_rate": 6.333307171680112e-06,
      "loss": 0.3701,
      "step": 27836
    },
    {
      "epoch": 1.2774539947684824,
      "grad_norm": 0.44929060339927673,
      "learning_rate": 6.3330708603585865e-06,
      "loss": 0.3913,
      "step": 27837
    },
    {
      "epoch": 1.277499885273737,
      "grad_norm": 0.48981133103370667,
      "learning_rate": 6.332834545831446e-06,
      "loss": 0.3828,
      "step": 27838
    },
    {
      "epoch": 1.2775457757789912,
      "grad_norm": 0.4668658375740051,
      "learning_rate": 6.3325982280992605e-06,
      "loss": 0.3805,
      "step": 27839
    },
    {
      "epoch": 1.2775916662842457,
      "grad_norm": 0.5057257413864136,
      "learning_rate": 6.3323619071625945e-06,
      "loss": 0.4525,
      "step": 27840
    },
    {
      "epoch": 1.2776375567895002,
      "grad_norm": 0.44307488203048706,
      "learning_rate": 6.3321255830220195e-06,
      "loss": 0.2996,
      "step": 27841
    },
    {
      "epoch": 1.2776834472947547,
      "grad_norm": 0.49730661511421204,
      "learning_rate": 6.331889255678105e-06,
      "loss": 0.3831,
      "step": 27842
    },
    {
      "epoch": 1.2777293378000092,
      "grad_norm": 0.4450420141220093,
      "learning_rate": 6.3316529251314156e-06,
      "loss": 0.3415,
      "step": 27843
    },
    {
      "epoch": 1.2777752283052637,
      "grad_norm": 0.48045793175697327,
      "learning_rate": 6.33141659138252e-06,
      "loss": 0.359,
      "step": 27844
    },
    {
      "epoch": 1.2778211188105182,
      "grad_norm": 0.4715976417064667,
      "learning_rate": 6.33118025443199e-06,
      "loss": 0.3892,
      "step": 27845
    },
    {
      "epoch": 1.2778670093157727,
      "grad_norm": 0.5009045600891113,
      "learning_rate": 6.33094391428039e-06,
      "loss": 0.3689,
      "step": 27846
    },
    {
      "epoch": 1.277912899821027,
      "grad_norm": 0.4447994828224182,
      "learning_rate": 6.330707570928291e-06,
      "loss": 0.3284,
      "step": 27847
    },
    {
      "epoch": 1.2779587903262815,
      "grad_norm": 0.45731762051582336,
      "learning_rate": 6.33047122437626e-06,
      "loss": 0.3385,
      "step": 27848
    },
    {
      "epoch": 1.278004680831536,
      "grad_norm": 0.456718772649765,
      "learning_rate": 6.330234874624864e-06,
      "loss": 0.3533,
      "step": 27849
    },
    {
      "epoch": 1.2780505713367905,
      "grad_norm": 0.4643952548503876,
      "learning_rate": 6.329998521674674e-06,
      "loss": 0.3378,
      "step": 27850
    },
    {
      "epoch": 1.278096461842045,
      "grad_norm": 0.4562808871269226,
      "learning_rate": 6.329762165526259e-06,
      "loss": 0.3475,
      "step": 27851
    },
    {
      "epoch": 1.2781423523472992,
      "grad_norm": 0.42403310537338257,
      "learning_rate": 6.329525806180183e-06,
      "loss": 0.27,
      "step": 27852
    },
    {
      "epoch": 1.2781882428525537,
      "grad_norm": 0.497005432844162,
      "learning_rate": 6.329289443637019e-06,
      "loss": 0.3936,
      "step": 27853
    },
    {
      "epoch": 1.2782341333578082,
      "grad_norm": 0.4689309895038605,
      "learning_rate": 6.329053077897333e-06,
      "loss": 0.3557,
      "step": 27854
    },
    {
      "epoch": 1.2782800238630627,
      "grad_norm": 0.4359360635280609,
      "learning_rate": 6.328816708961694e-06,
      "loss": 0.3159,
      "step": 27855
    },
    {
      "epoch": 1.2783259143683172,
      "grad_norm": 0.4672968089580536,
      "learning_rate": 6.32858033683067e-06,
      "loss": 0.3799,
      "step": 27856
    },
    {
      "epoch": 1.2783718048735717,
      "grad_norm": 0.4896391034126282,
      "learning_rate": 6.328343961504828e-06,
      "loss": 0.3618,
      "step": 27857
    },
    {
      "epoch": 1.2784176953788262,
      "grad_norm": 0.5053120851516724,
      "learning_rate": 6.3281075829847395e-06,
      "loss": 0.4248,
      "step": 27858
    },
    {
      "epoch": 1.2784635858840807,
      "grad_norm": 0.4995259642601013,
      "learning_rate": 6.327871201270972e-06,
      "loss": 0.4058,
      "step": 27859
    },
    {
      "epoch": 1.278509476389335,
      "grad_norm": 0.4661610424518585,
      "learning_rate": 6.327634816364092e-06,
      "loss": 0.3305,
      "step": 27860
    },
    {
      "epoch": 1.2785553668945895,
      "grad_norm": 0.48423781991004944,
      "learning_rate": 6.327398428264669e-06,
      "loss": 0.3611,
      "step": 27861
    },
    {
      "epoch": 1.278601257399844,
      "grad_norm": 0.4712240695953369,
      "learning_rate": 6.327162036973273e-06,
      "loss": 0.3256,
      "step": 27862
    },
    {
      "epoch": 1.2786471479050985,
      "grad_norm": 0.4493459165096283,
      "learning_rate": 6.326925642490471e-06,
      "loss": 0.3117,
      "step": 27863
    },
    {
      "epoch": 1.278693038410353,
      "grad_norm": 0.5155889987945557,
      "learning_rate": 6.326689244816832e-06,
      "loss": 0.424,
      "step": 27864
    },
    {
      "epoch": 1.2787389289156073,
      "grad_norm": 0.45732903480529785,
      "learning_rate": 6.3264528439529236e-06,
      "loss": 0.323,
      "step": 27865
    },
    {
      "epoch": 1.2787848194208618,
      "grad_norm": 0.4880560338497162,
      "learning_rate": 6.326216439899314e-06,
      "loss": 0.3859,
      "step": 27866
    },
    {
      "epoch": 1.2788307099261162,
      "grad_norm": 0.45793065428733826,
      "learning_rate": 6.325980032656574e-06,
      "loss": 0.3431,
      "step": 27867
    },
    {
      "epoch": 1.2788766004313707,
      "grad_norm": 0.46137070655822754,
      "learning_rate": 6.325743622225269e-06,
      "loss": 0.3249,
      "step": 27868
    },
    {
      "epoch": 1.2789224909366252,
      "grad_norm": 0.4644947350025177,
      "learning_rate": 6.32550720860597e-06,
      "loss": 0.3684,
      "step": 27869
    },
    {
      "epoch": 1.2789683814418797,
      "grad_norm": 0.44817861914634705,
      "learning_rate": 6.325270791799244e-06,
      "loss": 0.3479,
      "step": 27870
    },
    {
      "epoch": 1.2790142719471342,
      "grad_norm": 0.485521137714386,
      "learning_rate": 6.32503437180566e-06,
      "loss": 0.3698,
      "step": 27871
    },
    {
      "epoch": 1.2790601624523887,
      "grad_norm": 0.6072576642036438,
      "learning_rate": 6.324797948625787e-06,
      "loss": 0.4526,
      "step": 27872
    },
    {
      "epoch": 1.279106052957643,
      "grad_norm": 0.48102983832359314,
      "learning_rate": 6.324561522260193e-06,
      "loss": 0.3412,
      "step": 27873
    },
    {
      "epoch": 1.2791519434628975,
      "grad_norm": 0.48291513323783875,
      "learning_rate": 6.324325092709446e-06,
      "loss": 0.3393,
      "step": 27874
    },
    {
      "epoch": 1.279197833968152,
      "grad_norm": 0.4775524437427521,
      "learning_rate": 6.324088659974116e-06,
      "loss": 0.3672,
      "step": 27875
    },
    {
      "epoch": 1.2792437244734065,
      "grad_norm": 0.5014622807502747,
      "learning_rate": 6.323852224054769e-06,
      "loss": 0.4466,
      "step": 27876
    },
    {
      "epoch": 1.2792896149786608,
      "grad_norm": 0.43454504013061523,
      "learning_rate": 6.3236157849519765e-06,
      "loss": 0.3206,
      "step": 27877
    },
    {
      "epoch": 1.2793355054839153,
      "grad_norm": 0.48976725339889526,
      "learning_rate": 6.323379342666305e-06,
      "loss": 0.3665,
      "step": 27878
    },
    {
      "epoch": 1.2793813959891698,
      "grad_norm": 0.4953702688217163,
      "learning_rate": 6.323142897198326e-06,
      "loss": 0.3923,
      "step": 27879
    },
    {
      "epoch": 1.2794272864944243,
      "grad_norm": 0.4747747480869293,
      "learning_rate": 6.322906448548604e-06,
      "loss": 0.3555,
      "step": 27880
    },
    {
      "epoch": 1.2794731769996788,
      "grad_norm": 0.4344126284122467,
      "learning_rate": 6.322669996717709e-06,
      "loss": 0.2973,
      "step": 27881
    },
    {
      "epoch": 1.2795190675049333,
      "grad_norm": 0.4542183578014374,
      "learning_rate": 6.322433541706212e-06,
      "loss": 0.3521,
      "step": 27882
    },
    {
      "epoch": 1.2795649580101878,
      "grad_norm": 0.4738132655620575,
      "learning_rate": 6.322197083514678e-06,
      "loss": 0.322,
      "step": 27883
    },
    {
      "epoch": 1.2796108485154423,
      "grad_norm": 0.45783182978630066,
      "learning_rate": 6.3219606221436766e-06,
      "loss": 0.3551,
      "step": 27884
    },
    {
      "epoch": 1.2796567390206965,
      "grad_norm": 0.4686388373374939,
      "learning_rate": 6.3217241575937805e-06,
      "loss": 0.3814,
      "step": 27885
    },
    {
      "epoch": 1.279702629525951,
      "grad_norm": 0.5053128004074097,
      "learning_rate": 6.321487689865553e-06,
      "loss": 0.4596,
      "step": 27886
    },
    {
      "epoch": 1.2797485200312055,
      "grad_norm": 0.4901399612426758,
      "learning_rate": 6.3212512189595645e-06,
      "loss": 0.4471,
      "step": 27887
    },
    {
      "epoch": 1.27979441053646,
      "grad_norm": 0.4393260180950165,
      "learning_rate": 6.321014744876384e-06,
      "loss": 0.3062,
      "step": 27888
    },
    {
      "epoch": 1.2798403010417145,
      "grad_norm": 0.49528950452804565,
      "learning_rate": 6.32077826761658e-06,
      "loss": 0.4384,
      "step": 27889
    },
    {
      "epoch": 1.2798861915469688,
      "grad_norm": 0.4651753604412079,
      "learning_rate": 6.32054178718072e-06,
      "loss": 0.374,
      "step": 27890
    },
    {
      "epoch": 1.2799320820522233,
      "grad_norm": 0.46469226479530334,
      "learning_rate": 6.320305303569376e-06,
      "loss": 0.3684,
      "step": 27891
    },
    {
      "epoch": 1.2799779725574778,
      "grad_norm": 0.4812019169330597,
      "learning_rate": 6.320068816783113e-06,
      "loss": 0.3874,
      "step": 27892
    },
    {
      "epoch": 1.2800238630627323,
      "grad_norm": 0.4360223412513733,
      "learning_rate": 6.319832326822502e-06,
      "loss": 0.3161,
      "step": 27893
    },
    {
      "epoch": 1.2800697535679868,
      "grad_norm": 0.523171603679657,
      "learning_rate": 6.319595833688109e-06,
      "loss": 0.4543,
      "step": 27894
    },
    {
      "epoch": 1.2801156440732413,
      "grad_norm": 0.5291123390197754,
      "learning_rate": 6.319359337380505e-06,
      "loss": 0.4026,
      "step": 27895
    },
    {
      "epoch": 1.2801615345784958,
      "grad_norm": 0.43482133746147156,
      "learning_rate": 6.31912283790026e-06,
      "loss": 0.2857,
      "step": 27896
    },
    {
      "epoch": 1.2802074250837503,
      "grad_norm": 0.46525150537490845,
      "learning_rate": 6.31888633524794e-06,
      "loss": 0.3572,
      "step": 27897
    },
    {
      "epoch": 1.2802533155890046,
      "grad_norm": 0.45712149143218994,
      "learning_rate": 6.318649829424114e-06,
      "loss": 0.3175,
      "step": 27898
    },
    {
      "epoch": 1.280299206094259,
      "grad_norm": 0.4696604609489441,
      "learning_rate": 6.318413320429352e-06,
      "loss": 0.3428,
      "step": 27899
    },
    {
      "epoch": 1.2803450965995136,
      "grad_norm": 0.4610048234462738,
      "learning_rate": 6.318176808264221e-06,
      "loss": 0.382,
      "step": 27900
    },
    {
      "epoch": 1.280390987104768,
      "grad_norm": 0.4528370797634125,
      "learning_rate": 6.317940292929292e-06,
      "loss": 0.3834,
      "step": 27901
    },
    {
      "epoch": 1.2804368776100226,
      "grad_norm": 0.461145281791687,
      "learning_rate": 6.3177037744251335e-06,
      "loss": 0.3133,
      "step": 27902
    },
    {
      "epoch": 1.2804827681152768,
      "grad_norm": 0.4779565632343292,
      "learning_rate": 6.317467252752312e-06,
      "loss": 0.3341,
      "step": 27903
    },
    {
      "epoch": 1.2805286586205313,
      "grad_norm": 0.5112033486366272,
      "learning_rate": 6.317230727911398e-06,
      "loss": 0.393,
      "step": 27904
    },
    {
      "epoch": 1.2805745491257858,
      "grad_norm": 0.48595502972602844,
      "learning_rate": 6.316994199902958e-06,
      "loss": 0.4026,
      "step": 27905
    },
    {
      "epoch": 1.2806204396310403,
      "grad_norm": 0.4899119734764099,
      "learning_rate": 6.316757668727564e-06,
      "loss": 0.4013,
      "step": 27906
    },
    {
      "epoch": 1.2806663301362948,
      "grad_norm": 0.48145240545272827,
      "learning_rate": 6.316521134385784e-06,
      "loss": 0.3462,
      "step": 27907
    },
    {
      "epoch": 1.2807122206415493,
      "grad_norm": 0.47885310649871826,
      "learning_rate": 6.316284596878185e-06,
      "loss": 0.4113,
      "step": 27908
    },
    {
      "epoch": 1.2807581111468038,
      "grad_norm": 0.4890669882297516,
      "learning_rate": 6.316048056205337e-06,
      "loss": 0.3958,
      "step": 27909
    },
    {
      "epoch": 1.2808040016520583,
      "grad_norm": 0.4389622211456299,
      "learning_rate": 6.31581151236781e-06,
      "loss": 0.2998,
      "step": 27910
    },
    {
      "epoch": 1.2808498921573126,
      "grad_norm": 0.4524305760860443,
      "learning_rate": 6.3155749653661705e-06,
      "loss": 0.3378,
      "step": 27911
    },
    {
      "epoch": 1.280895782662567,
      "grad_norm": 0.5244722962379456,
      "learning_rate": 6.315338415200989e-06,
      "loss": 0.3106,
      "step": 27912
    },
    {
      "epoch": 1.2809416731678216,
      "grad_norm": 0.46941402554512024,
      "learning_rate": 6.315101861872834e-06,
      "loss": 0.3822,
      "step": 27913
    },
    {
      "epoch": 1.280987563673076,
      "grad_norm": 0.4887068569660187,
      "learning_rate": 6.314865305382274e-06,
      "loss": 0.3536,
      "step": 27914
    },
    {
      "epoch": 1.2810334541783304,
      "grad_norm": 0.5552395582199097,
      "learning_rate": 6.314628745729876e-06,
      "loss": 0.4114,
      "step": 27915
    },
    {
      "epoch": 1.2810793446835849,
      "grad_norm": 0.481122225522995,
      "learning_rate": 6.314392182916213e-06,
      "loss": 0.4299,
      "step": 27916
    },
    {
      "epoch": 1.2811252351888394,
      "grad_norm": 0.42810603976249695,
      "learning_rate": 6.31415561694185e-06,
      "loss": 0.2749,
      "step": 27917
    },
    {
      "epoch": 1.2811711256940939,
      "grad_norm": 0.4875357449054718,
      "learning_rate": 6.31391904780736e-06,
      "loss": 0.3853,
      "step": 27918
    },
    {
      "epoch": 1.2812170161993484,
      "grad_norm": 0.4573619067668915,
      "learning_rate": 6.313682475513307e-06,
      "loss": 0.3495,
      "step": 27919
    },
    {
      "epoch": 1.2812629067046029,
      "grad_norm": 0.4879096448421478,
      "learning_rate": 6.3134459000602635e-06,
      "loss": 0.3623,
      "step": 27920
    },
    {
      "epoch": 1.2813087972098574,
      "grad_norm": 0.4448162317276001,
      "learning_rate": 6.313209321448796e-06,
      "loss": 0.3152,
      "step": 27921
    },
    {
      "epoch": 1.2813546877151119,
      "grad_norm": 0.4345643222332001,
      "learning_rate": 6.312972739679476e-06,
      "loss": 0.3223,
      "step": 27922
    },
    {
      "epoch": 1.2814005782203661,
      "grad_norm": 0.4322528541088104,
      "learning_rate": 6.31273615475287e-06,
      "loss": 0.2858,
      "step": 27923
    },
    {
      "epoch": 1.2814464687256206,
      "grad_norm": 0.5135278105735779,
      "learning_rate": 6.312499566669549e-06,
      "loss": 0.3911,
      "step": 27924
    },
    {
      "epoch": 1.2814923592308751,
      "grad_norm": 0.46670961380004883,
      "learning_rate": 6.312262975430079e-06,
      "loss": 0.3437,
      "step": 27925
    },
    {
      "epoch": 1.2815382497361296,
      "grad_norm": 0.4446643888950348,
      "learning_rate": 6.312026381035031e-06,
      "loss": 0.3172,
      "step": 27926
    },
    {
      "epoch": 1.2815841402413841,
      "grad_norm": 0.4797717034816742,
      "learning_rate": 6.311789783484975e-06,
      "loss": 0.3902,
      "step": 27927
    },
    {
      "epoch": 1.2816300307466384,
      "grad_norm": 0.49759751558303833,
      "learning_rate": 6.311553182780478e-06,
      "loss": 0.409,
      "step": 27928
    },
    {
      "epoch": 1.281675921251893,
      "grad_norm": 0.48105165362358093,
      "learning_rate": 6.311316578922109e-06,
      "loss": 0.4148,
      "step": 27929
    },
    {
      "epoch": 1.2817218117571474,
      "grad_norm": 0.450704962015152,
      "learning_rate": 6.311079971910438e-06,
      "loss": 0.3413,
      "step": 27930
    },
    {
      "epoch": 1.2817677022624019,
      "grad_norm": 0.44499722123146057,
      "learning_rate": 6.310843361746033e-06,
      "loss": 0.3054,
      "step": 27931
    },
    {
      "epoch": 1.2818135927676564,
      "grad_norm": 0.46481916308403015,
      "learning_rate": 6.310606748429463e-06,
      "loss": 0.4073,
      "step": 27932
    },
    {
      "epoch": 1.2818594832729109,
      "grad_norm": 0.46222159266471863,
      "learning_rate": 6.3103701319612985e-06,
      "loss": 0.3373,
      "step": 27933
    },
    {
      "epoch": 1.2819053737781654,
      "grad_norm": 0.4440707862377167,
      "learning_rate": 6.3101335123421085e-06,
      "loss": 0.3275,
      "step": 27934
    },
    {
      "epoch": 1.2819512642834199,
      "grad_norm": 0.45265746116638184,
      "learning_rate": 6.3098968895724574e-06,
      "loss": 0.3214,
      "step": 27935
    },
    {
      "epoch": 1.2819971547886742,
      "grad_norm": 0.4968338906764984,
      "learning_rate": 6.309660263652921e-06,
      "loss": 0.3438,
      "step": 27936
    },
    {
      "epoch": 1.2820430452939287,
      "grad_norm": 0.47855669260025024,
      "learning_rate": 6.309423634584063e-06,
      "loss": 0.3631,
      "step": 27937
    },
    {
      "epoch": 1.2820889357991831,
      "grad_norm": 0.47526058554649353,
      "learning_rate": 6.309187002366456e-06,
      "loss": 0.3582,
      "step": 27938
    },
    {
      "epoch": 1.2821348263044376,
      "grad_norm": 0.4880504310131073,
      "learning_rate": 6.308950367000666e-06,
      "loss": 0.4343,
      "step": 27939
    },
    {
      "epoch": 1.2821807168096921,
      "grad_norm": 0.4558505415916443,
      "learning_rate": 6.308713728487264e-06,
      "loss": 0.3352,
      "step": 27940
    },
    {
      "epoch": 1.2822266073149464,
      "grad_norm": 0.4458456337451935,
      "learning_rate": 6.308477086826819e-06,
      "loss": 0.3777,
      "step": 27941
    },
    {
      "epoch": 1.282272497820201,
      "grad_norm": 0.5169133543968201,
      "learning_rate": 6.308240442019899e-06,
      "loss": 0.3909,
      "step": 27942
    },
    {
      "epoch": 1.2823183883254554,
      "grad_norm": 0.5154356360435486,
      "learning_rate": 6.3080037940670736e-06,
      "loss": 0.4837,
      "step": 27943
    },
    {
      "epoch": 1.28236427883071,
      "grad_norm": 0.5061883330345154,
      "learning_rate": 6.307767142968912e-06,
      "loss": 0.4431,
      "step": 27944
    },
    {
      "epoch": 1.2824101693359644,
      "grad_norm": 0.49453747272491455,
      "learning_rate": 6.307530488725983e-06,
      "loss": 0.3729,
      "step": 27945
    },
    {
      "epoch": 1.282456059841219,
      "grad_norm": 0.45170867443084717,
      "learning_rate": 6.307293831338856e-06,
      "loss": 0.3173,
      "step": 27946
    },
    {
      "epoch": 1.2825019503464734,
      "grad_norm": 0.43428173661231995,
      "learning_rate": 6.3070571708081e-06,
      "loss": 0.3111,
      "step": 27947
    },
    {
      "epoch": 1.282547840851728,
      "grad_norm": 0.4775456190109253,
      "learning_rate": 6.306820507134285e-06,
      "loss": 0.3464,
      "step": 27948
    },
    {
      "epoch": 1.2825937313569822,
      "grad_norm": 0.45871850848197937,
      "learning_rate": 6.306583840317977e-06,
      "loss": 0.3635,
      "step": 27949
    },
    {
      "epoch": 1.2826396218622367,
      "grad_norm": 0.4917430281639099,
      "learning_rate": 6.306347170359749e-06,
      "loss": 0.3302,
      "step": 27950
    },
    {
      "epoch": 1.2826855123674912,
      "grad_norm": 0.44096633791923523,
      "learning_rate": 6.306110497260169e-06,
      "loss": 0.291,
      "step": 27951
    },
    {
      "epoch": 1.2827314028727457,
      "grad_norm": 0.4854171872138977,
      "learning_rate": 6.3058738210198035e-06,
      "loss": 0.3807,
      "step": 27952
    },
    {
      "epoch": 1.2827772933780002,
      "grad_norm": 0.4535027742385864,
      "learning_rate": 6.305637141639225e-06,
      "loss": 0.3188,
      "step": 27953
    },
    {
      "epoch": 1.2828231838832544,
      "grad_norm": 0.5031104683876038,
      "learning_rate": 6.3054004591189995e-06,
      "loss": 0.3868,
      "step": 27954
    },
    {
      "epoch": 1.282869074388509,
      "grad_norm": 0.511268675327301,
      "learning_rate": 6.305163773459699e-06,
      "loss": 0.4064,
      "step": 27955
    },
    {
      "epoch": 1.2829149648937634,
      "grad_norm": 0.49902039766311646,
      "learning_rate": 6.304927084661893e-06,
      "loss": 0.4054,
      "step": 27956
    },
    {
      "epoch": 1.282960855399018,
      "grad_norm": 0.4851906895637512,
      "learning_rate": 6.304690392726147e-06,
      "loss": 0.3505,
      "step": 27957
    },
    {
      "epoch": 1.2830067459042724,
      "grad_norm": 0.44817718863487244,
      "learning_rate": 6.304453697653033e-06,
      "loss": 0.3679,
      "step": 27958
    },
    {
      "epoch": 1.283052636409527,
      "grad_norm": 0.45906683802604675,
      "learning_rate": 6.3042169994431194e-06,
      "loss": 0.3813,
      "step": 27959
    },
    {
      "epoch": 1.2830985269147814,
      "grad_norm": 0.4838658273220062,
      "learning_rate": 6.303980298096975e-06,
      "loss": 0.3445,
      "step": 27960
    },
    {
      "epoch": 1.283144417420036,
      "grad_norm": 0.4532667398452759,
      "learning_rate": 6.30374359361517e-06,
      "loss": 0.3289,
      "step": 27961
    },
    {
      "epoch": 1.2831903079252902,
      "grad_norm": 0.5090999603271484,
      "learning_rate": 6.303506885998275e-06,
      "loss": 0.4119,
      "step": 27962
    },
    {
      "epoch": 1.2832361984305447,
      "grad_norm": 0.44353389739990234,
      "learning_rate": 6.303270175246855e-06,
      "loss": 0.2971,
      "step": 27963
    },
    {
      "epoch": 1.2832820889357992,
      "grad_norm": 0.45460933446884155,
      "learning_rate": 6.303033461361483e-06,
      "loss": 0.3096,
      "step": 27964
    },
    {
      "epoch": 1.2833279794410537,
      "grad_norm": 0.5018517374992371,
      "learning_rate": 6.302796744342725e-06,
      "loss": 0.4098,
      "step": 27965
    },
    {
      "epoch": 1.283373869946308,
      "grad_norm": 0.47725749015808105,
      "learning_rate": 6.3025600241911544e-06,
      "loss": 0.309,
      "step": 27966
    },
    {
      "epoch": 1.2834197604515625,
      "grad_norm": 0.46066346764564514,
      "learning_rate": 6.302323300907337e-06,
      "loss": 0.3804,
      "step": 27967
    },
    {
      "epoch": 1.283465650956817,
      "grad_norm": 0.4454992711544037,
      "learning_rate": 6.302086574491842e-06,
      "loss": 0.3211,
      "step": 27968
    },
    {
      "epoch": 1.2835115414620715,
      "grad_norm": 0.4430641233921051,
      "learning_rate": 6.30184984494524e-06,
      "loss": 0.3084,
      "step": 27969
    },
    {
      "epoch": 1.283557431967326,
      "grad_norm": 0.4556950330734253,
      "learning_rate": 6.301613112268101e-06,
      "loss": 0.3343,
      "step": 27970
    },
    {
      "epoch": 1.2836033224725805,
      "grad_norm": 0.4609871208667755,
      "learning_rate": 6.301376376460993e-06,
      "loss": 0.3627,
      "step": 27971
    },
    {
      "epoch": 1.283649212977835,
      "grad_norm": 0.46355926990509033,
      "learning_rate": 6.301139637524486e-06,
      "loss": 0.3697,
      "step": 27972
    },
    {
      "epoch": 1.2836951034830895,
      "grad_norm": 0.4374755918979645,
      "learning_rate": 6.300902895459149e-06,
      "loss": 0.3374,
      "step": 27973
    },
    {
      "epoch": 1.2837409939883437,
      "grad_norm": 0.5259857177734375,
      "learning_rate": 6.300666150265549e-06,
      "loss": 0.3814,
      "step": 27974
    },
    {
      "epoch": 1.2837868844935982,
      "grad_norm": 0.46450692415237427,
      "learning_rate": 6.300429401944259e-06,
      "loss": 0.3353,
      "step": 27975
    },
    {
      "epoch": 1.2838327749988527,
      "grad_norm": 0.45779046416282654,
      "learning_rate": 6.300192650495847e-06,
      "loss": 0.3714,
      "step": 27976
    },
    {
      "epoch": 1.2838786655041072,
      "grad_norm": 0.45795565843582153,
      "learning_rate": 6.299955895920882e-06,
      "loss": 0.3299,
      "step": 27977
    },
    {
      "epoch": 1.2839245560093617,
      "grad_norm": 0.495059609413147,
      "learning_rate": 6.299719138219933e-06,
      "loss": 0.3935,
      "step": 27978
    },
    {
      "epoch": 1.283970446514616,
      "grad_norm": 0.47626081109046936,
      "learning_rate": 6.299482377393569e-06,
      "loss": 0.3286,
      "step": 27979
    },
    {
      "epoch": 1.2840163370198705,
      "grad_norm": 0.4811919033527374,
      "learning_rate": 6.2992456134423595e-06,
      "loss": 0.3813,
      "step": 27980
    },
    {
      "epoch": 1.284062227525125,
      "grad_norm": 0.4993738830089569,
      "learning_rate": 6.2990088463668765e-06,
      "loss": 0.4374,
      "step": 27981
    },
    {
      "epoch": 1.2841081180303795,
      "grad_norm": 0.49372154474258423,
      "learning_rate": 6.298772076167685e-06,
      "loss": 0.396,
      "step": 27982
    },
    {
      "epoch": 1.284154008535634,
      "grad_norm": 0.4576811194419861,
      "learning_rate": 6.298535302845358e-06,
      "loss": 0.3555,
      "step": 27983
    },
    {
      "epoch": 1.2841998990408885,
      "grad_norm": 0.5154021382331848,
      "learning_rate": 6.298298526400464e-06,
      "loss": 0.4067,
      "step": 27984
    },
    {
      "epoch": 1.284245789546143,
      "grad_norm": 0.5037956237792969,
      "learning_rate": 6.29806174683357e-06,
      "loss": 0.342,
      "step": 27985
    },
    {
      "epoch": 1.2842916800513975,
      "grad_norm": 0.47566887736320496,
      "learning_rate": 6.297824964145248e-06,
      "loss": 0.3654,
      "step": 27986
    },
    {
      "epoch": 1.2843375705566518,
      "grad_norm": 0.5180149674415588,
      "learning_rate": 6.297588178336067e-06,
      "loss": 0.401,
      "step": 27987
    },
    {
      "epoch": 1.2843834610619063,
      "grad_norm": 0.4895417094230652,
      "learning_rate": 6.297351389406596e-06,
      "loss": 0.3987,
      "step": 27988
    },
    {
      "epoch": 1.2844293515671608,
      "grad_norm": 0.4662227928638458,
      "learning_rate": 6.2971145973574044e-06,
      "loss": 0.3356,
      "step": 27989
    },
    {
      "epoch": 1.2844752420724153,
      "grad_norm": 0.5006363391876221,
      "learning_rate": 6.296877802189061e-06,
      "loss": 0.4051,
      "step": 27990
    },
    {
      "epoch": 1.2845211325776698,
      "grad_norm": 0.4571870267391205,
      "learning_rate": 6.296641003902134e-06,
      "loss": 0.3287,
      "step": 27991
    },
    {
      "epoch": 1.284567023082924,
      "grad_norm": 0.5016179084777832,
      "learning_rate": 6.2964042024971975e-06,
      "loss": 0.3375,
      "step": 27992
    },
    {
      "epoch": 1.2846129135881785,
      "grad_norm": 0.49141407012939453,
      "learning_rate": 6.296167397974817e-06,
      "loss": 0.3696,
      "step": 27993
    },
    {
      "epoch": 1.284658804093433,
      "grad_norm": 0.5017837285995483,
      "learning_rate": 6.2959305903355635e-06,
      "loss": 0.4086,
      "step": 27994
    },
    {
      "epoch": 1.2847046945986875,
      "grad_norm": 0.4881719946861267,
      "learning_rate": 6.295693779580005e-06,
      "loss": 0.4277,
      "step": 27995
    },
    {
      "epoch": 1.284750585103942,
      "grad_norm": 0.4799562990665436,
      "learning_rate": 6.295456965708713e-06,
      "loss": 0.3243,
      "step": 27996
    },
    {
      "epoch": 1.2847964756091965,
      "grad_norm": 0.455783873796463,
      "learning_rate": 6.295220148722254e-06,
      "loss": 0.3909,
      "step": 27997
    },
    {
      "epoch": 1.284842366114451,
      "grad_norm": 0.4835767447948456,
      "learning_rate": 6.294983328621202e-06,
      "loss": 0.3856,
      "step": 27998
    },
    {
      "epoch": 1.2848882566197055,
      "grad_norm": 0.4736703932285309,
      "learning_rate": 6.2947465054061215e-06,
      "loss": 0.4078,
      "step": 27999
    },
    {
      "epoch": 1.2849341471249598,
      "grad_norm": 0.5019463896751404,
      "learning_rate": 6.294509679077585e-06,
      "loss": 0.4465,
      "step": 28000
    },
    {
      "epoch": 1.2849800376302143,
      "grad_norm": 0.4571838080883026,
      "learning_rate": 6.2942728496361625e-06,
      "loss": 0.3292,
      "step": 28001
    },
    {
      "epoch": 1.2850259281354688,
      "grad_norm": 0.4683063328266144,
      "learning_rate": 6.294036017082421e-06,
      "loss": 0.3556,
      "step": 28002
    },
    {
      "epoch": 1.2850718186407233,
      "grad_norm": 0.4869784414768219,
      "learning_rate": 6.293799181416931e-06,
      "loss": 0.4098,
      "step": 28003
    },
    {
      "epoch": 1.2851177091459776,
      "grad_norm": 0.46651285886764526,
      "learning_rate": 6.293562342640262e-06,
      "loss": 0.3221,
      "step": 28004
    },
    {
      "epoch": 1.285163599651232,
      "grad_norm": 0.5062577128410339,
      "learning_rate": 6.293325500752986e-06,
      "loss": 0.3812,
      "step": 28005
    },
    {
      "epoch": 1.2852094901564866,
      "grad_norm": 0.44168779253959656,
      "learning_rate": 6.293088655755669e-06,
      "loss": 0.3291,
      "step": 28006
    },
    {
      "epoch": 1.285255380661741,
      "grad_norm": 0.518917441368103,
      "learning_rate": 6.292851807648882e-06,
      "loss": 0.3835,
      "step": 28007
    },
    {
      "epoch": 1.2853012711669956,
      "grad_norm": 0.4333477020263672,
      "learning_rate": 6.292614956433194e-06,
      "loss": 0.3032,
      "step": 28008
    },
    {
      "epoch": 1.28534716167225,
      "grad_norm": 0.43126508593559265,
      "learning_rate": 6.292378102109175e-06,
      "loss": 0.2957,
      "step": 28009
    },
    {
      "epoch": 1.2853930521775045,
      "grad_norm": 0.4491947591304779,
      "learning_rate": 6.2921412446773955e-06,
      "loss": 0.3322,
      "step": 28010
    },
    {
      "epoch": 1.285438942682759,
      "grad_norm": 0.4510684311389923,
      "learning_rate": 6.2919043841384235e-06,
      "loss": 0.3245,
      "step": 28011
    },
    {
      "epoch": 1.2854848331880133,
      "grad_norm": 0.4779126048088074,
      "learning_rate": 6.291667520492829e-06,
      "loss": 0.393,
      "step": 28012
    },
    {
      "epoch": 1.2855307236932678,
      "grad_norm": 0.47748321294784546,
      "learning_rate": 6.2914306537411815e-06,
      "loss": 0.3868,
      "step": 28013
    },
    {
      "epoch": 1.2855766141985223,
      "grad_norm": 0.46777623891830444,
      "learning_rate": 6.291193783884051e-06,
      "loss": 0.3615,
      "step": 28014
    },
    {
      "epoch": 1.2856225047037768,
      "grad_norm": 0.46542301774024963,
      "learning_rate": 6.290956910922008e-06,
      "loss": 0.3441,
      "step": 28015
    },
    {
      "epoch": 1.2856683952090313,
      "grad_norm": 0.47854384779930115,
      "learning_rate": 6.29072003485562e-06,
      "loss": 0.4114,
      "step": 28016
    },
    {
      "epoch": 1.2857142857142856,
      "grad_norm": 0.5072178244590759,
      "learning_rate": 6.290483155685458e-06,
      "loss": 0.4176,
      "step": 28017
    },
    {
      "epoch": 1.28576017621954,
      "grad_norm": 0.5169182419776917,
      "learning_rate": 6.290246273412092e-06,
      "loss": 0.4133,
      "step": 28018
    },
    {
      "epoch": 1.2858060667247946,
      "grad_norm": 0.45771628618240356,
      "learning_rate": 6.290009388036089e-06,
      "loss": 0.3192,
      "step": 28019
    },
    {
      "epoch": 1.285851957230049,
      "grad_norm": 0.47772571444511414,
      "learning_rate": 6.2897724995580225e-06,
      "loss": 0.3721,
      "step": 28020
    },
    {
      "epoch": 1.2858978477353036,
      "grad_norm": 0.47401484847068787,
      "learning_rate": 6.289535607978461e-06,
      "loss": 0.3787,
      "step": 28021
    },
    {
      "epoch": 1.285943738240558,
      "grad_norm": 0.46425482630729675,
      "learning_rate": 6.289298713297971e-06,
      "loss": 0.3446,
      "step": 28022
    },
    {
      "epoch": 1.2859896287458126,
      "grad_norm": 0.4534512162208557,
      "learning_rate": 6.289061815517126e-06,
      "loss": 0.2999,
      "step": 28023
    },
    {
      "epoch": 1.286035519251067,
      "grad_norm": 0.4260218143463135,
      "learning_rate": 6.288824914636493e-06,
      "loss": 0.297,
      "step": 28024
    },
    {
      "epoch": 1.2860814097563213,
      "grad_norm": 0.4626574218273163,
      "learning_rate": 6.2885880106566425e-06,
      "loss": 0.3411,
      "step": 28025
    },
    {
      "epoch": 1.2861273002615758,
      "grad_norm": 0.4694831967353821,
      "learning_rate": 6.288351103578145e-06,
      "loss": 0.361,
      "step": 28026
    },
    {
      "epoch": 1.2861731907668303,
      "grad_norm": 0.47322583198547363,
      "learning_rate": 6.288114193401571e-06,
      "loss": 0.3463,
      "step": 28027
    },
    {
      "epoch": 1.2862190812720848,
      "grad_norm": 0.42018210887908936,
      "learning_rate": 6.287877280127488e-06,
      "loss": 0.2408,
      "step": 28028
    },
    {
      "epoch": 1.2862649717773393,
      "grad_norm": 0.4446282684803009,
      "learning_rate": 6.287640363756468e-06,
      "loss": 0.3537,
      "step": 28029
    },
    {
      "epoch": 1.2863108622825936,
      "grad_norm": 0.44221577048301697,
      "learning_rate": 6.287403444289077e-06,
      "loss": 0.3018,
      "step": 28030
    },
    {
      "epoch": 1.2863567527878481,
      "grad_norm": 0.4677393436431885,
      "learning_rate": 6.2871665217258875e-06,
      "loss": 0.3272,
      "step": 28031
    },
    {
      "epoch": 1.2864026432931026,
      "grad_norm": 0.514629065990448,
      "learning_rate": 6.2869295960674705e-06,
      "loss": 0.3446,
      "step": 28032
    },
    {
      "epoch": 1.286448533798357,
      "grad_norm": 0.43056970834732056,
      "learning_rate": 6.2866926673143924e-06,
      "loss": 0.263,
      "step": 28033
    },
    {
      "epoch": 1.2864944243036116,
      "grad_norm": 0.5042280554771423,
      "learning_rate": 6.286455735467224e-06,
      "loss": 0.4868,
      "step": 28034
    },
    {
      "epoch": 1.286540314808866,
      "grad_norm": 0.4717193841934204,
      "learning_rate": 6.2862188005265365e-06,
      "loss": 0.402,
      "step": 28035
    },
    {
      "epoch": 1.2865862053141206,
      "grad_norm": 0.48021817207336426,
      "learning_rate": 6.285981862492898e-06,
      "loss": 0.3892,
      "step": 28036
    },
    {
      "epoch": 1.286632095819375,
      "grad_norm": 0.49710315465927124,
      "learning_rate": 6.285744921366881e-06,
      "loss": 0.394,
      "step": 28037
    },
    {
      "epoch": 1.2866779863246294,
      "grad_norm": 0.43318331241607666,
      "learning_rate": 6.285507977149052e-06,
      "loss": 0.2841,
      "step": 28038
    },
    {
      "epoch": 1.2867238768298839,
      "grad_norm": 0.4696180820465088,
      "learning_rate": 6.285271029839981e-06,
      "loss": 0.3057,
      "step": 28039
    },
    {
      "epoch": 1.2867697673351384,
      "grad_norm": 0.46836262941360474,
      "learning_rate": 6.285034079440241e-06,
      "loss": 0.3766,
      "step": 28040
    },
    {
      "epoch": 1.2868156578403929,
      "grad_norm": 0.43587324023246765,
      "learning_rate": 6.284797125950397e-06,
      "loss": 0.2917,
      "step": 28041
    },
    {
      "epoch": 1.2868615483456474,
      "grad_norm": 0.4919810891151428,
      "learning_rate": 6.284560169371023e-06,
      "loss": 0.3502,
      "step": 28042
    },
    {
      "epoch": 1.2869074388509016,
      "grad_norm": 0.5080631971359253,
      "learning_rate": 6.284323209702688e-06,
      "loss": 0.4267,
      "step": 28043
    },
    {
      "epoch": 1.2869533293561561,
      "grad_norm": 0.45041918754577637,
      "learning_rate": 6.28408624694596e-06,
      "loss": 0.3629,
      "step": 28044
    },
    {
      "epoch": 1.2869992198614106,
      "grad_norm": 0.5092109441757202,
      "learning_rate": 6.283849281101409e-06,
      "loss": 0.395,
      "step": 28045
    },
    {
      "epoch": 1.2870451103666651,
      "grad_norm": 0.45699307322502136,
      "learning_rate": 6.2836123121696065e-06,
      "loss": 0.3538,
      "step": 28046
    },
    {
      "epoch": 1.2870910008719196,
      "grad_norm": 0.44629091024398804,
      "learning_rate": 6.2833753401511214e-06,
      "loss": 0.3173,
      "step": 28047
    },
    {
      "epoch": 1.2871368913771741,
      "grad_norm": 0.4377323389053345,
      "learning_rate": 6.2831383650465235e-06,
      "loss": 0.3246,
      "step": 28048
    },
    {
      "epoch": 1.2871827818824286,
      "grad_norm": 0.48890864849090576,
      "learning_rate": 6.282901386856385e-06,
      "loss": 0.4058,
      "step": 28049
    },
    {
      "epoch": 1.2872286723876831,
      "grad_norm": 0.4579782485961914,
      "learning_rate": 6.282664405581272e-06,
      "loss": 0.3218,
      "step": 28050
    },
    {
      "epoch": 1.2872745628929374,
      "grad_norm": 0.43510380387306213,
      "learning_rate": 6.282427421221755e-06,
      "loss": 0.2847,
      "step": 28051
    },
    {
      "epoch": 1.287320453398192,
      "grad_norm": 0.4762396812438965,
      "learning_rate": 6.282190433778405e-06,
      "loss": 0.3605,
      "step": 28052
    },
    {
      "epoch": 1.2873663439034464,
      "grad_norm": 0.45483747124671936,
      "learning_rate": 6.2819534432517915e-06,
      "loss": 0.3466,
      "step": 28053
    },
    {
      "epoch": 1.287412234408701,
      "grad_norm": 0.48180606961250305,
      "learning_rate": 6.2817164496424865e-06,
      "loss": 0.4147,
      "step": 28054
    },
    {
      "epoch": 1.2874581249139552,
      "grad_norm": 0.4472403824329376,
      "learning_rate": 6.281479452951056e-06,
      "loss": 0.2997,
      "step": 28055
    },
    {
      "epoch": 1.2875040154192097,
      "grad_norm": 0.4397311508655548,
      "learning_rate": 6.2812424531780716e-06,
      "loss": 0.3331,
      "step": 28056
    },
    {
      "epoch": 1.2875499059244642,
      "grad_norm": 0.49483487010002136,
      "learning_rate": 6.281005450324105e-06,
      "loss": 0.3491,
      "step": 28057
    },
    {
      "epoch": 1.2875957964297187,
      "grad_norm": 0.4228760600090027,
      "learning_rate": 6.280768444389724e-06,
      "loss": 0.2597,
      "step": 28058
    },
    {
      "epoch": 1.2876416869349732,
      "grad_norm": 0.5105610489845276,
      "learning_rate": 6.280531435375499e-06,
      "loss": 0.4204,
      "step": 28059
    },
    {
      "epoch": 1.2876875774402277,
      "grad_norm": 0.4278881251811981,
      "learning_rate": 6.280294423282001e-06,
      "loss": 0.2995,
      "step": 28060
    },
    {
      "epoch": 1.2877334679454822,
      "grad_norm": 0.4799186885356903,
      "learning_rate": 6.2800574081097965e-06,
      "loss": 0.4124,
      "step": 28061
    },
    {
      "epoch": 1.2877793584507367,
      "grad_norm": 0.5145034193992615,
      "learning_rate": 6.279820389859459e-06,
      "loss": 0.4406,
      "step": 28062
    },
    {
      "epoch": 1.287825248955991,
      "grad_norm": 0.4484533965587616,
      "learning_rate": 6.279583368531559e-06,
      "loss": 0.3257,
      "step": 28063
    },
    {
      "epoch": 1.2878711394612454,
      "grad_norm": 0.48334038257598877,
      "learning_rate": 6.2793463441266625e-06,
      "loss": 0.3922,
      "step": 28064
    },
    {
      "epoch": 1.2879170299665,
      "grad_norm": 0.49381059408187866,
      "learning_rate": 6.279109316645344e-06,
      "loss": 0.3571,
      "step": 28065
    },
    {
      "epoch": 1.2879629204717544,
      "grad_norm": 0.4796421527862549,
      "learning_rate": 6.278872286088171e-06,
      "loss": 0.3597,
      "step": 28066
    },
    {
      "epoch": 1.288008810977009,
      "grad_norm": 0.43482673168182373,
      "learning_rate": 6.278635252455711e-06,
      "loss": 0.3238,
      "step": 28067
    },
    {
      "epoch": 1.2880547014822632,
      "grad_norm": 0.47177642583847046,
      "learning_rate": 6.27839821574854e-06,
      "loss": 0.3606,
      "step": 28068
    },
    {
      "epoch": 1.2881005919875177,
      "grad_norm": 0.4608008563518524,
      "learning_rate": 6.278161175967223e-06,
      "loss": 0.3545,
      "step": 28069
    },
    {
      "epoch": 1.2881464824927722,
      "grad_norm": 0.4514561891555786,
      "learning_rate": 6.277924133112333e-06,
      "loss": 0.3217,
      "step": 28070
    },
    {
      "epoch": 1.2881923729980267,
      "grad_norm": 0.5066956281661987,
      "learning_rate": 6.277687087184438e-06,
      "loss": 0.2917,
      "step": 28071
    },
    {
      "epoch": 1.2882382635032812,
      "grad_norm": 0.4906873106956482,
      "learning_rate": 6.27745003818411e-06,
      "loss": 0.3635,
      "step": 28072
    },
    {
      "epoch": 1.2882841540085357,
      "grad_norm": 0.4679623544216156,
      "learning_rate": 6.277212986111916e-06,
      "loss": 0.3692,
      "step": 28073
    },
    {
      "epoch": 1.2883300445137902,
      "grad_norm": 0.46598020195961,
      "learning_rate": 6.276975930968429e-06,
      "loss": 0.3122,
      "step": 28074
    },
    {
      "epoch": 1.2883759350190447,
      "grad_norm": 0.4327479898929596,
      "learning_rate": 6.276738872754217e-06,
      "loss": 0.3091,
      "step": 28075
    },
    {
      "epoch": 1.288421825524299,
      "grad_norm": 0.5081113576889038,
      "learning_rate": 6.276501811469852e-06,
      "loss": 0.4075,
      "step": 28076
    },
    {
      "epoch": 1.2884677160295535,
      "grad_norm": 0.46634232997894287,
      "learning_rate": 6.276264747115903e-06,
      "loss": 0.3285,
      "step": 28077
    },
    {
      "epoch": 1.288513606534808,
      "grad_norm": 0.43505290150642395,
      "learning_rate": 6.27602767969294e-06,
      "loss": 0.3273,
      "step": 28078
    },
    {
      "epoch": 1.2885594970400625,
      "grad_norm": 0.46722015738487244,
      "learning_rate": 6.275790609201532e-06,
      "loss": 0.3797,
      "step": 28079
    },
    {
      "epoch": 1.288605387545317,
      "grad_norm": 0.4527287483215332,
      "learning_rate": 6.275553535642251e-06,
      "loss": 0.2893,
      "step": 28080
    },
    {
      "epoch": 1.2886512780505712,
      "grad_norm": 0.43942123651504517,
      "learning_rate": 6.275316459015667e-06,
      "loss": 0.3013,
      "step": 28081
    },
    {
      "epoch": 1.2886971685558257,
      "grad_norm": 0.4568745195865631,
      "learning_rate": 6.275079379322349e-06,
      "loss": 0.3836,
      "step": 28082
    },
    {
      "epoch": 1.2887430590610802,
      "grad_norm": 0.452274352312088,
      "learning_rate": 6.2748422965628675e-06,
      "loss": 0.3487,
      "step": 28083
    },
    {
      "epoch": 1.2887889495663347,
      "grad_norm": 0.4649752676486969,
      "learning_rate": 6.274605210737792e-06,
      "loss": 0.3436,
      "step": 28084
    },
    {
      "epoch": 1.2888348400715892,
      "grad_norm": 0.4587714374065399,
      "learning_rate": 6.274368121847694e-06,
      "loss": 0.3281,
      "step": 28085
    },
    {
      "epoch": 1.2888807305768437,
      "grad_norm": 0.4453941285610199,
      "learning_rate": 6.274131029893142e-06,
      "loss": 0.3082,
      "step": 28086
    },
    {
      "epoch": 1.2889266210820982,
      "grad_norm": 0.4791351854801178,
      "learning_rate": 6.273893934874708e-06,
      "loss": 0.3766,
      "step": 28087
    },
    {
      "epoch": 1.2889725115873527,
      "grad_norm": 0.47649839520454407,
      "learning_rate": 6.27365683679296e-06,
      "loss": 0.3301,
      "step": 28088
    },
    {
      "epoch": 1.289018402092607,
      "grad_norm": 0.5120162963867188,
      "learning_rate": 6.273419735648471e-06,
      "loss": 0.3997,
      "step": 28089
    },
    {
      "epoch": 1.2890642925978615,
      "grad_norm": 0.4805338978767395,
      "learning_rate": 6.273182631441808e-06,
      "loss": 0.4045,
      "step": 28090
    },
    {
      "epoch": 1.289110183103116,
      "grad_norm": 0.5130795836448669,
      "learning_rate": 6.272945524173543e-06,
      "loss": 0.4084,
      "step": 28091
    },
    {
      "epoch": 1.2891560736083705,
      "grad_norm": 0.49978095293045044,
      "learning_rate": 6.272708413844246e-06,
      "loss": 0.4578,
      "step": 28092
    },
    {
      "epoch": 1.2892019641136248,
      "grad_norm": 0.5015215873718262,
      "learning_rate": 6.272471300454486e-06,
      "loss": 0.4575,
      "step": 28093
    },
    {
      "epoch": 1.2892478546188793,
      "grad_norm": 0.4563833773136139,
      "learning_rate": 6.272234184004835e-06,
      "loss": 0.3083,
      "step": 28094
    },
    {
      "epoch": 1.2892937451241337,
      "grad_norm": 0.4357316493988037,
      "learning_rate": 6.271997064495863e-06,
      "loss": 0.3064,
      "step": 28095
    },
    {
      "epoch": 1.2893396356293882,
      "grad_norm": 0.47616976499557495,
      "learning_rate": 6.271759941928137e-06,
      "loss": 0.4032,
      "step": 28096
    },
    {
      "epoch": 1.2893855261346427,
      "grad_norm": 0.46328607201576233,
      "learning_rate": 6.271522816302232e-06,
      "loss": 0.3417,
      "step": 28097
    },
    {
      "epoch": 1.2894314166398972,
      "grad_norm": 0.47418463230133057,
      "learning_rate": 6.271285687618716e-06,
      "loss": 0.39,
      "step": 28098
    },
    {
      "epoch": 1.2894773071451517,
      "grad_norm": 0.4340604841709137,
      "learning_rate": 6.271048555878159e-06,
      "loss": 0.3276,
      "step": 28099
    },
    {
      "epoch": 1.2895231976504062,
      "grad_norm": 0.4773349165916443,
      "learning_rate": 6.27081142108113e-06,
      "loss": 0.4498,
      "step": 28100
    },
    {
      "epoch": 1.2895690881556605,
      "grad_norm": 0.4368971288204193,
      "learning_rate": 6.270574283228201e-06,
      "loss": 0.2995,
      "step": 28101
    },
    {
      "epoch": 1.289614978660915,
      "grad_norm": 0.43238645792007446,
      "learning_rate": 6.270337142319943e-06,
      "loss": 0.3204,
      "step": 28102
    },
    {
      "epoch": 1.2896608691661695,
      "grad_norm": 0.4700338542461395,
      "learning_rate": 6.270099998356924e-06,
      "loss": 0.3588,
      "step": 28103
    },
    {
      "epoch": 1.289706759671424,
      "grad_norm": 0.5410152673721313,
      "learning_rate": 6.269862851339716e-06,
      "loss": 0.2912,
      "step": 28104
    },
    {
      "epoch": 1.2897526501766785,
      "grad_norm": 0.48020198941230774,
      "learning_rate": 6.269625701268889e-06,
      "loss": 0.3536,
      "step": 28105
    },
    {
      "epoch": 1.2897985406819328,
      "grad_norm": 0.6920510530471802,
      "learning_rate": 6.269388548145012e-06,
      "loss": 0.3465,
      "step": 28106
    },
    {
      "epoch": 1.2898444311871873,
      "grad_norm": 0.5102250576019287,
      "learning_rate": 6.2691513919686555e-06,
      "loss": 0.4095,
      "step": 28107
    },
    {
      "epoch": 1.2898903216924418,
      "grad_norm": 0.4701817035675049,
      "learning_rate": 6.2689142327403925e-06,
      "loss": 0.3207,
      "step": 28108
    },
    {
      "epoch": 1.2899362121976963,
      "grad_norm": 0.435823917388916,
      "learning_rate": 6.26867707046079e-06,
      "loss": 0.3429,
      "step": 28109
    },
    {
      "epoch": 1.2899821027029508,
      "grad_norm": 0.5534860491752625,
      "learning_rate": 6.268439905130419e-06,
      "loss": 0.3097,
      "step": 28110
    },
    {
      "epoch": 1.2900279932082053,
      "grad_norm": 0.5199395418167114,
      "learning_rate": 6.268202736749852e-06,
      "loss": 0.3592,
      "step": 28111
    },
    {
      "epoch": 1.2900738837134598,
      "grad_norm": 0.4815235733985901,
      "learning_rate": 6.267965565319656e-06,
      "loss": 0.3306,
      "step": 28112
    },
    {
      "epoch": 1.2901197742187143,
      "grad_norm": 0.446381151676178,
      "learning_rate": 6.267728390840405e-06,
      "loss": 0.2692,
      "step": 28113
    },
    {
      "epoch": 1.2901656647239685,
      "grad_norm": 0.5055413246154785,
      "learning_rate": 6.267491213312666e-06,
      "loss": 0.419,
      "step": 28114
    },
    {
      "epoch": 1.290211555229223,
      "grad_norm": 0.5072341561317444,
      "learning_rate": 6.267254032737011e-06,
      "loss": 0.3564,
      "step": 28115
    },
    {
      "epoch": 1.2902574457344775,
      "grad_norm": 0.4785483181476593,
      "learning_rate": 6.26701684911401e-06,
      "loss": 0.3798,
      "step": 28116
    },
    {
      "epoch": 1.290303336239732,
      "grad_norm": 0.504549503326416,
      "learning_rate": 6.266779662444232e-06,
      "loss": 0.3771,
      "step": 28117
    },
    {
      "epoch": 1.2903492267449865,
      "grad_norm": 0.4654318690299988,
      "learning_rate": 6.26654247272825e-06,
      "loss": 0.3245,
      "step": 28118
    },
    {
      "epoch": 1.2903951172502408,
      "grad_norm": 0.46297597885131836,
      "learning_rate": 6.266305279966634e-06,
      "loss": 0.3523,
      "step": 28119
    },
    {
      "epoch": 1.2904410077554953,
      "grad_norm": 0.45811977982521057,
      "learning_rate": 6.266068084159952e-06,
      "loss": 0.3425,
      "step": 28120
    },
    {
      "epoch": 1.2904868982607498,
      "grad_norm": 0.48623883724212646,
      "learning_rate": 6.265830885308776e-06,
      "loss": 0.3405,
      "step": 28121
    },
    {
      "epoch": 1.2905327887660043,
      "grad_norm": 0.4823985695838928,
      "learning_rate": 6.265593683413676e-06,
      "loss": 0.3867,
      "step": 28122
    },
    {
      "epoch": 1.2905786792712588,
      "grad_norm": 0.4745047092437744,
      "learning_rate": 6.265356478475223e-06,
      "loss": 0.4005,
      "step": 28123
    },
    {
      "epoch": 1.2906245697765133,
      "grad_norm": 0.49596139788627625,
      "learning_rate": 6.265119270493988e-06,
      "loss": 0.3793,
      "step": 28124
    },
    {
      "epoch": 1.2906704602817678,
      "grad_norm": 0.4973370134830475,
      "learning_rate": 6.26488205947054e-06,
      "loss": 0.4494,
      "step": 28125
    },
    {
      "epoch": 1.2907163507870223,
      "grad_norm": 0.4480249285697937,
      "learning_rate": 6.2646448454054485e-06,
      "loss": 0.3388,
      "step": 28126
    },
    {
      "epoch": 1.2907622412922766,
      "grad_norm": 0.5113403797149658,
      "learning_rate": 6.264407628299286e-06,
      "loss": 0.4539,
      "step": 28127
    },
    {
      "epoch": 1.290808131797531,
      "grad_norm": 0.4826340973377228,
      "learning_rate": 6.264170408152623e-06,
      "loss": 0.3484,
      "step": 28128
    },
    {
      "epoch": 1.2908540223027856,
      "grad_norm": 0.4176933467388153,
      "learning_rate": 6.263933184966028e-06,
      "loss": 0.2705,
      "step": 28129
    },
    {
      "epoch": 1.29089991280804,
      "grad_norm": 0.4535112679004669,
      "learning_rate": 6.263695958740074e-06,
      "loss": 0.3543,
      "step": 28130
    },
    {
      "epoch": 1.2909458033132946,
      "grad_norm": 0.5444042086601257,
      "learning_rate": 6.263458729475329e-06,
      "loss": 0.4367,
      "step": 28131
    },
    {
      "epoch": 1.2909916938185488,
      "grad_norm": 0.4246787428855896,
      "learning_rate": 6.263221497172365e-06,
      "loss": 0.2671,
      "step": 28132
    },
    {
      "epoch": 1.2910375843238033,
      "grad_norm": 0.4835907518863678,
      "learning_rate": 6.262984261831751e-06,
      "loss": 0.3489,
      "step": 28133
    },
    {
      "epoch": 1.2910834748290578,
      "grad_norm": 0.47306519746780396,
      "learning_rate": 6.262747023454059e-06,
      "loss": 0.33,
      "step": 28134
    },
    {
      "epoch": 1.2911293653343123,
      "grad_norm": 0.4270380437374115,
      "learning_rate": 6.2625097820398596e-06,
      "loss": 0.2835,
      "step": 28135
    },
    {
      "epoch": 1.2911752558395668,
      "grad_norm": 0.46325618028640747,
      "learning_rate": 6.262272537589723e-06,
      "loss": 0.3682,
      "step": 28136
    },
    {
      "epoch": 1.2912211463448213,
      "grad_norm": 0.4826434254646301,
      "learning_rate": 6.262035290104218e-06,
      "loss": 0.4039,
      "step": 28137
    },
    {
      "epoch": 1.2912670368500758,
      "grad_norm": 0.45712363719940186,
      "learning_rate": 6.261798039583916e-06,
      "loss": 0.3793,
      "step": 28138
    },
    {
      "epoch": 1.2913129273553303,
      "grad_norm": 0.4697333872318268,
      "learning_rate": 6.261560786029389e-06,
      "loss": 0.3943,
      "step": 28139
    },
    {
      "epoch": 1.2913588178605846,
      "grad_norm": 0.5037807822227478,
      "learning_rate": 6.261323529441206e-06,
      "loss": 0.4085,
      "step": 28140
    },
    {
      "epoch": 1.291404708365839,
      "grad_norm": 0.4362018406391144,
      "learning_rate": 6.261086269819939e-06,
      "loss": 0.3536,
      "step": 28141
    },
    {
      "epoch": 1.2914505988710936,
      "grad_norm": 0.46879464387893677,
      "learning_rate": 6.260849007166157e-06,
      "loss": 0.3829,
      "step": 28142
    },
    {
      "epoch": 1.291496489376348,
      "grad_norm": 0.4510067105293274,
      "learning_rate": 6.260611741480431e-06,
      "loss": 0.3328,
      "step": 28143
    },
    {
      "epoch": 1.2915423798816024,
      "grad_norm": 0.4698733687400818,
      "learning_rate": 6.260374472763331e-06,
      "loss": 0.3529,
      "step": 28144
    },
    {
      "epoch": 1.2915882703868569,
      "grad_norm": 0.4291249215602875,
      "learning_rate": 6.26013720101543e-06,
      "loss": 0.3013,
      "step": 28145
    },
    {
      "epoch": 1.2916341608921114,
      "grad_norm": 0.46072617173194885,
      "learning_rate": 6.259899926237295e-06,
      "loss": 0.3379,
      "step": 28146
    },
    {
      "epoch": 1.2916800513973659,
      "grad_norm": 0.485603004693985,
      "learning_rate": 6.259662648429499e-06,
      "loss": 0.3651,
      "step": 28147
    },
    {
      "epoch": 1.2917259419026204,
      "grad_norm": 0.4442216157913208,
      "learning_rate": 6.2594253675926105e-06,
      "loss": 0.2849,
      "step": 28148
    },
    {
      "epoch": 1.2917718324078749,
      "grad_norm": 0.4689701795578003,
      "learning_rate": 6.259188083727203e-06,
      "loss": 0.3912,
      "step": 28149
    },
    {
      "epoch": 1.2918177229131294,
      "grad_norm": 0.4457116425037384,
      "learning_rate": 6.258950796833846e-06,
      "loss": 0.3408,
      "step": 28150
    },
    {
      "epoch": 1.2918636134183838,
      "grad_norm": 0.43954628705978394,
      "learning_rate": 6.258713506913108e-06,
      "loss": 0.3076,
      "step": 28151
    },
    {
      "epoch": 1.2919095039236381,
      "grad_norm": 0.4560212790966034,
      "learning_rate": 6.258476213965563e-06,
      "loss": 0.3593,
      "step": 28152
    },
    {
      "epoch": 1.2919553944288926,
      "grad_norm": 0.44616764783859253,
      "learning_rate": 6.258238917991779e-06,
      "loss": 0.3094,
      "step": 28153
    },
    {
      "epoch": 1.2920012849341471,
      "grad_norm": 0.46181008219718933,
      "learning_rate": 6.258001618992328e-06,
      "loss": 0.3503,
      "step": 28154
    },
    {
      "epoch": 1.2920471754394016,
      "grad_norm": 0.4951649606227875,
      "learning_rate": 6.2577643169677785e-06,
      "loss": 0.4082,
      "step": 28155
    },
    {
      "epoch": 1.2920930659446561,
      "grad_norm": 0.4375470280647278,
      "learning_rate": 6.257527011918705e-06,
      "loss": 0.3185,
      "step": 28156
    },
    {
      "epoch": 1.2921389564499104,
      "grad_norm": 0.4763641655445099,
      "learning_rate": 6.257289703845675e-06,
      "loss": 0.3792,
      "step": 28157
    },
    {
      "epoch": 1.292184846955165,
      "grad_norm": 0.4842449724674225,
      "learning_rate": 6.257052392749259e-06,
      "loss": 0.3769,
      "step": 28158
    },
    {
      "epoch": 1.2922307374604194,
      "grad_norm": 0.5028340220451355,
      "learning_rate": 6.2568150786300305e-06,
      "loss": 0.4051,
      "step": 28159
    },
    {
      "epoch": 1.2922766279656739,
      "grad_norm": 0.499295711517334,
      "learning_rate": 6.256577761488557e-06,
      "loss": 0.3924,
      "step": 28160
    },
    {
      "epoch": 1.2923225184709284,
      "grad_norm": 0.4599973261356354,
      "learning_rate": 6.25634044132541e-06,
      "loss": 0.3195,
      "step": 28161
    },
    {
      "epoch": 1.2923684089761829,
      "grad_norm": 0.49181047081947327,
      "learning_rate": 6.256103118141164e-06,
      "loss": 0.3546,
      "step": 28162
    },
    {
      "epoch": 1.2924142994814374,
      "grad_norm": 0.4558887183666229,
      "learning_rate": 6.255865791936384e-06,
      "loss": 0.3421,
      "step": 28163
    },
    {
      "epoch": 1.2924601899866919,
      "grad_norm": 0.49994340538978577,
      "learning_rate": 6.255628462711642e-06,
      "loss": 0.3873,
      "step": 28164
    },
    {
      "epoch": 1.2925060804919462,
      "grad_norm": 0.47716203331947327,
      "learning_rate": 6.255391130467513e-06,
      "loss": 0.3733,
      "step": 28165
    },
    {
      "epoch": 1.2925519709972006,
      "grad_norm": 0.4969201683998108,
      "learning_rate": 6.255153795204563e-06,
      "loss": 0.4054,
      "step": 28166
    },
    {
      "epoch": 1.2925978615024551,
      "grad_norm": 0.4709016680717468,
      "learning_rate": 6.254916456923364e-06,
      "loss": 0.3499,
      "step": 28167
    },
    {
      "epoch": 1.2926437520077096,
      "grad_norm": 0.44989922642707825,
      "learning_rate": 6.254679115624488e-06,
      "loss": 0.3179,
      "step": 28168
    },
    {
      "epoch": 1.2926896425129641,
      "grad_norm": 0.49080345034599304,
      "learning_rate": 6.254441771308503e-06,
      "loss": 0.3675,
      "step": 28169
    },
    {
      "epoch": 1.2927355330182184,
      "grad_norm": 0.444807231426239,
      "learning_rate": 6.2542044239759825e-06,
      "loss": 0.3354,
      "step": 28170
    },
    {
      "epoch": 1.292781423523473,
      "grad_norm": 0.44433027505874634,
      "learning_rate": 6.253967073627496e-06,
      "loss": 0.3119,
      "step": 28171
    },
    {
      "epoch": 1.2928273140287274,
      "grad_norm": 0.5566893815994263,
      "learning_rate": 6.253729720263614e-06,
      "loss": 0.4518,
      "step": 28172
    },
    {
      "epoch": 1.292873204533982,
      "grad_norm": 0.48656103014945984,
      "learning_rate": 6.253492363884908e-06,
      "loss": 0.3435,
      "step": 28173
    },
    {
      "epoch": 1.2929190950392364,
      "grad_norm": 0.4757208526134491,
      "learning_rate": 6.25325500449195e-06,
      "loss": 0.3762,
      "step": 28174
    },
    {
      "epoch": 1.292964985544491,
      "grad_norm": 0.4750703275203705,
      "learning_rate": 6.253017642085306e-06,
      "loss": 0.3832,
      "step": 28175
    },
    {
      "epoch": 1.2930108760497454,
      "grad_norm": 0.507757842540741,
      "learning_rate": 6.252780276665553e-06,
      "loss": 0.3553,
      "step": 28176
    },
    {
      "epoch": 1.293056766555,
      "grad_norm": 0.4610876739025116,
      "learning_rate": 6.252542908233257e-06,
      "loss": 0.3333,
      "step": 28177
    },
    {
      "epoch": 1.2931026570602542,
      "grad_norm": 0.448053240776062,
      "learning_rate": 6.252305536788992e-06,
      "loss": 0.3208,
      "step": 28178
    },
    {
      "epoch": 1.2931485475655087,
      "grad_norm": 0.47671446204185486,
      "learning_rate": 6.252068162333328e-06,
      "loss": 0.3969,
      "step": 28179
    },
    {
      "epoch": 1.2931944380707632,
      "grad_norm": 0.484769731760025,
      "learning_rate": 6.251830784866832e-06,
      "loss": 0.4313,
      "step": 28180
    },
    {
      "epoch": 1.2932403285760177,
      "grad_norm": 0.4288400709629059,
      "learning_rate": 6.2515934043900805e-06,
      "loss": 0.3216,
      "step": 28181
    },
    {
      "epoch": 1.293286219081272,
      "grad_norm": 0.48768994212150574,
      "learning_rate": 6.251356020903642e-06,
      "loss": 0.393,
      "step": 28182
    },
    {
      "epoch": 1.2933321095865264,
      "grad_norm": 0.4802165627479553,
      "learning_rate": 6.251118634408087e-06,
      "loss": 0.3804,
      "step": 28183
    },
    {
      "epoch": 1.293378000091781,
      "grad_norm": 0.4553995132446289,
      "learning_rate": 6.250881244903986e-06,
      "loss": 0.3245,
      "step": 28184
    },
    {
      "epoch": 1.2934238905970354,
      "grad_norm": 0.5096217393875122,
      "learning_rate": 6.250643852391911e-06,
      "loss": 0.4705,
      "step": 28185
    },
    {
      "epoch": 1.29346978110229,
      "grad_norm": 0.4675668179988861,
      "learning_rate": 6.250406456872431e-06,
      "loss": 0.3645,
      "step": 28186
    },
    {
      "epoch": 1.2935156716075444,
      "grad_norm": 0.44361206889152527,
      "learning_rate": 6.25016905834612e-06,
      "loss": 0.3019,
      "step": 28187
    },
    {
      "epoch": 1.293561562112799,
      "grad_norm": 0.4849269688129425,
      "learning_rate": 6.249931656813545e-06,
      "loss": 0.338,
      "step": 28188
    },
    {
      "epoch": 1.2936074526180534,
      "grad_norm": 0.4396539330482483,
      "learning_rate": 6.24969425227528e-06,
      "loss": 0.3303,
      "step": 28189
    },
    {
      "epoch": 1.2936533431233077,
      "grad_norm": 0.4313536584377289,
      "learning_rate": 6.2494568447318955e-06,
      "loss": 0.3233,
      "step": 28190
    },
    {
      "epoch": 1.2936992336285622,
      "grad_norm": 0.46855443716049194,
      "learning_rate": 6.24921943418396e-06,
      "loss": 0.3187,
      "step": 28191
    },
    {
      "epoch": 1.2937451241338167,
      "grad_norm": 0.4327333867549896,
      "learning_rate": 6.248982020632046e-06,
      "loss": 0.299,
      "step": 28192
    },
    {
      "epoch": 1.2937910146390712,
      "grad_norm": 0.4876275956630707,
      "learning_rate": 6.248744604076726e-06,
      "loss": 0.4235,
      "step": 28193
    },
    {
      "epoch": 1.2938369051443257,
      "grad_norm": 0.5048226118087769,
      "learning_rate": 6.248507184518567e-06,
      "loss": 0.3805,
      "step": 28194
    },
    {
      "epoch": 1.29388279564958,
      "grad_norm": 0.4671584665775299,
      "learning_rate": 6.248269761958146e-06,
      "loss": 0.3041,
      "step": 28195
    },
    {
      "epoch": 1.2939286861548345,
      "grad_norm": 0.4680176377296448,
      "learning_rate": 6.2480323363960274e-06,
      "loss": 0.3695,
      "step": 28196
    },
    {
      "epoch": 1.293974576660089,
      "grad_norm": 0.4423791468143463,
      "learning_rate": 6.247794907832785e-06,
      "loss": 0.3188,
      "step": 28197
    },
    {
      "epoch": 1.2940204671653435,
      "grad_norm": 0.46270719170570374,
      "learning_rate": 6.24755747626899e-06,
      "loss": 0.3352,
      "step": 28198
    },
    {
      "epoch": 1.294066357670598,
      "grad_norm": 0.4656553864479065,
      "learning_rate": 6.247320041705214e-06,
      "loss": 0.3331,
      "step": 28199
    },
    {
      "epoch": 1.2941122481758525,
      "grad_norm": 0.46538588404655457,
      "learning_rate": 6.247082604142026e-06,
      "loss": 0.3587,
      "step": 28200
    },
    {
      "epoch": 1.294158138681107,
      "grad_norm": 0.5083321332931519,
      "learning_rate": 6.246845163579999e-06,
      "loss": 0.4305,
      "step": 28201
    },
    {
      "epoch": 1.2942040291863615,
      "grad_norm": 0.5457224249839783,
      "learning_rate": 6.246607720019701e-06,
      "loss": 0.5146,
      "step": 28202
    },
    {
      "epoch": 1.2942499196916157,
      "grad_norm": 0.5039021372795105,
      "learning_rate": 6.246370273461705e-06,
      "loss": 0.4287,
      "step": 28203
    },
    {
      "epoch": 1.2942958101968702,
      "grad_norm": 0.47966036200523376,
      "learning_rate": 6.246132823906583e-06,
      "loss": 0.3393,
      "step": 28204
    },
    {
      "epoch": 1.2943417007021247,
      "grad_norm": 0.4261705279350281,
      "learning_rate": 6.245895371354904e-06,
      "loss": 0.2795,
      "step": 28205
    },
    {
      "epoch": 1.2943875912073792,
      "grad_norm": 0.47341057658195496,
      "learning_rate": 6.24565791580724e-06,
      "loss": 0.3718,
      "step": 28206
    },
    {
      "epoch": 1.2944334817126337,
      "grad_norm": 0.47819915413856506,
      "learning_rate": 6.2454204572641645e-06,
      "loss": 0.3834,
      "step": 28207
    },
    {
      "epoch": 1.294479372217888,
      "grad_norm": 0.47994464635849,
      "learning_rate": 6.245182995726242e-06,
      "loss": 0.4399,
      "step": 28208
    },
    {
      "epoch": 1.2945252627231425,
      "grad_norm": 0.4513194262981415,
      "learning_rate": 6.244945531194049e-06,
      "loss": 0.3292,
      "step": 28209
    },
    {
      "epoch": 1.294571153228397,
      "grad_norm": 0.47888657450675964,
      "learning_rate": 6.244708063668155e-06,
      "loss": 0.4301,
      "step": 28210
    },
    {
      "epoch": 1.2946170437336515,
      "grad_norm": 0.4408484101295471,
      "learning_rate": 6.244470593149131e-06,
      "loss": 0.3045,
      "step": 28211
    },
    {
      "epoch": 1.294662934238906,
      "grad_norm": 0.49501144886016846,
      "learning_rate": 6.244233119637547e-06,
      "loss": 0.4154,
      "step": 28212
    },
    {
      "epoch": 1.2947088247441605,
      "grad_norm": 0.4493750035762787,
      "learning_rate": 6.243995643133976e-06,
      "loss": 0.3593,
      "step": 28213
    },
    {
      "epoch": 1.294754715249415,
      "grad_norm": 0.5314096212387085,
      "learning_rate": 6.2437581636389875e-06,
      "loss": 0.5131,
      "step": 28214
    },
    {
      "epoch": 1.2948006057546695,
      "grad_norm": 0.4515240490436554,
      "learning_rate": 6.243520681153155e-06,
      "loss": 0.295,
      "step": 28215
    },
    {
      "epoch": 1.2948464962599238,
      "grad_norm": 0.47085344791412354,
      "learning_rate": 6.2432831956770454e-06,
      "loss": 0.3808,
      "step": 28216
    },
    {
      "epoch": 1.2948923867651783,
      "grad_norm": 0.4392392039299011,
      "learning_rate": 6.243045707211234e-06,
      "loss": 0.3087,
      "step": 28217
    },
    {
      "epoch": 1.2949382772704328,
      "grad_norm": 0.5048328042030334,
      "learning_rate": 6.242808215756288e-06,
      "loss": 0.4668,
      "step": 28218
    },
    {
      "epoch": 1.2949841677756873,
      "grad_norm": 0.4718846082687378,
      "learning_rate": 6.2425707213127815e-06,
      "loss": 0.3438,
      "step": 28219
    },
    {
      "epoch": 1.2950300582809418,
      "grad_norm": 0.47589579224586487,
      "learning_rate": 6.242333223881285e-06,
      "loss": 0.3538,
      "step": 28220
    },
    {
      "epoch": 1.295075948786196,
      "grad_norm": 0.4690905809402466,
      "learning_rate": 6.242095723462369e-06,
      "loss": 0.3895,
      "step": 28221
    },
    {
      "epoch": 1.2951218392914505,
      "grad_norm": 0.4229196608066559,
      "learning_rate": 6.241858220056605e-06,
      "loss": 0.2873,
      "step": 28222
    },
    {
      "epoch": 1.295167729796705,
      "grad_norm": 0.48218896985054016,
      "learning_rate": 6.241620713664562e-06,
      "loss": 0.3911,
      "step": 28223
    },
    {
      "epoch": 1.2952136203019595,
      "grad_norm": 0.48509618639945984,
      "learning_rate": 6.241383204286816e-06,
      "loss": 0.3555,
      "step": 28224
    },
    {
      "epoch": 1.295259510807214,
      "grad_norm": 0.4459286630153656,
      "learning_rate": 6.2411456919239335e-06,
      "loss": 0.2965,
      "step": 28225
    },
    {
      "epoch": 1.2953054013124685,
      "grad_norm": 0.46515509486198425,
      "learning_rate": 6.240908176576487e-06,
      "loss": 0.355,
      "step": 28226
    },
    {
      "epoch": 1.295351291817723,
      "grad_norm": 0.5120500326156616,
      "learning_rate": 6.240670658245049e-06,
      "loss": 0.4591,
      "step": 28227
    },
    {
      "epoch": 1.2953971823229775,
      "grad_norm": 0.5213598608970642,
      "learning_rate": 6.24043313693019e-06,
      "loss": 0.4478,
      "step": 28228
    },
    {
      "epoch": 1.2954430728282318,
      "grad_norm": 0.5212617516517639,
      "learning_rate": 6.240195612632478e-06,
      "loss": 0.4635,
      "step": 28229
    },
    {
      "epoch": 1.2954889633334863,
      "grad_norm": 0.47674497961997986,
      "learning_rate": 6.23995808535249e-06,
      "loss": 0.3728,
      "step": 28230
    },
    {
      "epoch": 1.2955348538387408,
      "grad_norm": 0.47173964977264404,
      "learning_rate": 6.2397205550907925e-06,
      "loss": 0.381,
      "step": 28231
    },
    {
      "epoch": 1.2955807443439953,
      "grad_norm": 0.47051575779914856,
      "learning_rate": 6.23948302184796e-06,
      "loss": 0.3345,
      "step": 28232
    },
    {
      "epoch": 1.2956266348492496,
      "grad_norm": 0.42756760120391846,
      "learning_rate": 6.239245485624561e-06,
      "loss": 0.2984,
      "step": 28233
    },
    {
      "epoch": 1.295672525354504,
      "grad_norm": 0.49231743812561035,
      "learning_rate": 6.239007946421167e-06,
      "loss": 0.389,
      "step": 28234
    },
    {
      "epoch": 1.2957184158597586,
      "grad_norm": 0.45892927050590515,
      "learning_rate": 6.238770404238351e-06,
      "loss": 0.344,
      "step": 28235
    },
    {
      "epoch": 1.295764306365013,
      "grad_norm": 0.46634969115257263,
      "learning_rate": 6.238532859076682e-06,
      "loss": 0.2992,
      "step": 28236
    },
    {
      "epoch": 1.2958101968702675,
      "grad_norm": 0.45209044218063354,
      "learning_rate": 6.238295310936733e-06,
      "loss": 0.3137,
      "step": 28237
    },
    {
      "epoch": 1.295856087375522,
      "grad_norm": 0.48368358612060547,
      "learning_rate": 6.238057759819075e-06,
      "loss": 0.4326,
      "step": 28238
    },
    {
      "epoch": 1.2959019778807765,
      "grad_norm": 0.48004165291786194,
      "learning_rate": 6.2378202057242785e-06,
      "loss": 0.3677,
      "step": 28239
    },
    {
      "epoch": 1.295947868386031,
      "grad_norm": 0.4553161561489105,
      "learning_rate": 6.237582648652916e-06,
      "loss": 0.3316,
      "step": 28240
    },
    {
      "epoch": 1.2959937588912853,
      "grad_norm": 0.4572404623031616,
      "learning_rate": 6.237345088605556e-06,
      "loss": 0.2856,
      "step": 28241
    },
    {
      "epoch": 1.2960396493965398,
      "grad_norm": 0.48816996812820435,
      "learning_rate": 6.237107525582773e-06,
      "loss": 0.385,
      "step": 28242
    },
    {
      "epoch": 1.2960855399017943,
      "grad_norm": 0.48697546124458313,
      "learning_rate": 6.236869959585136e-06,
      "loss": 0.3783,
      "step": 28243
    },
    {
      "epoch": 1.2961314304070488,
      "grad_norm": 0.4897775948047638,
      "learning_rate": 6.2366323906132185e-06,
      "loss": 0.413,
      "step": 28244
    },
    {
      "epoch": 1.2961773209123033,
      "grad_norm": 0.468680739402771,
      "learning_rate": 6.236394818667588e-06,
      "loss": 0.3582,
      "step": 28245
    },
    {
      "epoch": 1.2962232114175576,
      "grad_norm": 0.473310261964798,
      "learning_rate": 6.236157243748819e-06,
      "loss": 0.3965,
      "step": 28246
    },
    {
      "epoch": 1.296269101922812,
      "grad_norm": 0.44930437207221985,
      "learning_rate": 6.235919665857484e-06,
      "loss": 0.3288,
      "step": 28247
    },
    {
      "epoch": 1.2963149924280666,
      "grad_norm": 0.4284410774707794,
      "learning_rate": 6.23568208499415e-06,
      "loss": 0.3321,
      "step": 28248
    },
    {
      "epoch": 1.296360882933321,
      "grad_norm": 0.498472660779953,
      "learning_rate": 6.235444501159391e-06,
      "loss": 0.3642,
      "step": 28249
    },
    {
      "epoch": 1.2964067734385756,
      "grad_norm": 0.4493360221385956,
      "learning_rate": 6.23520691435378e-06,
      "loss": 0.3786,
      "step": 28250
    },
    {
      "epoch": 1.29645266394383,
      "grad_norm": 0.4264218509197235,
      "learning_rate": 6.234969324577883e-06,
      "loss": 0.2964,
      "step": 28251
    },
    {
      "epoch": 1.2964985544490846,
      "grad_norm": 0.4268330931663513,
      "learning_rate": 6.2347317318322774e-06,
      "loss": 0.3447,
      "step": 28252
    },
    {
      "epoch": 1.296544444954339,
      "grad_norm": 0.44102004170417786,
      "learning_rate": 6.234494136117529e-06,
      "loss": 0.3017,
      "step": 28253
    },
    {
      "epoch": 1.2965903354595933,
      "grad_norm": 0.4497438669204712,
      "learning_rate": 6.234256537434212e-06,
      "loss": 0.3793,
      "step": 28254
    },
    {
      "epoch": 1.2966362259648478,
      "grad_norm": 0.44043946266174316,
      "learning_rate": 6.2340189357829e-06,
      "loss": 0.3284,
      "step": 28255
    },
    {
      "epoch": 1.2966821164701023,
      "grad_norm": 0.4701074957847595,
      "learning_rate": 6.233781331164161e-06,
      "loss": 0.3397,
      "step": 28256
    },
    {
      "epoch": 1.2967280069753568,
      "grad_norm": 0.46878552436828613,
      "learning_rate": 6.233543723578567e-06,
      "loss": 0.3204,
      "step": 28257
    },
    {
      "epoch": 1.2967738974806113,
      "grad_norm": 0.516370415687561,
      "learning_rate": 6.233306113026689e-06,
      "loss": 0.4495,
      "step": 28258
    },
    {
      "epoch": 1.2968197879858656,
      "grad_norm": 0.5256620645523071,
      "learning_rate": 6.2330684995091e-06,
      "loss": 0.4481,
      "step": 28259
    },
    {
      "epoch": 1.29686567849112,
      "grad_norm": 0.45697155594825745,
      "learning_rate": 6.232830883026369e-06,
      "loss": 0.318,
      "step": 28260
    },
    {
      "epoch": 1.2969115689963746,
      "grad_norm": 0.4993169605731964,
      "learning_rate": 6.232593263579071e-06,
      "loss": 0.4147,
      "step": 28261
    },
    {
      "epoch": 1.296957459501629,
      "grad_norm": 0.4609990119934082,
      "learning_rate": 6.232355641167773e-06,
      "loss": 0.3249,
      "step": 28262
    },
    {
      "epoch": 1.2970033500068836,
      "grad_norm": 0.479338675737381,
      "learning_rate": 6.23211801579305e-06,
      "loss": 0.3672,
      "step": 28263
    },
    {
      "epoch": 1.297049240512138,
      "grad_norm": 0.47226232290267944,
      "learning_rate": 6.231880387455472e-06,
      "loss": 0.3919,
      "step": 28264
    },
    {
      "epoch": 1.2970951310173926,
      "grad_norm": 0.4605559706687927,
      "learning_rate": 6.2316427561556095e-06,
      "loss": 0.3614,
      "step": 28265
    },
    {
      "epoch": 1.297141021522647,
      "grad_norm": 0.4790906310081482,
      "learning_rate": 6.231405121894038e-06,
      "loss": 0.4085,
      "step": 28266
    },
    {
      "epoch": 1.2971869120279014,
      "grad_norm": 0.6127444505691528,
      "learning_rate": 6.231167484671322e-06,
      "loss": 0.4463,
      "step": 28267
    },
    {
      "epoch": 1.2972328025331559,
      "grad_norm": 0.4843725562095642,
      "learning_rate": 6.2309298444880385e-06,
      "loss": 0.3586,
      "step": 28268
    },
    {
      "epoch": 1.2972786930384104,
      "grad_norm": 0.4457184374332428,
      "learning_rate": 6.230692201344757e-06,
      "loss": 0.324,
      "step": 28269
    },
    {
      "epoch": 1.2973245835436649,
      "grad_norm": 0.4672364890575409,
      "learning_rate": 6.2304545552420485e-06,
      "loss": 0.3241,
      "step": 28270
    },
    {
      "epoch": 1.2973704740489191,
      "grad_norm": 0.502672016620636,
      "learning_rate": 6.230216906180486e-06,
      "loss": 0.3669,
      "step": 28271
    },
    {
      "epoch": 1.2974163645541736,
      "grad_norm": 0.46793845295906067,
      "learning_rate": 6.229979254160641e-06,
      "loss": 0.3768,
      "step": 28272
    },
    {
      "epoch": 1.2974622550594281,
      "grad_norm": 0.44317319989204407,
      "learning_rate": 6.229741599183082e-06,
      "loss": 0.2986,
      "step": 28273
    },
    {
      "epoch": 1.2975081455646826,
      "grad_norm": 0.4593021869659424,
      "learning_rate": 6.229503941248382e-06,
      "loss": 0.3457,
      "step": 28274
    },
    {
      "epoch": 1.2975540360699371,
      "grad_norm": 0.45081472396850586,
      "learning_rate": 6.229266280357116e-06,
      "loss": 0.3153,
      "step": 28275
    },
    {
      "epoch": 1.2975999265751916,
      "grad_norm": 0.4700773358345032,
      "learning_rate": 6.22902861650985e-06,
      "loss": 0.357,
      "step": 28276
    },
    {
      "epoch": 1.2976458170804461,
      "grad_norm": 0.48508551716804504,
      "learning_rate": 6.2287909497071595e-06,
      "loss": 0.3732,
      "step": 28277
    },
    {
      "epoch": 1.2976917075857006,
      "grad_norm": 0.434691846370697,
      "learning_rate": 6.228553279949615e-06,
      "loss": 0.3188,
      "step": 28278
    },
    {
      "epoch": 1.297737598090955,
      "grad_norm": 0.4558145999908447,
      "learning_rate": 6.2283156072377856e-06,
      "loss": 0.3531,
      "step": 28279
    },
    {
      "epoch": 1.2977834885962094,
      "grad_norm": 0.4659963846206665,
      "learning_rate": 6.228077931572247e-06,
      "loss": 0.3821,
      "step": 28280
    },
    {
      "epoch": 1.297829379101464,
      "grad_norm": 0.42511171102523804,
      "learning_rate": 6.2278402529535675e-06,
      "loss": 0.2894,
      "step": 28281
    },
    {
      "epoch": 1.2978752696067184,
      "grad_norm": 0.4993799030780792,
      "learning_rate": 6.22760257138232e-06,
      "loss": 0.4487,
      "step": 28282
    },
    {
      "epoch": 1.297921160111973,
      "grad_norm": 0.45986226201057434,
      "learning_rate": 6.227364886859077e-06,
      "loss": 0.3467,
      "step": 28283
    },
    {
      "epoch": 1.2979670506172272,
      "grad_norm": 0.4547714591026306,
      "learning_rate": 6.2271271993844066e-06,
      "loss": 0.3359,
      "step": 28284
    },
    {
      "epoch": 1.2980129411224817,
      "grad_norm": 0.474901407957077,
      "learning_rate": 6.226889508958882e-06,
      "loss": 0.3685,
      "step": 28285
    },
    {
      "epoch": 1.2980588316277362,
      "grad_norm": 0.41690266132354736,
      "learning_rate": 6.226651815583078e-06,
      "loss": 0.271,
      "step": 28286
    },
    {
      "epoch": 1.2981047221329907,
      "grad_norm": 0.4565311074256897,
      "learning_rate": 6.226414119257561e-06,
      "loss": 0.339,
      "step": 28287
    },
    {
      "epoch": 1.2981506126382452,
      "grad_norm": 0.4872124493122101,
      "learning_rate": 6.226176419982907e-06,
      "loss": 0.4063,
      "step": 28288
    },
    {
      "epoch": 1.2981965031434997,
      "grad_norm": 0.4359271228313446,
      "learning_rate": 6.225938717759685e-06,
      "loss": 0.3107,
      "step": 28289
    },
    {
      "epoch": 1.2982423936487542,
      "grad_norm": 0.47329047322273254,
      "learning_rate": 6.225701012588468e-06,
      "loss": 0.341,
      "step": 28290
    },
    {
      "epoch": 1.2982882841540087,
      "grad_norm": 0.4435488283634186,
      "learning_rate": 6.225463304469825e-06,
      "loss": 0.2882,
      "step": 28291
    },
    {
      "epoch": 1.298334174659263,
      "grad_norm": 0.47545769810676575,
      "learning_rate": 6.225225593404331e-06,
      "loss": 0.3836,
      "step": 28292
    },
    {
      "epoch": 1.2983800651645174,
      "grad_norm": 0.4938840866088867,
      "learning_rate": 6.224987879392557e-06,
      "loss": 0.4022,
      "step": 28293
    },
    {
      "epoch": 1.298425955669772,
      "grad_norm": 0.4281056821346283,
      "learning_rate": 6.224750162435073e-06,
      "loss": 0.2745,
      "step": 28294
    },
    {
      "epoch": 1.2984718461750264,
      "grad_norm": 0.44834575057029724,
      "learning_rate": 6.224512442532451e-06,
      "loss": 0.359,
      "step": 28295
    },
    {
      "epoch": 1.298517736680281,
      "grad_norm": 0.4557170867919922,
      "learning_rate": 6.2242747196852625e-06,
      "loss": 0.3059,
      "step": 28296
    },
    {
      "epoch": 1.2985636271855352,
      "grad_norm": 0.5180224180221558,
      "learning_rate": 6.22403699389408e-06,
      "loss": 0.3492,
      "step": 28297
    },
    {
      "epoch": 1.2986095176907897,
      "grad_norm": 0.4579494595527649,
      "learning_rate": 6.223799265159476e-06,
      "loss": 0.3378,
      "step": 28298
    },
    {
      "epoch": 1.2986554081960442,
      "grad_norm": 0.4819929301738739,
      "learning_rate": 6.2235615334820195e-06,
      "loss": 0.3668,
      "step": 28299
    },
    {
      "epoch": 1.2987012987012987,
      "grad_norm": 0.46763843297958374,
      "learning_rate": 6.223323798862285e-06,
      "loss": 0.3574,
      "step": 28300
    },
    {
      "epoch": 1.2987471892065532,
      "grad_norm": 0.4747743308544159,
      "learning_rate": 6.223086061300842e-06,
      "loss": 0.403,
      "step": 28301
    },
    {
      "epoch": 1.2987930797118077,
      "grad_norm": 0.44085994362831116,
      "learning_rate": 6.222848320798262e-06,
      "loss": 0.3486,
      "step": 28302
    },
    {
      "epoch": 1.2988389702170622,
      "grad_norm": 0.5846200585365295,
      "learning_rate": 6.22261057735512e-06,
      "loss": 0.408,
      "step": 28303
    },
    {
      "epoch": 1.2988848607223167,
      "grad_norm": 0.48225364089012146,
      "learning_rate": 6.222372830971985e-06,
      "loss": 0.3845,
      "step": 28304
    },
    {
      "epoch": 1.298930751227571,
      "grad_norm": 0.5132493376731873,
      "learning_rate": 6.222135081649428e-06,
      "loss": 0.3747,
      "step": 28305
    },
    {
      "epoch": 1.2989766417328255,
      "grad_norm": 0.4257825016975403,
      "learning_rate": 6.221897329388024e-06,
      "loss": 0.2875,
      "step": 28306
    },
    {
      "epoch": 1.29902253223808,
      "grad_norm": 0.4708353579044342,
      "learning_rate": 6.2216595741883405e-06,
      "loss": 0.374,
      "step": 28307
    },
    {
      "epoch": 1.2990684227433344,
      "grad_norm": 0.49343109130859375,
      "learning_rate": 6.221421816050951e-06,
      "loss": 0.371,
      "step": 28308
    },
    {
      "epoch": 1.299114313248589,
      "grad_norm": 0.4671335518360138,
      "learning_rate": 6.221184054976428e-06,
      "loss": 0.35,
      "step": 28309
    },
    {
      "epoch": 1.2991602037538432,
      "grad_norm": 0.4626711308956146,
      "learning_rate": 6.220946290965344e-06,
      "loss": 0.3825,
      "step": 28310
    },
    {
      "epoch": 1.2992060942590977,
      "grad_norm": 0.45057567954063416,
      "learning_rate": 6.220708524018268e-06,
      "loss": 0.3183,
      "step": 28311
    },
    {
      "epoch": 1.2992519847643522,
      "grad_norm": 0.47545817494392395,
      "learning_rate": 6.220470754135775e-06,
      "loss": 0.3945,
      "step": 28312
    },
    {
      "epoch": 1.2992978752696067,
      "grad_norm": 0.4843747913837433,
      "learning_rate": 6.220232981318432e-06,
      "loss": 0.3951,
      "step": 28313
    },
    {
      "epoch": 1.2993437657748612,
      "grad_norm": 0.4609571695327759,
      "learning_rate": 6.219995205566817e-06,
      "loss": 0.3216,
      "step": 28314
    },
    {
      "epoch": 1.2993896562801157,
      "grad_norm": 0.49624764919281006,
      "learning_rate": 6.2197574268814975e-06,
      "loss": 0.443,
      "step": 28315
    },
    {
      "epoch": 1.2994355467853702,
      "grad_norm": 0.46185392141342163,
      "learning_rate": 6.219519645263046e-06,
      "loss": 0.3311,
      "step": 28316
    },
    {
      "epoch": 1.2994814372906247,
      "grad_norm": 0.44932329654693604,
      "learning_rate": 6.219281860712034e-06,
      "loss": 0.2944,
      "step": 28317
    },
    {
      "epoch": 1.299527327795879,
      "grad_norm": 0.45799845457077026,
      "learning_rate": 6.2190440732290345e-06,
      "loss": 0.3128,
      "step": 28318
    },
    {
      "epoch": 1.2995732183011335,
      "grad_norm": 0.5325301289558411,
      "learning_rate": 6.218806282814619e-06,
      "loss": 0.4391,
      "step": 28319
    },
    {
      "epoch": 1.299619108806388,
      "grad_norm": 0.5052163600921631,
      "learning_rate": 6.218568489469359e-06,
      "loss": 0.416,
      "step": 28320
    },
    {
      "epoch": 1.2996649993116425,
      "grad_norm": 0.47152796387672424,
      "learning_rate": 6.218330693193827e-06,
      "loss": 0.3621,
      "step": 28321
    },
    {
      "epoch": 1.2997108898168968,
      "grad_norm": 0.5058289766311646,
      "learning_rate": 6.2180928939885935e-06,
      "loss": 0.4142,
      "step": 28322
    },
    {
      "epoch": 1.2997567803221513,
      "grad_norm": 0.4970706105232239,
      "learning_rate": 6.217855091854231e-06,
      "loss": 0.3947,
      "step": 28323
    },
    {
      "epoch": 1.2998026708274057,
      "grad_norm": 0.49988478422164917,
      "learning_rate": 6.217617286791311e-06,
      "loss": 0.3538,
      "step": 28324
    },
    {
      "epoch": 1.2998485613326602,
      "grad_norm": 0.47883984446525574,
      "learning_rate": 6.2173794788004074e-06,
      "loss": 0.4016,
      "step": 28325
    },
    {
      "epoch": 1.2998944518379147,
      "grad_norm": 0.48086363077163696,
      "learning_rate": 6.21714166788209e-06,
      "loss": 0.3781,
      "step": 28326
    },
    {
      "epoch": 1.2999403423431692,
      "grad_norm": 0.475839227437973,
      "learning_rate": 6.2169038540369285e-06,
      "loss": 0.3768,
      "step": 28327
    },
    {
      "epoch": 1.2999862328484237,
      "grad_norm": 0.43456992506980896,
      "learning_rate": 6.216666037265499e-06,
      "loss": 0.3142,
      "step": 28328
    },
    {
      "epoch": 1.3000321233536782,
      "grad_norm": 0.5148123502731323,
      "learning_rate": 6.216428217568372e-06,
      "loss": 0.3877,
      "step": 28329
    },
    {
      "epoch": 1.3000780138589325,
      "grad_norm": 0.47856128215789795,
      "learning_rate": 6.2161903949461176e-06,
      "loss": 0.3666,
      "step": 28330
    },
    {
      "epoch": 1.300123904364187,
      "grad_norm": 0.45078423619270325,
      "learning_rate": 6.2159525693993105e-06,
      "loss": 0.3584,
      "step": 28331
    },
    {
      "epoch": 1.3001697948694415,
      "grad_norm": 0.4434889853000641,
      "learning_rate": 6.215714740928523e-06,
      "loss": 0.324,
      "step": 28332
    },
    {
      "epoch": 1.300215685374696,
      "grad_norm": 0.4791339933872223,
      "learning_rate": 6.215476909534321e-06,
      "loss": 0.3505,
      "step": 28333
    },
    {
      "epoch": 1.3002615758799505,
      "grad_norm": 0.4591011106967926,
      "learning_rate": 6.215239075217284e-06,
      "loss": 0.3607,
      "step": 28334
    },
    {
      "epoch": 1.3003074663852048,
      "grad_norm": 0.4973049461841583,
      "learning_rate": 6.215001237977979e-06,
      "loss": 0.3913,
      "step": 28335
    },
    {
      "epoch": 1.3003533568904593,
      "grad_norm": 0.45025011897087097,
      "learning_rate": 6.2147633978169805e-06,
      "loss": 0.3052,
      "step": 28336
    },
    {
      "epoch": 1.3003992473957138,
      "grad_norm": 0.4808560609817505,
      "learning_rate": 6.214525554734859e-06,
      "loss": 0.3789,
      "step": 28337
    },
    {
      "epoch": 1.3004451379009683,
      "grad_norm": 0.45503878593444824,
      "learning_rate": 6.214287708732186e-06,
      "loss": 0.3557,
      "step": 28338
    },
    {
      "epoch": 1.3004910284062228,
      "grad_norm": 0.4446350932121277,
      "learning_rate": 6.214049859809535e-06,
      "loss": 0.3532,
      "step": 28339
    },
    {
      "epoch": 1.3005369189114773,
      "grad_norm": 0.47719836235046387,
      "learning_rate": 6.213812007967478e-06,
      "loss": 0.3446,
      "step": 28340
    },
    {
      "epoch": 1.3005828094167318,
      "grad_norm": 0.4537985026836395,
      "learning_rate": 6.213574153206584e-06,
      "loss": 0.409,
      "step": 28341
    },
    {
      "epoch": 1.3006286999219863,
      "grad_norm": 0.4644417464733124,
      "learning_rate": 6.21333629552743e-06,
      "loss": 0.3155,
      "step": 28342
    },
    {
      "epoch": 1.3006745904272405,
      "grad_norm": 0.4750857949256897,
      "learning_rate": 6.213098434930583e-06,
      "loss": 0.3529,
      "step": 28343
    },
    {
      "epoch": 1.300720480932495,
      "grad_norm": 0.4374186098575592,
      "learning_rate": 6.212860571416618e-06,
      "loss": 0.3264,
      "step": 28344
    },
    {
      "epoch": 1.3007663714377495,
      "grad_norm": 0.4876135587692261,
      "learning_rate": 6.212622704986105e-06,
      "loss": 0.3447,
      "step": 28345
    },
    {
      "epoch": 1.300812261943004,
      "grad_norm": 0.46964141726493835,
      "learning_rate": 6.21238483563962e-06,
      "loss": 0.3728,
      "step": 28346
    },
    {
      "epoch": 1.3008581524482585,
      "grad_norm": 0.5390739440917969,
      "learning_rate": 6.21214696337773e-06,
      "loss": 0.3837,
      "step": 28347
    },
    {
      "epoch": 1.3009040429535128,
      "grad_norm": 0.5065861940383911,
      "learning_rate": 6.21190908820101e-06,
      "loss": 0.3438,
      "step": 28348
    },
    {
      "epoch": 1.3009499334587673,
      "grad_norm": 0.48037227988243103,
      "learning_rate": 6.211671210110029e-06,
      "loss": 0.3316,
      "step": 28349
    },
    {
      "epoch": 1.3009958239640218,
      "grad_norm": 0.54294753074646,
      "learning_rate": 6.211433329105363e-06,
      "loss": 0.4446,
      "step": 28350
    },
    {
      "epoch": 1.3010417144692763,
      "grad_norm": 0.4697289764881134,
      "learning_rate": 6.2111954451875825e-06,
      "loss": 0.3726,
      "step": 28351
    },
    {
      "epoch": 1.3010876049745308,
      "grad_norm": 0.4775595963001251,
      "learning_rate": 6.210957558357258e-06,
      "loss": 0.3637,
      "step": 28352
    },
    {
      "epoch": 1.3011334954797853,
      "grad_norm": 0.6707074046134949,
      "learning_rate": 6.210719668614964e-06,
      "loss": 0.382,
      "step": 28353
    },
    {
      "epoch": 1.3011793859850398,
      "grad_norm": 0.43275824189186096,
      "learning_rate": 6.21048177596127e-06,
      "loss": 0.2764,
      "step": 28354
    },
    {
      "epoch": 1.3012252764902943,
      "grad_norm": 0.46235793828964233,
      "learning_rate": 6.2102438803967504e-06,
      "loss": 0.3744,
      "step": 28355
    },
    {
      "epoch": 1.3012711669955486,
      "grad_norm": 0.42501163482666016,
      "learning_rate": 6.210005981921975e-06,
      "loss": 0.2634,
      "step": 28356
    },
    {
      "epoch": 1.301317057500803,
      "grad_norm": 0.4884771704673767,
      "learning_rate": 6.209768080537518e-06,
      "loss": 0.3574,
      "step": 28357
    },
    {
      "epoch": 1.3013629480060576,
      "grad_norm": 0.5541201829910278,
      "learning_rate": 6.209530176243952e-06,
      "loss": 0.46,
      "step": 28358
    },
    {
      "epoch": 1.301408838511312,
      "grad_norm": 0.4862549304962158,
      "learning_rate": 6.209292269041844e-06,
      "loss": 0.3509,
      "step": 28359
    },
    {
      "epoch": 1.3014547290165663,
      "grad_norm": 0.5106180906295776,
      "learning_rate": 6.209054358931773e-06,
      "loss": 0.425,
      "step": 28360
    },
    {
      "epoch": 1.3015006195218208,
      "grad_norm": 0.43822890520095825,
      "learning_rate": 6.208816445914306e-06,
      "loss": 0.3214,
      "step": 28361
    },
    {
      "epoch": 1.3015465100270753,
      "grad_norm": 0.49049073457717896,
      "learning_rate": 6.208578529990019e-06,
      "loss": 0.3877,
      "step": 28362
    },
    {
      "epoch": 1.3015924005323298,
      "grad_norm": 0.4863899052143097,
      "learning_rate": 6.208340611159481e-06,
      "loss": 0.3604,
      "step": 28363
    },
    {
      "epoch": 1.3016382910375843,
      "grad_norm": 0.46089184284210205,
      "learning_rate": 6.208102689423265e-06,
      "loss": 0.3423,
      "step": 28364
    },
    {
      "epoch": 1.3016841815428388,
      "grad_norm": 0.5719490051269531,
      "learning_rate": 6.2078647647819435e-06,
      "loss": 0.48,
      "step": 28365
    },
    {
      "epoch": 1.3017300720480933,
      "grad_norm": 0.475108802318573,
      "learning_rate": 6.207626837236088e-06,
      "loss": 0.3777,
      "step": 28366
    },
    {
      "epoch": 1.3017759625533478,
      "grad_norm": 0.4528774917125702,
      "learning_rate": 6.2073889067862705e-06,
      "loss": 0.3712,
      "step": 28367
    },
    {
      "epoch": 1.301821853058602,
      "grad_norm": 0.4743521809577942,
      "learning_rate": 6.207150973433065e-06,
      "loss": 0.4073,
      "step": 28368
    },
    {
      "epoch": 1.3018677435638566,
      "grad_norm": 0.4361277222633362,
      "learning_rate": 6.206913037177042e-06,
      "loss": 0.301,
      "step": 28369
    },
    {
      "epoch": 1.301913634069111,
      "grad_norm": 0.5007175207138062,
      "learning_rate": 6.206675098018774e-06,
      "loss": 0.4149,
      "step": 28370
    },
    {
      "epoch": 1.3019595245743656,
      "grad_norm": 0.4634784460067749,
      "learning_rate": 6.206437155958834e-06,
      "loss": 0.3695,
      "step": 28371
    },
    {
      "epoch": 1.30200541507962,
      "grad_norm": 0.46629971265792847,
      "learning_rate": 6.206199210997792e-06,
      "loss": 0.3505,
      "step": 28372
    },
    {
      "epoch": 1.3020513055848744,
      "grad_norm": 0.42915579676628113,
      "learning_rate": 6.205961263136222e-06,
      "loss": 0.3249,
      "step": 28373
    },
    {
      "epoch": 1.3020971960901289,
      "grad_norm": 0.4242452085018158,
      "learning_rate": 6.2057233123746965e-06,
      "loss": 0.3291,
      "step": 28374
    },
    {
      "epoch": 1.3021430865953834,
      "grad_norm": 0.46684548258781433,
      "learning_rate": 6.205485358713788e-06,
      "loss": 0.3923,
      "step": 28375
    },
    {
      "epoch": 1.3021889771006379,
      "grad_norm": 0.46860218048095703,
      "learning_rate": 6.205247402154064e-06,
      "loss": 0.3548,
      "step": 28376
    },
    {
      "epoch": 1.3022348676058924,
      "grad_norm": 0.44261425733566284,
      "learning_rate": 6.2050094426961036e-06,
      "loss": 0.3258,
      "step": 28377
    },
    {
      "epoch": 1.3022807581111469,
      "grad_norm": 0.45004382729530334,
      "learning_rate": 6.204771480340474e-06,
      "loss": 0.3323,
      "step": 28378
    },
    {
      "epoch": 1.3023266486164014,
      "grad_norm": 0.42827728390693665,
      "learning_rate": 6.20453351508775e-06,
      "loss": 0.2778,
      "step": 28379
    },
    {
      "epoch": 1.3023725391216558,
      "grad_norm": 0.4929460287094116,
      "learning_rate": 6.204295546938505e-06,
      "loss": 0.4184,
      "step": 28380
    },
    {
      "epoch": 1.3024184296269101,
      "grad_norm": 0.4692342281341553,
      "learning_rate": 6.204057575893307e-06,
      "loss": 0.3864,
      "step": 28381
    },
    {
      "epoch": 1.3024643201321646,
      "grad_norm": 0.41919299960136414,
      "learning_rate": 6.203819601952731e-06,
      "loss": 0.2552,
      "step": 28382
    },
    {
      "epoch": 1.3025102106374191,
      "grad_norm": 0.48206856846809387,
      "learning_rate": 6.203581625117348e-06,
      "loss": 0.3526,
      "step": 28383
    },
    {
      "epoch": 1.3025561011426736,
      "grad_norm": 0.44326165318489075,
      "learning_rate": 6.203343645387731e-06,
      "loss": 0.2985,
      "step": 28384
    },
    {
      "epoch": 1.3026019916479281,
      "grad_norm": 0.4822196066379547,
      "learning_rate": 6.2031056627644535e-06,
      "loss": 0.3871,
      "step": 28385
    },
    {
      "epoch": 1.3026478821531824,
      "grad_norm": 0.44136297702789307,
      "learning_rate": 6.2028676772480865e-06,
      "loss": 0.285,
      "step": 28386
    },
    {
      "epoch": 1.3026937726584369,
      "grad_norm": 0.4884640872478485,
      "learning_rate": 6.202629688839201e-06,
      "loss": 0.3419,
      "step": 28387
    },
    {
      "epoch": 1.3027396631636914,
      "grad_norm": 0.43173038959503174,
      "learning_rate": 6.202391697538373e-06,
      "loss": 0.2726,
      "step": 28388
    },
    {
      "epoch": 1.3027855536689459,
      "grad_norm": 0.45792195200920105,
      "learning_rate": 6.202153703346171e-06,
      "loss": 0.3467,
      "step": 28389
    },
    {
      "epoch": 1.3028314441742004,
      "grad_norm": 0.49218297004699707,
      "learning_rate": 6.201915706263169e-06,
      "loss": 0.328,
      "step": 28390
    },
    {
      "epoch": 1.3028773346794549,
      "grad_norm": 0.4530586302280426,
      "learning_rate": 6.201677706289939e-06,
      "loss": 0.3038,
      "step": 28391
    },
    {
      "epoch": 1.3029232251847094,
      "grad_norm": 0.44972261786460876,
      "learning_rate": 6.201439703427054e-06,
      "loss": 0.3281,
      "step": 28392
    },
    {
      "epoch": 1.3029691156899639,
      "grad_norm": 0.5540485978126526,
      "learning_rate": 6.201201697675083e-06,
      "loss": 0.37,
      "step": 28393
    },
    {
      "epoch": 1.3030150061952182,
      "grad_norm": 0.4657600224018097,
      "learning_rate": 6.200963689034605e-06,
      "loss": 0.3585,
      "step": 28394
    },
    {
      "epoch": 1.3030608967004726,
      "grad_norm": 0.48026615381240845,
      "learning_rate": 6.200725677506186e-06,
      "loss": 0.3738,
      "step": 28395
    },
    {
      "epoch": 1.3031067872057271,
      "grad_norm": 0.44920414686203003,
      "learning_rate": 6.200487663090401e-06,
      "loss": 0.3532,
      "step": 28396
    },
    {
      "epoch": 1.3031526777109816,
      "grad_norm": 0.4611596465110779,
      "learning_rate": 6.200249645787823e-06,
      "loss": 0.3191,
      "step": 28397
    },
    {
      "epoch": 1.3031985682162361,
      "grad_norm": 0.44115522503852844,
      "learning_rate": 6.200011625599022e-06,
      "loss": 0.332,
      "step": 28398
    },
    {
      "epoch": 1.3032444587214904,
      "grad_norm": 0.4785827398300171,
      "learning_rate": 6.199773602524573e-06,
      "loss": 0.3534,
      "step": 28399
    },
    {
      "epoch": 1.303290349226745,
      "grad_norm": 0.5511216521263123,
      "learning_rate": 6.1995355765650455e-06,
      "loss": 0.449,
      "step": 28400
    },
    {
      "epoch": 1.3033362397319994,
      "grad_norm": 0.508884072303772,
      "learning_rate": 6.1992975477210145e-06,
      "loss": 0.4017,
      "step": 28401
    },
    {
      "epoch": 1.303382130237254,
      "grad_norm": 0.48739388585090637,
      "learning_rate": 6.199059515993052e-06,
      "loss": 0.396,
      "step": 28402
    },
    {
      "epoch": 1.3034280207425084,
      "grad_norm": 0.49885162711143494,
      "learning_rate": 6.198821481381729e-06,
      "loss": 0.3899,
      "step": 28403
    },
    {
      "epoch": 1.303473911247763,
      "grad_norm": 0.47940605878829956,
      "learning_rate": 6.198583443887619e-06,
      "loss": 0.3374,
      "step": 28404
    },
    {
      "epoch": 1.3035198017530174,
      "grad_norm": 0.49607333540916443,
      "learning_rate": 6.1983454035112946e-06,
      "loss": 0.4084,
      "step": 28405
    },
    {
      "epoch": 1.303565692258272,
      "grad_norm": 0.44045209884643555,
      "learning_rate": 6.198107360253326e-06,
      "loss": 0.3291,
      "step": 28406
    },
    {
      "epoch": 1.3036115827635262,
      "grad_norm": 0.5088791847229004,
      "learning_rate": 6.197869314114288e-06,
      "loss": 0.4133,
      "step": 28407
    },
    {
      "epoch": 1.3036574732687807,
      "grad_norm": 0.4871130883693695,
      "learning_rate": 6.197631265094754e-06,
      "loss": 0.4436,
      "step": 28408
    },
    {
      "epoch": 1.3037033637740352,
      "grad_norm": 0.530781090259552,
      "learning_rate": 6.1973932131952945e-06,
      "loss": 0.4311,
      "step": 28409
    },
    {
      "epoch": 1.3037492542792897,
      "grad_norm": 0.42919108271598816,
      "learning_rate": 6.19715515841648e-06,
      "loss": 0.2838,
      "step": 28410
    },
    {
      "epoch": 1.303795144784544,
      "grad_norm": 0.4410487413406372,
      "learning_rate": 6.196917100758887e-06,
      "loss": 0.3307,
      "step": 28411
    },
    {
      "epoch": 1.3038410352897984,
      "grad_norm": 0.4688359200954437,
      "learning_rate": 6.196679040223086e-06,
      "loss": 0.3708,
      "step": 28412
    },
    {
      "epoch": 1.303886925795053,
      "grad_norm": 0.47724464535713196,
      "learning_rate": 6.19644097680965e-06,
      "loss": 0.3128,
      "step": 28413
    },
    {
      "epoch": 1.3039328163003074,
      "grad_norm": 0.456680566072464,
      "learning_rate": 6.19620291051915e-06,
      "loss": 0.3219,
      "step": 28414
    },
    {
      "epoch": 1.303978706805562,
      "grad_norm": 0.47553175687789917,
      "learning_rate": 6.195964841352159e-06,
      "loss": 0.364,
      "step": 28415
    },
    {
      "epoch": 1.3040245973108164,
      "grad_norm": 0.45054730772972107,
      "learning_rate": 6.1957267693092516e-06,
      "loss": 0.319,
      "step": 28416
    },
    {
      "epoch": 1.304070487816071,
      "grad_norm": 0.4467466473579407,
      "learning_rate": 6.195488694390997e-06,
      "loss": 0.2866,
      "step": 28417
    },
    {
      "epoch": 1.3041163783213254,
      "grad_norm": 0.5217620730400085,
      "learning_rate": 6.195250616597972e-06,
      "loss": 0.4796,
      "step": 28418
    },
    {
      "epoch": 1.3041622688265797,
      "grad_norm": 0.5000486969947815,
      "learning_rate": 6.195012535930745e-06,
      "loss": 0.4225,
      "step": 28419
    },
    {
      "epoch": 1.3042081593318342,
      "grad_norm": 0.49872806668281555,
      "learning_rate": 6.194774452389889e-06,
      "loss": 0.3938,
      "step": 28420
    },
    {
      "epoch": 1.3042540498370887,
      "grad_norm": 0.4550837278366089,
      "learning_rate": 6.194536365975979e-06,
      "loss": 0.3249,
      "step": 28421
    },
    {
      "epoch": 1.3042999403423432,
      "grad_norm": 0.4734867513179779,
      "learning_rate": 6.194298276689585e-06,
      "loss": 0.3488,
      "step": 28422
    },
    {
      "epoch": 1.3043458308475977,
      "grad_norm": 0.4773116707801819,
      "learning_rate": 6.194060184531281e-06,
      "loss": 0.3632,
      "step": 28423
    },
    {
      "epoch": 1.304391721352852,
      "grad_norm": 0.4427802562713623,
      "learning_rate": 6.193822089501639e-06,
      "loss": 0.3135,
      "step": 28424
    },
    {
      "epoch": 1.3044376118581065,
      "grad_norm": 0.5037828087806702,
      "learning_rate": 6.193583991601232e-06,
      "loss": 0.3806,
      "step": 28425
    },
    {
      "epoch": 1.304483502363361,
      "grad_norm": 0.48367759585380554,
      "learning_rate": 6.193345890830632e-06,
      "loss": 0.3272,
      "step": 28426
    },
    {
      "epoch": 1.3045293928686155,
      "grad_norm": 0.4710841476917267,
      "learning_rate": 6.193107787190412e-06,
      "loss": 0.3728,
      "step": 28427
    },
    {
      "epoch": 1.30457528337387,
      "grad_norm": 0.48471495509147644,
      "learning_rate": 6.192869680681144e-06,
      "loss": 0.3808,
      "step": 28428
    },
    {
      "epoch": 1.3046211738791245,
      "grad_norm": 0.4482499957084656,
      "learning_rate": 6.192631571303401e-06,
      "loss": 0.3245,
      "step": 28429
    },
    {
      "epoch": 1.304667064384379,
      "grad_norm": 0.4741309583187103,
      "learning_rate": 6.192393459057756e-06,
      "loss": 0.3609,
      "step": 28430
    },
    {
      "epoch": 1.3047129548896335,
      "grad_norm": 0.4384614825248718,
      "learning_rate": 6.19215534394478e-06,
      "loss": 0.3385,
      "step": 28431
    },
    {
      "epoch": 1.3047588453948877,
      "grad_norm": 0.5107391476631165,
      "learning_rate": 6.191917225965048e-06,
      "loss": 0.4438,
      "step": 28432
    },
    {
      "epoch": 1.3048047359001422,
      "grad_norm": 0.44049322605133057,
      "learning_rate": 6.1916791051191295e-06,
      "loss": 0.2815,
      "step": 28433
    },
    {
      "epoch": 1.3048506264053967,
      "grad_norm": 0.43820860981941223,
      "learning_rate": 6.191440981407601e-06,
      "loss": 0.2984,
      "step": 28434
    },
    {
      "epoch": 1.3048965169106512,
      "grad_norm": 0.4667779803276062,
      "learning_rate": 6.191202854831031e-06,
      "loss": 0.3383,
      "step": 28435
    },
    {
      "epoch": 1.3049424074159057,
      "grad_norm": 0.47031718492507935,
      "learning_rate": 6.1909647253899955e-06,
      "loss": 0.3415,
      "step": 28436
    },
    {
      "epoch": 1.30498829792116,
      "grad_norm": 0.47843238711357117,
      "learning_rate": 6.190726593085063e-06,
      "loss": 0.3401,
      "step": 28437
    },
    {
      "epoch": 1.3050341884264145,
      "grad_norm": 0.49746066331863403,
      "learning_rate": 6.190488457916811e-06,
      "loss": 0.4139,
      "step": 28438
    },
    {
      "epoch": 1.305080078931669,
      "grad_norm": 0.47530797123908997,
      "learning_rate": 6.190250319885811e-06,
      "loss": 0.3762,
      "step": 28439
    },
    {
      "epoch": 1.3051259694369235,
      "grad_norm": 0.4877783954143524,
      "learning_rate": 6.190012178992634e-06,
      "loss": 0.4019,
      "step": 28440
    },
    {
      "epoch": 1.305171859942178,
      "grad_norm": 0.46580812335014343,
      "learning_rate": 6.189774035237852e-06,
      "loss": 0.369,
      "step": 28441
    },
    {
      "epoch": 1.3052177504474325,
      "grad_norm": 0.46480369567871094,
      "learning_rate": 6.18953588862204e-06,
      "loss": 0.3474,
      "step": 28442
    },
    {
      "epoch": 1.305263640952687,
      "grad_norm": 0.47290506958961487,
      "learning_rate": 6.18929773914577e-06,
      "loss": 0.3566,
      "step": 28443
    },
    {
      "epoch": 1.3053095314579415,
      "grad_norm": 0.43707790970802307,
      "learning_rate": 6.189059586809614e-06,
      "loss": 0.3216,
      "step": 28444
    },
    {
      "epoch": 1.3053554219631958,
      "grad_norm": 0.48596128821372986,
      "learning_rate": 6.188821431614145e-06,
      "loss": 0.3701,
      "step": 28445
    },
    {
      "epoch": 1.3054013124684503,
      "grad_norm": 0.46648699045181274,
      "learning_rate": 6.188583273559935e-06,
      "loss": 0.3376,
      "step": 28446
    },
    {
      "epoch": 1.3054472029737048,
      "grad_norm": 0.4911003112792969,
      "learning_rate": 6.188345112647558e-06,
      "loss": 0.4069,
      "step": 28447
    },
    {
      "epoch": 1.3054930934789593,
      "grad_norm": 0.4514539837837219,
      "learning_rate": 6.188106948877585e-06,
      "loss": 0.3508,
      "step": 28448
    },
    {
      "epoch": 1.3055389839842135,
      "grad_norm": 0.4720689654350281,
      "learning_rate": 6.18786878225059e-06,
      "loss": 0.3627,
      "step": 28449
    },
    {
      "epoch": 1.305584874489468,
      "grad_norm": 0.5006113648414612,
      "learning_rate": 6.187630612767146e-06,
      "loss": 0.379,
      "step": 28450
    },
    {
      "epoch": 1.3056307649947225,
      "grad_norm": 0.4580785036087036,
      "learning_rate": 6.187392440427826e-06,
      "loss": 0.3312,
      "step": 28451
    },
    {
      "epoch": 1.305676655499977,
      "grad_norm": 0.4630391597747803,
      "learning_rate": 6.1871542652332e-06,
      "loss": 0.3858,
      "step": 28452
    },
    {
      "epoch": 1.3057225460052315,
      "grad_norm": 0.4813069999217987,
      "learning_rate": 6.1869160871838456e-06,
      "loss": 0.3974,
      "step": 28453
    },
    {
      "epoch": 1.305768436510486,
      "grad_norm": 0.49546414613723755,
      "learning_rate": 6.1866779062803295e-06,
      "loss": 0.3618,
      "step": 28454
    },
    {
      "epoch": 1.3058143270157405,
      "grad_norm": 0.4998624324798584,
      "learning_rate": 6.186439722523227e-06,
      "loss": 0.3496,
      "step": 28455
    },
    {
      "epoch": 1.305860217520995,
      "grad_norm": 0.4507425129413605,
      "learning_rate": 6.186201535913115e-06,
      "loss": 0.3686,
      "step": 28456
    },
    {
      "epoch": 1.3059061080262493,
      "grad_norm": 0.3999234437942505,
      "learning_rate": 6.185963346450561e-06,
      "loss": 0.2334,
      "step": 28457
    },
    {
      "epoch": 1.3059519985315038,
      "grad_norm": 0.4932142198085785,
      "learning_rate": 6.18572515413614e-06,
      "loss": 0.3719,
      "step": 28458
    },
    {
      "epoch": 1.3059978890367583,
      "grad_norm": 0.45936155319213867,
      "learning_rate": 6.1854869589704234e-06,
      "loss": 0.3537,
      "step": 28459
    },
    {
      "epoch": 1.3060437795420128,
      "grad_norm": 0.4460750222206116,
      "learning_rate": 6.185248760953985e-06,
      "loss": 0.2843,
      "step": 28460
    },
    {
      "epoch": 1.3060896700472673,
      "grad_norm": 0.4730258584022522,
      "learning_rate": 6.185010560087398e-06,
      "loss": 0.3895,
      "step": 28461
    },
    {
      "epoch": 1.3061355605525216,
      "grad_norm": 0.46346402168273926,
      "learning_rate": 6.184772356371234e-06,
      "loss": 0.3535,
      "step": 28462
    },
    {
      "epoch": 1.306181451057776,
      "grad_norm": 0.46151021122932434,
      "learning_rate": 6.184534149806066e-06,
      "loss": 0.3312,
      "step": 28463
    },
    {
      "epoch": 1.3062273415630306,
      "grad_norm": 0.515189528465271,
      "learning_rate": 6.184295940392469e-06,
      "loss": 0.3874,
      "step": 28464
    },
    {
      "epoch": 1.306273232068285,
      "grad_norm": 0.4880011975765228,
      "learning_rate": 6.184057728131011e-06,
      "loss": 0.4162,
      "step": 28465
    },
    {
      "epoch": 1.3063191225735395,
      "grad_norm": 0.44706106185913086,
      "learning_rate": 6.183819513022269e-06,
      "loss": 0.3099,
      "step": 28466
    },
    {
      "epoch": 1.306365013078794,
      "grad_norm": 0.48643556237220764,
      "learning_rate": 6.183581295066817e-06,
      "loss": 0.3412,
      "step": 28467
    },
    {
      "epoch": 1.3064109035840485,
      "grad_norm": 0.4929741322994232,
      "learning_rate": 6.183343074265224e-06,
      "loss": 0.3529,
      "step": 28468
    },
    {
      "epoch": 1.306456794089303,
      "grad_norm": 0.47997915744781494,
      "learning_rate": 6.183104850618064e-06,
      "loss": 0.3008,
      "step": 28469
    },
    {
      "epoch": 1.3065026845945573,
      "grad_norm": 0.505718469619751,
      "learning_rate": 6.1828666241259095e-06,
      "loss": 0.3989,
      "step": 28470
    },
    {
      "epoch": 1.3065485750998118,
      "grad_norm": 0.4884539544582367,
      "learning_rate": 6.182628394789335e-06,
      "loss": 0.4117,
      "step": 28471
    },
    {
      "epoch": 1.3065944656050663,
      "grad_norm": 0.4639051854610443,
      "learning_rate": 6.182390162608913e-06,
      "loss": 0.3284,
      "step": 28472
    },
    {
      "epoch": 1.3066403561103208,
      "grad_norm": 0.4939505159854889,
      "learning_rate": 6.1821519275852156e-06,
      "loss": 0.412,
      "step": 28473
    },
    {
      "epoch": 1.3066862466155753,
      "grad_norm": 0.4454679787158966,
      "learning_rate": 6.181913689718815e-06,
      "loss": 0.3201,
      "step": 28474
    },
    {
      "epoch": 1.3067321371208296,
      "grad_norm": 0.4373287260532379,
      "learning_rate": 6.1816754490102845e-06,
      "loss": 0.2965,
      "step": 28475
    },
    {
      "epoch": 1.306778027626084,
      "grad_norm": 0.4283183515071869,
      "learning_rate": 6.1814372054602e-06,
      "loss": 0.277,
      "step": 28476
    },
    {
      "epoch": 1.3068239181313386,
      "grad_norm": 0.43781235814094543,
      "learning_rate": 6.18119895906913e-06,
      "loss": 0.309,
      "step": 28477
    },
    {
      "epoch": 1.306869808636593,
      "grad_norm": 0.4613669514656067,
      "learning_rate": 6.1809607098376515e-06,
      "loss": 0.3414,
      "step": 28478
    },
    {
      "epoch": 1.3069156991418476,
      "grad_norm": 0.44764384627342224,
      "learning_rate": 6.180722457766332e-06,
      "loss": 0.3411,
      "step": 28479
    },
    {
      "epoch": 1.306961589647102,
      "grad_norm": 0.47301849722862244,
      "learning_rate": 6.18048420285575e-06,
      "loss": 0.3546,
      "step": 28480
    },
    {
      "epoch": 1.3070074801523566,
      "grad_norm": 0.4886377453804016,
      "learning_rate": 6.180245945106475e-06,
      "loss": 0.326,
      "step": 28481
    },
    {
      "epoch": 1.307053370657611,
      "grad_norm": 0.4973951578140259,
      "learning_rate": 6.18000768451908e-06,
      "loss": 0.4175,
      "step": 28482
    },
    {
      "epoch": 1.3070992611628653,
      "grad_norm": 0.5406638979911804,
      "learning_rate": 6.179769421094141e-06,
      "loss": 0.4031,
      "step": 28483
    },
    {
      "epoch": 1.3071451516681198,
      "grad_norm": 0.4691663384437561,
      "learning_rate": 6.179531154832229e-06,
      "loss": 0.3605,
      "step": 28484
    },
    {
      "epoch": 1.3071910421733743,
      "grad_norm": 0.4627991318702698,
      "learning_rate": 6.179292885733915e-06,
      "loss": 0.3176,
      "step": 28485
    },
    {
      "epoch": 1.3072369326786288,
      "grad_norm": 0.4395746886730194,
      "learning_rate": 6.179054613799774e-06,
      "loss": 0.2944,
      "step": 28486
    },
    {
      "epoch": 1.3072828231838833,
      "grad_norm": 0.4376395344734192,
      "learning_rate": 6.17881633903038e-06,
      "loss": 0.2796,
      "step": 28487
    },
    {
      "epoch": 1.3073287136891376,
      "grad_norm": 0.4390043616294861,
      "learning_rate": 6.178578061426304e-06,
      "loss": 0.3046,
      "step": 28488
    },
    {
      "epoch": 1.307374604194392,
      "grad_norm": 0.4263704717159271,
      "learning_rate": 6.17833978098812e-06,
      "loss": 0.2943,
      "step": 28489
    },
    {
      "epoch": 1.3074204946996466,
      "grad_norm": 0.4778251349925995,
      "learning_rate": 6.178101497716402e-06,
      "loss": 0.3582,
      "step": 28490
    },
    {
      "epoch": 1.307466385204901,
      "grad_norm": 0.5198782682418823,
      "learning_rate": 6.17786321161172e-06,
      "loss": 0.4067,
      "step": 28491
    },
    {
      "epoch": 1.3075122757101556,
      "grad_norm": 0.4476759433746338,
      "learning_rate": 6.177624922674648e-06,
      "loss": 0.3016,
      "step": 28492
    },
    {
      "epoch": 1.30755816621541,
      "grad_norm": 0.4724082052707672,
      "learning_rate": 6.1773866309057615e-06,
      "loss": 0.3988,
      "step": 28493
    },
    {
      "epoch": 1.3076040567206646,
      "grad_norm": 0.49371281266212463,
      "learning_rate": 6.177148336305632e-06,
      "loss": 0.3919,
      "step": 28494
    },
    {
      "epoch": 1.307649947225919,
      "grad_norm": 0.4505222737789154,
      "learning_rate": 6.176910038874831e-06,
      "loss": 0.3482,
      "step": 28495
    },
    {
      "epoch": 1.3076958377311734,
      "grad_norm": 0.4916074275970459,
      "learning_rate": 6.176671738613933e-06,
      "loss": 0.3962,
      "step": 28496
    },
    {
      "epoch": 1.3077417282364279,
      "grad_norm": 0.471711665391922,
      "learning_rate": 6.17643343552351e-06,
      "loss": 0.3532,
      "step": 28497
    },
    {
      "epoch": 1.3077876187416824,
      "grad_norm": 0.46106070280075073,
      "learning_rate": 6.176195129604138e-06,
      "loss": 0.3353,
      "step": 28498
    },
    {
      "epoch": 1.3078335092469369,
      "grad_norm": 0.44245418906211853,
      "learning_rate": 6.175956820856386e-06,
      "loss": 0.3278,
      "step": 28499
    },
    {
      "epoch": 1.3078793997521911,
      "grad_norm": 0.47627493739128113,
      "learning_rate": 6.1757185092808305e-06,
      "loss": 0.3779,
      "step": 28500
    },
    {
      "epoch": 1.3079252902574456,
      "grad_norm": 0.45083844661712646,
      "learning_rate": 6.175480194878043e-06,
      "loss": 0.3726,
      "step": 28501
    },
    {
      "epoch": 1.3079711807627001,
      "grad_norm": 0.4742313325405121,
      "learning_rate": 6.175241877648596e-06,
      "loss": 0.3789,
      "step": 28502
    },
    {
      "epoch": 1.3080170712679546,
      "grad_norm": 0.5404680967330933,
      "learning_rate": 6.175003557593062e-06,
      "loss": 0.3533,
      "step": 28503
    },
    {
      "epoch": 1.3080629617732091,
      "grad_norm": 0.5208746790885925,
      "learning_rate": 6.174765234712017e-06,
      "loss": 0.4807,
      "step": 28504
    },
    {
      "epoch": 1.3081088522784636,
      "grad_norm": 0.4549843966960907,
      "learning_rate": 6.174526909006033e-06,
      "loss": 0.3604,
      "step": 28505
    },
    {
      "epoch": 1.3081547427837181,
      "grad_norm": 0.45503559708595276,
      "learning_rate": 6.17428858047568e-06,
      "loss": 0.3889,
      "step": 28506
    },
    {
      "epoch": 1.3082006332889726,
      "grad_norm": 0.43390995264053345,
      "learning_rate": 6.174050249121535e-06,
      "loss": 0.2902,
      "step": 28507
    },
    {
      "epoch": 1.308246523794227,
      "grad_norm": 0.49847468733787537,
      "learning_rate": 6.17381191494417e-06,
      "loss": 0.4126,
      "step": 28508
    },
    {
      "epoch": 1.3082924142994814,
      "grad_norm": 0.44737550616264343,
      "learning_rate": 6.173573577944158e-06,
      "loss": 0.3372,
      "step": 28509
    },
    {
      "epoch": 1.308338304804736,
      "grad_norm": 0.47744810581207275,
      "learning_rate": 6.17333523812207e-06,
      "loss": 0.4259,
      "step": 28510
    },
    {
      "epoch": 1.3083841953099904,
      "grad_norm": 0.5481723546981812,
      "learning_rate": 6.1730968954784835e-06,
      "loss": 0.5268,
      "step": 28511
    },
    {
      "epoch": 1.308430085815245,
      "grad_norm": 0.477098286151886,
      "learning_rate": 6.172858550013968e-06,
      "loss": 0.3497,
      "step": 28512
    },
    {
      "epoch": 1.3084759763204992,
      "grad_norm": 0.47116851806640625,
      "learning_rate": 6.172620201729097e-06,
      "loss": 0.3119,
      "step": 28513
    },
    {
      "epoch": 1.3085218668257537,
      "grad_norm": 0.4341926872730255,
      "learning_rate": 6.172381850624445e-06,
      "loss": 0.2937,
      "step": 28514
    },
    {
      "epoch": 1.3085677573310082,
      "grad_norm": 0.5140688419342041,
      "learning_rate": 6.172143496700587e-06,
      "loss": 0.4393,
      "step": 28515
    },
    {
      "epoch": 1.3086136478362627,
      "grad_norm": 0.714664101600647,
      "learning_rate": 6.171905139958092e-06,
      "loss": 0.3828,
      "step": 28516
    },
    {
      "epoch": 1.3086595383415172,
      "grad_norm": 0.48204636573791504,
      "learning_rate": 6.171666780397534e-06,
      "loss": 0.3353,
      "step": 28517
    },
    {
      "epoch": 1.3087054288467717,
      "grad_norm": 0.4299159348011017,
      "learning_rate": 6.171428418019488e-06,
      "loss": 0.3335,
      "step": 28518
    },
    {
      "epoch": 1.3087513193520262,
      "grad_norm": 0.49862587451934814,
      "learning_rate": 6.171190052824526e-06,
      "loss": 0.4326,
      "step": 28519
    },
    {
      "epoch": 1.3087972098572807,
      "grad_norm": 0.4644327461719513,
      "learning_rate": 6.170951684813222e-06,
      "loss": 0.3636,
      "step": 28520
    },
    {
      "epoch": 1.308843100362535,
      "grad_norm": 0.4814310073852539,
      "learning_rate": 6.17071331398615e-06,
      "loss": 0.3797,
      "step": 28521
    },
    {
      "epoch": 1.3088889908677894,
      "grad_norm": 0.49443548917770386,
      "learning_rate": 6.17047494034388e-06,
      "loss": 0.3874,
      "step": 28522
    },
    {
      "epoch": 1.308934881373044,
      "grad_norm": 0.4782508611679077,
      "learning_rate": 6.170236563886988e-06,
      "loss": 0.3733,
      "step": 28523
    },
    {
      "epoch": 1.3089807718782984,
      "grad_norm": 0.47029030323028564,
      "learning_rate": 6.1699981846160475e-06,
      "loss": 0.3413,
      "step": 28524
    },
    {
      "epoch": 1.309026662383553,
      "grad_norm": 0.47686097025871277,
      "learning_rate": 6.169759802531629e-06,
      "loss": 0.3726,
      "step": 28525
    },
    {
      "epoch": 1.3090725528888072,
      "grad_norm": 0.4316127598285675,
      "learning_rate": 6.1695214176343074e-06,
      "loss": 0.3175,
      "step": 28526
    },
    {
      "epoch": 1.3091184433940617,
      "grad_norm": 0.503305971622467,
      "learning_rate": 6.169283029924657e-06,
      "loss": 0.4216,
      "step": 28527
    },
    {
      "epoch": 1.3091643338993162,
      "grad_norm": 0.48264917731285095,
      "learning_rate": 6.169044639403249e-06,
      "loss": 0.3673,
      "step": 28528
    },
    {
      "epoch": 1.3092102244045707,
      "grad_norm": 0.4756075143814087,
      "learning_rate": 6.168806246070659e-06,
      "loss": 0.3806,
      "step": 28529
    },
    {
      "epoch": 1.3092561149098252,
      "grad_norm": 0.518028974533081,
      "learning_rate": 6.168567849927457e-06,
      "loss": 0.4301,
      "step": 28530
    },
    {
      "epoch": 1.3093020054150797,
      "grad_norm": 0.46427789330482483,
      "learning_rate": 6.168329450974217e-06,
      "loss": 0.3651,
      "step": 28531
    },
    {
      "epoch": 1.3093478959203342,
      "grad_norm": 0.4986291527748108,
      "learning_rate": 6.168091049211516e-06,
      "loss": 0.4542,
      "step": 28532
    },
    {
      "epoch": 1.3093937864255887,
      "grad_norm": 0.41481804847717285,
      "learning_rate": 6.167852644639925e-06,
      "loss": 0.2913,
      "step": 28533
    },
    {
      "epoch": 1.309439676930843,
      "grad_norm": 0.4598534405231476,
      "learning_rate": 6.167614237260015e-06,
      "loss": 0.3372,
      "step": 28534
    },
    {
      "epoch": 1.3094855674360975,
      "grad_norm": 0.44803857803344727,
      "learning_rate": 6.167375827072364e-06,
      "loss": 0.3402,
      "step": 28535
    },
    {
      "epoch": 1.309531457941352,
      "grad_norm": 0.4834659993648529,
      "learning_rate": 6.16713741407754e-06,
      "loss": 0.3986,
      "step": 28536
    },
    {
      "epoch": 1.3095773484466064,
      "grad_norm": 0.4776255488395691,
      "learning_rate": 6.166898998276121e-06,
      "loss": 0.3446,
      "step": 28537
    },
    {
      "epoch": 1.3096232389518607,
      "grad_norm": 0.4730452299118042,
      "learning_rate": 6.166660579668677e-06,
      "loss": 0.4089,
      "step": 28538
    },
    {
      "epoch": 1.3096691294571152,
      "grad_norm": 0.4686093330383301,
      "learning_rate": 6.166422158255781e-06,
      "loss": 0.3704,
      "step": 28539
    },
    {
      "epoch": 1.3097150199623697,
      "grad_norm": 0.5005902051925659,
      "learning_rate": 6.166183734038009e-06,
      "loss": 0.3907,
      "step": 28540
    },
    {
      "epoch": 1.3097609104676242,
      "grad_norm": 0.45917049050331116,
      "learning_rate": 6.165945307015933e-06,
      "loss": 0.3282,
      "step": 28541
    },
    {
      "epoch": 1.3098068009728787,
      "grad_norm": 0.4347369074821472,
      "learning_rate": 6.165706877190127e-06,
      "loss": 0.3175,
      "step": 28542
    },
    {
      "epoch": 1.3098526914781332,
      "grad_norm": 0.461350679397583,
      "learning_rate": 6.165468444561164e-06,
      "loss": 0.3034,
      "step": 28543
    },
    {
      "epoch": 1.3098985819833877,
      "grad_norm": 0.476781964302063,
      "learning_rate": 6.165230009129618e-06,
      "loss": 0.3833,
      "step": 28544
    },
    {
      "epoch": 1.3099444724886422,
      "grad_norm": 0.4625754654407501,
      "learning_rate": 6.164991570896061e-06,
      "loss": 0.3656,
      "step": 28545
    },
    {
      "epoch": 1.3099903629938965,
      "grad_norm": 0.4508261978626251,
      "learning_rate": 6.164753129861066e-06,
      "loss": 0.3136,
      "step": 28546
    },
    {
      "epoch": 1.310036253499151,
      "grad_norm": 0.4516075551509857,
      "learning_rate": 6.164514686025208e-06,
      "loss": 0.3564,
      "step": 28547
    },
    {
      "epoch": 1.3100821440044055,
      "grad_norm": 0.484607458114624,
      "learning_rate": 6.1642762393890595e-06,
      "loss": 0.3464,
      "step": 28548
    },
    {
      "epoch": 1.31012803450966,
      "grad_norm": 0.4777067005634308,
      "learning_rate": 6.164037789953195e-06,
      "loss": 0.3728,
      "step": 28549
    },
    {
      "epoch": 1.3101739250149145,
      "grad_norm": 0.4741908013820648,
      "learning_rate": 6.163799337718186e-06,
      "loss": 0.3439,
      "step": 28550
    },
    {
      "epoch": 1.3102198155201688,
      "grad_norm": 0.5139142274856567,
      "learning_rate": 6.163560882684606e-06,
      "loss": 0.4746,
      "step": 28551
    },
    {
      "epoch": 1.3102657060254232,
      "grad_norm": 0.4546709358692169,
      "learning_rate": 6.163322424853031e-06,
      "loss": 0.3308,
      "step": 28552
    },
    {
      "epoch": 1.3103115965306777,
      "grad_norm": 0.4857752025127411,
      "learning_rate": 6.163083964224031e-06,
      "loss": 0.3868,
      "step": 28553
    },
    {
      "epoch": 1.3103574870359322,
      "grad_norm": 0.47456473112106323,
      "learning_rate": 6.162845500798184e-06,
      "loss": 0.36,
      "step": 28554
    },
    {
      "epoch": 1.3104033775411867,
      "grad_norm": 0.49496662616729736,
      "learning_rate": 6.1626070345760595e-06,
      "loss": 0.4222,
      "step": 28555
    },
    {
      "epoch": 1.3104492680464412,
      "grad_norm": 0.4268829822540283,
      "learning_rate": 6.16236856555823e-06,
      "loss": 0.3263,
      "step": 28556
    },
    {
      "epoch": 1.3104951585516957,
      "grad_norm": 0.5098237991333008,
      "learning_rate": 6.1621300937452726e-06,
      "loss": 0.3907,
      "step": 28557
    },
    {
      "epoch": 1.3105410490569502,
      "grad_norm": 0.44447413086891174,
      "learning_rate": 6.1618916191377585e-06,
      "loss": 0.3339,
      "step": 28558
    },
    {
      "epoch": 1.3105869395622045,
      "grad_norm": 0.5054704546928406,
      "learning_rate": 6.161653141736264e-06,
      "loss": 0.4602,
      "step": 28559
    },
    {
      "epoch": 1.310632830067459,
      "grad_norm": 0.461489737033844,
      "learning_rate": 6.161414661541358e-06,
      "loss": 0.3764,
      "step": 28560
    },
    {
      "epoch": 1.3106787205727135,
      "grad_norm": 0.48149654269218445,
      "learning_rate": 6.161176178553616e-06,
      "loss": 0.3654,
      "step": 28561
    },
    {
      "epoch": 1.310724611077968,
      "grad_norm": 0.44392064213752747,
      "learning_rate": 6.160937692773612e-06,
      "loss": 0.3485,
      "step": 28562
    },
    {
      "epoch": 1.3107705015832225,
      "grad_norm": 0.4637071490287781,
      "learning_rate": 6.160699204201921e-06,
      "loss": 0.3375,
      "step": 28563
    },
    {
      "epoch": 1.3108163920884768,
      "grad_norm": 0.46568265557289124,
      "learning_rate": 6.160460712839113e-06,
      "loss": 0.319,
      "step": 28564
    },
    {
      "epoch": 1.3108622825937313,
      "grad_norm": 0.45059555768966675,
      "learning_rate": 6.160222218685763e-06,
      "loss": 0.3324,
      "step": 28565
    },
    {
      "epoch": 1.3109081730989858,
      "grad_norm": 0.47354385256767273,
      "learning_rate": 6.1599837217424464e-06,
      "loss": 0.3738,
      "step": 28566
    },
    {
      "epoch": 1.3109540636042403,
      "grad_norm": 0.4755570590496063,
      "learning_rate": 6.159745222009733e-06,
      "loss": 0.3454,
      "step": 28567
    },
    {
      "epoch": 1.3109999541094948,
      "grad_norm": 0.4850596785545349,
      "learning_rate": 6.1595067194882e-06,
      "loss": 0.4403,
      "step": 28568
    },
    {
      "epoch": 1.3110458446147493,
      "grad_norm": 0.49915286898612976,
      "learning_rate": 6.159268214178419e-06,
      "loss": 0.4438,
      "step": 28569
    },
    {
      "epoch": 1.3110917351200038,
      "grad_norm": 0.48258668184280396,
      "learning_rate": 6.1590297060809634e-06,
      "loss": 0.3903,
      "step": 28570
    },
    {
      "epoch": 1.3111376256252583,
      "grad_norm": 0.44674497842788696,
      "learning_rate": 6.158791195196407e-06,
      "loss": 0.3466,
      "step": 28571
    },
    {
      "epoch": 1.3111835161305125,
      "grad_norm": 0.47380977869033813,
      "learning_rate": 6.158552681525325e-06,
      "loss": 0.3686,
      "step": 28572
    },
    {
      "epoch": 1.311229406635767,
      "grad_norm": 0.44130846858024597,
      "learning_rate": 6.158314165068287e-06,
      "loss": 0.3436,
      "step": 28573
    },
    {
      "epoch": 1.3112752971410215,
      "grad_norm": 0.4399886727333069,
      "learning_rate": 6.15807564582587e-06,
      "loss": 0.3195,
      "step": 28574
    },
    {
      "epoch": 1.311321187646276,
      "grad_norm": 0.47643277049064636,
      "learning_rate": 6.157837123798648e-06,
      "loss": 0.3699,
      "step": 28575
    },
    {
      "epoch": 1.3113670781515305,
      "grad_norm": 0.4628371596336365,
      "learning_rate": 6.157598598987193e-06,
      "loss": 0.3854,
      "step": 28576
    },
    {
      "epoch": 1.3114129686567848,
      "grad_norm": 0.46979060769081116,
      "learning_rate": 6.1573600713920765e-06,
      "loss": 0.3697,
      "step": 28577
    },
    {
      "epoch": 1.3114588591620393,
      "grad_norm": 0.43809646368026733,
      "learning_rate": 6.157121541013875e-06,
      "loss": 0.2975,
      "step": 28578
    },
    {
      "epoch": 1.3115047496672938,
      "grad_norm": 0.48673689365386963,
      "learning_rate": 6.156883007853161e-06,
      "loss": 0.3443,
      "step": 28579
    },
    {
      "epoch": 1.3115506401725483,
      "grad_norm": 0.4550205171108246,
      "learning_rate": 6.156644471910511e-06,
      "loss": 0.3474,
      "step": 28580
    },
    {
      "epoch": 1.3115965306778028,
      "grad_norm": 0.46326082944869995,
      "learning_rate": 6.156405933186494e-06,
      "loss": 0.3179,
      "step": 28581
    },
    {
      "epoch": 1.3116424211830573,
      "grad_norm": 0.4759811460971832,
      "learning_rate": 6.156167391681687e-06,
      "loss": 0.3256,
      "step": 28582
    },
    {
      "epoch": 1.3116883116883118,
      "grad_norm": 0.47733089327812195,
      "learning_rate": 6.155928847396661e-06,
      "loss": 0.3304,
      "step": 28583
    },
    {
      "epoch": 1.3117342021935663,
      "grad_norm": 0.4848475158214569,
      "learning_rate": 6.155690300331991e-06,
      "loss": 0.3442,
      "step": 28584
    },
    {
      "epoch": 1.3117800926988206,
      "grad_norm": 0.4695901572704315,
      "learning_rate": 6.15545175048825e-06,
      "loss": 0.3261,
      "step": 28585
    },
    {
      "epoch": 1.311825983204075,
      "grad_norm": 0.4677128493785858,
      "learning_rate": 6.155213197866014e-06,
      "loss": 0.3208,
      "step": 28586
    },
    {
      "epoch": 1.3118718737093296,
      "grad_norm": 0.478785902261734,
      "learning_rate": 6.154974642465855e-06,
      "loss": 0.3797,
      "step": 28587
    },
    {
      "epoch": 1.311917764214584,
      "grad_norm": 0.43242666125297546,
      "learning_rate": 6.154736084288345e-06,
      "loss": 0.3033,
      "step": 28588
    },
    {
      "epoch": 1.3119636547198383,
      "grad_norm": 0.5031163692474365,
      "learning_rate": 6.154497523334061e-06,
      "loss": 0.3918,
      "step": 28589
    },
    {
      "epoch": 1.3120095452250928,
      "grad_norm": 0.4640864431858063,
      "learning_rate": 6.154258959603573e-06,
      "loss": 0.3663,
      "step": 28590
    },
    {
      "epoch": 1.3120554357303473,
      "grad_norm": 0.4770459234714508,
      "learning_rate": 6.1540203930974575e-06,
      "loss": 0.3563,
      "step": 28591
    },
    {
      "epoch": 1.3121013262356018,
      "grad_norm": 0.48488062620162964,
      "learning_rate": 6.1537818238162874e-06,
      "loss": 0.418,
      "step": 28592
    },
    {
      "epoch": 1.3121472167408563,
      "grad_norm": 0.4416406452655792,
      "learning_rate": 6.153543251760634e-06,
      "loss": 0.3619,
      "step": 28593
    },
    {
      "epoch": 1.3121931072461108,
      "grad_norm": 0.505152702331543,
      "learning_rate": 6.153304676931075e-06,
      "loss": 0.436,
      "step": 28594
    },
    {
      "epoch": 1.3122389977513653,
      "grad_norm": 0.4532804489135742,
      "learning_rate": 6.1530660993281824e-06,
      "loss": 0.33,
      "step": 28595
    },
    {
      "epoch": 1.3122848882566198,
      "grad_norm": 0.4559321701526642,
      "learning_rate": 6.152827518952529e-06,
      "loss": 0.339,
      "step": 28596
    },
    {
      "epoch": 1.312330778761874,
      "grad_norm": 0.47121185064315796,
      "learning_rate": 6.15258893580469e-06,
      "loss": 0.3846,
      "step": 28597
    },
    {
      "epoch": 1.3123766692671286,
      "grad_norm": 0.4465380907058716,
      "learning_rate": 6.1523503498852375e-06,
      "loss": 0.3448,
      "step": 28598
    },
    {
      "epoch": 1.312422559772383,
      "grad_norm": 0.43431711196899414,
      "learning_rate": 6.152111761194747e-06,
      "loss": 0.2726,
      "step": 28599
    },
    {
      "epoch": 1.3124684502776376,
      "grad_norm": 0.470115065574646,
      "learning_rate": 6.1518731697337905e-06,
      "loss": 0.3722,
      "step": 28600
    },
    {
      "epoch": 1.312514340782892,
      "grad_norm": 0.4716876447200775,
      "learning_rate": 6.151634575502943e-06,
      "loss": 0.3637,
      "step": 28601
    },
    {
      "epoch": 1.3125602312881464,
      "grad_norm": 0.4862874150276184,
      "learning_rate": 6.1513959785027764e-06,
      "loss": 0.3013,
      "step": 28602
    },
    {
      "epoch": 1.3126061217934009,
      "grad_norm": 0.42651960253715515,
      "learning_rate": 6.15115737873387e-06,
      "loss": 0.3022,
      "step": 28603
    },
    {
      "epoch": 1.3126520122986554,
      "grad_norm": 0.46749043464660645,
      "learning_rate": 6.150918776196789e-06,
      "loss": 0.3815,
      "step": 28604
    },
    {
      "epoch": 1.3126979028039099,
      "grad_norm": 0.46974918246269226,
      "learning_rate": 6.150680170892113e-06,
      "loss": 0.3741,
      "step": 28605
    },
    {
      "epoch": 1.3127437933091644,
      "grad_norm": 0.49506789445877075,
      "learning_rate": 6.150441562820415e-06,
      "loss": 0.3798,
      "step": 28606
    },
    {
      "epoch": 1.3127896838144189,
      "grad_norm": 0.4572712481021881,
      "learning_rate": 6.150202951982267e-06,
      "loss": 0.3512,
      "step": 28607
    },
    {
      "epoch": 1.3128355743196733,
      "grad_norm": 0.4933372437953949,
      "learning_rate": 6.149964338378245e-06,
      "loss": 0.3827,
      "step": 28608
    },
    {
      "epoch": 1.3128814648249278,
      "grad_norm": 0.42919716238975525,
      "learning_rate": 6.149725722008922e-06,
      "loss": 0.3089,
      "step": 28609
    },
    {
      "epoch": 1.3129273553301821,
      "grad_norm": 0.4745995104312897,
      "learning_rate": 6.14948710287487e-06,
      "loss": 0.3309,
      "step": 28610
    },
    {
      "epoch": 1.3129732458354366,
      "grad_norm": 0.5293872952461243,
      "learning_rate": 6.149248480976666e-06,
      "loss": 0.4771,
      "step": 28611
    },
    {
      "epoch": 1.3130191363406911,
      "grad_norm": 0.4576182961463928,
      "learning_rate": 6.14900985631488e-06,
      "loss": 0.3734,
      "step": 28612
    },
    {
      "epoch": 1.3130650268459456,
      "grad_norm": 0.4510018229484558,
      "learning_rate": 6.148771228890089e-06,
      "loss": 0.3272,
      "step": 28613
    },
    {
      "epoch": 1.3131109173512001,
      "grad_norm": 0.4822823405265808,
      "learning_rate": 6.1485325987028675e-06,
      "loss": 0.4001,
      "step": 28614
    },
    {
      "epoch": 1.3131568078564544,
      "grad_norm": 0.479226291179657,
      "learning_rate": 6.148293965753786e-06,
      "loss": 0.3757,
      "step": 28615
    },
    {
      "epoch": 1.3132026983617089,
      "grad_norm": 0.47885024547576904,
      "learning_rate": 6.148055330043419e-06,
      "loss": 0.422,
      "step": 28616
    },
    {
      "epoch": 1.3132485888669634,
      "grad_norm": 0.5850602984428406,
      "learning_rate": 6.147816691572342e-06,
      "loss": 0.3542,
      "step": 28617
    },
    {
      "epoch": 1.3132944793722179,
      "grad_norm": 0.5081523656845093,
      "learning_rate": 6.147578050341129e-06,
      "loss": 0.4153,
      "step": 28618
    },
    {
      "epoch": 1.3133403698774724,
      "grad_norm": 0.4890824258327484,
      "learning_rate": 6.147339406350352e-06,
      "loss": 0.3518,
      "step": 28619
    },
    {
      "epoch": 1.3133862603827269,
      "grad_norm": 0.5079331994056702,
      "learning_rate": 6.147100759600587e-06,
      "loss": 0.3607,
      "step": 28620
    },
    {
      "epoch": 1.3134321508879814,
      "grad_norm": 0.47907009720802307,
      "learning_rate": 6.146862110092405e-06,
      "loss": 0.4094,
      "step": 28621
    },
    {
      "epoch": 1.3134780413932359,
      "grad_norm": 0.4969707727432251,
      "learning_rate": 6.146623457826382e-06,
      "loss": 0.3658,
      "step": 28622
    },
    {
      "epoch": 1.3135239318984901,
      "grad_norm": 0.4608663320541382,
      "learning_rate": 6.146384802803093e-06,
      "loss": 0.3578,
      "step": 28623
    },
    {
      "epoch": 1.3135698224037446,
      "grad_norm": 0.455472469329834,
      "learning_rate": 6.146146145023108e-06,
      "loss": 0.3626,
      "step": 28624
    },
    {
      "epoch": 1.3136157129089991,
      "grad_norm": 0.4790176749229431,
      "learning_rate": 6.145907484487006e-06,
      "loss": 0.3742,
      "step": 28625
    },
    {
      "epoch": 1.3136616034142536,
      "grad_norm": 0.4668760597705841,
      "learning_rate": 6.145668821195356e-06,
      "loss": 0.3469,
      "step": 28626
    },
    {
      "epoch": 1.313707493919508,
      "grad_norm": 0.48299357295036316,
      "learning_rate": 6.145430155148735e-06,
      "loss": 0.3808,
      "step": 28627
    },
    {
      "epoch": 1.3137533844247624,
      "grad_norm": 0.4882251024246216,
      "learning_rate": 6.145191486347715e-06,
      "loss": 0.388,
      "step": 28628
    },
    {
      "epoch": 1.313799274930017,
      "grad_norm": 0.4279222786426544,
      "learning_rate": 6.144952814792872e-06,
      "loss": 0.2825,
      "step": 28629
    },
    {
      "epoch": 1.3138451654352714,
      "grad_norm": 0.5230405926704407,
      "learning_rate": 6.144714140484779e-06,
      "loss": 0.3894,
      "step": 28630
    },
    {
      "epoch": 1.313891055940526,
      "grad_norm": 0.4978478252887726,
      "learning_rate": 6.144475463424009e-06,
      "loss": 0.3642,
      "step": 28631
    },
    {
      "epoch": 1.3139369464457804,
      "grad_norm": 0.47039347887039185,
      "learning_rate": 6.144236783611137e-06,
      "loss": 0.3923,
      "step": 28632
    },
    {
      "epoch": 1.313982836951035,
      "grad_norm": 0.4302489757537842,
      "learning_rate": 6.143998101046736e-06,
      "loss": 0.2861,
      "step": 28633
    },
    {
      "epoch": 1.3140287274562894,
      "grad_norm": 0.46030721068382263,
      "learning_rate": 6.143759415731382e-06,
      "loss": 0.3553,
      "step": 28634
    },
    {
      "epoch": 1.3140746179615437,
      "grad_norm": 0.44281432032585144,
      "learning_rate": 6.143520727665646e-06,
      "loss": 0.2706,
      "step": 28635
    },
    {
      "epoch": 1.3141205084667982,
      "grad_norm": 0.5107075572013855,
      "learning_rate": 6.143282036850105e-06,
      "loss": 0.4007,
      "step": 28636
    },
    {
      "epoch": 1.3141663989720527,
      "grad_norm": 0.46805277466773987,
      "learning_rate": 6.143043343285331e-06,
      "loss": 0.4281,
      "step": 28637
    },
    {
      "epoch": 1.3142122894773072,
      "grad_norm": 0.48218128085136414,
      "learning_rate": 6.142804646971898e-06,
      "loss": 0.3595,
      "step": 28638
    },
    {
      "epoch": 1.3142581799825617,
      "grad_norm": 0.5096114873886108,
      "learning_rate": 6.1425659479103805e-06,
      "loss": 0.4215,
      "step": 28639
    },
    {
      "epoch": 1.314304070487816,
      "grad_norm": 0.4957711398601532,
      "learning_rate": 6.142327246101354e-06,
      "loss": 0.4389,
      "step": 28640
    },
    {
      "epoch": 1.3143499609930704,
      "grad_norm": 0.45786359906196594,
      "learning_rate": 6.14208854154539e-06,
      "loss": 0.3458,
      "step": 28641
    },
    {
      "epoch": 1.314395851498325,
      "grad_norm": 0.4453641176223755,
      "learning_rate": 6.141849834243063e-06,
      "loss": 0.3449,
      "step": 28642
    },
    {
      "epoch": 1.3144417420035794,
      "grad_norm": 0.5161198973655701,
      "learning_rate": 6.141611124194948e-06,
      "loss": 0.3953,
      "step": 28643
    },
    {
      "epoch": 1.314487632508834,
      "grad_norm": 0.4899829626083374,
      "learning_rate": 6.141372411401618e-06,
      "loss": 0.3986,
      "step": 28644
    },
    {
      "epoch": 1.3145335230140884,
      "grad_norm": 0.4501853287220001,
      "learning_rate": 6.1411336958636495e-06,
      "loss": 0.3136,
      "step": 28645
    },
    {
      "epoch": 1.314579413519343,
      "grad_norm": 0.47929906845092773,
      "learning_rate": 6.140894977581612e-06,
      "loss": 0.3416,
      "step": 28646
    },
    {
      "epoch": 1.3146253040245974,
      "grad_norm": 0.46178215742111206,
      "learning_rate": 6.140656256556083e-06,
      "loss": 0.3903,
      "step": 28647
    },
    {
      "epoch": 1.3146711945298517,
      "grad_norm": 0.4913558065891266,
      "learning_rate": 6.140417532787637e-06,
      "loss": 0.3957,
      "step": 28648
    },
    {
      "epoch": 1.3147170850351062,
      "grad_norm": 0.45453307032585144,
      "learning_rate": 6.140178806276846e-06,
      "loss": 0.3552,
      "step": 28649
    },
    {
      "epoch": 1.3147629755403607,
      "grad_norm": 0.4590317904949188,
      "learning_rate": 6.1399400770242835e-06,
      "loss": 0.3189,
      "step": 28650
    },
    {
      "epoch": 1.3148088660456152,
      "grad_norm": 0.47262808680534363,
      "learning_rate": 6.139701345030526e-06,
      "loss": 0.3828,
      "step": 28651
    },
    {
      "epoch": 1.3148547565508697,
      "grad_norm": 0.46079179644584656,
      "learning_rate": 6.139462610296147e-06,
      "loss": 0.3459,
      "step": 28652
    },
    {
      "epoch": 1.314900647056124,
      "grad_norm": 0.48321473598480225,
      "learning_rate": 6.13922387282172e-06,
      "loss": 0.3425,
      "step": 28653
    },
    {
      "epoch": 1.3149465375613785,
      "grad_norm": 0.41569674015045166,
      "learning_rate": 6.138985132607819e-06,
      "loss": 0.2977,
      "step": 28654
    },
    {
      "epoch": 1.314992428066633,
      "grad_norm": 0.4783664047718048,
      "learning_rate": 6.138746389655017e-06,
      "loss": 0.3426,
      "step": 28655
    },
    {
      "epoch": 1.3150383185718875,
      "grad_norm": 0.5073363184928894,
      "learning_rate": 6.138507643963891e-06,
      "loss": 0.4288,
      "step": 28656
    },
    {
      "epoch": 1.315084209077142,
      "grad_norm": 0.5021705627441406,
      "learning_rate": 6.138268895535012e-06,
      "loss": 0.3827,
      "step": 28657
    },
    {
      "epoch": 1.3151300995823965,
      "grad_norm": 0.4829600155353546,
      "learning_rate": 6.138030144368957e-06,
      "loss": 0.4518,
      "step": 28658
    },
    {
      "epoch": 1.315175990087651,
      "grad_norm": 0.44453486800193787,
      "learning_rate": 6.137791390466299e-06,
      "loss": 0.316,
      "step": 28659
    },
    {
      "epoch": 1.3152218805929055,
      "grad_norm": 0.43608519434928894,
      "learning_rate": 6.1375526338276096e-06,
      "loss": 0.3149,
      "step": 28660
    },
    {
      "epoch": 1.3152677710981597,
      "grad_norm": 0.48824161291122437,
      "learning_rate": 6.137313874453466e-06,
      "loss": 0.3727,
      "step": 28661
    },
    {
      "epoch": 1.3153136616034142,
      "grad_norm": 0.46130818128585815,
      "learning_rate": 6.137075112344441e-06,
      "loss": 0.381,
      "step": 28662
    },
    {
      "epoch": 1.3153595521086687,
      "grad_norm": 0.4756026566028595,
      "learning_rate": 6.1368363475011116e-06,
      "loss": 0.3186,
      "step": 28663
    },
    {
      "epoch": 1.3154054426139232,
      "grad_norm": 0.506165087223053,
      "learning_rate": 6.136597579924047e-06,
      "loss": 0.3983,
      "step": 28664
    },
    {
      "epoch": 1.3154513331191777,
      "grad_norm": 0.4536078870296478,
      "learning_rate": 6.136358809613826e-06,
      "loss": 0.3175,
      "step": 28665
    },
    {
      "epoch": 1.315497223624432,
      "grad_norm": 0.43774083256721497,
      "learning_rate": 6.136120036571019e-06,
      "loss": 0.3168,
      "step": 28666
    },
    {
      "epoch": 1.3155431141296865,
      "grad_norm": 0.4618346691131592,
      "learning_rate": 6.135881260796202e-06,
      "loss": 0.3758,
      "step": 28667
    },
    {
      "epoch": 1.315589004634941,
      "grad_norm": 0.4773160219192505,
      "learning_rate": 6.1356424822899494e-06,
      "loss": 0.354,
      "step": 28668
    },
    {
      "epoch": 1.3156348951401955,
      "grad_norm": 0.4365794062614441,
      "learning_rate": 6.135403701052836e-06,
      "loss": 0.3176,
      "step": 28669
    },
    {
      "epoch": 1.31568078564545,
      "grad_norm": 0.5028301477432251,
      "learning_rate": 6.135164917085433e-06,
      "loss": 0.4069,
      "step": 28670
    },
    {
      "epoch": 1.3157266761507045,
      "grad_norm": 0.47298485040664673,
      "learning_rate": 6.134926130388319e-06,
      "loss": 0.3439,
      "step": 28671
    },
    {
      "epoch": 1.315772566655959,
      "grad_norm": 0.47862347960472107,
      "learning_rate": 6.134687340962064e-06,
      "loss": 0.375,
      "step": 28672
    },
    {
      "epoch": 1.3158184571612133,
      "grad_norm": 0.4909436106681824,
      "learning_rate": 6.134448548807246e-06,
      "loss": 0.3979,
      "step": 28673
    },
    {
      "epoch": 1.3158643476664678,
      "grad_norm": 0.47980645298957825,
      "learning_rate": 6.134209753924436e-06,
      "loss": 0.3455,
      "step": 28674
    },
    {
      "epoch": 1.3159102381717223,
      "grad_norm": 0.4563957452774048,
      "learning_rate": 6.133970956314209e-06,
      "loss": 0.3092,
      "step": 28675
    },
    {
      "epoch": 1.3159561286769768,
      "grad_norm": 0.46782323718070984,
      "learning_rate": 6.1337321559771415e-06,
      "loss": 0.3481,
      "step": 28676
    },
    {
      "epoch": 1.3160020191822313,
      "grad_norm": 0.4639793038368225,
      "learning_rate": 6.133493352913804e-06,
      "loss": 0.3211,
      "step": 28677
    },
    {
      "epoch": 1.3160479096874855,
      "grad_norm": 0.45938003063201904,
      "learning_rate": 6.133254547124773e-06,
      "loss": 0.3195,
      "step": 28678
    },
    {
      "epoch": 1.31609380019274,
      "grad_norm": 0.4461861252784729,
      "learning_rate": 6.133015738610622e-06,
      "loss": 0.3176,
      "step": 28679
    },
    {
      "epoch": 1.3161396906979945,
      "grad_norm": 0.4675120711326599,
      "learning_rate": 6.132776927371929e-06,
      "loss": 0.3625,
      "step": 28680
    },
    {
      "epoch": 1.316185581203249,
      "grad_norm": 0.4619564414024353,
      "learning_rate": 6.132538113409261e-06,
      "loss": 0.3193,
      "step": 28681
    },
    {
      "epoch": 1.3162314717085035,
      "grad_norm": 0.46000364422798157,
      "learning_rate": 6.132299296723199e-06,
      "loss": 0.3481,
      "step": 28682
    },
    {
      "epoch": 1.316277362213758,
      "grad_norm": 0.4760548174381256,
      "learning_rate": 6.132060477314313e-06,
      "loss": 0.3845,
      "step": 28683
    },
    {
      "epoch": 1.3163232527190125,
      "grad_norm": 0.47714975476264954,
      "learning_rate": 6.13182165518318e-06,
      "loss": 0.4089,
      "step": 28684
    },
    {
      "epoch": 1.316369143224267,
      "grad_norm": 0.514667809009552,
      "learning_rate": 6.131582830330373e-06,
      "loss": 0.4341,
      "step": 28685
    },
    {
      "epoch": 1.3164150337295213,
      "grad_norm": 0.47793710231781006,
      "learning_rate": 6.131344002756465e-06,
      "loss": 0.3515,
      "step": 28686
    },
    {
      "epoch": 1.3164609242347758,
      "grad_norm": 0.4991830587387085,
      "learning_rate": 6.131105172462033e-06,
      "loss": 0.367,
      "step": 28687
    },
    {
      "epoch": 1.3165068147400303,
      "grad_norm": 0.4995472729206085,
      "learning_rate": 6.13086633944765e-06,
      "loss": 0.4261,
      "step": 28688
    },
    {
      "epoch": 1.3165527052452848,
      "grad_norm": 0.4544256329536438,
      "learning_rate": 6.130627503713889e-06,
      "loss": 0.3499,
      "step": 28689
    },
    {
      "epoch": 1.3165985957505393,
      "grad_norm": 0.4764747619628906,
      "learning_rate": 6.130388665261328e-06,
      "loss": 0.3816,
      "step": 28690
    },
    {
      "epoch": 1.3166444862557936,
      "grad_norm": 0.481368750333786,
      "learning_rate": 6.130149824090538e-06,
      "loss": 0.3675,
      "step": 28691
    },
    {
      "epoch": 1.316690376761048,
      "grad_norm": 0.48215124011039734,
      "learning_rate": 6.1299109802020936e-06,
      "loss": 0.3744,
      "step": 28692
    },
    {
      "epoch": 1.3167362672663026,
      "grad_norm": 0.45092716813087463,
      "learning_rate": 6.129672133596572e-06,
      "loss": 0.3616,
      "step": 28693
    },
    {
      "epoch": 1.316782157771557,
      "grad_norm": 0.5013589859008789,
      "learning_rate": 6.129433284274543e-06,
      "loss": 0.3956,
      "step": 28694
    },
    {
      "epoch": 1.3168280482768115,
      "grad_norm": 0.4437345862388611,
      "learning_rate": 6.129194432236585e-06,
      "loss": 0.3213,
      "step": 28695
    },
    {
      "epoch": 1.316873938782066,
      "grad_norm": 0.5237861275672913,
      "learning_rate": 6.12895557748327e-06,
      "loss": 0.4029,
      "step": 28696
    },
    {
      "epoch": 1.3169198292873205,
      "grad_norm": 0.44656768441200256,
      "learning_rate": 6.128716720015173e-06,
      "loss": 0.3255,
      "step": 28697
    },
    {
      "epoch": 1.316965719792575,
      "grad_norm": 0.46499359607696533,
      "learning_rate": 6.128477859832869e-06,
      "loss": 0.3676,
      "step": 28698
    },
    {
      "epoch": 1.3170116102978293,
      "grad_norm": 0.46420058608055115,
      "learning_rate": 6.1282389969369316e-06,
      "loss": 0.3626,
      "step": 28699
    },
    {
      "epoch": 1.3170575008030838,
      "grad_norm": 0.46368592977523804,
      "learning_rate": 6.128000131327935e-06,
      "loss": 0.3572,
      "step": 28700
    },
    {
      "epoch": 1.3171033913083383,
      "grad_norm": 0.4811917841434479,
      "learning_rate": 6.127761263006455e-06,
      "loss": 0.3889,
      "step": 28701
    },
    {
      "epoch": 1.3171492818135928,
      "grad_norm": 0.45587828755378723,
      "learning_rate": 6.127522391973065e-06,
      "loss": 0.3427,
      "step": 28702
    },
    {
      "epoch": 1.3171951723188473,
      "grad_norm": 0.48763564229011536,
      "learning_rate": 6.12728351822834e-06,
      "loss": 0.3771,
      "step": 28703
    },
    {
      "epoch": 1.3172410628241016,
      "grad_norm": 0.4185521900653839,
      "learning_rate": 6.127044641772851e-06,
      "loss": 0.3012,
      "step": 28704
    },
    {
      "epoch": 1.317286953329356,
      "grad_norm": 0.4987112879753113,
      "learning_rate": 6.1268057626071785e-06,
      "loss": 0.4733,
      "step": 28705
    },
    {
      "epoch": 1.3173328438346106,
      "grad_norm": 0.48191723227500916,
      "learning_rate": 6.126566880731893e-06,
      "loss": 0.42,
      "step": 28706
    },
    {
      "epoch": 1.317378734339865,
      "grad_norm": 0.5103091597557068,
      "learning_rate": 6.1263279961475705e-06,
      "loss": 0.4178,
      "step": 28707
    },
    {
      "epoch": 1.3174246248451196,
      "grad_norm": 0.45616957545280457,
      "learning_rate": 6.126089108854784e-06,
      "loss": 0.3446,
      "step": 28708
    },
    {
      "epoch": 1.317470515350374,
      "grad_norm": 0.4339410662651062,
      "learning_rate": 6.125850218854107e-06,
      "loss": 0.3436,
      "step": 28709
    },
    {
      "epoch": 1.3175164058556286,
      "grad_norm": 0.4546510577201843,
      "learning_rate": 6.125611326146117e-06,
      "loss": 0.3351,
      "step": 28710
    },
    {
      "epoch": 1.317562296360883,
      "grad_norm": 0.4705520272254944,
      "learning_rate": 6.125372430731387e-06,
      "loss": 0.3167,
      "step": 28711
    },
    {
      "epoch": 1.3176081868661373,
      "grad_norm": 0.44667354226112366,
      "learning_rate": 6.125133532610492e-06,
      "loss": 0.3184,
      "step": 28712
    },
    {
      "epoch": 1.3176540773713918,
      "grad_norm": 0.482857882976532,
      "learning_rate": 6.124894631784005e-06,
      "loss": 0.348,
      "step": 28713
    },
    {
      "epoch": 1.3176999678766463,
      "grad_norm": 0.45483601093292236,
      "learning_rate": 6.124655728252502e-06,
      "loss": 0.2971,
      "step": 28714
    },
    {
      "epoch": 1.3177458583819008,
      "grad_norm": 0.4621037542819977,
      "learning_rate": 6.124416822016556e-06,
      "loss": 0.3932,
      "step": 28715
    },
    {
      "epoch": 1.3177917488871551,
      "grad_norm": 0.49478116631507874,
      "learning_rate": 6.124177913076743e-06,
      "loss": 0.3786,
      "step": 28716
    },
    {
      "epoch": 1.3178376393924096,
      "grad_norm": 0.45734739303588867,
      "learning_rate": 6.123939001433637e-06,
      "loss": 0.3459,
      "step": 28717
    },
    {
      "epoch": 1.317883529897664,
      "grad_norm": 0.5198931097984314,
      "learning_rate": 6.123700087087813e-06,
      "loss": 0.4695,
      "step": 28718
    },
    {
      "epoch": 1.3179294204029186,
      "grad_norm": 0.5072784423828125,
      "learning_rate": 6.123461170039844e-06,
      "loss": 0.4385,
      "step": 28719
    },
    {
      "epoch": 1.317975310908173,
      "grad_norm": 0.4468470513820648,
      "learning_rate": 6.123222250290306e-06,
      "loss": 0.3,
      "step": 28720
    },
    {
      "epoch": 1.3180212014134276,
      "grad_norm": 0.47366008162498474,
      "learning_rate": 6.122983327839772e-06,
      "loss": 0.3349,
      "step": 28721
    },
    {
      "epoch": 1.318067091918682,
      "grad_norm": 0.4719436764717102,
      "learning_rate": 6.122744402688818e-06,
      "loss": 0.4105,
      "step": 28722
    },
    {
      "epoch": 1.3181129824239366,
      "grad_norm": 0.4298560917377472,
      "learning_rate": 6.12250547483802e-06,
      "loss": 0.2977,
      "step": 28723
    },
    {
      "epoch": 1.3181588729291909,
      "grad_norm": 0.43682661652565,
      "learning_rate": 6.122266544287949e-06,
      "loss": 0.3409,
      "step": 28724
    },
    {
      "epoch": 1.3182047634344454,
      "grad_norm": 0.46767759323120117,
      "learning_rate": 6.122027611039181e-06,
      "loss": 0.3625,
      "step": 28725
    },
    {
      "epoch": 1.3182506539396999,
      "grad_norm": 0.4883808493614197,
      "learning_rate": 6.12178867509229e-06,
      "loss": 0.3525,
      "step": 28726
    },
    {
      "epoch": 1.3182965444449544,
      "grad_norm": 0.4817774295806885,
      "learning_rate": 6.121549736447852e-06,
      "loss": 0.4221,
      "step": 28727
    },
    {
      "epoch": 1.3183424349502089,
      "grad_norm": 0.5234060883522034,
      "learning_rate": 6.121310795106442e-06,
      "loss": 0.4346,
      "step": 28728
    },
    {
      "epoch": 1.3183883254554631,
      "grad_norm": 0.45140326023101807,
      "learning_rate": 6.121071851068632e-06,
      "loss": 0.3188,
      "step": 28729
    },
    {
      "epoch": 1.3184342159607176,
      "grad_norm": 0.46804988384246826,
      "learning_rate": 6.120832904334999e-06,
      "loss": 0.3647,
      "step": 28730
    },
    {
      "epoch": 1.3184801064659721,
      "grad_norm": 0.5010818839073181,
      "learning_rate": 6.120593954906115e-06,
      "loss": 0.4082,
      "step": 28731
    },
    {
      "epoch": 1.3185259969712266,
      "grad_norm": 0.5314507484436035,
      "learning_rate": 6.1203550027825576e-06,
      "loss": 0.3899,
      "step": 28732
    },
    {
      "epoch": 1.3185718874764811,
      "grad_norm": 0.4837670624256134,
      "learning_rate": 6.1201160479649005e-06,
      "loss": 0.3857,
      "step": 28733
    },
    {
      "epoch": 1.3186177779817356,
      "grad_norm": 0.4791371524333954,
      "learning_rate": 6.119877090453718e-06,
      "loss": 0.3845,
      "step": 28734
    },
    {
      "epoch": 1.3186636684869901,
      "grad_norm": 0.42712026834487915,
      "learning_rate": 6.119638130249582e-06,
      "loss": 0.2916,
      "step": 28735
    },
    {
      "epoch": 1.3187095589922446,
      "grad_norm": 0.4438757300376892,
      "learning_rate": 6.1193991673530716e-06,
      "loss": 0.3434,
      "step": 28736
    },
    {
      "epoch": 1.318755449497499,
      "grad_norm": 0.45712098479270935,
      "learning_rate": 6.119160201764759e-06,
      "loss": 0.3206,
      "step": 28737
    },
    {
      "epoch": 1.3188013400027534,
      "grad_norm": 0.5130568146705627,
      "learning_rate": 6.11892123348522e-06,
      "loss": 0.439,
      "step": 28738
    },
    {
      "epoch": 1.318847230508008,
      "grad_norm": 0.44837361574172974,
      "learning_rate": 6.118682262515028e-06,
      "loss": 0.3406,
      "step": 28739
    },
    {
      "epoch": 1.3188931210132624,
      "grad_norm": 0.4731118977069855,
      "learning_rate": 6.118443288854759e-06,
      "loss": 0.3967,
      "step": 28740
    },
    {
      "epoch": 1.318939011518517,
      "grad_norm": 0.46926721930503845,
      "learning_rate": 6.118204312504985e-06,
      "loss": 0.3784,
      "step": 28741
    },
    {
      "epoch": 1.3189849020237712,
      "grad_norm": 0.47100934386253357,
      "learning_rate": 6.117965333466283e-06,
      "loss": 0.3429,
      "step": 28742
    },
    {
      "epoch": 1.3190307925290257,
      "grad_norm": 0.4595562517642975,
      "learning_rate": 6.117726351739228e-06,
      "loss": 0.3985,
      "step": 28743
    },
    {
      "epoch": 1.3190766830342802,
      "grad_norm": 0.4745666980743408,
      "learning_rate": 6.117487367324394e-06,
      "loss": 0.3658,
      "step": 28744
    },
    {
      "epoch": 1.3191225735395347,
      "grad_norm": 0.4708336889743805,
      "learning_rate": 6.117248380222356e-06,
      "loss": 0.3359,
      "step": 28745
    },
    {
      "epoch": 1.3191684640447892,
      "grad_norm": 0.4649316370487213,
      "learning_rate": 6.117009390433687e-06,
      "loss": 0.3601,
      "step": 28746
    },
    {
      "epoch": 1.3192143545500437,
      "grad_norm": 0.49175596237182617,
      "learning_rate": 6.116770397958965e-06,
      "loss": 0.3862,
      "step": 28747
    },
    {
      "epoch": 1.3192602450552982,
      "grad_norm": 0.4427841901779175,
      "learning_rate": 6.11653140279876e-06,
      "loss": 0.3127,
      "step": 28748
    },
    {
      "epoch": 1.3193061355605527,
      "grad_norm": 0.3980049788951874,
      "learning_rate": 6.116292404953651e-06,
      "loss": 0.2641,
      "step": 28749
    },
    {
      "epoch": 1.319352026065807,
      "grad_norm": 0.523370623588562,
      "learning_rate": 6.116053404424212e-06,
      "loss": 0.4591,
      "step": 28750
    },
    {
      "epoch": 1.3193979165710614,
      "grad_norm": 0.47151267528533936,
      "learning_rate": 6.1158144012110145e-06,
      "loss": 0.3701,
      "step": 28751
    },
    {
      "epoch": 1.319443807076316,
      "grad_norm": 0.49633678793907166,
      "learning_rate": 6.115575395314636e-06,
      "loss": 0.3905,
      "step": 28752
    },
    {
      "epoch": 1.3194896975815704,
      "grad_norm": 0.48400455713272095,
      "learning_rate": 6.115336386735651e-06,
      "loss": 0.3985,
      "step": 28753
    },
    {
      "epoch": 1.319535588086825,
      "grad_norm": 0.48351743817329407,
      "learning_rate": 6.115097375474634e-06,
      "loss": 0.3817,
      "step": 28754
    },
    {
      "epoch": 1.3195814785920792,
      "grad_norm": 0.4967452883720398,
      "learning_rate": 6.11485836153216e-06,
      "loss": 0.3872,
      "step": 28755
    },
    {
      "epoch": 1.3196273690973337,
      "grad_norm": 0.46331241726875305,
      "learning_rate": 6.114619344908804e-06,
      "loss": 0.375,
      "step": 28756
    },
    {
      "epoch": 1.3196732596025882,
      "grad_norm": 0.5013239979743958,
      "learning_rate": 6.114380325605139e-06,
      "loss": 0.4031,
      "step": 28757
    },
    {
      "epoch": 1.3197191501078427,
      "grad_norm": 0.45188966393470764,
      "learning_rate": 6.114141303621743e-06,
      "loss": 0.3618,
      "step": 28758
    },
    {
      "epoch": 1.3197650406130972,
      "grad_norm": 0.52292799949646,
      "learning_rate": 6.113902278959187e-06,
      "loss": 0.4792,
      "step": 28759
    },
    {
      "epoch": 1.3198109311183517,
      "grad_norm": 0.45787322521209717,
      "learning_rate": 6.113663251618048e-06,
      "loss": 0.3082,
      "step": 28760
    },
    {
      "epoch": 1.3198568216236062,
      "grad_norm": 0.5172033309936523,
      "learning_rate": 6.113424221598902e-06,
      "loss": 0.4238,
      "step": 28761
    },
    {
      "epoch": 1.3199027121288605,
      "grad_norm": 0.47714874148368835,
      "learning_rate": 6.11318518890232e-06,
      "loss": 0.3141,
      "step": 28762
    },
    {
      "epoch": 1.319948602634115,
      "grad_norm": 0.49230214953422546,
      "learning_rate": 6.1129461535288785e-06,
      "loss": 0.3783,
      "step": 28763
    },
    {
      "epoch": 1.3199944931393695,
      "grad_norm": 0.49386778473854065,
      "learning_rate": 6.112707115479155e-06,
      "loss": 0.3934,
      "step": 28764
    },
    {
      "epoch": 1.320040383644624,
      "grad_norm": 0.4632430076599121,
      "learning_rate": 6.11246807475372e-06,
      "loss": 0.3342,
      "step": 28765
    },
    {
      "epoch": 1.3200862741498784,
      "grad_norm": 0.48206397891044617,
      "learning_rate": 6.112229031353151e-06,
      "loss": 0.3254,
      "step": 28766
    },
    {
      "epoch": 1.3201321646551327,
      "grad_norm": 0.43939170241355896,
      "learning_rate": 6.111989985278024e-06,
      "loss": 0.3064,
      "step": 28767
    },
    {
      "epoch": 1.3201780551603872,
      "grad_norm": 0.4627799093723297,
      "learning_rate": 6.11175093652891e-06,
      "loss": 0.325,
      "step": 28768
    },
    {
      "epoch": 1.3202239456656417,
      "grad_norm": 0.5178590416908264,
      "learning_rate": 6.111511885106386e-06,
      "loss": 0.3238,
      "step": 28769
    },
    {
      "epoch": 1.3202698361708962,
      "grad_norm": 0.4612278640270233,
      "learning_rate": 6.111272831011027e-06,
      "loss": 0.348,
      "step": 28770
    },
    {
      "epoch": 1.3203157266761507,
      "grad_norm": 0.448106586933136,
      "learning_rate": 6.111033774243409e-06,
      "loss": 0.3062,
      "step": 28771
    },
    {
      "epoch": 1.3203616171814052,
      "grad_norm": 0.47459936141967773,
      "learning_rate": 6.1107947148041046e-06,
      "loss": 0.3743,
      "step": 28772
    },
    {
      "epoch": 1.3204075076866597,
      "grad_norm": 0.4853750467300415,
      "learning_rate": 6.110555652693688e-06,
      "loss": 0.3799,
      "step": 28773
    },
    {
      "epoch": 1.3204533981919142,
      "grad_norm": 0.5146495699882507,
      "learning_rate": 6.1103165879127355e-06,
      "loss": 0.4869,
      "step": 28774
    },
    {
      "epoch": 1.3204992886971685,
      "grad_norm": 0.4930976927280426,
      "learning_rate": 6.110077520461824e-06,
      "loss": 0.2907,
      "step": 28775
    },
    {
      "epoch": 1.320545179202423,
      "grad_norm": 0.49036678671836853,
      "learning_rate": 6.109838450341525e-06,
      "loss": 0.4013,
      "step": 28776
    },
    {
      "epoch": 1.3205910697076775,
      "grad_norm": 0.47645196318626404,
      "learning_rate": 6.109599377552417e-06,
      "loss": 0.3734,
      "step": 28777
    },
    {
      "epoch": 1.320636960212932,
      "grad_norm": 0.45227330923080444,
      "learning_rate": 6.109360302095071e-06,
      "loss": 0.3111,
      "step": 28778
    },
    {
      "epoch": 1.3206828507181865,
      "grad_norm": 0.5247656106948853,
      "learning_rate": 6.109121223970063e-06,
      "loss": 0.4283,
      "step": 28779
    },
    {
      "epoch": 1.3207287412234407,
      "grad_norm": 0.40907084941864014,
      "learning_rate": 6.108882143177969e-06,
      "loss": 0.2603,
      "step": 28780
    },
    {
      "epoch": 1.3207746317286952,
      "grad_norm": 0.4714578688144684,
      "learning_rate": 6.108643059719364e-06,
      "loss": 0.3833,
      "step": 28781
    },
    {
      "epoch": 1.3208205222339497,
      "grad_norm": 0.45785200595855713,
      "learning_rate": 6.108403973594822e-06,
      "loss": 0.341,
      "step": 28782
    },
    {
      "epoch": 1.3208664127392042,
      "grad_norm": 0.47247743606567383,
      "learning_rate": 6.108164884804918e-06,
      "loss": 0.3735,
      "step": 28783
    },
    {
      "epoch": 1.3209123032444587,
      "grad_norm": 0.4861489534378052,
      "learning_rate": 6.107925793350228e-06,
      "loss": 0.3951,
      "step": 28784
    },
    {
      "epoch": 1.3209581937497132,
      "grad_norm": 0.4769836664199829,
      "learning_rate": 6.107686699231325e-06,
      "loss": 0.3886,
      "step": 28785
    },
    {
      "epoch": 1.3210040842549677,
      "grad_norm": 0.4841054677963257,
      "learning_rate": 6.107447602448784e-06,
      "loss": 0.4031,
      "step": 28786
    },
    {
      "epoch": 1.3210499747602222,
      "grad_norm": 0.4911634027957916,
      "learning_rate": 6.107208503003184e-06,
      "loss": 0.3792,
      "step": 28787
    },
    {
      "epoch": 1.3210958652654765,
      "grad_norm": 0.4687567949295044,
      "learning_rate": 6.106969400895095e-06,
      "loss": 0.3489,
      "step": 28788
    },
    {
      "epoch": 1.321141755770731,
      "grad_norm": 0.47050556540489197,
      "learning_rate": 6.106730296125096e-06,
      "loss": 0.3502,
      "step": 28789
    },
    {
      "epoch": 1.3211876462759855,
      "grad_norm": 0.550982654094696,
      "learning_rate": 6.106491188693759e-06,
      "loss": 0.4271,
      "step": 28790
    },
    {
      "epoch": 1.32123353678124,
      "grad_norm": 0.478751540184021,
      "learning_rate": 6.106252078601658e-06,
      "loss": 0.3733,
      "step": 28791
    },
    {
      "epoch": 1.3212794272864945,
      "grad_norm": 0.45403558015823364,
      "learning_rate": 6.106012965849372e-06,
      "loss": 0.3514,
      "step": 28792
    },
    {
      "epoch": 1.3213253177917488,
      "grad_norm": 0.5096951723098755,
      "learning_rate": 6.105773850437473e-06,
      "loss": 0.4177,
      "step": 28793
    },
    {
      "epoch": 1.3213712082970033,
      "grad_norm": 0.4999421536922455,
      "learning_rate": 6.105534732366537e-06,
      "loss": 0.4599,
      "step": 28794
    },
    {
      "epoch": 1.3214170988022578,
      "grad_norm": 0.44926419854164124,
      "learning_rate": 6.10529561163714e-06,
      "loss": 0.3015,
      "step": 28795
    },
    {
      "epoch": 1.3214629893075123,
      "grad_norm": 0.48875126242637634,
      "learning_rate": 6.105056488249854e-06,
      "loss": 0.3779,
      "step": 28796
    },
    {
      "epoch": 1.3215088798127668,
      "grad_norm": 0.5126500725746155,
      "learning_rate": 6.104817362205255e-06,
      "loss": 0.4675,
      "step": 28797
    },
    {
      "epoch": 1.3215547703180213,
      "grad_norm": 0.4920293688774109,
      "learning_rate": 6.104578233503922e-06,
      "loss": 0.37,
      "step": 28798
    },
    {
      "epoch": 1.3216006608232758,
      "grad_norm": 0.4848904013633728,
      "learning_rate": 6.104339102146426e-06,
      "loss": 0.3925,
      "step": 28799
    },
    {
      "epoch": 1.3216465513285303,
      "grad_norm": 0.4011046886444092,
      "learning_rate": 6.104099968133342e-06,
      "loss": 0.2607,
      "step": 28800
    },
    {
      "epoch": 1.3216924418337845,
      "grad_norm": 0.48108533024787903,
      "learning_rate": 6.103860831465248e-06,
      "loss": 0.3626,
      "step": 28801
    },
    {
      "epoch": 1.321738332339039,
      "grad_norm": 0.4518919289112091,
      "learning_rate": 6.103621692142715e-06,
      "loss": 0.3445,
      "step": 28802
    },
    {
      "epoch": 1.3217842228442935,
      "grad_norm": 0.43407300114631653,
      "learning_rate": 6.103382550166321e-06,
      "loss": 0.3009,
      "step": 28803
    },
    {
      "epoch": 1.321830113349548,
      "grad_norm": 0.4407515823841095,
      "learning_rate": 6.103143405536641e-06,
      "loss": 0.3267,
      "step": 28804
    },
    {
      "epoch": 1.3218760038548023,
      "grad_norm": 0.4812732934951782,
      "learning_rate": 6.1029042582542476e-06,
      "loss": 0.3565,
      "step": 28805
    },
    {
      "epoch": 1.3219218943600568,
      "grad_norm": 0.4357263147830963,
      "learning_rate": 6.1026651083197195e-06,
      "loss": 0.3004,
      "step": 28806
    },
    {
      "epoch": 1.3219677848653113,
      "grad_norm": 0.4157315790653229,
      "learning_rate": 6.102425955733628e-06,
      "loss": 0.2891,
      "step": 28807
    },
    {
      "epoch": 1.3220136753705658,
      "grad_norm": 0.49445128440856934,
      "learning_rate": 6.10218680049655e-06,
      "loss": 0.4284,
      "step": 28808
    },
    {
      "epoch": 1.3220595658758203,
      "grad_norm": 0.4910227358341217,
      "learning_rate": 6.101947642609062e-06,
      "loss": 0.4165,
      "step": 28809
    },
    {
      "epoch": 1.3221054563810748,
      "grad_norm": 0.48263034224510193,
      "learning_rate": 6.101708482071737e-06,
      "loss": 0.3931,
      "step": 28810
    },
    {
      "epoch": 1.3221513468863293,
      "grad_norm": 0.47633007168769836,
      "learning_rate": 6.10146931888515e-06,
      "loss": 0.3549,
      "step": 28811
    },
    {
      "epoch": 1.3221972373915838,
      "grad_norm": 0.4339072108268738,
      "learning_rate": 6.101230153049879e-06,
      "loss": 0.2901,
      "step": 28812
    },
    {
      "epoch": 1.322243127896838,
      "grad_norm": 0.45886462926864624,
      "learning_rate": 6.100990984566495e-06,
      "loss": 0.3436,
      "step": 28813
    },
    {
      "epoch": 1.3222890184020926,
      "grad_norm": 0.4765680730342865,
      "learning_rate": 6.100751813435576e-06,
      "loss": 0.3428,
      "step": 28814
    },
    {
      "epoch": 1.322334908907347,
      "grad_norm": 0.44993728399276733,
      "learning_rate": 6.1005126396576974e-06,
      "loss": 0.3583,
      "step": 28815
    },
    {
      "epoch": 1.3223807994126016,
      "grad_norm": 0.4260273873806,
      "learning_rate": 6.100273463233431e-06,
      "loss": 0.2781,
      "step": 28816
    },
    {
      "epoch": 1.322426689917856,
      "grad_norm": 0.48214298486709595,
      "learning_rate": 6.100034284163356e-06,
      "loss": 0.3684,
      "step": 28817
    },
    {
      "epoch": 1.3224725804231103,
      "grad_norm": 0.43614086508750916,
      "learning_rate": 6.099795102448045e-06,
      "loss": 0.338,
      "step": 28818
    },
    {
      "epoch": 1.3225184709283648,
      "grad_norm": 0.45613259077072144,
      "learning_rate": 6.099555918088074e-06,
      "loss": 0.3464,
      "step": 28819
    },
    {
      "epoch": 1.3225643614336193,
      "grad_norm": 0.4602162837982178,
      "learning_rate": 6.099316731084017e-06,
      "loss": 0.3385,
      "step": 28820
    },
    {
      "epoch": 1.3226102519388738,
      "grad_norm": 0.4515390992164612,
      "learning_rate": 6.099077541436452e-06,
      "loss": 0.3431,
      "step": 28821
    },
    {
      "epoch": 1.3226561424441283,
      "grad_norm": 0.46693888306617737,
      "learning_rate": 6.0988383491459514e-06,
      "loss": 0.3618,
      "step": 28822
    },
    {
      "epoch": 1.3227020329493828,
      "grad_norm": 0.5288559198379517,
      "learning_rate": 6.098599154213092e-06,
      "loss": 0.4536,
      "step": 28823
    },
    {
      "epoch": 1.3227479234546373,
      "grad_norm": 0.5215124487876892,
      "learning_rate": 6.098359956638447e-06,
      "loss": 0.4181,
      "step": 28824
    },
    {
      "epoch": 1.3227938139598918,
      "grad_norm": 0.4563755989074707,
      "learning_rate": 6.098120756422593e-06,
      "loss": 0.3309,
      "step": 28825
    },
    {
      "epoch": 1.322839704465146,
      "grad_norm": 0.505741536617279,
      "learning_rate": 6.0978815535661085e-06,
      "loss": 0.471,
      "step": 28826
    },
    {
      "epoch": 1.3228855949704006,
      "grad_norm": 0.4916747212409973,
      "learning_rate": 6.0976423480695615e-06,
      "loss": 0.3445,
      "step": 28827
    },
    {
      "epoch": 1.322931485475655,
      "grad_norm": 0.4388931691646576,
      "learning_rate": 6.097403139933532e-06,
      "loss": 0.3083,
      "step": 28828
    },
    {
      "epoch": 1.3229773759809096,
      "grad_norm": 0.4685806334018707,
      "learning_rate": 6.097163929158595e-06,
      "loss": 0.3806,
      "step": 28829
    },
    {
      "epoch": 1.323023266486164,
      "grad_norm": 0.4984460771083832,
      "learning_rate": 6.096924715745325e-06,
      "loss": 0.3961,
      "step": 28830
    },
    {
      "epoch": 1.3230691569914184,
      "grad_norm": 0.4493517577648163,
      "learning_rate": 6.096685499694297e-06,
      "loss": 0.3435,
      "step": 28831
    },
    {
      "epoch": 1.3231150474966729,
      "grad_norm": 0.48422810435295105,
      "learning_rate": 6.096446281006088e-06,
      "loss": 0.3803,
      "step": 28832
    },
    {
      "epoch": 1.3231609380019274,
      "grad_norm": 0.49591103196144104,
      "learning_rate": 6.096207059681268e-06,
      "loss": 0.4613,
      "step": 28833
    },
    {
      "epoch": 1.3232068285071819,
      "grad_norm": 0.4480378329753876,
      "learning_rate": 6.0959678357204185e-06,
      "loss": 0.305,
      "step": 28834
    },
    {
      "epoch": 1.3232527190124364,
      "grad_norm": 0.43708014488220215,
      "learning_rate": 6.095728609124112e-06,
      "loss": 0.33,
      "step": 28835
    },
    {
      "epoch": 1.3232986095176908,
      "grad_norm": 0.457009881734848,
      "learning_rate": 6.0954893798929235e-06,
      "loss": 0.3434,
      "step": 28836
    },
    {
      "epoch": 1.3233445000229453,
      "grad_norm": 0.47787559032440186,
      "learning_rate": 6.095250148027432e-06,
      "loss": 0.3901,
      "step": 28837
    },
    {
      "epoch": 1.3233903905281998,
      "grad_norm": 0.46493804454803467,
      "learning_rate": 6.095010913528206e-06,
      "loss": 0.3186,
      "step": 28838
    },
    {
      "epoch": 1.3234362810334541,
      "grad_norm": 0.47897231578826904,
      "learning_rate": 6.094771676395824e-06,
      "loss": 0.2995,
      "step": 28839
    },
    {
      "epoch": 1.3234821715387086,
      "grad_norm": 0.4451633095741272,
      "learning_rate": 6.0945324366308636e-06,
      "loss": 0.3042,
      "step": 28840
    },
    {
      "epoch": 1.3235280620439631,
      "grad_norm": 0.4683212339878082,
      "learning_rate": 6.094293194233897e-06,
      "loss": 0.336,
      "step": 28841
    },
    {
      "epoch": 1.3235739525492176,
      "grad_norm": 0.46995723247528076,
      "learning_rate": 6.0940539492055016e-06,
      "loss": 0.3423,
      "step": 28842
    },
    {
      "epoch": 1.3236198430544721,
      "grad_norm": 0.46864521503448486,
      "learning_rate": 6.093814701546251e-06,
      "loss": 0.3549,
      "step": 28843
    },
    {
      "epoch": 1.3236657335597264,
      "grad_norm": 0.4793601334095001,
      "learning_rate": 6.093575451256722e-06,
      "loss": 0.3764,
      "step": 28844
    },
    {
      "epoch": 1.3237116240649809,
      "grad_norm": 0.4651247262954712,
      "learning_rate": 6.0933361983374876e-06,
      "loss": 0.36,
      "step": 28845
    },
    {
      "epoch": 1.3237575145702354,
      "grad_norm": 0.48365968465805054,
      "learning_rate": 6.093096942789126e-06,
      "loss": 0.3849,
      "step": 28846
    },
    {
      "epoch": 1.3238034050754899,
      "grad_norm": 0.4886632561683655,
      "learning_rate": 6.09285768461221e-06,
      "loss": 0.3727,
      "step": 28847
    },
    {
      "epoch": 1.3238492955807444,
      "grad_norm": 0.4789888560771942,
      "learning_rate": 6.0926184238073185e-06,
      "loss": 0.3777,
      "step": 28848
    },
    {
      "epoch": 1.3238951860859989,
      "grad_norm": 0.44817668199539185,
      "learning_rate": 6.092379160375024e-06,
      "loss": 0.3378,
      "step": 28849
    },
    {
      "epoch": 1.3239410765912534,
      "grad_norm": 0.5150731801986694,
      "learning_rate": 6.092139894315901e-06,
      "loss": 0.405,
      "step": 28850
    },
    {
      "epoch": 1.3239869670965076,
      "grad_norm": 0.5056014060974121,
      "learning_rate": 6.0919006256305255e-06,
      "loss": 0.4361,
      "step": 28851
    },
    {
      "epoch": 1.3240328576017621,
      "grad_norm": 0.5149839520454407,
      "learning_rate": 6.091661354319476e-06,
      "loss": 0.4439,
      "step": 28852
    },
    {
      "epoch": 1.3240787481070166,
      "grad_norm": 0.4365101158618927,
      "learning_rate": 6.091422080383324e-06,
      "loss": 0.3577,
      "step": 28853
    },
    {
      "epoch": 1.3241246386122711,
      "grad_norm": 0.4697846472263336,
      "learning_rate": 6.0911828038226475e-06,
      "loss": 0.3603,
      "step": 28854
    },
    {
      "epoch": 1.3241705291175256,
      "grad_norm": 0.4836871027946472,
      "learning_rate": 6.09094352463802e-06,
      "loss": 0.3759,
      "step": 28855
    },
    {
      "epoch": 1.32421641962278,
      "grad_norm": 0.5125709176063538,
      "learning_rate": 6.090704242830017e-06,
      "loss": 0.4308,
      "step": 28856
    },
    {
      "epoch": 1.3242623101280344,
      "grad_norm": 0.4375849664211273,
      "learning_rate": 6.090464958399216e-06,
      "loss": 0.3054,
      "step": 28857
    },
    {
      "epoch": 1.324308200633289,
      "grad_norm": 0.4840388298034668,
      "learning_rate": 6.090225671346189e-06,
      "loss": 0.3877,
      "step": 28858
    },
    {
      "epoch": 1.3243540911385434,
      "grad_norm": 0.4650684595108032,
      "learning_rate": 6.089986381671516e-06,
      "loss": 0.4002,
      "step": 28859
    },
    {
      "epoch": 1.324399981643798,
      "grad_norm": 0.4824790954589844,
      "learning_rate": 6.089747089375768e-06,
      "loss": 0.4424,
      "step": 28860
    },
    {
      "epoch": 1.3244458721490524,
      "grad_norm": 0.4908510446548462,
      "learning_rate": 6.089507794459521e-06,
      "loss": 0.3785,
      "step": 28861
    },
    {
      "epoch": 1.324491762654307,
      "grad_norm": 0.4910569489002228,
      "learning_rate": 6.089268496923353e-06,
      "loss": 0.3366,
      "step": 28862
    },
    {
      "epoch": 1.3245376531595614,
      "grad_norm": 0.46018582582473755,
      "learning_rate": 6.089029196767838e-06,
      "loss": 0.3545,
      "step": 28863
    },
    {
      "epoch": 1.3245835436648157,
      "grad_norm": 0.50713050365448,
      "learning_rate": 6.088789893993553e-06,
      "loss": 0.3568,
      "step": 28864
    },
    {
      "epoch": 1.3246294341700702,
      "grad_norm": 0.5026248097419739,
      "learning_rate": 6.08855058860107e-06,
      "loss": 0.4023,
      "step": 28865
    },
    {
      "epoch": 1.3246753246753247,
      "grad_norm": 0.4650111794471741,
      "learning_rate": 6.088311280590967e-06,
      "loss": 0.3418,
      "step": 28866
    },
    {
      "epoch": 1.3247212151805792,
      "grad_norm": 0.42097482085227966,
      "learning_rate": 6.088071969963818e-06,
      "loss": 0.3022,
      "step": 28867
    },
    {
      "epoch": 1.3247671056858337,
      "grad_norm": 0.45232096314430237,
      "learning_rate": 6.087832656720199e-06,
      "loss": 0.3284,
      "step": 28868
    },
    {
      "epoch": 1.324812996191088,
      "grad_norm": 0.4476914405822754,
      "learning_rate": 6.087593340860688e-06,
      "loss": 0.2881,
      "step": 28869
    },
    {
      "epoch": 1.3248588866963424,
      "grad_norm": 0.4717225134372711,
      "learning_rate": 6.087354022385857e-06,
      "loss": 0.3248,
      "step": 28870
    },
    {
      "epoch": 1.324904777201597,
      "grad_norm": 0.42621830105781555,
      "learning_rate": 6.087114701296284e-06,
      "loss": 0.3406,
      "step": 28871
    },
    {
      "epoch": 1.3249506677068514,
      "grad_norm": 0.5103136897087097,
      "learning_rate": 6.086875377592542e-06,
      "loss": 0.322,
      "step": 28872
    },
    {
      "epoch": 1.324996558212106,
      "grad_norm": 0.48601871728897095,
      "learning_rate": 6.086636051275207e-06,
      "loss": 0.3731,
      "step": 28873
    },
    {
      "epoch": 1.3250424487173604,
      "grad_norm": 0.44567441940307617,
      "learning_rate": 6.086396722344856e-06,
      "loss": 0.308,
      "step": 28874
    },
    {
      "epoch": 1.325088339222615,
      "grad_norm": 0.4981589913368225,
      "learning_rate": 6.086157390802064e-06,
      "loss": 0.4166,
      "step": 28875
    },
    {
      "epoch": 1.3251342297278694,
      "grad_norm": 0.4742518961429596,
      "learning_rate": 6.085918056647407e-06,
      "loss": 0.3054,
      "step": 28876
    },
    {
      "epoch": 1.3251801202331237,
      "grad_norm": 0.46937453746795654,
      "learning_rate": 6.0856787198814585e-06,
      "loss": 0.3399,
      "step": 28877
    },
    {
      "epoch": 1.3252260107383782,
      "grad_norm": 0.5208996534347534,
      "learning_rate": 6.085439380504795e-06,
      "loss": 0.3902,
      "step": 28878
    },
    {
      "epoch": 1.3252719012436327,
      "grad_norm": 0.528114914894104,
      "learning_rate": 6.085200038517993e-06,
      "loss": 0.4347,
      "step": 28879
    },
    {
      "epoch": 1.3253177917488872,
      "grad_norm": 0.5162122845649719,
      "learning_rate": 6.0849606939216275e-06,
      "loss": 0.4032,
      "step": 28880
    },
    {
      "epoch": 1.3253636822541417,
      "grad_norm": 0.43470171093940735,
      "learning_rate": 6.0847213467162744e-06,
      "loss": 0.2754,
      "step": 28881
    },
    {
      "epoch": 1.325409572759396,
      "grad_norm": 0.4528656005859375,
      "learning_rate": 6.084481996902508e-06,
      "loss": 0.3063,
      "step": 28882
    },
    {
      "epoch": 1.3254554632646505,
      "grad_norm": 0.4595014154911041,
      "learning_rate": 6.0842426444809066e-06,
      "loss": 0.3483,
      "step": 28883
    },
    {
      "epoch": 1.325501353769905,
      "grad_norm": 0.5022663474082947,
      "learning_rate": 6.08400328945204e-06,
      "loss": 0.3372,
      "step": 28884
    },
    {
      "epoch": 1.3255472442751595,
      "grad_norm": 0.4986295700073242,
      "learning_rate": 6.083763931816492e-06,
      "loss": 0.4221,
      "step": 28885
    },
    {
      "epoch": 1.325593134780414,
      "grad_norm": 0.419863760471344,
      "learning_rate": 6.083524571574832e-06,
      "loss": 0.2692,
      "step": 28886
    },
    {
      "epoch": 1.3256390252856685,
      "grad_norm": 0.5348986983299255,
      "learning_rate": 6.0832852087276364e-06,
      "loss": 0.4852,
      "step": 28887
    },
    {
      "epoch": 1.325684915790923,
      "grad_norm": 0.459759384393692,
      "learning_rate": 6.0830458432754835e-06,
      "loss": 0.328,
      "step": 28888
    },
    {
      "epoch": 1.3257308062961775,
      "grad_norm": 0.4817798137664795,
      "learning_rate": 6.082806475218946e-06,
      "loss": 0.4039,
      "step": 28889
    },
    {
      "epoch": 1.3257766968014317,
      "grad_norm": 0.4742611348628998,
      "learning_rate": 6.082567104558599e-06,
      "loss": 0.3765,
      "step": 28890
    },
    {
      "epoch": 1.3258225873066862,
      "grad_norm": 0.526427686214447,
      "learning_rate": 6.0823277312950225e-06,
      "loss": 0.4169,
      "step": 28891
    },
    {
      "epoch": 1.3258684778119407,
      "grad_norm": 0.4859611392021179,
      "learning_rate": 6.082088355428789e-06,
      "loss": 0.3967,
      "step": 28892
    },
    {
      "epoch": 1.3259143683171952,
      "grad_norm": 0.45125874876976013,
      "learning_rate": 6.081848976960472e-06,
      "loss": 0.3414,
      "step": 28893
    },
    {
      "epoch": 1.3259602588224495,
      "grad_norm": 0.47578346729278564,
      "learning_rate": 6.081609595890653e-06,
      "loss": 0.356,
      "step": 28894
    },
    {
      "epoch": 1.326006149327704,
      "grad_norm": 0.4551136791706085,
      "learning_rate": 6.081370212219902e-06,
      "loss": 0.3339,
      "step": 28895
    },
    {
      "epoch": 1.3260520398329585,
      "grad_norm": 0.42970919609069824,
      "learning_rate": 6.081130825948799e-06,
      "loss": 0.2987,
      "step": 28896
    },
    {
      "epoch": 1.326097930338213,
      "grad_norm": 0.4999580681324005,
      "learning_rate": 6.080891437077917e-06,
      "loss": 0.3882,
      "step": 28897
    },
    {
      "epoch": 1.3261438208434675,
      "grad_norm": 0.46115192770957947,
      "learning_rate": 6.080652045607831e-06,
      "loss": 0.3756,
      "step": 28898
    },
    {
      "epoch": 1.326189711348722,
      "grad_norm": 0.49412643909454346,
      "learning_rate": 6.080412651539118e-06,
      "loss": 0.3947,
      "step": 28899
    },
    {
      "epoch": 1.3262356018539765,
      "grad_norm": 0.4687947928905487,
      "learning_rate": 6.0801732548723546e-06,
      "loss": 0.3754,
      "step": 28900
    },
    {
      "epoch": 1.326281492359231,
      "grad_norm": 0.471955806016922,
      "learning_rate": 6.079933855608114e-06,
      "loss": 0.3467,
      "step": 28901
    },
    {
      "epoch": 1.3263273828644853,
      "grad_norm": 0.45274701714515686,
      "learning_rate": 6.079694453746975e-06,
      "loss": 0.3632,
      "step": 28902
    },
    {
      "epoch": 1.3263732733697398,
      "grad_norm": 0.47636544704437256,
      "learning_rate": 6.079455049289512e-06,
      "loss": 0.3725,
      "step": 28903
    },
    {
      "epoch": 1.3264191638749943,
      "grad_norm": 0.49059930443763733,
      "learning_rate": 6.0792156422362984e-06,
      "loss": 0.3961,
      "step": 28904
    },
    {
      "epoch": 1.3264650543802488,
      "grad_norm": 0.4730585217475891,
      "learning_rate": 6.078976232587913e-06,
      "loss": 0.3731,
      "step": 28905
    },
    {
      "epoch": 1.3265109448855033,
      "grad_norm": 0.4916800558567047,
      "learning_rate": 6.07873682034493e-06,
      "loss": 0.3604,
      "step": 28906
    },
    {
      "epoch": 1.3265568353907575,
      "grad_norm": 0.5163220763206482,
      "learning_rate": 6.078497405507925e-06,
      "loss": 0.453,
      "step": 28907
    },
    {
      "epoch": 1.326602725896012,
      "grad_norm": 0.48936569690704346,
      "learning_rate": 6.078257988077475e-06,
      "loss": 0.372,
      "step": 28908
    },
    {
      "epoch": 1.3266486164012665,
      "grad_norm": 0.5114070177078247,
      "learning_rate": 6.078018568054154e-06,
      "loss": 0.3726,
      "step": 28909
    },
    {
      "epoch": 1.326694506906521,
      "grad_norm": 0.4688384532928467,
      "learning_rate": 6.077779145438538e-06,
      "loss": 0.3663,
      "step": 28910
    },
    {
      "epoch": 1.3267403974117755,
      "grad_norm": 0.488557368516922,
      "learning_rate": 6.077539720231204e-06,
      "loss": 0.3996,
      "step": 28911
    },
    {
      "epoch": 1.32678628791703,
      "grad_norm": 0.4941939413547516,
      "learning_rate": 6.077300292432727e-06,
      "loss": 0.3945,
      "step": 28912
    },
    {
      "epoch": 1.3268321784222845,
      "grad_norm": 0.4762782156467438,
      "learning_rate": 6.0770608620436835e-06,
      "loss": 0.3735,
      "step": 28913
    },
    {
      "epoch": 1.326878068927539,
      "grad_norm": 0.5439191460609436,
      "learning_rate": 6.076821429064649e-06,
      "loss": 0.4726,
      "step": 28914
    },
    {
      "epoch": 1.3269239594327933,
      "grad_norm": 0.47040224075317383,
      "learning_rate": 6.076581993496197e-06,
      "loss": 0.3459,
      "step": 28915
    },
    {
      "epoch": 1.3269698499380478,
      "grad_norm": 0.4570206105709076,
      "learning_rate": 6.0763425553389056e-06,
      "loss": 0.3575,
      "step": 28916
    },
    {
      "epoch": 1.3270157404433023,
      "grad_norm": 0.4774021506309509,
      "learning_rate": 6.07610311459335e-06,
      "loss": 0.3853,
      "step": 28917
    },
    {
      "epoch": 1.3270616309485568,
      "grad_norm": 0.4381569027900696,
      "learning_rate": 6.0758636712601075e-06,
      "loss": 0.3545,
      "step": 28918
    },
    {
      "epoch": 1.3271075214538113,
      "grad_norm": 0.5099719166755676,
      "learning_rate": 6.075624225339751e-06,
      "loss": 0.3976,
      "step": 28919
    },
    {
      "epoch": 1.3271534119590656,
      "grad_norm": 0.45369580388069153,
      "learning_rate": 6.075384776832858e-06,
      "loss": 0.3607,
      "step": 28920
    },
    {
      "epoch": 1.32719930246432,
      "grad_norm": 0.4879549443721771,
      "learning_rate": 6.075145325740004e-06,
      "loss": 0.3616,
      "step": 28921
    },
    {
      "epoch": 1.3272451929695745,
      "grad_norm": 0.4932876527309418,
      "learning_rate": 6.074905872061765e-06,
      "loss": 0.3684,
      "step": 28922
    },
    {
      "epoch": 1.327291083474829,
      "grad_norm": 0.46634000539779663,
      "learning_rate": 6.074666415798716e-06,
      "loss": 0.3865,
      "step": 28923
    },
    {
      "epoch": 1.3273369739800835,
      "grad_norm": 0.4864497184753418,
      "learning_rate": 6.074426956951435e-06,
      "loss": 0.3999,
      "step": 28924
    },
    {
      "epoch": 1.327382864485338,
      "grad_norm": 0.4584881365299225,
      "learning_rate": 6.074187495520496e-06,
      "loss": 0.3126,
      "step": 28925
    },
    {
      "epoch": 1.3274287549905925,
      "grad_norm": 0.6558611392974854,
      "learning_rate": 6.073948031506474e-06,
      "loss": 0.3467,
      "step": 28926
    },
    {
      "epoch": 1.327474645495847,
      "grad_norm": 0.4778306186199188,
      "learning_rate": 6.073708564909945e-06,
      "loss": 0.3927,
      "step": 28927
    },
    {
      "epoch": 1.3275205360011013,
      "grad_norm": 0.43326690793037415,
      "learning_rate": 6.073469095731488e-06,
      "loss": 0.3394,
      "step": 28928
    },
    {
      "epoch": 1.3275664265063558,
      "grad_norm": 0.46001869440078735,
      "learning_rate": 6.073229623971676e-06,
      "loss": 0.3159,
      "step": 28929
    },
    {
      "epoch": 1.3276123170116103,
      "grad_norm": 0.5068168044090271,
      "learning_rate": 6.072990149631084e-06,
      "loss": 0.3515,
      "step": 28930
    },
    {
      "epoch": 1.3276582075168648,
      "grad_norm": 0.4732169806957245,
      "learning_rate": 6.072750672710292e-06,
      "loss": 0.3656,
      "step": 28931
    },
    {
      "epoch": 1.3277040980221193,
      "grad_norm": 0.47115200757980347,
      "learning_rate": 6.07251119320987e-06,
      "loss": 0.3693,
      "step": 28932
    },
    {
      "epoch": 1.3277499885273736,
      "grad_norm": 0.457491010427475,
      "learning_rate": 6.072271711130399e-06,
      "loss": 0.3201,
      "step": 28933
    },
    {
      "epoch": 1.327795879032628,
      "grad_norm": 0.4461002051830292,
      "learning_rate": 6.072032226472453e-06,
      "loss": 0.2987,
      "step": 28934
    },
    {
      "epoch": 1.3278417695378826,
      "grad_norm": 0.5364105105400085,
      "learning_rate": 6.0717927392366086e-06,
      "loss": 0.4464,
      "step": 28935
    },
    {
      "epoch": 1.327887660043137,
      "grad_norm": 0.4862484931945801,
      "learning_rate": 6.07155324942344e-06,
      "loss": 0.399,
      "step": 28936
    },
    {
      "epoch": 1.3279335505483916,
      "grad_norm": 0.5264572501182556,
      "learning_rate": 6.0713137570335235e-06,
      "loss": 0.4773,
      "step": 28937
    },
    {
      "epoch": 1.327979441053646,
      "grad_norm": 0.4510900676250458,
      "learning_rate": 6.071074262067434e-06,
      "loss": 0.3387,
      "step": 28938
    },
    {
      "epoch": 1.3280253315589006,
      "grad_norm": 0.509158730506897,
      "learning_rate": 6.070834764525752e-06,
      "loss": 0.4329,
      "step": 28939
    },
    {
      "epoch": 1.3280712220641548,
      "grad_norm": 0.49007681012153625,
      "learning_rate": 6.07059526440905e-06,
      "loss": 0.3629,
      "step": 28940
    },
    {
      "epoch": 1.3281171125694093,
      "grad_norm": 0.46558547019958496,
      "learning_rate": 6.070355761717903e-06,
      "loss": 0.371,
      "step": 28941
    },
    {
      "epoch": 1.3281630030746638,
      "grad_norm": 0.47143054008483887,
      "learning_rate": 6.070116256452889e-06,
      "loss": 0.41,
      "step": 28942
    },
    {
      "epoch": 1.3282088935799183,
      "grad_norm": 0.5403587818145752,
      "learning_rate": 6.069876748614583e-06,
      "loss": 0.4573,
      "step": 28943
    },
    {
      "epoch": 1.3282547840851728,
      "grad_norm": 0.4359770119190216,
      "learning_rate": 6.06963723820356e-06,
      "loss": 0.2848,
      "step": 28944
    },
    {
      "epoch": 1.328300674590427,
      "grad_norm": 0.4860462546348572,
      "learning_rate": 6.069397725220399e-06,
      "loss": 0.396,
      "step": 28945
    },
    {
      "epoch": 1.3283465650956816,
      "grad_norm": 0.48685625195503235,
      "learning_rate": 6.069158209665673e-06,
      "loss": 0.4343,
      "step": 28946
    },
    {
      "epoch": 1.328392455600936,
      "grad_norm": 0.41911035776138306,
      "learning_rate": 6.068918691539958e-06,
      "loss": 0.2684,
      "step": 28947
    },
    {
      "epoch": 1.3284383461061906,
      "grad_norm": 0.4337615668773651,
      "learning_rate": 6.068679170843832e-06,
      "loss": 0.3004,
      "step": 28948
    },
    {
      "epoch": 1.328484236611445,
      "grad_norm": 0.5006253719329834,
      "learning_rate": 6.068439647577869e-06,
      "loss": 0.4114,
      "step": 28949
    },
    {
      "epoch": 1.3285301271166996,
      "grad_norm": 0.4485139548778534,
      "learning_rate": 6.068200121742647e-06,
      "loss": 0.3132,
      "step": 28950
    },
    {
      "epoch": 1.328576017621954,
      "grad_norm": 0.45289793610572815,
      "learning_rate": 6.067960593338741e-06,
      "loss": 0.3396,
      "step": 28951
    },
    {
      "epoch": 1.3286219081272086,
      "grad_norm": 0.4437621831893921,
      "learning_rate": 6.067721062366726e-06,
      "loss": 0.3356,
      "step": 28952
    },
    {
      "epoch": 1.3286677986324629,
      "grad_norm": 0.4823054075241089,
      "learning_rate": 6.067481528827179e-06,
      "loss": 0.3408,
      "step": 28953
    },
    {
      "epoch": 1.3287136891377174,
      "grad_norm": 0.4142948389053345,
      "learning_rate": 6.067241992720676e-06,
      "loss": 0.2937,
      "step": 28954
    },
    {
      "epoch": 1.3287595796429719,
      "grad_norm": 0.4394124448299408,
      "learning_rate": 6.0670024540477925e-06,
      "loss": 0.3239,
      "step": 28955
    },
    {
      "epoch": 1.3288054701482264,
      "grad_norm": 0.41455572843551636,
      "learning_rate": 6.066762912809106e-06,
      "loss": 0.258,
      "step": 28956
    },
    {
      "epoch": 1.3288513606534809,
      "grad_norm": 0.49027204513549805,
      "learning_rate": 6.06652336900519e-06,
      "loss": 0.3854,
      "step": 28957
    },
    {
      "epoch": 1.3288972511587351,
      "grad_norm": 0.4961969554424286,
      "learning_rate": 6.066283822636621e-06,
      "loss": 0.4236,
      "step": 28958
    },
    {
      "epoch": 1.3289431416639896,
      "grad_norm": 0.4746111035346985,
      "learning_rate": 6.066044273703978e-06,
      "loss": 0.34,
      "step": 28959
    },
    {
      "epoch": 1.3289890321692441,
      "grad_norm": 0.4441932737827301,
      "learning_rate": 6.065804722207834e-06,
      "loss": 0.3281,
      "step": 28960
    },
    {
      "epoch": 1.3290349226744986,
      "grad_norm": 0.45794716477394104,
      "learning_rate": 6.065565168148767e-06,
      "loss": 0.3418,
      "step": 28961
    },
    {
      "epoch": 1.3290808131797531,
      "grad_norm": 0.4718148112297058,
      "learning_rate": 6.065325611527352e-06,
      "loss": 0.3599,
      "step": 28962
    },
    {
      "epoch": 1.3291267036850076,
      "grad_norm": 0.4929414689540863,
      "learning_rate": 6.0650860523441634e-06,
      "loss": 0.4058,
      "step": 28963
    },
    {
      "epoch": 1.3291725941902621,
      "grad_norm": 0.46294182538986206,
      "learning_rate": 6.06484649059978e-06,
      "loss": 0.3041,
      "step": 28964
    },
    {
      "epoch": 1.3292184846955166,
      "grad_norm": 0.458987832069397,
      "learning_rate": 6.064606926294777e-06,
      "loss": 0.3618,
      "step": 28965
    },
    {
      "epoch": 1.329264375200771,
      "grad_norm": 0.44872206449508667,
      "learning_rate": 6.06436735942973e-06,
      "loss": 0.3237,
      "step": 28966
    },
    {
      "epoch": 1.3293102657060254,
      "grad_norm": 0.4783572852611542,
      "learning_rate": 6.064127790005216e-06,
      "loss": 0.3904,
      "step": 28967
    },
    {
      "epoch": 1.32935615621128,
      "grad_norm": 0.43776899576187134,
      "learning_rate": 6.063888218021811e-06,
      "loss": 0.339,
      "step": 28968
    },
    {
      "epoch": 1.3294020467165344,
      "grad_norm": 0.457673043012619,
      "learning_rate": 6.063648643480089e-06,
      "loss": 0.3797,
      "step": 28969
    },
    {
      "epoch": 1.329447937221789,
      "grad_norm": 0.4363172948360443,
      "learning_rate": 6.0634090663806286e-06,
      "loss": 0.2934,
      "step": 28970
    },
    {
      "epoch": 1.3294938277270432,
      "grad_norm": 0.4343571662902832,
      "learning_rate": 6.063169486724004e-06,
      "loss": 0.3189,
      "step": 28971
    },
    {
      "epoch": 1.3295397182322977,
      "grad_norm": 0.4812133014202118,
      "learning_rate": 6.062929904510795e-06,
      "loss": 0.3641,
      "step": 28972
    },
    {
      "epoch": 1.3295856087375522,
      "grad_norm": 0.47825053334236145,
      "learning_rate": 6.062690319741573e-06,
      "loss": 0.3338,
      "step": 28973
    },
    {
      "epoch": 1.3296314992428067,
      "grad_norm": 0.4277571439743042,
      "learning_rate": 6.062450732416916e-06,
      "loss": 0.2822,
      "step": 28974
    },
    {
      "epoch": 1.3296773897480612,
      "grad_norm": 0.5003933310508728,
      "learning_rate": 6.062211142537401e-06,
      "loss": 0.4247,
      "step": 28975
    },
    {
      "epoch": 1.3297232802533157,
      "grad_norm": 0.4528597891330719,
      "learning_rate": 6.061971550103603e-06,
      "loss": 0.331,
      "step": 28976
    },
    {
      "epoch": 1.3297691707585702,
      "grad_norm": 0.474589079618454,
      "learning_rate": 6.061731955116099e-06,
      "loss": 0.375,
      "step": 28977
    },
    {
      "epoch": 1.3298150612638246,
      "grad_norm": 0.48176315426826477,
      "learning_rate": 6.0614923575754645e-06,
      "loss": 0.3957,
      "step": 28978
    },
    {
      "epoch": 1.329860951769079,
      "grad_norm": 0.4220033586025238,
      "learning_rate": 6.061252757482276e-06,
      "loss": 0.294,
      "step": 28979
    },
    {
      "epoch": 1.3299068422743334,
      "grad_norm": 0.6418368220329285,
      "learning_rate": 6.06101315483711e-06,
      "loss": 0.4326,
      "step": 28980
    },
    {
      "epoch": 1.329952732779588,
      "grad_norm": 0.4186230003833771,
      "learning_rate": 6.060773549640541e-06,
      "loss": 0.2793,
      "step": 28981
    },
    {
      "epoch": 1.3299986232848424,
      "grad_norm": 0.46027758717536926,
      "learning_rate": 6.060533941893147e-06,
      "loss": 0.3611,
      "step": 28982
    },
    {
      "epoch": 1.3300445137900967,
      "grad_norm": 0.4906064569950104,
      "learning_rate": 6.060294331595504e-06,
      "loss": 0.3965,
      "step": 28983
    },
    {
      "epoch": 1.3300904042953512,
      "grad_norm": 0.46489042043685913,
      "learning_rate": 6.0600547187481885e-06,
      "loss": 0.2833,
      "step": 28984
    },
    {
      "epoch": 1.3301362948006057,
      "grad_norm": 0.5202928781509399,
      "learning_rate": 6.059815103351775e-06,
      "loss": 0.4042,
      "step": 28985
    },
    {
      "epoch": 1.3301821853058602,
      "grad_norm": 0.4678945243358612,
      "learning_rate": 6.05957548540684e-06,
      "loss": 0.3884,
      "step": 28986
    },
    {
      "epoch": 1.3302280758111147,
      "grad_norm": 0.42390763759613037,
      "learning_rate": 6.0593358649139615e-06,
      "loss": 0.2863,
      "step": 28987
    },
    {
      "epoch": 1.3302739663163692,
      "grad_norm": 0.4540550410747528,
      "learning_rate": 6.059096241873713e-06,
      "loss": 0.3131,
      "step": 28988
    },
    {
      "epoch": 1.3303198568216237,
      "grad_norm": 0.47083231806755066,
      "learning_rate": 6.058856616286675e-06,
      "loss": 0.3589,
      "step": 28989
    },
    {
      "epoch": 1.3303657473268782,
      "grad_norm": 0.4826127886772156,
      "learning_rate": 6.05861698815342e-06,
      "loss": 0.3811,
      "step": 28990
    },
    {
      "epoch": 1.3304116378321325,
      "grad_norm": 0.5192810297012329,
      "learning_rate": 6.058377357474523e-06,
      "loss": 0.4256,
      "step": 28991
    },
    {
      "epoch": 1.330457528337387,
      "grad_norm": 0.48335832357406616,
      "learning_rate": 6.058137724250563e-06,
      "loss": 0.3457,
      "step": 28992
    },
    {
      "epoch": 1.3305034188426414,
      "grad_norm": 0.481261283159256,
      "learning_rate": 6.057898088482118e-06,
      "loss": 0.3784,
      "step": 28993
    },
    {
      "epoch": 1.330549309347896,
      "grad_norm": 0.4477469027042389,
      "learning_rate": 6.057658450169761e-06,
      "loss": 0.3632,
      "step": 28994
    },
    {
      "epoch": 1.3305951998531504,
      "grad_norm": 0.4719911515712738,
      "learning_rate": 6.057418809314069e-06,
      "loss": 0.4089,
      "step": 28995
    },
    {
      "epoch": 1.3306410903584047,
      "grad_norm": 0.4626944661140442,
      "learning_rate": 6.057179165915619e-06,
      "loss": 0.3487,
      "step": 28996
    },
    {
      "epoch": 1.3306869808636592,
      "grad_norm": 0.450916588306427,
      "learning_rate": 6.056939519974986e-06,
      "loss": 0.3277,
      "step": 28997
    },
    {
      "epoch": 1.3307328713689137,
      "grad_norm": 0.42776787281036377,
      "learning_rate": 6.056699871492747e-06,
      "loss": 0.2854,
      "step": 28998
    },
    {
      "epoch": 1.3307787618741682,
      "grad_norm": 0.44800302386283875,
      "learning_rate": 6.056460220469479e-06,
      "loss": 0.3298,
      "step": 28999
    },
    {
      "epoch": 1.3308246523794227,
      "grad_norm": 0.46218302845954895,
      "learning_rate": 6.056220566905759e-06,
      "loss": 0.3484,
      "step": 29000
    },
    {
      "epoch": 1.3308705428846772,
      "grad_norm": 0.4502412974834442,
      "learning_rate": 6.055980910802159e-06,
      "loss": 0.3195,
      "step": 29001
    },
    {
      "epoch": 1.3309164333899317,
      "grad_norm": 0.4486466348171234,
      "learning_rate": 6.05574125215926e-06,
      "loss": 0.3761,
      "step": 29002
    },
    {
      "epoch": 1.3309623238951862,
      "grad_norm": 0.4678104519844055,
      "learning_rate": 6.055501590977635e-06,
      "loss": 0.3584,
      "step": 29003
    },
    {
      "epoch": 1.3310082144004405,
      "grad_norm": 0.4770292341709137,
      "learning_rate": 6.055261927257863e-06,
      "loss": 0.3524,
      "step": 29004
    },
    {
      "epoch": 1.331054104905695,
      "grad_norm": 0.48011305928230286,
      "learning_rate": 6.055022261000519e-06,
      "loss": 0.3169,
      "step": 29005
    },
    {
      "epoch": 1.3310999954109495,
      "grad_norm": 0.4528268873691559,
      "learning_rate": 6.054782592206178e-06,
      "loss": 0.322,
      "step": 29006
    },
    {
      "epoch": 1.331145885916204,
      "grad_norm": 0.4829457104206085,
      "learning_rate": 6.0545429208754205e-06,
      "loss": 0.3836,
      "step": 29007
    },
    {
      "epoch": 1.3311917764214585,
      "grad_norm": 0.47240519523620605,
      "learning_rate": 6.054303247008818e-06,
      "loss": 0.3461,
      "step": 29008
    },
    {
      "epoch": 1.3312376669267127,
      "grad_norm": 0.5094849467277527,
      "learning_rate": 6.05406357060695e-06,
      "loss": 0.429,
      "step": 29009
    },
    {
      "epoch": 1.3312835574319672,
      "grad_norm": 0.48100197315216064,
      "learning_rate": 6.053823891670391e-06,
      "loss": 0.3358,
      "step": 29010
    },
    {
      "epoch": 1.3313294479372217,
      "grad_norm": 0.5158067941665649,
      "learning_rate": 6.0535842101997186e-06,
      "loss": 0.4049,
      "step": 29011
    },
    {
      "epoch": 1.3313753384424762,
      "grad_norm": 0.4862521290779114,
      "learning_rate": 6.053344526195508e-06,
      "loss": 0.4207,
      "step": 29012
    },
    {
      "epoch": 1.3314212289477307,
      "grad_norm": 0.5173901915550232,
      "learning_rate": 6.053104839658337e-06,
      "loss": 0.3751,
      "step": 29013
    },
    {
      "epoch": 1.3314671194529852,
      "grad_norm": 0.45692718029022217,
      "learning_rate": 6.05286515058878e-06,
      "loss": 0.3559,
      "step": 29014
    },
    {
      "epoch": 1.3315130099582397,
      "grad_norm": 0.4491845369338989,
      "learning_rate": 6.052625458987415e-06,
      "loss": 0.322,
      "step": 29015
    },
    {
      "epoch": 1.3315589004634942,
      "grad_norm": 0.4488137364387512,
      "learning_rate": 6.052385764854821e-06,
      "loss": 0.3175,
      "step": 29016
    },
    {
      "epoch": 1.3316047909687485,
      "grad_norm": 0.5227344036102295,
      "learning_rate": 6.052146068191569e-06,
      "loss": 0.4253,
      "step": 29017
    },
    {
      "epoch": 1.331650681474003,
      "grad_norm": 0.5366879105567932,
      "learning_rate": 6.051906368998238e-06,
      "loss": 0.4597,
      "step": 29018
    },
    {
      "epoch": 1.3316965719792575,
      "grad_norm": 0.4673616290092468,
      "learning_rate": 6.051666667275403e-06,
      "loss": 0.3125,
      "step": 29019
    },
    {
      "epoch": 1.331742462484512,
      "grad_norm": 0.5072886943817139,
      "learning_rate": 6.051426963023642e-06,
      "loss": 0.4252,
      "step": 29020
    },
    {
      "epoch": 1.3317883529897665,
      "grad_norm": 0.5162843465805054,
      "learning_rate": 6.051187256243532e-06,
      "loss": 0.4086,
      "step": 29021
    },
    {
      "epoch": 1.3318342434950208,
      "grad_norm": 0.4688956141471863,
      "learning_rate": 6.050947546935648e-06,
      "loss": 0.3503,
      "step": 29022
    },
    {
      "epoch": 1.3318801340002753,
      "grad_norm": 0.500285804271698,
      "learning_rate": 6.050707835100567e-06,
      "loss": 0.4001,
      "step": 29023
    },
    {
      "epoch": 1.3319260245055298,
      "grad_norm": 0.4660649299621582,
      "learning_rate": 6.050468120738866e-06,
      "loss": 0.3053,
      "step": 29024
    },
    {
      "epoch": 1.3319719150107843,
      "grad_norm": 0.4897279143333435,
      "learning_rate": 6.050228403851118e-06,
      "loss": 0.3962,
      "step": 29025
    },
    {
      "epoch": 1.3320178055160388,
      "grad_norm": 0.4715103209018707,
      "learning_rate": 6.049988684437904e-06,
      "loss": 0.3629,
      "step": 29026
    },
    {
      "epoch": 1.3320636960212933,
      "grad_norm": 0.45087769627571106,
      "learning_rate": 6.0497489624998e-06,
      "loss": 0.3286,
      "step": 29027
    },
    {
      "epoch": 1.3321095865265478,
      "grad_norm": 0.45602095127105713,
      "learning_rate": 6.049509238037379e-06,
      "loss": 0.3176,
      "step": 29028
    },
    {
      "epoch": 1.332155477031802,
      "grad_norm": 0.5036959052085876,
      "learning_rate": 6.04926951105122e-06,
      "loss": 0.4548,
      "step": 29029
    },
    {
      "epoch": 1.3322013675370565,
      "grad_norm": 0.44432955980300903,
      "learning_rate": 6.0490297815419e-06,
      "loss": 0.3067,
      "step": 29030
    },
    {
      "epoch": 1.332247258042311,
      "grad_norm": 0.48377448320388794,
      "learning_rate": 6.048790049509992e-06,
      "loss": 0.4027,
      "step": 29031
    },
    {
      "epoch": 1.3322931485475655,
      "grad_norm": 0.4680689573287964,
      "learning_rate": 6.048550314956077e-06,
      "loss": 0.3543,
      "step": 29032
    },
    {
      "epoch": 1.33233903905282,
      "grad_norm": 0.47265562415122986,
      "learning_rate": 6.0483105778807295e-06,
      "loss": 0.3232,
      "step": 29033
    },
    {
      "epoch": 1.3323849295580743,
      "grad_norm": 0.4647786617279053,
      "learning_rate": 6.048070838284524e-06,
      "loss": 0.3292,
      "step": 29034
    },
    {
      "epoch": 1.3324308200633288,
      "grad_norm": 0.4825431704521179,
      "learning_rate": 6.047831096168042e-06,
      "loss": 0.4088,
      "step": 29035
    },
    {
      "epoch": 1.3324767105685833,
      "grad_norm": 0.4414055645465851,
      "learning_rate": 6.047591351531854e-06,
      "loss": 0.3186,
      "step": 29036
    },
    {
      "epoch": 1.3325226010738378,
      "grad_norm": 0.4920535385608673,
      "learning_rate": 6.04735160437654e-06,
      "loss": 0.3769,
      "step": 29037
    },
    {
      "epoch": 1.3325684915790923,
      "grad_norm": 0.44899627566337585,
      "learning_rate": 6.047111854702678e-06,
      "loss": 0.3417,
      "step": 29038
    },
    {
      "epoch": 1.3326143820843468,
      "grad_norm": 0.44957560300827026,
      "learning_rate": 6.04687210251084e-06,
      "loss": 0.3325,
      "step": 29039
    },
    {
      "epoch": 1.3326602725896013,
      "grad_norm": 0.48307403922080994,
      "learning_rate": 6.046632347801606e-06,
      "loss": 0.3824,
      "step": 29040
    },
    {
      "epoch": 1.3327061630948558,
      "grad_norm": 0.471717894077301,
      "learning_rate": 6.046392590575552e-06,
      "loss": 0.3361,
      "step": 29041
    },
    {
      "epoch": 1.33275205360011,
      "grad_norm": 0.4740654528141022,
      "learning_rate": 6.046152830833253e-06,
      "loss": 0.393,
      "step": 29042
    },
    {
      "epoch": 1.3327979441053646,
      "grad_norm": 0.5043904781341553,
      "learning_rate": 6.045913068575287e-06,
      "loss": 0.4055,
      "step": 29043
    },
    {
      "epoch": 1.332843834610619,
      "grad_norm": 0.4397449493408203,
      "learning_rate": 6.045673303802232e-06,
      "loss": 0.309,
      "step": 29044
    },
    {
      "epoch": 1.3328897251158736,
      "grad_norm": 0.489668071269989,
      "learning_rate": 6.04543353651466e-06,
      "loss": 0.3688,
      "step": 29045
    },
    {
      "epoch": 1.332935615621128,
      "grad_norm": 0.4514453113079071,
      "learning_rate": 6.045193766713151e-06,
      "loss": 0.3236,
      "step": 29046
    },
    {
      "epoch": 1.3329815061263823,
      "grad_norm": 0.47197476029396057,
      "learning_rate": 6.044953994398281e-06,
      "loss": 0.3832,
      "step": 29047
    },
    {
      "epoch": 1.3330273966316368,
      "grad_norm": 0.4888337254524231,
      "learning_rate": 6.0447142195706265e-06,
      "loss": 0.4154,
      "step": 29048
    },
    {
      "epoch": 1.3330732871368913,
      "grad_norm": 0.4410566985607147,
      "learning_rate": 6.044474442230765e-06,
      "loss": 0.3098,
      "step": 29049
    },
    {
      "epoch": 1.3331191776421458,
      "grad_norm": 0.46776658296585083,
      "learning_rate": 6.044234662379271e-06,
      "loss": 0.382,
      "step": 29050
    },
    {
      "epoch": 1.3331650681474003,
      "grad_norm": 0.47066453099250793,
      "learning_rate": 6.043994880016722e-06,
      "loss": 0.3828,
      "step": 29051
    },
    {
      "epoch": 1.3332109586526548,
      "grad_norm": 0.5110417008399963,
      "learning_rate": 6.0437550951436965e-06,
      "loss": 0.3837,
      "step": 29052
    },
    {
      "epoch": 1.3332568491579093,
      "grad_norm": 0.4114798605442047,
      "learning_rate": 6.043515307760768e-06,
      "loss": 0.2627,
      "step": 29053
    },
    {
      "epoch": 1.3333027396631638,
      "grad_norm": 0.46696728467941284,
      "learning_rate": 6.043275517868516e-06,
      "loss": 0.3637,
      "step": 29054
    },
    {
      "epoch": 1.333348630168418,
      "grad_norm": 0.4680844247341156,
      "learning_rate": 6.043035725467514e-06,
      "loss": 0.3729,
      "step": 29055
    },
    {
      "epoch": 1.3333945206736726,
      "grad_norm": 0.4294092655181885,
      "learning_rate": 6.042795930558341e-06,
      "loss": 0.2926,
      "step": 29056
    },
    {
      "epoch": 1.333440411178927,
      "grad_norm": 0.5240764021873474,
      "learning_rate": 6.042556133141572e-06,
      "loss": 0.4671,
      "step": 29057
    },
    {
      "epoch": 1.3334863016841816,
      "grad_norm": 0.44590306282043457,
      "learning_rate": 6.0423163332177845e-06,
      "loss": 0.3209,
      "step": 29058
    },
    {
      "epoch": 1.333532192189436,
      "grad_norm": 0.46013468503952026,
      "learning_rate": 6.0420765307875555e-06,
      "loss": 0.3272,
      "step": 29059
    },
    {
      "epoch": 1.3335780826946904,
      "grad_norm": 0.4538838267326355,
      "learning_rate": 6.041836725851462e-06,
      "loss": 0.3525,
      "step": 29060
    },
    {
      "epoch": 1.3336239731999449,
      "grad_norm": 0.46814173460006714,
      "learning_rate": 6.041596918410081e-06,
      "loss": 0.3442,
      "step": 29061
    },
    {
      "epoch": 1.3336698637051994,
      "grad_norm": 0.4819115400314331,
      "learning_rate": 6.041357108463985e-06,
      "loss": 0.3698,
      "step": 29062
    },
    {
      "epoch": 1.3337157542104539,
      "grad_norm": 0.48723849654197693,
      "learning_rate": 6.0411172960137555e-06,
      "loss": 0.361,
      "step": 29063
    },
    {
      "epoch": 1.3337616447157083,
      "grad_norm": 0.48952552676200867,
      "learning_rate": 6.040877481059968e-06,
      "loss": 0.3999,
      "step": 29064
    },
    {
      "epoch": 1.3338075352209628,
      "grad_norm": 0.49114885926246643,
      "learning_rate": 6.0406376636031994e-06,
      "loss": 0.3813,
      "step": 29065
    },
    {
      "epoch": 1.3338534257262173,
      "grad_norm": 0.4573996961116791,
      "learning_rate": 6.040397843644025e-06,
      "loss": 0.3697,
      "step": 29066
    },
    {
      "epoch": 1.3338993162314718,
      "grad_norm": 0.4565688371658325,
      "learning_rate": 6.040158021183021e-06,
      "loss": 0.354,
      "step": 29067
    },
    {
      "epoch": 1.3339452067367261,
      "grad_norm": 0.47614648938179016,
      "learning_rate": 6.0399181962207655e-06,
      "loss": 0.3609,
      "step": 29068
    },
    {
      "epoch": 1.3339910972419806,
      "grad_norm": 0.491591215133667,
      "learning_rate": 6.0396783687578356e-06,
      "loss": 0.3959,
      "step": 29069
    },
    {
      "epoch": 1.3340369877472351,
      "grad_norm": 0.4400084912776947,
      "learning_rate": 6.0394385387948066e-06,
      "loss": 0.3126,
      "step": 29070
    },
    {
      "epoch": 1.3340828782524896,
      "grad_norm": 0.4673841893672943,
      "learning_rate": 6.039198706332256e-06,
      "loss": 0.3864,
      "step": 29071
    },
    {
      "epoch": 1.3341287687577439,
      "grad_norm": 0.46085745096206665,
      "learning_rate": 6.038958871370763e-06,
      "loss": 0.3372,
      "step": 29072
    },
    {
      "epoch": 1.3341746592629984,
      "grad_norm": 0.48850929737091064,
      "learning_rate": 6.038719033910898e-06,
      "loss": 0.4339,
      "step": 29073
    },
    {
      "epoch": 1.3342205497682529,
      "grad_norm": 0.4771660268306732,
      "learning_rate": 6.038479193953244e-06,
      "loss": 0.4049,
      "step": 29074
    },
    {
      "epoch": 1.3342664402735074,
      "grad_norm": 0.47540417313575745,
      "learning_rate": 6.038239351498374e-06,
      "loss": 0.3767,
      "step": 29075
    },
    {
      "epoch": 1.3343123307787619,
      "grad_norm": 0.5048332810401917,
      "learning_rate": 6.0379995065468675e-06,
      "loss": 0.428,
      "step": 29076
    },
    {
      "epoch": 1.3343582212840164,
      "grad_norm": 0.48667386174201965,
      "learning_rate": 6.037759659099298e-06,
      "loss": 0.3817,
      "step": 29077
    },
    {
      "epoch": 1.3344041117892709,
      "grad_norm": 0.41540807485580444,
      "learning_rate": 6.037519809156246e-06,
      "loss": 0.2676,
      "step": 29078
    },
    {
      "epoch": 1.3344500022945254,
      "grad_norm": 0.5446277260780334,
      "learning_rate": 6.037279956718285e-06,
      "loss": 0.4665,
      "step": 29079
    },
    {
      "epoch": 1.3344958927997796,
      "grad_norm": 0.4949190616607666,
      "learning_rate": 6.037040101785993e-06,
      "loss": 0.4343,
      "step": 29080
    },
    {
      "epoch": 1.3345417833050341,
      "grad_norm": 0.4788063168525696,
      "learning_rate": 6.0368002443599475e-06,
      "loss": 0.3714,
      "step": 29081
    },
    {
      "epoch": 1.3345876738102886,
      "grad_norm": 0.5107227563858032,
      "learning_rate": 6.036560384440725e-06,
      "loss": 0.3988,
      "step": 29082
    },
    {
      "epoch": 1.3346335643155431,
      "grad_norm": 0.4969682991504669,
      "learning_rate": 6.0363205220289014e-06,
      "loss": 0.4347,
      "step": 29083
    },
    {
      "epoch": 1.3346794548207976,
      "grad_norm": 0.4305925965309143,
      "learning_rate": 6.036080657125054e-06,
      "loss": 0.2783,
      "step": 29084
    },
    {
      "epoch": 1.334725345326052,
      "grad_norm": 0.4621310234069824,
      "learning_rate": 6.035840789729759e-06,
      "loss": 0.39,
      "step": 29085
    },
    {
      "epoch": 1.3347712358313064,
      "grad_norm": 0.49027860164642334,
      "learning_rate": 6.035600919843594e-06,
      "loss": 0.4084,
      "step": 29086
    },
    {
      "epoch": 1.334817126336561,
      "grad_norm": 0.4194851517677307,
      "learning_rate": 6.035361047467137e-06,
      "loss": 0.2912,
      "step": 29087
    },
    {
      "epoch": 1.3348630168418154,
      "grad_norm": 0.4698309004306793,
      "learning_rate": 6.035121172600962e-06,
      "loss": 0.3502,
      "step": 29088
    },
    {
      "epoch": 1.33490890734707,
      "grad_norm": 0.5209037661552429,
      "learning_rate": 6.034881295245648e-06,
      "loss": 0.4788,
      "step": 29089
    },
    {
      "epoch": 1.3349547978523244,
      "grad_norm": 0.4469691514968872,
      "learning_rate": 6.03464141540177e-06,
      "loss": 0.3633,
      "step": 29090
    },
    {
      "epoch": 1.335000688357579,
      "grad_norm": 0.4826221466064453,
      "learning_rate": 6.034401533069907e-06,
      "loss": 0.3918,
      "step": 29091
    },
    {
      "epoch": 1.3350465788628334,
      "grad_norm": 0.48209673166275024,
      "learning_rate": 6.034161648250635e-06,
      "loss": 0.3297,
      "step": 29092
    },
    {
      "epoch": 1.3350924693680877,
      "grad_norm": 0.46209263801574707,
      "learning_rate": 6.03392176094453e-06,
      "loss": 0.3968,
      "step": 29093
    },
    {
      "epoch": 1.3351383598733422,
      "grad_norm": 0.4549507200717926,
      "learning_rate": 6.03368187115217e-06,
      "loss": 0.3401,
      "step": 29094
    },
    {
      "epoch": 1.3351842503785967,
      "grad_norm": 0.46352294087409973,
      "learning_rate": 6.0334419788741305e-06,
      "loss": 0.3601,
      "step": 29095
    },
    {
      "epoch": 1.3352301408838512,
      "grad_norm": 0.4696026146411896,
      "learning_rate": 6.03320208411099e-06,
      "loss": 0.3645,
      "step": 29096
    },
    {
      "epoch": 1.3352760313891057,
      "grad_norm": 0.43401142954826355,
      "learning_rate": 6.032962186863325e-06,
      "loss": 0.2856,
      "step": 29097
    },
    {
      "epoch": 1.33532192189436,
      "grad_norm": 0.4843392074108124,
      "learning_rate": 6.032722287131712e-06,
      "loss": 0.4358,
      "step": 29098
    },
    {
      "epoch": 1.3353678123996144,
      "grad_norm": 0.5364860892295837,
      "learning_rate": 6.032482384916726e-06,
      "loss": 0.4417,
      "step": 29099
    },
    {
      "epoch": 1.335413702904869,
      "grad_norm": 0.4722587466239929,
      "learning_rate": 6.032242480218947e-06,
      "loss": 0.3723,
      "step": 29100
    },
    {
      "epoch": 1.3354595934101234,
      "grad_norm": 0.4423934519290924,
      "learning_rate": 6.03200257303895e-06,
      "loss": 0.2997,
      "step": 29101
    },
    {
      "epoch": 1.335505483915378,
      "grad_norm": 0.4923318028450012,
      "learning_rate": 6.031762663377312e-06,
      "loss": 0.4,
      "step": 29102
    },
    {
      "epoch": 1.3355513744206324,
      "grad_norm": 0.7345426082611084,
      "learning_rate": 6.031522751234612e-06,
      "loss": 0.4046,
      "step": 29103
    },
    {
      "epoch": 1.335597264925887,
      "grad_norm": 0.4803784191608429,
      "learning_rate": 6.031282836611425e-06,
      "loss": 0.3589,
      "step": 29104
    },
    {
      "epoch": 1.3356431554311414,
      "grad_norm": 0.4763603210449219,
      "learning_rate": 6.031042919508327e-06,
      "loss": 0.3919,
      "step": 29105
    },
    {
      "epoch": 1.3356890459363957,
      "grad_norm": 0.43661871552467346,
      "learning_rate": 6.030802999925898e-06,
      "loss": 0.333,
      "step": 29106
    },
    {
      "epoch": 1.3357349364416502,
      "grad_norm": 0.48853445053100586,
      "learning_rate": 6.0305630778647115e-06,
      "loss": 0.3877,
      "step": 29107
    },
    {
      "epoch": 1.3357808269469047,
      "grad_norm": 0.46471449732780457,
      "learning_rate": 6.030323153325348e-06,
      "loss": 0.3201,
      "step": 29108
    },
    {
      "epoch": 1.3358267174521592,
      "grad_norm": 0.47538265585899353,
      "learning_rate": 6.030083226308381e-06,
      "loss": 0.3635,
      "step": 29109
    },
    {
      "epoch": 1.3358726079574137,
      "grad_norm": 0.5639995336532593,
      "learning_rate": 6.02984329681439e-06,
      "loss": 0.4227,
      "step": 29110
    },
    {
      "epoch": 1.335918498462668,
      "grad_norm": 0.49205997586250305,
      "learning_rate": 6.029603364843949e-06,
      "loss": 0.3726,
      "step": 29111
    },
    {
      "epoch": 1.3359643889679225,
      "grad_norm": 0.4508155882358551,
      "learning_rate": 6.029363430397638e-06,
      "loss": 0.3507,
      "step": 29112
    },
    {
      "epoch": 1.336010279473177,
      "grad_norm": 0.4871511161327362,
      "learning_rate": 6.0291234934760325e-06,
      "loss": 0.338,
      "step": 29113
    },
    {
      "epoch": 1.3360561699784315,
      "grad_norm": 0.49588778614997864,
      "learning_rate": 6.028883554079711e-06,
      "loss": 0.3799,
      "step": 29114
    },
    {
      "epoch": 1.336102060483686,
      "grad_norm": 0.4879857301712036,
      "learning_rate": 6.028643612209248e-06,
      "loss": 0.3989,
      "step": 29115
    },
    {
      "epoch": 1.3361479509889405,
      "grad_norm": 0.46847161650657654,
      "learning_rate": 6.028403667865223e-06,
      "loss": 0.3592,
      "step": 29116
    },
    {
      "epoch": 1.336193841494195,
      "grad_norm": 0.4904146194458008,
      "learning_rate": 6.02816372104821e-06,
      "loss": 0.3626,
      "step": 29117
    },
    {
      "epoch": 1.3362397319994492,
      "grad_norm": 0.46270591020584106,
      "learning_rate": 6.027923771758788e-06,
      "loss": 0.3229,
      "step": 29118
    },
    {
      "epoch": 1.3362856225047037,
      "grad_norm": 0.486944317817688,
      "learning_rate": 6.027683819997535e-06,
      "loss": 0.3783,
      "step": 29119
    },
    {
      "epoch": 1.3363315130099582,
      "grad_norm": 0.6919149160385132,
      "learning_rate": 6.027443865765027e-06,
      "loss": 0.3677,
      "step": 29120
    },
    {
      "epoch": 1.3363774035152127,
      "grad_norm": 0.48998329043388367,
      "learning_rate": 6.0272039090618395e-06,
      "loss": 0.3782,
      "step": 29121
    },
    {
      "epoch": 1.3364232940204672,
      "grad_norm": 0.48060932755470276,
      "learning_rate": 6.026963949888552e-06,
      "loss": 0.3729,
      "step": 29122
    },
    {
      "epoch": 1.3364691845257215,
      "grad_norm": 0.5158085227012634,
      "learning_rate": 6.026723988245738e-06,
      "loss": 0.4264,
      "step": 29123
    },
    {
      "epoch": 1.336515075030976,
      "grad_norm": 0.46642041206359863,
      "learning_rate": 6.026484024133978e-06,
      "loss": 0.3654,
      "step": 29124
    },
    {
      "epoch": 1.3365609655362305,
      "grad_norm": 0.45447760820388794,
      "learning_rate": 6.0262440575538495e-06,
      "loss": 0.3594,
      "step": 29125
    },
    {
      "epoch": 1.336606856041485,
      "grad_norm": 0.4749845266342163,
      "learning_rate": 6.026004088505927e-06,
      "loss": 0.4262,
      "step": 29126
    },
    {
      "epoch": 1.3366527465467395,
      "grad_norm": 0.45667627453804016,
      "learning_rate": 6.025764116990788e-06,
      "loss": 0.3211,
      "step": 29127
    },
    {
      "epoch": 1.336698637051994,
      "grad_norm": 0.503494381904602,
      "learning_rate": 6.025524143009009e-06,
      "loss": 0.4426,
      "step": 29128
    },
    {
      "epoch": 1.3367445275572485,
      "grad_norm": 0.4467994272708893,
      "learning_rate": 6.0252841665611704e-06,
      "loss": 0.3122,
      "step": 29129
    },
    {
      "epoch": 1.336790418062503,
      "grad_norm": 0.5051474571228027,
      "learning_rate": 6.025044187647846e-06,
      "loss": 0.3807,
      "step": 29130
    },
    {
      "epoch": 1.3368363085677573,
      "grad_norm": 0.46390625834465027,
      "learning_rate": 6.0248042062696145e-06,
      "loss": 0.3113,
      "step": 29131
    },
    {
      "epoch": 1.3368821990730118,
      "grad_norm": 0.5040279626846313,
      "learning_rate": 6.024564222427052e-06,
      "loss": 0.4103,
      "step": 29132
    },
    {
      "epoch": 1.3369280895782663,
      "grad_norm": 0.42201948165893555,
      "learning_rate": 6.024324236120735e-06,
      "loss": 0.2983,
      "step": 29133
    },
    {
      "epoch": 1.3369739800835208,
      "grad_norm": 0.4583841562271118,
      "learning_rate": 6.0240842473512415e-06,
      "loss": 0.3565,
      "step": 29134
    },
    {
      "epoch": 1.3370198705887752,
      "grad_norm": 0.4819902181625366,
      "learning_rate": 6.023844256119149e-06,
      "loss": 0.3512,
      "step": 29135
    },
    {
      "epoch": 1.3370657610940295,
      "grad_norm": 0.47656092047691345,
      "learning_rate": 6.023604262425035e-06,
      "loss": 0.3525,
      "step": 29136
    },
    {
      "epoch": 1.337111651599284,
      "grad_norm": 0.5208480358123779,
      "learning_rate": 6.0233642662694755e-06,
      "loss": 0.4017,
      "step": 29137
    },
    {
      "epoch": 1.3371575421045385,
      "grad_norm": 0.4388142228126526,
      "learning_rate": 6.023124267653046e-06,
      "loss": 0.2908,
      "step": 29138
    },
    {
      "epoch": 1.337203432609793,
      "grad_norm": 0.43300411105155945,
      "learning_rate": 6.022884266576328e-06,
      "loss": 0.2763,
      "step": 29139
    },
    {
      "epoch": 1.3372493231150475,
      "grad_norm": 0.4286147654056549,
      "learning_rate": 6.022644263039895e-06,
      "loss": 0.3175,
      "step": 29140
    },
    {
      "epoch": 1.337295213620302,
      "grad_norm": 0.47516924142837524,
      "learning_rate": 6.022404257044327e-06,
      "loss": 0.3727,
      "step": 29141
    },
    {
      "epoch": 1.3373411041255565,
      "grad_norm": 0.4633125066757202,
      "learning_rate": 6.022164248590197e-06,
      "loss": 0.3797,
      "step": 29142
    },
    {
      "epoch": 1.337386994630811,
      "grad_norm": 0.4670012295246124,
      "learning_rate": 6.021924237678085e-06,
      "loss": 0.3713,
      "step": 29143
    },
    {
      "epoch": 1.3374328851360653,
      "grad_norm": 0.4662891924381256,
      "learning_rate": 6.021684224308568e-06,
      "loss": 0.3578,
      "step": 29144
    },
    {
      "epoch": 1.3374787756413198,
      "grad_norm": 0.43505269289016724,
      "learning_rate": 6.021444208482223e-06,
      "loss": 0.3147,
      "step": 29145
    },
    {
      "epoch": 1.3375246661465743,
      "grad_norm": 0.4830741584300995,
      "learning_rate": 6.021204190199627e-06,
      "loss": 0.3963,
      "step": 29146
    },
    {
      "epoch": 1.3375705566518288,
      "grad_norm": 0.4690886437892914,
      "learning_rate": 6.020964169461359e-06,
      "loss": 0.3139,
      "step": 29147
    },
    {
      "epoch": 1.3376164471570833,
      "grad_norm": 0.4609993100166321,
      "learning_rate": 6.0207241462679935e-06,
      "loss": 0.3112,
      "step": 29148
    },
    {
      "epoch": 1.3376623376623376,
      "grad_norm": 0.49577590823173523,
      "learning_rate": 6.020484120620107e-06,
      "loss": 0.3925,
      "step": 29149
    },
    {
      "epoch": 1.337708228167592,
      "grad_norm": 0.529504120349884,
      "learning_rate": 6.020244092518278e-06,
      "loss": 0.3966,
      "step": 29150
    },
    {
      "epoch": 1.3377541186728465,
      "grad_norm": 0.4904326796531677,
      "learning_rate": 6.020004061963086e-06,
      "loss": 0.384,
      "step": 29151
    },
    {
      "epoch": 1.337800009178101,
      "grad_norm": 0.48486241698265076,
      "learning_rate": 6.019764028955107e-06,
      "loss": 0.3826,
      "step": 29152
    },
    {
      "epoch": 1.3378458996833555,
      "grad_norm": 0.46342960000038147,
      "learning_rate": 6.019523993494914e-06,
      "loss": 0.313,
      "step": 29153
    },
    {
      "epoch": 1.33789179018861,
      "grad_norm": 0.44884222745895386,
      "learning_rate": 6.0192839555830895e-06,
      "loss": 0.3525,
      "step": 29154
    },
    {
      "epoch": 1.3379376806938645,
      "grad_norm": 0.43570676445961,
      "learning_rate": 6.019043915220209e-06,
      "loss": 0.3099,
      "step": 29155
    },
    {
      "epoch": 1.337983571199119,
      "grad_norm": 0.4415733218193054,
      "learning_rate": 6.018803872406847e-06,
      "loss": 0.3582,
      "step": 29156
    },
    {
      "epoch": 1.3380294617043733,
      "grad_norm": 0.46555572748184204,
      "learning_rate": 6.018563827143586e-06,
      "loss": 0.3564,
      "step": 29157
    },
    {
      "epoch": 1.3380753522096278,
      "grad_norm": 0.4606015086174011,
      "learning_rate": 6.018323779431001e-06,
      "loss": 0.3365,
      "step": 29158
    },
    {
      "epoch": 1.3381212427148823,
      "grad_norm": 0.49656912684440613,
      "learning_rate": 6.018083729269667e-06,
      "loss": 0.4424,
      "step": 29159
    },
    {
      "epoch": 1.3381671332201368,
      "grad_norm": 0.5200998187065125,
      "learning_rate": 6.017843676660163e-06,
      "loss": 0.452,
      "step": 29160
    },
    {
      "epoch": 1.338213023725391,
      "grad_norm": 0.4752213656902313,
      "learning_rate": 6.017603621603066e-06,
      "loss": 0.4025,
      "step": 29161
    },
    {
      "epoch": 1.3382589142306456,
      "grad_norm": 0.5262995958328247,
      "learning_rate": 6.017363564098954e-06,
      "loss": 0.5039,
      "step": 29162
    },
    {
      "epoch": 1.3383048047359,
      "grad_norm": 0.4242538511753082,
      "learning_rate": 6.017123504148405e-06,
      "loss": 0.281,
      "step": 29163
    },
    {
      "epoch": 1.3383506952411546,
      "grad_norm": 0.4679216146469116,
      "learning_rate": 6.016883441751994e-06,
      "loss": 0.3451,
      "step": 29164
    },
    {
      "epoch": 1.338396585746409,
      "grad_norm": 0.4858410358428955,
      "learning_rate": 6.0166433769102995e-06,
      "loss": 0.4013,
      "step": 29165
    },
    {
      "epoch": 1.3384424762516636,
      "grad_norm": 0.4601886570453644,
      "learning_rate": 6.016403309623897e-06,
      "loss": 0.3516,
      "step": 29166
    },
    {
      "epoch": 1.338488366756918,
      "grad_norm": 0.5062465071678162,
      "learning_rate": 6.016163239893366e-06,
      "loss": 0.3948,
      "step": 29167
    },
    {
      "epoch": 1.3385342572621726,
      "grad_norm": 0.47910308837890625,
      "learning_rate": 6.0159231677192844e-06,
      "loss": 0.3026,
      "step": 29168
    },
    {
      "epoch": 1.3385801477674268,
      "grad_norm": 0.4425312578678131,
      "learning_rate": 6.015683093102227e-06,
      "loss": 0.2365,
      "step": 29169
    },
    {
      "epoch": 1.3386260382726813,
      "grad_norm": 0.3974357545375824,
      "learning_rate": 6.015443016042772e-06,
      "loss": 0.2507,
      "step": 29170
    },
    {
      "epoch": 1.3386719287779358,
      "grad_norm": 0.42777663469314575,
      "learning_rate": 6.015202936541499e-06,
      "loss": 0.2879,
      "step": 29171
    },
    {
      "epoch": 1.3387178192831903,
      "grad_norm": 0.49514320492744446,
      "learning_rate": 6.014962854598981e-06,
      "loss": 0.3383,
      "step": 29172
    },
    {
      "epoch": 1.3387637097884448,
      "grad_norm": 0.4531586468219757,
      "learning_rate": 6.0147227702158e-06,
      "loss": 0.3209,
      "step": 29173
    },
    {
      "epoch": 1.338809600293699,
      "grad_norm": 0.48989927768707275,
      "learning_rate": 6.014482683392529e-06,
      "loss": 0.458,
      "step": 29174
    },
    {
      "epoch": 1.3388554907989536,
      "grad_norm": 0.4495933949947357,
      "learning_rate": 6.014242594129748e-06,
      "loss": 0.3322,
      "step": 29175
    },
    {
      "epoch": 1.338901381304208,
      "grad_norm": 0.44451335072517395,
      "learning_rate": 6.014002502428033e-06,
      "loss": 0.3086,
      "step": 29176
    },
    {
      "epoch": 1.3389472718094626,
      "grad_norm": 0.5433889627456665,
      "learning_rate": 6.013762408287963e-06,
      "loss": 0.3257,
      "step": 29177
    },
    {
      "epoch": 1.338993162314717,
      "grad_norm": 0.4271731674671173,
      "learning_rate": 6.013522311710114e-06,
      "loss": 0.3107,
      "step": 29178
    },
    {
      "epoch": 1.3390390528199716,
      "grad_norm": 0.44732773303985596,
      "learning_rate": 6.013282212695064e-06,
      "loss": 0.3194,
      "step": 29179
    },
    {
      "epoch": 1.339084943325226,
      "grad_norm": 0.4863176941871643,
      "learning_rate": 6.0130421112433905e-06,
      "loss": 0.3817,
      "step": 29180
    },
    {
      "epoch": 1.3391308338304806,
      "grad_norm": 0.4189088046550751,
      "learning_rate": 6.01280200735567e-06,
      "loss": 0.2958,
      "step": 29181
    },
    {
      "epoch": 1.3391767243357349,
      "grad_norm": 0.45848438143730164,
      "learning_rate": 6.01256190103248e-06,
      "loss": 0.3663,
      "step": 29182
    },
    {
      "epoch": 1.3392226148409894,
      "grad_norm": 0.461305171251297,
      "learning_rate": 6.0123217922743975e-06,
      "loss": 0.3468,
      "step": 29183
    },
    {
      "epoch": 1.3392685053462439,
      "grad_norm": 0.4286729395389557,
      "learning_rate": 6.012081681082002e-06,
      "loss": 0.3124,
      "step": 29184
    },
    {
      "epoch": 1.3393143958514984,
      "grad_norm": 0.46640273928642273,
      "learning_rate": 6.011841567455869e-06,
      "loss": 0.3347,
      "step": 29185
    },
    {
      "epoch": 1.3393602863567529,
      "grad_norm": 0.45960262417793274,
      "learning_rate": 6.011601451396576e-06,
      "loss": 0.3709,
      "step": 29186
    },
    {
      "epoch": 1.3394061768620071,
      "grad_norm": 0.5025110840797424,
      "learning_rate": 6.0113613329046995e-06,
      "loss": 0.4115,
      "step": 29187
    },
    {
      "epoch": 1.3394520673672616,
      "grad_norm": 0.43345943093299866,
      "learning_rate": 6.011121211980821e-06,
      "loss": 0.3528,
      "step": 29188
    },
    {
      "epoch": 1.3394979578725161,
      "grad_norm": 0.47966325283050537,
      "learning_rate": 6.010881088625513e-06,
      "loss": 0.3672,
      "step": 29189
    },
    {
      "epoch": 1.3395438483777706,
      "grad_norm": 0.4822651445865631,
      "learning_rate": 6.010640962839356e-06,
      "loss": 0.4286,
      "step": 29190
    },
    {
      "epoch": 1.3395897388830251,
      "grad_norm": 0.46695074439048767,
      "learning_rate": 6.010400834622927e-06,
      "loss": 0.3477,
      "step": 29191
    },
    {
      "epoch": 1.3396356293882796,
      "grad_norm": 0.4966851770877838,
      "learning_rate": 6.010160703976802e-06,
      "loss": 0.3702,
      "step": 29192
    },
    {
      "epoch": 1.3396815198935341,
      "grad_norm": 0.511989951133728,
      "learning_rate": 6.009920570901559e-06,
      "loss": 0.4057,
      "step": 29193
    },
    {
      "epoch": 1.3397274103987886,
      "grad_norm": 0.4585323631763458,
      "learning_rate": 6.0096804353977765e-06,
      "loss": 0.3958,
      "step": 29194
    },
    {
      "epoch": 1.339773300904043,
      "grad_norm": 0.43832558393478394,
      "learning_rate": 6.009440297466031e-06,
      "loss": 0.3008,
      "step": 29195
    },
    {
      "epoch": 1.3398191914092974,
      "grad_norm": 0.48879435658454895,
      "learning_rate": 6.009200157106901e-06,
      "loss": 0.3257,
      "step": 29196
    },
    {
      "epoch": 1.339865081914552,
      "grad_norm": 0.4715830385684967,
      "learning_rate": 6.0089600143209615e-06,
      "loss": 0.3845,
      "step": 29197
    },
    {
      "epoch": 1.3399109724198064,
      "grad_norm": 0.4752826988697052,
      "learning_rate": 6.008719869108793e-06,
      "loss": 0.3562,
      "step": 29198
    },
    {
      "epoch": 1.3399568629250609,
      "grad_norm": 0.4479648768901825,
      "learning_rate": 6.00847972147097e-06,
      "loss": 0.3128,
      "step": 29199
    },
    {
      "epoch": 1.3400027534303152,
      "grad_norm": 0.514366090297699,
      "learning_rate": 6.0082395714080725e-06,
      "loss": 0.444,
      "step": 29200
    },
    {
      "epoch": 1.3400486439355697,
      "grad_norm": 0.47742584347724915,
      "learning_rate": 6.007999418920678e-06,
      "loss": 0.3997,
      "step": 29201
    },
    {
      "epoch": 1.3400945344408242,
      "grad_norm": 0.47242453694343567,
      "learning_rate": 6.007759264009362e-06,
      "loss": 0.4068,
      "step": 29202
    },
    {
      "epoch": 1.3401404249460787,
      "grad_norm": 0.47234049439430237,
      "learning_rate": 6.0075191066747025e-06,
      "loss": 0.3423,
      "step": 29203
    },
    {
      "epoch": 1.3401863154513332,
      "grad_norm": 0.5121144652366638,
      "learning_rate": 6.007278946917278e-06,
      "loss": 0.4467,
      "step": 29204
    },
    {
      "epoch": 1.3402322059565877,
      "grad_norm": 0.45934152603149414,
      "learning_rate": 6.007038784737665e-06,
      "loss": 0.3231,
      "step": 29205
    },
    {
      "epoch": 1.3402780964618421,
      "grad_norm": 0.4620693027973175,
      "learning_rate": 6.006798620136444e-06,
      "loss": 0.3653,
      "step": 29206
    },
    {
      "epoch": 1.3403239869670964,
      "grad_norm": 0.44651469588279724,
      "learning_rate": 6.006558453114187e-06,
      "loss": 0.3069,
      "step": 29207
    },
    {
      "epoch": 1.340369877472351,
      "grad_norm": 0.48637107014656067,
      "learning_rate": 6.006318283671478e-06,
      "loss": 0.3881,
      "step": 29208
    },
    {
      "epoch": 1.3404157679776054,
      "grad_norm": 0.4704943895339966,
      "learning_rate": 6.0060781118088875e-06,
      "loss": 0.3223,
      "step": 29209
    },
    {
      "epoch": 1.34046165848286,
      "grad_norm": 0.4406324326992035,
      "learning_rate": 6.005837937526997e-06,
      "loss": 0.308,
      "step": 29210
    },
    {
      "epoch": 1.3405075489881144,
      "grad_norm": 0.4804038405418396,
      "learning_rate": 6.005597760826386e-06,
      "loss": 0.4133,
      "step": 29211
    },
    {
      "epoch": 1.3405534394933687,
      "grad_norm": 0.45972132682800293,
      "learning_rate": 6.0053575817076294e-06,
      "loss": 0.3278,
      "step": 29212
    },
    {
      "epoch": 1.3405993299986232,
      "grad_norm": 0.46383020281791687,
      "learning_rate": 6.005117400171304e-06,
      "loss": 0.3179,
      "step": 29213
    },
    {
      "epoch": 1.3406452205038777,
      "grad_norm": 0.5002734065055847,
      "learning_rate": 6.004877216217989e-06,
      "loss": 0.3718,
      "step": 29214
    },
    {
      "epoch": 1.3406911110091322,
      "grad_norm": 0.42621147632598877,
      "learning_rate": 6.004637029848261e-06,
      "loss": 0.295,
      "step": 29215
    },
    {
      "epoch": 1.3407370015143867,
      "grad_norm": 0.4734596908092499,
      "learning_rate": 6.004396841062699e-06,
      "loss": 0.3351,
      "step": 29216
    },
    {
      "epoch": 1.3407828920196412,
      "grad_norm": 0.4835848808288574,
      "learning_rate": 6.004156649861879e-06,
      "loss": 0.3252,
      "step": 29217
    },
    {
      "epoch": 1.3408287825248957,
      "grad_norm": 0.4600129723548889,
      "learning_rate": 6.003916456246378e-06,
      "loss": 0.3439,
      "step": 29218
    },
    {
      "epoch": 1.3408746730301502,
      "grad_norm": 0.5287603139877319,
      "learning_rate": 6.0036762602167766e-06,
      "loss": 0.4619,
      "step": 29219
    },
    {
      "epoch": 1.3409205635354045,
      "grad_norm": 0.45106521248817444,
      "learning_rate": 6.00343606177365e-06,
      "loss": 0.3235,
      "step": 29220
    },
    {
      "epoch": 1.340966454040659,
      "grad_norm": 0.4701556861400604,
      "learning_rate": 6.003195860917576e-06,
      "loss": 0.3741,
      "step": 29221
    },
    {
      "epoch": 1.3410123445459134,
      "grad_norm": 0.5195840001106262,
      "learning_rate": 6.0029556576491325e-06,
      "loss": 0.3878,
      "step": 29222
    },
    {
      "epoch": 1.341058235051168,
      "grad_norm": 0.4832955598831177,
      "learning_rate": 6.002715451968898e-06,
      "loss": 0.3408,
      "step": 29223
    },
    {
      "epoch": 1.3411041255564224,
      "grad_norm": 0.6319329738616943,
      "learning_rate": 6.002475243877448e-06,
      "loss": 0.371,
      "step": 29224
    },
    {
      "epoch": 1.3411500160616767,
      "grad_norm": 0.47997426986694336,
      "learning_rate": 6.002235033375363e-06,
      "loss": 0.3273,
      "step": 29225
    },
    {
      "epoch": 1.3411959065669312,
      "grad_norm": 0.46783602237701416,
      "learning_rate": 6.0019948204632175e-06,
      "loss": 0.3058,
      "step": 29226
    },
    {
      "epoch": 1.3412417970721857,
      "grad_norm": 0.46668142080307007,
      "learning_rate": 6.001754605141591e-06,
      "loss": 0.3735,
      "step": 29227
    },
    {
      "epoch": 1.3412876875774402,
      "grad_norm": 0.4915667474269867,
      "learning_rate": 6.001514387411062e-06,
      "loss": 0.4122,
      "step": 29228
    },
    {
      "epoch": 1.3413335780826947,
      "grad_norm": 0.46938446164131165,
      "learning_rate": 6.001274167272206e-06,
      "loss": 0.3416,
      "step": 29229
    },
    {
      "epoch": 1.3413794685879492,
      "grad_norm": 0.45874449610710144,
      "learning_rate": 6.001033944725602e-06,
      "loss": 0.3811,
      "step": 29230
    },
    {
      "epoch": 1.3414253590932037,
      "grad_norm": 0.4958639144897461,
      "learning_rate": 6.0007937197718266e-06,
      "loss": 0.4073,
      "step": 29231
    },
    {
      "epoch": 1.3414712495984582,
      "grad_norm": 0.4950929284095764,
      "learning_rate": 6.00055349241146e-06,
      "loss": 0.3752,
      "step": 29232
    },
    {
      "epoch": 1.3415171401037125,
      "grad_norm": 0.5208605527877808,
      "learning_rate": 6.000313262645077e-06,
      "loss": 0.4061,
      "step": 29233
    },
    {
      "epoch": 1.341563030608967,
      "grad_norm": 0.4799065887928009,
      "learning_rate": 6.000073030473256e-06,
      "loss": 0.3618,
      "step": 29234
    },
    {
      "epoch": 1.3416089211142215,
      "grad_norm": 0.47568821907043457,
      "learning_rate": 5.999832795896574e-06,
      "loss": 0.3414,
      "step": 29235
    },
    {
      "epoch": 1.341654811619476,
      "grad_norm": 0.473205029964447,
      "learning_rate": 5.9995925589156126e-06,
      "loss": 0.3282,
      "step": 29236
    },
    {
      "epoch": 1.3417007021247305,
      "grad_norm": 0.45543763041496277,
      "learning_rate": 5.999352319530943e-06,
      "loss": 0.3263,
      "step": 29237
    },
    {
      "epoch": 1.3417465926299847,
      "grad_norm": 0.491754949092865,
      "learning_rate": 5.999112077743148e-06,
      "loss": 0.4378,
      "step": 29238
    },
    {
      "epoch": 1.3417924831352392,
      "grad_norm": 0.5080037713050842,
      "learning_rate": 5.998871833552806e-06,
      "loss": 0.4605,
      "step": 29239
    },
    {
      "epoch": 1.3418383736404937,
      "grad_norm": 0.49499720335006714,
      "learning_rate": 5.998631586960489e-06,
      "loss": 0.4204,
      "step": 29240
    },
    {
      "epoch": 1.3418842641457482,
      "grad_norm": 0.4667193591594696,
      "learning_rate": 5.99839133796678e-06,
      "loss": 0.3662,
      "step": 29241
    },
    {
      "epoch": 1.3419301546510027,
      "grad_norm": 0.47726163268089294,
      "learning_rate": 5.998151086572255e-06,
      "loss": 0.4058,
      "step": 29242
    },
    {
      "epoch": 1.3419760451562572,
      "grad_norm": 0.45335137844085693,
      "learning_rate": 5.997910832777491e-06,
      "loss": 0.3492,
      "step": 29243
    },
    {
      "epoch": 1.3420219356615117,
      "grad_norm": 0.49384984374046326,
      "learning_rate": 5.997670576583066e-06,
      "loss": 0.3954,
      "step": 29244
    },
    {
      "epoch": 1.3420678261667662,
      "grad_norm": 0.4665132462978363,
      "learning_rate": 5.997430317989561e-06,
      "loss": 0.3708,
      "step": 29245
    },
    {
      "epoch": 1.3421137166720205,
      "grad_norm": 0.4168911278247833,
      "learning_rate": 5.997190056997548e-06,
      "loss": 0.289,
      "step": 29246
    },
    {
      "epoch": 1.342159607177275,
      "grad_norm": 0.5059893131256104,
      "learning_rate": 5.996949793607608e-06,
      "loss": 0.4336,
      "step": 29247
    },
    {
      "epoch": 1.3422054976825295,
      "grad_norm": 0.4751115143299103,
      "learning_rate": 5.996709527820319e-06,
      "loss": 0.3896,
      "step": 29248
    },
    {
      "epoch": 1.342251388187784,
      "grad_norm": 0.4761570990085602,
      "learning_rate": 5.996469259636257e-06,
      "loss": 0.3447,
      "step": 29249
    },
    {
      "epoch": 1.3422972786930383,
      "grad_norm": 0.46875420212745667,
      "learning_rate": 5.996228989056004e-06,
      "loss": 0.3623,
      "step": 29250
    },
    {
      "epoch": 1.3423431691982928,
      "grad_norm": 0.4587228298187256,
      "learning_rate": 5.995988716080131e-06,
      "loss": 0.3514,
      "step": 29251
    },
    {
      "epoch": 1.3423890597035473,
      "grad_norm": 0.5551150441169739,
      "learning_rate": 5.995748440709221e-06,
      "loss": 0.3841,
      "step": 29252
    },
    {
      "epoch": 1.3424349502088018,
      "grad_norm": 0.45440763235092163,
      "learning_rate": 5.995508162943851e-06,
      "loss": 0.3104,
      "step": 29253
    },
    {
      "epoch": 1.3424808407140563,
      "grad_norm": 0.441155344247818,
      "learning_rate": 5.995267882784597e-06,
      "loss": 0.3069,
      "step": 29254
    },
    {
      "epoch": 1.3425267312193108,
      "grad_norm": 0.4764816462993622,
      "learning_rate": 5.995027600232038e-06,
      "loss": 0.3589,
      "step": 29255
    },
    {
      "epoch": 1.3425726217245653,
      "grad_norm": 0.47266799211502075,
      "learning_rate": 5.994787315286753e-06,
      "loss": 0.3876,
      "step": 29256
    },
    {
      "epoch": 1.3426185122298198,
      "grad_norm": 0.46778780221939087,
      "learning_rate": 5.994547027949317e-06,
      "loss": 0.4038,
      "step": 29257
    },
    {
      "epoch": 1.342664402735074,
      "grad_norm": 0.4520222842693329,
      "learning_rate": 5.994306738220308e-06,
      "loss": 0.3226,
      "step": 29258
    },
    {
      "epoch": 1.3427102932403285,
      "grad_norm": 0.4519561231136322,
      "learning_rate": 5.9940664461003086e-06,
      "loss": 0.3265,
      "step": 29259
    },
    {
      "epoch": 1.342756183745583,
      "grad_norm": 0.46091336011886597,
      "learning_rate": 5.99382615158989e-06,
      "loss": 0.3535,
      "step": 29260
    },
    {
      "epoch": 1.3428020742508375,
      "grad_norm": 0.4783730208873749,
      "learning_rate": 5.993585854689637e-06,
      "loss": 0.3242,
      "step": 29261
    },
    {
      "epoch": 1.342847964756092,
      "grad_norm": 0.48416611552238464,
      "learning_rate": 5.99334555540012e-06,
      "loss": 0.3499,
      "step": 29262
    },
    {
      "epoch": 1.3428938552613463,
      "grad_norm": 0.4815567135810852,
      "learning_rate": 5.99310525372192e-06,
      "loss": 0.3739,
      "step": 29263
    },
    {
      "epoch": 1.3429397457666008,
      "grad_norm": 0.4653523862361908,
      "learning_rate": 5.9928649496556175e-06,
      "loss": 0.372,
      "step": 29264
    },
    {
      "epoch": 1.3429856362718553,
      "grad_norm": 0.45613953471183777,
      "learning_rate": 5.9926246432017875e-06,
      "loss": 0.3495,
      "step": 29265
    },
    {
      "epoch": 1.3430315267771098,
      "grad_norm": 0.49128955602645874,
      "learning_rate": 5.9923843343610085e-06,
      "loss": 0.3955,
      "step": 29266
    },
    {
      "epoch": 1.3430774172823643,
      "grad_norm": 0.44284576177597046,
      "learning_rate": 5.992144023133859e-06,
      "loss": 0.3216,
      "step": 29267
    },
    {
      "epoch": 1.3431233077876188,
      "grad_norm": 0.496848464012146,
      "learning_rate": 5.991903709520915e-06,
      "loss": 0.3998,
      "step": 29268
    },
    {
      "epoch": 1.3431691982928733,
      "grad_norm": 0.4777551293373108,
      "learning_rate": 5.991663393522755e-06,
      "loss": 0.3358,
      "step": 29269
    },
    {
      "epoch": 1.3432150887981278,
      "grad_norm": 0.5262637138366699,
      "learning_rate": 5.991423075139959e-06,
      "loss": 0.4132,
      "step": 29270
    },
    {
      "epoch": 1.343260979303382,
      "grad_norm": 0.48929035663604736,
      "learning_rate": 5.991182754373104e-06,
      "loss": 0.3423,
      "step": 29271
    },
    {
      "epoch": 1.3433068698086366,
      "grad_norm": 0.4848423898220062,
      "learning_rate": 5.990942431222766e-06,
      "loss": 0.3328,
      "step": 29272
    },
    {
      "epoch": 1.343352760313891,
      "grad_norm": 0.4752117991447449,
      "learning_rate": 5.990702105689525e-06,
      "loss": 0.412,
      "step": 29273
    },
    {
      "epoch": 1.3433986508191456,
      "grad_norm": 0.4373140335083008,
      "learning_rate": 5.990461777773958e-06,
      "loss": 0.3372,
      "step": 29274
    },
    {
      "epoch": 1.3434445413244,
      "grad_norm": 0.46440523862838745,
      "learning_rate": 5.990221447476642e-06,
      "loss": 0.359,
      "step": 29275
    },
    {
      "epoch": 1.3434904318296543,
      "grad_norm": 0.46621769666671753,
      "learning_rate": 5.989981114798157e-06,
      "loss": 0.4351,
      "step": 29276
    },
    {
      "epoch": 1.3435363223349088,
      "grad_norm": 0.48791220784187317,
      "learning_rate": 5.9897407797390805e-06,
      "loss": 0.3422,
      "step": 29277
    },
    {
      "epoch": 1.3435822128401633,
      "grad_norm": 0.4702817499637604,
      "learning_rate": 5.989500442299989e-06,
      "loss": 0.3847,
      "step": 29278
    },
    {
      "epoch": 1.3436281033454178,
      "grad_norm": 0.4191371500492096,
      "learning_rate": 5.989260102481461e-06,
      "loss": 0.2699,
      "step": 29279
    },
    {
      "epoch": 1.3436739938506723,
      "grad_norm": 0.4672558307647705,
      "learning_rate": 5.989019760284074e-06,
      "loss": 0.3237,
      "step": 29280
    },
    {
      "epoch": 1.3437198843559268,
      "grad_norm": 0.4731612503528595,
      "learning_rate": 5.988779415708408e-06,
      "loss": 0.3365,
      "step": 29281
    },
    {
      "epoch": 1.3437657748611813,
      "grad_norm": 0.4600823223590851,
      "learning_rate": 5.988539068755039e-06,
      "loss": 0.3741,
      "step": 29282
    },
    {
      "epoch": 1.3438116653664358,
      "grad_norm": 0.4978110194206238,
      "learning_rate": 5.988298719424546e-06,
      "loss": 0.4088,
      "step": 29283
    },
    {
      "epoch": 1.34385755587169,
      "grad_norm": 0.4753190875053406,
      "learning_rate": 5.988058367717506e-06,
      "loss": 0.3013,
      "step": 29284
    },
    {
      "epoch": 1.3439034463769446,
      "grad_norm": 0.46011069416999817,
      "learning_rate": 5.987818013634497e-06,
      "loss": 0.3252,
      "step": 29285
    },
    {
      "epoch": 1.343949336882199,
      "grad_norm": 0.4473492205142975,
      "learning_rate": 5.987577657176097e-06,
      "loss": 0.3192,
      "step": 29286
    },
    {
      "epoch": 1.3439952273874536,
      "grad_norm": 0.4538627862930298,
      "learning_rate": 5.9873372983428855e-06,
      "loss": 0.3417,
      "step": 29287
    },
    {
      "epoch": 1.344041117892708,
      "grad_norm": 0.5157557725906372,
      "learning_rate": 5.987096937135441e-06,
      "loss": 0.4165,
      "step": 29288
    },
    {
      "epoch": 1.3440870083979624,
      "grad_norm": 0.5188693404197693,
      "learning_rate": 5.9868565735543375e-06,
      "loss": 0.4651,
      "step": 29289
    },
    {
      "epoch": 1.3441328989032169,
      "grad_norm": 0.4536497890949249,
      "learning_rate": 5.9866162076001555e-06,
      "loss": 0.3454,
      "step": 29290
    },
    {
      "epoch": 1.3441787894084714,
      "grad_norm": 0.46054399013519287,
      "learning_rate": 5.986375839273474e-06,
      "loss": 0.3482,
      "step": 29291
    },
    {
      "epoch": 1.3442246799137259,
      "grad_norm": 0.4543088674545288,
      "learning_rate": 5.986135468574868e-06,
      "loss": 0.3568,
      "step": 29292
    },
    {
      "epoch": 1.3442705704189803,
      "grad_norm": 0.4450756907463074,
      "learning_rate": 5.98589509550492e-06,
      "loss": 0.3319,
      "step": 29293
    },
    {
      "epoch": 1.3443164609242348,
      "grad_norm": 0.4665926992893219,
      "learning_rate": 5.9856547200642045e-06,
      "loss": 0.3656,
      "step": 29294
    },
    {
      "epoch": 1.3443623514294893,
      "grad_norm": 0.47160178422927856,
      "learning_rate": 5.985414342253301e-06,
      "loss": 0.345,
      "step": 29295
    },
    {
      "epoch": 1.3444082419347436,
      "grad_norm": 0.47906485199928284,
      "learning_rate": 5.985173962072786e-06,
      "loss": 0.412,
      "step": 29296
    },
    {
      "epoch": 1.3444541324399981,
      "grad_norm": 0.49599015712738037,
      "learning_rate": 5.984933579523238e-06,
      "loss": 0.402,
      "step": 29297
    },
    {
      "epoch": 1.3445000229452526,
      "grad_norm": 0.43745845556259155,
      "learning_rate": 5.9846931946052365e-06,
      "loss": 0.3526,
      "step": 29298
    },
    {
      "epoch": 1.3445459134505071,
      "grad_norm": 0.4768558442592621,
      "learning_rate": 5.984452807319359e-06,
      "loss": 0.3544,
      "step": 29299
    },
    {
      "epoch": 1.3445918039557616,
      "grad_norm": 0.4561288058757782,
      "learning_rate": 5.984212417666182e-06,
      "loss": 0.3838,
      "step": 29300
    },
    {
      "epoch": 1.3446376944610159,
      "grad_norm": 0.4625436067581177,
      "learning_rate": 5.983972025646286e-06,
      "loss": 0.377,
      "step": 29301
    },
    {
      "epoch": 1.3446835849662704,
      "grad_norm": 0.5314272046089172,
      "learning_rate": 5.983731631260247e-06,
      "loss": 0.3961,
      "step": 29302
    },
    {
      "epoch": 1.3447294754715249,
      "grad_norm": 0.47937533259391785,
      "learning_rate": 5.983491234508644e-06,
      "loss": 0.3598,
      "step": 29303
    },
    {
      "epoch": 1.3447753659767794,
      "grad_norm": 0.5026845335960388,
      "learning_rate": 5.9832508353920556e-06,
      "loss": 0.4013,
      "step": 29304
    },
    {
      "epoch": 1.3448212564820339,
      "grad_norm": 0.48508208990097046,
      "learning_rate": 5.9830104339110595e-06,
      "loss": 0.3907,
      "step": 29305
    },
    {
      "epoch": 1.3448671469872884,
      "grad_norm": 0.45594292879104614,
      "learning_rate": 5.982770030066232e-06,
      "loss": 0.35,
      "step": 29306
    },
    {
      "epoch": 1.3449130374925429,
      "grad_norm": 0.4764085114002228,
      "learning_rate": 5.982529623858154e-06,
      "loss": 0.3926,
      "step": 29307
    },
    {
      "epoch": 1.3449589279977974,
      "grad_norm": 0.43037670850753784,
      "learning_rate": 5.982289215287402e-06,
      "loss": 0.2791,
      "step": 29308
    },
    {
      "epoch": 1.3450048185030516,
      "grad_norm": 0.47275522351264954,
      "learning_rate": 5.9820488043545545e-06,
      "loss": 0.3858,
      "step": 29309
    },
    {
      "epoch": 1.3450507090083061,
      "grad_norm": 0.44493111968040466,
      "learning_rate": 5.981808391060191e-06,
      "loss": 0.3403,
      "step": 29310
    },
    {
      "epoch": 1.3450965995135606,
      "grad_norm": 0.4585282802581787,
      "learning_rate": 5.981567975404885e-06,
      "loss": 0.3438,
      "step": 29311
    },
    {
      "epoch": 1.3451424900188151,
      "grad_norm": 0.43969306349754333,
      "learning_rate": 5.981327557389221e-06,
      "loss": 0.3205,
      "step": 29312
    },
    {
      "epoch": 1.3451883805240696,
      "grad_norm": 0.5262542963027954,
      "learning_rate": 5.9810871370137715e-06,
      "loss": 0.4454,
      "step": 29313
    },
    {
      "epoch": 1.345234271029324,
      "grad_norm": 0.435591459274292,
      "learning_rate": 5.980846714279118e-06,
      "loss": 0.3106,
      "step": 29314
    },
    {
      "epoch": 1.3452801615345784,
      "grad_norm": 0.45104286074638367,
      "learning_rate": 5.9806062891858376e-06,
      "loss": 0.3516,
      "step": 29315
    },
    {
      "epoch": 1.345326052039833,
      "grad_norm": 0.5893418788909912,
      "learning_rate": 5.98036586173451e-06,
      "loss": 0.4378,
      "step": 29316
    },
    {
      "epoch": 1.3453719425450874,
      "grad_norm": 0.45053625106811523,
      "learning_rate": 5.98012543192571e-06,
      "loss": 0.3369,
      "step": 29317
    },
    {
      "epoch": 1.345417833050342,
      "grad_norm": 0.4683852791786194,
      "learning_rate": 5.9798849997600185e-06,
      "loss": 0.3786,
      "step": 29318
    },
    {
      "epoch": 1.3454637235555964,
      "grad_norm": 0.4448968470096588,
      "learning_rate": 5.979644565238012e-06,
      "loss": 0.3393,
      "step": 29319
    },
    {
      "epoch": 1.345509614060851,
      "grad_norm": 0.4790360629558563,
      "learning_rate": 5.979404128360271e-06,
      "loss": 0.4087,
      "step": 29320
    },
    {
      "epoch": 1.3455555045661054,
      "grad_norm": 0.45943203568458557,
      "learning_rate": 5.979163689127372e-06,
      "loss": 0.3557,
      "step": 29321
    },
    {
      "epoch": 1.3456013950713597,
      "grad_norm": 0.4400697648525238,
      "learning_rate": 5.978923247539892e-06,
      "loss": 0.3112,
      "step": 29322
    },
    {
      "epoch": 1.3456472855766142,
      "grad_norm": 0.45252957940101624,
      "learning_rate": 5.97868280359841e-06,
      "loss": 0.3261,
      "step": 29323
    },
    {
      "epoch": 1.3456931760818687,
      "grad_norm": 0.4798428416252136,
      "learning_rate": 5.978442357303507e-06,
      "loss": 0.4099,
      "step": 29324
    },
    {
      "epoch": 1.3457390665871232,
      "grad_norm": 0.4590834379196167,
      "learning_rate": 5.978201908655757e-06,
      "loss": 0.3107,
      "step": 29325
    },
    {
      "epoch": 1.3457849570923777,
      "grad_norm": 0.4403301477432251,
      "learning_rate": 5.977961457655742e-06,
      "loss": 0.3357,
      "step": 29326
    },
    {
      "epoch": 1.345830847597632,
      "grad_norm": 0.46569088101387024,
      "learning_rate": 5.977721004304038e-06,
      "loss": 0.343,
      "step": 29327
    },
    {
      "epoch": 1.3458767381028864,
      "grad_norm": 0.4549632668495178,
      "learning_rate": 5.977480548601222e-06,
      "loss": 0.3545,
      "step": 29328
    },
    {
      "epoch": 1.345922628608141,
      "grad_norm": 0.4642695486545563,
      "learning_rate": 5.977240090547876e-06,
      "loss": 0.3385,
      "step": 29329
    },
    {
      "epoch": 1.3459685191133954,
      "grad_norm": 0.49309054017066956,
      "learning_rate": 5.976999630144574e-06,
      "loss": 0.3843,
      "step": 29330
    },
    {
      "epoch": 1.34601440961865,
      "grad_norm": 0.44496285915374756,
      "learning_rate": 5.9767591673918975e-06,
      "loss": 0.3335,
      "step": 29331
    },
    {
      "epoch": 1.3460603001239044,
      "grad_norm": 0.4439498484134674,
      "learning_rate": 5.976518702290424e-06,
      "loss": 0.3199,
      "step": 29332
    },
    {
      "epoch": 1.346106190629159,
      "grad_norm": 0.4783628284931183,
      "learning_rate": 5.976278234840729e-06,
      "loss": 0.359,
      "step": 29333
    },
    {
      "epoch": 1.3461520811344134,
      "grad_norm": 0.4711025059223175,
      "learning_rate": 5.976037765043394e-06,
      "loss": 0.3652,
      "step": 29334
    },
    {
      "epoch": 1.3461979716396677,
      "grad_norm": 0.441425621509552,
      "learning_rate": 5.975797292898997e-06,
      "loss": 0.3146,
      "step": 29335
    },
    {
      "epoch": 1.3462438621449222,
      "grad_norm": 0.47010689973831177,
      "learning_rate": 5.975556818408115e-06,
      "loss": 0.3297,
      "step": 29336
    },
    {
      "epoch": 1.3462897526501767,
      "grad_norm": 0.4925490617752075,
      "learning_rate": 5.975316341571326e-06,
      "loss": 0.3939,
      "step": 29337
    },
    {
      "epoch": 1.3463356431554312,
      "grad_norm": 0.5271863341331482,
      "learning_rate": 5.9750758623892105e-06,
      "loss": 0.3532,
      "step": 29338
    },
    {
      "epoch": 1.3463815336606855,
      "grad_norm": 0.5285645723342896,
      "learning_rate": 5.974835380862344e-06,
      "loss": 0.4372,
      "step": 29339
    },
    {
      "epoch": 1.34642742416594,
      "grad_norm": 0.5414443612098694,
      "learning_rate": 5.974594896991306e-06,
      "loss": 0.4078,
      "step": 29340
    },
    {
      "epoch": 1.3464733146711945,
      "grad_norm": 0.4740733206272125,
      "learning_rate": 5.974354410776676e-06,
      "loss": 0.3625,
      "step": 29341
    },
    {
      "epoch": 1.346519205176449,
      "grad_norm": 0.48613595962524414,
      "learning_rate": 5.974113922219031e-06,
      "loss": 0.4014,
      "step": 29342
    },
    {
      "epoch": 1.3465650956817035,
      "grad_norm": 0.4630837142467499,
      "learning_rate": 5.97387343131895e-06,
      "loss": 0.3284,
      "step": 29343
    },
    {
      "epoch": 1.346610986186958,
      "grad_norm": 0.4600975215435028,
      "learning_rate": 5.9736329380770085e-06,
      "loss": 0.3571,
      "step": 29344
    },
    {
      "epoch": 1.3466568766922125,
      "grad_norm": 0.42264437675476074,
      "learning_rate": 5.973392442493789e-06,
      "loss": 0.2779,
      "step": 29345
    },
    {
      "epoch": 1.346702767197467,
      "grad_norm": 0.44337841868400574,
      "learning_rate": 5.973151944569866e-06,
      "loss": 0.3246,
      "step": 29346
    },
    {
      "epoch": 1.3467486577027212,
      "grad_norm": 0.5059216022491455,
      "learning_rate": 5.972911444305821e-06,
      "loss": 0.4,
      "step": 29347
    },
    {
      "epoch": 1.3467945482079757,
      "grad_norm": 0.4399208426475525,
      "learning_rate": 5.972670941702231e-06,
      "loss": 0.3285,
      "step": 29348
    },
    {
      "epoch": 1.3468404387132302,
      "grad_norm": 0.4736819565296173,
      "learning_rate": 5.9724304367596755e-06,
      "loss": 0.3099,
      "step": 29349
    },
    {
      "epoch": 1.3468863292184847,
      "grad_norm": 0.4642826318740845,
      "learning_rate": 5.97218992947873e-06,
      "loss": 0.3158,
      "step": 29350
    },
    {
      "epoch": 1.3469322197237392,
      "grad_norm": 0.464168518781662,
      "learning_rate": 5.971949419859975e-06,
      "loss": 0.3976,
      "step": 29351
    },
    {
      "epoch": 1.3469781102289935,
      "grad_norm": 0.46343088150024414,
      "learning_rate": 5.9717089079039885e-06,
      "loss": 0.3739,
      "step": 29352
    },
    {
      "epoch": 1.347024000734248,
      "grad_norm": 0.5674736499786377,
      "learning_rate": 5.971468393611349e-06,
      "loss": 0.3085,
      "step": 29353
    },
    {
      "epoch": 1.3470698912395025,
      "grad_norm": 0.5191990733146667,
      "learning_rate": 5.971227876982634e-06,
      "loss": 0.4908,
      "step": 29354
    },
    {
      "epoch": 1.347115781744757,
      "grad_norm": 0.4946841299533844,
      "learning_rate": 5.970987358018423e-06,
      "loss": 0.3823,
      "step": 29355
    },
    {
      "epoch": 1.3471616722500115,
      "grad_norm": 0.4473111927509308,
      "learning_rate": 5.970746836719293e-06,
      "loss": 0.3305,
      "step": 29356
    },
    {
      "epoch": 1.347207562755266,
      "grad_norm": 0.47448456287384033,
      "learning_rate": 5.970506313085824e-06,
      "loss": 0.3733,
      "step": 29357
    },
    {
      "epoch": 1.3472534532605205,
      "grad_norm": 0.4469602108001709,
      "learning_rate": 5.970265787118595e-06,
      "loss": 0.3276,
      "step": 29358
    },
    {
      "epoch": 1.347299343765775,
      "grad_norm": 0.5028883814811707,
      "learning_rate": 5.970025258818181e-06,
      "loss": 0.4114,
      "step": 29359
    },
    {
      "epoch": 1.3473452342710293,
      "grad_norm": 0.47064855694770813,
      "learning_rate": 5.969784728185164e-06,
      "loss": 0.315,
      "step": 29360
    },
    {
      "epoch": 1.3473911247762838,
      "grad_norm": 0.5612934231758118,
      "learning_rate": 5.969544195220118e-06,
      "loss": 0.3678,
      "step": 29361
    },
    {
      "epoch": 1.3474370152815383,
      "grad_norm": 0.6244404911994934,
      "learning_rate": 5.969303659923626e-06,
      "loss": 0.3793,
      "step": 29362
    },
    {
      "epoch": 1.3474829057867928,
      "grad_norm": 0.4776374101638794,
      "learning_rate": 5.969063122296265e-06,
      "loss": 0.3273,
      "step": 29363
    },
    {
      "epoch": 1.3475287962920472,
      "grad_norm": 0.5199674367904663,
      "learning_rate": 5.968822582338614e-06,
      "loss": 0.3704,
      "step": 29364
    },
    {
      "epoch": 1.3475746867973015,
      "grad_norm": 0.45372694730758667,
      "learning_rate": 5.968582040051247e-06,
      "loss": 0.3394,
      "step": 29365
    },
    {
      "epoch": 1.347620577302556,
      "grad_norm": 0.6402072906494141,
      "learning_rate": 5.968341495434749e-06,
      "loss": 0.4561,
      "step": 29366
    },
    {
      "epoch": 1.3476664678078105,
      "grad_norm": 0.5149707794189453,
      "learning_rate": 5.9681009484896944e-06,
      "loss": 0.4432,
      "step": 29367
    },
    {
      "epoch": 1.347712358313065,
      "grad_norm": 0.5099077224731445,
      "learning_rate": 5.967860399216662e-06,
      "loss": 0.4173,
      "step": 29368
    },
    {
      "epoch": 1.3477582488183195,
      "grad_norm": 0.46264445781707764,
      "learning_rate": 5.967619847616232e-06,
      "loss": 0.3654,
      "step": 29369
    },
    {
      "epoch": 1.347804139323574,
      "grad_norm": 0.4301794469356537,
      "learning_rate": 5.967379293688981e-06,
      "loss": 0.2864,
      "step": 29370
    },
    {
      "epoch": 1.3478500298288285,
      "grad_norm": 0.4316639006137848,
      "learning_rate": 5.967138737435487e-06,
      "loss": 0.2683,
      "step": 29371
    },
    {
      "epoch": 1.347895920334083,
      "grad_norm": 0.49088308215141296,
      "learning_rate": 5.966898178856332e-06,
      "loss": 0.3861,
      "step": 29372
    },
    {
      "epoch": 1.3479418108393373,
      "grad_norm": 0.4735565483570099,
      "learning_rate": 5.9666576179520895e-06,
      "loss": 0.3481,
      "step": 29373
    },
    {
      "epoch": 1.3479877013445918,
      "grad_norm": 0.4919901490211487,
      "learning_rate": 5.966417054723343e-06,
      "loss": 0.3967,
      "step": 29374
    },
    {
      "epoch": 1.3480335918498463,
      "grad_norm": 0.45784762501716614,
      "learning_rate": 5.9661764891706664e-06,
      "loss": 0.3617,
      "step": 29375
    },
    {
      "epoch": 1.3480794823551008,
      "grad_norm": 0.46149709820747375,
      "learning_rate": 5.965935921294641e-06,
      "loss": 0.3855,
      "step": 29376
    },
    {
      "epoch": 1.3481253728603553,
      "grad_norm": 0.4719356298446655,
      "learning_rate": 5.965695351095845e-06,
      "loss": 0.3653,
      "step": 29377
    },
    {
      "epoch": 1.3481712633656096,
      "grad_norm": 0.48319998383522034,
      "learning_rate": 5.965454778574855e-06,
      "loss": 0.3695,
      "step": 29378
    },
    {
      "epoch": 1.348217153870864,
      "grad_norm": 0.4846431314945221,
      "learning_rate": 5.965214203732252e-06,
      "loss": 0.3842,
      "step": 29379
    },
    {
      "epoch": 1.3482630443761185,
      "grad_norm": 0.5122337341308594,
      "learning_rate": 5.964973626568614e-06,
      "loss": 0.4154,
      "step": 29380
    },
    {
      "epoch": 1.348308934881373,
      "grad_norm": 0.47419509291648865,
      "learning_rate": 5.964733047084519e-06,
      "loss": 0.4186,
      "step": 29381
    },
    {
      "epoch": 1.3483548253866275,
      "grad_norm": 0.472273588180542,
      "learning_rate": 5.964492465280544e-06,
      "loss": 0.3713,
      "step": 29382
    },
    {
      "epoch": 1.348400715891882,
      "grad_norm": 0.46414420008659363,
      "learning_rate": 5.964251881157271e-06,
      "loss": 0.3691,
      "step": 29383
    },
    {
      "epoch": 1.3484466063971365,
      "grad_norm": 0.4281494915485382,
      "learning_rate": 5.964011294715275e-06,
      "loss": 0.3025,
      "step": 29384
    },
    {
      "epoch": 1.3484924969023908,
      "grad_norm": 0.47918418049812317,
      "learning_rate": 5.9637707059551375e-06,
      "loss": 0.3895,
      "step": 29385
    },
    {
      "epoch": 1.3485383874076453,
      "grad_norm": 0.47423022985458374,
      "learning_rate": 5.963530114877435e-06,
      "loss": 0.3758,
      "step": 29386
    },
    {
      "epoch": 1.3485842779128998,
      "grad_norm": 0.5077845454216003,
      "learning_rate": 5.963289521482746e-06,
      "loss": 0.4373,
      "step": 29387
    },
    {
      "epoch": 1.3486301684181543,
      "grad_norm": 0.4876904785633087,
      "learning_rate": 5.963048925771649e-06,
      "loss": 0.3778,
      "step": 29388
    },
    {
      "epoch": 1.3486760589234088,
      "grad_norm": 0.42401590943336487,
      "learning_rate": 5.962808327744726e-06,
      "loss": 0.2791,
      "step": 29389
    },
    {
      "epoch": 1.348721949428663,
      "grad_norm": 0.47485730051994324,
      "learning_rate": 5.9625677274025505e-06,
      "loss": 0.3777,
      "step": 29390
    },
    {
      "epoch": 1.3487678399339176,
      "grad_norm": 0.4733969271183014,
      "learning_rate": 5.9623271247457035e-06,
      "loss": 0.3516,
      "step": 29391
    },
    {
      "epoch": 1.348813730439172,
      "grad_norm": 0.44413211941719055,
      "learning_rate": 5.962086519774766e-06,
      "loss": 0.3498,
      "step": 29392
    },
    {
      "epoch": 1.3488596209444266,
      "grad_norm": 0.43104058504104614,
      "learning_rate": 5.961845912490312e-06,
      "loss": 0.3161,
      "step": 29393
    },
    {
      "epoch": 1.348905511449681,
      "grad_norm": 0.4933432936668396,
      "learning_rate": 5.961605302892922e-06,
      "loss": 0.3813,
      "step": 29394
    },
    {
      "epoch": 1.3489514019549356,
      "grad_norm": 0.49444133043289185,
      "learning_rate": 5.961364690983175e-06,
      "loss": 0.397,
      "step": 29395
    },
    {
      "epoch": 1.34899729246019,
      "grad_norm": 0.4660189151763916,
      "learning_rate": 5.9611240767616496e-06,
      "loss": 0.3517,
      "step": 29396
    },
    {
      "epoch": 1.3490431829654446,
      "grad_norm": 0.4854762554168701,
      "learning_rate": 5.960883460228925e-06,
      "loss": 0.3957,
      "step": 29397
    },
    {
      "epoch": 1.3490890734706988,
      "grad_norm": 0.4800436496734619,
      "learning_rate": 5.960642841385576e-06,
      "loss": 0.3725,
      "step": 29398
    },
    {
      "epoch": 1.3491349639759533,
      "grad_norm": 0.47272545099258423,
      "learning_rate": 5.960402220232187e-06,
      "loss": 0.3704,
      "step": 29399
    },
    {
      "epoch": 1.3491808544812078,
      "grad_norm": 0.4724104404449463,
      "learning_rate": 5.960161596769333e-06,
      "loss": 0.3767,
      "step": 29400
    },
    {
      "epoch": 1.3492267449864623,
      "grad_norm": 0.4776090979576111,
      "learning_rate": 5.959920970997592e-06,
      "loss": 0.3849,
      "step": 29401
    },
    {
      "epoch": 1.3492726354917168,
      "grad_norm": 0.46469059586524963,
      "learning_rate": 5.959680342917547e-06,
      "loss": 0.364,
      "step": 29402
    },
    {
      "epoch": 1.349318525996971,
      "grad_norm": 0.4276329278945923,
      "learning_rate": 5.959439712529772e-06,
      "loss": 0.2984,
      "step": 29403
    },
    {
      "epoch": 1.3493644165022256,
      "grad_norm": 0.44224753975868225,
      "learning_rate": 5.959199079834847e-06,
      "loss": 0.3053,
      "step": 29404
    },
    {
      "epoch": 1.34941030700748,
      "grad_norm": 0.4662752151489258,
      "learning_rate": 5.9589584448333505e-06,
      "loss": 0.3575,
      "step": 29405
    },
    {
      "epoch": 1.3494561975127346,
      "grad_norm": 0.4530302882194519,
      "learning_rate": 5.958717807525862e-06,
      "loss": 0.3401,
      "step": 29406
    },
    {
      "epoch": 1.349502088017989,
      "grad_norm": 0.4537728428840637,
      "learning_rate": 5.958477167912961e-06,
      "loss": 0.3228,
      "step": 29407
    },
    {
      "epoch": 1.3495479785232436,
      "grad_norm": 0.46529683470726013,
      "learning_rate": 5.958236525995225e-06,
      "loss": 0.3664,
      "step": 29408
    },
    {
      "epoch": 1.349593869028498,
      "grad_norm": 0.5176889300346375,
      "learning_rate": 5.957995881773231e-06,
      "loss": 0.4395,
      "step": 29409
    },
    {
      "epoch": 1.3496397595337526,
      "grad_norm": 0.43422916531562805,
      "learning_rate": 5.957755235247558e-06,
      "loss": 0.2814,
      "step": 29410
    },
    {
      "epoch": 1.3496856500390069,
      "grad_norm": 0.5037030577659607,
      "learning_rate": 5.957514586418789e-06,
      "loss": 0.362,
      "step": 29411
    },
    {
      "epoch": 1.3497315405442614,
      "grad_norm": 0.45877185463905334,
      "learning_rate": 5.957273935287498e-06,
      "loss": 0.3481,
      "step": 29412
    },
    {
      "epoch": 1.3497774310495159,
      "grad_norm": 0.4633530080318451,
      "learning_rate": 5.957033281854266e-06,
      "loss": 0.3626,
      "step": 29413
    },
    {
      "epoch": 1.3498233215547704,
      "grad_norm": 0.4563622772693634,
      "learning_rate": 5.956792626119672e-06,
      "loss": 0.3554,
      "step": 29414
    },
    {
      "epoch": 1.3498692120600249,
      "grad_norm": 0.4340989589691162,
      "learning_rate": 5.956551968084291e-06,
      "loss": 0.3199,
      "step": 29415
    },
    {
      "epoch": 1.3499151025652791,
      "grad_norm": 0.4400111138820648,
      "learning_rate": 5.956311307748706e-06,
      "loss": 0.2877,
      "step": 29416
    },
    {
      "epoch": 1.3499609930705336,
      "grad_norm": 0.4630332887172699,
      "learning_rate": 5.9560706451134945e-06,
      "loss": 0.359,
      "step": 29417
    },
    {
      "epoch": 1.3500068835757881,
      "grad_norm": 0.4595993757247925,
      "learning_rate": 5.955829980179234e-06,
      "loss": 0.355,
      "step": 29418
    },
    {
      "epoch": 1.3500527740810426,
      "grad_norm": 0.4341086149215698,
      "learning_rate": 5.955589312946504e-06,
      "loss": 0.3327,
      "step": 29419
    },
    {
      "epoch": 1.3500986645862971,
      "grad_norm": 0.4963456690311432,
      "learning_rate": 5.955348643415884e-06,
      "loss": 0.3799,
      "step": 29420
    },
    {
      "epoch": 1.3501445550915516,
      "grad_norm": 0.4864499270915985,
      "learning_rate": 5.955107971587951e-06,
      "loss": 0.3732,
      "step": 29421
    },
    {
      "epoch": 1.3501904455968061,
      "grad_norm": 0.44777753949165344,
      "learning_rate": 5.9548672974632845e-06,
      "loss": 0.3223,
      "step": 29422
    },
    {
      "epoch": 1.3502363361020606,
      "grad_norm": 0.4858568608760834,
      "learning_rate": 5.954626621042466e-06,
      "loss": 0.4134,
      "step": 29423
    },
    {
      "epoch": 1.350282226607315,
      "grad_norm": 0.48121047019958496,
      "learning_rate": 5.95438594232607e-06,
      "loss": 0.4098,
      "step": 29424
    },
    {
      "epoch": 1.3503281171125694,
      "grad_norm": 0.4450834393501282,
      "learning_rate": 5.954145261314678e-06,
      "loss": 0.3636,
      "step": 29425
    },
    {
      "epoch": 1.350374007617824,
      "grad_norm": 0.49920767545700073,
      "learning_rate": 5.953904578008867e-06,
      "loss": 0.3959,
      "step": 29426
    },
    {
      "epoch": 1.3504198981230784,
      "grad_norm": 0.4793969988822937,
      "learning_rate": 5.953663892409215e-06,
      "loss": 0.4305,
      "step": 29427
    },
    {
      "epoch": 1.3504657886283327,
      "grad_norm": 0.4778003692626953,
      "learning_rate": 5.953423204516304e-06,
      "loss": 0.3457,
      "step": 29428
    },
    {
      "epoch": 1.3505116791335872,
      "grad_norm": 0.4819778800010681,
      "learning_rate": 5.953182514330712e-06,
      "loss": 0.3322,
      "step": 29429
    },
    {
      "epoch": 1.3505575696388417,
      "grad_norm": 0.512312114238739,
      "learning_rate": 5.952941821853016e-06,
      "loss": 0.4285,
      "step": 29430
    },
    {
      "epoch": 1.3506034601440962,
      "grad_norm": 0.454806923866272,
      "learning_rate": 5.9527011270837954e-06,
      "loss": 0.3441,
      "step": 29431
    },
    {
      "epoch": 1.3506493506493507,
      "grad_norm": 0.48443689942359924,
      "learning_rate": 5.952460430023629e-06,
      "loss": 0.3778,
      "step": 29432
    },
    {
      "epoch": 1.3506952411546052,
      "grad_norm": 0.4988158941268921,
      "learning_rate": 5.952219730673095e-06,
      "loss": 0.4215,
      "step": 29433
    },
    {
      "epoch": 1.3507411316598597,
      "grad_norm": 0.47955095767974854,
      "learning_rate": 5.951979029032775e-06,
      "loss": 0.3968,
      "step": 29434
    },
    {
      "epoch": 1.3507870221651141,
      "grad_norm": 0.4715772271156311,
      "learning_rate": 5.951738325103247e-06,
      "loss": 0.3514,
      "step": 29435
    },
    {
      "epoch": 1.3508329126703684,
      "grad_norm": 0.44676560163497925,
      "learning_rate": 5.951497618885086e-06,
      "loss": 0.3419,
      "step": 29436
    },
    {
      "epoch": 1.350878803175623,
      "grad_norm": 0.4500732719898224,
      "learning_rate": 5.951256910378875e-06,
      "loss": 0.3249,
      "step": 29437
    },
    {
      "epoch": 1.3509246936808774,
      "grad_norm": 0.43639427423477173,
      "learning_rate": 5.95101619958519e-06,
      "loss": 0.3406,
      "step": 29438
    },
    {
      "epoch": 1.350970584186132,
      "grad_norm": 0.4610595405101776,
      "learning_rate": 5.9507754865046115e-06,
      "loss": 0.349,
      "step": 29439
    },
    {
      "epoch": 1.3510164746913864,
      "grad_norm": 0.4489709138870239,
      "learning_rate": 5.95053477113772e-06,
      "loss": 0.3178,
      "step": 29440
    },
    {
      "epoch": 1.3510623651966407,
      "grad_norm": 0.4721061885356903,
      "learning_rate": 5.950294053485091e-06,
      "loss": 0.3394,
      "step": 29441
    },
    {
      "epoch": 1.3511082557018952,
      "grad_norm": 0.4574808180332184,
      "learning_rate": 5.9500533335473044e-06,
      "loss": 0.3189,
      "step": 29442
    },
    {
      "epoch": 1.3511541462071497,
      "grad_norm": 0.4466267228126526,
      "learning_rate": 5.949812611324939e-06,
      "loss": 0.3021,
      "step": 29443
    },
    {
      "epoch": 1.3512000367124042,
      "grad_norm": 0.495392769575119,
      "learning_rate": 5.949571886818575e-06,
      "loss": 0.3962,
      "step": 29444
    },
    {
      "epoch": 1.3512459272176587,
      "grad_norm": 0.44819188117980957,
      "learning_rate": 5.949331160028789e-06,
      "loss": 0.3047,
      "step": 29445
    },
    {
      "epoch": 1.3512918177229132,
      "grad_norm": 0.4276049733161926,
      "learning_rate": 5.9490904309561636e-06,
      "loss": 0.2894,
      "step": 29446
    },
    {
      "epoch": 1.3513377082281677,
      "grad_norm": 0.48284846544265747,
      "learning_rate": 5.948849699601274e-06,
      "loss": 0.3784,
      "step": 29447
    },
    {
      "epoch": 1.3513835987334222,
      "grad_norm": 0.41465261578559875,
      "learning_rate": 5.9486089659647e-06,
      "loss": 0.2739,
      "step": 29448
    },
    {
      "epoch": 1.3514294892386765,
      "grad_norm": 0.48336705565452576,
      "learning_rate": 5.94836823004702e-06,
      "loss": 0.3271,
      "step": 29449
    },
    {
      "epoch": 1.351475379743931,
      "grad_norm": 0.4380035996437073,
      "learning_rate": 5.948127491848814e-06,
      "loss": 0.3159,
      "step": 29450
    },
    {
      "epoch": 1.3515212702491854,
      "grad_norm": 0.4483572244644165,
      "learning_rate": 5.947886751370664e-06,
      "loss": 0.309,
      "step": 29451
    },
    {
      "epoch": 1.35156716075444,
      "grad_norm": 0.4572576880455017,
      "learning_rate": 5.947646008613141e-06,
      "loss": 0.3648,
      "step": 29452
    },
    {
      "epoch": 1.3516130512596944,
      "grad_norm": 0.43247950077056885,
      "learning_rate": 5.947405263576831e-06,
      "loss": 0.2872,
      "step": 29453
    },
    {
      "epoch": 1.3516589417649487,
      "grad_norm": 0.4656652510166168,
      "learning_rate": 5.947164516262309e-06,
      "loss": 0.3734,
      "step": 29454
    },
    {
      "epoch": 1.3517048322702032,
      "grad_norm": 0.4362700879573822,
      "learning_rate": 5.946923766670156e-06,
      "loss": 0.3457,
      "step": 29455
    },
    {
      "epoch": 1.3517507227754577,
      "grad_norm": 0.539270281791687,
      "learning_rate": 5.94668301480095e-06,
      "loss": 0.3342,
      "step": 29456
    },
    {
      "epoch": 1.3517966132807122,
      "grad_norm": 0.4721450209617615,
      "learning_rate": 5.9464422606552705e-06,
      "loss": 0.3427,
      "step": 29457
    },
    {
      "epoch": 1.3518425037859667,
      "grad_norm": 0.5854961276054382,
      "learning_rate": 5.946201504233696e-06,
      "loss": 0.4711,
      "step": 29458
    },
    {
      "epoch": 1.3518883942912212,
      "grad_norm": 0.4998513162136078,
      "learning_rate": 5.945960745536805e-06,
      "loss": 0.4166,
      "step": 29459
    },
    {
      "epoch": 1.3519342847964757,
      "grad_norm": 0.44695258140563965,
      "learning_rate": 5.945719984565177e-06,
      "loss": 0.3235,
      "step": 29460
    },
    {
      "epoch": 1.3519801753017302,
      "grad_norm": 0.4576665163040161,
      "learning_rate": 5.9454792213193915e-06,
      "loss": 0.3471,
      "step": 29461
    },
    {
      "epoch": 1.3520260658069845,
      "grad_norm": 0.48912128806114197,
      "learning_rate": 5.945238455800027e-06,
      "loss": 0.3753,
      "step": 29462
    },
    {
      "epoch": 1.352071956312239,
      "grad_norm": 0.5700365304946899,
      "learning_rate": 5.944997688007662e-06,
      "loss": 0.3534,
      "step": 29463
    },
    {
      "epoch": 1.3521178468174935,
      "grad_norm": 0.4673402011394501,
      "learning_rate": 5.944756917942875e-06,
      "loss": 0.3513,
      "step": 29464
    },
    {
      "epoch": 1.352163737322748,
      "grad_norm": 0.44554173946380615,
      "learning_rate": 5.944516145606247e-06,
      "loss": 0.2995,
      "step": 29465
    },
    {
      "epoch": 1.3522096278280025,
      "grad_norm": 0.4523986876010895,
      "learning_rate": 5.944275370998355e-06,
      "loss": 0.3462,
      "step": 29466
    },
    {
      "epoch": 1.3522555183332567,
      "grad_norm": 0.44060468673706055,
      "learning_rate": 5.94403459411978e-06,
      "loss": 0.2846,
      "step": 29467
    },
    {
      "epoch": 1.3523014088385112,
      "grad_norm": 0.459499716758728,
      "learning_rate": 5.943793814971098e-06,
      "loss": 0.3061,
      "step": 29468
    },
    {
      "epoch": 1.3523472993437657,
      "grad_norm": 0.5119299292564392,
      "learning_rate": 5.943553033552891e-06,
      "loss": 0.4215,
      "step": 29469
    },
    {
      "epoch": 1.3523931898490202,
      "grad_norm": 0.5038403272628784,
      "learning_rate": 5.943312249865736e-06,
      "loss": 0.4109,
      "step": 29470
    },
    {
      "epoch": 1.3524390803542747,
      "grad_norm": 0.48753291368484497,
      "learning_rate": 5.9430714639102125e-06,
      "loss": 0.3683,
      "step": 29471
    },
    {
      "epoch": 1.3524849708595292,
      "grad_norm": 0.4845438599586487,
      "learning_rate": 5.9428306756869e-06,
      "loss": 0.3898,
      "step": 29472
    },
    {
      "epoch": 1.3525308613647837,
      "grad_norm": 0.4997783303260803,
      "learning_rate": 5.942589885196378e-06,
      "loss": 0.4343,
      "step": 29473
    },
    {
      "epoch": 1.352576751870038,
      "grad_norm": 0.5134733319282532,
      "learning_rate": 5.942349092439224e-06,
      "loss": 0.4508,
      "step": 29474
    },
    {
      "epoch": 1.3526226423752925,
      "grad_norm": 0.4712375998497009,
      "learning_rate": 5.942108297416018e-06,
      "loss": 0.3401,
      "step": 29475
    },
    {
      "epoch": 1.352668532880547,
      "grad_norm": 0.48030731081962585,
      "learning_rate": 5.9418675001273405e-06,
      "loss": 0.3452,
      "step": 29476
    },
    {
      "epoch": 1.3527144233858015,
      "grad_norm": 0.5054637789726257,
      "learning_rate": 5.941626700573767e-06,
      "loss": 0.3961,
      "step": 29477
    },
    {
      "epoch": 1.352760313891056,
      "grad_norm": 0.5077686309814453,
      "learning_rate": 5.941385898755879e-06,
      "loss": 0.3498,
      "step": 29478
    },
    {
      "epoch": 1.3528062043963103,
      "grad_norm": 0.45082950592041016,
      "learning_rate": 5.941145094674257e-06,
      "loss": 0.3531,
      "step": 29479
    },
    {
      "epoch": 1.3528520949015648,
      "grad_norm": 0.4849626421928406,
      "learning_rate": 5.940904288329474e-06,
      "loss": 0.3496,
      "step": 29480
    },
    {
      "epoch": 1.3528979854068193,
      "grad_norm": 0.4552357494831085,
      "learning_rate": 5.940663479722115e-06,
      "loss": 0.3087,
      "step": 29481
    },
    {
      "epoch": 1.3529438759120738,
      "grad_norm": 0.49145278334617615,
      "learning_rate": 5.940422668852759e-06,
      "loss": 0.3996,
      "step": 29482
    },
    {
      "epoch": 1.3529897664173283,
      "grad_norm": 0.4758847653865814,
      "learning_rate": 5.940181855721981e-06,
      "loss": 0.3608,
      "step": 29483
    },
    {
      "epoch": 1.3530356569225828,
      "grad_norm": 0.5474574565887451,
      "learning_rate": 5.9399410403303635e-06,
      "loss": 0.4008,
      "step": 29484
    },
    {
      "epoch": 1.3530815474278373,
      "grad_norm": 0.5052220821380615,
      "learning_rate": 5.939700222678484e-06,
      "loss": 0.2988,
      "step": 29485
    },
    {
      "epoch": 1.3531274379330918,
      "grad_norm": 0.42659851908683777,
      "learning_rate": 5.939459402766922e-06,
      "loss": 0.2652,
      "step": 29486
    },
    {
      "epoch": 1.353173328438346,
      "grad_norm": 0.4713885486125946,
      "learning_rate": 5.939218580596255e-06,
      "loss": 0.3551,
      "step": 29487
    },
    {
      "epoch": 1.3532192189436005,
      "grad_norm": 0.46077069640159607,
      "learning_rate": 5.938977756167067e-06,
      "loss": 0.3247,
      "step": 29488
    },
    {
      "epoch": 1.353265109448855,
      "grad_norm": 0.4927977919578552,
      "learning_rate": 5.938736929479932e-06,
      "loss": 0.3669,
      "step": 29489
    },
    {
      "epoch": 1.3533109999541095,
      "grad_norm": 0.5094456672668457,
      "learning_rate": 5.938496100535432e-06,
      "loss": 0.4342,
      "step": 29490
    },
    {
      "epoch": 1.353356890459364,
      "grad_norm": 0.5088447332382202,
      "learning_rate": 5.938255269334144e-06,
      "loss": 0.4305,
      "step": 29491
    },
    {
      "epoch": 1.3534027809646183,
      "grad_norm": 0.4570750594139099,
      "learning_rate": 5.938014435876649e-06,
      "loss": 0.2986,
      "step": 29492
    },
    {
      "epoch": 1.3534486714698728,
      "grad_norm": 0.47599947452545166,
      "learning_rate": 5.937773600163526e-06,
      "loss": 0.3742,
      "step": 29493
    },
    {
      "epoch": 1.3534945619751273,
      "grad_norm": 0.47962313890457153,
      "learning_rate": 5.937532762195351e-06,
      "loss": 0.3622,
      "step": 29494
    },
    {
      "epoch": 1.3535404524803818,
      "grad_norm": 0.49812382459640503,
      "learning_rate": 5.937291921972708e-06,
      "loss": 0.3844,
      "step": 29495
    },
    {
      "epoch": 1.3535863429856363,
      "grad_norm": 0.4808807969093323,
      "learning_rate": 5.937051079496172e-06,
      "loss": 0.3754,
      "step": 29496
    },
    {
      "epoch": 1.3536322334908908,
      "grad_norm": 0.5198098421096802,
      "learning_rate": 5.936810234766324e-06,
      "loss": 0.3717,
      "step": 29497
    },
    {
      "epoch": 1.3536781239961453,
      "grad_norm": 0.47564035654067993,
      "learning_rate": 5.9365693877837436e-06,
      "loss": 0.4212,
      "step": 29498
    },
    {
      "epoch": 1.3537240145013998,
      "grad_norm": 0.49062061309814453,
      "learning_rate": 5.93632853854901e-06,
      "loss": 0.3774,
      "step": 29499
    },
    {
      "epoch": 1.353769905006654,
      "grad_norm": 0.453936368227005,
      "learning_rate": 5.936087687062701e-06,
      "loss": 0.3483,
      "step": 29500
    },
    {
      "epoch": 1.3538157955119086,
      "grad_norm": 0.49795588850975037,
      "learning_rate": 5.935846833325397e-06,
      "loss": 0.4334,
      "step": 29501
    },
    {
      "epoch": 1.353861686017163,
      "grad_norm": 0.4456612467765808,
      "learning_rate": 5.935605977337678e-06,
      "loss": 0.2917,
      "step": 29502
    },
    {
      "epoch": 1.3539075765224176,
      "grad_norm": 0.4585441052913666,
      "learning_rate": 5.9353651191001195e-06,
      "loss": 0.3053,
      "step": 29503
    },
    {
      "epoch": 1.353953467027672,
      "grad_norm": 0.49219778180122375,
      "learning_rate": 5.9351242586133025e-06,
      "loss": 0.4228,
      "step": 29504
    },
    {
      "epoch": 1.3539993575329263,
      "grad_norm": 0.42750632762908936,
      "learning_rate": 5.93488339587781e-06,
      "loss": 0.3157,
      "step": 29505
    },
    {
      "epoch": 1.3540452480381808,
      "grad_norm": 0.49682432413101196,
      "learning_rate": 5.934642530894216e-06,
      "loss": 0.3929,
      "step": 29506
    },
    {
      "epoch": 1.3540911385434353,
      "grad_norm": 0.4230867922306061,
      "learning_rate": 5.934401663663103e-06,
      "loss": 0.3016,
      "step": 29507
    },
    {
      "epoch": 1.3541370290486898,
      "grad_norm": 0.46849682927131653,
      "learning_rate": 5.934160794185047e-06,
      "loss": 0.3334,
      "step": 29508
    },
    {
      "epoch": 1.3541829195539443,
      "grad_norm": 0.48456355929374695,
      "learning_rate": 5.933919922460629e-06,
      "loss": 0.3727,
      "step": 29509
    },
    {
      "epoch": 1.3542288100591988,
      "grad_norm": 0.44948720932006836,
      "learning_rate": 5.93367904849043e-06,
      "loss": 0.3208,
      "step": 29510
    },
    {
      "epoch": 1.3542747005644533,
      "grad_norm": 0.5145674347877502,
      "learning_rate": 5.933438172275028e-06,
      "loss": 0.4301,
      "step": 29511
    },
    {
      "epoch": 1.3543205910697078,
      "grad_norm": 0.5029955506324768,
      "learning_rate": 5.933197293815e-06,
      "loss": 0.426,
      "step": 29512
    },
    {
      "epoch": 1.354366481574962,
      "grad_norm": 0.5060005187988281,
      "learning_rate": 5.932956413110928e-06,
      "loss": 0.3944,
      "step": 29513
    },
    {
      "epoch": 1.3544123720802166,
      "grad_norm": 0.47717639803886414,
      "learning_rate": 5.9327155301633895e-06,
      "loss": 0.3455,
      "step": 29514
    },
    {
      "epoch": 1.354458262585471,
      "grad_norm": 0.4631054103374481,
      "learning_rate": 5.932474644972965e-06,
      "loss": 0.3643,
      "step": 29515
    },
    {
      "epoch": 1.3545041530907256,
      "grad_norm": 0.47242268919944763,
      "learning_rate": 5.9322337575402335e-06,
      "loss": 0.3257,
      "step": 29516
    },
    {
      "epoch": 1.3545500435959799,
      "grad_norm": 0.43661004304885864,
      "learning_rate": 5.931992867865774e-06,
      "loss": 0.3186,
      "step": 29517
    },
    {
      "epoch": 1.3545959341012344,
      "grad_norm": 0.45031511783599854,
      "learning_rate": 5.9317519759501665e-06,
      "loss": 0.3043,
      "step": 29518
    },
    {
      "epoch": 1.3546418246064889,
      "grad_norm": 0.4612464904785156,
      "learning_rate": 5.931511081793988e-06,
      "loss": 0.3574,
      "step": 29519
    },
    {
      "epoch": 1.3546877151117434,
      "grad_norm": 0.4424362778663635,
      "learning_rate": 5.931270185397821e-06,
      "loss": 0.3251,
      "step": 29520
    },
    {
      "epoch": 1.3547336056169978,
      "grad_norm": 0.48220860958099365,
      "learning_rate": 5.931029286762242e-06,
      "loss": 0.3487,
      "step": 29521
    },
    {
      "epoch": 1.3547794961222523,
      "grad_norm": 0.4638567864894867,
      "learning_rate": 5.9307883858878326e-06,
      "loss": 0.3239,
      "step": 29522
    },
    {
      "epoch": 1.3548253866275068,
      "grad_norm": 0.4949064254760742,
      "learning_rate": 5.9305474827751684e-06,
      "loss": 0.3514,
      "step": 29523
    },
    {
      "epoch": 1.3548712771327613,
      "grad_norm": 0.4567302167415619,
      "learning_rate": 5.930306577424834e-06,
      "loss": 0.3228,
      "step": 29524
    },
    {
      "epoch": 1.3549171676380156,
      "grad_norm": 0.6611633896827698,
      "learning_rate": 5.930065669837404e-06,
      "loss": 0.2822,
      "step": 29525
    },
    {
      "epoch": 1.3549630581432701,
      "grad_norm": 0.5450873970985413,
      "learning_rate": 5.92982476001346e-06,
      "loss": 0.46,
      "step": 29526
    },
    {
      "epoch": 1.3550089486485246,
      "grad_norm": 0.5070503354072571,
      "learning_rate": 5.929583847953581e-06,
      "loss": 0.412,
      "step": 29527
    },
    {
      "epoch": 1.3550548391537791,
      "grad_norm": 0.4776357114315033,
      "learning_rate": 5.929342933658346e-06,
      "loss": 0.3829,
      "step": 29528
    },
    {
      "epoch": 1.3551007296590336,
      "grad_norm": 0.4866112172603607,
      "learning_rate": 5.929102017128335e-06,
      "loss": 0.3712,
      "step": 29529
    },
    {
      "epoch": 1.3551466201642879,
      "grad_norm": 0.47708427906036377,
      "learning_rate": 5.928861098364127e-06,
      "loss": 0.3641,
      "step": 29530
    },
    {
      "epoch": 1.3551925106695424,
      "grad_norm": 0.4511614739894867,
      "learning_rate": 5.928620177366301e-06,
      "loss": 0.3198,
      "step": 29531
    },
    {
      "epoch": 1.3552384011747969,
      "grad_norm": 0.4654328525066376,
      "learning_rate": 5.9283792541354365e-06,
      "loss": 0.3308,
      "step": 29532
    },
    {
      "epoch": 1.3552842916800514,
      "grad_norm": 0.478515088558197,
      "learning_rate": 5.928138328672114e-06,
      "loss": 0.3496,
      "step": 29533
    },
    {
      "epoch": 1.3553301821853059,
      "grad_norm": 0.48931899666786194,
      "learning_rate": 5.92789740097691e-06,
      "loss": 0.3723,
      "step": 29534
    },
    {
      "epoch": 1.3553760726905604,
      "grad_norm": 0.4498971402645111,
      "learning_rate": 5.927656471050406e-06,
      "loss": 0.3397,
      "step": 29535
    },
    {
      "epoch": 1.3554219631958149,
      "grad_norm": 0.47696053981781006,
      "learning_rate": 5.927415538893182e-06,
      "loss": 0.4142,
      "step": 29536
    },
    {
      "epoch": 1.3554678537010694,
      "grad_norm": 0.4853246808052063,
      "learning_rate": 5.927174604505815e-06,
      "loss": 0.4221,
      "step": 29537
    },
    {
      "epoch": 1.3555137442063236,
      "grad_norm": 0.4427390992641449,
      "learning_rate": 5.926933667888887e-06,
      "loss": 0.3133,
      "step": 29538
    },
    {
      "epoch": 1.3555596347115781,
      "grad_norm": 0.5313991904258728,
      "learning_rate": 5.926692729042976e-06,
      "loss": 0.5289,
      "step": 29539
    },
    {
      "epoch": 1.3556055252168326,
      "grad_norm": 0.5311051607131958,
      "learning_rate": 5.92645178796866e-06,
      "loss": 0.4189,
      "step": 29540
    },
    {
      "epoch": 1.3556514157220871,
      "grad_norm": 0.4435186982154846,
      "learning_rate": 5.926210844666521e-06,
      "loss": 0.2881,
      "step": 29541
    },
    {
      "epoch": 1.3556973062273416,
      "grad_norm": 0.43815866112709045,
      "learning_rate": 5.925969899137136e-06,
      "loss": 0.3191,
      "step": 29542
    },
    {
      "epoch": 1.355743196732596,
      "grad_norm": 0.4523143470287323,
      "learning_rate": 5.925728951381088e-06,
      "loss": 0.3452,
      "step": 29543
    },
    {
      "epoch": 1.3557890872378504,
      "grad_norm": 0.4735391438007355,
      "learning_rate": 5.925488001398953e-06,
      "loss": 0.3454,
      "step": 29544
    },
    {
      "epoch": 1.355834977743105,
      "grad_norm": 0.5045285224914551,
      "learning_rate": 5.92524704919131e-06,
      "loss": 0.3619,
      "step": 29545
    },
    {
      "epoch": 1.3558808682483594,
      "grad_norm": 0.5694239735603333,
      "learning_rate": 5.92500609475874e-06,
      "loss": 0.3457,
      "step": 29546
    },
    {
      "epoch": 1.355926758753614,
      "grad_norm": 0.49713921546936035,
      "learning_rate": 5.924765138101824e-06,
      "loss": 0.3771,
      "step": 29547
    },
    {
      "epoch": 1.3559726492588684,
      "grad_norm": 0.44682055711746216,
      "learning_rate": 5.924524179221139e-06,
      "loss": 0.3086,
      "step": 29548
    },
    {
      "epoch": 1.356018539764123,
      "grad_norm": 0.47251468896865845,
      "learning_rate": 5.924283218117265e-06,
      "loss": 0.289,
      "step": 29549
    },
    {
      "epoch": 1.3560644302693774,
      "grad_norm": 0.530010461807251,
      "learning_rate": 5.924042254790783e-06,
      "loss": 0.3939,
      "step": 29550
    },
    {
      "epoch": 1.3561103207746317,
      "grad_norm": 0.46580496430397034,
      "learning_rate": 5.923801289242268e-06,
      "loss": 0.3048,
      "step": 29551
    },
    {
      "epoch": 1.3561562112798862,
      "grad_norm": 0.456935316324234,
      "learning_rate": 5.923560321472304e-06,
      "loss": 0.3395,
      "step": 29552
    },
    {
      "epoch": 1.3562021017851407,
      "grad_norm": 0.47044870257377625,
      "learning_rate": 5.923319351481469e-06,
      "loss": 0.3686,
      "step": 29553
    },
    {
      "epoch": 1.3562479922903952,
      "grad_norm": 0.43659117817878723,
      "learning_rate": 5.923078379270343e-06,
      "loss": 0.2964,
      "step": 29554
    },
    {
      "epoch": 1.3562938827956494,
      "grad_norm": 0.4753674566745758,
      "learning_rate": 5.9228374048395045e-06,
      "loss": 0.3931,
      "step": 29555
    },
    {
      "epoch": 1.356339773300904,
      "grad_norm": 0.48780912160873413,
      "learning_rate": 5.922596428189533e-06,
      "loss": 0.3721,
      "step": 29556
    },
    {
      "epoch": 1.3563856638061584,
      "grad_norm": 0.4867079257965088,
      "learning_rate": 5.922355449321008e-06,
      "loss": 0.3509,
      "step": 29557
    },
    {
      "epoch": 1.356431554311413,
      "grad_norm": 0.4791640639305115,
      "learning_rate": 5.92211446823451e-06,
      "loss": 0.3913,
      "step": 29558
    },
    {
      "epoch": 1.3564774448166674,
      "grad_norm": 0.4498705267906189,
      "learning_rate": 5.9218734849306156e-06,
      "loss": 0.3041,
      "step": 29559
    },
    {
      "epoch": 1.356523335321922,
      "grad_norm": 0.46047019958496094,
      "learning_rate": 5.921632499409909e-06,
      "loss": 0.3717,
      "step": 29560
    },
    {
      "epoch": 1.3565692258271764,
      "grad_norm": 0.454988569021225,
      "learning_rate": 5.921391511672967e-06,
      "loss": 0.3474,
      "step": 29561
    },
    {
      "epoch": 1.356615116332431,
      "grad_norm": 0.4630346894264221,
      "learning_rate": 5.921150521720368e-06,
      "loss": 0.3679,
      "step": 29562
    },
    {
      "epoch": 1.3566610068376852,
      "grad_norm": 0.438899427652359,
      "learning_rate": 5.920909529552692e-06,
      "loss": 0.2669,
      "step": 29563
    },
    {
      "epoch": 1.3567068973429397,
      "grad_norm": 0.44538411498069763,
      "learning_rate": 5.920668535170522e-06,
      "loss": 0.3031,
      "step": 29564
    },
    {
      "epoch": 1.3567527878481942,
      "grad_norm": 0.49406489729881287,
      "learning_rate": 5.920427538574432e-06,
      "loss": 0.3619,
      "step": 29565
    },
    {
      "epoch": 1.3567986783534487,
      "grad_norm": 0.4674498736858368,
      "learning_rate": 5.920186539765006e-06,
      "loss": 0.3454,
      "step": 29566
    },
    {
      "epoch": 1.3568445688587032,
      "grad_norm": 0.45554405450820923,
      "learning_rate": 5.919945538742822e-06,
      "loss": 0.3053,
      "step": 29567
    },
    {
      "epoch": 1.3568904593639575,
      "grad_norm": 0.4519006907939911,
      "learning_rate": 5.919704535508457e-06,
      "loss": 0.3474,
      "step": 29568
    },
    {
      "epoch": 1.356936349869212,
      "grad_norm": 0.42680203914642334,
      "learning_rate": 5.9194635300624935e-06,
      "loss": 0.309,
      "step": 29569
    },
    {
      "epoch": 1.3569822403744665,
      "grad_norm": 0.46383142471313477,
      "learning_rate": 5.919222522405512e-06,
      "loss": 0.4276,
      "step": 29570
    },
    {
      "epoch": 1.357028130879721,
      "grad_norm": 0.4380016028881073,
      "learning_rate": 5.91898151253809e-06,
      "loss": 0.3086,
      "step": 29571
    },
    {
      "epoch": 1.3570740213849755,
      "grad_norm": 0.5110563039779663,
      "learning_rate": 5.918740500460808e-06,
      "loss": 0.4029,
      "step": 29572
    },
    {
      "epoch": 1.35711991189023,
      "grad_norm": 0.4521673321723938,
      "learning_rate": 5.918499486174244e-06,
      "loss": 0.3422,
      "step": 29573
    },
    {
      "epoch": 1.3571658023954845,
      "grad_norm": 0.5332672595977783,
      "learning_rate": 5.918258469678978e-06,
      "loss": 0.4414,
      "step": 29574
    },
    {
      "epoch": 1.357211692900739,
      "grad_norm": 0.45664867758750916,
      "learning_rate": 5.918017450975592e-06,
      "loss": 0.3519,
      "step": 29575
    },
    {
      "epoch": 1.3572575834059932,
      "grad_norm": 0.49565690755844116,
      "learning_rate": 5.917776430064663e-06,
      "loss": 0.3823,
      "step": 29576
    },
    {
      "epoch": 1.3573034739112477,
      "grad_norm": 0.47229939699172974,
      "learning_rate": 5.917535406946771e-06,
      "loss": 0.4284,
      "step": 29577
    },
    {
      "epoch": 1.3573493644165022,
      "grad_norm": 0.46170151233673096,
      "learning_rate": 5.917294381622496e-06,
      "loss": 0.3335,
      "step": 29578
    },
    {
      "epoch": 1.3573952549217567,
      "grad_norm": 0.44030386209487915,
      "learning_rate": 5.917053354092417e-06,
      "loss": 0.3291,
      "step": 29579
    },
    {
      "epoch": 1.3574411454270112,
      "grad_norm": 0.47472819685935974,
      "learning_rate": 5.916812324357114e-06,
      "loss": 0.3307,
      "step": 29580
    },
    {
      "epoch": 1.3574870359322655,
      "grad_norm": 0.44266200065612793,
      "learning_rate": 5.916571292417168e-06,
      "loss": 0.2984,
      "step": 29581
    },
    {
      "epoch": 1.35753292643752,
      "grad_norm": 0.4729927182197571,
      "learning_rate": 5.916330258273158e-06,
      "loss": 0.3533,
      "step": 29582
    },
    {
      "epoch": 1.3575788169427745,
      "grad_norm": 0.4902630150318146,
      "learning_rate": 5.91608922192566e-06,
      "loss": 0.385,
      "step": 29583
    },
    {
      "epoch": 1.357624707448029,
      "grad_norm": 0.443960577249527,
      "learning_rate": 5.915848183375259e-06,
      "loss": 0.3035,
      "step": 29584
    },
    {
      "epoch": 1.3576705979532835,
      "grad_norm": 0.5070707201957703,
      "learning_rate": 5.9156071426225315e-06,
      "loss": 0.402,
      "step": 29585
    },
    {
      "epoch": 1.357716488458538,
      "grad_norm": 0.4142516553401947,
      "learning_rate": 5.915366099668058e-06,
      "loss": 0.2721,
      "step": 29586
    },
    {
      "epoch": 1.3577623789637925,
      "grad_norm": 0.4407435357570648,
      "learning_rate": 5.9151250545124185e-06,
      "loss": 0.307,
      "step": 29587
    },
    {
      "epoch": 1.357808269469047,
      "grad_norm": 0.4919806122779846,
      "learning_rate": 5.9148840071561895e-06,
      "loss": 0.4126,
      "step": 29588
    },
    {
      "epoch": 1.3578541599743013,
      "grad_norm": 0.45244964957237244,
      "learning_rate": 5.914642957599957e-06,
      "loss": 0.3053,
      "step": 29589
    },
    {
      "epoch": 1.3579000504795558,
      "grad_norm": 0.4528164863586426,
      "learning_rate": 5.914401905844294e-06,
      "loss": 0.3477,
      "step": 29590
    },
    {
      "epoch": 1.3579459409848103,
      "grad_norm": 0.4564751088619232,
      "learning_rate": 5.914160851889783e-06,
      "loss": 0.3083,
      "step": 29591
    },
    {
      "epoch": 1.3579918314900647,
      "grad_norm": 0.4766659736633301,
      "learning_rate": 5.913919795737004e-06,
      "loss": 0.3948,
      "step": 29592
    },
    {
      "epoch": 1.3580377219953192,
      "grad_norm": 0.4663122892379761,
      "learning_rate": 5.913678737386538e-06,
      "loss": 0.3889,
      "step": 29593
    },
    {
      "epoch": 1.3580836125005735,
      "grad_norm": 0.4822918772697449,
      "learning_rate": 5.9134376768389614e-06,
      "loss": 0.3392,
      "step": 29594
    },
    {
      "epoch": 1.358129503005828,
      "grad_norm": 0.4525180459022522,
      "learning_rate": 5.913196614094856e-06,
      "loss": 0.3315,
      "step": 29595
    },
    {
      "epoch": 1.3581753935110825,
      "grad_norm": 0.46415045857429504,
      "learning_rate": 5.912955549154801e-06,
      "loss": 0.352,
      "step": 29596
    },
    {
      "epoch": 1.358221284016337,
      "grad_norm": 0.444278359413147,
      "learning_rate": 5.912714482019376e-06,
      "loss": 0.3063,
      "step": 29597
    },
    {
      "epoch": 1.3582671745215915,
      "grad_norm": 0.4749598503112793,
      "learning_rate": 5.912473412689161e-06,
      "loss": 0.3797,
      "step": 29598
    },
    {
      "epoch": 1.358313065026846,
      "grad_norm": 0.4539699852466583,
      "learning_rate": 5.912232341164735e-06,
      "loss": 0.2921,
      "step": 29599
    },
    {
      "epoch": 1.3583589555321005,
      "grad_norm": 0.5553392171859741,
      "learning_rate": 5.911991267446677e-06,
      "loss": 0.4689,
      "step": 29600
    },
    {
      "epoch": 1.358404846037355,
      "grad_norm": 0.708074152469635,
      "learning_rate": 5.911750191535569e-06,
      "loss": 0.311,
      "step": 29601
    },
    {
      "epoch": 1.3584507365426093,
      "grad_norm": 0.43390798568725586,
      "learning_rate": 5.911509113431989e-06,
      "loss": 0.2958,
      "step": 29602
    },
    {
      "epoch": 1.3584966270478638,
      "grad_norm": 0.496028333902359,
      "learning_rate": 5.911268033136519e-06,
      "loss": 0.3847,
      "step": 29603
    },
    {
      "epoch": 1.3585425175531183,
      "grad_norm": 0.5099213123321533,
      "learning_rate": 5.911026950649736e-06,
      "loss": 0.468,
      "step": 29604
    },
    {
      "epoch": 1.3585884080583728,
      "grad_norm": 0.47072237730026245,
      "learning_rate": 5.9107858659722194e-06,
      "loss": 0.3506,
      "step": 29605
    },
    {
      "epoch": 1.358634298563627,
      "grad_norm": 0.512626588344574,
      "learning_rate": 5.910544779104552e-06,
      "loss": 0.3802,
      "step": 29606
    },
    {
      "epoch": 1.3586801890688815,
      "grad_norm": 0.4535597264766693,
      "learning_rate": 5.910303690047311e-06,
      "loss": 0.3169,
      "step": 29607
    },
    {
      "epoch": 1.358726079574136,
      "grad_norm": 0.5390753149986267,
      "learning_rate": 5.910062598801078e-06,
      "loss": 0.4112,
      "step": 29608
    },
    {
      "epoch": 1.3587719700793905,
      "grad_norm": 0.488546222448349,
      "learning_rate": 5.909821505366431e-06,
      "loss": 0.3922,
      "step": 29609
    },
    {
      "epoch": 1.358817860584645,
      "grad_norm": 0.43779709935188293,
      "learning_rate": 5.909580409743949e-06,
      "loss": 0.3188,
      "step": 29610
    },
    {
      "epoch": 1.3588637510898995,
      "grad_norm": 0.4747777581214905,
      "learning_rate": 5.909339311934213e-06,
      "loss": 0.3962,
      "step": 29611
    },
    {
      "epoch": 1.358909641595154,
      "grad_norm": 0.4844985902309418,
      "learning_rate": 5.909098211937806e-06,
      "loss": 0.364,
      "step": 29612
    },
    {
      "epoch": 1.3589555321004085,
      "grad_norm": 0.4262472093105316,
      "learning_rate": 5.9088571097553014e-06,
      "loss": 0.3346,
      "step": 29613
    },
    {
      "epoch": 1.3590014226056628,
      "grad_norm": 0.4293331801891327,
      "learning_rate": 5.908616005387285e-06,
      "loss": 0.2985,
      "step": 29614
    },
    {
      "epoch": 1.3590473131109173,
      "grad_norm": 0.4899218678474426,
      "learning_rate": 5.908374898834334e-06,
      "loss": 0.4435,
      "step": 29615
    },
    {
      "epoch": 1.3590932036161718,
      "grad_norm": 0.7056658267974854,
      "learning_rate": 5.908133790097026e-06,
      "loss": 0.3994,
      "step": 29616
    },
    {
      "epoch": 1.3591390941214263,
      "grad_norm": 0.4775899052619934,
      "learning_rate": 5.907892679175943e-06,
      "loss": 0.3449,
      "step": 29617
    },
    {
      "epoch": 1.3591849846266808,
      "grad_norm": 0.460114449262619,
      "learning_rate": 5.907651566071665e-06,
      "loss": 0.3747,
      "step": 29618
    },
    {
      "epoch": 1.359230875131935,
      "grad_norm": 0.4267895221710205,
      "learning_rate": 5.907410450784773e-06,
      "loss": 0.2802,
      "step": 29619
    },
    {
      "epoch": 1.3592767656371896,
      "grad_norm": 0.4801914095878601,
      "learning_rate": 5.907169333315843e-06,
      "loss": 0.3716,
      "step": 29620
    },
    {
      "epoch": 1.359322656142444,
      "grad_norm": 0.43569689989089966,
      "learning_rate": 5.906928213665459e-06,
      "loss": 0.293,
      "step": 29621
    },
    {
      "epoch": 1.3593685466476986,
      "grad_norm": 0.5279867649078369,
      "learning_rate": 5.906687091834197e-06,
      "loss": 0.465,
      "step": 29622
    },
    {
      "epoch": 1.359414437152953,
      "grad_norm": 0.4872879683971405,
      "learning_rate": 5.90644596782264e-06,
      "loss": 0.3753,
      "step": 29623
    },
    {
      "epoch": 1.3594603276582076,
      "grad_norm": 0.47108331322669983,
      "learning_rate": 5.906204841631365e-06,
      "loss": 0.3701,
      "step": 29624
    },
    {
      "epoch": 1.359506218163462,
      "grad_norm": 0.4660903513431549,
      "learning_rate": 5.905963713260956e-06,
      "loss": 0.3214,
      "step": 29625
    },
    {
      "epoch": 1.3595521086687166,
      "grad_norm": 0.4406264126300812,
      "learning_rate": 5.905722582711989e-06,
      "loss": 0.2727,
      "step": 29626
    },
    {
      "epoch": 1.3595979991739708,
      "grad_norm": 0.5067930817604065,
      "learning_rate": 5.905481449985045e-06,
      "loss": 0.3687,
      "step": 29627
    },
    {
      "epoch": 1.3596438896792253,
      "grad_norm": 0.46533942222595215,
      "learning_rate": 5.905240315080702e-06,
      "loss": 0.3562,
      "step": 29628
    },
    {
      "epoch": 1.3596897801844798,
      "grad_norm": 0.4441923201084137,
      "learning_rate": 5.904999177999544e-06,
      "loss": 0.3271,
      "step": 29629
    },
    {
      "epoch": 1.3597356706897343,
      "grad_norm": 0.48101547360420227,
      "learning_rate": 5.9047580387421485e-06,
      "loss": 0.3724,
      "step": 29630
    },
    {
      "epoch": 1.3597815611949888,
      "grad_norm": 0.4518243074417114,
      "learning_rate": 5.904516897309095e-06,
      "loss": 0.363,
      "step": 29631
    },
    {
      "epoch": 1.359827451700243,
      "grad_norm": 0.460448294878006,
      "learning_rate": 5.904275753700965e-06,
      "loss": 0.358,
      "step": 29632
    },
    {
      "epoch": 1.3598733422054976,
      "grad_norm": 0.45445820689201355,
      "learning_rate": 5.904034607918335e-06,
      "loss": 0.318,
      "step": 29633
    },
    {
      "epoch": 1.359919232710752,
      "grad_norm": 0.5103985667228699,
      "learning_rate": 5.9037934599617875e-06,
      "loss": 0.4108,
      "step": 29634
    },
    {
      "epoch": 1.3599651232160066,
      "grad_norm": 0.4557051658630371,
      "learning_rate": 5.903552309831903e-06,
      "loss": 0.3411,
      "step": 29635
    },
    {
      "epoch": 1.360011013721261,
      "grad_norm": 0.4576733708381653,
      "learning_rate": 5.9033111575292605e-06,
      "loss": 0.3583,
      "step": 29636
    },
    {
      "epoch": 1.3600569042265156,
      "grad_norm": 0.45510566234588623,
      "learning_rate": 5.90307000305444e-06,
      "loss": 0.3898,
      "step": 29637
    },
    {
      "epoch": 1.36010279473177,
      "grad_norm": 0.518474280834198,
      "learning_rate": 5.9028288464080194e-06,
      "loss": 0.4034,
      "step": 29638
    },
    {
      "epoch": 1.3601486852370246,
      "grad_norm": 0.4567142724990845,
      "learning_rate": 5.902587687590581e-06,
      "loss": 0.3546,
      "step": 29639
    },
    {
      "epoch": 1.3601945757422789,
      "grad_norm": 0.4524565637111664,
      "learning_rate": 5.902346526602704e-06,
      "loss": 0.3621,
      "step": 29640
    },
    {
      "epoch": 1.3602404662475334,
      "grad_norm": 0.46177491545677185,
      "learning_rate": 5.90210536344497e-06,
      "loss": 0.3546,
      "step": 29641
    },
    {
      "epoch": 1.3602863567527879,
      "grad_norm": 0.48046839237213135,
      "learning_rate": 5.901864198117956e-06,
      "loss": 0.382,
      "step": 29642
    },
    {
      "epoch": 1.3603322472580424,
      "grad_norm": 0.4983239769935608,
      "learning_rate": 5.901623030622243e-06,
      "loss": 0.4085,
      "step": 29643
    },
    {
      "epoch": 1.3603781377632966,
      "grad_norm": 0.45995792746543884,
      "learning_rate": 5.901381860958412e-06,
      "loss": 0.3405,
      "step": 29644
    },
    {
      "epoch": 1.3604240282685511,
      "grad_norm": 0.5116739273071289,
      "learning_rate": 5.90114068912704e-06,
      "loss": 0.4462,
      "step": 29645
    },
    {
      "epoch": 1.3604699187738056,
      "grad_norm": 0.4350147247314453,
      "learning_rate": 5.9008995151287115e-06,
      "loss": 0.3149,
      "step": 29646
    },
    {
      "epoch": 1.3605158092790601,
      "grad_norm": 0.46195584535598755,
      "learning_rate": 5.9006583389640035e-06,
      "loss": 0.3256,
      "step": 29647
    },
    {
      "epoch": 1.3605616997843146,
      "grad_norm": 0.4836227595806122,
      "learning_rate": 5.900417160633496e-06,
      "loss": 0.4146,
      "step": 29648
    },
    {
      "epoch": 1.3606075902895691,
      "grad_norm": 0.46304672956466675,
      "learning_rate": 5.90017598013777e-06,
      "loss": 0.3899,
      "step": 29649
    },
    {
      "epoch": 1.3606534807948236,
      "grad_norm": 0.4790802299976349,
      "learning_rate": 5.899934797477405e-06,
      "loss": 0.3299,
      "step": 29650
    },
    {
      "epoch": 1.3606993713000781,
      "grad_norm": 0.4604453444480896,
      "learning_rate": 5.899693612652979e-06,
      "loss": 0.363,
      "step": 29651
    },
    {
      "epoch": 1.3607452618053324,
      "grad_norm": 0.47053855657577515,
      "learning_rate": 5.899452425665076e-06,
      "loss": 0.3866,
      "step": 29652
    },
    {
      "epoch": 1.360791152310587,
      "grad_norm": 0.44615164399147034,
      "learning_rate": 5.8992112365142725e-06,
      "loss": 0.3292,
      "step": 29653
    },
    {
      "epoch": 1.3608370428158414,
      "grad_norm": 0.45640429854393005,
      "learning_rate": 5.898970045201151e-06,
      "loss": 0.3932,
      "step": 29654
    },
    {
      "epoch": 1.3608829333210959,
      "grad_norm": 0.45141783356666565,
      "learning_rate": 5.898728851726289e-06,
      "loss": 0.3954,
      "step": 29655
    },
    {
      "epoch": 1.3609288238263504,
      "grad_norm": 0.4146559536457062,
      "learning_rate": 5.898487656090268e-06,
      "loss": 0.3107,
      "step": 29656
    },
    {
      "epoch": 1.3609747143316047,
      "grad_norm": 0.4813310205936432,
      "learning_rate": 5.898246458293668e-06,
      "loss": 0.4233,
      "step": 29657
    },
    {
      "epoch": 1.3610206048368592,
      "grad_norm": 0.4341275990009308,
      "learning_rate": 5.8980052583370695e-06,
      "loss": 0.2935,
      "step": 29658
    },
    {
      "epoch": 1.3610664953421137,
      "grad_norm": 0.45984718203544617,
      "learning_rate": 5.89776405622105e-06,
      "loss": 0.3363,
      "step": 29659
    },
    {
      "epoch": 1.3611123858473682,
      "grad_norm": 0.44823405146598816,
      "learning_rate": 5.897522851946193e-06,
      "loss": 0.3323,
      "step": 29660
    },
    {
      "epoch": 1.3611582763526227,
      "grad_norm": 0.4772595465183258,
      "learning_rate": 5.8972816455130755e-06,
      "loss": 0.4058,
      "step": 29661
    },
    {
      "epoch": 1.3612041668578772,
      "grad_norm": 0.5200003385543823,
      "learning_rate": 5.897040436922279e-06,
      "loss": 0.3876,
      "step": 29662
    },
    {
      "epoch": 1.3612500573631316,
      "grad_norm": 0.45474350452423096,
      "learning_rate": 5.896799226174386e-06,
      "loss": 0.329,
      "step": 29663
    },
    {
      "epoch": 1.3612959478683861,
      "grad_norm": 0.4122833013534546,
      "learning_rate": 5.896558013269971e-06,
      "loss": 0.2982,
      "step": 29664
    },
    {
      "epoch": 1.3613418383736404,
      "grad_norm": 0.47715505957603455,
      "learning_rate": 5.896316798209616e-06,
      "loss": 0.3595,
      "step": 29665
    },
    {
      "epoch": 1.361387728878895,
      "grad_norm": 0.4455989599227905,
      "learning_rate": 5.896075580993904e-06,
      "loss": 0.3235,
      "step": 29666
    },
    {
      "epoch": 1.3614336193841494,
      "grad_norm": 0.44369298219680786,
      "learning_rate": 5.895834361623413e-06,
      "loss": 0.3215,
      "step": 29667
    },
    {
      "epoch": 1.361479509889404,
      "grad_norm": 0.4578772783279419,
      "learning_rate": 5.895593140098723e-06,
      "loss": 0.361,
      "step": 29668
    },
    {
      "epoch": 1.3615254003946584,
      "grad_norm": 0.46442487835884094,
      "learning_rate": 5.895351916420415e-06,
      "loss": 0.3505,
      "step": 29669
    },
    {
      "epoch": 1.3615712908999127,
      "grad_norm": 0.4709234833717346,
      "learning_rate": 5.895110690589065e-06,
      "loss": 0.4239,
      "step": 29670
    },
    {
      "epoch": 1.3616171814051672,
      "grad_norm": 0.46745041012763977,
      "learning_rate": 5.89486946260526e-06,
      "loss": 0.3482,
      "step": 29671
    },
    {
      "epoch": 1.3616630719104217,
      "grad_norm": 0.4968077540397644,
      "learning_rate": 5.894628232469573e-06,
      "loss": 0.3262,
      "step": 29672
    },
    {
      "epoch": 1.3617089624156762,
      "grad_norm": 0.4447762370109558,
      "learning_rate": 5.89438700018259e-06,
      "loss": 0.3057,
      "step": 29673
    },
    {
      "epoch": 1.3617548529209307,
      "grad_norm": 0.4621894657611847,
      "learning_rate": 5.894145765744888e-06,
      "loss": 0.343,
      "step": 29674
    },
    {
      "epoch": 1.3618007434261852,
      "grad_norm": 0.5195560455322266,
      "learning_rate": 5.893904529157048e-06,
      "loss": 0.4563,
      "step": 29675
    },
    {
      "epoch": 1.3618466339314397,
      "grad_norm": 0.4150986075401306,
      "learning_rate": 5.893663290419648e-06,
      "loss": 0.2835,
      "step": 29676
    },
    {
      "epoch": 1.3618925244366942,
      "grad_norm": 0.46529433131217957,
      "learning_rate": 5.893422049533272e-06,
      "loss": 0.3803,
      "step": 29677
    },
    {
      "epoch": 1.3619384149419484,
      "grad_norm": 0.47696444392204285,
      "learning_rate": 5.8931808064984965e-06,
      "loss": 0.3699,
      "step": 29678
    },
    {
      "epoch": 1.361984305447203,
      "grad_norm": 0.48936840891838074,
      "learning_rate": 5.892939561315903e-06,
      "loss": 0.4168,
      "step": 29679
    },
    {
      "epoch": 1.3620301959524574,
      "grad_norm": 0.6252825260162354,
      "learning_rate": 5.892698313986073e-06,
      "loss": 0.4106,
      "step": 29680
    },
    {
      "epoch": 1.362076086457712,
      "grad_norm": 0.4199279248714447,
      "learning_rate": 5.8924570645095835e-06,
      "loss": 0.2695,
      "step": 29681
    },
    {
      "epoch": 1.3621219769629664,
      "grad_norm": 0.5053243637084961,
      "learning_rate": 5.892215812887017e-06,
      "loss": 0.4676,
      "step": 29682
    },
    {
      "epoch": 1.3621678674682207,
      "grad_norm": 0.46395283937454224,
      "learning_rate": 5.891974559118953e-06,
      "loss": 0.3683,
      "step": 29683
    },
    {
      "epoch": 1.3622137579734752,
      "grad_norm": 0.43673619627952576,
      "learning_rate": 5.891733303205973e-06,
      "loss": 0.3242,
      "step": 29684
    },
    {
      "epoch": 1.3622596484787297,
      "grad_norm": 0.491452157497406,
      "learning_rate": 5.891492045148656e-06,
      "loss": 0.3822,
      "step": 29685
    },
    {
      "epoch": 1.3623055389839842,
      "grad_norm": 0.4200022518634796,
      "learning_rate": 5.891250784947581e-06,
      "loss": 0.2983,
      "step": 29686
    },
    {
      "epoch": 1.3623514294892387,
      "grad_norm": 0.4812476933002472,
      "learning_rate": 5.8910095226033285e-06,
      "loss": 0.3579,
      "step": 29687
    },
    {
      "epoch": 1.3623973199944932,
      "grad_norm": 0.5228948593139648,
      "learning_rate": 5.890768258116481e-06,
      "loss": 0.4348,
      "step": 29688
    },
    {
      "epoch": 1.3624432104997477,
      "grad_norm": 0.49590861797332764,
      "learning_rate": 5.890526991487615e-06,
      "loss": 0.3451,
      "step": 29689
    },
    {
      "epoch": 1.3624891010050022,
      "grad_norm": 0.47714853286743164,
      "learning_rate": 5.890285722717314e-06,
      "loss": 0.4093,
      "step": 29690
    },
    {
      "epoch": 1.3625349915102565,
      "grad_norm": 0.4981129467487335,
      "learning_rate": 5.890044451806156e-06,
      "loss": 0.4124,
      "step": 29691
    },
    {
      "epoch": 1.362580882015511,
      "grad_norm": 0.48129430413246155,
      "learning_rate": 5.889803178754723e-06,
      "loss": 0.3713,
      "step": 29692
    },
    {
      "epoch": 1.3626267725207655,
      "grad_norm": 0.4568685293197632,
      "learning_rate": 5.889561903563592e-06,
      "loss": 0.3547,
      "step": 29693
    },
    {
      "epoch": 1.36267266302602,
      "grad_norm": 0.44387710094451904,
      "learning_rate": 5.889320626233348e-06,
      "loss": 0.3066,
      "step": 29694
    },
    {
      "epoch": 1.3627185535312742,
      "grad_norm": 0.48319244384765625,
      "learning_rate": 5.889079346764567e-06,
      "loss": 0.4501,
      "step": 29695
    },
    {
      "epoch": 1.3627644440365287,
      "grad_norm": 0.5225914120674133,
      "learning_rate": 5.888838065157832e-06,
      "loss": 0.4526,
      "step": 29696
    },
    {
      "epoch": 1.3628103345417832,
      "grad_norm": 0.490865558385849,
      "learning_rate": 5.888596781413721e-06,
      "loss": 0.344,
      "step": 29697
    },
    {
      "epoch": 1.3628562250470377,
      "grad_norm": 0.45328599214553833,
      "learning_rate": 5.8883554955328145e-06,
      "loss": 0.3836,
      "step": 29698
    },
    {
      "epoch": 1.3629021155522922,
      "grad_norm": 0.4989314675331116,
      "learning_rate": 5.888114207515694e-06,
      "loss": 0.4061,
      "step": 29699
    },
    {
      "epoch": 1.3629480060575467,
      "grad_norm": 0.47671636939048767,
      "learning_rate": 5.88787291736294e-06,
      "loss": 0.3781,
      "step": 29700
    },
    {
      "epoch": 1.3629938965628012,
      "grad_norm": 0.46445339918136597,
      "learning_rate": 5.887631625075132e-06,
      "loss": 0.3732,
      "step": 29701
    },
    {
      "epoch": 1.3630397870680557,
      "grad_norm": 0.5187601447105408,
      "learning_rate": 5.887390330652849e-06,
      "loss": 0.4223,
      "step": 29702
    },
    {
      "epoch": 1.36308567757331,
      "grad_norm": 0.46434056758880615,
      "learning_rate": 5.887149034096673e-06,
      "loss": 0.3205,
      "step": 29703
    },
    {
      "epoch": 1.3631315680785645,
      "grad_norm": 0.45827803015708923,
      "learning_rate": 5.886907735407182e-06,
      "loss": 0.3129,
      "step": 29704
    },
    {
      "epoch": 1.363177458583819,
      "grad_norm": 0.531427264213562,
      "learning_rate": 5.88666643458496e-06,
      "loss": 0.4066,
      "step": 29705
    },
    {
      "epoch": 1.3632233490890735,
      "grad_norm": 0.4810008108615875,
      "learning_rate": 5.8864251316305846e-06,
      "loss": 0.3667,
      "step": 29706
    },
    {
      "epoch": 1.363269239594328,
      "grad_norm": 0.4745352864265442,
      "learning_rate": 5.8861838265446355e-06,
      "loss": 0.414,
      "step": 29707
    },
    {
      "epoch": 1.3633151300995823,
      "grad_norm": 0.443050742149353,
      "learning_rate": 5.8859425193276955e-06,
      "loss": 0.3185,
      "step": 29708
    },
    {
      "epoch": 1.3633610206048368,
      "grad_norm": 0.48126769065856934,
      "learning_rate": 5.885701209980342e-06,
      "loss": 0.3842,
      "step": 29709
    },
    {
      "epoch": 1.3634069111100913,
      "grad_norm": 0.44451019167900085,
      "learning_rate": 5.885459898503156e-06,
      "loss": 0.3105,
      "step": 29710
    },
    {
      "epoch": 1.3634528016153458,
      "grad_norm": 0.5091874599456787,
      "learning_rate": 5.88521858489672e-06,
      "loss": 0.3324,
      "step": 29711
    },
    {
      "epoch": 1.3634986921206003,
      "grad_norm": 0.44315865635871887,
      "learning_rate": 5.884977269161614e-06,
      "loss": 0.2847,
      "step": 29712
    },
    {
      "epoch": 1.3635445826258548,
      "grad_norm": 0.4910751283168793,
      "learning_rate": 5.884735951298413e-06,
      "loss": 0.4062,
      "step": 29713
    },
    {
      "epoch": 1.3635904731311093,
      "grad_norm": 0.5021066665649414,
      "learning_rate": 5.884494631307705e-06,
      "loss": 0.4075,
      "step": 29714
    },
    {
      "epoch": 1.3636363636363638,
      "grad_norm": 0.4840486943721771,
      "learning_rate": 5.8842533091900645e-06,
      "loss": 0.3505,
      "step": 29715
    },
    {
      "epoch": 1.363682254141618,
      "grad_norm": 0.4852827787399292,
      "learning_rate": 5.884011984946073e-06,
      "loss": 0.3441,
      "step": 29716
    },
    {
      "epoch": 1.3637281446468725,
      "grad_norm": 0.4716375470161438,
      "learning_rate": 5.883770658576313e-06,
      "loss": 0.323,
      "step": 29717
    },
    {
      "epoch": 1.363774035152127,
      "grad_norm": 0.44765496253967285,
      "learning_rate": 5.883529330081364e-06,
      "loss": 0.3534,
      "step": 29718
    },
    {
      "epoch": 1.3638199256573815,
      "grad_norm": 0.44826802611351013,
      "learning_rate": 5.883287999461806e-06,
      "loss": 0.3351,
      "step": 29719
    },
    {
      "epoch": 1.363865816162636,
      "grad_norm": 0.45793578028678894,
      "learning_rate": 5.883046666718219e-06,
      "loss": 0.3574,
      "step": 29720
    },
    {
      "epoch": 1.3639117066678903,
      "grad_norm": 0.4719219207763672,
      "learning_rate": 5.882805331851182e-06,
      "loss": 0.4078,
      "step": 29721
    },
    {
      "epoch": 1.3639575971731448,
      "grad_norm": 0.46208590269088745,
      "learning_rate": 5.882563994861278e-06,
      "loss": 0.3109,
      "step": 29722
    },
    {
      "epoch": 1.3640034876783993,
      "grad_norm": 0.4673776626586914,
      "learning_rate": 5.882322655749086e-06,
      "loss": 0.3209,
      "step": 29723
    },
    {
      "epoch": 1.3640493781836538,
      "grad_norm": 0.4862824082374573,
      "learning_rate": 5.882081314515186e-06,
      "loss": 0.3479,
      "step": 29724
    },
    {
      "epoch": 1.3640952686889083,
      "grad_norm": 0.5807982087135315,
      "learning_rate": 5.881839971160159e-06,
      "loss": 0.4647,
      "step": 29725
    },
    {
      "epoch": 1.3641411591941628,
      "grad_norm": 0.46980398893356323,
      "learning_rate": 5.881598625684586e-06,
      "loss": 0.363,
      "step": 29726
    },
    {
      "epoch": 1.3641870496994173,
      "grad_norm": 0.4606328010559082,
      "learning_rate": 5.881357278089045e-06,
      "loss": 0.3701,
      "step": 29727
    },
    {
      "epoch": 1.3642329402046718,
      "grad_norm": 0.4825633764266968,
      "learning_rate": 5.881115928374119e-06,
      "loss": 0.3775,
      "step": 29728
    },
    {
      "epoch": 1.364278830709926,
      "grad_norm": 0.5654458403587341,
      "learning_rate": 5.8808745765403875e-06,
      "loss": 0.4849,
      "step": 29729
    },
    {
      "epoch": 1.3643247212151806,
      "grad_norm": 0.4712643027305603,
      "learning_rate": 5.88063322258843e-06,
      "loss": 0.3819,
      "step": 29730
    },
    {
      "epoch": 1.364370611720435,
      "grad_norm": 0.4865855276584625,
      "learning_rate": 5.880391866518828e-06,
      "loss": 0.3717,
      "step": 29731
    },
    {
      "epoch": 1.3644165022256896,
      "grad_norm": 0.45772209763526917,
      "learning_rate": 5.880150508332161e-06,
      "loss": 0.3569,
      "step": 29732
    },
    {
      "epoch": 1.3644623927309438,
      "grad_norm": 0.46473848819732666,
      "learning_rate": 5.879909148029011e-06,
      "loss": 0.3002,
      "step": 29733
    },
    {
      "epoch": 1.3645082832361983,
      "grad_norm": 0.478754460811615,
      "learning_rate": 5.879667785609956e-06,
      "loss": 0.323,
      "step": 29734
    },
    {
      "epoch": 1.3645541737414528,
      "grad_norm": 0.48935818672180176,
      "learning_rate": 5.8794264210755765e-06,
      "loss": 0.365,
      "step": 29735
    },
    {
      "epoch": 1.3646000642467073,
      "grad_norm": 0.509634792804718,
      "learning_rate": 5.879185054426457e-06,
      "loss": 0.3662,
      "step": 29736
    },
    {
      "epoch": 1.3646459547519618,
      "grad_norm": 0.5099175572395325,
      "learning_rate": 5.878943685663172e-06,
      "loss": 0.4368,
      "step": 29737
    },
    {
      "epoch": 1.3646918452572163,
      "grad_norm": 0.48028722405433655,
      "learning_rate": 5.8787023147863065e-06,
      "loss": 0.3536,
      "step": 29738
    },
    {
      "epoch": 1.3647377357624708,
      "grad_norm": 0.6040860414505005,
      "learning_rate": 5.878460941796439e-06,
      "loss": 0.4118,
      "step": 29739
    },
    {
      "epoch": 1.3647836262677253,
      "grad_norm": 0.47918400168418884,
      "learning_rate": 5.87821956669415e-06,
      "loss": 0.4021,
      "step": 29740
    },
    {
      "epoch": 1.3648295167729796,
      "grad_norm": 0.45733773708343506,
      "learning_rate": 5.87797818948002e-06,
      "loss": 0.3205,
      "step": 29741
    },
    {
      "epoch": 1.364875407278234,
      "grad_norm": 0.47505882382392883,
      "learning_rate": 5.877736810154631e-06,
      "loss": 0.3556,
      "step": 29742
    },
    {
      "epoch": 1.3649212977834886,
      "grad_norm": 0.47317835688591003,
      "learning_rate": 5.877495428718559e-06,
      "loss": 0.3331,
      "step": 29743
    },
    {
      "epoch": 1.364967188288743,
      "grad_norm": 0.4912436008453369,
      "learning_rate": 5.87725404517239e-06,
      "loss": 0.4306,
      "step": 29744
    },
    {
      "epoch": 1.3650130787939976,
      "grad_norm": 0.5024107694625854,
      "learning_rate": 5.877012659516701e-06,
      "loss": 0.397,
      "step": 29745
    },
    {
      "epoch": 1.3650589692992519,
      "grad_norm": 0.49367591738700867,
      "learning_rate": 5.876771271752073e-06,
      "loss": 0.356,
      "step": 29746
    },
    {
      "epoch": 1.3651048598045064,
      "grad_norm": 0.45909377932548523,
      "learning_rate": 5.876529881879087e-06,
      "loss": 0.3503,
      "step": 29747
    },
    {
      "epoch": 1.3651507503097609,
      "grad_norm": 0.49692976474761963,
      "learning_rate": 5.876288489898324e-06,
      "loss": 0.3965,
      "step": 29748
    },
    {
      "epoch": 1.3651966408150153,
      "grad_norm": 0.4792816638946533,
      "learning_rate": 5.876047095810362e-06,
      "loss": 0.3616,
      "step": 29749
    },
    {
      "epoch": 1.3652425313202698,
      "grad_norm": 0.5081164240837097,
      "learning_rate": 5.8758056996157855e-06,
      "loss": 0.3832,
      "step": 29750
    },
    {
      "epoch": 1.3652884218255243,
      "grad_norm": 0.46000030636787415,
      "learning_rate": 5.875564301315172e-06,
      "loss": 0.3368,
      "step": 29751
    },
    {
      "epoch": 1.3653343123307788,
      "grad_norm": 0.4886743128299713,
      "learning_rate": 5.875322900909102e-06,
      "loss": 0.3676,
      "step": 29752
    },
    {
      "epoch": 1.3653802028360333,
      "grad_norm": 0.41099125146865845,
      "learning_rate": 5.8750814983981575e-06,
      "loss": 0.2865,
      "step": 29753
    },
    {
      "epoch": 1.3654260933412876,
      "grad_norm": 0.48294857144355774,
      "learning_rate": 5.8748400937829165e-06,
      "loss": 0.2894,
      "step": 29754
    },
    {
      "epoch": 1.3654719838465421,
      "grad_norm": 0.45478665828704834,
      "learning_rate": 5.874598687063964e-06,
      "loss": 0.3403,
      "step": 29755
    },
    {
      "epoch": 1.3655178743517966,
      "grad_norm": 0.477312296628952,
      "learning_rate": 5.874357278241876e-06,
      "loss": 0.4335,
      "step": 29756
    },
    {
      "epoch": 1.365563764857051,
      "grad_norm": 0.4832588732242584,
      "learning_rate": 5.874115867317234e-06,
      "loss": 0.3504,
      "step": 29757
    },
    {
      "epoch": 1.3656096553623056,
      "grad_norm": 0.5276463031768799,
      "learning_rate": 5.873874454290619e-06,
      "loss": 0.3717,
      "step": 29758
    },
    {
      "epoch": 1.3656555458675599,
      "grad_norm": 0.46185624599456787,
      "learning_rate": 5.873633039162614e-06,
      "loss": 0.3658,
      "step": 29759
    },
    {
      "epoch": 1.3657014363728144,
      "grad_norm": 0.4658348858356476,
      "learning_rate": 5.873391621933795e-06,
      "loss": 0.354,
      "step": 29760
    },
    {
      "epoch": 1.3657473268780689,
      "grad_norm": 0.46875959634780884,
      "learning_rate": 5.873150202604746e-06,
      "loss": 0.3708,
      "step": 29761
    },
    {
      "epoch": 1.3657932173833234,
      "grad_norm": 0.44907090067863464,
      "learning_rate": 5.872908781176046e-06,
      "loss": 0.3418,
      "step": 29762
    },
    {
      "epoch": 1.3658391078885779,
      "grad_norm": 0.43456974625587463,
      "learning_rate": 5.8726673576482764e-06,
      "loss": 0.3284,
      "step": 29763
    },
    {
      "epoch": 1.3658849983938324,
      "grad_norm": 0.470121830701828,
      "learning_rate": 5.8724259320220164e-06,
      "loss": 0.3493,
      "step": 29764
    },
    {
      "epoch": 1.3659308888990869,
      "grad_norm": 0.4517599046230316,
      "learning_rate": 5.872184504297849e-06,
      "loss": 0.3339,
      "step": 29765
    },
    {
      "epoch": 1.3659767794043414,
      "grad_norm": 0.5214707851409912,
      "learning_rate": 5.871943074476352e-06,
      "loss": 0.4562,
      "step": 29766
    },
    {
      "epoch": 1.3660226699095956,
      "grad_norm": 0.4566161036491394,
      "learning_rate": 5.8717016425581065e-06,
      "loss": 0.3468,
      "step": 29767
    },
    {
      "epoch": 1.3660685604148501,
      "grad_norm": 0.49475225806236267,
      "learning_rate": 5.8714602085436945e-06,
      "loss": 0.4085,
      "step": 29768
    },
    {
      "epoch": 1.3661144509201046,
      "grad_norm": 0.47037360072135925,
      "learning_rate": 5.871218772433694e-06,
      "loss": 0.3535,
      "step": 29769
    },
    {
      "epoch": 1.3661603414253591,
      "grad_norm": 0.5339037179946899,
      "learning_rate": 5.87097733422869e-06,
      "loss": 0.4859,
      "step": 29770
    },
    {
      "epoch": 1.3662062319306136,
      "grad_norm": 0.45391660928726196,
      "learning_rate": 5.870735893929258e-06,
      "loss": 0.3311,
      "step": 29771
    },
    {
      "epoch": 1.366252122435868,
      "grad_norm": 0.4685741066932678,
      "learning_rate": 5.870494451535983e-06,
      "loss": 0.3349,
      "step": 29772
    },
    {
      "epoch": 1.3662980129411224,
      "grad_norm": 0.4549446105957031,
      "learning_rate": 5.870253007049444e-06,
      "loss": 0.3371,
      "step": 29773
    },
    {
      "epoch": 1.366343903446377,
      "grad_norm": 0.48662102222442627,
      "learning_rate": 5.870011560470219e-06,
      "loss": 0.3696,
      "step": 29774
    },
    {
      "epoch": 1.3663897939516314,
      "grad_norm": 0.5081114768981934,
      "learning_rate": 5.869770111798892e-06,
      "loss": 0.4235,
      "step": 29775
    },
    {
      "epoch": 1.366435684456886,
      "grad_norm": 0.4568287432193756,
      "learning_rate": 5.869528661036041e-06,
      "loss": 0.3325,
      "step": 29776
    },
    {
      "epoch": 1.3664815749621404,
      "grad_norm": 0.4743413031101227,
      "learning_rate": 5.86928720818225e-06,
      "loss": 0.3693,
      "step": 29777
    },
    {
      "epoch": 1.366527465467395,
      "grad_norm": 0.43583944439888,
      "learning_rate": 5.869045753238096e-06,
      "loss": 0.3159,
      "step": 29778
    },
    {
      "epoch": 1.3665733559726494,
      "grad_norm": 0.5159749388694763,
      "learning_rate": 5.868804296204163e-06,
      "loss": 0.3859,
      "step": 29779
    },
    {
      "epoch": 1.3666192464779037,
      "grad_norm": 0.48418113589286804,
      "learning_rate": 5.868562837081028e-06,
      "loss": 0.3567,
      "step": 29780
    },
    {
      "epoch": 1.3666651369831582,
      "grad_norm": 0.48067766427993774,
      "learning_rate": 5.868321375869274e-06,
      "loss": 0.3796,
      "step": 29781
    },
    {
      "epoch": 1.3667110274884127,
      "grad_norm": 0.47048845887184143,
      "learning_rate": 5.868079912569482e-06,
      "loss": 0.3388,
      "step": 29782
    },
    {
      "epoch": 1.3667569179936672,
      "grad_norm": 0.44692379236221313,
      "learning_rate": 5.867838447182232e-06,
      "loss": 0.3078,
      "step": 29783
    },
    {
      "epoch": 1.3668028084989214,
      "grad_norm": 0.40212884545326233,
      "learning_rate": 5.867596979708102e-06,
      "loss": 0.2538,
      "step": 29784
    },
    {
      "epoch": 1.366848699004176,
      "grad_norm": 0.45300817489624023,
      "learning_rate": 5.867355510147677e-06,
      "loss": 0.2937,
      "step": 29785
    },
    {
      "epoch": 1.3668945895094304,
      "grad_norm": 0.47108668088912964,
      "learning_rate": 5.867114038501535e-06,
      "loss": 0.3774,
      "step": 29786
    },
    {
      "epoch": 1.366940480014685,
      "grad_norm": 0.44795137643814087,
      "learning_rate": 5.866872564770258e-06,
      "loss": 0.307,
      "step": 29787
    },
    {
      "epoch": 1.3669863705199394,
      "grad_norm": 0.4745928645133972,
      "learning_rate": 5.866631088954427e-06,
      "loss": 0.3765,
      "step": 29788
    },
    {
      "epoch": 1.367032261025194,
      "grad_norm": 0.4997924268245697,
      "learning_rate": 5.86638961105462e-06,
      "loss": 0.4638,
      "step": 29789
    },
    {
      "epoch": 1.3670781515304484,
      "grad_norm": 0.4842694103717804,
      "learning_rate": 5.866148131071421e-06,
      "loss": 0.4274,
      "step": 29790
    },
    {
      "epoch": 1.367124042035703,
      "grad_norm": 0.4479700028896332,
      "learning_rate": 5.865906649005408e-06,
      "loss": 0.3866,
      "step": 29791
    },
    {
      "epoch": 1.3671699325409572,
      "grad_norm": 0.4480702877044678,
      "learning_rate": 5.8656651648571625e-06,
      "loss": 0.3303,
      "step": 29792
    },
    {
      "epoch": 1.3672158230462117,
      "grad_norm": 0.46636658906936646,
      "learning_rate": 5.865423678627266e-06,
      "loss": 0.3187,
      "step": 29793
    },
    {
      "epoch": 1.3672617135514662,
      "grad_norm": 0.5072179436683655,
      "learning_rate": 5.8651821903163e-06,
      "loss": 0.2762,
      "step": 29794
    },
    {
      "epoch": 1.3673076040567207,
      "grad_norm": 0.4815240502357483,
      "learning_rate": 5.864940699924841e-06,
      "loss": 0.3339,
      "step": 29795
    },
    {
      "epoch": 1.3673534945619752,
      "grad_norm": 0.45585671067237854,
      "learning_rate": 5.864699207453476e-06,
      "loss": 0.392,
      "step": 29796
    },
    {
      "epoch": 1.3673993850672295,
      "grad_norm": 0.50896817445755,
      "learning_rate": 5.864457712902779e-06,
      "loss": 0.4446,
      "step": 29797
    },
    {
      "epoch": 1.367445275572484,
      "grad_norm": 0.45998165011405945,
      "learning_rate": 5.864216216273335e-06,
      "loss": 0.3547,
      "step": 29798
    },
    {
      "epoch": 1.3674911660777385,
      "grad_norm": 0.46132102608680725,
      "learning_rate": 5.863974717565727e-06,
      "loss": 0.3513,
      "step": 29799
    },
    {
      "epoch": 1.367537056582993,
      "grad_norm": 0.4766498804092407,
      "learning_rate": 5.863733216780529e-06,
      "loss": 0.3686,
      "step": 29800
    },
    {
      "epoch": 1.3675829470882475,
      "grad_norm": 0.4720718264579773,
      "learning_rate": 5.863491713918327e-06,
      "loss": 0.3643,
      "step": 29801
    },
    {
      "epoch": 1.367628837593502,
      "grad_norm": 0.5083315372467041,
      "learning_rate": 5.863250208979697e-06,
      "loss": 0.4043,
      "step": 29802
    },
    {
      "epoch": 1.3676747280987565,
      "grad_norm": 0.5806950330734253,
      "learning_rate": 5.863008701965225e-06,
      "loss": 0.3719,
      "step": 29803
    },
    {
      "epoch": 1.367720618604011,
      "grad_norm": 0.4860505163669586,
      "learning_rate": 5.8627671928754894e-06,
      "loss": 0.3574,
      "step": 29804
    },
    {
      "epoch": 1.3677665091092652,
      "grad_norm": 0.49794402718544006,
      "learning_rate": 5.862525681711071e-06,
      "loss": 0.4519,
      "step": 29805
    },
    {
      "epoch": 1.3678123996145197,
      "grad_norm": 0.4721835255622864,
      "learning_rate": 5.8622841684725494e-06,
      "loss": 0.384,
      "step": 29806
    },
    {
      "epoch": 1.3678582901197742,
      "grad_norm": 0.44490474462509155,
      "learning_rate": 5.862042653160509e-06,
      "loss": 0.3403,
      "step": 29807
    },
    {
      "epoch": 1.3679041806250287,
      "grad_norm": 0.4682622253894806,
      "learning_rate": 5.8618011357755255e-06,
      "loss": 0.3894,
      "step": 29808
    },
    {
      "epoch": 1.3679500711302832,
      "grad_norm": 0.47407492995262146,
      "learning_rate": 5.861559616318182e-06,
      "loss": 0.3896,
      "step": 29809
    },
    {
      "epoch": 1.3679959616355375,
      "grad_norm": 0.4773600995540619,
      "learning_rate": 5.861318094789064e-06,
      "loss": 0.3764,
      "step": 29810
    },
    {
      "epoch": 1.368041852140792,
      "grad_norm": 0.47023552656173706,
      "learning_rate": 5.861076571188743e-06,
      "loss": 0.3695,
      "step": 29811
    },
    {
      "epoch": 1.3680877426460465,
      "grad_norm": 0.45501163601875305,
      "learning_rate": 5.8608350455178055e-06,
      "loss": 0.3199,
      "step": 29812
    },
    {
      "epoch": 1.368133633151301,
      "grad_norm": 0.4484744071960449,
      "learning_rate": 5.860593517776833e-06,
      "loss": 0.3663,
      "step": 29813
    },
    {
      "epoch": 1.3681795236565555,
      "grad_norm": 0.48460522294044495,
      "learning_rate": 5.860351987966403e-06,
      "loss": 0.4191,
      "step": 29814
    },
    {
      "epoch": 1.36822541416181,
      "grad_norm": 0.4628541171550751,
      "learning_rate": 5.860110456087099e-06,
      "loss": 0.3599,
      "step": 29815
    },
    {
      "epoch": 1.3682713046670645,
      "grad_norm": 0.4088127613067627,
      "learning_rate": 5.859868922139501e-06,
      "loss": 0.2847,
      "step": 29816
    },
    {
      "epoch": 1.368317195172319,
      "grad_norm": 0.441959023475647,
      "learning_rate": 5.859627386124189e-06,
      "loss": 0.3252,
      "step": 29817
    },
    {
      "epoch": 1.3683630856775733,
      "grad_norm": 0.46995848417282104,
      "learning_rate": 5.859385848041746e-06,
      "loss": 0.3228,
      "step": 29818
    },
    {
      "epoch": 1.3684089761828278,
      "grad_norm": 0.49330806732177734,
      "learning_rate": 5.859144307892749e-06,
      "loss": 0.407,
      "step": 29819
    },
    {
      "epoch": 1.3684548666880822,
      "grad_norm": 0.4692334532737732,
      "learning_rate": 5.858902765677781e-06,
      "loss": 0.2939,
      "step": 29820
    },
    {
      "epoch": 1.3685007571933367,
      "grad_norm": 0.48432621359825134,
      "learning_rate": 5.8586612213974255e-06,
      "loss": 0.3504,
      "step": 29821
    },
    {
      "epoch": 1.368546647698591,
      "grad_norm": 0.4794270098209381,
      "learning_rate": 5.858419675052259e-06,
      "loss": 0.3324,
      "step": 29822
    },
    {
      "epoch": 1.3685925382038455,
      "grad_norm": 0.436763197183609,
      "learning_rate": 5.858178126642864e-06,
      "loss": 0.2853,
      "step": 29823
    },
    {
      "epoch": 1.3686384287091,
      "grad_norm": 0.46534958481788635,
      "learning_rate": 5.8579365761698225e-06,
      "loss": 0.3231,
      "step": 29824
    },
    {
      "epoch": 1.3686843192143545,
      "grad_norm": 0.4873819649219513,
      "learning_rate": 5.857695023633713e-06,
      "loss": 0.3292,
      "step": 29825
    },
    {
      "epoch": 1.368730209719609,
      "grad_norm": 0.4816637635231018,
      "learning_rate": 5.857453469035118e-06,
      "loss": 0.3632,
      "step": 29826
    },
    {
      "epoch": 1.3687761002248635,
      "grad_norm": 0.5007820129394531,
      "learning_rate": 5.857211912374619e-06,
      "loss": 0.3911,
      "step": 29827
    },
    {
      "epoch": 1.368821990730118,
      "grad_norm": 0.48338058590888977,
      "learning_rate": 5.856970353652794e-06,
      "loss": 0.3919,
      "step": 29828
    },
    {
      "epoch": 1.3688678812353725,
      "grad_norm": 0.4476061463356018,
      "learning_rate": 5.856728792870228e-06,
      "loss": 0.3197,
      "step": 29829
    },
    {
      "epoch": 1.3689137717406268,
      "grad_norm": 0.44476377964019775,
      "learning_rate": 5.856487230027499e-06,
      "loss": 0.3127,
      "step": 29830
    },
    {
      "epoch": 1.3689596622458813,
      "grad_norm": 0.43014779686927795,
      "learning_rate": 5.856245665125186e-06,
      "loss": 0.3216,
      "step": 29831
    },
    {
      "epoch": 1.3690055527511358,
      "grad_norm": 0.4995499849319458,
      "learning_rate": 5.856004098163877e-06,
      "loss": 0.404,
      "step": 29832
    },
    {
      "epoch": 1.3690514432563903,
      "grad_norm": 0.4835204482078552,
      "learning_rate": 5.855762529144145e-06,
      "loss": 0.4245,
      "step": 29833
    },
    {
      "epoch": 1.3690973337616448,
      "grad_norm": 0.4401048421859741,
      "learning_rate": 5.855520958066575e-06,
      "loss": 0.2939,
      "step": 29834
    },
    {
      "epoch": 1.369143224266899,
      "grad_norm": 0.4981495440006256,
      "learning_rate": 5.855279384931748e-06,
      "loss": 0.4341,
      "step": 29835
    },
    {
      "epoch": 1.3691891147721535,
      "grad_norm": 0.46675142645835876,
      "learning_rate": 5.8550378097402425e-06,
      "loss": 0.3068,
      "step": 29836
    },
    {
      "epoch": 1.369235005277408,
      "grad_norm": 0.429499089717865,
      "learning_rate": 5.854796232492642e-06,
      "loss": 0.2771,
      "step": 29837
    },
    {
      "epoch": 1.3692808957826625,
      "grad_norm": 0.46417662501335144,
      "learning_rate": 5.854554653189526e-06,
      "loss": 0.3314,
      "step": 29838
    },
    {
      "epoch": 1.369326786287917,
      "grad_norm": 0.48090627789497375,
      "learning_rate": 5.854313071831475e-06,
      "loss": 0.4058,
      "step": 29839
    },
    {
      "epoch": 1.3693726767931715,
      "grad_norm": 0.46481871604919434,
      "learning_rate": 5.854071488419071e-06,
      "loss": 0.3419,
      "step": 29840
    },
    {
      "epoch": 1.369418567298426,
      "grad_norm": 0.47211334109306335,
      "learning_rate": 5.853829902952895e-06,
      "loss": 0.3705,
      "step": 29841
    },
    {
      "epoch": 1.3694644578036805,
      "grad_norm": 0.4566355049610138,
      "learning_rate": 5.853588315433529e-06,
      "loss": 0.4015,
      "step": 29842
    },
    {
      "epoch": 1.3695103483089348,
      "grad_norm": 0.42488816380500793,
      "learning_rate": 5.853346725861551e-06,
      "loss": 0.2772,
      "step": 29843
    },
    {
      "epoch": 1.3695562388141893,
      "grad_norm": 0.45572802424430847,
      "learning_rate": 5.853105134237544e-06,
      "loss": 0.3232,
      "step": 29844
    },
    {
      "epoch": 1.3696021293194438,
      "grad_norm": 0.46379485726356506,
      "learning_rate": 5.8528635405620875e-06,
      "loss": 0.3482,
      "step": 29845
    },
    {
      "epoch": 1.3696480198246983,
      "grad_norm": 0.42551612854003906,
      "learning_rate": 5.852621944835763e-06,
      "loss": 0.2731,
      "step": 29846
    },
    {
      "epoch": 1.3696939103299528,
      "grad_norm": 0.44556811451911926,
      "learning_rate": 5.852380347059153e-06,
      "loss": 0.3596,
      "step": 29847
    },
    {
      "epoch": 1.369739800835207,
      "grad_norm": 0.49066853523254395,
      "learning_rate": 5.852138747232837e-06,
      "loss": 0.3527,
      "step": 29848
    },
    {
      "epoch": 1.3697856913404616,
      "grad_norm": 0.45625752210617065,
      "learning_rate": 5.851897145357396e-06,
      "loss": 0.3358,
      "step": 29849
    },
    {
      "epoch": 1.369831581845716,
      "grad_norm": 0.5072566270828247,
      "learning_rate": 5.851655541433412e-06,
      "loss": 0.4343,
      "step": 29850
    },
    {
      "epoch": 1.3698774723509706,
      "grad_norm": 0.5686078071594238,
      "learning_rate": 5.851413935461464e-06,
      "loss": 0.39,
      "step": 29851
    },
    {
      "epoch": 1.369923362856225,
      "grad_norm": 0.4519713819026947,
      "learning_rate": 5.851172327442135e-06,
      "loss": 0.3125,
      "step": 29852
    },
    {
      "epoch": 1.3699692533614796,
      "grad_norm": 0.5110547542572021,
      "learning_rate": 5.850930717376004e-06,
      "loss": 0.3631,
      "step": 29853
    },
    {
      "epoch": 1.370015143866734,
      "grad_norm": 0.468814492225647,
      "learning_rate": 5.850689105263655e-06,
      "loss": 0.3577,
      "step": 29854
    },
    {
      "epoch": 1.3700610343719886,
      "grad_norm": 0.4704006612300873,
      "learning_rate": 5.850447491105668e-06,
      "loss": 0.3431,
      "step": 29855
    },
    {
      "epoch": 1.3701069248772428,
      "grad_norm": 0.5195923447608948,
      "learning_rate": 5.850205874902621e-06,
      "loss": 0.4473,
      "step": 29856
    },
    {
      "epoch": 1.3701528153824973,
      "grad_norm": 0.42556577920913696,
      "learning_rate": 5.849964256655096e-06,
      "loss": 0.3121,
      "step": 29857
    },
    {
      "epoch": 1.3701987058877518,
      "grad_norm": 0.4783497452735901,
      "learning_rate": 5.8497226363636774e-06,
      "loss": 0.3121,
      "step": 29858
    },
    {
      "epoch": 1.3702445963930063,
      "grad_norm": 0.48916274309158325,
      "learning_rate": 5.849481014028945e-06,
      "loss": 0.3588,
      "step": 29859
    },
    {
      "epoch": 1.3702904868982608,
      "grad_norm": 0.4638911783695221,
      "learning_rate": 5.849239389651476e-06,
      "loss": 0.3267,
      "step": 29860
    },
    {
      "epoch": 1.370336377403515,
      "grad_norm": 0.4423004388809204,
      "learning_rate": 5.8489977632318565e-06,
      "loss": 0.3103,
      "step": 29861
    },
    {
      "epoch": 1.3703822679087696,
      "grad_norm": 0.5200011134147644,
      "learning_rate": 5.848756134770665e-06,
      "loss": 0.408,
      "step": 29862
    },
    {
      "epoch": 1.370428158414024,
      "grad_norm": 0.4414168894290924,
      "learning_rate": 5.848514504268481e-06,
      "loss": 0.3257,
      "step": 29863
    },
    {
      "epoch": 1.3704740489192786,
      "grad_norm": 0.45072516798973083,
      "learning_rate": 5.848272871725889e-06,
      "loss": 0.3265,
      "step": 29864
    },
    {
      "epoch": 1.370519939424533,
      "grad_norm": 0.47062426805496216,
      "learning_rate": 5.848031237143469e-06,
      "loss": 0.3443,
      "step": 29865
    },
    {
      "epoch": 1.3705658299297876,
      "grad_norm": 0.4317375123500824,
      "learning_rate": 5.847789600521802e-06,
      "loss": 0.3,
      "step": 29866
    },
    {
      "epoch": 1.370611720435042,
      "grad_norm": 0.4936336278915405,
      "learning_rate": 5.8475479618614665e-06,
      "loss": 0.4043,
      "step": 29867
    },
    {
      "epoch": 1.3706576109402966,
      "grad_norm": 0.43565696477890015,
      "learning_rate": 5.8473063211630465e-06,
      "loss": 0.3354,
      "step": 29868
    },
    {
      "epoch": 1.3707035014455509,
      "grad_norm": 0.5324808955192566,
      "learning_rate": 5.847064678427123e-06,
      "loss": 0.2976,
      "step": 29869
    },
    {
      "epoch": 1.3707493919508054,
      "grad_norm": 0.47710222005844116,
      "learning_rate": 5.846823033654275e-06,
      "loss": 0.3659,
      "step": 29870
    },
    {
      "epoch": 1.3707952824560599,
      "grad_norm": 0.4863049387931824,
      "learning_rate": 5.846581386845086e-06,
      "loss": 0.4032,
      "step": 29871
    },
    {
      "epoch": 1.3708411729613144,
      "grad_norm": 0.45570310950279236,
      "learning_rate": 5.846339738000135e-06,
      "loss": 0.3436,
      "step": 29872
    },
    {
      "epoch": 1.3708870634665686,
      "grad_norm": 0.44967561960220337,
      "learning_rate": 5.846098087120004e-06,
      "loss": 0.3041,
      "step": 29873
    },
    {
      "epoch": 1.3709329539718231,
      "grad_norm": 0.46124783158302307,
      "learning_rate": 5.8458564342052746e-06,
      "loss": 0.3523,
      "step": 29874
    },
    {
      "epoch": 1.3709788444770776,
      "grad_norm": 0.5000905990600586,
      "learning_rate": 5.845614779256527e-06,
      "loss": 0.373,
      "step": 29875
    },
    {
      "epoch": 1.3710247349823321,
      "grad_norm": 0.5006259083747864,
      "learning_rate": 5.845373122274345e-06,
      "loss": 0.4085,
      "step": 29876
    },
    {
      "epoch": 1.3710706254875866,
      "grad_norm": 0.4693033993244171,
      "learning_rate": 5.845131463259305e-06,
      "loss": 0.3331,
      "step": 29877
    },
    {
      "epoch": 1.3711165159928411,
      "grad_norm": 0.47834503650665283,
      "learning_rate": 5.844889802211992e-06,
      "loss": 0.336,
      "step": 29878
    },
    {
      "epoch": 1.3711624064980956,
      "grad_norm": 0.48930051922798157,
      "learning_rate": 5.8446481391329835e-06,
      "loss": 0.3078,
      "step": 29879
    },
    {
      "epoch": 1.3712082970033501,
      "grad_norm": 0.46694740653038025,
      "learning_rate": 5.844406474022865e-06,
      "loss": 0.3557,
      "step": 29880
    },
    {
      "epoch": 1.3712541875086044,
      "grad_norm": 0.4398307800292969,
      "learning_rate": 5.844164806882216e-06,
      "loss": 0.3517,
      "step": 29881
    },
    {
      "epoch": 1.371300078013859,
      "grad_norm": 0.4710296392440796,
      "learning_rate": 5.843923137711613e-06,
      "loss": 0.364,
      "step": 29882
    },
    {
      "epoch": 1.3713459685191134,
      "grad_norm": 0.5010955333709717,
      "learning_rate": 5.843681466511646e-06,
      "loss": 0.3987,
      "step": 29883
    },
    {
      "epoch": 1.3713918590243679,
      "grad_norm": 0.4906069040298462,
      "learning_rate": 5.843439793282888e-06,
      "loss": 0.4369,
      "step": 29884
    },
    {
      "epoch": 1.3714377495296224,
      "grad_norm": 0.48778411746025085,
      "learning_rate": 5.843198118025925e-06,
      "loss": 0.3644,
      "step": 29885
    },
    {
      "epoch": 1.3714836400348767,
      "grad_norm": 0.5281990170478821,
      "learning_rate": 5.842956440741336e-06,
      "loss": 0.4228,
      "step": 29886
    },
    {
      "epoch": 1.3715295305401312,
      "grad_norm": 0.46646177768707275,
      "learning_rate": 5.8427147614297045e-06,
      "loss": 0.3872,
      "step": 29887
    },
    {
      "epoch": 1.3715754210453857,
      "grad_norm": 0.5040508508682251,
      "learning_rate": 5.842473080091608e-06,
      "loss": 0.4155,
      "step": 29888
    },
    {
      "epoch": 1.3716213115506402,
      "grad_norm": 0.45189303159713745,
      "learning_rate": 5.84223139672763e-06,
      "loss": 0.3162,
      "step": 29889
    },
    {
      "epoch": 1.3716672020558947,
      "grad_norm": 0.4873679578304291,
      "learning_rate": 5.841989711338351e-06,
      "loss": 0.3594,
      "step": 29890
    },
    {
      "epoch": 1.3717130925611491,
      "grad_norm": 0.4925362467765808,
      "learning_rate": 5.841748023924354e-06,
      "loss": 0.3764,
      "step": 29891
    },
    {
      "epoch": 1.3717589830664036,
      "grad_norm": 0.4965896010398865,
      "learning_rate": 5.8415063344862174e-06,
      "loss": 0.3616,
      "step": 29892
    },
    {
      "epoch": 1.3718048735716581,
      "grad_norm": 0.48598533868789673,
      "learning_rate": 5.841264643024524e-06,
      "loss": 0.3819,
      "step": 29893
    },
    {
      "epoch": 1.3718507640769124,
      "grad_norm": 0.454355925321579,
      "learning_rate": 5.841022949539854e-06,
      "loss": 0.3312,
      "step": 29894
    },
    {
      "epoch": 1.371896654582167,
      "grad_norm": 0.5029056668281555,
      "learning_rate": 5.84078125403279e-06,
      "loss": 0.4331,
      "step": 29895
    },
    {
      "epoch": 1.3719425450874214,
      "grad_norm": 0.4234856069087982,
      "learning_rate": 5.840539556503913e-06,
      "loss": 0.2954,
      "step": 29896
    },
    {
      "epoch": 1.371988435592676,
      "grad_norm": 0.4185897409915924,
      "learning_rate": 5.840297856953803e-06,
      "loss": 0.2829,
      "step": 29897
    },
    {
      "epoch": 1.3720343260979304,
      "grad_norm": 0.47505441308021545,
      "learning_rate": 5.840056155383042e-06,
      "loss": 0.3716,
      "step": 29898
    },
    {
      "epoch": 1.3720802166031847,
      "grad_norm": 0.4962650239467621,
      "learning_rate": 5.839814451792211e-06,
      "loss": 0.3718,
      "step": 29899
    },
    {
      "epoch": 1.3721261071084392,
      "grad_norm": 0.4963543117046356,
      "learning_rate": 5.839572746181891e-06,
      "loss": 0.3934,
      "step": 29900
    },
    {
      "epoch": 1.3721719976136937,
      "grad_norm": 0.4675716459751129,
      "learning_rate": 5.839331038552664e-06,
      "loss": 0.3784,
      "step": 29901
    },
    {
      "epoch": 1.3722178881189482,
      "grad_norm": 0.44355955719947815,
      "learning_rate": 5.83908932890511e-06,
      "loss": 0.3691,
      "step": 29902
    },
    {
      "epoch": 1.3722637786242027,
      "grad_norm": 0.4902159571647644,
      "learning_rate": 5.838847617239812e-06,
      "loss": 0.3804,
      "step": 29903
    },
    {
      "epoch": 1.3723096691294572,
      "grad_norm": 0.46450403332710266,
      "learning_rate": 5.83860590355735e-06,
      "loss": 0.3261,
      "step": 29904
    },
    {
      "epoch": 1.3723555596347117,
      "grad_norm": 0.43805646896362305,
      "learning_rate": 5.838364187858304e-06,
      "loss": 0.3408,
      "step": 29905
    },
    {
      "epoch": 1.3724014501399662,
      "grad_norm": 0.43881484866142273,
      "learning_rate": 5.838122470143258e-06,
      "loss": 0.3719,
      "step": 29906
    },
    {
      "epoch": 1.3724473406452204,
      "grad_norm": 0.45728597044944763,
      "learning_rate": 5.8378807504127925e-06,
      "loss": 0.3676,
      "step": 29907
    },
    {
      "epoch": 1.372493231150475,
      "grad_norm": 0.47228026390075684,
      "learning_rate": 5.837639028667488e-06,
      "loss": 0.3451,
      "step": 29908
    },
    {
      "epoch": 1.3725391216557294,
      "grad_norm": 0.4513946771621704,
      "learning_rate": 5.837397304907926e-06,
      "loss": 0.3381,
      "step": 29909
    },
    {
      "epoch": 1.372585012160984,
      "grad_norm": 0.5032436847686768,
      "learning_rate": 5.837155579134686e-06,
      "loss": 0.4114,
      "step": 29910
    },
    {
      "epoch": 1.3726309026662382,
      "grad_norm": 0.5276680588722229,
      "learning_rate": 5.836913851348353e-06,
      "loss": 0.3682,
      "step": 29911
    },
    {
      "epoch": 1.3726767931714927,
      "grad_norm": 0.5179101824760437,
      "learning_rate": 5.8366721215495045e-06,
      "loss": 0.3855,
      "step": 29912
    },
    {
      "epoch": 1.3727226836767472,
      "grad_norm": 0.4972294867038727,
      "learning_rate": 5.836430389738727e-06,
      "loss": 0.4112,
      "step": 29913
    },
    {
      "epoch": 1.3727685741820017,
      "grad_norm": 0.46858328580856323,
      "learning_rate": 5.836188655916595e-06,
      "loss": 0.3092,
      "step": 29914
    },
    {
      "epoch": 1.3728144646872562,
      "grad_norm": 0.49752476811408997,
      "learning_rate": 5.835946920083695e-06,
      "loss": 0.4202,
      "step": 29915
    },
    {
      "epoch": 1.3728603551925107,
      "grad_norm": 0.4393417239189148,
      "learning_rate": 5.835705182240605e-06,
      "loss": 0.3146,
      "step": 29916
    },
    {
      "epoch": 1.3729062456977652,
      "grad_norm": 0.48807623982429504,
      "learning_rate": 5.835463442387909e-06,
      "loss": 0.3482,
      "step": 29917
    },
    {
      "epoch": 1.3729521362030197,
      "grad_norm": 0.5008551478385925,
      "learning_rate": 5.835221700526187e-06,
      "loss": 0.4159,
      "step": 29918
    },
    {
      "epoch": 1.372998026708274,
      "grad_norm": 0.43300560116767883,
      "learning_rate": 5.834979956656021e-06,
      "loss": 0.2896,
      "step": 29919
    },
    {
      "epoch": 1.3730439172135285,
      "grad_norm": 0.4733290374279022,
      "learning_rate": 5.834738210777992e-06,
      "loss": 0.3731,
      "step": 29920
    },
    {
      "epoch": 1.373089807718783,
      "grad_norm": 0.47431910037994385,
      "learning_rate": 5.834496462892678e-06,
      "loss": 0.3775,
      "step": 29921
    },
    {
      "epoch": 1.3731356982240375,
      "grad_norm": 0.4977334141731262,
      "learning_rate": 5.8342547130006655e-06,
      "loss": 0.3799,
      "step": 29922
    },
    {
      "epoch": 1.373181588729292,
      "grad_norm": 0.47961804270744324,
      "learning_rate": 5.8340129611025345e-06,
      "loss": 0.3457,
      "step": 29923
    },
    {
      "epoch": 1.3732274792345462,
      "grad_norm": 0.46254727244377136,
      "learning_rate": 5.833771207198865e-06,
      "loss": 0.3383,
      "step": 29924
    },
    {
      "epoch": 1.3732733697398007,
      "grad_norm": 0.4781051576137543,
      "learning_rate": 5.833529451290237e-06,
      "loss": 0.3606,
      "step": 29925
    },
    {
      "epoch": 1.3733192602450552,
      "grad_norm": 0.48764723539352417,
      "learning_rate": 5.833287693377236e-06,
      "loss": 0.3223,
      "step": 29926
    },
    {
      "epoch": 1.3733651507503097,
      "grad_norm": 0.45057082176208496,
      "learning_rate": 5.833045933460441e-06,
      "loss": 0.3641,
      "step": 29927
    },
    {
      "epoch": 1.3734110412555642,
      "grad_norm": 0.44263991713523865,
      "learning_rate": 5.832804171540432e-06,
      "loss": 0.3003,
      "step": 29928
    },
    {
      "epoch": 1.3734569317608187,
      "grad_norm": 0.47315505146980286,
      "learning_rate": 5.832562407617792e-06,
      "loss": 0.384,
      "step": 29929
    },
    {
      "epoch": 1.3735028222660732,
      "grad_norm": 0.4774218201637268,
      "learning_rate": 5.832320641693104e-06,
      "loss": 0.3602,
      "step": 29930
    },
    {
      "epoch": 1.3735487127713277,
      "grad_norm": 0.44227614998817444,
      "learning_rate": 5.832078873766946e-06,
      "loss": 0.3091,
      "step": 29931
    },
    {
      "epoch": 1.373594603276582,
      "grad_norm": 0.49108612537384033,
      "learning_rate": 5.831837103839901e-06,
      "loss": 0.398,
      "step": 29932
    },
    {
      "epoch": 1.3736404937818365,
      "grad_norm": 0.5225463509559631,
      "learning_rate": 5.83159533191255e-06,
      "loss": 0.4523,
      "step": 29933
    },
    {
      "epoch": 1.373686384287091,
      "grad_norm": 0.44263342022895813,
      "learning_rate": 5.831353557985476e-06,
      "loss": 0.3365,
      "step": 29934
    },
    {
      "epoch": 1.3737322747923455,
      "grad_norm": 0.5050567388534546,
      "learning_rate": 5.83111178205926e-06,
      "loss": 0.4021,
      "step": 29935
    },
    {
      "epoch": 1.3737781652976,
      "grad_norm": 0.4229198098182678,
      "learning_rate": 5.8308700041344804e-06,
      "loss": 0.2815,
      "step": 29936
    },
    {
      "epoch": 1.3738240558028543,
      "grad_norm": 0.5005334615707397,
      "learning_rate": 5.8306282242117206e-06,
      "loss": 0.3399,
      "step": 29937
    },
    {
      "epoch": 1.3738699463081088,
      "grad_norm": 0.4750683903694153,
      "learning_rate": 5.830386442291562e-06,
      "loss": 0.3955,
      "step": 29938
    },
    {
      "epoch": 1.3739158368133633,
      "grad_norm": 0.4602978825569153,
      "learning_rate": 5.830144658374585e-06,
      "loss": 0.3259,
      "step": 29939
    },
    {
      "epoch": 1.3739617273186178,
      "grad_norm": 0.47690650820732117,
      "learning_rate": 5.829902872461376e-06,
      "loss": 0.3891,
      "step": 29940
    },
    {
      "epoch": 1.3740076178238723,
      "grad_norm": 0.5023176074028015,
      "learning_rate": 5.82966108455251e-06,
      "loss": 0.3992,
      "step": 29941
    },
    {
      "epoch": 1.3740535083291268,
      "grad_norm": 0.5280569195747375,
      "learning_rate": 5.829419294648571e-06,
      "loss": 0.4455,
      "step": 29942
    },
    {
      "epoch": 1.3740993988343813,
      "grad_norm": 0.48837098479270935,
      "learning_rate": 5.829177502750141e-06,
      "loss": 0.3954,
      "step": 29943
    },
    {
      "epoch": 1.3741452893396358,
      "grad_norm": 0.4593237638473511,
      "learning_rate": 5.8289357088578e-06,
      "loss": 0.3378,
      "step": 29944
    },
    {
      "epoch": 1.37419117984489,
      "grad_norm": 0.4931478202342987,
      "learning_rate": 5.828693912972131e-06,
      "loss": 0.3545,
      "step": 29945
    },
    {
      "epoch": 1.3742370703501445,
      "grad_norm": 0.43119341135025024,
      "learning_rate": 5.828452115093714e-06,
      "loss": 0.3037,
      "step": 29946
    },
    {
      "epoch": 1.374282960855399,
      "grad_norm": 0.46201884746551514,
      "learning_rate": 5.828210315223132e-06,
      "loss": 0.3064,
      "step": 29947
    },
    {
      "epoch": 1.3743288513606535,
      "grad_norm": 0.49133220314979553,
      "learning_rate": 5.827968513360965e-06,
      "loss": 0.3737,
      "step": 29948
    },
    {
      "epoch": 1.374374741865908,
      "grad_norm": 0.4481653869152069,
      "learning_rate": 5.827726709507795e-06,
      "loss": 0.322,
      "step": 29949
    },
    {
      "epoch": 1.3744206323711623,
      "grad_norm": 0.5017622709274292,
      "learning_rate": 5.827484903664202e-06,
      "loss": 0.3499,
      "step": 29950
    },
    {
      "epoch": 1.3744665228764168,
      "grad_norm": 0.5003820061683655,
      "learning_rate": 5.8272430958307725e-06,
      "loss": 0.4117,
      "step": 29951
    },
    {
      "epoch": 1.3745124133816713,
      "grad_norm": 0.41425642371177673,
      "learning_rate": 5.8270012860080825e-06,
      "loss": 0.2387,
      "step": 29952
    },
    {
      "epoch": 1.3745583038869258,
      "grad_norm": 0.47807297110557556,
      "learning_rate": 5.826759474196715e-06,
      "loss": 0.3733,
      "step": 29953
    },
    {
      "epoch": 1.3746041943921803,
      "grad_norm": 0.49599745869636536,
      "learning_rate": 5.826517660397253e-06,
      "loss": 0.452,
      "step": 29954
    },
    {
      "epoch": 1.3746500848974348,
      "grad_norm": 0.4883154630661011,
      "learning_rate": 5.826275844610275e-06,
      "loss": 0.4042,
      "step": 29955
    },
    {
      "epoch": 1.3746959754026893,
      "grad_norm": 0.44748809933662415,
      "learning_rate": 5.826034026836366e-06,
      "loss": 0.3406,
      "step": 29956
    },
    {
      "epoch": 1.3747418659079438,
      "grad_norm": 0.5185078382492065,
      "learning_rate": 5.825792207076106e-06,
      "loss": 0.44,
      "step": 29957
    },
    {
      "epoch": 1.374787756413198,
      "grad_norm": 0.4805600643157959,
      "learning_rate": 5.825550385330076e-06,
      "loss": 0.3756,
      "step": 29958
    },
    {
      "epoch": 1.3748336469184526,
      "grad_norm": 0.428946852684021,
      "learning_rate": 5.825308561598856e-06,
      "loss": 0.2906,
      "step": 29959
    },
    {
      "epoch": 1.374879537423707,
      "grad_norm": 0.45514506101608276,
      "learning_rate": 5.8250667358830315e-06,
      "loss": 0.3388,
      "step": 29960
    },
    {
      "epoch": 1.3749254279289616,
      "grad_norm": 0.48010337352752686,
      "learning_rate": 5.824824908183181e-06,
      "loss": 0.356,
      "step": 29961
    },
    {
      "epoch": 1.3749713184342158,
      "grad_norm": 0.4652189314365387,
      "learning_rate": 5.824583078499888e-06,
      "loss": 0.4319,
      "step": 29962
    },
    {
      "epoch": 1.3750172089394703,
      "grad_norm": 0.49779486656188965,
      "learning_rate": 5.824341246833732e-06,
      "loss": 0.3602,
      "step": 29963
    },
    {
      "epoch": 1.3750630994447248,
      "grad_norm": 0.5026893019676208,
      "learning_rate": 5.824099413185295e-06,
      "loss": 0.4436,
      "step": 29964
    },
    {
      "epoch": 1.3751089899499793,
      "grad_norm": 0.5119026899337769,
      "learning_rate": 5.823857577555161e-06,
      "loss": 0.4486,
      "step": 29965
    },
    {
      "epoch": 1.3751548804552338,
      "grad_norm": 0.48339319229125977,
      "learning_rate": 5.823615739943909e-06,
      "loss": 0.4345,
      "step": 29966
    },
    {
      "epoch": 1.3752007709604883,
      "grad_norm": 0.49107396602630615,
      "learning_rate": 5.82337390035212e-06,
      "loss": 0.3742,
      "step": 29967
    },
    {
      "epoch": 1.3752466614657428,
      "grad_norm": 0.5385260581970215,
      "learning_rate": 5.823132058780378e-06,
      "loss": 0.4199,
      "step": 29968
    },
    {
      "epoch": 1.3752925519709973,
      "grad_norm": 0.4572819471359253,
      "learning_rate": 5.822890215229261e-06,
      "loss": 0.3115,
      "step": 29969
    },
    {
      "epoch": 1.3753384424762516,
      "grad_norm": 0.45857003331184387,
      "learning_rate": 5.822648369699353e-06,
      "loss": 0.3324,
      "step": 29970
    },
    {
      "epoch": 1.375384332981506,
      "grad_norm": 0.522088885307312,
      "learning_rate": 5.822406522191237e-06,
      "loss": 0.4261,
      "step": 29971
    },
    {
      "epoch": 1.3754302234867606,
      "grad_norm": 0.4853614866733551,
      "learning_rate": 5.822164672705492e-06,
      "loss": 0.335,
      "step": 29972
    },
    {
      "epoch": 1.375476113992015,
      "grad_norm": 0.5195671916007996,
      "learning_rate": 5.8219228212427e-06,
      "loss": 0.4521,
      "step": 29973
    },
    {
      "epoch": 1.3755220044972696,
      "grad_norm": 0.4595966935157776,
      "learning_rate": 5.821680967803444e-06,
      "loss": 0.3161,
      "step": 29974
    },
    {
      "epoch": 1.3755678950025239,
      "grad_norm": 0.48904651403427124,
      "learning_rate": 5.821439112388305e-06,
      "loss": 0.3889,
      "step": 29975
    },
    {
      "epoch": 1.3756137855077784,
      "grad_norm": 0.4679660499095917,
      "learning_rate": 5.821197254997862e-06,
      "loss": 0.3426,
      "step": 29976
    },
    {
      "epoch": 1.3756596760130329,
      "grad_norm": 0.42805859446525574,
      "learning_rate": 5.820955395632701e-06,
      "loss": 0.2938,
      "step": 29977
    },
    {
      "epoch": 1.3757055665182873,
      "grad_norm": 0.4697129726409912,
      "learning_rate": 5.8207135342934006e-06,
      "loss": 0.3226,
      "step": 29978
    },
    {
      "epoch": 1.3757514570235418,
      "grad_norm": 0.4855487644672394,
      "learning_rate": 5.8204716709805435e-06,
      "loss": 0.3762,
      "step": 29979
    },
    {
      "epoch": 1.3757973475287963,
      "grad_norm": 0.4444408416748047,
      "learning_rate": 5.82022980569471e-06,
      "loss": 0.3467,
      "step": 29980
    },
    {
      "epoch": 1.3758432380340508,
      "grad_norm": 0.4726026654243469,
      "learning_rate": 5.819987938436483e-06,
      "loss": 0.3746,
      "step": 29981
    },
    {
      "epoch": 1.3758891285393053,
      "grad_norm": 0.4688367545604706,
      "learning_rate": 5.819746069206445e-06,
      "loss": 0.3572,
      "step": 29982
    },
    {
      "epoch": 1.3759350190445596,
      "grad_norm": 0.5091916918754578,
      "learning_rate": 5.819504198005175e-06,
      "loss": 0.4536,
      "step": 29983
    },
    {
      "epoch": 1.3759809095498141,
      "grad_norm": 0.48533016443252563,
      "learning_rate": 5.819262324833257e-06,
      "loss": 0.3619,
      "step": 29984
    },
    {
      "epoch": 1.3760268000550686,
      "grad_norm": 0.4556562602519989,
      "learning_rate": 5.819020449691272e-06,
      "loss": 0.3351,
      "step": 29985
    },
    {
      "epoch": 1.376072690560323,
      "grad_norm": 0.4366835057735443,
      "learning_rate": 5.8187785725798e-06,
      "loss": 0.3352,
      "step": 29986
    },
    {
      "epoch": 1.3761185810655776,
      "grad_norm": 0.4405767023563385,
      "learning_rate": 5.818536693499424e-06,
      "loss": 0.3346,
      "step": 29987
    },
    {
      "epoch": 1.3761644715708319,
      "grad_norm": 0.47440850734710693,
      "learning_rate": 5.818294812450727e-06,
      "loss": 0.4103,
      "step": 29988
    },
    {
      "epoch": 1.3762103620760864,
      "grad_norm": 0.44162923097610474,
      "learning_rate": 5.818052929434288e-06,
      "loss": 0.2736,
      "step": 29989
    },
    {
      "epoch": 1.3762562525813409,
      "grad_norm": 0.49869540333747864,
      "learning_rate": 5.8178110444506896e-06,
      "loss": 0.4282,
      "step": 29990
    },
    {
      "epoch": 1.3763021430865954,
      "grad_norm": 0.42945775389671326,
      "learning_rate": 5.817569157500516e-06,
      "loss": 0.2881,
      "step": 29991
    },
    {
      "epoch": 1.3763480335918499,
      "grad_norm": 0.45787739753723145,
      "learning_rate": 5.817327268584344e-06,
      "loss": 0.3653,
      "step": 29992
    },
    {
      "epoch": 1.3763939240971044,
      "grad_norm": 0.4477265179157257,
      "learning_rate": 5.817085377702758e-06,
      "loss": 0.3217,
      "step": 29993
    },
    {
      "epoch": 1.3764398146023589,
      "grad_norm": 0.5165705680847168,
      "learning_rate": 5.816843484856341e-06,
      "loss": 0.451,
      "step": 29994
    },
    {
      "epoch": 1.3764857051076134,
      "grad_norm": 0.4712737500667572,
      "learning_rate": 5.816601590045673e-06,
      "loss": 0.3996,
      "step": 29995
    },
    {
      "epoch": 1.3765315956128676,
      "grad_norm": 0.5163758993148804,
      "learning_rate": 5.816359693271335e-06,
      "loss": 0.443,
      "step": 29996
    },
    {
      "epoch": 1.3765774861181221,
      "grad_norm": 0.4562827944755554,
      "learning_rate": 5.816117794533911e-06,
      "loss": 0.3281,
      "step": 29997
    },
    {
      "epoch": 1.3766233766233766,
      "grad_norm": 0.44356945157051086,
      "learning_rate": 5.8158758938339795e-06,
      "loss": 0.3463,
      "step": 29998
    },
    {
      "epoch": 1.3766692671286311,
      "grad_norm": 0.47437626123428345,
      "learning_rate": 5.815633991172125e-06,
      "loss": 0.4258,
      "step": 29999
    },
    {
      "epoch": 1.3767151576338854,
      "grad_norm": 0.5096777081489563,
      "learning_rate": 5.815392086548929e-06,
      "loss": 0.3751,
      "step": 30000
    },
    {
      "epoch": 1.37676104813914,
      "grad_norm": 0.4830615520477295,
      "learning_rate": 5.81515017996497e-06,
      "loss": 0.3185,
      "step": 30001
    },
    {
      "epoch": 1.3768069386443944,
      "grad_norm": 0.4878954589366913,
      "learning_rate": 5.814908271420835e-06,
      "loss": 0.3561,
      "step": 30002
    },
    {
      "epoch": 1.376852829149649,
      "grad_norm": 0.3972504734992981,
      "learning_rate": 5.814666360917101e-06,
      "loss": 0.2538,
      "step": 30003
    },
    {
      "epoch": 1.3768987196549034,
      "grad_norm": 0.5065438747406006,
      "learning_rate": 5.81442444845435e-06,
      "loss": 0.4021,
      "step": 30004
    },
    {
      "epoch": 1.376944610160158,
      "grad_norm": 0.49227166175842285,
      "learning_rate": 5.814182534033168e-06,
      "loss": 0.3427,
      "step": 30005
    },
    {
      "epoch": 1.3769905006654124,
      "grad_norm": 0.5042336583137512,
      "learning_rate": 5.813940617654134e-06,
      "loss": 0.3602,
      "step": 30006
    },
    {
      "epoch": 1.377036391170667,
      "grad_norm": 0.41764673590660095,
      "learning_rate": 5.813698699317828e-06,
      "loss": 0.2805,
      "step": 30007
    },
    {
      "epoch": 1.3770822816759212,
      "grad_norm": 0.4617655277252197,
      "learning_rate": 5.8134567790248345e-06,
      "loss": 0.3692,
      "step": 30008
    },
    {
      "epoch": 1.3771281721811757,
      "grad_norm": 0.4986257553100586,
      "learning_rate": 5.813214856775733e-06,
      "loss": 0.3811,
      "step": 30009
    },
    {
      "epoch": 1.3771740626864302,
      "grad_norm": 0.4710755944252014,
      "learning_rate": 5.8129729325711075e-06,
      "loss": 0.3512,
      "step": 30010
    },
    {
      "epoch": 1.3772199531916847,
      "grad_norm": 0.4673018455505371,
      "learning_rate": 5.81273100641154e-06,
      "loss": 0.3545,
      "step": 30011
    },
    {
      "epoch": 1.3772658436969392,
      "grad_norm": 0.5035912394523621,
      "learning_rate": 5.812489078297609e-06,
      "loss": 0.4364,
      "step": 30012
    },
    {
      "epoch": 1.3773117342021934,
      "grad_norm": 0.42631104588508606,
      "learning_rate": 5.812247148229898e-06,
      "loss": 0.2956,
      "step": 30013
    },
    {
      "epoch": 1.377357624707448,
      "grad_norm": 0.4278251826763153,
      "learning_rate": 5.812005216208989e-06,
      "loss": 0.309,
      "step": 30014
    },
    {
      "epoch": 1.3774035152127024,
      "grad_norm": 0.4798485040664673,
      "learning_rate": 5.8117632822354646e-06,
      "loss": 0.3443,
      "step": 30015
    },
    {
      "epoch": 1.377449405717957,
      "grad_norm": 0.4846261143684387,
      "learning_rate": 5.811521346309905e-06,
      "loss": 0.3293,
      "step": 30016
    },
    {
      "epoch": 1.3774952962232114,
      "grad_norm": 0.4743522107601166,
      "learning_rate": 5.8112794084328936e-06,
      "loss": 0.3447,
      "step": 30017
    },
    {
      "epoch": 1.377541186728466,
      "grad_norm": 0.4767338037490845,
      "learning_rate": 5.811037468605011e-06,
      "loss": 0.3243,
      "step": 30018
    },
    {
      "epoch": 1.3775870772337204,
      "grad_norm": 0.5213478803634644,
      "learning_rate": 5.810795526826839e-06,
      "loss": 0.4027,
      "step": 30019
    },
    {
      "epoch": 1.377632967738975,
      "grad_norm": 0.49750325083732605,
      "learning_rate": 5.810553583098959e-06,
      "loss": 0.3683,
      "step": 30020
    },
    {
      "epoch": 1.3776788582442292,
      "grad_norm": 0.49233412742614746,
      "learning_rate": 5.8103116374219535e-06,
      "loss": 0.3644,
      "step": 30021
    },
    {
      "epoch": 1.3777247487494837,
      "grad_norm": 0.5257917642593384,
      "learning_rate": 5.810069689796407e-06,
      "loss": 0.4044,
      "step": 30022
    },
    {
      "epoch": 1.3777706392547382,
      "grad_norm": 0.4458836615085602,
      "learning_rate": 5.809827740222895e-06,
      "loss": 0.3248,
      "step": 30023
    },
    {
      "epoch": 1.3778165297599927,
      "grad_norm": 0.45984727144241333,
      "learning_rate": 5.809585788702004e-06,
      "loss": 0.3303,
      "step": 30024
    },
    {
      "epoch": 1.3778624202652472,
      "grad_norm": 0.4440291225910187,
      "learning_rate": 5.809343835234316e-06,
      "loss": 0.3194,
      "step": 30025
    },
    {
      "epoch": 1.3779083107705015,
      "grad_norm": 0.5417717099189758,
      "learning_rate": 5.809101879820409e-06,
      "loss": 0.4939,
      "step": 30026
    },
    {
      "epoch": 1.377954201275756,
      "grad_norm": 0.49395695328712463,
      "learning_rate": 5.80885992246087e-06,
      "loss": 0.4245,
      "step": 30027
    },
    {
      "epoch": 1.3780000917810105,
      "grad_norm": 0.4934785068035126,
      "learning_rate": 5.808617963156277e-06,
      "loss": 0.3919,
      "step": 30028
    },
    {
      "epoch": 1.378045982286265,
      "grad_norm": 0.4426972270011902,
      "learning_rate": 5.808376001907212e-06,
      "loss": 0.2934,
      "step": 30029
    },
    {
      "epoch": 1.3780918727915195,
      "grad_norm": 0.5040740370750427,
      "learning_rate": 5.80813403871426e-06,
      "loss": 0.4,
      "step": 30030
    },
    {
      "epoch": 1.378137763296774,
      "grad_norm": 0.4542422592639923,
      "learning_rate": 5.807892073577999e-06,
      "loss": 0.3041,
      "step": 30031
    },
    {
      "epoch": 1.3781836538020285,
      "grad_norm": 0.5049039721488953,
      "learning_rate": 5.807650106499012e-06,
      "loss": 0.4533,
      "step": 30032
    },
    {
      "epoch": 1.378229544307283,
      "grad_norm": 0.4647717773914337,
      "learning_rate": 5.807408137477885e-06,
      "loss": 0.333,
      "step": 30033
    },
    {
      "epoch": 1.3782754348125372,
      "grad_norm": 0.48151805996894836,
      "learning_rate": 5.807166166515192e-06,
      "loss": 0.363,
      "step": 30034
    },
    {
      "epoch": 1.3783213253177917,
      "grad_norm": 0.5103830695152283,
      "learning_rate": 5.806924193611521e-06,
      "loss": 0.4319,
      "step": 30035
    },
    {
      "epoch": 1.3783672158230462,
      "grad_norm": 0.4830034673213959,
      "learning_rate": 5.806682218767452e-06,
      "loss": 0.3676,
      "step": 30036
    },
    {
      "epoch": 1.3784131063283007,
      "grad_norm": 0.4696045517921448,
      "learning_rate": 5.806440241983566e-06,
      "loss": 0.3781,
      "step": 30037
    },
    {
      "epoch": 1.3784589968335552,
      "grad_norm": 0.4406489431858063,
      "learning_rate": 5.806198263260447e-06,
      "loss": 0.3138,
      "step": 30038
    },
    {
      "epoch": 1.3785048873388095,
      "grad_norm": 0.4668993651866913,
      "learning_rate": 5.805956282598676e-06,
      "loss": 0.3482,
      "step": 30039
    },
    {
      "epoch": 1.378550777844064,
      "grad_norm": 0.4303893446922302,
      "learning_rate": 5.805714299998833e-06,
      "loss": 0.3006,
      "step": 30040
    },
    {
      "epoch": 1.3785966683493185,
      "grad_norm": 0.47575873136520386,
      "learning_rate": 5.805472315461501e-06,
      "loss": 0.3584,
      "step": 30041
    },
    {
      "epoch": 1.378642558854573,
      "grad_norm": 0.5100101828575134,
      "learning_rate": 5.805230328987265e-06,
      "loss": 0.4079,
      "step": 30042
    },
    {
      "epoch": 1.3786884493598275,
      "grad_norm": 0.4731379449367523,
      "learning_rate": 5.8049883405767015e-06,
      "loss": 0.3598,
      "step": 30043
    },
    {
      "epoch": 1.378734339865082,
      "grad_norm": 0.44419899582862854,
      "learning_rate": 5.804746350230396e-06,
      "loss": 0.3162,
      "step": 30044
    },
    {
      "epoch": 1.3787802303703365,
      "grad_norm": 0.4433366656303406,
      "learning_rate": 5.80450435794893e-06,
      "loss": 0.313,
      "step": 30045
    },
    {
      "epoch": 1.378826120875591,
      "grad_norm": 0.4854978919029236,
      "learning_rate": 5.804262363732885e-06,
      "loss": 0.4027,
      "step": 30046
    },
    {
      "epoch": 1.3788720113808453,
      "grad_norm": 0.44491836428642273,
      "learning_rate": 5.804020367582842e-06,
      "loss": 0.3308,
      "step": 30047
    },
    {
      "epoch": 1.3789179018860998,
      "grad_norm": 0.5081619620323181,
      "learning_rate": 5.8037783694993835e-06,
      "loss": 0.4722,
      "step": 30048
    },
    {
      "epoch": 1.3789637923913542,
      "grad_norm": 0.44249504804611206,
      "learning_rate": 5.803536369483092e-06,
      "loss": 0.3117,
      "step": 30049
    },
    {
      "epoch": 1.3790096828966087,
      "grad_norm": 0.4390316307544708,
      "learning_rate": 5.803294367534551e-06,
      "loss": 0.3155,
      "step": 30050
    },
    {
      "epoch": 1.379055573401863,
      "grad_norm": 0.44082918763160706,
      "learning_rate": 5.803052363654338e-06,
      "loss": 0.3115,
      "step": 30051
    },
    {
      "epoch": 1.3791014639071175,
      "grad_norm": 0.44473499059677124,
      "learning_rate": 5.802810357843038e-06,
      "loss": 0.2948,
      "step": 30052
    },
    {
      "epoch": 1.379147354412372,
      "grad_norm": 0.4447748363018036,
      "learning_rate": 5.802568350101234e-06,
      "loss": 0.351,
      "step": 30053
    },
    {
      "epoch": 1.3791932449176265,
      "grad_norm": 0.4902257025241852,
      "learning_rate": 5.802326340429505e-06,
      "loss": 0.4176,
      "step": 30054
    },
    {
      "epoch": 1.379239135422881,
      "grad_norm": 0.4404298961162567,
      "learning_rate": 5.8020843288284344e-06,
      "loss": 0.3164,
      "step": 30055
    },
    {
      "epoch": 1.3792850259281355,
      "grad_norm": 0.47950705885887146,
      "learning_rate": 5.801842315298606e-06,
      "loss": 0.3564,
      "step": 30056
    },
    {
      "epoch": 1.37933091643339,
      "grad_norm": 0.4463609457015991,
      "learning_rate": 5.801600299840597e-06,
      "loss": 0.3132,
      "step": 30057
    },
    {
      "epoch": 1.3793768069386445,
      "grad_norm": 0.4774281680583954,
      "learning_rate": 5.801358282454994e-06,
      "loss": 0.3508,
      "step": 30058
    },
    {
      "epoch": 1.3794226974438988,
      "grad_norm": 0.5036218166351318,
      "learning_rate": 5.8011162631423765e-06,
      "loss": 0.4499,
      "step": 30059
    },
    {
      "epoch": 1.3794685879491533,
      "grad_norm": 0.46814975142478943,
      "learning_rate": 5.800874241903329e-06,
      "loss": 0.3515,
      "step": 30060
    },
    {
      "epoch": 1.3795144784544078,
      "grad_norm": 0.4841305911540985,
      "learning_rate": 5.80063221873843e-06,
      "loss": 0.3357,
      "step": 30061
    },
    {
      "epoch": 1.3795603689596623,
      "grad_norm": 0.5619818568229675,
      "learning_rate": 5.800390193648265e-06,
      "loss": 0.3162,
      "step": 30062
    },
    {
      "epoch": 1.3796062594649168,
      "grad_norm": 0.5121968388557434,
      "learning_rate": 5.8001481666334114e-06,
      "loss": 0.431,
      "step": 30063
    },
    {
      "epoch": 1.379652149970171,
      "grad_norm": 0.5101747512817383,
      "learning_rate": 5.799906137694455e-06,
      "loss": 0.4058,
      "step": 30064
    },
    {
      "epoch": 1.3796980404754255,
      "grad_norm": 0.583008348941803,
      "learning_rate": 5.799664106831977e-06,
      "loss": 0.3597,
      "step": 30065
    },
    {
      "epoch": 1.37974393098068,
      "grad_norm": 0.5467517971992493,
      "learning_rate": 5.79942207404656e-06,
      "loss": 0.4141,
      "step": 30066
    },
    {
      "epoch": 1.3797898214859345,
      "grad_norm": 0.5037644505500793,
      "learning_rate": 5.799180039338786e-06,
      "loss": 0.4007,
      "step": 30067
    },
    {
      "epoch": 1.379835711991189,
      "grad_norm": 0.5072501301765442,
      "learning_rate": 5.7989380027092335e-06,
      "loss": 0.4386,
      "step": 30068
    },
    {
      "epoch": 1.3798816024964435,
      "grad_norm": 0.5426491498947144,
      "learning_rate": 5.798695964158487e-06,
      "loss": 0.4431,
      "step": 30069
    },
    {
      "epoch": 1.379927493001698,
      "grad_norm": 0.4687478840351105,
      "learning_rate": 5.798453923687132e-06,
      "loss": 0.363,
      "step": 30070
    },
    {
      "epoch": 1.3799733835069525,
      "grad_norm": 0.4715091586112976,
      "learning_rate": 5.798211881295746e-06,
      "loss": 0.3161,
      "step": 30071
    },
    {
      "epoch": 1.3800192740122068,
      "grad_norm": 0.47338390350341797,
      "learning_rate": 5.797969836984911e-06,
      "loss": 0.3861,
      "step": 30072
    },
    {
      "epoch": 1.3800651645174613,
      "grad_norm": 0.5061212182044983,
      "learning_rate": 5.797727790755212e-06,
      "loss": 0.3869,
      "step": 30073
    },
    {
      "epoch": 1.3801110550227158,
      "grad_norm": 0.4940020740032196,
      "learning_rate": 5.797485742607229e-06,
      "loss": 0.3861,
      "step": 30074
    },
    {
      "epoch": 1.3801569455279703,
      "grad_norm": 0.4310090243816376,
      "learning_rate": 5.7972436925415436e-06,
      "loss": 0.3,
      "step": 30075
    },
    {
      "epoch": 1.3802028360332248,
      "grad_norm": 0.5196375846862793,
      "learning_rate": 5.797001640558739e-06,
      "loss": 0.4616,
      "step": 30076
    },
    {
      "epoch": 1.380248726538479,
      "grad_norm": 0.4669770896434784,
      "learning_rate": 5.796759586659398e-06,
      "loss": 0.3147,
      "step": 30077
    },
    {
      "epoch": 1.3802946170437336,
      "grad_norm": 0.4549065828323364,
      "learning_rate": 5.7965175308441015e-06,
      "loss": 0.3422,
      "step": 30078
    },
    {
      "epoch": 1.380340507548988,
      "grad_norm": 0.42525434494018555,
      "learning_rate": 5.796275473113432e-06,
      "loss": 0.2884,
      "step": 30079
    },
    {
      "epoch": 1.3803863980542426,
      "grad_norm": 0.4323212802410126,
      "learning_rate": 5.79603341346797e-06,
      "loss": 0.312,
      "step": 30080
    },
    {
      "epoch": 1.380432288559497,
      "grad_norm": 0.4833287000656128,
      "learning_rate": 5.7957913519083e-06,
      "loss": 0.3692,
      "step": 30081
    },
    {
      "epoch": 1.3804781790647516,
      "grad_norm": 0.4570082128047943,
      "learning_rate": 5.795549288435003e-06,
      "loss": 0.3071,
      "step": 30082
    },
    {
      "epoch": 1.380524069570006,
      "grad_norm": 0.46827077865600586,
      "learning_rate": 5.795307223048659e-06,
      "loss": 0.3557,
      "step": 30083
    },
    {
      "epoch": 1.3805699600752606,
      "grad_norm": 0.48298442363739014,
      "learning_rate": 5.795065155749855e-06,
      "loss": 0.3215,
      "step": 30084
    },
    {
      "epoch": 1.3806158505805148,
      "grad_norm": 0.4812818169593811,
      "learning_rate": 5.794823086539168e-06,
      "loss": 0.4099,
      "step": 30085
    },
    {
      "epoch": 1.3806617410857693,
      "grad_norm": 0.4831850826740265,
      "learning_rate": 5.794581015417182e-06,
      "loss": 0.3506,
      "step": 30086
    },
    {
      "epoch": 1.3807076315910238,
      "grad_norm": 0.41549769043922424,
      "learning_rate": 5.794338942384482e-06,
      "loss": 0.2626,
      "step": 30087
    },
    {
      "epoch": 1.3807535220962783,
      "grad_norm": 0.4679988920688629,
      "learning_rate": 5.794096867441646e-06,
      "loss": 0.3475,
      "step": 30088
    },
    {
      "epoch": 1.3807994126015326,
      "grad_norm": 0.46672746539115906,
      "learning_rate": 5.793854790589257e-06,
      "loss": 0.3828,
      "step": 30089
    },
    {
      "epoch": 1.380845303106787,
      "grad_norm": 0.48221760988235474,
      "learning_rate": 5.793612711827899e-06,
      "loss": 0.4281,
      "step": 30090
    },
    {
      "epoch": 1.3808911936120416,
      "grad_norm": 0.4793669283390045,
      "learning_rate": 5.793370631158153e-06,
      "loss": 0.3571,
      "step": 30091
    },
    {
      "epoch": 1.380937084117296,
      "grad_norm": 0.4874010980129242,
      "learning_rate": 5.793128548580601e-06,
      "loss": 0.3814,
      "step": 30092
    },
    {
      "epoch": 1.3809829746225506,
      "grad_norm": 0.4862196445465088,
      "learning_rate": 5.792886464095825e-06,
      "loss": 0.341,
      "step": 30093
    },
    {
      "epoch": 1.381028865127805,
      "grad_norm": 0.48247984051704407,
      "learning_rate": 5.792644377704406e-06,
      "loss": 0.3304,
      "step": 30094
    },
    {
      "epoch": 1.3810747556330596,
      "grad_norm": 0.4899463653564453,
      "learning_rate": 5.79240228940693e-06,
      "loss": 0.3884,
      "step": 30095
    },
    {
      "epoch": 1.381120646138314,
      "grad_norm": 0.48217883706092834,
      "learning_rate": 5.792160199203974e-06,
      "loss": 0.3745,
      "step": 30096
    },
    {
      "epoch": 1.3811665366435684,
      "grad_norm": 0.48121970891952515,
      "learning_rate": 5.791918107096124e-06,
      "loss": 0.3768,
      "step": 30097
    },
    {
      "epoch": 1.3812124271488229,
      "grad_norm": 0.4415617883205414,
      "learning_rate": 5.79167601308396e-06,
      "loss": 0.3633,
      "step": 30098
    },
    {
      "epoch": 1.3812583176540774,
      "grad_norm": 0.46626052260398865,
      "learning_rate": 5.791433917168067e-06,
      "loss": 0.3114,
      "step": 30099
    },
    {
      "epoch": 1.3813042081593319,
      "grad_norm": 0.47387996315956116,
      "learning_rate": 5.791191819349022e-06,
      "loss": 0.3188,
      "step": 30100
    },
    {
      "epoch": 1.3813500986645864,
      "grad_norm": 0.4824138879776001,
      "learning_rate": 5.790949719627414e-06,
      "loss": 0.3922,
      "step": 30101
    },
    {
      "epoch": 1.3813959891698406,
      "grad_norm": 0.4569844901561737,
      "learning_rate": 5.790707618003819e-06,
      "loss": 0.3071,
      "step": 30102
    },
    {
      "epoch": 1.3814418796750951,
      "grad_norm": 0.4678370952606201,
      "learning_rate": 5.790465514478823e-06,
      "loss": 0.3386,
      "step": 30103
    },
    {
      "epoch": 1.3814877701803496,
      "grad_norm": 0.46363189816474915,
      "learning_rate": 5.790223409053007e-06,
      "loss": 0.342,
      "step": 30104
    },
    {
      "epoch": 1.3815336606856041,
      "grad_norm": 0.4456811249256134,
      "learning_rate": 5.789981301726953e-06,
      "loss": 0.3305,
      "step": 30105
    },
    {
      "epoch": 1.3815795511908586,
      "grad_norm": 0.44500625133514404,
      "learning_rate": 5.789739192501242e-06,
      "loss": 0.3297,
      "step": 30106
    },
    {
      "epoch": 1.3816254416961131,
      "grad_norm": 0.5086700916290283,
      "learning_rate": 5.789497081376458e-06,
      "loss": 0.4421,
      "step": 30107
    },
    {
      "epoch": 1.3816713322013676,
      "grad_norm": 0.469319224357605,
      "learning_rate": 5.789254968353183e-06,
      "loss": 0.3695,
      "step": 30108
    },
    {
      "epoch": 1.3817172227066221,
      "grad_norm": 0.4956066608428955,
      "learning_rate": 5.7890128534319986e-06,
      "loss": 0.3949,
      "step": 30109
    },
    {
      "epoch": 1.3817631132118764,
      "grad_norm": 0.47333189845085144,
      "learning_rate": 5.788770736613489e-06,
      "loss": 0.3892,
      "step": 30110
    },
    {
      "epoch": 1.381809003717131,
      "grad_norm": 0.44093748927116394,
      "learning_rate": 5.788528617898232e-06,
      "loss": 0.3378,
      "step": 30111
    },
    {
      "epoch": 1.3818548942223854,
      "grad_norm": 0.5100197196006775,
      "learning_rate": 5.788286497286816e-06,
      "loss": 0.4026,
      "step": 30112
    },
    {
      "epoch": 1.3819007847276399,
      "grad_norm": 0.4195316433906555,
      "learning_rate": 5.788044374779816e-06,
      "loss": 0.272,
      "step": 30113
    },
    {
      "epoch": 1.3819466752328944,
      "grad_norm": 0.493714839220047,
      "learning_rate": 5.787802250377821e-06,
      "loss": 0.4634,
      "step": 30114
    },
    {
      "epoch": 1.3819925657381487,
      "grad_norm": 0.4993947446346283,
      "learning_rate": 5.787560124081409e-06,
      "loss": 0.3796,
      "step": 30115
    },
    {
      "epoch": 1.3820384562434032,
      "grad_norm": 0.44466516375541687,
      "learning_rate": 5.787317995891163e-06,
      "loss": 0.3132,
      "step": 30116
    },
    {
      "epoch": 1.3820843467486577,
      "grad_norm": 0.4807978570461273,
      "learning_rate": 5.7870758658076654e-06,
      "loss": 0.343,
      "step": 30117
    },
    {
      "epoch": 1.3821302372539122,
      "grad_norm": 0.47198548913002014,
      "learning_rate": 5.7868337338315006e-06,
      "loss": 0.3975,
      "step": 30118
    },
    {
      "epoch": 1.3821761277591667,
      "grad_norm": 0.4122215211391449,
      "learning_rate": 5.786591599963247e-06,
      "loss": 0.2694,
      "step": 30119
    },
    {
      "epoch": 1.3822220182644211,
      "grad_norm": 0.43830570578575134,
      "learning_rate": 5.78634946420349e-06,
      "loss": 0.3405,
      "step": 30120
    },
    {
      "epoch": 1.3822679087696756,
      "grad_norm": 0.5233800411224365,
      "learning_rate": 5.786107326552811e-06,
      "loss": 0.4738,
      "step": 30121
    },
    {
      "epoch": 1.3823137992749301,
      "grad_norm": 0.4553917646408081,
      "learning_rate": 5.785865187011791e-06,
      "loss": 0.3148,
      "step": 30122
    },
    {
      "epoch": 1.3823596897801844,
      "grad_norm": 0.4600346088409424,
      "learning_rate": 5.785623045581013e-06,
      "loss": 0.3581,
      "step": 30123
    },
    {
      "epoch": 1.382405580285439,
      "grad_norm": 0.44866496324539185,
      "learning_rate": 5.785380902261061e-06,
      "loss": 0.3316,
      "step": 30124
    },
    {
      "epoch": 1.3824514707906934,
      "grad_norm": 0.42523252964019775,
      "learning_rate": 5.785138757052515e-06,
      "loss": 0.3097,
      "step": 30125
    },
    {
      "epoch": 1.382497361295948,
      "grad_norm": 0.4859412908554077,
      "learning_rate": 5.784896609955958e-06,
      "loss": 0.4136,
      "step": 30126
    },
    {
      "epoch": 1.3825432518012024,
      "grad_norm": 0.5036506652832031,
      "learning_rate": 5.784654460971972e-06,
      "loss": 0.3529,
      "step": 30127
    },
    {
      "epoch": 1.3825891423064567,
      "grad_norm": 0.48324862122535706,
      "learning_rate": 5.78441231010114e-06,
      "loss": 0.4217,
      "step": 30128
    },
    {
      "epoch": 1.3826350328117112,
      "grad_norm": 0.4600822925567627,
      "learning_rate": 5.784170157344044e-06,
      "loss": 0.3407,
      "step": 30129
    },
    {
      "epoch": 1.3826809233169657,
      "grad_norm": 0.44016775488853455,
      "learning_rate": 5.7839280027012666e-06,
      "loss": 0.3281,
      "step": 30130
    },
    {
      "epoch": 1.3827268138222202,
      "grad_norm": 0.4400809407234192,
      "learning_rate": 5.78368584617339e-06,
      "loss": 0.3033,
      "step": 30131
    },
    {
      "epoch": 1.3827727043274747,
      "grad_norm": 0.41949546337127686,
      "learning_rate": 5.783443687760997e-06,
      "loss": 0.2674,
      "step": 30132
    },
    {
      "epoch": 1.3828185948327292,
      "grad_norm": 0.4590534269809723,
      "learning_rate": 5.783201527464667e-06,
      "loss": 0.3461,
      "step": 30133
    },
    {
      "epoch": 1.3828644853379837,
      "grad_norm": 0.4224918782711029,
      "learning_rate": 5.7829593652849845e-06,
      "loss": 0.2772,
      "step": 30134
    },
    {
      "epoch": 1.3829103758432382,
      "grad_norm": 0.437712699174881,
      "learning_rate": 5.782717201222533e-06,
      "loss": 0.304,
      "step": 30135
    },
    {
      "epoch": 1.3829562663484924,
      "grad_norm": 0.48794978857040405,
      "learning_rate": 5.782475035277895e-06,
      "loss": 0.3941,
      "step": 30136
    },
    {
      "epoch": 1.383002156853747,
      "grad_norm": 0.4472852945327759,
      "learning_rate": 5.782232867451649e-06,
      "loss": 0.3613,
      "step": 30137
    },
    {
      "epoch": 1.3830480473590014,
      "grad_norm": 0.47399502992630005,
      "learning_rate": 5.7819906977443805e-06,
      "loss": 0.3418,
      "step": 30138
    },
    {
      "epoch": 1.383093937864256,
      "grad_norm": 0.46424493193626404,
      "learning_rate": 5.7817485261566716e-06,
      "loss": 0.324,
      "step": 30139
    },
    {
      "epoch": 1.3831398283695102,
      "grad_norm": 0.43304571509361267,
      "learning_rate": 5.7815063526891034e-06,
      "loss": 0.3182,
      "step": 30140
    },
    {
      "epoch": 1.3831857188747647,
      "grad_norm": 0.46367359161376953,
      "learning_rate": 5.781264177342261e-06,
      "loss": 0.3248,
      "step": 30141
    },
    {
      "epoch": 1.3832316093800192,
      "grad_norm": 0.482255220413208,
      "learning_rate": 5.781022000116724e-06,
      "loss": 0.3747,
      "step": 30142
    },
    {
      "epoch": 1.3832774998852737,
      "grad_norm": 0.4829370677471161,
      "learning_rate": 5.780779821013074e-06,
      "loss": 0.365,
      "step": 30143
    },
    {
      "epoch": 1.3833233903905282,
      "grad_norm": 0.4703187346458435,
      "learning_rate": 5.780537640031897e-06,
      "loss": 0.3641,
      "step": 30144
    },
    {
      "epoch": 1.3833692808957827,
      "grad_norm": 0.47962355613708496,
      "learning_rate": 5.7802954571737725e-06,
      "loss": 0.3876,
      "step": 30145
    },
    {
      "epoch": 1.3834151714010372,
      "grad_norm": 0.4452257454395294,
      "learning_rate": 5.780053272439283e-06,
      "loss": 0.3185,
      "step": 30146
    },
    {
      "epoch": 1.3834610619062917,
      "grad_norm": 0.44089433550834656,
      "learning_rate": 5.779811085829014e-06,
      "loss": 0.3576,
      "step": 30147
    },
    {
      "epoch": 1.383506952411546,
      "grad_norm": 0.465387225151062,
      "learning_rate": 5.779568897343543e-06,
      "loss": 0.3876,
      "step": 30148
    },
    {
      "epoch": 1.3835528429168005,
      "grad_norm": 0.4795728027820587,
      "learning_rate": 5.779326706983456e-06,
      "loss": 0.3551,
      "step": 30149
    },
    {
      "epoch": 1.383598733422055,
      "grad_norm": 0.4223843812942505,
      "learning_rate": 5.779084514749333e-06,
      "loss": 0.3172,
      "step": 30150
    },
    {
      "epoch": 1.3836446239273095,
      "grad_norm": 0.49916842579841614,
      "learning_rate": 5.778842320641758e-06,
      "loss": 0.4036,
      "step": 30151
    },
    {
      "epoch": 1.383690514432564,
      "grad_norm": 0.4564020037651062,
      "learning_rate": 5.778600124661313e-06,
      "loss": 0.3218,
      "step": 30152
    },
    {
      "epoch": 1.3837364049378182,
      "grad_norm": 0.44122374057769775,
      "learning_rate": 5.778357926808582e-06,
      "loss": 0.3078,
      "step": 30153
    },
    {
      "epoch": 1.3837822954430727,
      "grad_norm": 0.4379919171333313,
      "learning_rate": 5.778115727084144e-06,
      "loss": 0.3266,
      "step": 30154
    },
    {
      "epoch": 1.3838281859483272,
      "grad_norm": 0.4965982437133789,
      "learning_rate": 5.7778735254885845e-06,
      "loss": 0.3704,
      "step": 30155
    },
    {
      "epoch": 1.3838740764535817,
      "grad_norm": 0.4463886320590973,
      "learning_rate": 5.777631322022483e-06,
      "loss": 0.3167,
      "step": 30156
    },
    {
      "epoch": 1.3839199669588362,
      "grad_norm": 0.4470362961292267,
      "learning_rate": 5.777389116686426e-06,
      "loss": 0.3436,
      "step": 30157
    },
    {
      "epoch": 1.3839658574640907,
      "grad_norm": 0.4523601531982422,
      "learning_rate": 5.777146909480993e-06,
      "loss": 0.3049,
      "step": 30158
    },
    {
      "epoch": 1.3840117479693452,
      "grad_norm": 0.4852578938007355,
      "learning_rate": 5.776904700406765e-06,
      "loss": 0.4004,
      "step": 30159
    },
    {
      "epoch": 1.3840576384745997,
      "grad_norm": 0.44050097465515137,
      "learning_rate": 5.776662489464327e-06,
      "loss": 0.3283,
      "step": 30160
    },
    {
      "epoch": 1.384103528979854,
      "grad_norm": 0.4994003474712372,
      "learning_rate": 5.776420276654262e-06,
      "loss": 0.4142,
      "step": 30161
    },
    {
      "epoch": 1.3841494194851085,
      "grad_norm": 0.43885889649391174,
      "learning_rate": 5.7761780619771505e-06,
      "loss": 0.3022,
      "step": 30162
    },
    {
      "epoch": 1.384195309990363,
      "grad_norm": 0.4919586479663849,
      "learning_rate": 5.775935845433575e-06,
      "loss": 0.4391,
      "step": 30163
    },
    {
      "epoch": 1.3842412004956175,
      "grad_norm": 0.4547795057296753,
      "learning_rate": 5.775693627024121e-06,
      "loss": 0.3732,
      "step": 30164
    },
    {
      "epoch": 1.384287091000872,
      "grad_norm": 0.42514944076538086,
      "learning_rate": 5.775451406749367e-06,
      "loss": 0.2962,
      "step": 30165
    },
    {
      "epoch": 1.3843329815061263,
      "grad_norm": 0.47637006640434265,
      "learning_rate": 5.775209184609898e-06,
      "loss": 0.3486,
      "step": 30166
    },
    {
      "epoch": 1.3843788720113808,
      "grad_norm": 0.4872959554195404,
      "learning_rate": 5.7749669606062945e-06,
      "loss": 0.4134,
      "step": 30167
    },
    {
      "epoch": 1.3844247625166353,
      "grad_norm": 0.5105438232421875,
      "learning_rate": 5.774724734739141e-06,
      "loss": 0.4293,
      "step": 30168
    },
    {
      "epoch": 1.3844706530218898,
      "grad_norm": 0.44854724407196045,
      "learning_rate": 5.77448250700902e-06,
      "loss": 0.3299,
      "step": 30169
    },
    {
      "epoch": 1.3845165435271443,
      "grad_norm": 0.4842880964279175,
      "learning_rate": 5.77424027741651e-06,
      "loss": 0.3974,
      "step": 30170
    },
    {
      "epoch": 1.3845624340323988,
      "grad_norm": 0.4838508367538452,
      "learning_rate": 5.773998045962197e-06,
      "loss": 0.3725,
      "step": 30171
    },
    {
      "epoch": 1.3846083245376533,
      "grad_norm": 0.467365026473999,
      "learning_rate": 5.7737558126466655e-06,
      "loss": 0.3704,
      "step": 30172
    },
    {
      "epoch": 1.3846542150429078,
      "grad_norm": 0.5171970129013062,
      "learning_rate": 5.773513577470493e-06,
      "loss": 0.3323,
      "step": 30173
    },
    {
      "epoch": 1.384700105548162,
      "grad_norm": 0.5004127025604248,
      "learning_rate": 5.773271340434265e-06,
      "loss": 0.4512,
      "step": 30174
    },
    {
      "epoch": 1.3847459960534165,
      "grad_norm": 0.4914829432964325,
      "learning_rate": 5.773029101538566e-06,
      "loss": 0.394,
      "step": 30175
    },
    {
      "epoch": 1.384791886558671,
      "grad_norm": 0.5040169954299927,
      "learning_rate": 5.772786860783972e-06,
      "loss": 0.3659,
      "step": 30176
    },
    {
      "epoch": 1.3848377770639255,
      "grad_norm": 0.4755440354347229,
      "learning_rate": 5.772544618171072e-06,
      "loss": 0.3515,
      "step": 30177
    },
    {
      "epoch": 1.3848836675691798,
      "grad_norm": 0.48394501209259033,
      "learning_rate": 5.772302373700446e-06,
      "loss": 0.3935,
      "step": 30178
    },
    {
      "epoch": 1.3849295580744343,
      "grad_norm": 0.4634353220462799,
      "learning_rate": 5.772060127372676e-06,
      "loss": 0.3548,
      "step": 30179
    },
    {
      "epoch": 1.3849754485796888,
      "grad_norm": 0.4337379038333893,
      "learning_rate": 5.771817879188345e-06,
      "loss": 0.3236,
      "step": 30180
    },
    {
      "epoch": 1.3850213390849433,
      "grad_norm": 0.5086374282836914,
      "learning_rate": 5.7715756291480355e-06,
      "loss": 0.4427,
      "step": 30181
    },
    {
      "epoch": 1.3850672295901978,
      "grad_norm": 0.46645882725715637,
      "learning_rate": 5.77133337725233e-06,
      "loss": 0.3494,
      "step": 30182
    },
    {
      "epoch": 1.3851131200954523,
      "grad_norm": 0.4445298910140991,
      "learning_rate": 5.771091123501812e-06,
      "loss": 0.3159,
      "step": 30183
    },
    {
      "epoch": 1.3851590106007068,
      "grad_norm": 0.45073261857032776,
      "learning_rate": 5.770848867897062e-06,
      "loss": 0.3553,
      "step": 30184
    },
    {
      "epoch": 1.3852049011059613,
      "grad_norm": 0.48292627930641174,
      "learning_rate": 5.770606610438665e-06,
      "loss": 0.3594,
      "step": 30185
    },
    {
      "epoch": 1.3852507916112156,
      "grad_norm": 0.6269731521606445,
      "learning_rate": 5.770364351127202e-06,
      "loss": 0.3412,
      "step": 30186
    },
    {
      "epoch": 1.38529668211647,
      "grad_norm": 0.45274046063423157,
      "learning_rate": 5.770122089963255e-06,
      "loss": 0.3418,
      "step": 30187
    },
    {
      "epoch": 1.3853425726217246,
      "grad_norm": 0.4876083433628082,
      "learning_rate": 5.769879826947406e-06,
      "loss": 0.3592,
      "step": 30188
    },
    {
      "epoch": 1.385388463126979,
      "grad_norm": 0.46288368105888367,
      "learning_rate": 5.769637562080242e-06,
      "loss": 0.3374,
      "step": 30189
    },
    {
      "epoch": 1.3854343536322336,
      "grad_norm": 0.469482958316803,
      "learning_rate": 5.7693952953623435e-06,
      "loss": 0.3391,
      "step": 30190
    },
    {
      "epoch": 1.3854802441374878,
      "grad_norm": 0.476589560508728,
      "learning_rate": 5.7691530267942884e-06,
      "loss": 0.2171,
      "step": 30191
    },
    {
      "epoch": 1.3855261346427423,
      "grad_norm": 0.47293582558631897,
      "learning_rate": 5.768910756376666e-06,
      "loss": 0.3792,
      "step": 30192
    },
    {
      "epoch": 1.3855720251479968,
      "grad_norm": 0.4677846133708954,
      "learning_rate": 5.768668484110054e-06,
      "loss": 0.3554,
      "step": 30193
    },
    {
      "epoch": 1.3856179156532513,
      "grad_norm": 0.5350527167320251,
      "learning_rate": 5.768426209995039e-06,
      "loss": 0.4428,
      "step": 30194
    },
    {
      "epoch": 1.3856638061585058,
      "grad_norm": 0.4820314943790436,
      "learning_rate": 5.768183934032199e-06,
      "loss": 0.4073,
      "step": 30195
    },
    {
      "epoch": 1.3857096966637603,
      "grad_norm": 0.48067232966423035,
      "learning_rate": 5.767941656222121e-06,
      "loss": 0.3594,
      "step": 30196
    },
    {
      "epoch": 1.3857555871690148,
      "grad_norm": 0.46545538306236267,
      "learning_rate": 5.7676993765653845e-06,
      "loss": 0.3498,
      "step": 30197
    },
    {
      "epoch": 1.3858014776742693,
      "grad_norm": 0.4139038622379303,
      "learning_rate": 5.767457095062573e-06,
      "loss": 0.2658,
      "step": 30198
    },
    {
      "epoch": 1.3858473681795236,
      "grad_norm": 0.4817889332771301,
      "learning_rate": 5.76721481171427e-06,
      "loss": 0.3786,
      "step": 30199
    },
    {
      "epoch": 1.385893258684778,
      "grad_norm": 0.4949606657028198,
      "learning_rate": 5.766972526521059e-06,
      "loss": 0.3678,
      "step": 30200
    },
    {
      "epoch": 1.3859391491900326,
      "grad_norm": 0.44682562351226807,
      "learning_rate": 5.766730239483519e-06,
      "loss": 0.2934,
      "step": 30201
    },
    {
      "epoch": 1.385985039695287,
      "grad_norm": 0.45298531651496887,
      "learning_rate": 5.766487950602236e-06,
      "loss": 0.3326,
      "step": 30202
    },
    {
      "epoch": 1.3860309302005416,
      "grad_norm": 0.4452357590198517,
      "learning_rate": 5.766245659877791e-06,
      "loss": 0.3343,
      "step": 30203
    },
    {
      "epoch": 1.3860768207057959,
      "grad_norm": 0.4115660488605499,
      "learning_rate": 5.766003367310767e-06,
      "loss": 0.298,
      "step": 30204
    },
    {
      "epoch": 1.3861227112110504,
      "grad_norm": 0.5188589692115784,
      "learning_rate": 5.7657610729017455e-06,
      "loss": 0.4691,
      "step": 30205
    },
    {
      "epoch": 1.3861686017163048,
      "grad_norm": 0.520616888999939,
      "learning_rate": 5.765518776651311e-06,
      "loss": 0.4486,
      "step": 30206
    },
    {
      "epoch": 1.3862144922215593,
      "grad_norm": 0.434212863445282,
      "learning_rate": 5.7652764785600464e-06,
      "loss": 0.3102,
      "step": 30207
    },
    {
      "epoch": 1.3862603827268138,
      "grad_norm": 0.4577532112598419,
      "learning_rate": 5.765034178628532e-06,
      "loss": 0.3422,
      "step": 30208
    },
    {
      "epoch": 1.3863062732320683,
      "grad_norm": 0.4571246802806854,
      "learning_rate": 5.764791876857354e-06,
      "loss": 0.3516,
      "step": 30209
    },
    {
      "epoch": 1.3863521637373228,
      "grad_norm": 0.5221338272094727,
      "learning_rate": 5.764549573247091e-06,
      "loss": 0.4845,
      "step": 30210
    },
    {
      "epoch": 1.3863980542425773,
      "grad_norm": 0.46580061316490173,
      "learning_rate": 5.764307267798328e-06,
      "loss": 0.3506,
      "step": 30211
    },
    {
      "epoch": 1.3864439447478316,
      "grad_norm": 0.4762353003025055,
      "learning_rate": 5.764064960511648e-06,
      "loss": 0.371,
      "step": 30212
    },
    {
      "epoch": 1.3864898352530861,
      "grad_norm": 0.4462132751941681,
      "learning_rate": 5.763822651387632e-06,
      "loss": 0.3251,
      "step": 30213
    },
    {
      "epoch": 1.3865357257583406,
      "grad_norm": 0.46403101086616516,
      "learning_rate": 5.763580340426864e-06,
      "loss": 0.3493,
      "step": 30214
    },
    {
      "epoch": 1.386581616263595,
      "grad_norm": 0.477291464805603,
      "learning_rate": 5.763338027629925e-06,
      "loss": 0.3676,
      "step": 30215
    },
    {
      "epoch": 1.3866275067688496,
      "grad_norm": 0.44286972284317017,
      "learning_rate": 5.7630957129974e-06,
      "loss": 0.3256,
      "step": 30216
    },
    {
      "epoch": 1.3866733972741039,
      "grad_norm": 0.4290679693222046,
      "learning_rate": 5.762853396529872e-06,
      "loss": 0.271,
      "step": 30217
    },
    {
      "epoch": 1.3867192877793584,
      "grad_norm": 0.46706992387771606,
      "learning_rate": 5.7626110782279224e-06,
      "loss": 0.3235,
      "step": 30218
    },
    {
      "epoch": 1.3867651782846129,
      "grad_norm": 0.4916428029537201,
      "learning_rate": 5.762368758092131e-06,
      "loss": 0.402,
      "step": 30219
    },
    {
      "epoch": 1.3868110687898674,
      "grad_norm": 0.44531404972076416,
      "learning_rate": 5.762126436123086e-06,
      "loss": 0.3371,
      "step": 30220
    },
    {
      "epoch": 1.3868569592951219,
      "grad_norm": 0.4945046603679657,
      "learning_rate": 5.761884112321365e-06,
      "loss": 0.422,
      "step": 30221
    },
    {
      "epoch": 1.3869028498003764,
      "grad_norm": 0.5105440020561218,
      "learning_rate": 5.761641786687555e-06,
      "loss": 0.4286,
      "step": 30222
    },
    {
      "epoch": 1.3869487403056309,
      "grad_norm": 0.48724180459976196,
      "learning_rate": 5.761399459222238e-06,
      "loss": 0.4184,
      "step": 30223
    },
    {
      "epoch": 1.3869946308108854,
      "grad_norm": 0.4847036302089691,
      "learning_rate": 5.761157129925992e-06,
      "loss": 0.3592,
      "step": 30224
    },
    {
      "epoch": 1.3870405213161396,
      "grad_norm": 0.44796520471572876,
      "learning_rate": 5.760914798799407e-06,
      "loss": 0.3212,
      "step": 30225
    },
    {
      "epoch": 1.3870864118213941,
      "grad_norm": 0.48685550689697266,
      "learning_rate": 5.7606724658430605e-06,
      "loss": 0.3882,
      "step": 30226
    },
    {
      "epoch": 1.3871323023266486,
      "grad_norm": 0.4488457143306732,
      "learning_rate": 5.760430131057535e-06,
      "loss": 0.2855,
      "step": 30227
    },
    {
      "epoch": 1.3871781928319031,
      "grad_norm": 0.5084857940673828,
      "learning_rate": 5.760187794443418e-06,
      "loss": 0.42,
      "step": 30228
    },
    {
      "epoch": 1.3872240833371574,
      "grad_norm": 0.47038644552230835,
      "learning_rate": 5.759945456001288e-06,
      "loss": 0.3737,
      "step": 30229
    },
    {
      "epoch": 1.387269973842412,
      "grad_norm": 0.5137220025062561,
      "learning_rate": 5.7597031157317285e-06,
      "loss": 0.4094,
      "step": 30230
    },
    {
      "epoch": 1.3873158643476664,
      "grad_norm": 0.4809383451938629,
      "learning_rate": 5.759460773635323e-06,
      "loss": 0.3597,
      "step": 30231
    },
    {
      "epoch": 1.387361754852921,
      "grad_norm": 0.4539553225040436,
      "learning_rate": 5.759218429712655e-06,
      "loss": 0.3278,
      "step": 30232
    },
    {
      "epoch": 1.3874076453581754,
      "grad_norm": 0.49070674180984497,
      "learning_rate": 5.758976083964305e-06,
      "loss": 0.3976,
      "step": 30233
    },
    {
      "epoch": 1.38745353586343,
      "grad_norm": 0.4861520528793335,
      "learning_rate": 5.758733736390859e-06,
      "loss": 0.3549,
      "step": 30234
    },
    {
      "epoch": 1.3874994263686844,
      "grad_norm": 0.4564046859741211,
      "learning_rate": 5.758491386992896e-06,
      "loss": 0.3725,
      "step": 30235
    },
    {
      "epoch": 1.387545316873939,
      "grad_norm": 0.6546860337257385,
      "learning_rate": 5.758249035771001e-06,
      "loss": 0.4455,
      "step": 30236
    },
    {
      "epoch": 1.3875912073791932,
      "grad_norm": 0.47364097833633423,
      "learning_rate": 5.7580066827257565e-06,
      "loss": 0.3232,
      "step": 30237
    },
    {
      "epoch": 1.3876370978844477,
      "grad_norm": 0.4762275815010071,
      "learning_rate": 5.757764327857745e-06,
      "loss": 0.3477,
      "step": 30238
    },
    {
      "epoch": 1.3876829883897022,
      "grad_norm": 0.49329712986946106,
      "learning_rate": 5.7575219711675504e-06,
      "loss": 0.4184,
      "step": 30239
    },
    {
      "epoch": 1.3877288788949567,
      "grad_norm": 0.47955426573753357,
      "learning_rate": 5.757279612655754e-06,
      "loss": 0.3294,
      "step": 30240
    },
    {
      "epoch": 1.3877747694002112,
      "grad_norm": 0.46467211842536926,
      "learning_rate": 5.757037252322939e-06,
      "loss": 0.3467,
      "step": 30241
    },
    {
      "epoch": 1.3878206599054654,
      "grad_norm": 0.48041290044784546,
      "learning_rate": 5.756794890169688e-06,
      "loss": 0.3194,
      "step": 30242
    },
    {
      "epoch": 1.38786655041072,
      "grad_norm": 0.5120606422424316,
      "learning_rate": 5.756552526196584e-06,
      "loss": 0.34,
      "step": 30243
    },
    {
      "epoch": 1.3879124409159744,
      "grad_norm": 0.4717693328857422,
      "learning_rate": 5.756310160404209e-06,
      "loss": 0.3314,
      "step": 30244
    },
    {
      "epoch": 1.387958331421229,
      "grad_norm": 0.5047916769981384,
      "learning_rate": 5.7560677927931505e-06,
      "loss": 0.3913,
      "step": 30245
    },
    {
      "epoch": 1.3880042219264834,
      "grad_norm": 0.44254550337791443,
      "learning_rate": 5.755825423363984e-06,
      "loss": 0.288,
      "step": 30246
    },
    {
      "epoch": 1.388050112431738,
      "grad_norm": 0.4101138114929199,
      "learning_rate": 5.7555830521172975e-06,
      "loss": 0.2498,
      "step": 30247
    },
    {
      "epoch": 1.3880960029369924,
      "grad_norm": 0.4936481714248657,
      "learning_rate": 5.755340679053671e-06,
      "loss": 0.4302,
      "step": 30248
    },
    {
      "epoch": 1.388141893442247,
      "grad_norm": 0.49934646487236023,
      "learning_rate": 5.75509830417369e-06,
      "loss": 0.4434,
      "step": 30249
    },
    {
      "epoch": 1.3881877839475012,
      "grad_norm": 0.6422379016876221,
      "learning_rate": 5.754855927477936e-06,
      "loss": 0.451,
      "step": 30250
    },
    {
      "epoch": 1.3882336744527557,
      "grad_norm": 0.4839049279689789,
      "learning_rate": 5.754613548966991e-06,
      "loss": 0.3668,
      "step": 30251
    },
    {
      "epoch": 1.3882795649580102,
      "grad_norm": 0.47374632954597473,
      "learning_rate": 5.7543711686414385e-06,
      "loss": 0.378,
      "step": 30252
    },
    {
      "epoch": 1.3883254554632647,
      "grad_norm": 0.48342493176460266,
      "learning_rate": 5.75412878650186e-06,
      "loss": 0.3828,
      "step": 30253
    },
    {
      "epoch": 1.3883713459685192,
      "grad_norm": 0.475362628698349,
      "learning_rate": 5.753886402548843e-06,
      "loss": 0.3243,
      "step": 30254
    },
    {
      "epoch": 1.3884172364737735,
      "grad_norm": 0.46324992179870605,
      "learning_rate": 5.753644016782964e-06,
      "loss": 0.3022,
      "step": 30255
    },
    {
      "epoch": 1.388463126979028,
      "grad_norm": 0.5234882235527039,
      "learning_rate": 5.7534016292048114e-06,
      "loss": 0.4761,
      "step": 30256
    },
    {
      "epoch": 1.3885090174842825,
      "grad_norm": 0.42796188592910767,
      "learning_rate": 5.753159239814965e-06,
      "loss": 0.3046,
      "step": 30257
    },
    {
      "epoch": 1.388554907989537,
      "grad_norm": 0.41637393832206726,
      "learning_rate": 5.752916848614007e-06,
      "loss": 0.2803,
      "step": 30258
    },
    {
      "epoch": 1.3886007984947915,
      "grad_norm": 0.46950602531433105,
      "learning_rate": 5.7526744556025235e-06,
      "loss": 0.3731,
      "step": 30259
    },
    {
      "epoch": 1.388646689000046,
      "grad_norm": 0.45282888412475586,
      "learning_rate": 5.752432060781093e-06,
      "loss": 0.3263,
      "step": 30260
    },
    {
      "epoch": 1.3886925795053005,
      "grad_norm": 0.4992215037345886,
      "learning_rate": 5.752189664150304e-06,
      "loss": 0.4165,
      "step": 30261
    },
    {
      "epoch": 1.388738470010555,
      "grad_norm": 0.4870789051055908,
      "learning_rate": 5.751947265710736e-06,
      "loss": 0.3835,
      "step": 30262
    },
    {
      "epoch": 1.3887843605158092,
      "grad_norm": 0.4718581736087799,
      "learning_rate": 5.751704865462969e-06,
      "loss": 0.3965,
      "step": 30263
    },
    {
      "epoch": 1.3888302510210637,
      "grad_norm": 0.4494836628437042,
      "learning_rate": 5.751462463407591e-06,
      "loss": 0.3158,
      "step": 30264
    },
    {
      "epoch": 1.3888761415263182,
      "grad_norm": 0.5015130043029785,
      "learning_rate": 5.751220059545183e-06,
      "loss": 0.3793,
      "step": 30265
    },
    {
      "epoch": 1.3889220320315727,
      "grad_norm": 0.4693112075328827,
      "learning_rate": 5.750977653876327e-06,
      "loss": 0.3912,
      "step": 30266
    },
    {
      "epoch": 1.388967922536827,
      "grad_norm": 0.46782997250556946,
      "learning_rate": 5.750735246401609e-06,
      "loss": 0.3707,
      "step": 30267
    },
    {
      "epoch": 1.3890138130420815,
      "grad_norm": 0.4616799056529999,
      "learning_rate": 5.750492837121608e-06,
      "loss": 0.3452,
      "step": 30268
    },
    {
      "epoch": 1.389059703547336,
      "grad_norm": 0.46908581256866455,
      "learning_rate": 5.750250426036909e-06,
      "loss": 0.3781,
      "step": 30269
    },
    {
      "epoch": 1.3891055940525905,
      "grad_norm": 0.46836769580841064,
      "learning_rate": 5.750008013148093e-06,
      "loss": 0.3188,
      "step": 30270
    },
    {
      "epoch": 1.389151484557845,
      "grad_norm": 0.44338929653167725,
      "learning_rate": 5.7497655984557456e-06,
      "loss": 0.3233,
      "step": 30271
    },
    {
      "epoch": 1.3891973750630995,
      "grad_norm": 0.5079687237739563,
      "learning_rate": 5.749523181960449e-06,
      "loss": 0.445,
      "step": 30272
    },
    {
      "epoch": 1.389243265568354,
      "grad_norm": 0.4805227518081665,
      "learning_rate": 5.7492807636627854e-06,
      "loss": 0.365,
      "step": 30273
    },
    {
      "epoch": 1.3892891560736085,
      "grad_norm": 0.4633045494556427,
      "learning_rate": 5.749038343563338e-06,
      "loss": 0.3541,
      "step": 30274
    },
    {
      "epoch": 1.3893350465788628,
      "grad_norm": 0.45519891381263733,
      "learning_rate": 5.74879592166269e-06,
      "loss": 0.3108,
      "step": 30275
    },
    {
      "epoch": 1.3893809370841173,
      "grad_norm": 0.4301750957965851,
      "learning_rate": 5.748553497961424e-06,
      "loss": 0.324,
      "step": 30276
    },
    {
      "epoch": 1.3894268275893717,
      "grad_norm": 0.4371340572834015,
      "learning_rate": 5.748311072460122e-06,
      "loss": 0.3569,
      "step": 30277
    },
    {
      "epoch": 1.3894727180946262,
      "grad_norm": 0.48944544792175293,
      "learning_rate": 5.748068645159371e-06,
      "loss": 0.3631,
      "step": 30278
    },
    {
      "epoch": 1.3895186085998807,
      "grad_norm": 0.5014194846153259,
      "learning_rate": 5.7478262160597485e-06,
      "loss": 0.34,
      "step": 30279
    },
    {
      "epoch": 1.389564499105135,
      "grad_norm": 0.5651270151138306,
      "learning_rate": 5.74758378516184e-06,
      "loss": 0.4017,
      "step": 30280
    },
    {
      "epoch": 1.3896103896103895,
      "grad_norm": 0.4608399271965027,
      "learning_rate": 5.747341352466228e-06,
      "loss": 0.341,
      "step": 30281
    },
    {
      "epoch": 1.389656280115644,
      "grad_norm": 0.42807790637016296,
      "learning_rate": 5.747098917973498e-06,
      "loss": 0.3006,
      "step": 30282
    },
    {
      "epoch": 1.3897021706208985,
      "grad_norm": 0.45664551854133606,
      "learning_rate": 5.74685648168423e-06,
      "loss": 0.3467,
      "step": 30283
    },
    {
      "epoch": 1.389748061126153,
      "grad_norm": 0.4479440152645111,
      "learning_rate": 5.746614043599007e-06,
      "loss": 0.3274,
      "step": 30284
    },
    {
      "epoch": 1.3897939516314075,
      "grad_norm": 0.4336516857147217,
      "learning_rate": 5.7463716037184146e-06,
      "loss": 0.2725,
      "step": 30285
    },
    {
      "epoch": 1.389839842136662,
      "grad_norm": 0.486752450466156,
      "learning_rate": 5.746129162043032e-06,
      "loss": 0.3679,
      "step": 30286
    },
    {
      "epoch": 1.3898857326419165,
      "grad_norm": 0.4881468713283539,
      "learning_rate": 5.7458867185734445e-06,
      "loss": 0.3937,
      "step": 30287
    },
    {
      "epoch": 1.3899316231471708,
      "grad_norm": 0.4904804527759552,
      "learning_rate": 5.745644273310236e-06,
      "loss": 0.3533,
      "step": 30288
    },
    {
      "epoch": 1.3899775136524253,
      "grad_norm": 0.4749709665775299,
      "learning_rate": 5.745401826253989e-06,
      "loss": 0.342,
      "step": 30289
    },
    {
      "epoch": 1.3900234041576798,
      "grad_norm": 0.45477306842803955,
      "learning_rate": 5.745159377405284e-06,
      "loss": 0.332,
      "step": 30290
    },
    {
      "epoch": 1.3900692946629343,
      "grad_norm": 0.4279290437698364,
      "learning_rate": 5.744916926764708e-06,
      "loss": 0.2988,
      "step": 30291
    },
    {
      "epoch": 1.3901151851681888,
      "grad_norm": 0.5339832305908203,
      "learning_rate": 5.744674474332839e-06,
      "loss": 0.5179,
      "step": 30292
    },
    {
      "epoch": 1.390161075673443,
      "grad_norm": 0.4888162314891815,
      "learning_rate": 5.744432020110265e-06,
      "loss": 0.4337,
      "step": 30293
    },
    {
      "epoch": 1.3902069661786975,
      "grad_norm": 0.4568067491054535,
      "learning_rate": 5.744189564097566e-06,
      "loss": 0.3553,
      "step": 30294
    },
    {
      "epoch": 1.390252856683952,
      "grad_norm": 0.45538386702537537,
      "learning_rate": 5.743947106295327e-06,
      "loss": 0.3066,
      "step": 30295
    },
    {
      "epoch": 1.3902987471892065,
      "grad_norm": 0.46293267607688904,
      "learning_rate": 5.743704646704129e-06,
      "loss": 0.3931,
      "step": 30296
    },
    {
      "epoch": 1.390344637694461,
      "grad_norm": 0.4840135872364044,
      "learning_rate": 5.7434621853245564e-06,
      "loss": 0.4031,
      "step": 30297
    },
    {
      "epoch": 1.3903905281997155,
      "grad_norm": 0.46634942293167114,
      "learning_rate": 5.743219722157192e-06,
      "loss": 0.3561,
      "step": 30298
    },
    {
      "epoch": 1.39043641870497,
      "grad_norm": 0.45144978165626526,
      "learning_rate": 5.742977257202619e-06,
      "loss": 0.3306,
      "step": 30299
    },
    {
      "epoch": 1.3904823092102245,
      "grad_norm": 0.48121321201324463,
      "learning_rate": 5.74273479046142e-06,
      "loss": 0.3892,
      "step": 30300
    },
    {
      "epoch": 1.3905281997154788,
      "grad_norm": 0.4638969600200653,
      "learning_rate": 5.7424923219341775e-06,
      "loss": 0.3483,
      "step": 30301
    },
    {
      "epoch": 1.3905740902207333,
      "grad_norm": 0.4213906228542328,
      "learning_rate": 5.742249851621476e-06,
      "loss": 0.2578,
      "step": 30302
    },
    {
      "epoch": 1.3906199807259878,
      "grad_norm": 0.45766401290893555,
      "learning_rate": 5.7420073795238975e-06,
      "loss": 0.3611,
      "step": 30303
    },
    {
      "epoch": 1.3906658712312423,
      "grad_norm": 0.4397094249725342,
      "learning_rate": 5.741764905642026e-06,
      "loss": 0.3303,
      "step": 30304
    },
    {
      "epoch": 1.3907117617364968,
      "grad_norm": 0.4559842348098755,
      "learning_rate": 5.741522429976445e-06,
      "loss": 0.3247,
      "step": 30305
    },
    {
      "epoch": 1.390757652241751,
      "grad_norm": 0.4601464569568634,
      "learning_rate": 5.741279952527734e-06,
      "loss": 0.3472,
      "step": 30306
    },
    {
      "epoch": 1.3908035427470056,
      "grad_norm": 0.4247780442237854,
      "learning_rate": 5.74103747329648e-06,
      "loss": 0.2693,
      "step": 30307
    },
    {
      "epoch": 1.39084943325226,
      "grad_norm": 0.4816422462463379,
      "learning_rate": 5.740794992283265e-06,
      "loss": 0.3354,
      "step": 30308
    },
    {
      "epoch": 1.3908953237575146,
      "grad_norm": 0.479301393032074,
      "learning_rate": 5.74055250948867e-06,
      "loss": 0.394,
      "step": 30309
    },
    {
      "epoch": 1.390941214262769,
      "grad_norm": 0.45645663142204285,
      "learning_rate": 5.740310024913283e-06,
      "loss": 0.3593,
      "step": 30310
    },
    {
      "epoch": 1.3909871047680236,
      "grad_norm": 0.48338621854782104,
      "learning_rate": 5.740067538557683e-06,
      "loss": 0.3991,
      "step": 30311
    },
    {
      "epoch": 1.391032995273278,
      "grad_norm": 0.5078062415122986,
      "learning_rate": 5.739825050422453e-06,
      "loss": 0.4444,
      "step": 30312
    },
    {
      "epoch": 1.3910788857785326,
      "grad_norm": 0.45488956570625305,
      "learning_rate": 5.739582560508179e-06,
      "loss": 0.3236,
      "step": 30313
    },
    {
      "epoch": 1.3911247762837868,
      "grad_norm": 0.47081366181373596,
      "learning_rate": 5.73934006881544e-06,
      "loss": 0.3398,
      "step": 30314
    },
    {
      "epoch": 1.3911706667890413,
      "grad_norm": 0.42333802580833435,
      "learning_rate": 5.739097575344823e-06,
      "loss": 0.3013,
      "step": 30315
    },
    {
      "epoch": 1.3912165572942958,
      "grad_norm": 0.4540907144546509,
      "learning_rate": 5.73885508009691e-06,
      "loss": 0.3266,
      "step": 30316
    },
    {
      "epoch": 1.3912624477995503,
      "grad_norm": 0.4784020185470581,
      "learning_rate": 5.738612583072282e-06,
      "loss": 0.3964,
      "step": 30317
    },
    {
      "epoch": 1.3913083383048046,
      "grad_norm": 0.47597798705101013,
      "learning_rate": 5.738370084271525e-06,
      "loss": 0.3313,
      "step": 30318
    },
    {
      "epoch": 1.391354228810059,
      "grad_norm": 0.4556713104248047,
      "learning_rate": 5.738127583695223e-06,
      "loss": 0.3367,
      "step": 30319
    },
    {
      "epoch": 1.3914001193153136,
      "grad_norm": 0.46421167254447937,
      "learning_rate": 5.737885081343953e-06,
      "loss": 0.34,
      "step": 30320
    },
    {
      "epoch": 1.391446009820568,
      "grad_norm": 0.46209293603897095,
      "learning_rate": 5.7376425772183055e-06,
      "loss": 0.3206,
      "step": 30321
    },
    {
      "epoch": 1.3914919003258226,
      "grad_norm": 0.4941738247871399,
      "learning_rate": 5.737400071318859e-06,
      "loss": 0.3648,
      "step": 30322
    },
    {
      "epoch": 1.391537790831077,
      "grad_norm": 0.47378218173980713,
      "learning_rate": 5.737157563646198e-06,
      "loss": 0.3431,
      "step": 30323
    },
    {
      "epoch": 1.3915836813363316,
      "grad_norm": 0.4601399600505829,
      "learning_rate": 5.736915054200907e-06,
      "loss": 0.372,
      "step": 30324
    },
    {
      "epoch": 1.391629571841586,
      "grad_norm": 0.41920557618141174,
      "learning_rate": 5.7366725429835655e-06,
      "loss": 0.2705,
      "step": 30325
    },
    {
      "epoch": 1.3916754623468404,
      "grad_norm": 0.443963885307312,
      "learning_rate": 5.736430029994761e-06,
      "loss": 0.3202,
      "step": 30326
    },
    {
      "epoch": 1.3917213528520949,
      "grad_norm": 0.4723348021507263,
      "learning_rate": 5.7361875152350745e-06,
      "loss": 0.391,
      "step": 30327
    },
    {
      "epoch": 1.3917672433573494,
      "grad_norm": 0.45168864727020264,
      "learning_rate": 5.735944998705088e-06,
      "loss": 0.3293,
      "step": 30328
    },
    {
      "epoch": 1.3918131338626039,
      "grad_norm": 0.4439443349838257,
      "learning_rate": 5.735702480405387e-06,
      "loss": 0.3344,
      "step": 30329
    },
    {
      "epoch": 1.3918590243678584,
      "grad_norm": 0.453037291765213,
      "learning_rate": 5.735459960336555e-06,
      "loss": 0.3353,
      "step": 30330
    },
    {
      "epoch": 1.3919049148731126,
      "grad_norm": 0.5007305145263672,
      "learning_rate": 5.735217438499171e-06,
      "loss": 0.3852,
      "step": 30331
    },
    {
      "epoch": 1.3919508053783671,
      "grad_norm": 0.4656926393508911,
      "learning_rate": 5.7349749148938235e-06,
      "loss": 0.3038,
      "step": 30332
    },
    {
      "epoch": 1.3919966958836216,
      "grad_norm": 0.495773047208786,
      "learning_rate": 5.734732389521094e-06,
      "loss": 0.3622,
      "step": 30333
    },
    {
      "epoch": 1.3920425863888761,
      "grad_norm": 0.4706609845161438,
      "learning_rate": 5.734489862381563e-06,
      "loss": 0.3788,
      "step": 30334
    },
    {
      "epoch": 1.3920884768941306,
      "grad_norm": 0.47529056668281555,
      "learning_rate": 5.734247333475815e-06,
      "loss": 0.3956,
      "step": 30335
    },
    {
      "epoch": 1.3921343673993851,
      "grad_norm": 0.4827890694141388,
      "learning_rate": 5.734004802804436e-06,
      "loss": 0.379,
      "step": 30336
    },
    {
      "epoch": 1.3921802579046396,
      "grad_norm": 0.4604039192199707,
      "learning_rate": 5.733762270368006e-06,
      "loss": 0.3466,
      "step": 30337
    },
    {
      "epoch": 1.3922261484098941,
      "grad_norm": 0.48655077815055847,
      "learning_rate": 5.733519736167109e-06,
      "loss": 0.384,
      "step": 30338
    },
    {
      "epoch": 1.3922720389151484,
      "grad_norm": 0.4173821806907654,
      "learning_rate": 5.733277200202329e-06,
      "loss": 0.281,
      "step": 30339
    },
    {
      "epoch": 1.3923179294204029,
      "grad_norm": 0.814814031124115,
      "learning_rate": 5.733034662474249e-06,
      "loss": 0.3498,
      "step": 30340
    },
    {
      "epoch": 1.3923638199256574,
      "grad_norm": 0.4419386088848114,
      "learning_rate": 5.732792122983452e-06,
      "loss": 0.3552,
      "step": 30341
    },
    {
      "epoch": 1.3924097104309119,
      "grad_norm": 0.4876553416252136,
      "learning_rate": 5.732549581730521e-06,
      "loss": 0.3726,
      "step": 30342
    },
    {
      "epoch": 1.3924556009361664,
      "grad_norm": 0.5092206001281738,
      "learning_rate": 5.732307038716041e-06,
      "loss": 0.4212,
      "step": 30343
    },
    {
      "epoch": 1.3925014914414207,
      "grad_norm": 0.41890591382980347,
      "learning_rate": 5.732064493940592e-06,
      "loss": 0.2814,
      "step": 30344
    },
    {
      "epoch": 1.3925473819466752,
      "grad_norm": 0.5132268667221069,
      "learning_rate": 5.731821947404759e-06,
      "loss": 0.4234,
      "step": 30345
    },
    {
      "epoch": 1.3925932724519297,
      "grad_norm": 0.4383687973022461,
      "learning_rate": 5.731579399109124e-06,
      "loss": 0.2965,
      "step": 30346
    },
    {
      "epoch": 1.3926391629571842,
      "grad_norm": 0.5155216455459595,
      "learning_rate": 5.7313368490542735e-06,
      "loss": 0.439,
      "step": 30347
    },
    {
      "epoch": 1.3926850534624386,
      "grad_norm": 0.47807371616363525,
      "learning_rate": 5.731094297240789e-06,
      "loss": 0.3535,
      "step": 30348
    },
    {
      "epoch": 1.3927309439676931,
      "grad_norm": 0.4623567461967468,
      "learning_rate": 5.730851743669251e-06,
      "loss": 0.3075,
      "step": 30349
    },
    {
      "epoch": 1.3927768344729476,
      "grad_norm": 0.5822451114654541,
      "learning_rate": 5.730609188340247e-06,
      "loss": 0.4935,
      "step": 30350
    },
    {
      "epoch": 1.3928227249782021,
      "grad_norm": 0.4866446852684021,
      "learning_rate": 5.730366631254358e-06,
      "loss": 0.3808,
      "step": 30351
    },
    {
      "epoch": 1.3928686154834564,
      "grad_norm": 0.4441816806793213,
      "learning_rate": 5.7301240724121676e-06,
      "loss": 0.3067,
      "step": 30352
    },
    {
      "epoch": 1.392914505988711,
      "grad_norm": 0.4770102798938751,
      "learning_rate": 5.729881511814261e-06,
      "loss": 0.3764,
      "step": 30353
    },
    {
      "epoch": 1.3929603964939654,
      "grad_norm": 0.5705597996711731,
      "learning_rate": 5.729638949461218e-06,
      "loss": 0.4614,
      "step": 30354
    },
    {
      "epoch": 1.39300628699922,
      "grad_norm": 0.5419921875,
      "learning_rate": 5.729396385353622e-06,
      "loss": 0.4295,
      "step": 30355
    },
    {
      "epoch": 1.3930521775044742,
      "grad_norm": 0.4608101546764374,
      "learning_rate": 5.729153819492061e-06,
      "loss": 0.3168,
      "step": 30356
    },
    {
      "epoch": 1.3930980680097287,
      "grad_norm": 0.4511256217956543,
      "learning_rate": 5.728911251877114e-06,
      "loss": 0.3604,
      "step": 30357
    },
    {
      "epoch": 1.3931439585149832,
      "grad_norm": 0.46918269991874695,
      "learning_rate": 5.728668682509366e-06,
      "loss": 0.3906,
      "step": 30358
    },
    {
      "epoch": 1.3931898490202377,
      "grad_norm": 0.4779897630214691,
      "learning_rate": 5.728426111389399e-06,
      "loss": 0.3544,
      "step": 30359
    },
    {
      "epoch": 1.3932357395254922,
      "grad_norm": 0.4973537027835846,
      "learning_rate": 5.728183538517798e-06,
      "loss": 0.3654,
      "step": 30360
    },
    {
      "epoch": 1.3932816300307467,
      "grad_norm": 0.4682236909866333,
      "learning_rate": 5.727940963895145e-06,
      "loss": 0.3087,
      "step": 30361
    },
    {
      "epoch": 1.3933275205360012,
      "grad_norm": 0.4664536118507385,
      "learning_rate": 5.727698387522023e-06,
      "loss": 0.352,
      "step": 30362
    },
    {
      "epoch": 1.3933734110412557,
      "grad_norm": 0.4596472680568695,
      "learning_rate": 5.727455809399015e-06,
      "loss": 0.3148,
      "step": 30363
    },
    {
      "epoch": 1.39341930154651,
      "grad_norm": 0.433256059885025,
      "learning_rate": 5.727213229526708e-06,
      "loss": 0.3233,
      "step": 30364
    },
    {
      "epoch": 1.3934651920517644,
      "grad_norm": 0.46075862646102905,
      "learning_rate": 5.726970647905682e-06,
      "loss": 0.3163,
      "step": 30365
    },
    {
      "epoch": 1.393511082557019,
      "grad_norm": 0.44754964113235474,
      "learning_rate": 5.72672806453652e-06,
      "loss": 0.3334,
      "step": 30366
    },
    {
      "epoch": 1.3935569730622734,
      "grad_norm": 0.4604916572570801,
      "learning_rate": 5.726485479419808e-06,
      "loss": 0.3162,
      "step": 30367
    },
    {
      "epoch": 1.393602863567528,
      "grad_norm": 0.43873023986816406,
      "learning_rate": 5.726242892556125e-06,
      "loss": 0.2928,
      "step": 30368
    },
    {
      "epoch": 1.3936487540727822,
      "grad_norm": 0.4664902091026306,
      "learning_rate": 5.72600030394606e-06,
      "loss": 0.3309,
      "step": 30369
    },
    {
      "epoch": 1.3936946445780367,
      "grad_norm": 0.4777691662311554,
      "learning_rate": 5.725757713590192e-06,
      "loss": 0.3391,
      "step": 30370
    },
    {
      "epoch": 1.3937405350832912,
      "grad_norm": 0.4571942985057831,
      "learning_rate": 5.725515121489106e-06,
      "loss": 0.3316,
      "step": 30371
    },
    {
      "epoch": 1.3937864255885457,
      "grad_norm": 0.4715847074985504,
      "learning_rate": 5.725272527643384e-06,
      "loss": 0.3414,
      "step": 30372
    },
    {
      "epoch": 1.3938323160938002,
      "grad_norm": 0.4987230896949768,
      "learning_rate": 5.725029932053612e-06,
      "loss": 0.3017,
      "step": 30373
    },
    {
      "epoch": 1.3938782065990547,
      "grad_norm": 0.4543333947658539,
      "learning_rate": 5.724787334720371e-06,
      "loss": 0.3415,
      "step": 30374
    },
    {
      "epoch": 1.3939240971043092,
      "grad_norm": 0.47051939368247986,
      "learning_rate": 5.724544735644247e-06,
      "loss": 0.3669,
      "step": 30375
    },
    {
      "epoch": 1.3939699876095637,
      "grad_norm": 0.46173739433288574,
      "learning_rate": 5.72430213482582e-06,
      "loss": 0.3717,
      "step": 30376
    },
    {
      "epoch": 1.394015878114818,
      "grad_norm": 0.44817879796028137,
      "learning_rate": 5.7240595322656746e-06,
      "loss": 0.3331,
      "step": 30377
    },
    {
      "epoch": 1.3940617686200725,
      "grad_norm": 0.5127765536308289,
      "learning_rate": 5.7238169279643955e-06,
      "loss": 0.416,
      "step": 30378
    },
    {
      "epoch": 1.394107659125327,
      "grad_norm": 0.45980104804039,
      "learning_rate": 5.723574321922564e-06,
      "loss": 0.3607,
      "step": 30379
    },
    {
      "epoch": 1.3941535496305815,
      "grad_norm": 0.41525402665138245,
      "learning_rate": 5.723331714140765e-06,
      "loss": 0.2849,
      "step": 30380
    },
    {
      "epoch": 1.394199440135836,
      "grad_norm": 0.4880722463130951,
      "learning_rate": 5.723089104619583e-06,
      "loss": 0.3629,
      "step": 30381
    },
    {
      "epoch": 1.3942453306410902,
      "grad_norm": 0.5014582276344299,
      "learning_rate": 5.722846493359598e-06,
      "loss": 0.4393,
      "step": 30382
    },
    {
      "epoch": 1.3942912211463447,
      "grad_norm": 0.47979748249053955,
      "learning_rate": 5.722603880361396e-06,
      "loss": 0.3569,
      "step": 30383
    },
    {
      "epoch": 1.3943371116515992,
      "grad_norm": 0.4602370262145996,
      "learning_rate": 5.72236126562556e-06,
      "loss": 0.3135,
      "step": 30384
    },
    {
      "epoch": 1.3943830021568537,
      "grad_norm": 0.434594064950943,
      "learning_rate": 5.722118649152673e-06,
      "loss": 0.2995,
      "step": 30385
    },
    {
      "epoch": 1.3944288926621082,
      "grad_norm": 0.47161078453063965,
      "learning_rate": 5.7218760309433185e-06,
      "loss": 0.4037,
      "step": 30386
    },
    {
      "epoch": 1.3944747831673627,
      "grad_norm": 0.430989533662796,
      "learning_rate": 5.7216334109980795e-06,
      "loss": 0.2803,
      "step": 30387
    },
    {
      "epoch": 1.3945206736726172,
      "grad_norm": 0.5023766160011292,
      "learning_rate": 5.72139078931754e-06,
      "loss": 0.4191,
      "step": 30388
    },
    {
      "epoch": 1.3945665641778717,
      "grad_norm": 0.46908140182495117,
      "learning_rate": 5.7211481659022835e-06,
      "loss": 0.3875,
      "step": 30389
    },
    {
      "epoch": 1.394612454683126,
      "grad_norm": 0.4324429929256439,
      "learning_rate": 5.7209055407528945e-06,
      "loss": 0.3171,
      "step": 30390
    },
    {
      "epoch": 1.3946583451883805,
      "grad_norm": 0.4551837146282196,
      "learning_rate": 5.720662913869954e-06,
      "loss": 0.3637,
      "step": 30391
    },
    {
      "epoch": 1.394704235693635,
      "grad_norm": 0.5051344037055969,
      "learning_rate": 5.720420285254047e-06,
      "loss": 0.2904,
      "step": 30392
    },
    {
      "epoch": 1.3947501261988895,
      "grad_norm": 0.4744405448436737,
      "learning_rate": 5.720177654905756e-06,
      "loss": 0.424,
      "step": 30393
    },
    {
      "epoch": 1.394796016704144,
      "grad_norm": 0.46851974725723267,
      "learning_rate": 5.719935022825666e-06,
      "loss": 0.3601,
      "step": 30394
    },
    {
      "epoch": 1.3948419072093983,
      "grad_norm": 0.4627271890640259,
      "learning_rate": 5.719692389014359e-06,
      "loss": 0.3651,
      "step": 30395
    },
    {
      "epoch": 1.3948877977146528,
      "grad_norm": 0.49190598726272583,
      "learning_rate": 5.7194497534724184e-06,
      "loss": 0.3825,
      "step": 30396
    },
    {
      "epoch": 1.3949336882199073,
      "grad_norm": 0.43936887383461,
      "learning_rate": 5.7192071162004285e-06,
      "loss": 0.3002,
      "step": 30397
    },
    {
      "epoch": 1.3949795787251618,
      "grad_norm": 0.44454583525657654,
      "learning_rate": 5.718964477198973e-06,
      "loss": 0.3083,
      "step": 30398
    },
    {
      "epoch": 1.3950254692304163,
      "grad_norm": 0.41476377844810486,
      "learning_rate": 5.718721836468634e-06,
      "loss": 0.3055,
      "step": 30399
    },
    {
      "epoch": 1.3950713597356708,
      "grad_norm": 0.5318158268928528,
      "learning_rate": 5.718479194009996e-06,
      "loss": 0.421,
      "step": 30400
    },
    {
      "epoch": 1.3951172502409253,
      "grad_norm": 0.47496166825294495,
      "learning_rate": 5.718236549823642e-06,
      "loss": 0.3457,
      "step": 30401
    },
    {
      "epoch": 1.3951631407461798,
      "grad_norm": 0.49589404463768005,
      "learning_rate": 5.717993903910157e-06,
      "loss": 0.394,
      "step": 30402
    },
    {
      "epoch": 1.395209031251434,
      "grad_norm": 0.537624180316925,
      "learning_rate": 5.717751256270123e-06,
      "loss": 0.4489,
      "step": 30403
    },
    {
      "epoch": 1.3952549217566885,
      "grad_norm": 0.46196869015693665,
      "learning_rate": 5.717508606904124e-06,
      "loss": 0.322,
      "step": 30404
    },
    {
      "epoch": 1.395300812261943,
      "grad_norm": 0.4932910203933716,
      "learning_rate": 5.717265955812742e-06,
      "loss": 0.4482,
      "step": 30405
    },
    {
      "epoch": 1.3953467027671975,
      "grad_norm": 0.46956688165664673,
      "learning_rate": 5.717023302996563e-06,
      "loss": 0.3515,
      "step": 30406
    },
    {
      "epoch": 1.3953925932724518,
      "grad_norm": 0.462916761636734,
      "learning_rate": 5.716780648456167e-06,
      "loss": 0.3531,
      "step": 30407
    },
    {
      "epoch": 1.3954384837777063,
      "grad_norm": 0.4539948105812073,
      "learning_rate": 5.716537992192143e-06,
      "loss": 0.3351,
      "step": 30408
    },
    {
      "epoch": 1.3954843742829608,
      "grad_norm": 0.45970776677131653,
      "learning_rate": 5.71629533420507e-06,
      "loss": 0.3227,
      "step": 30409
    },
    {
      "epoch": 1.3955302647882153,
      "grad_norm": 0.48853880167007446,
      "learning_rate": 5.7160526744955315e-06,
      "loss": 0.3674,
      "step": 30410
    },
    {
      "epoch": 1.3955761552934698,
      "grad_norm": 0.43366432189941406,
      "learning_rate": 5.715810013064112e-06,
      "loss": 0.3086,
      "step": 30411
    },
    {
      "epoch": 1.3956220457987243,
      "grad_norm": 0.4934180676937103,
      "learning_rate": 5.715567349911398e-06,
      "loss": 0.405,
      "step": 30412
    },
    {
      "epoch": 1.3956679363039788,
      "grad_norm": 0.4571032226085663,
      "learning_rate": 5.71532468503797e-06,
      "loss": 0.3462,
      "step": 30413
    },
    {
      "epoch": 1.3957138268092333,
      "grad_norm": 0.4463425874710083,
      "learning_rate": 5.71508201844441e-06,
      "loss": 0.3429,
      "step": 30414
    },
    {
      "epoch": 1.3957597173144876,
      "grad_norm": 0.48190680146217346,
      "learning_rate": 5.714839350131305e-06,
      "loss": 0.3521,
      "step": 30415
    },
    {
      "epoch": 1.395805607819742,
      "grad_norm": 0.48486289381980896,
      "learning_rate": 5.714596680099236e-06,
      "loss": 0.336,
      "step": 30416
    },
    {
      "epoch": 1.3958514983249966,
      "grad_norm": 0.48747485876083374,
      "learning_rate": 5.714354008348787e-06,
      "loss": 0.411,
      "step": 30417
    },
    {
      "epoch": 1.395897388830251,
      "grad_norm": 0.4783124327659607,
      "learning_rate": 5.714111334880543e-06,
      "loss": 0.3824,
      "step": 30418
    },
    {
      "epoch": 1.3959432793355055,
      "grad_norm": 0.4417360723018646,
      "learning_rate": 5.713868659695087e-06,
      "loss": 0.3566,
      "step": 30419
    },
    {
      "epoch": 1.3959891698407598,
      "grad_norm": 0.4791170358657837,
      "learning_rate": 5.713625982793001e-06,
      "loss": 0.3778,
      "step": 30420
    },
    {
      "epoch": 1.3960350603460143,
      "grad_norm": 0.4776018559932709,
      "learning_rate": 5.713383304174871e-06,
      "loss": 0.3683,
      "step": 30421
    },
    {
      "epoch": 1.3960809508512688,
      "grad_norm": 0.48346200585365295,
      "learning_rate": 5.713140623841278e-06,
      "loss": 0.312,
      "step": 30422
    },
    {
      "epoch": 1.3961268413565233,
      "grad_norm": 0.48635196685791016,
      "learning_rate": 5.712897941792808e-06,
      "loss": 0.3903,
      "step": 30423
    },
    {
      "epoch": 1.3961727318617778,
      "grad_norm": 0.49332302808761597,
      "learning_rate": 5.712655258030043e-06,
      "loss": 0.4068,
      "step": 30424
    },
    {
      "epoch": 1.3962186223670323,
      "grad_norm": 0.4469403624534607,
      "learning_rate": 5.712412572553567e-06,
      "loss": 0.3602,
      "step": 30425
    },
    {
      "epoch": 1.3962645128722868,
      "grad_norm": 0.47400718927383423,
      "learning_rate": 5.712169885363964e-06,
      "loss": 0.375,
      "step": 30426
    },
    {
      "epoch": 1.3963104033775413,
      "grad_norm": 0.43531715869903564,
      "learning_rate": 5.711927196461816e-06,
      "loss": 0.312,
      "step": 30427
    },
    {
      "epoch": 1.3963562938827956,
      "grad_norm": 0.4611225128173828,
      "learning_rate": 5.711684505847709e-06,
      "loss": 0.3157,
      "step": 30428
    },
    {
      "epoch": 1.39640218438805,
      "grad_norm": 0.444152295589447,
      "learning_rate": 5.711441813522225e-06,
      "loss": 0.3063,
      "step": 30429
    },
    {
      "epoch": 1.3964480748933046,
      "grad_norm": 0.4856477975845337,
      "learning_rate": 5.711199119485948e-06,
      "loss": 0.3835,
      "step": 30430
    },
    {
      "epoch": 1.396493965398559,
      "grad_norm": 0.4897124767303467,
      "learning_rate": 5.710956423739462e-06,
      "loss": 0.3477,
      "step": 30431
    },
    {
      "epoch": 1.3965398559038136,
      "grad_norm": 0.44401347637176514,
      "learning_rate": 5.71071372628335e-06,
      "loss": 0.2994,
      "step": 30432
    },
    {
      "epoch": 1.3965857464090679,
      "grad_norm": 0.4477541744709015,
      "learning_rate": 5.710471027118195e-06,
      "loss": 0.317,
      "step": 30433
    },
    {
      "epoch": 1.3966316369143223,
      "grad_norm": 0.4925900995731354,
      "learning_rate": 5.710228326244581e-06,
      "loss": 0.3822,
      "step": 30434
    },
    {
      "epoch": 1.3966775274195768,
      "grad_norm": 0.4776746332645416,
      "learning_rate": 5.709985623663094e-06,
      "loss": 0.406,
      "step": 30435
    },
    {
      "epoch": 1.3967234179248313,
      "grad_norm": 0.4376874268054962,
      "learning_rate": 5.709742919374315e-06,
      "loss": 0.2961,
      "step": 30436
    },
    {
      "epoch": 1.3967693084300858,
      "grad_norm": 0.4580034911632538,
      "learning_rate": 5.709500213378827e-06,
      "loss": 0.3446,
      "step": 30437
    },
    {
      "epoch": 1.3968151989353403,
      "grad_norm": 0.43879827857017517,
      "learning_rate": 5.709257505677217e-06,
      "loss": 0.3004,
      "step": 30438
    },
    {
      "epoch": 1.3968610894405948,
      "grad_norm": 0.450599730014801,
      "learning_rate": 5.709014796270065e-06,
      "loss": 0.3382,
      "step": 30439
    },
    {
      "epoch": 1.3969069799458493,
      "grad_norm": 0.45005470514297485,
      "learning_rate": 5.708772085157956e-06,
      "loss": 0.3343,
      "step": 30440
    },
    {
      "epoch": 1.3969528704511036,
      "grad_norm": 0.47958171367645264,
      "learning_rate": 5.708529372341476e-06,
      "loss": 0.3449,
      "step": 30441
    },
    {
      "epoch": 1.396998760956358,
      "grad_norm": 0.48262518644332886,
      "learning_rate": 5.708286657821204e-06,
      "loss": 0.4306,
      "step": 30442
    },
    {
      "epoch": 1.3970446514616126,
      "grad_norm": 0.46672382950782776,
      "learning_rate": 5.708043941597728e-06,
      "loss": 0.2987,
      "step": 30443
    },
    {
      "epoch": 1.397090541966867,
      "grad_norm": 0.45540308952331543,
      "learning_rate": 5.707801223671627e-06,
      "loss": 0.3599,
      "step": 30444
    },
    {
      "epoch": 1.3971364324721214,
      "grad_norm": 0.4659494161605835,
      "learning_rate": 5.707558504043489e-06,
      "loss": 0.3434,
      "step": 30445
    },
    {
      "epoch": 1.3971823229773759,
      "grad_norm": 0.4493085443973541,
      "learning_rate": 5.707315782713898e-06,
      "loss": 0.3248,
      "step": 30446
    },
    {
      "epoch": 1.3972282134826304,
      "grad_norm": 0.4677808880805969,
      "learning_rate": 5.707073059683433e-06,
      "loss": 0.3942,
      "step": 30447
    },
    {
      "epoch": 1.3972741039878849,
      "grad_norm": 0.4951171278953552,
      "learning_rate": 5.706830334952681e-06,
      "loss": 0.3819,
      "step": 30448
    },
    {
      "epoch": 1.3973199944931394,
      "grad_norm": 0.5140604972839355,
      "learning_rate": 5.706587608522227e-06,
      "loss": 0.4249,
      "step": 30449
    },
    {
      "epoch": 1.3973658849983939,
      "grad_norm": 0.4192191958427429,
      "learning_rate": 5.70634488039265e-06,
      "loss": 0.2682,
      "step": 30450
    },
    {
      "epoch": 1.3974117755036484,
      "grad_norm": 0.451386421918869,
      "learning_rate": 5.706102150564539e-06,
      "loss": 0.328,
      "step": 30451
    },
    {
      "epoch": 1.3974576660089029,
      "grad_norm": 0.4777073264122009,
      "learning_rate": 5.705859419038475e-06,
      "loss": 0.3629,
      "step": 30452
    },
    {
      "epoch": 1.3975035565141571,
      "grad_norm": 0.48843997716903687,
      "learning_rate": 5.705616685815041e-06,
      "loss": 0.3306,
      "step": 30453
    },
    {
      "epoch": 1.3975494470194116,
      "grad_norm": 0.47370651364326477,
      "learning_rate": 5.70537395089482e-06,
      "loss": 0.3506,
      "step": 30454
    },
    {
      "epoch": 1.3975953375246661,
      "grad_norm": 0.48216167092323303,
      "learning_rate": 5.7051312142784e-06,
      "loss": 0.3271,
      "step": 30455
    },
    {
      "epoch": 1.3976412280299206,
      "grad_norm": 0.46302226185798645,
      "learning_rate": 5.7048884759663604e-06,
      "loss": 0.3428,
      "step": 30456
    },
    {
      "epoch": 1.3976871185351751,
      "grad_norm": 0.4696814715862274,
      "learning_rate": 5.704645735959288e-06,
      "loss": 0.3673,
      "step": 30457
    },
    {
      "epoch": 1.3977330090404294,
      "grad_norm": 0.4597233235836029,
      "learning_rate": 5.704402994257764e-06,
      "loss": 0.3265,
      "step": 30458
    },
    {
      "epoch": 1.397778899545684,
      "grad_norm": 0.476111501455307,
      "learning_rate": 5.704160250862372e-06,
      "loss": 0.4011,
      "step": 30459
    },
    {
      "epoch": 1.3978247900509384,
      "grad_norm": 0.44585949182510376,
      "learning_rate": 5.7039175057736985e-06,
      "loss": 0.3282,
      "step": 30460
    },
    {
      "epoch": 1.397870680556193,
      "grad_norm": 0.4638659358024597,
      "learning_rate": 5.703674758992324e-06,
      "loss": 0.3734,
      "step": 30461
    },
    {
      "epoch": 1.3979165710614474,
      "grad_norm": 0.4854170083999634,
      "learning_rate": 5.703432010518836e-06,
      "loss": 0.4127,
      "step": 30462
    },
    {
      "epoch": 1.397962461566702,
      "grad_norm": 0.44458135962486267,
      "learning_rate": 5.7031892603538144e-06,
      "loss": 0.3023,
      "step": 30463
    },
    {
      "epoch": 1.3980083520719564,
      "grad_norm": 0.4612308740615845,
      "learning_rate": 5.702946508497844e-06,
      "loss": 0.3905,
      "step": 30464
    },
    {
      "epoch": 1.398054242577211,
      "grad_norm": 0.5073607563972473,
      "learning_rate": 5.7027037549515106e-06,
      "loss": 0.4023,
      "step": 30465
    },
    {
      "epoch": 1.3981001330824652,
      "grad_norm": 0.48198723793029785,
      "learning_rate": 5.702460999715396e-06,
      "loss": 0.3604,
      "step": 30466
    },
    {
      "epoch": 1.3981460235877197,
      "grad_norm": 0.4853844940662384,
      "learning_rate": 5.702218242790084e-06,
      "loss": 0.3888,
      "step": 30467
    },
    {
      "epoch": 1.3981919140929742,
      "grad_norm": 0.5036321878433228,
      "learning_rate": 5.7019754841761585e-06,
      "loss": 0.4339,
      "step": 30468
    },
    {
      "epoch": 1.3982378045982287,
      "grad_norm": 0.48087015748023987,
      "learning_rate": 5.701732723874205e-06,
      "loss": 0.3836,
      "step": 30469
    },
    {
      "epoch": 1.3982836951034832,
      "grad_norm": 0.45086729526519775,
      "learning_rate": 5.701489961884804e-06,
      "loss": 0.3054,
      "step": 30470
    },
    {
      "epoch": 1.3983295856087374,
      "grad_norm": 0.46627333760261536,
      "learning_rate": 5.701247198208543e-06,
      "loss": 0.4083,
      "step": 30471
    },
    {
      "epoch": 1.398375476113992,
      "grad_norm": 0.5135111808776855,
      "learning_rate": 5.701004432846002e-06,
      "loss": 0.4369,
      "step": 30472
    },
    {
      "epoch": 1.3984213666192464,
      "grad_norm": 0.47604984045028687,
      "learning_rate": 5.700761665797767e-06,
      "loss": 0.3637,
      "step": 30473
    },
    {
      "epoch": 1.398467257124501,
      "grad_norm": 0.47560882568359375,
      "learning_rate": 5.700518897064423e-06,
      "loss": 0.3434,
      "step": 30474
    },
    {
      "epoch": 1.3985131476297554,
      "grad_norm": 0.4974267780780792,
      "learning_rate": 5.70027612664655e-06,
      "loss": 0.4142,
      "step": 30475
    },
    {
      "epoch": 1.39855903813501,
      "grad_norm": 0.47538191080093384,
      "learning_rate": 5.700033354544734e-06,
      "loss": 0.3365,
      "step": 30476
    },
    {
      "epoch": 1.3986049286402644,
      "grad_norm": 0.5176913142204285,
      "learning_rate": 5.69979058075956e-06,
      "loss": 0.3822,
      "step": 30477
    },
    {
      "epoch": 1.398650819145519,
      "grad_norm": 0.45130783319473267,
      "learning_rate": 5.69954780529161e-06,
      "loss": 0.332,
      "step": 30478
    },
    {
      "epoch": 1.3986967096507732,
      "grad_norm": 0.47908908128738403,
      "learning_rate": 5.699305028141469e-06,
      "loss": 0.359,
      "step": 30479
    },
    {
      "epoch": 1.3987426001560277,
      "grad_norm": 0.4527153968811035,
      "learning_rate": 5.699062249309719e-06,
      "loss": 0.3823,
      "step": 30480
    },
    {
      "epoch": 1.3987884906612822,
      "grad_norm": 0.45467329025268555,
      "learning_rate": 5.698819468796945e-06,
      "loss": 0.3252,
      "step": 30481
    },
    {
      "epoch": 1.3988343811665367,
      "grad_norm": 0.45557960867881775,
      "learning_rate": 5.69857668660373e-06,
      "loss": 0.3653,
      "step": 30482
    },
    {
      "epoch": 1.3988802716717912,
      "grad_norm": 0.45011451840400696,
      "learning_rate": 5.69833390273066e-06,
      "loss": 0.3077,
      "step": 30483
    },
    {
      "epoch": 1.3989261621770455,
      "grad_norm": 0.4902711808681488,
      "learning_rate": 5.698091117178317e-06,
      "loss": 0.3942,
      "step": 30484
    },
    {
      "epoch": 1.3989720526823,
      "grad_norm": 0.4374467432498932,
      "learning_rate": 5.697848329947285e-06,
      "loss": 0.2941,
      "step": 30485
    },
    {
      "epoch": 1.3990179431875545,
      "grad_norm": 0.4423973262310028,
      "learning_rate": 5.6976055410381495e-06,
      "loss": 0.3232,
      "step": 30486
    },
    {
      "epoch": 1.399063833692809,
      "grad_norm": 0.4958360493183136,
      "learning_rate": 5.697362750451489e-06,
      "loss": 0.3902,
      "step": 30487
    },
    {
      "epoch": 1.3991097241980635,
      "grad_norm": 0.44962233304977417,
      "learning_rate": 5.6971199581878954e-06,
      "loss": 0.3173,
      "step": 30488
    },
    {
      "epoch": 1.399155614703318,
      "grad_norm": 0.4194338023662567,
      "learning_rate": 5.696877164247946e-06,
      "loss": 0.2835,
      "step": 30489
    },
    {
      "epoch": 1.3992015052085724,
      "grad_norm": 0.46637168526649475,
      "learning_rate": 5.696634368632228e-06,
      "loss": 0.3657,
      "step": 30490
    },
    {
      "epoch": 1.399247395713827,
      "grad_norm": 0.4376131594181061,
      "learning_rate": 5.696391571341324e-06,
      "loss": 0.3076,
      "step": 30491
    },
    {
      "epoch": 1.3992932862190812,
      "grad_norm": 0.4737323820590973,
      "learning_rate": 5.6961487723758165e-06,
      "loss": 0.3699,
      "step": 30492
    },
    {
      "epoch": 1.3993391767243357,
      "grad_norm": 0.48057979345321655,
      "learning_rate": 5.6959059717362916e-06,
      "loss": 0.361,
      "step": 30493
    },
    {
      "epoch": 1.3993850672295902,
      "grad_norm": 0.46984413266181946,
      "learning_rate": 5.6956631694233344e-06,
      "loss": 0.3396,
      "step": 30494
    },
    {
      "epoch": 1.3994309577348447,
      "grad_norm": 0.47452211380004883,
      "learning_rate": 5.695420365437526e-06,
      "loss": 0.3804,
      "step": 30495
    },
    {
      "epoch": 1.399476848240099,
      "grad_norm": 0.5063121914863586,
      "learning_rate": 5.6951775597794504e-06,
      "loss": 0.3905,
      "step": 30496
    },
    {
      "epoch": 1.3995227387453535,
      "grad_norm": 0.4413262903690338,
      "learning_rate": 5.6949347524496925e-06,
      "loss": 0.2889,
      "step": 30497
    },
    {
      "epoch": 1.399568629250608,
      "grad_norm": 0.5324788689613342,
      "learning_rate": 5.694691943448834e-06,
      "loss": 0.3573,
      "step": 30498
    },
    {
      "epoch": 1.3996145197558625,
      "grad_norm": 0.47010743618011475,
      "learning_rate": 5.694449132777463e-06,
      "loss": 0.3626,
      "step": 30499
    },
    {
      "epoch": 1.399660410261117,
      "grad_norm": 0.48463723063468933,
      "learning_rate": 5.694206320436161e-06,
      "loss": 0.4044,
      "step": 30500
    },
    {
      "epoch": 1.3997063007663715,
      "grad_norm": 0.4952201843261719,
      "learning_rate": 5.693963506425512e-06,
      "loss": 0.3539,
      "step": 30501
    },
    {
      "epoch": 1.399752191271626,
      "grad_norm": 0.4416428804397583,
      "learning_rate": 5.693720690746098e-06,
      "loss": 0.2946,
      "step": 30502
    },
    {
      "epoch": 1.3997980817768805,
      "grad_norm": 0.47428685426712036,
      "learning_rate": 5.6934778733985064e-06,
      "loss": 0.3646,
      "step": 30503
    },
    {
      "epoch": 1.3998439722821348,
      "grad_norm": 0.47907447814941406,
      "learning_rate": 5.693235054383318e-06,
      "loss": 0.4347,
      "step": 30504
    },
    {
      "epoch": 1.3998898627873892,
      "grad_norm": 0.3950011730194092,
      "learning_rate": 5.692992233701119e-06,
      "loss": 0.2598,
      "step": 30505
    },
    {
      "epoch": 1.3999357532926437,
      "grad_norm": 0.49776071310043335,
      "learning_rate": 5.692749411352493e-06,
      "loss": 0.3811,
      "step": 30506
    },
    {
      "epoch": 1.3999816437978982,
      "grad_norm": 0.4954083263874054,
      "learning_rate": 5.692506587338022e-06,
      "loss": 0.4057,
      "step": 30507
    },
    {
      "epoch": 1.4000275343031527,
      "grad_norm": 0.46958890557289124,
      "learning_rate": 5.692263761658292e-06,
      "loss": 0.4367,
      "step": 30508
    },
    {
      "epoch": 1.400073424808407,
      "grad_norm": 0.4891435205936432,
      "learning_rate": 5.692020934313886e-06,
      "loss": 0.3689,
      "step": 30509
    },
    {
      "epoch": 1.4001193153136615,
      "grad_norm": 0.469858855009079,
      "learning_rate": 5.691778105305387e-06,
      "loss": 0.3772,
      "step": 30510
    },
    {
      "epoch": 1.400165205818916,
      "grad_norm": 0.4417387545108795,
      "learning_rate": 5.691535274633381e-06,
      "loss": 0.2982,
      "step": 30511
    },
    {
      "epoch": 1.4002110963241705,
      "grad_norm": 0.4273539185523987,
      "learning_rate": 5.691292442298451e-06,
      "loss": 0.3181,
      "step": 30512
    },
    {
      "epoch": 1.400256986829425,
      "grad_norm": 0.4822871685028076,
      "learning_rate": 5.69104960830118e-06,
      "loss": 0.3987,
      "step": 30513
    },
    {
      "epoch": 1.4003028773346795,
      "grad_norm": 0.4981745779514313,
      "learning_rate": 5.6908067726421545e-06,
      "loss": 0.4462,
      "step": 30514
    },
    {
      "epoch": 1.400348767839934,
      "grad_norm": 0.4563388228416443,
      "learning_rate": 5.690563935321954e-06,
      "loss": 0.3496,
      "step": 30515
    },
    {
      "epoch": 1.4003946583451885,
      "grad_norm": 0.48132598400115967,
      "learning_rate": 5.690321096341168e-06,
      "loss": 0.362,
      "step": 30516
    },
    {
      "epoch": 1.4004405488504428,
      "grad_norm": 0.45878010988235474,
      "learning_rate": 5.690078255700377e-06,
      "loss": 0.3385,
      "step": 30517
    },
    {
      "epoch": 1.4004864393556973,
      "grad_norm": 0.47036391496658325,
      "learning_rate": 5.689835413400164e-06,
      "loss": 0.3601,
      "step": 30518
    },
    {
      "epoch": 1.4005323298609518,
      "grad_norm": 0.4814489781856537,
      "learning_rate": 5.6895925694411154e-06,
      "loss": 0.3484,
      "step": 30519
    },
    {
      "epoch": 1.4005782203662063,
      "grad_norm": 0.47754746675491333,
      "learning_rate": 5.689349723823816e-06,
      "loss": 0.3896,
      "step": 30520
    },
    {
      "epoch": 1.4006241108714608,
      "grad_norm": 0.47672638297080994,
      "learning_rate": 5.689106876548845e-06,
      "loss": 0.406,
      "step": 30521
    },
    {
      "epoch": 1.400670001376715,
      "grad_norm": 0.4592185914516449,
      "learning_rate": 5.6888640276167915e-06,
      "loss": 0.3492,
      "step": 30522
    },
    {
      "epoch": 1.4007158918819695,
      "grad_norm": 0.4650770425796509,
      "learning_rate": 5.6886211770282385e-06,
      "loss": 0.3439,
      "step": 30523
    },
    {
      "epoch": 1.400761782387224,
      "grad_norm": 0.47770148515701294,
      "learning_rate": 5.6883783247837665e-06,
      "loss": 0.3803,
      "step": 30524
    },
    {
      "epoch": 1.4008076728924785,
      "grad_norm": 0.4395122230052948,
      "learning_rate": 5.6881354708839635e-06,
      "loss": 0.3212,
      "step": 30525
    },
    {
      "epoch": 1.400853563397733,
      "grad_norm": 0.4546579420566559,
      "learning_rate": 5.6878926153294105e-06,
      "loss": 0.3717,
      "step": 30526
    },
    {
      "epoch": 1.4008994539029875,
      "grad_norm": 0.47088274359703064,
      "learning_rate": 5.687649758120694e-06,
      "loss": 0.3025,
      "step": 30527
    },
    {
      "epoch": 1.400945344408242,
      "grad_norm": 0.4586959481239319,
      "learning_rate": 5.687406899258396e-06,
      "loss": 0.3827,
      "step": 30528
    },
    {
      "epoch": 1.4009912349134965,
      "grad_norm": 0.4332188367843628,
      "learning_rate": 5.6871640387431026e-06,
      "loss": 0.3234,
      "step": 30529
    },
    {
      "epoch": 1.4010371254187508,
      "grad_norm": 0.5075878500938416,
      "learning_rate": 5.686921176575395e-06,
      "loss": 0.3696,
      "step": 30530
    },
    {
      "epoch": 1.4010830159240053,
      "grad_norm": 0.4826306104660034,
      "learning_rate": 5.68667831275586e-06,
      "loss": 0.3764,
      "step": 30531
    },
    {
      "epoch": 1.4011289064292598,
      "grad_norm": 0.4077684283256531,
      "learning_rate": 5.686435447285079e-06,
      "loss": 0.2897,
      "step": 30532
    },
    {
      "epoch": 1.4011747969345143,
      "grad_norm": 0.4237208068370819,
      "learning_rate": 5.686192580163639e-06,
      "loss": 0.2835,
      "step": 30533
    },
    {
      "epoch": 1.4012206874397686,
      "grad_norm": 0.4674409031867981,
      "learning_rate": 5.6859497113921225e-06,
      "loss": 0.3727,
      "step": 30534
    },
    {
      "epoch": 1.401266577945023,
      "grad_norm": 0.4776623249053955,
      "learning_rate": 5.685706840971111e-06,
      "loss": 0.3822,
      "step": 30535
    },
    {
      "epoch": 1.4013124684502776,
      "grad_norm": 0.46475669741630554,
      "learning_rate": 5.685463968901193e-06,
      "loss": 0.3488,
      "step": 30536
    },
    {
      "epoch": 1.401358358955532,
      "grad_norm": 0.4328990876674652,
      "learning_rate": 5.68522109518295e-06,
      "loss": 0.2959,
      "step": 30537
    },
    {
      "epoch": 1.4014042494607866,
      "grad_norm": 0.46911895275115967,
      "learning_rate": 5.684978219816968e-06,
      "loss": 0.3437,
      "step": 30538
    },
    {
      "epoch": 1.401450139966041,
      "grad_norm": 0.4853106737136841,
      "learning_rate": 5.684735342803828e-06,
      "loss": 0.3793,
      "step": 30539
    },
    {
      "epoch": 1.4014960304712956,
      "grad_norm": 0.44736698269844055,
      "learning_rate": 5.684492464144116e-06,
      "loss": 0.3057,
      "step": 30540
    },
    {
      "epoch": 1.40154192097655,
      "grad_norm": 0.4573589861392975,
      "learning_rate": 5.684249583838414e-06,
      "loss": 0.3831,
      "step": 30541
    },
    {
      "epoch": 1.4015878114818043,
      "grad_norm": 0.44389578700065613,
      "learning_rate": 5.68400670188731e-06,
      "loss": 0.3423,
      "step": 30542
    },
    {
      "epoch": 1.4016337019870588,
      "grad_norm": 0.45849719643592834,
      "learning_rate": 5.683763818291383e-06,
      "loss": 0.3039,
      "step": 30543
    },
    {
      "epoch": 1.4016795924923133,
      "grad_norm": 0.4361306130886078,
      "learning_rate": 5.683520933051223e-06,
      "loss": 0.2926,
      "step": 30544
    },
    {
      "epoch": 1.4017254829975678,
      "grad_norm": 0.49260056018829346,
      "learning_rate": 5.683278046167411e-06,
      "loss": 0.3785,
      "step": 30545
    },
    {
      "epoch": 1.4017713735028223,
      "grad_norm": 0.46770408749580383,
      "learning_rate": 5.683035157640529e-06,
      "loss": 0.3693,
      "step": 30546
    },
    {
      "epoch": 1.4018172640080766,
      "grad_norm": 0.4235199987888336,
      "learning_rate": 5.682792267471162e-06,
      "loss": 0.287,
      "step": 30547
    },
    {
      "epoch": 1.401863154513331,
      "grad_norm": 0.4623138904571533,
      "learning_rate": 5.682549375659897e-06,
      "loss": 0.3785,
      "step": 30548
    },
    {
      "epoch": 1.4019090450185856,
      "grad_norm": 0.47290629148483276,
      "learning_rate": 5.6823064822073175e-06,
      "loss": 0.3849,
      "step": 30549
    },
    {
      "epoch": 1.40195493552384,
      "grad_norm": 0.48674890398979187,
      "learning_rate": 5.682063587114004e-06,
      "loss": 0.4225,
      "step": 30550
    },
    {
      "epoch": 1.4020008260290946,
      "grad_norm": 0.44226357340812683,
      "learning_rate": 5.681820690380544e-06,
      "loss": 0.3243,
      "step": 30551
    },
    {
      "epoch": 1.402046716534349,
      "grad_norm": 0.44181811809539795,
      "learning_rate": 5.681577792007519e-06,
      "loss": 0.3169,
      "step": 30552
    },
    {
      "epoch": 1.4020926070396036,
      "grad_norm": 0.495848149061203,
      "learning_rate": 5.681334891995516e-06,
      "loss": 0.3747,
      "step": 30553
    },
    {
      "epoch": 1.402138497544858,
      "grad_norm": 0.48354411125183105,
      "learning_rate": 5.681091990345116e-06,
      "loss": 0.3536,
      "step": 30554
    },
    {
      "epoch": 1.4021843880501124,
      "grad_norm": 0.4623820185661316,
      "learning_rate": 5.680849087056907e-06,
      "loss": 0.3622,
      "step": 30555
    },
    {
      "epoch": 1.4022302785553669,
      "grad_norm": 0.4764816462993622,
      "learning_rate": 5.68060618213147e-06,
      "loss": 0.3797,
      "step": 30556
    },
    {
      "epoch": 1.4022761690606214,
      "grad_norm": 0.45645153522491455,
      "learning_rate": 5.680363275569388e-06,
      "loss": 0.3657,
      "step": 30557
    },
    {
      "epoch": 1.4023220595658759,
      "grad_norm": 0.4388754665851593,
      "learning_rate": 5.680120367371248e-06,
      "loss": 0.335,
      "step": 30558
    },
    {
      "epoch": 1.4023679500711304,
      "grad_norm": 0.5064692497253418,
      "learning_rate": 5.679877457537635e-06,
      "loss": 0.3976,
      "step": 30559
    },
    {
      "epoch": 1.4024138405763846,
      "grad_norm": 0.5059183835983276,
      "learning_rate": 5.679634546069129e-06,
      "loss": 0.3498,
      "step": 30560
    },
    {
      "epoch": 1.4024597310816391,
      "grad_norm": 0.5017886757850647,
      "learning_rate": 5.6793916329663176e-06,
      "loss": 0.4308,
      "step": 30561
    },
    {
      "epoch": 1.4025056215868936,
      "grad_norm": 0.4760131239891052,
      "learning_rate": 5.679148718229784e-06,
      "loss": 0.3618,
      "step": 30562
    },
    {
      "epoch": 1.4025515120921481,
      "grad_norm": 0.44018077850341797,
      "learning_rate": 5.6789058018601104e-06,
      "loss": 0.3218,
      "step": 30563
    },
    {
      "epoch": 1.4025974025974026,
      "grad_norm": 0.4927705228328705,
      "learning_rate": 5.678662883857884e-06,
      "loss": 0.3612,
      "step": 30564
    },
    {
      "epoch": 1.4026432931026571,
      "grad_norm": 0.4663553535938263,
      "learning_rate": 5.678419964223688e-06,
      "loss": 0.3448,
      "step": 30565
    },
    {
      "epoch": 1.4026891836079116,
      "grad_norm": 0.4771214723587036,
      "learning_rate": 5.6781770429581066e-06,
      "loss": 0.3763,
      "step": 30566
    },
    {
      "epoch": 1.4027350741131661,
      "grad_norm": 0.4704783260822296,
      "learning_rate": 5.677934120061721e-06,
      "loss": 0.3892,
      "step": 30567
    },
    {
      "epoch": 1.4027809646184204,
      "grad_norm": 0.4526841938495636,
      "learning_rate": 5.67769119553512e-06,
      "loss": 0.3691,
      "step": 30568
    },
    {
      "epoch": 1.4028268551236749,
      "grad_norm": 0.4877703785896301,
      "learning_rate": 5.6774482693788835e-06,
      "loss": 0.4102,
      "step": 30569
    },
    {
      "epoch": 1.4028727456289294,
      "grad_norm": 0.517266035079956,
      "learning_rate": 5.6772053415936e-06,
      "loss": 0.3549,
      "step": 30570
    },
    {
      "epoch": 1.4029186361341839,
      "grad_norm": 0.4898463785648346,
      "learning_rate": 5.676962412179851e-06,
      "loss": 0.4367,
      "step": 30571
    },
    {
      "epoch": 1.4029645266394384,
      "grad_norm": 0.456590861082077,
      "learning_rate": 5.67671948113822e-06,
      "loss": 0.3689,
      "step": 30572
    },
    {
      "epoch": 1.4030104171446927,
      "grad_norm": 0.491571307182312,
      "learning_rate": 5.676476548469293e-06,
      "loss": 0.3672,
      "step": 30573
    },
    {
      "epoch": 1.4030563076499472,
      "grad_norm": 0.4826657772064209,
      "learning_rate": 5.676233614173652e-06,
      "loss": 0.3896,
      "step": 30574
    },
    {
      "epoch": 1.4031021981552017,
      "grad_norm": 0.4708399772644043,
      "learning_rate": 5.675990678251884e-06,
      "loss": 0.374,
      "step": 30575
    },
    {
      "epoch": 1.4031480886604561,
      "grad_norm": 0.5163946747779846,
      "learning_rate": 5.675747740704571e-06,
      "loss": 0.4693,
      "step": 30576
    },
    {
      "epoch": 1.4031939791657106,
      "grad_norm": 0.486020028591156,
      "learning_rate": 5.675504801532299e-06,
      "loss": 0.4305,
      "step": 30577
    },
    {
      "epoch": 1.4032398696709651,
      "grad_norm": 0.4489801824092865,
      "learning_rate": 5.6752618607356505e-06,
      "loss": 0.3307,
      "step": 30578
    },
    {
      "epoch": 1.4032857601762196,
      "grad_norm": 0.49930375814437866,
      "learning_rate": 5.675018918315211e-06,
      "loss": 0.3732,
      "step": 30579
    },
    {
      "epoch": 1.4033316506814741,
      "grad_norm": 0.4781452417373657,
      "learning_rate": 5.674775974271561e-06,
      "loss": 0.3717,
      "step": 30580
    },
    {
      "epoch": 1.4033775411867284,
      "grad_norm": 0.5085635185241699,
      "learning_rate": 5.674533028605291e-06,
      "loss": 0.4462,
      "step": 30581
    },
    {
      "epoch": 1.403423431691983,
      "grad_norm": 0.4564163386821747,
      "learning_rate": 5.674290081316982e-06,
      "loss": 0.3163,
      "step": 30582
    },
    {
      "epoch": 1.4034693221972374,
      "grad_norm": 0.5115766525268555,
      "learning_rate": 5.674047132407217e-06,
      "loss": 0.3875,
      "step": 30583
    },
    {
      "epoch": 1.403515212702492,
      "grad_norm": 0.4586496651172638,
      "learning_rate": 5.673804181876581e-06,
      "loss": 0.3294,
      "step": 30584
    },
    {
      "epoch": 1.4035611032077462,
      "grad_norm": 0.4568221867084503,
      "learning_rate": 5.673561229725659e-06,
      "loss": 0.3358,
      "step": 30585
    },
    {
      "epoch": 1.4036069937130007,
      "grad_norm": 0.49388402700424194,
      "learning_rate": 5.6733182759550354e-06,
      "loss": 0.4421,
      "step": 30586
    },
    {
      "epoch": 1.4036528842182552,
      "grad_norm": 0.4518837630748749,
      "learning_rate": 5.673075320565293e-06,
      "loss": 0.332,
      "step": 30587
    },
    {
      "epoch": 1.4036987747235097,
      "grad_norm": 0.4668740928173065,
      "learning_rate": 5.672832363557018e-06,
      "loss": 0.349,
      "step": 30588
    },
    {
      "epoch": 1.4037446652287642,
      "grad_norm": 0.5386171936988831,
      "learning_rate": 5.6725894049307924e-06,
      "loss": 0.5308,
      "step": 30589
    },
    {
      "epoch": 1.4037905557340187,
      "grad_norm": 0.46664607524871826,
      "learning_rate": 5.672346444687202e-06,
      "loss": 0.3385,
      "step": 30590
    },
    {
      "epoch": 1.4038364462392732,
      "grad_norm": 0.48087185621261597,
      "learning_rate": 5.67210348282683e-06,
      "loss": 0.4125,
      "step": 30591
    },
    {
      "epoch": 1.4038823367445277,
      "grad_norm": 0.4826028645038605,
      "learning_rate": 5.671860519350263e-06,
      "loss": 0.3186,
      "step": 30592
    },
    {
      "epoch": 1.403928227249782,
      "grad_norm": 0.4580307900905609,
      "learning_rate": 5.671617554258083e-06,
      "loss": 0.318,
      "step": 30593
    },
    {
      "epoch": 1.4039741177550364,
      "grad_norm": 0.450836181640625,
      "learning_rate": 5.6713745875508745e-06,
      "loss": 0.3475,
      "step": 30594
    },
    {
      "epoch": 1.404020008260291,
      "grad_norm": 0.4636431038379669,
      "learning_rate": 5.67113161922922e-06,
      "loss": 0.3135,
      "step": 30595
    },
    {
      "epoch": 1.4040658987655454,
      "grad_norm": 0.48314210772514343,
      "learning_rate": 5.670888649293709e-06,
      "loss": 0.344,
      "step": 30596
    },
    {
      "epoch": 1.4041117892708,
      "grad_norm": 0.4551754891872406,
      "learning_rate": 5.670645677744919e-06,
      "loss": 0.3151,
      "step": 30597
    },
    {
      "epoch": 1.4041576797760542,
      "grad_norm": 0.4140892028808594,
      "learning_rate": 5.670402704583441e-06,
      "loss": 0.2445,
      "step": 30598
    },
    {
      "epoch": 1.4042035702813087,
      "grad_norm": 0.4736423194408417,
      "learning_rate": 5.670159729809855e-06,
      "loss": 0.3405,
      "step": 30599
    },
    {
      "epoch": 1.4042494607865632,
      "grad_norm": 0.49598556756973267,
      "learning_rate": 5.669916753424745e-06,
      "loss": 0.3872,
      "step": 30600
    },
    {
      "epoch": 1.4042953512918177,
      "grad_norm": 0.4861445128917694,
      "learning_rate": 5.669673775428698e-06,
      "loss": 0.4017,
      "step": 30601
    },
    {
      "epoch": 1.4043412417970722,
      "grad_norm": 0.466854989528656,
      "learning_rate": 5.669430795822297e-06,
      "loss": 0.3338,
      "step": 30602
    },
    {
      "epoch": 1.4043871323023267,
      "grad_norm": 0.5214053392410278,
      "learning_rate": 5.6691878146061265e-06,
      "loss": 0.3376,
      "step": 30603
    },
    {
      "epoch": 1.4044330228075812,
      "grad_norm": 0.5264842510223389,
      "learning_rate": 5.66894483178077e-06,
      "loss": 0.442,
      "step": 30604
    },
    {
      "epoch": 1.4044789133128357,
      "grad_norm": 0.4776712954044342,
      "learning_rate": 5.668701847346813e-06,
      "loss": 0.3482,
      "step": 30605
    },
    {
      "epoch": 1.40452480381809,
      "grad_norm": 0.5053424835205078,
      "learning_rate": 5.668458861304838e-06,
      "loss": 0.4322,
      "step": 30606
    },
    {
      "epoch": 1.4045706943233445,
      "grad_norm": 0.5037473440170288,
      "learning_rate": 5.66821587365543e-06,
      "loss": 0.396,
      "step": 30607
    },
    {
      "epoch": 1.404616584828599,
      "grad_norm": 0.4608955383300781,
      "learning_rate": 5.6679728843991745e-06,
      "loss": 0.3676,
      "step": 30608
    },
    {
      "epoch": 1.4046624753338535,
      "grad_norm": 0.49622461199760437,
      "learning_rate": 5.667729893536655e-06,
      "loss": 0.408,
      "step": 30609
    },
    {
      "epoch": 1.404708365839108,
      "grad_norm": 0.4593330919742584,
      "learning_rate": 5.667486901068457e-06,
      "loss": 0.3604,
      "step": 30610
    },
    {
      "epoch": 1.4047542563443622,
      "grad_norm": 0.46884584426879883,
      "learning_rate": 5.6672439069951624e-06,
      "loss": 0.3118,
      "step": 30611
    },
    {
      "epoch": 1.4048001468496167,
      "grad_norm": 0.47514376044273376,
      "learning_rate": 5.667000911317357e-06,
      "loss": 0.3358,
      "step": 30612
    },
    {
      "epoch": 1.4048460373548712,
      "grad_norm": 0.5239102244377136,
      "learning_rate": 5.666757914035625e-06,
      "loss": 0.3936,
      "step": 30613
    },
    {
      "epoch": 1.4048919278601257,
      "grad_norm": 0.48989957571029663,
      "learning_rate": 5.666514915150551e-06,
      "loss": 0.353,
      "step": 30614
    },
    {
      "epoch": 1.4049378183653802,
      "grad_norm": 0.449582576751709,
      "learning_rate": 5.6662719146627176e-06,
      "loss": 0.3497,
      "step": 30615
    },
    {
      "epoch": 1.4049837088706347,
      "grad_norm": 0.507612943649292,
      "learning_rate": 5.666028912572712e-06,
      "loss": 0.4163,
      "step": 30616
    },
    {
      "epoch": 1.4050295993758892,
      "grad_norm": 0.4762733578681946,
      "learning_rate": 5.665785908881116e-06,
      "loss": 0.3628,
      "step": 30617
    },
    {
      "epoch": 1.4050754898811437,
      "grad_norm": 0.477347195148468,
      "learning_rate": 5.665542903588516e-06,
      "loss": 0.3802,
      "step": 30618
    },
    {
      "epoch": 1.405121380386398,
      "grad_norm": 0.5213756561279297,
      "learning_rate": 5.665299896695494e-06,
      "loss": 0.4043,
      "step": 30619
    },
    {
      "epoch": 1.4051672708916525,
      "grad_norm": 0.4820301830768585,
      "learning_rate": 5.665056888202637e-06,
      "loss": 0.4263,
      "step": 30620
    },
    {
      "epoch": 1.405213161396907,
      "grad_norm": 0.48144298791885376,
      "learning_rate": 5.664813878110529e-06,
      "loss": 0.343,
      "step": 30621
    },
    {
      "epoch": 1.4052590519021615,
      "grad_norm": 0.4894770681858063,
      "learning_rate": 5.6645708664197505e-06,
      "loss": 0.3812,
      "step": 30622
    },
    {
      "epoch": 1.4053049424074158,
      "grad_norm": 0.4747690260410309,
      "learning_rate": 5.66432785313089e-06,
      "loss": 0.3432,
      "step": 30623
    },
    {
      "epoch": 1.4053508329126703,
      "grad_norm": 0.4587125778198242,
      "learning_rate": 5.6640848382445324e-06,
      "loss": 0.2986,
      "step": 30624
    },
    {
      "epoch": 1.4053967234179248,
      "grad_norm": 0.4678109586238861,
      "learning_rate": 5.663841821761259e-06,
      "loss": 0.3418,
      "step": 30625
    },
    {
      "epoch": 1.4054426139231793,
      "grad_norm": 0.4801413118839264,
      "learning_rate": 5.663598803681656e-06,
      "loss": 0.3682,
      "step": 30626
    },
    {
      "epoch": 1.4054885044284338,
      "grad_norm": 0.46379968523979187,
      "learning_rate": 5.6633557840063066e-06,
      "loss": 0.3162,
      "step": 30627
    },
    {
      "epoch": 1.4055343949336883,
      "grad_norm": 0.4812333285808563,
      "learning_rate": 5.663112762735796e-06,
      "loss": 0.3591,
      "step": 30628
    },
    {
      "epoch": 1.4055802854389428,
      "grad_norm": 0.44547685980796814,
      "learning_rate": 5.6628697398707076e-06,
      "loss": 0.3447,
      "step": 30629
    },
    {
      "epoch": 1.4056261759441973,
      "grad_norm": 0.5347281098365784,
      "learning_rate": 5.662626715411629e-06,
      "loss": 0.4761,
      "step": 30630
    },
    {
      "epoch": 1.4056720664494515,
      "grad_norm": 0.4757799506187439,
      "learning_rate": 5.662383689359141e-06,
      "loss": 0.3481,
      "step": 30631
    },
    {
      "epoch": 1.405717956954706,
      "grad_norm": 0.46899133920669556,
      "learning_rate": 5.662140661713829e-06,
      "loss": 0.36,
      "step": 30632
    },
    {
      "epoch": 1.4057638474599605,
      "grad_norm": 0.45681729912757874,
      "learning_rate": 5.661897632476279e-06,
      "loss": 0.3167,
      "step": 30633
    },
    {
      "epoch": 1.405809737965215,
      "grad_norm": 0.43412426114082336,
      "learning_rate": 5.661654601647072e-06,
      "loss": 0.3392,
      "step": 30634
    },
    {
      "epoch": 1.4058556284704695,
      "grad_norm": 0.46264445781707764,
      "learning_rate": 5.661411569226797e-06,
      "loss": 0.32,
      "step": 30635
    },
    {
      "epoch": 1.4059015189757238,
      "grad_norm": 0.6127398014068604,
      "learning_rate": 5.661168535216035e-06,
      "loss": 0.2972,
      "step": 30636
    },
    {
      "epoch": 1.4059474094809783,
      "grad_norm": 0.4390941262245178,
      "learning_rate": 5.660925499615369e-06,
      "loss": 0.3139,
      "step": 30637
    },
    {
      "epoch": 1.4059932999862328,
      "grad_norm": 0.43178120255470276,
      "learning_rate": 5.660682462425388e-06,
      "loss": 0.2964,
      "step": 30638
    },
    {
      "epoch": 1.4060391904914873,
      "grad_norm": 0.48744189739227295,
      "learning_rate": 5.6604394236466745e-06,
      "loss": 0.3671,
      "step": 30639
    },
    {
      "epoch": 1.4060850809967418,
      "grad_norm": 0.47400999069213867,
      "learning_rate": 5.660196383279811e-06,
      "loss": 0.3113,
      "step": 30640
    },
    {
      "epoch": 1.4061309715019963,
      "grad_norm": 0.5465152859687805,
      "learning_rate": 5.659953341325386e-06,
      "loss": 0.5219,
      "step": 30641
    },
    {
      "epoch": 1.4061768620072508,
      "grad_norm": 0.46050935983657837,
      "learning_rate": 5.6597102977839804e-06,
      "loss": 0.3637,
      "step": 30642
    },
    {
      "epoch": 1.4062227525125053,
      "grad_norm": 0.48843950033187866,
      "learning_rate": 5.659467252656179e-06,
      "loss": 0.3873,
      "step": 30643
    },
    {
      "epoch": 1.4062686430177596,
      "grad_norm": 0.49041104316711426,
      "learning_rate": 5.659224205942568e-06,
      "loss": 0.3621,
      "step": 30644
    },
    {
      "epoch": 1.406314533523014,
      "grad_norm": 0.46887555718421936,
      "learning_rate": 5.6589811576437295e-06,
      "loss": 0.3447,
      "step": 30645
    },
    {
      "epoch": 1.4063604240282686,
      "grad_norm": 0.47729310393333435,
      "learning_rate": 5.6587381077602485e-06,
      "loss": 0.3407,
      "step": 30646
    },
    {
      "epoch": 1.406406314533523,
      "grad_norm": 0.4840472638607025,
      "learning_rate": 5.658495056292714e-06,
      "loss": 0.3925,
      "step": 30647
    },
    {
      "epoch": 1.4064522050387775,
      "grad_norm": 0.43364202976226807,
      "learning_rate": 5.658252003241704e-06,
      "loss": 0.3,
      "step": 30648
    },
    {
      "epoch": 1.4064980955440318,
      "grad_norm": 0.5036024451255798,
      "learning_rate": 5.658008948607806e-06,
      "loss": 0.4112,
      "step": 30649
    },
    {
      "epoch": 1.4065439860492863,
      "grad_norm": 0.507006824016571,
      "learning_rate": 5.657765892391604e-06,
      "loss": 0.4343,
      "step": 30650
    },
    {
      "epoch": 1.4065898765545408,
      "grad_norm": 0.4379733204841614,
      "learning_rate": 5.657522834593683e-06,
      "loss": 0.3271,
      "step": 30651
    },
    {
      "epoch": 1.4066357670597953,
      "grad_norm": 0.484536349773407,
      "learning_rate": 5.657279775214628e-06,
      "loss": 0.3578,
      "step": 30652
    },
    {
      "epoch": 1.4066816575650498,
      "grad_norm": 0.47118768095970154,
      "learning_rate": 5.657036714255022e-06,
      "loss": 0.427,
      "step": 30653
    },
    {
      "epoch": 1.4067275480703043,
      "grad_norm": 0.474059522151947,
      "learning_rate": 5.656793651715449e-06,
      "loss": 0.3944,
      "step": 30654
    },
    {
      "epoch": 1.4067734385755588,
      "grad_norm": 0.4684872329235077,
      "learning_rate": 5.656550587596496e-06,
      "loss": 0.3495,
      "step": 30655
    },
    {
      "epoch": 1.4068193290808133,
      "grad_norm": 0.45506152510643005,
      "learning_rate": 5.656307521898744e-06,
      "loss": 0.2978,
      "step": 30656
    },
    {
      "epoch": 1.4068652195860676,
      "grad_norm": 0.45992323756217957,
      "learning_rate": 5.656064454622781e-06,
      "loss": 0.3611,
      "step": 30657
    },
    {
      "epoch": 1.406911110091322,
      "grad_norm": 0.44445157051086426,
      "learning_rate": 5.6558213857691916e-06,
      "loss": 0.3235,
      "step": 30658
    },
    {
      "epoch": 1.4069570005965766,
      "grad_norm": 0.5436306595802307,
      "learning_rate": 5.6555783153385564e-06,
      "loss": 0.495,
      "step": 30659
    },
    {
      "epoch": 1.407002891101831,
      "grad_norm": 0.4242517054080963,
      "learning_rate": 5.6553352433314626e-06,
      "loss": 0.3088,
      "step": 30660
    },
    {
      "epoch": 1.4070487816070856,
      "grad_norm": 0.4567176103591919,
      "learning_rate": 5.655092169748495e-06,
      "loss": 0.3595,
      "step": 30661
    },
    {
      "epoch": 1.4070946721123399,
      "grad_norm": 0.4350859522819519,
      "learning_rate": 5.654849094590237e-06,
      "loss": 0.2966,
      "step": 30662
    },
    {
      "epoch": 1.4071405626175943,
      "grad_norm": 0.4829665720462799,
      "learning_rate": 5.654606017857274e-06,
      "loss": 0.3867,
      "step": 30663
    },
    {
      "epoch": 1.4071864531228488,
      "grad_norm": 0.4963776469230652,
      "learning_rate": 5.654362939550191e-06,
      "loss": 0.3994,
      "step": 30664
    },
    {
      "epoch": 1.4072323436281033,
      "grad_norm": 0.445199579000473,
      "learning_rate": 5.654119859669569e-06,
      "loss": 0.3409,
      "step": 30665
    },
    {
      "epoch": 1.4072782341333578,
      "grad_norm": 0.4864162802696228,
      "learning_rate": 5.653876778215996e-06,
      "loss": 0.3871,
      "step": 30666
    },
    {
      "epoch": 1.4073241246386123,
      "grad_norm": 0.48371148109436035,
      "learning_rate": 5.6536336951900576e-06,
      "loss": 0.3984,
      "step": 30667
    },
    {
      "epoch": 1.4073700151438668,
      "grad_norm": 0.47050681710243225,
      "learning_rate": 5.653390610592336e-06,
      "loss": 0.3548,
      "step": 30668
    },
    {
      "epoch": 1.4074159056491213,
      "grad_norm": 0.46876898407936096,
      "learning_rate": 5.653147524423416e-06,
      "loss": 0.2987,
      "step": 30669
    },
    {
      "epoch": 1.4074617961543756,
      "grad_norm": 0.5129265189170837,
      "learning_rate": 5.652904436683882e-06,
      "loss": 0.4216,
      "step": 30670
    },
    {
      "epoch": 1.40750768665963,
      "grad_norm": 0.45877495408058167,
      "learning_rate": 5.652661347374317e-06,
      "loss": 0.3037,
      "step": 30671
    },
    {
      "epoch": 1.4075535771648846,
      "grad_norm": 0.47052499651908875,
      "learning_rate": 5.65241825649531e-06,
      "loss": 0.3779,
      "step": 30672
    },
    {
      "epoch": 1.407599467670139,
      "grad_norm": 0.4405198395252228,
      "learning_rate": 5.652175164047442e-06,
      "loss": 0.3179,
      "step": 30673
    },
    {
      "epoch": 1.4076453581753934,
      "grad_norm": 0.4603905975818634,
      "learning_rate": 5.651932070031298e-06,
      "loss": 0.358,
      "step": 30674
    },
    {
      "epoch": 1.4076912486806479,
      "grad_norm": 0.48367437720298767,
      "learning_rate": 5.651688974447466e-06,
      "loss": 0.3433,
      "step": 30675
    },
    {
      "epoch": 1.4077371391859024,
      "grad_norm": 0.48867806792259216,
      "learning_rate": 5.651445877296524e-06,
      "loss": 0.3574,
      "step": 30676
    },
    {
      "epoch": 1.4077830296911569,
      "grad_norm": 0.47831350564956665,
      "learning_rate": 5.651202778579061e-06,
      "loss": 0.3936,
      "step": 30677
    },
    {
      "epoch": 1.4078289201964114,
      "grad_norm": 0.4969753623008728,
      "learning_rate": 5.650959678295663e-06,
      "loss": 0.3591,
      "step": 30678
    },
    {
      "epoch": 1.4078748107016659,
      "grad_norm": 0.4149496853351593,
      "learning_rate": 5.650716576446911e-06,
      "loss": 0.2859,
      "step": 30679
    },
    {
      "epoch": 1.4079207012069204,
      "grad_norm": 0.49920201301574707,
      "learning_rate": 5.650473473033391e-06,
      "loss": 0.4488,
      "step": 30680
    },
    {
      "epoch": 1.4079665917121749,
      "grad_norm": 0.4971385598182678,
      "learning_rate": 5.650230368055689e-06,
      "loss": 0.3979,
      "step": 30681
    },
    {
      "epoch": 1.4080124822174291,
      "grad_norm": 0.4720684885978699,
      "learning_rate": 5.649987261514386e-06,
      "loss": 0.3284,
      "step": 30682
    },
    {
      "epoch": 1.4080583727226836,
      "grad_norm": 0.48297712206840515,
      "learning_rate": 5.64974415341007e-06,
      "loss": 0.3899,
      "step": 30683
    },
    {
      "epoch": 1.4081042632279381,
      "grad_norm": 0.4882252514362335,
      "learning_rate": 5.649501043743324e-06,
      "loss": 0.4061,
      "step": 30684
    },
    {
      "epoch": 1.4081501537331926,
      "grad_norm": 0.4824000298976898,
      "learning_rate": 5.649257932514733e-06,
      "loss": 0.3629,
      "step": 30685
    },
    {
      "epoch": 1.4081960442384471,
      "grad_norm": 0.4858934283256531,
      "learning_rate": 5.6490148197248824e-06,
      "loss": 0.3759,
      "step": 30686
    },
    {
      "epoch": 1.4082419347437014,
      "grad_norm": 0.4833122491836548,
      "learning_rate": 5.648771705374355e-06,
      "loss": 0.2909,
      "step": 30687
    },
    {
      "epoch": 1.408287825248956,
      "grad_norm": 0.4304989278316498,
      "learning_rate": 5.648528589463736e-06,
      "loss": 0.296,
      "step": 30688
    },
    {
      "epoch": 1.4083337157542104,
      "grad_norm": 0.4500770568847656,
      "learning_rate": 5.648285471993611e-06,
      "loss": 0.2979,
      "step": 30689
    },
    {
      "epoch": 1.408379606259465,
      "grad_norm": 0.4763033092021942,
      "learning_rate": 5.648042352964564e-06,
      "loss": 0.3892,
      "step": 30690
    },
    {
      "epoch": 1.4084254967647194,
      "grad_norm": 0.49361369013786316,
      "learning_rate": 5.6477992323771805e-06,
      "loss": 0.3684,
      "step": 30691
    },
    {
      "epoch": 1.408471387269974,
      "grad_norm": 0.5281588435173035,
      "learning_rate": 5.647556110232044e-06,
      "loss": 0.4647,
      "step": 30692
    },
    {
      "epoch": 1.4085172777752284,
      "grad_norm": 0.4348994195461273,
      "learning_rate": 5.647312986529738e-06,
      "loss": 0.2558,
      "step": 30693
    },
    {
      "epoch": 1.408563168280483,
      "grad_norm": 0.4772498607635498,
      "learning_rate": 5.647069861270849e-06,
      "loss": 0.4025,
      "step": 30694
    },
    {
      "epoch": 1.4086090587857372,
      "grad_norm": 0.5009322166442871,
      "learning_rate": 5.6468267344559626e-06,
      "loss": 0.4053,
      "step": 30695
    },
    {
      "epoch": 1.4086549492909917,
      "grad_norm": 0.458627313375473,
      "learning_rate": 5.646583606085662e-06,
      "loss": 0.3244,
      "step": 30696
    },
    {
      "epoch": 1.4087008397962462,
      "grad_norm": 0.4330346882343292,
      "learning_rate": 5.64634047616053e-06,
      "loss": 0.3047,
      "step": 30697
    },
    {
      "epoch": 1.4087467303015007,
      "grad_norm": 0.48149943351745605,
      "learning_rate": 5.646097344681155e-06,
      "loss": 0.3571,
      "step": 30698
    },
    {
      "epoch": 1.4087926208067552,
      "grad_norm": 0.46391624212265015,
      "learning_rate": 5.645854211648119e-06,
      "loss": 0.332,
      "step": 30699
    },
    {
      "epoch": 1.4088385113120094,
      "grad_norm": 0.47340190410614014,
      "learning_rate": 5.645611077062008e-06,
      "loss": 0.3722,
      "step": 30700
    },
    {
      "epoch": 1.408884401817264,
      "grad_norm": 0.4731406271457672,
      "learning_rate": 5.645367940923405e-06,
      "loss": 0.3216,
      "step": 30701
    },
    {
      "epoch": 1.4089302923225184,
      "grad_norm": 0.4492392838001251,
      "learning_rate": 5.645124803232897e-06,
      "loss": 0.3504,
      "step": 30702
    },
    {
      "epoch": 1.408976182827773,
      "grad_norm": 0.4633786082267761,
      "learning_rate": 5.6448816639910675e-06,
      "loss": 0.3331,
      "step": 30703
    },
    {
      "epoch": 1.4090220733330274,
      "grad_norm": 0.4437253177165985,
      "learning_rate": 5.6446385231985e-06,
      "loss": 0.3103,
      "step": 30704
    },
    {
      "epoch": 1.409067963838282,
      "grad_norm": 0.48084208369255066,
      "learning_rate": 5.64439538085578e-06,
      "loss": 0.3622,
      "step": 30705
    },
    {
      "epoch": 1.4091138543435364,
      "grad_norm": 0.46057942509651184,
      "learning_rate": 5.644152236963494e-06,
      "loss": 0.325,
      "step": 30706
    },
    {
      "epoch": 1.409159744848791,
      "grad_norm": 0.43466031551361084,
      "learning_rate": 5.643909091522225e-06,
      "loss": 0.3006,
      "step": 30707
    },
    {
      "epoch": 1.4092056353540452,
      "grad_norm": 0.4857945740222931,
      "learning_rate": 5.643665944532557e-06,
      "loss": 0.4165,
      "step": 30708
    },
    {
      "epoch": 1.4092515258592997,
      "grad_norm": 0.4735317528247833,
      "learning_rate": 5.643422795995076e-06,
      "loss": 0.4095,
      "step": 30709
    },
    {
      "epoch": 1.4092974163645542,
      "grad_norm": 0.4613579213619232,
      "learning_rate": 5.643179645910365e-06,
      "loss": 0.355,
      "step": 30710
    },
    {
      "epoch": 1.4093433068698087,
      "grad_norm": 0.4570719301700592,
      "learning_rate": 5.642936494279011e-06,
      "loss": 0.3289,
      "step": 30711
    },
    {
      "epoch": 1.409389197375063,
      "grad_norm": 0.4825561046600342,
      "learning_rate": 5.642693341101598e-06,
      "loss": 0.3876,
      "step": 30712
    },
    {
      "epoch": 1.4094350878803175,
      "grad_norm": 0.5210810899734497,
      "learning_rate": 5.642450186378711e-06,
      "loss": 0.3863,
      "step": 30713
    },
    {
      "epoch": 1.409480978385572,
      "grad_norm": 0.4759581983089447,
      "learning_rate": 5.642207030110932e-06,
      "loss": 0.382,
      "step": 30714
    },
    {
      "epoch": 1.4095268688908265,
      "grad_norm": 0.46867358684539795,
      "learning_rate": 5.64196387229885e-06,
      "loss": 0.3337,
      "step": 30715
    },
    {
      "epoch": 1.409572759396081,
      "grad_norm": 0.46694397926330566,
      "learning_rate": 5.641720712943045e-06,
      "loss": 0.3879,
      "step": 30716
    },
    {
      "epoch": 1.4096186499013355,
      "grad_norm": 0.46989521384239197,
      "learning_rate": 5.641477552044107e-06,
      "loss": 0.3545,
      "step": 30717
    },
    {
      "epoch": 1.40966454040659,
      "grad_norm": 0.4884621798992157,
      "learning_rate": 5.641234389602617e-06,
      "loss": 0.3638,
      "step": 30718
    },
    {
      "epoch": 1.4097104309118444,
      "grad_norm": 0.5010775923728943,
      "learning_rate": 5.640991225619159e-06,
      "loss": 0.3973,
      "step": 30719
    },
    {
      "epoch": 1.4097563214170987,
      "grad_norm": 0.48933959007263184,
      "learning_rate": 5.640748060094322e-06,
      "loss": 0.3903,
      "step": 30720
    },
    {
      "epoch": 1.4098022119223532,
      "grad_norm": 0.48443931341171265,
      "learning_rate": 5.640504893028687e-06,
      "loss": 0.3875,
      "step": 30721
    },
    {
      "epoch": 1.4098481024276077,
      "grad_norm": 0.4882057011127472,
      "learning_rate": 5.640261724422838e-06,
      "loss": 0.3985,
      "step": 30722
    },
    {
      "epoch": 1.4098939929328622,
      "grad_norm": 0.4784524142742157,
      "learning_rate": 5.640018554277365e-06,
      "loss": 0.3646,
      "step": 30723
    },
    {
      "epoch": 1.4099398834381167,
      "grad_norm": 0.44518914818763733,
      "learning_rate": 5.639775382592847e-06,
      "loss": 0.3191,
      "step": 30724
    },
    {
      "epoch": 1.409985773943371,
      "grad_norm": 0.48890867829322815,
      "learning_rate": 5.6395322093698716e-06,
      "loss": 0.3787,
      "step": 30725
    },
    {
      "epoch": 1.4100316644486255,
      "grad_norm": 0.49693554639816284,
      "learning_rate": 5.639289034609024e-06,
      "loss": 0.3962,
      "step": 30726
    },
    {
      "epoch": 1.41007755495388,
      "grad_norm": 0.4443364143371582,
      "learning_rate": 5.639045858310886e-06,
      "loss": 0.3268,
      "step": 30727
    },
    {
      "epoch": 1.4101234454591345,
      "grad_norm": 0.4881758391857147,
      "learning_rate": 5.638802680476047e-06,
      "loss": 0.3997,
      "step": 30728
    },
    {
      "epoch": 1.410169335964389,
      "grad_norm": 0.4745207130908966,
      "learning_rate": 5.638559501105087e-06,
      "loss": 0.359,
      "step": 30729
    },
    {
      "epoch": 1.4102152264696435,
      "grad_norm": 0.4794755280017853,
      "learning_rate": 5.638316320198594e-06,
      "loss": 0.3756,
      "step": 30730
    },
    {
      "epoch": 1.410261116974898,
      "grad_norm": 0.4616711735725403,
      "learning_rate": 5.638073137757151e-06,
      "loss": 0.3417,
      "step": 30731
    },
    {
      "epoch": 1.4103070074801525,
      "grad_norm": 0.43901970982551575,
      "learning_rate": 5.637829953781344e-06,
      "loss": 0.281,
      "step": 30732
    },
    {
      "epoch": 1.4103528979854068,
      "grad_norm": 0.4389268755912781,
      "learning_rate": 5.637586768271758e-06,
      "loss": 0.3257,
      "step": 30733
    },
    {
      "epoch": 1.4103987884906612,
      "grad_norm": 0.46237727999687195,
      "learning_rate": 5.637343581228976e-06,
      "loss": 0.3599,
      "step": 30734
    },
    {
      "epoch": 1.4104446789959157,
      "grad_norm": 0.4500422775745392,
      "learning_rate": 5.637100392653585e-06,
      "loss": 0.341,
      "step": 30735
    },
    {
      "epoch": 1.4104905695011702,
      "grad_norm": 0.48443707823753357,
      "learning_rate": 5.636857202546167e-06,
      "loss": 0.4267,
      "step": 30736
    },
    {
      "epoch": 1.4105364600064247,
      "grad_norm": 0.4469890892505646,
      "learning_rate": 5.63661401090731e-06,
      "loss": 0.3115,
      "step": 30737
    },
    {
      "epoch": 1.410582350511679,
      "grad_norm": 0.5021235346794128,
      "learning_rate": 5.636370817737596e-06,
      "loss": 0.4222,
      "step": 30738
    },
    {
      "epoch": 1.4106282410169335,
      "grad_norm": 0.48171862959861755,
      "learning_rate": 5.636127623037612e-06,
      "loss": 0.3703,
      "step": 30739
    },
    {
      "epoch": 1.410674131522188,
      "grad_norm": 0.47213053703308105,
      "learning_rate": 5.635884426807941e-06,
      "loss": 0.3496,
      "step": 30740
    },
    {
      "epoch": 1.4107200220274425,
      "grad_norm": 0.4211561977863312,
      "learning_rate": 5.635641229049169e-06,
      "loss": 0.2877,
      "step": 30741
    },
    {
      "epoch": 1.410765912532697,
      "grad_norm": 0.4373224079608917,
      "learning_rate": 5.6353980297618805e-06,
      "loss": 0.3303,
      "step": 30742
    },
    {
      "epoch": 1.4108118030379515,
      "grad_norm": 0.4817468822002411,
      "learning_rate": 5.635154828946661e-06,
      "loss": 0.3652,
      "step": 30743
    },
    {
      "epoch": 1.410857693543206,
      "grad_norm": 0.46277809143066406,
      "learning_rate": 5.6349116266040925e-06,
      "loss": 0.362,
      "step": 30744
    },
    {
      "epoch": 1.4109035840484605,
      "grad_norm": 0.44873908162117004,
      "learning_rate": 5.6346684227347634e-06,
      "loss": 0.3154,
      "step": 30745
    },
    {
      "epoch": 1.4109494745537148,
      "grad_norm": 0.48887234926223755,
      "learning_rate": 5.634425217339258e-06,
      "loss": 0.4277,
      "step": 30746
    },
    {
      "epoch": 1.4109953650589693,
      "grad_norm": 0.463225781917572,
      "learning_rate": 5.634182010418158e-06,
      "loss": 0.3524,
      "step": 30747
    },
    {
      "epoch": 1.4110412555642238,
      "grad_norm": 0.46243152022361755,
      "learning_rate": 5.633938801972051e-06,
      "loss": 0.3589,
      "step": 30748
    },
    {
      "epoch": 1.4110871460694783,
      "grad_norm": 0.4806060194969177,
      "learning_rate": 5.633695592001522e-06,
      "loss": 0.3603,
      "step": 30749
    },
    {
      "epoch": 1.4111330365747328,
      "grad_norm": 0.4719722867012024,
      "learning_rate": 5.633452380507155e-06,
      "loss": 0.3512,
      "step": 30750
    },
    {
      "epoch": 1.411178927079987,
      "grad_norm": 0.4405514597892761,
      "learning_rate": 5.633209167489536e-06,
      "loss": 0.355,
      "step": 30751
    },
    {
      "epoch": 1.4112248175852415,
      "grad_norm": 0.47409629821777344,
      "learning_rate": 5.632965952949246e-06,
      "loss": 0.3436,
      "step": 30752
    },
    {
      "epoch": 1.411270708090496,
      "grad_norm": 0.4613935947418213,
      "learning_rate": 5.632722736886875e-06,
      "loss": 0.3478,
      "step": 30753
    },
    {
      "epoch": 1.4113165985957505,
      "grad_norm": 0.45856964588165283,
      "learning_rate": 5.632479519303005e-06,
      "loss": 0.3378,
      "step": 30754
    },
    {
      "epoch": 1.411362489101005,
      "grad_norm": 0.4496823847293854,
      "learning_rate": 5.63223630019822e-06,
      "loss": 0.3174,
      "step": 30755
    },
    {
      "epoch": 1.4114083796062595,
      "grad_norm": 0.44037240743637085,
      "learning_rate": 5.631993079573108e-06,
      "loss": 0.2915,
      "step": 30756
    },
    {
      "epoch": 1.411454270111514,
      "grad_norm": 0.444921612739563,
      "learning_rate": 5.631749857428253e-06,
      "loss": 0.3347,
      "step": 30757
    },
    {
      "epoch": 1.4115001606167685,
      "grad_norm": 0.4797303378582001,
      "learning_rate": 5.6315066337642355e-06,
      "loss": 0.3707,
      "step": 30758
    },
    {
      "epoch": 1.4115460511220228,
      "grad_norm": 0.5092024207115173,
      "learning_rate": 5.631263408581646e-06,
      "loss": 0.4175,
      "step": 30759
    },
    {
      "epoch": 1.4115919416272773,
      "grad_norm": 0.5355422496795654,
      "learning_rate": 5.631020181881067e-06,
      "loss": 0.3838,
      "step": 30760
    },
    {
      "epoch": 1.4116378321325318,
      "grad_norm": 0.4820060729980469,
      "learning_rate": 5.630776953663084e-06,
      "loss": 0.3458,
      "step": 30761
    },
    {
      "epoch": 1.4116837226377863,
      "grad_norm": 0.4782482385635376,
      "learning_rate": 5.6305337239282805e-06,
      "loss": 0.4286,
      "step": 30762
    },
    {
      "epoch": 1.4117296131430406,
      "grad_norm": 0.4851239025592804,
      "learning_rate": 5.630290492677244e-06,
      "loss": 0.3507,
      "step": 30763
    },
    {
      "epoch": 1.411775503648295,
      "grad_norm": 0.4915924072265625,
      "learning_rate": 5.630047259910556e-06,
      "loss": 0.3632,
      "step": 30764
    },
    {
      "epoch": 1.4118213941535496,
      "grad_norm": 0.44733747839927673,
      "learning_rate": 5.629804025628804e-06,
      "loss": 0.2955,
      "step": 30765
    },
    {
      "epoch": 1.411867284658804,
      "grad_norm": 0.43608808517456055,
      "learning_rate": 5.629560789832572e-06,
      "loss": 0.318,
      "step": 30766
    },
    {
      "epoch": 1.4119131751640586,
      "grad_norm": 0.5141822099685669,
      "learning_rate": 5.629317552522446e-06,
      "loss": 0.4107,
      "step": 30767
    },
    {
      "epoch": 1.411959065669313,
      "grad_norm": 0.48339518904685974,
      "learning_rate": 5.629074313699009e-06,
      "loss": 0.3796,
      "step": 30768
    },
    {
      "epoch": 1.4120049561745676,
      "grad_norm": 0.441035658121109,
      "learning_rate": 5.6288310733628475e-06,
      "loss": 0.3165,
      "step": 30769
    },
    {
      "epoch": 1.412050846679822,
      "grad_norm": 0.45699751377105713,
      "learning_rate": 5.628587831514544e-06,
      "loss": 0.3244,
      "step": 30770
    },
    {
      "epoch": 1.4120967371850763,
      "grad_norm": 0.4463905096054077,
      "learning_rate": 5.628344588154686e-06,
      "loss": 0.3185,
      "step": 30771
    },
    {
      "epoch": 1.4121426276903308,
      "grad_norm": 0.5068268179893494,
      "learning_rate": 5.628101343283858e-06,
      "loss": 0.4086,
      "step": 30772
    },
    {
      "epoch": 1.4121885181955853,
      "grad_norm": 0.49815142154693604,
      "learning_rate": 5.6278580969026444e-06,
      "loss": 0.376,
      "step": 30773
    },
    {
      "epoch": 1.4122344087008398,
      "grad_norm": 0.4651508927345276,
      "learning_rate": 5.6276148490116316e-06,
      "loss": 0.3421,
      "step": 30774
    },
    {
      "epoch": 1.4122802992060943,
      "grad_norm": 0.45601505041122437,
      "learning_rate": 5.6273715996114e-06,
      "loss": 0.3526,
      "step": 30775
    },
    {
      "epoch": 1.4123261897113486,
      "grad_norm": 0.4383566379547119,
      "learning_rate": 5.627128348702538e-06,
      "loss": 0.3144,
      "step": 30776
    },
    {
      "epoch": 1.412372080216603,
      "grad_norm": 0.4120192527770996,
      "learning_rate": 5.626885096285633e-06,
      "loss": 0.3014,
      "step": 30777
    },
    {
      "epoch": 1.4124179707218576,
      "grad_norm": 0.44541439414024353,
      "learning_rate": 5.626641842361267e-06,
      "loss": 0.2891,
      "step": 30778
    },
    {
      "epoch": 1.412463861227112,
      "grad_norm": 0.4743247926235199,
      "learning_rate": 5.626398586930023e-06,
      "loss": 0.347,
      "step": 30779
    },
    {
      "epoch": 1.4125097517323666,
      "grad_norm": 0.4687708616256714,
      "learning_rate": 5.62615532999249e-06,
      "loss": 0.3129,
      "step": 30780
    },
    {
      "epoch": 1.412555642237621,
      "grad_norm": 0.41564494371414185,
      "learning_rate": 5.625912071549248e-06,
      "loss": 0.2624,
      "step": 30781
    },
    {
      "epoch": 1.4126015327428756,
      "grad_norm": 0.43975740671157837,
      "learning_rate": 5.625668811600888e-06,
      "loss": 0.2809,
      "step": 30782
    },
    {
      "epoch": 1.41264742324813,
      "grad_norm": 0.4882320165634155,
      "learning_rate": 5.625425550147993e-06,
      "loss": 0.3371,
      "step": 30783
    },
    {
      "epoch": 1.4126933137533844,
      "grad_norm": 0.451289564371109,
      "learning_rate": 5.625182287191145e-06,
      "loss": 0.3021,
      "step": 30784
    },
    {
      "epoch": 1.4127392042586389,
      "grad_norm": 0.46049511432647705,
      "learning_rate": 5.624939022730931e-06,
      "loss": 0.3398,
      "step": 30785
    },
    {
      "epoch": 1.4127850947638934,
      "grad_norm": 0.4688246250152588,
      "learning_rate": 5.624695756767935e-06,
      "loss": 0.3753,
      "step": 30786
    },
    {
      "epoch": 1.4128309852691479,
      "grad_norm": 0.4541654586791992,
      "learning_rate": 5.624452489302743e-06,
      "loss": 0.3402,
      "step": 30787
    },
    {
      "epoch": 1.4128768757744024,
      "grad_norm": 0.5195730924606323,
      "learning_rate": 5.624209220335941e-06,
      "loss": 0.4128,
      "step": 30788
    },
    {
      "epoch": 1.4129227662796566,
      "grad_norm": 0.5064204335212708,
      "learning_rate": 5.623965949868113e-06,
      "loss": 0.4544,
      "step": 30789
    },
    {
      "epoch": 1.4129686567849111,
      "grad_norm": 0.4945853650569916,
      "learning_rate": 5.623722677899842e-06,
      "loss": 0.3834,
      "step": 30790
    },
    {
      "epoch": 1.4130145472901656,
      "grad_norm": 0.45019251108169556,
      "learning_rate": 5.623479404431717e-06,
      "loss": 0.3206,
      "step": 30791
    },
    {
      "epoch": 1.4130604377954201,
      "grad_norm": 0.48744407296180725,
      "learning_rate": 5.623236129464319e-06,
      "loss": 0.427,
      "step": 30792
    },
    {
      "epoch": 1.4131063283006746,
      "grad_norm": 0.44211477041244507,
      "learning_rate": 5.622992852998237e-06,
      "loss": 0.3379,
      "step": 30793
    },
    {
      "epoch": 1.4131522188059291,
      "grad_norm": 0.4443802535533905,
      "learning_rate": 5.622749575034052e-06,
      "loss": 0.282,
      "step": 30794
    },
    {
      "epoch": 1.4131981093111836,
      "grad_norm": 0.4637661576271057,
      "learning_rate": 5.622506295572351e-06,
      "loss": 0.3568,
      "step": 30795
    },
    {
      "epoch": 1.4132439998164381,
      "grad_norm": 0.45468705892562866,
      "learning_rate": 5.622263014613719e-06,
      "loss": 0.3247,
      "step": 30796
    },
    {
      "epoch": 1.4132898903216924,
      "grad_norm": 0.4522338807582855,
      "learning_rate": 5.622019732158741e-06,
      "loss": 0.3029,
      "step": 30797
    },
    {
      "epoch": 1.4133357808269469,
      "grad_norm": 0.45286786556243896,
      "learning_rate": 5.621776448208002e-06,
      "loss": 0.3419,
      "step": 30798
    },
    {
      "epoch": 1.4133816713322014,
      "grad_norm": 0.5250926613807678,
      "learning_rate": 5.621533162762086e-06,
      "loss": 0.3537,
      "step": 30799
    },
    {
      "epoch": 1.4134275618374559,
      "grad_norm": 0.4815371036529541,
      "learning_rate": 5.621289875821581e-06,
      "loss": 0.3945,
      "step": 30800
    },
    {
      "epoch": 1.4134734523427102,
      "grad_norm": 0.4450767934322357,
      "learning_rate": 5.6210465873870675e-06,
      "loss": 0.3141,
      "step": 30801
    },
    {
      "epoch": 1.4135193428479647,
      "grad_norm": 0.45520344376564026,
      "learning_rate": 5.620803297459134e-06,
      "loss": 0.3443,
      "step": 30802
    },
    {
      "epoch": 1.4135652333532192,
      "grad_norm": 0.4648921489715576,
      "learning_rate": 5.620560006038365e-06,
      "loss": 0.3363,
      "step": 30803
    },
    {
      "epoch": 1.4136111238584737,
      "grad_norm": 0.43136176466941833,
      "learning_rate": 5.620316713125344e-06,
      "loss": 0.2792,
      "step": 30804
    },
    {
      "epoch": 1.4136570143637281,
      "grad_norm": 0.4508459270000458,
      "learning_rate": 5.620073418720659e-06,
      "loss": 0.3164,
      "step": 30805
    },
    {
      "epoch": 1.4137029048689826,
      "grad_norm": 0.4485633075237274,
      "learning_rate": 5.61983012282489e-06,
      "loss": 0.355,
      "step": 30806
    },
    {
      "epoch": 1.4137487953742371,
      "grad_norm": 0.4914606511592865,
      "learning_rate": 5.619586825438628e-06,
      "loss": 0.3541,
      "step": 30807
    },
    {
      "epoch": 1.4137946858794916,
      "grad_norm": 0.4921012818813324,
      "learning_rate": 5.619343526562454e-06,
      "loss": 0.418,
      "step": 30808
    },
    {
      "epoch": 1.413840576384746,
      "grad_norm": 0.5061198472976685,
      "learning_rate": 5.619100226196954e-06,
      "loss": 0.4026,
      "step": 30809
    },
    {
      "epoch": 1.4138864668900004,
      "grad_norm": 0.5178914070129395,
      "learning_rate": 5.6188569243427135e-06,
      "loss": 0.4659,
      "step": 30810
    },
    {
      "epoch": 1.413932357395255,
      "grad_norm": 0.4768790006637573,
      "learning_rate": 5.618613621000319e-06,
      "loss": 0.3339,
      "step": 30811
    },
    {
      "epoch": 1.4139782479005094,
      "grad_norm": 0.49426522850990295,
      "learning_rate": 5.618370316170352e-06,
      "loss": 0.4321,
      "step": 30812
    },
    {
      "epoch": 1.414024138405764,
      "grad_norm": 0.4731158912181854,
      "learning_rate": 5.618127009853401e-06,
      "loss": 0.383,
      "step": 30813
    },
    {
      "epoch": 1.4140700289110182,
      "grad_norm": 0.48745959997177124,
      "learning_rate": 5.617883702050047e-06,
      "loss": 0.4198,
      "step": 30814
    },
    {
      "epoch": 1.4141159194162727,
      "grad_norm": 0.44403496384620667,
      "learning_rate": 5.6176403927608814e-06,
      "loss": 0.3322,
      "step": 30815
    },
    {
      "epoch": 1.4141618099215272,
      "grad_norm": 0.4661827087402344,
      "learning_rate": 5.617397081986484e-06,
      "loss": 0.3352,
      "step": 30816
    },
    {
      "epoch": 1.4142077004267817,
      "grad_norm": 0.4513093829154968,
      "learning_rate": 5.61715376972744e-06,
      "loss": 0.3203,
      "step": 30817
    },
    {
      "epoch": 1.4142535909320362,
      "grad_norm": 0.4757383167743683,
      "learning_rate": 5.616910455984337e-06,
      "loss": 0.4375,
      "step": 30818
    },
    {
      "epoch": 1.4142994814372907,
      "grad_norm": 0.4529770612716675,
      "learning_rate": 5.61666714075776e-06,
      "loss": 0.3456,
      "step": 30819
    },
    {
      "epoch": 1.4143453719425452,
      "grad_norm": 0.49846771359443665,
      "learning_rate": 5.616423824048291e-06,
      "loss": 0.4144,
      "step": 30820
    },
    {
      "epoch": 1.4143912624477997,
      "grad_norm": 0.46628206968307495,
      "learning_rate": 5.61618050585652e-06,
      "loss": 0.375,
      "step": 30821
    },
    {
      "epoch": 1.414437152953054,
      "grad_norm": 0.46206483244895935,
      "learning_rate": 5.615937186183029e-06,
      "loss": 0.3679,
      "step": 30822
    },
    {
      "epoch": 1.4144830434583084,
      "grad_norm": 0.48857581615448,
      "learning_rate": 5.615693865028402e-06,
      "loss": 0.3742,
      "step": 30823
    },
    {
      "epoch": 1.414528933963563,
      "grad_norm": 0.46295231580734253,
      "learning_rate": 5.615450542393225e-06,
      "loss": 0.2921,
      "step": 30824
    },
    {
      "epoch": 1.4145748244688174,
      "grad_norm": 0.49245187640190125,
      "learning_rate": 5.615207218278085e-06,
      "loss": 0.3819,
      "step": 30825
    },
    {
      "epoch": 1.414620714974072,
      "grad_norm": 0.449907124042511,
      "learning_rate": 5.614963892683567e-06,
      "loss": 0.3027,
      "step": 30826
    },
    {
      "epoch": 1.4146666054793262,
      "grad_norm": 0.4931392967700958,
      "learning_rate": 5.614720565610252e-06,
      "loss": 0.3888,
      "step": 30827
    },
    {
      "epoch": 1.4147124959845807,
      "grad_norm": 0.47935810685157776,
      "learning_rate": 5.61447723705873e-06,
      "loss": 0.3843,
      "step": 30828
    },
    {
      "epoch": 1.4147583864898352,
      "grad_norm": 0.48280128836631775,
      "learning_rate": 5.614233907029583e-06,
      "loss": 0.4114,
      "step": 30829
    },
    {
      "epoch": 1.4148042769950897,
      "grad_norm": 0.5052943229675293,
      "learning_rate": 5.613990575523398e-06,
      "loss": 0.4056,
      "step": 30830
    },
    {
      "epoch": 1.4148501675003442,
      "grad_norm": 0.48071202635765076,
      "learning_rate": 5.6137472425407604e-06,
      "loss": 0.3942,
      "step": 30831
    },
    {
      "epoch": 1.4148960580055987,
      "grad_norm": 0.4723033010959625,
      "learning_rate": 5.613503908082254e-06,
      "loss": 0.2998,
      "step": 30832
    },
    {
      "epoch": 1.4149419485108532,
      "grad_norm": 0.6170492172241211,
      "learning_rate": 5.6132605721484635e-06,
      "loss": 0.4795,
      "step": 30833
    },
    {
      "epoch": 1.4149878390161077,
      "grad_norm": 0.45418781042099,
      "learning_rate": 5.613017234739975e-06,
      "loss": 0.3166,
      "step": 30834
    },
    {
      "epoch": 1.415033729521362,
      "grad_norm": 0.44432970881462097,
      "learning_rate": 5.612773895857373e-06,
      "loss": 0.3242,
      "step": 30835
    },
    {
      "epoch": 1.4150796200266165,
      "grad_norm": 0.48921099305152893,
      "learning_rate": 5.6125305555012445e-06,
      "loss": 0.4322,
      "step": 30836
    },
    {
      "epoch": 1.415125510531871,
      "grad_norm": 0.47543594241142273,
      "learning_rate": 5.612287213672173e-06,
      "loss": 0.3817,
      "step": 30837
    },
    {
      "epoch": 1.4151714010371255,
      "grad_norm": 0.48448383808135986,
      "learning_rate": 5.612043870370744e-06,
      "loss": 0.358,
      "step": 30838
    },
    {
      "epoch": 1.41521729154238,
      "grad_norm": 0.47895464301109314,
      "learning_rate": 5.611800525597541e-06,
      "loss": 0.3935,
      "step": 30839
    },
    {
      "epoch": 1.4152631820476342,
      "grad_norm": 0.44931748509407043,
      "learning_rate": 5.611557179353153e-06,
      "loss": 0.3232,
      "step": 30840
    },
    {
      "epoch": 1.4153090725528887,
      "grad_norm": 0.5104110240936279,
      "learning_rate": 5.611313831638161e-06,
      "loss": 0.387,
      "step": 30841
    },
    {
      "epoch": 1.4153549630581432,
      "grad_norm": 0.47373420000076294,
      "learning_rate": 5.611070482453154e-06,
      "loss": 0.4025,
      "step": 30842
    },
    {
      "epoch": 1.4154008535633977,
      "grad_norm": 0.487661749124527,
      "learning_rate": 5.610827131798715e-06,
      "loss": 0.3783,
      "step": 30843
    },
    {
      "epoch": 1.4154467440686522,
      "grad_norm": 0.46824395656585693,
      "learning_rate": 5.610583779675429e-06,
      "loss": 0.3652,
      "step": 30844
    },
    {
      "epoch": 1.4154926345739067,
      "grad_norm": 0.49245521426200867,
      "learning_rate": 5.610340426083882e-06,
      "loss": 0.3817,
      "step": 30845
    },
    {
      "epoch": 1.4155385250791612,
      "grad_norm": 0.4526039659976959,
      "learning_rate": 5.610097071024658e-06,
      "loss": 0.3703,
      "step": 30846
    },
    {
      "epoch": 1.4155844155844157,
      "grad_norm": 0.48301881551742554,
      "learning_rate": 5.609853714498346e-06,
      "loss": 0.375,
      "step": 30847
    },
    {
      "epoch": 1.41563030608967,
      "grad_norm": 0.4459870457649231,
      "learning_rate": 5.609610356505526e-06,
      "loss": 0.2999,
      "step": 30848
    },
    {
      "epoch": 1.4156761965949245,
      "grad_norm": 0.47432318329811096,
      "learning_rate": 5.609366997046785e-06,
      "loss": 0.3177,
      "step": 30849
    },
    {
      "epoch": 1.415722087100179,
      "grad_norm": 0.43323707580566406,
      "learning_rate": 5.60912363612271e-06,
      "loss": 0.3027,
      "step": 30850
    },
    {
      "epoch": 1.4157679776054335,
      "grad_norm": 0.47800540924072266,
      "learning_rate": 5.6088802737338834e-06,
      "loss": 0.3604,
      "step": 30851
    },
    {
      "epoch": 1.4158138681106878,
      "grad_norm": 0.46721842885017395,
      "learning_rate": 5.608636909880892e-06,
      "loss": 0.3639,
      "step": 30852
    },
    {
      "epoch": 1.4158597586159423,
      "grad_norm": 0.46183836460113525,
      "learning_rate": 5.608393544564322e-06,
      "loss": 0.3257,
      "step": 30853
    },
    {
      "epoch": 1.4159056491211968,
      "grad_norm": 0.4921756684780121,
      "learning_rate": 5.608150177784758e-06,
      "loss": 0.3916,
      "step": 30854
    },
    {
      "epoch": 1.4159515396264513,
      "grad_norm": 0.4852762818336487,
      "learning_rate": 5.6079068095427835e-06,
      "loss": 0.3868,
      "step": 30855
    },
    {
      "epoch": 1.4159974301317058,
      "grad_norm": 0.44553765654563904,
      "learning_rate": 5.607663439838986e-06,
      "loss": 0.3186,
      "step": 30856
    },
    {
      "epoch": 1.4160433206369603,
      "grad_norm": 0.47788533568382263,
      "learning_rate": 5.607420068673949e-06,
      "loss": 0.3654,
      "step": 30857
    },
    {
      "epoch": 1.4160892111422148,
      "grad_norm": 0.45754408836364746,
      "learning_rate": 5.6071766960482585e-06,
      "loss": 0.3741,
      "step": 30858
    },
    {
      "epoch": 1.4161351016474693,
      "grad_norm": 0.498123437166214,
      "learning_rate": 5.606933321962501e-06,
      "loss": 0.3773,
      "step": 30859
    },
    {
      "epoch": 1.4161809921527235,
      "grad_norm": 0.4949764907360077,
      "learning_rate": 5.606689946417258e-06,
      "loss": 0.4187,
      "step": 30860
    },
    {
      "epoch": 1.416226882657978,
      "grad_norm": 0.47138383984565735,
      "learning_rate": 5.606446569413118e-06,
      "loss": 0.3677,
      "step": 30861
    },
    {
      "epoch": 1.4162727731632325,
      "grad_norm": 0.48738282918930054,
      "learning_rate": 5.606203190950667e-06,
      "loss": 0.4039,
      "step": 30862
    },
    {
      "epoch": 1.416318663668487,
      "grad_norm": 0.4584967792034149,
      "learning_rate": 5.605959811030487e-06,
      "loss": 0.3289,
      "step": 30863
    },
    {
      "epoch": 1.4163645541737415,
      "grad_norm": 0.4889296293258667,
      "learning_rate": 5.605716429653166e-06,
      "loss": 0.3874,
      "step": 30864
    },
    {
      "epoch": 1.4164104446789958,
      "grad_norm": 0.47574689984321594,
      "learning_rate": 5.605473046819287e-06,
      "loss": 0.2948,
      "step": 30865
    },
    {
      "epoch": 1.4164563351842503,
      "grad_norm": 0.4602375328540802,
      "learning_rate": 5.6052296625294375e-06,
      "loss": 0.2953,
      "step": 30866
    },
    {
      "epoch": 1.4165022256895048,
      "grad_norm": 0.44930317997932434,
      "learning_rate": 5.604986276784201e-06,
      "loss": 0.362,
      "step": 30867
    },
    {
      "epoch": 1.4165481161947593,
      "grad_norm": 0.46380865573883057,
      "learning_rate": 5.604742889584164e-06,
      "loss": 0.3139,
      "step": 30868
    },
    {
      "epoch": 1.4165940067000138,
      "grad_norm": 0.49410802125930786,
      "learning_rate": 5.60449950092991e-06,
      "loss": 0.4068,
      "step": 30869
    },
    {
      "epoch": 1.4166398972052683,
      "grad_norm": 0.49883976578712463,
      "learning_rate": 5.604256110822027e-06,
      "loss": 0.3873,
      "step": 30870
    },
    {
      "epoch": 1.4166857877105228,
      "grad_norm": 0.47129490971565247,
      "learning_rate": 5.604012719261098e-06,
      "loss": 0.3455,
      "step": 30871
    },
    {
      "epoch": 1.4167316782157773,
      "grad_norm": 0.45425158739089966,
      "learning_rate": 5.603769326247709e-06,
      "loss": 0.3247,
      "step": 30872
    },
    {
      "epoch": 1.4167775687210316,
      "grad_norm": 0.5123706459999084,
      "learning_rate": 5.603525931782445e-06,
      "loss": 0.3211,
      "step": 30873
    },
    {
      "epoch": 1.416823459226286,
      "grad_norm": 0.68637615442276,
      "learning_rate": 5.603282535865891e-06,
      "loss": 0.4701,
      "step": 30874
    },
    {
      "epoch": 1.4168693497315406,
      "grad_norm": 0.4929622411727905,
      "learning_rate": 5.603039138498635e-06,
      "loss": 0.3703,
      "step": 30875
    },
    {
      "epoch": 1.416915240236795,
      "grad_norm": 0.49409058690071106,
      "learning_rate": 5.602795739681259e-06,
      "loss": 0.3932,
      "step": 30876
    },
    {
      "epoch": 1.4169611307420495,
      "grad_norm": 0.4800538122653961,
      "learning_rate": 5.60255233941435e-06,
      "loss": 0.3435,
      "step": 30877
    },
    {
      "epoch": 1.4170070212473038,
      "grad_norm": 0.47643357515335083,
      "learning_rate": 5.602308937698492e-06,
      "loss": 0.3738,
      "step": 30878
    },
    {
      "epoch": 1.4170529117525583,
      "grad_norm": 0.5361646413803101,
      "learning_rate": 5.602065534534273e-06,
      "loss": 0.4618,
      "step": 30879
    },
    {
      "epoch": 1.4170988022578128,
      "grad_norm": 0.5524381399154663,
      "learning_rate": 5.601822129922273e-06,
      "loss": 0.4279,
      "step": 30880
    },
    {
      "epoch": 1.4171446927630673,
      "grad_norm": 0.5012064576148987,
      "learning_rate": 5.601578723863085e-06,
      "loss": 0.3724,
      "step": 30881
    },
    {
      "epoch": 1.4171905832683218,
      "grad_norm": 0.49646779894828796,
      "learning_rate": 5.601335316357286e-06,
      "loss": 0.3688,
      "step": 30882
    },
    {
      "epoch": 1.4172364737735763,
      "grad_norm": 0.46183857321739197,
      "learning_rate": 5.601091907405467e-06,
      "loss": 0.3285,
      "step": 30883
    },
    {
      "epoch": 1.4172823642788308,
      "grad_norm": 0.4562946856021881,
      "learning_rate": 5.600848497008212e-06,
      "loss": 0.3202,
      "step": 30884
    },
    {
      "epoch": 1.4173282547840853,
      "grad_norm": 0.5227342844009399,
      "learning_rate": 5.600605085166106e-06,
      "loss": 0.4564,
      "step": 30885
    },
    {
      "epoch": 1.4173741452893396,
      "grad_norm": 0.45754769444465637,
      "learning_rate": 5.600361671879734e-06,
      "loss": 0.316,
      "step": 30886
    },
    {
      "epoch": 1.417420035794594,
      "grad_norm": 0.5234241485595703,
      "learning_rate": 5.600118257149681e-06,
      "loss": 0.4744,
      "step": 30887
    },
    {
      "epoch": 1.4174659262998486,
      "grad_norm": 0.4899563491344452,
      "learning_rate": 5.599874840976534e-06,
      "loss": 0.383,
      "step": 30888
    },
    {
      "epoch": 1.417511816805103,
      "grad_norm": 0.449245423078537,
      "learning_rate": 5.5996314233608766e-06,
      "loss": 0.339,
      "step": 30889
    },
    {
      "epoch": 1.4175577073103574,
      "grad_norm": 0.4816667139530182,
      "learning_rate": 5.599388004303296e-06,
      "loss": 0.3604,
      "step": 30890
    },
    {
      "epoch": 1.4176035978156118,
      "grad_norm": 0.46153369545936584,
      "learning_rate": 5.599144583804375e-06,
      "loss": 0.3398,
      "step": 30891
    },
    {
      "epoch": 1.4176494883208663,
      "grad_norm": 0.5026823878288269,
      "learning_rate": 5.598901161864701e-06,
      "loss": 0.4169,
      "step": 30892
    },
    {
      "epoch": 1.4176953788261208,
      "grad_norm": 0.5115267038345337,
      "learning_rate": 5.5986577384848595e-06,
      "loss": 0.4208,
      "step": 30893
    },
    {
      "epoch": 1.4177412693313753,
      "grad_norm": 0.49240148067474365,
      "learning_rate": 5.5984143136654335e-06,
      "loss": 0.3377,
      "step": 30894
    },
    {
      "epoch": 1.4177871598366298,
      "grad_norm": 0.43318167328834534,
      "learning_rate": 5.59817088740701e-06,
      "loss": 0.3134,
      "step": 30895
    },
    {
      "epoch": 1.4178330503418843,
      "grad_norm": 0.46684059500694275,
      "learning_rate": 5.597927459710176e-06,
      "loss": 0.35,
      "step": 30896
    },
    {
      "epoch": 1.4178789408471388,
      "grad_norm": 0.4637692868709564,
      "learning_rate": 5.597684030575515e-06,
      "loss": 0.4069,
      "step": 30897
    },
    {
      "epoch": 1.4179248313523931,
      "grad_norm": 0.4730170667171478,
      "learning_rate": 5.597440600003612e-06,
      "loss": 0.3423,
      "step": 30898
    },
    {
      "epoch": 1.4179707218576476,
      "grad_norm": 0.4782581925392151,
      "learning_rate": 5.597197167995052e-06,
      "loss": 0.3814,
      "step": 30899
    },
    {
      "epoch": 1.418016612362902,
      "grad_norm": 0.5052204132080078,
      "learning_rate": 5.596953734550422e-06,
      "loss": 0.4053,
      "step": 30900
    },
    {
      "epoch": 1.4180625028681566,
      "grad_norm": 0.5245323777198792,
      "learning_rate": 5.5967102996703075e-06,
      "loss": 0.4144,
      "step": 30901
    },
    {
      "epoch": 1.418108393373411,
      "grad_norm": 0.508316159248352,
      "learning_rate": 5.596466863355291e-06,
      "loss": 0.3997,
      "step": 30902
    },
    {
      "epoch": 1.4181542838786654,
      "grad_norm": 0.47295576333999634,
      "learning_rate": 5.596223425605962e-06,
      "loss": 0.349,
      "step": 30903
    },
    {
      "epoch": 1.4182001743839199,
      "grad_norm": 0.4769926965236664,
      "learning_rate": 5.5959799864229035e-06,
      "loss": 0.3725,
      "step": 30904
    },
    {
      "epoch": 1.4182460648891744,
      "grad_norm": 0.5032974481582642,
      "learning_rate": 5.5957365458067e-06,
      "loss": 0.4556,
      "step": 30905
    },
    {
      "epoch": 1.4182919553944289,
      "grad_norm": 0.45182308554649353,
      "learning_rate": 5.595493103757939e-06,
      "loss": 0.3608,
      "step": 30906
    },
    {
      "epoch": 1.4183378458996834,
      "grad_norm": 0.4422386586666107,
      "learning_rate": 5.595249660277205e-06,
      "loss": 0.3116,
      "step": 30907
    },
    {
      "epoch": 1.4183837364049379,
      "grad_norm": 0.44333624839782715,
      "learning_rate": 5.595006215365084e-06,
      "loss": 0.3214,
      "step": 30908
    },
    {
      "epoch": 1.4184296269101924,
      "grad_norm": 0.47620120644569397,
      "learning_rate": 5.594762769022159e-06,
      "loss": 0.3668,
      "step": 30909
    },
    {
      "epoch": 1.4184755174154469,
      "grad_norm": 0.44359755516052246,
      "learning_rate": 5.594519321249019e-06,
      "loss": 0.3179,
      "step": 30910
    },
    {
      "epoch": 1.4185214079207011,
      "grad_norm": 0.4887865483760834,
      "learning_rate": 5.594275872046247e-06,
      "loss": 0.3646,
      "step": 30911
    },
    {
      "epoch": 1.4185672984259556,
      "grad_norm": 0.5266855955123901,
      "learning_rate": 5.5940324214144286e-06,
      "loss": 0.3124,
      "step": 30912
    },
    {
      "epoch": 1.4186131889312101,
      "grad_norm": 0.47665849328041077,
      "learning_rate": 5.59378896935415e-06,
      "loss": 0.3569,
      "step": 30913
    },
    {
      "epoch": 1.4186590794364646,
      "grad_norm": 0.4883550703525543,
      "learning_rate": 5.593545515865998e-06,
      "loss": 0.3918,
      "step": 30914
    },
    {
      "epoch": 1.4187049699417191,
      "grad_norm": 0.4549596309661865,
      "learning_rate": 5.593302060950555e-06,
      "loss": 0.2765,
      "step": 30915
    },
    {
      "epoch": 1.4187508604469734,
      "grad_norm": 0.4398154318332672,
      "learning_rate": 5.593058604608408e-06,
      "loss": 0.3291,
      "step": 30916
    },
    {
      "epoch": 1.418796750952228,
      "grad_norm": 0.5812874436378479,
      "learning_rate": 5.592815146840141e-06,
      "loss": 0.5129,
      "step": 30917
    },
    {
      "epoch": 1.4188426414574824,
      "grad_norm": 0.4330728054046631,
      "learning_rate": 5.592571687646342e-06,
      "loss": 0.3089,
      "step": 30918
    },
    {
      "epoch": 1.418888531962737,
      "grad_norm": 0.46022188663482666,
      "learning_rate": 5.592328227027596e-06,
      "loss": 0.2851,
      "step": 30919
    },
    {
      "epoch": 1.4189344224679914,
      "grad_norm": 0.4766254425048828,
      "learning_rate": 5.592084764984486e-06,
      "loss": 0.34,
      "step": 30920
    },
    {
      "epoch": 1.418980312973246,
      "grad_norm": 0.48329663276672363,
      "learning_rate": 5.5918413015176e-06,
      "loss": 0.3755,
      "step": 30921
    },
    {
      "epoch": 1.4190262034785004,
      "grad_norm": 0.42892536520957947,
      "learning_rate": 5.59159783662752e-06,
      "loss": 0.287,
      "step": 30922
    },
    {
      "epoch": 1.419072093983755,
      "grad_norm": 0.5095816850662231,
      "learning_rate": 5.591354370314835e-06,
      "loss": 0.3925,
      "step": 30923
    },
    {
      "epoch": 1.4191179844890092,
      "grad_norm": 0.5774825811386108,
      "learning_rate": 5.591110902580132e-06,
      "loss": 0.4593,
      "step": 30924
    },
    {
      "epoch": 1.4191638749942637,
      "grad_norm": 0.47800755500793457,
      "learning_rate": 5.590867433423992e-06,
      "loss": 0.382,
      "step": 30925
    },
    {
      "epoch": 1.4192097654995182,
      "grad_norm": 0.5057810544967651,
      "learning_rate": 5.590623962847001e-06,
      "loss": 0.3699,
      "step": 30926
    },
    {
      "epoch": 1.4192556560047727,
      "grad_norm": 0.4692411422729492,
      "learning_rate": 5.5903804908497486e-06,
      "loss": 0.3623,
      "step": 30927
    },
    {
      "epoch": 1.4193015465100272,
      "grad_norm": 0.43551987409591675,
      "learning_rate": 5.590137017432815e-06,
      "loss": 0.314,
      "step": 30928
    },
    {
      "epoch": 1.4193474370152814,
      "grad_norm": 0.48797404766082764,
      "learning_rate": 5.589893542596789e-06,
      "loss": 0.4336,
      "step": 30929
    },
    {
      "epoch": 1.419393327520536,
      "grad_norm": 0.4378354549407959,
      "learning_rate": 5.589650066342256e-06,
      "loss": 0.3055,
      "step": 30930
    },
    {
      "epoch": 1.4194392180257904,
      "grad_norm": 0.47631052136421204,
      "learning_rate": 5.5894065886698e-06,
      "loss": 0.3635,
      "step": 30931
    },
    {
      "epoch": 1.419485108531045,
      "grad_norm": 0.47139057517051697,
      "learning_rate": 5.589163109580007e-06,
      "loss": 0.3739,
      "step": 30932
    },
    {
      "epoch": 1.4195309990362994,
      "grad_norm": 0.44874078035354614,
      "learning_rate": 5.588919629073462e-06,
      "loss": 0.3272,
      "step": 30933
    },
    {
      "epoch": 1.419576889541554,
      "grad_norm": 0.4764311611652374,
      "learning_rate": 5.588676147150752e-06,
      "loss": 0.3375,
      "step": 30934
    },
    {
      "epoch": 1.4196227800468084,
      "grad_norm": 0.45631274580955505,
      "learning_rate": 5.588432663812462e-06,
      "loss": 0.3332,
      "step": 30935
    },
    {
      "epoch": 1.419668670552063,
      "grad_norm": 0.456581711769104,
      "learning_rate": 5.588189179059177e-06,
      "loss": 0.3428,
      "step": 30936
    },
    {
      "epoch": 1.4197145610573172,
      "grad_norm": 0.46663832664489746,
      "learning_rate": 5.587945692891482e-06,
      "loss": 0.3339,
      "step": 30937
    },
    {
      "epoch": 1.4197604515625717,
      "grad_norm": 0.4635070264339447,
      "learning_rate": 5.587702205309964e-06,
      "loss": 0.3667,
      "step": 30938
    },
    {
      "epoch": 1.4198063420678262,
      "grad_norm": 0.4456779360771179,
      "learning_rate": 5.587458716315207e-06,
      "loss": 0.3137,
      "step": 30939
    },
    {
      "epoch": 1.4198522325730807,
      "grad_norm": 0.4748457670211792,
      "learning_rate": 5.587215225907798e-06,
      "loss": 0.3526,
      "step": 30940
    },
    {
      "epoch": 1.419898123078335,
      "grad_norm": 0.4397049844264984,
      "learning_rate": 5.586971734088321e-06,
      "loss": 0.2946,
      "step": 30941
    },
    {
      "epoch": 1.4199440135835895,
      "grad_norm": 0.5057998895645142,
      "learning_rate": 5.586728240857362e-06,
      "loss": 0.3774,
      "step": 30942
    },
    {
      "epoch": 1.419989904088844,
      "grad_norm": 0.4631330668926239,
      "learning_rate": 5.586484746215507e-06,
      "loss": 0.3484,
      "step": 30943
    },
    {
      "epoch": 1.4200357945940985,
      "grad_norm": 0.4738018810749054,
      "learning_rate": 5.586241250163342e-06,
      "loss": 0.3793,
      "step": 30944
    },
    {
      "epoch": 1.420081685099353,
      "grad_norm": 0.4386574923992157,
      "learning_rate": 5.585997752701451e-06,
      "loss": 0.3163,
      "step": 30945
    },
    {
      "epoch": 1.4201275756046075,
      "grad_norm": 0.4675652086734772,
      "learning_rate": 5.585754253830421e-06,
      "loss": 0.3708,
      "step": 30946
    },
    {
      "epoch": 1.420173466109862,
      "grad_norm": 0.45826563239097595,
      "learning_rate": 5.585510753550837e-06,
      "loss": 0.3425,
      "step": 30947
    },
    {
      "epoch": 1.4202193566151164,
      "grad_norm": 0.3999793529510498,
      "learning_rate": 5.585267251863284e-06,
      "loss": 0.2985,
      "step": 30948
    },
    {
      "epoch": 1.4202652471203707,
      "grad_norm": 0.4877398610115051,
      "learning_rate": 5.585023748768348e-06,
      "loss": 0.349,
      "step": 30949
    },
    {
      "epoch": 1.4203111376256252,
      "grad_norm": 0.48880425095558167,
      "learning_rate": 5.584780244266613e-06,
      "loss": 0.3755,
      "step": 30950
    },
    {
      "epoch": 1.4203570281308797,
      "grad_norm": 0.47869059443473816,
      "learning_rate": 5.584536738358668e-06,
      "loss": 0.3465,
      "step": 30951
    },
    {
      "epoch": 1.4204029186361342,
      "grad_norm": 0.5100029110908508,
      "learning_rate": 5.584293231045097e-06,
      "loss": 0.3831,
      "step": 30952
    },
    {
      "epoch": 1.4204488091413887,
      "grad_norm": 0.46094179153442383,
      "learning_rate": 5.584049722326484e-06,
      "loss": 0.334,
      "step": 30953
    },
    {
      "epoch": 1.420494699646643,
      "grad_norm": 0.47298967838287354,
      "learning_rate": 5.583806212203416e-06,
      "loss": 0.3424,
      "step": 30954
    },
    {
      "epoch": 1.4205405901518975,
      "grad_norm": 0.4833521842956543,
      "learning_rate": 5.583562700676478e-06,
      "loss": 0.3494,
      "step": 30955
    },
    {
      "epoch": 1.420586480657152,
      "grad_norm": 0.440070778131485,
      "learning_rate": 5.583319187746255e-06,
      "loss": 0.336,
      "step": 30956
    },
    {
      "epoch": 1.4206323711624065,
      "grad_norm": 0.5047367811203003,
      "learning_rate": 5.583075673413336e-06,
      "loss": 0.3632,
      "step": 30957
    },
    {
      "epoch": 1.420678261667661,
      "grad_norm": 0.47020480036735535,
      "learning_rate": 5.582832157678303e-06,
      "loss": 0.347,
      "step": 30958
    },
    {
      "epoch": 1.4207241521729155,
      "grad_norm": 0.48552921414375305,
      "learning_rate": 5.582588640541742e-06,
      "loss": 0.3683,
      "step": 30959
    },
    {
      "epoch": 1.42077004267817,
      "grad_norm": 0.48459237813949585,
      "learning_rate": 5.582345122004239e-06,
      "loss": 0.4275,
      "step": 30960
    },
    {
      "epoch": 1.4208159331834245,
      "grad_norm": 0.46578142046928406,
      "learning_rate": 5.582101602066381e-06,
      "loss": 0.3575,
      "step": 30961
    },
    {
      "epoch": 1.4208618236886787,
      "grad_norm": 0.48758113384246826,
      "learning_rate": 5.581858080728752e-06,
      "loss": 0.3432,
      "step": 30962
    },
    {
      "epoch": 1.4209077141939332,
      "grad_norm": 0.4580640494823456,
      "learning_rate": 5.581614557991938e-06,
      "loss": 0.3834,
      "step": 30963
    },
    {
      "epoch": 1.4209536046991877,
      "grad_norm": 0.4301185607910156,
      "learning_rate": 5.581371033856524e-06,
      "loss": 0.2891,
      "step": 30964
    },
    {
      "epoch": 1.4209994952044422,
      "grad_norm": 0.4643731415271759,
      "learning_rate": 5.581127508323096e-06,
      "loss": 0.3061,
      "step": 30965
    },
    {
      "epoch": 1.4210453857096967,
      "grad_norm": 0.46613380312919617,
      "learning_rate": 5.58088398139224e-06,
      "loss": 0.3435,
      "step": 30966
    },
    {
      "epoch": 1.421091276214951,
      "grad_norm": 0.46103614568710327,
      "learning_rate": 5.580640453064541e-06,
      "loss": 0.3541,
      "step": 30967
    },
    {
      "epoch": 1.4211371667202055,
      "grad_norm": 0.4556327760219574,
      "learning_rate": 5.580396923340586e-06,
      "loss": 0.3281,
      "step": 30968
    },
    {
      "epoch": 1.42118305722546,
      "grad_norm": 0.4736635982990265,
      "learning_rate": 5.580153392220959e-06,
      "loss": 0.3556,
      "step": 30969
    },
    {
      "epoch": 1.4212289477307145,
      "grad_norm": 0.46170416474342346,
      "learning_rate": 5.579909859706246e-06,
      "loss": 0.3558,
      "step": 30970
    },
    {
      "epoch": 1.421274838235969,
      "grad_norm": 0.507839560508728,
      "learning_rate": 5.579666325797033e-06,
      "loss": 0.3766,
      "step": 30971
    },
    {
      "epoch": 1.4213207287412235,
      "grad_norm": 0.4694567322731018,
      "learning_rate": 5.579422790493906e-06,
      "loss": 0.3368,
      "step": 30972
    },
    {
      "epoch": 1.421366619246478,
      "grad_norm": 0.44808900356292725,
      "learning_rate": 5.57917925379745e-06,
      "loss": 0.3117,
      "step": 30973
    },
    {
      "epoch": 1.4214125097517325,
      "grad_norm": 0.48612546920776367,
      "learning_rate": 5.578935715708249e-06,
      "loss": 0.3724,
      "step": 30974
    },
    {
      "epoch": 1.4214584002569868,
      "grad_norm": 0.4376981854438782,
      "learning_rate": 5.578692176226892e-06,
      "loss": 0.2955,
      "step": 30975
    },
    {
      "epoch": 1.4215042907622413,
      "grad_norm": 0.44477537274360657,
      "learning_rate": 5.578448635353962e-06,
      "loss": 0.3214,
      "step": 30976
    },
    {
      "epoch": 1.4215501812674958,
      "grad_norm": 0.4642540216445923,
      "learning_rate": 5.5782050930900455e-06,
      "loss": 0.3536,
      "step": 30977
    },
    {
      "epoch": 1.4215960717727503,
      "grad_norm": 0.4852762520313263,
      "learning_rate": 5.577961549435729e-06,
      "loss": 0.372,
      "step": 30978
    },
    {
      "epoch": 1.4216419622780045,
      "grad_norm": 0.44843393564224243,
      "learning_rate": 5.577718004391598e-06,
      "loss": 0.3458,
      "step": 30979
    },
    {
      "epoch": 1.421687852783259,
      "grad_norm": 0.46489760279655457,
      "learning_rate": 5.577474457958238e-06,
      "loss": 0.4139,
      "step": 30980
    },
    {
      "epoch": 1.4217337432885135,
      "grad_norm": 0.48675012588500977,
      "learning_rate": 5.577230910136231e-06,
      "loss": 0.4099,
      "step": 30981
    },
    {
      "epoch": 1.421779633793768,
      "grad_norm": 0.4463357627391815,
      "learning_rate": 5.576987360926167e-06,
      "loss": 0.3154,
      "step": 30982
    },
    {
      "epoch": 1.4218255242990225,
      "grad_norm": 0.4637179374694824,
      "learning_rate": 5.576743810328632e-06,
      "loss": 0.3538,
      "step": 30983
    },
    {
      "epoch": 1.421871414804277,
      "grad_norm": 0.4994390606880188,
      "learning_rate": 5.576500258344211e-06,
      "loss": 0.4532,
      "step": 30984
    },
    {
      "epoch": 1.4219173053095315,
      "grad_norm": 0.43484872579574585,
      "learning_rate": 5.5762567049734864e-06,
      "loss": 0.2709,
      "step": 30985
    },
    {
      "epoch": 1.421963195814786,
      "grad_norm": 0.45192837715148926,
      "learning_rate": 5.576013150217046e-06,
      "loss": 0.3246,
      "step": 30986
    },
    {
      "epoch": 1.4220090863200403,
      "grad_norm": 0.48731496930122375,
      "learning_rate": 5.575769594075477e-06,
      "loss": 0.4002,
      "step": 30987
    },
    {
      "epoch": 1.4220549768252948,
      "grad_norm": 0.45242056250572205,
      "learning_rate": 5.575526036549364e-06,
      "loss": 0.3529,
      "step": 30988
    },
    {
      "epoch": 1.4221008673305493,
      "grad_norm": 0.4376673400402069,
      "learning_rate": 5.575282477639292e-06,
      "loss": 0.2943,
      "step": 30989
    },
    {
      "epoch": 1.4221467578358038,
      "grad_norm": 0.4633028209209442,
      "learning_rate": 5.575038917345848e-06,
      "loss": 0.356,
      "step": 30990
    },
    {
      "epoch": 1.4221926483410583,
      "grad_norm": 0.46812039613723755,
      "learning_rate": 5.574795355669616e-06,
      "loss": 0.3322,
      "step": 30991
    },
    {
      "epoch": 1.4222385388463126,
      "grad_norm": 0.44561320543289185,
      "learning_rate": 5.574551792611184e-06,
      "loss": 0.3323,
      "step": 30992
    },
    {
      "epoch": 1.422284429351567,
      "grad_norm": 0.4743368923664093,
      "learning_rate": 5.574308228171134e-06,
      "loss": 0.3734,
      "step": 30993
    },
    {
      "epoch": 1.4223303198568216,
      "grad_norm": 0.4766879975795746,
      "learning_rate": 5.574064662350055e-06,
      "loss": 0.3529,
      "step": 30994
    },
    {
      "epoch": 1.422376210362076,
      "grad_norm": 0.4752628803253174,
      "learning_rate": 5.573821095148533e-06,
      "loss": 0.3583,
      "step": 30995
    },
    {
      "epoch": 1.4224221008673306,
      "grad_norm": 0.4821039140224457,
      "learning_rate": 5.573577526567151e-06,
      "loss": 0.403,
      "step": 30996
    },
    {
      "epoch": 1.422467991372585,
      "grad_norm": 0.4664444327354431,
      "learning_rate": 5.573333956606497e-06,
      "loss": 0.3419,
      "step": 30997
    },
    {
      "epoch": 1.4225138818778396,
      "grad_norm": 0.4838399291038513,
      "learning_rate": 5.5730903852671536e-06,
      "loss": 0.3796,
      "step": 30998
    },
    {
      "epoch": 1.422559772383094,
      "grad_norm": 0.48404639959335327,
      "learning_rate": 5.57284681254971e-06,
      "loss": 0.3687,
      "step": 30999
    },
    {
      "epoch": 1.4226056628883483,
      "grad_norm": 0.5291709899902344,
      "learning_rate": 5.5726032384547516e-06,
      "loss": 0.4816,
      "step": 31000
    },
    {
      "epoch": 1.4226515533936028,
      "grad_norm": 0.490590900182724,
      "learning_rate": 5.572359662982862e-06,
      "loss": 0.3345,
      "step": 31001
    },
    {
      "epoch": 1.4226974438988573,
      "grad_norm": 0.48723292350769043,
      "learning_rate": 5.572116086134628e-06,
      "loss": 0.4069,
      "step": 31002
    },
    {
      "epoch": 1.4227433344041118,
      "grad_norm": 0.44026732444763184,
      "learning_rate": 5.571872507910636e-06,
      "loss": 0.2971,
      "step": 31003
    },
    {
      "epoch": 1.4227892249093663,
      "grad_norm": 0.5076603293418884,
      "learning_rate": 5.571628928311471e-06,
      "loss": 0.4092,
      "step": 31004
    },
    {
      "epoch": 1.4228351154146206,
      "grad_norm": 0.49760347604751587,
      "learning_rate": 5.571385347337718e-06,
      "loss": 0.4052,
      "step": 31005
    },
    {
      "epoch": 1.422881005919875,
      "grad_norm": 0.46346408128738403,
      "learning_rate": 5.571141764989966e-06,
      "loss": 0.3899,
      "step": 31006
    },
    {
      "epoch": 1.4229268964251296,
      "grad_norm": 0.465949147939682,
      "learning_rate": 5.570898181268795e-06,
      "loss": 0.3499,
      "step": 31007
    },
    {
      "epoch": 1.422972786930384,
      "grad_norm": 0.44134849309921265,
      "learning_rate": 5.570654596174796e-06,
      "loss": 0.3274,
      "step": 31008
    },
    {
      "epoch": 1.4230186774356386,
      "grad_norm": 0.48117873072624207,
      "learning_rate": 5.570411009708553e-06,
      "loss": 0.3799,
      "step": 31009
    },
    {
      "epoch": 1.423064567940893,
      "grad_norm": 0.4314451813697815,
      "learning_rate": 5.570167421870651e-06,
      "loss": 0.3018,
      "step": 31010
    },
    {
      "epoch": 1.4231104584461476,
      "grad_norm": 0.45402801036834717,
      "learning_rate": 5.569923832661676e-06,
      "loss": 0.3188,
      "step": 31011
    },
    {
      "epoch": 1.423156348951402,
      "grad_norm": 0.47225436568260193,
      "learning_rate": 5.569680242082215e-06,
      "loss": 0.3711,
      "step": 31012
    },
    {
      "epoch": 1.4232022394566564,
      "grad_norm": 0.4351010024547577,
      "learning_rate": 5.569436650132851e-06,
      "loss": 0.3159,
      "step": 31013
    },
    {
      "epoch": 1.4232481299619109,
      "grad_norm": 0.4244071841239929,
      "learning_rate": 5.569193056814174e-06,
      "loss": 0.2713,
      "step": 31014
    },
    {
      "epoch": 1.4232940204671654,
      "grad_norm": 0.4773484170436859,
      "learning_rate": 5.568949462126765e-06,
      "loss": 0.3774,
      "step": 31015
    },
    {
      "epoch": 1.4233399109724199,
      "grad_norm": 0.4773140251636505,
      "learning_rate": 5.568705866071212e-06,
      "loss": 0.3796,
      "step": 31016
    },
    {
      "epoch": 1.4233858014776744,
      "grad_norm": 0.458549439907074,
      "learning_rate": 5.568462268648104e-06,
      "loss": 0.3221,
      "step": 31017
    },
    {
      "epoch": 1.4234316919829286,
      "grad_norm": 0.4530276358127594,
      "learning_rate": 5.568218669858021e-06,
      "loss": 0.3012,
      "step": 31018
    },
    {
      "epoch": 1.4234775824881831,
      "grad_norm": 0.47206175327301025,
      "learning_rate": 5.5679750697015525e-06,
      "loss": 0.3511,
      "step": 31019
    },
    {
      "epoch": 1.4235234729934376,
      "grad_norm": 0.4554614722728729,
      "learning_rate": 5.5677314681792824e-06,
      "loss": 0.3371,
      "step": 31020
    },
    {
      "epoch": 1.4235693634986921,
      "grad_norm": 0.5060295462608337,
      "learning_rate": 5.567487865291798e-06,
      "loss": 0.3804,
      "step": 31021
    },
    {
      "epoch": 1.4236152540039466,
      "grad_norm": 0.4525175392627716,
      "learning_rate": 5.567244261039684e-06,
      "loss": 0.3095,
      "step": 31022
    },
    {
      "epoch": 1.4236611445092011,
      "grad_norm": 0.47621452808380127,
      "learning_rate": 5.567000655423528e-06,
      "loss": 0.3213,
      "step": 31023
    },
    {
      "epoch": 1.4237070350144556,
      "grad_norm": 0.5312054753303528,
      "learning_rate": 5.566757048443911e-06,
      "loss": 0.4759,
      "step": 31024
    },
    {
      "epoch": 1.4237529255197099,
      "grad_norm": 0.4629071056842804,
      "learning_rate": 5.566513440101424e-06,
      "loss": 0.3148,
      "step": 31025
    },
    {
      "epoch": 1.4237988160249644,
      "grad_norm": 0.4769214689731598,
      "learning_rate": 5.566269830396652e-06,
      "loss": 0.315,
      "step": 31026
    },
    {
      "epoch": 1.4238447065302189,
      "grad_norm": 0.4603554904460907,
      "learning_rate": 5.566026219330179e-06,
      "loss": 0.3008,
      "step": 31027
    },
    {
      "epoch": 1.4238905970354734,
      "grad_norm": 0.47094807028770447,
      "learning_rate": 5.565782606902593e-06,
      "loss": 0.3839,
      "step": 31028
    },
    {
      "epoch": 1.4239364875407279,
      "grad_norm": 0.47544899582862854,
      "learning_rate": 5.565538993114476e-06,
      "loss": 0.3589,
      "step": 31029
    },
    {
      "epoch": 1.4239823780459822,
      "grad_norm": 0.44159430265426636,
      "learning_rate": 5.5652953779664175e-06,
      "loss": 0.3506,
      "step": 31030
    },
    {
      "epoch": 1.4240282685512367,
      "grad_norm": 0.49555739760398865,
      "learning_rate": 5.565051761459001e-06,
      "loss": 0.4314,
      "step": 31031
    },
    {
      "epoch": 1.4240741590564912,
      "grad_norm": 0.49806538224220276,
      "learning_rate": 5.564808143592815e-06,
      "loss": 0.3826,
      "step": 31032
    },
    {
      "epoch": 1.4241200495617456,
      "grad_norm": 0.6368608474731445,
      "learning_rate": 5.564564524368443e-06,
      "loss": 0.4214,
      "step": 31033
    },
    {
      "epoch": 1.4241659400670001,
      "grad_norm": 0.4357253909111023,
      "learning_rate": 5.564320903786471e-06,
      "loss": 0.2951,
      "step": 31034
    },
    {
      "epoch": 1.4242118305722546,
      "grad_norm": 0.4655635356903076,
      "learning_rate": 5.564077281847485e-06,
      "loss": 0.3427,
      "step": 31035
    },
    {
      "epoch": 1.4242577210775091,
      "grad_norm": 0.4947997033596039,
      "learning_rate": 5.563833658552071e-06,
      "loss": 0.3758,
      "step": 31036
    },
    {
      "epoch": 1.4243036115827636,
      "grad_norm": 0.43814054131507874,
      "learning_rate": 5.563590033900816e-06,
      "loss": 0.3188,
      "step": 31037
    },
    {
      "epoch": 1.424349502088018,
      "grad_norm": 0.4973623752593994,
      "learning_rate": 5.563346407894304e-06,
      "loss": 0.4133,
      "step": 31038
    },
    {
      "epoch": 1.4243953925932724,
      "grad_norm": 0.46660125255584717,
      "learning_rate": 5.563102780533124e-06,
      "loss": 0.3931,
      "step": 31039
    },
    {
      "epoch": 1.424441283098527,
      "grad_norm": 0.4456760287284851,
      "learning_rate": 5.562859151817858e-06,
      "loss": 0.309,
      "step": 31040
    },
    {
      "epoch": 1.4244871736037814,
      "grad_norm": 0.4747001826763153,
      "learning_rate": 5.562615521749093e-06,
      "loss": 0.3349,
      "step": 31041
    },
    {
      "epoch": 1.424533064109036,
      "grad_norm": 0.4795725643634796,
      "learning_rate": 5.562371890327415e-06,
      "loss": 0.3747,
      "step": 31042
    },
    {
      "epoch": 1.4245789546142902,
      "grad_norm": 0.46141675114631653,
      "learning_rate": 5.562128257553411e-06,
      "loss": 0.3332,
      "step": 31043
    },
    {
      "epoch": 1.4246248451195447,
      "grad_norm": 0.4970603287220001,
      "learning_rate": 5.561884623427666e-06,
      "loss": 0.3582,
      "step": 31044
    },
    {
      "epoch": 1.4246707356247992,
      "grad_norm": 0.46043330430984497,
      "learning_rate": 5.561640987950765e-06,
      "loss": 0.3223,
      "step": 31045
    },
    {
      "epoch": 1.4247166261300537,
      "grad_norm": 0.46806102991104126,
      "learning_rate": 5.561397351123294e-06,
      "loss": 0.3244,
      "step": 31046
    },
    {
      "epoch": 1.4247625166353082,
      "grad_norm": 0.47579801082611084,
      "learning_rate": 5.56115371294584e-06,
      "loss": 0.4038,
      "step": 31047
    },
    {
      "epoch": 1.4248084071405627,
      "grad_norm": 0.48727986216545105,
      "learning_rate": 5.56091007341899e-06,
      "loss": 0.4001,
      "step": 31048
    },
    {
      "epoch": 1.4248542976458172,
      "grad_norm": 0.4433022439479828,
      "learning_rate": 5.560666432543327e-06,
      "loss": 0.3263,
      "step": 31049
    },
    {
      "epoch": 1.4249001881510717,
      "grad_norm": 0.4921528995037079,
      "learning_rate": 5.560422790319438e-06,
      "loss": 0.4102,
      "step": 31050
    },
    {
      "epoch": 1.424946078656326,
      "grad_norm": 0.47830930352211,
      "learning_rate": 5.560179146747911e-06,
      "loss": 0.3792,
      "step": 31051
    },
    {
      "epoch": 1.4249919691615804,
      "grad_norm": 0.48027366399765015,
      "learning_rate": 5.559935501829328e-06,
      "loss": 0.3962,
      "step": 31052
    },
    {
      "epoch": 1.425037859666835,
      "grad_norm": 0.4601075053215027,
      "learning_rate": 5.5596918555642765e-06,
      "loss": 0.3656,
      "step": 31053
    },
    {
      "epoch": 1.4250837501720894,
      "grad_norm": 0.482686847448349,
      "learning_rate": 5.559448207953344e-06,
      "loss": 0.3989,
      "step": 31054
    },
    {
      "epoch": 1.425129640677344,
      "grad_norm": 0.5216338634490967,
      "learning_rate": 5.559204558997115e-06,
      "loss": 0.4139,
      "step": 31055
    },
    {
      "epoch": 1.4251755311825982,
      "grad_norm": 0.5063871741294861,
      "learning_rate": 5.558960908696175e-06,
      "loss": 0.4114,
      "step": 31056
    },
    {
      "epoch": 1.4252214216878527,
      "grad_norm": 0.46411412954330444,
      "learning_rate": 5.558717257051111e-06,
      "loss": 0.3285,
      "step": 31057
    },
    {
      "epoch": 1.4252673121931072,
      "grad_norm": 0.4624720811843872,
      "learning_rate": 5.558473604062507e-06,
      "loss": 0.3206,
      "step": 31058
    },
    {
      "epoch": 1.4253132026983617,
      "grad_norm": 0.46868404746055603,
      "learning_rate": 5.558229949730951e-06,
      "loss": 0.3501,
      "step": 31059
    },
    {
      "epoch": 1.4253590932036162,
      "grad_norm": 0.49551478028297424,
      "learning_rate": 5.557986294057029e-06,
      "loss": 0.3578,
      "step": 31060
    },
    {
      "epoch": 1.4254049837088707,
      "grad_norm": 0.48992183804512024,
      "learning_rate": 5.557742637041326e-06,
      "loss": 0.4218,
      "step": 31061
    },
    {
      "epoch": 1.4254508742141252,
      "grad_norm": 0.46859657764434814,
      "learning_rate": 5.557498978684428e-06,
      "loss": 0.3568,
      "step": 31062
    },
    {
      "epoch": 1.4254967647193797,
      "grad_norm": 0.3955368101596832,
      "learning_rate": 5.55725531898692e-06,
      "loss": 0.2502,
      "step": 31063
    },
    {
      "epoch": 1.425542655224634,
      "grad_norm": 0.49963104724884033,
      "learning_rate": 5.557011657949388e-06,
      "loss": 0.4049,
      "step": 31064
    },
    {
      "epoch": 1.4255885457298885,
      "grad_norm": 0.45654359459877014,
      "learning_rate": 5.5567679955724206e-06,
      "loss": 0.3348,
      "step": 31065
    },
    {
      "epoch": 1.425634436235143,
      "grad_norm": 0.4742346405982971,
      "learning_rate": 5.5565243318566005e-06,
      "loss": 0.372,
      "step": 31066
    },
    {
      "epoch": 1.4256803267403975,
      "grad_norm": 0.48425766825675964,
      "learning_rate": 5.556280666802516e-06,
      "loss": 0.4301,
      "step": 31067
    },
    {
      "epoch": 1.4257262172456517,
      "grad_norm": 0.5147684812545776,
      "learning_rate": 5.556037000410752e-06,
      "loss": 0.3932,
      "step": 31068
    },
    {
      "epoch": 1.4257721077509062,
      "grad_norm": 0.4624393582344055,
      "learning_rate": 5.555793332681894e-06,
      "loss": 0.3342,
      "step": 31069
    },
    {
      "epoch": 1.4258179982561607,
      "grad_norm": 0.501838743686676,
      "learning_rate": 5.555549663616529e-06,
      "loss": 0.3893,
      "step": 31070
    },
    {
      "epoch": 1.4258638887614152,
      "grad_norm": 0.4835980534553528,
      "learning_rate": 5.555305993215242e-06,
      "loss": 0.3741,
      "step": 31071
    },
    {
      "epoch": 1.4259097792666697,
      "grad_norm": 0.47363075613975525,
      "learning_rate": 5.555062321478621e-06,
      "loss": 0.3694,
      "step": 31072
    },
    {
      "epoch": 1.4259556697719242,
      "grad_norm": 0.4904089868068695,
      "learning_rate": 5.554818648407247e-06,
      "loss": 0.3685,
      "step": 31073
    },
    {
      "epoch": 1.4260015602771787,
      "grad_norm": 0.4854147136211395,
      "learning_rate": 5.554574974001712e-06,
      "loss": 0.4037,
      "step": 31074
    },
    {
      "epoch": 1.4260474507824332,
      "grad_norm": 0.49199244379997253,
      "learning_rate": 5.554331298262597e-06,
      "loss": 0.3741,
      "step": 31075
    },
    {
      "epoch": 1.4260933412876875,
      "grad_norm": 0.4391774535179138,
      "learning_rate": 5.554087621190492e-06,
      "loss": 0.3082,
      "step": 31076
    },
    {
      "epoch": 1.426139231792942,
      "grad_norm": 0.4108867645263672,
      "learning_rate": 5.553843942785981e-06,
      "loss": 0.2799,
      "step": 31077
    },
    {
      "epoch": 1.4261851222981965,
      "grad_norm": 0.4756261110305786,
      "learning_rate": 5.55360026304965e-06,
      "loss": 0.3414,
      "step": 31078
    },
    {
      "epoch": 1.426231012803451,
      "grad_norm": 0.5060372352600098,
      "learning_rate": 5.553356581982085e-06,
      "loss": 0.4111,
      "step": 31079
    },
    {
      "epoch": 1.4262769033087055,
      "grad_norm": 0.48658743500709534,
      "learning_rate": 5.553112899583871e-06,
      "loss": 0.3694,
      "step": 31080
    },
    {
      "epoch": 1.4263227938139598,
      "grad_norm": 0.4890486001968384,
      "learning_rate": 5.552869215855595e-06,
      "loss": 0.383,
      "step": 31081
    },
    {
      "epoch": 1.4263686843192143,
      "grad_norm": 0.43703097105026245,
      "learning_rate": 5.552625530797845e-06,
      "loss": 0.2834,
      "step": 31082
    },
    {
      "epoch": 1.4264145748244688,
      "grad_norm": 0.43353942036628723,
      "learning_rate": 5.552381844411204e-06,
      "loss": 0.2827,
      "step": 31083
    },
    {
      "epoch": 1.4264604653297233,
      "grad_norm": 0.525862991809845,
      "learning_rate": 5.552138156696258e-06,
      "loss": 0.4573,
      "step": 31084
    },
    {
      "epoch": 1.4265063558349778,
      "grad_norm": 0.4623962938785553,
      "learning_rate": 5.551894467653596e-06,
      "loss": 0.3122,
      "step": 31085
    },
    {
      "epoch": 1.4265522463402323,
      "grad_norm": 0.41249749064445496,
      "learning_rate": 5.5516507772838e-06,
      "loss": 0.283,
      "step": 31086
    },
    {
      "epoch": 1.4265981368454868,
      "grad_norm": 0.44757911562919617,
      "learning_rate": 5.551407085587459e-06,
      "loss": 0.3486,
      "step": 31087
    },
    {
      "epoch": 1.4266440273507413,
      "grad_norm": 0.4539002776145935,
      "learning_rate": 5.551163392565159e-06,
      "loss": 0.285,
      "step": 31088
    },
    {
      "epoch": 1.4266899178559955,
      "grad_norm": 0.4802192449569702,
      "learning_rate": 5.550919698217483e-06,
      "loss": 0.3981,
      "step": 31089
    },
    {
      "epoch": 1.42673580836125,
      "grad_norm": 0.501649022102356,
      "learning_rate": 5.550676002545019e-06,
      "loss": 0.4167,
      "step": 31090
    },
    {
      "epoch": 1.4267816988665045,
      "grad_norm": 0.47513073682785034,
      "learning_rate": 5.550432305548355e-06,
      "loss": 0.3622,
      "step": 31091
    },
    {
      "epoch": 1.426827589371759,
      "grad_norm": 0.48464372754096985,
      "learning_rate": 5.5501886072280735e-06,
      "loss": 0.357,
      "step": 31092
    },
    {
      "epoch": 1.4268734798770135,
      "grad_norm": 0.4413989782333374,
      "learning_rate": 5.549944907584763e-06,
      "loss": 0.3072,
      "step": 31093
    },
    {
      "epoch": 1.4269193703822678,
      "grad_norm": 0.46637436747550964,
      "learning_rate": 5.549701206619008e-06,
      "loss": 0.3735,
      "step": 31094
    },
    {
      "epoch": 1.4269652608875223,
      "grad_norm": 0.4858378767967224,
      "learning_rate": 5.549457504331395e-06,
      "loss": 0.4421,
      "step": 31095
    },
    {
      "epoch": 1.4270111513927768,
      "grad_norm": 0.4464746415615082,
      "learning_rate": 5.549213800722511e-06,
      "loss": 0.3459,
      "step": 31096
    },
    {
      "epoch": 1.4270570418980313,
      "grad_norm": 0.4108494222164154,
      "learning_rate": 5.548970095792939e-06,
      "loss": 0.2617,
      "step": 31097
    },
    {
      "epoch": 1.4271029324032858,
      "grad_norm": 0.4329231083393097,
      "learning_rate": 5.54872638954327e-06,
      "loss": 0.2919,
      "step": 31098
    },
    {
      "epoch": 1.4271488229085403,
      "grad_norm": 0.43960508704185486,
      "learning_rate": 5.548482681974086e-06,
      "loss": 0.3043,
      "step": 31099
    },
    {
      "epoch": 1.4271947134137948,
      "grad_norm": 0.45808500051498413,
      "learning_rate": 5.548238973085973e-06,
      "loss": 0.3257,
      "step": 31100
    },
    {
      "epoch": 1.4272406039190493,
      "grad_norm": 0.5050986409187317,
      "learning_rate": 5.547995262879518e-06,
      "loss": 0.4834,
      "step": 31101
    },
    {
      "epoch": 1.4272864944243036,
      "grad_norm": 0.4525831341743469,
      "learning_rate": 5.547751551355311e-06,
      "loss": 0.3665,
      "step": 31102
    },
    {
      "epoch": 1.427332384929558,
      "grad_norm": 0.48764780163764954,
      "learning_rate": 5.54750783851393e-06,
      "loss": 0.3894,
      "step": 31103
    },
    {
      "epoch": 1.4273782754348125,
      "grad_norm": 0.48833557963371277,
      "learning_rate": 5.547264124355969e-06,
      "loss": 0.3572,
      "step": 31104
    },
    {
      "epoch": 1.427424165940067,
      "grad_norm": 0.4694660007953644,
      "learning_rate": 5.547020408882009e-06,
      "loss": 0.3097,
      "step": 31105
    },
    {
      "epoch": 1.4274700564453215,
      "grad_norm": 0.436752587556839,
      "learning_rate": 5.546776692092637e-06,
      "loss": 0.2971,
      "step": 31106
    },
    {
      "epoch": 1.4275159469505758,
      "grad_norm": 0.47643283009529114,
      "learning_rate": 5.54653297398844e-06,
      "loss": 0.3513,
      "step": 31107
    },
    {
      "epoch": 1.4275618374558303,
      "grad_norm": 0.4848998486995697,
      "learning_rate": 5.546289254570004e-06,
      "loss": 0.4039,
      "step": 31108
    },
    {
      "epoch": 1.4276077279610848,
      "grad_norm": 0.5048577785491943,
      "learning_rate": 5.546045533837915e-06,
      "loss": 0.4136,
      "step": 31109
    },
    {
      "epoch": 1.4276536184663393,
      "grad_norm": 0.5138494968414307,
      "learning_rate": 5.5458018117927595e-06,
      "loss": 0.412,
      "step": 31110
    },
    {
      "epoch": 1.4276995089715938,
      "grad_norm": 0.49820584058761597,
      "learning_rate": 5.545558088435121e-06,
      "loss": 0.45,
      "step": 31111
    },
    {
      "epoch": 1.4277453994768483,
      "grad_norm": 0.46815353631973267,
      "learning_rate": 5.545314363765588e-06,
      "loss": 0.3617,
      "step": 31112
    },
    {
      "epoch": 1.4277912899821028,
      "grad_norm": 0.46897268295288086,
      "learning_rate": 5.545070637784746e-06,
      "loss": 0.3874,
      "step": 31113
    },
    {
      "epoch": 1.427837180487357,
      "grad_norm": 0.4410501718521118,
      "learning_rate": 5.544826910493182e-06,
      "loss": 0.2994,
      "step": 31114
    },
    {
      "epoch": 1.4278830709926116,
      "grad_norm": 0.45134487748146057,
      "learning_rate": 5.544583181891481e-06,
      "loss": 0.3156,
      "step": 31115
    },
    {
      "epoch": 1.427928961497866,
      "grad_norm": 0.5071738958358765,
      "learning_rate": 5.544339451980229e-06,
      "loss": 0.4318,
      "step": 31116
    },
    {
      "epoch": 1.4279748520031206,
      "grad_norm": 0.4897975027561188,
      "learning_rate": 5.544095720760012e-06,
      "loss": 0.3989,
      "step": 31117
    },
    {
      "epoch": 1.428020742508375,
      "grad_norm": 0.4546639621257782,
      "learning_rate": 5.5438519882314165e-06,
      "loss": 0.3189,
      "step": 31118
    },
    {
      "epoch": 1.4280666330136293,
      "grad_norm": 0.47649550437927246,
      "learning_rate": 5.54360825439503e-06,
      "loss": 0.3675,
      "step": 31119
    },
    {
      "epoch": 1.4281125235188838,
      "grad_norm": 0.45490825176239014,
      "learning_rate": 5.543364519251437e-06,
      "loss": 0.3352,
      "step": 31120
    },
    {
      "epoch": 1.4281584140241383,
      "grad_norm": 0.42464736104011536,
      "learning_rate": 5.5431207828012226e-06,
      "loss": 0.3063,
      "step": 31121
    },
    {
      "epoch": 1.4282043045293928,
      "grad_norm": 0.503308892250061,
      "learning_rate": 5.542877045044975e-06,
      "loss": 0.4001,
      "step": 31122
    },
    {
      "epoch": 1.4282501950346473,
      "grad_norm": 0.4990735650062561,
      "learning_rate": 5.542633305983279e-06,
      "loss": 0.4327,
      "step": 31123
    },
    {
      "epoch": 1.4282960855399018,
      "grad_norm": 0.4514814615249634,
      "learning_rate": 5.54238956561672e-06,
      "loss": 0.328,
      "step": 31124
    },
    {
      "epoch": 1.4283419760451563,
      "grad_norm": 0.4717971682548523,
      "learning_rate": 5.542145823945888e-06,
      "loss": 0.3303,
      "step": 31125
    },
    {
      "epoch": 1.4283878665504108,
      "grad_norm": 0.4817515015602112,
      "learning_rate": 5.541902080971366e-06,
      "loss": 0.3877,
      "step": 31126
    },
    {
      "epoch": 1.428433757055665,
      "grad_norm": 0.47787919640541077,
      "learning_rate": 5.541658336693739e-06,
      "loss": 0.3638,
      "step": 31127
    },
    {
      "epoch": 1.4284796475609196,
      "grad_norm": 0.46937263011932373,
      "learning_rate": 5.541414591113595e-06,
      "loss": 0.3665,
      "step": 31128
    },
    {
      "epoch": 1.428525538066174,
      "grad_norm": 0.49982351064682007,
      "learning_rate": 5.54117084423152e-06,
      "loss": 0.3577,
      "step": 31129
    },
    {
      "epoch": 1.4285714285714286,
      "grad_norm": 0.4852502942085266,
      "learning_rate": 5.5409270960481e-06,
      "loss": 0.3897,
      "step": 31130
    },
    {
      "epoch": 1.428617319076683,
      "grad_norm": 0.43196403980255127,
      "learning_rate": 5.540683346563921e-06,
      "loss": 0.287,
      "step": 31131
    },
    {
      "epoch": 1.4286632095819374,
      "grad_norm": 0.4504820704460144,
      "learning_rate": 5.5404395957795686e-06,
      "loss": 0.3455,
      "step": 31132
    },
    {
      "epoch": 1.4287091000871919,
      "grad_norm": 0.4444904923439026,
      "learning_rate": 5.54019584369563e-06,
      "loss": 0.3208,
      "step": 31133
    },
    {
      "epoch": 1.4287549905924464,
      "grad_norm": 0.474321186542511,
      "learning_rate": 5.5399520903126915e-06,
      "loss": 0.3226,
      "step": 31134
    },
    {
      "epoch": 1.4288008810977009,
      "grad_norm": 0.4553077518939972,
      "learning_rate": 5.539708335631338e-06,
      "loss": 0.2677,
      "step": 31135
    },
    {
      "epoch": 1.4288467716029554,
      "grad_norm": 0.46007683873176575,
      "learning_rate": 5.5394645796521565e-06,
      "loss": 0.3416,
      "step": 31136
    },
    {
      "epoch": 1.4288926621082099,
      "grad_norm": 0.49282917380332947,
      "learning_rate": 5.5392208223757326e-06,
      "loss": 0.4046,
      "step": 31137
    },
    {
      "epoch": 1.4289385526134644,
      "grad_norm": 0.49968212842941284,
      "learning_rate": 5.538977063802653e-06,
      "loss": 0.4205,
      "step": 31138
    },
    {
      "epoch": 1.4289844431187189,
      "grad_norm": 0.4658733010292053,
      "learning_rate": 5.538733303933504e-06,
      "loss": 0.4071,
      "step": 31139
    },
    {
      "epoch": 1.4290303336239731,
      "grad_norm": 0.4665359556674957,
      "learning_rate": 5.538489542768871e-06,
      "loss": 0.3726,
      "step": 31140
    },
    {
      "epoch": 1.4290762241292276,
      "grad_norm": 0.4611313045024872,
      "learning_rate": 5.538245780309342e-06,
      "loss": 0.3474,
      "step": 31141
    },
    {
      "epoch": 1.4291221146344821,
      "grad_norm": 0.46552059054374695,
      "learning_rate": 5.5380020165555e-06,
      "loss": 0.3262,
      "step": 31142
    },
    {
      "epoch": 1.4291680051397366,
      "grad_norm": 0.5336297154426575,
      "learning_rate": 5.5377582515079335e-06,
      "loss": 0.4692,
      "step": 31143
    },
    {
      "epoch": 1.4292138956449911,
      "grad_norm": 0.4543493092060089,
      "learning_rate": 5.537514485167228e-06,
      "loss": 0.341,
      "step": 31144
    },
    {
      "epoch": 1.4292597861502454,
      "grad_norm": 0.4760379195213318,
      "learning_rate": 5.5372707175339694e-06,
      "loss": 0.3639,
      "step": 31145
    },
    {
      "epoch": 1.4293056766555,
      "grad_norm": 0.458626925945282,
      "learning_rate": 5.5370269486087435e-06,
      "loss": 0.345,
      "step": 31146
    },
    {
      "epoch": 1.4293515671607544,
      "grad_norm": 0.49291160702705383,
      "learning_rate": 5.53678317839214e-06,
      "loss": 0.4093,
      "step": 31147
    },
    {
      "epoch": 1.429397457666009,
      "grad_norm": 0.47705790400505066,
      "learning_rate": 5.5365394068847414e-06,
      "loss": 0.3618,
      "step": 31148
    },
    {
      "epoch": 1.4294433481712634,
      "grad_norm": 0.5001669526100159,
      "learning_rate": 5.536295634087133e-06,
      "loss": 0.3629,
      "step": 31149
    },
    {
      "epoch": 1.429489238676518,
      "grad_norm": 0.5112882852554321,
      "learning_rate": 5.536051859999906e-06,
      "loss": 0.3868,
      "step": 31150
    },
    {
      "epoch": 1.4295351291817724,
      "grad_norm": 0.5265322327613831,
      "learning_rate": 5.53580808462364e-06,
      "loss": 0.4315,
      "step": 31151
    },
    {
      "epoch": 1.4295810196870269,
      "grad_norm": 0.4835872948169708,
      "learning_rate": 5.535564307958927e-06,
      "loss": 0.3669,
      "step": 31152
    },
    {
      "epoch": 1.4296269101922812,
      "grad_norm": 0.5137470364570618,
      "learning_rate": 5.535320530006352e-06,
      "loss": 0.407,
      "step": 31153
    },
    {
      "epoch": 1.4296728006975357,
      "grad_norm": 0.4499948024749756,
      "learning_rate": 5.535076750766497e-06,
      "loss": 0.3054,
      "step": 31154
    },
    {
      "epoch": 1.4297186912027902,
      "grad_norm": 0.4675934314727783,
      "learning_rate": 5.534832970239952e-06,
      "loss": 0.3289,
      "step": 31155
    },
    {
      "epoch": 1.4297645817080447,
      "grad_norm": 0.4311208426952362,
      "learning_rate": 5.534589188427304e-06,
      "loss": 0.3085,
      "step": 31156
    },
    {
      "epoch": 1.429810472213299,
      "grad_norm": 0.45414820313453674,
      "learning_rate": 5.534345405329136e-06,
      "loss": 0.3176,
      "step": 31157
    },
    {
      "epoch": 1.4298563627185534,
      "grad_norm": 0.4811494052410126,
      "learning_rate": 5.534101620946037e-06,
      "loss": 0.3969,
      "step": 31158
    },
    {
      "epoch": 1.429902253223808,
      "grad_norm": 0.4965008795261383,
      "learning_rate": 5.5338578352785935e-06,
      "loss": 0.4224,
      "step": 31159
    },
    {
      "epoch": 1.4299481437290624,
      "grad_norm": 0.44886913895606995,
      "learning_rate": 5.533614048327387e-06,
      "loss": 0.2961,
      "step": 31160
    },
    {
      "epoch": 1.429994034234317,
      "grad_norm": 0.5075212121009827,
      "learning_rate": 5.53337026009301e-06,
      "loss": 0.4441,
      "step": 31161
    },
    {
      "epoch": 1.4300399247395714,
      "grad_norm": 0.489182710647583,
      "learning_rate": 5.533126470576044e-06,
      "loss": 0.3582,
      "step": 31162
    },
    {
      "epoch": 1.430085815244826,
      "grad_norm": 0.483430951833725,
      "learning_rate": 5.5328826797770786e-06,
      "loss": 0.3936,
      "step": 31163
    },
    {
      "epoch": 1.4301317057500804,
      "grad_norm": 0.4501614272594452,
      "learning_rate": 5.5326388876966985e-06,
      "loss": 0.3104,
      "step": 31164
    },
    {
      "epoch": 1.4301775962553347,
      "grad_norm": 0.5078386068344116,
      "learning_rate": 5.5323950943354885e-06,
      "loss": 0.41,
      "step": 31165
    },
    {
      "epoch": 1.4302234867605892,
      "grad_norm": 0.5015767216682434,
      "learning_rate": 5.532151299694037e-06,
      "loss": 0.3932,
      "step": 31166
    },
    {
      "epoch": 1.4302693772658437,
      "grad_norm": 0.4077334403991699,
      "learning_rate": 5.5319075037729296e-06,
      "loss": 0.2484,
      "step": 31167
    },
    {
      "epoch": 1.4303152677710982,
      "grad_norm": 0.4816370904445648,
      "learning_rate": 5.531663706572753e-06,
      "loss": 0.4079,
      "step": 31168
    },
    {
      "epoch": 1.4303611582763527,
      "grad_norm": 0.5312691926956177,
      "learning_rate": 5.531419908094093e-06,
      "loss": 0.4152,
      "step": 31169
    },
    {
      "epoch": 1.430407048781607,
      "grad_norm": 0.3909851312637329,
      "learning_rate": 5.531176108337536e-06,
      "loss": 0.2429,
      "step": 31170
    },
    {
      "epoch": 1.4304529392868615,
      "grad_norm": 0.45513197779655457,
      "learning_rate": 5.530932307303667e-06,
      "loss": 0.3629,
      "step": 31171
    },
    {
      "epoch": 1.430498829792116,
      "grad_norm": 0.46619272232055664,
      "learning_rate": 5.530688504993073e-06,
      "loss": 0.3643,
      "step": 31172
    },
    {
      "epoch": 1.4305447202973705,
      "grad_norm": 0.4528715908527374,
      "learning_rate": 5.530444701406342e-06,
      "loss": 0.3727,
      "step": 31173
    },
    {
      "epoch": 1.430590610802625,
      "grad_norm": 0.4613390564918518,
      "learning_rate": 5.530200896544059e-06,
      "loss": 0.3231,
      "step": 31174
    },
    {
      "epoch": 1.4306365013078794,
      "grad_norm": 0.45612916350364685,
      "learning_rate": 5.529957090406811e-06,
      "loss": 0.3004,
      "step": 31175
    },
    {
      "epoch": 1.430682391813134,
      "grad_norm": 0.48451218008995056,
      "learning_rate": 5.529713282995182e-06,
      "loss": 0.3987,
      "step": 31176
    },
    {
      "epoch": 1.4307282823183884,
      "grad_norm": 0.5055718421936035,
      "learning_rate": 5.5294694743097584e-06,
      "loss": 0.3429,
      "step": 31177
    },
    {
      "epoch": 1.4307741728236427,
      "grad_norm": 0.5270825028419495,
      "learning_rate": 5.52922566435113e-06,
      "loss": 0.4964,
      "step": 31178
    },
    {
      "epoch": 1.4308200633288972,
      "grad_norm": 0.48701563477516174,
      "learning_rate": 5.528981853119881e-06,
      "loss": 0.4508,
      "step": 31179
    },
    {
      "epoch": 1.4308659538341517,
      "grad_norm": 0.43516382575035095,
      "learning_rate": 5.528738040616596e-06,
      "loss": 0.2705,
      "step": 31180
    },
    {
      "epoch": 1.4309118443394062,
      "grad_norm": 0.49163514375686646,
      "learning_rate": 5.528494226841864e-06,
      "loss": 0.4035,
      "step": 31181
    },
    {
      "epoch": 1.4309577348446607,
      "grad_norm": 0.4601171612739563,
      "learning_rate": 5.52825041179627e-06,
      "loss": 0.4085,
      "step": 31182
    },
    {
      "epoch": 1.431003625349915,
      "grad_norm": 0.4760420024394989,
      "learning_rate": 5.5280065954803996e-06,
      "loss": 0.3558,
      "step": 31183
    },
    {
      "epoch": 1.4310495158551695,
      "grad_norm": 0.47134125232696533,
      "learning_rate": 5.527762777894841e-06,
      "loss": 0.3684,
      "step": 31184
    },
    {
      "epoch": 1.431095406360424,
      "grad_norm": 0.457643061876297,
      "learning_rate": 5.52751895904018e-06,
      "loss": 0.4,
      "step": 31185
    },
    {
      "epoch": 1.4311412968656785,
      "grad_norm": 0.4472745954990387,
      "learning_rate": 5.527275138917001e-06,
      "loss": 0.349,
      "step": 31186
    },
    {
      "epoch": 1.431187187370933,
      "grad_norm": 0.6127449870109558,
      "learning_rate": 5.527031317525893e-06,
      "loss": 0.3052,
      "step": 31187
    },
    {
      "epoch": 1.4312330778761875,
      "grad_norm": 0.44817066192626953,
      "learning_rate": 5.52678749486744e-06,
      "loss": 0.3489,
      "step": 31188
    },
    {
      "epoch": 1.431278968381442,
      "grad_norm": 0.45312389731407166,
      "learning_rate": 5.526543670942228e-06,
      "loss": 0.3376,
      "step": 31189
    },
    {
      "epoch": 1.4313248588866965,
      "grad_norm": 0.46881818771362305,
      "learning_rate": 5.526299845750847e-06,
      "loss": 0.3293,
      "step": 31190
    },
    {
      "epoch": 1.4313707493919507,
      "grad_norm": 0.4609656035900116,
      "learning_rate": 5.526056019293882e-06,
      "loss": 0.3058,
      "step": 31191
    },
    {
      "epoch": 1.4314166398972052,
      "grad_norm": 0.4606319069862366,
      "learning_rate": 5.525812191571916e-06,
      "loss": 0.363,
      "step": 31192
    },
    {
      "epoch": 1.4314625304024597,
      "grad_norm": 0.4652095139026642,
      "learning_rate": 5.525568362585538e-06,
      "loss": 0.3067,
      "step": 31193
    },
    {
      "epoch": 1.4315084209077142,
      "grad_norm": 0.5363459587097168,
      "learning_rate": 5.5253245323353335e-06,
      "loss": 0.3693,
      "step": 31194
    },
    {
      "epoch": 1.4315543114129687,
      "grad_norm": 0.46269989013671875,
      "learning_rate": 5.525080700821891e-06,
      "loss": 0.3779,
      "step": 31195
    },
    {
      "epoch": 1.431600201918223,
      "grad_norm": 0.49151402711868286,
      "learning_rate": 5.524836868045794e-06,
      "loss": 0.4102,
      "step": 31196
    },
    {
      "epoch": 1.4316460924234775,
      "grad_norm": 0.48053231835365295,
      "learning_rate": 5.5245930340076304e-06,
      "loss": 0.4145,
      "step": 31197
    },
    {
      "epoch": 1.431691982928732,
      "grad_norm": 0.5108349919319153,
      "learning_rate": 5.5243491987079854e-06,
      "loss": 0.4194,
      "step": 31198
    },
    {
      "epoch": 1.4317378734339865,
      "grad_norm": 0.4650034010410309,
      "learning_rate": 5.524105362147446e-06,
      "loss": 0.3513,
      "step": 31199
    },
    {
      "epoch": 1.431783763939241,
      "grad_norm": 0.521092414855957,
      "learning_rate": 5.523861524326599e-06,
      "loss": 0.4219,
      "step": 31200
    },
    {
      "epoch": 1.4318296544444955,
      "grad_norm": 0.4538710117340088,
      "learning_rate": 5.523617685246031e-06,
      "loss": 0.3436,
      "step": 31201
    },
    {
      "epoch": 1.43187554494975,
      "grad_norm": 0.44621777534484863,
      "learning_rate": 5.523373844906328e-06,
      "loss": 0.2826,
      "step": 31202
    },
    {
      "epoch": 1.4319214354550043,
      "grad_norm": 0.4550861418247223,
      "learning_rate": 5.5231300033080746e-06,
      "loss": 0.3303,
      "step": 31203
    },
    {
      "epoch": 1.4319673259602588,
      "grad_norm": 0.4539366066455841,
      "learning_rate": 5.522886160451861e-06,
      "loss": 0.3526,
      "step": 31204
    },
    {
      "epoch": 1.4320132164655133,
      "grad_norm": 0.4691370129585266,
      "learning_rate": 5.522642316338268e-06,
      "loss": 0.3471,
      "step": 31205
    },
    {
      "epoch": 1.4320591069707678,
      "grad_norm": 0.47325608134269714,
      "learning_rate": 5.5223984709678855e-06,
      "loss": 0.3606,
      "step": 31206
    },
    {
      "epoch": 1.4321049974760223,
      "grad_norm": 0.4400242567062378,
      "learning_rate": 5.522154624341303e-06,
      "loss": 0.2855,
      "step": 31207
    },
    {
      "epoch": 1.4321508879812765,
      "grad_norm": 0.4369378089904785,
      "learning_rate": 5.521910776459101e-06,
      "loss": 0.2835,
      "step": 31208
    },
    {
      "epoch": 1.432196778486531,
      "grad_norm": 0.514861524105072,
      "learning_rate": 5.521666927321868e-06,
      "loss": 0.401,
      "step": 31209
    },
    {
      "epoch": 1.4322426689917855,
      "grad_norm": 0.48621705174446106,
      "learning_rate": 5.521423076930192e-06,
      "loss": 0.3744,
      "step": 31210
    },
    {
      "epoch": 1.43228855949704,
      "grad_norm": 0.45788395404815674,
      "learning_rate": 5.5211792252846565e-06,
      "loss": 0.3367,
      "step": 31211
    },
    {
      "epoch": 1.4323344500022945,
      "grad_norm": 0.6031245589256287,
      "learning_rate": 5.52093537238585e-06,
      "loss": 0.3573,
      "step": 31212
    },
    {
      "epoch": 1.432380340507549,
      "grad_norm": 0.4817715287208557,
      "learning_rate": 5.520691518234359e-06,
      "loss": 0.3352,
      "step": 31213
    },
    {
      "epoch": 1.4324262310128035,
      "grad_norm": 0.5083053112030029,
      "learning_rate": 5.520447662830768e-06,
      "loss": 0.4267,
      "step": 31214
    },
    {
      "epoch": 1.432472121518058,
      "grad_norm": 0.4644688367843628,
      "learning_rate": 5.520203806175666e-06,
      "loss": 0.3051,
      "step": 31215
    },
    {
      "epoch": 1.4325180120233123,
      "grad_norm": 0.4444335699081421,
      "learning_rate": 5.519959948269637e-06,
      "loss": 0.3295,
      "step": 31216
    },
    {
      "epoch": 1.4325639025285668,
      "grad_norm": 0.5276976227760315,
      "learning_rate": 5.519716089113269e-06,
      "loss": 0.4076,
      "step": 31217
    },
    {
      "epoch": 1.4326097930338213,
      "grad_norm": 0.4655589461326599,
      "learning_rate": 5.519472228707149e-06,
      "loss": 0.3442,
      "step": 31218
    },
    {
      "epoch": 1.4326556835390758,
      "grad_norm": 0.47360605001449585,
      "learning_rate": 5.51922836705186e-06,
      "loss": 0.3694,
      "step": 31219
    },
    {
      "epoch": 1.4327015740443303,
      "grad_norm": 0.4911370277404785,
      "learning_rate": 5.518984504147991e-06,
      "loss": 0.3873,
      "step": 31220
    },
    {
      "epoch": 1.4327474645495846,
      "grad_norm": 0.481213241815567,
      "learning_rate": 5.518740639996129e-06,
      "loss": 0.3892,
      "step": 31221
    },
    {
      "epoch": 1.432793355054839,
      "grad_norm": 0.4377342462539673,
      "learning_rate": 5.5184967745968585e-06,
      "loss": 0.2878,
      "step": 31222
    },
    {
      "epoch": 1.4328392455600936,
      "grad_norm": 0.48144635558128357,
      "learning_rate": 5.518252907950769e-06,
      "loss": 0.3644,
      "step": 31223
    },
    {
      "epoch": 1.432885136065348,
      "grad_norm": 0.5083449482917786,
      "learning_rate": 5.518009040058443e-06,
      "loss": 0.4303,
      "step": 31224
    },
    {
      "epoch": 1.4329310265706026,
      "grad_norm": 0.48671770095825195,
      "learning_rate": 5.5177651709204694e-06,
      "loss": 0.3371,
      "step": 31225
    },
    {
      "epoch": 1.432976917075857,
      "grad_norm": 0.44579946994781494,
      "learning_rate": 5.517521300537434e-06,
      "loss": 0.3096,
      "step": 31226
    },
    {
      "epoch": 1.4330228075811116,
      "grad_norm": 0.6146912574768066,
      "learning_rate": 5.517277428909922e-06,
      "loss": 0.3723,
      "step": 31227
    },
    {
      "epoch": 1.433068698086366,
      "grad_norm": 0.4635695517063141,
      "learning_rate": 5.517033556038521e-06,
      "loss": 0.3495,
      "step": 31228
    },
    {
      "epoch": 1.4331145885916203,
      "grad_norm": 0.44822603464126587,
      "learning_rate": 5.5167896819238195e-06,
      "loss": 0.3015,
      "step": 31229
    },
    {
      "epoch": 1.4331604790968748,
      "grad_norm": 0.456224262714386,
      "learning_rate": 5.516545806566402e-06,
      "loss": 0.3287,
      "step": 31230
    },
    {
      "epoch": 1.4332063696021293,
      "grad_norm": 0.513614296913147,
      "learning_rate": 5.516301929966852e-06,
      "loss": 0.3896,
      "step": 31231
    },
    {
      "epoch": 1.4332522601073838,
      "grad_norm": 0.5015686750411987,
      "learning_rate": 5.516058052125761e-06,
      "loss": 0.3593,
      "step": 31232
    },
    {
      "epoch": 1.4332981506126383,
      "grad_norm": 0.4730696678161621,
      "learning_rate": 5.515814173043712e-06,
      "loss": 0.3399,
      "step": 31233
    },
    {
      "epoch": 1.4333440411178926,
      "grad_norm": 0.46964579820632935,
      "learning_rate": 5.515570292721294e-06,
      "loss": 0.3347,
      "step": 31234
    },
    {
      "epoch": 1.433389931623147,
      "grad_norm": 0.492269366979599,
      "learning_rate": 5.515326411159092e-06,
      "loss": 0.4147,
      "step": 31235
    },
    {
      "epoch": 1.4334358221284016,
      "grad_norm": 0.44471505284309387,
      "learning_rate": 5.515082528357693e-06,
      "loss": 0.2868,
      "step": 31236
    },
    {
      "epoch": 1.433481712633656,
      "grad_norm": 0.48176226019859314,
      "learning_rate": 5.5148386443176815e-06,
      "loss": 0.3311,
      "step": 31237
    },
    {
      "epoch": 1.4335276031389106,
      "grad_norm": 0.45948708057403564,
      "learning_rate": 5.514594759039646e-06,
      "loss": 0.3677,
      "step": 31238
    },
    {
      "epoch": 1.433573493644165,
      "grad_norm": 0.42645886540412903,
      "learning_rate": 5.5143508725241725e-06,
      "loss": 0.2767,
      "step": 31239
    },
    {
      "epoch": 1.4336193841494196,
      "grad_norm": 0.4710995852947235,
      "learning_rate": 5.51410698477185e-06,
      "loss": 0.373,
      "step": 31240
    },
    {
      "epoch": 1.433665274654674,
      "grad_norm": 0.4659832715988159,
      "learning_rate": 5.513863095783259e-06,
      "loss": 0.3272,
      "step": 31241
    },
    {
      "epoch": 1.4337111651599284,
      "grad_norm": 0.4803202450275421,
      "learning_rate": 5.513619205558991e-06,
      "loss": 0.3602,
      "step": 31242
    },
    {
      "epoch": 1.4337570556651829,
      "grad_norm": 0.4392968714237213,
      "learning_rate": 5.51337531409963e-06,
      "loss": 0.3137,
      "step": 31243
    },
    {
      "epoch": 1.4338029461704374,
      "grad_norm": 0.48227450251579285,
      "learning_rate": 5.513131421405764e-06,
      "loss": 0.3511,
      "step": 31244
    },
    {
      "epoch": 1.4338488366756919,
      "grad_norm": 0.529499351978302,
      "learning_rate": 5.51288752747798e-06,
      "loss": 0.4758,
      "step": 31245
    },
    {
      "epoch": 1.4338947271809461,
      "grad_norm": 0.48742902278900146,
      "learning_rate": 5.5126436323168634e-06,
      "loss": 0.4367,
      "step": 31246
    },
    {
      "epoch": 1.4339406176862006,
      "grad_norm": 0.4881955683231354,
      "learning_rate": 5.512399735922999e-06,
      "loss": 0.4018,
      "step": 31247
    },
    {
      "epoch": 1.4339865081914551,
      "grad_norm": 0.4701627790927887,
      "learning_rate": 5.5121558382969744e-06,
      "loss": 0.4219,
      "step": 31248
    },
    {
      "epoch": 1.4340323986967096,
      "grad_norm": 0.47178494930267334,
      "learning_rate": 5.511911939439378e-06,
      "loss": 0.3488,
      "step": 31249
    },
    {
      "epoch": 1.4340782892019641,
      "grad_norm": 0.4015968143939972,
      "learning_rate": 5.511668039350795e-06,
      "loss": 0.2792,
      "step": 31250
    },
    {
      "epoch": 1.4341241797072186,
      "grad_norm": 0.4519443213939667,
      "learning_rate": 5.5114241380318116e-06,
      "loss": 0.3211,
      "step": 31251
    },
    {
      "epoch": 1.4341700702124731,
      "grad_norm": 0.496487021446228,
      "learning_rate": 5.511180235483015e-06,
      "loss": 0.4518,
      "step": 31252
    },
    {
      "epoch": 1.4342159607177276,
      "grad_norm": 0.4863892197608948,
      "learning_rate": 5.51093633170499e-06,
      "loss": 0.3567,
      "step": 31253
    },
    {
      "epoch": 1.4342618512229819,
      "grad_norm": 0.46944448351860046,
      "learning_rate": 5.510692426698325e-06,
      "loss": 0.3489,
      "step": 31254
    },
    {
      "epoch": 1.4343077417282364,
      "grad_norm": 0.46215325593948364,
      "learning_rate": 5.510448520463607e-06,
      "loss": 0.3363,
      "step": 31255
    },
    {
      "epoch": 1.4343536322334909,
      "grad_norm": 0.4589986503124237,
      "learning_rate": 5.510204613001421e-06,
      "loss": 0.3232,
      "step": 31256
    },
    {
      "epoch": 1.4343995227387454,
      "grad_norm": 0.4844457507133484,
      "learning_rate": 5.509960704312354e-06,
      "loss": 0.4059,
      "step": 31257
    },
    {
      "epoch": 1.4344454132439999,
      "grad_norm": 0.4951876699924469,
      "learning_rate": 5.509716794396991e-06,
      "loss": 0.3688,
      "step": 31258
    },
    {
      "epoch": 1.4344913037492542,
      "grad_norm": 0.4735114276409149,
      "learning_rate": 5.5094728832559195e-06,
      "loss": 0.3922,
      "step": 31259
    },
    {
      "epoch": 1.4345371942545087,
      "grad_norm": 0.480099618434906,
      "learning_rate": 5.50922897088973e-06,
      "loss": 0.4053,
      "step": 31260
    },
    {
      "epoch": 1.4345830847597631,
      "grad_norm": 0.4813135862350464,
      "learning_rate": 5.508985057299002e-06,
      "loss": 0.3508,
      "step": 31261
    },
    {
      "epoch": 1.4346289752650176,
      "grad_norm": 0.44718798995018005,
      "learning_rate": 5.508741142484327e-06,
      "loss": 0.3548,
      "step": 31262
    },
    {
      "epoch": 1.4346748657702721,
      "grad_norm": 0.4617826044559479,
      "learning_rate": 5.508497226446291e-06,
      "loss": 0.38,
      "step": 31263
    },
    {
      "epoch": 1.4347207562755266,
      "grad_norm": 0.46927306056022644,
      "learning_rate": 5.508253309185478e-06,
      "loss": 0.3404,
      "step": 31264
    },
    {
      "epoch": 1.4347666467807811,
      "grad_norm": 0.46597418189048767,
      "learning_rate": 5.508009390702477e-06,
      "loss": 0.3936,
      "step": 31265
    },
    {
      "epoch": 1.4348125372860356,
      "grad_norm": 0.5003175735473633,
      "learning_rate": 5.507765470997875e-06,
      "loss": 0.3911,
      "step": 31266
    },
    {
      "epoch": 1.43485842779129,
      "grad_norm": 0.5048842430114746,
      "learning_rate": 5.507521550072256e-06,
      "loss": 0.4244,
      "step": 31267
    },
    {
      "epoch": 1.4349043182965444,
      "grad_norm": 0.47882556915283203,
      "learning_rate": 5.507277627926206e-06,
      "loss": 0.396,
      "step": 31268
    },
    {
      "epoch": 1.434950208801799,
      "grad_norm": 0.434061735868454,
      "learning_rate": 5.507033704560316e-06,
      "loss": 0.2819,
      "step": 31269
    },
    {
      "epoch": 1.4349960993070534,
      "grad_norm": 0.5054728388786316,
      "learning_rate": 5.506789779975169e-06,
      "loss": 0.4042,
      "step": 31270
    },
    {
      "epoch": 1.435041989812308,
      "grad_norm": 0.4878963530063629,
      "learning_rate": 5.5065458541713525e-06,
      "loss": 0.3871,
      "step": 31271
    },
    {
      "epoch": 1.4350878803175622,
      "grad_norm": 0.4776059091091156,
      "learning_rate": 5.506301927149453e-06,
      "loss": 0.3289,
      "step": 31272
    },
    {
      "epoch": 1.4351337708228167,
      "grad_norm": 0.4663805663585663,
      "learning_rate": 5.506057998910059e-06,
      "loss": 0.3422,
      "step": 31273
    },
    {
      "epoch": 1.4351796613280712,
      "grad_norm": 0.4405117332935333,
      "learning_rate": 5.505814069453755e-06,
      "loss": 0.3162,
      "step": 31274
    },
    {
      "epoch": 1.4352255518333257,
      "grad_norm": 0.5175184607505798,
      "learning_rate": 5.5055701387811255e-06,
      "loss": 0.4795,
      "step": 31275
    },
    {
      "epoch": 1.4352714423385802,
      "grad_norm": 0.5051531195640564,
      "learning_rate": 5.505326206892759e-06,
      "loss": 0.3496,
      "step": 31276
    },
    {
      "epoch": 1.4353173328438347,
      "grad_norm": 0.4543939232826233,
      "learning_rate": 5.505082273789245e-06,
      "loss": 0.3776,
      "step": 31277
    },
    {
      "epoch": 1.4353632233490892,
      "grad_norm": 0.4529600143432617,
      "learning_rate": 5.504838339471167e-06,
      "loss": 0.348,
      "step": 31278
    },
    {
      "epoch": 1.4354091138543437,
      "grad_norm": 0.5172673463821411,
      "learning_rate": 5.50459440393911e-06,
      "loss": 0.3801,
      "step": 31279
    },
    {
      "epoch": 1.435455004359598,
      "grad_norm": 0.46863535046577454,
      "learning_rate": 5.504350467193666e-06,
      "loss": 0.3913,
      "step": 31280
    },
    {
      "epoch": 1.4355008948648524,
      "grad_norm": 0.4577239155769348,
      "learning_rate": 5.504106529235416e-06,
      "loss": 0.3241,
      "step": 31281
    },
    {
      "epoch": 1.435546785370107,
      "grad_norm": 0.44955503940582275,
      "learning_rate": 5.503862590064949e-06,
      "loss": 0.3673,
      "step": 31282
    },
    {
      "epoch": 1.4355926758753614,
      "grad_norm": 0.4839995503425598,
      "learning_rate": 5.503618649682853e-06,
      "loss": 0.3756,
      "step": 31283
    },
    {
      "epoch": 1.435638566380616,
      "grad_norm": 0.44301360845565796,
      "learning_rate": 5.503374708089712e-06,
      "loss": 0.314,
      "step": 31284
    },
    {
      "epoch": 1.4356844568858702,
      "grad_norm": 0.5052185654640198,
      "learning_rate": 5.503130765286113e-06,
      "loss": 0.44,
      "step": 31285
    },
    {
      "epoch": 1.4357303473911247,
      "grad_norm": 0.4602861702442169,
      "learning_rate": 5.502886821272644e-06,
      "loss": 0.3258,
      "step": 31286
    },
    {
      "epoch": 1.4357762378963792,
      "grad_norm": 0.4827623963356018,
      "learning_rate": 5.502642876049891e-06,
      "loss": 0.3573,
      "step": 31287
    },
    {
      "epoch": 1.4358221284016337,
      "grad_norm": 0.47578364610671997,
      "learning_rate": 5.502398929618441e-06,
      "loss": 0.3967,
      "step": 31288
    },
    {
      "epoch": 1.4358680189068882,
      "grad_norm": 0.4548254609107971,
      "learning_rate": 5.50215498197888e-06,
      "loss": 0.3211,
      "step": 31289
    },
    {
      "epoch": 1.4359139094121427,
      "grad_norm": 0.4505936801433563,
      "learning_rate": 5.5019110331317936e-06,
      "loss": 0.3495,
      "step": 31290
    },
    {
      "epoch": 1.4359597999173972,
      "grad_norm": 0.4768708348274231,
      "learning_rate": 5.501667083077771e-06,
      "loss": 0.3276,
      "step": 31291
    },
    {
      "epoch": 1.4360056904226515,
      "grad_norm": 0.44940635561943054,
      "learning_rate": 5.501423131817396e-06,
      "loss": 0.3067,
      "step": 31292
    },
    {
      "epoch": 1.436051580927906,
      "grad_norm": 0.46306875348091125,
      "learning_rate": 5.501179179351257e-06,
      "loss": 0.3469,
      "step": 31293
    },
    {
      "epoch": 1.4360974714331605,
      "grad_norm": 0.4432179629802704,
      "learning_rate": 5.500935225679941e-06,
      "loss": 0.3394,
      "step": 31294
    },
    {
      "epoch": 1.436143361938415,
      "grad_norm": 0.48616695404052734,
      "learning_rate": 5.500691270804034e-06,
      "loss": 0.386,
      "step": 31295
    },
    {
      "epoch": 1.4361892524436695,
      "grad_norm": 0.45936718583106995,
      "learning_rate": 5.500447314724121e-06,
      "loss": 0.3246,
      "step": 31296
    },
    {
      "epoch": 1.4362351429489237,
      "grad_norm": 0.45469799637794495,
      "learning_rate": 5.500203357440792e-06,
      "loss": 0.3324,
      "step": 31297
    },
    {
      "epoch": 1.4362810334541782,
      "grad_norm": 0.4351957440376282,
      "learning_rate": 5.499959398954631e-06,
      "loss": 0.2883,
      "step": 31298
    },
    {
      "epoch": 1.4363269239594327,
      "grad_norm": 0.4529068171977997,
      "learning_rate": 5.499715439266226e-06,
      "loss": 0.3075,
      "step": 31299
    },
    {
      "epoch": 1.4363728144646872,
      "grad_norm": 0.48000335693359375,
      "learning_rate": 5.499471478376163e-06,
      "loss": 0.3785,
      "step": 31300
    },
    {
      "epoch": 1.4364187049699417,
      "grad_norm": 0.4580353796482086,
      "learning_rate": 5.4992275162850275e-06,
      "loss": 0.3354,
      "step": 31301
    },
    {
      "epoch": 1.4364645954751962,
      "grad_norm": 0.4898914694786072,
      "learning_rate": 5.498983552993408e-06,
      "loss": 0.411,
      "step": 31302
    },
    {
      "epoch": 1.4365104859804507,
      "grad_norm": 0.4478312134742737,
      "learning_rate": 5.49873958850189e-06,
      "loss": 0.3099,
      "step": 31303
    },
    {
      "epoch": 1.4365563764857052,
      "grad_norm": 0.46175700426101685,
      "learning_rate": 5.498495622811061e-06,
      "loss": 0.3741,
      "step": 31304
    },
    {
      "epoch": 1.4366022669909595,
      "grad_norm": 0.480903685092926,
      "learning_rate": 5.498251655921508e-06,
      "loss": 0.3552,
      "step": 31305
    },
    {
      "epoch": 1.436648157496214,
      "grad_norm": 0.4578835964202881,
      "learning_rate": 5.498007687833818e-06,
      "loss": 0.3696,
      "step": 31306
    },
    {
      "epoch": 1.4366940480014685,
      "grad_norm": 0.504242479801178,
      "learning_rate": 5.497763718548574e-06,
      "loss": 0.4595,
      "step": 31307
    },
    {
      "epoch": 1.436739938506723,
      "grad_norm": 0.462536096572876,
      "learning_rate": 5.497519748066368e-06,
      "loss": 0.336,
      "step": 31308
    },
    {
      "epoch": 1.4367858290119775,
      "grad_norm": 0.48290562629699707,
      "learning_rate": 5.497275776387782e-06,
      "loss": 0.4001,
      "step": 31309
    },
    {
      "epoch": 1.4368317195172318,
      "grad_norm": 0.44656771421432495,
      "learning_rate": 5.4970318035134065e-06,
      "loss": 0.2911,
      "step": 31310
    },
    {
      "epoch": 1.4368776100224863,
      "grad_norm": 0.46598732471466064,
      "learning_rate": 5.496787829443825e-06,
      "loss": 0.3483,
      "step": 31311
    },
    {
      "epoch": 1.4369235005277408,
      "grad_norm": 0.4733452796936035,
      "learning_rate": 5.496543854179626e-06,
      "loss": 0.3688,
      "step": 31312
    },
    {
      "epoch": 1.4369693910329953,
      "grad_norm": 0.4787349998950958,
      "learning_rate": 5.496299877721395e-06,
      "loss": 0.3962,
      "step": 31313
    },
    {
      "epoch": 1.4370152815382498,
      "grad_norm": 0.46761354804039,
      "learning_rate": 5.49605590006972e-06,
      "loss": 0.3196,
      "step": 31314
    },
    {
      "epoch": 1.4370611720435043,
      "grad_norm": 0.447812020778656,
      "learning_rate": 5.4958119212251874e-06,
      "loss": 0.3332,
      "step": 31315
    },
    {
      "epoch": 1.4371070625487588,
      "grad_norm": 0.4714793264865875,
      "learning_rate": 5.4955679411883835e-06,
      "loss": 0.3638,
      "step": 31316
    },
    {
      "epoch": 1.4371529530540132,
      "grad_norm": 0.5205245614051819,
      "learning_rate": 5.495323959959895e-06,
      "loss": 0.4365,
      "step": 31317
    },
    {
      "epoch": 1.4371988435592675,
      "grad_norm": 0.4729923903942108,
      "learning_rate": 5.4950799775403086e-06,
      "loss": 0.3774,
      "step": 31318
    },
    {
      "epoch": 1.437244734064522,
      "grad_norm": 0.519253671169281,
      "learning_rate": 5.494835993930211e-06,
      "loss": 0.4506,
      "step": 31319
    },
    {
      "epoch": 1.4372906245697765,
      "grad_norm": 0.4829270839691162,
      "learning_rate": 5.49459200913019e-06,
      "loss": 0.3955,
      "step": 31320
    },
    {
      "epoch": 1.437336515075031,
      "grad_norm": 0.44445738196372986,
      "learning_rate": 5.4943480231408305e-06,
      "loss": 0.3221,
      "step": 31321
    },
    {
      "epoch": 1.4373824055802855,
      "grad_norm": 0.4662395417690277,
      "learning_rate": 5.494104035962721e-06,
      "loss": 0.3613,
      "step": 31322
    },
    {
      "epoch": 1.4374282960855398,
      "grad_norm": 0.46019259095191956,
      "learning_rate": 5.4938600475964455e-06,
      "loss": 0.3238,
      "step": 31323
    },
    {
      "epoch": 1.4374741865907943,
      "grad_norm": 0.4775780737400055,
      "learning_rate": 5.493616058042592e-06,
      "loss": 0.3515,
      "step": 31324
    },
    {
      "epoch": 1.4375200770960488,
      "grad_norm": 0.49702268838882446,
      "learning_rate": 5.49337206730175e-06,
      "loss": 0.4333,
      "step": 31325
    },
    {
      "epoch": 1.4375659676013033,
      "grad_norm": 0.4497983455657959,
      "learning_rate": 5.493128075374503e-06,
      "loss": 0.3013,
      "step": 31326
    },
    {
      "epoch": 1.4376118581065578,
      "grad_norm": 0.4566693902015686,
      "learning_rate": 5.4928840822614384e-06,
      "loss": 0.3357,
      "step": 31327
    },
    {
      "epoch": 1.4376577486118123,
      "grad_norm": 0.449724942445755,
      "learning_rate": 5.492640087963143e-06,
      "loss": 0.3233,
      "step": 31328
    },
    {
      "epoch": 1.4377036391170668,
      "grad_norm": 0.5228531360626221,
      "learning_rate": 5.492396092480204e-06,
      "loss": 0.4378,
      "step": 31329
    },
    {
      "epoch": 1.4377495296223213,
      "grad_norm": 0.45829933881759644,
      "learning_rate": 5.492152095813208e-06,
      "loss": 0.3558,
      "step": 31330
    },
    {
      "epoch": 1.4377954201275756,
      "grad_norm": 0.44357579946517944,
      "learning_rate": 5.491908097962742e-06,
      "loss": 0.3203,
      "step": 31331
    },
    {
      "epoch": 1.43784131063283,
      "grad_norm": 0.444955438375473,
      "learning_rate": 5.491664098929391e-06,
      "loss": 0.2875,
      "step": 31332
    },
    {
      "epoch": 1.4378872011380845,
      "grad_norm": 0.5146957635879517,
      "learning_rate": 5.491420098713743e-06,
      "loss": 0.3701,
      "step": 31333
    },
    {
      "epoch": 1.437933091643339,
      "grad_norm": 0.46581801772117615,
      "learning_rate": 5.491176097316386e-06,
      "loss": 0.3622,
      "step": 31334
    },
    {
      "epoch": 1.4379789821485933,
      "grad_norm": 0.47365114092826843,
      "learning_rate": 5.490932094737904e-06,
      "loss": 0.3511,
      "step": 31335
    },
    {
      "epoch": 1.4380248726538478,
      "grad_norm": 0.4519864022731781,
      "learning_rate": 5.490688090978885e-06,
      "loss": 0.3157,
      "step": 31336
    },
    {
      "epoch": 1.4380707631591023,
      "grad_norm": 0.4523589611053467,
      "learning_rate": 5.490444086039918e-06,
      "loss": 0.3483,
      "step": 31337
    },
    {
      "epoch": 1.4381166536643568,
      "grad_norm": 0.47337013483047485,
      "learning_rate": 5.4902000799215875e-06,
      "loss": 0.3948,
      "step": 31338
    },
    {
      "epoch": 1.4381625441696113,
      "grad_norm": 0.47752612829208374,
      "learning_rate": 5.489956072624481e-06,
      "loss": 0.3795,
      "step": 31339
    },
    {
      "epoch": 1.4382084346748658,
      "grad_norm": 0.4637202024459839,
      "learning_rate": 5.489712064149182e-06,
      "loss": 0.35,
      "step": 31340
    },
    {
      "epoch": 1.4382543251801203,
      "grad_norm": 0.4736252427101135,
      "learning_rate": 5.489468054496281e-06,
      "loss": 0.334,
      "step": 31341
    },
    {
      "epoch": 1.4383002156853748,
      "grad_norm": 0.4821629822254181,
      "learning_rate": 5.489224043666366e-06,
      "loss": 0.3967,
      "step": 31342
    },
    {
      "epoch": 1.438346106190629,
      "grad_norm": 0.4589228928089142,
      "learning_rate": 5.4889800316600195e-06,
      "loss": 0.3323,
      "step": 31343
    },
    {
      "epoch": 1.4383919966958836,
      "grad_norm": 0.5332475900650024,
      "learning_rate": 5.48873601847783e-06,
      "loss": 0.5006,
      "step": 31344
    },
    {
      "epoch": 1.438437887201138,
      "grad_norm": 0.5547856092453003,
      "learning_rate": 5.488492004120387e-06,
      "loss": 0.4859,
      "step": 31345
    },
    {
      "epoch": 1.4384837777063926,
      "grad_norm": 0.44414642453193665,
      "learning_rate": 5.488247988588272e-06,
      "loss": 0.3392,
      "step": 31346
    },
    {
      "epoch": 1.438529668211647,
      "grad_norm": 0.4363952577114105,
      "learning_rate": 5.488003971882076e-06,
      "loss": 0.3314,
      "step": 31347
    },
    {
      "epoch": 1.4385755587169013,
      "grad_norm": 0.43330708146095276,
      "learning_rate": 5.487759954002385e-06,
      "loss": 0.2811,
      "step": 31348
    },
    {
      "epoch": 1.4386214492221558,
      "grad_norm": 0.4809993803501129,
      "learning_rate": 5.487515934949786e-06,
      "loss": 0.3886,
      "step": 31349
    },
    {
      "epoch": 1.4386673397274103,
      "grad_norm": 0.495016872882843,
      "learning_rate": 5.487271914724863e-06,
      "loss": 0.4226,
      "step": 31350
    },
    {
      "epoch": 1.4387132302326648,
      "grad_norm": 0.4973916709423065,
      "learning_rate": 5.487027893328206e-06,
      "loss": 0.3914,
      "step": 31351
    },
    {
      "epoch": 1.4387591207379193,
      "grad_norm": 0.4934927821159363,
      "learning_rate": 5.486783870760399e-06,
      "loss": 0.4294,
      "step": 31352
    },
    {
      "epoch": 1.4388050112431738,
      "grad_norm": 0.4328587055206299,
      "learning_rate": 5.486539847022033e-06,
      "loss": 0.3206,
      "step": 31353
    },
    {
      "epoch": 1.4388509017484283,
      "grad_norm": 0.45245832204818726,
      "learning_rate": 5.486295822113692e-06,
      "loss": 0.351,
      "step": 31354
    },
    {
      "epoch": 1.4388967922536828,
      "grad_norm": 0.5030629634857178,
      "learning_rate": 5.486051796035961e-06,
      "loss": 0.4261,
      "step": 31355
    },
    {
      "epoch": 1.438942682758937,
      "grad_norm": 0.4464128315448761,
      "learning_rate": 5.485807768789431e-06,
      "loss": 0.3272,
      "step": 31356
    },
    {
      "epoch": 1.4389885732641916,
      "grad_norm": 0.41742241382598877,
      "learning_rate": 5.485563740374684e-06,
      "loss": 0.2826,
      "step": 31357
    },
    {
      "epoch": 1.439034463769446,
      "grad_norm": 0.4848165214061737,
      "learning_rate": 5.485319710792311e-06,
      "loss": 0.3989,
      "step": 31358
    },
    {
      "epoch": 1.4390803542747006,
      "grad_norm": 0.43734264373779297,
      "learning_rate": 5.485075680042897e-06,
      "loss": 0.286,
      "step": 31359
    },
    {
      "epoch": 1.439126244779955,
      "grad_norm": 0.497589111328125,
      "learning_rate": 5.48483164812703e-06,
      "loss": 0.4194,
      "step": 31360
    },
    {
      "epoch": 1.4391721352852094,
      "grad_norm": 0.4430268406867981,
      "learning_rate": 5.484587615045295e-06,
      "loss": 0.3282,
      "step": 31361
    },
    {
      "epoch": 1.4392180257904639,
      "grad_norm": 0.5103000402450562,
      "learning_rate": 5.4843435807982805e-06,
      "loss": 0.4599,
      "step": 31362
    },
    {
      "epoch": 1.4392639162957184,
      "grad_norm": 0.4488638639450073,
      "learning_rate": 5.484099545386571e-06,
      "loss": 0.3391,
      "step": 31363
    },
    {
      "epoch": 1.4393098068009729,
      "grad_norm": 0.4482279121875763,
      "learning_rate": 5.483855508810756e-06,
      "loss": 0.339,
      "step": 31364
    },
    {
      "epoch": 1.4393556973062274,
      "grad_norm": 0.46682676672935486,
      "learning_rate": 5.483611471071422e-06,
      "loss": 0.3513,
      "step": 31365
    },
    {
      "epoch": 1.4394015878114819,
      "grad_norm": 0.4459417462348938,
      "learning_rate": 5.483367432169154e-06,
      "loss": 0.3308,
      "step": 31366
    },
    {
      "epoch": 1.4394474783167364,
      "grad_norm": 0.49146759510040283,
      "learning_rate": 5.483123392104539e-06,
      "loss": 0.3956,
      "step": 31367
    },
    {
      "epoch": 1.4394933688219909,
      "grad_norm": 0.48244786262512207,
      "learning_rate": 5.482879350878165e-06,
      "loss": 0.3767,
      "step": 31368
    },
    {
      "epoch": 1.4395392593272451,
      "grad_norm": 0.5042799115180969,
      "learning_rate": 5.482635308490619e-06,
      "loss": 0.4157,
      "step": 31369
    },
    {
      "epoch": 1.4395851498324996,
      "grad_norm": 0.4869748055934906,
      "learning_rate": 5.482391264942488e-06,
      "loss": 0.3814,
      "step": 31370
    },
    {
      "epoch": 1.4396310403377541,
      "grad_norm": 0.46865376830101013,
      "learning_rate": 5.48214722023436e-06,
      "loss": 0.3993,
      "step": 31371
    },
    {
      "epoch": 1.4396769308430086,
      "grad_norm": 0.4360770583152771,
      "learning_rate": 5.481903174366816e-06,
      "loss": 0.3036,
      "step": 31372
    },
    {
      "epoch": 1.4397228213482631,
      "grad_norm": 0.46867313981056213,
      "learning_rate": 5.48165912734045e-06,
      "loss": 0.3226,
      "step": 31373
    },
    {
      "epoch": 1.4397687118535174,
      "grad_norm": 0.5339978337287903,
      "learning_rate": 5.481415079155844e-06,
      "loss": 0.4624,
      "step": 31374
    },
    {
      "epoch": 1.439814602358772,
      "grad_norm": 0.5256969332695007,
      "learning_rate": 5.481171029813588e-06,
      "loss": 0.3014,
      "step": 31375
    },
    {
      "epoch": 1.4398604928640264,
      "grad_norm": 0.49928709864616394,
      "learning_rate": 5.480926979314267e-06,
      "loss": 0.401,
      "step": 31376
    },
    {
      "epoch": 1.439906383369281,
      "grad_norm": 0.4604565501213074,
      "learning_rate": 5.4806829276584685e-06,
      "loss": 0.3206,
      "step": 31377
    },
    {
      "epoch": 1.4399522738745354,
      "grad_norm": 0.46290624141693115,
      "learning_rate": 5.480438874846778e-06,
      "loss": 0.343,
      "step": 31378
    },
    {
      "epoch": 1.43999816437979,
      "grad_norm": 0.5096548795700073,
      "learning_rate": 5.480194820879786e-06,
      "loss": 0.3708,
      "step": 31379
    },
    {
      "epoch": 1.4400440548850444,
      "grad_norm": 0.4929756820201874,
      "learning_rate": 5.479950765758075e-06,
      "loss": 0.3529,
      "step": 31380
    },
    {
      "epoch": 1.4400899453902987,
      "grad_norm": 0.48528236150741577,
      "learning_rate": 5.479706709482235e-06,
      "loss": 0.353,
      "step": 31381
    },
    {
      "epoch": 1.4401358358955532,
      "grad_norm": 0.4715818166732788,
      "learning_rate": 5.479462652052853e-06,
      "loss": 0.3413,
      "step": 31382
    },
    {
      "epoch": 1.4401817264008077,
      "grad_norm": 0.45701804757118225,
      "learning_rate": 5.479218593470512e-06,
      "loss": 0.3133,
      "step": 31383
    },
    {
      "epoch": 1.4402276169060622,
      "grad_norm": 0.4825209975242615,
      "learning_rate": 5.478974533735802e-06,
      "loss": 0.3372,
      "step": 31384
    },
    {
      "epoch": 1.4402735074113167,
      "grad_norm": 0.5008915066719055,
      "learning_rate": 5.478730472849311e-06,
      "loss": 0.4343,
      "step": 31385
    },
    {
      "epoch": 1.440319397916571,
      "grad_norm": 0.4563448131084442,
      "learning_rate": 5.478486410811623e-06,
      "loss": 0.3627,
      "step": 31386
    },
    {
      "epoch": 1.4403652884218254,
      "grad_norm": 0.4518575966358185,
      "learning_rate": 5.478242347623327e-06,
      "loss": 0.3132,
      "step": 31387
    },
    {
      "epoch": 1.44041117892708,
      "grad_norm": 0.49911078810691833,
      "learning_rate": 5.4779982832850085e-06,
      "loss": 0.4573,
      "step": 31388
    },
    {
      "epoch": 1.4404570694323344,
      "grad_norm": 0.4592586159706116,
      "learning_rate": 5.477754217797254e-06,
      "loss": 0.3367,
      "step": 31389
    },
    {
      "epoch": 1.440502959937589,
      "grad_norm": 0.5126767754554749,
      "learning_rate": 5.477510151160653e-06,
      "loss": 0.455,
      "step": 31390
    },
    {
      "epoch": 1.4405488504428434,
      "grad_norm": 0.4522346556186676,
      "learning_rate": 5.4772660833757895e-06,
      "loss": 0.3122,
      "step": 31391
    },
    {
      "epoch": 1.440594740948098,
      "grad_norm": 0.45370566844940186,
      "learning_rate": 5.477022014443254e-06,
      "loss": 0.3567,
      "step": 31392
    },
    {
      "epoch": 1.4406406314533524,
      "grad_norm": 0.4503306746482849,
      "learning_rate": 5.476777944363629e-06,
      "loss": 0.3383,
      "step": 31393
    },
    {
      "epoch": 1.4406865219586067,
      "grad_norm": 0.46526479721069336,
      "learning_rate": 5.476533873137504e-06,
      "loss": 0.35,
      "step": 31394
    },
    {
      "epoch": 1.4407324124638612,
      "grad_norm": 0.4870149493217468,
      "learning_rate": 5.476289800765465e-06,
      "loss": 0.3774,
      "step": 31395
    },
    {
      "epoch": 1.4407783029691157,
      "grad_norm": 0.5725382566452026,
      "learning_rate": 5.4760457272481e-06,
      "loss": 0.454,
      "step": 31396
    },
    {
      "epoch": 1.4408241934743702,
      "grad_norm": 0.4231787323951721,
      "learning_rate": 5.475801652585997e-06,
      "loss": 0.3002,
      "step": 31397
    },
    {
      "epoch": 1.4408700839796247,
      "grad_norm": 0.45858556032180786,
      "learning_rate": 5.475557576779739e-06,
      "loss": 0.3516,
      "step": 31398
    },
    {
      "epoch": 1.440915974484879,
      "grad_norm": 0.478611558675766,
      "learning_rate": 5.475313499829915e-06,
      "loss": 0.3871,
      "step": 31399
    },
    {
      "epoch": 1.4409618649901335,
      "grad_norm": 0.4636435806751251,
      "learning_rate": 5.475069421737113e-06,
      "loss": 0.3555,
      "step": 31400
    },
    {
      "epoch": 1.441007755495388,
      "grad_norm": 0.484809547662735,
      "learning_rate": 5.474825342501917e-06,
      "loss": 0.3671,
      "step": 31401
    },
    {
      "epoch": 1.4410536460006425,
      "grad_norm": 0.457040935754776,
      "learning_rate": 5.474581262124918e-06,
      "loss": 0.3306,
      "step": 31402
    },
    {
      "epoch": 1.441099536505897,
      "grad_norm": 0.46439191699028015,
      "learning_rate": 5.474337180606702e-06,
      "loss": 0.3337,
      "step": 31403
    },
    {
      "epoch": 1.4411454270111514,
      "grad_norm": 0.46279504895210266,
      "learning_rate": 5.474093097947852e-06,
      "loss": 0.3326,
      "step": 31404
    },
    {
      "epoch": 1.441191317516406,
      "grad_norm": 0.4923481345176697,
      "learning_rate": 5.4738490141489585e-06,
      "loss": 0.3981,
      "step": 31405
    },
    {
      "epoch": 1.4412372080216604,
      "grad_norm": 0.4916951656341553,
      "learning_rate": 5.473604929210607e-06,
      "loss": 0.3998,
      "step": 31406
    },
    {
      "epoch": 1.4412830985269147,
      "grad_norm": 0.44359204173088074,
      "learning_rate": 5.473360843133388e-06,
      "loss": 0.3152,
      "step": 31407
    },
    {
      "epoch": 1.4413289890321692,
      "grad_norm": 0.458746999502182,
      "learning_rate": 5.473116755917883e-06,
      "loss": 0.3676,
      "step": 31408
    },
    {
      "epoch": 1.4413748795374237,
      "grad_norm": 0.5041618943214417,
      "learning_rate": 5.472872667564681e-06,
      "loss": 0.3937,
      "step": 31409
    },
    {
      "epoch": 1.4414207700426782,
      "grad_norm": 0.4622243046760559,
      "learning_rate": 5.472628578074373e-06,
      "loss": 0.3614,
      "step": 31410
    },
    {
      "epoch": 1.4414666605479327,
      "grad_norm": 0.4347517788410187,
      "learning_rate": 5.472384487447539e-06,
      "loss": 0.3264,
      "step": 31411
    },
    {
      "epoch": 1.441512551053187,
      "grad_norm": 0.4699138402938843,
      "learning_rate": 5.47214039568477e-06,
      "loss": 0.3739,
      "step": 31412
    },
    {
      "epoch": 1.4415584415584415,
      "grad_norm": 0.4471346139907837,
      "learning_rate": 5.4718963027866535e-06,
      "loss": 0.3192,
      "step": 31413
    },
    {
      "epoch": 1.441604332063696,
      "grad_norm": 0.4525623321533203,
      "learning_rate": 5.4716522087537745e-06,
      "loss": 0.3429,
      "step": 31414
    },
    {
      "epoch": 1.4416502225689505,
      "grad_norm": 0.4725325107574463,
      "learning_rate": 5.47140811358672e-06,
      "loss": 0.3374,
      "step": 31415
    },
    {
      "epoch": 1.441696113074205,
      "grad_norm": 0.4657646417617798,
      "learning_rate": 5.47116401728608e-06,
      "loss": 0.3482,
      "step": 31416
    },
    {
      "epoch": 1.4417420035794595,
      "grad_norm": 0.48165014386177063,
      "learning_rate": 5.470919919852437e-06,
      "loss": 0.382,
      "step": 31417
    },
    {
      "epoch": 1.441787894084714,
      "grad_norm": 0.4800724387168884,
      "learning_rate": 5.470675821286381e-06,
      "loss": 0.3851,
      "step": 31418
    },
    {
      "epoch": 1.4418337845899685,
      "grad_norm": 0.452603816986084,
      "learning_rate": 5.470431721588499e-06,
      "loss": 0.3103,
      "step": 31419
    },
    {
      "epoch": 1.4418796750952227,
      "grad_norm": 0.46072864532470703,
      "learning_rate": 5.470187620759377e-06,
      "loss": 0.3573,
      "step": 31420
    },
    {
      "epoch": 1.4419255656004772,
      "grad_norm": 0.4626932144165039,
      "learning_rate": 5.469943518799602e-06,
      "loss": 0.2985,
      "step": 31421
    },
    {
      "epoch": 1.4419714561057317,
      "grad_norm": 0.46423834562301636,
      "learning_rate": 5.469699415709761e-06,
      "loss": 0.3322,
      "step": 31422
    },
    {
      "epoch": 1.4420173466109862,
      "grad_norm": 0.49719539284706116,
      "learning_rate": 5.469455311490441e-06,
      "loss": 0.3844,
      "step": 31423
    },
    {
      "epoch": 1.4420632371162405,
      "grad_norm": 0.4634324014186859,
      "learning_rate": 5.469211206142229e-06,
      "loss": 0.3389,
      "step": 31424
    },
    {
      "epoch": 1.442109127621495,
      "grad_norm": 0.44302046298980713,
      "learning_rate": 5.468967099665712e-06,
      "loss": 0.3508,
      "step": 31425
    },
    {
      "epoch": 1.4421550181267495,
      "grad_norm": 0.4605623483657837,
      "learning_rate": 5.468722992061477e-06,
      "loss": 0.3658,
      "step": 31426
    },
    {
      "epoch": 1.442200908632004,
      "grad_norm": 0.47131460905075073,
      "learning_rate": 5.4684788833301124e-06,
      "loss": 0.3663,
      "step": 31427
    },
    {
      "epoch": 1.4422467991372585,
      "grad_norm": 0.4544435739517212,
      "learning_rate": 5.468234773472203e-06,
      "loss": 0.299,
      "step": 31428
    },
    {
      "epoch": 1.442292689642513,
      "grad_norm": 0.44661569595336914,
      "learning_rate": 5.467990662488336e-06,
      "loss": 0.2662,
      "step": 31429
    },
    {
      "epoch": 1.4423385801477675,
      "grad_norm": 0.44164687395095825,
      "learning_rate": 5.467746550379101e-06,
      "loss": 0.319,
      "step": 31430
    },
    {
      "epoch": 1.442384470653022,
      "grad_norm": 0.48393216729164124,
      "learning_rate": 5.467502437145082e-06,
      "loss": 0.3513,
      "step": 31431
    },
    {
      "epoch": 1.4424303611582763,
      "grad_norm": 0.5150102972984314,
      "learning_rate": 5.467258322786867e-06,
      "loss": 0.451,
      "step": 31432
    },
    {
      "epoch": 1.4424762516635308,
      "grad_norm": 0.496931791305542,
      "learning_rate": 5.4670142073050436e-06,
      "loss": 0.4072,
      "step": 31433
    },
    {
      "epoch": 1.4425221421687853,
      "grad_norm": 0.4517005383968353,
      "learning_rate": 5.4667700907001986e-06,
      "loss": 0.3274,
      "step": 31434
    },
    {
      "epoch": 1.4425680326740398,
      "grad_norm": 0.4690845310688019,
      "learning_rate": 5.466525972972919e-06,
      "loss": 0.372,
      "step": 31435
    },
    {
      "epoch": 1.4426139231792943,
      "grad_norm": 0.4751775562763214,
      "learning_rate": 5.4662818541237914e-06,
      "loss": 0.35,
      "step": 31436
    },
    {
      "epoch": 1.4426598136845485,
      "grad_norm": 0.4207060933113098,
      "learning_rate": 5.466037734153403e-06,
      "loss": 0.309,
      "step": 31437
    },
    {
      "epoch": 1.442705704189803,
      "grad_norm": 0.5052752494812012,
      "learning_rate": 5.4657936130623415e-06,
      "loss": 0.4932,
      "step": 31438
    },
    {
      "epoch": 1.4427515946950575,
      "grad_norm": 0.42586788535118103,
      "learning_rate": 5.465549490851192e-06,
      "loss": 0.2952,
      "step": 31439
    },
    {
      "epoch": 1.442797485200312,
      "grad_norm": 0.4814181923866272,
      "learning_rate": 5.465305367520544e-06,
      "loss": 0.3829,
      "step": 31440
    },
    {
      "epoch": 1.4428433757055665,
      "grad_norm": 0.4755573868751526,
      "learning_rate": 5.465061243070984e-06,
      "loss": 0.34,
      "step": 31441
    },
    {
      "epoch": 1.442889266210821,
      "grad_norm": 0.41971951723098755,
      "learning_rate": 5.464817117503098e-06,
      "loss": 0.3047,
      "step": 31442
    },
    {
      "epoch": 1.4429351567160755,
      "grad_norm": 0.48504501581192017,
      "learning_rate": 5.464572990817471e-06,
      "loss": 0.3744,
      "step": 31443
    },
    {
      "epoch": 1.44298104722133,
      "grad_norm": 0.4887520372867584,
      "learning_rate": 5.4643288630146966e-06,
      "loss": 0.3717,
      "step": 31444
    },
    {
      "epoch": 1.4430269377265843,
      "grad_norm": 0.48134979605674744,
      "learning_rate": 5.464084734095355e-06,
      "loss": 0.3691,
      "step": 31445
    },
    {
      "epoch": 1.4430728282318388,
      "grad_norm": 0.46565812826156616,
      "learning_rate": 5.463840604060038e-06,
      "loss": 0.3518,
      "step": 31446
    },
    {
      "epoch": 1.4431187187370933,
      "grad_norm": 0.41107267141342163,
      "learning_rate": 5.46359647290933e-06,
      "loss": 0.2712,
      "step": 31447
    },
    {
      "epoch": 1.4431646092423478,
      "grad_norm": 0.4478841722011566,
      "learning_rate": 5.463352340643818e-06,
      "loss": 0.318,
      "step": 31448
    },
    {
      "epoch": 1.4432104997476023,
      "grad_norm": 0.47210371494293213,
      "learning_rate": 5.4631082072640905e-06,
      "loss": 0.3609,
      "step": 31449
    },
    {
      "epoch": 1.4432563902528566,
      "grad_norm": 0.43305447697639465,
      "learning_rate": 5.462864072770734e-06,
      "loss": 0.2593,
      "step": 31450
    },
    {
      "epoch": 1.443302280758111,
      "grad_norm": 0.5071636438369751,
      "learning_rate": 5.462619937164335e-06,
      "loss": 0.4478,
      "step": 31451
    },
    {
      "epoch": 1.4433481712633656,
      "grad_norm": 0.4485457241535187,
      "learning_rate": 5.462375800445483e-06,
      "loss": 0.3333,
      "step": 31452
    },
    {
      "epoch": 1.44339406176862,
      "grad_norm": 0.4690832495689392,
      "learning_rate": 5.462131662614762e-06,
      "loss": 0.3258,
      "step": 31453
    },
    {
      "epoch": 1.4434399522738746,
      "grad_norm": 0.4451025128364563,
      "learning_rate": 5.461887523672758e-06,
      "loss": 0.3338,
      "step": 31454
    },
    {
      "epoch": 1.443485842779129,
      "grad_norm": 0.4559609889984131,
      "learning_rate": 5.4616433836200636e-06,
      "loss": 0.3326,
      "step": 31455
    },
    {
      "epoch": 1.4435317332843836,
      "grad_norm": 0.4676848351955414,
      "learning_rate": 5.46139924245726e-06,
      "loss": 0.3385,
      "step": 31456
    },
    {
      "epoch": 1.443577623789638,
      "grad_norm": 0.5128927230834961,
      "learning_rate": 5.461155100184938e-06,
      "loss": 0.3918,
      "step": 31457
    },
    {
      "epoch": 1.4436235142948923,
      "grad_norm": 0.47096022963523865,
      "learning_rate": 5.460910956803684e-06,
      "loss": 0.3908,
      "step": 31458
    },
    {
      "epoch": 1.4436694048001468,
      "grad_norm": 0.42631372809410095,
      "learning_rate": 5.460666812314084e-06,
      "loss": 0.2618,
      "step": 31459
    },
    {
      "epoch": 1.4437152953054013,
      "grad_norm": 0.4620649814605713,
      "learning_rate": 5.460422666716725e-06,
      "loss": 0.3385,
      "step": 31460
    },
    {
      "epoch": 1.4437611858106558,
      "grad_norm": 0.4816322326660156,
      "learning_rate": 5.460178520012196e-06,
      "loss": 0.3824,
      "step": 31461
    },
    {
      "epoch": 1.4438070763159103,
      "grad_norm": 0.47888118028640747,
      "learning_rate": 5.459934372201082e-06,
      "loss": 0.3892,
      "step": 31462
    },
    {
      "epoch": 1.4438529668211646,
      "grad_norm": 0.4660278260707855,
      "learning_rate": 5.459690223283972e-06,
      "loss": 0.359,
      "step": 31463
    },
    {
      "epoch": 1.443898857326419,
      "grad_norm": 0.45937272906303406,
      "learning_rate": 5.4594460732614515e-06,
      "loss": 0.3214,
      "step": 31464
    },
    {
      "epoch": 1.4439447478316736,
      "grad_norm": 0.46098822355270386,
      "learning_rate": 5.459201922134108e-06,
      "loss": 0.3603,
      "step": 31465
    },
    {
      "epoch": 1.443990638336928,
      "grad_norm": 0.4859407842159271,
      "learning_rate": 5.458957769902529e-06,
      "loss": 0.3799,
      "step": 31466
    },
    {
      "epoch": 1.4440365288421826,
      "grad_norm": 0.4993225038051605,
      "learning_rate": 5.458713616567301e-06,
      "loss": 0.3992,
      "step": 31467
    },
    {
      "epoch": 1.444082419347437,
      "grad_norm": 0.47318485379219055,
      "learning_rate": 5.458469462129013e-06,
      "loss": 0.3283,
      "step": 31468
    },
    {
      "epoch": 1.4441283098526916,
      "grad_norm": 0.44621405005455017,
      "learning_rate": 5.458225306588251e-06,
      "loss": 0.3268,
      "step": 31469
    },
    {
      "epoch": 1.4441742003579459,
      "grad_norm": 0.43175822496414185,
      "learning_rate": 5.457981149945599e-06,
      "loss": 0.3309,
      "step": 31470
    },
    {
      "epoch": 1.4442200908632004,
      "grad_norm": 0.46265751123428345,
      "learning_rate": 5.457736992201648e-06,
      "loss": 0.3633,
      "step": 31471
    },
    {
      "epoch": 1.4442659813684549,
      "grad_norm": 0.4149712920188904,
      "learning_rate": 5.457492833356986e-06,
      "loss": 0.2723,
      "step": 31472
    },
    {
      "epoch": 1.4443118718737094,
      "grad_norm": 0.4646666646003723,
      "learning_rate": 5.457248673412194e-06,
      "loss": 0.3783,
      "step": 31473
    },
    {
      "epoch": 1.4443577623789638,
      "grad_norm": 0.4596672058105469,
      "learning_rate": 5.457004512367867e-06,
      "loss": 0.3479,
      "step": 31474
    },
    {
      "epoch": 1.4444036528842181,
      "grad_norm": 0.4868861138820648,
      "learning_rate": 5.456760350224588e-06,
      "loss": 0.3715,
      "step": 31475
    },
    {
      "epoch": 1.4444495433894726,
      "grad_norm": 0.4642239511013031,
      "learning_rate": 5.456516186982943e-06,
      "loss": 0.3445,
      "step": 31476
    },
    {
      "epoch": 1.4444954338947271,
      "grad_norm": 0.4699695110321045,
      "learning_rate": 5.456272022643521e-06,
      "loss": 0.3801,
      "step": 31477
    },
    {
      "epoch": 1.4445413243999816,
      "grad_norm": 0.4731456935405731,
      "learning_rate": 5.4560278572069095e-06,
      "loss": 0.3329,
      "step": 31478
    },
    {
      "epoch": 1.4445872149052361,
      "grad_norm": 0.4280276596546173,
      "learning_rate": 5.455783690673694e-06,
      "loss": 0.3311,
      "step": 31479
    },
    {
      "epoch": 1.4446331054104906,
      "grad_norm": 0.4406444728374481,
      "learning_rate": 5.455539523044463e-06,
      "loss": 0.3103,
      "step": 31480
    },
    {
      "epoch": 1.4446789959157451,
      "grad_norm": 0.4373532235622406,
      "learning_rate": 5.455295354319803e-06,
      "loss": 0.3202,
      "step": 31481
    },
    {
      "epoch": 1.4447248864209996,
      "grad_norm": 0.4357433617115021,
      "learning_rate": 5.455051184500302e-06,
      "loss": 0.2938,
      "step": 31482
    },
    {
      "epoch": 1.4447707769262539,
      "grad_norm": 0.45730534195899963,
      "learning_rate": 5.454807013586545e-06,
      "loss": 0.3648,
      "step": 31483
    },
    {
      "epoch": 1.4448166674315084,
      "grad_norm": 0.46353477239608765,
      "learning_rate": 5.454562841579122e-06,
      "loss": 0.32,
      "step": 31484
    },
    {
      "epoch": 1.4448625579367629,
      "grad_norm": 0.4984106719493866,
      "learning_rate": 5.45431866847862e-06,
      "loss": 0.4263,
      "step": 31485
    },
    {
      "epoch": 1.4449084484420174,
      "grad_norm": 0.4432547986507416,
      "learning_rate": 5.454074494285624e-06,
      "loss": 0.2921,
      "step": 31486
    },
    {
      "epoch": 1.4449543389472719,
      "grad_norm": 0.4568859040737152,
      "learning_rate": 5.4538303190007206e-06,
      "loss": 0.3262,
      "step": 31487
    },
    {
      "epoch": 1.4450002294525262,
      "grad_norm": 0.4592471420764923,
      "learning_rate": 5.4535861426244984e-06,
      "loss": 0.3128,
      "step": 31488
    },
    {
      "epoch": 1.4450461199577807,
      "grad_norm": 0.4604605734348297,
      "learning_rate": 5.453341965157547e-06,
      "loss": 0.3492,
      "step": 31489
    },
    {
      "epoch": 1.4450920104630351,
      "grad_norm": 0.43592938780784607,
      "learning_rate": 5.453097786600451e-06,
      "loss": 0.3406,
      "step": 31490
    },
    {
      "epoch": 1.4451379009682896,
      "grad_norm": 0.4726647436618805,
      "learning_rate": 5.452853606953796e-06,
      "loss": 0.3332,
      "step": 31491
    },
    {
      "epoch": 1.4451837914735441,
      "grad_norm": 0.4880056381225586,
      "learning_rate": 5.452609426218173e-06,
      "loss": 0.3687,
      "step": 31492
    },
    {
      "epoch": 1.4452296819787986,
      "grad_norm": 0.47249162197113037,
      "learning_rate": 5.452365244394165e-06,
      "loss": 0.3713,
      "step": 31493
    },
    {
      "epoch": 1.4452755724840531,
      "grad_norm": 0.45968177914619446,
      "learning_rate": 5.452121061482363e-06,
      "loss": 0.3794,
      "step": 31494
    },
    {
      "epoch": 1.4453214629893076,
      "grad_norm": 0.4668162763118744,
      "learning_rate": 5.451876877483352e-06,
      "loss": 0.3412,
      "step": 31495
    },
    {
      "epoch": 1.445367353494562,
      "grad_norm": 0.41201239824295044,
      "learning_rate": 5.45163269239772e-06,
      "loss": 0.2639,
      "step": 31496
    },
    {
      "epoch": 1.4454132439998164,
      "grad_norm": 0.497307687997818,
      "learning_rate": 5.451388506226054e-06,
      "loss": 0.4233,
      "step": 31497
    },
    {
      "epoch": 1.445459134505071,
      "grad_norm": 0.4361111521720886,
      "learning_rate": 5.451144318968941e-06,
      "loss": 0.2984,
      "step": 31498
    },
    {
      "epoch": 1.4455050250103254,
      "grad_norm": 0.4693741798400879,
      "learning_rate": 5.450900130626967e-06,
      "loss": 0.3479,
      "step": 31499
    },
    {
      "epoch": 1.44555091551558,
      "grad_norm": 0.5227579474449158,
      "learning_rate": 5.450655941200722e-06,
      "loss": 0.4007,
      "step": 31500
    },
    {
      "epoch": 1.4455968060208342,
      "grad_norm": 0.5005751848220825,
      "learning_rate": 5.450411750690793e-06,
      "loss": 0.3959,
      "step": 31501
    },
    {
      "epoch": 1.4456426965260887,
      "grad_norm": 0.4914509057998657,
      "learning_rate": 5.450167559097763e-06,
      "loss": 0.3707,
      "step": 31502
    },
    {
      "epoch": 1.4456885870313432,
      "grad_norm": 0.46135854721069336,
      "learning_rate": 5.449923366422224e-06,
      "loss": 0.364,
      "step": 31503
    },
    {
      "epoch": 1.4457344775365977,
      "grad_norm": 0.4501906931400299,
      "learning_rate": 5.44967917266476e-06,
      "loss": 0.3078,
      "step": 31504
    },
    {
      "epoch": 1.4457803680418522,
      "grad_norm": 0.46237432956695557,
      "learning_rate": 5.44943497782596e-06,
      "loss": 0.3332,
      "step": 31505
    },
    {
      "epoch": 1.4458262585471067,
      "grad_norm": 0.48096054792404175,
      "learning_rate": 5.4491907819064106e-06,
      "loss": 0.3919,
      "step": 31506
    },
    {
      "epoch": 1.4458721490523612,
      "grad_norm": 0.45157212018966675,
      "learning_rate": 5.4489465849067005e-06,
      "loss": 0.3137,
      "step": 31507
    },
    {
      "epoch": 1.4459180395576157,
      "grad_norm": 0.47216635942459106,
      "learning_rate": 5.448702386827414e-06,
      "loss": 0.3668,
      "step": 31508
    },
    {
      "epoch": 1.44596393006287,
      "grad_norm": 0.47318604588508606,
      "learning_rate": 5.44845818766914e-06,
      "loss": 0.3304,
      "step": 31509
    },
    {
      "epoch": 1.4460098205681244,
      "grad_norm": 0.4905325174331665,
      "learning_rate": 5.448213987432466e-06,
      "loss": 0.4142,
      "step": 31510
    },
    {
      "epoch": 1.446055711073379,
      "grad_norm": 0.4934482276439667,
      "learning_rate": 5.4479697861179795e-06,
      "loss": 0.3646,
      "step": 31511
    },
    {
      "epoch": 1.4461016015786334,
      "grad_norm": 0.43877291679382324,
      "learning_rate": 5.447725583726267e-06,
      "loss": 0.3135,
      "step": 31512
    },
    {
      "epoch": 1.4461474920838877,
      "grad_norm": 0.44059619307518005,
      "learning_rate": 5.4474813802579145e-06,
      "loss": 0.3216,
      "step": 31513
    },
    {
      "epoch": 1.4461933825891422,
      "grad_norm": 0.48452332615852356,
      "learning_rate": 5.44723717571351e-06,
      "loss": 0.3824,
      "step": 31514
    },
    {
      "epoch": 1.4462392730943967,
      "grad_norm": 0.49147099256515503,
      "learning_rate": 5.446992970093643e-06,
      "loss": 0.3841,
      "step": 31515
    },
    {
      "epoch": 1.4462851635996512,
      "grad_norm": 0.49230459332466125,
      "learning_rate": 5.446748763398897e-06,
      "loss": 0.3908,
      "step": 31516
    },
    {
      "epoch": 1.4463310541049057,
      "grad_norm": 0.45573940873146057,
      "learning_rate": 5.446504555629864e-06,
      "loss": 0.357,
      "step": 31517
    },
    {
      "epoch": 1.4463769446101602,
      "grad_norm": 0.4334750473499298,
      "learning_rate": 5.446260346787126e-06,
      "loss": 0.3074,
      "step": 31518
    },
    {
      "epoch": 1.4464228351154147,
      "grad_norm": 0.4894486367702484,
      "learning_rate": 5.4460161368712735e-06,
      "loss": 0.3793,
      "step": 31519
    },
    {
      "epoch": 1.4464687256206692,
      "grad_norm": 0.5001718401908875,
      "learning_rate": 5.445771925882895e-06,
      "loss": 0.381,
      "step": 31520
    },
    {
      "epoch": 1.4465146161259235,
      "grad_norm": 0.48809054493904114,
      "learning_rate": 5.445527713822573e-06,
      "loss": 0.3563,
      "step": 31521
    },
    {
      "epoch": 1.446560506631178,
      "grad_norm": 0.482994019985199,
      "learning_rate": 5.4452835006908986e-06,
      "loss": 0.3875,
      "step": 31522
    },
    {
      "epoch": 1.4466063971364325,
      "grad_norm": 0.5072069764137268,
      "learning_rate": 5.4450392864884584e-06,
      "loss": 0.3783,
      "step": 31523
    },
    {
      "epoch": 1.446652287641687,
      "grad_norm": 0.4639210104942322,
      "learning_rate": 5.444795071215837e-06,
      "loss": 0.3638,
      "step": 31524
    },
    {
      "epoch": 1.4466981781469415,
      "grad_norm": 0.43786078691482544,
      "learning_rate": 5.444550854873625e-06,
      "loss": 0.3097,
      "step": 31525
    },
    {
      "epoch": 1.4467440686521957,
      "grad_norm": 0.45528531074523926,
      "learning_rate": 5.44430663746241e-06,
      "loss": 0.3035,
      "step": 31526
    },
    {
      "epoch": 1.4467899591574502,
      "grad_norm": 0.4688328504562378,
      "learning_rate": 5.444062418982775e-06,
      "loss": 0.3823,
      "step": 31527
    },
    {
      "epoch": 1.4468358496627047,
      "grad_norm": 0.4502474069595337,
      "learning_rate": 5.443818199435312e-06,
      "loss": 0.3171,
      "step": 31528
    },
    {
      "epoch": 1.4468817401679592,
      "grad_norm": 0.4974532723426819,
      "learning_rate": 5.4435739788206065e-06,
      "loss": 0.4246,
      "step": 31529
    },
    {
      "epoch": 1.4469276306732137,
      "grad_norm": 0.7088391184806824,
      "learning_rate": 5.443329757139245e-06,
      "loss": 0.2902,
      "step": 31530
    },
    {
      "epoch": 1.4469735211784682,
      "grad_norm": 0.5362744927406311,
      "learning_rate": 5.4430855343918144e-06,
      "loss": 0.5075,
      "step": 31531
    },
    {
      "epoch": 1.4470194116837227,
      "grad_norm": 0.4511568248271942,
      "learning_rate": 5.442841310578905e-06,
      "loss": 0.3422,
      "step": 31532
    },
    {
      "epoch": 1.4470653021889772,
      "grad_norm": 0.5412117838859558,
      "learning_rate": 5.4425970857011016e-06,
      "loss": 0.4104,
      "step": 31533
    },
    {
      "epoch": 1.4471111926942315,
      "grad_norm": 0.4334595203399658,
      "learning_rate": 5.442352859758991e-06,
      "loss": 0.2889,
      "step": 31534
    },
    {
      "epoch": 1.447157083199486,
      "grad_norm": 0.4764520525932312,
      "learning_rate": 5.4421086327531604e-06,
      "loss": 0.3629,
      "step": 31535
    },
    {
      "epoch": 1.4472029737047405,
      "grad_norm": 0.4808928966522217,
      "learning_rate": 5.441864404684198e-06,
      "loss": 0.3618,
      "step": 31536
    },
    {
      "epoch": 1.447248864209995,
      "grad_norm": 0.44103720784187317,
      "learning_rate": 5.441620175552693e-06,
      "loss": 0.3268,
      "step": 31537
    },
    {
      "epoch": 1.4472947547152495,
      "grad_norm": 0.5004603266716003,
      "learning_rate": 5.4413759453592305e-06,
      "loss": 0.4154,
      "step": 31538
    },
    {
      "epoch": 1.4473406452205038,
      "grad_norm": 0.45953577756881714,
      "learning_rate": 5.4411317141043984e-06,
      "loss": 0.3117,
      "step": 31539
    },
    {
      "epoch": 1.4473865357257583,
      "grad_norm": 0.4854617714881897,
      "learning_rate": 5.440887481788784e-06,
      "loss": 0.359,
      "step": 31540
    },
    {
      "epoch": 1.4474324262310128,
      "grad_norm": 0.4850221574306488,
      "learning_rate": 5.440643248412972e-06,
      "loss": 0.4112,
      "step": 31541
    },
    {
      "epoch": 1.4474783167362673,
      "grad_norm": 0.4714001715183258,
      "learning_rate": 5.440399013977553e-06,
      "loss": 0.3725,
      "step": 31542
    },
    {
      "epoch": 1.4475242072415218,
      "grad_norm": 0.4845878481864929,
      "learning_rate": 5.4401547784831155e-06,
      "loss": 0.4192,
      "step": 31543
    },
    {
      "epoch": 1.4475700977467763,
      "grad_norm": 0.514161229133606,
      "learning_rate": 5.439910541930243e-06,
      "loss": 0.4018,
      "step": 31544
    },
    {
      "epoch": 1.4476159882520307,
      "grad_norm": 0.48714280128479004,
      "learning_rate": 5.439666304319524e-06,
      "loss": 0.3926,
      "step": 31545
    },
    {
      "epoch": 1.4476618787572852,
      "grad_norm": 0.4493511915206909,
      "learning_rate": 5.439422065651547e-06,
      "loss": 0.3108,
      "step": 31546
    },
    {
      "epoch": 1.4477077692625395,
      "grad_norm": 0.5395213961601257,
      "learning_rate": 5.439177825926899e-06,
      "loss": 0.4787,
      "step": 31547
    },
    {
      "epoch": 1.447753659767794,
      "grad_norm": 0.4332595765590668,
      "learning_rate": 5.438933585146166e-06,
      "loss": 0.3108,
      "step": 31548
    },
    {
      "epoch": 1.4477995502730485,
      "grad_norm": 0.45727410912513733,
      "learning_rate": 5.438689343309937e-06,
      "loss": 0.3276,
      "step": 31549
    },
    {
      "epoch": 1.447845440778303,
      "grad_norm": 0.5744962096214294,
      "learning_rate": 5.4384451004187985e-06,
      "loss": 0.3841,
      "step": 31550
    },
    {
      "epoch": 1.4478913312835575,
      "grad_norm": 0.43793392181396484,
      "learning_rate": 5.438200856473339e-06,
      "loss": 0.2916,
      "step": 31551
    },
    {
      "epoch": 1.4479372217888118,
      "grad_norm": 0.4581752121448517,
      "learning_rate": 5.4379566114741425e-06,
      "loss": 0.3608,
      "step": 31552
    },
    {
      "epoch": 1.4479831122940663,
      "grad_norm": 0.47944313287734985,
      "learning_rate": 5.437712365421799e-06,
      "loss": 0.3882,
      "step": 31553
    },
    {
      "epoch": 1.4480290027993208,
      "grad_norm": 0.4713372588157654,
      "learning_rate": 5.437468118316897e-06,
      "loss": 0.3672,
      "step": 31554
    },
    {
      "epoch": 1.4480748933045753,
      "grad_norm": 0.463824599981308,
      "learning_rate": 5.437223870160021e-06,
      "loss": 0.357,
      "step": 31555
    },
    {
      "epoch": 1.4481207838098298,
      "grad_norm": 0.47978249192237854,
      "learning_rate": 5.43697962095176e-06,
      "loss": 0.3291,
      "step": 31556
    },
    {
      "epoch": 1.4481666743150843,
      "grad_norm": 0.45111119747161865,
      "learning_rate": 5.4367353706927e-06,
      "loss": 0.31,
      "step": 31557
    },
    {
      "epoch": 1.4482125648203388,
      "grad_norm": 0.4915436804294586,
      "learning_rate": 5.43649111938343e-06,
      "loss": 0.3713,
      "step": 31558
    },
    {
      "epoch": 1.448258455325593,
      "grad_norm": 0.44064241647720337,
      "learning_rate": 5.4362468670245364e-06,
      "loss": 0.2947,
      "step": 31559
    },
    {
      "epoch": 1.4483043458308476,
      "grad_norm": 0.4814739525318146,
      "learning_rate": 5.436002613616608e-06,
      "loss": 0.3938,
      "step": 31560
    },
    {
      "epoch": 1.448350236336102,
      "grad_norm": 0.5456748604774475,
      "learning_rate": 5.435758359160231e-06,
      "loss": 0.4717,
      "step": 31561
    },
    {
      "epoch": 1.4483961268413565,
      "grad_norm": 0.4593081474304199,
      "learning_rate": 5.43551410365599e-06,
      "loss": 0.3437,
      "step": 31562
    },
    {
      "epoch": 1.448442017346611,
      "grad_norm": 0.48476284742355347,
      "learning_rate": 5.435269847104477e-06,
      "loss": 0.4118,
      "step": 31563
    },
    {
      "epoch": 1.4484879078518653,
      "grad_norm": 0.4382380247116089,
      "learning_rate": 5.435025589506277e-06,
      "loss": 0.3213,
      "step": 31564
    },
    {
      "epoch": 1.4485337983571198,
      "grad_norm": 0.5273840427398682,
      "learning_rate": 5.434781330861978e-06,
      "loss": 0.4325,
      "step": 31565
    },
    {
      "epoch": 1.4485796888623743,
      "grad_norm": 0.43704164028167725,
      "learning_rate": 5.434537071172167e-06,
      "loss": 0.3017,
      "step": 31566
    },
    {
      "epoch": 1.4486255793676288,
      "grad_norm": 0.4042474627494812,
      "learning_rate": 5.4342928104374304e-06,
      "loss": 0.2748,
      "step": 31567
    },
    {
      "epoch": 1.4486714698728833,
      "grad_norm": 0.47312068939208984,
      "learning_rate": 5.434048548658358e-06,
      "loss": 0.384,
      "step": 31568
    },
    {
      "epoch": 1.4487173603781378,
      "grad_norm": 0.42987897992134094,
      "learning_rate": 5.433804285835536e-06,
      "loss": 0.2709,
      "step": 31569
    },
    {
      "epoch": 1.4487632508833923,
      "grad_norm": 0.4516170024871826,
      "learning_rate": 5.433560021969551e-06,
      "loss": 0.3073,
      "step": 31570
    },
    {
      "epoch": 1.4488091413886468,
      "grad_norm": 0.47726088762283325,
      "learning_rate": 5.433315757060991e-06,
      "loss": 0.3984,
      "step": 31571
    },
    {
      "epoch": 1.448855031893901,
      "grad_norm": 0.44944024085998535,
      "learning_rate": 5.4330714911104435e-06,
      "loss": 0.3167,
      "step": 31572
    },
    {
      "epoch": 1.4489009223991556,
      "grad_norm": 0.46468833088874817,
      "learning_rate": 5.432827224118495e-06,
      "loss": 0.3506,
      "step": 31573
    },
    {
      "epoch": 1.44894681290441,
      "grad_norm": 0.44470399618148804,
      "learning_rate": 5.432582956085735e-06,
      "loss": 0.2963,
      "step": 31574
    },
    {
      "epoch": 1.4489927034096646,
      "grad_norm": 0.44895508885383606,
      "learning_rate": 5.432338687012748e-06,
      "loss": 0.3317,
      "step": 31575
    },
    {
      "epoch": 1.449038593914919,
      "grad_norm": 0.5199688673019409,
      "learning_rate": 5.432094416900124e-06,
      "loss": 0.42,
      "step": 31576
    },
    {
      "epoch": 1.4490844844201733,
      "grad_norm": 0.43362748622894287,
      "learning_rate": 5.43185014574845e-06,
      "loss": 0.2725,
      "step": 31577
    },
    {
      "epoch": 1.4491303749254278,
      "grad_norm": 0.45270612835884094,
      "learning_rate": 5.431605873558311e-06,
      "loss": 0.363,
      "step": 31578
    },
    {
      "epoch": 1.4491762654306823,
      "grad_norm": 0.49447566270828247,
      "learning_rate": 5.431361600330296e-06,
      "loss": 0.4202,
      "step": 31579
    },
    {
      "epoch": 1.4492221559359368,
      "grad_norm": 0.40922412276268005,
      "learning_rate": 5.4311173260649945e-06,
      "loss": 0.2857,
      "step": 31580
    },
    {
      "epoch": 1.4492680464411913,
      "grad_norm": 0.49631357192993164,
      "learning_rate": 5.430873050762989e-06,
      "loss": 0.3942,
      "step": 31581
    },
    {
      "epoch": 1.4493139369464458,
      "grad_norm": 0.4973682463169098,
      "learning_rate": 5.430628774424873e-06,
      "loss": 0.4185,
      "step": 31582
    },
    {
      "epoch": 1.4493598274517003,
      "grad_norm": 0.4822652041912079,
      "learning_rate": 5.43038449705123e-06,
      "loss": 0.4198,
      "step": 31583
    },
    {
      "epoch": 1.4494057179569548,
      "grad_norm": 0.5054507851600647,
      "learning_rate": 5.430140218642646e-06,
      "loss": 0.4229,
      "step": 31584
    },
    {
      "epoch": 1.449451608462209,
      "grad_norm": 0.484998881816864,
      "learning_rate": 5.429895939199712e-06,
      "loss": 0.3969,
      "step": 31585
    },
    {
      "epoch": 1.4494974989674636,
      "grad_norm": 0.46955549716949463,
      "learning_rate": 5.429651658723014e-06,
      "loss": 0.349,
      "step": 31586
    },
    {
      "epoch": 1.449543389472718,
      "grad_norm": 0.44023317098617554,
      "learning_rate": 5.42940737721314e-06,
      "loss": 0.3327,
      "step": 31587
    },
    {
      "epoch": 1.4495892799779726,
      "grad_norm": 0.4595310688018799,
      "learning_rate": 5.429163094670675e-06,
      "loss": 0.3281,
      "step": 31588
    },
    {
      "epoch": 1.449635170483227,
      "grad_norm": 0.41622549295425415,
      "learning_rate": 5.42891881109621e-06,
      "loss": 0.2747,
      "step": 31589
    },
    {
      "epoch": 1.4496810609884814,
      "grad_norm": 0.47133758664131165,
      "learning_rate": 5.428674526490328e-06,
      "loss": 0.4077,
      "step": 31590
    },
    {
      "epoch": 1.4497269514937359,
      "grad_norm": 0.4732748568058014,
      "learning_rate": 5.428430240853621e-06,
      "loss": 0.3376,
      "step": 31591
    },
    {
      "epoch": 1.4497728419989904,
      "grad_norm": 0.47463351488113403,
      "learning_rate": 5.428185954186675e-06,
      "loss": 0.3604,
      "step": 31592
    },
    {
      "epoch": 1.4498187325042449,
      "grad_norm": 0.4565639793872833,
      "learning_rate": 5.427941666490077e-06,
      "loss": 0.3554,
      "step": 31593
    },
    {
      "epoch": 1.4498646230094994,
      "grad_norm": 0.43262723088264465,
      "learning_rate": 5.427697377764414e-06,
      "loss": 0.3041,
      "step": 31594
    },
    {
      "epoch": 1.4499105135147539,
      "grad_norm": 0.4597509205341339,
      "learning_rate": 5.427453088010273e-06,
      "loss": 0.3314,
      "step": 31595
    },
    {
      "epoch": 1.4499564040200084,
      "grad_norm": 0.4963235855102539,
      "learning_rate": 5.4272087972282414e-06,
      "loss": 0.3883,
      "step": 31596
    },
    {
      "epoch": 1.4500022945252629,
      "grad_norm": 0.497183620929718,
      "learning_rate": 5.426964505418909e-06,
      "loss": 0.4004,
      "step": 31597
    },
    {
      "epoch": 1.4500481850305171,
      "grad_norm": 0.537064790725708,
      "learning_rate": 5.426720212582862e-06,
      "loss": 0.4417,
      "step": 31598
    },
    {
      "epoch": 1.4500940755357716,
      "grad_norm": 0.5430310964584351,
      "learning_rate": 5.426475918720686e-06,
      "loss": 0.3434,
      "step": 31599
    },
    {
      "epoch": 1.4501399660410261,
      "grad_norm": 0.499609112739563,
      "learning_rate": 5.426231623832971e-06,
      "loss": 0.4599,
      "step": 31600
    },
    {
      "epoch": 1.4501858565462806,
      "grad_norm": 0.4501103460788727,
      "learning_rate": 5.4259873279203035e-06,
      "loss": 0.321,
      "step": 31601
    },
    {
      "epoch": 1.450231747051535,
      "grad_norm": 0.4918352961540222,
      "learning_rate": 5.4257430309832714e-06,
      "loss": 0.3589,
      "step": 31602
    },
    {
      "epoch": 1.4502776375567894,
      "grad_norm": 0.45906490087509155,
      "learning_rate": 5.42549873302246e-06,
      "loss": 0.3443,
      "step": 31603
    },
    {
      "epoch": 1.450323528062044,
      "grad_norm": 0.4886839687824249,
      "learning_rate": 5.4252544340384604e-06,
      "loss": 0.3787,
      "step": 31604
    },
    {
      "epoch": 1.4503694185672984,
      "grad_norm": 0.47225892543792725,
      "learning_rate": 5.425010134031858e-06,
      "loss": 0.3265,
      "step": 31605
    },
    {
      "epoch": 1.450415309072553,
      "grad_norm": 0.4561483561992645,
      "learning_rate": 5.424765833003239e-06,
      "loss": 0.2961,
      "step": 31606
    },
    {
      "epoch": 1.4504611995778074,
      "grad_norm": 0.4830673336982727,
      "learning_rate": 5.424521530953193e-06,
      "loss": 0.4097,
      "step": 31607
    },
    {
      "epoch": 1.450507090083062,
      "grad_norm": 0.4718051552772522,
      "learning_rate": 5.424277227882307e-06,
      "loss": 0.3467,
      "step": 31608
    },
    {
      "epoch": 1.4505529805883164,
      "grad_norm": 0.46575626730918884,
      "learning_rate": 5.424032923791168e-06,
      "loss": 0.3914,
      "step": 31609
    },
    {
      "epoch": 1.4505988710935707,
      "grad_norm": 0.4473998248577118,
      "learning_rate": 5.423788618680364e-06,
      "loss": 0.3269,
      "step": 31610
    },
    {
      "epoch": 1.4506447615988252,
      "grad_norm": 0.4723009765148163,
      "learning_rate": 5.423544312550481e-06,
      "loss": 0.4364,
      "step": 31611
    },
    {
      "epoch": 1.4506906521040797,
      "grad_norm": 0.5076924562454224,
      "learning_rate": 5.423300005402108e-06,
      "loss": 0.4052,
      "step": 31612
    },
    {
      "epoch": 1.4507365426093342,
      "grad_norm": 0.4647425413131714,
      "learning_rate": 5.423055697235832e-06,
      "loss": 0.3597,
      "step": 31613
    },
    {
      "epoch": 1.4507824331145887,
      "grad_norm": 0.46228864789009094,
      "learning_rate": 5.42281138805224e-06,
      "loss": 0.2995,
      "step": 31614
    },
    {
      "epoch": 1.450828323619843,
      "grad_norm": 0.47446712851524353,
      "learning_rate": 5.422567077851922e-06,
      "loss": 0.3526,
      "step": 31615
    },
    {
      "epoch": 1.4508742141250974,
      "grad_norm": 0.4463285207748413,
      "learning_rate": 5.422322766635463e-06,
      "loss": 0.3148,
      "step": 31616
    },
    {
      "epoch": 1.450920104630352,
      "grad_norm": 0.4402204751968384,
      "learning_rate": 5.4220784544034495e-06,
      "loss": 0.3053,
      "step": 31617
    },
    {
      "epoch": 1.4509659951356064,
      "grad_norm": 0.5054766535758972,
      "learning_rate": 5.421834141156471e-06,
      "loss": 0.4464,
      "step": 31618
    },
    {
      "epoch": 1.451011885640861,
      "grad_norm": 0.47673550248146057,
      "learning_rate": 5.4215898268951165e-06,
      "loss": 0.3626,
      "step": 31619
    },
    {
      "epoch": 1.4510577761461154,
      "grad_norm": 0.5053106546401978,
      "learning_rate": 5.42134551161997e-06,
      "loss": 0.3768,
      "step": 31620
    },
    {
      "epoch": 1.45110366665137,
      "grad_norm": 0.49903103709220886,
      "learning_rate": 5.421101195331619e-06,
      "loss": 0.4187,
      "step": 31621
    },
    {
      "epoch": 1.4511495571566244,
      "grad_norm": 0.5214585065841675,
      "learning_rate": 5.420856878030655e-06,
      "loss": 0.4405,
      "step": 31622
    },
    {
      "epoch": 1.4511954476618787,
      "grad_norm": 0.4655395746231079,
      "learning_rate": 5.420612559717662e-06,
      "loss": 0.3906,
      "step": 31623
    },
    {
      "epoch": 1.4512413381671332,
      "grad_norm": 0.47630926966667175,
      "learning_rate": 5.420368240393228e-06,
      "loss": 0.3315,
      "step": 31624
    },
    {
      "epoch": 1.4512872286723877,
      "grad_norm": 0.4782724380493164,
      "learning_rate": 5.4201239200579425e-06,
      "loss": 0.3705,
      "step": 31625
    },
    {
      "epoch": 1.4513331191776422,
      "grad_norm": 0.47093111276626587,
      "learning_rate": 5.419879598712391e-06,
      "loss": 0.3573,
      "step": 31626
    },
    {
      "epoch": 1.4513790096828967,
      "grad_norm": 0.4574928283691406,
      "learning_rate": 5.41963527635716e-06,
      "loss": 0.3442,
      "step": 31627
    },
    {
      "epoch": 1.451424900188151,
      "grad_norm": 0.45991677045822144,
      "learning_rate": 5.4193909529928416e-06,
      "loss": 0.3417,
      "step": 31628
    },
    {
      "epoch": 1.4514707906934055,
      "grad_norm": 0.46784305572509766,
      "learning_rate": 5.419146628620018e-06,
      "loss": 0.3367,
      "step": 31629
    },
    {
      "epoch": 1.45151668119866,
      "grad_norm": 0.4615319073200226,
      "learning_rate": 5.418902303239279e-06,
      "loss": 0.4017,
      "step": 31630
    },
    {
      "epoch": 1.4515625717039145,
      "grad_norm": 0.5220027565956116,
      "learning_rate": 5.418657976851213e-06,
      "loss": 0.455,
      "step": 31631
    },
    {
      "epoch": 1.451608462209169,
      "grad_norm": 0.4927975535392761,
      "learning_rate": 5.418413649456406e-06,
      "loss": 0.3506,
      "step": 31632
    },
    {
      "epoch": 1.4516543527144234,
      "grad_norm": 0.4587079882621765,
      "learning_rate": 5.4181693210554485e-06,
      "loss": 0.3428,
      "step": 31633
    },
    {
      "epoch": 1.451700243219678,
      "grad_norm": 0.44178688526153564,
      "learning_rate": 5.417924991648923e-06,
      "loss": 0.3466,
      "step": 31634
    },
    {
      "epoch": 1.4517461337249324,
      "grad_norm": 0.47204509377479553,
      "learning_rate": 5.4176806612374215e-06,
      "loss": 0.3405,
      "step": 31635
    },
    {
      "epoch": 1.4517920242301867,
      "grad_norm": 0.4837011992931366,
      "learning_rate": 5.417436329821528e-06,
      "loss": 0.4131,
      "step": 31636
    },
    {
      "epoch": 1.4518379147354412,
      "grad_norm": 0.48836588859558105,
      "learning_rate": 5.417191997401834e-06,
      "loss": 0.4062,
      "step": 31637
    },
    {
      "epoch": 1.4518838052406957,
      "grad_norm": 0.45542967319488525,
      "learning_rate": 5.416947663978924e-06,
      "loss": 0.3178,
      "step": 31638
    },
    {
      "epoch": 1.4519296957459502,
      "grad_norm": 0.42953771352767944,
      "learning_rate": 5.416703329553387e-06,
      "loss": 0.322,
      "step": 31639
    },
    {
      "epoch": 1.4519755862512047,
      "grad_norm": 0.46427950263023376,
      "learning_rate": 5.416458994125808e-06,
      "loss": 0.3411,
      "step": 31640
    },
    {
      "epoch": 1.452021476756459,
      "grad_norm": 0.4875447452068329,
      "learning_rate": 5.416214657696778e-06,
      "loss": 0.3666,
      "step": 31641
    },
    {
      "epoch": 1.4520673672617135,
      "grad_norm": 0.513375461101532,
      "learning_rate": 5.415970320266884e-06,
      "loss": 0.4459,
      "step": 31642
    },
    {
      "epoch": 1.452113257766968,
      "grad_norm": 0.4543962776660919,
      "learning_rate": 5.4157259818367125e-06,
      "loss": 0.366,
      "step": 31643
    },
    {
      "epoch": 1.4521591482722225,
      "grad_norm": 0.45241397619247437,
      "learning_rate": 5.415481642406849e-06,
      "loss": 0.3233,
      "step": 31644
    },
    {
      "epoch": 1.452205038777477,
      "grad_norm": 0.5078842043876648,
      "learning_rate": 5.415237301977886e-06,
      "loss": 0.4535,
      "step": 31645
    },
    {
      "epoch": 1.4522509292827315,
      "grad_norm": 0.4627383053302765,
      "learning_rate": 5.414992960550406e-06,
      "loss": 0.3397,
      "step": 31646
    },
    {
      "epoch": 1.452296819787986,
      "grad_norm": 0.5406371355056763,
      "learning_rate": 5.414748618125001e-06,
      "loss": 0.4739,
      "step": 31647
    },
    {
      "epoch": 1.4523427102932402,
      "grad_norm": 0.4885638356208801,
      "learning_rate": 5.414504274702256e-06,
      "loss": 0.3873,
      "step": 31648
    },
    {
      "epoch": 1.4523886007984947,
      "grad_norm": 0.5156583786010742,
      "learning_rate": 5.414259930282758e-06,
      "loss": 0.415,
      "step": 31649
    },
    {
      "epoch": 1.4524344913037492,
      "grad_norm": 0.44994184374809265,
      "learning_rate": 5.414015584867096e-06,
      "loss": 0.3301,
      "step": 31650
    },
    {
      "epoch": 1.4524803818090037,
      "grad_norm": 0.45658576488494873,
      "learning_rate": 5.413771238455857e-06,
      "loss": 0.3248,
      "step": 31651
    },
    {
      "epoch": 1.4525262723142582,
      "grad_norm": 0.5044172406196594,
      "learning_rate": 5.413526891049629e-06,
      "loss": 0.3986,
      "step": 31652
    },
    {
      "epoch": 1.4525721628195125,
      "grad_norm": 0.4753064215183258,
      "learning_rate": 5.413282542649001e-06,
      "loss": 0.3965,
      "step": 31653
    },
    {
      "epoch": 1.452618053324767,
      "grad_norm": 0.4664745628833771,
      "learning_rate": 5.413038193254556e-06,
      "loss": 0.3506,
      "step": 31654
    },
    {
      "epoch": 1.4526639438300215,
      "grad_norm": 0.4888538122177124,
      "learning_rate": 5.412793842866886e-06,
      "loss": 0.3906,
      "step": 31655
    },
    {
      "epoch": 1.452709834335276,
      "grad_norm": 0.43374478816986084,
      "learning_rate": 5.4125494914865776e-06,
      "loss": 0.2834,
      "step": 31656
    },
    {
      "epoch": 1.4527557248405305,
      "grad_norm": 0.4544423520565033,
      "learning_rate": 5.412305139114216e-06,
      "loss": 0.3292,
      "step": 31657
    },
    {
      "epoch": 1.452801615345785,
      "grad_norm": 0.4773028492927551,
      "learning_rate": 5.412060785750393e-06,
      "loss": 0.3658,
      "step": 31658
    },
    {
      "epoch": 1.4528475058510395,
      "grad_norm": 0.47520604729652405,
      "learning_rate": 5.4118164313956936e-06,
      "loss": 0.3688,
      "step": 31659
    },
    {
      "epoch": 1.452893396356294,
      "grad_norm": 0.4983828663825989,
      "learning_rate": 5.411572076050704e-06,
      "loss": 0.3931,
      "step": 31660
    },
    {
      "epoch": 1.4529392868615483,
      "grad_norm": 0.4865405559539795,
      "learning_rate": 5.411327719716013e-06,
      "loss": 0.346,
      "step": 31661
    },
    {
      "epoch": 1.4529851773668028,
      "grad_norm": 0.4596133828163147,
      "learning_rate": 5.411083362392211e-06,
      "loss": 0.323,
      "step": 31662
    },
    {
      "epoch": 1.4530310678720573,
      "grad_norm": 0.463183730840683,
      "learning_rate": 5.4108390040798815e-06,
      "loss": 0.3429,
      "step": 31663
    },
    {
      "epoch": 1.4530769583773118,
      "grad_norm": 0.4940626621246338,
      "learning_rate": 5.410594644779615e-06,
      "loss": 0.3925,
      "step": 31664
    },
    {
      "epoch": 1.4531228488825663,
      "grad_norm": 0.4246724247932434,
      "learning_rate": 5.410350284491998e-06,
      "loss": 0.3208,
      "step": 31665
    },
    {
      "epoch": 1.4531687393878205,
      "grad_norm": 0.45821335911750793,
      "learning_rate": 5.410105923217617e-06,
      "loss": 0.3585,
      "step": 31666
    },
    {
      "epoch": 1.453214629893075,
      "grad_norm": 0.4330706298351288,
      "learning_rate": 5.409861560957062e-06,
      "loss": 0.2908,
      "step": 31667
    },
    {
      "epoch": 1.4532605203983295,
      "grad_norm": 0.44777190685272217,
      "learning_rate": 5.409617197710918e-06,
      "loss": 0.3318,
      "step": 31668
    },
    {
      "epoch": 1.453306410903584,
      "grad_norm": 0.44717422127723694,
      "learning_rate": 5.409372833479775e-06,
      "loss": 0.3287,
      "step": 31669
    },
    {
      "epoch": 1.4533523014088385,
      "grad_norm": 0.4453507959842682,
      "learning_rate": 5.40912846826422e-06,
      "loss": 0.2963,
      "step": 31670
    },
    {
      "epoch": 1.453398191914093,
      "grad_norm": 0.4331429600715637,
      "learning_rate": 5.408884102064837e-06,
      "loss": 0.2897,
      "step": 31671
    },
    {
      "epoch": 1.4534440824193475,
      "grad_norm": 0.48981642723083496,
      "learning_rate": 5.408639734882219e-06,
      "loss": 0.3694,
      "step": 31672
    },
    {
      "epoch": 1.453489972924602,
      "grad_norm": 0.4838477373123169,
      "learning_rate": 5.408395366716952e-06,
      "loss": 0.3563,
      "step": 31673
    },
    {
      "epoch": 1.4535358634298563,
      "grad_norm": 0.4682222902774811,
      "learning_rate": 5.408150997569622e-06,
      "loss": 0.366,
      "step": 31674
    },
    {
      "epoch": 1.4535817539351108,
      "grad_norm": 0.42475154995918274,
      "learning_rate": 5.407906627440819e-06,
      "loss": 0.2972,
      "step": 31675
    },
    {
      "epoch": 1.4536276444403653,
      "grad_norm": 0.4580738842487335,
      "learning_rate": 5.407662256331129e-06,
      "loss": 0.3021,
      "step": 31676
    },
    {
      "epoch": 1.4536735349456198,
      "grad_norm": 0.48073810338974,
      "learning_rate": 5.407417884241139e-06,
      "loss": 0.3635,
      "step": 31677
    },
    {
      "epoch": 1.4537194254508743,
      "grad_norm": 0.5158578753471375,
      "learning_rate": 5.407173511171436e-06,
      "loss": 0.4804,
      "step": 31678
    },
    {
      "epoch": 1.4537653159561286,
      "grad_norm": 0.4345202147960663,
      "learning_rate": 5.406929137122611e-06,
      "loss": 0.315,
      "step": 31679
    },
    {
      "epoch": 1.453811206461383,
      "grad_norm": 0.48770198225975037,
      "learning_rate": 5.406684762095251e-06,
      "loss": 0.3861,
      "step": 31680
    },
    {
      "epoch": 1.4538570969666376,
      "grad_norm": 0.4548685848712921,
      "learning_rate": 5.406440386089942e-06,
      "loss": 0.3156,
      "step": 31681
    },
    {
      "epoch": 1.453902987471892,
      "grad_norm": 0.4704999029636383,
      "learning_rate": 5.406196009107272e-06,
      "loss": 0.3198,
      "step": 31682
    },
    {
      "epoch": 1.4539488779771466,
      "grad_norm": 0.44958823919296265,
      "learning_rate": 5.4059516311478276e-06,
      "loss": 0.3172,
      "step": 31683
    },
    {
      "epoch": 1.453994768482401,
      "grad_norm": 0.46236634254455566,
      "learning_rate": 5.405707252212199e-06,
      "loss": 0.3178,
      "step": 31684
    },
    {
      "epoch": 1.4540406589876556,
      "grad_norm": 0.4707605540752411,
      "learning_rate": 5.4054628723009705e-06,
      "loss": 0.3904,
      "step": 31685
    },
    {
      "epoch": 1.45408654949291,
      "grad_norm": 0.43966299295425415,
      "learning_rate": 5.405218491414734e-06,
      "loss": 0.3162,
      "step": 31686
    },
    {
      "epoch": 1.4541324399981643,
      "grad_norm": 0.5429790616035461,
      "learning_rate": 5.4049741095540755e-06,
      "loss": 0.3775,
      "step": 31687
    },
    {
      "epoch": 1.4541783305034188,
      "grad_norm": 0.5067269802093506,
      "learning_rate": 5.40472972671958e-06,
      "loss": 0.4017,
      "step": 31688
    },
    {
      "epoch": 1.4542242210086733,
      "grad_norm": 0.48501306772232056,
      "learning_rate": 5.404485342911837e-06,
      "loss": 0.3803,
      "step": 31689
    },
    {
      "epoch": 1.4542701115139278,
      "grad_norm": 0.4812372326850891,
      "learning_rate": 5.404240958131436e-06,
      "loss": 0.3623,
      "step": 31690
    },
    {
      "epoch": 1.454316002019182,
      "grad_norm": 0.4632980525493622,
      "learning_rate": 5.403996572378963e-06,
      "loss": 0.3815,
      "step": 31691
    },
    {
      "epoch": 1.4543618925244366,
      "grad_norm": 0.42121750116348267,
      "learning_rate": 5.403752185655005e-06,
      "loss": 0.2948,
      "step": 31692
    },
    {
      "epoch": 1.454407783029691,
      "grad_norm": 0.5320842266082764,
      "learning_rate": 5.40350779796015e-06,
      "loss": 0.4286,
      "step": 31693
    },
    {
      "epoch": 1.4544536735349456,
      "grad_norm": 0.47411030530929565,
      "learning_rate": 5.403263409294986e-06,
      "loss": 0.3627,
      "step": 31694
    },
    {
      "epoch": 1.4544995640402,
      "grad_norm": 0.46880394220352173,
      "learning_rate": 5.4030190196601015e-06,
      "loss": 0.3701,
      "step": 31695
    },
    {
      "epoch": 1.4545454545454546,
      "grad_norm": 0.4768487215042114,
      "learning_rate": 5.402774629056084e-06,
      "loss": 0.388,
      "step": 31696
    },
    {
      "epoch": 1.454591345050709,
      "grad_norm": 0.486568808555603,
      "learning_rate": 5.402530237483521e-06,
      "loss": 0.3951,
      "step": 31697
    },
    {
      "epoch": 1.4546372355559636,
      "grad_norm": 0.42959803342819214,
      "learning_rate": 5.402285844942999e-06,
      "loss": 0.2645,
      "step": 31698
    },
    {
      "epoch": 1.4546831260612179,
      "grad_norm": 0.46651342511177063,
      "learning_rate": 5.402041451435106e-06,
      "loss": 0.4098,
      "step": 31699
    },
    {
      "epoch": 1.4547290165664724,
      "grad_norm": 0.5194833874702454,
      "learning_rate": 5.4017970569604295e-06,
      "loss": 0.4452,
      "step": 31700
    },
    {
      "epoch": 1.4547749070717269,
      "grad_norm": 0.48392850160598755,
      "learning_rate": 5.40155266151956e-06,
      "loss": 0.3941,
      "step": 31701
    },
    {
      "epoch": 1.4548207975769814,
      "grad_norm": 0.48023393750190735,
      "learning_rate": 5.401308265113082e-06,
      "loss": 0.3736,
      "step": 31702
    },
    {
      "epoch": 1.4548666880822358,
      "grad_norm": 0.4706224203109741,
      "learning_rate": 5.401063867741584e-06,
      "loss": 0.3447,
      "step": 31703
    },
    {
      "epoch": 1.4549125785874901,
      "grad_norm": 0.46051982045173645,
      "learning_rate": 5.4008194694056545e-06,
      "loss": 0.3691,
      "step": 31704
    },
    {
      "epoch": 1.4549584690927446,
      "grad_norm": 0.45851796865463257,
      "learning_rate": 5.40057507010588e-06,
      "loss": 0.3666,
      "step": 31705
    },
    {
      "epoch": 1.4550043595979991,
      "grad_norm": 0.4726441502571106,
      "learning_rate": 5.400330669842849e-06,
      "loss": 0.3435,
      "step": 31706
    },
    {
      "epoch": 1.4550502501032536,
      "grad_norm": 0.4611124098300934,
      "learning_rate": 5.400086268617151e-06,
      "loss": 0.3733,
      "step": 31707
    },
    {
      "epoch": 1.4550961406085081,
      "grad_norm": 0.4720187783241272,
      "learning_rate": 5.39984186642937e-06,
      "loss": 0.3392,
      "step": 31708
    },
    {
      "epoch": 1.4551420311137626,
      "grad_norm": 0.43098124861717224,
      "learning_rate": 5.399597463280095e-06,
      "loss": 0.3112,
      "step": 31709
    },
    {
      "epoch": 1.455187921619017,
      "grad_norm": 0.452161580324173,
      "learning_rate": 5.399353059169916e-06,
      "loss": 0.3045,
      "step": 31710
    },
    {
      "epoch": 1.4552338121242716,
      "grad_norm": 0.4700487554073334,
      "learning_rate": 5.399108654099418e-06,
      "loss": 0.3659,
      "step": 31711
    },
    {
      "epoch": 1.4552797026295259,
      "grad_norm": 0.4566735625267029,
      "learning_rate": 5.398864248069189e-06,
      "loss": 0.3804,
      "step": 31712
    },
    {
      "epoch": 1.4553255931347804,
      "grad_norm": 0.5041570663452148,
      "learning_rate": 5.398619841079818e-06,
      "loss": 0.417,
      "step": 31713
    },
    {
      "epoch": 1.4553714836400349,
      "grad_norm": 0.5136959552764893,
      "learning_rate": 5.3983754331318915e-06,
      "loss": 0.4331,
      "step": 31714
    },
    {
      "epoch": 1.4554173741452894,
      "grad_norm": 0.40874212980270386,
      "learning_rate": 5.398131024226e-06,
      "loss": 0.2639,
      "step": 31715
    },
    {
      "epoch": 1.4554632646505439,
      "grad_norm": 0.46759033203125,
      "learning_rate": 5.397886614362726e-06,
      "loss": 0.3704,
      "step": 31716
    },
    {
      "epoch": 1.4555091551557982,
      "grad_norm": 0.47300708293914795,
      "learning_rate": 5.3976422035426615e-06,
      "loss": 0.3936,
      "step": 31717
    },
    {
      "epoch": 1.4555550456610526,
      "grad_norm": 0.45352593064308167,
      "learning_rate": 5.397397791766393e-06,
      "loss": 0.3832,
      "step": 31718
    },
    {
      "epoch": 1.4556009361663071,
      "grad_norm": 0.5446633100509644,
      "learning_rate": 5.39715337903451e-06,
      "loss": 0.4925,
      "step": 31719
    },
    {
      "epoch": 1.4556468266715616,
      "grad_norm": 0.4448098838329315,
      "learning_rate": 5.396908965347596e-06,
      "loss": 0.2956,
      "step": 31720
    },
    {
      "epoch": 1.4556927171768161,
      "grad_norm": 0.44262444972991943,
      "learning_rate": 5.396664550706242e-06,
      "loss": 0.3123,
      "step": 31721
    },
    {
      "epoch": 1.4557386076820706,
      "grad_norm": 0.4554646611213684,
      "learning_rate": 5.396420135111035e-06,
      "loss": 0.3376,
      "step": 31722
    },
    {
      "epoch": 1.4557844981873251,
      "grad_norm": 0.4622972309589386,
      "learning_rate": 5.396175718562563e-06,
      "loss": 0.3306,
      "step": 31723
    },
    {
      "epoch": 1.4558303886925796,
      "grad_norm": 0.4878419041633606,
      "learning_rate": 5.3959313010614135e-06,
      "loss": 0.3912,
      "step": 31724
    },
    {
      "epoch": 1.455876279197834,
      "grad_norm": 0.44706910848617554,
      "learning_rate": 5.395686882608173e-06,
      "loss": 0.316,
      "step": 31725
    },
    {
      "epoch": 1.4559221697030884,
      "grad_norm": 0.44238710403442383,
      "learning_rate": 5.395442463203431e-06,
      "loss": 0.3369,
      "step": 31726
    },
    {
      "epoch": 1.455968060208343,
      "grad_norm": 0.4819006621837616,
      "learning_rate": 5.395198042847776e-06,
      "loss": 0.3574,
      "step": 31727
    },
    {
      "epoch": 1.4560139507135974,
      "grad_norm": 0.5023280382156372,
      "learning_rate": 5.394953621541792e-06,
      "loss": 0.3969,
      "step": 31728
    },
    {
      "epoch": 1.456059841218852,
      "grad_norm": 0.5112073421478271,
      "learning_rate": 5.394709199286071e-06,
      "loss": 0.4437,
      "step": 31729
    },
    {
      "epoch": 1.4561057317241062,
      "grad_norm": 0.49302011728286743,
      "learning_rate": 5.3944647760812e-06,
      "loss": 0.3514,
      "step": 31730
    },
    {
      "epoch": 1.4561516222293607,
      "grad_norm": 0.45961514115333557,
      "learning_rate": 5.394220351927763e-06,
      "loss": 0.3285,
      "step": 31731
    },
    {
      "epoch": 1.4561975127346152,
      "grad_norm": 0.4241871237754822,
      "learning_rate": 5.393975926826352e-06,
      "loss": 0.3173,
      "step": 31732
    },
    {
      "epoch": 1.4562434032398697,
      "grad_norm": 0.46221402287483215,
      "learning_rate": 5.3937315007775526e-06,
      "loss": 0.3309,
      "step": 31733
    },
    {
      "epoch": 1.4562892937451242,
      "grad_norm": 0.44979017972946167,
      "learning_rate": 5.393487073781953e-06,
      "loss": 0.3167,
      "step": 31734
    },
    {
      "epoch": 1.4563351842503787,
      "grad_norm": 0.4798332154750824,
      "learning_rate": 5.393242645840143e-06,
      "loss": 0.3991,
      "step": 31735
    },
    {
      "epoch": 1.4563810747556332,
      "grad_norm": 0.4460448622703552,
      "learning_rate": 5.392998216952706e-06,
      "loss": 0.3168,
      "step": 31736
    },
    {
      "epoch": 1.4564269652608874,
      "grad_norm": 0.4373162090778351,
      "learning_rate": 5.392753787120234e-06,
      "loss": 0.2747,
      "step": 31737
    },
    {
      "epoch": 1.456472855766142,
      "grad_norm": 0.49573132395744324,
      "learning_rate": 5.392509356343313e-06,
      "loss": 0.398,
      "step": 31738
    },
    {
      "epoch": 1.4565187462713964,
      "grad_norm": 0.4487608075141907,
      "learning_rate": 5.39226492462253e-06,
      "loss": 0.3212,
      "step": 31739
    },
    {
      "epoch": 1.456564636776651,
      "grad_norm": 0.482695609331131,
      "learning_rate": 5.392020491958474e-06,
      "loss": 0.3853,
      "step": 31740
    },
    {
      "epoch": 1.4566105272819054,
      "grad_norm": 0.47517502307891846,
      "learning_rate": 5.391776058351735e-06,
      "loss": 0.3778,
      "step": 31741
    },
    {
      "epoch": 1.4566564177871597,
      "grad_norm": 0.46444737911224365,
      "learning_rate": 5.391531623802893e-06,
      "loss": 0.3443,
      "step": 31742
    },
    {
      "epoch": 1.4567023082924142,
      "grad_norm": 0.5038636326789856,
      "learning_rate": 5.391287188312544e-06,
      "loss": 0.4268,
      "step": 31743
    },
    {
      "epoch": 1.4567481987976687,
      "grad_norm": 0.4671814441680908,
      "learning_rate": 5.391042751881274e-06,
      "loss": 0.3648,
      "step": 31744
    },
    {
      "epoch": 1.4567940893029232,
      "grad_norm": 0.5118071436882019,
      "learning_rate": 5.390798314509669e-06,
      "loss": 0.4689,
      "step": 31745
    },
    {
      "epoch": 1.4568399798081777,
      "grad_norm": 0.4481141269207001,
      "learning_rate": 5.3905538761983155e-06,
      "loss": 0.3426,
      "step": 31746
    },
    {
      "epoch": 1.4568858703134322,
      "grad_norm": 0.4869726300239563,
      "learning_rate": 5.390309436947805e-06,
      "loss": 0.399,
      "step": 31747
    },
    {
      "epoch": 1.4569317608186867,
      "grad_norm": 0.45881137251853943,
      "learning_rate": 5.390064996758723e-06,
      "loss": 0.3683,
      "step": 31748
    },
    {
      "epoch": 1.4569776513239412,
      "grad_norm": 0.49615198373794556,
      "learning_rate": 5.389820555631658e-06,
      "loss": 0.3428,
      "step": 31749
    },
    {
      "epoch": 1.4570235418291955,
      "grad_norm": 0.463264137506485,
      "learning_rate": 5.389576113567196e-06,
      "loss": 0.3802,
      "step": 31750
    },
    {
      "epoch": 1.45706943233445,
      "grad_norm": 0.5372828245162964,
      "learning_rate": 5.389331670565928e-06,
      "loss": 0.453,
      "step": 31751
    },
    {
      "epoch": 1.4571153228397045,
      "grad_norm": 0.4853935241699219,
      "learning_rate": 5.38908722662844e-06,
      "loss": 0.393,
      "step": 31752
    },
    {
      "epoch": 1.457161213344959,
      "grad_norm": 0.4778353273868561,
      "learning_rate": 5.3888427817553205e-06,
      "loss": 0.3854,
      "step": 31753
    },
    {
      "epoch": 1.4572071038502135,
      "grad_norm": 0.4788605570793152,
      "learning_rate": 5.388598335947155e-06,
      "loss": 0.3577,
      "step": 31754
    },
    {
      "epoch": 1.4572529943554677,
      "grad_norm": 0.580269992351532,
      "learning_rate": 5.3883538892045354e-06,
      "loss": 0.3927,
      "step": 31755
    },
    {
      "epoch": 1.4572988848607222,
      "grad_norm": 0.43341192603111267,
      "learning_rate": 5.388109441528046e-06,
      "loss": 0.3225,
      "step": 31756
    },
    {
      "epoch": 1.4573447753659767,
      "grad_norm": 0.4422518312931061,
      "learning_rate": 5.387864992918276e-06,
      "loss": 0.2914,
      "step": 31757
    },
    {
      "epoch": 1.4573906658712312,
      "grad_norm": 0.5456418991088867,
      "learning_rate": 5.387620543375814e-06,
      "loss": 0.3489,
      "step": 31758
    },
    {
      "epoch": 1.4574365563764857,
      "grad_norm": 0.454874187707901,
      "learning_rate": 5.387376092901245e-06,
      "loss": 0.3548,
      "step": 31759
    },
    {
      "epoch": 1.4574824468817402,
      "grad_norm": 0.5220717787742615,
      "learning_rate": 5.387131641495159e-06,
      "loss": 0.4282,
      "step": 31760
    },
    {
      "epoch": 1.4575283373869947,
      "grad_norm": 0.5027706623077393,
      "learning_rate": 5.386887189158145e-06,
      "loss": 0.4412,
      "step": 31761
    },
    {
      "epoch": 1.4575742278922492,
      "grad_norm": 0.5147600769996643,
      "learning_rate": 5.386642735890791e-06,
      "loss": 0.3795,
      "step": 31762
    },
    {
      "epoch": 1.4576201183975035,
      "grad_norm": 0.5040339827537537,
      "learning_rate": 5.38639828169368e-06,
      "loss": 0.4402,
      "step": 31763
    },
    {
      "epoch": 1.457666008902758,
      "grad_norm": 0.4829963445663452,
      "learning_rate": 5.386153826567404e-06,
      "loss": 0.3577,
      "step": 31764
    },
    {
      "epoch": 1.4577118994080125,
      "grad_norm": 0.49548453092575073,
      "learning_rate": 5.38590937051255e-06,
      "loss": 0.4109,
      "step": 31765
    },
    {
      "epoch": 1.457757789913267,
      "grad_norm": 0.4876136779785156,
      "learning_rate": 5.385664913529706e-06,
      "loss": 0.3866,
      "step": 31766
    },
    {
      "epoch": 1.4578036804185215,
      "grad_norm": 0.46445268392562866,
      "learning_rate": 5.38542045561946e-06,
      "loss": 0.3495,
      "step": 31767
    },
    {
      "epoch": 1.4578495709237758,
      "grad_norm": 0.4918076694011688,
      "learning_rate": 5.385175996782399e-06,
      "loss": 0.3771,
      "step": 31768
    },
    {
      "epoch": 1.4578954614290303,
      "grad_norm": 0.46131467819213867,
      "learning_rate": 5.384931537019111e-06,
      "loss": 0.3647,
      "step": 31769
    },
    {
      "epoch": 1.4579413519342848,
      "grad_norm": 0.43847426772117615,
      "learning_rate": 5.384687076330184e-06,
      "loss": 0.2933,
      "step": 31770
    },
    {
      "epoch": 1.4579872424395393,
      "grad_norm": 0.47957944869995117,
      "learning_rate": 5.384442614716206e-06,
      "loss": 0.3892,
      "step": 31771
    },
    {
      "epoch": 1.4580331329447938,
      "grad_norm": 0.48010751605033875,
      "learning_rate": 5.384198152177765e-06,
      "loss": 0.4048,
      "step": 31772
    },
    {
      "epoch": 1.4580790234500483,
      "grad_norm": 0.44766220450401306,
      "learning_rate": 5.38395368871545e-06,
      "loss": 0.3355,
      "step": 31773
    },
    {
      "epoch": 1.4581249139553027,
      "grad_norm": 0.4655357599258423,
      "learning_rate": 5.383709224329846e-06,
      "loss": 0.3722,
      "step": 31774
    },
    {
      "epoch": 1.4581708044605572,
      "grad_norm": 0.46345436573028564,
      "learning_rate": 5.383464759021544e-06,
      "loss": 0.341,
      "step": 31775
    },
    {
      "epoch": 1.4582166949658115,
      "grad_norm": 0.4936167299747467,
      "learning_rate": 5.383220292791128e-06,
      "loss": 0.4237,
      "step": 31776
    },
    {
      "epoch": 1.458262585471066,
      "grad_norm": 0.46812498569488525,
      "learning_rate": 5.3829758256391895e-06,
      "loss": 0.34,
      "step": 31777
    },
    {
      "epoch": 1.4583084759763205,
      "grad_norm": 0.4803893566131592,
      "learning_rate": 5.3827313575663155e-06,
      "loss": 0.3865,
      "step": 31778
    },
    {
      "epoch": 1.458354366481575,
      "grad_norm": 0.44903597235679626,
      "learning_rate": 5.382486888573093e-06,
      "loss": 0.33,
      "step": 31779
    },
    {
      "epoch": 1.4584002569868293,
      "grad_norm": 0.4652249217033386,
      "learning_rate": 5.38224241866011e-06,
      "loss": 0.3265,
      "step": 31780
    },
    {
      "epoch": 1.4584461474920838,
      "grad_norm": 0.5255602598190308,
      "learning_rate": 5.381997947827954e-06,
      "loss": 0.4483,
      "step": 31781
    },
    {
      "epoch": 1.4584920379973383,
      "grad_norm": 0.4681764841079712,
      "learning_rate": 5.381753476077214e-06,
      "loss": 0.3246,
      "step": 31782
    },
    {
      "epoch": 1.4585379285025928,
      "grad_norm": 0.45895472168922424,
      "learning_rate": 5.381509003408477e-06,
      "loss": 0.3661,
      "step": 31783
    },
    {
      "epoch": 1.4585838190078473,
      "grad_norm": 0.48549172282218933,
      "learning_rate": 5.3812645298223335e-06,
      "loss": 0.4143,
      "step": 31784
    },
    {
      "epoch": 1.4586297095131018,
      "grad_norm": 0.4738878309726715,
      "learning_rate": 5.3810200553193655e-06,
      "loss": 0.3418,
      "step": 31785
    },
    {
      "epoch": 1.4586756000183563,
      "grad_norm": 0.47453609108924866,
      "learning_rate": 5.380775579900167e-06,
      "loss": 0.3822,
      "step": 31786
    },
    {
      "epoch": 1.4587214905236108,
      "grad_norm": 0.4774502217769623,
      "learning_rate": 5.380531103565323e-06,
      "loss": 0.4415,
      "step": 31787
    },
    {
      "epoch": 1.458767381028865,
      "grad_norm": 0.48874348402023315,
      "learning_rate": 5.380286626315422e-06,
      "loss": 0.3896,
      "step": 31788
    },
    {
      "epoch": 1.4588132715341195,
      "grad_norm": 0.48349034786224365,
      "learning_rate": 5.3800421481510515e-06,
      "loss": 0.3891,
      "step": 31789
    },
    {
      "epoch": 1.458859162039374,
      "grad_norm": 0.46257421374320984,
      "learning_rate": 5.379797669072798e-06,
      "loss": 0.3315,
      "step": 31790
    },
    {
      "epoch": 1.4589050525446285,
      "grad_norm": 0.4699471592903137,
      "learning_rate": 5.379553189081252e-06,
      "loss": 0.3937,
      "step": 31791
    },
    {
      "epoch": 1.458950943049883,
      "grad_norm": 0.4804019331932068,
      "learning_rate": 5.379308708177001e-06,
      "loss": 0.3654,
      "step": 31792
    },
    {
      "epoch": 1.4589968335551373,
      "grad_norm": 0.45206400752067566,
      "learning_rate": 5.379064226360632e-06,
      "loss": 0.3551,
      "step": 31793
    },
    {
      "epoch": 1.4590427240603918,
      "grad_norm": 0.4776676297187805,
      "learning_rate": 5.378819743632734e-06,
      "loss": 0.3538,
      "step": 31794
    },
    {
      "epoch": 1.4590886145656463,
      "grad_norm": 0.44597911834716797,
      "learning_rate": 5.378575259993893e-06,
      "loss": 0.3637,
      "step": 31795
    },
    {
      "epoch": 1.4591345050709008,
      "grad_norm": 0.4166402220726013,
      "learning_rate": 5.3783307754446976e-06,
      "loss": 0.2861,
      "step": 31796
    },
    {
      "epoch": 1.4591803955761553,
      "grad_norm": 0.4976336658000946,
      "learning_rate": 5.378086289985738e-06,
      "loss": 0.3887,
      "step": 31797
    },
    {
      "epoch": 1.4592262860814098,
      "grad_norm": 0.4735249876976013,
      "learning_rate": 5.3778418036175985e-06,
      "loss": 0.3545,
      "step": 31798
    },
    {
      "epoch": 1.4592721765866643,
      "grad_norm": 0.46378806233406067,
      "learning_rate": 5.377597316340869e-06,
      "loss": 0.3275,
      "step": 31799
    },
    {
      "epoch": 1.4593180670919188,
      "grad_norm": 0.44812238216400146,
      "learning_rate": 5.377352828156137e-06,
      "loss": 0.3366,
      "step": 31800
    },
    {
      "epoch": 1.459363957597173,
      "grad_norm": 0.5025234222412109,
      "learning_rate": 5.377108339063991e-06,
      "loss": 0.3653,
      "step": 31801
    },
    {
      "epoch": 1.4594098481024276,
      "grad_norm": 0.4662858843803406,
      "learning_rate": 5.376863849065017e-06,
      "loss": 0.4061,
      "step": 31802
    },
    {
      "epoch": 1.459455738607682,
      "grad_norm": 0.4641208052635193,
      "learning_rate": 5.376619358159807e-06,
      "loss": 0.3338,
      "step": 31803
    },
    {
      "epoch": 1.4595016291129366,
      "grad_norm": 0.4493691027164459,
      "learning_rate": 5.376374866348944e-06,
      "loss": 0.3231,
      "step": 31804
    },
    {
      "epoch": 1.459547519618191,
      "grad_norm": 0.47502127289772034,
      "learning_rate": 5.376130373633019e-06,
      "loss": 0.3631,
      "step": 31805
    },
    {
      "epoch": 1.4595934101234453,
      "grad_norm": 0.5112781524658203,
      "learning_rate": 5.37588588001262e-06,
      "loss": 0.439,
      "step": 31806
    },
    {
      "epoch": 1.4596393006286998,
      "grad_norm": 0.5035282969474792,
      "learning_rate": 5.375641385488333e-06,
      "loss": 0.4192,
      "step": 31807
    },
    {
      "epoch": 1.4596851911339543,
      "grad_norm": 0.4740073084831238,
      "learning_rate": 5.375396890060746e-06,
      "loss": 0.3395,
      "step": 31808
    },
    {
      "epoch": 1.4597310816392088,
      "grad_norm": 0.4311659336090088,
      "learning_rate": 5.375152393730451e-06,
      "loss": 0.322,
      "step": 31809
    },
    {
      "epoch": 1.4597769721444633,
      "grad_norm": 0.48666098713874817,
      "learning_rate": 5.3749078964980315e-06,
      "loss": 0.343,
      "step": 31810
    },
    {
      "epoch": 1.4598228626497178,
      "grad_norm": 0.44900816679000854,
      "learning_rate": 5.374663398364076e-06,
      "loss": 0.3264,
      "step": 31811
    },
    {
      "epoch": 1.4598687531549723,
      "grad_norm": 0.4609900414943695,
      "learning_rate": 5.374418899329175e-06,
      "loss": 0.3812,
      "step": 31812
    },
    {
      "epoch": 1.4599146436602268,
      "grad_norm": 0.45296239852905273,
      "learning_rate": 5.374174399393914e-06,
      "loss": 0.3575,
      "step": 31813
    },
    {
      "epoch": 1.459960534165481,
      "grad_norm": 0.4739902913570404,
      "learning_rate": 5.373929898558881e-06,
      "loss": 0.346,
      "step": 31814
    },
    {
      "epoch": 1.4600064246707356,
      "grad_norm": 0.4945509135723114,
      "learning_rate": 5.3736853968246635e-06,
      "loss": 0.384,
      "step": 31815
    },
    {
      "epoch": 1.46005231517599,
      "grad_norm": 0.4569809138774872,
      "learning_rate": 5.3734408941918525e-06,
      "loss": 0.3309,
      "step": 31816
    },
    {
      "epoch": 1.4600982056812446,
      "grad_norm": 0.4636821150779724,
      "learning_rate": 5.373196390661034e-06,
      "loss": 0.3383,
      "step": 31817
    },
    {
      "epoch": 1.460144096186499,
      "grad_norm": 0.4706510901451111,
      "learning_rate": 5.372951886232795e-06,
      "loss": 0.3125,
      "step": 31818
    },
    {
      "epoch": 1.4601899866917534,
      "grad_norm": 0.48894205689430237,
      "learning_rate": 5.372707380907725e-06,
      "loss": 0.4102,
      "step": 31819
    },
    {
      "epoch": 1.4602358771970079,
      "grad_norm": 0.4381929337978363,
      "learning_rate": 5.3724628746864105e-06,
      "loss": 0.2922,
      "step": 31820
    },
    {
      "epoch": 1.4602817677022624,
      "grad_norm": 0.5284765362739563,
      "learning_rate": 5.372218367569442e-06,
      "loss": 0.3897,
      "step": 31821
    },
    {
      "epoch": 1.4603276582075169,
      "grad_norm": 0.47020918130874634,
      "learning_rate": 5.371973859557404e-06,
      "loss": 0.3815,
      "step": 31822
    },
    {
      "epoch": 1.4603735487127714,
      "grad_norm": 0.4818968176841736,
      "learning_rate": 5.371729350650889e-06,
      "loss": 0.3741,
      "step": 31823
    },
    {
      "epoch": 1.4604194392180259,
      "grad_norm": 0.44925743341445923,
      "learning_rate": 5.371484840850478e-06,
      "loss": 0.301,
      "step": 31824
    },
    {
      "epoch": 1.4604653297232804,
      "grad_norm": 0.5215743184089661,
      "learning_rate": 5.371240330156766e-06,
      "loss": 0.3676,
      "step": 31825
    },
    {
      "epoch": 1.4605112202285346,
      "grad_norm": 0.5002943873405457,
      "learning_rate": 5.370995818570338e-06,
      "loss": 0.3457,
      "step": 31826
    },
    {
      "epoch": 1.4605571107337891,
      "grad_norm": 0.4734555184841156,
      "learning_rate": 5.3707513060917824e-06,
      "loss": 0.3988,
      "step": 31827
    },
    {
      "epoch": 1.4606030012390436,
      "grad_norm": 0.48917949199676514,
      "learning_rate": 5.370506792721687e-06,
      "loss": 0.3829,
      "step": 31828
    },
    {
      "epoch": 1.4606488917442981,
      "grad_norm": 0.5055498480796814,
      "learning_rate": 5.370262278460638e-06,
      "loss": 0.3936,
      "step": 31829
    },
    {
      "epoch": 1.4606947822495526,
      "grad_norm": 0.4865299165248871,
      "learning_rate": 5.370017763309225e-06,
      "loss": 0.3901,
      "step": 31830
    },
    {
      "epoch": 1.460740672754807,
      "grad_norm": 0.4380596876144409,
      "learning_rate": 5.369773247268038e-06,
      "loss": 0.3105,
      "step": 31831
    },
    {
      "epoch": 1.4607865632600614,
      "grad_norm": 0.46879491209983826,
      "learning_rate": 5.369528730337663e-06,
      "loss": 0.3504,
      "step": 31832
    },
    {
      "epoch": 1.460832453765316,
      "grad_norm": 0.4532281160354614,
      "learning_rate": 5.369284212518687e-06,
      "loss": 0.3356,
      "step": 31833
    },
    {
      "epoch": 1.4608783442705704,
      "grad_norm": 0.45346301794052124,
      "learning_rate": 5.3690396938116995e-06,
      "loss": 0.3074,
      "step": 31834
    },
    {
      "epoch": 1.460924234775825,
      "grad_norm": 0.45121511816978455,
      "learning_rate": 5.368795174217288e-06,
      "loss": 0.2993,
      "step": 31835
    },
    {
      "epoch": 1.4609701252810794,
      "grad_norm": 0.46691641211509705,
      "learning_rate": 5.36855065373604e-06,
      "loss": 0.3304,
      "step": 31836
    },
    {
      "epoch": 1.4610160157863339,
      "grad_norm": 0.46297675371170044,
      "learning_rate": 5.368306132368545e-06,
      "loss": 0.3425,
      "step": 31837
    },
    {
      "epoch": 1.4610619062915884,
      "grad_norm": 0.46319469809532166,
      "learning_rate": 5.368061610115389e-06,
      "loss": 0.335,
      "step": 31838
    },
    {
      "epoch": 1.4611077967968427,
      "grad_norm": 0.4951958656311035,
      "learning_rate": 5.367817086977161e-06,
      "loss": 0.3893,
      "step": 31839
    },
    {
      "epoch": 1.4611536873020972,
      "grad_norm": 0.47001132369041443,
      "learning_rate": 5.36757256295445e-06,
      "loss": 0.3771,
      "step": 31840
    },
    {
      "epoch": 1.4611995778073517,
      "grad_norm": 0.45216435194015503,
      "learning_rate": 5.367328038047841e-06,
      "loss": 0.3441,
      "step": 31841
    },
    {
      "epoch": 1.4612454683126062,
      "grad_norm": 0.4867473840713501,
      "learning_rate": 5.367083512257926e-06,
      "loss": 0.3918,
      "step": 31842
    },
    {
      "epoch": 1.4612913588178607,
      "grad_norm": 0.45032796263694763,
      "learning_rate": 5.366838985585291e-06,
      "loss": 0.3408,
      "step": 31843
    },
    {
      "epoch": 1.461337249323115,
      "grad_norm": 0.4190317392349243,
      "learning_rate": 5.366594458030523e-06,
      "loss": 0.2429,
      "step": 31844
    },
    {
      "epoch": 1.4613831398283694,
      "grad_norm": 0.5045154094696045,
      "learning_rate": 5.366349929594212e-06,
      "loss": 0.3539,
      "step": 31845
    },
    {
      "epoch": 1.461429030333624,
      "grad_norm": 0.5018970370292664,
      "learning_rate": 5.3661054002769445e-06,
      "loss": 0.3724,
      "step": 31846
    },
    {
      "epoch": 1.4614749208388784,
      "grad_norm": 0.7125783562660217,
      "learning_rate": 5.365860870079307e-06,
      "loss": 0.4157,
      "step": 31847
    },
    {
      "epoch": 1.461520811344133,
      "grad_norm": 0.4515393376350403,
      "learning_rate": 5.365616339001892e-06,
      "loss": 0.31,
      "step": 31848
    },
    {
      "epoch": 1.4615667018493874,
      "grad_norm": 0.4798808991909027,
      "learning_rate": 5.365371807045285e-06,
      "loss": 0.3497,
      "step": 31849
    },
    {
      "epoch": 1.461612592354642,
      "grad_norm": 0.4762817323207855,
      "learning_rate": 5.365127274210074e-06,
      "loss": 0.3379,
      "step": 31850
    },
    {
      "epoch": 1.4616584828598964,
      "grad_norm": 0.4323571026325226,
      "learning_rate": 5.364882740496847e-06,
      "loss": 0.277,
      "step": 31851
    },
    {
      "epoch": 1.4617043733651507,
      "grad_norm": 0.5357112288475037,
      "learning_rate": 5.364638205906192e-06,
      "loss": 0.4157,
      "step": 31852
    },
    {
      "epoch": 1.4617502638704052,
      "grad_norm": 0.4910373389720917,
      "learning_rate": 5.364393670438697e-06,
      "loss": 0.3624,
      "step": 31853
    },
    {
      "epoch": 1.4617961543756597,
      "grad_norm": 0.45407652854919434,
      "learning_rate": 5.364149134094951e-06,
      "loss": 0.3042,
      "step": 31854
    },
    {
      "epoch": 1.4618420448809142,
      "grad_norm": 0.5146250128746033,
      "learning_rate": 5.363904596875541e-06,
      "loss": 0.3831,
      "step": 31855
    },
    {
      "epoch": 1.4618879353861687,
      "grad_norm": 0.4933314621448517,
      "learning_rate": 5.363660058781054e-06,
      "loss": 0.3884,
      "step": 31856
    },
    {
      "epoch": 1.461933825891423,
      "grad_norm": 0.49251964688301086,
      "learning_rate": 5.36341551981208e-06,
      "loss": 0.332,
      "step": 31857
    },
    {
      "epoch": 1.4619797163966775,
      "grad_norm": 0.4987245798110962,
      "learning_rate": 5.363170979969207e-06,
      "loss": 0.4071,
      "step": 31858
    },
    {
      "epoch": 1.462025606901932,
      "grad_norm": 0.458042711019516,
      "learning_rate": 5.362926439253022e-06,
      "loss": 0.3153,
      "step": 31859
    },
    {
      "epoch": 1.4620714974071864,
      "grad_norm": 0.4696270823478699,
      "learning_rate": 5.362681897664115e-06,
      "loss": 0.3869,
      "step": 31860
    },
    {
      "epoch": 1.462117387912441,
      "grad_norm": 0.4788892865180969,
      "learning_rate": 5.362437355203071e-06,
      "loss": 0.3523,
      "step": 31861
    },
    {
      "epoch": 1.4621632784176954,
      "grad_norm": 0.43843215703964233,
      "learning_rate": 5.362192811870479e-06,
      "loss": 0.3214,
      "step": 31862
    },
    {
      "epoch": 1.46220916892295,
      "grad_norm": 0.4562893211841583,
      "learning_rate": 5.3619482676669285e-06,
      "loss": 0.3109,
      "step": 31863
    },
    {
      "epoch": 1.4622550594282044,
      "grad_norm": 0.45047619938850403,
      "learning_rate": 5.361703722593006e-06,
      "loss": 0.3431,
      "step": 31864
    },
    {
      "epoch": 1.4623009499334587,
      "grad_norm": 0.49364861845970154,
      "learning_rate": 5.361459176649302e-06,
      "loss": 0.4432,
      "step": 31865
    },
    {
      "epoch": 1.4623468404387132,
      "grad_norm": 0.46983715891838074,
      "learning_rate": 5.361214629836401e-06,
      "loss": 0.3794,
      "step": 31866
    },
    {
      "epoch": 1.4623927309439677,
      "grad_norm": 0.44851234555244446,
      "learning_rate": 5.360970082154892e-06,
      "loss": 0.3093,
      "step": 31867
    },
    {
      "epoch": 1.4624386214492222,
      "grad_norm": 0.45752227306365967,
      "learning_rate": 5.360725533605366e-06,
      "loss": 0.3596,
      "step": 31868
    },
    {
      "epoch": 1.4624845119544765,
      "grad_norm": 0.47107943892478943,
      "learning_rate": 5.3604809841884075e-06,
      "loss": 0.3419,
      "step": 31869
    },
    {
      "epoch": 1.462530402459731,
      "grad_norm": 0.45457693934440613,
      "learning_rate": 5.360236433904606e-06,
      "loss": 0.3498,
      "step": 31870
    },
    {
      "epoch": 1.4625762929649855,
      "grad_norm": 0.4456236958503723,
      "learning_rate": 5.359991882754552e-06,
      "loss": 0.3447,
      "step": 31871
    },
    {
      "epoch": 1.46262218347024,
      "grad_norm": 0.49727508425712585,
      "learning_rate": 5.359747330738828e-06,
      "loss": 0.3716,
      "step": 31872
    },
    {
      "epoch": 1.4626680739754945,
      "grad_norm": 0.4723261594772339,
      "learning_rate": 5.359502777858025e-06,
      "loss": 0.3799,
      "step": 31873
    },
    {
      "epoch": 1.462713964480749,
      "grad_norm": 0.48986056447029114,
      "learning_rate": 5.359258224112735e-06,
      "loss": 0.351,
      "step": 31874
    },
    {
      "epoch": 1.4627598549860035,
      "grad_norm": 0.49697962403297424,
      "learning_rate": 5.359013669503539e-06,
      "loss": 0.3706,
      "step": 31875
    },
    {
      "epoch": 1.462805745491258,
      "grad_norm": 0.4182332456111908,
      "learning_rate": 5.358769114031029e-06,
      "loss": 0.279,
      "step": 31876
    },
    {
      "epoch": 1.4628516359965122,
      "grad_norm": 0.4793785810470581,
      "learning_rate": 5.358524557695795e-06,
      "loss": 0.3492,
      "step": 31877
    },
    {
      "epoch": 1.4628975265017667,
      "grad_norm": 0.5213657021522522,
      "learning_rate": 5.358280000498419e-06,
      "loss": 0.419,
      "step": 31878
    },
    {
      "epoch": 1.4629434170070212,
      "grad_norm": 0.4321836531162262,
      "learning_rate": 5.3580354424394956e-06,
      "loss": 0.2958,
      "step": 31879
    },
    {
      "epoch": 1.4629893075122757,
      "grad_norm": 0.45759573578834534,
      "learning_rate": 5.357790883519608e-06,
      "loss": 0.3048,
      "step": 31880
    },
    {
      "epoch": 1.4630351980175302,
      "grad_norm": 0.5028917789459229,
      "learning_rate": 5.357546323739347e-06,
      "loss": 0.4082,
      "step": 31881
    },
    {
      "epoch": 1.4630810885227845,
      "grad_norm": 0.5087604522705078,
      "learning_rate": 5.357301763099301e-06,
      "loss": 0.451,
      "step": 31882
    },
    {
      "epoch": 1.463126979028039,
      "grad_norm": 0.47791752219200134,
      "learning_rate": 5.357057201600054e-06,
      "loss": 0.3806,
      "step": 31883
    },
    {
      "epoch": 1.4631728695332935,
      "grad_norm": 0.47763291001319885,
      "learning_rate": 5.3568126392422e-06,
      "loss": 0.3599,
      "step": 31884
    },
    {
      "epoch": 1.463218760038548,
      "grad_norm": 0.46538180112838745,
      "learning_rate": 5.356568076026323e-06,
      "loss": 0.3705,
      "step": 31885
    },
    {
      "epoch": 1.4632646505438025,
      "grad_norm": 0.45281895995140076,
      "learning_rate": 5.356323511953012e-06,
      "loss": 0.3423,
      "step": 31886
    },
    {
      "epoch": 1.463310541049057,
      "grad_norm": 0.49143868684768677,
      "learning_rate": 5.356078947022856e-06,
      "loss": 0.414,
      "step": 31887
    },
    {
      "epoch": 1.4633564315543115,
      "grad_norm": 0.5179584622383118,
      "learning_rate": 5.355834381236443e-06,
      "loss": 0.4076,
      "step": 31888
    },
    {
      "epoch": 1.463402322059566,
      "grad_norm": 0.4880094826221466,
      "learning_rate": 5.355589814594359e-06,
      "loss": 0.3787,
      "step": 31889
    },
    {
      "epoch": 1.4634482125648203,
      "grad_norm": 0.4866298735141754,
      "learning_rate": 5.355345247097194e-06,
      "loss": 0.3991,
      "step": 31890
    },
    {
      "epoch": 1.4634941030700748,
      "grad_norm": 0.4958963990211487,
      "learning_rate": 5.355100678745539e-06,
      "loss": 0.4125,
      "step": 31891
    },
    {
      "epoch": 1.4635399935753293,
      "grad_norm": 0.4967672526836395,
      "learning_rate": 5.354856109539977e-06,
      "loss": 0.3868,
      "step": 31892
    },
    {
      "epoch": 1.4635858840805838,
      "grad_norm": 0.4800032377243042,
      "learning_rate": 5.354611539481096e-06,
      "loss": 0.3916,
      "step": 31893
    },
    {
      "epoch": 1.4636317745858383,
      "grad_norm": 0.48168689012527466,
      "learning_rate": 5.354366968569489e-06,
      "loss": 0.3314,
      "step": 31894
    },
    {
      "epoch": 1.4636776650910925,
      "grad_norm": 0.456110417842865,
      "learning_rate": 5.354122396805739e-06,
      "loss": 0.342,
      "step": 31895
    },
    {
      "epoch": 1.463723555596347,
      "grad_norm": 0.48731985688209534,
      "learning_rate": 5.353877824190437e-06,
      "loss": 0.3985,
      "step": 31896
    },
    {
      "epoch": 1.4637694461016015,
      "grad_norm": 0.4670856297016144,
      "learning_rate": 5.353633250724171e-06,
      "loss": 0.3715,
      "step": 31897
    },
    {
      "epoch": 1.463815336606856,
      "grad_norm": 0.4790845513343811,
      "learning_rate": 5.353388676407529e-06,
      "loss": 0.3274,
      "step": 31898
    },
    {
      "epoch": 1.4638612271121105,
      "grad_norm": 0.4402669072151184,
      "learning_rate": 5.353144101241099e-06,
      "loss": 0.2907,
      "step": 31899
    },
    {
      "epoch": 1.463907117617365,
      "grad_norm": 0.49588459730148315,
      "learning_rate": 5.352899525225468e-06,
      "loss": 0.4017,
      "step": 31900
    },
    {
      "epoch": 1.4639530081226195,
      "grad_norm": 0.43539783358573914,
      "learning_rate": 5.352654948361224e-06,
      "loss": 0.2993,
      "step": 31901
    },
    {
      "epoch": 1.463998898627874,
      "grad_norm": 0.47025546431541443,
      "learning_rate": 5.3524103706489584e-06,
      "loss": 0.3658,
      "step": 31902
    },
    {
      "epoch": 1.4640447891331283,
      "grad_norm": 0.5211172103881836,
      "learning_rate": 5.352165792089257e-06,
      "loss": 0.4068,
      "step": 31903
    },
    {
      "epoch": 1.4640906796383828,
      "grad_norm": 0.44131729006767273,
      "learning_rate": 5.351921212682706e-06,
      "loss": 0.2993,
      "step": 31904
    },
    {
      "epoch": 1.4641365701436373,
      "grad_norm": 0.4939483106136322,
      "learning_rate": 5.351676632429897e-06,
      "loss": 0.4259,
      "step": 31905
    },
    {
      "epoch": 1.4641824606488918,
      "grad_norm": 0.48727333545684814,
      "learning_rate": 5.3514320513314165e-06,
      "loss": 0.3981,
      "step": 31906
    },
    {
      "epoch": 1.4642283511541463,
      "grad_norm": 0.42784011363983154,
      "learning_rate": 5.351187469387852e-06,
      "loss": 0.282,
      "step": 31907
    },
    {
      "epoch": 1.4642742416594006,
      "grad_norm": 0.46634697914123535,
      "learning_rate": 5.350942886599794e-06,
      "loss": 0.3432,
      "step": 31908
    },
    {
      "epoch": 1.464320132164655,
      "grad_norm": 0.49110129475593567,
      "learning_rate": 5.350698302967829e-06,
      "loss": 0.3756,
      "step": 31909
    },
    {
      "epoch": 1.4643660226699096,
      "grad_norm": 0.47671616077423096,
      "learning_rate": 5.3504537184925445e-06,
      "loss": 0.3842,
      "step": 31910
    },
    {
      "epoch": 1.464411913175164,
      "grad_norm": 0.49569159746170044,
      "learning_rate": 5.350209133174529e-06,
      "loss": 0.4218,
      "step": 31911
    },
    {
      "epoch": 1.4644578036804186,
      "grad_norm": 0.5032116770744324,
      "learning_rate": 5.3499645470143715e-06,
      "loss": 0.3871,
      "step": 31912
    },
    {
      "epoch": 1.464503694185673,
      "grad_norm": 0.4802033007144928,
      "learning_rate": 5.349719960012659e-06,
      "loss": 0.3735,
      "step": 31913
    },
    {
      "epoch": 1.4645495846909276,
      "grad_norm": 0.5009698867797852,
      "learning_rate": 5.349475372169982e-06,
      "loss": 0.4292,
      "step": 31914
    },
    {
      "epoch": 1.4645954751961818,
      "grad_norm": 0.47545239329338074,
      "learning_rate": 5.349230783486926e-06,
      "loss": 0.3496,
      "step": 31915
    },
    {
      "epoch": 1.4646413657014363,
      "grad_norm": 0.4688998758792877,
      "learning_rate": 5.348986193964081e-06,
      "loss": 0.378,
      "step": 31916
    },
    {
      "epoch": 1.4646872562066908,
      "grad_norm": 0.4508756101131439,
      "learning_rate": 5.348741603602033e-06,
      "loss": 0.3678,
      "step": 31917
    },
    {
      "epoch": 1.4647331467119453,
      "grad_norm": 0.5321968793869019,
      "learning_rate": 5.348497012401371e-06,
      "loss": 0.4689,
      "step": 31918
    },
    {
      "epoch": 1.4647790372171998,
      "grad_norm": 0.46053096652030945,
      "learning_rate": 5.348252420362686e-06,
      "loss": 0.3355,
      "step": 31919
    },
    {
      "epoch": 1.464824927722454,
      "grad_norm": 0.4595707654953003,
      "learning_rate": 5.348007827486563e-06,
      "loss": 0.3379,
      "step": 31920
    },
    {
      "epoch": 1.4648708182277086,
      "grad_norm": 0.6103337407112122,
      "learning_rate": 5.347763233773591e-06,
      "loss": 0.3984,
      "step": 31921
    },
    {
      "epoch": 1.464916708732963,
      "grad_norm": 0.4572864770889282,
      "learning_rate": 5.347518639224358e-06,
      "loss": 0.379,
      "step": 31922
    },
    {
      "epoch": 1.4649625992382176,
      "grad_norm": 0.582602858543396,
      "learning_rate": 5.347274043839451e-06,
      "loss": 0.4006,
      "step": 31923
    },
    {
      "epoch": 1.465008489743472,
      "grad_norm": 0.46633827686309814,
      "learning_rate": 5.347029447619462e-06,
      "loss": 0.3706,
      "step": 31924
    },
    {
      "epoch": 1.4650543802487266,
      "grad_norm": 0.48213115334510803,
      "learning_rate": 5.346784850564975e-06,
      "loss": 0.3948,
      "step": 31925
    },
    {
      "epoch": 1.465100270753981,
      "grad_norm": 0.5211972594261169,
      "learning_rate": 5.3465402526765806e-06,
      "loss": 0.3227,
      "step": 31926
    },
    {
      "epoch": 1.4651461612592356,
      "grad_norm": 0.41819635033607483,
      "learning_rate": 5.346295653954866e-06,
      "loss": 0.2801,
      "step": 31927
    },
    {
      "epoch": 1.4651920517644899,
      "grad_norm": 0.45224785804748535,
      "learning_rate": 5.34605105440042e-06,
      "loss": 0.3536,
      "step": 31928
    },
    {
      "epoch": 1.4652379422697444,
      "grad_norm": 0.4287731647491455,
      "learning_rate": 5.345806454013829e-06,
      "loss": 0.3155,
      "step": 31929
    },
    {
      "epoch": 1.4652838327749989,
      "grad_norm": 0.46453845500946045,
      "learning_rate": 5.345561852795684e-06,
      "loss": 0.3188,
      "step": 31930
    },
    {
      "epoch": 1.4653297232802533,
      "grad_norm": 0.5049861073493958,
      "learning_rate": 5.345317250746572e-06,
      "loss": 0.427,
      "step": 31931
    },
    {
      "epoch": 1.4653756137855078,
      "grad_norm": 0.44409817457199097,
      "learning_rate": 5.34507264786708e-06,
      "loss": 0.3505,
      "step": 31932
    },
    {
      "epoch": 1.4654215042907621,
      "grad_norm": 0.4532390236854553,
      "learning_rate": 5.3448280441577986e-06,
      "loss": 0.3195,
      "step": 31933
    },
    {
      "epoch": 1.4654673947960166,
      "grad_norm": 0.4819427728652954,
      "learning_rate": 5.3445834396193134e-06,
      "loss": 0.4109,
      "step": 31934
    },
    {
      "epoch": 1.4655132853012711,
      "grad_norm": 0.4371897876262665,
      "learning_rate": 5.344338834252214e-06,
      "loss": 0.3078,
      "step": 31935
    },
    {
      "epoch": 1.4655591758065256,
      "grad_norm": 0.45668089389801025,
      "learning_rate": 5.344094228057089e-06,
      "loss": 0.3386,
      "step": 31936
    },
    {
      "epoch": 1.4656050663117801,
      "grad_norm": 0.4692455530166626,
      "learning_rate": 5.343849621034525e-06,
      "loss": 0.3192,
      "step": 31937
    },
    {
      "epoch": 1.4656509568170346,
      "grad_norm": 0.47319069504737854,
      "learning_rate": 5.343605013185112e-06,
      "loss": 0.3565,
      "step": 31938
    },
    {
      "epoch": 1.465696847322289,
      "grad_norm": 0.4688185751438141,
      "learning_rate": 5.343360404509436e-06,
      "loss": 0.3426,
      "step": 31939
    },
    {
      "epoch": 1.4657427378275436,
      "grad_norm": 0.4668067693710327,
      "learning_rate": 5.343115795008088e-06,
      "loss": 0.3926,
      "step": 31940
    },
    {
      "epoch": 1.4657886283327979,
      "grad_norm": 0.46265363693237305,
      "learning_rate": 5.342871184681655e-06,
      "loss": 0.3471,
      "step": 31941
    },
    {
      "epoch": 1.4658345188380524,
      "grad_norm": 0.4943835735321045,
      "learning_rate": 5.342626573530725e-06,
      "loss": 0.3139,
      "step": 31942
    },
    {
      "epoch": 1.4658804093433069,
      "grad_norm": 0.49796488881111145,
      "learning_rate": 5.342381961555885e-06,
      "loss": 0.3771,
      "step": 31943
    },
    {
      "epoch": 1.4659262998485614,
      "grad_norm": 0.46749645471572876,
      "learning_rate": 5.342137348757726e-06,
      "loss": 0.3448,
      "step": 31944
    },
    {
      "epoch": 1.4659721903538159,
      "grad_norm": 0.49825426936149597,
      "learning_rate": 5.341892735136833e-06,
      "loss": 0.4084,
      "step": 31945
    },
    {
      "epoch": 1.4660180808590701,
      "grad_norm": 0.4281926453113556,
      "learning_rate": 5.341648120693796e-06,
      "loss": 0.2861,
      "step": 31946
    },
    {
      "epoch": 1.4660639713643246,
      "grad_norm": 0.4827999174594879,
      "learning_rate": 5.341403505429206e-06,
      "loss": 0.2663,
      "step": 31947
    },
    {
      "epoch": 1.4661098618695791,
      "grad_norm": 0.5395724177360535,
      "learning_rate": 5.341158889343645e-06,
      "loss": 0.4527,
      "step": 31948
    },
    {
      "epoch": 1.4661557523748336,
      "grad_norm": 0.472086101770401,
      "learning_rate": 5.340914272437705e-06,
      "loss": 0.3765,
      "step": 31949
    },
    {
      "epoch": 1.4662016428800881,
      "grad_norm": 0.4563523828983307,
      "learning_rate": 5.340669654711974e-06,
      "loss": 0.35,
      "step": 31950
    },
    {
      "epoch": 1.4662475333853426,
      "grad_norm": 0.4705981910228729,
      "learning_rate": 5.34042503616704e-06,
      "loss": 0.3167,
      "step": 31951
    },
    {
      "epoch": 1.4662934238905971,
      "grad_norm": 0.4338723421096802,
      "learning_rate": 5.340180416803492e-06,
      "loss": 0.3037,
      "step": 31952
    },
    {
      "epoch": 1.4663393143958516,
      "grad_norm": 0.47680768370628357,
      "learning_rate": 5.339935796621918e-06,
      "loss": 0.3752,
      "step": 31953
    },
    {
      "epoch": 1.466385204901106,
      "grad_norm": 0.4851015508174896,
      "learning_rate": 5.339691175622904e-06,
      "loss": 0.3885,
      "step": 31954
    },
    {
      "epoch": 1.4664310954063604,
      "grad_norm": 0.45983853936195374,
      "learning_rate": 5.339446553807041e-06,
      "loss": 0.3699,
      "step": 31955
    },
    {
      "epoch": 1.466476985911615,
      "grad_norm": 0.5084765553474426,
      "learning_rate": 5.339201931174916e-06,
      "loss": 0.3832,
      "step": 31956
    },
    {
      "epoch": 1.4665228764168694,
      "grad_norm": 0.5049951076507568,
      "learning_rate": 5.338957307727117e-06,
      "loss": 0.4217,
      "step": 31957
    },
    {
      "epoch": 1.4665687669221237,
      "grad_norm": 0.4427330195903778,
      "learning_rate": 5.338712683464233e-06,
      "loss": 0.3357,
      "step": 31958
    },
    {
      "epoch": 1.4666146574273782,
      "grad_norm": 0.5348069667816162,
      "learning_rate": 5.338468058386852e-06,
      "loss": 0.4284,
      "step": 31959
    },
    {
      "epoch": 1.4666605479326327,
      "grad_norm": 0.4667181074619293,
      "learning_rate": 5.338223432495562e-06,
      "loss": 0.3662,
      "step": 31960
    },
    {
      "epoch": 1.4667064384378872,
      "grad_norm": 0.49890953302383423,
      "learning_rate": 5.337978805790951e-06,
      "loss": 0.4072,
      "step": 31961
    },
    {
      "epoch": 1.4667523289431417,
      "grad_norm": 0.4647672772407532,
      "learning_rate": 5.337734178273608e-06,
      "loss": 0.4058,
      "step": 31962
    },
    {
      "epoch": 1.4667982194483962,
      "grad_norm": 0.46787896752357483,
      "learning_rate": 5.337489549944122e-06,
      "loss": 0.3638,
      "step": 31963
    },
    {
      "epoch": 1.4668441099536507,
      "grad_norm": 0.5214434266090393,
      "learning_rate": 5.3372449208030785e-06,
      "loss": 0.4037,
      "step": 31964
    },
    {
      "epoch": 1.4668900004589052,
      "grad_norm": 0.4922039806842804,
      "learning_rate": 5.337000290851067e-06,
      "loss": 0.3502,
      "step": 31965
    },
    {
      "epoch": 1.4669358909641594,
      "grad_norm": 0.43606099486351013,
      "learning_rate": 5.3367556600886775e-06,
      "loss": 0.3256,
      "step": 31966
    },
    {
      "epoch": 1.466981781469414,
      "grad_norm": 0.4347285032272339,
      "learning_rate": 5.336511028516496e-06,
      "loss": 0.3331,
      "step": 31967
    },
    {
      "epoch": 1.4670276719746684,
      "grad_norm": 0.5161684155464172,
      "learning_rate": 5.336266396135113e-06,
      "loss": 0.4209,
      "step": 31968
    },
    {
      "epoch": 1.467073562479923,
      "grad_norm": 0.4454857409000397,
      "learning_rate": 5.336021762945115e-06,
      "loss": 0.3337,
      "step": 31969
    },
    {
      "epoch": 1.4671194529851774,
      "grad_norm": 0.4700000286102295,
      "learning_rate": 5.33577712894709e-06,
      "loss": 0.351,
      "step": 31970
    },
    {
      "epoch": 1.4671653434904317,
      "grad_norm": 0.465406209230423,
      "learning_rate": 5.335532494141628e-06,
      "loss": 0.3446,
      "step": 31971
    },
    {
      "epoch": 1.4672112339956862,
      "grad_norm": 0.5046120882034302,
      "learning_rate": 5.3352878585293145e-06,
      "loss": 0.4171,
      "step": 31972
    },
    {
      "epoch": 1.4672571245009407,
      "grad_norm": 0.49165287613868713,
      "learning_rate": 5.33504322211074e-06,
      "loss": 0.3787,
      "step": 31973
    },
    {
      "epoch": 1.4673030150061952,
      "grad_norm": 0.45873063802719116,
      "learning_rate": 5.334798584886495e-06,
      "loss": 0.3318,
      "step": 31974
    },
    {
      "epoch": 1.4673489055114497,
      "grad_norm": 0.4731920063495636,
      "learning_rate": 5.334553946857162e-06,
      "loss": 0.3932,
      "step": 31975
    },
    {
      "epoch": 1.4673947960167042,
      "grad_norm": 0.4208911657333374,
      "learning_rate": 5.3343093080233334e-06,
      "loss": 0.3071,
      "step": 31976
    },
    {
      "epoch": 1.4674406865219587,
      "grad_norm": 0.45921406149864197,
      "learning_rate": 5.334064668385595e-06,
      "loss": 0.3685,
      "step": 31977
    },
    {
      "epoch": 1.4674865770272132,
      "grad_norm": 0.52915358543396,
      "learning_rate": 5.3338200279445396e-06,
      "loss": 0.3357,
      "step": 31978
    },
    {
      "epoch": 1.4675324675324675,
      "grad_norm": 0.4377219080924988,
      "learning_rate": 5.3335753867007516e-06,
      "loss": 0.3031,
      "step": 31979
    },
    {
      "epoch": 1.467578358037722,
      "grad_norm": 0.5103516578674316,
      "learning_rate": 5.333330744654817e-06,
      "loss": 0.3654,
      "step": 31980
    },
    {
      "epoch": 1.4676242485429765,
      "grad_norm": 0.4653784930706024,
      "learning_rate": 5.333086101807331e-06,
      "loss": 0.3675,
      "step": 31981
    },
    {
      "epoch": 1.467670139048231,
      "grad_norm": 0.5007166266441345,
      "learning_rate": 5.332841458158875e-06,
      "loss": 0.4159,
      "step": 31982
    },
    {
      "epoch": 1.4677160295534855,
      "grad_norm": 0.5113239884376526,
      "learning_rate": 5.332596813710041e-06,
      "loss": 0.3993,
      "step": 31983
    },
    {
      "epoch": 1.4677619200587397,
      "grad_norm": 0.4820045828819275,
      "learning_rate": 5.332352168461418e-06,
      "loss": 0.3886,
      "step": 31984
    },
    {
      "epoch": 1.4678078105639942,
      "grad_norm": 0.48079735040664673,
      "learning_rate": 5.332107522413594e-06,
      "loss": 0.3641,
      "step": 31985
    },
    {
      "epoch": 1.4678537010692487,
      "grad_norm": 0.4428570568561554,
      "learning_rate": 5.331862875567153e-06,
      "loss": 0.2904,
      "step": 31986
    },
    {
      "epoch": 1.4678995915745032,
      "grad_norm": 0.4756988286972046,
      "learning_rate": 5.331618227922688e-06,
      "loss": 0.3731,
      "step": 31987
    },
    {
      "epoch": 1.4679454820797577,
      "grad_norm": 0.458570271730423,
      "learning_rate": 5.3313735794807855e-06,
      "loss": 0.3698,
      "step": 31988
    },
    {
      "epoch": 1.4679913725850122,
      "grad_norm": 0.4607965052127838,
      "learning_rate": 5.331128930242036e-06,
      "loss": 0.3521,
      "step": 31989
    },
    {
      "epoch": 1.4680372630902667,
      "grad_norm": 0.4442620277404785,
      "learning_rate": 5.330884280207024e-06,
      "loss": 0.3381,
      "step": 31990
    },
    {
      "epoch": 1.4680831535955212,
      "grad_norm": 0.5599028468132019,
      "learning_rate": 5.330639629376339e-06,
      "loss": 0.4966,
      "step": 31991
    },
    {
      "epoch": 1.4681290441007755,
      "grad_norm": 0.5482249855995178,
      "learning_rate": 5.3303949777505725e-06,
      "loss": 0.386,
      "step": 31992
    },
    {
      "epoch": 1.46817493460603,
      "grad_norm": 0.5110315680503845,
      "learning_rate": 5.330150325330308e-06,
      "loss": 0.4475,
      "step": 31993
    },
    {
      "epoch": 1.4682208251112845,
      "grad_norm": 0.4702593684196472,
      "learning_rate": 5.3299056721161365e-06,
      "loss": 0.3766,
      "step": 31994
    },
    {
      "epoch": 1.468266715616539,
      "grad_norm": 0.41059210896492004,
      "learning_rate": 5.329661018108647e-06,
      "loss": 0.3244,
      "step": 31995
    },
    {
      "epoch": 1.4683126061217933,
      "grad_norm": 0.4426000714302063,
      "learning_rate": 5.329416363308427e-06,
      "loss": 0.3486,
      "step": 31996
    },
    {
      "epoch": 1.4683584966270478,
      "grad_norm": 0.44065266847610474,
      "learning_rate": 5.3291717077160636e-06,
      "loss": 0.3589,
      "step": 31997
    },
    {
      "epoch": 1.4684043871323023,
      "grad_norm": 0.4203678071498871,
      "learning_rate": 5.328927051332147e-06,
      "loss": 0.2945,
      "step": 31998
    },
    {
      "epoch": 1.4684502776375568,
      "grad_norm": 0.43769046664237976,
      "learning_rate": 5.328682394157264e-06,
      "loss": 0.2856,
      "step": 31999
    },
    {
      "epoch": 1.4684961681428113,
      "grad_norm": 0.4644850194454193,
      "learning_rate": 5.328437736192004e-06,
      "loss": 0.3368,
      "step": 32000
    },
    {
      "epoch": 1.4685420586480658,
      "grad_norm": 0.46735361218452454,
      "learning_rate": 5.328193077436955e-06,
      "loss": 0.3206,
      "step": 32001
    },
    {
      "epoch": 1.4685879491533202,
      "grad_norm": 0.5124819278717041,
      "learning_rate": 5.327948417892704e-06,
      "loss": 0.4113,
      "step": 32002
    },
    {
      "epoch": 1.4686338396585747,
      "grad_norm": 0.4593260586261749,
      "learning_rate": 5.327703757559841e-06,
      "loss": 0.3472,
      "step": 32003
    },
    {
      "epoch": 1.468679730163829,
      "grad_norm": 0.45220068097114563,
      "learning_rate": 5.327459096438955e-06,
      "loss": 0.3432,
      "step": 32004
    },
    {
      "epoch": 1.4687256206690835,
      "grad_norm": 0.47802838683128357,
      "learning_rate": 5.327214434530632e-06,
      "loss": 0.3327,
      "step": 32005
    },
    {
      "epoch": 1.468771511174338,
      "grad_norm": 0.4243192970752716,
      "learning_rate": 5.326969771835463e-06,
      "loss": 0.3045,
      "step": 32006
    },
    {
      "epoch": 1.4688174016795925,
      "grad_norm": 0.45395082235336304,
      "learning_rate": 5.326725108354035e-06,
      "loss": 0.3166,
      "step": 32007
    },
    {
      "epoch": 1.468863292184847,
      "grad_norm": 0.46591436862945557,
      "learning_rate": 5.326480444086935e-06,
      "loss": 0.3331,
      "step": 32008
    },
    {
      "epoch": 1.4689091826901013,
      "grad_norm": 0.5051455497741699,
      "learning_rate": 5.326235779034753e-06,
      "loss": 0.4232,
      "step": 32009
    },
    {
      "epoch": 1.4689550731953558,
      "grad_norm": 0.5273759365081787,
      "learning_rate": 5.325991113198076e-06,
      "loss": 0.44,
      "step": 32010
    },
    {
      "epoch": 1.4690009637006103,
      "grad_norm": 0.48473554849624634,
      "learning_rate": 5.325746446577494e-06,
      "loss": 0.3928,
      "step": 32011
    },
    {
      "epoch": 1.4690468542058648,
      "grad_norm": 0.49563896656036377,
      "learning_rate": 5.325501779173597e-06,
      "loss": 0.3685,
      "step": 32012
    },
    {
      "epoch": 1.4690927447111193,
      "grad_norm": 0.41852867603302,
      "learning_rate": 5.325257110986966e-06,
      "loss": 0.2782,
      "step": 32013
    },
    {
      "epoch": 1.4691386352163738,
      "grad_norm": 0.4839380085468292,
      "learning_rate": 5.325012442018197e-06,
      "loss": 0.3628,
      "step": 32014
    },
    {
      "epoch": 1.4691845257216283,
      "grad_norm": 0.44317543506622314,
      "learning_rate": 5.324767772267876e-06,
      "loss": 0.305,
      "step": 32015
    },
    {
      "epoch": 1.4692304162268828,
      "grad_norm": 0.4602982699871063,
      "learning_rate": 5.32452310173659e-06,
      "loss": 0.3765,
      "step": 32016
    },
    {
      "epoch": 1.469276306732137,
      "grad_norm": 0.447652131319046,
      "learning_rate": 5.324278430424929e-06,
      "loss": 0.2998,
      "step": 32017
    },
    {
      "epoch": 1.4693221972373915,
      "grad_norm": 0.4808535873889923,
      "learning_rate": 5.324033758333482e-06,
      "loss": 0.4026,
      "step": 32018
    },
    {
      "epoch": 1.469368087742646,
      "grad_norm": 0.4307735860347748,
      "learning_rate": 5.323789085462834e-06,
      "loss": 0.2991,
      "step": 32019
    },
    {
      "epoch": 1.4694139782479005,
      "grad_norm": 0.4704866111278534,
      "learning_rate": 5.323544411813576e-06,
      "loss": 0.3274,
      "step": 32020
    },
    {
      "epoch": 1.469459868753155,
      "grad_norm": 0.43960535526275635,
      "learning_rate": 5.323299737386296e-06,
      "loss": 0.3541,
      "step": 32021
    },
    {
      "epoch": 1.4695057592584093,
      "grad_norm": 0.4882011115550995,
      "learning_rate": 5.3230550621815824e-06,
      "loss": 0.3884,
      "step": 32022
    },
    {
      "epoch": 1.4695516497636638,
      "grad_norm": 0.46917080879211426,
      "learning_rate": 5.322810386200022e-06,
      "loss": 0.4322,
      "step": 32023
    },
    {
      "epoch": 1.4695975402689183,
      "grad_norm": 0.4688405394554138,
      "learning_rate": 5.322565709442207e-06,
      "loss": 0.3787,
      "step": 32024
    },
    {
      "epoch": 1.4696434307741728,
      "grad_norm": 0.4486179053783417,
      "learning_rate": 5.322321031908721e-06,
      "loss": 0.3274,
      "step": 32025
    },
    {
      "epoch": 1.4696893212794273,
      "grad_norm": 0.49341434240341187,
      "learning_rate": 5.322076353600156e-06,
      "loss": 0.3607,
      "step": 32026
    },
    {
      "epoch": 1.4697352117846818,
      "grad_norm": 0.4980112910270691,
      "learning_rate": 5.321831674517098e-06,
      "loss": 0.3966,
      "step": 32027
    },
    {
      "epoch": 1.4697811022899363,
      "grad_norm": 0.4310759902000427,
      "learning_rate": 5.321586994660137e-06,
      "loss": 0.2847,
      "step": 32028
    },
    {
      "epoch": 1.4698269927951908,
      "grad_norm": 0.4371199607849121,
      "learning_rate": 5.3213423140298615e-06,
      "loss": 0.3049,
      "step": 32029
    },
    {
      "epoch": 1.469872883300445,
      "grad_norm": 0.5193871259689331,
      "learning_rate": 5.321097632626858e-06,
      "loss": 0.338,
      "step": 32030
    },
    {
      "epoch": 1.4699187738056996,
      "grad_norm": 0.4821956753730774,
      "learning_rate": 5.320852950451716e-06,
      "loss": 0.3678,
      "step": 32031
    },
    {
      "epoch": 1.469964664310954,
      "grad_norm": 0.4550100862979889,
      "learning_rate": 5.320608267505024e-06,
      "loss": 0.3292,
      "step": 32032
    },
    {
      "epoch": 1.4700105548162086,
      "grad_norm": 0.4735218286514282,
      "learning_rate": 5.320363583787371e-06,
      "loss": 0.3515,
      "step": 32033
    },
    {
      "epoch": 1.470056445321463,
      "grad_norm": 0.3999912142753601,
      "learning_rate": 5.320118899299345e-06,
      "loss": 0.2634,
      "step": 32034
    },
    {
      "epoch": 1.4701023358267173,
      "grad_norm": 0.46924370527267456,
      "learning_rate": 5.319874214041533e-06,
      "loss": 0.3743,
      "step": 32035
    },
    {
      "epoch": 1.4701482263319718,
      "grad_norm": 0.5325684547424316,
      "learning_rate": 5.3196295280145235e-06,
      "loss": 0.4135,
      "step": 32036
    },
    {
      "epoch": 1.4701941168372263,
      "grad_norm": 0.4592367112636566,
      "learning_rate": 5.319384841218908e-06,
      "loss": 0.3383,
      "step": 32037
    },
    {
      "epoch": 1.4702400073424808,
      "grad_norm": 0.4844416677951813,
      "learning_rate": 5.319140153655272e-06,
      "loss": 0.3443,
      "step": 32038
    },
    {
      "epoch": 1.4702858978477353,
      "grad_norm": 0.45764854550361633,
      "learning_rate": 5.318895465324205e-06,
      "loss": 0.3134,
      "step": 32039
    },
    {
      "epoch": 1.4703317883529898,
      "grad_norm": 0.5871613025665283,
      "learning_rate": 5.318650776226294e-06,
      "loss": 0.3807,
      "step": 32040
    },
    {
      "epoch": 1.4703776788582443,
      "grad_norm": 0.46543848514556885,
      "learning_rate": 5.31840608636213e-06,
      "loss": 0.3446,
      "step": 32041
    },
    {
      "epoch": 1.4704235693634988,
      "grad_norm": 0.5003018975257874,
      "learning_rate": 5.318161395732299e-06,
      "loss": 0.357,
      "step": 32042
    },
    {
      "epoch": 1.470469459868753,
      "grad_norm": 0.4572242498397827,
      "learning_rate": 5.31791670433739e-06,
      "loss": 0.3339,
      "step": 32043
    },
    {
      "epoch": 1.4705153503740076,
      "grad_norm": 0.43412917852401733,
      "learning_rate": 5.317672012177993e-06,
      "loss": 0.3087,
      "step": 32044
    },
    {
      "epoch": 1.470561240879262,
      "grad_norm": 0.5248821973800659,
      "learning_rate": 5.3174273192546924e-06,
      "loss": 0.405,
      "step": 32045
    },
    {
      "epoch": 1.4706071313845166,
      "grad_norm": 0.4379834234714508,
      "learning_rate": 5.317182625568082e-06,
      "loss": 0.2952,
      "step": 32046
    },
    {
      "epoch": 1.4706530218897709,
      "grad_norm": 0.46651822328567505,
      "learning_rate": 5.316937931118745e-06,
      "loss": 0.3465,
      "step": 32047
    },
    {
      "epoch": 1.4706989123950254,
      "grad_norm": 0.4583413004875183,
      "learning_rate": 5.3166932359072735e-06,
      "loss": 0.3244,
      "step": 32048
    },
    {
      "epoch": 1.4707448029002799,
      "grad_norm": 0.4537734091281891,
      "learning_rate": 5.316448539934256e-06,
      "loss": 0.353,
      "step": 32049
    },
    {
      "epoch": 1.4707906934055344,
      "grad_norm": 0.5175872445106506,
      "learning_rate": 5.316203843200278e-06,
      "loss": 0.4384,
      "step": 32050
    },
    {
      "epoch": 1.4708365839107889,
      "grad_norm": 0.48653343319892883,
      "learning_rate": 5.31595914570593e-06,
      "loss": 0.4267,
      "step": 32051
    },
    {
      "epoch": 1.4708824744160434,
      "grad_norm": 0.4833705425262451,
      "learning_rate": 5.315714447451801e-06,
      "loss": 0.4135,
      "step": 32052
    },
    {
      "epoch": 1.4709283649212979,
      "grad_norm": 0.4492006301879883,
      "learning_rate": 5.315469748438477e-06,
      "loss": 0.3139,
      "step": 32053
    },
    {
      "epoch": 1.4709742554265524,
      "grad_norm": 0.5261073112487793,
      "learning_rate": 5.315225048666547e-06,
      "loss": 0.4262,
      "step": 32054
    },
    {
      "epoch": 1.4710201459318066,
      "grad_norm": 0.4762416183948517,
      "learning_rate": 5.314980348136602e-06,
      "loss": 0.3655,
      "step": 32055
    },
    {
      "epoch": 1.4710660364370611,
      "grad_norm": 0.5034158229827881,
      "learning_rate": 5.3147356468492285e-06,
      "loss": 0.3864,
      "step": 32056
    },
    {
      "epoch": 1.4711119269423156,
      "grad_norm": 0.46422988176345825,
      "learning_rate": 5.314490944805015e-06,
      "loss": 0.3773,
      "step": 32057
    },
    {
      "epoch": 1.4711578174475701,
      "grad_norm": 0.44940030574798584,
      "learning_rate": 5.314246242004549e-06,
      "loss": 0.3538,
      "step": 32058
    },
    {
      "epoch": 1.4712037079528246,
      "grad_norm": 0.5211067199707031,
      "learning_rate": 5.31400153844842e-06,
      "loss": 0.3886,
      "step": 32059
    },
    {
      "epoch": 1.471249598458079,
      "grad_norm": 0.45785287022590637,
      "learning_rate": 5.313756834137218e-06,
      "loss": 0.3139,
      "step": 32060
    },
    {
      "epoch": 1.4712954889633334,
      "grad_norm": 0.47487348318099976,
      "learning_rate": 5.31351212907153e-06,
      "loss": 0.4065,
      "step": 32061
    },
    {
      "epoch": 1.471341379468588,
      "grad_norm": 0.6778917908668518,
      "learning_rate": 5.313267423251942e-06,
      "loss": 0.2664,
      "step": 32062
    },
    {
      "epoch": 1.4713872699738424,
      "grad_norm": 0.4605705738067627,
      "learning_rate": 5.313022716679047e-06,
      "loss": 0.3435,
      "step": 32063
    },
    {
      "epoch": 1.471433160479097,
      "grad_norm": 0.4908944070339203,
      "learning_rate": 5.31277800935343e-06,
      "loss": 0.4296,
      "step": 32064
    },
    {
      "epoch": 1.4714790509843514,
      "grad_norm": 0.45814260840415955,
      "learning_rate": 5.3125333012756795e-06,
      "loss": 0.347,
      "step": 32065
    },
    {
      "epoch": 1.4715249414896059,
      "grad_norm": 0.43747884035110474,
      "learning_rate": 5.312288592446388e-06,
      "loss": 0.3082,
      "step": 32066
    },
    {
      "epoch": 1.4715708319948604,
      "grad_norm": 0.46185818314552307,
      "learning_rate": 5.312043882866139e-06,
      "loss": 0.3405,
      "step": 32067
    },
    {
      "epoch": 1.4716167225001147,
      "grad_norm": 0.46715840697288513,
      "learning_rate": 5.311799172535522e-06,
      "loss": 0.327,
      "step": 32068
    },
    {
      "epoch": 1.4716626130053692,
      "grad_norm": 0.46411746740341187,
      "learning_rate": 5.3115544614551286e-06,
      "loss": 0.3629,
      "step": 32069
    },
    {
      "epoch": 1.4717085035106237,
      "grad_norm": 0.456381231546402,
      "learning_rate": 5.311309749625544e-06,
      "loss": 0.3341,
      "step": 32070
    },
    {
      "epoch": 1.4717543940158782,
      "grad_norm": 0.4469074606895447,
      "learning_rate": 5.3110650370473584e-06,
      "loss": 0.3651,
      "step": 32071
    },
    {
      "epoch": 1.4718002845211327,
      "grad_norm": 0.45176440477371216,
      "learning_rate": 5.310820323721159e-06,
      "loss": 0.3439,
      "step": 32072
    },
    {
      "epoch": 1.471846175026387,
      "grad_norm": 0.47577545046806335,
      "learning_rate": 5.310575609647533e-06,
      "loss": 0.3474,
      "step": 32073
    },
    {
      "epoch": 1.4718920655316414,
      "grad_norm": 0.4674552083015442,
      "learning_rate": 5.310330894827074e-06,
      "loss": 0.3557,
      "step": 32074
    },
    {
      "epoch": 1.471937956036896,
      "grad_norm": 0.4747660160064697,
      "learning_rate": 5.310086179260365e-06,
      "loss": 0.4241,
      "step": 32075
    },
    {
      "epoch": 1.4719838465421504,
      "grad_norm": 0.45715874433517456,
      "learning_rate": 5.309841462947997e-06,
      "loss": 0.3199,
      "step": 32076
    },
    {
      "epoch": 1.472029737047405,
      "grad_norm": 0.5239123106002808,
      "learning_rate": 5.30959674589056e-06,
      "loss": 0.4113,
      "step": 32077
    },
    {
      "epoch": 1.4720756275526594,
      "grad_norm": 0.4803256094455719,
      "learning_rate": 5.309352028088638e-06,
      "loss": 0.3735,
      "step": 32078
    },
    {
      "epoch": 1.472121518057914,
      "grad_norm": 0.499620258808136,
      "learning_rate": 5.309107309542822e-06,
      "loss": 0.4205,
      "step": 32079
    },
    {
      "epoch": 1.4721674085631684,
      "grad_norm": 0.4901861250400543,
      "learning_rate": 5.308862590253702e-06,
      "loss": 0.3712,
      "step": 32080
    },
    {
      "epoch": 1.4722132990684227,
      "grad_norm": 0.483675479888916,
      "learning_rate": 5.3086178702218635e-06,
      "loss": 0.3499,
      "step": 32081
    },
    {
      "epoch": 1.4722591895736772,
      "grad_norm": 0.4482039511203766,
      "learning_rate": 5.308373149447897e-06,
      "loss": 0.2974,
      "step": 32082
    },
    {
      "epoch": 1.4723050800789317,
      "grad_norm": 0.4688926339149475,
      "learning_rate": 5.308128427932392e-06,
      "loss": 0.3238,
      "step": 32083
    },
    {
      "epoch": 1.4723509705841862,
      "grad_norm": 0.46346724033355713,
      "learning_rate": 5.307883705675932e-06,
      "loss": 0.3899,
      "step": 32084
    },
    {
      "epoch": 1.4723968610894405,
      "grad_norm": 0.47251787781715393,
      "learning_rate": 5.307638982679111e-06,
      "loss": 0.4171,
      "step": 32085
    },
    {
      "epoch": 1.472442751594695,
      "grad_norm": 0.43615666031837463,
      "learning_rate": 5.307394258942515e-06,
      "loss": 0.2724,
      "step": 32086
    },
    {
      "epoch": 1.4724886420999495,
      "grad_norm": 0.4318934381008148,
      "learning_rate": 5.307149534466734e-06,
      "loss": 0.2977,
      "step": 32087
    },
    {
      "epoch": 1.472534532605204,
      "grad_norm": 0.45368871092796326,
      "learning_rate": 5.306904809252355e-06,
      "loss": 0.3191,
      "step": 32088
    },
    {
      "epoch": 1.4725804231104584,
      "grad_norm": 0.49371564388275146,
      "learning_rate": 5.306660083299966e-06,
      "loss": 0.4144,
      "step": 32089
    },
    {
      "epoch": 1.472626313615713,
      "grad_norm": 0.45591971278190613,
      "learning_rate": 5.306415356610155e-06,
      "loss": 0.3639,
      "step": 32090
    },
    {
      "epoch": 1.4726722041209674,
      "grad_norm": 0.45543017983436584,
      "learning_rate": 5.306170629183513e-06,
      "loss": 0.3441,
      "step": 32091
    },
    {
      "epoch": 1.472718094626222,
      "grad_norm": 0.43691474199295044,
      "learning_rate": 5.305925901020628e-06,
      "loss": 0.3233,
      "step": 32092
    },
    {
      "epoch": 1.4727639851314762,
      "grad_norm": 0.4881291687488556,
      "learning_rate": 5.305681172122088e-06,
      "loss": 0.4285,
      "step": 32093
    },
    {
      "epoch": 1.4728098756367307,
      "grad_norm": 0.46096301078796387,
      "learning_rate": 5.30543644248848e-06,
      "loss": 0.3269,
      "step": 32094
    },
    {
      "epoch": 1.4728557661419852,
      "grad_norm": 0.46370741724967957,
      "learning_rate": 5.305191712120395e-06,
      "loss": 0.3652,
      "step": 32095
    },
    {
      "epoch": 1.4729016566472397,
      "grad_norm": 0.598700225353241,
      "learning_rate": 5.304946981018419e-06,
      "loss": 0.3996,
      "step": 32096
    },
    {
      "epoch": 1.4729475471524942,
      "grad_norm": 0.4296024739742279,
      "learning_rate": 5.3047022491831415e-06,
      "loss": 0.3206,
      "step": 32097
    },
    {
      "epoch": 1.4729934376577485,
      "grad_norm": 0.44607773423194885,
      "learning_rate": 5.304457516615151e-06,
      "loss": 0.3374,
      "step": 32098
    },
    {
      "epoch": 1.473039328163003,
      "grad_norm": 0.48436278104782104,
      "learning_rate": 5.304212783315038e-06,
      "loss": 0.4402,
      "step": 32099
    },
    {
      "epoch": 1.4730852186682575,
      "grad_norm": 0.44917213916778564,
      "learning_rate": 5.303968049283389e-06,
      "loss": 0.3144,
      "step": 32100
    },
    {
      "epoch": 1.473131109173512,
      "grad_norm": 0.48056715726852417,
      "learning_rate": 5.3037233145207925e-06,
      "loss": 0.3178,
      "step": 32101
    },
    {
      "epoch": 1.4731769996787665,
      "grad_norm": 0.49721914529800415,
      "learning_rate": 5.303478579027836e-06,
      "loss": 0.4529,
      "step": 32102
    },
    {
      "epoch": 1.473222890184021,
      "grad_norm": 0.46930742263793945,
      "learning_rate": 5.303233842805111e-06,
      "loss": 0.4143,
      "step": 32103
    },
    {
      "epoch": 1.4732687806892755,
      "grad_norm": 0.44961121678352356,
      "learning_rate": 5.302989105853203e-06,
      "loss": 0.3229,
      "step": 32104
    },
    {
      "epoch": 1.47331467119453,
      "grad_norm": 0.4435790777206421,
      "learning_rate": 5.302744368172703e-06,
      "loss": 0.3282,
      "step": 32105
    },
    {
      "epoch": 1.4733605616997842,
      "grad_norm": 0.5323469638824463,
      "learning_rate": 5.302499629764198e-06,
      "loss": 0.4331,
      "step": 32106
    },
    {
      "epoch": 1.4734064522050387,
      "grad_norm": 0.4732191562652588,
      "learning_rate": 5.302254890628275e-06,
      "loss": 0.3269,
      "step": 32107
    },
    {
      "epoch": 1.4734523427102932,
      "grad_norm": 0.4918588399887085,
      "learning_rate": 5.302010150765527e-06,
      "loss": 0.4352,
      "step": 32108
    },
    {
      "epoch": 1.4734982332155477,
      "grad_norm": 0.4445362091064453,
      "learning_rate": 5.301765410176539e-06,
      "loss": 0.3237,
      "step": 32109
    },
    {
      "epoch": 1.4735441237208022,
      "grad_norm": 0.4067014157772064,
      "learning_rate": 5.301520668861902e-06,
      "loss": 0.278,
      "step": 32110
    },
    {
      "epoch": 1.4735900142260565,
      "grad_norm": 0.4547867774963379,
      "learning_rate": 5.3012759268222e-06,
      "loss": 0.3225,
      "step": 32111
    },
    {
      "epoch": 1.473635904731311,
      "grad_norm": 0.44877687096595764,
      "learning_rate": 5.301031184058026e-06,
      "loss": 0.3078,
      "step": 32112
    },
    {
      "epoch": 1.4736817952365655,
      "grad_norm": 0.5194283127784729,
      "learning_rate": 5.300786440569966e-06,
      "loss": 0.4406,
      "step": 32113
    },
    {
      "epoch": 1.47372768574182,
      "grad_norm": 0.5044810175895691,
      "learning_rate": 5.300541696358611e-06,
      "loss": 0.4203,
      "step": 32114
    },
    {
      "epoch": 1.4737735762470745,
      "grad_norm": 0.4562418758869171,
      "learning_rate": 5.3002969514245485e-06,
      "loss": 0.2764,
      "step": 32115
    },
    {
      "epoch": 1.473819466752329,
      "grad_norm": 0.47908076643943787,
      "learning_rate": 5.300052205768364e-06,
      "loss": 0.3461,
      "step": 32116
    },
    {
      "epoch": 1.4738653572575835,
      "grad_norm": 0.4281238615512848,
      "learning_rate": 5.299807459390651e-06,
      "loss": 0.2408,
      "step": 32117
    },
    {
      "epoch": 1.473911247762838,
      "grad_norm": 0.44104140996932983,
      "learning_rate": 5.299562712291995e-06,
      "loss": 0.3247,
      "step": 32118
    },
    {
      "epoch": 1.4739571382680923,
      "grad_norm": 0.46291911602020264,
      "learning_rate": 5.299317964472984e-06,
      "loss": 0.3629,
      "step": 32119
    },
    {
      "epoch": 1.4740030287733468,
      "grad_norm": 0.47003173828125,
      "learning_rate": 5.299073215934209e-06,
      "loss": 0.3418,
      "step": 32120
    },
    {
      "epoch": 1.4740489192786013,
      "grad_norm": 0.46763402223587036,
      "learning_rate": 5.298828466676258e-06,
      "loss": 0.3611,
      "step": 32121
    },
    {
      "epoch": 1.4740948097838558,
      "grad_norm": 0.5117161273956299,
      "learning_rate": 5.298583716699718e-06,
      "loss": 0.4045,
      "step": 32122
    },
    {
      "epoch": 1.4741407002891103,
      "grad_norm": 0.4899384677410126,
      "learning_rate": 5.298338966005179e-06,
      "loss": 0.3735,
      "step": 32123
    },
    {
      "epoch": 1.4741865907943645,
      "grad_norm": 0.48555582761764526,
      "learning_rate": 5.298094214593227e-06,
      "loss": 0.4109,
      "step": 32124
    },
    {
      "epoch": 1.474232481299619,
      "grad_norm": 0.4935511350631714,
      "learning_rate": 5.297849462464455e-06,
      "loss": 0.3251,
      "step": 32125
    },
    {
      "epoch": 1.4742783718048735,
      "grad_norm": 0.48091232776641846,
      "learning_rate": 5.297604709619449e-06,
      "loss": 0.4142,
      "step": 32126
    },
    {
      "epoch": 1.474324262310128,
      "grad_norm": 0.4515378773212433,
      "learning_rate": 5.297359956058795e-06,
      "loss": 0.2981,
      "step": 32127
    },
    {
      "epoch": 1.4743701528153825,
      "grad_norm": 0.4305891990661621,
      "learning_rate": 5.297115201783086e-06,
      "loss": 0.2811,
      "step": 32128
    },
    {
      "epoch": 1.474416043320637,
      "grad_norm": 0.4781722128391266,
      "learning_rate": 5.296870446792907e-06,
      "loss": 0.3764,
      "step": 32129
    },
    {
      "epoch": 1.4744619338258915,
      "grad_norm": 0.5054376721382141,
      "learning_rate": 5.29662569108885e-06,
      "loss": 0.3978,
      "step": 32130
    },
    {
      "epoch": 1.474507824331146,
      "grad_norm": 0.43782874941825867,
      "learning_rate": 5.296380934671503e-06,
      "loss": 0.3156,
      "step": 32131
    },
    {
      "epoch": 1.4745537148364003,
      "grad_norm": 0.47763556241989136,
      "learning_rate": 5.296136177541451e-06,
      "loss": 0.4283,
      "step": 32132
    },
    {
      "epoch": 1.4745996053416548,
      "grad_norm": 0.4743375778198242,
      "learning_rate": 5.295891419699286e-06,
      "loss": 0.3578,
      "step": 32133
    },
    {
      "epoch": 1.4746454958469093,
      "grad_norm": 0.47007471323013306,
      "learning_rate": 5.295646661145596e-06,
      "loss": 0.3753,
      "step": 32134
    },
    {
      "epoch": 1.4746913863521638,
      "grad_norm": 0.4671744406223297,
      "learning_rate": 5.295401901880968e-06,
      "loss": 0.31,
      "step": 32135
    },
    {
      "epoch": 1.474737276857418,
      "grad_norm": 0.4932475984096527,
      "learning_rate": 5.295157141905992e-06,
      "loss": 0.3852,
      "step": 32136
    },
    {
      "epoch": 1.4747831673626726,
      "grad_norm": 0.46660348773002625,
      "learning_rate": 5.294912381221257e-06,
      "loss": 0.345,
      "step": 32137
    },
    {
      "epoch": 1.474829057867927,
      "grad_norm": 0.46788978576660156,
      "learning_rate": 5.2946676198273494e-06,
      "loss": 0.387,
      "step": 32138
    },
    {
      "epoch": 1.4748749483731816,
      "grad_norm": 0.4677716791629791,
      "learning_rate": 5.29442285772486e-06,
      "loss": 0.3948,
      "step": 32139
    },
    {
      "epoch": 1.474920838878436,
      "grad_norm": 0.44886189699172974,
      "learning_rate": 5.294178094914376e-06,
      "loss": 0.3123,
      "step": 32140
    },
    {
      "epoch": 1.4749667293836906,
      "grad_norm": 0.46136191487312317,
      "learning_rate": 5.293933331396487e-06,
      "loss": 0.3558,
      "step": 32141
    },
    {
      "epoch": 1.475012619888945,
      "grad_norm": 0.4687035083770752,
      "learning_rate": 5.293688567171781e-06,
      "loss": 0.3391,
      "step": 32142
    },
    {
      "epoch": 1.4750585103941996,
      "grad_norm": 0.4563540518283844,
      "learning_rate": 5.293443802240847e-06,
      "loss": 0.3216,
      "step": 32143
    },
    {
      "epoch": 1.4751044008994538,
      "grad_norm": 0.4705131947994232,
      "learning_rate": 5.293199036604273e-06,
      "loss": 0.3701,
      "step": 32144
    },
    {
      "epoch": 1.4751502914047083,
      "grad_norm": 0.5243456959724426,
      "learning_rate": 5.292954270262648e-06,
      "loss": 0.4167,
      "step": 32145
    },
    {
      "epoch": 1.4751961819099628,
      "grad_norm": 0.5293105244636536,
      "learning_rate": 5.29270950321656e-06,
      "loss": 0.4437,
      "step": 32146
    },
    {
      "epoch": 1.4752420724152173,
      "grad_norm": 0.44688811898231506,
      "learning_rate": 5.292464735466599e-06,
      "loss": 0.3426,
      "step": 32147
    },
    {
      "epoch": 1.4752879629204718,
      "grad_norm": 0.4468919336795807,
      "learning_rate": 5.292219967013353e-06,
      "loss": 0.3188,
      "step": 32148
    },
    {
      "epoch": 1.475333853425726,
      "grad_norm": 0.4285844564437866,
      "learning_rate": 5.291975197857408e-06,
      "loss": 0.302,
      "step": 32149
    },
    {
      "epoch": 1.4753797439309806,
      "grad_norm": 0.4776487350463867,
      "learning_rate": 5.291730427999355e-06,
      "loss": 0.399,
      "step": 32150
    },
    {
      "epoch": 1.475425634436235,
      "grad_norm": 0.47976982593536377,
      "learning_rate": 5.291485657439785e-06,
      "loss": 0.3555,
      "step": 32151
    },
    {
      "epoch": 1.4754715249414896,
      "grad_norm": 0.43079400062561035,
      "learning_rate": 5.291240886179281e-06,
      "loss": 0.3161,
      "step": 32152
    },
    {
      "epoch": 1.475517415446744,
      "grad_norm": 0.475788950920105,
      "learning_rate": 5.290996114218436e-06,
      "loss": 0.3519,
      "step": 32153
    },
    {
      "epoch": 1.4755633059519986,
      "grad_norm": 0.5033673644065857,
      "learning_rate": 5.290751341557838e-06,
      "loss": 0.4372,
      "step": 32154
    },
    {
      "epoch": 1.475609196457253,
      "grad_norm": 0.4571196734905243,
      "learning_rate": 5.290506568198074e-06,
      "loss": 0.3269,
      "step": 32155
    },
    {
      "epoch": 1.4756550869625076,
      "grad_norm": 0.43819674849510193,
      "learning_rate": 5.2902617941397334e-06,
      "loss": 0.3036,
      "step": 32156
    },
    {
      "epoch": 1.4757009774677619,
      "grad_norm": 0.4568757712841034,
      "learning_rate": 5.2900170193834046e-06,
      "loss": 0.3655,
      "step": 32157
    },
    {
      "epoch": 1.4757468679730164,
      "grad_norm": 0.4275202453136444,
      "learning_rate": 5.289772243929676e-06,
      "loss": 0.2739,
      "step": 32158
    },
    {
      "epoch": 1.4757927584782708,
      "grad_norm": 0.4915981590747833,
      "learning_rate": 5.289527467779139e-06,
      "loss": 0.4168,
      "step": 32159
    },
    {
      "epoch": 1.4758386489835253,
      "grad_norm": 0.4561948776245117,
      "learning_rate": 5.289282690932377e-06,
      "loss": 0.3607,
      "step": 32160
    },
    {
      "epoch": 1.4758845394887798,
      "grad_norm": 0.4902079999446869,
      "learning_rate": 5.289037913389982e-06,
      "loss": 0.3705,
      "step": 32161
    },
    {
      "epoch": 1.4759304299940341,
      "grad_norm": 0.45454102754592896,
      "learning_rate": 5.2887931351525424e-06,
      "loss": 0.2949,
      "step": 32162
    },
    {
      "epoch": 1.4759763204992886,
      "grad_norm": 0.4169740080833435,
      "learning_rate": 5.288548356220645e-06,
      "loss": 0.2931,
      "step": 32163
    },
    {
      "epoch": 1.4760222110045431,
      "grad_norm": 0.45563727617263794,
      "learning_rate": 5.288303576594882e-06,
      "loss": 0.3207,
      "step": 32164
    },
    {
      "epoch": 1.4760681015097976,
      "grad_norm": 0.45957526564598083,
      "learning_rate": 5.28805879627584e-06,
      "loss": 0.3259,
      "step": 32165
    },
    {
      "epoch": 1.4761139920150521,
      "grad_norm": 0.4953639507293701,
      "learning_rate": 5.287814015264107e-06,
      "loss": 0.3838,
      "step": 32166
    },
    {
      "epoch": 1.4761598825203066,
      "grad_norm": 0.4996055066585541,
      "learning_rate": 5.2875692335602705e-06,
      "loss": 0.4255,
      "step": 32167
    },
    {
      "epoch": 1.476205773025561,
      "grad_norm": 0.47862255573272705,
      "learning_rate": 5.2873244511649224e-06,
      "loss": 0.3787,
      "step": 32168
    },
    {
      "epoch": 1.4762516635308156,
      "grad_norm": 0.4911181926727295,
      "learning_rate": 5.2870796680786505e-06,
      "loss": 0.3761,
      "step": 32169
    },
    {
      "epoch": 1.4762975540360699,
      "grad_norm": 0.43438398838043213,
      "learning_rate": 5.286834884302041e-06,
      "loss": 0.3189,
      "step": 32170
    },
    {
      "epoch": 1.4763434445413244,
      "grad_norm": 0.49739471077919006,
      "learning_rate": 5.2865900998356845e-06,
      "loss": 0.3793,
      "step": 32171
    },
    {
      "epoch": 1.4763893350465789,
      "grad_norm": 0.48677533864974976,
      "learning_rate": 5.286345314680169e-06,
      "loss": 0.3802,
      "step": 32172
    },
    {
      "epoch": 1.4764352255518334,
      "grad_norm": 0.4675295650959015,
      "learning_rate": 5.2861005288360845e-06,
      "loss": 0.35,
      "step": 32173
    },
    {
      "epoch": 1.4764811160570877,
      "grad_norm": 0.4477384090423584,
      "learning_rate": 5.2858557423040165e-06,
      "loss": 0.3341,
      "step": 32174
    },
    {
      "epoch": 1.4765270065623421,
      "grad_norm": 0.41926464438438416,
      "learning_rate": 5.285610955084558e-06,
      "loss": 0.292,
      "step": 32175
    },
    {
      "epoch": 1.4765728970675966,
      "grad_norm": 0.4473084807395935,
      "learning_rate": 5.285366167178295e-06,
      "loss": 0.2846,
      "step": 32176
    },
    {
      "epoch": 1.4766187875728511,
      "grad_norm": 0.450969398021698,
      "learning_rate": 5.285121378585815e-06,
      "loss": 0.3367,
      "step": 32177
    },
    {
      "epoch": 1.4766646780781056,
      "grad_norm": 0.46748706698417664,
      "learning_rate": 5.284876589307708e-06,
      "loss": 0.3904,
      "step": 32178
    },
    {
      "epoch": 1.4767105685833601,
      "grad_norm": 0.4565579295158386,
      "learning_rate": 5.284631799344564e-06,
      "loss": 0.3298,
      "step": 32179
    },
    {
      "epoch": 1.4767564590886146,
      "grad_norm": 0.4720044434070587,
      "learning_rate": 5.284387008696971e-06,
      "loss": 0.3543,
      "step": 32180
    },
    {
      "epoch": 1.4768023495938691,
      "grad_norm": 0.4883100390434265,
      "learning_rate": 5.284142217365514e-06,
      "loss": 0.3532,
      "step": 32181
    },
    {
      "epoch": 1.4768482400991234,
      "grad_norm": 0.48219752311706543,
      "learning_rate": 5.283897425350788e-06,
      "loss": 0.3726,
      "step": 32182
    },
    {
      "epoch": 1.476894130604378,
      "grad_norm": 0.46640753746032715,
      "learning_rate": 5.283652632653376e-06,
      "loss": 0.324,
      "step": 32183
    },
    {
      "epoch": 1.4769400211096324,
      "grad_norm": 0.49780508875846863,
      "learning_rate": 5.283407839273869e-06,
      "loss": 0.4142,
      "step": 32184
    },
    {
      "epoch": 1.476985911614887,
      "grad_norm": 0.4580658972263336,
      "learning_rate": 5.283163045212857e-06,
      "loss": 0.3527,
      "step": 32185
    },
    {
      "epoch": 1.4770318021201414,
      "grad_norm": 0.47155308723449707,
      "learning_rate": 5.282918250470926e-06,
      "loss": 0.3651,
      "step": 32186
    },
    {
      "epoch": 1.4770776926253957,
      "grad_norm": 0.5855438709259033,
      "learning_rate": 5.2826734550486665e-06,
      "loss": 0.4308,
      "step": 32187
    },
    {
      "epoch": 1.4771235831306502,
      "grad_norm": 0.4865555465221405,
      "learning_rate": 5.282428658946667e-06,
      "loss": 0.421,
      "step": 32188
    },
    {
      "epoch": 1.4771694736359047,
      "grad_norm": 0.5320597887039185,
      "learning_rate": 5.282183862165514e-06,
      "loss": 0.3529,
      "step": 32189
    },
    {
      "epoch": 1.4772153641411592,
      "grad_norm": 0.47090551257133484,
      "learning_rate": 5.2819390647058e-06,
      "loss": 0.3681,
      "step": 32190
    },
    {
      "epoch": 1.4772612546464137,
      "grad_norm": 0.5162542462348938,
      "learning_rate": 5.281694266568112e-06,
      "loss": 0.4189,
      "step": 32191
    },
    {
      "epoch": 1.4773071451516682,
      "grad_norm": 0.48198291659355164,
      "learning_rate": 5.281449467753036e-06,
      "loss": 0.3873,
      "step": 32192
    },
    {
      "epoch": 1.4773530356569227,
      "grad_norm": 0.45941951870918274,
      "learning_rate": 5.281204668261164e-06,
      "loss": 0.3788,
      "step": 32193
    },
    {
      "epoch": 1.4773989261621772,
      "grad_norm": 0.44771191477775574,
      "learning_rate": 5.280959868093083e-06,
      "loss": 0.3224,
      "step": 32194
    },
    {
      "epoch": 1.4774448166674314,
      "grad_norm": 0.5035176873207092,
      "learning_rate": 5.2807150672493816e-06,
      "loss": 0.4214,
      "step": 32195
    },
    {
      "epoch": 1.477490707172686,
      "grad_norm": 0.4409049451351166,
      "learning_rate": 5.2804702657306495e-06,
      "loss": 0.2641,
      "step": 32196
    },
    {
      "epoch": 1.4775365976779404,
      "grad_norm": 0.48864734172821045,
      "learning_rate": 5.280225463537476e-06,
      "loss": 0.3558,
      "step": 32197
    },
    {
      "epoch": 1.477582488183195,
      "grad_norm": 0.50663822889328,
      "learning_rate": 5.279980660670448e-06,
      "loss": 0.3584,
      "step": 32198
    },
    {
      "epoch": 1.4776283786884494,
      "grad_norm": 0.45244652032852173,
      "learning_rate": 5.279735857130155e-06,
      "loss": 0.3552,
      "step": 32199
    },
    {
      "epoch": 1.4776742691937037,
      "grad_norm": 0.5134270191192627,
      "learning_rate": 5.279491052917185e-06,
      "loss": 0.4434,
      "step": 32200
    },
    {
      "epoch": 1.4777201596989582,
      "grad_norm": 0.441841185092926,
      "learning_rate": 5.279246248032127e-06,
      "loss": 0.3022,
      "step": 32201
    },
    {
      "epoch": 1.4777660502042127,
      "grad_norm": 0.455996572971344,
      "learning_rate": 5.279001442475572e-06,
      "loss": 0.3428,
      "step": 32202
    },
    {
      "epoch": 1.4778119407094672,
      "grad_norm": 0.49613645672798157,
      "learning_rate": 5.278756636248104e-06,
      "loss": 0.3667,
      "step": 32203
    },
    {
      "epoch": 1.4778578312147217,
      "grad_norm": 0.4697081744670868,
      "learning_rate": 5.278511829350316e-06,
      "loss": 0.3463,
      "step": 32204
    },
    {
      "epoch": 1.4779037217199762,
      "grad_norm": 0.44042932987213135,
      "learning_rate": 5.278267021782794e-06,
      "loss": 0.3127,
      "step": 32205
    },
    {
      "epoch": 1.4779496122252307,
      "grad_norm": 0.5432997941970825,
      "learning_rate": 5.278022213546128e-06,
      "loss": 0.5186,
      "step": 32206
    },
    {
      "epoch": 1.4779955027304852,
      "grad_norm": 0.5859554409980774,
      "learning_rate": 5.277777404640906e-06,
      "loss": 0.3792,
      "step": 32207
    },
    {
      "epoch": 1.4780413932357395,
      "grad_norm": 0.42461588978767395,
      "learning_rate": 5.277532595067718e-06,
      "loss": 0.2833,
      "step": 32208
    },
    {
      "epoch": 1.478087283740994,
      "grad_norm": 0.4361141622066498,
      "learning_rate": 5.2772877848271506e-06,
      "loss": 0.3243,
      "step": 32209
    },
    {
      "epoch": 1.4781331742462485,
      "grad_norm": 0.45226922631263733,
      "learning_rate": 5.277042973919795e-06,
      "loss": 0.3091,
      "step": 32210
    },
    {
      "epoch": 1.478179064751503,
      "grad_norm": 0.45459383726119995,
      "learning_rate": 5.2767981623462365e-06,
      "loss": 0.3537,
      "step": 32211
    },
    {
      "epoch": 1.4782249552567575,
      "grad_norm": 0.4496331810951233,
      "learning_rate": 5.276553350107066e-06,
      "loss": 0.3471,
      "step": 32212
    },
    {
      "epoch": 1.4782708457620117,
      "grad_norm": 0.49105075001716614,
      "learning_rate": 5.276308537202874e-06,
      "loss": 0.4306,
      "step": 32213
    },
    {
      "epoch": 1.4783167362672662,
      "grad_norm": 0.6723960041999817,
      "learning_rate": 5.276063723634246e-06,
      "loss": 0.353,
      "step": 32214
    },
    {
      "epoch": 1.4783626267725207,
      "grad_norm": 0.4893755316734314,
      "learning_rate": 5.275818909401771e-06,
      "loss": 0.3834,
      "step": 32215
    },
    {
      "epoch": 1.4784085172777752,
      "grad_norm": 0.5055813789367676,
      "learning_rate": 5.27557409450604e-06,
      "loss": 0.3832,
      "step": 32216
    },
    {
      "epoch": 1.4784544077830297,
      "grad_norm": 0.6979690194129944,
      "learning_rate": 5.275329278947638e-06,
      "loss": 0.3284,
      "step": 32217
    },
    {
      "epoch": 1.4785002982882842,
      "grad_norm": 0.4800167679786682,
      "learning_rate": 5.2750844627271594e-06,
      "loss": 0.3489,
      "step": 32218
    },
    {
      "epoch": 1.4785461887935387,
      "grad_norm": 0.5017383694648743,
      "learning_rate": 5.274839645845188e-06,
      "loss": 0.4234,
      "step": 32219
    },
    {
      "epoch": 1.4785920792987932,
      "grad_norm": 0.4398152232170105,
      "learning_rate": 5.2745948283023126e-06,
      "loss": 0.2997,
      "step": 32220
    },
    {
      "epoch": 1.4786379698040475,
      "grad_norm": 0.48357245326042175,
      "learning_rate": 5.274350010099125e-06,
      "loss": 0.4023,
      "step": 32221
    },
    {
      "epoch": 1.478683860309302,
      "grad_norm": 0.4752514660358429,
      "learning_rate": 5.27410519123621e-06,
      "loss": 0.3683,
      "step": 32222
    },
    {
      "epoch": 1.4787297508145565,
      "grad_norm": 0.44536012411117554,
      "learning_rate": 5.27386037171416e-06,
      "loss": 0.2829,
      "step": 32223
    },
    {
      "epoch": 1.478775641319811,
      "grad_norm": 0.46751123666763306,
      "learning_rate": 5.273615551533564e-06,
      "loss": 0.2962,
      "step": 32224
    },
    {
      "epoch": 1.4788215318250653,
      "grad_norm": 0.4844211935997009,
      "learning_rate": 5.273370730695008e-06,
      "loss": 0.3782,
      "step": 32225
    },
    {
      "epoch": 1.4788674223303198,
      "grad_norm": 0.4923376441001892,
      "learning_rate": 5.273125909199079e-06,
      "loss": 0.4338,
      "step": 32226
    },
    {
      "epoch": 1.4789133128355743,
      "grad_norm": 0.4800521433353424,
      "learning_rate": 5.272881087046372e-06,
      "loss": 0.3419,
      "step": 32227
    },
    {
      "epoch": 1.4789592033408288,
      "grad_norm": 0.5042484998703003,
      "learning_rate": 5.2726362642374695e-06,
      "loss": 0.3546,
      "step": 32228
    },
    {
      "epoch": 1.4790050938460833,
      "grad_norm": 0.4690392017364502,
      "learning_rate": 5.272391440772965e-06,
      "loss": 0.3362,
      "step": 32229
    },
    {
      "epoch": 1.4790509843513377,
      "grad_norm": 0.4888511002063751,
      "learning_rate": 5.272146616653444e-06,
      "loss": 0.3994,
      "step": 32230
    },
    {
      "epoch": 1.4790968748565922,
      "grad_norm": 0.4465472996234894,
      "learning_rate": 5.271901791879496e-06,
      "loss": 0.3252,
      "step": 32231
    },
    {
      "epoch": 1.4791427653618467,
      "grad_norm": 0.4621584415435791,
      "learning_rate": 5.271656966451709e-06,
      "loss": 0.3453,
      "step": 32232
    },
    {
      "epoch": 1.479188655867101,
      "grad_norm": 0.4941501319408417,
      "learning_rate": 5.271412140370674e-06,
      "loss": 0.3861,
      "step": 32233
    },
    {
      "epoch": 1.4792345463723555,
      "grad_norm": 0.4701978862285614,
      "learning_rate": 5.271167313636978e-06,
      "loss": 0.3554,
      "step": 32234
    },
    {
      "epoch": 1.47928043687761,
      "grad_norm": 0.44477006793022156,
      "learning_rate": 5.270922486251211e-06,
      "loss": 0.3368,
      "step": 32235
    },
    {
      "epoch": 1.4793263273828645,
      "grad_norm": 0.46687495708465576,
      "learning_rate": 5.270677658213961e-06,
      "loss": 0.3784,
      "step": 32236
    },
    {
      "epoch": 1.479372217888119,
      "grad_norm": 0.46737995743751526,
      "learning_rate": 5.270432829525816e-06,
      "loss": 0.3469,
      "step": 32237
    },
    {
      "epoch": 1.4794181083933733,
      "grad_norm": 0.4454015791416168,
      "learning_rate": 5.270188000187366e-06,
      "loss": 0.3089,
      "step": 32238
    },
    {
      "epoch": 1.4794639988986278,
      "grad_norm": 0.46454596519470215,
      "learning_rate": 5.269943170199197e-06,
      "loss": 0.3817,
      "step": 32239
    },
    {
      "epoch": 1.4795098894038823,
      "grad_norm": 0.4369686245918274,
      "learning_rate": 5.269698339561903e-06,
      "loss": 0.284,
      "step": 32240
    },
    {
      "epoch": 1.4795557799091368,
      "grad_norm": 0.4287276864051819,
      "learning_rate": 5.269453508276069e-06,
      "loss": 0.2818,
      "step": 32241
    },
    {
      "epoch": 1.4796016704143913,
      "grad_norm": 0.45772886276245117,
      "learning_rate": 5.269208676342283e-06,
      "loss": 0.3319,
      "step": 32242
    },
    {
      "epoch": 1.4796475609196458,
      "grad_norm": 0.47653254866600037,
      "learning_rate": 5.268963843761135e-06,
      "loss": 0.3294,
      "step": 32243
    },
    {
      "epoch": 1.4796934514249003,
      "grad_norm": 0.5394758582115173,
      "learning_rate": 5.268719010533214e-06,
      "loss": 0.3902,
      "step": 32244
    },
    {
      "epoch": 1.4797393419301548,
      "grad_norm": 0.45063939690589905,
      "learning_rate": 5.2684741766591085e-06,
      "loss": 0.3184,
      "step": 32245
    },
    {
      "epoch": 1.479785232435409,
      "grad_norm": 0.4706563353538513,
      "learning_rate": 5.268229342139408e-06,
      "loss": 0.4076,
      "step": 32246
    },
    {
      "epoch": 1.4798311229406635,
      "grad_norm": 0.5012007355690002,
      "learning_rate": 5.267984506974702e-06,
      "loss": 0.407,
      "step": 32247
    },
    {
      "epoch": 1.479877013445918,
      "grad_norm": 0.48534274101257324,
      "learning_rate": 5.267739671165576e-06,
      "loss": 0.3283,
      "step": 32248
    },
    {
      "epoch": 1.4799229039511725,
      "grad_norm": 0.4272443354129791,
      "learning_rate": 5.2674948347126195e-06,
      "loss": 0.2831,
      "step": 32249
    },
    {
      "epoch": 1.479968794456427,
      "grad_norm": 0.45840010046958923,
      "learning_rate": 5.267249997616425e-06,
      "loss": 0.3179,
      "step": 32250
    },
    {
      "epoch": 1.4800146849616813,
      "grad_norm": 0.4599660038948059,
      "learning_rate": 5.267005159877577e-06,
      "loss": 0.3739,
      "step": 32251
    },
    {
      "epoch": 1.4800605754669358,
      "grad_norm": 0.4887057840824127,
      "learning_rate": 5.266760321496666e-06,
      "loss": 0.3721,
      "step": 32252
    },
    {
      "epoch": 1.4801064659721903,
      "grad_norm": 0.4779162108898163,
      "learning_rate": 5.266515482474281e-06,
      "loss": 0.4151,
      "step": 32253
    },
    {
      "epoch": 1.4801523564774448,
      "grad_norm": 0.46435460448265076,
      "learning_rate": 5.26627064281101e-06,
      "loss": 0.3655,
      "step": 32254
    },
    {
      "epoch": 1.4801982469826993,
      "grad_norm": 0.4353535771369934,
      "learning_rate": 5.2660258025074425e-06,
      "loss": 0.2925,
      "step": 32255
    },
    {
      "epoch": 1.4802441374879538,
      "grad_norm": 0.4641478657722473,
      "learning_rate": 5.265780961564167e-06,
      "loss": 0.3714,
      "step": 32256
    },
    {
      "epoch": 1.4802900279932083,
      "grad_norm": 0.4852233827114105,
      "learning_rate": 5.265536119981772e-06,
      "loss": 0.3744,
      "step": 32257
    },
    {
      "epoch": 1.4803359184984628,
      "grad_norm": 0.42776840925216675,
      "learning_rate": 5.265291277760846e-06,
      "loss": 0.2918,
      "step": 32258
    },
    {
      "epoch": 1.480381809003717,
      "grad_norm": 0.45558902621269226,
      "learning_rate": 5.265046434901979e-06,
      "loss": 0.3256,
      "step": 32259
    },
    {
      "epoch": 1.4804276995089716,
      "grad_norm": 0.4512421190738678,
      "learning_rate": 5.264801591405757e-06,
      "loss": 0.331,
      "step": 32260
    },
    {
      "epoch": 1.480473590014226,
      "grad_norm": 0.4861771762371063,
      "learning_rate": 5.264556747272773e-06,
      "loss": 0.3928,
      "step": 32261
    },
    {
      "epoch": 1.4805194805194806,
      "grad_norm": 0.44187191128730774,
      "learning_rate": 5.2643119025036135e-06,
      "loss": 0.3047,
      "step": 32262
    },
    {
      "epoch": 1.4805653710247348,
      "grad_norm": 0.4591623842716217,
      "learning_rate": 5.264067057098866e-06,
      "loss": 0.3472,
      "step": 32263
    },
    {
      "epoch": 1.4806112615299893,
      "grad_norm": 0.46147510409355164,
      "learning_rate": 5.263822211059122e-06,
      "loss": 0.4002,
      "step": 32264
    },
    {
      "epoch": 1.4806571520352438,
      "grad_norm": 0.5028408765792847,
      "learning_rate": 5.2635773643849675e-06,
      "loss": 0.3535,
      "step": 32265
    },
    {
      "epoch": 1.4807030425404983,
      "grad_norm": 0.5251145958900452,
      "learning_rate": 5.263332517076992e-06,
      "loss": 0.446,
      "step": 32266
    },
    {
      "epoch": 1.4807489330457528,
      "grad_norm": 0.4643406569957733,
      "learning_rate": 5.263087669135787e-06,
      "loss": 0.3698,
      "step": 32267
    },
    {
      "epoch": 1.4807948235510073,
      "grad_norm": 0.4679146409034729,
      "learning_rate": 5.262842820561938e-06,
      "loss": 0.3567,
      "step": 32268
    },
    {
      "epoch": 1.4808407140562618,
      "grad_norm": 0.46273306012153625,
      "learning_rate": 5.262597971356035e-06,
      "loss": 0.3765,
      "step": 32269
    },
    {
      "epoch": 1.4808866045615163,
      "grad_norm": 0.42155322432518005,
      "learning_rate": 5.262353121518667e-06,
      "loss": 0.2734,
      "step": 32270
    },
    {
      "epoch": 1.4809324950667706,
      "grad_norm": 0.43986257910728455,
      "learning_rate": 5.262108271050422e-06,
      "loss": 0.3089,
      "step": 32271
    },
    {
      "epoch": 1.480978385572025,
      "grad_norm": 0.4873642325401306,
      "learning_rate": 5.261863419951891e-06,
      "loss": 0.3635,
      "step": 32272
    },
    {
      "epoch": 1.4810242760772796,
      "grad_norm": 0.502087414264679,
      "learning_rate": 5.261618568223661e-06,
      "loss": 0.3935,
      "step": 32273
    },
    {
      "epoch": 1.481070166582534,
      "grad_norm": 0.4453190565109253,
      "learning_rate": 5.261373715866319e-06,
      "loss": 0.3367,
      "step": 32274
    },
    {
      "epoch": 1.4811160570877886,
      "grad_norm": 0.4663076400756836,
      "learning_rate": 5.261128862880457e-06,
      "loss": 0.3405,
      "step": 32275
    },
    {
      "epoch": 1.4811619475930429,
      "grad_norm": 0.43091490864753723,
      "learning_rate": 5.260884009266661e-06,
      "loss": 0.2768,
      "step": 32276
    },
    {
      "epoch": 1.4812078380982974,
      "grad_norm": 0.4805290102958679,
      "learning_rate": 5.260639155025522e-06,
      "loss": 0.3561,
      "step": 32277
    },
    {
      "epoch": 1.4812537286035519,
      "grad_norm": 0.5315728187561035,
      "learning_rate": 5.26039430015763e-06,
      "loss": 0.4363,
      "step": 32278
    },
    {
      "epoch": 1.4812996191088064,
      "grad_norm": 0.49049949645996094,
      "learning_rate": 5.260149444663571e-06,
      "loss": 0.3421,
      "step": 32279
    },
    {
      "epoch": 1.4813455096140609,
      "grad_norm": 0.49210989475250244,
      "learning_rate": 5.259904588543933e-06,
      "loss": 0.3901,
      "step": 32280
    },
    {
      "epoch": 1.4813914001193154,
      "grad_norm": 0.4296182692050934,
      "learning_rate": 5.259659731799307e-06,
      "loss": 0.3098,
      "step": 32281
    },
    {
      "epoch": 1.4814372906245699,
      "grad_norm": 0.5041384696960449,
      "learning_rate": 5.259414874430282e-06,
      "loss": 0.41,
      "step": 32282
    },
    {
      "epoch": 1.4814831811298244,
      "grad_norm": 0.4752067029476166,
      "learning_rate": 5.259170016437446e-06,
      "loss": 0.3963,
      "step": 32283
    },
    {
      "epoch": 1.4815290716350786,
      "grad_norm": 0.5466283559799194,
      "learning_rate": 5.2589251578213884e-06,
      "loss": 0.2981,
      "step": 32284
    },
    {
      "epoch": 1.4815749621403331,
      "grad_norm": 0.4475005865097046,
      "learning_rate": 5.2586802985826965e-06,
      "loss": 0.3144,
      "step": 32285
    },
    {
      "epoch": 1.4816208526455876,
      "grad_norm": 0.5330756902694702,
      "learning_rate": 5.258435438721961e-06,
      "loss": 0.435,
      "step": 32286
    },
    {
      "epoch": 1.4816667431508421,
      "grad_norm": 0.48924750089645386,
      "learning_rate": 5.258190578239769e-06,
      "loss": 0.357,
      "step": 32287
    },
    {
      "epoch": 1.4817126336560966,
      "grad_norm": 0.4536020755767822,
      "learning_rate": 5.25794571713671e-06,
      "loss": 0.3328,
      "step": 32288
    },
    {
      "epoch": 1.481758524161351,
      "grad_norm": 0.6382410526275635,
      "learning_rate": 5.257700855413373e-06,
      "loss": 0.3403,
      "step": 32289
    },
    {
      "epoch": 1.4818044146666054,
      "grad_norm": 0.4127758741378784,
      "learning_rate": 5.257455993070349e-06,
      "loss": 0.2692,
      "step": 32290
    },
    {
      "epoch": 1.48185030517186,
      "grad_norm": 0.45960569381713867,
      "learning_rate": 5.257211130108222e-06,
      "loss": 0.3781,
      "step": 32291
    },
    {
      "epoch": 1.4818961956771144,
      "grad_norm": 0.4709983468055725,
      "learning_rate": 5.256966266527585e-06,
      "loss": 0.3833,
      "step": 32292
    },
    {
      "epoch": 1.481942086182369,
      "grad_norm": 0.4444618821144104,
      "learning_rate": 5.2567214023290235e-06,
      "loss": 0.3394,
      "step": 32293
    },
    {
      "epoch": 1.4819879766876234,
      "grad_norm": 0.457006573677063,
      "learning_rate": 5.25647653751313e-06,
      "loss": 0.317,
      "step": 32294
    },
    {
      "epoch": 1.4820338671928779,
      "grad_norm": 0.4663785696029663,
      "learning_rate": 5.2562316720804904e-06,
      "loss": 0.3268,
      "step": 32295
    },
    {
      "epoch": 1.4820797576981324,
      "grad_norm": 0.45807361602783203,
      "learning_rate": 5.255986806031694e-06,
      "loss": 0.3586,
      "step": 32296
    },
    {
      "epoch": 1.4821256482033867,
      "grad_norm": 0.5094810128211975,
      "learning_rate": 5.25574193936733e-06,
      "loss": 0.4504,
      "step": 32297
    },
    {
      "epoch": 1.4821715387086412,
      "grad_norm": 0.49613264203071594,
      "learning_rate": 5.255497072087989e-06,
      "loss": 0.3469,
      "step": 32298
    },
    {
      "epoch": 1.4822174292138957,
      "grad_norm": 0.4654496908187866,
      "learning_rate": 5.255252204194257e-06,
      "loss": 0.357,
      "step": 32299
    },
    {
      "epoch": 1.4822633197191502,
      "grad_norm": 0.4875797629356384,
      "learning_rate": 5.255007335686725e-06,
      "loss": 0.3411,
      "step": 32300
    },
    {
      "epoch": 1.4823092102244046,
      "grad_norm": 0.47078442573547363,
      "learning_rate": 5.25476246656598e-06,
      "loss": 0.391,
      "step": 32301
    },
    {
      "epoch": 1.482355100729659,
      "grad_norm": 0.46430617570877075,
      "learning_rate": 5.2545175968326115e-06,
      "loss": 0.4009,
      "step": 32302
    },
    {
      "epoch": 1.4824009912349134,
      "grad_norm": 0.45111218094825745,
      "learning_rate": 5.2542727264872085e-06,
      "loss": 0.3337,
      "step": 32303
    },
    {
      "epoch": 1.482446881740168,
      "grad_norm": 0.5206025838851929,
      "learning_rate": 5.25402785553036e-06,
      "loss": 0.4645,
      "step": 32304
    },
    {
      "epoch": 1.4824927722454224,
      "grad_norm": 0.4681684374809265,
      "learning_rate": 5.253782983962656e-06,
      "loss": 0.376,
      "step": 32305
    },
    {
      "epoch": 1.482538662750677,
      "grad_norm": 0.47294217348098755,
      "learning_rate": 5.2535381117846825e-06,
      "loss": 0.3636,
      "step": 32306
    },
    {
      "epoch": 1.4825845532559314,
      "grad_norm": 0.4847589135169983,
      "learning_rate": 5.253293238997029e-06,
      "loss": 0.4077,
      "step": 32307
    },
    {
      "epoch": 1.482630443761186,
      "grad_norm": 0.4365226626396179,
      "learning_rate": 5.253048365600286e-06,
      "loss": 0.3256,
      "step": 32308
    },
    {
      "epoch": 1.4826763342664404,
      "grad_norm": 0.45794156193733215,
      "learning_rate": 5.252803491595043e-06,
      "loss": 0.3799,
      "step": 32309
    },
    {
      "epoch": 1.4827222247716947,
      "grad_norm": 0.43100976943969727,
      "learning_rate": 5.252558616981886e-06,
      "loss": 0.3047,
      "step": 32310
    },
    {
      "epoch": 1.4827681152769492,
      "grad_norm": 0.5044086575508118,
      "learning_rate": 5.252313741761406e-06,
      "loss": 0.4274,
      "step": 32311
    },
    {
      "epoch": 1.4828140057822037,
      "grad_norm": 0.4636385142803192,
      "learning_rate": 5.25206886593419e-06,
      "loss": 0.3955,
      "step": 32312
    },
    {
      "epoch": 1.4828598962874582,
      "grad_norm": 0.4615718424320221,
      "learning_rate": 5.2518239895008285e-06,
      "loss": 0.3069,
      "step": 32313
    },
    {
      "epoch": 1.4829057867927125,
      "grad_norm": 0.4650508463382721,
      "learning_rate": 5.2515791124619085e-06,
      "loss": 0.3308,
      "step": 32314
    },
    {
      "epoch": 1.482951677297967,
      "grad_norm": 0.466871052980423,
      "learning_rate": 5.251334234818023e-06,
      "loss": 0.3862,
      "step": 32315
    },
    {
      "epoch": 1.4829975678032215,
      "grad_norm": 0.45182231068611145,
      "learning_rate": 5.251089356569756e-06,
      "loss": 0.356,
      "step": 32316
    },
    {
      "epoch": 1.483043458308476,
      "grad_norm": 0.441617876291275,
      "learning_rate": 5.2508444777176996e-06,
      "loss": 0.2881,
      "step": 32317
    },
    {
      "epoch": 1.4830893488137304,
      "grad_norm": 0.4877391755580902,
      "learning_rate": 5.25059959826244e-06,
      "loss": 0.3611,
      "step": 32318
    },
    {
      "epoch": 1.483135239318985,
      "grad_norm": 0.4494521915912628,
      "learning_rate": 5.250354718204568e-06,
      "loss": 0.3606,
      "step": 32319
    },
    {
      "epoch": 1.4831811298242394,
      "grad_norm": 0.4764392375946045,
      "learning_rate": 5.250109837544673e-06,
      "loss": 0.3567,
      "step": 32320
    },
    {
      "epoch": 1.483227020329494,
      "grad_norm": 0.48611682653427124,
      "learning_rate": 5.24986495628334e-06,
      "loss": 0.3997,
      "step": 32321
    },
    {
      "epoch": 1.4832729108347482,
      "grad_norm": 0.5012817978858948,
      "learning_rate": 5.249620074421164e-06,
      "loss": 0.4123,
      "step": 32322
    },
    {
      "epoch": 1.4833188013400027,
      "grad_norm": 0.4377935230731964,
      "learning_rate": 5.249375191958729e-06,
      "loss": 0.3091,
      "step": 32323
    },
    {
      "epoch": 1.4833646918452572,
      "grad_norm": 0.48370859026908875,
      "learning_rate": 5.249130308896624e-06,
      "loss": 0.288,
      "step": 32324
    },
    {
      "epoch": 1.4834105823505117,
      "grad_norm": 0.45328912138938904,
      "learning_rate": 5.24888542523544e-06,
      "loss": 0.3471,
      "step": 32325
    },
    {
      "epoch": 1.4834564728557662,
      "grad_norm": 0.47834107279777527,
      "learning_rate": 5.248640540975767e-06,
      "loss": 0.3894,
      "step": 32326
    },
    {
      "epoch": 1.4835023633610205,
      "grad_norm": 0.47861477732658386,
      "learning_rate": 5.248395656118191e-06,
      "loss": 0.3708,
      "step": 32327
    },
    {
      "epoch": 1.483548253866275,
      "grad_norm": 0.5039728879928589,
      "learning_rate": 5.2481507706633e-06,
      "loss": 0.4615,
      "step": 32328
    },
    {
      "epoch": 1.4835941443715295,
      "grad_norm": 0.4393763542175293,
      "learning_rate": 5.247905884611687e-06,
      "loss": 0.3145,
      "step": 32329
    },
    {
      "epoch": 1.483640034876784,
      "grad_norm": 0.4937411844730377,
      "learning_rate": 5.247660997963937e-06,
      "loss": 0.4024,
      "step": 32330
    },
    {
      "epoch": 1.4836859253820385,
      "grad_norm": 0.4825166165828705,
      "learning_rate": 5.24741611072064e-06,
      "loss": 0.4018,
      "step": 32331
    },
    {
      "epoch": 1.483731815887293,
      "grad_norm": 0.5424790978431702,
      "learning_rate": 5.247171222882388e-06,
      "loss": 0.4045,
      "step": 32332
    },
    {
      "epoch": 1.4837777063925475,
      "grad_norm": 0.41208282113075256,
      "learning_rate": 5.246926334449766e-06,
      "loss": 0.2672,
      "step": 32333
    },
    {
      "epoch": 1.483823596897802,
      "grad_norm": 0.45391663908958435,
      "learning_rate": 5.246681445423363e-06,
      "loss": 0.3338,
      "step": 32334
    },
    {
      "epoch": 1.4838694874030562,
      "grad_norm": 0.5171379446983337,
      "learning_rate": 5.24643655580377e-06,
      "loss": 0.436,
      "step": 32335
    },
    {
      "epoch": 1.4839153779083107,
      "grad_norm": 0.46104681491851807,
      "learning_rate": 5.246191665591574e-06,
      "loss": 0.3358,
      "step": 32336
    },
    {
      "epoch": 1.4839612684135652,
      "grad_norm": 0.4757278859615326,
      "learning_rate": 5.245946774787366e-06,
      "loss": 0.3098,
      "step": 32337
    },
    {
      "epoch": 1.4840071589188197,
      "grad_norm": 0.46621063351631165,
      "learning_rate": 5.245701883391734e-06,
      "loss": 0.3207,
      "step": 32338
    },
    {
      "epoch": 1.4840530494240742,
      "grad_norm": 0.4821682870388031,
      "learning_rate": 5.245456991405264e-06,
      "loss": 0.3918,
      "step": 32339
    },
    {
      "epoch": 1.4840989399293285,
      "grad_norm": 0.4943676292896271,
      "learning_rate": 5.2452120988285505e-06,
      "loss": 0.3877,
      "step": 32340
    },
    {
      "epoch": 1.484144830434583,
      "grad_norm": 0.45784708857536316,
      "learning_rate": 5.244967205662177e-06,
      "loss": 0.3254,
      "step": 32341
    },
    {
      "epoch": 1.4841907209398375,
      "grad_norm": 0.4543229341506958,
      "learning_rate": 5.2447223119067345e-06,
      "loss": 0.3261,
      "step": 32342
    },
    {
      "epoch": 1.484236611445092,
      "grad_norm": 0.45193830132484436,
      "learning_rate": 5.244477417562814e-06,
      "loss": 0.29,
      "step": 32343
    },
    {
      "epoch": 1.4842825019503465,
      "grad_norm": 0.5345727205276489,
      "learning_rate": 5.244232522631002e-06,
      "loss": 0.3467,
      "step": 32344
    },
    {
      "epoch": 1.484328392455601,
      "grad_norm": 0.48128750920295715,
      "learning_rate": 5.2439876271118865e-06,
      "loss": 0.3923,
      "step": 32345
    },
    {
      "epoch": 1.4843742829608555,
      "grad_norm": 0.45415621995925903,
      "learning_rate": 5.243742731006059e-06,
      "loss": 0.3143,
      "step": 32346
    },
    {
      "epoch": 1.48442017346611,
      "grad_norm": 0.45187845826148987,
      "learning_rate": 5.243497834314106e-06,
      "loss": 0.2751,
      "step": 32347
    },
    {
      "epoch": 1.4844660639713643,
      "grad_norm": 0.48964959383010864,
      "learning_rate": 5.243252937036619e-06,
      "loss": 0.3606,
      "step": 32348
    },
    {
      "epoch": 1.4845119544766188,
      "grad_norm": 0.47665393352508545,
      "learning_rate": 5.243008039174185e-06,
      "loss": 0.3824,
      "step": 32349
    },
    {
      "epoch": 1.4845578449818733,
      "grad_norm": 0.45537135004997253,
      "learning_rate": 5.242763140727393e-06,
      "loss": 0.3521,
      "step": 32350
    },
    {
      "epoch": 1.4846037354871278,
      "grad_norm": 0.5056536793708801,
      "learning_rate": 5.242518241696833e-06,
      "loss": 0.4333,
      "step": 32351
    },
    {
      "epoch": 1.484649625992382,
      "grad_norm": 0.46078360080718994,
      "learning_rate": 5.242273342083092e-06,
      "loss": 0.3357,
      "step": 32352
    },
    {
      "epoch": 1.4846955164976365,
      "grad_norm": 0.45424604415893555,
      "learning_rate": 5.24202844188676e-06,
      "loss": 0.3517,
      "step": 32353
    },
    {
      "epoch": 1.484741407002891,
      "grad_norm": 0.48669958114624023,
      "learning_rate": 5.241783541108427e-06,
      "loss": 0.3621,
      "step": 32354
    },
    {
      "epoch": 1.4847872975081455,
      "grad_norm": 0.46680521965026855,
      "learning_rate": 5.241538639748681e-06,
      "loss": 0.3291,
      "step": 32355
    },
    {
      "epoch": 1.4848331880134,
      "grad_norm": 0.47588977217674255,
      "learning_rate": 5.24129373780811e-06,
      "loss": 0.3283,
      "step": 32356
    },
    {
      "epoch": 1.4848790785186545,
      "grad_norm": 0.5736688375473022,
      "learning_rate": 5.241048835287305e-06,
      "loss": 0.3394,
      "step": 32357
    },
    {
      "epoch": 1.484924969023909,
      "grad_norm": 0.443052738904953,
      "learning_rate": 5.240803932186852e-06,
      "loss": 0.3372,
      "step": 32358
    },
    {
      "epoch": 1.4849708595291635,
      "grad_norm": 0.4627077877521515,
      "learning_rate": 5.2405590285073425e-06,
      "loss": 0.336,
      "step": 32359
    },
    {
      "epoch": 1.4850167500344178,
      "grad_norm": 0.48419806361198425,
      "learning_rate": 5.240314124249365e-06,
      "loss": 0.3496,
      "step": 32360
    },
    {
      "epoch": 1.4850626405396723,
      "grad_norm": 0.4560392200946808,
      "learning_rate": 5.2400692194135065e-06,
      "loss": 0.3324,
      "step": 32361
    },
    {
      "epoch": 1.4851085310449268,
      "grad_norm": 0.4768802225589752,
      "learning_rate": 5.239824314000357e-06,
      "loss": 0.3035,
      "step": 32362
    },
    {
      "epoch": 1.4851544215501813,
      "grad_norm": 0.4935522973537445,
      "learning_rate": 5.239579408010507e-06,
      "loss": 0.4155,
      "step": 32363
    },
    {
      "epoch": 1.4852003120554358,
      "grad_norm": 0.49052897095680237,
      "learning_rate": 5.239334501444543e-06,
      "loss": 0.4197,
      "step": 32364
    },
    {
      "epoch": 1.48524620256069,
      "grad_norm": 0.461058109998703,
      "learning_rate": 5.239089594303056e-06,
      "loss": 0.3499,
      "step": 32365
    },
    {
      "epoch": 1.4852920930659446,
      "grad_norm": 0.49561581015586853,
      "learning_rate": 5.238844686586635e-06,
      "loss": 0.3627,
      "step": 32366
    },
    {
      "epoch": 1.485337983571199,
      "grad_norm": 0.44844287633895874,
      "learning_rate": 5.238599778295865e-06,
      "loss": 0.3284,
      "step": 32367
    },
    {
      "epoch": 1.4853838740764536,
      "grad_norm": 0.5191134810447693,
      "learning_rate": 5.23835486943134e-06,
      "loss": 0.349,
      "step": 32368
    },
    {
      "epoch": 1.485429764581708,
      "grad_norm": 0.46778756380081177,
      "learning_rate": 5.238109959993646e-06,
      "loss": 0.3685,
      "step": 32369
    },
    {
      "epoch": 1.4854756550869626,
      "grad_norm": 0.517098605632782,
      "learning_rate": 5.237865049983374e-06,
      "loss": 0.4609,
      "step": 32370
    },
    {
      "epoch": 1.485521545592217,
      "grad_norm": 0.4473532736301422,
      "learning_rate": 5.237620139401111e-06,
      "loss": 0.3454,
      "step": 32371
    },
    {
      "epoch": 1.4855674360974715,
      "grad_norm": 0.4636293649673462,
      "learning_rate": 5.237375228247446e-06,
      "loss": 0.3634,
      "step": 32372
    },
    {
      "epoch": 1.4856133266027258,
      "grad_norm": 0.4639020562171936,
      "learning_rate": 5.237130316522967e-06,
      "loss": 0.3723,
      "step": 32373
    },
    {
      "epoch": 1.4856592171079803,
      "grad_norm": 0.5386571288108826,
      "learning_rate": 5.236885404228267e-06,
      "loss": 0.4864,
      "step": 32374
    },
    {
      "epoch": 1.4857051076132348,
      "grad_norm": 0.509666919708252,
      "learning_rate": 5.2366404913639315e-06,
      "loss": 0.4246,
      "step": 32375
    },
    {
      "epoch": 1.4857509981184893,
      "grad_norm": 0.4744151830673218,
      "learning_rate": 5.2363955779305495e-06,
      "loss": 0.3399,
      "step": 32376
    },
    {
      "epoch": 1.4857968886237438,
      "grad_norm": 0.478560209274292,
      "learning_rate": 5.2361506639287126e-06,
      "loss": 0.4371,
      "step": 32377
    },
    {
      "epoch": 1.485842779128998,
      "grad_norm": 0.44602057337760925,
      "learning_rate": 5.235905749359006e-06,
      "loss": 0.3476,
      "step": 32378
    },
    {
      "epoch": 1.4858886696342526,
      "grad_norm": 0.49468302726745605,
      "learning_rate": 5.235660834222021e-06,
      "loss": 0.3869,
      "step": 32379
    },
    {
      "epoch": 1.485934560139507,
      "grad_norm": 0.4631974399089813,
      "learning_rate": 5.235415918518346e-06,
      "loss": 0.3551,
      "step": 32380
    },
    {
      "epoch": 1.4859804506447616,
      "grad_norm": 0.4616166055202484,
      "learning_rate": 5.235171002248571e-06,
      "loss": 0.3298,
      "step": 32381
    },
    {
      "epoch": 1.486026341150016,
      "grad_norm": 0.4785248637199402,
      "learning_rate": 5.234926085413284e-06,
      "loss": 0.3504,
      "step": 32382
    },
    {
      "epoch": 1.4860722316552706,
      "grad_norm": 0.4794504940509796,
      "learning_rate": 5.234681168013073e-06,
      "loss": 0.4118,
      "step": 32383
    },
    {
      "epoch": 1.486118122160525,
      "grad_norm": 0.49514123797416687,
      "learning_rate": 5.234436250048528e-06,
      "loss": 0.3695,
      "step": 32384
    },
    {
      "epoch": 1.4861640126657796,
      "grad_norm": 0.462048202753067,
      "learning_rate": 5.234191331520237e-06,
      "loss": 0.3685,
      "step": 32385
    },
    {
      "epoch": 1.4862099031710339,
      "grad_norm": 0.4800395667552948,
      "learning_rate": 5.233946412428791e-06,
      "loss": 0.387,
      "step": 32386
    },
    {
      "epoch": 1.4862557936762884,
      "grad_norm": 0.7294582724571228,
      "learning_rate": 5.233701492774778e-06,
      "loss": 0.3456,
      "step": 32387
    },
    {
      "epoch": 1.4863016841815428,
      "grad_norm": 0.45245125889778137,
      "learning_rate": 5.233456572558786e-06,
      "loss": 0.3244,
      "step": 32388
    },
    {
      "epoch": 1.4863475746867973,
      "grad_norm": 0.4890178442001343,
      "learning_rate": 5.233211651781405e-06,
      "loss": 0.4162,
      "step": 32389
    },
    {
      "epoch": 1.4863934651920518,
      "grad_norm": 0.45102447271347046,
      "learning_rate": 5.232966730443222e-06,
      "loss": 0.2726,
      "step": 32390
    },
    {
      "epoch": 1.4864393556973061,
      "grad_norm": 0.4705100357532501,
      "learning_rate": 5.23272180854483e-06,
      "loss": 0.3299,
      "step": 32391
    },
    {
      "epoch": 1.4864852462025606,
      "grad_norm": 0.5006636381149292,
      "learning_rate": 5.232476886086815e-06,
      "loss": 0.3854,
      "step": 32392
    },
    {
      "epoch": 1.4865311367078151,
      "grad_norm": 0.48193633556365967,
      "learning_rate": 5.232231963069765e-06,
      "loss": 0.3778,
      "step": 32393
    },
    {
      "epoch": 1.4865770272130696,
      "grad_norm": 0.484199583530426,
      "learning_rate": 5.231987039494272e-06,
      "loss": 0.4163,
      "step": 32394
    },
    {
      "epoch": 1.486622917718324,
      "grad_norm": 0.5995803475379944,
      "learning_rate": 5.231742115360921e-06,
      "loss": 0.3724,
      "step": 32395
    },
    {
      "epoch": 1.4866688082235786,
      "grad_norm": 0.46093231439590454,
      "learning_rate": 5.231497190670305e-06,
      "loss": 0.3119,
      "step": 32396
    },
    {
      "epoch": 1.486714698728833,
      "grad_norm": 0.4827044904232025,
      "learning_rate": 5.231252265423011e-06,
      "loss": 0.3728,
      "step": 32397
    },
    {
      "epoch": 1.4867605892340876,
      "grad_norm": 0.47655683755874634,
      "learning_rate": 5.231007339619629e-06,
      "loss": 0.3582,
      "step": 32398
    },
    {
      "epoch": 1.4868064797393419,
      "grad_norm": 0.474122017621994,
      "learning_rate": 5.230762413260747e-06,
      "loss": 0.3698,
      "step": 32399
    },
    {
      "epoch": 1.4868523702445964,
      "grad_norm": 0.4862482249736786,
      "learning_rate": 5.230517486346954e-06,
      "loss": 0.3652,
      "step": 32400
    },
    {
      "epoch": 1.4868982607498509,
      "grad_norm": 0.5885858535766602,
      "learning_rate": 5.230272558878839e-06,
      "loss": 0.3548,
      "step": 32401
    },
    {
      "epoch": 1.4869441512551054,
      "grad_norm": 0.5089755058288574,
      "learning_rate": 5.230027630856992e-06,
      "loss": 0.422,
      "step": 32402
    },
    {
      "epoch": 1.4869900417603596,
      "grad_norm": 0.4889802038669586,
      "learning_rate": 5.2297827022820014e-06,
      "loss": 0.3282,
      "step": 32403
    },
    {
      "epoch": 1.4870359322656141,
      "grad_norm": 0.4968697130680084,
      "learning_rate": 5.229537773154454e-06,
      "loss": 0.4023,
      "step": 32404
    },
    {
      "epoch": 1.4870818227708686,
      "grad_norm": 0.4942907691001892,
      "learning_rate": 5.229292843474942e-06,
      "loss": 0.4031,
      "step": 32405
    },
    {
      "epoch": 1.4871277132761231,
      "grad_norm": 0.4551313519477844,
      "learning_rate": 5.229047913244052e-06,
      "loss": 0.3495,
      "step": 32406
    },
    {
      "epoch": 1.4871736037813776,
      "grad_norm": 0.43653157353401184,
      "learning_rate": 5.228802982462374e-06,
      "loss": 0.3077,
      "step": 32407
    },
    {
      "epoch": 1.4872194942866321,
      "grad_norm": 0.48374438285827637,
      "learning_rate": 5.228558051130499e-06,
      "loss": 0.3829,
      "step": 32408
    },
    {
      "epoch": 1.4872653847918866,
      "grad_norm": 0.5199885368347168,
      "learning_rate": 5.2283131192490135e-06,
      "loss": 0.411,
      "step": 32409
    },
    {
      "epoch": 1.4873112752971411,
      "grad_norm": 0.4609754681587219,
      "learning_rate": 5.228068186818505e-06,
      "loss": 0.3636,
      "step": 32410
    },
    {
      "epoch": 1.4873571658023954,
      "grad_norm": 0.4843488335609436,
      "learning_rate": 5.227823253839568e-06,
      "loss": 0.388,
      "step": 32411
    },
    {
      "epoch": 1.48740305630765,
      "grad_norm": 0.4751400351524353,
      "learning_rate": 5.227578320312785e-06,
      "loss": 0.3866,
      "step": 32412
    },
    {
      "epoch": 1.4874489468129044,
      "grad_norm": 0.4935431480407715,
      "learning_rate": 5.227333386238748e-06,
      "loss": 0.3811,
      "step": 32413
    },
    {
      "epoch": 1.487494837318159,
      "grad_norm": 0.42536357045173645,
      "learning_rate": 5.227088451618049e-06,
      "loss": 0.2898,
      "step": 32414
    },
    {
      "epoch": 1.4875407278234134,
      "grad_norm": 0.47140803933143616,
      "learning_rate": 5.226843516451271e-06,
      "loss": 0.3941,
      "step": 32415
    },
    {
      "epoch": 1.4875866183286677,
      "grad_norm": 0.4453476369380951,
      "learning_rate": 5.226598580739006e-06,
      "loss": 0.3107,
      "step": 32416
    },
    {
      "epoch": 1.4876325088339222,
      "grad_norm": 0.5224695801734924,
      "learning_rate": 5.2263536444818445e-06,
      "loss": 0.413,
      "step": 32417
    },
    {
      "epoch": 1.4876783993391767,
      "grad_norm": 0.46678829193115234,
      "learning_rate": 5.226108707680372e-06,
      "loss": 0.3622,
      "step": 32418
    },
    {
      "epoch": 1.4877242898444312,
      "grad_norm": 0.5104742050170898,
      "learning_rate": 5.225863770335181e-06,
      "loss": 0.429,
      "step": 32419
    },
    {
      "epoch": 1.4877701803496857,
      "grad_norm": 0.48558154702186584,
      "learning_rate": 5.225618832446859e-06,
      "loss": 0.3427,
      "step": 32420
    },
    {
      "epoch": 1.4878160708549402,
      "grad_norm": 0.4677046537399292,
      "learning_rate": 5.225373894015994e-06,
      "loss": 0.3886,
      "step": 32421
    },
    {
      "epoch": 1.4878619613601947,
      "grad_norm": 0.4536486268043518,
      "learning_rate": 5.225128955043176e-06,
      "loss": 0.3307,
      "step": 32422
    },
    {
      "epoch": 1.4879078518654492,
      "grad_norm": 0.45813390612602234,
      "learning_rate": 5.224884015528994e-06,
      "loss": 0.3514,
      "step": 32423
    },
    {
      "epoch": 1.4879537423707034,
      "grad_norm": 0.46142256259918213,
      "learning_rate": 5.224639075474036e-06,
      "loss": 0.3857,
      "step": 32424
    },
    {
      "epoch": 1.487999632875958,
      "grad_norm": 0.471737265586853,
      "learning_rate": 5.224394134878895e-06,
      "loss": 0.3903,
      "step": 32425
    },
    {
      "epoch": 1.4880455233812124,
      "grad_norm": 0.46197670698165894,
      "learning_rate": 5.224149193744155e-06,
      "loss": 0.3494,
      "step": 32426
    },
    {
      "epoch": 1.488091413886467,
      "grad_norm": 0.4563344717025757,
      "learning_rate": 5.223904252070406e-06,
      "loss": 0.4207,
      "step": 32427
    },
    {
      "epoch": 1.4881373043917214,
      "grad_norm": 0.46322908997535706,
      "learning_rate": 5.2236593098582385e-06,
      "loss": 0.3455,
      "step": 32428
    },
    {
      "epoch": 1.4881831948969757,
      "grad_norm": 0.4734077453613281,
      "learning_rate": 5.2234143671082415e-06,
      "loss": 0.3605,
      "step": 32429
    },
    {
      "epoch": 1.4882290854022302,
      "grad_norm": 0.4871661067008972,
      "learning_rate": 5.223169423821004e-06,
      "loss": 0.3812,
      "step": 32430
    },
    {
      "epoch": 1.4882749759074847,
      "grad_norm": 0.4972514808177948,
      "learning_rate": 5.222924479997114e-06,
      "loss": 0.4036,
      "step": 32431
    },
    {
      "epoch": 1.4883208664127392,
      "grad_norm": 0.464763879776001,
      "learning_rate": 5.222679535637159e-06,
      "loss": 0.3558,
      "step": 32432
    },
    {
      "epoch": 1.4883667569179937,
      "grad_norm": 0.472945898771286,
      "learning_rate": 5.222434590741733e-06,
      "loss": 0.3378,
      "step": 32433
    },
    {
      "epoch": 1.4884126474232482,
      "grad_norm": 0.468758761882782,
      "learning_rate": 5.22218964531142e-06,
      "loss": 0.3347,
      "step": 32434
    },
    {
      "epoch": 1.4884585379285027,
      "grad_norm": 0.5076015591621399,
      "learning_rate": 5.221944699346811e-06,
      "loss": 0.4377,
      "step": 32435
    },
    {
      "epoch": 1.4885044284337572,
      "grad_norm": 0.5030063986778259,
      "learning_rate": 5.221699752848498e-06,
      "loss": 0.434,
      "step": 32436
    },
    {
      "epoch": 1.4885503189390115,
      "grad_norm": 0.5112724900245667,
      "learning_rate": 5.221454805817063e-06,
      "loss": 0.4284,
      "step": 32437
    },
    {
      "epoch": 1.488596209444266,
      "grad_norm": 0.4544411599636078,
      "learning_rate": 5.221209858253101e-06,
      "loss": 0.3338,
      "step": 32438
    },
    {
      "epoch": 1.4886420999495205,
      "grad_norm": 0.566962718963623,
      "learning_rate": 5.2209649101572005e-06,
      "loss": 0.4587,
      "step": 32439
    },
    {
      "epoch": 1.488687990454775,
      "grad_norm": 0.4672989249229431,
      "learning_rate": 5.220719961529947e-06,
      "loss": 0.3463,
      "step": 32440
    },
    {
      "epoch": 1.4887338809600292,
      "grad_norm": 0.4605965316295624,
      "learning_rate": 5.220475012371932e-06,
      "loss": 0.3104,
      "step": 32441
    },
    {
      "epoch": 1.4887797714652837,
      "grad_norm": 0.42739707231521606,
      "learning_rate": 5.220230062683746e-06,
      "loss": 0.3049,
      "step": 32442
    },
    {
      "epoch": 1.4888256619705382,
      "grad_norm": 0.4944610893726349,
      "learning_rate": 5.2199851124659725e-06,
      "loss": 0.3941,
      "step": 32443
    },
    {
      "epoch": 1.4888715524757927,
      "grad_norm": 0.46612539887428284,
      "learning_rate": 5.219740161719207e-06,
      "loss": 0.3496,
      "step": 32444
    },
    {
      "epoch": 1.4889174429810472,
      "grad_norm": 0.4866597354412079,
      "learning_rate": 5.219495210444036e-06,
      "loss": 0.36,
      "step": 32445
    },
    {
      "epoch": 1.4889633334863017,
      "grad_norm": 0.4737932085990906,
      "learning_rate": 5.219250258641046e-06,
      "loss": 0.3687,
      "step": 32446
    },
    {
      "epoch": 1.4890092239915562,
      "grad_norm": 0.4782184958457947,
      "learning_rate": 5.2190053063108305e-06,
      "loss": 0.3499,
      "step": 32447
    },
    {
      "epoch": 1.4890551144968107,
      "grad_norm": 0.4779413938522339,
      "learning_rate": 5.218760353453975e-06,
      "loss": 0.3313,
      "step": 32448
    },
    {
      "epoch": 1.489101005002065,
      "grad_norm": 0.4625546634197235,
      "learning_rate": 5.218515400071071e-06,
      "loss": 0.3323,
      "step": 32449
    },
    {
      "epoch": 1.4891468955073195,
      "grad_norm": 0.4414535164833069,
      "learning_rate": 5.218270446162707e-06,
      "loss": 0.305,
      "step": 32450
    },
    {
      "epoch": 1.489192786012574,
      "grad_norm": 0.45431190729141235,
      "learning_rate": 5.218025491729469e-06,
      "loss": 0.3032,
      "step": 32451
    },
    {
      "epoch": 1.4892386765178285,
      "grad_norm": 0.4497077167034149,
      "learning_rate": 5.21778053677195e-06,
      "loss": 0.3209,
      "step": 32452
    },
    {
      "epoch": 1.489284567023083,
      "grad_norm": 0.4962162971496582,
      "learning_rate": 5.217535581290739e-06,
      "loss": 0.3764,
      "step": 32453
    },
    {
      "epoch": 1.4893304575283373,
      "grad_norm": 0.4543689489364624,
      "learning_rate": 5.217290625286421e-06,
      "loss": 0.3096,
      "step": 32454
    },
    {
      "epoch": 1.4893763480335918,
      "grad_norm": 0.46530142426490784,
      "learning_rate": 5.217045668759588e-06,
      "loss": 0.3354,
      "step": 32455
    },
    {
      "epoch": 1.4894222385388463,
      "grad_norm": 0.4765910804271698,
      "learning_rate": 5.2168007117108295e-06,
      "loss": 0.3676,
      "step": 32456
    },
    {
      "epoch": 1.4894681290441008,
      "grad_norm": 0.48544952273368835,
      "learning_rate": 5.216555754140734e-06,
      "loss": 0.3254,
      "step": 32457
    },
    {
      "epoch": 1.4895140195493553,
      "grad_norm": 0.4873560070991516,
      "learning_rate": 5.21631079604989e-06,
      "loss": 0.3587,
      "step": 32458
    },
    {
      "epoch": 1.4895599100546097,
      "grad_norm": 0.4649731516838074,
      "learning_rate": 5.216065837438887e-06,
      "loss": 0.3481,
      "step": 32459
    },
    {
      "epoch": 1.4896058005598642,
      "grad_norm": 0.4719788432121277,
      "learning_rate": 5.215820878308313e-06,
      "loss": 0.3656,
      "step": 32460
    },
    {
      "epoch": 1.4896516910651187,
      "grad_norm": 0.4643266499042511,
      "learning_rate": 5.215575918658758e-06,
      "loss": 0.369,
      "step": 32461
    },
    {
      "epoch": 1.489697581570373,
      "grad_norm": 0.5103459358215332,
      "learning_rate": 5.215330958490812e-06,
      "loss": 0.4008,
      "step": 32462
    },
    {
      "epoch": 1.4897434720756275,
      "grad_norm": 0.4633001685142517,
      "learning_rate": 5.215085997805063e-06,
      "loss": 0.3829,
      "step": 32463
    },
    {
      "epoch": 1.489789362580882,
      "grad_norm": 0.47449991106987,
      "learning_rate": 5.214841036602099e-06,
      "loss": 0.3571,
      "step": 32464
    },
    {
      "epoch": 1.4898352530861365,
      "grad_norm": 0.4497971534729004,
      "learning_rate": 5.214596074882511e-06,
      "loss": 0.3309,
      "step": 32465
    },
    {
      "epoch": 1.489881143591391,
      "grad_norm": 0.4497924745082855,
      "learning_rate": 5.214351112646886e-06,
      "loss": 0.3567,
      "step": 32466
    },
    {
      "epoch": 1.4899270340966453,
      "grad_norm": 0.47269895672798157,
      "learning_rate": 5.214106149895814e-06,
      "loss": 0.3398,
      "step": 32467
    },
    {
      "epoch": 1.4899729246018998,
      "grad_norm": 0.5033291578292847,
      "learning_rate": 5.213861186629886e-06,
      "loss": 0.3666,
      "step": 32468
    },
    {
      "epoch": 1.4900188151071543,
      "grad_norm": 0.47006574273109436,
      "learning_rate": 5.2136162228496885e-06,
      "loss": 0.3436,
      "step": 32469
    },
    {
      "epoch": 1.4900647056124088,
      "grad_norm": 0.47165316343307495,
      "learning_rate": 5.213371258555812e-06,
      "loss": 0.3542,
      "step": 32470
    },
    {
      "epoch": 1.4901105961176633,
      "grad_norm": 0.47586381435394287,
      "learning_rate": 5.213126293748844e-06,
      "loss": 0.3584,
      "step": 32471
    },
    {
      "epoch": 1.4901564866229178,
      "grad_norm": 0.48044487833976746,
      "learning_rate": 5.212881328429374e-06,
      "loss": 0.3645,
      "step": 32472
    },
    {
      "epoch": 1.4902023771281723,
      "grad_norm": 0.47151049971580505,
      "learning_rate": 5.212636362597994e-06,
      "loss": 0.3601,
      "step": 32473
    },
    {
      "epoch": 1.4902482676334268,
      "grad_norm": 0.48227182030677795,
      "learning_rate": 5.212391396255289e-06,
      "loss": 0.3594,
      "step": 32474
    },
    {
      "epoch": 1.490294158138681,
      "grad_norm": 0.4740387797355652,
      "learning_rate": 5.21214642940185e-06,
      "loss": 0.3979,
      "step": 32475
    },
    {
      "epoch": 1.4903400486439355,
      "grad_norm": 0.46698257327079773,
      "learning_rate": 5.211901462038266e-06,
      "loss": 0.3795,
      "step": 32476
    },
    {
      "epoch": 1.49038593914919,
      "grad_norm": 0.5277775526046753,
      "learning_rate": 5.211656494165125e-06,
      "loss": 0.378,
      "step": 32477
    },
    {
      "epoch": 1.4904318296544445,
      "grad_norm": 0.4319857656955719,
      "learning_rate": 5.211411525783017e-06,
      "loss": 0.3053,
      "step": 32478
    },
    {
      "epoch": 1.490477720159699,
      "grad_norm": 0.4771716594696045,
      "learning_rate": 5.211166556892532e-06,
      "loss": 0.3962,
      "step": 32479
    },
    {
      "epoch": 1.4905236106649533,
      "grad_norm": 0.48755839467048645,
      "learning_rate": 5.210921587494257e-06,
      "loss": 0.3585,
      "step": 32480
    },
    {
      "epoch": 1.4905695011702078,
      "grad_norm": 0.42155900597572327,
      "learning_rate": 5.210676617588782e-06,
      "loss": 0.2816,
      "step": 32481
    },
    {
      "epoch": 1.4906153916754623,
      "grad_norm": 0.4692254364490509,
      "learning_rate": 5.210431647176698e-06,
      "loss": 0.383,
      "step": 32482
    },
    {
      "epoch": 1.4906612821807168,
      "grad_norm": 0.45834192633628845,
      "learning_rate": 5.210186676258591e-06,
      "loss": 0.3391,
      "step": 32483
    },
    {
      "epoch": 1.4907071726859713,
      "grad_norm": 0.46294498443603516,
      "learning_rate": 5.209941704835052e-06,
      "loss": 0.334,
      "step": 32484
    },
    {
      "epoch": 1.4907530631912258,
      "grad_norm": 0.4737560451030731,
      "learning_rate": 5.20969673290667e-06,
      "loss": 0.3317,
      "step": 32485
    },
    {
      "epoch": 1.4907989536964803,
      "grad_norm": 0.45365896821022034,
      "learning_rate": 5.209451760474031e-06,
      "loss": 0.3242,
      "step": 32486
    },
    {
      "epoch": 1.4908448442017348,
      "grad_norm": 0.47143787145614624,
      "learning_rate": 5.20920678753773e-06,
      "loss": 0.3566,
      "step": 32487
    },
    {
      "epoch": 1.490890734706989,
      "grad_norm": 0.49440547823905945,
      "learning_rate": 5.20896181409835e-06,
      "loss": 0.34,
      "step": 32488
    },
    {
      "epoch": 1.4909366252122436,
      "grad_norm": 0.53519207239151,
      "learning_rate": 5.2087168401564824e-06,
      "loss": 0.4628,
      "step": 32489
    },
    {
      "epoch": 1.490982515717498,
      "grad_norm": 0.5015696883201599,
      "learning_rate": 5.208471865712719e-06,
      "loss": 0.4151,
      "step": 32490
    },
    {
      "epoch": 1.4910284062227526,
      "grad_norm": 0.4778095483779907,
      "learning_rate": 5.208226890767646e-06,
      "loss": 0.3479,
      "step": 32491
    },
    {
      "epoch": 1.4910742967280068,
      "grad_norm": 0.4882211685180664,
      "learning_rate": 5.207981915321853e-06,
      "loss": 0.3874,
      "step": 32492
    },
    {
      "epoch": 1.4911201872332613,
      "grad_norm": 0.5058813691139221,
      "learning_rate": 5.207736939375928e-06,
      "loss": 0.3914,
      "step": 32493
    },
    {
      "epoch": 1.4911660777385158,
      "grad_norm": 0.449704110622406,
      "learning_rate": 5.207491962930462e-06,
      "loss": 0.3366,
      "step": 32494
    },
    {
      "epoch": 1.4912119682437703,
      "grad_norm": 0.4503457546234131,
      "learning_rate": 5.207246985986043e-06,
      "loss": 0.3591,
      "step": 32495
    },
    {
      "epoch": 1.4912578587490248,
      "grad_norm": 0.4372318983078003,
      "learning_rate": 5.2070020085432615e-06,
      "loss": 0.3264,
      "step": 32496
    },
    {
      "epoch": 1.4913037492542793,
      "grad_norm": 0.460175484418869,
      "learning_rate": 5.206757030602705e-06,
      "loss": 0.333,
      "step": 32497
    },
    {
      "epoch": 1.4913496397595338,
      "grad_norm": 0.4229927659034729,
      "learning_rate": 5.2065120521649615e-06,
      "loss": 0.3136,
      "step": 32498
    },
    {
      "epoch": 1.4913955302647883,
      "grad_norm": 0.48736435174942017,
      "learning_rate": 5.206267073230624e-06,
      "loss": 0.4042,
      "step": 32499
    },
    {
      "epoch": 1.4914414207700426,
      "grad_norm": 0.4254434108734131,
      "learning_rate": 5.206022093800277e-06,
      "loss": 0.3126,
      "step": 32500
    },
    {
      "epoch": 1.491487311275297,
      "grad_norm": 0.46921733021736145,
      "learning_rate": 5.205777113874514e-06,
      "loss": 0.4057,
      "step": 32501
    },
    {
      "epoch": 1.4915332017805516,
      "grad_norm": 0.47904855012893677,
      "learning_rate": 5.205532133453922e-06,
      "loss": 0.386,
      "step": 32502
    },
    {
      "epoch": 1.491579092285806,
      "grad_norm": 0.45980754494667053,
      "learning_rate": 5.205287152539088e-06,
      "loss": 0.3579,
      "step": 32503
    },
    {
      "epoch": 1.4916249827910606,
      "grad_norm": 0.5059465169906616,
      "learning_rate": 5.205042171130606e-06,
      "loss": 0.4508,
      "step": 32504
    },
    {
      "epoch": 1.4916708732963149,
      "grad_norm": 0.4449034035205841,
      "learning_rate": 5.204797189229059e-06,
      "loss": 0.3738,
      "step": 32505
    },
    {
      "epoch": 1.4917167638015694,
      "grad_norm": 0.48877862095832825,
      "learning_rate": 5.204552206835042e-06,
      "loss": 0.3965,
      "step": 32506
    },
    {
      "epoch": 1.4917626543068239,
      "grad_norm": 0.474043071269989,
      "learning_rate": 5.204307223949141e-06,
      "loss": 0.3383,
      "step": 32507
    },
    {
      "epoch": 1.4918085448120784,
      "grad_norm": 0.47350358963012695,
      "learning_rate": 5.2040622405719456e-06,
      "loss": 0.3897,
      "step": 32508
    },
    {
      "epoch": 1.4918544353173329,
      "grad_norm": 0.459407776594162,
      "learning_rate": 5.203817256704044e-06,
      "loss": 0.3215,
      "step": 32509
    },
    {
      "epoch": 1.4919003258225874,
      "grad_norm": 0.4775444269180298,
      "learning_rate": 5.203572272346027e-06,
      "loss": 0.3631,
      "step": 32510
    },
    {
      "epoch": 1.4919462163278419,
      "grad_norm": 0.44606396555900574,
      "learning_rate": 5.203327287498482e-06,
      "loss": 0.342,
      "step": 32511
    },
    {
      "epoch": 1.4919921068330964,
      "grad_norm": 0.45047304034233093,
      "learning_rate": 5.203082302162e-06,
      "loss": 0.3272,
      "step": 32512
    },
    {
      "epoch": 1.4920379973383506,
      "grad_norm": 0.5457999110221863,
      "learning_rate": 5.202837316337169e-06,
      "loss": 0.3529,
      "step": 32513
    },
    {
      "epoch": 1.4920838878436051,
      "grad_norm": 0.4722496569156647,
      "learning_rate": 5.202592330024578e-06,
      "loss": 0.4096,
      "step": 32514
    },
    {
      "epoch": 1.4921297783488596,
      "grad_norm": 0.5037813782691956,
      "learning_rate": 5.202347343224817e-06,
      "loss": 0.4141,
      "step": 32515
    },
    {
      "epoch": 1.4921756688541141,
      "grad_norm": 0.5722838640213013,
      "learning_rate": 5.2021023559384734e-06,
      "loss": 0.4265,
      "step": 32516
    },
    {
      "epoch": 1.4922215593593686,
      "grad_norm": 0.48118695616722107,
      "learning_rate": 5.201857368166139e-06,
      "loss": 0.3208,
      "step": 32517
    },
    {
      "epoch": 1.492267449864623,
      "grad_norm": 0.48233741521835327,
      "learning_rate": 5.201612379908401e-06,
      "loss": 0.3852,
      "step": 32518
    },
    {
      "epoch": 1.4923133403698774,
      "grad_norm": 0.4746929705142975,
      "learning_rate": 5.201367391165847e-06,
      "loss": 0.3612,
      "step": 32519
    },
    {
      "epoch": 1.492359230875132,
      "grad_norm": 0.4887195825576782,
      "learning_rate": 5.201122401939069e-06,
      "loss": 0.4005,
      "step": 32520
    },
    {
      "epoch": 1.4924051213803864,
      "grad_norm": 0.4761020541191101,
      "learning_rate": 5.200877412228655e-06,
      "loss": 0.364,
      "step": 32521
    },
    {
      "epoch": 1.4924510118856409,
      "grad_norm": 0.4892801344394684,
      "learning_rate": 5.200632422035194e-06,
      "loss": 0.3459,
      "step": 32522
    },
    {
      "epoch": 1.4924969023908954,
      "grad_norm": 0.4638475477695465,
      "learning_rate": 5.2003874313592775e-06,
      "loss": 0.3397,
      "step": 32523
    },
    {
      "epoch": 1.4925427928961499,
      "grad_norm": 0.48983097076416016,
      "learning_rate": 5.200142440201491e-06,
      "loss": 0.3717,
      "step": 32524
    },
    {
      "epoch": 1.4925886834014044,
      "grad_norm": 0.4548947215080261,
      "learning_rate": 5.199897448562424e-06,
      "loss": 0.3735,
      "step": 32525
    },
    {
      "epoch": 1.4926345739066587,
      "grad_norm": 0.4728935956954956,
      "learning_rate": 5.199652456442668e-06,
      "loss": 0.3753,
      "step": 32526
    },
    {
      "epoch": 1.4926804644119132,
      "grad_norm": 0.5063821077346802,
      "learning_rate": 5.199407463842811e-06,
      "loss": 0.4168,
      "step": 32527
    },
    {
      "epoch": 1.4927263549171677,
      "grad_norm": 0.45788809657096863,
      "learning_rate": 5.1991624707634415e-06,
      "loss": 0.3297,
      "step": 32528
    },
    {
      "epoch": 1.4927722454224222,
      "grad_norm": 0.4555186331272125,
      "learning_rate": 5.198917477205149e-06,
      "loss": 0.3524,
      "step": 32529
    },
    {
      "epoch": 1.4928181359276764,
      "grad_norm": 0.4603189527988434,
      "learning_rate": 5.198672483168523e-06,
      "loss": 0.2956,
      "step": 32530
    },
    {
      "epoch": 1.492864026432931,
      "grad_norm": 0.4926038384437561,
      "learning_rate": 5.1984274886541515e-06,
      "loss": 0.3459,
      "step": 32531
    },
    {
      "epoch": 1.4929099169381854,
      "grad_norm": 0.46656593680381775,
      "learning_rate": 5.198182493662627e-06,
      "loss": 0.3778,
      "step": 32532
    },
    {
      "epoch": 1.49295580744344,
      "grad_norm": 0.4872576594352722,
      "learning_rate": 5.197937498194533e-06,
      "loss": 0.3497,
      "step": 32533
    },
    {
      "epoch": 1.4930016979486944,
      "grad_norm": 0.48604539036750793,
      "learning_rate": 5.197692502250465e-06,
      "loss": 0.3485,
      "step": 32534
    },
    {
      "epoch": 1.493047588453949,
      "grad_norm": 0.47751009464263916,
      "learning_rate": 5.197447505831007e-06,
      "loss": 0.3476,
      "step": 32535
    },
    {
      "epoch": 1.4930934789592034,
      "grad_norm": 0.46607398986816406,
      "learning_rate": 5.197202508936749e-06,
      "loss": 0.3471,
      "step": 32536
    },
    {
      "epoch": 1.493139369464458,
      "grad_norm": 0.4623022675514221,
      "learning_rate": 5.196957511568282e-06,
      "loss": 0.3566,
      "step": 32537
    },
    {
      "epoch": 1.4931852599697122,
      "grad_norm": 0.4550822377204895,
      "learning_rate": 5.196712513726195e-06,
      "loss": 0.3157,
      "step": 32538
    },
    {
      "epoch": 1.4932311504749667,
      "grad_norm": 0.541007399559021,
      "learning_rate": 5.196467515411078e-06,
      "loss": 0.461,
      "step": 32539
    },
    {
      "epoch": 1.4932770409802212,
      "grad_norm": 0.43232396245002747,
      "learning_rate": 5.196222516623516e-06,
      "loss": 0.3283,
      "step": 32540
    },
    {
      "epoch": 1.4933229314854757,
      "grad_norm": 0.4820578694343567,
      "learning_rate": 5.195977517364103e-06,
      "loss": 0.3694,
      "step": 32541
    },
    {
      "epoch": 1.4933688219907302,
      "grad_norm": 0.45731353759765625,
      "learning_rate": 5.1957325176334224e-06,
      "loss": 0.3422,
      "step": 32542
    },
    {
      "epoch": 1.4934147124959845,
      "grad_norm": 0.5556846261024475,
      "learning_rate": 5.19548751743207e-06,
      "loss": 0.4527,
      "step": 32543
    },
    {
      "epoch": 1.493460603001239,
      "grad_norm": 0.46742039918899536,
      "learning_rate": 5.19524251676063e-06,
      "loss": 0.4024,
      "step": 32544
    },
    {
      "epoch": 1.4935064935064934,
      "grad_norm": 0.44558605551719666,
      "learning_rate": 5.194997515619695e-06,
      "loss": 0.2982,
      "step": 32545
    },
    {
      "epoch": 1.493552384011748,
      "grad_norm": 0.49691998958587646,
      "learning_rate": 5.194752514009852e-06,
      "loss": 0.4825,
      "step": 32546
    },
    {
      "epoch": 1.4935982745170024,
      "grad_norm": 0.4831225275993347,
      "learning_rate": 5.19450751193169e-06,
      "loss": 0.4455,
      "step": 32547
    },
    {
      "epoch": 1.493644165022257,
      "grad_norm": 0.4995240271091461,
      "learning_rate": 5.194262509385799e-06,
      "loss": 0.3554,
      "step": 32548
    },
    {
      "epoch": 1.4936900555275114,
      "grad_norm": 0.4603930115699768,
      "learning_rate": 5.194017506372769e-06,
      "loss": 0.3691,
      "step": 32549
    },
    {
      "epoch": 1.493735946032766,
      "grad_norm": 0.5147192478179932,
      "learning_rate": 5.193772502893188e-06,
      "loss": 0.4557,
      "step": 32550
    },
    {
      "epoch": 1.4937818365380202,
      "grad_norm": 0.49076563119888306,
      "learning_rate": 5.1935274989476434e-06,
      "loss": 0.3789,
      "step": 32551
    },
    {
      "epoch": 1.4938277270432747,
      "grad_norm": 0.48677903413772583,
      "learning_rate": 5.1932824945367285e-06,
      "loss": 0.3984,
      "step": 32552
    },
    {
      "epoch": 1.4938736175485292,
      "grad_norm": 0.4535476863384247,
      "learning_rate": 5.193037489661029e-06,
      "loss": 0.317,
      "step": 32553
    },
    {
      "epoch": 1.4939195080537837,
      "grad_norm": 0.4989411234855652,
      "learning_rate": 5.192792484321134e-06,
      "loss": 0.3981,
      "step": 32554
    },
    {
      "epoch": 1.4939653985590382,
      "grad_norm": 0.47385919094085693,
      "learning_rate": 5.1925474785176366e-06,
      "loss": 0.3963,
      "step": 32555
    },
    {
      "epoch": 1.4940112890642925,
      "grad_norm": 0.46761974692344666,
      "learning_rate": 5.192302472251123e-06,
      "loss": 0.3861,
      "step": 32556
    },
    {
      "epoch": 1.494057179569547,
      "grad_norm": 0.4709824025630951,
      "learning_rate": 5.192057465522181e-06,
      "loss": 0.3264,
      "step": 32557
    },
    {
      "epoch": 1.4941030700748015,
      "grad_norm": 0.42888540029525757,
      "learning_rate": 5.1918124583314026e-06,
      "loss": 0.2885,
      "step": 32558
    },
    {
      "epoch": 1.494148960580056,
      "grad_norm": 0.46583297848701477,
      "learning_rate": 5.1915674506793745e-06,
      "loss": 0.3388,
      "step": 32559
    },
    {
      "epoch": 1.4941948510853105,
      "grad_norm": 0.45026281476020813,
      "learning_rate": 5.191322442566688e-06,
      "loss": 0.3339,
      "step": 32560
    },
    {
      "epoch": 1.494240741590565,
      "grad_norm": 0.46284499764442444,
      "learning_rate": 5.191077433993933e-06,
      "loss": 0.374,
      "step": 32561
    },
    {
      "epoch": 1.4942866320958195,
      "grad_norm": 0.4848873019218445,
      "learning_rate": 5.190832424961695e-06,
      "loss": 0.3413,
      "step": 32562
    },
    {
      "epoch": 1.494332522601074,
      "grad_norm": 0.48311084508895874,
      "learning_rate": 5.190587415470565e-06,
      "loss": 0.3538,
      "step": 32563
    },
    {
      "epoch": 1.4943784131063282,
      "grad_norm": 0.570075511932373,
      "learning_rate": 5.1903424055211324e-06,
      "loss": 0.3483,
      "step": 32564
    },
    {
      "epoch": 1.4944243036115827,
      "grad_norm": 0.4560588598251343,
      "learning_rate": 5.1900973951139865e-06,
      "loss": 0.3247,
      "step": 32565
    },
    {
      "epoch": 1.4944701941168372,
      "grad_norm": 0.5002174377441406,
      "learning_rate": 5.189852384249718e-06,
      "loss": 0.4003,
      "step": 32566
    },
    {
      "epoch": 1.4945160846220917,
      "grad_norm": 0.47151753306388855,
      "learning_rate": 5.189607372928914e-06,
      "loss": 0.3459,
      "step": 32567
    },
    {
      "epoch": 1.4945619751273462,
      "grad_norm": 0.4523603022098541,
      "learning_rate": 5.189362361152163e-06,
      "loss": 0.3667,
      "step": 32568
    },
    {
      "epoch": 1.4946078656326005,
      "grad_norm": 0.4433008134365082,
      "learning_rate": 5.189117348920056e-06,
      "loss": 0.3013,
      "step": 32569
    },
    {
      "epoch": 1.494653756137855,
      "grad_norm": 0.5028434991836548,
      "learning_rate": 5.1888723362331806e-06,
      "loss": 0.4026,
      "step": 32570
    },
    {
      "epoch": 1.4946996466431095,
      "grad_norm": 0.517037034034729,
      "learning_rate": 5.188627323092128e-06,
      "loss": 0.4529,
      "step": 32571
    },
    {
      "epoch": 1.494745537148364,
      "grad_norm": 0.4504440724849701,
      "learning_rate": 5.1883823094974864e-06,
      "loss": 0.3621,
      "step": 32572
    },
    {
      "epoch": 1.4947914276536185,
      "grad_norm": 0.4609016180038452,
      "learning_rate": 5.188137295449843e-06,
      "loss": 0.3703,
      "step": 32573
    },
    {
      "epoch": 1.494837318158873,
      "grad_norm": 0.5202422142028809,
      "learning_rate": 5.18789228094979e-06,
      "loss": 0.4068,
      "step": 32574
    },
    {
      "epoch": 1.4948832086641275,
      "grad_norm": 0.4575437307357788,
      "learning_rate": 5.187647265997916e-06,
      "loss": 0.3681,
      "step": 32575
    },
    {
      "epoch": 1.494929099169382,
      "grad_norm": 0.4750671088695526,
      "learning_rate": 5.1874022505948086e-06,
      "loss": 0.3486,
      "step": 32576
    },
    {
      "epoch": 1.4949749896746363,
      "grad_norm": 0.46575966477394104,
      "learning_rate": 5.187157234741058e-06,
      "loss": 0.3862,
      "step": 32577
    },
    {
      "epoch": 1.4950208801798908,
      "grad_norm": 0.48859214782714844,
      "learning_rate": 5.186912218437255e-06,
      "loss": 0.4036,
      "step": 32578
    },
    {
      "epoch": 1.4950667706851453,
      "grad_norm": 0.48272618651390076,
      "learning_rate": 5.186667201683984e-06,
      "loss": 0.354,
      "step": 32579
    },
    {
      "epoch": 1.4951126611903998,
      "grad_norm": 0.48283910751342773,
      "learning_rate": 5.186422184481839e-06,
      "loss": 0.39,
      "step": 32580
    },
    {
      "epoch": 1.495158551695654,
      "grad_norm": 0.45446208119392395,
      "learning_rate": 5.186177166831408e-06,
      "loss": 0.381,
      "step": 32581
    },
    {
      "epoch": 1.4952044422009085,
      "grad_norm": 0.48283690214157104,
      "learning_rate": 5.185932148733279e-06,
      "loss": 0.358,
      "step": 32582
    },
    {
      "epoch": 1.495250332706163,
      "grad_norm": 0.4367370903491974,
      "learning_rate": 5.185687130188042e-06,
      "loss": 0.2933,
      "step": 32583
    },
    {
      "epoch": 1.4952962232114175,
      "grad_norm": 0.4511030316352844,
      "learning_rate": 5.185442111196286e-06,
      "loss": 0.3017,
      "step": 32584
    },
    {
      "epoch": 1.495342113716672,
      "grad_norm": 0.49175339937210083,
      "learning_rate": 5.185197091758599e-06,
      "loss": 0.4095,
      "step": 32585
    },
    {
      "epoch": 1.4953880042219265,
      "grad_norm": 0.4730582535266876,
      "learning_rate": 5.184952071875573e-06,
      "loss": 0.3217,
      "step": 32586
    },
    {
      "epoch": 1.495433894727181,
      "grad_norm": 0.5141056776046753,
      "learning_rate": 5.184707051547795e-06,
      "loss": 0.4114,
      "step": 32587
    },
    {
      "epoch": 1.4954797852324355,
      "grad_norm": 0.5044583082199097,
      "learning_rate": 5.184462030775856e-06,
      "loss": 0.3971,
      "step": 32588
    },
    {
      "epoch": 1.4955256757376898,
      "grad_norm": 0.42000943422317505,
      "learning_rate": 5.184217009560342e-06,
      "loss": 0.303,
      "step": 32589
    },
    {
      "epoch": 1.4955715662429443,
      "grad_norm": 0.45065370202064514,
      "learning_rate": 5.183971987901846e-06,
      "loss": 0.3526,
      "step": 32590
    },
    {
      "epoch": 1.4956174567481988,
      "grad_norm": 0.46547064185142517,
      "learning_rate": 5.183726965800955e-06,
      "loss": 0.3551,
      "step": 32591
    },
    {
      "epoch": 1.4956633472534533,
      "grad_norm": 0.45704925060272217,
      "learning_rate": 5.1834819432582585e-06,
      "loss": 0.3515,
      "step": 32592
    },
    {
      "epoch": 1.4957092377587078,
      "grad_norm": 0.48265931010246277,
      "learning_rate": 5.183236920274346e-06,
      "loss": 0.3658,
      "step": 32593
    },
    {
      "epoch": 1.495755128263962,
      "grad_norm": 0.4727610647678375,
      "learning_rate": 5.182991896849806e-06,
      "loss": 0.3772,
      "step": 32594
    },
    {
      "epoch": 1.4958010187692166,
      "grad_norm": 0.4863734841346741,
      "learning_rate": 5.182746872985229e-06,
      "loss": 0.3668,
      "step": 32595
    },
    {
      "epoch": 1.495846909274471,
      "grad_norm": 1.7209581136703491,
      "learning_rate": 5.182501848681203e-06,
      "loss": 0.307,
      "step": 32596
    },
    {
      "epoch": 1.4958927997797256,
      "grad_norm": 0.45055389404296875,
      "learning_rate": 5.182256823938318e-06,
      "loss": 0.3202,
      "step": 32597
    },
    {
      "epoch": 1.49593869028498,
      "grad_norm": 0.46957501769065857,
      "learning_rate": 5.182011798757161e-06,
      "loss": 0.3666,
      "step": 32598
    },
    {
      "epoch": 1.4959845807902346,
      "grad_norm": 0.4440155625343323,
      "learning_rate": 5.181766773138326e-06,
      "loss": 0.2994,
      "step": 32599
    },
    {
      "epoch": 1.496030471295489,
      "grad_norm": 0.45915642380714417,
      "learning_rate": 5.181521747082397e-06,
      "loss": 0.3553,
      "step": 32600
    },
    {
      "epoch": 1.4960763618007435,
      "grad_norm": 0.43989524245262146,
      "learning_rate": 5.181276720589967e-06,
      "loss": 0.3188,
      "step": 32601
    },
    {
      "epoch": 1.4961222523059978,
      "grad_norm": 0.5142198204994202,
      "learning_rate": 5.181031693661622e-06,
      "loss": 0.4522,
      "step": 32602
    },
    {
      "epoch": 1.4961681428112523,
      "grad_norm": 0.46187636256217957,
      "learning_rate": 5.180786666297954e-06,
      "loss": 0.348,
      "step": 32603
    },
    {
      "epoch": 1.4962140333165068,
      "grad_norm": 0.5160707235336304,
      "learning_rate": 5.180541638499552e-06,
      "loss": 0.3872,
      "step": 32604
    },
    {
      "epoch": 1.4962599238217613,
      "grad_norm": 0.4476722776889801,
      "learning_rate": 5.180296610267003e-06,
      "loss": 0.3154,
      "step": 32605
    },
    {
      "epoch": 1.4963058143270158,
      "grad_norm": 0.5009312033653259,
      "learning_rate": 5.180051581600899e-06,
      "loss": 0.3848,
      "step": 32606
    },
    {
      "epoch": 1.49635170483227,
      "grad_norm": 0.4788958430290222,
      "learning_rate": 5.179806552501826e-06,
      "loss": 0.3441,
      "step": 32607
    },
    {
      "epoch": 1.4963975953375246,
      "grad_norm": 0.4858202338218689,
      "learning_rate": 5.179561522970376e-06,
      "loss": 0.3245,
      "step": 32608
    },
    {
      "epoch": 1.496443485842779,
      "grad_norm": 0.4013120234012604,
      "learning_rate": 5.179316493007136e-06,
      "loss": 0.2091,
      "step": 32609
    },
    {
      "epoch": 1.4964893763480336,
      "grad_norm": 0.4568566679954529,
      "learning_rate": 5.179071462612698e-06,
      "loss": 0.3044,
      "step": 32610
    },
    {
      "epoch": 1.496535266853288,
      "grad_norm": 0.44588610529899597,
      "learning_rate": 5.178826431787649e-06,
      "loss": 0.2768,
      "step": 32611
    },
    {
      "epoch": 1.4965811573585426,
      "grad_norm": 0.4667956233024597,
      "learning_rate": 5.17858140053258e-06,
      "loss": 0.3684,
      "step": 32612
    },
    {
      "epoch": 1.496627047863797,
      "grad_norm": 0.4453117549419403,
      "learning_rate": 5.178336368848077e-06,
      "loss": 0.3085,
      "step": 32613
    },
    {
      "epoch": 1.4966729383690516,
      "grad_norm": 0.45313918590545654,
      "learning_rate": 5.178091336734733e-06,
      "loss": 0.3313,
      "step": 32614
    },
    {
      "epoch": 1.4967188288743059,
      "grad_norm": 0.4657720923423767,
      "learning_rate": 5.1778463041931346e-06,
      "loss": 0.3481,
      "step": 32615
    },
    {
      "epoch": 1.4967647193795603,
      "grad_norm": 0.4799053370952606,
      "learning_rate": 5.177601271223872e-06,
      "loss": 0.3385,
      "step": 32616
    },
    {
      "epoch": 1.4968106098848148,
      "grad_norm": 0.4610821008682251,
      "learning_rate": 5.177356237827535e-06,
      "loss": 0.3312,
      "step": 32617
    },
    {
      "epoch": 1.4968565003900693,
      "grad_norm": 0.4326995313167572,
      "learning_rate": 5.177111204004711e-06,
      "loss": 0.3285,
      "step": 32618
    },
    {
      "epoch": 1.4969023908953236,
      "grad_norm": 0.46440601348876953,
      "learning_rate": 5.17686616975599e-06,
      "loss": 0.3228,
      "step": 32619
    },
    {
      "epoch": 1.4969482814005781,
      "grad_norm": 0.4918068051338196,
      "learning_rate": 5.176621135081965e-06,
      "loss": 0.3217,
      "step": 32620
    },
    {
      "epoch": 1.4969941719058326,
      "grad_norm": 0.4461206793785095,
      "learning_rate": 5.1763760999832205e-06,
      "loss": 0.3031,
      "step": 32621
    },
    {
      "epoch": 1.4970400624110871,
      "grad_norm": 0.5044770836830139,
      "learning_rate": 5.176131064460346e-06,
      "loss": 0.4036,
      "step": 32622
    },
    {
      "epoch": 1.4970859529163416,
      "grad_norm": 0.4721382260322571,
      "learning_rate": 5.175886028513933e-06,
      "loss": 0.3581,
      "step": 32623
    },
    {
      "epoch": 1.497131843421596,
      "grad_norm": 0.45932653546333313,
      "learning_rate": 5.175640992144568e-06,
      "loss": 0.3339,
      "step": 32624
    },
    {
      "epoch": 1.4971777339268506,
      "grad_norm": 0.5004351139068604,
      "learning_rate": 5.175395955352842e-06,
      "loss": 0.4254,
      "step": 32625
    },
    {
      "epoch": 1.497223624432105,
      "grad_norm": 0.4440430998802185,
      "learning_rate": 5.175150918139346e-06,
      "loss": 0.3442,
      "step": 32626
    },
    {
      "epoch": 1.4972695149373594,
      "grad_norm": 0.4274478554725647,
      "learning_rate": 5.174905880504666e-06,
      "loss": 0.3258,
      "step": 32627
    },
    {
      "epoch": 1.4973154054426139,
      "grad_norm": 0.48320508003234863,
      "learning_rate": 5.174660842449392e-06,
      "loss": 0.4057,
      "step": 32628
    },
    {
      "epoch": 1.4973612959478684,
      "grad_norm": 0.4785393476486206,
      "learning_rate": 5.174415803974116e-06,
      "loss": 0.3463,
      "step": 32629
    },
    {
      "epoch": 1.4974071864531229,
      "grad_norm": 0.442158043384552,
      "learning_rate": 5.174170765079423e-06,
      "loss": 0.3081,
      "step": 32630
    },
    {
      "epoch": 1.4974530769583774,
      "grad_norm": 0.4580605626106262,
      "learning_rate": 5.173925725765905e-06,
      "loss": 0.3671,
      "step": 32631
    },
    {
      "epoch": 1.4974989674636316,
      "grad_norm": 0.46944111585617065,
      "learning_rate": 5.173680686034151e-06,
      "loss": 0.372,
      "step": 32632
    },
    {
      "epoch": 1.4975448579688861,
      "grad_norm": 0.4412649869918823,
      "learning_rate": 5.173435645884749e-06,
      "loss": 0.3415,
      "step": 32633
    },
    {
      "epoch": 1.4975907484741406,
      "grad_norm": 0.44685855507850647,
      "learning_rate": 5.17319060531829e-06,
      "loss": 0.3098,
      "step": 32634
    },
    {
      "epoch": 1.4976366389793951,
      "grad_norm": 0.4748915433883667,
      "learning_rate": 5.172945564335361e-06,
      "loss": 0.4001,
      "step": 32635
    },
    {
      "epoch": 1.4976825294846496,
      "grad_norm": 0.48031970858573914,
      "learning_rate": 5.172700522936552e-06,
      "loss": 0.3513,
      "step": 32636
    },
    {
      "epoch": 1.4977284199899041,
      "grad_norm": 0.5012335777282715,
      "learning_rate": 5.172455481122456e-06,
      "loss": 0.4244,
      "step": 32637
    },
    {
      "epoch": 1.4977743104951586,
      "grad_norm": 0.43695971369743347,
      "learning_rate": 5.1722104388936555e-06,
      "loss": 0.3219,
      "step": 32638
    },
    {
      "epoch": 1.4978202010004131,
      "grad_norm": 0.5308743715286255,
      "learning_rate": 5.171965396250745e-06,
      "loss": 0.4515,
      "step": 32639
    },
    {
      "epoch": 1.4978660915056674,
      "grad_norm": 0.4448179006576538,
      "learning_rate": 5.1717203531943115e-06,
      "loss": 0.3204,
      "step": 32640
    },
    {
      "epoch": 1.497911982010922,
      "grad_norm": 0.517703652381897,
      "learning_rate": 5.171475309724944e-06,
      "loss": 0.5369,
      "step": 32641
    },
    {
      "epoch": 1.4979578725161764,
      "grad_norm": 0.43918561935424805,
      "learning_rate": 5.171230265843234e-06,
      "loss": 0.3329,
      "step": 32642
    },
    {
      "epoch": 1.498003763021431,
      "grad_norm": 0.467804491519928,
      "learning_rate": 5.170985221549769e-06,
      "loss": 0.3163,
      "step": 32643
    },
    {
      "epoch": 1.4980496535266854,
      "grad_norm": 0.43559709191322327,
      "learning_rate": 5.170740176845137e-06,
      "loss": 0.2675,
      "step": 32644
    },
    {
      "epoch": 1.4980955440319397,
      "grad_norm": 0.4466151297092438,
      "learning_rate": 5.170495131729929e-06,
      "loss": 0.331,
      "step": 32645
    },
    {
      "epoch": 1.4981414345371942,
      "grad_norm": 0.4450010657310486,
      "learning_rate": 5.170250086204735e-06,
      "loss": 0.3635,
      "step": 32646
    },
    {
      "epoch": 1.4981873250424487,
      "grad_norm": 0.4183286428451538,
      "learning_rate": 5.170005040270143e-06,
      "loss": 0.3092,
      "step": 32647
    },
    {
      "epoch": 1.4982332155477032,
      "grad_norm": 0.45541322231292725,
      "learning_rate": 5.169759993926744e-06,
      "loss": 0.3683,
      "step": 32648
    },
    {
      "epoch": 1.4982791060529577,
      "grad_norm": 0.49249210953712463,
      "learning_rate": 5.1695149471751236e-06,
      "loss": 0.3931,
      "step": 32649
    },
    {
      "epoch": 1.4983249965582122,
      "grad_norm": 0.45634743571281433,
      "learning_rate": 5.1692699000158745e-06,
      "loss": 0.3253,
      "step": 32650
    },
    {
      "epoch": 1.4983708870634667,
      "grad_norm": 0.47145089507102966,
      "learning_rate": 5.169024852449585e-06,
      "loss": 0.392,
      "step": 32651
    },
    {
      "epoch": 1.4984167775687212,
      "grad_norm": 0.4584760367870331,
      "learning_rate": 5.168779804476842e-06,
      "loss": 0.3571,
      "step": 32652
    },
    {
      "epoch": 1.4984626680739754,
      "grad_norm": 0.41557759046554565,
      "learning_rate": 5.168534756098238e-06,
      "loss": 0.2923,
      "step": 32653
    },
    {
      "epoch": 1.49850855857923,
      "grad_norm": 0.46119579672813416,
      "learning_rate": 5.168289707314362e-06,
      "loss": 0.2863,
      "step": 32654
    },
    {
      "epoch": 1.4985544490844844,
      "grad_norm": 0.4361475110054016,
      "learning_rate": 5.168044658125802e-06,
      "loss": 0.3395,
      "step": 32655
    },
    {
      "epoch": 1.498600339589739,
      "grad_norm": 0.46039947867393494,
      "learning_rate": 5.167799608533146e-06,
      "loss": 0.3845,
      "step": 32656
    },
    {
      "epoch": 1.4986462300949934,
      "grad_norm": 0.44175317883491516,
      "learning_rate": 5.167554558536987e-06,
      "loss": 0.3256,
      "step": 32657
    },
    {
      "epoch": 1.4986921206002477,
      "grad_norm": 0.4562526047229767,
      "learning_rate": 5.1673095081379095e-06,
      "loss": 0.3389,
      "step": 32658
    },
    {
      "epoch": 1.4987380111055022,
      "grad_norm": 0.5073937177658081,
      "learning_rate": 5.1670644573365094e-06,
      "loss": 0.3749,
      "step": 32659
    },
    {
      "epoch": 1.4987839016107567,
      "grad_norm": 0.4532459080219269,
      "learning_rate": 5.16681940613337e-06,
      "loss": 0.3883,
      "step": 32660
    },
    {
      "epoch": 1.4988297921160112,
      "grad_norm": 0.6352558135986328,
      "learning_rate": 5.166574354529081e-06,
      "loss": 0.4997,
      "step": 32661
    },
    {
      "epoch": 1.4988756826212657,
      "grad_norm": 0.4959786832332611,
      "learning_rate": 5.166329302524235e-06,
      "loss": 0.4019,
      "step": 32662
    },
    {
      "epoch": 1.4989215731265202,
      "grad_norm": 0.46843844652175903,
      "learning_rate": 5.16608425011942e-06,
      "loss": 0.3443,
      "step": 32663
    },
    {
      "epoch": 1.4989674636317747,
      "grad_norm": 0.4782419800758362,
      "learning_rate": 5.1658391973152245e-06,
      "loss": 0.4028,
      "step": 32664
    },
    {
      "epoch": 1.4990133541370292,
      "grad_norm": 0.4790678322315216,
      "learning_rate": 5.165594144112238e-06,
      "loss": 0.3774,
      "step": 32665
    },
    {
      "epoch": 1.4990592446422835,
      "grad_norm": 0.4578644037246704,
      "learning_rate": 5.165349090511048e-06,
      "loss": 0.3521,
      "step": 32666
    },
    {
      "epoch": 1.499105135147538,
      "grad_norm": 0.4387330114841461,
      "learning_rate": 5.165104036512247e-06,
      "loss": 0.32,
      "step": 32667
    },
    {
      "epoch": 1.4991510256527925,
      "grad_norm": 0.48777422308921814,
      "learning_rate": 5.164858982116423e-06,
      "loss": 0.4076,
      "step": 32668
    },
    {
      "epoch": 1.499196916158047,
      "grad_norm": 0.5067765116691589,
      "learning_rate": 5.1646139273241645e-06,
      "loss": 0.3739,
      "step": 32669
    },
    {
      "epoch": 1.4992428066633012,
      "grad_norm": 0.5063968896865845,
      "learning_rate": 5.164368872136063e-06,
      "loss": 0.3676,
      "step": 32670
    },
    {
      "epoch": 1.4992886971685557,
      "grad_norm": 0.5246476531028748,
      "learning_rate": 5.164123816552706e-06,
      "loss": 0.4213,
      "step": 32671
    },
    {
      "epoch": 1.4993345876738102,
      "grad_norm": 0.5739502310752869,
      "learning_rate": 5.163878760574682e-06,
      "loss": 0.4129,
      "step": 32672
    },
    {
      "epoch": 1.4993804781790647,
      "grad_norm": 0.4550324082374573,
      "learning_rate": 5.163633704202581e-06,
      "loss": 0.3495,
      "step": 32673
    },
    {
      "epoch": 1.4994263686843192,
      "grad_norm": 0.44659504294395447,
      "learning_rate": 5.163388647436993e-06,
      "loss": 0.3085,
      "step": 32674
    },
    {
      "epoch": 1.4994722591895737,
      "grad_norm": 0.4911981523036957,
      "learning_rate": 5.163143590278508e-06,
      "loss": 0.3894,
      "step": 32675
    },
    {
      "epoch": 1.4995181496948282,
      "grad_norm": 0.44962671399116516,
      "learning_rate": 5.162898532727713e-06,
      "loss": 0.3283,
      "step": 32676
    },
    {
      "epoch": 1.4995640402000827,
      "grad_norm": 0.4686223864555359,
      "learning_rate": 5.1626534747852e-06,
      "loss": 0.377,
      "step": 32677
    },
    {
      "epoch": 1.499609930705337,
      "grad_norm": 0.4423569142818451,
      "learning_rate": 5.162408416451554e-06,
      "loss": 0.3192,
      "step": 32678
    },
    {
      "epoch": 1.4996558212105915,
      "grad_norm": 0.4755551815032959,
      "learning_rate": 5.162163357727369e-06,
      "loss": 0.3216,
      "step": 32679
    },
    {
      "epoch": 1.499701711715846,
      "grad_norm": 0.4955936074256897,
      "learning_rate": 5.161918298613231e-06,
      "loss": 0.4336,
      "step": 32680
    },
    {
      "epoch": 1.4997476022211005,
      "grad_norm": 0.45997166633605957,
      "learning_rate": 5.161673239109732e-06,
      "loss": 0.3833,
      "step": 32681
    },
    {
      "epoch": 1.499793492726355,
      "grad_norm": 0.43228137493133545,
      "learning_rate": 5.1614281792174605e-06,
      "loss": 0.316,
      "step": 32682
    },
    {
      "epoch": 1.4998393832316093,
      "grad_norm": 0.460433691740036,
      "learning_rate": 5.161183118937003e-06,
      "loss": 0.3313,
      "step": 32683
    },
    {
      "epoch": 1.4998852737368638,
      "grad_norm": 0.6957261562347412,
      "learning_rate": 5.160938058268951e-06,
      "loss": 0.3829,
      "step": 32684
    },
    {
      "epoch": 1.4999311642421183,
      "grad_norm": 0.5120862126350403,
      "learning_rate": 5.160692997213896e-06,
      "loss": 0.4335,
      "step": 32685
    },
    {
      "epoch": 1.4999770547473728,
      "grad_norm": 0.4699173867702484,
      "learning_rate": 5.160447935772423e-06,
      "loss": 0.3048,
      "step": 32686
    },
    {
      "epoch": 1.5000229452526272,
      "grad_norm": 0.489084392786026,
      "learning_rate": 5.1602028739451235e-06,
      "loss": 0.3968,
      "step": 32687
    },
    {
      "epoch": 1.5000688357578817,
      "grad_norm": 0.48347148299217224,
      "learning_rate": 5.159957811732588e-06,
      "loss": 0.3644,
      "step": 32688
    },
    {
      "epoch": 1.5001147262631362,
      "grad_norm": 0.4803374409675598,
      "learning_rate": 5.159712749135404e-06,
      "loss": 0.3729,
      "step": 32689
    },
    {
      "epoch": 1.5001606167683907,
      "grad_norm": 0.44852668046951294,
      "learning_rate": 5.159467686154159e-06,
      "loss": 0.3284,
      "step": 32690
    },
    {
      "epoch": 1.5002065072736452,
      "grad_norm": 0.47035688161849976,
      "learning_rate": 5.159222622789448e-06,
      "loss": 0.3443,
      "step": 32691
    },
    {
      "epoch": 1.5002523977788995,
      "grad_norm": 0.47691479325294495,
      "learning_rate": 5.158977559041856e-06,
      "loss": 0.3743,
      "step": 32692
    },
    {
      "epoch": 1.500298288284154,
      "grad_norm": 0.5123530626296997,
      "learning_rate": 5.158732494911971e-06,
      "loss": 0.4253,
      "step": 32693
    },
    {
      "epoch": 1.5003441787894085,
      "grad_norm": 0.5266831517219543,
      "learning_rate": 5.158487430400386e-06,
      "loss": 0.3428,
      "step": 32694
    },
    {
      "epoch": 1.5003900692946628,
      "grad_norm": 0.42494234442710876,
      "learning_rate": 5.158242365507689e-06,
      "loss": 0.3002,
      "step": 32695
    },
    {
      "epoch": 1.5004359597999173,
      "grad_norm": 0.4968528151512146,
      "learning_rate": 5.157997300234468e-06,
      "loss": 0.4039,
      "step": 32696
    },
    {
      "epoch": 1.5004818503051718,
      "grad_norm": 0.47835028171539307,
      "learning_rate": 5.1577522345813155e-06,
      "loss": 0.3356,
      "step": 32697
    },
    {
      "epoch": 1.5005277408104263,
      "grad_norm": 0.5209726691246033,
      "learning_rate": 5.157507168548816e-06,
      "loss": 0.3967,
      "step": 32698
    },
    {
      "epoch": 1.5005736313156808,
      "grad_norm": 0.5262799859046936,
      "learning_rate": 5.157262102137564e-06,
      "loss": 0.4171,
      "step": 32699
    },
    {
      "epoch": 1.5006195218209353,
      "grad_norm": 0.5023545026779175,
      "learning_rate": 5.157017035348144e-06,
      "loss": 0.415,
      "step": 32700
    },
    {
      "epoch": 1.5006654123261898,
      "grad_norm": 0.4932379126548767,
      "learning_rate": 5.156771968181148e-06,
      "loss": 0.3946,
      "step": 32701
    },
    {
      "epoch": 1.5007113028314443,
      "grad_norm": 0.4738796055316925,
      "learning_rate": 5.156526900637166e-06,
      "loss": 0.3475,
      "step": 32702
    },
    {
      "epoch": 1.5007571933366988,
      "grad_norm": 0.4888899326324463,
      "learning_rate": 5.156281832716787e-06,
      "loss": 0.361,
      "step": 32703
    },
    {
      "epoch": 1.500803083841953,
      "grad_norm": 0.5066118836402893,
      "learning_rate": 5.156036764420597e-06,
      "loss": 0.4418,
      "step": 32704
    },
    {
      "epoch": 1.5008489743472075,
      "grad_norm": 0.4515625238418579,
      "learning_rate": 5.1557916957491895e-06,
      "loss": 0.3576,
      "step": 32705
    },
    {
      "epoch": 1.500894864852462,
      "grad_norm": 0.4855920672416687,
      "learning_rate": 5.15554662670315e-06,
      "loss": 0.3495,
      "step": 32706
    },
    {
      "epoch": 1.5009407553577165,
      "grad_norm": 0.4401746690273285,
      "learning_rate": 5.155301557283073e-06,
      "loss": 0.3074,
      "step": 32707
    },
    {
      "epoch": 1.5009866458629708,
      "grad_norm": 0.5005365014076233,
      "learning_rate": 5.155056487489544e-06,
      "loss": 0.3806,
      "step": 32708
    },
    {
      "epoch": 1.5010325363682253,
      "grad_norm": 0.5161315202713013,
      "learning_rate": 5.154811417323152e-06,
      "loss": 0.3768,
      "step": 32709
    },
    {
      "epoch": 1.5010784268734798,
      "grad_norm": 0.4449882209300995,
      "learning_rate": 5.1545663467844875e-06,
      "loss": 0.3072,
      "step": 32710
    },
    {
      "epoch": 1.5011243173787343,
      "grad_norm": 0.4637942612171173,
      "learning_rate": 5.15432127587414e-06,
      "loss": 0.3507,
      "step": 32711
    },
    {
      "epoch": 1.5011702078839888,
      "grad_norm": 0.4692681133747101,
      "learning_rate": 5.154076204592699e-06,
      "loss": 0.3544,
      "step": 32712
    },
    {
      "epoch": 1.5012160983892433,
      "grad_norm": 0.4487364888191223,
      "learning_rate": 5.153831132940753e-06,
      "loss": 0.3485,
      "step": 32713
    },
    {
      "epoch": 1.5012619888944978,
      "grad_norm": 0.4455161988735199,
      "learning_rate": 5.153586060918892e-06,
      "loss": 0.3514,
      "step": 32714
    },
    {
      "epoch": 1.5013078793997523,
      "grad_norm": 0.4687962234020233,
      "learning_rate": 5.153340988527704e-06,
      "loss": 0.3282,
      "step": 32715
    },
    {
      "epoch": 1.5013537699050068,
      "grad_norm": 0.46618759632110596,
      "learning_rate": 5.153095915767781e-06,
      "loss": 0.323,
      "step": 32716
    },
    {
      "epoch": 1.501399660410261,
      "grad_norm": 0.5234248042106628,
      "learning_rate": 5.1528508426397085e-06,
      "loss": 0.4018,
      "step": 32717
    },
    {
      "epoch": 1.5014455509155156,
      "grad_norm": 0.47964876890182495,
      "learning_rate": 5.15260576914408e-06,
      "loss": 0.4102,
      "step": 32718
    },
    {
      "epoch": 1.50149144142077,
      "grad_norm": 0.48703891038894653,
      "learning_rate": 5.152360695281482e-06,
      "loss": 0.3589,
      "step": 32719
    },
    {
      "epoch": 1.5015373319260243,
      "grad_norm": 0.480169415473938,
      "learning_rate": 5.152115621052503e-06,
      "loss": 0.4371,
      "step": 32720
    },
    {
      "epoch": 1.5015832224312788,
      "grad_norm": 0.46981653571128845,
      "learning_rate": 5.151870546457736e-06,
      "loss": 0.3737,
      "step": 32721
    },
    {
      "epoch": 1.5016291129365333,
      "grad_norm": 0.4615938663482666,
      "learning_rate": 5.151625471497768e-06,
      "loss": 0.3296,
      "step": 32722
    },
    {
      "epoch": 1.5016750034417878,
      "grad_norm": 0.43904951214790344,
      "learning_rate": 5.151380396173188e-06,
      "loss": 0.2751,
      "step": 32723
    },
    {
      "epoch": 1.5017208939470423,
      "grad_norm": 0.5143400430679321,
      "learning_rate": 5.151135320484586e-06,
      "loss": 0.377,
      "step": 32724
    },
    {
      "epoch": 1.5017667844522968,
      "grad_norm": 0.44622352719306946,
      "learning_rate": 5.1508902444325525e-06,
      "loss": 0.3144,
      "step": 32725
    },
    {
      "epoch": 1.5018126749575513,
      "grad_norm": 0.43638700246810913,
      "learning_rate": 5.150645168017674e-06,
      "loss": 0.3222,
      "step": 32726
    },
    {
      "epoch": 1.5018585654628058,
      "grad_norm": 0.4574262201786041,
      "learning_rate": 5.150400091240541e-06,
      "loss": 0.3954,
      "step": 32727
    },
    {
      "epoch": 1.5019044559680603,
      "grad_norm": 0.4736306965351105,
      "learning_rate": 5.150155014101744e-06,
      "loss": 0.3351,
      "step": 32728
    },
    {
      "epoch": 1.5019503464733148,
      "grad_norm": 0.4707596004009247,
      "learning_rate": 5.149909936601874e-06,
      "loss": 0.3608,
      "step": 32729
    },
    {
      "epoch": 1.501996236978569,
      "grad_norm": 0.42518213391304016,
      "learning_rate": 5.149664858741516e-06,
      "loss": 0.3002,
      "step": 32730
    },
    {
      "epoch": 1.5020421274838236,
      "grad_norm": 0.4604383707046509,
      "learning_rate": 5.149419780521259e-06,
      "loss": 0.3361,
      "step": 32731
    },
    {
      "epoch": 1.502088017989078,
      "grad_norm": 0.47936442494392395,
      "learning_rate": 5.149174701941696e-06,
      "loss": 0.363,
      "step": 32732
    },
    {
      "epoch": 1.5021339084943324,
      "grad_norm": 0.4596308171749115,
      "learning_rate": 5.148929623003416e-06,
      "loss": 0.3124,
      "step": 32733
    },
    {
      "epoch": 1.5021797989995869,
      "grad_norm": 0.4643476903438568,
      "learning_rate": 5.148684543707006e-06,
      "loss": 0.3332,
      "step": 32734
    },
    {
      "epoch": 1.5022256895048414,
      "grad_norm": 0.42815670371055603,
      "learning_rate": 5.148439464053059e-06,
      "loss": 0.2833,
      "step": 32735
    },
    {
      "epoch": 1.5022715800100959,
      "grad_norm": 0.4539550244808197,
      "learning_rate": 5.14819438404216e-06,
      "loss": 0.3238,
      "step": 32736
    },
    {
      "epoch": 1.5023174705153504,
      "grad_norm": 0.4566689133644104,
      "learning_rate": 5.1479493036749e-06,
      "loss": 0.3574,
      "step": 32737
    },
    {
      "epoch": 1.5023633610206049,
      "grad_norm": 0.4827084541320801,
      "learning_rate": 5.147704222951868e-06,
      "loss": 0.3854,
      "step": 32738
    },
    {
      "epoch": 1.5024092515258594,
      "grad_norm": 0.4791339337825775,
      "learning_rate": 5.147459141873656e-06,
      "loss": 0.4275,
      "step": 32739
    },
    {
      "epoch": 1.5024551420311139,
      "grad_norm": 0.49924468994140625,
      "learning_rate": 5.147214060440851e-06,
      "loss": 0.412,
      "step": 32740
    },
    {
      "epoch": 1.5025010325363684,
      "grad_norm": 0.45989498496055603,
      "learning_rate": 5.1469689786540415e-06,
      "loss": 0.3671,
      "step": 32741
    },
    {
      "epoch": 1.5025469230416229,
      "grad_norm": 0.5214325785636902,
      "learning_rate": 5.146723896513819e-06,
      "loss": 0.4592,
      "step": 32742
    },
    {
      "epoch": 1.5025928135468771,
      "grad_norm": 0.4800948202610016,
      "learning_rate": 5.146478814020771e-06,
      "loss": 0.369,
      "step": 32743
    },
    {
      "epoch": 1.5026387040521316,
      "grad_norm": 0.530949592590332,
      "learning_rate": 5.146233731175489e-06,
      "loss": 0.3479,
      "step": 32744
    },
    {
      "epoch": 1.5026845945573861,
      "grad_norm": 0.4928324222564697,
      "learning_rate": 5.145988647978559e-06,
      "loss": 0.3526,
      "step": 32745
    },
    {
      "epoch": 1.5027304850626404,
      "grad_norm": 0.5022740364074707,
      "learning_rate": 5.1457435644305746e-06,
      "loss": 0.3649,
      "step": 32746
    },
    {
      "epoch": 1.502776375567895,
      "grad_norm": 0.49258044362068176,
      "learning_rate": 5.145498480532122e-06,
      "loss": 0.3714,
      "step": 32747
    },
    {
      "epoch": 1.5028222660731494,
      "grad_norm": 0.47175848484039307,
      "learning_rate": 5.14525339628379e-06,
      "loss": 0.3502,
      "step": 32748
    },
    {
      "epoch": 1.502868156578404,
      "grad_norm": 0.4350002408027649,
      "learning_rate": 5.1450083116861704e-06,
      "loss": 0.2896,
      "step": 32749
    },
    {
      "epoch": 1.5029140470836584,
      "grad_norm": 0.45505622029304504,
      "learning_rate": 5.1447632267398524e-06,
      "loss": 0.3508,
      "step": 32750
    },
    {
      "epoch": 1.5029599375889129,
      "grad_norm": 0.4473058581352234,
      "learning_rate": 5.144518141445425e-06,
      "loss": 0.3339,
      "step": 32751
    },
    {
      "epoch": 1.5030058280941674,
      "grad_norm": 0.4679861068725586,
      "learning_rate": 5.1442730558034745e-06,
      "loss": 0.3637,
      "step": 32752
    },
    {
      "epoch": 1.5030517185994219,
      "grad_norm": 0.47852495312690735,
      "learning_rate": 5.144027969814595e-06,
      "loss": 0.3405,
      "step": 32753
    },
    {
      "epoch": 1.5030976091046764,
      "grad_norm": 0.4426574409008026,
      "learning_rate": 5.143782883479374e-06,
      "loss": 0.3279,
      "step": 32754
    },
    {
      "epoch": 1.5031434996099307,
      "grad_norm": 0.49324122071266174,
      "learning_rate": 5.143537796798398e-06,
      "loss": 0.3969,
      "step": 32755
    },
    {
      "epoch": 1.5031893901151852,
      "grad_norm": 0.4641503691673279,
      "learning_rate": 5.14329270977226e-06,
      "loss": 0.3379,
      "step": 32756
    },
    {
      "epoch": 1.5032352806204397,
      "grad_norm": 0.4565747380256653,
      "learning_rate": 5.14304762240155e-06,
      "loss": 0.3504,
      "step": 32757
    },
    {
      "epoch": 1.5032811711256941,
      "grad_norm": 0.4210914373397827,
      "learning_rate": 5.142802534686855e-06,
      "loss": 0.2787,
      "step": 32758
    },
    {
      "epoch": 1.5033270616309484,
      "grad_norm": 0.4932003319263458,
      "learning_rate": 5.142557446628764e-06,
      "loss": 0.3989,
      "step": 32759
    },
    {
      "epoch": 1.503372952136203,
      "grad_norm": 0.45669296383857727,
      "learning_rate": 5.142312358227868e-06,
      "loss": 0.4044,
      "step": 32760
    },
    {
      "epoch": 1.5034188426414574,
      "grad_norm": 0.47934290766716003,
      "learning_rate": 5.142067269484755e-06,
      "loss": 0.4095,
      "step": 32761
    },
    {
      "epoch": 1.503464733146712,
      "grad_norm": 0.41430023312568665,
      "learning_rate": 5.1418221804000164e-06,
      "loss": 0.2636,
      "step": 32762
    },
    {
      "epoch": 1.5035106236519664,
      "grad_norm": 0.49233314394950867,
      "learning_rate": 5.141577090974239e-06,
      "loss": 0.422,
      "step": 32763
    },
    {
      "epoch": 1.503556514157221,
      "grad_norm": 0.45914387702941895,
      "learning_rate": 5.141332001208014e-06,
      "loss": 0.3369,
      "step": 32764
    },
    {
      "epoch": 1.5036024046624754,
      "grad_norm": 0.509056568145752,
      "learning_rate": 5.141086911101929e-06,
      "loss": 0.4102,
      "step": 32765
    },
    {
      "epoch": 1.50364829516773,
      "grad_norm": 0.493757426738739,
      "learning_rate": 5.140841820656575e-06,
      "loss": 0.3962,
      "step": 32766
    },
    {
      "epoch": 1.5036941856729844,
      "grad_norm": 0.5295786261558533,
      "learning_rate": 5.140596729872541e-06,
      "loss": 0.4891,
      "step": 32767
    },
    {
      "epoch": 1.5037400761782387,
      "grad_norm": 0.4630822241306305,
      "learning_rate": 5.140351638750417e-06,
      "loss": 0.3274,
      "step": 32768
    },
    {
      "epoch": 1.5037859666834932,
      "grad_norm": 0.4026663601398468,
      "learning_rate": 5.140106547290791e-06,
      "loss": 0.2616,
      "step": 32769
    },
    {
      "epoch": 1.5038318571887477,
      "grad_norm": 0.4737588167190552,
      "learning_rate": 5.139861455494254e-06,
      "loss": 0.3581,
      "step": 32770
    },
    {
      "epoch": 1.503877747694002,
      "grad_norm": 0.4924485981464386,
      "learning_rate": 5.139616363361393e-06,
      "loss": 0.419,
      "step": 32771
    },
    {
      "epoch": 1.5039236381992565,
      "grad_norm": 0.48611074686050415,
      "learning_rate": 5.139371270892799e-06,
      "loss": 0.3675,
      "step": 32772
    },
    {
      "epoch": 1.503969528704511,
      "grad_norm": 0.4732033908367157,
      "learning_rate": 5.139126178089061e-06,
      "loss": 0.3647,
      "step": 32773
    },
    {
      "epoch": 1.5040154192097654,
      "grad_norm": 0.48148900270462036,
      "learning_rate": 5.138881084950769e-06,
      "loss": 0.3458,
      "step": 32774
    },
    {
      "epoch": 1.50406130971502,
      "grad_norm": 0.4908605217933655,
      "learning_rate": 5.13863599147851e-06,
      "loss": 0.3629,
      "step": 32775
    },
    {
      "epoch": 1.5041072002202744,
      "grad_norm": 0.43375730514526367,
      "learning_rate": 5.138390897672877e-06,
      "loss": 0.312,
      "step": 32776
    },
    {
      "epoch": 1.504153090725529,
      "grad_norm": 0.5111060738563538,
      "learning_rate": 5.138145803534457e-06,
      "loss": 0.4097,
      "step": 32777
    },
    {
      "epoch": 1.5041989812307834,
      "grad_norm": 0.43197059631347656,
      "learning_rate": 5.137900709063841e-06,
      "loss": 0.3071,
      "step": 32778
    },
    {
      "epoch": 1.504244871736038,
      "grad_norm": 0.49968311190605164,
      "learning_rate": 5.137655614261616e-06,
      "loss": 0.423,
      "step": 32779
    },
    {
      "epoch": 1.5042907622412924,
      "grad_norm": 0.4544350206851959,
      "learning_rate": 5.137410519128373e-06,
      "loss": 0.3163,
      "step": 32780
    },
    {
      "epoch": 1.5043366527465467,
      "grad_norm": 0.4515538215637207,
      "learning_rate": 5.137165423664702e-06,
      "loss": 0.2797,
      "step": 32781
    },
    {
      "epoch": 1.5043825432518012,
      "grad_norm": 0.4915199279785156,
      "learning_rate": 5.13692032787119e-06,
      "loss": 0.3675,
      "step": 32782
    },
    {
      "epoch": 1.5044284337570557,
      "grad_norm": 0.4520937502384186,
      "learning_rate": 5.136675231748429e-06,
      "loss": 0.3587,
      "step": 32783
    },
    {
      "epoch": 1.50447432426231,
      "grad_norm": 0.48954692482948303,
      "learning_rate": 5.136430135297007e-06,
      "loss": 0.3674,
      "step": 32784
    },
    {
      "epoch": 1.5045202147675645,
      "grad_norm": 0.43955543637275696,
      "learning_rate": 5.136185038517513e-06,
      "loss": 0.2879,
      "step": 32785
    },
    {
      "epoch": 1.504566105272819,
      "grad_norm": 0.46372300386428833,
      "learning_rate": 5.135939941410536e-06,
      "loss": 0.3749,
      "step": 32786
    },
    {
      "epoch": 1.5046119957780735,
      "grad_norm": 0.45749542117118835,
      "learning_rate": 5.135694843976669e-06,
      "loss": 0.3357,
      "step": 32787
    },
    {
      "epoch": 1.504657886283328,
      "grad_norm": 0.48393115401268005,
      "learning_rate": 5.135449746216496e-06,
      "loss": 0.2938,
      "step": 32788
    },
    {
      "epoch": 1.5047037767885825,
      "grad_norm": 0.4295254051685333,
      "learning_rate": 5.13520464813061e-06,
      "loss": 0.3198,
      "step": 32789
    },
    {
      "epoch": 1.504749667293837,
      "grad_norm": 0.4525389075279236,
      "learning_rate": 5.134959549719602e-06,
      "loss": 0.3462,
      "step": 32790
    },
    {
      "epoch": 1.5047955577990915,
      "grad_norm": 0.6302730441093445,
      "learning_rate": 5.134714450984055e-06,
      "loss": 0.309,
      "step": 32791
    },
    {
      "epoch": 1.504841448304346,
      "grad_norm": 0.48473483324050903,
      "learning_rate": 5.134469351924564e-06,
      "loss": 0.4143,
      "step": 32792
    },
    {
      "epoch": 1.5048873388096002,
      "grad_norm": 0.4580526649951935,
      "learning_rate": 5.134224252541718e-06,
      "loss": 0.3311,
      "step": 32793
    },
    {
      "epoch": 1.5049332293148547,
      "grad_norm": 0.4836895763874054,
      "learning_rate": 5.133979152836104e-06,
      "loss": 0.3791,
      "step": 32794
    },
    {
      "epoch": 1.5049791198201092,
      "grad_norm": 0.46281442046165466,
      "learning_rate": 5.1337340528083125e-06,
      "loss": 0.3135,
      "step": 32795
    },
    {
      "epoch": 1.5050250103253637,
      "grad_norm": 0.46044954657554626,
      "learning_rate": 5.1334889524589316e-06,
      "loss": 0.3254,
      "step": 32796
    },
    {
      "epoch": 1.505070900830618,
      "grad_norm": 0.48662540316581726,
      "learning_rate": 5.133243851788552e-06,
      "loss": 0.354,
      "step": 32797
    },
    {
      "epoch": 1.5051167913358725,
      "grad_norm": 0.44564583897590637,
      "learning_rate": 5.132998750797764e-06,
      "loss": 0.3536,
      "step": 32798
    },
    {
      "epoch": 1.505162681841127,
      "grad_norm": 0.48173239827156067,
      "learning_rate": 5.132753649487156e-06,
      "loss": 0.3926,
      "step": 32799
    },
    {
      "epoch": 1.5052085723463815,
      "grad_norm": 0.43214672803878784,
      "learning_rate": 5.132508547857317e-06,
      "loss": 0.2912,
      "step": 32800
    },
    {
      "epoch": 1.505254462851636,
      "grad_norm": 0.5324264168739319,
      "learning_rate": 5.132263445908837e-06,
      "loss": 0.4409,
      "step": 32801
    },
    {
      "epoch": 1.5053003533568905,
      "grad_norm": 0.45244595408439636,
      "learning_rate": 5.132018343642305e-06,
      "loss": 0.31,
      "step": 32802
    },
    {
      "epoch": 1.505346243862145,
      "grad_norm": 0.5400252342224121,
      "learning_rate": 5.13177324105831e-06,
      "loss": 0.342,
      "step": 32803
    },
    {
      "epoch": 1.5053921343673995,
      "grad_norm": 0.45666155219078064,
      "learning_rate": 5.131528138157443e-06,
      "loss": 0.3714,
      "step": 32804
    },
    {
      "epoch": 1.505438024872654,
      "grad_norm": 0.5679479241371155,
      "learning_rate": 5.131283034940292e-06,
      "loss": 0.468,
      "step": 32805
    },
    {
      "epoch": 1.5054839153779083,
      "grad_norm": 0.47452667355537415,
      "learning_rate": 5.131037931407447e-06,
      "loss": 0.3576,
      "step": 32806
    },
    {
      "epoch": 1.5055298058831628,
      "grad_norm": 0.4930656850337982,
      "learning_rate": 5.130792827559496e-06,
      "loss": 0.4241,
      "step": 32807
    },
    {
      "epoch": 1.5055756963884173,
      "grad_norm": 0.45299631357192993,
      "learning_rate": 5.1305477233970304e-06,
      "loss": 0.2887,
      "step": 32808
    },
    {
      "epoch": 1.5056215868936715,
      "grad_norm": 0.45337390899658203,
      "learning_rate": 5.130302618920637e-06,
      "loss": 0.3073,
      "step": 32809
    },
    {
      "epoch": 1.505667477398926,
      "grad_norm": 0.473055899143219,
      "learning_rate": 5.13005751413091e-06,
      "loss": 0.3737,
      "step": 32810
    },
    {
      "epoch": 1.5057133679041805,
      "grad_norm": 0.4456495940685272,
      "learning_rate": 5.129812409028434e-06,
      "loss": 0.3251,
      "step": 32811
    },
    {
      "epoch": 1.505759258409435,
      "grad_norm": 0.4496329128742218,
      "learning_rate": 5.129567303613801e-06,
      "loss": 0.3344,
      "step": 32812
    },
    {
      "epoch": 1.5058051489146895,
      "grad_norm": 0.47687768936157227,
      "learning_rate": 5.129322197887598e-06,
      "loss": 0.3531,
      "step": 32813
    },
    {
      "epoch": 1.505851039419944,
      "grad_norm": 0.4649117588996887,
      "learning_rate": 5.1290770918504174e-06,
      "loss": 0.4048,
      "step": 32814
    },
    {
      "epoch": 1.5058969299251985,
      "grad_norm": 0.4566599428653717,
      "learning_rate": 5.128831985502846e-06,
      "loss": 0.3753,
      "step": 32815
    },
    {
      "epoch": 1.505942820430453,
      "grad_norm": 0.46511006355285645,
      "learning_rate": 5.1285868788454765e-06,
      "loss": 0.3163,
      "step": 32816
    },
    {
      "epoch": 1.5059887109357075,
      "grad_norm": 0.4653542637825012,
      "learning_rate": 5.128341771878893e-06,
      "loss": 0.3611,
      "step": 32817
    },
    {
      "epoch": 1.506034601440962,
      "grad_norm": 0.46036452054977417,
      "learning_rate": 5.12809666460369e-06,
      "loss": 0.3316,
      "step": 32818
    },
    {
      "epoch": 1.5060804919462163,
      "grad_norm": 0.48076075315475464,
      "learning_rate": 5.127851557020454e-06,
      "loss": 0.3492,
      "step": 32819
    },
    {
      "epoch": 1.5061263824514708,
      "grad_norm": 0.5332815647125244,
      "learning_rate": 5.127606449129776e-06,
      "loss": 0.4524,
      "step": 32820
    },
    {
      "epoch": 1.5061722729567253,
      "grad_norm": 0.4529314339160919,
      "learning_rate": 5.1273613409322455e-06,
      "loss": 0.343,
      "step": 32821
    },
    {
      "epoch": 1.5062181634619796,
      "grad_norm": 0.48692092299461365,
      "learning_rate": 5.1271162324284505e-06,
      "loss": 0.3909,
      "step": 32822
    },
    {
      "epoch": 1.506264053967234,
      "grad_norm": 0.48001629114151,
      "learning_rate": 5.126871123618979e-06,
      "loss": 0.4067,
      "step": 32823
    },
    {
      "epoch": 1.5063099444724886,
      "grad_norm": 0.537373423576355,
      "learning_rate": 5.126626014504426e-06,
      "loss": 0.4757,
      "step": 32824
    },
    {
      "epoch": 1.506355834977743,
      "grad_norm": 0.4413423538208008,
      "learning_rate": 5.126380905085375e-06,
      "loss": 0.3317,
      "step": 32825
    },
    {
      "epoch": 1.5064017254829976,
      "grad_norm": 0.44824352860450745,
      "learning_rate": 5.1261357953624194e-06,
      "loss": 0.3212,
      "step": 32826
    },
    {
      "epoch": 1.506447615988252,
      "grad_norm": 0.5113248229026794,
      "learning_rate": 5.125890685336147e-06,
      "loss": 0.4355,
      "step": 32827
    },
    {
      "epoch": 1.5064935064935066,
      "grad_norm": 0.46396464109420776,
      "learning_rate": 5.125645575007145e-06,
      "loss": 0.3319,
      "step": 32828
    },
    {
      "epoch": 1.506539396998761,
      "grad_norm": 0.4569416344165802,
      "learning_rate": 5.125400464376007e-06,
      "loss": 0.3792,
      "step": 32829
    },
    {
      "epoch": 1.5065852875040155,
      "grad_norm": 0.5107754468917847,
      "learning_rate": 5.12515535344332e-06,
      "loss": 0.4297,
      "step": 32830
    },
    {
      "epoch": 1.50663117800927,
      "grad_norm": 0.45438775420188904,
      "learning_rate": 5.124910242209673e-06,
      "loss": 0.3419,
      "step": 32831
    },
    {
      "epoch": 1.5066770685145243,
      "grad_norm": 0.4571935832500458,
      "learning_rate": 5.124665130675658e-06,
      "loss": 0.3613,
      "step": 32832
    },
    {
      "epoch": 1.5067229590197788,
      "grad_norm": 0.4636681377887726,
      "learning_rate": 5.124420018841861e-06,
      "loss": 0.3202,
      "step": 32833
    },
    {
      "epoch": 1.5067688495250333,
      "grad_norm": 0.46161359548568726,
      "learning_rate": 5.1241749067088745e-06,
      "loss": 0.344,
      "step": 32834
    },
    {
      "epoch": 1.5068147400302876,
      "grad_norm": 0.4809042513370514,
      "learning_rate": 5.123929794277286e-06,
      "loss": 0.336,
      "step": 32835
    },
    {
      "epoch": 1.506860630535542,
      "grad_norm": 0.4723934233188629,
      "learning_rate": 5.123684681547685e-06,
      "loss": 0.3201,
      "step": 32836
    },
    {
      "epoch": 1.5069065210407966,
      "grad_norm": 0.462736576795578,
      "learning_rate": 5.123439568520661e-06,
      "loss": 0.3462,
      "step": 32837
    },
    {
      "epoch": 1.506952411546051,
      "grad_norm": 0.5029008388519287,
      "learning_rate": 5.123194455196806e-06,
      "loss": 0.4139,
      "step": 32838
    },
    {
      "epoch": 1.5069983020513056,
      "grad_norm": 0.49763917922973633,
      "learning_rate": 5.1229493415767065e-06,
      "loss": 0.3982,
      "step": 32839
    },
    {
      "epoch": 1.50704419255656,
      "grad_norm": 0.4570232033729553,
      "learning_rate": 5.12270422766095e-06,
      "loss": 0.3344,
      "step": 32840
    },
    {
      "epoch": 1.5070900830618146,
      "grad_norm": 0.44540923833847046,
      "learning_rate": 5.122459113450131e-06,
      "loss": 0.335,
      "step": 32841
    },
    {
      "epoch": 1.507135973567069,
      "grad_norm": 0.4682774841785431,
      "learning_rate": 5.1222139989448365e-06,
      "loss": 0.339,
      "step": 32842
    },
    {
      "epoch": 1.5071818640723236,
      "grad_norm": 0.4959350824356079,
      "learning_rate": 5.1219688841456546e-06,
      "loss": 0.4294,
      "step": 32843
    },
    {
      "epoch": 1.5072277545775778,
      "grad_norm": 0.445987343788147,
      "learning_rate": 5.121723769053178e-06,
      "loss": 0.2967,
      "step": 32844
    },
    {
      "epoch": 1.5072736450828323,
      "grad_norm": 0.4685248136520386,
      "learning_rate": 5.121478653667993e-06,
      "loss": 0.3371,
      "step": 32845
    },
    {
      "epoch": 1.5073195355880868,
      "grad_norm": 0.4719371199607849,
      "learning_rate": 5.121233537990691e-06,
      "loss": 0.3816,
      "step": 32846
    },
    {
      "epoch": 1.5073654260933413,
      "grad_norm": 0.4594067335128784,
      "learning_rate": 5.120988422021859e-06,
      "loss": 0.3772,
      "step": 32847
    },
    {
      "epoch": 1.5074113165985956,
      "grad_norm": 0.4767225682735443,
      "learning_rate": 5.120743305762088e-06,
      "loss": 0.3507,
      "step": 32848
    },
    {
      "epoch": 1.5074572071038501,
      "grad_norm": 0.4705440104007721,
      "learning_rate": 5.12049818921197e-06,
      "loss": 0.3273,
      "step": 32849
    },
    {
      "epoch": 1.5075030976091046,
      "grad_norm": 0.4943653643131256,
      "learning_rate": 5.120253072372089e-06,
      "loss": 0.3963,
      "step": 32850
    },
    {
      "epoch": 1.5075489881143591,
      "grad_norm": 0.46062716841697693,
      "learning_rate": 5.120007955243039e-06,
      "loss": 0.2931,
      "step": 32851
    },
    {
      "epoch": 1.5075948786196136,
      "grad_norm": 0.47785094380378723,
      "learning_rate": 5.119762837825407e-06,
      "loss": 0.3766,
      "step": 32852
    },
    {
      "epoch": 1.507640769124868,
      "grad_norm": 0.44761592149734497,
      "learning_rate": 5.119517720119782e-06,
      "loss": 0.2986,
      "step": 32853
    },
    {
      "epoch": 1.5076866596301226,
      "grad_norm": 0.4577309787273407,
      "learning_rate": 5.1192726021267566e-06,
      "loss": 0.3507,
      "step": 32854
    },
    {
      "epoch": 1.507732550135377,
      "grad_norm": 0.4630909562110901,
      "learning_rate": 5.119027483846919e-06,
      "loss": 0.3161,
      "step": 32855
    },
    {
      "epoch": 1.5077784406406316,
      "grad_norm": 0.4636571705341339,
      "learning_rate": 5.118782365280855e-06,
      "loss": 0.3431,
      "step": 32856
    },
    {
      "epoch": 1.5078243311458859,
      "grad_norm": 0.515229344367981,
      "learning_rate": 5.1185372464291586e-06,
      "loss": 0.451,
      "step": 32857
    },
    {
      "epoch": 1.5078702216511404,
      "grad_norm": 0.4891173243522644,
      "learning_rate": 5.118292127292417e-06,
      "loss": 0.3743,
      "step": 32858
    },
    {
      "epoch": 1.5079161121563949,
      "grad_norm": 0.49580472707748413,
      "learning_rate": 5.118047007871221e-06,
      "loss": 0.4029,
      "step": 32859
    },
    {
      "epoch": 1.5079620026616491,
      "grad_norm": 0.5243240594863892,
      "learning_rate": 5.1178018881661595e-06,
      "loss": 0.4444,
      "step": 32860
    },
    {
      "epoch": 1.5080078931669036,
      "grad_norm": 0.485748827457428,
      "learning_rate": 5.117556768177821e-06,
      "loss": 0.3937,
      "step": 32861
    },
    {
      "epoch": 1.5080537836721581,
      "grad_norm": 0.4666086733341217,
      "learning_rate": 5.117311647906794e-06,
      "loss": 0.3582,
      "step": 32862
    },
    {
      "epoch": 1.5080996741774126,
      "grad_norm": 0.4761119484901428,
      "learning_rate": 5.117066527353671e-06,
      "loss": 0.4144,
      "step": 32863
    },
    {
      "epoch": 1.5081455646826671,
      "grad_norm": 0.47545504570007324,
      "learning_rate": 5.11682140651904e-06,
      "loss": 0.3332,
      "step": 32864
    },
    {
      "epoch": 1.5081914551879216,
      "grad_norm": 0.47717857360839844,
      "learning_rate": 5.116576285403491e-06,
      "loss": 0.3713,
      "step": 32865
    },
    {
      "epoch": 1.5082373456931761,
      "grad_norm": 0.4473412334918976,
      "learning_rate": 5.116331164007612e-06,
      "loss": 0.2819,
      "step": 32866
    },
    {
      "epoch": 1.5082832361984306,
      "grad_norm": 0.43218520283699036,
      "learning_rate": 5.116086042331993e-06,
      "loss": 0.3104,
      "step": 32867
    },
    {
      "epoch": 1.5083291267036851,
      "grad_norm": 0.4776930510997772,
      "learning_rate": 5.115840920377223e-06,
      "loss": 0.3887,
      "step": 32868
    },
    {
      "epoch": 1.5083750172089396,
      "grad_norm": 0.46142086386680603,
      "learning_rate": 5.115595798143894e-06,
      "loss": 0.3332,
      "step": 32869
    },
    {
      "epoch": 1.508420907714194,
      "grad_norm": 0.4706965982913971,
      "learning_rate": 5.115350675632592e-06,
      "loss": 0.3651,
      "step": 32870
    },
    {
      "epoch": 1.5084667982194484,
      "grad_norm": 0.5128078460693359,
      "learning_rate": 5.11510555284391e-06,
      "loss": 0.4355,
      "step": 32871
    },
    {
      "epoch": 1.508512688724703,
      "grad_norm": 0.418459415435791,
      "learning_rate": 5.1148604297784346e-06,
      "loss": 0.2931,
      "step": 32872
    },
    {
      "epoch": 1.5085585792299572,
      "grad_norm": 0.48294806480407715,
      "learning_rate": 5.114615306436755e-06,
      "loss": 0.3655,
      "step": 32873
    },
    {
      "epoch": 1.5086044697352117,
      "grad_norm": 0.47281691431999207,
      "learning_rate": 5.114370182819461e-06,
      "loss": 0.3462,
      "step": 32874
    },
    {
      "epoch": 1.5086503602404662,
      "grad_norm": 0.4664763808250427,
      "learning_rate": 5.114125058927146e-06,
      "loss": 0.304,
      "step": 32875
    },
    {
      "epoch": 1.5086962507457207,
      "grad_norm": 0.4891469478607178,
      "learning_rate": 5.113879934760395e-06,
      "loss": 0.4151,
      "step": 32876
    },
    {
      "epoch": 1.5087421412509752,
      "grad_norm": 0.5529966950416565,
      "learning_rate": 5.1136348103197985e-06,
      "loss": 0.3969,
      "step": 32877
    },
    {
      "epoch": 1.5087880317562297,
      "grad_norm": 0.43017295002937317,
      "learning_rate": 5.113389685605946e-06,
      "loss": 0.274,
      "step": 32878
    },
    {
      "epoch": 1.5088339222614842,
      "grad_norm": 0.449070006608963,
      "learning_rate": 5.113144560619426e-06,
      "loss": 0.3108,
      "step": 32879
    },
    {
      "epoch": 1.5088798127667387,
      "grad_norm": 0.5158154964447021,
      "learning_rate": 5.1128994353608305e-06,
      "loss": 0.406,
      "step": 32880
    },
    {
      "epoch": 1.5089257032719932,
      "grad_norm": 0.520153284072876,
      "learning_rate": 5.1126543098307466e-06,
      "loss": 0.479,
      "step": 32881
    },
    {
      "epoch": 1.5089715937772474,
      "grad_norm": 0.481496125459671,
      "learning_rate": 5.112409184029766e-06,
      "loss": 0.3899,
      "step": 32882
    },
    {
      "epoch": 1.509017484282502,
      "grad_norm": 0.5117810368537903,
      "learning_rate": 5.1121640579584765e-06,
      "loss": 0.4212,
      "step": 32883
    },
    {
      "epoch": 1.5090633747877564,
      "grad_norm": 0.42805805802345276,
      "learning_rate": 5.111918931617467e-06,
      "loss": 0.2614,
      "step": 32884
    },
    {
      "epoch": 1.509109265293011,
      "grad_norm": 0.5474495887756348,
      "learning_rate": 5.111673805007326e-06,
      "loss": 0.5101,
      "step": 32885
    },
    {
      "epoch": 1.5091551557982652,
      "grad_norm": 0.4785231351852417,
      "learning_rate": 5.111428678128648e-06,
      "loss": 0.3519,
      "step": 32886
    },
    {
      "epoch": 1.5092010463035197,
      "grad_norm": 0.47195032238960266,
      "learning_rate": 5.111183550982019e-06,
      "loss": 0.3296,
      "step": 32887
    },
    {
      "epoch": 1.5092469368087742,
      "grad_norm": 0.49945446848869324,
      "learning_rate": 5.110938423568027e-06,
      "loss": 0.4029,
      "step": 32888
    },
    {
      "epoch": 1.5092928273140287,
      "grad_norm": 0.48852723836898804,
      "learning_rate": 5.110693295887264e-06,
      "loss": 0.3284,
      "step": 32889
    },
    {
      "epoch": 1.5093387178192832,
      "grad_norm": 0.4610905051231384,
      "learning_rate": 5.110448167940318e-06,
      "loss": 0.3132,
      "step": 32890
    },
    {
      "epoch": 1.5093846083245377,
      "grad_norm": 0.4688301384449005,
      "learning_rate": 5.110203039727779e-06,
      "loss": 0.3762,
      "step": 32891
    },
    {
      "epoch": 1.5094304988297922,
      "grad_norm": 0.48501455783843994,
      "learning_rate": 5.109957911250236e-06,
      "loss": 0.3383,
      "step": 32892
    },
    {
      "epoch": 1.5094763893350467,
      "grad_norm": 0.5033056735992432,
      "learning_rate": 5.10971278250828e-06,
      "loss": 0.4053,
      "step": 32893
    },
    {
      "epoch": 1.5095222798403012,
      "grad_norm": 0.4347902238368988,
      "learning_rate": 5.1094676535025e-06,
      "loss": 0.2833,
      "step": 32894
    },
    {
      "epoch": 1.5095681703455555,
      "grad_norm": 0.4654155671596527,
      "learning_rate": 5.109222524233484e-06,
      "loss": 0.3303,
      "step": 32895
    },
    {
      "epoch": 1.50961406085081,
      "grad_norm": 0.47166234254837036,
      "learning_rate": 5.1089773947018215e-06,
      "loss": 0.377,
      "step": 32896
    },
    {
      "epoch": 1.5096599513560645,
      "grad_norm": 0.48635968565940857,
      "learning_rate": 5.108732264908104e-06,
      "loss": 0.3967,
      "step": 32897
    },
    {
      "epoch": 1.5097058418613187,
      "grad_norm": 0.47026216983795166,
      "learning_rate": 5.108487134852919e-06,
      "loss": 0.3459,
      "step": 32898
    },
    {
      "epoch": 1.5097517323665732,
      "grad_norm": 0.43051543831825256,
      "learning_rate": 5.108242004536857e-06,
      "loss": 0.3146,
      "step": 32899
    },
    {
      "epoch": 1.5097976228718277,
      "grad_norm": 0.4713224172592163,
      "learning_rate": 5.107996873960508e-06,
      "loss": 0.3799,
      "step": 32900
    },
    {
      "epoch": 1.5098435133770822,
      "grad_norm": 0.4791678190231323,
      "learning_rate": 5.107751743124458e-06,
      "loss": 0.3621,
      "step": 32901
    },
    {
      "epoch": 1.5098894038823367,
      "grad_norm": 0.4685980975627899,
      "learning_rate": 5.1075066120293e-06,
      "loss": 0.4133,
      "step": 32902
    },
    {
      "epoch": 1.5099352943875912,
      "grad_norm": 0.4428599774837494,
      "learning_rate": 5.107261480675625e-06,
      "loss": 0.3247,
      "step": 32903
    },
    {
      "epoch": 1.5099811848928457,
      "grad_norm": 0.5112189054489136,
      "learning_rate": 5.107016349064018e-06,
      "loss": 0.3064,
      "step": 32904
    },
    {
      "epoch": 1.5100270753981002,
      "grad_norm": 0.549166202545166,
      "learning_rate": 5.10677121719507e-06,
      "loss": 0.4021,
      "step": 32905
    },
    {
      "epoch": 1.5100729659033547,
      "grad_norm": 0.45137226581573486,
      "learning_rate": 5.106526085069372e-06,
      "loss": 0.3501,
      "step": 32906
    },
    {
      "epoch": 1.5101188564086092,
      "grad_norm": 0.4434645473957062,
      "learning_rate": 5.106280952687512e-06,
      "loss": 0.2721,
      "step": 32907
    },
    {
      "epoch": 1.5101647469138635,
      "grad_norm": 0.4948899745941162,
      "learning_rate": 5.106035820050081e-06,
      "loss": 0.4279,
      "step": 32908
    },
    {
      "epoch": 1.510210637419118,
      "grad_norm": 0.4634902775287628,
      "learning_rate": 5.105790687157666e-06,
      "loss": 0.3354,
      "step": 32909
    },
    {
      "epoch": 1.5102565279243725,
      "grad_norm": 0.45112544298171997,
      "learning_rate": 5.105545554010858e-06,
      "loss": 0.3014,
      "step": 32910
    },
    {
      "epoch": 1.5103024184296268,
      "grad_norm": 0.5241010785102844,
      "learning_rate": 5.105300420610246e-06,
      "loss": 0.4362,
      "step": 32911
    },
    {
      "epoch": 1.5103483089348813,
      "grad_norm": 0.4839531183242798,
      "learning_rate": 5.10505528695642e-06,
      "loss": 0.3111,
      "step": 32912
    },
    {
      "epoch": 1.5103941994401358,
      "grad_norm": 0.5071544647216797,
      "learning_rate": 5.104810153049969e-06,
      "loss": 0.3804,
      "step": 32913
    },
    {
      "epoch": 1.5104400899453903,
      "grad_norm": 0.46169421076774597,
      "learning_rate": 5.104565018891482e-06,
      "loss": 0.3777,
      "step": 32914
    },
    {
      "epoch": 1.5104859804506447,
      "grad_norm": 0.44316917657852173,
      "learning_rate": 5.104319884481551e-06,
      "loss": 0.3275,
      "step": 32915
    },
    {
      "epoch": 1.5105318709558992,
      "grad_norm": 0.46173954010009766,
      "learning_rate": 5.104074749820762e-06,
      "loss": 0.3838,
      "step": 32916
    },
    {
      "epoch": 1.5105777614611537,
      "grad_norm": 0.46743935346603394,
      "learning_rate": 5.103829614909707e-06,
      "loss": 0.3278,
      "step": 32917
    },
    {
      "epoch": 1.5106236519664082,
      "grad_norm": 0.47058355808258057,
      "learning_rate": 5.1035844797489735e-06,
      "loss": 0.3732,
      "step": 32918
    },
    {
      "epoch": 1.5106695424716627,
      "grad_norm": 0.4932444095611572,
      "learning_rate": 5.103339344339153e-06,
      "loss": 0.3802,
      "step": 32919
    },
    {
      "epoch": 1.5107154329769172,
      "grad_norm": 0.47195911407470703,
      "learning_rate": 5.1030942086808346e-06,
      "loss": 0.3793,
      "step": 32920
    },
    {
      "epoch": 1.5107613234821715,
      "grad_norm": 0.4412645995616913,
      "learning_rate": 5.102849072774605e-06,
      "loss": 0.3213,
      "step": 32921
    },
    {
      "epoch": 1.510807213987426,
      "grad_norm": 0.4529319703578949,
      "learning_rate": 5.102603936621056e-06,
      "loss": 0.3323,
      "step": 32922
    },
    {
      "epoch": 1.5108531044926805,
      "grad_norm": 0.4406147599220276,
      "learning_rate": 5.102358800220779e-06,
      "loss": 0.3213,
      "step": 32923
    },
    {
      "epoch": 1.5108989949979348,
      "grad_norm": 0.4675256907939911,
      "learning_rate": 5.1021136635743594e-06,
      "loss": 0.3726,
      "step": 32924
    },
    {
      "epoch": 1.5109448855031893,
      "grad_norm": 0.4172089993953705,
      "learning_rate": 5.101868526682391e-06,
      "loss": 0.296,
      "step": 32925
    },
    {
      "epoch": 1.5109907760084438,
      "grad_norm": 0.4600245952606201,
      "learning_rate": 5.101623389545459e-06,
      "loss": 0.3414,
      "step": 32926
    },
    {
      "epoch": 1.5110366665136983,
      "grad_norm": 0.4454044699668884,
      "learning_rate": 5.101378252164155e-06,
      "loss": 0.3407,
      "step": 32927
    },
    {
      "epoch": 1.5110825570189528,
      "grad_norm": 0.49111026525497437,
      "learning_rate": 5.101133114539069e-06,
      "loss": 0.3794,
      "step": 32928
    },
    {
      "epoch": 1.5111284475242073,
      "grad_norm": 0.47979646921157837,
      "learning_rate": 5.10088797667079e-06,
      "loss": 0.3699,
      "step": 32929
    },
    {
      "epoch": 1.5111743380294618,
      "grad_norm": 0.4436599910259247,
      "learning_rate": 5.100642838559907e-06,
      "loss": 0.3358,
      "step": 32930
    },
    {
      "epoch": 1.5112202285347163,
      "grad_norm": 0.49318814277648926,
      "learning_rate": 5.100397700207009e-06,
      "loss": 0.4268,
      "step": 32931
    },
    {
      "epoch": 1.5112661190399708,
      "grad_norm": 0.4757654368877411,
      "learning_rate": 5.100152561612687e-06,
      "loss": 0.3318,
      "step": 32932
    },
    {
      "epoch": 1.511312009545225,
      "grad_norm": 0.47588858008384705,
      "learning_rate": 5.099907422777529e-06,
      "loss": 0.3609,
      "step": 32933
    },
    {
      "epoch": 1.5113579000504795,
      "grad_norm": 0.4523887634277344,
      "learning_rate": 5.099662283702126e-06,
      "loss": 0.3118,
      "step": 32934
    },
    {
      "epoch": 1.511403790555734,
      "grad_norm": 0.5391321778297424,
      "learning_rate": 5.0994171443870654e-06,
      "loss": 0.4016,
      "step": 32935
    },
    {
      "epoch": 1.5114496810609885,
      "grad_norm": 0.48915722966194153,
      "learning_rate": 5.09917200483294e-06,
      "loss": 0.4505,
      "step": 32936
    },
    {
      "epoch": 1.5114955715662428,
      "grad_norm": 0.4565589427947998,
      "learning_rate": 5.098926865040336e-06,
      "loss": 0.3421,
      "step": 32937
    },
    {
      "epoch": 1.5115414620714973,
      "grad_norm": 0.4710758626461029,
      "learning_rate": 5.098681725009843e-06,
      "loss": 0.3861,
      "step": 32938
    },
    {
      "epoch": 1.5115873525767518,
      "grad_norm": 0.5026084184646606,
      "learning_rate": 5.0984365847420524e-06,
      "loss": 0.4225,
      "step": 32939
    },
    {
      "epoch": 1.5116332430820063,
      "grad_norm": 0.4483088552951813,
      "learning_rate": 5.098191444237553e-06,
      "loss": 0.3192,
      "step": 32940
    },
    {
      "epoch": 1.5116791335872608,
      "grad_norm": 0.45624521374702454,
      "learning_rate": 5.0979463034969355e-06,
      "loss": 0.257,
      "step": 32941
    },
    {
      "epoch": 1.5117250240925153,
      "grad_norm": 0.4575149118900299,
      "learning_rate": 5.097701162520787e-06,
      "loss": 0.3217,
      "step": 32942
    },
    {
      "epoch": 1.5117709145977698,
      "grad_norm": 0.4449019134044647,
      "learning_rate": 5.097456021309698e-06,
      "loss": 0.3016,
      "step": 32943
    },
    {
      "epoch": 1.5118168051030243,
      "grad_norm": 0.5556275844573975,
      "learning_rate": 5.097210879864258e-06,
      "loss": 0.4079,
      "step": 32944
    },
    {
      "epoch": 1.5118626956082788,
      "grad_norm": 0.5077406764030457,
      "learning_rate": 5.096965738185056e-06,
      "loss": 0.3994,
      "step": 32945
    },
    {
      "epoch": 1.511908586113533,
      "grad_norm": 0.46166062355041504,
      "learning_rate": 5.0967205962726815e-06,
      "loss": 0.3027,
      "step": 32946
    },
    {
      "epoch": 1.5119544766187876,
      "grad_norm": 0.4965742528438568,
      "learning_rate": 5.096475454127726e-06,
      "loss": 0.3712,
      "step": 32947
    },
    {
      "epoch": 1.512000367124042,
      "grad_norm": 0.4740668535232544,
      "learning_rate": 5.096230311750779e-06,
      "loss": 0.3701,
      "step": 32948
    },
    {
      "epoch": 1.5120462576292963,
      "grad_norm": 0.4472162425518036,
      "learning_rate": 5.095985169142424e-06,
      "loss": 0.3122,
      "step": 32949
    },
    {
      "epoch": 1.5120921481345508,
      "grad_norm": 0.4320763349533081,
      "learning_rate": 5.095740026303258e-06,
      "loss": 0.3196,
      "step": 32950
    },
    {
      "epoch": 1.5121380386398053,
      "grad_norm": 0.44756749272346497,
      "learning_rate": 5.0954948832338666e-06,
      "loss": 0.3226,
      "step": 32951
    },
    {
      "epoch": 1.5121839291450598,
      "grad_norm": 0.5062116980552673,
      "learning_rate": 5.095249739934841e-06,
      "loss": 0.4349,
      "step": 32952
    },
    {
      "epoch": 1.5122298196503143,
      "grad_norm": 0.4963386356830597,
      "learning_rate": 5.095004596406769e-06,
      "loss": 0.3861,
      "step": 32953
    },
    {
      "epoch": 1.5122757101555688,
      "grad_norm": 0.49955645203590393,
      "learning_rate": 5.0947594526502416e-06,
      "loss": 0.4271,
      "step": 32954
    },
    {
      "epoch": 1.5123216006608233,
      "grad_norm": 0.4615998864173889,
      "learning_rate": 5.094514308665847e-06,
      "loss": 0.3515,
      "step": 32955
    },
    {
      "epoch": 1.5123674911660778,
      "grad_norm": 0.4534638822078705,
      "learning_rate": 5.094269164454174e-06,
      "loss": 0.3241,
      "step": 32956
    },
    {
      "epoch": 1.5124133816713323,
      "grad_norm": 0.4982343316078186,
      "learning_rate": 5.094024020015817e-06,
      "loss": 0.3808,
      "step": 32957
    },
    {
      "epoch": 1.5124592721765868,
      "grad_norm": 0.4697071611881256,
      "learning_rate": 5.093778875351359e-06,
      "loss": 0.3814,
      "step": 32958
    },
    {
      "epoch": 1.512505162681841,
      "grad_norm": 0.4335629940032959,
      "learning_rate": 5.093533730461394e-06,
      "loss": 0.3123,
      "step": 32959
    },
    {
      "epoch": 1.5125510531870956,
      "grad_norm": 0.4794343113899231,
      "learning_rate": 5.093288585346509e-06,
      "loss": 0.3598,
      "step": 32960
    },
    {
      "epoch": 1.51259694369235,
      "grad_norm": 0.4779466986656189,
      "learning_rate": 5.093043440007294e-06,
      "loss": 0.3088,
      "step": 32961
    },
    {
      "epoch": 1.5126428341976044,
      "grad_norm": 0.47442707419395447,
      "learning_rate": 5.09279829444434e-06,
      "loss": 0.3799,
      "step": 32962
    },
    {
      "epoch": 1.5126887247028589,
      "grad_norm": 0.4841737151145935,
      "learning_rate": 5.0925531486582344e-06,
      "loss": 0.3631,
      "step": 32963
    },
    {
      "epoch": 1.5127346152081134,
      "grad_norm": 0.44799405336380005,
      "learning_rate": 5.092308002649567e-06,
      "loss": 0.3415,
      "step": 32964
    },
    {
      "epoch": 1.5127805057133679,
      "grad_norm": 0.4868413209915161,
      "learning_rate": 5.09206285641893e-06,
      "loss": 0.3745,
      "step": 32965
    },
    {
      "epoch": 1.5128263962186224,
      "grad_norm": 0.4384042024612427,
      "learning_rate": 5.09181770996691e-06,
      "loss": 0.2923,
      "step": 32966
    },
    {
      "epoch": 1.5128722867238769,
      "grad_norm": 0.5177088379859924,
      "learning_rate": 5.091572563294096e-06,
      "loss": 0.453,
      "step": 32967
    },
    {
      "epoch": 1.5129181772291314,
      "grad_norm": 0.5143230557441711,
      "learning_rate": 5.091327416401082e-06,
      "loss": 0.4383,
      "step": 32968
    },
    {
      "epoch": 1.5129640677343859,
      "grad_norm": 0.45902279019355774,
      "learning_rate": 5.091082269288452e-06,
      "loss": 0.3324,
      "step": 32969
    },
    {
      "epoch": 1.5130099582396404,
      "grad_norm": 0.447106271982193,
      "learning_rate": 5.090837121956799e-06,
      "loss": 0.2901,
      "step": 32970
    },
    {
      "epoch": 1.5130558487448946,
      "grad_norm": 0.4388471245765686,
      "learning_rate": 5.090591974406711e-06,
      "loss": 0.3315,
      "step": 32971
    },
    {
      "epoch": 1.5131017392501491,
      "grad_norm": 0.43902021646499634,
      "learning_rate": 5.090346826638777e-06,
      "loss": 0.32,
      "step": 32972
    },
    {
      "epoch": 1.5131476297554036,
      "grad_norm": 0.47846272587776184,
      "learning_rate": 5.0901016786535885e-06,
      "loss": 0.3861,
      "step": 32973
    },
    {
      "epoch": 1.5131935202606581,
      "grad_norm": 0.4471953809261322,
      "learning_rate": 5.089856530451734e-06,
      "loss": 0.3023,
      "step": 32974
    },
    {
      "epoch": 1.5132394107659124,
      "grad_norm": 0.4588596522808075,
      "learning_rate": 5.089611382033802e-06,
      "loss": 0.3268,
      "step": 32975
    },
    {
      "epoch": 1.513285301271167,
      "grad_norm": 0.47428715229034424,
      "learning_rate": 5.089366233400384e-06,
      "loss": 0.38,
      "step": 32976
    },
    {
      "epoch": 1.5133311917764214,
      "grad_norm": 0.4670547842979431,
      "learning_rate": 5.089121084552066e-06,
      "loss": 0.3931,
      "step": 32977
    },
    {
      "epoch": 1.513377082281676,
      "grad_norm": 0.4484016001224518,
      "learning_rate": 5.088875935489442e-06,
      "loss": 0.3377,
      "step": 32978
    },
    {
      "epoch": 1.5134229727869304,
      "grad_norm": 0.4753055274486542,
      "learning_rate": 5.0886307862131e-06,
      "loss": 0.3814,
      "step": 32979
    },
    {
      "epoch": 1.5134688632921849,
      "grad_norm": 0.4561261832714081,
      "learning_rate": 5.088385636723628e-06,
      "loss": 0.3688,
      "step": 32980
    },
    {
      "epoch": 1.5135147537974394,
      "grad_norm": 0.4856032431125641,
      "learning_rate": 5.088140487021617e-06,
      "loss": 0.4027,
      "step": 32981
    },
    {
      "epoch": 1.5135606443026939,
      "grad_norm": 0.4726117253303528,
      "learning_rate": 5.087895337107656e-06,
      "loss": 0.3359,
      "step": 32982
    },
    {
      "epoch": 1.5136065348079484,
      "grad_norm": 0.4547535479068756,
      "learning_rate": 5.087650186982333e-06,
      "loss": 0.3491,
      "step": 32983
    },
    {
      "epoch": 1.5136524253132027,
      "grad_norm": 0.4858349859714508,
      "learning_rate": 5.08740503664624e-06,
      "loss": 0.3907,
      "step": 32984
    },
    {
      "epoch": 1.5136983158184572,
      "grad_norm": 0.4974118769168854,
      "learning_rate": 5.087159886099966e-06,
      "loss": 0.3962,
      "step": 32985
    },
    {
      "epoch": 1.5137442063237116,
      "grad_norm": 0.4718133211135864,
      "learning_rate": 5.086914735344099e-06,
      "loss": 0.339,
      "step": 32986
    },
    {
      "epoch": 1.513790096828966,
      "grad_norm": 0.46721309423446655,
      "learning_rate": 5.086669584379229e-06,
      "loss": 0.3667,
      "step": 32987
    },
    {
      "epoch": 1.5138359873342204,
      "grad_norm": 0.4504622220993042,
      "learning_rate": 5.086424433205948e-06,
      "loss": 0.3165,
      "step": 32988
    },
    {
      "epoch": 1.513881877839475,
      "grad_norm": 0.4800935387611389,
      "learning_rate": 5.086179281824841e-06,
      "loss": 0.346,
      "step": 32989
    },
    {
      "epoch": 1.5139277683447294,
      "grad_norm": 0.4713888466358185,
      "learning_rate": 5.0859341302365016e-06,
      "loss": 0.3486,
      "step": 32990
    },
    {
      "epoch": 1.513973658849984,
      "grad_norm": 0.5469886660575867,
      "learning_rate": 5.085688978441518e-06,
      "loss": 0.443,
      "step": 32991
    },
    {
      "epoch": 1.5140195493552384,
      "grad_norm": 0.4701731503009796,
      "learning_rate": 5.085443826440479e-06,
      "loss": 0.3332,
      "step": 32992
    },
    {
      "epoch": 1.514065439860493,
      "grad_norm": 0.4823015630245209,
      "learning_rate": 5.085198674233975e-06,
      "loss": 0.3761,
      "step": 32993
    },
    {
      "epoch": 1.5141113303657474,
      "grad_norm": 0.4821377992630005,
      "learning_rate": 5.084953521822594e-06,
      "loss": 0.3324,
      "step": 32994
    },
    {
      "epoch": 1.514157220871002,
      "grad_norm": 0.5180010795593262,
      "learning_rate": 5.0847083692069276e-06,
      "loss": 0.4829,
      "step": 32995
    },
    {
      "epoch": 1.5142031113762564,
      "grad_norm": 0.4725053012371063,
      "learning_rate": 5.084463216387565e-06,
      "loss": 0.378,
      "step": 32996
    },
    {
      "epoch": 1.5142490018815107,
      "grad_norm": 0.4603176712989807,
      "learning_rate": 5.084218063365093e-06,
      "loss": 0.3552,
      "step": 32997
    },
    {
      "epoch": 1.5142948923867652,
      "grad_norm": 0.47787052392959595,
      "learning_rate": 5.083972910140104e-06,
      "loss": 0.3515,
      "step": 32998
    },
    {
      "epoch": 1.5143407828920197,
      "grad_norm": 0.49880218505859375,
      "learning_rate": 5.083727756713187e-06,
      "loss": 0.427,
      "step": 32999
    },
    {
      "epoch": 1.514386673397274,
      "grad_norm": 0.4680313766002655,
      "learning_rate": 5.0834826030849305e-06,
      "loss": 0.349,
      "step": 33000
    },
    {
      "epoch": 1.5144325639025285,
      "grad_norm": 0.4542843997478485,
      "learning_rate": 5.083237449255926e-06,
      "loss": 0.3091,
      "step": 33001
    },
    {
      "epoch": 1.514478454407783,
      "grad_norm": 0.5088198184967041,
      "learning_rate": 5.082992295226761e-06,
      "loss": 0.4181,
      "step": 33002
    },
    {
      "epoch": 1.5145243449130374,
      "grad_norm": 0.45194828510284424,
      "learning_rate": 5.082747140998024e-06,
      "loss": 0.3156,
      "step": 33003
    },
    {
      "epoch": 1.514570235418292,
      "grad_norm": 0.46553921699523926,
      "learning_rate": 5.0825019865703075e-06,
      "loss": 0.3532,
      "step": 33004
    },
    {
      "epoch": 1.5146161259235464,
      "grad_norm": 0.49687710404396057,
      "learning_rate": 5.0822568319442e-06,
      "loss": 0.4048,
      "step": 33005
    },
    {
      "epoch": 1.514662016428801,
      "grad_norm": 0.42273738980293274,
      "learning_rate": 5.082011677120292e-06,
      "loss": 0.2688,
      "step": 33006
    },
    {
      "epoch": 1.5147079069340554,
      "grad_norm": 0.45425301790237427,
      "learning_rate": 5.081766522099171e-06,
      "loss": 0.3582,
      "step": 33007
    },
    {
      "epoch": 1.51475379743931,
      "grad_norm": 0.4508279860019684,
      "learning_rate": 5.081521366881425e-06,
      "loss": 0.3351,
      "step": 33008
    },
    {
      "epoch": 1.5147996879445644,
      "grad_norm": 0.5306543111801147,
      "learning_rate": 5.081276211467649e-06,
      "loss": 0.4035,
      "step": 33009
    },
    {
      "epoch": 1.5148455784498187,
      "grad_norm": 0.4281502962112427,
      "learning_rate": 5.081031055858428e-06,
      "loss": 0.2977,
      "step": 33010
    },
    {
      "epoch": 1.5148914689550732,
      "grad_norm": 0.5367514491081238,
      "learning_rate": 5.080785900054352e-06,
      "loss": 0.3218,
      "step": 33011
    },
    {
      "epoch": 1.5149373594603277,
      "grad_norm": 0.4951612651348114,
      "learning_rate": 5.080540744056013e-06,
      "loss": 0.3737,
      "step": 33012
    },
    {
      "epoch": 1.514983249965582,
      "grad_norm": 0.49979129433631897,
      "learning_rate": 5.0802955878639995e-06,
      "loss": 0.4201,
      "step": 33013
    },
    {
      "epoch": 1.5150291404708365,
      "grad_norm": 0.4428306519985199,
      "learning_rate": 5.080050431478898e-06,
      "loss": 0.3516,
      "step": 33014
    },
    {
      "epoch": 1.515075030976091,
      "grad_norm": 0.491031289100647,
      "learning_rate": 5.079805274901302e-06,
      "loss": 0.348,
      "step": 33015
    },
    {
      "epoch": 1.5151209214813455,
      "grad_norm": 0.4457719027996063,
      "learning_rate": 5.079560118131799e-06,
      "loss": 0.2981,
      "step": 33016
    },
    {
      "epoch": 1.5151668119866,
      "grad_norm": 0.4750492572784424,
      "learning_rate": 5.0793149611709806e-06,
      "loss": 0.3848,
      "step": 33017
    },
    {
      "epoch": 1.5152127024918545,
      "grad_norm": 0.47274166345596313,
      "learning_rate": 5.079069804019432e-06,
      "loss": 0.3604,
      "step": 33018
    },
    {
      "epoch": 1.515258592997109,
      "grad_norm": 0.48137491941452026,
      "learning_rate": 5.078824646677748e-06,
      "loss": 0.3937,
      "step": 33019
    },
    {
      "epoch": 1.5153044835023635,
      "grad_norm": 0.48941293358802795,
      "learning_rate": 5.078579489146513e-06,
      "loss": 0.3785,
      "step": 33020
    },
    {
      "epoch": 1.515350374007618,
      "grad_norm": 0.46263161301612854,
      "learning_rate": 5.078334331426321e-06,
      "loss": 0.316,
      "step": 33021
    },
    {
      "epoch": 1.5153962645128722,
      "grad_norm": 0.4745839834213257,
      "learning_rate": 5.078089173517759e-06,
      "loss": 0.3489,
      "step": 33022
    },
    {
      "epoch": 1.5154421550181267,
      "grad_norm": 0.4619411826133728,
      "learning_rate": 5.077844015421418e-06,
      "loss": 0.3348,
      "step": 33023
    },
    {
      "epoch": 1.5154880455233812,
      "grad_norm": 0.49199873208999634,
      "learning_rate": 5.0775988571378875e-06,
      "loss": 0.4442,
      "step": 33024
    },
    {
      "epoch": 1.5155339360286355,
      "grad_norm": 0.45103219151496887,
      "learning_rate": 5.0773536986677544e-06,
      "loss": 0.3184,
      "step": 33025
    },
    {
      "epoch": 1.51557982653389,
      "grad_norm": 0.4490686357021332,
      "learning_rate": 5.0771085400116096e-06,
      "loss": 0.3403,
      "step": 33026
    },
    {
      "epoch": 1.5156257170391445,
      "grad_norm": 0.4395928680896759,
      "learning_rate": 5.076863381170045e-06,
      "loss": 0.3235,
      "step": 33027
    },
    {
      "epoch": 1.515671607544399,
      "grad_norm": 0.45402446389198303,
      "learning_rate": 5.0766182221436475e-06,
      "loss": 0.3164,
      "step": 33028
    },
    {
      "epoch": 1.5157174980496535,
      "grad_norm": 0.545629620552063,
      "learning_rate": 5.076373062933007e-06,
      "loss": 0.3969,
      "step": 33029
    },
    {
      "epoch": 1.515763388554908,
      "grad_norm": 0.4608049988746643,
      "learning_rate": 5.0761279035387135e-06,
      "loss": 0.3336,
      "step": 33030
    },
    {
      "epoch": 1.5158092790601625,
      "grad_norm": 0.4742632806301117,
      "learning_rate": 5.075882743961357e-06,
      "loss": 0.3575,
      "step": 33031
    },
    {
      "epoch": 1.515855169565417,
      "grad_norm": 0.4876706898212433,
      "learning_rate": 5.0756375842015245e-06,
      "loss": 0.3916,
      "step": 33032
    },
    {
      "epoch": 1.5159010600706715,
      "grad_norm": 0.43666982650756836,
      "learning_rate": 5.07539242425981e-06,
      "loss": 0.3123,
      "step": 33033
    },
    {
      "epoch": 1.515946950575926,
      "grad_norm": 0.4772433936595917,
      "learning_rate": 5.0751472641367995e-06,
      "loss": 0.3999,
      "step": 33034
    },
    {
      "epoch": 1.5159928410811803,
      "grad_norm": 0.46065327525138855,
      "learning_rate": 5.074902103833082e-06,
      "loss": 0.3581,
      "step": 33035
    },
    {
      "epoch": 1.5160387315864348,
      "grad_norm": 0.44228842854499817,
      "learning_rate": 5.074656943349252e-06,
      "loss": 0.3082,
      "step": 33036
    },
    {
      "epoch": 1.5160846220916893,
      "grad_norm": 0.5365902781486511,
      "learning_rate": 5.074411782685893e-06,
      "loss": 0.4621,
      "step": 33037
    },
    {
      "epoch": 1.5161305125969435,
      "grad_norm": 0.46863114833831787,
      "learning_rate": 5.074166621843598e-06,
      "loss": 0.3259,
      "step": 33038
    },
    {
      "epoch": 1.516176403102198,
      "grad_norm": 0.45192092657089233,
      "learning_rate": 5.073921460822956e-06,
      "loss": 0.322,
      "step": 33039
    },
    {
      "epoch": 1.5162222936074525,
      "grad_norm": 0.47045275568962097,
      "learning_rate": 5.0736762996245546e-06,
      "loss": 0.3892,
      "step": 33040
    },
    {
      "epoch": 1.516268184112707,
      "grad_norm": 0.45325735211372375,
      "learning_rate": 5.0734311382489875e-06,
      "loss": 0.342,
      "step": 33041
    },
    {
      "epoch": 1.5163140746179615,
      "grad_norm": 0.46869683265686035,
      "learning_rate": 5.0731859766968395e-06,
      "loss": 0.3876,
      "step": 33042
    },
    {
      "epoch": 1.516359965123216,
      "grad_norm": 0.5243753790855408,
      "learning_rate": 5.072940814968702e-06,
      "loss": 0.4647,
      "step": 33043
    },
    {
      "epoch": 1.5164058556284705,
      "grad_norm": 0.4870217442512512,
      "learning_rate": 5.072695653065167e-06,
      "loss": 0.3562,
      "step": 33044
    },
    {
      "epoch": 1.516451746133725,
      "grad_norm": 0.4675758481025696,
      "learning_rate": 5.072450490986821e-06,
      "loss": 0.3736,
      "step": 33045
    },
    {
      "epoch": 1.5164976366389795,
      "grad_norm": 0.49644336104393005,
      "learning_rate": 5.072205328734254e-06,
      "loss": 0.3937,
      "step": 33046
    },
    {
      "epoch": 1.516543527144234,
      "grad_norm": 0.42883315682411194,
      "learning_rate": 5.071960166308056e-06,
      "loss": 0.2877,
      "step": 33047
    },
    {
      "epoch": 1.5165894176494883,
      "grad_norm": 0.4704356789588928,
      "learning_rate": 5.071715003708817e-06,
      "loss": 0.3175,
      "step": 33048
    },
    {
      "epoch": 1.5166353081547428,
      "grad_norm": 0.4845125377178192,
      "learning_rate": 5.071469840937125e-06,
      "loss": 0.3531,
      "step": 33049
    },
    {
      "epoch": 1.5166811986599973,
      "grad_norm": 0.4675728678703308,
      "learning_rate": 5.071224677993573e-06,
      "loss": 0.3955,
      "step": 33050
    },
    {
      "epoch": 1.5167270891652516,
      "grad_norm": 0.47753751277923584,
      "learning_rate": 5.070979514878747e-06,
      "loss": 0.3734,
      "step": 33051
    },
    {
      "epoch": 1.516772979670506,
      "grad_norm": 0.46507972478866577,
      "learning_rate": 5.070734351593236e-06,
      "loss": 0.3556,
      "step": 33052
    },
    {
      "epoch": 1.5168188701757606,
      "grad_norm": 0.4897947907447815,
      "learning_rate": 5.070489188137633e-06,
      "loss": 0.4292,
      "step": 33053
    },
    {
      "epoch": 1.516864760681015,
      "grad_norm": 0.46886664628982544,
      "learning_rate": 5.070244024512524e-06,
      "loss": 0.3781,
      "step": 33054
    },
    {
      "epoch": 1.5169106511862696,
      "grad_norm": 0.46459871530532837,
      "learning_rate": 5.069998860718503e-06,
      "loss": 0.3459,
      "step": 33055
    },
    {
      "epoch": 1.516956541691524,
      "grad_norm": 0.47577965259552,
      "learning_rate": 5.0697536967561555e-06,
      "loss": 0.3872,
      "step": 33056
    },
    {
      "epoch": 1.5170024321967785,
      "grad_norm": 0.4306287467479706,
      "learning_rate": 5.06950853262607e-06,
      "loss": 0.3193,
      "step": 33057
    },
    {
      "epoch": 1.517048322702033,
      "grad_norm": 0.44365522265434265,
      "learning_rate": 5.069263368328843e-06,
      "loss": 0.313,
      "step": 33058
    },
    {
      "epoch": 1.5170942132072875,
      "grad_norm": 0.45083296298980713,
      "learning_rate": 5.069018203865056e-06,
      "loss": 0.28,
      "step": 33059
    },
    {
      "epoch": 1.5171401037125418,
      "grad_norm": 0.4979355037212372,
      "learning_rate": 5.068773039235302e-06,
      "loss": 0.3927,
      "step": 33060
    },
    {
      "epoch": 1.5171859942177963,
      "grad_norm": 0.4147357940673828,
      "learning_rate": 5.068527874440173e-06,
      "loss": 0.2573,
      "step": 33061
    },
    {
      "epoch": 1.5172318847230508,
      "grad_norm": 0.46807530522346497,
      "learning_rate": 5.068282709480254e-06,
      "loss": 0.3594,
      "step": 33062
    },
    {
      "epoch": 1.5172777752283053,
      "grad_norm": 0.47481852769851685,
      "learning_rate": 5.068037544356137e-06,
      "loss": 0.3624,
      "step": 33063
    },
    {
      "epoch": 1.5173236657335596,
      "grad_norm": 0.4877435564994812,
      "learning_rate": 5.067792379068413e-06,
      "loss": 0.4044,
      "step": 33064
    },
    {
      "epoch": 1.517369556238814,
      "grad_norm": 0.46689414978027344,
      "learning_rate": 5.067547213617668e-06,
      "loss": 0.3507,
      "step": 33065
    },
    {
      "epoch": 1.5174154467440686,
      "grad_norm": 0.5162866711616516,
      "learning_rate": 5.067302048004493e-06,
      "loss": 0.3488,
      "step": 33066
    },
    {
      "epoch": 1.517461337249323,
      "grad_norm": 0.4684656262397766,
      "learning_rate": 5.067056882229478e-06,
      "loss": 0.3723,
      "step": 33067
    },
    {
      "epoch": 1.5175072277545776,
      "grad_norm": 0.4358789324760437,
      "learning_rate": 5.066811716293213e-06,
      "loss": 0.2943,
      "step": 33068
    },
    {
      "epoch": 1.517553118259832,
      "grad_norm": 0.5011264681816101,
      "learning_rate": 5.066566550196285e-06,
      "loss": 0.3924,
      "step": 33069
    },
    {
      "epoch": 1.5175990087650866,
      "grad_norm": 0.4692758619785309,
      "learning_rate": 5.066321383939288e-06,
      "loss": 0.3895,
      "step": 33070
    },
    {
      "epoch": 1.517644899270341,
      "grad_norm": 0.46320345997810364,
      "learning_rate": 5.0660762175228075e-06,
      "loss": 0.39,
      "step": 33071
    },
    {
      "epoch": 1.5176907897755956,
      "grad_norm": 0.46558186411857605,
      "learning_rate": 5.065831050947437e-06,
      "loss": 0.3546,
      "step": 33072
    },
    {
      "epoch": 1.5177366802808498,
      "grad_norm": 0.4773818850517273,
      "learning_rate": 5.065585884213761e-06,
      "loss": 0.3487,
      "step": 33073
    },
    {
      "epoch": 1.5177825707861043,
      "grad_norm": 0.4680628478527069,
      "learning_rate": 5.065340717322372e-06,
      "loss": 0.3576,
      "step": 33074
    },
    {
      "epoch": 1.5178284612913588,
      "grad_norm": 0.45360931754112244,
      "learning_rate": 5.0650955502738595e-06,
      "loss": 0.3144,
      "step": 33075
    },
    {
      "epoch": 1.5178743517966131,
      "grad_norm": 0.4445614516735077,
      "learning_rate": 5.064850383068813e-06,
      "loss": 0.3297,
      "step": 33076
    },
    {
      "epoch": 1.5179202423018676,
      "grad_norm": 0.48551374673843384,
      "learning_rate": 5.064605215707822e-06,
      "loss": 0.4392,
      "step": 33077
    },
    {
      "epoch": 1.5179661328071221,
      "grad_norm": 0.529626727104187,
      "learning_rate": 5.064360048191475e-06,
      "loss": 0.4724,
      "step": 33078
    },
    {
      "epoch": 1.5180120233123766,
      "grad_norm": 0.47567227482795715,
      "learning_rate": 5.064114880520363e-06,
      "loss": 0.3656,
      "step": 33079
    },
    {
      "epoch": 1.518057913817631,
      "grad_norm": 0.4787372052669525,
      "learning_rate": 5.0638697126950746e-06,
      "loss": 0.374,
      "step": 33080
    },
    {
      "epoch": 1.5181038043228856,
      "grad_norm": 0.44507479667663574,
      "learning_rate": 5.0636245447162004e-06,
      "loss": 0.3376,
      "step": 33081
    },
    {
      "epoch": 1.51814969482814,
      "grad_norm": 0.46953943371772766,
      "learning_rate": 5.063379376584328e-06,
      "loss": 0.3931,
      "step": 33082
    },
    {
      "epoch": 1.5181955853333946,
      "grad_norm": 0.4642258882522583,
      "learning_rate": 5.063134208300049e-06,
      "loss": 0.3238,
      "step": 33083
    },
    {
      "epoch": 1.518241475838649,
      "grad_norm": 0.45909154415130615,
      "learning_rate": 5.062889039863953e-06,
      "loss": 0.3199,
      "step": 33084
    },
    {
      "epoch": 1.5182873663439036,
      "grad_norm": 0.45715761184692383,
      "learning_rate": 5.062643871276627e-06,
      "loss": 0.3142,
      "step": 33085
    },
    {
      "epoch": 1.5183332568491579,
      "grad_norm": 0.481799840927124,
      "learning_rate": 5.062398702538662e-06,
      "loss": 0.4009,
      "step": 33086
    },
    {
      "epoch": 1.5183791473544124,
      "grad_norm": 0.4709456264972687,
      "learning_rate": 5.06215353365065e-06,
      "loss": 0.3932,
      "step": 33087
    },
    {
      "epoch": 1.5184250378596669,
      "grad_norm": 0.44289568066596985,
      "learning_rate": 5.061908364613177e-06,
      "loss": 0.2747,
      "step": 33088
    },
    {
      "epoch": 1.5184709283649211,
      "grad_norm": 0.4894202649593353,
      "learning_rate": 5.061663195426835e-06,
      "loss": 0.3492,
      "step": 33089
    },
    {
      "epoch": 1.5185168188701756,
      "grad_norm": 0.4981745183467865,
      "learning_rate": 5.061418026092211e-06,
      "loss": 0.4332,
      "step": 33090
    },
    {
      "epoch": 1.5185627093754301,
      "grad_norm": 0.4492800235748291,
      "learning_rate": 5.0611728566098965e-06,
      "loss": 0.3548,
      "step": 33091
    },
    {
      "epoch": 1.5186085998806846,
      "grad_norm": 0.46527099609375,
      "learning_rate": 5.06092768698048e-06,
      "loss": 0.3617,
      "step": 33092
    },
    {
      "epoch": 1.5186544903859391,
      "grad_norm": 0.49841877818107605,
      "learning_rate": 5.060682517204552e-06,
      "loss": 0.4057,
      "step": 33093
    },
    {
      "epoch": 1.5187003808911936,
      "grad_norm": 0.49266549944877625,
      "learning_rate": 5.060437347282703e-06,
      "loss": 0.3755,
      "step": 33094
    },
    {
      "epoch": 1.5187462713964481,
      "grad_norm": 0.4791688323020935,
      "learning_rate": 5.06019217721552e-06,
      "loss": 0.4298,
      "step": 33095
    },
    {
      "epoch": 1.5187921619017026,
      "grad_norm": 0.4628967344760895,
      "learning_rate": 5.059947007003595e-06,
      "loss": 0.3454,
      "step": 33096
    },
    {
      "epoch": 1.5188380524069571,
      "grad_norm": 0.4693751633167267,
      "learning_rate": 5.059701836647515e-06,
      "loss": 0.3902,
      "step": 33097
    },
    {
      "epoch": 1.5188839429122116,
      "grad_norm": 0.47565728425979614,
      "learning_rate": 5.059456666147871e-06,
      "loss": 0.372,
      "step": 33098
    },
    {
      "epoch": 1.518929833417466,
      "grad_norm": 0.47959795594215393,
      "learning_rate": 5.059211495505254e-06,
      "loss": 0.3895,
      "step": 33099
    },
    {
      "epoch": 1.5189757239227204,
      "grad_norm": 0.46603432297706604,
      "learning_rate": 5.05896632472025e-06,
      "loss": 0.3139,
      "step": 33100
    },
    {
      "epoch": 1.519021614427975,
      "grad_norm": 0.46787333488464355,
      "learning_rate": 5.0587211537934525e-06,
      "loss": 0.3288,
      "step": 33101
    },
    {
      "epoch": 1.5190675049332292,
      "grad_norm": 0.49200358986854553,
      "learning_rate": 5.058475982725447e-06,
      "loss": 0.3843,
      "step": 33102
    },
    {
      "epoch": 1.5191133954384837,
      "grad_norm": 0.466325044631958,
      "learning_rate": 5.058230811516827e-06,
      "loss": 0.3702,
      "step": 33103
    },
    {
      "epoch": 1.5191592859437382,
      "grad_norm": 0.4988346993923187,
      "learning_rate": 5.057985640168179e-06,
      "loss": 0.3754,
      "step": 33104
    },
    {
      "epoch": 1.5192051764489927,
      "grad_norm": 0.4740409851074219,
      "learning_rate": 5.057740468680096e-06,
      "loss": 0.4087,
      "step": 33105
    },
    {
      "epoch": 1.5192510669542472,
      "grad_norm": 0.4210430383682251,
      "learning_rate": 5.0574952970531645e-06,
      "loss": 0.2668,
      "step": 33106
    },
    {
      "epoch": 1.5192969574595017,
      "grad_norm": 0.46198585629463196,
      "learning_rate": 5.0572501252879735e-06,
      "loss": 0.3778,
      "step": 33107
    },
    {
      "epoch": 1.5193428479647562,
      "grad_norm": 0.49492454528808594,
      "learning_rate": 5.057004953385114e-06,
      "loss": 0.4157,
      "step": 33108
    },
    {
      "epoch": 1.5193887384700107,
      "grad_norm": 0.42376402020454407,
      "learning_rate": 5.056759781345177e-06,
      "loss": 0.3022,
      "step": 33109
    },
    {
      "epoch": 1.5194346289752652,
      "grad_norm": 0.4857349395751953,
      "learning_rate": 5.0565146091687516e-06,
      "loss": 0.4018,
      "step": 33110
    },
    {
      "epoch": 1.5194805194805194,
      "grad_norm": 0.42582279443740845,
      "learning_rate": 5.056269436856424e-06,
      "loss": 0.2865,
      "step": 33111
    },
    {
      "epoch": 1.519526409985774,
      "grad_norm": 0.48234376311302185,
      "learning_rate": 5.056024264408788e-06,
      "loss": 0.3801,
      "step": 33112
    },
    {
      "epoch": 1.5195723004910284,
      "grad_norm": 0.4814351499080658,
      "learning_rate": 5.055779091826429e-06,
      "loss": 0.3548,
      "step": 33113
    },
    {
      "epoch": 1.5196181909962827,
      "grad_norm": 0.4461040496826172,
      "learning_rate": 5.05553391910994e-06,
      "loss": 0.3313,
      "step": 33114
    },
    {
      "epoch": 1.5196640815015372,
      "grad_norm": 0.4580996036529541,
      "learning_rate": 5.05528874625991e-06,
      "loss": 0.3268,
      "step": 33115
    },
    {
      "epoch": 1.5197099720067917,
      "grad_norm": 0.5304268598556519,
      "learning_rate": 5.055043573276928e-06,
      "loss": 0.3692,
      "step": 33116
    },
    {
      "epoch": 1.5197558625120462,
      "grad_norm": 0.4541924297809601,
      "learning_rate": 5.054798400161584e-06,
      "loss": 0.3254,
      "step": 33117
    },
    {
      "epoch": 1.5198017530173007,
      "grad_norm": 0.4889978766441345,
      "learning_rate": 5.054553226914466e-06,
      "loss": 0.3762,
      "step": 33118
    },
    {
      "epoch": 1.5198476435225552,
      "grad_norm": 0.4738158881664276,
      "learning_rate": 5.054308053536165e-06,
      "loss": 0.3742,
      "step": 33119
    },
    {
      "epoch": 1.5198935340278097,
      "grad_norm": 0.4959319531917572,
      "learning_rate": 5.054062880027271e-06,
      "loss": 0.3216,
      "step": 33120
    },
    {
      "epoch": 1.5199394245330642,
      "grad_norm": 0.4813539683818817,
      "learning_rate": 5.053817706388372e-06,
      "loss": 0.3656,
      "step": 33121
    },
    {
      "epoch": 1.5199853150383187,
      "grad_norm": 0.4650638997554779,
      "learning_rate": 5.053572532620059e-06,
      "loss": 0.3225,
      "step": 33122
    },
    {
      "epoch": 1.5200312055435732,
      "grad_norm": 0.45099329948425293,
      "learning_rate": 5.0533273587229205e-06,
      "loss": 0.3139,
      "step": 33123
    },
    {
      "epoch": 1.5200770960488275,
      "grad_norm": 0.4505697190761566,
      "learning_rate": 5.053082184697547e-06,
      "loss": 0.3299,
      "step": 33124
    },
    {
      "epoch": 1.520122986554082,
      "grad_norm": 0.4831925332546234,
      "learning_rate": 5.052837010544526e-06,
      "loss": 0.3844,
      "step": 33125
    },
    {
      "epoch": 1.5201688770593365,
      "grad_norm": 0.472883939743042,
      "learning_rate": 5.0525918362644515e-06,
      "loss": 0.3529,
      "step": 33126
    },
    {
      "epoch": 1.5202147675645907,
      "grad_norm": 0.4358261823654175,
      "learning_rate": 5.0523466618579085e-06,
      "loss": 0.2886,
      "step": 33127
    },
    {
      "epoch": 1.5202606580698452,
      "grad_norm": 0.4866802990436554,
      "learning_rate": 5.052101487325488e-06,
      "loss": 0.3727,
      "step": 33128
    },
    {
      "epoch": 1.5203065485750997,
      "grad_norm": 0.44990724325180054,
      "learning_rate": 5.051856312667781e-06,
      "loss": 0.2941,
      "step": 33129
    },
    {
      "epoch": 1.5203524390803542,
      "grad_norm": 0.4570745825767517,
      "learning_rate": 5.051611137885375e-06,
      "loss": 0.359,
      "step": 33130
    },
    {
      "epoch": 1.5203983295856087,
      "grad_norm": 0.47600966691970825,
      "learning_rate": 5.0513659629788605e-06,
      "loss": 0.3545,
      "step": 33131
    },
    {
      "epoch": 1.5204442200908632,
      "grad_norm": 0.4671674072742462,
      "learning_rate": 5.051120787948828e-06,
      "loss": 0.3582,
      "step": 33132
    },
    {
      "epoch": 1.5204901105961177,
      "grad_norm": 0.42479634284973145,
      "learning_rate": 5.050875612795865e-06,
      "loss": 0.3294,
      "step": 33133
    },
    {
      "epoch": 1.5205360011013722,
      "grad_norm": 0.46646901965141296,
      "learning_rate": 5.050630437520562e-06,
      "loss": 0.3258,
      "step": 33134
    },
    {
      "epoch": 1.5205818916066267,
      "grad_norm": 0.4532846510410309,
      "learning_rate": 5.05038526212351e-06,
      "loss": 0.3527,
      "step": 33135
    },
    {
      "epoch": 1.5206277821118812,
      "grad_norm": 0.45455634593963623,
      "learning_rate": 5.050140086605297e-06,
      "loss": 0.3427,
      "step": 33136
    },
    {
      "epoch": 1.5206736726171355,
      "grad_norm": 0.48840072751045227,
      "learning_rate": 5.049894910966513e-06,
      "loss": 0.3993,
      "step": 33137
    },
    {
      "epoch": 1.52071956312239,
      "grad_norm": 0.45427241921424866,
      "learning_rate": 5.049649735207748e-06,
      "loss": 0.3153,
      "step": 33138
    },
    {
      "epoch": 1.5207654536276445,
      "grad_norm": 0.473233699798584,
      "learning_rate": 5.0494045593295894e-06,
      "loss": 0.3629,
      "step": 33139
    },
    {
      "epoch": 1.5208113441328988,
      "grad_norm": 0.4572136104106903,
      "learning_rate": 5.04915938333263e-06,
      "loss": 0.3384,
      "step": 33140
    },
    {
      "epoch": 1.5208572346381533,
      "grad_norm": 0.4904002249240875,
      "learning_rate": 5.048914207217457e-06,
      "loss": 0.4717,
      "step": 33141
    },
    {
      "epoch": 1.5209031251434078,
      "grad_norm": 0.45605236291885376,
      "learning_rate": 5.048669030984661e-06,
      "loss": 0.3395,
      "step": 33142
    },
    {
      "epoch": 1.5209490156486623,
      "grad_norm": 0.481944739818573,
      "learning_rate": 5.048423854634832e-06,
      "loss": 0.3915,
      "step": 33143
    },
    {
      "epoch": 1.5209949061539167,
      "grad_norm": 0.4432668685913086,
      "learning_rate": 5.048178678168557e-06,
      "loss": 0.2883,
      "step": 33144
    },
    {
      "epoch": 1.5210407966591712,
      "grad_norm": 0.4559985101222992,
      "learning_rate": 5.047933501586429e-06,
      "loss": 0.3476,
      "step": 33145
    },
    {
      "epoch": 1.5210866871644257,
      "grad_norm": 0.4293455183506012,
      "learning_rate": 5.047688324889035e-06,
      "loss": 0.2774,
      "step": 33146
    },
    {
      "epoch": 1.5211325776696802,
      "grad_norm": 0.5065532922744751,
      "learning_rate": 5.047443148076966e-06,
      "loss": 0.3841,
      "step": 33147
    },
    {
      "epoch": 1.5211784681749347,
      "grad_norm": 0.45357683300971985,
      "learning_rate": 5.047197971150813e-06,
      "loss": 0.3223,
      "step": 33148
    },
    {
      "epoch": 1.521224358680189,
      "grad_norm": 0.4754711091518402,
      "learning_rate": 5.046952794111162e-06,
      "loss": 0.3765,
      "step": 33149
    },
    {
      "epoch": 1.5212702491854435,
      "grad_norm": 0.46898791193962097,
      "learning_rate": 5.046707616958604e-06,
      "loss": 0.3527,
      "step": 33150
    },
    {
      "epoch": 1.521316139690698,
      "grad_norm": 0.45028096437454224,
      "learning_rate": 5.046462439693729e-06,
      "loss": 0.34,
      "step": 33151
    },
    {
      "epoch": 1.5213620301959525,
      "grad_norm": 0.44368526339530945,
      "learning_rate": 5.046217262317128e-06,
      "loss": 0.3106,
      "step": 33152
    },
    {
      "epoch": 1.5214079207012068,
      "grad_norm": 0.43974897265434265,
      "learning_rate": 5.045972084829388e-06,
      "loss": 0.2911,
      "step": 33153
    },
    {
      "epoch": 1.5214538112064613,
      "grad_norm": 0.4517296850681305,
      "learning_rate": 5.045726907231101e-06,
      "loss": 0.328,
      "step": 33154
    },
    {
      "epoch": 1.5214997017117158,
      "grad_norm": 0.46778419613838196,
      "learning_rate": 5.0454817295228534e-06,
      "loss": 0.3587,
      "step": 33155
    },
    {
      "epoch": 1.5215455922169703,
      "grad_norm": 0.4485432207584381,
      "learning_rate": 5.045236551705236e-06,
      "loss": 0.3204,
      "step": 33156
    },
    {
      "epoch": 1.5215914827222248,
      "grad_norm": 0.4924410581588745,
      "learning_rate": 5.044991373778842e-06,
      "loss": 0.3958,
      "step": 33157
    },
    {
      "epoch": 1.5216373732274793,
      "grad_norm": 0.4524105191230774,
      "learning_rate": 5.044746195744254e-06,
      "loss": 0.3442,
      "step": 33158
    },
    {
      "epoch": 1.5216832637327338,
      "grad_norm": 0.470284104347229,
      "learning_rate": 5.044501017602069e-06,
      "loss": 0.3771,
      "step": 33159
    },
    {
      "epoch": 1.5217291542379883,
      "grad_norm": 0.4767764210700989,
      "learning_rate": 5.044255839352872e-06,
      "loss": 0.3552,
      "step": 33160
    },
    {
      "epoch": 1.5217750447432428,
      "grad_norm": 0.4740636646747589,
      "learning_rate": 5.044010660997253e-06,
      "loss": 0.3455,
      "step": 33161
    },
    {
      "epoch": 1.521820935248497,
      "grad_norm": 0.46428853273391724,
      "learning_rate": 5.0437654825358015e-06,
      "loss": 0.3653,
      "step": 33162
    },
    {
      "epoch": 1.5218668257537515,
      "grad_norm": 0.43491166830062866,
      "learning_rate": 5.04352030396911e-06,
      "loss": 0.2986,
      "step": 33163
    },
    {
      "epoch": 1.521912716259006,
      "grad_norm": 0.4821775257587433,
      "learning_rate": 5.043275125297765e-06,
      "loss": 0.4169,
      "step": 33164
    },
    {
      "epoch": 1.5219586067642603,
      "grad_norm": 0.45576924085617065,
      "learning_rate": 5.0430299465223566e-06,
      "loss": 0.329,
      "step": 33165
    },
    {
      "epoch": 1.5220044972695148,
      "grad_norm": 0.49046722054481506,
      "learning_rate": 5.042784767643476e-06,
      "loss": 0.4104,
      "step": 33166
    },
    {
      "epoch": 1.5220503877747693,
      "grad_norm": 0.4616661071777344,
      "learning_rate": 5.04253958866171e-06,
      "loss": 0.3569,
      "step": 33167
    },
    {
      "epoch": 1.5220962782800238,
      "grad_norm": 0.46663323044776917,
      "learning_rate": 5.04229440957765e-06,
      "loss": 0.3227,
      "step": 33168
    },
    {
      "epoch": 1.5221421687852783,
      "grad_norm": 0.4276491701602936,
      "learning_rate": 5.042049230391885e-06,
      "loss": 0.2919,
      "step": 33169
    },
    {
      "epoch": 1.5221880592905328,
      "grad_norm": 0.48580917716026306,
      "learning_rate": 5.041804051105006e-06,
      "loss": 0.4082,
      "step": 33170
    },
    {
      "epoch": 1.5222339497957873,
      "grad_norm": 0.46233025193214417,
      "learning_rate": 5.041558871717603e-06,
      "loss": 0.311,
      "step": 33171
    },
    {
      "epoch": 1.5222798403010418,
      "grad_norm": 0.4323754906654358,
      "learning_rate": 5.04131369223026e-06,
      "loss": 0.3058,
      "step": 33172
    },
    {
      "epoch": 1.5223257308062963,
      "grad_norm": 0.4929503798484802,
      "learning_rate": 5.041068512643572e-06,
      "loss": 0.3542,
      "step": 33173
    },
    {
      "epoch": 1.5223716213115508,
      "grad_norm": 0.4551067650318146,
      "learning_rate": 5.040823332958128e-06,
      "loss": 0.3408,
      "step": 33174
    },
    {
      "epoch": 1.522417511816805,
      "grad_norm": 0.46326974034309387,
      "learning_rate": 5.040578153174518e-06,
      "loss": 0.3554,
      "step": 33175
    },
    {
      "epoch": 1.5224634023220596,
      "grad_norm": 0.46471384167671204,
      "learning_rate": 5.040332973293328e-06,
      "loss": 0.3366,
      "step": 33176
    },
    {
      "epoch": 1.522509292827314,
      "grad_norm": 0.5243381261825562,
      "learning_rate": 5.040087793315152e-06,
      "loss": 0.4453,
      "step": 33177
    },
    {
      "epoch": 1.5225551833325683,
      "grad_norm": 0.4876064658164978,
      "learning_rate": 5.039842613240574e-06,
      "loss": 0.3668,
      "step": 33178
    },
    {
      "epoch": 1.5226010738378228,
      "grad_norm": 0.4409368634223938,
      "learning_rate": 5.039597433070189e-06,
      "loss": 0.3536,
      "step": 33179
    },
    {
      "epoch": 1.5226469643430773,
      "grad_norm": 0.43520769476890564,
      "learning_rate": 5.039352252804586e-06,
      "loss": 0.3231,
      "step": 33180
    },
    {
      "epoch": 1.5226928548483318,
      "grad_norm": 0.4593453109264374,
      "learning_rate": 5.039107072444353e-06,
      "loss": 0.3438,
      "step": 33181
    },
    {
      "epoch": 1.5227387453535863,
      "grad_norm": 0.448498398065567,
      "learning_rate": 5.038861891990079e-06,
      "loss": 0.3034,
      "step": 33182
    },
    {
      "epoch": 1.5227846358588408,
      "grad_norm": 0.44905877113342285,
      "learning_rate": 5.038616711442354e-06,
      "loss": 0.3345,
      "step": 33183
    },
    {
      "epoch": 1.5228305263640953,
      "grad_norm": 0.4815274775028229,
      "learning_rate": 5.038371530801769e-06,
      "loss": 0.361,
      "step": 33184
    },
    {
      "epoch": 1.5228764168693498,
      "grad_norm": 0.4651052951812744,
      "learning_rate": 5.038126350068912e-06,
      "loss": 0.3377,
      "step": 33185
    },
    {
      "epoch": 1.5229223073746043,
      "grad_norm": 0.48631033301353455,
      "learning_rate": 5.037881169244374e-06,
      "loss": 0.3258,
      "step": 33186
    },
    {
      "epoch": 1.5229681978798588,
      "grad_norm": 0.48365604877471924,
      "learning_rate": 5.037635988328743e-06,
      "loss": 0.43,
      "step": 33187
    },
    {
      "epoch": 1.523014088385113,
      "grad_norm": 0.4597786068916321,
      "learning_rate": 5.03739080732261e-06,
      "loss": 0.3288,
      "step": 33188
    },
    {
      "epoch": 1.5230599788903676,
      "grad_norm": 0.447753369808197,
      "learning_rate": 5.0371456262265616e-06,
      "loss": 0.3199,
      "step": 33189
    },
    {
      "epoch": 1.523105869395622,
      "grad_norm": 0.4303513467311859,
      "learning_rate": 5.036900445041191e-06,
      "loss": 0.317,
      "step": 33190
    },
    {
      "epoch": 1.5231517599008764,
      "grad_norm": 0.4908033609390259,
      "learning_rate": 5.036655263767087e-06,
      "loss": 0.3649,
      "step": 33191
    },
    {
      "epoch": 1.5231976504061309,
      "grad_norm": 0.42502832412719727,
      "learning_rate": 5.03641008240484e-06,
      "loss": 0.3276,
      "step": 33192
    },
    {
      "epoch": 1.5232435409113854,
      "grad_norm": 0.4938699007034302,
      "learning_rate": 5.036164900955036e-06,
      "loss": 0.3807,
      "step": 33193
    },
    {
      "epoch": 1.5232894314166399,
      "grad_norm": 0.5009687542915344,
      "learning_rate": 5.035919719418268e-06,
      "loss": 0.4001,
      "step": 33194
    },
    {
      "epoch": 1.5233353219218944,
      "grad_norm": 0.46142950654029846,
      "learning_rate": 5.0356745377951236e-06,
      "loss": 0.3323,
      "step": 33195
    },
    {
      "epoch": 1.5233812124271489,
      "grad_norm": 0.4864007234573364,
      "learning_rate": 5.0354293560861935e-06,
      "loss": 0.3347,
      "step": 33196
    },
    {
      "epoch": 1.5234271029324034,
      "grad_norm": 0.4798075258731842,
      "learning_rate": 5.035184174292068e-06,
      "loss": 0.3681,
      "step": 33197
    },
    {
      "epoch": 1.5234729934376579,
      "grad_norm": 0.5086239576339722,
      "learning_rate": 5.034938992413334e-06,
      "loss": 0.3986,
      "step": 33198
    },
    {
      "epoch": 1.5235188839429123,
      "grad_norm": 0.45581093430519104,
      "learning_rate": 5.034693810450583e-06,
      "loss": 0.3113,
      "step": 33199
    },
    {
      "epoch": 1.5235647744481666,
      "grad_norm": 0.43493783473968506,
      "learning_rate": 5.034448628404405e-06,
      "loss": 0.3288,
      "step": 33200
    },
    {
      "epoch": 1.5236106649534211,
      "grad_norm": 0.475620836019516,
      "learning_rate": 5.034203446275388e-06,
      "loss": 0.349,
      "step": 33201
    },
    {
      "epoch": 1.5236565554586756,
      "grad_norm": 0.4633347690105438,
      "learning_rate": 5.033958264064124e-06,
      "loss": 0.3602,
      "step": 33202
    },
    {
      "epoch": 1.52370244596393,
      "grad_norm": 0.469974547624588,
      "learning_rate": 5.033713081771201e-06,
      "loss": 0.3947,
      "step": 33203
    },
    {
      "epoch": 1.5237483364691844,
      "grad_norm": 0.4608430862426758,
      "learning_rate": 5.0334678993972075e-06,
      "loss": 0.3348,
      "step": 33204
    },
    {
      "epoch": 1.523794226974439,
      "grad_norm": 0.5178831815719604,
      "learning_rate": 5.033222716942735e-06,
      "loss": 0.4424,
      "step": 33205
    },
    {
      "epoch": 1.5238401174796934,
      "grad_norm": 0.5134355425834656,
      "learning_rate": 5.032977534408372e-06,
      "loss": 0.41,
      "step": 33206
    },
    {
      "epoch": 1.5238860079849479,
      "grad_norm": 0.47879335284233093,
      "learning_rate": 5.0327323517947094e-06,
      "loss": 0.3531,
      "step": 33207
    },
    {
      "epoch": 1.5239318984902024,
      "grad_norm": 0.4919058084487915,
      "learning_rate": 5.032487169102334e-06,
      "loss": 0.3916,
      "step": 33208
    },
    {
      "epoch": 1.5239777889954569,
      "grad_norm": 0.44016921520233154,
      "learning_rate": 5.032241986331838e-06,
      "loss": 0.2985,
      "step": 33209
    },
    {
      "epoch": 1.5240236795007114,
      "grad_norm": 0.48360544443130493,
      "learning_rate": 5.031996803483811e-06,
      "loss": 0.3909,
      "step": 33210
    },
    {
      "epoch": 1.5240695700059659,
      "grad_norm": 0.5125842094421387,
      "learning_rate": 5.031751620558841e-06,
      "loss": 0.4284,
      "step": 33211
    },
    {
      "epoch": 1.5241154605112204,
      "grad_norm": 0.4602748453617096,
      "learning_rate": 5.0315064375575175e-06,
      "loss": 0.3717,
      "step": 33212
    },
    {
      "epoch": 1.5241613510164747,
      "grad_norm": 0.5097305774688721,
      "learning_rate": 5.031261254480433e-06,
      "loss": 0.4198,
      "step": 33213
    },
    {
      "epoch": 1.5242072415217292,
      "grad_norm": 0.4983460307121277,
      "learning_rate": 5.031016071328174e-06,
      "loss": 0.3966,
      "step": 33214
    },
    {
      "epoch": 1.5242531320269836,
      "grad_norm": 0.5090299844741821,
      "learning_rate": 5.030770888101331e-06,
      "loss": 0.444,
      "step": 33215
    },
    {
      "epoch": 1.524299022532238,
      "grad_norm": 0.44167542457580566,
      "learning_rate": 5.030525704800493e-06,
      "loss": 0.358,
      "step": 33216
    },
    {
      "epoch": 1.5243449130374924,
      "grad_norm": 0.45539042353630066,
      "learning_rate": 5.0302805214262514e-06,
      "loss": 0.3045,
      "step": 33217
    },
    {
      "epoch": 1.524390803542747,
      "grad_norm": 0.4757000505924225,
      "learning_rate": 5.030035337979195e-06,
      "loss": 0.3231,
      "step": 33218
    },
    {
      "epoch": 1.5244366940480014,
      "grad_norm": 0.46485990285873413,
      "learning_rate": 5.029790154459914e-06,
      "loss": 0.3754,
      "step": 33219
    },
    {
      "epoch": 1.524482584553256,
      "grad_norm": 0.4744590222835541,
      "learning_rate": 5.029544970868995e-06,
      "loss": 0.3242,
      "step": 33220
    },
    {
      "epoch": 1.5245284750585104,
      "grad_norm": 0.45173555612564087,
      "learning_rate": 5.0292997872070306e-06,
      "loss": 0.3301,
      "step": 33221
    },
    {
      "epoch": 1.524574365563765,
      "grad_norm": 0.44999220967292786,
      "learning_rate": 5.029054603474609e-06,
      "loss": 0.315,
      "step": 33222
    },
    {
      "epoch": 1.5246202560690194,
      "grad_norm": 0.47932276129722595,
      "learning_rate": 5.02880941967232e-06,
      "loss": 0.4138,
      "step": 33223
    },
    {
      "epoch": 1.524666146574274,
      "grad_norm": 0.4254818260669708,
      "learning_rate": 5.028564235800755e-06,
      "loss": 0.2855,
      "step": 33224
    },
    {
      "epoch": 1.5247120370795284,
      "grad_norm": 0.477897584438324,
      "learning_rate": 5.0283190518605e-06,
      "loss": 0.3753,
      "step": 33225
    },
    {
      "epoch": 1.5247579275847827,
      "grad_norm": 0.4940219521522522,
      "learning_rate": 5.028073867852148e-06,
      "loss": 0.4059,
      "step": 33226
    },
    {
      "epoch": 1.5248038180900372,
      "grad_norm": 0.4944014847278595,
      "learning_rate": 5.027828683776286e-06,
      "loss": 0.3923,
      "step": 33227
    },
    {
      "epoch": 1.5248497085952917,
      "grad_norm": 0.5033700466156006,
      "learning_rate": 5.027583499633506e-06,
      "loss": 0.4049,
      "step": 33228
    },
    {
      "epoch": 1.524895599100546,
      "grad_norm": 0.5869184136390686,
      "learning_rate": 5.027338315424397e-06,
      "loss": 0.3402,
      "step": 33229
    },
    {
      "epoch": 1.5249414896058004,
      "grad_norm": 0.514778196811676,
      "learning_rate": 5.0270931311495465e-06,
      "loss": 0.3018,
      "step": 33230
    },
    {
      "epoch": 1.524987380111055,
      "grad_norm": 0.48651888966560364,
      "learning_rate": 5.026847946809547e-06,
      "loss": 0.3844,
      "step": 33231
    },
    {
      "epoch": 1.5250332706163094,
      "grad_norm": 0.4721430838108063,
      "learning_rate": 5.026602762404984e-06,
      "loss": 0.3633,
      "step": 33232
    },
    {
      "epoch": 1.525079161121564,
      "grad_norm": 0.4642482399940491,
      "learning_rate": 5.026357577936451e-06,
      "loss": 0.3281,
      "step": 33233
    },
    {
      "epoch": 1.5251250516268184,
      "grad_norm": 0.40732863545417786,
      "learning_rate": 5.0261123934045374e-06,
      "loss": 0.2431,
      "step": 33234
    },
    {
      "epoch": 1.525170942132073,
      "grad_norm": 0.452980101108551,
      "learning_rate": 5.025867208809831e-06,
      "loss": 0.3078,
      "step": 33235
    },
    {
      "epoch": 1.5252168326373274,
      "grad_norm": 0.4706696569919586,
      "learning_rate": 5.025622024152923e-06,
      "loss": 0.3403,
      "step": 33236
    },
    {
      "epoch": 1.525262723142582,
      "grad_norm": 0.623918890953064,
      "learning_rate": 5.025376839434401e-06,
      "loss": 0.3699,
      "step": 33237
    },
    {
      "epoch": 1.5253086136478362,
      "grad_norm": 0.5010150671005249,
      "learning_rate": 5.025131654654856e-06,
      "loss": 0.4035,
      "step": 33238
    },
    {
      "epoch": 1.5253545041530907,
      "grad_norm": 0.4390171468257904,
      "learning_rate": 5.024886469814878e-06,
      "loss": 0.3126,
      "step": 33239
    },
    {
      "epoch": 1.5254003946583452,
      "grad_norm": 0.46011510491371155,
      "learning_rate": 5.024641284915055e-06,
      "loss": 0.358,
      "step": 33240
    },
    {
      "epoch": 1.5254462851635997,
      "grad_norm": 0.5175339579582214,
      "learning_rate": 5.024396099955977e-06,
      "loss": 0.3876,
      "step": 33241
    },
    {
      "epoch": 1.525492175668854,
      "grad_norm": 0.45749351382255554,
      "learning_rate": 5.024150914938236e-06,
      "loss": 0.315,
      "step": 33242
    },
    {
      "epoch": 1.5255380661741085,
      "grad_norm": 0.47081509232521057,
      "learning_rate": 5.023905729862419e-06,
      "loss": 0.372,
      "step": 33243
    },
    {
      "epoch": 1.525583956679363,
      "grad_norm": 0.4796488285064697,
      "learning_rate": 5.023660544729114e-06,
      "loss": 0.368,
      "step": 33244
    },
    {
      "epoch": 1.5256298471846175,
      "grad_norm": 0.4658336639404297,
      "learning_rate": 5.023415359538915e-06,
      "loss": 0.3685,
      "step": 33245
    },
    {
      "epoch": 1.525675737689872,
      "grad_norm": 0.467491090297699,
      "learning_rate": 5.02317017429241e-06,
      "loss": 0.3484,
      "step": 33246
    },
    {
      "epoch": 1.5257216281951265,
      "grad_norm": 0.5008457899093628,
      "learning_rate": 5.022924988990187e-06,
      "loss": 0.3518,
      "step": 33247
    },
    {
      "epoch": 1.525767518700381,
      "grad_norm": 0.4612867534160614,
      "learning_rate": 5.022679803632836e-06,
      "loss": 0.356,
      "step": 33248
    },
    {
      "epoch": 1.5258134092056355,
      "grad_norm": 0.47690388560295105,
      "learning_rate": 5.022434618220948e-06,
      "loss": 0.3715,
      "step": 33249
    },
    {
      "epoch": 1.52585929971089,
      "grad_norm": 0.4424048364162445,
      "learning_rate": 5.022189432755111e-06,
      "loss": 0.2964,
      "step": 33250
    },
    {
      "epoch": 1.5259051902161442,
      "grad_norm": 0.4892492890357971,
      "learning_rate": 5.021944247235918e-06,
      "loss": 0.3392,
      "step": 33251
    },
    {
      "epoch": 1.5259510807213987,
      "grad_norm": 0.49987250566482544,
      "learning_rate": 5.021699061663954e-06,
      "loss": 0.4293,
      "step": 33252
    },
    {
      "epoch": 1.5259969712266532,
      "grad_norm": 0.46126264333724976,
      "learning_rate": 5.0214538760398115e-06,
      "loss": 0.3126,
      "step": 33253
    },
    {
      "epoch": 1.5260428617319075,
      "grad_norm": 0.4869365394115448,
      "learning_rate": 5.021208690364078e-06,
      "loss": 0.4293,
      "step": 33254
    },
    {
      "epoch": 1.526088752237162,
      "grad_norm": 0.4923984408378601,
      "learning_rate": 5.020963504637344e-06,
      "loss": 0.3664,
      "step": 33255
    },
    {
      "epoch": 1.5261346427424165,
      "grad_norm": 0.4820711016654968,
      "learning_rate": 5.020718318860201e-06,
      "loss": 0.3787,
      "step": 33256
    },
    {
      "epoch": 1.526180533247671,
      "grad_norm": 0.49870505928993225,
      "learning_rate": 5.020473133033235e-06,
      "loss": 0.3915,
      "step": 33257
    },
    {
      "epoch": 1.5262264237529255,
      "grad_norm": 0.4556707739830017,
      "learning_rate": 5.0202279471570396e-06,
      "loss": 0.3668,
      "step": 33258
    },
    {
      "epoch": 1.52627231425818,
      "grad_norm": 0.47212743759155273,
      "learning_rate": 5.019982761232201e-06,
      "loss": 0.399,
      "step": 33259
    },
    {
      "epoch": 1.5263182047634345,
      "grad_norm": 0.41015321016311646,
      "learning_rate": 5.0197375752593115e-06,
      "loss": 0.2524,
      "step": 33260
    },
    {
      "epoch": 1.526364095268689,
      "grad_norm": 0.5067545175552368,
      "learning_rate": 5.019492389238958e-06,
      "loss": 0.3974,
      "step": 33261
    },
    {
      "epoch": 1.5264099857739435,
      "grad_norm": 0.435400128364563,
      "learning_rate": 5.019247203171733e-06,
      "loss": 0.2743,
      "step": 33262
    },
    {
      "epoch": 1.526455876279198,
      "grad_norm": 0.44996505975723267,
      "learning_rate": 5.019002017058223e-06,
      "loss": 0.3255,
      "step": 33263
    },
    {
      "epoch": 1.5265017667844523,
      "grad_norm": 0.4678812325000763,
      "learning_rate": 5.01875683089902e-06,
      "loss": 0.3779,
      "step": 33264
    },
    {
      "epoch": 1.5265476572897068,
      "grad_norm": 0.44305774569511414,
      "learning_rate": 5.0185116446947125e-06,
      "loss": 0.3178,
      "step": 33265
    },
    {
      "epoch": 1.5265935477949613,
      "grad_norm": 0.4542619585990906,
      "learning_rate": 5.01826645844589e-06,
      "loss": 0.3207,
      "step": 33266
    },
    {
      "epoch": 1.5266394383002155,
      "grad_norm": 0.44277340173721313,
      "learning_rate": 5.018021272153143e-06,
      "loss": 0.2882,
      "step": 33267
    },
    {
      "epoch": 1.52668532880547,
      "grad_norm": 0.4699394106864929,
      "learning_rate": 5.017776085817061e-06,
      "loss": 0.3545,
      "step": 33268
    },
    {
      "epoch": 1.5267312193107245,
      "grad_norm": 0.39964911341667175,
      "learning_rate": 5.017530899438231e-06,
      "loss": 0.2771,
      "step": 33269
    },
    {
      "epoch": 1.526777109815979,
      "grad_norm": 0.4964044392108917,
      "learning_rate": 5.017285713017247e-06,
      "loss": 0.3539,
      "step": 33270
    },
    {
      "epoch": 1.5268230003212335,
      "grad_norm": 0.4560113847255707,
      "learning_rate": 5.017040526554695e-06,
      "loss": 0.3872,
      "step": 33271
    },
    {
      "epoch": 1.526868890826488,
      "grad_norm": 0.5335861444473267,
      "learning_rate": 5.016795340051166e-06,
      "loss": 0.4502,
      "step": 33272
    },
    {
      "epoch": 1.5269147813317425,
      "grad_norm": 0.4584243595600128,
      "learning_rate": 5.016550153507251e-06,
      "loss": 0.3302,
      "step": 33273
    },
    {
      "epoch": 1.526960671836997,
      "grad_norm": 0.49385321140289307,
      "learning_rate": 5.016304966923536e-06,
      "loss": 0.4404,
      "step": 33274
    },
    {
      "epoch": 1.5270065623422515,
      "grad_norm": 0.516526460647583,
      "learning_rate": 5.016059780300614e-06,
      "loss": 0.3891,
      "step": 33275
    },
    {
      "epoch": 1.527052452847506,
      "grad_norm": 0.4815191328525543,
      "learning_rate": 5.015814593639073e-06,
      "loss": 0.3725,
      "step": 33276
    },
    {
      "epoch": 1.5270983433527603,
      "grad_norm": 0.46832793951034546,
      "learning_rate": 5.015569406939502e-06,
      "loss": 0.3308,
      "step": 33277
    },
    {
      "epoch": 1.5271442338580148,
      "grad_norm": 0.5530183911323547,
      "learning_rate": 5.015324220202493e-06,
      "loss": 0.3361,
      "step": 33278
    },
    {
      "epoch": 1.5271901243632693,
      "grad_norm": 0.4593767821788788,
      "learning_rate": 5.0150790334286335e-06,
      "loss": 0.3135,
      "step": 33279
    },
    {
      "epoch": 1.5272360148685236,
      "grad_norm": 0.413675457239151,
      "learning_rate": 5.0148338466185145e-06,
      "loss": 0.2705,
      "step": 33280
    },
    {
      "epoch": 1.527281905373778,
      "grad_norm": 0.44036993384361267,
      "learning_rate": 5.014588659772723e-06,
      "loss": 0.3465,
      "step": 33281
    },
    {
      "epoch": 1.5273277958790326,
      "grad_norm": 0.47713518142700195,
      "learning_rate": 5.014343472891852e-06,
      "loss": 0.3678,
      "step": 33282
    },
    {
      "epoch": 1.527373686384287,
      "grad_norm": 0.44120803475379944,
      "learning_rate": 5.014098285976487e-06,
      "loss": 0.3206,
      "step": 33283
    },
    {
      "epoch": 1.5274195768895416,
      "grad_norm": 0.4527011215686798,
      "learning_rate": 5.013853099027223e-06,
      "loss": 0.3285,
      "step": 33284
    },
    {
      "epoch": 1.527465467394796,
      "grad_norm": 0.536106526851654,
      "learning_rate": 5.013607912044646e-06,
      "loss": 0.4905,
      "step": 33285
    },
    {
      "epoch": 1.5275113579000505,
      "grad_norm": 0.44855910539627075,
      "learning_rate": 5.013362725029345e-06,
      "loss": 0.34,
      "step": 33286
    },
    {
      "epoch": 1.527557248405305,
      "grad_norm": 0.47430697083473206,
      "learning_rate": 5.013117537981912e-06,
      "loss": 0.4075,
      "step": 33287
    },
    {
      "epoch": 1.5276031389105595,
      "grad_norm": 0.4752584397792816,
      "learning_rate": 5.012872350902934e-06,
      "loss": 0.3565,
      "step": 33288
    },
    {
      "epoch": 1.5276490294158138,
      "grad_norm": 0.47045838832855225,
      "learning_rate": 5.0126271637930046e-06,
      "loss": 0.342,
      "step": 33289
    },
    {
      "epoch": 1.5276949199210683,
      "grad_norm": 0.5077087879180908,
      "learning_rate": 5.0123819766527095e-06,
      "loss": 0.4233,
      "step": 33290
    },
    {
      "epoch": 1.5277408104263228,
      "grad_norm": 0.4747716188430786,
      "learning_rate": 5.012136789482639e-06,
      "loss": 0.2936,
      "step": 33291
    },
    {
      "epoch": 1.527786700931577,
      "grad_norm": 0.5030711889266968,
      "learning_rate": 5.011891602283383e-06,
      "loss": 0.3794,
      "step": 33292
    },
    {
      "epoch": 1.5278325914368316,
      "grad_norm": 0.45925700664520264,
      "learning_rate": 5.011646415055533e-06,
      "loss": 0.33,
      "step": 33293
    },
    {
      "epoch": 1.527878481942086,
      "grad_norm": 0.5080087780952454,
      "learning_rate": 5.011401227799676e-06,
      "loss": 0.3775,
      "step": 33294
    },
    {
      "epoch": 1.5279243724473406,
      "grad_norm": 0.49982157349586487,
      "learning_rate": 5.011156040516404e-06,
      "loss": 0.3761,
      "step": 33295
    },
    {
      "epoch": 1.527970262952595,
      "grad_norm": 0.46139052510261536,
      "learning_rate": 5.010910853206304e-06,
      "loss": 0.3649,
      "step": 33296
    },
    {
      "epoch": 1.5280161534578496,
      "grad_norm": 0.45476680994033813,
      "learning_rate": 5.010665665869967e-06,
      "loss": 0.318,
      "step": 33297
    },
    {
      "epoch": 1.528062043963104,
      "grad_norm": 0.43298497796058655,
      "learning_rate": 5.010420478507981e-06,
      "loss": 0.2903,
      "step": 33298
    },
    {
      "epoch": 1.5281079344683586,
      "grad_norm": 0.4857610762119293,
      "learning_rate": 5.01017529112094e-06,
      "loss": 0.3567,
      "step": 33299
    },
    {
      "epoch": 1.528153824973613,
      "grad_norm": 0.5234953165054321,
      "learning_rate": 5.00993010370943e-06,
      "loss": 0.458,
      "step": 33300
    },
    {
      "epoch": 1.5281997154788676,
      "grad_norm": 0.4120674133300781,
      "learning_rate": 5.00968491627404e-06,
      "loss": 0.2565,
      "step": 33301
    },
    {
      "epoch": 1.5282456059841218,
      "grad_norm": 0.45423442125320435,
      "learning_rate": 5.0094397288153615e-06,
      "loss": 0.3514,
      "step": 33302
    },
    {
      "epoch": 1.5282914964893763,
      "grad_norm": 0.46962639689445496,
      "learning_rate": 5.009194541333982e-06,
      "loss": 0.3324,
      "step": 33303
    },
    {
      "epoch": 1.5283373869946308,
      "grad_norm": 0.48889923095703125,
      "learning_rate": 5.008949353830495e-06,
      "loss": 0.3592,
      "step": 33304
    },
    {
      "epoch": 1.5283832774998851,
      "grad_norm": 0.437005877494812,
      "learning_rate": 5.008704166305486e-06,
      "loss": 0.3053,
      "step": 33305
    },
    {
      "epoch": 1.5284291680051396,
      "grad_norm": 0.5269743204116821,
      "learning_rate": 5.008458978759548e-06,
      "loss": 0.4889,
      "step": 33306
    },
    {
      "epoch": 1.5284750585103941,
      "grad_norm": 0.4587824046611786,
      "learning_rate": 5.008213791193267e-06,
      "loss": 0.3495,
      "step": 33307
    },
    {
      "epoch": 1.5285209490156486,
      "grad_norm": 0.49283888936042786,
      "learning_rate": 5.0079686036072336e-06,
      "loss": 0.3453,
      "step": 33308
    },
    {
      "epoch": 1.528566839520903,
      "grad_norm": 0.4359007477760315,
      "learning_rate": 5.007723416002039e-06,
      "loss": 0.2895,
      "step": 33309
    },
    {
      "epoch": 1.5286127300261576,
      "grad_norm": 0.467785120010376,
      "learning_rate": 5.007478228378273e-06,
      "loss": 0.353,
      "step": 33310
    },
    {
      "epoch": 1.528658620531412,
      "grad_norm": 0.48227131366729736,
      "learning_rate": 5.0072330407365245e-06,
      "loss": 0.3926,
      "step": 33311
    },
    {
      "epoch": 1.5287045110366666,
      "grad_norm": 0.47962939739227295,
      "learning_rate": 5.006987853077382e-06,
      "loss": 0.3826,
      "step": 33312
    },
    {
      "epoch": 1.528750401541921,
      "grad_norm": 0.47314366698265076,
      "learning_rate": 5.006742665401436e-06,
      "loss": 0.3542,
      "step": 33313
    },
    {
      "epoch": 1.5287962920471756,
      "grad_norm": 0.5658924579620361,
      "learning_rate": 5.006497477709275e-06,
      "loss": 0.3853,
      "step": 33314
    },
    {
      "epoch": 1.5288421825524299,
      "grad_norm": 0.46082544326782227,
      "learning_rate": 5.006252290001491e-06,
      "loss": 0.3292,
      "step": 33315
    },
    {
      "epoch": 1.5288880730576844,
      "grad_norm": 0.4744125306606293,
      "learning_rate": 5.006007102278672e-06,
      "loss": 0.3721,
      "step": 33316
    },
    {
      "epoch": 1.5289339635629389,
      "grad_norm": 0.4767659604549408,
      "learning_rate": 5.005761914541408e-06,
      "loss": 0.3277,
      "step": 33317
    },
    {
      "epoch": 1.5289798540681931,
      "grad_norm": 0.5028297305107117,
      "learning_rate": 5.005516726790288e-06,
      "loss": 0.3879,
      "step": 33318
    },
    {
      "epoch": 1.5290257445734476,
      "grad_norm": 0.4611974060535431,
      "learning_rate": 5.005271539025902e-06,
      "loss": 0.3367,
      "step": 33319
    },
    {
      "epoch": 1.5290716350787021,
      "grad_norm": 0.49858367443084717,
      "learning_rate": 5.0050263512488396e-06,
      "loss": 0.4284,
      "step": 33320
    },
    {
      "epoch": 1.5291175255839566,
      "grad_norm": 0.4822252094745636,
      "learning_rate": 5.004781163459691e-06,
      "loss": 0.3776,
      "step": 33321
    },
    {
      "epoch": 1.5291634160892111,
      "grad_norm": 0.49153226613998413,
      "learning_rate": 5.004535975659045e-06,
      "loss": 0.3953,
      "step": 33322
    },
    {
      "epoch": 1.5292093065944656,
      "grad_norm": 0.4621138870716095,
      "learning_rate": 5.00429078784749e-06,
      "loss": 0.3597,
      "step": 33323
    },
    {
      "epoch": 1.5292551970997201,
      "grad_norm": 0.48682576417922974,
      "learning_rate": 5.004045600025619e-06,
      "loss": 0.394,
      "step": 33324
    },
    {
      "epoch": 1.5293010876049746,
      "grad_norm": 0.4705032408237457,
      "learning_rate": 5.0038004121940185e-06,
      "loss": 0.3457,
      "step": 33325
    },
    {
      "epoch": 1.5293469781102291,
      "grad_norm": 0.47932180762290955,
      "learning_rate": 5.003555224353278e-06,
      "loss": 0.3776,
      "step": 33326
    },
    {
      "epoch": 1.5293928686154834,
      "grad_norm": 0.47892990708351135,
      "learning_rate": 5.003310036503991e-06,
      "loss": 0.3615,
      "step": 33327
    },
    {
      "epoch": 1.529438759120738,
      "grad_norm": 0.4796202778816223,
      "learning_rate": 5.0030648486467435e-06,
      "loss": 0.3597,
      "step": 33328
    },
    {
      "epoch": 1.5294846496259924,
      "grad_norm": 0.48022696375846863,
      "learning_rate": 5.002819660782125e-06,
      "loss": 0.3787,
      "step": 33329
    },
    {
      "epoch": 1.529530540131247,
      "grad_norm": 0.4858448803424835,
      "learning_rate": 5.002574472910727e-06,
      "loss": 0.3676,
      "step": 33330
    },
    {
      "epoch": 1.5295764306365012,
      "grad_norm": 0.5513169765472412,
      "learning_rate": 5.002329285033137e-06,
      "loss": 0.4413,
      "step": 33331
    },
    {
      "epoch": 1.5296223211417557,
      "grad_norm": 0.44000500440597534,
      "learning_rate": 5.002084097149949e-06,
      "loss": 0.2916,
      "step": 33332
    },
    {
      "epoch": 1.5296682116470102,
      "grad_norm": 0.49267736077308655,
      "learning_rate": 5.001838909261747e-06,
      "loss": 0.408,
      "step": 33333
    },
    {
      "epoch": 1.5297141021522647,
      "grad_norm": 0.4637821614742279,
      "learning_rate": 5.001593721369123e-06,
      "loss": 0.3576,
      "step": 33334
    },
    {
      "epoch": 1.5297599926575192,
      "grad_norm": 0.43816161155700684,
      "learning_rate": 5.001348533472666e-06,
      "loss": 0.3423,
      "step": 33335
    },
    {
      "epoch": 1.5298058831627737,
      "grad_norm": 0.46674075722694397,
      "learning_rate": 5.001103345572967e-06,
      "loss": 0.3229,
      "step": 33336
    },
    {
      "epoch": 1.5298517736680282,
      "grad_norm": 0.47178855538368225,
      "learning_rate": 5.000858157670614e-06,
      "loss": 0.3874,
      "step": 33337
    },
    {
      "epoch": 1.5298976641732827,
      "grad_norm": 0.47776514291763306,
      "learning_rate": 5.000612969766199e-06,
      "loss": 0.3899,
      "step": 33338
    },
    {
      "epoch": 1.5299435546785372,
      "grad_norm": 0.4624825119972229,
      "learning_rate": 5.0003677818603105e-06,
      "loss": 0.3463,
      "step": 33339
    },
    {
      "epoch": 1.5299894451837914,
      "grad_norm": 0.41650810837745667,
      "learning_rate": 5.0001225939535345e-06,
      "loss": 0.2644,
      "step": 33340
    },
    {
      "epoch": 1.530035335689046,
      "grad_norm": 0.4418177306652069,
      "learning_rate": 4.999877406046466e-06,
      "loss": 0.3218,
      "step": 33341
    },
    {
      "epoch": 1.5300812261943004,
      "grad_norm": 0.5067244172096252,
      "learning_rate": 4.999632218139693e-06,
      "loss": 0.4056,
      "step": 33342
    },
    {
      "epoch": 1.5301271166995547,
      "grad_norm": 0.4835446774959564,
      "learning_rate": 4.999387030233803e-06,
      "loss": 0.3032,
      "step": 33343
    },
    {
      "epoch": 1.5301730072048092,
      "grad_norm": 0.4892168641090393,
      "learning_rate": 4.999141842329387e-06,
      "loss": 0.4081,
      "step": 33344
    },
    {
      "epoch": 1.5302188977100637,
      "grad_norm": 0.4792460501194,
      "learning_rate": 4.9988966544270355e-06,
      "loss": 0.3712,
      "step": 33345
    },
    {
      "epoch": 1.5302647882153182,
      "grad_norm": 0.5082064270973206,
      "learning_rate": 4.998651466527335e-06,
      "loss": 0.4559,
      "step": 33346
    },
    {
      "epoch": 1.5303106787205727,
      "grad_norm": 0.47889474034309387,
      "learning_rate": 4.998406278630879e-06,
      "loss": 0.3725,
      "step": 33347
    },
    {
      "epoch": 1.5303565692258272,
      "grad_norm": 0.48877692222595215,
      "learning_rate": 4.998161090738255e-06,
      "loss": 0.4291,
      "step": 33348
    },
    {
      "epoch": 1.5304024597310817,
      "grad_norm": 0.49434706568717957,
      "learning_rate": 4.997915902850054e-06,
      "loss": 0.34,
      "step": 33349
    },
    {
      "epoch": 1.5304483502363362,
      "grad_norm": 0.5326569080352783,
      "learning_rate": 4.9976707149668634e-06,
      "loss": 0.4728,
      "step": 33350
    },
    {
      "epoch": 1.5304942407415907,
      "grad_norm": 0.46771886944770813,
      "learning_rate": 4.997425527089275e-06,
      "loss": 0.3932,
      "step": 33351
    },
    {
      "epoch": 1.5305401312468452,
      "grad_norm": 0.47147440910339355,
      "learning_rate": 4.997180339217876e-06,
      "loss": 0.3661,
      "step": 33352
    },
    {
      "epoch": 1.5305860217520995,
      "grad_norm": 0.44111692905426025,
      "learning_rate": 4.996935151353257e-06,
      "loss": 0.2915,
      "step": 33353
    },
    {
      "epoch": 1.530631912257354,
      "grad_norm": 0.45354127883911133,
      "learning_rate": 4.996689963496011e-06,
      "loss": 0.2927,
      "step": 33354
    },
    {
      "epoch": 1.5306778027626085,
      "grad_norm": 0.4173949360847473,
      "learning_rate": 4.996444775646722e-06,
      "loss": 0.3002,
      "step": 33355
    },
    {
      "epoch": 1.5307236932678627,
      "grad_norm": 0.47706976532936096,
      "learning_rate": 4.996199587805983e-06,
      "loss": 0.4036,
      "step": 33356
    },
    {
      "epoch": 1.5307695837731172,
      "grad_norm": 0.4859236776828766,
      "learning_rate": 4.995954399974384e-06,
      "loss": 0.4167,
      "step": 33357
    },
    {
      "epoch": 1.5308154742783717,
      "grad_norm": 0.4598395824432373,
      "learning_rate": 4.995709212152509e-06,
      "loss": 0.3502,
      "step": 33358
    },
    {
      "epoch": 1.5308613647836262,
      "grad_norm": 0.5090404748916626,
      "learning_rate": 4.995464024340956e-06,
      "loss": 0.4159,
      "step": 33359
    },
    {
      "epoch": 1.5309072552888807,
      "grad_norm": 0.43861424922943115,
      "learning_rate": 4.995218836540311e-06,
      "loss": 0.3206,
      "step": 33360
    },
    {
      "epoch": 1.5309531457941352,
      "grad_norm": 0.4799749553203583,
      "learning_rate": 4.994973648751161e-06,
      "loss": 0.4103,
      "step": 33361
    },
    {
      "epoch": 1.5309990362993897,
      "grad_norm": 0.44542351365089417,
      "learning_rate": 4.994728460974099e-06,
      "loss": 0.3232,
      "step": 33362
    },
    {
      "epoch": 1.5310449268046442,
      "grad_norm": 0.47915101051330566,
      "learning_rate": 4.9944832732097135e-06,
      "loss": 0.4032,
      "step": 33363
    },
    {
      "epoch": 1.5310908173098987,
      "grad_norm": 0.4975765347480774,
      "learning_rate": 4.994238085458592e-06,
      "loss": 0.3773,
      "step": 33364
    },
    {
      "epoch": 1.5311367078151532,
      "grad_norm": 0.4953908920288086,
      "learning_rate": 4.993992897721329e-06,
      "loss": 0.3833,
      "step": 33365
    },
    {
      "epoch": 1.5311825983204075,
      "grad_norm": 0.4649415612220764,
      "learning_rate": 4.993747709998511e-06,
      "loss": 0.3534,
      "step": 33366
    },
    {
      "epoch": 1.531228488825662,
      "grad_norm": 0.4955553710460663,
      "learning_rate": 4.993502522290725e-06,
      "loss": 0.4145,
      "step": 33367
    },
    {
      "epoch": 1.5312743793309165,
      "grad_norm": 0.47564664483070374,
      "learning_rate": 4.993257334598565e-06,
      "loss": 0.3509,
      "step": 33368
    },
    {
      "epoch": 1.5313202698361708,
      "grad_norm": 0.4572297930717468,
      "learning_rate": 4.99301214692262e-06,
      "loss": 0.3175,
      "step": 33369
    },
    {
      "epoch": 1.5313661603414253,
      "grad_norm": 0.4961814284324646,
      "learning_rate": 4.992766959263477e-06,
      "loss": 0.3761,
      "step": 33370
    },
    {
      "epoch": 1.5314120508466798,
      "grad_norm": 0.4717697203159332,
      "learning_rate": 4.992521771621728e-06,
      "loss": 0.3824,
      "step": 33371
    },
    {
      "epoch": 1.5314579413519342,
      "grad_norm": 0.45403072237968445,
      "learning_rate": 4.992276583997962e-06,
      "loss": 0.3045,
      "step": 33372
    },
    {
      "epoch": 1.5315038318571887,
      "grad_norm": 0.49734988808631897,
      "learning_rate": 4.992031396392766e-06,
      "loss": 0.429,
      "step": 33373
    },
    {
      "epoch": 1.5315497223624432,
      "grad_norm": 0.49346253275871277,
      "learning_rate": 4.991786208806735e-06,
      "loss": 0.3808,
      "step": 33374
    },
    {
      "epoch": 1.5315956128676977,
      "grad_norm": 0.44582995772361755,
      "learning_rate": 4.991541021240455e-06,
      "loss": 0.3086,
      "step": 33375
    },
    {
      "epoch": 1.5316415033729522,
      "grad_norm": 0.44004857540130615,
      "learning_rate": 4.991295833694516e-06,
      "loss": 0.3515,
      "step": 33376
    },
    {
      "epoch": 1.5316873938782067,
      "grad_norm": 0.5284396409988403,
      "learning_rate": 4.991050646169506e-06,
      "loss": 0.4389,
      "step": 33377
    },
    {
      "epoch": 1.531733284383461,
      "grad_norm": 0.41716140508651733,
      "learning_rate": 4.9908054586660184e-06,
      "loss": 0.2475,
      "step": 33378
    },
    {
      "epoch": 1.5317791748887155,
      "grad_norm": 0.4962914288043976,
      "learning_rate": 4.990560271184641e-06,
      "loss": 0.3886,
      "step": 33379
    },
    {
      "epoch": 1.53182506539397,
      "grad_norm": 0.46365785598754883,
      "learning_rate": 4.990315083725961e-06,
      "loss": 0.3376,
      "step": 33380
    },
    {
      "epoch": 1.5318709558992243,
      "grad_norm": 0.5083673000335693,
      "learning_rate": 4.990069896290572e-06,
      "loss": 0.4284,
      "step": 33381
    },
    {
      "epoch": 1.5319168464044788,
      "grad_norm": 0.44102224707603455,
      "learning_rate": 4.989824708879061e-06,
      "loss": 0.295,
      "step": 33382
    },
    {
      "epoch": 1.5319627369097333,
      "grad_norm": 0.562372088432312,
      "learning_rate": 4.989579521492019e-06,
      "loss": 0.4321,
      "step": 33383
    },
    {
      "epoch": 1.5320086274149878,
      "grad_norm": 0.4857560694217682,
      "learning_rate": 4.989334334130034e-06,
      "loss": 0.3881,
      "step": 33384
    },
    {
      "epoch": 1.5320545179202423,
      "grad_norm": 0.43600326776504517,
      "learning_rate": 4.989089146793698e-06,
      "loss": 0.2853,
      "step": 33385
    },
    {
      "epoch": 1.5321004084254968,
      "grad_norm": 0.5014874339103699,
      "learning_rate": 4.988843959483597e-06,
      "loss": 0.4352,
      "step": 33386
    },
    {
      "epoch": 1.5321462989307513,
      "grad_norm": 0.4651113450527191,
      "learning_rate": 4.988598772200325e-06,
      "loss": 0.3544,
      "step": 33387
    },
    {
      "epoch": 1.5321921894360058,
      "grad_norm": 0.43831777572631836,
      "learning_rate": 4.988353584944469e-06,
      "loss": 0.3054,
      "step": 33388
    },
    {
      "epoch": 1.5322380799412603,
      "grad_norm": 0.4669461250305176,
      "learning_rate": 4.988108397716617e-06,
      "loss": 0.3663,
      "step": 33389
    },
    {
      "epoch": 1.5322839704465148,
      "grad_norm": 0.4665217399597168,
      "learning_rate": 4.987863210517362e-06,
      "loss": 0.334,
      "step": 33390
    },
    {
      "epoch": 1.532329860951769,
      "grad_norm": 0.46735626459121704,
      "learning_rate": 4.987618023347293e-06,
      "loss": 0.3623,
      "step": 33391
    },
    {
      "epoch": 1.5323757514570235,
      "grad_norm": 0.4254129230976105,
      "learning_rate": 4.987372836206997e-06,
      "loss": 0.2881,
      "step": 33392
    },
    {
      "epoch": 1.532421641962278,
      "grad_norm": 0.49478113651275635,
      "learning_rate": 4.9871276490970665e-06,
      "loss": 0.3741,
      "step": 33393
    },
    {
      "epoch": 1.5324675324675323,
      "grad_norm": 0.5142353773117065,
      "learning_rate": 4.986882462018091e-06,
      "loss": 0.3573,
      "step": 33394
    },
    {
      "epoch": 1.5325134229727868,
      "grad_norm": 0.4721316397190094,
      "learning_rate": 4.986637274970655e-06,
      "loss": 0.383,
      "step": 33395
    },
    {
      "epoch": 1.5325593134780413,
      "grad_norm": 0.47167855501174927,
      "learning_rate": 4.986392087955356e-06,
      "loss": 0.3489,
      "step": 33396
    },
    {
      "epoch": 1.5326052039832958,
      "grad_norm": 0.42862963676452637,
      "learning_rate": 4.986146900972778e-06,
      "loss": 0.3075,
      "step": 33397
    },
    {
      "epoch": 1.5326510944885503,
      "grad_norm": 0.45063650608062744,
      "learning_rate": 4.985901714023513e-06,
      "loss": 0.3736,
      "step": 33398
    },
    {
      "epoch": 1.5326969849938048,
      "grad_norm": 0.5179423689842224,
      "learning_rate": 4.98565652710815e-06,
      "loss": 0.4584,
      "step": 33399
    },
    {
      "epoch": 1.5327428754990593,
      "grad_norm": 0.499192476272583,
      "learning_rate": 4.985411340227279e-06,
      "loss": 0.3778,
      "step": 33400
    },
    {
      "epoch": 1.5327887660043138,
      "grad_norm": 0.4588735103607178,
      "learning_rate": 4.985166153381486e-06,
      "loss": 0.3006,
      "step": 33401
    },
    {
      "epoch": 1.5328346565095683,
      "grad_norm": 0.47090548276901245,
      "learning_rate": 4.984920966571367e-06,
      "loss": 0.3517,
      "step": 33402
    },
    {
      "epoch": 1.5328805470148228,
      "grad_norm": 0.4428320527076721,
      "learning_rate": 4.984675779797508e-06,
      "loss": 0.2874,
      "step": 33403
    },
    {
      "epoch": 1.532926437520077,
      "grad_norm": 0.47406089305877686,
      "learning_rate": 4.984430593060498e-06,
      "loss": 0.3309,
      "step": 33404
    },
    {
      "epoch": 1.5329723280253316,
      "grad_norm": 0.47259098291397095,
      "learning_rate": 4.984185406360929e-06,
      "loss": 0.3811,
      "step": 33405
    },
    {
      "epoch": 1.533018218530586,
      "grad_norm": 0.4673413932323456,
      "learning_rate": 4.983940219699388e-06,
      "loss": 0.3007,
      "step": 33406
    },
    {
      "epoch": 1.5330641090358403,
      "grad_norm": 0.46608853340148926,
      "learning_rate": 4.983695033076466e-06,
      "loss": 0.3262,
      "step": 33407
    },
    {
      "epoch": 1.5331099995410948,
      "grad_norm": 0.4304957091808319,
      "learning_rate": 4.98344984649275e-06,
      "loss": 0.3045,
      "step": 33408
    },
    {
      "epoch": 1.5331558900463493,
      "grad_norm": 0.48709484934806824,
      "learning_rate": 4.983204659948836e-06,
      "loss": 0.3583,
      "step": 33409
    },
    {
      "epoch": 1.5332017805516038,
      "grad_norm": 0.4840950667858124,
      "learning_rate": 4.9829594734453075e-06,
      "loss": 0.3896,
      "step": 33410
    },
    {
      "epoch": 1.5332476710568583,
      "grad_norm": 0.5038446187973022,
      "learning_rate": 4.982714286982754e-06,
      "loss": 0.3546,
      "step": 33411
    },
    {
      "epoch": 1.5332935615621128,
      "grad_norm": 0.48782816529273987,
      "learning_rate": 4.9824691005617695e-06,
      "loss": 0.3755,
      "step": 33412
    },
    {
      "epoch": 1.5333394520673673,
      "grad_norm": 0.4720396101474762,
      "learning_rate": 4.982223914182942e-06,
      "loss": 0.3112,
      "step": 33413
    },
    {
      "epoch": 1.5333853425726218,
      "grad_norm": 0.49455755949020386,
      "learning_rate": 4.981978727846859e-06,
      "loss": 0.397,
      "step": 33414
    },
    {
      "epoch": 1.5334312330778763,
      "grad_norm": 0.44888776540756226,
      "learning_rate": 4.981733541554112e-06,
      "loss": 0.3339,
      "step": 33415
    },
    {
      "epoch": 1.5334771235831306,
      "grad_norm": 0.49241727590560913,
      "learning_rate": 4.98148835530529e-06,
      "loss": 0.3465,
      "step": 33416
    },
    {
      "epoch": 1.533523014088385,
      "grad_norm": 0.49762701988220215,
      "learning_rate": 4.981243169100981e-06,
      "loss": 0.3642,
      "step": 33417
    },
    {
      "epoch": 1.5335689045936396,
      "grad_norm": 0.5293100476264954,
      "learning_rate": 4.980997982941779e-06,
      "loss": 0.3965,
      "step": 33418
    },
    {
      "epoch": 1.533614795098894,
      "grad_norm": 0.5477155447006226,
      "learning_rate": 4.98075279682827e-06,
      "loss": 0.4733,
      "step": 33419
    },
    {
      "epoch": 1.5336606856041484,
      "grad_norm": 0.47349607944488525,
      "learning_rate": 4.9805076107610425e-06,
      "loss": 0.3271,
      "step": 33420
    },
    {
      "epoch": 1.5337065761094029,
      "grad_norm": 0.514231264591217,
      "learning_rate": 4.98026242474069e-06,
      "loss": 0.3945,
      "step": 33421
    },
    {
      "epoch": 1.5337524666146574,
      "grad_norm": 0.5172644257545471,
      "learning_rate": 4.9800172387678005e-06,
      "loss": 0.4287,
      "step": 33422
    },
    {
      "epoch": 1.5337983571199119,
      "grad_norm": 0.4736723303794861,
      "learning_rate": 4.979772052842961e-06,
      "loss": 0.3518,
      "step": 33423
    },
    {
      "epoch": 1.5338442476251664,
      "grad_norm": 0.4812438189983368,
      "learning_rate": 4.9795268669667655e-06,
      "loss": 0.3316,
      "step": 33424
    },
    {
      "epoch": 1.5338901381304209,
      "grad_norm": 0.4548267424106598,
      "learning_rate": 4.979281681139801e-06,
      "loss": 0.3185,
      "step": 33425
    },
    {
      "epoch": 1.5339360286356754,
      "grad_norm": 0.49130624532699585,
      "learning_rate": 4.979036495362656e-06,
      "loss": 0.3595,
      "step": 33426
    },
    {
      "epoch": 1.5339819191409299,
      "grad_norm": 0.45011934638023376,
      "learning_rate": 4.978791309635924e-06,
      "loss": 0.3027,
      "step": 33427
    },
    {
      "epoch": 1.5340278096461843,
      "grad_norm": 0.46906059980392456,
      "learning_rate": 4.978546123960192e-06,
      "loss": 0.3664,
      "step": 33428
    },
    {
      "epoch": 1.5340737001514386,
      "grad_norm": 0.4785364866256714,
      "learning_rate": 4.978300938336047e-06,
      "loss": 0.3607,
      "step": 33429
    },
    {
      "epoch": 1.5341195906566931,
      "grad_norm": 0.5014561414718628,
      "learning_rate": 4.978055752764083e-06,
      "loss": 0.3961,
      "step": 33430
    },
    {
      "epoch": 1.5341654811619476,
      "grad_norm": 0.4950072467327118,
      "learning_rate": 4.97781056724489e-06,
      "loss": 0.3964,
      "step": 33431
    },
    {
      "epoch": 1.534211371667202,
      "grad_norm": 0.48121893405914307,
      "learning_rate": 4.977565381779053e-06,
      "loss": 0.392,
      "step": 33432
    },
    {
      "epoch": 1.5342572621724564,
      "grad_norm": 0.46876922249794006,
      "learning_rate": 4.977320196367165e-06,
      "loss": 0.3399,
      "step": 33433
    },
    {
      "epoch": 1.534303152677711,
      "grad_norm": 0.4675857722759247,
      "learning_rate": 4.977075011009816e-06,
      "loss": 0.361,
      "step": 33434
    },
    {
      "epoch": 1.5343490431829654,
      "grad_norm": 0.4779052138328552,
      "learning_rate": 4.976829825707592e-06,
      "loss": 0.3653,
      "step": 33435
    },
    {
      "epoch": 1.5343949336882199,
      "grad_norm": 0.492640882730484,
      "learning_rate": 4.976584640461086e-06,
      "loss": 0.413,
      "step": 33436
    },
    {
      "epoch": 1.5344408241934744,
      "grad_norm": 0.5001112222671509,
      "learning_rate": 4.976339455270888e-06,
      "loss": 0.4531,
      "step": 33437
    },
    {
      "epoch": 1.5344867146987289,
      "grad_norm": 0.44833678007125854,
      "learning_rate": 4.976094270137582e-06,
      "loss": 0.2971,
      "step": 33438
    },
    {
      "epoch": 1.5345326052039834,
      "grad_norm": 0.48467087745666504,
      "learning_rate": 4.975849085061766e-06,
      "loss": 0.4089,
      "step": 33439
    },
    {
      "epoch": 1.5345784957092379,
      "grad_norm": 0.4982539117336273,
      "learning_rate": 4.975603900044024e-06,
      "loss": 0.4146,
      "step": 33440
    },
    {
      "epoch": 1.5346243862144924,
      "grad_norm": 0.49855053424835205,
      "learning_rate": 4.975358715084947e-06,
      "loss": 0.4076,
      "step": 33441
    },
    {
      "epoch": 1.5346702767197467,
      "grad_norm": 0.5264787077903748,
      "learning_rate": 4.975113530185124e-06,
      "loss": 0.4335,
      "step": 33442
    },
    {
      "epoch": 1.5347161672250011,
      "grad_norm": 0.48854929208755493,
      "learning_rate": 4.974868345345146e-06,
      "loss": 0.366,
      "step": 33443
    },
    {
      "epoch": 1.5347620577302556,
      "grad_norm": 0.47693586349487305,
      "learning_rate": 4.974623160565601e-06,
      "loss": 0.3906,
      "step": 33444
    },
    {
      "epoch": 1.53480794823551,
      "grad_norm": 0.4990052282810211,
      "learning_rate": 4.9743779758470774e-06,
      "loss": 0.3768,
      "step": 33445
    },
    {
      "epoch": 1.5348538387407644,
      "grad_norm": 0.4892053008079529,
      "learning_rate": 4.9741327911901695e-06,
      "loss": 0.3533,
      "step": 33446
    },
    {
      "epoch": 1.534899729246019,
      "grad_norm": 0.4960637390613556,
      "learning_rate": 4.973887606595464e-06,
      "loss": 0.3822,
      "step": 33447
    },
    {
      "epoch": 1.5349456197512734,
      "grad_norm": 0.46830883622169495,
      "learning_rate": 4.9736424220635496e-06,
      "loss": 0.3663,
      "step": 33448
    },
    {
      "epoch": 1.534991510256528,
      "grad_norm": 0.4579513669013977,
      "learning_rate": 4.9733972375950175e-06,
      "loss": 0.3515,
      "step": 33449
    },
    {
      "epoch": 1.5350374007617824,
      "grad_norm": 0.5010212063789368,
      "learning_rate": 4.973152053190457e-06,
      "loss": 0.3808,
      "step": 33450
    },
    {
      "epoch": 1.535083291267037,
      "grad_norm": 0.438052773475647,
      "learning_rate": 4.972906868850454e-06,
      "loss": 0.2938,
      "step": 33451
    },
    {
      "epoch": 1.5351291817722914,
      "grad_norm": 0.4766504764556885,
      "learning_rate": 4.972661684575605e-06,
      "loss": 0.3363,
      "step": 33452
    },
    {
      "epoch": 1.535175072277546,
      "grad_norm": 0.4499192535877228,
      "learning_rate": 4.972416500366496e-06,
      "loss": 0.3017,
      "step": 33453
    },
    {
      "epoch": 1.5352209627828004,
      "grad_norm": 0.4784931242465973,
      "learning_rate": 4.972171316223714e-06,
      "loss": 0.3567,
      "step": 33454
    },
    {
      "epoch": 1.5352668532880547,
      "grad_norm": 0.5144932270050049,
      "learning_rate": 4.971926132147853e-06,
      "loss": 0.4282,
      "step": 33455
    },
    {
      "epoch": 1.5353127437933092,
      "grad_norm": 0.4516366422176361,
      "learning_rate": 4.971680948139501e-06,
      "loss": 0.3523,
      "step": 33456
    },
    {
      "epoch": 1.5353586342985637,
      "grad_norm": 0.4766368865966797,
      "learning_rate": 4.971435764199247e-06,
      "loss": 0.3099,
      "step": 33457
    },
    {
      "epoch": 1.535404524803818,
      "grad_norm": 0.4583054482936859,
      "learning_rate": 4.9711905803276815e-06,
      "loss": 0.3359,
      "step": 33458
    },
    {
      "epoch": 1.5354504153090724,
      "grad_norm": 0.4670715034008026,
      "learning_rate": 4.970945396525393e-06,
      "loss": 0.3649,
      "step": 33459
    },
    {
      "epoch": 1.535496305814327,
      "grad_norm": 0.4758431613445282,
      "learning_rate": 4.97070021279297e-06,
      "loss": 0.3384,
      "step": 33460
    },
    {
      "epoch": 1.5355421963195814,
      "grad_norm": 0.4251309037208557,
      "learning_rate": 4.970455029131006e-06,
      "loss": 0.2563,
      "step": 33461
    },
    {
      "epoch": 1.535588086824836,
      "grad_norm": 0.4490491449832916,
      "learning_rate": 4.970209845540089e-06,
      "loss": 0.2967,
      "step": 33462
    },
    {
      "epoch": 1.5356339773300904,
      "grad_norm": 0.5228750705718994,
      "learning_rate": 4.9699646620208054e-06,
      "loss": 0.404,
      "step": 33463
    },
    {
      "epoch": 1.535679867835345,
      "grad_norm": 0.45339563488960266,
      "learning_rate": 4.969719478573749e-06,
      "loss": 0.3601,
      "step": 33464
    },
    {
      "epoch": 1.5357257583405994,
      "grad_norm": 0.4756806790828705,
      "learning_rate": 4.969474295199508e-06,
      "loss": 0.3664,
      "step": 33465
    },
    {
      "epoch": 1.535771648845854,
      "grad_norm": 0.500033438205719,
      "learning_rate": 4.969229111898669e-06,
      "loss": 0.3748,
      "step": 33466
    },
    {
      "epoch": 1.5358175393511082,
      "grad_norm": 0.49815234541893005,
      "learning_rate": 4.968983928671827e-06,
      "loss": 0.3906,
      "step": 33467
    },
    {
      "epoch": 1.5358634298563627,
      "grad_norm": 0.4770985245704651,
      "learning_rate": 4.968738745519568e-06,
      "loss": 0.313,
      "step": 33468
    },
    {
      "epoch": 1.5359093203616172,
      "grad_norm": 0.4826403856277466,
      "learning_rate": 4.9684935624424825e-06,
      "loss": 0.3428,
      "step": 33469
    },
    {
      "epoch": 1.5359552108668715,
      "grad_norm": 0.4563618302345276,
      "learning_rate": 4.9682483794411606e-06,
      "loss": 0.3729,
      "step": 33470
    },
    {
      "epoch": 1.536001101372126,
      "grad_norm": 0.475603848695755,
      "learning_rate": 4.968003196516191e-06,
      "loss": 0.3406,
      "step": 33471
    },
    {
      "epoch": 1.5360469918773805,
      "grad_norm": 0.4416847229003906,
      "learning_rate": 4.967758013668164e-06,
      "loss": 0.2676,
      "step": 33472
    },
    {
      "epoch": 1.536092882382635,
      "grad_norm": 0.4959782361984253,
      "learning_rate": 4.9675128308976665e-06,
      "loss": 0.4389,
      "step": 33473
    },
    {
      "epoch": 1.5361387728878895,
      "grad_norm": 0.4590708911418915,
      "learning_rate": 4.967267648205293e-06,
      "loss": 0.3322,
      "step": 33474
    },
    {
      "epoch": 1.536184663393144,
      "grad_norm": 0.48880016803741455,
      "learning_rate": 4.967022465591631e-06,
      "loss": 0.4122,
      "step": 33475
    },
    {
      "epoch": 1.5362305538983985,
      "grad_norm": 0.4766637682914734,
      "learning_rate": 4.9667772830572665e-06,
      "loss": 0.333,
      "step": 33476
    },
    {
      "epoch": 1.536276444403653,
      "grad_norm": 0.4620327651500702,
      "learning_rate": 4.966532100602794e-06,
      "loss": 0.2889,
      "step": 33477
    },
    {
      "epoch": 1.5363223349089075,
      "grad_norm": 0.4541313052177429,
      "learning_rate": 4.966286918228802e-06,
      "loss": 0.3651,
      "step": 33478
    },
    {
      "epoch": 1.536368225414162,
      "grad_norm": 0.4923425614833832,
      "learning_rate": 4.966041735935877e-06,
      "loss": 0.3885,
      "step": 33479
    },
    {
      "epoch": 1.5364141159194162,
      "grad_norm": 0.45403826236724854,
      "learning_rate": 4.9657965537246134e-06,
      "loss": 0.3027,
      "step": 33480
    },
    {
      "epoch": 1.5364600064246707,
      "grad_norm": 0.4706946611404419,
      "learning_rate": 4.965551371595597e-06,
      "loss": 0.3707,
      "step": 33481
    },
    {
      "epoch": 1.5365058969299252,
      "grad_norm": 0.48963186144828796,
      "learning_rate": 4.965306189549418e-06,
      "loss": 0.4482,
      "step": 33482
    },
    {
      "epoch": 1.5365517874351795,
      "grad_norm": 0.4674972891807556,
      "learning_rate": 4.9650610075866675e-06,
      "loss": 0.3471,
      "step": 33483
    },
    {
      "epoch": 1.536597677940434,
      "grad_norm": 0.46600818634033203,
      "learning_rate": 4.9648158257079345e-06,
      "loss": 0.3787,
      "step": 33484
    },
    {
      "epoch": 1.5366435684456885,
      "grad_norm": 0.475747287273407,
      "learning_rate": 4.964570643913807e-06,
      "loss": 0.4019,
      "step": 33485
    },
    {
      "epoch": 1.536689458950943,
      "grad_norm": 0.4632922410964966,
      "learning_rate": 4.964325462204878e-06,
      "loss": 0.3427,
      "step": 33486
    },
    {
      "epoch": 1.5367353494561975,
      "grad_norm": 0.4941183626651764,
      "learning_rate": 4.964080280581735e-06,
      "loss": 0.3473,
      "step": 33487
    },
    {
      "epoch": 1.536781239961452,
      "grad_norm": 0.48881468176841736,
      "learning_rate": 4.963835099044965e-06,
      "loss": 0.3417,
      "step": 33488
    },
    {
      "epoch": 1.5368271304667065,
      "grad_norm": 0.47942933440208435,
      "learning_rate": 4.963589917595162e-06,
      "loss": 0.3721,
      "step": 33489
    },
    {
      "epoch": 1.536873020971961,
      "grad_norm": 0.4614333510398865,
      "learning_rate": 4.963344736232914e-06,
      "loss": 0.3326,
      "step": 33490
    },
    {
      "epoch": 1.5369189114772155,
      "grad_norm": 0.4957965910434723,
      "learning_rate": 4.963099554958809e-06,
      "loss": 0.3323,
      "step": 33491
    },
    {
      "epoch": 1.53696480198247,
      "grad_norm": 0.47933393716812134,
      "learning_rate": 4.96285437377344e-06,
      "loss": 0.3866,
      "step": 33492
    },
    {
      "epoch": 1.5370106924877243,
      "grad_norm": 0.45818474888801575,
      "learning_rate": 4.9626091926773935e-06,
      "loss": 0.3597,
      "step": 33493
    },
    {
      "epoch": 1.5370565829929788,
      "grad_norm": 0.49736732244491577,
      "learning_rate": 4.962364011671258e-06,
      "loss": 0.4185,
      "step": 33494
    },
    {
      "epoch": 1.5371024734982333,
      "grad_norm": 0.4605158567428589,
      "learning_rate": 4.962118830755627e-06,
      "loss": 0.3551,
      "step": 33495
    },
    {
      "epoch": 1.5371483640034875,
      "grad_norm": 0.46528831124305725,
      "learning_rate": 4.9618736499310894e-06,
      "loss": 0.3461,
      "step": 33496
    },
    {
      "epoch": 1.537194254508742,
      "grad_norm": 0.4810307025909424,
      "learning_rate": 4.961628469198232e-06,
      "loss": 0.3419,
      "step": 33497
    },
    {
      "epoch": 1.5372401450139965,
      "grad_norm": 0.4557763636112213,
      "learning_rate": 4.961383288557647e-06,
      "loss": 0.3544,
      "step": 33498
    },
    {
      "epoch": 1.537286035519251,
      "grad_norm": 0.4526757299900055,
      "learning_rate": 4.961138108009923e-06,
      "loss": 0.3499,
      "step": 33499
    },
    {
      "epoch": 1.5373319260245055,
      "grad_norm": 0.4954931139945984,
      "learning_rate": 4.960892927555648e-06,
      "loss": 0.4136,
      "step": 33500
    },
    {
      "epoch": 1.53737781652976,
      "grad_norm": 0.5315516591072083,
      "learning_rate": 4.960647747195415e-06,
      "loss": 0.4032,
      "step": 33501
    },
    {
      "epoch": 1.5374237070350145,
      "grad_norm": 0.48933500051498413,
      "learning_rate": 4.960402566929812e-06,
      "loss": 0.4347,
      "step": 33502
    },
    {
      "epoch": 1.537469597540269,
      "grad_norm": 0.5072292685508728,
      "learning_rate": 4.960157386759426e-06,
      "loss": 0.3955,
      "step": 33503
    },
    {
      "epoch": 1.5375154880455235,
      "grad_norm": 0.5325661897659302,
      "learning_rate": 4.95991220668485e-06,
      "loss": 0.4201,
      "step": 33504
    },
    {
      "epoch": 1.5375613785507778,
      "grad_norm": 0.44953417778015137,
      "learning_rate": 4.959667026706673e-06,
      "loss": 0.3376,
      "step": 33505
    },
    {
      "epoch": 1.5376072690560323,
      "grad_norm": 0.5028049945831299,
      "learning_rate": 4.9594218468254854e-06,
      "loss": 0.3721,
      "step": 33506
    },
    {
      "epoch": 1.5376531595612868,
      "grad_norm": 0.4753458499908447,
      "learning_rate": 4.959176667041873e-06,
      "loss": 0.3525,
      "step": 33507
    },
    {
      "epoch": 1.5376990500665413,
      "grad_norm": 0.4624435007572174,
      "learning_rate": 4.958931487356429e-06,
      "loss": 0.3043,
      "step": 33508
    },
    {
      "epoch": 1.5377449405717956,
      "grad_norm": 0.5080615878105164,
      "learning_rate": 4.958686307769742e-06,
      "loss": 0.3729,
      "step": 33509
    },
    {
      "epoch": 1.53779083107705,
      "grad_norm": 0.46574658155441284,
      "learning_rate": 4.9584411282824e-06,
      "loss": 0.3344,
      "step": 33510
    },
    {
      "epoch": 1.5378367215823046,
      "grad_norm": 0.4834032654762268,
      "learning_rate": 4.958195948894995e-06,
      "loss": 0.3527,
      "step": 33511
    },
    {
      "epoch": 1.537882612087559,
      "grad_norm": 0.4529985189437866,
      "learning_rate": 4.957950769608116e-06,
      "loss": 0.323,
      "step": 33512
    },
    {
      "epoch": 1.5379285025928136,
      "grad_norm": 0.44337084889411926,
      "learning_rate": 4.957705590422351e-06,
      "loss": 0.2973,
      "step": 33513
    },
    {
      "epoch": 1.537974393098068,
      "grad_norm": 0.4465126097202301,
      "learning_rate": 4.9574604113382914e-06,
      "loss": 0.3197,
      "step": 33514
    },
    {
      "epoch": 1.5380202836033225,
      "grad_norm": 0.4845251441001892,
      "learning_rate": 4.957215232356527e-06,
      "loss": 0.3644,
      "step": 33515
    },
    {
      "epoch": 1.538066174108577,
      "grad_norm": 0.4352024495601654,
      "learning_rate": 4.956970053477643e-06,
      "loss": 0.3145,
      "step": 33516
    },
    {
      "epoch": 1.5381120646138315,
      "grad_norm": 0.42236316204071045,
      "learning_rate": 4.956724874702236e-06,
      "loss": 0.2947,
      "step": 33517
    },
    {
      "epoch": 1.5381579551190858,
      "grad_norm": 0.506375253200531,
      "learning_rate": 4.956479696030891e-06,
      "loss": 0.4013,
      "step": 33518
    },
    {
      "epoch": 1.5382038456243403,
      "grad_norm": 0.4780620038509369,
      "learning_rate": 4.9562345174641984e-06,
      "loss": 0.4045,
      "step": 33519
    },
    {
      "epoch": 1.5382497361295948,
      "grad_norm": 0.46596357226371765,
      "learning_rate": 4.955989339002749e-06,
      "loss": 0.3391,
      "step": 33520
    },
    {
      "epoch": 1.538295626634849,
      "grad_norm": 0.5346869826316833,
      "learning_rate": 4.9557441606471306e-06,
      "loss": 0.4813,
      "step": 33521
    },
    {
      "epoch": 1.5383415171401036,
      "grad_norm": 0.4888061583042145,
      "learning_rate": 4.955498982397933e-06,
      "loss": 0.3987,
      "step": 33522
    },
    {
      "epoch": 1.538387407645358,
      "grad_norm": 0.4588027000427246,
      "learning_rate": 4.955253804255747e-06,
      "loss": 0.3819,
      "step": 33523
    },
    {
      "epoch": 1.5384332981506126,
      "grad_norm": 0.45934948325157166,
      "learning_rate": 4.955008626221162e-06,
      "loss": 0.3275,
      "step": 33524
    },
    {
      "epoch": 1.538479188655867,
      "grad_norm": 0.4335435926914215,
      "learning_rate": 4.954763448294764e-06,
      "loss": 0.2672,
      "step": 33525
    },
    {
      "epoch": 1.5385250791611216,
      "grad_norm": 0.46984604001045227,
      "learning_rate": 4.954518270477148e-06,
      "loss": 0.262,
      "step": 33526
    },
    {
      "epoch": 1.538570969666376,
      "grad_norm": 0.4909984767436981,
      "learning_rate": 4.954273092768902e-06,
      "loss": 0.397,
      "step": 33527
    },
    {
      "epoch": 1.5386168601716306,
      "grad_norm": 0.47724518179893494,
      "learning_rate": 4.954027915170614e-06,
      "loss": 0.36,
      "step": 33528
    },
    {
      "epoch": 1.538662750676885,
      "grad_norm": 0.45673078298568726,
      "learning_rate": 4.953782737682874e-06,
      "loss": 0.3481,
      "step": 33529
    },
    {
      "epoch": 1.5387086411821396,
      "grad_norm": 0.42836862802505493,
      "learning_rate": 4.953537560306273e-06,
      "loss": 0.319,
      "step": 33530
    },
    {
      "epoch": 1.5387545316873938,
      "grad_norm": 0.46772244572639465,
      "learning_rate": 4.953292383041397e-06,
      "loss": 0.3153,
      "step": 33531
    },
    {
      "epoch": 1.5388004221926483,
      "grad_norm": 0.5043237209320068,
      "learning_rate": 4.95304720588884e-06,
      "loss": 0.3643,
      "step": 33532
    },
    {
      "epoch": 1.5388463126979028,
      "grad_norm": 0.48656636476516724,
      "learning_rate": 4.952802028849189e-06,
      "loss": 0.3963,
      "step": 33533
    },
    {
      "epoch": 1.5388922032031571,
      "grad_norm": 0.4500691890716553,
      "learning_rate": 4.952556851923035e-06,
      "loss": 0.2868,
      "step": 33534
    },
    {
      "epoch": 1.5389380937084116,
      "grad_norm": 0.4585524797439575,
      "learning_rate": 4.9523116751109665e-06,
      "loss": 0.3379,
      "step": 33535
    },
    {
      "epoch": 1.5389839842136661,
      "grad_norm": 0.4980064034461975,
      "learning_rate": 4.952066498413573e-06,
      "loss": 0.4013,
      "step": 33536
    },
    {
      "epoch": 1.5390298747189206,
      "grad_norm": 0.507173478603363,
      "learning_rate": 4.951821321831445e-06,
      "loss": 0.3856,
      "step": 33537
    },
    {
      "epoch": 1.539075765224175,
      "grad_norm": 0.5247698426246643,
      "learning_rate": 4.95157614536517e-06,
      "loss": 0.3778,
      "step": 33538
    },
    {
      "epoch": 1.5391216557294296,
      "grad_norm": 0.4858540892601013,
      "learning_rate": 4.95133096901534e-06,
      "loss": 0.3764,
      "step": 33539
    },
    {
      "epoch": 1.539167546234684,
      "grad_norm": 0.4381698668003082,
      "learning_rate": 4.951085792782545e-06,
      "loss": 0.2703,
      "step": 33540
    },
    {
      "epoch": 1.5392134367399386,
      "grad_norm": 0.45550626516342163,
      "learning_rate": 4.950840616667371e-06,
      "loss": 0.329,
      "step": 33541
    },
    {
      "epoch": 1.539259327245193,
      "grad_norm": 0.46883624792099,
      "learning_rate": 4.950595440670412e-06,
      "loss": 0.4082,
      "step": 33542
    },
    {
      "epoch": 1.5393052177504476,
      "grad_norm": 0.4338158667087555,
      "learning_rate": 4.9503502647922546e-06,
      "loss": 0.2846,
      "step": 33543
    },
    {
      "epoch": 1.5393511082557019,
      "grad_norm": 0.5083686113357544,
      "learning_rate": 4.950105089033489e-06,
      "loss": 0.3285,
      "step": 33544
    },
    {
      "epoch": 1.5393969987609564,
      "grad_norm": 0.4538721740245819,
      "learning_rate": 4.949859913394705e-06,
      "loss": 0.3284,
      "step": 33545
    },
    {
      "epoch": 1.5394428892662109,
      "grad_norm": 0.4593503475189209,
      "learning_rate": 4.949614737876492e-06,
      "loss": 0.3321,
      "step": 33546
    },
    {
      "epoch": 1.5394887797714651,
      "grad_norm": 0.4546606242656708,
      "learning_rate": 4.949369562479438e-06,
      "loss": 0.3298,
      "step": 33547
    },
    {
      "epoch": 1.5395346702767196,
      "grad_norm": 0.46268191933631897,
      "learning_rate": 4.949124387204136e-06,
      "loss": 0.378,
      "step": 33548
    },
    {
      "epoch": 1.5395805607819741,
      "grad_norm": 0.5112650394439697,
      "learning_rate": 4.948879212051174e-06,
      "loss": 0.3885,
      "step": 33549
    },
    {
      "epoch": 1.5396264512872286,
      "grad_norm": 0.48819857835769653,
      "learning_rate": 4.94863403702114e-06,
      "loss": 0.4141,
      "step": 33550
    },
    {
      "epoch": 1.5396723417924831,
      "grad_norm": 0.4744936525821686,
      "learning_rate": 4.948388862114627e-06,
      "loss": 0.4149,
      "step": 33551
    },
    {
      "epoch": 1.5397182322977376,
      "grad_norm": 0.4827585518360138,
      "learning_rate": 4.948143687332222e-06,
      "loss": 0.3807,
      "step": 33552
    },
    {
      "epoch": 1.5397641228029921,
      "grad_norm": 0.4800211191177368,
      "learning_rate": 4.947898512674513e-06,
      "loss": 0.3655,
      "step": 33553
    },
    {
      "epoch": 1.5398100133082466,
      "grad_norm": 0.48756587505340576,
      "learning_rate": 4.947653338142093e-06,
      "loss": 0.3834,
      "step": 33554
    },
    {
      "epoch": 1.5398559038135011,
      "grad_norm": 0.48141101002693176,
      "learning_rate": 4.947408163735551e-06,
      "loss": 0.3545,
      "step": 33555
    },
    {
      "epoch": 1.5399017943187554,
      "grad_norm": 0.471162348985672,
      "learning_rate": 4.947162989455474e-06,
      "loss": 0.3492,
      "step": 33556
    },
    {
      "epoch": 1.53994768482401,
      "grad_norm": 0.47999197244644165,
      "learning_rate": 4.946917815302455e-06,
      "loss": 0.388,
      "step": 33557
    },
    {
      "epoch": 1.5399935753292644,
      "grad_norm": 0.5018584132194519,
      "learning_rate": 4.946672641277081e-06,
      "loss": 0.3897,
      "step": 33558
    },
    {
      "epoch": 1.5400394658345187,
      "grad_norm": 0.4959656596183777,
      "learning_rate": 4.946427467379942e-06,
      "loss": 0.3913,
      "step": 33559
    },
    {
      "epoch": 1.5400853563397732,
      "grad_norm": 0.5107811093330383,
      "learning_rate": 4.946182293611629e-06,
      "loss": 0.4037,
      "step": 33560
    },
    {
      "epoch": 1.5401312468450277,
      "grad_norm": 0.4421091377735138,
      "learning_rate": 4.945937119972731e-06,
      "loss": 0.2774,
      "step": 33561
    },
    {
      "epoch": 1.5401771373502822,
      "grad_norm": 0.4308513104915619,
      "learning_rate": 4.945691946463836e-06,
      "loss": 0.306,
      "step": 33562
    },
    {
      "epoch": 1.5402230278555367,
      "grad_norm": 0.4444197118282318,
      "learning_rate": 4.945446773085535e-06,
      "loss": 0.3001,
      "step": 33563
    },
    {
      "epoch": 1.5402689183607912,
      "grad_norm": 0.47759947180747986,
      "learning_rate": 4.945201599838419e-06,
      "loss": 0.4016,
      "step": 33564
    },
    {
      "epoch": 1.5403148088660457,
      "grad_norm": 0.4484749734401703,
      "learning_rate": 4.9449564267230724e-06,
      "loss": 0.3116,
      "step": 33565
    },
    {
      "epoch": 1.5403606993713002,
      "grad_norm": 0.46381714940071106,
      "learning_rate": 4.944711253740091e-06,
      "loss": 0.3193,
      "step": 33566
    },
    {
      "epoch": 1.5404065898765547,
      "grad_norm": 0.5293952226638794,
      "learning_rate": 4.944466080890062e-06,
      "loss": 0.4635,
      "step": 33567
    },
    {
      "epoch": 1.5404524803818092,
      "grad_norm": 0.44792816042900085,
      "learning_rate": 4.9442209081735715e-06,
      "loss": 0.3316,
      "step": 33568
    },
    {
      "epoch": 1.5404983708870634,
      "grad_norm": 0.5117705464363098,
      "learning_rate": 4.943975735591214e-06,
      "loss": 0.3919,
      "step": 33569
    },
    {
      "epoch": 1.540544261392318,
      "grad_norm": 0.4713311791419983,
      "learning_rate": 4.943730563143578e-06,
      "loss": 0.3861,
      "step": 33570
    },
    {
      "epoch": 1.5405901518975724,
      "grad_norm": 0.4796547293663025,
      "learning_rate": 4.943485390831252e-06,
      "loss": 0.3881,
      "step": 33571
    },
    {
      "epoch": 1.5406360424028267,
      "grad_norm": 0.4474563002586365,
      "learning_rate": 4.943240218654825e-06,
      "loss": 0.3081,
      "step": 33572
    },
    {
      "epoch": 1.5406819329080812,
      "grad_norm": 0.5067110061645508,
      "learning_rate": 4.9429950466148874e-06,
      "loss": 0.3579,
      "step": 33573
    },
    {
      "epoch": 1.5407278234133357,
      "grad_norm": 0.4425121247768402,
      "learning_rate": 4.94274987471203e-06,
      "loss": 0.3273,
      "step": 33574
    },
    {
      "epoch": 1.5407737139185902,
      "grad_norm": 0.436112642288208,
      "learning_rate": 4.942504702946838e-06,
      "loss": 0.2652,
      "step": 33575
    },
    {
      "epoch": 1.5408196044238447,
      "grad_norm": 0.47832804918289185,
      "learning_rate": 4.9422595313199065e-06,
      "loss": 0.3643,
      "step": 33576
    },
    {
      "epoch": 1.5408654949290992,
      "grad_norm": 0.49406495690345764,
      "learning_rate": 4.942014359831822e-06,
      "loss": 0.4523,
      "step": 33577
    },
    {
      "epoch": 1.5409113854343537,
      "grad_norm": 0.45660126209259033,
      "learning_rate": 4.941769188483174e-06,
      "loss": 0.3612,
      "step": 33578
    },
    {
      "epoch": 1.5409572759396082,
      "grad_norm": 0.43226879835128784,
      "learning_rate": 4.941524017274554e-06,
      "loss": 0.3537,
      "step": 33579
    },
    {
      "epoch": 1.5410031664448627,
      "grad_norm": 0.4497494697570801,
      "learning_rate": 4.94127884620655e-06,
      "loss": 0.3312,
      "step": 33580
    },
    {
      "epoch": 1.5410490569501172,
      "grad_norm": 0.4486096501350403,
      "learning_rate": 4.94103367527975e-06,
      "loss": 0.3361,
      "step": 33581
    },
    {
      "epoch": 1.5410949474553715,
      "grad_norm": 0.4650793671607971,
      "learning_rate": 4.940788504494748e-06,
      "loss": 0.3462,
      "step": 33582
    },
    {
      "epoch": 1.541140837960626,
      "grad_norm": 0.5677633881568909,
      "learning_rate": 4.94054333385213e-06,
      "loss": 0.4902,
      "step": 33583
    },
    {
      "epoch": 1.5411867284658805,
      "grad_norm": 0.4618867337703705,
      "learning_rate": 4.940298163352486e-06,
      "loss": 0.3343,
      "step": 33584
    },
    {
      "epoch": 1.5412326189711347,
      "grad_norm": 0.5067033767700195,
      "learning_rate": 4.940052992996407e-06,
      "loss": 0.3897,
      "step": 33585
    },
    {
      "epoch": 1.5412785094763892,
      "grad_norm": 0.5166358351707458,
      "learning_rate": 4.939807822784482e-06,
      "loss": 0.44,
      "step": 33586
    },
    {
      "epoch": 1.5413243999816437,
      "grad_norm": 0.48250359296798706,
      "learning_rate": 4.939562652717298e-06,
      "loss": 0.3924,
      "step": 33587
    },
    {
      "epoch": 1.5413702904868982,
      "grad_norm": 0.45088499784469604,
      "learning_rate": 4.939317482795449e-06,
      "loss": 0.305,
      "step": 33588
    },
    {
      "epoch": 1.5414161809921527,
      "grad_norm": 0.46606260538101196,
      "learning_rate": 4.939072313019522e-06,
      "loss": 0.3457,
      "step": 33589
    },
    {
      "epoch": 1.5414620714974072,
      "grad_norm": 0.484151691198349,
      "learning_rate": 4.938827143390104e-06,
      "loss": 0.3814,
      "step": 33590
    },
    {
      "epoch": 1.5415079620026617,
      "grad_norm": 0.4708734154701233,
      "learning_rate": 4.938581973907791e-06,
      "loss": 0.3373,
      "step": 33591
    },
    {
      "epoch": 1.5415538525079162,
      "grad_norm": 0.45363467931747437,
      "learning_rate": 4.938336804573168e-06,
      "loss": 0.3345,
      "step": 33592
    },
    {
      "epoch": 1.5415997430131707,
      "grad_norm": 0.5060080289840698,
      "learning_rate": 4.938091635386824e-06,
      "loss": 0.3991,
      "step": 33593
    },
    {
      "epoch": 1.541645633518425,
      "grad_norm": 0.46949246525764465,
      "learning_rate": 4.937846466349352e-06,
      "loss": 0.362,
      "step": 33594
    },
    {
      "epoch": 1.5416915240236795,
      "grad_norm": 0.4366362988948822,
      "learning_rate": 4.937601297461339e-06,
      "loss": 0.297,
      "step": 33595
    },
    {
      "epoch": 1.541737414528934,
      "grad_norm": 0.49495115876197815,
      "learning_rate": 4.937356128723373e-06,
      "loss": 0.3648,
      "step": 33596
    },
    {
      "epoch": 1.5417833050341885,
      "grad_norm": 0.47930920124053955,
      "learning_rate": 4.937110960136049e-06,
      "loss": 0.3915,
      "step": 33597
    },
    {
      "epoch": 1.5418291955394428,
      "grad_norm": 0.44499385356903076,
      "learning_rate": 4.936865791699953e-06,
      "loss": 0.3198,
      "step": 33598
    },
    {
      "epoch": 1.5418750860446973,
      "grad_norm": 0.45275044441223145,
      "learning_rate": 4.936620623415672e-06,
      "loss": 0.3223,
      "step": 33599
    },
    {
      "epoch": 1.5419209765499517,
      "grad_norm": 0.4935304522514343,
      "learning_rate": 4.936375455283802e-06,
      "loss": 0.3444,
      "step": 33600
    },
    {
      "epoch": 1.5419668670552062,
      "grad_norm": 0.4994998574256897,
      "learning_rate": 4.936130287304928e-06,
      "loss": 0.4172,
      "step": 33601
    },
    {
      "epoch": 1.5420127575604607,
      "grad_norm": 0.46916863322257996,
      "learning_rate": 4.935885119479638e-06,
      "loss": 0.3931,
      "step": 33602
    },
    {
      "epoch": 1.5420586480657152,
      "grad_norm": 0.4675274193286896,
      "learning_rate": 4.9356399518085256e-06,
      "loss": 0.3485,
      "step": 33603
    },
    {
      "epoch": 1.5421045385709697,
      "grad_norm": 0.5057913661003113,
      "learning_rate": 4.93539478429218e-06,
      "loss": 0.3885,
      "step": 33604
    },
    {
      "epoch": 1.5421504290762242,
      "grad_norm": 0.4735415279865265,
      "learning_rate": 4.935149616931189e-06,
      "loss": 0.37,
      "step": 33605
    },
    {
      "epoch": 1.5421963195814787,
      "grad_norm": 0.4587606191635132,
      "learning_rate": 4.934904449726141e-06,
      "loss": 0.3049,
      "step": 33606
    },
    {
      "epoch": 1.542242210086733,
      "grad_norm": 0.5668028593063354,
      "learning_rate": 4.93465928267763e-06,
      "loss": 0.465,
      "step": 33607
    },
    {
      "epoch": 1.5422881005919875,
      "grad_norm": 0.4814692437648773,
      "learning_rate": 4.934414115786242e-06,
      "loss": 0.3679,
      "step": 33608
    },
    {
      "epoch": 1.542333991097242,
      "grad_norm": 0.5004754662513733,
      "learning_rate": 4.934168949052564e-06,
      "loss": 0.3914,
      "step": 33609
    },
    {
      "epoch": 1.5423798816024963,
      "grad_norm": 0.4443148374557495,
      "learning_rate": 4.933923782477193e-06,
      "loss": 0.3163,
      "step": 33610
    },
    {
      "epoch": 1.5424257721077508,
      "grad_norm": 0.5136976838111877,
      "learning_rate": 4.933678616060714e-06,
      "loss": 0.3177,
      "step": 33611
    },
    {
      "epoch": 1.5424716626130053,
      "grad_norm": 0.4516414403915405,
      "learning_rate": 4.933433449803715e-06,
      "loss": 0.3265,
      "step": 33612
    },
    {
      "epoch": 1.5425175531182598,
      "grad_norm": 0.48030439019203186,
      "learning_rate": 4.9331882837067884e-06,
      "loss": 0.3425,
      "step": 33613
    },
    {
      "epoch": 1.5425634436235143,
      "grad_norm": 0.47294512391090393,
      "learning_rate": 4.932943117770524e-06,
      "loss": 0.3638,
      "step": 33614
    },
    {
      "epoch": 1.5426093341287688,
      "grad_norm": 0.4611269533634186,
      "learning_rate": 4.932697951995509e-06,
      "loss": 0.3431,
      "step": 33615
    },
    {
      "epoch": 1.5426552246340233,
      "grad_norm": 0.4959776997566223,
      "learning_rate": 4.932452786382335e-06,
      "loss": 0.3477,
      "step": 33616
    },
    {
      "epoch": 1.5427011151392778,
      "grad_norm": 0.48662444949150085,
      "learning_rate": 4.932207620931591e-06,
      "loss": 0.3364,
      "step": 33617
    },
    {
      "epoch": 1.5427470056445323,
      "grad_norm": 0.5055026412010193,
      "learning_rate": 4.931962455643863e-06,
      "loss": 0.4492,
      "step": 33618
    },
    {
      "epoch": 1.5427928961497868,
      "grad_norm": 0.49302318692207336,
      "learning_rate": 4.931717290519747e-06,
      "loss": 0.3854,
      "step": 33619
    },
    {
      "epoch": 1.542838786655041,
      "grad_norm": 0.4855456054210663,
      "learning_rate": 4.931472125559829e-06,
      "loss": 0.341,
      "step": 33620
    },
    {
      "epoch": 1.5428846771602955,
      "grad_norm": 0.5016556978225708,
      "learning_rate": 4.931226960764698e-06,
      "loss": 0.3932,
      "step": 33621
    },
    {
      "epoch": 1.54293056766555,
      "grad_norm": 0.523742139339447,
      "learning_rate": 4.930981796134946e-06,
      "loss": 0.46,
      "step": 33622
    },
    {
      "epoch": 1.5429764581708043,
      "grad_norm": 0.5382898449897766,
      "learning_rate": 4.930736631671161e-06,
      "loss": 0.4623,
      "step": 33623
    },
    {
      "epoch": 1.5430223486760588,
      "grad_norm": 0.5127125382423401,
      "learning_rate": 4.930491467373929e-06,
      "loss": 0.4361,
      "step": 33624
    },
    {
      "epoch": 1.5430682391813133,
      "grad_norm": 0.4651939868927002,
      "learning_rate": 4.930246303243846e-06,
      "loss": 0.3789,
      "step": 33625
    },
    {
      "epoch": 1.5431141296865678,
      "grad_norm": 0.47741934657096863,
      "learning_rate": 4.930001139281499e-06,
      "loss": 0.352,
      "step": 33626
    },
    {
      "epoch": 1.5431600201918223,
      "grad_norm": 0.47967615723609924,
      "learning_rate": 4.929755975487476e-06,
      "loss": 0.3321,
      "step": 33627
    },
    {
      "epoch": 1.5432059106970768,
      "grad_norm": 0.5015032887458801,
      "learning_rate": 4.929510811862369e-06,
      "loss": 0.4331,
      "step": 33628
    },
    {
      "epoch": 1.5432518012023313,
      "grad_norm": 0.4766474962234497,
      "learning_rate": 4.929265648406766e-06,
      "loss": 0.3785,
      "step": 33629
    },
    {
      "epoch": 1.5432976917075858,
      "grad_norm": 0.498293936252594,
      "learning_rate": 4.929020485121254e-06,
      "loss": 0.4095,
      "step": 33630
    },
    {
      "epoch": 1.5433435822128403,
      "grad_norm": 0.546868622303009,
      "learning_rate": 4.928775322006428e-06,
      "loss": 0.3927,
      "step": 33631
    },
    {
      "epoch": 1.5433894727180948,
      "grad_norm": 0.5303363800048828,
      "learning_rate": 4.928530159062876e-06,
      "loss": 0.4343,
      "step": 33632
    },
    {
      "epoch": 1.543435363223349,
      "grad_norm": 0.47333669662475586,
      "learning_rate": 4.928284996291183e-06,
      "loss": 0.3722,
      "step": 33633
    },
    {
      "epoch": 1.5434812537286036,
      "grad_norm": 0.4681225121021271,
      "learning_rate": 4.928039833691945e-06,
      "loss": 0.3517,
      "step": 33634
    },
    {
      "epoch": 1.543527144233858,
      "grad_norm": 0.49124595522880554,
      "learning_rate": 4.927794671265748e-06,
      "loss": 0.3695,
      "step": 33635
    },
    {
      "epoch": 1.5435730347391123,
      "grad_norm": 0.4366133511066437,
      "learning_rate": 4.927549509013181e-06,
      "loss": 0.2735,
      "step": 33636
    },
    {
      "epoch": 1.5436189252443668,
      "grad_norm": 0.48148953914642334,
      "learning_rate": 4.927304346934834e-06,
      "loss": 0.3843,
      "step": 33637
    },
    {
      "epoch": 1.5436648157496213,
      "grad_norm": 0.48762959241867065,
      "learning_rate": 4.927059185031299e-06,
      "loss": 0.4392,
      "step": 33638
    },
    {
      "epoch": 1.5437107062548758,
      "grad_norm": 0.5201305747032166,
      "learning_rate": 4.926814023303163e-06,
      "loss": 0.4282,
      "step": 33639
    },
    {
      "epoch": 1.5437565967601303,
      "grad_norm": 0.4859296977519989,
      "learning_rate": 4.926568861751014e-06,
      "loss": 0.3947,
      "step": 33640
    },
    {
      "epoch": 1.5438024872653848,
      "grad_norm": 0.47841450572013855,
      "learning_rate": 4.926323700375446e-06,
      "loss": 0.3566,
      "step": 33641
    },
    {
      "epoch": 1.5438483777706393,
      "grad_norm": 0.44055384397506714,
      "learning_rate": 4.926078539177046e-06,
      "loss": 0.3354,
      "step": 33642
    },
    {
      "epoch": 1.5438942682758938,
      "grad_norm": 0.48486483097076416,
      "learning_rate": 4.925833378156403e-06,
      "loss": 0.3561,
      "step": 33643
    },
    {
      "epoch": 1.5439401587811483,
      "grad_norm": 0.49194344878196716,
      "learning_rate": 4.925588217314109e-06,
      "loss": 0.3833,
      "step": 33644
    },
    {
      "epoch": 1.5439860492864026,
      "grad_norm": 0.48740342259407043,
      "learning_rate": 4.925343056650751e-06,
      "loss": 0.4077,
      "step": 33645
    },
    {
      "epoch": 1.544031939791657,
      "grad_norm": 0.4789520800113678,
      "learning_rate": 4.925097896166917e-06,
      "loss": 0.3763,
      "step": 33646
    },
    {
      "epoch": 1.5440778302969116,
      "grad_norm": 0.44591280817985535,
      "learning_rate": 4.924852735863202e-06,
      "loss": 0.31,
      "step": 33647
    },
    {
      "epoch": 1.5441237208021659,
      "grad_norm": 0.47929877042770386,
      "learning_rate": 4.924607575740192e-06,
      "loss": 0.3907,
      "step": 33648
    },
    {
      "epoch": 1.5441696113074204,
      "grad_norm": 0.43600186705589294,
      "learning_rate": 4.9243624157984755e-06,
      "loss": 0.3173,
      "step": 33649
    },
    {
      "epoch": 1.5442155018126749,
      "grad_norm": 0.4479968547821045,
      "learning_rate": 4.924117256038645e-06,
      "loss": 0.3051,
      "step": 33650
    },
    {
      "epoch": 1.5442613923179294,
      "grad_norm": 0.47753340005874634,
      "learning_rate": 4.923872096461288e-06,
      "loss": 0.3659,
      "step": 33651
    },
    {
      "epoch": 1.5443072828231839,
      "grad_norm": 0.47188812494277954,
      "learning_rate": 4.923626937066994e-06,
      "loss": 0.368,
      "step": 33652
    },
    {
      "epoch": 1.5443531733284384,
      "grad_norm": 0.5013169050216675,
      "learning_rate": 4.923381777856353e-06,
      "loss": 0.3894,
      "step": 33653
    },
    {
      "epoch": 1.5443990638336929,
      "grad_norm": 0.44498521089553833,
      "learning_rate": 4.923136618829958e-06,
      "loss": 0.3007,
      "step": 33654
    },
    {
      "epoch": 1.5444449543389474,
      "grad_norm": 0.4475548565387726,
      "learning_rate": 4.92289145998839e-06,
      "loss": 0.3414,
      "step": 33655
    },
    {
      "epoch": 1.5444908448442018,
      "grad_norm": 0.4787669777870178,
      "learning_rate": 4.922646301332247e-06,
      "loss": 0.3788,
      "step": 33656
    },
    {
      "epoch": 1.5445367353494563,
      "grad_norm": 0.4321731626987457,
      "learning_rate": 4.922401142862115e-06,
      "loss": 0.3173,
      "step": 33657
    },
    {
      "epoch": 1.5445826258547106,
      "grad_norm": 0.5022450685501099,
      "learning_rate": 4.9221559845785824e-06,
      "loss": 0.3322,
      "step": 33658
    },
    {
      "epoch": 1.5446285163599651,
      "grad_norm": 0.4767553210258484,
      "learning_rate": 4.921910826482242e-06,
      "loss": 0.3268,
      "step": 33659
    },
    {
      "epoch": 1.5446744068652196,
      "grad_norm": 0.4745975434780121,
      "learning_rate": 4.92166566857368e-06,
      "loss": 0.4289,
      "step": 33660
    },
    {
      "epoch": 1.544720297370474,
      "grad_norm": 0.5147960782051086,
      "learning_rate": 4.9214205108534866e-06,
      "loss": 0.4689,
      "step": 33661
    },
    {
      "epoch": 1.5447661878757284,
      "grad_norm": 0.47956761717796326,
      "learning_rate": 4.921175353322253e-06,
      "loss": 0.4306,
      "step": 33662
    },
    {
      "epoch": 1.544812078380983,
      "grad_norm": 0.45985832810401917,
      "learning_rate": 4.920930195980569e-06,
      "loss": 0.3608,
      "step": 33663
    },
    {
      "epoch": 1.5448579688862374,
      "grad_norm": 0.4663658142089844,
      "learning_rate": 4.920685038829021e-06,
      "loss": 0.374,
      "step": 33664
    },
    {
      "epoch": 1.5449038593914919,
      "grad_norm": 0.487166166305542,
      "learning_rate": 4.920439881868202e-06,
      "loss": 0.3584,
      "step": 33665
    },
    {
      "epoch": 1.5449497498967464,
      "grad_norm": 0.48380008339881897,
      "learning_rate": 4.9201947250987e-06,
      "loss": 0.3852,
      "step": 33666
    },
    {
      "epoch": 1.5449956404020009,
      "grad_norm": 0.44474971294403076,
      "learning_rate": 4.919949568521102e-06,
      "loss": 0.3135,
      "step": 33667
    },
    {
      "epoch": 1.5450415309072554,
      "grad_norm": 0.511354386806488,
      "learning_rate": 4.919704412136002e-06,
      "loss": 0.3823,
      "step": 33668
    },
    {
      "epoch": 1.5450874214125099,
      "grad_norm": 0.4838235676288605,
      "learning_rate": 4.919459255943988e-06,
      "loss": 0.3903,
      "step": 33669
    },
    {
      "epoch": 1.5451333119177644,
      "grad_norm": 0.4718272089958191,
      "learning_rate": 4.919214099945649e-06,
      "loss": 0.3437,
      "step": 33670
    },
    {
      "epoch": 1.5451792024230186,
      "grad_norm": 0.909521758556366,
      "learning_rate": 4.918968944141573e-06,
      "loss": 0.3736,
      "step": 33671
    },
    {
      "epoch": 1.5452250929282731,
      "grad_norm": 0.5044870972633362,
      "learning_rate": 4.918723788532353e-06,
      "loss": 0.3897,
      "step": 33672
    },
    {
      "epoch": 1.5452709834335276,
      "grad_norm": 0.4717123508453369,
      "learning_rate": 4.9184786331185765e-06,
      "loss": 0.3565,
      "step": 33673
    },
    {
      "epoch": 1.545316873938782,
      "grad_norm": 0.44002804160118103,
      "learning_rate": 4.9182334779008305e-06,
      "loss": 0.2778,
      "step": 33674
    },
    {
      "epoch": 1.5453627644440364,
      "grad_norm": 0.504742443561554,
      "learning_rate": 4.91798832287971e-06,
      "loss": 0.3756,
      "step": 33675
    },
    {
      "epoch": 1.545408654949291,
      "grad_norm": 0.42752397060394287,
      "learning_rate": 4.917743168055802e-06,
      "loss": 0.3017,
      "step": 33676
    },
    {
      "epoch": 1.5454545454545454,
      "grad_norm": 0.44653525948524475,
      "learning_rate": 4.917498013429693e-06,
      "loss": 0.348,
      "step": 33677
    },
    {
      "epoch": 1.5455004359598,
      "grad_norm": 0.5015146136283875,
      "learning_rate": 4.9172528590019775e-06,
      "loss": 0.4103,
      "step": 33678
    },
    {
      "epoch": 1.5455463264650544,
      "grad_norm": 0.43168386816978455,
      "learning_rate": 4.9170077047732425e-06,
      "loss": 0.2808,
      "step": 33679
    },
    {
      "epoch": 1.545592216970309,
      "grad_norm": 0.47256985306739807,
      "learning_rate": 4.916762550744076e-06,
      "loss": 0.3495,
      "step": 33680
    },
    {
      "epoch": 1.5456381074755634,
      "grad_norm": 0.48931726813316345,
      "learning_rate": 4.916517396915071e-06,
      "loss": 0.4532,
      "step": 33681
    },
    {
      "epoch": 1.545683997980818,
      "grad_norm": 0.46851399540901184,
      "learning_rate": 4.9162722432868156e-06,
      "loss": 0.3702,
      "step": 33682
    },
    {
      "epoch": 1.5457298884860722,
      "grad_norm": 0.4916781187057495,
      "learning_rate": 4.916027089859897e-06,
      "loss": 0.3554,
      "step": 33683
    },
    {
      "epoch": 1.5457757789913267,
      "grad_norm": 0.47459301352500916,
      "learning_rate": 4.915781936634909e-06,
      "loss": 0.3749,
      "step": 33684
    },
    {
      "epoch": 1.5458216694965812,
      "grad_norm": 0.45352235436439514,
      "learning_rate": 4.915536783612438e-06,
      "loss": 0.2996,
      "step": 33685
    },
    {
      "epoch": 1.5458675600018357,
      "grad_norm": 0.4767163097858429,
      "learning_rate": 4.915291630793074e-06,
      "loss": 0.3613,
      "step": 33686
    },
    {
      "epoch": 1.54591345050709,
      "grad_norm": 0.5026130080223083,
      "learning_rate": 4.9150464781774074e-06,
      "loss": 0.4085,
      "step": 33687
    },
    {
      "epoch": 1.5459593410123444,
      "grad_norm": 0.5052232146263123,
      "learning_rate": 4.914801325766027e-06,
      "loss": 0.4035,
      "step": 33688
    },
    {
      "epoch": 1.546005231517599,
      "grad_norm": 0.4421968460083008,
      "learning_rate": 4.914556173559522e-06,
      "loss": 0.3008,
      "step": 33689
    },
    {
      "epoch": 1.5460511220228534,
      "grad_norm": 0.4664362072944641,
      "learning_rate": 4.9143110215584825e-06,
      "loss": 0.3868,
      "step": 33690
    },
    {
      "epoch": 1.546097012528108,
      "grad_norm": 0.5112592577934265,
      "learning_rate": 4.914065869763499e-06,
      "loss": 0.3991,
      "step": 33691
    },
    {
      "epoch": 1.5461429030333624,
      "grad_norm": 0.4529246687889099,
      "learning_rate": 4.913820718175159e-06,
      "loss": 0.2979,
      "step": 33692
    },
    {
      "epoch": 1.546188793538617,
      "grad_norm": 0.49521708488464355,
      "learning_rate": 4.913575566794054e-06,
      "loss": 0.3917,
      "step": 33693
    },
    {
      "epoch": 1.5462346840438714,
      "grad_norm": 0.47084107995033264,
      "learning_rate": 4.913330415620773e-06,
      "loss": 0.3448,
      "step": 33694
    },
    {
      "epoch": 1.546280574549126,
      "grad_norm": 0.47486555576324463,
      "learning_rate": 4.913085264655901e-06,
      "loss": 0.3585,
      "step": 33695
    },
    {
      "epoch": 1.5463264650543802,
      "grad_norm": 0.47531548142433167,
      "learning_rate": 4.912840113900035e-06,
      "loss": 0.3732,
      "step": 33696
    },
    {
      "epoch": 1.5463723555596347,
      "grad_norm": 0.467709481716156,
      "learning_rate": 4.9125949633537614e-06,
      "loss": 0.334,
      "step": 33697
    },
    {
      "epoch": 1.5464182460648892,
      "grad_norm": 0.4966362416744232,
      "learning_rate": 4.912349813017667e-06,
      "loss": 0.3814,
      "step": 33698
    },
    {
      "epoch": 1.5464641365701435,
      "grad_norm": 0.5105735659599304,
      "learning_rate": 4.912104662892346e-06,
      "loss": 0.4094,
      "step": 33699
    },
    {
      "epoch": 1.546510027075398,
      "grad_norm": 0.4202808141708374,
      "learning_rate": 4.911859512978385e-06,
      "loss": 0.2941,
      "step": 33700
    },
    {
      "epoch": 1.5465559175806525,
      "grad_norm": 0.47547367215156555,
      "learning_rate": 4.911614363276374e-06,
      "loss": 0.3877,
      "step": 33701
    },
    {
      "epoch": 1.546601808085907,
      "grad_norm": 0.4933498501777649,
      "learning_rate": 4.911369213786901e-06,
      "loss": 0.3992,
      "step": 33702
    },
    {
      "epoch": 1.5466476985911615,
      "grad_norm": 0.46759510040283203,
      "learning_rate": 4.911124064510559e-06,
      "loss": 0.3371,
      "step": 33703
    },
    {
      "epoch": 1.546693589096416,
      "grad_norm": 0.47216543555259705,
      "learning_rate": 4.910878915447935e-06,
      "loss": 0.3447,
      "step": 33704
    },
    {
      "epoch": 1.5467394796016705,
      "grad_norm": 0.5177017450332642,
      "learning_rate": 4.910633766599617e-06,
      "loss": 0.3955,
      "step": 33705
    },
    {
      "epoch": 1.546785370106925,
      "grad_norm": 0.48004966974258423,
      "learning_rate": 4.910388617966199e-06,
      "loss": 0.3952,
      "step": 33706
    },
    {
      "epoch": 1.5468312606121795,
      "grad_norm": 0.4812425971031189,
      "learning_rate": 4.910143469548269e-06,
      "loss": 0.3341,
      "step": 33707
    },
    {
      "epoch": 1.546877151117434,
      "grad_norm": 0.5259968042373657,
      "learning_rate": 4.909898321346413e-06,
      "loss": 0.3306,
      "step": 33708
    },
    {
      "epoch": 1.5469230416226882,
      "grad_norm": 0.4751369059085846,
      "learning_rate": 4.909653173361225e-06,
      "loss": 0.3628,
      "step": 33709
    },
    {
      "epoch": 1.5469689321279427,
      "grad_norm": 0.501651406288147,
      "learning_rate": 4.909408025593292e-06,
      "loss": 0.4191,
      "step": 33710
    },
    {
      "epoch": 1.5470148226331972,
      "grad_norm": 0.5016651153564453,
      "learning_rate": 4.909162878043202e-06,
      "loss": 0.423,
      "step": 33711
    },
    {
      "epoch": 1.5470607131384515,
      "grad_norm": 0.48306190967559814,
      "learning_rate": 4.9089177307115495e-06,
      "loss": 0.3661,
      "step": 33712
    },
    {
      "epoch": 1.547106603643706,
      "grad_norm": 0.48112261295318604,
      "learning_rate": 4.90867258359892e-06,
      "loss": 0.3718,
      "step": 33713
    },
    {
      "epoch": 1.5471524941489605,
      "grad_norm": 0.4686889350414276,
      "learning_rate": 4.908427436705904e-06,
      "loss": 0.3199,
      "step": 33714
    },
    {
      "epoch": 1.547198384654215,
      "grad_norm": 0.4698586165904999,
      "learning_rate": 4.908182290033092e-06,
      "loss": 0.3481,
      "step": 33715
    },
    {
      "epoch": 1.5472442751594695,
      "grad_norm": 0.47841984033584595,
      "learning_rate": 4.907937143581072e-06,
      "loss": 0.3543,
      "step": 33716
    },
    {
      "epoch": 1.547290165664724,
      "grad_norm": 0.4799274802207947,
      "learning_rate": 4.907691997350432e-06,
      "loss": 0.3854,
      "step": 33717
    },
    {
      "epoch": 1.5473360561699785,
      "grad_norm": 0.44813692569732666,
      "learning_rate": 4.907446851341766e-06,
      "loss": 0.3274,
      "step": 33718
    },
    {
      "epoch": 1.547381946675233,
      "grad_norm": 0.49787187576293945,
      "learning_rate": 4.907201705555662e-06,
      "loss": 0.3446,
      "step": 33719
    },
    {
      "epoch": 1.5474278371804875,
      "grad_norm": 0.5546039342880249,
      "learning_rate": 4.906956559992707e-06,
      "loss": 0.4069,
      "step": 33720
    },
    {
      "epoch": 1.547473727685742,
      "grad_norm": 0.49720388650894165,
      "learning_rate": 4.906711414653493e-06,
      "loss": 0.3823,
      "step": 33721
    },
    {
      "epoch": 1.5475196181909963,
      "grad_norm": 0.4946902096271515,
      "learning_rate": 4.906466269538609e-06,
      "loss": 0.435,
      "step": 33722
    },
    {
      "epoch": 1.5475655086962508,
      "grad_norm": 0.4508228302001953,
      "learning_rate": 4.906221124648642e-06,
      "loss": 0.3277,
      "step": 33723
    },
    {
      "epoch": 1.5476113992015053,
      "grad_norm": 0.4800426959991455,
      "learning_rate": 4.905975979984186e-06,
      "loss": 0.3583,
      "step": 33724
    },
    {
      "epoch": 1.5476572897067595,
      "grad_norm": 0.48433807492256165,
      "learning_rate": 4.905730835545827e-06,
      "loss": 0.4073,
      "step": 33725
    },
    {
      "epoch": 1.547703180212014,
      "grad_norm": 0.4893975853919983,
      "learning_rate": 4.905485691334153e-06,
      "loss": 0.4089,
      "step": 33726
    },
    {
      "epoch": 1.5477490707172685,
      "grad_norm": 0.48508909344673157,
      "learning_rate": 4.90524054734976e-06,
      "loss": 0.4032,
      "step": 33727
    },
    {
      "epoch": 1.547794961222523,
      "grad_norm": 0.44432926177978516,
      "learning_rate": 4.904995403593232e-06,
      "loss": 0.3373,
      "step": 33728
    },
    {
      "epoch": 1.5478408517277775,
      "grad_norm": 0.45035335421562195,
      "learning_rate": 4.90475026006516e-06,
      "loss": 0.3163,
      "step": 33729
    },
    {
      "epoch": 1.547886742233032,
      "grad_norm": 0.5301679968833923,
      "learning_rate": 4.904505116766134e-06,
      "loss": 0.3468,
      "step": 33730
    },
    {
      "epoch": 1.5479326327382865,
      "grad_norm": 0.43505701422691345,
      "learning_rate": 4.904259973696744e-06,
      "loss": 0.2741,
      "step": 33731
    },
    {
      "epoch": 1.547978523243541,
      "grad_norm": 0.4907168745994568,
      "learning_rate": 4.904014830857575e-06,
      "loss": 0.3747,
      "step": 33732
    },
    {
      "epoch": 1.5480244137487955,
      "grad_norm": 0.500189483165741,
      "learning_rate": 4.903769688249224e-06,
      "loss": 0.4033,
      "step": 33733
    },
    {
      "epoch": 1.5480703042540498,
      "grad_norm": 0.4848182499408722,
      "learning_rate": 4.903524545872275e-06,
      "loss": 0.3313,
      "step": 33734
    },
    {
      "epoch": 1.5481161947593043,
      "grad_norm": 0.45743417739868164,
      "learning_rate": 4.903279403727319e-06,
      "loss": 0.3007,
      "step": 33735
    },
    {
      "epoch": 1.5481620852645588,
      "grad_norm": 0.47727182507514954,
      "learning_rate": 4.903034261814945e-06,
      "loss": 0.3791,
      "step": 33736
    },
    {
      "epoch": 1.548207975769813,
      "grad_norm": 0.45982953906059265,
      "learning_rate": 4.902789120135744e-06,
      "loss": 0.3835,
      "step": 33737
    },
    {
      "epoch": 1.5482538662750676,
      "grad_norm": 0.5072396397590637,
      "learning_rate": 4.902543978690305e-06,
      "loss": 0.485,
      "step": 33738
    },
    {
      "epoch": 1.548299756780322,
      "grad_norm": 0.5088459253311157,
      "learning_rate": 4.902298837479214e-06,
      "loss": 0.394,
      "step": 33739
    },
    {
      "epoch": 1.5483456472855766,
      "grad_norm": 0.49588534235954285,
      "learning_rate": 4.902053696503065e-06,
      "loss": 0.4047,
      "step": 33740
    },
    {
      "epoch": 1.548391537790831,
      "grad_norm": 0.48816725611686707,
      "learning_rate": 4.901808555762448e-06,
      "loss": 0.3825,
      "step": 33741
    },
    {
      "epoch": 1.5484374282960855,
      "grad_norm": 0.5040444135665894,
      "learning_rate": 4.9015634152579475e-06,
      "loss": 0.3718,
      "step": 33742
    },
    {
      "epoch": 1.54848331880134,
      "grad_norm": 0.536422073841095,
      "learning_rate": 4.901318274990159e-06,
      "loss": 0.3599,
      "step": 33743
    },
    {
      "epoch": 1.5485292093065945,
      "grad_norm": 0.4810895323753357,
      "learning_rate": 4.901073134959667e-06,
      "loss": 0.3683,
      "step": 33744
    },
    {
      "epoch": 1.548575099811849,
      "grad_norm": 0.45391982793807983,
      "learning_rate": 4.900827995167063e-06,
      "loss": 0.3394,
      "step": 33745
    },
    {
      "epoch": 1.5486209903171035,
      "grad_norm": 0.5132395029067993,
      "learning_rate": 4.900582855612936e-06,
      "loss": 0.4163,
      "step": 33746
    },
    {
      "epoch": 1.5486668808223578,
      "grad_norm": 0.48213669657707214,
      "learning_rate": 4.9003377162978764e-06,
      "loss": 0.3303,
      "step": 33747
    },
    {
      "epoch": 1.5487127713276123,
      "grad_norm": 0.4752504825592041,
      "learning_rate": 4.900092577222472e-06,
      "loss": 0.3435,
      "step": 33748
    },
    {
      "epoch": 1.5487586618328668,
      "grad_norm": 0.5047745108604431,
      "learning_rate": 4.899847438387315e-06,
      "loss": 0.3299,
      "step": 33749
    },
    {
      "epoch": 1.548804552338121,
      "grad_norm": 0.4779907763004303,
      "learning_rate": 4.899602299792993e-06,
      "loss": 0.3895,
      "step": 33750
    },
    {
      "epoch": 1.5488504428433756,
      "grad_norm": 0.4793146848678589,
      "learning_rate": 4.899357161440095e-06,
      "loss": 0.3781,
      "step": 33751
    },
    {
      "epoch": 1.54889633334863,
      "grad_norm": 0.4452204704284668,
      "learning_rate": 4.8991120233292116e-06,
      "loss": 0.3125,
      "step": 33752
    },
    {
      "epoch": 1.5489422238538846,
      "grad_norm": 0.46993136405944824,
      "learning_rate": 4.898866885460933e-06,
      "loss": 0.3661,
      "step": 33753
    },
    {
      "epoch": 1.548988114359139,
      "grad_norm": 0.47236761450767517,
      "learning_rate": 4.898621747835845e-06,
      "loss": 0.3645,
      "step": 33754
    },
    {
      "epoch": 1.5490340048643936,
      "grad_norm": 0.4964770972728729,
      "learning_rate": 4.898376610454542e-06,
      "loss": 0.4016,
      "step": 33755
    },
    {
      "epoch": 1.549079895369648,
      "grad_norm": 0.4356682598590851,
      "learning_rate": 4.89813147331761e-06,
      "loss": 0.263,
      "step": 33756
    },
    {
      "epoch": 1.5491257858749026,
      "grad_norm": 0.4814671277999878,
      "learning_rate": 4.8978863364256405e-06,
      "loss": 0.4095,
      "step": 33757
    },
    {
      "epoch": 1.549171676380157,
      "grad_norm": 0.4583836495876312,
      "learning_rate": 4.897641199779222e-06,
      "loss": 0.3413,
      "step": 33758
    },
    {
      "epoch": 1.5492175668854116,
      "grad_norm": 0.4807807207107544,
      "learning_rate": 4.8973960633789445e-06,
      "loss": 0.4078,
      "step": 33759
    },
    {
      "epoch": 1.5492634573906658,
      "grad_norm": 0.44903725385665894,
      "learning_rate": 4.897150927225395e-06,
      "loss": 0.2791,
      "step": 33760
    },
    {
      "epoch": 1.5493093478959203,
      "grad_norm": 0.45138269662857056,
      "learning_rate": 4.896905791319167e-06,
      "loss": 0.3398,
      "step": 33761
    },
    {
      "epoch": 1.5493552384011748,
      "grad_norm": 0.4773988425731659,
      "learning_rate": 4.8966606556608485e-06,
      "loss": 0.394,
      "step": 33762
    },
    {
      "epoch": 1.5494011289064291,
      "grad_norm": 0.5426749587059021,
      "learning_rate": 4.896415520251027e-06,
      "loss": 0.4537,
      "step": 33763
    },
    {
      "epoch": 1.5494470194116836,
      "grad_norm": 0.5032060146331787,
      "learning_rate": 4.896170385090295e-06,
      "loss": 0.3815,
      "step": 33764
    },
    {
      "epoch": 1.549492909916938,
      "grad_norm": 0.45466119050979614,
      "learning_rate": 4.89592525017924e-06,
      "loss": 0.3511,
      "step": 33765
    },
    {
      "epoch": 1.5495388004221926,
      "grad_norm": 0.46169033646583557,
      "learning_rate": 4.895680115518452e-06,
      "loss": 0.3609,
      "step": 33766
    },
    {
      "epoch": 1.549584690927447,
      "grad_norm": 0.45012903213500977,
      "learning_rate": 4.895434981108519e-06,
      "loss": 0.3172,
      "step": 33767
    },
    {
      "epoch": 1.5496305814327016,
      "grad_norm": 0.47289618849754333,
      "learning_rate": 4.895189846950034e-06,
      "loss": 0.3804,
      "step": 33768
    },
    {
      "epoch": 1.549676471937956,
      "grad_norm": 0.4494435489177704,
      "learning_rate": 4.894944713043583e-06,
      "loss": 0.3955,
      "step": 33769
    },
    {
      "epoch": 1.5497223624432106,
      "grad_norm": 0.48021695017814636,
      "learning_rate": 4.894699579389755e-06,
      "loss": 0.384,
      "step": 33770
    },
    {
      "epoch": 1.549768252948465,
      "grad_norm": 0.4484626054763794,
      "learning_rate": 4.894454445989144e-06,
      "loss": 0.307,
      "step": 33771
    },
    {
      "epoch": 1.5498141434537194,
      "grad_norm": 0.5259761214256287,
      "learning_rate": 4.8942093128423366e-06,
      "loss": 0.4155,
      "step": 33772
    },
    {
      "epoch": 1.5498600339589739,
      "grad_norm": 0.4955519139766693,
      "learning_rate": 4.893964179949921e-06,
      "loss": 0.3946,
      "step": 33773
    },
    {
      "epoch": 1.5499059244642284,
      "grad_norm": 0.39875203371047974,
      "learning_rate": 4.8937190473124895e-06,
      "loss": 0.266,
      "step": 33774
    },
    {
      "epoch": 1.5499518149694829,
      "grad_norm": 0.45695504546165466,
      "learning_rate": 4.89347391493063e-06,
      "loss": 0.3552,
      "step": 33775
    },
    {
      "epoch": 1.5499977054747371,
      "grad_norm": 0.44218578934669495,
      "learning_rate": 4.89322878280493e-06,
      "loss": 0.2991,
      "step": 33776
    },
    {
      "epoch": 1.5500435959799916,
      "grad_norm": 0.41821911931037903,
      "learning_rate": 4.892983650935983e-06,
      "loss": 0.2794,
      "step": 33777
    },
    {
      "epoch": 1.5500894864852461,
      "grad_norm": 0.5973734855651855,
      "learning_rate": 4.8927385193243775e-06,
      "loss": 0.4021,
      "step": 33778
    },
    {
      "epoch": 1.5501353769905006,
      "grad_norm": 0.4495666027069092,
      "learning_rate": 4.892493387970699e-06,
      "loss": 0.3455,
      "step": 33779
    },
    {
      "epoch": 1.5501812674957551,
      "grad_norm": 0.4741359353065491,
      "learning_rate": 4.8922482568755425e-06,
      "loss": 0.3859,
      "step": 33780
    },
    {
      "epoch": 1.5502271580010096,
      "grad_norm": 0.4612380266189575,
      "learning_rate": 4.892003126039494e-06,
      "loss": 0.3678,
      "step": 33781
    },
    {
      "epoch": 1.5502730485062641,
      "grad_norm": 0.5034440159797668,
      "learning_rate": 4.891757995463144e-06,
      "loss": 0.4165,
      "step": 33782
    },
    {
      "epoch": 1.5503189390115186,
      "grad_norm": 0.43555501103401184,
      "learning_rate": 4.891512865147081e-06,
      "loss": 0.3019,
      "step": 33783
    },
    {
      "epoch": 1.5503648295167731,
      "grad_norm": 0.46017640829086304,
      "learning_rate": 4.891267735091898e-06,
      "loss": 0.3691,
      "step": 33784
    },
    {
      "epoch": 1.5504107200220274,
      "grad_norm": 0.48416104912757874,
      "learning_rate": 4.8910226052981785e-06,
      "loss": 0.4106,
      "step": 33785
    },
    {
      "epoch": 1.550456610527282,
      "grad_norm": 0.4599337577819824,
      "learning_rate": 4.890777475766518e-06,
      "loss": 0.324,
      "step": 33786
    },
    {
      "epoch": 1.5505025010325364,
      "grad_norm": 0.4213723838329315,
      "learning_rate": 4.890532346497502e-06,
      "loss": 0.2552,
      "step": 33787
    },
    {
      "epoch": 1.5505483915377907,
      "grad_norm": 0.4804609715938568,
      "learning_rate": 4.89028721749172e-06,
      "loss": 0.3838,
      "step": 33788
    },
    {
      "epoch": 1.5505942820430452,
      "grad_norm": 0.45456886291503906,
      "learning_rate": 4.890042088749765e-06,
      "loss": 0.3154,
      "step": 33789
    },
    {
      "epoch": 1.5506401725482997,
      "grad_norm": 0.4597671926021576,
      "learning_rate": 4.889796960272223e-06,
      "loss": 0.3319,
      "step": 33790
    },
    {
      "epoch": 1.5506860630535542,
      "grad_norm": 0.4779617488384247,
      "learning_rate": 4.889551832059682e-06,
      "loss": 0.3744,
      "step": 33791
    },
    {
      "epoch": 1.5507319535588087,
      "grad_norm": 0.4518517255783081,
      "learning_rate": 4.889306704112737e-06,
      "loss": 0.3044,
      "step": 33792
    },
    {
      "epoch": 1.5507778440640632,
      "grad_norm": 0.5017098188400269,
      "learning_rate": 4.889061576431975e-06,
      "loss": 0.3672,
      "step": 33793
    },
    {
      "epoch": 1.5508237345693177,
      "grad_norm": 0.42467671632766724,
      "learning_rate": 4.888816449017983e-06,
      "loss": 0.3012,
      "step": 33794
    },
    {
      "epoch": 1.5508696250745722,
      "grad_norm": 0.48134827613830566,
      "learning_rate": 4.8885713218713535e-06,
      "loss": 0.3808,
      "step": 33795
    },
    {
      "epoch": 1.5509155155798267,
      "grad_norm": 0.42205482721328735,
      "learning_rate": 4.8883261949926745e-06,
      "loss": 0.3089,
      "step": 33796
    },
    {
      "epoch": 1.5509614060850812,
      "grad_norm": 0.4670848548412323,
      "learning_rate": 4.888081068382534e-06,
      "loss": 0.357,
      "step": 33797
    },
    {
      "epoch": 1.5510072965903354,
      "grad_norm": 0.432318776845932,
      "learning_rate": 4.887835942041525e-06,
      "loss": 0.3048,
      "step": 33798
    },
    {
      "epoch": 1.55105318709559,
      "grad_norm": 0.48607152700424194,
      "learning_rate": 4.887590815970235e-06,
      "loss": 0.387,
      "step": 33799
    },
    {
      "epoch": 1.5510990776008444,
      "grad_norm": 0.45279988646507263,
      "learning_rate": 4.887345690169254e-06,
      "loss": 0.3218,
      "step": 33800
    },
    {
      "epoch": 1.5511449681060987,
      "grad_norm": 0.47188666462898254,
      "learning_rate": 4.88710056463917e-06,
      "loss": 0.3581,
      "step": 33801
    },
    {
      "epoch": 1.5511908586113532,
      "grad_norm": 0.5018243193626404,
      "learning_rate": 4.886855439380576e-06,
      "loss": 0.4253,
      "step": 33802
    },
    {
      "epoch": 1.5512367491166077,
      "grad_norm": 0.48465701937675476,
      "learning_rate": 4.8866103143940565e-06,
      "loss": 0.3564,
      "step": 33803
    },
    {
      "epoch": 1.5512826396218622,
      "grad_norm": 0.45508936047554016,
      "learning_rate": 4.886365189680203e-06,
      "loss": 0.3554,
      "step": 33804
    },
    {
      "epoch": 1.5513285301271167,
      "grad_norm": 0.43549874424934387,
      "learning_rate": 4.886120065239606e-06,
      "loss": 0.3096,
      "step": 33805
    },
    {
      "epoch": 1.5513744206323712,
      "grad_norm": 0.44993752241134644,
      "learning_rate": 4.8858749410728556e-06,
      "loss": 0.3267,
      "step": 33806
    },
    {
      "epoch": 1.5514203111376257,
      "grad_norm": 0.44739818572998047,
      "learning_rate": 4.885629817180539e-06,
      "loss": 0.3356,
      "step": 33807
    },
    {
      "epoch": 1.5514662016428802,
      "grad_norm": 0.5050371289253235,
      "learning_rate": 4.8853846935632465e-06,
      "loss": 0.402,
      "step": 33808
    },
    {
      "epoch": 1.5515120921481347,
      "grad_norm": 0.4535248279571533,
      "learning_rate": 4.885139570221568e-06,
      "loss": 0.3169,
      "step": 33809
    },
    {
      "epoch": 1.5515579826533892,
      "grad_norm": 0.44346967339515686,
      "learning_rate": 4.884894447156091e-06,
      "loss": 0.3021,
      "step": 33810
    },
    {
      "epoch": 1.5516038731586435,
      "grad_norm": 0.49705013632774353,
      "learning_rate": 4.884649324367409e-06,
      "loss": 0.3744,
      "step": 33811
    },
    {
      "epoch": 1.551649763663898,
      "grad_norm": 0.5165277719497681,
      "learning_rate": 4.8844042018561086e-06,
      "loss": 0.4018,
      "step": 33812
    },
    {
      "epoch": 1.5516956541691524,
      "grad_norm": 0.4656561315059662,
      "learning_rate": 4.884159079622778e-06,
      "loss": 0.3681,
      "step": 33813
    },
    {
      "epoch": 1.5517415446744067,
      "grad_norm": 0.4630248546600342,
      "learning_rate": 4.883913957668008e-06,
      "loss": 0.3793,
      "step": 33814
    },
    {
      "epoch": 1.5517874351796612,
      "grad_norm": 0.47022202610969543,
      "learning_rate": 4.883668835992391e-06,
      "loss": 0.3635,
      "step": 33815
    },
    {
      "epoch": 1.5518333256849157,
      "grad_norm": 0.4275858402252197,
      "learning_rate": 4.883423714596511e-06,
      "loss": 0.2863,
      "step": 33816
    },
    {
      "epoch": 1.5518792161901702,
      "grad_norm": 0.4666968286037445,
      "learning_rate": 4.883178593480962e-06,
      "loss": 0.3365,
      "step": 33817
    },
    {
      "epoch": 1.5519251066954247,
      "grad_norm": 0.4844528138637543,
      "learning_rate": 4.882933472646332e-06,
      "loss": 0.3906,
      "step": 33818
    },
    {
      "epoch": 1.5519709972006792,
      "grad_norm": 0.4875427782535553,
      "learning_rate": 4.882688352093206e-06,
      "loss": 0.382,
      "step": 33819
    },
    {
      "epoch": 1.5520168877059337,
      "grad_norm": 0.4495276212692261,
      "learning_rate": 4.882443231822181e-06,
      "loss": 0.3041,
      "step": 33820
    },
    {
      "epoch": 1.5520627782111882,
      "grad_norm": 0.48284369707107544,
      "learning_rate": 4.882198111833843e-06,
      "loss": 0.3244,
      "step": 33821
    },
    {
      "epoch": 1.5521086687164427,
      "grad_norm": 0.4668222963809967,
      "learning_rate": 4.88195299212878e-06,
      "loss": 0.3972,
      "step": 33822
    },
    {
      "epoch": 1.552154559221697,
      "grad_norm": 0.4829729199409485,
      "learning_rate": 4.881707872707584e-06,
      "loss": 0.3964,
      "step": 33823
    },
    {
      "epoch": 1.5522004497269515,
      "grad_norm": 0.46594879031181335,
      "learning_rate": 4.881462753570843e-06,
      "loss": 0.3513,
      "step": 33824
    },
    {
      "epoch": 1.552246340232206,
      "grad_norm": 0.46571871638298035,
      "learning_rate": 4.881217634719145e-06,
      "loss": 0.3806,
      "step": 33825
    },
    {
      "epoch": 1.5522922307374603,
      "grad_norm": 0.4769730269908905,
      "learning_rate": 4.880972516153083e-06,
      "loss": 0.3401,
      "step": 33826
    },
    {
      "epoch": 1.5523381212427148,
      "grad_norm": 0.44877541065216064,
      "learning_rate": 4.880727397873244e-06,
      "loss": 0.3344,
      "step": 33827
    },
    {
      "epoch": 1.5523840117479693,
      "grad_norm": 0.4549187123775482,
      "learning_rate": 4.880482279880218e-06,
      "loss": 0.3295,
      "step": 33828
    },
    {
      "epoch": 1.5524299022532237,
      "grad_norm": 0.5007297396659851,
      "learning_rate": 4.880237162174595e-06,
      "loss": 0.413,
      "step": 33829
    },
    {
      "epoch": 1.5524757927584782,
      "grad_norm": 0.545470654964447,
      "learning_rate": 4.879992044756964e-06,
      "loss": 0.4126,
      "step": 33830
    },
    {
      "epoch": 1.5525216832637327,
      "grad_norm": 0.4608706831932068,
      "learning_rate": 4.879746927627911e-06,
      "loss": 0.3789,
      "step": 33831
    },
    {
      "epoch": 1.5525675737689872,
      "grad_norm": 0.5089528560638428,
      "learning_rate": 4.879501810788032e-06,
      "loss": 0.3845,
      "step": 33832
    },
    {
      "epoch": 1.5526134642742417,
      "grad_norm": 0.5185964107513428,
      "learning_rate": 4.879256694237914e-06,
      "loss": 0.3725,
      "step": 33833
    },
    {
      "epoch": 1.5526593547794962,
      "grad_norm": 0.46582528948783875,
      "learning_rate": 4.8790115779781445e-06,
      "loss": 0.3253,
      "step": 33834
    },
    {
      "epoch": 1.5527052452847507,
      "grad_norm": 0.43996596336364746,
      "learning_rate": 4.878766462009311e-06,
      "loss": 0.312,
      "step": 33835
    },
    {
      "epoch": 1.552751135790005,
      "grad_norm": 0.44419464468955994,
      "learning_rate": 4.878521346332009e-06,
      "loss": 0.305,
      "step": 33836
    },
    {
      "epoch": 1.5527970262952595,
      "grad_norm": 0.448348730802536,
      "learning_rate": 4.878276230946824e-06,
      "loss": 0.3398,
      "step": 33837
    },
    {
      "epoch": 1.552842916800514,
      "grad_norm": 0.47825533151626587,
      "learning_rate": 4.878031115854346e-06,
      "loss": 0.4338,
      "step": 33838
    },
    {
      "epoch": 1.5528888073057683,
      "grad_norm": 0.5453977584838867,
      "learning_rate": 4.877786001055165e-06,
      "loss": 0.4315,
      "step": 33839
    },
    {
      "epoch": 1.5529346978110228,
      "grad_norm": 0.47256314754486084,
      "learning_rate": 4.877540886549871e-06,
      "loss": 0.3726,
      "step": 33840
    },
    {
      "epoch": 1.5529805883162773,
      "grad_norm": 0.4944014251232147,
      "learning_rate": 4.87729577233905e-06,
      "loss": 0.3931,
      "step": 33841
    },
    {
      "epoch": 1.5530264788215318,
      "grad_norm": 0.44876396656036377,
      "learning_rate": 4.877050658423296e-06,
      "loss": 0.3089,
      "step": 33842
    },
    {
      "epoch": 1.5530723693267863,
      "grad_norm": 0.47473201155662537,
      "learning_rate": 4.876805544803196e-06,
      "loss": 0.3499,
      "step": 33843
    },
    {
      "epoch": 1.5531182598320408,
      "grad_norm": 0.46753138303756714,
      "learning_rate": 4.8765604314793395e-06,
      "loss": 0.371,
      "step": 33844
    },
    {
      "epoch": 1.5531641503372953,
      "grad_norm": 0.46229952573776245,
      "learning_rate": 4.8763153184523165e-06,
      "loss": 0.3837,
      "step": 33845
    },
    {
      "epoch": 1.5532100408425498,
      "grad_norm": 0.4687906801700592,
      "learning_rate": 4.8760702057227165e-06,
      "loss": 0.3846,
      "step": 33846
    },
    {
      "epoch": 1.5532559313478043,
      "grad_norm": 0.49077412486076355,
      "learning_rate": 4.875825093291126e-06,
      "loss": 0.4523,
      "step": 33847
    },
    {
      "epoch": 1.5533018218530588,
      "grad_norm": 0.4721636474132538,
      "learning_rate": 4.87557998115814e-06,
      "loss": 0.3509,
      "step": 33848
    },
    {
      "epoch": 1.553347712358313,
      "grad_norm": 0.4530836343765259,
      "learning_rate": 4.875334869324344e-06,
      "loss": 0.3197,
      "step": 33849
    },
    {
      "epoch": 1.5533936028635675,
      "grad_norm": 0.4735397696495056,
      "learning_rate": 4.875089757790328e-06,
      "loss": 0.3672,
      "step": 33850
    },
    {
      "epoch": 1.553439493368822,
      "grad_norm": 0.45790648460388184,
      "learning_rate": 4.874844646556682e-06,
      "loss": 0.3592,
      "step": 33851
    },
    {
      "epoch": 1.5534853838740763,
      "grad_norm": 0.46510395407676697,
      "learning_rate": 4.874599535623995e-06,
      "loss": 0.3506,
      "step": 33852
    },
    {
      "epoch": 1.5535312743793308,
      "grad_norm": 0.4455546438694,
      "learning_rate": 4.874354424992855e-06,
      "loss": 0.2868,
      "step": 33853
    },
    {
      "epoch": 1.5535771648845853,
      "grad_norm": 0.4468923509120941,
      "learning_rate": 4.8741093146638555e-06,
      "loss": 0.304,
      "step": 33854
    },
    {
      "epoch": 1.5536230553898398,
      "grad_norm": 0.4555032551288605,
      "learning_rate": 4.873864204637583e-06,
      "loss": 0.3476,
      "step": 33855
    },
    {
      "epoch": 1.5536689458950943,
      "grad_norm": 0.509987473487854,
      "learning_rate": 4.873619094914626e-06,
      "loss": 0.4246,
      "step": 33856
    },
    {
      "epoch": 1.5537148364003488,
      "grad_norm": 0.5118496417999268,
      "learning_rate": 4.873373985495576e-06,
      "loss": 0.416,
      "step": 33857
    },
    {
      "epoch": 1.5537607269056033,
      "grad_norm": 0.5022135972976685,
      "learning_rate": 4.873128876381022e-06,
      "loss": 0.3652,
      "step": 33858
    },
    {
      "epoch": 1.5538066174108578,
      "grad_norm": 0.5145373344421387,
      "learning_rate": 4.872883767571551e-06,
      "loss": 0.3888,
      "step": 33859
    },
    {
      "epoch": 1.5538525079161123,
      "grad_norm": 0.4669910967350006,
      "learning_rate": 4.872638659067756e-06,
      "loss": 0.3428,
      "step": 33860
    },
    {
      "epoch": 1.5538983984213666,
      "grad_norm": 0.49525442719459534,
      "learning_rate": 4.872393550870226e-06,
      "loss": 0.3538,
      "step": 33861
    },
    {
      "epoch": 1.553944288926621,
      "grad_norm": 0.49172645807266235,
      "learning_rate": 4.872148442979546e-06,
      "loss": 0.3442,
      "step": 33862
    },
    {
      "epoch": 1.5539901794318756,
      "grad_norm": 0.45497870445251465,
      "learning_rate": 4.8719033353963104e-06,
      "loss": 0.3487,
      "step": 33863
    },
    {
      "epoch": 1.55403606993713,
      "grad_norm": 0.4599754214286804,
      "learning_rate": 4.871658228121108e-06,
      "loss": 0.3309,
      "step": 33864
    },
    {
      "epoch": 1.5540819604423843,
      "grad_norm": 0.4922204315662384,
      "learning_rate": 4.871413121154527e-06,
      "loss": 0.3742,
      "step": 33865
    },
    {
      "epoch": 1.5541278509476388,
      "grad_norm": 0.4618149995803833,
      "learning_rate": 4.871168014497154e-06,
      "loss": 0.3987,
      "step": 33866
    },
    {
      "epoch": 1.5541737414528933,
      "grad_norm": 0.47546014189720154,
      "learning_rate": 4.870922908149584e-06,
      "loss": 0.3534,
      "step": 33867
    },
    {
      "epoch": 1.5542196319581478,
      "grad_norm": 0.5193238854408264,
      "learning_rate": 4.870677802112404e-06,
      "loss": 0.368,
      "step": 33868
    },
    {
      "epoch": 1.5542655224634023,
      "grad_norm": 0.47295698523521423,
      "learning_rate": 4.8704326963862e-06,
      "loss": 0.3828,
      "step": 33869
    },
    {
      "epoch": 1.5543114129686568,
      "grad_norm": 0.4816264510154724,
      "learning_rate": 4.870187590971567e-06,
      "loss": 0.338,
      "step": 33870
    },
    {
      "epoch": 1.5543573034739113,
      "grad_norm": 0.49063390493392944,
      "learning_rate": 4.869942485869091e-06,
      "loss": 0.4119,
      "step": 33871
    },
    {
      "epoch": 1.5544031939791658,
      "grad_norm": 0.4739210307598114,
      "learning_rate": 4.869697381079363e-06,
      "loss": 0.3441,
      "step": 33872
    },
    {
      "epoch": 1.5544490844844203,
      "grad_norm": 0.43170270323753357,
      "learning_rate": 4.869452276602971e-06,
      "loss": 0.2702,
      "step": 33873
    },
    {
      "epoch": 1.5544949749896746,
      "grad_norm": 0.5101765990257263,
      "learning_rate": 4.869207172440506e-06,
      "loss": 0.4116,
      "step": 33874
    },
    {
      "epoch": 1.554540865494929,
      "grad_norm": 0.46140211820602417,
      "learning_rate": 4.868962068592554e-06,
      "loss": 0.3369,
      "step": 33875
    },
    {
      "epoch": 1.5545867560001836,
      "grad_norm": 0.47751447558403015,
      "learning_rate": 4.868716965059709e-06,
      "loss": 0.3614,
      "step": 33876
    },
    {
      "epoch": 1.5546326465054379,
      "grad_norm": 0.556000292301178,
      "learning_rate": 4.868471861842559e-06,
      "loss": 0.3779,
      "step": 33877
    },
    {
      "epoch": 1.5546785370106924,
      "grad_norm": 0.4481445550918579,
      "learning_rate": 4.868226758941691e-06,
      "loss": 0.36,
      "step": 33878
    },
    {
      "epoch": 1.5547244275159469,
      "grad_norm": 0.4806668162345886,
      "learning_rate": 4.867981656357696e-06,
      "loss": 0.3993,
      "step": 33879
    },
    {
      "epoch": 1.5547703180212014,
      "grad_norm": 0.4933127164840698,
      "learning_rate": 4.867736554091165e-06,
      "loss": 0.437,
      "step": 33880
    },
    {
      "epoch": 1.5548162085264559,
      "grad_norm": 0.514406681060791,
      "learning_rate": 4.867491452142684e-06,
      "loss": 0.4964,
      "step": 33881
    },
    {
      "epoch": 1.5548620990317104,
      "grad_norm": 0.4856951832771301,
      "learning_rate": 4.867246350512846e-06,
      "loss": 0.3641,
      "step": 33882
    },
    {
      "epoch": 1.5549079895369649,
      "grad_norm": 0.4681156873703003,
      "learning_rate": 4.867001249202238e-06,
      "loss": 0.3538,
      "step": 33883
    },
    {
      "epoch": 1.5549538800422193,
      "grad_norm": 0.49690771102905273,
      "learning_rate": 4.8667561482114485e-06,
      "loss": 0.3906,
      "step": 33884
    },
    {
      "epoch": 1.5549997705474738,
      "grad_norm": 0.4789225459098816,
      "learning_rate": 4.86651104754107e-06,
      "loss": 0.3344,
      "step": 33885
    },
    {
      "epoch": 1.5550456610527283,
      "grad_norm": 0.44636818766593933,
      "learning_rate": 4.86626594719169e-06,
      "loss": 0.3151,
      "step": 33886
    },
    {
      "epoch": 1.5550915515579826,
      "grad_norm": 0.47394487261772156,
      "learning_rate": 4.866020847163898e-06,
      "loss": 0.336,
      "step": 33887
    },
    {
      "epoch": 1.5551374420632371,
      "grad_norm": 0.4233240485191345,
      "learning_rate": 4.865775747458284e-06,
      "loss": 0.3012,
      "step": 33888
    },
    {
      "epoch": 1.5551833325684916,
      "grad_norm": 0.4341496229171753,
      "learning_rate": 4.865530648075437e-06,
      "loss": 0.3237,
      "step": 33889
    },
    {
      "epoch": 1.555229223073746,
      "grad_norm": 0.46069103479385376,
      "learning_rate": 4.865285549015944e-06,
      "loss": 0.3352,
      "step": 33890
    },
    {
      "epoch": 1.5552751135790004,
      "grad_norm": 0.46911177039146423,
      "learning_rate": 4.8650404502804e-06,
      "loss": 0.365,
      "step": 33891
    },
    {
      "epoch": 1.5553210040842549,
      "grad_norm": 0.4479244649410248,
      "learning_rate": 4.8647953518693906e-06,
      "loss": 0.3249,
      "step": 33892
    },
    {
      "epoch": 1.5553668945895094,
      "grad_norm": 0.4458649158477783,
      "learning_rate": 4.8645502537835045e-06,
      "loss": 0.2635,
      "step": 33893
    },
    {
      "epoch": 1.5554127850947639,
      "grad_norm": 0.4572649300098419,
      "learning_rate": 4.864305156023333e-06,
      "loss": 0.3201,
      "step": 33894
    },
    {
      "epoch": 1.5554586756000184,
      "grad_norm": 0.552188515663147,
      "learning_rate": 4.864060058589465e-06,
      "loss": 0.4649,
      "step": 33895
    },
    {
      "epoch": 1.5555045661052729,
      "grad_norm": 0.4331375062465668,
      "learning_rate": 4.863814961482487e-06,
      "loss": 0.2927,
      "step": 33896
    },
    {
      "epoch": 1.5555504566105274,
      "grad_norm": 0.507425844669342,
      "learning_rate": 4.863569864702994e-06,
      "loss": 0.4455,
      "step": 33897
    },
    {
      "epoch": 1.5555963471157819,
      "grad_norm": 0.4967997968196869,
      "learning_rate": 4.863324768251573e-06,
      "loss": 0.4153,
      "step": 33898
    },
    {
      "epoch": 1.5556422376210364,
      "grad_norm": 0.4184669256210327,
      "learning_rate": 4.8630796721288126e-06,
      "loss": 0.2793,
      "step": 33899
    },
    {
      "epoch": 1.5556881281262906,
      "grad_norm": 0.48595064878463745,
      "learning_rate": 4.8628345763353e-06,
      "loss": 0.3584,
      "step": 33900
    },
    {
      "epoch": 1.5557340186315451,
      "grad_norm": 0.4529782235622406,
      "learning_rate": 4.862589480871628e-06,
      "loss": 0.3472,
      "step": 33901
    },
    {
      "epoch": 1.5557799091367996,
      "grad_norm": 0.6260877251625061,
      "learning_rate": 4.862344385738386e-06,
      "loss": 0.3535,
      "step": 33902
    },
    {
      "epoch": 1.555825799642054,
      "grad_norm": 0.5443751215934753,
      "learning_rate": 4.8620992909361605e-06,
      "loss": 0.3639,
      "step": 33903
    },
    {
      "epoch": 1.5558716901473084,
      "grad_norm": 0.4476192891597748,
      "learning_rate": 4.861854196465545e-06,
      "loss": 0.3083,
      "step": 33904
    },
    {
      "epoch": 1.555917580652563,
      "grad_norm": 0.5101179480552673,
      "learning_rate": 4.861609102327125e-06,
      "loss": 0.3973,
      "step": 33905
    },
    {
      "epoch": 1.5559634711578174,
      "grad_norm": 0.456595242023468,
      "learning_rate": 4.86136400852149e-06,
      "loss": 0.3082,
      "step": 33906
    },
    {
      "epoch": 1.556009361663072,
      "grad_norm": 0.47844791412353516,
      "learning_rate": 4.861118915049233e-06,
      "loss": 0.3799,
      "step": 33907
    },
    {
      "epoch": 1.5560552521683264,
      "grad_norm": 0.5073503255844116,
      "learning_rate": 4.860873821910942e-06,
      "loss": 0.4056,
      "step": 33908
    },
    {
      "epoch": 1.556101142673581,
      "grad_norm": 0.47735118865966797,
      "learning_rate": 4.860628729107203e-06,
      "loss": 0.3456,
      "step": 33909
    },
    {
      "epoch": 1.5561470331788354,
      "grad_norm": 0.47785258293151855,
      "learning_rate": 4.860383636638609e-06,
      "loss": 0.3569,
      "step": 33910
    },
    {
      "epoch": 1.55619292368409,
      "grad_norm": 0.4961502254009247,
      "learning_rate": 4.860138544505749e-06,
      "loss": 0.3668,
      "step": 33911
    },
    {
      "epoch": 1.5562388141893442,
      "grad_norm": 0.5004842877388,
      "learning_rate": 4.85989345270921e-06,
      "loss": 0.4564,
      "step": 33912
    },
    {
      "epoch": 1.5562847046945987,
      "grad_norm": 0.458984911441803,
      "learning_rate": 4.859648361249585e-06,
      "loss": 0.3555,
      "step": 33913
    },
    {
      "epoch": 1.5563305951998532,
      "grad_norm": 0.49291127920150757,
      "learning_rate": 4.8594032701274605e-06,
      "loss": 0.4534,
      "step": 33914
    },
    {
      "epoch": 1.5563764857051074,
      "grad_norm": 0.4456261098384857,
      "learning_rate": 4.859158179343425e-06,
      "loss": 0.3136,
      "step": 33915
    },
    {
      "epoch": 1.556422376210362,
      "grad_norm": 0.49355629086494446,
      "learning_rate": 4.858913088898072e-06,
      "loss": 0.3645,
      "step": 33916
    },
    {
      "epoch": 1.5564682667156164,
      "grad_norm": 0.45331108570098877,
      "learning_rate": 4.8586679987919884e-06,
      "loss": 0.358,
      "step": 33917
    },
    {
      "epoch": 1.556514157220871,
      "grad_norm": 0.48834094405174255,
      "learning_rate": 4.858422909025762e-06,
      "loss": 0.3444,
      "step": 33918
    },
    {
      "epoch": 1.5565600477261254,
      "grad_norm": 0.49738866090774536,
      "learning_rate": 4.858177819599985e-06,
      "loss": 0.385,
      "step": 33919
    },
    {
      "epoch": 1.55660593823138,
      "grad_norm": 0.46537724137306213,
      "learning_rate": 4.857932730515246e-06,
      "loss": 0.3607,
      "step": 33920
    },
    {
      "epoch": 1.5566518287366344,
      "grad_norm": 0.4307282269001007,
      "learning_rate": 4.857687641772133e-06,
      "loss": 0.2924,
      "step": 33921
    },
    {
      "epoch": 1.556697719241889,
      "grad_norm": 0.487947553396225,
      "learning_rate": 4.8574425533712376e-06,
      "loss": 0.3758,
      "step": 33922
    },
    {
      "epoch": 1.5567436097471434,
      "grad_norm": 0.4825795292854309,
      "learning_rate": 4.857197465313147e-06,
      "loss": 0.4028,
      "step": 33923
    },
    {
      "epoch": 1.556789500252398,
      "grad_norm": 0.47242680191993713,
      "learning_rate": 4.856952377598451e-06,
      "loss": 0.3635,
      "step": 33924
    },
    {
      "epoch": 1.5568353907576522,
      "grad_norm": 0.4348750114440918,
      "learning_rate": 4.8567072902277405e-06,
      "loss": 0.3598,
      "step": 33925
    },
    {
      "epoch": 1.5568812812629067,
      "grad_norm": 0.44803258776664734,
      "learning_rate": 4.856462203201603e-06,
      "loss": 0.3264,
      "step": 33926
    },
    {
      "epoch": 1.5569271717681612,
      "grad_norm": 0.41285648941993713,
      "learning_rate": 4.856217116520627e-06,
      "loss": 0.2523,
      "step": 33927
    },
    {
      "epoch": 1.5569730622734155,
      "grad_norm": 0.4857197701931,
      "learning_rate": 4.8559720301854055e-06,
      "loss": 0.3668,
      "step": 33928
    },
    {
      "epoch": 1.55701895277867,
      "grad_norm": 0.44854727387428284,
      "learning_rate": 4.855726944196526e-06,
      "loss": 0.347,
      "step": 33929
    },
    {
      "epoch": 1.5570648432839245,
      "grad_norm": 0.4953623414039612,
      "learning_rate": 4.855481858554578e-06,
      "loss": 0.3463,
      "step": 33930
    },
    {
      "epoch": 1.557110733789179,
      "grad_norm": 0.46095573902130127,
      "learning_rate": 4.855236773260149e-06,
      "loss": 0.3247,
      "step": 33931
    },
    {
      "epoch": 1.5571566242944335,
      "grad_norm": 0.45630112290382385,
      "learning_rate": 4.854991688313831e-06,
      "loss": 0.3528,
      "step": 33932
    },
    {
      "epoch": 1.557202514799688,
      "grad_norm": 0.48347222805023193,
      "learning_rate": 4.854746603716212e-06,
      "loss": 0.3713,
      "step": 33933
    },
    {
      "epoch": 1.5572484053049425,
      "grad_norm": 0.4526872932910919,
      "learning_rate": 4.85450151946788e-06,
      "loss": 0.2854,
      "step": 33934
    },
    {
      "epoch": 1.557294295810197,
      "grad_norm": 0.44880005717277527,
      "learning_rate": 4.854256435569428e-06,
      "loss": 0.3217,
      "step": 33935
    },
    {
      "epoch": 1.5573401863154515,
      "grad_norm": 0.4592137336730957,
      "learning_rate": 4.854011352021443e-06,
      "loss": 0.3149,
      "step": 33936
    },
    {
      "epoch": 1.557386076820706,
      "grad_norm": 0.4712860584259033,
      "learning_rate": 4.853766268824513e-06,
      "loss": 0.3554,
      "step": 33937
    },
    {
      "epoch": 1.5574319673259602,
      "grad_norm": 0.47146177291870117,
      "learning_rate": 4.8535211859792305e-06,
      "loss": 0.355,
      "step": 33938
    },
    {
      "epoch": 1.5574778578312147,
      "grad_norm": 0.44241660833358765,
      "learning_rate": 4.853276103486183e-06,
      "loss": 0.347,
      "step": 33939
    },
    {
      "epoch": 1.5575237483364692,
      "grad_norm": 0.4568447172641754,
      "learning_rate": 4.8530310213459585e-06,
      "loss": 0.3133,
      "step": 33940
    },
    {
      "epoch": 1.5575696388417235,
      "grad_norm": 0.42443621158599854,
      "learning_rate": 4.8527859395591504e-06,
      "loss": 0.2782,
      "step": 33941
    },
    {
      "epoch": 1.557615529346978,
      "grad_norm": 0.515719473361969,
      "learning_rate": 4.852540858126345e-06,
      "loss": 0.4186,
      "step": 33942
    },
    {
      "epoch": 1.5576614198522325,
      "grad_norm": 0.48515060544013977,
      "learning_rate": 4.852295777048131e-06,
      "loss": 0.3629,
      "step": 33943
    },
    {
      "epoch": 1.557707310357487,
      "grad_norm": 0.4753115773200989,
      "learning_rate": 4.852050696325101e-06,
      "loss": 0.3363,
      "step": 33944
    },
    {
      "epoch": 1.5577532008627415,
      "grad_norm": 0.49088215827941895,
      "learning_rate": 4.851805615957842e-06,
      "loss": 0.3581,
      "step": 33945
    },
    {
      "epoch": 1.557799091367996,
      "grad_norm": 0.46973127126693726,
      "learning_rate": 4.851560535946943e-06,
      "loss": 0.3565,
      "step": 33946
    },
    {
      "epoch": 1.5578449818732505,
      "grad_norm": 0.4709365963935852,
      "learning_rate": 4.851315456292995e-06,
      "loss": 0.3577,
      "step": 33947
    },
    {
      "epoch": 1.557890872378505,
      "grad_norm": 0.4910786747932434,
      "learning_rate": 4.851070376996586e-06,
      "loss": 0.3675,
      "step": 33948
    },
    {
      "epoch": 1.5579367628837595,
      "grad_norm": 0.47230836749076843,
      "learning_rate": 4.850825298058304e-06,
      "loss": 0.3772,
      "step": 33949
    },
    {
      "epoch": 1.5579826533890138,
      "grad_norm": 0.4961956739425659,
      "learning_rate": 4.850580219478742e-06,
      "loss": 0.4197,
      "step": 33950
    },
    {
      "epoch": 1.5580285438942683,
      "grad_norm": 0.5157335996627808,
      "learning_rate": 4.850335141258488e-06,
      "loss": 0.4177,
      "step": 33951
    },
    {
      "epoch": 1.5580744343995228,
      "grad_norm": 0.4636906385421753,
      "learning_rate": 4.850090063398129e-06,
      "loss": 0.3721,
      "step": 33952
    },
    {
      "epoch": 1.5581203249047773,
      "grad_norm": 0.4237014949321747,
      "learning_rate": 4.849844985898257e-06,
      "loss": 0.2916,
      "step": 33953
    },
    {
      "epoch": 1.5581662154100315,
      "grad_norm": 0.45102742314338684,
      "learning_rate": 4.84959990875946e-06,
      "loss": 0.3904,
      "step": 33954
    },
    {
      "epoch": 1.558212105915286,
      "grad_norm": 0.4883011281490326,
      "learning_rate": 4.849354831982327e-06,
      "loss": 0.3798,
      "step": 33955
    },
    {
      "epoch": 1.5582579964205405,
      "grad_norm": 0.4765281081199646,
      "learning_rate": 4.849109755567449e-06,
      "loss": 0.3675,
      "step": 33956
    },
    {
      "epoch": 1.558303886925795,
      "grad_norm": 0.47588610649108887,
      "learning_rate": 4.848864679515415e-06,
      "loss": 0.3783,
      "step": 33957
    },
    {
      "epoch": 1.5583497774310495,
      "grad_norm": 0.475922167301178,
      "learning_rate": 4.848619603826813e-06,
      "loss": 0.3743,
      "step": 33958
    },
    {
      "epoch": 1.558395667936304,
      "grad_norm": 0.4717026948928833,
      "learning_rate": 4.848374528502234e-06,
      "loss": 0.3588,
      "step": 33959
    },
    {
      "epoch": 1.5584415584415585,
      "grad_norm": 0.4554486572742462,
      "learning_rate": 4.848129453542266e-06,
      "loss": 0.3006,
      "step": 33960
    },
    {
      "epoch": 1.558487448946813,
      "grad_norm": 0.474334180355072,
      "learning_rate": 4.847884378947497e-06,
      "loss": 0.3537,
      "step": 33961
    },
    {
      "epoch": 1.5585333394520675,
      "grad_norm": 0.5182164907455444,
      "learning_rate": 4.84763930471852e-06,
      "loss": 0.429,
      "step": 33962
    },
    {
      "epoch": 1.5585792299573218,
      "grad_norm": 0.5130590796470642,
      "learning_rate": 4.847394230855922e-06,
      "loss": 0.426,
      "step": 33963
    },
    {
      "epoch": 1.5586251204625763,
      "grad_norm": 0.4523591995239258,
      "learning_rate": 4.847149157360293e-06,
      "loss": 0.3373,
      "step": 33964
    },
    {
      "epoch": 1.5586710109678308,
      "grad_norm": 0.49015969038009644,
      "learning_rate": 4.846904084232221e-06,
      "loss": 0.3784,
      "step": 33965
    },
    {
      "epoch": 1.558716901473085,
      "grad_norm": 0.4569879174232483,
      "learning_rate": 4.846659011472297e-06,
      "loss": 0.3484,
      "step": 33966
    },
    {
      "epoch": 1.5587627919783396,
      "grad_norm": 0.5073875784873962,
      "learning_rate": 4.846413939081111e-06,
      "loss": 0.4182,
      "step": 33967
    },
    {
      "epoch": 1.558808682483594,
      "grad_norm": 0.4717338979244232,
      "learning_rate": 4.846168867059249e-06,
      "loss": 0.3355,
      "step": 33968
    },
    {
      "epoch": 1.5588545729888486,
      "grad_norm": 0.4369087815284729,
      "learning_rate": 4.845923795407303e-06,
      "loss": 0.3232,
      "step": 33969
    },
    {
      "epoch": 1.558900463494103,
      "grad_norm": 0.4897747039794922,
      "learning_rate": 4.845678724125863e-06,
      "loss": 0.366,
      "step": 33970
    },
    {
      "epoch": 1.5589463539993575,
      "grad_norm": 0.43615251779556274,
      "learning_rate": 4.845433653215513e-06,
      "loss": 0.2948,
      "step": 33971
    },
    {
      "epoch": 1.558992244504612,
      "grad_norm": 0.4515037536621094,
      "learning_rate": 4.84518858267685e-06,
      "loss": 0.3914,
      "step": 33972
    },
    {
      "epoch": 1.5590381350098665,
      "grad_norm": 0.48962491750717163,
      "learning_rate": 4.844943512510458e-06,
      "loss": 0.4044,
      "step": 33973
    },
    {
      "epoch": 1.559084025515121,
      "grad_norm": 0.4202832579612732,
      "learning_rate": 4.844698442716928e-06,
      "loss": 0.3008,
      "step": 33974
    },
    {
      "epoch": 1.5591299160203755,
      "grad_norm": 0.4623993933200836,
      "learning_rate": 4.8444533732968504e-06,
      "loss": 0.3411,
      "step": 33975
    },
    {
      "epoch": 1.5591758065256298,
      "grad_norm": 0.48730501532554626,
      "learning_rate": 4.844208304250813e-06,
      "loss": 0.362,
      "step": 33976
    },
    {
      "epoch": 1.5592216970308843,
      "grad_norm": 0.4625155031681061,
      "learning_rate": 4.843963235579403e-06,
      "loss": 0.3407,
      "step": 33977
    },
    {
      "epoch": 1.5592675875361388,
      "grad_norm": 0.4783336818218231,
      "learning_rate": 4.843718167283215e-06,
      "loss": 0.3721,
      "step": 33978
    },
    {
      "epoch": 1.559313478041393,
      "grad_norm": 0.4403071403503418,
      "learning_rate": 4.843473099362835e-06,
      "loss": 0.2825,
      "step": 33979
    },
    {
      "epoch": 1.5593593685466476,
      "grad_norm": 0.45905882120132446,
      "learning_rate": 4.8432280318188515e-06,
      "loss": 0.3395,
      "step": 33980
    },
    {
      "epoch": 1.559405259051902,
      "grad_norm": 0.5128629207611084,
      "learning_rate": 4.842982964651858e-06,
      "loss": 0.4177,
      "step": 33981
    },
    {
      "epoch": 1.5594511495571566,
      "grad_norm": 0.49125218391418457,
      "learning_rate": 4.842737897862439e-06,
      "loss": 0.3845,
      "step": 33982
    },
    {
      "epoch": 1.559497040062411,
      "grad_norm": 0.5071790218353271,
      "learning_rate": 4.842492831451184e-06,
      "loss": 0.4003,
      "step": 33983
    },
    {
      "epoch": 1.5595429305676656,
      "grad_norm": 0.48845842480659485,
      "learning_rate": 4.842247765418686e-06,
      "loss": 0.3466,
      "step": 33984
    },
    {
      "epoch": 1.55958882107292,
      "grad_norm": 0.4492848813533783,
      "learning_rate": 4.842002699765532e-06,
      "loss": 0.3156,
      "step": 33985
    },
    {
      "epoch": 1.5596347115781746,
      "grad_norm": 0.4443824887275696,
      "learning_rate": 4.841757634492312e-06,
      "loss": 0.3005,
      "step": 33986
    },
    {
      "epoch": 1.559680602083429,
      "grad_norm": 0.4452267587184906,
      "learning_rate": 4.841512569599615e-06,
      "loss": 0.3316,
      "step": 33987
    },
    {
      "epoch": 1.5597264925886836,
      "grad_norm": 0.49753159284591675,
      "learning_rate": 4.84126750508803e-06,
      "loss": 0.4079,
      "step": 33988
    },
    {
      "epoch": 1.5597723830939378,
      "grad_norm": 0.46810078620910645,
      "learning_rate": 4.841022440958145e-06,
      "loss": 0.3619,
      "step": 33989
    },
    {
      "epoch": 1.5598182735991923,
      "grad_norm": 0.4785688519477844,
      "learning_rate": 4.840777377210554e-06,
      "loss": 0.3596,
      "step": 33990
    },
    {
      "epoch": 1.5598641641044468,
      "grad_norm": 0.5039102435112,
      "learning_rate": 4.8405323138458415e-06,
      "loss": 0.3872,
      "step": 33991
    },
    {
      "epoch": 1.5599100546097011,
      "grad_norm": 0.469425767660141,
      "learning_rate": 4.840287250864597e-06,
      "loss": 0.3478,
      "step": 33992
    },
    {
      "epoch": 1.5599559451149556,
      "grad_norm": 0.4289983808994293,
      "learning_rate": 4.840042188267413e-06,
      "loss": 0.287,
      "step": 33993
    },
    {
      "epoch": 1.56000183562021,
      "grad_norm": 0.4520646929740906,
      "learning_rate": 4.839797126054877e-06,
      "loss": 0.3216,
      "step": 33994
    },
    {
      "epoch": 1.5600477261254646,
      "grad_norm": 0.4336785674095154,
      "learning_rate": 4.839552064227579e-06,
      "loss": 0.3121,
      "step": 33995
    },
    {
      "epoch": 1.560093616630719,
      "grad_norm": 0.4593063294887543,
      "learning_rate": 4.839307002786106e-06,
      "loss": 0.3403,
      "step": 33996
    },
    {
      "epoch": 1.5601395071359736,
      "grad_norm": 0.5014958381652832,
      "learning_rate": 4.83906194173105e-06,
      "loss": 0.3729,
      "step": 33997
    },
    {
      "epoch": 1.560185397641228,
      "grad_norm": 0.48236924409866333,
      "learning_rate": 4.838816881063e-06,
      "loss": 0.4369,
      "step": 33998
    },
    {
      "epoch": 1.5602312881464826,
      "grad_norm": 0.44015175104141235,
      "learning_rate": 4.838571820782542e-06,
      "loss": 0.3018,
      "step": 33999
    },
    {
      "epoch": 1.560277178651737,
      "grad_norm": 0.4763874411582947,
      "learning_rate": 4.838326760890269e-06,
      "loss": 0.3874,
      "step": 34000
    },
    {
      "epoch": 1.5603230691569914,
      "grad_norm": 0.47120100259780884,
      "learning_rate": 4.83808170138677e-06,
      "loss": 0.3173,
      "step": 34001
    },
    {
      "epoch": 1.5603689596622459,
      "grad_norm": 0.4392794668674469,
      "learning_rate": 4.837836642272632e-06,
      "loss": 0.2955,
      "step": 34002
    },
    {
      "epoch": 1.5604148501675004,
      "grad_norm": 0.4662547707557678,
      "learning_rate": 4.837591583548447e-06,
      "loss": 0.3381,
      "step": 34003
    },
    {
      "epoch": 1.5604607406727546,
      "grad_norm": 0.48285943269729614,
      "learning_rate": 4.837346525214804e-06,
      "loss": 0.3631,
      "step": 34004
    },
    {
      "epoch": 1.5605066311780091,
      "grad_norm": 0.46197715401649475,
      "learning_rate": 4.837101467272287e-06,
      "loss": 0.3778,
      "step": 34005
    },
    {
      "epoch": 1.5605525216832636,
      "grad_norm": 0.4374086558818817,
      "learning_rate": 4.836856409721493e-06,
      "loss": 0.3063,
      "step": 34006
    },
    {
      "epoch": 1.5605984121885181,
      "grad_norm": 0.4273760914802551,
      "learning_rate": 4.836611352563008e-06,
      "loss": 0.2891,
      "step": 34007
    },
    {
      "epoch": 1.5606443026937726,
      "grad_norm": 0.45668232440948486,
      "learning_rate": 4.836366295797419e-06,
      "loss": 0.3816,
      "step": 34008
    },
    {
      "epoch": 1.5606901931990271,
      "grad_norm": 0.4518336057662964,
      "learning_rate": 4.83612123942532e-06,
      "loss": 0.3523,
      "step": 34009
    },
    {
      "epoch": 1.5607360837042816,
      "grad_norm": 0.48512694239616394,
      "learning_rate": 4.835876183447296e-06,
      "loss": 0.395,
      "step": 34010
    },
    {
      "epoch": 1.5607819742095361,
      "grad_norm": 0.43445539474487305,
      "learning_rate": 4.835631127863937e-06,
      "loss": 0.3218,
      "step": 34011
    },
    {
      "epoch": 1.5608278647147906,
      "grad_norm": 0.4571727216243744,
      "learning_rate": 4.835386072675836e-06,
      "loss": 0.3799,
      "step": 34012
    },
    {
      "epoch": 1.5608737552200451,
      "grad_norm": 0.4810318350791931,
      "learning_rate": 4.835141017883579e-06,
      "loss": 0.3923,
      "step": 34013
    },
    {
      "epoch": 1.5609196457252994,
      "grad_norm": 0.4938943386077881,
      "learning_rate": 4.834895963487754e-06,
      "loss": 0.3989,
      "step": 34014
    },
    {
      "epoch": 1.560965536230554,
      "grad_norm": 0.4688873291015625,
      "learning_rate": 4.8346509094889525e-06,
      "loss": 0.3548,
      "step": 34015
    },
    {
      "epoch": 1.5610114267358084,
      "grad_norm": 0.498010516166687,
      "learning_rate": 4.834405855887765e-06,
      "loss": 0.3907,
      "step": 34016
    },
    {
      "epoch": 1.5610573172410627,
      "grad_norm": 0.44251519441604614,
      "learning_rate": 4.834160802684777e-06,
      "loss": 0.3361,
      "step": 34017
    },
    {
      "epoch": 1.5611032077463172,
      "grad_norm": 0.5103425979614258,
      "learning_rate": 4.833915749880582e-06,
      "loss": 0.4208,
      "step": 34018
    },
    {
      "epoch": 1.5611490982515717,
      "grad_norm": 0.45707976818084717,
      "learning_rate": 4.833670697475767e-06,
      "loss": 0.3433,
      "step": 34019
    },
    {
      "epoch": 1.5611949887568262,
      "grad_norm": 0.4793717861175537,
      "learning_rate": 4.833425645470919e-06,
      "loss": 0.3376,
      "step": 34020
    },
    {
      "epoch": 1.5612408792620807,
      "grad_norm": 0.4782590866088867,
      "learning_rate": 4.8331805938666316e-06,
      "loss": 0.3822,
      "step": 34021
    },
    {
      "epoch": 1.5612867697673352,
      "grad_norm": 0.42645901441574097,
      "learning_rate": 4.832935542663493e-06,
      "loss": 0.2685,
      "step": 34022
    },
    {
      "epoch": 1.5613326602725897,
      "grad_norm": 0.5145801305770874,
      "learning_rate": 4.83269049186209e-06,
      "loss": 0.3422,
      "step": 34023
    },
    {
      "epoch": 1.5613785507778442,
      "grad_norm": 0.47988617420196533,
      "learning_rate": 4.832445441463014e-06,
      "loss": 0.3738,
      "step": 34024
    },
    {
      "epoch": 1.5614244412830987,
      "grad_norm": 0.4541892111301422,
      "learning_rate": 4.832200391466855e-06,
      "loss": 0.3607,
      "step": 34025
    },
    {
      "epoch": 1.5614703317883531,
      "grad_norm": 0.48689472675323486,
      "learning_rate": 4.831955341874199e-06,
      "loss": 0.375,
      "step": 34026
    },
    {
      "epoch": 1.5615162222936074,
      "grad_norm": 0.5040265321731567,
      "learning_rate": 4.831710292685639e-06,
      "loss": 0.422,
      "step": 34027
    },
    {
      "epoch": 1.561562112798862,
      "grad_norm": 0.4834090769290924,
      "learning_rate": 4.831465243901763e-06,
      "loss": 0.3658,
      "step": 34028
    },
    {
      "epoch": 1.5616080033041164,
      "grad_norm": 0.4912855923175812,
      "learning_rate": 4.83122019552316e-06,
      "loss": 0.3417,
      "step": 34029
    },
    {
      "epoch": 1.5616538938093707,
      "grad_norm": 0.45810794830322266,
      "learning_rate": 4.830975147550417e-06,
      "loss": 0.3398,
      "step": 34030
    },
    {
      "epoch": 1.5616997843146252,
      "grad_norm": 0.47721803188323975,
      "learning_rate": 4.830730099984128e-06,
      "loss": 0.3506,
      "step": 34031
    },
    {
      "epoch": 1.5617456748198797,
      "grad_norm": 0.48147934675216675,
      "learning_rate": 4.830485052824879e-06,
      "loss": 0.4135,
      "step": 34032
    },
    {
      "epoch": 1.5617915653251342,
      "grad_norm": 0.45746269822120667,
      "learning_rate": 4.830240006073258e-06,
      "loss": 0.3532,
      "step": 34033
    },
    {
      "epoch": 1.5618374558303887,
      "grad_norm": 0.47519248723983765,
      "learning_rate": 4.829994959729858e-06,
      "loss": 0.3568,
      "step": 34034
    },
    {
      "epoch": 1.5618833463356432,
      "grad_norm": 0.4243800640106201,
      "learning_rate": 4.829749913795267e-06,
      "loss": 0.2965,
      "step": 34035
    },
    {
      "epoch": 1.5619292368408977,
      "grad_norm": 0.4840381443500519,
      "learning_rate": 4.829504868270071e-06,
      "loss": 0.3805,
      "step": 34036
    },
    {
      "epoch": 1.5619751273461522,
      "grad_norm": 0.4488910436630249,
      "learning_rate": 4.829259823154864e-06,
      "loss": 0.3313,
      "step": 34037
    },
    {
      "epoch": 1.5620210178514067,
      "grad_norm": 0.46201086044311523,
      "learning_rate": 4.829014778450234e-06,
      "loss": 0.3824,
      "step": 34038
    },
    {
      "epoch": 1.562066908356661,
      "grad_norm": 0.4562823474407196,
      "learning_rate": 4.828769734156768e-06,
      "loss": 0.3247,
      "step": 34039
    },
    {
      "epoch": 1.5621127988619155,
      "grad_norm": 0.4547005295753479,
      "learning_rate": 4.828524690275058e-06,
      "loss": 0.3168,
      "step": 34040
    },
    {
      "epoch": 1.56215868936717,
      "grad_norm": 0.47842517495155334,
      "learning_rate": 4.828279646805691e-06,
      "loss": 0.3479,
      "step": 34041
    },
    {
      "epoch": 1.5622045798724244,
      "grad_norm": 0.4355309009552002,
      "learning_rate": 4.828034603749256e-06,
      "loss": 0.2759,
      "step": 34042
    },
    {
      "epoch": 1.5622504703776787,
      "grad_norm": 0.49969831109046936,
      "learning_rate": 4.827789561106346e-06,
      "loss": 0.4011,
      "step": 34043
    },
    {
      "epoch": 1.5622963608829332,
      "grad_norm": 0.4737112820148468,
      "learning_rate": 4.827544518877547e-06,
      "loss": 0.3617,
      "step": 34044
    },
    {
      "epoch": 1.5623422513881877,
      "grad_norm": 0.45174118876457214,
      "learning_rate": 4.827299477063449e-06,
      "loss": 0.3112,
      "step": 34045
    },
    {
      "epoch": 1.5623881418934422,
      "grad_norm": 0.4617692828178406,
      "learning_rate": 4.827054435664641e-06,
      "loss": 0.3662,
      "step": 34046
    },
    {
      "epoch": 1.5624340323986967,
      "grad_norm": 0.4640926122665405,
      "learning_rate": 4.8268093946817125e-06,
      "loss": 0.3432,
      "step": 34047
    },
    {
      "epoch": 1.5624799229039512,
      "grad_norm": 0.45087364315986633,
      "learning_rate": 4.826564354115251e-06,
      "loss": 0.3364,
      "step": 34048
    },
    {
      "epoch": 1.5625258134092057,
      "grad_norm": 0.4648876190185547,
      "learning_rate": 4.8263193139658496e-06,
      "loss": 0.3828,
      "step": 34049
    },
    {
      "epoch": 1.5625717039144602,
      "grad_norm": 0.4506319463253021,
      "learning_rate": 4.826074274234096e-06,
      "loss": 0.3398,
      "step": 34050
    },
    {
      "epoch": 1.5626175944197147,
      "grad_norm": 0.4686102271080017,
      "learning_rate": 4.825829234920578e-06,
      "loss": 0.3428,
      "step": 34051
    },
    {
      "epoch": 1.562663484924969,
      "grad_norm": 0.5071650147438049,
      "learning_rate": 4.825584196025886e-06,
      "loss": 0.3691,
      "step": 34052
    },
    {
      "epoch": 1.5627093754302235,
      "grad_norm": 0.4780280888080597,
      "learning_rate": 4.8253391575506085e-06,
      "loss": 0.3536,
      "step": 34053
    },
    {
      "epoch": 1.562755265935478,
      "grad_norm": 0.41944676637649536,
      "learning_rate": 4.825094119495334e-06,
      "loss": 0.2997,
      "step": 34054
    },
    {
      "epoch": 1.5628011564407323,
      "grad_norm": 0.46455785632133484,
      "learning_rate": 4.8248490818606545e-06,
      "loss": 0.3131,
      "step": 34055
    },
    {
      "epoch": 1.5628470469459868,
      "grad_norm": 0.4990754723548889,
      "learning_rate": 4.82460404464716e-06,
      "loss": 0.3579,
      "step": 34056
    },
    {
      "epoch": 1.5628929374512412,
      "grad_norm": 0.4536486864089966,
      "learning_rate": 4.824359007855433e-06,
      "loss": 0.3371,
      "step": 34057
    },
    {
      "epoch": 1.5629388279564957,
      "grad_norm": 0.4606002867221832,
      "learning_rate": 4.824113971486069e-06,
      "loss": 0.3218,
      "step": 34058
    },
    {
      "epoch": 1.5629847184617502,
      "grad_norm": 0.503911554813385,
      "learning_rate": 4.823868935539656e-06,
      "loss": 0.4206,
      "step": 34059
    },
    {
      "epoch": 1.5630306089670047,
      "grad_norm": 0.4997771084308624,
      "learning_rate": 4.823623900016783e-06,
      "loss": 0.412,
      "step": 34060
    },
    {
      "epoch": 1.5630764994722592,
      "grad_norm": 0.48437240719795227,
      "learning_rate": 4.823378864918037e-06,
      "loss": 0.4098,
      "step": 34061
    },
    {
      "epoch": 1.5631223899775137,
      "grad_norm": 0.5180996060371399,
      "learning_rate": 4.8231338302440104e-06,
      "loss": 0.4011,
      "step": 34062
    },
    {
      "epoch": 1.5631682804827682,
      "grad_norm": 0.47583699226379395,
      "learning_rate": 4.822888795995291e-06,
      "loss": 0.3855,
      "step": 34063
    },
    {
      "epoch": 1.5632141709880227,
      "grad_norm": 0.4277336001396179,
      "learning_rate": 4.822643762172466e-06,
      "loss": 0.2978,
      "step": 34064
    },
    {
      "epoch": 1.563260061493277,
      "grad_norm": 0.48672977089881897,
      "learning_rate": 4.82239872877613e-06,
      "loss": 0.3907,
      "step": 34065
    },
    {
      "epoch": 1.5633059519985315,
      "grad_norm": 0.4795621633529663,
      "learning_rate": 4.822153695806868e-06,
      "loss": 0.3968,
      "step": 34066
    },
    {
      "epoch": 1.563351842503786,
      "grad_norm": 0.3996714949607849,
      "learning_rate": 4.821908663265269e-06,
      "loss": 0.2446,
      "step": 34067
    },
    {
      "epoch": 1.5633977330090403,
      "grad_norm": 0.4766741991043091,
      "learning_rate": 4.821663631151925e-06,
      "loss": 0.3754,
      "step": 34068
    },
    {
      "epoch": 1.5634436235142948,
      "grad_norm": 0.4672122597694397,
      "learning_rate": 4.821418599467424e-06,
      "loss": 0.3716,
      "step": 34069
    },
    {
      "epoch": 1.5634895140195493,
      "grad_norm": 0.4888591468334198,
      "learning_rate": 4.8211735682123515e-06,
      "loss": 0.3927,
      "step": 34070
    },
    {
      "epoch": 1.5635354045248038,
      "grad_norm": 0.5323549509048462,
      "learning_rate": 4.820928537387304e-06,
      "loss": 0.4709,
      "step": 34071
    },
    {
      "epoch": 1.5635812950300583,
      "grad_norm": 0.4848652184009552,
      "learning_rate": 4.8206835069928644e-06,
      "loss": 0.386,
      "step": 34072
    },
    {
      "epoch": 1.5636271855353128,
      "grad_norm": 0.43395841121673584,
      "learning_rate": 4.820438477029625e-06,
      "loss": 0.2939,
      "step": 34073
    },
    {
      "epoch": 1.5636730760405673,
      "grad_norm": 0.42950427532196045,
      "learning_rate": 4.820193447498175e-06,
      "loss": 0.2506,
      "step": 34074
    },
    {
      "epoch": 1.5637189665458218,
      "grad_norm": 0.4448913335800171,
      "learning_rate": 4.8199484183991044e-06,
      "loss": 0.3698,
      "step": 34075
    },
    {
      "epoch": 1.5637648570510763,
      "grad_norm": 0.4335491359233856,
      "learning_rate": 4.819703389732997e-06,
      "loss": 0.3255,
      "step": 34076
    },
    {
      "epoch": 1.5638107475563308,
      "grad_norm": 0.4486835300922394,
      "learning_rate": 4.819458361500449e-06,
      "loss": 0.3066,
      "step": 34077
    },
    {
      "epoch": 1.563856638061585,
      "grad_norm": 0.4256046712398529,
      "learning_rate": 4.819213333702048e-06,
      "loss": 0.295,
      "step": 34078
    },
    {
      "epoch": 1.5639025285668395,
      "grad_norm": 0.46987104415893555,
      "learning_rate": 4.818968306338378e-06,
      "loss": 0.3363,
      "step": 34079
    },
    {
      "epoch": 1.563948419072094,
      "grad_norm": 0.48400285840034485,
      "learning_rate": 4.818723279410035e-06,
      "loss": 0.3743,
      "step": 34080
    },
    {
      "epoch": 1.5639943095773483,
      "grad_norm": 0.4564042091369629,
      "learning_rate": 4.818478252917604e-06,
      "loss": 0.297,
      "step": 34081
    },
    {
      "epoch": 1.5640402000826028,
      "grad_norm": 0.46303045749664307,
      "learning_rate": 4.818233226861676e-06,
      "loss": 0.3235,
      "step": 34082
    },
    {
      "epoch": 1.5640860905878573,
      "grad_norm": 0.4713778495788574,
      "learning_rate": 4.81798820124284e-06,
      "loss": 0.3575,
      "step": 34083
    },
    {
      "epoch": 1.5641319810931118,
      "grad_norm": 0.5021894574165344,
      "learning_rate": 4.817743176061685e-06,
      "loss": 0.3805,
      "step": 34084
    },
    {
      "epoch": 1.5641778715983663,
      "grad_norm": 0.4478503465652466,
      "learning_rate": 4.817498151318798e-06,
      "loss": 0.2923,
      "step": 34085
    },
    {
      "epoch": 1.5642237621036208,
      "grad_norm": 0.46586576104164124,
      "learning_rate": 4.817253127014773e-06,
      "loss": 0.3668,
      "step": 34086
    },
    {
      "epoch": 1.5642696526088753,
      "grad_norm": 0.43985575437545776,
      "learning_rate": 4.817008103150196e-06,
      "loss": 0.3208,
      "step": 34087
    },
    {
      "epoch": 1.5643155431141298,
      "grad_norm": 0.5239508152008057,
      "learning_rate": 4.8167630797256554e-06,
      "loss": 0.4144,
      "step": 34088
    },
    {
      "epoch": 1.5643614336193843,
      "grad_norm": 0.5419965982437134,
      "learning_rate": 4.816518056741743e-06,
      "loss": 0.4223,
      "step": 34089
    },
    {
      "epoch": 1.5644073241246386,
      "grad_norm": 0.44531160593032837,
      "learning_rate": 4.816273034199047e-06,
      "loss": 0.331,
      "step": 34090
    },
    {
      "epoch": 1.564453214629893,
      "grad_norm": 0.45661619305610657,
      "learning_rate": 4.816028012098154e-06,
      "loss": 0.3363,
      "step": 34091
    },
    {
      "epoch": 1.5644991051351476,
      "grad_norm": 0.4739121198654175,
      "learning_rate": 4.815782990439658e-06,
      "loss": 0.3863,
      "step": 34092
    },
    {
      "epoch": 1.5645449956404018,
      "grad_norm": 0.390281081199646,
      "learning_rate": 4.815537969224145e-06,
      "loss": 0.2511,
      "step": 34093
    },
    {
      "epoch": 1.5645908861456563,
      "grad_norm": 0.4408140480518341,
      "learning_rate": 4.815292948452207e-06,
      "loss": 0.3199,
      "step": 34094
    },
    {
      "epoch": 1.5646367766509108,
      "grad_norm": 0.4515714645385742,
      "learning_rate": 4.815047928124428e-06,
      "loss": 0.3418,
      "step": 34095
    },
    {
      "epoch": 1.5646826671561653,
      "grad_norm": 0.43468818068504333,
      "learning_rate": 4.814802908241402e-06,
      "loss": 0.3364,
      "step": 34096
    },
    {
      "epoch": 1.5647285576614198,
      "grad_norm": 0.527330219745636,
      "learning_rate": 4.814557888803717e-06,
      "loss": 0.4352,
      "step": 34097
    },
    {
      "epoch": 1.5647744481666743,
      "grad_norm": 0.46747714281082153,
      "learning_rate": 4.814312869811959e-06,
      "loss": 0.333,
      "step": 34098
    },
    {
      "epoch": 1.5648203386719288,
      "grad_norm": 0.46116548776626587,
      "learning_rate": 4.814067851266723e-06,
      "loss": 0.3168,
      "step": 34099
    },
    {
      "epoch": 1.5648662291771833,
      "grad_norm": 0.43344518542289734,
      "learning_rate": 4.813822833168596e-06,
      "loss": 0.3143,
      "step": 34100
    },
    {
      "epoch": 1.5649121196824378,
      "grad_norm": 0.4438704550266266,
      "learning_rate": 4.813577815518162e-06,
      "loss": 0.3295,
      "step": 34101
    },
    {
      "epoch": 1.5649580101876923,
      "grad_norm": 0.43321946263313293,
      "learning_rate": 4.813332798316017e-06,
      "loss": 0.2744,
      "step": 34102
    },
    {
      "epoch": 1.5650039006929466,
      "grad_norm": 0.4740264117717743,
      "learning_rate": 4.8130877815627485e-06,
      "loss": 0.3326,
      "step": 34103
    },
    {
      "epoch": 1.565049791198201,
      "grad_norm": 0.4633159041404724,
      "learning_rate": 4.812842765258943e-06,
      "loss": 0.3567,
      "step": 34104
    },
    {
      "epoch": 1.5650956817034556,
      "grad_norm": 0.4954968988895416,
      "learning_rate": 4.812597749405193e-06,
      "loss": 0.3875,
      "step": 34105
    },
    {
      "epoch": 1.5651415722087099,
      "grad_norm": 0.4982178807258606,
      "learning_rate": 4.812352734002087e-06,
      "loss": 0.4469,
      "step": 34106
    },
    {
      "epoch": 1.5651874627139644,
      "grad_norm": 0.4895936846733093,
      "learning_rate": 4.81210771905021e-06,
      "loss": 0.3594,
      "step": 34107
    },
    {
      "epoch": 1.5652333532192189,
      "grad_norm": 0.4916960299015045,
      "learning_rate": 4.811862704550158e-06,
      "loss": 0.3417,
      "step": 34108
    },
    {
      "epoch": 1.5652792437244734,
      "grad_norm": 0.4567507803440094,
      "learning_rate": 4.811617690502516e-06,
      "loss": 0.3092,
      "step": 34109
    },
    {
      "epoch": 1.5653251342297279,
      "grad_norm": 0.44885534048080444,
      "learning_rate": 4.8113726769078726e-06,
      "loss": 0.336,
      "step": 34110
    },
    {
      "epoch": 1.5653710247349824,
      "grad_norm": 0.4257388412952423,
      "learning_rate": 4.81112766376682e-06,
      "loss": 0.2859,
      "step": 34111
    },
    {
      "epoch": 1.5654169152402369,
      "grad_norm": 0.4598163068294525,
      "learning_rate": 4.810882651079946e-06,
      "loss": 0.3342,
      "step": 34112
    },
    {
      "epoch": 1.5654628057454913,
      "grad_norm": 0.43642479181289673,
      "learning_rate": 4.810637638847837e-06,
      "loss": 0.3102,
      "step": 34113
    },
    {
      "epoch": 1.5655086962507458,
      "grad_norm": 0.5114754438400269,
      "learning_rate": 4.810392627071087e-06,
      "loss": 0.4316,
      "step": 34114
    },
    {
      "epoch": 1.5655545867560003,
      "grad_norm": 0.4800611138343811,
      "learning_rate": 4.8101476157502835e-06,
      "loss": 0.3557,
      "step": 34115
    },
    {
      "epoch": 1.5656004772612546,
      "grad_norm": 0.4600962996482849,
      "learning_rate": 4.809902604886013e-06,
      "loss": 0.3676,
      "step": 34116
    },
    {
      "epoch": 1.5656463677665091,
      "grad_norm": 0.456121563911438,
      "learning_rate": 4.809657594478868e-06,
      "loss": 0.3175,
      "step": 34117
    },
    {
      "epoch": 1.5656922582717636,
      "grad_norm": 0.5247764587402344,
      "learning_rate": 4.809412584529437e-06,
      "loss": 0.4591,
      "step": 34118
    },
    {
      "epoch": 1.565738148777018,
      "grad_norm": 0.44318264722824097,
      "learning_rate": 4.809167575038306e-06,
      "loss": 0.3368,
      "step": 34119
    },
    {
      "epoch": 1.5657840392822724,
      "grad_norm": 0.4649055302143097,
      "learning_rate": 4.808922566006069e-06,
      "loss": 0.3509,
      "step": 34120
    },
    {
      "epoch": 1.5658299297875269,
      "grad_norm": 0.42826753854751587,
      "learning_rate": 4.8086775574333124e-06,
      "loss": 0.2989,
      "step": 34121
    },
    {
      "epoch": 1.5658758202927814,
      "grad_norm": 0.4467582702636719,
      "learning_rate": 4.8084325493206255e-06,
      "loss": 0.3214,
      "step": 34122
    },
    {
      "epoch": 1.5659217107980359,
      "grad_norm": 0.4620489776134491,
      "learning_rate": 4.808187541668599e-06,
      "loss": 0.3586,
      "step": 34123
    },
    {
      "epoch": 1.5659676013032904,
      "grad_norm": 0.4536444842815399,
      "learning_rate": 4.807942534477821e-06,
      "loss": 0.3361,
      "step": 34124
    },
    {
      "epoch": 1.5660134918085449,
      "grad_norm": 0.4625416398048401,
      "learning_rate": 4.807697527748879e-06,
      "loss": 0.3724,
      "step": 34125
    },
    {
      "epoch": 1.5660593823137994,
      "grad_norm": 0.4803650677204132,
      "learning_rate": 4.807452521482365e-06,
      "loss": 0.3573,
      "step": 34126
    },
    {
      "epoch": 1.5661052728190539,
      "grad_norm": 0.5104102492332458,
      "learning_rate": 4.807207515678867e-06,
      "loss": 0.3246,
      "step": 34127
    },
    {
      "epoch": 1.5661511633243081,
      "grad_norm": 0.4634581208229065,
      "learning_rate": 4.806962510338974e-06,
      "loss": 0.3635,
      "step": 34128
    },
    {
      "epoch": 1.5661970538295626,
      "grad_norm": 0.4899982810020447,
      "learning_rate": 4.806717505463273e-06,
      "loss": 0.437,
      "step": 34129
    },
    {
      "epoch": 1.5662429443348171,
      "grad_norm": 0.485330194234848,
      "learning_rate": 4.806472501052357e-06,
      "loss": 0.3696,
      "step": 34130
    },
    {
      "epoch": 1.5662888348400716,
      "grad_norm": 0.4710325002670288,
      "learning_rate": 4.806227497106814e-06,
      "loss": 0.4068,
      "step": 34131
    },
    {
      "epoch": 1.566334725345326,
      "grad_norm": 0.49292463064193726,
      "learning_rate": 4.805982493627232e-06,
      "loss": 0.3812,
      "step": 34132
    },
    {
      "epoch": 1.5663806158505804,
      "grad_norm": 0.4320945739746094,
      "learning_rate": 4.805737490614202e-06,
      "loss": 0.2572,
      "step": 34133
    },
    {
      "epoch": 1.566426506355835,
      "grad_norm": 0.46861350536346436,
      "learning_rate": 4.805492488068312e-06,
      "loss": 0.396,
      "step": 34134
    },
    {
      "epoch": 1.5664723968610894,
      "grad_norm": 0.5624412298202515,
      "learning_rate": 4.805247485990149e-06,
      "loss": 0.3063,
      "step": 34135
    },
    {
      "epoch": 1.566518287366344,
      "grad_norm": 0.44513922929763794,
      "learning_rate": 4.805002484380306e-06,
      "loss": 0.3538,
      "step": 34136
    },
    {
      "epoch": 1.5665641778715984,
      "grad_norm": 0.4849376082420349,
      "learning_rate": 4.804757483239371e-06,
      "loss": 0.3716,
      "step": 34137
    },
    {
      "epoch": 1.566610068376853,
      "grad_norm": 0.508667528629303,
      "learning_rate": 4.804512482567931e-06,
      "loss": 0.3884,
      "step": 34138
    },
    {
      "epoch": 1.5666559588821074,
      "grad_norm": 0.4708024263381958,
      "learning_rate": 4.804267482366578e-06,
      "loss": 0.3486,
      "step": 34139
    },
    {
      "epoch": 1.566701849387362,
      "grad_norm": 0.4759385883808136,
      "learning_rate": 4.804022482635901e-06,
      "loss": 0.3555,
      "step": 34140
    },
    {
      "epoch": 1.5667477398926162,
      "grad_norm": 0.5195858478546143,
      "learning_rate": 4.803777483376484e-06,
      "loss": 0.4318,
      "step": 34141
    },
    {
      "epoch": 1.5667936303978707,
      "grad_norm": 0.4676797688007355,
      "learning_rate": 4.8035324845889246e-06,
      "loss": 0.3377,
      "step": 34142
    },
    {
      "epoch": 1.5668395209031252,
      "grad_norm": 0.4206415116786957,
      "learning_rate": 4.803287486273806e-06,
      "loss": 0.2664,
      "step": 34143
    },
    {
      "epoch": 1.5668854114083794,
      "grad_norm": 0.477561891078949,
      "learning_rate": 4.803042488431718e-06,
      "loss": 0.3829,
      "step": 34144
    },
    {
      "epoch": 1.566931301913634,
      "grad_norm": 0.4732050597667694,
      "learning_rate": 4.802797491063252e-06,
      "loss": 0.3805,
      "step": 34145
    },
    {
      "epoch": 1.5669771924188884,
      "grad_norm": 0.4746980369091034,
      "learning_rate": 4.802552494168995e-06,
      "loss": 0.3592,
      "step": 34146
    },
    {
      "epoch": 1.567023082924143,
      "grad_norm": 0.4743090569972992,
      "learning_rate": 4.802307497749537e-06,
      "loss": 0.3497,
      "step": 34147
    },
    {
      "epoch": 1.5670689734293974,
      "grad_norm": 0.4263301491737366,
      "learning_rate": 4.802062501805468e-06,
      "loss": 0.2692,
      "step": 34148
    },
    {
      "epoch": 1.567114863934652,
      "grad_norm": 0.5128852128982544,
      "learning_rate": 4.801817506337376e-06,
      "loss": 0.406,
      "step": 34149
    },
    {
      "epoch": 1.5671607544399064,
      "grad_norm": 0.43033042550086975,
      "learning_rate": 4.8015725113458485e-06,
      "loss": 0.2896,
      "step": 34150
    },
    {
      "epoch": 1.567206644945161,
      "grad_norm": 0.4806380271911621,
      "learning_rate": 4.801327516831478e-06,
      "loss": 0.3602,
      "step": 34151
    },
    {
      "epoch": 1.5672525354504154,
      "grad_norm": 0.43870940804481506,
      "learning_rate": 4.801082522794853e-06,
      "loss": 0.3172,
      "step": 34152
    },
    {
      "epoch": 1.56729842595567,
      "grad_norm": 0.474739670753479,
      "learning_rate": 4.800837529236559e-06,
      "loss": 0.3485,
      "step": 34153
    },
    {
      "epoch": 1.5673443164609242,
      "grad_norm": 0.488978773355484,
      "learning_rate": 4.80059253615719e-06,
      "loss": 0.3419,
      "step": 34154
    },
    {
      "epoch": 1.5673902069661787,
      "grad_norm": 0.4817551076412201,
      "learning_rate": 4.800347543557334e-06,
      "loss": 0.3756,
      "step": 34155
    },
    {
      "epoch": 1.5674360974714332,
      "grad_norm": 0.4170075058937073,
      "learning_rate": 4.800102551437576e-06,
      "loss": 0.2632,
      "step": 34156
    },
    {
      "epoch": 1.5674819879766875,
      "grad_norm": 0.45500314235687256,
      "learning_rate": 4.79985755979851e-06,
      "loss": 0.3362,
      "step": 34157
    },
    {
      "epoch": 1.567527878481942,
      "grad_norm": 0.47636979818344116,
      "learning_rate": 4.799612568640724e-06,
      "loss": 0.3839,
      "step": 34158
    },
    {
      "epoch": 1.5675737689871965,
      "grad_norm": 0.46664899587631226,
      "learning_rate": 4.799367577964807e-06,
      "loss": 0.3696,
      "step": 34159
    },
    {
      "epoch": 1.567619659492451,
      "grad_norm": 0.546671450138092,
      "learning_rate": 4.799122587771346e-06,
      "loss": 0.3555,
      "step": 34160
    },
    {
      "epoch": 1.5676655499977055,
      "grad_norm": 0.488545686006546,
      "learning_rate": 4.798877598060932e-06,
      "loss": 0.3409,
      "step": 34161
    },
    {
      "epoch": 1.56771144050296,
      "grad_norm": 0.44966620206832886,
      "learning_rate": 4.798632608834155e-06,
      "loss": 0.3389,
      "step": 34162
    },
    {
      "epoch": 1.5677573310082145,
      "grad_norm": 0.47057265043258667,
      "learning_rate": 4.798387620091601e-06,
      "loss": 0.3792,
      "step": 34163
    },
    {
      "epoch": 1.567803221513469,
      "grad_norm": 0.47142040729522705,
      "learning_rate": 4.7981426318338624e-06,
      "loss": 0.3777,
      "step": 34164
    },
    {
      "epoch": 1.5678491120187235,
      "grad_norm": 0.46498560905456543,
      "learning_rate": 4.797897644061527e-06,
      "loss": 0.3643,
      "step": 34165
    },
    {
      "epoch": 1.567895002523978,
      "grad_norm": 0.4516003131866455,
      "learning_rate": 4.797652656775184e-06,
      "loss": 0.3476,
      "step": 34166
    },
    {
      "epoch": 1.5679408930292322,
      "grad_norm": 0.44859495759010315,
      "learning_rate": 4.7974076699754234e-06,
      "loss": 0.3452,
      "step": 34167
    },
    {
      "epoch": 1.5679867835344867,
      "grad_norm": 0.47780841588974,
      "learning_rate": 4.797162683662834e-06,
      "loss": 0.388,
      "step": 34168
    },
    {
      "epoch": 1.5680326740397412,
      "grad_norm": 0.41988447308540344,
      "learning_rate": 4.796917697838002e-06,
      "loss": 0.3009,
      "step": 34169
    },
    {
      "epoch": 1.5680785645449955,
      "grad_norm": 0.4456281065940857,
      "learning_rate": 4.79667271250152e-06,
      "loss": 0.2994,
      "step": 34170
    },
    {
      "epoch": 1.56812445505025,
      "grad_norm": 0.4825037121772766,
      "learning_rate": 4.796427727653976e-06,
      "loss": 0.3701,
      "step": 34171
    },
    {
      "epoch": 1.5681703455555045,
      "grad_norm": 0.46703827381134033,
      "learning_rate": 4.796182743295957e-06,
      "loss": 0.3723,
      "step": 34172
    },
    {
      "epoch": 1.568216236060759,
      "grad_norm": 0.4766594171524048,
      "learning_rate": 4.795937759428056e-06,
      "loss": 0.3976,
      "step": 34173
    },
    {
      "epoch": 1.5682621265660135,
      "grad_norm": 0.4836577773094177,
      "learning_rate": 4.7956927760508615e-06,
      "loss": 0.4145,
      "step": 34174
    },
    {
      "epoch": 1.568308017071268,
      "grad_norm": 0.4990769326686859,
      "learning_rate": 4.7954477931649596e-06,
      "loss": 0.3345,
      "step": 34175
    },
    {
      "epoch": 1.5683539075765225,
      "grad_norm": 0.441350519657135,
      "learning_rate": 4.795202810770942e-06,
      "loss": 0.2888,
      "step": 34176
    },
    {
      "epoch": 1.568399798081777,
      "grad_norm": 0.47990211844444275,
      "learning_rate": 4.794957828869397e-06,
      "loss": 0.3361,
      "step": 34177
    },
    {
      "epoch": 1.5684456885870315,
      "grad_norm": 0.46796905994415283,
      "learning_rate": 4.794712847460912e-06,
      "loss": 0.3484,
      "step": 34178
    },
    {
      "epoch": 1.5684915790922858,
      "grad_norm": 0.4664016664028168,
      "learning_rate": 4.79446786654608e-06,
      "loss": 0.3224,
      "step": 34179
    },
    {
      "epoch": 1.5685374695975403,
      "grad_norm": 0.43120306730270386,
      "learning_rate": 4.794222886125487e-06,
      "loss": 0.2798,
      "step": 34180
    },
    {
      "epoch": 1.5685833601027948,
      "grad_norm": 0.45119747519493103,
      "learning_rate": 4.793977906199723e-06,
      "loss": 0.3149,
      "step": 34181
    },
    {
      "epoch": 1.568629250608049,
      "grad_norm": 0.4717291295528412,
      "learning_rate": 4.7937329267693776e-06,
      "loss": 0.3465,
      "step": 34182
    },
    {
      "epoch": 1.5686751411133035,
      "grad_norm": 0.5033001899719238,
      "learning_rate": 4.79348794783504e-06,
      "loss": 0.4083,
      "step": 34183
    },
    {
      "epoch": 1.568721031618558,
      "grad_norm": 0.5068426132202148,
      "learning_rate": 4.793242969397296e-06,
      "loss": 0.433,
      "step": 34184
    },
    {
      "epoch": 1.5687669221238125,
      "grad_norm": 0.4615301191806793,
      "learning_rate": 4.79299799145674e-06,
      "loss": 0.3624,
      "step": 34185
    },
    {
      "epoch": 1.568812812629067,
      "grad_norm": 0.5054537057876587,
      "learning_rate": 4.7927530140139575e-06,
      "loss": 0.4236,
      "step": 34186
    },
    {
      "epoch": 1.5688587031343215,
      "grad_norm": 0.45371052622795105,
      "learning_rate": 4.7925080370695385e-06,
      "loss": 0.3225,
      "step": 34187
    },
    {
      "epoch": 1.568904593639576,
      "grad_norm": 0.47033756971359253,
      "learning_rate": 4.792263060624073e-06,
      "loss": 0.3205,
      "step": 34188
    },
    {
      "epoch": 1.5689504841448305,
      "grad_norm": 0.4672520160675049,
      "learning_rate": 4.79201808467815e-06,
      "loss": 0.3325,
      "step": 34189
    },
    {
      "epoch": 1.568996374650085,
      "grad_norm": 0.46877363324165344,
      "learning_rate": 4.791773109232354e-06,
      "loss": 0.3716,
      "step": 34190
    },
    {
      "epoch": 1.5690422651553395,
      "grad_norm": 0.4696226418018341,
      "learning_rate": 4.791528134287283e-06,
      "loss": 0.347,
      "step": 34191
    },
    {
      "epoch": 1.5690881556605938,
      "grad_norm": 0.4276796281337738,
      "learning_rate": 4.791283159843518e-06,
      "loss": 0.2654,
      "step": 34192
    },
    {
      "epoch": 1.5691340461658483,
      "grad_norm": 0.47269725799560547,
      "learning_rate": 4.791038185901653e-06,
      "loss": 0.3592,
      "step": 34193
    },
    {
      "epoch": 1.5691799366711028,
      "grad_norm": 0.4391658902168274,
      "learning_rate": 4.790793212462272e-06,
      "loss": 0.302,
      "step": 34194
    },
    {
      "epoch": 1.569225827176357,
      "grad_norm": 0.46739891171455383,
      "learning_rate": 4.79054823952597e-06,
      "loss": 0.3401,
      "step": 34195
    },
    {
      "epoch": 1.5692717176816116,
      "grad_norm": 0.4465655982494354,
      "learning_rate": 4.790303267093333e-06,
      "loss": 0.3205,
      "step": 34196
    },
    {
      "epoch": 1.569317608186866,
      "grad_norm": 0.5543652772903442,
      "learning_rate": 4.79005829516495e-06,
      "loss": 0.499,
      "step": 34197
    },
    {
      "epoch": 1.5693634986921206,
      "grad_norm": 0.44123291969299316,
      "learning_rate": 4.789813323741411e-06,
      "loss": 0.3184,
      "step": 34198
    },
    {
      "epoch": 1.569409389197375,
      "grad_norm": 0.9951797127723694,
      "learning_rate": 4.789568352823304e-06,
      "loss": 0.4241,
      "step": 34199
    },
    {
      "epoch": 1.5694552797026295,
      "grad_norm": 0.518595814704895,
      "learning_rate": 4.789323382411218e-06,
      "loss": 0.3935,
      "step": 34200
    },
    {
      "epoch": 1.569501170207884,
      "grad_norm": 0.5077909827232361,
      "learning_rate": 4.7890784125057435e-06,
      "loss": 0.3957,
      "step": 34201
    },
    {
      "epoch": 1.5695470607131385,
      "grad_norm": 0.44167929887771606,
      "learning_rate": 4.78883344310747e-06,
      "loss": 0.2878,
      "step": 34202
    },
    {
      "epoch": 1.569592951218393,
      "grad_norm": 0.4390799105167389,
      "learning_rate": 4.788588474216984e-06,
      "loss": 0.298,
      "step": 34203
    },
    {
      "epoch": 1.5696388417236475,
      "grad_norm": 0.44792142510414124,
      "learning_rate": 4.788343505834877e-06,
      "loss": 0.3597,
      "step": 34204
    },
    {
      "epoch": 1.5696847322289018,
      "grad_norm": 0.4795165956020355,
      "learning_rate": 4.788098537961737e-06,
      "loss": 0.3339,
      "step": 34205
    },
    {
      "epoch": 1.5697306227341563,
      "grad_norm": 0.4769688844680786,
      "learning_rate": 4.787853570598151e-06,
      "loss": 0.3513,
      "step": 34206
    },
    {
      "epoch": 1.5697765132394108,
      "grad_norm": 0.4868701100349426,
      "learning_rate": 4.787608603744712e-06,
      "loss": 0.3512,
      "step": 34207
    },
    {
      "epoch": 1.569822403744665,
      "grad_norm": 0.4186769127845764,
      "learning_rate": 4.787363637402008e-06,
      "loss": 0.2903,
      "step": 34208
    },
    {
      "epoch": 1.5698682942499196,
      "grad_norm": 0.4714920222759247,
      "learning_rate": 4.7871186715706266e-06,
      "loss": 0.3919,
      "step": 34209
    },
    {
      "epoch": 1.569914184755174,
      "grad_norm": 0.475122332572937,
      "learning_rate": 4.786873706251158e-06,
      "loss": 0.3476,
      "step": 34210
    },
    {
      "epoch": 1.5699600752604286,
      "grad_norm": 0.46369802951812744,
      "learning_rate": 4.78662874144419e-06,
      "loss": 0.3482,
      "step": 34211
    },
    {
      "epoch": 1.570005965765683,
      "grad_norm": 0.46473565697669983,
      "learning_rate": 4.786383777150312e-06,
      "loss": 0.3476,
      "step": 34212
    },
    {
      "epoch": 1.5700518562709376,
      "grad_norm": 0.4648815393447876,
      "learning_rate": 4.786138813370116e-06,
      "loss": 0.3365,
      "step": 34213
    },
    {
      "epoch": 1.570097746776192,
      "grad_norm": 0.48359236121177673,
      "learning_rate": 4.785893850104187e-06,
      "loss": 0.3704,
      "step": 34214
    },
    {
      "epoch": 1.5701436372814466,
      "grad_norm": 0.45097431540489197,
      "learning_rate": 4.785648887353115e-06,
      "loss": 0.3388,
      "step": 34215
    },
    {
      "epoch": 1.570189527786701,
      "grad_norm": 0.49320828914642334,
      "learning_rate": 4.785403925117491e-06,
      "loss": 0.3925,
      "step": 34216
    },
    {
      "epoch": 1.5702354182919553,
      "grad_norm": 0.4789159595966339,
      "learning_rate": 4.785158963397903e-06,
      "loss": 0.398,
      "step": 34217
    },
    {
      "epoch": 1.5702813087972098,
      "grad_norm": 0.4815405309200287,
      "learning_rate": 4.784914002194939e-06,
      "loss": 0.351,
      "step": 34218
    },
    {
      "epoch": 1.5703271993024643,
      "grad_norm": 0.46454918384552,
      "learning_rate": 4.7846690415091896e-06,
      "loss": 0.3572,
      "step": 34219
    },
    {
      "epoch": 1.5703730898077188,
      "grad_norm": 0.48363715410232544,
      "learning_rate": 4.784424081341243e-06,
      "loss": 0.3929,
      "step": 34220
    },
    {
      "epoch": 1.5704189803129731,
      "grad_norm": 0.45669469237327576,
      "learning_rate": 4.784179121691687e-06,
      "loss": 0.3271,
      "step": 34221
    },
    {
      "epoch": 1.5704648708182276,
      "grad_norm": 0.4062530994415283,
      "learning_rate": 4.783934162561114e-06,
      "loss": 0.2591,
      "step": 34222
    },
    {
      "epoch": 1.570510761323482,
      "grad_norm": 0.5039300918579102,
      "learning_rate": 4.783689203950111e-06,
      "loss": 0.3835,
      "step": 34223
    },
    {
      "epoch": 1.5705566518287366,
      "grad_norm": 0.49327167868614197,
      "learning_rate": 4.783444245859268e-06,
      "loss": 0.3464,
      "step": 34224
    },
    {
      "epoch": 1.570602542333991,
      "grad_norm": 0.4999888241291046,
      "learning_rate": 4.783199288289171e-06,
      "loss": 0.3878,
      "step": 34225
    },
    {
      "epoch": 1.5706484328392456,
      "grad_norm": 0.43679288029670715,
      "learning_rate": 4.782954331240413e-06,
      "loss": 0.3191,
      "step": 34226
    },
    {
      "epoch": 1.5706943233445,
      "grad_norm": 0.4738425016403198,
      "learning_rate": 4.7827093747135815e-06,
      "loss": 0.3629,
      "step": 34227
    },
    {
      "epoch": 1.5707402138497546,
      "grad_norm": 0.46536460518836975,
      "learning_rate": 4.782464418709263e-06,
      "loss": 0.3416,
      "step": 34228
    },
    {
      "epoch": 1.570786104355009,
      "grad_norm": 0.5771710276603699,
      "learning_rate": 4.782219463228051e-06,
      "loss": 0.3923,
      "step": 34229
    },
    {
      "epoch": 1.5708319948602634,
      "grad_norm": 0.4636612832546234,
      "learning_rate": 4.781974508270532e-06,
      "loss": 0.3378,
      "step": 34230
    },
    {
      "epoch": 1.5708778853655179,
      "grad_norm": 0.43743571639060974,
      "learning_rate": 4.781729553837295e-06,
      "loss": 0.3274,
      "step": 34231
    },
    {
      "epoch": 1.5709237758707724,
      "grad_norm": 0.4824206531047821,
      "learning_rate": 4.7814845999289305e-06,
      "loss": 0.3621,
      "step": 34232
    },
    {
      "epoch": 1.5709696663760266,
      "grad_norm": 0.45043420791625977,
      "learning_rate": 4.781239646546026e-06,
      "loss": 0.3136,
      "step": 34233
    },
    {
      "epoch": 1.5710155568812811,
      "grad_norm": 0.46529099345207214,
      "learning_rate": 4.78099469368917e-06,
      "loss": 0.3439,
      "step": 34234
    },
    {
      "epoch": 1.5710614473865356,
      "grad_norm": 0.4561481773853302,
      "learning_rate": 4.780749741358955e-06,
      "loss": 0.332,
      "step": 34235
    },
    {
      "epoch": 1.5711073378917901,
      "grad_norm": 0.49924275279045105,
      "learning_rate": 4.7805047895559676e-06,
      "loss": 0.4402,
      "step": 34236
    },
    {
      "epoch": 1.5711532283970446,
      "grad_norm": 0.48606640100479126,
      "learning_rate": 4.780259838280794e-06,
      "loss": 0.368,
      "step": 34237
    },
    {
      "epoch": 1.5711991189022991,
      "grad_norm": 0.44855543971061707,
      "learning_rate": 4.780014887534028e-06,
      "loss": 0.3563,
      "step": 34238
    },
    {
      "epoch": 1.5712450094075536,
      "grad_norm": 0.4757257401943207,
      "learning_rate": 4.779769937316258e-06,
      "loss": 0.3406,
      "step": 34239
    },
    {
      "epoch": 1.5712908999128081,
      "grad_norm": 0.45499545335769653,
      "learning_rate": 4.779524987628069e-06,
      "loss": 0.3636,
      "step": 34240
    },
    {
      "epoch": 1.5713367904180626,
      "grad_norm": 0.4529830813407898,
      "learning_rate": 4.779280038470055e-06,
      "loss": 0.3125,
      "step": 34241
    },
    {
      "epoch": 1.5713826809233171,
      "grad_norm": 0.450122207403183,
      "learning_rate": 4.779035089842803e-06,
      "loss": 0.3131,
      "step": 34242
    },
    {
      "epoch": 1.5714285714285714,
      "grad_norm": 0.44742995500564575,
      "learning_rate": 4.778790141746899e-06,
      "loss": 0.2988,
      "step": 34243
    },
    {
      "epoch": 1.571474461933826,
      "grad_norm": 0.4290243983268738,
      "learning_rate": 4.778545194182937e-06,
      "loss": 0.2892,
      "step": 34244
    },
    {
      "epoch": 1.5715203524390804,
      "grad_norm": 0.44179683923721313,
      "learning_rate": 4.778300247151505e-06,
      "loss": 0.3292,
      "step": 34245
    },
    {
      "epoch": 1.5715662429443347,
      "grad_norm": 0.4820263087749481,
      "learning_rate": 4.778055300653189e-06,
      "loss": 0.3378,
      "step": 34246
    },
    {
      "epoch": 1.5716121334495892,
      "grad_norm": 0.45052969455718994,
      "learning_rate": 4.777810354688581e-06,
      "loss": 0.3183,
      "step": 34247
    },
    {
      "epoch": 1.5716580239548437,
      "grad_norm": 0.4890190064907074,
      "learning_rate": 4.77756540925827e-06,
      "loss": 0.3422,
      "step": 34248
    },
    {
      "epoch": 1.5717039144600982,
      "grad_norm": 0.4480682909488678,
      "learning_rate": 4.77732046436284e-06,
      "loss": 0.3448,
      "step": 34249
    },
    {
      "epoch": 1.5717498049653527,
      "grad_norm": 0.4340474605560303,
      "learning_rate": 4.777075520002887e-06,
      "loss": 0.3099,
      "step": 34250
    },
    {
      "epoch": 1.5717956954706072,
      "grad_norm": 0.5461264252662659,
      "learning_rate": 4.776830576178998e-06,
      "loss": 0.4269,
      "step": 34251
    },
    {
      "epoch": 1.5718415859758617,
      "grad_norm": 0.48331037163734436,
      "learning_rate": 4.7765856328917585e-06,
      "loss": 0.3697,
      "step": 34252
    },
    {
      "epoch": 1.5718874764811162,
      "grad_norm": 0.44253432750701904,
      "learning_rate": 4.776340690141762e-06,
      "loss": 0.3097,
      "step": 34253
    },
    {
      "epoch": 1.5719333669863707,
      "grad_norm": 0.4688449800014496,
      "learning_rate": 4.776095747929596e-06,
      "loss": 0.3254,
      "step": 34254
    },
    {
      "epoch": 1.5719792574916251,
      "grad_norm": 0.4992218017578125,
      "learning_rate": 4.775850806255845e-06,
      "loss": 0.37,
      "step": 34255
    },
    {
      "epoch": 1.5720251479968794,
      "grad_norm": 0.48047375679016113,
      "learning_rate": 4.775605865121106e-06,
      "loss": 0.3926,
      "step": 34256
    },
    {
      "epoch": 1.572071038502134,
      "grad_norm": 0.4353560209274292,
      "learning_rate": 4.7753609245259644e-06,
      "loss": 0.3177,
      "step": 34257
    },
    {
      "epoch": 1.5721169290073884,
      "grad_norm": 0.4713706374168396,
      "learning_rate": 4.7751159844710085e-06,
      "loss": 0.3953,
      "step": 34258
    },
    {
      "epoch": 1.5721628195126427,
      "grad_norm": 0.48382440209388733,
      "learning_rate": 4.774871044956825e-06,
      "loss": 0.3195,
      "step": 34259
    },
    {
      "epoch": 1.5722087100178972,
      "grad_norm": 0.4641120135784149,
      "learning_rate": 4.774626105984008e-06,
      "loss": 0.3647,
      "step": 34260
    },
    {
      "epoch": 1.5722546005231517,
      "grad_norm": 0.4703112840652466,
      "learning_rate": 4.774381167553143e-06,
      "loss": 0.363,
      "step": 34261
    },
    {
      "epoch": 1.5723004910284062,
      "grad_norm": 0.4731237292289734,
      "learning_rate": 4.774136229664821e-06,
      "loss": 0.387,
      "step": 34262
    },
    {
      "epoch": 1.5723463815336607,
      "grad_norm": 0.493058979511261,
      "learning_rate": 4.77389129231963e-06,
      "loss": 0.4028,
      "step": 34263
    },
    {
      "epoch": 1.5723922720389152,
      "grad_norm": 0.4944364130496979,
      "learning_rate": 4.773646355518158e-06,
      "loss": 0.3621,
      "step": 34264
    },
    {
      "epoch": 1.5724381625441697,
      "grad_norm": 0.4646797180175781,
      "learning_rate": 4.773401419260995e-06,
      "loss": 0.3789,
      "step": 34265
    },
    {
      "epoch": 1.5724840530494242,
      "grad_norm": 0.4787401854991913,
      "learning_rate": 4.77315648354873e-06,
      "loss": 0.4034,
      "step": 34266
    },
    {
      "epoch": 1.5725299435546787,
      "grad_norm": 0.45806455612182617,
      "learning_rate": 4.772911548381954e-06,
      "loss": 0.2936,
      "step": 34267
    },
    {
      "epoch": 1.572575834059933,
      "grad_norm": 0.4789261221885681,
      "learning_rate": 4.772666613761252e-06,
      "loss": 0.3535,
      "step": 34268
    },
    {
      "epoch": 1.5726217245651875,
      "grad_norm": 0.46552324295043945,
      "learning_rate": 4.772421679687216e-06,
      "loss": 0.3398,
      "step": 34269
    },
    {
      "epoch": 1.572667615070442,
      "grad_norm": 0.4678603708744049,
      "learning_rate": 4.772176746160435e-06,
      "loss": 0.329,
      "step": 34270
    },
    {
      "epoch": 1.5727135055756962,
      "grad_norm": 0.4397526681423187,
      "learning_rate": 4.771931813181495e-06,
      "loss": 0.3033,
      "step": 34271
    },
    {
      "epoch": 1.5727593960809507,
      "grad_norm": 0.4484117031097412,
      "learning_rate": 4.771686880750987e-06,
      "loss": 0.3258,
      "step": 34272
    },
    {
      "epoch": 1.5728052865862052,
      "grad_norm": 0.4853927791118622,
      "learning_rate": 4.771441948869502e-06,
      "loss": 0.3384,
      "step": 34273
    },
    {
      "epoch": 1.5728511770914597,
      "grad_norm": 0.45818454027175903,
      "learning_rate": 4.771197017537625e-06,
      "loss": 0.3108,
      "step": 34274
    },
    {
      "epoch": 1.5728970675967142,
      "grad_norm": 0.4507461488246918,
      "learning_rate": 4.770952086755949e-06,
      "loss": 0.2873,
      "step": 34275
    },
    {
      "epoch": 1.5729429581019687,
      "grad_norm": 0.44860732555389404,
      "learning_rate": 4.7707071565250605e-06,
      "loss": 0.3003,
      "step": 34276
    },
    {
      "epoch": 1.5729888486072232,
      "grad_norm": 0.5549141764640808,
      "learning_rate": 4.770462226845547e-06,
      "loss": 0.3647,
      "step": 34277
    },
    {
      "epoch": 1.5730347391124777,
      "grad_norm": 0.4198443591594696,
      "learning_rate": 4.770217297718e-06,
      "loss": 0.2805,
      "step": 34278
    },
    {
      "epoch": 1.5730806296177322,
      "grad_norm": 0.46926406025886536,
      "learning_rate": 4.769972369143011e-06,
      "loss": 0.3821,
      "step": 34279
    },
    {
      "epoch": 1.5731265201229867,
      "grad_norm": 0.4389936029911041,
      "learning_rate": 4.7697274411211615e-06,
      "loss": 0.341,
      "step": 34280
    },
    {
      "epoch": 1.573172410628241,
      "grad_norm": 0.46562328934669495,
      "learning_rate": 4.769482513653047e-06,
      "loss": 0.3503,
      "step": 34281
    },
    {
      "epoch": 1.5732183011334955,
      "grad_norm": 0.4734017848968506,
      "learning_rate": 4.7692375867392545e-06,
      "loss": 0.3886,
      "step": 34282
    },
    {
      "epoch": 1.57326419163875,
      "grad_norm": 0.5125962495803833,
      "learning_rate": 4.7689926603803725e-06,
      "loss": 0.3927,
      "step": 34283
    },
    {
      "epoch": 1.5733100821440043,
      "grad_norm": 0.4991873502731323,
      "learning_rate": 4.76874773457699e-06,
      "loss": 0.4365,
      "step": 34284
    },
    {
      "epoch": 1.5733559726492587,
      "grad_norm": 0.4640202522277832,
      "learning_rate": 4.768502809329697e-06,
      "loss": 0.3498,
      "step": 34285
    },
    {
      "epoch": 1.5734018631545132,
      "grad_norm": 0.43064460158348083,
      "learning_rate": 4.7682578846390794e-06,
      "loss": 0.3084,
      "step": 34286
    },
    {
      "epoch": 1.5734477536597677,
      "grad_norm": 0.5220872759819031,
      "learning_rate": 4.76801296050573e-06,
      "loss": 0.4229,
      "step": 34287
    },
    {
      "epoch": 1.5734936441650222,
      "grad_norm": 0.5002781748771667,
      "learning_rate": 4.767768036930236e-06,
      "loss": 0.4249,
      "step": 34288
    },
    {
      "epoch": 1.5735395346702767,
      "grad_norm": 0.4616413116455078,
      "learning_rate": 4.767523113913188e-06,
      "loss": 0.3422,
      "step": 34289
    },
    {
      "epoch": 1.5735854251755312,
      "grad_norm": 0.46458911895751953,
      "learning_rate": 4.767278191455172e-06,
      "loss": 0.3258,
      "step": 34290
    },
    {
      "epoch": 1.5736313156807857,
      "grad_norm": 0.4459450840950012,
      "learning_rate": 4.767033269556779e-06,
      "loss": 0.3385,
      "step": 34291
    },
    {
      "epoch": 1.5736772061860402,
      "grad_norm": 0.4159854054450989,
      "learning_rate": 4.7667883482185975e-06,
      "loss": 0.2662,
      "step": 34292
    },
    {
      "epoch": 1.5737230966912947,
      "grad_norm": 0.4617675542831421,
      "learning_rate": 4.7665434274412145e-06,
      "loss": 0.327,
      "step": 34293
    },
    {
      "epoch": 1.573768987196549,
      "grad_norm": 0.49081090092658997,
      "learning_rate": 4.766298507225223e-06,
      "loss": 0.3825,
      "step": 34294
    },
    {
      "epoch": 1.5738148777018035,
      "grad_norm": 0.45218536257743835,
      "learning_rate": 4.7660535875712105e-06,
      "loss": 0.3218,
      "step": 34295
    },
    {
      "epoch": 1.573860768207058,
      "grad_norm": 0.4742080271244049,
      "learning_rate": 4.7658086684797636e-06,
      "loss": 0.3635,
      "step": 34296
    },
    {
      "epoch": 1.5739066587123123,
      "grad_norm": 0.4839315712451935,
      "learning_rate": 4.765563749951474e-06,
      "loss": 0.3748,
      "step": 34297
    },
    {
      "epoch": 1.5739525492175668,
      "grad_norm": 0.5110636353492737,
      "learning_rate": 4.765318831986929e-06,
      "loss": 0.3995,
      "step": 34298
    },
    {
      "epoch": 1.5739984397228213,
      "grad_norm": 0.4119517207145691,
      "learning_rate": 4.765073914586717e-06,
      "loss": 0.2342,
      "step": 34299
    },
    {
      "epoch": 1.5740443302280758,
      "grad_norm": 0.4538610279560089,
      "learning_rate": 4.76482899775143e-06,
      "loss": 0.3613,
      "step": 34300
    },
    {
      "epoch": 1.5740902207333303,
      "grad_norm": 0.4636071026325226,
      "learning_rate": 4.764584081481655e-06,
      "loss": 0.3573,
      "step": 34301
    },
    {
      "epoch": 1.5741361112385848,
      "grad_norm": 0.47652003169059753,
      "learning_rate": 4.764339165777979e-06,
      "loss": 0.342,
      "step": 34302
    },
    {
      "epoch": 1.5741820017438393,
      "grad_norm": 0.4804535210132599,
      "learning_rate": 4.7640942506409955e-06,
      "loss": 0.3357,
      "step": 34303
    },
    {
      "epoch": 1.5742278922490938,
      "grad_norm": 0.47071439027786255,
      "learning_rate": 4.76384933607129e-06,
      "loss": 0.3578,
      "step": 34304
    },
    {
      "epoch": 1.5742737827543483,
      "grad_norm": 0.46876317262649536,
      "learning_rate": 4.763604422069451e-06,
      "loss": 0.3319,
      "step": 34305
    },
    {
      "epoch": 1.5743196732596025,
      "grad_norm": 0.48963093757629395,
      "learning_rate": 4.763359508636071e-06,
      "loss": 0.3898,
      "step": 34306
    },
    {
      "epoch": 1.574365563764857,
      "grad_norm": 0.5622459053993225,
      "learning_rate": 4.763114595771736e-06,
      "loss": 0.4548,
      "step": 34307
    },
    {
      "epoch": 1.5744114542701115,
      "grad_norm": 0.491474986076355,
      "learning_rate": 4.762869683477033e-06,
      "loss": 0.354,
      "step": 34308
    },
    {
      "epoch": 1.574457344775366,
      "grad_norm": 0.42723217606544495,
      "learning_rate": 4.762624771752556e-06,
      "loss": 0.2997,
      "step": 34309
    },
    {
      "epoch": 1.5745032352806203,
      "grad_norm": 0.475632905960083,
      "learning_rate": 4.762379860598892e-06,
      "loss": 0.3394,
      "step": 34310
    },
    {
      "epoch": 1.5745491257858748,
      "grad_norm": 0.47174015641212463,
      "learning_rate": 4.762134950016628e-06,
      "loss": 0.355,
      "step": 34311
    },
    {
      "epoch": 1.5745950162911293,
      "grad_norm": 0.44658371806144714,
      "learning_rate": 4.761890040006355e-06,
      "loss": 0.3171,
      "step": 34312
    },
    {
      "epoch": 1.5746409067963838,
      "grad_norm": 0.48598629236221313,
      "learning_rate": 4.761645130568663e-06,
      "loss": 0.3298,
      "step": 34313
    },
    {
      "epoch": 1.5746867973016383,
      "grad_norm": 0.49269095063209534,
      "learning_rate": 4.761400221704135e-06,
      "loss": 0.3751,
      "step": 34314
    },
    {
      "epoch": 1.5747326878068928,
      "grad_norm": 0.5349605083465576,
      "learning_rate": 4.761155313413367e-06,
      "loss": 0.4361,
      "step": 34315
    },
    {
      "epoch": 1.5747785783121473,
      "grad_norm": 0.498081237077713,
      "learning_rate": 4.7609104056969455e-06,
      "loss": 0.429,
      "step": 34316
    },
    {
      "epoch": 1.5748244688174018,
      "grad_norm": 0.4983614683151245,
      "learning_rate": 4.760665498555457e-06,
      "loss": 0.3954,
      "step": 34317
    },
    {
      "epoch": 1.5748703593226563,
      "grad_norm": 0.4624408483505249,
      "learning_rate": 4.760420591989495e-06,
      "loss": 0.3643,
      "step": 34318
    },
    {
      "epoch": 1.5749162498279106,
      "grad_norm": 0.488615483045578,
      "learning_rate": 4.760175685999644e-06,
      "loss": 0.3905,
      "step": 34319
    },
    {
      "epoch": 1.574962140333165,
      "grad_norm": 0.43220335245132446,
      "learning_rate": 4.759930780586494e-06,
      "loss": 0.2832,
      "step": 34320
    },
    {
      "epoch": 1.5750080308384196,
      "grad_norm": 0.478853702545166,
      "learning_rate": 4.759685875750637e-06,
      "loss": 0.3576,
      "step": 34321
    },
    {
      "epoch": 1.5750539213436738,
      "grad_norm": 0.4924907386302948,
      "learning_rate": 4.759440971492659e-06,
      "loss": 0.4253,
      "step": 34322
    },
    {
      "epoch": 1.5750998118489283,
      "grad_norm": 0.47425708174705505,
      "learning_rate": 4.759196067813151e-06,
      "loss": 0.3665,
      "step": 34323
    },
    {
      "epoch": 1.5751457023541828,
      "grad_norm": 0.4653985798358917,
      "learning_rate": 4.758951164712697e-06,
      "loss": 0.3873,
      "step": 34324
    },
    {
      "epoch": 1.5751915928594373,
      "grad_norm": 0.4825315475463867,
      "learning_rate": 4.758706262191892e-06,
      "loss": 0.3667,
      "step": 34325
    },
    {
      "epoch": 1.5752374833646918,
      "grad_norm": 0.4883703887462616,
      "learning_rate": 4.758461360251321e-06,
      "loss": 0.3823,
      "step": 34326
    },
    {
      "epoch": 1.5752833738699463,
      "grad_norm": 0.4457126557826996,
      "learning_rate": 4.758216458891574e-06,
      "loss": 0.3409,
      "step": 34327
    },
    {
      "epoch": 1.5753292643752008,
      "grad_norm": 0.44325193762779236,
      "learning_rate": 4.757971558113241e-06,
      "loss": 0.2965,
      "step": 34328
    },
    {
      "epoch": 1.5753751548804553,
      "grad_norm": 0.44058743119239807,
      "learning_rate": 4.7577266579169104e-06,
      "loss": 0.3097,
      "step": 34329
    },
    {
      "epoch": 1.5754210453857098,
      "grad_norm": 0.4704897105693817,
      "learning_rate": 4.757481758303168e-06,
      "loss": 0.401,
      "step": 34330
    },
    {
      "epoch": 1.5754669358909643,
      "grad_norm": 0.4642285406589508,
      "learning_rate": 4.757236859272609e-06,
      "loss": 0.3256,
      "step": 34331
    },
    {
      "epoch": 1.5755128263962186,
      "grad_norm": 0.4991973340511322,
      "learning_rate": 4.756991960825817e-06,
      "loss": 0.2969,
      "step": 34332
    },
    {
      "epoch": 1.575558716901473,
      "grad_norm": 0.45864588022232056,
      "learning_rate": 4.756747062963382e-06,
      "loss": 0.3523,
      "step": 34333
    },
    {
      "epoch": 1.5756046074067276,
      "grad_norm": 0.47496944665908813,
      "learning_rate": 4.756502165685895e-06,
      "loss": 0.3945,
      "step": 34334
    },
    {
      "epoch": 1.5756504979119819,
      "grad_norm": 0.4338351786136627,
      "learning_rate": 4.756257268993943e-06,
      "loss": 0.2993,
      "step": 34335
    },
    {
      "epoch": 1.5756963884172364,
      "grad_norm": 0.5312501192092896,
      "learning_rate": 4.7560123728881135e-06,
      "loss": 0.3927,
      "step": 34336
    },
    {
      "epoch": 1.5757422789224909,
      "grad_norm": 0.5302611589431763,
      "learning_rate": 4.755767477369e-06,
      "loss": 0.4067,
      "step": 34337
    },
    {
      "epoch": 1.5757881694277454,
      "grad_norm": 0.485738068819046,
      "learning_rate": 4.755522582437187e-06,
      "loss": 0.3868,
      "step": 34338
    },
    {
      "epoch": 1.5758340599329999,
      "grad_norm": 0.477957546710968,
      "learning_rate": 4.7552776880932655e-06,
      "loss": 0.3648,
      "step": 34339
    },
    {
      "epoch": 1.5758799504382544,
      "grad_norm": 0.49350807070732117,
      "learning_rate": 4.755032794337824e-06,
      "loss": 0.3746,
      "step": 34340
    },
    {
      "epoch": 1.5759258409435088,
      "grad_norm": 0.44400399923324585,
      "learning_rate": 4.754787901171453e-06,
      "loss": 0.3179,
      "step": 34341
    },
    {
      "epoch": 1.5759717314487633,
      "grad_norm": 0.4642792344093323,
      "learning_rate": 4.754543008594736e-06,
      "loss": 0.3393,
      "step": 34342
    },
    {
      "epoch": 1.5760176219540178,
      "grad_norm": 0.4435717463493347,
      "learning_rate": 4.754298116608268e-06,
      "loss": 0.3187,
      "step": 34343
    },
    {
      "epoch": 1.5760635124592723,
      "grad_norm": 0.44980141520500183,
      "learning_rate": 4.754053225212636e-06,
      "loss": 0.2971,
      "step": 34344
    },
    {
      "epoch": 1.5761094029645266,
      "grad_norm": 0.5081884264945984,
      "learning_rate": 4.753808334408427e-06,
      "loss": 0.3972,
      "step": 34345
    },
    {
      "epoch": 1.5761552934697811,
      "grad_norm": 0.5386309623718262,
      "learning_rate": 4.753563444196232e-06,
      "loss": 0.4065,
      "step": 34346
    },
    {
      "epoch": 1.5762011839750356,
      "grad_norm": 0.48380935192108154,
      "learning_rate": 4.753318554576639e-06,
      "loss": 0.4014,
      "step": 34347
    },
    {
      "epoch": 1.57624707448029,
      "grad_norm": 0.4552316963672638,
      "learning_rate": 4.753073665550236e-06,
      "loss": 0.3113,
      "step": 34348
    },
    {
      "epoch": 1.5762929649855444,
      "grad_norm": 0.4361666142940521,
      "learning_rate": 4.752828777117614e-06,
      "loss": 0.3196,
      "step": 34349
    },
    {
      "epoch": 1.5763388554907989,
      "grad_norm": 0.4735325574874878,
      "learning_rate": 4.752583889279361e-06,
      "loss": 0.3143,
      "step": 34350
    },
    {
      "epoch": 1.5763847459960534,
      "grad_norm": 0.4618490934371948,
      "learning_rate": 4.752339002036063e-06,
      "loss": 0.3311,
      "step": 34351
    },
    {
      "epoch": 1.5764306365013079,
      "grad_norm": 0.4503174126148224,
      "learning_rate": 4.752094115388315e-06,
      "loss": 0.3268,
      "step": 34352
    },
    {
      "epoch": 1.5764765270065624,
      "grad_norm": 0.447312593460083,
      "learning_rate": 4.751849229336701e-06,
      "loss": 0.326,
      "step": 34353
    },
    {
      "epoch": 1.5765224175118169,
      "grad_norm": 0.4811963438987732,
      "learning_rate": 4.751604343881811e-06,
      "loss": 0.3692,
      "step": 34354
    },
    {
      "epoch": 1.5765683080170714,
      "grad_norm": 0.4374050199985504,
      "learning_rate": 4.751359459024235e-06,
      "loss": 0.3211,
      "step": 34355
    },
    {
      "epoch": 1.5766141985223259,
      "grad_norm": 0.44680067896842957,
      "learning_rate": 4.7511145747645614e-06,
      "loss": 0.3461,
      "step": 34356
    },
    {
      "epoch": 1.5766600890275801,
      "grad_norm": 0.4474058449268341,
      "learning_rate": 4.750869691103378e-06,
      "loss": 0.3348,
      "step": 34357
    },
    {
      "epoch": 1.5767059795328346,
      "grad_norm": 0.5072599649429321,
      "learning_rate": 4.750624808041273e-06,
      "loss": 0.4137,
      "step": 34358
    },
    {
      "epoch": 1.5767518700380891,
      "grad_norm": 0.5107139945030212,
      "learning_rate": 4.750379925578838e-06,
      "loss": 0.4494,
      "step": 34359
    },
    {
      "epoch": 1.5767977605433434,
      "grad_norm": 0.49037083983421326,
      "learning_rate": 4.750135043716661e-06,
      "loss": 0.3716,
      "step": 34360
    },
    {
      "epoch": 1.576843651048598,
      "grad_norm": 0.460433691740036,
      "learning_rate": 4.74989016245533e-06,
      "loss": 0.3412,
      "step": 34361
    },
    {
      "epoch": 1.5768895415538524,
      "grad_norm": 0.48525089025497437,
      "learning_rate": 4.749645281795434e-06,
      "loss": 0.4065,
      "step": 34362
    },
    {
      "epoch": 1.576935432059107,
      "grad_norm": 0.4107433259487152,
      "learning_rate": 4.749400401737562e-06,
      "loss": 0.2415,
      "step": 34363
    },
    {
      "epoch": 1.5769813225643614,
      "grad_norm": 0.4531669020652771,
      "learning_rate": 4.749155522282301e-06,
      "loss": 0.351,
      "step": 34364
    },
    {
      "epoch": 1.577027213069616,
      "grad_norm": 0.44392192363739014,
      "learning_rate": 4.748910643430245e-06,
      "loss": 0.3266,
      "step": 34365
    },
    {
      "epoch": 1.5770731035748704,
      "grad_norm": 0.4626510441303253,
      "learning_rate": 4.748665765181979e-06,
      "loss": 0.3606,
      "step": 34366
    },
    {
      "epoch": 1.577118994080125,
      "grad_norm": 0.4385238289833069,
      "learning_rate": 4.7484208875380914e-06,
      "loss": 0.3202,
      "step": 34367
    },
    {
      "epoch": 1.5771648845853794,
      "grad_norm": 0.4789722263813019,
      "learning_rate": 4.748176010499173e-06,
      "loss": 0.3724,
      "step": 34368
    },
    {
      "epoch": 1.577210775090634,
      "grad_norm": 0.45046213269233704,
      "learning_rate": 4.7479311340658115e-06,
      "loss": 0.3102,
      "step": 34369
    },
    {
      "epoch": 1.5772566655958882,
      "grad_norm": 0.45858317613601685,
      "learning_rate": 4.747686258238597e-06,
      "loss": 0.3367,
      "step": 34370
    },
    {
      "epoch": 1.5773025561011427,
      "grad_norm": 0.4763321876525879,
      "learning_rate": 4.747441383018116e-06,
      "loss": 0.3392,
      "step": 34371
    },
    {
      "epoch": 1.5773484466063972,
      "grad_norm": 0.4977017343044281,
      "learning_rate": 4.74719650840496e-06,
      "loss": 0.3626,
      "step": 34372
    },
    {
      "epoch": 1.5773943371116514,
      "grad_norm": 0.4347671866416931,
      "learning_rate": 4.746951634399714e-06,
      "loss": 0.3195,
      "step": 34373
    },
    {
      "epoch": 1.577440227616906,
      "grad_norm": 0.4792231619358063,
      "learning_rate": 4.746706761002972e-06,
      "loss": 0.3731,
      "step": 34374
    },
    {
      "epoch": 1.5774861181221604,
      "grad_norm": 0.43972858786582947,
      "learning_rate": 4.74646188821532e-06,
      "loss": 0.32,
      "step": 34375
    },
    {
      "epoch": 1.577532008627415,
      "grad_norm": 0.45526033639907837,
      "learning_rate": 4.746217016037346e-06,
      "loss": 0.3119,
      "step": 34376
    },
    {
      "epoch": 1.5775778991326694,
      "grad_norm": 0.42201414704322815,
      "learning_rate": 4.745972144469641e-06,
      "loss": 0.3001,
      "step": 34377
    },
    {
      "epoch": 1.577623789637924,
      "grad_norm": 0.43599236011505127,
      "learning_rate": 4.745727273512793e-06,
      "loss": 0.2981,
      "step": 34378
    },
    {
      "epoch": 1.5776696801431784,
      "grad_norm": 0.4688419997692108,
      "learning_rate": 4.745482403167389e-06,
      "loss": 0.3361,
      "step": 34379
    },
    {
      "epoch": 1.577715570648433,
      "grad_norm": 0.469624400138855,
      "learning_rate": 4.745237533434021e-06,
      "loss": 0.3369,
      "step": 34380
    },
    {
      "epoch": 1.5777614611536874,
      "grad_norm": 0.5070335268974304,
      "learning_rate": 4.744992664313277e-06,
      "loss": 0.443,
      "step": 34381
    },
    {
      "epoch": 1.577807351658942,
      "grad_norm": 0.4825596809387207,
      "learning_rate": 4.744747795805744e-06,
      "loss": 0.3459,
      "step": 34382
    },
    {
      "epoch": 1.5778532421641962,
      "grad_norm": 0.47976213693618774,
      "learning_rate": 4.7445029279120124e-06,
      "loss": 0.3265,
      "step": 34383
    },
    {
      "epoch": 1.5778991326694507,
      "grad_norm": 0.4541701376438141,
      "learning_rate": 4.744258060632671e-06,
      "loss": 0.3665,
      "step": 34384
    },
    {
      "epoch": 1.5779450231747052,
      "grad_norm": 0.4815753400325775,
      "learning_rate": 4.744013193968306e-06,
      "loss": 0.3918,
      "step": 34385
    },
    {
      "epoch": 1.5779909136799595,
      "grad_norm": 0.5039643049240112,
      "learning_rate": 4.74376832791951e-06,
      "loss": 0.3702,
      "step": 34386
    },
    {
      "epoch": 1.578036804185214,
      "grad_norm": 0.46420979499816895,
      "learning_rate": 4.743523462486871e-06,
      "loss": 0.3417,
      "step": 34387
    },
    {
      "epoch": 1.5780826946904685,
      "grad_norm": 0.4726588726043701,
      "learning_rate": 4.743278597670978e-06,
      "loss": 0.3472,
      "step": 34388
    },
    {
      "epoch": 1.578128585195723,
      "grad_norm": 0.44611823558807373,
      "learning_rate": 4.743033733472416e-06,
      "loss": 0.3012,
      "step": 34389
    },
    {
      "epoch": 1.5781744757009775,
      "grad_norm": 0.48438596725463867,
      "learning_rate": 4.74278886989178e-06,
      "loss": 0.3762,
      "step": 34390
    },
    {
      "epoch": 1.578220366206232,
      "grad_norm": 0.42503178119659424,
      "learning_rate": 4.742544006929654e-06,
      "loss": 0.3053,
      "step": 34391
    },
    {
      "epoch": 1.5782662567114865,
      "grad_norm": 0.46989893913269043,
      "learning_rate": 4.742299144586628e-06,
      "loss": 0.3484,
      "step": 34392
    },
    {
      "epoch": 1.578312147216741,
      "grad_norm": 0.45106247067451477,
      "learning_rate": 4.742054282863292e-06,
      "loss": 0.2922,
      "step": 34393
    },
    {
      "epoch": 1.5783580377219955,
      "grad_norm": 0.519317626953125,
      "learning_rate": 4.7418094217602335e-06,
      "loss": 0.4764,
      "step": 34394
    },
    {
      "epoch": 1.5784039282272497,
      "grad_norm": 0.4333176016807556,
      "learning_rate": 4.74156456127804e-06,
      "loss": 0.311,
      "step": 34395
    },
    {
      "epoch": 1.5784498187325042,
      "grad_norm": 0.4908159077167511,
      "learning_rate": 4.741319701417305e-06,
      "loss": 0.3912,
      "step": 34396
    },
    {
      "epoch": 1.5784957092377587,
      "grad_norm": 0.47922369837760925,
      "learning_rate": 4.741074842178614e-06,
      "loss": 0.4072,
      "step": 34397
    },
    {
      "epoch": 1.5785415997430132,
      "grad_norm": 0.44962525367736816,
      "learning_rate": 4.740829983562556e-06,
      "loss": 0.3539,
      "step": 34398
    },
    {
      "epoch": 1.5785874902482675,
      "grad_norm": 0.45752912759780884,
      "learning_rate": 4.7405851255697196e-06,
      "loss": 0.3128,
      "step": 34399
    },
    {
      "epoch": 1.578633380753522,
      "grad_norm": 0.4441259205341339,
      "learning_rate": 4.740340268200695e-06,
      "loss": 0.2774,
      "step": 34400
    },
    {
      "epoch": 1.5786792712587765,
      "grad_norm": 0.4637984037399292,
      "learning_rate": 4.740095411456068e-06,
      "loss": 0.3197,
      "step": 34401
    },
    {
      "epoch": 1.578725161764031,
      "grad_norm": 0.490584135055542,
      "learning_rate": 4.739850555336432e-06,
      "loss": 0.3632,
      "step": 34402
    },
    {
      "epoch": 1.5787710522692855,
      "grad_norm": 0.48650720715522766,
      "learning_rate": 4.7396056998423725e-06,
      "loss": 0.4131,
      "step": 34403
    },
    {
      "epoch": 1.57881694277454,
      "grad_norm": 0.47464945912361145,
      "learning_rate": 4.739360844974478e-06,
      "loss": 0.3429,
      "step": 34404
    },
    {
      "epoch": 1.5788628332797945,
      "grad_norm": 0.48285701870918274,
      "learning_rate": 4.73911599073334e-06,
      "loss": 0.3741,
      "step": 34405
    },
    {
      "epoch": 1.578908723785049,
      "grad_norm": 0.4599708318710327,
      "learning_rate": 4.7388711371195454e-06,
      "loss": 0.3155,
      "step": 34406
    },
    {
      "epoch": 1.5789546142903035,
      "grad_norm": 0.4671148955821991,
      "learning_rate": 4.738626284133682e-06,
      "loss": 0.3523,
      "step": 34407
    },
    {
      "epoch": 1.5790005047955578,
      "grad_norm": 0.5202903151512146,
      "learning_rate": 4.738381431776341e-06,
      "loss": 0.4559,
      "step": 34408
    },
    {
      "epoch": 1.5790463953008123,
      "grad_norm": 0.5136663913726807,
      "learning_rate": 4.73813658004811e-06,
      "loss": 0.4289,
      "step": 34409
    },
    {
      "epoch": 1.5790922858060668,
      "grad_norm": 0.48985564708709717,
      "learning_rate": 4.737891728949578e-06,
      "loss": 0.3463,
      "step": 34410
    },
    {
      "epoch": 1.579138176311321,
      "grad_norm": 0.49128296971321106,
      "learning_rate": 4.7376468784813334e-06,
      "loss": 0.3787,
      "step": 34411
    },
    {
      "epoch": 1.5791840668165755,
      "grad_norm": 0.44338926672935486,
      "learning_rate": 4.737402028643966e-06,
      "loss": 0.3158,
      "step": 34412
    },
    {
      "epoch": 1.57922995732183,
      "grad_norm": 0.47099071741104126,
      "learning_rate": 4.737157179438062e-06,
      "loss": 0.3998,
      "step": 34413
    },
    {
      "epoch": 1.5792758478270845,
      "grad_norm": 0.461872398853302,
      "learning_rate": 4.736912330864214e-06,
      "loss": 0.3558,
      "step": 34414
    },
    {
      "epoch": 1.579321738332339,
      "grad_norm": 0.482786625623703,
      "learning_rate": 4.7366674829230096e-06,
      "loss": 0.4006,
      "step": 34415
    },
    {
      "epoch": 1.5793676288375935,
      "grad_norm": 0.4322293698787689,
      "learning_rate": 4.7364226356150325e-06,
      "loss": 0.3336,
      "step": 34416
    },
    {
      "epoch": 1.579413519342848,
      "grad_norm": 0.4366132616996765,
      "learning_rate": 4.736177788940879e-06,
      "loss": 0.2899,
      "step": 34417
    },
    {
      "epoch": 1.5794594098481025,
      "grad_norm": 0.48266324400901794,
      "learning_rate": 4.735932942901136e-06,
      "loss": 0.3877,
      "step": 34418
    },
    {
      "epoch": 1.579505300353357,
      "grad_norm": 0.44812893867492676,
      "learning_rate": 4.735688097496387e-06,
      "loss": 0.2957,
      "step": 34419
    },
    {
      "epoch": 1.5795511908586115,
      "grad_norm": 0.4843303859233856,
      "learning_rate": 4.735443252727228e-06,
      "loss": 0.3638,
      "step": 34420
    },
    {
      "epoch": 1.5795970813638658,
      "grad_norm": 0.4521855413913727,
      "learning_rate": 4.735198408594244e-06,
      "loss": 0.2963,
      "step": 34421
    },
    {
      "epoch": 1.5796429718691203,
      "grad_norm": 0.6072704195976257,
      "learning_rate": 4.734953565098024e-06,
      "loss": 0.4828,
      "step": 34422
    },
    {
      "epoch": 1.5796888623743748,
      "grad_norm": 0.4128320813179016,
      "learning_rate": 4.7347087222391545e-06,
      "loss": 0.2606,
      "step": 34423
    },
    {
      "epoch": 1.579734752879629,
      "grad_norm": 0.49240946769714355,
      "learning_rate": 4.734463880018229e-06,
      "loss": 0.4032,
      "step": 34424
    },
    {
      "epoch": 1.5797806433848836,
      "grad_norm": 0.4533618688583374,
      "learning_rate": 4.734219038435835e-06,
      "loss": 0.3007,
      "step": 34425
    },
    {
      "epoch": 1.579826533890138,
      "grad_norm": 0.4841307997703552,
      "learning_rate": 4.733974197492558e-06,
      "loss": 0.355,
      "step": 34426
    },
    {
      "epoch": 1.5798724243953925,
      "grad_norm": 0.4938371181488037,
      "learning_rate": 4.7337293571889914e-06,
      "loss": 0.3939,
      "step": 34427
    },
    {
      "epoch": 1.579918314900647,
      "grad_norm": 0.46043580770492554,
      "learning_rate": 4.733484517525721e-06,
      "loss": 0.333,
      "step": 34428
    },
    {
      "epoch": 1.5799642054059015,
      "grad_norm": 0.4985210597515106,
      "learning_rate": 4.733239678503334e-06,
      "loss": 0.4072,
      "step": 34429
    },
    {
      "epoch": 1.580010095911156,
      "grad_norm": 0.45864444971084595,
      "learning_rate": 4.732994840122424e-06,
      "loss": 0.3443,
      "step": 34430
    },
    {
      "epoch": 1.5800559864164105,
      "grad_norm": 0.47806692123413086,
      "learning_rate": 4.732750002383577e-06,
      "loss": 0.3439,
      "step": 34431
    },
    {
      "epoch": 1.580101876921665,
      "grad_norm": 0.44951456785202026,
      "learning_rate": 4.7325051652873805e-06,
      "loss": 0.3151,
      "step": 34432
    },
    {
      "epoch": 1.5801477674269195,
      "grad_norm": 0.4935435354709625,
      "learning_rate": 4.732260328834426e-06,
      "loss": 0.37,
      "step": 34433
    },
    {
      "epoch": 1.5801936579321738,
      "grad_norm": 0.45162340998649597,
      "learning_rate": 4.732015493025301e-06,
      "loss": 0.3355,
      "step": 34434
    },
    {
      "epoch": 1.5802395484374283,
      "grad_norm": 0.4715065658092499,
      "learning_rate": 4.731770657860591e-06,
      "loss": 0.3739,
      "step": 34435
    },
    {
      "epoch": 1.5802854389426828,
      "grad_norm": 0.49243614077568054,
      "learning_rate": 4.731525823340893e-06,
      "loss": 0.3403,
      "step": 34436
    },
    {
      "epoch": 1.580331329447937,
      "grad_norm": 0.4590493440628052,
      "learning_rate": 4.731280989466788e-06,
      "loss": 0.3072,
      "step": 34437
    },
    {
      "epoch": 1.5803772199531916,
      "grad_norm": 0.49236413836479187,
      "learning_rate": 4.731036156238865e-06,
      "loss": 0.3745,
      "step": 34438
    },
    {
      "epoch": 1.580423110458446,
      "grad_norm": 0.46603336930274963,
      "learning_rate": 4.730791323657719e-06,
      "loss": 0.3957,
      "step": 34439
    },
    {
      "epoch": 1.5804690009637006,
      "grad_norm": 0.5645825862884521,
      "learning_rate": 4.730546491723934e-06,
      "loss": 0.4418,
      "step": 34440
    },
    {
      "epoch": 1.580514891468955,
      "grad_norm": 0.42145490646362305,
      "learning_rate": 4.730301660438099e-06,
      "loss": 0.3007,
      "step": 34441
    },
    {
      "epoch": 1.5805607819742096,
      "grad_norm": 0.47927382588386536,
      "learning_rate": 4.730056829800803e-06,
      "loss": 0.3513,
      "step": 34442
    },
    {
      "epoch": 1.580606672479464,
      "grad_norm": 0.4265795052051544,
      "learning_rate": 4.729811999812637e-06,
      "loss": 0.2761,
      "step": 34443
    },
    {
      "epoch": 1.5806525629847186,
      "grad_norm": 0.4777577519416809,
      "learning_rate": 4.729567170474185e-06,
      "loss": 0.3789,
      "step": 34444
    },
    {
      "epoch": 1.580698453489973,
      "grad_norm": 0.45921361446380615,
      "learning_rate": 4.72932234178604e-06,
      "loss": 0.3484,
      "step": 34445
    },
    {
      "epoch": 1.5807443439952273,
      "grad_norm": 0.47235363721847534,
      "learning_rate": 4.72907751374879e-06,
      "loss": 0.3943,
      "step": 34446
    },
    {
      "epoch": 1.5807902345004818,
      "grad_norm": 0.43517857789993286,
      "learning_rate": 4.728832686363023e-06,
      "loss": 0.3028,
      "step": 34447
    },
    {
      "epoch": 1.5808361250057363,
      "grad_norm": 0.5322061777114868,
      "learning_rate": 4.728587859629327e-06,
      "loss": 0.3722,
      "step": 34448
    },
    {
      "epoch": 1.5808820155109906,
      "grad_norm": 0.5091196894645691,
      "learning_rate": 4.728343033548293e-06,
      "loss": 0.2865,
      "step": 34449
    },
    {
      "epoch": 1.580927906016245,
      "grad_norm": 0.45518651604652405,
      "learning_rate": 4.728098208120505e-06,
      "loss": 0.3417,
      "step": 34450
    },
    {
      "epoch": 1.5809737965214996,
      "grad_norm": 0.4545513987541199,
      "learning_rate": 4.727853383346558e-06,
      "loss": 0.3546,
      "step": 34451
    },
    {
      "epoch": 1.581019687026754,
      "grad_norm": 0.473846971988678,
      "learning_rate": 4.727608559227037e-06,
      "loss": 0.3546,
      "step": 34452
    },
    {
      "epoch": 1.5810655775320086,
      "grad_norm": 0.49696484208106995,
      "learning_rate": 4.727363735762532e-06,
      "loss": 0.4247,
      "step": 34453
    },
    {
      "epoch": 1.581111468037263,
      "grad_norm": 0.45082351565361023,
      "learning_rate": 4.7271189129536295e-06,
      "loss": 0.3117,
      "step": 34454
    },
    {
      "epoch": 1.5811573585425176,
      "grad_norm": 0.4791335165500641,
      "learning_rate": 4.7268740908009216e-06,
      "loss": 0.3674,
      "step": 34455
    },
    {
      "epoch": 1.581203249047772,
      "grad_norm": 0.43058332800865173,
      "learning_rate": 4.726629269304995e-06,
      "loss": 0.2767,
      "step": 34456
    },
    {
      "epoch": 1.5812491395530266,
      "grad_norm": 0.4964630901813507,
      "learning_rate": 4.726384448466437e-06,
      "loss": 0.4038,
      "step": 34457
    },
    {
      "epoch": 1.581295030058281,
      "grad_norm": 0.44527408480644226,
      "learning_rate": 4.726139628285841e-06,
      "loss": 0.3645,
      "step": 34458
    },
    {
      "epoch": 1.5813409205635354,
      "grad_norm": 0.4617421627044678,
      "learning_rate": 4.725894808763791e-06,
      "loss": 0.3664,
      "step": 34459
    },
    {
      "epoch": 1.5813868110687899,
      "grad_norm": 0.5003541111946106,
      "learning_rate": 4.725649989900876e-06,
      "loss": 0.3968,
      "step": 34460
    },
    {
      "epoch": 1.5814327015740444,
      "grad_norm": 0.486860066652298,
      "learning_rate": 4.725405171697689e-06,
      "loss": 0.3814,
      "step": 34461
    },
    {
      "epoch": 1.5814785920792986,
      "grad_norm": 0.483610600233078,
      "learning_rate": 4.725160354154815e-06,
      "loss": 0.4072,
      "step": 34462
    },
    {
      "epoch": 1.5815244825845531,
      "grad_norm": 0.45767226815223694,
      "learning_rate": 4.724915537272843e-06,
      "loss": 0.3331,
      "step": 34463
    },
    {
      "epoch": 1.5815703730898076,
      "grad_norm": 0.4369322657585144,
      "learning_rate": 4.7246707210523626e-06,
      "loss": 0.322,
      "step": 34464
    },
    {
      "epoch": 1.5816162635950621,
      "grad_norm": 0.4593113958835602,
      "learning_rate": 4.724425905493963e-06,
      "loss": 0.299,
      "step": 34465
    },
    {
      "epoch": 1.5816621541003166,
      "grad_norm": 0.46761929988861084,
      "learning_rate": 4.724181090598229e-06,
      "loss": 0.3435,
      "step": 34466
    },
    {
      "epoch": 1.5817080446055711,
      "grad_norm": 0.44399577379226685,
      "learning_rate": 4.723936276365756e-06,
      "loss": 0.314,
      "step": 34467
    },
    {
      "epoch": 1.5817539351108256,
      "grad_norm": 0.479021817445755,
      "learning_rate": 4.723691462797128e-06,
      "loss": 0.37,
      "step": 34468
    },
    {
      "epoch": 1.5817998256160801,
      "grad_norm": 0.4903155565261841,
      "learning_rate": 4.723446649892934e-06,
      "loss": 0.4675,
      "step": 34469
    },
    {
      "epoch": 1.5818457161213346,
      "grad_norm": 0.45112359523773193,
      "learning_rate": 4.723201837653765e-06,
      "loss": 0.2817,
      "step": 34470
    },
    {
      "epoch": 1.5818916066265891,
      "grad_norm": 0.4757433533668518,
      "learning_rate": 4.722957026080208e-06,
      "loss": 0.3584,
      "step": 34471
    },
    {
      "epoch": 1.5819374971318434,
      "grad_norm": 0.46068274974823,
      "learning_rate": 4.72271221517285e-06,
      "loss": 0.3391,
      "step": 34472
    },
    {
      "epoch": 1.581983387637098,
      "grad_norm": 0.4295575022697449,
      "learning_rate": 4.7224674049322834e-06,
      "loss": 0.3029,
      "step": 34473
    },
    {
      "epoch": 1.5820292781423524,
      "grad_norm": 0.4634479582309723,
      "learning_rate": 4.722222595359095e-06,
      "loss": 0.3445,
      "step": 34474
    },
    {
      "epoch": 1.5820751686476067,
      "grad_norm": 0.46829524636268616,
      "learning_rate": 4.721977786453873e-06,
      "loss": 0.3813,
      "step": 34475
    },
    {
      "epoch": 1.5821210591528612,
      "grad_norm": 0.4492100179195404,
      "learning_rate": 4.721732978217207e-06,
      "loss": 0.3321,
      "step": 34476
    },
    {
      "epoch": 1.5821669496581157,
      "grad_norm": 0.4966422915458679,
      "learning_rate": 4.721488170649687e-06,
      "loss": 0.3577,
      "step": 34477
    },
    {
      "epoch": 1.5822128401633702,
      "grad_norm": 0.43524259328842163,
      "learning_rate": 4.721243363751896e-06,
      "loss": 0.2972,
      "step": 34478
    },
    {
      "epoch": 1.5822587306686247,
      "grad_norm": 0.4802364408969879,
      "learning_rate": 4.720998557524429e-06,
      "loss": 0.3793,
      "step": 34479
    },
    {
      "epoch": 1.5823046211738792,
      "grad_norm": 0.5156641602516174,
      "learning_rate": 4.720753751967875e-06,
      "loss": 0.4036,
      "step": 34480
    },
    {
      "epoch": 1.5823505116791337,
      "grad_norm": 0.483347624540329,
      "learning_rate": 4.720508947082816e-06,
      "loss": 0.3862,
      "step": 34481
    },
    {
      "epoch": 1.5823964021843882,
      "grad_norm": 0.4823780953884125,
      "learning_rate": 4.720264142869846e-06,
      "loss": 0.3742,
      "step": 34482
    },
    {
      "epoch": 1.5824422926896426,
      "grad_norm": 0.47350820899009705,
      "learning_rate": 4.720019339329554e-06,
      "loss": 0.3767,
      "step": 34483
    },
    {
      "epoch": 1.582488183194897,
      "grad_norm": 0.5104520916938782,
      "learning_rate": 4.719774536462525e-06,
      "loss": 0.4216,
      "step": 34484
    },
    {
      "epoch": 1.5825340737001514,
      "grad_norm": 0.4646252393722534,
      "learning_rate": 4.719529734269351e-06,
      "loss": 0.29,
      "step": 34485
    },
    {
      "epoch": 1.582579964205406,
      "grad_norm": 0.46553856134414673,
      "learning_rate": 4.71928493275062e-06,
      "loss": 0.3952,
      "step": 34486
    },
    {
      "epoch": 1.5826258547106604,
      "grad_norm": 0.4512983560562134,
      "learning_rate": 4.71904013190692e-06,
      "loss": 0.2926,
      "step": 34487
    },
    {
      "epoch": 1.5826717452159147,
      "grad_norm": 0.5158006548881531,
      "learning_rate": 4.718795331738838e-06,
      "loss": 0.3937,
      "step": 34488
    },
    {
      "epoch": 1.5827176357211692,
      "grad_norm": 0.4445599317550659,
      "learning_rate": 4.718550532246966e-06,
      "loss": 0.3221,
      "step": 34489
    },
    {
      "epoch": 1.5827635262264237,
      "grad_norm": 0.444424033164978,
      "learning_rate": 4.718305733431891e-06,
      "loss": 0.2861,
      "step": 34490
    },
    {
      "epoch": 1.5828094167316782,
      "grad_norm": 0.46928074955940247,
      "learning_rate": 4.718060935294202e-06,
      "loss": 0.3281,
      "step": 34491
    },
    {
      "epoch": 1.5828553072369327,
      "grad_norm": 0.4669865369796753,
      "learning_rate": 4.717816137834487e-06,
      "loss": 0.3264,
      "step": 34492
    },
    {
      "epoch": 1.5829011977421872,
      "grad_norm": 0.473935067653656,
      "learning_rate": 4.717571341053336e-06,
      "loss": 0.3441,
      "step": 34493
    },
    {
      "epoch": 1.5829470882474417,
      "grad_norm": 0.4499046504497528,
      "learning_rate": 4.717326544951334e-06,
      "loss": 0.3332,
      "step": 34494
    },
    {
      "epoch": 1.5829929787526962,
      "grad_norm": 0.4681978225708008,
      "learning_rate": 4.717081749529075e-06,
      "loss": 0.3397,
      "step": 34495
    },
    {
      "epoch": 1.5830388692579507,
      "grad_norm": 0.45647066831588745,
      "learning_rate": 4.716836954787145e-06,
      "loss": 0.302,
      "step": 34496
    },
    {
      "epoch": 1.583084759763205,
      "grad_norm": 0.43028759956359863,
      "learning_rate": 4.716592160726132e-06,
      "loss": 0.3116,
      "step": 34497
    },
    {
      "epoch": 1.5831306502684594,
      "grad_norm": 0.49799495935440063,
      "learning_rate": 4.716347367346626e-06,
      "loss": 0.4451,
      "step": 34498
    },
    {
      "epoch": 1.583176540773714,
      "grad_norm": 0.4543262720108032,
      "learning_rate": 4.716102574649215e-06,
      "loss": 0.3147,
      "step": 34499
    },
    {
      "epoch": 1.5832224312789682,
      "grad_norm": 0.4617965519428253,
      "learning_rate": 4.715857782634486e-06,
      "loss": 0.3449,
      "step": 34500
    },
    {
      "epoch": 1.5832683217842227,
      "grad_norm": 0.4661722183227539,
      "learning_rate": 4.7156129913030316e-06,
      "loss": 0.3697,
      "step": 34501
    },
    {
      "epoch": 1.5833142122894772,
      "grad_norm": 0.4829505681991577,
      "learning_rate": 4.715368200655439e-06,
      "loss": 0.3797,
      "step": 34502
    },
    {
      "epoch": 1.5833601027947317,
      "grad_norm": 0.4805639982223511,
      "learning_rate": 4.715123410692292e-06,
      "loss": 0.29,
      "step": 34503
    },
    {
      "epoch": 1.5834059932999862,
      "grad_norm": 0.43287238478660583,
      "learning_rate": 4.714878621414187e-06,
      "loss": 0.3068,
      "step": 34504
    },
    {
      "epoch": 1.5834518838052407,
      "grad_norm": 0.504483163356781,
      "learning_rate": 4.714633832821708e-06,
      "loss": 0.4195,
      "step": 34505
    },
    {
      "epoch": 1.5834977743104952,
      "grad_norm": 0.45353904366493225,
      "learning_rate": 4.714389044915443e-06,
      "loss": 0.3217,
      "step": 34506
    },
    {
      "epoch": 1.5835436648157497,
      "grad_norm": 0.5309372544288635,
      "learning_rate": 4.714144257695984e-06,
      "loss": 0.4362,
      "step": 34507
    },
    {
      "epoch": 1.5835895553210042,
      "grad_norm": 0.49125128984451294,
      "learning_rate": 4.713899471163918e-06,
      "loss": 0.4112,
      "step": 34508
    },
    {
      "epoch": 1.5836354458262587,
      "grad_norm": 0.9372045397758484,
      "learning_rate": 4.713654685319831e-06,
      "loss": 0.3898,
      "step": 34509
    },
    {
      "epoch": 1.583681336331513,
      "grad_norm": 0.4888708293437958,
      "learning_rate": 4.713409900164317e-06,
      "loss": 0.3527,
      "step": 34510
    },
    {
      "epoch": 1.5837272268367675,
      "grad_norm": 0.4270753264427185,
      "learning_rate": 4.713165115697961e-06,
      "loss": 0.3033,
      "step": 34511
    },
    {
      "epoch": 1.583773117342022,
      "grad_norm": 0.47595661878585815,
      "learning_rate": 4.712920331921351e-06,
      "loss": 0.3412,
      "step": 34512
    },
    {
      "epoch": 1.5838190078472763,
      "grad_norm": 0.4398542046546936,
      "learning_rate": 4.712675548835079e-06,
      "loss": 0.3221,
      "step": 34513
    },
    {
      "epoch": 1.5838648983525307,
      "grad_norm": 0.4746125042438507,
      "learning_rate": 4.71243076643973e-06,
      "loss": 0.3501,
      "step": 34514
    },
    {
      "epoch": 1.5839107888577852,
      "grad_norm": 0.4795666038990021,
      "learning_rate": 4.712185984735894e-06,
      "loss": 0.3811,
      "step": 34515
    },
    {
      "epoch": 1.5839566793630397,
      "grad_norm": 0.4941750168800354,
      "learning_rate": 4.711941203724162e-06,
      "loss": 0.3534,
      "step": 34516
    },
    {
      "epoch": 1.5840025698682942,
      "grad_norm": 0.4700463116168976,
      "learning_rate": 4.7116964234051185e-06,
      "loss": 0.3921,
      "step": 34517
    },
    {
      "epoch": 1.5840484603735487,
      "grad_norm": 0.4921993017196655,
      "learning_rate": 4.711451643779356e-06,
      "loss": 0.3998,
      "step": 34518
    },
    {
      "epoch": 1.5840943508788032,
      "grad_norm": 0.4448760151863098,
      "learning_rate": 4.711206864847459e-06,
      "loss": 0.3125,
      "step": 34519
    },
    {
      "epoch": 1.5841402413840577,
      "grad_norm": 0.4679434895515442,
      "learning_rate": 4.710962086610021e-06,
      "loss": 0.3365,
      "step": 34520
    },
    {
      "epoch": 1.5841861318893122,
      "grad_norm": 0.46245276927948,
      "learning_rate": 4.710717309067626e-06,
      "loss": 0.3467,
      "step": 34521
    },
    {
      "epoch": 1.5842320223945667,
      "grad_norm": 0.5030012130737305,
      "learning_rate": 4.710472532220863e-06,
      "loss": 0.3513,
      "step": 34522
    },
    {
      "epoch": 1.584277912899821,
      "grad_norm": 0.49735182523727417,
      "learning_rate": 4.710227756070325e-06,
      "loss": 0.3702,
      "step": 34523
    },
    {
      "epoch": 1.5843238034050755,
      "grad_norm": 0.450185090303421,
      "learning_rate": 4.709982980616599e-06,
      "loss": 0.3151,
      "step": 34524
    },
    {
      "epoch": 1.58436969391033,
      "grad_norm": 0.5023850798606873,
      "learning_rate": 4.709738205860268e-06,
      "loss": 0.3852,
      "step": 34525
    },
    {
      "epoch": 1.5844155844155843,
      "grad_norm": 0.4783794581890106,
      "learning_rate": 4.709493431801928e-06,
      "loss": 0.4325,
      "step": 34526
    },
    {
      "epoch": 1.5844614749208388,
      "grad_norm": 0.4516247808933258,
      "learning_rate": 4.709248658442165e-06,
      "loss": 0.3669,
      "step": 34527
    },
    {
      "epoch": 1.5845073654260933,
      "grad_norm": 0.4789596199989319,
      "learning_rate": 4.7090038857815655e-06,
      "loss": 0.351,
      "step": 34528
    },
    {
      "epoch": 1.5845532559313478,
      "grad_norm": 0.4462050199508667,
      "learning_rate": 4.70875911382072e-06,
      "loss": 0.3061,
      "step": 34529
    },
    {
      "epoch": 1.5845991464366023,
      "grad_norm": 0.45963054895401,
      "learning_rate": 4.708514342560218e-06,
      "loss": 0.3138,
      "step": 34530
    },
    {
      "epoch": 1.5846450369418568,
      "grad_norm": 0.49401775002479553,
      "learning_rate": 4.708269572000645e-06,
      "loss": 0.3855,
      "step": 34531
    },
    {
      "epoch": 1.5846909274471113,
      "grad_norm": 0.4915311336517334,
      "learning_rate": 4.708024802142593e-06,
      "loss": 0.3649,
      "step": 34532
    },
    {
      "epoch": 1.5847368179523658,
      "grad_norm": 0.5864157676696777,
      "learning_rate": 4.70778003298665e-06,
      "loss": 0.324,
      "step": 34533
    },
    {
      "epoch": 1.5847827084576203,
      "grad_norm": 0.4741992652416229,
      "learning_rate": 4.707535264533402e-06,
      "loss": 0.4122,
      "step": 34534
    },
    {
      "epoch": 1.5848285989628745,
      "grad_norm": 0.4905097484588623,
      "learning_rate": 4.707290496783441e-06,
      "loss": 0.3845,
      "step": 34535
    },
    {
      "epoch": 1.584874489468129,
      "grad_norm": 0.4716007709503174,
      "learning_rate": 4.707045729737354e-06,
      "loss": 0.3296,
      "step": 34536
    },
    {
      "epoch": 1.5849203799733835,
      "grad_norm": 0.47660380601882935,
      "learning_rate": 4.706800963395728e-06,
      "loss": 0.3862,
      "step": 34537
    },
    {
      "epoch": 1.5849662704786378,
      "grad_norm": 0.4517204165458679,
      "learning_rate": 4.706556197759154e-06,
      "loss": 0.3163,
      "step": 34538
    },
    {
      "epoch": 1.5850121609838923,
      "grad_norm": 0.43394896388053894,
      "learning_rate": 4.70631143282822e-06,
      "loss": 0.3079,
      "step": 34539
    },
    {
      "epoch": 1.5850580514891468,
      "grad_norm": 0.4525814950466156,
      "learning_rate": 4.706066668603514e-06,
      "loss": 0.3005,
      "step": 34540
    },
    {
      "epoch": 1.5851039419944013,
      "grad_norm": 0.47550779581069946,
      "learning_rate": 4.705821905085625e-06,
      "loss": 0.4208,
      "step": 34541
    },
    {
      "epoch": 1.5851498324996558,
      "grad_norm": 0.4417058527469635,
      "learning_rate": 4.705577142275142e-06,
      "loss": 0.2707,
      "step": 34542
    },
    {
      "epoch": 1.5851957230049103,
      "grad_norm": 0.47244930267333984,
      "learning_rate": 4.7053323801726505e-06,
      "loss": 0.3307,
      "step": 34543
    },
    {
      "epoch": 1.5852416135101648,
      "grad_norm": 0.4288998544216156,
      "learning_rate": 4.705087618778745e-06,
      "loss": 0.2981,
      "step": 34544
    },
    {
      "epoch": 1.5852875040154193,
      "grad_norm": 0.5020086169242859,
      "learning_rate": 4.704842858094009e-06,
      "loss": 0.3998,
      "step": 34545
    },
    {
      "epoch": 1.5853333945206738,
      "grad_norm": 0.4833795726299286,
      "learning_rate": 4.704598098119033e-06,
      "loss": 0.3512,
      "step": 34546
    },
    {
      "epoch": 1.5853792850259283,
      "grad_norm": 0.4819575548171997,
      "learning_rate": 4.704353338854406e-06,
      "loss": 0.3524,
      "step": 34547
    },
    {
      "epoch": 1.5854251755311826,
      "grad_norm": 0.47497355937957764,
      "learning_rate": 4.704108580300716e-06,
      "loss": 0.3286,
      "step": 34548
    },
    {
      "epoch": 1.585471066036437,
      "grad_norm": 0.49087902903556824,
      "learning_rate": 4.703863822458548e-06,
      "loss": 0.419,
      "step": 34549
    },
    {
      "epoch": 1.5855169565416916,
      "grad_norm": 0.48719003796577454,
      "learning_rate": 4.703619065328499e-06,
      "loss": 0.3792,
      "step": 34550
    },
    {
      "epoch": 1.5855628470469458,
      "grad_norm": 0.48690444231033325,
      "learning_rate": 4.703374308911151e-06,
      "loss": 0.3838,
      "step": 34551
    },
    {
      "epoch": 1.5856087375522003,
      "grad_norm": 0.5013701915740967,
      "learning_rate": 4.703129553207094e-06,
      "loss": 0.4197,
      "step": 34552
    },
    {
      "epoch": 1.5856546280574548,
      "grad_norm": 0.46105894446372986,
      "learning_rate": 4.702884798216915e-06,
      "loss": 0.3296,
      "step": 34553
    },
    {
      "epoch": 1.5857005185627093,
      "grad_norm": 0.4834175407886505,
      "learning_rate": 4.702640043941206e-06,
      "loss": 0.388,
      "step": 34554
    },
    {
      "epoch": 1.5857464090679638,
      "grad_norm": 0.467048317193985,
      "learning_rate": 4.702395290380554e-06,
      "loss": 0.3724,
      "step": 34555
    },
    {
      "epoch": 1.5857922995732183,
      "grad_norm": 0.5070849657058716,
      "learning_rate": 4.702150537535547e-06,
      "loss": 0.4464,
      "step": 34556
    },
    {
      "epoch": 1.5858381900784728,
      "grad_norm": 0.4568125903606415,
      "learning_rate": 4.701905785406774e-06,
      "loss": 0.3509,
      "step": 34557
    },
    {
      "epoch": 1.5858840805837273,
      "grad_norm": 0.46787765622138977,
      "learning_rate": 4.701661033994824e-06,
      "loss": 0.3284,
      "step": 34558
    },
    {
      "epoch": 1.5859299710889818,
      "grad_norm": 0.4700010418891907,
      "learning_rate": 4.701416283300283e-06,
      "loss": 0.3524,
      "step": 34559
    },
    {
      "epoch": 1.5859758615942363,
      "grad_norm": 0.46363264322280884,
      "learning_rate": 4.7011715333237436e-06,
      "loss": 0.3195,
      "step": 34560
    },
    {
      "epoch": 1.5860217520994906,
      "grad_norm": 0.5072803497314453,
      "learning_rate": 4.7009267840657915e-06,
      "loss": 0.4066,
      "step": 34561
    },
    {
      "epoch": 1.586067642604745,
      "grad_norm": 0.4677402973175049,
      "learning_rate": 4.700682035527016e-06,
      "loss": 0.3934,
      "step": 34562
    },
    {
      "epoch": 1.5861135331099996,
      "grad_norm": 0.48642536997795105,
      "learning_rate": 4.700437287708007e-06,
      "loss": 0.3319,
      "step": 34563
    },
    {
      "epoch": 1.5861594236152539,
      "grad_norm": 0.4711402654647827,
      "learning_rate": 4.700192540609351e-06,
      "loss": 0.3159,
      "step": 34564
    },
    {
      "epoch": 1.5862053141205084,
      "grad_norm": 0.4461587369441986,
      "learning_rate": 4.699947794231636e-06,
      "loss": 0.3624,
      "step": 34565
    },
    {
      "epoch": 1.5862512046257629,
      "grad_norm": 0.46647313237190247,
      "learning_rate": 4.699703048575454e-06,
      "loss": 0.3327,
      "step": 34566
    },
    {
      "epoch": 1.5862970951310174,
      "grad_norm": 0.509233295917511,
      "learning_rate": 4.699458303641391e-06,
      "loss": 0.43,
      "step": 34567
    },
    {
      "epoch": 1.5863429856362719,
      "grad_norm": 0.48189833760261536,
      "learning_rate": 4.699213559430034e-06,
      "loss": 0.3556,
      "step": 34568
    },
    {
      "epoch": 1.5863888761415263,
      "grad_norm": 0.4503464698791504,
      "learning_rate": 4.6989688159419755e-06,
      "loss": 0.3184,
      "step": 34569
    },
    {
      "epoch": 1.5864347666467808,
      "grad_norm": 0.5226192474365234,
      "learning_rate": 4.6987240731778015e-06,
      "loss": 0.3992,
      "step": 34570
    },
    {
      "epoch": 1.5864806571520353,
      "grad_norm": 0.4482278823852539,
      "learning_rate": 4.698479331138099e-06,
      "loss": 0.3264,
      "step": 34571
    },
    {
      "epoch": 1.5865265476572898,
      "grad_norm": 0.5040515065193176,
      "learning_rate": 4.698234589823463e-06,
      "loss": 0.3854,
      "step": 34572
    },
    {
      "epoch": 1.5865724381625441,
      "grad_norm": 0.5213207602500916,
      "learning_rate": 4.697989849234475e-06,
      "loss": 0.4311,
      "step": 34573
    },
    {
      "epoch": 1.5866183286677986,
      "grad_norm": 0.4529222548007965,
      "learning_rate": 4.697745109371724e-06,
      "loss": 0.3559,
      "step": 34574
    },
    {
      "epoch": 1.5866642191730531,
      "grad_norm": 0.49292704463005066,
      "learning_rate": 4.697500370235804e-06,
      "loss": 0.4109,
      "step": 34575
    },
    {
      "epoch": 1.5867101096783076,
      "grad_norm": 0.4504312574863434,
      "learning_rate": 4.697255631827299e-06,
      "loss": 0.3267,
      "step": 34576
    },
    {
      "epoch": 1.5867560001835619,
      "grad_norm": 0.4388074278831482,
      "learning_rate": 4.697010894146798e-06,
      "loss": 0.2933,
      "step": 34577
    },
    {
      "epoch": 1.5868018906888164,
      "grad_norm": 0.46784502267837524,
      "learning_rate": 4.696766157194891e-06,
      "loss": 0.3368,
      "step": 34578
    },
    {
      "epoch": 1.5868477811940709,
      "grad_norm": 0.4724850654602051,
      "learning_rate": 4.696521420972165e-06,
      "loss": 0.3367,
      "step": 34579
    },
    {
      "epoch": 1.5868936716993254,
      "grad_norm": 0.496127724647522,
      "learning_rate": 4.696276685479208e-06,
      "loss": 0.4084,
      "step": 34580
    },
    {
      "epoch": 1.5869395622045799,
      "grad_norm": 0.4329172968864441,
      "learning_rate": 4.696031950716612e-06,
      "loss": 0.2939,
      "step": 34581
    },
    {
      "epoch": 1.5869854527098344,
      "grad_norm": 0.4585079252719879,
      "learning_rate": 4.6957872166849636e-06,
      "loss": 0.3265,
      "step": 34582
    },
    {
      "epoch": 1.5870313432150889,
      "grad_norm": 0.4721071124076843,
      "learning_rate": 4.695542483384849e-06,
      "loss": 0.3711,
      "step": 34583
    },
    {
      "epoch": 1.5870772337203434,
      "grad_norm": 0.47259363532066345,
      "learning_rate": 4.695297750816859e-06,
      "loss": 0.3584,
      "step": 34584
    },
    {
      "epoch": 1.5871231242255979,
      "grad_norm": 0.4456488788127899,
      "learning_rate": 4.695053018981584e-06,
      "loss": 0.3567,
      "step": 34585
    },
    {
      "epoch": 1.5871690147308521,
      "grad_norm": 0.47606053948402405,
      "learning_rate": 4.694808287879609e-06,
      "loss": 0.3779,
      "step": 34586
    },
    {
      "epoch": 1.5872149052361066,
      "grad_norm": 0.4712086617946625,
      "learning_rate": 4.694563557511521e-06,
      "loss": 0.3306,
      "step": 34587
    },
    {
      "epoch": 1.5872607957413611,
      "grad_norm": 0.4578312337398529,
      "learning_rate": 4.694318827877914e-06,
      "loss": 0.3642,
      "step": 34588
    },
    {
      "epoch": 1.5873066862466154,
      "grad_norm": 0.46725407242774963,
      "learning_rate": 4.694074098979374e-06,
      "loss": 0.343,
      "step": 34589
    },
    {
      "epoch": 1.58735257675187,
      "grad_norm": 0.47350719571113586,
      "learning_rate": 4.6938293708164875e-06,
      "loss": 0.3736,
      "step": 34590
    },
    {
      "epoch": 1.5873984672571244,
      "grad_norm": 0.5089979767799377,
      "learning_rate": 4.693584643389846e-06,
      "loss": 0.4287,
      "step": 34591
    },
    {
      "epoch": 1.587444357762379,
      "grad_norm": 0.47449058294296265,
      "learning_rate": 4.693339916700037e-06,
      "loss": 0.3931,
      "step": 34592
    },
    {
      "epoch": 1.5874902482676334,
      "grad_norm": 0.4670298397541046,
      "learning_rate": 4.693095190747646e-06,
      "loss": 0.3434,
      "step": 34593
    },
    {
      "epoch": 1.587536138772888,
      "grad_norm": 0.42573282122612,
      "learning_rate": 4.692850465533268e-06,
      "loss": 0.2989,
      "step": 34594
    },
    {
      "epoch": 1.5875820292781424,
      "grad_norm": 0.419572651386261,
      "learning_rate": 4.692605741057487e-06,
      "loss": 0.3066,
      "step": 34595
    },
    {
      "epoch": 1.587627919783397,
      "grad_norm": 0.5445533394813538,
      "learning_rate": 4.692361017320889e-06,
      "loss": 0.3746,
      "step": 34596
    },
    {
      "epoch": 1.5876738102886514,
      "grad_norm": 0.49602025747299194,
      "learning_rate": 4.692116294324068e-06,
      "loss": 0.3649,
      "step": 34597
    },
    {
      "epoch": 1.587719700793906,
      "grad_norm": 0.5027179718017578,
      "learning_rate": 4.691871572067611e-06,
      "loss": 0.4138,
      "step": 34598
    },
    {
      "epoch": 1.5877655912991602,
      "grad_norm": 0.4785214066505432,
      "learning_rate": 4.691626850552104e-06,
      "loss": 0.3838,
      "step": 34599
    },
    {
      "epoch": 1.5878114818044147,
      "grad_norm": 0.4304395914077759,
      "learning_rate": 4.691382129778138e-06,
      "loss": 0.2985,
      "step": 34600
    },
    {
      "epoch": 1.5878573723096692,
      "grad_norm": 0.4875289499759674,
      "learning_rate": 4.691137409746301e-06,
      "loss": 0.3981,
      "step": 34601
    },
    {
      "epoch": 1.5879032628149234,
      "grad_norm": 0.4775334298610687,
      "learning_rate": 4.690892690457178e-06,
      "loss": 0.3871,
      "step": 34602
    },
    {
      "epoch": 1.587949153320178,
      "grad_norm": 0.4600880742073059,
      "learning_rate": 4.690647971911364e-06,
      "loss": 0.3516,
      "step": 34603
    },
    {
      "epoch": 1.5879950438254324,
      "grad_norm": 0.465518593788147,
      "learning_rate": 4.690403254109443e-06,
      "loss": 0.3578,
      "step": 34604
    },
    {
      "epoch": 1.588040934330687,
      "grad_norm": 0.4696153998374939,
      "learning_rate": 4.690158537052004e-06,
      "loss": 0.3879,
      "step": 34605
    },
    {
      "epoch": 1.5880868248359414,
      "grad_norm": 0.46333110332489014,
      "learning_rate": 4.689913820739636e-06,
      "loss": 0.3513,
      "step": 34606
    },
    {
      "epoch": 1.588132715341196,
      "grad_norm": 0.48700082302093506,
      "learning_rate": 4.689669105172929e-06,
      "loss": 0.3861,
      "step": 34607
    },
    {
      "epoch": 1.5881786058464504,
      "grad_norm": 0.47353264689445496,
      "learning_rate": 4.689424390352466e-06,
      "loss": 0.3946,
      "step": 34608
    },
    {
      "epoch": 1.588224496351705,
      "grad_norm": 0.430731862783432,
      "learning_rate": 4.689179676278843e-06,
      "loss": 0.3014,
      "step": 34609
    },
    {
      "epoch": 1.5882703868569594,
      "grad_norm": 0.4611452519893646,
      "learning_rate": 4.688934962952644e-06,
      "loss": 0.4035,
      "step": 34610
    },
    {
      "epoch": 1.588316277362214,
      "grad_norm": 0.5709488391876221,
      "learning_rate": 4.688690250374457e-06,
      "loss": 0.3988,
      "step": 34611
    },
    {
      "epoch": 1.5883621678674682,
      "grad_norm": 0.4743790626525879,
      "learning_rate": 4.688445538544873e-06,
      "loss": 0.4052,
      "step": 34612
    },
    {
      "epoch": 1.5884080583727227,
      "grad_norm": 0.46625393629074097,
      "learning_rate": 4.688200827464479e-06,
      "loss": 0.3481,
      "step": 34613
    },
    {
      "epoch": 1.5884539488779772,
      "grad_norm": 0.471148818731308,
      "learning_rate": 4.6879561171338615e-06,
      "loss": 0.3401,
      "step": 34614
    },
    {
      "epoch": 1.5884998393832315,
      "grad_norm": 0.4626951813697815,
      "learning_rate": 4.6877114075536135e-06,
      "loss": 0.3391,
      "step": 34615
    },
    {
      "epoch": 1.588545729888486,
      "grad_norm": 0.443096399307251,
      "learning_rate": 4.687466698724321e-06,
      "loss": 0.3111,
      "step": 34616
    },
    {
      "epoch": 1.5885916203937405,
      "grad_norm": 0.4645618796348572,
      "learning_rate": 4.687221990646573e-06,
      "loss": 0.3431,
      "step": 34617
    },
    {
      "epoch": 1.588637510898995,
      "grad_norm": 0.44557464122772217,
      "learning_rate": 4.686977283320954e-06,
      "loss": 0.3145,
      "step": 34618
    },
    {
      "epoch": 1.5886834014042495,
      "grad_norm": 0.48213982582092285,
      "learning_rate": 4.68673257674806e-06,
      "loss": 0.3815,
      "step": 34619
    },
    {
      "epoch": 1.588729291909504,
      "grad_norm": 0.46545517444610596,
      "learning_rate": 4.686487870928473e-06,
      "loss": 0.3293,
      "step": 34620
    },
    {
      "epoch": 1.5887751824147585,
      "grad_norm": 0.4708687663078308,
      "learning_rate": 4.6862431658627834e-06,
      "loss": 0.3559,
      "step": 34621
    },
    {
      "epoch": 1.588821072920013,
      "grad_norm": 0.4602172076702118,
      "learning_rate": 4.685998461551581e-06,
      "loss": 0.3382,
      "step": 34622
    },
    {
      "epoch": 1.5888669634252675,
      "grad_norm": 0.4869634509086609,
      "learning_rate": 4.685753757995453e-06,
      "loss": 0.3693,
      "step": 34623
    },
    {
      "epoch": 1.5889128539305217,
      "grad_norm": 0.46521803736686707,
      "learning_rate": 4.685509055194986e-06,
      "loss": 0.3357,
      "step": 34624
    },
    {
      "epoch": 1.5889587444357762,
      "grad_norm": 0.4544336497783661,
      "learning_rate": 4.685264353150773e-06,
      "loss": 0.3236,
      "step": 34625
    },
    {
      "epoch": 1.5890046349410307,
      "grad_norm": 0.49167975783348083,
      "learning_rate": 4.6850196518634e-06,
      "loss": 0.4445,
      "step": 34626
    },
    {
      "epoch": 1.589050525446285,
      "grad_norm": 0.5972442030906677,
      "learning_rate": 4.684774951333453e-06,
      "loss": 0.3391,
      "step": 34627
    },
    {
      "epoch": 1.5890964159515395,
      "grad_norm": 0.5371578931808472,
      "learning_rate": 4.684530251561526e-06,
      "loss": 0.4293,
      "step": 34628
    },
    {
      "epoch": 1.589142306456794,
      "grad_norm": 0.442306250333786,
      "learning_rate": 4.6842855525482026e-06,
      "loss": 0.3387,
      "step": 34629
    },
    {
      "epoch": 1.5891881969620485,
      "grad_norm": 0.4762756824493408,
      "learning_rate": 4.684040854294071e-06,
      "loss": 0.38,
      "step": 34630
    },
    {
      "epoch": 1.589234087467303,
      "grad_norm": 0.47143983840942383,
      "learning_rate": 4.683796156799723e-06,
      "loss": 0.3546,
      "step": 34631
    },
    {
      "epoch": 1.5892799779725575,
      "grad_norm": 0.5182498693466187,
      "learning_rate": 4.683551460065746e-06,
      "loss": 0.4613,
      "step": 34632
    },
    {
      "epoch": 1.589325868477812,
      "grad_norm": 0.4570844769477844,
      "learning_rate": 4.683306764092727e-06,
      "loss": 0.3273,
      "step": 34633
    },
    {
      "epoch": 1.5893717589830665,
      "grad_norm": 0.4789077639579773,
      "learning_rate": 4.683062068881256e-06,
      "loss": 0.3696,
      "step": 34634
    },
    {
      "epoch": 1.589417649488321,
      "grad_norm": 0.4777088165283203,
      "learning_rate": 4.682817374431921e-06,
      "loss": 0.3649,
      "step": 34635
    },
    {
      "epoch": 1.5894635399935755,
      "grad_norm": 0.4895303547382355,
      "learning_rate": 4.6825726807453075e-06,
      "loss": 0.3757,
      "step": 34636
    },
    {
      "epoch": 1.5895094304988298,
      "grad_norm": 0.477053701877594,
      "learning_rate": 4.682327987822009e-06,
      "loss": 0.3617,
      "step": 34637
    },
    {
      "epoch": 1.5895553210040843,
      "grad_norm": 0.46935850381851196,
      "learning_rate": 4.682083295662612e-06,
      "loss": 0.3415,
      "step": 34638
    },
    {
      "epoch": 1.5896012115093388,
      "grad_norm": 0.4935864806175232,
      "learning_rate": 4.681838604267702e-06,
      "loss": 0.3716,
      "step": 34639
    },
    {
      "epoch": 1.589647102014593,
      "grad_norm": 0.440094530582428,
      "learning_rate": 4.681593913637872e-06,
      "loss": 0.2922,
      "step": 34640
    },
    {
      "epoch": 1.5896929925198475,
      "grad_norm": 0.4910375773906708,
      "learning_rate": 4.681349223773708e-06,
      "loss": 0.4138,
      "step": 34641
    },
    {
      "epoch": 1.589738883025102,
      "grad_norm": 0.4861021041870117,
      "learning_rate": 4.6811045346757964e-06,
      "loss": 0.3913,
      "step": 34642
    },
    {
      "epoch": 1.5897847735303565,
      "grad_norm": 0.4904944598674774,
      "learning_rate": 4.6808598463447295e-06,
      "loss": 0.4097,
      "step": 34643
    },
    {
      "epoch": 1.589830664035611,
      "grad_norm": 0.4447546899318695,
      "learning_rate": 4.680615158781094e-06,
      "loss": 0.3336,
      "step": 34644
    },
    {
      "epoch": 1.5898765545408655,
      "grad_norm": 0.5124903321266174,
      "learning_rate": 4.680370471985476e-06,
      "loss": 0.4252,
      "step": 34645
    },
    {
      "epoch": 1.58992244504612,
      "grad_norm": 0.4588734209537506,
      "learning_rate": 4.680125785958468e-06,
      "loss": 0.3611,
      "step": 34646
    },
    {
      "epoch": 1.5899683355513745,
      "grad_norm": 0.49763569235801697,
      "learning_rate": 4.6798811007006575e-06,
      "loss": 0.3902,
      "step": 34647
    },
    {
      "epoch": 1.590014226056629,
      "grad_norm": 0.5236087441444397,
      "learning_rate": 4.67963641621263e-06,
      "loss": 0.4889,
      "step": 34648
    },
    {
      "epoch": 1.5900601165618835,
      "grad_norm": 0.8617398142814636,
      "learning_rate": 4.679391732494977e-06,
      "loss": 0.3431,
      "step": 34649
    },
    {
      "epoch": 1.5901060070671378,
      "grad_norm": 0.44143134355545044,
      "learning_rate": 4.679147049548286e-06,
      "loss": 0.2891,
      "step": 34650
    },
    {
      "epoch": 1.5901518975723923,
      "grad_norm": 0.4836884140968323,
      "learning_rate": 4.678902367373145e-06,
      "loss": 0.3478,
      "step": 34651
    },
    {
      "epoch": 1.5901977880776468,
      "grad_norm": 0.470175176858902,
      "learning_rate": 4.67865768597014e-06,
      "loss": 0.3429,
      "step": 34652
    },
    {
      "epoch": 1.590243678582901,
      "grad_norm": 0.47863808274269104,
      "learning_rate": 4.678413005339864e-06,
      "loss": 0.3558,
      "step": 34653
    },
    {
      "epoch": 1.5902895690881556,
      "grad_norm": 0.5038286447525024,
      "learning_rate": 4.678168325482903e-06,
      "loss": 0.3877,
      "step": 34654
    },
    {
      "epoch": 1.59033545959341,
      "grad_norm": 0.4705699384212494,
      "learning_rate": 4.6779236463998454e-06,
      "loss": 0.3735,
      "step": 34655
    },
    {
      "epoch": 1.5903813500986645,
      "grad_norm": 0.5444666743278503,
      "learning_rate": 4.6776789680912806e-06,
      "loss": 0.5396,
      "step": 34656
    },
    {
      "epoch": 1.590427240603919,
      "grad_norm": 0.4743742048740387,
      "learning_rate": 4.677434290557796e-06,
      "loss": 0.3331,
      "step": 34657
    },
    {
      "epoch": 1.5904731311091735,
      "grad_norm": 0.4315963685512543,
      "learning_rate": 4.677189613799978e-06,
      "loss": 0.2878,
      "step": 34658
    },
    {
      "epoch": 1.590519021614428,
      "grad_norm": 0.5056748986244202,
      "learning_rate": 4.676944937818419e-06,
      "loss": 0.4405,
      "step": 34659
    },
    {
      "epoch": 1.5905649121196825,
      "grad_norm": 0.47059184312820435,
      "learning_rate": 4.676700262613707e-06,
      "loss": 0.3645,
      "step": 34660
    },
    {
      "epoch": 1.590610802624937,
      "grad_norm": 0.4842603802680969,
      "learning_rate": 4.676455588186425e-06,
      "loss": 0.3663,
      "step": 34661
    },
    {
      "epoch": 1.5906566931301913,
      "grad_norm": 0.5041903853416443,
      "learning_rate": 4.676210914537168e-06,
      "loss": 0.4425,
      "step": 34662
    },
    {
      "epoch": 1.5907025836354458,
      "grad_norm": 0.4720912277698517,
      "learning_rate": 4.6759662416665205e-06,
      "loss": 0.3609,
      "step": 34663
    },
    {
      "epoch": 1.5907484741407003,
      "grad_norm": 0.46274659037590027,
      "learning_rate": 4.675721569575072e-06,
      "loss": 0.3577,
      "step": 34664
    },
    {
      "epoch": 1.5907943646459548,
      "grad_norm": 0.4845477044582367,
      "learning_rate": 4.675476898263411e-06,
      "loss": 0.3493,
      "step": 34665
    },
    {
      "epoch": 1.590840255151209,
      "grad_norm": 0.42338109016418457,
      "learning_rate": 4.675232227732127e-06,
      "loss": 0.2864,
      "step": 34666
    },
    {
      "epoch": 1.5908861456564636,
      "grad_norm": 0.4466056525707245,
      "learning_rate": 4.674987557981803e-06,
      "loss": 0.3235,
      "step": 34667
    },
    {
      "epoch": 1.590932036161718,
      "grad_norm": 0.47225505113601685,
      "learning_rate": 4.6747428890130345e-06,
      "loss": 0.3572,
      "step": 34668
    },
    {
      "epoch": 1.5909779266669726,
      "grad_norm": 0.5123305916786194,
      "learning_rate": 4.674498220826407e-06,
      "loss": 0.3971,
      "step": 34669
    },
    {
      "epoch": 1.591023817172227,
      "grad_norm": 0.4775237739086151,
      "learning_rate": 4.674253553422507e-06,
      "loss": 0.3934,
      "step": 34670
    },
    {
      "epoch": 1.5910697076774816,
      "grad_norm": 0.46348837018013,
      "learning_rate": 4.674008886801925e-06,
      "loss": 0.3426,
      "step": 34671
    },
    {
      "epoch": 1.591115598182736,
      "grad_norm": 0.5293883681297302,
      "learning_rate": 4.673764220965249e-06,
      "loss": 0.4223,
      "step": 34672
    },
    {
      "epoch": 1.5911614886879906,
      "grad_norm": 0.48984840512275696,
      "learning_rate": 4.673519555913066e-06,
      "loss": 0.3676,
      "step": 34673
    },
    {
      "epoch": 1.591207379193245,
      "grad_norm": 0.5124698877334595,
      "learning_rate": 4.673274891645967e-06,
      "loss": 0.4064,
      "step": 34674
    },
    {
      "epoch": 1.5912532696984993,
      "grad_norm": 0.5050203800201416,
      "learning_rate": 4.673030228164539e-06,
      "loss": 0.4332,
      "step": 34675
    },
    {
      "epoch": 1.5912991602037538,
      "grad_norm": 0.43710777163505554,
      "learning_rate": 4.672785565469368e-06,
      "loss": 0.31,
      "step": 34676
    },
    {
      "epoch": 1.5913450507090083,
      "grad_norm": 0.48700010776519775,
      "learning_rate": 4.672540903561046e-06,
      "loss": 0.3618,
      "step": 34677
    },
    {
      "epoch": 1.5913909412142626,
      "grad_norm": 0.4482556879520416,
      "learning_rate": 4.67229624244016e-06,
      "loss": 0.3464,
      "step": 34678
    },
    {
      "epoch": 1.591436831719517,
      "grad_norm": 0.45000559091567993,
      "learning_rate": 4.672051582107296e-06,
      "loss": 0.3439,
      "step": 34679
    },
    {
      "epoch": 1.5914827222247716,
      "grad_norm": 0.5079840421676636,
      "learning_rate": 4.671806922563047e-06,
      "loss": 0.3928,
      "step": 34680
    },
    {
      "epoch": 1.591528612730026,
      "grad_norm": 0.4834607243537903,
      "learning_rate": 4.671562263807997e-06,
      "loss": 0.382,
      "step": 34681
    },
    {
      "epoch": 1.5915745032352806,
      "grad_norm": 0.4709584414958954,
      "learning_rate": 4.671317605842739e-06,
      "loss": 0.3498,
      "step": 34682
    },
    {
      "epoch": 1.591620393740535,
      "grad_norm": 0.4898265302181244,
      "learning_rate": 4.671072948667854e-06,
      "loss": 0.3935,
      "step": 34683
    },
    {
      "epoch": 1.5916662842457896,
      "grad_norm": 0.4682188034057617,
      "learning_rate": 4.670828292283937e-06,
      "loss": 0.3933,
      "step": 34684
    },
    {
      "epoch": 1.591712174751044,
      "grad_norm": 0.4772651195526123,
      "learning_rate": 4.6705836366915755e-06,
      "loss": 0.3473,
      "step": 34685
    },
    {
      "epoch": 1.5917580652562986,
      "grad_norm": 0.4357205927371979,
      "learning_rate": 4.670338981891354e-06,
      "loss": 0.2978,
      "step": 34686
    },
    {
      "epoch": 1.591803955761553,
      "grad_norm": 0.4465365409851074,
      "learning_rate": 4.670094327883865e-06,
      "loss": 0.3359,
      "step": 34687
    },
    {
      "epoch": 1.5918498462668074,
      "grad_norm": 0.4873638153076172,
      "learning_rate": 4.669849674669695e-06,
      "loss": 0.4333,
      "step": 34688
    },
    {
      "epoch": 1.5918957367720619,
      "grad_norm": 0.4412349462509155,
      "learning_rate": 4.669605022249429e-06,
      "loss": 0.3036,
      "step": 34689
    },
    {
      "epoch": 1.5919416272773164,
      "grad_norm": 0.4592774212360382,
      "learning_rate": 4.6693603706236625e-06,
      "loss": 0.3121,
      "step": 34690
    },
    {
      "epoch": 1.5919875177825706,
      "grad_norm": 0.4286041259765625,
      "learning_rate": 4.6691157197929785e-06,
      "loss": 0.2859,
      "step": 34691
    },
    {
      "epoch": 1.5920334082878251,
      "grad_norm": 0.4584428668022156,
      "learning_rate": 4.668871069757966e-06,
      "loss": 0.3564,
      "step": 34692
    },
    {
      "epoch": 1.5920792987930796,
      "grad_norm": 0.46054837107658386,
      "learning_rate": 4.668626420519216e-06,
      "loss": 0.3284,
      "step": 34693
    },
    {
      "epoch": 1.5921251892983341,
      "grad_norm": 0.48860853910446167,
      "learning_rate": 4.668381772077314e-06,
      "loss": 0.3898,
      "step": 34694
    },
    {
      "epoch": 1.5921710798035886,
      "grad_norm": 0.4533975422382355,
      "learning_rate": 4.668137124432848e-06,
      "loss": 0.3614,
      "step": 34695
    },
    {
      "epoch": 1.5922169703088431,
      "grad_norm": 0.4359634816646576,
      "learning_rate": 4.667892477586409e-06,
      "loss": 0.2796,
      "step": 34696
    },
    {
      "epoch": 1.5922628608140976,
      "grad_norm": 0.44313374161720276,
      "learning_rate": 4.667647831538583e-06,
      "loss": 0.3098,
      "step": 34697
    },
    {
      "epoch": 1.5923087513193521,
      "grad_norm": 0.46732357144355774,
      "learning_rate": 4.667403186289959e-06,
      "loss": 0.3537,
      "step": 34698
    },
    {
      "epoch": 1.5923546418246066,
      "grad_norm": 0.4540899693965912,
      "learning_rate": 4.667158541841126e-06,
      "loss": 0.3112,
      "step": 34699
    },
    {
      "epoch": 1.5924005323298611,
      "grad_norm": 0.45713433623313904,
      "learning_rate": 4.6669138981926724e-06,
      "loss": 0.2922,
      "step": 34700
    },
    {
      "epoch": 1.5924464228351154,
      "grad_norm": 0.4429115951061249,
      "learning_rate": 4.6666692553451826e-06,
      "loss": 0.3187,
      "step": 34701
    },
    {
      "epoch": 1.59249231334037,
      "grad_norm": 0.4501625597476959,
      "learning_rate": 4.666424613299251e-06,
      "loss": 0.3345,
      "step": 34702
    },
    {
      "epoch": 1.5925382038456244,
      "grad_norm": 0.4606688618659973,
      "learning_rate": 4.666179972055462e-06,
      "loss": 0.2694,
      "step": 34703
    },
    {
      "epoch": 1.5925840943508787,
      "grad_norm": 0.5207648277282715,
      "learning_rate": 4.665935331614405e-06,
      "loss": 0.4544,
      "step": 34704
    },
    {
      "epoch": 1.5926299848561332,
      "grad_norm": 0.46853891015052795,
      "learning_rate": 4.665690691976668e-06,
      "loss": 0.3795,
      "step": 34705
    },
    {
      "epoch": 1.5926758753613877,
      "grad_norm": 0.4817691445350647,
      "learning_rate": 4.66544605314284e-06,
      "loss": 0.3978,
      "step": 34706
    },
    {
      "epoch": 1.5927217658666422,
      "grad_norm": 0.5696911215782166,
      "learning_rate": 4.665201415113507e-06,
      "loss": 0.5315,
      "step": 34707
    },
    {
      "epoch": 1.5927676563718967,
      "grad_norm": 0.4373834431171417,
      "learning_rate": 4.6649567778892605e-06,
      "loss": 0.3149,
      "step": 34708
    },
    {
      "epoch": 1.5928135468771512,
      "grad_norm": 0.4728403687477112,
      "learning_rate": 4.664712141470687e-06,
      "loss": 0.3658,
      "step": 34709
    },
    {
      "epoch": 1.5928594373824057,
      "grad_norm": 0.44083866477012634,
      "learning_rate": 4.664467505858373e-06,
      "loss": 0.3095,
      "step": 34710
    },
    {
      "epoch": 1.5929053278876601,
      "grad_norm": 0.47908201813697815,
      "learning_rate": 4.6642228710529106e-06,
      "loss": 0.339,
      "step": 34711
    },
    {
      "epoch": 1.5929512183929146,
      "grad_norm": 0.4389486014842987,
      "learning_rate": 4.663978237054887e-06,
      "loss": 0.3009,
      "step": 34712
    },
    {
      "epoch": 1.592997108898169,
      "grad_norm": 0.4782083332538605,
      "learning_rate": 4.6637336038648875e-06,
      "loss": 0.4075,
      "step": 34713
    },
    {
      "epoch": 1.5930429994034234,
      "grad_norm": 0.4747216999530792,
      "learning_rate": 4.6634889714835044e-06,
      "loss": 0.3509,
      "step": 34714
    },
    {
      "epoch": 1.593088889908678,
      "grad_norm": 0.5163630843162537,
      "learning_rate": 4.663244339911324e-06,
      "loss": 0.4079,
      "step": 34715
    },
    {
      "epoch": 1.5931347804139322,
      "grad_norm": 0.44474077224731445,
      "learning_rate": 4.662999709148935e-06,
      "loss": 0.3266,
      "step": 34716
    },
    {
      "epoch": 1.5931806709191867,
      "grad_norm": 0.44529733061790466,
      "learning_rate": 4.662755079196922e-06,
      "loss": 0.3286,
      "step": 34717
    },
    {
      "epoch": 1.5932265614244412,
      "grad_norm": 0.45347124338150024,
      "learning_rate": 4.662510450055881e-06,
      "loss": 0.3454,
      "step": 34718
    },
    {
      "epoch": 1.5932724519296957,
      "grad_norm": 0.4771330654621124,
      "learning_rate": 4.662265821726394e-06,
      "loss": 0.3701,
      "step": 34719
    },
    {
      "epoch": 1.5933183424349502,
      "grad_norm": 0.4837397038936615,
      "learning_rate": 4.6620211942090494e-06,
      "loss": 0.3854,
      "step": 34720
    },
    {
      "epoch": 1.5933642329402047,
      "grad_norm": 0.4684503972530365,
      "learning_rate": 4.66177656750444e-06,
      "loss": 0.3208,
      "step": 34721
    },
    {
      "epoch": 1.5934101234454592,
      "grad_norm": 0.44609692692756653,
      "learning_rate": 4.66153194161315e-06,
      "loss": 0.3063,
      "step": 34722
    },
    {
      "epoch": 1.5934560139507137,
      "grad_norm": 0.47343870997428894,
      "learning_rate": 4.661287316535768e-06,
      "loss": 0.3461,
      "step": 34723
    },
    {
      "epoch": 1.5935019044559682,
      "grad_norm": 0.4844398498535156,
      "learning_rate": 4.661042692272884e-06,
      "loss": 0.4094,
      "step": 34724
    },
    {
      "epoch": 1.5935477949612227,
      "grad_norm": 0.5398646593093872,
      "learning_rate": 4.660798068825086e-06,
      "loss": 0.4016,
      "step": 34725
    },
    {
      "epoch": 1.593593685466477,
      "grad_norm": 0.44693177938461304,
      "learning_rate": 4.66055344619296e-06,
      "loss": 0.315,
      "step": 34726
    },
    {
      "epoch": 1.5936395759717314,
      "grad_norm": 0.4499146640300751,
      "learning_rate": 4.660308824377097e-06,
      "loss": 0.3207,
      "step": 34727
    },
    {
      "epoch": 1.593685466476986,
      "grad_norm": 0.41190117597579956,
      "learning_rate": 4.6600642033780845e-06,
      "loss": 0.2952,
      "step": 34728
    },
    {
      "epoch": 1.5937313569822402,
      "grad_norm": 0.4760834872722626,
      "learning_rate": 4.6598195831965085e-06,
      "loss": 0.3782,
      "step": 34729
    },
    {
      "epoch": 1.5937772474874947,
      "grad_norm": 0.5275144577026367,
      "learning_rate": 4.659574963832961e-06,
      "loss": 0.4145,
      "step": 34730
    },
    {
      "epoch": 1.5938231379927492,
      "grad_norm": 0.41282689571380615,
      "learning_rate": 4.6593303452880275e-06,
      "loss": 0.2934,
      "step": 34731
    },
    {
      "epoch": 1.5938690284980037,
      "grad_norm": 0.48382943868637085,
      "learning_rate": 4.659085727562296e-06,
      "loss": 0.3895,
      "step": 34732
    },
    {
      "epoch": 1.5939149190032582,
      "grad_norm": 0.49519088864326477,
      "learning_rate": 4.658841110656357e-06,
      "loss": 0.3904,
      "step": 34733
    },
    {
      "epoch": 1.5939608095085127,
      "grad_norm": 0.537339985370636,
      "learning_rate": 4.658596494570798e-06,
      "loss": 0.3815,
      "step": 34734
    },
    {
      "epoch": 1.5940067000137672,
      "grad_norm": 0.42528122663497925,
      "learning_rate": 4.658351879306205e-06,
      "loss": 0.2916,
      "step": 34735
    },
    {
      "epoch": 1.5940525905190217,
      "grad_norm": 0.4270685017108917,
      "learning_rate": 4.6581072648631685e-06,
      "loss": 0.2967,
      "step": 34736
    },
    {
      "epoch": 1.5940984810242762,
      "grad_norm": 0.46985676884651184,
      "learning_rate": 4.657862651242277e-06,
      "loss": 0.3095,
      "step": 34737
    },
    {
      "epoch": 1.5941443715295307,
      "grad_norm": 0.5448096394538879,
      "learning_rate": 4.657618038444115e-06,
      "loss": 0.4252,
      "step": 34738
    },
    {
      "epoch": 1.594190262034785,
      "grad_norm": 0.48673850297927856,
      "learning_rate": 4.657373426469276e-06,
      "loss": 0.3731,
      "step": 34739
    },
    {
      "epoch": 1.5942361525400395,
      "grad_norm": 0.45394930243492126,
      "learning_rate": 4.657128815318346e-06,
      "loss": 0.313,
      "step": 34740
    },
    {
      "epoch": 1.594282043045294,
      "grad_norm": 0.5110794305801392,
      "learning_rate": 4.656884204991912e-06,
      "loss": 0.4121,
      "step": 34741
    },
    {
      "epoch": 1.5943279335505482,
      "grad_norm": 0.4714236557483673,
      "learning_rate": 4.656639595490564e-06,
      "loss": 0.3441,
      "step": 34742
    },
    {
      "epoch": 1.5943738240558027,
      "grad_norm": 0.4964030981063843,
      "learning_rate": 4.65639498681489e-06,
      "loss": 0.4028,
      "step": 34743
    },
    {
      "epoch": 1.5944197145610572,
      "grad_norm": 0.4547504484653473,
      "learning_rate": 4.656150378965475e-06,
      "loss": 0.3657,
      "step": 34744
    },
    {
      "epoch": 1.5944656050663117,
      "grad_norm": 0.4467400014400482,
      "learning_rate": 4.655905771942912e-06,
      "loss": 0.3198,
      "step": 34745
    },
    {
      "epoch": 1.5945114955715662,
      "grad_norm": 0.4882299304008484,
      "learning_rate": 4.655661165747787e-06,
      "loss": 0.3602,
      "step": 34746
    },
    {
      "epoch": 1.5945573860768207,
      "grad_norm": 0.44205793738365173,
      "learning_rate": 4.655416560380688e-06,
      "loss": 0.3375,
      "step": 34747
    },
    {
      "epoch": 1.5946032765820752,
      "grad_norm": 0.47099974751472473,
      "learning_rate": 4.655171955842202e-06,
      "loss": 0.3417,
      "step": 34748
    },
    {
      "epoch": 1.5946491670873297,
      "grad_norm": 0.4626399278640747,
      "learning_rate": 4.654927352132921e-06,
      "loss": 0.3004,
      "step": 34749
    },
    {
      "epoch": 1.5946950575925842,
      "grad_norm": 0.46260422468185425,
      "learning_rate": 4.65468274925343e-06,
      "loss": 0.3232,
      "step": 34750
    },
    {
      "epoch": 1.5947409480978385,
      "grad_norm": 0.555467963218689,
      "learning_rate": 4.654438147204317e-06,
      "loss": 0.3975,
      "step": 34751
    },
    {
      "epoch": 1.594786838603093,
      "grad_norm": 0.4430099129676819,
      "learning_rate": 4.654193545986172e-06,
      "loss": 0.3169,
      "step": 34752
    },
    {
      "epoch": 1.5948327291083475,
      "grad_norm": 0.5208683013916016,
      "learning_rate": 4.653948945599583e-06,
      "loss": 0.4242,
      "step": 34753
    },
    {
      "epoch": 1.594878619613602,
      "grad_norm": 0.4529822766780853,
      "learning_rate": 4.6537043460451345e-06,
      "loss": 0.3775,
      "step": 34754
    },
    {
      "epoch": 1.5949245101188563,
      "grad_norm": 0.43357154726982117,
      "learning_rate": 4.653459747323421e-06,
      "loss": 0.3186,
      "step": 34755
    },
    {
      "epoch": 1.5949704006241108,
      "grad_norm": 0.45859864354133606,
      "learning_rate": 4.653215149435027e-06,
      "loss": 0.3187,
      "step": 34756
    },
    {
      "epoch": 1.5950162911293653,
      "grad_norm": 0.5375958681106567,
      "learning_rate": 4.65297055238054e-06,
      "loss": 0.4376,
      "step": 34757
    },
    {
      "epoch": 1.5950621816346198,
      "grad_norm": 0.44979578256607056,
      "learning_rate": 4.65272595616055e-06,
      "loss": 0.3116,
      "step": 34758
    },
    {
      "epoch": 1.5951080721398743,
      "grad_norm": 0.4660671651363373,
      "learning_rate": 4.652481360775645e-06,
      "loss": 0.3921,
      "step": 34759
    },
    {
      "epoch": 1.5951539626451288,
      "grad_norm": 0.4531121850013733,
      "learning_rate": 4.65223676622641e-06,
      "loss": 0.3281,
      "step": 34760
    },
    {
      "epoch": 1.5951998531503833,
      "grad_norm": 0.5443975925445557,
      "learning_rate": 4.651992172513438e-06,
      "loss": 0.403,
      "step": 34761
    },
    {
      "epoch": 1.5952457436556378,
      "grad_norm": 0.4304085373878479,
      "learning_rate": 4.651747579637316e-06,
      "loss": 0.279,
      "step": 34762
    },
    {
      "epoch": 1.5952916341608923,
      "grad_norm": 0.4876463711261749,
      "learning_rate": 4.651502987598629e-06,
      "loss": 0.4351,
      "step": 34763
    },
    {
      "epoch": 1.5953375246661465,
      "grad_norm": 0.4766169488430023,
      "learning_rate": 4.651258396397968e-06,
      "loss": 0.3516,
      "step": 34764
    },
    {
      "epoch": 1.595383415171401,
      "grad_norm": 0.5024241209030151,
      "learning_rate": 4.651013806035921e-06,
      "loss": 0.3935,
      "step": 34765
    },
    {
      "epoch": 1.5954293056766555,
      "grad_norm": 0.4957285225391388,
      "learning_rate": 4.650769216513075e-06,
      "loss": 0.3942,
      "step": 34766
    },
    {
      "epoch": 1.5954751961819098,
      "grad_norm": 0.4947311580181122,
      "learning_rate": 4.650524627830019e-06,
      "loss": 0.4343,
      "step": 34767
    },
    {
      "epoch": 1.5955210866871643,
      "grad_norm": 0.47733622789382935,
      "learning_rate": 4.650280039987342e-06,
      "loss": 0.38,
      "step": 34768
    },
    {
      "epoch": 1.5955669771924188,
      "grad_norm": 0.494078665971756,
      "learning_rate": 4.650035452985629e-06,
      "loss": 0.3964,
      "step": 34769
    },
    {
      "epoch": 1.5956128676976733,
      "grad_norm": 0.3970520794391632,
      "learning_rate": 4.649790866825473e-06,
      "loss": 0.2624,
      "step": 34770
    },
    {
      "epoch": 1.5956587582029278,
      "grad_norm": 0.4800978899002075,
      "learning_rate": 4.649546281507458e-06,
      "loss": 0.3802,
      "step": 34771
    },
    {
      "epoch": 1.5957046487081823,
      "grad_norm": 0.5012110471725464,
      "learning_rate": 4.649301697032172e-06,
      "loss": 0.4484,
      "step": 34772
    },
    {
      "epoch": 1.5957505392134368,
      "grad_norm": 0.4730996787548065,
      "learning_rate": 4.649057113400208e-06,
      "loss": 0.3741,
      "step": 34773
    },
    {
      "epoch": 1.5957964297186913,
      "grad_norm": 0.44102975726127625,
      "learning_rate": 4.64881253061215e-06,
      "loss": 0.2918,
      "step": 34774
    },
    {
      "epoch": 1.5958423202239458,
      "grad_norm": 0.49235406517982483,
      "learning_rate": 4.648567948668584e-06,
      "loss": 0.4044,
      "step": 34775
    },
    {
      "epoch": 1.5958882107292003,
      "grad_norm": 0.44022056460380554,
      "learning_rate": 4.648323367570104e-06,
      "loss": 0.3514,
      "step": 34776
    },
    {
      "epoch": 1.5959341012344546,
      "grad_norm": 0.6699870824813843,
      "learning_rate": 4.648078787317295e-06,
      "loss": 0.3233,
      "step": 34777
    },
    {
      "epoch": 1.595979991739709,
      "grad_norm": 0.5118640661239624,
      "learning_rate": 4.647834207910745e-06,
      "loss": 0.4141,
      "step": 34778
    },
    {
      "epoch": 1.5960258822449636,
      "grad_norm": 0.4721646308898926,
      "learning_rate": 4.647589629351043e-06,
      "loss": 0.3568,
      "step": 34779
    },
    {
      "epoch": 1.5960717727502178,
      "grad_norm": 0.4855617880821228,
      "learning_rate": 4.647345051638777e-06,
      "loss": 0.3613,
      "step": 34780
    },
    {
      "epoch": 1.5961176632554723,
      "grad_norm": 0.5117548704147339,
      "learning_rate": 4.647100474774534e-06,
      "loss": 0.3943,
      "step": 34781
    },
    {
      "epoch": 1.5961635537607268,
      "grad_norm": 0.5065207481384277,
      "learning_rate": 4.646855898758903e-06,
      "loss": 0.4252,
      "step": 34782
    },
    {
      "epoch": 1.5962094442659813,
      "grad_norm": 0.4753648638725281,
      "learning_rate": 4.646611323592472e-06,
      "loss": 0.3159,
      "step": 34783
    },
    {
      "epoch": 1.5962553347712358,
      "grad_norm": 0.5063811540603638,
      "learning_rate": 4.64636674927583e-06,
      "loss": 0.3566,
      "step": 34784
    },
    {
      "epoch": 1.5963012252764903,
      "grad_norm": 0.4984270930290222,
      "learning_rate": 4.6461221758095635e-06,
      "loss": 0.4068,
      "step": 34785
    },
    {
      "epoch": 1.5963471157817448,
      "grad_norm": 0.5074690580368042,
      "learning_rate": 4.645877603194263e-06,
      "loss": 0.4476,
      "step": 34786
    },
    {
      "epoch": 1.5963930062869993,
      "grad_norm": 0.4590400457382202,
      "learning_rate": 4.645633031430514e-06,
      "loss": 0.3239,
      "step": 34787
    },
    {
      "epoch": 1.5964388967922538,
      "grad_norm": 0.45170119404792786,
      "learning_rate": 4.645388460518904e-06,
      "loss": 0.3172,
      "step": 34788
    },
    {
      "epoch": 1.5964847872975083,
      "grad_norm": 0.46132713556289673,
      "learning_rate": 4.645143890460026e-06,
      "loss": 0.3515,
      "step": 34789
    },
    {
      "epoch": 1.5965306778027626,
      "grad_norm": 0.6673550605773926,
      "learning_rate": 4.644899321254464e-06,
      "loss": 0.3615,
      "step": 34790
    },
    {
      "epoch": 1.596576568308017,
      "grad_norm": 0.4102596044540405,
      "learning_rate": 4.644654752902806e-06,
      "loss": 0.2762,
      "step": 34791
    },
    {
      "epoch": 1.5966224588132716,
      "grad_norm": 0.4621727168560028,
      "learning_rate": 4.6444101854056415e-06,
      "loss": 0.3069,
      "step": 34792
    },
    {
      "epoch": 1.5966683493185259,
      "grad_norm": 0.45002245903015137,
      "learning_rate": 4.6441656187635595e-06,
      "loss": 0.3143,
      "step": 34793
    },
    {
      "epoch": 1.5967142398237804,
      "grad_norm": 0.41082170605659485,
      "learning_rate": 4.643921052977144e-06,
      "loss": 0.2788,
      "step": 34794
    },
    {
      "epoch": 1.5967601303290349,
      "grad_norm": 0.44898471236228943,
      "learning_rate": 4.6436764880469895e-06,
      "loss": 0.2974,
      "step": 34795
    },
    {
      "epoch": 1.5968060208342894,
      "grad_norm": 0.8298036456108093,
      "learning_rate": 4.64343192397368e-06,
      "loss": 0.3275,
      "step": 34796
    },
    {
      "epoch": 1.5968519113395439,
      "grad_norm": 0.4956530034542084,
      "learning_rate": 4.643187360757801e-06,
      "loss": 0.3454,
      "step": 34797
    },
    {
      "epoch": 1.5968978018447983,
      "grad_norm": 0.4920119643211365,
      "learning_rate": 4.642942798399947e-06,
      "loss": 0.3731,
      "step": 34798
    },
    {
      "epoch": 1.5969436923500528,
      "grad_norm": 0.5594425201416016,
      "learning_rate": 4.642698236900702e-06,
      "loss": 0.3948,
      "step": 34799
    },
    {
      "epoch": 1.5969895828553073,
      "grad_norm": 0.5012873411178589,
      "learning_rate": 4.642453676260654e-06,
      "loss": 0.326,
      "step": 34800
    },
    {
      "epoch": 1.5970354733605618,
      "grad_norm": 0.4863239526748657,
      "learning_rate": 4.642209116480394e-06,
      "loss": 0.3539,
      "step": 34801
    },
    {
      "epoch": 1.5970813638658161,
      "grad_norm": 0.6023303270339966,
      "learning_rate": 4.641964557560507e-06,
      "loss": 0.366,
      "step": 34802
    },
    {
      "epoch": 1.5971272543710706,
      "grad_norm": 0.4957040846347809,
      "learning_rate": 4.641719999501581e-06,
      "loss": 0.3697,
      "step": 34803
    },
    {
      "epoch": 1.5971731448763251,
      "grad_norm": 0.4713757634162903,
      "learning_rate": 4.641475442304207e-06,
      "loss": 0.3196,
      "step": 34804
    },
    {
      "epoch": 1.5972190353815794,
      "grad_norm": 0.5363430976867676,
      "learning_rate": 4.641230885968972e-06,
      "loss": 0.3747,
      "step": 34805
    },
    {
      "epoch": 1.5972649258868339,
      "grad_norm": 0.4569535553455353,
      "learning_rate": 4.6409863304964616e-06,
      "loss": 0.3122,
      "step": 34806
    },
    {
      "epoch": 1.5973108163920884,
      "grad_norm": 0.43456047773361206,
      "learning_rate": 4.640741775887267e-06,
      "loss": 0.2966,
      "step": 34807
    },
    {
      "epoch": 1.5973567068973429,
      "grad_norm": 0.4803799092769623,
      "learning_rate": 4.640497222141976e-06,
      "loss": 0.345,
      "step": 34808
    },
    {
      "epoch": 1.5974025974025974,
      "grad_norm": 0.4783639907836914,
      "learning_rate": 4.640252669261172e-06,
      "loss": 0.3921,
      "step": 34809
    },
    {
      "epoch": 1.5974484879078519,
      "grad_norm": 0.5058416128158569,
      "learning_rate": 4.64000811724545e-06,
      "loss": 0.3076,
      "step": 34810
    },
    {
      "epoch": 1.5974943784131064,
      "grad_norm": 0.5651657581329346,
      "learning_rate": 4.639763566095395e-06,
      "loss": 0.4238,
      "step": 34811
    },
    {
      "epoch": 1.5975402689183609,
      "grad_norm": 0.5287007093429565,
      "learning_rate": 4.6395190158115925e-06,
      "loss": 0.439,
      "step": 34812
    },
    {
      "epoch": 1.5975861594236154,
      "grad_norm": 0.46182140707969666,
      "learning_rate": 4.639274466394635e-06,
      "loss": 0.3346,
      "step": 34813
    },
    {
      "epoch": 1.5976320499288699,
      "grad_norm": 0.48368632793426514,
      "learning_rate": 4.639029917845109e-06,
      "loss": 0.3619,
      "step": 34814
    },
    {
      "epoch": 1.5976779404341241,
      "grad_norm": 0.5247377753257751,
      "learning_rate": 4.638785370163602e-06,
      "loss": 0.3754,
      "step": 34815
    },
    {
      "epoch": 1.5977238309393786,
      "grad_norm": 0.47451063990592957,
      "learning_rate": 4.6385408233507e-06,
      "loss": 0.2921,
      "step": 34816
    },
    {
      "epoch": 1.5977697214446331,
      "grad_norm": 0.44768038392066956,
      "learning_rate": 4.638296277406996e-06,
      "loss": 0.3258,
      "step": 34817
    },
    {
      "epoch": 1.5978156119498874,
      "grad_norm": 0.4526967406272888,
      "learning_rate": 4.638051732333074e-06,
      "loss": 0.326,
      "step": 34818
    },
    {
      "epoch": 1.597861502455142,
      "grad_norm": 0.4210854768753052,
      "learning_rate": 4.637807188129522e-06,
      "loss": 0.3045,
      "step": 34819
    },
    {
      "epoch": 1.5979073929603964,
      "grad_norm": 0.4827062487602234,
      "learning_rate": 4.637562644796931e-06,
      "loss": 0.3444,
      "step": 34820
    },
    {
      "epoch": 1.597953283465651,
      "grad_norm": 0.46279793977737427,
      "learning_rate": 4.637318102335888e-06,
      "loss": 0.3574,
      "step": 34821
    },
    {
      "epoch": 1.5979991739709054,
      "grad_norm": 0.5102381706237793,
      "learning_rate": 4.637073560746979e-06,
      "loss": 0.3975,
      "step": 34822
    },
    {
      "epoch": 1.59804506447616,
      "grad_norm": 0.46953868865966797,
      "learning_rate": 4.636829020030794e-06,
      "loss": 0.3363,
      "step": 34823
    },
    {
      "epoch": 1.5980909549814144,
      "grad_norm": 0.49957776069641113,
      "learning_rate": 4.636584480187922e-06,
      "loss": 0.3883,
      "step": 34824
    },
    {
      "epoch": 1.598136845486669,
      "grad_norm": 0.47378212213516235,
      "learning_rate": 4.636339941218946e-06,
      "loss": 0.3538,
      "step": 34825
    },
    {
      "epoch": 1.5981827359919234,
      "grad_norm": 0.45736268162727356,
      "learning_rate": 4.636095403124461e-06,
      "loss": 0.3431,
      "step": 34826
    },
    {
      "epoch": 1.598228626497178,
      "grad_norm": 0.48532652854919434,
      "learning_rate": 4.635850865905051e-06,
      "loss": 0.3563,
      "step": 34827
    },
    {
      "epoch": 1.5982745170024322,
      "grad_norm": 0.48605310916900635,
      "learning_rate": 4.635606329561304e-06,
      "loss": 0.4214,
      "step": 34828
    },
    {
      "epoch": 1.5983204075076867,
      "grad_norm": 0.5004994869232178,
      "learning_rate": 4.6353617940938096e-06,
      "loss": 0.4341,
      "step": 34829
    },
    {
      "epoch": 1.5983662980129412,
      "grad_norm": 0.4945167899131775,
      "learning_rate": 4.635117259503155e-06,
      "loss": 0.4265,
      "step": 34830
    },
    {
      "epoch": 1.5984121885181954,
      "grad_norm": 0.5651324391365051,
      "learning_rate": 4.634872725789926e-06,
      "loss": 0.3942,
      "step": 34831
    },
    {
      "epoch": 1.59845807902345,
      "grad_norm": 0.4603986442089081,
      "learning_rate": 4.634628192954716e-06,
      "loss": 0.3313,
      "step": 34832
    },
    {
      "epoch": 1.5985039695287044,
      "grad_norm": 0.4635562598705292,
      "learning_rate": 4.634383660998109e-06,
      "loss": 0.358,
      "step": 34833
    },
    {
      "epoch": 1.598549860033959,
      "grad_norm": 0.4482724964618683,
      "learning_rate": 4.634139129920693e-06,
      "loss": 0.3288,
      "step": 34834
    },
    {
      "epoch": 1.5985957505392134,
      "grad_norm": 0.44460728764533997,
      "learning_rate": 4.633894599723058e-06,
      "loss": 0.3258,
      "step": 34835
    },
    {
      "epoch": 1.598641641044468,
      "grad_norm": 0.47591206431388855,
      "learning_rate": 4.63365007040579e-06,
      "loss": 0.3747,
      "step": 34836
    },
    {
      "epoch": 1.5986875315497224,
      "grad_norm": 0.47765445709228516,
      "learning_rate": 4.6334055419694775e-06,
      "loss": 0.3715,
      "step": 34837
    },
    {
      "epoch": 1.598733422054977,
      "grad_norm": 0.4598182439804077,
      "learning_rate": 4.63316101441471e-06,
      "loss": 0.3199,
      "step": 34838
    },
    {
      "epoch": 1.5987793125602314,
      "grad_norm": 0.44339531660079956,
      "learning_rate": 4.632916487742077e-06,
      "loss": 0.3296,
      "step": 34839
    },
    {
      "epoch": 1.5988252030654857,
      "grad_norm": 0.47321388125419617,
      "learning_rate": 4.632671961952159e-06,
      "loss": 0.365,
      "step": 34840
    },
    {
      "epoch": 1.5988710935707402,
      "grad_norm": 0.46899881958961487,
      "learning_rate": 4.632427437045551e-06,
      "loss": 0.3348,
      "step": 34841
    },
    {
      "epoch": 1.5989169840759947,
      "grad_norm": 0.49198564887046814,
      "learning_rate": 4.63218291302284e-06,
      "loss": 0.3822,
      "step": 34842
    },
    {
      "epoch": 1.5989628745812492,
      "grad_norm": 0.4984564185142517,
      "learning_rate": 4.631938389884612e-06,
      "loss": 0.4179,
      "step": 34843
    },
    {
      "epoch": 1.5990087650865035,
      "grad_norm": 0.4637838900089264,
      "learning_rate": 4.631693867631457e-06,
      "loss": 0.3799,
      "step": 34844
    },
    {
      "epoch": 1.599054655591758,
      "grad_norm": 0.4733388423919678,
      "learning_rate": 4.631449346263962e-06,
      "loss": 0.3649,
      "step": 34845
    },
    {
      "epoch": 1.5991005460970125,
      "grad_norm": 0.4993596076965332,
      "learning_rate": 4.631204825782715e-06,
      "loss": 0.3573,
      "step": 34846
    },
    {
      "epoch": 1.599146436602267,
      "grad_norm": 0.47080856561660767,
      "learning_rate": 4.630960306188301e-06,
      "loss": 0.3679,
      "step": 34847
    },
    {
      "epoch": 1.5991923271075215,
      "grad_norm": 0.45036759972572327,
      "learning_rate": 4.630715787481314e-06,
      "loss": 0.3096,
      "step": 34848
    },
    {
      "epoch": 1.599238217612776,
      "grad_norm": 0.42406123876571655,
      "learning_rate": 4.630471269662339e-06,
      "loss": 0.269,
      "step": 34849
    },
    {
      "epoch": 1.5992841081180305,
      "grad_norm": 0.5076342225074768,
      "learning_rate": 4.6302267527319625e-06,
      "loss": 0.4281,
      "step": 34850
    },
    {
      "epoch": 1.599329998623285,
      "grad_norm": 0.47667980194091797,
      "learning_rate": 4.629982236690776e-06,
      "loss": 0.3622,
      "step": 34851
    },
    {
      "epoch": 1.5993758891285395,
      "grad_norm": 0.4962298274040222,
      "learning_rate": 4.629737721539364e-06,
      "loss": 0.3633,
      "step": 34852
    },
    {
      "epoch": 1.5994217796337937,
      "grad_norm": 0.4731343984603882,
      "learning_rate": 4.629493207278315e-06,
      "loss": 0.3313,
      "step": 34853
    },
    {
      "epoch": 1.5994676701390482,
      "grad_norm": 0.5986219644546509,
      "learning_rate": 4.629248693908219e-06,
      "loss": 0.4005,
      "step": 34854
    },
    {
      "epoch": 1.5995135606443027,
      "grad_norm": 0.43405160307884216,
      "learning_rate": 4.629004181429663e-06,
      "loss": 0.3282,
      "step": 34855
    },
    {
      "epoch": 1.599559451149557,
      "grad_norm": 0.49386027455329895,
      "learning_rate": 4.628759669843235e-06,
      "loss": 0.369,
      "step": 34856
    },
    {
      "epoch": 1.5996053416548115,
      "grad_norm": 0.4660702645778656,
      "learning_rate": 4.628515159149523e-06,
      "loss": 0.3087,
      "step": 34857
    },
    {
      "epoch": 1.599651232160066,
      "grad_norm": 0.4737514853477478,
      "learning_rate": 4.628270649349115e-06,
      "loss": 0.353,
      "step": 34858
    },
    {
      "epoch": 1.5996971226653205,
      "grad_norm": 0.5699807405471802,
      "learning_rate": 4.628026140442596e-06,
      "loss": 0.3356,
      "step": 34859
    },
    {
      "epoch": 1.599743013170575,
      "grad_norm": 0.49088722467422485,
      "learning_rate": 4.627781632430559e-06,
      "loss": 0.3491,
      "step": 34860
    },
    {
      "epoch": 1.5997889036758295,
      "grad_norm": 0.45973482728004456,
      "learning_rate": 4.627537125313591e-06,
      "loss": 0.3163,
      "step": 34861
    },
    {
      "epoch": 1.599834794181084,
      "grad_norm": 0.43102431297302246,
      "learning_rate": 4.627292619092276e-06,
      "loss": 0.2719,
      "step": 34862
    },
    {
      "epoch": 1.5998806846863385,
      "grad_norm": 0.42631447315216064,
      "learning_rate": 4.627048113767206e-06,
      "loss": 0.2965,
      "step": 34863
    },
    {
      "epoch": 1.599926575191593,
      "grad_norm": 0.5110950469970703,
      "learning_rate": 4.626803609338968e-06,
      "loss": 0.3697,
      "step": 34864
    },
    {
      "epoch": 1.5999724656968475,
      "grad_norm": 0.45644357800483704,
      "learning_rate": 4.626559105808149e-06,
      "loss": 0.3551,
      "step": 34865
    },
    {
      "epoch": 1.6000183562021018,
      "grad_norm": 0.48917511105537415,
      "learning_rate": 4.626314603175337e-06,
      "loss": 0.3487,
      "step": 34866
    },
    {
      "epoch": 1.6000642467073563,
      "grad_norm": 0.46688464283943176,
      "learning_rate": 4.6260701014411215e-06,
      "loss": 0.3697,
      "step": 34867
    },
    {
      "epoch": 1.6001101372126108,
      "grad_norm": 0.4899645447731018,
      "learning_rate": 4.625825600606087e-06,
      "loss": 0.406,
      "step": 34868
    },
    {
      "epoch": 1.600156027717865,
      "grad_norm": 0.4923345744609833,
      "learning_rate": 4.625581100670827e-06,
      "loss": 0.3917,
      "step": 34869
    },
    {
      "epoch": 1.6002019182231195,
      "grad_norm": 0.44675078988075256,
      "learning_rate": 4.625336601635925e-06,
      "loss": 0.3067,
      "step": 34870
    },
    {
      "epoch": 1.600247808728374,
      "grad_norm": 0.46831583976745605,
      "learning_rate": 4.625092103501969e-06,
      "loss": 0.3761,
      "step": 34871
    },
    {
      "epoch": 1.6002936992336285,
      "grad_norm": 0.48872241377830505,
      "learning_rate": 4.62484760626955e-06,
      "loss": 0.3916,
      "step": 34872
    },
    {
      "epoch": 1.600339589738883,
      "grad_norm": 0.4432581961154938,
      "learning_rate": 4.624603109939254e-06,
      "loss": 0.3175,
      "step": 34873
    },
    {
      "epoch": 1.6003854802441375,
      "grad_norm": 0.4905319809913635,
      "learning_rate": 4.624358614511667e-06,
      "loss": 0.4025,
      "step": 34874
    },
    {
      "epoch": 1.600431370749392,
      "grad_norm": 0.56778883934021,
      "learning_rate": 4.624114119987381e-06,
      "loss": 0.3333,
      "step": 34875
    },
    {
      "epoch": 1.6004772612546465,
      "grad_norm": 0.5066177248954773,
      "learning_rate": 4.623869626366982e-06,
      "loss": 0.3829,
      "step": 34876
    },
    {
      "epoch": 1.600523151759901,
      "grad_norm": 0.4770393967628479,
      "learning_rate": 4.623625133651056e-06,
      "loss": 0.4121,
      "step": 34877
    },
    {
      "epoch": 1.6005690422651553,
      "grad_norm": 0.4968225955963135,
      "learning_rate": 4.623380641840195e-06,
      "loss": 0.4383,
      "step": 34878
    },
    {
      "epoch": 1.6006149327704098,
      "grad_norm": 0.4286067485809326,
      "learning_rate": 4.623136150934984e-06,
      "loss": 0.3046,
      "step": 34879
    },
    {
      "epoch": 1.6006608232756643,
      "grad_norm": 0.4685288667678833,
      "learning_rate": 4.622891660936012e-06,
      "loss": 0.3421,
      "step": 34880
    },
    {
      "epoch": 1.6007067137809188,
      "grad_norm": 0.4423278272151947,
      "learning_rate": 4.622647171843864e-06,
      "loss": 0.2903,
      "step": 34881
    },
    {
      "epoch": 1.600752604286173,
      "grad_norm": 0.4528770446777344,
      "learning_rate": 4.622402683659133e-06,
      "loss": 0.3523,
      "step": 34882
    },
    {
      "epoch": 1.6007984947914276,
      "grad_norm": 0.5023056864738464,
      "learning_rate": 4.622158196382405e-06,
      "loss": 0.3986,
      "step": 34883
    },
    {
      "epoch": 1.600844385296682,
      "grad_norm": 0.43100085854530334,
      "learning_rate": 4.621913710014264e-06,
      "loss": 0.2952,
      "step": 34884
    },
    {
      "epoch": 1.6008902758019365,
      "grad_norm": 0.4778672158718109,
      "learning_rate": 4.621669224555303e-06,
      "loss": 0.4142,
      "step": 34885
    },
    {
      "epoch": 1.600936166307191,
      "grad_norm": 0.4654640555381775,
      "learning_rate": 4.621424740006109e-06,
      "loss": 0.3503,
      "step": 34886
    },
    {
      "epoch": 1.6009820568124455,
      "grad_norm": 0.4785864055156708,
      "learning_rate": 4.621180256367268e-06,
      "loss": 0.4147,
      "step": 34887
    },
    {
      "epoch": 1.6010279473177,
      "grad_norm": 0.4578460156917572,
      "learning_rate": 4.6209357736393694e-06,
      "loss": 0.333,
      "step": 34888
    },
    {
      "epoch": 1.6010738378229545,
      "grad_norm": 0.5140366554260254,
      "learning_rate": 4.6206912918230004e-06,
      "loss": 0.3725,
      "step": 34889
    },
    {
      "epoch": 1.601119728328209,
      "grad_norm": 0.48435184359550476,
      "learning_rate": 4.620446810918748e-06,
      "loss": 0.3707,
      "step": 34890
    },
    {
      "epoch": 1.6011656188334633,
      "grad_norm": 0.4748537540435791,
      "learning_rate": 4.620202330927203e-06,
      "loss": 0.3644,
      "step": 34891
    },
    {
      "epoch": 1.6012115093387178,
      "grad_norm": 0.49680015444755554,
      "learning_rate": 4.619957851848951e-06,
      "loss": 0.3686,
      "step": 34892
    },
    {
      "epoch": 1.6012573998439723,
      "grad_norm": 0.4802763760089874,
      "learning_rate": 4.61971337368458e-06,
      "loss": 0.3566,
      "step": 34893
    },
    {
      "epoch": 1.6013032903492266,
      "grad_norm": 0.4818015992641449,
      "learning_rate": 4.619468896434679e-06,
      "loss": 0.3539,
      "step": 34894
    },
    {
      "epoch": 1.601349180854481,
      "grad_norm": 0.5197352766990662,
      "learning_rate": 4.619224420099835e-06,
      "loss": 0.3973,
      "step": 34895
    },
    {
      "epoch": 1.6013950713597356,
      "grad_norm": 0.5193723440170288,
      "learning_rate": 4.6189799446806344e-06,
      "loss": 0.4046,
      "step": 34896
    },
    {
      "epoch": 1.60144096186499,
      "grad_norm": 0.41530877351760864,
      "learning_rate": 4.618735470177669e-06,
      "loss": 0.2855,
      "step": 34897
    },
    {
      "epoch": 1.6014868523702446,
      "grad_norm": 0.505231499671936,
      "learning_rate": 4.6184909965915245e-06,
      "loss": 0.4059,
      "step": 34898
    },
    {
      "epoch": 1.601532742875499,
      "grad_norm": 0.46907904744148254,
      "learning_rate": 4.618246523922787e-06,
      "loss": 0.3279,
      "step": 34899
    },
    {
      "epoch": 1.6015786333807536,
      "grad_norm": 0.5181283354759216,
      "learning_rate": 4.6180020521720475e-06,
      "loss": 0.3189,
      "step": 34900
    },
    {
      "epoch": 1.601624523886008,
      "grad_norm": 0.42142054438591003,
      "learning_rate": 4.6177575813398925e-06,
      "loss": 0.2732,
      "step": 34901
    },
    {
      "epoch": 1.6016704143912626,
      "grad_norm": 0.4701882004737854,
      "learning_rate": 4.617513111426908e-06,
      "loss": 0.361,
      "step": 34902
    },
    {
      "epoch": 1.601716304896517,
      "grad_norm": 0.4183480441570282,
      "learning_rate": 4.617268642433685e-06,
      "loss": 0.2721,
      "step": 34903
    },
    {
      "epoch": 1.6017621954017713,
      "grad_norm": 0.4720074236392975,
      "learning_rate": 4.617024174360811e-06,
      "loss": 0.3306,
      "step": 34904
    },
    {
      "epoch": 1.6018080859070258,
      "grad_norm": 0.4673229157924652,
      "learning_rate": 4.616779707208872e-06,
      "loss": 0.3966,
      "step": 34905
    },
    {
      "epoch": 1.6018539764122803,
      "grad_norm": 0.5110482573509216,
      "learning_rate": 4.616535240978458e-06,
      "loss": 0.4048,
      "step": 34906
    },
    {
      "epoch": 1.6018998669175346,
      "grad_norm": 0.4976640045642853,
      "learning_rate": 4.616290775670155e-06,
      "loss": 0.4057,
      "step": 34907
    },
    {
      "epoch": 1.601945757422789,
      "grad_norm": 0.4749062657356262,
      "learning_rate": 4.616046311284551e-06,
      "loss": 0.3912,
      "step": 34908
    },
    {
      "epoch": 1.6019916479280436,
      "grad_norm": 0.4628889560699463,
      "learning_rate": 4.615801847822236e-06,
      "loss": 0.4016,
      "step": 34909
    },
    {
      "epoch": 1.602037538433298,
      "grad_norm": 0.5357798337936401,
      "learning_rate": 4.615557385283795e-06,
      "loss": 0.4365,
      "step": 34910
    },
    {
      "epoch": 1.6020834289385526,
      "grad_norm": 0.49128544330596924,
      "learning_rate": 4.6153129236698185e-06,
      "loss": 0.3982,
      "step": 34911
    },
    {
      "epoch": 1.602129319443807,
      "grad_norm": 0.4567388892173767,
      "learning_rate": 4.61506846298089e-06,
      "loss": 0.3549,
      "step": 34912
    },
    {
      "epoch": 1.6021752099490616,
      "grad_norm": 0.45051223039627075,
      "learning_rate": 4.614824003217603e-06,
      "loss": 0.3122,
      "step": 34913
    },
    {
      "epoch": 1.602221100454316,
      "grad_norm": 0.4980641007423401,
      "learning_rate": 4.614579544380542e-06,
      "loss": 0.3671,
      "step": 34914
    },
    {
      "epoch": 1.6022669909595706,
      "grad_norm": 0.47340893745422363,
      "learning_rate": 4.614335086470295e-06,
      "loss": 0.3652,
      "step": 34915
    },
    {
      "epoch": 1.602312881464825,
      "grad_norm": 0.4830954074859619,
      "learning_rate": 4.614090629487452e-06,
      "loss": 0.3798,
      "step": 34916
    },
    {
      "epoch": 1.6023587719700794,
      "grad_norm": 0.4438508450984955,
      "learning_rate": 4.613846173432599e-06,
      "loss": 0.361,
      "step": 34917
    },
    {
      "epoch": 1.6024046624753339,
      "grad_norm": 0.46878087520599365,
      "learning_rate": 4.613601718306321e-06,
      "loss": 0.4119,
      "step": 34918
    },
    {
      "epoch": 1.6024505529805884,
      "grad_norm": 0.4310548007488251,
      "learning_rate": 4.613357264109212e-06,
      "loss": 0.3367,
      "step": 34919
    },
    {
      "epoch": 1.6024964434858426,
      "grad_norm": 0.4591136574745178,
      "learning_rate": 4.613112810841856e-06,
      "loss": 0.2926,
      "step": 34920
    },
    {
      "epoch": 1.6025423339910971,
      "grad_norm": 0.4821411371231079,
      "learning_rate": 4.612868358504841e-06,
      "loss": 0.3985,
      "step": 34921
    },
    {
      "epoch": 1.6025882244963516,
      "grad_norm": 0.47148606181144714,
      "learning_rate": 4.6126239070987565e-06,
      "loss": 0.3793,
      "step": 34922
    },
    {
      "epoch": 1.6026341150016061,
      "grad_norm": 0.4340754747390747,
      "learning_rate": 4.612379456624189e-06,
      "loss": 0.328,
      "step": 34923
    },
    {
      "epoch": 1.6026800055068606,
      "grad_norm": 0.49296459555625916,
      "learning_rate": 4.612135007081724e-06,
      "loss": 0.3398,
      "step": 34924
    },
    {
      "epoch": 1.6027258960121151,
      "grad_norm": 0.4422965943813324,
      "learning_rate": 4.6118905584719554e-06,
      "loss": 0.3235,
      "step": 34925
    },
    {
      "epoch": 1.6027717865173696,
      "grad_norm": 0.45875436067581177,
      "learning_rate": 4.611646110795467e-06,
      "loss": 0.32,
      "step": 34926
    },
    {
      "epoch": 1.6028176770226241,
      "grad_norm": 0.483359158039093,
      "learning_rate": 4.611401664052845e-06,
      "loss": 0.4032,
      "step": 34927
    },
    {
      "epoch": 1.6028635675278786,
      "grad_norm": 0.47151580452919006,
      "learning_rate": 4.611157218244682e-06,
      "loss": 0.3194,
      "step": 34928
    },
    {
      "epoch": 1.602909458033133,
      "grad_norm": 0.4892735183238983,
      "learning_rate": 4.610912773371562e-06,
      "loss": 0.4292,
      "step": 34929
    },
    {
      "epoch": 1.6029553485383874,
      "grad_norm": 0.5474965572357178,
      "learning_rate": 4.610668329434073e-06,
      "loss": 0.412,
      "step": 34930
    },
    {
      "epoch": 1.603001239043642,
      "grad_norm": 0.445709764957428,
      "learning_rate": 4.6104238864328054e-06,
      "loss": 0.2953,
      "step": 34931
    },
    {
      "epoch": 1.6030471295488964,
      "grad_norm": 0.47554370760917664,
      "learning_rate": 4.610179444368345e-06,
      "loss": 0.3905,
      "step": 34932
    },
    {
      "epoch": 1.6030930200541507,
      "grad_norm": 0.4601365625858307,
      "learning_rate": 4.609935003241278e-06,
      "loss": 0.3563,
      "step": 34933
    },
    {
      "epoch": 1.6031389105594052,
      "grad_norm": 0.47242823243141174,
      "learning_rate": 4.6096905630521966e-06,
      "loss": 0.3517,
      "step": 34934
    },
    {
      "epoch": 1.6031848010646597,
      "grad_norm": 0.470455527305603,
      "learning_rate": 4.609446123801686e-06,
      "loss": 0.3156,
      "step": 34935
    },
    {
      "epoch": 1.6032306915699142,
      "grad_norm": 0.4755844473838806,
      "learning_rate": 4.609201685490333e-06,
      "loss": 0.3582,
      "step": 34936
    },
    {
      "epoch": 1.6032765820751687,
      "grad_norm": 0.4406697750091553,
      "learning_rate": 4.608957248118728e-06,
      "loss": 0.2844,
      "step": 34937
    },
    {
      "epoch": 1.6033224725804232,
      "grad_norm": 0.5107644200325012,
      "learning_rate": 4.608712811687457e-06,
      "loss": 0.3518,
      "step": 34938
    },
    {
      "epoch": 1.6033683630856777,
      "grad_norm": 0.41964754462242126,
      "learning_rate": 4.608468376197106e-06,
      "loss": 0.2654,
      "step": 34939
    },
    {
      "epoch": 1.6034142535909321,
      "grad_norm": 0.5450202822685242,
      "learning_rate": 4.608223941648268e-06,
      "loss": 0.3195,
      "step": 34940
    },
    {
      "epoch": 1.6034601440961866,
      "grad_norm": 0.5180357098579407,
      "learning_rate": 4.607979508041527e-06,
      "loss": 0.4253,
      "step": 34941
    },
    {
      "epoch": 1.603506034601441,
      "grad_norm": 0.44706830382347107,
      "learning_rate": 4.607735075377471e-06,
      "loss": 0.3458,
      "step": 34942
    },
    {
      "epoch": 1.6035519251066954,
      "grad_norm": 0.5048439502716064,
      "learning_rate": 4.607490643656688e-06,
      "loss": 0.3575,
      "step": 34943
    },
    {
      "epoch": 1.60359781561195,
      "grad_norm": 0.4818957448005676,
      "learning_rate": 4.607246212879768e-06,
      "loss": 0.3827,
      "step": 34944
    },
    {
      "epoch": 1.6036437061172042,
      "grad_norm": 0.480134516954422,
      "learning_rate": 4.607001783047296e-06,
      "loss": 0.4046,
      "step": 34945
    },
    {
      "epoch": 1.6036895966224587,
      "grad_norm": 0.47946253418922424,
      "learning_rate": 4.606757354159858e-06,
      "loss": 0.3342,
      "step": 34946
    },
    {
      "epoch": 1.6037354871277132,
      "grad_norm": 0.4680171310901642,
      "learning_rate": 4.606512926218048e-06,
      "loss": 0.3463,
      "step": 34947
    },
    {
      "epoch": 1.6037813776329677,
      "grad_norm": 0.4684879183769226,
      "learning_rate": 4.606268499222449e-06,
      "loss": 0.4038,
      "step": 34948
    },
    {
      "epoch": 1.6038272681382222,
      "grad_norm": 0.46269285678863525,
      "learning_rate": 4.606024073173649e-06,
      "loss": 0.3487,
      "step": 34949
    },
    {
      "epoch": 1.6038731586434767,
      "grad_norm": 0.4904498755931854,
      "learning_rate": 4.605779648072238e-06,
      "loss": 0.3969,
      "step": 34950
    },
    {
      "epoch": 1.6039190491487312,
      "grad_norm": 0.44177570939064026,
      "learning_rate": 4.6055352239188035e-06,
      "loss": 0.294,
      "step": 34951
    },
    {
      "epoch": 1.6039649396539857,
      "grad_norm": 0.4468318819999695,
      "learning_rate": 4.6052908007139305e-06,
      "loss": 0.3515,
      "step": 34952
    },
    {
      "epoch": 1.6040108301592402,
      "grad_norm": 0.427727609872818,
      "learning_rate": 4.6050463784582095e-06,
      "loss": 0.2729,
      "step": 34953
    },
    {
      "epoch": 1.6040567206644947,
      "grad_norm": 0.4571574330329895,
      "learning_rate": 4.604801957152228e-06,
      "loss": 0.3657,
      "step": 34954
    },
    {
      "epoch": 1.604102611169749,
      "grad_norm": 0.47174879908561707,
      "learning_rate": 4.604557536796569e-06,
      "loss": 0.3387,
      "step": 34955
    },
    {
      "epoch": 1.6041485016750034,
      "grad_norm": 0.44028279185295105,
      "learning_rate": 4.604313117391828e-06,
      "loss": 0.3213,
      "step": 34956
    },
    {
      "epoch": 1.604194392180258,
      "grad_norm": 0.4432438313961029,
      "learning_rate": 4.604068698938589e-06,
      "loss": 0.3194,
      "step": 34957
    },
    {
      "epoch": 1.6042402826855122,
      "grad_norm": 0.48729759454727173,
      "learning_rate": 4.6038242814374385e-06,
      "loss": 0.3959,
      "step": 34958
    },
    {
      "epoch": 1.6042861731907667,
      "grad_norm": 0.4271671175956726,
      "learning_rate": 4.603579864888967e-06,
      "loss": 0.2718,
      "step": 34959
    },
    {
      "epoch": 1.6043320636960212,
      "grad_norm": 0.45247501134872437,
      "learning_rate": 4.60333544929376e-06,
      "loss": 0.3168,
      "step": 34960
    },
    {
      "epoch": 1.6043779542012757,
      "grad_norm": 0.5119667649269104,
      "learning_rate": 4.603091034652405e-06,
      "loss": 0.3251,
      "step": 34961
    },
    {
      "epoch": 1.6044238447065302,
      "grad_norm": 0.4681013822555542,
      "learning_rate": 4.602846620965492e-06,
      "loss": 0.3724,
      "step": 34962
    },
    {
      "epoch": 1.6044697352117847,
      "grad_norm": 0.45092830061912537,
      "learning_rate": 4.6026022082336085e-06,
      "loss": 0.3511,
      "step": 34963
    },
    {
      "epoch": 1.6045156257170392,
      "grad_norm": 0.5134447813034058,
      "learning_rate": 4.602357796457339e-06,
      "loss": 0.4324,
      "step": 34964
    },
    {
      "epoch": 1.6045615162222937,
      "grad_norm": 0.4170004725456238,
      "learning_rate": 4.6021133856372755e-06,
      "loss": 0.2772,
      "step": 34965
    },
    {
      "epoch": 1.6046074067275482,
      "grad_norm": 0.4513907730579376,
      "learning_rate": 4.601868975774003e-06,
      "loss": 0.2992,
      "step": 34966
    },
    {
      "epoch": 1.6046532972328025,
      "grad_norm": 0.44555824995040894,
      "learning_rate": 4.6016245668681085e-06,
      "loss": 0.3266,
      "step": 34967
    },
    {
      "epoch": 1.604699187738057,
      "grad_norm": 0.4610910415649414,
      "learning_rate": 4.601380158920183e-06,
      "loss": 0.3007,
      "step": 34968
    },
    {
      "epoch": 1.6047450782433115,
      "grad_norm": 0.4259757101535797,
      "learning_rate": 4.601135751930812e-06,
      "loss": 0.2806,
      "step": 34969
    },
    {
      "epoch": 1.604790968748566,
      "grad_norm": 0.551280677318573,
      "learning_rate": 4.600891345900583e-06,
      "loss": 0.4716,
      "step": 34970
    },
    {
      "epoch": 1.6048368592538202,
      "grad_norm": 0.4932110905647278,
      "learning_rate": 4.600646940830086e-06,
      "loss": 0.4521,
      "step": 34971
    },
    {
      "epoch": 1.6048827497590747,
      "grad_norm": 0.4629458785057068,
      "learning_rate": 4.600402536719906e-06,
      "loss": 0.3553,
      "step": 34972
    },
    {
      "epoch": 1.6049286402643292,
      "grad_norm": 0.48184314370155334,
      "learning_rate": 4.60015813357063e-06,
      "loss": 0.3714,
      "step": 34973
    },
    {
      "epoch": 1.6049745307695837,
      "grad_norm": 0.5285406708717346,
      "learning_rate": 4.599913731382851e-06,
      "loss": 0.5326,
      "step": 34974
    },
    {
      "epoch": 1.6050204212748382,
      "grad_norm": 0.46760788559913635,
      "learning_rate": 4.599669330157152e-06,
      "loss": 0.3404,
      "step": 34975
    },
    {
      "epoch": 1.6050663117800927,
      "grad_norm": 0.45487216114997864,
      "learning_rate": 4.599424929894122e-06,
      "loss": 0.3195,
      "step": 34976
    },
    {
      "epoch": 1.6051122022853472,
      "grad_norm": 0.42622101306915283,
      "learning_rate": 4.599180530594346e-06,
      "loss": 0.2965,
      "step": 34977
    },
    {
      "epoch": 1.6051580927906017,
      "grad_norm": 0.48806440830230713,
      "learning_rate": 4.5989361322584165e-06,
      "loss": 0.4017,
      "step": 34978
    },
    {
      "epoch": 1.6052039832958562,
      "grad_norm": 0.4414578378200531,
      "learning_rate": 4.59869173488692e-06,
      "loss": 0.2864,
      "step": 34979
    },
    {
      "epoch": 1.6052498738011105,
      "grad_norm": 0.46192610263824463,
      "learning_rate": 4.598447338480442e-06,
      "loss": 0.3431,
      "step": 34980
    },
    {
      "epoch": 1.605295764306365,
      "grad_norm": 0.5118463635444641,
      "learning_rate": 4.598202943039571e-06,
      "loss": 0.4174,
      "step": 34981
    },
    {
      "epoch": 1.6053416548116195,
      "grad_norm": 0.48298126459121704,
      "learning_rate": 4.597958548564897e-06,
      "loss": 0.3425,
      "step": 34982
    },
    {
      "epoch": 1.6053875453168738,
      "grad_norm": 0.47807198762893677,
      "learning_rate": 4.597714155057003e-06,
      "loss": 0.326,
      "step": 34983
    },
    {
      "epoch": 1.6054334358221283,
      "grad_norm": 0.5108973979949951,
      "learning_rate": 4.597469762516481e-06,
      "loss": 0.4253,
      "step": 34984
    },
    {
      "epoch": 1.6054793263273828,
      "grad_norm": 0.4417910575866699,
      "learning_rate": 4.5972253709439176e-06,
      "loss": 0.3258,
      "step": 34985
    },
    {
      "epoch": 1.6055252168326373,
      "grad_norm": 0.4860847592353821,
      "learning_rate": 4.596980980339899e-06,
      "loss": 0.3759,
      "step": 34986
    },
    {
      "epoch": 1.6055711073378918,
      "grad_norm": 0.4457682967185974,
      "learning_rate": 4.596736590705015e-06,
      "loss": 0.3149,
      "step": 34987
    },
    {
      "epoch": 1.6056169978431463,
      "grad_norm": 0.4593941867351532,
      "learning_rate": 4.5964922020398515e-06,
      "loss": 0.3082,
      "step": 34988
    },
    {
      "epoch": 1.6056628883484008,
      "grad_norm": 0.40664488077163696,
      "learning_rate": 4.596247814344996e-06,
      "loss": 0.2702,
      "step": 34989
    },
    {
      "epoch": 1.6057087788536553,
      "grad_norm": 0.4798348844051361,
      "learning_rate": 4.596003427621038e-06,
      "loss": 0.3714,
      "step": 34990
    },
    {
      "epoch": 1.6057546693589098,
      "grad_norm": 0.48690807819366455,
      "learning_rate": 4.5957590418685655e-06,
      "loss": 0.3834,
      "step": 34991
    },
    {
      "epoch": 1.6058005598641643,
      "grad_norm": 0.4945811331272125,
      "learning_rate": 4.595514657088163e-06,
      "loss": 0.3859,
      "step": 34992
    },
    {
      "epoch": 1.6058464503694185,
      "grad_norm": 0.46243277192115784,
      "learning_rate": 4.595270273280422e-06,
      "loss": 0.3429,
      "step": 34993
    },
    {
      "epoch": 1.605892340874673,
      "grad_norm": 0.496953547000885,
      "learning_rate": 4.595025890445928e-06,
      "loss": 0.4086,
      "step": 34994
    },
    {
      "epoch": 1.6059382313799275,
      "grad_norm": 0.497714638710022,
      "learning_rate": 4.594781508585266e-06,
      "loss": 0.3612,
      "step": 34995
    },
    {
      "epoch": 1.6059841218851818,
      "grad_norm": 0.4800831973552704,
      "learning_rate": 4.59453712769903e-06,
      "loss": 0.3651,
      "step": 34996
    },
    {
      "epoch": 1.6060300123904363,
      "grad_norm": 0.4632662236690521,
      "learning_rate": 4.5942927477878044e-06,
      "loss": 0.3492,
      "step": 34997
    },
    {
      "epoch": 1.6060759028956908,
      "grad_norm": 0.47459548711776733,
      "learning_rate": 4.594048368852173e-06,
      "loss": 0.3465,
      "step": 34998
    },
    {
      "epoch": 1.6061217934009453,
      "grad_norm": 0.49554672837257385,
      "learning_rate": 4.59380399089273e-06,
      "loss": 0.3921,
      "step": 34999
    },
    {
      "epoch": 1.6061676839061998,
      "grad_norm": 0.4706018567085266,
      "learning_rate": 4.59355961391006e-06,
      "loss": 0.3624,
      "step": 35000
    },
    {
      "epoch": 1.6062135744114543,
      "grad_norm": 0.46854308247566223,
      "learning_rate": 4.59331523790475e-06,
      "loss": 0.3659,
      "step": 35001
    },
    {
      "epoch": 1.6062594649167088,
      "grad_norm": 0.46793970465660095,
      "learning_rate": 4.5930708628773894e-06,
      "loss": 0.3359,
      "step": 35002
    },
    {
      "epoch": 1.6063053554219633,
      "grad_norm": 0.4724282920360565,
      "learning_rate": 4.592826488828565e-06,
      "loss": 0.4074,
      "step": 35003
    },
    {
      "epoch": 1.6063512459272178,
      "grad_norm": 0.4942350685596466,
      "learning_rate": 4.592582115758862e-06,
      "loss": 0.3454,
      "step": 35004
    },
    {
      "epoch": 1.6063971364324723,
      "grad_norm": 0.44197481870651245,
      "learning_rate": 4.592337743668873e-06,
      "loss": 0.3143,
      "step": 35005
    },
    {
      "epoch": 1.6064430269377266,
      "grad_norm": 0.44394582509994507,
      "learning_rate": 4.592093372559183e-06,
      "loss": 0.3826,
      "step": 35006
    },
    {
      "epoch": 1.606488917442981,
      "grad_norm": 0.5206359028816223,
      "learning_rate": 4.591849002430379e-06,
      "loss": 0.39,
      "step": 35007
    },
    {
      "epoch": 1.6065348079482356,
      "grad_norm": 0.4473307132720947,
      "learning_rate": 4.591604633283049e-06,
      "loss": 0.3287,
      "step": 35008
    },
    {
      "epoch": 1.6065806984534898,
      "grad_norm": 0.39300963282585144,
      "learning_rate": 4.5913602651177824e-06,
      "loss": 0.254,
      "step": 35009
    },
    {
      "epoch": 1.6066265889587443,
      "grad_norm": 0.45773744583129883,
      "learning_rate": 4.5911158979351645e-06,
      "loss": 0.3756,
      "step": 35010
    },
    {
      "epoch": 1.6066724794639988,
      "grad_norm": 0.4700591266155243,
      "learning_rate": 4.5908715317357825e-06,
      "loss": 0.3788,
      "step": 35011
    },
    {
      "epoch": 1.6067183699692533,
      "grad_norm": 0.44999608397483826,
      "learning_rate": 4.590627166520227e-06,
      "loss": 0.2889,
      "step": 35012
    },
    {
      "epoch": 1.6067642604745078,
      "grad_norm": 0.4843432605266571,
      "learning_rate": 4.590382802289084e-06,
      "loss": 0.3788,
      "step": 35013
    },
    {
      "epoch": 1.6068101509797623,
      "grad_norm": 0.46214428544044495,
      "learning_rate": 4.59013843904294e-06,
      "loss": 0.3299,
      "step": 35014
    },
    {
      "epoch": 1.6068560414850168,
      "grad_norm": 0.4524962604045868,
      "learning_rate": 4.589894076782385e-06,
      "loss": 0.345,
      "step": 35015
    },
    {
      "epoch": 1.6069019319902713,
      "grad_norm": 0.47244617342948914,
      "learning_rate": 4.589649715508005e-06,
      "loss": 0.428,
      "step": 35016
    },
    {
      "epoch": 1.6069478224955258,
      "grad_norm": 0.4420822262763977,
      "learning_rate": 4.589405355220386e-06,
      "loss": 0.3157,
      "step": 35017
    },
    {
      "epoch": 1.60699371300078,
      "grad_norm": 0.42337849736213684,
      "learning_rate": 4.58916099592012e-06,
      "loss": 0.2683,
      "step": 35018
    },
    {
      "epoch": 1.6070396035060346,
      "grad_norm": 0.4918625056743622,
      "learning_rate": 4.588916637607792e-06,
      "loss": 0.3971,
      "step": 35019
    },
    {
      "epoch": 1.607085494011289,
      "grad_norm": 0.44537779688835144,
      "learning_rate": 4.5886722802839865e-06,
      "loss": 0.3095,
      "step": 35020
    },
    {
      "epoch": 1.6071313845165436,
      "grad_norm": 0.6261507272720337,
      "learning_rate": 4.5884279239492975e-06,
      "loss": 0.4151,
      "step": 35021
    },
    {
      "epoch": 1.6071772750217979,
      "grad_norm": 0.4522685110569,
      "learning_rate": 4.58818356860431e-06,
      "loss": 0.3084,
      "step": 35022
    },
    {
      "epoch": 1.6072231655270524,
      "grad_norm": 0.48193398118019104,
      "learning_rate": 4.587939214249609e-06,
      "loss": 0.3745,
      "step": 35023
    },
    {
      "epoch": 1.6072690560323069,
      "grad_norm": 0.45002859830856323,
      "learning_rate": 4.587694860885785e-06,
      "loss": 0.3298,
      "step": 35024
    },
    {
      "epoch": 1.6073149465375614,
      "grad_norm": 0.4775749146938324,
      "learning_rate": 4.587450508513425e-06,
      "loss": 0.3124,
      "step": 35025
    },
    {
      "epoch": 1.6073608370428158,
      "grad_norm": 0.4787912964820862,
      "learning_rate": 4.587206157133115e-06,
      "loss": 0.3941,
      "step": 35026
    },
    {
      "epoch": 1.6074067275480703,
      "grad_norm": 0.48591986298561096,
      "learning_rate": 4.586961806745445e-06,
      "loss": 0.3709,
      "step": 35027
    },
    {
      "epoch": 1.6074526180533248,
      "grad_norm": 0.49338817596435547,
      "learning_rate": 4.586717457351002e-06,
      "loss": 0.4213,
      "step": 35028
    },
    {
      "epoch": 1.6074985085585793,
      "grad_norm": 0.4702754616737366,
      "learning_rate": 4.586473108950371e-06,
      "loss": 0.3812,
      "step": 35029
    },
    {
      "epoch": 1.6075443990638338,
      "grad_norm": 0.4584999978542328,
      "learning_rate": 4.586228761544144e-06,
      "loss": 0.333,
      "step": 35030
    },
    {
      "epoch": 1.6075902895690881,
      "grad_norm": 0.45198914408683777,
      "learning_rate": 4.585984415132906e-06,
      "loss": 0.3395,
      "step": 35031
    },
    {
      "epoch": 1.6076361800743426,
      "grad_norm": 0.505451500415802,
      "learning_rate": 4.5857400697172425e-06,
      "loss": 0.3819,
      "step": 35032
    },
    {
      "epoch": 1.6076820705795971,
      "grad_norm": 0.49722203612327576,
      "learning_rate": 4.585495725297746e-06,
      "loss": 0.3612,
      "step": 35033
    },
    {
      "epoch": 1.6077279610848514,
      "grad_norm": 0.4900634288787842,
      "learning_rate": 4.585251381875001e-06,
      "loss": 0.3631,
      "step": 35034
    },
    {
      "epoch": 1.6077738515901059,
      "grad_norm": 0.43958601355552673,
      "learning_rate": 4.585007039449594e-06,
      "loss": 0.3253,
      "step": 35035
    },
    {
      "epoch": 1.6078197420953604,
      "grad_norm": 0.4619571566581726,
      "learning_rate": 4.584762698022116e-06,
      "loss": 0.3361,
      "step": 35036
    },
    {
      "epoch": 1.6078656326006149,
      "grad_norm": 0.47435927391052246,
      "learning_rate": 4.584518357593152e-06,
      "loss": 0.3125,
      "step": 35037
    },
    {
      "epoch": 1.6079115231058694,
      "grad_norm": 0.42160356044769287,
      "learning_rate": 4.584274018163288e-06,
      "loss": 0.2796,
      "step": 35038
    },
    {
      "epoch": 1.6079574136111239,
      "grad_norm": 0.4869803786277771,
      "learning_rate": 4.584029679733117e-06,
      "loss": 0.3574,
      "step": 35039
    },
    {
      "epoch": 1.6080033041163784,
      "grad_norm": 0.44030439853668213,
      "learning_rate": 4.583785342303224e-06,
      "loss": 0.2779,
      "step": 35040
    },
    {
      "epoch": 1.6080491946216329,
      "grad_norm": 0.48289743065834045,
      "learning_rate": 4.583541005874192e-06,
      "loss": 0.3653,
      "step": 35041
    },
    {
      "epoch": 1.6080950851268874,
      "grad_norm": 0.45983409881591797,
      "learning_rate": 4.583296670446615e-06,
      "loss": 0.2891,
      "step": 35042
    },
    {
      "epoch": 1.6081409756321419,
      "grad_norm": 0.45919692516326904,
      "learning_rate": 4.583052336021078e-06,
      "loss": 0.3127,
      "step": 35043
    },
    {
      "epoch": 1.6081868661373961,
      "grad_norm": 0.4721025824546814,
      "learning_rate": 4.582808002598168e-06,
      "loss": 0.3638,
      "step": 35044
    },
    {
      "epoch": 1.6082327566426506,
      "grad_norm": 0.46395421028137207,
      "learning_rate": 4.582563670178473e-06,
      "loss": 0.3389,
      "step": 35045
    },
    {
      "epoch": 1.6082786471479051,
      "grad_norm": 0.4705526828765869,
      "learning_rate": 4.582319338762581e-06,
      "loss": 0.3754,
      "step": 35046
    },
    {
      "epoch": 1.6083245376531594,
      "grad_norm": 0.48528969287872314,
      "learning_rate": 4.582075008351079e-06,
      "loss": 0.3405,
      "step": 35047
    },
    {
      "epoch": 1.608370428158414,
      "grad_norm": 0.5044940710067749,
      "learning_rate": 4.581830678944553e-06,
      "loss": 0.4444,
      "step": 35048
    },
    {
      "epoch": 1.6084163186636684,
      "grad_norm": 0.5420824289321899,
      "learning_rate": 4.5815863505435945e-06,
      "loss": 0.5006,
      "step": 35049
    },
    {
      "epoch": 1.608462209168923,
      "grad_norm": 0.45507562160491943,
      "learning_rate": 4.581342023148789e-06,
      "loss": 0.3111,
      "step": 35050
    },
    {
      "epoch": 1.6085080996741774,
      "grad_norm": 0.48054614663124084,
      "learning_rate": 4.581097696760722e-06,
      "loss": 0.388,
      "step": 35051
    },
    {
      "epoch": 1.608553990179432,
      "grad_norm": 0.47234001755714417,
      "learning_rate": 4.580853371379984e-06,
      "loss": 0.3648,
      "step": 35052
    },
    {
      "epoch": 1.6085998806846864,
      "grad_norm": 0.45196667313575745,
      "learning_rate": 4.580609047007162e-06,
      "loss": 0.307,
      "step": 35053
    },
    {
      "epoch": 1.608645771189941,
      "grad_norm": 0.5005353689193726,
      "learning_rate": 4.58036472364284e-06,
      "loss": 0.3713,
      "step": 35054
    },
    {
      "epoch": 1.6086916616951954,
      "grad_norm": 0.5007423162460327,
      "learning_rate": 4.58012040128761e-06,
      "loss": 0.3275,
      "step": 35055
    },
    {
      "epoch": 1.6087375522004497,
      "grad_norm": 0.462198942899704,
      "learning_rate": 4.579876079942059e-06,
      "loss": 0.3652,
      "step": 35056
    },
    {
      "epoch": 1.6087834427057042,
      "grad_norm": 0.9332163333892822,
      "learning_rate": 4.579631759606773e-06,
      "loss": 0.4063,
      "step": 35057
    },
    {
      "epoch": 1.6088293332109587,
      "grad_norm": 0.4544518291950226,
      "learning_rate": 4.57938744028234e-06,
      "loss": 0.3079,
      "step": 35058
    },
    {
      "epoch": 1.6088752237162132,
      "grad_norm": 0.45464959740638733,
      "learning_rate": 4.579143121969347e-06,
      "loss": 0.3192,
      "step": 35059
    },
    {
      "epoch": 1.6089211142214674,
      "grad_norm": 0.46009159088134766,
      "learning_rate": 4.578898804668381e-06,
      "loss": 0.3738,
      "step": 35060
    },
    {
      "epoch": 1.608967004726722,
      "grad_norm": 0.4810626804828644,
      "learning_rate": 4.5786544883800315e-06,
      "loss": 0.4268,
      "step": 35061
    },
    {
      "epoch": 1.6090128952319764,
      "grad_norm": 0.4706176817417145,
      "learning_rate": 4.578410173104887e-06,
      "loss": 0.3641,
      "step": 35062
    },
    {
      "epoch": 1.609058785737231,
      "grad_norm": 0.48172464966773987,
      "learning_rate": 4.578165858843529e-06,
      "loss": 0.3782,
      "step": 35063
    },
    {
      "epoch": 1.6091046762424854,
      "grad_norm": 0.7682211995124817,
      "learning_rate": 4.577921545596551e-06,
      "loss": 0.373,
      "step": 35064
    },
    {
      "epoch": 1.60915056674774,
      "grad_norm": 0.47874733805656433,
      "learning_rate": 4.577677233364539e-06,
      "loss": 0.3358,
      "step": 35065
    },
    {
      "epoch": 1.6091964572529944,
      "grad_norm": 0.4851420819759369,
      "learning_rate": 4.577432922148079e-06,
      "loss": 0.3432,
      "step": 35066
    },
    {
      "epoch": 1.609242347758249,
      "grad_norm": 0.457152396440506,
      "learning_rate": 4.577188611947761e-06,
      "loss": 0.3756,
      "step": 35067
    },
    {
      "epoch": 1.6092882382635034,
      "grad_norm": 0.5096479654312134,
      "learning_rate": 4.57694430276417e-06,
      "loss": 0.4299,
      "step": 35068
    },
    {
      "epoch": 1.6093341287687577,
      "grad_norm": 0.4651198983192444,
      "learning_rate": 4.576699994597892e-06,
      "loss": 0.3187,
      "step": 35069
    },
    {
      "epoch": 1.6093800192740122,
      "grad_norm": 0.5461528897285461,
      "learning_rate": 4.57645568744952e-06,
      "loss": 0.4453,
      "step": 35070
    },
    {
      "epoch": 1.6094259097792667,
      "grad_norm": 0.44050878286361694,
      "learning_rate": 4.576211381319639e-06,
      "loss": 0.3159,
      "step": 35071
    },
    {
      "epoch": 1.609471800284521,
      "grad_norm": 0.4955904483795166,
      "learning_rate": 4.575967076208833e-06,
      "loss": 0.3812,
      "step": 35072
    },
    {
      "epoch": 1.6095176907897755,
      "grad_norm": 0.4605967700481415,
      "learning_rate": 4.5757227721176945e-06,
      "loss": 0.3286,
      "step": 35073
    },
    {
      "epoch": 1.60956358129503,
      "grad_norm": 0.49206939339637756,
      "learning_rate": 4.575478469046808e-06,
      "loss": 0.4125,
      "step": 35074
    },
    {
      "epoch": 1.6096094718002845,
      "grad_norm": 0.4736131727695465,
      "learning_rate": 4.575234166996763e-06,
      "loss": 0.3891,
      "step": 35075
    },
    {
      "epoch": 1.609655362305539,
      "grad_norm": 0.5139959454536438,
      "learning_rate": 4.574989865968143e-06,
      "loss": 0.4505,
      "step": 35076
    },
    {
      "epoch": 1.6097012528107935,
      "grad_norm": 0.457712322473526,
      "learning_rate": 4.57474556596154e-06,
      "loss": 0.3508,
      "step": 35077
    },
    {
      "epoch": 1.609747143316048,
      "grad_norm": 0.46004945039749146,
      "learning_rate": 4.5745012669775405e-06,
      "loss": 0.3284,
      "step": 35078
    },
    {
      "epoch": 1.6097930338213025,
      "grad_norm": 0.8467329144477844,
      "learning_rate": 4.57425696901673e-06,
      "loss": 0.3715,
      "step": 35079
    },
    {
      "epoch": 1.609838924326557,
      "grad_norm": 0.4870775043964386,
      "learning_rate": 4.574012672079697e-06,
      "loss": 0.399,
      "step": 35080
    },
    {
      "epoch": 1.6098848148318115,
      "grad_norm": 0.4898388981819153,
      "learning_rate": 4.5737683761670304e-06,
      "loss": 0.3847,
      "step": 35081
    },
    {
      "epoch": 1.6099307053370657,
      "grad_norm": 0.47500932216644287,
      "learning_rate": 4.573524081279313e-06,
      "loss": 0.3699,
      "step": 35082
    },
    {
      "epoch": 1.6099765958423202,
      "grad_norm": 0.4638137221336365,
      "learning_rate": 4.5732797874171395e-06,
      "loss": 0.3206,
      "step": 35083
    },
    {
      "epoch": 1.6100224863475747,
      "grad_norm": 0.44559910893440247,
      "learning_rate": 4.573035494581094e-06,
      "loss": 0.3124,
      "step": 35084
    },
    {
      "epoch": 1.610068376852829,
      "grad_norm": 0.42249950766563416,
      "learning_rate": 4.572791202771759e-06,
      "loss": 0.279,
      "step": 35085
    },
    {
      "epoch": 1.6101142673580835,
      "grad_norm": 0.4786585569381714,
      "learning_rate": 4.5725469119897295e-06,
      "loss": 0.3613,
      "step": 35086
    },
    {
      "epoch": 1.610160157863338,
      "grad_norm": 0.4372207522392273,
      "learning_rate": 4.572302622235589e-06,
      "loss": 0.339,
      "step": 35087
    },
    {
      "epoch": 1.6102060483685925,
      "grad_norm": 0.49437057971954346,
      "learning_rate": 4.572058333509925e-06,
      "loss": 0.3851,
      "step": 35088
    },
    {
      "epoch": 1.610251938873847,
      "grad_norm": 0.8243091106414795,
      "learning_rate": 4.571814045813327e-06,
      "loss": 0.3276,
      "step": 35089
    },
    {
      "epoch": 1.6102978293791015,
      "grad_norm": 0.48955899477005005,
      "learning_rate": 4.5715697591463805e-06,
      "loss": 0.353,
      "step": 35090
    },
    {
      "epoch": 1.610343719884356,
      "grad_norm": 0.4993062913417816,
      "learning_rate": 4.571325473509672e-06,
      "loss": 0.3942,
      "step": 35091
    },
    {
      "epoch": 1.6103896103896105,
      "grad_norm": 0.49777162075042725,
      "learning_rate": 4.571081188903792e-06,
      "loss": 0.3697,
      "step": 35092
    },
    {
      "epoch": 1.610435500894865,
      "grad_norm": 0.4667954444885254,
      "learning_rate": 4.570836905329326e-06,
      "loss": 0.3616,
      "step": 35093
    },
    {
      "epoch": 1.6104813914001195,
      "grad_norm": 0.4534285366535187,
      "learning_rate": 4.570592622786862e-06,
      "loss": 0.33,
      "step": 35094
    },
    {
      "epoch": 1.6105272819053738,
      "grad_norm": 0.5129544138908386,
      "learning_rate": 4.570348341276987e-06,
      "loss": 0.4325,
      "step": 35095
    },
    {
      "epoch": 1.6105731724106283,
      "grad_norm": 0.48183056712150574,
      "learning_rate": 4.57010406080029e-06,
      "loss": 0.3672,
      "step": 35096
    },
    {
      "epoch": 1.6106190629158827,
      "grad_norm": 0.4934347867965698,
      "learning_rate": 4.5698597813573545e-06,
      "loss": 0.3836,
      "step": 35097
    },
    {
      "epoch": 1.610664953421137,
      "grad_norm": 0.7495447397232056,
      "learning_rate": 4.569615502948772e-06,
      "loss": 0.4272,
      "step": 35098
    },
    {
      "epoch": 1.6107108439263915,
      "grad_norm": 0.45012184977531433,
      "learning_rate": 4.5693712255751285e-06,
      "loss": 0.3135,
      "step": 35099
    },
    {
      "epoch": 1.610756734431646,
      "grad_norm": 0.4588436782360077,
      "learning_rate": 4.56912694923701e-06,
      "loss": 0.3192,
      "step": 35100
    },
    {
      "epoch": 1.6108026249369005,
      "grad_norm": 0.47536221146583557,
      "learning_rate": 4.568882673935007e-06,
      "loss": 0.3688,
      "step": 35101
    },
    {
      "epoch": 1.610848515442155,
      "grad_norm": 0.4482388198375702,
      "learning_rate": 4.5686383996697056e-06,
      "loss": 0.2877,
      "step": 35102
    },
    {
      "epoch": 1.6108944059474095,
      "grad_norm": 0.46007096767425537,
      "learning_rate": 4.568394126441689e-06,
      "loss": 0.3422,
      "step": 35103
    },
    {
      "epoch": 1.610940296452664,
      "grad_norm": 0.4419252872467041,
      "learning_rate": 4.568149854251552e-06,
      "loss": 0.3318,
      "step": 35104
    },
    {
      "epoch": 1.6109861869579185,
      "grad_norm": 0.5034579038619995,
      "learning_rate": 4.5679055830998766e-06,
      "loss": 0.3407,
      "step": 35105
    },
    {
      "epoch": 1.611032077463173,
      "grad_norm": 0.6039446592330933,
      "learning_rate": 4.567661312987252e-06,
      "loss": 0.3206,
      "step": 35106
    },
    {
      "epoch": 1.6110779679684273,
      "grad_norm": 0.4649903476238251,
      "learning_rate": 4.567417043914267e-06,
      "loss": 0.3338,
      "step": 35107
    },
    {
      "epoch": 1.6111238584736818,
      "grad_norm": 0.4459214210510254,
      "learning_rate": 4.5671727758815064e-06,
      "loss": 0.3114,
      "step": 35108
    },
    {
      "epoch": 1.6111697489789363,
      "grad_norm": 0.7183945178985596,
      "learning_rate": 4.566928508889559e-06,
      "loss": 0.4423,
      "step": 35109
    },
    {
      "epoch": 1.6112156394841908,
      "grad_norm": 0.47298669815063477,
      "learning_rate": 4.566684242939011e-06,
      "loss": 0.3921,
      "step": 35110
    },
    {
      "epoch": 1.611261529989445,
      "grad_norm": 0.43747541308403015,
      "learning_rate": 4.566439978030452e-06,
      "loss": 0.3031,
      "step": 35111
    },
    {
      "epoch": 1.6113074204946995,
      "grad_norm": 0.4505838453769684,
      "learning_rate": 4.566195714164467e-06,
      "loss": 0.322,
      "step": 35112
    },
    {
      "epoch": 1.611353310999954,
      "grad_norm": 0.5738093852996826,
      "learning_rate": 4.5659514513416424e-06,
      "loss": 0.472,
      "step": 35113
    },
    {
      "epoch": 1.6113992015052085,
      "grad_norm": 0.4550623893737793,
      "learning_rate": 4.56570718956257e-06,
      "loss": 0.3115,
      "step": 35114
    },
    {
      "epoch": 1.611445092010463,
      "grad_norm": 0.4841283857822418,
      "learning_rate": 4.565462928827835e-06,
      "loss": 0.3386,
      "step": 35115
    },
    {
      "epoch": 1.6114909825157175,
      "grad_norm": 0.46640875935554504,
      "learning_rate": 4.565218669138024e-06,
      "loss": 0.387,
      "step": 35116
    },
    {
      "epoch": 1.611536873020972,
      "grad_norm": 0.47442761063575745,
      "learning_rate": 4.564974410493725e-06,
      "loss": 0.3906,
      "step": 35117
    },
    {
      "epoch": 1.6115827635262265,
      "grad_norm": 0.47995996475219727,
      "learning_rate": 4.564730152895525e-06,
      "loss": 0.363,
      "step": 35118
    },
    {
      "epoch": 1.611628654031481,
      "grad_norm": 0.45361706614494324,
      "learning_rate": 4.56448589634401e-06,
      "loss": 0.2989,
      "step": 35119
    },
    {
      "epoch": 1.6116745445367353,
      "grad_norm": 0.5169030427932739,
      "learning_rate": 4.564241640839772e-06,
      "loss": 0.3516,
      "step": 35120
    },
    {
      "epoch": 1.6117204350419898,
      "grad_norm": 0.46917563676834106,
      "learning_rate": 4.5639973863833945e-06,
      "loss": 0.3247,
      "step": 35121
    },
    {
      "epoch": 1.6117663255472443,
      "grad_norm": 0.4739153981208801,
      "learning_rate": 4.563753132975464e-06,
      "loss": 0.4287,
      "step": 35122
    },
    {
      "epoch": 1.6118122160524986,
      "grad_norm": 0.47725197672843933,
      "learning_rate": 4.563508880616571e-06,
      "loss": 0.3578,
      "step": 35123
    },
    {
      "epoch": 1.611858106557753,
      "grad_norm": 0.4230272173881531,
      "learning_rate": 4.563264629307302e-06,
      "loss": 0.303,
      "step": 35124
    },
    {
      "epoch": 1.6119039970630076,
      "grad_norm": 0.476823627948761,
      "learning_rate": 4.563020379048241e-06,
      "loss": 0.3285,
      "step": 35125
    },
    {
      "epoch": 1.611949887568262,
      "grad_norm": 0.4682890772819519,
      "learning_rate": 4.5627761298399804e-06,
      "loss": 0.3332,
      "step": 35126
    },
    {
      "epoch": 1.6119957780735166,
      "grad_norm": 0.47459834814071655,
      "learning_rate": 4.562531881683105e-06,
      "loss": 0.4255,
      "step": 35127
    },
    {
      "epoch": 1.612041668578771,
      "grad_norm": 0.4748800992965698,
      "learning_rate": 4.562287634578201e-06,
      "loss": 0.3197,
      "step": 35128
    },
    {
      "epoch": 1.6120875590840256,
      "grad_norm": 0.4671265184879303,
      "learning_rate": 4.562043388525858e-06,
      "loss": 0.3562,
      "step": 35129
    },
    {
      "epoch": 1.61213344958928,
      "grad_norm": 0.4574284851551056,
      "learning_rate": 4.561799143526664e-06,
      "loss": 0.2941,
      "step": 35130
    },
    {
      "epoch": 1.6121793400945346,
      "grad_norm": 0.5040728449821472,
      "learning_rate": 4.561554899581202e-06,
      "loss": 0.4249,
      "step": 35131
    },
    {
      "epoch": 1.612225230599789,
      "grad_norm": 0.4620607793331146,
      "learning_rate": 4.561310656690065e-06,
      "loss": 0.3358,
      "step": 35132
    },
    {
      "epoch": 1.6122711211050433,
      "grad_norm": 0.4788861572742462,
      "learning_rate": 4.561066414853836e-06,
      "loss": 0.3303,
      "step": 35133
    },
    {
      "epoch": 1.6123170116102978,
      "grad_norm": 0.4754139482975006,
      "learning_rate": 4.560822174073102e-06,
      "loss": 0.3321,
      "step": 35134
    },
    {
      "epoch": 1.6123629021155523,
      "grad_norm": 0.46730247139930725,
      "learning_rate": 4.560577934348454e-06,
      "loss": 0.3793,
      "step": 35135
    },
    {
      "epoch": 1.6124087926208066,
      "grad_norm": 0.45752984285354614,
      "learning_rate": 4.560333695680478e-06,
      "loss": 0.3569,
      "step": 35136
    },
    {
      "epoch": 1.612454683126061,
      "grad_norm": 0.4766443371772766,
      "learning_rate": 4.560089458069758e-06,
      "loss": 0.4199,
      "step": 35137
    },
    {
      "epoch": 1.6125005736313156,
      "grad_norm": 0.4542084336280823,
      "learning_rate": 4.559845221516886e-06,
      "loss": 0.3573,
      "step": 35138
    },
    {
      "epoch": 1.61254646413657,
      "grad_norm": 0.537150502204895,
      "learning_rate": 4.5596009860224476e-06,
      "loss": 0.4694,
      "step": 35139
    },
    {
      "epoch": 1.6125923546418246,
      "grad_norm": 0.5264257788658142,
      "learning_rate": 4.5593567515870305e-06,
      "loss": 0.4516,
      "step": 35140
    },
    {
      "epoch": 1.612638245147079,
      "grad_norm": 0.4658612012863159,
      "learning_rate": 4.559112518211218e-06,
      "loss": 0.34,
      "step": 35141
    },
    {
      "epoch": 1.6126841356523336,
      "grad_norm": 0.5397464036941528,
      "learning_rate": 4.558868285895603e-06,
      "loss": 0.4702,
      "step": 35142
    },
    {
      "epoch": 1.612730026157588,
      "grad_norm": 0.482726126909256,
      "learning_rate": 4.558624054640771e-06,
      "loss": 0.3949,
      "step": 35143
    },
    {
      "epoch": 1.6127759166628426,
      "grad_norm": 0.4795769453048706,
      "learning_rate": 4.558379824447308e-06,
      "loss": 0.3266,
      "step": 35144
    },
    {
      "epoch": 1.6128218071680969,
      "grad_norm": 0.49074628949165344,
      "learning_rate": 4.558135595315803e-06,
      "loss": 0.3638,
      "step": 35145
    },
    {
      "epoch": 1.6128676976733514,
      "grad_norm": 0.4977889657020569,
      "learning_rate": 4.557891367246842e-06,
      "loss": 0.4562,
      "step": 35146
    },
    {
      "epoch": 1.6129135881786059,
      "grad_norm": 0.4551311731338501,
      "learning_rate": 4.55764714024101e-06,
      "loss": 0.2948,
      "step": 35147
    },
    {
      "epoch": 1.6129594786838604,
      "grad_norm": 0.4656989276409149,
      "learning_rate": 4.557402914298901e-06,
      "loss": 0.3634,
      "step": 35148
    },
    {
      "epoch": 1.6130053691891146,
      "grad_norm": 0.4575183391571045,
      "learning_rate": 4.557158689421097e-06,
      "loss": 0.3569,
      "step": 35149
    },
    {
      "epoch": 1.6130512596943691,
      "grad_norm": 0.4720980226993561,
      "learning_rate": 4.5569144656081855e-06,
      "loss": 0.345,
      "step": 35150
    },
    {
      "epoch": 1.6130971501996236,
      "grad_norm": 0.4591463506221771,
      "learning_rate": 4.556670242860757e-06,
      "loss": 0.3243,
      "step": 35151
    },
    {
      "epoch": 1.6131430407048781,
      "grad_norm": 0.48612621426582336,
      "learning_rate": 4.556426021179395e-06,
      "loss": 0.369,
      "step": 35152
    },
    {
      "epoch": 1.6131889312101326,
      "grad_norm": 0.4449271261692047,
      "learning_rate": 4.556181800564689e-06,
      "loss": 0.2802,
      "step": 35153
    },
    {
      "epoch": 1.6132348217153871,
      "grad_norm": 0.4914563298225403,
      "learning_rate": 4.555937581017226e-06,
      "loss": 0.4161,
      "step": 35154
    },
    {
      "epoch": 1.6132807122206416,
      "grad_norm": 0.45053595304489136,
      "learning_rate": 4.555693362537593e-06,
      "loss": 0.3382,
      "step": 35155
    },
    {
      "epoch": 1.6133266027258961,
      "grad_norm": 0.4463652968406677,
      "learning_rate": 4.555449145126375e-06,
      "loss": 0.345,
      "step": 35156
    },
    {
      "epoch": 1.6133724932311506,
      "grad_norm": 0.48425188660621643,
      "learning_rate": 4.555204928784164e-06,
      "loss": 0.3543,
      "step": 35157
    },
    {
      "epoch": 1.613418383736405,
      "grad_norm": 0.4901755154132843,
      "learning_rate": 4.554960713511544e-06,
      "loss": 0.3845,
      "step": 35158
    },
    {
      "epoch": 1.6134642742416594,
      "grad_norm": 0.5169627070426941,
      "learning_rate": 4.554716499309103e-06,
      "loss": 0.4247,
      "step": 35159
    },
    {
      "epoch": 1.6135101647469139,
      "grad_norm": 0.5189130902290344,
      "learning_rate": 4.554472286177429e-06,
      "loss": 0.4124,
      "step": 35160
    },
    {
      "epoch": 1.6135560552521682,
      "grad_norm": 0.49915915727615356,
      "learning_rate": 4.554228074117108e-06,
      "loss": 0.3925,
      "step": 35161
    },
    {
      "epoch": 1.6136019457574227,
      "grad_norm": 0.502739667892456,
      "learning_rate": 4.553983863128726e-06,
      "loss": 0.3996,
      "step": 35162
    },
    {
      "epoch": 1.6136478362626772,
      "grad_norm": 0.509726881980896,
      "learning_rate": 4.5537396532128745e-06,
      "loss": 0.4056,
      "step": 35163
    },
    {
      "epoch": 1.6136937267679317,
      "grad_norm": 0.4734102785587311,
      "learning_rate": 4.553495444370138e-06,
      "loss": 0.3524,
      "step": 35164
    },
    {
      "epoch": 1.6137396172731862,
      "grad_norm": 0.4455709159374237,
      "learning_rate": 4.553251236601103e-06,
      "loss": 0.3209,
      "step": 35165
    },
    {
      "epoch": 1.6137855077784407,
      "grad_norm": 0.48856645822525024,
      "learning_rate": 4.5530070299063586e-06,
      "loss": 0.3916,
      "step": 35166
    },
    {
      "epoch": 1.6138313982836952,
      "grad_norm": 0.48706477880477905,
      "learning_rate": 4.552762824286492e-06,
      "loss": 0.3467,
      "step": 35167
    },
    {
      "epoch": 1.6138772887889496,
      "grad_norm": 0.4798251688480377,
      "learning_rate": 4.552518619742086e-06,
      "loss": 0.3963,
      "step": 35168
    },
    {
      "epoch": 1.6139231792942041,
      "grad_norm": 0.47593826055526733,
      "learning_rate": 4.552274416273735e-06,
      "loss": 0.3691,
      "step": 35169
    },
    {
      "epoch": 1.6139690697994586,
      "grad_norm": 0.4565679132938385,
      "learning_rate": 4.552030213882022e-06,
      "loss": 0.3308,
      "step": 35170
    },
    {
      "epoch": 1.614014960304713,
      "grad_norm": 0.501885712146759,
      "learning_rate": 4.551786012567535e-06,
      "loss": 0.3932,
      "step": 35171
    },
    {
      "epoch": 1.6140608508099674,
      "grad_norm": 0.4721248745918274,
      "learning_rate": 4.55154181233086e-06,
      "loss": 0.3326,
      "step": 35172
    },
    {
      "epoch": 1.614106741315222,
      "grad_norm": 0.4689807593822479,
      "learning_rate": 4.551297613172588e-06,
      "loss": 0.3865,
      "step": 35173
    },
    {
      "epoch": 1.6141526318204762,
      "grad_norm": 0.4811904728412628,
      "learning_rate": 4.551053415093302e-06,
      "loss": 0.3748,
      "step": 35174
    },
    {
      "epoch": 1.6141985223257307,
      "grad_norm": 0.4928191900253296,
      "learning_rate": 4.55080921809359e-06,
      "loss": 0.4368,
      "step": 35175
    },
    {
      "epoch": 1.6142444128309852,
      "grad_norm": 0.4447808265686035,
      "learning_rate": 4.550565022174041e-06,
      "loss": 0.3218,
      "step": 35176
    },
    {
      "epoch": 1.6142903033362397,
      "grad_norm": 0.45746299624443054,
      "learning_rate": 4.550320827335243e-06,
      "loss": 0.3444,
      "step": 35177
    },
    {
      "epoch": 1.6143361938414942,
      "grad_norm": 0.46120956540107727,
      "learning_rate": 4.550076633577777e-06,
      "loss": 0.3299,
      "step": 35178
    },
    {
      "epoch": 1.6143820843467487,
      "grad_norm": 0.46369606256484985,
      "learning_rate": 4.549832440902238e-06,
      "loss": 0.3993,
      "step": 35179
    },
    {
      "epoch": 1.6144279748520032,
      "grad_norm": 0.5029719471931458,
      "learning_rate": 4.54958824930921e-06,
      "loss": 0.4189,
      "step": 35180
    },
    {
      "epoch": 1.6144738653572577,
      "grad_norm": 0.42532849311828613,
      "learning_rate": 4.549344058799279e-06,
      "loss": 0.2462,
      "step": 35181
    },
    {
      "epoch": 1.6145197558625122,
      "grad_norm": 0.5017949938774109,
      "learning_rate": 4.549099869373034e-06,
      "loss": 0.4349,
      "step": 35182
    },
    {
      "epoch": 1.6145656463677667,
      "grad_norm": 0.4536321461200714,
      "learning_rate": 4.548855681031062e-06,
      "loss": 0.3182,
      "step": 35183
    },
    {
      "epoch": 1.614611536873021,
      "grad_norm": 0.46532362699508667,
      "learning_rate": 4.548611493773947e-06,
      "loss": 0.3613,
      "step": 35184
    },
    {
      "epoch": 1.6146574273782754,
      "grad_norm": 0.4836522340774536,
      "learning_rate": 4.548367307602281e-06,
      "loss": 0.3756,
      "step": 35185
    },
    {
      "epoch": 1.61470331788353,
      "grad_norm": 0.49262407422065735,
      "learning_rate": 4.54812312251665e-06,
      "loss": 0.3743,
      "step": 35186
    },
    {
      "epoch": 1.6147492083887842,
      "grad_norm": 0.42764127254486084,
      "learning_rate": 4.547878938517638e-06,
      "loss": 0.2776,
      "step": 35187
    },
    {
      "epoch": 1.6147950988940387,
      "grad_norm": 0.4750308394432068,
      "learning_rate": 4.5476347556058355e-06,
      "loss": 0.3896,
      "step": 35188
    },
    {
      "epoch": 1.6148409893992932,
      "grad_norm": 0.5065581798553467,
      "learning_rate": 4.5473905737818295e-06,
      "loss": 0.3656,
      "step": 35189
    },
    {
      "epoch": 1.6148868799045477,
      "grad_norm": 0.4487263858318329,
      "learning_rate": 4.547146393046204e-06,
      "loss": 0.3511,
      "step": 35190
    },
    {
      "epoch": 1.6149327704098022,
      "grad_norm": 0.4229077696800232,
      "learning_rate": 4.5469022133995504e-06,
      "loss": 0.2773,
      "step": 35191
    },
    {
      "epoch": 1.6149786609150567,
      "grad_norm": 0.4307233989238739,
      "learning_rate": 4.546658034842455e-06,
      "loss": 0.2867,
      "step": 35192
    },
    {
      "epoch": 1.6150245514203112,
      "grad_norm": 0.5289615988731384,
      "learning_rate": 4.5464138573755015e-06,
      "loss": 0.4621,
      "step": 35193
    },
    {
      "epoch": 1.6150704419255657,
      "grad_norm": 0.42924270033836365,
      "learning_rate": 4.546169680999281e-06,
      "loss": 0.2887,
      "step": 35194
    },
    {
      "epoch": 1.6151163324308202,
      "grad_norm": 0.48323026299476624,
      "learning_rate": 4.54592550571438e-06,
      "loss": 0.3493,
      "step": 35195
    },
    {
      "epoch": 1.6151622229360745,
      "grad_norm": 0.43896231055259705,
      "learning_rate": 4.545681331521381e-06,
      "loss": 0.2999,
      "step": 35196
    },
    {
      "epoch": 1.615208113441329,
      "grad_norm": 0.4671870768070221,
      "learning_rate": 4.54543715842088e-06,
      "loss": 0.3563,
      "step": 35197
    },
    {
      "epoch": 1.6152540039465835,
      "grad_norm": 0.4926178753376007,
      "learning_rate": 4.545192986413457e-06,
      "loss": 0.3897,
      "step": 35198
    },
    {
      "epoch": 1.615299894451838,
      "grad_norm": 0.47545766830444336,
      "learning_rate": 4.544948815499699e-06,
      "loss": 0.3849,
      "step": 35199
    },
    {
      "epoch": 1.6153457849570922,
      "grad_norm": 0.5472874641418457,
      "learning_rate": 4.544704645680198e-06,
      "loss": 0.3144,
      "step": 35200
    },
    {
      "epoch": 1.6153916754623467,
      "grad_norm": 0.43695104122161865,
      "learning_rate": 4.544460476955539e-06,
      "loss": 0.3507,
      "step": 35201
    },
    {
      "epoch": 1.6154375659676012,
      "grad_norm": 0.4742199778556824,
      "learning_rate": 4.544216309326308e-06,
      "loss": 0.3605,
      "step": 35202
    },
    {
      "epoch": 1.6154834564728557,
      "grad_norm": 0.5102025270462036,
      "learning_rate": 4.543972142793093e-06,
      "loss": 0.4282,
      "step": 35203
    },
    {
      "epoch": 1.6155293469781102,
      "grad_norm": 0.40152162313461304,
      "learning_rate": 4.543727977356481e-06,
      "loss": 0.2609,
      "step": 35204
    },
    {
      "epoch": 1.6155752374833647,
      "grad_norm": 0.4831485152244568,
      "learning_rate": 4.54348381301706e-06,
      "loss": 0.3428,
      "step": 35205
    },
    {
      "epoch": 1.6156211279886192,
      "grad_norm": 0.4464220404624939,
      "learning_rate": 4.543239649775414e-06,
      "loss": 0.338,
      "step": 35206
    },
    {
      "epoch": 1.6156670184938737,
      "grad_norm": 0.4370708167552948,
      "learning_rate": 4.5429954876321345e-06,
      "loss": 0.3012,
      "step": 35207
    },
    {
      "epoch": 1.6157129089991282,
      "grad_norm": 0.4686935842037201,
      "learning_rate": 4.542751326587807e-06,
      "loss": 0.3219,
      "step": 35208
    },
    {
      "epoch": 1.6157587995043825,
      "grad_norm": 0.44972705841064453,
      "learning_rate": 4.542507166643017e-06,
      "loss": 0.3193,
      "step": 35209
    },
    {
      "epoch": 1.615804690009637,
      "grad_norm": 0.4709707200527191,
      "learning_rate": 4.542263007798353e-06,
      "loss": 0.3628,
      "step": 35210
    },
    {
      "epoch": 1.6158505805148915,
      "grad_norm": 0.49618273973464966,
      "learning_rate": 4.5420188500544035e-06,
      "loss": 0.4264,
      "step": 35211
    },
    {
      "epoch": 1.6158964710201458,
      "grad_norm": 0.4657321274280548,
      "learning_rate": 4.541774693411751e-06,
      "loss": 0.3133,
      "step": 35212
    },
    {
      "epoch": 1.6159423615254003,
      "grad_norm": 0.4910042881965637,
      "learning_rate": 4.541530537870988e-06,
      "loss": 0.3734,
      "step": 35213
    },
    {
      "epoch": 1.6159882520306548,
      "grad_norm": 0.4698801040649414,
      "learning_rate": 4.5412863834326995e-06,
      "loss": 0.3667,
      "step": 35214
    },
    {
      "epoch": 1.6160341425359093,
      "grad_norm": 0.47721824049949646,
      "learning_rate": 4.541042230097472e-06,
      "loss": 0.3565,
      "step": 35215
    },
    {
      "epoch": 1.6160800330411638,
      "grad_norm": 0.4822738766670227,
      "learning_rate": 4.5407980778658925e-06,
      "loss": 0.3703,
      "step": 35216
    },
    {
      "epoch": 1.6161259235464183,
      "grad_norm": 0.45121315121650696,
      "learning_rate": 4.54055392673855e-06,
      "loss": 0.3515,
      "step": 35217
    },
    {
      "epoch": 1.6161718140516728,
      "grad_norm": 0.5132384896278381,
      "learning_rate": 4.540309776716028e-06,
      "loss": 0.4246,
      "step": 35218
    },
    {
      "epoch": 1.6162177045569273,
      "grad_norm": 0.6126204133033752,
      "learning_rate": 4.540065627798919e-06,
      "loss": 0.4341,
      "step": 35219
    },
    {
      "epoch": 1.6162635950621818,
      "grad_norm": 0.5258728861808777,
      "learning_rate": 4.539821479987806e-06,
      "loss": 0.4129,
      "step": 35220
    },
    {
      "epoch": 1.6163094855674363,
      "grad_norm": 0.4921393394470215,
      "learning_rate": 4.5395773332832745e-06,
      "loss": 0.4004,
      "step": 35221
    },
    {
      "epoch": 1.6163553760726905,
      "grad_norm": 0.46148261427879333,
      "learning_rate": 4.539333187685917e-06,
      "loss": 0.3604,
      "step": 35222
    },
    {
      "epoch": 1.616401266577945,
      "grad_norm": 0.43393704295158386,
      "learning_rate": 4.539089043196318e-06,
      "loss": 0.3284,
      "step": 35223
    },
    {
      "epoch": 1.6164471570831995,
      "grad_norm": 0.4742489159107208,
      "learning_rate": 4.538844899815063e-06,
      "loss": 0.3742,
      "step": 35224
    },
    {
      "epoch": 1.6164930475884538,
      "grad_norm": 0.43717989325523376,
      "learning_rate": 4.538600757542742e-06,
      "loss": 0.2817,
      "step": 35225
    },
    {
      "epoch": 1.6165389380937083,
      "grad_norm": 0.4664948880672455,
      "learning_rate": 4.538356616379939e-06,
      "loss": 0.3289,
      "step": 35226
    },
    {
      "epoch": 1.6165848285989628,
      "grad_norm": 0.4446718096733093,
      "learning_rate": 4.538112476327242e-06,
      "loss": 0.3602,
      "step": 35227
    },
    {
      "epoch": 1.6166307191042173,
      "grad_norm": 0.48142009973526,
      "learning_rate": 4.53786833738524e-06,
      "loss": 0.3907,
      "step": 35228
    },
    {
      "epoch": 1.6166766096094718,
      "grad_norm": 0.5219006538391113,
      "learning_rate": 4.537624199554519e-06,
      "loss": 0.4319,
      "step": 35229
    },
    {
      "epoch": 1.6167225001147263,
      "grad_norm": 0.46773040294647217,
      "learning_rate": 4.537380062835665e-06,
      "loss": 0.3394,
      "step": 35230
    },
    {
      "epoch": 1.6167683906199808,
      "grad_norm": 0.484846830368042,
      "learning_rate": 4.537135927229267e-06,
      "loss": 0.3543,
      "step": 35231
    },
    {
      "epoch": 1.6168142811252353,
      "grad_norm": 0.4613887071609497,
      "learning_rate": 4.536891792735911e-06,
      "loss": 0.3505,
      "step": 35232
    },
    {
      "epoch": 1.6168601716304898,
      "grad_norm": 0.48618024587631226,
      "learning_rate": 4.536647659356182e-06,
      "loss": 0.3862,
      "step": 35233
    },
    {
      "epoch": 1.616906062135744,
      "grad_norm": 0.42986661195755005,
      "learning_rate": 4.536403527090671e-06,
      "loss": 0.2958,
      "step": 35234
    },
    {
      "epoch": 1.6169519526409986,
      "grad_norm": 0.4768902659416199,
      "learning_rate": 4.5361593959399635e-06,
      "loss": 0.3644,
      "step": 35235
    },
    {
      "epoch": 1.616997843146253,
      "grad_norm": 0.44302046298980713,
      "learning_rate": 4.5359152659046454e-06,
      "loss": 0.3526,
      "step": 35236
    },
    {
      "epoch": 1.6170437336515076,
      "grad_norm": 0.4578110873699188,
      "learning_rate": 4.535671136985305e-06,
      "loss": 0.3414,
      "step": 35237
    },
    {
      "epoch": 1.6170896241567618,
      "grad_norm": 0.44232332706451416,
      "learning_rate": 4.5354270091825295e-06,
      "loss": 0.3033,
      "step": 35238
    },
    {
      "epoch": 1.6171355146620163,
      "grad_norm": 0.4418845772743225,
      "learning_rate": 4.535182882496905e-06,
      "loss": 0.3276,
      "step": 35239
    },
    {
      "epoch": 1.6171814051672708,
      "grad_norm": 0.44349023699760437,
      "learning_rate": 4.534938756929017e-06,
      "loss": 0.3007,
      "step": 35240
    },
    {
      "epoch": 1.6172272956725253,
      "grad_norm": 0.4558194577693939,
      "learning_rate": 4.534694632479458e-06,
      "loss": 0.3445,
      "step": 35241
    },
    {
      "epoch": 1.6172731861777798,
      "grad_norm": 0.4796501696109772,
      "learning_rate": 4.53445050914881e-06,
      "loss": 0.3613,
      "step": 35242
    },
    {
      "epoch": 1.6173190766830343,
      "grad_norm": 0.5377784371376038,
      "learning_rate": 4.53420638693766e-06,
      "loss": 0.4803,
      "step": 35243
    },
    {
      "epoch": 1.6173649671882888,
      "grad_norm": 0.4884701371192932,
      "learning_rate": 4.533962265846598e-06,
      "loss": 0.3865,
      "step": 35244
    },
    {
      "epoch": 1.6174108576935433,
      "grad_norm": 0.45140549540519714,
      "learning_rate": 4.533718145876211e-06,
      "loss": 0.3294,
      "step": 35245
    },
    {
      "epoch": 1.6174567481987978,
      "grad_norm": 0.4348629117012024,
      "learning_rate": 4.533474027027083e-06,
      "loss": 0.3166,
      "step": 35246
    },
    {
      "epoch": 1.617502638704052,
      "grad_norm": 0.46847599744796753,
      "learning_rate": 4.533229909299803e-06,
      "loss": 0.3594,
      "step": 35247
    },
    {
      "epoch": 1.6175485292093066,
      "grad_norm": 1.1716820001602173,
      "learning_rate": 4.532985792694958e-06,
      "loss": 0.3672,
      "step": 35248
    },
    {
      "epoch": 1.617594419714561,
      "grad_norm": 0.45622408390045166,
      "learning_rate": 4.532741677213133e-06,
      "loss": 0.3154,
      "step": 35249
    },
    {
      "epoch": 1.6176403102198154,
      "grad_norm": 0.497847318649292,
      "learning_rate": 4.532497562854919e-06,
      "loss": 0.3698,
      "step": 35250
    },
    {
      "epoch": 1.6176862007250699,
      "grad_norm": 0.46620234847068787,
      "learning_rate": 4.532253449620901e-06,
      "loss": 0.3319,
      "step": 35251
    },
    {
      "epoch": 1.6177320912303244,
      "grad_norm": 0.4813525378704071,
      "learning_rate": 4.532009337511665e-06,
      "loss": 0.3751,
      "step": 35252
    },
    {
      "epoch": 1.6177779817355789,
      "grad_norm": 0.48124200105667114,
      "learning_rate": 4.531765226527799e-06,
      "loss": 0.3379,
      "step": 35253
    },
    {
      "epoch": 1.6178238722408333,
      "grad_norm": 0.501890242099762,
      "learning_rate": 4.53152111666989e-06,
      "loss": 0.3363,
      "step": 35254
    },
    {
      "epoch": 1.6178697627460878,
      "grad_norm": 0.45758265256881714,
      "learning_rate": 4.531277007938523e-06,
      "loss": 0.3432,
      "step": 35255
    },
    {
      "epoch": 1.6179156532513423,
      "grad_norm": 0.4688650071620941,
      "learning_rate": 4.531032900334289e-06,
      "loss": 0.3733,
      "step": 35256
    },
    {
      "epoch": 1.6179615437565968,
      "grad_norm": 0.5282744765281677,
      "learning_rate": 4.530788793857773e-06,
      "loss": 0.4149,
      "step": 35257
    },
    {
      "epoch": 1.6180074342618513,
      "grad_norm": 0.45652472972869873,
      "learning_rate": 4.53054468850956e-06,
      "loss": 0.3235,
      "step": 35258
    },
    {
      "epoch": 1.6180533247671058,
      "grad_norm": 0.4732910692691803,
      "learning_rate": 4.530300584290241e-06,
      "loss": 0.3854,
      "step": 35259
    },
    {
      "epoch": 1.6180992152723601,
      "grad_norm": 0.4966532588005066,
      "learning_rate": 4.5300564812004e-06,
      "loss": 0.3892,
      "step": 35260
    },
    {
      "epoch": 1.6181451057776146,
      "grad_norm": 0.489690363407135,
      "learning_rate": 4.529812379240624e-06,
      "loss": 0.3744,
      "step": 35261
    },
    {
      "epoch": 1.618190996282869,
      "grad_norm": 0.4421101212501526,
      "learning_rate": 4.5295682784115014e-06,
      "loss": 0.3061,
      "step": 35262
    },
    {
      "epoch": 1.6182368867881234,
      "grad_norm": 0.5143552422523499,
      "learning_rate": 4.52932417871362e-06,
      "loss": 0.4552,
      "step": 35263
    },
    {
      "epoch": 1.6182827772933779,
      "grad_norm": 0.4567424952983856,
      "learning_rate": 4.5290800801475634e-06,
      "loss": 0.3279,
      "step": 35264
    },
    {
      "epoch": 1.6183286677986324,
      "grad_norm": 0.45537069439888,
      "learning_rate": 4.528835982713922e-06,
      "loss": 0.3341,
      "step": 35265
    },
    {
      "epoch": 1.6183745583038869,
      "grad_norm": 0.49809014797210693,
      "learning_rate": 4.528591886413281e-06,
      "loss": 0.3896,
      "step": 35266
    },
    {
      "epoch": 1.6184204488091414,
      "grad_norm": 0.4608822464942932,
      "learning_rate": 4.528347791246227e-06,
      "loss": 0.3349,
      "step": 35267
    },
    {
      "epoch": 1.6184663393143959,
      "grad_norm": 0.48798486590385437,
      "learning_rate": 4.528103697213348e-06,
      "loss": 0.4052,
      "step": 35268
    },
    {
      "epoch": 1.6185122298196504,
      "grad_norm": 0.42442312836647034,
      "learning_rate": 4.527859604315232e-06,
      "loss": 0.3289,
      "step": 35269
    },
    {
      "epoch": 1.6185581203249049,
      "grad_norm": 0.4998888373374939,
      "learning_rate": 4.527615512552464e-06,
      "loss": 0.3886,
      "step": 35270
    },
    {
      "epoch": 1.6186040108301594,
      "grad_norm": 0.49102166295051575,
      "learning_rate": 4.52737142192563e-06,
      "loss": 0.3889,
      "step": 35271
    },
    {
      "epoch": 1.6186499013354139,
      "grad_norm": 0.5066962838172913,
      "learning_rate": 4.52712733243532e-06,
      "loss": 0.421,
      "step": 35272
    },
    {
      "epoch": 1.6186957918406681,
      "grad_norm": 0.4994352459907532,
      "learning_rate": 4.5268832440821194e-06,
      "loss": 0.397,
      "step": 35273
    },
    {
      "epoch": 1.6187416823459226,
      "grad_norm": 0.47642555832862854,
      "learning_rate": 4.526639156866614e-06,
      "loss": 0.388,
      "step": 35274
    },
    {
      "epoch": 1.6187875728511771,
      "grad_norm": 0.48843249678611755,
      "learning_rate": 4.526395070789394e-06,
      "loss": 0.3768,
      "step": 35275
    },
    {
      "epoch": 1.6188334633564314,
      "grad_norm": 0.47652867436408997,
      "learning_rate": 4.526150985851044e-06,
      "loss": 0.3909,
      "step": 35276
    },
    {
      "epoch": 1.618879353861686,
      "grad_norm": 0.4179336130619049,
      "learning_rate": 4.525906902052149e-06,
      "loss": 0.2755,
      "step": 35277
    },
    {
      "epoch": 1.6189252443669404,
      "grad_norm": 0.4650818705558777,
      "learning_rate": 4.525662819393301e-06,
      "loss": 0.3983,
      "step": 35278
    },
    {
      "epoch": 1.618971134872195,
      "grad_norm": 0.472674161195755,
      "learning_rate": 4.525418737875084e-06,
      "loss": 0.3765,
      "step": 35279
    },
    {
      "epoch": 1.6190170253774494,
      "grad_norm": 0.46041741967201233,
      "learning_rate": 4.5251746574980835e-06,
      "loss": 0.3875,
      "step": 35280
    },
    {
      "epoch": 1.619062915882704,
      "grad_norm": 0.47982272505760193,
      "learning_rate": 4.52493057826289e-06,
      "loss": 0.3719,
      "step": 35281
    },
    {
      "epoch": 1.6191088063879584,
      "grad_norm": 0.48932552337646484,
      "learning_rate": 4.524686500170087e-06,
      "loss": 0.4149,
      "step": 35282
    },
    {
      "epoch": 1.619154696893213,
      "grad_norm": 0.4422452747821808,
      "learning_rate": 4.524442423220262e-06,
      "loss": 0.3321,
      "step": 35283
    },
    {
      "epoch": 1.6192005873984674,
      "grad_norm": 0.48200148344039917,
      "learning_rate": 4.524198347414005e-06,
      "loss": 0.379,
      "step": 35284
    },
    {
      "epoch": 1.6192464779037217,
      "grad_norm": 0.48795610666275024,
      "learning_rate": 4.523954272751902e-06,
      "loss": 0.3566,
      "step": 35285
    },
    {
      "epoch": 1.6192923684089762,
      "grad_norm": 0.44464004039764404,
      "learning_rate": 4.523710199234535e-06,
      "loss": 0.2775,
      "step": 35286
    },
    {
      "epoch": 1.6193382589142307,
      "grad_norm": 0.47872304916381836,
      "learning_rate": 4.5234661268624965e-06,
      "loss": 0.3424,
      "step": 35287
    },
    {
      "epoch": 1.6193841494194852,
      "grad_norm": 0.4593062698841095,
      "learning_rate": 4.523222055636373e-06,
      "loss": 0.346,
      "step": 35288
    },
    {
      "epoch": 1.6194300399247394,
      "grad_norm": 0.5061553716659546,
      "learning_rate": 4.522977985556748e-06,
      "loss": 0.3816,
      "step": 35289
    },
    {
      "epoch": 1.619475930429994,
      "grad_norm": 0.46625834703445435,
      "learning_rate": 4.522733916624211e-06,
      "loss": 0.3283,
      "step": 35290
    },
    {
      "epoch": 1.6195218209352484,
      "grad_norm": 0.5105793476104736,
      "learning_rate": 4.5224898488393494e-06,
      "loss": 0.4052,
      "step": 35291
    },
    {
      "epoch": 1.619567711440503,
      "grad_norm": 0.5003365874290466,
      "learning_rate": 4.522245782202746e-06,
      "loss": 0.3392,
      "step": 35292
    },
    {
      "epoch": 1.6196136019457574,
      "grad_norm": 0.4674070477485657,
      "learning_rate": 4.522001716714993e-06,
      "loss": 0.3786,
      "step": 35293
    },
    {
      "epoch": 1.619659492451012,
      "grad_norm": 0.469468891620636,
      "learning_rate": 4.521757652376675e-06,
      "loss": 0.3316,
      "step": 35294
    },
    {
      "epoch": 1.6197053829562664,
      "grad_norm": 0.4948015511035919,
      "learning_rate": 4.521513589188378e-06,
      "loss": 0.3848,
      "step": 35295
    },
    {
      "epoch": 1.619751273461521,
      "grad_norm": 0.4885922074317932,
      "learning_rate": 4.521269527150691e-06,
      "loss": 0.3615,
      "step": 35296
    },
    {
      "epoch": 1.6197971639667754,
      "grad_norm": 0.43567728996276855,
      "learning_rate": 4.521025466264199e-06,
      "loss": 0.327,
      "step": 35297
    },
    {
      "epoch": 1.6198430544720297,
      "grad_norm": 0.4396163821220398,
      "learning_rate": 4.5207814065294884e-06,
      "loss": 0.3082,
      "step": 35298
    },
    {
      "epoch": 1.6198889449772842,
      "grad_norm": 0.4649001359939575,
      "learning_rate": 4.520537347947149e-06,
      "loss": 0.3429,
      "step": 35299
    },
    {
      "epoch": 1.6199348354825387,
      "grad_norm": 0.499847412109375,
      "learning_rate": 4.5202932905177656e-06,
      "loss": 0.4404,
      "step": 35300
    },
    {
      "epoch": 1.619980725987793,
      "grad_norm": 0.4450450539588928,
      "learning_rate": 4.520049234241926e-06,
      "loss": 0.3148,
      "step": 35301
    },
    {
      "epoch": 1.6200266164930475,
      "grad_norm": 0.5095664262771606,
      "learning_rate": 4.519805179120216e-06,
      "loss": 0.4186,
      "step": 35302
    },
    {
      "epoch": 1.620072506998302,
      "grad_norm": 0.45161136984825134,
      "learning_rate": 4.5195611251532226e-06,
      "loss": 0.3467,
      "step": 35303
    },
    {
      "epoch": 1.6201183975035565,
      "grad_norm": 0.5368122458457947,
      "learning_rate": 4.519317072341534e-06,
      "loss": 0.4899,
      "step": 35304
    },
    {
      "epoch": 1.620164288008811,
      "grad_norm": 0.4838189482688904,
      "learning_rate": 4.519073020685734e-06,
      "loss": 0.4231,
      "step": 35305
    },
    {
      "epoch": 1.6202101785140655,
      "grad_norm": 0.48127108812332153,
      "learning_rate": 4.518828970186414e-06,
      "loss": 0.3601,
      "step": 35306
    },
    {
      "epoch": 1.62025606901932,
      "grad_norm": 0.4628502130508423,
      "learning_rate": 4.518584920844159e-06,
      "loss": 0.3345,
      "step": 35307
    },
    {
      "epoch": 1.6203019595245745,
      "grad_norm": 0.4806426763534546,
      "learning_rate": 4.518340872659552e-06,
      "loss": 0.3208,
      "step": 35308
    },
    {
      "epoch": 1.620347850029829,
      "grad_norm": 0.46664339303970337,
      "learning_rate": 4.518096825633185e-06,
      "loss": 0.2977,
      "step": 35309
    },
    {
      "epoch": 1.6203937405350834,
      "grad_norm": 0.46739232540130615,
      "learning_rate": 4.5178527797656436e-06,
      "loss": 0.3429,
      "step": 35310
    },
    {
      "epoch": 1.6204396310403377,
      "grad_norm": 0.47474464774131775,
      "learning_rate": 4.5176087350575125e-06,
      "loss": 0.3324,
      "step": 35311
    },
    {
      "epoch": 1.6204855215455922,
      "grad_norm": 0.520048975944519,
      "learning_rate": 4.517364691509382e-06,
      "loss": 0.2796,
      "step": 35312
    },
    {
      "epoch": 1.6205314120508467,
      "grad_norm": 0.4948219358921051,
      "learning_rate": 4.5171206491218365e-06,
      "loss": 0.3703,
      "step": 35313
    },
    {
      "epoch": 1.620577302556101,
      "grad_norm": 0.43446993827819824,
      "learning_rate": 4.5168766078954615e-06,
      "loss": 0.3111,
      "step": 35314
    },
    {
      "epoch": 1.6206231930613555,
      "grad_norm": 0.4698401391506195,
      "learning_rate": 4.516632567830848e-06,
      "loss": 0.4173,
      "step": 35315
    },
    {
      "epoch": 1.62066908356661,
      "grad_norm": 0.4574570953845978,
      "learning_rate": 4.516388528928581e-06,
      "loss": 0.3504,
      "step": 35316
    },
    {
      "epoch": 1.6207149740718645,
      "grad_norm": 0.46265166997909546,
      "learning_rate": 4.516144491189245e-06,
      "loss": 0.3615,
      "step": 35317
    },
    {
      "epoch": 1.620760864577119,
      "grad_norm": 0.43952587246894836,
      "learning_rate": 4.51590045461343e-06,
      "loss": 0.3084,
      "step": 35318
    },
    {
      "epoch": 1.6208067550823735,
      "grad_norm": 0.48864656686782837,
      "learning_rate": 4.515656419201723e-06,
      "loss": 0.3839,
      "step": 35319
    },
    {
      "epoch": 1.620852645587628,
      "grad_norm": 0.45447131991386414,
      "learning_rate": 4.515412384954706e-06,
      "loss": 0.3373,
      "step": 35320
    },
    {
      "epoch": 1.6208985360928825,
      "grad_norm": 0.47784101963043213,
      "learning_rate": 4.515168351872971e-06,
      "loss": 0.4052,
      "step": 35321
    },
    {
      "epoch": 1.620944426598137,
      "grad_norm": 0.45506900548934937,
      "learning_rate": 4.5149243199571044e-06,
      "loss": 0.3264,
      "step": 35322
    },
    {
      "epoch": 1.6209903171033913,
      "grad_norm": 0.5313686728477478,
      "learning_rate": 4.51468028920769e-06,
      "loss": 0.37,
      "step": 35323
    },
    {
      "epoch": 1.6210362076086458,
      "grad_norm": 0.47584855556488037,
      "learning_rate": 4.514436259625317e-06,
      "loss": 0.3625,
      "step": 35324
    },
    {
      "epoch": 1.6210820981139002,
      "grad_norm": 0.4910423159599304,
      "learning_rate": 4.514192231210573e-06,
      "loss": 0.3418,
      "step": 35325
    },
    {
      "epoch": 1.6211279886191547,
      "grad_norm": 0.4454311728477478,
      "learning_rate": 4.51394820396404e-06,
      "loss": 0.3492,
      "step": 35326
    },
    {
      "epoch": 1.621173879124409,
      "grad_norm": 0.4458179473876953,
      "learning_rate": 4.51370417788631e-06,
      "loss": 0.3004,
      "step": 35327
    },
    {
      "epoch": 1.6212197696296635,
      "grad_norm": 0.4452100098133087,
      "learning_rate": 4.5134601529779685e-06,
      "loss": 0.2968,
      "step": 35328
    },
    {
      "epoch": 1.621265660134918,
      "grad_norm": 0.5456843972206116,
      "learning_rate": 4.5132161292396e-06,
      "loss": 0.4768,
      "step": 35329
    },
    {
      "epoch": 1.6213115506401725,
      "grad_norm": 0.47966283559799194,
      "learning_rate": 4.5129721066717955e-06,
      "loss": 0.3927,
      "step": 35330
    },
    {
      "epoch": 1.621357441145427,
      "grad_norm": 0.5243122577667236,
      "learning_rate": 4.5127280852751385e-06,
      "loss": 0.4663,
      "step": 35331
    },
    {
      "epoch": 1.6214033316506815,
      "grad_norm": 0.5283436179161072,
      "learning_rate": 4.512484065050216e-06,
      "loss": 0.4614,
      "step": 35332
    },
    {
      "epoch": 1.621449222155936,
      "grad_norm": 0.5018191337585449,
      "learning_rate": 4.512240045997616e-06,
      "loss": 0.4029,
      "step": 35333
    },
    {
      "epoch": 1.6214951126611905,
      "grad_norm": 0.44825559854507446,
      "learning_rate": 4.511996028117925e-06,
      "loss": 0.3212,
      "step": 35334
    },
    {
      "epoch": 1.621541003166445,
      "grad_norm": 0.4860967695713043,
      "learning_rate": 4.511752011411728e-06,
      "loss": 0.4162,
      "step": 35335
    },
    {
      "epoch": 1.6215868936716993,
      "grad_norm": 0.4744742810726166,
      "learning_rate": 4.511507995879615e-06,
      "loss": 0.4195,
      "step": 35336
    },
    {
      "epoch": 1.6216327841769538,
      "grad_norm": 0.4749017655849457,
      "learning_rate": 4.511263981522171e-06,
      "loss": 0.3712,
      "step": 35337
    },
    {
      "epoch": 1.6216786746822083,
      "grad_norm": 0.45471271872520447,
      "learning_rate": 4.511019968339982e-06,
      "loss": 0.2993,
      "step": 35338
    },
    {
      "epoch": 1.6217245651874626,
      "grad_norm": 0.42778950929641724,
      "learning_rate": 4.510775956333636e-06,
      "loss": 0.3015,
      "step": 35339
    },
    {
      "epoch": 1.621770455692717,
      "grad_norm": 0.46196243166923523,
      "learning_rate": 4.51053194550372e-06,
      "loss": 0.3482,
      "step": 35340
    },
    {
      "epoch": 1.6218163461979715,
      "grad_norm": 0.46847328543663025,
      "learning_rate": 4.51028793585082e-06,
      "loss": 0.3945,
      "step": 35341
    },
    {
      "epoch": 1.621862236703226,
      "grad_norm": 0.4768366515636444,
      "learning_rate": 4.510043927375522e-06,
      "loss": 0.3702,
      "step": 35342
    },
    {
      "epoch": 1.6219081272084805,
      "grad_norm": 0.49599939584732056,
      "learning_rate": 4.509799920078413e-06,
      "loss": 0.3961,
      "step": 35343
    },
    {
      "epoch": 1.621954017713735,
      "grad_norm": 0.4418201446533203,
      "learning_rate": 4.509555913960083e-06,
      "loss": 0.3051,
      "step": 35344
    },
    {
      "epoch": 1.6219999082189895,
      "grad_norm": 0.49317970871925354,
      "learning_rate": 4.509311909021115e-06,
      "loss": 0.4164,
      "step": 35345
    },
    {
      "epoch": 1.622045798724244,
      "grad_norm": 0.48890888690948486,
      "learning_rate": 4.509067905262097e-06,
      "loss": 0.4,
      "step": 35346
    },
    {
      "epoch": 1.6220916892294985,
      "grad_norm": 0.4389429986476898,
      "learning_rate": 4.508823902683617e-06,
      "loss": 0.2775,
      "step": 35347
    },
    {
      "epoch": 1.622137579734753,
      "grad_norm": 0.4804004728794098,
      "learning_rate": 4.5085799012862575e-06,
      "loss": 0.3154,
      "step": 35348
    },
    {
      "epoch": 1.6221834702400073,
      "grad_norm": 0.44557854533195496,
      "learning_rate": 4.50833590107061e-06,
      "loss": 0.3035,
      "step": 35349
    },
    {
      "epoch": 1.6222293607452618,
      "grad_norm": 0.4692220091819763,
      "learning_rate": 4.508091902037261e-06,
      "loss": 0.3504,
      "step": 35350
    },
    {
      "epoch": 1.6222752512505163,
      "grad_norm": 0.4567122757434845,
      "learning_rate": 4.507847904186793e-06,
      "loss": 0.3245,
      "step": 35351
    },
    {
      "epoch": 1.6223211417557706,
      "grad_norm": 0.49859338998794556,
      "learning_rate": 4.507603907519797e-06,
      "loss": 0.4147,
      "step": 35352
    },
    {
      "epoch": 1.622367032261025,
      "grad_norm": 0.5595880746841431,
      "learning_rate": 4.507359912036858e-06,
      "loss": 0.4829,
      "step": 35353
    },
    {
      "epoch": 1.6224129227662796,
      "grad_norm": 0.4808178246021271,
      "learning_rate": 4.507115917738562e-06,
      "loss": 0.4108,
      "step": 35354
    },
    {
      "epoch": 1.622458813271534,
      "grad_norm": 0.478484570980072,
      "learning_rate": 4.5068719246254985e-06,
      "loss": 0.3565,
      "step": 35355
    },
    {
      "epoch": 1.6225047037767886,
      "grad_norm": 0.5046977400779724,
      "learning_rate": 4.506627932698252e-06,
      "loss": 0.4142,
      "step": 35356
    },
    {
      "epoch": 1.622550594282043,
      "grad_norm": 0.46329638361930847,
      "learning_rate": 4.506383941957407e-06,
      "loss": 0.3184,
      "step": 35357
    },
    {
      "epoch": 1.6225964847872976,
      "grad_norm": 0.46192467212677,
      "learning_rate": 4.506139952403556e-06,
      "loss": 0.3659,
      "step": 35358
    },
    {
      "epoch": 1.622642375292552,
      "grad_norm": 0.4539344906806946,
      "learning_rate": 4.505895964037282e-06,
      "loss": 0.3492,
      "step": 35359
    },
    {
      "epoch": 1.6226882657978066,
      "grad_norm": 0.4501603841781616,
      "learning_rate": 4.50565197685917e-06,
      "loss": 0.3356,
      "step": 35360
    },
    {
      "epoch": 1.622734156303061,
      "grad_norm": 0.4797610938549042,
      "learning_rate": 4.505407990869812e-06,
      "loss": 0.3798,
      "step": 35361
    },
    {
      "epoch": 1.6227800468083153,
      "grad_norm": 0.48196348547935486,
      "learning_rate": 4.50516400606979e-06,
      "loss": 0.3728,
      "step": 35362
    },
    {
      "epoch": 1.6228259373135698,
      "grad_norm": 0.5280942320823669,
      "learning_rate": 4.504920022459691e-06,
      "loss": 0.4388,
      "step": 35363
    },
    {
      "epoch": 1.6228718278188243,
      "grad_norm": 0.45943018794059753,
      "learning_rate": 4.504676040040106e-06,
      "loss": 0.3152,
      "step": 35364
    },
    {
      "epoch": 1.6229177183240786,
      "grad_norm": 0.44655418395996094,
      "learning_rate": 4.504432058811618e-06,
      "loss": 0.3405,
      "step": 35365
    },
    {
      "epoch": 1.622963608829333,
      "grad_norm": 0.5128242373466492,
      "learning_rate": 4.504188078774813e-06,
      "loss": 0.4068,
      "step": 35366
    },
    {
      "epoch": 1.6230094993345876,
      "grad_norm": 0.4285009503364563,
      "learning_rate": 4.503944099930281e-06,
      "loss": 0.2535,
      "step": 35367
    },
    {
      "epoch": 1.623055389839842,
      "grad_norm": 0.5098015666007996,
      "learning_rate": 4.503700122278606e-06,
      "loss": 0.3909,
      "step": 35368
    },
    {
      "epoch": 1.6231012803450966,
      "grad_norm": 0.4700107276439667,
      "learning_rate": 4.503456145820377e-06,
      "loss": 0.3745,
      "step": 35369
    },
    {
      "epoch": 1.623147170850351,
      "grad_norm": 0.42465415596961975,
      "learning_rate": 4.503212170556176e-06,
      "loss": 0.3074,
      "step": 35370
    },
    {
      "epoch": 1.6231930613556056,
      "grad_norm": 0.5148603916168213,
      "learning_rate": 4.502968196486596e-06,
      "loss": 0.4515,
      "step": 35371
    },
    {
      "epoch": 1.62323895186086,
      "grad_norm": 0.4716781973838806,
      "learning_rate": 4.502724223612219e-06,
      "loss": 0.3312,
      "step": 35372
    },
    {
      "epoch": 1.6232848423661146,
      "grad_norm": 0.49580928683280945,
      "learning_rate": 4.502480251933634e-06,
      "loss": 0.3642,
      "step": 35373
    },
    {
      "epoch": 1.6233307328713689,
      "grad_norm": 0.4970363974571228,
      "learning_rate": 4.502236281451427e-06,
      "loss": 0.3948,
      "step": 35374
    },
    {
      "epoch": 1.6233766233766234,
      "grad_norm": 0.45346924662590027,
      "learning_rate": 4.501992312166185e-06,
      "loss": 0.2856,
      "step": 35375
    },
    {
      "epoch": 1.6234225138818779,
      "grad_norm": 0.479053795337677,
      "learning_rate": 4.501748344078493e-06,
      "loss": 0.4066,
      "step": 35376
    },
    {
      "epoch": 1.6234684043871324,
      "grad_norm": 0.5144193172454834,
      "learning_rate": 4.50150437718894e-06,
      "loss": 0.4022,
      "step": 35377
    },
    {
      "epoch": 1.6235142948923866,
      "grad_norm": 0.46112629771232605,
      "learning_rate": 4.501260411498112e-06,
      "loss": 0.3209,
      "step": 35378
    },
    {
      "epoch": 1.6235601853976411,
      "grad_norm": 0.45200371742248535,
      "learning_rate": 4.501016447006593e-06,
      "loss": 0.361,
      "step": 35379
    },
    {
      "epoch": 1.6236060759028956,
      "grad_norm": 0.4522021412849426,
      "learning_rate": 4.500772483714974e-06,
      "loss": 0.2926,
      "step": 35380
    },
    {
      "epoch": 1.6236519664081501,
      "grad_norm": 0.49623537063598633,
      "learning_rate": 4.50052852162384e-06,
      "loss": 0.3708,
      "step": 35381
    },
    {
      "epoch": 1.6236978569134046,
      "grad_norm": 0.4409749507904053,
      "learning_rate": 4.500284560733776e-06,
      "loss": 0.3059,
      "step": 35382
    },
    {
      "epoch": 1.6237437474186591,
      "grad_norm": 0.48686155676841736,
      "learning_rate": 4.500040601045371e-06,
      "loss": 0.3627,
      "step": 35383
    },
    {
      "epoch": 1.6237896379239136,
      "grad_norm": 0.5141722559928894,
      "learning_rate": 4.49979664255921e-06,
      "loss": 0.406,
      "step": 35384
    },
    {
      "epoch": 1.6238355284291681,
      "grad_norm": 0.4541027545928955,
      "learning_rate": 4.499552685275879e-06,
      "loss": 0.3222,
      "step": 35385
    },
    {
      "epoch": 1.6238814189344226,
      "grad_norm": 0.4378875195980072,
      "learning_rate": 4.499308729195968e-06,
      "loss": 0.2946,
      "step": 35386
    },
    {
      "epoch": 1.623927309439677,
      "grad_norm": 0.4457743167877197,
      "learning_rate": 4.499064774320061e-06,
      "loss": 0.3542,
      "step": 35387
    },
    {
      "epoch": 1.6239731999449314,
      "grad_norm": 0.4205888509750366,
      "learning_rate": 4.498820820648743e-06,
      "loss": 0.2757,
      "step": 35388
    },
    {
      "epoch": 1.6240190904501859,
      "grad_norm": 0.47514307498931885,
      "learning_rate": 4.498576868182605e-06,
      "loss": 0.3713,
      "step": 35389
    },
    {
      "epoch": 1.6240649809554402,
      "grad_norm": 0.46261709928512573,
      "learning_rate": 4.498332916922232e-06,
      "loss": 0.3235,
      "step": 35390
    },
    {
      "epoch": 1.6241108714606947,
      "grad_norm": 0.4553653597831726,
      "learning_rate": 4.498088966868206e-06,
      "loss": 0.3252,
      "step": 35391
    },
    {
      "epoch": 1.6241567619659492,
      "grad_norm": 0.47684141993522644,
      "learning_rate": 4.497845018021122e-06,
      "loss": 0.3229,
      "step": 35392
    },
    {
      "epoch": 1.6242026524712037,
      "grad_norm": 0.4928971230983734,
      "learning_rate": 4.497601070381561e-06,
      "loss": 0.3746,
      "step": 35393
    },
    {
      "epoch": 1.6242485429764582,
      "grad_norm": 0.4919544756412506,
      "learning_rate": 4.49735712395011e-06,
      "loss": 0.4021,
      "step": 35394
    },
    {
      "epoch": 1.6242944334817127,
      "grad_norm": 0.5035430192947388,
      "learning_rate": 4.497113178727357e-06,
      "loss": 0.4064,
      "step": 35395
    },
    {
      "epoch": 1.6243403239869671,
      "grad_norm": 0.5140207409858704,
      "learning_rate": 4.496869234713889e-06,
      "loss": 0.434,
      "step": 35396
    },
    {
      "epoch": 1.6243862144922216,
      "grad_norm": 0.48337191343307495,
      "learning_rate": 4.496625291910288e-06,
      "loss": 0.3935,
      "step": 35397
    },
    {
      "epoch": 1.6244321049974761,
      "grad_norm": 0.49859920144081116,
      "learning_rate": 4.49638135031715e-06,
      "loss": 0.3835,
      "step": 35398
    },
    {
      "epoch": 1.6244779955027306,
      "grad_norm": 0.48298388719558716,
      "learning_rate": 4.496137409935053e-06,
      "loss": 0.3939,
      "step": 35399
    },
    {
      "epoch": 1.624523886007985,
      "grad_norm": 0.4352428615093231,
      "learning_rate": 4.495893470764584e-06,
      "loss": 0.2523,
      "step": 35400
    },
    {
      "epoch": 1.6245697765132394,
      "grad_norm": 0.47244688868522644,
      "learning_rate": 4.495649532806336e-06,
      "loss": 0.3159,
      "step": 35401
    },
    {
      "epoch": 1.624615667018494,
      "grad_norm": 0.4874855577945709,
      "learning_rate": 4.49540559606089e-06,
      "loss": 0.3457,
      "step": 35402
    },
    {
      "epoch": 1.6246615575237482,
      "grad_norm": 0.4939901530742645,
      "learning_rate": 4.495161660528836e-06,
      "loss": 0.4127,
      "step": 35403
    },
    {
      "epoch": 1.6247074480290027,
      "grad_norm": 0.44046100974082947,
      "learning_rate": 4.494917726210756e-06,
      "loss": 0.283,
      "step": 35404
    },
    {
      "epoch": 1.6247533385342572,
      "grad_norm": 0.5058900713920593,
      "learning_rate": 4.4946737931072425e-06,
      "loss": 0.4453,
      "step": 35405
    },
    {
      "epoch": 1.6247992290395117,
      "grad_norm": 0.5065587162971497,
      "learning_rate": 4.494429861218878e-06,
      "loss": 0.4043,
      "step": 35406
    },
    {
      "epoch": 1.6248451195447662,
      "grad_norm": 0.45777061581611633,
      "learning_rate": 4.494185930546247e-06,
      "loss": 0.3342,
      "step": 35407
    },
    {
      "epoch": 1.6248910100500207,
      "grad_norm": 0.4357583224773407,
      "learning_rate": 4.493942001089943e-06,
      "loss": 0.3225,
      "step": 35408
    },
    {
      "epoch": 1.6249369005552752,
      "grad_norm": 0.47999265789985657,
      "learning_rate": 4.493698072850548e-06,
      "loss": 0.3396,
      "step": 35409
    },
    {
      "epoch": 1.6249827910605297,
      "grad_norm": 0.49701592326164246,
      "learning_rate": 4.4934541458286475e-06,
      "loss": 0.4065,
      "step": 35410
    },
    {
      "epoch": 1.6250286815657842,
      "grad_norm": 0.45117515325546265,
      "learning_rate": 4.493210220024832e-06,
      "loss": 0.3137,
      "step": 35411
    },
    {
      "epoch": 1.6250745720710384,
      "grad_norm": 0.45930275321006775,
      "learning_rate": 4.492966295439686e-06,
      "loss": 0.3068,
      "step": 35412
    },
    {
      "epoch": 1.625120462576293,
      "grad_norm": 0.44015687704086304,
      "learning_rate": 4.492722372073793e-06,
      "loss": 0.3124,
      "step": 35413
    },
    {
      "epoch": 1.6251663530815474,
      "grad_norm": 0.40956366062164307,
      "learning_rate": 4.492478449927746e-06,
      "loss": 0.2789,
      "step": 35414
    },
    {
      "epoch": 1.625212243586802,
      "grad_norm": 0.4509166181087494,
      "learning_rate": 4.492234529002127e-06,
      "loss": 0.3298,
      "step": 35415
    },
    {
      "epoch": 1.6252581340920562,
      "grad_norm": 0.5024508833885193,
      "learning_rate": 4.491990609297523e-06,
      "loss": 0.4077,
      "step": 35416
    },
    {
      "epoch": 1.6253040245973107,
      "grad_norm": 0.48165079951286316,
      "learning_rate": 4.491746690814523e-06,
      "loss": 0.4089,
      "step": 35417
    },
    {
      "epoch": 1.6253499151025652,
      "grad_norm": 0.4518864154815674,
      "learning_rate": 4.491502773553711e-06,
      "loss": 0.3262,
      "step": 35418
    },
    {
      "epoch": 1.6253958056078197,
      "grad_norm": 0.4665011763572693,
      "learning_rate": 4.491258857515673e-06,
      "loss": 0.326,
      "step": 35419
    },
    {
      "epoch": 1.6254416961130742,
      "grad_norm": 0.44789183139801025,
      "learning_rate": 4.4910149427009994e-06,
      "loss": 0.3254,
      "step": 35420
    },
    {
      "epoch": 1.6254875866183287,
      "grad_norm": 0.4629696011543274,
      "learning_rate": 4.490771029110274e-06,
      "loss": 0.3278,
      "step": 35421
    },
    {
      "epoch": 1.6255334771235832,
      "grad_norm": 0.44808197021484375,
      "learning_rate": 4.49052711674408e-06,
      "loss": 0.3458,
      "step": 35422
    },
    {
      "epoch": 1.6255793676288377,
      "grad_norm": 0.4735610783100128,
      "learning_rate": 4.49028320560301e-06,
      "loss": 0.3797,
      "step": 35423
    },
    {
      "epoch": 1.6256252581340922,
      "grad_norm": 0.4541337788105011,
      "learning_rate": 4.490039295687649e-06,
      "loss": 0.3137,
      "step": 35424
    },
    {
      "epoch": 1.6256711486393465,
      "grad_norm": 0.4597019851207733,
      "learning_rate": 4.489795386998581e-06,
      "loss": 0.3299,
      "step": 35425
    },
    {
      "epoch": 1.625717039144601,
      "grad_norm": 0.4339398741722107,
      "learning_rate": 4.489551479536395e-06,
      "loss": 0.2896,
      "step": 35426
    },
    {
      "epoch": 1.6257629296498555,
      "grad_norm": 0.46574637293815613,
      "learning_rate": 4.4893075733016765e-06,
      "loss": 0.3291,
      "step": 35427
    },
    {
      "epoch": 1.6258088201551097,
      "grad_norm": 0.4440774619579315,
      "learning_rate": 4.48906366829501e-06,
      "loss": 0.3258,
      "step": 35428
    },
    {
      "epoch": 1.6258547106603642,
      "grad_norm": 0.45186764001846313,
      "learning_rate": 4.488819764516986e-06,
      "loss": 0.3171,
      "step": 35429
    },
    {
      "epoch": 1.6259006011656187,
      "grad_norm": 0.46767154335975647,
      "learning_rate": 4.488575861968189e-06,
      "loss": 0.3365,
      "step": 35430
    },
    {
      "epoch": 1.6259464916708732,
      "grad_norm": 0.4650920629501343,
      "learning_rate": 4.488331960649206e-06,
      "loss": 0.3768,
      "step": 35431
    },
    {
      "epoch": 1.6259923821761277,
      "grad_norm": 0.4853613078594208,
      "learning_rate": 4.488088060560623e-06,
      "loss": 0.3833,
      "step": 35432
    },
    {
      "epoch": 1.6260382726813822,
      "grad_norm": 0.4423483610153198,
      "learning_rate": 4.487844161703027e-06,
      "loss": 0.3139,
      "step": 35433
    },
    {
      "epoch": 1.6260841631866367,
      "grad_norm": 0.48707109689712524,
      "learning_rate": 4.487600264077004e-06,
      "loss": 0.3913,
      "step": 35434
    },
    {
      "epoch": 1.6261300536918912,
      "grad_norm": 0.44890594482421875,
      "learning_rate": 4.487356367683139e-06,
      "loss": 0.2955,
      "step": 35435
    },
    {
      "epoch": 1.6261759441971457,
      "grad_norm": 0.46130481362342834,
      "learning_rate": 4.4871124725220215e-06,
      "loss": 0.3662,
      "step": 35436
    },
    {
      "epoch": 1.6262218347024002,
      "grad_norm": 0.449642151594162,
      "learning_rate": 4.486868578594237e-06,
      "loss": 0.3097,
      "step": 35437
    },
    {
      "epoch": 1.6262677252076545,
      "grad_norm": 0.437250554561615,
      "learning_rate": 4.486624685900371e-06,
      "loss": 0.2977,
      "step": 35438
    },
    {
      "epoch": 1.626313615712909,
      "grad_norm": 0.4712527096271515,
      "learning_rate": 4.486380794441011e-06,
      "loss": 0.3208,
      "step": 35439
    },
    {
      "epoch": 1.6263595062181635,
      "grad_norm": 0.48606249690055847,
      "learning_rate": 4.486136904216743e-06,
      "loss": 0.361,
      "step": 35440
    },
    {
      "epoch": 1.6264053967234178,
      "grad_norm": 0.4611618220806122,
      "learning_rate": 4.485893015228151e-06,
      "loss": 0.3163,
      "step": 35441
    },
    {
      "epoch": 1.6264512872286723,
      "grad_norm": 0.46257397532463074,
      "learning_rate": 4.485649127475828e-06,
      "loss": 0.3119,
      "step": 35442
    },
    {
      "epoch": 1.6264971777339268,
      "grad_norm": 0.5052624940872192,
      "learning_rate": 4.485405240960356e-06,
      "loss": 0.4335,
      "step": 35443
    },
    {
      "epoch": 1.6265430682391813,
      "grad_norm": 0.4491192102432251,
      "learning_rate": 4.485161355682319e-06,
      "loss": 0.3355,
      "step": 35444
    },
    {
      "epoch": 1.6265889587444358,
      "grad_norm": 0.6520278453826904,
      "learning_rate": 4.48491747164231e-06,
      "loss": 0.444,
      "step": 35445
    },
    {
      "epoch": 1.6266348492496903,
      "grad_norm": 0.4951401650905609,
      "learning_rate": 4.48467358884091e-06,
      "loss": 0.3589,
      "step": 35446
    },
    {
      "epoch": 1.6266807397549448,
      "grad_norm": 0.45782414078712463,
      "learning_rate": 4.484429707278707e-06,
      "loss": 0.3391,
      "step": 35447
    },
    {
      "epoch": 1.6267266302601993,
      "grad_norm": 0.46797433495521545,
      "learning_rate": 4.484185826956289e-06,
      "loss": 0.3183,
      "step": 35448
    },
    {
      "epoch": 1.6267725207654538,
      "grad_norm": 0.503954291343689,
      "learning_rate": 4.4839419478742415e-06,
      "loss": 0.3823,
      "step": 35449
    },
    {
      "epoch": 1.6268184112707083,
      "grad_norm": 0.46666282415390015,
      "learning_rate": 4.483698070033148e-06,
      "loss": 0.329,
      "step": 35450
    },
    {
      "epoch": 1.6268643017759625,
      "grad_norm": 0.4709049165248871,
      "learning_rate": 4.4834541934336e-06,
      "loss": 0.3067,
      "step": 35451
    },
    {
      "epoch": 1.626910192281217,
      "grad_norm": 0.4874848425388336,
      "learning_rate": 4.483210318076182e-06,
      "loss": 0.4211,
      "step": 35452
    },
    {
      "epoch": 1.6269560827864715,
      "grad_norm": 0.4769006073474884,
      "learning_rate": 4.4829664439614786e-06,
      "loss": 0.3616,
      "step": 35453
    },
    {
      "epoch": 1.6270019732917258,
      "grad_norm": 0.47058728337287903,
      "learning_rate": 4.4827225710900795e-06,
      "loss": 0.3959,
      "step": 35454
    },
    {
      "epoch": 1.6270478637969803,
      "grad_norm": 0.44224813580513,
      "learning_rate": 4.482478699462569e-06,
      "loss": 0.3241,
      "step": 35455
    },
    {
      "epoch": 1.6270937543022348,
      "grad_norm": 0.4598153829574585,
      "learning_rate": 4.482234829079531e-06,
      "loss": 0.3504,
      "step": 35456
    },
    {
      "epoch": 1.6271396448074893,
      "grad_norm": 0.5308739542961121,
      "learning_rate": 4.481990959941558e-06,
      "loss": 0.4136,
      "step": 35457
    },
    {
      "epoch": 1.6271855353127438,
      "grad_norm": 0.4537332355976105,
      "learning_rate": 4.481747092049233e-06,
      "loss": 0.2794,
      "step": 35458
    },
    {
      "epoch": 1.6272314258179983,
      "grad_norm": 0.649353563785553,
      "learning_rate": 4.4815032254031415e-06,
      "loss": 0.4316,
      "step": 35459
    },
    {
      "epoch": 1.6272773163232528,
      "grad_norm": 0.508204460144043,
      "learning_rate": 4.481259360003872e-06,
      "loss": 0.3581,
      "step": 35460
    },
    {
      "epoch": 1.6273232068285073,
      "grad_norm": 0.4447965621948242,
      "learning_rate": 4.48101549585201e-06,
      "loss": 0.3163,
      "step": 35461
    },
    {
      "epoch": 1.6273690973337618,
      "grad_norm": 0.4591626524925232,
      "learning_rate": 4.480771632948141e-06,
      "loss": 0.3193,
      "step": 35462
    },
    {
      "epoch": 1.627414987839016,
      "grad_norm": 0.49186432361602783,
      "learning_rate": 4.480527771292853e-06,
      "loss": 0.3673,
      "step": 35463
    },
    {
      "epoch": 1.6274608783442706,
      "grad_norm": 0.4688226282596588,
      "learning_rate": 4.480283910886733e-06,
      "loss": 0.3399,
      "step": 35464
    },
    {
      "epoch": 1.627506768849525,
      "grad_norm": 0.4897063970565796,
      "learning_rate": 4.4800400517303635e-06,
      "loss": 0.3743,
      "step": 35465
    },
    {
      "epoch": 1.6275526593547793,
      "grad_norm": 0.4215324819087982,
      "learning_rate": 4.479796193824335e-06,
      "loss": 0.259,
      "step": 35466
    },
    {
      "epoch": 1.6275985498600338,
      "grad_norm": 0.48412153124809265,
      "learning_rate": 4.479552337169233e-06,
      "loss": 0.3711,
      "step": 35467
    },
    {
      "epoch": 1.6276444403652883,
      "grad_norm": 0.5110636949539185,
      "learning_rate": 4.4793084817656435e-06,
      "loss": 0.4398,
      "step": 35468
    },
    {
      "epoch": 1.6276903308705428,
      "grad_norm": 0.4544069766998291,
      "learning_rate": 4.479064627614151e-06,
      "loss": 0.3661,
      "step": 35469
    },
    {
      "epoch": 1.6277362213757973,
      "grad_norm": 0.49742844700813293,
      "learning_rate": 4.478820774715346e-06,
      "loss": 0.4153,
      "step": 35470
    },
    {
      "epoch": 1.6277821118810518,
      "grad_norm": 0.47196000814437866,
      "learning_rate": 4.478576923069812e-06,
      "loss": 0.3621,
      "step": 35471
    },
    {
      "epoch": 1.6278280023863063,
      "grad_norm": 0.45216888189315796,
      "learning_rate": 4.4783330726781325e-06,
      "loss": 0.2966,
      "step": 35472
    },
    {
      "epoch": 1.6278738928915608,
      "grad_norm": 0.48925426602363586,
      "learning_rate": 4.478089223540902e-06,
      "loss": 0.4005,
      "step": 35473
    },
    {
      "epoch": 1.6279197833968153,
      "grad_norm": 0.43423992395401,
      "learning_rate": 4.4778453756587e-06,
      "loss": 0.3161,
      "step": 35474
    },
    {
      "epoch": 1.6279656739020698,
      "grad_norm": 0.4603155553340912,
      "learning_rate": 4.4776015290321145e-06,
      "loss": 0.3181,
      "step": 35475
    },
    {
      "epoch": 1.628011564407324,
      "grad_norm": 0.4592399597167969,
      "learning_rate": 4.477357683661734e-06,
      "loss": 0.3862,
      "step": 35476
    },
    {
      "epoch": 1.6280574549125786,
      "grad_norm": 0.49150657653808594,
      "learning_rate": 4.477113839548143e-06,
      "loss": 0.3746,
      "step": 35477
    },
    {
      "epoch": 1.628103345417833,
      "grad_norm": 0.4394973814487457,
      "learning_rate": 4.476869996691926e-06,
      "loss": 0.3272,
      "step": 35478
    },
    {
      "epoch": 1.6281492359230874,
      "grad_norm": 0.48618602752685547,
      "learning_rate": 4.476626155093673e-06,
      "loss": 0.3625,
      "step": 35479
    },
    {
      "epoch": 1.6281951264283419,
      "grad_norm": 0.46169352531433105,
      "learning_rate": 4.47638231475397e-06,
      "loss": 0.3445,
      "step": 35480
    },
    {
      "epoch": 1.6282410169335964,
      "grad_norm": 0.4708074927330017,
      "learning_rate": 4.4761384756734015e-06,
      "loss": 0.4214,
      "step": 35481
    },
    {
      "epoch": 1.6282869074388509,
      "grad_norm": 0.4659002423286438,
      "learning_rate": 4.4758946378525545e-06,
      "loss": 0.3545,
      "step": 35482
    },
    {
      "epoch": 1.6283327979441053,
      "grad_norm": 0.4801311790943146,
      "learning_rate": 4.475650801292016e-06,
      "loss": 0.3643,
      "step": 35483
    },
    {
      "epoch": 1.6283786884493598,
      "grad_norm": 0.458286315202713,
      "learning_rate": 4.47540696599237e-06,
      "loss": 0.3169,
      "step": 35484
    },
    {
      "epoch": 1.6284245789546143,
      "grad_norm": 0.49893537163734436,
      "learning_rate": 4.475163131954207e-06,
      "loss": 0.4242,
      "step": 35485
    },
    {
      "epoch": 1.6284704694598688,
      "grad_norm": 0.4759639799594879,
      "learning_rate": 4.4749192991781115e-06,
      "loss": 0.3748,
      "step": 35486
    },
    {
      "epoch": 1.6285163599651233,
      "grad_norm": 0.46488863229751587,
      "learning_rate": 4.4746754676646665e-06,
      "loss": 0.3196,
      "step": 35487
    },
    {
      "epoch": 1.6285622504703778,
      "grad_norm": 0.467911034822464,
      "learning_rate": 4.474431637414463e-06,
      "loss": 0.3213,
      "step": 35488
    },
    {
      "epoch": 1.6286081409756321,
      "grad_norm": 0.4602161645889282,
      "learning_rate": 4.474187808428086e-06,
      "loss": 0.338,
      "step": 35489
    },
    {
      "epoch": 1.6286540314808866,
      "grad_norm": 0.47880834341049194,
      "learning_rate": 4.47394398070612e-06,
      "loss": 0.3308,
      "step": 35490
    },
    {
      "epoch": 1.628699921986141,
      "grad_norm": 0.5082428455352783,
      "learning_rate": 4.4737001542491534e-06,
      "loss": 0.3946,
      "step": 35491
    },
    {
      "epoch": 1.6287458124913954,
      "grad_norm": 0.4887165427207947,
      "learning_rate": 4.473456329057773e-06,
      "loss": 0.3858,
      "step": 35492
    },
    {
      "epoch": 1.6287917029966499,
      "grad_norm": 0.4803290367126465,
      "learning_rate": 4.473212505132561e-06,
      "loss": 0.362,
      "step": 35493
    },
    {
      "epoch": 1.6288375935019044,
      "grad_norm": 0.4641997516155243,
      "learning_rate": 4.472968682474109e-06,
      "loss": 0.3409,
      "step": 35494
    },
    {
      "epoch": 1.6288834840071589,
      "grad_norm": 0.45477670431137085,
      "learning_rate": 4.472724861083001e-06,
      "loss": 0.3248,
      "step": 35495
    },
    {
      "epoch": 1.6289293745124134,
      "grad_norm": 0.5027965903282166,
      "learning_rate": 4.472481040959822e-06,
      "loss": 0.3931,
      "step": 35496
    },
    {
      "epoch": 1.6289752650176679,
      "grad_norm": 0.49711543321609497,
      "learning_rate": 4.47223722210516e-06,
      "loss": 0.4319,
      "step": 35497
    },
    {
      "epoch": 1.6290211555229224,
      "grad_norm": 0.46250882744789124,
      "learning_rate": 4.471993404519602e-06,
      "loss": 0.3301,
      "step": 35498
    },
    {
      "epoch": 1.6290670460281769,
      "grad_norm": 0.48355385661125183,
      "learning_rate": 4.471749588203733e-06,
      "loss": 0.3912,
      "step": 35499
    },
    {
      "epoch": 1.6291129365334314,
      "grad_norm": 0.4857760965824127,
      "learning_rate": 4.471505773158137e-06,
      "loss": 0.395,
      "step": 35500
    },
    {
      "epoch": 1.6291588270386856,
      "grad_norm": 0.5373702049255371,
      "learning_rate": 4.471261959383406e-06,
      "loss": 0.41,
      "step": 35501
    },
    {
      "epoch": 1.6292047175439401,
      "grad_norm": 0.4733721911907196,
      "learning_rate": 4.471018146880122e-06,
      "loss": 0.3898,
      "step": 35502
    },
    {
      "epoch": 1.6292506080491946,
      "grad_norm": 0.4943593144416809,
      "learning_rate": 4.4707743356488715e-06,
      "loss": 0.4354,
      "step": 35503
    },
    {
      "epoch": 1.6292964985544491,
      "grad_norm": 0.48634272813796997,
      "learning_rate": 4.470530525690242e-06,
      "loss": 0.4063,
      "step": 35504
    },
    {
      "epoch": 1.6293423890597034,
      "grad_norm": 0.5147380828857422,
      "learning_rate": 4.4702867170048215e-06,
      "loss": 0.378,
      "step": 35505
    },
    {
      "epoch": 1.629388279564958,
      "grad_norm": 0.487272173166275,
      "learning_rate": 4.470042909593191e-06,
      "loss": 0.3879,
      "step": 35506
    },
    {
      "epoch": 1.6294341700702124,
      "grad_norm": 0.4719718098640442,
      "learning_rate": 4.4697991034559415e-06,
      "loss": 0.4019,
      "step": 35507
    },
    {
      "epoch": 1.629480060575467,
      "grad_norm": 0.4661698043346405,
      "learning_rate": 4.469555298593661e-06,
      "loss": 0.3731,
      "step": 35508
    },
    {
      "epoch": 1.6295259510807214,
      "grad_norm": 0.4525972902774811,
      "learning_rate": 4.4693114950069275e-06,
      "loss": 0.2952,
      "step": 35509
    },
    {
      "epoch": 1.629571841585976,
      "grad_norm": 0.4766845405101776,
      "learning_rate": 4.469067692696334e-06,
      "loss": 0.3507,
      "step": 35510
    },
    {
      "epoch": 1.6296177320912304,
      "grad_norm": 0.5005804896354675,
      "learning_rate": 4.468823891662467e-06,
      "loss": 0.3847,
      "step": 35511
    },
    {
      "epoch": 1.629663622596485,
      "grad_norm": 0.49273112416267395,
      "learning_rate": 4.468580091905908e-06,
      "loss": 0.3962,
      "step": 35512
    },
    {
      "epoch": 1.6297095131017394,
      "grad_norm": 0.4786682724952698,
      "learning_rate": 4.468336293427249e-06,
      "loss": 0.3729,
      "step": 35513
    },
    {
      "epoch": 1.6297554036069937,
      "grad_norm": 0.4951097369194031,
      "learning_rate": 4.468092496227072e-06,
      "loss": 0.3763,
      "step": 35514
    },
    {
      "epoch": 1.6298012941122482,
      "grad_norm": 0.48214292526245117,
      "learning_rate": 4.467848700305963e-06,
      "loss": 0.3517,
      "step": 35515
    },
    {
      "epoch": 1.6298471846175027,
      "grad_norm": 0.4995957612991333,
      "learning_rate": 4.467604905664512e-06,
      "loss": 0.3974,
      "step": 35516
    },
    {
      "epoch": 1.629893075122757,
      "grad_norm": 0.4466254711151123,
      "learning_rate": 4.467361112303304e-06,
      "loss": 0.3325,
      "step": 35517
    },
    {
      "epoch": 1.6299389656280114,
      "grad_norm": 0.45246148109436035,
      "learning_rate": 4.467117320222923e-06,
      "loss": 0.3376,
      "step": 35518
    },
    {
      "epoch": 1.629984856133266,
      "grad_norm": 0.4883652329444885,
      "learning_rate": 4.466873529423957e-06,
      "loss": 0.4325,
      "step": 35519
    },
    {
      "epoch": 1.6300307466385204,
      "grad_norm": 0.45610594749450684,
      "learning_rate": 4.466629739906993e-06,
      "loss": 0.3206,
      "step": 35520
    },
    {
      "epoch": 1.630076637143775,
      "grad_norm": 0.5404648184776306,
      "learning_rate": 4.4663859516726124e-06,
      "loss": 0.394,
      "step": 35521
    },
    {
      "epoch": 1.6301225276490294,
      "grad_norm": 0.46589526534080505,
      "learning_rate": 4.466142164721409e-06,
      "loss": 0.3738,
      "step": 35522
    },
    {
      "epoch": 1.630168418154284,
      "grad_norm": 0.47595810890197754,
      "learning_rate": 4.4658983790539645e-06,
      "loss": 0.3888,
      "step": 35523
    },
    {
      "epoch": 1.6302143086595384,
      "grad_norm": 0.4296239912509918,
      "learning_rate": 4.465654594670864e-06,
      "loss": 0.2968,
      "step": 35524
    },
    {
      "epoch": 1.630260199164793,
      "grad_norm": 0.45196154713630676,
      "learning_rate": 4.465410811572698e-06,
      "loss": 0.3306,
      "step": 35525
    },
    {
      "epoch": 1.6303060896700474,
      "grad_norm": 0.4577309191226959,
      "learning_rate": 4.46516702976005e-06,
      "loss": 0.3368,
      "step": 35526
    },
    {
      "epoch": 1.6303519801753017,
      "grad_norm": 0.47257304191589355,
      "learning_rate": 4.4649232492335034e-06,
      "loss": 0.3618,
      "step": 35527
    },
    {
      "epoch": 1.6303978706805562,
      "grad_norm": 0.4611373543739319,
      "learning_rate": 4.46467946999365e-06,
      "loss": 0.3647,
      "step": 35528
    },
    {
      "epoch": 1.6304437611858107,
      "grad_norm": 0.4811505973339081,
      "learning_rate": 4.464435692041075e-06,
      "loss": 0.3351,
      "step": 35529
    },
    {
      "epoch": 1.630489651691065,
      "grad_norm": 0.4493474066257477,
      "learning_rate": 4.4641919153763596e-06,
      "loss": 0.3341,
      "step": 35530
    },
    {
      "epoch": 1.6305355421963195,
      "grad_norm": 0.5177345275878906,
      "learning_rate": 4.463948140000096e-06,
      "loss": 0.4794,
      "step": 35531
    },
    {
      "epoch": 1.630581432701574,
      "grad_norm": 0.4485916197299957,
      "learning_rate": 4.4637043659128675e-06,
      "loss": 0.2953,
      "step": 35532
    },
    {
      "epoch": 1.6306273232068285,
      "grad_norm": 0.4676695764064789,
      "learning_rate": 4.463460593115261e-06,
      "loss": 0.3367,
      "step": 35533
    },
    {
      "epoch": 1.630673213712083,
      "grad_norm": 0.49476614594459534,
      "learning_rate": 4.463216821607862e-06,
      "loss": 0.3665,
      "step": 35534
    },
    {
      "epoch": 1.6307191042173375,
      "grad_norm": 0.46426495909690857,
      "learning_rate": 4.462973051391257e-06,
      "loss": 0.3586,
      "step": 35535
    },
    {
      "epoch": 1.630764994722592,
      "grad_norm": 0.46032726764678955,
      "learning_rate": 4.462729282466033e-06,
      "loss": 0.285,
      "step": 35536
    },
    {
      "epoch": 1.6308108852278465,
      "grad_norm": 0.45155343413352966,
      "learning_rate": 4.462485514832773e-06,
      "loss": 0.3387,
      "step": 35537
    },
    {
      "epoch": 1.630856775733101,
      "grad_norm": 0.4509875178337097,
      "learning_rate": 4.462241748492069e-06,
      "loss": 0.3292,
      "step": 35538
    },
    {
      "epoch": 1.6309026662383554,
      "grad_norm": 0.44780299067497253,
      "learning_rate": 4.461997983444502e-06,
      "loss": 0.326,
      "step": 35539
    },
    {
      "epoch": 1.6309485567436097,
      "grad_norm": 0.44531455636024475,
      "learning_rate": 4.461754219690661e-06,
      "loss": 0.3416,
      "step": 35540
    },
    {
      "epoch": 1.6309944472488642,
      "grad_norm": 0.49277830123901367,
      "learning_rate": 4.46151045723113e-06,
      "loss": 0.4095,
      "step": 35541
    },
    {
      "epoch": 1.6310403377541187,
      "grad_norm": 0.48743683099746704,
      "learning_rate": 4.461266696066498e-06,
      "loss": 0.4087,
      "step": 35542
    },
    {
      "epoch": 1.631086228259373,
      "grad_norm": 0.4748031497001648,
      "learning_rate": 4.461022936197347e-06,
      "loss": 0.3238,
      "step": 35543
    },
    {
      "epoch": 1.6311321187646275,
      "grad_norm": 0.48686230182647705,
      "learning_rate": 4.460779177624268e-06,
      "loss": 0.3568,
      "step": 35544
    },
    {
      "epoch": 1.631178009269882,
      "grad_norm": 0.482170432806015,
      "learning_rate": 4.460535420347845e-06,
      "loss": 0.3542,
      "step": 35545
    },
    {
      "epoch": 1.6312238997751365,
      "grad_norm": 0.5064695477485657,
      "learning_rate": 4.460291664368663e-06,
      "loss": 0.4667,
      "step": 35546
    },
    {
      "epoch": 1.631269790280391,
      "grad_norm": 0.4757748246192932,
      "learning_rate": 4.46004790968731e-06,
      "loss": 0.3633,
      "step": 35547
    },
    {
      "epoch": 1.6313156807856455,
      "grad_norm": 0.4961652457714081,
      "learning_rate": 4.4598041563043714e-06,
      "loss": 0.4134,
      "step": 35548
    },
    {
      "epoch": 1.6313615712909,
      "grad_norm": 0.47700512409210205,
      "learning_rate": 4.459560404220431e-06,
      "loss": 0.3392,
      "step": 35549
    },
    {
      "epoch": 1.6314074617961545,
      "grad_norm": 0.44466179609298706,
      "learning_rate": 4.45931665343608e-06,
      "loss": 0.2977,
      "step": 35550
    },
    {
      "epoch": 1.631453352301409,
      "grad_norm": 0.5151621699333191,
      "learning_rate": 4.459072903951901e-06,
      "loss": 0.3584,
      "step": 35551
    },
    {
      "epoch": 1.6314992428066633,
      "grad_norm": 0.4676557779312134,
      "learning_rate": 4.458829155768481e-06,
      "loss": 0.333,
      "step": 35552
    },
    {
      "epoch": 1.6315451333119178,
      "grad_norm": 0.4812327027320862,
      "learning_rate": 4.458585408886406e-06,
      "loss": 0.4154,
      "step": 35553
    },
    {
      "epoch": 1.6315910238171722,
      "grad_norm": 0.508109450340271,
      "learning_rate": 4.458341663306263e-06,
      "loss": 0.4355,
      "step": 35554
    },
    {
      "epoch": 1.6316369143224265,
      "grad_norm": 0.4853738844394684,
      "learning_rate": 4.458097919028636e-06,
      "loss": 0.369,
      "step": 35555
    },
    {
      "epoch": 1.631682804827681,
      "grad_norm": 0.46401435136795044,
      "learning_rate": 4.4578541760541135e-06,
      "loss": 0.3513,
      "step": 35556
    },
    {
      "epoch": 1.6317286953329355,
      "grad_norm": 0.4852491617202759,
      "learning_rate": 4.4576104343832806e-06,
      "loss": 0.4132,
      "step": 35557
    },
    {
      "epoch": 1.63177458583819,
      "grad_norm": 0.5148482918739319,
      "learning_rate": 4.457366694016722e-06,
      "loss": 0.4633,
      "step": 35558
    },
    {
      "epoch": 1.6318204763434445,
      "grad_norm": 0.4928186535835266,
      "learning_rate": 4.457122954955026e-06,
      "loss": 0.3693,
      "step": 35559
    },
    {
      "epoch": 1.631866366848699,
      "grad_norm": 0.49567824602127075,
      "learning_rate": 4.456879217198778e-06,
      "loss": 0.3873,
      "step": 35560
    },
    {
      "epoch": 1.6319122573539535,
      "grad_norm": 0.5124966502189636,
      "learning_rate": 4.456635480748564e-06,
      "loss": 0.3893,
      "step": 35561
    },
    {
      "epoch": 1.631958147859208,
      "grad_norm": 0.5041531324386597,
      "learning_rate": 4.456391745604971e-06,
      "loss": 0.3597,
      "step": 35562
    },
    {
      "epoch": 1.6320040383644625,
      "grad_norm": 0.5168278217315674,
      "learning_rate": 4.456148011768584e-06,
      "loss": 0.4082,
      "step": 35563
    },
    {
      "epoch": 1.632049928869717,
      "grad_norm": 0.44661450386047363,
      "learning_rate": 4.455904279239987e-06,
      "loss": 0.3471,
      "step": 35564
    },
    {
      "epoch": 1.6320958193749713,
      "grad_norm": 0.48150816559791565,
      "learning_rate": 4.4556605480197714e-06,
      "loss": 0.3888,
      "step": 35565
    },
    {
      "epoch": 1.6321417098802258,
      "grad_norm": 0.4782267212867737,
      "learning_rate": 4.45541681810852e-06,
      "loss": 0.391,
      "step": 35566
    },
    {
      "epoch": 1.6321876003854803,
      "grad_norm": 0.48155367374420166,
      "learning_rate": 4.45517308950682e-06,
      "loss": 0.3625,
      "step": 35567
    },
    {
      "epoch": 1.6322334908907346,
      "grad_norm": 0.4995434284210205,
      "learning_rate": 4.454929362215255e-06,
      "loss": 0.385,
      "step": 35568
    },
    {
      "epoch": 1.632279381395989,
      "grad_norm": 0.4260987937450409,
      "learning_rate": 4.454685636234414e-06,
      "loss": 0.2639,
      "step": 35569
    },
    {
      "epoch": 1.6323252719012435,
      "grad_norm": 0.4724951982498169,
      "learning_rate": 4.454441911564881e-06,
      "loss": 0.3524,
      "step": 35570
    },
    {
      "epoch": 1.632371162406498,
      "grad_norm": 0.5226155519485474,
      "learning_rate": 4.454198188207242e-06,
      "loss": 0.4696,
      "step": 35571
    },
    {
      "epoch": 1.6324170529117525,
      "grad_norm": 0.4589334726333618,
      "learning_rate": 4.4539544661620855e-06,
      "loss": 0.3065,
      "step": 35572
    },
    {
      "epoch": 1.632462943417007,
      "grad_norm": 0.46255263686180115,
      "learning_rate": 4.453710745429997e-06,
      "loss": 0.3716,
      "step": 35573
    },
    {
      "epoch": 1.6325088339222615,
      "grad_norm": 0.4969880282878876,
      "learning_rate": 4.4534670260115606e-06,
      "loss": 0.4059,
      "step": 35574
    },
    {
      "epoch": 1.632554724427516,
      "grad_norm": 0.439707487821579,
      "learning_rate": 4.4532233079073645e-06,
      "loss": 0.3086,
      "step": 35575
    },
    {
      "epoch": 1.6326006149327705,
      "grad_norm": 0.4809582829475403,
      "learning_rate": 4.452979591117993e-06,
      "loss": 0.3789,
      "step": 35576
    },
    {
      "epoch": 1.632646505438025,
      "grad_norm": 0.44056758284568787,
      "learning_rate": 4.4527358756440326e-06,
      "loss": 0.3115,
      "step": 35577
    },
    {
      "epoch": 1.6326923959432793,
      "grad_norm": 0.5248442888259888,
      "learning_rate": 4.452492161486071e-06,
      "loss": 0.3129,
      "step": 35578
    },
    {
      "epoch": 1.6327382864485338,
      "grad_norm": 0.4645281136035919,
      "learning_rate": 4.452248448644693e-06,
      "loss": 0.3136,
      "step": 35579
    },
    {
      "epoch": 1.6327841769537883,
      "grad_norm": 0.46180009841918945,
      "learning_rate": 4.452004737120481e-06,
      "loss": 0.3209,
      "step": 35580
    },
    {
      "epoch": 1.6328300674590426,
      "grad_norm": 0.47050777077674866,
      "learning_rate": 4.451761026914028e-06,
      "loss": 0.3689,
      "step": 35581
    },
    {
      "epoch": 1.632875957964297,
      "grad_norm": 0.47126999497413635,
      "learning_rate": 4.451517318025917e-06,
      "loss": 0.3514,
      "step": 35582
    },
    {
      "epoch": 1.6329218484695516,
      "grad_norm": 0.45324668288230896,
      "learning_rate": 4.451273610456732e-06,
      "loss": 0.3456,
      "step": 35583
    },
    {
      "epoch": 1.632967738974806,
      "grad_norm": 0.5103207230567932,
      "learning_rate": 4.451029904207062e-06,
      "loss": 0.348,
      "step": 35584
    },
    {
      "epoch": 1.6330136294800606,
      "grad_norm": 0.45469173789024353,
      "learning_rate": 4.450786199277492e-06,
      "loss": 0.3173,
      "step": 35585
    },
    {
      "epoch": 1.633059519985315,
      "grad_norm": 0.46323975920677185,
      "learning_rate": 4.450542495668605e-06,
      "loss": 0.3609,
      "step": 35586
    },
    {
      "epoch": 1.6331054104905696,
      "grad_norm": 0.45869237184524536,
      "learning_rate": 4.4502987933809935e-06,
      "loss": 0.3383,
      "step": 35587
    },
    {
      "epoch": 1.633151300995824,
      "grad_norm": 0.47390031814575195,
      "learning_rate": 4.4500550924152385e-06,
      "loss": 0.3581,
      "step": 35588
    },
    {
      "epoch": 1.6331971915010786,
      "grad_norm": 0.4739677906036377,
      "learning_rate": 4.449811392771927e-06,
      "loss": 0.3352,
      "step": 35589
    },
    {
      "epoch": 1.6332430820063328,
      "grad_norm": 0.49291273951530457,
      "learning_rate": 4.449567694451646e-06,
      "loss": 0.3978,
      "step": 35590
    },
    {
      "epoch": 1.6332889725115873,
      "grad_norm": 0.4669926166534424,
      "learning_rate": 4.449323997454982e-06,
      "loss": 0.3408,
      "step": 35591
    },
    {
      "epoch": 1.6333348630168418,
      "grad_norm": 0.4422406852245331,
      "learning_rate": 4.449080301782517e-06,
      "loss": 0.2799,
      "step": 35592
    },
    {
      "epoch": 1.6333807535220963,
      "grad_norm": 0.45932871103286743,
      "learning_rate": 4.448836607434842e-06,
      "loss": 0.3624,
      "step": 35593
    },
    {
      "epoch": 1.6334266440273506,
      "grad_norm": 0.46716800332069397,
      "learning_rate": 4.448592914412542e-06,
      "loss": 0.3712,
      "step": 35594
    },
    {
      "epoch": 1.633472534532605,
      "grad_norm": 0.48791664838790894,
      "learning_rate": 4.4483492227162e-06,
      "loss": 0.365,
      "step": 35595
    },
    {
      "epoch": 1.6335184250378596,
      "grad_norm": 0.4919716417789459,
      "learning_rate": 4.448105532346406e-06,
      "loss": 0.4296,
      "step": 35596
    },
    {
      "epoch": 1.633564315543114,
      "grad_norm": 0.4869931936264038,
      "learning_rate": 4.447861843303743e-06,
      "loss": 0.3734,
      "step": 35597
    },
    {
      "epoch": 1.6336102060483686,
      "grad_norm": 0.5010695457458496,
      "learning_rate": 4.447618155588798e-06,
      "loss": 0.4287,
      "step": 35598
    },
    {
      "epoch": 1.633656096553623,
      "grad_norm": 0.42879530787467957,
      "learning_rate": 4.447374469202157e-06,
      "loss": 0.2804,
      "step": 35599
    },
    {
      "epoch": 1.6337019870588776,
      "grad_norm": 0.5159753561019897,
      "learning_rate": 4.4471307841444064e-06,
      "loss": 0.3996,
      "step": 35600
    },
    {
      "epoch": 1.633747877564132,
      "grad_norm": 0.5035858154296875,
      "learning_rate": 4.446887100416132e-06,
      "loss": 0.4228,
      "step": 35601
    },
    {
      "epoch": 1.6337937680693866,
      "grad_norm": 0.4661635458469391,
      "learning_rate": 4.446643418017917e-06,
      "loss": 0.3696,
      "step": 35602
    },
    {
      "epoch": 1.6338396585746409,
      "grad_norm": 0.45550912618637085,
      "learning_rate": 4.446399736950352e-06,
      "loss": 0.3488,
      "step": 35603
    },
    {
      "epoch": 1.6338855490798954,
      "grad_norm": 0.4925895929336548,
      "learning_rate": 4.446156057214021e-06,
      "loss": 0.3892,
      "step": 35604
    },
    {
      "epoch": 1.6339314395851499,
      "grad_norm": 0.4821862578392029,
      "learning_rate": 4.445912378809509e-06,
      "loss": 0.4019,
      "step": 35605
    },
    {
      "epoch": 1.6339773300904041,
      "grad_norm": 0.5001386404037476,
      "learning_rate": 4.445668701737404e-06,
      "loss": 0.3891,
      "step": 35606
    },
    {
      "epoch": 1.6340232205956586,
      "grad_norm": 0.47938376665115356,
      "learning_rate": 4.445425025998291e-06,
      "loss": 0.3918,
      "step": 35607
    },
    {
      "epoch": 1.6340691111009131,
      "grad_norm": 0.5095259547233582,
      "learning_rate": 4.445181351592754e-06,
      "loss": 0.3743,
      "step": 35608
    },
    {
      "epoch": 1.6341150016061676,
      "grad_norm": 0.5015885829925537,
      "learning_rate": 4.444937678521382e-06,
      "loss": 0.3679,
      "step": 35609
    },
    {
      "epoch": 1.6341608921114221,
      "grad_norm": 0.48548418283462524,
      "learning_rate": 4.4446940067847596e-06,
      "loss": 0.4005,
      "step": 35610
    },
    {
      "epoch": 1.6342067826166766,
      "grad_norm": 0.4505102038383484,
      "learning_rate": 4.444450336383472e-06,
      "loss": 0.3495,
      "step": 35611
    },
    {
      "epoch": 1.6342526731219311,
      "grad_norm": 0.5070098638534546,
      "learning_rate": 4.444206667318107e-06,
      "loss": 0.4246,
      "step": 35612
    },
    {
      "epoch": 1.6342985636271856,
      "grad_norm": 0.480770081281662,
      "learning_rate": 4.443962999589251e-06,
      "loss": 0.3877,
      "step": 35613
    },
    {
      "epoch": 1.6343444541324401,
      "grad_norm": 0.45760998129844666,
      "learning_rate": 4.443719333197484e-06,
      "loss": 0.3288,
      "step": 35614
    },
    {
      "epoch": 1.6343903446376946,
      "grad_norm": 0.49743637442588806,
      "learning_rate": 4.4434756681434e-06,
      "loss": 0.4411,
      "step": 35615
    },
    {
      "epoch": 1.634436235142949,
      "grad_norm": 0.48857566714286804,
      "learning_rate": 4.443232004427581e-06,
      "loss": 0.4169,
      "step": 35616
    },
    {
      "epoch": 1.6344821256482034,
      "grad_norm": 0.4636232256889343,
      "learning_rate": 4.442988342050613e-06,
      "loss": 0.3434,
      "step": 35617
    },
    {
      "epoch": 1.6345280161534579,
      "grad_norm": 0.41896358132362366,
      "learning_rate": 4.442744681013082e-06,
      "loss": 0.2875,
      "step": 35618
    },
    {
      "epoch": 1.6345739066587122,
      "grad_norm": 0.48809313774108887,
      "learning_rate": 4.442501021315575e-06,
      "loss": 0.3755,
      "step": 35619
    },
    {
      "epoch": 1.6346197971639667,
      "grad_norm": 0.4602876305580139,
      "learning_rate": 4.442257362958674e-06,
      "loss": 0.3694,
      "step": 35620
    },
    {
      "epoch": 1.6346656876692212,
      "grad_norm": 0.472431480884552,
      "learning_rate": 4.442013705942972e-06,
      "loss": 0.3619,
      "step": 35621
    },
    {
      "epoch": 1.6347115781744757,
      "grad_norm": 0.4966350197792053,
      "learning_rate": 4.441770050269051e-06,
      "loss": 0.409,
      "step": 35622
    },
    {
      "epoch": 1.6347574686797302,
      "grad_norm": 0.4724692702293396,
      "learning_rate": 4.441526395937493e-06,
      "loss": 0.4084,
      "step": 35623
    },
    {
      "epoch": 1.6348033591849847,
      "grad_norm": 0.4232155978679657,
      "learning_rate": 4.44128274294889e-06,
      "loss": 0.3098,
      "step": 35624
    },
    {
      "epoch": 1.6348492496902391,
      "grad_norm": 0.5240747928619385,
      "learning_rate": 4.4410390913038264e-06,
      "loss": 0.4283,
      "step": 35625
    },
    {
      "epoch": 1.6348951401954936,
      "grad_norm": 0.444128155708313,
      "learning_rate": 4.4407954410028865e-06,
      "loss": 0.287,
      "step": 35626
    },
    {
      "epoch": 1.6349410307007481,
      "grad_norm": 0.48619914054870605,
      "learning_rate": 4.440551792046657e-06,
      "loss": 0.3735,
      "step": 35627
    },
    {
      "epoch": 1.6349869212060026,
      "grad_norm": 0.5236396789550781,
      "learning_rate": 4.440308144435725e-06,
      "loss": 0.446,
      "step": 35628
    },
    {
      "epoch": 1.635032811711257,
      "grad_norm": 0.4722157120704651,
      "learning_rate": 4.440064498170673e-06,
      "loss": 0.3797,
      "step": 35629
    },
    {
      "epoch": 1.6350787022165114,
      "grad_norm": 0.43018248677253723,
      "learning_rate": 4.43982085325209e-06,
      "loss": 0.2873,
      "step": 35630
    },
    {
      "epoch": 1.635124592721766,
      "grad_norm": 0.5092353224754333,
      "learning_rate": 4.439577209680563e-06,
      "loss": 0.4322,
      "step": 35631
    },
    {
      "epoch": 1.6351704832270202,
      "grad_norm": 0.48995721340179443,
      "learning_rate": 4.439333567456675e-06,
      "loss": 0.4257,
      "step": 35632
    },
    {
      "epoch": 1.6352163737322747,
      "grad_norm": 0.46206921339035034,
      "learning_rate": 4.439089926581011e-06,
      "loss": 0.3162,
      "step": 35633
    },
    {
      "epoch": 1.6352622642375292,
      "grad_norm": 0.5321056246757507,
      "learning_rate": 4.438846287054161e-06,
      "loss": 0.4445,
      "step": 35634
    },
    {
      "epoch": 1.6353081547427837,
      "grad_norm": 0.4654044806957245,
      "learning_rate": 4.4386026488767075e-06,
      "loss": 0.3604,
      "step": 35635
    },
    {
      "epoch": 1.6353540452480382,
      "grad_norm": 0.4397132396697998,
      "learning_rate": 4.438359012049236e-06,
      "loss": 0.326,
      "step": 35636
    },
    {
      "epoch": 1.6353999357532927,
      "grad_norm": 0.45030948519706726,
      "learning_rate": 4.4381153765723366e-06,
      "loss": 0.364,
      "step": 35637
    },
    {
      "epoch": 1.6354458262585472,
      "grad_norm": 0.48009103536605835,
      "learning_rate": 4.437871742446591e-06,
      "loss": 0.3527,
      "step": 35638
    },
    {
      "epoch": 1.6354917167638017,
      "grad_norm": 0.4942985475063324,
      "learning_rate": 4.4376281096725855e-06,
      "loss": 0.3784,
      "step": 35639
    },
    {
      "epoch": 1.6355376072690562,
      "grad_norm": 0.4986535906791687,
      "learning_rate": 4.437384478250909e-06,
      "loss": 0.4117,
      "step": 35640
    },
    {
      "epoch": 1.6355834977743104,
      "grad_norm": 0.43956705927848816,
      "learning_rate": 4.4371408481821445e-06,
      "loss": 0.2991,
      "step": 35641
    },
    {
      "epoch": 1.635629388279565,
      "grad_norm": 0.472791850566864,
      "learning_rate": 4.436897219466877e-06,
      "loss": 0.3327,
      "step": 35642
    },
    {
      "epoch": 1.6356752787848194,
      "grad_norm": 0.47992491722106934,
      "learning_rate": 4.436653592105697e-06,
      "loss": 0.4125,
      "step": 35643
    },
    {
      "epoch": 1.6357211692900737,
      "grad_norm": 0.47765377163887024,
      "learning_rate": 4.436409966099185e-06,
      "loss": 0.3617,
      "step": 35644
    },
    {
      "epoch": 1.6357670597953282,
      "grad_norm": 0.5129196047782898,
      "learning_rate": 4.4361663414479286e-06,
      "loss": 0.4269,
      "step": 35645
    },
    {
      "epoch": 1.6358129503005827,
      "grad_norm": 0.5107359886169434,
      "learning_rate": 4.435922718152516e-06,
      "loss": 0.4252,
      "step": 35646
    },
    {
      "epoch": 1.6358588408058372,
      "grad_norm": 0.4762175381183624,
      "learning_rate": 4.435679096213531e-06,
      "loss": 0.3418,
      "step": 35647
    },
    {
      "epoch": 1.6359047313110917,
      "grad_norm": 0.4651995897293091,
      "learning_rate": 4.435435475631559e-06,
      "loss": 0.3623,
      "step": 35648
    },
    {
      "epoch": 1.6359506218163462,
      "grad_norm": 0.4553166329860687,
      "learning_rate": 4.435191856407187e-06,
      "loss": 0.3648,
      "step": 35649
    },
    {
      "epoch": 1.6359965123216007,
      "grad_norm": 0.43616724014282227,
      "learning_rate": 4.434948238541001e-06,
      "loss": 0.3024,
      "step": 35650
    },
    {
      "epoch": 1.6360424028268552,
      "grad_norm": 0.48796477913856506,
      "learning_rate": 4.434704622033583e-06,
      "loss": 0.4176,
      "step": 35651
    },
    {
      "epoch": 1.6360882933321097,
      "grad_norm": 0.48956993222236633,
      "learning_rate": 4.4344610068855245e-06,
      "loss": 0.3833,
      "step": 35652
    },
    {
      "epoch": 1.6361341838373642,
      "grad_norm": 0.47219109535217285,
      "learning_rate": 4.434217393097409e-06,
      "loss": 0.3959,
      "step": 35653
    },
    {
      "epoch": 1.6361800743426185,
      "grad_norm": 0.47809115052223206,
      "learning_rate": 4.433973780669821e-06,
      "loss": 0.3956,
      "step": 35654
    },
    {
      "epoch": 1.636225964847873,
      "grad_norm": 0.46596962213516235,
      "learning_rate": 4.433730169603349e-06,
      "loss": 0.3642,
      "step": 35655
    },
    {
      "epoch": 1.6362718553531275,
      "grad_norm": 0.4518429636955261,
      "learning_rate": 4.433486559898577e-06,
      "loss": 0.3208,
      "step": 35656
    },
    {
      "epoch": 1.6363177458583817,
      "grad_norm": 0.5009412169456482,
      "learning_rate": 4.433242951556088e-06,
      "loss": 0.4411,
      "step": 35657
    },
    {
      "epoch": 1.6363636363636362,
      "grad_norm": 0.43304285407066345,
      "learning_rate": 4.4329993445764744e-06,
      "loss": 0.2858,
      "step": 35658
    },
    {
      "epoch": 1.6364095268688907,
      "grad_norm": 0.46118709444999695,
      "learning_rate": 4.432755738960317e-06,
      "loss": 0.3386,
      "step": 35659
    },
    {
      "epoch": 1.6364554173741452,
      "grad_norm": 0.49600446224212646,
      "learning_rate": 4.432512134708203e-06,
      "loss": 0.3669,
      "step": 35660
    },
    {
      "epoch": 1.6365013078793997,
      "grad_norm": 0.4674980640411377,
      "learning_rate": 4.432268531820718e-06,
      "loss": 0.3494,
      "step": 35661
    },
    {
      "epoch": 1.6365471983846542,
      "grad_norm": 0.5115658640861511,
      "learning_rate": 4.43202493029845e-06,
      "loss": 0.4434,
      "step": 35662
    },
    {
      "epoch": 1.6365930888899087,
      "grad_norm": 0.4620290994644165,
      "learning_rate": 4.431781330141981e-06,
      "loss": 0.3318,
      "step": 35663
    },
    {
      "epoch": 1.6366389793951632,
      "grad_norm": 0.46119529008865356,
      "learning_rate": 4.431537731351897e-06,
      "loss": 0.3202,
      "step": 35664
    },
    {
      "epoch": 1.6366848699004177,
      "grad_norm": 0.48112353682518005,
      "learning_rate": 4.4312941339287894e-06,
      "loss": 0.3252,
      "step": 35665
    },
    {
      "epoch": 1.6367307604056722,
      "grad_norm": 0.49085715413093567,
      "learning_rate": 4.431050537873238e-06,
      "loss": 0.4159,
      "step": 35666
    },
    {
      "epoch": 1.6367766509109265,
      "grad_norm": 0.4530125558376312,
      "learning_rate": 4.430806943185828e-06,
      "loss": 0.3263,
      "step": 35667
    },
    {
      "epoch": 1.636822541416181,
      "grad_norm": 0.5141487717628479,
      "learning_rate": 4.43056334986715e-06,
      "loss": 0.3582,
      "step": 35668
    },
    {
      "epoch": 1.6368684319214355,
      "grad_norm": 0.4410403072834015,
      "learning_rate": 4.430319757917788e-06,
      "loss": 0.2958,
      "step": 35669
    },
    {
      "epoch": 1.6369143224266898,
      "grad_norm": 0.4575383961200714,
      "learning_rate": 4.430076167338326e-06,
      "loss": 0.3047,
      "step": 35670
    },
    {
      "epoch": 1.6369602129319443,
      "grad_norm": 0.5113713145256042,
      "learning_rate": 4.429832578129352e-06,
      "loss": 0.4092,
      "step": 35671
    },
    {
      "epoch": 1.6370061034371988,
      "grad_norm": 0.445078581571579,
      "learning_rate": 4.42958899029145e-06,
      "loss": 0.2902,
      "step": 35672
    },
    {
      "epoch": 1.6370519939424533,
      "grad_norm": 0.4611130356788635,
      "learning_rate": 4.429345403825205e-06,
      "loss": 0.2989,
      "step": 35673
    },
    {
      "epoch": 1.6370978844477078,
      "grad_norm": 0.4543219208717346,
      "learning_rate": 4.4291018187312055e-06,
      "loss": 0.3287,
      "step": 35674
    },
    {
      "epoch": 1.6371437749529623,
      "grad_norm": 0.46497899293899536,
      "learning_rate": 4.428858235010037e-06,
      "loss": 0.2495,
      "step": 35675
    },
    {
      "epoch": 1.6371896654582168,
      "grad_norm": 0.46645691990852356,
      "learning_rate": 4.428614652662283e-06,
      "loss": 0.33,
      "step": 35676
    },
    {
      "epoch": 1.6372355559634713,
      "grad_norm": 0.5070371031761169,
      "learning_rate": 4.428371071688531e-06,
      "loss": 0.3813,
      "step": 35677
    },
    {
      "epoch": 1.6372814464687258,
      "grad_norm": 0.4763984978199005,
      "learning_rate": 4.4281274920893656e-06,
      "loss": 0.3436,
      "step": 35678
    },
    {
      "epoch": 1.63732733697398,
      "grad_norm": 0.4638059437274933,
      "learning_rate": 4.427883913865372e-06,
      "loss": 0.3455,
      "step": 35679
    },
    {
      "epoch": 1.6373732274792345,
      "grad_norm": 0.45774126052856445,
      "learning_rate": 4.427640337017139e-06,
      "loss": 0.3073,
      "step": 35680
    },
    {
      "epoch": 1.637419117984489,
      "grad_norm": 0.4965127408504486,
      "learning_rate": 4.42739676154525e-06,
      "loss": 0.3966,
      "step": 35681
    },
    {
      "epoch": 1.6374650084897435,
      "grad_norm": 0.46903929114341736,
      "learning_rate": 4.427153187450291e-06,
      "loss": 0.3807,
      "step": 35682
    },
    {
      "epoch": 1.6375108989949978,
      "grad_norm": 0.5189263224601746,
      "learning_rate": 4.426909614732847e-06,
      "loss": 0.4119,
      "step": 35683
    },
    {
      "epoch": 1.6375567895002523,
      "grad_norm": 0.5395647287368774,
      "learning_rate": 4.426666043393506e-06,
      "loss": 0.4608,
      "step": 35684
    },
    {
      "epoch": 1.6376026800055068,
      "grad_norm": 0.4999631643295288,
      "learning_rate": 4.42642247343285e-06,
      "loss": 0.3856,
      "step": 35685
    },
    {
      "epoch": 1.6376485705107613,
      "grad_norm": 0.5340934991836548,
      "learning_rate": 4.426178904851468e-06,
      "loss": 0.5455,
      "step": 35686
    },
    {
      "epoch": 1.6376944610160158,
      "grad_norm": 0.4663400650024414,
      "learning_rate": 4.425935337649947e-06,
      "loss": 0.345,
      "step": 35687
    },
    {
      "epoch": 1.6377403515212703,
      "grad_norm": 0.49454426765441895,
      "learning_rate": 4.425691771828866e-06,
      "loss": 0.3882,
      "step": 35688
    },
    {
      "epoch": 1.6377862420265248,
      "grad_norm": 0.40748757123947144,
      "learning_rate": 4.425448207388819e-06,
      "loss": 0.2545,
      "step": 35689
    },
    {
      "epoch": 1.6378321325317793,
      "grad_norm": 0.4889836013317108,
      "learning_rate": 4.425204644330385e-06,
      "loss": 0.3993,
      "step": 35690
    },
    {
      "epoch": 1.6378780230370338,
      "grad_norm": 0.46511900424957275,
      "learning_rate": 4.4249610826541534e-06,
      "loss": 0.3789,
      "step": 35691
    },
    {
      "epoch": 1.637923913542288,
      "grad_norm": 0.46183717250823975,
      "learning_rate": 4.424717522360709e-06,
      "loss": 0.3374,
      "step": 35692
    },
    {
      "epoch": 1.6379698040475426,
      "grad_norm": 0.44178566336631775,
      "learning_rate": 4.424473963450638e-06,
      "loss": 0.3271,
      "step": 35693
    },
    {
      "epoch": 1.638015694552797,
      "grad_norm": 0.48861458897590637,
      "learning_rate": 4.424230405924523e-06,
      "loss": 0.4527,
      "step": 35694
    },
    {
      "epoch": 1.6380615850580513,
      "grad_norm": 0.45539265871047974,
      "learning_rate": 4.4239868497829544e-06,
      "loss": 0.3409,
      "step": 35695
    },
    {
      "epoch": 1.6381074755633058,
      "grad_norm": 0.45385536551475525,
      "learning_rate": 4.423743295026516e-06,
      "loss": 0.3377,
      "step": 35696
    },
    {
      "epoch": 1.6381533660685603,
      "grad_norm": 0.47125518321990967,
      "learning_rate": 4.423499741655792e-06,
      "loss": 0.3548,
      "step": 35697
    },
    {
      "epoch": 1.6381992565738148,
      "grad_norm": 0.40974971652030945,
      "learning_rate": 4.423256189671369e-06,
      "loss": 0.252,
      "step": 35698
    },
    {
      "epoch": 1.6382451470790693,
      "grad_norm": 0.4433223307132721,
      "learning_rate": 4.423012639073834e-06,
      "loss": 0.3281,
      "step": 35699
    },
    {
      "epoch": 1.6382910375843238,
      "grad_norm": 0.46210217475891113,
      "learning_rate": 4.4227690898637706e-06,
      "loss": 0.3633,
      "step": 35700
    },
    {
      "epoch": 1.6383369280895783,
      "grad_norm": 0.4420068860054016,
      "learning_rate": 4.422525542041765e-06,
      "loss": 0.3456,
      "step": 35701
    },
    {
      "epoch": 1.6383828185948328,
      "grad_norm": 0.4367007315158844,
      "learning_rate": 4.4222819956084034e-06,
      "loss": 0.3227,
      "step": 35702
    },
    {
      "epoch": 1.6384287091000873,
      "grad_norm": 0.4843018054962158,
      "learning_rate": 4.422038450564272e-06,
      "loss": 0.3768,
      "step": 35703
    },
    {
      "epoch": 1.6384745996053418,
      "grad_norm": 0.47956719994544983,
      "learning_rate": 4.421794906909955e-06,
      "loss": 0.3962,
      "step": 35704
    },
    {
      "epoch": 1.638520490110596,
      "grad_norm": 0.45831501483917236,
      "learning_rate": 4.421551364646039e-06,
      "loss": 0.2951,
      "step": 35705
    },
    {
      "epoch": 1.6385663806158506,
      "grad_norm": 0.49406731128692627,
      "learning_rate": 4.42130782377311e-06,
      "loss": 0.4475,
      "step": 35706
    },
    {
      "epoch": 1.638612271121105,
      "grad_norm": 0.44962841272354126,
      "learning_rate": 4.4210642842917515e-06,
      "loss": 0.3341,
      "step": 35707
    },
    {
      "epoch": 1.6386581616263594,
      "grad_norm": 0.4390642046928406,
      "learning_rate": 4.4208207462025524e-06,
      "loss": 0.3234,
      "step": 35708
    },
    {
      "epoch": 1.6387040521316139,
      "grad_norm": 0.48307743668556213,
      "learning_rate": 4.420577209506097e-06,
      "loss": 0.4099,
      "step": 35709
    },
    {
      "epoch": 1.6387499426368684,
      "grad_norm": 0.48022863268852234,
      "learning_rate": 4.4203336742029675e-06,
      "loss": 0.3806,
      "step": 35710
    },
    {
      "epoch": 1.6387958331421228,
      "grad_norm": 0.46951502561569214,
      "learning_rate": 4.420090140293756e-06,
      "loss": 0.3416,
      "step": 35711
    },
    {
      "epoch": 1.6388417236473773,
      "grad_norm": 0.4614250659942627,
      "learning_rate": 4.419846607779043e-06,
      "loss": 0.3378,
      "step": 35712
    },
    {
      "epoch": 1.6388876141526318,
      "grad_norm": 0.4302309453487396,
      "learning_rate": 4.419603076659416e-06,
      "loss": 0.3015,
      "step": 35713
    },
    {
      "epoch": 1.6389335046578863,
      "grad_norm": 0.4283803403377533,
      "learning_rate": 4.419359546935461e-06,
      "loss": 0.2891,
      "step": 35714
    },
    {
      "epoch": 1.6389793951631408,
      "grad_norm": 0.4737962484359741,
      "learning_rate": 4.419116018607762e-06,
      "loss": 0.4054,
      "step": 35715
    },
    {
      "epoch": 1.6390252856683953,
      "grad_norm": 0.4839833974838257,
      "learning_rate": 4.418872491676905e-06,
      "loss": 0.392,
      "step": 35716
    },
    {
      "epoch": 1.6390711761736498,
      "grad_norm": 0.47726672887802124,
      "learning_rate": 4.418628966143478e-06,
      "loss": 0.3963,
      "step": 35717
    },
    {
      "epoch": 1.6391170666789041,
      "grad_norm": 0.45434364676475525,
      "learning_rate": 4.418385442008065e-06,
      "loss": 0.321,
      "step": 35718
    },
    {
      "epoch": 1.6391629571841586,
      "grad_norm": 0.5004035830497742,
      "learning_rate": 4.418141919271249e-06,
      "loss": 0.4608,
      "step": 35719
    },
    {
      "epoch": 1.639208847689413,
      "grad_norm": 0.47498834133148193,
      "learning_rate": 4.4178983979336206e-06,
      "loss": 0.3719,
      "step": 35720
    },
    {
      "epoch": 1.6392547381946674,
      "grad_norm": 0.5097767114639282,
      "learning_rate": 4.417654877995762e-06,
      "loss": 0.4361,
      "step": 35721
    },
    {
      "epoch": 1.6393006286999219,
      "grad_norm": 0.4756438732147217,
      "learning_rate": 4.417411359458259e-06,
      "loss": 0.3419,
      "step": 35722
    },
    {
      "epoch": 1.6393465192051764,
      "grad_norm": 0.5038793683052063,
      "learning_rate": 4.417167842321699e-06,
      "loss": 0.3645,
      "step": 35723
    },
    {
      "epoch": 1.6393924097104309,
      "grad_norm": 0.48670700192451477,
      "learning_rate": 4.4169243265866656e-06,
      "loss": 0.3894,
      "step": 35724
    },
    {
      "epoch": 1.6394383002156854,
      "grad_norm": 0.4713616669178009,
      "learning_rate": 4.416680812253745e-06,
      "loss": 0.3841,
      "step": 35725
    },
    {
      "epoch": 1.6394841907209399,
      "grad_norm": 0.496724009513855,
      "learning_rate": 4.416437299323523e-06,
      "loss": 0.405,
      "step": 35726
    },
    {
      "epoch": 1.6395300812261944,
      "grad_norm": 0.48274707794189453,
      "learning_rate": 4.416193787796587e-06,
      "loss": 0.4238,
      "step": 35727
    },
    {
      "epoch": 1.6395759717314489,
      "grad_norm": 0.472925066947937,
      "learning_rate": 4.415950277673519e-06,
      "loss": 0.332,
      "step": 35728
    },
    {
      "epoch": 1.6396218622367034,
      "grad_norm": 0.5082917809486389,
      "learning_rate": 4.415706768954904e-06,
      "loss": 0.4158,
      "step": 35729
    },
    {
      "epoch": 1.6396677527419576,
      "grad_norm": 0.46859049797058105,
      "learning_rate": 4.4154632616413335e-06,
      "loss": 0.3702,
      "step": 35730
    },
    {
      "epoch": 1.6397136432472121,
      "grad_norm": 0.45591527223587036,
      "learning_rate": 4.415219755733389e-06,
      "loss": 0.3626,
      "step": 35731
    },
    {
      "epoch": 1.6397595337524666,
      "grad_norm": 0.4450148642063141,
      "learning_rate": 4.414976251231654e-06,
      "loss": 0.2943,
      "step": 35732
    },
    {
      "epoch": 1.639805424257721,
      "grad_norm": 0.4886983335018158,
      "learning_rate": 4.414732748136719e-06,
      "loss": 0.356,
      "step": 35733
    },
    {
      "epoch": 1.6398513147629754,
      "grad_norm": 0.4273974299430847,
      "learning_rate": 4.414489246449166e-06,
      "loss": 0.2776,
      "step": 35734
    },
    {
      "epoch": 1.63989720526823,
      "grad_norm": 0.45374929904937744,
      "learning_rate": 4.414245746169581e-06,
      "loss": 0.3289,
      "step": 35735
    },
    {
      "epoch": 1.6399430957734844,
      "grad_norm": 0.43119311332702637,
      "learning_rate": 4.414002247298551e-06,
      "loss": 0.2809,
      "step": 35736
    },
    {
      "epoch": 1.639988986278739,
      "grad_norm": 0.47092658281326294,
      "learning_rate": 4.41375874983666e-06,
      "loss": 0.3749,
      "step": 35737
    },
    {
      "epoch": 1.6400348767839934,
      "grad_norm": 0.47063392400741577,
      "learning_rate": 4.4135152537844935e-06,
      "loss": 0.3305,
      "step": 35738
    },
    {
      "epoch": 1.640080767289248,
      "grad_norm": 0.4680700898170471,
      "learning_rate": 4.413271759142639e-06,
      "loss": 0.3773,
      "step": 35739
    },
    {
      "epoch": 1.6401266577945024,
      "grad_norm": 0.5166497230529785,
      "learning_rate": 4.413028265911681e-06,
      "loss": 0.4445,
      "step": 35740
    },
    {
      "epoch": 1.640172548299757,
      "grad_norm": 0.4919517934322357,
      "learning_rate": 4.412784774092203e-06,
      "loss": 0.3635,
      "step": 35741
    },
    {
      "epoch": 1.6402184388050114,
      "grad_norm": 0.46016979217529297,
      "learning_rate": 4.412541283684794e-06,
      "loss": 0.3072,
      "step": 35742
    },
    {
      "epoch": 1.6402643293102657,
      "grad_norm": 0.46814101934432983,
      "learning_rate": 4.412297794690038e-06,
      "loss": 0.3468,
      "step": 35743
    },
    {
      "epoch": 1.6403102198155202,
      "grad_norm": 0.5016466975212097,
      "learning_rate": 4.4120543071085185e-06,
      "loss": 0.3945,
      "step": 35744
    },
    {
      "epoch": 1.6403561103207747,
      "grad_norm": 0.46005645394325256,
      "learning_rate": 4.411810820940825e-06,
      "loss": 0.3127,
      "step": 35745
    },
    {
      "epoch": 1.640402000826029,
      "grad_norm": 0.45622026920318604,
      "learning_rate": 4.411567336187539e-06,
      "loss": 0.3247,
      "step": 35746
    },
    {
      "epoch": 1.6404478913312834,
      "grad_norm": 0.5015740394592285,
      "learning_rate": 4.411323852849249e-06,
      "loss": 0.374,
      "step": 35747
    },
    {
      "epoch": 1.640493781836538,
      "grad_norm": 0.4581682085990906,
      "learning_rate": 4.411080370926539e-06,
      "loss": 0.32,
      "step": 35748
    },
    {
      "epoch": 1.6405396723417924,
      "grad_norm": 0.4913271963596344,
      "learning_rate": 4.410836890419995e-06,
      "loss": 0.4027,
      "step": 35749
    },
    {
      "epoch": 1.640585562847047,
      "grad_norm": 0.4913686513900757,
      "learning_rate": 4.410593411330201e-06,
      "loss": 0.3897,
      "step": 35750
    },
    {
      "epoch": 1.6406314533523014,
      "grad_norm": 0.4772675037384033,
      "learning_rate": 4.4103499336577445e-06,
      "loss": 0.4017,
      "step": 35751
    },
    {
      "epoch": 1.640677343857556,
      "grad_norm": 0.45257773995399475,
      "learning_rate": 4.410106457403212e-06,
      "loss": 0.3262,
      "step": 35752
    },
    {
      "epoch": 1.6407232343628104,
      "grad_norm": 0.46693772077560425,
      "learning_rate": 4.409862982567185e-06,
      "loss": 0.3501,
      "step": 35753
    },
    {
      "epoch": 1.640769124868065,
      "grad_norm": 0.4553653299808502,
      "learning_rate": 4.409619509150253e-06,
      "loss": 0.3772,
      "step": 35754
    },
    {
      "epoch": 1.6408150153733194,
      "grad_norm": 0.48222243785858154,
      "learning_rate": 4.4093760371529994e-06,
      "loss": 0.3874,
      "step": 35755
    },
    {
      "epoch": 1.6408609058785737,
      "grad_norm": 0.5167502164840698,
      "learning_rate": 4.409132566576008e-06,
      "loss": 0.4269,
      "step": 35756
    },
    {
      "epoch": 1.6409067963838282,
      "grad_norm": 0.48042425513267517,
      "learning_rate": 4.40888909741987e-06,
      "loss": 0.3749,
      "step": 35757
    },
    {
      "epoch": 1.6409526868890827,
      "grad_norm": 0.49616631865501404,
      "learning_rate": 4.408645629685166e-06,
      "loss": 0.3191,
      "step": 35758
    },
    {
      "epoch": 1.640998577394337,
      "grad_norm": 0.47820743918418884,
      "learning_rate": 4.40840216337248e-06,
      "loss": 0.3984,
      "step": 35759
    },
    {
      "epoch": 1.6410444678995915,
      "grad_norm": 0.4152669310569763,
      "learning_rate": 4.408158698482402e-06,
      "loss": 0.294,
      "step": 35760
    },
    {
      "epoch": 1.641090358404846,
      "grad_norm": 0.41199105978012085,
      "learning_rate": 4.407915235015516e-06,
      "loss": 0.2794,
      "step": 35761
    },
    {
      "epoch": 1.6411362489101005,
      "grad_norm": 0.4818490743637085,
      "learning_rate": 4.407671772972406e-06,
      "loss": 0.4046,
      "step": 35762
    },
    {
      "epoch": 1.641182139415355,
      "grad_norm": 0.47653257846832275,
      "learning_rate": 4.407428312353659e-06,
      "loss": 0.3434,
      "step": 35763
    },
    {
      "epoch": 1.6412280299206095,
      "grad_norm": 0.4745029807090759,
      "learning_rate": 4.4071848531598605e-06,
      "loss": 0.3408,
      "step": 35764
    },
    {
      "epoch": 1.641273920425864,
      "grad_norm": 0.4521040916442871,
      "learning_rate": 4.406941395391595e-06,
      "loss": 0.3199,
      "step": 35765
    },
    {
      "epoch": 1.6413198109311185,
      "grad_norm": 0.5148151516914368,
      "learning_rate": 4.4066979390494455e-06,
      "loss": 0.4018,
      "step": 35766
    },
    {
      "epoch": 1.641365701436373,
      "grad_norm": 0.5304829478263855,
      "learning_rate": 4.406454484134004e-06,
      "loss": 0.4389,
      "step": 35767
    },
    {
      "epoch": 1.6414115919416272,
      "grad_norm": 0.453359991312027,
      "learning_rate": 4.406211030645851e-06,
      "loss": 0.363,
      "step": 35768
    },
    {
      "epoch": 1.6414574824468817,
      "grad_norm": 0.48904773592948914,
      "learning_rate": 4.405967578585571e-06,
      "loss": 0.402,
      "step": 35769
    },
    {
      "epoch": 1.6415033729521362,
      "grad_norm": 0.49641793966293335,
      "learning_rate": 4.405724127953754e-06,
      "loss": 0.3901,
      "step": 35770
    },
    {
      "epoch": 1.6415492634573907,
      "grad_norm": 0.4273126721382141,
      "learning_rate": 4.405480678750983e-06,
      "loss": 0.3186,
      "step": 35771
    },
    {
      "epoch": 1.641595153962645,
      "grad_norm": 0.46698257327079773,
      "learning_rate": 4.405237230977841e-06,
      "loss": 0.3447,
      "step": 35772
    },
    {
      "epoch": 1.6416410444678995,
      "grad_norm": 0.4783484637737274,
      "learning_rate": 4.404993784634918e-06,
      "loss": 0.3653,
      "step": 35773
    },
    {
      "epoch": 1.641686934973154,
      "grad_norm": 0.4641822874546051,
      "learning_rate": 4.404750339722796e-06,
      "loss": 0.3076,
      "step": 35774
    },
    {
      "epoch": 1.6417328254784085,
      "grad_norm": 0.4685385823249817,
      "learning_rate": 4.404506896242062e-06,
      "loss": 0.2974,
      "step": 35775
    },
    {
      "epoch": 1.641778715983663,
      "grad_norm": 0.46848735213279724,
      "learning_rate": 4.404263454193301e-06,
      "loss": 0.3766,
      "step": 35776
    },
    {
      "epoch": 1.6418246064889175,
      "grad_norm": 0.4908662438392639,
      "learning_rate": 4.404020013577099e-06,
      "loss": 0.3875,
      "step": 35777
    },
    {
      "epoch": 1.641870496994172,
      "grad_norm": 0.4622827172279358,
      "learning_rate": 4.403776574394039e-06,
      "loss": 0.3268,
      "step": 35778
    },
    {
      "epoch": 1.6419163874994265,
      "grad_norm": 0.5139374732971191,
      "learning_rate": 4.40353313664471e-06,
      "loss": 0.3581,
      "step": 35779
    },
    {
      "epoch": 1.641962278004681,
      "grad_norm": 0.4291189908981323,
      "learning_rate": 4.403289700329695e-06,
      "loss": 0.2732,
      "step": 35780
    },
    {
      "epoch": 1.6420081685099353,
      "grad_norm": 0.4418382942676544,
      "learning_rate": 4.4030462654495784e-06,
      "loss": 0.3084,
      "step": 35781
    },
    {
      "epoch": 1.6420540590151897,
      "grad_norm": 0.4305156171321869,
      "learning_rate": 4.402802832004949e-06,
      "loss": 0.2827,
      "step": 35782
    },
    {
      "epoch": 1.6420999495204442,
      "grad_norm": 0.4875820279121399,
      "learning_rate": 4.40255939999639e-06,
      "loss": 0.3633,
      "step": 35783
    },
    {
      "epoch": 1.6421458400256985,
      "grad_norm": 0.45124122500419617,
      "learning_rate": 4.402315969424486e-06,
      "loss": 0.3739,
      "step": 35784
    },
    {
      "epoch": 1.642191730530953,
      "grad_norm": 0.4058588445186615,
      "learning_rate": 4.402072540289825e-06,
      "loss": 0.2511,
      "step": 35785
    },
    {
      "epoch": 1.6422376210362075,
      "grad_norm": 0.47803589701652527,
      "learning_rate": 4.4018291125929905e-06,
      "loss": 0.3323,
      "step": 35786
    },
    {
      "epoch": 1.642283511541462,
      "grad_norm": 0.4509625732898712,
      "learning_rate": 4.4015856863345665e-06,
      "loss": 0.3208,
      "step": 35787
    },
    {
      "epoch": 1.6423294020467165,
      "grad_norm": 0.4389108419418335,
      "learning_rate": 4.401342261515142e-06,
      "loss": 0.3152,
      "step": 35788
    },
    {
      "epoch": 1.642375292551971,
      "grad_norm": 0.4610974192619324,
      "learning_rate": 4.4010988381353e-06,
      "loss": 0.3349,
      "step": 35789
    },
    {
      "epoch": 1.6424211830572255,
      "grad_norm": 0.4365984797477722,
      "learning_rate": 4.400855416195625e-06,
      "loss": 0.3258,
      "step": 35790
    },
    {
      "epoch": 1.64246707356248,
      "grad_norm": 0.4406372010707855,
      "learning_rate": 4.400611995696706e-06,
      "loss": 0.3096,
      "step": 35791
    },
    {
      "epoch": 1.6425129640677345,
      "grad_norm": 0.46938857436180115,
      "learning_rate": 4.400368576639125e-06,
      "loss": 0.3348,
      "step": 35792
    },
    {
      "epoch": 1.642558854572989,
      "grad_norm": 0.4995099902153015,
      "learning_rate": 4.400125159023466e-06,
      "loss": 0.4216,
      "step": 35793
    },
    {
      "epoch": 1.6426047450782433,
      "grad_norm": 0.4595467150211334,
      "learning_rate": 4.3998817428503195e-06,
      "loss": 0.3556,
      "step": 35794
    },
    {
      "epoch": 1.6426506355834978,
      "grad_norm": 0.4711065888404846,
      "learning_rate": 4.399638328120268e-06,
      "loss": 0.3722,
      "step": 35795
    },
    {
      "epoch": 1.6426965260887523,
      "grad_norm": 0.44299525022506714,
      "learning_rate": 4.399394914833897e-06,
      "loss": 0.3163,
      "step": 35796
    },
    {
      "epoch": 1.6427424165940065,
      "grad_norm": 0.500495970249176,
      "learning_rate": 4.399151502991789e-06,
      "loss": 0.3963,
      "step": 35797
    },
    {
      "epoch": 1.642788307099261,
      "grad_norm": 0.5117055773735046,
      "learning_rate": 4.398908092594535e-06,
      "loss": 0.4474,
      "step": 35798
    },
    {
      "epoch": 1.6428341976045155,
      "grad_norm": 0.46225759387016296,
      "learning_rate": 4.398664683642716e-06,
      "loss": 0.3199,
      "step": 35799
    },
    {
      "epoch": 1.64288008810977,
      "grad_norm": 0.4997924864292145,
      "learning_rate": 4.3984212761369176e-06,
      "loss": 0.407,
      "step": 35800
    },
    {
      "epoch": 1.6429259786150245,
      "grad_norm": 0.49980151653289795,
      "learning_rate": 4.398177870077728e-06,
      "loss": 0.3741,
      "step": 35801
    },
    {
      "epoch": 1.642971869120279,
      "grad_norm": 0.4796699583530426,
      "learning_rate": 4.397934465465731e-06,
      "loss": 0.4081,
      "step": 35802
    },
    {
      "epoch": 1.6430177596255335,
      "grad_norm": 0.4289434850215912,
      "learning_rate": 4.397691062301508e-06,
      "loss": 0.3122,
      "step": 35803
    },
    {
      "epoch": 1.643063650130788,
      "grad_norm": 0.5133978128433228,
      "learning_rate": 4.397447660585651e-06,
      "loss": 0.3105,
      "step": 35804
    },
    {
      "epoch": 1.6431095406360425,
      "grad_norm": 0.4320638179779053,
      "learning_rate": 4.3972042603187425e-06,
      "loss": 0.2922,
      "step": 35805
    },
    {
      "epoch": 1.643155431141297,
      "grad_norm": 0.4650939404964447,
      "learning_rate": 4.396960861501366e-06,
      "loss": 0.3638,
      "step": 35806
    },
    {
      "epoch": 1.6432013216465513,
      "grad_norm": 0.4647216498851776,
      "learning_rate": 4.3967174641341095e-06,
      "loss": 0.3077,
      "step": 35807
    },
    {
      "epoch": 1.6432472121518058,
      "grad_norm": 0.43632805347442627,
      "learning_rate": 4.3964740682175576e-06,
      "loss": 0.3227,
      "step": 35808
    },
    {
      "epoch": 1.6432931026570603,
      "grad_norm": 0.4735108017921448,
      "learning_rate": 4.396230673752292e-06,
      "loss": 0.3608,
      "step": 35809
    },
    {
      "epoch": 1.6433389931623146,
      "grad_norm": 0.4759521484375,
      "learning_rate": 4.395987280738904e-06,
      "loss": 0.3487,
      "step": 35810
    },
    {
      "epoch": 1.643384883667569,
      "grad_norm": 0.5054999589920044,
      "learning_rate": 4.395743889177975e-06,
      "loss": 0.4497,
      "step": 35811
    },
    {
      "epoch": 1.6434307741728236,
      "grad_norm": 0.5128865242004395,
      "learning_rate": 4.39550049907009e-06,
      "loss": 0.4271,
      "step": 35812
    },
    {
      "epoch": 1.643476664678078,
      "grad_norm": 0.5182333588600159,
      "learning_rate": 4.395257110415838e-06,
      "loss": 0.3868,
      "step": 35813
    },
    {
      "epoch": 1.6435225551833326,
      "grad_norm": 0.47527754306793213,
      "learning_rate": 4.3950137232158005e-06,
      "loss": 0.3721,
      "step": 35814
    },
    {
      "epoch": 1.643568445688587,
      "grad_norm": 0.46385812759399414,
      "learning_rate": 4.394770337470563e-06,
      "loss": 0.326,
      "step": 35815
    },
    {
      "epoch": 1.6436143361938416,
      "grad_norm": 0.48085999488830566,
      "learning_rate": 4.3945269531807135e-06,
      "loss": 0.3634,
      "step": 35816
    },
    {
      "epoch": 1.643660226699096,
      "grad_norm": 0.4314105808734894,
      "learning_rate": 4.394283570346835e-06,
      "loss": 0.3147,
      "step": 35817
    },
    {
      "epoch": 1.6437061172043506,
      "grad_norm": 0.45863136649131775,
      "learning_rate": 4.394040188969513e-06,
      "loss": 0.3744,
      "step": 35818
    },
    {
      "epoch": 1.6437520077096048,
      "grad_norm": 0.43358537554740906,
      "learning_rate": 4.3937968090493345e-06,
      "loss": 0.333,
      "step": 35819
    },
    {
      "epoch": 1.6437978982148593,
      "grad_norm": 0.464619517326355,
      "learning_rate": 4.3935534305868825e-06,
      "loss": 0.3169,
      "step": 35820
    },
    {
      "epoch": 1.6438437887201138,
      "grad_norm": 0.4987611770629883,
      "learning_rate": 4.393310053582741e-06,
      "loss": 0.3952,
      "step": 35821
    },
    {
      "epoch": 1.643889679225368,
      "grad_norm": 0.5382171869277954,
      "learning_rate": 4.393066678037501e-06,
      "loss": 0.5477,
      "step": 35822
    },
    {
      "epoch": 1.6439355697306226,
      "grad_norm": 0.4668377637863159,
      "learning_rate": 4.392823303951743e-06,
      "loss": 0.3447,
      "step": 35823
    },
    {
      "epoch": 1.643981460235877,
      "grad_norm": 0.43894046545028687,
      "learning_rate": 4.392579931326051e-06,
      "loss": 0.3162,
      "step": 35824
    },
    {
      "epoch": 1.6440273507411316,
      "grad_norm": 0.45414823293685913,
      "learning_rate": 4.392336560161015e-06,
      "loss": 0.3249,
      "step": 35825
    },
    {
      "epoch": 1.644073241246386,
      "grad_norm": 0.4436976909637451,
      "learning_rate": 4.392093190457217e-06,
      "loss": 0.333,
      "step": 35826
    },
    {
      "epoch": 1.6441191317516406,
      "grad_norm": 0.49503812193870544,
      "learning_rate": 4.391849822215244e-06,
      "loss": 0.3477,
      "step": 35827
    },
    {
      "epoch": 1.644165022256895,
      "grad_norm": 0.41574251651763916,
      "learning_rate": 4.391606455435679e-06,
      "loss": 0.2649,
      "step": 35828
    },
    {
      "epoch": 1.6442109127621496,
      "grad_norm": 0.463836133480072,
      "learning_rate": 4.391363090119109e-06,
      "loss": 0.3153,
      "step": 35829
    },
    {
      "epoch": 1.644256803267404,
      "grad_norm": 0.48424381017684937,
      "learning_rate": 4.391119726266119e-06,
      "loss": 0.3733,
      "step": 35830
    },
    {
      "epoch": 1.6443026937726586,
      "grad_norm": 0.470198392868042,
      "learning_rate": 4.390876363877292e-06,
      "loss": 0.3901,
      "step": 35831
    },
    {
      "epoch": 1.6443485842779129,
      "grad_norm": 0.42296624183654785,
      "learning_rate": 4.390633002953216e-06,
      "loss": 0.2977,
      "step": 35832
    },
    {
      "epoch": 1.6443944747831674,
      "grad_norm": 0.4460691809654236,
      "learning_rate": 4.390389643494477e-06,
      "loss": 0.3148,
      "step": 35833
    },
    {
      "epoch": 1.6444403652884219,
      "grad_norm": 0.44360625743865967,
      "learning_rate": 4.390146285501657e-06,
      "loss": 0.3729,
      "step": 35834
    },
    {
      "epoch": 1.6444862557936761,
      "grad_norm": 0.4957050681114197,
      "learning_rate": 4.389902928975343e-06,
      "loss": 0.3751,
      "step": 35835
    },
    {
      "epoch": 1.6445321462989306,
      "grad_norm": 0.45793092250823975,
      "learning_rate": 4.38965957391612e-06,
      "loss": 0.3063,
      "step": 35836
    },
    {
      "epoch": 1.6445780368041851,
      "grad_norm": 0.44571229815483093,
      "learning_rate": 4.389416220324572e-06,
      "loss": 0.3092,
      "step": 35837
    },
    {
      "epoch": 1.6446239273094396,
      "grad_norm": 0.513238251209259,
      "learning_rate": 4.389172868201286e-06,
      "loss": 0.4177,
      "step": 35838
    },
    {
      "epoch": 1.6446698178146941,
      "grad_norm": 0.45076093077659607,
      "learning_rate": 4.388929517546847e-06,
      "loss": 0.3169,
      "step": 35839
    },
    {
      "epoch": 1.6447157083199486,
      "grad_norm": 0.4670800566673279,
      "learning_rate": 4.38868616836184e-06,
      "loss": 0.4033,
      "step": 35840
    },
    {
      "epoch": 1.6447615988252031,
      "grad_norm": 0.4193473160266876,
      "learning_rate": 4.388442820646849e-06,
      "loss": 0.2853,
      "step": 35841
    },
    {
      "epoch": 1.6448074893304576,
      "grad_norm": 0.5827734470367432,
      "learning_rate": 4.38819947440246e-06,
      "loss": 0.2997,
      "step": 35842
    },
    {
      "epoch": 1.6448533798357121,
      "grad_norm": 0.467144250869751,
      "learning_rate": 4.387956129629257e-06,
      "loss": 0.3226,
      "step": 35843
    },
    {
      "epoch": 1.6448992703409666,
      "grad_norm": 0.46242573857307434,
      "learning_rate": 4.387712786327829e-06,
      "loss": 0.3611,
      "step": 35844
    },
    {
      "epoch": 1.6449451608462209,
      "grad_norm": 0.4546317756175995,
      "learning_rate": 4.387469444498758e-06,
      "loss": 0.3245,
      "step": 35845
    },
    {
      "epoch": 1.6449910513514754,
      "grad_norm": 0.46148136258125305,
      "learning_rate": 4.387226104142628e-06,
      "loss": 0.3511,
      "step": 35846
    },
    {
      "epoch": 1.6450369418567299,
      "grad_norm": 0.4385524094104767,
      "learning_rate": 4.386982765260027e-06,
      "loss": 0.3423,
      "step": 35847
    },
    {
      "epoch": 1.6450828323619842,
      "grad_norm": 0.4744969606399536,
      "learning_rate": 4.386739427851538e-06,
      "loss": 0.3797,
      "step": 35848
    },
    {
      "epoch": 1.6451287228672387,
      "grad_norm": 0.49610549211502075,
      "learning_rate": 4.386496091917748e-06,
      "loss": 0.3459,
      "step": 35849
    },
    {
      "epoch": 1.6451746133724932,
      "grad_norm": 0.4768170118331909,
      "learning_rate": 4.386252757459242e-06,
      "loss": 0.3728,
      "step": 35850
    },
    {
      "epoch": 1.6452205038777477,
      "grad_norm": 0.45942503213882446,
      "learning_rate": 4.386009424476603e-06,
      "loss": 0.3367,
      "step": 35851
    },
    {
      "epoch": 1.6452663943830022,
      "grad_norm": 0.4893627166748047,
      "learning_rate": 4.3857660929704175e-06,
      "loss": 0.3338,
      "step": 35852
    },
    {
      "epoch": 1.6453122848882566,
      "grad_norm": 0.4380716383457184,
      "learning_rate": 4.385522762941271e-06,
      "loss": 0.2978,
      "step": 35853
    },
    {
      "epoch": 1.6453581753935111,
      "grad_norm": 0.43358251452445984,
      "learning_rate": 4.385279434389749e-06,
      "loss": 0.2782,
      "step": 35854
    },
    {
      "epoch": 1.6454040658987656,
      "grad_norm": 0.4585760831832886,
      "learning_rate": 4.385036107316435e-06,
      "loss": 0.3574,
      "step": 35855
    },
    {
      "epoch": 1.6454499564040201,
      "grad_norm": 0.45548945665359497,
      "learning_rate": 4.384792781721916e-06,
      "loss": 0.3078,
      "step": 35856
    },
    {
      "epoch": 1.6454958469092744,
      "grad_norm": 0.436739444732666,
      "learning_rate": 4.3845494576067766e-06,
      "loss": 0.2932,
      "step": 35857
    },
    {
      "epoch": 1.645541737414529,
      "grad_norm": 0.44653865694999695,
      "learning_rate": 4.384306134971599e-06,
      "loss": 0.3448,
      "step": 35858
    },
    {
      "epoch": 1.6455876279197834,
      "grad_norm": 0.47348302602767944,
      "learning_rate": 4.384062813816972e-06,
      "loss": 0.3685,
      "step": 35859
    },
    {
      "epoch": 1.645633518425038,
      "grad_norm": 0.4350849390029907,
      "learning_rate": 4.3838194941434815e-06,
      "loss": 0.3003,
      "step": 35860
    },
    {
      "epoch": 1.6456794089302922,
      "grad_norm": 0.5129213929176331,
      "learning_rate": 4.38357617595171e-06,
      "loss": 0.3823,
      "step": 35861
    },
    {
      "epoch": 1.6457252994355467,
      "grad_norm": 0.47791916131973267,
      "learning_rate": 4.383332859242241e-06,
      "loss": 0.3466,
      "step": 35862
    },
    {
      "epoch": 1.6457711899408012,
      "grad_norm": 0.48637765645980835,
      "learning_rate": 4.383089544015664e-06,
      "loss": 0.4034,
      "step": 35863
    },
    {
      "epoch": 1.6458170804460557,
      "grad_norm": 0.512289822101593,
      "learning_rate": 4.382846230272562e-06,
      "loss": 0.3976,
      "step": 35864
    },
    {
      "epoch": 1.6458629709513102,
      "grad_norm": 0.44551882147789,
      "learning_rate": 4.382602918013517e-06,
      "loss": 0.302,
      "step": 35865
    },
    {
      "epoch": 1.6459088614565647,
      "grad_norm": 0.5206678509712219,
      "learning_rate": 4.38235960723912e-06,
      "loss": 0.4688,
      "step": 35866
    },
    {
      "epoch": 1.6459547519618192,
      "grad_norm": 0.4506447911262512,
      "learning_rate": 4.382116297949954e-06,
      "loss": 0.3204,
      "step": 35867
    },
    {
      "epoch": 1.6460006424670737,
      "grad_norm": 0.4454910457134247,
      "learning_rate": 4.3818729901466e-06,
      "loss": 0.3236,
      "step": 35868
    },
    {
      "epoch": 1.6460465329723282,
      "grad_norm": 0.4584294259548187,
      "learning_rate": 4.381629683829649e-06,
      "loss": 0.3385,
      "step": 35869
    },
    {
      "epoch": 1.6460924234775824,
      "grad_norm": 0.4795841872692108,
      "learning_rate": 4.381386378999683e-06,
      "loss": 0.4148,
      "step": 35870
    },
    {
      "epoch": 1.646138313982837,
      "grad_norm": 0.4646975100040436,
      "learning_rate": 4.381143075657287e-06,
      "loss": 0.3615,
      "step": 35871
    },
    {
      "epoch": 1.6461842044880914,
      "grad_norm": 0.4283987581729889,
      "learning_rate": 4.380899773803048e-06,
      "loss": 0.3121,
      "step": 35872
    },
    {
      "epoch": 1.6462300949933457,
      "grad_norm": 0.4476897120475769,
      "learning_rate": 4.380656473437549e-06,
      "loss": 0.3546,
      "step": 35873
    },
    {
      "epoch": 1.6462759854986002,
      "grad_norm": 0.4907616674900055,
      "learning_rate": 4.380413174561373e-06,
      "loss": 0.3613,
      "step": 35874
    },
    {
      "epoch": 1.6463218760038547,
      "grad_norm": 0.4547068774700165,
      "learning_rate": 4.38016987717511e-06,
      "loss": 0.3313,
      "step": 35875
    },
    {
      "epoch": 1.6463677665091092,
      "grad_norm": 0.444036066532135,
      "learning_rate": 4.379926581279344e-06,
      "loss": 0.2901,
      "step": 35876
    },
    {
      "epoch": 1.6464136570143637,
      "grad_norm": 0.5038992166519165,
      "learning_rate": 4.379683286874657e-06,
      "loss": 0.4248,
      "step": 35877
    },
    {
      "epoch": 1.6464595475196182,
      "grad_norm": 0.47008755803108215,
      "learning_rate": 4.379439993961637e-06,
      "loss": 0.3634,
      "step": 35878
    },
    {
      "epoch": 1.6465054380248727,
      "grad_norm": 0.4997967481613159,
      "learning_rate": 4.379196702540867e-06,
      "loss": 0.4123,
      "step": 35879
    },
    {
      "epoch": 1.6465513285301272,
      "grad_norm": 0.42728912830352783,
      "learning_rate": 4.378953412612933e-06,
      "loss": 0.2981,
      "step": 35880
    },
    {
      "epoch": 1.6465972190353817,
      "grad_norm": 0.4909107983112335,
      "learning_rate": 4.378710124178421e-06,
      "loss": 0.3482,
      "step": 35881
    },
    {
      "epoch": 1.6466431095406362,
      "grad_norm": 0.4335554242134094,
      "learning_rate": 4.378466837237915e-06,
      "loss": 0.3034,
      "step": 35882
    },
    {
      "epoch": 1.6466890000458905,
      "grad_norm": 0.5149171948432922,
      "learning_rate": 4.378223551791999e-06,
      "loss": 0.409,
      "step": 35883
    },
    {
      "epoch": 1.646734890551145,
      "grad_norm": 0.46609222888946533,
      "learning_rate": 4.37798026784126e-06,
      "loss": 0.3466,
      "step": 35884
    },
    {
      "epoch": 1.6467807810563995,
      "grad_norm": 0.47231432795524597,
      "learning_rate": 4.377736985386283e-06,
      "loss": 0.3509,
      "step": 35885
    },
    {
      "epoch": 1.6468266715616537,
      "grad_norm": 0.47837042808532715,
      "learning_rate": 4.377493704427649e-06,
      "loss": 0.3681,
      "step": 35886
    },
    {
      "epoch": 1.6468725620669082,
      "grad_norm": 0.4613659977912903,
      "learning_rate": 4.377250424965949e-06,
      "loss": 0.3349,
      "step": 35887
    },
    {
      "epoch": 1.6469184525721627,
      "grad_norm": 0.4442126154899597,
      "learning_rate": 4.377007147001765e-06,
      "loss": 0.3255,
      "step": 35888
    },
    {
      "epoch": 1.6469643430774172,
      "grad_norm": 0.5007584691047668,
      "learning_rate": 4.37676387053568e-06,
      "loss": 0.4189,
      "step": 35889
    },
    {
      "epoch": 1.6470102335826717,
      "grad_norm": 0.507055938243866,
      "learning_rate": 4.376520595568284e-06,
      "loss": 0.3774,
      "step": 35890
    },
    {
      "epoch": 1.6470561240879262,
      "grad_norm": 0.4995647072792053,
      "learning_rate": 4.3762773221001596e-06,
      "loss": 0.3867,
      "step": 35891
    },
    {
      "epoch": 1.6471020145931807,
      "grad_norm": 0.46633684635162354,
      "learning_rate": 4.3760340501318896e-06,
      "loss": 0.3419,
      "step": 35892
    },
    {
      "epoch": 1.6471479050984352,
      "grad_norm": 0.45598554611206055,
      "learning_rate": 4.3757907796640605e-06,
      "loss": 0.3724,
      "step": 35893
    },
    {
      "epoch": 1.6471937956036897,
      "grad_norm": 0.5060083866119385,
      "learning_rate": 4.3755475106972585e-06,
      "loss": 0.4215,
      "step": 35894
    },
    {
      "epoch": 1.6472396861089442,
      "grad_norm": 0.490540474653244,
      "learning_rate": 4.375304243232068e-06,
      "loss": 0.3369,
      "step": 35895
    },
    {
      "epoch": 1.6472855766141985,
      "grad_norm": 0.4229942560195923,
      "learning_rate": 4.375060977269071e-06,
      "loss": 0.2657,
      "step": 35896
    },
    {
      "epoch": 1.647331467119453,
      "grad_norm": 0.5268293619155884,
      "learning_rate": 4.374817712808858e-06,
      "loss": 0.4335,
      "step": 35897
    },
    {
      "epoch": 1.6473773576247075,
      "grad_norm": 0.5111556053161621,
      "learning_rate": 4.3745744498520105e-06,
      "loss": 0.4356,
      "step": 35898
    },
    {
      "epoch": 1.6474232481299618,
      "grad_norm": 0.5039180517196655,
      "learning_rate": 4.3743311883991124e-06,
      "loss": 0.4458,
      "step": 35899
    },
    {
      "epoch": 1.6474691386352163,
      "grad_norm": 0.4656873643398285,
      "learning_rate": 4.374087928450752e-06,
      "loss": 0.3786,
      "step": 35900
    },
    {
      "epoch": 1.6475150291404708,
      "grad_norm": 0.47338080406188965,
      "learning_rate": 4.373844670007513e-06,
      "loss": 0.3552,
      "step": 35901
    },
    {
      "epoch": 1.6475609196457253,
      "grad_norm": 0.4701687693595886,
      "learning_rate": 4.3736014130699775e-06,
      "loss": 0.3615,
      "step": 35902
    },
    {
      "epoch": 1.6476068101509798,
      "grad_norm": 0.4791070818901062,
      "learning_rate": 4.373358157638735e-06,
      "loss": 0.3625,
      "step": 35903
    },
    {
      "epoch": 1.6476527006562343,
      "grad_norm": 0.4878644645214081,
      "learning_rate": 4.373114903714368e-06,
      "loss": 0.3469,
      "step": 35904
    },
    {
      "epoch": 1.6476985911614888,
      "grad_norm": 0.49195224046707153,
      "learning_rate": 4.372871651297461e-06,
      "loss": 0.3448,
      "step": 35905
    },
    {
      "epoch": 1.6477444816667433,
      "grad_norm": 0.5160082578659058,
      "learning_rate": 4.372628400388601e-06,
      "loss": 0.406,
      "step": 35906
    },
    {
      "epoch": 1.6477903721719978,
      "grad_norm": 0.4769684076309204,
      "learning_rate": 4.372385150988372e-06,
      "loss": 0.3449,
      "step": 35907
    },
    {
      "epoch": 1.647836262677252,
      "grad_norm": 0.4544503092765808,
      "learning_rate": 4.372141903097356e-06,
      "loss": 0.3145,
      "step": 35908
    },
    {
      "epoch": 1.6478821531825065,
      "grad_norm": 0.4470287263393402,
      "learning_rate": 4.3718986567161424e-06,
      "loss": 0.3144,
      "step": 35909
    },
    {
      "epoch": 1.647928043687761,
      "grad_norm": 0.46876898407936096,
      "learning_rate": 4.371655411845315e-06,
      "loss": 0.3279,
      "step": 35910
    },
    {
      "epoch": 1.6479739341930153,
      "grad_norm": 0.4714542627334595,
      "learning_rate": 4.371412168485457e-06,
      "loss": 0.3475,
      "step": 35911
    },
    {
      "epoch": 1.6480198246982698,
      "grad_norm": 0.5600622296333313,
      "learning_rate": 4.371168926637155e-06,
      "loss": 0.4228,
      "step": 35912
    },
    {
      "epoch": 1.6480657152035243,
      "grad_norm": 0.4761614203453064,
      "learning_rate": 4.370925686300993e-06,
      "loss": 0.3253,
      "step": 35913
    },
    {
      "epoch": 1.6481116057087788,
      "grad_norm": 0.4816685616970062,
      "learning_rate": 4.370682447477555e-06,
      "loss": 0.3794,
      "step": 35914
    },
    {
      "epoch": 1.6481574962140333,
      "grad_norm": 0.44472235441207886,
      "learning_rate": 4.370439210167429e-06,
      "loss": 0.3187,
      "step": 35915
    },
    {
      "epoch": 1.6482033867192878,
      "grad_norm": 0.41309282183647156,
      "learning_rate": 4.370195974371198e-06,
      "loss": 0.2922,
      "step": 35916
    },
    {
      "epoch": 1.6482492772245423,
      "grad_norm": 0.4674730598926544,
      "learning_rate": 4.369952740089444e-06,
      "loss": 0.326,
      "step": 35917
    },
    {
      "epoch": 1.6482951677297968,
      "grad_norm": 0.46466493606567383,
      "learning_rate": 4.369709507322757e-06,
      "loss": 0.3589,
      "step": 35918
    },
    {
      "epoch": 1.6483410582350513,
      "grad_norm": 0.48003995418548584,
      "learning_rate": 4.36946627607172e-06,
      "loss": 0.3584,
      "step": 35919
    },
    {
      "epoch": 1.6483869487403058,
      "grad_norm": 0.471615195274353,
      "learning_rate": 4.369223046336917e-06,
      "loss": 0.3971,
      "step": 35920
    },
    {
      "epoch": 1.64843283924556,
      "grad_norm": 0.4379487931728363,
      "learning_rate": 4.368979818118935e-06,
      "loss": 0.3028,
      "step": 35921
    },
    {
      "epoch": 1.6484787297508146,
      "grad_norm": 0.5069451332092285,
      "learning_rate": 4.368736591418356e-06,
      "loss": 0.3168,
      "step": 35922
    },
    {
      "epoch": 1.648524620256069,
      "grad_norm": 0.4645574986934662,
      "learning_rate": 4.3684933662357645e-06,
      "loss": 0.3733,
      "step": 35923
    },
    {
      "epoch": 1.6485705107613233,
      "grad_norm": 0.4728159010410309,
      "learning_rate": 4.36825014257175e-06,
      "loss": 0.3195,
      "step": 35924
    },
    {
      "epoch": 1.6486164012665778,
      "grad_norm": 0.505206823348999,
      "learning_rate": 4.368006920426894e-06,
      "loss": 0.424,
      "step": 35925
    },
    {
      "epoch": 1.6486622917718323,
      "grad_norm": 0.47774896025657654,
      "learning_rate": 4.3677636998017815e-06,
      "loss": 0.3855,
      "step": 35926
    },
    {
      "epoch": 1.6487081822770868,
      "grad_norm": 0.48680517077445984,
      "learning_rate": 4.367520480696997e-06,
      "loss": 0.3614,
      "step": 35927
    },
    {
      "epoch": 1.6487540727823413,
      "grad_norm": 0.4690234363079071,
      "learning_rate": 4.367277263113127e-06,
      "loss": 0.3414,
      "step": 35928
    },
    {
      "epoch": 1.6487999632875958,
      "grad_norm": 0.4303583800792694,
      "learning_rate": 4.367034047050755e-06,
      "loss": 0.3021,
      "step": 35929
    },
    {
      "epoch": 1.6488458537928503,
      "grad_norm": 0.4407225549221039,
      "learning_rate": 4.366790832510466e-06,
      "loss": 0.3068,
      "step": 35930
    },
    {
      "epoch": 1.6488917442981048,
      "grad_norm": 0.4914962351322174,
      "learning_rate": 4.366547619492846e-06,
      "loss": 0.3737,
      "step": 35931
    },
    {
      "epoch": 1.6489376348033593,
      "grad_norm": 0.4584190845489502,
      "learning_rate": 4.3663044079984795e-06,
      "loss": 0.3229,
      "step": 35932
    },
    {
      "epoch": 1.6489835253086138,
      "grad_norm": 0.48588070273399353,
      "learning_rate": 4.366061198027949e-06,
      "loss": 0.3676,
      "step": 35933
    },
    {
      "epoch": 1.649029415813868,
      "grad_norm": 0.43435367941856384,
      "learning_rate": 4.3658179895818436e-06,
      "loss": 0.2826,
      "step": 35934
    },
    {
      "epoch": 1.6490753063191226,
      "grad_norm": 0.49441584944725037,
      "learning_rate": 4.365574782660745e-06,
      "loss": 0.3399,
      "step": 35935
    },
    {
      "epoch": 1.649121196824377,
      "grad_norm": 0.4394732415676117,
      "learning_rate": 4.365331577265238e-06,
      "loss": 0.3213,
      "step": 35936
    },
    {
      "epoch": 1.6491670873296314,
      "grad_norm": 0.45605823397636414,
      "learning_rate": 4.365088373395909e-06,
      "loss": 0.3194,
      "step": 35937
    },
    {
      "epoch": 1.6492129778348859,
      "grad_norm": 0.49027353525161743,
      "learning_rate": 4.3648451710533425e-06,
      "loss": 0.3707,
      "step": 35938
    },
    {
      "epoch": 1.6492588683401403,
      "grad_norm": 0.4711936414241791,
      "learning_rate": 4.36460197023812e-06,
      "loss": 0.3239,
      "step": 35939
    },
    {
      "epoch": 1.6493047588453948,
      "grad_norm": 0.49379420280456543,
      "learning_rate": 4.364358770950832e-06,
      "loss": 0.3686,
      "step": 35940
    },
    {
      "epoch": 1.6493506493506493,
      "grad_norm": 0.483435720205307,
      "learning_rate": 4.364115573192061e-06,
      "loss": 0.4135,
      "step": 35941
    },
    {
      "epoch": 1.6493965398559038,
      "grad_norm": 0.4706214368343353,
      "learning_rate": 4.36387237696239e-06,
      "loss": 0.3312,
      "step": 35942
    },
    {
      "epoch": 1.6494424303611583,
      "grad_norm": 0.4465349018573761,
      "learning_rate": 4.363629182262405e-06,
      "loss": 0.2737,
      "step": 35943
    },
    {
      "epoch": 1.6494883208664128,
      "grad_norm": 0.44056040048599243,
      "learning_rate": 4.363385989092693e-06,
      "loss": 0.2673,
      "step": 35944
    },
    {
      "epoch": 1.6495342113716673,
      "grad_norm": 0.4617215692996979,
      "learning_rate": 4.363142797453833e-06,
      "loss": 0.3521,
      "step": 35945
    },
    {
      "epoch": 1.6495801018769216,
      "grad_norm": 0.47313228249549866,
      "learning_rate": 4.362899607346416e-06,
      "loss": 0.3588,
      "step": 35946
    },
    {
      "epoch": 1.649625992382176,
      "grad_norm": 0.47361859679222107,
      "learning_rate": 4.362656418771025e-06,
      "loss": 0.3339,
      "step": 35947
    },
    {
      "epoch": 1.6496718828874306,
      "grad_norm": 0.4966105818748474,
      "learning_rate": 4.362413231728243e-06,
      "loss": 0.4392,
      "step": 35948
    },
    {
      "epoch": 1.649717773392685,
      "grad_norm": 0.46930089592933655,
      "learning_rate": 4.362170046218657e-06,
      "loss": 0.3753,
      "step": 35949
    },
    {
      "epoch": 1.6497636638979394,
      "grad_norm": 0.48905280232429504,
      "learning_rate": 4.36192686224285e-06,
      "loss": 0.4378,
      "step": 35950
    },
    {
      "epoch": 1.6498095544031939,
      "grad_norm": 0.43733733892440796,
      "learning_rate": 4.3616836798014065e-06,
      "loss": 0.324,
      "step": 35951
    },
    {
      "epoch": 1.6498554449084484,
      "grad_norm": 0.4886120855808258,
      "learning_rate": 4.361440498894914e-06,
      "loss": 0.3397,
      "step": 35952
    },
    {
      "epoch": 1.6499013354137029,
      "grad_norm": 0.4351097047328949,
      "learning_rate": 4.361197319523955e-06,
      "loss": 0.3163,
      "step": 35953
    },
    {
      "epoch": 1.6499472259189574,
      "grad_norm": 0.4852837026119232,
      "learning_rate": 4.3609541416891145e-06,
      "loss": 0.4078,
      "step": 35954
    },
    {
      "epoch": 1.6499931164242119,
      "grad_norm": 0.46513986587524414,
      "learning_rate": 4.360710965390978e-06,
      "loss": 0.3245,
      "step": 35955
    },
    {
      "epoch": 1.6500390069294664,
      "grad_norm": 0.507186770439148,
      "learning_rate": 4.36046779063013e-06,
      "loss": 0.432,
      "step": 35956
    },
    {
      "epoch": 1.6500848974347209,
      "grad_norm": 0.5067645311355591,
      "learning_rate": 4.360224617407155e-06,
      "loss": 0.3308,
      "step": 35957
    },
    {
      "epoch": 1.6501307879399754,
      "grad_norm": 0.4566802382469177,
      "learning_rate": 4.359981445722638e-06,
      "loss": 0.3515,
      "step": 35958
    },
    {
      "epoch": 1.6501766784452296,
      "grad_norm": 0.46597445011138916,
      "learning_rate": 4.359738275577163e-06,
      "loss": 0.3704,
      "step": 35959
    },
    {
      "epoch": 1.6502225689504841,
      "grad_norm": 0.5354997515678406,
      "learning_rate": 4.359495106971317e-06,
      "loss": 0.3049,
      "step": 35960
    },
    {
      "epoch": 1.6502684594557386,
      "grad_norm": 0.5137149691581726,
      "learning_rate": 4.35925193990568e-06,
      "loss": 0.4244,
      "step": 35961
    },
    {
      "epoch": 1.650314349960993,
      "grad_norm": 0.45069023966789246,
      "learning_rate": 4.359008774380842e-06,
      "loss": 0.328,
      "step": 35962
    },
    {
      "epoch": 1.6503602404662474,
      "grad_norm": 0.4509267508983612,
      "learning_rate": 4.358765610397386e-06,
      "loss": 0.3241,
      "step": 35963
    },
    {
      "epoch": 1.650406130971502,
      "grad_norm": 0.4760028123855591,
      "learning_rate": 4.358522447955895e-06,
      "loss": 0.3872,
      "step": 35964
    },
    {
      "epoch": 1.6504520214767564,
      "grad_norm": 0.4613112509250641,
      "learning_rate": 4.3582792870569555e-06,
      "loss": 0.3752,
      "step": 35965
    },
    {
      "epoch": 1.650497911982011,
      "grad_norm": 0.4791675806045532,
      "learning_rate": 4.358036127701153e-06,
      "loss": 0.3846,
      "step": 35966
    },
    {
      "epoch": 1.6505438024872654,
      "grad_norm": 0.5006075501441956,
      "learning_rate": 4.357792969889068e-06,
      "loss": 0.3942,
      "step": 35967
    },
    {
      "epoch": 1.65058969299252,
      "grad_norm": 0.48373717069625854,
      "learning_rate": 4.357549813621291e-06,
      "loss": 0.3792,
      "step": 35968
    },
    {
      "epoch": 1.6506355834977744,
      "grad_norm": 0.4603436589241028,
      "learning_rate": 4.357306658898404e-06,
      "loss": 0.3444,
      "step": 35969
    },
    {
      "epoch": 1.650681474003029,
      "grad_norm": 0.5049374103546143,
      "learning_rate": 4.35706350572099e-06,
      "loss": 0.407,
      "step": 35970
    },
    {
      "epoch": 1.6507273645082834,
      "grad_norm": 0.4457077383995056,
      "learning_rate": 4.356820354089636e-06,
      "loss": 0.3016,
      "step": 35971
    },
    {
      "epoch": 1.6507732550135377,
      "grad_norm": 0.47533726692199707,
      "learning_rate": 4.356577204004927e-06,
      "loss": 0.3865,
      "step": 35972
    },
    {
      "epoch": 1.6508191455187922,
      "grad_norm": 0.48755377531051636,
      "learning_rate": 4.356334055467444e-06,
      "loss": 0.4039,
      "step": 35973
    },
    {
      "epoch": 1.6508650360240467,
      "grad_norm": 0.458965539932251,
      "learning_rate": 4.356090908477777e-06,
      "loss": 0.3505,
      "step": 35974
    },
    {
      "epoch": 1.650910926529301,
      "grad_norm": 0.41512060165405273,
      "learning_rate": 4.355847763036508e-06,
      "loss": 0.2691,
      "step": 35975
    },
    {
      "epoch": 1.6509568170345554,
      "grad_norm": 0.4832877218723297,
      "learning_rate": 4.35560461914422e-06,
      "loss": 0.3993,
      "step": 35976
    },
    {
      "epoch": 1.65100270753981,
      "grad_norm": 0.4928421974182129,
      "learning_rate": 4.355361476801502e-06,
      "loss": 0.4171,
      "step": 35977
    },
    {
      "epoch": 1.6510485980450644,
      "grad_norm": 0.44953614473342896,
      "learning_rate": 4.355118336008936e-06,
      "loss": 0.311,
      "step": 35978
    },
    {
      "epoch": 1.651094488550319,
      "grad_norm": 0.5108081102371216,
      "learning_rate": 4.354875196767104e-06,
      "loss": 0.4951,
      "step": 35979
    },
    {
      "epoch": 1.6511403790555734,
      "grad_norm": 0.44888004660606384,
      "learning_rate": 4.3546320590765965e-06,
      "loss": 0.3418,
      "step": 35980
    },
    {
      "epoch": 1.651186269560828,
      "grad_norm": 0.48100563883781433,
      "learning_rate": 4.3543889229379946e-06,
      "loss": 0.383,
      "step": 35981
    },
    {
      "epoch": 1.6512321600660824,
      "grad_norm": 0.4599229693412781,
      "learning_rate": 4.354145788351882e-06,
      "loss": 0.3037,
      "step": 35982
    },
    {
      "epoch": 1.651278050571337,
      "grad_norm": 0.4645184576511383,
      "learning_rate": 4.353902655318846e-06,
      "loss": 0.3327,
      "step": 35983
    },
    {
      "epoch": 1.6513239410765914,
      "grad_norm": 0.4297705292701721,
      "learning_rate": 4.353659523839471e-06,
      "loss": 0.2854,
      "step": 35984
    },
    {
      "epoch": 1.6513698315818457,
      "grad_norm": 0.4643173813819885,
      "learning_rate": 4.35341639391434e-06,
      "loss": 0.3433,
      "step": 35985
    },
    {
      "epoch": 1.6514157220871002,
      "grad_norm": 0.4644688069820404,
      "learning_rate": 4.353173265544039e-06,
      "loss": 0.3648,
      "step": 35986
    },
    {
      "epoch": 1.6514616125923547,
      "grad_norm": 0.4398758113384247,
      "learning_rate": 4.352930138729152e-06,
      "loss": 0.3092,
      "step": 35987
    },
    {
      "epoch": 1.651507503097609,
      "grad_norm": 0.4798223078250885,
      "learning_rate": 4.352687013470262e-06,
      "loss": 0.3498,
      "step": 35988
    },
    {
      "epoch": 1.6515533936028635,
      "grad_norm": 0.45280513167381287,
      "learning_rate": 4.352443889767958e-06,
      "loss": 0.3379,
      "step": 35989
    },
    {
      "epoch": 1.651599284108118,
      "grad_norm": 0.47199442982673645,
      "learning_rate": 4.352200767622821e-06,
      "loss": 0.3739,
      "step": 35990
    },
    {
      "epoch": 1.6516451746133725,
      "grad_norm": 0.4413686990737915,
      "learning_rate": 4.3519576470354375e-06,
      "loss": 0.3361,
      "step": 35991
    },
    {
      "epoch": 1.651691065118627,
      "grad_norm": 0.480246365070343,
      "learning_rate": 4.3517145280063895e-06,
      "loss": 0.35,
      "step": 35992
    },
    {
      "epoch": 1.6517369556238815,
      "grad_norm": 0.4901658296585083,
      "learning_rate": 4.351471410536265e-06,
      "loss": 0.447,
      "step": 35993
    },
    {
      "epoch": 1.651782846129136,
      "grad_norm": 0.483841210603714,
      "learning_rate": 4.351228294625648e-06,
      "loss": 0.3705,
      "step": 35994
    },
    {
      "epoch": 1.6518287366343904,
      "grad_norm": 0.4929962158203125,
      "learning_rate": 4.350985180275119e-06,
      "loss": 0.4044,
      "step": 35995
    },
    {
      "epoch": 1.651874627139645,
      "grad_norm": 0.4916427433490753,
      "learning_rate": 4.350742067485268e-06,
      "loss": 0.3808,
      "step": 35996
    },
    {
      "epoch": 1.6519205176448992,
      "grad_norm": 0.4855644106864929,
      "learning_rate": 4.350498956256678e-06,
      "loss": 0.3451,
      "step": 35997
    },
    {
      "epoch": 1.6519664081501537,
      "grad_norm": 0.4811583459377289,
      "learning_rate": 4.350255846589931e-06,
      "loss": 0.401,
      "step": 35998
    },
    {
      "epoch": 1.6520122986554082,
      "grad_norm": 0.4791715145111084,
      "learning_rate": 4.350012738485616e-06,
      "loss": 0.3607,
      "step": 35999
    },
    {
      "epoch": 1.6520581891606625,
      "grad_norm": 0.4449109733104706,
      "learning_rate": 4.349769631944314e-06,
      "loss": 0.3439,
      "step": 36000
    },
    {
      "epoch": 1.652104079665917,
      "grad_norm": 0.4651741683483124,
      "learning_rate": 4.349526526966609e-06,
      "loss": 0.3127,
      "step": 36001
    },
    {
      "epoch": 1.6521499701711715,
      "grad_norm": 0.5020276308059692,
      "learning_rate": 4.349283423553091e-06,
      "loss": 0.3915,
      "step": 36002
    },
    {
      "epoch": 1.652195860676426,
      "grad_norm": 0.4172147810459137,
      "learning_rate": 4.3490403217043395e-06,
      "loss": 0.2881,
      "step": 36003
    },
    {
      "epoch": 1.6522417511816805,
      "grad_norm": 0.4468369781970978,
      "learning_rate": 4.348797221420939e-06,
      "loss": 0.3293,
      "step": 36004
    },
    {
      "epoch": 1.652287641686935,
      "grad_norm": 0.4693075120449066,
      "learning_rate": 4.3485541227034765e-06,
      "loss": 0.3699,
      "step": 36005
    },
    {
      "epoch": 1.6523335321921895,
      "grad_norm": 0.4877202808856964,
      "learning_rate": 4.3483110255525365e-06,
      "loss": 0.3863,
      "step": 36006
    },
    {
      "epoch": 1.652379422697444,
      "grad_norm": 0.46974435448646545,
      "learning_rate": 4.348067929968702e-06,
      "loss": 0.3358,
      "step": 36007
    },
    {
      "epoch": 1.6524253132026985,
      "grad_norm": 0.46888473629951477,
      "learning_rate": 4.34782483595256e-06,
      "loss": 0.3764,
      "step": 36008
    },
    {
      "epoch": 1.652471203707953,
      "grad_norm": 0.49411866068840027,
      "learning_rate": 4.347581743504693e-06,
      "loss": 0.412,
      "step": 36009
    },
    {
      "epoch": 1.6525170942132072,
      "grad_norm": 0.45911452174186707,
      "learning_rate": 4.347338652625683e-06,
      "loss": 0.3089,
      "step": 36010
    },
    {
      "epoch": 1.6525629847184617,
      "grad_norm": 0.5009966492652893,
      "learning_rate": 4.347095563316121e-06,
      "loss": 0.4446,
      "step": 36011
    },
    {
      "epoch": 1.6526088752237162,
      "grad_norm": 0.45252281427383423,
      "learning_rate": 4.346852475576587e-06,
      "loss": 0.3268,
      "step": 36012
    },
    {
      "epoch": 1.6526547657289705,
      "grad_norm": 0.4503854811191559,
      "learning_rate": 4.346609389407665e-06,
      "loss": 0.3382,
      "step": 36013
    },
    {
      "epoch": 1.652700656234225,
      "grad_norm": 0.4583148658275604,
      "learning_rate": 4.346366304809943e-06,
      "loss": 0.3738,
      "step": 36014
    },
    {
      "epoch": 1.6527465467394795,
      "grad_norm": 0.5285967588424683,
      "learning_rate": 4.346123221784005e-06,
      "loss": 0.3902,
      "step": 36015
    },
    {
      "epoch": 1.652792437244734,
      "grad_norm": 0.5204439759254456,
      "learning_rate": 4.345880140330431e-06,
      "loss": 0.4404,
      "step": 36016
    },
    {
      "epoch": 1.6528383277499885,
      "grad_norm": 0.4820834994316101,
      "learning_rate": 4.345637060449811e-06,
      "loss": 0.3961,
      "step": 36017
    },
    {
      "epoch": 1.652884218255243,
      "grad_norm": 0.5019481182098389,
      "learning_rate": 4.345393982142728e-06,
      "loss": 0.4407,
      "step": 36018
    },
    {
      "epoch": 1.6529301087604975,
      "grad_norm": 0.4796825349330902,
      "learning_rate": 4.345150905409764e-06,
      "loss": 0.3526,
      "step": 36019
    },
    {
      "epoch": 1.652975999265752,
      "grad_norm": 0.4567500948905945,
      "learning_rate": 4.3449078302515065e-06,
      "loss": 0.3025,
      "step": 36020
    },
    {
      "epoch": 1.6530218897710065,
      "grad_norm": 0.44055721163749695,
      "learning_rate": 4.344664756668539e-06,
      "loss": 0.3273,
      "step": 36021
    },
    {
      "epoch": 1.653067780276261,
      "grad_norm": 0.4587423503398895,
      "learning_rate": 4.344421684661444e-06,
      "loss": 0.3692,
      "step": 36022
    },
    {
      "epoch": 1.6531136707815153,
      "grad_norm": 0.4305415749549866,
      "learning_rate": 4.34417861423081e-06,
      "loss": 0.3093,
      "step": 36023
    },
    {
      "epoch": 1.6531595612867698,
      "grad_norm": 0.4753001034259796,
      "learning_rate": 4.34393554537722e-06,
      "loss": 0.3536,
      "step": 36024
    },
    {
      "epoch": 1.6532054517920243,
      "grad_norm": 0.4818207323551178,
      "learning_rate": 4.343692478101258e-06,
      "loss": 0.415,
      "step": 36025
    },
    {
      "epoch": 1.6532513422972785,
      "grad_norm": 0.4840905964374542,
      "learning_rate": 4.343449412403506e-06,
      "loss": 0.3752,
      "step": 36026
    },
    {
      "epoch": 1.653297232802533,
      "grad_norm": 0.4998270571231842,
      "learning_rate": 4.343206348284553e-06,
      "loss": 0.4485,
      "step": 36027
    },
    {
      "epoch": 1.6533431233077875,
      "grad_norm": 0.4937484860420227,
      "learning_rate": 4.342963285744981e-06,
      "loss": 0.3527,
      "step": 36028
    },
    {
      "epoch": 1.653389013813042,
      "grad_norm": 0.48509126901626587,
      "learning_rate": 4.342720224785374e-06,
      "loss": 0.3522,
      "step": 36029
    },
    {
      "epoch": 1.6534349043182965,
      "grad_norm": 0.5056251287460327,
      "learning_rate": 4.342477165406319e-06,
      "loss": 0.4346,
      "step": 36030
    },
    {
      "epoch": 1.653480794823551,
      "grad_norm": 0.4848411977291107,
      "learning_rate": 4.342234107608398e-06,
      "loss": 0.3561,
      "step": 36031
    },
    {
      "epoch": 1.6535266853288055,
      "grad_norm": 0.47766000032424927,
      "learning_rate": 4.341991051392195e-06,
      "loss": 0.3973,
      "step": 36032
    },
    {
      "epoch": 1.65357257583406,
      "grad_norm": 0.44374167919158936,
      "learning_rate": 4.341747996758297e-06,
      "loss": 0.3393,
      "step": 36033
    },
    {
      "epoch": 1.6536184663393145,
      "grad_norm": 0.4642241895198822,
      "learning_rate": 4.341504943707288e-06,
      "loss": 0.3559,
      "step": 36034
    },
    {
      "epoch": 1.6536643568445688,
      "grad_norm": 0.47755149006843567,
      "learning_rate": 4.341261892239751e-06,
      "loss": 0.3804,
      "step": 36035
    },
    {
      "epoch": 1.6537102473498233,
      "grad_norm": 0.48947200179100037,
      "learning_rate": 4.341018842356272e-06,
      "loss": 0.36,
      "step": 36036
    },
    {
      "epoch": 1.6537561378550778,
      "grad_norm": 0.4508672058582306,
      "learning_rate": 4.340775794057435e-06,
      "loss": 0.3083,
      "step": 36037
    },
    {
      "epoch": 1.6538020283603323,
      "grad_norm": 0.4680686593055725,
      "learning_rate": 4.340532747343822e-06,
      "loss": 0.3894,
      "step": 36038
    },
    {
      "epoch": 1.6538479188655866,
      "grad_norm": 0.45429444313049316,
      "learning_rate": 4.340289702216021e-06,
      "loss": 0.3402,
      "step": 36039
    },
    {
      "epoch": 1.653893809370841,
      "grad_norm": 0.46961474418640137,
      "learning_rate": 4.340046658674616e-06,
      "loss": 0.3906,
      "step": 36040
    },
    {
      "epoch": 1.6539396998760956,
      "grad_norm": 0.49488213658332825,
      "learning_rate": 4.339803616720189e-06,
      "loss": 0.4053,
      "step": 36041
    },
    {
      "epoch": 1.65398559038135,
      "grad_norm": 0.4506267309188843,
      "learning_rate": 4.339560576353327e-06,
      "loss": 0.3358,
      "step": 36042
    },
    {
      "epoch": 1.6540314808866046,
      "grad_norm": 0.49704161286354065,
      "learning_rate": 4.339317537574613e-06,
      "loss": 0.4338,
      "step": 36043
    },
    {
      "epoch": 1.654077371391859,
      "grad_norm": 0.4659401476383209,
      "learning_rate": 4.33907450038463e-06,
      "loss": 0.3571,
      "step": 36044
    },
    {
      "epoch": 1.6541232618971136,
      "grad_norm": 0.4463460445404053,
      "learning_rate": 4.338831464783967e-06,
      "loss": 0.3099,
      "step": 36045
    },
    {
      "epoch": 1.654169152402368,
      "grad_norm": 0.49807626008987427,
      "learning_rate": 4.338588430773206e-06,
      "loss": 0.4087,
      "step": 36046
    },
    {
      "epoch": 1.6542150429076226,
      "grad_norm": 0.4588109254837036,
      "learning_rate": 4.3383453983529275e-06,
      "loss": 0.3338,
      "step": 36047
    },
    {
      "epoch": 1.6542609334128768,
      "grad_norm": 0.49700185656547546,
      "learning_rate": 4.338102367523723e-06,
      "loss": 0.377,
      "step": 36048
    },
    {
      "epoch": 1.6543068239181313,
      "grad_norm": 0.4630831182003021,
      "learning_rate": 4.337859338286172e-06,
      "loss": 0.3253,
      "step": 36049
    },
    {
      "epoch": 1.6543527144233858,
      "grad_norm": 0.4813178777694702,
      "learning_rate": 4.33761631064086e-06,
      "loss": 0.3906,
      "step": 36050
    },
    {
      "epoch": 1.65439860492864,
      "grad_norm": 0.4512578248977661,
      "learning_rate": 4.337373284588372e-06,
      "loss": 0.3298,
      "step": 36051
    },
    {
      "epoch": 1.6544444954338946,
      "grad_norm": 0.47649335861206055,
      "learning_rate": 4.337130260129293e-06,
      "loss": 0.3187,
      "step": 36052
    },
    {
      "epoch": 1.654490385939149,
      "grad_norm": 0.4801841676235199,
      "learning_rate": 4.336887237264205e-06,
      "loss": 0.3263,
      "step": 36053
    },
    {
      "epoch": 1.6545362764444036,
      "grad_norm": 0.4341984689235687,
      "learning_rate": 4.336644215993695e-06,
      "loss": 0.274,
      "step": 36054
    },
    {
      "epoch": 1.654582166949658,
      "grad_norm": 0.5016672015190125,
      "learning_rate": 4.3364011963183465e-06,
      "loss": 0.4242,
      "step": 36055
    },
    {
      "epoch": 1.6546280574549126,
      "grad_norm": 0.45307016372680664,
      "learning_rate": 4.336158178238744e-06,
      "loss": 0.3075,
      "step": 36056
    },
    {
      "epoch": 1.654673947960167,
      "grad_norm": 0.40461796522140503,
      "learning_rate": 4.335915161755469e-06,
      "loss": 0.2756,
      "step": 36057
    },
    {
      "epoch": 1.6547198384654216,
      "grad_norm": 0.47295016050338745,
      "learning_rate": 4.335672146869111e-06,
      "loss": 0.3766,
      "step": 36058
    },
    {
      "epoch": 1.654765728970676,
      "grad_norm": 0.7599577903747559,
      "learning_rate": 4.335429133580252e-06,
      "loss": 0.4031,
      "step": 36059
    },
    {
      "epoch": 1.6548116194759306,
      "grad_norm": 0.5119338631629944,
      "learning_rate": 4.335186121889473e-06,
      "loss": 0.4259,
      "step": 36060
    },
    {
      "epoch": 1.6548575099811849,
      "grad_norm": 0.4601072072982788,
      "learning_rate": 4.334943111797364e-06,
      "loss": 0.3298,
      "step": 36061
    },
    {
      "epoch": 1.6549034004864394,
      "grad_norm": 0.4406202435493469,
      "learning_rate": 4.334700103304508e-06,
      "loss": 0.2869,
      "step": 36062
    },
    {
      "epoch": 1.6549492909916939,
      "grad_norm": 0.46977996826171875,
      "learning_rate": 4.334457096411486e-06,
      "loss": 0.3142,
      "step": 36063
    },
    {
      "epoch": 1.6549951814969481,
      "grad_norm": 0.46411871910095215,
      "learning_rate": 4.334214091118886e-06,
      "loss": 0.3385,
      "step": 36064
    },
    {
      "epoch": 1.6550410720022026,
      "grad_norm": 0.6432812213897705,
      "learning_rate": 4.333971087427291e-06,
      "loss": 0.477,
      "step": 36065
    },
    {
      "epoch": 1.6550869625074571,
      "grad_norm": 0.43838033080101013,
      "learning_rate": 4.333728085337283e-06,
      "loss": 0.3091,
      "step": 36066
    },
    {
      "epoch": 1.6551328530127116,
      "grad_norm": 0.4661412537097931,
      "learning_rate": 4.333485084849451e-06,
      "loss": 0.3612,
      "step": 36067
    },
    {
      "epoch": 1.6551787435179661,
      "grad_norm": 0.4695759117603302,
      "learning_rate": 4.333242085964378e-06,
      "loss": 0.3653,
      "step": 36068
    },
    {
      "epoch": 1.6552246340232206,
      "grad_norm": 0.4891836941242218,
      "learning_rate": 4.332999088682644e-06,
      "loss": 0.3873,
      "step": 36069
    },
    {
      "epoch": 1.6552705245284751,
      "grad_norm": 0.4200678765773773,
      "learning_rate": 4.332756093004839e-06,
      "loss": 0.268,
      "step": 36070
    },
    {
      "epoch": 1.6553164150337296,
      "grad_norm": 0.47114959359169006,
      "learning_rate": 4.332513098931546e-06,
      "loss": 0.3422,
      "step": 36071
    },
    {
      "epoch": 1.6553623055389841,
      "grad_norm": 0.456096887588501,
      "learning_rate": 4.332270106463346e-06,
      "loss": 0.3206,
      "step": 36072
    },
    {
      "epoch": 1.6554081960442386,
      "grad_norm": 0.44225654006004333,
      "learning_rate": 4.332027115600827e-06,
      "loss": 0.3111,
      "step": 36073
    },
    {
      "epoch": 1.6554540865494929,
      "grad_norm": 0.4714425206184387,
      "learning_rate": 4.331784126344572e-06,
      "loss": 0.3094,
      "step": 36074
    },
    {
      "epoch": 1.6554999770547474,
      "grad_norm": 0.4704779088497162,
      "learning_rate": 4.331541138695163e-06,
      "loss": 0.3913,
      "step": 36075
    },
    {
      "epoch": 1.6555458675600019,
      "grad_norm": 0.524785578250885,
      "learning_rate": 4.331298152653189e-06,
      "loss": 0.4411,
      "step": 36076
    },
    {
      "epoch": 1.6555917580652562,
      "grad_norm": 0.4599447250366211,
      "learning_rate": 4.3310551682192315e-06,
      "loss": 0.3369,
      "step": 36077
    },
    {
      "epoch": 1.6556376485705107,
      "grad_norm": 0.46362996101379395,
      "learning_rate": 4.330812185393874e-06,
      "loss": 0.3549,
      "step": 36078
    },
    {
      "epoch": 1.6556835390757652,
      "grad_norm": 0.4824897050857544,
      "learning_rate": 4.330569204177704e-06,
      "loss": 0.39,
      "step": 36079
    },
    {
      "epoch": 1.6557294295810197,
      "grad_norm": 0.5028586983680725,
      "learning_rate": 4.3303262245713035e-06,
      "loss": 0.462,
      "step": 36080
    },
    {
      "epoch": 1.6557753200862741,
      "grad_norm": 0.42033007740974426,
      "learning_rate": 4.330083246575255e-06,
      "loss": 0.2855,
      "step": 36081
    },
    {
      "epoch": 1.6558212105915286,
      "grad_norm": 0.44922661781311035,
      "learning_rate": 4.329840270190146e-06,
      "loss": 0.3485,
      "step": 36082
    },
    {
      "epoch": 1.6558671010967831,
      "grad_norm": 0.47912341356277466,
      "learning_rate": 4.329597295416561e-06,
      "loss": 0.3547,
      "step": 36083
    },
    {
      "epoch": 1.6559129916020376,
      "grad_norm": 0.45944979786872864,
      "learning_rate": 4.32935432225508e-06,
      "loss": 0.3054,
      "step": 36084
    },
    {
      "epoch": 1.6559588821072921,
      "grad_norm": 0.4869399964809418,
      "learning_rate": 4.329111350706293e-06,
      "loss": 0.3863,
      "step": 36085
    },
    {
      "epoch": 1.6560047726125464,
      "grad_norm": 0.4809008538722992,
      "learning_rate": 4.328868380770781e-06,
      "loss": 0.3497,
      "step": 36086
    },
    {
      "epoch": 1.656050663117801,
      "grad_norm": 0.45028677582740784,
      "learning_rate": 4.328625412449126e-06,
      "loss": 0.3225,
      "step": 36087
    },
    {
      "epoch": 1.6560965536230554,
      "grad_norm": 0.45062077045440674,
      "learning_rate": 4.328382445741918e-06,
      "loss": 0.3199,
      "step": 36088
    },
    {
      "epoch": 1.6561424441283097,
      "grad_norm": 0.45899972319602966,
      "learning_rate": 4.328139480649738e-06,
      "loss": 0.3479,
      "step": 36089
    },
    {
      "epoch": 1.6561883346335642,
      "grad_norm": 0.4497840106487274,
      "learning_rate": 4.327896517173172e-06,
      "loss": 0.3516,
      "step": 36090
    },
    {
      "epoch": 1.6562342251388187,
      "grad_norm": 0.4907016456127167,
      "learning_rate": 4.3276535553127985e-06,
      "loss": 0.4042,
      "step": 36091
    },
    {
      "epoch": 1.6562801156440732,
      "grad_norm": 0.47974255681037903,
      "learning_rate": 4.327410595069209e-06,
      "loss": 0.4061,
      "step": 36092
    },
    {
      "epoch": 1.6563260061493277,
      "grad_norm": 0.4152029752731323,
      "learning_rate": 4.3271676364429844e-06,
      "loss": 0.271,
      "step": 36093
    },
    {
      "epoch": 1.6563718966545822,
      "grad_norm": 0.5030942559242249,
      "learning_rate": 4.326924679434708e-06,
      "loss": 0.358,
      "step": 36094
    },
    {
      "epoch": 1.6564177871598367,
      "grad_norm": 0.46497970819473267,
      "learning_rate": 4.326681724044966e-06,
      "loss": 0.3338,
      "step": 36095
    },
    {
      "epoch": 1.6564636776650912,
      "grad_norm": 0.5144133567810059,
      "learning_rate": 4.326438770274343e-06,
      "loss": 0.4023,
      "step": 36096
    },
    {
      "epoch": 1.6565095681703457,
      "grad_norm": 0.49707159399986267,
      "learning_rate": 4.3261958181234195e-06,
      "loss": 0.4081,
      "step": 36097
    },
    {
      "epoch": 1.6565554586756002,
      "grad_norm": 0.48274490237236023,
      "learning_rate": 4.325952867592785e-06,
      "loss": 0.3898,
      "step": 36098
    },
    {
      "epoch": 1.6566013491808544,
      "grad_norm": 0.4326018989086151,
      "learning_rate": 4.32570991868302e-06,
      "loss": 0.2839,
      "step": 36099
    },
    {
      "epoch": 1.656647239686109,
      "grad_norm": 0.4729047119617462,
      "learning_rate": 4.32546697139471e-06,
      "loss": 0.3725,
      "step": 36100
    },
    {
      "epoch": 1.6566931301913634,
      "grad_norm": 0.4617598056793213,
      "learning_rate": 4.32522402572844e-06,
      "loss": 0.3688,
      "step": 36101
    },
    {
      "epoch": 1.6567390206966177,
      "grad_norm": 0.5105594992637634,
      "learning_rate": 4.324981081684793e-06,
      "loss": 0.3889,
      "step": 36102
    },
    {
      "epoch": 1.6567849112018722,
      "grad_norm": 0.4739111065864563,
      "learning_rate": 4.32473813926435e-06,
      "loss": 0.3469,
      "step": 36103
    },
    {
      "epoch": 1.6568308017071267,
      "grad_norm": 0.48888498544692993,
      "learning_rate": 4.324495198467702e-06,
      "loss": 0.3535,
      "step": 36104
    },
    {
      "epoch": 1.6568766922123812,
      "grad_norm": 0.4776265621185303,
      "learning_rate": 4.324252259295431e-06,
      "loss": 0.3261,
      "step": 36105
    },
    {
      "epoch": 1.6569225827176357,
      "grad_norm": 0.4853723347187042,
      "learning_rate": 4.324009321748117e-06,
      "loss": 0.3499,
      "step": 36106
    },
    {
      "epoch": 1.6569684732228902,
      "grad_norm": 0.4828360974788666,
      "learning_rate": 4.323766385826349e-06,
      "loss": 0.3837,
      "step": 36107
    },
    {
      "epoch": 1.6570143637281447,
      "grad_norm": 0.44398459792137146,
      "learning_rate": 4.32352345153071e-06,
      "loss": 0.3195,
      "step": 36108
    },
    {
      "epoch": 1.6570602542333992,
      "grad_norm": 0.44199803471565247,
      "learning_rate": 4.323280518861781e-06,
      "loss": 0.3167,
      "step": 36109
    },
    {
      "epoch": 1.6571061447386537,
      "grad_norm": 0.4572855830192566,
      "learning_rate": 4.323037587820151e-06,
      "loss": 0.3762,
      "step": 36110
    },
    {
      "epoch": 1.6571520352439082,
      "grad_norm": 0.48942917585372925,
      "learning_rate": 4.3227946584064016e-06,
      "loss": 0.3846,
      "step": 36111
    },
    {
      "epoch": 1.6571979257491625,
      "grad_norm": 0.4529446065425873,
      "learning_rate": 4.3225517306211165e-06,
      "loss": 0.3138,
      "step": 36112
    },
    {
      "epoch": 1.657243816254417,
      "grad_norm": 0.4470978081226349,
      "learning_rate": 4.322308804464882e-06,
      "loss": 0.3151,
      "step": 36113
    },
    {
      "epoch": 1.6572897067596715,
      "grad_norm": 0.5017333030700684,
      "learning_rate": 4.3220658799382805e-06,
      "loss": 0.4201,
      "step": 36114
    },
    {
      "epoch": 1.6573355972649257,
      "grad_norm": 0.4790981411933899,
      "learning_rate": 4.321822957041896e-06,
      "loss": 0.3667,
      "step": 36115
    },
    {
      "epoch": 1.6573814877701802,
      "grad_norm": 0.491029292345047,
      "learning_rate": 4.321580035776314e-06,
      "loss": 0.3782,
      "step": 36116
    },
    {
      "epoch": 1.6574273782754347,
      "grad_norm": 0.5177568793296814,
      "learning_rate": 4.321337116142118e-06,
      "loss": 0.4502,
      "step": 36117
    },
    {
      "epoch": 1.6574732687806892,
      "grad_norm": 0.6120824813842773,
      "learning_rate": 4.321094198139889e-06,
      "loss": 0.3858,
      "step": 36118
    },
    {
      "epoch": 1.6575191592859437,
      "grad_norm": 0.46056845784187317,
      "learning_rate": 4.320851281770217e-06,
      "loss": 0.3209,
      "step": 36119
    },
    {
      "epoch": 1.6575650497911982,
      "grad_norm": 0.45297276973724365,
      "learning_rate": 4.320608367033684e-06,
      "loss": 0.3242,
      "step": 36120
    },
    {
      "epoch": 1.6576109402964527,
      "grad_norm": 0.4537056088447571,
      "learning_rate": 4.320365453930872e-06,
      "loss": 0.3665,
      "step": 36121
    },
    {
      "epoch": 1.6576568308017072,
      "grad_norm": 0.4922426640987396,
      "learning_rate": 4.3201225424623675e-06,
      "loss": 0.3313,
      "step": 36122
    },
    {
      "epoch": 1.6577027213069617,
      "grad_norm": 0.4928635358810425,
      "learning_rate": 4.319879632628753e-06,
      "loss": 0.3933,
      "step": 36123
    },
    {
      "epoch": 1.657748611812216,
      "grad_norm": 0.45508816838264465,
      "learning_rate": 4.3196367244306145e-06,
      "loss": 0.3112,
      "step": 36124
    },
    {
      "epoch": 1.6577945023174705,
      "grad_norm": 0.4624754786491394,
      "learning_rate": 4.319393817868533e-06,
      "loss": 0.3378,
      "step": 36125
    },
    {
      "epoch": 1.657840392822725,
      "grad_norm": 0.4666912853717804,
      "learning_rate": 4.319150912943096e-06,
      "loss": 0.3615,
      "step": 36126
    },
    {
      "epoch": 1.6578862833279795,
      "grad_norm": 0.4899422824382782,
      "learning_rate": 4.318908009654885e-06,
      "loss": 0.3428,
      "step": 36127
    },
    {
      "epoch": 1.6579321738332338,
      "grad_norm": 0.49616503715515137,
      "learning_rate": 4.318665108004486e-06,
      "loss": 0.4475,
      "step": 36128
    },
    {
      "epoch": 1.6579780643384883,
      "grad_norm": 0.44999024271965027,
      "learning_rate": 4.3184222079924835e-06,
      "loss": 0.3114,
      "step": 36129
    },
    {
      "epoch": 1.6580239548437428,
      "grad_norm": 0.45895087718963623,
      "learning_rate": 4.318179309619459e-06,
      "loss": 0.3611,
      "step": 36130
    },
    {
      "epoch": 1.6580698453489973,
      "grad_norm": 0.4932117760181427,
      "learning_rate": 4.317936412885997e-06,
      "loss": 0.3907,
      "step": 36131
    },
    {
      "epoch": 1.6581157358542518,
      "grad_norm": 0.4635425806045532,
      "learning_rate": 4.317693517792685e-06,
      "loss": 0.3398,
      "step": 36132
    },
    {
      "epoch": 1.6581616263595063,
      "grad_norm": 0.46430590748786926,
      "learning_rate": 4.317450624340103e-06,
      "loss": 0.4197,
      "step": 36133
    },
    {
      "epoch": 1.6582075168647608,
      "grad_norm": 0.42157426476478577,
      "learning_rate": 4.317207732528838e-06,
      "loss": 0.329,
      "step": 36134
    },
    {
      "epoch": 1.6582534073700153,
      "grad_norm": 0.44508829712867737,
      "learning_rate": 4.316964842359473e-06,
      "loss": 0.3118,
      "step": 36135
    },
    {
      "epoch": 1.6582992978752698,
      "grad_norm": 0.4645518660545349,
      "learning_rate": 4.316721953832592e-06,
      "loss": 0.3519,
      "step": 36136
    },
    {
      "epoch": 1.658345188380524,
      "grad_norm": 0.4577440619468689,
      "learning_rate": 4.316479066948778e-06,
      "loss": 0.3271,
      "step": 36137
    },
    {
      "epoch": 1.6583910788857785,
      "grad_norm": 0.4876921474933624,
      "learning_rate": 4.316236181708618e-06,
      "loss": 0.3933,
      "step": 36138
    },
    {
      "epoch": 1.658436969391033,
      "grad_norm": 0.4513891339302063,
      "learning_rate": 4.3159932981126925e-06,
      "loss": 0.3452,
      "step": 36139
    },
    {
      "epoch": 1.6584828598962873,
      "grad_norm": 0.46797114610671997,
      "learning_rate": 4.315750416161586e-06,
      "loss": 0.3789,
      "step": 36140
    },
    {
      "epoch": 1.6585287504015418,
      "grad_norm": 0.541668176651001,
      "learning_rate": 4.3155075358558865e-06,
      "loss": 0.5096,
      "step": 36141
    },
    {
      "epoch": 1.6585746409067963,
      "grad_norm": 0.48768866062164307,
      "learning_rate": 4.3152646571961745e-06,
      "loss": 0.3839,
      "step": 36142
    },
    {
      "epoch": 1.6586205314120508,
      "grad_norm": 0.482728511095047,
      "learning_rate": 4.3150217801830345e-06,
      "loss": 0.3538,
      "step": 36143
    },
    {
      "epoch": 1.6586664219173053,
      "grad_norm": 0.4559321701526642,
      "learning_rate": 4.314778904817052e-06,
      "loss": 0.3357,
      "step": 36144
    },
    {
      "epoch": 1.6587123124225598,
      "grad_norm": 0.48633721470832825,
      "learning_rate": 4.314536031098809e-06,
      "loss": 0.3972,
      "step": 36145
    },
    {
      "epoch": 1.6587582029278143,
      "grad_norm": 0.4220765233039856,
      "learning_rate": 4.314293159028889e-06,
      "loss": 0.288,
      "step": 36146
    },
    {
      "epoch": 1.6588040934330688,
      "grad_norm": 0.49434608221054077,
      "learning_rate": 4.314050288607879e-06,
      "loss": 0.3514,
      "step": 36147
    },
    {
      "epoch": 1.6588499839383233,
      "grad_norm": 0.4648244380950928,
      "learning_rate": 4.313807419836363e-06,
      "loss": 0.3464,
      "step": 36148
    },
    {
      "epoch": 1.6588958744435778,
      "grad_norm": 0.45056143403053284,
      "learning_rate": 4.313564552714921e-06,
      "loss": 0.3098,
      "step": 36149
    },
    {
      "epoch": 1.658941764948832,
      "grad_norm": 0.5130603313446045,
      "learning_rate": 4.313321687244142e-06,
      "loss": 0.426,
      "step": 36150
    },
    {
      "epoch": 1.6589876554540866,
      "grad_norm": 0.48223114013671875,
      "learning_rate": 4.313078823424607e-06,
      "loss": 0.3467,
      "step": 36151
    },
    {
      "epoch": 1.659033545959341,
      "grad_norm": 0.4116940498352051,
      "learning_rate": 4.312835961256898e-06,
      "loss": 0.2823,
      "step": 36152
    },
    {
      "epoch": 1.6590794364645953,
      "grad_norm": 0.4911724030971527,
      "learning_rate": 4.312593100741604e-06,
      "loss": 0.4004,
      "step": 36153
    },
    {
      "epoch": 1.6591253269698498,
      "grad_norm": 0.4769153594970703,
      "learning_rate": 4.312350241879307e-06,
      "loss": 0.3514,
      "step": 36154
    },
    {
      "epoch": 1.6591712174751043,
      "grad_norm": 0.4908784329891205,
      "learning_rate": 4.31210738467059e-06,
      "loss": 0.3687,
      "step": 36155
    },
    {
      "epoch": 1.6592171079803588,
      "grad_norm": 0.4965973198413849,
      "learning_rate": 4.311864529116038e-06,
      "loss": 0.3667,
      "step": 36156
    },
    {
      "epoch": 1.6592629984856133,
      "grad_norm": 0.4554583430290222,
      "learning_rate": 4.311621675216235e-06,
      "loss": 0.3192,
      "step": 36157
    },
    {
      "epoch": 1.6593088889908678,
      "grad_norm": 0.44832873344421387,
      "learning_rate": 4.311378822971765e-06,
      "loss": 0.3497,
      "step": 36158
    },
    {
      "epoch": 1.6593547794961223,
      "grad_norm": 0.5041817426681519,
      "learning_rate": 4.311135972383209e-06,
      "loss": 0.3883,
      "step": 36159
    },
    {
      "epoch": 1.6594006700013768,
      "grad_norm": 0.44811132550239563,
      "learning_rate": 4.310893123451156e-06,
      "loss": 0.3295,
      "step": 36160
    },
    {
      "epoch": 1.6594465605066313,
      "grad_norm": 0.47344714403152466,
      "learning_rate": 4.310650276176187e-06,
      "loss": 0.3631,
      "step": 36161
    },
    {
      "epoch": 1.6594924510118858,
      "grad_norm": 0.46713128685951233,
      "learning_rate": 4.3104074305588845e-06,
      "loss": 0.3661,
      "step": 36162
    },
    {
      "epoch": 1.65953834151714,
      "grad_norm": 0.4695574641227722,
      "learning_rate": 4.3101645865998364e-06,
      "loss": 0.3791,
      "step": 36163
    },
    {
      "epoch": 1.6595842320223946,
      "grad_norm": 0.441419780254364,
      "learning_rate": 4.309921744299625e-06,
      "loss": 0.2945,
      "step": 36164
    },
    {
      "epoch": 1.659630122527649,
      "grad_norm": 0.49451547861099243,
      "learning_rate": 4.309678903658833e-06,
      "loss": 0.4572,
      "step": 36165
    },
    {
      "epoch": 1.6596760130329034,
      "grad_norm": 0.5003684163093567,
      "learning_rate": 4.309436064678047e-06,
      "loss": 0.3841,
      "step": 36166
    },
    {
      "epoch": 1.6597219035381579,
      "grad_norm": 0.49283352494239807,
      "learning_rate": 4.309193227357849e-06,
      "loss": 0.3539,
      "step": 36167
    },
    {
      "epoch": 1.6597677940434123,
      "grad_norm": 0.45494237542152405,
      "learning_rate": 4.30895039169882e-06,
      "loss": 0.3574,
      "step": 36168
    },
    {
      "epoch": 1.6598136845486668,
      "grad_norm": 0.4487442672252655,
      "learning_rate": 4.30870755770155e-06,
      "loss": 0.3281,
      "step": 36169
    },
    {
      "epoch": 1.6598595750539213,
      "grad_norm": 0.4158509075641632,
      "learning_rate": 4.308464725366621e-06,
      "loss": 0.2899,
      "step": 36170
    },
    {
      "epoch": 1.6599054655591758,
      "grad_norm": 0.5776636004447937,
      "learning_rate": 4.308221894694614e-06,
      "loss": 0.3803,
      "step": 36171
    },
    {
      "epoch": 1.6599513560644303,
      "grad_norm": 0.47545215487480164,
      "learning_rate": 4.307979065686117e-06,
      "loss": 0.3867,
      "step": 36172
    },
    {
      "epoch": 1.6599972465696848,
      "grad_norm": 0.4764041006565094,
      "learning_rate": 4.307736238341711e-06,
      "loss": 0.3773,
      "step": 36173
    },
    {
      "epoch": 1.6600431370749393,
      "grad_norm": 0.4419650137424469,
      "learning_rate": 4.307493412661979e-06,
      "loss": 0.3482,
      "step": 36174
    },
    {
      "epoch": 1.6600890275801936,
      "grad_norm": 0.5252270698547363,
      "learning_rate": 4.3072505886475086e-06,
      "loss": 0.3428,
      "step": 36175
    },
    {
      "epoch": 1.660134918085448,
      "grad_norm": 0.4509217441082001,
      "learning_rate": 4.307007766298882e-06,
      "loss": 0.3138,
      "step": 36176
    },
    {
      "epoch": 1.6601808085907026,
      "grad_norm": 0.4895867109298706,
      "learning_rate": 4.3067649456166825e-06,
      "loss": 0.4179,
      "step": 36177
    },
    {
      "epoch": 1.6602266990959569,
      "grad_norm": 0.48821520805358887,
      "learning_rate": 4.306522126601495e-06,
      "loss": 0.3845,
      "step": 36178
    },
    {
      "epoch": 1.6602725896012114,
      "grad_norm": 0.5249105095863342,
      "learning_rate": 4.306279309253903e-06,
      "loss": 0.4358,
      "step": 36179
    },
    {
      "epoch": 1.6603184801064659,
      "grad_norm": 0.4857627749443054,
      "learning_rate": 4.306036493574489e-06,
      "loss": 0.4024,
      "step": 36180
    },
    {
      "epoch": 1.6603643706117204,
      "grad_norm": 0.4672257602214813,
      "learning_rate": 4.305793679563841e-06,
      "loss": 0.3368,
      "step": 36181
    },
    {
      "epoch": 1.6604102611169749,
      "grad_norm": 0.46526116132736206,
      "learning_rate": 4.305550867222539e-06,
      "loss": 0.3339,
      "step": 36182
    },
    {
      "epoch": 1.6604561516222294,
      "grad_norm": 0.4322390854358673,
      "learning_rate": 4.305308056551165e-06,
      "loss": 0.2941,
      "step": 36183
    },
    {
      "epoch": 1.6605020421274839,
      "grad_norm": 0.4829897880554199,
      "learning_rate": 4.305065247550309e-06,
      "loss": 0.3522,
      "step": 36184
    },
    {
      "epoch": 1.6605479326327384,
      "grad_norm": 0.47197550535202026,
      "learning_rate": 4.304822440220552e-06,
      "loss": 0.354,
      "step": 36185
    },
    {
      "epoch": 1.6605938231379929,
      "grad_norm": 0.4809637665748596,
      "learning_rate": 4.304579634562477e-06,
      "loss": 0.4257,
      "step": 36186
    },
    {
      "epoch": 1.6606397136432474,
      "grad_norm": 0.46118879318237305,
      "learning_rate": 4.304336830576667e-06,
      "loss": 0.3149,
      "step": 36187
    },
    {
      "epoch": 1.6606856041485016,
      "grad_norm": 0.4638432264328003,
      "learning_rate": 4.304094028263709e-06,
      "loss": 0.3533,
      "step": 36188
    },
    {
      "epoch": 1.6607314946537561,
      "grad_norm": 0.4894821345806122,
      "learning_rate": 4.303851227624186e-06,
      "loss": 0.413,
      "step": 36189
    },
    {
      "epoch": 1.6607773851590106,
      "grad_norm": 0.5373408794403076,
      "learning_rate": 4.303608428658678e-06,
      "loss": 0.4382,
      "step": 36190
    },
    {
      "epoch": 1.660823275664265,
      "grad_norm": 0.4608858823776245,
      "learning_rate": 4.303365631367774e-06,
      "loss": 0.3224,
      "step": 36191
    },
    {
      "epoch": 1.6608691661695194,
      "grad_norm": 0.42793843150138855,
      "learning_rate": 4.303122835752056e-06,
      "loss": 0.2998,
      "step": 36192
    },
    {
      "epoch": 1.660915056674774,
      "grad_norm": 0.45485982298851013,
      "learning_rate": 4.302880041812106e-06,
      "loss": 0.3379,
      "step": 36193
    },
    {
      "epoch": 1.6609609471800284,
      "grad_norm": 0.45809122920036316,
      "learning_rate": 4.3026372495485115e-06,
      "loss": 0.3262,
      "step": 36194
    },
    {
      "epoch": 1.661006837685283,
      "grad_norm": 0.46151870489120483,
      "learning_rate": 4.302394458961854e-06,
      "loss": 0.3288,
      "step": 36195
    },
    {
      "epoch": 1.6610527281905374,
      "grad_norm": 0.5126960277557373,
      "learning_rate": 4.302151670052715e-06,
      "loss": 0.4593,
      "step": 36196
    },
    {
      "epoch": 1.661098618695792,
      "grad_norm": 0.46597525477409363,
      "learning_rate": 4.301908882821684e-06,
      "loss": 0.374,
      "step": 36197
    },
    {
      "epoch": 1.6611445092010464,
      "grad_norm": 0.49306201934814453,
      "learning_rate": 4.301666097269341e-06,
      "loss": 0.3846,
      "step": 36198
    },
    {
      "epoch": 1.661190399706301,
      "grad_norm": 0.47155600786209106,
      "learning_rate": 4.30142331339627e-06,
      "loss": 0.3528,
      "step": 36199
    },
    {
      "epoch": 1.6612362902115554,
      "grad_norm": 0.4647429883480072,
      "learning_rate": 4.301180531203057e-06,
      "loss": 0.3726,
      "step": 36200
    },
    {
      "epoch": 1.6612821807168097,
      "grad_norm": 0.4472130835056305,
      "learning_rate": 4.300937750690284e-06,
      "loss": 0.34,
      "step": 36201
    },
    {
      "epoch": 1.6613280712220642,
      "grad_norm": 0.4980207085609436,
      "learning_rate": 4.300694971858532e-06,
      "loss": 0.3967,
      "step": 36202
    },
    {
      "epoch": 1.6613739617273187,
      "grad_norm": 0.5025799870491028,
      "learning_rate": 4.300452194708392e-06,
      "loss": 0.4018,
      "step": 36203
    },
    {
      "epoch": 1.661419852232573,
      "grad_norm": 0.4571799635887146,
      "learning_rate": 4.3002094192404425e-06,
      "loss": 0.3503,
      "step": 36204
    },
    {
      "epoch": 1.6614657427378274,
      "grad_norm": 0.4764159619808197,
      "learning_rate": 4.299966645455266e-06,
      "loss": 0.3489,
      "step": 36205
    },
    {
      "epoch": 1.661511633243082,
      "grad_norm": 0.5104197263717651,
      "learning_rate": 4.299723873353451e-06,
      "loss": 0.3898,
      "step": 36206
    },
    {
      "epoch": 1.6615575237483364,
      "grad_norm": 0.45109322667121887,
      "learning_rate": 4.29948110293558e-06,
      "loss": 0.3726,
      "step": 36207
    },
    {
      "epoch": 1.661603414253591,
      "grad_norm": 0.4569368064403534,
      "learning_rate": 4.299238334202234e-06,
      "loss": 0.3392,
      "step": 36208
    },
    {
      "epoch": 1.6616493047588454,
      "grad_norm": 0.44139808416366577,
      "learning_rate": 4.2989955671539994e-06,
      "loss": 0.3122,
      "step": 36209
    },
    {
      "epoch": 1.6616951952641,
      "grad_norm": 0.45661434531211853,
      "learning_rate": 4.2987528017914596e-06,
      "loss": 0.3131,
      "step": 36210
    },
    {
      "epoch": 1.6617410857693544,
      "grad_norm": 0.47844335436820984,
      "learning_rate": 4.298510038115195e-06,
      "loss": 0.3591,
      "step": 36211
    },
    {
      "epoch": 1.661786976274609,
      "grad_norm": 0.4727175831794739,
      "learning_rate": 4.298267276125796e-06,
      "loss": 0.3984,
      "step": 36212
    },
    {
      "epoch": 1.6618328667798632,
      "grad_norm": 0.5041042566299438,
      "learning_rate": 4.298024515823842e-06,
      "loss": 0.3878,
      "step": 36213
    },
    {
      "epoch": 1.6618787572851177,
      "grad_norm": 0.4534107744693756,
      "learning_rate": 4.297781757209917e-06,
      "loss": 0.3335,
      "step": 36214
    },
    {
      "epoch": 1.6619246477903722,
      "grad_norm": 0.47263941168785095,
      "learning_rate": 4.2975390002846056e-06,
      "loss": 0.3751,
      "step": 36215
    },
    {
      "epoch": 1.6619705382956267,
      "grad_norm": 0.46041783690452576,
      "learning_rate": 4.297296245048491e-06,
      "loss": 0.3647,
      "step": 36216
    },
    {
      "epoch": 1.662016428800881,
      "grad_norm": 0.5919760465621948,
      "learning_rate": 4.297053491502156e-06,
      "loss": 0.3277,
      "step": 36217
    },
    {
      "epoch": 1.6620623193061355,
      "grad_norm": 0.4689157009124756,
      "learning_rate": 4.296810739646186e-06,
      "loss": 0.3828,
      "step": 36218
    },
    {
      "epoch": 1.66210820981139,
      "grad_norm": 0.49491897225379944,
      "learning_rate": 4.296567989481166e-06,
      "loss": 0.4025,
      "step": 36219
    },
    {
      "epoch": 1.6621541003166445,
      "grad_norm": 0.4740844964981079,
      "learning_rate": 4.296325241007677e-06,
      "loss": 0.3683,
      "step": 36220
    },
    {
      "epoch": 1.662199990821899,
      "grad_norm": 0.5126505494117737,
      "learning_rate": 4.296082494226303e-06,
      "loss": 0.3932,
      "step": 36221
    },
    {
      "epoch": 1.6622458813271535,
      "grad_norm": 0.45973503589630127,
      "learning_rate": 4.29583974913763e-06,
      "loss": 0.333,
      "step": 36222
    },
    {
      "epoch": 1.662291771832408,
      "grad_norm": 0.46231767535209656,
      "learning_rate": 4.295597005742239e-06,
      "loss": 0.3485,
      "step": 36223
    },
    {
      "epoch": 1.6623376623376624,
      "grad_norm": 0.4517453908920288,
      "learning_rate": 4.295354264040713e-06,
      "loss": 0.3204,
      "step": 36224
    },
    {
      "epoch": 1.662383552842917,
      "grad_norm": 0.4346649646759033,
      "learning_rate": 4.295111524033641e-06,
      "loss": 0.2728,
      "step": 36225
    },
    {
      "epoch": 1.6624294433481712,
      "grad_norm": 0.48042160272598267,
      "learning_rate": 4.294868785721603e-06,
      "loss": 0.3267,
      "step": 36226
    },
    {
      "epoch": 1.6624753338534257,
      "grad_norm": 0.5077573657035828,
      "learning_rate": 4.29462604910518e-06,
      "loss": 0.3318,
      "step": 36227
    },
    {
      "epoch": 1.6625212243586802,
      "grad_norm": 0.43770632147789,
      "learning_rate": 4.294383314184961e-06,
      "loss": 0.304,
      "step": 36228
    },
    {
      "epoch": 1.6625671148639345,
      "grad_norm": 0.4767512381076813,
      "learning_rate": 4.294140580961528e-06,
      "loss": 0.3989,
      "step": 36229
    },
    {
      "epoch": 1.662613005369189,
      "grad_norm": 0.46568331122398376,
      "learning_rate": 4.293897849435462e-06,
      "loss": 0.3664,
      "step": 36230
    },
    {
      "epoch": 1.6626588958744435,
      "grad_norm": 0.4980507493019104,
      "learning_rate": 4.293655119607351e-06,
      "loss": 0.3803,
      "step": 36231
    },
    {
      "epoch": 1.662704786379698,
      "grad_norm": 0.4911334812641144,
      "learning_rate": 4.293412391477776e-06,
      "loss": 0.3696,
      "step": 36232
    },
    {
      "epoch": 1.6627506768849525,
      "grad_norm": 0.4816778898239136,
      "learning_rate": 4.293169665047319e-06,
      "loss": 0.3095,
      "step": 36233
    },
    {
      "epoch": 1.662796567390207,
      "grad_norm": 0.45551568269729614,
      "learning_rate": 4.292926940316568e-06,
      "loss": 0.3765,
      "step": 36234
    },
    {
      "epoch": 1.6628424578954615,
      "grad_norm": 0.4306444823741913,
      "learning_rate": 4.292684217286105e-06,
      "loss": 0.2949,
      "step": 36235
    },
    {
      "epoch": 1.662888348400716,
      "grad_norm": 0.4980430006980896,
      "learning_rate": 4.292441495956511e-06,
      "loss": 0.4266,
      "step": 36236
    },
    {
      "epoch": 1.6629342389059705,
      "grad_norm": 0.4745534360408783,
      "learning_rate": 4.292198776328374e-06,
      "loss": 0.3737,
      "step": 36237
    },
    {
      "epoch": 1.662980129411225,
      "grad_norm": 0.5083572864532471,
      "learning_rate": 4.291956058402276e-06,
      "loss": 0.3724,
      "step": 36238
    },
    {
      "epoch": 1.6630260199164792,
      "grad_norm": 0.46010252833366394,
      "learning_rate": 4.291713342178797e-06,
      "loss": 0.3415,
      "step": 36239
    },
    {
      "epoch": 1.6630719104217337,
      "grad_norm": 0.4652763307094574,
      "learning_rate": 4.291470627658526e-06,
      "loss": 0.3795,
      "step": 36240
    },
    {
      "epoch": 1.6631178009269882,
      "grad_norm": 0.48567676544189453,
      "learning_rate": 4.291227914842045e-06,
      "loss": 0.3609,
      "step": 36241
    },
    {
      "epoch": 1.6631636914322425,
      "grad_norm": 0.4545653760433197,
      "learning_rate": 4.290985203729936e-06,
      "loss": 0.3113,
      "step": 36242
    },
    {
      "epoch": 1.663209581937497,
      "grad_norm": 0.453704297542572,
      "learning_rate": 4.290742494322785e-06,
      "loss": 0.3219,
      "step": 36243
    },
    {
      "epoch": 1.6632554724427515,
      "grad_norm": 0.4520237445831299,
      "learning_rate": 4.290499786621174e-06,
      "loss": 0.348,
      "step": 36244
    },
    {
      "epoch": 1.663301362948006,
      "grad_norm": 0.484774112701416,
      "learning_rate": 4.290257080625686e-06,
      "loss": 0.3431,
      "step": 36245
    },
    {
      "epoch": 1.6633472534532605,
      "grad_norm": 0.47661012411117554,
      "learning_rate": 4.2900143763369066e-06,
      "loss": 0.3929,
      "step": 36246
    },
    {
      "epoch": 1.663393143958515,
      "grad_norm": 0.4306141436100006,
      "learning_rate": 4.2897716737554204e-06,
      "loss": 0.3201,
      "step": 36247
    },
    {
      "epoch": 1.6634390344637695,
      "grad_norm": 0.48564422130584717,
      "learning_rate": 4.289528972881806e-06,
      "loss": 0.3521,
      "step": 36248
    },
    {
      "epoch": 1.663484924969024,
      "grad_norm": 0.48704737424850464,
      "learning_rate": 4.289286273716652e-06,
      "loss": 0.3803,
      "step": 36249
    },
    {
      "epoch": 1.6635308154742785,
      "grad_norm": 0.5362771153450012,
      "learning_rate": 4.28904357626054e-06,
      "loss": 0.4631,
      "step": 36250
    },
    {
      "epoch": 1.663576705979533,
      "grad_norm": 0.47447440028190613,
      "learning_rate": 4.288800880514053e-06,
      "loss": 0.411,
      "step": 36251
    },
    {
      "epoch": 1.6636225964847873,
      "grad_norm": 0.4667125642299652,
      "learning_rate": 4.288558186477776e-06,
      "loss": 0.4023,
      "step": 36252
    },
    {
      "epoch": 1.6636684869900418,
      "grad_norm": 0.4606325328350067,
      "learning_rate": 4.288315494152293e-06,
      "loss": 0.3004,
      "step": 36253
    },
    {
      "epoch": 1.6637143774952963,
      "grad_norm": 0.45616137981414795,
      "learning_rate": 4.2880728035381855e-06,
      "loss": 0.3438,
      "step": 36254
    },
    {
      "epoch": 1.6637602680005505,
      "grad_norm": 0.47695672512054443,
      "learning_rate": 4.287830114636037e-06,
      "loss": 0.4234,
      "step": 36255
    },
    {
      "epoch": 1.663806158505805,
      "grad_norm": 0.4499322175979614,
      "learning_rate": 4.287587427446434e-06,
      "loss": 0.3426,
      "step": 36256
    },
    {
      "epoch": 1.6638520490110595,
      "grad_norm": 0.5467880368232727,
      "learning_rate": 4.287344741969959e-06,
      "loss": 0.494,
      "step": 36257
    },
    {
      "epoch": 1.663897939516314,
      "grad_norm": 0.47830554842948914,
      "learning_rate": 4.287102058207193e-06,
      "loss": 0.3596,
      "step": 36258
    },
    {
      "epoch": 1.6639438300215685,
      "grad_norm": 0.49418142437934875,
      "learning_rate": 4.286859376158723e-06,
      "loss": 0.3798,
      "step": 36259
    },
    {
      "epoch": 1.663989720526823,
      "grad_norm": 0.4680376350879669,
      "learning_rate": 4.286616695825132e-06,
      "loss": 0.3375,
      "step": 36260
    },
    {
      "epoch": 1.6640356110320775,
      "grad_norm": 0.4352279603481293,
      "learning_rate": 4.286374017207e-06,
      "loss": 0.344,
      "step": 36261
    },
    {
      "epoch": 1.664081501537332,
      "grad_norm": 0.45345935225486755,
      "learning_rate": 4.286131340304914e-06,
      "loss": 0.3184,
      "step": 36262
    },
    {
      "epoch": 1.6641273920425865,
      "grad_norm": 0.4989250600337982,
      "learning_rate": 4.2858886651194585e-06,
      "loss": 0.3871,
      "step": 36263
    },
    {
      "epoch": 1.6641732825478408,
      "grad_norm": 0.46464794874191284,
      "learning_rate": 4.285645991651214e-06,
      "loss": 0.303,
      "step": 36264
    },
    {
      "epoch": 1.6642191730530953,
      "grad_norm": 0.48235201835632324,
      "learning_rate": 4.285403319900766e-06,
      "loss": 0.3663,
      "step": 36265
    },
    {
      "epoch": 1.6642650635583498,
      "grad_norm": 0.45667579770088196,
      "learning_rate": 4.285160649868698e-06,
      "loss": 0.3623,
      "step": 36266
    },
    {
      "epoch": 1.664310954063604,
      "grad_norm": 0.4917313754558563,
      "learning_rate": 4.284917981555591e-06,
      "loss": 0.4021,
      "step": 36267
    },
    {
      "epoch": 1.6643568445688586,
      "grad_norm": 0.47283604741096497,
      "learning_rate": 4.284675314962032e-06,
      "loss": 0.3846,
      "step": 36268
    },
    {
      "epoch": 1.664402735074113,
      "grad_norm": 0.4974950850009918,
      "learning_rate": 4.284432650088604e-06,
      "loss": 0.394,
      "step": 36269
    },
    {
      "epoch": 1.6644486255793676,
      "grad_norm": 0.46044453978538513,
      "learning_rate": 4.284189986935887e-06,
      "loss": 0.3414,
      "step": 36270
    },
    {
      "epoch": 1.664494516084622,
      "grad_norm": 0.4786221981048584,
      "learning_rate": 4.283947325504469e-06,
      "loss": 0.3834,
      "step": 36271
    },
    {
      "epoch": 1.6645404065898766,
      "grad_norm": 0.4130963683128357,
      "learning_rate": 4.283704665794932e-06,
      "loss": 0.2798,
      "step": 36272
    },
    {
      "epoch": 1.664586297095131,
      "grad_norm": 0.46508416533470154,
      "learning_rate": 4.283462007807859e-06,
      "loss": 0.3562,
      "step": 36273
    },
    {
      "epoch": 1.6646321876003856,
      "grad_norm": 0.46477949619293213,
      "learning_rate": 4.283219351543834e-06,
      "loss": 0.3611,
      "step": 36274
    },
    {
      "epoch": 1.66467807810564,
      "grad_norm": 0.5179890990257263,
      "learning_rate": 4.28297669700344e-06,
      "loss": 0.4346,
      "step": 36275
    },
    {
      "epoch": 1.6647239686108946,
      "grad_norm": 0.4880692660808563,
      "learning_rate": 4.282734044187259e-06,
      "loss": 0.3738,
      "step": 36276
    },
    {
      "epoch": 1.6647698591161488,
      "grad_norm": 0.4676433503627777,
      "learning_rate": 4.282491393095878e-06,
      "loss": 0.3155,
      "step": 36277
    },
    {
      "epoch": 1.6648157496214033,
      "grad_norm": 0.5174286961555481,
      "learning_rate": 4.282248743729879e-06,
      "loss": 0.5062,
      "step": 36278
    },
    {
      "epoch": 1.6648616401266578,
      "grad_norm": 0.47100773453712463,
      "learning_rate": 4.282006096089844e-06,
      "loss": 0.3913,
      "step": 36279
    },
    {
      "epoch": 1.664907530631912,
      "grad_norm": 0.40979066491127014,
      "learning_rate": 4.281763450176359e-06,
      "loss": 0.276,
      "step": 36280
    },
    {
      "epoch": 1.6649534211371666,
      "grad_norm": 0.5318982601165771,
      "learning_rate": 4.281520805990006e-06,
      "loss": 0.4064,
      "step": 36281
    },
    {
      "epoch": 1.664999311642421,
      "grad_norm": 0.49839577078819275,
      "learning_rate": 4.281278163531366e-06,
      "loss": 0.405,
      "step": 36282
    },
    {
      "epoch": 1.6650452021476756,
      "grad_norm": 0.482532262802124,
      "learning_rate": 4.281035522801028e-06,
      "loss": 0.3993,
      "step": 36283
    },
    {
      "epoch": 1.66509109265293,
      "grad_norm": 0.49582916498184204,
      "learning_rate": 4.280792883799572e-06,
      "loss": 0.4073,
      "step": 36284
    },
    {
      "epoch": 1.6651369831581846,
      "grad_norm": 0.4148208200931549,
      "learning_rate": 4.280550246527584e-06,
      "loss": 0.2603,
      "step": 36285
    },
    {
      "epoch": 1.665182873663439,
      "grad_norm": 0.4479190707206726,
      "learning_rate": 4.280307610985643e-06,
      "loss": 0.3198,
      "step": 36286
    },
    {
      "epoch": 1.6652287641686936,
      "grad_norm": 0.6520950198173523,
      "learning_rate": 4.2800649771743365e-06,
      "loss": 0.3625,
      "step": 36287
    },
    {
      "epoch": 1.665274654673948,
      "grad_norm": 0.43715718388557434,
      "learning_rate": 4.279822345094246e-06,
      "loss": 0.3138,
      "step": 36288
    },
    {
      "epoch": 1.6653205451792026,
      "grad_norm": 0.43279600143432617,
      "learning_rate": 4.2795797147459535e-06,
      "loss": 0.3124,
      "step": 36289
    },
    {
      "epoch": 1.6653664356844569,
      "grad_norm": 0.464556485414505,
      "learning_rate": 4.279337086130047e-06,
      "loss": 0.3525,
      "step": 36290
    },
    {
      "epoch": 1.6654123261897114,
      "grad_norm": 0.444024920463562,
      "learning_rate": 4.279094459247108e-06,
      "loss": 0.3354,
      "step": 36291
    },
    {
      "epoch": 1.6654582166949659,
      "grad_norm": 0.4661685526371002,
      "learning_rate": 4.278851834097717e-06,
      "loss": 0.3395,
      "step": 36292
    },
    {
      "epoch": 1.6655041072002201,
      "grad_norm": 0.48338985443115234,
      "learning_rate": 4.2786092106824605e-06,
      "loss": 0.3819,
      "step": 36293
    },
    {
      "epoch": 1.6655499977054746,
      "grad_norm": 0.45614129304885864,
      "learning_rate": 4.278366589001922e-06,
      "loss": 0.3138,
      "step": 36294
    },
    {
      "epoch": 1.6655958882107291,
      "grad_norm": 0.46891483664512634,
      "learning_rate": 4.278123969056683e-06,
      "loss": 0.3951,
      "step": 36295
    },
    {
      "epoch": 1.6656417787159836,
      "grad_norm": 0.46651163697242737,
      "learning_rate": 4.277881350847329e-06,
      "loss": 0.3283,
      "step": 36296
    },
    {
      "epoch": 1.6656876692212381,
      "grad_norm": 0.45361328125,
      "learning_rate": 4.277638734374442e-06,
      "loss": 0.3488,
      "step": 36297
    },
    {
      "epoch": 1.6657335597264926,
      "grad_norm": 0.47035273909568787,
      "learning_rate": 4.277396119638605e-06,
      "loss": 0.3747,
      "step": 36298
    },
    {
      "epoch": 1.6657794502317471,
      "grad_norm": 0.47960934042930603,
      "learning_rate": 4.2771535066404035e-06,
      "loss": 0.391,
      "step": 36299
    },
    {
      "epoch": 1.6658253407370016,
      "grad_norm": 0.5951382517814636,
      "learning_rate": 4.276910895380419e-06,
      "loss": 0.3287,
      "step": 36300
    },
    {
      "epoch": 1.6658712312422561,
      "grad_norm": 0.45369866490364075,
      "learning_rate": 4.276668285859235e-06,
      "loss": 0.3608,
      "step": 36301
    },
    {
      "epoch": 1.6659171217475104,
      "grad_norm": 0.44291356205940247,
      "learning_rate": 4.276425678077438e-06,
      "loss": 0.3372,
      "step": 36302
    },
    {
      "epoch": 1.6659630122527649,
      "grad_norm": 0.5147892236709595,
      "learning_rate": 4.276183072035607e-06,
      "loss": 0.4503,
      "step": 36303
    },
    {
      "epoch": 1.6660089027580194,
      "grad_norm": 0.476492702960968,
      "learning_rate": 4.275940467734325e-06,
      "loss": 0.311,
      "step": 36304
    },
    {
      "epoch": 1.6660547932632739,
      "grad_norm": 0.5348714590072632,
      "learning_rate": 4.275697865174182e-06,
      "loss": 0.4501,
      "step": 36305
    },
    {
      "epoch": 1.6661006837685282,
      "grad_norm": 0.4763276278972626,
      "learning_rate": 4.275455264355755e-06,
      "loss": 0.3797,
      "step": 36306
    },
    {
      "epoch": 1.6661465742737827,
      "grad_norm": 0.4355514347553253,
      "learning_rate": 4.275212665279629e-06,
      "loss": 0.3148,
      "step": 36307
    },
    {
      "epoch": 1.6661924647790372,
      "grad_norm": 0.523844301700592,
      "learning_rate": 4.274970067946389e-06,
      "loss": 0.3998,
      "step": 36308
    },
    {
      "epoch": 1.6662383552842917,
      "grad_norm": 0.46772611141204834,
      "learning_rate": 4.274727472356617e-06,
      "loss": 0.3653,
      "step": 36309
    },
    {
      "epoch": 1.6662842457895461,
      "grad_norm": 0.4746570885181427,
      "learning_rate": 4.274484878510895e-06,
      "loss": 0.3743,
      "step": 36310
    },
    {
      "epoch": 1.6663301362948006,
      "grad_norm": 0.46243345737457275,
      "learning_rate": 4.274242286409809e-06,
      "loss": 0.3071,
      "step": 36311
    },
    {
      "epoch": 1.6663760268000551,
      "grad_norm": 0.4791312515735626,
      "learning_rate": 4.273999696053941e-06,
      "loss": 0.3587,
      "step": 36312
    },
    {
      "epoch": 1.6664219173053096,
      "grad_norm": 0.41274598240852356,
      "learning_rate": 4.273757107443875e-06,
      "loss": 0.29,
      "step": 36313
    },
    {
      "epoch": 1.6664678078105641,
      "grad_norm": 0.4645887017250061,
      "learning_rate": 4.273514520580194e-06,
      "loss": 0.3557,
      "step": 36314
    },
    {
      "epoch": 1.6665136983158184,
      "grad_norm": 0.48121514916419983,
      "learning_rate": 4.273271935463482e-06,
      "loss": 0.3544,
      "step": 36315
    },
    {
      "epoch": 1.666559588821073,
      "grad_norm": 0.4817105829715729,
      "learning_rate": 4.27302935209432e-06,
      "loss": 0.4152,
      "step": 36316
    },
    {
      "epoch": 1.6666054793263274,
      "grad_norm": 0.49046632647514343,
      "learning_rate": 4.272786770473294e-06,
      "loss": 0.4726,
      "step": 36317
    },
    {
      "epoch": 1.6666513698315817,
      "grad_norm": 0.5043453574180603,
      "learning_rate": 4.2725441906009866e-06,
      "loss": 0.3956,
      "step": 36318
    },
    {
      "epoch": 1.6666972603368362,
      "grad_norm": 0.46656617522239685,
      "learning_rate": 4.272301612477981e-06,
      "loss": 0.3222,
      "step": 36319
    },
    {
      "epoch": 1.6667431508420907,
      "grad_norm": 0.45955711603164673,
      "learning_rate": 4.272059036104857e-06,
      "loss": 0.3464,
      "step": 36320
    },
    {
      "epoch": 1.6667890413473452,
      "grad_norm": 0.48475947976112366,
      "learning_rate": 4.271816461482205e-06,
      "loss": 0.373,
      "step": 36321
    },
    {
      "epoch": 1.6668349318525997,
      "grad_norm": 0.4875313341617584,
      "learning_rate": 4.271573888610603e-06,
      "loss": 0.3308,
      "step": 36322
    },
    {
      "epoch": 1.6668808223578542,
      "grad_norm": 0.42728909850120544,
      "learning_rate": 4.2713313174906354e-06,
      "loss": 0.3115,
      "step": 36323
    },
    {
      "epoch": 1.6669267128631087,
      "grad_norm": 0.4502031207084656,
      "learning_rate": 4.271088748122888e-06,
      "loss": 0.3179,
      "step": 36324
    },
    {
      "epoch": 1.6669726033683632,
      "grad_norm": 0.42734870314598083,
      "learning_rate": 4.270846180507941e-06,
      "loss": 0.2855,
      "step": 36325
    },
    {
      "epoch": 1.6670184938736177,
      "grad_norm": 0.44200703501701355,
      "learning_rate": 4.270603614646378e-06,
      "loss": 0.3442,
      "step": 36326
    },
    {
      "epoch": 1.6670643843788722,
      "grad_norm": 0.4745289981365204,
      "learning_rate": 4.270361050538784e-06,
      "loss": 0.3119,
      "step": 36327
    },
    {
      "epoch": 1.6671102748841264,
      "grad_norm": 0.47574561834335327,
      "learning_rate": 4.270118488185742e-06,
      "loss": 0.325,
      "step": 36328
    },
    {
      "epoch": 1.667156165389381,
      "grad_norm": 0.45527952909469604,
      "learning_rate": 4.269875927587833e-06,
      "loss": 0.3307,
      "step": 36329
    },
    {
      "epoch": 1.6672020558946354,
      "grad_norm": 0.4937826693058014,
      "learning_rate": 4.269633368745643e-06,
      "loss": 0.3332,
      "step": 36330
    },
    {
      "epoch": 1.6672479463998897,
      "grad_norm": 0.4832055866718292,
      "learning_rate": 4.269390811659755e-06,
      "loss": 0.4036,
      "step": 36331
    },
    {
      "epoch": 1.6672938369051442,
      "grad_norm": 0.4986433982849121,
      "learning_rate": 4.269148256330749e-06,
      "loss": 0.3779,
      "step": 36332
    },
    {
      "epoch": 1.6673397274103987,
      "grad_norm": 0.49590370059013367,
      "learning_rate": 4.268905702759213e-06,
      "loss": 0.4297,
      "step": 36333
    },
    {
      "epoch": 1.6673856179156532,
      "grad_norm": 0.45368897914886475,
      "learning_rate": 4.268663150945727e-06,
      "loss": 0.2966,
      "step": 36334
    },
    {
      "epoch": 1.6674315084209077,
      "grad_norm": 0.4347810447216034,
      "learning_rate": 4.268420600890876e-06,
      "loss": 0.3172,
      "step": 36335
    },
    {
      "epoch": 1.6674773989261622,
      "grad_norm": 0.4685761332511902,
      "learning_rate": 4.268178052595244e-06,
      "loss": 0.3468,
      "step": 36336
    },
    {
      "epoch": 1.6675232894314167,
      "grad_norm": 0.44975653290748596,
      "learning_rate": 4.267935506059411e-06,
      "loss": 0.3183,
      "step": 36337
    },
    {
      "epoch": 1.6675691799366712,
      "grad_norm": 0.470132052898407,
      "learning_rate": 4.267692961283961e-06,
      "loss": 0.3384,
      "step": 36338
    },
    {
      "epoch": 1.6676150704419257,
      "grad_norm": 0.523624062538147,
      "learning_rate": 4.267450418269481e-06,
      "loss": 0.4281,
      "step": 36339
    },
    {
      "epoch": 1.6676609609471802,
      "grad_norm": 0.4622613787651062,
      "learning_rate": 4.267207877016549e-06,
      "loss": 0.3283,
      "step": 36340
    },
    {
      "epoch": 1.6677068514524345,
      "grad_norm": 0.48685312271118164,
      "learning_rate": 4.2669653375257516e-06,
      "loss": 0.3775,
      "step": 36341
    },
    {
      "epoch": 1.667752741957689,
      "grad_norm": 0.49793750047683716,
      "learning_rate": 4.266722799797672e-06,
      "loss": 0.3989,
      "step": 36342
    },
    {
      "epoch": 1.6677986324629435,
      "grad_norm": 0.4666235148906708,
      "learning_rate": 4.266480263832892e-06,
      "loss": 0.3266,
      "step": 36343
    },
    {
      "epoch": 1.6678445229681977,
      "grad_norm": 0.4603443741798401,
      "learning_rate": 4.266237729631995e-06,
      "loss": 0.3695,
      "step": 36344
    },
    {
      "epoch": 1.6678904134734522,
      "grad_norm": 0.4639434516429901,
      "learning_rate": 4.265995197195566e-06,
      "loss": 0.3934,
      "step": 36345
    },
    {
      "epoch": 1.6679363039787067,
      "grad_norm": 0.47402071952819824,
      "learning_rate": 4.265752666524186e-06,
      "loss": 0.372,
      "step": 36346
    },
    {
      "epoch": 1.6679821944839612,
      "grad_norm": 0.4229109287261963,
      "learning_rate": 4.265510137618438e-06,
      "loss": 0.2922,
      "step": 36347
    },
    {
      "epoch": 1.6680280849892157,
      "grad_norm": 0.4681171178817749,
      "learning_rate": 4.265267610478908e-06,
      "loss": 0.3341,
      "step": 36348
    },
    {
      "epoch": 1.6680739754944702,
      "grad_norm": 0.43394458293914795,
      "learning_rate": 4.265025085106178e-06,
      "loss": 0.3142,
      "step": 36349
    },
    {
      "epoch": 1.6681198659997247,
      "grad_norm": 0.5006320476531982,
      "learning_rate": 4.26478256150083e-06,
      "loss": 0.3873,
      "step": 36350
    },
    {
      "epoch": 1.6681657565049792,
      "grad_norm": 0.47395649552345276,
      "learning_rate": 4.2645400396634475e-06,
      "loss": 0.4092,
      "step": 36351
    },
    {
      "epoch": 1.6682116470102337,
      "grad_norm": 0.5267159342765808,
      "learning_rate": 4.2642975195946146e-06,
      "loss": 0.2888,
      "step": 36352
    },
    {
      "epoch": 1.668257537515488,
      "grad_norm": 0.5016491413116455,
      "learning_rate": 4.264055001294914e-06,
      "loss": 0.4657,
      "step": 36353
    },
    {
      "epoch": 1.6683034280207425,
      "grad_norm": 0.44743773341178894,
      "learning_rate": 4.263812484764926e-06,
      "loss": 0.3689,
      "step": 36354
    },
    {
      "epoch": 1.668349318525997,
      "grad_norm": 0.4475707411766052,
      "learning_rate": 4.26356997000524e-06,
      "loss": 0.2916,
      "step": 36355
    },
    {
      "epoch": 1.6683952090312513,
      "grad_norm": 0.48303112387657166,
      "learning_rate": 4.263327457016435e-06,
      "loss": 0.4169,
      "step": 36356
    },
    {
      "epoch": 1.6684410995365058,
      "grad_norm": 0.5823234915733337,
      "learning_rate": 4.263084945799095e-06,
      "loss": 0.4071,
      "step": 36357
    },
    {
      "epoch": 1.6684869900417603,
      "grad_norm": 0.511863648891449,
      "learning_rate": 4.262842436353803e-06,
      "loss": 0.4691,
      "step": 36358
    },
    {
      "epoch": 1.6685328805470148,
      "grad_norm": 0.46549227833747864,
      "learning_rate": 4.2625999286811435e-06,
      "loss": 0.3123,
      "step": 36359
    },
    {
      "epoch": 1.6685787710522693,
      "grad_norm": 0.4582907259464264,
      "learning_rate": 4.262357422781697e-06,
      "loss": 0.3396,
      "step": 36360
    },
    {
      "epoch": 1.6686246615575238,
      "grad_norm": 0.45790350437164307,
      "learning_rate": 4.2621149186560486e-06,
      "loss": 0.3424,
      "step": 36361
    },
    {
      "epoch": 1.6686705520627783,
      "grad_norm": 0.4680982530117035,
      "learning_rate": 4.261872416304781e-06,
      "loss": 0.343,
      "step": 36362
    },
    {
      "epoch": 1.6687164425680328,
      "grad_norm": 0.4952363967895508,
      "learning_rate": 4.261629915728475e-06,
      "loss": 0.4213,
      "step": 36363
    },
    {
      "epoch": 1.6687623330732873,
      "grad_norm": 0.43381890654563904,
      "learning_rate": 4.261387416927718e-06,
      "loss": 0.3196,
      "step": 36364
    },
    {
      "epoch": 1.6688082235785417,
      "grad_norm": 0.4501165747642517,
      "learning_rate": 4.261144919903092e-06,
      "loss": 0.3825,
      "step": 36365
    },
    {
      "epoch": 1.668854114083796,
      "grad_norm": 0.4674747586250305,
      "learning_rate": 4.260902424655178e-06,
      "loss": 0.3454,
      "step": 36366
    },
    {
      "epoch": 1.6689000045890505,
      "grad_norm": 0.44054991006851196,
      "learning_rate": 4.260659931184561e-06,
      "loss": 0.3089,
      "step": 36367
    },
    {
      "epoch": 1.668945895094305,
      "grad_norm": 0.48704859614372253,
      "learning_rate": 4.260417439491824e-06,
      "loss": 0.3911,
      "step": 36368
    },
    {
      "epoch": 1.6689917855995593,
      "grad_norm": 0.457518994808197,
      "learning_rate": 4.260174949577548e-06,
      "loss": 0.3368,
      "step": 36369
    },
    {
      "epoch": 1.6690376761048138,
      "grad_norm": 0.5266293287277222,
      "learning_rate": 4.2599324614423185e-06,
      "loss": 0.3607,
      "step": 36370
    },
    {
      "epoch": 1.6690835666100683,
      "grad_norm": 0.49156007170677185,
      "learning_rate": 4.259689975086718e-06,
      "loss": 0.4164,
      "step": 36371
    },
    {
      "epoch": 1.6691294571153228,
      "grad_norm": 0.46097517013549805,
      "learning_rate": 4.25944749051133e-06,
      "loss": 0.3095,
      "step": 36372
    },
    {
      "epoch": 1.6691753476205773,
      "grad_norm": 0.4610563814640045,
      "learning_rate": 4.259205007716737e-06,
      "loss": 0.327,
      "step": 36373
    },
    {
      "epoch": 1.6692212381258318,
      "grad_norm": 0.441128671169281,
      "learning_rate": 4.258962526703522e-06,
      "loss": 0.3061,
      "step": 36374
    },
    {
      "epoch": 1.6692671286310863,
      "grad_norm": 0.5286633968353271,
      "learning_rate": 4.2587200474722666e-06,
      "loss": 0.3806,
      "step": 36375
    },
    {
      "epoch": 1.6693130191363408,
      "grad_norm": 0.4847819209098816,
      "learning_rate": 4.258477570023557e-06,
      "loss": 0.3887,
      "step": 36376
    },
    {
      "epoch": 1.6693589096415953,
      "grad_norm": 0.4531690776348114,
      "learning_rate": 4.258235094357975e-06,
      "loss": 0.3292,
      "step": 36377
    },
    {
      "epoch": 1.6694048001468498,
      "grad_norm": 0.4556334912776947,
      "learning_rate": 4.2579926204761025e-06,
      "loss": 0.3241,
      "step": 36378
    },
    {
      "epoch": 1.669450690652104,
      "grad_norm": 0.4550252854824066,
      "learning_rate": 4.257750148378525e-06,
      "loss": 0.3072,
      "step": 36379
    },
    {
      "epoch": 1.6694965811573586,
      "grad_norm": 0.4946868419647217,
      "learning_rate": 4.257507678065824e-06,
      "loss": 0.3578,
      "step": 36380
    },
    {
      "epoch": 1.669542471662613,
      "grad_norm": 0.47052067518234253,
      "learning_rate": 4.25726520953858e-06,
      "loss": 0.3768,
      "step": 36381
    },
    {
      "epoch": 1.6695883621678673,
      "grad_norm": 0.5159062147140503,
      "learning_rate": 4.257022742797382e-06,
      "loss": 0.4535,
      "step": 36382
    },
    {
      "epoch": 1.6696342526731218,
      "grad_norm": 0.49265843629837036,
      "learning_rate": 4.2567802778428105e-06,
      "loss": 0.4313,
      "step": 36383
    },
    {
      "epoch": 1.6696801431783763,
      "grad_norm": 0.5001285076141357,
      "learning_rate": 4.256537814675446e-06,
      "loss": 0.3804,
      "step": 36384
    },
    {
      "epoch": 1.6697260336836308,
      "grad_norm": 0.4813125729560852,
      "learning_rate": 4.256295353295872e-06,
      "loss": 0.3178,
      "step": 36385
    },
    {
      "epoch": 1.6697719241888853,
      "grad_norm": 0.46900615096092224,
      "learning_rate": 4.2560528937046746e-06,
      "loss": 0.3558,
      "step": 36386
    },
    {
      "epoch": 1.6698178146941398,
      "grad_norm": 0.48367422819137573,
      "learning_rate": 4.2558104359024355e-06,
      "loss": 0.3386,
      "step": 36387
    },
    {
      "epoch": 1.6698637051993943,
      "grad_norm": 0.4695724844932556,
      "learning_rate": 4.2555679798897365e-06,
      "loss": 0.3169,
      "step": 36388
    },
    {
      "epoch": 1.6699095957046488,
      "grad_norm": 0.4959152042865753,
      "learning_rate": 4.255325525667162e-06,
      "loss": 0.3695,
      "step": 36389
    },
    {
      "epoch": 1.6699554862099033,
      "grad_norm": 0.5036804676055908,
      "learning_rate": 4.255083073235296e-06,
      "loss": 0.3928,
      "step": 36390
    },
    {
      "epoch": 1.6700013767151576,
      "grad_norm": 0.476836621761322,
      "learning_rate": 4.254840622594717e-06,
      "loss": 0.3718,
      "step": 36391
    },
    {
      "epoch": 1.670047267220412,
      "grad_norm": 0.4662763774394989,
      "learning_rate": 4.254598173746013e-06,
      "loss": 0.3415,
      "step": 36392
    },
    {
      "epoch": 1.6700931577256666,
      "grad_norm": 0.4697283208370209,
      "learning_rate": 4.254355726689765e-06,
      "loss": 0.3134,
      "step": 36393
    },
    {
      "epoch": 1.670139048230921,
      "grad_norm": 0.4429141879081726,
      "learning_rate": 4.2541132814265555e-06,
      "loss": 0.3212,
      "step": 36394
    },
    {
      "epoch": 1.6701849387361754,
      "grad_norm": 0.4392108619213104,
      "learning_rate": 4.253870837956969e-06,
      "loss": 0.2844,
      "step": 36395
    },
    {
      "epoch": 1.6702308292414298,
      "grad_norm": 0.4867120683193207,
      "learning_rate": 4.253628396281588e-06,
      "loss": 0.3862,
      "step": 36396
    },
    {
      "epoch": 1.6702767197466843,
      "grad_norm": 0.49286913871765137,
      "learning_rate": 4.253385956400993e-06,
      "loss": 0.3555,
      "step": 36397
    },
    {
      "epoch": 1.6703226102519388,
      "grad_norm": 0.44488343596458435,
      "learning_rate": 4.253143518315771e-06,
      "loss": 0.2991,
      "step": 36398
    },
    {
      "epoch": 1.6703685007571933,
      "grad_norm": 0.49190205335617065,
      "learning_rate": 4.2529010820265035e-06,
      "loss": 0.3736,
      "step": 36399
    },
    {
      "epoch": 1.6704143912624478,
      "grad_norm": 0.4606250524520874,
      "learning_rate": 4.252658647533772e-06,
      "loss": 0.384,
      "step": 36400
    },
    {
      "epoch": 1.6704602817677023,
      "grad_norm": 0.4886568486690521,
      "learning_rate": 4.252416214838162e-06,
      "loss": 0.393,
      "step": 36401
    },
    {
      "epoch": 1.6705061722729568,
      "grad_norm": 0.5127386450767517,
      "learning_rate": 4.252173783940254e-06,
      "loss": 0.4342,
      "step": 36402
    },
    {
      "epoch": 1.6705520627782113,
      "grad_norm": 0.5012575387954712,
      "learning_rate": 4.25193135484063e-06,
      "loss": 0.3985,
      "step": 36403
    },
    {
      "epoch": 1.6705979532834656,
      "grad_norm": 0.4899231493473053,
      "learning_rate": 4.251688927539879e-06,
      "loss": 0.3888,
      "step": 36404
    },
    {
      "epoch": 1.67064384378872,
      "grad_norm": 0.48674607276916504,
      "learning_rate": 4.251446502038579e-06,
      "loss": 0.3425,
      "step": 36405
    },
    {
      "epoch": 1.6706897342939746,
      "grad_norm": 0.5360870361328125,
      "learning_rate": 4.251204078337311e-06,
      "loss": 0.4129,
      "step": 36406
    },
    {
      "epoch": 1.6707356247992289,
      "grad_norm": 0.49139028787612915,
      "learning_rate": 4.250961656436663e-06,
      "loss": 0.3708,
      "step": 36407
    },
    {
      "epoch": 1.6707815153044834,
      "grad_norm": 0.43626487255096436,
      "learning_rate": 4.250719236337216e-06,
      "loss": 0.3377,
      "step": 36408
    },
    {
      "epoch": 1.6708274058097379,
      "grad_norm": 0.4327443540096283,
      "learning_rate": 4.250476818039552e-06,
      "loss": 0.2853,
      "step": 36409
    },
    {
      "epoch": 1.6708732963149924,
      "grad_norm": 0.49165529012680054,
      "learning_rate": 4.250234401544255e-06,
      "loss": 0.4109,
      "step": 36410
    },
    {
      "epoch": 1.6709191868202469,
      "grad_norm": 0.49875110387802124,
      "learning_rate": 4.249991986851909e-06,
      "loss": 0.4288,
      "step": 36411
    },
    {
      "epoch": 1.6709650773255014,
      "grad_norm": 0.4527752697467804,
      "learning_rate": 4.249749573963092e-06,
      "loss": 0.3532,
      "step": 36412
    },
    {
      "epoch": 1.6710109678307559,
      "grad_norm": 0.49607908725738525,
      "learning_rate": 4.2495071628783936e-06,
      "loss": 0.4127,
      "step": 36413
    },
    {
      "epoch": 1.6710568583360104,
      "grad_norm": 0.4266645014286041,
      "learning_rate": 4.249264753598394e-06,
      "loss": 0.2665,
      "step": 36414
    },
    {
      "epoch": 1.6711027488412649,
      "grad_norm": 0.45072200894355774,
      "learning_rate": 4.249022346123674e-06,
      "loss": 0.35,
      "step": 36415
    },
    {
      "epoch": 1.6711486393465194,
      "grad_norm": 0.5091655254364014,
      "learning_rate": 4.248779940454818e-06,
      "loss": 0.3738,
      "step": 36416
    },
    {
      "epoch": 1.6711945298517736,
      "grad_norm": 0.48325029015541077,
      "learning_rate": 4.2485375365924105e-06,
      "loss": 0.3964,
      "step": 36417
    },
    {
      "epoch": 1.6712404203570281,
      "grad_norm": 0.45113036036491394,
      "learning_rate": 4.248295134537032e-06,
      "loss": 0.3519,
      "step": 36418
    },
    {
      "epoch": 1.6712863108622826,
      "grad_norm": 0.4783253073692322,
      "learning_rate": 4.248052734289267e-06,
      "loss": 0.3684,
      "step": 36419
    },
    {
      "epoch": 1.671332201367537,
      "grad_norm": 0.4224016070365906,
      "learning_rate": 4.247810335849698e-06,
      "loss": 0.3012,
      "step": 36420
    },
    {
      "epoch": 1.6713780918727914,
      "grad_norm": 0.43874800205230713,
      "learning_rate": 4.247567939218908e-06,
      "loss": 0.3345,
      "step": 36421
    },
    {
      "epoch": 1.671423982378046,
      "grad_norm": 0.49213725328445435,
      "learning_rate": 4.247325544397478e-06,
      "loss": 0.3383,
      "step": 36422
    },
    {
      "epoch": 1.6714698728833004,
      "grad_norm": 0.47545698285102844,
      "learning_rate": 4.247083151385994e-06,
      "loss": 0.3738,
      "step": 36423
    },
    {
      "epoch": 1.671515763388555,
      "grad_norm": 0.5239338874816895,
      "learning_rate": 4.246840760185038e-06,
      "loss": 0.4821,
      "step": 36424
    },
    {
      "epoch": 1.6715616538938094,
      "grad_norm": 0.5066376328468323,
      "learning_rate": 4.246598370795189e-06,
      "loss": 0.3769,
      "step": 36425
    },
    {
      "epoch": 1.671607544399064,
      "grad_norm": 0.5015862584114075,
      "learning_rate": 4.246355983217038e-06,
      "loss": 0.4395,
      "step": 36426
    },
    {
      "epoch": 1.6716534349043184,
      "grad_norm": 0.4790043234825134,
      "learning_rate": 4.24611359745116e-06,
      "loss": 0.3731,
      "step": 36427
    },
    {
      "epoch": 1.671699325409573,
      "grad_norm": 0.5384829640388489,
      "learning_rate": 4.24587121349814e-06,
      "loss": 0.3323,
      "step": 36428
    },
    {
      "epoch": 1.6717452159148274,
      "grad_norm": 0.49120527505874634,
      "learning_rate": 4.245628831358564e-06,
      "loss": 0.3923,
      "step": 36429
    },
    {
      "epoch": 1.6717911064200817,
      "grad_norm": 0.45001354813575745,
      "learning_rate": 4.2453864510330115e-06,
      "loss": 0.3103,
      "step": 36430
    },
    {
      "epoch": 1.6718369969253362,
      "grad_norm": 0.4836060702800751,
      "learning_rate": 4.245144072522066e-06,
      "loss": 0.3925,
      "step": 36431
    },
    {
      "epoch": 1.6718828874305907,
      "grad_norm": 0.446071594953537,
      "learning_rate": 4.244901695826312e-06,
      "loss": 0.3004,
      "step": 36432
    },
    {
      "epoch": 1.671928777935845,
      "grad_norm": 0.45469948649406433,
      "learning_rate": 4.244659320946331e-06,
      "loss": 0.302,
      "step": 36433
    },
    {
      "epoch": 1.6719746684410994,
      "grad_norm": 0.5117505788803101,
      "learning_rate": 4.244416947882703e-06,
      "loss": 0.3725,
      "step": 36434
    },
    {
      "epoch": 1.672020558946354,
      "grad_norm": 0.468723863363266,
      "learning_rate": 4.244174576636017e-06,
      "loss": 0.3618,
      "step": 36435
    },
    {
      "epoch": 1.6720664494516084,
      "grad_norm": 0.4763523042201996,
      "learning_rate": 4.243932207206852e-06,
      "loss": 0.3774,
      "step": 36436
    },
    {
      "epoch": 1.672112339956863,
      "grad_norm": 0.4621734619140625,
      "learning_rate": 4.24368983959579e-06,
      "loss": 0.3366,
      "step": 36437
    },
    {
      "epoch": 1.6721582304621174,
      "grad_norm": 0.46940046548843384,
      "learning_rate": 4.243447473803418e-06,
      "loss": 0.3645,
      "step": 36438
    },
    {
      "epoch": 1.672204120967372,
      "grad_norm": 0.5135679244995117,
      "learning_rate": 4.243205109830314e-06,
      "loss": 0.4165,
      "step": 36439
    },
    {
      "epoch": 1.6722500114726264,
      "grad_norm": 0.42299628257751465,
      "learning_rate": 4.242962747677062e-06,
      "loss": 0.3182,
      "step": 36440
    },
    {
      "epoch": 1.672295901977881,
      "grad_norm": 0.47843003273010254,
      "learning_rate": 4.242720387344247e-06,
      "loss": 0.3624,
      "step": 36441
    },
    {
      "epoch": 1.6723417924831352,
      "grad_norm": 0.4899647831916809,
      "learning_rate": 4.242478028832451e-06,
      "loss": 0.4137,
      "step": 36442
    },
    {
      "epoch": 1.6723876829883897,
      "grad_norm": 0.45200687646865845,
      "learning_rate": 4.242235672142255e-06,
      "loss": 0.3074,
      "step": 36443
    },
    {
      "epoch": 1.6724335734936442,
      "grad_norm": 0.44975486397743225,
      "learning_rate": 4.241993317274244e-06,
      "loss": 0.3713,
      "step": 36444
    },
    {
      "epoch": 1.6724794639988985,
      "grad_norm": 0.47753891348838806,
      "learning_rate": 4.241750964229001e-06,
      "loss": 0.3791,
      "step": 36445
    },
    {
      "epoch": 1.672525354504153,
      "grad_norm": 0.49888068437576294,
      "learning_rate": 4.241508613007104e-06,
      "loss": 0.3942,
      "step": 36446
    },
    {
      "epoch": 1.6725712450094075,
      "grad_norm": 0.4564834237098694,
      "learning_rate": 4.241266263609142e-06,
      "loss": 0.3792,
      "step": 36447
    },
    {
      "epoch": 1.672617135514662,
      "grad_norm": 0.4786907732486725,
      "learning_rate": 4.241023916035696e-06,
      "loss": 0.3578,
      "step": 36448
    },
    {
      "epoch": 1.6726630260199165,
      "grad_norm": 0.5032262802124023,
      "learning_rate": 4.2407815702873474e-06,
      "loss": 0.3851,
      "step": 36449
    },
    {
      "epoch": 1.672708916525171,
      "grad_norm": 0.4884079694747925,
      "learning_rate": 4.240539226364677e-06,
      "loss": 0.366,
      "step": 36450
    },
    {
      "epoch": 1.6727548070304255,
      "grad_norm": 0.4401054084300995,
      "learning_rate": 4.240296884268272e-06,
      "loss": 0.3711,
      "step": 36451
    },
    {
      "epoch": 1.67280069753568,
      "grad_norm": 0.4238280653953552,
      "learning_rate": 4.240054543998714e-06,
      "loss": 0.2835,
      "step": 36452
    },
    {
      "epoch": 1.6728465880409344,
      "grad_norm": 0.4716249406337738,
      "learning_rate": 4.239812205556583e-06,
      "loss": 0.408,
      "step": 36453
    },
    {
      "epoch": 1.672892478546189,
      "grad_norm": 0.5285786986351013,
      "learning_rate": 4.239569868942466e-06,
      "loss": 0.3072,
      "step": 36454
    },
    {
      "epoch": 1.6729383690514432,
      "grad_norm": 0.45675376057624817,
      "learning_rate": 4.239327534156943e-06,
      "loss": 0.3523,
      "step": 36455
    },
    {
      "epoch": 1.6729842595566977,
      "grad_norm": 0.4795709550380707,
      "learning_rate": 4.239085201200594e-06,
      "loss": 0.3572,
      "step": 36456
    },
    {
      "epoch": 1.6730301500619522,
      "grad_norm": 0.5046348571777344,
      "learning_rate": 4.2388428700740085e-06,
      "loss": 0.413,
      "step": 36457
    },
    {
      "epoch": 1.6730760405672065,
      "grad_norm": 0.43525412678718567,
      "learning_rate": 4.238600540777765e-06,
      "loss": 0.3134,
      "step": 36458
    },
    {
      "epoch": 1.673121931072461,
      "grad_norm": 0.4625495374202728,
      "learning_rate": 4.238358213312446e-06,
      "loss": 0.3447,
      "step": 36459
    },
    {
      "epoch": 1.6731678215777155,
      "grad_norm": 0.49690213799476624,
      "learning_rate": 4.238115887678636e-06,
      "loss": 0.3657,
      "step": 36460
    },
    {
      "epoch": 1.67321371208297,
      "grad_norm": 0.46920424699783325,
      "learning_rate": 4.237873563876917e-06,
      "loss": 0.3871,
      "step": 36461
    },
    {
      "epoch": 1.6732596025882245,
      "grad_norm": 0.5055617690086365,
      "learning_rate": 4.237631241907869e-06,
      "loss": 0.3971,
      "step": 36462
    },
    {
      "epoch": 1.673305493093479,
      "grad_norm": 0.46127983927726746,
      "learning_rate": 4.23738892177208e-06,
      "loss": 0.3493,
      "step": 36463
    },
    {
      "epoch": 1.6733513835987335,
      "grad_norm": 0.4607318937778473,
      "learning_rate": 4.237146603470129e-06,
      "loss": 0.3142,
      "step": 36464
    },
    {
      "epoch": 1.673397274103988,
      "grad_norm": 0.4585719108581543,
      "learning_rate": 4.2369042870025996e-06,
      "loss": 0.3642,
      "step": 36465
    },
    {
      "epoch": 1.6734431646092425,
      "grad_norm": 0.4570196866989136,
      "learning_rate": 4.236661972370076e-06,
      "loss": 0.3219,
      "step": 36466
    },
    {
      "epoch": 1.673489055114497,
      "grad_norm": 0.4818902909755707,
      "learning_rate": 4.236419659573138e-06,
      "loss": 0.4103,
      "step": 36467
    },
    {
      "epoch": 1.6735349456197512,
      "grad_norm": 0.46282175183296204,
      "learning_rate": 4.236177348612368e-06,
      "loss": 0.3713,
      "step": 36468
    },
    {
      "epoch": 1.6735808361250057,
      "grad_norm": 0.48059719800949097,
      "learning_rate": 4.235935039488353e-06,
      "loss": 0.3556,
      "step": 36469
    },
    {
      "epoch": 1.6736267266302602,
      "grad_norm": 0.41766485571861267,
      "learning_rate": 4.235692732201674e-06,
      "loss": 0.243,
      "step": 36470
    },
    {
      "epoch": 1.6736726171355145,
      "grad_norm": 0.4788079559803009,
      "learning_rate": 4.23545042675291e-06,
      "loss": 0.3702,
      "step": 36471
    },
    {
      "epoch": 1.673718507640769,
      "grad_norm": 0.4729735255241394,
      "learning_rate": 4.235208123142648e-06,
      "loss": 0.3611,
      "step": 36472
    },
    {
      "epoch": 1.6737643981460235,
      "grad_norm": 0.5080539584159851,
      "learning_rate": 4.2349658213714685e-06,
      "loss": 0.4052,
      "step": 36473
    },
    {
      "epoch": 1.673810288651278,
      "grad_norm": 0.47496187686920166,
      "learning_rate": 4.234723521439954e-06,
      "loss": 0.3665,
      "step": 36474
    },
    {
      "epoch": 1.6738561791565325,
      "grad_norm": 0.48652276396751404,
      "learning_rate": 4.23448122334869e-06,
      "loss": 0.3991,
      "step": 36475
    },
    {
      "epoch": 1.673902069661787,
      "grad_norm": 0.461581826210022,
      "learning_rate": 4.234238927098256e-06,
      "loss": 0.3487,
      "step": 36476
    },
    {
      "epoch": 1.6739479601670415,
      "grad_norm": 0.4595324695110321,
      "learning_rate": 4.233996632689234e-06,
      "loss": 0.3121,
      "step": 36477
    },
    {
      "epoch": 1.673993850672296,
      "grad_norm": 0.4366900324821472,
      "learning_rate": 4.233754340122211e-06,
      "loss": 0.3316,
      "step": 36478
    },
    {
      "epoch": 1.6740397411775505,
      "grad_norm": 0.5074285864830017,
      "learning_rate": 4.233512049397766e-06,
      "loss": 0.4095,
      "step": 36479
    },
    {
      "epoch": 1.6740856316828048,
      "grad_norm": 0.5864399075508118,
      "learning_rate": 4.233269760516483e-06,
      "loss": 0.3881,
      "step": 36480
    },
    {
      "epoch": 1.6741315221880593,
      "grad_norm": 0.5089992880821228,
      "learning_rate": 4.233027473478944e-06,
      "loss": 0.3959,
      "step": 36481
    },
    {
      "epoch": 1.6741774126933138,
      "grad_norm": 0.45408928394317627,
      "learning_rate": 4.232785188285732e-06,
      "loss": 0.3136,
      "step": 36482
    },
    {
      "epoch": 1.6742233031985683,
      "grad_norm": 0.460478812456131,
      "learning_rate": 4.232542904937429e-06,
      "loss": 0.3516,
      "step": 36483
    },
    {
      "epoch": 1.6742691937038225,
      "grad_norm": 0.49170345067977905,
      "learning_rate": 4.232300623434616e-06,
      "loss": 0.4301,
      "step": 36484
    },
    {
      "epoch": 1.674315084209077,
      "grad_norm": 0.4590580463409424,
      "learning_rate": 4.232058343777881e-06,
      "loss": 0.3217,
      "step": 36485
    },
    {
      "epoch": 1.6743609747143315,
      "grad_norm": 0.46955394744873047,
      "learning_rate": 4.231816065967803e-06,
      "loss": 0.3372,
      "step": 36486
    },
    {
      "epoch": 1.674406865219586,
      "grad_norm": 0.47277143597602844,
      "learning_rate": 4.231573790004964e-06,
      "loss": 0.3314,
      "step": 36487
    },
    {
      "epoch": 1.6744527557248405,
      "grad_norm": 0.4322127103805542,
      "learning_rate": 4.2313315158899476e-06,
      "loss": 0.2644,
      "step": 36488
    },
    {
      "epoch": 1.674498646230095,
      "grad_norm": 0.44472944736480713,
      "learning_rate": 4.231089243623337e-06,
      "loss": 0.3432,
      "step": 36489
    },
    {
      "epoch": 1.6745445367353495,
      "grad_norm": 0.46111059188842773,
      "learning_rate": 4.2308469732057115e-06,
      "loss": 0.3654,
      "step": 36490
    },
    {
      "epoch": 1.674590427240604,
      "grad_norm": 0.4555404484272003,
      "learning_rate": 4.230604704637659e-06,
      "loss": 0.3479,
      "step": 36491
    },
    {
      "epoch": 1.6746363177458585,
      "grad_norm": 0.42896705865859985,
      "learning_rate": 4.23036243791976e-06,
      "loss": 0.3281,
      "step": 36492
    },
    {
      "epoch": 1.6746822082511128,
      "grad_norm": 0.4827493131160736,
      "learning_rate": 4.230120173052593e-06,
      "loss": 0.3391,
      "step": 36493
    },
    {
      "epoch": 1.6747280987563673,
      "grad_norm": 0.5437570810317993,
      "learning_rate": 4.2298779100367465e-06,
      "loss": 0.4725,
      "step": 36494
    },
    {
      "epoch": 1.6747739892616218,
      "grad_norm": 0.4679453670978546,
      "learning_rate": 4.2296356488728e-06,
      "loss": 0.3912,
      "step": 36495
    },
    {
      "epoch": 1.674819879766876,
      "grad_norm": 0.49863117933273315,
      "learning_rate": 4.229393389561337e-06,
      "loss": 0.4458,
      "step": 36496
    },
    {
      "epoch": 1.6748657702721306,
      "grad_norm": 0.47238606214523315,
      "learning_rate": 4.22915113210294e-06,
      "loss": 0.3721,
      "step": 36497
    },
    {
      "epoch": 1.674911660777385,
      "grad_norm": 0.47224247455596924,
      "learning_rate": 4.228908876498191e-06,
      "loss": 0.3289,
      "step": 36498
    },
    {
      "epoch": 1.6749575512826396,
      "grad_norm": 0.47474759817123413,
      "learning_rate": 4.228666622747671e-06,
      "loss": 0.2893,
      "step": 36499
    },
    {
      "epoch": 1.675003441787894,
      "grad_norm": 0.43323126435279846,
      "learning_rate": 4.228424370851966e-06,
      "loss": 0.3064,
      "step": 36500
    },
    {
      "epoch": 1.6750493322931486,
      "grad_norm": 0.4749133586883545,
      "learning_rate": 4.228182120811657e-06,
      "loss": 0.3569,
      "step": 36501
    },
    {
      "epoch": 1.675095222798403,
      "grad_norm": 0.5038188099861145,
      "learning_rate": 4.227939872627326e-06,
      "loss": 0.3965,
      "step": 36502
    },
    {
      "epoch": 1.6751411133036576,
      "grad_norm": 0.4963737428188324,
      "learning_rate": 4.2276976262995564e-06,
      "loss": 0.3473,
      "step": 36503
    },
    {
      "epoch": 1.675187003808912,
      "grad_norm": 0.4400983154773712,
      "learning_rate": 4.22745538182893e-06,
      "loss": 0.3113,
      "step": 36504
    },
    {
      "epoch": 1.6752328943141666,
      "grad_norm": 0.47862768173217773,
      "learning_rate": 4.2272131392160276e-06,
      "loss": 0.3621,
      "step": 36505
    },
    {
      "epoch": 1.6752787848194208,
      "grad_norm": 0.48373058438301086,
      "learning_rate": 4.226970898461436e-06,
      "loss": 0.4425,
      "step": 36506
    },
    {
      "epoch": 1.6753246753246753,
      "grad_norm": 0.48168671131134033,
      "learning_rate": 4.226728659565736e-06,
      "loss": 0.3945,
      "step": 36507
    },
    {
      "epoch": 1.6753705658299298,
      "grad_norm": 0.46919965744018555,
      "learning_rate": 4.226486422529508e-06,
      "loss": 0.3444,
      "step": 36508
    },
    {
      "epoch": 1.675416456335184,
      "grad_norm": 0.4852716326713562,
      "learning_rate": 4.226244187353337e-06,
      "loss": 0.4048,
      "step": 36509
    },
    {
      "epoch": 1.6754623468404386,
      "grad_norm": 0.45921608805656433,
      "learning_rate": 4.226001954037804e-06,
      "loss": 0.3193,
      "step": 36510
    },
    {
      "epoch": 1.675508237345693,
      "grad_norm": 0.45602890849113464,
      "learning_rate": 4.225759722583491e-06,
      "loss": 0.3498,
      "step": 36511
    },
    {
      "epoch": 1.6755541278509476,
      "grad_norm": 0.4616810381412506,
      "learning_rate": 4.225517492990983e-06,
      "loss": 0.3588,
      "step": 36512
    },
    {
      "epoch": 1.675600018356202,
      "grad_norm": 0.4362317621707916,
      "learning_rate": 4.225275265260861e-06,
      "loss": 0.2841,
      "step": 36513
    },
    {
      "epoch": 1.6756459088614566,
      "grad_norm": 0.4725174903869629,
      "learning_rate": 4.225033039393708e-06,
      "loss": 0.3518,
      "step": 36514
    },
    {
      "epoch": 1.675691799366711,
      "grad_norm": 0.437983900308609,
      "learning_rate": 4.224790815390104e-06,
      "loss": 0.2818,
      "step": 36515
    },
    {
      "epoch": 1.6757376898719656,
      "grad_norm": 0.48707646131515503,
      "learning_rate": 4.224548593250635e-06,
      "loss": 0.4123,
      "step": 36516
    },
    {
      "epoch": 1.67578358037722,
      "grad_norm": 0.4063223898410797,
      "learning_rate": 4.224306372975881e-06,
      "loss": 0.2543,
      "step": 36517
    },
    {
      "epoch": 1.6758294708824746,
      "grad_norm": 0.4906246066093445,
      "learning_rate": 4.2240641545664255e-06,
      "loss": 0.3504,
      "step": 36518
    },
    {
      "epoch": 1.6758753613877289,
      "grad_norm": 0.4598824381828308,
      "learning_rate": 4.223821938022851e-06,
      "loss": 0.3345,
      "step": 36519
    },
    {
      "epoch": 1.6759212518929834,
      "grad_norm": 0.5146784782409668,
      "learning_rate": 4.22357972334574e-06,
      "loss": 0.3948,
      "step": 36520
    },
    {
      "epoch": 1.6759671423982379,
      "grad_norm": 0.48112761974334717,
      "learning_rate": 4.2233375105356736e-06,
      "loss": 0.3725,
      "step": 36521
    },
    {
      "epoch": 1.6760130329034921,
      "grad_norm": 0.46727386116981506,
      "learning_rate": 4.223095299593236e-06,
      "loss": 0.3317,
      "step": 36522
    },
    {
      "epoch": 1.6760589234087466,
      "grad_norm": 0.47835224866867065,
      "learning_rate": 4.22285309051901e-06,
      "loss": 0.3741,
      "step": 36523
    },
    {
      "epoch": 1.6761048139140011,
      "grad_norm": 0.5095685720443726,
      "learning_rate": 4.222610883313576e-06,
      "loss": 0.4152,
      "step": 36524
    },
    {
      "epoch": 1.6761507044192556,
      "grad_norm": 0.44336411356925964,
      "learning_rate": 4.222368677977518e-06,
      "loss": 0.3456,
      "step": 36525
    },
    {
      "epoch": 1.6761965949245101,
      "grad_norm": 0.5016524791717529,
      "learning_rate": 4.222126474511418e-06,
      "loss": 0.4254,
      "step": 36526
    },
    {
      "epoch": 1.6762424854297646,
      "grad_norm": 0.464347779750824,
      "learning_rate": 4.221884272915857e-06,
      "loss": 0.3741,
      "step": 36527
    },
    {
      "epoch": 1.6762883759350191,
      "grad_norm": 0.45197755098342896,
      "learning_rate": 4.22164207319142e-06,
      "loss": 0.3,
      "step": 36528
    },
    {
      "epoch": 1.6763342664402736,
      "grad_norm": 0.4850073754787445,
      "learning_rate": 4.221399875338688e-06,
      "loss": 0.3815,
      "step": 36529
    },
    {
      "epoch": 1.676380156945528,
      "grad_norm": 0.45495352149009705,
      "learning_rate": 4.221157679358243e-06,
      "loss": 0.342,
      "step": 36530
    },
    {
      "epoch": 1.6764260474507824,
      "grad_norm": 0.46351560950279236,
      "learning_rate": 4.220915485250669e-06,
      "loss": 0.3318,
      "step": 36531
    },
    {
      "epoch": 1.6764719379560369,
      "grad_norm": 0.4632202386856079,
      "learning_rate": 4.220673293016547e-06,
      "loss": 0.3236,
      "step": 36532
    },
    {
      "epoch": 1.6765178284612914,
      "grad_norm": 0.44485148787498474,
      "learning_rate": 4.220431102656458e-06,
      "loss": 0.2849,
      "step": 36533
    },
    {
      "epoch": 1.6765637189665457,
      "grad_norm": 0.4886535704135895,
      "learning_rate": 4.220188914170988e-06,
      "loss": 0.41,
      "step": 36534
    },
    {
      "epoch": 1.6766096094718002,
      "grad_norm": 0.5439357757568359,
      "learning_rate": 4.219946727560718e-06,
      "loss": 0.3961,
      "step": 36535
    },
    {
      "epoch": 1.6766554999770547,
      "grad_norm": 0.4195284843444824,
      "learning_rate": 4.219704542826228e-06,
      "loss": 0.2581,
      "step": 36536
    },
    {
      "epoch": 1.6767013904823092,
      "grad_norm": 0.4579995274543762,
      "learning_rate": 4.219462359968105e-06,
      "loss": 0.3577,
      "step": 36537
    },
    {
      "epoch": 1.6767472809875636,
      "grad_norm": 0.4592399299144745,
      "learning_rate": 4.219220178986927e-06,
      "loss": 0.3338,
      "step": 36538
    },
    {
      "epoch": 1.6767931714928181,
      "grad_norm": 0.44765183329582214,
      "learning_rate": 4.2189779998832784e-06,
      "loss": 0.3213,
      "step": 36539
    },
    {
      "epoch": 1.6768390619980726,
      "grad_norm": 0.47102999687194824,
      "learning_rate": 4.218735822657741e-06,
      "loss": 0.3626,
      "step": 36540
    },
    {
      "epoch": 1.6768849525033271,
      "grad_norm": 0.4997973144054413,
      "learning_rate": 4.218493647310898e-06,
      "loss": 0.3614,
      "step": 36541
    },
    {
      "epoch": 1.6769308430085816,
      "grad_norm": 0.45439597964286804,
      "learning_rate": 4.218251473843329e-06,
      "loss": 0.3634,
      "step": 36542
    },
    {
      "epoch": 1.6769767335138361,
      "grad_norm": 0.4429270923137665,
      "learning_rate": 4.21800930225562e-06,
      "loss": 0.3147,
      "step": 36543
    },
    {
      "epoch": 1.6770226240190904,
      "grad_norm": 0.4388974606990814,
      "learning_rate": 4.2177671325483525e-06,
      "loss": 0.2994,
      "step": 36544
    },
    {
      "epoch": 1.677068514524345,
      "grad_norm": 0.4201744496822357,
      "learning_rate": 4.217524964722108e-06,
      "loss": 0.3031,
      "step": 36545
    },
    {
      "epoch": 1.6771144050295994,
      "grad_norm": 0.43443751335144043,
      "learning_rate": 4.217282798777468e-06,
      "loss": 0.323,
      "step": 36546
    },
    {
      "epoch": 1.6771602955348537,
      "grad_norm": 0.49964815378189087,
      "learning_rate": 4.217040634715016e-06,
      "loss": 0.3865,
      "step": 36547
    },
    {
      "epoch": 1.6772061860401082,
      "grad_norm": 0.48234280943870544,
      "learning_rate": 4.216798472535336e-06,
      "loss": 0.3611,
      "step": 36548
    },
    {
      "epoch": 1.6772520765453627,
      "grad_norm": 0.5201814770698547,
      "learning_rate": 4.216556312239005e-06,
      "loss": 0.3978,
      "step": 36549
    },
    {
      "epoch": 1.6772979670506172,
      "grad_norm": 0.44239118695259094,
      "learning_rate": 4.216314153826612e-06,
      "loss": 0.3046,
      "step": 36550
    },
    {
      "epoch": 1.6773438575558717,
      "grad_norm": 0.46834179759025574,
      "learning_rate": 4.216071997298735e-06,
      "loss": 0.4121,
      "step": 36551
    },
    {
      "epoch": 1.6773897480611262,
      "grad_norm": 0.422545850276947,
      "learning_rate": 4.215829842655957e-06,
      "loss": 0.2767,
      "step": 36552
    },
    {
      "epoch": 1.6774356385663807,
      "grad_norm": 0.4376762807369232,
      "learning_rate": 4.215587689898861e-06,
      "loss": 0.3155,
      "step": 36553
    },
    {
      "epoch": 1.6774815290716352,
      "grad_norm": 0.49101999402046204,
      "learning_rate": 4.215345539028029e-06,
      "loss": 0.4726,
      "step": 36554
    },
    {
      "epoch": 1.6775274195768897,
      "grad_norm": 0.4575490355491638,
      "learning_rate": 4.215103390044043e-06,
      "loss": 0.3315,
      "step": 36555
    },
    {
      "epoch": 1.6775733100821442,
      "grad_norm": 0.42061564326286316,
      "learning_rate": 4.214861242947486e-06,
      "loss": 0.2793,
      "step": 36556
    },
    {
      "epoch": 1.6776192005873984,
      "grad_norm": 0.5177093148231506,
      "learning_rate": 4.214619097738941e-06,
      "loss": 0.3703,
      "step": 36557
    },
    {
      "epoch": 1.677665091092653,
      "grad_norm": 0.4535965919494629,
      "learning_rate": 4.214376954418987e-06,
      "loss": 0.3034,
      "step": 36558
    },
    {
      "epoch": 1.6777109815979074,
      "grad_norm": 0.4873185157775879,
      "learning_rate": 4.214134812988211e-06,
      "loss": 0.4155,
      "step": 36559
    },
    {
      "epoch": 1.6777568721031617,
      "grad_norm": 0.45540252327919006,
      "learning_rate": 4.2138926734471915e-06,
      "loss": 0.3303,
      "step": 36560
    },
    {
      "epoch": 1.6778027626084162,
      "grad_norm": 0.44114428758621216,
      "learning_rate": 4.2136505357965115e-06,
      "loss": 0.3401,
      "step": 36561
    },
    {
      "epoch": 1.6778486531136707,
      "grad_norm": 0.527583658695221,
      "learning_rate": 4.213408400036755e-06,
      "loss": 0.4142,
      "step": 36562
    },
    {
      "epoch": 1.6778945436189252,
      "grad_norm": 0.4526752531528473,
      "learning_rate": 4.213166266168503e-06,
      "loss": 0.303,
      "step": 36563
    },
    {
      "epoch": 1.6779404341241797,
      "grad_norm": 0.44314995408058167,
      "learning_rate": 4.2129241341923345e-06,
      "loss": 0.3176,
      "step": 36564
    },
    {
      "epoch": 1.6779863246294342,
      "grad_norm": 0.441885769367218,
      "learning_rate": 4.2126820041088385e-06,
      "loss": 0.3276,
      "step": 36565
    },
    {
      "epoch": 1.6780322151346887,
      "grad_norm": 0.4680526554584503,
      "learning_rate": 4.212439875918593e-06,
      "loss": 0.3655,
      "step": 36566
    },
    {
      "epoch": 1.6780781056399432,
      "grad_norm": 0.4734300971031189,
      "learning_rate": 4.212197749622181e-06,
      "loss": 0.3358,
      "step": 36567
    },
    {
      "epoch": 1.6781239961451977,
      "grad_norm": 0.4728740155696869,
      "learning_rate": 4.2119556252201845e-06,
      "loss": 0.3522,
      "step": 36568
    },
    {
      "epoch": 1.678169886650452,
      "grad_norm": 0.45828431844711304,
      "learning_rate": 4.211713502713187e-06,
      "loss": 0.3324,
      "step": 36569
    },
    {
      "epoch": 1.6782157771557065,
      "grad_norm": 0.6216670274734497,
      "learning_rate": 4.211471382101768e-06,
      "loss": 0.4285,
      "step": 36570
    },
    {
      "epoch": 1.678261667660961,
      "grad_norm": 0.47429198026657104,
      "learning_rate": 4.2112292633865125e-06,
      "loss": 0.409,
      "step": 36571
    },
    {
      "epoch": 1.6783075581662155,
      "grad_norm": 0.5526126623153687,
      "learning_rate": 4.210987146568002e-06,
      "loss": 0.3891,
      "step": 36572
    },
    {
      "epoch": 1.6783534486714697,
      "grad_norm": 0.5467831492424011,
      "learning_rate": 4.210745031646817e-06,
      "loss": 0.4537,
      "step": 36573
    },
    {
      "epoch": 1.6783993391767242,
      "grad_norm": 0.48331567645072937,
      "learning_rate": 4.210502918623543e-06,
      "loss": 0.3883,
      "step": 36574
    },
    {
      "epoch": 1.6784452296819787,
      "grad_norm": 0.4123234748840332,
      "learning_rate": 4.21026080749876e-06,
      "loss": 0.2909,
      "step": 36575
    },
    {
      "epoch": 1.6784911201872332,
      "grad_norm": 0.4490203559398651,
      "learning_rate": 4.210018698273048e-06,
      "loss": 0.3628,
      "step": 36576
    },
    {
      "epoch": 1.6785370106924877,
      "grad_norm": 0.5207685232162476,
      "learning_rate": 4.209776590946994e-06,
      "loss": 0.4364,
      "step": 36577
    },
    {
      "epoch": 1.6785829011977422,
      "grad_norm": 0.4506654143333435,
      "learning_rate": 4.209534485521178e-06,
      "loss": 0.3241,
      "step": 36578
    },
    {
      "epoch": 1.6786287917029967,
      "grad_norm": 0.5041995644569397,
      "learning_rate": 4.209292381996183e-06,
      "loss": 0.421,
      "step": 36579
    },
    {
      "epoch": 1.6786746822082512,
      "grad_norm": 0.46286600828170776,
      "learning_rate": 4.209050280372587e-06,
      "loss": 0.337,
      "step": 36580
    },
    {
      "epoch": 1.6787205727135057,
      "grad_norm": 0.4742344617843628,
      "learning_rate": 4.2088081806509786e-06,
      "loss": 0.3774,
      "step": 36581
    },
    {
      "epoch": 1.67876646321876,
      "grad_norm": 0.44923317432403564,
      "learning_rate": 4.208566082831936e-06,
      "loss": 0.3547,
      "step": 36582
    },
    {
      "epoch": 1.6788123537240145,
      "grad_norm": 0.46849343180656433,
      "learning_rate": 4.208323986916041e-06,
      "loss": 0.3245,
      "step": 36583
    },
    {
      "epoch": 1.678858244229269,
      "grad_norm": 0.5013726353645325,
      "learning_rate": 4.208081892903878e-06,
      "loss": 0.3793,
      "step": 36584
    },
    {
      "epoch": 1.6789041347345233,
      "grad_norm": 0.46768638491630554,
      "learning_rate": 4.207839800796029e-06,
      "loss": 0.3715,
      "step": 36585
    },
    {
      "epoch": 1.6789500252397778,
      "grad_norm": 0.4959128499031067,
      "learning_rate": 4.207597710593072e-06,
      "loss": 0.3738,
      "step": 36586
    },
    {
      "epoch": 1.6789959157450323,
      "grad_norm": 0.4266280233860016,
      "learning_rate": 4.207355622295595e-06,
      "loss": 0.2903,
      "step": 36587
    },
    {
      "epoch": 1.6790418062502868,
      "grad_norm": 0.4653206765651703,
      "learning_rate": 4.207113535904177e-06,
      "loss": 0.3206,
      "step": 36588
    },
    {
      "epoch": 1.6790876967555413,
      "grad_norm": 0.40866971015930176,
      "learning_rate": 4.2068714514194e-06,
      "loss": 0.2627,
      "step": 36589
    },
    {
      "epoch": 1.6791335872607958,
      "grad_norm": 0.4527736008167267,
      "learning_rate": 4.206629368841849e-06,
      "loss": 0.3282,
      "step": 36590
    },
    {
      "epoch": 1.6791794777660503,
      "grad_norm": 0.474046915769577,
      "learning_rate": 4.206387288172102e-06,
      "loss": 0.3646,
      "step": 36591
    },
    {
      "epoch": 1.6792253682713048,
      "grad_norm": 0.469391405582428,
      "learning_rate": 4.206145209410743e-06,
      "loss": 0.3528,
      "step": 36592
    },
    {
      "epoch": 1.6792712587765593,
      "grad_norm": 0.4496285915374756,
      "learning_rate": 4.2059031325583546e-06,
      "loss": 0.3261,
      "step": 36593
    },
    {
      "epoch": 1.6793171492818137,
      "grad_norm": 0.4801749587059021,
      "learning_rate": 4.20566105761552e-06,
      "loss": 0.3818,
      "step": 36594
    },
    {
      "epoch": 1.679363039787068,
      "grad_norm": 0.46251434087753296,
      "learning_rate": 4.205418984582818e-06,
      "loss": 0.3504,
      "step": 36595
    },
    {
      "epoch": 1.6794089302923225,
      "grad_norm": 0.4858919084072113,
      "learning_rate": 4.205176913460833e-06,
      "loss": 0.4285,
      "step": 36596
    },
    {
      "epoch": 1.679454820797577,
      "grad_norm": 0.509089469909668,
      "learning_rate": 4.204934844250148e-06,
      "loss": 0.49,
      "step": 36597
    },
    {
      "epoch": 1.6795007113028313,
      "grad_norm": 0.48630544543266296,
      "learning_rate": 4.204692776951341e-06,
      "loss": 0.3693,
      "step": 36598
    },
    {
      "epoch": 1.6795466018080858,
      "grad_norm": 0.4375717043876648,
      "learning_rate": 4.204450711564999e-06,
      "loss": 0.2816,
      "step": 36599
    },
    {
      "epoch": 1.6795924923133403,
      "grad_norm": 0.45944130420684814,
      "learning_rate": 4.204208648091702e-06,
      "loss": 0.3363,
      "step": 36600
    },
    {
      "epoch": 1.6796383828185948,
      "grad_norm": 0.4906388819217682,
      "learning_rate": 4.20396658653203e-06,
      "loss": 0.3919,
      "step": 36601
    },
    {
      "epoch": 1.6796842733238493,
      "grad_norm": 0.496921569108963,
      "learning_rate": 4.20372452688657e-06,
      "loss": 0.3781,
      "step": 36602
    },
    {
      "epoch": 1.6797301638291038,
      "grad_norm": 0.4289204478263855,
      "learning_rate": 4.203482469155901e-06,
      "loss": 0.2901,
      "step": 36603
    },
    {
      "epoch": 1.6797760543343583,
      "grad_norm": 0.54576575756073,
      "learning_rate": 4.203240413340602e-06,
      "loss": 0.4721,
      "step": 36604
    },
    {
      "epoch": 1.6798219448396128,
      "grad_norm": 0.5237581729888916,
      "learning_rate": 4.202998359441262e-06,
      "loss": 0.457,
      "step": 36605
    },
    {
      "epoch": 1.6798678353448673,
      "grad_norm": 0.4975808560848236,
      "learning_rate": 4.202756307458458e-06,
      "loss": 0.3955,
      "step": 36606
    },
    {
      "epoch": 1.6799137258501218,
      "grad_norm": 0.47764453291893005,
      "learning_rate": 4.202514257392772e-06,
      "loss": 0.3387,
      "step": 36607
    },
    {
      "epoch": 1.679959616355376,
      "grad_norm": 0.4803623557090759,
      "learning_rate": 4.202272209244789e-06,
      "loss": 0.3959,
      "step": 36608
    },
    {
      "epoch": 1.6800055068606305,
      "grad_norm": 0.4637603461742401,
      "learning_rate": 4.20203016301509e-06,
      "loss": 0.3679,
      "step": 36609
    },
    {
      "epoch": 1.680051397365885,
      "grad_norm": 0.48946255445480347,
      "learning_rate": 4.201788118704255e-06,
      "loss": 0.4384,
      "step": 36610
    },
    {
      "epoch": 1.6800972878711393,
      "grad_norm": 0.4636344909667969,
      "learning_rate": 4.20154607631287e-06,
      "loss": 0.3614,
      "step": 36611
    },
    {
      "epoch": 1.6801431783763938,
      "grad_norm": 0.4655325710773468,
      "learning_rate": 4.201304035841514e-06,
      "loss": 0.3295,
      "step": 36612
    },
    {
      "epoch": 1.6801890688816483,
      "grad_norm": 0.43468889594078064,
      "learning_rate": 4.20106199729077e-06,
      "loss": 0.3045,
      "step": 36613
    },
    {
      "epoch": 1.6802349593869028,
      "grad_norm": 0.4637826979160309,
      "learning_rate": 4.200819960661217e-06,
      "loss": 0.3618,
      "step": 36614
    },
    {
      "epoch": 1.6802808498921573,
      "grad_norm": 0.47906258702278137,
      "learning_rate": 4.200577925953442e-06,
      "loss": 0.375,
      "step": 36615
    },
    {
      "epoch": 1.6803267403974118,
      "grad_norm": 0.43818292021751404,
      "learning_rate": 4.200335893168024e-06,
      "loss": 0.2931,
      "step": 36616
    },
    {
      "epoch": 1.6803726309026663,
      "grad_norm": 0.4494995176792145,
      "learning_rate": 4.200093862305546e-06,
      "loss": 0.3417,
      "step": 36617
    },
    {
      "epoch": 1.6804185214079208,
      "grad_norm": 0.43340763449668884,
      "learning_rate": 4.19985183336659e-06,
      "loss": 0.2826,
      "step": 36618
    },
    {
      "epoch": 1.6804644119131753,
      "grad_norm": 0.47955751419067383,
      "learning_rate": 4.199609806351739e-06,
      "loss": 0.42,
      "step": 36619
    },
    {
      "epoch": 1.6805103024184296,
      "grad_norm": 0.5102643966674805,
      "learning_rate": 4.1993677812615704e-06,
      "loss": 0.4368,
      "step": 36620
    },
    {
      "epoch": 1.680556192923684,
      "grad_norm": 0.4862624406814575,
      "learning_rate": 4.1991257580966725e-06,
      "loss": 0.3554,
      "step": 36621
    },
    {
      "epoch": 1.6806020834289386,
      "grad_norm": 0.4391551613807678,
      "learning_rate": 4.198883736857624e-06,
      "loss": 0.2856,
      "step": 36622
    },
    {
      "epoch": 1.6806479739341929,
      "grad_norm": 0.4572699964046478,
      "learning_rate": 4.198641717545007e-06,
      "loss": 0.3449,
      "step": 36623
    },
    {
      "epoch": 1.6806938644394473,
      "grad_norm": 0.45318472385406494,
      "learning_rate": 4.198399700159404e-06,
      "loss": 0.2873,
      "step": 36624
    },
    {
      "epoch": 1.6807397549447018,
      "grad_norm": 0.45232975482940674,
      "learning_rate": 4.198157684701397e-06,
      "loss": 0.2846,
      "step": 36625
    },
    {
      "epoch": 1.6807856454499563,
      "grad_norm": 0.49212661385536194,
      "learning_rate": 4.1979156711715655e-06,
      "loss": 0.3748,
      "step": 36626
    },
    {
      "epoch": 1.6808315359552108,
      "grad_norm": 0.4476878345012665,
      "learning_rate": 4.197673659570497e-06,
      "loss": 0.3439,
      "step": 36627
    },
    {
      "epoch": 1.6808774264604653,
      "grad_norm": 0.46549829840660095,
      "learning_rate": 4.197431649898769e-06,
      "loss": 0.3511,
      "step": 36628
    },
    {
      "epoch": 1.6809233169657198,
      "grad_norm": 0.4481625258922577,
      "learning_rate": 4.197189642156962e-06,
      "loss": 0.2946,
      "step": 36629
    },
    {
      "epoch": 1.6809692074709743,
      "grad_norm": 0.44091692566871643,
      "learning_rate": 4.196947636345663e-06,
      "loss": 0.2857,
      "step": 36630
    },
    {
      "epoch": 1.6810150979762288,
      "grad_norm": 0.4902280271053314,
      "learning_rate": 4.1967056324654515e-06,
      "loss": 0.4617,
      "step": 36631
    },
    {
      "epoch": 1.6810609884814833,
      "grad_norm": 0.41542696952819824,
      "learning_rate": 4.196463630516909e-06,
      "loss": 0.2866,
      "step": 36632
    },
    {
      "epoch": 1.6811068789867376,
      "grad_norm": 0.44120338559150696,
      "learning_rate": 4.196221630500617e-06,
      "loss": 0.3132,
      "step": 36633
    },
    {
      "epoch": 1.681152769491992,
      "grad_norm": 0.42308467626571655,
      "learning_rate": 4.19597963241716e-06,
      "loss": 0.29,
      "step": 36634
    },
    {
      "epoch": 1.6811986599972466,
      "grad_norm": 0.46878519654273987,
      "learning_rate": 4.195737636267116e-06,
      "loss": 0.3561,
      "step": 36635
    },
    {
      "epoch": 1.6812445505025009,
      "grad_norm": 0.4856399893760681,
      "learning_rate": 4.195495642051071e-06,
      "loss": 0.3465,
      "step": 36636
    },
    {
      "epoch": 1.6812904410077554,
      "grad_norm": 0.43609169125556946,
      "learning_rate": 4.195253649769605e-06,
      "loss": 0.2943,
      "step": 36637
    },
    {
      "epoch": 1.6813363315130099,
      "grad_norm": 0.4336220622062683,
      "learning_rate": 4.195011659423299e-06,
      "loss": 0.2822,
      "step": 36638
    },
    {
      "epoch": 1.6813822220182644,
      "grad_norm": 0.4066895842552185,
      "learning_rate": 4.1947696710127375e-06,
      "loss": 0.2661,
      "step": 36639
    },
    {
      "epoch": 1.6814281125235189,
      "grad_norm": 0.43248075246810913,
      "learning_rate": 4.1945276845385e-06,
      "loss": 0.297,
      "step": 36640
    },
    {
      "epoch": 1.6814740030287734,
      "grad_norm": 0.4402267634868622,
      "learning_rate": 4.194285700001167e-06,
      "loss": 0.305,
      "step": 36641
    },
    {
      "epoch": 1.6815198935340279,
      "grad_norm": 0.5321704745292664,
      "learning_rate": 4.194043717401325e-06,
      "loss": 0.4413,
      "step": 36642
    },
    {
      "epoch": 1.6815657840392824,
      "grad_norm": 0.43868669867515564,
      "learning_rate": 4.193801736739554e-06,
      "loss": 0.2725,
      "step": 36643
    },
    {
      "epoch": 1.6816116745445369,
      "grad_norm": 0.44483572244644165,
      "learning_rate": 4.193559758016435e-06,
      "loss": 0.3074,
      "step": 36644
    },
    {
      "epoch": 1.6816575650497914,
      "grad_norm": 0.46181023120880127,
      "learning_rate": 4.193317781232549e-06,
      "loss": 0.3242,
      "step": 36645
    },
    {
      "epoch": 1.6817034555550456,
      "grad_norm": 0.46997594833374023,
      "learning_rate": 4.19307580638848e-06,
      "loss": 0.3778,
      "step": 36646
    },
    {
      "epoch": 1.6817493460603001,
      "grad_norm": 0.531737208366394,
      "learning_rate": 4.19283383348481e-06,
      "loss": 0.3996,
      "step": 36647
    },
    {
      "epoch": 1.6817952365655546,
      "grad_norm": 0.4286327660083771,
      "learning_rate": 4.192591862522117e-06,
      "loss": 0.3108,
      "step": 36648
    },
    {
      "epoch": 1.681841127070809,
      "grad_norm": 0.46070733666419983,
      "learning_rate": 4.192349893500989e-06,
      "loss": 0.3984,
      "step": 36649
    },
    {
      "epoch": 1.6818870175760634,
      "grad_norm": 0.44115087389945984,
      "learning_rate": 4.192107926422003e-06,
      "loss": 0.2904,
      "step": 36650
    },
    {
      "epoch": 1.681932908081318,
      "grad_norm": 0.5475233793258667,
      "learning_rate": 4.191865961285742e-06,
      "loss": 0.37,
      "step": 36651
    },
    {
      "epoch": 1.6819787985865724,
      "grad_norm": 0.5091083645820618,
      "learning_rate": 4.191623998092789e-06,
      "loss": 0.4183,
      "step": 36652
    },
    {
      "epoch": 1.682024689091827,
      "grad_norm": 0.4824158847332001,
      "learning_rate": 4.191382036843725e-06,
      "loss": 0.3679,
      "step": 36653
    },
    {
      "epoch": 1.6820705795970814,
      "grad_norm": 0.473499596118927,
      "learning_rate": 4.191140077539131e-06,
      "loss": 0.3872,
      "step": 36654
    },
    {
      "epoch": 1.682116470102336,
      "grad_norm": 0.4657679796218872,
      "learning_rate": 4.190898120179592e-06,
      "loss": 0.3078,
      "step": 36655
    },
    {
      "epoch": 1.6821623606075904,
      "grad_norm": 0.449124813079834,
      "learning_rate": 4.190656164765687e-06,
      "loss": 0.3552,
      "step": 36656
    },
    {
      "epoch": 1.6822082511128449,
      "grad_norm": 0.44133663177490234,
      "learning_rate": 4.190414211297996e-06,
      "loss": 0.2953,
      "step": 36657
    },
    {
      "epoch": 1.6822541416180992,
      "grad_norm": 0.49887338280677795,
      "learning_rate": 4.190172259777106e-06,
      "loss": 0.3696,
      "step": 36658
    },
    {
      "epoch": 1.6823000321233537,
      "grad_norm": 0.46963536739349365,
      "learning_rate": 4.189930310203596e-06,
      "loss": 0.3009,
      "step": 36659
    },
    {
      "epoch": 1.6823459226286082,
      "grad_norm": 0.4850354492664337,
      "learning_rate": 4.1896883625780464e-06,
      "loss": 0.3766,
      "step": 36660
    },
    {
      "epoch": 1.6823918131338627,
      "grad_norm": 0.4611857533454895,
      "learning_rate": 4.189446416901042e-06,
      "loss": 0.3602,
      "step": 36661
    },
    {
      "epoch": 1.682437703639117,
      "grad_norm": 0.4727003276348114,
      "learning_rate": 4.189204473173163e-06,
      "loss": 0.3614,
      "step": 36662
    },
    {
      "epoch": 1.6824835941443714,
      "grad_norm": 0.43927904963493347,
      "learning_rate": 4.18896253139499e-06,
      "loss": 0.2916,
      "step": 36663
    },
    {
      "epoch": 1.682529484649626,
      "grad_norm": 0.47729185223579407,
      "learning_rate": 4.188720591567107e-06,
      "loss": 0.3871,
      "step": 36664
    },
    {
      "epoch": 1.6825753751548804,
      "grad_norm": 0.47336286306381226,
      "learning_rate": 4.1884786536900966e-06,
      "loss": 0.379,
      "step": 36665
    },
    {
      "epoch": 1.682621265660135,
      "grad_norm": 0.49842989444732666,
      "learning_rate": 4.188236717764536e-06,
      "loss": 0.3521,
      "step": 36666
    },
    {
      "epoch": 1.6826671561653894,
      "grad_norm": 0.4778773784637451,
      "learning_rate": 4.187994783791012e-06,
      "loss": 0.3877,
      "step": 36667
    },
    {
      "epoch": 1.682713046670644,
      "grad_norm": 0.48494231700897217,
      "learning_rate": 4.187752851770104e-06,
      "loss": 0.4289,
      "step": 36668
    },
    {
      "epoch": 1.6827589371758984,
      "grad_norm": 0.5036440491676331,
      "learning_rate": 4.187510921702392e-06,
      "loss": 0.4221,
      "step": 36669
    },
    {
      "epoch": 1.682804827681153,
      "grad_norm": 0.5084603428840637,
      "learning_rate": 4.187268993588462e-06,
      "loss": 0.4015,
      "step": 36670
    },
    {
      "epoch": 1.6828507181864072,
      "grad_norm": 0.49309879541397095,
      "learning_rate": 4.187027067428894e-06,
      "loss": 0.4272,
      "step": 36671
    },
    {
      "epoch": 1.6828966086916617,
      "grad_norm": 0.46676939725875854,
      "learning_rate": 4.186785143224267e-06,
      "loss": 0.356,
      "step": 36672
    },
    {
      "epoch": 1.6829424991969162,
      "grad_norm": 0.4170481860637665,
      "learning_rate": 4.186543220975166e-06,
      "loss": 0.2735,
      "step": 36673
    },
    {
      "epoch": 1.6829883897021705,
      "grad_norm": 0.47602564096450806,
      "learning_rate": 4.186301300682173e-06,
      "loss": 0.3743,
      "step": 36674
    },
    {
      "epoch": 1.683034280207425,
      "grad_norm": 0.46793580055236816,
      "learning_rate": 4.186059382345868e-06,
      "loss": 0.311,
      "step": 36675
    },
    {
      "epoch": 1.6830801707126795,
      "grad_norm": 0.49261555075645447,
      "learning_rate": 4.185817465966833e-06,
      "loss": 0.3774,
      "step": 36676
    },
    {
      "epoch": 1.683126061217934,
      "grad_norm": 0.5102073550224304,
      "learning_rate": 4.185575551545651e-06,
      "loss": 0.3542,
      "step": 36677
    },
    {
      "epoch": 1.6831719517231885,
      "grad_norm": 0.4897017478942871,
      "learning_rate": 4.185333639082903e-06,
      "loss": 0.3907,
      "step": 36678
    },
    {
      "epoch": 1.683217842228443,
      "grad_norm": 0.5146334171295166,
      "learning_rate": 4.185091728579167e-06,
      "loss": 0.4676,
      "step": 36679
    },
    {
      "epoch": 1.6832637327336974,
      "grad_norm": 0.4781786799430847,
      "learning_rate": 4.184849820035031e-06,
      "loss": 0.3661,
      "step": 36680
    },
    {
      "epoch": 1.683309623238952,
      "grad_norm": 0.48679864406585693,
      "learning_rate": 4.184607913451074e-06,
      "loss": 0.352,
      "step": 36681
    },
    {
      "epoch": 1.6833555137442064,
      "grad_norm": 0.4676544964313507,
      "learning_rate": 4.184366008827876e-06,
      "loss": 0.3276,
      "step": 36682
    },
    {
      "epoch": 1.683401404249461,
      "grad_norm": 0.4873086214065552,
      "learning_rate": 4.184124106166022e-06,
      "loss": 0.3772,
      "step": 36683
    },
    {
      "epoch": 1.6834472947547152,
      "grad_norm": 0.4920465648174286,
      "learning_rate": 4.183882205466092e-06,
      "loss": 0.416,
      "step": 36684
    },
    {
      "epoch": 1.6834931852599697,
      "grad_norm": 0.46206143498420715,
      "learning_rate": 4.1836403067286655e-06,
      "loss": 0.3562,
      "step": 36685
    },
    {
      "epoch": 1.6835390757652242,
      "grad_norm": 0.4901551604270935,
      "learning_rate": 4.183398409954328e-06,
      "loss": 0.4383,
      "step": 36686
    },
    {
      "epoch": 1.6835849662704785,
      "grad_norm": 0.4095803201198578,
      "learning_rate": 4.18315651514366e-06,
      "loss": 0.2852,
      "step": 36687
    },
    {
      "epoch": 1.683630856775733,
      "grad_norm": 0.4333679974079132,
      "learning_rate": 4.182914622297243e-06,
      "loss": 0.3068,
      "step": 36688
    },
    {
      "epoch": 1.6836767472809875,
      "grad_norm": 0.505587100982666,
      "learning_rate": 4.182672731415658e-06,
      "loss": 0.4101,
      "step": 36689
    },
    {
      "epoch": 1.683722637786242,
      "grad_norm": 0.5026230216026306,
      "learning_rate": 4.1824308424994875e-06,
      "loss": 0.4129,
      "step": 36690
    },
    {
      "epoch": 1.6837685282914965,
      "grad_norm": 0.47114697098731995,
      "learning_rate": 4.18218895554931e-06,
      "loss": 0.3705,
      "step": 36691
    },
    {
      "epoch": 1.683814418796751,
      "grad_norm": 0.4381970763206482,
      "learning_rate": 4.1819470705657126e-06,
      "loss": 0.3196,
      "step": 36692
    },
    {
      "epoch": 1.6838603093020055,
      "grad_norm": 0.46089982986450195,
      "learning_rate": 4.181705187549276e-06,
      "loss": 0.3551,
      "step": 36693
    },
    {
      "epoch": 1.68390619980726,
      "grad_norm": 0.4817427098751068,
      "learning_rate": 4.181463306500576e-06,
      "loss": 0.3567,
      "step": 36694
    },
    {
      "epoch": 1.6839520903125145,
      "grad_norm": 0.47132623195648193,
      "learning_rate": 4.181221427420201e-06,
      "loss": 0.3701,
      "step": 36695
    },
    {
      "epoch": 1.683997980817769,
      "grad_norm": 0.45564770698547363,
      "learning_rate": 4.18097955030873e-06,
      "loss": 0.4068,
      "step": 36696
    },
    {
      "epoch": 1.6840438713230232,
      "grad_norm": 0.5715989470481873,
      "learning_rate": 4.1807376751667435e-06,
      "loss": 0.3548,
      "step": 36697
    },
    {
      "epoch": 1.6840897618282777,
      "grad_norm": 0.5544700026512146,
      "learning_rate": 4.180495801994826e-06,
      "loss": 0.4487,
      "step": 36698
    },
    {
      "epoch": 1.6841356523335322,
      "grad_norm": 0.49110493063926697,
      "learning_rate": 4.180253930793557e-06,
      "loss": 0.3573,
      "step": 36699
    },
    {
      "epoch": 1.6841815428387865,
      "grad_norm": 0.47974082827568054,
      "learning_rate": 4.180012061563517e-06,
      "loss": 0.4064,
      "step": 36700
    },
    {
      "epoch": 1.684227433344041,
      "grad_norm": 0.48985666036605835,
      "learning_rate": 4.1797701943052904e-06,
      "loss": 0.4135,
      "step": 36701
    },
    {
      "epoch": 1.6842733238492955,
      "grad_norm": 0.43646353483200073,
      "learning_rate": 4.179528329019458e-06,
      "loss": 0.2989,
      "step": 36702
    },
    {
      "epoch": 1.68431921435455,
      "grad_norm": 0.4515102505683899,
      "learning_rate": 4.1792864657066e-06,
      "loss": 0.3387,
      "step": 36703
    },
    {
      "epoch": 1.6843651048598045,
      "grad_norm": 0.473971426486969,
      "learning_rate": 4.1790446043673e-06,
      "loss": 0.3517,
      "step": 36704
    },
    {
      "epoch": 1.684410995365059,
      "grad_norm": 0.46834880113601685,
      "learning_rate": 4.178802745002139e-06,
      "loss": 0.3488,
      "step": 36705
    },
    {
      "epoch": 1.6844568858703135,
      "grad_norm": 0.4526287913322449,
      "learning_rate": 4.178560887611696e-06,
      "loss": 0.3193,
      "step": 36706
    },
    {
      "epoch": 1.684502776375568,
      "grad_norm": 0.45599237084388733,
      "learning_rate": 4.178319032196556e-06,
      "loss": 0.328,
      "step": 36707
    },
    {
      "epoch": 1.6845486668808225,
      "grad_norm": 0.4968269169330597,
      "learning_rate": 4.178077178757301e-06,
      "loss": 0.3561,
      "step": 36708
    },
    {
      "epoch": 1.6845945573860768,
      "grad_norm": 0.496565043926239,
      "learning_rate": 4.177835327294509e-06,
      "loss": 0.3738,
      "step": 36709
    },
    {
      "epoch": 1.6846404478913313,
      "grad_norm": 0.5355930924415588,
      "learning_rate": 4.177593477808764e-06,
      "loss": 0.394,
      "step": 36710
    },
    {
      "epoch": 1.6846863383965858,
      "grad_norm": 0.41289010643959045,
      "learning_rate": 4.177351630300648e-06,
      "loss": 0.2449,
      "step": 36711
    },
    {
      "epoch": 1.68473222890184,
      "grad_norm": 0.4664144814014435,
      "learning_rate": 4.177109784770742e-06,
      "loss": 0.3619,
      "step": 36712
    },
    {
      "epoch": 1.6847781194070945,
      "grad_norm": 0.44762948155403137,
      "learning_rate": 4.176867941219624e-06,
      "loss": 0.3167,
      "step": 36713
    },
    {
      "epoch": 1.684824009912349,
      "grad_norm": 0.4577914774417877,
      "learning_rate": 4.176626099647881e-06,
      "loss": 0.3687,
      "step": 36714
    },
    {
      "epoch": 1.6848699004176035,
      "grad_norm": 0.5063039660453796,
      "learning_rate": 4.176384260056094e-06,
      "loss": 0.4125,
      "step": 36715
    },
    {
      "epoch": 1.684915790922858,
      "grad_norm": 0.5064852833747864,
      "learning_rate": 4.17614242244484e-06,
      "loss": 0.4282,
      "step": 36716
    },
    {
      "epoch": 1.6849616814281125,
      "grad_norm": 0.5046652555465698,
      "learning_rate": 4.1759005868147054e-06,
      "loss": 0.4354,
      "step": 36717
    },
    {
      "epoch": 1.685007571933367,
      "grad_norm": 0.526836633682251,
      "learning_rate": 4.1756587531662695e-06,
      "loss": 0.4302,
      "step": 36718
    },
    {
      "epoch": 1.6850534624386215,
      "grad_norm": 0.43159839510917664,
      "learning_rate": 4.175416921500113e-06,
      "loss": 0.2865,
      "step": 36719
    },
    {
      "epoch": 1.685099352943876,
      "grad_norm": 0.48072636127471924,
      "learning_rate": 4.175175091816821e-06,
      "loss": 0.3526,
      "step": 36720
    },
    {
      "epoch": 1.6851452434491305,
      "grad_norm": 0.47590675950050354,
      "learning_rate": 4.17493326411697e-06,
      "loss": 0.3454,
      "step": 36721
    },
    {
      "epoch": 1.6851911339543848,
      "grad_norm": 0.4662701189517975,
      "learning_rate": 4.1746914384011435e-06,
      "loss": 0.3507,
      "step": 36722
    },
    {
      "epoch": 1.6852370244596393,
      "grad_norm": 0.4876888692378998,
      "learning_rate": 4.174449614669927e-06,
      "loss": 0.3744,
      "step": 36723
    },
    {
      "epoch": 1.6852829149648938,
      "grad_norm": 0.47026485204696655,
      "learning_rate": 4.174207792923897e-06,
      "loss": 0.3421,
      "step": 36724
    },
    {
      "epoch": 1.685328805470148,
      "grad_norm": 0.5658932328224182,
      "learning_rate": 4.173965973163635e-06,
      "loss": 0.4592,
      "step": 36725
    },
    {
      "epoch": 1.6853746959754026,
      "grad_norm": 0.4143156409263611,
      "learning_rate": 4.173724155389727e-06,
      "loss": 0.2783,
      "step": 36726
    },
    {
      "epoch": 1.685420586480657,
      "grad_norm": 0.4525379240512848,
      "learning_rate": 4.17348233960275e-06,
      "loss": 0.3031,
      "step": 36727
    },
    {
      "epoch": 1.6854664769859116,
      "grad_norm": 0.489981472492218,
      "learning_rate": 4.173240525803286e-06,
      "loss": 0.3831,
      "step": 36728
    },
    {
      "epoch": 1.685512367491166,
      "grad_norm": 0.4131324291229248,
      "learning_rate": 4.172998713991919e-06,
      "loss": 0.3011,
      "step": 36729
    },
    {
      "epoch": 1.6855582579964206,
      "grad_norm": 0.4798654019832611,
      "learning_rate": 4.172756904169229e-06,
      "loss": 0.3561,
      "step": 36730
    },
    {
      "epoch": 1.685604148501675,
      "grad_norm": 0.44545382261276245,
      "learning_rate": 4.172515096335798e-06,
      "loss": 0.3654,
      "step": 36731
    },
    {
      "epoch": 1.6856500390069296,
      "grad_norm": 0.4579126238822937,
      "learning_rate": 4.172273290492207e-06,
      "loss": 0.3395,
      "step": 36732
    },
    {
      "epoch": 1.685695929512184,
      "grad_norm": 0.4803103804588318,
      "learning_rate": 4.172031486639037e-06,
      "loss": 0.3774,
      "step": 36733
    },
    {
      "epoch": 1.6857418200174386,
      "grad_norm": 0.43887925148010254,
      "learning_rate": 4.171789684776869e-06,
      "loss": 0.3183,
      "step": 36734
    },
    {
      "epoch": 1.6857877105226928,
      "grad_norm": 0.46445730328559875,
      "learning_rate": 4.1715478849062866e-06,
      "loss": 0.3233,
      "step": 36735
    },
    {
      "epoch": 1.6858336010279473,
      "grad_norm": 0.45388227701187134,
      "learning_rate": 4.171306087027871e-06,
      "loss": 0.3394,
      "step": 36736
    },
    {
      "epoch": 1.6858794915332018,
      "grad_norm": 0.5232885479927063,
      "learning_rate": 4.171064291142201e-06,
      "loss": 0.4215,
      "step": 36737
    },
    {
      "epoch": 1.685925382038456,
      "grad_norm": 0.5060933232307434,
      "learning_rate": 4.17082249724986e-06,
      "loss": 0.3982,
      "step": 36738
    },
    {
      "epoch": 1.6859712725437106,
      "grad_norm": 0.49918821454048157,
      "learning_rate": 4.17058070535143e-06,
      "loss": 0.4111,
      "step": 36739
    },
    {
      "epoch": 1.686017163048965,
      "grad_norm": 0.4382694363594055,
      "learning_rate": 4.170338915447491e-06,
      "loss": 0.3298,
      "step": 36740
    },
    {
      "epoch": 1.6860630535542196,
      "grad_norm": 0.457792192697525,
      "learning_rate": 4.170097127538626e-06,
      "loss": 0.3527,
      "step": 36741
    },
    {
      "epoch": 1.686108944059474,
      "grad_norm": 0.4837076663970947,
      "learning_rate": 4.1698553416254154e-06,
      "loss": 0.4135,
      "step": 36742
    },
    {
      "epoch": 1.6861548345647286,
      "grad_norm": 0.4626867473125458,
      "learning_rate": 4.16961355770844e-06,
      "loss": 0.3844,
      "step": 36743
    },
    {
      "epoch": 1.686200725069983,
      "grad_norm": 0.4796828329563141,
      "learning_rate": 4.16937177578828e-06,
      "loss": 0.3795,
      "step": 36744
    },
    {
      "epoch": 1.6862466155752376,
      "grad_norm": 0.48276805877685547,
      "learning_rate": 4.169129995865522e-06,
      "loss": 0.3525,
      "step": 36745
    },
    {
      "epoch": 1.686292506080492,
      "grad_norm": 0.4635700583457947,
      "learning_rate": 4.168888217940743e-06,
      "loss": 0.3367,
      "step": 36746
    },
    {
      "epoch": 1.6863383965857464,
      "grad_norm": 0.4601075053215027,
      "learning_rate": 4.168646442014525e-06,
      "loss": 0.3814,
      "step": 36747
    },
    {
      "epoch": 1.6863842870910009,
      "grad_norm": 0.519730269908905,
      "learning_rate": 4.16840466808745e-06,
      "loss": 0.3889,
      "step": 36748
    },
    {
      "epoch": 1.6864301775962554,
      "grad_norm": 0.4684511125087738,
      "learning_rate": 4.1681628961601005e-06,
      "loss": 0.3351,
      "step": 36749
    },
    {
      "epoch": 1.6864760681015099,
      "grad_norm": 0.44276687502861023,
      "learning_rate": 4.1679211262330545e-06,
      "loss": 0.3178,
      "step": 36750
    },
    {
      "epoch": 1.6865219586067641,
      "grad_norm": 0.4810556173324585,
      "learning_rate": 4.167679358306898e-06,
      "loss": 0.3741,
      "step": 36751
    },
    {
      "epoch": 1.6865678491120186,
      "grad_norm": 0.47345873713493347,
      "learning_rate": 4.1674375923822085e-06,
      "loss": 0.3404,
      "step": 36752
    },
    {
      "epoch": 1.6866137396172731,
      "grad_norm": 0.5055721998214722,
      "learning_rate": 4.167195828459569e-06,
      "loss": 0.3501,
      "step": 36753
    },
    {
      "epoch": 1.6866596301225276,
      "grad_norm": 0.46325966715812683,
      "learning_rate": 4.166954066539561e-06,
      "loss": 0.3773,
      "step": 36754
    },
    {
      "epoch": 1.6867055206277821,
      "grad_norm": 0.4470631182193756,
      "learning_rate": 4.166712306622766e-06,
      "loss": 0.3117,
      "step": 36755
    },
    {
      "epoch": 1.6867514111330366,
      "grad_norm": 0.4749334454536438,
      "learning_rate": 4.166470548709762e-06,
      "loss": 0.3295,
      "step": 36756
    },
    {
      "epoch": 1.6867973016382911,
      "grad_norm": 0.4975185692310333,
      "learning_rate": 4.166228792801137e-06,
      "loss": 0.3941,
      "step": 36757
    },
    {
      "epoch": 1.6868431921435456,
      "grad_norm": 0.524256706237793,
      "learning_rate": 4.165987038897468e-06,
      "loss": 0.4603,
      "step": 36758
    },
    {
      "epoch": 1.6868890826488,
      "grad_norm": 0.5346419811248779,
      "learning_rate": 4.1657452869993345e-06,
      "loss": 0.4157,
      "step": 36759
    },
    {
      "epoch": 1.6869349731540544,
      "grad_norm": 0.4835657477378845,
      "learning_rate": 4.165503537107323e-06,
      "loss": 0.4022,
      "step": 36760
    },
    {
      "epoch": 1.6869808636593089,
      "grad_norm": 0.45903512835502625,
      "learning_rate": 4.1652617892220115e-06,
      "loss": 0.3883,
      "step": 36761
    },
    {
      "epoch": 1.6870267541645634,
      "grad_norm": 0.5147690773010254,
      "learning_rate": 4.165020043343981e-06,
      "loss": 0.3504,
      "step": 36762
    },
    {
      "epoch": 1.6870726446698177,
      "grad_norm": 0.5026307702064514,
      "learning_rate": 4.164778299473815e-06,
      "loss": 0.437,
      "step": 36763
    },
    {
      "epoch": 1.6871185351750722,
      "grad_norm": 0.4412024915218353,
      "learning_rate": 4.164536557612093e-06,
      "loss": 0.2976,
      "step": 36764
    },
    {
      "epoch": 1.6871644256803267,
      "grad_norm": 0.4594278931617737,
      "learning_rate": 4.164294817759395e-06,
      "loss": 0.3712,
      "step": 36765
    },
    {
      "epoch": 1.6872103161855811,
      "grad_norm": 0.43304747343063354,
      "learning_rate": 4.164053079916306e-06,
      "loss": 0.3124,
      "step": 36766
    },
    {
      "epoch": 1.6872562066908356,
      "grad_norm": 0.5000492930412292,
      "learning_rate": 4.163811344083407e-06,
      "loss": 0.3843,
      "step": 36767
    },
    {
      "epoch": 1.6873020971960901,
      "grad_norm": 0.4477241337299347,
      "learning_rate": 4.163569610261275e-06,
      "loss": 0.3578,
      "step": 36768
    },
    {
      "epoch": 1.6873479877013446,
      "grad_norm": 0.46870797872543335,
      "learning_rate": 4.163327878450496e-06,
      "loss": 0.3504,
      "step": 36769
    },
    {
      "epoch": 1.6873938782065991,
      "grad_norm": 0.5244657397270203,
      "learning_rate": 4.16308614865165e-06,
      "loss": 0.4142,
      "step": 36770
    },
    {
      "epoch": 1.6874397687118536,
      "grad_norm": 0.48102545738220215,
      "learning_rate": 4.1628444208653144e-06,
      "loss": 0.3865,
      "step": 36771
    },
    {
      "epoch": 1.6874856592171081,
      "grad_norm": 0.43701088428497314,
      "learning_rate": 4.162602695092076e-06,
      "loss": 0.3125,
      "step": 36772
    },
    {
      "epoch": 1.6875315497223624,
      "grad_norm": 0.4430621862411499,
      "learning_rate": 4.162360971332514e-06,
      "loss": 0.2991,
      "step": 36773
    },
    {
      "epoch": 1.687577440227617,
      "grad_norm": 0.48940935730934143,
      "learning_rate": 4.162119249587208e-06,
      "loss": 0.3689,
      "step": 36774
    },
    {
      "epoch": 1.6876233307328714,
      "grad_norm": 0.4404969811439514,
      "learning_rate": 4.1618775298567425e-06,
      "loss": 0.306,
      "step": 36775
    },
    {
      "epoch": 1.6876692212381257,
      "grad_norm": 0.4469551146030426,
      "learning_rate": 4.161635812141698e-06,
      "loss": 0.3023,
      "step": 36776
    },
    {
      "epoch": 1.6877151117433802,
      "grad_norm": 0.4747537672519684,
      "learning_rate": 4.161394096442653e-06,
      "loss": 0.3999,
      "step": 36777
    },
    {
      "epoch": 1.6877610022486347,
      "grad_norm": 0.44679808616638184,
      "learning_rate": 4.161152382760189e-06,
      "loss": 0.2989,
      "step": 36778
    },
    {
      "epoch": 1.6878068927538892,
      "grad_norm": 0.5224229097366333,
      "learning_rate": 4.160910671094891e-06,
      "loss": 0.4162,
      "step": 36779
    },
    {
      "epoch": 1.6878527832591437,
      "grad_norm": 0.5652673244476318,
      "learning_rate": 4.160668961447339e-06,
      "loss": 0.3119,
      "step": 36780
    },
    {
      "epoch": 1.6878986737643982,
      "grad_norm": 0.4801797568798065,
      "learning_rate": 4.16042725381811e-06,
      "loss": 0.3464,
      "step": 36781
    },
    {
      "epoch": 1.6879445642696527,
      "grad_norm": 0.4744786024093628,
      "learning_rate": 4.160185548207791e-06,
      "loss": 0.3721,
      "step": 36782
    },
    {
      "epoch": 1.6879904547749072,
      "grad_norm": 0.46331149339675903,
      "learning_rate": 4.15994384461696e-06,
      "loss": 0.342,
      "step": 36783
    },
    {
      "epoch": 1.6880363452801617,
      "grad_norm": 0.4478573501110077,
      "learning_rate": 4.159702143046199e-06,
      "loss": 0.3336,
      "step": 36784
    },
    {
      "epoch": 1.6880822357854162,
      "grad_norm": 0.4576170742511749,
      "learning_rate": 4.159460443496089e-06,
      "loss": 0.3282,
      "step": 36785
    },
    {
      "epoch": 1.6881281262906704,
      "grad_norm": 0.4604507088661194,
      "learning_rate": 4.159218745967212e-06,
      "loss": 0.3607,
      "step": 36786
    },
    {
      "epoch": 1.688174016795925,
      "grad_norm": 0.48022356629371643,
      "learning_rate": 4.158977050460146e-06,
      "loss": 0.346,
      "step": 36787
    },
    {
      "epoch": 1.6882199073011794,
      "grad_norm": 0.4428163766860962,
      "learning_rate": 4.1587353569754775e-06,
      "loss": 0.3018,
      "step": 36788
    },
    {
      "epoch": 1.6882657978064337,
      "grad_norm": 0.47617003321647644,
      "learning_rate": 4.158493665513784e-06,
      "loss": 0.4376,
      "step": 36789
    },
    {
      "epoch": 1.6883116883116882,
      "grad_norm": 0.50802081823349,
      "learning_rate": 4.158251976075648e-06,
      "loss": 0.4479,
      "step": 36790
    },
    {
      "epoch": 1.6883575788169427,
      "grad_norm": 0.48347902297973633,
      "learning_rate": 4.15801028866165e-06,
      "loss": 0.3666,
      "step": 36791
    },
    {
      "epoch": 1.6884034693221972,
      "grad_norm": 0.4668214023113251,
      "learning_rate": 4.157768603272373e-06,
      "loss": 0.3715,
      "step": 36792
    },
    {
      "epoch": 1.6884493598274517,
      "grad_norm": 0.4578152298927307,
      "learning_rate": 4.157526919908393e-06,
      "loss": 0.3593,
      "step": 36793
    },
    {
      "epoch": 1.6884952503327062,
      "grad_norm": 0.4864543378353119,
      "learning_rate": 4.157285238570298e-06,
      "loss": 0.3672,
      "step": 36794
    },
    {
      "epoch": 1.6885411408379607,
      "grad_norm": 0.44036224484443665,
      "learning_rate": 4.1570435592586656e-06,
      "loss": 0.341,
      "step": 36795
    },
    {
      "epoch": 1.6885870313432152,
      "grad_norm": 0.4474756717681885,
      "learning_rate": 4.156801881974076e-06,
      "loss": 0.3262,
      "step": 36796
    },
    {
      "epoch": 1.6886329218484697,
      "grad_norm": 0.46045470237731934,
      "learning_rate": 4.156560206717113e-06,
      "loss": 0.3515,
      "step": 36797
    },
    {
      "epoch": 1.688678812353724,
      "grad_norm": 0.44514498114585876,
      "learning_rate": 4.156318533488357e-06,
      "loss": 0.3427,
      "step": 36798
    },
    {
      "epoch": 1.6887247028589785,
      "grad_norm": 0.46188172698020935,
      "learning_rate": 4.156076862288386e-06,
      "loss": 0.3424,
      "step": 36799
    },
    {
      "epoch": 1.688770593364233,
      "grad_norm": 0.43751174211502075,
      "learning_rate": 4.155835193117787e-06,
      "loss": 0.3283,
      "step": 36800
    },
    {
      "epoch": 1.6888164838694872,
      "grad_norm": 0.43324941396713257,
      "learning_rate": 4.155593525977136e-06,
      "loss": 0.2944,
      "step": 36801
    },
    {
      "epoch": 1.6888623743747417,
      "grad_norm": 0.44718584418296814,
      "learning_rate": 4.1553518608670165e-06,
      "loss": 0.3405,
      "step": 36802
    },
    {
      "epoch": 1.6889082648799962,
      "grad_norm": 0.4858654737472534,
      "learning_rate": 4.155110197788011e-06,
      "loss": 0.3638,
      "step": 36803
    },
    {
      "epoch": 1.6889541553852507,
      "grad_norm": 0.4785882830619812,
      "learning_rate": 4.154868536740697e-06,
      "loss": 0.4021,
      "step": 36804
    },
    {
      "epoch": 1.6890000458905052,
      "grad_norm": 0.45162996649742126,
      "learning_rate": 4.154626877725656e-06,
      "loss": 0.3409,
      "step": 36805
    },
    {
      "epoch": 1.6890459363957597,
      "grad_norm": 0.4666168987751007,
      "learning_rate": 4.154385220743474e-06,
      "loss": 0.3595,
      "step": 36806
    },
    {
      "epoch": 1.6890918269010142,
      "grad_norm": 0.46007511019706726,
      "learning_rate": 4.154143565794727e-06,
      "loss": 0.3107,
      "step": 36807
    },
    {
      "epoch": 1.6891377174062687,
      "grad_norm": 0.44796469807624817,
      "learning_rate": 4.153901912879998e-06,
      "loss": 0.3204,
      "step": 36808
    },
    {
      "epoch": 1.6891836079115232,
      "grad_norm": 0.432830274105072,
      "learning_rate": 4.153660261999866e-06,
      "loss": 0.2955,
      "step": 36809
    },
    {
      "epoch": 1.6892294984167777,
      "grad_norm": 0.48359212279319763,
      "learning_rate": 4.153418613154916e-06,
      "loss": 0.3659,
      "step": 36810
    },
    {
      "epoch": 1.689275388922032,
      "grad_norm": 0.47564437985420227,
      "learning_rate": 4.153176966345727e-06,
      "loss": 0.3724,
      "step": 36811
    },
    {
      "epoch": 1.6893212794272865,
      "grad_norm": 0.46529024839401245,
      "learning_rate": 4.152935321572879e-06,
      "loss": 0.3573,
      "step": 36812
    },
    {
      "epoch": 1.689367169932541,
      "grad_norm": 0.4471629858016968,
      "learning_rate": 4.152693678836955e-06,
      "loss": 0.3107,
      "step": 36813
    },
    {
      "epoch": 1.6894130604377953,
      "grad_norm": 0.49422475695610046,
      "learning_rate": 4.152452038138536e-06,
      "loss": 0.4159,
      "step": 36814
    },
    {
      "epoch": 1.6894589509430498,
      "grad_norm": 0.4844529628753662,
      "learning_rate": 4.1522103994782e-06,
      "loss": 0.4023,
      "step": 36815
    },
    {
      "epoch": 1.6895048414483043,
      "grad_norm": 0.46729210019111633,
      "learning_rate": 4.151968762856532e-06,
      "loss": 0.3818,
      "step": 36816
    },
    {
      "epoch": 1.6895507319535588,
      "grad_norm": 0.4574427306652069,
      "learning_rate": 4.151727128274112e-06,
      "loss": 0.3159,
      "step": 36817
    },
    {
      "epoch": 1.6895966224588133,
      "grad_norm": 0.48064419627189636,
      "learning_rate": 4.151485495731519e-06,
      "loss": 0.3334,
      "step": 36818
    },
    {
      "epoch": 1.6896425129640678,
      "grad_norm": 0.5360739827156067,
      "learning_rate": 4.151243865229337e-06,
      "loss": 0.4429,
      "step": 36819
    },
    {
      "epoch": 1.6896884034693223,
      "grad_norm": 0.4431733787059784,
      "learning_rate": 4.151002236768146e-06,
      "loss": 0.3413,
      "step": 36820
    },
    {
      "epoch": 1.6897342939745768,
      "grad_norm": 0.4831811487674713,
      "learning_rate": 4.150760610348525e-06,
      "loss": 0.3755,
      "step": 36821
    },
    {
      "epoch": 1.6897801844798312,
      "grad_norm": 0.4504460096359253,
      "learning_rate": 4.150518985971057e-06,
      "loss": 0.3207,
      "step": 36822
    },
    {
      "epoch": 1.6898260749850857,
      "grad_norm": 0.44837692379951477,
      "learning_rate": 4.150277363636323e-06,
      "loss": 0.3197,
      "step": 36823
    },
    {
      "epoch": 1.68987196549034,
      "grad_norm": 0.4410548806190491,
      "learning_rate": 4.150035743344904e-06,
      "loss": 0.3264,
      "step": 36824
    },
    {
      "epoch": 1.6899178559955945,
      "grad_norm": 0.45388856530189514,
      "learning_rate": 4.149794125097382e-06,
      "loss": 0.3628,
      "step": 36825
    },
    {
      "epoch": 1.689963746500849,
      "grad_norm": 0.46156802773475647,
      "learning_rate": 4.149552508894336e-06,
      "loss": 0.3548,
      "step": 36826
    },
    {
      "epoch": 1.6900096370061033,
      "grad_norm": 0.44656237959861755,
      "learning_rate": 4.149310894736345e-06,
      "loss": 0.3126,
      "step": 36827
    },
    {
      "epoch": 1.6900555275113578,
      "grad_norm": 0.4617288410663605,
      "learning_rate": 4.149069282623997e-06,
      "loss": 0.3664,
      "step": 36828
    },
    {
      "epoch": 1.6901014180166123,
      "grad_norm": 0.43301883339881897,
      "learning_rate": 4.148827672557867e-06,
      "loss": 0.36,
      "step": 36829
    },
    {
      "epoch": 1.6901473085218668,
      "grad_norm": 0.4307457506656647,
      "learning_rate": 4.148586064538537e-06,
      "loss": 0.3187,
      "step": 36830
    },
    {
      "epoch": 1.6901931990271213,
      "grad_norm": 0.4437965452671051,
      "learning_rate": 4.14834445856659e-06,
      "loss": 0.31,
      "step": 36831
    },
    {
      "epoch": 1.6902390895323758,
      "grad_norm": 0.49264201521873474,
      "learning_rate": 4.1481028546426055e-06,
      "loss": 0.4367,
      "step": 36832
    },
    {
      "epoch": 1.6902849800376303,
      "grad_norm": 0.4288793206214905,
      "learning_rate": 4.147861252767164e-06,
      "loss": 0.2752,
      "step": 36833
    },
    {
      "epoch": 1.6903308705428848,
      "grad_norm": 0.4393137991428375,
      "learning_rate": 4.147619652940848e-06,
      "loss": 0.2999,
      "step": 36834
    },
    {
      "epoch": 1.6903767610481393,
      "grad_norm": 0.507400393486023,
      "learning_rate": 4.147378055164239e-06,
      "loss": 0.3783,
      "step": 36835
    },
    {
      "epoch": 1.6904226515533936,
      "grad_norm": 0.4294240176677704,
      "learning_rate": 4.147136459437913e-06,
      "loss": 0.255,
      "step": 36836
    },
    {
      "epoch": 1.690468542058648,
      "grad_norm": 0.4532887935638428,
      "learning_rate": 4.146894865762458e-06,
      "loss": 0.3629,
      "step": 36837
    },
    {
      "epoch": 1.6905144325639025,
      "grad_norm": 0.44023844599723816,
      "learning_rate": 4.146653274138451e-06,
      "loss": 0.2768,
      "step": 36838
    },
    {
      "epoch": 1.690560323069157,
      "grad_norm": 0.5166060328483582,
      "learning_rate": 4.146411684566472e-06,
      "loss": 0.4024,
      "step": 36839
    },
    {
      "epoch": 1.6906062135744113,
      "grad_norm": 0.46761175990104675,
      "learning_rate": 4.146170097047106e-06,
      "loss": 0.3531,
      "step": 36840
    },
    {
      "epoch": 1.6906521040796658,
      "grad_norm": 0.4524780213832855,
      "learning_rate": 4.14592851158093e-06,
      "loss": 0.3162,
      "step": 36841
    },
    {
      "epoch": 1.6906979945849203,
      "grad_norm": 0.42384982109069824,
      "learning_rate": 4.1456869281685265e-06,
      "loss": 0.2664,
      "step": 36842
    },
    {
      "epoch": 1.6907438850901748,
      "grad_norm": 0.4624957740306854,
      "learning_rate": 4.145445346810475e-06,
      "loss": 0.3504,
      "step": 36843
    },
    {
      "epoch": 1.6907897755954293,
      "grad_norm": 0.44703394174575806,
      "learning_rate": 4.1452037675073595e-06,
      "loss": 0.3383,
      "step": 36844
    },
    {
      "epoch": 1.6908356661006838,
      "grad_norm": 0.4698556065559387,
      "learning_rate": 4.144962190259759e-06,
      "loss": 0.3737,
      "step": 36845
    },
    {
      "epoch": 1.6908815566059383,
      "grad_norm": 0.5036715269088745,
      "learning_rate": 4.144720615068254e-06,
      "loss": 0.4416,
      "step": 36846
    },
    {
      "epoch": 1.6909274471111928,
      "grad_norm": 0.5077207088470459,
      "learning_rate": 4.1444790419334265e-06,
      "loss": 0.4312,
      "step": 36847
    },
    {
      "epoch": 1.6909733376164473,
      "grad_norm": 0.433074951171875,
      "learning_rate": 4.144237470855857e-06,
      "loss": 0.3227,
      "step": 36848
    },
    {
      "epoch": 1.6910192281217016,
      "grad_norm": 0.4638296663761139,
      "learning_rate": 4.143995901836124e-06,
      "loss": 0.3823,
      "step": 36849
    },
    {
      "epoch": 1.691065118626956,
      "grad_norm": 0.44176164269447327,
      "learning_rate": 4.1437543348748145e-06,
      "loss": 0.3273,
      "step": 36850
    },
    {
      "epoch": 1.6911110091322106,
      "grad_norm": 0.4198774993419647,
      "learning_rate": 4.143512769972504e-06,
      "loss": 0.2833,
      "step": 36851
    },
    {
      "epoch": 1.6911568996374648,
      "grad_norm": 0.45586130023002625,
      "learning_rate": 4.143271207129773e-06,
      "loss": 0.3561,
      "step": 36852
    },
    {
      "epoch": 1.6912027901427193,
      "grad_norm": 0.45588189363479614,
      "learning_rate": 4.143029646347207e-06,
      "loss": 0.3477,
      "step": 36853
    },
    {
      "epoch": 1.6912486806479738,
      "grad_norm": 0.4814203083515167,
      "learning_rate": 4.142788087625383e-06,
      "loss": 0.3879,
      "step": 36854
    },
    {
      "epoch": 1.6912945711532283,
      "grad_norm": 0.5031703114509583,
      "learning_rate": 4.142546530964883e-06,
      "loss": 0.3952,
      "step": 36855
    },
    {
      "epoch": 1.6913404616584828,
      "grad_norm": 0.4574336111545563,
      "learning_rate": 4.142304976366289e-06,
      "loss": 0.3487,
      "step": 36856
    },
    {
      "epoch": 1.6913863521637373,
      "grad_norm": 0.47126877307891846,
      "learning_rate": 4.14206342383018e-06,
      "loss": 0.3388,
      "step": 36857
    },
    {
      "epoch": 1.6914322426689918,
      "grad_norm": 0.4507964551448822,
      "learning_rate": 4.141821873357136e-06,
      "loss": 0.3207,
      "step": 36858
    },
    {
      "epoch": 1.6914781331742463,
      "grad_norm": 0.4789820909500122,
      "learning_rate": 4.141580324947742e-06,
      "loss": 0.3792,
      "step": 36859
    },
    {
      "epoch": 1.6915240236795008,
      "grad_norm": 0.49611976742744446,
      "learning_rate": 4.141338778602576e-06,
      "loss": 0.3927,
      "step": 36860
    },
    {
      "epoch": 1.6915699141847553,
      "grad_norm": 0.5025447010993958,
      "learning_rate": 4.141097234322219e-06,
      "loss": 0.4073,
      "step": 36861
    },
    {
      "epoch": 1.6916158046900096,
      "grad_norm": 0.45788344740867615,
      "learning_rate": 4.140855692107253e-06,
      "loss": 0.3306,
      "step": 36862
    },
    {
      "epoch": 1.691661695195264,
      "grad_norm": 0.49800029397010803,
      "learning_rate": 4.140614151958257e-06,
      "loss": 0.4112,
      "step": 36863
    },
    {
      "epoch": 1.6917075857005186,
      "grad_norm": 0.4580008387565613,
      "learning_rate": 4.140372613875811e-06,
      "loss": 0.3347,
      "step": 36864
    },
    {
      "epoch": 1.6917534762057729,
      "grad_norm": 0.4278797209262848,
      "learning_rate": 4.1401310778604994e-06,
      "loss": 0.2699,
      "step": 36865
    },
    {
      "epoch": 1.6917993667110274,
      "grad_norm": 0.4615432918071747,
      "learning_rate": 4.139889543912902e-06,
      "loss": 0.3223,
      "step": 36866
    },
    {
      "epoch": 1.6918452572162819,
      "grad_norm": 0.4696485102176666,
      "learning_rate": 4.139648012033598e-06,
      "loss": 0.3395,
      "step": 36867
    },
    {
      "epoch": 1.6918911477215364,
      "grad_norm": 0.4438839256763458,
      "learning_rate": 4.139406482223169e-06,
      "loss": 0.3085,
      "step": 36868
    },
    {
      "epoch": 1.6919370382267909,
      "grad_norm": 0.489225298166275,
      "learning_rate": 4.139164954482196e-06,
      "loss": 0.4206,
      "step": 36869
    },
    {
      "epoch": 1.6919829287320454,
      "grad_norm": 0.44891610741615295,
      "learning_rate": 4.138923428811257e-06,
      "loss": 0.3026,
      "step": 36870
    },
    {
      "epoch": 1.6920288192372999,
      "grad_norm": 0.4640611410140991,
      "learning_rate": 4.138681905210938e-06,
      "loss": 0.3238,
      "step": 36871
    },
    {
      "epoch": 1.6920747097425544,
      "grad_norm": 0.467511922121048,
      "learning_rate": 4.138440383681819e-06,
      "loss": 0.3525,
      "step": 36872
    },
    {
      "epoch": 1.6921206002478089,
      "grad_norm": 0.4589094817638397,
      "learning_rate": 4.138198864224478e-06,
      "loss": 0.3531,
      "step": 36873
    },
    {
      "epoch": 1.6921664907530634,
      "grad_norm": 0.5083353519439697,
      "learning_rate": 4.137957346839493e-06,
      "loss": 0.4155,
      "step": 36874
    },
    {
      "epoch": 1.6922123812583176,
      "grad_norm": 0.4752688705921173,
      "learning_rate": 4.137715831527451e-06,
      "loss": 0.3467,
      "step": 36875
    },
    {
      "epoch": 1.6922582717635721,
      "grad_norm": 0.4769943058490753,
      "learning_rate": 4.1374743182889306e-06,
      "loss": 0.3147,
      "step": 36876
    },
    {
      "epoch": 1.6923041622688266,
      "grad_norm": 0.47080984711647034,
      "learning_rate": 4.137232807124512e-06,
      "loss": 0.3257,
      "step": 36877
    },
    {
      "epoch": 1.692350052774081,
      "grad_norm": 0.4451020658016205,
      "learning_rate": 4.1369912980347766e-06,
      "loss": 0.313,
      "step": 36878
    },
    {
      "epoch": 1.6923959432793354,
      "grad_norm": 0.515381395816803,
      "learning_rate": 4.136749791020305e-06,
      "loss": 0.4387,
      "step": 36879
    },
    {
      "epoch": 1.69244183378459,
      "grad_norm": 0.45998769998550415,
      "learning_rate": 4.136508286081675e-06,
      "loss": 0.3181,
      "step": 36880
    },
    {
      "epoch": 1.6924877242898444,
      "grad_norm": 0.47594398260116577,
      "learning_rate": 4.136266783219472e-06,
      "loss": 0.3783,
      "step": 36881
    },
    {
      "epoch": 1.692533614795099,
      "grad_norm": 0.4827595055103302,
      "learning_rate": 4.136025282434276e-06,
      "loss": 0.3814,
      "step": 36882
    },
    {
      "epoch": 1.6925795053003534,
      "grad_norm": 0.48962658643722534,
      "learning_rate": 4.135783783726665e-06,
      "loss": 0.358,
      "step": 36883
    },
    {
      "epoch": 1.692625395805608,
      "grad_norm": 0.4838423728942871,
      "learning_rate": 4.135542287097222e-06,
      "loss": 0.3848,
      "step": 36884
    },
    {
      "epoch": 1.6926712863108624,
      "grad_norm": 0.43589261174201965,
      "learning_rate": 4.135300792546527e-06,
      "loss": 0.2918,
      "step": 36885
    },
    {
      "epoch": 1.6927171768161169,
      "grad_norm": 0.5090653300285339,
      "learning_rate": 4.135059300075158e-06,
      "loss": 0.4372,
      "step": 36886
    },
    {
      "epoch": 1.6927630673213712,
      "grad_norm": 0.4632788896560669,
      "learning_rate": 4.134817809683702e-06,
      "loss": 0.3147,
      "step": 36887
    },
    {
      "epoch": 1.6928089578266257,
      "grad_norm": 0.4514651894569397,
      "learning_rate": 4.1345763213727345e-06,
      "loss": 0.3364,
      "step": 36888
    },
    {
      "epoch": 1.6928548483318802,
      "grad_norm": 0.48137402534484863,
      "learning_rate": 4.134334835142838e-06,
      "loss": 0.3794,
      "step": 36889
    },
    {
      "epoch": 1.6929007388371344,
      "grad_norm": 0.49424150586128235,
      "learning_rate": 4.1340933509945935e-06,
      "loss": 0.4263,
      "step": 36890
    },
    {
      "epoch": 1.692946629342389,
      "grad_norm": 0.4852655529975891,
      "learning_rate": 4.133851868928581e-06,
      "loss": 0.4279,
      "step": 36891
    },
    {
      "epoch": 1.6929925198476434,
      "grad_norm": 0.4533689022064209,
      "learning_rate": 4.133610388945379e-06,
      "loss": 0.3004,
      "step": 36892
    },
    {
      "epoch": 1.693038410352898,
      "grad_norm": 0.4615727365016937,
      "learning_rate": 4.1333689110455734e-06,
      "loss": 0.3356,
      "step": 36893
    },
    {
      "epoch": 1.6930843008581524,
      "grad_norm": 0.45964616537094116,
      "learning_rate": 4.133127435229743e-06,
      "loss": 0.3314,
      "step": 36894
    },
    {
      "epoch": 1.693130191363407,
      "grad_norm": 0.47771501541137695,
      "learning_rate": 4.1328859614984646e-06,
      "loss": 0.4177,
      "step": 36895
    },
    {
      "epoch": 1.6931760818686614,
      "grad_norm": 0.49474000930786133,
      "learning_rate": 4.1326444898523235e-06,
      "loss": 0.4055,
      "step": 36896
    },
    {
      "epoch": 1.693221972373916,
      "grad_norm": 0.44006356596946716,
      "learning_rate": 4.132403020291899e-06,
      "loss": 0.2787,
      "step": 36897
    },
    {
      "epoch": 1.6932678628791704,
      "grad_norm": 0.5134905576705933,
      "learning_rate": 4.13216155281777e-06,
      "loss": 0.4495,
      "step": 36898
    },
    {
      "epoch": 1.693313753384425,
      "grad_norm": 0.4804705083370209,
      "learning_rate": 4.13192008743052e-06,
      "loss": 0.3592,
      "step": 36899
    },
    {
      "epoch": 1.6933596438896792,
      "grad_norm": 0.4158957302570343,
      "learning_rate": 4.131678624130728e-06,
      "loss": 0.283,
      "step": 36900
    },
    {
      "epoch": 1.6934055343949337,
      "grad_norm": 0.4449012875556946,
      "learning_rate": 4.131437162918972e-06,
      "loss": 0.3637,
      "step": 36901
    },
    {
      "epoch": 1.6934514249001882,
      "grad_norm": 0.45994657278060913,
      "learning_rate": 4.131195703795839e-06,
      "loss": 0.3445,
      "step": 36902
    },
    {
      "epoch": 1.6934973154054425,
      "grad_norm": 0.4867658019065857,
      "learning_rate": 4.1309542467619055e-06,
      "loss": 0.3813,
      "step": 36903
    },
    {
      "epoch": 1.693543205910697,
      "grad_norm": 0.4464125633239746,
      "learning_rate": 4.1307127918177515e-06,
      "loss": 0.3406,
      "step": 36904
    },
    {
      "epoch": 1.6935890964159515,
      "grad_norm": 0.426437109708786,
      "learning_rate": 4.1304713389639604e-06,
      "loss": 0.2947,
      "step": 36905
    },
    {
      "epoch": 1.693634986921206,
      "grad_norm": 0.4777913987636566,
      "learning_rate": 4.130229888201111e-06,
      "loss": 0.3689,
      "step": 36906
    },
    {
      "epoch": 1.6936808774264605,
      "grad_norm": 0.4622422158718109,
      "learning_rate": 4.129988439529784e-06,
      "loss": 0.3253,
      "step": 36907
    },
    {
      "epoch": 1.693726767931715,
      "grad_norm": 0.4693111777305603,
      "learning_rate": 4.129746992950558e-06,
      "loss": 0.3708,
      "step": 36908
    },
    {
      "epoch": 1.6937726584369694,
      "grad_norm": 0.45766136050224304,
      "learning_rate": 4.1295055484640184e-06,
      "loss": 0.3261,
      "step": 36909
    },
    {
      "epoch": 1.693818548942224,
      "grad_norm": 0.47616150975227356,
      "learning_rate": 4.129264106070743e-06,
      "loss": 0.3451,
      "step": 36910
    },
    {
      "epoch": 1.6938644394474784,
      "grad_norm": 0.474323570728302,
      "learning_rate": 4.129022665771312e-06,
      "loss": 0.3377,
      "step": 36911
    },
    {
      "epoch": 1.693910329952733,
      "grad_norm": 0.4845379590988159,
      "learning_rate": 4.128781227566307e-06,
      "loss": 0.3958,
      "step": 36912
    },
    {
      "epoch": 1.6939562204579872,
      "grad_norm": 0.5337485074996948,
      "learning_rate": 4.128539791456309e-06,
      "loss": 0.4577,
      "step": 36913
    },
    {
      "epoch": 1.6940021109632417,
      "grad_norm": 0.4459736943244934,
      "learning_rate": 4.128298357441894e-06,
      "loss": 0.3171,
      "step": 36914
    },
    {
      "epoch": 1.6940480014684962,
      "grad_norm": 0.47228631377220154,
      "learning_rate": 4.12805692552365e-06,
      "loss": 0.3302,
      "step": 36915
    },
    {
      "epoch": 1.6940938919737505,
      "grad_norm": 0.4573717415332794,
      "learning_rate": 4.127815495702154e-06,
      "loss": 0.3497,
      "step": 36916
    },
    {
      "epoch": 1.694139782479005,
      "grad_norm": 0.48213067650794983,
      "learning_rate": 4.127574067977984e-06,
      "loss": 0.3515,
      "step": 36917
    },
    {
      "epoch": 1.6941856729842595,
      "grad_norm": 0.4555206894874573,
      "learning_rate": 4.127332642351725e-06,
      "loss": 0.3353,
      "step": 36918
    },
    {
      "epoch": 1.694231563489514,
      "grad_norm": 0.44800934195518494,
      "learning_rate": 4.1270912188239555e-06,
      "loss": 0.3062,
      "step": 36919
    },
    {
      "epoch": 1.6942774539947685,
      "grad_norm": 0.4445018470287323,
      "learning_rate": 4.1268497973952545e-06,
      "loss": 0.3145,
      "step": 36920
    },
    {
      "epoch": 1.694323344500023,
      "grad_norm": 0.49942195415496826,
      "learning_rate": 4.126608378066207e-06,
      "loss": 0.4066,
      "step": 36921
    },
    {
      "epoch": 1.6943692350052775,
      "grad_norm": 0.460245281457901,
      "learning_rate": 4.126366960837389e-06,
      "loss": 0.3483,
      "step": 36922
    },
    {
      "epoch": 1.694415125510532,
      "grad_norm": 0.4961283802986145,
      "learning_rate": 4.126125545709381e-06,
      "loss": 0.4086,
      "step": 36923
    },
    {
      "epoch": 1.6944610160157865,
      "grad_norm": 0.430593341588974,
      "learning_rate": 4.125884132682767e-06,
      "loss": 0.2751,
      "step": 36924
    },
    {
      "epoch": 1.6945069065210407,
      "grad_norm": 0.4765610992908478,
      "learning_rate": 4.125642721758127e-06,
      "loss": 0.3396,
      "step": 36925
    },
    {
      "epoch": 1.6945527970262952,
      "grad_norm": 0.4605938494205475,
      "learning_rate": 4.125401312936038e-06,
      "loss": 0.3167,
      "step": 36926
    },
    {
      "epoch": 1.6945986875315497,
      "grad_norm": 0.42799097299575806,
      "learning_rate": 4.125159906217084e-06,
      "loss": 0.2984,
      "step": 36927
    },
    {
      "epoch": 1.6946445780368042,
      "grad_norm": 0.4860045611858368,
      "learning_rate": 4.124918501601846e-06,
      "loss": 0.3459,
      "step": 36928
    },
    {
      "epoch": 1.6946904685420585,
      "grad_norm": 0.4884439706802368,
      "learning_rate": 4.124677099090898e-06,
      "loss": 0.4201,
      "step": 36929
    },
    {
      "epoch": 1.694736359047313,
      "grad_norm": 0.5269063711166382,
      "learning_rate": 4.124435698684829e-06,
      "loss": 0.4422,
      "step": 36930
    },
    {
      "epoch": 1.6947822495525675,
      "grad_norm": 0.42389658093452454,
      "learning_rate": 4.124194300384216e-06,
      "loss": 0.268,
      "step": 36931
    },
    {
      "epoch": 1.694828140057822,
      "grad_norm": 0.4852316379547119,
      "learning_rate": 4.123952904189638e-06,
      "loss": 0.3641,
      "step": 36932
    },
    {
      "epoch": 1.6948740305630765,
      "grad_norm": 0.48655593395233154,
      "learning_rate": 4.1237115101016775e-06,
      "loss": 0.3818,
      "step": 36933
    },
    {
      "epoch": 1.694919921068331,
      "grad_norm": 0.47353994846343994,
      "learning_rate": 4.123470118120915e-06,
      "loss": 0.3287,
      "step": 36934
    },
    {
      "epoch": 1.6949658115735855,
      "grad_norm": 0.4697560667991638,
      "learning_rate": 4.123228728247927e-06,
      "loss": 0.3717,
      "step": 36935
    },
    {
      "epoch": 1.69501170207884,
      "grad_norm": 0.42585283517837524,
      "learning_rate": 4.1229873404833e-06,
      "loss": 0.3098,
      "step": 36936
    },
    {
      "epoch": 1.6950575925840945,
      "grad_norm": 0.4320126175880432,
      "learning_rate": 4.122745954827611e-06,
      "loss": 0.2928,
      "step": 36937
    },
    {
      "epoch": 1.6951034830893488,
      "grad_norm": 0.4783463776111603,
      "learning_rate": 4.122504571281443e-06,
      "loss": 0.3861,
      "step": 36938
    },
    {
      "epoch": 1.6951493735946033,
      "grad_norm": 0.4963923990726471,
      "learning_rate": 4.122263189845372e-06,
      "loss": 0.3756,
      "step": 36939
    },
    {
      "epoch": 1.6951952640998578,
      "grad_norm": 0.4609127938747406,
      "learning_rate": 4.1220218105199815e-06,
      "loss": 0.3329,
      "step": 36940
    },
    {
      "epoch": 1.695241154605112,
      "grad_norm": 0.4785782992839813,
      "learning_rate": 4.121780433305852e-06,
      "loss": 0.35,
      "step": 36941
    },
    {
      "epoch": 1.6952870451103665,
      "grad_norm": 0.4581732749938965,
      "learning_rate": 4.121539058203563e-06,
      "loss": 0.3468,
      "step": 36942
    },
    {
      "epoch": 1.695332935615621,
      "grad_norm": 0.4667890667915344,
      "learning_rate": 4.121297685213695e-06,
      "loss": 0.3338,
      "step": 36943
    },
    {
      "epoch": 1.6953788261208755,
      "grad_norm": 0.44299349188804626,
      "learning_rate": 4.12105631433683e-06,
      "loss": 0.3241,
      "step": 36944
    },
    {
      "epoch": 1.69542471662613,
      "grad_norm": 0.4530942440032959,
      "learning_rate": 4.1208149455735445e-06,
      "loss": 0.3035,
      "step": 36945
    },
    {
      "epoch": 1.6954706071313845,
      "grad_norm": 0.4427909851074219,
      "learning_rate": 4.120573578924424e-06,
      "loss": 0.3247,
      "step": 36946
    },
    {
      "epoch": 1.695516497636639,
      "grad_norm": 0.4766167998313904,
      "learning_rate": 4.120332214390046e-06,
      "loss": 0.3769,
      "step": 36947
    },
    {
      "epoch": 1.6955623881418935,
      "grad_norm": 0.449705570936203,
      "learning_rate": 4.120090851970991e-06,
      "loss": 0.3215,
      "step": 36948
    },
    {
      "epoch": 1.695608278647148,
      "grad_norm": 0.4725184142589569,
      "learning_rate": 4.1198494916678406e-06,
      "loss": 0.3392,
      "step": 36949
    },
    {
      "epoch": 1.6956541691524025,
      "grad_norm": 0.4967580735683441,
      "learning_rate": 4.119608133481174e-06,
      "loss": 0.3738,
      "step": 36950
    },
    {
      "epoch": 1.6957000596576568,
      "grad_norm": 0.4509488642215729,
      "learning_rate": 4.11936677741157e-06,
      "loss": 0.3501,
      "step": 36951
    },
    {
      "epoch": 1.6957459501629113,
      "grad_norm": 0.4471164345741272,
      "learning_rate": 4.119125423459613e-06,
      "loss": 0.3482,
      "step": 36952
    },
    {
      "epoch": 1.6957918406681658,
      "grad_norm": 0.4719114303588867,
      "learning_rate": 4.118884071625882e-06,
      "loss": 0.3665,
      "step": 36953
    },
    {
      "epoch": 1.69583773117342,
      "grad_norm": 0.5000113248825073,
      "learning_rate": 4.118642721910955e-06,
      "loss": 0.348,
      "step": 36954
    },
    {
      "epoch": 1.6958836216786746,
      "grad_norm": 0.4456068277359009,
      "learning_rate": 4.118401374315416e-06,
      "loss": 0.3441,
      "step": 36955
    },
    {
      "epoch": 1.695929512183929,
      "grad_norm": 0.44855162501335144,
      "learning_rate": 4.1181600288398426e-06,
      "loss": 0.3387,
      "step": 36956
    },
    {
      "epoch": 1.6959754026891836,
      "grad_norm": 0.5031652450561523,
      "learning_rate": 4.1179186854848145e-06,
      "loss": 0.4061,
      "step": 36957
    },
    {
      "epoch": 1.696021293194438,
      "grad_norm": 0.4833148419857025,
      "learning_rate": 4.117677344250915e-06,
      "loss": 0.391,
      "step": 36958
    },
    {
      "epoch": 1.6960671836996926,
      "grad_norm": 0.48636889457702637,
      "learning_rate": 4.1174360051387235e-06,
      "loss": 0.3508,
      "step": 36959
    },
    {
      "epoch": 1.696113074204947,
      "grad_norm": 0.4749028980731964,
      "learning_rate": 4.1171946681488185e-06,
      "loss": 0.3888,
      "step": 36960
    },
    {
      "epoch": 1.6961589647102016,
      "grad_norm": 0.4741758406162262,
      "learning_rate": 4.116953333281783e-06,
      "loss": 0.3767,
      "step": 36961
    },
    {
      "epoch": 1.696204855215456,
      "grad_norm": 0.4679613709449768,
      "learning_rate": 4.1167120005381965e-06,
      "loss": 0.3384,
      "step": 36962
    },
    {
      "epoch": 1.6962507457207106,
      "grad_norm": 0.539152204990387,
      "learning_rate": 4.116470669918636e-06,
      "loss": 0.4761,
      "step": 36963
    },
    {
      "epoch": 1.6962966362259648,
      "grad_norm": 0.49847641587257385,
      "learning_rate": 4.116229341423688e-06,
      "loss": 0.3574,
      "step": 36964
    },
    {
      "epoch": 1.6963425267312193,
      "grad_norm": 0.5194189548492432,
      "learning_rate": 4.115988015053928e-06,
      "loss": 0.4081,
      "step": 36965
    },
    {
      "epoch": 1.6963884172364738,
      "grad_norm": 0.49501198530197144,
      "learning_rate": 4.115746690809936e-06,
      "loss": 0.3572,
      "step": 36966
    },
    {
      "epoch": 1.696434307741728,
      "grad_norm": 0.4745776355266571,
      "learning_rate": 4.115505368692297e-06,
      "loss": 0.4067,
      "step": 36967
    },
    {
      "epoch": 1.6964801982469826,
      "grad_norm": 0.4701031744480133,
      "learning_rate": 4.1152640487015875e-06,
      "loss": 0.3402,
      "step": 36968
    },
    {
      "epoch": 1.696526088752237,
      "grad_norm": 0.4699636995792389,
      "learning_rate": 4.115022730838389e-06,
      "loss": 0.3221,
      "step": 36969
    },
    {
      "epoch": 1.6965719792574916,
      "grad_norm": 0.5030108690261841,
      "learning_rate": 4.114781415103281e-06,
      "loss": 0.4353,
      "step": 36970
    },
    {
      "epoch": 1.696617869762746,
      "grad_norm": 0.4906654357910156,
      "learning_rate": 4.114540101496845e-06,
      "loss": 0.3561,
      "step": 36971
    },
    {
      "epoch": 1.6966637602680006,
      "grad_norm": 0.4536774158477783,
      "learning_rate": 4.114298790019661e-06,
      "loss": 0.3296,
      "step": 36972
    },
    {
      "epoch": 1.696709650773255,
      "grad_norm": 0.5202628970146179,
      "learning_rate": 4.114057480672305e-06,
      "loss": 0.3958,
      "step": 36973
    },
    {
      "epoch": 1.6967555412785096,
      "grad_norm": 0.4745253920555115,
      "learning_rate": 4.113816173455365e-06,
      "loss": 0.3696,
      "step": 36974
    },
    {
      "epoch": 1.696801431783764,
      "grad_norm": 0.4701108932495117,
      "learning_rate": 4.113574868369418e-06,
      "loss": 0.3482,
      "step": 36975
    },
    {
      "epoch": 1.6968473222890184,
      "grad_norm": 0.5014048218727112,
      "learning_rate": 4.113333565415041e-06,
      "loss": 0.4462,
      "step": 36976
    },
    {
      "epoch": 1.6968932127942729,
      "grad_norm": 0.4417453110218048,
      "learning_rate": 4.113092264592818e-06,
      "loss": 0.342,
      "step": 36977
    },
    {
      "epoch": 1.6969391032995274,
      "grad_norm": 0.43841734528541565,
      "learning_rate": 4.112850965903329e-06,
      "loss": 0.3077,
      "step": 36978
    },
    {
      "epoch": 1.6969849938047816,
      "grad_norm": 0.4786892533302307,
      "learning_rate": 4.112609669347152e-06,
      "loss": 0.3592,
      "step": 36979
    },
    {
      "epoch": 1.6970308843100361,
      "grad_norm": 0.43335777521133423,
      "learning_rate": 4.112368374924869e-06,
      "loss": 0.3061,
      "step": 36980
    },
    {
      "epoch": 1.6970767748152906,
      "grad_norm": 0.455598920583725,
      "learning_rate": 4.112127082637062e-06,
      "loss": 0.3128,
      "step": 36981
    },
    {
      "epoch": 1.6971226653205451,
      "grad_norm": 0.42871060967445374,
      "learning_rate": 4.111885792484306e-06,
      "loss": 0.2862,
      "step": 36982
    },
    {
      "epoch": 1.6971685558257996,
      "grad_norm": 0.43354126811027527,
      "learning_rate": 4.111644504467186e-06,
      "loss": 0.2968,
      "step": 36983
    },
    {
      "epoch": 1.6972144463310541,
      "grad_norm": 0.5511528253555298,
      "learning_rate": 4.111403218586282e-06,
      "loss": 0.3639,
      "step": 36984
    },
    {
      "epoch": 1.6972603368363086,
      "grad_norm": 0.47169139981269836,
      "learning_rate": 4.111161934842169e-06,
      "loss": 0.3449,
      "step": 36985
    },
    {
      "epoch": 1.6973062273415631,
      "grad_norm": 0.480385959148407,
      "learning_rate": 4.1109206532354355e-06,
      "loss": 0.4005,
      "step": 36986
    },
    {
      "epoch": 1.6973521178468176,
      "grad_norm": 0.4801252782344818,
      "learning_rate": 4.110679373766655e-06,
      "loss": 0.3955,
      "step": 36987
    },
    {
      "epoch": 1.697398008352072,
      "grad_norm": 0.4385213255882263,
      "learning_rate": 4.110438096436408e-06,
      "loss": 0.2744,
      "step": 36988
    },
    {
      "epoch": 1.6974438988573264,
      "grad_norm": 0.4913046061992645,
      "learning_rate": 4.110196821245279e-06,
      "loss": 0.3776,
      "step": 36989
    },
    {
      "epoch": 1.6974897893625809,
      "grad_norm": 0.4658375680446625,
      "learning_rate": 4.109955548193846e-06,
      "loss": 0.3108,
      "step": 36990
    },
    {
      "epoch": 1.6975356798678354,
      "grad_norm": 0.4557473957538605,
      "learning_rate": 4.109714277282688e-06,
      "loss": 0.3547,
      "step": 36991
    },
    {
      "epoch": 1.6975815703730897,
      "grad_norm": 0.4565676152706146,
      "learning_rate": 4.109473008512387e-06,
      "loss": 0.349,
      "step": 36992
    },
    {
      "epoch": 1.6976274608783442,
      "grad_norm": 0.48243609070777893,
      "learning_rate": 4.1092317418835225e-06,
      "loss": 0.4108,
      "step": 36993
    },
    {
      "epoch": 1.6976733513835986,
      "grad_norm": 0.48242318630218506,
      "learning_rate": 4.108990477396672e-06,
      "loss": 0.3857,
      "step": 36994
    },
    {
      "epoch": 1.6977192418888531,
      "grad_norm": 0.4577937126159668,
      "learning_rate": 4.108749215052421e-06,
      "loss": 0.3565,
      "step": 36995
    },
    {
      "epoch": 1.6977651323941076,
      "grad_norm": 0.473869264125824,
      "learning_rate": 4.1085079548513464e-06,
      "loss": 0.3959,
      "step": 36996
    },
    {
      "epoch": 1.6978110228993621,
      "grad_norm": 0.4573453962802887,
      "learning_rate": 4.108266696794028e-06,
      "loss": 0.3865,
      "step": 36997
    },
    {
      "epoch": 1.6978569134046166,
      "grad_norm": 0.4452921450138092,
      "learning_rate": 4.1080254408810475e-06,
      "loss": 0.3142,
      "step": 36998
    },
    {
      "epoch": 1.6979028039098711,
      "grad_norm": 0.5026951432228088,
      "learning_rate": 4.107784187112984e-06,
      "loss": 0.3865,
      "step": 36999
    },
    {
      "epoch": 1.6979486944151256,
      "grad_norm": 0.48680049180984497,
      "learning_rate": 4.1075429354904165e-06,
      "loss": 0.3786,
      "step": 37000
    },
    {
      "epoch": 1.6979945849203801,
      "grad_norm": 0.4942714273929596,
      "learning_rate": 4.107301686013928e-06,
      "loss": 0.391,
      "step": 37001
    },
    {
      "epoch": 1.6980404754256344,
      "grad_norm": 0.45103299617767334,
      "learning_rate": 4.1070604386840976e-06,
      "loss": 0.3156,
      "step": 37002
    },
    {
      "epoch": 1.698086365930889,
      "grad_norm": 0.42651480436325073,
      "learning_rate": 4.106819193501504e-06,
      "loss": 0.2787,
      "step": 37003
    },
    {
      "epoch": 1.6981322564361434,
      "grad_norm": 0.5122444033622742,
      "learning_rate": 4.10657795046673e-06,
      "loss": 0.3899,
      "step": 37004
    },
    {
      "epoch": 1.6981781469413977,
      "grad_norm": 0.473503053188324,
      "learning_rate": 4.106336709580353e-06,
      "loss": 0.389,
      "step": 37005
    },
    {
      "epoch": 1.6982240374466522,
      "grad_norm": 0.4633901119232178,
      "learning_rate": 4.106095470842955e-06,
      "loss": 0.334,
      "step": 37006
    },
    {
      "epoch": 1.6982699279519067,
      "grad_norm": 0.4847634732723236,
      "learning_rate": 4.1058542342551125e-06,
      "loss": 0.4163,
      "step": 37007
    },
    {
      "epoch": 1.6983158184571612,
      "grad_norm": 0.437478244304657,
      "learning_rate": 4.105612999817412e-06,
      "loss": 0.317,
      "step": 37008
    },
    {
      "epoch": 1.6983617089624157,
      "grad_norm": 0.4665999412536621,
      "learning_rate": 4.105371767530429e-06,
      "loss": 0.3677,
      "step": 37009
    },
    {
      "epoch": 1.6984075994676702,
      "grad_norm": 0.49108701944351196,
      "learning_rate": 4.105130537394742e-06,
      "loss": 0.4048,
      "step": 37010
    },
    {
      "epoch": 1.6984534899729247,
      "grad_norm": 0.48684123158454895,
      "learning_rate": 4.104889309410936e-06,
      "loss": 0.3626,
      "step": 37011
    },
    {
      "epoch": 1.6984993804781792,
      "grad_norm": 0.4597502648830414,
      "learning_rate": 4.104648083579588e-06,
      "loss": 0.3044,
      "step": 37012
    },
    {
      "epoch": 1.6985452709834337,
      "grad_norm": 0.4922095835208893,
      "learning_rate": 4.104406859901279e-06,
      "loss": 0.3678,
      "step": 37013
    },
    {
      "epoch": 1.698591161488688,
      "grad_norm": 0.45404890179634094,
      "learning_rate": 4.104165638376589e-06,
      "loss": 0.33,
      "step": 37014
    },
    {
      "epoch": 1.6986370519939424,
      "grad_norm": 0.4564456045627594,
      "learning_rate": 4.103924419006098e-06,
      "loss": 0.3013,
      "step": 37015
    },
    {
      "epoch": 1.698682942499197,
      "grad_norm": 0.5030897855758667,
      "learning_rate": 4.103683201790384e-06,
      "loss": 0.3745,
      "step": 37016
    },
    {
      "epoch": 1.6987288330044514,
      "grad_norm": 0.48292121291160583,
      "learning_rate": 4.1034419867300315e-06,
      "loss": 0.3684,
      "step": 37017
    },
    {
      "epoch": 1.6987747235097057,
      "grad_norm": 0.50263512134552,
      "learning_rate": 4.1032007738256175e-06,
      "loss": 0.3628,
      "step": 37018
    },
    {
      "epoch": 1.6988206140149602,
      "grad_norm": 0.4769405126571655,
      "learning_rate": 4.1029595630777215e-06,
      "loss": 0.3586,
      "step": 37019
    },
    {
      "epoch": 1.6988665045202147,
      "grad_norm": 0.44763803482055664,
      "learning_rate": 4.102718354486926e-06,
      "loss": 0.3308,
      "step": 37020
    },
    {
      "epoch": 1.6989123950254692,
      "grad_norm": 0.45846760272979736,
      "learning_rate": 4.102477148053809e-06,
      "loss": 0.3034,
      "step": 37021
    },
    {
      "epoch": 1.6989582855307237,
      "grad_norm": 0.49356159567832947,
      "learning_rate": 4.10223594377895e-06,
      "loss": 0.4278,
      "step": 37022
    },
    {
      "epoch": 1.6990041760359782,
      "grad_norm": 0.4713141620159149,
      "learning_rate": 4.101994741662932e-06,
      "loss": 0.3995,
      "step": 37023
    },
    {
      "epoch": 1.6990500665412327,
      "grad_norm": 0.4768821597099304,
      "learning_rate": 4.101753541706333e-06,
      "loss": 0.3236,
      "step": 37024
    },
    {
      "epoch": 1.6990959570464872,
      "grad_norm": 0.45070868730545044,
      "learning_rate": 4.101512343909733e-06,
      "loss": 0.3242,
      "step": 37025
    },
    {
      "epoch": 1.6991418475517417,
      "grad_norm": 0.49519988894462585,
      "learning_rate": 4.101271148273712e-06,
      "loss": 0.4296,
      "step": 37026
    },
    {
      "epoch": 1.699187738056996,
      "grad_norm": 0.46270689368247986,
      "learning_rate": 4.101029954798851e-06,
      "loss": 0.3618,
      "step": 37027
    },
    {
      "epoch": 1.6992336285622505,
      "grad_norm": 0.4458094835281372,
      "learning_rate": 4.1007887634857274e-06,
      "loss": 0.3274,
      "step": 37028
    },
    {
      "epoch": 1.699279519067505,
      "grad_norm": 0.44086140394210815,
      "learning_rate": 4.100547574334925e-06,
      "loss": 0.2969,
      "step": 37029
    },
    {
      "epoch": 1.6993254095727592,
      "grad_norm": 0.4878346025943756,
      "learning_rate": 4.1003063873470225e-06,
      "loss": 0.408,
      "step": 37030
    },
    {
      "epoch": 1.6993713000780137,
      "grad_norm": 0.48761337995529175,
      "learning_rate": 4.100065202522596e-06,
      "loss": 0.4047,
      "step": 37031
    },
    {
      "epoch": 1.6994171905832682,
      "grad_norm": 0.4447671175003052,
      "learning_rate": 4.0998240198622305e-06,
      "loss": 0.292,
      "step": 37032
    },
    {
      "epoch": 1.6994630810885227,
      "grad_norm": 0.4682255685329437,
      "learning_rate": 4.099582839366505e-06,
      "loss": 0.3737,
      "step": 37033
    },
    {
      "epoch": 1.6995089715937772,
      "grad_norm": 0.46399277448654175,
      "learning_rate": 4.099341661035997e-06,
      "loss": 0.3315,
      "step": 37034
    },
    {
      "epoch": 1.6995548620990317,
      "grad_norm": 0.48087140917778015,
      "learning_rate": 4.099100484871289e-06,
      "loss": 0.3575,
      "step": 37035
    },
    {
      "epoch": 1.6996007526042862,
      "grad_norm": 0.4813202917575836,
      "learning_rate": 4.0988593108729605e-06,
      "loss": 0.4075,
      "step": 37036
    },
    {
      "epoch": 1.6996466431095407,
      "grad_norm": 0.49177423119544983,
      "learning_rate": 4.098618139041591e-06,
      "loss": 0.403,
      "step": 37037
    },
    {
      "epoch": 1.6996925336147952,
      "grad_norm": 0.5182430744171143,
      "learning_rate": 4.0983769693777574e-06,
      "loss": 0.4402,
      "step": 37038
    },
    {
      "epoch": 1.6997384241200497,
      "grad_norm": 0.4809049665927887,
      "learning_rate": 4.098135801882045e-06,
      "loss": 0.3671,
      "step": 37039
    },
    {
      "epoch": 1.699784314625304,
      "grad_norm": 0.4783908724784851,
      "learning_rate": 4.097894636555032e-06,
      "loss": 0.3559,
      "step": 37040
    },
    {
      "epoch": 1.6998302051305585,
      "grad_norm": 0.46669304370880127,
      "learning_rate": 4.097653473397296e-06,
      "loss": 0.2959,
      "step": 37041
    },
    {
      "epoch": 1.699876095635813,
      "grad_norm": 0.46887364983558655,
      "learning_rate": 4.097412312409421e-06,
      "loss": 0.2994,
      "step": 37042
    },
    {
      "epoch": 1.6999219861410673,
      "grad_norm": 0.46676933765411377,
      "learning_rate": 4.097171153591983e-06,
      "loss": 0.3938,
      "step": 37043
    },
    {
      "epoch": 1.6999678766463218,
      "grad_norm": 0.46536609530448914,
      "learning_rate": 4.096929996945562e-06,
      "loss": 0.3525,
      "step": 37044
    },
    {
      "epoch": 1.7000137671515763,
      "grad_norm": 0.42363160848617554,
      "learning_rate": 4.09668884247074e-06,
      "loss": 0.2981,
      "step": 37045
    },
    {
      "epoch": 1.7000596576568308,
      "grad_norm": 0.4613165557384491,
      "learning_rate": 4.096447690168098e-06,
      "loss": 0.3306,
      "step": 37046
    },
    {
      "epoch": 1.7001055481620853,
      "grad_norm": 0.49438849091529846,
      "learning_rate": 4.0962065400382125e-06,
      "loss": 0.3742,
      "step": 37047
    },
    {
      "epoch": 1.7001514386673398,
      "grad_norm": 0.4550308883190155,
      "learning_rate": 4.095965392081667e-06,
      "loss": 0.3546,
      "step": 37048
    },
    {
      "epoch": 1.7001973291725943,
      "grad_norm": 0.46803539991378784,
      "learning_rate": 4.095724246299038e-06,
      "loss": 0.3539,
      "step": 37049
    },
    {
      "epoch": 1.7002432196778487,
      "grad_norm": 0.45837870240211487,
      "learning_rate": 4.095483102690905e-06,
      "loss": 0.346,
      "step": 37050
    },
    {
      "epoch": 1.7002891101831032,
      "grad_norm": 0.4857238829135895,
      "learning_rate": 4.095241961257852e-06,
      "loss": 0.4067,
      "step": 37051
    },
    {
      "epoch": 1.7003350006883577,
      "grad_norm": 0.4473886787891388,
      "learning_rate": 4.095000822000458e-06,
      "loss": 0.3103,
      "step": 37052
    },
    {
      "epoch": 1.700380891193612,
      "grad_norm": 0.4641798436641693,
      "learning_rate": 4.094759684919298e-06,
      "loss": 0.3361,
      "step": 37053
    },
    {
      "epoch": 1.7004267816988665,
      "grad_norm": 0.4464186429977417,
      "learning_rate": 4.094518550014957e-06,
      "loss": 0.3375,
      "step": 37054
    },
    {
      "epoch": 1.700472672204121,
      "grad_norm": 0.5191483497619629,
      "learning_rate": 4.094277417288014e-06,
      "loss": 0.3262,
      "step": 37055
    },
    {
      "epoch": 1.7005185627093753,
      "grad_norm": 0.5633576512336731,
      "learning_rate": 4.094036286739045e-06,
      "loss": 0.4973,
      "step": 37056
    },
    {
      "epoch": 1.7005644532146298,
      "grad_norm": 0.5799922943115234,
      "learning_rate": 4.093795158368636e-06,
      "loss": 0.3863,
      "step": 37057
    },
    {
      "epoch": 1.7006103437198843,
      "grad_norm": 0.4663558900356293,
      "learning_rate": 4.093554032177363e-06,
      "loss": 0.2877,
      "step": 37058
    },
    {
      "epoch": 1.7006562342251388,
      "grad_norm": 0.5218594074249268,
      "learning_rate": 4.0933129081658025e-06,
      "loss": 0.3695,
      "step": 37059
    },
    {
      "epoch": 1.7007021247303933,
      "grad_norm": 0.4873916208744049,
      "learning_rate": 4.093071786334542e-06,
      "loss": 0.3537,
      "step": 37060
    },
    {
      "epoch": 1.7007480152356478,
      "grad_norm": 0.44444769620895386,
      "learning_rate": 4.092830666684158e-06,
      "loss": 0.2896,
      "step": 37061
    },
    {
      "epoch": 1.7007939057409023,
      "grad_norm": 0.47003471851348877,
      "learning_rate": 4.092589549215229e-06,
      "loss": 0.3474,
      "step": 37062
    },
    {
      "epoch": 1.7008397962461568,
      "grad_norm": 0.4852973520755768,
      "learning_rate": 4.092348433928336e-06,
      "loss": 0.3661,
      "step": 37063
    },
    {
      "epoch": 1.7008856867514113,
      "grad_norm": 0.46563276648521423,
      "learning_rate": 4.092107320824059e-06,
      "loss": 0.3427,
      "step": 37064
    },
    {
      "epoch": 1.7009315772566656,
      "grad_norm": 0.4912428557872772,
      "learning_rate": 4.091866209902975e-06,
      "loss": 0.3955,
      "step": 37065
    },
    {
      "epoch": 1.70097746776192,
      "grad_norm": 0.4934949278831482,
      "learning_rate": 4.091625101165668e-06,
      "loss": 0.3848,
      "step": 37066
    },
    {
      "epoch": 1.7010233582671745,
      "grad_norm": 0.4814314544200897,
      "learning_rate": 4.091383994612716e-06,
      "loss": 0.3699,
      "step": 37067
    },
    {
      "epoch": 1.7010692487724288,
      "grad_norm": 0.4549257457256317,
      "learning_rate": 4.0911428902446985e-06,
      "loss": 0.2974,
      "step": 37068
    },
    {
      "epoch": 1.7011151392776833,
      "grad_norm": 0.45683541893959045,
      "learning_rate": 4.090901788062196e-06,
      "loss": 0.3483,
      "step": 37069
    },
    {
      "epoch": 1.7011610297829378,
      "grad_norm": 0.4526069164276123,
      "learning_rate": 4.0906606880657875e-06,
      "loss": 0.3255,
      "step": 37070
    },
    {
      "epoch": 1.7012069202881923,
      "grad_norm": 0.49994105100631714,
      "learning_rate": 4.090419590256054e-06,
      "loss": 0.3707,
      "step": 37071
    },
    {
      "epoch": 1.7012528107934468,
      "grad_norm": 0.38907769322395325,
      "learning_rate": 4.090178494633571e-06,
      "loss": 0.2404,
      "step": 37072
    },
    {
      "epoch": 1.7012987012987013,
      "grad_norm": 0.4601591229438782,
      "learning_rate": 4.089937401198924e-06,
      "loss": 0.4,
      "step": 37073
    },
    {
      "epoch": 1.7013445918039558,
      "grad_norm": 0.5101577043533325,
      "learning_rate": 4.089696309952692e-06,
      "loss": 0.3895,
      "step": 37074
    },
    {
      "epoch": 1.7013904823092103,
      "grad_norm": 0.4520625174045563,
      "learning_rate": 4.0894552208954495e-06,
      "loss": 0.3092,
      "step": 37075
    },
    {
      "epoch": 1.7014363728144648,
      "grad_norm": 0.476667195558548,
      "learning_rate": 4.089214134027782e-06,
      "loss": 0.3858,
      "step": 37076
    },
    {
      "epoch": 1.7014822633197193,
      "grad_norm": 0.47687771916389465,
      "learning_rate": 4.088973049350267e-06,
      "loss": 0.3344,
      "step": 37077
    },
    {
      "epoch": 1.7015281538249736,
      "grad_norm": 0.40557971596717834,
      "learning_rate": 4.088731966863483e-06,
      "loss": 0.2828,
      "step": 37078
    },
    {
      "epoch": 1.701574044330228,
      "grad_norm": 0.4913073182106018,
      "learning_rate": 4.088490886568012e-06,
      "loss": 0.3756,
      "step": 37079
    },
    {
      "epoch": 1.7016199348354826,
      "grad_norm": 0.47396451234817505,
      "learning_rate": 4.088249808464434e-06,
      "loss": 0.3503,
      "step": 37080
    },
    {
      "epoch": 1.7016658253407368,
      "grad_norm": 0.4691883325576782,
      "learning_rate": 4.088008732553324e-06,
      "loss": 0.3561,
      "step": 37081
    },
    {
      "epoch": 1.7017117158459913,
      "grad_norm": 0.4907822608947754,
      "learning_rate": 4.087767658835267e-06,
      "loss": 0.3526,
      "step": 37082
    },
    {
      "epoch": 1.7017576063512458,
      "grad_norm": 0.5208011865615845,
      "learning_rate": 4.087526587310842e-06,
      "loss": 0.4337,
      "step": 37083
    },
    {
      "epoch": 1.7018034968565003,
      "grad_norm": 0.4618575870990753,
      "learning_rate": 4.087285517980626e-06,
      "loss": 0.2975,
      "step": 37084
    },
    {
      "epoch": 1.7018493873617548,
      "grad_norm": 0.44503843784332275,
      "learning_rate": 4.087044450845201e-06,
      "loss": 0.285,
      "step": 37085
    },
    {
      "epoch": 1.7018952778670093,
      "grad_norm": 0.4864273965358734,
      "learning_rate": 4.0868033859051465e-06,
      "loss": 0.3988,
      "step": 37086
    },
    {
      "epoch": 1.7019411683722638,
      "grad_norm": 0.49956944584846497,
      "learning_rate": 4.086562323161039e-06,
      "loss": 0.4196,
      "step": 37087
    },
    {
      "epoch": 1.7019870588775183,
      "grad_norm": 0.44127318263053894,
      "learning_rate": 4.086321262613463e-06,
      "loss": 0.3239,
      "step": 37088
    },
    {
      "epoch": 1.7020329493827728,
      "grad_norm": 0.49322354793548584,
      "learning_rate": 4.0860802042629964e-06,
      "loss": 0.4267,
      "step": 37089
    },
    {
      "epoch": 1.7020788398880273,
      "grad_norm": 0.4673802852630615,
      "learning_rate": 4.085839148110218e-06,
      "loss": 0.3372,
      "step": 37090
    },
    {
      "epoch": 1.7021247303932816,
      "grad_norm": 0.4939850866794586,
      "learning_rate": 4.085598094155708e-06,
      "loss": 0.3476,
      "step": 37091
    },
    {
      "epoch": 1.702170620898536,
      "grad_norm": 0.49505728483200073,
      "learning_rate": 4.085357042400047e-06,
      "loss": 0.3968,
      "step": 37092
    },
    {
      "epoch": 1.7022165114037906,
      "grad_norm": 0.4444194436073303,
      "learning_rate": 4.08511599284381e-06,
      "loss": 0.3266,
      "step": 37093
    },
    {
      "epoch": 1.7022624019090449,
      "grad_norm": 0.47243350744247437,
      "learning_rate": 4.084874945487583e-06,
      "loss": 0.3892,
      "step": 37094
    },
    {
      "epoch": 1.7023082924142994,
      "grad_norm": 0.5057694315910339,
      "learning_rate": 4.084633900331943e-06,
      "loss": 0.3765,
      "step": 37095
    },
    {
      "epoch": 1.7023541829195539,
      "grad_norm": 0.4706217348575592,
      "learning_rate": 4.084392857377468e-06,
      "loss": 0.3491,
      "step": 37096
    },
    {
      "epoch": 1.7024000734248084,
      "grad_norm": 0.46859052777290344,
      "learning_rate": 4.084151816624742e-06,
      "loss": 0.3516,
      "step": 37097
    },
    {
      "epoch": 1.7024459639300629,
      "grad_norm": 0.7105806469917297,
      "learning_rate": 4.083910778074341e-06,
      "loss": 0.4175,
      "step": 37098
    },
    {
      "epoch": 1.7024918544353174,
      "grad_norm": 0.4650220572948456,
      "learning_rate": 4.083669741726844e-06,
      "loss": 0.3105,
      "step": 37099
    },
    {
      "epoch": 1.7025377449405719,
      "grad_norm": 0.5274021029472351,
      "learning_rate": 4.083428707582833e-06,
      "loss": 0.3938,
      "step": 37100
    },
    {
      "epoch": 1.7025836354458264,
      "grad_norm": 0.4464446008205414,
      "learning_rate": 4.083187675642887e-06,
      "loss": 0.2969,
      "step": 37101
    },
    {
      "epoch": 1.7026295259510809,
      "grad_norm": 0.48198747634887695,
      "learning_rate": 4.082946645907585e-06,
      "loss": 0.4273,
      "step": 37102
    },
    {
      "epoch": 1.7026754164563351,
      "grad_norm": 0.49331995844841003,
      "learning_rate": 4.0827056183775045e-06,
      "loss": 0.3963,
      "step": 37103
    },
    {
      "epoch": 1.7027213069615896,
      "grad_norm": 0.4680737853050232,
      "learning_rate": 4.082464593053231e-06,
      "loss": 0.3542,
      "step": 37104
    },
    {
      "epoch": 1.7027671974668441,
      "grad_norm": 0.4616132378578186,
      "learning_rate": 4.0822235699353394e-06,
      "loss": 0.339,
      "step": 37105
    },
    {
      "epoch": 1.7028130879720986,
      "grad_norm": 0.5102261304855347,
      "learning_rate": 4.08198254902441e-06,
      "loss": 0.407,
      "step": 37106
    },
    {
      "epoch": 1.702858978477353,
      "grad_norm": 0.5707585215568542,
      "learning_rate": 4.081741530321024e-06,
      "loss": 0.4773,
      "step": 37107
    },
    {
      "epoch": 1.7029048689826074,
      "grad_norm": 0.49624764919281006,
      "learning_rate": 4.081500513825759e-06,
      "loss": 0.4216,
      "step": 37108
    },
    {
      "epoch": 1.702950759487862,
      "grad_norm": 0.4800606369972229,
      "learning_rate": 4.081259499539193e-06,
      "loss": 0.3578,
      "step": 37109
    },
    {
      "epoch": 1.7029966499931164,
      "grad_norm": 0.5712722539901733,
      "learning_rate": 4.081018487461911e-06,
      "loss": 0.4809,
      "step": 37110
    },
    {
      "epoch": 1.703042540498371,
      "grad_norm": 0.4546302556991577,
      "learning_rate": 4.0807774775944895e-06,
      "loss": 0.3085,
      "step": 37111
    },
    {
      "epoch": 1.7030884310036254,
      "grad_norm": 0.43546995520591736,
      "learning_rate": 4.0805364699375064e-06,
      "loss": 0.3383,
      "step": 37112
    },
    {
      "epoch": 1.70313432150888,
      "grad_norm": 0.44864436984062195,
      "learning_rate": 4.0802954644915444e-06,
      "loss": 0.3192,
      "step": 37113
    },
    {
      "epoch": 1.7031802120141344,
      "grad_norm": 0.5189633369445801,
      "learning_rate": 4.080054461257181e-06,
      "loss": 0.4266,
      "step": 37114
    },
    {
      "epoch": 1.7032261025193889,
      "grad_norm": 0.5082626342773438,
      "learning_rate": 4.0798134602349945e-06,
      "loss": 0.3866,
      "step": 37115
    },
    {
      "epoch": 1.7032719930246432,
      "grad_norm": 0.4611167907714844,
      "learning_rate": 4.079572461425569e-06,
      "loss": 0.3434,
      "step": 37116
    },
    {
      "epoch": 1.7033178835298977,
      "grad_norm": 0.4593517482280731,
      "learning_rate": 4.0793314648294806e-06,
      "loss": 0.3618,
      "step": 37117
    },
    {
      "epoch": 1.7033637740351522,
      "grad_norm": 0.44533324241638184,
      "learning_rate": 4.079090470447308e-06,
      "loss": 0.2862,
      "step": 37118
    },
    {
      "epoch": 1.7034096645404064,
      "grad_norm": 0.4900319278240204,
      "learning_rate": 4.0788494782796335e-06,
      "loss": 0.3657,
      "step": 37119
    },
    {
      "epoch": 1.703455555045661,
      "grad_norm": 0.5279750823974609,
      "learning_rate": 4.078608488327036e-06,
      "loss": 0.3911,
      "step": 37120
    },
    {
      "epoch": 1.7035014455509154,
      "grad_norm": 0.5033696293830872,
      "learning_rate": 4.078367500590092e-06,
      "loss": 0.4531,
      "step": 37121
    },
    {
      "epoch": 1.70354733605617,
      "grad_norm": 0.47592678666114807,
      "learning_rate": 4.078126515069385e-06,
      "loss": 0.3454,
      "step": 37122
    },
    {
      "epoch": 1.7035932265614244,
      "grad_norm": 0.47349923849105835,
      "learning_rate": 4.077885531765493e-06,
      "loss": 0.3836,
      "step": 37123
    },
    {
      "epoch": 1.703639117066679,
      "grad_norm": 0.49263712763786316,
      "learning_rate": 4.077644550678993e-06,
      "loss": 0.4334,
      "step": 37124
    },
    {
      "epoch": 1.7036850075719334,
      "grad_norm": 0.4827812910079956,
      "learning_rate": 4.077403571810469e-06,
      "loss": 0.3741,
      "step": 37125
    },
    {
      "epoch": 1.703730898077188,
      "grad_norm": 0.45588603615760803,
      "learning_rate": 4.077162595160498e-06,
      "loss": 0.3399,
      "step": 37126
    },
    {
      "epoch": 1.7037767885824424,
      "grad_norm": 0.4579865634441376,
      "learning_rate": 4.076921620729658e-06,
      "loss": 0.3311,
      "step": 37127
    },
    {
      "epoch": 1.703822679087697,
      "grad_norm": 0.46900448203086853,
      "learning_rate": 4.076680648518532e-06,
      "loss": 0.3216,
      "step": 37128
    },
    {
      "epoch": 1.7038685695929512,
      "grad_norm": 0.4781333804130554,
      "learning_rate": 4.076439678527698e-06,
      "loss": 0.3111,
      "step": 37129
    },
    {
      "epoch": 1.7039144600982057,
      "grad_norm": 0.4797394275665283,
      "learning_rate": 4.076198710757733e-06,
      "loss": 0.4071,
      "step": 37130
    },
    {
      "epoch": 1.7039603506034602,
      "grad_norm": 0.4584585726261139,
      "learning_rate": 4.075957745209219e-06,
      "loss": 0.323,
      "step": 37131
    },
    {
      "epoch": 1.7040062411087145,
      "grad_norm": 0.46239206194877625,
      "learning_rate": 4.075716781882737e-06,
      "loss": 0.3222,
      "step": 37132
    },
    {
      "epoch": 1.704052131613969,
      "grad_norm": 0.5300318598747253,
      "learning_rate": 4.075475820778862e-06,
      "loss": 0.3756,
      "step": 37133
    },
    {
      "epoch": 1.7040980221192235,
      "grad_norm": 0.48300135135650635,
      "learning_rate": 4.075234861898177e-06,
      "loss": 0.3874,
      "step": 37134
    },
    {
      "epoch": 1.704143912624478,
      "grad_norm": 0.44051608443260193,
      "learning_rate": 4.074993905241261e-06,
      "loss": 0.34,
      "step": 37135
    },
    {
      "epoch": 1.7041898031297325,
      "grad_norm": 0.4496227204799652,
      "learning_rate": 4.074752950808692e-06,
      "loss": 0.3287,
      "step": 37136
    },
    {
      "epoch": 1.704235693634987,
      "grad_norm": 0.52524334192276,
      "learning_rate": 4.074511998601049e-06,
      "loss": 0.5143,
      "step": 37137
    },
    {
      "epoch": 1.7042815841402414,
      "grad_norm": 0.46143487095832825,
      "learning_rate": 4.074271048618914e-06,
      "loss": 0.3636,
      "step": 37138
    },
    {
      "epoch": 1.704327474645496,
      "grad_norm": 0.4894908666610718,
      "learning_rate": 4.0740301008628655e-06,
      "loss": 0.4186,
      "step": 37139
    },
    {
      "epoch": 1.7043733651507504,
      "grad_norm": 0.42181724309921265,
      "learning_rate": 4.073789155333481e-06,
      "loss": 0.2919,
      "step": 37140
    },
    {
      "epoch": 1.704419255656005,
      "grad_norm": 0.44759997725486755,
      "learning_rate": 4.073548212031342e-06,
      "loss": 0.3237,
      "step": 37141
    },
    {
      "epoch": 1.7044651461612592,
      "grad_norm": 0.42724454402923584,
      "learning_rate": 4.073307270957027e-06,
      "loss": 0.2999,
      "step": 37142
    },
    {
      "epoch": 1.7045110366665137,
      "grad_norm": 0.48375964164733887,
      "learning_rate": 4.073066332111115e-06,
      "loss": 0.4095,
      "step": 37143
    },
    {
      "epoch": 1.7045569271717682,
      "grad_norm": 0.48780426383018494,
      "learning_rate": 4.072825395494186e-06,
      "loss": 0.3998,
      "step": 37144
    },
    {
      "epoch": 1.7046028176770225,
      "grad_norm": 0.5033005475997925,
      "learning_rate": 4.072584461106821e-06,
      "loss": 0.4568,
      "step": 37145
    },
    {
      "epoch": 1.704648708182277,
      "grad_norm": 0.4485622048377991,
      "learning_rate": 4.0723435289495945e-06,
      "loss": 0.3403,
      "step": 37146
    },
    {
      "epoch": 1.7046945986875315,
      "grad_norm": 0.45271849632263184,
      "learning_rate": 4.072102599023091e-06,
      "loss": 0.3707,
      "step": 37147
    },
    {
      "epoch": 1.704740489192786,
      "grad_norm": 0.43735307455062866,
      "learning_rate": 4.0718616713278885e-06,
      "loss": 0.3403,
      "step": 37148
    },
    {
      "epoch": 1.7047863796980405,
      "grad_norm": 0.4794866740703583,
      "learning_rate": 4.071620745864564e-06,
      "loss": 0.3779,
      "step": 37149
    },
    {
      "epoch": 1.704832270203295,
      "grad_norm": 0.4712761640548706,
      "learning_rate": 4.0713798226337005e-06,
      "loss": 0.3812,
      "step": 37150
    },
    {
      "epoch": 1.7048781607085495,
      "grad_norm": 0.509405791759491,
      "learning_rate": 4.071138901635875e-06,
      "loss": 0.3885,
      "step": 37151
    },
    {
      "epoch": 1.704924051213804,
      "grad_norm": 0.4988270699977875,
      "learning_rate": 4.070897982871665e-06,
      "loss": 0.3889,
      "step": 37152
    },
    {
      "epoch": 1.7049699417190585,
      "grad_norm": 0.4369191825389862,
      "learning_rate": 4.070657066341655e-06,
      "loss": 0.3321,
      "step": 37153
    },
    {
      "epoch": 1.7050158322243127,
      "grad_norm": 0.49078813195228577,
      "learning_rate": 4.07041615204642e-06,
      "loss": 0.4246,
      "step": 37154
    },
    {
      "epoch": 1.7050617227295672,
      "grad_norm": 0.4961918890476227,
      "learning_rate": 4.070175239986541e-06,
      "loss": 0.3688,
      "step": 37155
    },
    {
      "epoch": 1.7051076132348217,
      "grad_norm": 0.4577355980873108,
      "learning_rate": 4.069934330162598e-06,
      "loss": 0.323,
      "step": 37156
    },
    {
      "epoch": 1.705153503740076,
      "grad_norm": 0.4761107563972473,
      "learning_rate": 4.069693422575169e-06,
      "loss": 0.3504,
      "step": 37157
    },
    {
      "epoch": 1.7051993942453305,
      "grad_norm": 0.46491101384162903,
      "learning_rate": 4.0694525172248315e-06,
      "loss": 0.3409,
      "step": 37158
    },
    {
      "epoch": 1.705245284750585,
      "grad_norm": 0.4526561498641968,
      "learning_rate": 4.06921161411217e-06,
      "loss": 0.3291,
      "step": 37159
    },
    {
      "epoch": 1.7052911752558395,
      "grad_norm": 0.4282483756542206,
      "learning_rate": 4.06897071323776e-06,
      "loss": 0.2818,
      "step": 37160
    },
    {
      "epoch": 1.705337065761094,
      "grad_norm": 0.44433850049972534,
      "learning_rate": 4.06872981460218e-06,
      "loss": 0.3024,
      "step": 37161
    },
    {
      "epoch": 1.7053829562663485,
      "grad_norm": 0.46502140164375305,
      "learning_rate": 4.068488918206013e-06,
      "loss": 0.375,
      "step": 37162
    },
    {
      "epoch": 1.705428846771603,
      "grad_norm": 0.5303376913070679,
      "learning_rate": 4.068248024049836e-06,
      "loss": 0.4066,
      "step": 37163
    },
    {
      "epoch": 1.7054747372768575,
      "grad_norm": 0.46291354298591614,
      "learning_rate": 4.068007132134226e-06,
      "loss": 0.39,
      "step": 37164
    },
    {
      "epoch": 1.705520627782112,
      "grad_norm": 0.4432717561721802,
      "learning_rate": 4.067766242459767e-06,
      "loss": 0.3483,
      "step": 37165
    },
    {
      "epoch": 1.7055665182873665,
      "grad_norm": 0.5276281833648682,
      "learning_rate": 4.067525355027037e-06,
      "loss": 0.4376,
      "step": 37166
    },
    {
      "epoch": 1.7056124087926208,
      "grad_norm": 0.43291765451431274,
      "learning_rate": 4.067284469836612e-06,
      "loss": 0.319,
      "step": 37167
    },
    {
      "epoch": 1.7056582992978753,
      "grad_norm": 0.491380512714386,
      "learning_rate": 4.067043586889073e-06,
      "loss": 0.3916,
      "step": 37168
    },
    {
      "epoch": 1.7057041898031298,
      "grad_norm": 0.4536169171333313,
      "learning_rate": 4.066802706185001e-06,
      "loss": 0.3187,
      "step": 37169
    },
    {
      "epoch": 1.705750080308384,
      "grad_norm": 0.460276335477829,
      "learning_rate": 4.0665618277249745e-06,
      "loss": 0.3325,
      "step": 37170
    },
    {
      "epoch": 1.7057959708136385,
      "grad_norm": 0.44502002000808716,
      "learning_rate": 4.066320951509571e-06,
      "loss": 0.3407,
      "step": 37171
    },
    {
      "epoch": 1.705841861318893,
      "grad_norm": 0.50029057264328,
      "learning_rate": 4.0660800775393715e-06,
      "loss": 0.4023,
      "step": 37172
    },
    {
      "epoch": 1.7058877518241475,
      "grad_norm": 0.47610247135162354,
      "learning_rate": 4.065839205814956e-06,
      "loss": 0.4003,
      "step": 37173
    },
    {
      "epoch": 1.705933642329402,
      "grad_norm": 0.4778377413749695,
      "learning_rate": 4.065598336336899e-06,
      "loss": 0.4003,
      "step": 37174
    },
    {
      "epoch": 1.7059795328346565,
      "grad_norm": 0.47889193892478943,
      "learning_rate": 4.065357469105785e-06,
      "loss": 0.3875,
      "step": 37175
    },
    {
      "epoch": 1.706025423339911,
      "grad_norm": 0.531707763671875,
      "learning_rate": 4.065116604122192e-06,
      "loss": 0.4382,
      "step": 37176
    },
    {
      "epoch": 1.7060713138451655,
      "grad_norm": 0.46462202072143555,
      "learning_rate": 4.0648757413866975e-06,
      "loss": 0.3327,
      "step": 37177
    },
    {
      "epoch": 1.70611720435042,
      "grad_norm": 0.4666639566421509,
      "learning_rate": 4.064634880899882e-06,
      "loss": 0.3516,
      "step": 37178
    },
    {
      "epoch": 1.7061630948556745,
      "grad_norm": 0.5211677551269531,
      "learning_rate": 4.064394022662325e-06,
      "loss": 0.4127,
      "step": 37179
    },
    {
      "epoch": 1.7062089853609288,
      "grad_norm": 0.4819161891937256,
      "learning_rate": 4.064153166674603e-06,
      "loss": 0.3506,
      "step": 37180
    },
    {
      "epoch": 1.7062548758661833,
      "grad_norm": 0.5249783396720886,
      "learning_rate": 4.0639123129372995e-06,
      "loss": 0.4482,
      "step": 37181
    },
    {
      "epoch": 1.7063007663714378,
      "grad_norm": 0.5111711025238037,
      "learning_rate": 4.063671461450991e-06,
      "loss": 0.4053,
      "step": 37182
    },
    {
      "epoch": 1.706346656876692,
      "grad_norm": 0.4787129759788513,
      "learning_rate": 4.063430612216257e-06,
      "loss": 0.3952,
      "step": 37183
    },
    {
      "epoch": 1.7063925473819466,
      "grad_norm": 0.47377315163612366,
      "learning_rate": 4.063189765233677e-06,
      "loss": 0.3734,
      "step": 37184
    },
    {
      "epoch": 1.706438437887201,
      "grad_norm": 0.4762354791164398,
      "learning_rate": 4.06294892050383e-06,
      "loss": 0.374,
      "step": 37185
    },
    {
      "epoch": 1.7064843283924556,
      "grad_norm": 0.46650922298431396,
      "learning_rate": 4.062708078027293e-06,
      "loss": 0.3815,
      "step": 37186
    },
    {
      "epoch": 1.70653021889771,
      "grad_norm": 0.4503220021724701,
      "learning_rate": 4.06246723780465e-06,
      "loss": 0.3286,
      "step": 37187
    },
    {
      "epoch": 1.7065761094029646,
      "grad_norm": 0.4705676734447479,
      "learning_rate": 4.062226399836478e-06,
      "loss": 0.3555,
      "step": 37188
    },
    {
      "epoch": 1.706621999908219,
      "grad_norm": 0.4887832701206207,
      "learning_rate": 4.061985564123352e-06,
      "loss": 0.3875,
      "step": 37189
    },
    {
      "epoch": 1.7066678904134736,
      "grad_norm": 0.4402146637439728,
      "learning_rate": 4.061744730665857e-06,
      "loss": 0.3039,
      "step": 37190
    },
    {
      "epoch": 1.706713780918728,
      "grad_norm": 0.4440721273422241,
      "learning_rate": 4.06150389946457e-06,
      "loss": 0.2953,
      "step": 37191
    },
    {
      "epoch": 1.7067596714239823,
      "grad_norm": 0.5012357234954834,
      "learning_rate": 4.0612630705200685e-06,
      "loss": 0.4609,
      "step": 37192
    },
    {
      "epoch": 1.7068055619292368,
      "grad_norm": 0.47520551085472107,
      "learning_rate": 4.061022243832934e-06,
      "loss": 0.329,
      "step": 37193
    },
    {
      "epoch": 1.7068514524344913,
      "grad_norm": 0.4874781668186188,
      "learning_rate": 4.060781419403746e-06,
      "loss": 0.3749,
      "step": 37194
    },
    {
      "epoch": 1.7068973429397458,
      "grad_norm": 0.502419650554657,
      "learning_rate": 4.060540597233078e-06,
      "loss": 0.4328,
      "step": 37195
    },
    {
      "epoch": 1.706943233445,
      "grad_norm": 0.48548516631126404,
      "learning_rate": 4.060299777321517e-06,
      "loss": 0.3993,
      "step": 37196
    },
    {
      "epoch": 1.7069891239502546,
      "grad_norm": 0.4321601092815399,
      "learning_rate": 4.060058959669638e-06,
      "loss": 0.3074,
      "step": 37197
    },
    {
      "epoch": 1.707035014455509,
      "grad_norm": 0.47215190529823303,
      "learning_rate": 4.05981814427802e-06,
      "loss": 0.4133,
      "step": 37198
    },
    {
      "epoch": 1.7070809049607636,
      "grad_norm": 0.47352322936058044,
      "learning_rate": 4.059577331147243e-06,
      "loss": 0.342,
      "step": 37199
    },
    {
      "epoch": 1.707126795466018,
      "grad_norm": 0.49360090494155884,
      "learning_rate": 4.059336520277886e-06,
      "loss": 0.4076,
      "step": 37200
    },
    {
      "epoch": 1.7071726859712726,
      "grad_norm": 0.4465285539627075,
      "learning_rate": 4.059095711670527e-06,
      "loss": 0.3298,
      "step": 37201
    },
    {
      "epoch": 1.707218576476527,
      "grad_norm": 0.4660680294036865,
      "learning_rate": 4.058854905325746e-06,
      "loss": 0.3594,
      "step": 37202
    },
    {
      "epoch": 1.7072644669817816,
      "grad_norm": 0.5783329010009766,
      "learning_rate": 4.058614101244122e-06,
      "loss": 0.4123,
      "step": 37203
    },
    {
      "epoch": 1.707310357487036,
      "grad_norm": 0.4617738127708435,
      "learning_rate": 4.058373299426235e-06,
      "loss": 0.3675,
      "step": 37204
    },
    {
      "epoch": 1.7073562479922904,
      "grad_norm": 0.47495830059051514,
      "learning_rate": 4.058132499872661e-06,
      "loss": 0.3382,
      "step": 37205
    },
    {
      "epoch": 1.7074021384975449,
      "grad_norm": 0.4865867495536804,
      "learning_rate": 4.057891702583983e-06,
      "loss": 0.4318,
      "step": 37206
    },
    {
      "epoch": 1.7074480290027994,
      "grad_norm": 0.47814252972602844,
      "learning_rate": 4.057650907560777e-06,
      "loss": 0.3801,
      "step": 37207
    },
    {
      "epoch": 1.7074939195080536,
      "grad_norm": 0.4484092593193054,
      "learning_rate": 4.057410114803622e-06,
      "loss": 0.3126,
      "step": 37208
    },
    {
      "epoch": 1.7075398100133081,
      "grad_norm": 0.45347321033477783,
      "learning_rate": 4.0571693243131015e-06,
      "loss": 0.3328,
      "step": 37209
    },
    {
      "epoch": 1.7075857005185626,
      "grad_norm": 0.46406063437461853,
      "learning_rate": 4.05692853608979e-06,
      "loss": 0.3584,
      "step": 37210
    },
    {
      "epoch": 1.7076315910238171,
      "grad_norm": 0.4286377727985382,
      "learning_rate": 4.056687750134265e-06,
      "loss": 0.318,
      "step": 37211
    },
    {
      "epoch": 1.7076774815290716,
      "grad_norm": 0.49172669649124146,
      "learning_rate": 4.056446966447111e-06,
      "loss": 0.3725,
      "step": 37212
    },
    {
      "epoch": 1.7077233720343261,
      "grad_norm": 0.4806927442550659,
      "learning_rate": 4.056206185028903e-06,
      "loss": 0.3857,
      "step": 37213
    },
    {
      "epoch": 1.7077692625395806,
      "grad_norm": 0.49589765071868896,
      "learning_rate": 4.055965405880223e-06,
      "loss": 0.387,
      "step": 37214
    },
    {
      "epoch": 1.707815153044835,
      "grad_norm": 0.49459123611450195,
      "learning_rate": 4.055724629001647e-06,
      "loss": 0.4132,
      "step": 37215
    },
    {
      "epoch": 1.7078610435500896,
      "grad_norm": 0.4940870404243469,
      "learning_rate": 4.055483854393756e-06,
      "loss": 0.4572,
      "step": 37216
    },
    {
      "epoch": 1.707906934055344,
      "grad_norm": 0.47986456751823425,
      "learning_rate": 4.0552430820571255e-06,
      "loss": 0.4022,
      "step": 37217
    },
    {
      "epoch": 1.7079528245605984,
      "grad_norm": 0.448303759098053,
      "learning_rate": 4.0550023119923395e-06,
      "loss": 0.3567,
      "step": 37218
    },
    {
      "epoch": 1.7079987150658529,
      "grad_norm": 0.4745594561100006,
      "learning_rate": 4.054761544199975e-06,
      "loss": 0.372,
      "step": 37219
    },
    {
      "epoch": 1.7080446055711074,
      "grad_norm": 0.5128031373023987,
      "learning_rate": 4.054520778680609e-06,
      "loss": 0.4857,
      "step": 37220
    },
    {
      "epoch": 1.7080904960763617,
      "grad_norm": 0.48289692401885986,
      "learning_rate": 4.054280015434824e-06,
      "loss": 0.3718,
      "step": 37221
    },
    {
      "epoch": 1.7081363865816162,
      "grad_norm": 0.4999849200248718,
      "learning_rate": 4.0540392544631965e-06,
      "loss": 0.4086,
      "step": 37222
    },
    {
      "epoch": 1.7081822770868706,
      "grad_norm": 0.47292983531951904,
      "learning_rate": 4.053798495766305e-06,
      "loss": 0.3098,
      "step": 37223
    },
    {
      "epoch": 1.7082281675921251,
      "grad_norm": 0.4618285000324249,
      "learning_rate": 4.05355773934473e-06,
      "loss": 0.3409,
      "step": 37224
    },
    {
      "epoch": 1.7082740580973796,
      "grad_norm": 0.531160295009613,
      "learning_rate": 4.053316985199051e-06,
      "loss": 0.3811,
      "step": 37225
    },
    {
      "epoch": 1.7083199486026341,
      "grad_norm": 0.4821196496486664,
      "learning_rate": 4.053076233329845e-06,
      "loss": 0.3813,
      "step": 37226
    },
    {
      "epoch": 1.7083658391078886,
      "grad_norm": 0.5254929661750793,
      "learning_rate": 4.0528354837376925e-06,
      "loss": 0.4099,
      "step": 37227
    },
    {
      "epoch": 1.7084117296131431,
      "grad_norm": 0.45511549711227417,
      "learning_rate": 4.052594736423171e-06,
      "loss": 0.3534,
      "step": 37228
    },
    {
      "epoch": 1.7084576201183976,
      "grad_norm": 0.42991873621940613,
      "learning_rate": 4.052353991386859e-06,
      "loss": 0.2792,
      "step": 37229
    },
    {
      "epoch": 1.708503510623652,
      "grad_norm": 0.46137696504592896,
      "learning_rate": 4.052113248629338e-06,
      "loss": 0.3106,
      "step": 37230
    },
    {
      "epoch": 1.7085494011289064,
      "grad_norm": 0.43926867842674255,
      "learning_rate": 4.051872508151187e-06,
      "loss": 0.3031,
      "step": 37231
    },
    {
      "epoch": 1.708595291634161,
      "grad_norm": 0.47060224413871765,
      "learning_rate": 4.05163176995298e-06,
      "loss": 0.3461,
      "step": 37232
    },
    {
      "epoch": 1.7086411821394154,
      "grad_norm": 0.46438634395599365,
      "learning_rate": 4.051391034035301e-06,
      "loss": 0.3577,
      "step": 37233
    },
    {
      "epoch": 1.7086870726446697,
      "grad_norm": 0.47936832904815674,
      "learning_rate": 4.051150300398729e-06,
      "loss": 0.3988,
      "step": 37234
    },
    {
      "epoch": 1.7087329631499242,
      "grad_norm": 0.43051546812057495,
      "learning_rate": 4.050909569043839e-06,
      "loss": 0.2831,
      "step": 37235
    },
    {
      "epoch": 1.7087788536551787,
      "grad_norm": 0.4612168073654175,
      "learning_rate": 4.0506688399712115e-06,
      "loss": 0.3627,
      "step": 37236
    },
    {
      "epoch": 1.7088247441604332,
      "grad_norm": 0.45353445410728455,
      "learning_rate": 4.050428113181427e-06,
      "loss": 0.3218,
      "step": 37237
    },
    {
      "epoch": 1.7088706346656877,
      "grad_norm": 0.5035279393196106,
      "learning_rate": 4.0501873886750634e-06,
      "loss": 0.3295,
      "step": 37238
    },
    {
      "epoch": 1.7089165251709422,
      "grad_norm": 0.41984808444976807,
      "learning_rate": 4.049946666452696e-06,
      "loss": 0.2999,
      "step": 37239
    },
    {
      "epoch": 1.7089624156761967,
      "grad_norm": 0.482452392578125,
      "learning_rate": 4.049705946514911e-06,
      "loss": 0.3922,
      "step": 37240
    },
    {
      "epoch": 1.7090083061814512,
      "grad_norm": 0.6534234881401062,
      "learning_rate": 4.0494652288622825e-06,
      "loss": 0.4853,
      "step": 37241
    },
    {
      "epoch": 1.7090541966867057,
      "grad_norm": 0.4699498116970062,
      "learning_rate": 4.049224513495389e-06,
      "loss": 0.3556,
      "step": 37242
    },
    {
      "epoch": 1.70910008719196,
      "grad_norm": 0.4807699918746948,
      "learning_rate": 4.048983800414812e-06,
      "loss": 0.3912,
      "step": 37243
    },
    {
      "epoch": 1.7091459776972144,
      "grad_norm": 0.4927505552768707,
      "learning_rate": 4.048743089621127e-06,
      "loss": 0.4045,
      "step": 37244
    },
    {
      "epoch": 1.709191868202469,
      "grad_norm": 0.4810812771320343,
      "learning_rate": 4.048502381114915e-06,
      "loss": 0.3669,
      "step": 37245
    },
    {
      "epoch": 1.7092377587077232,
      "grad_norm": 0.5033617615699768,
      "learning_rate": 4.048261674896755e-06,
      "loss": 0.3704,
      "step": 37246
    },
    {
      "epoch": 1.7092836492129777,
      "grad_norm": 0.4786507487297058,
      "learning_rate": 4.048020970967226e-06,
      "loss": 0.3909,
      "step": 37247
    },
    {
      "epoch": 1.7093295397182322,
      "grad_norm": 0.4453367292881012,
      "learning_rate": 4.047780269326905e-06,
      "loss": 0.3297,
      "step": 37248
    },
    {
      "epoch": 1.7093754302234867,
      "grad_norm": 0.4948914349079132,
      "learning_rate": 4.047539569976373e-06,
      "loss": 0.3985,
      "step": 37249
    },
    {
      "epoch": 1.7094213207287412,
      "grad_norm": 0.47703272104263306,
      "learning_rate": 4.047298872916207e-06,
      "loss": 0.3578,
      "step": 37250
    },
    {
      "epoch": 1.7094672112339957,
      "grad_norm": 0.48056522011756897,
      "learning_rate": 4.047058178146985e-06,
      "loss": 0.3912,
      "step": 37251
    },
    {
      "epoch": 1.7095131017392502,
      "grad_norm": 0.4651651978492737,
      "learning_rate": 4.04681748566929e-06,
      "loss": 0.3118,
      "step": 37252
    },
    {
      "epoch": 1.7095589922445047,
      "grad_norm": 0.432786226272583,
      "learning_rate": 4.0465767954836975e-06,
      "loss": 0.2708,
      "step": 37253
    },
    {
      "epoch": 1.7096048827497592,
      "grad_norm": 0.47514307498931885,
      "learning_rate": 4.046336107590785e-06,
      "loss": 0.3866,
      "step": 37254
    },
    {
      "epoch": 1.7096507732550137,
      "grad_norm": 0.487131804227829,
      "learning_rate": 4.046095421991136e-06,
      "loss": 0.3375,
      "step": 37255
    },
    {
      "epoch": 1.709696663760268,
      "grad_norm": 0.4613759517669678,
      "learning_rate": 4.045854738685325e-06,
      "loss": 0.3221,
      "step": 37256
    },
    {
      "epoch": 1.7097425542655225,
      "grad_norm": 0.5064923763275146,
      "learning_rate": 4.045614057673931e-06,
      "loss": 0.4493,
      "step": 37257
    },
    {
      "epoch": 1.709788444770777,
      "grad_norm": 0.44667404890060425,
      "learning_rate": 4.045373378957536e-06,
      "loss": 0.3596,
      "step": 37258
    },
    {
      "epoch": 1.7098343352760312,
      "grad_norm": 0.45310187339782715,
      "learning_rate": 4.045132702536716e-06,
      "loss": 0.356,
      "step": 37259
    },
    {
      "epoch": 1.7098802257812857,
      "grad_norm": 0.42656636238098145,
      "learning_rate": 4.044892028412049e-06,
      "loss": 0.3157,
      "step": 37260
    },
    {
      "epoch": 1.7099261162865402,
      "grad_norm": 0.45634812116622925,
      "learning_rate": 4.044651356584117e-06,
      "loss": 0.2996,
      "step": 37261
    },
    {
      "epoch": 1.7099720067917947,
      "grad_norm": 0.4932784140110016,
      "learning_rate": 4.044410687053498e-06,
      "loss": 0.3787,
      "step": 37262
    },
    {
      "epoch": 1.7100178972970492,
      "grad_norm": 0.43509674072265625,
      "learning_rate": 4.044170019820767e-06,
      "loss": 0.3027,
      "step": 37263
    },
    {
      "epoch": 1.7100637878023037,
      "grad_norm": 0.4595159590244293,
      "learning_rate": 4.043929354886508e-06,
      "loss": 0.3401,
      "step": 37264
    },
    {
      "epoch": 1.7101096783075582,
      "grad_norm": 0.4910777807235718,
      "learning_rate": 4.043688692251296e-06,
      "loss": 0.4285,
      "step": 37265
    },
    {
      "epoch": 1.7101555688128127,
      "grad_norm": 0.47283264994621277,
      "learning_rate": 4.043448031915712e-06,
      "loss": 0.3039,
      "step": 37266
    },
    {
      "epoch": 1.7102014593180672,
      "grad_norm": 0.5002070069313049,
      "learning_rate": 4.04320737388033e-06,
      "loss": 0.3848,
      "step": 37267
    },
    {
      "epoch": 1.7102473498233217,
      "grad_norm": 0.500329852104187,
      "learning_rate": 4.042966718145735e-06,
      "loss": 0.4079,
      "step": 37268
    },
    {
      "epoch": 1.710293240328576,
      "grad_norm": 0.4287467896938324,
      "learning_rate": 4.042726064712503e-06,
      "loss": 0.3133,
      "step": 37269
    },
    {
      "epoch": 1.7103391308338305,
      "grad_norm": 0.45750361680984497,
      "learning_rate": 4.042485413581212e-06,
      "loss": 0.3319,
      "step": 37270
    },
    {
      "epoch": 1.710385021339085,
      "grad_norm": 0.4609622657299042,
      "learning_rate": 4.0422447647524424e-06,
      "loss": 0.3658,
      "step": 37271
    },
    {
      "epoch": 1.7104309118443393,
      "grad_norm": 0.5328750014305115,
      "learning_rate": 4.042004118226772e-06,
      "loss": 0.415,
      "step": 37272
    },
    {
      "epoch": 1.7104768023495938,
      "grad_norm": 0.4722987115383148,
      "learning_rate": 4.041763474004777e-06,
      "loss": 0.3503,
      "step": 37273
    },
    {
      "epoch": 1.7105226928548483,
      "grad_norm": 0.4802601635456085,
      "learning_rate": 4.04152283208704e-06,
      "loss": 0.3597,
      "step": 37274
    },
    {
      "epoch": 1.7105685833601028,
      "grad_norm": 0.4661281406879425,
      "learning_rate": 4.04128219247414e-06,
      "loss": 0.3344,
      "step": 37275
    },
    {
      "epoch": 1.7106144738653573,
      "grad_norm": 0.48276233673095703,
      "learning_rate": 4.0410415551666495e-06,
      "loss": 0.3217,
      "step": 37276
    },
    {
      "epoch": 1.7106603643706118,
      "grad_norm": 0.47436198592185974,
      "learning_rate": 4.040800920165155e-06,
      "loss": 0.3533,
      "step": 37277
    },
    {
      "epoch": 1.7107062548758663,
      "grad_norm": 0.5083417296409607,
      "learning_rate": 4.040560287470231e-06,
      "loss": 0.4449,
      "step": 37278
    },
    {
      "epoch": 1.7107521453811207,
      "grad_norm": 0.502764105796814,
      "learning_rate": 4.040319657082455e-06,
      "loss": 0.3807,
      "step": 37279
    },
    {
      "epoch": 1.7107980358863752,
      "grad_norm": 0.4862821698188782,
      "learning_rate": 4.040079029002409e-06,
      "loss": 0.3982,
      "step": 37280
    },
    {
      "epoch": 1.7108439263916295,
      "grad_norm": 0.4743708372116089,
      "learning_rate": 4.0398384032306694e-06,
      "loss": 0.3786,
      "step": 37281
    },
    {
      "epoch": 1.710889816896884,
      "grad_norm": 0.4454227387905121,
      "learning_rate": 4.039597779767814e-06,
      "loss": 0.3083,
      "step": 37282
    },
    {
      "epoch": 1.7109357074021385,
      "grad_norm": 0.4496665298938751,
      "learning_rate": 4.039357158614425e-06,
      "loss": 0.3254,
      "step": 37283
    },
    {
      "epoch": 1.710981597907393,
      "grad_norm": 0.4640202224254608,
      "learning_rate": 4.039116539771078e-06,
      "loss": 0.3255,
      "step": 37284
    },
    {
      "epoch": 1.7110274884126473,
      "grad_norm": 0.456325888633728,
      "learning_rate": 4.038875923238352e-06,
      "loss": 0.3318,
      "step": 37285
    },
    {
      "epoch": 1.7110733789179018,
      "grad_norm": 0.46854954957962036,
      "learning_rate": 4.038635309016827e-06,
      "loss": 0.3089,
      "step": 37286
    },
    {
      "epoch": 1.7111192694231563,
      "grad_norm": 0.47335946559906006,
      "learning_rate": 4.03839469710708e-06,
      "loss": 0.3754,
      "step": 37287
    },
    {
      "epoch": 1.7111651599284108,
      "grad_norm": 0.5227986574172974,
      "learning_rate": 4.038154087509689e-06,
      "loss": 0.4124,
      "step": 37288
    },
    {
      "epoch": 1.7112110504336653,
      "grad_norm": 0.5032994151115417,
      "learning_rate": 4.037913480225236e-06,
      "loss": 0.3725,
      "step": 37289
    },
    {
      "epoch": 1.7112569409389198,
      "grad_norm": 0.5717638731002808,
      "learning_rate": 4.037672875254297e-06,
      "loss": 0.3159,
      "step": 37290
    },
    {
      "epoch": 1.7113028314441743,
      "grad_norm": 0.4264291226863861,
      "learning_rate": 4.03743227259745e-06,
      "loss": 0.2861,
      "step": 37291
    },
    {
      "epoch": 1.7113487219494288,
      "grad_norm": 0.44525814056396484,
      "learning_rate": 4.037191672255277e-06,
      "loss": 0.297,
      "step": 37292
    },
    {
      "epoch": 1.7113946124546833,
      "grad_norm": 0.5035507678985596,
      "learning_rate": 4.036951074228352e-06,
      "loss": 0.3295,
      "step": 37293
    },
    {
      "epoch": 1.7114405029599375,
      "grad_norm": 0.43730267882347107,
      "learning_rate": 4.036710478517254e-06,
      "loss": 0.2824,
      "step": 37294
    },
    {
      "epoch": 1.711486393465192,
      "grad_norm": 0.49828433990478516,
      "learning_rate": 4.036469885122567e-06,
      "loss": 0.3226,
      "step": 37295
    },
    {
      "epoch": 1.7115322839704465,
      "grad_norm": 0.48228147625923157,
      "learning_rate": 4.036229294044864e-06,
      "loss": 0.3638,
      "step": 37296
    },
    {
      "epoch": 1.7115781744757008,
      "grad_norm": 0.495972603559494,
      "learning_rate": 4.035988705284725e-06,
      "loss": 0.349,
      "step": 37297
    },
    {
      "epoch": 1.7116240649809553,
      "grad_norm": 0.4948122799396515,
      "learning_rate": 4.0357481188427305e-06,
      "loss": 0.3537,
      "step": 37298
    },
    {
      "epoch": 1.7116699554862098,
      "grad_norm": 0.44758141040802,
      "learning_rate": 4.035507534719457e-06,
      "loss": 0.3252,
      "step": 37299
    },
    {
      "epoch": 1.7117158459914643,
      "grad_norm": 0.4738883972167969,
      "learning_rate": 4.035266952915484e-06,
      "loss": 0.3553,
      "step": 37300
    },
    {
      "epoch": 1.7117617364967188,
      "grad_norm": 0.4827274978160858,
      "learning_rate": 4.035026373431387e-06,
      "loss": 0.3836,
      "step": 37301
    },
    {
      "epoch": 1.7118076270019733,
      "grad_norm": 0.5196304321289062,
      "learning_rate": 4.034785796267749e-06,
      "loss": 0.4134,
      "step": 37302
    },
    {
      "epoch": 1.7118535175072278,
      "grad_norm": 0.47551608085632324,
      "learning_rate": 4.034545221425147e-06,
      "loss": 0.3786,
      "step": 37303
    },
    {
      "epoch": 1.7118994080124823,
      "grad_norm": 0.48373571038246155,
      "learning_rate": 4.034304648904156e-06,
      "loss": 0.3692,
      "step": 37304
    },
    {
      "epoch": 1.7119452985177368,
      "grad_norm": 0.4583427608013153,
      "learning_rate": 4.03406407870536e-06,
      "loss": 0.3214,
      "step": 37305
    },
    {
      "epoch": 1.7119911890229913,
      "grad_norm": 0.4982357323169708,
      "learning_rate": 4.033823510829335e-06,
      "loss": 0.4239,
      "step": 37306
    },
    {
      "epoch": 1.7120370795282456,
      "grad_norm": 0.4551074802875519,
      "learning_rate": 4.03358294527666e-06,
      "loss": 0.3636,
      "step": 37307
    },
    {
      "epoch": 1.7120829700335,
      "grad_norm": 0.47619926929473877,
      "learning_rate": 4.033342382047911e-06,
      "loss": 0.4229,
      "step": 37308
    },
    {
      "epoch": 1.7121288605387546,
      "grad_norm": 0.4577862620353699,
      "learning_rate": 4.033101821143671e-06,
      "loss": 0.3404,
      "step": 37309
    },
    {
      "epoch": 1.7121747510440088,
      "grad_norm": 0.4380928575992584,
      "learning_rate": 4.032861262564513e-06,
      "loss": 0.3483,
      "step": 37310
    },
    {
      "epoch": 1.7122206415492633,
      "grad_norm": 0.4592551589012146,
      "learning_rate": 4.032620706311021e-06,
      "loss": 0.3878,
      "step": 37311
    },
    {
      "epoch": 1.7122665320545178,
      "grad_norm": 0.4703465700149536,
      "learning_rate": 4.03238015238377e-06,
      "loss": 0.352,
      "step": 37312
    },
    {
      "epoch": 1.7123124225597723,
      "grad_norm": 0.4664079248905182,
      "learning_rate": 4.032139600783339e-06,
      "loss": 0.3273,
      "step": 37313
    },
    {
      "epoch": 1.7123583130650268,
      "grad_norm": 0.48494628071784973,
      "learning_rate": 4.031899051510307e-06,
      "loss": 0.3484,
      "step": 37314
    },
    {
      "epoch": 1.7124042035702813,
      "grad_norm": 0.4541513919830322,
      "learning_rate": 4.031658504565253e-06,
      "loss": 0.3646,
      "step": 37315
    },
    {
      "epoch": 1.7124500940755358,
      "grad_norm": 0.4337867498397827,
      "learning_rate": 4.031417959948753e-06,
      "loss": 0.2987,
      "step": 37316
    },
    {
      "epoch": 1.7124959845807903,
      "grad_norm": 0.4954255521297455,
      "learning_rate": 4.031177417661388e-06,
      "loss": 0.4118,
      "step": 37317
    },
    {
      "epoch": 1.7125418750860448,
      "grad_norm": 0.5105871558189392,
      "learning_rate": 4.030936877703736e-06,
      "loss": 0.4075,
      "step": 37318
    },
    {
      "epoch": 1.712587765591299,
      "grad_norm": 0.49367764592170715,
      "learning_rate": 4.0306963400763745e-06,
      "loss": 0.3406,
      "step": 37319
    },
    {
      "epoch": 1.7126336560965536,
      "grad_norm": 0.5028600692749023,
      "learning_rate": 4.030455804779883e-06,
      "loss": 0.4086,
      "step": 37320
    },
    {
      "epoch": 1.712679546601808,
      "grad_norm": 0.48203909397125244,
      "learning_rate": 4.030215271814839e-06,
      "loss": 0.4042,
      "step": 37321
    },
    {
      "epoch": 1.7127254371070626,
      "grad_norm": 0.4726642072200775,
      "learning_rate": 4.02997474118182e-06,
      "loss": 0.33,
      "step": 37322
    },
    {
      "epoch": 1.7127713276123169,
      "grad_norm": 0.5309221744537354,
      "learning_rate": 4.029734212881408e-06,
      "loss": 0.4032,
      "step": 37323
    },
    {
      "epoch": 1.7128172181175714,
      "grad_norm": 0.4851519465446472,
      "learning_rate": 4.0294936869141776e-06,
      "loss": 0.3516,
      "step": 37324
    },
    {
      "epoch": 1.7128631086228259,
      "grad_norm": 0.44537079334259033,
      "learning_rate": 4.029253163280707e-06,
      "loss": 0.3075,
      "step": 37325
    },
    {
      "epoch": 1.7129089991280804,
      "grad_norm": 0.4342617392539978,
      "learning_rate": 4.0290126419815785e-06,
      "loss": 0.3292,
      "step": 37326
    },
    {
      "epoch": 1.7129548896333349,
      "grad_norm": 0.47228944301605225,
      "learning_rate": 4.028772123017368e-06,
      "loss": 0.3569,
      "step": 37327
    },
    {
      "epoch": 1.7130007801385894,
      "grad_norm": 0.5286669135093689,
      "learning_rate": 4.028531606388653e-06,
      "loss": 0.3839,
      "step": 37328
    },
    {
      "epoch": 1.7130466706438439,
      "grad_norm": 0.4851683974266052,
      "learning_rate": 4.028291092096013e-06,
      "loss": 0.3847,
      "step": 37329
    },
    {
      "epoch": 1.7130925611490984,
      "grad_norm": 0.4726741313934326,
      "learning_rate": 4.028050580140027e-06,
      "loss": 0.3779,
      "step": 37330
    },
    {
      "epoch": 1.7131384516543529,
      "grad_norm": 0.4778311252593994,
      "learning_rate": 4.027810070521273e-06,
      "loss": 0.3543,
      "step": 37331
    },
    {
      "epoch": 1.7131843421596071,
      "grad_norm": 0.49756529927253723,
      "learning_rate": 4.027569563240326e-06,
      "loss": 0.3535,
      "step": 37332
    },
    {
      "epoch": 1.7132302326648616,
      "grad_norm": 0.4910120964050293,
      "learning_rate": 4.02732905829777e-06,
      "loss": 0.4271,
      "step": 37333
    },
    {
      "epoch": 1.7132761231701161,
      "grad_norm": 0.47159597277641296,
      "learning_rate": 4.027088555694181e-06,
      "loss": 0.3272,
      "step": 37334
    },
    {
      "epoch": 1.7133220136753704,
      "grad_norm": 0.46531111001968384,
      "learning_rate": 4.026848055430135e-06,
      "loss": 0.3644,
      "step": 37335
    },
    {
      "epoch": 1.713367904180625,
      "grad_norm": 0.4460766613483429,
      "learning_rate": 4.026607557506214e-06,
      "loss": 0.3306,
      "step": 37336
    },
    {
      "epoch": 1.7134137946858794,
      "grad_norm": 0.49223431944847107,
      "learning_rate": 4.026367061922994e-06,
      "loss": 0.4107,
      "step": 37337
    },
    {
      "epoch": 1.713459685191134,
      "grad_norm": 0.4411812722682953,
      "learning_rate": 4.026126568681052e-06,
      "loss": 0.2922,
      "step": 37338
    },
    {
      "epoch": 1.7135055756963884,
      "grad_norm": 0.4565831422805786,
      "learning_rate": 4.02588607778097e-06,
      "loss": 0.3368,
      "step": 37339
    },
    {
      "epoch": 1.713551466201643,
      "grad_norm": 0.5011718273162842,
      "learning_rate": 4.025645589223326e-06,
      "loss": 0.379,
      "step": 37340
    },
    {
      "epoch": 1.7135973567068974,
      "grad_norm": 0.45208147168159485,
      "learning_rate": 4.025405103008694e-06,
      "loss": 0.3199,
      "step": 37341
    },
    {
      "epoch": 1.7136432472121519,
      "grad_norm": 0.47122251987457275,
      "learning_rate": 4.025164619137658e-06,
      "loss": 0.3283,
      "step": 37342
    },
    {
      "epoch": 1.7136891377174064,
      "grad_norm": 0.47793155908584595,
      "learning_rate": 4.024924137610792e-06,
      "loss": 0.3656,
      "step": 37343
    },
    {
      "epoch": 1.7137350282226609,
      "grad_norm": 0.4862075448036194,
      "learning_rate": 4.024683658428675e-06,
      "loss": 0.335,
      "step": 37344
    },
    {
      "epoch": 1.7137809187279152,
      "grad_norm": 0.4497852623462677,
      "learning_rate": 4.024443181591887e-06,
      "loss": 0.3136,
      "step": 37345
    },
    {
      "epoch": 1.7138268092331697,
      "grad_norm": 0.457185834646225,
      "learning_rate": 4.024202707101006e-06,
      "loss": 0.3414,
      "step": 37346
    },
    {
      "epoch": 1.7138726997384242,
      "grad_norm": 0.4834270477294922,
      "learning_rate": 4.023962234956607e-06,
      "loss": 0.3761,
      "step": 37347
    },
    {
      "epoch": 1.7139185902436784,
      "grad_norm": 0.45955580472946167,
      "learning_rate": 4.023721765159272e-06,
      "loss": 0.3563,
      "step": 37348
    },
    {
      "epoch": 1.713964480748933,
      "grad_norm": 0.4962072968482971,
      "learning_rate": 4.023481297709579e-06,
      "loss": 0.3704,
      "step": 37349
    },
    {
      "epoch": 1.7140103712541874,
      "grad_norm": 0.48209697008132935,
      "learning_rate": 4.023240832608104e-06,
      "loss": 0.3511,
      "step": 37350
    },
    {
      "epoch": 1.714056261759442,
      "grad_norm": 0.45875638723373413,
      "learning_rate": 4.023000369855427e-06,
      "loss": 0.3985,
      "step": 37351
    },
    {
      "epoch": 1.7141021522646964,
      "grad_norm": 0.4975643754005432,
      "learning_rate": 4.022759909452127e-06,
      "loss": 0.3827,
      "step": 37352
    },
    {
      "epoch": 1.714148042769951,
      "grad_norm": 0.4814925491809845,
      "learning_rate": 4.0225194513987776e-06,
      "loss": 0.4078,
      "step": 37353
    },
    {
      "epoch": 1.7141939332752054,
      "grad_norm": 0.43571585416793823,
      "learning_rate": 4.0222789956959635e-06,
      "loss": 0.3173,
      "step": 37354
    },
    {
      "epoch": 1.71423982378046,
      "grad_norm": 0.5095200538635254,
      "learning_rate": 4.02203854234426e-06,
      "loss": 0.4408,
      "step": 37355
    },
    {
      "epoch": 1.7142857142857144,
      "grad_norm": 0.43497005105018616,
      "learning_rate": 4.021798091344243e-06,
      "loss": 0.3166,
      "step": 37356
    },
    {
      "epoch": 1.714331604790969,
      "grad_norm": 0.45422419905662537,
      "learning_rate": 4.021557642696494e-06,
      "loss": 0.3567,
      "step": 37357
    },
    {
      "epoch": 1.7143774952962232,
      "grad_norm": 0.44679537415504456,
      "learning_rate": 4.021317196401591e-06,
      "loss": 0.295,
      "step": 37358
    },
    {
      "epoch": 1.7144233858014777,
      "grad_norm": 0.45649194717407227,
      "learning_rate": 4.021076752460108e-06,
      "loss": 0.3431,
      "step": 37359
    },
    {
      "epoch": 1.7144692763067322,
      "grad_norm": 0.4441635012626648,
      "learning_rate": 4.02083631087263e-06,
      "loss": 0.2946,
      "step": 37360
    },
    {
      "epoch": 1.7145151668119865,
      "grad_norm": 0.6025425791740417,
      "learning_rate": 4.020595871639731e-06,
      "loss": 0.4325,
      "step": 37361
    },
    {
      "epoch": 1.714561057317241,
      "grad_norm": 0.48729264736175537,
      "learning_rate": 4.020355434761989e-06,
      "loss": 0.394,
      "step": 37362
    },
    {
      "epoch": 1.7146069478224955,
      "grad_norm": 0.46638697385787964,
      "learning_rate": 4.020115000239983e-06,
      "loss": 0.3257,
      "step": 37363
    },
    {
      "epoch": 1.71465283832775,
      "grad_norm": 0.44915318489074707,
      "learning_rate": 4.019874568074292e-06,
      "loss": 0.3291,
      "step": 37364
    },
    {
      "epoch": 1.7146987288330044,
      "grad_norm": 0.4523506760597229,
      "learning_rate": 4.019634138265493e-06,
      "loss": 0.2986,
      "step": 37365
    },
    {
      "epoch": 1.714744619338259,
      "grad_norm": 0.4175998866558075,
      "learning_rate": 4.019393710814163e-06,
      "loss": 0.3367,
      "step": 37366
    },
    {
      "epoch": 1.7147905098435134,
      "grad_norm": 0.45889878273010254,
      "learning_rate": 4.019153285720884e-06,
      "loss": 0.3417,
      "step": 37367
    },
    {
      "epoch": 1.714836400348768,
      "grad_norm": 0.4952715337276459,
      "learning_rate": 4.018912862986231e-06,
      "loss": 0.4329,
      "step": 37368
    },
    {
      "epoch": 1.7148822908540224,
      "grad_norm": 0.4462283253669739,
      "learning_rate": 4.018672442610781e-06,
      "loss": 0.3043,
      "step": 37369
    },
    {
      "epoch": 1.7149281813592767,
      "grad_norm": 0.5120095014572144,
      "learning_rate": 4.018432024595116e-06,
      "loss": 0.4132,
      "step": 37370
    },
    {
      "epoch": 1.7149740718645312,
      "grad_norm": 0.4621037244796753,
      "learning_rate": 4.018191608939812e-06,
      "loss": 0.3354,
      "step": 37371
    },
    {
      "epoch": 1.7150199623697857,
      "grad_norm": 0.4522314965724945,
      "learning_rate": 4.017951195645446e-06,
      "loss": 0.3497,
      "step": 37372
    },
    {
      "epoch": 1.7150658528750402,
      "grad_norm": 0.4655211865901947,
      "learning_rate": 4.017710784712599e-06,
      "loss": 0.3649,
      "step": 37373
    },
    {
      "epoch": 1.7151117433802945,
      "grad_norm": 0.48257794976234436,
      "learning_rate": 4.0174703761418486e-06,
      "loss": 0.4107,
      "step": 37374
    },
    {
      "epoch": 1.715157633885549,
      "grad_norm": 0.45438098907470703,
      "learning_rate": 4.017229969933768e-06,
      "loss": 0.3178,
      "step": 37375
    },
    {
      "epoch": 1.7152035243908035,
      "grad_norm": 0.453106552362442,
      "learning_rate": 4.016989566088942e-06,
      "loss": 0.3797,
      "step": 37376
    },
    {
      "epoch": 1.715249414896058,
      "grad_norm": 0.4805621802806854,
      "learning_rate": 4.016749164607946e-06,
      "loss": 0.3785,
      "step": 37377
    },
    {
      "epoch": 1.7152953054013125,
      "grad_norm": 0.4582268297672272,
      "learning_rate": 4.016508765491356e-06,
      "loss": 0.3843,
      "step": 37378
    },
    {
      "epoch": 1.715341195906567,
      "grad_norm": 0.47891950607299805,
      "learning_rate": 4.016268368739754e-06,
      "loss": 0.4186,
      "step": 37379
    },
    {
      "epoch": 1.7153870864118215,
      "grad_norm": 0.4825790524482727,
      "learning_rate": 4.016027974353716e-06,
      "loss": 0.3558,
      "step": 37380
    },
    {
      "epoch": 1.715432976917076,
      "grad_norm": 0.4485175907611847,
      "learning_rate": 4.015787582333818e-06,
      "loss": 0.3359,
      "step": 37381
    },
    {
      "epoch": 1.7154788674223305,
      "grad_norm": 0.46191123127937317,
      "learning_rate": 4.015547192680642e-06,
      "loss": 0.344,
      "step": 37382
    },
    {
      "epoch": 1.7155247579275847,
      "grad_norm": 0.4657062888145447,
      "learning_rate": 4.015306805394765e-06,
      "loss": 0.3186,
      "step": 37383
    },
    {
      "epoch": 1.7155706484328392,
      "grad_norm": 0.4660350978374481,
      "learning_rate": 4.015066420476763e-06,
      "loss": 0.3336,
      "step": 37384
    },
    {
      "epoch": 1.7156165389380937,
      "grad_norm": 0.4915370047092438,
      "learning_rate": 4.014826037927216e-06,
      "loss": 0.3514,
      "step": 37385
    },
    {
      "epoch": 1.715662429443348,
      "grad_norm": 0.5144296884536743,
      "learning_rate": 4.0145856577467015e-06,
      "loss": 0.3126,
      "step": 37386
    },
    {
      "epoch": 1.7157083199486025,
      "grad_norm": 0.4534396529197693,
      "learning_rate": 4.014345279935796e-06,
      "loss": 0.3493,
      "step": 37387
    },
    {
      "epoch": 1.715754210453857,
      "grad_norm": 0.47214728593826294,
      "learning_rate": 4.014104904495082e-06,
      "loss": 0.3876,
      "step": 37388
    },
    {
      "epoch": 1.7158001009591115,
      "grad_norm": 0.5021201968193054,
      "learning_rate": 4.0138645314251335e-06,
      "loss": 0.3942,
      "step": 37389
    },
    {
      "epoch": 1.715845991464366,
      "grad_norm": 0.5134985446929932,
      "learning_rate": 4.013624160726527e-06,
      "loss": 0.452,
      "step": 37390
    },
    {
      "epoch": 1.7158918819696205,
      "grad_norm": 0.512238621711731,
      "learning_rate": 4.013383792399845e-06,
      "loss": 0.4439,
      "step": 37391
    },
    {
      "epoch": 1.715937772474875,
      "grad_norm": 0.4435451924800873,
      "learning_rate": 4.013143426445665e-06,
      "loss": 0.3203,
      "step": 37392
    },
    {
      "epoch": 1.7159836629801295,
      "grad_norm": 0.44603249430656433,
      "learning_rate": 4.012903062864561e-06,
      "loss": 0.3101,
      "step": 37393
    },
    {
      "epoch": 1.716029553485384,
      "grad_norm": 0.46990033984184265,
      "learning_rate": 4.012662701657115e-06,
      "loss": 0.3466,
      "step": 37394
    },
    {
      "epoch": 1.7160754439906385,
      "grad_norm": 0.4934183359146118,
      "learning_rate": 4.012422342823904e-06,
      "loss": 0.3931,
      "step": 37395
    },
    {
      "epoch": 1.7161213344958928,
      "grad_norm": 0.4806623160839081,
      "learning_rate": 4.012181986365506e-06,
      "loss": 0.3985,
      "step": 37396
    },
    {
      "epoch": 1.7161672250011473,
      "grad_norm": 0.48924383521080017,
      "learning_rate": 4.0119416322824956e-06,
      "loss": 0.3561,
      "step": 37397
    },
    {
      "epoch": 1.7162131155064018,
      "grad_norm": 0.48046043515205383,
      "learning_rate": 4.011701280575456e-06,
      "loss": 0.3845,
      "step": 37398
    },
    {
      "epoch": 1.716259006011656,
      "grad_norm": 0.489839106798172,
      "learning_rate": 4.011460931244963e-06,
      "loss": 0.3783,
      "step": 37399
    },
    {
      "epoch": 1.7163048965169105,
      "grad_norm": 0.490020751953125,
      "learning_rate": 4.0112205842915936e-06,
      "loss": 0.3422,
      "step": 37400
    },
    {
      "epoch": 1.716350787022165,
      "grad_norm": 0.4628543555736542,
      "learning_rate": 4.010980239715928e-06,
      "loss": 0.3482,
      "step": 37401
    },
    {
      "epoch": 1.7163966775274195,
      "grad_norm": 0.4331725239753723,
      "learning_rate": 4.0107398975185415e-06,
      "loss": 0.3366,
      "step": 37402
    },
    {
      "epoch": 1.716442568032674,
      "grad_norm": 0.44855016469955444,
      "learning_rate": 4.0104995577000125e-06,
      "loss": 0.32,
      "step": 37403
    },
    {
      "epoch": 1.7164884585379285,
      "grad_norm": 0.4676326811313629,
      "learning_rate": 4.010259220260921e-06,
      "loss": 0.3487,
      "step": 37404
    },
    {
      "epoch": 1.716534349043183,
      "grad_norm": 0.49320530891418457,
      "learning_rate": 4.010018885201844e-06,
      "loss": 0.4193,
      "step": 37405
    },
    {
      "epoch": 1.7165802395484375,
      "grad_norm": 0.4477243423461914,
      "learning_rate": 4.009778552523358e-06,
      "loss": 0.3132,
      "step": 37406
    },
    {
      "epoch": 1.716626130053692,
      "grad_norm": 0.5851663947105408,
      "learning_rate": 4.009538222226044e-06,
      "loss": 0.3872,
      "step": 37407
    },
    {
      "epoch": 1.7166720205589463,
      "grad_norm": 0.6327536106109619,
      "learning_rate": 4.0092978943104775e-06,
      "loss": 0.3158,
      "step": 37408
    },
    {
      "epoch": 1.7167179110642008,
      "grad_norm": 0.5061814785003662,
      "learning_rate": 4.009057568777234e-06,
      "loss": 0.4366,
      "step": 37409
    },
    {
      "epoch": 1.7167638015694553,
      "grad_norm": 0.47651946544647217,
      "learning_rate": 4.008817245626898e-06,
      "loss": 0.3467,
      "step": 37410
    },
    {
      "epoch": 1.7168096920747098,
      "grad_norm": 0.4591066539287567,
      "learning_rate": 4.008576924860042e-06,
      "loss": 0.2832,
      "step": 37411
    },
    {
      "epoch": 1.716855582579964,
      "grad_norm": 0.49064821004867554,
      "learning_rate": 4.008336606477244e-06,
      "loss": 0.3977,
      "step": 37412
    },
    {
      "epoch": 1.7169014730852186,
      "grad_norm": 0.4682297110557556,
      "learning_rate": 4.008096290479087e-06,
      "loss": 0.3447,
      "step": 37413
    },
    {
      "epoch": 1.716947363590473,
      "grad_norm": 0.43156325817108154,
      "learning_rate": 4.0078559768661435e-06,
      "loss": 0.2768,
      "step": 37414
    },
    {
      "epoch": 1.7169932540957276,
      "grad_norm": 0.5454644560813904,
      "learning_rate": 4.007615665638993e-06,
      "loss": 0.4423,
      "step": 37415
    },
    {
      "epoch": 1.717039144600982,
      "grad_norm": 0.4622335731983185,
      "learning_rate": 4.007375356798214e-06,
      "loss": 0.3481,
      "step": 37416
    },
    {
      "epoch": 1.7170850351062366,
      "grad_norm": 0.5088116526603699,
      "learning_rate": 4.007135050344385e-06,
      "loss": 0.3448,
      "step": 37417
    },
    {
      "epoch": 1.717130925611491,
      "grad_norm": 0.4551600217819214,
      "learning_rate": 4.00689474627808e-06,
      "loss": 0.3614,
      "step": 37418
    },
    {
      "epoch": 1.7171768161167456,
      "grad_norm": 0.44858092069625854,
      "learning_rate": 4.006654444599882e-06,
      "loss": 0.2912,
      "step": 37419
    },
    {
      "epoch": 1.717222706622,
      "grad_norm": 0.8074127435684204,
      "learning_rate": 4.006414145310367e-06,
      "loss": 0.4605,
      "step": 37420
    },
    {
      "epoch": 1.7172685971272543,
      "grad_norm": 0.45256632566452026,
      "learning_rate": 4.00617384841011e-06,
      "loss": 0.3092,
      "step": 37421
    },
    {
      "epoch": 1.7173144876325088,
      "grad_norm": 0.5214881300926208,
      "learning_rate": 4.005933553899693e-06,
      "loss": 0.4056,
      "step": 37422
    },
    {
      "epoch": 1.7173603781377633,
      "grad_norm": 0.5207011103630066,
      "learning_rate": 4.005693261779693e-06,
      "loss": 0.4571,
      "step": 37423
    },
    {
      "epoch": 1.7174062686430176,
      "grad_norm": 0.4614640176296234,
      "learning_rate": 4.005452972050684e-06,
      "loss": 0.3546,
      "step": 37424
    },
    {
      "epoch": 1.717452159148272,
      "grad_norm": 0.4548502564430237,
      "learning_rate": 4.005212684713249e-06,
      "loss": 0.312,
      "step": 37425
    },
    {
      "epoch": 1.7174980496535266,
      "grad_norm": 0.4582907259464264,
      "learning_rate": 4.0049723997679625e-06,
      "loss": 0.358,
      "step": 37426
    },
    {
      "epoch": 1.717543940158781,
      "grad_norm": 0.45684152841567993,
      "learning_rate": 4.004732117215403e-06,
      "loss": 0.3248,
      "step": 37427
    },
    {
      "epoch": 1.7175898306640356,
      "grad_norm": 0.461478054523468,
      "learning_rate": 4.00449183705615e-06,
      "loss": 0.3546,
      "step": 37428
    },
    {
      "epoch": 1.71763572116929,
      "grad_norm": 0.47843286395072937,
      "learning_rate": 4.0042515592907806e-06,
      "loss": 0.4121,
      "step": 37429
    },
    {
      "epoch": 1.7176816116745446,
      "grad_norm": 0.4785355031490326,
      "learning_rate": 4.004011283919871e-06,
      "loss": 0.3736,
      "step": 37430
    },
    {
      "epoch": 1.717727502179799,
      "grad_norm": 0.4617527723312378,
      "learning_rate": 4.003771010943997e-06,
      "loss": 0.3743,
      "step": 37431
    },
    {
      "epoch": 1.7177733926850536,
      "grad_norm": 0.4967412054538727,
      "learning_rate": 4.003530740363745e-06,
      "loss": 0.4163,
      "step": 37432
    },
    {
      "epoch": 1.717819283190308,
      "grad_norm": 0.5080995559692383,
      "learning_rate": 4.003290472179684e-06,
      "loss": 0.2781,
      "step": 37433
    },
    {
      "epoch": 1.7178651736955624,
      "grad_norm": 0.46165111660957336,
      "learning_rate": 4.003050206392393e-06,
      "loss": 0.3413,
      "step": 37434
    },
    {
      "epoch": 1.7179110642008169,
      "grad_norm": 0.5243955254554749,
      "learning_rate": 4.002809943002453e-06,
      "loss": 0.4108,
      "step": 37435
    },
    {
      "epoch": 1.7179569547060713,
      "grad_norm": 0.4620725214481354,
      "learning_rate": 4.002569682010442e-06,
      "loss": 0.3328,
      "step": 37436
    },
    {
      "epoch": 1.7180028452113256,
      "grad_norm": 0.4952905774116516,
      "learning_rate": 4.0023294234169345e-06,
      "loss": 0.4072,
      "step": 37437
    },
    {
      "epoch": 1.7180487357165801,
      "grad_norm": 0.4835174083709717,
      "learning_rate": 4.00208916722251e-06,
      "loss": 0.3606,
      "step": 37438
    },
    {
      "epoch": 1.7180946262218346,
      "grad_norm": 0.46597862243652344,
      "learning_rate": 4.001848913427747e-06,
      "loss": 0.3806,
      "step": 37439
    },
    {
      "epoch": 1.7181405167270891,
      "grad_norm": 0.43003422021865845,
      "learning_rate": 4.001608662033221e-06,
      "loss": 0.2815,
      "step": 37440
    },
    {
      "epoch": 1.7181864072323436,
      "grad_norm": 0.4521295726299286,
      "learning_rate": 4.0013684130395116e-06,
      "loss": 0.3371,
      "step": 37441
    },
    {
      "epoch": 1.7182322977375981,
      "grad_norm": 0.455587774515152,
      "learning_rate": 4.0011281664471966e-06,
      "loss": 0.3238,
      "step": 37442
    },
    {
      "epoch": 1.7182781882428526,
      "grad_norm": 0.4918869435787201,
      "learning_rate": 4.000887922256852e-06,
      "loss": 0.3693,
      "step": 37443
    },
    {
      "epoch": 1.718324078748107,
      "grad_norm": 0.4982035160064697,
      "learning_rate": 4.000647680469058e-06,
      "loss": 0.4172,
      "step": 37444
    },
    {
      "epoch": 1.7183699692533616,
      "grad_norm": 0.45832544565200806,
      "learning_rate": 4.000407441084391e-06,
      "loss": 0.2756,
      "step": 37445
    },
    {
      "epoch": 1.718415859758616,
      "grad_norm": 0.469639390707016,
      "learning_rate": 4.0001672041034256e-06,
      "loss": 0.361,
      "step": 37446
    },
    {
      "epoch": 1.7184617502638704,
      "grad_norm": 0.4746950566768646,
      "learning_rate": 3.999926969526745e-06,
      "loss": 0.375,
      "step": 37447
    },
    {
      "epoch": 1.7185076407691249,
      "grad_norm": 0.423507422208786,
      "learning_rate": 3.999686737354925e-06,
      "loss": 0.3029,
      "step": 37448
    },
    {
      "epoch": 1.7185535312743794,
      "grad_norm": 0.46763208508491516,
      "learning_rate": 3.999446507588541e-06,
      "loss": 0.3644,
      "step": 37449
    },
    {
      "epoch": 1.7185994217796337,
      "grad_norm": 0.4662920832633972,
      "learning_rate": 3.999206280228174e-06,
      "loss": 0.3689,
      "step": 37450
    },
    {
      "epoch": 1.7186453122848881,
      "grad_norm": 0.47742247581481934,
      "learning_rate": 3.9989660552744e-06,
      "loss": 0.3873,
      "step": 37451
    },
    {
      "epoch": 1.7186912027901426,
      "grad_norm": 0.47880977392196655,
      "learning_rate": 3.998725832727794e-06,
      "loss": 0.3335,
      "step": 37452
    },
    {
      "epoch": 1.7187370932953971,
      "grad_norm": 0.47136253118515015,
      "learning_rate": 3.9984856125889384e-06,
      "loss": 0.3596,
      "step": 37453
    },
    {
      "epoch": 1.7187829838006516,
      "grad_norm": 0.43218478560447693,
      "learning_rate": 3.9982453948584105e-06,
      "loss": 0.3041,
      "step": 37454
    },
    {
      "epoch": 1.7188288743059061,
      "grad_norm": 0.4914086163043976,
      "learning_rate": 3.998005179536783e-06,
      "loss": 0.432,
      "step": 37455
    },
    {
      "epoch": 1.7188747648111606,
      "grad_norm": 0.5185924768447876,
      "learning_rate": 3.997764966624639e-06,
      "loss": 0.4798,
      "step": 37456
    },
    {
      "epoch": 1.7189206553164151,
      "grad_norm": 0.4142240285873413,
      "learning_rate": 3.997524756122554e-06,
      "loss": 0.29,
      "step": 37457
    },
    {
      "epoch": 1.7189665458216696,
      "grad_norm": 0.45898813009262085,
      "learning_rate": 3.997284548031103e-06,
      "loss": 0.382,
      "step": 37458
    },
    {
      "epoch": 1.719012436326924,
      "grad_norm": 0.46060672402381897,
      "learning_rate": 3.997044342350868e-06,
      "loss": 0.3469,
      "step": 37459
    },
    {
      "epoch": 1.7190583268321784,
      "grad_norm": 0.48611119389533997,
      "learning_rate": 3.996804139082427e-06,
      "loss": 0.3503,
      "step": 37460
    },
    {
      "epoch": 1.719104217337433,
      "grad_norm": 0.4571055471897125,
      "learning_rate": 3.996563938226351e-06,
      "loss": 0.3383,
      "step": 37461
    },
    {
      "epoch": 1.7191501078426874,
      "grad_norm": 0.5042287111282349,
      "learning_rate": 3.996323739783224e-06,
      "loss": 0.4256,
      "step": 37462
    },
    {
      "epoch": 1.7191959983479417,
      "grad_norm": 0.45242840051651,
      "learning_rate": 3.996083543753623e-06,
      "loss": 0.3158,
      "step": 37463
    },
    {
      "epoch": 1.7192418888531962,
      "grad_norm": 0.5024271011352539,
      "learning_rate": 3.995843350138123e-06,
      "loss": 0.3983,
      "step": 37464
    },
    {
      "epoch": 1.7192877793584507,
      "grad_norm": 0.4561421275138855,
      "learning_rate": 3.995603158937303e-06,
      "loss": 0.346,
      "step": 37465
    },
    {
      "epoch": 1.7193336698637052,
      "grad_norm": 0.4468078315258026,
      "learning_rate": 3.99536297015174e-06,
      "loss": 0.3128,
      "step": 37466
    },
    {
      "epoch": 1.7193795603689597,
      "grad_norm": 0.5193073153495789,
      "learning_rate": 3.995122783782014e-06,
      "loss": 0.422,
      "step": 37467
    },
    {
      "epoch": 1.7194254508742142,
      "grad_norm": 0.44783931970596313,
      "learning_rate": 3.994882599828697e-06,
      "loss": 0.3198,
      "step": 37468
    },
    {
      "epoch": 1.7194713413794687,
      "grad_norm": 0.4439665675163269,
      "learning_rate": 3.994642418292372e-06,
      "loss": 0.3472,
      "step": 37469
    },
    {
      "epoch": 1.7195172318847232,
      "grad_norm": 0.46491897106170654,
      "learning_rate": 3.994402239173615e-06,
      "loss": 0.3198,
      "step": 37470
    },
    {
      "epoch": 1.7195631223899777,
      "grad_norm": 0.5054397583007812,
      "learning_rate": 3.994162062473002e-06,
      "loss": 0.4048,
      "step": 37471
    },
    {
      "epoch": 1.719609012895232,
      "grad_norm": 0.41765931248664856,
      "learning_rate": 3.993921888191114e-06,
      "loss": 0.2432,
      "step": 37472
    },
    {
      "epoch": 1.7196549034004864,
      "grad_norm": 0.4711093008518219,
      "learning_rate": 3.993681716328526e-06,
      "loss": 0.3601,
      "step": 37473
    },
    {
      "epoch": 1.719700793905741,
      "grad_norm": 0.45483720302581787,
      "learning_rate": 3.993441546885813e-06,
      "loss": 0.316,
      "step": 37474
    },
    {
      "epoch": 1.7197466844109952,
      "grad_norm": 0.44771209359169006,
      "learning_rate": 3.993201379863558e-06,
      "loss": 0.2952,
      "step": 37475
    },
    {
      "epoch": 1.7197925749162497,
      "grad_norm": 0.48636239767074585,
      "learning_rate": 3.9929612152623364e-06,
      "loss": 0.3318,
      "step": 37476
    },
    {
      "epoch": 1.7198384654215042,
      "grad_norm": 0.4718059003353119,
      "learning_rate": 3.992721053082723e-06,
      "loss": 0.3671,
      "step": 37477
    },
    {
      "epoch": 1.7198843559267587,
      "grad_norm": 0.46593016386032104,
      "learning_rate": 3.992480893325299e-06,
      "loss": 0.3454,
      "step": 37478
    },
    {
      "epoch": 1.7199302464320132,
      "grad_norm": 0.47570833563804626,
      "learning_rate": 3.99224073599064e-06,
      "loss": 0.4017,
      "step": 37479
    },
    {
      "epoch": 1.7199761369372677,
      "grad_norm": 0.4842776954174042,
      "learning_rate": 3.992000581079324e-06,
      "loss": 0.3554,
      "step": 37480
    },
    {
      "epoch": 1.7200220274425222,
      "grad_norm": 0.4806846082210541,
      "learning_rate": 3.991760428591928e-06,
      "loss": 0.4014,
      "step": 37481
    },
    {
      "epoch": 1.7200679179477767,
      "grad_norm": 0.4526643455028534,
      "learning_rate": 3.991520278529032e-06,
      "loss": 0.3194,
      "step": 37482
    },
    {
      "epoch": 1.7201138084530312,
      "grad_norm": 0.45557910203933716,
      "learning_rate": 3.991280130891208e-06,
      "loss": 0.3378,
      "step": 37483
    },
    {
      "epoch": 1.7201596989582857,
      "grad_norm": 0.5032175779342651,
      "learning_rate": 3.991039985679039e-06,
      "loss": 0.4044,
      "step": 37484
    },
    {
      "epoch": 1.72020558946354,
      "grad_norm": 0.46120551228523254,
      "learning_rate": 3.990799842893101e-06,
      "loss": 0.3491,
      "step": 37485
    },
    {
      "epoch": 1.7202514799687945,
      "grad_norm": 0.4485439658164978,
      "learning_rate": 3.9905597025339705e-06,
      "loss": 0.3157,
      "step": 37486
    },
    {
      "epoch": 1.720297370474049,
      "grad_norm": 0.4824945628643036,
      "learning_rate": 3.990319564602225e-06,
      "loss": 0.364,
      "step": 37487
    },
    {
      "epoch": 1.7203432609793032,
      "grad_norm": 0.4498058557510376,
      "learning_rate": 3.990079429098443e-06,
      "loss": 0.3433,
      "step": 37488
    },
    {
      "epoch": 1.7203891514845577,
      "grad_norm": 0.44333308935165405,
      "learning_rate": 3.989839296023198e-06,
      "loss": 0.2932,
      "step": 37489
    },
    {
      "epoch": 1.7204350419898122,
      "grad_norm": 0.45488208532333374,
      "learning_rate": 3.989599165377075e-06,
      "loss": 0.3268,
      "step": 37490
    },
    {
      "epoch": 1.7204809324950667,
      "grad_norm": 0.4659055471420288,
      "learning_rate": 3.989359037160645e-06,
      "loss": 0.346,
      "step": 37491
    },
    {
      "epoch": 1.7205268230003212,
      "grad_norm": 0.46532806754112244,
      "learning_rate": 3.989118911374488e-06,
      "loss": 0.3609,
      "step": 37492
    },
    {
      "epoch": 1.7205727135055757,
      "grad_norm": 0.4649726450443268,
      "learning_rate": 3.988878788019181e-06,
      "loss": 0.336,
      "step": 37493
    },
    {
      "epoch": 1.7206186040108302,
      "grad_norm": 0.47610193490982056,
      "learning_rate": 3.988638667095301e-06,
      "loss": 0.3645,
      "step": 37494
    },
    {
      "epoch": 1.7206644945160847,
      "grad_norm": 0.5477766990661621,
      "learning_rate": 3.988398548603427e-06,
      "loss": 0.4065,
      "step": 37495
    },
    {
      "epoch": 1.7207103850213392,
      "grad_norm": 0.4297437071800232,
      "learning_rate": 3.9881584325441325e-06,
      "loss": 0.2932,
      "step": 37496
    },
    {
      "epoch": 1.7207562755265935,
      "grad_norm": 0.4719291925430298,
      "learning_rate": 3.987918318918e-06,
      "loss": 0.3666,
      "step": 37497
    },
    {
      "epoch": 1.720802166031848,
      "grad_norm": 0.46739521622657776,
      "learning_rate": 3.987678207725605e-06,
      "loss": 0.3823,
      "step": 37498
    },
    {
      "epoch": 1.7208480565371025,
      "grad_norm": 0.44146203994750977,
      "learning_rate": 3.987438098967522e-06,
      "loss": 0.3443,
      "step": 37499
    },
    {
      "epoch": 1.720893947042357,
      "grad_norm": 0.5081411004066467,
      "learning_rate": 3.987197992644333e-06,
      "loss": 0.4601,
      "step": 37500
    },
    {
      "epoch": 1.7209398375476113,
      "grad_norm": 0.5735171437263489,
      "learning_rate": 3.986957888756612e-06,
      "loss": 0.3333,
      "step": 37501
    },
    {
      "epoch": 1.7209857280528658,
      "grad_norm": 0.4871867299079895,
      "learning_rate": 3.9867177873049375e-06,
      "loss": 0.3904,
      "step": 37502
    },
    {
      "epoch": 1.7210316185581203,
      "grad_norm": 0.40315142273902893,
      "learning_rate": 3.986477688289888e-06,
      "loss": 0.2663,
      "step": 37503
    },
    {
      "epoch": 1.7210775090633748,
      "grad_norm": 0.5050402879714966,
      "learning_rate": 3.986237591712039e-06,
      "loss": 0.4533,
      "step": 37504
    },
    {
      "epoch": 1.7211233995686293,
      "grad_norm": 0.49255913496017456,
      "learning_rate": 3.985997497571968e-06,
      "loss": 0.4059,
      "step": 37505
    },
    {
      "epoch": 1.7211692900738838,
      "grad_norm": 0.4671362042427063,
      "learning_rate": 3.985757405870254e-06,
      "loss": 0.3257,
      "step": 37506
    },
    {
      "epoch": 1.7212151805791382,
      "grad_norm": 0.49167948961257935,
      "learning_rate": 3.985517316607473e-06,
      "loss": 0.3897,
      "step": 37507
    },
    {
      "epoch": 1.7212610710843927,
      "grad_norm": 0.4676930606365204,
      "learning_rate": 3.985277229784203e-06,
      "loss": 0.354,
      "step": 37508
    },
    {
      "epoch": 1.7213069615896472,
      "grad_norm": 0.47411277890205383,
      "learning_rate": 3.9850371454010205e-06,
      "loss": 0.3917,
      "step": 37509
    },
    {
      "epoch": 1.7213528520949015,
      "grad_norm": 0.5109216570854187,
      "learning_rate": 3.984797063458504e-06,
      "loss": 0.4725,
      "step": 37510
    },
    {
      "epoch": 1.721398742600156,
      "grad_norm": 0.46160703897476196,
      "learning_rate": 3.984556983957228e-06,
      "loss": 0.3445,
      "step": 37511
    },
    {
      "epoch": 1.7214446331054105,
      "grad_norm": 0.49744412302970886,
      "learning_rate": 3.984316906897774e-06,
      "loss": 0.3575,
      "step": 37512
    },
    {
      "epoch": 1.7214905236106648,
      "grad_norm": 0.5047242045402527,
      "learning_rate": 3.984076832280718e-06,
      "loss": 0.4503,
      "step": 37513
    },
    {
      "epoch": 1.7215364141159193,
      "grad_norm": 0.49746882915496826,
      "learning_rate": 3.983836760106634e-06,
      "loss": 0.4197,
      "step": 37514
    },
    {
      "epoch": 1.7215823046211738,
      "grad_norm": 0.4703180193901062,
      "learning_rate": 3.9835966903761046e-06,
      "loss": 0.3384,
      "step": 37515
    },
    {
      "epoch": 1.7216281951264283,
      "grad_norm": 0.507028341293335,
      "learning_rate": 3.983356623089704e-06,
      "loss": 0.4222,
      "step": 37516
    },
    {
      "epoch": 1.7216740856316828,
      "grad_norm": 0.5000210404396057,
      "learning_rate": 3.983116558248007e-06,
      "loss": 0.4687,
      "step": 37517
    },
    {
      "epoch": 1.7217199761369373,
      "grad_norm": 0.45217305421829224,
      "learning_rate": 3.9828764958515955e-06,
      "loss": 0.3185,
      "step": 37518
    },
    {
      "epoch": 1.7217658666421918,
      "grad_norm": 0.4977262616157532,
      "learning_rate": 3.9826364359010465e-06,
      "loss": 0.3643,
      "step": 37519
    },
    {
      "epoch": 1.7218117571474463,
      "grad_norm": 0.4937879145145416,
      "learning_rate": 3.982396378396934e-06,
      "loss": 0.4067,
      "step": 37520
    },
    {
      "epoch": 1.7218576476527008,
      "grad_norm": 0.46037471294403076,
      "learning_rate": 3.982156323339838e-06,
      "loss": 0.3141,
      "step": 37521
    },
    {
      "epoch": 1.7219035381579553,
      "grad_norm": 0.47940704226493835,
      "learning_rate": 3.9819162707303346e-06,
      "loss": 0.4117,
      "step": 37522
    },
    {
      "epoch": 1.7219494286632095,
      "grad_norm": 0.46581918001174927,
      "learning_rate": 3.981676220569e-06,
      "loss": 0.3048,
      "step": 37523
    },
    {
      "epoch": 1.721995319168464,
      "grad_norm": 0.4672877788543701,
      "learning_rate": 3.981436172856415e-06,
      "loss": 0.3669,
      "step": 37524
    },
    {
      "epoch": 1.7220412096737185,
      "grad_norm": 0.48470765352249146,
      "learning_rate": 3.981196127593154e-06,
      "loss": 0.3808,
      "step": 37525
    },
    {
      "epoch": 1.7220871001789728,
      "grad_norm": 0.4611252546310425,
      "learning_rate": 3.980956084779792e-06,
      "loss": 0.3359,
      "step": 37526
    },
    {
      "epoch": 1.7221329906842273,
      "grad_norm": 0.4747268259525299,
      "learning_rate": 3.980716044416911e-06,
      "loss": 0.3514,
      "step": 37527
    },
    {
      "epoch": 1.7221788811894818,
      "grad_norm": 0.4880480468273163,
      "learning_rate": 3.980476006505087e-06,
      "loss": 0.3847,
      "step": 37528
    },
    {
      "epoch": 1.7222247716947363,
      "grad_norm": 0.46068698167800903,
      "learning_rate": 3.980235971044896e-06,
      "loss": 0.3191,
      "step": 37529
    },
    {
      "epoch": 1.7222706621999908,
      "grad_norm": 0.461694598197937,
      "learning_rate": 3.979995938036915e-06,
      "loss": 0.3521,
      "step": 37530
    },
    {
      "epoch": 1.7223165527052453,
      "grad_norm": 0.5194443464279175,
      "learning_rate": 3.979755907481723e-06,
      "loss": 0.4,
      "step": 37531
    },
    {
      "epoch": 1.7223624432104998,
      "grad_norm": 0.42952805757522583,
      "learning_rate": 3.979515879379895e-06,
      "loss": 0.3075,
      "step": 37532
    },
    {
      "epoch": 1.7224083337157543,
      "grad_norm": 0.44474998116493225,
      "learning_rate": 3.979275853732009e-06,
      "loss": 0.3105,
      "step": 37533
    },
    {
      "epoch": 1.7224542242210088,
      "grad_norm": 0.45281049609184265,
      "learning_rate": 3.979035830538643e-06,
      "loss": 0.3415,
      "step": 37534
    },
    {
      "epoch": 1.7225001147262633,
      "grad_norm": 0.48055335879325867,
      "learning_rate": 3.9787958098003736e-06,
      "loss": 0.4269,
      "step": 37535
    },
    {
      "epoch": 1.7225460052315176,
      "grad_norm": 0.4424835443496704,
      "learning_rate": 3.978555791517777e-06,
      "loss": 0.2859,
      "step": 37536
    },
    {
      "epoch": 1.722591895736772,
      "grad_norm": 0.5206233859062195,
      "learning_rate": 3.978315775691433e-06,
      "loss": 0.432,
      "step": 37537
    },
    {
      "epoch": 1.7226377862420266,
      "grad_norm": 0.480898380279541,
      "learning_rate": 3.978075762321917e-06,
      "loss": 0.3403,
      "step": 37538
    },
    {
      "epoch": 1.7226836767472808,
      "grad_norm": 0.4597509503364563,
      "learning_rate": 3.977835751409804e-06,
      "loss": 0.3443,
      "step": 37539
    },
    {
      "epoch": 1.7227295672525353,
      "grad_norm": 0.4521420896053314,
      "learning_rate": 3.977595742955676e-06,
      "loss": 0.349,
      "step": 37540
    },
    {
      "epoch": 1.7227754577577898,
      "grad_norm": 0.5329156517982483,
      "learning_rate": 3.9773557369601065e-06,
      "loss": 0.3258,
      "step": 37541
    },
    {
      "epoch": 1.7228213482630443,
      "grad_norm": 0.4797748327255249,
      "learning_rate": 3.977115733423673e-06,
      "loss": 0.377,
      "step": 37542
    },
    {
      "epoch": 1.7228672387682988,
      "grad_norm": 0.4856626093387604,
      "learning_rate": 3.9768757323469545e-06,
      "loss": 0.4252,
      "step": 37543
    },
    {
      "epoch": 1.7229131292735533,
      "grad_norm": 0.4284609258174896,
      "learning_rate": 3.976635733730527e-06,
      "loss": 0.2918,
      "step": 37544
    },
    {
      "epoch": 1.7229590197788078,
      "grad_norm": 0.45278462767601013,
      "learning_rate": 3.976395737574967e-06,
      "loss": 0.3045,
      "step": 37545
    },
    {
      "epoch": 1.7230049102840623,
      "grad_norm": 0.4712797999382019,
      "learning_rate": 3.976155743880853e-06,
      "loss": 0.3507,
      "step": 37546
    },
    {
      "epoch": 1.7230508007893168,
      "grad_norm": 0.44265469908714294,
      "learning_rate": 3.975915752648761e-06,
      "loss": 0.3118,
      "step": 37547
    },
    {
      "epoch": 1.723096691294571,
      "grad_norm": 0.5240386724472046,
      "learning_rate": 3.975675763879266e-06,
      "loss": 0.4425,
      "step": 37548
    },
    {
      "epoch": 1.7231425817998256,
      "grad_norm": 0.4849725663661957,
      "learning_rate": 3.97543577757295e-06,
      "loss": 0.3769,
      "step": 37549
    },
    {
      "epoch": 1.72318847230508,
      "grad_norm": 0.42730268836021423,
      "learning_rate": 3.975195793730387e-06,
      "loss": 0.2778,
      "step": 37550
    },
    {
      "epoch": 1.7232343628103346,
      "grad_norm": 0.4534895718097687,
      "learning_rate": 3.974955812352155e-06,
      "loss": 0.3372,
      "step": 37551
    },
    {
      "epoch": 1.7232802533155889,
      "grad_norm": 0.4546175003051758,
      "learning_rate": 3.974715833438831e-06,
      "loss": 0.3271,
      "step": 37552
    },
    {
      "epoch": 1.7233261438208434,
      "grad_norm": 0.44587600231170654,
      "learning_rate": 3.974475856990992e-06,
      "loss": 0.3213,
      "step": 37553
    },
    {
      "epoch": 1.7233720343260979,
      "grad_norm": 0.47271448373794556,
      "learning_rate": 3.974235883009213e-06,
      "loss": 0.3754,
      "step": 37554
    },
    {
      "epoch": 1.7234179248313524,
      "grad_norm": 0.456817090511322,
      "learning_rate": 3.973995911494074e-06,
      "loss": 0.332,
      "step": 37555
    },
    {
      "epoch": 1.7234638153366069,
      "grad_norm": 0.5080356001853943,
      "learning_rate": 3.973755942446152e-06,
      "loss": 0.4085,
      "step": 37556
    },
    {
      "epoch": 1.7235097058418614,
      "grad_norm": 0.46196117997169495,
      "learning_rate": 3.973515975866021e-06,
      "loss": 0.3553,
      "step": 37557
    },
    {
      "epoch": 1.7235555963471159,
      "grad_norm": 0.48965105414390564,
      "learning_rate": 3.9732760117542626e-06,
      "loss": 0.3845,
      "step": 37558
    },
    {
      "epoch": 1.7236014868523704,
      "grad_norm": 0.4894275367259979,
      "learning_rate": 3.973036050111451e-06,
      "loss": 0.3812,
      "step": 37559
    },
    {
      "epoch": 1.7236473773576249,
      "grad_norm": 0.475813090801239,
      "learning_rate": 3.972796090938163e-06,
      "loss": 0.3675,
      "step": 37560
    },
    {
      "epoch": 1.7236932678628791,
      "grad_norm": 0.513232409954071,
      "learning_rate": 3.972556134234975e-06,
      "loss": 0.3494,
      "step": 37561
    },
    {
      "epoch": 1.7237391583681336,
      "grad_norm": 0.47769877314567566,
      "learning_rate": 3.972316180002466e-06,
      "loss": 0.3701,
      "step": 37562
    },
    {
      "epoch": 1.7237850488733881,
      "grad_norm": 0.46627339720726013,
      "learning_rate": 3.972076228241213e-06,
      "loss": 0.3765,
      "step": 37563
    },
    {
      "epoch": 1.7238309393786424,
      "grad_norm": 0.4687279462814331,
      "learning_rate": 3.971836278951791e-06,
      "loss": 0.3838,
      "step": 37564
    },
    {
      "epoch": 1.723876829883897,
      "grad_norm": 0.4852904975414276,
      "learning_rate": 3.971596332134779e-06,
      "loss": 0.3742,
      "step": 37565
    },
    {
      "epoch": 1.7239227203891514,
      "grad_norm": 0.48786213994026184,
      "learning_rate": 3.9713563877907545e-06,
      "loss": 0.3496,
      "step": 37566
    },
    {
      "epoch": 1.723968610894406,
      "grad_norm": 0.4731210172176361,
      "learning_rate": 3.971116445920291e-06,
      "loss": 0.3234,
      "step": 37567
    },
    {
      "epoch": 1.7240145013996604,
      "grad_norm": 0.46663519740104675,
      "learning_rate": 3.970876506523969e-06,
      "loss": 0.3229,
      "step": 37568
    },
    {
      "epoch": 1.724060391904915,
      "grad_norm": 0.46064475178718567,
      "learning_rate": 3.970636569602364e-06,
      "loss": 0.3446,
      "step": 37569
    },
    {
      "epoch": 1.7241062824101694,
      "grad_norm": 0.44724735617637634,
      "learning_rate": 3.970396635156052e-06,
      "loss": 0.3074,
      "step": 37570
    },
    {
      "epoch": 1.7241521729154239,
      "grad_norm": 0.4779791831970215,
      "learning_rate": 3.970156703185613e-06,
      "loss": 0.4032,
      "step": 37571
    },
    {
      "epoch": 1.7241980634206784,
      "grad_norm": 0.4674805998802185,
      "learning_rate": 3.969916773691621e-06,
      "loss": 0.3605,
      "step": 37572
    },
    {
      "epoch": 1.7242439539259329,
      "grad_norm": 0.46409332752227783,
      "learning_rate": 3.969676846674654e-06,
      "loss": 0.3939,
      "step": 37573
    },
    {
      "epoch": 1.7242898444311872,
      "grad_norm": 0.4558377265930176,
      "learning_rate": 3.969436922135289e-06,
      "loss": 0.3379,
      "step": 37574
    },
    {
      "epoch": 1.7243357349364417,
      "grad_norm": 0.4678764045238495,
      "learning_rate": 3.9691970000741045e-06,
      "loss": 0.3789,
      "step": 37575
    },
    {
      "epoch": 1.7243816254416962,
      "grad_norm": 0.491243839263916,
      "learning_rate": 3.968957080491674e-06,
      "loss": 0.3562,
      "step": 37576
    },
    {
      "epoch": 1.7244275159469504,
      "grad_norm": 0.5193041563034058,
      "learning_rate": 3.968717163388576e-06,
      "loss": 0.3935,
      "step": 37577
    },
    {
      "epoch": 1.724473406452205,
      "grad_norm": 0.48554202914237976,
      "learning_rate": 3.96847724876539e-06,
      "loss": 0.4166,
      "step": 37578
    },
    {
      "epoch": 1.7245192969574594,
      "grad_norm": 0.46968919038772583,
      "learning_rate": 3.968237336622688e-06,
      "loss": 0.3632,
      "step": 37579
    },
    {
      "epoch": 1.724565187462714,
      "grad_norm": 0.48958495259284973,
      "learning_rate": 3.967997426961052e-06,
      "loss": 0.3468,
      "step": 37580
    },
    {
      "epoch": 1.7246110779679684,
      "grad_norm": 0.47594383358955383,
      "learning_rate": 3.967757519781056e-06,
      "loss": 0.3825,
      "step": 37581
    },
    {
      "epoch": 1.724656968473223,
      "grad_norm": 0.4329048991203308,
      "learning_rate": 3.967517615083275e-06,
      "loss": 0.2605,
      "step": 37582
    },
    {
      "epoch": 1.7247028589784774,
      "grad_norm": 0.48690247535705566,
      "learning_rate": 3.967277712868291e-06,
      "loss": 0.3682,
      "step": 37583
    },
    {
      "epoch": 1.724748749483732,
      "grad_norm": 0.43930667638778687,
      "learning_rate": 3.967037813136677e-06,
      "loss": 0.2801,
      "step": 37584
    },
    {
      "epoch": 1.7247946399889864,
      "grad_norm": 0.5230001211166382,
      "learning_rate": 3.966797915889011e-06,
      "loss": 0.423,
      "step": 37585
    },
    {
      "epoch": 1.7248405304942407,
      "grad_norm": 0.44060665369033813,
      "learning_rate": 3.96655802112587e-06,
      "loss": 0.2968,
      "step": 37586
    },
    {
      "epoch": 1.7248864209994952,
      "grad_norm": 0.5006231069564819,
      "learning_rate": 3.966318128847832e-06,
      "loss": 0.4017,
      "step": 37587
    },
    {
      "epoch": 1.7249323115047497,
      "grad_norm": 0.586255669593811,
      "learning_rate": 3.96607823905547e-06,
      "loss": 0.3447,
      "step": 37588
    },
    {
      "epoch": 1.7249782020100042,
      "grad_norm": 0.4721508324146271,
      "learning_rate": 3.965838351749367e-06,
      "loss": 0.3599,
      "step": 37589
    },
    {
      "epoch": 1.7250240925152585,
      "grad_norm": 0.46438029408454895,
      "learning_rate": 3.9655984669300955e-06,
      "loss": 0.3333,
      "step": 37590
    },
    {
      "epoch": 1.725069983020513,
      "grad_norm": 0.5636918544769287,
      "learning_rate": 3.9653585845982304e-06,
      "loss": 0.3103,
      "step": 37591
    },
    {
      "epoch": 1.7251158735257675,
      "grad_norm": 0.48524752259254456,
      "learning_rate": 3.965118704754354e-06,
      "loss": 0.3317,
      "step": 37592
    },
    {
      "epoch": 1.725161764031022,
      "grad_norm": 0.4652059078216553,
      "learning_rate": 3.964878827399039e-06,
      "loss": 0.347,
      "step": 37593
    },
    {
      "epoch": 1.7252076545362764,
      "grad_norm": 0.49668341875076294,
      "learning_rate": 3.964638952532866e-06,
      "loss": 0.3466,
      "step": 37594
    },
    {
      "epoch": 1.725253545041531,
      "grad_norm": 0.5006192922592163,
      "learning_rate": 3.964399080156407e-06,
      "loss": 0.3853,
      "step": 37595
    },
    {
      "epoch": 1.7252994355467854,
      "grad_norm": 0.4834129512310028,
      "learning_rate": 3.9641592102702435e-06,
      "loss": 0.4054,
      "step": 37596
    },
    {
      "epoch": 1.72534532605204,
      "grad_norm": 0.45366448163986206,
      "learning_rate": 3.963919342874949e-06,
      "loss": 0.3282,
      "step": 37597
    },
    {
      "epoch": 1.7253912165572944,
      "grad_norm": 0.5185109972953796,
      "learning_rate": 3.963679477971099e-06,
      "loss": 0.4652,
      "step": 37598
    },
    {
      "epoch": 1.7254371070625487,
      "grad_norm": 0.47021418809890747,
      "learning_rate": 3.963439615559277e-06,
      "loss": 0.3185,
      "step": 37599
    },
    {
      "epoch": 1.7254829975678032,
      "grad_norm": 0.4304482936859131,
      "learning_rate": 3.963199755640053e-06,
      "loss": 0.3021,
      "step": 37600
    },
    {
      "epoch": 1.7255288880730577,
      "grad_norm": 0.46824756264686584,
      "learning_rate": 3.962959898214008e-06,
      "loss": 0.3588,
      "step": 37601
    },
    {
      "epoch": 1.725574778578312,
      "grad_norm": 0.46092677116394043,
      "learning_rate": 3.9627200432817165e-06,
      "loss": 0.3285,
      "step": 37602
    },
    {
      "epoch": 1.7256206690835665,
      "grad_norm": 0.4817688465118408,
      "learning_rate": 3.962480190843757e-06,
      "loss": 0.2963,
      "step": 37603
    },
    {
      "epoch": 1.725666559588821,
      "grad_norm": 0.46024349331855774,
      "learning_rate": 3.962240340900701e-06,
      "loss": 0.323,
      "step": 37604
    },
    {
      "epoch": 1.7257124500940755,
      "grad_norm": 0.45891401171684265,
      "learning_rate": 3.962000493453133e-06,
      "loss": 0.3325,
      "step": 37605
    },
    {
      "epoch": 1.72575834059933,
      "grad_norm": 0.44002100825309753,
      "learning_rate": 3.961760648501627e-06,
      "loss": 0.3108,
      "step": 37606
    },
    {
      "epoch": 1.7258042311045845,
      "grad_norm": 0.4563017785549164,
      "learning_rate": 3.961520806046757e-06,
      "loss": 0.3751,
      "step": 37607
    },
    {
      "epoch": 1.725850121609839,
      "grad_norm": 0.45861729979515076,
      "learning_rate": 3.961280966089103e-06,
      "loss": 0.3197,
      "step": 37608
    },
    {
      "epoch": 1.7258960121150935,
      "grad_norm": 0.44539135694503784,
      "learning_rate": 3.9610411286292396e-06,
      "loss": 0.3256,
      "step": 37609
    },
    {
      "epoch": 1.725941902620348,
      "grad_norm": 0.4862193167209625,
      "learning_rate": 3.960801293667743e-06,
      "loss": 0.3344,
      "step": 37610
    },
    {
      "epoch": 1.7259877931256025,
      "grad_norm": 0.5091336965560913,
      "learning_rate": 3.960561461205195e-06,
      "loss": 0.3597,
      "step": 37611
    },
    {
      "epoch": 1.7260336836308567,
      "grad_norm": 0.4904113709926605,
      "learning_rate": 3.960321631242167e-06,
      "loss": 0.4448,
      "step": 37612
    },
    {
      "epoch": 1.7260795741361112,
      "grad_norm": 0.46362507343292236,
      "learning_rate": 3.960081803779235e-06,
      "loss": 0.3664,
      "step": 37613
    },
    {
      "epoch": 1.7261254646413657,
      "grad_norm": 0.46124547719955444,
      "learning_rate": 3.9598419788169804e-06,
      "loss": 0.3519,
      "step": 37614
    },
    {
      "epoch": 1.72617135514662,
      "grad_norm": 0.43686968088150024,
      "learning_rate": 3.959602156355977e-06,
      "loss": 0.2956,
      "step": 37615
    },
    {
      "epoch": 1.7262172456518745,
      "grad_norm": 0.44320154190063477,
      "learning_rate": 3.959362336396802e-06,
      "loss": 0.2927,
      "step": 37616
    },
    {
      "epoch": 1.726263136157129,
      "grad_norm": 0.4559209942817688,
      "learning_rate": 3.959122518940033e-06,
      "loss": 0.3176,
      "step": 37617
    },
    {
      "epoch": 1.7263090266623835,
      "grad_norm": 0.45440229773521423,
      "learning_rate": 3.958882703986245e-06,
      "loss": 0.3099,
      "step": 37618
    },
    {
      "epoch": 1.726354917167638,
      "grad_norm": 0.47660914063453674,
      "learning_rate": 3.958642891536015e-06,
      "loss": 0.355,
      "step": 37619
    },
    {
      "epoch": 1.7264008076728925,
      "grad_norm": 0.49120286107063293,
      "learning_rate": 3.958403081589921e-06,
      "loss": 0.4522,
      "step": 37620
    },
    {
      "epoch": 1.726446698178147,
      "grad_norm": 0.4136184751987457,
      "learning_rate": 3.958163274148539e-06,
      "loss": 0.2641,
      "step": 37621
    },
    {
      "epoch": 1.7264925886834015,
      "grad_norm": 0.426470011472702,
      "learning_rate": 3.9579234692124445e-06,
      "loss": 0.2966,
      "step": 37622
    },
    {
      "epoch": 1.726538479188656,
      "grad_norm": 0.48396599292755127,
      "learning_rate": 3.957683666782216e-06,
      "loss": 0.3776,
      "step": 37623
    },
    {
      "epoch": 1.7265843696939105,
      "grad_norm": 0.5111770033836365,
      "learning_rate": 3.95744386685843e-06,
      "loss": 0.3858,
      "step": 37624
    },
    {
      "epoch": 1.7266302601991648,
      "grad_norm": 0.49605560302734375,
      "learning_rate": 3.957204069441663e-06,
      "loss": 0.3924,
      "step": 37625
    },
    {
      "epoch": 1.7266761507044193,
      "grad_norm": 0.5005434155464172,
      "learning_rate": 3.9569642745324875e-06,
      "loss": 0.3049,
      "step": 37626
    },
    {
      "epoch": 1.7267220412096738,
      "grad_norm": 0.4701997637748718,
      "learning_rate": 3.956724482131487e-06,
      "loss": 0.3478,
      "step": 37627
    },
    {
      "epoch": 1.726767931714928,
      "grad_norm": 0.4853617250919342,
      "learning_rate": 3.9564846922392335e-06,
      "loss": 0.3837,
      "step": 37628
    },
    {
      "epoch": 1.7268138222201825,
      "grad_norm": 0.4502929151058197,
      "learning_rate": 3.956244904856304e-06,
      "loss": 0.2981,
      "step": 37629
    },
    {
      "epoch": 1.726859712725437,
      "grad_norm": 0.47922638058662415,
      "learning_rate": 3.9560051199832786e-06,
      "loss": 0.3876,
      "step": 37630
    },
    {
      "epoch": 1.7269056032306915,
      "grad_norm": 0.48648831248283386,
      "learning_rate": 3.9557653376207306e-06,
      "loss": 0.398,
      "step": 37631
    },
    {
      "epoch": 1.726951493735946,
      "grad_norm": 0.5228362083435059,
      "learning_rate": 3.9555255577692354e-06,
      "loss": 0.3621,
      "step": 37632
    },
    {
      "epoch": 1.7269973842412005,
      "grad_norm": 0.49243059754371643,
      "learning_rate": 3.955285780429375e-06,
      "loss": 0.379,
      "step": 37633
    },
    {
      "epoch": 1.727043274746455,
      "grad_norm": 0.4640369713306427,
      "learning_rate": 3.9550460056017205e-06,
      "loss": 0.371,
      "step": 37634
    },
    {
      "epoch": 1.7270891652517095,
      "grad_norm": 0.45532137155532837,
      "learning_rate": 3.95480623328685e-06,
      "loss": 0.3516,
      "step": 37635
    },
    {
      "epoch": 1.727135055756964,
      "grad_norm": 0.4885496199131012,
      "learning_rate": 3.954566463485342e-06,
      "loss": 0.4224,
      "step": 37636
    },
    {
      "epoch": 1.7271809462622183,
      "grad_norm": 0.4878074824810028,
      "learning_rate": 3.954326696197771e-06,
      "loss": 0.3815,
      "step": 37637
    },
    {
      "epoch": 1.7272268367674728,
      "grad_norm": 0.4718996584415436,
      "learning_rate": 3.9540869314247134e-06,
      "loss": 0.3504,
      "step": 37638
    },
    {
      "epoch": 1.7272727272727273,
      "grad_norm": 0.48672056198120117,
      "learning_rate": 3.953847169166748e-06,
      "loss": 0.3842,
      "step": 37639
    },
    {
      "epoch": 1.7273186177779818,
      "grad_norm": 0.4533197283744812,
      "learning_rate": 3.9536074094244506e-06,
      "loss": 0.3049,
      "step": 37640
    },
    {
      "epoch": 1.727364508283236,
      "grad_norm": 0.5071361064910889,
      "learning_rate": 3.953367652198395e-06,
      "loss": 0.4153,
      "step": 37641
    },
    {
      "epoch": 1.7274103987884906,
      "grad_norm": 0.4767429828643799,
      "learning_rate": 3.953127897489161e-06,
      "loss": 0.3705,
      "step": 37642
    },
    {
      "epoch": 1.727456289293745,
      "grad_norm": 0.5048102736473083,
      "learning_rate": 3.952888145297324e-06,
      "loss": 0.4114,
      "step": 37643
    },
    {
      "epoch": 1.7275021797989996,
      "grad_norm": 0.44948458671569824,
      "learning_rate": 3.95264839562346e-06,
      "loss": 0.3199,
      "step": 37644
    },
    {
      "epoch": 1.727548070304254,
      "grad_norm": 0.44682055711746216,
      "learning_rate": 3.952408648468147e-06,
      "loss": 0.3363,
      "step": 37645
    },
    {
      "epoch": 1.7275939608095086,
      "grad_norm": 0.4394311308860779,
      "learning_rate": 3.952168903831961e-06,
      "loss": 0.3313,
      "step": 37646
    },
    {
      "epoch": 1.727639851314763,
      "grad_norm": 0.4514541029930115,
      "learning_rate": 3.951929161715476e-06,
      "loss": 0.308,
      "step": 37647
    },
    {
      "epoch": 1.7276857418200176,
      "grad_norm": 0.47513025999069214,
      "learning_rate": 3.951689422119272e-06,
      "loss": 0.4092,
      "step": 37648
    },
    {
      "epoch": 1.727731632325272,
      "grad_norm": 0.4893389940261841,
      "learning_rate": 3.951449685043924e-06,
      "loss": 0.4037,
      "step": 37649
    },
    {
      "epoch": 1.7277775228305263,
      "grad_norm": 0.4138801395893097,
      "learning_rate": 3.951209950490008e-06,
      "loss": 0.2879,
      "step": 37650
    },
    {
      "epoch": 1.7278234133357808,
      "grad_norm": 0.47534239292144775,
      "learning_rate": 3.950970218458103e-06,
      "loss": 0.3527,
      "step": 37651
    },
    {
      "epoch": 1.7278693038410353,
      "grad_norm": 0.4440337121486664,
      "learning_rate": 3.950730488948782e-06,
      "loss": 0.3022,
      "step": 37652
    },
    {
      "epoch": 1.7279151943462896,
      "grad_norm": 0.4509468972682953,
      "learning_rate": 3.950490761962621e-06,
      "loss": 0.3574,
      "step": 37653
    },
    {
      "epoch": 1.727961084851544,
      "grad_norm": 0.44328397512435913,
      "learning_rate": 3.950251037500201e-06,
      "loss": 0.3021,
      "step": 37654
    },
    {
      "epoch": 1.7280069753567986,
      "grad_norm": 0.4531576931476593,
      "learning_rate": 3.950011315562097e-06,
      "loss": 0.3178,
      "step": 37655
    },
    {
      "epoch": 1.728052865862053,
      "grad_norm": 0.46454399824142456,
      "learning_rate": 3.949771596148881e-06,
      "loss": 0.344,
      "step": 37656
    },
    {
      "epoch": 1.7280987563673076,
      "grad_norm": 0.4286445379257202,
      "learning_rate": 3.949531879261136e-06,
      "loss": 0.2772,
      "step": 37657
    },
    {
      "epoch": 1.728144646872562,
      "grad_norm": 0.43322786688804626,
      "learning_rate": 3.949292164899434e-06,
      "loss": 0.2975,
      "step": 37658
    },
    {
      "epoch": 1.7281905373778166,
      "grad_norm": 0.4554278254508972,
      "learning_rate": 3.9490524530643536e-06,
      "loss": 0.3424,
      "step": 37659
    },
    {
      "epoch": 1.728236427883071,
      "grad_norm": 0.46346452832221985,
      "learning_rate": 3.948812743756469e-06,
      "loss": 0.3086,
      "step": 37660
    },
    {
      "epoch": 1.7282823183883256,
      "grad_norm": 0.46050119400024414,
      "learning_rate": 3.948573036976359e-06,
      "loss": 0.33,
      "step": 37661
    },
    {
      "epoch": 1.72832820889358,
      "grad_norm": 0.4652964770793915,
      "learning_rate": 3.9483333327246e-06,
      "loss": 0.3424,
      "step": 37662
    },
    {
      "epoch": 1.7283740993988344,
      "grad_norm": 0.5044764280319214,
      "learning_rate": 3.948093631001763e-06,
      "loss": 0.4018,
      "step": 37663
    },
    {
      "epoch": 1.7284199899040888,
      "grad_norm": 0.45207908749580383,
      "learning_rate": 3.947853931808433e-06,
      "loss": 0.3416,
      "step": 37664
    },
    {
      "epoch": 1.7284658804093433,
      "grad_norm": 0.47938260436058044,
      "learning_rate": 3.947614235145182e-06,
      "loss": 0.3747,
      "step": 37665
    },
    {
      "epoch": 1.7285117709145976,
      "grad_norm": 0.46454739570617676,
      "learning_rate": 3.947374541012584e-06,
      "loss": 0.3637,
      "step": 37666
    },
    {
      "epoch": 1.7285576614198521,
      "grad_norm": 0.46638867259025574,
      "learning_rate": 3.9471348494112205e-06,
      "loss": 0.3425,
      "step": 37667
    },
    {
      "epoch": 1.7286035519251066,
      "grad_norm": 0.5074900388717651,
      "learning_rate": 3.946895160341665e-06,
      "loss": 0.3478,
      "step": 37668
    },
    {
      "epoch": 1.7286494424303611,
      "grad_norm": 0.45006459951400757,
      "learning_rate": 3.946655473804492e-06,
      "loss": 0.3087,
      "step": 37669
    },
    {
      "epoch": 1.7286953329356156,
      "grad_norm": 0.43827497959136963,
      "learning_rate": 3.946415789800283e-06,
      "loss": 0.3041,
      "step": 37670
    },
    {
      "epoch": 1.7287412234408701,
      "grad_norm": 0.4709852635860443,
      "learning_rate": 3.946176108329611e-06,
      "loss": 0.3781,
      "step": 37671
    },
    {
      "epoch": 1.7287871139461246,
      "grad_norm": 0.4881029725074768,
      "learning_rate": 3.945936429393051e-06,
      "loss": 0.4109,
      "step": 37672
    },
    {
      "epoch": 1.728833004451379,
      "grad_norm": 0.4328114986419678,
      "learning_rate": 3.945696752991183e-06,
      "loss": 0.3104,
      "step": 37673
    },
    {
      "epoch": 1.7288788949566336,
      "grad_norm": 0.4756580591201782,
      "learning_rate": 3.945457079124582e-06,
      "loss": 0.3628,
      "step": 37674
    },
    {
      "epoch": 1.7289247854618879,
      "grad_norm": 0.4868132472038269,
      "learning_rate": 3.9452174077938216e-06,
      "loss": 0.3697,
      "step": 37675
    },
    {
      "epoch": 1.7289706759671424,
      "grad_norm": 0.5099665522575378,
      "learning_rate": 3.944977738999482e-06,
      "loss": 0.3589,
      "step": 37676
    },
    {
      "epoch": 1.7290165664723969,
      "grad_norm": 0.5095929503440857,
      "learning_rate": 3.9447380727421395e-06,
      "loss": 0.3257,
      "step": 37677
    },
    {
      "epoch": 1.7290624569776514,
      "grad_norm": 0.49904799461364746,
      "learning_rate": 3.944498409022366e-06,
      "loss": 0.3728,
      "step": 37678
    },
    {
      "epoch": 1.7291083474829056,
      "grad_norm": 0.5024864673614502,
      "learning_rate": 3.944258747840742e-06,
      "loss": 0.343,
      "step": 37679
    },
    {
      "epoch": 1.7291542379881601,
      "grad_norm": 0.5154918432235718,
      "learning_rate": 3.944019089197842e-06,
      "loss": 0.3713,
      "step": 37680
    },
    {
      "epoch": 1.7292001284934146,
      "grad_norm": 0.4523521065711975,
      "learning_rate": 3.943779433094244e-06,
      "loss": 0.3316,
      "step": 37681
    },
    {
      "epoch": 1.7292460189986691,
      "grad_norm": 0.48860740661621094,
      "learning_rate": 3.943539779530522e-06,
      "loss": 0.3721,
      "step": 37682
    },
    {
      "epoch": 1.7292919095039236,
      "grad_norm": 0.4496825933456421,
      "learning_rate": 3.943300128507255e-06,
      "loss": 0.3331,
      "step": 37683
    },
    {
      "epoch": 1.7293378000091781,
      "grad_norm": 0.50135737657547,
      "learning_rate": 3.943060480025015e-06,
      "loss": 0.3845,
      "step": 37684
    },
    {
      "epoch": 1.7293836905144326,
      "grad_norm": 0.4964587390422821,
      "learning_rate": 3.942820834084382e-06,
      "loss": 0.3659,
      "step": 37685
    },
    {
      "epoch": 1.7294295810196871,
      "grad_norm": 0.45602425932884216,
      "learning_rate": 3.942581190685932e-06,
      "loss": 0.338,
      "step": 37686
    },
    {
      "epoch": 1.7294754715249416,
      "grad_norm": 0.4918994605541229,
      "learning_rate": 3.94234154983024e-06,
      "loss": 0.3833,
      "step": 37687
    },
    {
      "epoch": 1.729521362030196,
      "grad_norm": 0.44704511761665344,
      "learning_rate": 3.942101911517883e-06,
      "loss": 0.3239,
      "step": 37688
    },
    {
      "epoch": 1.7295672525354504,
      "grad_norm": 0.4794505536556244,
      "learning_rate": 3.941862275749438e-06,
      "loss": 0.4064,
      "step": 37689
    },
    {
      "epoch": 1.729613143040705,
      "grad_norm": 0.44472235441207886,
      "learning_rate": 3.941622642525479e-06,
      "loss": 0.2972,
      "step": 37690
    },
    {
      "epoch": 1.7296590335459592,
      "grad_norm": 0.4283420443534851,
      "learning_rate": 3.941383011846583e-06,
      "loss": 0.3,
      "step": 37691
    },
    {
      "epoch": 1.7297049240512137,
      "grad_norm": 0.4611454904079437,
      "learning_rate": 3.941143383713327e-06,
      "loss": 0.3427,
      "step": 37692
    },
    {
      "epoch": 1.7297508145564682,
      "grad_norm": 0.4861791431903839,
      "learning_rate": 3.9409037581262885e-06,
      "loss": 0.3976,
      "step": 37693
    },
    {
      "epoch": 1.7297967050617227,
      "grad_norm": 0.4751395881175995,
      "learning_rate": 3.94066413508604e-06,
      "loss": 0.4043,
      "step": 37694
    },
    {
      "epoch": 1.7298425955669772,
      "grad_norm": 0.5078848600387573,
      "learning_rate": 3.940424514593162e-06,
      "loss": 0.4517,
      "step": 37695
    },
    {
      "epoch": 1.7298884860722317,
      "grad_norm": 0.5129518508911133,
      "learning_rate": 3.940184896648228e-06,
      "loss": 0.4431,
      "step": 37696
    },
    {
      "epoch": 1.7299343765774862,
      "grad_norm": 0.4845578968524933,
      "learning_rate": 3.939945281251812e-06,
      "loss": 0.3846,
      "step": 37697
    },
    {
      "epoch": 1.7299802670827407,
      "grad_norm": 0.5167884230613708,
      "learning_rate": 3.939705668404497e-06,
      "loss": 0.444,
      "step": 37698
    },
    {
      "epoch": 1.7300261575879952,
      "grad_norm": 0.44016215205192566,
      "learning_rate": 3.939466058106855e-06,
      "loss": 0.299,
      "step": 37699
    },
    {
      "epoch": 1.7300720480932497,
      "grad_norm": 0.5090606212615967,
      "learning_rate": 3.93922645035946e-06,
      "loss": 0.3952,
      "step": 37700
    },
    {
      "epoch": 1.730117938598504,
      "grad_norm": 0.5415244698524475,
      "learning_rate": 3.938986845162892e-06,
      "loss": 0.4439,
      "step": 37701
    },
    {
      "epoch": 1.7301638291037584,
      "grad_norm": 0.4660992920398712,
      "learning_rate": 3.938747242517726e-06,
      "loss": 0.3299,
      "step": 37702
    },
    {
      "epoch": 1.730209719609013,
      "grad_norm": 0.49812474846839905,
      "learning_rate": 3.938507642424536e-06,
      "loss": 0.3593,
      "step": 37703
    },
    {
      "epoch": 1.7302556101142672,
      "grad_norm": 0.49513715505599976,
      "learning_rate": 3.938268044883903e-06,
      "loss": 0.4197,
      "step": 37704
    },
    {
      "epoch": 1.7303015006195217,
      "grad_norm": 0.4584709703922272,
      "learning_rate": 3.938028449896399e-06,
      "loss": 0.2836,
      "step": 37705
    },
    {
      "epoch": 1.7303473911247762,
      "grad_norm": 0.4992996156215668,
      "learning_rate": 3.9377888574626e-06,
      "loss": 0.4415,
      "step": 37706
    },
    {
      "epoch": 1.7303932816300307,
      "grad_norm": 0.4919486939907074,
      "learning_rate": 3.937549267583085e-06,
      "loss": 0.4121,
      "step": 37707
    },
    {
      "epoch": 1.7304391721352852,
      "grad_norm": 0.45202216506004333,
      "learning_rate": 3.93730968025843e-06,
      "loss": 0.3438,
      "step": 37708
    },
    {
      "epoch": 1.7304850626405397,
      "grad_norm": 0.46126124262809753,
      "learning_rate": 3.937070095489207e-06,
      "loss": 0.3484,
      "step": 37709
    },
    {
      "epoch": 1.7305309531457942,
      "grad_norm": 0.5129637718200684,
      "learning_rate": 3.9368305132759966e-06,
      "loss": 0.4102,
      "step": 37710
    },
    {
      "epoch": 1.7305768436510487,
      "grad_norm": 0.4908868968486786,
      "learning_rate": 3.936590933619374e-06,
      "loss": 0.412,
      "step": 37711
    },
    {
      "epoch": 1.7306227341563032,
      "grad_norm": 0.48706531524658203,
      "learning_rate": 3.936351356519912e-06,
      "loss": 0.3735,
      "step": 37712
    },
    {
      "epoch": 1.7306686246615577,
      "grad_norm": 0.4465389549732208,
      "learning_rate": 3.936111781978191e-06,
      "loss": 0.3509,
      "step": 37713
    },
    {
      "epoch": 1.730714515166812,
      "grad_norm": 0.46393975615501404,
      "learning_rate": 3.935872209994785e-06,
      "loss": 0.3484,
      "step": 37714
    },
    {
      "epoch": 1.7307604056720665,
      "grad_norm": 0.4549098014831543,
      "learning_rate": 3.935632640570271e-06,
      "loss": 0.3192,
      "step": 37715
    },
    {
      "epoch": 1.730806296177321,
      "grad_norm": 0.45554542541503906,
      "learning_rate": 3.935393073705225e-06,
      "loss": 0.3158,
      "step": 37716
    },
    {
      "epoch": 1.7308521866825752,
      "grad_norm": 0.4349183738231659,
      "learning_rate": 3.935153509400222e-06,
      "loss": 0.2848,
      "step": 37717
    },
    {
      "epoch": 1.7308980771878297,
      "grad_norm": 0.46370723843574524,
      "learning_rate": 3.9349139476558365e-06,
      "loss": 0.3544,
      "step": 37718
    },
    {
      "epoch": 1.7309439676930842,
      "grad_norm": 0.45701155066490173,
      "learning_rate": 3.93467438847265e-06,
      "loss": 0.3235,
      "step": 37719
    },
    {
      "epoch": 1.7309898581983387,
      "grad_norm": 0.4520294964313507,
      "learning_rate": 3.934434831851235e-06,
      "loss": 0.3273,
      "step": 37720
    },
    {
      "epoch": 1.7310357487035932,
      "grad_norm": 0.4941861629486084,
      "learning_rate": 3.9341952777921665e-06,
      "loss": 0.3496,
      "step": 37721
    },
    {
      "epoch": 1.7310816392088477,
      "grad_norm": 0.5370991230010986,
      "learning_rate": 3.933955726296023e-06,
      "loss": 0.3949,
      "step": 37722
    },
    {
      "epoch": 1.7311275297141022,
      "grad_norm": 0.5831584930419922,
      "learning_rate": 3.9337161773633795e-06,
      "loss": 0.3201,
      "step": 37723
    },
    {
      "epoch": 1.7311734202193567,
      "grad_norm": 0.5078880190849304,
      "learning_rate": 3.933476630994812e-06,
      "loss": 0.4142,
      "step": 37724
    },
    {
      "epoch": 1.7312193107246112,
      "grad_norm": 0.5126383900642395,
      "learning_rate": 3.933237087190896e-06,
      "loss": 0.3703,
      "step": 37725
    },
    {
      "epoch": 1.7312652012298655,
      "grad_norm": 0.45886334776878357,
      "learning_rate": 3.932997545952209e-06,
      "loss": 0.3131,
      "step": 37726
    },
    {
      "epoch": 1.73131109173512,
      "grad_norm": 0.4797188639640808,
      "learning_rate": 3.9327580072793266e-06,
      "loss": 0.3468,
      "step": 37727
    },
    {
      "epoch": 1.7313569822403745,
      "grad_norm": 0.454182505607605,
      "learning_rate": 3.932518471172822e-06,
      "loss": 0.3686,
      "step": 37728
    },
    {
      "epoch": 1.731402872745629,
      "grad_norm": 0.4468729794025421,
      "learning_rate": 3.9322789376332755e-06,
      "loss": 0.3082,
      "step": 37729
    },
    {
      "epoch": 1.7314487632508833,
      "grad_norm": 0.4744400978088379,
      "learning_rate": 3.932039406661261e-06,
      "loss": 0.3309,
      "step": 37730
    },
    {
      "epoch": 1.7314946537561378,
      "grad_norm": 0.46409016847610474,
      "learning_rate": 3.931799878257354e-06,
      "loss": 0.3167,
      "step": 37731
    },
    {
      "epoch": 1.7315405442613923,
      "grad_norm": 0.44316425919532776,
      "learning_rate": 3.931560352422132e-06,
      "loss": 0.3327,
      "step": 37732
    },
    {
      "epoch": 1.7315864347666468,
      "grad_norm": 0.4968279302120209,
      "learning_rate": 3.93132082915617e-06,
      "loss": 0.3943,
      "step": 37733
    },
    {
      "epoch": 1.7316323252719013,
      "grad_norm": 0.5221280455589294,
      "learning_rate": 3.931081308460042e-06,
      "loss": 0.4385,
      "step": 37734
    },
    {
      "epoch": 1.7316782157771557,
      "grad_norm": 0.5274427533149719,
      "learning_rate": 3.930841790334329e-06,
      "loss": 0.4187,
      "step": 37735
    },
    {
      "epoch": 1.7317241062824102,
      "grad_norm": 0.5154575109481812,
      "learning_rate": 3.930602274779604e-06,
      "loss": 0.3648,
      "step": 37736
    },
    {
      "epoch": 1.7317699967876647,
      "grad_norm": 0.4670393764972687,
      "learning_rate": 3.930362761796441e-06,
      "loss": 0.3628,
      "step": 37737
    },
    {
      "epoch": 1.7318158872929192,
      "grad_norm": 0.46661749482154846,
      "learning_rate": 3.930123251385419e-06,
      "loss": 0.3605,
      "step": 37738
    },
    {
      "epoch": 1.7318617777981735,
      "grad_norm": 0.45243117213249207,
      "learning_rate": 3.929883743547114e-06,
      "loss": 0.3389,
      "step": 37739
    },
    {
      "epoch": 1.731907668303428,
      "grad_norm": 0.5083144903182983,
      "learning_rate": 3.929644238282098e-06,
      "loss": 0.4425,
      "step": 37740
    },
    {
      "epoch": 1.7319535588086825,
      "grad_norm": 0.48450958728790283,
      "learning_rate": 3.929404735590951e-06,
      "loss": 0.3812,
      "step": 37741
    },
    {
      "epoch": 1.7319994493139368,
      "grad_norm": 0.45165616273880005,
      "learning_rate": 3.929165235474249e-06,
      "loss": 0.3324,
      "step": 37742
    },
    {
      "epoch": 1.7320453398191913,
      "grad_norm": 0.47355028986930847,
      "learning_rate": 3.928925737932566e-06,
      "loss": 0.3723,
      "step": 37743
    },
    {
      "epoch": 1.7320912303244458,
      "grad_norm": 0.47502601146698,
      "learning_rate": 3.928686242966478e-06,
      "loss": 0.4058,
      "step": 37744
    },
    {
      "epoch": 1.7321371208297003,
      "grad_norm": 0.4867856204509735,
      "learning_rate": 3.9284467505765625e-06,
      "loss": 0.3834,
      "step": 37745
    },
    {
      "epoch": 1.7321830113349548,
      "grad_norm": 0.46754777431488037,
      "learning_rate": 3.928207260763394e-06,
      "loss": 0.366,
      "step": 37746
    },
    {
      "epoch": 1.7322289018402093,
      "grad_norm": 0.42955413460731506,
      "learning_rate": 3.927967773527549e-06,
      "loss": 0.2755,
      "step": 37747
    },
    {
      "epoch": 1.7322747923454638,
      "grad_norm": 0.45646870136260986,
      "learning_rate": 3.927728288869603e-06,
      "loss": 0.3253,
      "step": 37748
    },
    {
      "epoch": 1.7323206828507183,
      "grad_norm": 0.48685693740844727,
      "learning_rate": 3.9274888067901305e-06,
      "loss": 0.3828,
      "step": 37749
    },
    {
      "epoch": 1.7323665733559728,
      "grad_norm": 0.4678124189376831,
      "learning_rate": 3.927249327289711e-06,
      "loss": 0.3494,
      "step": 37750
    },
    {
      "epoch": 1.7324124638612273,
      "grad_norm": 0.5095535516738892,
      "learning_rate": 3.9270098503689165e-06,
      "loss": 0.4217,
      "step": 37751
    },
    {
      "epoch": 1.7324583543664815,
      "grad_norm": 0.4811328947544098,
      "learning_rate": 3.926770376028326e-06,
      "loss": 0.3633,
      "step": 37752
    },
    {
      "epoch": 1.732504244871736,
      "grad_norm": 0.4730220139026642,
      "learning_rate": 3.926530904268514e-06,
      "loss": 0.3438,
      "step": 37753
    },
    {
      "epoch": 1.7325501353769905,
      "grad_norm": 0.4898549020290375,
      "learning_rate": 3.926291435090057e-06,
      "loss": 0.4089,
      "step": 37754
    },
    {
      "epoch": 1.7325960258822448,
      "grad_norm": 0.4313989579677582,
      "learning_rate": 3.926051968493527e-06,
      "loss": 0.2905,
      "step": 37755
    },
    {
      "epoch": 1.7326419163874993,
      "grad_norm": 0.47588685154914856,
      "learning_rate": 3.9258125044795056e-06,
      "loss": 0.332,
      "step": 37756
    },
    {
      "epoch": 1.7326878068927538,
      "grad_norm": 0.4357886016368866,
      "learning_rate": 3.925573043048567e-06,
      "loss": 0.3184,
      "step": 37757
    },
    {
      "epoch": 1.7327336973980083,
      "grad_norm": 0.471975177526474,
      "learning_rate": 3.925333584201285e-06,
      "loss": 0.3612,
      "step": 37758
    },
    {
      "epoch": 1.7327795879032628,
      "grad_norm": 0.5200879573822021,
      "learning_rate": 3.925094127938236e-06,
      "loss": 0.3357,
      "step": 37759
    },
    {
      "epoch": 1.7328254784085173,
      "grad_norm": 0.5017543435096741,
      "learning_rate": 3.924854674259998e-06,
      "loss": 0.3778,
      "step": 37760
    },
    {
      "epoch": 1.7328713689137718,
      "grad_norm": 0.47081589698791504,
      "learning_rate": 3.9246152231671445e-06,
      "loss": 0.3762,
      "step": 37761
    },
    {
      "epoch": 1.7329172594190263,
      "grad_norm": 0.4721234142780304,
      "learning_rate": 3.92437577466025e-06,
      "loss": 0.345,
      "step": 37762
    },
    {
      "epoch": 1.7329631499242808,
      "grad_norm": 0.447214275598526,
      "learning_rate": 3.924136328739893e-06,
      "loss": 0.3457,
      "step": 37763
    },
    {
      "epoch": 1.733009040429535,
      "grad_norm": 0.4461555778980255,
      "learning_rate": 3.923896885406651e-06,
      "loss": 0.3062,
      "step": 37764
    },
    {
      "epoch": 1.7330549309347896,
      "grad_norm": 0.4548856019973755,
      "learning_rate": 3.923657444661095e-06,
      "loss": 0.2986,
      "step": 37765
    },
    {
      "epoch": 1.733100821440044,
      "grad_norm": 0.4415822923183441,
      "learning_rate": 3.923418006503805e-06,
      "loss": 0.3201,
      "step": 37766
    },
    {
      "epoch": 1.7331467119452986,
      "grad_norm": 0.4948274791240692,
      "learning_rate": 3.923178570935354e-06,
      "loss": 0.3761,
      "step": 37767
    },
    {
      "epoch": 1.7331926024505528,
      "grad_norm": 0.5065581798553467,
      "learning_rate": 3.922939137956317e-06,
      "loss": 0.4441,
      "step": 37768
    },
    {
      "epoch": 1.7332384929558073,
      "grad_norm": 0.47803249955177307,
      "learning_rate": 3.9226997075672745e-06,
      "loss": 0.3648,
      "step": 37769
    },
    {
      "epoch": 1.7332843834610618,
      "grad_norm": 0.4461725652217865,
      "learning_rate": 3.922460279768798e-06,
      "loss": 0.3262,
      "step": 37770
    },
    {
      "epoch": 1.7333302739663163,
      "grad_norm": 0.44367727637290955,
      "learning_rate": 3.9222208545614625e-06,
      "loss": 0.2871,
      "step": 37771
    },
    {
      "epoch": 1.7333761644715708,
      "grad_norm": 0.47054678201675415,
      "learning_rate": 3.921981431945848e-06,
      "loss": 0.3315,
      "step": 37772
    },
    {
      "epoch": 1.7334220549768253,
      "grad_norm": 0.4531335234642029,
      "learning_rate": 3.921742011922527e-06,
      "loss": 0.3242,
      "step": 37773
    },
    {
      "epoch": 1.7334679454820798,
      "grad_norm": 0.4755464494228363,
      "learning_rate": 3.9215025944920765e-06,
      "loss": 0.3813,
      "step": 37774
    },
    {
      "epoch": 1.7335138359873343,
      "grad_norm": 0.4542332887649536,
      "learning_rate": 3.921263179655072e-06,
      "loss": 0.3454,
      "step": 37775
    },
    {
      "epoch": 1.7335597264925888,
      "grad_norm": 0.517070472240448,
      "learning_rate": 3.92102376741209e-06,
      "loss": 0.4176,
      "step": 37776
    },
    {
      "epoch": 1.733605616997843,
      "grad_norm": 0.49232161045074463,
      "learning_rate": 3.9207843577637015e-06,
      "loss": 0.3553,
      "step": 37777
    },
    {
      "epoch": 1.7336515075030976,
      "grad_norm": 0.4544234871864319,
      "learning_rate": 3.92054495071049e-06,
      "loss": 0.3456,
      "step": 37778
    },
    {
      "epoch": 1.733697398008352,
      "grad_norm": 0.4365238845348358,
      "learning_rate": 3.920305546253026e-06,
      "loss": 0.2888,
      "step": 37779
    },
    {
      "epoch": 1.7337432885136064,
      "grad_norm": 0.4615754187107086,
      "learning_rate": 3.920066144391886e-06,
      "loss": 0.3342,
      "step": 37780
    },
    {
      "epoch": 1.7337891790188609,
      "grad_norm": 0.43486255407333374,
      "learning_rate": 3.919826745127647e-06,
      "loss": 0.314,
      "step": 37781
    },
    {
      "epoch": 1.7338350695241154,
      "grad_norm": 0.42503219842910767,
      "learning_rate": 3.919587348460883e-06,
      "loss": 0.2808,
      "step": 37782
    },
    {
      "epoch": 1.7338809600293699,
      "grad_norm": 0.4473850429058075,
      "learning_rate": 3.9193479543921695e-06,
      "loss": 0.3265,
      "step": 37783
    },
    {
      "epoch": 1.7339268505346244,
      "grad_norm": 0.509564995765686,
      "learning_rate": 3.919108562922084e-06,
      "loss": 0.4426,
      "step": 37784
    },
    {
      "epoch": 1.7339727410398789,
      "grad_norm": 0.4935932457447052,
      "learning_rate": 3.918869174051202e-06,
      "loss": 0.3907,
      "step": 37785
    },
    {
      "epoch": 1.7340186315451334,
      "grad_norm": 0.47518447041511536,
      "learning_rate": 3.918629787780098e-06,
      "loss": 0.3655,
      "step": 37786
    },
    {
      "epoch": 1.7340645220503879,
      "grad_norm": 0.46330776810646057,
      "learning_rate": 3.918390404109348e-06,
      "loss": 0.2894,
      "step": 37787
    },
    {
      "epoch": 1.7341104125556424,
      "grad_norm": 0.4758063554763794,
      "learning_rate": 3.918151023039528e-06,
      "loss": 0.3599,
      "step": 37788
    },
    {
      "epoch": 1.7341563030608969,
      "grad_norm": 0.4561474323272705,
      "learning_rate": 3.917911644571214e-06,
      "loss": 0.3115,
      "step": 37789
    },
    {
      "epoch": 1.7342021935661511,
      "grad_norm": 0.4917207360267639,
      "learning_rate": 3.917672268704979e-06,
      "loss": 0.3709,
      "step": 37790
    },
    {
      "epoch": 1.7342480840714056,
      "grad_norm": 0.43501728773117065,
      "learning_rate": 3.917432895441402e-06,
      "loss": 0.3161,
      "step": 37791
    },
    {
      "epoch": 1.7342939745766601,
      "grad_norm": 0.5072606205940247,
      "learning_rate": 3.917193524781058e-06,
      "loss": 0.4011,
      "step": 37792
    },
    {
      "epoch": 1.7343398650819144,
      "grad_norm": 0.4619543254375458,
      "learning_rate": 3.916954156724518e-06,
      "loss": 0.3611,
      "step": 37793
    },
    {
      "epoch": 1.734385755587169,
      "grad_norm": 0.4629977345466614,
      "learning_rate": 3.916714791272365e-06,
      "loss": 0.3181,
      "step": 37794
    },
    {
      "epoch": 1.7344316460924234,
      "grad_norm": 0.49492716789245605,
      "learning_rate": 3.916475428425171e-06,
      "loss": 0.4117,
      "step": 37795
    },
    {
      "epoch": 1.734477536597678,
      "grad_norm": 0.4954071640968323,
      "learning_rate": 3.9162360681835095e-06,
      "loss": 0.41,
      "step": 37796
    },
    {
      "epoch": 1.7345234271029324,
      "grad_norm": 0.4620422422885895,
      "learning_rate": 3.9159967105479604e-06,
      "loss": 0.3468,
      "step": 37797
    },
    {
      "epoch": 1.734569317608187,
      "grad_norm": 0.4937390685081482,
      "learning_rate": 3.915757355519097e-06,
      "loss": 0.3839,
      "step": 37798
    },
    {
      "epoch": 1.7346152081134414,
      "grad_norm": 0.4778202772140503,
      "learning_rate": 3.915518003097492e-06,
      "loss": 0.3784,
      "step": 37799
    },
    {
      "epoch": 1.7346610986186959,
      "grad_norm": 0.49725303053855896,
      "learning_rate": 3.915278653283726e-06,
      "loss": 0.3772,
      "step": 37800
    },
    {
      "epoch": 1.7347069891239504,
      "grad_norm": 0.4928148686885834,
      "learning_rate": 3.915039306078373e-06,
      "loss": 0.4283,
      "step": 37801
    },
    {
      "epoch": 1.7347528796292049,
      "grad_norm": 0.4940696358680725,
      "learning_rate": 3.914799961482008e-06,
      "loss": 0.3998,
      "step": 37802
    },
    {
      "epoch": 1.7347987701344592,
      "grad_norm": 0.5216817259788513,
      "learning_rate": 3.914560619495206e-06,
      "loss": 0.371,
      "step": 37803
    },
    {
      "epoch": 1.7348446606397137,
      "grad_norm": 0.48414674401283264,
      "learning_rate": 3.914321280118544e-06,
      "loss": 0.3836,
      "step": 37804
    },
    {
      "epoch": 1.7348905511449682,
      "grad_norm": 0.48860999941825867,
      "learning_rate": 3.914081943352594e-06,
      "loss": 0.3954,
      "step": 37805
    },
    {
      "epoch": 1.7349364416502224,
      "grad_norm": 0.44745299220085144,
      "learning_rate": 3.913842609197937e-06,
      "loss": 0.3297,
      "step": 37806
    },
    {
      "epoch": 1.734982332155477,
      "grad_norm": 0.4577964246273041,
      "learning_rate": 3.913603277655145e-06,
      "loss": 0.3199,
      "step": 37807
    },
    {
      "epoch": 1.7350282226607314,
      "grad_norm": 0.43561914563179016,
      "learning_rate": 3.913363948724794e-06,
      "loss": 0.2885,
      "step": 37808
    },
    {
      "epoch": 1.735074113165986,
      "grad_norm": 0.4532536566257477,
      "learning_rate": 3.91312462240746e-06,
      "loss": 0.3231,
      "step": 37809
    },
    {
      "epoch": 1.7351200036712404,
      "grad_norm": 0.4688645601272583,
      "learning_rate": 3.9128852987037184e-06,
      "loss": 0.3641,
      "step": 37810
    },
    {
      "epoch": 1.735165894176495,
      "grad_norm": 0.5063457489013672,
      "learning_rate": 3.912645977614143e-06,
      "loss": 0.3543,
      "step": 37811
    },
    {
      "epoch": 1.7352117846817494,
      "grad_norm": 0.5305487513542175,
      "learning_rate": 3.912406659139314e-06,
      "loss": 0.4038,
      "step": 37812
    },
    {
      "epoch": 1.735257675187004,
      "grad_norm": 0.5343021154403687,
      "learning_rate": 3.9121673432798015e-06,
      "loss": 0.3274,
      "step": 37813
    },
    {
      "epoch": 1.7353035656922584,
      "grad_norm": 0.4330636262893677,
      "learning_rate": 3.911928030036182e-06,
      "loss": 0.3117,
      "step": 37814
    },
    {
      "epoch": 1.7353494561975127,
      "grad_norm": 0.4431147575378418,
      "learning_rate": 3.911688719409034e-06,
      "loss": 0.2912,
      "step": 37815
    },
    {
      "epoch": 1.7353953467027672,
      "grad_norm": 0.45175331830978394,
      "learning_rate": 3.911449411398932e-06,
      "loss": 0.3238,
      "step": 37816
    },
    {
      "epoch": 1.7354412372080217,
      "grad_norm": 0.49868541955947876,
      "learning_rate": 3.911210106006449e-06,
      "loss": 0.4066,
      "step": 37817
    },
    {
      "epoch": 1.7354871277132762,
      "grad_norm": 0.4436154365539551,
      "learning_rate": 3.910970803232163e-06,
      "loss": 0.301,
      "step": 37818
    },
    {
      "epoch": 1.7355330182185305,
      "grad_norm": 0.4512239992618561,
      "learning_rate": 3.910731503076648e-06,
      "loss": 0.3159,
      "step": 37819
    },
    {
      "epoch": 1.735578908723785,
      "grad_norm": 0.4715718924999237,
      "learning_rate": 3.910492205540479e-06,
      "loss": 0.3429,
      "step": 37820
    },
    {
      "epoch": 1.7356247992290394,
      "grad_norm": 0.4154037535190582,
      "learning_rate": 3.910252910624234e-06,
      "loss": 0.2823,
      "step": 37821
    },
    {
      "epoch": 1.735670689734294,
      "grad_norm": 0.45371514558792114,
      "learning_rate": 3.910013618328487e-06,
      "loss": 0.3496,
      "step": 37822
    },
    {
      "epoch": 1.7357165802395484,
      "grad_norm": 0.5113012194633484,
      "learning_rate": 3.909774328653812e-06,
      "loss": 0.4332,
      "step": 37823
    },
    {
      "epoch": 1.735762470744803,
      "grad_norm": 0.46074530482292175,
      "learning_rate": 3.909535041600786e-06,
      "loss": 0.3601,
      "step": 37824
    },
    {
      "epoch": 1.7358083612500574,
      "grad_norm": 0.4375602900981903,
      "learning_rate": 3.909295757169985e-06,
      "loss": 0.2642,
      "step": 37825
    },
    {
      "epoch": 1.735854251755312,
      "grad_norm": 0.48137366771698,
      "learning_rate": 3.909056475361983e-06,
      "loss": 0.3385,
      "step": 37826
    },
    {
      "epoch": 1.7359001422605664,
      "grad_norm": 0.4824962317943573,
      "learning_rate": 3.908817196177353e-06,
      "loss": 0.364,
      "step": 37827
    },
    {
      "epoch": 1.7359460327658207,
      "grad_norm": 0.468479186296463,
      "learning_rate": 3.9085779196166775e-06,
      "loss": 0.3318,
      "step": 37828
    },
    {
      "epoch": 1.7359919232710752,
      "grad_norm": 0.47652795910835266,
      "learning_rate": 3.908338645680526e-06,
      "loss": 0.3164,
      "step": 37829
    },
    {
      "epoch": 1.7360378137763297,
      "grad_norm": 0.4329882860183716,
      "learning_rate": 3.908099374369475e-06,
      "loss": 0.3087,
      "step": 37830
    },
    {
      "epoch": 1.736083704281584,
      "grad_norm": 0.49736976623535156,
      "learning_rate": 3.907860105684101e-06,
      "loss": 0.4145,
      "step": 37831
    },
    {
      "epoch": 1.7361295947868385,
      "grad_norm": 0.4967638850212097,
      "learning_rate": 3.907620839624979e-06,
      "loss": 0.3636,
      "step": 37832
    },
    {
      "epoch": 1.736175485292093,
      "grad_norm": 0.48689496517181396,
      "learning_rate": 3.907381576192682e-06,
      "loss": 0.3768,
      "step": 37833
    },
    {
      "epoch": 1.7362213757973475,
      "grad_norm": 0.4676749110221863,
      "learning_rate": 3.907142315387791e-06,
      "loss": 0.3757,
      "step": 37834
    },
    {
      "epoch": 1.736267266302602,
      "grad_norm": 0.4551714062690735,
      "learning_rate": 3.906903057210876e-06,
      "loss": 0.3577,
      "step": 37835
    },
    {
      "epoch": 1.7363131568078565,
      "grad_norm": 0.43191415071487427,
      "learning_rate": 3.906663801662513e-06,
      "loss": 0.3272,
      "step": 37836
    },
    {
      "epoch": 1.736359047313111,
      "grad_norm": 0.5016480684280396,
      "learning_rate": 3.90642454874328e-06,
      "loss": 0.4546,
      "step": 37837
    },
    {
      "epoch": 1.7364049378183655,
      "grad_norm": 0.48935431241989136,
      "learning_rate": 3.906185298453751e-06,
      "loss": 0.4106,
      "step": 37838
    },
    {
      "epoch": 1.73645082832362,
      "grad_norm": 0.46445828676223755,
      "learning_rate": 3.9059460507945e-06,
      "loss": 0.3447,
      "step": 37839
    },
    {
      "epoch": 1.7364967188288745,
      "grad_norm": 0.46387794613838196,
      "learning_rate": 3.905706805766105e-06,
      "loss": 0.3653,
      "step": 37840
    },
    {
      "epoch": 1.7365426093341287,
      "grad_norm": 0.45771172642707825,
      "learning_rate": 3.905467563369139e-06,
      "loss": 0.3098,
      "step": 37841
    },
    {
      "epoch": 1.7365884998393832,
      "grad_norm": 0.5334724187850952,
      "learning_rate": 3.905228323604176e-06,
      "loss": 0.4786,
      "step": 37842
    },
    {
      "epoch": 1.7366343903446377,
      "grad_norm": 0.4668572247028351,
      "learning_rate": 3.904989086471796e-06,
      "loss": 0.3292,
      "step": 37843
    },
    {
      "epoch": 1.736680280849892,
      "grad_norm": 0.460836261510849,
      "learning_rate": 3.9047498519725714e-06,
      "loss": 0.2868,
      "step": 37844
    },
    {
      "epoch": 1.7367261713551465,
      "grad_norm": 0.46095919609069824,
      "learning_rate": 3.9045106201070764e-06,
      "loss": 0.3335,
      "step": 37845
    },
    {
      "epoch": 1.736772061860401,
      "grad_norm": 0.47260582447052,
      "learning_rate": 3.904271390875889e-06,
      "loss": 0.3854,
      "step": 37846
    },
    {
      "epoch": 1.7368179523656555,
      "grad_norm": 0.4793805181980133,
      "learning_rate": 3.904032164279583e-06,
      "loss": 0.361,
      "step": 37847
    },
    {
      "epoch": 1.73686384287091,
      "grad_norm": 0.4562554359436035,
      "learning_rate": 3.903792940318732e-06,
      "loss": 0.3444,
      "step": 37848
    },
    {
      "epoch": 1.7369097333761645,
      "grad_norm": 0.4470905363559723,
      "learning_rate": 3.903553718993915e-06,
      "loss": 0.3289,
      "step": 37849
    },
    {
      "epoch": 1.736955623881419,
      "grad_norm": 0.48299121856689453,
      "learning_rate": 3.903314500305704e-06,
      "loss": 0.3625,
      "step": 37850
    },
    {
      "epoch": 1.7370015143866735,
      "grad_norm": 0.4796854555606842,
      "learning_rate": 3.903075284254676e-06,
      "loss": 0.3558,
      "step": 37851
    },
    {
      "epoch": 1.737047404891928,
      "grad_norm": 0.4653775691986084,
      "learning_rate": 3.902836070841406e-06,
      "loss": 0.3361,
      "step": 37852
    },
    {
      "epoch": 1.7370932953971823,
      "grad_norm": 0.4681016504764557,
      "learning_rate": 3.902596860066469e-06,
      "loss": 0.343,
      "step": 37853
    },
    {
      "epoch": 1.7371391859024368,
      "grad_norm": 0.7698867321014404,
      "learning_rate": 3.902357651930441e-06,
      "loss": 0.3819,
      "step": 37854
    },
    {
      "epoch": 1.7371850764076913,
      "grad_norm": 0.4971826374530792,
      "learning_rate": 3.902118446433893e-06,
      "loss": 0.389,
      "step": 37855
    },
    {
      "epoch": 1.7372309669129458,
      "grad_norm": 0.4718954265117645,
      "learning_rate": 3.901879243577408e-06,
      "loss": 0.3738,
      "step": 37856
    },
    {
      "epoch": 1.7372768574182,
      "grad_norm": 0.5012049674987793,
      "learning_rate": 3.901640043361556e-06,
      "loss": 0.443,
      "step": 37857
    },
    {
      "epoch": 1.7373227479234545,
      "grad_norm": 0.4853041470050812,
      "learning_rate": 3.90140084578691e-06,
      "loss": 0.3881,
      "step": 37858
    },
    {
      "epoch": 1.737368638428709,
      "grad_norm": 0.45671334862709045,
      "learning_rate": 3.90116165085405e-06,
      "loss": 0.3253,
      "step": 37859
    },
    {
      "epoch": 1.7374145289339635,
      "grad_norm": 0.5007564425468445,
      "learning_rate": 3.900922458563551e-06,
      "loss": 0.3469,
      "step": 37860
    },
    {
      "epoch": 1.737460419439218,
      "grad_norm": 0.44194164872169495,
      "learning_rate": 3.9006832689159845e-06,
      "loss": 0.3252,
      "step": 37861
    },
    {
      "epoch": 1.7375063099444725,
      "grad_norm": 0.46637746691703796,
      "learning_rate": 3.9004440819119285e-06,
      "loss": 0.3578,
      "step": 37862
    },
    {
      "epoch": 1.737552200449727,
      "grad_norm": 0.4666733741760254,
      "learning_rate": 3.900204897551958e-06,
      "loss": 0.3566,
      "step": 37863
    },
    {
      "epoch": 1.7375980909549815,
      "grad_norm": 0.4632923901081085,
      "learning_rate": 3.899965715836645e-06,
      "loss": 0.3248,
      "step": 37864
    },
    {
      "epoch": 1.737643981460236,
      "grad_norm": 0.45875582098960876,
      "learning_rate": 3.89972653676657e-06,
      "loss": 0.3055,
      "step": 37865
    },
    {
      "epoch": 1.7376898719654903,
      "grad_norm": 0.4747883081436157,
      "learning_rate": 3.899487360342305e-06,
      "loss": 0.3704,
      "step": 37866
    },
    {
      "epoch": 1.7377357624707448,
      "grad_norm": 0.45304328203201294,
      "learning_rate": 3.899248186564425e-06,
      "loss": 0.3078,
      "step": 37867
    },
    {
      "epoch": 1.7377816529759993,
      "grad_norm": 0.5072611570358276,
      "learning_rate": 3.899009015433506e-06,
      "loss": 0.4126,
      "step": 37868
    },
    {
      "epoch": 1.7378275434812536,
      "grad_norm": 0.4828420877456665,
      "learning_rate": 3.8987698469501236e-06,
      "loss": 0.3757,
      "step": 37869
    },
    {
      "epoch": 1.737873433986508,
      "grad_norm": 0.44230419397354126,
      "learning_rate": 3.898530681114849e-06,
      "loss": 0.3161,
      "step": 37870
    },
    {
      "epoch": 1.7379193244917626,
      "grad_norm": 0.4832528233528137,
      "learning_rate": 3.898291517928264e-06,
      "loss": 0.3794,
      "step": 37871
    },
    {
      "epoch": 1.737965214997017,
      "grad_norm": 0.49374139308929443,
      "learning_rate": 3.8980523573909394e-06,
      "loss": 0.361,
      "step": 37872
    },
    {
      "epoch": 1.7380111055022716,
      "grad_norm": 0.5273470282554626,
      "learning_rate": 3.8978131995034504e-06,
      "loss": 0.3839,
      "step": 37873
    },
    {
      "epoch": 1.738056996007526,
      "grad_norm": 0.46764346957206726,
      "learning_rate": 3.897574044266373e-06,
      "loss": 0.3841,
      "step": 37874
    },
    {
      "epoch": 1.7381028865127806,
      "grad_norm": 0.4615870416164398,
      "learning_rate": 3.897334891680283e-06,
      "loss": 0.354,
      "step": 37875
    },
    {
      "epoch": 1.738148777018035,
      "grad_norm": 0.48016002774238586,
      "learning_rate": 3.897095741745752e-06,
      "loss": 0.3849,
      "step": 37876
    },
    {
      "epoch": 1.7381946675232895,
      "grad_norm": 0.4906712472438812,
      "learning_rate": 3.89685659446336e-06,
      "loss": 0.373,
      "step": 37877
    },
    {
      "epoch": 1.738240558028544,
      "grad_norm": 0.4949273467063904,
      "learning_rate": 3.896617449833681e-06,
      "loss": 0.4256,
      "step": 37878
    },
    {
      "epoch": 1.7382864485337983,
      "grad_norm": 0.4806094765663147,
      "learning_rate": 3.896378307857285e-06,
      "loss": 0.3768,
      "step": 37879
    },
    {
      "epoch": 1.7383323390390528,
      "grad_norm": 0.47996315360069275,
      "learning_rate": 3.8961391685347536e-06,
      "loss": 0.3747,
      "step": 37880
    },
    {
      "epoch": 1.7383782295443073,
      "grad_norm": 0.45010027289390564,
      "learning_rate": 3.895900031866659e-06,
      "loss": 0.3464,
      "step": 37881
    },
    {
      "epoch": 1.7384241200495616,
      "grad_norm": 0.48724475502967834,
      "learning_rate": 3.895660897853575e-06,
      "loss": 0.4224,
      "step": 37882
    },
    {
      "epoch": 1.738470010554816,
      "grad_norm": 0.4635736644268036,
      "learning_rate": 3.895421766496079e-06,
      "loss": 0.3609,
      "step": 37883
    },
    {
      "epoch": 1.7385159010600706,
      "grad_norm": 0.42947572469711304,
      "learning_rate": 3.895182637794746e-06,
      "loss": 0.3239,
      "step": 37884
    },
    {
      "epoch": 1.738561791565325,
      "grad_norm": 0.4659508466720581,
      "learning_rate": 3.8949435117501465e-06,
      "loss": 0.3622,
      "step": 37885
    },
    {
      "epoch": 1.7386076820705796,
      "grad_norm": 0.528142511844635,
      "learning_rate": 3.8947043883628626e-06,
      "loss": 0.4175,
      "step": 37886
    },
    {
      "epoch": 1.738653572575834,
      "grad_norm": 0.44024258852005005,
      "learning_rate": 3.894465267633465e-06,
      "loss": 0.3277,
      "step": 37887
    },
    {
      "epoch": 1.7386994630810886,
      "grad_norm": 0.47341403365135193,
      "learning_rate": 3.894226149562529e-06,
      "loss": 0.3291,
      "step": 37888
    },
    {
      "epoch": 1.738745353586343,
      "grad_norm": 0.4834555685520172,
      "learning_rate": 3.89398703415063e-06,
      "loss": 0.3265,
      "step": 37889
    },
    {
      "epoch": 1.7387912440915976,
      "grad_norm": 0.44103726744651794,
      "learning_rate": 3.893747921398344e-06,
      "loss": 0.3373,
      "step": 37890
    },
    {
      "epoch": 1.738837134596852,
      "grad_norm": 0.4566248953342438,
      "learning_rate": 3.893508811306245e-06,
      "loss": 0.3048,
      "step": 37891
    },
    {
      "epoch": 1.7388830251021063,
      "grad_norm": 0.519572377204895,
      "learning_rate": 3.893269703874906e-06,
      "loss": 0.427,
      "step": 37892
    },
    {
      "epoch": 1.7389289156073608,
      "grad_norm": 0.48425421118736267,
      "learning_rate": 3.893030599104905e-06,
      "loss": 0.3811,
      "step": 37893
    },
    {
      "epoch": 1.7389748061126153,
      "grad_norm": 0.4950339198112488,
      "learning_rate": 3.892791496996818e-06,
      "loss": 0.3774,
      "step": 37894
    },
    {
      "epoch": 1.7390206966178696,
      "grad_norm": 0.48766353726387024,
      "learning_rate": 3.892552397551216e-06,
      "loss": 0.3651,
      "step": 37895
    },
    {
      "epoch": 1.7390665871231241,
      "grad_norm": 0.48620134592056274,
      "learning_rate": 3.892313300768677e-06,
      "loss": 0.3561,
      "step": 37896
    },
    {
      "epoch": 1.7391124776283786,
      "grad_norm": 0.4747154116630554,
      "learning_rate": 3.892074206649775e-06,
      "loss": 0.3152,
      "step": 37897
    },
    {
      "epoch": 1.7391583681336331,
      "grad_norm": 0.4623135030269623,
      "learning_rate": 3.891835115195083e-06,
      "loss": 0.3372,
      "step": 37898
    },
    {
      "epoch": 1.7392042586388876,
      "grad_norm": 0.4672326147556305,
      "learning_rate": 3.89159602640518e-06,
      "loss": 0.3594,
      "step": 37899
    },
    {
      "epoch": 1.739250149144142,
      "grad_norm": 0.49602916836738586,
      "learning_rate": 3.891356940280639e-06,
      "loss": 0.3748,
      "step": 37900
    },
    {
      "epoch": 1.7392960396493966,
      "grad_norm": 0.5135919451713562,
      "learning_rate": 3.891117856822032e-06,
      "loss": 0.4026,
      "step": 37901
    },
    {
      "epoch": 1.739341930154651,
      "grad_norm": 0.4555191993713379,
      "learning_rate": 3.890878776029938e-06,
      "loss": 0.3252,
      "step": 37902
    },
    {
      "epoch": 1.7393878206599056,
      "grad_norm": 0.4648953676223755,
      "learning_rate": 3.890639697904932e-06,
      "loss": 0.3169,
      "step": 37903
    },
    {
      "epoch": 1.7394337111651599,
      "grad_norm": 0.4850652515888214,
      "learning_rate": 3.890400622447586e-06,
      "loss": 0.4249,
      "step": 37904
    },
    {
      "epoch": 1.7394796016704144,
      "grad_norm": 0.45225778222084045,
      "learning_rate": 3.8901615496584764e-06,
      "loss": 0.3069,
      "step": 37905
    },
    {
      "epoch": 1.7395254921756689,
      "grad_norm": 0.4264868199825287,
      "learning_rate": 3.889922479538178e-06,
      "loss": 0.2836,
      "step": 37906
    },
    {
      "epoch": 1.7395713826809232,
      "grad_norm": 0.4594583213329315,
      "learning_rate": 3.889683412087264e-06,
      "loss": 0.3166,
      "step": 37907
    },
    {
      "epoch": 1.7396172731861776,
      "grad_norm": 0.44186604022979736,
      "learning_rate": 3.889444347306313e-06,
      "loss": 0.3094,
      "step": 37908
    },
    {
      "epoch": 1.7396631636914321,
      "grad_norm": 0.470024436712265,
      "learning_rate": 3.889205285195898e-06,
      "loss": 0.341,
      "step": 37909
    },
    {
      "epoch": 1.7397090541966866,
      "grad_norm": 0.5121283531188965,
      "learning_rate": 3.888966225756593e-06,
      "loss": 0.4008,
      "step": 37910
    },
    {
      "epoch": 1.7397549447019411,
      "grad_norm": 0.5117537975311279,
      "learning_rate": 3.888727168988974e-06,
      "loss": 0.4012,
      "step": 37911
    },
    {
      "epoch": 1.7398008352071956,
      "grad_norm": 0.5018085837364197,
      "learning_rate": 3.888488114893616e-06,
      "loss": 0.4497,
      "step": 37912
    },
    {
      "epoch": 1.7398467257124501,
      "grad_norm": 0.4352177083492279,
      "learning_rate": 3.888249063471091e-06,
      "loss": 0.2829,
      "step": 37913
    },
    {
      "epoch": 1.7398926162177046,
      "grad_norm": 0.4980044364929199,
      "learning_rate": 3.888010014721978e-06,
      "loss": 0.433,
      "step": 37914
    },
    {
      "epoch": 1.7399385067229591,
      "grad_norm": 0.49845001101493835,
      "learning_rate": 3.887770968646849e-06,
      "loss": 0.4092,
      "step": 37915
    },
    {
      "epoch": 1.7399843972282136,
      "grad_norm": 0.47982701659202576,
      "learning_rate": 3.887531925246281e-06,
      "loss": 0.3676,
      "step": 37916
    },
    {
      "epoch": 1.740030287733468,
      "grad_norm": 0.44795164465904236,
      "learning_rate": 3.8872928845208465e-06,
      "loss": 0.3471,
      "step": 37917
    },
    {
      "epoch": 1.7400761782387224,
      "grad_norm": 0.4580914080142975,
      "learning_rate": 3.887053846471122e-06,
      "loss": 0.3083,
      "step": 37918
    },
    {
      "epoch": 1.740122068743977,
      "grad_norm": 0.5032056570053101,
      "learning_rate": 3.886814811097683e-06,
      "loss": 0.4782,
      "step": 37919
    },
    {
      "epoch": 1.7401679592492312,
      "grad_norm": 0.49358004331588745,
      "learning_rate": 3.8865757784011e-06,
      "loss": 0.3868,
      "step": 37920
    },
    {
      "epoch": 1.7402138497544857,
      "grad_norm": 0.4595424234867096,
      "learning_rate": 3.886336748381953e-06,
      "loss": 0.3818,
      "step": 37921
    },
    {
      "epoch": 1.7402597402597402,
      "grad_norm": 0.48590585589408875,
      "learning_rate": 3.886097721040816e-06,
      "loss": 0.4278,
      "step": 37922
    },
    {
      "epoch": 1.7403056307649947,
      "grad_norm": 0.4943716526031494,
      "learning_rate": 3.885858696378258e-06,
      "loss": 0.4097,
      "step": 37923
    },
    {
      "epoch": 1.7403515212702492,
      "grad_norm": 0.4562951624393463,
      "learning_rate": 3.885619674394862e-06,
      "loss": 0.3746,
      "step": 37924
    },
    {
      "epoch": 1.7403974117755037,
      "grad_norm": 0.4828827977180481,
      "learning_rate": 3.885380655091198e-06,
      "loss": 0.3777,
      "step": 37925
    },
    {
      "epoch": 1.7404433022807582,
      "grad_norm": 0.45610445737838745,
      "learning_rate": 3.885141638467841e-06,
      "loss": 0.3376,
      "step": 37926
    },
    {
      "epoch": 1.7404891927860127,
      "grad_norm": 0.45092371106147766,
      "learning_rate": 3.884902624525367e-06,
      "loss": 0.318,
      "step": 37927
    },
    {
      "epoch": 1.7405350832912672,
      "grad_norm": 0.4492829740047455,
      "learning_rate": 3.884663613264351e-06,
      "loss": 0.2999,
      "step": 37928
    },
    {
      "epoch": 1.7405809737965217,
      "grad_norm": 0.4760184586048126,
      "learning_rate": 3.884424604685365e-06,
      "loss": 0.3597,
      "step": 37929
    },
    {
      "epoch": 1.740626864301776,
      "grad_norm": 0.4797089099884033,
      "learning_rate": 3.884185598788988e-06,
      "loss": 0.4083,
      "step": 37930
    },
    {
      "epoch": 1.7406727548070304,
      "grad_norm": 0.42375004291534424,
      "learning_rate": 3.883946595575792e-06,
      "loss": 0.261,
      "step": 37931
    },
    {
      "epoch": 1.740718645312285,
      "grad_norm": 0.44299307465553284,
      "learning_rate": 3.883707595046351e-06,
      "loss": 0.3042,
      "step": 37932
    },
    {
      "epoch": 1.7407645358175392,
      "grad_norm": 0.5549852848052979,
      "learning_rate": 3.883468597201241e-06,
      "loss": 0.4073,
      "step": 37933
    },
    {
      "epoch": 1.7408104263227937,
      "grad_norm": 0.4480770528316498,
      "learning_rate": 3.883229602041038e-06,
      "loss": 0.3319,
      "step": 37934
    },
    {
      "epoch": 1.7408563168280482,
      "grad_norm": 0.4775012135505676,
      "learning_rate": 3.882990609566314e-06,
      "loss": 0.349,
      "step": 37935
    },
    {
      "epoch": 1.7409022073333027,
      "grad_norm": 0.45379960536956787,
      "learning_rate": 3.882751619777645e-06,
      "loss": 0.3435,
      "step": 37936
    },
    {
      "epoch": 1.7409480978385572,
      "grad_norm": 0.44584015011787415,
      "learning_rate": 3.882512632675608e-06,
      "loss": 0.3017,
      "step": 37937
    },
    {
      "epoch": 1.7409939883438117,
      "grad_norm": 0.48601266741752625,
      "learning_rate": 3.882273648260773e-06,
      "loss": 0.4027,
      "step": 37938
    },
    {
      "epoch": 1.7410398788490662,
      "grad_norm": 0.45220714807510376,
      "learning_rate": 3.882034666533718e-06,
      "loss": 0.314,
      "step": 37939
    },
    {
      "epoch": 1.7410857693543207,
      "grad_norm": 0.47872194647789,
      "learning_rate": 3.881795687495017e-06,
      "loss": 0.4294,
      "step": 37940
    },
    {
      "epoch": 1.7411316598595752,
      "grad_norm": 0.5045514702796936,
      "learning_rate": 3.881556711145242e-06,
      "loss": 0.4023,
      "step": 37941
    },
    {
      "epoch": 1.7411775503648295,
      "grad_norm": 0.496366947889328,
      "learning_rate": 3.881317737484973e-06,
      "loss": 0.4119,
      "step": 37942
    },
    {
      "epoch": 1.741223440870084,
      "grad_norm": 0.4768097996711731,
      "learning_rate": 3.881078766514782e-06,
      "loss": 0.4053,
      "step": 37943
    },
    {
      "epoch": 1.7412693313753385,
      "grad_norm": 0.4879123866558075,
      "learning_rate": 3.880839798235242e-06,
      "loss": 0.4046,
      "step": 37944
    },
    {
      "epoch": 1.741315221880593,
      "grad_norm": 0.4374123811721802,
      "learning_rate": 3.880600832646929e-06,
      "loss": 0.3049,
      "step": 37945
    },
    {
      "epoch": 1.7413611123858472,
      "grad_norm": 0.45772024989128113,
      "learning_rate": 3.880361869750419e-06,
      "loss": 0.3988,
      "step": 37946
    },
    {
      "epoch": 1.7414070028911017,
      "grad_norm": 0.40565404295921326,
      "learning_rate": 3.880122909546284e-06,
      "loss": 0.2447,
      "step": 37947
    },
    {
      "epoch": 1.7414528933963562,
      "grad_norm": 0.47302478551864624,
      "learning_rate": 3.879883952035102e-06,
      "loss": 0.3428,
      "step": 37948
    },
    {
      "epoch": 1.7414987839016107,
      "grad_norm": 0.4119592607021332,
      "learning_rate": 3.879644997217444e-06,
      "loss": 0.2555,
      "step": 37949
    },
    {
      "epoch": 1.7415446744068652,
      "grad_norm": 0.4624257981777191,
      "learning_rate": 3.879406045093885e-06,
      "loss": 0.3596,
      "step": 37950
    },
    {
      "epoch": 1.7415905649121197,
      "grad_norm": 0.4962199926376343,
      "learning_rate": 3.879167095665002e-06,
      "loss": 0.3725,
      "step": 37951
    },
    {
      "epoch": 1.7416364554173742,
      "grad_norm": 0.47904813289642334,
      "learning_rate": 3.878928148931369e-06,
      "loss": 0.3225,
      "step": 37952
    },
    {
      "epoch": 1.7416823459226287,
      "grad_norm": 0.5103809833526611,
      "learning_rate": 3.8786892048935605e-06,
      "loss": 0.4498,
      "step": 37953
    },
    {
      "epoch": 1.7417282364278832,
      "grad_norm": 0.4917394518852234,
      "learning_rate": 3.8784502635521485e-06,
      "loss": 0.4135,
      "step": 37954
    },
    {
      "epoch": 1.7417741269331375,
      "grad_norm": 0.462625116109848,
      "learning_rate": 3.878211324907711e-06,
      "loss": 0.3163,
      "step": 37955
    },
    {
      "epoch": 1.741820017438392,
      "grad_norm": 0.44750088453292847,
      "learning_rate": 3.877972388960822e-06,
      "loss": 0.3124,
      "step": 37956
    },
    {
      "epoch": 1.7418659079436465,
      "grad_norm": 0.4470539093017578,
      "learning_rate": 3.877733455712052e-06,
      "loss": 0.3197,
      "step": 37957
    },
    {
      "epoch": 1.7419117984489008,
      "grad_norm": 0.45596978068351746,
      "learning_rate": 3.877494525161982e-06,
      "loss": 0.3208,
      "step": 37958
    },
    {
      "epoch": 1.7419576889541553,
      "grad_norm": 0.48154351115226746,
      "learning_rate": 3.877255597311183e-06,
      "loss": 0.3892,
      "step": 37959
    },
    {
      "epoch": 1.7420035794594098,
      "grad_norm": 0.4488265812397003,
      "learning_rate": 3.877016672160229e-06,
      "loss": 0.3268,
      "step": 37960
    },
    {
      "epoch": 1.7420494699646643,
      "grad_norm": 0.4380034804344177,
      "learning_rate": 3.876777749709696e-06,
      "loss": 0.3005,
      "step": 37961
    },
    {
      "epoch": 1.7420953604699188,
      "grad_norm": 0.45686468482017517,
      "learning_rate": 3.876538829960158e-06,
      "loss": 0.3474,
      "step": 37962
    },
    {
      "epoch": 1.7421412509751733,
      "grad_norm": 0.47951850295066833,
      "learning_rate": 3.876299912912188e-06,
      "loss": 0.4007,
      "step": 37963
    },
    {
      "epoch": 1.7421871414804277,
      "grad_norm": 0.43697425723075867,
      "learning_rate": 3.876060998566364e-06,
      "loss": 0.2841,
      "step": 37964
    },
    {
      "epoch": 1.7422330319856822,
      "grad_norm": 0.4656594693660736,
      "learning_rate": 3.875822086923258e-06,
      "loss": 0.3883,
      "step": 37965
    },
    {
      "epoch": 1.7422789224909367,
      "grad_norm": 0.4658859372138977,
      "learning_rate": 3.875583177983444e-06,
      "loss": 0.387,
      "step": 37966
    },
    {
      "epoch": 1.7423248129961912,
      "grad_norm": 0.4475056529045105,
      "learning_rate": 3.8753442717475e-06,
      "loss": 0.2877,
      "step": 37967
    },
    {
      "epoch": 1.7423707035014455,
      "grad_norm": 0.4683990776538849,
      "learning_rate": 3.875105368215997e-06,
      "loss": 0.3246,
      "step": 37968
    },
    {
      "epoch": 1.7424165940067,
      "grad_norm": 0.45770785212516785,
      "learning_rate": 3.87486646738951e-06,
      "loss": 0.3271,
      "step": 37969
    },
    {
      "epoch": 1.7424624845119545,
      "grad_norm": 0.49078628420829773,
      "learning_rate": 3.874627569268615e-06,
      "loss": 0.3655,
      "step": 37970
    },
    {
      "epoch": 1.7425083750172088,
      "grad_norm": 0.4394998848438263,
      "learning_rate": 3.874388673853885e-06,
      "loss": 0.3003,
      "step": 37971
    },
    {
      "epoch": 1.7425542655224633,
      "grad_norm": 0.5385946035385132,
      "learning_rate": 3.874149781145894e-06,
      "loss": 0.2971,
      "step": 37972
    },
    {
      "epoch": 1.7426001560277178,
      "grad_norm": 0.46914124488830566,
      "learning_rate": 3.873910891145218e-06,
      "loss": 0.3531,
      "step": 37973
    },
    {
      "epoch": 1.7426460465329723,
      "grad_norm": 0.4598560333251953,
      "learning_rate": 3.873672003852432e-06,
      "loss": 0.3665,
      "step": 37974
    },
    {
      "epoch": 1.7426919370382268,
      "grad_norm": 0.4359232783317566,
      "learning_rate": 3.8734331192681075e-06,
      "loss": 0.2751,
      "step": 37975
    },
    {
      "epoch": 1.7427378275434813,
      "grad_norm": 0.46403488516807556,
      "learning_rate": 3.873194237392823e-06,
      "loss": 0.371,
      "step": 37976
    },
    {
      "epoch": 1.7427837180487358,
      "grad_norm": 0.4754049777984619,
      "learning_rate": 3.87295535822715e-06,
      "loss": 0.2499,
      "step": 37977
    },
    {
      "epoch": 1.7428296085539903,
      "grad_norm": 0.5008671879768372,
      "learning_rate": 3.872716481771661e-06,
      "loss": 0.4276,
      "step": 37978
    },
    {
      "epoch": 1.7428754990592448,
      "grad_norm": 0.5081888437271118,
      "learning_rate": 3.872477608026936e-06,
      "loss": 0.4419,
      "step": 37979
    },
    {
      "epoch": 1.7429213895644993,
      "grad_norm": 0.4507344663143158,
      "learning_rate": 3.872238736993547e-06,
      "loss": 0.3403,
      "step": 37980
    },
    {
      "epoch": 1.7429672800697535,
      "grad_norm": 0.5313845872879028,
      "learning_rate": 3.871999868672066e-06,
      "loss": 0.4081,
      "step": 37981
    },
    {
      "epoch": 1.743013170575008,
      "grad_norm": 0.4921160638332367,
      "learning_rate": 3.87176100306307e-06,
      "loss": 0.4115,
      "step": 37982
    },
    {
      "epoch": 1.7430590610802625,
      "grad_norm": 0.481585830450058,
      "learning_rate": 3.8715221401671335e-06,
      "loss": 0.3867,
      "step": 37983
    },
    {
      "epoch": 1.7431049515855168,
      "grad_norm": 0.4809044301509857,
      "learning_rate": 3.871283279984828e-06,
      "loss": 0.4208,
      "step": 37984
    },
    {
      "epoch": 1.7431508420907713,
      "grad_norm": 0.5015039443969727,
      "learning_rate": 3.871044422516731e-06,
      "loss": 0.3829,
      "step": 37985
    },
    {
      "epoch": 1.7431967325960258,
      "grad_norm": 0.47089001536369324,
      "learning_rate": 3.870805567763417e-06,
      "loss": 0.3234,
      "step": 37986
    },
    {
      "epoch": 1.7432426231012803,
      "grad_norm": 0.4653850793838501,
      "learning_rate": 3.870566715725459e-06,
      "loss": 0.3586,
      "step": 37987
    },
    {
      "epoch": 1.7432885136065348,
      "grad_norm": 0.45627743005752563,
      "learning_rate": 3.870327866403431e-06,
      "loss": 0.3234,
      "step": 37988
    },
    {
      "epoch": 1.7433344041117893,
      "grad_norm": 0.4074941575527191,
      "learning_rate": 3.870089019797907e-06,
      "loss": 0.2681,
      "step": 37989
    },
    {
      "epoch": 1.7433802946170438,
      "grad_norm": 0.48102593421936035,
      "learning_rate": 3.8698501759094646e-06,
      "loss": 0.3678,
      "step": 37990
    },
    {
      "epoch": 1.7434261851222983,
      "grad_norm": 0.47702229022979736,
      "learning_rate": 3.869611334738673e-06,
      "loss": 0.3366,
      "step": 37991
    },
    {
      "epoch": 1.7434720756275528,
      "grad_norm": 0.5347232222557068,
      "learning_rate": 3.8693724962861116e-06,
      "loss": 0.3559,
      "step": 37992
    },
    {
      "epoch": 1.743517966132807,
      "grad_norm": 0.5036343932151794,
      "learning_rate": 3.869133660552352e-06,
      "loss": 0.3582,
      "step": 37993
    },
    {
      "epoch": 1.7435638566380616,
      "grad_norm": 0.4728985130786896,
      "learning_rate": 3.868894827537968e-06,
      "loss": 0.3347,
      "step": 37994
    },
    {
      "epoch": 1.743609747143316,
      "grad_norm": 0.4648227393627167,
      "learning_rate": 3.868655997243536e-06,
      "loss": 0.3324,
      "step": 37995
    },
    {
      "epoch": 1.7436556376485703,
      "grad_norm": 0.46075934171676636,
      "learning_rate": 3.8684171696696295e-06,
      "loss": 0.3398,
      "step": 37996
    },
    {
      "epoch": 1.7437015281538248,
      "grad_norm": 0.4379268288612366,
      "learning_rate": 3.868178344816822e-06,
      "loss": 0.3261,
      "step": 37997
    },
    {
      "epoch": 1.7437474186590793,
      "grad_norm": 0.4336170256137848,
      "learning_rate": 3.867939522685688e-06,
      "loss": 0.2966,
      "step": 37998
    },
    {
      "epoch": 1.7437933091643338,
      "grad_norm": 0.4741629958152771,
      "learning_rate": 3.8677007032768035e-06,
      "loss": 0.3554,
      "step": 37999
    },
    {
      "epoch": 1.7438391996695883,
      "grad_norm": 0.45813441276550293,
      "learning_rate": 3.867461886590738e-06,
      "loss": 0.3228,
      "step": 38000
    },
    {
      "epoch": 1.7438850901748428,
      "grad_norm": 0.49603018164634705,
      "learning_rate": 3.867223072628073e-06,
      "loss": 0.3591,
      "step": 38001
    },
    {
      "epoch": 1.7439309806800973,
      "grad_norm": 0.49939361214637756,
      "learning_rate": 3.8669842613893785e-06,
      "loss": 0.3934,
      "step": 38002
    },
    {
      "epoch": 1.7439768711853518,
      "grad_norm": 0.47434869408607483,
      "learning_rate": 3.866745452875228e-06,
      "loss": 0.2877,
      "step": 38003
    },
    {
      "epoch": 1.7440227616906063,
      "grad_norm": 0.4390914738178253,
      "learning_rate": 3.866506647086198e-06,
      "loss": 0.3025,
      "step": 38004
    },
    {
      "epoch": 1.7440686521958608,
      "grad_norm": 0.47859013080596924,
      "learning_rate": 3.866267844022862e-06,
      "loss": 0.4251,
      "step": 38005
    },
    {
      "epoch": 1.744114542701115,
      "grad_norm": 0.4661463499069214,
      "learning_rate": 3.866029043685791e-06,
      "loss": 0.3569,
      "step": 38006
    },
    {
      "epoch": 1.7441604332063696,
      "grad_norm": 0.4449816346168518,
      "learning_rate": 3.865790246075565e-06,
      "loss": 0.3233,
      "step": 38007
    },
    {
      "epoch": 1.744206323711624,
      "grad_norm": 0.46287795901298523,
      "learning_rate": 3.865551451192756e-06,
      "loss": 0.3712,
      "step": 38008
    },
    {
      "epoch": 1.7442522142168784,
      "grad_norm": 0.4136410057544708,
      "learning_rate": 3.8653126590379364e-06,
      "loss": 0.2648,
      "step": 38009
    },
    {
      "epoch": 1.7442981047221329,
      "grad_norm": 0.488627552986145,
      "learning_rate": 3.865073869611683e-06,
      "loss": 0.4028,
      "step": 38010
    },
    {
      "epoch": 1.7443439952273874,
      "grad_norm": 0.45532307028770447,
      "learning_rate": 3.8648350829145676e-06,
      "loss": 0.3242,
      "step": 38011
    },
    {
      "epoch": 1.7443898857326419,
      "grad_norm": 0.46150466799736023,
      "learning_rate": 3.864596298947164e-06,
      "loss": 0.3391,
      "step": 38012
    },
    {
      "epoch": 1.7444357762378964,
      "grad_norm": 0.4998036026954651,
      "learning_rate": 3.864357517710051e-06,
      "loss": 0.3808,
      "step": 38013
    },
    {
      "epoch": 1.7444816667431509,
      "grad_norm": 0.4728691875934601,
      "learning_rate": 3.8641187392038e-06,
      "loss": 0.3578,
      "step": 38014
    },
    {
      "epoch": 1.7445275572484054,
      "grad_norm": 0.45582035183906555,
      "learning_rate": 3.863879963428981e-06,
      "loss": 0.3239,
      "step": 38015
    },
    {
      "epoch": 1.7445734477536599,
      "grad_norm": 0.44844090938568115,
      "learning_rate": 3.863641190386176e-06,
      "loss": 0.332,
      "step": 38016
    },
    {
      "epoch": 1.7446193382589144,
      "grad_norm": 0.5106685161590576,
      "learning_rate": 3.863402420075954e-06,
      "loss": 0.4414,
      "step": 38017
    },
    {
      "epoch": 1.7446652287641689,
      "grad_norm": 0.4613772928714752,
      "learning_rate": 3.863163652498891e-06,
      "loss": 0.3741,
      "step": 38018
    },
    {
      "epoch": 1.7447111192694231,
      "grad_norm": 0.46264442801475525,
      "learning_rate": 3.8629248876555595e-06,
      "loss": 0.3397,
      "step": 38019
    },
    {
      "epoch": 1.7447570097746776,
      "grad_norm": 0.48016098141670227,
      "learning_rate": 3.862686125546536e-06,
      "loss": 0.3738,
      "step": 38020
    },
    {
      "epoch": 1.7448029002799321,
      "grad_norm": 0.4793284237384796,
      "learning_rate": 3.862447366172393e-06,
      "loss": 0.3708,
      "step": 38021
    },
    {
      "epoch": 1.7448487907851864,
      "grad_norm": 0.4508094787597656,
      "learning_rate": 3.862208609533703e-06,
      "loss": 0.3317,
      "step": 38022
    },
    {
      "epoch": 1.744894681290441,
      "grad_norm": 0.4547170400619507,
      "learning_rate": 3.861969855631045e-06,
      "loss": 0.335,
      "step": 38023
    },
    {
      "epoch": 1.7449405717956954,
      "grad_norm": 0.4202292263507843,
      "learning_rate": 3.861731104464989e-06,
      "loss": 0.2755,
      "step": 38024
    },
    {
      "epoch": 1.74498646230095,
      "grad_norm": 0.46441975235939026,
      "learning_rate": 3.861492356036111e-06,
      "loss": 0.3803,
      "step": 38025
    },
    {
      "epoch": 1.7450323528062044,
      "grad_norm": 0.44824299216270447,
      "learning_rate": 3.861253610344984e-06,
      "loss": 0.3549,
      "step": 38026
    },
    {
      "epoch": 1.7450782433114589,
      "grad_norm": 0.4311133325099945,
      "learning_rate": 3.861014867392183e-06,
      "loss": 0.2776,
      "step": 38027
    },
    {
      "epoch": 1.7451241338167134,
      "grad_norm": 0.49576759338378906,
      "learning_rate": 3.860776127178281e-06,
      "loss": 0.4285,
      "step": 38028
    },
    {
      "epoch": 1.7451700243219679,
      "grad_norm": 0.4565376341342926,
      "learning_rate": 3.860537389703854e-06,
      "loss": 0.3213,
      "step": 38029
    },
    {
      "epoch": 1.7452159148272224,
      "grad_norm": 0.482695072889328,
      "learning_rate": 3.860298654969475e-06,
      "loss": 0.334,
      "step": 38030
    },
    {
      "epoch": 1.7452618053324767,
      "grad_norm": 0.4559727609157562,
      "learning_rate": 3.8600599229757165e-06,
      "loss": 0.3173,
      "step": 38031
    },
    {
      "epoch": 1.7453076958377312,
      "grad_norm": 0.41081032156944275,
      "learning_rate": 3.859821193723155e-06,
      "loss": 0.2458,
      "step": 38032
    },
    {
      "epoch": 1.7453535863429857,
      "grad_norm": 0.4661165475845337,
      "learning_rate": 3.859582467212365e-06,
      "loss": 0.2931,
      "step": 38033
    },
    {
      "epoch": 1.7453994768482402,
      "grad_norm": 0.4763866364955902,
      "learning_rate": 3.859343743443916e-06,
      "loss": 0.3963,
      "step": 38034
    },
    {
      "epoch": 1.7454453673534944,
      "grad_norm": 0.4529687762260437,
      "learning_rate": 3.859105022418389e-06,
      "loss": 0.3032,
      "step": 38035
    },
    {
      "epoch": 1.745491257858749,
      "grad_norm": 0.48756447434425354,
      "learning_rate": 3.858866304136354e-06,
      "loss": 0.4326,
      "step": 38036
    },
    {
      "epoch": 1.7455371483640034,
      "grad_norm": 0.4894818365573883,
      "learning_rate": 3.858627588598382e-06,
      "loss": 0.4243,
      "step": 38037
    },
    {
      "epoch": 1.745583038869258,
      "grad_norm": 0.5077701210975647,
      "learning_rate": 3.858388875805054e-06,
      "loss": 0.4161,
      "step": 38038
    },
    {
      "epoch": 1.7456289293745124,
      "grad_norm": 0.4977981448173523,
      "learning_rate": 3.8581501657569384e-06,
      "loss": 0.3705,
      "step": 38039
    },
    {
      "epoch": 1.745674819879767,
      "grad_norm": 0.4803827702999115,
      "learning_rate": 3.857911458454612e-06,
      "loss": 0.3498,
      "step": 38040
    },
    {
      "epoch": 1.7457207103850214,
      "grad_norm": 0.4764993488788605,
      "learning_rate": 3.857672753898648e-06,
      "loss": 0.3306,
      "step": 38041
    },
    {
      "epoch": 1.745766600890276,
      "grad_norm": 0.45429080724716187,
      "learning_rate": 3.857434052089621e-06,
      "loss": 0.3558,
      "step": 38042
    },
    {
      "epoch": 1.7458124913955304,
      "grad_norm": 0.5428577065467834,
      "learning_rate": 3.8571953530281025e-06,
      "loss": 0.44,
      "step": 38043
    },
    {
      "epoch": 1.7458583819007847,
      "grad_norm": 0.4856216311454773,
      "learning_rate": 3.8569566567146705e-06,
      "loss": 0.3831,
      "step": 38044
    },
    {
      "epoch": 1.7459042724060392,
      "grad_norm": 0.6090908646583557,
      "learning_rate": 3.856717963149897e-06,
      "loss": 0.3348,
      "step": 38045
    },
    {
      "epoch": 1.7459501629112937,
      "grad_norm": 0.47545376420021057,
      "learning_rate": 3.856479272334355e-06,
      "loss": 0.3393,
      "step": 38046
    },
    {
      "epoch": 1.745996053416548,
      "grad_norm": 0.45884227752685547,
      "learning_rate": 3.85624058426862e-06,
      "loss": 0.3537,
      "step": 38047
    },
    {
      "epoch": 1.7460419439218025,
      "grad_norm": 0.5147340893745422,
      "learning_rate": 3.856001898953265e-06,
      "loss": 0.3987,
      "step": 38048
    },
    {
      "epoch": 1.746087834427057,
      "grad_norm": 0.49896541237831116,
      "learning_rate": 3.8557632163888635e-06,
      "loss": 0.3771,
      "step": 38049
    },
    {
      "epoch": 1.7461337249323114,
      "grad_norm": 0.4938095211982727,
      "learning_rate": 3.8555245365759915e-06,
      "loss": 0.3527,
      "step": 38050
    },
    {
      "epoch": 1.746179615437566,
      "grad_norm": 0.4150000512599945,
      "learning_rate": 3.855285859515223e-06,
      "loss": 0.2832,
      "step": 38051
    },
    {
      "epoch": 1.7462255059428204,
      "grad_norm": 0.48580825328826904,
      "learning_rate": 3.85504718520713e-06,
      "loss": 0.372,
      "step": 38052
    },
    {
      "epoch": 1.746271396448075,
      "grad_norm": 0.5207895040512085,
      "learning_rate": 3.854808513652286e-06,
      "loss": 0.3678,
      "step": 38053
    },
    {
      "epoch": 1.7463172869533294,
      "grad_norm": 0.44285261631011963,
      "learning_rate": 3.854569844851267e-06,
      "loss": 0.3262,
      "step": 38054
    },
    {
      "epoch": 1.746363177458584,
      "grad_norm": 0.479082316160202,
      "learning_rate": 3.8543311788046465e-06,
      "loss": 0.3351,
      "step": 38055
    },
    {
      "epoch": 1.7464090679638384,
      "grad_norm": 0.49444156885147095,
      "learning_rate": 3.8540925155129955e-06,
      "loss": 0.3765,
      "step": 38056
    },
    {
      "epoch": 1.7464549584690927,
      "grad_norm": 0.4872749447822571,
      "learning_rate": 3.8538538549768934e-06,
      "loss": 0.3967,
      "step": 38057
    },
    {
      "epoch": 1.7465008489743472,
      "grad_norm": 0.45521658658981323,
      "learning_rate": 3.85361519719691e-06,
      "loss": 0.3224,
      "step": 38058
    },
    {
      "epoch": 1.7465467394796017,
      "grad_norm": 0.4507570266723633,
      "learning_rate": 3.853376542173619e-06,
      "loss": 0.3335,
      "step": 38059
    },
    {
      "epoch": 1.746592629984856,
      "grad_norm": 0.4997755289077759,
      "learning_rate": 3.853137889907597e-06,
      "loss": 0.3853,
      "step": 38060
    },
    {
      "epoch": 1.7466385204901105,
      "grad_norm": 0.5147897601127625,
      "learning_rate": 3.852899240399416e-06,
      "loss": 0.3902,
      "step": 38061
    },
    {
      "epoch": 1.746684410995365,
      "grad_norm": 0.46618714928627014,
      "learning_rate": 3.852660593649649e-06,
      "loss": 0.3459,
      "step": 38062
    },
    {
      "epoch": 1.7467303015006195,
      "grad_norm": 0.47780168056488037,
      "learning_rate": 3.8524219496588735e-06,
      "loss": 0.3754,
      "step": 38063
    },
    {
      "epoch": 1.746776192005874,
      "grad_norm": 0.46038898825645447,
      "learning_rate": 3.85218330842766e-06,
      "loss": 0.3218,
      "step": 38064
    },
    {
      "epoch": 1.7468220825111285,
      "grad_norm": 0.4934273660182953,
      "learning_rate": 3.851944669956582e-06,
      "loss": 0.387,
      "step": 38065
    },
    {
      "epoch": 1.746867973016383,
      "grad_norm": 0.4990277588367462,
      "learning_rate": 3.851706034246216e-06,
      "loss": 0.3233,
      "step": 38066
    },
    {
      "epoch": 1.7469138635216375,
      "grad_norm": 0.44005143642425537,
      "learning_rate": 3.851467401297135e-06,
      "loss": 0.2958,
      "step": 38067
    },
    {
      "epoch": 1.746959754026892,
      "grad_norm": 0.4454624652862549,
      "learning_rate": 3.851228771109911e-06,
      "loss": 0.2985,
      "step": 38068
    },
    {
      "epoch": 1.7470056445321465,
      "grad_norm": 0.44387415051460266,
      "learning_rate": 3.850990143685121e-06,
      "loss": 0.3138,
      "step": 38069
    },
    {
      "epoch": 1.7470515350374007,
      "grad_norm": 0.4584024250507355,
      "learning_rate": 3.850751519023337e-06,
      "loss": 0.3072,
      "step": 38070
    },
    {
      "epoch": 1.7470974255426552,
      "grad_norm": 0.468189537525177,
      "learning_rate": 3.85051289712513e-06,
      "loss": 0.3715,
      "step": 38071
    },
    {
      "epoch": 1.7471433160479097,
      "grad_norm": 0.5106051564216614,
      "learning_rate": 3.850274277991079e-06,
      "loss": 0.4245,
      "step": 38072
    },
    {
      "epoch": 1.747189206553164,
      "grad_norm": 0.4444620609283447,
      "learning_rate": 3.850035661621756e-06,
      "loss": 0.3358,
      "step": 38073
    },
    {
      "epoch": 1.7472350970584185,
      "grad_norm": 0.48902514576911926,
      "learning_rate": 3.849797048017733e-06,
      "loss": 0.369,
      "step": 38074
    },
    {
      "epoch": 1.747280987563673,
      "grad_norm": 0.47217416763305664,
      "learning_rate": 3.8495584371795865e-06,
      "loss": 0.3359,
      "step": 38075
    },
    {
      "epoch": 1.7473268780689275,
      "grad_norm": 0.4627039134502411,
      "learning_rate": 3.849319829107888e-06,
      "loss": 0.3189,
      "step": 38076
    },
    {
      "epoch": 1.747372768574182,
      "grad_norm": 0.48779672384262085,
      "learning_rate": 3.849081223803211e-06,
      "loss": 0.3575,
      "step": 38077
    },
    {
      "epoch": 1.7474186590794365,
      "grad_norm": 0.4559279978275299,
      "learning_rate": 3.8488426212661325e-06,
      "loss": 0.3155,
      "step": 38078
    },
    {
      "epoch": 1.747464549584691,
      "grad_norm": 0.46457433700561523,
      "learning_rate": 3.848604021497224e-06,
      "loss": 0.3368,
      "step": 38079
    },
    {
      "epoch": 1.7475104400899455,
      "grad_norm": 0.49840980768203735,
      "learning_rate": 3.848365424497058e-06,
      "loss": 0.3647,
      "step": 38080
    },
    {
      "epoch": 1.7475563305952,
      "grad_norm": 0.4659261703491211,
      "learning_rate": 3.84812683026621e-06,
      "loss": 0.3632,
      "step": 38081
    },
    {
      "epoch": 1.7476022211004543,
      "grad_norm": 0.454520046710968,
      "learning_rate": 3.8478882388052555e-06,
      "loss": 0.326,
      "step": 38082
    },
    {
      "epoch": 1.7476481116057088,
      "grad_norm": 0.4582139849662781,
      "learning_rate": 3.847649650114764e-06,
      "loss": 0.385,
      "step": 38083
    },
    {
      "epoch": 1.7476940021109633,
      "grad_norm": 0.4651276469230652,
      "learning_rate": 3.8474110641953115e-06,
      "loss": 0.3444,
      "step": 38084
    },
    {
      "epoch": 1.7477398926162175,
      "grad_norm": 0.4231316149234772,
      "learning_rate": 3.8471724810474734e-06,
      "loss": 0.2776,
      "step": 38085
    },
    {
      "epoch": 1.747785783121472,
      "grad_norm": 0.5295804738998413,
      "learning_rate": 3.84693390067182e-06,
      "loss": 0.4412,
      "step": 38086
    },
    {
      "epoch": 1.7478316736267265,
      "grad_norm": 0.4610322415828705,
      "learning_rate": 3.846695323068926e-06,
      "loss": 0.3405,
      "step": 38087
    },
    {
      "epoch": 1.747877564131981,
      "grad_norm": 0.4613892138004303,
      "learning_rate": 3.846456748239367e-06,
      "loss": 0.3285,
      "step": 38088
    },
    {
      "epoch": 1.7479234546372355,
      "grad_norm": 0.4432855546474457,
      "learning_rate": 3.846218176183715e-06,
      "loss": 0.3348,
      "step": 38089
    },
    {
      "epoch": 1.74796934514249,
      "grad_norm": 0.49812015891075134,
      "learning_rate": 3.845979606902544e-06,
      "loss": 0.3864,
      "step": 38090
    },
    {
      "epoch": 1.7480152356477445,
      "grad_norm": 0.5012561678886414,
      "learning_rate": 3.8457410403964285e-06,
      "loss": 0.3881,
      "step": 38091
    },
    {
      "epoch": 1.748061126152999,
      "grad_norm": 0.48442694544792175,
      "learning_rate": 3.8455024766659415e-06,
      "loss": 0.3635,
      "step": 38092
    },
    {
      "epoch": 1.7481070166582535,
      "grad_norm": 0.5005574822425842,
      "learning_rate": 3.845263915711655e-06,
      "loss": 0.4273,
      "step": 38093
    },
    {
      "epoch": 1.748152907163508,
      "grad_norm": 0.46281829476356506,
      "learning_rate": 3.845025357534147e-06,
      "loss": 0.3189,
      "step": 38094
    },
    {
      "epoch": 1.7481987976687623,
      "grad_norm": 0.4967372417449951,
      "learning_rate": 3.8447868021339864e-06,
      "loss": 0.4053,
      "step": 38095
    },
    {
      "epoch": 1.7482446881740168,
      "grad_norm": 0.4903258979320526,
      "learning_rate": 3.8445482495117496e-06,
      "loss": 0.3645,
      "step": 38096
    },
    {
      "epoch": 1.7482905786792713,
      "grad_norm": 0.4525153934955597,
      "learning_rate": 3.8443096996680095e-06,
      "loss": 0.3657,
      "step": 38097
    },
    {
      "epoch": 1.7483364691845256,
      "grad_norm": 0.46167466044425964,
      "learning_rate": 3.8440711526033404e-06,
      "loss": 0.3123,
      "step": 38098
    },
    {
      "epoch": 1.74838235968978,
      "grad_norm": 0.47946059703826904,
      "learning_rate": 3.843832608318313e-06,
      "loss": 0.3762,
      "step": 38099
    },
    {
      "epoch": 1.7484282501950346,
      "grad_norm": 0.481432169675827,
      "learning_rate": 3.843594066813507e-06,
      "loss": 0.4092,
      "step": 38100
    },
    {
      "epoch": 1.748474140700289,
      "grad_norm": 0.49810469150543213,
      "learning_rate": 3.8433555280894915e-06,
      "loss": 0.3522,
      "step": 38101
    },
    {
      "epoch": 1.7485200312055436,
      "grad_norm": 0.46181628108024597,
      "learning_rate": 3.843116992146838e-06,
      "loss": 0.3354,
      "step": 38102
    },
    {
      "epoch": 1.748565921710798,
      "grad_norm": 0.4687594771385193,
      "learning_rate": 3.842878458986126e-06,
      "loss": 0.3236,
      "step": 38103
    },
    {
      "epoch": 1.7486118122160526,
      "grad_norm": 0.5257999897003174,
      "learning_rate": 3.842639928607925e-06,
      "loss": 0.345,
      "step": 38104
    },
    {
      "epoch": 1.748657702721307,
      "grad_norm": 0.4715394079685211,
      "learning_rate": 3.84240140101281e-06,
      "loss": 0.3105,
      "step": 38105
    },
    {
      "epoch": 1.7487035932265615,
      "grad_norm": 0.4646730422973633,
      "learning_rate": 3.842162876201354e-06,
      "loss": 0.3191,
      "step": 38106
    },
    {
      "epoch": 1.748749483731816,
      "grad_norm": 0.4432922303676605,
      "learning_rate": 3.841924354174131e-06,
      "loss": 0.278,
      "step": 38107
    },
    {
      "epoch": 1.7487953742370703,
      "grad_norm": 0.5247595906257629,
      "learning_rate": 3.8416858349317134e-06,
      "loss": 0.4545,
      "step": 38108
    },
    {
      "epoch": 1.7488412647423248,
      "grad_norm": 0.4653107821941376,
      "learning_rate": 3.841447318474677e-06,
      "loss": 0.3733,
      "step": 38109
    },
    {
      "epoch": 1.7488871552475793,
      "grad_norm": 0.454339861869812,
      "learning_rate": 3.8412088048035945e-06,
      "loss": 0.3437,
      "step": 38110
    },
    {
      "epoch": 1.7489330457528336,
      "grad_norm": 0.45869478583335876,
      "learning_rate": 3.840970293919037e-06,
      "loss": 0.3529,
      "step": 38111
    },
    {
      "epoch": 1.748978936258088,
      "grad_norm": 0.4230233430862427,
      "learning_rate": 3.840731785821583e-06,
      "loss": 0.2824,
      "step": 38112
    },
    {
      "epoch": 1.7490248267633426,
      "grad_norm": 0.49271175265312195,
      "learning_rate": 3.840493280511802e-06,
      "loss": 0.3704,
      "step": 38113
    },
    {
      "epoch": 1.749070717268597,
      "grad_norm": 0.466296911239624,
      "learning_rate": 3.840254777990266e-06,
      "loss": 0.3688,
      "step": 38114
    },
    {
      "epoch": 1.7491166077738516,
      "grad_norm": 0.5173819661140442,
      "learning_rate": 3.840016278257555e-06,
      "loss": 0.4086,
      "step": 38115
    },
    {
      "epoch": 1.749162498279106,
      "grad_norm": 0.4452902674674988,
      "learning_rate": 3.839777781314238e-06,
      "loss": 0.3332,
      "step": 38116
    },
    {
      "epoch": 1.7492083887843606,
      "grad_norm": 0.4867652356624603,
      "learning_rate": 3.83953928716089e-06,
      "loss": 0.375,
      "step": 38117
    },
    {
      "epoch": 1.749254279289615,
      "grad_norm": 0.4646952450275421,
      "learning_rate": 3.839300795798081e-06,
      "loss": 0.3542,
      "step": 38118
    },
    {
      "epoch": 1.7493001697948696,
      "grad_norm": 0.49498888850212097,
      "learning_rate": 3.839062307226389e-06,
      "loss": 0.4197,
      "step": 38119
    },
    {
      "epoch": 1.7493460603001239,
      "grad_norm": 0.4770100712776184,
      "learning_rate": 3.838823821446387e-06,
      "loss": 0.3374,
      "step": 38120
    },
    {
      "epoch": 1.7493919508053783,
      "grad_norm": 0.4847092032432556,
      "learning_rate": 3.838585338458644e-06,
      "loss": 0.3486,
      "step": 38121
    },
    {
      "epoch": 1.7494378413106328,
      "grad_norm": 0.5174691081047058,
      "learning_rate": 3.838346858263739e-06,
      "loss": 0.3694,
      "step": 38122
    },
    {
      "epoch": 1.7494837318158873,
      "grad_norm": 0.4751739799976349,
      "learning_rate": 3.838108380862243e-06,
      "loss": 0.3836,
      "step": 38123
    },
    {
      "epoch": 1.7495296223211416,
      "grad_norm": 0.5592171549797058,
      "learning_rate": 3.837869906254728e-06,
      "loss": 0.3699,
      "step": 38124
    },
    {
      "epoch": 1.7495755128263961,
      "grad_norm": 0.44661763310432434,
      "learning_rate": 3.837631434441771e-06,
      "loss": 0.3082,
      "step": 38125
    },
    {
      "epoch": 1.7496214033316506,
      "grad_norm": 0.4721638560295105,
      "learning_rate": 3.837392965423943e-06,
      "loss": 0.3444,
      "step": 38126
    },
    {
      "epoch": 1.7496672938369051,
      "grad_norm": 0.46432891488075256,
      "learning_rate": 3.837154499201818e-06,
      "loss": 0.3388,
      "step": 38127
    },
    {
      "epoch": 1.7497131843421596,
      "grad_norm": 0.4756113886833191,
      "learning_rate": 3.83691603577597e-06,
      "loss": 0.3784,
      "step": 38128
    },
    {
      "epoch": 1.749759074847414,
      "grad_norm": 0.46746310591697693,
      "learning_rate": 3.836677575146971e-06,
      "loss": 0.3493,
      "step": 38129
    },
    {
      "epoch": 1.7498049653526686,
      "grad_norm": 0.47583332657814026,
      "learning_rate": 3.836439117315393e-06,
      "loss": 0.3876,
      "step": 38130
    },
    {
      "epoch": 1.749850855857923,
      "grad_norm": 0.46496549248695374,
      "learning_rate": 3.836200662281815e-06,
      "loss": 0.3808,
      "step": 38131
    },
    {
      "epoch": 1.7498967463631776,
      "grad_norm": 0.4402986764907837,
      "learning_rate": 3.835962210046807e-06,
      "loss": 0.3103,
      "step": 38132
    },
    {
      "epoch": 1.7499426368684319,
      "grad_norm": 0.6384602189064026,
      "learning_rate": 3.835723760610941e-06,
      "loss": 0.3013,
      "step": 38133
    },
    {
      "epoch": 1.7499885273736864,
      "grad_norm": 0.4458513855934143,
      "learning_rate": 3.835485313974794e-06,
      "loss": 0.3268,
      "step": 38134
    },
    {
      "epoch": 1.7500344178789409,
      "grad_norm": 0.46175071597099304,
      "learning_rate": 3.835246870138936e-06,
      "loss": 0.3634,
      "step": 38135
    },
    {
      "epoch": 1.7500803083841951,
      "grad_norm": 0.4617447257041931,
      "learning_rate": 3.83500842910394e-06,
      "loss": 0.3779,
      "step": 38136
    },
    {
      "epoch": 1.7501261988894496,
      "grad_norm": 0.46795785427093506,
      "learning_rate": 3.834769990870383e-06,
      "loss": 0.4061,
      "step": 38137
    },
    {
      "epoch": 1.7501720893947041,
      "grad_norm": 0.45606985688209534,
      "learning_rate": 3.8345315554388364e-06,
      "loss": 0.318,
      "step": 38138
    },
    {
      "epoch": 1.7502179798999586,
      "grad_norm": 0.441455215215683,
      "learning_rate": 3.834293122809873e-06,
      "loss": 0.3284,
      "step": 38139
    },
    {
      "epoch": 1.7502638704052131,
      "grad_norm": 0.4359261989593506,
      "learning_rate": 3.8340546929840675e-06,
      "loss": 0.331,
      "step": 38140
    },
    {
      "epoch": 1.7503097609104676,
      "grad_norm": 0.49580463767051697,
      "learning_rate": 3.833816265961993e-06,
      "loss": 0.3246,
      "step": 38141
    },
    {
      "epoch": 1.7503556514157221,
      "grad_norm": 0.4948081374168396,
      "learning_rate": 3.833577841744219e-06,
      "loss": 0.3516,
      "step": 38142
    },
    {
      "epoch": 1.7504015419209766,
      "grad_norm": 0.4832538962364197,
      "learning_rate": 3.833339420331324e-06,
      "loss": 0.381,
      "step": 38143
    },
    {
      "epoch": 1.7504474324262311,
      "grad_norm": 0.4935844838619232,
      "learning_rate": 3.8331010017238815e-06,
      "loss": 0.3631,
      "step": 38144
    },
    {
      "epoch": 1.7504933229314856,
      "grad_norm": 0.4307432770729065,
      "learning_rate": 3.832862585922461e-06,
      "loss": 0.2874,
      "step": 38145
    },
    {
      "epoch": 1.75053921343674,
      "grad_norm": 0.4742552936077118,
      "learning_rate": 3.832624172927638e-06,
      "loss": 0.3332,
      "step": 38146
    },
    {
      "epoch": 1.7505851039419944,
      "grad_norm": 0.4865756630897522,
      "learning_rate": 3.8323857627399855e-06,
      "loss": 0.3702,
      "step": 38147
    },
    {
      "epoch": 1.750630994447249,
      "grad_norm": 0.4902043640613556,
      "learning_rate": 3.832147355360077e-06,
      "loss": 0.3228,
      "step": 38148
    },
    {
      "epoch": 1.7506768849525032,
      "grad_norm": 0.5285575985908508,
      "learning_rate": 3.8319089507884845e-06,
      "loss": 0.4122,
      "step": 38149
    },
    {
      "epoch": 1.7507227754577577,
      "grad_norm": 0.4738818109035492,
      "learning_rate": 3.8316705490257836e-06,
      "loss": 0.3088,
      "step": 38150
    },
    {
      "epoch": 1.7507686659630122,
      "grad_norm": 0.453727662563324,
      "learning_rate": 3.831432150072546e-06,
      "loss": 0.3107,
      "step": 38151
    },
    {
      "epoch": 1.7508145564682667,
      "grad_norm": 0.5346817374229431,
      "learning_rate": 3.831193753929343e-06,
      "loss": 0.2537,
      "step": 38152
    },
    {
      "epoch": 1.7508604469735212,
      "grad_norm": 0.4578160047531128,
      "learning_rate": 3.830955360596753e-06,
      "loss": 0.361,
      "step": 38153
    },
    {
      "epoch": 1.7509063374787757,
      "grad_norm": 0.45820382237434387,
      "learning_rate": 3.830716970075345e-06,
      "loss": 0.309,
      "step": 38154
    },
    {
      "epoch": 1.7509522279840302,
      "grad_norm": 0.4526442289352417,
      "learning_rate": 3.830478582365693e-06,
      "loss": 0.3002,
      "step": 38155
    },
    {
      "epoch": 1.7509981184892847,
      "grad_norm": 0.5117863416671753,
      "learning_rate": 3.830240197468373e-06,
      "loss": 0.4113,
      "step": 38156
    },
    {
      "epoch": 1.7510440089945392,
      "grad_norm": 0.4909150302410126,
      "learning_rate": 3.830001815383956e-06,
      "loss": 0.3647,
      "step": 38157
    },
    {
      "epoch": 1.7510898994997937,
      "grad_norm": 0.4600817859172821,
      "learning_rate": 3.829763436113012e-06,
      "loss": 0.3703,
      "step": 38158
    },
    {
      "epoch": 1.751135790005048,
      "grad_norm": 0.46346205472946167,
      "learning_rate": 3.829525059656121e-06,
      "loss": 0.347,
      "step": 38159
    },
    {
      "epoch": 1.7511816805103024,
      "grad_norm": 0.49247220158576965,
      "learning_rate": 3.8292866860138525e-06,
      "loss": 0.4084,
      "step": 38160
    },
    {
      "epoch": 1.751227571015557,
      "grad_norm": 0.5336139798164368,
      "learning_rate": 3.8290483151867785e-06,
      "loss": 0.4194,
      "step": 38161
    },
    {
      "epoch": 1.7512734615208112,
      "grad_norm": 0.45903024077415466,
      "learning_rate": 3.828809947175475e-06,
      "loss": 0.35,
      "step": 38162
    },
    {
      "epoch": 1.7513193520260657,
      "grad_norm": 0.503702700138092,
      "learning_rate": 3.828571581980514e-06,
      "loss": 0.3654,
      "step": 38163
    },
    {
      "epoch": 1.7513652425313202,
      "grad_norm": 0.509530782699585,
      "learning_rate": 3.828333219602466e-06,
      "loss": 0.365,
      "step": 38164
    },
    {
      "epoch": 1.7514111330365747,
      "grad_norm": 0.4682491421699524,
      "learning_rate": 3.82809486004191e-06,
      "loss": 0.3694,
      "step": 38165
    },
    {
      "epoch": 1.7514570235418292,
      "grad_norm": 0.4417170584201813,
      "learning_rate": 3.827856503299416e-06,
      "loss": 0.2936,
      "step": 38166
    },
    {
      "epoch": 1.7515029140470837,
      "grad_norm": 0.48972633481025696,
      "learning_rate": 3.827618149375555e-06,
      "loss": 0.3991,
      "step": 38167
    },
    {
      "epoch": 1.7515488045523382,
      "grad_norm": 0.4649178683757782,
      "learning_rate": 3.827379798270904e-06,
      "loss": 0.3636,
      "step": 38168
    },
    {
      "epoch": 1.7515946950575927,
      "grad_norm": 0.5182351469993591,
      "learning_rate": 3.8271414499860345e-06,
      "loss": 0.3988,
      "step": 38169
    },
    {
      "epoch": 1.7516405855628472,
      "grad_norm": 0.46883293986320496,
      "learning_rate": 3.826903104521517e-06,
      "loss": 0.3451,
      "step": 38170
    },
    {
      "epoch": 1.7516864760681015,
      "grad_norm": 0.46674907207489014,
      "learning_rate": 3.8266647618779315e-06,
      "loss": 0.3236,
      "step": 38171
    },
    {
      "epoch": 1.751732366573356,
      "grad_norm": 0.4476717412471771,
      "learning_rate": 3.826426422055845e-06,
      "loss": 0.3049,
      "step": 38172
    },
    {
      "epoch": 1.7517782570786105,
      "grad_norm": 0.47447434067726135,
      "learning_rate": 3.826188085055831e-06,
      "loss": 0.3508,
      "step": 38173
    },
    {
      "epoch": 1.7518241475838647,
      "grad_norm": 0.4956931471824646,
      "learning_rate": 3.825949750878466e-06,
      "loss": 0.4407,
      "step": 38174
    },
    {
      "epoch": 1.7518700380891192,
      "grad_norm": 0.4717135429382324,
      "learning_rate": 3.8257114195243215e-06,
      "loss": 0.3339,
      "step": 38175
    },
    {
      "epoch": 1.7519159285943737,
      "grad_norm": 0.5097389221191406,
      "learning_rate": 3.825473090993969e-06,
      "loss": 0.3977,
      "step": 38176
    },
    {
      "epoch": 1.7519618190996282,
      "grad_norm": 0.43335920572280884,
      "learning_rate": 3.8252347652879845e-06,
      "loss": 0.3137,
      "step": 38177
    },
    {
      "epoch": 1.7520077096048827,
      "grad_norm": 0.49420949816703796,
      "learning_rate": 3.8249964424069394e-06,
      "loss": 0.3953,
      "step": 38178
    },
    {
      "epoch": 1.7520536001101372,
      "grad_norm": 0.5198296308517456,
      "learning_rate": 3.824758122351405e-06,
      "loss": 0.4862,
      "step": 38179
    },
    {
      "epoch": 1.7520994906153917,
      "grad_norm": 0.49265122413635254,
      "learning_rate": 3.8245198051219585e-06,
      "loss": 0.4401,
      "step": 38180
    },
    {
      "epoch": 1.7521453811206462,
      "grad_norm": 0.4839162230491638,
      "learning_rate": 3.824281490719171e-06,
      "loss": 0.3413,
      "step": 38181
    },
    {
      "epoch": 1.7521912716259007,
      "grad_norm": 0.48862046003341675,
      "learning_rate": 3.824043179143615e-06,
      "loss": 0.3696,
      "step": 38182
    },
    {
      "epoch": 1.7522371621311552,
      "grad_norm": 0.44298094511032104,
      "learning_rate": 3.823804870395863e-06,
      "loss": 0.2985,
      "step": 38183
    },
    {
      "epoch": 1.7522830526364095,
      "grad_norm": 0.4870200455188751,
      "learning_rate": 3.823566564476491e-06,
      "loss": 0.3229,
      "step": 38184
    },
    {
      "epoch": 1.752328943141664,
      "grad_norm": 0.4738350212574005,
      "learning_rate": 3.823328261386069e-06,
      "loss": 0.3569,
      "step": 38185
    },
    {
      "epoch": 1.7523748336469185,
      "grad_norm": 0.4842919707298279,
      "learning_rate": 3.82308996112517e-06,
      "loss": 0.3548,
      "step": 38186
    },
    {
      "epoch": 1.7524207241521728,
      "grad_norm": 0.43787020444869995,
      "learning_rate": 3.82285166369437e-06,
      "loss": 0.3055,
      "step": 38187
    },
    {
      "epoch": 1.7524666146574273,
      "grad_norm": 0.49981147050857544,
      "learning_rate": 3.82261336909424e-06,
      "loss": 0.3632,
      "step": 38188
    },
    {
      "epoch": 1.7525125051626818,
      "grad_norm": 0.5303622484207153,
      "learning_rate": 3.822375077325352e-06,
      "loss": 0.4237,
      "step": 38189
    },
    {
      "epoch": 1.7525583956679363,
      "grad_norm": 0.43769294023513794,
      "learning_rate": 3.822136788388282e-06,
      "loss": 0.3236,
      "step": 38190
    },
    {
      "epoch": 1.7526042861731908,
      "grad_norm": 0.463581383228302,
      "learning_rate": 3.821898502283601e-06,
      "loss": 0.3084,
      "step": 38191
    },
    {
      "epoch": 1.7526501766784452,
      "grad_norm": 0.4965786933898926,
      "learning_rate": 3.82166021901188e-06,
      "loss": 0.4554,
      "step": 38192
    },
    {
      "epoch": 1.7526960671836997,
      "grad_norm": 0.4936651885509491,
      "learning_rate": 3.821421938573697e-06,
      "loss": 0.3791,
      "step": 38193
    },
    {
      "epoch": 1.7527419576889542,
      "grad_norm": 0.5074048042297363,
      "learning_rate": 3.821183660969622e-06,
      "loss": 0.3743,
      "step": 38194
    },
    {
      "epoch": 1.7527878481942087,
      "grad_norm": 0.5093694925308228,
      "learning_rate": 3.820945386200226e-06,
      "loss": 0.4438,
      "step": 38195
    },
    {
      "epoch": 1.7528337386994632,
      "grad_norm": 0.4397948086261749,
      "learning_rate": 3.820707114266086e-06,
      "loss": 0.3425,
      "step": 38196
    },
    {
      "epoch": 1.7528796292047175,
      "grad_norm": 0.49644970893859863,
      "learning_rate": 3.820468845167773e-06,
      "loss": 0.4014,
      "step": 38197
    },
    {
      "epoch": 1.752925519709972,
      "grad_norm": 0.4853942394256592,
      "learning_rate": 3.82023057890586e-06,
      "loss": 0.3985,
      "step": 38198
    },
    {
      "epoch": 1.7529714102152265,
      "grad_norm": 0.5195972919464111,
      "learning_rate": 3.819992315480921e-06,
      "loss": 0.4851,
      "step": 38199
    },
    {
      "epoch": 1.7530173007204808,
      "grad_norm": 0.42381638288497925,
      "learning_rate": 3.819754054893527e-06,
      "loss": 0.2741,
      "step": 38200
    },
    {
      "epoch": 1.7530631912257353,
      "grad_norm": 0.4638938009738922,
      "learning_rate": 3.819515797144251e-06,
      "loss": 0.3702,
      "step": 38201
    },
    {
      "epoch": 1.7531090817309898,
      "grad_norm": 0.4675822854042053,
      "learning_rate": 3.8192775422336685e-06,
      "loss": 0.3311,
      "step": 38202
    },
    {
      "epoch": 1.7531549722362443,
      "grad_norm": 0.4215841293334961,
      "learning_rate": 3.819039290162351e-06,
      "loss": 0.2788,
      "step": 38203
    },
    {
      "epoch": 1.7532008627414988,
      "grad_norm": 0.561294674873352,
      "learning_rate": 3.818801040930871e-06,
      "loss": 0.4176,
      "step": 38204
    },
    {
      "epoch": 1.7532467532467533,
      "grad_norm": 0.4887683689594269,
      "learning_rate": 3.818562794539801e-06,
      "loss": 0.3823,
      "step": 38205
    },
    {
      "epoch": 1.7532926437520078,
      "grad_norm": 0.5062733888626099,
      "learning_rate": 3.818324550989716e-06,
      "loss": 0.3712,
      "step": 38206
    },
    {
      "epoch": 1.7533385342572623,
      "grad_norm": 0.4431309103965759,
      "learning_rate": 3.818086310281185e-06,
      "loss": 0.3052,
      "step": 38207
    },
    {
      "epoch": 1.7533844247625168,
      "grad_norm": 0.7142221927642822,
      "learning_rate": 3.817848072414786e-06,
      "loss": 0.3849,
      "step": 38208
    },
    {
      "epoch": 1.753430315267771,
      "grad_norm": 0.44148916006088257,
      "learning_rate": 3.817609837391088e-06,
      "loss": 0.303,
      "step": 38209
    },
    {
      "epoch": 1.7534762057730255,
      "grad_norm": 0.5489116311073303,
      "learning_rate": 3.817371605210665e-06,
      "loss": 0.4632,
      "step": 38210
    },
    {
      "epoch": 1.75352209627828,
      "grad_norm": 0.43401098251342773,
      "learning_rate": 3.817133375874091e-06,
      "loss": 0.3572,
      "step": 38211
    },
    {
      "epoch": 1.7535679867835345,
      "grad_norm": 0.4292687177658081,
      "learning_rate": 3.816895149381938e-06,
      "loss": 0.2786,
      "step": 38212
    },
    {
      "epoch": 1.7536138772887888,
      "grad_norm": 0.4996427595615387,
      "learning_rate": 3.816656925734777e-06,
      "loss": 0.384,
      "step": 38213
    },
    {
      "epoch": 1.7536597677940433,
      "grad_norm": 0.4408896863460541,
      "learning_rate": 3.816418704933184e-06,
      "loss": 0.2822,
      "step": 38214
    },
    {
      "epoch": 1.7537056582992978,
      "grad_norm": 0.4812081456184387,
      "learning_rate": 3.8161804869777315e-06,
      "loss": 0.4075,
      "step": 38215
    },
    {
      "epoch": 1.7537515488045523,
      "grad_norm": 0.47232586145401,
      "learning_rate": 3.815942271868991e-06,
      "loss": 0.3126,
      "step": 38216
    },
    {
      "epoch": 1.7537974393098068,
      "grad_norm": 0.46676433086395264,
      "learning_rate": 3.815704059607534e-06,
      "loss": 0.3109,
      "step": 38217
    },
    {
      "epoch": 1.7538433298150613,
      "grad_norm": 0.4698551595211029,
      "learning_rate": 3.8154658501939355e-06,
      "loss": 0.3201,
      "step": 38218
    },
    {
      "epoch": 1.7538892203203158,
      "grad_norm": 0.4659937620162964,
      "learning_rate": 3.8152276436287686e-06,
      "loss": 0.3305,
      "step": 38219
    },
    {
      "epoch": 1.7539351108255703,
      "grad_norm": 0.4518466591835022,
      "learning_rate": 3.8149894399126045e-06,
      "loss": 0.3299,
      "step": 38220
    },
    {
      "epoch": 1.7539810013308248,
      "grad_norm": 0.42783045768737793,
      "learning_rate": 3.814751239046017e-06,
      "loss": 0.2663,
      "step": 38221
    },
    {
      "epoch": 1.754026891836079,
      "grad_norm": 0.5260459780693054,
      "learning_rate": 3.814513041029579e-06,
      "loss": 0.4022,
      "step": 38222
    },
    {
      "epoch": 1.7540727823413336,
      "grad_norm": 0.48999151587486267,
      "learning_rate": 3.814274845863861e-06,
      "loss": 0.3774,
      "step": 38223
    },
    {
      "epoch": 1.754118672846588,
      "grad_norm": 0.5121983289718628,
      "learning_rate": 3.81403665354944e-06,
      "loss": 0.3911,
      "step": 38224
    },
    {
      "epoch": 1.7541645633518423,
      "grad_norm": 0.47604820132255554,
      "learning_rate": 3.8137984640868863e-06,
      "loss": 0.3423,
      "step": 38225
    },
    {
      "epoch": 1.7542104538570968,
      "grad_norm": 0.471549928188324,
      "learning_rate": 3.813560277476772e-06,
      "loss": 0.3698,
      "step": 38226
    },
    {
      "epoch": 1.7542563443623513,
      "grad_norm": 0.43529465794563293,
      "learning_rate": 3.8133220937196718e-06,
      "loss": 0.2943,
      "step": 38227
    },
    {
      "epoch": 1.7543022348676058,
      "grad_norm": 0.4500519633293152,
      "learning_rate": 3.813083912816158e-06,
      "loss": 0.3436,
      "step": 38228
    },
    {
      "epoch": 1.7543481253728603,
      "grad_norm": 0.4286739230155945,
      "learning_rate": 3.8128457347668e-06,
      "loss": 0.2709,
      "step": 38229
    },
    {
      "epoch": 1.7543940158781148,
      "grad_norm": 0.507461428642273,
      "learning_rate": 3.8126075595721756e-06,
      "loss": 0.3778,
      "step": 38230
    },
    {
      "epoch": 1.7544399063833693,
      "grad_norm": 0.4735884666442871,
      "learning_rate": 3.812369387232855e-06,
      "loss": 0.3886,
      "step": 38231
    },
    {
      "epoch": 1.7544857968886238,
      "grad_norm": 0.4654468595981598,
      "learning_rate": 3.8121312177494107e-06,
      "loss": 0.3396,
      "step": 38232
    },
    {
      "epoch": 1.7545316873938783,
      "grad_norm": 0.4904736280441284,
      "learning_rate": 3.8118930511224168e-06,
      "loss": 0.3678,
      "step": 38233
    },
    {
      "epoch": 1.7545775778991328,
      "grad_norm": 0.44794490933418274,
      "learning_rate": 3.811654887352445e-06,
      "loss": 0.3165,
      "step": 38234
    },
    {
      "epoch": 1.754623468404387,
      "grad_norm": 0.45835959911346436,
      "learning_rate": 3.8114167264400658e-06,
      "loss": 0.3273,
      "step": 38235
    },
    {
      "epoch": 1.7546693589096416,
      "grad_norm": 0.47494685649871826,
      "learning_rate": 3.8111785683858563e-06,
      "loss": 0.3811,
      "step": 38236
    },
    {
      "epoch": 1.754715249414896,
      "grad_norm": 0.5015307068824768,
      "learning_rate": 3.8109404131903886e-06,
      "loss": 0.4401,
      "step": 38237
    },
    {
      "epoch": 1.7547611399201504,
      "grad_norm": 0.45531246066093445,
      "learning_rate": 3.810702260854231e-06,
      "loss": 0.3172,
      "step": 38238
    },
    {
      "epoch": 1.7548070304254049,
      "grad_norm": 0.4644160270690918,
      "learning_rate": 3.8104641113779606e-06,
      "loss": 0.3797,
      "step": 38239
    },
    {
      "epoch": 1.7548529209306594,
      "grad_norm": 0.49182045459747314,
      "learning_rate": 3.8102259647621494e-06,
      "loss": 0.347,
      "step": 38240
    },
    {
      "epoch": 1.7548988114359139,
      "grad_norm": 0.4740894138813019,
      "learning_rate": 3.8099878210073676e-06,
      "loss": 0.3179,
      "step": 38241
    },
    {
      "epoch": 1.7549447019411684,
      "grad_norm": 0.47984302043914795,
      "learning_rate": 3.80974968011419e-06,
      "loss": 0.4062,
      "step": 38242
    },
    {
      "epoch": 1.7549905924464229,
      "grad_norm": 0.48271089792251587,
      "learning_rate": 3.8095115420831895e-06,
      "loss": 0.39,
      "step": 38243
    },
    {
      "epoch": 1.7550364829516774,
      "grad_norm": 0.4634411036968231,
      "learning_rate": 3.809273406914936e-06,
      "loss": 0.3611,
      "step": 38244
    },
    {
      "epoch": 1.7550823734569319,
      "grad_norm": 0.45366501808166504,
      "learning_rate": 3.809035274610006e-06,
      "loss": 0.3476,
      "step": 38245
    },
    {
      "epoch": 1.7551282639621864,
      "grad_norm": 0.454681932926178,
      "learning_rate": 3.8087971451689708e-06,
      "loss": 0.3877,
      "step": 38246
    },
    {
      "epoch": 1.7551741544674409,
      "grad_norm": 0.4686877727508545,
      "learning_rate": 3.808559018592402e-06,
      "loss": 0.3166,
      "step": 38247
    },
    {
      "epoch": 1.7552200449726951,
      "grad_norm": 0.5123105645179749,
      "learning_rate": 3.8083208948808713e-06,
      "loss": 0.4276,
      "step": 38248
    },
    {
      "epoch": 1.7552659354779496,
      "grad_norm": 0.4363461434841156,
      "learning_rate": 3.8080827740349545e-06,
      "loss": 0.3271,
      "step": 38249
    },
    {
      "epoch": 1.7553118259832041,
      "grad_norm": 0.4420645833015442,
      "learning_rate": 3.8078446560552223e-06,
      "loss": 0.3274,
      "step": 38250
    },
    {
      "epoch": 1.7553577164884584,
      "grad_norm": 0.5362547039985657,
      "learning_rate": 3.807606540942245e-06,
      "loss": 0.4109,
      "step": 38251
    },
    {
      "epoch": 1.755403606993713,
      "grad_norm": 0.42891788482666016,
      "learning_rate": 3.8073684286966e-06,
      "loss": 0.2917,
      "step": 38252
    },
    {
      "epoch": 1.7554494974989674,
      "grad_norm": 0.5154009461402893,
      "learning_rate": 3.807130319318858e-06,
      "loss": 0.4774,
      "step": 38253
    },
    {
      "epoch": 1.755495388004222,
      "grad_norm": 0.5612766146659851,
      "learning_rate": 3.806892212809589e-06,
      "loss": 0.4641,
      "step": 38254
    },
    {
      "epoch": 1.7555412785094764,
      "grad_norm": 0.5181286334991455,
      "learning_rate": 3.80665410916937e-06,
      "loss": 0.5061,
      "step": 38255
    },
    {
      "epoch": 1.7555871690147309,
      "grad_norm": 0.531748354434967,
      "learning_rate": 3.80641600839877e-06,
      "loss": 0.412,
      "step": 38256
    },
    {
      "epoch": 1.7556330595199854,
      "grad_norm": 0.4841928482055664,
      "learning_rate": 3.806177910498361e-06,
      "loss": 0.3972,
      "step": 38257
    },
    {
      "epoch": 1.7556789500252399,
      "grad_norm": 0.4579367935657501,
      "learning_rate": 3.80593981546872e-06,
      "loss": 0.3325,
      "step": 38258
    },
    {
      "epoch": 1.7557248405304944,
      "grad_norm": 0.6023992300033569,
      "learning_rate": 3.8057017233104175e-06,
      "loss": 0.4487,
      "step": 38259
    },
    {
      "epoch": 1.7557707310357487,
      "grad_norm": 0.46867311000823975,
      "learning_rate": 3.8054636340240225e-06,
      "loss": 0.333,
      "step": 38260
    },
    {
      "epoch": 1.7558166215410032,
      "grad_norm": 0.45162853598594666,
      "learning_rate": 3.8052255476101124e-06,
      "loss": 0.3372,
      "step": 38261
    },
    {
      "epoch": 1.7558625120462577,
      "grad_norm": 0.46779870986938477,
      "learning_rate": 3.8049874640692575e-06,
      "loss": 0.3574,
      "step": 38262
    },
    {
      "epoch": 1.755908402551512,
      "grad_norm": 0.44593721628189087,
      "learning_rate": 3.80474938340203e-06,
      "loss": 0.2959,
      "step": 38263
    },
    {
      "epoch": 1.7559542930567664,
      "grad_norm": 0.4924672842025757,
      "learning_rate": 3.804511305609004e-06,
      "loss": 0.4088,
      "step": 38264
    },
    {
      "epoch": 1.756000183562021,
      "grad_norm": 0.4211216866970062,
      "learning_rate": 3.804273230690751e-06,
      "loss": 0.3098,
      "step": 38265
    },
    {
      "epoch": 1.7560460740672754,
      "grad_norm": 0.46275594830513,
      "learning_rate": 3.8040351586478406e-06,
      "loss": 0.3809,
      "step": 38266
    },
    {
      "epoch": 1.75609196457253,
      "grad_norm": 0.4629899561405182,
      "learning_rate": 3.803797089480852e-06,
      "loss": 0.3583,
      "step": 38267
    },
    {
      "epoch": 1.7561378550777844,
      "grad_norm": 0.4310542047023773,
      "learning_rate": 3.8035590231903524e-06,
      "loss": 0.2912,
      "step": 38268
    },
    {
      "epoch": 1.756183745583039,
      "grad_norm": 0.47964125871658325,
      "learning_rate": 3.803320959776915e-06,
      "loss": 0.3584,
      "step": 38269
    },
    {
      "epoch": 1.7562296360882934,
      "grad_norm": 0.47627997398376465,
      "learning_rate": 3.803082899241114e-06,
      "loss": 0.3475,
      "step": 38270
    },
    {
      "epoch": 1.756275526593548,
      "grad_norm": 0.4328688979148865,
      "learning_rate": 3.802844841583521e-06,
      "loss": 0.2954,
      "step": 38271
    },
    {
      "epoch": 1.7563214170988024,
      "grad_norm": 0.5056161880493164,
      "learning_rate": 3.8026067868047068e-06,
      "loss": 0.3967,
      "step": 38272
    },
    {
      "epoch": 1.7563673076040567,
      "grad_norm": 0.4901103377342224,
      "learning_rate": 3.8023687349052467e-06,
      "loss": 0.398,
      "step": 38273
    },
    {
      "epoch": 1.7564131981093112,
      "grad_norm": 0.455577552318573,
      "learning_rate": 3.802130685885712e-06,
      "loss": 0.3302,
      "step": 38274
    },
    {
      "epoch": 1.7564590886145657,
      "grad_norm": 0.5018193125724792,
      "learning_rate": 3.8018926397466736e-06,
      "loss": 0.4099,
      "step": 38275
    },
    {
      "epoch": 1.75650497911982,
      "grad_norm": 0.462346613407135,
      "learning_rate": 3.801654596488707e-06,
      "loss": 0.3692,
      "step": 38276
    },
    {
      "epoch": 1.7565508696250745,
      "grad_norm": 0.4271996319293976,
      "learning_rate": 3.8014165561123827e-06,
      "loss": 0.2911,
      "step": 38277
    },
    {
      "epoch": 1.756596760130329,
      "grad_norm": 0.5613418221473694,
      "learning_rate": 3.801178518618271e-06,
      "loss": 0.3896,
      "step": 38278
    },
    {
      "epoch": 1.7566426506355834,
      "grad_norm": 0.4786616861820221,
      "learning_rate": 3.8009404840069484e-06,
      "loss": 0.3597,
      "step": 38279
    },
    {
      "epoch": 1.756688541140838,
      "grad_norm": 0.4913674592971802,
      "learning_rate": 3.800702452278986e-06,
      "loss": 0.3986,
      "step": 38280
    },
    {
      "epoch": 1.7567344316460924,
      "grad_norm": 0.45116549730300903,
      "learning_rate": 3.800464423434957e-06,
      "loss": 0.3164,
      "step": 38281
    },
    {
      "epoch": 1.756780322151347,
      "grad_norm": 0.4954787492752075,
      "learning_rate": 3.800226397475429e-06,
      "loss": 0.4334,
      "step": 38282
    },
    {
      "epoch": 1.7568262126566014,
      "grad_norm": 0.46176910400390625,
      "learning_rate": 3.7999883744009795e-06,
      "loss": 0.3436,
      "step": 38283
    },
    {
      "epoch": 1.756872103161856,
      "grad_norm": 0.4726191461086273,
      "learning_rate": 3.7997503542121798e-06,
      "loss": 0.3411,
      "step": 38284
    },
    {
      "epoch": 1.7569179936671104,
      "grad_norm": 0.47916683554649353,
      "learning_rate": 3.7995123369096004e-06,
      "loss": 0.3353,
      "step": 38285
    },
    {
      "epoch": 1.7569638841723647,
      "grad_norm": 0.45721709728240967,
      "learning_rate": 3.799274322493816e-06,
      "loss": 0.3292,
      "step": 38286
    },
    {
      "epoch": 1.7570097746776192,
      "grad_norm": 0.46589529514312744,
      "learning_rate": 3.7990363109653984e-06,
      "loss": 0.3393,
      "step": 38287
    },
    {
      "epoch": 1.7570556651828737,
      "grad_norm": 0.44673770666122437,
      "learning_rate": 3.7987983023249163e-06,
      "loss": 0.324,
      "step": 38288
    },
    {
      "epoch": 1.757101555688128,
      "grad_norm": 0.4309248924255371,
      "learning_rate": 3.798560296572948e-06,
      "loss": 0.3154,
      "step": 38289
    },
    {
      "epoch": 1.7571474461933825,
      "grad_norm": 0.52397221326828,
      "learning_rate": 3.7983222937100626e-06,
      "loss": 0.4381,
      "step": 38290
    },
    {
      "epoch": 1.757193336698637,
      "grad_norm": 0.4323873519897461,
      "learning_rate": 3.7980842937368325e-06,
      "loss": 0.2796,
      "step": 38291
    },
    {
      "epoch": 1.7572392272038915,
      "grad_norm": 0.4782414734363556,
      "learning_rate": 3.797846296653831e-06,
      "loss": 0.3544,
      "step": 38292
    },
    {
      "epoch": 1.757285117709146,
      "grad_norm": 0.4502119719982147,
      "learning_rate": 3.7976083024616297e-06,
      "loss": 0.3156,
      "step": 38293
    },
    {
      "epoch": 1.7573310082144005,
      "grad_norm": 0.4780610501766205,
      "learning_rate": 3.7973703111607984e-06,
      "loss": 0.397,
      "step": 38294
    },
    {
      "epoch": 1.757376898719655,
      "grad_norm": 0.5238561034202576,
      "learning_rate": 3.7971323227519147e-06,
      "loss": 0.3922,
      "step": 38295
    },
    {
      "epoch": 1.7574227892249095,
      "grad_norm": 0.4771853983402252,
      "learning_rate": 3.7968943372355477e-06,
      "loss": 0.4037,
      "step": 38296
    },
    {
      "epoch": 1.757468679730164,
      "grad_norm": 0.5051767826080322,
      "learning_rate": 3.796656354612269e-06,
      "loss": 0.4289,
      "step": 38297
    },
    {
      "epoch": 1.7575145702354182,
      "grad_norm": 0.4776654839515686,
      "learning_rate": 3.7964183748826534e-06,
      "loss": 0.3367,
      "step": 38298
    },
    {
      "epoch": 1.7575604607406727,
      "grad_norm": 0.4678038954734802,
      "learning_rate": 3.7961803980472716e-06,
      "loss": 0.4064,
      "step": 38299
    },
    {
      "epoch": 1.7576063512459272,
      "grad_norm": 0.47737982869148254,
      "learning_rate": 3.795942424106694e-06,
      "loss": 0.3681,
      "step": 38300
    },
    {
      "epoch": 1.7576522417511817,
      "grad_norm": 0.45834699273109436,
      "learning_rate": 3.7957044530614966e-06,
      "loss": 0.3271,
      "step": 38301
    },
    {
      "epoch": 1.757698132256436,
      "grad_norm": 0.4882853329181671,
      "learning_rate": 3.79546648491225e-06,
      "loss": 0.3643,
      "step": 38302
    },
    {
      "epoch": 1.7577440227616905,
      "grad_norm": 0.443279892206192,
      "learning_rate": 3.795228519659526e-06,
      "loss": 0.3291,
      "step": 38303
    },
    {
      "epoch": 1.757789913266945,
      "grad_norm": 0.4696867763996124,
      "learning_rate": 3.7949905573038977e-06,
      "loss": 0.3687,
      "step": 38304
    },
    {
      "epoch": 1.7578358037721995,
      "grad_norm": 0.48926839232444763,
      "learning_rate": 3.7947525978459364e-06,
      "loss": 0.3365,
      "step": 38305
    },
    {
      "epoch": 1.757881694277454,
      "grad_norm": 0.5036697387695312,
      "learning_rate": 3.7945146412862145e-06,
      "loss": 0.3829,
      "step": 38306
    },
    {
      "epoch": 1.7579275847827085,
      "grad_norm": 0.4913530647754669,
      "learning_rate": 3.794276687625305e-06,
      "loss": 0.3769,
      "step": 38307
    },
    {
      "epoch": 1.757973475287963,
      "grad_norm": 0.45507287979125977,
      "learning_rate": 3.794038736863779e-06,
      "loss": 0.3604,
      "step": 38308
    },
    {
      "epoch": 1.7580193657932175,
      "grad_norm": 0.4666888117790222,
      "learning_rate": 3.7938007890022084e-06,
      "loss": 0.349,
      "step": 38309
    },
    {
      "epoch": 1.758065256298472,
      "grad_norm": 0.49028727412223816,
      "learning_rate": 3.793562844041168e-06,
      "loss": 0.3938,
      "step": 38310
    },
    {
      "epoch": 1.7581111468037263,
      "grad_norm": 0.5133760571479797,
      "learning_rate": 3.793324901981227e-06,
      "loss": 0.3208,
      "step": 38311
    },
    {
      "epoch": 1.7581570373089808,
      "grad_norm": 0.5037102103233337,
      "learning_rate": 3.79308696282296e-06,
      "loss": 0.403,
      "step": 38312
    },
    {
      "epoch": 1.7582029278142353,
      "grad_norm": 0.4614470899105072,
      "learning_rate": 3.792849026566936e-06,
      "loss": 0.3263,
      "step": 38313
    },
    {
      "epoch": 1.7582488183194895,
      "grad_norm": 0.4743252694606781,
      "learning_rate": 3.7926110932137304e-06,
      "loss": 0.3867,
      "step": 38314
    },
    {
      "epoch": 1.758294708824744,
      "grad_norm": 0.4464605152606964,
      "learning_rate": 3.792373162763915e-06,
      "loss": 0.2822,
      "step": 38315
    },
    {
      "epoch": 1.7583405993299985,
      "grad_norm": 0.4677067995071411,
      "learning_rate": 3.7921352352180577e-06,
      "loss": 0.337,
      "step": 38316
    },
    {
      "epoch": 1.758386489835253,
      "grad_norm": 0.4541931450366974,
      "learning_rate": 3.7918973105767365e-06,
      "loss": 0.328,
      "step": 38317
    },
    {
      "epoch": 1.7584323803405075,
      "grad_norm": 0.4460824131965637,
      "learning_rate": 3.791659388840521e-06,
      "loss": 0.2998,
      "step": 38318
    },
    {
      "epoch": 1.758478270845762,
      "grad_norm": 0.5064753890037537,
      "learning_rate": 3.7914214700099827e-06,
      "loss": 0.3477,
      "step": 38319
    },
    {
      "epoch": 1.7585241613510165,
      "grad_norm": 0.4383980929851532,
      "learning_rate": 3.7911835540856944e-06,
      "loss": 0.2813,
      "step": 38320
    },
    {
      "epoch": 1.758570051856271,
      "grad_norm": 0.4959060847759247,
      "learning_rate": 3.7909456410682288e-06,
      "loss": 0.3468,
      "step": 38321
    },
    {
      "epoch": 1.7586159423615255,
      "grad_norm": 0.47594988346099854,
      "learning_rate": 3.790707730958155e-06,
      "loss": 0.3899,
      "step": 38322
    },
    {
      "epoch": 1.75866183286678,
      "grad_norm": 0.48165324330329895,
      "learning_rate": 3.79046982375605e-06,
      "loss": 0.4432,
      "step": 38323
    },
    {
      "epoch": 1.7587077233720343,
      "grad_norm": 0.4453853964805603,
      "learning_rate": 3.790231919462483e-06,
      "loss": 0.3513,
      "step": 38324
    },
    {
      "epoch": 1.7587536138772888,
      "grad_norm": 0.46017348766326904,
      "learning_rate": 3.7899940180780255e-06,
      "loss": 0.3282,
      "step": 38325
    },
    {
      "epoch": 1.7587995043825433,
      "grad_norm": 0.4543053209781647,
      "learning_rate": 3.7897561196032512e-06,
      "loss": 0.3499,
      "step": 38326
    },
    {
      "epoch": 1.7588453948877976,
      "grad_norm": 0.45544353127479553,
      "learning_rate": 3.7895182240387318e-06,
      "loss": 0.3782,
      "step": 38327
    },
    {
      "epoch": 1.758891285393052,
      "grad_norm": 0.5120173692703247,
      "learning_rate": 3.789280331385038e-06,
      "loss": 0.4256,
      "step": 38328
    },
    {
      "epoch": 1.7589371758983066,
      "grad_norm": 0.5207133293151855,
      "learning_rate": 3.7890424416427436e-06,
      "loss": 0.4154,
      "step": 38329
    },
    {
      "epoch": 1.758983066403561,
      "grad_norm": 0.456525057554245,
      "learning_rate": 3.7888045548124205e-06,
      "loss": 0.3274,
      "step": 38330
    },
    {
      "epoch": 1.7590289569088156,
      "grad_norm": 0.46602892875671387,
      "learning_rate": 3.788566670894637e-06,
      "loss": 0.3501,
      "step": 38331
    },
    {
      "epoch": 1.75907484741407,
      "grad_norm": 0.4597901403903961,
      "learning_rate": 3.788328789889971e-06,
      "loss": 0.3428,
      "step": 38332
    },
    {
      "epoch": 1.7591207379193246,
      "grad_norm": 0.4388744831085205,
      "learning_rate": 3.788090911798993e-06,
      "loss": 0.3013,
      "step": 38333
    },
    {
      "epoch": 1.759166628424579,
      "grad_norm": 0.48054590821266174,
      "learning_rate": 3.7878530366222716e-06,
      "loss": 0.3544,
      "step": 38334
    },
    {
      "epoch": 1.7592125189298335,
      "grad_norm": 0.4662417769432068,
      "learning_rate": 3.787615164360382e-06,
      "loss": 0.3623,
      "step": 38335
    },
    {
      "epoch": 1.759258409435088,
      "grad_norm": 0.48504161834716797,
      "learning_rate": 3.787377295013896e-06,
      "loss": 0.4041,
      "step": 38336
    },
    {
      "epoch": 1.7593042999403423,
      "grad_norm": 0.5425921082496643,
      "learning_rate": 3.7871394285833817e-06,
      "loss": 0.3966,
      "step": 38337
    },
    {
      "epoch": 1.7593501904455968,
      "grad_norm": 0.4686540961265564,
      "learning_rate": 3.7869015650694175e-06,
      "loss": 0.3232,
      "step": 38338
    },
    {
      "epoch": 1.7593960809508513,
      "grad_norm": 0.4686158001422882,
      "learning_rate": 3.7866637044725717e-06,
      "loss": 0.3438,
      "step": 38339
    },
    {
      "epoch": 1.7594419714561056,
      "grad_norm": 0.4754941165447235,
      "learning_rate": 3.7864258467934156e-06,
      "loss": 0.3807,
      "step": 38340
    },
    {
      "epoch": 1.75948786196136,
      "grad_norm": 0.4595724940299988,
      "learning_rate": 3.786187992032524e-06,
      "loss": 0.3804,
      "step": 38341
    },
    {
      "epoch": 1.7595337524666146,
      "grad_norm": 0.46796467900276184,
      "learning_rate": 3.785950140190466e-06,
      "loss": 0.3638,
      "step": 38342
    },
    {
      "epoch": 1.759579642971869,
      "grad_norm": 0.5049145817756653,
      "learning_rate": 3.785712291267814e-06,
      "loss": 0.4168,
      "step": 38343
    },
    {
      "epoch": 1.7596255334771236,
      "grad_norm": 0.45139241218566895,
      "learning_rate": 3.785474445265142e-06,
      "loss": 0.3092,
      "step": 38344
    },
    {
      "epoch": 1.759671423982378,
      "grad_norm": 0.4832097291946411,
      "learning_rate": 3.785236602183021e-06,
      "loss": 0.3738,
      "step": 38345
    },
    {
      "epoch": 1.7597173144876326,
      "grad_norm": 0.4438186585903168,
      "learning_rate": 3.7849987620220226e-06,
      "loss": 0.2817,
      "step": 38346
    },
    {
      "epoch": 1.759763204992887,
      "grad_norm": 0.4807962477207184,
      "learning_rate": 3.784760924782717e-06,
      "loss": 0.3749,
      "step": 38347
    },
    {
      "epoch": 1.7598090954981416,
      "grad_norm": 0.500275194644928,
      "learning_rate": 3.7845230904656793e-06,
      "loss": 0.311,
      "step": 38348
    },
    {
      "epoch": 1.7598549860033958,
      "grad_norm": 0.4842105507850647,
      "learning_rate": 3.7842852590714806e-06,
      "loss": 0.4042,
      "step": 38349
    },
    {
      "epoch": 1.7599008765086503,
      "grad_norm": 0.45828014612197876,
      "learning_rate": 3.7840474306006904e-06,
      "loss": 0.2946,
      "step": 38350
    },
    {
      "epoch": 1.7599467670139048,
      "grad_norm": 0.47050970792770386,
      "learning_rate": 3.7838096050538837e-06,
      "loss": 0.3763,
      "step": 38351
    },
    {
      "epoch": 1.7599926575191591,
      "grad_norm": 0.46474409103393555,
      "learning_rate": 3.783571782431631e-06,
      "loss": 0.3647,
      "step": 38352
    },
    {
      "epoch": 1.7600385480244136,
      "grad_norm": 0.47763368487358093,
      "learning_rate": 3.7833339627345017e-06,
      "loss": 0.377,
      "step": 38353
    },
    {
      "epoch": 1.7600844385296681,
      "grad_norm": 0.5136001706123352,
      "learning_rate": 3.7830961459630728e-06,
      "loss": 0.3972,
      "step": 38354
    },
    {
      "epoch": 1.7601303290349226,
      "grad_norm": 0.45928817987442017,
      "learning_rate": 3.7828583321179136e-06,
      "loss": 0.3103,
      "step": 38355
    },
    {
      "epoch": 1.7601762195401771,
      "grad_norm": 0.43964308500289917,
      "learning_rate": 3.7826205211995946e-06,
      "loss": 0.3174,
      "step": 38356
    },
    {
      "epoch": 1.7602221100454316,
      "grad_norm": 0.47442957758903503,
      "learning_rate": 3.7823827132086897e-06,
      "loss": 0.3611,
      "step": 38357
    },
    {
      "epoch": 1.760268000550686,
      "grad_norm": 0.43624040484428406,
      "learning_rate": 3.782144908145771e-06,
      "loss": 0.3083,
      "step": 38358
    },
    {
      "epoch": 1.7603138910559406,
      "grad_norm": 0.5192256569862366,
      "learning_rate": 3.781907106011407e-06,
      "loss": 0.4205,
      "step": 38359
    },
    {
      "epoch": 1.760359781561195,
      "grad_norm": 0.502558171749115,
      "learning_rate": 3.7816693068061746e-06,
      "loss": 0.416,
      "step": 38360
    },
    {
      "epoch": 1.7604056720664496,
      "grad_norm": 0.4580441415309906,
      "learning_rate": 3.781431510530642e-06,
      "loss": 0.3364,
      "step": 38361
    },
    {
      "epoch": 1.7604515625717039,
      "grad_norm": 0.48370984196662903,
      "learning_rate": 3.781193717185382e-06,
      "loss": 0.3511,
      "step": 38362
    },
    {
      "epoch": 1.7604974530769584,
      "grad_norm": 0.46518009901046753,
      "learning_rate": 3.7809559267709663e-06,
      "loss": 0.3778,
      "step": 38363
    },
    {
      "epoch": 1.7605433435822129,
      "grad_norm": 0.43130454421043396,
      "learning_rate": 3.780718139287968e-06,
      "loss": 0.3222,
      "step": 38364
    },
    {
      "epoch": 1.7605892340874671,
      "grad_norm": 0.4705457091331482,
      "learning_rate": 3.7804803547369552e-06,
      "loss": 0.3457,
      "step": 38365
    },
    {
      "epoch": 1.7606351245927216,
      "grad_norm": 0.485038697719574,
      "learning_rate": 3.7802425731185037e-06,
      "loss": 0.3747,
      "step": 38366
    },
    {
      "epoch": 1.7606810150979761,
      "grad_norm": 0.5070366263389587,
      "learning_rate": 3.7800047944331846e-06,
      "loss": 0.4022,
      "step": 38367
    },
    {
      "epoch": 1.7607269056032306,
      "grad_norm": 0.4587084949016571,
      "learning_rate": 3.779767018681568e-06,
      "loss": 0.3471,
      "step": 38368
    },
    {
      "epoch": 1.7607727961084851,
      "grad_norm": 0.43464237451553345,
      "learning_rate": 3.7795292458642274e-06,
      "loss": 0.3485,
      "step": 38369
    },
    {
      "epoch": 1.7608186866137396,
      "grad_norm": 0.47145384550094604,
      "learning_rate": 3.7792914759817333e-06,
      "loss": 0.3344,
      "step": 38370
    },
    {
      "epoch": 1.7608645771189941,
      "grad_norm": 0.49127840995788574,
      "learning_rate": 3.7790537090346564e-06,
      "loss": 0.4116,
      "step": 38371
    },
    {
      "epoch": 1.7609104676242486,
      "grad_norm": 0.42733117938041687,
      "learning_rate": 3.7788159450235728e-06,
      "loss": 0.3343,
      "step": 38372
    },
    {
      "epoch": 1.7609563581295031,
      "grad_norm": 0.4438866376876831,
      "learning_rate": 3.778578183949051e-06,
      "loss": 0.322,
      "step": 38373
    },
    {
      "epoch": 1.7610022486347576,
      "grad_norm": 0.47877466678619385,
      "learning_rate": 3.7783404258116607e-06,
      "loss": 0.3544,
      "step": 38374
    },
    {
      "epoch": 1.761048139140012,
      "grad_norm": 0.4426877796649933,
      "learning_rate": 3.7781026706119784e-06,
      "loss": 0.3578,
      "step": 38375
    },
    {
      "epoch": 1.7610940296452664,
      "grad_norm": 0.4676334261894226,
      "learning_rate": 3.777864918350573e-06,
      "loss": 0.3668,
      "step": 38376
    },
    {
      "epoch": 1.761139920150521,
      "grad_norm": 0.4337848126888275,
      "learning_rate": 3.777627169028017e-06,
      "loss": 0.3165,
      "step": 38377
    },
    {
      "epoch": 1.7611858106557752,
      "grad_norm": 0.49028557538986206,
      "learning_rate": 3.777389422644881e-06,
      "loss": 0.388,
      "step": 38378
    },
    {
      "epoch": 1.7612317011610297,
      "grad_norm": 0.4711563289165497,
      "learning_rate": 3.777151679201738e-06,
      "loss": 0.4113,
      "step": 38379
    },
    {
      "epoch": 1.7612775916662842,
      "grad_norm": 0.4533214867115021,
      "learning_rate": 3.7769139386991606e-06,
      "loss": 0.3405,
      "step": 38380
    },
    {
      "epoch": 1.7613234821715387,
      "grad_norm": 0.45808663964271545,
      "learning_rate": 3.7766762011377154e-06,
      "loss": 0.3106,
      "step": 38381
    },
    {
      "epoch": 1.7613693726767932,
      "grad_norm": 0.46038907766342163,
      "learning_rate": 3.776438466517981e-06,
      "loss": 0.3795,
      "step": 38382
    },
    {
      "epoch": 1.7614152631820477,
      "grad_norm": 0.470052570104599,
      "learning_rate": 3.776200734840526e-06,
      "loss": 0.3401,
      "step": 38383
    },
    {
      "epoch": 1.7614611536873022,
      "grad_norm": 0.5138224959373474,
      "learning_rate": 3.7759630061059204e-06,
      "loss": 0.368,
      "step": 38384
    },
    {
      "epoch": 1.7615070441925567,
      "grad_norm": 0.4590214490890503,
      "learning_rate": 3.775725280314739e-06,
      "loss": 0.3594,
      "step": 38385
    },
    {
      "epoch": 1.7615529346978112,
      "grad_norm": 0.46629631519317627,
      "learning_rate": 3.775487557467552e-06,
      "loss": 0.3241,
      "step": 38386
    },
    {
      "epoch": 1.7615988252030654,
      "grad_norm": 0.48703789710998535,
      "learning_rate": 3.7752498375649283e-06,
      "loss": 0.3456,
      "step": 38387
    },
    {
      "epoch": 1.76164471570832,
      "grad_norm": 0.4532487690448761,
      "learning_rate": 3.775012120607444e-06,
      "loss": 0.369,
      "step": 38388
    },
    {
      "epoch": 1.7616906062135744,
      "grad_norm": 0.4913702607154846,
      "learning_rate": 3.77477440659567e-06,
      "loss": 0.3736,
      "step": 38389
    },
    {
      "epoch": 1.761736496718829,
      "grad_norm": 0.5234541893005371,
      "learning_rate": 3.7745366955301755e-06,
      "loss": 0.3576,
      "step": 38390
    },
    {
      "epoch": 1.7617823872240832,
      "grad_norm": 0.4744754731655121,
      "learning_rate": 3.7742989874115338e-06,
      "loss": 0.3805,
      "step": 38391
    },
    {
      "epoch": 1.7618282777293377,
      "grad_norm": 0.4563060998916626,
      "learning_rate": 3.7740612822403167e-06,
      "loss": 0.3087,
      "step": 38392
    },
    {
      "epoch": 1.7618741682345922,
      "grad_norm": 0.43083345890045166,
      "learning_rate": 3.773823580017093e-06,
      "loss": 0.2872,
      "step": 38393
    },
    {
      "epoch": 1.7619200587398467,
      "grad_norm": 0.5092521905899048,
      "learning_rate": 3.7735858807424402e-06,
      "loss": 0.4334,
      "step": 38394
    },
    {
      "epoch": 1.7619659492451012,
      "grad_norm": 0.46459004282951355,
      "learning_rate": 3.773348184416925e-06,
      "loss": 0.3155,
      "step": 38395
    },
    {
      "epoch": 1.7620118397503557,
      "grad_norm": 0.4866071045398712,
      "learning_rate": 3.7731104910411183e-06,
      "loss": 0.3932,
      "step": 38396
    },
    {
      "epoch": 1.7620577302556102,
      "grad_norm": 0.48907604813575745,
      "learning_rate": 3.7728728006155955e-06,
      "loss": 0.3739,
      "step": 38397
    },
    {
      "epoch": 1.7621036207608647,
      "grad_norm": 0.4651419520378113,
      "learning_rate": 3.7726351131409267e-06,
      "loss": 0.3507,
      "step": 38398
    },
    {
      "epoch": 1.7621495112661192,
      "grad_norm": 0.46683207154273987,
      "learning_rate": 3.772397428617681e-06,
      "loss": 0.3305,
      "step": 38399
    },
    {
      "epoch": 1.7621954017713735,
      "grad_norm": 0.48388731479644775,
      "learning_rate": 3.772159747046434e-06,
      "loss": 0.3705,
      "step": 38400
    },
    {
      "epoch": 1.762241292276628,
      "grad_norm": 0.4656377136707306,
      "learning_rate": 3.7719220684277553e-06,
      "loss": 0.3439,
      "step": 38401
    },
    {
      "epoch": 1.7622871827818825,
      "grad_norm": 0.4899931252002716,
      "learning_rate": 3.7716843927622144e-06,
      "loss": 0.3925,
      "step": 38402
    },
    {
      "epoch": 1.7623330732871367,
      "grad_norm": 0.48617154359817505,
      "learning_rate": 3.7714467200503867e-06,
      "loss": 0.4062,
      "step": 38403
    },
    {
      "epoch": 1.7623789637923912,
      "grad_norm": 0.5124910473823547,
      "learning_rate": 3.7712090502928417e-06,
      "loss": 0.4406,
      "step": 38404
    },
    {
      "epoch": 1.7624248542976457,
      "grad_norm": 0.5015543103218079,
      "learning_rate": 3.7709713834901505e-06,
      "loss": 0.4344,
      "step": 38405
    },
    {
      "epoch": 1.7624707448029002,
      "grad_norm": 0.47744908928871155,
      "learning_rate": 3.7707337196428863e-06,
      "loss": 0.3776,
      "step": 38406
    },
    {
      "epoch": 1.7625166353081547,
      "grad_norm": 0.4517383575439453,
      "learning_rate": 3.7704960587516194e-06,
      "loss": 0.2975,
      "step": 38407
    },
    {
      "epoch": 1.7625625258134092,
      "grad_norm": 0.4984171688556671,
      "learning_rate": 3.7702584008169184e-06,
      "loss": 0.4094,
      "step": 38408
    },
    {
      "epoch": 1.7626084163186637,
      "grad_norm": 0.6187036633491516,
      "learning_rate": 3.770020745839361e-06,
      "loss": 0.3674,
      "step": 38409
    },
    {
      "epoch": 1.7626543068239182,
      "grad_norm": 0.543114423751831,
      "learning_rate": 3.7697830938195157e-06,
      "loss": 0.5127,
      "step": 38410
    },
    {
      "epoch": 1.7627001973291727,
      "grad_norm": 0.5098135471343994,
      "learning_rate": 3.769545444757953e-06,
      "loss": 0.4156,
      "step": 38411
    },
    {
      "epoch": 1.7627460878344272,
      "grad_norm": 0.5018177032470703,
      "learning_rate": 3.7693077986552445e-06,
      "loss": 0.4319,
      "step": 38412
    },
    {
      "epoch": 1.7627919783396815,
      "grad_norm": 0.4527531862258911,
      "learning_rate": 3.7690701555119635e-06,
      "loss": 0.3372,
      "step": 38413
    },
    {
      "epoch": 1.762837868844936,
      "grad_norm": 0.41302385926246643,
      "learning_rate": 3.7688325153286803e-06,
      "loss": 0.2613,
      "step": 38414
    },
    {
      "epoch": 1.7628837593501905,
      "grad_norm": 0.4281408190727234,
      "learning_rate": 3.768594878105964e-06,
      "loss": 0.3146,
      "step": 38415
    },
    {
      "epoch": 1.7629296498554448,
      "grad_norm": 0.5146076679229736,
      "learning_rate": 3.7683572438443917e-06,
      "loss": 0.4166,
      "step": 38416
    },
    {
      "epoch": 1.7629755403606993,
      "grad_norm": 0.4726836085319519,
      "learning_rate": 3.7681196125445303e-06,
      "loss": 0.3326,
      "step": 38417
    },
    {
      "epoch": 1.7630214308659538,
      "grad_norm": 0.4964827001094818,
      "learning_rate": 3.7678819842069503e-06,
      "loss": 0.3747,
      "step": 38418
    },
    {
      "epoch": 1.7630673213712083,
      "grad_norm": 0.48425567150115967,
      "learning_rate": 3.7676443588322276e-06,
      "loss": 0.3793,
      "step": 38419
    },
    {
      "epoch": 1.7631132118764627,
      "grad_norm": 0.4698258936405182,
      "learning_rate": 3.767406736420931e-06,
      "loss": 0.3643,
      "step": 38420
    },
    {
      "epoch": 1.7631591023817172,
      "grad_norm": 0.4701162874698639,
      "learning_rate": 3.7671691169736312e-06,
      "loss": 0.3813,
      "step": 38421
    },
    {
      "epoch": 1.7632049928869717,
      "grad_norm": 0.44547238945961,
      "learning_rate": 3.766931500490902e-06,
      "loss": 0.3073,
      "step": 38422
    },
    {
      "epoch": 1.7632508833922262,
      "grad_norm": 0.4657565951347351,
      "learning_rate": 3.7666938869733128e-06,
      "loss": 0.3454,
      "step": 38423
    },
    {
      "epoch": 1.7632967738974807,
      "grad_norm": 0.47479498386383057,
      "learning_rate": 3.766456276421434e-06,
      "loss": 0.3495,
      "step": 38424
    },
    {
      "epoch": 1.7633426644027352,
      "grad_norm": 0.4870316982269287,
      "learning_rate": 3.7662186688358403e-06,
      "loss": 0.3398,
      "step": 38425
    },
    {
      "epoch": 1.7633885549079895,
      "grad_norm": 0.46854791045188904,
      "learning_rate": 3.7659810642171013e-06,
      "loss": 0.33,
      "step": 38426
    },
    {
      "epoch": 1.763434445413244,
      "grad_norm": 0.48503589630126953,
      "learning_rate": 3.765743462565787e-06,
      "loss": 0.3936,
      "step": 38427
    },
    {
      "epoch": 1.7634803359184985,
      "grad_norm": 0.4690198302268982,
      "learning_rate": 3.765505863882472e-06,
      "loss": 0.359,
      "step": 38428
    },
    {
      "epoch": 1.7635262264237528,
      "grad_norm": 0.4582815170288086,
      "learning_rate": 3.7652682681677255e-06,
      "loss": 0.3611,
      "step": 38429
    },
    {
      "epoch": 1.7635721169290073,
      "grad_norm": 0.4690004289150238,
      "learning_rate": 3.7650306754221167e-06,
      "loss": 0.3468,
      "step": 38430
    },
    {
      "epoch": 1.7636180074342618,
      "grad_norm": 0.46284356713294983,
      "learning_rate": 3.7647930856462215e-06,
      "loss": 0.3853,
      "step": 38431
    },
    {
      "epoch": 1.7636638979395163,
      "grad_norm": 0.49339282512664795,
      "learning_rate": 3.7645554988406095e-06,
      "loss": 0.3891,
      "step": 38432
    },
    {
      "epoch": 1.7637097884447708,
      "grad_norm": 0.46456676721572876,
      "learning_rate": 3.7643179150058504e-06,
      "loss": 0.3125,
      "step": 38433
    },
    {
      "epoch": 1.7637556789500253,
      "grad_norm": 0.449203759431839,
      "learning_rate": 3.764080334142518e-06,
      "loss": 0.2988,
      "step": 38434
    },
    {
      "epoch": 1.7638015694552798,
      "grad_norm": 0.4307943880558014,
      "learning_rate": 3.7638427562511816e-06,
      "loss": 0.269,
      "step": 38435
    },
    {
      "epoch": 1.7638474599605343,
      "grad_norm": 0.4483773410320282,
      "learning_rate": 3.7636051813324116e-06,
      "loss": 0.3573,
      "step": 38436
    },
    {
      "epoch": 1.7638933504657888,
      "grad_norm": 0.44686639308929443,
      "learning_rate": 3.763367609386783e-06,
      "loss": 0.3197,
      "step": 38437
    },
    {
      "epoch": 1.763939240971043,
      "grad_norm": 0.47998306155204773,
      "learning_rate": 3.7631300404148663e-06,
      "loss": 0.4057,
      "step": 38438
    },
    {
      "epoch": 1.7639851314762975,
      "grad_norm": 0.46391916275024414,
      "learning_rate": 3.762892474417228e-06,
      "loss": 0.3209,
      "step": 38439
    },
    {
      "epoch": 1.764031021981552,
      "grad_norm": 0.45572006702423096,
      "learning_rate": 3.762654911394445e-06,
      "loss": 0.323,
      "step": 38440
    },
    {
      "epoch": 1.7640769124868063,
      "grad_norm": 0.4794723689556122,
      "learning_rate": 3.7624173513470863e-06,
      "loss": 0.3525,
      "step": 38441
    },
    {
      "epoch": 1.7641228029920608,
      "grad_norm": 0.48541536927223206,
      "learning_rate": 3.7621797942757223e-06,
      "loss": 0.3735,
      "step": 38442
    },
    {
      "epoch": 1.7641686934973153,
      "grad_norm": 0.4728507399559021,
      "learning_rate": 3.7619422401809265e-06,
      "loss": 0.3731,
      "step": 38443
    },
    {
      "epoch": 1.7642145840025698,
      "grad_norm": 0.48088783025741577,
      "learning_rate": 3.7617046890632686e-06,
      "loss": 0.375,
      "step": 38444
    },
    {
      "epoch": 1.7642604745078243,
      "grad_norm": 0.5236520171165466,
      "learning_rate": 3.76146714092332e-06,
      "loss": 0.3701,
      "step": 38445
    },
    {
      "epoch": 1.7643063650130788,
      "grad_norm": 0.5070548057556152,
      "learning_rate": 3.7612295957616503e-06,
      "loss": 0.411,
      "step": 38446
    },
    {
      "epoch": 1.7643522555183333,
      "grad_norm": 0.4297960102558136,
      "learning_rate": 3.7609920535788343e-06,
      "loss": 0.3047,
      "step": 38447
    },
    {
      "epoch": 1.7643981460235878,
      "grad_norm": 0.4578765630722046,
      "learning_rate": 3.7607545143754414e-06,
      "loss": 0.3043,
      "step": 38448
    },
    {
      "epoch": 1.7644440365288423,
      "grad_norm": 0.4711192846298218,
      "learning_rate": 3.7605169781520418e-06,
      "loss": 0.3905,
      "step": 38449
    },
    {
      "epoch": 1.7644899270340968,
      "grad_norm": 0.43182337284088135,
      "learning_rate": 3.7602794449092088e-06,
      "loss": 0.3116,
      "step": 38450
    },
    {
      "epoch": 1.764535817539351,
      "grad_norm": 0.4665738046169281,
      "learning_rate": 3.760041914647512e-06,
      "loss": 0.3858,
      "step": 38451
    },
    {
      "epoch": 1.7645817080446056,
      "grad_norm": 0.5155542492866516,
      "learning_rate": 3.7598043873675216e-06,
      "loss": 0.4002,
      "step": 38452
    },
    {
      "epoch": 1.76462759854986,
      "grad_norm": 0.5507570505142212,
      "learning_rate": 3.759566863069812e-06,
      "loss": 0.3141,
      "step": 38453
    },
    {
      "epoch": 1.7646734890551143,
      "grad_norm": 0.4773794710636139,
      "learning_rate": 3.7593293417549526e-06,
      "loss": 0.3884,
      "step": 38454
    },
    {
      "epoch": 1.7647193795603688,
      "grad_norm": 0.513459324836731,
      "learning_rate": 3.759091823423514e-06,
      "loss": 0.3295,
      "step": 38455
    },
    {
      "epoch": 1.7647652700656233,
      "grad_norm": 0.4660806357860565,
      "learning_rate": 3.7588543080760686e-06,
      "loss": 0.3717,
      "step": 38456
    },
    {
      "epoch": 1.7648111605708778,
      "grad_norm": 0.4965137541294098,
      "learning_rate": 3.7586167957131864e-06,
      "loss": 0.3504,
      "step": 38457
    },
    {
      "epoch": 1.7648570510761323,
      "grad_norm": 0.43501007556915283,
      "learning_rate": 3.7583792863354372e-06,
      "loss": 0.2893,
      "step": 38458
    },
    {
      "epoch": 1.7649029415813868,
      "grad_norm": 0.45114362239837646,
      "learning_rate": 3.758141779943396e-06,
      "loss": 0.3087,
      "step": 38459
    },
    {
      "epoch": 1.7649488320866413,
      "grad_norm": 0.5181567668914795,
      "learning_rate": 3.7579042765376337e-06,
      "loss": 0.4398,
      "step": 38460
    },
    {
      "epoch": 1.7649947225918958,
      "grad_norm": 0.4726133644580841,
      "learning_rate": 3.7576667761187157e-06,
      "loss": 0.4069,
      "step": 38461
    },
    {
      "epoch": 1.7650406130971503,
      "grad_norm": 0.4695392847061157,
      "learning_rate": 3.757429278687219e-06,
      "loss": 0.3408,
      "step": 38462
    },
    {
      "epoch": 1.7650865036024048,
      "grad_norm": 0.4348408877849579,
      "learning_rate": 3.7571917842437135e-06,
      "loss": 0.33,
      "step": 38463
    },
    {
      "epoch": 1.765132394107659,
      "grad_norm": 0.4362618923187256,
      "learning_rate": 3.756954292788768e-06,
      "loss": 0.326,
      "step": 38464
    },
    {
      "epoch": 1.7651782846129136,
      "grad_norm": 0.43866854906082153,
      "learning_rate": 3.7567168043229562e-06,
      "loss": 0.3238,
      "step": 38465
    },
    {
      "epoch": 1.765224175118168,
      "grad_norm": 0.4723884165287018,
      "learning_rate": 3.756479318846848e-06,
      "loss": 0.3629,
      "step": 38466
    },
    {
      "epoch": 1.7652700656234224,
      "grad_norm": 0.47802889347076416,
      "learning_rate": 3.756241836361012e-06,
      "loss": 0.4092,
      "step": 38467
    },
    {
      "epoch": 1.7653159561286769,
      "grad_norm": 0.5006721019744873,
      "learning_rate": 3.756004356866025e-06,
      "loss": 0.3728,
      "step": 38468
    },
    {
      "epoch": 1.7653618466339314,
      "grad_norm": 0.4914933145046234,
      "learning_rate": 3.755766880362454e-06,
      "loss": 0.4036,
      "step": 38469
    },
    {
      "epoch": 1.7654077371391859,
      "grad_norm": 0.42574670910835266,
      "learning_rate": 3.75552940685087e-06,
      "loss": 0.326,
      "step": 38470
    },
    {
      "epoch": 1.7654536276444404,
      "grad_norm": 0.4239595830440521,
      "learning_rate": 3.755291936331846e-06,
      "loss": 0.2976,
      "step": 38471
    },
    {
      "epoch": 1.7654995181496949,
      "grad_norm": 0.49394649267196655,
      "learning_rate": 3.755054468805953e-06,
      "loss": 0.4333,
      "step": 38472
    },
    {
      "epoch": 1.7655454086549494,
      "grad_norm": 0.47833678126335144,
      "learning_rate": 3.754817004273758e-06,
      "loss": 0.3767,
      "step": 38473
    },
    {
      "epoch": 1.7655912991602039,
      "grad_norm": 0.4767918884754181,
      "learning_rate": 3.754579542735838e-06,
      "loss": 0.3681,
      "step": 38474
    },
    {
      "epoch": 1.7656371896654584,
      "grad_norm": 0.441020667552948,
      "learning_rate": 3.75434208419276e-06,
      "loss": 0.3175,
      "step": 38475
    },
    {
      "epoch": 1.7656830801707126,
      "grad_norm": 0.4671199917793274,
      "learning_rate": 3.754104628645098e-06,
      "loss": 0.3389,
      "step": 38476
    },
    {
      "epoch": 1.7657289706759671,
      "grad_norm": 0.4706002473831177,
      "learning_rate": 3.753867176093418e-06,
      "loss": 0.3395,
      "step": 38477
    },
    {
      "epoch": 1.7657748611812216,
      "grad_norm": 0.4754515588283539,
      "learning_rate": 3.7536297265382966e-06,
      "loss": 0.3471,
      "step": 38478
    },
    {
      "epoch": 1.7658207516864761,
      "grad_norm": 0.5018705725669861,
      "learning_rate": 3.7533922799803017e-06,
      "loss": 0.4064,
      "step": 38479
    },
    {
      "epoch": 1.7658666421917304,
      "grad_norm": 0.47097817063331604,
      "learning_rate": 3.7531548364200034e-06,
      "loss": 0.374,
      "step": 38480
    },
    {
      "epoch": 1.765912532696985,
      "grad_norm": 0.5272670388221741,
      "learning_rate": 3.7529173958579755e-06,
      "loss": 0.4541,
      "step": 38481
    },
    {
      "epoch": 1.7659584232022394,
      "grad_norm": 0.49096012115478516,
      "learning_rate": 3.752679958294789e-06,
      "loss": 0.3458,
      "step": 38482
    },
    {
      "epoch": 1.766004313707494,
      "grad_norm": 0.4396880269050598,
      "learning_rate": 3.7524425237310107e-06,
      "loss": 0.2898,
      "step": 38483
    },
    {
      "epoch": 1.7660502042127484,
      "grad_norm": 0.485205739736557,
      "learning_rate": 3.7522050921672167e-06,
      "loss": 0.3428,
      "step": 38484
    },
    {
      "epoch": 1.7660960947180029,
      "grad_norm": 0.49671176075935364,
      "learning_rate": 3.7519676636039747e-06,
      "loss": 0.3844,
      "step": 38485
    },
    {
      "epoch": 1.7661419852232574,
      "grad_norm": 0.46306127309799194,
      "learning_rate": 3.7517302380418564e-06,
      "loss": 0.3151,
      "step": 38486
    },
    {
      "epoch": 1.7661878757285119,
      "grad_norm": 0.46388280391693115,
      "learning_rate": 3.7514928154814335e-06,
      "loss": 0.3477,
      "step": 38487
    },
    {
      "epoch": 1.7662337662337664,
      "grad_norm": 0.47771385312080383,
      "learning_rate": 3.7512553959232766e-06,
      "loss": 0.3615,
      "step": 38488
    },
    {
      "epoch": 1.7662796567390207,
      "grad_norm": 0.4718706011772156,
      "learning_rate": 3.751017979367954e-06,
      "loss": 0.3745,
      "step": 38489
    },
    {
      "epoch": 1.7663255472442752,
      "grad_norm": 0.45186829566955566,
      "learning_rate": 3.750780565816041e-06,
      "loss": 0.2889,
      "step": 38490
    },
    {
      "epoch": 1.7663714377495296,
      "grad_norm": 0.4666984975337982,
      "learning_rate": 3.750543155268107e-06,
      "loss": 0.3121,
      "step": 38491
    },
    {
      "epoch": 1.766417328254784,
      "grad_norm": 0.48314914107322693,
      "learning_rate": 3.7503057477247206e-06,
      "loss": 0.3662,
      "step": 38492
    },
    {
      "epoch": 1.7664632187600384,
      "grad_norm": 0.4488626718521118,
      "learning_rate": 3.750068343186456e-06,
      "loss": 0.3248,
      "step": 38493
    },
    {
      "epoch": 1.766509109265293,
      "grad_norm": 0.46268755197525024,
      "learning_rate": 3.7498309416538823e-06,
      "loss": 0.3524,
      "step": 38494
    },
    {
      "epoch": 1.7665549997705474,
      "grad_norm": 0.4513900578022003,
      "learning_rate": 3.7495935431275687e-06,
      "loss": 0.3227,
      "step": 38495
    },
    {
      "epoch": 1.766600890275802,
      "grad_norm": 0.4730130732059479,
      "learning_rate": 3.7493561476080904e-06,
      "loss": 0.3669,
      "step": 38496
    },
    {
      "epoch": 1.7666467807810564,
      "grad_norm": 0.43243369460105896,
      "learning_rate": 3.7491187550960156e-06,
      "loss": 0.3078,
      "step": 38497
    },
    {
      "epoch": 1.766692671286311,
      "grad_norm": 0.43778154253959656,
      "learning_rate": 3.748881365591914e-06,
      "loss": 0.3274,
      "step": 38498
    },
    {
      "epoch": 1.7667385617915654,
      "grad_norm": 0.5096520781517029,
      "learning_rate": 3.7486439790963593e-06,
      "loss": 0.3694,
      "step": 38499
    },
    {
      "epoch": 1.76678445229682,
      "grad_norm": 0.4443760812282562,
      "learning_rate": 3.7484065956099208e-06,
      "loss": 0.3358,
      "step": 38500
    },
    {
      "epoch": 1.7668303428020744,
      "grad_norm": 0.466034471988678,
      "learning_rate": 3.7481692151331673e-06,
      "loss": 0.3333,
      "step": 38501
    },
    {
      "epoch": 1.7668762333073287,
      "grad_norm": 0.5084569454193115,
      "learning_rate": 3.747931837666674e-06,
      "loss": 0.4001,
      "step": 38502
    },
    {
      "epoch": 1.7669221238125832,
      "grad_norm": 0.4870623052120209,
      "learning_rate": 3.747694463211009e-06,
      "loss": 0.3799,
      "step": 38503
    },
    {
      "epoch": 1.7669680143178377,
      "grad_norm": 0.41490522027015686,
      "learning_rate": 3.747457091766743e-06,
      "loss": 0.2732,
      "step": 38504
    },
    {
      "epoch": 1.767013904823092,
      "grad_norm": 0.43872538208961487,
      "learning_rate": 3.747219723334448e-06,
      "loss": 0.3114,
      "step": 38505
    },
    {
      "epoch": 1.7670597953283464,
      "grad_norm": 0.4772000014781952,
      "learning_rate": 3.7469823579146945e-06,
      "loss": 0.3621,
      "step": 38506
    },
    {
      "epoch": 1.767105685833601,
      "grad_norm": 0.5230588912963867,
      "learning_rate": 3.746744995508052e-06,
      "loss": 0.4489,
      "step": 38507
    },
    {
      "epoch": 1.7671515763388554,
      "grad_norm": 0.46467316150665283,
      "learning_rate": 3.7465076361150934e-06,
      "loss": 0.3333,
      "step": 38508
    },
    {
      "epoch": 1.76719746684411,
      "grad_norm": 0.4688573181629181,
      "learning_rate": 3.7462702797363883e-06,
      "loss": 0.313,
      "step": 38509
    },
    {
      "epoch": 1.7672433573493644,
      "grad_norm": 0.4833219349384308,
      "learning_rate": 3.746032926372507e-06,
      "loss": 0.3831,
      "step": 38510
    },
    {
      "epoch": 1.767289247854619,
      "grad_norm": 0.4736151099205017,
      "learning_rate": 3.7457955760240187e-06,
      "loss": 0.3825,
      "step": 38511
    },
    {
      "epoch": 1.7673351383598734,
      "grad_norm": 0.45076894760131836,
      "learning_rate": 3.745558228691498e-06,
      "loss": 0.3298,
      "step": 38512
    },
    {
      "epoch": 1.767381028865128,
      "grad_norm": 0.4703231751918793,
      "learning_rate": 3.745320884375515e-06,
      "loss": 0.3525,
      "step": 38513
    },
    {
      "epoch": 1.7674269193703824,
      "grad_norm": 0.4730297923088074,
      "learning_rate": 3.7450835430766375e-06,
      "loss": 0.3508,
      "step": 38514
    },
    {
      "epoch": 1.7674728098756367,
      "grad_norm": 0.4682832360267639,
      "learning_rate": 3.7448462047954393e-06,
      "loss": 0.3498,
      "step": 38515
    },
    {
      "epoch": 1.7675187003808912,
      "grad_norm": 0.45303937792778015,
      "learning_rate": 3.7446088695324894e-06,
      "loss": 0.3689,
      "step": 38516
    },
    {
      "epoch": 1.7675645908861457,
      "grad_norm": 0.4568033516407013,
      "learning_rate": 3.744371537288357e-06,
      "loss": 0.3396,
      "step": 38517
    },
    {
      "epoch": 1.7676104813914,
      "grad_norm": 0.47557491064071655,
      "learning_rate": 3.744134208063617e-06,
      "loss": 0.3846,
      "step": 38518
    },
    {
      "epoch": 1.7676563718966545,
      "grad_norm": 0.4927780330181122,
      "learning_rate": 3.7438968818588383e-06,
      "loss": 0.3953,
      "step": 38519
    },
    {
      "epoch": 1.767702262401909,
      "grad_norm": 0.45397743582725525,
      "learning_rate": 3.743659558674589e-06,
      "loss": 0.3453,
      "step": 38520
    },
    {
      "epoch": 1.7677481529071635,
      "grad_norm": 0.48580628633499146,
      "learning_rate": 3.7434222385114444e-06,
      "loss": 0.3895,
      "step": 38521
    },
    {
      "epoch": 1.767794043412418,
      "grad_norm": 0.5023244619369507,
      "learning_rate": 3.7431849213699724e-06,
      "loss": 0.3733,
      "step": 38522
    },
    {
      "epoch": 1.7678399339176725,
      "grad_norm": 0.4831523299217224,
      "learning_rate": 3.742947607250741e-06,
      "loss": 0.4029,
      "step": 38523
    },
    {
      "epoch": 1.767885824422927,
      "grad_norm": 0.5166612267494202,
      "learning_rate": 3.7427102961543267e-06,
      "loss": 0.413,
      "step": 38524
    },
    {
      "epoch": 1.7679317149281815,
      "grad_norm": 0.4437985420227051,
      "learning_rate": 3.742472988081297e-06,
      "loss": 0.309,
      "step": 38525
    },
    {
      "epoch": 1.767977605433436,
      "grad_norm": 0.44597673416137695,
      "learning_rate": 3.742235683032222e-06,
      "loss": 0.3313,
      "step": 38526
    },
    {
      "epoch": 1.7680234959386902,
      "grad_norm": 0.4579927623271942,
      "learning_rate": 3.7419983810076744e-06,
      "loss": 0.3483,
      "step": 38527
    },
    {
      "epoch": 1.7680693864439447,
      "grad_norm": 0.5150508284568787,
      "learning_rate": 3.741761082008223e-06,
      "loss": 0.3084,
      "step": 38528
    },
    {
      "epoch": 1.7681152769491992,
      "grad_norm": 0.4623219668865204,
      "learning_rate": 3.741523786034439e-06,
      "loss": 0.3807,
      "step": 38529
    },
    {
      "epoch": 1.7681611674544535,
      "grad_norm": 0.4210550785064697,
      "learning_rate": 3.741286493086893e-06,
      "loss": 0.28,
      "step": 38530
    },
    {
      "epoch": 1.768207057959708,
      "grad_norm": 0.4599069356918335,
      "learning_rate": 3.741049203166157e-06,
      "loss": 0.36,
      "step": 38531
    },
    {
      "epoch": 1.7682529484649625,
      "grad_norm": 0.4746827781200409,
      "learning_rate": 3.7408119162727974e-06,
      "loss": 0.3854,
      "step": 38532
    },
    {
      "epoch": 1.768298838970217,
      "grad_norm": 0.49421611428260803,
      "learning_rate": 3.74057463240739e-06,
      "loss": 0.349,
      "step": 38533
    },
    {
      "epoch": 1.7683447294754715,
      "grad_norm": 0.48385533690452576,
      "learning_rate": 3.7403373515705034e-06,
      "loss": 0.3301,
      "step": 38534
    },
    {
      "epoch": 1.768390619980726,
      "grad_norm": 0.5149655342102051,
      "learning_rate": 3.7401000737627067e-06,
      "loss": 0.4547,
      "step": 38535
    },
    {
      "epoch": 1.7684365104859805,
      "grad_norm": 0.5138331055641174,
      "learning_rate": 3.739862798984573e-06,
      "loss": 0.4158,
      "step": 38536
    },
    {
      "epoch": 1.768482400991235,
      "grad_norm": 0.4657612144947052,
      "learning_rate": 3.739625527236671e-06,
      "loss": 0.313,
      "step": 38537
    },
    {
      "epoch": 1.7685282914964895,
      "grad_norm": 0.48610982298851013,
      "learning_rate": 3.7393882585195697e-06,
      "loss": 0.3434,
      "step": 38538
    },
    {
      "epoch": 1.768574182001744,
      "grad_norm": 0.45144006609916687,
      "learning_rate": 3.7391509928338445e-06,
      "loss": 0.3643,
      "step": 38539
    },
    {
      "epoch": 1.7686200725069983,
      "grad_norm": 0.49678999185562134,
      "learning_rate": 3.7389137301800626e-06,
      "loss": 0.4023,
      "step": 38540
    },
    {
      "epoch": 1.7686659630122528,
      "grad_norm": 0.5130404233932495,
      "learning_rate": 3.738676470558795e-06,
      "loss": 0.4507,
      "step": 38541
    },
    {
      "epoch": 1.7687118535175073,
      "grad_norm": 0.4708603620529175,
      "learning_rate": 3.738439213970612e-06,
      "loss": 0.3293,
      "step": 38542
    },
    {
      "epoch": 1.7687577440227615,
      "grad_norm": 0.4638685882091522,
      "learning_rate": 3.738201960416086e-06,
      "loss": 0.312,
      "step": 38543
    },
    {
      "epoch": 1.768803634528016,
      "grad_norm": 0.47997885942459106,
      "learning_rate": 3.737964709895785e-06,
      "loss": 0.3685,
      "step": 38544
    },
    {
      "epoch": 1.7688495250332705,
      "grad_norm": 0.5013919472694397,
      "learning_rate": 3.7377274624102787e-06,
      "loss": 0.4057,
      "step": 38545
    },
    {
      "epoch": 1.768895415538525,
      "grad_norm": 0.5012370944023132,
      "learning_rate": 3.7374902179601413e-06,
      "loss": 0.3857,
      "step": 38546
    },
    {
      "epoch": 1.7689413060437795,
      "grad_norm": 0.4549535810947418,
      "learning_rate": 3.737252976545942e-06,
      "loss": 0.3459,
      "step": 38547
    },
    {
      "epoch": 1.768987196549034,
      "grad_norm": 0.4928508996963501,
      "learning_rate": 3.7370157381682496e-06,
      "loss": 0.3592,
      "step": 38548
    },
    {
      "epoch": 1.7690330870542885,
      "grad_norm": 0.5145472288131714,
      "learning_rate": 3.7367785028276373e-06,
      "loss": 0.3304,
      "step": 38549
    },
    {
      "epoch": 1.769078977559543,
      "grad_norm": 0.49751579761505127,
      "learning_rate": 3.7365412705246735e-06,
      "loss": 0.4007,
      "step": 38550
    },
    {
      "epoch": 1.7691248680647975,
      "grad_norm": 0.504457950592041,
      "learning_rate": 3.7363040412599284e-06,
      "loss": 0.3367,
      "step": 38551
    },
    {
      "epoch": 1.769170758570052,
      "grad_norm": 0.5629599690437317,
      "learning_rate": 3.7360668150339733e-06,
      "loss": 0.4776,
      "step": 38552
    },
    {
      "epoch": 1.7692166490753063,
      "grad_norm": 0.46189168095588684,
      "learning_rate": 3.7358295918473796e-06,
      "loss": 0.3324,
      "step": 38553
    },
    {
      "epoch": 1.7692625395805608,
      "grad_norm": 0.452391654253006,
      "learning_rate": 3.7355923717007146e-06,
      "loss": 0.3348,
      "step": 38554
    },
    {
      "epoch": 1.7693084300858153,
      "grad_norm": 0.528039276599884,
      "learning_rate": 3.7353551545945528e-06,
      "loss": 0.4218,
      "step": 38555
    },
    {
      "epoch": 1.7693543205910696,
      "grad_norm": 0.4705066680908203,
      "learning_rate": 3.7351179405294624e-06,
      "loss": 0.359,
      "step": 38556
    },
    {
      "epoch": 1.769400211096324,
      "grad_norm": 0.4617433547973633,
      "learning_rate": 3.7348807295060136e-06,
      "loss": 0.3767,
      "step": 38557
    },
    {
      "epoch": 1.7694461016015786,
      "grad_norm": 0.45232275128364563,
      "learning_rate": 3.734643521524778e-06,
      "loss": 0.2715,
      "step": 38558
    },
    {
      "epoch": 1.769491992106833,
      "grad_norm": 0.4697965681552887,
      "learning_rate": 3.7344063165863258e-06,
      "loss": 0.35,
      "step": 38559
    },
    {
      "epoch": 1.7695378826120876,
      "grad_norm": 0.4364967942237854,
      "learning_rate": 3.7341691146912244e-06,
      "loss": 0.2956,
      "step": 38560
    },
    {
      "epoch": 1.769583773117342,
      "grad_norm": 0.4612088203430176,
      "learning_rate": 3.7339319158400494e-06,
      "loss": 0.3802,
      "step": 38561
    },
    {
      "epoch": 1.7696296636225965,
      "grad_norm": 0.4626777470111847,
      "learning_rate": 3.7336947200333683e-06,
      "loss": 0.327,
      "step": 38562
    },
    {
      "epoch": 1.769675554127851,
      "grad_norm": 0.4789682924747467,
      "learning_rate": 3.733457527271751e-06,
      "loss": 0.3553,
      "step": 38563
    },
    {
      "epoch": 1.7697214446331055,
      "grad_norm": 0.5410568118095398,
      "learning_rate": 3.733220337555769e-06,
      "loss": 0.4301,
      "step": 38564
    },
    {
      "epoch": 1.7697673351383598,
      "grad_norm": 0.49810272455215454,
      "learning_rate": 3.732983150885993e-06,
      "loss": 0.4071,
      "step": 38565
    },
    {
      "epoch": 1.7698132256436143,
      "grad_norm": 0.4807017147541046,
      "learning_rate": 3.7327459672629896e-06,
      "loss": 0.3663,
      "step": 38566
    },
    {
      "epoch": 1.7698591161488688,
      "grad_norm": 0.5166071057319641,
      "learning_rate": 3.7325087866873348e-06,
      "loss": 0.3873,
      "step": 38567
    },
    {
      "epoch": 1.7699050066541233,
      "grad_norm": 0.44316667318344116,
      "learning_rate": 3.7322716091595967e-06,
      "loss": 0.312,
      "step": 38568
    },
    {
      "epoch": 1.7699508971593776,
      "grad_norm": 0.4966508746147156,
      "learning_rate": 3.732034434680344e-06,
      "loss": 0.3535,
      "step": 38569
    },
    {
      "epoch": 1.769996787664632,
      "grad_norm": 0.45816951990127563,
      "learning_rate": 3.731797263250149e-06,
      "loss": 0.373,
      "step": 38570
    },
    {
      "epoch": 1.7700426781698866,
      "grad_norm": 0.43862634897232056,
      "learning_rate": 3.7315600948695817e-06,
      "loss": 0.3178,
      "step": 38571
    },
    {
      "epoch": 1.770088568675141,
      "grad_norm": 0.4678722321987152,
      "learning_rate": 3.7313229295392095e-06,
      "loss": 0.3589,
      "step": 38572
    },
    {
      "epoch": 1.7701344591803956,
      "grad_norm": 0.4796691834926605,
      "learning_rate": 3.731085767259609e-06,
      "loss": 0.3839,
      "step": 38573
    },
    {
      "epoch": 1.77018034968565,
      "grad_norm": 0.5083514451980591,
      "learning_rate": 3.7308486080313454e-06,
      "loss": 0.3955,
      "step": 38574
    },
    {
      "epoch": 1.7702262401909046,
      "grad_norm": 0.4831717610359192,
      "learning_rate": 3.730611451854991e-06,
      "loss": 0.3205,
      "step": 38575
    },
    {
      "epoch": 1.770272130696159,
      "grad_norm": 0.5091385245323181,
      "learning_rate": 3.730374298731112e-06,
      "loss": 0.418,
      "step": 38576
    },
    {
      "epoch": 1.7703180212014136,
      "grad_norm": 0.47544386982917786,
      "learning_rate": 3.7301371486602856e-06,
      "loss": 0.3693,
      "step": 38577
    },
    {
      "epoch": 1.7703639117066678,
      "grad_norm": 0.49333903193473816,
      "learning_rate": 3.7299000016430775e-06,
      "loss": 0.3388,
      "step": 38578
    },
    {
      "epoch": 1.7704098022119223,
      "grad_norm": 0.456682950258255,
      "learning_rate": 3.7296628576800585e-06,
      "loss": 0.3318,
      "step": 38579
    },
    {
      "epoch": 1.7704556927171768,
      "grad_norm": 0.4832608997821808,
      "learning_rate": 3.7294257167718003e-06,
      "loss": 0.3696,
      "step": 38580
    },
    {
      "epoch": 1.7705015832224311,
      "grad_norm": 0.5153594017028809,
      "learning_rate": 3.729188578918872e-06,
      "loss": 0.4137,
      "step": 38581
    },
    {
      "epoch": 1.7705474737276856,
      "grad_norm": 0.48623257875442505,
      "learning_rate": 3.728951444121842e-06,
      "loss": 0.3985,
      "step": 38582
    },
    {
      "epoch": 1.7705933642329401,
      "grad_norm": 0.471584677696228,
      "learning_rate": 3.7287143123812853e-06,
      "loss": 0.3419,
      "step": 38583
    },
    {
      "epoch": 1.7706392547381946,
      "grad_norm": 0.48362842202186584,
      "learning_rate": 3.7284771836977695e-06,
      "loss": 0.363,
      "step": 38584
    },
    {
      "epoch": 1.770685145243449,
      "grad_norm": 0.46947935223579407,
      "learning_rate": 3.728240058071863e-06,
      "loss": 0.3579,
      "step": 38585
    },
    {
      "epoch": 1.7707310357487036,
      "grad_norm": 0.5046108365058899,
      "learning_rate": 3.728002935504139e-06,
      "loss": 0.3986,
      "step": 38586
    },
    {
      "epoch": 1.770776926253958,
      "grad_norm": 0.4774002134799957,
      "learning_rate": 3.727765815995167e-06,
      "loss": 0.3508,
      "step": 38587
    },
    {
      "epoch": 1.7708228167592126,
      "grad_norm": 0.45514386892318726,
      "learning_rate": 3.727528699545514e-06,
      "loss": 0.324,
      "step": 38588
    },
    {
      "epoch": 1.770868707264467,
      "grad_norm": 0.45323365926742554,
      "learning_rate": 3.727291586155756e-06,
      "loss": 0.3402,
      "step": 38589
    },
    {
      "epoch": 1.7709145977697216,
      "grad_norm": 0.4561525285243988,
      "learning_rate": 3.727054475826459e-06,
      "loss": 0.3366,
      "step": 38590
    },
    {
      "epoch": 1.7709604882749759,
      "grad_norm": 0.4826277494430542,
      "learning_rate": 3.7268173685581934e-06,
      "loss": 0.3291,
      "step": 38591
    },
    {
      "epoch": 1.7710063787802304,
      "grad_norm": 0.4709374010562897,
      "learning_rate": 3.7265802643515314e-06,
      "loss": 0.3579,
      "step": 38592
    },
    {
      "epoch": 1.7710522692854849,
      "grad_norm": 0.43209028244018555,
      "learning_rate": 3.7263431632070413e-06,
      "loss": 0.3096,
      "step": 38593
    },
    {
      "epoch": 1.7710981597907391,
      "grad_norm": 0.4100930392742157,
      "learning_rate": 3.7261060651252923e-06,
      "loss": 0.2351,
      "step": 38594
    },
    {
      "epoch": 1.7711440502959936,
      "grad_norm": 0.4591045379638672,
      "learning_rate": 3.7258689701068596e-06,
      "loss": 0.3585,
      "step": 38595
    },
    {
      "epoch": 1.7711899408012481,
      "grad_norm": 0.477417916059494,
      "learning_rate": 3.7256318781523086e-06,
      "loss": 0.3769,
      "step": 38596
    },
    {
      "epoch": 1.7712358313065026,
      "grad_norm": 0.5884917974472046,
      "learning_rate": 3.7253947892622088e-06,
      "loss": 0.4087,
      "step": 38597
    },
    {
      "epoch": 1.7712817218117571,
      "grad_norm": 0.45911407470703125,
      "learning_rate": 3.7251577034371346e-06,
      "loss": 0.3406,
      "step": 38598
    },
    {
      "epoch": 1.7713276123170116,
      "grad_norm": 0.39112111926078796,
      "learning_rate": 3.7249206206776534e-06,
      "loss": 0.253,
      "step": 38599
    },
    {
      "epoch": 1.7713735028222661,
      "grad_norm": 0.5277090072631836,
      "learning_rate": 3.724683540984334e-06,
      "loss": 0.4292,
      "step": 38600
    },
    {
      "epoch": 1.7714193933275206,
      "grad_norm": 0.46726396679878235,
      "learning_rate": 3.7244464643577506e-06,
      "loss": 0.3346,
      "step": 38601
    },
    {
      "epoch": 1.7714652838327751,
      "grad_norm": 0.48324766755104065,
      "learning_rate": 3.72420939079847e-06,
      "loss": 0.313,
      "step": 38602
    },
    {
      "epoch": 1.7715111743380296,
      "grad_norm": 0.4578716456890106,
      "learning_rate": 3.723972320307061e-06,
      "loss": 0.3533,
      "step": 38603
    },
    {
      "epoch": 1.771557064843284,
      "grad_norm": 0.48574161529541016,
      "learning_rate": 3.7237352528840983e-06,
      "loss": 0.3971,
      "step": 38604
    },
    {
      "epoch": 1.7716029553485384,
      "grad_norm": 0.44975003600120544,
      "learning_rate": 3.7234981885301492e-06,
      "loss": 0.3302,
      "step": 38605
    },
    {
      "epoch": 1.771648845853793,
      "grad_norm": 0.4717950224876404,
      "learning_rate": 3.723261127245784e-06,
      "loss": 0.3288,
      "step": 38606
    },
    {
      "epoch": 1.7716947363590472,
      "grad_norm": 0.4624135196208954,
      "learning_rate": 3.723024069031572e-06,
      "loss": 0.3412,
      "step": 38607
    },
    {
      "epoch": 1.7717406268643017,
      "grad_norm": 0.4835161566734314,
      "learning_rate": 3.7227870138880856e-06,
      "loss": 0.4036,
      "step": 38608
    },
    {
      "epoch": 1.7717865173695562,
      "grad_norm": 0.5001975893974304,
      "learning_rate": 3.722549961815893e-06,
      "loss": 0.3976,
      "step": 38609
    },
    {
      "epoch": 1.7718324078748107,
      "grad_norm": 0.4461921453475952,
      "learning_rate": 3.722312912815562e-06,
      "loss": 0.3312,
      "step": 38610
    },
    {
      "epoch": 1.7718782983800652,
      "grad_norm": 0.48167791962623596,
      "learning_rate": 3.722075866887668e-06,
      "loss": 0.3514,
      "step": 38611
    },
    {
      "epoch": 1.7719241888853197,
      "grad_norm": 0.4678392708301544,
      "learning_rate": 3.721838824032778e-06,
      "loss": 0.3124,
      "step": 38612
    },
    {
      "epoch": 1.7719700793905742,
      "grad_norm": 0.49929267168045044,
      "learning_rate": 3.7216017842514606e-06,
      "loss": 0.4381,
      "step": 38613
    },
    {
      "epoch": 1.7720159698958287,
      "grad_norm": 0.4935516119003296,
      "learning_rate": 3.721364747544289e-06,
      "loss": 0.3584,
      "step": 38614
    },
    {
      "epoch": 1.7720618604010832,
      "grad_norm": 0.4555259644985199,
      "learning_rate": 3.7211277139118317e-06,
      "loss": 0.3354,
      "step": 38615
    },
    {
      "epoch": 1.7721077509063374,
      "grad_norm": 0.47761666774749756,
      "learning_rate": 3.7208906833546565e-06,
      "loss": 0.3878,
      "step": 38616
    },
    {
      "epoch": 1.772153641411592,
      "grad_norm": 0.4073365032672882,
      "learning_rate": 3.7206536558733388e-06,
      "loss": 0.2379,
      "step": 38617
    },
    {
      "epoch": 1.7721995319168464,
      "grad_norm": 0.5379689931869507,
      "learning_rate": 3.7204166314684442e-06,
      "loss": 0.4588,
      "step": 38618
    },
    {
      "epoch": 1.7722454224221007,
      "grad_norm": 0.5229946970939636,
      "learning_rate": 3.720179610140541e-06,
      "loss": 0.4252,
      "step": 38619
    },
    {
      "epoch": 1.7722913129273552,
      "grad_norm": 0.5249634385108948,
      "learning_rate": 3.7199425918902043e-06,
      "loss": 0.4164,
      "step": 38620
    },
    {
      "epoch": 1.7723372034326097,
      "grad_norm": 0.4545176923274994,
      "learning_rate": 3.7197055767180023e-06,
      "loss": 0.3598,
      "step": 38621
    },
    {
      "epoch": 1.7723830939378642,
      "grad_norm": 0.45977604389190674,
      "learning_rate": 3.7194685646245026e-06,
      "loss": 0.3393,
      "step": 38622
    },
    {
      "epoch": 1.7724289844431187,
      "grad_norm": 0.46226322650909424,
      "learning_rate": 3.719231555610278e-06,
      "loss": 0.3447,
      "step": 38623
    },
    {
      "epoch": 1.7724748749483732,
      "grad_norm": 0.4780139625072479,
      "learning_rate": 3.718994549675897e-06,
      "loss": 0.3284,
      "step": 38624
    },
    {
      "epoch": 1.7725207654536277,
      "grad_norm": 0.48105674982070923,
      "learning_rate": 3.7187575468219284e-06,
      "loss": 0.3685,
      "step": 38625
    },
    {
      "epoch": 1.7725666559588822,
      "grad_norm": 0.4751576781272888,
      "learning_rate": 3.7185205470489455e-06,
      "loss": 0.3606,
      "step": 38626
    },
    {
      "epoch": 1.7726125464641367,
      "grad_norm": 0.45569533109664917,
      "learning_rate": 3.7182835503575164e-06,
      "loss": 0.367,
      "step": 38627
    },
    {
      "epoch": 1.7726584369693912,
      "grad_norm": 0.49231454730033875,
      "learning_rate": 3.7180465567482093e-06,
      "loss": 0.4382,
      "step": 38628
    },
    {
      "epoch": 1.7727043274746455,
      "grad_norm": 0.46188369393348694,
      "learning_rate": 3.717809566221597e-06,
      "loss": 0.3188,
      "step": 38629
    },
    {
      "epoch": 1.7727502179799,
      "grad_norm": 0.47314122319221497,
      "learning_rate": 3.717572578778248e-06,
      "loss": 0.3766,
      "step": 38630
    },
    {
      "epoch": 1.7727961084851545,
      "grad_norm": 0.46679896116256714,
      "learning_rate": 3.7173355944187296e-06,
      "loss": 0.3345,
      "step": 38631
    },
    {
      "epoch": 1.7728419989904087,
      "grad_norm": 0.4729922413825989,
      "learning_rate": 3.717098613143617e-06,
      "loss": 0.3463,
      "step": 38632
    },
    {
      "epoch": 1.7728878894956632,
      "grad_norm": 0.4638116657733917,
      "learning_rate": 3.716861634953477e-06,
      "loss": 0.3232,
      "step": 38633
    },
    {
      "epoch": 1.7729337800009177,
      "grad_norm": 0.5012961626052856,
      "learning_rate": 3.7166246598488785e-06,
      "loss": 0.4239,
      "step": 38634
    },
    {
      "epoch": 1.7729796705061722,
      "grad_norm": 0.47806546092033386,
      "learning_rate": 3.716387687830394e-06,
      "loss": 0.3657,
      "step": 38635
    },
    {
      "epoch": 1.7730255610114267,
      "grad_norm": 0.4350389540195465,
      "learning_rate": 3.7161507188985925e-06,
      "loss": 0.3039,
      "step": 38636
    },
    {
      "epoch": 1.7730714515166812,
      "grad_norm": 0.44631901383399963,
      "learning_rate": 3.715913753054041e-06,
      "loss": 0.3075,
      "step": 38637
    },
    {
      "epoch": 1.7731173420219357,
      "grad_norm": 0.478247731924057,
      "learning_rate": 3.7156767902973134e-06,
      "loss": 0.291,
      "step": 38638
    },
    {
      "epoch": 1.7731632325271902,
      "grad_norm": 0.4301947057247162,
      "learning_rate": 3.7154398306289786e-06,
      "loss": 0.2781,
      "step": 38639
    },
    {
      "epoch": 1.7732091230324447,
      "grad_norm": 0.4958086609840393,
      "learning_rate": 3.715202874049605e-06,
      "loss": 0.3881,
      "step": 38640
    },
    {
      "epoch": 1.7732550135376992,
      "grad_norm": 0.4955841302871704,
      "learning_rate": 3.714965920559761e-06,
      "loss": 0.4484,
      "step": 38641
    },
    {
      "epoch": 1.7733009040429535,
      "grad_norm": 0.48201343417167664,
      "learning_rate": 3.71472897016002e-06,
      "loss": 0.3414,
      "step": 38642
    },
    {
      "epoch": 1.773346794548208,
      "grad_norm": 0.43385186791419983,
      "learning_rate": 3.7144920228509507e-06,
      "loss": 0.2703,
      "step": 38643
    },
    {
      "epoch": 1.7733926850534625,
      "grad_norm": 0.44730186462402344,
      "learning_rate": 3.714255078633121e-06,
      "loss": 0.3163,
      "step": 38644
    },
    {
      "epoch": 1.7734385755587168,
      "grad_norm": 0.45939165353775024,
      "learning_rate": 3.714018137507103e-06,
      "loss": 0.3636,
      "step": 38645
    },
    {
      "epoch": 1.7734844660639713,
      "grad_norm": 0.4454156160354614,
      "learning_rate": 3.7137811994734656e-06,
      "loss": 0.3249,
      "step": 38646
    },
    {
      "epoch": 1.7735303565692258,
      "grad_norm": 0.4642527997493744,
      "learning_rate": 3.7135442645327767e-06,
      "loss": 0.3777,
      "step": 38647
    },
    {
      "epoch": 1.7735762470744802,
      "grad_norm": 0.5166807174682617,
      "learning_rate": 3.7133073326856096e-06,
      "loss": 0.3876,
      "step": 38648
    },
    {
      "epoch": 1.7736221375797347,
      "grad_norm": 0.5027990341186523,
      "learning_rate": 3.713070403932532e-06,
      "loss": 0.3554,
      "step": 38649
    },
    {
      "epoch": 1.7736680280849892,
      "grad_norm": 0.4712843596935272,
      "learning_rate": 3.7128334782741134e-06,
      "loss": 0.3671,
      "step": 38650
    },
    {
      "epoch": 1.7737139185902437,
      "grad_norm": 0.48860350251197815,
      "learning_rate": 3.7125965557109246e-06,
      "loss": 0.4057,
      "step": 38651
    },
    {
      "epoch": 1.7737598090954982,
      "grad_norm": 0.41781896352767944,
      "learning_rate": 3.7123596362435355e-06,
      "loss": 0.276,
      "step": 38652
    },
    {
      "epoch": 1.7738056996007527,
      "grad_norm": 0.45310860872268677,
      "learning_rate": 3.7121227198725125e-06,
      "loss": 0.363,
      "step": 38653
    },
    {
      "epoch": 1.773851590106007,
      "grad_norm": 0.45763465762138367,
      "learning_rate": 3.7118858065984296e-06,
      "loss": 0.3304,
      "step": 38654
    },
    {
      "epoch": 1.7738974806112615,
      "grad_norm": 0.4800237715244293,
      "learning_rate": 3.7116488964218555e-06,
      "loss": 0.3594,
      "step": 38655
    },
    {
      "epoch": 1.773943371116516,
      "grad_norm": 0.4406096041202545,
      "learning_rate": 3.7114119893433574e-06,
      "loss": 0.3069,
      "step": 38656
    },
    {
      "epoch": 1.7739892616217705,
      "grad_norm": 0.499500036239624,
      "learning_rate": 3.7111750853635087e-06,
      "loss": 0.4048,
      "step": 38657
    },
    {
      "epoch": 1.7740351521270248,
      "grad_norm": 0.47643256187438965,
      "learning_rate": 3.710938184482877e-06,
      "loss": 0.3282,
      "step": 38658
    },
    {
      "epoch": 1.7740810426322793,
      "grad_norm": 0.4817681312561035,
      "learning_rate": 3.71070128670203e-06,
      "loss": 0.3941,
      "step": 38659
    },
    {
      "epoch": 1.7741269331375338,
      "grad_norm": 0.44607314467430115,
      "learning_rate": 3.710464392021541e-06,
      "loss": 0.2939,
      "step": 38660
    },
    {
      "epoch": 1.7741728236427883,
      "grad_norm": 0.464729905128479,
      "learning_rate": 3.7102275004419795e-06,
      "loss": 0.3481,
      "step": 38661
    },
    {
      "epoch": 1.7742187141480428,
      "grad_norm": 0.44852015376091003,
      "learning_rate": 3.7099906119639108e-06,
      "loss": 0.3166,
      "step": 38662
    },
    {
      "epoch": 1.7742646046532973,
      "grad_norm": 0.45701828598976135,
      "learning_rate": 3.7097537265879096e-06,
      "loss": 0.371,
      "step": 38663
    },
    {
      "epoch": 1.7743104951585518,
      "grad_norm": 0.4807978868484497,
      "learning_rate": 3.7095168443145435e-06,
      "loss": 0.3828,
      "step": 38664
    },
    {
      "epoch": 1.7743563856638063,
      "grad_norm": 0.47495120763778687,
      "learning_rate": 3.7092799651443813e-06,
      "loss": 0.376,
      "step": 38665
    },
    {
      "epoch": 1.7744022761690608,
      "grad_norm": 0.44755518436431885,
      "learning_rate": 3.709043089077994e-06,
      "loss": 0.3507,
      "step": 38666
    },
    {
      "epoch": 1.774448166674315,
      "grad_norm": 0.43564754724502563,
      "learning_rate": 3.7088062161159504e-06,
      "loss": 0.268,
      "step": 38667
    },
    {
      "epoch": 1.7744940571795695,
      "grad_norm": 0.4905352294445038,
      "learning_rate": 3.708569346258819e-06,
      "loss": 0.4,
      "step": 38668
    },
    {
      "epoch": 1.774539947684824,
      "grad_norm": 0.4904656410217285,
      "learning_rate": 3.708332479507172e-06,
      "loss": 0.4405,
      "step": 38669
    },
    {
      "epoch": 1.7745858381900783,
      "grad_norm": 0.4862385392189026,
      "learning_rate": 3.708095615861578e-06,
      "loss": 0.3361,
      "step": 38670
    },
    {
      "epoch": 1.7746317286953328,
      "grad_norm": 0.45290136337280273,
      "learning_rate": 3.7078587553226057e-06,
      "loss": 0.3321,
      "step": 38671
    },
    {
      "epoch": 1.7746776192005873,
      "grad_norm": 0.45138660073280334,
      "learning_rate": 3.7076218978908262e-06,
      "loss": 0.3043,
      "step": 38672
    },
    {
      "epoch": 1.7747235097058418,
      "grad_norm": 0.4899376630783081,
      "learning_rate": 3.707385043566808e-06,
      "loss": 0.3472,
      "step": 38673
    },
    {
      "epoch": 1.7747694002110963,
      "grad_norm": 0.4827486276626587,
      "learning_rate": 3.707148192351121e-06,
      "loss": 0.3601,
      "step": 38674
    },
    {
      "epoch": 1.7748152907163508,
      "grad_norm": 0.4980613589286804,
      "learning_rate": 3.7069113442443323e-06,
      "loss": 0.3874,
      "step": 38675
    },
    {
      "epoch": 1.7748611812216053,
      "grad_norm": 0.4982292056083679,
      "learning_rate": 3.7066744992470155e-06,
      "loss": 0.416,
      "step": 38676
    },
    {
      "epoch": 1.7749070717268598,
      "grad_norm": 0.4814682900905609,
      "learning_rate": 3.7064376573597386e-06,
      "loss": 0.3767,
      "step": 38677
    },
    {
      "epoch": 1.7749529622321143,
      "grad_norm": 0.4303734004497528,
      "learning_rate": 3.70620081858307e-06,
      "loss": 0.2975,
      "step": 38678
    },
    {
      "epoch": 1.7749988527373688,
      "grad_norm": 0.48041343688964844,
      "learning_rate": 3.705963982917581e-06,
      "loss": 0.3469,
      "step": 38679
    },
    {
      "epoch": 1.775044743242623,
      "grad_norm": 0.46200627088546753,
      "learning_rate": 3.7057271503638404e-06,
      "loss": 0.3197,
      "step": 38680
    },
    {
      "epoch": 1.7750906337478776,
      "grad_norm": 0.44478148221969604,
      "learning_rate": 3.7054903209224146e-06,
      "loss": 0.3108,
      "step": 38681
    },
    {
      "epoch": 1.775136524253132,
      "grad_norm": 0.5038337707519531,
      "learning_rate": 3.705253494593879e-06,
      "loss": 0.4357,
      "step": 38682
    },
    {
      "epoch": 1.7751824147583863,
      "grad_norm": 0.5064499378204346,
      "learning_rate": 3.7050166713788017e-06,
      "loss": 0.4007,
      "step": 38683
    },
    {
      "epoch": 1.7752283052636408,
      "grad_norm": 0.4801899492740631,
      "learning_rate": 3.7047798512777463e-06,
      "loss": 0.3398,
      "step": 38684
    },
    {
      "epoch": 1.7752741957688953,
      "grad_norm": 0.4947572946548462,
      "learning_rate": 3.704543034291289e-06,
      "loss": 0.4237,
      "step": 38685
    },
    {
      "epoch": 1.7753200862741498,
      "grad_norm": 0.5049558281898499,
      "learning_rate": 3.7043062204199966e-06,
      "loss": 0.3945,
      "step": 38686
    },
    {
      "epoch": 1.7753659767794043,
      "grad_norm": 0.4876173138618469,
      "learning_rate": 3.704069409664438e-06,
      "loss": 0.4194,
      "step": 38687
    },
    {
      "epoch": 1.7754118672846588,
      "grad_norm": 0.4788006842136383,
      "learning_rate": 3.7038326020251847e-06,
      "loss": 0.3779,
      "step": 38688
    },
    {
      "epoch": 1.7754577577899133,
      "grad_norm": 0.4422045052051544,
      "learning_rate": 3.7035957975028046e-06,
      "loss": 0.3291,
      "step": 38689
    },
    {
      "epoch": 1.7755036482951678,
      "grad_norm": 0.4834446310997009,
      "learning_rate": 3.7033589960978657e-06,
      "loss": 0.3616,
      "step": 38690
    },
    {
      "epoch": 1.7755495388004223,
      "grad_norm": 0.4987007975578308,
      "learning_rate": 3.703122197810941e-06,
      "loss": 0.4333,
      "step": 38691
    },
    {
      "epoch": 1.7755954293056768,
      "grad_norm": 0.5029178261756897,
      "learning_rate": 3.7028854026425985e-06,
      "loss": 0.3655,
      "step": 38692
    },
    {
      "epoch": 1.775641319810931,
      "grad_norm": 0.48140189051628113,
      "learning_rate": 3.7026486105934057e-06,
      "loss": 0.3916,
      "step": 38693
    },
    {
      "epoch": 1.7756872103161856,
      "grad_norm": 0.45967769622802734,
      "learning_rate": 3.702411821663935e-06,
      "loss": 0.3664,
      "step": 38694
    },
    {
      "epoch": 1.77573310082144,
      "grad_norm": 0.4670740067958832,
      "learning_rate": 3.7021750358547537e-06,
      "loss": 0.3758,
      "step": 38695
    },
    {
      "epoch": 1.7757789913266944,
      "grad_norm": 0.4445522129535675,
      "learning_rate": 3.7019382531664304e-06,
      "loss": 0.313,
      "step": 38696
    },
    {
      "epoch": 1.7758248818319489,
      "grad_norm": 0.4579584300518036,
      "learning_rate": 3.7017014735995378e-06,
      "loss": 0.3576,
      "step": 38697
    },
    {
      "epoch": 1.7758707723372034,
      "grad_norm": 0.4726782739162445,
      "learning_rate": 3.701464697154643e-06,
      "loss": 0.3711,
      "step": 38698
    },
    {
      "epoch": 1.7759166628424579,
      "grad_norm": 0.44019168615341187,
      "learning_rate": 3.7012279238323157e-06,
      "loss": 0.3156,
      "step": 38699
    },
    {
      "epoch": 1.7759625533477124,
      "grad_norm": 0.49122607707977295,
      "learning_rate": 3.7009911536331256e-06,
      "loss": 0.408,
      "step": 38700
    },
    {
      "epoch": 1.7760084438529669,
      "grad_norm": 0.48665308952331543,
      "learning_rate": 3.700754386557642e-06,
      "loss": 0.3942,
      "step": 38701
    },
    {
      "epoch": 1.7760543343582214,
      "grad_norm": 0.48522692918777466,
      "learning_rate": 3.700517622606432e-06,
      "loss": 0.4371,
      "step": 38702
    },
    {
      "epoch": 1.7761002248634759,
      "grad_norm": 0.4556894302368164,
      "learning_rate": 3.700280861780069e-06,
      "loss": 0.3491,
      "step": 38703
    },
    {
      "epoch": 1.7761461153687303,
      "grad_norm": 0.4846763014793396,
      "learning_rate": 3.70004410407912e-06,
      "loss": 0.3573,
      "step": 38704
    },
    {
      "epoch": 1.7761920058739846,
      "grad_norm": 0.4335854649543762,
      "learning_rate": 3.6998073495041566e-06,
      "loss": 0.285,
      "step": 38705
    },
    {
      "epoch": 1.7762378963792391,
      "grad_norm": 0.4490160644054413,
      "learning_rate": 3.6995705980557415e-06,
      "loss": 0.3462,
      "step": 38706
    },
    {
      "epoch": 1.7762837868844936,
      "grad_norm": 0.4872885048389435,
      "learning_rate": 3.6993338497344523e-06,
      "loss": 0.4052,
      "step": 38707
    },
    {
      "epoch": 1.776329677389748,
      "grad_norm": 0.484502375125885,
      "learning_rate": 3.699097104540854e-06,
      "loss": 0.3651,
      "step": 38708
    },
    {
      "epoch": 1.7763755678950024,
      "grad_norm": 0.4643760919570923,
      "learning_rate": 3.698860362475516e-06,
      "loss": 0.3484,
      "step": 38709
    },
    {
      "epoch": 1.776421458400257,
      "grad_norm": 0.4756908118724823,
      "learning_rate": 3.6986236235390084e-06,
      "loss": 0.3923,
      "step": 38710
    },
    {
      "epoch": 1.7764673489055114,
      "grad_norm": 0.45895692706108093,
      "learning_rate": 3.698386887731901e-06,
      "loss": 0.3717,
      "step": 38711
    },
    {
      "epoch": 1.7765132394107659,
      "grad_norm": 0.4976475238800049,
      "learning_rate": 3.69815015505476e-06,
      "loss": 0.4033,
      "step": 38712
    },
    {
      "epoch": 1.7765591299160204,
      "grad_norm": 0.5010587573051453,
      "learning_rate": 3.6979134255081593e-06,
      "loss": 0.3719,
      "step": 38713
    },
    {
      "epoch": 1.7766050204212749,
      "grad_norm": 0.4220401346683502,
      "learning_rate": 3.6976766990926655e-06,
      "loss": 0.2854,
      "step": 38714
    },
    {
      "epoch": 1.7766509109265294,
      "grad_norm": 0.4740554988384247,
      "learning_rate": 3.6974399758088477e-06,
      "loss": 0.2831,
      "step": 38715
    },
    {
      "epoch": 1.7766968014317839,
      "grad_norm": 0.5042394995689392,
      "learning_rate": 3.6972032556572755e-06,
      "loss": 0.4602,
      "step": 38716
    },
    {
      "epoch": 1.7767426919370384,
      "grad_norm": 0.5015743970870972,
      "learning_rate": 3.6969665386385196e-06,
      "loss": 0.4087,
      "step": 38717
    },
    {
      "epoch": 1.7767885824422927,
      "grad_norm": 0.6407734751701355,
      "learning_rate": 3.696729824753145e-06,
      "loss": 0.3711,
      "step": 38718
    },
    {
      "epoch": 1.7768344729475471,
      "grad_norm": 0.43201109766960144,
      "learning_rate": 3.696493114001727e-06,
      "loss": 0.3092,
      "step": 38719
    },
    {
      "epoch": 1.7768803634528016,
      "grad_norm": 0.5038405656814575,
      "learning_rate": 3.6962564063848305e-06,
      "loss": 0.3617,
      "step": 38720
    },
    {
      "epoch": 1.776926253958056,
      "grad_norm": 0.5605885982513428,
      "learning_rate": 3.6960197019030253e-06,
      "loss": 0.4122,
      "step": 38721
    },
    {
      "epoch": 1.7769721444633104,
      "grad_norm": 0.4641634225845337,
      "learning_rate": 3.695783000556882e-06,
      "loss": 0.324,
      "step": 38722
    },
    {
      "epoch": 1.777018034968565,
      "grad_norm": 0.5137197971343994,
      "learning_rate": 3.6955463023469697e-06,
      "loss": 0.4215,
      "step": 38723
    },
    {
      "epoch": 1.7770639254738194,
      "grad_norm": 0.45366814732551575,
      "learning_rate": 3.6953096072738536e-06,
      "loss": 0.3015,
      "step": 38724
    },
    {
      "epoch": 1.777109815979074,
      "grad_norm": 0.4421251118183136,
      "learning_rate": 3.6950729153381094e-06,
      "loss": 0.3037,
      "step": 38725
    },
    {
      "epoch": 1.7771557064843284,
      "grad_norm": 0.48795151710510254,
      "learning_rate": 3.694836226540302e-06,
      "loss": 0.3866,
      "step": 38726
    },
    {
      "epoch": 1.777201596989583,
      "grad_norm": 0.47505733370780945,
      "learning_rate": 3.694599540881001e-06,
      "loss": 0.3639,
      "step": 38727
    },
    {
      "epoch": 1.7772474874948374,
      "grad_norm": 0.46536684036254883,
      "learning_rate": 3.694362858360777e-06,
      "loss": 0.3741,
      "step": 38728
    },
    {
      "epoch": 1.777293378000092,
      "grad_norm": 0.47876566648483276,
      "learning_rate": 3.6941261789801986e-06,
      "loss": 0.3818,
      "step": 38729
    },
    {
      "epoch": 1.7773392685053464,
      "grad_norm": 0.4812789559364319,
      "learning_rate": 3.693889502739833e-06,
      "loss": 0.3834,
      "step": 38730
    },
    {
      "epoch": 1.7773851590106007,
      "grad_norm": 0.4632786214351654,
      "learning_rate": 3.6936528296402525e-06,
      "loss": 0.3392,
      "step": 38731
    },
    {
      "epoch": 1.7774310495158552,
      "grad_norm": 0.5048632621765137,
      "learning_rate": 3.693416159682024e-06,
      "loss": 0.4023,
      "step": 38732
    },
    {
      "epoch": 1.7774769400211097,
      "grad_norm": 0.5229439735412598,
      "learning_rate": 3.693179492865716e-06,
      "loss": 0.4856,
      "step": 38733
    },
    {
      "epoch": 1.777522830526364,
      "grad_norm": 0.459413081407547,
      "learning_rate": 3.6929428291919e-06,
      "loss": 0.3966,
      "step": 38734
    },
    {
      "epoch": 1.7775687210316184,
      "grad_norm": 0.4646683633327484,
      "learning_rate": 3.692706168661145e-06,
      "loss": 0.3507,
      "step": 38735
    },
    {
      "epoch": 1.777614611536873,
      "grad_norm": 0.4666522145271301,
      "learning_rate": 3.692469511274017e-06,
      "loss": 0.3485,
      "step": 38736
    },
    {
      "epoch": 1.7776605020421274,
      "grad_norm": 0.4557129144668579,
      "learning_rate": 3.6922328570310896e-06,
      "loss": 0.3367,
      "step": 38737
    },
    {
      "epoch": 1.777706392547382,
      "grad_norm": 0.4737633168697357,
      "learning_rate": 3.6919962059329285e-06,
      "loss": 0.3506,
      "step": 38738
    },
    {
      "epoch": 1.7777522830526364,
      "grad_norm": 0.44450145959854126,
      "learning_rate": 3.6917595579801036e-06,
      "loss": 0.2819,
      "step": 38739
    },
    {
      "epoch": 1.777798173557891,
      "grad_norm": 0.47432535886764526,
      "learning_rate": 3.6915229131731824e-06,
      "loss": 0.3699,
      "step": 38740
    },
    {
      "epoch": 1.7778440640631454,
      "grad_norm": 0.509697437286377,
      "learning_rate": 3.6912862715127374e-06,
      "loss": 0.4015,
      "step": 38741
    },
    {
      "epoch": 1.7778899545684,
      "grad_norm": 0.5004681348800659,
      "learning_rate": 3.6910496329993358e-06,
      "loss": 0.3721,
      "step": 38742
    },
    {
      "epoch": 1.7779358450736542,
      "grad_norm": 0.45762088894844055,
      "learning_rate": 3.6908129976335454e-06,
      "loss": 0.3351,
      "step": 38743
    },
    {
      "epoch": 1.7779817355789087,
      "grad_norm": 0.4737938940525055,
      "learning_rate": 3.690576365415938e-06,
      "loss": 0.3768,
      "step": 38744
    },
    {
      "epoch": 1.7780276260841632,
      "grad_norm": 0.5007230043411255,
      "learning_rate": 3.6903397363470817e-06,
      "loss": 0.4158,
      "step": 38745
    },
    {
      "epoch": 1.7780735165894177,
      "grad_norm": 0.4866727292537689,
      "learning_rate": 3.690103110427542e-06,
      "loss": 0.3705,
      "step": 38746
    },
    {
      "epoch": 1.778119407094672,
      "grad_norm": 0.5128588080406189,
      "learning_rate": 3.689866487657894e-06,
      "loss": 0.4548,
      "step": 38747
    },
    {
      "epoch": 1.7781652975999265,
      "grad_norm": 0.467977911233902,
      "learning_rate": 3.6896298680387023e-06,
      "loss": 0.3722,
      "step": 38748
    },
    {
      "epoch": 1.778211188105181,
      "grad_norm": 0.48930954933166504,
      "learning_rate": 3.6893932515705372e-06,
      "loss": 0.403,
      "step": 38749
    },
    {
      "epoch": 1.7782570786104355,
      "grad_norm": 0.44622647762298584,
      "learning_rate": 3.689156638253968e-06,
      "loss": 0.321,
      "step": 38750
    },
    {
      "epoch": 1.77830296911569,
      "grad_norm": 0.4731053411960602,
      "learning_rate": 3.688920028089564e-06,
      "loss": 0.3479,
      "step": 38751
    },
    {
      "epoch": 1.7783488596209445,
      "grad_norm": 0.4504857063293457,
      "learning_rate": 3.688683421077892e-06,
      "loss": 0.344,
      "step": 38752
    },
    {
      "epoch": 1.778394750126199,
      "grad_norm": 0.4367230236530304,
      "learning_rate": 3.6884468172195243e-06,
      "loss": 0.2951,
      "step": 38753
    },
    {
      "epoch": 1.7784406406314535,
      "grad_norm": 0.46515780687332153,
      "learning_rate": 3.6882102165150273e-06,
      "loss": 0.339,
      "step": 38754
    },
    {
      "epoch": 1.778486531136708,
      "grad_norm": 0.5011507272720337,
      "learning_rate": 3.687973618964969e-06,
      "loss": 0.3815,
      "step": 38755
    },
    {
      "epoch": 1.7785324216419622,
      "grad_norm": 0.4553022086620331,
      "learning_rate": 3.687737024569922e-06,
      "loss": 0.3238,
      "step": 38756
    },
    {
      "epoch": 1.7785783121472167,
      "grad_norm": 0.49520811438560486,
      "learning_rate": 3.687500433330453e-06,
      "loss": 0.4362,
      "step": 38757
    },
    {
      "epoch": 1.7786242026524712,
      "grad_norm": 0.48133385181427,
      "learning_rate": 3.687263845247131e-06,
      "loss": 0.3839,
      "step": 38758
    },
    {
      "epoch": 1.7786700931577255,
      "grad_norm": 0.5521310567855835,
      "learning_rate": 3.687027260320526e-06,
      "loss": 0.4693,
      "step": 38759
    },
    {
      "epoch": 1.77871598366298,
      "grad_norm": 0.45161595940589905,
      "learning_rate": 3.6867906785512053e-06,
      "loss": 0.3386,
      "step": 38760
    },
    {
      "epoch": 1.7787618741682345,
      "grad_norm": 0.4778243601322174,
      "learning_rate": 3.686554099939737e-06,
      "loss": 0.3418,
      "step": 38761
    },
    {
      "epoch": 1.778807764673489,
      "grad_norm": 0.4270366132259369,
      "learning_rate": 3.6863175244866935e-06,
      "loss": 0.2846,
      "step": 38762
    },
    {
      "epoch": 1.7788536551787435,
      "grad_norm": 0.451988160610199,
      "learning_rate": 3.686080952192642e-06,
      "loss": 0.3292,
      "step": 38763
    },
    {
      "epoch": 1.778899545683998,
      "grad_norm": 0.46123743057250977,
      "learning_rate": 3.68584438305815e-06,
      "loss": 0.3764,
      "step": 38764
    },
    {
      "epoch": 1.7789454361892525,
      "grad_norm": 0.47436368465423584,
      "learning_rate": 3.6856078170837883e-06,
      "loss": 0.3247,
      "step": 38765
    },
    {
      "epoch": 1.778991326694507,
      "grad_norm": 0.46316099166870117,
      "learning_rate": 3.6853712542701248e-06,
      "loss": 0.2978,
      "step": 38766
    },
    {
      "epoch": 1.7790372171997615,
      "grad_norm": 0.4840412139892578,
      "learning_rate": 3.6851346946177267e-06,
      "loss": 0.3492,
      "step": 38767
    },
    {
      "epoch": 1.779083107705016,
      "grad_norm": 0.5125594139099121,
      "learning_rate": 3.6848981381271675e-06,
      "loss": 0.4203,
      "step": 38768
    },
    {
      "epoch": 1.7791289982102703,
      "grad_norm": 0.4330027997493744,
      "learning_rate": 3.6846615847990117e-06,
      "loss": 0.317,
      "step": 38769
    },
    {
      "epoch": 1.7791748887155248,
      "grad_norm": 0.4930709898471832,
      "learning_rate": 3.684425034633831e-06,
      "loss": 0.4334,
      "step": 38770
    },
    {
      "epoch": 1.7792207792207793,
      "grad_norm": 0.44742172956466675,
      "learning_rate": 3.6841884876321916e-06,
      "loss": 0.3254,
      "step": 38771
    },
    {
      "epoch": 1.7792666697260335,
      "grad_norm": 0.4393977224826813,
      "learning_rate": 3.6839519437946644e-06,
      "loss": 0.3074,
      "step": 38772
    },
    {
      "epoch": 1.779312560231288,
      "grad_norm": 0.4759564697742462,
      "learning_rate": 3.683715403121817e-06,
      "loss": 0.3845,
      "step": 38773
    },
    {
      "epoch": 1.7793584507365425,
      "grad_norm": 0.5070866346359253,
      "learning_rate": 3.6834788656142186e-06,
      "loss": 0.4476,
      "step": 38774
    },
    {
      "epoch": 1.779404341241797,
      "grad_norm": 0.5410648584365845,
      "learning_rate": 3.683242331272438e-06,
      "loss": 0.4626,
      "step": 38775
    },
    {
      "epoch": 1.7794502317470515,
      "grad_norm": 0.4763728380203247,
      "learning_rate": 3.6830058000970446e-06,
      "loss": 0.3513,
      "step": 38776
    },
    {
      "epoch": 1.779496122252306,
      "grad_norm": 0.47258642315864563,
      "learning_rate": 3.6827692720886044e-06,
      "loss": 0.3699,
      "step": 38777
    },
    {
      "epoch": 1.7795420127575605,
      "grad_norm": 0.43259844183921814,
      "learning_rate": 3.68253274724769e-06,
      "loss": 0.2787,
      "step": 38778
    },
    {
      "epoch": 1.779587903262815,
      "grad_norm": 0.4752890169620514,
      "learning_rate": 3.6822962255748695e-06,
      "loss": 0.3359,
      "step": 38779
    },
    {
      "epoch": 1.7796337937680695,
      "grad_norm": 0.4541025459766388,
      "learning_rate": 3.6820597070707085e-06,
      "loss": 0.3366,
      "step": 38780
    },
    {
      "epoch": 1.779679684273324,
      "grad_norm": 0.4203289747238159,
      "learning_rate": 3.68182319173578e-06,
      "loss": 0.311,
      "step": 38781
    },
    {
      "epoch": 1.7797255747785783,
      "grad_norm": 0.4727313816547394,
      "learning_rate": 3.68158667957065e-06,
      "loss": 0.3277,
      "step": 38782
    },
    {
      "epoch": 1.7797714652838328,
      "grad_norm": 0.44337737560272217,
      "learning_rate": 3.6813501705758866e-06,
      "loss": 0.3415,
      "step": 38783
    },
    {
      "epoch": 1.7798173557890873,
      "grad_norm": 0.4635860025882721,
      "learning_rate": 3.6811136647520613e-06,
      "loss": 0.3336,
      "step": 38784
    },
    {
      "epoch": 1.7798632462943416,
      "grad_norm": 0.4659208059310913,
      "learning_rate": 3.680877162099742e-06,
      "loss": 0.3785,
      "step": 38785
    },
    {
      "epoch": 1.779909136799596,
      "grad_norm": 0.49302011728286743,
      "learning_rate": 3.6806406626194948e-06,
      "loss": 0.4302,
      "step": 38786
    },
    {
      "epoch": 1.7799550273048506,
      "grad_norm": 0.4949454367160797,
      "learning_rate": 3.6804041663118915e-06,
      "loss": 0.3986,
      "step": 38787
    },
    {
      "epoch": 1.780000917810105,
      "grad_norm": 0.529241681098938,
      "learning_rate": 3.6801676731775003e-06,
      "loss": 0.4568,
      "step": 38788
    },
    {
      "epoch": 1.7800468083153596,
      "grad_norm": 0.4531041085720062,
      "learning_rate": 3.6799311832168873e-06,
      "loss": 0.3785,
      "step": 38789
    },
    {
      "epoch": 1.780092698820614,
      "grad_norm": 0.47543302178382874,
      "learning_rate": 3.679694696430625e-06,
      "loss": 0.3984,
      "step": 38790
    },
    {
      "epoch": 1.7801385893258685,
      "grad_norm": 0.5123245120048523,
      "learning_rate": 3.679458212819281e-06,
      "loss": 0.4844,
      "step": 38791
    },
    {
      "epoch": 1.780184479831123,
      "grad_norm": 0.41209524869918823,
      "learning_rate": 3.6792217323834212e-06,
      "loss": 0.2648,
      "step": 38792
    },
    {
      "epoch": 1.7802303703363775,
      "grad_norm": 0.48457223176956177,
      "learning_rate": 3.678985255123617e-06,
      "loss": 0.3697,
      "step": 38793
    },
    {
      "epoch": 1.7802762608416318,
      "grad_norm": 0.44528916478157043,
      "learning_rate": 3.6787487810404376e-06,
      "loss": 0.3151,
      "step": 38794
    },
    {
      "epoch": 1.7803221513468863,
      "grad_norm": 0.4946439266204834,
      "learning_rate": 3.678512310134448e-06,
      "loss": 0.3392,
      "step": 38795
    },
    {
      "epoch": 1.7803680418521408,
      "grad_norm": 0.4705406427383423,
      "learning_rate": 3.6782758424062216e-06,
      "loss": 0.3356,
      "step": 38796
    },
    {
      "epoch": 1.780413932357395,
      "grad_norm": 0.515501081943512,
      "learning_rate": 3.678039377856324e-06,
      "loss": 0.3195,
      "step": 38797
    },
    {
      "epoch": 1.7804598228626496,
      "grad_norm": 0.490554541349411,
      "learning_rate": 3.6778029164853225e-06,
      "loss": 0.3691,
      "step": 38798
    },
    {
      "epoch": 1.780505713367904,
      "grad_norm": 0.45429137349128723,
      "learning_rate": 3.67756645829379e-06,
      "loss": 0.2844,
      "step": 38799
    },
    {
      "epoch": 1.7805516038731586,
      "grad_norm": 0.49002280831336975,
      "learning_rate": 3.6773300032822924e-06,
      "loss": 0.4043,
      "step": 38800
    },
    {
      "epoch": 1.780597494378413,
      "grad_norm": 0.4559251368045807,
      "learning_rate": 3.6770935514513973e-06,
      "loss": 0.3291,
      "step": 38801
    },
    {
      "epoch": 1.7806433848836676,
      "grad_norm": 0.48697417974472046,
      "learning_rate": 3.6768571028016764e-06,
      "loss": 0.3625,
      "step": 38802
    },
    {
      "epoch": 1.780689275388922,
      "grad_norm": 0.4785788655281067,
      "learning_rate": 3.676620657333696e-06,
      "loss": 0.3294,
      "step": 38803
    },
    {
      "epoch": 1.7807351658941766,
      "grad_norm": 0.5466011762619019,
      "learning_rate": 3.6763842150480256e-06,
      "loss": 0.4518,
      "step": 38804
    },
    {
      "epoch": 1.780781056399431,
      "grad_norm": 0.48355022072792053,
      "learning_rate": 3.676147775945231e-06,
      "loss": 0.316,
      "step": 38805
    },
    {
      "epoch": 1.7808269469046856,
      "grad_norm": 0.49268755316734314,
      "learning_rate": 3.675911340025886e-06,
      "loss": 0.4082,
      "step": 38806
    },
    {
      "epoch": 1.7808728374099398,
      "grad_norm": 0.49337977170944214,
      "learning_rate": 3.6756749072905564e-06,
      "loss": 0.3769,
      "step": 38807
    },
    {
      "epoch": 1.7809187279151943,
      "grad_norm": 0.4707299470901489,
      "learning_rate": 3.675438477739809e-06,
      "loss": 0.3564,
      "step": 38808
    },
    {
      "epoch": 1.7809646184204488,
      "grad_norm": 0.4616657495498657,
      "learning_rate": 3.675202051374215e-06,
      "loss": 0.3357,
      "step": 38809
    },
    {
      "epoch": 1.7810105089257031,
      "grad_norm": 0.4482233226299286,
      "learning_rate": 3.6749656281943426e-06,
      "loss": 0.2876,
      "step": 38810
    },
    {
      "epoch": 1.7810563994309576,
      "grad_norm": 0.46304577589035034,
      "learning_rate": 3.6747292082007568e-06,
      "loss": 0.35,
      "step": 38811
    },
    {
      "epoch": 1.7811022899362121,
      "grad_norm": 0.49315279722213745,
      "learning_rate": 3.6744927913940308e-06,
      "loss": 0.3942,
      "step": 38812
    },
    {
      "epoch": 1.7811481804414666,
      "grad_norm": 0.4711878299713135,
      "learning_rate": 3.674256377774732e-06,
      "loss": 0.3207,
      "step": 38813
    },
    {
      "epoch": 1.781194070946721,
      "grad_norm": 0.40258875489234924,
      "learning_rate": 3.6740199673434274e-06,
      "loss": 0.2689,
      "step": 38814
    },
    {
      "epoch": 1.7812399614519756,
      "grad_norm": 0.4317987263202667,
      "learning_rate": 3.6737835601006868e-06,
      "loss": 0.3013,
      "step": 38815
    },
    {
      "epoch": 1.78128585195723,
      "grad_norm": 0.4454007148742676,
      "learning_rate": 3.673547156047078e-06,
      "loss": 0.3055,
      "step": 38816
    },
    {
      "epoch": 1.7813317424624846,
      "grad_norm": 0.4814406931400299,
      "learning_rate": 3.673310755183168e-06,
      "loss": 0.3774,
      "step": 38817
    },
    {
      "epoch": 1.781377632967739,
      "grad_norm": 0.4965106248855591,
      "learning_rate": 3.6730743575095306e-06,
      "loss": 0.4161,
      "step": 38818
    },
    {
      "epoch": 1.7814235234729936,
      "grad_norm": 0.4762606918811798,
      "learning_rate": 3.6728379630267285e-06,
      "loss": 0.3598,
      "step": 38819
    },
    {
      "epoch": 1.7814694139782479,
      "grad_norm": 0.5015295147895813,
      "learning_rate": 3.67260157173533e-06,
      "loss": 0.3975,
      "step": 38820
    },
    {
      "epoch": 1.7815153044835024,
      "grad_norm": 0.4930337965488434,
      "learning_rate": 3.6723651836359093e-06,
      "loss": 0.3986,
      "step": 38821
    },
    {
      "epoch": 1.7815611949887569,
      "grad_norm": 0.5053561925888062,
      "learning_rate": 3.67212879872903e-06,
      "loss": 0.3981,
      "step": 38822
    },
    {
      "epoch": 1.7816070854940111,
      "grad_norm": 0.47690486907958984,
      "learning_rate": 3.6718924170152613e-06,
      "loss": 0.3431,
      "step": 38823
    },
    {
      "epoch": 1.7816529759992656,
      "grad_norm": 0.44550928473472595,
      "learning_rate": 3.6716560384951733e-06,
      "loss": 0.3472,
      "step": 38824
    },
    {
      "epoch": 1.7816988665045201,
      "grad_norm": 0.4321156442165375,
      "learning_rate": 3.6714196631693333e-06,
      "loss": 0.3021,
      "step": 38825
    },
    {
      "epoch": 1.7817447570097746,
      "grad_norm": 0.49916934967041016,
      "learning_rate": 3.6711832910383072e-06,
      "loss": 0.3839,
      "step": 38826
    },
    {
      "epoch": 1.7817906475150291,
      "grad_norm": 0.5040802359580994,
      "learning_rate": 3.6709469221026685e-06,
      "loss": 0.4304,
      "step": 38827
    },
    {
      "epoch": 1.7818365380202836,
      "grad_norm": 0.4874526858329773,
      "learning_rate": 3.670710556362983e-06,
      "loss": 0.3768,
      "step": 38828
    },
    {
      "epoch": 1.7818824285255381,
      "grad_norm": 0.49648502469062805,
      "learning_rate": 3.6704741938198175e-06,
      "loss": 0.3865,
      "step": 38829
    },
    {
      "epoch": 1.7819283190307926,
      "grad_norm": 0.488092839717865,
      "learning_rate": 3.6702378344737433e-06,
      "loss": 0.3679,
      "step": 38830
    },
    {
      "epoch": 1.7819742095360471,
      "grad_norm": 0.43595221638679504,
      "learning_rate": 3.6700014783253272e-06,
      "loss": 0.2814,
      "step": 38831
    },
    {
      "epoch": 1.7820201000413014,
      "grad_norm": 0.5059211254119873,
      "learning_rate": 3.669765125375136e-06,
      "loss": 0.4126,
      "step": 38832
    },
    {
      "epoch": 1.782065990546556,
      "grad_norm": 0.4665108621120453,
      "learning_rate": 3.669528775623742e-06,
      "loss": 0.3476,
      "step": 38833
    },
    {
      "epoch": 1.7821118810518104,
      "grad_norm": 0.48190823197364807,
      "learning_rate": 3.6692924290717112e-06,
      "loss": 0.3727,
      "step": 38834
    },
    {
      "epoch": 1.782157771557065,
      "grad_norm": 0.4643876552581787,
      "learning_rate": 3.669056085719612e-06,
      "loss": 0.3117,
      "step": 38835
    },
    {
      "epoch": 1.7822036620623192,
      "grad_norm": 0.4390116035938263,
      "learning_rate": 3.6688197455680117e-06,
      "loss": 0.2814,
      "step": 38836
    },
    {
      "epoch": 1.7822495525675737,
      "grad_norm": 0.43688899278640747,
      "learning_rate": 3.668583408617481e-06,
      "loss": 0.3113,
      "step": 38837
    },
    {
      "epoch": 1.7822954430728282,
      "grad_norm": 0.4819587469100952,
      "learning_rate": 3.6683470748685874e-06,
      "loss": 0.3429,
      "step": 38838
    },
    {
      "epoch": 1.7823413335780827,
      "grad_norm": 0.4955873191356659,
      "learning_rate": 3.6681107443218968e-06,
      "loss": 0.4393,
      "step": 38839
    },
    {
      "epoch": 1.7823872240833372,
      "grad_norm": 0.4806351065635681,
      "learning_rate": 3.6678744169779817e-06,
      "loss": 0.369,
      "step": 38840
    },
    {
      "epoch": 1.7824331145885917,
      "grad_norm": 0.46430543065071106,
      "learning_rate": 3.6676380928374076e-06,
      "loss": 0.3599,
      "step": 38841
    },
    {
      "epoch": 1.7824790050938462,
      "grad_norm": 0.48802515864372253,
      "learning_rate": 3.6674017719007416e-06,
      "loss": 0.3782,
      "step": 38842
    },
    {
      "epoch": 1.7825248955991007,
      "grad_norm": 0.4700908064842224,
      "learning_rate": 3.6671654541685554e-06,
      "loss": 0.3611,
      "step": 38843
    },
    {
      "epoch": 1.7825707861043552,
      "grad_norm": 0.46627581119537354,
      "learning_rate": 3.666929139641416e-06,
      "loss": 0.3305,
      "step": 38844
    },
    {
      "epoch": 1.7826166766096094,
      "grad_norm": 0.48393192887306213,
      "learning_rate": 3.66669282831989e-06,
      "loss": 0.3422,
      "step": 38845
    },
    {
      "epoch": 1.782662567114864,
      "grad_norm": 0.47725164890289307,
      "learning_rate": 3.6664565202045477e-06,
      "loss": 0.4048,
      "step": 38846
    },
    {
      "epoch": 1.7827084576201184,
      "grad_norm": 0.4746381640434265,
      "learning_rate": 3.666220215295957e-06,
      "loss": 0.3328,
      "step": 38847
    },
    {
      "epoch": 1.7827543481253727,
      "grad_norm": 0.4609190821647644,
      "learning_rate": 3.665983913594684e-06,
      "loss": 0.3586,
      "step": 38848
    },
    {
      "epoch": 1.7828002386306272,
      "grad_norm": 0.467178076505661,
      "learning_rate": 3.6657476151013004e-06,
      "loss": 0.3801,
      "step": 38849
    },
    {
      "epoch": 1.7828461291358817,
      "grad_norm": 0.468646377325058,
      "learning_rate": 3.665511319816373e-06,
      "loss": 0.3179,
      "step": 38850
    },
    {
      "epoch": 1.7828920196411362,
      "grad_norm": 0.46147310733795166,
      "learning_rate": 3.665275027740468e-06,
      "loss": 0.3302,
      "step": 38851
    },
    {
      "epoch": 1.7829379101463907,
      "grad_norm": 0.46726518869400024,
      "learning_rate": 3.6650387388741564e-06,
      "loss": 0.3947,
      "step": 38852
    },
    {
      "epoch": 1.7829838006516452,
      "grad_norm": 0.4900877773761749,
      "learning_rate": 3.6648024532180055e-06,
      "loss": 0.3399,
      "step": 38853
    },
    {
      "epoch": 1.7830296911568997,
      "grad_norm": 0.5002977848052979,
      "learning_rate": 3.6645661707725812e-06,
      "loss": 0.4062,
      "step": 38854
    },
    {
      "epoch": 1.7830755816621542,
      "grad_norm": 0.4390370547771454,
      "learning_rate": 3.6643298915384562e-06,
      "loss": 0.3224,
      "step": 38855
    },
    {
      "epoch": 1.7831214721674087,
      "grad_norm": 0.4955928325653076,
      "learning_rate": 3.6640936155161955e-06,
      "loss": 0.3495,
      "step": 38856
    },
    {
      "epoch": 1.7831673626726632,
      "grad_norm": 0.5008420944213867,
      "learning_rate": 3.6638573427063677e-06,
      "loss": 0.389,
      "step": 38857
    },
    {
      "epoch": 1.7832132531779175,
      "grad_norm": 0.4896942675113678,
      "learning_rate": 3.663621073109542e-06,
      "loss": 0.3412,
      "step": 38858
    },
    {
      "epoch": 1.783259143683172,
      "grad_norm": 0.470429927110672,
      "learning_rate": 3.663384806726286e-06,
      "loss": 0.3435,
      "step": 38859
    },
    {
      "epoch": 1.7833050341884265,
      "grad_norm": 0.45611026883125305,
      "learning_rate": 3.6631485435571657e-06,
      "loss": 0.3995,
      "step": 38860
    },
    {
      "epoch": 1.7833509246936807,
      "grad_norm": 0.4429650902748108,
      "learning_rate": 3.662912283602753e-06,
      "loss": 0.3205,
      "step": 38861
    },
    {
      "epoch": 1.7833968151989352,
      "grad_norm": 0.4269215166568756,
      "learning_rate": 3.662676026863616e-06,
      "loss": 0.2734,
      "step": 38862
    },
    {
      "epoch": 1.7834427057041897,
      "grad_norm": 0.3998905420303345,
      "learning_rate": 3.662439773340317e-06,
      "loss": 0.2469,
      "step": 38863
    },
    {
      "epoch": 1.7834885962094442,
      "grad_norm": 0.45013442635536194,
      "learning_rate": 3.6622035230334307e-06,
      "loss": 0.3342,
      "step": 38864
    },
    {
      "epoch": 1.7835344867146987,
      "grad_norm": 0.45822131633758545,
      "learning_rate": 3.6619672759435234e-06,
      "loss": 0.3271,
      "step": 38865
    },
    {
      "epoch": 1.7835803772199532,
      "grad_norm": 0.4669106602668762,
      "learning_rate": 3.66173103207116e-06,
      "loss": 0.2837,
      "step": 38866
    },
    {
      "epoch": 1.7836262677252077,
      "grad_norm": 0.5053828954696655,
      "learning_rate": 3.661494791416913e-06,
      "loss": 0.3949,
      "step": 38867
    },
    {
      "epoch": 1.7836721582304622,
      "grad_norm": 0.4411364197731018,
      "learning_rate": 3.661258553981348e-06,
      "loss": 0.306,
      "step": 38868
    },
    {
      "epoch": 1.7837180487357167,
      "grad_norm": 0.4495849013328552,
      "learning_rate": 3.661022319765035e-06,
      "loss": 0.3022,
      "step": 38869
    },
    {
      "epoch": 1.7837639392409712,
      "grad_norm": 0.4716343879699707,
      "learning_rate": 3.6607860887685377e-06,
      "loss": 0.3653,
      "step": 38870
    },
    {
      "epoch": 1.7838098297462255,
      "grad_norm": 0.4861082434654236,
      "learning_rate": 3.6605498609924295e-06,
      "loss": 0.3693,
      "step": 38871
    },
    {
      "epoch": 1.78385572025148,
      "grad_norm": 0.5092144012451172,
      "learning_rate": 3.660313636437276e-06,
      "loss": 0.3975,
      "step": 38872
    },
    {
      "epoch": 1.7839016107567345,
      "grad_norm": 0.4512619376182556,
      "learning_rate": 3.6600774151036445e-06,
      "loss": 0.3345,
      "step": 38873
    },
    {
      "epoch": 1.7839475012619888,
      "grad_norm": 0.4386513829231262,
      "learning_rate": 3.6598411969921043e-06,
      "loss": 0.2944,
      "step": 38874
    },
    {
      "epoch": 1.7839933917672433,
      "grad_norm": 0.5139645338058472,
      "learning_rate": 3.6596049821032242e-06,
      "loss": 0.3791,
      "step": 38875
    },
    {
      "epoch": 1.7840392822724978,
      "grad_norm": 0.4300140142440796,
      "learning_rate": 3.659368770437568e-06,
      "loss": 0.2792,
      "step": 38876
    },
    {
      "epoch": 1.7840851727777522,
      "grad_norm": 0.48955217003822327,
      "learning_rate": 3.6591325619957098e-06,
      "loss": 0.4023,
      "step": 38877
    },
    {
      "epoch": 1.7841310632830067,
      "grad_norm": 0.46486377716064453,
      "learning_rate": 3.6588963567782138e-06,
      "loss": 0.3222,
      "step": 38878
    },
    {
      "epoch": 1.7841769537882612,
      "grad_norm": 0.47650933265686035,
      "learning_rate": 3.6586601547856484e-06,
      "loss": 0.4016,
      "step": 38879
    },
    {
      "epoch": 1.7842228442935157,
      "grad_norm": 0.45058494806289673,
      "learning_rate": 3.658423956018582e-06,
      "loss": 0.2922,
      "step": 38880
    },
    {
      "epoch": 1.7842687347987702,
      "grad_norm": 0.47090595960617065,
      "learning_rate": 3.6581877604775835e-06,
      "loss": 0.334,
      "step": 38881
    },
    {
      "epoch": 1.7843146253040247,
      "grad_norm": 0.4944796860218048,
      "learning_rate": 3.6579515681632174e-06,
      "loss": 0.423,
      "step": 38882
    },
    {
      "epoch": 1.784360515809279,
      "grad_norm": 0.44962990283966064,
      "learning_rate": 3.6577153790760565e-06,
      "loss": 0.3288,
      "step": 38883
    },
    {
      "epoch": 1.7844064063145335,
      "grad_norm": 0.5113868713378906,
      "learning_rate": 3.6574791932166676e-06,
      "loss": 0.3828,
      "step": 38884
    },
    {
      "epoch": 1.784452296819788,
      "grad_norm": 0.4317995607852936,
      "learning_rate": 3.657243010585614e-06,
      "loss": 0.3175,
      "step": 38885
    },
    {
      "epoch": 1.7844981873250423,
      "grad_norm": 0.4848055839538574,
      "learning_rate": 3.6570068311834692e-06,
      "loss": 0.369,
      "step": 38886
    },
    {
      "epoch": 1.7845440778302968,
      "grad_norm": 0.46412521600723267,
      "learning_rate": 3.6567706550108e-06,
      "loss": 0.3515,
      "step": 38887
    },
    {
      "epoch": 1.7845899683355513,
      "grad_norm": 0.4597375988960266,
      "learning_rate": 3.656534482068172e-06,
      "loss": 0.3519,
      "step": 38888
    },
    {
      "epoch": 1.7846358588408058,
      "grad_norm": 0.5018007755279541,
      "learning_rate": 3.6562983123561556e-06,
      "loss": 0.3984,
      "step": 38889
    },
    {
      "epoch": 1.7846817493460603,
      "grad_norm": 0.4584071934223175,
      "learning_rate": 3.6560621458753177e-06,
      "loss": 0.3693,
      "step": 38890
    },
    {
      "epoch": 1.7847276398513148,
      "grad_norm": 0.46777623891830444,
      "learning_rate": 3.655825982626223e-06,
      "loss": 0.3255,
      "step": 38891
    },
    {
      "epoch": 1.7847735303565693,
      "grad_norm": 0.4695948362350464,
      "learning_rate": 3.655589822609446e-06,
      "loss": 0.4168,
      "step": 38892
    },
    {
      "epoch": 1.7848194208618238,
      "grad_norm": 0.4416811466217041,
      "learning_rate": 3.6553536658255506e-06,
      "loss": 0.3242,
      "step": 38893
    },
    {
      "epoch": 1.7848653113670783,
      "grad_norm": 0.4720553755760193,
      "learning_rate": 3.655117512275105e-06,
      "loss": 0.3679,
      "step": 38894
    },
    {
      "epoch": 1.7849112018723328,
      "grad_norm": 0.4780040681362152,
      "learning_rate": 3.654881361958678e-06,
      "loss": 0.4092,
      "step": 38895
    },
    {
      "epoch": 1.784957092377587,
      "grad_norm": 0.4819720685482025,
      "learning_rate": 3.6546452148768363e-06,
      "loss": 0.3539,
      "step": 38896
    },
    {
      "epoch": 1.7850029828828415,
      "grad_norm": 0.4870342016220093,
      "learning_rate": 3.654409071030146e-06,
      "loss": 0.3948,
      "step": 38897
    },
    {
      "epoch": 1.785048873388096,
      "grad_norm": 0.49844029545783997,
      "learning_rate": 3.6541729304191796e-06,
      "loss": 0.4164,
      "step": 38898
    },
    {
      "epoch": 1.7850947638933503,
      "grad_norm": 0.44270724058151245,
      "learning_rate": 3.653936793044503e-06,
      "loss": 0.3331,
      "step": 38899
    },
    {
      "epoch": 1.7851406543986048,
      "grad_norm": 0.433416485786438,
      "learning_rate": 3.653700658906683e-06,
      "loss": 0.288,
      "step": 38900
    },
    {
      "epoch": 1.7851865449038593,
      "grad_norm": 0.5079478621482849,
      "learning_rate": 3.6534645280062874e-06,
      "loss": 0.3823,
      "step": 38901
    },
    {
      "epoch": 1.7852324354091138,
      "grad_norm": 0.50795978307724,
      "learning_rate": 3.6532284003438858e-06,
      "loss": 0.4769,
      "step": 38902
    },
    {
      "epoch": 1.7852783259143683,
      "grad_norm": 0.48330941796302795,
      "learning_rate": 3.6529922759200443e-06,
      "loss": 0.4197,
      "step": 38903
    },
    {
      "epoch": 1.7853242164196228,
      "grad_norm": 0.47169798612594604,
      "learning_rate": 3.652756154735329e-06,
      "loss": 0.3457,
      "step": 38904
    },
    {
      "epoch": 1.7853701069248773,
      "grad_norm": 0.4668859541416168,
      "learning_rate": 3.6525200367903123e-06,
      "loss": 0.3788,
      "step": 38905
    },
    {
      "epoch": 1.7854159974301318,
      "grad_norm": 0.5085064768791199,
      "learning_rate": 3.652283922085561e-06,
      "loss": 0.4408,
      "step": 38906
    },
    {
      "epoch": 1.7854618879353863,
      "grad_norm": 0.4649844467639923,
      "learning_rate": 3.6520478106216383e-06,
      "loss": 0.3241,
      "step": 38907
    },
    {
      "epoch": 1.7855077784406408,
      "grad_norm": 0.47638532519340515,
      "learning_rate": 3.6518117023991175e-06,
      "loss": 0.3679,
      "step": 38908
    },
    {
      "epoch": 1.785553668945895,
      "grad_norm": 0.453876256942749,
      "learning_rate": 3.6515755974185637e-06,
      "loss": 0.3212,
      "step": 38909
    },
    {
      "epoch": 1.7855995594511496,
      "grad_norm": 0.4389156699180603,
      "learning_rate": 3.651339495680544e-06,
      "loss": 0.3213,
      "step": 38910
    },
    {
      "epoch": 1.785645449956404,
      "grad_norm": 0.5007078051567078,
      "learning_rate": 3.651103397185628e-06,
      "loss": 0.3579,
      "step": 38911
    },
    {
      "epoch": 1.7856913404616583,
      "grad_norm": 0.42395341396331787,
      "learning_rate": 3.6508673019343827e-06,
      "loss": 0.2898,
      "step": 38912
    },
    {
      "epoch": 1.7857372309669128,
      "grad_norm": 0.47447890043258667,
      "learning_rate": 3.6506312099273735e-06,
      "loss": 0.3543,
      "step": 38913
    },
    {
      "epoch": 1.7857831214721673,
      "grad_norm": 0.4424329102039337,
      "learning_rate": 3.650395121165173e-06,
      "loss": 0.3258,
      "step": 38914
    },
    {
      "epoch": 1.7858290119774218,
      "grad_norm": 0.4706698954105377,
      "learning_rate": 3.6501590356483457e-06,
      "loss": 0.3294,
      "step": 38915
    },
    {
      "epoch": 1.7858749024826763,
      "grad_norm": 0.5204449892044067,
      "learning_rate": 3.649922953377459e-06,
      "loss": 0.4157,
      "step": 38916
    },
    {
      "epoch": 1.7859207929879308,
      "grad_norm": 0.4657539129257202,
      "learning_rate": 3.649686874353083e-06,
      "loss": 0.3868,
      "step": 38917
    },
    {
      "epoch": 1.7859666834931853,
      "grad_norm": 0.4737354815006256,
      "learning_rate": 3.6494507985757833e-06,
      "loss": 0.3558,
      "step": 38918
    },
    {
      "epoch": 1.7860125739984398,
      "grad_norm": 0.4947808086872101,
      "learning_rate": 3.6492147260461264e-06,
      "loss": 0.4197,
      "step": 38919
    },
    {
      "epoch": 1.7860584645036943,
      "grad_norm": 0.4777209758758545,
      "learning_rate": 3.648978656764684e-06,
      "loss": 0.3441,
      "step": 38920
    },
    {
      "epoch": 1.7861043550089486,
      "grad_norm": 0.4520593285560608,
      "learning_rate": 3.6487425907320216e-06,
      "loss": 0.3005,
      "step": 38921
    },
    {
      "epoch": 1.786150245514203,
      "grad_norm": 0.48971161246299744,
      "learning_rate": 3.6485065279487052e-06,
      "loss": 0.3684,
      "step": 38922
    },
    {
      "epoch": 1.7861961360194576,
      "grad_norm": 0.4493216574192047,
      "learning_rate": 3.6482704684153056e-06,
      "loss": 0.2938,
      "step": 38923
    },
    {
      "epoch": 1.786242026524712,
      "grad_norm": 0.5002160668373108,
      "learning_rate": 3.6480344121323887e-06,
      "loss": 0.4173,
      "step": 38924
    },
    {
      "epoch": 1.7862879170299664,
      "grad_norm": 0.47588375210762024,
      "learning_rate": 3.6477983591005205e-06,
      "loss": 0.3991,
      "step": 38925
    },
    {
      "epoch": 1.7863338075352209,
      "grad_norm": 0.46012020111083984,
      "learning_rate": 3.6475623093202716e-06,
      "loss": 0.3772,
      "step": 38926
    },
    {
      "epoch": 1.7863796980404754,
      "grad_norm": 0.5056272149085999,
      "learning_rate": 3.6473262627922096e-06,
      "loss": 0.3689,
      "step": 38927
    },
    {
      "epoch": 1.7864255885457299,
      "grad_norm": 0.4490322768688202,
      "learning_rate": 3.6470902195169e-06,
      "loss": 0.3371,
      "step": 38928
    },
    {
      "epoch": 1.7864714790509844,
      "grad_norm": 0.48112401366233826,
      "learning_rate": 3.646854179494912e-06,
      "loss": 0.3808,
      "step": 38929
    },
    {
      "epoch": 1.7865173695562389,
      "grad_norm": 0.5166167616844177,
      "learning_rate": 3.6466181427268126e-06,
      "loss": 0.3785,
      "step": 38930
    },
    {
      "epoch": 1.7865632600614934,
      "grad_norm": 0.48401951789855957,
      "learning_rate": 3.6463821092131687e-06,
      "loss": 0.4247,
      "step": 38931
    },
    {
      "epoch": 1.7866091505667479,
      "grad_norm": 0.4511071443557739,
      "learning_rate": 3.6461460789545496e-06,
      "loss": 0.3534,
      "step": 38932
    },
    {
      "epoch": 1.7866550410720023,
      "grad_norm": 0.45948219299316406,
      "learning_rate": 3.645910051951521e-06,
      "loss": 0.3295,
      "step": 38933
    },
    {
      "epoch": 1.7867009315772566,
      "grad_norm": 0.4965776205062866,
      "learning_rate": 3.645674028204653e-06,
      "loss": 0.3858,
      "step": 38934
    },
    {
      "epoch": 1.7867468220825111,
      "grad_norm": 0.4239000678062439,
      "learning_rate": 3.645438007714508e-06,
      "loss": 0.2756,
      "step": 38935
    },
    {
      "epoch": 1.7867927125877656,
      "grad_norm": 0.4805695414543152,
      "learning_rate": 3.64520199048166e-06,
      "loss": 0.3741,
      "step": 38936
    },
    {
      "epoch": 1.78683860309302,
      "grad_norm": 0.44096726179122925,
      "learning_rate": 3.6449659765066736e-06,
      "loss": 0.3547,
      "step": 38937
    },
    {
      "epoch": 1.7868844935982744,
      "grad_norm": 0.4842705726623535,
      "learning_rate": 3.6447299657901146e-06,
      "loss": 0.33,
      "step": 38938
    },
    {
      "epoch": 1.786930384103529,
      "grad_norm": 0.5217254161834717,
      "learning_rate": 3.644493958332554e-06,
      "loss": 0.3885,
      "step": 38939
    },
    {
      "epoch": 1.7869762746087834,
      "grad_norm": 0.42562544345855713,
      "learning_rate": 3.6442579541345573e-06,
      "loss": 0.2793,
      "step": 38940
    },
    {
      "epoch": 1.7870221651140379,
      "grad_norm": 0.4950166940689087,
      "learning_rate": 3.6440219531966905e-06,
      "loss": 0.3796,
      "step": 38941
    },
    {
      "epoch": 1.7870680556192924,
      "grad_norm": 0.479701966047287,
      "learning_rate": 3.643785955519525e-06,
      "loss": 0.4008,
      "step": 38942
    },
    {
      "epoch": 1.7871139461245469,
      "grad_norm": 0.5827389359474182,
      "learning_rate": 3.6435499611036262e-06,
      "loss": 0.344,
      "step": 38943
    },
    {
      "epoch": 1.7871598366298014,
      "grad_norm": 0.44716349244117737,
      "learning_rate": 3.64331396994956e-06,
      "loss": 0.3113,
      "step": 38944
    },
    {
      "epoch": 1.7872057271350559,
      "grad_norm": 0.48132508993148804,
      "learning_rate": 3.6430779820578975e-06,
      "loss": 0.3503,
      "step": 38945
    },
    {
      "epoch": 1.7872516176403104,
      "grad_norm": 0.44393736124038696,
      "learning_rate": 3.642841997429204e-06,
      "loss": 0.3471,
      "step": 38946
    },
    {
      "epoch": 1.7872975081455647,
      "grad_norm": 0.4960499107837677,
      "learning_rate": 3.642606016064045e-06,
      "loss": 0.4239,
      "step": 38947
    },
    {
      "epoch": 1.7873433986508191,
      "grad_norm": 0.5084317922592163,
      "learning_rate": 3.6423700379629923e-06,
      "loss": 0.3677,
      "step": 38948
    },
    {
      "epoch": 1.7873892891560736,
      "grad_norm": 0.46935632824897766,
      "learning_rate": 3.6421340631266105e-06,
      "loss": 0.3646,
      "step": 38949
    },
    {
      "epoch": 1.787435179661328,
      "grad_norm": 0.4705304801464081,
      "learning_rate": 3.641898091555467e-06,
      "loss": 0.33,
      "step": 38950
    },
    {
      "epoch": 1.7874810701665824,
      "grad_norm": 0.5175275802612305,
      "learning_rate": 3.6416621232501315e-06,
      "loss": 0.4203,
      "step": 38951
    },
    {
      "epoch": 1.787526960671837,
      "grad_norm": 0.47787007689476013,
      "learning_rate": 3.6414261582111698e-06,
      "loss": 0.3764,
      "step": 38952
    },
    {
      "epoch": 1.7875728511770914,
      "grad_norm": 0.5212996602058411,
      "learning_rate": 3.6411901964391476e-06,
      "loss": 0.3889,
      "step": 38953
    },
    {
      "epoch": 1.787618741682346,
      "grad_norm": 0.4915921092033386,
      "learning_rate": 3.6409542379346363e-06,
      "loss": 0.3825,
      "step": 38954
    },
    {
      "epoch": 1.7876646321876004,
      "grad_norm": 0.4641614258289337,
      "learning_rate": 3.6407182826982e-06,
      "loss": 0.3352,
      "step": 38955
    },
    {
      "epoch": 1.787710522692855,
      "grad_norm": 0.4913608431816101,
      "learning_rate": 3.640482330730406e-06,
      "loss": 0.4027,
      "step": 38956
    },
    {
      "epoch": 1.7877564131981094,
      "grad_norm": 0.4311918020248413,
      "learning_rate": 3.6402463820318245e-06,
      "loss": 0.2775,
      "step": 38957
    },
    {
      "epoch": 1.787802303703364,
      "grad_norm": 0.4445841908454895,
      "learning_rate": 3.6400104366030214e-06,
      "loss": 0.3316,
      "step": 38958
    },
    {
      "epoch": 1.7878481942086184,
      "grad_norm": 0.46535274386405945,
      "learning_rate": 3.6397744944445634e-06,
      "loss": 0.3247,
      "step": 38959
    },
    {
      "epoch": 1.7878940847138727,
      "grad_norm": 0.5080114603042603,
      "learning_rate": 3.6395385555570184e-06,
      "loss": 0.4207,
      "step": 38960
    },
    {
      "epoch": 1.7879399752191272,
      "grad_norm": 0.49811017513275146,
      "learning_rate": 3.639302619940955e-06,
      "loss": 0.3645,
      "step": 38961
    },
    {
      "epoch": 1.7879858657243817,
      "grad_norm": 0.4890167713165283,
      "learning_rate": 3.6390666875969367e-06,
      "loss": 0.3981,
      "step": 38962
    },
    {
      "epoch": 1.788031756229636,
      "grad_norm": 0.46625998616218567,
      "learning_rate": 3.6388307585255354e-06,
      "loss": 0.3186,
      "step": 38963
    },
    {
      "epoch": 1.7880776467348904,
      "grad_norm": 0.4739624559879303,
      "learning_rate": 3.6385948327273168e-06,
      "loss": 0.3471,
      "step": 38964
    },
    {
      "epoch": 1.788123537240145,
      "grad_norm": 0.4863038659095764,
      "learning_rate": 3.638358910202847e-06,
      "loss": 0.4081,
      "step": 38965
    },
    {
      "epoch": 1.7881694277453994,
      "grad_norm": 0.46976757049560547,
      "learning_rate": 3.638122990952695e-06,
      "loss": 0.3246,
      "step": 38966
    },
    {
      "epoch": 1.788215318250654,
      "grad_norm": 0.4728696346282959,
      "learning_rate": 3.6378870749774276e-06,
      "loss": 0.359,
      "step": 38967
    },
    {
      "epoch": 1.7882612087559084,
      "grad_norm": 0.4206938147544861,
      "learning_rate": 3.637651162277612e-06,
      "loss": 0.2771,
      "step": 38968
    },
    {
      "epoch": 1.788307099261163,
      "grad_norm": 0.4470050632953644,
      "learning_rate": 3.637415252853813e-06,
      "loss": 0.3245,
      "step": 38969
    },
    {
      "epoch": 1.7883529897664174,
      "grad_norm": 0.4776450991630554,
      "learning_rate": 3.637179346706603e-06,
      "loss": 0.3695,
      "step": 38970
    },
    {
      "epoch": 1.788398880271672,
      "grad_norm": 0.4869869351387024,
      "learning_rate": 3.636943443836546e-06,
      "loss": 0.4111,
      "step": 38971
    },
    {
      "epoch": 1.7884447707769262,
      "grad_norm": 0.7510921359062195,
      "learning_rate": 3.636707544244209e-06,
      "loss": 0.3653,
      "step": 38972
    },
    {
      "epoch": 1.7884906612821807,
      "grad_norm": 0.45729202032089233,
      "learning_rate": 3.6364716479301608e-06,
      "loss": 0.313,
      "step": 38973
    },
    {
      "epoch": 1.7885365517874352,
      "grad_norm": 0.4710065424442291,
      "learning_rate": 3.6362357548949677e-06,
      "loss": 0.3615,
      "step": 38974
    },
    {
      "epoch": 1.7885824422926895,
      "grad_norm": 0.46029266715049744,
      "learning_rate": 3.635999865139197e-06,
      "loss": 0.3311,
      "step": 38975
    },
    {
      "epoch": 1.788628332797944,
      "grad_norm": 0.4766286313533783,
      "learning_rate": 3.635763978663417e-06,
      "loss": 0.3486,
      "step": 38976
    },
    {
      "epoch": 1.7886742233031985,
      "grad_norm": 0.5313324928283691,
      "learning_rate": 3.6355280954681938e-06,
      "loss": 0.4835,
      "step": 38977
    },
    {
      "epoch": 1.788720113808453,
      "grad_norm": 0.45634564757347107,
      "learning_rate": 3.6352922155540925e-06,
      "loss": 0.3235,
      "step": 38978
    },
    {
      "epoch": 1.7887660043137075,
      "grad_norm": 0.5241215229034424,
      "learning_rate": 3.6350563389216853e-06,
      "loss": 0.4652,
      "step": 38979
    },
    {
      "epoch": 1.788811894818962,
      "grad_norm": 0.49338221549987793,
      "learning_rate": 3.634820465571537e-06,
      "loss": 0.3953,
      "step": 38980
    },
    {
      "epoch": 1.7888577853242165,
      "grad_norm": 0.4995090663433075,
      "learning_rate": 3.6345845955042125e-06,
      "loss": 0.3794,
      "step": 38981
    },
    {
      "epoch": 1.788903675829471,
      "grad_norm": 0.5188331007957458,
      "learning_rate": 3.6343487287202828e-06,
      "loss": 0.4912,
      "step": 38982
    },
    {
      "epoch": 1.7889495663347255,
      "grad_norm": 0.4594517648220062,
      "learning_rate": 3.6341128652203137e-06,
      "loss": 0.3146,
      "step": 38983
    },
    {
      "epoch": 1.78899545683998,
      "grad_norm": 0.45125338435173035,
      "learning_rate": 3.63387700500487e-06,
      "loss": 0.2914,
      "step": 38984
    },
    {
      "epoch": 1.7890413473452342,
      "grad_norm": 0.45458483695983887,
      "learning_rate": 3.633641148074523e-06,
      "loss": 0.3191,
      "step": 38985
    },
    {
      "epoch": 1.7890872378504887,
      "grad_norm": 0.4788898527622223,
      "learning_rate": 3.633405294429837e-06,
      "loss": 0.3793,
      "step": 38986
    },
    {
      "epoch": 1.7891331283557432,
      "grad_norm": 0.4719712734222412,
      "learning_rate": 3.6331694440713794e-06,
      "loss": 0.3484,
      "step": 38987
    },
    {
      "epoch": 1.7891790188609975,
      "grad_norm": 0.4692332446575165,
      "learning_rate": 3.6329335969997194e-06,
      "loss": 0.345,
      "step": 38988
    },
    {
      "epoch": 1.789224909366252,
      "grad_norm": 0.4610622227191925,
      "learning_rate": 3.6326977532154223e-06,
      "loss": 0.3435,
      "step": 38989
    },
    {
      "epoch": 1.7892707998715065,
      "grad_norm": 0.5087058544158936,
      "learning_rate": 3.632461912719053e-06,
      "loss": 0.4207,
      "step": 38990
    },
    {
      "epoch": 1.789316690376761,
      "grad_norm": 0.46746012568473816,
      "learning_rate": 3.632226075511184e-06,
      "loss": 0.3435,
      "step": 38991
    },
    {
      "epoch": 1.7893625808820155,
      "grad_norm": 0.44698840379714966,
      "learning_rate": 3.63199024159238e-06,
      "loss": 0.2914,
      "step": 38992
    },
    {
      "epoch": 1.78940847138727,
      "grad_norm": 0.4543249011039734,
      "learning_rate": 3.6317544109632053e-06,
      "loss": 0.3245,
      "step": 38993
    },
    {
      "epoch": 1.7894543618925245,
      "grad_norm": 0.4537357985973358,
      "learning_rate": 3.6315185836242316e-06,
      "loss": 0.3166,
      "step": 38994
    },
    {
      "epoch": 1.789500252397779,
      "grad_norm": 0.5167016983032227,
      "learning_rate": 3.6312827595760236e-06,
      "loss": 0.3891,
      "step": 38995
    },
    {
      "epoch": 1.7895461429030335,
      "grad_norm": 0.46405622363090515,
      "learning_rate": 3.631046938819146e-06,
      "loss": 0.3656,
      "step": 38996
    },
    {
      "epoch": 1.789592033408288,
      "grad_norm": 0.4733262062072754,
      "learning_rate": 3.630811121354172e-06,
      "loss": 0.3798,
      "step": 38997
    },
    {
      "epoch": 1.7896379239135423,
      "grad_norm": 0.47339415550231934,
      "learning_rate": 3.630575307181664e-06,
      "loss": 0.3283,
      "step": 38998
    },
    {
      "epoch": 1.7896838144187968,
      "grad_norm": 0.47460290789604187,
      "learning_rate": 3.63033949630219e-06,
      "loss": 0.3494,
      "step": 38999
    },
    {
      "epoch": 1.7897297049240513,
      "grad_norm": 0.4535413086414337,
      "learning_rate": 3.6301036887163155e-06,
      "loss": 0.3439,
      "step": 39000
    },
    {
      "epoch": 1.7897755954293055,
      "grad_norm": 0.49388542771339417,
      "learning_rate": 3.6298678844246106e-06,
      "loss": 0.3786,
      "step": 39001
    },
    {
      "epoch": 1.78982148593456,
      "grad_norm": 0.5035730600357056,
      "learning_rate": 3.6296320834276415e-06,
      "loss": 0.2787,
      "step": 39002
    },
    {
      "epoch": 1.7898673764398145,
      "grad_norm": 0.4609580934047699,
      "learning_rate": 3.629396285725974e-06,
      "loss": 0.3514,
      "step": 39003
    },
    {
      "epoch": 1.789913266945069,
      "grad_norm": 0.48449817299842834,
      "learning_rate": 3.6291604913201766e-06,
      "loss": 0.4319,
      "step": 39004
    },
    {
      "epoch": 1.7899591574503235,
      "grad_norm": 0.48680898547172546,
      "learning_rate": 3.628924700210816e-06,
      "loss": 0.3513,
      "step": 39005
    },
    {
      "epoch": 1.790005047955578,
      "grad_norm": 0.4859076738357544,
      "learning_rate": 3.628688912398456e-06,
      "loss": 0.3882,
      "step": 39006
    },
    {
      "epoch": 1.7900509384608325,
      "grad_norm": 0.5115410685539246,
      "learning_rate": 3.628453127883669e-06,
      "loss": 0.3573,
      "step": 39007
    },
    {
      "epoch": 1.790096828966087,
      "grad_norm": 0.4633951485157013,
      "learning_rate": 3.628217346667019e-06,
      "loss": 0.3648,
      "step": 39008
    },
    {
      "epoch": 1.7901427194713415,
      "grad_norm": 0.4623551368713379,
      "learning_rate": 3.627981568749072e-06,
      "loss": 0.3704,
      "step": 39009
    },
    {
      "epoch": 1.7901886099765958,
      "grad_norm": 0.43444451689720154,
      "learning_rate": 3.6277457941303985e-06,
      "loss": 0.2962,
      "step": 39010
    },
    {
      "epoch": 1.7902345004818503,
      "grad_norm": 0.4922162592411041,
      "learning_rate": 3.6275100228115623e-06,
      "loss": 0.4003,
      "step": 39011
    },
    {
      "epoch": 1.7902803909871048,
      "grad_norm": 0.4627329707145691,
      "learning_rate": 3.6272742547931294e-06,
      "loss": 0.3272,
      "step": 39012
    },
    {
      "epoch": 1.7903262814923593,
      "grad_norm": 0.4639261066913605,
      "learning_rate": 3.6270384900756704e-06,
      "loss": 0.3468,
      "step": 39013
    },
    {
      "epoch": 1.7903721719976136,
      "grad_norm": 0.45098528265953064,
      "learning_rate": 3.6268027286597513e-06,
      "loss": 0.2997,
      "step": 39014
    },
    {
      "epoch": 1.790418062502868,
      "grad_norm": 0.5043196678161621,
      "learning_rate": 3.626566970545937e-06,
      "loss": 0.4683,
      "step": 39015
    },
    {
      "epoch": 1.7904639530081226,
      "grad_norm": 0.4593127369880676,
      "learning_rate": 3.626331215734796e-06,
      "loss": 0.3349,
      "step": 39016
    },
    {
      "epoch": 1.790509843513377,
      "grad_norm": 0.4710067808628082,
      "learning_rate": 3.626095464226895e-06,
      "loss": 0.3252,
      "step": 39017
    },
    {
      "epoch": 1.7905557340186316,
      "grad_norm": 0.48087239265441895,
      "learning_rate": 3.6258597160228e-06,
      "loss": 0.4204,
      "step": 39018
    },
    {
      "epoch": 1.790601624523886,
      "grad_norm": 0.4525650143623352,
      "learning_rate": 3.6256239711230805e-06,
      "loss": 0.3231,
      "step": 39019
    },
    {
      "epoch": 1.7906475150291405,
      "grad_norm": 0.4923372268676758,
      "learning_rate": 3.6253882295283017e-06,
      "loss": 0.3331,
      "step": 39020
    },
    {
      "epoch": 1.790693405534395,
      "grad_norm": 0.444284588098526,
      "learning_rate": 3.625152491239027e-06,
      "loss": 0.264,
      "step": 39021
    },
    {
      "epoch": 1.7907392960396495,
      "grad_norm": 0.4350729286670685,
      "learning_rate": 3.62491675625583e-06,
      "loss": 0.3228,
      "step": 39022
    },
    {
      "epoch": 1.7907851865449038,
      "grad_norm": 0.48174792528152466,
      "learning_rate": 3.624681024579273e-06,
      "loss": 0.373,
      "step": 39023
    },
    {
      "epoch": 1.7908310770501583,
      "grad_norm": 0.4338656961917877,
      "learning_rate": 3.6244452962099235e-06,
      "loss": 0.331,
      "step": 39024
    },
    {
      "epoch": 1.7908769675554128,
      "grad_norm": 0.4945756196975708,
      "learning_rate": 3.62420957114835e-06,
      "loss": 0.4106,
      "step": 39025
    },
    {
      "epoch": 1.790922858060667,
      "grad_norm": 0.5030802488327026,
      "learning_rate": 3.6239738493951183e-06,
      "loss": 0.3657,
      "step": 39026
    },
    {
      "epoch": 1.7909687485659216,
      "grad_norm": 0.4823955297470093,
      "learning_rate": 3.623738130950793e-06,
      "loss": 0.4072,
      "step": 39027
    },
    {
      "epoch": 1.791014639071176,
      "grad_norm": 0.47356152534484863,
      "learning_rate": 3.6235024158159452e-06,
      "loss": 0.2981,
      "step": 39028
    },
    {
      "epoch": 1.7910605295764306,
      "grad_norm": 0.49216094613075256,
      "learning_rate": 3.62326670399114e-06,
      "loss": 0.34,
      "step": 39029
    },
    {
      "epoch": 1.791106420081685,
      "grad_norm": 0.46532973647117615,
      "learning_rate": 3.623030995476943e-06,
      "loss": 0.317,
      "step": 39030
    },
    {
      "epoch": 1.7911523105869396,
      "grad_norm": 0.44334882497787476,
      "learning_rate": 3.6227952902739223e-06,
      "loss": 0.3013,
      "step": 39031
    },
    {
      "epoch": 1.791198201092194,
      "grad_norm": 0.4597153961658478,
      "learning_rate": 3.622559588382645e-06,
      "loss": 0.3174,
      "step": 39032
    },
    {
      "epoch": 1.7912440915974486,
      "grad_norm": 0.4653659462928772,
      "learning_rate": 3.622323889803676e-06,
      "loss": 0.3389,
      "step": 39033
    },
    {
      "epoch": 1.791289982102703,
      "grad_norm": 0.43161216378211975,
      "learning_rate": 3.622088194537582e-06,
      "loss": 0.3173,
      "step": 39034
    },
    {
      "epoch": 1.7913358726079576,
      "grad_norm": 0.45239779353141785,
      "learning_rate": 3.621852502584933e-06,
      "loss": 0.3332,
      "step": 39035
    },
    {
      "epoch": 1.7913817631132118,
      "grad_norm": 0.464103102684021,
      "learning_rate": 3.621616813946294e-06,
      "loss": 0.3302,
      "step": 39036
    },
    {
      "epoch": 1.7914276536184663,
      "grad_norm": 0.4936441481113434,
      "learning_rate": 3.6213811286222295e-06,
      "loss": 0.3725,
      "step": 39037
    },
    {
      "epoch": 1.7914735441237208,
      "grad_norm": 0.4331585764884949,
      "learning_rate": 3.62114544661331e-06,
      "loss": 0.313,
      "step": 39038
    },
    {
      "epoch": 1.7915194346289751,
      "grad_norm": 0.4732377827167511,
      "learning_rate": 3.6209097679201004e-06,
      "loss": 0.3492,
      "step": 39039
    },
    {
      "epoch": 1.7915653251342296,
      "grad_norm": 0.4589652419090271,
      "learning_rate": 3.620674092543166e-06,
      "loss": 0.328,
      "step": 39040
    },
    {
      "epoch": 1.7916112156394841,
      "grad_norm": 0.49584299325942993,
      "learning_rate": 3.620438420483078e-06,
      "loss": 0.3991,
      "step": 39041
    },
    {
      "epoch": 1.7916571061447386,
      "grad_norm": 0.4257907569408417,
      "learning_rate": 3.6202027517403983e-06,
      "loss": 0.2823,
      "step": 39042
    },
    {
      "epoch": 1.791702996649993,
      "grad_norm": 0.5027692317962646,
      "learning_rate": 3.6199670863156946e-06,
      "loss": 0.3663,
      "step": 39043
    },
    {
      "epoch": 1.7917488871552476,
      "grad_norm": 0.4778556525707245,
      "learning_rate": 3.6197314242095355e-06,
      "loss": 0.3867,
      "step": 39044
    },
    {
      "epoch": 1.791794777660502,
      "grad_norm": 0.46024858951568604,
      "learning_rate": 3.6194957654224874e-06,
      "loss": 0.3309,
      "step": 39045
    },
    {
      "epoch": 1.7918406681657566,
      "grad_norm": 0.44204387068748474,
      "learning_rate": 3.619260109955115e-06,
      "loss": 0.3388,
      "step": 39046
    },
    {
      "epoch": 1.791886558671011,
      "grad_norm": 0.4797074794769287,
      "learning_rate": 3.6190244578079874e-06,
      "loss": 0.3699,
      "step": 39047
    },
    {
      "epoch": 1.7919324491762656,
      "grad_norm": 0.5228578448295593,
      "learning_rate": 3.61878880898167e-06,
      "loss": 0.3833,
      "step": 39048
    },
    {
      "epoch": 1.7919783396815199,
      "grad_norm": 0.44349056482315063,
      "learning_rate": 3.618553163476728e-06,
      "loss": 0.354,
      "step": 39049
    },
    {
      "epoch": 1.7920242301867744,
      "grad_norm": 0.4855322539806366,
      "learning_rate": 3.6183175212937316e-06,
      "loss": 0.3612,
      "step": 39050
    },
    {
      "epoch": 1.7920701206920289,
      "grad_norm": 0.4381566047668457,
      "learning_rate": 3.618081882433245e-06,
      "loss": 0.3054,
      "step": 39051
    },
    {
      "epoch": 1.7921160111972831,
      "grad_norm": 0.4582786560058594,
      "learning_rate": 3.617846246895834e-06,
      "loss": 0.3125,
      "step": 39052
    },
    {
      "epoch": 1.7921619017025376,
      "grad_norm": 0.4975954294204712,
      "learning_rate": 3.6176106146820687e-06,
      "loss": 0.3563,
      "step": 39053
    },
    {
      "epoch": 1.7922077922077921,
      "grad_norm": 0.4677339494228363,
      "learning_rate": 3.617374985792513e-06,
      "loss": 0.3566,
      "step": 39054
    },
    {
      "epoch": 1.7922536827130466,
      "grad_norm": 0.45274174213409424,
      "learning_rate": 3.6171393602277317e-06,
      "loss": 0.3276,
      "step": 39055
    },
    {
      "epoch": 1.7922995732183011,
      "grad_norm": 0.4715751111507416,
      "learning_rate": 3.6169037379882964e-06,
      "loss": 0.3506,
      "step": 39056
    },
    {
      "epoch": 1.7923454637235556,
      "grad_norm": 0.4670037627220154,
      "learning_rate": 3.616668119074771e-06,
      "loss": 0.3582,
      "step": 39057
    },
    {
      "epoch": 1.7923913542288101,
      "grad_norm": 0.48028427362442017,
      "learning_rate": 3.6164325034877212e-06,
      "loss": 0.3835,
      "step": 39058
    },
    {
      "epoch": 1.7924372447340646,
      "grad_norm": 0.4367833137512207,
      "learning_rate": 3.616196891227715e-06,
      "loss": 0.2766,
      "step": 39059
    },
    {
      "epoch": 1.7924831352393191,
      "grad_norm": 0.45414289832115173,
      "learning_rate": 3.6159612822953194e-06,
      "loss": 0.3225,
      "step": 39060
    },
    {
      "epoch": 1.7925290257445734,
      "grad_norm": 0.4873010516166687,
      "learning_rate": 3.6157256766910975e-06,
      "loss": 0.3948,
      "step": 39061
    },
    {
      "epoch": 1.792574916249828,
      "grad_norm": 0.5049685835838318,
      "learning_rate": 3.6154900744156206e-06,
      "loss": 0.4023,
      "step": 39062
    },
    {
      "epoch": 1.7926208067550824,
      "grad_norm": 0.48493272066116333,
      "learning_rate": 3.615254475469454e-06,
      "loss": 0.3748,
      "step": 39063
    },
    {
      "epoch": 1.7926666972603367,
      "grad_norm": 0.4843459129333496,
      "learning_rate": 3.6150188798531626e-06,
      "loss": 0.3768,
      "step": 39064
    },
    {
      "epoch": 1.7927125877655912,
      "grad_norm": 0.41103413701057434,
      "learning_rate": 3.6147832875673115e-06,
      "loss": 0.2656,
      "step": 39065
    },
    {
      "epoch": 1.7927584782708457,
      "grad_norm": 0.4740593731403351,
      "learning_rate": 3.6145476986124716e-06,
      "loss": 0.3596,
      "step": 39066
    },
    {
      "epoch": 1.7928043687761002,
      "grad_norm": 0.43567797541618347,
      "learning_rate": 3.6143121129892067e-06,
      "loss": 0.3372,
      "step": 39067
    },
    {
      "epoch": 1.7928502592813547,
      "grad_norm": 0.6817604303359985,
      "learning_rate": 3.614076530698083e-06,
      "loss": 0.4235,
      "step": 39068
    },
    {
      "epoch": 1.7928961497866092,
      "grad_norm": 0.45818445086479187,
      "learning_rate": 3.6138409517396694e-06,
      "loss": 0.3514,
      "step": 39069
    },
    {
      "epoch": 1.7929420402918637,
      "grad_norm": 0.511385440826416,
      "learning_rate": 3.6136053761145307e-06,
      "loss": 0.4487,
      "step": 39070
    },
    {
      "epoch": 1.7929879307971182,
      "grad_norm": 0.4798769950866699,
      "learning_rate": 3.613369803823231e-06,
      "loss": 0.3972,
      "step": 39071
    },
    {
      "epoch": 1.7930338213023727,
      "grad_norm": 0.44799596071243286,
      "learning_rate": 3.613134234866341e-06,
      "loss": 0.3007,
      "step": 39072
    },
    {
      "epoch": 1.7930797118076272,
      "grad_norm": 0.49659913778305054,
      "learning_rate": 3.612898669244427e-06,
      "loss": 0.3843,
      "step": 39073
    },
    {
      "epoch": 1.7931256023128814,
      "grad_norm": 0.4618549048900604,
      "learning_rate": 3.6126631069580514e-06,
      "loss": 0.3815,
      "step": 39074
    },
    {
      "epoch": 1.793171492818136,
      "grad_norm": 0.4708144962787628,
      "learning_rate": 3.6124275480077846e-06,
      "loss": 0.3607,
      "step": 39075
    },
    {
      "epoch": 1.7932173833233904,
      "grad_norm": 0.46132534742355347,
      "learning_rate": 3.612191992394192e-06,
      "loss": 0.3375,
      "step": 39076
    },
    {
      "epoch": 1.7932632738286447,
      "grad_norm": 0.4657265543937683,
      "learning_rate": 3.6119564401178374e-06,
      "loss": 0.3272,
      "step": 39077
    },
    {
      "epoch": 1.7933091643338992,
      "grad_norm": 0.5100289583206177,
      "learning_rate": 3.611720891179291e-06,
      "loss": 0.3804,
      "step": 39078
    },
    {
      "epoch": 1.7933550548391537,
      "grad_norm": 0.4835840165615082,
      "learning_rate": 3.6114853455791187e-06,
      "loss": 0.3905,
      "step": 39079
    },
    {
      "epoch": 1.7934009453444082,
      "grad_norm": 0.47456595301628113,
      "learning_rate": 3.6112498033178844e-06,
      "loss": 0.3666,
      "step": 39080
    },
    {
      "epoch": 1.7934468358496627,
      "grad_norm": 0.47419407963752747,
      "learning_rate": 3.6110142643961564e-06,
      "loss": 0.3849,
      "step": 39081
    },
    {
      "epoch": 1.7934927263549172,
      "grad_norm": 0.4323647916316986,
      "learning_rate": 3.610778728814502e-06,
      "loss": 0.2864,
      "step": 39082
    },
    {
      "epoch": 1.7935386168601717,
      "grad_norm": 0.49279850721359253,
      "learning_rate": 3.6105431965734832e-06,
      "loss": 0.4066,
      "step": 39083
    },
    {
      "epoch": 1.7935845073654262,
      "grad_norm": 0.4407173693180084,
      "learning_rate": 3.610307667673672e-06,
      "loss": 0.3092,
      "step": 39084
    },
    {
      "epoch": 1.7936303978706807,
      "grad_norm": 0.4560415744781494,
      "learning_rate": 3.610072142115634e-06,
      "loss": 0.3203,
      "step": 39085
    },
    {
      "epoch": 1.7936762883759352,
      "grad_norm": 0.5087391138076782,
      "learning_rate": 3.609836619899929e-06,
      "loss": 0.4515,
      "step": 39086
    },
    {
      "epoch": 1.7937221788811895,
      "grad_norm": 0.49655935168266296,
      "learning_rate": 3.6096011010271315e-06,
      "loss": 0.416,
      "step": 39087
    },
    {
      "epoch": 1.793768069386444,
      "grad_norm": 0.4519485533237457,
      "learning_rate": 3.6093655854978043e-06,
      "loss": 0.358,
      "step": 39088
    },
    {
      "epoch": 1.7938139598916985,
      "grad_norm": 0.4522759020328522,
      "learning_rate": 3.609130073312513e-06,
      "loss": 0.3599,
      "step": 39089
    },
    {
      "epoch": 1.7938598503969527,
      "grad_norm": 0.4722626507282257,
      "learning_rate": 3.6088945644718254e-06,
      "loss": 0.3417,
      "step": 39090
    },
    {
      "epoch": 1.7939057409022072,
      "grad_norm": 0.4826270639896393,
      "learning_rate": 3.608659058976308e-06,
      "loss": 0.3995,
      "step": 39091
    },
    {
      "epoch": 1.7939516314074617,
      "grad_norm": 0.43914300203323364,
      "learning_rate": 3.608423556826524e-06,
      "loss": 0.3073,
      "step": 39092
    },
    {
      "epoch": 1.7939975219127162,
      "grad_norm": 0.5353196263313293,
      "learning_rate": 3.6081880580230445e-06,
      "loss": 0.4576,
      "step": 39093
    },
    {
      "epoch": 1.7940434124179707,
      "grad_norm": 0.4994310736656189,
      "learning_rate": 3.6079525625664335e-06,
      "loss": 0.4249,
      "step": 39094
    },
    {
      "epoch": 1.7940893029232252,
      "grad_norm": 0.4786619544029236,
      "learning_rate": 3.6077170704572556e-06,
      "loss": 0.3972,
      "step": 39095
    },
    {
      "epoch": 1.7941351934284797,
      "grad_norm": 0.42954838275909424,
      "learning_rate": 3.60748158169608e-06,
      "loss": 0.2709,
      "step": 39096
    },
    {
      "epoch": 1.7941810839337342,
      "grad_norm": 0.47246503829956055,
      "learning_rate": 3.607246096283472e-06,
      "loss": 0.3517,
      "step": 39097
    },
    {
      "epoch": 1.7942269744389887,
      "grad_norm": 0.4846789836883545,
      "learning_rate": 3.607010614219998e-06,
      "loss": 0.3869,
      "step": 39098
    },
    {
      "epoch": 1.794272864944243,
      "grad_norm": 0.4607219696044922,
      "learning_rate": 3.606775135506221e-06,
      "loss": 0.3122,
      "step": 39099
    },
    {
      "epoch": 1.7943187554494975,
      "grad_norm": 0.4825912117958069,
      "learning_rate": 3.6065396601427115e-06,
      "loss": 0.3736,
      "step": 39100
    },
    {
      "epoch": 1.794364645954752,
      "grad_norm": 0.4601112902164459,
      "learning_rate": 3.6063041881300353e-06,
      "loss": 0.2945,
      "step": 39101
    },
    {
      "epoch": 1.7944105364600065,
      "grad_norm": 0.4749048948287964,
      "learning_rate": 3.606068719468757e-06,
      "loss": 0.3611,
      "step": 39102
    },
    {
      "epoch": 1.7944564269652608,
      "grad_norm": 0.4865274727344513,
      "learning_rate": 3.605833254159444e-06,
      "loss": 0.4343,
      "step": 39103
    },
    {
      "epoch": 1.7945023174705153,
      "grad_norm": 0.4507814049720764,
      "learning_rate": 3.605597792202662e-06,
      "loss": 0.3232,
      "step": 39104
    },
    {
      "epoch": 1.7945482079757697,
      "grad_norm": 0.4889148771762848,
      "learning_rate": 3.605362333598975e-06,
      "loss": 0.3817,
      "step": 39105
    },
    {
      "epoch": 1.7945940984810242,
      "grad_norm": 0.5521024465560913,
      "learning_rate": 3.6051268783489533e-06,
      "loss": 0.421,
      "step": 39106
    },
    {
      "epoch": 1.7946399889862787,
      "grad_norm": 0.45276516675949097,
      "learning_rate": 3.604891426453163e-06,
      "loss": 0.3036,
      "step": 39107
    },
    {
      "epoch": 1.7946858794915332,
      "grad_norm": 0.5247492790222168,
      "learning_rate": 3.6046559779121654e-06,
      "loss": 0.3812,
      "step": 39108
    },
    {
      "epoch": 1.7947317699967877,
      "grad_norm": 0.5347162485122681,
      "learning_rate": 3.604420532726531e-06,
      "loss": 0.4396,
      "step": 39109
    },
    {
      "epoch": 1.7947776605020422,
      "grad_norm": 0.44859734177589417,
      "learning_rate": 3.604185090896826e-06,
      "loss": 0.2907,
      "step": 39110
    },
    {
      "epoch": 1.7948235510072967,
      "grad_norm": 0.4471120238304138,
      "learning_rate": 3.6039496524236126e-06,
      "loss": 0.2985,
      "step": 39111
    },
    {
      "epoch": 1.794869441512551,
      "grad_norm": 0.47846418619155884,
      "learning_rate": 3.6037142173074625e-06,
      "loss": 0.3386,
      "step": 39112
    },
    {
      "epoch": 1.7949153320178055,
      "grad_norm": 0.5252884030342102,
      "learning_rate": 3.6034787855489382e-06,
      "loss": 0.3861,
      "step": 39113
    },
    {
      "epoch": 1.79496122252306,
      "grad_norm": 0.4815131425857544,
      "learning_rate": 3.603243357148605e-06,
      "loss": 0.344,
      "step": 39114
    },
    {
      "epoch": 1.7950071130283143,
      "grad_norm": 0.4899124801158905,
      "learning_rate": 3.603007932107032e-06,
      "loss": 0.3709,
      "step": 39115
    },
    {
      "epoch": 1.7950530035335688,
      "grad_norm": 0.4672159254550934,
      "learning_rate": 3.6027725104247845e-06,
      "loss": 0.3206,
      "step": 39116
    },
    {
      "epoch": 1.7950988940388233,
      "grad_norm": 0.4296744167804718,
      "learning_rate": 3.6025370921024267e-06,
      "loss": 0.2646,
      "step": 39117
    },
    {
      "epoch": 1.7951447845440778,
      "grad_norm": 0.4530632495880127,
      "learning_rate": 3.6023016771405273e-06,
      "loss": 0.3545,
      "step": 39118
    },
    {
      "epoch": 1.7951906750493323,
      "grad_norm": 0.45011207461357117,
      "learning_rate": 3.6020662655396516e-06,
      "loss": 0.3062,
      "step": 39119
    },
    {
      "epoch": 1.7952365655545868,
      "grad_norm": 0.4747127592563629,
      "learning_rate": 3.601830857300363e-06,
      "loss": 0.3605,
      "step": 39120
    },
    {
      "epoch": 1.7952824560598413,
      "grad_norm": 0.45756855607032776,
      "learning_rate": 3.6015954524232323e-06,
      "loss": 0.3371,
      "step": 39121
    },
    {
      "epoch": 1.7953283465650958,
      "grad_norm": 0.49475839734077454,
      "learning_rate": 3.6013600509088232e-06,
      "loss": 0.4214,
      "step": 39122
    },
    {
      "epoch": 1.7953742370703503,
      "grad_norm": 0.5522712469100952,
      "learning_rate": 3.6011246527577e-06,
      "loss": 0.5556,
      "step": 39123
    },
    {
      "epoch": 1.7954201275756048,
      "grad_norm": 0.4747956693172455,
      "learning_rate": 3.6008892579704317e-06,
      "loss": 0.3296,
      "step": 39124
    },
    {
      "epoch": 1.795466018080859,
      "grad_norm": 0.49176618456840515,
      "learning_rate": 3.6006538665475838e-06,
      "loss": 0.3699,
      "step": 39125
    },
    {
      "epoch": 1.7955119085861135,
      "grad_norm": 0.4626230299472809,
      "learning_rate": 3.600418478489719e-06,
      "loss": 0.3345,
      "step": 39126
    },
    {
      "epoch": 1.795557799091368,
      "grad_norm": 0.48788586258888245,
      "learning_rate": 3.6001830937974085e-06,
      "loss": 0.4083,
      "step": 39127
    },
    {
      "epoch": 1.7956036895966223,
      "grad_norm": 0.464763879776001,
      "learning_rate": 3.5999477124712158e-06,
      "loss": 0.3071,
      "step": 39128
    },
    {
      "epoch": 1.7956495801018768,
      "grad_norm": 0.47066062688827515,
      "learning_rate": 3.599712334511708e-06,
      "loss": 0.3639,
      "step": 39129
    },
    {
      "epoch": 1.7956954706071313,
      "grad_norm": 0.5380046963691711,
      "learning_rate": 3.5994769599194466e-06,
      "loss": 0.4034,
      "step": 39130
    },
    {
      "epoch": 1.7957413611123858,
      "grad_norm": 0.4737645983695984,
      "learning_rate": 3.599241588695004e-06,
      "loss": 0.3861,
      "step": 39131
    },
    {
      "epoch": 1.7957872516176403,
      "grad_norm": 0.4689282178878784,
      "learning_rate": 3.5990062208389432e-06,
      "loss": 0.3842,
      "step": 39132
    },
    {
      "epoch": 1.7958331421228948,
      "grad_norm": 0.4675632417201996,
      "learning_rate": 3.5987708563518286e-06,
      "loss": 0.3605,
      "step": 39133
    },
    {
      "epoch": 1.7958790326281493,
      "grad_norm": 0.4577837288379669,
      "learning_rate": 3.598535495234229e-06,
      "loss": 0.3186,
      "step": 39134
    },
    {
      "epoch": 1.7959249231334038,
      "grad_norm": 0.47404882311820984,
      "learning_rate": 3.59830013748671e-06,
      "loss": 0.4139,
      "step": 39135
    },
    {
      "epoch": 1.7959708136386583,
      "grad_norm": 0.4490429162979126,
      "learning_rate": 3.5980647831098337e-06,
      "loss": 0.3329,
      "step": 39136
    },
    {
      "epoch": 1.7960167041439128,
      "grad_norm": 0.4627262055873871,
      "learning_rate": 3.5978294321041716e-06,
      "loss": 0.3382,
      "step": 39137
    },
    {
      "epoch": 1.796062594649167,
      "grad_norm": 0.48307734727859497,
      "learning_rate": 3.5975940844702873e-06,
      "loss": 0.3752,
      "step": 39138
    },
    {
      "epoch": 1.7961084851544216,
      "grad_norm": 0.4380726218223572,
      "learning_rate": 3.5973587402087452e-06,
      "loss": 0.3177,
      "step": 39139
    },
    {
      "epoch": 1.796154375659676,
      "grad_norm": 0.4519878327846527,
      "learning_rate": 3.5971233993201138e-06,
      "loss": 0.3018,
      "step": 39140
    },
    {
      "epoch": 1.7962002661649303,
      "grad_norm": 0.4569428861141205,
      "learning_rate": 3.5968880618049583e-06,
      "loss": 0.361,
      "step": 39141
    },
    {
      "epoch": 1.7962461566701848,
      "grad_norm": 0.48241445422172546,
      "learning_rate": 3.5966527276638417e-06,
      "loss": 0.3313,
      "step": 39142
    },
    {
      "epoch": 1.7962920471754393,
      "grad_norm": 0.45483192801475525,
      "learning_rate": 3.596417396897335e-06,
      "loss": 0.3306,
      "step": 39143
    },
    {
      "epoch": 1.7963379376806938,
      "grad_norm": 0.4606817066669464,
      "learning_rate": 3.596182069506e-06,
      "loss": 0.3619,
      "step": 39144
    },
    {
      "epoch": 1.7963838281859483,
      "grad_norm": 0.4854259490966797,
      "learning_rate": 3.5959467454904048e-06,
      "loss": 0.3707,
      "step": 39145
    },
    {
      "epoch": 1.7964297186912028,
      "grad_norm": 0.5069246888160706,
      "learning_rate": 3.595711424851115e-06,
      "loss": 0.4497,
      "step": 39146
    },
    {
      "epoch": 1.7964756091964573,
      "grad_norm": 0.5490174889564514,
      "learning_rate": 3.5954761075886955e-06,
      "loss": 0.3823,
      "step": 39147
    },
    {
      "epoch": 1.7965214997017118,
      "grad_norm": 0.5013524889945984,
      "learning_rate": 3.595240793703711e-06,
      "loss": 0.3303,
      "step": 39148
    },
    {
      "epoch": 1.7965673902069663,
      "grad_norm": 0.49147921800613403,
      "learning_rate": 3.5950054831967306e-06,
      "loss": 0.3872,
      "step": 39149
    },
    {
      "epoch": 1.7966132807122206,
      "grad_norm": 0.43856003880500793,
      "learning_rate": 3.5947701760683197e-06,
      "loss": 0.3059,
      "step": 39150
    },
    {
      "epoch": 1.796659171217475,
      "grad_norm": 0.4646542966365814,
      "learning_rate": 3.594534872319041e-06,
      "loss": 0.2933,
      "step": 39151
    },
    {
      "epoch": 1.7967050617227296,
      "grad_norm": 0.4598725140094757,
      "learning_rate": 3.5942995719494632e-06,
      "loss": 0.3446,
      "step": 39152
    },
    {
      "epoch": 1.7967509522279839,
      "grad_norm": 0.4479806423187256,
      "learning_rate": 3.5940642749601523e-06,
      "loss": 0.3003,
      "step": 39153
    },
    {
      "epoch": 1.7967968427332384,
      "grad_norm": 0.446286141872406,
      "learning_rate": 3.5938289813516713e-06,
      "loss": 0.3056,
      "step": 39154
    },
    {
      "epoch": 1.7968427332384929,
      "grad_norm": 0.4434756636619568,
      "learning_rate": 3.5935936911245895e-06,
      "loss": 0.3224,
      "step": 39155
    },
    {
      "epoch": 1.7968886237437474,
      "grad_norm": 0.4779556095600128,
      "learning_rate": 3.5933584042794705e-06,
      "loss": 0.3583,
      "step": 39156
    },
    {
      "epoch": 1.7969345142490019,
      "grad_norm": 0.4515618681907654,
      "learning_rate": 3.593123120816878e-06,
      "loss": 0.3115,
      "step": 39157
    },
    {
      "epoch": 1.7969804047542564,
      "grad_norm": 0.502406656742096,
      "learning_rate": 3.5928878407373836e-06,
      "loss": 0.4788,
      "step": 39158
    },
    {
      "epoch": 1.7970262952595109,
      "grad_norm": 0.46805617213249207,
      "learning_rate": 3.5926525640415493e-06,
      "loss": 0.3382,
      "step": 39159
    },
    {
      "epoch": 1.7970721857647654,
      "grad_norm": 0.4763306677341461,
      "learning_rate": 3.5924172907299403e-06,
      "loss": 0.426,
      "step": 39160
    },
    {
      "epoch": 1.7971180762700198,
      "grad_norm": 0.6413353681564331,
      "learning_rate": 3.5921820208031246e-06,
      "loss": 0.3616,
      "step": 39161
    },
    {
      "epoch": 1.7971639667752743,
      "grad_norm": 0.47044244408607483,
      "learning_rate": 3.591946754261667e-06,
      "loss": 0.3395,
      "step": 39162
    },
    {
      "epoch": 1.7972098572805286,
      "grad_norm": 0.48470616340637207,
      "learning_rate": 3.591711491106133e-06,
      "loss": 0.3707,
      "step": 39163
    },
    {
      "epoch": 1.7972557477857831,
      "grad_norm": 0.5120109915733337,
      "learning_rate": 3.591476231337086e-06,
      "loss": 0.3909,
      "step": 39164
    },
    {
      "epoch": 1.7973016382910376,
      "grad_norm": 0.4529309570789337,
      "learning_rate": 3.5912409749550964e-06,
      "loss": 0.3021,
      "step": 39165
    },
    {
      "epoch": 1.797347528796292,
      "grad_norm": 0.45821020007133484,
      "learning_rate": 3.5910057219607275e-06,
      "loss": 0.3116,
      "step": 39166
    },
    {
      "epoch": 1.7973934193015464,
      "grad_norm": 0.6176456809043884,
      "learning_rate": 3.5907704723545443e-06,
      "loss": 0.3359,
      "step": 39167
    },
    {
      "epoch": 1.7974393098068009,
      "grad_norm": 0.5167381167411804,
      "learning_rate": 3.590535226137114e-06,
      "loss": 0.4386,
      "step": 39168
    },
    {
      "epoch": 1.7974852003120554,
      "grad_norm": 0.4389282166957855,
      "learning_rate": 3.5902999833090025e-06,
      "loss": 0.314,
      "step": 39169
    },
    {
      "epoch": 1.7975310908173099,
      "grad_norm": 0.4669302701950073,
      "learning_rate": 3.5900647438707713e-06,
      "loss": 0.2942,
      "step": 39170
    },
    {
      "epoch": 1.7975769813225644,
      "grad_norm": 0.49780312180519104,
      "learning_rate": 3.589829507822993e-06,
      "loss": 0.4061,
      "step": 39171
    },
    {
      "epoch": 1.7976228718278189,
      "grad_norm": 0.5416058897972107,
      "learning_rate": 3.589594275166228e-06,
      "loss": 0.4354,
      "step": 39172
    },
    {
      "epoch": 1.7976687623330734,
      "grad_norm": 0.5023727416992188,
      "learning_rate": 3.5893590459010443e-06,
      "loss": 0.4201,
      "step": 39173
    },
    {
      "epoch": 1.7977146528383279,
      "grad_norm": 0.4762422442436218,
      "learning_rate": 3.5891238200280064e-06,
      "loss": 0.3606,
      "step": 39174
    },
    {
      "epoch": 1.7977605433435824,
      "grad_norm": 0.4768833816051483,
      "learning_rate": 3.588888597547682e-06,
      "loss": 0.363,
      "step": 39175
    },
    {
      "epoch": 1.7978064338488366,
      "grad_norm": 0.43534204363822937,
      "learning_rate": 3.5886533784606327e-06,
      "loss": 0.2905,
      "step": 39176
    },
    {
      "epoch": 1.7978523243540911,
      "grad_norm": 0.48054373264312744,
      "learning_rate": 3.588418162767428e-06,
      "loss": 0.3591,
      "step": 39177
    },
    {
      "epoch": 1.7978982148593456,
      "grad_norm": 0.47345200181007385,
      "learning_rate": 3.588182950468633e-06,
      "loss": 0.3539,
      "step": 39178
    },
    {
      "epoch": 1.7979441053646,
      "grad_norm": 0.47555145621299744,
      "learning_rate": 3.587947741564809e-06,
      "loss": 0.3893,
      "step": 39179
    },
    {
      "epoch": 1.7979899958698544,
      "grad_norm": 0.4615512490272522,
      "learning_rate": 3.5877125360565277e-06,
      "loss": 0.3358,
      "step": 39180
    },
    {
      "epoch": 1.798035886375109,
      "grad_norm": 0.4767516553401947,
      "learning_rate": 3.5874773339443517e-06,
      "loss": 0.3412,
      "step": 39181
    },
    {
      "epoch": 1.7980817768803634,
      "grad_norm": 0.4808574914932251,
      "learning_rate": 3.5872421352288465e-06,
      "loss": 0.3538,
      "step": 39182
    },
    {
      "epoch": 1.798127667385618,
      "grad_norm": 0.4516178071498871,
      "learning_rate": 3.587006939910579e-06,
      "loss": 0.3174,
      "step": 39183
    },
    {
      "epoch": 1.7981735578908724,
      "grad_norm": 0.5146178603172302,
      "learning_rate": 3.586771747990113e-06,
      "loss": 0.4089,
      "step": 39184
    },
    {
      "epoch": 1.798219448396127,
      "grad_norm": 0.48138606548309326,
      "learning_rate": 3.586536559468014e-06,
      "loss": 0.3112,
      "step": 39185
    },
    {
      "epoch": 1.7982653389013814,
      "grad_norm": 0.47517290711402893,
      "learning_rate": 3.5863013743448504e-06,
      "loss": 0.3524,
      "step": 39186
    },
    {
      "epoch": 1.798311229406636,
      "grad_norm": 0.4804094731807709,
      "learning_rate": 3.5860661926211848e-06,
      "loss": 0.3652,
      "step": 39187
    },
    {
      "epoch": 1.7983571199118902,
      "grad_norm": 0.42782214283943176,
      "learning_rate": 3.585831014297584e-06,
      "loss": 0.2905,
      "step": 39188
    },
    {
      "epoch": 1.7984030104171447,
      "grad_norm": 0.46503114700317383,
      "learning_rate": 3.5855958393746137e-06,
      "loss": 0.3095,
      "step": 39189
    },
    {
      "epoch": 1.7984489009223992,
      "grad_norm": 0.4764620363712311,
      "learning_rate": 3.5853606678528387e-06,
      "loss": 0.3768,
      "step": 39190
    },
    {
      "epoch": 1.7984947914276537,
      "grad_norm": 0.5156913995742798,
      "learning_rate": 3.5851254997328233e-06,
      "loss": 0.4519,
      "step": 39191
    },
    {
      "epoch": 1.798540681932908,
      "grad_norm": 0.5220249891281128,
      "learning_rate": 3.5848903350151364e-06,
      "loss": 0.3099,
      "step": 39192
    },
    {
      "epoch": 1.7985865724381624,
      "grad_norm": 0.48183974623680115,
      "learning_rate": 3.5846551737003417e-06,
      "loss": 0.3724,
      "step": 39193
    },
    {
      "epoch": 1.798632462943417,
      "grad_norm": 0.4872881770133972,
      "learning_rate": 3.584420015789003e-06,
      "loss": 0.3658,
      "step": 39194
    },
    {
      "epoch": 1.7986783534486714,
      "grad_norm": 0.45851632952690125,
      "learning_rate": 3.5841848612816892e-06,
      "loss": 0.3264,
      "step": 39195
    },
    {
      "epoch": 1.798724243953926,
      "grad_norm": 0.47330886125564575,
      "learning_rate": 3.583949710178964e-06,
      "loss": 0.372,
      "step": 39196
    },
    {
      "epoch": 1.7987701344591804,
      "grad_norm": 0.47900545597076416,
      "learning_rate": 3.5837145624813923e-06,
      "loss": 0.3632,
      "step": 39197
    },
    {
      "epoch": 1.798816024964435,
      "grad_norm": 0.4801865816116333,
      "learning_rate": 3.5834794181895395e-06,
      "loss": 0.3722,
      "step": 39198
    },
    {
      "epoch": 1.7988619154696894,
      "grad_norm": 0.4970323443412781,
      "learning_rate": 3.583244277303972e-06,
      "loss": 0.3688,
      "step": 39199
    },
    {
      "epoch": 1.798907805974944,
      "grad_norm": 0.4540555477142334,
      "learning_rate": 3.5830091398252554e-06,
      "loss": 0.3148,
      "step": 39200
    },
    {
      "epoch": 1.7989536964801982,
      "grad_norm": 0.43361538648605347,
      "learning_rate": 3.5827740057539524e-06,
      "loss": 0.3166,
      "step": 39201
    },
    {
      "epoch": 1.7989995869854527,
      "grad_norm": 0.4528296887874603,
      "learning_rate": 3.582538875090633e-06,
      "loss": 0.3316,
      "step": 39202
    },
    {
      "epoch": 1.7990454774907072,
      "grad_norm": 0.5063188672065735,
      "learning_rate": 3.5823037478358594e-06,
      "loss": 0.3972,
      "step": 39203
    },
    {
      "epoch": 1.7990913679959615,
      "grad_norm": 0.4593522548675537,
      "learning_rate": 3.5820686239901982e-06,
      "loss": 0.3319,
      "step": 39204
    },
    {
      "epoch": 1.799137258501216,
      "grad_norm": 0.44395455718040466,
      "learning_rate": 3.581833503554214e-06,
      "loss": 0.3233,
      "step": 39205
    },
    {
      "epoch": 1.7991831490064705,
      "grad_norm": 0.46326175332069397,
      "learning_rate": 3.581598386528473e-06,
      "loss": 0.3467,
      "step": 39206
    },
    {
      "epoch": 1.799229039511725,
      "grad_norm": 0.4977503716945648,
      "learning_rate": 3.581363272913539e-06,
      "loss": 0.4487,
      "step": 39207
    },
    {
      "epoch": 1.7992749300169795,
      "grad_norm": 0.46444225311279297,
      "learning_rate": 3.5811281627099803e-06,
      "loss": 0.3423,
      "step": 39208
    },
    {
      "epoch": 1.799320820522234,
      "grad_norm": 0.45476028323173523,
      "learning_rate": 3.58089305591836e-06,
      "loss": 0.3373,
      "step": 39209
    },
    {
      "epoch": 1.7993667110274885,
      "grad_norm": 0.4991852343082428,
      "learning_rate": 3.580657952539244e-06,
      "loss": 0.4728,
      "step": 39210
    },
    {
      "epoch": 1.799412601532743,
      "grad_norm": 0.49933263659477234,
      "learning_rate": 3.5804228525731977e-06,
      "loss": 0.4309,
      "step": 39211
    },
    {
      "epoch": 1.7994584920379975,
      "grad_norm": 0.5126903653144836,
      "learning_rate": 3.5801877560207863e-06,
      "loss": 0.4679,
      "step": 39212
    },
    {
      "epoch": 1.799504382543252,
      "grad_norm": 0.4869692623615265,
      "learning_rate": 3.579952662882574e-06,
      "loss": 0.3939,
      "step": 39213
    },
    {
      "epoch": 1.7995502730485062,
      "grad_norm": 0.4746963381767273,
      "learning_rate": 3.579717573159129e-06,
      "loss": 0.3677,
      "step": 39214
    },
    {
      "epoch": 1.7995961635537607,
      "grad_norm": 0.48424577713012695,
      "learning_rate": 3.579482486851015e-06,
      "loss": 0.3564,
      "step": 39215
    },
    {
      "epoch": 1.7996420540590152,
      "grad_norm": 0.47278037667274475,
      "learning_rate": 3.5792474039587965e-06,
      "loss": 0.3499,
      "step": 39216
    },
    {
      "epoch": 1.7996879445642695,
      "grad_norm": 0.4383920431137085,
      "learning_rate": 3.5790123244830398e-06,
      "loss": 0.3128,
      "step": 39217
    },
    {
      "epoch": 1.799733835069524,
      "grad_norm": 0.5058495998382568,
      "learning_rate": 3.578777248424311e-06,
      "loss": 0.4028,
      "step": 39218
    },
    {
      "epoch": 1.7997797255747785,
      "grad_norm": 0.48429593443870544,
      "learning_rate": 3.5785421757831717e-06,
      "loss": 0.3683,
      "step": 39219
    },
    {
      "epoch": 1.799825616080033,
      "grad_norm": 0.46059587597846985,
      "learning_rate": 3.5783071065601936e-06,
      "loss": 0.3534,
      "step": 39220
    },
    {
      "epoch": 1.7998715065852875,
      "grad_norm": 0.494500070810318,
      "learning_rate": 3.5780720407559366e-06,
      "loss": 0.387,
      "step": 39221
    },
    {
      "epoch": 1.799917397090542,
      "grad_norm": 0.4394480884075165,
      "learning_rate": 3.5778369783709652e-06,
      "loss": 0.3126,
      "step": 39222
    },
    {
      "epoch": 1.7999632875957965,
      "grad_norm": 0.47112613916397095,
      "learning_rate": 3.5776019194058496e-06,
      "loss": 0.392,
      "step": 39223
    },
    {
      "epoch": 1.800009178101051,
      "grad_norm": 0.5653722882270813,
      "learning_rate": 3.5773668638611526e-06,
      "loss": 0.4628,
      "step": 39224
    },
    {
      "epoch": 1.8000550686063055,
      "grad_norm": 0.4742116928100586,
      "learning_rate": 3.5771318117374382e-06,
      "loss": 0.3573,
      "step": 39225
    },
    {
      "epoch": 1.80010095911156,
      "grad_norm": 0.46739962697029114,
      "learning_rate": 3.5768967630352736e-06,
      "loss": 0.3554,
      "step": 39226
    },
    {
      "epoch": 1.8001468496168143,
      "grad_norm": 0.4979287385940552,
      "learning_rate": 3.5766617177552234e-06,
      "loss": 0.4271,
      "step": 39227
    },
    {
      "epoch": 1.8001927401220688,
      "grad_norm": 0.4359785318374634,
      "learning_rate": 3.576426675897852e-06,
      "loss": 0.3175,
      "step": 39228
    },
    {
      "epoch": 1.8002386306273233,
      "grad_norm": 0.4938175678253174,
      "learning_rate": 3.5761916374637237e-06,
      "loss": 0.4281,
      "step": 39229
    },
    {
      "epoch": 1.8002845211325775,
      "grad_norm": 0.47553664445877075,
      "learning_rate": 3.5759566024534066e-06,
      "loss": 0.3468,
      "step": 39230
    },
    {
      "epoch": 1.800330411637832,
      "grad_norm": 0.4512454569339752,
      "learning_rate": 3.5757215708674647e-06,
      "loss": 0.3218,
      "step": 39231
    },
    {
      "epoch": 1.8003763021430865,
      "grad_norm": 0.48382073640823364,
      "learning_rate": 3.5754865427064623e-06,
      "loss": 0.3793,
      "step": 39232
    },
    {
      "epoch": 1.800422192648341,
      "grad_norm": 0.5018970966339111,
      "learning_rate": 3.575251517970966e-06,
      "loss": 0.4885,
      "step": 39233
    },
    {
      "epoch": 1.8004680831535955,
      "grad_norm": 0.45490139722824097,
      "learning_rate": 3.5750164966615397e-06,
      "loss": 0.3094,
      "step": 39234
    },
    {
      "epoch": 1.80051397365885,
      "grad_norm": 0.4589248597621918,
      "learning_rate": 3.5747814787787473e-06,
      "loss": 0.3045,
      "step": 39235
    },
    {
      "epoch": 1.8005598641641045,
      "grad_norm": 0.46435531973838806,
      "learning_rate": 3.574546464323157e-06,
      "loss": 0.3371,
      "step": 39236
    },
    {
      "epoch": 1.800605754669359,
      "grad_norm": 0.44815996289253235,
      "learning_rate": 3.5743114532953335e-06,
      "loss": 0.3346,
      "step": 39237
    },
    {
      "epoch": 1.8006516451746135,
      "grad_norm": 0.45247188210487366,
      "learning_rate": 3.5740764456958395e-06,
      "loss": 0.3192,
      "step": 39238
    },
    {
      "epoch": 1.8006975356798678,
      "grad_norm": 0.5225508809089661,
      "learning_rate": 3.5738414415252424e-06,
      "loss": 0.4473,
      "step": 39239
    },
    {
      "epoch": 1.8007434261851223,
      "grad_norm": 0.44614115357398987,
      "learning_rate": 3.573606440784107e-06,
      "loss": 0.3072,
      "step": 39240
    },
    {
      "epoch": 1.8007893166903768,
      "grad_norm": 0.4881133735179901,
      "learning_rate": 3.573371443472996e-06,
      "loss": 0.3637,
      "step": 39241
    },
    {
      "epoch": 1.800835207195631,
      "grad_norm": 0.5156430602073669,
      "learning_rate": 3.5731364495924793e-06,
      "loss": 0.4287,
      "step": 39242
    },
    {
      "epoch": 1.8008810977008856,
      "grad_norm": 0.4470885992050171,
      "learning_rate": 3.5729014591431177e-06,
      "loss": 0.3381,
      "step": 39243
    },
    {
      "epoch": 1.80092698820614,
      "grad_norm": 0.4537081718444824,
      "learning_rate": 3.5726664721254757e-06,
      "loss": 0.3387,
      "step": 39244
    },
    {
      "epoch": 1.8009728787113946,
      "grad_norm": 0.44785383343696594,
      "learning_rate": 3.572431488540123e-06,
      "loss": 0.3373,
      "step": 39245
    },
    {
      "epoch": 1.801018769216649,
      "grad_norm": 0.4318164587020874,
      "learning_rate": 3.572196508387622e-06,
      "loss": 0.3189,
      "step": 39246
    },
    {
      "epoch": 1.8010646597219035,
      "grad_norm": 0.4455227255821228,
      "learning_rate": 3.571961531668536e-06,
      "loss": 0.3223,
      "step": 39247
    },
    {
      "epoch": 1.801110550227158,
      "grad_norm": 0.4470912516117096,
      "learning_rate": 3.571726558383433e-06,
      "loss": 0.3165,
      "step": 39248
    },
    {
      "epoch": 1.8011564407324125,
      "grad_norm": 0.48179107904434204,
      "learning_rate": 3.5714915885328773e-06,
      "loss": 0.4318,
      "step": 39249
    },
    {
      "epoch": 1.801202331237667,
      "grad_norm": 0.47391849756240845,
      "learning_rate": 3.571256622117432e-06,
      "loss": 0.3477,
      "step": 39250
    },
    {
      "epoch": 1.8012482217429215,
      "grad_norm": 0.47040629386901855,
      "learning_rate": 3.5710216591376646e-06,
      "loss": 0.4035,
      "step": 39251
    },
    {
      "epoch": 1.8012941122481758,
      "grad_norm": 0.4455244243144989,
      "learning_rate": 3.570786699594139e-06,
      "loss": 0.3391,
      "step": 39252
    },
    {
      "epoch": 1.8013400027534303,
      "grad_norm": 0.4383828639984131,
      "learning_rate": 3.5705517434874193e-06,
      "loss": 0.3174,
      "step": 39253
    },
    {
      "epoch": 1.8013858932586848,
      "grad_norm": 0.5050778985023499,
      "learning_rate": 3.5703167908180734e-06,
      "loss": 0.4118,
      "step": 39254
    },
    {
      "epoch": 1.801431783763939,
      "grad_norm": 0.48553112149238586,
      "learning_rate": 3.570081841586664e-06,
      "loss": 0.4189,
      "step": 39255
    },
    {
      "epoch": 1.8014776742691936,
      "grad_norm": 0.479735791683197,
      "learning_rate": 3.569846895793754e-06,
      "loss": 0.3649,
      "step": 39256
    },
    {
      "epoch": 1.801523564774448,
      "grad_norm": 0.4841137230396271,
      "learning_rate": 3.569611953439913e-06,
      "loss": 0.3941,
      "step": 39257
    },
    {
      "epoch": 1.8015694552797026,
      "grad_norm": 0.4801064729690552,
      "learning_rate": 3.5693770145257046e-06,
      "loss": 0.3781,
      "step": 39258
    },
    {
      "epoch": 1.801615345784957,
      "grad_norm": 0.4492424428462982,
      "learning_rate": 3.569142079051691e-06,
      "loss": 0.2972,
      "step": 39259
    },
    {
      "epoch": 1.8016612362902116,
      "grad_norm": 0.45552802085876465,
      "learning_rate": 3.568907147018441e-06,
      "loss": 0.3026,
      "step": 39260
    },
    {
      "epoch": 1.801707126795466,
      "grad_norm": 0.4365113079547882,
      "learning_rate": 3.5686722184265167e-06,
      "loss": 0.3017,
      "step": 39261
    },
    {
      "epoch": 1.8017530173007206,
      "grad_norm": 0.4780440032482147,
      "learning_rate": 3.5684372932764843e-06,
      "loss": 0.3656,
      "step": 39262
    },
    {
      "epoch": 1.801798907805975,
      "grad_norm": 0.4473513960838318,
      "learning_rate": 3.568202371568907e-06,
      "loss": 0.3178,
      "step": 39263
    },
    {
      "epoch": 1.8018447983112296,
      "grad_norm": 0.4802020192146301,
      "learning_rate": 3.5679674533043533e-06,
      "loss": 0.3675,
      "step": 39264
    },
    {
      "epoch": 1.8018906888164838,
      "grad_norm": 0.44099804759025574,
      "learning_rate": 3.5677325384833854e-06,
      "loss": 0.2683,
      "step": 39265
    },
    {
      "epoch": 1.8019365793217383,
      "grad_norm": 0.4765324294567108,
      "learning_rate": 3.5674976271065666e-06,
      "loss": 0.3488,
      "step": 39266
    },
    {
      "epoch": 1.8019824698269928,
      "grad_norm": 0.49827513098716736,
      "learning_rate": 3.5672627191744656e-06,
      "loss": 0.3293,
      "step": 39267
    },
    {
      "epoch": 1.8020283603322471,
      "grad_norm": 0.48104366660118103,
      "learning_rate": 3.5670278146876457e-06,
      "loss": 0.351,
      "step": 39268
    },
    {
      "epoch": 1.8020742508375016,
      "grad_norm": 0.45449671149253845,
      "learning_rate": 3.566792913646671e-06,
      "loss": 0.3244,
      "step": 39269
    },
    {
      "epoch": 1.802120141342756,
      "grad_norm": 0.48197853565216064,
      "learning_rate": 3.566558016052108e-06,
      "loss": 0.3993,
      "step": 39270
    },
    {
      "epoch": 1.8021660318480106,
      "grad_norm": 0.4611546993255615,
      "learning_rate": 3.56632312190452e-06,
      "loss": 0.3567,
      "step": 39271
    },
    {
      "epoch": 1.802211922353265,
      "grad_norm": 0.533311128616333,
      "learning_rate": 3.566088231204471e-06,
      "loss": 0.4052,
      "step": 39272
    },
    {
      "epoch": 1.8022578128585196,
      "grad_norm": 0.4383503496646881,
      "learning_rate": 3.565853343952529e-06,
      "loss": 0.3416,
      "step": 39273
    },
    {
      "epoch": 1.802303703363774,
      "grad_norm": 0.4511098265647888,
      "learning_rate": 3.565618460149257e-06,
      "loss": 0.366,
      "step": 39274
    },
    {
      "epoch": 1.8023495938690286,
      "grad_norm": 0.41594845056533813,
      "learning_rate": 3.565383579795219e-06,
      "loss": 0.2797,
      "step": 39275
    },
    {
      "epoch": 1.802395484374283,
      "grad_norm": 0.4979259967803955,
      "learning_rate": 3.5651487028909815e-06,
      "loss": 0.4277,
      "step": 39276
    },
    {
      "epoch": 1.8024413748795374,
      "grad_norm": 0.512932300567627,
      "learning_rate": 3.564913829437109e-06,
      "loss": 0.4053,
      "step": 39277
    },
    {
      "epoch": 1.8024872653847919,
      "grad_norm": 0.5362216234207153,
      "learning_rate": 3.5646789594341636e-06,
      "loss": 0.4779,
      "step": 39278
    },
    {
      "epoch": 1.8025331558900464,
      "grad_norm": 0.4687674343585968,
      "learning_rate": 3.564444092882714e-06,
      "loss": 0.3484,
      "step": 39279
    },
    {
      "epoch": 1.8025790463953009,
      "grad_norm": 0.48018693923950195,
      "learning_rate": 3.564209229783324e-06,
      "loss": 0.3652,
      "step": 39280
    },
    {
      "epoch": 1.8026249369005551,
      "grad_norm": 0.5012412071228027,
      "learning_rate": 3.5639743701365558e-06,
      "loss": 0.3301,
      "step": 39281
    },
    {
      "epoch": 1.8026708274058096,
      "grad_norm": 0.4927400052547455,
      "learning_rate": 3.5637395139429776e-06,
      "loss": 0.4132,
      "step": 39282
    },
    {
      "epoch": 1.8027167179110641,
      "grad_norm": 0.488741397857666,
      "learning_rate": 3.563504661203152e-06,
      "loss": 0.346,
      "step": 39283
    },
    {
      "epoch": 1.8027626084163186,
      "grad_norm": 0.45236778259277344,
      "learning_rate": 3.563269811917642e-06,
      "loss": 0.318,
      "step": 39284
    },
    {
      "epoch": 1.8028084989215731,
      "grad_norm": 0.43687424063682556,
      "learning_rate": 3.563034966087018e-06,
      "loss": 0.3018,
      "step": 39285
    },
    {
      "epoch": 1.8028543894268276,
      "grad_norm": 0.449699342250824,
      "learning_rate": 3.5628001237118415e-06,
      "loss": 0.3258,
      "step": 39286
    },
    {
      "epoch": 1.8029002799320821,
      "grad_norm": 0.49896955490112305,
      "learning_rate": 3.5625652847926745e-06,
      "loss": 0.3916,
      "step": 39287
    },
    {
      "epoch": 1.8029461704373366,
      "grad_norm": 0.4704723060131073,
      "learning_rate": 3.5623304493300858e-06,
      "loss": 0.3645,
      "step": 39288
    },
    {
      "epoch": 1.8029920609425911,
      "grad_norm": 0.4715275466442108,
      "learning_rate": 3.5620956173246396e-06,
      "loss": 0.3821,
      "step": 39289
    },
    {
      "epoch": 1.8030379514478454,
      "grad_norm": 0.44830822944641113,
      "learning_rate": 3.5618607887768976e-06,
      "loss": 0.359,
      "step": 39290
    },
    {
      "epoch": 1.8030838419531,
      "grad_norm": 0.48768895864486694,
      "learning_rate": 3.5616259636874284e-06,
      "loss": 0.3403,
      "step": 39291
    },
    {
      "epoch": 1.8031297324583544,
      "grad_norm": 0.46323302388191223,
      "learning_rate": 3.5613911420567937e-06,
      "loss": 0.3589,
      "step": 39292
    },
    {
      "epoch": 1.8031756229636087,
      "grad_norm": 0.4593289792537689,
      "learning_rate": 3.5611563238855606e-06,
      "loss": 0.3523,
      "step": 39293
    },
    {
      "epoch": 1.8032215134688632,
      "grad_norm": 0.4646664559841156,
      "learning_rate": 3.5609215091742894e-06,
      "loss": 0.3654,
      "step": 39294
    },
    {
      "epoch": 1.8032674039741177,
      "grad_norm": 0.47845137119293213,
      "learning_rate": 3.56068669792355e-06,
      "loss": 0.402,
      "step": 39295
    },
    {
      "epoch": 1.8033132944793722,
      "grad_norm": 0.45610469579696655,
      "learning_rate": 3.5604518901339047e-06,
      "loss": 0.3212,
      "step": 39296
    },
    {
      "epoch": 1.8033591849846267,
      "grad_norm": 0.45342209935188293,
      "learning_rate": 3.560217085805918e-06,
      "loss": 0.3303,
      "step": 39297
    },
    {
      "epoch": 1.8034050754898812,
      "grad_norm": 0.4680362045764923,
      "learning_rate": 3.559982284940155e-06,
      "loss": 0.3553,
      "step": 39298
    },
    {
      "epoch": 1.8034509659951357,
      "grad_norm": 0.4510551691055298,
      "learning_rate": 3.5597474875371807e-06,
      "loss": 0.3319,
      "step": 39299
    },
    {
      "epoch": 1.8034968565003902,
      "grad_norm": 0.4679715037345886,
      "learning_rate": 3.5595126935975565e-06,
      "loss": 0.3931,
      "step": 39300
    },
    {
      "epoch": 1.8035427470056447,
      "grad_norm": 0.4478391408920288,
      "learning_rate": 3.5592779031218516e-06,
      "loss": 0.3035,
      "step": 39301
    },
    {
      "epoch": 1.8035886375108992,
      "grad_norm": 0.4669497311115265,
      "learning_rate": 3.55904311611063e-06,
      "loss": 0.3131,
      "step": 39302
    },
    {
      "epoch": 1.8036345280161534,
      "grad_norm": 0.46841856837272644,
      "learning_rate": 3.5588083325644525e-06,
      "loss": 0.3568,
      "step": 39303
    },
    {
      "epoch": 1.803680418521408,
      "grad_norm": 0.4601190686225891,
      "learning_rate": 3.558573552483887e-06,
      "loss": 0.3624,
      "step": 39304
    },
    {
      "epoch": 1.8037263090266624,
      "grad_norm": 0.4474479556083679,
      "learning_rate": 3.558338775869498e-06,
      "loss": 0.3176,
      "step": 39305
    },
    {
      "epoch": 1.8037721995319167,
      "grad_norm": 0.49226072430610657,
      "learning_rate": 3.558104002721847e-06,
      "loss": 0.4149,
      "step": 39306
    },
    {
      "epoch": 1.8038180900371712,
      "grad_norm": 0.4580143392086029,
      "learning_rate": 3.5578692330415033e-06,
      "loss": 0.3644,
      "step": 39307
    },
    {
      "epoch": 1.8038639805424257,
      "grad_norm": 0.47971004247665405,
      "learning_rate": 3.557634466829029e-06,
      "loss": 0.3454,
      "step": 39308
    },
    {
      "epoch": 1.8039098710476802,
      "grad_norm": 0.4903052747249603,
      "learning_rate": 3.557399704084986e-06,
      "loss": 0.3939,
      "step": 39309
    },
    {
      "epoch": 1.8039557615529347,
      "grad_norm": 0.4632745683193207,
      "learning_rate": 3.5571649448099434e-06,
      "loss": 0.3599,
      "step": 39310
    },
    {
      "epoch": 1.8040016520581892,
      "grad_norm": 0.4452897608280182,
      "learning_rate": 3.5569301890044638e-06,
      "loss": 0.3073,
      "step": 39311
    },
    {
      "epoch": 1.8040475425634437,
      "grad_norm": 0.48747268319129944,
      "learning_rate": 3.556695436669111e-06,
      "loss": 0.3791,
      "step": 39312
    },
    {
      "epoch": 1.8040934330686982,
      "grad_norm": 0.46609342098236084,
      "learning_rate": 3.5564606878044505e-06,
      "loss": 0.3628,
      "step": 39313
    },
    {
      "epoch": 1.8041393235739527,
      "grad_norm": 0.4931948184967041,
      "learning_rate": 3.5562259424110467e-06,
      "loss": 0.4007,
      "step": 39314
    },
    {
      "epoch": 1.8041852140792072,
      "grad_norm": 0.43707239627838135,
      "learning_rate": 3.5559912004894616e-06,
      "loss": 0.3075,
      "step": 39315
    },
    {
      "epoch": 1.8042311045844615,
      "grad_norm": 0.46779289841651917,
      "learning_rate": 3.555756462040264e-06,
      "loss": 0.3916,
      "step": 39316
    },
    {
      "epoch": 1.804276995089716,
      "grad_norm": 0.4659048020839691,
      "learning_rate": 3.555521727064017e-06,
      "loss": 0.3492,
      "step": 39317
    },
    {
      "epoch": 1.8043228855949704,
      "grad_norm": 0.4051838219165802,
      "learning_rate": 3.555286995561282e-06,
      "loss": 0.2534,
      "step": 39318
    },
    {
      "epoch": 1.8043687761002247,
      "grad_norm": 0.4775054454803467,
      "learning_rate": 3.555052267532628e-06,
      "loss": 0.3838,
      "step": 39319
    },
    {
      "epoch": 1.8044146666054792,
      "grad_norm": 0.49684667587280273,
      "learning_rate": 3.554817542978617e-06,
      "loss": 0.38,
      "step": 39320
    },
    {
      "epoch": 1.8044605571107337,
      "grad_norm": 0.47997039556503296,
      "learning_rate": 3.5545828218998112e-06,
      "loss": 0.3848,
      "step": 39321
    },
    {
      "epoch": 1.8045064476159882,
      "grad_norm": 0.476949542760849,
      "learning_rate": 3.5543481042967805e-06,
      "loss": 0.3535,
      "step": 39322
    },
    {
      "epoch": 1.8045523381212427,
      "grad_norm": 0.466804563999176,
      "learning_rate": 3.554113390170085e-06,
      "loss": 0.3486,
      "step": 39323
    },
    {
      "epoch": 1.8045982286264972,
      "grad_norm": 0.47255849838256836,
      "learning_rate": 3.5538786795202907e-06,
      "loss": 0.3825,
      "step": 39324
    },
    {
      "epoch": 1.8046441191317517,
      "grad_norm": 0.4365609288215637,
      "learning_rate": 3.553643972347962e-06,
      "loss": 0.2882,
      "step": 39325
    },
    {
      "epoch": 1.8046900096370062,
      "grad_norm": 0.475857138633728,
      "learning_rate": 3.5534092686536637e-06,
      "loss": 0.3609,
      "step": 39326
    },
    {
      "epoch": 1.8047359001422607,
      "grad_norm": 0.4938315749168396,
      "learning_rate": 3.5531745684379594e-06,
      "loss": 0.4078,
      "step": 39327
    },
    {
      "epoch": 1.804781790647515,
      "grad_norm": 0.44474807381629944,
      "learning_rate": 3.5529398717014113e-06,
      "loss": 0.3269,
      "step": 39328
    },
    {
      "epoch": 1.8048276811527695,
      "grad_norm": 0.4790645241737366,
      "learning_rate": 3.5527051784445884e-06,
      "loss": 0.36,
      "step": 39329
    },
    {
      "epoch": 1.804873571658024,
      "grad_norm": 0.471356600522995,
      "learning_rate": 3.552470488668054e-06,
      "loss": 0.3692,
      "step": 39330
    },
    {
      "epoch": 1.8049194621632783,
      "grad_norm": 0.48099663853645325,
      "learning_rate": 3.5522358023723672e-06,
      "loss": 0.3841,
      "step": 39331
    },
    {
      "epoch": 1.8049653526685328,
      "grad_norm": 0.4951941668987274,
      "learning_rate": 3.5520011195580994e-06,
      "loss": 0.423,
      "step": 39332
    },
    {
      "epoch": 1.8050112431737872,
      "grad_norm": 0.43567320704460144,
      "learning_rate": 3.551766440225811e-06,
      "loss": 0.3142,
      "step": 39333
    },
    {
      "epoch": 1.8050571336790417,
      "grad_norm": 0.44170209765434265,
      "learning_rate": 3.551531764376067e-06,
      "loss": 0.3099,
      "step": 39334
    },
    {
      "epoch": 1.8051030241842962,
      "grad_norm": 0.4575502574443817,
      "learning_rate": 3.5512970920094324e-06,
      "loss": 0.3487,
      "step": 39335
    },
    {
      "epoch": 1.8051489146895507,
      "grad_norm": 0.44578108191490173,
      "learning_rate": 3.5510624231264717e-06,
      "loss": 0.3134,
      "step": 39336
    },
    {
      "epoch": 1.8051948051948052,
      "grad_norm": 0.5043043494224548,
      "learning_rate": 3.5508277577277463e-06,
      "loss": 0.3656,
      "step": 39337
    },
    {
      "epoch": 1.8052406957000597,
      "grad_norm": 0.4623585343360901,
      "learning_rate": 3.5505930958138246e-06,
      "loss": 0.3224,
      "step": 39338
    },
    {
      "epoch": 1.8052865862053142,
      "grad_norm": 0.4920845627784729,
      "learning_rate": 3.5503584373852694e-06,
      "loss": 0.3543,
      "step": 39339
    },
    {
      "epoch": 1.8053324767105687,
      "grad_norm": 0.4601145386695862,
      "learning_rate": 3.550123782442643e-06,
      "loss": 0.3085,
      "step": 39340
    },
    {
      "epoch": 1.805378367215823,
      "grad_norm": 0.4655556082725525,
      "learning_rate": 3.549889130986513e-06,
      "loss": 0.3584,
      "step": 39341
    },
    {
      "epoch": 1.8054242577210775,
      "grad_norm": 0.4654712975025177,
      "learning_rate": 3.549654483017442e-06,
      "loss": 0.3731,
      "step": 39342
    },
    {
      "epoch": 1.805470148226332,
      "grad_norm": 0.4468616247177124,
      "learning_rate": 3.5494198385359915e-06,
      "loss": 0.2959,
      "step": 39343
    },
    {
      "epoch": 1.8055160387315863,
      "grad_norm": 0.4409073293209076,
      "learning_rate": 3.5491851975427307e-06,
      "loss": 0.3208,
      "step": 39344
    },
    {
      "epoch": 1.8055619292368408,
      "grad_norm": 0.44044968485832214,
      "learning_rate": 3.548950560038222e-06,
      "loss": 0.3097,
      "step": 39345
    },
    {
      "epoch": 1.8056078197420953,
      "grad_norm": 0.486675500869751,
      "learning_rate": 3.5487159260230274e-06,
      "loss": 0.36,
      "step": 39346
    },
    {
      "epoch": 1.8056537102473498,
      "grad_norm": 0.4384382665157318,
      "learning_rate": 3.5484812954977153e-06,
      "loss": 0.3169,
      "step": 39347
    },
    {
      "epoch": 1.8056996007526043,
      "grad_norm": 0.525726854801178,
      "learning_rate": 3.548246668462847e-06,
      "loss": 0.4639,
      "step": 39348
    },
    {
      "epoch": 1.8057454912578588,
      "grad_norm": 0.5005178451538086,
      "learning_rate": 3.548012044918985e-06,
      "loss": 0.4902,
      "step": 39349
    },
    {
      "epoch": 1.8057913817631133,
      "grad_norm": 0.4615524113178253,
      "learning_rate": 3.5477774248666972e-06,
      "loss": 0.3306,
      "step": 39350
    },
    {
      "epoch": 1.8058372722683678,
      "grad_norm": 0.44902145862579346,
      "learning_rate": 3.5475428083065478e-06,
      "loss": 0.3294,
      "step": 39351
    },
    {
      "epoch": 1.8058831627736223,
      "grad_norm": 0.4812130928039551,
      "learning_rate": 3.5473081952390976e-06,
      "loss": 0.4089,
      "step": 39352
    },
    {
      "epoch": 1.8059290532788768,
      "grad_norm": 0.46974262595176697,
      "learning_rate": 3.547073585664914e-06,
      "loss": 0.3825,
      "step": 39353
    },
    {
      "epoch": 1.805974943784131,
      "grad_norm": 0.4758954346179962,
      "learning_rate": 3.54683897958456e-06,
      "loss": 0.3981,
      "step": 39354
    },
    {
      "epoch": 1.8060208342893855,
      "grad_norm": 0.467261403799057,
      "learning_rate": 3.546604376998597e-06,
      "loss": 0.3829,
      "step": 39355
    },
    {
      "epoch": 1.80606672479464,
      "grad_norm": 0.47148534655570984,
      "learning_rate": 3.546369777907596e-06,
      "loss": 0.3446,
      "step": 39356
    },
    {
      "epoch": 1.8061126152998943,
      "grad_norm": 0.48792725801467896,
      "learning_rate": 3.5461351823121144e-06,
      "loss": 0.3627,
      "step": 39357
    },
    {
      "epoch": 1.8061585058051488,
      "grad_norm": 0.49115055799484253,
      "learning_rate": 3.5459005902127196e-06,
      "loss": 0.3435,
      "step": 39358
    },
    {
      "epoch": 1.8062043963104033,
      "grad_norm": 0.5524871945381165,
      "learning_rate": 3.545666001609973e-06,
      "loss": 0.26,
      "step": 39359
    },
    {
      "epoch": 1.8062502868156578,
      "grad_norm": 0.47939959168434143,
      "learning_rate": 3.5454314165044424e-06,
      "loss": 0.395,
      "step": 39360
    },
    {
      "epoch": 1.8062961773209123,
      "grad_norm": 0.44979482889175415,
      "learning_rate": 3.5451968348966904e-06,
      "loss": 0.2927,
      "step": 39361
    },
    {
      "epoch": 1.8063420678261668,
      "grad_norm": 0.5196375846862793,
      "learning_rate": 3.54496225678728e-06,
      "loss": 0.4358,
      "step": 39362
    },
    {
      "epoch": 1.8063879583314213,
      "grad_norm": 0.47665971517562866,
      "learning_rate": 3.544727682176776e-06,
      "loss": 0.3448,
      "step": 39363
    },
    {
      "epoch": 1.8064338488366758,
      "grad_norm": 0.4489773213863373,
      "learning_rate": 3.5444931110657444e-06,
      "loss": 0.3692,
      "step": 39364
    },
    {
      "epoch": 1.8064797393419303,
      "grad_norm": 0.46411487460136414,
      "learning_rate": 3.5442585434547438e-06,
      "loss": 0.3509,
      "step": 39365
    },
    {
      "epoch": 1.8065256298471846,
      "grad_norm": 0.46334949135780334,
      "learning_rate": 3.5440239793443453e-06,
      "loss": 0.3336,
      "step": 39366
    },
    {
      "epoch": 1.806571520352439,
      "grad_norm": 0.5342000722885132,
      "learning_rate": 3.543789418735109e-06,
      "loss": 0.3344,
      "step": 39367
    },
    {
      "epoch": 1.8066174108576936,
      "grad_norm": 0.46767657995224,
      "learning_rate": 3.5435548616275984e-06,
      "loss": 0.3667,
      "step": 39368
    },
    {
      "epoch": 1.806663301362948,
      "grad_norm": 0.48962831497192383,
      "learning_rate": 3.54332030802238e-06,
      "loss": 0.4198,
      "step": 39369
    },
    {
      "epoch": 1.8067091918682023,
      "grad_norm": 0.4334319829940796,
      "learning_rate": 3.5430857579200166e-06,
      "loss": 0.3215,
      "step": 39370
    },
    {
      "epoch": 1.8067550823734568,
      "grad_norm": 0.47955387830734253,
      "learning_rate": 3.54285121132107e-06,
      "loss": 0.3411,
      "step": 39371
    },
    {
      "epoch": 1.8068009728787113,
      "grad_norm": 0.4400034546852112,
      "learning_rate": 3.5426166682261088e-06,
      "loss": 0.3161,
      "step": 39372
    },
    {
      "epoch": 1.8068468633839658,
      "grad_norm": 0.4896361827850342,
      "learning_rate": 3.542382128635694e-06,
      "loss": 0.3996,
      "step": 39373
    },
    {
      "epoch": 1.8068927538892203,
      "grad_norm": 0.4794846475124359,
      "learning_rate": 3.542147592550389e-06,
      "loss": 0.3863,
      "step": 39374
    },
    {
      "epoch": 1.8069386443944748,
      "grad_norm": 0.40737876296043396,
      "learning_rate": 3.5419130599707603e-06,
      "loss": 0.2705,
      "step": 39375
    },
    {
      "epoch": 1.8069845348997293,
      "grad_norm": 0.5043819546699524,
      "learning_rate": 3.54167853089737e-06,
      "loss": 0.3775,
      "step": 39376
    },
    {
      "epoch": 1.8070304254049838,
      "grad_norm": 0.4714910686016083,
      "learning_rate": 3.5414440053307807e-06,
      "loss": 0.364,
      "step": 39377
    },
    {
      "epoch": 1.8070763159102383,
      "grad_norm": 0.46037811040878296,
      "learning_rate": 3.541209483271562e-06,
      "loss": 0.3614,
      "step": 39378
    },
    {
      "epoch": 1.8071222064154926,
      "grad_norm": 0.46615785360336304,
      "learning_rate": 3.540974964720273e-06,
      "loss": 0.3823,
      "step": 39379
    },
    {
      "epoch": 1.807168096920747,
      "grad_norm": 0.45109298825263977,
      "learning_rate": 3.5407404496774755e-06,
      "loss": 0.34,
      "step": 39380
    },
    {
      "epoch": 1.8072139874260016,
      "grad_norm": 0.43808576464653015,
      "learning_rate": 3.54050593814374e-06,
      "loss": 0.3272,
      "step": 39381
    },
    {
      "epoch": 1.8072598779312559,
      "grad_norm": 0.4434596002101898,
      "learning_rate": 3.540271430119626e-06,
      "loss": 0.3229,
      "step": 39382
    },
    {
      "epoch": 1.8073057684365104,
      "grad_norm": 0.459803968667984,
      "learning_rate": 3.540036925605699e-06,
      "loss": 0.3423,
      "step": 39383
    },
    {
      "epoch": 1.8073516589417649,
      "grad_norm": 0.49817344546318054,
      "learning_rate": 3.539802424602522e-06,
      "loss": 0.4044,
      "step": 39384
    },
    {
      "epoch": 1.8073975494470194,
      "grad_norm": 0.4696045219898224,
      "learning_rate": 3.53956792711066e-06,
      "loss": 0.3688,
      "step": 39385
    },
    {
      "epoch": 1.8074434399522739,
      "grad_norm": 0.46462154388427734,
      "learning_rate": 3.539333433130674e-06,
      "loss": 0.3407,
      "step": 39386
    },
    {
      "epoch": 1.8074893304575284,
      "grad_norm": 0.46397772431373596,
      "learning_rate": 3.539098942663132e-06,
      "loss": 0.3251,
      "step": 39387
    },
    {
      "epoch": 1.8075352209627829,
      "grad_norm": 0.47480034828186035,
      "learning_rate": 3.538864455708596e-06,
      "loss": 0.3022,
      "step": 39388
    },
    {
      "epoch": 1.8075811114680373,
      "grad_norm": 0.4780513644218445,
      "learning_rate": 3.5386299722676287e-06,
      "loss": 0.3282,
      "step": 39389
    },
    {
      "epoch": 1.8076270019732918,
      "grad_norm": 0.5047063231468201,
      "learning_rate": 3.538395492340796e-06,
      "loss": 0.3823,
      "step": 39390
    },
    {
      "epoch": 1.8076728924785463,
      "grad_norm": 0.4939988851547241,
      "learning_rate": 3.5381610159286615e-06,
      "loss": 0.3479,
      "step": 39391
    },
    {
      "epoch": 1.8077187829838006,
      "grad_norm": 0.480477511882782,
      "learning_rate": 3.5379265430317877e-06,
      "loss": 0.3687,
      "step": 39392
    },
    {
      "epoch": 1.8077646734890551,
      "grad_norm": 0.5041800737380981,
      "learning_rate": 3.537692073650738e-06,
      "loss": 0.3896,
      "step": 39393
    },
    {
      "epoch": 1.8078105639943096,
      "grad_norm": 0.46361982822418213,
      "learning_rate": 3.5374576077860784e-06,
      "loss": 0.3301,
      "step": 39394
    },
    {
      "epoch": 1.807856454499564,
      "grad_norm": 0.46986833214759827,
      "learning_rate": 3.537223145438372e-06,
      "loss": 0.4069,
      "step": 39395
    },
    {
      "epoch": 1.8079023450048184,
      "grad_norm": 0.45814529061317444,
      "learning_rate": 3.5369886866081815e-06,
      "loss": 0.3544,
      "step": 39396
    },
    {
      "epoch": 1.8079482355100729,
      "grad_norm": 0.4591224193572998,
      "learning_rate": 3.536754231296072e-06,
      "loss": 0.3537,
      "step": 39397
    },
    {
      "epoch": 1.8079941260153274,
      "grad_norm": 0.450774222612381,
      "learning_rate": 3.5365197795026075e-06,
      "loss": 0.3323,
      "step": 39398
    },
    {
      "epoch": 1.8080400165205819,
      "grad_norm": 0.469013512134552,
      "learning_rate": 3.5362853312283484e-06,
      "loss": 0.3306,
      "step": 39399
    },
    {
      "epoch": 1.8080859070258364,
      "grad_norm": 0.49245598912239075,
      "learning_rate": 3.536050886473864e-06,
      "loss": 0.4283,
      "step": 39400
    },
    {
      "epoch": 1.8081317975310909,
      "grad_norm": 0.45118993520736694,
      "learning_rate": 3.535816445239715e-06,
      "loss": 0.3218,
      "step": 39401
    },
    {
      "epoch": 1.8081776880363454,
      "grad_norm": 0.46114370226860046,
      "learning_rate": 3.5355820075264625e-06,
      "loss": 0.3748,
      "step": 39402
    },
    {
      "epoch": 1.8082235785415999,
      "grad_norm": 0.4903721809387207,
      "learning_rate": 3.5353475733346754e-06,
      "loss": 0.423,
      "step": 39403
    },
    {
      "epoch": 1.8082694690468544,
      "grad_norm": 0.4909888803958893,
      "learning_rate": 3.5351131426649156e-06,
      "loss": 0.3463,
      "step": 39404
    },
    {
      "epoch": 1.8083153595521086,
      "grad_norm": 0.4454742968082428,
      "learning_rate": 3.5348787155177444e-06,
      "loss": 0.2991,
      "step": 39405
    },
    {
      "epoch": 1.8083612500573631,
      "grad_norm": 0.4895438253879547,
      "learning_rate": 3.5346442918937297e-06,
      "loss": 0.4121,
      "step": 39406
    },
    {
      "epoch": 1.8084071405626176,
      "grad_norm": 0.4469958245754242,
      "learning_rate": 3.5344098717934318e-06,
      "loss": 0.3253,
      "step": 39407
    },
    {
      "epoch": 1.808453031067872,
      "grad_norm": 0.4424133598804474,
      "learning_rate": 3.534175455217414e-06,
      "loss": 0.3299,
      "step": 39408
    },
    {
      "epoch": 1.8084989215731264,
      "grad_norm": 0.44898784160614014,
      "learning_rate": 3.5339410421662435e-06,
      "loss": 0.3382,
      "step": 39409
    },
    {
      "epoch": 1.808544812078381,
      "grad_norm": 0.45608261227607727,
      "learning_rate": 3.5337066326404823e-06,
      "loss": 0.3448,
      "step": 39410
    },
    {
      "epoch": 1.8085907025836354,
      "grad_norm": 0.4978122413158417,
      "learning_rate": 3.5334722266406927e-06,
      "loss": 0.3897,
      "step": 39411
    },
    {
      "epoch": 1.80863659308889,
      "grad_norm": 0.5066713094711304,
      "learning_rate": 3.5332378241674414e-06,
      "loss": 0.382,
      "step": 39412
    },
    {
      "epoch": 1.8086824835941444,
      "grad_norm": 0.4606563150882721,
      "learning_rate": 3.53300342522129e-06,
      "loss": 0.3644,
      "step": 39413
    },
    {
      "epoch": 1.808728374099399,
      "grad_norm": 0.4733734726905823,
      "learning_rate": 3.5327690298027993e-06,
      "loss": 0.3553,
      "step": 39414
    },
    {
      "epoch": 1.8087742646046534,
      "grad_norm": 0.4480231702327728,
      "learning_rate": 3.5325346379125385e-06,
      "loss": 0.3402,
      "step": 39415
    },
    {
      "epoch": 1.808820155109908,
      "grad_norm": 0.4656098783016205,
      "learning_rate": 3.532300249551069e-06,
      "loss": 0.356,
      "step": 39416
    },
    {
      "epoch": 1.8088660456151622,
      "grad_norm": 0.46409595012664795,
      "learning_rate": 3.5320658647189534e-06,
      "loss": 0.35,
      "step": 39417
    },
    {
      "epoch": 1.8089119361204167,
      "grad_norm": 0.49375781416893005,
      "learning_rate": 3.5318314834167565e-06,
      "loss": 0.3652,
      "step": 39418
    },
    {
      "epoch": 1.8089578266256712,
      "grad_norm": 0.46811777353286743,
      "learning_rate": 3.5315971056450416e-06,
      "loss": 0.3565,
      "step": 39419
    },
    {
      "epoch": 1.8090037171309254,
      "grad_norm": 0.45803651213645935,
      "learning_rate": 3.5313627314043703e-06,
      "loss": 0.3253,
      "step": 39420
    },
    {
      "epoch": 1.80904960763618,
      "grad_norm": 0.48711204528808594,
      "learning_rate": 3.53112836069531e-06,
      "loss": 0.3519,
      "step": 39421
    },
    {
      "epoch": 1.8090954981414344,
      "grad_norm": 0.42546218633651733,
      "learning_rate": 3.530893993518424e-06,
      "loss": 0.3217,
      "step": 39422
    },
    {
      "epoch": 1.809141388646689,
      "grad_norm": 0.46847471594810486,
      "learning_rate": 3.53065962987427e-06,
      "loss": 0.3682,
      "step": 39423
    },
    {
      "epoch": 1.8091872791519434,
      "grad_norm": 0.44000476598739624,
      "learning_rate": 3.530425269763419e-06,
      "loss": 0.3155,
      "step": 39424
    },
    {
      "epoch": 1.809233169657198,
      "grad_norm": 0.4481617212295532,
      "learning_rate": 3.5301909131864316e-06,
      "loss": 0.3478,
      "step": 39425
    },
    {
      "epoch": 1.8092790601624524,
      "grad_norm": 0.4471161961555481,
      "learning_rate": 3.5299565601438707e-06,
      "loss": 0.334,
      "step": 39426
    },
    {
      "epoch": 1.809324950667707,
      "grad_norm": 0.44624075293540955,
      "learning_rate": 3.5297222106362994e-06,
      "loss": 0.3,
      "step": 39427
    },
    {
      "epoch": 1.8093708411729614,
      "grad_norm": 0.46490681171417236,
      "learning_rate": 3.529487864664283e-06,
      "loss": 0.3534,
      "step": 39428
    },
    {
      "epoch": 1.809416731678216,
      "grad_norm": 0.4589875638484955,
      "learning_rate": 3.5292535222283842e-06,
      "loss": 0.31,
      "step": 39429
    },
    {
      "epoch": 1.8094626221834702,
      "grad_norm": 0.49487200379371643,
      "learning_rate": 3.5290191833291644e-06,
      "loss": 0.3905,
      "step": 39430
    },
    {
      "epoch": 1.8095085126887247,
      "grad_norm": 0.4707028269767761,
      "learning_rate": 3.5287848479671915e-06,
      "loss": 0.3588,
      "step": 39431
    },
    {
      "epoch": 1.8095544031939792,
      "grad_norm": 0.5007294416427612,
      "learning_rate": 3.5285505161430266e-06,
      "loss": 0.4416,
      "step": 39432
    },
    {
      "epoch": 1.8096002936992335,
      "grad_norm": 0.4758753478527069,
      "learning_rate": 3.5283161878572318e-06,
      "loss": 0.3899,
      "step": 39433
    },
    {
      "epoch": 1.809646184204488,
      "grad_norm": 0.4538108706474304,
      "learning_rate": 3.528081863110374e-06,
      "loss": 0.3132,
      "step": 39434
    },
    {
      "epoch": 1.8096920747097425,
      "grad_norm": 0.4754985570907593,
      "learning_rate": 3.5278475419030145e-06,
      "loss": 0.3737,
      "step": 39435
    },
    {
      "epoch": 1.809737965214997,
      "grad_norm": 0.46852758526802063,
      "learning_rate": 3.527613224235714e-06,
      "loss": 0.3627,
      "step": 39436
    },
    {
      "epoch": 1.8097838557202515,
      "grad_norm": 0.44394734501838684,
      "learning_rate": 3.5273789101090426e-06,
      "loss": 0.3329,
      "step": 39437
    },
    {
      "epoch": 1.809829746225506,
      "grad_norm": 0.479248970746994,
      "learning_rate": 3.5271445995235593e-06,
      "loss": 0.3426,
      "step": 39438
    },
    {
      "epoch": 1.8098756367307605,
      "grad_norm": 0.4800010919570923,
      "learning_rate": 3.526910292479827e-06,
      "loss": 0.3421,
      "step": 39439
    },
    {
      "epoch": 1.809921527236015,
      "grad_norm": 0.4706912338733673,
      "learning_rate": 3.5266759889784126e-06,
      "loss": 0.383,
      "step": 39440
    },
    {
      "epoch": 1.8099674177412695,
      "grad_norm": 0.46369218826293945,
      "learning_rate": 3.526441689019877e-06,
      "loss": 0.3164,
      "step": 39441
    },
    {
      "epoch": 1.810013308246524,
      "grad_norm": 0.44786131381988525,
      "learning_rate": 3.5262073926047815e-06,
      "loss": 0.309,
      "step": 39442
    },
    {
      "epoch": 1.8100591987517782,
      "grad_norm": 0.47272443771362305,
      "learning_rate": 3.525973099733695e-06,
      "loss": 0.3603,
      "step": 39443
    },
    {
      "epoch": 1.8101050892570327,
      "grad_norm": 0.465867280960083,
      "learning_rate": 3.5257388104071784e-06,
      "loss": 0.3342,
      "step": 39444
    },
    {
      "epoch": 1.8101509797622872,
      "grad_norm": 0.4450068473815918,
      "learning_rate": 3.525504524625791e-06,
      "loss": 0.3257,
      "step": 39445
    },
    {
      "epoch": 1.8101968702675415,
      "grad_norm": 0.4671347737312317,
      "learning_rate": 3.525270242390103e-06,
      "loss": 0.347,
      "step": 39446
    },
    {
      "epoch": 1.810242760772796,
      "grad_norm": 0.46415403485298157,
      "learning_rate": 3.525035963700675e-06,
      "loss": 0.3755,
      "step": 39447
    },
    {
      "epoch": 1.8102886512780505,
      "grad_norm": 0.45591476559638977,
      "learning_rate": 3.5248016885580675e-06,
      "loss": 0.3422,
      "step": 39448
    },
    {
      "epoch": 1.810334541783305,
      "grad_norm": 0.47728389501571655,
      "learning_rate": 3.5245674169628487e-06,
      "loss": 0.3878,
      "step": 39449
    },
    {
      "epoch": 1.8103804322885595,
      "grad_norm": 0.48697972297668457,
      "learning_rate": 3.5243331489155785e-06,
      "loss": 0.377,
      "step": 39450
    },
    {
      "epoch": 1.810426322793814,
      "grad_norm": 0.45997846126556396,
      "learning_rate": 3.52409888441682e-06,
      "loss": 0.3149,
      "step": 39451
    },
    {
      "epoch": 1.8104722132990685,
      "grad_norm": 0.5034828186035156,
      "learning_rate": 3.52386462346714e-06,
      "loss": 0.3945,
      "step": 39452
    },
    {
      "epoch": 1.810518103804323,
      "grad_norm": 0.4767049252986908,
      "learning_rate": 3.523630366067099e-06,
      "loss": 0.354,
      "step": 39453
    },
    {
      "epoch": 1.8105639943095775,
      "grad_norm": 0.4790947735309601,
      "learning_rate": 3.5233961122172604e-06,
      "loss": 0.389,
      "step": 39454
    },
    {
      "epoch": 1.8106098848148318,
      "grad_norm": 0.4993637800216675,
      "learning_rate": 3.5231618619181897e-06,
      "loss": 0.4014,
      "step": 39455
    },
    {
      "epoch": 1.8106557753200863,
      "grad_norm": 0.48440149426460266,
      "learning_rate": 3.5229276151704477e-06,
      "loss": 0.3964,
      "step": 39456
    },
    {
      "epoch": 1.8107016658253408,
      "grad_norm": 0.456769198179245,
      "learning_rate": 3.5226933719745994e-06,
      "loss": 0.3339,
      "step": 39457
    },
    {
      "epoch": 1.8107475563305953,
      "grad_norm": 0.4684979021549225,
      "learning_rate": 3.5224591323312044e-06,
      "loss": 0.4113,
      "step": 39458
    },
    {
      "epoch": 1.8107934468358495,
      "grad_norm": 0.4555327594280243,
      "learning_rate": 3.5222248962408314e-06,
      "loss": 0.3255,
      "step": 39459
    },
    {
      "epoch": 1.810839337341104,
      "grad_norm": 0.45053476095199585,
      "learning_rate": 3.521990663704041e-06,
      "loss": 0.3169,
      "step": 39460
    },
    {
      "epoch": 1.8108852278463585,
      "grad_norm": 0.4760741889476776,
      "learning_rate": 3.521756434721396e-06,
      "loss": 0.3577,
      "step": 39461
    },
    {
      "epoch": 1.810931118351613,
      "grad_norm": 0.872757077217102,
      "learning_rate": 3.521522209293461e-06,
      "loss": 0.3982,
      "step": 39462
    },
    {
      "epoch": 1.8109770088568675,
      "grad_norm": 0.45069801807403564,
      "learning_rate": 3.5212879874207986e-06,
      "loss": 0.3297,
      "step": 39463
    },
    {
      "epoch": 1.811022899362122,
      "grad_norm": 0.48185452818870544,
      "learning_rate": 3.5210537691039694e-06,
      "loss": 0.408,
      "step": 39464
    },
    {
      "epoch": 1.8110687898673765,
      "grad_norm": 0.4508371353149414,
      "learning_rate": 3.5208195543435407e-06,
      "loss": 0.3422,
      "step": 39465
    },
    {
      "epoch": 1.811114680372631,
      "grad_norm": 0.4416200518608093,
      "learning_rate": 3.520585343140076e-06,
      "loss": 0.2842,
      "step": 39466
    },
    {
      "epoch": 1.8111605708778855,
      "grad_norm": 0.47118598222732544,
      "learning_rate": 3.5203511354941333e-06,
      "loss": 0.3642,
      "step": 39467
    },
    {
      "epoch": 1.8112064613831398,
      "grad_norm": 0.4564858078956604,
      "learning_rate": 3.5201169314062816e-06,
      "loss": 0.3088,
      "step": 39468
    },
    {
      "epoch": 1.8112523518883943,
      "grad_norm": 0.4718751609325409,
      "learning_rate": 3.5198827308770812e-06,
      "loss": 0.3295,
      "step": 39469
    },
    {
      "epoch": 1.8112982423936488,
      "grad_norm": 0.4917490482330322,
      "learning_rate": 3.5196485339070947e-06,
      "loss": 0.347,
      "step": 39470
    },
    {
      "epoch": 1.811344132898903,
      "grad_norm": 0.4902651309967041,
      "learning_rate": 3.5194143404968874e-06,
      "loss": 0.3817,
      "step": 39471
    },
    {
      "epoch": 1.8113900234041576,
      "grad_norm": 0.47088849544525146,
      "learning_rate": 3.5191801506470213e-06,
      "loss": 0.3771,
      "step": 39472
    },
    {
      "epoch": 1.811435913909412,
      "grad_norm": 0.4848845601081848,
      "learning_rate": 3.518945964358057e-06,
      "loss": 0.4023,
      "step": 39473
    },
    {
      "epoch": 1.8114818044146666,
      "grad_norm": 0.506091296672821,
      "learning_rate": 3.5187117816305632e-06,
      "loss": 0.4172,
      "step": 39474
    },
    {
      "epoch": 1.811527694919921,
      "grad_norm": 0.4335012137889862,
      "learning_rate": 3.5184776024650996e-06,
      "loss": 0.2788,
      "step": 39475
    },
    {
      "epoch": 1.8115735854251755,
      "grad_norm": 0.4631637930870056,
      "learning_rate": 3.518243426862229e-06,
      "loss": 0.3762,
      "step": 39476
    },
    {
      "epoch": 1.81161947593043,
      "grad_norm": 0.4419138431549072,
      "learning_rate": 3.518009254822516e-06,
      "loss": 0.2997,
      "step": 39477
    },
    {
      "epoch": 1.8116653664356845,
      "grad_norm": 0.45471277832984924,
      "learning_rate": 3.5177750863465234e-06,
      "loss": 0.3216,
      "step": 39478
    },
    {
      "epoch": 1.811711256940939,
      "grad_norm": 0.47711506485939026,
      "learning_rate": 3.517540921434811e-06,
      "loss": 0.3999,
      "step": 39479
    },
    {
      "epoch": 1.8117571474461935,
      "grad_norm": 0.4533890187740326,
      "learning_rate": 3.517306760087947e-06,
      "loss": 0.2658,
      "step": 39480
    },
    {
      "epoch": 1.8118030379514478,
      "grad_norm": 0.4536905288696289,
      "learning_rate": 3.5170726023064935e-06,
      "loss": 0.3946,
      "step": 39481
    },
    {
      "epoch": 1.8118489284567023,
      "grad_norm": 0.4744662344455719,
      "learning_rate": 3.51683844809101e-06,
      "loss": 0.4002,
      "step": 39482
    },
    {
      "epoch": 1.8118948189619568,
      "grad_norm": 0.45185065269470215,
      "learning_rate": 3.5166042974420637e-06,
      "loss": 0.3079,
      "step": 39483
    },
    {
      "epoch": 1.811940709467211,
      "grad_norm": 0.46054282784461975,
      "learning_rate": 3.5163701503602154e-06,
      "loss": 0.321,
      "step": 39484
    },
    {
      "epoch": 1.8119865999724656,
      "grad_norm": 0.4445771872997284,
      "learning_rate": 3.516136006846027e-06,
      "loss": 0.3468,
      "step": 39485
    },
    {
      "epoch": 1.81203249047772,
      "grad_norm": 0.4504624307155609,
      "learning_rate": 3.5159018669000655e-06,
      "loss": 0.2975,
      "step": 39486
    },
    {
      "epoch": 1.8120783809829746,
      "grad_norm": 0.48114365339279175,
      "learning_rate": 3.5156677305228913e-06,
      "loss": 0.394,
      "step": 39487
    },
    {
      "epoch": 1.812124271488229,
      "grad_norm": 0.45713376998901367,
      "learning_rate": 3.5154335977150665e-06,
      "loss": 0.3418,
      "step": 39488
    },
    {
      "epoch": 1.8121701619934836,
      "grad_norm": 0.4792759120464325,
      "learning_rate": 3.515199468477156e-06,
      "loss": 0.3454,
      "step": 39489
    },
    {
      "epoch": 1.812216052498738,
      "grad_norm": 0.480305552482605,
      "learning_rate": 3.514965342809723e-06,
      "loss": 0.3518,
      "step": 39490
    },
    {
      "epoch": 1.8122619430039926,
      "grad_norm": 0.4601585566997528,
      "learning_rate": 3.51473122071333e-06,
      "loss": 0.3818,
      "step": 39491
    },
    {
      "epoch": 1.812307833509247,
      "grad_norm": 0.496074378490448,
      "learning_rate": 3.5144971021885375e-06,
      "loss": 0.3787,
      "step": 39492
    },
    {
      "epoch": 1.8123537240145016,
      "grad_norm": 0.45184215903282166,
      "learning_rate": 3.5142629872359123e-06,
      "loss": 0.2937,
      "step": 39493
    },
    {
      "epoch": 1.8123996145197558,
      "grad_norm": 0.4206579923629761,
      "learning_rate": 3.514028875856016e-06,
      "loss": 0.273,
      "step": 39494
    },
    {
      "epoch": 1.8124455050250103,
      "grad_norm": 0.4677528440952301,
      "learning_rate": 3.5137947680494085e-06,
      "loss": 0.2968,
      "step": 39495
    },
    {
      "epoch": 1.8124913955302648,
      "grad_norm": 0.46379512548446655,
      "learning_rate": 3.5135606638166573e-06,
      "loss": 0.348,
      "step": 39496
    },
    {
      "epoch": 1.8125372860355191,
      "grad_norm": 0.44144296646118164,
      "learning_rate": 3.5133265631583246e-06,
      "loss": 0.3031,
      "step": 39497
    },
    {
      "epoch": 1.8125831765407736,
      "grad_norm": 0.496247798204422,
      "learning_rate": 3.513092466074971e-06,
      "loss": 0.4057,
      "step": 39498
    },
    {
      "epoch": 1.812629067046028,
      "grad_norm": 0.4584907591342926,
      "learning_rate": 3.512858372567162e-06,
      "loss": 0.3355,
      "step": 39499
    },
    {
      "epoch": 1.8126749575512826,
      "grad_norm": 0.4650803506374359,
      "learning_rate": 3.5126242826354583e-06,
      "loss": 0.3202,
      "step": 39500
    },
    {
      "epoch": 1.812720848056537,
      "grad_norm": 0.4555077850818634,
      "learning_rate": 3.5123901962804217e-06,
      "loss": 0.3461,
      "step": 39501
    },
    {
      "epoch": 1.8127667385617916,
      "grad_norm": 0.4598619341850281,
      "learning_rate": 3.51215611350262e-06,
      "loss": 0.3265,
      "step": 39502
    },
    {
      "epoch": 1.812812629067046,
      "grad_norm": 0.557809591293335,
      "learning_rate": 3.511922034302613e-06,
      "loss": 0.4538,
      "step": 39503
    },
    {
      "epoch": 1.8128585195723006,
      "grad_norm": 0.4756147861480713,
      "learning_rate": 3.5116879586809627e-06,
      "loss": 0.3297,
      "step": 39504
    },
    {
      "epoch": 1.812904410077555,
      "grad_norm": 0.4510299861431122,
      "learning_rate": 3.5114538866382342e-06,
      "loss": 0.3422,
      "step": 39505
    },
    {
      "epoch": 1.8129503005828094,
      "grad_norm": 0.45085158944129944,
      "learning_rate": 3.5112198181749903e-06,
      "loss": 0.316,
      "step": 39506
    },
    {
      "epoch": 1.8129961910880639,
      "grad_norm": 0.44742006063461304,
      "learning_rate": 3.510985753291789e-06,
      "loss": 0.305,
      "step": 39507
    },
    {
      "epoch": 1.8130420815933184,
      "grad_norm": 0.44663214683532715,
      "learning_rate": 3.5107516919892003e-06,
      "loss": 0.2969,
      "step": 39508
    },
    {
      "epoch": 1.8130879720985726,
      "grad_norm": 0.5223094820976257,
      "learning_rate": 3.510517634267784e-06,
      "loss": 0.3692,
      "step": 39509
    },
    {
      "epoch": 1.8131338626038271,
      "grad_norm": 0.48271459341049194,
      "learning_rate": 3.510283580128101e-06,
      "loss": 0.4002,
      "step": 39510
    },
    {
      "epoch": 1.8131797531090816,
      "grad_norm": 0.454841673374176,
      "learning_rate": 3.510049529570717e-06,
      "loss": 0.3208,
      "step": 39511
    },
    {
      "epoch": 1.8132256436143361,
      "grad_norm": 0.5297150611877441,
      "learning_rate": 3.5098154825961933e-06,
      "loss": 0.3868,
      "step": 39512
    },
    {
      "epoch": 1.8132715341195906,
      "grad_norm": 0.4768126308917999,
      "learning_rate": 3.509581439205092e-06,
      "loss": 0.3791,
      "step": 39513
    },
    {
      "epoch": 1.8133174246248451,
      "grad_norm": 0.47430965304374695,
      "learning_rate": 3.5093473993979785e-06,
      "loss": 0.3676,
      "step": 39514
    },
    {
      "epoch": 1.8133633151300996,
      "grad_norm": 0.44978076219558716,
      "learning_rate": 3.5091133631754136e-06,
      "loss": 0.3432,
      "step": 39515
    },
    {
      "epoch": 1.8134092056353541,
      "grad_norm": 0.44673579931259155,
      "learning_rate": 3.5088793305379593e-06,
      "loss": 0.3359,
      "step": 39516
    },
    {
      "epoch": 1.8134550961406086,
      "grad_norm": 0.484850138425827,
      "learning_rate": 3.508645301486181e-06,
      "loss": 0.3923,
      "step": 39517
    },
    {
      "epoch": 1.8135009866458631,
      "grad_norm": 0.4703046977519989,
      "learning_rate": 3.50841127602064e-06,
      "loss": 0.3233,
      "step": 39518
    },
    {
      "epoch": 1.8135468771511174,
      "grad_norm": 0.6792561411857605,
      "learning_rate": 3.5081772541418984e-06,
      "loss": 0.2875,
      "step": 39519
    },
    {
      "epoch": 1.813592767656372,
      "grad_norm": 0.4948335289955139,
      "learning_rate": 3.5079432358505208e-06,
      "loss": 0.4213,
      "step": 39520
    },
    {
      "epoch": 1.8136386581616264,
      "grad_norm": 0.44160768389701843,
      "learning_rate": 3.507709221147069e-06,
      "loss": 0.3116,
      "step": 39521
    },
    {
      "epoch": 1.8136845486668807,
      "grad_norm": 0.5229886174201965,
      "learning_rate": 3.507475210032104e-06,
      "loss": 0.4239,
      "step": 39522
    },
    {
      "epoch": 1.8137304391721352,
      "grad_norm": 0.4474203884601593,
      "learning_rate": 3.5072412025061897e-06,
      "loss": 0.3137,
      "step": 39523
    },
    {
      "epoch": 1.8137763296773897,
      "grad_norm": 0.5003618001937866,
      "learning_rate": 3.5070071985698907e-06,
      "loss": 0.3556,
      "step": 39524
    },
    {
      "epoch": 1.8138222201826442,
      "grad_norm": 0.49806690216064453,
      "learning_rate": 3.506773198223768e-06,
      "loss": 0.4491,
      "step": 39525
    },
    {
      "epoch": 1.8138681106878987,
      "grad_norm": 0.5106737017631531,
      "learning_rate": 3.5065392014683834e-06,
      "loss": 0.3634,
      "step": 39526
    },
    {
      "epoch": 1.8139140011931532,
      "grad_norm": 0.44777336716651917,
      "learning_rate": 3.506305208304302e-06,
      "loss": 0.2697,
      "step": 39527
    },
    {
      "epoch": 1.8139598916984077,
      "grad_norm": 0.4923311769962311,
      "learning_rate": 3.506071218732085e-06,
      "loss": 0.3538,
      "step": 39528
    },
    {
      "epoch": 1.8140057822036622,
      "grad_norm": 0.47008460760116577,
      "learning_rate": 3.505837232752293e-06,
      "loss": 0.3446,
      "step": 39529
    },
    {
      "epoch": 1.8140516727089167,
      "grad_norm": 0.465802937746048,
      "learning_rate": 3.5056032503654935e-06,
      "loss": 0.3525,
      "step": 39530
    },
    {
      "epoch": 1.8140975632141711,
      "grad_norm": 0.44240227341651917,
      "learning_rate": 3.505369271572246e-06,
      "loss": 0.292,
      "step": 39531
    },
    {
      "epoch": 1.8141434537194254,
      "grad_norm": 0.49259674549102783,
      "learning_rate": 3.5051352963731128e-06,
      "loss": 0.3902,
      "step": 39532
    },
    {
      "epoch": 1.81418934422468,
      "grad_norm": 0.4702433943748474,
      "learning_rate": 3.5049013247686585e-06,
      "loss": 0.3459,
      "step": 39533
    },
    {
      "epoch": 1.8142352347299344,
      "grad_norm": 0.4792638123035431,
      "learning_rate": 3.5046673567594445e-06,
      "loss": 0.3794,
      "step": 39534
    },
    {
      "epoch": 1.8142811252351887,
      "grad_norm": 0.4250314235687256,
      "learning_rate": 3.5044333923460327e-06,
      "loss": 0.2885,
      "step": 39535
    },
    {
      "epoch": 1.8143270157404432,
      "grad_norm": 0.5330049395561218,
      "learning_rate": 3.5041994315289873e-06,
      "loss": 0.487,
      "step": 39536
    },
    {
      "epoch": 1.8143729062456977,
      "grad_norm": 0.5002133250236511,
      "learning_rate": 3.5039654743088707e-06,
      "loss": 0.3617,
      "step": 39537
    },
    {
      "epoch": 1.8144187967509522,
      "grad_norm": 0.4636618494987488,
      "learning_rate": 3.503731520686243e-06,
      "loss": 0.3323,
      "step": 39538
    },
    {
      "epoch": 1.8144646872562067,
      "grad_norm": 0.43951013684272766,
      "learning_rate": 3.5034975706616702e-06,
      "loss": 0.2902,
      "step": 39539
    },
    {
      "epoch": 1.8145105777614612,
      "grad_norm": 0.4752309024333954,
      "learning_rate": 3.5032636242357138e-06,
      "loss": 0.3893,
      "step": 39540
    },
    {
      "epoch": 1.8145564682667157,
      "grad_norm": 0.46331796050071716,
      "learning_rate": 3.503029681408935e-06,
      "loss": 0.3157,
      "step": 39541
    },
    {
      "epoch": 1.8146023587719702,
      "grad_norm": 0.45354852080345154,
      "learning_rate": 3.502795742181898e-06,
      "loss": 0.3448,
      "step": 39542
    },
    {
      "epoch": 1.8146482492772247,
      "grad_norm": 0.40669703483581543,
      "learning_rate": 3.5025618065551647e-06,
      "loss": 0.2301,
      "step": 39543
    },
    {
      "epoch": 1.814694139782479,
      "grad_norm": 0.4454670250415802,
      "learning_rate": 3.5023278745292965e-06,
      "loss": 0.3272,
      "step": 39544
    },
    {
      "epoch": 1.8147400302877335,
      "grad_norm": 0.49089691042900085,
      "learning_rate": 3.5020939461048586e-06,
      "loss": 0.4403,
      "step": 39545
    },
    {
      "epoch": 1.814785920792988,
      "grad_norm": 0.48121514916419983,
      "learning_rate": 3.501860021282412e-06,
      "loss": 0.4,
      "step": 39546
    },
    {
      "epoch": 1.8148318112982424,
      "grad_norm": 0.5374740958213806,
      "learning_rate": 3.5016261000625173e-06,
      "loss": 0.4917,
      "step": 39547
    },
    {
      "epoch": 1.8148777018034967,
      "grad_norm": 0.43312427401542664,
      "learning_rate": 3.5013921824457408e-06,
      "loss": 0.2955,
      "step": 39548
    },
    {
      "epoch": 1.8149235923087512,
      "grad_norm": 0.41755566000938416,
      "learning_rate": 3.5011582684326435e-06,
      "loss": 0.2531,
      "step": 39549
    },
    {
      "epoch": 1.8149694828140057,
      "grad_norm": 0.4619131088256836,
      "learning_rate": 3.500924358023785e-06,
      "loss": 0.3508,
      "step": 39550
    },
    {
      "epoch": 1.8150153733192602,
      "grad_norm": 0.4567030966281891,
      "learning_rate": 3.5006904512197326e-06,
      "loss": 0.3725,
      "step": 39551
    },
    {
      "epoch": 1.8150612638245147,
      "grad_norm": 0.46277812123298645,
      "learning_rate": 3.5004565480210464e-06,
      "loss": 0.352,
      "step": 39552
    },
    {
      "epoch": 1.8151071543297692,
      "grad_norm": 0.46202561259269714,
      "learning_rate": 3.5002226484282873e-06,
      "loss": 0.3186,
      "step": 39553
    },
    {
      "epoch": 1.8151530448350237,
      "grad_norm": 0.45498454570770264,
      "learning_rate": 3.4999887524420206e-06,
      "loss": 0.3371,
      "step": 39554
    },
    {
      "epoch": 1.8151989353402782,
      "grad_norm": 0.44943398237228394,
      "learning_rate": 3.4997548600628084e-06,
      "loss": 0.3429,
      "step": 39555
    },
    {
      "epoch": 1.8152448258455327,
      "grad_norm": 0.4773716926574707,
      "learning_rate": 3.4995209712912117e-06,
      "loss": 0.3838,
      "step": 39556
    },
    {
      "epoch": 1.815290716350787,
      "grad_norm": 0.469123512506485,
      "learning_rate": 3.4992870861277928e-06,
      "loss": 0.3748,
      "step": 39557
    },
    {
      "epoch": 1.8153366068560415,
      "grad_norm": 0.49857503175735474,
      "learning_rate": 3.499053204573115e-06,
      "loss": 0.3868,
      "step": 39558
    },
    {
      "epoch": 1.815382497361296,
      "grad_norm": 0.47888579964637756,
      "learning_rate": 3.4988193266277415e-06,
      "loss": 0.3547,
      "step": 39559
    },
    {
      "epoch": 1.8154283878665503,
      "grad_norm": 0.48772895336151123,
      "learning_rate": 3.4985854522922306e-06,
      "loss": 0.3711,
      "step": 39560
    },
    {
      "epoch": 1.8154742783718048,
      "grad_norm": 0.4486040771007538,
      "learning_rate": 3.498351581567151e-06,
      "loss": 0.3202,
      "step": 39561
    },
    {
      "epoch": 1.8155201688770592,
      "grad_norm": 0.4901120066642761,
      "learning_rate": 3.498117714453061e-06,
      "loss": 0.4046,
      "step": 39562
    },
    {
      "epoch": 1.8155660593823137,
      "grad_norm": 0.4414820969104767,
      "learning_rate": 3.497883850950523e-06,
      "loss": 0.2689,
      "step": 39563
    },
    {
      "epoch": 1.8156119498875682,
      "grad_norm": 0.4740545153617859,
      "learning_rate": 3.497649991060102e-06,
      "loss": 0.3197,
      "step": 39564
    },
    {
      "epoch": 1.8156578403928227,
      "grad_norm": 0.44108715653419495,
      "learning_rate": 3.4974161347823582e-06,
      "loss": 0.3159,
      "step": 39565
    },
    {
      "epoch": 1.8157037308980772,
      "grad_norm": 0.48185884952545166,
      "learning_rate": 3.4971822821178524e-06,
      "loss": 0.3494,
      "step": 39566
    },
    {
      "epoch": 1.8157496214033317,
      "grad_norm": 0.4806457459926605,
      "learning_rate": 3.496948433067151e-06,
      "loss": 0.38,
      "step": 39567
    },
    {
      "epoch": 1.8157955119085862,
      "grad_norm": 0.4542843699455261,
      "learning_rate": 3.496714587630814e-06,
      "loss": 0.3339,
      "step": 39568
    },
    {
      "epoch": 1.8158414024138407,
      "grad_norm": 0.4481528103351593,
      "learning_rate": 3.496480745809403e-06,
      "loss": 0.3208,
      "step": 39569
    },
    {
      "epoch": 1.815887292919095,
      "grad_norm": 0.462266206741333,
      "learning_rate": 3.4962469076034823e-06,
      "loss": 0.3661,
      "step": 39570
    },
    {
      "epoch": 1.8159331834243495,
      "grad_norm": 0.5014846920967102,
      "learning_rate": 3.4960130730136135e-06,
      "loss": 0.3898,
      "step": 39571
    },
    {
      "epoch": 1.815979073929604,
      "grad_norm": 0.4919555187225342,
      "learning_rate": 3.495779242040356e-06,
      "loss": 0.3668,
      "step": 39572
    },
    {
      "epoch": 1.8160249644348583,
      "grad_norm": 0.4625389277935028,
      "learning_rate": 3.4955454146842773e-06,
      "loss": 0.3419,
      "step": 39573
    },
    {
      "epoch": 1.8160708549401128,
      "grad_norm": 0.5019527673721313,
      "learning_rate": 3.495311590945937e-06,
      "loss": 0.4436,
      "step": 39574
    },
    {
      "epoch": 1.8161167454453673,
      "grad_norm": 0.44494757056236267,
      "learning_rate": 3.4950777708258955e-06,
      "loss": 0.3392,
      "step": 39575
    },
    {
      "epoch": 1.8161626359506218,
      "grad_norm": 0.5316281318664551,
      "learning_rate": 3.4948439543247185e-06,
      "loss": 0.3736,
      "step": 39576
    },
    {
      "epoch": 1.8162085264558763,
      "grad_norm": 0.4386516809463501,
      "learning_rate": 3.494610141442967e-06,
      "loss": 0.3147,
      "step": 39577
    },
    {
      "epoch": 1.8162544169611308,
      "grad_norm": 0.48993417620658875,
      "learning_rate": 3.4943763321812007e-06,
      "loss": 0.3961,
      "step": 39578
    },
    {
      "epoch": 1.8163003074663853,
      "grad_norm": 0.44720199704170227,
      "learning_rate": 3.4941425265399873e-06,
      "loss": 0.2872,
      "step": 39579
    },
    {
      "epoch": 1.8163461979716398,
      "grad_norm": 0.4452161192893982,
      "learning_rate": 3.4939087245198855e-06,
      "loss": 0.3391,
      "step": 39580
    },
    {
      "epoch": 1.8163920884768943,
      "grad_norm": 0.4964442849159241,
      "learning_rate": 3.493674926121455e-06,
      "loss": 0.4122,
      "step": 39581
    },
    {
      "epoch": 1.8164379789821488,
      "grad_norm": 0.48770570755004883,
      "learning_rate": 3.493441131345263e-06,
      "loss": 0.3681,
      "step": 39582
    },
    {
      "epoch": 1.816483869487403,
      "grad_norm": 0.47472161054611206,
      "learning_rate": 3.4932073401918697e-06,
      "loss": 0.3521,
      "step": 39583
    },
    {
      "epoch": 1.8165297599926575,
      "grad_norm": 0.4979342222213745,
      "learning_rate": 3.492973552661837e-06,
      "loss": 0.4258,
      "step": 39584
    },
    {
      "epoch": 1.816575650497912,
      "grad_norm": 0.48419591784477234,
      "learning_rate": 3.4927397687557273e-06,
      "loss": 0.3359,
      "step": 39585
    },
    {
      "epoch": 1.8166215410031663,
      "grad_norm": 0.46609562635421753,
      "learning_rate": 3.492505988474103e-06,
      "loss": 0.3275,
      "step": 39586
    },
    {
      "epoch": 1.8166674315084208,
      "grad_norm": 0.45768651366233826,
      "learning_rate": 3.492272211817526e-06,
      "loss": 0.3579,
      "step": 39587
    },
    {
      "epoch": 1.8167133220136753,
      "grad_norm": 0.44072335958480835,
      "learning_rate": 3.492038438786556e-06,
      "loss": 0.342,
      "step": 39588
    },
    {
      "epoch": 1.8167592125189298,
      "grad_norm": 0.4868362843990326,
      "learning_rate": 3.4918046693817608e-06,
      "loss": 0.3388,
      "step": 39589
    },
    {
      "epoch": 1.8168051030241843,
      "grad_norm": 0.46968239545822144,
      "learning_rate": 3.4915709036036983e-06,
      "loss": 0.3913,
      "step": 39590
    },
    {
      "epoch": 1.8168509935294388,
      "grad_norm": 0.4459097981452942,
      "learning_rate": 3.4913371414529313e-06,
      "loss": 0.3265,
      "step": 39591
    },
    {
      "epoch": 1.8168968840346933,
      "grad_norm": 0.46137523651123047,
      "learning_rate": 3.4911033829300232e-06,
      "loss": 0.3474,
      "step": 39592
    },
    {
      "epoch": 1.8169427745399478,
      "grad_norm": 0.48670485615730286,
      "learning_rate": 3.4908696280355346e-06,
      "loss": 0.3889,
      "step": 39593
    },
    {
      "epoch": 1.8169886650452023,
      "grad_norm": 0.4040960371494293,
      "learning_rate": 3.4906358767700273e-06,
      "loss": 0.2715,
      "step": 39594
    },
    {
      "epoch": 1.8170345555504566,
      "grad_norm": 0.4574456214904785,
      "learning_rate": 3.4904021291340663e-06,
      "loss": 0.344,
      "step": 39595
    },
    {
      "epoch": 1.817080446055711,
      "grad_norm": 0.4248758852481842,
      "learning_rate": 3.4901683851282108e-06,
      "loss": 0.286,
      "step": 39596
    },
    {
      "epoch": 1.8171263365609656,
      "grad_norm": 0.4577995836734772,
      "learning_rate": 3.4899346447530234e-06,
      "loss": 0.3684,
      "step": 39597
    },
    {
      "epoch": 1.8171722270662198,
      "grad_norm": 0.4645821750164032,
      "learning_rate": 3.489700908009067e-06,
      "loss": 0.3707,
      "step": 39598
    },
    {
      "epoch": 1.8172181175714743,
      "grad_norm": 0.4526720941066742,
      "learning_rate": 3.4894671748969045e-06,
      "loss": 0.3757,
      "step": 39599
    },
    {
      "epoch": 1.8172640080767288,
      "grad_norm": 0.4931142032146454,
      "learning_rate": 3.489233445417094e-06,
      "loss": 0.4197,
      "step": 39600
    },
    {
      "epoch": 1.8173098985819833,
      "grad_norm": 0.4752224385738373,
      "learning_rate": 3.488999719570203e-06,
      "loss": 0.3769,
      "step": 39601
    },
    {
      "epoch": 1.8173557890872378,
      "grad_norm": 0.4186309576034546,
      "learning_rate": 3.48876599735679e-06,
      "loss": 0.2721,
      "step": 39602
    },
    {
      "epoch": 1.8174016795924923,
      "grad_norm": 0.5018913745880127,
      "learning_rate": 3.4885322787774154e-06,
      "loss": 0.3865,
      "step": 39603
    },
    {
      "epoch": 1.8174475700977468,
      "grad_norm": 0.4992775321006775,
      "learning_rate": 3.488298563832646e-06,
      "loss": 0.3606,
      "step": 39604
    },
    {
      "epoch": 1.8174934606030013,
      "grad_norm": 0.46387362480163574,
      "learning_rate": 3.488064852523042e-06,
      "loss": 0.3069,
      "step": 39605
    },
    {
      "epoch": 1.8175393511082558,
      "grad_norm": 0.4529910683631897,
      "learning_rate": 3.487831144849163e-06,
      "loss": 0.3333,
      "step": 39606
    },
    {
      "epoch": 1.8175852416135103,
      "grad_norm": 0.43581998348236084,
      "learning_rate": 3.4875974408115743e-06,
      "loss": 0.291,
      "step": 39607
    },
    {
      "epoch": 1.8176311321187646,
      "grad_norm": 0.4874977469444275,
      "learning_rate": 3.4873637404108364e-06,
      "loss": 0.3811,
      "step": 39608
    },
    {
      "epoch": 1.817677022624019,
      "grad_norm": 0.45020362734794617,
      "learning_rate": 3.487130043647509e-06,
      "loss": 0.3546,
      "step": 39609
    },
    {
      "epoch": 1.8177229131292736,
      "grad_norm": 0.4834462106227875,
      "learning_rate": 3.4868963505221587e-06,
      "loss": 0.3446,
      "step": 39610
    },
    {
      "epoch": 1.8177688036345279,
      "grad_norm": 0.48098015785217285,
      "learning_rate": 3.486662661035345e-06,
      "loss": 0.3807,
      "step": 39611
    },
    {
      "epoch": 1.8178146941397824,
      "grad_norm": 0.46889030933380127,
      "learning_rate": 3.486428975187629e-06,
      "loss": 0.3222,
      "step": 39612
    },
    {
      "epoch": 1.8178605846450369,
      "grad_norm": 0.5324971079826355,
      "learning_rate": 3.486195292979575e-06,
      "loss": 0.3927,
      "step": 39613
    },
    {
      "epoch": 1.8179064751502914,
      "grad_norm": 0.4587648808956146,
      "learning_rate": 3.485961614411743e-06,
      "loss": 0.3361,
      "step": 39614
    },
    {
      "epoch": 1.8179523656555459,
      "grad_norm": 0.48495808243751526,
      "learning_rate": 3.485727939484693e-06,
      "loss": 0.3672,
      "step": 39615
    },
    {
      "epoch": 1.8179982561608004,
      "grad_norm": 0.46631932258605957,
      "learning_rate": 3.4854942681989925e-06,
      "loss": 0.3236,
      "step": 39616
    },
    {
      "epoch": 1.8180441466660548,
      "grad_norm": 0.5099856853485107,
      "learning_rate": 3.4852606005552e-06,
      "loss": 0.3846,
      "step": 39617
    },
    {
      "epoch": 1.8180900371713093,
      "grad_norm": 0.46327638626098633,
      "learning_rate": 3.4850269365538765e-06,
      "loss": 0.337,
      "step": 39618
    },
    {
      "epoch": 1.8181359276765638,
      "grad_norm": 0.47863510251045227,
      "learning_rate": 3.484793276195586e-06,
      "loss": 0.3854,
      "step": 39619
    },
    {
      "epoch": 1.8181818181818183,
      "grad_norm": 0.44759026169776917,
      "learning_rate": 3.4845596194808896e-06,
      "loss": 0.2916,
      "step": 39620
    },
    {
      "epoch": 1.8182277086870726,
      "grad_norm": 0.5106325149536133,
      "learning_rate": 3.484325966410349e-06,
      "loss": 0.4466,
      "step": 39621
    },
    {
      "epoch": 1.8182735991923271,
      "grad_norm": 0.496844083070755,
      "learning_rate": 3.484092316984524e-06,
      "loss": 0.3674,
      "step": 39622
    },
    {
      "epoch": 1.8183194896975816,
      "grad_norm": 0.511273205280304,
      "learning_rate": 3.483858671203982e-06,
      "loss": 0.4868,
      "step": 39623
    },
    {
      "epoch": 1.818365380202836,
      "grad_norm": 0.4763088822364807,
      "learning_rate": 3.4836250290692797e-06,
      "loss": 0.3836,
      "step": 39624
    },
    {
      "epoch": 1.8184112707080904,
      "grad_norm": 0.44941946864128113,
      "learning_rate": 3.483391390580979e-06,
      "loss": 0.3035,
      "step": 39625
    },
    {
      "epoch": 1.8184571612133449,
      "grad_norm": 0.5031437873840332,
      "learning_rate": 3.483157755739645e-06,
      "loss": 0.3765,
      "step": 39626
    },
    {
      "epoch": 1.8185030517185994,
      "grad_norm": 0.5027937293052673,
      "learning_rate": 3.482924124545838e-06,
      "loss": 0.4134,
      "step": 39627
    },
    {
      "epoch": 1.8185489422238539,
      "grad_norm": 0.725335955619812,
      "learning_rate": 3.482690497000119e-06,
      "loss": 0.3855,
      "step": 39628
    },
    {
      "epoch": 1.8185948327291084,
      "grad_norm": 0.4535479247570038,
      "learning_rate": 3.4824568731030516e-06,
      "loss": 0.3283,
      "step": 39629
    },
    {
      "epoch": 1.8186407232343629,
      "grad_norm": 0.48404204845428467,
      "learning_rate": 3.4822232528551958e-06,
      "loss": 0.3548,
      "step": 39630
    },
    {
      "epoch": 1.8186866137396174,
      "grad_norm": 0.4953702688217163,
      "learning_rate": 3.4819896362571126e-06,
      "loss": 0.3959,
      "step": 39631
    },
    {
      "epoch": 1.8187325042448719,
      "grad_norm": 0.45875510573387146,
      "learning_rate": 3.4817560233093674e-06,
      "loss": 0.3477,
      "step": 39632
    },
    {
      "epoch": 1.8187783947501261,
      "grad_norm": 0.5192646384239197,
      "learning_rate": 3.4815224140125194e-06,
      "loss": 0.4271,
      "step": 39633
    },
    {
      "epoch": 1.8188242852553806,
      "grad_norm": 0.49065494537353516,
      "learning_rate": 3.481288808367129e-06,
      "loss": 0.3562,
      "step": 39634
    },
    {
      "epoch": 1.8188701757606351,
      "grad_norm": 0.5205522775650024,
      "learning_rate": 3.4810552063737617e-06,
      "loss": 0.4019,
      "step": 39635
    },
    {
      "epoch": 1.8189160662658896,
      "grad_norm": 0.47686171531677246,
      "learning_rate": 3.480821608032977e-06,
      "loss": 0.3451,
      "step": 39636
    },
    {
      "epoch": 1.818961956771144,
      "grad_norm": 0.46159252524375916,
      "learning_rate": 3.480588013345335e-06,
      "loss": 0.3408,
      "step": 39637
    },
    {
      "epoch": 1.8190078472763984,
      "grad_norm": 0.4689064025878906,
      "learning_rate": 3.4803544223114017e-06,
      "loss": 0.361,
      "step": 39638
    },
    {
      "epoch": 1.819053737781653,
      "grad_norm": 0.46284064650535583,
      "learning_rate": 3.4801208349317357e-06,
      "loss": 0.3181,
      "step": 39639
    },
    {
      "epoch": 1.8190996282869074,
      "grad_norm": 0.5090509057044983,
      "learning_rate": 3.4798872512068983e-06,
      "loss": 0.39,
      "step": 39640
    },
    {
      "epoch": 1.819145518792162,
      "grad_norm": 0.5186017155647278,
      "learning_rate": 3.479653671137454e-06,
      "loss": 0.492,
      "step": 39641
    },
    {
      "epoch": 1.8191914092974164,
      "grad_norm": 0.4499276876449585,
      "learning_rate": 3.479420094723962e-06,
      "loss": 0.3489,
      "step": 39642
    },
    {
      "epoch": 1.819237299802671,
      "grad_norm": 0.4829350411891937,
      "learning_rate": 3.4791865219669828e-06,
      "loss": 0.3446,
      "step": 39643
    },
    {
      "epoch": 1.8192831903079254,
      "grad_norm": 0.4944691061973572,
      "learning_rate": 3.4789529528670827e-06,
      "loss": 0.4094,
      "step": 39644
    },
    {
      "epoch": 1.81932908081318,
      "grad_norm": 0.4556099474430084,
      "learning_rate": 3.4787193874248217e-06,
      "loss": 0.3385,
      "step": 39645
    },
    {
      "epoch": 1.8193749713184342,
      "grad_norm": 0.48509979248046875,
      "learning_rate": 3.4784858256407567e-06,
      "loss": 0.3911,
      "step": 39646
    },
    {
      "epoch": 1.8194208618236887,
      "grad_norm": 0.4380609393119812,
      "learning_rate": 3.478252267515455e-06,
      "loss": 0.2747,
      "step": 39647
    },
    {
      "epoch": 1.8194667523289432,
      "grad_norm": 0.43105247616767883,
      "learning_rate": 3.478018713049477e-06,
      "loss": 0.2827,
      "step": 39648
    },
    {
      "epoch": 1.8195126428341974,
      "grad_norm": 0.4686887562274933,
      "learning_rate": 3.4777851622433828e-06,
      "loss": 0.3793,
      "step": 39649
    },
    {
      "epoch": 1.819558533339452,
      "grad_norm": 0.42547518014907837,
      "learning_rate": 3.477551615097735e-06,
      "loss": 0.2996,
      "step": 39650
    },
    {
      "epoch": 1.8196044238447064,
      "grad_norm": 0.49002817273139954,
      "learning_rate": 3.477318071613096e-06,
      "loss": 0.4055,
      "step": 39651
    },
    {
      "epoch": 1.819650314349961,
      "grad_norm": 0.45447611808776855,
      "learning_rate": 3.477084531790024e-06,
      "loss": 0.3375,
      "step": 39652
    },
    {
      "epoch": 1.8196962048552154,
      "grad_norm": 0.47052499651908875,
      "learning_rate": 3.4768509956290854e-06,
      "loss": 0.3714,
      "step": 39653
    },
    {
      "epoch": 1.81974209536047,
      "grad_norm": 0.426779568195343,
      "learning_rate": 3.4766174631308393e-06,
      "loss": 0.3238,
      "step": 39654
    },
    {
      "epoch": 1.8197879858657244,
      "grad_norm": 0.5042814016342163,
      "learning_rate": 3.4763839342958473e-06,
      "loss": 0.4499,
      "step": 39655
    },
    {
      "epoch": 1.819833876370979,
      "grad_norm": 0.4874879717826843,
      "learning_rate": 3.4761504091246695e-06,
      "loss": 0.3832,
      "step": 39656
    },
    {
      "epoch": 1.8198797668762334,
      "grad_norm": 0.4270375370979309,
      "learning_rate": 3.4759168876178707e-06,
      "loss": 0.3033,
      "step": 39657
    },
    {
      "epoch": 1.819925657381488,
      "grad_norm": 0.4860326051712036,
      "learning_rate": 3.4756833697760107e-06,
      "loss": 0.3467,
      "step": 39658
    },
    {
      "epoch": 1.8199715478867422,
      "grad_norm": 0.464802086353302,
      "learning_rate": 3.4754498555996484e-06,
      "loss": 0.3607,
      "step": 39659
    },
    {
      "epoch": 1.8200174383919967,
      "grad_norm": 0.5333609580993652,
      "learning_rate": 3.4752163450893506e-06,
      "loss": 0.4076,
      "step": 39660
    },
    {
      "epoch": 1.8200633288972512,
      "grad_norm": 0.4679904282093048,
      "learning_rate": 3.4749828382456764e-06,
      "loss": 0.3054,
      "step": 39661
    },
    {
      "epoch": 1.8201092194025055,
      "grad_norm": 0.4843520522117615,
      "learning_rate": 3.4747493350691864e-06,
      "loss": 0.3493,
      "step": 39662
    },
    {
      "epoch": 1.82015510990776,
      "grad_norm": 0.4795750677585602,
      "learning_rate": 3.4745158355604435e-06,
      "loss": 0.3948,
      "step": 39663
    },
    {
      "epoch": 1.8202010004130145,
      "grad_norm": 0.4739072620868683,
      "learning_rate": 3.474282339720009e-06,
      "loss": 0.3475,
      "step": 39664
    },
    {
      "epoch": 1.820246890918269,
      "grad_norm": 0.4935581386089325,
      "learning_rate": 3.474048847548441e-06,
      "loss": 0.3389,
      "step": 39665
    },
    {
      "epoch": 1.8202927814235235,
      "grad_norm": 0.45583951473236084,
      "learning_rate": 3.4738153590463065e-06,
      "loss": 0.338,
      "step": 39666
    },
    {
      "epoch": 1.820338671928778,
      "grad_norm": 0.4920426905155182,
      "learning_rate": 3.4735818742141653e-06,
      "loss": 0.4099,
      "step": 39667
    },
    {
      "epoch": 1.8203845624340325,
      "grad_norm": 0.4615534543991089,
      "learning_rate": 3.473348393052575e-06,
      "loss": 0.3699,
      "step": 39668
    },
    {
      "epoch": 1.820430452939287,
      "grad_norm": 0.4846288859844208,
      "learning_rate": 3.473114915562102e-06,
      "loss": 0.3577,
      "step": 39669
    },
    {
      "epoch": 1.8204763434445415,
      "grad_norm": 0.4416280686855316,
      "learning_rate": 3.4728814417433054e-06,
      "loss": 0.2895,
      "step": 39670
    },
    {
      "epoch": 1.8205222339497957,
      "grad_norm": 0.47974348068237305,
      "learning_rate": 3.4726479715967466e-06,
      "loss": 0.3443,
      "step": 39671
    },
    {
      "epoch": 1.8205681244550502,
      "grad_norm": 0.4343111217021942,
      "learning_rate": 3.472414505122988e-06,
      "loss": 0.2719,
      "step": 39672
    },
    {
      "epoch": 1.8206140149603047,
      "grad_norm": 0.551262617111206,
      "learning_rate": 3.4721810423225904e-06,
      "loss": 0.4476,
      "step": 39673
    },
    {
      "epoch": 1.8206599054655592,
      "grad_norm": 0.46013858914375305,
      "learning_rate": 3.471947583196113e-06,
      "loss": 0.327,
      "step": 39674
    },
    {
      "epoch": 1.8207057959708135,
      "grad_norm": 0.44072896242141724,
      "learning_rate": 3.4717141277441217e-06,
      "loss": 0.2941,
      "step": 39675
    },
    {
      "epoch": 1.820751686476068,
      "grad_norm": 0.4894869029521942,
      "learning_rate": 3.471480675967176e-06,
      "loss": 0.3947,
      "step": 39676
    },
    {
      "epoch": 1.8207975769813225,
      "grad_norm": 0.4676510989665985,
      "learning_rate": 3.4712472278658348e-06,
      "loss": 0.357,
      "step": 39677
    },
    {
      "epoch": 1.820843467486577,
      "grad_norm": 0.46327951550483704,
      "learning_rate": 3.4710137834406634e-06,
      "loss": 0.3319,
      "step": 39678
    },
    {
      "epoch": 1.8208893579918315,
      "grad_norm": 0.5456066131591797,
      "learning_rate": 3.470780342692221e-06,
      "loss": 0.4153,
      "step": 39679
    },
    {
      "epoch": 1.820935248497086,
      "grad_norm": 0.4520506262779236,
      "learning_rate": 3.4705469056210673e-06,
      "loss": 0.3432,
      "step": 39680
    },
    {
      "epoch": 1.8209811390023405,
      "grad_norm": 0.4608404040336609,
      "learning_rate": 3.470313472227768e-06,
      "loss": 0.341,
      "step": 39681
    },
    {
      "epoch": 1.821027029507595,
      "grad_norm": 0.5059654712677002,
      "learning_rate": 3.4700800425128813e-06,
      "loss": 0.4258,
      "step": 39682
    },
    {
      "epoch": 1.8210729200128495,
      "grad_norm": 0.44469496607780457,
      "learning_rate": 3.469846616476968e-06,
      "loss": 0.2854,
      "step": 39683
    },
    {
      "epoch": 1.8211188105181038,
      "grad_norm": 0.4834306836128235,
      "learning_rate": 3.4696131941205925e-06,
      "loss": 0.3455,
      "step": 39684
    },
    {
      "epoch": 1.8211647010233583,
      "grad_norm": 0.48792532086372375,
      "learning_rate": 3.469379775444314e-06,
      "loss": 0.3789,
      "step": 39685
    },
    {
      "epoch": 1.8212105915286128,
      "grad_norm": 0.5039854645729065,
      "learning_rate": 3.4691463604486937e-06,
      "loss": 0.3293,
      "step": 39686
    },
    {
      "epoch": 1.821256482033867,
      "grad_norm": 0.4919971525669098,
      "learning_rate": 3.468912949134291e-06,
      "loss": 0.3886,
      "step": 39687
    },
    {
      "epoch": 1.8213023725391215,
      "grad_norm": 0.47926145792007446,
      "learning_rate": 3.4686795415016723e-06,
      "loss": 0.3757,
      "step": 39688
    },
    {
      "epoch": 1.821348263044376,
      "grad_norm": 0.433773934841156,
      "learning_rate": 3.4684461375513965e-06,
      "loss": 0.2907,
      "step": 39689
    },
    {
      "epoch": 1.8213941535496305,
      "grad_norm": 0.44634684920310974,
      "learning_rate": 3.468212737284021e-06,
      "loss": 0.3116,
      "step": 39690
    },
    {
      "epoch": 1.821440044054885,
      "grad_norm": 0.47449058294296265,
      "learning_rate": 3.4679793407001135e-06,
      "loss": 0.3359,
      "step": 39691
    },
    {
      "epoch": 1.8214859345601395,
      "grad_norm": 0.48565399646759033,
      "learning_rate": 3.4677459478002316e-06,
      "loss": 0.3563,
      "step": 39692
    },
    {
      "epoch": 1.821531825065394,
      "grad_norm": 0.48586633801460266,
      "learning_rate": 3.467512558584936e-06,
      "loss": 0.347,
      "step": 39693
    },
    {
      "epoch": 1.8215777155706485,
      "grad_norm": 0.426209956407547,
      "learning_rate": 3.4672791730547907e-06,
      "loss": 0.3191,
      "step": 39694
    },
    {
      "epoch": 1.821623606075903,
      "grad_norm": 0.50107741355896,
      "learning_rate": 3.467045791210354e-06,
      "loss": 0.4237,
      "step": 39695
    },
    {
      "epoch": 1.8216694965811575,
      "grad_norm": 0.5386788249015808,
      "learning_rate": 3.4668124130521874e-06,
      "loss": 0.2958,
      "step": 39696
    },
    {
      "epoch": 1.8217153870864118,
      "grad_norm": 0.453836053609848,
      "learning_rate": 3.4665790385808545e-06,
      "loss": 0.3383,
      "step": 39697
    },
    {
      "epoch": 1.8217612775916663,
      "grad_norm": 0.43805840611457825,
      "learning_rate": 3.466345667796915e-06,
      "loss": 0.2965,
      "step": 39698
    },
    {
      "epoch": 1.8218071680969208,
      "grad_norm": 0.48212528228759766,
      "learning_rate": 3.4661123007009294e-06,
      "loss": 0.3358,
      "step": 39699
    },
    {
      "epoch": 1.821853058602175,
      "grad_norm": 0.43631258606910706,
      "learning_rate": 3.4658789372934608e-06,
      "loss": 0.2946,
      "step": 39700
    },
    {
      "epoch": 1.8218989491074296,
      "grad_norm": 0.4848869740962982,
      "learning_rate": 3.4656455775750684e-06,
      "loss": 0.3886,
      "step": 39701
    },
    {
      "epoch": 1.821944839612684,
      "grad_norm": 0.47595199942588806,
      "learning_rate": 3.465412221546313e-06,
      "loss": 0.3449,
      "step": 39702
    },
    {
      "epoch": 1.8219907301179386,
      "grad_norm": 0.4694417417049408,
      "learning_rate": 3.4651788692077587e-06,
      "loss": 0.3416,
      "step": 39703
    },
    {
      "epoch": 1.822036620623193,
      "grad_norm": 0.4631176292896271,
      "learning_rate": 3.4649455205599646e-06,
      "loss": 0.3132,
      "step": 39704
    },
    {
      "epoch": 1.8220825111284475,
      "grad_norm": 0.4707501530647278,
      "learning_rate": 3.4647121756034913e-06,
      "loss": 0.3813,
      "step": 39705
    },
    {
      "epoch": 1.822128401633702,
      "grad_norm": 0.532524824142456,
      "learning_rate": 3.4644788343389014e-06,
      "loss": 0.4556,
      "step": 39706
    },
    {
      "epoch": 1.8221742921389565,
      "grad_norm": 0.44256794452667236,
      "learning_rate": 3.4642454967667555e-06,
      "loss": 0.3314,
      "step": 39707
    },
    {
      "epoch": 1.822220182644211,
      "grad_norm": 0.5229761600494385,
      "learning_rate": 3.4640121628876122e-06,
      "loss": 0.4271,
      "step": 39708
    },
    {
      "epoch": 1.8222660731494655,
      "grad_norm": 0.47077512741088867,
      "learning_rate": 3.463778832702037e-06,
      "loss": 0.3966,
      "step": 39709
    },
    {
      "epoch": 1.8223119636547198,
      "grad_norm": 0.4409750699996948,
      "learning_rate": 3.463545506210589e-06,
      "loss": 0.292,
      "step": 39710
    },
    {
      "epoch": 1.8223578541599743,
      "grad_norm": 0.48615244030952454,
      "learning_rate": 3.463312183413828e-06,
      "loss": 0.4027,
      "step": 39711
    },
    {
      "epoch": 1.8224037446652288,
      "grad_norm": 0.44817638397216797,
      "learning_rate": 3.463078864312317e-06,
      "loss": 0.333,
      "step": 39712
    },
    {
      "epoch": 1.822449635170483,
      "grad_norm": 0.43933331966400146,
      "learning_rate": 3.4628455489066172e-06,
      "loss": 0.3531,
      "step": 39713
    },
    {
      "epoch": 1.8224955256757376,
      "grad_norm": 0.4364699721336365,
      "learning_rate": 3.4626122371972866e-06,
      "loss": 0.3056,
      "step": 39714
    },
    {
      "epoch": 1.822541416180992,
      "grad_norm": 0.4388059973716736,
      "learning_rate": 3.4623789291848897e-06,
      "loss": 0.3113,
      "step": 39715
    },
    {
      "epoch": 1.8225873066862466,
      "grad_norm": 0.47817790508270264,
      "learning_rate": 3.462145624869986e-06,
      "loss": 0.3662,
      "step": 39716
    },
    {
      "epoch": 1.822633197191501,
      "grad_norm": 0.4888109266757965,
      "learning_rate": 3.461912324253135e-06,
      "loss": 0.3798,
      "step": 39717
    },
    {
      "epoch": 1.8226790876967556,
      "grad_norm": 0.47051936388015747,
      "learning_rate": 3.4616790273349017e-06,
      "loss": 0.3636,
      "step": 39718
    },
    {
      "epoch": 1.82272497820201,
      "grad_norm": 0.42735761404037476,
      "learning_rate": 3.4614457341158446e-06,
      "loss": 0.3046,
      "step": 39719
    },
    {
      "epoch": 1.8227708687072646,
      "grad_norm": 0.4589006006717682,
      "learning_rate": 3.4612124445965245e-06,
      "loss": 0.3529,
      "step": 39720
    },
    {
      "epoch": 1.822816759212519,
      "grad_norm": 0.4616396725177765,
      "learning_rate": 3.460979158777502e-06,
      "loss": 0.3573,
      "step": 39721
    },
    {
      "epoch": 1.8228626497177733,
      "grad_norm": 0.44229286909103394,
      "learning_rate": 3.4607458766593403e-06,
      "loss": 0.3098,
      "step": 39722
    },
    {
      "epoch": 1.8229085402230278,
      "grad_norm": 0.47998663783073425,
      "learning_rate": 3.460512598242599e-06,
      "loss": 0.3398,
      "step": 39723
    },
    {
      "epoch": 1.8229544307282823,
      "grad_norm": 0.4813268780708313,
      "learning_rate": 3.4602793235278364e-06,
      "loss": 0.3262,
      "step": 39724
    },
    {
      "epoch": 1.8230003212335368,
      "grad_norm": 0.4682084023952484,
      "learning_rate": 3.4600460525156187e-06,
      "loss": 0.3211,
      "step": 39725
    },
    {
      "epoch": 1.8230462117387911,
      "grad_norm": 0.48063281178474426,
      "learning_rate": 3.4598127852065037e-06,
      "loss": 0.3826,
      "step": 39726
    },
    {
      "epoch": 1.8230921022440456,
      "grad_norm": 0.4624738097190857,
      "learning_rate": 3.459579521601052e-06,
      "loss": 0.3566,
      "step": 39727
    },
    {
      "epoch": 1.8231379927493,
      "grad_norm": 0.459591269493103,
      "learning_rate": 3.4593462616998263e-06,
      "loss": 0.3184,
      "step": 39728
    },
    {
      "epoch": 1.8231838832545546,
      "grad_norm": 0.47669076919555664,
      "learning_rate": 3.459113005503387e-06,
      "loss": 0.3811,
      "step": 39729
    },
    {
      "epoch": 1.823229773759809,
      "grad_norm": 0.4679020643234253,
      "learning_rate": 3.4588797530122923e-06,
      "loss": 0.3628,
      "step": 39730
    },
    {
      "epoch": 1.8232756642650636,
      "grad_norm": 0.5673671364784241,
      "learning_rate": 3.4586465042271077e-06,
      "loss": 0.4215,
      "step": 39731
    },
    {
      "epoch": 1.823321554770318,
      "grad_norm": 0.5005654692649841,
      "learning_rate": 3.4584132591483914e-06,
      "loss": 0.4395,
      "step": 39732
    },
    {
      "epoch": 1.8233674452755726,
      "grad_norm": 0.45318859815597534,
      "learning_rate": 3.4581800177767042e-06,
      "loss": 0.3503,
      "step": 39733
    },
    {
      "epoch": 1.823413335780827,
      "grad_norm": 0.488800585269928,
      "learning_rate": 3.4579467801126078e-06,
      "loss": 0.3922,
      "step": 39734
    },
    {
      "epoch": 1.8234592262860814,
      "grad_norm": 0.4423036277294159,
      "learning_rate": 3.4577135461566636e-06,
      "loss": 0.3122,
      "step": 39735
    },
    {
      "epoch": 1.8235051167913359,
      "grad_norm": 0.48434388637542725,
      "learning_rate": 3.4574803159094294e-06,
      "loss": 0.3617,
      "step": 39736
    },
    {
      "epoch": 1.8235510072965904,
      "grad_norm": 0.5010190010070801,
      "learning_rate": 3.45724708937147e-06,
      "loss": 0.3892,
      "step": 39737
    },
    {
      "epoch": 1.8235968978018446,
      "grad_norm": 0.43173491954803467,
      "learning_rate": 3.457013866543344e-06,
      "loss": 0.2992,
      "step": 39738
    },
    {
      "epoch": 1.8236427883070991,
      "grad_norm": 0.48391714692115784,
      "learning_rate": 3.456780647425611e-06,
      "loss": 0.3222,
      "step": 39739
    },
    {
      "epoch": 1.8236886788123536,
      "grad_norm": 0.41770222783088684,
      "learning_rate": 3.4565474320188354e-06,
      "loss": 0.2762,
      "step": 39740
    },
    {
      "epoch": 1.8237345693176081,
      "grad_norm": 0.43311479687690735,
      "learning_rate": 3.456314220323576e-06,
      "loss": 0.2665,
      "step": 39741
    },
    {
      "epoch": 1.8237804598228626,
      "grad_norm": 0.468510240316391,
      "learning_rate": 3.4560810123403933e-06,
      "loss": 0.3814,
      "step": 39742
    },
    {
      "epoch": 1.8238263503281171,
      "grad_norm": 0.43205875158309937,
      "learning_rate": 3.4558478080698487e-06,
      "loss": 0.2997,
      "step": 39743
    },
    {
      "epoch": 1.8238722408333716,
      "grad_norm": 0.6720637083053589,
      "learning_rate": 3.455614607512503e-06,
      "loss": 0.4674,
      "step": 39744
    },
    {
      "epoch": 1.8239181313386261,
      "grad_norm": 0.443794846534729,
      "learning_rate": 3.4553814106689143e-06,
      "loss": 0.3176,
      "step": 39745
    },
    {
      "epoch": 1.8239640218438806,
      "grad_norm": 0.4892192482948303,
      "learning_rate": 3.455148217539649e-06,
      "loss": 0.3601,
      "step": 39746
    },
    {
      "epoch": 1.8240099123491351,
      "grad_norm": 0.4681258797645569,
      "learning_rate": 3.4549150281252635e-06,
      "loss": 0.3648,
      "step": 39747
    },
    {
      "epoch": 1.8240558028543894,
      "grad_norm": 0.4436596632003784,
      "learning_rate": 3.4546818424263194e-06,
      "loss": 0.3328,
      "step": 39748
    },
    {
      "epoch": 1.824101693359644,
      "grad_norm": 0.4754773676395416,
      "learning_rate": 3.454448660443378e-06,
      "loss": 0.3998,
      "step": 39749
    },
    {
      "epoch": 1.8241475838648984,
      "grad_norm": 0.5269674062728882,
      "learning_rate": 3.4542154821770006e-06,
      "loss": 0.4163,
      "step": 39750
    },
    {
      "epoch": 1.8241934743701527,
      "grad_norm": 0.43985024094581604,
      "learning_rate": 3.453982307627747e-06,
      "loss": 0.3125,
      "step": 39751
    },
    {
      "epoch": 1.8242393648754072,
      "grad_norm": 0.49593809247016907,
      "learning_rate": 3.453749136796176e-06,
      "loss": 0.3766,
      "step": 39752
    },
    {
      "epoch": 1.8242852553806617,
      "grad_norm": 0.4544205963611603,
      "learning_rate": 3.453515969682852e-06,
      "loss": 0.2961,
      "step": 39753
    },
    {
      "epoch": 1.8243311458859162,
      "grad_norm": 0.490241140127182,
      "learning_rate": 3.453282806288334e-06,
      "loss": 0.3405,
      "step": 39754
    },
    {
      "epoch": 1.8243770363911707,
      "grad_norm": 0.4823947250843048,
      "learning_rate": 3.4530496466131817e-06,
      "loss": 0.3778,
      "step": 39755
    },
    {
      "epoch": 1.8244229268964252,
      "grad_norm": 0.48063433170318604,
      "learning_rate": 3.452816490657958e-06,
      "loss": 0.3695,
      "step": 39756
    },
    {
      "epoch": 1.8244688174016797,
      "grad_norm": 0.45817458629608154,
      "learning_rate": 3.4525833384232222e-06,
      "loss": 0.3478,
      "step": 39757
    },
    {
      "epoch": 1.8245147079069342,
      "grad_norm": 0.4703139662742615,
      "learning_rate": 3.452350189909534e-06,
      "loss": 0.3877,
      "step": 39758
    },
    {
      "epoch": 1.8245605984121887,
      "grad_norm": 0.46057528257369995,
      "learning_rate": 3.4521170451174566e-06,
      "loss": 0.3769,
      "step": 39759
    },
    {
      "epoch": 1.824606488917443,
      "grad_norm": 0.6564388871192932,
      "learning_rate": 3.4518839040475486e-06,
      "loss": 0.3968,
      "step": 39760
    },
    {
      "epoch": 1.8246523794226974,
      "grad_norm": 0.4606430232524872,
      "learning_rate": 3.4516507667003695e-06,
      "loss": 0.3093,
      "step": 39761
    },
    {
      "epoch": 1.824698269927952,
      "grad_norm": 0.47507789731025696,
      "learning_rate": 3.4514176330764837e-06,
      "loss": 0.356,
      "step": 39762
    },
    {
      "epoch": 1.8247441604332064,
      "grad_norm": 0.4683626592159271,
      "learning_rate": 3.4511845031764496e-06,
      "loss": 0.3369,
      "step": 39763
    },
    {
      "epoch": 1.8247900509384607,
      "grad_norm": 0.4159708023071289,
      "learning_rate": 3.4509513770008273e-06,
      "loss": 0.2911,
      "step": 39764
    },
    {
      "epoch": 1.8248359414437152,
      "grad_norm": 0.4397841989994049,
      "learning_rate": 3.450718254550178e-06,
      "loss": 0.3243,
      "step": 39765
    },
    {
      "epoch": 1.8248818319489697,
      "grad_norm": 0.4882018268108368,
      "learning_rate": 3.4504851358250637e-06,
      "loss": 0.3496,
      "step": 39766
    },
    {
      "epoch": 1.8249277224542242,
      "grad_norm": 0.4699988067150116,
      "learning_rate": 3.4502520208260403e-06,
      "loss": 0.3518,
      "step": 39767
    },
    {
      "epoch": 1.8249736129594787,
      "grad_norm": 0.45117104053497314,
      "learning_rate": 3.4500189095536752e-06,
      "loss": 0.3208,
      "step": 39768
    },
    {
      "epoch": 1.8250195034647332,
      "grad_norm": 0.45605313777923584,
      "learning_rate": 3.449785802008524e-06,
      "loss": 0.3329,
      "step": 39769
    },
    {
      "epoch": 1.8250653939699877,
      "grad_norm": 0.4631207287311554,
      "learning_rate": 3.4495526981911488e-06,
      "loss": 0.3425,
      "step": 39770
    },
    {
      "epoch": 1.8251112844752422,
      "grad_norm": 0.4506734609603882,
      "learning_rate": 3.4493195981021097e-06,
      "loss": 0.3316,
      "step": 39771
    },
    {
      "epoch": 1.8251571749804967,
      "grad_norm": 0.4647207260131836,
      "learning_rate": 3.449086501741969e-06,
      "loss": 0.3362,
      "step": 39772
    },
    {
      "epoch": 1.825203065485751,
      "grad_norm": 0.4753653407096863,
      "learning_rate": 3.4488534091112828e-06,
      "loss": 0.3367,
      "step": 39773
    },
    {
      "epoch": 1.8252489559910055,
      "grad_norm": 0.513957142829895,
      "learning_rate": 3.448620320210617e-06,
      "loss": 0.3708,
      "step": 39774
    },
    {
      "epoch": 1.82529484649626,
      "grad_norm": 0.444884717464447,
      "learning_rate": 3.44838723504053e-06,
      "loss": 0.3327,
      "step": 39775
    },
    {
      "epoch": 1.8253407370015142,
      "grad_norm": 0.4631721079349518,
      "learning_rate": 3.4481541536015805e-06,
      "loss": 0.3553,
      "step": 39776
    },
    {
      "epoch": 1.8253866275067687,
      "grad_norm": 0.4578433036804199,
      "learning_rate": 3.4479210758943316e-06,
      "loss": 0.3217,
      "step": 39777
    },
    {
      "epoch": 1.8254325180120232,
      "grad_norm": 0.4629529118537903,
      "learning_rate": 3.4476880019193425e-06,
      "loss": 0.3538,
      "step": 39778
    },
    {
      "epoch": 1.8254784085172777,
      "grad_norm": 0.4903775751590729,
      "learning_rate": 3.447454931677172e-06,
      "loss": 0.3959,
      "step": 39779
    },
    {
      "epoch": 1.8255242990225322,
      "grad_norm": 0.4303648769855499,
      "learning_rate": 3.4472218651683853e-06,
      "loss": 0.2586,
      "step": 39780
    },
    {
      "epoch": 1.8255701895277867,
      "grad_norm": 0.4740014672279358,
      "learning_rate": 3.4469888023935396e-06,
      "loss": 0.3578,
      "step": 39781
    },
    {
      "epoch": 1.8256160800330412,
      "grad_norm": 0.4490221440792084,
      "learning_rate": 3.4467557433531927e-06,
      "loss": 0.2949,
      "step": 39782
    },
    {
      "epoch": 1.8256619705382957,
      "grad_norm": 0.5850761532783508,
      "learning_rate": 3.446522688047911e-06,
      "loss": 0.411,
      "step": 39783
    },
    {
      "epoch": 1.8257078610435502,
      "grad_norm": 0.4485396146774292,
      "learning_rate": 3.4462896364782505e-06,
      "loss": 0.3439,
      "step": 39784
    },
    {
      "epoch": 1.8257537515488047,
      "grad_norm": 0.4747643768787384,
      "learning_rate": 3.4460565886447746e-06,
      "loss": 0.3826,
      "step": 39785
    },
    {
      "epoch": 1.825799642054059,
      "grad_norm": 0.503736138343811,
      "learning_rate": 3.4458235445480404e-06,
      "loss": 0.3549,
      "step": 39786
    },
    {
      "epoch": 1.8258455325593135,
      "grad_norm": 0.46490034461021423,
      "learning_rate": 3.4455905041886113e-06,
      "loss": 0.3133,
      "step": 39787
    },
    {
      "epoch": 1.825891423064568,
      "grad_norm": 0.4739399552345276,
      "learning_rate": 3.4453574675670463e-06,
      "loss": 0.3926,
      "step": 39788
    },
    {
      "epoch": 1.8259373135698223,
      "grad_norm": 0.4864806830883026,
      "learning_rate": 3.4451244346839043e-06,
      "loss": 0.4054,
      "step": 39789
    },
    {
      "epoch": 1.8259832040750767,
      "grad_norm": 0.44184496998786926,
      "learning_rate": 3.4448914055397488e-06,
      "loss": 0.3281,
      "step": 39790
    },
    {
      "epoch": 1.8260290945803312,
      "grad_norm": 0.4813521206378937,
      "learning_rate": 3.4446583801351386e-06,
      "loss": 0.4117,
      "step": 39791
    },
    {
      "epoch": 1.8260749850855857,
      "grad_norm": 0.43435606360435486,
      "learning_rate": 3.4444253584706334e-06,
      "loss": 0.3239,
      "step": 39792
    },
    {
      "epoch": 1.8261208755908402,
      "grad_norm": 0.4891532361507416,
      "learning_rate": 3.4441923405467947e-06,
      "loss": 0.3756,
      "step": 39793
    },
    {
      "epoch": 1.8261667660960947,
      "grad_norm": 0.45238667726516724,
      "learning_rate": 3.4439593263641835e-06,
      "loss": 0.3422,
      "step": 39794
    },
    {
      "epoch": 1.8262126566013492,
      "grad_norm": 0.46222174167633057,
      "learning_rate": 3.4437263159233557e-06,
      "loss": 0.3502,
      "step": 39795
    },
    {
      "epoch": 1.8262585471066037,
      "grad_norm": 0.43490520119667053,
      "learning_rate": 3.443493309224878e-06,
      "loss": 0.2861,
      "step": 39796
    },
    {
      "epoch": 1.8263044376118582,
      "grad_norm": 0.5129135251045227,
      "learning_rate": 3.4432603062693073e-06,
      "loss": 0.3775,
      "step": 39797
    },
    {
      "epoch": 1.8263503281171127,
      "grad_norm": 0.47910308837890625,
      "learning_rate": 3.443027307057203e-06,
      "loss": 0.3496,
      "step": 39798
    },
    {
      "epoch": 1.826396218622367,
      "grad_norm": 0.47066253423690796,
      "learning_rate": 3.442794311589128e-06,
      "loss": 0.3689,
      "step": 39799
    },
    {
      "epoch": 1.8264421091276215,
      "grad_norm": 0.4875509738922119,
      "learning_rate": 3.4425613198656417e-06,
      "loss": 0.3551,
      "step": 39800
    },
    {
      "epoch": 1.826487999632876,
      "grad_norm": 0.4715254604816437,
      "learning_rate": 3.442328331887301e-06,
      "loss": 0.3456,
      "step": 39801
    },
    {
      "epoch": 1.8265338901381303,
      "grad_norm": 0.4701811969280243,
      "learning_rate": 3.4420953476546724e-06,
      "loss": 0.3607,
      "step": 39802
    },
    {
      "epoch": 1.8265797806433848,
      "grad_norm": 0.4681417644023895,
      "learning_rate": 3.4418623671683114e-06,
      "loss": 0.3312,
      "step": 39803
    },
    {
      "epoch": 1.8266256711486393,
      "grad_norm": 0.46512049436569214,
      "learning_rate": 3.4416293904287777e-06,
      "loss": 0.3428,
      "step": 39804
    },
    {
      "epoch": 1.8266715616538938,
      "grad_norm": 0.46033889055252075,
      "learning_rate": 3.441396417436636e-06,
      "loss": 0.3423,
      "step": 39805
    },
    {
      "epoch": 1.8267174521591483,
      "grad_norm": 0.476309210062027,
      "learning_rate": 3.441163448192444e-06,
      "loss": 0.34,
      "step": 39806
    },
    {
      "epoch": 1.8267633426644028,
      "grad_norm": 0.49267256259918213,
      "learning_rate": 3.4409304826967606e-06,
      "loss": 0.3852,
      "step": 39807
    },
    {
      "epoch": 1.8268092331696573,
      "grad_norm": 0.4587656557559967,
      "learning_rate": 3.440697520950148e-06,
      "loss": 0.3373,
      "step": 39808
    },
    {
      "epoch": 1.8268551236749118,
      "grad_norm": 0.45361581444740295,
      "learning_rate": 3.4404645629531665e-06,
      "loss": 0.3508,
      "step": 39809
    },
    {
      "epoch": 1.8269010141801663,
      "grad_norm": 0.4888070225715637,
      "learning_rate": 3.440231608706373e-06,
      "loss": 0.402,
      "step": 39810
    },
    {
      "epoch": 1.8269469046854205,
      "grad_norm": 0.4693959653377533,
      "learning_rate": 3.4399986582103324e-06,
      "loss": 0.3154,
      "step": 39811
    },
    {
      "epoch": 1.826992795190675,
      "grad_norm": 0.49927473068237305,
      "learning_rate": 3.4397657114656025e-06,
      "loss": 0.3884,
      "step": 39812
    },
    {
      "epoch": 1.8270386856959295,
      "grad_norm": 0.41470178961753845,
      "learning_rate": 3.439532768472743e-06,
      "loss": 0.2709,
      "step": 39813
    },
    {
      "epoch": 1.827084576201184,
      "grad_norm": 0.46712177991867065,
      "learning_rate": 3.4392998292323153e-06,
      "loss": 0.3596,
      "step": 39814
    },
    {
      "epoch": 1.8271304667064383,
      "grad_norm": 0.47077158093452454,
      "learning_rate": 3.439066893744879e-06,
      "loss": 0.3682,
      "step": 39815
    },
    {
      "epoch": 1.8271763572116928,
      "grad_norm": 0.5060456991195679,
      "learning_rate": 3.4388339620109947e-06,
      "loss": 0.4576,
      "step": 39816
    },
    {
      "epoch": 1.8272222477169473,
      "grad_norm": 0.469989538192749,
      "learning_rate": 3.4386010340312192e-06,
      "loss": 0.357,
      "step": 39817
    },
    {
      "epoch": 1.8272681382222018,
      "grad_norm": 0.4360527992248535,
      "learning_rate": 3.4383681098061185e-06,
      "loss": 0.2945,
      "step": 39818
    },
    {
      "epoch": 1.8273140287274563,
      "grad_norm": 0.46837764978408813,
      "learning_rate": 3.4381351893362486e-06,
      "loss": 0.3872,
      "step": 39819
    },
    {
      "epoch": 1.8273599192327108,
      "grad_norm": 0.4442499279975891,
      "learning_rate": 3.43790227262217e-06,
      "loss": 0.2968,
      "step": 39820
    },
    {
      "epoch": 1.8274058097379653,
      "grad_norm": 0.443032830953598,
      "learning_rate": 3.437669359664444e-06,
      "loss": 0.2933,
      "step": 39821
    },
    {
      "epoch": 1.8274517002432198,
      "grad_norm": 0.41946423053741455,
      "learning_rate": 3.4374364504636305e-06,
      "loss": 0.2815,
      "step": 39822
    },
    {
      "epoch": 1.8274975907484743,
      "grad_norm": 0.49188661575317383,
      "learning_rate": 3.4372035450202872e-06,
      "loss": 0.389,
      "step": 39823
    },
    {
      "epoch": 1.8275434812537286,
      "grad_norm": 0.4841740131378174,
      "learning_rate": 3.4369706433349793e-06,
      "loss": 0.3602,
      "step": 39824
    },
    {
      "epoch": 1.827589371758983,
      "grad_norm": 0.48956382274627686,
      "learning_rate": 3.4367377454082628e-06,
      "loss": 0.3884,
      "step": 39825
    },
    {
      "epoch": 1.8276352622642376,
      "grad_norm": 0.49822378158569336,
      "learning_rate": 3.4365048512406962e-06,
      "loss": 0.4251,
      "step": 39826
    },
    {
      "epoch": 1.8276811527694918,
      "grad_norm": 0.5036949515342712,
      "learning_rate": 3.4362719608328443e-06,
      "loss": 0.3642,
      "step": 39827
    },
    {
      "epoch": 1.8277270432747463,
      "grad_norm": 0.47843703627586365,
      "learning_rate": 3.4360390741852646e-06,
      "loss": 0.3588,
      "step": 39828
    },
    {
      "epoch": 1.8277729337800008,
      "grad_norm": 0.4904283285140991,
      "learning_rate": 3.4358061912985163e-06,
      "loss": 0.3453,
      "step": 39829
    },
    {
      "epoch": 1.8278188242852553,
      "grad_norm": 0.44428929686546326,
      "learning_rate": 3.4355733121731614e-06,
      "loss": 0.3184,
      "step": 39830
    },
    {
      "epoch": 1.8278647147905098,
      "grad_norm": 0.514350414276123,
      "learning_rate": 3.4353404368097585e-06,
      "loss": 0.4048,
      "step": 39831
    },
    {
      "epoch": 1.8279106052957643,
      "grad_norm": 0.48060178756713867,
      "learning_rate": 3.435107565208867e-06,
      "loss": 0.3627,
      "step": 39832
    },
    {
      "epoch": 1.8279564958010188,
      "grad_norm": 0.5014491677284241,
      "learning_rate": 3.4348746973710495e-06,
      "loss": 0.4083,
      "step": 39833
    },
    {
      "epoch": 1.8280023863062733,
      "grad_norm": 0.4820750951766968,
      "learning_rate": 3.4346418332968644e-06,
      "loss": 0.3833,
      "step": 39834
    },
    {
      "epoch": 1.8280482768115278,
      "grad_norm": 0.43934452533721924,
      "learning_rate": 3.4344089729868705e-06,
      "loss": 0.3061,
      "step": 39835
    },
    {
      "epoch": 1.8280941673167823,
      "grad_norm": 0.43806639313697815,
      "learning_rate": 3.4341761164416297e-06,
      "loss": 0.2665,
      "step": 39836
    },
    {
      "epoch": 1.8281400578220366,
      "grad_norm": 0.45623186230659485,
      "learning_rate": 3.433943263661702e-06,
      "loss": 0.2958,
      "step": 39837
    },
    {
      "epoch": 1.828185948327291,
      "grad_norm": 0.46764761209487915,
      "learning_rate": 3.433710414647644e-06,
      "loss": 0.359,
      "step": 39838
    },
    {
      "epoch": 1.8282318388325456,
      "grad_norm": 0.44870442152023315,
      "learning_rate": 3.43347756940002e-06,
      "loss": 0.2754,
      "step": 39839
    },
    {
      "epoch": 1.8282777293377999,
      "grad_norm": 0.5139065980911255,
      "learning_rate": 3.433244727919388e-06,
      "loss": 0.4244,
      "step": 39840
    },
    {
      "epoch": 1.8283236198430544,
      "grad_norm": 0.4684973955154419,
      "learning_rate": 3.433011890206307e-06,
      "loss": 0.34,
      "step": 39841
    },
    {
      "epoch": 1.8283695103483089,
      "grad_norm": 0.43400582671165466,
      "learning_rate": 3.4327790562613387e-06,
      "loss": 0.3011,
      "step": 39842
    },
    {
      "epoch": 1.8284154008535634,
      "grad_norm": 0.44612035155296326,
      "learning_rate": 3.432546226085043e-06,
      "loss": 0.3126,
      "step": 39843
    },
    {
      "epoch": 1.8284612913588179,
      "grad_norm": 0.4521428942680359,
      "learning_rate": 3.4323133996779755e-06,
      "loss": 0.342,
      "step": 39844
    },
    {
      "epoch": 1.8285071818640724,
      "grad_norm": 0.4564052224159241,
      "learning_rate": 3.4320805770407028e-06,
      "loss": 0.3322,
      "step": 39845
    },
    {
      "epoch": 1.8285530723693268,
      "grad_norm": 0.49334412813186646,
      "learning_rate": 3.431847758173782e-06,
      "loss": 0.3541,
      "step": 39846
    },
    {
      "epoch": 1.8285989628745813,
      "grad_norm": 0.5017808675765991,
      "learning_rate": 3.431614943077769e-06,
      "loss": 0.4264,
      "step": 39847
    },
    {
      "epoch": 1.8286448533798358,
      "grad_norm": 0.4284824728965759,
      "learning_rate": 3.43138213175323e-06,
      "loss": 0.2848,
      "step": 39848
    },
    {
      "epoch": 1.8286907438850901,
      "grad_norm": 0.4694102704524994,
      "learning_rate": 3.431149324200721e-06,
      "loss": 0.3354,
      "step": 39849
    },
    {
      "epoch": 1.8287366343903446,
      "grad_norm": 0.479158878326416,
      "learning_rate": 3.430916520420803e-06,
      "loss": 0.3564,
      "step": 39850
    },
    {
      "epoch": 1.8287825248955991,
      "grad_norm": 0.5006948113441467,
      "learning_rate": 3.4306837204140346e-06,
      "loss": 0.374,
      "step": 39851
    },
    {
      "epoch": 1.8288284154008536,
      "grad_norm": 0.46287286281585693,
      "learning_rate": 3.430450924180978e-06,
      "loss": 0.3488,
      "step": 39852
    },
    {
      "epoch": 1.8288743059061079,
      "grad_norm": 0.48436230421066284,
      "learning_rate": 3.4302181317221905e-06,
      "loss": 0.3773,
      "step": 39853
    },
    {
      "epoch": 1.8289201964113624,
      "grad_norm": 0.4432602524757385,
      "learning_rate": 3.4299853430382318e-06,
      "loss": 0.3326,
      "step": 39854
    },
    {
      "epoch": 1.8289660869166169,
      "grad_norm": 0.4736188054084778,
      "learning_rate": 3.429752558129664e-06,
      "loss": 0.3141,
      "step": 39855
    },
    {
      "epoch": 1.8290119774218714,
      "grad_norm": 0.48585787415504456,
      "learning_rate": 3.429519776997047e-06,
      "loss": 0.3587,
      "step": 39856
    },
    {
      "epoch": 1.8290578679271259,
      "grad_norm": 0.5037298202514648,
      "learning_rate": 3.4292869996409374e-06,
      "loss": 0.3874,
      "step": 39857
    },
    {
      "epoch": 1.8291037584323804,
      "grad_norm": 0.47064104676246643,
      "learning_rate": 3.4290542260618987e-06,
      "loss": 0.3342,
      "step": 39858
    },
    {
      "epoch": 1.8291496489376349,
      "grad_norm": 0.4904780685901642,
      "learning_rate": 3.4288214562604876e-06,
      "loss": 0.3616,
      "step": 39859
    },
    {
      "epoch": 1.8291955394428894,
      "grad_norm": 0.42936259508132935,
      "learning_rate": 3.428588690237263e-06,
      "loss": 0.2876,
      "step": 39860
    },
    {
      "epoch": 1.8292414299481439,
      "grad_norm": 0.44854018092155457,
      "learning_rate": 3.4283559279927895e-06,
      "loss": 0.3426,
      "step": 39861
    },
    {
      "epoch": 1.8292873204533981,
      "grad_norm": 0.4780154228210449,
      "learning_rate": 3.428123169527624e-06,
      "loss": 0.3507,
      "step": 39862
    },
    {
      "epoch": 1.8293332109586526,
      "grad_norm": 0.4782347083091736,
      "learning_rate": 3.4278904148423247e-06,
      "loss": 0.3607,
      "step": 39863
    },
    {
      "epoch": 1.8293791014639071,
      "grad_norm": 0.4915837347507477,
      "learning_rate": 3.4276576639374542e-06,
      "loss": 0.3442,
      "step": 39864
    },
    {
      "epoch": 1.8294249919691614,
      "grad_norm": 0.47087234258651733,
      "learning_rate": 3.4274249168135702e-06,
      "loss": 0.3416,
      "step": 39865
    },
    {
      "epoch": 1.829470882474416,
      "grad_norm": 0.4413788914680481,
      "learning_rate": 3.427192173471231e-06,
      "loss": 0.3168,
      "step": 39866
    },
    {
      "epoch": 1.8295167729796704,
      "grad_norm": 0.5047722458839417,
      "learning_rate": 3.4269594339110014e-06,
      "loss": 0.3886,
      "step": 39867
    },
    {
      "epoch": 1.829562663484925,
      "grad_norm": 0.43959546089172363,
      "learning_rate": 3.426726698133438e-06,
      "loss": 0.3056,
      "step": 39868
    },
    {
      "epoch": 1.8296085539901794,
      "grad_norm": 0.4711773991584778,
      "learning_rate": 3.4264939661390973e-06,
      "loss": 0.3377,
      "step": 39869
    },
    {
      "epoch": 1.829654444495434,
      "grad_norm": 0.4531530439853668,
      "learning_rate": 3.426261237928544e-06,
      "loss": 0.3111,
      "step": 39870
    },
    {
      "epoch": 1.8297003350006884,
      "grad_norm": 0.4908592402935028,
      "learning_rate": 3.4260285135023353e-06,
      "loss": 0.3614,
      "step": 39871
    },
    {
      "epoch": 1.829746225505943,
      "grad_norm": 0.4415791928768158,
      "learning_rate": 3.425795792861031e-06,
      "loss": 0.3429,
      "step": 39872
    },
    {
      "epoch": 1.8297921160111974,
      "grad_norm": 0.48063236474990845,
      "learning_rate": 3.4255630760051916e-06,
      "loss": 0.3899,
      "step": 39873
    },
    {
      "epoch": 1.829838006516452,
      "grad_norm": 0.5113020539283752,
      "learning_rate": 3.4253303629353762e-06,
      "loss": 0.4156,
      "step": 39874
    },
    {
      "epoch": 1.8298838970217062,
      "grad_norm": 0.46374186873435974,
      "learning_rate": 3.425097653652142e-06,
      "loss": 0.3063,
      "step": 39875
    },
    {
      "epoch": 1.8299297875269607,
      "grad_norm": 0.4975281357765198,
      "learning_rate": 3.4248649481560535e-06,
      "loss": 0.3633,
      "step": 39876
    },
    {
      "epoch": 1.8299756780322152,
      "grad_norm": 0.4965936541557312,
      "learning_rate": 3.4246322464476676e-06,
      "loss": 0.3655,
      "step": 39877
    },
    {
      "epoch": 1.8300215685374694,
      "grad_norm": 0.4819747805595398,
      "learning_rate": 3.424399548527543e-06,
      "loss": 0.3423,
      "step": 39878
    },
    {
      "epoch": 1.830067459042724,
      "grad_norm": 0.48193904757499695,
      "learning_rate": 3.4241668543962403e-06,
      "loss": 0.384,
      "step": 39879
    },
    {
      "epoch": 1.8301133495479784,
      "grad_norm": 0.4784378409385681,
      "learning_rate": 3.42393416405432e-06,
      "loss": 0.3506,
      "step": 39880
    },
    {
      "epoch": 1.830159240053233,
      "grad_norm": 0.49161818623542786,
      "learning_rate": 3.423701477502338e-06,
      "loss": 0.4141,
      "step": 39881
    },
    {
      "epoch": 1.8302051305584874,
      "grad_norm": 0.4426211416721344,
      "learning_rate": 3.4234687947408586e-06,
      "loss": 0.3017,
      "step": 39882
    },
    {
      "epoch": 1.830251021063742,
      "grad_norm": 0.5334861874580383,
      "learning_rate": 3.4232361157704395e-06,
      "loss": 0.4947,
      "step": 39883
    },
    {
      "epoch": 1.8302969115689964,
      "grad_norm": 0.4907417893409729,
      "learning_rate": 3.4230034405916397e-06,
      "loss": 0.379,
      "step": 39884
    },
    {
      "epoch": 1.830342802074251,
      "grad_norm": 0.43633753061294556,
      "learning_rate": 3.4227707692050178e-06,
      "loss": 0.2903,
      "step": 39885
    },
    {
      "epoch": 1.8303886925795054,
      "grad_norm": 0.44748276472091675,
      "learning_rate": 3.4225381016111354e-06,
      "loss": 0.3034,
      "step": 39886
    },
    {
      "epoch": 1.83043458308476,
      "grad_norm": 0.45776447653770447,
      "learning_rate": 3.422305437810551e-06,
      "loss": 0.3642,
      "step": 39887
    },
    {
      "epoch": 1.8304804735900142,
      "grad_norm": 0.461580365896225,
      "learning_rate": 3.422072777803822e-06,
      "loss": 0.3569,
      "step": 39888
    },
    {
      "epoch": 1.8305263640952687,
      "grad_norm": 0.46489113569259644,
      "learning_rate": 3.421840121591512e-06,
      "loss": 0.34,
      "step": 39889
    },
    {
      "epoch": 1.8305722546005232,
      "grad_norm": 0.48917025327682495,
      "learning_rate": 3.4216074691741795e-06,
      "loss": 0.3658,
      "step": 39890
    },
    {
      "epoch": 1.8306181451057775,
      "grad_norm": 0.4313655197620392,
      "learning_rate": 3.42137482055238e-06,
      "loss": 0.2858,
      "step": 39891
    },
    {
      "epoch": 1.830664035611032,
      "grad_norm": 0.4934520125389099,
      "learning_rate": 3.4211421757266784e-06,
      "loss": 0.3974,
      "step": 39892
    },
    {
      "epoch": 1.8307099261162865,
      "grad_norm": 0.4365563690662384,
      "learning_rate": 3.4209095346976305e-06,
      "loss": 0.2897,
      "step": 39893
    },
    {
      "epoch": 1.830755816621541,
      "grad_norm": 0.4462529718875885,
      "learning_rate": 3.420676897465796e-06,
      "loss": 0.2998,
      "step": 39894
    },
    {
      "epoch": 1.8308017071267955,
      "grad_norm": 0.4518919289112091,
      "learning_rate": 3.420444264031737e-06,
      "loss": 0.3034,
      "step": 39895
    },
    {
      "epoch": 1.83084759763205,
      "grad_norm": 0.45529159903526306,
      "learning_rate": 3.4202116343960103e-06,
      "loss": 0.3316,
      "step": 39896
    },
    {
      "epoch": 1.8308934881373045,
      "grad_norm": 0.47908708453178406,
      "learning_rate": 3.4199790085591734e-06,
      "loss": 0.403,
      "step": 39897
    },
    {
      "epoch": 1.830939378642559,
      "grad_norm": 0.4375518560409546,
      "learning_rate": 3.4197463865217918e-06,
      "loss": 0.281,
      "step": 39898
    },
    {
      "epoch": 1.8309852691478135,
      "grad_norm": 0.504301130771637,
      "learning_rate": 3.4195137682844205e-06,
      "loss": 0.3515,
      "step": 39899
    },
    {
      "epoch": 1.8310311596530677,
      "grad_norm": 0.4544859230518341,
      "learning_rate": 3.4192811538476184e-06,
      "loss": 0.3547,
      "step": 39900
    },
    {
      "epoch": 1.8310770501583222,
      "grad_norm": 0.5025752186775208,
      "learning_rate": 3.419048543211948e-06,
      "loss": 0.4656,
      "step": 39901
    },
    {
      "epoch": 1.8311229406635767,
      "grad_norm": 0.46962785720825195,
      "learning_rate": 3.418815936377966e-06,
      "loss": 0.3573,
      "step": 39902
    },
    {
      "epoch": 1.8311688311688312,
      "grad_norm": 0.4398167133331299,
      "learning_rate": 3.4185833333462317e-06,
      "loss": 0.3252,
      "step": 39903
    },
    {
      "epoch": 1.8312147216740855,
      "grad_norm": 0.4696708023548126,
      "learning_rate": 3.4183507341173073e-06,
      "loss": 0.3451,
      "step": 39904
    },
    {
      "epoch": 1.83126061217934,
      "grad_norm": 0.499142587184906,
      "learning_rate": 3.4181181386917497e-06,
      "loss": 0.4089,
      "step": 39905
    },
    {
      "epoch": 1.8313065026845945,
      "grad_norm": 0.43776917457580566,
      "learning_rate": 3.417885547070118e-06,
      "loss": 0.3156,
      "step": 39906
    },
    {
      "epoch": 1.831352393189849,
      "grad_norm": 0.45756399631500244,
      "learning_rate": 3.417652959252973e-06,
      "loss": 0.313,
      "step": 39907
    },
    {
      "epoch": 1.8313982836951035,
      "grad_norm": 0.5262242555618286,
      "learning_rate": 3.417420375240873e-06,
      "loss": 0.3805,
      "step": 39908
    },
    {
      "epoch": 1.831444174200358,
      "grad_norm": 0.4564991891384125,
      "learning_rate": 3.4171877950343755e-06,
      "loss": 0.3181,
      "step": 39909
    },
    {
      "epoch": 1.8314900647056125,
      "grad_norm": 0.4668429493904114,
      "learning_rate": 3.4169552186340444e-06,
      "loss": 0.3389,
      "step": 39910
    },
    {
      "epoch": 1.831535955210867,
      "grad_norm": 0.4947572350502014,
      "learning_rate": 3.4167226460404357e-06,
      "loss": 0.399,
      "step": 39911
    },
    {
      "epoch": 1.8315818457161215,
      "grad_norm": 0.45476171374320984,
      "learning_rate": 3.4164900772541087e-06,
      "loss": 0.3436,
      "step": 39912
    },
    {
      "epoch": 1.8316277362213758,
      "grad_norm": 0.4689728319644928,
      "learning_rate": 3.4162575122756248e-06,
      "loss": 0.352,
      "step": 39913
    },
    {
      "epoch": 1.8316736267266303,
      "grad_norm": 0.47441449761390686,
      "learning_rate": 3.416024951105541e-06,
      "loss": 0.3581,
      "step": 39914
    },
    {
      "epoch": 1.8317195172318848,
      "grad_norm": 0.48141568899154663,
      "learning_rate": 3.4157923937444177e-06,
      "loss": 0.3907,
      "step": 39915
    },
    {
      "epoch": 1.831765407737139,
      "grad_norm": 0.4737670123577118,
      "learning_rate": 3.4155598401928126e-06,
      "loss": 0.3807,
      "step": 39916
    },
    {
      "epoch": 1.8318112982423935,
      "grad_norm": 0.5022450089454651,
      "learning_rate": 3.4153272904512867e-06,
      "loss": 0.4341,
      "step": 39917
    },
    {
      "epoch": 1.831857188747648,
      "grad_norm": 0.4362969696521759,
      "learning_rate": 3.4150947445203985e-06,
      "loss": 0.3085,
      "step": 39918
    },
    {
      "epoch": 1.8319030792529025,
      "grad_norm": 0.45722270011901855,
      "learning_rate": 3.414862202400706e-06,
      "loss": 0.3178,
      "step": 39919
    },
    {
      "epoch": 1.831948969758157,
      "grad_norm": 0.5092957615852356,
      "learning_rate": 3.414629664092771e-06,
      "loss": 0.3642,
      "step": 39920
    },
    {
      "epoch": 1.8319948602634115,
      "grad_norm": 0.4718768298625946,
      "learning_rate": 3.4143971295971514e-06,
      "loss": 0.3758,
      "step": 39921
    },
    {
      "epoch": 1.832040750768666,
      "grad_norm": 0.4614443778991699,
      "learning_rate": 3.4141645989144055e-06,
      "loss": 0.3404,
      "step": 39922
    },
    {
      "epoch": 1.8320866412739205,
      "grad_norm": 0.44428136944770813,
      "learning_rate": 3.4139320720450943e-06,
      "loss": 0.3066,
      "step": 39923
    },
    {
      "epoch": 1.832132531779175,
      "grad_norm": 0.5068833231925964,
      "learning_rate": 3.413699548989775e-06,
      "loss": 0.3566,
      "step": 39924
    },
    {
      "epoch": 1.8321784222844295,
      "grad_norm": 0.4967723488807678,
      "learning_rate": 3.4134670297490064e-06,
      "loss": 0.3694,
      "step": 39925
    },
    {
      "epoch": 1.8322243127896838,
      "grad_norm": 0.46272847056388855,
      "learning_rate": 3.413234514323351e-06,
      "loss": 0.3386,
      "step": 39926
    },
    {
      "epoch": 1.8322702032949383,
      "grad_norm": 0.4361560046672821,
      "learning_rate": 3.4130020027133653e-06,
      "loss": 0.2914,
      "step": 39927
    },
    {
      "epoch": 1.8323160938001928,
      "grad_norm": 0.46583291888237,
      "learning_rate": 3.4127694949196083e-06,
      "loss": 0.3607,
      "step": 39928
    },
    {
      "epoch": 1.832361984305447,
      "grad_norm": 0.4693363606929779,
      "learning_rate": 3.41253699094264e-06,
      "loss": 0.3481,
      "step": 39929
    },
    {
      "epoch": 1.8324078748107016,
      "grad_norm": 0.43972843885421753,
      "learning_rate": 3.4123044907830205e-06,
      "loss": 0.3416,
      "step": 39930
    },
    {
      "epoch": 1.832453765315956,
      "grad_norm": 0.45557788014411926,
      "learning_rate": 3.412071994441305e-06,
      "loss": 0.3111,
      "step": 39931
    },
    {
      "epoch": 1.8324996558212105,
      "grad_norm": 0.44942227005958557,
      "learning_rate": 3.411839501918057e-06,
      "loss": 0.2779,
      "step": 39932
    },
    {
      "epoch": 1.832545546326465,
      "grad_norm": 0.4538784921169281,
      "learning_rate": 3.4116070132138335e-06,
      "loss": 0.3231,
      "step": 39933
    },
    {
      "epoch": 1.8325914368317195,
      "grad_norm": 0.47214362025260925,
      "learning_rate": 3.4113745283291934e-06,
      "loss": 0.3921,
      "step": 39934
    },
    {
      "epoch": 1.832637327336974,
      "grad_norm": 0.4535367786884308,
      "learning_rate": 3.4111420472646965e-06,
      "loss": 0.3406,
      "step": 39935
    },
    {
      "epoch": 1.8326832178422285,
      "grad_norm": 0.4443082809448242,
      "learning_rate": 3.4109095700209022e-06,
      "loss": 0.3302,
      "step": 39936
    },
    {
      "epoch": 1.832729108347483,
      "grad_norm": 0.4912261366844177,
      "learning_rate": 3.4106770965983675e-06,
      "loss": 0.3939,
      "step": 39937
    },
    {
      "epoch": 1.8327749988527373,
      "grad_norm": 0.4652918875217438,
      "learning_rate": 3.4104446269976543e-06,
      "loss": 0.3852,
      "step": 39938
    },
    {
      "epoch": 1.8328208893579918,
      "grad_norm": 0.44766363501548767,
      "learning_rate": 3.4102121612193196e-06,
      "loss": 0.3308,
      "step": 39939
    },
    {
      "epoch": 1.8328667798632463,
      "grad_norm": 0.48971042037010193,
      "learning_rate": 3.409979699263921e-06,
      "loss": 0.3374,
      "step": 39940
    },
    {
      "epoch": 1.8329126703685008,
      "grad_norm": 0.4273504912853241,
      "learning_rate": 3.409747241132021e-06,
      "loss": 0.3065,
      "step": 39941
    },
    {
      "epoch": 1.832958560873755,
      "grad_norm": 0.4703972637653351,
      "learning_rate": 3.4095147868241772e-06,
      "loss": 0.367,
      "step": 39942
    },
    {
      "epoch": 1.8330044513790096,
      "grad_norm": 0.48062074184417725,
      "learning_rate": 3.4092823363409477e-06,
      "loss": 0.3815,
      "step": 39943
    },
    {
      "epoch": 1.833050341884264,
      "grad_norm": 0.4576103985309601,
      "learning_rate": 3.409049889682893e-06,
      "loss": 0.3412,
      "step": 39944
    },
    {
      "epoch": 1.8330962323895186,
      "grad_norm": 0.490282267332077,
      "learning_rate": 3.4088174468505713e-06,
      "loss": 0.3745,
      "step": 39945
    },
    {
      "epoch": 1.833142122894773,
      "grad_norm": 0.45636287331581116,
      "learning_rate": 3.408585007844539e-06,
      "loss": 0.3342,
      "step": 39946
    },
    {
      "epoch": 1.8331880134000276,
      "grad_norm": 0.4622512757778168,
      "learning_rate": 3.40835257266536e-06,
      "loss": 0.3209,
      "step": 39947
    },
    {
      "epoch": 1.833233903905282,
      "grad_norm": 0.49249333143234253,
      "learning_rate": 3.408120141313591e-06,
      "loss": 0.3949,
      "step": 39948
    },
    {
      "epoch": 1.8332797944105366,
      "grad_norm": 0.4692722260951996,
      "learning_rate": 3.40788771378979e-06,
      "loss": 0.3698,
      "step": 39949
    },
    {
      "epoch": 1.833325684915791,
      "grad_norm": 0.4519585967063904,
      "learning_rate": 3.407655290094516e-06,
      "loss": 0.3191,
      "step": 39950
    },
    {
      "epoch": 1.8333715754210453,
      "grad_norm": 0.4598078429698944,
      "learning_rate": 3.4074228702283287e-06,
      "loss": 0.326,
      "step": 39951
    },
    {
      "epoch": 1.8334174659262998,
      "grad_norm": 0.4811152219772339,
      "learning_rate": 3.407190454191788e-06,
      "loss": 0.3936,
      "step": 39952
    },
    {
      "epoch": 1.8334633564315543,
      "grad_norm": 0.43513578176498413,
      "learning_rate": 3.406958041985448e-06,
      "loss": 0.3211,
      "step": 39953
    },
    {
      "epoch": 1.8335092469368086,
      "grad_norm": 0.4456690549850464,
      "learning_rate": 3.4067256336098744e-06,
      "loss": 0.2995,
      "step": 39954
    },
    {
      "epoch": 1.833555137442063,
      "grad_norm": 0.4899369776248932,
      "learning_rate": 3.406493229065622e-06,
      "loss": 0.361,
      "step": 39955
    },
    {
      "epoch": 1.8336010279473176,
      "grad_norm": 0.4825911521911621,
      "learning_rate": 3.40626082835325e-06,
      "loss": 0.3508,
      "step": 39956
    },
    {
      "epoch": 1.833646918452572,
      "grad_norm": 0.4323596656322479,
      "learning_rate": 3.406028431473318e-06,
      "loss": 0.2819,
      "step": 39957
    },
    {
      "epoch": 1.8336928089578266,
      "grad_norm": 0.4366082549095154,
      "learning_rate": 3.4057960384263857e-06,
      "loss": 0.3289,
      "step": 39958
    },
    {
      "epoch": 1.833738699463081,
      "grad_norm": 0.47617030143737793,
      "learning_rate": 3.405563649213009e-06,
      "loss": 0.3799,
      "step": 39959
    },
    {
      "epoch": 1.8337845899683356,
      "grad_norm": 0.48050934076309204,
      "learning_rate": 3.40533126383375e-06,
      "loss": 0.4027,
      "step": 39960
    },
    {
      "epoch": 1.83383048047359,
      "grad_norm": 0.4609653353691101,
      "learning_rate": 3.405098882289165e-06,
      "loss": 0.3463,
      "step": 39961
    },
    {
      "epoch": 1.8338763709788446,
      "grad_norm": 0.5062091946601868,
      "learning_rate": 3.404866504579813e-06,
      "loss": 0.3904,
      "step": 39962
    },
    {
      "epoch": 1.833922261484099,
      "grad_norm": 0.4677774906158447,
      "learning_rate": 3.4046341307062547e-06,
      "loss": 0.3391,
      "step": 39963
    },
    {
      "epoch": 1.8339681519893534,
      "grad_norm": 0.47597381472587585,
      "learning_rate": 3.4044017606690485e-06,
      "loss": 0.3594,
      "step": 39964
    },
    {
      "epoch": 1.8340140424946079,
      "grad_norm": 0.508150577545166,
      "learning_rate": 3.4041693944687515e-06,
      "loss": 0.4189,
      "step": 39965
    },
    {
      "epoch": 1.8340599329998624,
      "grad_norm": 0.4767831265926361,
      "learning_rate": 3.403937032105924e-06,
      "loss": 0.3583,
      "step": 39966
    },
    {
      "epoch": 1.8341058235051166,
      "grad_norm": 0.473952054977417,
      "learning_rate": 3.403704673581124e-06,
      "loss": 0.3991,
      "step": 39967
    },
    {
      "epoch": 1.8341517140103711,
      "grad_norm": 0.5166555643081665,
      "learning_rate": 3.4034723188949092e-06,
      "loss": 0.4863,
      "step": 39968
    },
    {
      "epoch": 1.8341976045156256,
      "grad_norm": 0.48794183135032654,
      "learning_rate": 3.4032399680478413e-06,
      "loss": 0.4529,
      "step": 39969
    },
    {
      "epoch": 1.8342434950208801,
      "grad_norm": 0.46893325448036194,
      "learning_rate": 3.403007621040477e-06,
      "loss": 0.3336,
      "step": 39970
    },
    {
      "epoch": 1.8342893855261346,
      "grad_norm": 0.4389420747756958,
      "learning_rate": 3.4027752778733747e-06,
      "loss": 0.3264,
      "step": 39971
    },
    {
      "epoch": 1.8343352760313891,
      "grad_norm": 0.46775805950164795,
      "learning_rate": 3.402542938547094e-06,
      "loss": 0.3536,
      "step": 39972
    },
    {
      "epoch": 1.8343811665366436,
      "grad_norm": 0.47164833545684814,
      "learning_rate": 3.4023106030621943e-06,
      "loss": 0.4151,
      "step": 39973
    },
    {
      "epoch": 1.8344270570418981,
      "grad_norm": 0.48729804158210754,
      "learning_rate": 3.4020782714192314e-06,
      "loss": 0.3967,
      "step": 39974
    },
    {
      "epoch": 1.8344729475471526,
      "grad_norm": 0.48050767183303833,
      "learning_rate": 3.401845943618768e-06,
      "loss": 0.3341,
      "step": 39975
    },
    {
      "epoch": 1.8345188380524071,
      "grad_norm": 0.47942933440208435,
      "learning_rate": 3.4016136196613596e-06,
      "loss": 0.3809,
      "step": 39976
    },
    {
      "epoch": 1.8345647285576614,
      "grad_norm": 0.4591526687145233,
      "learning_rate": 3.401381299547566e-06,
      "loss": 0.3164,
      "step": 39977
    },
    {
      "epoch": 1.834610619062916,
      "grad_norm": 0.44258183240890503,
      "learning_rate": 3.4011489832779465e-06,
      "loss": 0.3123,
      "step": 39978
    },
    {
      "epoch": 1.8346565095681704,
      "grad_norm": 0.45343753695487976,
      "learning_rate": 3.4009166708530596e-06,
      "loss": 0.3209,
      "step": 39979
    },
    {
      "epoch": 1.8347024000734247,
      "grad_norm": 0.4733816087245941,
      "learning_rate": 3.400684362273462e-06,
      "loss": 0.3789,
      "step": 39980
    },
    {
      "epoch": 1.8347482905786792,
      "grad_norm": 0.5549359321594238,
      "learning_rate": 3.4004520575397137e-06,
      "loss": 0.4606,
      "step": 39981
    },
    {
      "epoch": 1.8347941810839337,
      "grad_norm": 0.4625313878059387,
      "learning_rate": 3.4002197566523748e-06,
      "loss": 0.3515,
      "step": 39982
    },
    {
      "epoch": 1.8348400715891882,
      "grad_norm": 0.5009883046150208,
      "learning_rate": 3.399987459612002e-06,
      "loss": 0.3661,
      "step": 39983
    },
    {
      "epoch": 1.8348859620944427,
      "grad_norm": 0.43865346908569336,
      "learning_rate": 3.399755166419152e-06,
      "loss": 0.31,
      "step": 39984
    },
    {
      "epoch": 1.8349318525996972,
      "grad_norm": 0.4846648573875427,
      "learning_rate": 3.3995228770743882e-06,
      "loss": 0.369,
      "step": 39985
    },
    {
      "epoch": 1.8349777431049517,
      "grad_norm": 0.4774952530860901,
      "learning_rate": 3.399290591578267e-06,
      "loss": 0.4023,
      "step": 39986
    },
    {
      "epoch": 1.8350236336102062,
      "grad_norm": 0.48021814227104187,
      "learning_rate": 3.399058309931346e-06,
      "loss": 0.3413,
      "step": 39987
    },
    {
      "epoch": 1.8350695241154606,
      "grad_norm": 0.48949164152145386,
      "learning_rate": 3.398826032134185e-06,
      "loss": 0.3826,
      "step": 39988
    },
    {
      "epoch": 1.835115414620715,
      "grad_norm": 0.44999539852142334,
      "learning_rate": 3.3985937581873416e-06,
      "loss": 0.3031,
      "step": 39989
    },
    {
      "epoch": 1.8351613051259694,
      "grad_norm": 0.4809440076351166,
      "learning_rate": 3.398361488091373e-06,
      "loss": 0.3567,
      "step": 39990
    },
    {
      "epoch": 1.835207195631224,
      "grad_norm": 0.49652642011642456,
      "learning_rate": 3.398129221846842e-06,
      "loss": 0.3798,
      "step": 39991
    },
    {
      "epoch": 1.8352530861364784,
      "grad_norm": 0.5037232041358948,
      "learning_rate": 3.397896959454304e-06,
      "loss": 0.398,
      "step": 39992
    },
    {
      "epoch": 1.8352989766417327,
      "grad_norm": 0.46513646841049194,
      "learning_rate": 3.397664700914318e-06,
      "loss": 0.3739,
      "step": 39993
    },
    {
      "epoch": 1.8353448671469872,
      "grad_norm": 0.4537631869316101,
      "learning_rate": 3.397432446227443e-06,
      "loss": 0.383,
      "step": 39994
    },
    {
      "epoch": 1.8353907576522417,
      "grad_norm": 0.4685698449611664,
      "learning_rate": 3.3972001953942373e-06,
      "loss": 0.3475,
      "step": 39995
    },
    {
      "epoch": 1.8354366481574962,
      "grad_norm": 0.5096879601478577,
      "learning_rate": 3.3969679484152575e-06,
      "loss": 0.3597,
      "step": 39996
    },
    {
      "epoch": 1.8354825386627507,
      "grad_norm": 0.48225119709968567,
      "learning_rate": 3.3967357052910653e-06,
      "loss": 0.4235,
      "step": 39997
    },
    {
      "epoch": 1.8355284291680052,
      "grad_norm": 0.4618787467479706,
      "learning_rate": 3.3965034660222186e-06,
      "loss": 0.354,
      "step": 39998
    },
    {
      "epoch": 1.8355743196732597,
      "grad_norm": 0.4447525143623352,
      "learning_rate": 3.396271230609274e-06,
      "loss": 0.3368,
      "step": 39999
    },
    {
      "epoch": 1.8356202101785142,
      "grad_norm": 0.5123956203460693,
      "learning_rate": 3.3960389990527908e-06,
      "loss": 0.3257,
      "step": 40000
    },
    {
      "epoch": 1.8356661006837687,
      "grad_norm": 0.4180506765842438,
      "learning_rate": 3.395806771353328e-06,
      "loss": 0.2989,
      "step": 40001
    },
    {
      "epoch": 1.835711991189023,
      "grad_norm": 0.4920891523361206,
      "learning_rate": 3.395574547511442e-06,
      "loss": 0.4188,
      "step": 40002
    },
    {
      "epoch": 1.8357578816942774,
      "grad_norm": 0.48719072341918945,
      "learning_rate": 3.3953423275276963e-06,
      "loss": 0.381,
      "step": 40003
    },
    {
      "epoch": 1.835803772199532,
      "grad_norm": 0.44342103600502014,
      "learning_rate": 3.395110111402644e-06,
      "loss": 0.2798,
      "step": 40004
    },
    {
      "epoch": 1.8358496627047862,
      "grad_norm": 0.4603930711746216,
      "learning_rate": 3.3948778991368435e-06,
      "loss": 0.3287,
      "step": 40005
    },
    {
      "epoch": 1.8358955532100407,
      "grad_norm": 0.5122914910316467,
      "learning_rate": 3.3946456907308577e-06,
      "loss": 0.4086,
      "step": 40006
    },
    {
      "epoch": 1.8359414437152952,
      "grad_norm": 0.4727425277233124,
      "learning_rate": 3.394413486185242e-06,
      "loss": 0.3528,
      "step": 40007
    },
    {
      "epoch": 1.8359873342205497,
      "grad_norm": 0.44000017642974854,
      "learning_rate": 3.394181285500554e-06,
      "loss": 0.3033,
      "step": 40008
    },
    {
      "epoch": 1.8360332247258042,
      "grad_norm": 0.4536668360233307,
      "learning_rate": 3.393949088677354e-06,
      "loss": 0.3525,
      "step": 40009
    },
    {
      "epoch": 1.8360791152310587,
      "grad_norm": 0.4706602990627289,
      "learning_rate": 3.3937168957161996e-06,
      "loss": 0.3976,
      "step": 40010
    },
    {
      "epoch": 1.8361250057363132,
      "grad_norm": 0.48928114771842957,
      "learning_rate": 3.3934847066176475e-06,
      "loss": 0.3642,
      "step": 40011
    },
    {
      "epoch": 1.8361708962415677,
      "grad_norm": 0.44901615381240845,
      "learning_rate": 3.3932525213822594e-06,
      "loss": 0.2943,
      "step": 40012
    },
    {
      "epoch": 1.8362167867468222,
      "grad_norm": 0.47857481241226196,
      "learning_rate": 3.393020340010592e-06,
      "loss": 0.348,
      "step": 40013
    },
    {
      "epoch": 1.8362626772520767,
      "grad_norm": 0.4598817229270935,
      "learning_rate": 3.392788162503203e-06,
      "loss": 0.3114,
      "step": 40014
    },
    {
      "epoch": 1.836308567757331,
      "grad_norm": 0.5410502552986145,
      "learning_rate": 3.3925559888606506e-06,
      "loss": 0.3703,
      "step": 40015
    },
    {
      "epoch": 1.8363544582625855,
      "grad_norm": 0.5212858319282532,
      "learning_rate": 3.3923238190834952e-06,
      "loss": 0.4095,
      "step": 40016
    },
    {
      "epoch": 1.83640034876784,
      "grad_norm": 0.4732181131839752,
      "learning_rate": 3.392091653172293e-06,
      "loss": 0.3397,
      "step": 40017
    },
    {
      "epoch": 1.8364462392730942,
      "grad_norm": 0.40973031520843506,
      "learning_rate": 3.391859491127601e-06,
      "loss": 0.269,
      "step": 40018
    },
    {
      "epoch": 1.8364921297783487,
      "grad_norm": 0.5094745755195618,
      "learning_rate": 3.3916273329499816e-06,
      "loss": 0.4242,
      "step": 40019
    },
    {
      "epoch": 1.8365380202836032,
      "grad_norm": 0.4824801981449127,
      "learning_rate": 3.3913951786399905e-06,
      "loss": 0.3583,
      "step": 40020
    },
    {
      "epoch": 1.8365839107888577,
      "grad_norm": 0.47572124004364014,
      "learning_rate": 3.3911630281981853e-06,
      "loss": 0.3883,
      "step": 40021
    },
    {
      "epoch": 1.8366298012941122,
      "grad_norm": 0.43510639667510986,
      "learning_rate": 3.390930881625126e-06,
      "loss": 0.3041,
      "step": 40022
    },
    {
      "epoch": 1.8366756917993667,
      "grad_norm": 0.466013640165329,
      "learning_rate": 3.3906987389213714e-06,
      "loss": 0.3411,
      "step": 40023
    },
    {
      "epoch": 1.8367215823046212,
      "grad_norm": 0.4450278580188751,
      "learning_rate": 3.390466600087475e-06,
      "loss": 0.2795,
      "step": 40024
    },
    {
      "epoch": 1.8367674728098757,
      "grad_norm": 0.4858452081680298,
      "learning_rate": 3.390234465124003e-06,
      "loss": 0.3821,
      "step": 40025
    },
    {
      "epoch": 1.8368133633151302,
      "grad_norm": 0.4891620874404907,
      "learning_rate": 3.390002334031507e-06,
      "loss": 0.3647,
      "step": 40026
    },
    {
      "epoch": 1.8368592538203845,
      "grad_norm": 0.5070965886116028,
      "learning_rate": 3.3897702068105447e-06,
      "loss": 0.337,
      "step": 40027
    },
    {
      "epoch": 1.836905144325639,
      "grad_norm": 0.4650188982486725,
      "learning_rate": 3.3895380834616796e-06,
      "loss": 0.384,
      "step": 40028
    },
    {
      "epoch": 1.8369510348308935,
      "grad_norm": 0.44363224506378174,
      "learning_rate": 3.389305963985467e-06,
      "loss": 0.3625,
      "step": 40029
    },
    {
      "epoch": 1.836996925336148,
      "grad_norm": 0.4735911786556244,
      "learning_rate": 3.389073848382464e-06,
      "loss": 0.4323,
      "step": 40030
    },
    {
      "epoch": 1.8370428158414023,
      "grad_norm": 0.4961511194705963,
      "learning_rate": 3.3888417366532312e-06,
      "loss": 0.3957,
      "step": 40031
    },
    {
      "epoch": 1.8370887063466568,
      "grad_norm": 0.45797258615493774,
      "learning_rate": 3.3886096287983254e-06,
      "loss": 0.3411,
      "step": 40032
    },
    {
      "epoch": 1.8371345968519113,
      "grad_norm": 0.4434041380882263,
      "learning_rate": 3.388377524818303e-06,
      "loss": 0.3089,
      "step": 40033
    },
    {
      "epoch": 1.8371804873571658,
      "grad_norm": 0.5157313346862793,
      "learning_rate": 3.3881454247137263e-06,
      "loss": 0.4628,
      "step": 40034
    },
    {
      "epoch": 1.8372263778624203,
      "grad_norm": 0.44423922896385193,
      "learning_rate": 3.3879133284851507e-06,
      "loss": 0.2925,
      "step": 40035
    },
    {
      "epoch": 1.8372722683676748,
      "grad_norm": 0.503124475479126,
      "learning_rate": 3.387681236133134e-06,
      "loss": 0.4036,
      "step": 40036
    },
    {
      "epoch": 1.8373181588729293,
      "grad_norm": 0.4812272787094116,
      "learning_rate": 3.387449147658236e-06,
      "loss": 0.3848,
      "step": 40037
    },
    {
      "epoch": 1.8373640493781838,
      "grad_norm": 0.4713943302631378,
      "learning_rate": 3.3872170630610136e-06,
      "loss": 0.3592,
      "step": 40038
    },
    {
      "epoch": 1.8374099398834383,
      "grad_norm": 0.4816942811012268,
      "learning_rate": 3.3869849823420233e-06,
      "loss": 0.3589,
      "step": 40039
    },
    {
      "epoch": 1.8374558303886925,
      "grad_norm": 0.46792346239089966,
      "learning_rate": 3.3867529055018273e-06,
      "loss": 0.3496,
      "step": 40040
    },
    {
      "epoch": 1.837501720893947,
      "grad_norm": 0.4468141198158264,
      "learning_rate": 3.3865208325409816e-06,
      "loss": 0.3265,
      "step": 40041
    },
    {
      "epoch": 1.8375476113992015,
      "grad_norm": 0.5125146508216858,
      "learning_rate": 3.3862887634600423e-06,
      "loss": 0.4246,
      "step": 40042
    },
    {
      "epoch": 1.8375935019044558,
      "grad_norm": 0.4288843870162964,
      "learning_rate": 3.3860566982595706e-06,
      "loss": 0.2954,
      "step": 40043
    },
    {
      "epoch": 1.8376393924097103,
      "grad_norm": 0.4696047008037567,
      "learning_rate": 3.3858246369401234e-06,
      "loss": 0.3496,
      "step": 40044
    },
    {
      "epoch": 1.8376852829149648,
      "grad_norm": 0.5032174587249756,
      "learning_rate": 3.385592579502258e-06,
      "loss": 0.4147,
      "step": 40045
    },
    {
      "epoch": 1.8377311734202193,
      "grad_norm": 0.4543551802635193,
      "learning_rate": 3.385360525946531e-06,
      "loss": 0.3246,
      "step": 40046
    },
    {
      "epoch": 1.8377770639254738,
      "grad_norm": 0.5061081647872925,
      "learning_rate": 3.385128476273506e-06,
      "loss": 0.4123,
      "step": 40047
    },
    {
      "epoch": 1.8378229544307283,
      "grad_norm": 0.5673410296440125,
      "learning_rate": 3.384896430483736e-06,
      "loss": 0.3442,
      "step": 40048
    },
    {
      "epoch": 1.8378688449359828,
      "grad_norm": 0.48610663414001465,
      "learning_rate": 3.3846643885777776e-06,
      "loss": 0.3161,
      "step": 40049
    },
    {
      "epoch": 1.8379147354412373,
      "grad_norm": 0.48675185441970825,
      "learning_rate": 3.3844323505561944e-06,
      "loss": 0.3947,
      "step": 40050
    },
    {
      "epoch": 1.8379606259464918,
      "grad_norm": 0.4737752676010132,
      "learning_rate": 3.3842003164195414e-06,
      "loss": 0.3529,
      "step": 40051
    },
    {
      "epoch": 1.8380065164517463,
      "grad_norm": 0.48218345642089844,
      "learning_rate": 3.3839682861683754e-06,
      "loss": 0.3773,
      "step": 40052
    },
    {
      "epoch": 1.8380524069570006,
      "grad_norm": 0.49974116683006287,
      "learning_rate": 3.3837362598032564e-06,
      "loss": 0.3468,
      "step": 40053
    },
    {
      "epoch": 1.838098297462255,
      "grad_norm": 0.47853338718414307,
      "learning_rate": 3.383504237324742e-06,
      "loss": 0.3706,
      "step": 40054
    },
    {
      "epoch": 1.8381441879675096,
      "grad_norm": 0.4632374942302704,
      "learning_rate": 3.383272218733388e-06,
      "loss": 0.2881,
      "step": 40055
    },
    {
      "epoch": 1.8381900784727638,
      "grad_norm": 0.45719119906425476,
      "learning_rate": 3.383040204029756e-06,
      "loss": 0.377,
      "step": 40056
    },
    {
      "epoch": 1.8382359689780183,
      "grad_norm": 0.4733566641807556,
      "learning_rate": 3.382808193214402e-06,
      "loss": 0.3559,
      "step": 40057
    },
    {
      "epoch": 1.8382818594832728,
      "grad_norm": 0.48723509907722473,
      "learning_rate": 3.3825761862878826e-06,
      "loss": 0.3596,
      "step": 40058
    },
    {
      "epoch": 1.8383277499885273,
      "grad_norm": 0.45164161920547485,
      "learning_rate": 3.3823441832507574e-06,
      "loss": 0.3423,
      "step": 40059
    },
    {
      "epoch": 1.8383736404937818,
      "grad_norm": 0.5071130394935608,
      "learning_rate": 3.382112184103585e-06,
      "loss": 0.36,
      "step": 40060
    },
    {
      "epoch": 1.8384195309990363,
      "grad_norm": 0.5026705861091614,
      "learning_rate": 3.3818801888469193e-06,
      "loss": 0.3787,
      "step": 40061
    },
    {
      "epoch": 1.8384654215042908,
      "grad_norm": 0.4690433740615845,
      "learning_rate": 3.3816481974813233e-06,
      "loss": 0.3274,
      "step": 40062
    },
    {
      "epoch": 1.8385113120095453,
      "grad_norm": 0.45508500933647156,
      "learning_rate": 3.381416210007353e-06,
      "loss": 0.3035,
      "step": 40063
    },
    {
      "epoch": 1.8385572025147998,
      "grad_norm": 0.47315606474876404,
      "learning_rate": 3.3811842264255647e-06,
      "loss": 0.3813,
      "step": 40064
    },
    {
      "epoch": 1.8386030930200543,
      "grad_norm": 0.4633285701274872,
      "learning_rate": 3.3809522467365185e-06,
      "loss": 0.3739,
      "step": 40065
    },
    {
      "epoch": 1.8386489835253086,
      "grad_norm": 0.47519776225090027,
      "learning_rate": 3.3807202709407704e-06,
      "loss": 0.3888,
      "step": 40066
    },
    {
      "epoch": 1.838694874030563,
      "grad_norm": 0.46747300028800964,
      "learning_rate": 3.3804882990388776e-06,
      "loss": 0.3697,
      "step": 40067
    },
    {
      "epoch": 1.8387407645358176,
      "grad_norm": 0.4418640732765198,
      "learning_rate": 3.3802563310314018e-06,
      "loss": 0.3264,
      "step": 40068
    },
    {
      "epoch": 1.8387866550410719,
      "grad_norm": 0.4547782838344574,
      "learning_rate": 3.3800243669188994e-06,
      "loss": 0.3229,
      "step": 40069
    },
    {
      "epoch": 1.8388325455463264,
      "grad_norm": 0.4975554049015045,
      "learning_rate": 3.3797924067019227e-06,
      "loss": 0.4063,
      "step": 40070
    },
    {
      "epoch": 1.8388784360515809,
      "grad_norm": 0.4877837896347046,
      "learning_rate": 3.379560450381037e-06,
      "loss": 0.3302,
      "step": 40071
    },
    {
      "epoch": 1.8389243265568354,
      "grad_norm": 0.44827812910079956,
      "learning_rate": 3.379328497956797e-06,
      "loss": 0.3152,
      "step": 40072
    },
    {
      "epoch": 1.8389702170620899,
      "grad_norm": 0.4726530611515045,
      "learning_rate": 3.379096549429759e-06,
      "loss": 0.371,
      "step": 40073
    },
    {
      "epoch": 1.8390161075673443,
      "grad_norm": 0.4561325013637543,
      "learning_rate": 3.3788646048004837e-06,
      "loss": 0.3226,
      "step": 40074
    },
    {
      "epoch": 1.8390619980725988,
      "grad_norm": 0.4795105755329132,
      "learning_rate": 3.3786326640695266e-06,
      "loss": 0.3675,
      "step": 40075
    },
    {
      "epoch": 1.8391078885778533,
      "grad_norm": 0.45084723830223083,
      "learning_rate": 3.3784007272374453e-06,
      "loss": 0.3233,
      "step": 40076
    },
    {
      "epoch": 1.8391537790831078,
      "grad_norm": 0.5003656148910522,
      "learning_rate": 3.3781687943047995e-06,
      "loss": 0.4007,
      "step": 40077
    },
    {
      "epoch": 1.8391996695883621,
      "grad_norm": 0.437844455242157,
      "learning_rate": 3.3779368652721467e-06,
      "loss": 0.3367,
      "step": 40078
    },
    {
      "epoch": 1.8392455600936166,
      "grad_norm": 0.4764179587364197,
      "learning_rate": 3.377704940140043e-06,
      "loss": 0.373,
      "step": 40079
    },
    {
      "epoch": 1.8392914505988711,
      "grad_norm": 0.42635950446128845,
      "learning_rate": 3.3774730189090464e-06,
      "loss": 0.3024,
      "step": 40080
    },
    {
      "epoch": 1.8393373411041256,
      "grad_norm": 0.49560028314590454,
      "learning_rate": 3.3772411015797165e-06,
      "loss": 0.324,
      "step": 40081
    },
    {
      "epoch": 1.8393832316093799,
      "grad_norm": 0.4866359233856201,
      "learning_rate": 3.377009188152609e-06,
      "loss": 0.4122,
      "step": 40082
    },
    {
      "epoch": 1.8394291221146344,
      "grad_norm": 0.44516223669052124,
      "learning_rate": 3.37677727862828e-06,
      "loss": 0.3139,
      "step": 40083
    },
    {
      "epoch": 1.8394750126198889,
      "grad_norm": 0.43794363737106323,
      "learning_rate": 3.3765453730072916e-06,
      "loss": 0.3193,
      "step": 40084
    },
    {
      "epoch": 1.8395209031251434,
      "grad_norm": 0.4666297733783722,
      "learning_rate": 3.376313471290199e-06,
      "loss": 0.3556,
      "step": 40085
    },
    {
      "epoch": 1.8395667936303979,
      "grad_norm": 0.4612573981285095,
      "learning_rate": 3.37608157347756e-06,
      "loss": 0.3044,
      "step": 40086
    },
    {
      "epoch": 1.8396126841356524,
      "grad_norm": 0.47304508090019226,
      "learning_rate": 3.3758496795699323e-06,
      "loss": 0.3371,
      "step": 40087
    },
    {
      "epoch": 1.8396585746409069,
      "grad_norm": 0.482916921377182,
      "learning_rate": 3.3756177895678733e-06,
      "loss": 0.3614,
      "step": 40088
    },
    {
      "epoch": 1.8397044651461614,
      "grad_norm": 0.4894770383834839,
      "learning_rate": 3.3753859034719393e-06,
      "loss": 0.4044,
      "step": 40089
    },
    {
      "epoch": 1.8397503556514159,
      "grad_norm": 0.44892576336860657,
      "learning_rate": 3.3751540212826916e-06,
      "loss": 0.3172,
      "step": 40090
    },
    {
      "epoch": 1.8397962461566701,
      "grad_norm": 0.49046286940574646,
      "learning_rate": 3.3749221430006866e-06,
      "loss": 0.4191,
      "step": 40091
    },
    {
      "epoch": 1.8398421366619246,
      "grad_norm": 0.46104514598846436,
      "learning_rate": 3.374690268626478e-06,
      "loss": 0.3533,
      "step": 40092
    },
    {
      "epoch": 1.8398880271671791,
      "grad_norm": 0.45382875204086304,
      "learning_rate": 3.374458398160628e-06,
      "loss": 0.3229,
      "step": 40093
    },
    {
      "epoch": 1.8399339176724334,
      "grad_norm": 0.47095879912376404,
      "learning_rate": 3.374226531603693e-06,
      "loss": 0.3532,
      "step": 40094
    },
    {
      "epoch": 1.839979808177688,
      "grad_norm": 0.4647656977176666,
      "learning_rate": 3.3739946689562286e-06,
      "loss": 0.3423,
      "step": 40095
    },
    {
      "epoch": 1.8400256986829424,
      "grad_norm": 0.49705958366394043,
      "learning_rate": 3.373762810218796e-06,
      "loss": 0.4279,
      "step": 40096
    },
    {
      "epoch": 1.840071589188197,
      "grad_norm": 0.47082722187042236,
      "learning_rate": 3.3735309553919494e-06,
      "loss": 0.3739,
      "step": 40097
    },
    {
      "epoch": 1.8401174796934514,
      "grad_norm": 0.4391995966434479,
      "learning_rate": 3.3732991044762452e-06,
      "loss": 0.3108,
      "step": 40098
    },
    {
      "epoch": 1.840163370198706,
      "grad_norm": 0.5253812074661255,
      "learning_rate": 3.3730672574722463e-06,
      "loss": 0.5041,
      "step": 40099
    },
    {
      "epoch": 1.8402092607039604,
      "grad_norm": 0.45840689539909363,
      "learning_rate": 3.3728354143805063e-06,
      "loss": 0.3031,
      "step": 40100
    },
    {
      "epoch": 1.840255151209215,
      "grad_norm": 0.4417295455932617,
      "learning_rate": 3.3726035752015833e-06,
      "loss": 0.3125,
      "step": 40101
    },
    {
      "epoch": 1.8403010417144694,
      "grad_norm": 0.44637274742126465,
      "learning_rate": 3.3723717399360355e-06,
      "loss": 0.3617,
      "step": 40102
    },
    {
      "epoch": 1.840346932219724,
      "grad_norm": 0.48672690987586975,
      "learning_rate": 3.3721399085844208e-06,
      "loss": 0.3894,
      "step": 40103
    },
    {
      "epoch": 1.8403928227249782,
      "grad_norm": 0.47997018694877625,
      "learning_rate": 3.3719080811472934e-06,
      "loss": 0.3617,
      "step": 40104
    },
    {
      "epoch": 1.8404387132302327,
      "grad_norm": 0.5309159755706787,
      "learning_rate": 3.371676257625215e-06,
      "loss": 0.4415,
      "step": 40105
    },
    {
      "epoch": 1.8404846037354872,
      "grad_norm": 0.4676118493080139,
      "learning_rate": 3.3714444380187415e-06,
      "loss": 0.3464,
      "step": 40106
    },
    {
      "epoch": 1.8405304942407414,
      "grad_norm": 0.4376239478588104,
      "learning_rate": 3.3712126223284292e-06,
      "loss": 0.3585,
      "step": 40107
    },
    {
      "epoch": 1.840576384745996,
      "grad_norm": 0.5081982016563416,
      "learning_rate": 3.370980810554837e-06,
      "loss": 0.4088,
      "step": 40108
    },
    {
      "epoch": 1.8406222752512504,
      "grad_norm": 0.433517187833786,
      "learning_rate": 3.3707490026985225e-06,
      "loss": 0.287,
      "step": 40109
    },
    {
      "epoch": 1.840668165756505,
      "grad_norm": 0.4637381136417389,
      "learning_rate": 3.3705171987600394e-06,
      "loss": 0.3651,
      "step": 40110
    },
    {
      "epoch": 1.8407140562617594,
      "grad_norm": 0.42848601937294006,
      "learning_rate": 3.370285398739951e-06,
      "loss": 0.2665,
      "step": 40111
    },
    {
      "epoch": 1.840759946767014,
      "grad_norm": 0.4670964777469635,
      "learning_rate": 3.370053602638812e-06,
      "loss": 0.3781,
      "step": 40112
    },
    {
      "epoch": 1.8408058372722684,
      "grad_norm": 0.48157966136932373,
      "learning_rate": 3.36982181045718e-06,
      "loss": 0.3828,
      "step": 40113
    },
    {
      "epoch": 1.840851727777523,
      "grad_norm": 0.4564974904060364,
      "learning_rate": 3.3695900221956103e-06,
      "loss": 0.3036,
      "step": 40114
    },
    {
      "epoch": 1.8408976182827774,
      "grad_norm": 0.4663805663585663,
      "learning_rate": 3.369358237854663e-06,
      "loss": 0.3081,
      "step": 40115
    },
    {
      "epoch": 1.8409435087880317,
      "grad_norm": 0.45845264196395874,
      "learning_rate": 3.369126457434895e-06,
      "loss": 0.375,
      "step": 40116
    },
    {
      "epoch": 1.8409893992932862,
      "grad_norm": 0.5117235779762268,
      "learning_rate": 3.368894680936861e-06,
      "loss": 0.4131,
      "step": 40117
    },
    {
      "epoch": 1.8410352897985407,
      "grad_norm": 0.48235753178596497,
      "learning_rate": 3.368662908361122e-06,
      "loss": 0.3301,
      "step": 40118
    },
    {
      "epoch": 1.8410811803037952,
      "grad_norm": 0.4922463893890381,
      "learning_rate": 3.368431139708235e-06,
      "loss": 0.4437,
      "step": 40119
    },
    {
      "epoch": 1.8411270708090495,
      "grad_norm": 0.494913250207901,
      "learning_rate": 3.3681993749787528e-06,
      "loss": 0.3971,
      "step": 40120
    },
    {
      "epoch": 1.841172961314304,
      "grad_norm": 0.4681732654571533,
      "learning_rate": 3.367967614173239e-06,
      "loss": 0.3196,
      "step": 40121
    },
    {
      "epoch": 1.8412188518195585,
      "grad_norm": 0.4512799382209778,
      "learning_rate": 3.367735857292247e-06,
      "loss": 0.3567,
      "step": 40122
    },
    {
      "epoch": 1.841264742324813,
      "grad_norm": 0.46655550599098206,
      "learning_rate": 3.3675041043363348e-06,
      "loss": 0.321,
      "step": 40123
    },
    {
      "epoch": 1.8413106328300675,
      "grad_norm": 0.4183458685874939,
      "learning_rate": 3.3672723553060615e-06,
      "loss": 0.2795,
      "step": 40124
    },
    {
      "epoch": 1.841356523335322,
      "grad_norm": 0.4655725657939911,
      "learning_rate": 3.367040610201982e-06,
      "loss": 0.3598,
      "step": 40125
    },
    {
      "epoch": 1.8414024138405765,
      "grad_norm": 0.47940540313720703,
      "learning_rate": 3.3668088690246525e-06,
      "loss": 0.3774,
      "step": 40126
    },
    {
      "epoch": 1.841448304345831,
      "grad_norm": 0.4671495258808136,
      "learning_rate": 3.3665771317746343e-06,
      "loss": 0.3299,
      "step": 40127
    },
    {
      "epoch": 1.8414941948510855,
      "grad_norm": 0.4431859850883484,
      "learning_rate": 3.366345398452483e-06,
      "loss": 0.3199,
      "step": 40128
    },
    {
      "epoch": 1.8415400853563397,
      "grad_norm": 0.49852409958839417,
      "learning_rate": 3.366113669058754e-06,
      "loss": 0.3868,
      "step": 40129
    },
    {
      "epoch": 1.8415859758615942,
      "grad_norm": 0.4486108124256134,
      "learning_rate": 3.365881943594007e-06,
      "loss": 0.3232,
      "step": 40130
    },
    {
      "epoch": 1.8416318663668487,
      "grad_norm": 0.4684428870677948,
      "learning_rate": 3.3656502220587988e-06,
      "loss": 0.3312,
      "step": 40131
    },
    {
      "epoch": 1.841677756872103,
      "grad_norm": 0.5015115141868591,
      "learning_rate": 3.3654185044536838e-06,
      "loss": 0.4206,
      "step": 40132
    },
    {
      "epoch": 1.8417236473773575,
      "grad_norm": 0.4728427827358246,
      "learning_rate": 3.3651867907792226e-06,
      "loss": 0.3735,
      "step": 40133
    },
    {
      "epoch": 1.841769537882612,
      "grad_norm": 0.4692358076572418,
      "learning_rate": 3.364955081035972e-06,
      "loss": 0.3534,
      "step": 40134
    },
    {
      "epoch": 1.8418154283878665,
      "grad_norm": 0.5298478603363037,
      "learning_rate": 3.364723375224487e-06,
      "loss": 0.4887,
      "step": 40135
    },
    {
      "epoch": 1.841861318893121,
      "grad_norm": 0.4001864790916443,
      "learning_rate": 3.364491673345327e-06,
      "loss": 0.2354,
      "step": 40136
    },
    {
      "epoch": 1.8419072093983755,
      "grad_norm": 0.4565076231956482,
      "learning_rate": 3.364259975399049e-06,
      "loss": 0.3357,
      "step": 40137
    },
    {
      "epoch": 1.84195309990363,
      "grad_norm": 0.47771820425987244,
      "learning_rate": 3.364028281386208e-06,
      "loss": 0.3482,
      "step": 40138
    },
    {
      "epoch": 1.8419989904088845,
      "grad_norm": 0.43700429797172546,
      "learning_rate": 3.363796591307364e-06,
      "loss": 0.3108,
      "step": 40139
    },
    {
      "epoch": 1.842044880914139,
      "grad_norm": 0.46043312549591064,
      "learning_rate": 3.3635649051630724e-06,
      "loss": 0.3275,
      "step": 40140
    },
    {
      "epoch": 1.8420907714193935,
      "grad_norm": 0.49332576990127563,
      "learning_rate": 3.3633332229538888e-06,
      "loss": 0.3946,
      "step": 40141
    },
    {
      "epoch": 1.8421366619246478,
      "grad_norm": 0.48606953024864197,
      "learning_rate": 3.3631015446803743e-06,
      "loss": 0.3843,
      "step": 40142
    },
    {
      "epoch": 1.8421825524299023,
      "grad_norm": 0.46592041850090027,
      "learning_rate": 3.362869870343084e-06,
      "loss": 0.3852,
      "step": 40143
    },
    {
      "epoch": 1.8422284429351568,
      "grad_norm": 0.46210891008377075,
      "learning_rate": 3.362638199942575e-06,
      "loss": 0.3676,
      "step": 40144
    },
    {
      "epoch": 1.842274333440411,
      "grad_norm": 0.4840258061885834,
      "learning_rate": 3.3624065334794038e-06,
      "loss": 0.3515,
      "step": 40145
    },
    {
      "epoch": 1.8423202239456655,
      "grad_norm": 0.47677192091941833,
      "learning_rate": 3.3621748709541287e-06,
      "loss": 0.3967,
      "step": 40146
    },
    {
      "epoch": 1.84236611445092,
      "grad_norm": 0.4676186442375183,
      "learning_rate": 3.3619432123673057e-06,
      "loss": 0.338,
      "step": 40147
    },
    {
      "epoch": 1.8424120049561745,
      "grad_norm": 0.5050968527793884,
      "learning_rate": 3.3617115577194904e-06,
      "loss": 0.3246,
      "step": 40148
    },
    {
      "epoch": 1.842457895461429,
      "grad_norm": 0.4328094720840454,
      "learning_rate": 3.3614799070112448e-06,
      "loss": 0.3241,
      "step": 40149
    },
    {
      "epoch": 1.8425037859666835,
      "grad_norm": 0.4765894114971161,
      "learning_rate": 3.3612482602431216e-06,
      "loss": 0.4017,
      "step": 40150
    },
    {
      "epoch": 1.842549676471938,
      "grad_norm": 0.4436030983924866,
      "learning_rate": 3.361016617415679e-06,
      "loss": 0.322,
      "step": 40151
    },
    {
      "epoch": 1.8425955669771925,
      "grad_norm": 0.43980351090431213,
      "learning_rate": 3.3607849785294744e-06,
      "loss": 0.2932,
      "step": 40152
    },
    {
      "epoch": 1.842641457482447,
      "grad_norm": 0.47944140434265137,
      "learning_rate": 3.3605533435850646e-06,
      "loss": 0.3966,
      "step": 40153
    },
    {
      "epoch": 1.8426873479877015,
      "grad_norm": 0.46652156114578247,
      "learning_rate": 3.360321712583005e-06,
      "loss": 0.3319,
      "step": 40154
    },
    {
      "epoch": 1.8427332384929558,
      "grad_norm": 0.44899260997772217,
      "learning_rate": 3.360090085523856e-06,
      "loss": 0.3314,
      "step": 40155
    },
    {
      "epoch": 1.8427791289982103,
      "grad_norm": 0.4919794797897339,
      "learning_rate": 3.359858462408173e-06,
      "loss": 0.3534,
      "step": 40156
    },
    {
      "epoch": 1.8428250195034648,
      "grad_norm": 0.42925024032592773,
      "learning_rate": 3.359626843236512e-06,
      "loss": 0.2904,
      "step": 40157
    },
    {
      "epoch": 1.842870910008719,
      "grad_norm": 0.6350916624069214,
      "learning_rate": 3.359395228009431e-06,
      "loss": 0.3624,
      "step": 40158
    },
    {
      "epoch": 1.8429168005139736,
      "grad_norm": 0.48383086919784546,
      "learning_rate": 3.359163616727487e-06,
      "loss": 0.3685,
      "step": 40159
    },
    {
      "epoch": 1.842962691019228,
      "grad_norm": 0.5049861073493958,
      "learning_rate": 3.3589320093912357e-06,
      "loss": 0.4063,
      "step": 40160
    },
    {
      "epoch": 1.8430085815244825,
      "grad_norm": 0.44479402899742126,
      "learning_rate": 3.3587004060012364e-06,
      "loss": 0.2988,
      "step": 40161
    },
    {
      "epoch": 1.843054472029737,
      "grad_norm": 0.4592954218387604,
      "learning_rate": 3.3584688065580438e-06,
      "loss": 0.2894,
      "step": 40162
    },
    {
      "epoch": 1.8431003625349915,
      "grad_norm": 0.4851171672344208,
      "learning_rate": 3.358237211062214e-06,
      "loss": 0.3329,
      "step": 40163
    },
    {
      "epoch": 1.843146253040246,
      "grad_norm": 0.504614531993866,
      "learning_rate": 3.3580056195143074e-06,
      "loss": 0.3513,
      "step": 40164
    },
    {
      "epoch": 1.8431921435455005,
      "grad_norm": 0.4932461977005005,
      "learning_rate": 3.3577740319148792e-06,
      "loss": 0.3722,
      "step": 40165
    },
    {
      "epoch": 1.843238034050755,
      "grad_norm": 0.4877512454986572,
      "learning_rate": 3.357542448264486e-06,
      "loss": 0.3156,
      "step": 40166
    },
    {
      "epoch": 1.8432839245560093,
      "grad_norm": 0.4547358751296997,
      "learning_rate": 3.3573108685636844e-06,
      "loss": 0.3297,
      "step": 40167
    },
    {
      "epoch": 1.8433298150612638,
      "grad_norm": 0.4972550868988037,
      "learning_rate": 3.3570792928130325e-06,
      "loss": 0.3372,
      "step": 40168
    },
    {
      "epoch": 1.8433757055665183,
      "grad_norm": 0.4829994738101959,
      "learning_rate": 3.356847721013084e-06,
      "loss": 0.3804,
      "step": 40169
    },
    {
      "epoch": 1.8434215960717728,
      "grad_norm": 0.5092733502388,
      "learning_rate": 3.3566161531644004e-06,
      "loss": 0.2968,
      "step": 40170
    },
    {
      "epoch": 1.843467486577027,
      "grad_norm": 0.4555870592594147,
      "learning_rate": 3.3563845892675363e-06,
      "loss": 0.3331,
      "step": 40171
    },
    {
      "epoch": 1.8435133770822816,
      "grad_norm": 0.45496371388435364,
      "learning_rate": 3.3561530293230473e-06,
      "loss": 0.357,
      "step": 40172
    },
    {
      "epoch": 1.843559267587536,
      "grad_norm": 0.49132633209228516,
      "learning_rate": 3.3559214733314925e-06,
      "loss": 0.3814,
      "step": 40173
    },
    {
      "epoch": 1.8436051580927906,
      "grad_norm": 0.5186668634414673,
      "learning_rate": 3.355689921293428e-06,
      "loss": 0.3899,
      "step": 40174
    },
    {
      "epoch": 1.843651048598045,
      "grad_norm": 0.46897244453430176,
      "learning_rate": 3.355458373209408e-06,
      "loss": 0.3547,
      "step": 40175
    },
    {
      "epoch": 1.8436969391032996,
      "grad_norm": 0.4857618808746338,
      "learning_rate": 3.3552268290799937e-06,
      "loss": 0.4085,
      "step": 40176
    },
    {
      "epoch": 1.843742829608554,
      "grad_norm": 0.5028660893440247,
      "learning_rate": 3.35499528890574e-06,
      "loss": 0.3995,
      "step": 40177
    },
    {
      "epoch": 1.8437887201138086,
      "grad_norm": 0.5205161571502686,
      "learning_rate": 3.3547637526872028e-06,
      "loss": 0.4577,
      "step": 40178
    },
    {
      "epoch": 1.843834610619063,
      "grad_norm": 0.5011703968048096,
      "learning_rate": 3.3545322204249384e-06,
      "loss": 0.4123,
      "step": 40179
    },
    {
      "epoch": 1.8438805011243173,
      "grad_norm": 0.47178688645362854,
      "learning_rate": 3.3543006921195064e-06,
      "loss": 0.384,
      "step": 40180
    },
    {
      "epoch": 1.8439263916295718,
      "grad_norm": 0.4854075014591217,
      "learning_rate": 3.3540691677714622e-06,
      "loss": 0.4124,
      "step": 40181
    },
    {
      "epoch": 1.8439722821348263,
      "grad_norm": 0.4899073839187622,
      "learning_rate": 3.3538376473813603e-06,
      "loss": 0.3615,
      "step": 40182
    },
    {
      "epoch": 1.8440181726400806,
      "grad_norm": 0.4421781897544861,
      "learning_rate": 3.3536061309497602e-06,
      "loss": 0.3044,
      "step": 40183
    },
    {
      "epoch": 1.844064063145335,
      "grad_norm": 0.43765202164649963,
      "learning_rate": 3.3533746184772188e-06,
      "loss": 0.3155,
      "step": 40184
    },
    {
      "epoch": 1.8441099536505896,
      "grad_norm": 0.44694969058036804,
      "learning_rate": 3.3531431099642887e-06,
      "loss": 0.3042,
      "step": 40185
    },
    {
      "epoch": 1.844155844155844,
      "grad_norm": 0.4438680112361908,
      "learning_rate": 3.352911605411532e-06,
      "loss": 0.3261,
      "step": 40186
    },
    {
      "epoch": 1.8442017346610986,
      "grad_norm": 0.4938766062259674,
      "learning_rate": 3.3526801048195035e-06,
      "loss": 0.3868,
      "step": 40187
    },
    {
      "epoch": 1.844247625166353,
      "grad_norm": 0.48017898201942444,
      "learning_rate": 3.3524486081887575e-06,
      "loss": 0.3381,
      "step": 40188
    },
    {
      "epoch": 1.8442935156716076,
      "grad_norm": 0.42440158128738403,
      "learning_rate": 3.352217115519855e-06,
      "loss": 0.314,
      "step": 40189
    },
    {
      "epoch": 1.844339406176862,
      "grad_norm": 0.4698415696620941,
      "learning_rate": 3.351985626813349e-06,
      "loss": 0.3545,
      "step": 40190
    },
    {
      "epoch": 1.8443852966821166,
      "grad_norm": 0.436884343624115,
      "learning_rate": 3.3517541420697952e-06,
      "loss": 0.2957,
      "step": 40191
    },
    {
      "epoch": 1.844431187187371,
      "grad_norm": 0.46428993344306946,
      "learning_rate": 3.3515226612897555e-06,
      "loss": 0.3316,
      "step": 40192
    },
    {
      "epoch": 1.8444770776926254,
      "grad_norm": 0.45856600999832153,
      "learning_rate": 3.3512911844737834e-06,
      "loss": 0.3529,
      "step": 40193
    },
    {
      "epoch": 1.8445229681978799,
      "grad_norm": 0.4722120761871338,
      "learning_rate": 3.351059711622434e-06,
      "loss": 0.3573,
      "step": 40194
    },
    {
      "epoch": 1.8445688587031344,
      "grad_norm": 0.46723583340644836,
      "learning_rate": 3.3508282427362672e-06,
      "loss": 0.3437,
      "step": 40195
    },
    {
      "epoch": 1.8446147492083886,
      "grad_norm": 0.4561867415904999,
      "learning_rate": 3.350596777815838e-06,
      "loss": 0.3107,
      "step": 40196
    },
    {
      "epoch": 1.8446606397136431,
      "grad_norm": 0.4521401524543762,
      "learning_rate": 3.3503653168617007e-06,
      "loss": 0.2826,
      "step": 40197
    },
    {
      "epoch": 1.8447065302188976,
      "grad_norm": 0.5296445488929749,
      "learning_rate": 3.350133859874416e-06,
      "loss": 0.4357,
      "step": 40198
    },
    {
      "epoch": 1.8447524207241521,
      "grad_norm": 0.4156520366668701,
      "learning_rate": 3.3499024068545393e-06,
      "loss": 0.2797,
      "step": 40199
    },
    {
      "epoch": 1.8447983112294066,
      "grad_norm": 0.45170968770980835,
      "learning_rate": 3.349670957802625e-06,
      "loss": 0.332,
      "step": 40200
    },
    {
      "epoch": 1.8448442017346611,
      "grad_norm": 0.45238637924194336,
      "learning_rate": 3.3494395127192324e-06,
      "loss": 0.3183,
      "step": 40201
    },
    {
      "epoch": 1.8448900922399156,
      "grad_norm": 0.47506338357925415,
      "learning_rate": 3.349208071604917e-06,
      "loss": 0.3732,
      "step": 40202
    },
    {
      "epoch": 1.8449359827451701,
      "grad_norm": 0.4922519624233246,
      "learning_rate": 3.3489766344602326e-06,
      "loss": 0.3427,
      "step": 40203
    },
    {
      "epoch": 1.8449818732504246,
      "grad_norm": 0.44472962617874146,
      "learning_rate": 3.3487452012857412e-06,
      "loss": 0.2997,
      "step": 40204
    },
    {
      "epoch": 1.845027763755679,
      "grad_norm": 0.5015663504600525,
      "learning_rate": 3.348513772081996e-06,
      "loss": 0.3654,
      "step": 40205
    },
    {
      "epoch": 1.8450736542609334,
      "grad_norm": 0.4426925778388977,
      "learning_rate": 3.348282346849552e-06,
      "loss": 0.2898,
      "step": 40206
    },
    {
      "epoch": 1.845119544766188,
      "grad_norm": 0.4806230366230011,
      "learning_rate": 3.3480509255889695e-06,
      "loss": 0.3556,
      "step": 40207
    },
    {
      "epoch": 1.8451654352714424,
      "grad_norm": 0.5011109709739685,
      "learning_rate": 3.3478195083008034e-06,
      "loss": 0.3942,
      "step": 40208
    },
    {
      "epoch": 1.8452113257766967,
      "grad_norm": 0.46690890192985535,
      "learning_rate": 3.34758809498561e-06,
      "loss": 0.3474,
      "step": 40209
    },
    {
      "epoch": 1.8452572162819512,
      "grad_norm": 0.4850022494792938,
      "learning_rate": 3.3473566856439443e-06,
      "loss": 0.4063,
      "step": 40210
    },
    {
      "epoch": 1.8453031067872057,
      "grad_norm": 0.4774045944213867,
      "learning_rate": 3.3471252802763656e-06,
      "loss": 0.3288,
      "step": 40211
    },
    {
      "epoch": 1.8453489972924602,
      "grad_norm": 0.4652615785598755,
      "learning_rate": 3.3468938788834284e-06,
      "loss": 0.3268,
      "step": 40212
    },
    {
      "epoch": 1.8453948877977147,
      "grad_norm": 0.47561171650886536,
      "learning_rate": 3.346662481465688e-06,
      "loss": 0.4047,
      "step": 40213
    },
    {
      "epoch": 1.8454407783029692,
      "grad_norm": 0.5156642198562622,
      "learning_rate": 3.3464310880237038e-06,
      "loss": 0.4138,
      "step": 40214
    },
    {
      "epoch": 1.8454866688082237,
      "grad_norm": 0.5278937220573425,
      "learning_rate": 3.346199698558032e-06,
      "loss": 0.4586,
      "step": 40215
    },
    {
      "epoch": 1.8455325593134781,
      "grad_norm": 0.47812986373901367,
      "learning_rate": 3.3459683130692267e-06,
      "loss": 0.3479,
      "step": 40216
    },
    {
      "epoch": 1.8455784498187326,
      "grad_norm": 0.48146477341651917,
      "learning_rate": 3.345736931557846e-06,
      "loss": 0.3689,
      "step": 40217
    },
    {
      "epoch": 1.845624340323987,
      "grad_norm": 0.4844745695590973,
      "learning_rate": 3.3455055540244462e-06,
      "loss": 0.3814,
      "step": 40218
    },
    {
      "epoch": 1.8456702308292414,
      "grad_norm": 0.46681469678878784,
      "learning_rate": 3.3452741804695823e-06,
      "loss": 0.3221,
      "step": 40219
    },
    {
      "epoch": 1.845716121334496,
      "grad_norm": 0.46164578199386597,
      "learning_rate": 3.3450428108938127e-06,
      "loss": 0.2937,
      "step": 40220
    },
    {
      "epoch": 1.8457620118397502,
      "grad_norm": 0.5032753348350525,
      "learning_rate": 3.344811445297693e-06,
      "loss": 0.3943,
      "step": 40221
    },
    {
      "epoch": 1.8458079023450047,
      "grad_norm": 0.44146233797073364,
      "learning_rate": 3.344580083681778e-06,
      "loss": 0.3139,
      "step": 40222
    },
    {
      "epoch": 1.8458537928502592,
      "grad_norm": 0.47278133034706116,
      "learning_rate": 3.344348726046627e-06,
      "loss": 0.3575,
      "step": 40223
    },
    {
      "epoch": 1.8458996833555137,
      "grad_norm": 0.4297723174095154,
      "learning_rate": 3.3441173723927944e-06,
      "loss": 0.2986,
      "step": 40224
    },
    {
      "epoch": 1.8459455738607682,
      "grad_norm": 0.44753748178482056,
      "learning_rate": 3.343886022720836e-06,
      "loss": 0.3506,
      "step": 40225
    },
    {
      "epoch": 1.8459914643660227,
      "grad_norm": 0.4846854507923126,
      "learning_rate": 3.3436546770313107e-06,
      "loss": 0.3588,
      "step": 40226
    },
    {
      "epoch": 1.8460373548712772,
      "grad_norm": 0.4135468304157257,
      "learning_rate": 3.3434233353247724e-06,
      "loss": 0.2632,
      "step": 40227
    },
    {
      "epoch": 1.8460832453765317,
      "grad_norm": 0.4615086317062378,
      "learning_rate": 3.3431919976017768e-06,
      "loss": 0.3176,
      "step": 40228
    },
    {
      "epoch": 1.8461291358817862,
      "grad_norm": 0.4799922704696655,
      "learning_rate": 3.342960663862883e-06,
      "loss": 0.3589,
      "step": 40229
    },
    {
      "epoch": 1.8461750263870407,
      "grad_norm": 0.4557317793369293,
      "learning_rate": 3.342729334108646e-06,
      "loss": 0.3121,
      "step": 40230
    },
    {
      "epoch": 1.846220916892295,
      "grad_norm": 0.4909101724624634,
      "learning_rate": 3.342498008339621e-06,
      "loss": 0.3941,
      "step": 40231
    },
    {
      "epoch": 1.8462668073975494,
      "grad_norm": 0.46094390749931335,
      "learning_rate": 3.3422666865563658e-06,
      "loss": 0.3567,
      "step": 40232
    },
    {
      "epoch": 1.846312697902804,
      "grad_norm": 0.49138572812080383,
      "learning_rate": 3.3420353687594365e-06,
      "loss": 0.4046,
      "step": 40233
    },
    {
      "epoch": 1.8463585884080582,
      "grad_norm": 0.49135449528694153,
      "learning_rate": 3.3418040549493863e-06,
      "loss": 0.3534,
      "step": 40234
    },
    {
      "epoch": 1.8464044789133127,
      "grad_norm": 0.5155145525932312,
      "learning_rate": 3.3415727451267767e-06,
      "loss": 0.4434,
      "step": 40235
    },
    {
      "epoch": 1.8464503694185672,
      "grad_norm": 0.43690621852874756,
      "learning_rate": 3.3413414392921605e-06,
      "loss": 0.2789,
      "step": 40236
    },
    {
      "epoch": 1.8464962599238217,
      "grad_norm": 0.43873971700668335,
      "learning_rate": 3.341110137446094e-06,
      "loss": 0.3014,
      "step": 40237
    },
    {
      "epoch": 1.8465421504290762,
      "grad_norm": 0.4224017560482025,
      "learning_rate": 3.340878839589135e-06,
      "loss": 0.3039,
      "step": 40238
    },
    {
      "epoch": 1.8465880409343307,
      "grad_norm": 0.48476657271385193,
      "learning_rate": 3.340647545721839e-06,
      "loss": 0.3586,
      "step": 40239
    },
    {
      "epoch": 1.8466339314395852,
      "grad_norm": 0.4460589587688446,
      "learning_rate": 3.3404162558447595e-06,
      "loss": 0.3212,
      "step": 40240
    },
    {
      "epoch": 1.8466798219448397,
      "grad_norm": 0.4813358187675476,
      "learning_rate": 3.340184969958458e-06,
      "loss": 0.3462,
      "step": 40241
    },
    {
      "epoch": 1.8467257124500942,
      "grad_norm": 0.4737553596496582,
      "learning_rate": 3.3399536880634863e-06,
      "loss": 0.3371,
      "step": 40242
    },
    {
      "epoch": 1.8467716029553487,
      "grad_norm": 0.4661926031112671,
      "learning_rate": 3.3397224101604032e-06,
      "loss": 0.3431,
      "step": 40243
    },
    {
      "epoch": 1.846817493460603,
      "grad_norm": 0.47274938225746155,
      "learning_rate": 3.339491136249763e-06,
      "loss": 0.3791,
      "step": 40244
    },
    {
      "epoch": 1.8468633839658575,
      "grad_norm": 0.4470342695713043,
      "learning_rate": 3.339259866332123e-06,
      "loss": 0.3085,
      "step": 40245
    },
    {
      "epoch": 1.846909274471112,
      "grad_norm": 0.44906309247016907,
      "learning_rate": 3.3390286004080388e-06,
      "loss": 0.348,
      "step": 40246
    },
    {
      "epoch": 1.8469551649763662,
      "grad_norm": 0.47322139143943787,
      "learning_rate": 3.338797338478065e-06,
      "loss": 0.3477,
      "step": 40247
    },
    {
      "epoch": 1.8470010554816207,
      "grad_norm": 0.45633676648139954,
      "learning_rate": 3.3385660805427612e-06,
      "loss": 0.3317,
      "step": 40248
    },
    {
      "epoch": 1.8470469459868752,
      "grad_norm": 0.4716099202632904,
      "learning_rate": 3.3383348266026815e-06,
      "loss": 0.3829,
      "step": 40249
    },
    {
      "epoch": 1.8470928364921297,
      "grad_norm": 0.520809531211853,
      "learning_rate": 3.33810357665838e-06,
      "loss": 0.4052,
      "step": 40250
    },
    {
      "epoch": 1.8471387269973842,
      "grad_norm": 0.47469982504844666,
      "learning_rate": 3.3378723307104167e-06,
      "loss": 0.3602,
      "step": 40251
    },
    {
      "epoch": 1.8471846175026387,
      "grad_norm": 0.4338953495025635,
      "learning_rate": 3.3376410887593464e-06,
      "loss": 0.2839,
      "step": 40252
    },
    {
      "epoch": 1.8472305080078932,
      "grad_norm": 0.5152711868286133,
      "learning_rate": 3.3374098508057227e-06,
      "loss": 0.4096,
      "step": 40253
    },
    {
      "epoch": 1.8472763985131477,
      "grad_norm": 0.46566447615623474,
      "learning_rate": 3.337178616850106e-06,
      "loss": 0.3282,
      "step": 40254
    },
    {
      "epoch": 1.8473222890184022,
      "grad_norm": 0.4541769027709961,
      "learning_rate": 3.336947386893049e-06,
      "loss": 0.3584,
      "step": 40255
    },
    {
      "epoch": 1.8473681795236565,
      "grad_norm": 0.4525289833545685,
      "learning_rate": 3.336716160935106e-06,
      "loss": 0.3516,
      "step": 40256
    },
    {
      "epoch": 1.847414070028911,
      "grad_norm": 0.48179036378860474,
      "learning_rate": 3.336484938976839e-06,
      "loss": 0.3348,
      "step": 40257
    },
    {
      "epoch": 1.8474599605341655,
      "grad_norm": 0.4722668528556824,
      "learning_rate": 3.3362537210188e-06,
      "loss": 0.3466,
      "step": 40258
    },
    {
      "epoch": 1.8475058510394198,
      "grad_norm": 0.43033891916275024,
      "learning_rate": 3.336022507061545e-06,
      "loss": 0.2623,
      "step": 40259
    },
    {
      "epoch": 1.8475517415446743,
      "grad_norm": 0.4403129816055298,
      "learning_rate": 3.3357912971056318e-06,
      "loss": 0.3273,
      "step": 40260
    },
    {
      "epoch": 1.8475976320499288,
      "grad_norm": 0.47655537724494934,
      "learning_rate": 3.335560091151615e-06,
      "loss": 0.356,
      "step": 40261
    },
    {
      "epoch": 1.8476435225551833,
      "grad_norm": 0.5336896181106567,
      "learning_rate": 3.3353288892000486e-06,
      "loss": 0.4452,
      "step": 40262
    },
    {
      "epoch": 1.8476894130604378,
      "grad_norm": 0.4733862578868866,
      "learning_rate": 3.3350976912514932e-06,
      "loss": 0.4135,
      "step": 40263
    },
    {
      "epoch": 1.8477353035656923,
      "grad_norm": 0.43277111649513245,
      "learning_rate": 3.3348664973065026e-06,
      "loss": 0.3023,
      "step": 40264
    },
    {
      "epoch": 1.8477811940709468,
      "grad_norm": 0.49459731578826904,
      "learning_rate": 3.3346353073656307e-06,
      "loss": 0.3522,
      "step": 40265
    },
    {
      "epoch": 1.8478270845762013,
      "grad_norm": 0.4786777198314667,
      "learning_rate": 3.334404121429437e-06,
      "loss": 0.3697,
      "step": 40266
    },
    {
      "epoch": 1.8478729750814558,
      "grad_norm": 0.4714864194393158,
      "learning_rate": 3.3341729394984755e-06,
      "loss": 0.3509,
      "step": 40267
    },
    {
      "epoch": 1.8479188655867103,
      "grad_norm": 0.44262534379959106,
      "learning_rate": 3.3339417615733004e-06,
      "loss": 0.2833,
      "step": 40268
    },
    {
      "epoch": 1.8479647560919645,
      "grad_norm": 0.46255114674568176,
      "learning_rate": 3.3337105876544718e-06,
      "loss": 0.335,
      "step": 40269
    },
    {
      "epoch": 1.848010646597219,
      "grad_norm": 0.4758565127849579,
      "learning_rate": 3.333479417742544e-06,
      "loss": 0.3614,
      "step": 40270
    },
    {
      "epoch": 1.8480565371024735,
      "grad_norm": 0.4858417510986328,
      "learning_rate": 3.333248251838069e-06,
      "loss": 0.4198,
      "step": 40271
    },
    {
      "epoch": 1.8481024276077278,
      "grad_norm": 0.4302600026130676,
      "learning_rate": 3.333017089941608e-06,
      "loss": 0.2885,
      "step": 40272
    },
    {
      "epoch": 1.8481483181129823,
      "grad_norm": 0.5097487568855286,
      "learning_rate": 3.3327859320537153e-06,
      "loss": 0.4601,
      "step": 40273
    },
    {
      "epoch": 1.8481942086182368,
      "grad_norm": 0.4900307357311249,
      "learning_rate": 3.3325547781749455e-06,
      "loss": 0.3324,
      "step": 40274
    },
    {
      "epoch": 1.8482400991234913,
      "grad_norm": 0.4893048405647278,
      "learning_rate": 3.332323628305855e-06,
      "loss": 0.3806,
      "step": 40275
    },
    {
      "epoch": 1.8482859896287458,
      "grad_norm": 0.5331499576568604,
      "learning_rate": 3.3320924824470008e-06,
      "loss": 0.4681,
      "step": 40276
    },
    {
      "epoch": 1.8483318801340003,
      "grad_norm": 0.48338252305984497,
      "learning_rate": 3.3318613405989376e-06,
      "loss": 0.2752,
      "step": 40277
    },
    {
      "epoch": 1.8483777706392548,
      "grad_norm": 0.4524998366832733,
      "learning_rate": 3.3316302027622188e-06,
      "loss": 0.3381,
      "step": 40278
    },
    {
      "epoch": 1.8484236611445093,
      "grad_norm": 0.48449498414993286,
      "learning_rate": 3.3313990689374054e-06,
      "loss": 0.3887,
      "step": 40279
    },
    {
      "epoch": 1.8484695516497638,
      "grad_norm": 0.48849570751190186,
      "learning_rate": 3.3311679391250507e-06,
      "loss": 0.3984,
      "step": 40280
    },
    {
      "epoch": 1.8485154421550183,
      "grad_norm": 0.5508852005004883,
      "learning_rate": 3.3309368133257093e-06,
      "loss": 0.4589,
      "step": 40281
    },
    {
      "epoch": 1.8485613326602726,
      "grad_norm": 0.5098567605018616,
      "learning_rate": 3.3307056915399383e-06,
      "loss": 0.4117,
      "step": 40282
    },
    {
      "epoch": 1.848607223165527,
      "grad_norm": 0.43602102994918823,
      "learning_rate": 3.330474573768294e-06,
      "loss": 0.3172,
      "step": 40283
    },
    {
      "epoch": 1.8486531136707816,
      "grad_norm": 0.506976842880249,
      "learning_rate": 3.33024346001133e-06,
      "loss": 0.443,
      "step": 40284
    },
    {
      "epoch": 1.8486990041760358,
      "grad_norm": 0.45459091663360596,
      "learning_rate": 3.3300123502696047e-06,
      "loss": 0.3248,
      "step": 40285
    },
    {
      "epoch": 1.8487448946812903,
      "grad_norm": 0.5261184573173523,
      "learning_rate": 3.3297812445436728e-06,
      "loss": 0.3752,
      "step": 40286
    },
    {
      "epoch": 1.8487907851865448,
      "grad_norm": 0.44434213638305664,
      "learning_rate": 3.3295501428340886e-06,
      "loss": 0.301,
      "step": 40287
    },
    {
      "epoch": 1.8488366756917993,
      "grad_norm": 0.4724124073982239,
      "learning_rate": 3.3293190451414104e-06,
      "loss": 0.3754,
      "step": 40288
    },
    {
      "epoch": 1.8488825661970538,
      "grad_norm": 0.5689067244529724,
      "learning_rate": 3.3290879514661926e-06,
      "loss": 0.4046,
      "step": 40289
    },
    {
      "epoch": 1.8489284567023083,
      "grad_norm": 0.5080040097236633,
      "learning_rate": 3.328856861808989e-06,
      "loss": 0.4193,
      "step": 40290
    },
    {
      "epoch": 1.8489743472075628,
      "grad_norm": 0.47288674116134644,
      "learning_rate": 3.3286257761703588e-06,
      "loss": 0.369,
      "step": 40291
    },
    {
      "epoch": 1.8490202377128173,
      "grad_norm": 0.46641141176223755,
      "learning_rate": 3.328394694550858e-06,
      "loss": 0.322,
      "step": 40292
    },
    {
      "epoch": 1.8490661282180718,
      "grad_norm": 0.46567028760910034,
      "learning_rate": 3.3281636169510366e-06,
      "loss": 0.3303,
      "step": 40293
    },
    {
      "epoch": 1.849112018723326,
      "grad_norm": 0.4862552583217621,
      "learning_rate": 3.327932543371456e-06,
      "loss": 0.3449,
      "step": 40294
    },
    {
      "epoch": 1.8491579092285806,
      "grad_norm": 0.5060927271842957,
      "learning_rate": 3.32770147381267e-06,
      "loss": 0.3841,
      "step": 40295
    },
    {
      "epoch": 1.849203799733835,
      "grad_norm": 0.4582996666431427,
      "learning_rate": 3.327470408275234e-06,
      "loss": 0.3368,
      "step": 40296
    },
    {
      "epoch": 1.8492496902390896,
      "grad_norm": 0.41669654846191406,
      "learning_rate": 3.327239346759704e-06,
      "loss": 0.2736,
      "step": 40297
    },
    {
      "epoch": 1.8492955807443439,
      "grad_norm": 0.5514963269233704,
      "learning_rate": 3.3270082892666354e-06,
      "loss": 0.3806,
      "step": 40298
    },
    {
      "epoch": 1.8493414712495984,
      "grad_norm": 0.5049227476119995,
      "learning_rate": 3.326777235796582e-06,
      "loss": 0.4137,
      "step": 40299
    },
    {
      "epoch": 1.8493873617548529,
      "grad_norm": 0.4483928084373474,
      "learning_rate": 3.3265461863501036e-06,
      "loss": 0.2924,
      "step": 40300
    },
    {
      "epoch": 1.8494332522601074,
      "grad_norm": 0.4311830401420593,
      "learning_rate": 3.3263151409277527e-06,
      "loss": 0.2638,
      "step": 40301
    },
    {
      "epoch": 1.8494791427653618,
      "grad_norm": 0.5194509029388428,
      "learning_rate": 3.3260840995300853e-06,
      "loss": 0.4682,
      "step": 40302
    },
    {
      "epoch": 1.8495250332706163,
      "grad_norm": 0.4780785143375397,
      "learning_rate": 3.325853062157658e-06,
      "loss": 0.3734,
      "step": 40303
    },
    {
      "epoch": 1.8495709237758708,
      "grad_norm": 0.4819948971271515,
      "learning_rate": 3.3256220288110263e-06,
      "loss": 0.3773,
      "step": 40304
    },
    {
      "epoch": 1.8496168142811253,
      "grad_norm": 0.47465479373931885,
      "learning_rate": 3.3253909994907425e-06,
      "loss": 0.3794,
      "step": 40305
    },
    {
      "epoch": 1.8496627047863798,
      "grad_norm": 0.4980343282222748,
      "learning_rate": 3.3251599741973674e-06,
      "loss": 0.3844,
      "step": 40306
    },
    {
      "epoch": 1.8497085952916341,
      "grad_norm": 0.48781460523605347,
      "learning_rate": 3.324928952931453e-06,
      "loss": 0.4202,
      "step": 40307
    },
    {
      "epoch": 1.8497544857968886,
      "grad_norm": 0.4286201298236847,
      "learning_rate": 3.3246979356935573e-06,
      "loss": 0.2888,
      "step": 40308
    },
    {
      "epoch": 1.8498003763021431,
      "grad_norm": 0.4626065492630005,
      "learning_rate": 3.324466922484232e-06,
      "loss": 0.3528,
      "step": 40309
    },
    {
      "epoch": 1.8498462668073974,
      "grad_norm": 0.5187747478485107,
      "learning_rate": 3.3242359133040366e-06,
      "loss": 0.4055,
      "step": 40310
    },
    {
      "epoch": 1.8498921573126519,
      "grad_norm": 0.681985080242157,
      "learning_rate": 3.324004908153524e-06,
      "loss": 0.3167,
      "step": 40311
    },
    {
      "epoch": 1.8499380478179064,
      "grad_norm": 0.4886026382446289,
      "learning_rate": 3.3237739070332496e-06,
      "loss": 0.3698,
      "step": 40312
    },
    {
      "epoch": 1.8499839383231609,
      "grad_norm": 0.47821518778800964,
      "learning_rate": 3.3235429099437723e-06,
      "loss": 0.3825,
      "step": 40313
    },
    {
      "epoch": 1.8500298288284154,
      "grad_norm": 0.5394052863121033,
      "learning_rate": 3.3233119168856454e-06,
      "loss": 0.4648,
      "step": 40314
    },
    {
      "epoch": 1.8500757193336699,
      "grad_norm": 0.4873966574668884,
      "learning_rate": 3.323080927859421e-06,
      "loss": 0.3259,
      "step": 40315
    },
    {
      "epoch": 1.8501216098389244,
      "grad_norm": 0.4885316491127014,
      "learning_rate": 3.3228499428656603e-06,
      "loss": 0.4246,
      "step": 40316
    },
    {
      "epoch": 1.8501675003441789,
      "grad_norm": 0.4414200782775879,
      "learning_rate": 3.322618961904916e-06,
      "loss": 0.2915,
      "step": 40317
    },
    {
      "epoch": 1.8502133908494334,
      "grad_norm": 0.5047275424003601,
      "learning_rate": 3.3223879849777426e-06,
      "loss": 0.3513,
      "step": 40318
    },
    {
      "epoch": 1.8502592813546879,
      "grad_norm": 0.5115063786506653,
      "learning_rate": 3.3221570120846974e-06,
      "loss": 0.4235,
      "step": 40319
    },
    {
      "epoch": 1.8503051718599421,
      "grad_norm": 0.5689840912818909,
      "learning_rate": 3.3219260432263357e-06,
      "loss": 0.3681,
      "step": 40320
    },
    {
      "epoch": 1.8503510623651966,
      "grad_norm": 0.517750084400177,
      "learning_rate": 3.3216950784032097e-06,
      "loss": 0.388,
      "step": 40321
    },
    {
      "epoch": 1.8503969528704511,
      "grad_norm": 0.49009838700294495,
      "learning_rate": 3.321464117615879e-06,
      "loss": 0.3593,
      "step": 40322
    },
    {
      "epoch": 1.8504428433757054,
      "grad_norm": 0.4579828977584839,
      "learning_rate": 3.3212331608648986e-06,
      "loss": 0.3769,
      "step": 40323
    },
    {
      "epoch": 1.85048873388096,
      "grad_norm": 0.4815659523010254,
      "learning_rate": 3.32100220815082e-06,
      "loss": 0.3907,
      "step": 40324
    },
    {
      "epoch": 1.8505346243862144,
      "grad_norm": 0.4690357744693756,
      "learning_rate": 3.320771259474204e-06,
      "loss": 0.3501,
      "step": 40325
    },
    {
      "epoch": 1.850580514891469,
      "grad_norm": 0.4366646409034729,
      "learning_rate": 3.320540314835602e-06,
      "loss": 0.2868,
      "step": 40326
    },
    {
      "epoch": 1.8506264053967234,
      "grad_norm": 0.49968916177749634,
      "learning_rate": 3.3203093742355685e-06,
      "loss": 0.3309,
      "step": 40327
    },
    {
      "epoch": 1.850672295901978,
      "grad_norm": 0.4759405553340912,
      "learning_rate": 3.3200784376746632e-06,
      "loss": 0.3633,
      "step": 40328
    },
    {
      "epoch": 1.8507181864072324,
      "grad_norm": 0.4625864028930664,
      "learning_rate": 3.3198475051534386e-06,
      "loss": 0.3927,
      "step": 40329
    },
    {
      "epoch": 1.850764076912487,
      "grad_norm": 0.4834825396537781,
      "learning_rate": 3.3196165766724497e-06,
      "loss": 0.3378,
      "step": 40330
    },
    {
      "epoch": 1.8508099674177414,
      "grad_norm": 0.4236920177936554,
      "learning_rate": 3.319385652232253e-06,
      "loss": 0.3097,
      "step": 40331
    },
    {
      "epoch": 1.850855857922996,
      "grad_norm": 0.5523902773857117,
      "learning_rate": 3.3191547318334043e-06,
      "loss": 0.4958,
      "step": 40332
    },
    {
      "epoch": 1.8509017484282502,
      "grad_norm": 0.43546047806739807,
      "learning_rate": 3.318923815476456e-06,
      "loss": 0.2831,
      "step": 40333
    },
    {
      "epoch": 1.8509476389335047,
      "grad_norm": 0.47303926944732666,
      "learning_rate": 3.318692903161967e-06,
      "loss": 0.3639,
      "step": 40334
    },
    {
      "epoch": 1.8509935294387592,
      "grad_norm": 0.43903714418411255,
      "learning_rate": 3.3184619948904905e-06,
      "loss": 0.3352,
      "step": 40335
    },
    {
      "epoch": 1.8510394199440134,
      "grad_norm": 0.5200567245483398,
      "learning_rate": 3.318231090662582e-06,
      "loss": 0.3702,
      "step": 40336
    },
    {
      "epoch": 1.851085310449268,
      "grad_norm": 0.47132304310798645,
      "learning_rate": 3.318000190478798e-06,
      "loss": 0.3095,
      "step": 40337
    },
    {
      "epoch": 1.8511312009545224,
      "grad_norm": 0.44837722182273865,
      "learning_rate": 3.317769294339692e-06,
      "loss": 0.2888,
      "step": 40338
    },
    {
      "epoch": 1.851177091459777,
      "grad_norm": 0.44245871901512146,
      "learning_rate": 3.31753840224582e-06,
      "loss": 0.324,
      "step": 40339
    },
    {
      "epoch": 1.8512229819650314,
      "grad_norm": 0.5134694576263428,
      "learning_rate": 3.317307514197737e-06,
      "loss": 0.4247,
      "step": 40340
    },
    {
      "epoch": 1.851268872470286,
      "grad_norm": 0.44494450092315674,
      "learning_rate": 3.3170766301959985e-06,
      "loss": 0.3162,
      "step": 40341
    },
    {
      "epoch": 1.8513147629755404,
      "grad_norm": 0.47726595401763916,
      "learning_rate": 3.31684575024116e-06,
      "loss": 0.3303,
      "step": 40342
    },
    {
      "epoch": 1.851360653480795,
      "grad_norm": 0.44975194334983826,
      "learning_rate": 3.3166148743337734e-06,
      "loss": 0.323,
      "step": 40343
    },
    {
      "epoch": 1.8514065439860494,
      "grad_norm": 0.5017104148864746,
      "learning_rate": 3.3163840024743998e-06,
      "loss": 0.4164,
      "step": 40344
    },
    {
      "epoch": 1.8514524344913037,
      "grad_norm": 0.4444129168987274,
      "learning_rate": 3.316153134663591e-06,
      "loss": 0.3172,
      "step": 40345
    },
    {
      "epoch": 1.8514983249965582,
      "grad_norm": 0.5043652057647705,
      "learning_rate": 3.315922270901901e-06,
      "loss": 0.393,
      "step": 40346
    },
    {
      "epoch": 1.8515442155018127,
      "grad_norm": 0.4725661873817444,
      "learning_rate": 3.315691411189888e-06,
      "loss": 0.3351,
      "step": 40347
    },
    {
      "epoch": 1.851590106007067,
      "grad_norm": 0.4680109918117523,
      "learning_rate": 3.315460555528105e-06,
      "loss": 0.3614,
      "step": 40348
    },
    {
      "epoch": 1.8516359965123215,
      "grad_norm": 0.44956454634666443,
      "learning_rate": 3.3152297039171066e-06,
      "loss": 0.3262,
      "step": 40349
    },
    {
      "epoch": 1.851681887017576,
      "grad_norm": 0.447170615196228,
      "learning_rate": 3.31499885635745e-06,
      "loss": 0.3391,
      "step": 40350
    },
    {
      "epoch": 1.8517277775228305,
      "grad_norm": 0.44262704253196716,
      "learning_rate": 3.31476801284969e-06,
      "loss": 0.2813,
      "step": 40351
    },
    {
      "epoch": 1.851773668028085,
      "grad_norm": 0.4534452557563782,
      "learning_rate": 3.31453717339438e-06,
      "loss": 0.373,
      "step": 40352
    },
    {
      "epoch": 1.8518195585333395,
      "grad_norm": 0.45108890533447266,
      "learning_rate": 3.314306337992077e-06,
      "loss": 0.3058,
      "step": 40353
    },
    {
      "epoch": 1.851865449038594,
      "grad_norm": 0.46963900327682495,
      "learning_rate": 3.314075506643335e-06,
      "loss": 0.3667,
      "step": 40354
    },
    {
      "epoch": 1.8519113395438485,
      "grad_norm": 0.5113492608070374,
      "learning_rate": 3.3138446793487077e-06,
      "loss": 0.3574,
      "step": 40355
    },
    {
      "epoch": 1.851957230049103,
      "grad_norm": 0.47289562225341797,
      "learning_rate": 3.3136138561087537e-06,
      "loss": 0.347,
      "step": 40356
    },
    {
      "epoch": 1.8520031205543575,
      "grad_norm": 0.45599907636642456,
      "learning_rate": 3.313383036924026e-06,
      "loss": 0.333,
      "step": 40357
    },
    {
      "epoch": 1.8520490110596117,
      "grad_norm": 0.4962487816810608,
      "learning_rate": 3.3131522217950794e-06,
      "loss": 0.4054,
      "step": 40358
    },
    {
      "epoch": 1.8520949015648662,
      "grad_norm": 0.4903612434864044,
      "learning_rate": 3.3129214107224696e-06,
      "loss": 0.3681,
      "step": 40359
    },
    {
      "epoch": 1.8521407920701207,
      "grad_norm": 0.4585355520248413,
      "learning_rate": 3.312690603706752e-06,
      "loss": 0.3355,
      "step": 40360
    },
    {
      "epoch": 1.852186682575375,
      "grad_norm": 0.5049539804458618,
      "learning_rate": 3.3124598007484793e-06,
      "loss": 0.3997,
      "step": 40361
    },
    {
      "epoch": 1.8522325730806295,
      "grad_norm": 0.4485749900341034,
      "learning_rate": 3.3122290018482094e-06,
      "loss": 0.3335,
      "step": 40362
    },
    {
      "epoch": 1.852278463585884,
      "grad_norm": 0.5551913976669312,
      "learning_rate": 3.311998207006496e-06,
      "loss": 0.4805,
      "step": 40363
    },
    {
      "epoch": 1.8523243540911385,
      "grad_norm": 0.46636343002319336,
      "learning_rate": 3.311767416223892e-06,
      "loss": 0.3563,
      "step": 40364
    },
    {
      "epoch": 1.852370244596393,
      "grad_norm": 0.474102646112442,
      "learning_rate": 3.311536629500957e-06,
      "loss": 0.3618,
      "step": 40365
    },
    {
      "epoch": 1.8524161351016475,
      "grad_norm": 0.49373120069503784,
      "learning_rate": 3.311305846838243e-06,
      "loss": 0.3934,
      "step": 40366
    },
    {
      "epoch": 1.852462025606902,
      "grad_norm": 0.4369644522666931,
      "learning_rate": 3.3110750682363054e-06,
      "loss": 0.3107,
      "step": 40367
    },
    {
      "epoch": 1.8525079161121565,
      "grad_norm": 0.4594891369342804,
      "learning_rate": 3.3108442936956997e-06,
      "loss": 0.3058,
      "step": 40368
    },
    {
      "epoch": 1.852553806617411,
      "grad_norm": 0.47270017862319946,
      "learning_rate": 3.31061352321698e-06,
      "loss": 0.3534,
      "step": 40369
    },
    {
      "epoch": 1.8525996971226655,
      "grad_norm": 0.4611585736274719,
      "learning_rate": 3.3103827568006996e-06,
      "loss": 0.3446,
      "step": 40370
    },
    {
      "epoch": 1.8526455876279198,
      "grad_norm": 0.49016016721725464,
      "learning_rate": 3.3101519944474183e-06,
      "loss": 0.373,
      "step": 40371
    },
    {
      "epoch": 1.8526914781331743,
      "grad_norm": 0.4962975084781647,
      "learning_rate": 3.3099212361576873e-06,
      "loss": 0.3924,
      "step": 40372
    },
    {
      "epoch": 1.8527373686384287,
      "grad_norm": 0.44242244958877563,
      "learning_rate": 3.309690481932062e-06,
      "loss": 0.3512,
      "step": 40373
    },
    {
      "epoch": 1.852783259143683,
      "grad_norm": 0.4601365327835083,
      "learning_rate": 3.3094597317710977e-06,
      "loss": 0.3653,
      "step": 40374
    },
    {
      "epoch": 1.8528291496489375,
      "grad_norm": 0.4547540843486786,
      "learning_rate": 3.309228985675349e-06,
      "loss": 0.304,
      "step": 40375
    },
    {
      "epoch": 1.852875040154192,
      "grad_norm": 0.45722106099128723,
      "learning_rate": 3.3089982436453714e-06,
      "loss": 0.3573,
      "step": 40376
    },
    {
      "epoch": 1.8529209306594465,
      "grad_norm": 0.4594864845275879,
      "learning_rate": 3.308767505681718e-06,
      "loss": 0.3615,
      "step": 40377
    },
    {
      "epoch": 1.852966821164701,
      "grad_norm": 0.4543708264827728,
      "learning_rate": 3.3085367717849465e-06,
      "loss": 0.3179,
      "step": 40378
    },
    {
      "epoch": 1.8530127116699555,
      "grad_norm": 0.4803710877895355,
      "learning_rate": 3.3083060419556102e-06,
      "loss": 0.2553,
      "step": 40379
    },
    {
      "epoch": 1.85305860217521,
      "grad_norm": 0.45421984791755676,
      "learning_rate": 3.308075316194263e-06,
      "loss": 0.3481,
      "step": 40380
    },
    {
      "epoch": 1.8531044926804645,
      "grad_norm": 0.48005250096321106,
      "learning_rate": 3.3078445945014614e-06,
      "loss": 0.3431,
      "step": 40381
    },
    {
      "epoch": 1.853150383185719,
      "grad_norm": 0.4485762417316437,
      "learning_rate": 3.3076138768777597e-06,
      "loss": 0.3273,
      "step": 40382
    },
    {
      "epoch": 1.8531962736909733,
      "grad_norm": 0.431786447763443,
      "learning_rate": 3.3073831633237117e-06,
      "loss": 0.3012,
      "step": 40383
    },
    {
      "epoch": 1.8532421641962278,
      "grad_norm": 0.4938342273235321,
      "learning_rate": 3.3071524538398735e-06,
      "loss": 0.343,
      "step": 40384
    },
    {
      "epoch": 1.8532880547014823,
      "grad_norm": 0.4701195955276489,
      "learning_rate": 3.3069217484268e-06,
      "loss": 0.3463,
      "step": 40385
    },
    {
      "epoch": 1.8533339452067368,
      "grad_norm": 0.47680822014808655,
      "learning_rate": 3.3066910470850432e-06,
      "loss": 0.3482,
      "step": 40386
    },
    {
      "epoch": 1.853379835711991,
      "grad_norm": 0.4916174113750458,
      "learning_rate": 3.3064603498151617e-06,
      "loss": 0.39,
      "step": 40387
    },
    {
      "epoch": 1.8534257262172456,
      "grad_norm": 0.47379714250564575,
      "learning_rate": 3.3062296566177087e-06,
      "loss": 0.3868,
      "step": 40388
    },
    {
      "epoch": 1.8534716167225,
      "grad_norm": 0.4527122676372528,
      "learning_rate": 3.305998967493238e-06,
      "loss": 0.3119,
      "step": 40389
    },
    {
      "epoch": 1.8535175072277545,
      "grad_norm": 0.47671982645988464,
      "learning_rate": 3.305768282442306e-06,
      "loss": 0.324,
      "step": 40390
    },
    {
      "epoch": 1.853563397733009,
      "grad_norm": 0.4962898790836334,
      "learning_rate": 3.305537601465466e-06,
      "loss": 0.3897,
      "step": 40391
    },
    {
      "epoch": 1.8536092882382635,
      "grad_norm": 0.5074235796928406,
      "learning_rate": 3.305306924563272e-06,
      "loss": 0.3738,
      "step": 40392
    },
    {
      "epoch": 1.853655178743518,
      "grad_norm": 0.45273494720458984,
      "learning_rate": 3.3050762517362824e-06,
      "loss": 0.3228,
      "step": 40393
    },
    {
      "epoch": 1.8537010692487725,
      "grad_norm": 0.49693411588668823,
      "learning_rate": 3.3048455829850486e-06,
      "loss": 0.3997,
      "step": 40394
    },
    {
      "epoch": 1.853746959754027,
      "grad_norm": 0.4722270369529724,
      "learning_rate": 3.3046149183101252e-06,
      "loss": 0.358,
      "step": 40395
    },
    {
      "epoch": 1.8537928502592813,
      "grad_norm": 0.47275233268737793,
      "learning_rate": 3.304384257712069e-06,
      "loss": 0.3652,
      "step": 40396
    },
    {
      "epoch": 1.8538387407645358,
      "grad_norm": 0.46848970651626587,
      "learning_rate": 3.304153601191434e-06,
      "loss": 0.3751,
      "step": 40397
    },
    {
      "epoch": 1.8538846312697903,
      "grad_norm": 0.4266578257083893,
      "learning_rate": 3.3039229487487717e-06,
      "loss": 0.281,
      "step": 40398
    },
    {
      "epoch": 1.8539305217750446,
      "grad_norm": 0.4510553181171417,
      "learning_rate": 3.3036923003846417e-06,
      "loss": 0.3301,
      "step": 40399
    },
    {
      "epoch": 1.853976412280299,
      "grad_norm": 0.4405747354030609,
      "learning_rate": 3.3034616560995963e-06,
      "loss": 0.2968,
      "step": 40400
    },
    {
      "epoch": 1.8540223027855536,
      "grad_norm": 0.507162868976593,
      "learning_rate": 3.3032310158941893e-06,
      "loss": 0.4171,
      "step": 40401
    },
    {
      "epoch": 1.854068193290808,
      "grad_norm": 0.4775787591934204,
      "learning_rate": 3.303000379768977e-06,
      "loss": 0.3502,
      "step": 40402
    },
    {
      "epoch": 1.8541140837960626,
      "grad_norm": 0.4950236678123474,
      "learning_rate": 3.3027697477245135e-06,
      "loss": 0.375,
      "step": 40403
    },
    {
      "epoch": 1.854159974301317,
      "grad_norm": 0.4665931761264801,
      "learning_rate": 3.3025391197613503e-06,
      "loss": 0.3595,
      "step": 40404
    },
    {
      "epoch": 1.8542058648065716,
      "grad_norm": 0.46330899000167847,
      "learning_rate": 3.3023084958800487e-06,
      "loss": 0.341,
      "step": 40405
    },
    {
      "epoch": 1.854251755311826,
      "grad_norm": 0.4906168282032013,
      "learning_rate": 3.302077876081158e-06,
      "loss": 0.3849,
      "step": 40406
    },
    {
      "epoch": 1.8542976458170806,
      "grad_norm": 0.4449165463447571,
      "learning_rate": 3.301847260365234e-06,
      "loss": 0.306,
      "step": 40407
    },
    {
      "epoch": 1.854343536322335,
      "grad_norm": 0.4921450614929199,
      "learning_rate": 3.3016166487328293e-06,
      "loss": 0.3547,
      "step": 40408
    },
    {
      "epoch": 1.8543894268275893,
      "grad_norm": 0.4533887803554535,
      "learning_rate": 3.301386041184503e-06,
      "loss": 0.3449,
      "step": 40409
    },
    {
      "epoch": 1.8544353173328438,
      "grad_norm": 0.40465766191482544,
      "learning_rate": 3.301155437720807e-06,
      "loss": 0.2498,
      "step": 40410
    },
    {
      "epoch": 1.8544812078380983,
      "grad_norm": 0.4959333837032318,
      "learning_rate": 3.300924838342295e-06,
      "loss": 0.3238,
      "step": 40411
    },
    {
      "epoch": 1.8545270983433526,
      "grad_norm": 0.5592528581619263,
      "learning_rate": 3.300694243049524e-06,
      "loss": 0.3555,
      "step": 40412
    },
    {
      "epoch": 1.854572988848607,
      "grad_norm": 0.49321016669273376,
      "learning_rate": 3.300463651843046e-06,
      "loss": 0.3944,
      "step": 40413
    },
    {
      "epoch": 1.8546188793538616,
      "grad_norm": 0.4886530637741089,
      "learning_rate": 3.3002330647234147e-06,
      "loss": 0.3601,
      "step": 40414
    },
    {
      "epoch": 1.854664769859116,
      "grad_norm": 0.46757471561431885,
      "learning_rate": 3.3000024816911892e-06,
      "loss": 0.3713,
      "step": 40415
    },
    {
      "epoch": 1.8547106603643706,
      "grad_norm": 0.6948164701461792,
      "learning_rate": 3.2997719027469205e-06,
      "loss": 0.2637,
      "step": 40416
    },
    {
      "epoch": 1.854756550869625,
      "grad_norm": 0.5493487119674683,
      "learning_rate": 3.299541327891163e-06,
      "loss": 0.392,
      "step": 40417
    },
    {
      "epoch": 1.8548024413748796,
      "grad_norm": 0.47073256969451904,
      "learning_rate": 3.299310757124473e-06,
      "loss": 0.3356,
      "step": 40418
    },
    {
      "epoch": 1.854848331880134,
      "grad_norm": 0.4546600878238678,
      "learning_rate": 3.2990801904474035e-06,
      "loss": 0.3254,
      "step": 40419
    },
    {
      "epoch": 1.8548942223853886,
      "grad_norm": 0.5111712217330933,
      "learning_rate": 3.2988496278605078e-06,
      "loss": 0.4136,
      "step": 40420
    },
    {
      "epoch": 1.854940112890643,
      "grad_norm": 0.49000680446624756,
      "learning_rate": 3.2986190693643437e-06,
      "loss": 0.415,
      "step": 40421
    },
    {
      "epoch": 1.8549860033958974,
      "grad_norm": 0.47147336602211,
      "learning_rate": 3.298388514959464e-06,
      "loss": 0.3591,
      "step": 40422
    },
    {
      "epoch": 1.8550318939011519,
      "grad_norm": 0.4704801142215729,
      "learning_rate": 3.2981579646464213e-06,
      "loss": 0.3326,
      "step": 40423
    },
    {
      "epoch": 1.8550777844064064,
      "grad_norm": 0.48693814873695374,
      "learning_rate": 3.2979274184257725e-06,
      "loss": 0.3991,
      "step": 40424
    },
    {
      "epoch": 1.8551236749116606,
      "grad_norm": 0.5025617480278015,
      "learning_rate": 3.297696876298072e-06,
      "loss": 0.3588,
      "step": 40425
    },
    {
      "epoch": 1.8551695654169151,
      "grad_norm": 0.48962709307670593,
      "learning_rate": 3.2974663382638705e-06,
      "loss": 0.3623,
      "step": 40426
    },
    {
      "epoch": 1.8552154559221696,
      "grad_norm": 0.5061325430870056,
      "learning_rate": 3.2972358043237286e-06,
      "loss": 0.4197,
      "step": 40427
    },
    {
      "epoch": 1.8552613464274241,
      "grad_norm": 0.54630446434021,
      "learning_rate": 3.2970052744781956e-06,
      "loss": 0.4393,
      "step": 40428
    },
    {
      "epoch": 1.8553072369326786,
      "grad_norm": 0.44675150513648987,
      "learning_rate": 3.2967747487278256e-06,
      "loss": 0.3368,
      "step": 40429
    },
    {
      "epoch": 1.8553531274379331,
      "grad_norm": 0.45063328742980957,
      "learning_rate": 3.2965442270731774e-06,
      "loss": 0.3318,
      "step": 40430
    },
    {
      "epoch": 1.8553990179431876,
      "grad_norm": 0.4517485201358795,
      "learning_rate": 3.2963137095148022e-06,
      "loss": 0.3408,
      "step": 40431
    },
    {
      "epoch": 1.8554449084484421,
      "grad_norm": 0.4778563380241394,
      "learning_rate": 3.2960831960532535e-06,
      "loss": 0.3267,
      "step": 40432
    },
    {
      "epoch": 1.8554907989536966,
      "grad_norm": 0.46979695558547974,
      "learning_rate": 3.295852686689088e-06,
      "loss": 0.3883,
      "step": 40433
    },
    {
      "epoch": 1.855536689458951,
      "grad_norm": 0.4589351415634155,
      "learning_rate": 3.2956221814228593e-06,
      "loss": 0.3199,
      "step": 40434
    },
    {
      "epoch": 1.8555825799642054,
      "grad_norm": 0.49148833751678467,
      "learning_rate": 3.2953916802551193e-06,
      "loss": 0.4179,
      "step": 40435
    },
    {
      "epoch": 1.85562847046946,
      "grad_norm": 0.48249709606170654,
      "learning_rate": 3.2951611831864266e-06,
      "loss": 0.3449,
      "step": 40436
    },
    {
      "epoch": 1.8556743609747142,
      "grad_norm": 0.4216356873512268,
      "learning_rate": 3.2949306902173327e-06,
      "loss": 0.2628,
      "step": 40437
    },
    {
      "epoch": 1.8557202514799687,
      "grad_norm": 0.4652479588985443,
      "learning_rate": 3.294700201348393e-06,
      "loss": 0.3744,
      "step": 40438
    },
    {
      "epoch": 1.8557661419852232,
      "grad_norm": 0.4533633887767792,
      "learning_rate": 3.2944697165801598e-06,
      "loss": 0.3312,
      "step": 40439
    },
    {
      "epoch": 1.8558120324904777,
      "grad_norm": 0.47569209337234497,
      "learning_rate": 3.294239235913189e-06,
      "loss": 0.3598,
      "step": 40440
    },
    {
      "epoch": 1.8558579229957322,
      "grad_norm": 0.4762534201145172,
      "learning_rate": 3.294008759348035e-06,
      "loss": 0.3888,
      "step": 40441
    },
    {
      "epoch": 1.8559038135009867,
      "grad_norm": 0.4759160876274109,
      "learning_rate": 3.2937782868852497e-06,
      "loss": 0.4382,
      "step": 40442
    },
    {
      "epoch": 1.8559497040062412,
      "grad_norm": 0.49226096272468567,
      "learning_rate": 3.293547818525391e-06,
      "loss": 0.3813,
      "step": 40443
    },
    {
      "epoch": 1.8559955945114956,
      "grad_norm": 0.45572954416275024,
      "learning_rate": 3.2933173542690112e-06,
      "loss": 0.3551,
      "step": 40444
    },
    {
      "epoch": 1.8560414850167501,
      "grad_norm": 0.47553253173828125,
      "learning_rate": 3.293086894116663e-06,
      "loss": 0.3827,
      "step": 40445
    },
    {
      "epoch": 1.8560873755220046,
      "grad_norm": 0.44638678431510925,
      "learning_rate": 3.2928564380689036e-06,
      "loss": 0.2872,
      "step": 40446
    },
    {
      "epoch": 1.856133266027259,
      "grad_norm": 0.4659976363182068,
      "learning_rate": 3.2926259861262855e-06,
      "loss": 0.3679,
      "step": 40447
    },
    {
      "epoch": 1.8561791565325134,
      "grad_norm": 0.4820985496044159,
      "learning_rate": 3.2923955382893614e-06,
      "loss": 0.3361,
      "step": 40448
    },
    {
      "epoch": 1.856225047037768,
      "grad_norm": 0.4595310389995575,
      "learning_rate": 3.2921650945586905e-06,
      "loss": 0.3569,
      "step": 40449
    },
    {
      "epoch": 1.8562709375430222,
      "grad_norm": 0.5015496015548706,
      "learning_rate": 3.2919346549348215e-06,
      "loss": 0.3795,
      "step": 40450
    },
    {
      "epoch": 1.8563168280482767,
      "grad_norm": 0.46801725029945374,
      "learning_rate": 3.2917042194183093e-06,
      "loss": 0.3564,
      "step": 40451
    },
    {
      "epoch": 1.8563627185535312,
      "grad_norm": 0.4736338257789612,
      "learning_rate": 3.2914737880097103e-06,
      "loss": 0.3722,
      "step": 40452
    },
    {
      "epoch": 1.8564086090587857,
      "grad_norm": 0.49746474623680115,
      "learning_rate": 3.2912433607095782e-06,
      "loss": 0.3622,
      "step": 40453
    },
    {
      "epoch": 1.8564544995640402,
      "grad_norm": 0.45160624384880066,
      "learning_rate": 3.2910129375184665e-06,
      "loss": 0.3278,
      "step": 40454
    },
    {
      "epoch": 1.8565003900692947,
      "grad_norm": 0.5271814465522766,
      "learning_rate": 3.290782518436929e-06,
      "loss": 0.4645,
      "step": 40455
    },
    {
      "epoch": 1.8565462805745492,
      "grad_norm": 0.4491087794303894,
      "learning_rate": 3.2905521034655207e-06,
      "loss": 0.3067,
      "step": 40456
    },
    {
      "epoch": 1.8565921710798037,
      "grad_norm": 0.4486611485481262,
      "learning_rate": 3.2903216926047925e-06,
      "loss": 0.338,
      "step": 40457
    },
    {
      "epoch": 1.8566380615850582,
      "grad_norm": 0.4695137143135071,
      "learning_rate": 3.2900912858553035e-06,
      "loss": 0.3674,
      "step": 40458
    },
    {
      "epoch": 1.8566839520903127,
      "grad_norm": 0.45977285504341125,
      "learning_rate": 3.289860883217605e-06,
      "loss": 0.3595,
      "step": 40459
    },
    {
      "epoch": 1.856729842595567,
      "grad_norm": 0.4911457598209381,
      "learning_rate": 3.289630484692251e-06,
      "loss": 0.3421,
      "step": 40460
    },
    {
      "epoch": 1.8567757331008214,
      "grad_norm": 0.4917941391468048,
      "learning_rate": 3.289400090279796e-06,
      "loss": 0.3954,
      "step": 40461
    },
    {
      "epoch": 1.856821623606076,
      "grad_norm": 0.4886419475078583,
      "learning_rate": 3.2891696999807944e-06,
      "loss": 0.3747,
      "step": 40462
    },
    {
      "epoch": 1.8568675141113302,
      "grad_norm": 0.4697672426700592,
      "learning_rate": 3.288939313795798e-06,
      "loss": 0.292,
      "step": 40463
    },
    {
      "epoch": 1.8569134046165847,
      "grad_norm": 0.4708062708377838,
      "learning_rate": 3.288708931725364e-06,
      "loss": 0.3285,
      "step": 40464
    },
    {
      "epoch": 1.8569592951218392,
      "grad_norm": 0.47901931405067444,
      "learning_rate": 3.2884785537700452e-06,
      "loss": 0.3298,
      "step": 40465
    },
    {
      "epoch": 1.8570051856270937,
      "grad_norm": 0.4290608763694763,
      "learning_rate": 3.2882481799303946e-06,
      "loss": 0.2688,
      "step": 40466
    },
    {
      "epoch": 1.8570510761323482,
      "grad_norm": 0.45043766498565674,
      "learning_rate": 3.288017810206967e-06,
      "loss": 0.3236,
      "step": 40467
    },
    {
      "epoch": 1.8570969666376027,
      "grad_norm": 0.4941132366657257,
      "learning_rate": 3.2877874446003167e-06,
      "loss": 0.4004,
      "step": 40468
    },
    {
      "epoch": 1.8571428571428572,
      "grad_norm": 0.478247731924057,
      "learning_rate": 3.2875570831109956e-06,
      "loss": 0.34,
      "step": 40469
    },
    {
      "epoch": 1.8571887476481117,
      "grad_norm": 0.5179359316825867,
      "learning_rate": 3.2873267257395607e-06,
      "loss": 0.3823,
      "step": 40470
    },
    {
      "epoch": 1.8572346381533662,
      "grad_norm": 0.4760451018810272,
      "learning_rate": 3.2870963724865656e-06,
      "loss": 0.3535,
      "step": 40471
    },
    {
      "epoch": 1.8572805286586205,
      "grad_norm": 0.49085113406181335,
      "learning_rate": 3.286866023352562e-06,
      "loss": 0.3616,
      "step": 40472
    },
    {
      "epoch": 1.857326419163875,
      "grad_norm": 0.44673463702201843,
      "learning_rate": 3.286635678338104e-06,
      "loss": 0.3649,
      "step": 40473
    },
    {
      "epoch": 1.8573723096691295,
      "grad_norm": 0.46099433302879333,
      "learning_rate": 3.286405337443747e-06,
      "loss": 0.3279,
      "step": 40474
    },
    {
      "epoch": 1.857418200174384,
      "grad_norm": 0.44094225764274597,
      "learning_rate": 3.2861750006700453e-06,
      "loss": 0.3351,
      "step": 40475
    },
    {
      "epoch": 1.8574640906796382,
      "grad_norm": 0.47792744636535645,
      "learning_rate": 3.28594466801755e-06,
      "loss": 0.4039,
      "step": 40476
    },
    {
      "epoch": 1.8575099811848927,
      "grad_norm": 0.478287935256958,
      "learning_rate": 3.2857143394868184e-06,
      "loss": 0.3688,
      "step": 40477
    },
    {
      "epoch": 1.8575558716901472,
      "grad_norm": 0.48180848360061646,
      "learning_rate": 3.285484015078403e-06,
      "loss": 0.3792,
      "step": 40478
    },
    {
      "epoch": 1.8576017621954017,
      "grad_norm": 0.48857712745666504,
      "learning_rate": 3.2852536947928547e-06,
      "loss": 0.3795,
      "step": 40479
    },
    {
      "epoch": 1.8576476527006562,
      "grad_norm": 0.4339185655117035,
      "learning_rate": 3.2850233786307327e-06,
      "loss": 0.2812,
      "step": 40480
    },
    {
      "epoch": 1.8576935432059107,
      "grad_norm": 0.5463237762451172,
      "learning_rate": 3.2847930665925876e-06,
      "loss": 0.3457,
      "step": 40481
    },
    {
      "epoch": 1.8577394337111652,
      "grad_norm": 0.4906579554080963,
      "learning_rate": 3.2845627586789732e-06,
      "loss": 0.3874,
      "step": 40482
    },
    {
      "epoch": 1.8577853242164197,
      "grad_norm": 0.47759681940078735,
      "learning_rate": 3.2843324548904453e-06,
      "loss": 0.3848,
      "step": 40483
    },
    {
      "epoch": 1.8578312147216742,
      "grad_norm": 0.4853752851486206,
      "learning_rate": 3.2841021552275553e-06,
      "loss": 0.4042,
      "step": 40484
    },
    {
      "epoch": 1.8578771052269285,
      "grad_norm": 0.4175878167152405,
      "learning_rate": 3.283871859690857e-06,
      "loss": 0.3003,
      "step": 40485
    },
    {
      "epoch": 1.857922995732183,
      "grad_norm": 0.5099867582321167,
      "learning_rate": 3.2836415682809077e-06,
      "loss": 0.4504,
      "step": 40486
    },
    {
      "epoch": 1.8579688862374375,
      "grad_norm": 0.47399142384529114,
      "learning_rate": 3.2834112809982576e-06,
      "loss": 0.3742,
      "step": 40487
    },
    {
      "epoch": 1.8580147767426918,
      "grad_norm": 0.4545018970966339,
      "learning_rate": 3.2831809978434608e-06,
      "loss": 0.3144,
      "step": 40488
    },
    {
      "epoch": 1.8580606672479463,
      "grad_norm": 0.48586878180503845,
      "learning_rate": 3.2829507188170733e-06,
      "loss": 0.35,
      "step": 40489
    },
    {
      "epoch": 1.8581065577532008,
      "grad_norm": 0.4599045515060425,
      "learning_rate": 3.282720443919647e-06,
      "loss": 0.347,
      "step": 40490
    },
    {
      "epoch": 1.8581524482584553,
      "grad_norm": 0.49765047430992126,
      "learning_rate": 3.2824901731517343e-06,
      "loss": 0.3953,
      "step": 40491
    },
    {
      "epoch": 1.8581983387637098,
      "grad_norm": 0.5206354260444641,
      "learning_rate": 3.2822599065138927e-06,
      "loss": 0.4812,
      "step": 40492
    },
    {
      "epoch": 1.8582442292689643,
      "grad_norm": 0.4445689618587494,
      "learning_rate": 3.2820296440066747e-06,
      "loss": 0.2921,
      "step": 40493
    },
    {
      "epoch": 1.8582901197742188,
      "grad_norm": 0.47815993428230286,
      "learning_rate": 3.2817993856306307e-06,
      "loss": 0.3638,
      "step": 40494
    },
    {
      "epoch": 1.8583360102794733,
      "grad_norm": 0.42235100269317627,
      "learning_rate": 3.2815691313863186e-06,
      "loss": 0.2681,
      "step": 40495
    },
    {
      "epoch": 1.8583819007847278,
      "grad_norm": 0.42937490344047546,
      "learning_rate": 3.2813388812742897e-06,
      "loss": 0.2933,
      "step": 40496
    },
    {
      "epoch": 1.8584277912899823,
      "grad_norm": 0.47735100984573364,
      "learning_rate": 3.281108635295098e-06,
      "loss": 0.379,
      "step": 40497
    },
    {
      "epoch": 1.8584736817952365,
      "grad_norm": 0.4386392831802368,
      "learning_rate": 3.280878393449299e-06,
      "loss": 0.3121,
      "step": 40498
    },
    {
      "epoch": 1.858519572300491,
      "grad_norm": 0.4631252586841583,
      "learning_rate": 3.280648155737444e-06,
      "loss": 0.3436,
      "step": 40499
    },
    {
      "epoch": 1.8585654628057455,
      "grad_norm": 0.45992547273635864,
      "learning_rate": 3.2804179221600862e-06,
      "loss": 0.4004,
      "step": 40500
    },
    {
      "epoch": 1.8586113533109998,
      "grad_norm": 0.508155345916748,
      "learning_rate": 3.2801876927177823e-06,
      "loss": 0.4546,
      "step": 40501
    },
    {
      "epoch": 1.8586572438162543,
      "grad_norm": 0.4374437928199768,
      "learning_rate": 3.279957467411084e-06,
      "loss": 0.3227,
      "step": 40502
    },
    {
      "epoch": 1.8587031343215088,
      "grad_norm": 0.4607063829898834,
      "learning_rate": 3.2797272462405456e-06,
      "loss": 0.3417,
      "step": 40503
    },
    {
      "epoch": 1.8587490248267633,
      "grad_norm": 0.4352935552597046,
      "learning_rate": 3.2794970292067186e-06,
      "loss": 0.297,
      "step": 40504
    },
    {
      "epoch": 1.8587949153320178,
      "grad_norm": 0.49879348278045654,
      "learning_rate": 3.2792668163101597e-06,
      "loss": 0.3855,
      "step": 40505
    },
    {
      "epoch": 1.8588408058372723,
      "grad_norm": 0.46867433190345764,
      "learning_rate": 3.279036607551421e-06,
      "loss": 0.3317,
      "step": 40506
    },
    {
      "epoch": 1.8588866963425268,
      "grad_norm": 0.4629073441028595,
      "learning_rate": 3.278806402931054e-06,
      "loss": 0.3186,
      "step": 40507
    },
    {
      "epoch": 1.8589325868477813,
      "grad_norm": 0.42354097962379456,
      "learning_rate": 3.2785762024496166e-06,
      "loss": 0.2919,
      "step": 40508
    },
    {
      "epoch": 1.8589784773530358,
      "grad_norm": 0.4578912556171417,
      "learning_rate": 3.27834600610766e-06,
      "loss": 0.2997,
      "step": 40509
    },
    {
      "epoch": 1.8590243678582903,
      "grad_norm": 0.4417024552822113,
      "learning_rate": 3.2781158139057367e-06,
      "loss": 0.2944,
      "step": 40510
    },
    {
      "epoch": 1.8590702583635446,
      "grad_norm": 0.4865308403968811,
      "learning_rate": 3.2778856258444023e-06,
      "loss": 0.3874,
      "step": 40511
    },
    {
      "epoch": 1.859116148868799,
      "grad_norm": 0.4919701814651489,
      "learning_rate": 3.27765544192421e-06,
      "loss": 0.4023,
      "step": 40512
    },
    {
      "epoch": 1.8591620393740536,
      "grad_norm": 0.4876241385936737,
      "learning_rate": 3.2774252621457105e-06,
      "loss": 0.3991,
      "step": 40513
    },
    {
      "epoch": 1.8592079298793078,
      "grad_norm": 0.4662734270095825,
      "learning_rate": 3.2771950865094614e-06,
      "loss": 0.3734,
      "step": 40514
    },
    {
      "epoch": 1.8592538203845623,
      "grad_norm": 0.45262831449508667,
      "learning_rate": 3.276964915016016e-06,
      "loss": 0.3714,
      "step": 40515
    },
    {
      "epoch": 1.8592997108898168,
      "grad_norm": 0.4360681474208832,
      "learning_rate": 3.2767347476659227e-06,
      "loss": 0.2844,
      "step": 40516
    },
    {
      "epoch": 1.8593456013950713,
      "grad_norm": 0.4859558045864105,
      "learning_rate": 3.276504584459741e-06,
      "loss": 0.3359,
      "step": 40517
    },
    {
      "epoch": 1.8593914919003258,
      "grad_norm": 0.5471981167793274,
      "learning_rate": 3.2762744253980205e-06,
      "loss": 0.4847,
      "step": 40518
    },
    {
      "epoch": 1.8594373824055803,
      "grad_norm": 0.4872472286224365,
      "learning_rate": 3.2760442704813162e-06,
      "loss": 0.4234,
      "step": 40519
    },
    {
      "epoch": 1.8594832729108348,
      "grad_norm": 0.48090410232543945,
      "learning_rate": 3.275814119710182e-06,
      "loss": 0.3834,
      "step": 40520
    },
    {
      "epoch": 1.8595291634160893,
      "grad_norm": 0.46259573101997375,
      "learning_rate": 3.2755839730851707e-06,
      "loss": 0.3672,
      "step": 40521
    },
    {
      "epoch": 1.8595750539213438,
      "grad_norm": 0.44553032517433167,
      "learning_rate": 3.2753538306068335e-06,
      "loss": 0.3021,
      "step": 40522
    },
    {
      "epoch": 1.859620944426598,
      "grad_norm": 0.4630652666091919,
      "learning_rate": 3.275123692275728e-06,
      "loss": 0.3453,
      "step": 40523
    },
    {
      "epoch": 1.8596668349318526,
      "grad_norm": 0.48131272196769714,
      "learning_rate": 3.2748935580924057e-06,
      "loss": 0.366,
      "step": 40524
    },
    {
      "epoch": 1.859712725437107,
      "grad_norm": 0.4392104744911194,
      "learning_rate": 3.2746634280574196e-06,
      "loss": 0.3137,
      "step": 40525
    },
    {
      "epoch": 1.8597586159423614,
      "grad_norm": 0.504638135433197,
      "learning_rate": 3.2744333021713237e-06,
      "loss": 0.4026,
      "step": 40526
    },
    {
      "epoch": 1.8598045064476159,
      "grad_norm": 0.47061944007873535,
      "learning_rate": 3.2742031804346707e-06,
      "loss": 0.4089,
      "step": 40527
    },
    {
      "epoch": 1.8598503969528704,
      "grad_norm": 0.51091468334198,
      "learning_rate": 3.273973062848013e-06,
      "loss": 0.4545,
      "step": 40528
    },
    {
      "epoch": 1.8598962874581249,
      "grad_norm": 0.45529428124427795,
      "learning_rate": 3.273742949411908e-06,
      "loss": 0.3668,
      "step": 40529
    },
    {
      "epoch": 1.8599421779633794,
      "grad_norm": 0.4591984450817108,
      "learning_rate": 3.2735128401269055e-06,
      "loss": 0.3225,
      "step": 40530
    },
    {
      "epoch": 1.8599880684686338,
      "grad_norm": 0.4552486836910248,
      "learning_rate": 3.2732827349935585e-06,
      "loss": 0.335,
      "step": 40531
    },
    {
      "epoch": 1.8600339589738883,
      "grad_norm": 0.4579552412033081,
      "learning_rate": 3.273052634012423e-06,
      "loss": 0.3444,
      "step": 40532
    },
    {
      "epoch": 1.8600798494791428,
      "grad_norm": 0.4971858263015747,
      "learning_rate": 3.2728225371840507e-06,
      "loss": 0.3721,
      "step": 40533
    },
    {
      "epoch": 1.8601257399843973,
      "grad_norm": 0.49052736163139343,
      "learning_rate": 3.272592444508993e-06,
      "loss": 0.3941,
      "step": 40534
    },
    {
      "epoch": 1.8601716304896518,
      "grad_norm": 0.48033231496810913,
      "learning_rate": 3.2723623559878083e-06,
      "loss": 0.3236,
      "step": 40535
    },
    {
      "epoch": 1.8602175209949061,
      "grad_norm": 0.47988998889923096,
      "learning_rate": 3.2721322716210457e-06,
      "loss": 0.3573,
      "step": 40536
    },
    {
      "epoch": 1.8602634115001606,
      "grad_norm": 0.48498162627220154,
      "learning_rate": 3.271902191409261e-06,
      "loss": 0.3692,
      "step": 40537
    },
    {
      "epoch": 1.860309302005415,
      "grad_norm": 0.46444427967071533,
      "learning_rate": 3.271672115353003e-06,
      "loss": 0.3292,
      "step": 40538
    },
    {
      "epoch": 1.8603551925106694,
      "grad_norm": 0.4641346037387848,
      "learning_rate": 3.2714420434528305e-06,
      "loss": 0.3411,
      "step": 40539
    },
    {
      "epoch": 1.8604010830159239,
      "grad_norm": 0.4939061999320984,
      "learning_rate": 3.2712119757092936e-06,
      "loss": 0.4078,
      "step": 40540
    },
    {
      "epoch": 1.8604469735211784,
      "grad_norm": 0.47170281410217285,
      "learning_rate": 3.2709819121229457e-06,
      "loss": 0.3402,
      "step": 40541
    },
    {
      "epoch": 1.8604928640264329,
      "grad_norm": 0.5147426724433899,
      "learning_rate": 3.2707518526943416e-06,
      "loss": 0.4621,
      "step": 40542
    },
    {
      "epoch": 1.8605387545316874,
      "grad_norm": 0.45042237639427185,
      "learning_rate": 3.270521797424034e-06,
      "loss": 0.3181,
      "step": 40543
    },
    {
      "epoch": 1.8605846450369419,
      "grad_norm": 0.43745186924934387,
      "learning_rate": 3.2702917463125727e-06,
      "loss": 0.2944,
      "step": 40544
    },
    {
      "epoch": 1.8606305355421964,
      "grad_norm": 0.4619556963443756,
      "learning_rate": 3.2700616993605166e-06,
      "loss": 0.3252,
      "step": 40545
    },
    {
      "epoch": 1.8606764260474509,
      "grad_norm": 0.5352588295936584,
      "learning_rate": 3.269831656568416e-06,
      "loss": 0.3448,
      "step": 40546
    },
    {
      "epoch": 1.8607223165527054,
      "grad_norm": 0.4807080626487732,
      "learning_rate": 3.269601617936823e-06,
      "loss": 0.3714,
      "step": 40547
    },
    {
      "epoch": 1.8607682070579599,
      "grad_norm": 0.45593440532684326,
      "learning_rate": 3.2693715834662933e-06,
      "loss": 0.3477,
      "step": 40548
    },
    {
      "epoch": 1.8608140975632141,
      "grad_norm": 0.4112125337123871,
      "learning_rate": 3.269141553157378e-06,
      "loss": 0.281,
      "step": 40549
    },
    {
      "epoch": 1.8608599880684686,
      "grad_norm": 0.5036879181861877,
      "learning_rate": 3.2689115270106296e-06,
      "loss": 0.3991,
      "step": 40550
    },
    {
      "epoch": 1.8609058785737231,
      "grad_norm": 0.44279763102531433,
      "learning_rate": 3.2686815050266047e-06,
      "loss": 0.2933,
      "step": 40551
    },
    {
      "epoch": 1.8609517690789774,
      "grad_norm": 0.5018121004104614,
      "learning_rate": 3.2684514872058536e-06,
      "loss": 0.3782,
      "step": 40552
    },
    {
      "epoch": 1.860997659584232,
      "grad_norm": 0.5405827164649963,
      "learning_rate": 3.2682214735489305e-06,
      "loss": 0.4422,
      "step": 40553
    },
    {
      "epoch": 1.8610435500894864,
      "grad_norm": 0.4483787417411804,
      "learning_rate": 3.267991464056388e-06,
      "loss": 0.2834,
      "step": 40554
    },
    {
      "epoch": 1.861089440594741,
      "grad_norm": 0.4710089862346649,
      "learning_rate": 3.26776145872878e-06,
      "loss": 0.3488,
      "step": 40555
    },
    {
      "epoch": 1.8611353310999954,
      "grad_norm": 0.5031662583351135,
      "learning_rate": 3.2675314575666567e-06,
      "loss": 0.382,
      "step": 40556
    },
    {
      "epoch": 1.86118122160525,
      "grad_norm": 0.4852766692638397,
      "learning_rate": 3.2673014605705756e-06,
      "loss": 0.4361,
      "step": 40557
    },
    {
      "epoch": 1.8612271121105044,
      "grad_norm": 0.44936904311180115,
      "learning_rate": 3.2670714677410877e-06,
      "loss": 0.3462,
      "step": 40558
    },
    {
      "epoch": 1.861273002615759,
      "grad_norm": 0.4642120897769928,
      "learning_rate": 3.266841479078745e-06,
      "loss": 0.3416,
      "step": 40559
    },
    {
      "epoch": 1.8613188931210134,
      "grad_norm": 0.4379352927207947,
      "learning_rate": 3.2666114945841022e-06,
      "loss": 0.2628,
      "step": 40560
    },
    {
      "epoch": 1.8613647836262677,
      "grad_norm": 0.47870543599128723,
      "learning_rate": 3.266381514257712e-06,
      "loss": 0.3578,
      "step": 40561
    },
    {
      "epoch": 1.8614106741315222,
      "grad_norm": 0.5182487964630127,
      "learning_rate": 3.266151538100125e-06,
      "loss": 0.4522,
      "step": 40562
    },
    {
      "epoch": 1.8614565646367767,
      "grad_norm": 0.4527634382247925,
      "learning_rate": 3.2659215661118994e-06,
      "loss": 0.3019,
      "step": 40563
    },
    {
      "epoch": 1.8615024551420312,
      "grad_norm": 0.44533082842826843,
      "learning_rate": 3.2656915982935843e-06,
      "loss": 0.3166,
      "step": 40564
    },
    {
      "epoch": 1.8615483456472854,
      "grad_norm": 0.48734965920448303,
      "learning_rate": 3.265461634645731e-06,
      "loss": 0.3316,
      "step": 40565
    },
    {
      "epoch": 1.86159423615254,
      "grad_norm": 0.46406134963035583,
      "learning_rate": 3.2652316751688974e-06,
      "loss": 0.4028,
      "step": 40566
    },
    {
      "epoch": 1.8616401266577944,
      "grad_norm": 0.41516074538230896,
      "learning_rate": 3.265001719863634e-06,
      "loss": 0.2588,
      "step": 40567
    },
    {
      "epoch": 1.861686017163049,
      "grad_norm": 0.4555928409099579,
      "learning_rate": 3.264771768730495e-06,
      "loss": 0.361,
      "step": 40568
    },
    {
      "epoch": 1.8617319076683034,
      "grad_norm": 0.48160600662231445,
      "learning_rate": 3.26454182177003e-06,
      "loss": 0.4037,
      "step": 40569
    },
    {
      "epoch": 1.861777798173558,
      "grad_norm": 0.616723895072937,
      "learning_rate": 3.2643118789827953e-06,
      "loss": 0.3449,
      "step": 40570
    },
    {
      "epoch": 1.8618236886788124,
      "grad_norm": 0.45792356133461,
      "learning_rate": 3.2640819403693423e-06,
      "loss": 0.363,
      "step": 40571
    },
    {
      "epoch": 1.861869579184067,
      "grad_norm": 0.4561545252799988,
      "learning_rate": 3.263852005930223e-06,
      "loss": 0.3254,
      "step": 40572
    },
    {
      "epoch": 1.8619154696893214,
      "grad_norm": 0.45796236395835876,
      "learning_rate": 3.2636220756659937e-06,
      "loss": 0.3573,
      "step": 40573
    },
    {
      "epoch": 1.8619613601945757,
      "grad_norm": 0.4802989065647125,
      "learning_rate": 3.2633921495772053e-06,
      "loss": 0.3506,
      "step": 40574
    },
    {
      "epoch": 1.8620072506998302,
      "grad_norm": 0.6010094285011292,
      "learning_rate": 3.263162227664409e-06,
      "loss": 0.5173,
      "step": 40575
    },
    {
      "epoch": 1.8620531412050847,
      "grad_norm": 0.44536447525024414,
      "learning_rate": 3.2629323099281617e-06,
      "loss": 0.2904,
      "step": 40576
    },
    {
      "epoch": 1.862099031710339,
      "grad_norm": 0.44536781311035156,
      "learning_rate": 3.2627023963690128e-06,
      "loss": 0.2703,
      "step": 40577
    },
    {
      "epoch": 1.8621449222155935,
      "grad_norm": 0.4282838702201843,
      "learning_rate": 3.2624724869875147e-06,
      "loss": 0.2965,
      "step": 40578
    },
    {
      "epoch": 1.862190812720848,
      "grad_norm": 0.44729307293891907,
      "learning_rate": 3.2622425817842242e-06,
      "loss": 0.3085,
      "step": 40579
    },
    {
      "epoch": 1.8622367032261025,
      "grad_norm": 0.47040292620658875,
      "learning_rate": 3.262012680759692e-06,
      "loss": 0.3117,
      "step": 40580
    },
    {
      "epoch": 1.862282593731357,
      "grad_norm": 0.48567116260528564,
      "learning_rate": 3.261782783914469e-06,
      "loss": 0.3847,
      "step": 40581
    },
    {
      "epoch": 1.8623284842366115,
      "grad_norm": 0.4573099613189697,
      "learning_rate": 3.2615528912491113e-06,
      "loss": 0.3193,
      "step": 40582
    },
    {
      "epoch": 1.862374374741866,
      "grad_norm": 0.4340274930000305,
      "learning_rate": 3.2613230027641696e-06,
      "loss": 0.2949,
      "step": 40583
    },
    {
      "epoch": 1.8624202652471205,
      "grad_norm": 0.4891645610332489,
      "learning_rate": 3.2610931184601958e-06,
      "loss": 0.37,
      "step": 40584
    },
    {
      "epoch": 1.862466155752375,
      "grad_norm": 0.47471436858177185,
      "learning_rate": 3.2608632383377474e-06,
      "loss": 0.3567,
      "step": 40585
    },
    {
      "epoch": 1.8625120462576294,
      "grad_norm": 0.49920758605003357,
      "learning_rate": 3.260633362397373e-06,
      "loss": 0.378,
      "step": 40586
    },
    {
      "epoch": 1.8625579367628837,
      "grad_norm": 0.48607760667800903,
      "learning_rate": 3.260403490639624e-06,
      "loss": 0.4201,
      "step": 40587
    },
    {
      "epoch": 1.8626038272681382,
      "grad_norm": 0.49308910965919495,
      "learning_rate": 3.2601736230650583e-06,
      "loss": 0.3755,
      "step": 40588
    },
    {
      "epoch": 1.8626497177733927,
      "grad_norm": 0.4899759888648987,
      "learning_rate": 3.2599437596742255e-06,
      "loss": 0.3856,
      "step": 40589
    },
    {
      "epoch": 1.862695608278647,
      "grad_norm": 0.45371419191360474,
      "learning_rate": 3.259713900467678e-06,
      "loss": 0.319,
      "step": 40590
    },
    {
      "epoch": 1.8627414987839015,
      "grad_norm": 0.44646432995796204,
      "learning_rate": 3.2594840454459695e-06,
      "loss": 0.3402,
      "step": 40591
    },
    {
      "epoch": 1.862787389289156,
      "grad_norm": 0.46981915831565857,
      "learning_rate": 3.2592541946096536e-06,
      "loss": 0.3567,
      "step": 40592
    },
    {
      "epoch": 1.8628332797944105,
      "grad_norm": 0.4852912724018097,
      "learning_rate": 3.259024347959279e-06,
      "loss": 0.3918,
      "step": 40593
    },
    {
      "epoch": 1.862879170299665,
      "grad_norm": 0.47390133142471313,
      "learning_rate": 3.258794505495404e-06,
      "loss": 0.3559,
      "step": 40594
    },
    {
      "epoch": 1.8629250608049195,
      "grad_norm": 0.45345208048820496,
      "learning_rate": 3.258564667218579e-06,
      "loss": 0.3533,
      "step": 40595
    },
    {
      "epoch": 1.862970951310174,
      "grad_norm": 0.4535030126571655,
      "learning_rate": 3.2583348331293545e-06,
      "loss": 0.3477,
      "step": 40596
    },
    {
      "epoch": 1.8630168418154285,
      "grad_norm": 0.4521172046661377,
      "learning_rate": 3.2581050032282866e-06,
      "loss": 0.333,
      "step": 40597
    },
    {
      "epoch": 1.863062732320683,
      "grad_norm": 0.43522173166275024,
      "learning_rate": 3.2578751775159267e-06,
      "loss": 0.3031,
      "step": 40598
    },
    {
      "epoch": 1.8631086228259375,
      "grad_norm": 0.4725697636604309,
      "learning_rate": 3.2576453559928245e-06,
      "loss": 0.3582,
      "step": 40599
    },
    {
      "epoch": 1.8631545133311918,
      "grad_norm": 0.47011974453926086,
      "learning_rate": 3.2574155386595373e-06,
      "loss": 0.3827,
      "step": 40600
    },
    {
      "epoch": 1.8632004038364463,
      "grad_norm": 0.45221254229545593,
      "learning_rate": 3.257185725516616e-06,
      "loss": 0.3121,
      "step": 40601
    },
    {
      "epoch": 1.8632462943417007,
      "grad_norm": 0.4872588813304901,
      "learning_rate": 3.2569559165646136e-06,
      "loss": 0.3795,
      "step": 40602
    },
    {
      "epoch": 1.863292184846955,
      "grad_norm": 0.4899713397026062,
      "learning_rate": 3.2567261118040804e-06,
      "loss": 0.4439,
      "step": 40603
    },
    {
      "epoch": 1.8633380753522095,
      "grad_norm": 0.4765571057796478,
      "learning_rate": 3.2564963112355717e-06,
      "loss": 0.3656,
      "step": 40604
    },
    {
      "epoch": 1.863383965857464,
      "grad_norm": 0.4983557164669037,
      "learning_rate": 3.256266514859639e-06,
      "loss": 0.3777,
      "step": 40605
    },
    {
      "epoch": 1.8634298563627185,
      "grad_norm": 0.45017385482788086,
      "learning_rate": 3.256036722676833e-06,
      "loss": 0.3547,
      "step": 40606
    },
    {
      "epoch": 1.863475746867973,
      "grad_norm": 0.4200134575366974,
      "learning_rate": 3.2558069346877117e-06,
      "loss": 0.2997,
      "step": 40607
    },
    {
      "epoch": 1.8635216373732275,
      "grad_norm": 0.45535385608673096,
      "learning_rate": 3.255577150892823e-06,
      "loss": 0.2901,
      "step": 40608
    },
    {
      "epoch": 1.863567527878482,
      "grad_norm": 0.4469875395298004,
      "learning_rate": 3.2553473712927185e-06,
      "loss": 0.3252,
      "step": 40609
    },
    {
      "epoch": 1.8636134183837365,
      "grad_norm": 0.47536611557006836,
      "learning_rate": 3.2551175958879556e-06,
      "loss": 0.3495,
      "step": 40610
    },
    {
      "epoch": 1.863659308888991,
      "grad_norm": 0.4794268310070038,
      "learning_rate": 3.254887824679084e-06,
      "loss": 0.3242,
      "step": 40611
    },
    {
      "epoch": 1.8637051993942453,
      "grad_norm": 0.45534518361091614,
      "learning_rate": 3.254658057666654e-06,
      "loss": 0.3257,
      "step": 40612
    },
    {
      "epoch": 1.8637510898994998,
      "grad_norm": 0.5086663961410522,
      "learning_rate": 3.254428294851223e-06,
      "loss": 0.4096,
      "step": 40613
    },
    {
      "epoch": 1.8637969804047543,
      "grad_norm": 0.4981512725353241,
      "learning_rate": 3.254198536233341e-06,
      "loss": 0.3848,
      "step": 40614
    },
    {
      "epoch": 1.8638428709100086,
      "grad_norm": 0.46958738565444946,
      "learning_rate": 3.253968781813558e-06,
      "loss": 0.3654,
      "step": 40615
    },
    {
      "epoch": 1.863888761415263,
      "grad_norm": 0.4885769188404083,
      "learning_rate": 3.253739031592431e-06,
      "loss": 0.3853,
      "step": 40616
    },
    {
      "epoch": 1.8639346519205175,
      "grad_norm": 0.5148939490318298,
      "learning_rate": 3.253509285570511e-06,
      "loss": 0.3992,
      "step": 40617
    },
    {
      "epoch": 1.863980542425772,
      "grad_norm": 0.4883454442024231,
      "learning_rate": 3.253279543748349e-06,
      "loss": 0.3275,
      "step": 40618
    },
    {
      "epoch": 1.8640264329310265,
      "grad_norm": 0.4816766381263733,
      "learning_rate": 3.2530498061264994e-06,
      "loss": 0.3657,
      "step": 40619
    },
    {
      "epoch": 1.864072323436281,
      "grad_norm": 0.46653613448143005,
      "learning_rate": 3.2528200727055128e-06,
      "loss": 0.3817,
      "step": 40620
    },
    {
      "epoch": 1.8641182139415355,
      "grad_norm": 0.44647395610809326,
      "learning_rate": 3.2525903434859413e-06,
      "loss": 0.3254,
      "step": 40621
    },
    {
      "epoch": 1.86416410444679,
      "grad_norm": 0.4859084486961365,
      "learning_rate": 3.25236061846834e-06,
      "loss": 0.4367,
      "step": 40622
    },
    {
      "epoch": 1.8642099949520445,
      "grad_norm": 0.48087742924690247,
      "learning_rate": 3.2521308976532605e-06,
      "loss": 0.393,
      "step": 40623
    },
    {
      "epoch": 1.864255885457299,
      "grad_norm": 0.5647234320640564,
      "learning_rate": 3.251901181041253e-06,
      "loss": 0.4576,
      "step": 40624
    },
    {
      "epoch": 1.8643017759625533,
      "grad_norm": 0.4743705987930298,
      "learning_rate": 3.2516714686328725e-06,
      "loss": 0.377,
      "step": 40625
    },
    {
      "epoch": 1.8643476664678078,
      "grad_norm": 0.46758297085762024,
      "learning_rate": 3.251441760428671e-06,
      "loss": 0.3771,
      "step": 40626
    },
    {
      "epoch": 1.8643935569730623,
      "grad_norm": 0.4738147258758545,
      "learning_rate": 3.251212056429197e-06,
      "loss": 0.3869,
      "step": 40627
    },
    {
      "epoch": 1.8644394474783166,
      "grad_norm": 0.4917566776275635,
      "learning_rate": 3.250982356635009e-06,
      "loss": 0.4008,
      "step": 40628
    },
    {
      "epoch": 1.864485337983571,
      "grad_norm": 0.4821847379207611,
      "learning_rate": 3.250752661046657e-06,
      "loss": 0.3543,
      "step": 40629
    },
    {
      "epoch": 1.8645312284888256,
      "grad_norm": 0.4127359688282013,
      "learning_rate": 3.25052296966469e-06,
      "loss": 0.2379,
      "step": 40630
    },
    {
      "epoch": 1.86457711899408,
      "grad_norm": 0.44102901220321655,
      "learning_rate": 3.250293282489665e-06,
      "loss": 0.2933,
      "step": 40631
    },
    {
      "epoch": 1.8646230094993346,
      "grad_norm": 0.4247455894947052,
      "learning_rate": 3.250063599522133e-06,
      "loss": 0.2902,
      "step": 40632
    },
    {
      "epoch": 1.864668900004589,
      "grad_norm": 0.4821682274341583,
      "learning_rate": 3.249833920762644e-06,
      "loss": 0.371,
      "step": 40633
    },
    {
      "epoch": 1.8647147905098436,
      "grad_norm": 0.5072263479232788,
      "learning_rate": 3.2496042462117533e-06,
      "loss": 0.3563,
      "step": 40634
    },
    {
      "epoch": 1.864760681015098,
      "grad_norm": 0.4682324230670929,
      "learning_rate": 3.2493745758700116e-06,
      "loss": 0.3776,
      "step": 40635
    },
    {
      "epoch": 1.8648065715203526,
      "grad_norm": 0.47768300771713257,
      "learning_rate": 3.2491449097379725e-06,
      "loss": 0.4293,
      "step": 40636
    },
    {
      "epoch": 1.864852462025607,
      "grad_norm": 0.49399155378341675,
      "learning_rate": 3.2489152478161843e-06,
      "loss": 0.3825,
      "step": 40637
    },
    {
      "epoch": 1.8648983525308613,
      "grad_norm": 0.48155924677848816,
      "learning_rate": 3.248685590105205e-06,
      "loss": 0.3794,
      "step": 40638
    },
    {
      "epoch": 1.8649442430361158,
      "grad_norm": 0.43247830867767334,
      "learning_rate": 3.2484559366055834e-06,
      "loss": 0.3257,
      "step": 40639
    },
    {
      "epoch": 1.8649901335413703,
      "grad_norm": 0.4682311415672302,
      "learning_rate": 3.2482262873178714e-06,
      "loss": 0.3321,
      "step": 40640
    },
    {
      "epoch": 1.8650360240466246,
      "grad_norm": 0.5370655655860901,
      "learning_rate": 3.247996642242624e-06,
      "loss": 0.3883,
      "step": 40641
    },
    {
      "epoch": 1.865081914551879,
      "grad_norm": 0.48995935916900635,
      "learning_rate": 3.247767001380392e-06,
      "loss": 0.4086,
      "step": 40642
    },
    {
      "epoch": 1.8651278050571336,
      "grad_norm": 0.4899439811706543,
      "learning_rate": 3.2475373647317245e-06,
      "loss": 0.3775,
      "step": 40643
    },
    {
      "epoch": 1.865173695562388,
      "grad_norm": 0.4759605824947357,
      "learning_rate": 3.2473077322971793e-06,
      "loss": 0.3963,
      "step": 40644
    },
    {
      "epoch": 1.8652195860676426,
      "grad_norm": 0.4989371597766876,
      "learning_rate": 3.247078104077305e-06,
      "loss": 0.4227,
      "step": 40645
    },
    {
      "epoch": 1.865265476572897,
      "grad_norm": 0.4951787292957306,
      "learning_rate": 3.246848480072654e-06,
      "loss": 0.3427,
      "step": 40646
    },
    {
      "epoch": 1.8653113670781516,
      "grad_norm": 0.5179735422134399,
      "learning_rate": 3.24661886028378e-06,
      "loss": 0.3529,
      "step": 40647
    },
    {
      "epoch": 1.865357257583406,
      "grad_norm": 0.42660200595855713,
      "learning_rate": 3.2463892447112344e-06,
      "loss": 0.2982,
      "step": 40648
    },
    {
      "epoch": 1.8654031480886606,
      "grad_norm": 0.4724346995353699,
      "learning_rate": 3.2461596333555666e-06,
      "loss": 0.3793,
      "step": 40649
    },
    {
      "epoch": 1.8654490385939149,
      "grad_norm": 0.4424279034137726,
      "learning_rate": 3.2459300262173342e-06,
      "loss": 0.3076,
      "step": 40650
    },
    {
      "epoch": 1.8654949290991694,
      "grad_norm": 0.4944688677787781,
      "learning_rate": 3.2457004232970875e-06,
      "loss": 0.3178,
      "step": 40651
    },
    {
      "epoch": 1.8655408196044239,
      "grad_norm": 0.4386030435562134,
      "learning_rate": 3.245470824595374e-06,
      "loss": 0.3235,
      "step": 40652
    },
    {
      "epoch": 1.8655867101096784,
      "grad_norm": 0.4818091094493866,
      "learning_rate": 3.245241230112752e-06,
      "loss": 0.3868,
      "step": 40653
    },
    {
      "epoch": 1.8656326006149326,
      "grad_norm": 0.508457362651825,
      "learning_rate": 3.2450116398497712e-06,
      "loss": 0.4103,
      "step": 40654
    },
    {
      "epoch": 1.8656784911201871,
      "grad_norm": 0.4548834562301636,
      "learning_rate": 3.2447820538069817e-06,
      "loss": 0.3171,
      "step": 40655
    },
    {
      "epoch": 1.8657243816254416,
      "grad_norm": 0.47062554955482483,
      "learning_rate": 3.2445524719849387e-06,
      "loss": 0.3422,
      "step": 40656
    },
    {
      "epoch": 1.8657702721306961,
      "grad_norm": 0.4630744159221649,
      "learning_rate": 3.2443228943841932e-06,
      "loss": 0.3536,
      "step": 40657
    },
    {
      "epoch": 1.8658161626359506,
      "grad_norm": 0.47432541847229004,
      "learning_rate": 3.244093321005295e-06,
      "loss": 0.3958,
      "step": 40658
    },
    {
      "epoch": 1.8658620531412051,
      "grad_norm": 0.4837040603160858,
      "learning_rate": 3.2438637518487997e-06,
      "loss": 0.3682,
      "step": 40659
    },
    {
      "epoch": 1.8659079436464596,
      "grad_norm": 0.45332592725753784,
      "learning_rate": 3.2436341869152584e-06,
      "loss": 0.3025,
      "step": 40660
    },
    {
      "epoch": 1.8659538341517141,
      "grad_norm": 0.4613487422466278,
      "learning_rate": 3.243404626205221e-06,
      "loss": 0.3473,
      "step": 40661
    },
    {
      "epoch": 1.8659997246569686,
      "grad_norm": 0.5561777353286743,
      "learning_rate": 3.2431750697192426e-06,
      "loss": 0.5404,
      "step": 40662
    },
    {
      "epoch": 1.866045615162223,
      "grad_norm": 0.5318770408630371,
      "learning_rate": 3.242945517457874e-06,
      "loss": 0.4351,
      "step": 40663
    },
    {
      "epoch": 1.8660915056674774,
      "grad_norm": 0.5948088765144348,
      "learning_rate": 3.2427159694216637e-06,
      "loss": 0.4462,
      "step": 40664
    },
    {
      "epoch": 1.8661373961727319,
      "grad_norm": 0.4870043992996216,
      "learning_rate": 3.24248642561117e-06,
      "loss": 0.3482,
      "step": 40665
    },
    {
      "epoch": 1.8661832866779862,
      "grad_norm": 0.4509469270706177,
      "learning_rate": 3.242256886026941e-06,
      "loss": 0.3209,
      "step": 40666
    },
    {
      "epoch": 1.8662291771832407,
      "grad_norm": 0.4758235216140747,
      "learning_rate": 3.24202735066953e-06,
      "loss": 0.3175,
      "step": 40667
    },
    {
      "epoch": 1.8662750676884952,
      "grad_norm": 0.4881424605846405,
      "learning_rate": 3.2417978195394874e-06,
      "loss": 0.3469,
      "step": 40668
    },
    {
      "epoch": 1.8663209581937497,
      "grad_norm": 0.4640918970108032,
      "learning_rate": 3.2415682926373667e-06,
      "loss": 0.3168,
      "step": 40669
    },
    {
      "epoch": 1.8663668486990042,
      "grad_norm": 0.48344627022743225,
      "learning_rate": 3.241338769963719e-06,
      "loss": 0.3809,
      "step": 40670
    },
    {
      "epoch": 1.8664127392042587,
      "grad_norm": 0.4823472499847412,
      "learning_rate": 3.2411092515190952e-06,
      "loss": 0.3963,
      "step": 40671
    },
    {
      "epoch": 1.8664586297095132,
      "grad_norm": 0.48238396644592285,
      "learning_rate": 3.24087973730405e-06,
      "loss": 0.3965,
      "step": 40672
    },
    {
      "epoch": 1.8665045202147676,
      "grad_norm": 0.481507271528244,
      "learning_rate": 3.2406502273191355e-06,
      "loss": 0.3515,
      "step": 40673
    },
    {
      "epoch": 1.8665504107200221,
      "grad_norm": 0.4900887906551361,
      "learning_rate": 3.2404207215648986e-06,
      "loss": 0.415,
      "step": 40674
    },
    {
      "epoch": 1.8665963012252766,
      "grad_norm": 0.4664182662963867,
      "learning_rate": 3.240191220041896e-06,
      "loss": 0.3403,
      "step": 40675
    },
    {
      "epoch": 1.866642191730531,
      "grad_norm": 0.46869534254074097,
      "learning_rate": 3.2399617227506793e-06,
      "loss": 0.3513,
      "step": 40676
    },
    {
      "epoch": 1.8666880822357854,
      "grad_norm": 0.4409872889518738,
      "learning_rate": 3.2397322296917972e-06,
      "loss": 0.2962,
      "step": 40677
    },
    {
      "epoch": 1.86673397274104,
      "grad_norm": 0.4614500403404236,
      "learning_rate": 3.2395027408658047e-06,
      "loss": 0.3405,
      "step": 40678
    },
    {
      "epoch": 1.8667798632462942,
      "grad_norm": 0.4410739243030548,
      "learning_rate": 3.2392732562732533e-06,
      "loss": 0.3003,
      "step": 40679
    },
    {
      "epoch": 1.8668257537515487,
      "grad_norm": 0.4981977045536041,
      "learning_rate": 3.239043775914691e-06,
      "loss": 0.4197,
      "step": 40680
    },
    {
      "epoch": 1.8668716442568032,
      "grad_norm": 0.45733121037483215,
      "learning_rate": 3.2388142997906745e-06,
      "loss": 0.317,
      "step": 40681
    },
    {
      "epoch": 1.8669175347620577,
      "grad_norm": 0.4625927209854126,
      "learning_rate": 3.2385848279017546e-06,
      "loss": 0.3599,
      "step": 40682
    },
    {
      "epoch": 1.8669634252673122,
      "grad_norm": 0.4650525152683258,
      "learning_rate": 3.2383553602484806e-06,
      "loss": 0.345,
      "step": 40683
    },
    {
      "epoch": 1.8670093157725667,
      "grad_norm": 0.45474961400032043,
      "learning_rate": 3.2381258968314076e-06,
      "loss": 0.3273,
      "step": 40684
    },
    {
      "epoch": 1.8670552062778212,
      "grad_norm": 0.4604373872280121,
      "learning_rate": 3.237896437651085e-06,
      "loss": 0.3825,
      "step": 40685
    },
    {
      "epoch": 1.8671010967830757,
      "grad_norm": 0.48153409361839294,
      "learning_rate": 3.2376669827080643e-06,
      "loss": 0.3252,
      "step": 40686
    },
    {
      "epoch": 1.8671469872883302,
      "grad_norm": 0.454925537109375,
      "learning_rate": 3.2374375320028993e-06,
      "loss": 0.3119,
      "step": 40687
    },
    {
      "epoch": 1.8671928777935847,
      "grad_norm": 0.4897676706314087,
      "learning_rate": 3.237208085536141e-06,
      "loss": 0.3642,
      "step": 40688
    },
    {
      "epoch": 1.867238768298839,
      "grad_norm": 0.4363034963607788,
      "learning_rate": 3.2369786433083406e-06,
      "loss": 0.2901,
      "step": 40689
    },
    {
      "epoch": 1.8672846588040934,
      "grad_norm": 0.4592253267765045,
      "learning_rate": 3.23674920532005e-06,
      "loss": 0.3765,
      "step": 40690
    },
    {
      "epoch": 1.867330549309348,
      "grad_norm": 0.4582436978816986,
      "learning_rate": 3.2365197715718223e-06,
      "loss": 0.3372,
      "step": 40691
    },
    {
      "epoch": 1.8673764398146022,
      "grad_norm": 0.4628022611141205,
      "learning_rate": 3.236290342064205e-06,
      "loss": 0.3311,
      "step": 40692
    },
    {
      "epoch": 1.8674223303198567,
      "grad_norm": 0.4433578550815582,
      "learning_rate": 3.2360609167977555e-06,
      "loss": 0.335,
      "step": 40693
    },
    {
      "epoch": 1.8674682208251112,
      "grad_norm": 0.461154967546463,
      "learning_rate": 3.2358314957730218e-06,
      "loss": 0.3267,
      "step": 40694
    },
    {
      "epoch": 1.8675141113303657,
      "grad_norm": 0.5114935040473938,
      "learning_rate": 3.235602078990556e-06,
      "loss": 0.4301,
      "step": 40695
    },
    {
      "epoch": 1.8675600018356202,
      "grad_norm": 0.4680079221725464,
      "learning_rate": 3.235372666450912e-06,
      "loss": 0.3603,
      "step": 40696
    },
    {
      "epoch": 1.8676058923408747,
      "grad_norm": 0.3999522626399994,
      "learning_rate": 3.235143258154639e-06,
      "loss": 0.2853,
      "step": 40697
    },
    {
      "epoch": 1.8676517828461292,
      "grad_norm": 0.4605233669281006,
      "learning_rate": 3.2349138541022896e-06,
      "loss": 0.3368,
      "step": 40698
    },
    {
      "epoch": 1.8676976733513837,
      "grad_norm": 0.41906335949897766,
      "learning_rate": 3.2346844542944154e-06,
      "loss": 0.2735,
      "step": 40699
    },
    {
      "epoch": 1.8677435638566382,
      "grad_norm": 0.48962685465812683,
      "learning_rate": 3.2344550587315683e-06,
      "loss": 0.3979,
      "step": 40700
    },
    {
      "epoch": 1.8677894543618925,
      "grad_norm": 0.4788654148578644,
      "learning_rate": 3.234225667414299e-06,
      "loss": 0.3561,
      "step": 40701
    },
    {
      "epoch": 1.867835344867147,
      "grad_norm": 0.4790509343147278,
      "learning_rate": 3.233996280343158e-06,
      "loss": 0.3845,
      "step": 40702
    },
    {
      "epoch": 1.8678812353724015,
      "grad_norm": 0.46765217185020447,
      "learning_rate": 3.2337668975187005e-06,
      "loss": 0.3424,
      "step": 40703
    },
    {
      "epoch": 1.8679271258776557,
      "grad_norm": 0.4677450954914093,
      "learning_rate": 3.2335375189414763e-06,
      "loss": 0.323,
      "step": 40704
    },
    {
      "epoch": 1.8679730163829102,
      "grad_norm": 0.44587984681129456,
      "learning_rate": 3.233308144612036e-06,
      "loss": 0.2896,
      "step": 40705
    },
    {
      "epoch": 1.8680189068881647,
      "grad_norm": 0.4748303294181824,
      "learning_rate": 3.2330787745309324e-06,
      "loss": 0.2968,
      "step": 40706
    },
    {
      "epoch": 1.8680647973934192,
      "grad_norm": 0.4378405809402466,
      "learning_rate": 3.2328494086987173e-06,
      "loss": 0.2884,
      "step": 40707
    },
    {
      "epoch": 1.8681106878986737,
      "grad_norm": 0.46037033200263977,
      "learning_rate": 3.2326200471159395e-06,
      "loss": 0.3358,
      "step": 40708
    },
    {
      "epoch": 1.8681565784039282,
      "grad_norm": 0.4521106779575348,
      "learning_rate": 3.2323906897831546e-06,
      "loss": 0.3252,
      "step": 40709
    },
    {
      "epoch": 1.8682024689091827,
      "grad_norm": 0.49802979826927185,
      "learning_rate": 3.232161336700912e-06,
      "loss": 0.3849,
      "step": 40710
    },
    {
      "epoch": 1.8682483594144372,
      "grad_norm": 0.48265233635902405,
      "learning_rate": 3.231931987869763e-06,
      "loss": 0.3574,
      "step": 40711
    },
    {
      "epoch": 1.8682942499196917,
      "grad_norm": 0.47643181681632996,
      "learning_rate": 3.23170264329026e-06,
      "loss": 0.35,
      "step": 40712
    },
    {
      "epoch": 1.8683401404249462,
      "grad_norm": 0.46723902225494385,
      "learning_rate": 3.2314733029629543e-06,
      "loss": 0.3498,
      "step": 40713
    },
    {
      "epoch": 1.8683860309302005,
      "grad_norm": 0.5109745860099792,
      "learning_rate": 3.2312439668883954e-06,
      "loss": 0.4404,
      "step": 40714
    },
    {
      "epoch": 1.868431921435455,
      "grad_norm": 0.4574137330055237,
      "learning_rate": 3.231014635067138e-06,
      "loss": 0.3384,
      "step": 40715
    },
    {
      "epoch": 1.8684778119407095,
      "grad_norm": 0.5264354944229126,
      "learning_rate": 3.2307853074997327e-06,
      "loss": 0.4929,
      "step": 40716
    },
    {
      "epoch": 1.8685237024459638,
      "grad_norm": 0.4549841582775116,
      "learning_rate": 3.2305559841867284e-06,
      "loss": 0.317,
      "step": 40717
    },
    {
      "epoch": 1.8685695929512183,
      "grad_norm": 0.5012626051902771,
      "learning_rate": 3.2303266651286803e-06,
      "loss": 0.3953,
      "step": 40718
    },
    {
      "epoch": 1.8686154834564728,
      "grad_norm": 0.46170321106910706,
      "learning_rate": 3.230097350326138e-06,
      "loss": 0.3788,
      "step": 40719
    },
    {
      "epoch": 1.8686613739617273,
      "grad_norm": 0.48067739605903625,
      "learning_rate": 3.2298680397796523e-06,
      "loss": 0.4375,
      "step": 40720
    },
    {
      "epoch": 1.8687072644669818,
      "grad_norm": 0.4140966236591339,
      "learning_rate": 3.2296387334897754e-06,
      "loss": 0.2683,
      "step": 40721
    },
    {
      "epoch": 1.8687531549722363,
      "grad_norm": 0.4758281707763672,
      "learning_rate": 3.229409431457059e-06,
      "loss": 0.3765,
      "step": 40722
    },
    {
      "epoch": 1.8687990454774908,
      "grad_norm": 0.4974440336227417,
      "learning_rate": 3.2291801336820523e-06,
      "loss": 0.4189,
      "step": 40723
    },
    {
      "epoch": 1.8688449359827453,
      "grad_norm": 0.4582734704017639,
      "learning_rate": 3.2289508401653102e-06,
      "loss": 0.3569,
      "step": 40724
    },
    {
      "epoch": 1.8688908264879998,
      "grad_norm": 0.44212210178375244,
      "learning_rate": 3.228721550907383e-06,
      "loss": 0.3011,
      "step": 40725
    },
    {
      "epoch": 1.8689367169932543,
      "grad_norm": 0.46050137281417847,
      "learning_rate": 3.2284922659088192e-06,
      "loss": 0.3383,
      "step": 40726
    },
    {
      "epoch": 1.8689826074985085,
      "grad_norm": 0.46425989270210266,
      "learning_rate": 3.228262985170174e-06,
      "loss": 0.3399,
      "step": 40727
    },
    {
      "epoch": 1.869028498003763,
      "grad_norm": 0.4565785527229309,
      "learning_rate": 3.2280337086919977e-06,
      "loss": 0.3678,
      "step": 40728
    },
    {
      "epoch": 1.8690743885090175,
      "grad_norm": 0.46119794249534607,
      "learning_rate": 3.2278044364748384e-06,
      "loss": 0.3426,
      "step": 40729
    },
    {
      "epoch": 1.8691202790142718,
      "grad_norm": 0.4771638512611389,
      "learning_rate": 3.2275751685192523e-06,
      "loss": 0.3586,
      "step": 40730
    },
    {
      "epoch": 1.8691661695195263,
      "grad_norm": 0.4892936050891876,
      "learning_rate": 3.2273459048257894e-06,
      "loss": 0.3837,
      "step": 40731
    },
    {
      "epoch": 1.8692120600247808,
      "grad_norm": 0.43033915758132935,
      "learning_rate": 3.2271166453949987e-06,
      "loss": 0.3314,
      "step": 40732
    },
    {
      "epoch": 1.8692579505300353,
      "grad_norm": 0.44673073291778564,
      "learning_rate": 3.2268873902274328e-06,
      "loss": 0.2924,
      "step": 40733
    },
    {
      "epoch": 1.8693038410352898,
      "grad_norm": 0.4479477107524872,
      "learning_rate": 3.226658139323644e-06,
      "loss": 0.318,
      "step": 40734
    },
    {
      "epoch": 1.8693497315405443,
      "grad_norm": 0.5080916285514832,
      "learning_rate": 3.2264288926841823e-06,
      "loss": 0.355,
      "step": 40735
    },
    {
      "epoch": 1.8693956220457988,
      "grad_norm": 0.49496474862098694,
      "learning_rate": 3.2261996503095977e-06,
      "loss": 0.3956,
      "step": 40736
    },
    {
      "epoch": 1.8694415125510533,
      "grad_norm": 0.5329535007476807,
      "learning_rate": 3.225970412200445e-06,
      "loss": 0.4474,
      "step": 40737
    },
    {
      "epoch": 1.8694874030563078,
      "grad_norm": 0.46709516644477844,
      "learning_rate": 3.2257411783572734e-06,
      "loss": 0.3553,
      "step": 40738
    },
    {
      "epoch": 1.869533293561562,
      "grad_norm": 0.4768551290035248,
      "learning_rate": 3.225511948780634e-06,
      "loss": 0.3523,
      "step": 40739
    },
    {
      "epoch": 1.8695791840668166,
      "grad_norm": 0.4548031985759735,
      "learning_rate": 3.2252827234710786e-06,
      "loss": 0.3476,
      "step": 40740
    },
    {
      "epoch": 1.869625074572071,
      "grad_norm": 0.48026639223098755,
      "learning_rate": 3.2250535024291585e-06,
      "loss": 0.4007,
      "step": 40741
    },
    {
      "epoch": 1.8696709650773256,
      "grad_norm": 0.4627038538455963,
      "learning_rate": 3.2248242856554234e-06,
      "loss": 0.3371,
      "step": 40742
    },
    {
      "epoch": 1.8697168555825798,
      "grad_norm": 0.5128834247589111,
      "learning_rate": 3.2245950731504265e-06,
      "loss": 0.4841,
      "step": 40743
    },
    {
      "epoch": 1.8697627460878343,
      "grad_norm": 0.49921223521232605,
      "learning_rate": 3.224365864914718e-06,
      "loss": 0.3757,
      "step": 40744
    },
    {
      "epoch": 1.8698086365930888,
      "grad_norm": 0.49540871381759644,
      "learning_rate": 3.2241366609488478e-06,
      "loss": 0.3381,
      "step": 40745
    },
    {
      "epoch": 1.8698545270983433,
      "grad_norm": 0.5509725213050842,
      "learning_rate": 3.22390746125337e-06,
      "loss": 0.4604,
      "step": 40746
    },
    {
      "epoch": 1.8699004176035978,
      "grad_norm": 0.4361392557621002,
      "learning_rate": 3.2236782658288344e-06,
      "loss": 0.3221,
      "step": 40747
    },
    {
      "epoch": 1.8699463081088523,
      "grad_norm": 0.4969726800918579,
      "learning_rate": 3.2234490746757907e-06,
      "loss": 0.3893,
      "step": 40748
    },
    {
      "epoch": 1.8699921986141068,
      "grad_norm": 0.4852102994918823,
      "learning_rate": 3.2232198877947923e-06,
      "loss": 0.3675,
      "step": 40749
    },
    {
      "epoch": 1.8700380891193613,
      "grad_norm": 0.5214864611625671,
      "learning_rate": 3.2229907051863895e-06,
      "loss": 0.4187,
      "step": 40750
    },
    {
      "epoch": 1.8700839796246158,
      "grad_norm": 0.5014107823371887,
      "learning_rate": 3.2227615268511316e-06,
      "loss": 0.4053,
      "step": 40751
    },
    {
      "epoch": 1.87012987012987,
      "grad_norm": 0.5290302038192749,
      "learning_rate": 3.2225323527895725e-06,
      "loss": 0.3582,
      "step": 40752
    },
    {
      "epoch": 1.8701757606351246,
      "grad_norm": 0.43171751499176025,
      "learning_rate": 3.2223031830022623e-06,
      "loss": 0.3452,
      "step": 40753
    },
    {
      "epoch": 1.870221651140379,
      "grad_norm": 0.5005162954330444,
      "learning_rate": 3.222074017489751e-06,
      "loss": 0.3972,
      "step": 40754
    },
    {
      "epoch": 1.8702675416456334,
      "grad_norm": 0.4486680030822754,
      "learning_rate": 3.221844856252592e-06,
      "loss": 0.3196,
      "step": 40755
    },
    {
      "epoch": 1.8703134321508879,
      "grad_norm": 0.5238682627677917,
      "learning_rate": 3.2216156992913337e-06,
      "loss": 0.4265,
      "step": 40756
    },
    {
      "epoch": 1.8703593226561424,
      "grad_norm": 0.4576411545276642,
      "learning_rate": 3.2213865466065276e-06,
      "loss": 0.3246,
      "step": 40757
    },
    {
      "epoch": 1.8704052131613969,
      "grad_norm": 0.4694410264492035,
      "learning_rate": 3.221157398198727e-06,
      "loss": 0.3393,
      "step": 40758
    },
    {
      "epoch": 1.8704511036666513,
      "grad_norm": 0.4602578580379486,
      "learning_rate": 3.2209282540684815e-06,
      "loss": 0.3472,
      "step": 40759
    },
    {
      "epoch": 1.8704969941719058,
      "grad_norm": 0.498900830745697,
      "learning_rate": 3.2206991142163414e-06,
      "loss": 0.3917,
      "step": 40760
    },
    {
      "epoch": 1.8705428846771603,
      "grad_norm": 0.5118532180786133,
      "learning_rate": 3.220469978642859e-06,
      "loss": 0.3872,
      "step": 40761
    },
    {
      "epoch": 1.8705887751824148,
      "grad_norm": 0.4716373383998871,
      "learning_rate": 3.2202408473485847e-06,
      "loss": 0.3007,
      "step": 40762
    },
    {
      "epoch": 1.8706346656876693,
      "grad_norm": 0.4739513099193573,
      "learning_rate": 3.2200117203340675e-06,
      "loss": 0.3495,
      "step": 40763
    },
    {
      "epoch": 1.8706805561929238,
      "grad_norm": 0.4786577820777893,
      "learning_rate": 3.219782597599863e-06,
      "loss": 0.3826,
      "step": 40764
    },
    {
      "epoch": 1.8707264466981781,
      "grad_norm": 0.4648756980895996,
      "learning_rate": 3.219553479146519e-06,
      "loss": 0.3167,
      "step": 40765
    },
    {
      "epoch": 1.8707723372034326,
      "grad_norm": 0.4703684151172638,
      "learning_rate": 3.2193243649745865e-06,
      "loss": 0.354,
      "step": 40766
    },
    {
      "epoch": 1.870818227708687,
      "grad_norm": 0.4296219050884247,
      "learning_rate": 3.2190952550846157e-06,
      "loss": 0.3106,
      "step": 40767
    },
    {
      "epoch": 1.8708641182139414,
      "grad_norm": 0.4723515808582306,
      "learning_rate": 3.21886614947716e-06,
      "loss": 0.3527,
      "step": 40768
    },
    {
      "epoch": 1.8709100087191959,
      "grad_norm": 0.49085506796836853,
      "learning_rate": 3.2186370481527697e-06,
      "loss": 0.376,
      "step": 40769
    },
    {
      "epoch": 1.8709558992244504,
      "grad_norm": 0.4819372594356537,
      "learning_rate": 3.218407951111993e-06,
      "loss": 0.3669,
      "step": 40770
    },
    {
      "epoch": 1.8710017897297049,
      "grad_norm": 0.45792436599731445,
      "learning_rate": 3.2181788583553847e-06,
      "loss": 0.3404,
      "step": 40771
    },
    {
      "epoch": 1.8710476802349594,
      "grad_norm": 0.45170557498931885,
      "learning_rate": 3.2179497698834936e-06,
      "loss": 0.3165,
      "step": 40772
    },
    {
      "epoch": 1.8710935707402139,
      "grad_norm": 0.4720611274242401,
      "learning_rate": 3.217720685696869e-06,
      "loss": 0.3269,
      "step": 40773
    },
    {
      "epoch": 1.8711394612454684,
      "grad_norm": 0.5174765586853027,
      "learning_rate": 3.2174916057960657e-06,
      "loss": 0.431,
      "step": 40774
    },
    {
      "epoch": 1.8711853517507229,
      "grad_norm": 0.43505552411079407,
      "learning_rate": 3.217262530181633e-06,
      "loss": 0.2947,
      "step": 40775
    },
    {
      "epoch": 1.8712312422559774,
      "grad_norm": 0.49187368154525757,
      "learning_rate": 3.2170334588541195e-06,
      "loss": 0.397,
      "step": 40776
    },
    {
      "epoch": 1.8712771327612319,
      "grad_norm": 0.45217183232307434,
      "learning_rate": 3.216804391814079e-06,
      "loss": 0.3569,
      "step": 40777
    },
    {
      "epoch": 1.8713230232664861,
      "grad_norm": 0.515798807144165,
      "learning_rate": 3.216575329062061e-06,
      "loss": 0.3655,
      "step": 40778
    },
    {
      "epoch": 1.8713689137717406,
      "grad_norm": 0.48925110697746277,
      "learning_rate": 3.2163462705986146e-06,
      "loss": 0.3839,
      "step": 40779
    },
    {
      "epoch": 1.8714148042769951,
      "grad_norm": 0.4585813283920288,
      "learning_rate": 3.2161172164242947e-06,
      "loss": 0.3579,
      "step": 40780
    },
    {
      "epoch": 1.8714606947822494,
      "grad_norm": 0.4877050817012787,
      "learning_rate": 3.21588816653965e-06,
      "loss": 0.4291,
      "step": 40781
    },
    {
      "epoch": 1.871506585287504,
      "grad_norm": 0.45397013425827026,
      "learning_rate": 3.2156591209452294e-06,
      "loss": 0.3347,
      "step": 40782
    },
    {
      "epoch": 1.8715524757927584,
      "grad_norm": 0.470426470041275,
      "learning_rate": 3.2154300796415873e-06,
      "loss": 0.3607,
      "step": 40783
    },
    {
      "epoch": 1.871598366298013,
      "grad_norm": 0.4714231789112091,
      "learning_rate": 3.2152010426292725e-06,
      "loss": 0.3864,
      "step": 40784
    },
    {
      "epoch": 1.8716442568032674,
      "grad_norm": 0.49862444400787354,
      "learning_rate": 3.2149720099088335e-06,
      "loss": 0.3903,
      "step": 40785
    },
    {
      "epoch": 1.871690147308522,
      "grad_norm": 0.48576605319976807,
      "learning_rate": 3.214742981480826e-06,
      "loss": 0.341,
      "step": 40786
    },
    {
      "epoch": 1.8717360378137764,
      "grad_norm": 0.5064970254898071,
      "learning_rate": 3.2145139573457982e-06,
      "loss": 0.4283,
      "step": 40787
    },
    {
      "epoch": 1.871781928319031,
      "grad_norm": 0.46467825770378113,
      "learning_rate": 3.214284937504298e-06,
      "loss": 0.3244,
      "step": 40788
    },
    {
      "epoch": 1.8718278188242854,
      "grad_norm": 0.46187037229537964,
      "learning_rate": 3.2140559219568816e-06,
      "loss": 0.3546,
      "step": 40789
    },
    {
      "epoch": 1.8718737093295397,
      "grad_norm": 0.46945130825042725,
      "learning_rate": 3.213826910704097e-06,
      "loss": 0.365,
      "step": 40790
    },
    {
      "epoch": 1.8719195998347942,
      "grad_norm": 0.44744163751602173,
      "learning_rate": 3.2135979037464937e-06,
      "loss": 0.335,
      "step": 40791
    },
    {
      "epoch": 1.8719654903400487,
      "grad_norm": 0.46984878182411194,
      "learning_rate": 3.2133689010846247e-06,
      "loss": 0.3903,
      "step": 40792
    },
    {
      "epoch": 1.872011380845303,
      "grad_norm": 0.491263747215271,
      "learning_rate": 3.2131399027190402e-06,
      "loss": 0.4119,
      "step": 40793
    },
    {
      "epoch": 1.8720572713505574,
      "grad_norm": 0.49642932415008545,
      "learning_rate": 3.2129109086502875e-06,
      "loss": 0.4466,
      "step": 40794
    },
    {
      "epoch": 1.872103161855812,
      "grad_norm": 0.4380612373352051,
      "learning_rate": 3.212681918878922e-06,
      "loss": 0.2798,
      "step": 40795
    },
    {
      "epoch": 1.8721490523610664,
      "grad_norm": 0.49369558691978455,
      "learning_rate": 3.212452933405493e-06,
      "loss": 0.4024,
      "step": 40796
    },
    {
      "epoch": 1.872194942866321,
      "grad_norm": 0.4712003469467163,
      "learning_rate": 3.2122239522305497e-06,
      "loss": 0.3895,
      "step": 40797
    },
    {
      "epoch": 1.8722408333715754,
      "grad_norm": 0.5141485333442688,
      "learning_rate": 3.211994975354643e-06,
      "loss": 0.4379,
      "step": 40798
    },
    {
      "epoch": 1.87228672387683,
      "grad_norm": 0.47650057077407837,
      "learning_rate": 3.2117660027783255e-06,
      "loss": 0.4034,
      "step": 40799
    },
    {
      "epoch": 1.8723326143820844,
      "grad_norm": 0.48307308554649353,
      "learning_rate": 3.2115370345021457e-06,
      "loss": 0.3513,
      "step": 40800
    },
    {
      "epoch": 1.872378504887339,
      "grad_norm": 0.4398488998413086,
      "learning_rate": 3.2113080705266532e-06,
      "loss": 0.3352,
      "step": 40801
    },
    {
      "epoch": 1.8724243953925934,
      "grad_norm": 0.43941035866737366,
      "learning_rate": 3.211079110852402e-06,
      "loss": 0.3154,
      "step": 40802
    },
    {
      "epoch": 1.8724702858978477,
      "grad_norm": 0.4663524627685547,
      "learning_rate": 3.2108501554799413e-06,
      "loss": 0.3112,
      "step": 40803
    },
    {
      "epoch": 1.8725161764031022,
      "grad_norm": 0.5122995972633362,
      "learning_rate": 3.2106212044098204e-06,
      "loss": 0.3462,
      "step": 40804
    },
    {
      "epoch": 1.8725620669083567,
      "grad_norm": 0.4860347509384155,
      "learning_rate": 3.2103922576425918e-06,
      "loss": 0.3854,
      "step": 40805
    },
    {
      "epoch": 1.872607957413611,
      "grad_norm": 0.4439956843852997,
      "learning_rate": 3.2101633151788046e-06,
      "loss": 0.3431,
      "step": 40806
    },
    {
      "epoch": 1.8726538479188655,
      "grad_norm": 0.4780729413032532,
      "learning_rate": 3.209934377019007e-06,
      "loss": 0.3713,
      "step": 40807
    },
    {
      "epoch": 1.87269973842412,
      "grad_norm": 0.42868635058403015,
      "learning_rate": 3.2097054431637563e-06,
      "loss": 0.2993,
      "step": 40808
    },
    {
      "epoch": 1.8727456289293745,
      "grad_norm": 0.4646168053150177,
      "learning_rate": 3.2094765136135985e-06,
      "loss": 0.402,
      "step": 40809
    },
    {
      "epoch": 1.872791519434629,
      "grad_norm": 0.48864880204200745,
      "learning_rate": 3.2092475883690814e-06,
      "loss": 0.3893,
      "step": 40810
    },
    {
      "epoch": 1.8728374099398835,
      "grad_norm": 0.46659091114997864,
      "learning_rate": 3.209018667430761e-06,
      "loss": 0.3604,
      "step": 40811
    },
    {
      "epoch": 1.872883300445138,
      "grad_norm": 0.4875173270702362,
      "learning_rate": 3.208789750799186e-06,
      "loss": 0.3908,
      "step": 40812
    },
    {
      "epoch": 1.8729291909503925,
      "grad_norm": 0.46357446908950806,
      "learning_rate": 3.2085608384749052e-06,
      "loss": 0.3046,
      "step": 40813
    },
    {
      "epoch": 1.872975081455647,
      "grad_norm": 0.45003461837768555,
      "learning_rate": 3.2083319304584704e-06,
      "loss": 0.3257,
      "step": 40814
    },
    {
      "epoch": 1.8730209719609014,
      "grad_norm": 0.48325875401496887,
      "learning_rate": 3.2081030267504327e-06,
      "loss": 0.3697,
      "step": 40815
    },
    {
      "epoch": 1.8730668624661557,
      "grad_norm": 0.456366628408432,
      "learning_rate": 3.207874127351339e-06,
      "loss": 0.3577,
      "step": 40816
    },
    {
      "epoch": 1.8731127529714102,
      "grad_norm": 0.4352400600910187,
      "learning_rate": 3.2076452322617452e-06,
      "loss": 0.2966,
      "step": 40817
    },
    {
      "epoch": 1.8731586434766647,
      "grad_norm": 0.4380384385585785,
      "learning_rate": 3.2074163414821984e-06,
      "loss": 0.3234,
      "step": 40818
    },
    {
      "epoch": 1.873204533981919,
      "grad_norm": 0.45777013897895813,
      "learning_rate": 3.2071874550132488e-06,
      "loss": 0.3723,
      "step": 40819
    },
    {
      "epoch": 1.8732504244871735,
      "grad_norm": 0.4996108114719391,
      "learning_rate": 3.2069585728554482e-06,
      "loss": 0.3907,
      "step": 40820
    },
    {
      "epoch": 1.873296314992428,
      "grad_norm": 0.4152979552745819,
      "learning_rate": 3.206729695009346e-06,
      "loss": 0.2805,
      "step": 40821
    },
    {
      "epoch": 1.8733422054976825,
      "grad_norm": 0.46005675196647644,
      "learning_rate": 3.206500821475492e-06,
      "loss": 0.3365,
      "step": 40822
    },
    {
      "epoch": 1.873388096002937,
      "grad_norm": 0.453836053609848,
      "learning_rate": 3.206271952254439e-06,
      "loss": 0.3421,
      "step": 40823
    },
    {
      "epoch": 1.8734339865081915,
      "grad_norm": 0.445949912071228,
      "learning_rate": 3.206043087346735e-06,
      "loss": 0.2786,
      "step": 40824
    },
    {
      "epoch": 1.873479877013446,
      "grad_norm": 0.49423670768737793,
      "learning_rate": 3.2058142267529314e-06,
      "loss": 0.388,
      "step": 40825
    },
    {
      "epoch": 1.8735257675187005,
      "grad_norm": 0.5177964568138123,
      "learning_rate": 3.205585370473579e-06,
      "loss": 0.4089,
      "step": 40826
    },
    {
      "epoch": 1.873571658023955,
      "grad_norm": 0.5953711867332458,
      "learning_rate": 3.205356518509227e-06,
      "loss": 0.3317,
      "step": 40827
    },
    {
      "epoch": 1.8736175485292093,
      "grad_norm": 0.4540679156780243,
      "learning_rate": 3.2051276708604244e-06,
      "loss": 0.3396,
      "step": 40828
    },
    {
      "epoch": 1.8736634390344638,
      "grad_norm": 0.4874948263168335,
      "learning_rate": 3.2048988275277258e-06,
      "loss": 0.3846,
      "step": 40829
    },
    {
      "epoch": 1.8737093295397182,
      "grad_norm": 0.47497689723968506,
      "learning_rate": 3.2046699885116794e-06,
      "loss": 0.4012,
      "step": 40830
    },
    {
      "epoch": 1.8737552200449727,
      "grad_norm": 0.4538644552230835,
      "learning_rate": 3.204441153812834e-06,
      "loss": 0.3195,
      "step": 40831
    },
    {
      "epoch": 1.873801110550227,
      "grad_norm": 0.43262597918510437,
      "learning_rate": 3.204212323431739e-06,
      "loss": 0.2884,
      "step": 40832
    },
    {
      "epoch": 1.8738470010554815,
      "grad_norm": 0.4953249394893646,
      "learning_rate": 3.203983497368949e-06,
      "loss": 0.4047,
      "step": 40833
    },
    {
      "epoch": 1.873892891560736,
      "grad_norm": 0.47787341475486755,
      "learning_rate": 3.2037546756250114e-06,
      "loss": 0.4127,
      "step": 40834
    },
    {
      "epoch": 1.8739387820659905,
      "grad_norm": 0.462568074464798,
      "learning_rate": 3.2035258582004757e-06,
      "loss": 0.3298,
      "step": 40835
    },
    {
      "epoch": 1.873984672571245,
      "grad_norm": 0.45099908113479614,
      "learning_rate": 3.2032970450958952e-06,
      "loss": 0.3368,
      "step": 40836
    },
    {
      "epoch": 1.8740305630764995,
      "grad_norm": 0.45052874088287354,
      "learning_rate": 3.2030682363118177e-06,
      "loss": 0.3168,
      "step": 40837
    },
    {
      "epoch": 1.874076453581754,
      "grad_norm": 0.49139031767845154,
      "learning_rate": 3.2028394318487923e-06,
      "loss": 0.3865,
      "step": 40838
    },
    {
      "epoch": 1.8741223440870085,
      "grad_norm": 0.4749772548675537,
      "learning_rate": 3.2026106317073722e-06,
      "loss": 0.3881,
      "step": 40839
    },
    {
      "epoch": 1.874168234592263,
      "grad_norm": 0.43694987893104553,
      "learning_rate": 3.2023818358881063e-06,
      "loss": 0.3048,
      "step": 40840
    },
    {
      "epoch": 1.8742141250975173,
      "grad_norm": 0.48097196221351624,
      "learning_rate": 3.202153044391544e-06,
      "loss": 0.3467,
      "step": 40841
    },
    {
      "epoch": 1.8742600156027718,
      "grad_norm": 0.4816828966140747,
      "learning_rate": 3.2019242572182376e-06,
      "loss": 0.3548,
      "step": 40842
    },
    {
      "epoch": 1.8743059061080263,
      "grad_norm": 0.5016543865203857,
      "learning_rate": 3.2016954743687355e-06,
      "loss": 0.4121,
      "step": 40843
    },
    {
      "epoch": 1.8743517966132806,
      "grad_norm": 0.5262869596481323,
      "learning_rate": 3.2014666958435857e-06,
      "loss": 0.4247,
      "step": 40844
    },
    {
      "epoch": 1.874397687118535,
      "grad_norm": 0.5393515229225159,
      "learning_rate": 3.201237921643344e-06,
      "loss": 0.4274,
      "step": 40845
    },
    {
      "epoch": 1.8744435776237895,
      "grad_norm": 0.45059239864349365,
      "learning_rate": 3.2010091517685572e-06,
      "loss": 0.3048,
      "step": 40846
    },
    {
      "epoch": 1.874489468129044,
      "grad_norm": 0.49902456998825073,
      "learning_rate": 3.200780386219774e-06,
      "loss": 0.3616,
      "step": 40847
    },
    {
      "epoch": 1.8745353586342985,
      "grad_norm": 0.45046401023864746,
      "learning_rate": 3.200551624997547e-06,
      "loss": 0.3449,
      "step": 40848
    },
    {
      "epoch": 1.874581249139553,
      "grad_norm": 0.4514561891555786,
      "learning_rate": 3.200322868102426e-06,
      "loss": 0.3045,
      "step": 40849
    },
    {
      "epoch": 1.8746271396448075,
      "grad_norm": 0.5060063600540161,
      "learning_rate": 3.200094115534958e-06,
      "loss": 0.3806,
      "step": 40850
    },
    {
      "epoch": 1.874673030150062,
      "grad_norm": 0.5106275677680969,
      "learning_rate": 3.1998653672956983e-06,
      "loss": 0.4362,
      "step": 40851
    },
    {
      "epoch": 1.8747189206553165,
      "grad_norm": 0.4617348313331604,
      "learning_rate": 3.1996366233851944e-06,
      "loss": 0.3821,
      "step": 40852
    },
    {
      "epoch": 1.874764811160571,
      "grad_norm": 0.48427310585975647,
      "learning_rate": 3.1994078838039932e-06,
      "loss": 0.3649,
      "step": 40853
    },
    {
      "epoch": 1.8748107016658253,
      "grad_norm": 0.531186044216156,
      "learning_rate": 3.1991791485526504e-06,
      "loss": 0.3839,
      "step": 40854
    },
    {
      "epoch": 1.8748565921710798,
      "grad_norm": 0.473624050617218,
      "learning_rate": 3.198950417631713e-06,
      "loss": 0.3645,
      "step": 40855
    },
    {
      "epoch": 1.8749024826763343,
      "grad_norm": 0.43522363901138306,
      "learning_rate": 3.1987216910417308e-06,
      "loss": 0.2711,
      "step": 40856
    },
    {
      "epoch": 1.8749483731815886,
      "grad_norm": 0.4460555613040924,
      "learning_rate": 3.1984929687832555e-06,
      "loss": 0.2942,
      "step": 40857
    },
    {
      "epoch": 1.874994263686843,
      "grad_norm": 0.44664376974105835,
      "learning_rate": 3.1982642508568357e-06,
      "loss": 0.3197,
      "step": 40858
    },
    {
      "epoch": 1.8750401541920976,
      "grad_norm": 0.4478331506252289,
      "learning_rate": 3.19803553726302e-06,
      "loss": 0.2979,
      "step": 40859
    },
    {
      "epoch": 1.875086044697352,
      "grad_norm": 0.4893341660499573,
      "learning_rate": 3.197806828002362e-06,
      "loss": 0.3687,
      "step": 40860
    },
    {
      "epoch": 1.8751319352026066,
      "grad_norm": 0.501018226146698,
      "learning_rate": 3.1975781230754093e-06,
      "loss": 0.3704,
      "step": 40861
    },
    {
      "epoch": 1.875177825707861,
      "grad_norm": 0.4379594326019287,
      "learning_rate": 3.1973494224827116e-06,
      "loss": 0.2966,
      "step": 40862
    },
    {
      "epoch": 1.8752237162131156,
      "grad_norm": 0.46309545636177063,
      "learning_rate": 3.1971207262248206e-06,
      "loss": 0.3093,
      "step": 40863
    },
    {
      "epoch": 1.87526960671837,
      "grad_norm": 0.4987621307373047,
      "learning_rate": 3.1968920343022857e-06,
      "loss": 0.3722,
      "step": 40864
    },
    {
      "epoch": 1.8753154972236246,
      "grad_norm": 0.4991621673107147,
      "learning_rate": 3.196663346715656e-06,
      "loss": 0.3766,
      "step": 40865
    },
    {
      "epoch": 1.875361387728879,
      "grad_norm": 0.46523022651672363,
      "learning_rate": 3.19643466346548e-06,
      "loss": 0.3028,
      "step": 40866
    },
    {
      "epoch": 1.8754072782341333,
      "grad_norm": 0.48662468791007996,
      "learning_rate": 3.1962059845523113e-06,
      "loss": 0.3469,
      "step": 40867
    },
    {
      "epoch": 1.8754531687393878,
      "grad_norm": 0.47083377838134766,
      "learning_rate": 3.1959773099766977e-06,
      "loss": 0.3278,
      "step": 40868
    },
    {
      "epoch": 1.8754990592446423,
      "grad_norm": 0.4565489888191223,
      "learning_rate": 3.1957486397391892e-06,
      "loss": 0.356,
      "step": 40869
    },
    {
      "epoch": 1.8755449497498966,
      "grad_norm": 0.4829057455062866,
      "learning_rate": 3.195519973840335e-06,
      "loss": 0.3353,
      "step": 40870
    },
    {
      "epoch": 1.875590840255151,
      "grad_norm": 0.5008401274681091,
      "learning_rate": 3.195291312280687e-06,
      "loss": 0.3817,
      "step": 40871
    },
    {
      "epoch": 1.8756367307604056,
      "grad_norm": 0.5003997683525085,
      "learning_rate": 3.1950626550607923e-06,
      "loss": 0.4231,
      "step": 40872
    },
    {
      "epoch": 1.87568262126566,
      "grad_norm": 0.46256083250045776,
      "learning_rate": 3.194834002181203e-06,
      "loss": 0.347,
      "step": 40873
    },
    {
      "epoch": 1.8757285117709146,
      "grad_norm": 0.47234123945236206,
      "learning_rate": 3.1946053536424704e-06,
      "loss": 0.4095,
      "step": 40874
    },
    {
      "epoch": 1.875774402276169,
      "grad_norm": 0.4805777370929718,
      "learning_rate": 3.1943767094451382e-06,
      "loss": 0.3817,
      "step": 40875
    },
    {
      "epoch": 1.8758202927814236,
      "grad_norm": 0.43107733130455017,
      "learning_rate": 3.1941480695897626e-06,
      "loss": 0.2919,
      "step": 40876
    },
    {
      "epoch": 1.875866183286678,
      "grad_norm": 0.48459169268608093,
      "learning_rate": 3.1939194340768907e-06,
      "loss": 0.3633,
      "step": 40877
    },
    {
      "epoch": 1.8759120737919326,
      "grad_norm": 0.46466493606567383,
      "learning_rate": 3.193690802907072e-06,
      "loss": 0.3403,
      "step": 40878
    },
    {
      "epoch": 1.8759579642971869,
      "grad_norm": 0.4675995409488678,
      "learning_rate": 3.193462176080857e-06,
      "loss": 0.354,
      "step": 40879
    },
    {
      "epoch": 1.8760038548024414,
      "grad_norm": 0.4871419668197632,
      "learning_rate": 3.1932335535987962e-06,
      "loss": 0.3496,
      "step": 40880
    },
    {
      "epoch": 1.8760497453076959,
      "grad_norm": 0.47044265270233154,
      "learning_rate": 3.1930049354614356e-06,
      "loss": 0.3739,
      "step": 40881
    },
    {
      "epoch": 1.8760956358129501,
      "grad_norm": 0.4486694037914276,
      "learning_rate": 3.1927763216693307e-06,
      "loss": 0.3186,
      "step": 40882
    },
    {
      "epoch": 1.8761415263182046,
      "grad_norm": 0.48766204714775085,
      "learning_rate": 3.1925477122230282e-06,
      "loss": 0.3911,
      "step": 40883
    },
    {
      "epoch": 1.8761874168234591,
      "grad_norm": 0.44279924035072327,
      "learning_rate": 3.1923191071230763e-06,
      "loss": 0.283,
      "step": 40884
    },
    {
      "epoch": 1.8762333073287136,
      "grad_norm": 0.41964468359947205,
      "learning_rate": 3.192090506370028e-06,
      "loss": 0.2896,
      "step": 40885
    },
    {
      "epoch": 1.8762791978339681,
      "grad_norm": 0.477951318025589,
      "learning_rate": 3.1918619099644314e-06,
      "loss": 0.38,
      "step": 40886
    },
    {
      "epoch": 1.8763250883392226,
      "grad_norm": 0.46327531337738037,
      "learning_rate": 3.1916333179068344e-06,
      "loss": 0.3224,
      "step": 40887
    },
    {
      "epoch": 1.8763709788444771,
      "grad_norm": 0.4303705394268036,
      "learning_rate": 3.1914047301977903e-06,
      "loss": 0.2808,
      "step": 40888
    },
    {
      "epoch": 1.8764168693497316,
      "grad_norm": 0.5067707300186157,
      "learning_rate": 3.191176146837847e-06,
      "loss": 0.4242,
      "step": 40889
    },
    {
      "epoch": 1.8764627598549861,
      "grad_norm": 0.4907696545124054,
      "learning_rate": 3.190947567827554e-06,
      "loss": 0.3859,
      "step": 40890
    },
    {
      "epoch": 1.8765086503602406,
      "grad_norm": 0.6529475450515747,
      "learning_rate": 3.190718993167461e-06,
      "loss": 0.3242,
      "step": 40891
    },
    {
      "epoch": 1.876554540865495,
      "grad_norm": 0.48794758319854736,
      "learning_rate": 3.190490422858118e-06,
      "loss": 0.3517,
      "step": 40892
    },
    {
      "epoch": 1.8766004313707494,
      "grad_norm": 0.4901316165924072,
      "learning_rate": 3.190261856900072e-06,
      "loss": 0.3943,
      "step": 40893
    },
    {
      "epoch": 1.8766463218760039,
      "grad_norm": 0.45873603224754333,
      "learning_rate": 3.1900332952938782e-06,
      "loss": 0.3328,
      "step": 40894
    },
    {
      "epoch": 1.8766922123812582,
      "grad_norm": 0.41371187567710876,
      "learning_rate": 3.189804738040082e-06,
      "loss": 0.2929,
      "step": 40895
    },
    {
      "epoch": 1.8767381028865127,
      "grad_norm": 0.47034165263175964,
      "learning_rate": 3.189576185139236e-06,
      "loss": 0.3617,
      "step": 40896
    },
    {
      "epoch": 1.8767839933917672,
      "grad_norm": 0.4725523293018341,
      "learning_rate": 3.1893476365918847e-06,
      "loss": 0.3388,
      "step": 40897
    },
    {
      "epoch": 1.8768298838970217,
      "grad_norm": 0.48386532068252563,
      "learning_rate": 3.1891190923985827e-06,
      "loss": 0.3824,
      "step": 40898
    },
    {
      "epoch": 1.8768757744022762,
      "grad_norm": 0.46412011981010437,
      "learning_rate": 3.188890552559878e-06,
      "loss": 0.3354,
      "step": 40899
    },
    {
      "epoch": 1.8769216649075307,
      "grad_norm": 0.48253268003463745,
      "learning_rate": 3.1886620170763183e-06,
      "loss": 0.3564,
      "step": 40900
    },
    {
      "epoch": 1.8769675554127851,
      "grad_norm": 0.5229529738426208,
      "learning_rate": 3.1884334859484563e-06,
      "loss": 0.4508,
      "step": 40901
    },
    {
      "epoch": 1.8770134459180396,
      "grad_norm": 0.4572031795978546,
      "learning_rate": 3.18820495917684e-06,
      "loss": 0.3096,
      "step": 40902
    },
    {
      "epoch": 1.8770593364232941,
      "grad_norm": 0.45701250433921814,
      "learning_rate": 3.187976436762017e-06,
      "loss": 0.3153,
      "step": 40903
    },
    {
      "epoch": 1.8771052269285486,
      "grad_norm": 0.4692665636539459,
      "learning_rate": 3.18774791870454e-06,
      "loss": 0.3121,
      "step": 40904
    },
    {
      "epoch": 1.877151117433803,
      "grad_norm": 0.479989618062973,
      "learning_rate": 3.1875194050049585e-06,
      "loss": 0.3533,
      "step": 40905
    },
    {
      "epoch": 1.8771970079390574,
      "grad_norm": 0.4459161162376404,
      "learning_rate": 3.187290895663819e-06,
      "loss": 0.306,
      "step": 40906
    },
    {
      "epoch": 1.877242898444312,
      "grad_norm": 0.45808088779449463,
      "learning_rate": 3.187062390681674e-06,
      "loss": 0.3459,
      "step": 40907
    },
    {
      "epoch": 1.8772887889495662,
      "grad_norm": 0.41858750581741333,
      "learning_rate": 3.1868338900590717e-06,
      "loss": 0.3018,
      "step": 40908
    },
    {
      "epoch": 1.8773346794548207,
      "grad_norm": 0.4467853009700775,
      "learning_rate": 3.1866053937965596e-06,
      "loss": 0.3082,
      "step": 40909
    },
    {
      "epoch": 1.8773805699600752,
      "grad_norm": 0.43415582180023193,
      "learning_rate": 3.186376901894691e-06,
      "loss": 0.2892,
      "step": 40910
    },
    {
      "epoch": 1.8774264604653297,
      "grad_norm": 0.5375862717628479,
      "learning_rate": 3.1861484143540144e-06,
      "loss": 0.4403,
      "step": 40911
    },
    {
      "epoch": 1.8774723509705842,
      "grad_norm": 0.503567099571228,
      "learning_rate": 3.1859199311750765e-06,
      "loss": 0.314,
      "step": 40912
    },
    {
      "epoch": 1.8775182414758387,
      "grad_norm": 0.4473890960216522,
      "learning_rate": 3.18569145235843e-06,
      "loss": 0.3093,
      "step": 40913
    },
    {
      "epoch": 1.8775641319810932,
      "grad_norm": 0.4627523124217987,
      "learning_rate": 3.1854629779046232e-06,
      "loss": 0.3559,
      "step": 40914
    },
    {
      "epoch": 1.8776100224863477,
      "grad_norm": 0.4521934986114502,
      "learning_rate": 3.1852345078142027e-06,
      "loss": 0.3672,
      "step": 40915
    },
    {
      "epoch": 1.8776559129916022,
      "grad_norm": 0.4474486708641052,
      "learning_rate": 3.1850060420877226e-06,
      "loss": 0.3161,
      "step": 40916
    },
    {
      "epoch": 1.8777018034968564,
      "grad_norm": 0.45620644092559814,
      "learning_rate": 3.184777580725731e-06,
      "loss": 0.3643,
      "step": 40917
    },
    {
      "epoch": 1.877747694002111,
      "grad_norm": 0.46167102456092834,
      "learning_rate": 3.184549123728774e-06,
      "loss": 0.3207,
      "step": 40918
    },
    {
      "epoch": 1.8777935845073654,
      "grad_norm": 0.43175017833709717,
      "learning_rate": 3.1843206710974053e-06,
      "loss": 0.3455,
      "step": 40919
    },
    {
      "epoch": 1.87783947501262,
      "grad_norm": 0.7083204984664917,
      "learning_rate": 3.184092222832172e-06,
      "loss": 0.3604,
      "step": 40920
    },
    {
      "epoch": 1.8778853655178742,
      "grad_norm": 0.474498450756073,
      "learning_rate": 3.183863778933623e-06,
      "loss": 0.3942,
      "step": 40921
    },
    {
      "epoch": 1.8779312560231287,
      "grad_norm": 0.47159677743911743,
      "learning_rate": 3.1836353394023088e-06,
      "loss": 0.3494,
      "step": 40922
    },
    {
      "epoch": 1.8779771465283832,
      "grad_norm": 0.4765138328075409,
      "learning_rate": 3.1834069042387794e-06,
      "loss": 0.4053,
      "step": 40923
    },
    {
      "epoch": 1.8780230370336377,
      "grad_norm": 0.4741875231266022,
      "learning_rate": 3.1831784734435806e-06,
      "loss": 0.341,
      "step": 40924
    },
    {
      "epoch": 1.8780689275388922,
      "grad_norm": 0.4413011372089386,
      "learning_rate": 3.182950047017266e-06,
      "loss": 0.2939,
      "step": 40925
    },
    {
      "epoch": 1.8781148180441467,
      "grad_norm": 0.476239413022995,
      "learning_rate": 3.182721624960383e-06,
      "loss": 0.3774,
      "step": 40926
    },
    {
      "epoch": 1.8781607085494012,
      "grad_norm": 0.47231870889663696,
      "learning_rate": 3.1824932072734803e-06,
      "loss": 0.3491,
      "step": 40927
    },
    {
      "epoch": 1.8782065990546557,
      "grad_norm": 0.512829601764679,
      "learning_rate": 3.1822647939571084e-06,
      "loss": 0.4153,
      "step": 40928
    },
    {
      "epoch": 1.8782524895599102,
      "grad_norm": 0.4762999415397644,
      "learning_rate": 3.1820363850118165e-06,
      "loss": 0.3358,
      "step": 40929
    },
    {
      "epoch": 1.8782983800651645,
      "grad_norm": 0.45543399453163147,
      "learning_rate": 3.181807980438153e-06,
      "loss": 0.3569,
      "step": 40930
    },
    {
      "epoch": 1.878344270570419,
      "grad_norm": 0.4704321622848511,
      "learning_rate": 3.1815795802366654e-06,
      "loss": 0.2989,
      "step": 40931
    },
    {
      "epoch": 1.8783901610756735,
      "grad_norm": 0.4486519992351532,
      "learning_rate": 3.181351184407907e-06,
      "loss": 0.3274,
      "step": 40932
    },
    {
      "epoch": 1.8784360515809277,
      "grad_norm": 0.4495529234409332,
      "learning_rate": 3.181122792952425e-06,
      "loss": 0.3044,
      "step": 40933
    },
    {
      "epoch": 1.8784819420861822,
      "grad_norm": 0.46509629487991333,
      "learning_rate": 3.180894405870768e-06,
      "loss": 0.3178,
      "step": 40934
    },
    {
      "epoch": 1.8785278325914367,
      "grad_norm": 0.4702320694923401,
      "learning_rate": 3.1806660231634867e-06,
      "loss": 0.4017,
      "step": 40935
    },
    {
      "epoch": 1.8785737230966912,
      "grad_norm": 0.4962367117404938,
      "learning_rate": 3.1804376448311293e-06,
      "loss": 0.388,
      "step": 40936
    },
    {
      "epoch": 1.8786196136019457,
      "grad_norm": 0.4944383502006531,
      "learning_rate": 3.1802092708742426e-06,
      "loss": 0.3905,
      "step": 40937
    },
    {
      "epoch": 1.8786655041072002,
      "grad_norm": 0.4689026474952698,
      "learning_rate": 3.1799809012933806e-06,
      "loss": 0.3787,
      "step": 40938
    },
    {
      "epoch": 1.8787113946124547,
      "grad_norm": 0.45295166969299316,
      "learning_rate": 3.1797525360890903e-06,
      "loss": 0.3055,
      "step": 40939
    },
    {
      "epoch": 1.8787572851177092,
      "grad_norm": 0.4577741324901581,
      "learning_rate": 3.17952417526192e-06,
      "loss": 0.3134,
      "step": 40940
    },
    {
      "epoch": 1.8788031756229637,
      "grad_norm": 0.49297085404396057,
      "learning_rate": 3.1792958188124203e-06,
      "loss": 0.351,
      "step": 40941
    },
    {
      "epoch": 1.8788490661282182,
      "grad_norm": 0.45363950729370117,
      "learning_rate": 3.179067466741139e-06,
      "loss": 0.3225,
      "step": 40942
    },
    {
      "epoch": 1.8788949566334725,
      "grad_norm": 0.49120309948921204,
      "learning_rate": 3.1788391190486253e-06,
      "loss": 0.3848,
      "step": 40943
    },
    {
      "epoch": 1.878940847138727,
      "grad_norm": 0.444976806640625,
      "learning_rate": 3.178610775735429e-06,
      "loss": 0.3348,
      "step": 40944
    },
    {
      "epoch": 1.8789867376439815,
      "grad_norm": 0.4482094347476959,
      "learning_rate": 3.1783824368020997e-06,
      "loss": 0.3154,
      "step": 40945
    },
    {
      "epoch": 1.8790326281492358,
      "grad_norm": 0.5034446120262146,
      "learning_rate": 3.1781541022491836e-06,
      "loss": 0.3338,
      "step": 40946
    },
    {
      "epoch": 1.8790785186544903,
      "grad_norm": 0.5176222324371338,
      "learning_rate": 3.1779257720772337e-06,
      "loss": 0.4615,
      "step": 40947
    },
    {
      "epoch": 1.8791244091597448,
      "grad_norm": 0.4652732312679291,
      "learning_rate": 3.1776974462867972e-06,
      "loss": 0.323,
      "step": 40948
    },
    {
      "epoch": 1.8791702996649993,
      "grad_norm": 0.4904159605503082,
      "learning_rate": 3.177469124878423e-06,
      "loss": 0.3534,
      "step": 40949
    },
    {
      "epoch": 1.8792161901702538,
      "grad_norm": 0.5092717409133911,
      "learning_rate": 3.17724080785266e-06,
      "loss": 0.4567,
      "step": 40950
    },
    {
      "epoch": 1.8792620806755083,
      "grad_norm": 0.4551880359649658,
      "learning_rate": 3.177012495210058e-06,
      "loss": 0.3277,
      "step": 40951
    },
    {
      "epoch": 1.8793079711807628,
      "grad_norm": 0.48032402992248535,
      "learning_rate": 3.176784186951164e-06,
      "loss": 0.3606,
      "step": 40952
    },
    {
      "epoch": 1.8793538616860173,
      "grad_norm": 0.5019930005073547,
      "learning_rate": 3.1765558830765306e-06,
      "loss": 0.3446,
      "step": 40953
    },
    {
      "epoch": 1.8793997521912718,
      "grad_norm": 0.457221120595932,
      "learning_rate": 3.1763275835867048e-06,
      "loss": 0.346,
      "step": 40954
    },
    {
      "epoch": 1.8794456426965263,
      "grad_norm": 0.425237774848938,
      "learning_rate": 3.1760992884822344e-06,
      "loss": 0.2834,
      "step": 40955
    },
    {
      "epoch": 1.8794915332017805,
      "grad_norm": 0.48285725712776184,
      "learning_rate": 3.17587099776367e-06,
      "loss": 0.4012,
      "step": 40956
    },
    {
      "epoch": 1.879537423707035,
      "grad_norm": 0.4501810073852539,
      "learning_rate": 3.1756427114315613e-06,
      "loss": 0.3587,
      "step": 40957
    },
    {
      "epoch": 1.8795833142122895,
      "grad_norm": 0.4360103905200958,
      "learning_rate": 3.175414429486454e-06,
      "loss": 0.2812,
      "step": 40958
    },
    {
      "epoch": 1.8796292047175438,
      "grad_norm": 0.4685761034488678,
      "learning_rate": 3.1751861519289008e-06,
      "loss": 0.3559,
      "step": 40959
    },
    {
      "epoch": 1.8796750952227983,
      "grad_norm": 0.4544246196746826,
      "learning_rate": 3.174957878759448e-06,
      "loss": 0.3458,
      "step": 40960
    },
    {
      "epoch": 1.8797209857280528,
      "grad_norm": 0.47219225764274597,
      "learning_rate": 3.1747296099786475e-06,
      "loss": 0.3163,
      "step": 40961
    },
    {
      "epoch": 1.8797668762333073,
      "grad_norm": 0.456236332654953,
      "learning_rate": 3.1745013455870433e-06,
      "loss": 0.317,
      "step": 40962
    },
    {
      "epoch": 1.8798127667385618,
      "grad_norm": 0.5169612169265747,
      "learning_rate": 3.1742730855851896e-06,
      "loss": 0.4315,
      "step": 40963
    },
    {
      "epoch": 1.8798586572438163,
      "grad_norm": 0.46480226516723633,
      "learning_rate": 3.174044829973632e-06,
      "loss": 0.3369,
      "step": 40964
    },
    {
      "epoch": 1.8799045477490708,
      "grad_norm": 0.4524090886116028,
      "learning_rate": 3.1738165787529198e-06,
      "loss": 0.3061,
      "step": 40965
    },
    {
      "epoch": 1.8799504382543253,
      "grad_norm": 0.4854663908481598,
      "learning_rate": 3.1735883319236027e-06,
      "loss": 0.4362,
      "step": 40966
    },
    {
      "epoch": 1.8799963287595798,
      "grad_norm": 0.4476641118526459,
      "learning_rate": 3.1733600894862303e-06,
      "loss": 0.2879,
      "step": 40967
    },
    {
      "epoch": 1.880042219264834,
      "grad_norm": 0.4982832670211792,
      "learning_rate": 3.173131851441348e-06,
      "loss": 0.3742,
      "step": 40968
    },
    {
      "epoch": 1.8800881097700886,
      "grad_norm": 0.47871583700180054,
      "learning_rate": 3.1729036177895084e-06,
      "loss": 0.3744,
      "step": 40969
    },
    {
      "epoch": 1.880134000275343,
      "grad_norm": 0.5018702149391174,
      "learning_rate": 3.1726753885312593e-06,
      "loss": 0.4088,
      "step": 40970
    },
    {
      "epoch": 1.8801798907805973,
      "grad_norm": 0.49285808205604553,
      "learning_rate": 3.172447163667148e-06,
      "loss": 0.4066,
      "step": 40971
    },
    {
      "epoch": 1.8802257812858518,
      "grad_norm": 0.4498308300971985,
      "learning_rate": 3.1722189431977264e-06,
      "loss": 0.3391,
      "step": 40972
    },
    {
      "epoch": 1.8802716717911063,
      "grad_norm": 0.4803049862384796,
      "learning_rate": 3.171990727123541e-06,
      "loss": 0.3504,
      "step": 40973
    },
    {
      "epoch": 1.8803175622963608,
      "grad_norm": 0.46708038449287415,
      "learning_rate": 3.171762515445138e-06,
      "loss": 0.3738,
      "step": 40974
    },
    {
      "epoch": 1.8803634528016153,
      "grad_norm": 0.4638459384441376,
      "learning_rate": 3.1715343081630725e-06,
      "loss": 0.3382,
      "step": 40975
    },
    {
      "epoch": 1.8804093433068698,
      "grad_norm": 0.5472952723503113,
      "learning_rate": 3.1713061052778887e-06,
      "loss": 0.3378,
      "step": 40976
    },
    {
      "epoch": 1.8804552338121243,
      "grad_norm": 0.39143460988998413,
      "learning_rate": 3.1710779067901358e-06,
      "loss": 0.2337,
      "step": 40977
    },
    {
      "epoch": 1.8805011243173788,
      "grad_norm": 0.44960275292396545,
      "learning_rate": 3.1708497127003646e-06,
      "loss": 0.3567,
      "step": 40978
    },
    {
      "epoch": 1.8805470148226333,
      "grad_norm": 0.4635041356086731,
      "learning_rate": 3.170621523009123e-06,
      "loss": 0.3112,
      "step": 40979
    },
    {
      "epoch": 1.8805929053278878,
      "grad_norm": 0.4710862934589386,
      "learning_rate": 3.1703933377169567e-06,
      "loss": 0.3256,
      "step": 40980
    },
    {
      "epoch": 1.880638795833142,
      "grad_norm": 0.42451804876327515,
      "learning_rate": 3.1701651568244193e-06,
      "loss": 0.3285,
      "step": 40981
    },
    {
      "epoch": 1.8806846863383966,
      "grad_norm": 0.49581775069236755,
      "learning_rate": 3.1699369803320566e-06,
      "loss": 0.4141,
      "step": 40982
    },
    {
      "epoch": 1.880730576843651,
      "grad_norm": 0.4580715000629425,
      "learning_rate": 3.169708808240417e-06,
      "loss": 0.3457,
      "step": 40983
    },
    {
      "epoch": 1.8807764673489054,
      "grad_norm": 0.5822945833206177,
      "learning_rate": 3.1694806405500517e-06,
      "loss": 0.4298,
      "step": 40984
    },
    {
      "epoch": 1.8808223578541599,
      "grad_norm": 0.46547731757164,
      "learning_rate": 3.1692524772615076e-06,
      "loss": 0.3195,
      "step": 40985
    },
    {
      "epoch": 1.8808682483594144,
      "grad_norm": 0.4505749046802521,
      "learning_rate": 3.1690243183753313e-06,
      "loss": 0.3176,
      "step": 40986
    },
    {
      "epoch": 1.8809141388646688,
      "grad_norm": 0.46239930391311646,
      "learning_rate": 3.1687961638920763e-06,
      "loss": 0.3257,
      "step": 40987
    },
    {
      "epoch": 1.8809600293699233,
      "grad_norm": 0.46887993812561035,
      "learning_rate": 3.1685680138122876e-06,
      "loss": 0.3266,
      "step": 40988
    },
    {
      "epoch": 1.8810059198751778,
      "grad_norm": 0.44191545248031616,
      "learning_rate": 3.168339868136513e-06,
      "loss": 0.3081,
      "step": 40989
    },
    {
      "epoch": 1.8810518103804323,
      "grad_norm": 0.470470666885376,
      "learning_rate": 3.168111726865305e-06,
      "loss": 0.3695,
      "step": 40990
    },
    {
      "epoch": 1.8810977008856868,
      "grad_norm": 0.4549897313117981,
      "learning_rate": 3.1678835899992107e-06,
      "loss": 0.3018,
      "step": 40991
    },
    {
      "epoch": 1.8811435913909413,
      "grad_norm": 0.43316784501075745,
      "learning_rate": 3.1676554575387763e-06,
      "loss": 0.302,
      "step": 40992
    },
    {
      "epoch": 1.8811894818961958,
      "grad_norm": 0.4470137357711792,
      "learning_rate": 3.1674273294845535e-06,
      "loss": 0.3216,
      "step": 40993
    },
    {
      "epoch": 1.8812353724014501,
      "grad_norm": 0.46059149503707886,
      "learning_rate": 3.167199205837089e-06,
      "loss": 0.3283,
      "step": 40994
    },
    {
      "epoch": 1.8812812629067046,
      "grad_norm": 0.44102296233177185,
      "learning_rate": 3.166971086596933e-06,
      "loss": 0.3092,
      "step": 40995
    },
    {
      "epoch": 1.881327153411959,
      "grad_norm": 0.49818354845046997,
      "learning_rate": 3.1667429717646304e-06,
      "loss": 0.426,
      "step": 40996
    },
    {
      "epoch": 1.8813730439172134,
      "grad_norm": 0.4413798451423645,
      "learning_rate": 3.1665148613407348e-06,
      "loss": 0.3063,
      "step": 40997
    },
    {
      "epoch": 1.8814189344224679,
      "grad_norm": 0.4837610423564911,
      "learning_rate": 3.166286755325792e-06,
      "loss": 0.4054,
      "step": 40998
    },
    {
      "epoch": 1.8814648249277224,
      "grad_norm": 0.4820041060447693,
      "learning_rate": 3.1660586537203497e-06,
      "loss": 0.3746,
      "step": 40999
    },
    {
      "epoch": 1.8815107154329769,
      "grad_norm": 0.4675370752811432,
      "learning_rate": 3.1658305565249594e-06,
      "loss": 0.3214,
      "step": 41000
    },
    {
      "epoch": 1.8815566059382314,
      "grad_norm": 0.48801398277282715,
      "learning_rate": 3.165602463740166e-06,
      "loss": 0.3843,
      "step": 41001
    },
    {
      "epoch": 1.8816024964434859,
      "grad_norm": 0.49266982078552246,
      "learning_rate": 3.165374375366519e-06,
      "loss": 0.3881,
      "step": 41002
    },
    {
      "epoch": 1.8816483869487404,
      "grad_norm": 0.473346471786499,
      "learning_rate": 3.1651462914045695e-06,
      "loss": 0.3793,
      "step": 41003
    },
    {
      "epoch": 1.8816942774539949,
      "grad_norm": 0.47626182436943054,
      "learning_rate": 3.164918211854864e-06,
      "loss": 0.381,
      "step": 41004
    },
    {
      "epoch": 1.8817401679592494,
      "grad_norm": 0.4672994315624237,
      "learning_rate": 3.1646901367179495e-06,
      "loss": 0.3598,
      "step": 41005
    },
    {
      "epoch": 1.8817860584645036,
      "grad_norm": 0.4838709831237793,
      "learning_rate": 3.164462065994378e-06,
      "loss": 0.3767,
      "step": 41006
    },
    {
      "epoch": 1.8818319489697581,
      "grad_norm": 0.4869977533817291,
      "learning_rate": 3.1642339996846947e-06,
      "loss": 0.4107,
      "step": 41007
    },
    {
      "epoch": 1.8818778394750126,
      "grad_norm": 0.4710145592689514,
      "learning_rate": 3.164005937789448e-06,
      "loss": 0.3517,
      "step": 41008
    },
    {
      "epoch": 1.8819237299802671,
      "grad_norm": 0.48436930775642395,
      "learning_rate": 3.1637778803091906e-06,
      "loss": 0.3879,
      "step": 41009
    },
    {
      "epoch": 1.8819696204855214,
      "grad_norm": 0.4550233483314514,
      "learning_rate": 3.1635498272444663e-06,
      "loss": 0.3587,
      "step": 41010
    },
    {
      "epoch": 1.882015510990776,
      "grad_norm": 0.4757569134235382,
      "learning_rate": 3.1633217785958236e-06,
      "loss": 0.4098,
      "step": 41011
    },
    {
      "epoch": 1.8820614014960304,
      "grad_norm": 0.5197775363922119,
      "learning_rate": 3.163093734363815e-06,
      "loss": 0.4454,
      "step": 41012
    },
    {
      "epoch": 1.882107292001285,
      "grad_norm": 0.4945377707481384,
      "learning_rate": 3.162865694548985e-06,
      "loss": 0.3993,
      "step": 41013
    },
    {
      "epoch": 1.8821531825065394,
      "grad_norm": 0.46583765745162964,
      "learning_rate": 3.1626376591518827e-06,
      "loss": 0.3524,
      "step": 41014
    },
    {
      "epoch": 1.882199073011794,
      "grad_norm": 0.4493637979030609,
      "learning_rate": 3.1624096281730575e-06,
      "loss": 0.294,
      "step": 41015
    },
    {
      "epoch": 1.8822449635170484,
      "grad_norm": 0.47812148928642273,
      "learning_rate": 3.162181601613058e-06,
      "loss": 0.3623,
      "step": 41016
    },
    {
      "epoch": 1.882290854022303,
      "grad_norm": 0.44607001543045044,
      "learning_rate": 3.1619535794724295e-06,
      "loss": 0.3312,
      "step": 41017
    },
    {
      "epoch": 1.8823367445275574,
      "grad_norm": 0.5113836526870728,
      "learning_rate": 3.161725561751724e-06,
      "loss": 0.4008,
      "step": 41018
    },
    {
      "epoch": 1.8823826350328117,
      "grad_norm": 0.4481378495693207,
      "learning_rate": 3.1614975484514886e-06,
      "loss": 0.2826,
      "step": 41019
    },
    {
      "epoch": 1.8824285255380662,
      "grad_norm": 0.4838877320289612,
      "learning_rate": 3.16126953957227e-06,
      "loss": 0.3627,
      "step": 41020
    },
    {
      "epoch": 1.8824744160433207,
      "grad_norm": 0.44637173414230347,
      "learning_rate": 3.1610415351146195e-06,
      "loss": 0.3072,
      "step": 41021
    },
    {
      "epoch": 1.882520306548575,
      "grad_norm": 0.4835214912891388,
      "learning_rate": 3.1608135350790836e-06,
      "loss": 0.374,
      "step": 41022
    },
    {
      "epoch": 1.8825661970538294,
      "grad_norm": 0.49482837319374084,
      "learning_rate": 3.1605855394662087e-06,
      "loss": 0.3754,
      "step": 41023
    },
    {
      "epoch": 1.882612087559084,
      "grad_norm": 0.5678021311759949,
      "learning_rate": 3.160357548276547e-06,
      "loss": 0.3885,
      "step": 41024
    },
    {
      "epoch": 1.8826579780643384,
      "grad_norm": 0.5226348638534546,
      "learning_rate": 3.1601295615106444e-06,
      "loss": 0.4199,
      "step": 41025
    },
    {
      "epoch": 1.882703868569593,
      "grad_norm": 0.4846934378147125,
      "learning_rate": 3.1599015791690503e-06,
      "loss": 0.371,
      "step": 41026
    },
    {
      "epoch": 1.8827497590748474,
      "grad_norm": 0.4854075610637665,
      "learning_rate": 3.15967360125231e-06,
      "loss": 0.3882,
      "step": 41027
    },
    {
      "epoch": 1.882795649580102,
      "grad_norm": 0.494087278842926,
      "learning_rate": 3.159445627760976e-06,
      "loss": 0.3795,
      "step": 41028
    },
    {
      "epoch": 1.8828415400853564,
      "grad_norm": 0.5188537836074829,
      "learning_rate": 3.1592176586955937e-06,
      "loss": 0.4179,
      "step": 41029
    },
    {
      "epoch": 1.882887430590611,
      "grad_norm": 0.45918014645576477,
      "learning_rate": 3.15898969405671e-06,
      "loss": 0.3781,
      "step": 41030
    },
    {
      "epoch": 1.8829333210958654,
      "grad_norm": 0.4742754399776459,
      "learning_rate": 3.1587617338448782e-06,
      "loss": 0.37,
      "step": 41031
    },
    {
      "epoch": 1.8829792116011197,
      "grad_norm": 0.5049639344215393,
      "learning_rate": 3.1585337780606427e-06,
      "loss": 0.3932,
      "step": 41032
    },
    {
      "epoch": 1.8830251021063742,
      "grad_norm": 0.4636766314506531,
      "learning_rate": 3.15830582670455e-06,
      "loss": 0.3197,
      "step": 41033
    },
    {
      "epoch": 1.8830709926116287,
      "grad_norm": 0.48075205087661743,
      "learning_rate": 3.158077879777152e-06,
      "loss": 0.3689,
      "step": 41034
    },
    {
      "epoch": 1.883116883116883,
      "grad_norm": 0.4679269790649414,
      "learning_rate": 3.157849937278996e-06,
      "loss": 0.3965,
      "step": 41035
    },
    {
      "epoch": 1.8831627736221375,
      "grad_norm": 0.4851629137992859,
      "learning_rate": 3.1576219992106282e-06,
      "loss": 0.3489,
      "step": 41036
    },
    {
      "epoch": 1.883208664127392,
      "grad_norm": 0.4370095729827881,
      "learning_rate": 3.157394065572599e-06,
      "loss": 0.3428,
      "step": 41037
    },
    {
      "epoch": 1.8832545546326465,
      "grad_norm": 0.4753059148788452,
      "learning_rate": 3.157166136365456e-06,
      "loss": 0.3576,
      "step": 41038
    },
    {
      "epoch": 1.883300445137901,
      "grad_norm": 0.5060460567474365,
      "learning_rate": 3.156938211589744e-06,
      "loss": 0.4029,
      "step": 41039
    },
    {
      "epoch": 1.8833463356431555,
      "grad_norm": 0.45769980549812317,
      "learning_rate": 3.156710291246017e-06,
      "loss": 0.3456,
      "step": 41040
    },
    {
      "epoch": 1.88339222614841,
      "grad_norm": 0.4897744059562683,
      "learning_rate": 3.156482375334819e-06,
      "loss": 0.3776,
      "step": 41041
    },
    {
      "epoch": 1.8834381166536645,
      "grad_norm": 0.4726833403110504,
      "learning_rate": 3.1562544638566983e-06,
      "loss": 0.3617,
      "step": 41042
    },
    {
      "epoch": 1.883484007158919,
      "grad_norm": 0.4847007989883423,
      "learning_rate": 3.156026556812205e-06,
      "loss": 0.3881,
      "step": 41043
    },
    {
      "epoch": 1.8835298976641734,
      "grad_norm": 0.4632507562637329,
      "learning_rate": 3.155798654201886e-06,
      "loss": 0.3046,
      "step": 41044
    },
    {
      "epoch": 1.8835757881694277,
      "grad_norm": 0.4409697949886322,
      "learning_rate": 3.1555707560262865e-06,
      "loss": 0.2842,
      "step": 41045
    },
    {
      "epoch": 1.8836216786746822,
      "grad_norm": 0.49303868412971497,
      "learning_rate": 3.1553428622859595e-06,
      "loss": 0.3752,
      "step": 41046
    },
    {
      "epoch": 1.8836675691799367,
      "grad_norm": 0.4547823369503021,
      "learning_rate": 3.1551149729814513e-06,
      "loss": 0.331,
      "step": 41047
    },
    {
      "epoch": 1.883713459685191,
      "grad_norm": 0.505851149559021,
      "learning_rate": 3.1548870881133074e-06,
      "loss": 0.4515,
      "step": 41048
    },
    {
      "epoch": 1.8837593501904455,
      "grad_norm": 0.4911642372608185,
      "learning_rate": 3.154659207682079e-06,
      "loss": 0.4169,
      "step": 41049
    },
    {
      "epoch": 1.8838052406957,
      "grad_norm": 0.49238285422325134,
      "learning_rate": 3.1544313316883134e-06,
      "loss": 0.3733,
      "step": 41050
    },
    {
      "epoch": 1.8838511312009545,
      "grad_norm": 0.47842520475387573,
      "learning_rate": 3.1542034601325556e-06,
      "loss": 0.3821,
      "step": 41051
    },
    {
      "epoch": 1.883897021706209,
      "grad_norm": 0.4504016041755676,
      "learning_rate": 3.153975593015358e-06,
      "loss": 0.3187,
      "step": 41052
    },
    {
      "epoch": 1.8839429122114635,
      "grad_norm": 0.48080721497535706,
      "learning_rate": 3.153747730337267e-06,
      "loss": 0.3358,
      "step": 41053
    },
    {
      "epoch": 1.883988802716718,
      "grad_norm": 0.4527626633644104,
      "learning_rate": 3.1535198720988276e-06,
      "loss": 0.339,
      "step": 41054
    },
    {
      "epoch": 1.8840346932219725,
      "grad_norm": 0.4792722165584564,
      "learning_rate": 3.153292018300592e-06,
      "loss": 0.3841,
      "step": 41055
    },
    {
      "epoch": 1.884080583727227,
      "grad_norm": 0.4764436185359955,
      "learning_rate": 3.1530641689431064e-06,
      "loss": 0.3289,
      "step": 41056
    },
    {
      "epoch": 1.8841264742324813,
      "grad_norm": 0.49971815943717957,
      "learning_rate": 3.1528363240269168e-06,
      "loss": 0.3801,
      "step": 41057
    },
    {
      "epoch": 1.8841723647377357,
      "grad_norm": 0.46734774112701416,
      "learning_rate": 3.1526084835525743e-06,
      "loss": 0.3466,
      "step": 41058
    },
    {
      "epoch": 1.8842182552429902,
      "grad_norm": 0.47285333275794983,
      "learning_rate": 3.1523806475206253e-06,
      "loss": 0.3479,
      "step": 41059
    },
    {
      "epoch": 1.8842641457482445,
      "grad_norm": 0.45007196068763733,
      "learning_rate": 3.152152815931618e-06,
      "loss": 0.3375,
      "step": 41060
    },
    {
      "epoch": 1.884310036253499,
      "grad_norm": 0.4664475619792938,
      "learning_rate": 3.151924988786098e-06,
      "loss": 0.3346,
      "step": 41061
    },
    {
      "epoch": 1.8843559267587535,
      "grad_norm": 0.4767315089702606,
      "learning_rate": 3.151697166084617e-06,
      "loss": 0.3529,
      "step": 41062
    },
    {
      "epoch": 1.884401817264008,
      "grad_norm": 0.4740633964538574,
      "learning_rate": 3.1514693478277207e-06,
      "loss": 0.3967,
      "step": 41063
    },
    {
      "epoch": 1.8844477077692625,
      "grad_norm": 0.4482031464576721,
      "learning_rate": 3.1512415340159565e-06,
      "loss": 0.3454,
      "step": 41064
    },
    {
      "epoch": 1.884493598274517,
      "grad_norm": 0.4896005094051361,
      "learning_rate": 3.151013724649874e-06,
      "loss": 0.411,
      "step": 41065
    },
    {
      "epoch": 1.8845394887797715,
      "grad_norm": 0.4585939943790436,
      "learning_rate": 3.1507859197300194e-06,
      "loss": 0.356,
      "step": 41066
    },
    {
      "epoch": 1.884585379285026,
      "grad_norm": 0.4270206689834595,
      "learning_rate": 3.15055811925694e-06,
      "loss": 0.2783,
      "step": 41067
    },
    {
      "epoch": 1.8846312697902805,
      "grad_norm": 0.563872754573822,
      "learning_rate": 3.150330323231186e-06,
      "loss": 0.3809,
      "step": 41068
    },
    {
      "epoch": 1.884677160295535,
      "grad_norm": 0.4903283715248108,
      "learning_rate": 3.150102531653304e-06,
      "loss": 0.3609,
      "step": 41069
    },
    {
      "epoch": 1.8847230508007893,
      "grad_norm": 0.47412610054016113,
      "learning_rate": 3.1498747445238403e-06,
      "loss": 0.3893,
      "step": 41070
    },
    {
      "epoch": 1.8847689413060438,
      "grad_norm": 0.4898699223995209,
      "learning_rate": 3.1496469618433453e-06,
      "loss": 0.3793,
      "step": 41071
    },
    {
      "epoch": 1.8848148318112983,
      "grad_norm": 0.43060028553009033,
      "learning_rate": 3.1494191836123643e-06,
      "loss": 0.3091,
      "step": 41072
    },
    {
      "epoch": 1.8848607223165526,
      "grad_norm": 0.5120705366134644,
      "learning_rate": 3.1491914098314458e-06,
      "loss": 0.4497,
      "step": 41073
    },
    {
      "epoch": 1.884906612821807,
      "grad_norm": 0.44911623001098633,
      "learning_rate": 3.1489636405011386e-06,
      "loss": 0.3402,
      "step": 41074
    },
    {
      "epoch": 1.8849525033270615,
      "grad_norm": 0.49238765239715576,
      "learning_rate": 3.148735875621991e-06,
      "loss": 0.3687,
      "step": 41075
    },
    {
      "epoch": 1.884998393832316,
      "grad_norm": 0.4685404598712921,
      "learning_rate": 3.1485081151945464e-06,
      "loss": 0.32,
      "step": 41076
    },
    {
      "epoch": 1.8850442843375705,
      "grad_norm": 0.51351398229599,
      "learning_rate": 3.1482803592193574e-06,
      "loss": 0.4102,
      "step": 41077
    },
    {
      "epoch": 1.885090174842825,
      "grad_norm": 0.4447522461414337,
      "learning_rate": 3.1480526076969697e-06,
      "loss": 0.2875,
      "step": 41078
    },
    {
      "epoch": 1.8851360653480795,
      "grad_norm": 0.4599219262599945,
      "learning_rate": 3.14782486062793e-06,
      "loss": 0.3353,
      "step": 41079
    },
    {
      "epoch": 1.885181955853334,
      "grad_norm": 0.4919465482234955,
      "learning_rate": 3.1475971180127877e-06,
      "loss": 0.37,
      "step": 41080
    },
    {
      "epoch": 1.8852278463585885,
      "grad_norm": 0.4829567074775696,
      "learning_rate": 3.14736937985209e-06,
      "loss": 0.358,
      "step": 41081
    },
    {
      "epoch": 1.885273736863843,
      "grad_norm": 0.4866565763950348,
      "learning_rate": 3.147141646146382e-06,
      "loss": 0.3741,
      "step": 41082
    },
    {
      "epoch": 1.8853196273690973,
      "grad_norm": 0.4796319901943207,
      "learning_rate": 3.146913916896216e-06,
      "loss": 0.3211,
      "step": 41083
    },
    {
      "epoch": 1.8853655178743518,
      "grad_norm": 0.5006620287895203,
      "learning_rate": 3.1466861921021367e-06,
      "loss": 0.3985,
      "step": 41084
    },
    {
      "epoch": 1.8854114083796063,
      "grad_norm": 0.47266310453414917,
      "learning_rate": 3.1464584717646917e-06,
      "loss": 0.2927,
      "step": 41085
    },
    {
      "epoch": 1.8854572988848606,
      "grad_norm": 0.44848233461380005,
      "learning_rate": 3.1462307558844298e-06,
      "loss": 0.2782,
      "step": 41086
    },
    {
      "epoch": 1.885503189390115,
      "grad_norm": 0.4683627784252167,
      "learning_rate": 3.1460030444618973e-06,
      "loss": 0.388,
      "step": 41087
    },
    {
      "epoch": 1.8855490798953696,
      "grad_norm": 0.43223220109939575,
      "learning_rate": 3.1457753374976406e-06,
      "loss": 0.2855,
      "step": 41088
    },
    {
      "epoch": 1.885594970400624,
      "grad_norm": 0.4398265480995178,
      "learning_rate": 3.145547634992212e-06,
      "loss": 0.2903,
      "step": 41089
    },
    {
      "epoch": 1.8856408609058786,
      "grad_norm": 0.47206953167915344,
      "learning_rate": 3.145319936946155e-06,
      "loss": 0.3422,
      "step": 41090
    },
    {
      "epoch": 1.885686751411133,
      "grad_norm": 0.46676135063171387,
      "learning_rate": 3.1450922433600175e-06,
      "loss": 0.3552,
      "step": 41091
    },
    {
      "epoch": 1.8857326419163876,
      "grad_norm": 0.45267415046691895,
      "learning_rate": 3.144864554234349e-06,
      "loss": 0.316,
      "step": 41092
    },
    {
      "epoch": 1.885778532421642,
      "grad_norm": 0.48060664534568787,
      "learning_rate": 3.1446368695696955e-06,
      "loss": 0.4128,
      "step": 41093
    },
    {
      "epoch": 1.8858244229268966,
      "grad_norm": 0.474813312292099,
      "learning_rate": 3.1444091893666044e-06,
      "loss": 0.3638,
      "step": 41094
    },
    {
      "epoch": 1.8858703134321508,
      "grad_norm": 0.4663175642490387,
      "learning_rate": 3.144181513625622e-06,
      "loss": 0.3778,
      "step": 41095
    },
    {
      "epoch": 1.8859162039374053,
      "grad_norm": 0.4853506088256836,
      "learning_rate": 3.143953842347299e-06,
      "loss": 0.3867,
      "step": 41096
    },
    {
      "epoch": 1.8859620944426598,
      "grad_norm": 0.4963219165802002,
      "learning_rate": 3.143726175532183e-06,
      "loss": 0.4107,
      "step": 41097
    },
    {
      "epoch": 1.8860079849479143,
      "grad_norm": 0.4717501997947693,
      "learning_rate": 3.143498513180816e-06,
      "loss": 0.3985,
      "step": 41098
    },
    {
      "epoch": 1.8860538754531686,
      "grad_norm": 0.488343745470047,
      "learning_rate": 3.1432708552937513e-06,
      "loss": 0.3744,
      "step": 41099
    },
    {
      "epoch": 1.886099765958423,
      "grad_norm": 0.43021586537361145,
      "learning_rate": 3.143043201871534e-06,
      "loss": 0.3084,
      "step": 41100
    },
    {
      "epoch": 1.8861456564636776,
      "grad_norm": 0.4764815866947174,
      "learning_rate": 3.1428155529147108e-06,
      "loss": 0.4293,
      "step": 41101
    },
    {
      "epoch": 1.886191546968932,
      "grad_norm": 0.4853217303752899,
      "learning_rate": 3.142587908423832e-06,
      "loss": 0.422,
      "step": 41102
    },
    {
      "epoch": 1.8862374374741866,
      "grad_norm": 0.45550957322120667,
      "learning_rate": 3.142360268399442e-06,
      "loss": 0.3156,
      "step": 41103
    },
    {
      "epoch": 1.886283327979441,
      "grad_norm": 0.45159295201301575,
      "learning_rate": 3.1421326328420877e-06,
      "loss": 0.3281,
      "step": 41104
    },
    {
      "epoch": 1.8863292184846956,
      "grad_norm": 0.4706265330314636,
      "learning_rate": 3.1419050017523196e-06,
      "loss": 0.3621,
      "step": 41105
    },
    {
      "epoch": 1.88637510898995,
      "grad_norm": 0.4876958131790161,
      "learning_rate": 3.1416773751306837e-06,
      "loss": 0.3979,
      "step": 41106
    },
    {
      "epoch": 1.8864209994952046,
      "grad_norm": 0.4814624488353729,
      "learning_rate": 3.1414497529777265e-06,
      "loss": 0.2631,
      "step": 41107
    },
    {
      "epoch": 1.8864668900004589,
      "grad_norm": 0.4697282314300537,
      "learning_rate": 3.141222135293997e-06,
      "loss": 0.3668,
      "step": 41108
    },
    {
      "epoch": 1.8865127805057134,
      "grad_norm": 0.4303424656391144,
      "learning_rate": 3.1409945220800416e-06,
      "loss": 0.2821,
      "step": 41109
    },
    {
      "epoch": 1.8865586710109679,
      "grad_norm": 0.5009654760360718,
      "learning_rate": 3.1407669133364053e-06,
      "loss": 0.3547,
      "step": 41110
    },
    {
      "epoch": 1.8866045615162221,
      "grad_norm": 0.49545609951019287,
      "learning_rate": 3.1405393090636406e-06,
      "loss": 0.3871,
      "step": 41111
    },
    {
      "epoch": 1.8866504520214766,
      "grad_norm": 0.5043784379959106,
      "learning_rate": 3.1403117092622908e-06,
      "loss": 0.4654,
      "step": 41112
    },
    {
      "epoch": 1.8866963425267311,
      "grad_norm": 0.4593999981880188,
      "learning_rate": 3.140084113932904e-06,
      "loss": 0.3082,
      "step": 41113
    },
    {
      "epoch": 1.8867422330319856,
      "grad_norm": 0.470515638589859,
      "learning_rate": 3.139856523076029e-06,
      "loss": 0.3537,
      "step": 41114
    },
    {
      "epoch": 1.8867881235372401,
      "grad_norm": 0.5026741027832031,
      "learning_rate": 3.139628936692212e-06,
      "loss": 0.3472,
      "step": 41115
    },
    {
      "epoch": 1.8868340140424946,
      "grad_norm": 0.4406096339225769,
      "learning_rate": 3.1394013547819986e-06,
      "loss": 0.293,
      "step": 41116
    },
    {
      "epoch": 1.8868799045477491,
      "grad_norm": 0.4873218238353729,
      "learning_rate": 3.1391737773459393e-06,
      "loss": 0.3715,
      "step": 41117
    },
    {
      "epoch": 1.8869257950530036,
      "grad_norm": 0.4864557087421417,
      "learning_rate": 3.1389462043845797e-06,
      "loss": 0.4058,
      "step": 41118
    },
    {
      "epoch": 1.8869716855582581,
      "grad_norm": 0.47774580121040344,
      "learning_rate": 3.138718635898466e-06,
      "loss": 0.3288,
      "step": 41119
    },
    {
      "epoch": 1.8870175760635126,
      "grad_norm": 0.4742848575115204,
      "learning_rate": 3.138491071888148e-06,
      "loss": 0.3673,
      "step": 41120
    },
    {
      "epoch": 1.887063466568767,
      "grad_norm": 0.5039926767349243,
      "learning_rate": 3.138263512354171e-06,
      "loss": 0.4256,
      "step": 41121
    },
    {
      "epoch": 1.8871093570740214,
      "grad_norm": 0.5023134350776672,
      "learning_rate": 3.138035957297082e-06,
      "loss": 0.3829,
      "step": 41122
    },
    {
      "epoch": 1.8871552475792759,
      "grad_norm": 0.4216409921646118,
      "learning_rate": 3.1378084067174298e-06,
      "loss": 0.3106,
      "step": 41123
    },
    {
      "epoch": 1.8872011380845302,
      "grad_norm": 0.5173583030700684,
      "learning_rate": 3.137580860615761e-06,
      "loss": 0.4224,
      "step": 41124
    },
    {
      "epoch": 1.8872470285897847,
      "grad_norm": 0.46903419494628906,
      "learning_rate": 3.137353318992622e-06,
      "loss": 0.3681,
      "step": 41125
    },
    {
      "epoch": 1.8872929190950392,
      "grad_norm": 0.4946678578853607,
      "learning_rate": 3.137125781848558e-06,
      "loss": 0.397,
      "step": 41126
    },
    {
      "epoch": 1.8873388096002937,
      "grad_norm": 0.4374212324619293,
      "learning_rate": 3.136898249184121e-06,
      "loss": 0.3165,
      "step": 41127
    },
    {
      "epoch": 1.8873847001055482,
      "grad_norm": 0.4598016142845154,
      "learning_rate": 3.136670720999856e-06,
      "loss": 0.3139,
      "step": 41128
    },
    {
      "epoch": 1.8874305906108026,
      "grad_norm": 0.4759996831417084,
      "learning_rate": 3.1364431972963083e-06,
      "loss": 0.368,
      "step": 41129
    },
    {
      "epoch": 1.8874764811160571,
      "grad_norm": 0.46867337822914124,
      "learning_rate": 3.1362156780740278e-06,
      "loss": 0.3628,
      "step": 41130
    },
    {
      "epoch": 1.8875223716213116,
      "grad_norm": 0.4929969608783722,
      "learning_rate": 3.1359881633335604e-06,
      "loss": 0.3668,
      "step": 41131
    },
    {
      "epoch": 1.8875682621265661,
      "grad_norm": 0.46970969438552856,
      "learning_rate": 3.1357606530754516e-06,
      "loss": 0.3651,
      "step": 41132
    },
    {
      "epoch": 1.8876141526318206,
      "grad_norm": 0.5175648331642151,
      "learning_rate": 3.135533147300251e-06,
      "loss": 0.4367,
      "step": 41133
    },
    {
      "epoch": 1.887660043137075,
      "grad_norm": 0.4960569441318512,
      "learning_rate": 3.1353056460085053e-06,
      "loss": 0.3887,
      "step": 41134
    },
    {
      "epoch": 1.8877059336423294,
      "grad_norm": 0.4537118971347809,
      "learning_rate": 3.1350781492007596e-06,
      "loss": 0.3115,
      "step": 41135
    },
    {
      "epoch": 1.887751824147584,
      "grad_norm": 0.45467454195022583,
      "learning_rate": 3.134850656877564e-06,
      "loss": 0.3477,
      "step": 41136
    },
    {
      "epoch": 1.8877977146528382,
      "grad_norm": 0.49898117780685425,
      "learning_rate": 3.1346231690394642e-06,
      "loss": 0.3703,
      "step": 41137
    },
    {
      "epoch": 1.8878436051580927,
      "grad_norm": 0.5164281129837036,
      "learning_rate": 3.1343956856870038e-06,
      "loss": 0.4199,
      "step": 41138
    },
    {
      "epoch": 1.8878894956633472,
      "grad_norm": 0.4712604880332947,
      "learning_rate": 3.1341682068207356e-06,
      "loss": 0.38,
      "step": 41139
    },
    {
      "epoch": 1.8879353861686017,
      "grad_norm": 0.47745200991630554,
      "learning_rate": 3.133940732441204e-06,
      "loss": 0.3138,
      "step": 41140
    },
    {
      "epoch": 1.8879812766738562,
      "grad_norm": 0.46911901235580444,
      "learning_rate": 3.133713262548955e-06,
      "loss": 0.3767,
      "step": 41141
    },
    {
      "epoch": 1.8880271671791107,
      "grad_norm": 0.43628227710723877,
      "learning_rate": 3.1334857971445383e-06,
      "loss": 0.2957,
      "step": 41142
    },
    {
      "epoch": 1.8880730576843652,
      "grad_norm": 0.49944472312927246,
      "learning_rate": 3.1332583362284983e-06,
      "loss": 0.3996,
      "step": 41143
    },
    {
      "epoch": 1.8881189481896197,
      "grad_norm": 0.5038762092590332,
      "learning_rate": 3.133030879801382e-06,
      "loss": 0.4479,
      "step": 41144
    },
    {
      "epoch": 1.8881648386948742,
      "grad_norm": 0.4964870810508728,
      "learning_rate": 3.1328034278637383e-06,
      "loss": 0.4066,
      "step": 41145
    },
    {
      "epoch": 1.8882107292001284,
      "grad_norm": 0.4704322814941406,
      "learning_rate": 3.1325759804161125e-06,
      "loss": 0.3971,
      "step": 41146
    },
    {
      "epoch": 1.888256619705383,
      "grad_norm": 0.42865249514579773,
      "learning_rate": 3.132348537459051e-06,
      "loss": 0.28,
      "step": 41147
    },
    {
      "epoch": 1.8883025102106374,
      "grad_norm": 0.5313159227371216,
      "learning_rate": 3.1321210989931026e-06,
      "loss": 0.4086,
      "step": 41148
    },
    {
      "epoch": 1.8883484007158917,
      "grad_norm": 0.47520172595977783,
      "learning_rate": 3.131893665018815e-06,
      "loss": 0.4053,
      "step": 41149
    },
    {
      "epoch": 1.8883942912211462,
      "grad_norm": 0.45428359508514404,
      "learning_rate": 3.1316662355367318e-06,
      "loss": 0.351,
      "step": 41150
    },
    {
      "epoch": 1.8884401817264007,
      "grad_norm": 0.49090445041656494,
      "learning_rate": 3.1314388105474024e-06,
      "loss": 0.4065,
      "step": 41151
    },
    {
      "epoch": 1.8884860722316552,
      "grad_norm": 0.47239595651626587,
      "learning_rate": 3.131211390051373e-06,
      "loss": 0.373,
      "step": 41152
    },
    {
      "epoch": 1.8885319627369097,
      "grad_norm": 0.4654393494129181,
      "learning_rate": 3.1309839740491888e-06,
      "loss": 0.3068,
      "step": 41153
    },
    {
      "epoch": 1.8885778532421642,
      "grad_norm": 0.464420348405838,
      "learning_rate": 3.1307565625413992e-06,
      "loss": 0.3623,
      "step": 41154
    },
    {
      "epoch": 1.8886237437474187,
      "grad_norm": 0.46534568071365356,
      "learning_rate": 3.130529155528551e-06,
      "loss": 0.3512,
      "step": 41155
    },
    {
      "epoch": 1.8886696342526732,
      "grad_norm": 0.4454861283302307,
      "learning_rate": 3.1303017530111895e-06,
      "loss": 0.2926,
      "step": 41156
    },
    {
      "epoch": 1.8887155247579277,
      "grad_norm": 0.491476446390152,
      "learning_rate": 3.130074354989863e-06,
      "loss": 0.3885,
      "step": 41157
    },
    {
      "epoch": 1.8887614152631822,
      "grad_norm": 0.4414600133895874,
      "learning_rate": 3.1298469614651173e-06,
      "loss": 0.3222,
      "step": 41158
    },
    {
      "epoch": 1.8888073057684365,
      "grad_norm": 0.4600023031234741,
      "learning_rate": 3.129619572437499e-06,
      "loss": 0.3221,
      "step": 41159
    },
    {
      "epoch": 1.888853196273691,
      "grad_norm": 0.47206708788871765,
      "learning_rate": 3.129392187907554e-06,
      "loss": 0.3631,
      "step": 41160
    },
    {
      "epoch": 1.8888990867789455,
      "grad_norm": 0.4247168302536011,
      "learning_rate": 3.129164807875832e-06,
      "loss": 0.2685,
      "step": 41161
    },
    {
      "epoch": 1.8889449772841997,
      "grad_norm": 0.45369696617126465,
      "learning_rate": 3.1289374323428777e-06,
      "loss": 0.3454,
      "step": 41162
    },
    {
      "epoch": 1.8889908677894542,
      "grad_norm": 0.503206193447113,
      "learning_rate": 3.128710061309238e-06,
      "loss": 0.4105,
      "step": 41163
    },
    {
      "epoch": 1.8890367582947087,
      "grad_norm": 0.4733751714229584,
      "learning_rate": 3.128482694775461e-06,
      "loss": 0.4266,
      "step": 41164
    },
    {
      "epoch": 1.8890826487999632,
      "grad_norm": 0.5182019472122192,
      "learning_rate": 3.1282553327420927e-06,
      "loss": 0.4561,
      "step": 41165
    },
    {
      "epoch": 1.8891285393052177,
      "grad_norm": 0.5134178400039673,
      "learning_rate": 3.1280279752096787e-06,
      "loss": 0.4314,
      "step": 41166
    },
    {
      "epoch": 1.8891744298104722,
      "grad_norm": 0.47933104634284973,
      "learning_rate": 3.127800622178767e-06,
      "loss": 0.3303,
      "step": 41167
    },
    {
      "epoch": 1.8892203203157267,
      "grad_norm": 0.47664764523506165,
      "learning_rate": 3.127573273649904e-06,
      "loss": 0.3569,
      "step": 41168
    },
    {
      "epoch": 1.8892662108209812,
      "grad_norm": 0.4786016345024109,
      "learning_rate": 3.127345929623634e-06,
      "loss": 0.3714,
      "step": 41169
    },
    {
      "epoch": 1.8893121013262357,
      "grad_norm": 0.5402268171310425,
      "learning_rate": 3.1271185901005085e-06,
      "loss": 0.274,
      "step": 41170
    },
    {
      "epoch": 1.8893579918314902,
      "grad_norm": 0.452197790145874,
      "learning_rate": 3.126891255081072e-06,
      "loss": 0.3019,
      "step": 41171
    },
    {
      "epoch": 1.8894038823367445,
      "grad_norm": 0.46202072501182556,
      "learning_rate": 3.1266639245658693e-06,
      "loss": 0.3485,
      "step": 41172
    },
    {
      "epoch": 1.889449772841999,
      "grad_norm": 0.4801209270954132,
      "learning_rate": 3.12643659855545e-06,
      "loss": 0.3533,
      "step": 41173
    },
    {
      "epoch": 1.8894956633472535,
      "grad_norm": 0.4799911677837372,
      "learning_rate": 3.1262092770503594e-06,
      "loss": 0.3139,
      "step": 41174
    },
    {
      "epoch": 1.8895415538525078,
      "grad_norm": 0.5055767893791199,
      "learning_rate": 3.1259819600511413e-06,
      "loss": 0.3819,
      "step": 41175
    },
    {
      "epoch": 1.8895874443577623,
      "grad_norm": 0.47848111391067505,
      "learning_rate": 3.1257546475583477e-06,
      "loss": 0.3693,
      "step": 41176
    },
    {
      "epoch": 1.8896333348630168,
      "grad_norm": 0.46945539116859436,
      "learning_rate": 3.125527339572523e-06,
      "loss": 0.3297,
      "step": 41177
    },
    {
      "epoch": 1.8896792253682713,
      "grad_norm": 0.4737134277820587,
      "learning_rate": 3.125300036094212e-06,
      "loss": 0.3501,
      "step": 41178
    },
    {
      "epoch": 1.8897251158735258,
      "grad_norm": 0.5007602572441101,
      "learning_rate": 3.125072737123964e-06,
      "loss": 0.3514,
      "step": 41179
    },
    {
      "epoch": 1.8897710063787803,
      "grad_norm": 0.46914568543434143,
      "learning_rate": 3.1248454426623243e-06,
      "loss": 0.3264,
      "step": 41180
    },
    {
      "epoch": 1.8898168968840348,
      "grad_norm": 0.4685961902141571,
      "learning_rate": 3.1246181527098372e-06,
      "loss": 0.3648,
      "step": 41181
    },
    {
      "epoch": 1.8898627873892893,
      "grad_norm": 0.45776647329330444,
      "learning_rate": 3.1243908672670535e-06,
      "loss": 0.3161,
      "step": 41182
    },
    {
      "epoch": 1.8899086778945438,
      "grad_norm": 0.49521705508232117,
      "learning_rate": 3.1241635863345182e-06,
      "loss": 0.3615,
      "step": 41183
    },
    {
      "epoch": 1.889954568399798,
      "grad_norm": 0.5050050020217896,
      "learning_rate": 3.123936309912776e-06,
      "loss": 0.3841,
      "step": 41184
    },
    {
      "epoch": 1.8900004589050525,
      "grad_norm": 0.4600909650325775,
      "learning_rate": 3.1237090380023766e-06,
      "loss": 0.324,
      "step": 41185
    },
    {
      "epoch": 1.890046349410307,
      "grad_norm": 0.44461163878440857,
      "learning_rate": 3.1234817706038646e-06,
      "loss": 0.3337,
      "step": 41186
    },
    {
      "epoch": 1.8900922399155615,
      "grad_norm": 0.47812700271606445,
      "learning_rate": 3.123254507717784e-06,
      "loss": 0.3798,
      "step": 41187
    },
    {
      "epoch": 1.8901381304208158,
      "grad_norm": 0.4376629590988159,
      "learning_rate": 3.123027249344688e-06,
      "loss": 0.3105,
      "step": 41188
    },
    {
      "epoch": 1.8901840209260703,
      "grad_norm": 0.4467106759548187,
      "learning_rate": 3.1227999954851174e-06,
      "loss": 0.3124,
      "step": 41189
    },
    {
      "epoch": 1.8902299114313248,
      "grad_norm": 0.45303601026535034,
      "learning_rate": 3.1225727461396207e-06,
      "loss": 0.335,
      "step": 41190
    },
    {
      "epoch": 1.8902758019365793,
      "grad_norm": 0.43162739276885986,
      "learning_rate": 3.1223455013087413e-06,
      "loss": 0.3018,
      "step": 41191
    },
    {
      "epoch": 1.8903216924418338,
      "grad_norm": 0.5173735618591309,
      "learning_rate": 3.122118260993031e-06,
      "loss": 0.4223,
      "step": 41192
    },
    {
      "epoch": 1.8903675829470883,
      "grad_norm": 0.49723488092422485,
      "learning_rate": 3.1218910251930334e-06,
      "loss": 0.4246,
      "step": 41193
    },
    {
      "epoch": 1.8904134734523428,
      "grad_norm": 0.47979727387428284,
      "learning_rate": 3.121663793909294e-06,
      "loss": 0.3515,
      "step": 41194
    },
    {
      "epoch": 1.8904593639575973,
      "grad_norm": 0.47939544916152954,
      "learning_rate": 3.1214365671423613e-06,
      "loss": 0.3536,
      "step": 41195
    },
    {
      "epoch": 1.8905052544628518,
      "grad_norm": 0.489620178937912,
      "learning_rate": 3.121209344892781e-06,
      "loss": 0.3915,
      "step": 41196
    },
    {
      "epoch": 1.890551144968106,
      "grad_norm": 0.42947274446487427,
      "learning_rate": 3.1209821271610975e-06,
      "loss": 0.2993,
      "step": 41197
    },
    {
      "epoch": 1.8905970354733606,
      "grad_norm": 0.47395825386047363,
      "learning_rate": 3.1207549139478604e-06,
      "loss": 0.3827,
      "step": 41198
    },
    {
      "epoch": 1.890642925978615,
      "grad_norm": 0.45382463932037354,
      "learning_rate": 3.1205277052536153e-06,
      "loss": 0.3151,
      "step": 41199
    },
    {
      "epoch": 1.8906888164838693,
      "grad_norm": 0.4740877151489258,
      "learning_rate": 3.1203005010789062e-06,
      "loss": 0.35,
      "step": 41200
    },
    {
      "epoch": 1.8907347069891238,
      "grad_norm": 0.49499279260635376,
      "learning_rate": 3.1200733014242822e-06,
      "loss": 0.3585,
      "step": 41201
    },
    {
      "epoch": 1.8907805974943783,
      "grad_norm": 0.49770835041999817,
      "learning_rate": 3.119846106290289e-06,
      "loss": 0.4134,
      "step": 41202
    },
    {
      "epoch": 1.8908264879996328,
      "grad_norm": 0.43906986713409424,
      "learning_rate": 3.1196189156774704e-06,
      "loss": 0.2694,
      "step": 41203
    },
    {
      "epoch": 1.8908723785048873,
      "grad_norm": 0.4907604157924652,
      "learning_rate": 3.119391729586377e-06,
      "loss": 0.4021,
      "step": 41204
    },
    {
      "epoch": 1.8909182690101418,
      "grad_norm": 0.4510626196861267,
      "learning_rate": 3.1191645480175525e-06,
      "loss": 0.3285,
      "step": 41205
    },
    {
      "epoch": 1.8909641595153963,
      "grad_norm": 0.46524322032928467,
      "learning_rate": 3.118937370971543e-06,
      "loss": 0.3512,
      "step": 41206
    },
    {
      "epoch": 1.8910100500206508,
      "grad_norm": 0.443034827709198,
      "learning_rate": 3.1187101984488964e-06,
      "loss": 0.301,
      "step": 41207
    },
    {
      "epoch": 1.8910559405259053,
      "grad_norm": 0.438945472240448,
      "learning_rate": 3.1184830304501577e-06,
      "loss": 0.2988,
      "step": 41208
    },
    {
      "epoch": 1.8911018310311598,
      "grad_norm": 0.4736744165420532,
      "learning_rate": 3.1182558669758722e-06,
      "loss": 0.3938,
      "step": 41209
    },
    {
      "epoch": 1.891147721536414,
      "grad_norm": 0.4627220630645752,
      "learning_rate": 3.1180287080265896e-06,
      "loss": 0.3261,
      "step": 41210
    },
    {
      "epoch": 1.8911936120416686,
      "grad_norm": 0.4554779827594757,
      "learning_rate": 3.1178015536028537e-06,
      "loss": 0.3777,
      "step": 41211
    },
    {
      "epoch": 1.891239502546923,
      "grad_norm": 0.4913612902164459,
      "learning_rate": 3.1175744037052087e-06,
      "loss": 0.3542,
      "step": 41212
    },
    {
      "epoch": 1.8912853930521774,
      "grad_norm": 0.46813344955444336,
      "learning_rate": 3.1173472583342052e-06,
      "loss": 0.3623,
      "step": 41213
    },
    {
      "epoch": 1.8913312835574319,
      "grad_norm": 0.47300752997398376,
      "learning_rate": 3.1171201174903874e-06,
      "loss": 0.3688,
      "step": 41214
    },
    {
      "epoch": 1.8913771740626864,
      "grad_norm": 0.4756969213485718,
      "learning_rate": 3.1168929811743e-06,
      "loss": 0.3585,
      "step": 41215
    },
    {
      "epoch": 1.8914230645679408,
      "grad_norm": 0.49719470739364624,
      "learning_rate": 3.116665849386492e-06,
      "loss": 0.3913,
      "step": 41216
    },
    {
      "epoch": 1.8914689550731953,
      "grad_norm": 0.4546404778957367,
      "learning_rate": 3.116438722127508e-06,
      "loss": 0.3677,
      "step": 41217
    },
    {
      "epoch": 1.8915148455784498,
      "grad_norm": 0.49796062707901,
      "learning_rate": 3.1162115993978935e-06,
      "loss": 0.3948,
      "step": 41218
    },
    {
      "epoch": 1.8915607360837043,
      "grad_norm": 0.45443177223205566,
      "learning_rate": 3.1159844811981966e-06,
      "loss": 0.3339,
      "step": 41219
    },
    {
      "epoch": 1.8916066265889588,
      "grad_norm": 0.4904322028160095,
      "learning_rate": 3.1157573675289627e-06,
      "loss": 0.4108,
      "step": 41220
    },
    {
      "epoch": 1.8916525170942133,
      "grad_norm": 0.48188769817352295,
      "learning_rate": 3.1155302583907367e-06,
      "loss": 0.3704,
      "step": 41221
    },
    {
      "epoch": 1.8916984075994678,
      "grad_norm": 0.4205312132835388,
      "learning_rate": 3.115303153784066e-06,
      "loss": 0.2688,
      "step": 41222
    },
    {
      "epoch": 1.891744298104722,
      "grad_norm": 0.4634118676185608,
      "learning_rate": 3.115076053709497e-06,
      "loss": 0.4037,
      "step": 41223
    },
    {
      "epoch": 1.8917901886099766,
      "grad_norm": 0.45413535833358765,
      "learning_rate": 3.114848958167575e-06,
      "loss": 0.3101,
      "step": 41224
    },
    {
      "epoch": 1.891836079115231,
      "grad_norm": 0.5066378116607666,
      "learning_rate": 3.1146218671588445e-06,
      "loss": 0.4179,
      "step": 41225
    },
    {
      "epoch": 1.8918819696204854,
      "grad_norm": 0.5349546074867249,
      "learning_rate": 3.114394780683855e-06,
      "loss": 0.4787,
      "step": 41226
    },
    {
      "epoch": 1.8919278601257399,
      "grad_norm": 0.465223103761673,
      "learning_rate": 3.1141676987431512e-06,
      "loss": 0.3077,
      "step": 41227
    },
    {
      "epoch": 1.8919737506309944,
      "grad_norm": 0.4973445534706116,
      "learning_rate": 3.113940621337278e-06,
      "loss": 0.3871,
      "step": 41228
    },
    {
      "epoch": 1.8920196411362489,
      "grad_norm": 0.5124292969703674,
      "learning_rate": 3.113713548466783e-06,
      "loss": 0.3834,
      "step": 41229
    },
    {
      "epoch": 1.8920655316415034,
      "grad_norm": 0.46148982644081116,
      "learning_rate": 3.1134864801322117e-06,
      "loss": 0.3496,
      "step": 41230
    },
    {
      "epoch": 1.8921114221467579,
      "grad_norm": 0.4394114911556244,
      "learning_rate": 3.113259416334108e-06,
      "loss": 0.2817,
      "step": 41231
    },
    {
      "epoch": 1.8921573126520124,
      "grad_norm": 0.48006749153137207,
      "learning_rate": 3.1130323570730236e-06,
      "loss": 0.3798,
      "step": 41232
    },
    {
      "epoch": 1.8922032031572669,
      "grad_norm": 0.4865095317363739,
      "learning_rate": 3.1128053023494985e-06,
      "loss": 0.3768,
      "step": 41233
    },
    {
      "epoch": 1.8922490936625214,
      "grad_norm": 0.47790881991386414,
      "learning_rate": 3.11257825216408e-06,
      "loss": 0.3464,
      "step": 41234
    },
    {
      "epoch": 1.8922949841677756,
      "grad_norm": 0.4721854329109192,
      "learning_rate": 3.112351206517317e-06,
      "loss": 0.3443,
      "step": 41235
    },
    {
      "epoch": 1.8923408746730301,
      "grad_norm": 0.4781986176967621,
      "learning_rate": 3.1121241654097536e-06,
      "loss": 0.3303,
      "step": 41236
    },
    {
      "epoch": 1.8923867651782846,
      "grad_norm": 0.46346694231033325,
      "learning_rate": 3.1118971288419346e-06,
      "loss": 0.3485,
      "step": 41237
    },
    {
      "epoch": 1.892432655683539,
      "grad_norm": 0.49548622965812683,
      "learning_rate": 3.1116700968144088e-06,
      "loss": 0.3765,
      "step": 41238
    },
    {
      "epoch": 1.8924785461887934,
      "grad_norm": 0.519116997718811,
      "learning_rate": 3.1114430693277194e-06,
      "loss": 0.3722,
      "step": 41239
    },
    {
      "epoch": 1.892524436694048,
      "grad_norm": 0.46004951000213623,
      "learning_rate": 3.1112160463824116e-06,
      "loss": 0.3107,
      "step": 41240
    },
    {
      "epoch": 1.8925703271993024,
      "grad_norm": 0.5163196921348572,
      "learning_rate": 3.1109890279790355e-06,
      "loss": 0.3601,
      "step": 41241
    },
    {
      "epoch": 1.892616217704557,
      "grad_norm": 0.4503540098667145,
      "learning_rate": 3.110762014118134e-06,
      "loss": 0.3353,
      "step": 41242
    },
    {
      "epoch": 1.8926621082098114,
      "grad_norm": 0.47328004240989685,
      "learning_rate": 3.110535004800253e-06,
      "loss": 0.3617,
      "step": 41243
    },
    {
      "epoch": 1.892707998715066,
      "grad_norm": 0.5275623798370361,
      "learning_rate": 3.11030800002594e-06,
      "loss": 0.4354,
      "step": 41244
    },
    {
      "epoch": 1.8927538892203204,
      "grad_norm": 0.4528310298919678,
      "learning_rate": 3.11008099979574e-06,
      "loss": 0.298,
      "step": 41245
    },
    {
      "epoch": 1.892799779725575,
      "grad_norm": 0.512881338596344,
      "learning_rate": 3.1098540041101965e-06,
      "loss": 0.3709,
      "step": 41246
    },
    {
      "epoch": 1.8928456702308294,
      "grad_norm": 0.5487818121910095,
      "learning_rate": 3.109627012969859e-06,
      "loss": 0.3622,
      "step": 41247
    },
    {
      "epoch": 1.8928915607360837,
      "grad_norm": 0.5063799619674683,
      "learning_rate": 3.1094000263752734e-06,
      "loss": 0.3653,
      "step": 41248
    },
    {
      "epoch": 1.8929374512413382,
      "grad_norm": 0.45389240980148315,
      "learning_rate": 3.1091730443269817e-06,
      "loss": 0.3706,
      "step": 41249
    },
    {
      "epoch": 1.8929833417465927,
      "grad_norm": 0.4750155508518219,
      "learning_rate": 3.108946066825533e-06,
      "loss": 0.3347,
      "step": 41250
    },
    {
      "epoch": 1.893029232251847,
      "grad_norm": 0.4781514108181,
      "learning_rate": 3.1087190938714733e-06,
      "loss": 0.411,
      "step": 41251
    },
    {
      "epoch": 1.8930751227571014,
      "grad_norm": 0.47568875551223755,
      "learning_rate": 3.1084921254653443e-06,
      "loss": 0.3692,
      "step": 41252
    },
    {
      "epoch": 1.893121013262356,
      "grad_norm": 0.4677448570728302,
      "learning_rate": 3.1082651616076975e-06,
      "loss": 0.3596,
      "step": 41253
    },
    {
      "epoch": 1.8931669037676104,
      "grad_norm": 0.4624066948890686,
      "learning_rate": 3.1080382022990763e-06,
      "loss": 0.3178,
      "step": 41254
    },
    {
      "epoch": 1.893212794272865,
      "grad_norm": 0.4592594504356384,
      "learning_rate": 3.107811247540026e-06,
      "loss": 0.3263,
      "step": 41255
    },
    {
      "epoch": 1.8932586847781194,
      "grad_norm": 0.44488954544067383,
      "learning_rate": 3.1075842973310893e-06,
      "loss": 0.3203,
      "step": 41256
    },
    {
      "epoch": 1.893304575283374,
      "grad_norm": 0.46758103370666504,
      "learning_rate": 3.107357351672818e-06,
      "loss": 0.3454,
      "step": 41257
    },
    {
      "epoch": 1.8933504657886284,
      "grad_norm": 0.4891447424888611,
      "learning_rate": 3.107130410565754e-06,
      "loss": 0.3765,
      "step": 41258
    },
    {
      "epoch": 1.893396356293883,
      "grad_norm": 0.4446899890899658,
      "learning_rate": 3.1069034740104443e-06,
      "loss": 0.2949,
      "step": 41259
    },
    {
      "epoch": 1.8934422467991374,
      "grad_norm": 0.49757182598114014,
      "learning_rate": 3.106676542007434e-06,
      "loss": 0.4049,
      "step": 41260
    },
    {
      "epoch": 1.8934881373043917,
      "grad_norm": 0.45456501841545105,
      "learning_rate": 3.10644961455727e-06,
      "loss": 0.328,
      "step": 41261
    },
    {
      "epoch": 1.8935340278096462,
      "grad_norm": 0.4532433748245239,
      "learning_rate": 3.106222691660494e-06,
      "loss": 0.3442,
      "step": 41262
    },
    {
      "epoch": 1.8935799183149007,
      "grad_norm": 0.44578391313552856,
      "learning_rate": 3.1059957733176575e-06,
      "loss": 0.3056,
      "step": 41263
    },
    {
      "epoch": 1.893625808820155,
      "grad_norm": 0.4677031338214874,
      "learning_rate": 3.1057688595293035e-06,
      "loss": 0.405,
      "step": 41264
    },
    {
      "epoch": 1.8936716993254095,
      "grad_norm": 0.45704105496406555,
      "learning_rate": 3.105541950295976e-06,
      "loss": 0.3168,
      "step": 41265
    },
    {
      "epoch": 1.893717589830664,
      "grad_norm": 0.5308483242988586,
      "learning_rate": 3.1053150456182235e-06,
      "loss": 0.4504,
      "step": 41266
    },
    {
      "epoch": 1.8937634803359185,
      "grad_norm": 0.4926720857620239,
      "learning_rate": 3.1050881454965902e-06,
      "loss": 0.4562,
      "step": 41267
    },
    {
      "epoch": 1.893809370841173,
      "grad_norm": 0.46268779039382935,
      "learning_rate": 3.10486124993162e-06,
      "loss": 0.334,
      "step": 41268
    },
    {
      "epoch": 1.8938552613464275,
      "grad_norm": 0.48834484815597534,
      "learning_rate": 3.1046343589238624e-06,
      "loss": 0.3575,
      "step": 41269
    },
    {
      "epoch": 1.893901151851682,
      "grad_norm": 0.44369015097618103,
      "learning_rate": 3.1044074724738614e-06,
      "loss": 0.3197,
      "step": 41270
    },
    {
      "epoch": 1.8939470423569364,
      "grad_norm": 0.4686090350151062,
      "learning_rate": 3.10418059058216e-06,
      "loss": 0.3683,
      "step": 41271
    },
    {
      "epoch": 1.893992932862191,
      "grad_norm": 0.45741167664527893,
      "learning_rate": 3.103953713249308e-06,
      "loss": 0.3265,
      "step": 41272
    },
    {
      "epoch": 1.8940388233674452,
      "grad_norm": 0.4879246652126312,
      "learning_rate": 3.1037268404758487e-06,
      "loss": 0.3848,
      "step": 41273
    },
    {
      "epoch": 1.8940847138726997,
      "grad_norm": 0.6793796420097351,
      "learning_rate": 3.1034999722623256e-06,
      "loss": 0.4359,
      "step": 41274
    },
    {
      "epoch": 1.8941306043779542,
      "grad_norm": 0.4598791003227234,
      "learning_rate": 3.103273108609289e-06,
      "loss": 0.3038,
      "step": 41275
    },
    {
      "epoch": 1.8941764948832087,
      "grad_norm": 0.48693540692329407,
      "learning_rate": 3.103046249517282e-06,
      "loss": 0.3599,
      "step": 41276
    },
    {
      "epoch": 1.894222385388463,
      "grad_norm": 0.42047783732414246,
      "learning_rate": 3.102819394986848e-06,
      "loss": 0.2703,
      "step": 41277
    },
    {
      "epoch": 1.8942682758937175,
      "grad_norm": 0.488725870847702,
      "learning_rate": 3.102592545018536e-06,
      "loss": 0.3713,
      "step": 41278
    },
    {
      "epoch": 1.894314166398972,
      "grad_norm": 0.41957828402519226,
      "learning_rate": 3.1023656996128905e-06,
      "loss": 0.2784,
      "step": 41279
    },
    {
      "epoch": 1.8943600569042265,
      "grad_norm": 0.5231638550758362,
      "learning_rate": 3.1021388587704553e-06,
      "loss": 0.4678,
      "step": 41280
    },
    {
      "epoch": 1.894405947409481,
      "grad_norm": 0.49268701672554016,
      "learning_rate": 3.1019120224917785e-06,
      "loss": 0.3901,
      "step": 41281
    },
    {
      "epoch": 1.8944518379147355,
      "grad_norm": 0.4971865713596344,
      "learning_rate": 3.1016851907774036e-06,
      "loss": 0.3854,
      "step": 41282
    },
    {
      "epoch": 1.89449772841999,
      "grad_norm": 0.4569043219089508,
      "learning_rate": 3.101458363627875e-06,
      "loss": 0.3276,
      "step": 41283
    },
    {
      "epoch": 1.8945436189252445,
      "grad_norm": 0.4339468777179718,
      "learning_rate": 3.101231541043742e-06,
      "loss": 0.2848,
      "step": 41284
    },
    {
      "epoch": 1.894589509430499,
      "grad_norm": 0.4613548517227173,
      "learning_rate": 3.101004723025548e-06,
      "loss": 0.335,
      "step": 41285
    },
    {
      "epoch": 1.8946353999357533,
      "grad_norm": 0.46577420830726624,
      "learning_rate": 3.100777909573837e-06,
      "loss": 0.3385,
      "step": 41286
    },
    {
      "epoch": 1.8946812904410077,
      "grad_norm": 0.5025984048843384,
      "learning_rate": 3.100551100689157e-06,
      "loss": 0.3984,
      "step": 41287
    },
    {
      "epoch": 1.8947271809462622,
      "grad_norm": 0.45395809412002563,
      "learning_rate": 3.100324296372052e-06,
      "loss": 0.2846,
      "step": 41288
    },
    {
      "epoch": 1.8947730714515165,
      "grad_norm": 0.616338312625885,
      "learning_rate": 3.100097496623068e-06,
      "loss": 0.3821,
      "step": 41289
    },
    {
      "epoch": 1.894818961956771,
      "grad_norm": 0.5045035481452942,
      "learning_rate": 3.0998707014427463e-06,
      "loss": 0.4621,
      "step": 41290
    },
    {
      "epoch": 1.8948648524620255,
      "grad_norm": 0.5059893131256104,
      "learning_rate": 3.0996439108316396e-06,
      "loss": 0.4147,
      "step": 41291
    },
    {
      "epoch": 1.89491074296728,
      "grad_norm": 0.4637081027030945,
      "learning_rate": 3.0994171247902887e-06,
      "loss": 0.3254,
      "step": 41292
    },
    {
      "epoch": 1.8949566334725345,
      "grad_norm": 0.4381686747074127,
      "learning_rate": 3.099190343319239e-06,
      "loss": 0.3061,
      "step": 41293
    },
    {
      "epoch": 1.895002523977789,
      "grad_norm": 0.4915980398654938,
      "learning_rate": 3.0989635664190374e-06,
      "loss": 0.3628,
      "step": 41294
    },
    {
      "epoch": 1.8950484144830435,
      "grad_norm": 0.44747138023376465,
      "learning_rate": 3.0987367940902284e-06,
      "loss": 0.2774,
      "step": 41295
    },
    {
      "epoch": 1.895094304988298,
      "grad_norm": 0.46230950951576233,
      "learning_rate": 3.0985100263333555e-06,
      "loss": 0.3386,
      "step": 41296
    },
    {
      "epoch": 1.8951401954935525,
      "grad_norm": 0.4835541844367981,
      "learning_rate": 3.0982832631489677e-06,
      "loss": 0.3719,
      "step": 41297
    },
    {
      "epoch": 1.895186085998807,
      "grad_norm": 0.47275274991989136,
      "learning_rate": 3.09805650453761e-06,
      "loss": 0.3345,
      "step": 41298
    },
    {
      "epoch": 1.8952319765040613,
      "grad_norm": 0.4877709746360779,
      "learning_rate": 3.0978297504998226e-06,
      "loss": 0.3793,
      "step": 41299
    },
    {
      "epoch": 1.8952778670093158,
      "grad_norm": 0.4718673825263977,
      "learning_rate": 3.097603001036157e-06,
      "loss": 0.316,
      "step": 41300
    },
    {
      "epoch": 1.8953237575145703,
      "grad_norm": 0.5149658918380737,
      "learning_rate": 3.097376256147155e-06,
      "loss": 0.3889,
      "step": 41301
    },
    {
      "epoch": 1.8953696480198245,
      "grad_norm": 0.47058600187301636,
      "learning_rate": 3.097149515833362e-06,
      "loss": 0.3772,
      "step": 41302
    },
    {
      "epoch": 1.895415538525079,
      "grad_norm": 0.5015552043914795,
      "learning_rate": 3.0969227800953242e-06,
      "loss": 0.4107,
      "step": 41303
    },
    {
      "epoch": 1.8954614290303335,
      "grad_norm": 0.49075189232826233,
      "learning_rate": 3.0966960489335874e-06,
      "loss": 0.415,
      "step": 41304
    },
    {
      "epoch": 1.895507319535588,
      "grad_norm": 0.48438775539398193,
      "learning_rate": 3.0964693223486936e-06,
      "loss": 0.3718,
      "step": 41305
    },
    {
      "epoch": 1.8955532100408425,
      "grad_norm": 0.5056449174880981,
      "learning_rate": 3.096242600341192e-06,
      "loss": 0.4169,
      "step": 41306
    },
    {
      "epoch": 1.895599100546097,
      "grad_norm": 0.47764265537261963,
      "learning_rate": 3.096015882911626e-06,
      "loss": 0.3948,
      "step": 41307
    },
    {
      "epoch": 1.8956449910513515,
      "grad_norm": 0.455504447221756,
      "learning_rate": 3.0957891700605404e-06,
      "loss": 0.3168,
      "step": 41308
    },
    {
      "epoch": 1.895690881556606,
      "grad_norm": 0.4424023926258087,
      "learning_rate": 3.0955624617884815e-06,
      "loss": 0.3338,
      "step": 41309
    },
    {
      "epoch": 1.8957367720618605,
      "grad_norm": 0.4814959466457367,
      "learning_rate": 3.095335758095994e-06,
      "loss": 0.4062,
      "step": 41310
    },
    {
      "epoch": 1.895782662567115,
      "grad_norm": 0.47504523396492004,
      "learning_rate": 3.0951090589836215e-06,
      "loss": 0.3702,
      "step": 41311
    },
    {
      "epoch": 1.8958285530723693,
      "grad_norm": 0.47444871068000793,
      "learning_rate": 3.0948823644519118e-06,
      "loss": 0.3807,
      "step": 41312
    },
    {
      "epoch": 1.8958744435776238,
      "grad_norm": 0.4389927387237549,
      "learning_rate": 3.094655674501409e-06,
      "loss": 0.3003,
      "step": 41313
    },
    {
      "epoch": 1.8959203340828783,
      "grad_norm": 0.448261559009552,
      "learning_rate": 3.0944289891326564e-06,
      "loss": 0.323,
      "step": 41314
    },
    {
      "epoch": 1.8959662245881326,
      "grad_norm": 0.5008325576782227,
      "learning_rate": 3.094202308346202e-06,
      "loss": 0.3856,
      "step": 41315
    },
    {
      "epoch": 1.896012115093387,
      "grad_norm": 0.48241421580314636,
      "learning_rate": 3.0939756321425896e-06,
      "loss": 0.4205,
      "step": 41316
    },
    {
      "epoch": 1.8960580055986416,
      "grad_norm": 0.46519938111305237,
      "learning_rate": 3.0937489605223625e-06,
      "loss": 0.3458,
      "step": 41317
    },
    {
      "epoch": 1.896103896103896,
      "grad_norm": 0.5029230713844299,
      "learning_rate": 3.0935222934860686e-06,
      "loss": 0.4233,
      "step": 41318
    },
    {
      "epoch": 1.8961497866091506,
      "grad_norm": 0.520279586315155,
      "learning_rate": 3.093295631034253e-06,
      "loss": 0.4367,
      "step": 41319
    },
    {
      "epoch": 1.896195677114405,
      "grad_norm": 0.48823970556259155,
      "learning_rate": 3.093068973167458e-06,
      "loss": 0.4057,
      "step": 41320
    },
    {
      "epoch": 1.8962415676196596,
      "grad_norm": 0.492681086063385,
      "learning_rate": 3.0928423198862317e-06,
      "loss": 0.3797,
      "step": 41321
    },
    {
      "epoch": 1.896287458124914,
      "grad_norm": 0.4697648584842682,
      "learning_rate": 3.092615671191117e-06,
      "loss": 0.351,
      "step": 41322
    },
    {
      "epoch": 1.8963333486301686,
      "grad_norm": 0.45003649592399597,
      "learning_rate": 3.0923890270826606e-06,
      "loss": 0.3104,
      "step": 41323
    },
    {
      "epoch": 1.8963792391354228,
      "grad_norm": 0.4379443824291229,
      "learning_rate": 3.092162387561405e-06,
      "loss": 0.2739,
      "step": 41324
    },
    {
      "epoch": 1.8964251296406773,
      "grad_norm": 0.4780958592891693,
      "learning_rate": 3.091935752627898e-06,
      "loss": 0.4001,
      "step": 41325
    },
    {
      "epoch": 1.8964710201459318,
      "grad_norm": 0.5721418857574463,
      "learning_rate": 3.0917091222826827e-06,
      "loss": 0.3488,
      "step": 41326
    },
    {
      "epoch": 1.896516910651186,
      "grad_norm": 0.4809184968471527,
      "learning_rate": 3.0914824965263035e-06,
      "loss": 0.3193,
      "step": 41327
    },
    {
      "epoch": 1.8965628011564406,
      "grad_norm": 0.47580093145370483,
      "learning_rate": 3.091255875359308e-06,
      "loss": 0.3565,
      "step": 41328
    },
    {
      "epoch": 1.896608691661695,
      "grad_norm": 0.42898058891296387,
      "learning_rate": 3.0910292587822398e-06,
      "loss": 0.2872,
      "step": 41329
    },
    {
      "epoch": 1.8966545821669496,
      "grad_norm": 0.5047294497489929,
      "learning_rate": 3.0908026467956424e-06,
      "loss": 0.369,
      "step": 41330
    },
    {
      "epoch": 1.896700472672204,
      "grad_norm": 0.532032310962677,
      "learning_rate": 3.0905760394000635e-06,
      "loss": 0.4555,
      "step": 41331
    },
    {
      "epoch": 1.8967463631774586,
      "grad_norm": 0.4287479817867279,
      "learning_rate": 3.090349436596046e-06,
      "loss": 0.3268,
      "step": 41332
    },
    {
      "epoch": 1.896792253682713,
      "grad_norm": 0.49266374111175537,
      "learning_rate": 3.090122838384134e-06,
      "loss": 0.3498,
      "step": 41333
    },
    {
      "epoch": 1.8968381441879676,
      "grad_norm": 0.4340055286884308,
      "learning_rate": 3.089896244764876e-06,
      "loss": 0.3046,
      "step": 41334
    },
    {
      "epoch": 1.896884034693222,
      "grad_norm": 0.49657538533210754,
      "learning_rate": 3.0896696557388143e-06,
      "loss": 0.3749,
      "step": 41335
    },
    {
      "epoch": 1.8969299251984766,
      "grad_norm": 0.4315204918384552,
      "learning_rate": 3.089443071306494e-06,
      "loss": 0.2831,
      "step": 41336
    },
    {
      "epoch": 1.8969758157037309,
      "grad_norm": 0.4750882685184479,
      "learning_rate": 3.08921649146846e-06,
      "loss": 0.3881,
      "step": 41337
    },
    {
      "epoch": 1.8970217062089854,
      "grad_norm": 0.4684036374092102,
      "learning_rate": 3.088989916225258e-06,
      "loss": 0.3728,
      "step": 41338
    },
    {
      "epoch": 1.8970675967142399,
      "grad_norm": 0.4727766215801239,
      "learning_rate": 3.088763345577429e-06,
      "loss": 0.3652,
      "step": 41339
    },
    {
      "epoch": 1.8971134872194941,
      "grad_norm": 0.4784647226333618,
      "learning_rate": 3.088536779525524e-06,
      "loss": 0.3527,
      "step": 41340
    },
    {
      "epoch": 1.8971593777247486,
      "grad_norm": 0.4774829149246216,
      "learning_rate": 3.088310218070084e-06,
      "loss": 0.3589,
      "step": 41341
    },
    {
      "epoch": 1.8972052682300031,
      "grad_norm": 0.466403603553772,
      "learning_rate": 3.088083661211655e-06,
      "loss": 0.3461,
      "step": 41342
    },
    {
      "epoch": 1.8972511587352576,
      "grad_norm": 0.46614235639572144,
      "learning_rate": 3.0878571089507813e-06,
      "loss": 0.347,
      "step": 41343
    },
    {
      "epoch": 1.8972970492405121,
      "grad_norm": 0.4803677499294281,
      "learning_rate": 3.0876305612880074e-06,
      "loss": 0.3675,
      "step": 41344
    },
    {
      "epoch": 1.8973429397457666,
      "grad_norm": 0.4387318789958954,
      "learning_rate": 3.0874040182238787e-06,
      "loss": 0.2991,
      "step": 41345
    },
    {
      "epoch": 1.8973888302510211,
      "grad_norm": 0.45115095376968384,
      "learning_rate": 3.0871774797589394e-06,
      "loss": 0.3493,
      "step": 41346
    },
    {
      "epoch": 1.8974347207562756,
      "grad_norm": 0.4409339427947998,
      "learning_rate": 3.086950945893735e-06,
      "loss": 0.3319,
      "step": 41347
    },
    {
      "epoch": 1.8974806112615301,
      "grad_norm": 0.4229160249233246,
      "learning_rate": 3.086724416628808e-06,
      "loss": 0.2514,
      "step": 41348
    },
    {
      "epoch": 1.8975265017667846,
      "grad_norm": 0.4249306917190552,
      "learning_rate": 3.086497891964707e-06,
      "loss": 0.2648,
      "step": 41349
    },
    {
      "epoch": 1.8975723922720389,
      "grad_norm": 0.44979265332221985,
      "learning_rate": 3.0862713719019745e-06,
      "loss": 0.2969,
      "step": 41350
    },
    {
      "epoch": 1.8976182827772934,
      "grad_norm": 0.5150836706161499,
      "learning_rate": 3.0860448564411545e-06,
      "loss": 0.4053,
      "step": 41351
    },
    {
      "epoch": 1.8976641732825479,
      "grad_norm": 0.4928245544433594,
      "learning_rate": 3.0858183455827935e-06,
      "loss": 0.3899,
      "step": 41352
    },
    {
      "epoch": 1.8977100637878022,
      "grad_norm": 0.5218529105186462,
      "learning_rate": 3.085591839327435e-06,
      "loss": 0.4374,
      "step": 41353
    },
    {
      "epoch": 1.8977559542930567,
      "grad_norm": 0.47860127687454224,
      "learning_rate": 3.0853653376756245e-06,
      "loss": 0.4128,
      "step": 41354
    },
    {
      "epoch": 1.8978018447983112,
      "grad_norm": 0.4865467846393585,
      "learning_rate": 3.085138840627903e-06,
      "loss": 0.3441,
      "step": 41355
    },
    {
      "epoch": 1.8978477353035657,
      "grad_norm": 0.4956628382205963,
      "learning_rate": 3.0849123481848207e-06,
      "loss": 0.3607,
      "step": 41356
    },
    {
      "epoch": 1.8978936258088202,
      "grad_norm": 0.4887450635433197,
      "learning_rate": 3.0846858603469204e-06,
      "loss": 0.3904,
      "step": 41357
    },
    {
      "epoch": 1.8979395163140746,
      "grad_norm": 0.48269015550613403,
      "learning_rate": 3.084459377114745e-06,
      "loss": 0.396,
      "step": 41358
    },
    {
      "epoch": 1.8979854068193291,
      "grad_norm": 0.4926266372203827,
      "learning_rate": 3.0842328984888402e-06,
      "loss": 0.3542,
      "step": 41359
    },
    {
      "epoch": 1.8980312973245836,
      "grad_norm": 0.4579259753227234,
      "learning_rate": 3.0840064244697522e-06,
      "loss": 0.3229,
      "step": 41360
    },
    {
      "epoch": 1.8980771878298381,
      "grad_norm": 0.48977261781692505,
      "learning_rate": 3.0837799550580203e-06,
      "loss": 0.3927,
      "step": 41361
    },
    {
      "epoch": 1.8981230783350924,
      "grad_norm": 0.471478670835495,
      "learning_rate": 3.0835534902541963e-06,
      "loss": 0.3622,
      "step": 41362
    },
    {
      "epoch": 1.898168968840347,
      "grad_norm": 0.47682449221611023,
      "learning_rate": 3.0833270300588207e-06,
      "loss": 0.3209,
      "step": 41363
    },
    {
      "epoch": 1.8982148593456014,
      "grad_norm": 0.44824767112731934,
      "learning_rate": 3.083100574472438e-06,
      "loss": 0.3045,
      "step": 41364
    },
    {
      "epoch": 1.898260749850856,
      "grad_norm": 0.4394513666629791,
      "learning_rate": 3.082874123495594e-06,
      "loss": 0.3019,
      "step": 41365
    },
    {
      "epoch": 1.8983066403561102,
      "grad_norm": 0.59175705909729,
      "learning_rate": 3.082647677128833e-06,
      "loss": 0.317,
      "step": 41366
    },
    {
      "epoch": 1.8983525308613647,
      "grad_norm": 0.4853340983390808,
      "learning_rate": 3.082421235372698e-06,
      "loss": 0.3961,
      "step": 41367
    },
    {
      "epoch": 1.8983984213666192,
      "grad_norm": 0.44530442357063293,
      "learning_rate": 3.0821947982277357e-06,
      "loss": 0.3147,
      "step": 41368
    },
    {
      "epoch": 1.8984443118718737,
      "grad_norm": 0.5093839168548584,
      "learning_rate": 3.0819683656944897e-06,
      "loss": 0.3993,
      "step": 41369
    },
    {
      "epoch": 1.8984902023771282,
      "grad_norm": 0.47955745458602905,
      "learning_rate": 3.0817419377735025e-06,
      "loss": 0.3868,
      "step": 41370
    },
    {
      "epoch": 1.8985360928823827,
      "grad_norm": 0.45728230476379395,
      "learning_rate": 3.0815155144653224e-06,
      "loss": 0.3433,
      "step": 41371
    },
    {
      "epoch": 1.8985819833876372,
      "grad_norm": 0.5199812650680542,
      "learning_rate": 3.081289095770492e-06,
      "loss": 0.4399,
      "step": 41372
    },
    {
      "epoch": 1.8986278738928917,
      "grad_norm": 0.4781954288482666,
      "learning_rate": 3.081062681689555e-06,
      "loss": 0.3955,
      "step": 41373
    },
    {
      "epoch": 1.8986737643981462,
      "grad_norm": 0.46593138575553894,
      "learning_rate": 3.0808362722230573e-06,
      "loss": 0.3334,
      "step": 41374
    },
    {
      "epoch": 1.8987196549034004,
      "grad_norm": 0.5000401139259338,
      "learning_rate": 3.080609867371543e-06,
      "loss": 0.3672,
      "step": 41375
    },
    {
      "epoch": 1.898765545408655,
      "grad_norm": 0.4434860050678253,
      "learning_rate": 3.0803834671355537e-06,
      "loss": 0.3083,
      "step": 41376
    },
    {
      "epoch": 1.8988114359139094,
      "grad_norm": 0.45429325103759766,
      "learning_rate": 3.0801570715156385e-06,
      "loss": 0.3505,
      "step": 41377
    },
    {
      "epoch": 1.8988573264191637,
      "grad_norm": 0.45842641592025757,
      "learning_rate": 3.0799306805123396e-06,
      "loss": 0.3183,
      "step": 41378
    },
    {
      "epoch": 1.8989032169244182,
      "grad_norm": 0.4495009183883667,
      "learning_rate": 3.079704294126201e-06,
      "loss": 0.3245,
      "step": 41379
    },
    {
      "epoch": 1.8989491074296727,
      "grad_norm": 0.45997318625450134,
      "learning_rate": 3.079477912357768e-06,
      "loss": 0.3187,
      "step": 41380
    },
    {
      "epoch": 1.8989949979349272,
      "grad_norm": 0.5034911036491394,
      "learning_rate": 3.079251535207585e-06,
      "loss": 0.3867,
      "step": 41381
    },
    {
      "epoch": 1.8990408884401817,
      "grad_norm": 0.513751745223999,
      "learning_rate": 3.0790251626761924e-06,
      "loss": 0.3804,
      "step": 41382
    },
    {
      "epoch": 1.8990867789454362,
      "grad_norm": 0.4622642993927002,
      "learning_rate": 3.0787987947641416e-06,
      "loss": 0.3111,
      "step": 41383
    },
    {
      "epoch": 1.8991326694506907,
      "grad_norm": 0.45239636301994324,
      "learning_rate": 3.0785724314719724e-06,
      "loss": 0.3239,
      "step": 41384
    },
    {
      "epoch": 1.8991785599559452,
      "grad_norm": 0.4533463418483734,
      "learning_rate": 3.0783460728002294e-06,
      "loss": 0.3062,
      "step": 41385
    },
    {
      "epoch": 1.8992244504611997,
      "grad_norm": 0.46175503730773926,
      "learning_rate": 3.078119718749459e-06,
      "loss": 0.3166,
      "step": 41386
    },
    {
      "epoch": 1.8992703409664542,
      "grad_norm": 0.463453471660614,
      "learning_rate": 3.077893369320204e-06,
      "loss": 0.2907,
      "step": 41387
    },
    {
      "epoch": 1.8993162314717085,
      "grad_norm": 0.46581870317459106,
      "learning_rate": 3.077667024513008e-06,
      "loss": 0.364,
      "step": 41388
    },
    {
      "epoch": 1.899362121976963,
      "grad_norm": 0.47171932458877563,
      "learning_rate": 3.0774406843284156e-06,
      "loss": 0.3762,
      "step": 41389
    },
    {
      "epoch": 1.8994080124822175,
      "grad_norm": 0.4148736298084259,
      "learning_rate": 3.0772143487669726e-06,
      "loss": 0.2718,
      "step": 41390
    },
    {
      "epoch": 1.8994539029874717,
      "grad_norm": 0.471902996301651,
      "learning_rate": 3.0769880178292224e-06,
      "loss": 0.3556,
      "step": 41391
    },
    {
      "epoch": 1.8994997934927262,
      "grad_norm": 0.46473681926727295,
      "learning_rate": 3.076761691515707e-06,
      "loss": 0.3541,
      "step": 41392
    },
    {
      "epoch": 1.8995456839979807,
      "grad_norm": 0.4966817796230316,
      "learning_rate": 3.076535369826975e-06,
      "loss": 0.3616,
      "step": 41393
    },
    {
      "epoch": 1.8995915745032352,
      "grad_norm": 0.47986501455307007,
      "learning_rate": 3.076309052763568e-06,
      "loss": 0.3725,
      "step": 41394
    },
    {
      "epoch": 1.8996374650084897,
      "grad_norm": 0.48514416813850403,
      "learning_rate": 3.0760827403260295e-06,
      "loss": 0.3471,
      "step": 41395
    },
    {
      "epoch": 1.8996833555137442,
      "grad_norm": 0.46562033891677856,
      "learning_rate": 3.0758564325149067e-06,
      "loss": 0.3352,
      "step": 41396
    },
    {
      "epoch": 1.8997292460189987,
      "grad_norm": 0.5356138944625854,
      "learning_rate": 3.0756301293307407e-06,
      "loss": 0.4204,
      "step": 41397
    },
    {
      "epoch": 1.8997751365242532,
      "grad_norm": 0.5223627686500549,
      "learning_rate": 3.0754038307740754e-06,
      "loss": 0.3927,
      "step": 41398
    },
    {
      "epoch": 1.8998210270295077,
      "grad_norm": 0.5011340379714966,
      "learning_rate": 3.0751775368454584e-06,
      "loss": 0.415,
      "step": 41399
    },
    {
      "epoch": 1.8998669175347622,
      "grad_norm": 0.4960937201976776,
      "learning_rate": 3.0749512475454317e-06,
      "loss": 0.374,
      "step": 41400
    },
    {
      "epoch": 1.8999128080400165,
      "grad_norm": 0.4673342704772949,
      "learning_rate": 3.074724962874539e-06,
      "loss": 0.3275,
      "step": 41401
    },
    {
      "epoch": 1.899958698545271,
      "grad_norm": 0.4454406499862671,
      "learning_rate": 3.0744986828333263e-06,
      "loss": 0.3398,
      "step": 41402
    },
    {
      "epoch": 1.9000045890505255,
      "grad_norm": 0.5034834146499634,
      "learning_rate": 3.074272407422336e-06,
      "loss": 0.403,
      "step": 41403
    },
    {
      "epoch": 1.9000504795557798,
      "grad_norm": 0.4703139662742615,
      "learning_rate": 3.0740461366421113e-06,
      "loss": 0.3676,
      "step": 41404
    },
    {
      "epoch": 1.9000963700610343,
      "grad_norm": 0.49312445521354675,
      "learning_rate": 3.0738198704931997e-06,
      "loss": 0.3433,
      "step": 41405
    },
    {
      "epoch": 1.9001422605662888,
      "grad_norm": 0.5127609372138977,
      "learning_rate": 3.073593608976143e-06,
      "loss": 0.3545,
      "step": 41406
    },
    {
      "epoch": 1.9001881510715433,
      "grad_norm": 0.48808130621910095,
      "learning_rate": 3.0733673520914853e-06,
      "loss": 0.4447,
      "step": 41407
    },
    {
      "epoch": 1.9002340415767978,
      "grad_norm": 0.44427812099456787,
      "learning_rate": 3.073141099839772e-06,
      "loss": 0.2883,
      "step": 41408
    },
    {
      "epoch": 1.9002799320820523,
      "grad_norm": 0.40720510482788086,
      "learning_rate": 3.0729148522215456e-06,
      "loss": 0.3052,
      "step": 41409
    },
    {
      "epoch": 1.9003258225873068,
      "grad_norm": 0.5038220882415771,
      "learning_rate": 3.07268860923735e-06,
      "loss": 0.3989,
      "step": 41410
    },
    {
      "epoch": 1.9003717130925613,
      "grad_norm": 0.49350884556770325,
      "learning_rate": 3.0724623708877315e-06,
      "loss": 0.3693,
      "step": 41411
    },
    {
      "epoch": 1.9004176035978158,
      "grad_norm": 0.4786263704299927,
      "learning_rate": 3.0722361371732323e-06,
      "loss": 0.3814,
      "step": 41412
    },
    {
      "epoch": 1.90046349410307,
      "grad_norm": 0.4418644607067108,
      "learning_rate": 3.072009908094395e-06,
      "loss": 0.3395,
      "step": 41413
    },
    {
      "epoch": 1.9005093846083245,
      "grad_norm": 0.4409025311470032,
      "learning_rate": 3.0717836836517675e-06,
      "loss": 0.3663,
      "step": 41414
    },
    {
      "epoch": 1.900555275113579,
      "grad_norm": 0.47292205691337585,
      "learning_rate": 3.071557463845891e-06,
      "loss": 0.32,
      "step": 41415
    },
    {
      "epoch": 1.9006011656188333,
      "grad_norm": 0.41654813289642334,
      "learning_rate": 3.0713312486773096e-06,
      "loss": 0.2831,
      "step": 41416
    },
    {
      "epoch": 1.9006470561240878,
      "grad_norm": 0.4755242168903351,
      "learning_rate": 3.071105038146569e-06,
      "loss": 0.3581,
      "step": 41417
    },
    {
      "epoch": 1.9006929466293423,
      "grad_norm": 0.5076202750205994,
      "learning_rate": 3.0708788322542115e-06,
      "loss": 0.4556,
      "step": 41418
    },
    {
      "epoch": 1.9007388371345968,
      "grad_norm": 0.4687141478061676,
      "learning_rate": 3.0706526310007817e-06,
      "loss": 0.3359,
      "step": 41419
    },
    {
      "epoch": 1.9007847276398513,
      "grad_norm": 0.4413447082042694,
      "learning_rate": 3.0704264343868216e-06,
      "loss": 0.2843,
      "step": 41420
    },
    {
      "epoch": 1.9008306181451058,
      "grad_norm": 0.42279112339019775,
      "learning_rate": 3.0702002424128787e-06,
      "loss": 0.2658,
      "step": 41421
    },
    {
      "epoch": 1.9008765086503603,
      "grad_norm": 0.4256211221218109,
      "learning_rate": 3.0699740550794956e-06,
      "loss": 0.3106,
      "step": 41422
    },
    {
      "epoch": 1.9009223991556148,
      "grad_norm": 0.4700208604335785,
      "learning_rate": 3.0697478723872144e-06,
      "loss": 0.3451,
      "step": 41423
    },
    {
      "epoch": 1.9009682896608693,
      "grad_norm": 0.4352574348449707,
      "learning_rate": 3.0695216943365814e-06,
      "loss": 0.2805,
      "step": 41424
    },
    {
      "epoch": 1.9010141801661238,
      "grad_norm": 0.428152471780777,
      "learning_rate": 3.0692955209281396e-06,
      "loss": 0.2823,
      "step": 41425
    },
    {
      "epoch": 1.901060070671378,
      "grad_norm": 0.49111267924308777,
      "learning_rate": 3.0690693521624306e-06,
      "loss": 0.404,
      "step": 41426
    },
    {
      "epoch": 1.9011059611766326,
      "grad_norm": 0.45435822010040283,
      "learning_rate": 3.0688431880400025e-06,
      "loss": 0.3155,
      "step": 41427
    },
    {
      "epoch": 1.901151851681887,
      "grad_norm": 0.4784923493862152,
      "learning_rate": 3.068617028561397e-06,
      "loss": 0.3716,
      "step": 41428
    },
    {
      "epoch": 1.9011977421871413,
      "grad_norm": 0.488253116607666,
      "learning_rate": 3.0683908737271566e-06,
      "loss": 0.3995,
      "step": 41429
    },
    {
      "epoch": 1.9012436326923958,
      "grad_norm": 0.450344055891037,
      "learning_rate": 3.068164723537828e-06,
      "loss": 0.3403,
      "step": 41430
    },
    {
      "epoch": 1.9012895231976503,
      "grad_norm": 0.41516032814979553,
      "learning_rate": 3.067938577993953e-06,
      "loss": 0.2663,
      "step": 41431
    },
    {
      "epoch": 1.9013354137029048,
      "grad_norm": 0.48087143898010254,
      "learning_rate": 3.0677124370960747e-06,
      "loss": 0.361,
      "step": 41432
    },
    {
      "epoch": 1.9013813042081593,
      "grad_norm": 0.49434953927993774,
      "learning_rate": 3.067486300844741e-06,
      "loss": 0.3945,
      "step": 41433
    },
    {
      "epoch": 1.9014271947134138,
      "grad_norm": 0.45705169439315796,
      "learning_rate": 3.0672601692404914e-06,
      "loss": 0.3053,
      "step": 41434
    },
    {
      "epoch": 1.9014730852186683,
      "grad_norm": 0.49436134099960327,
      "learning_rate": 3.0670340422838696e-06,
      "loss": 0.2696,
      "step": 41435
    },
    {
      "epoch": 1.9015189757239228,
      "grad_norm": 0.5084503293037415,
      "learning_rate": 3.066807919975422e-06,
      "loss": 0.4113,
      "step": 41436
    },
    {
      "epoch": 1.9015648662291773,
      "grad_norm": 0.5224766731262207,
      "learning_rate": 3.0665818023156924e-06,
      "loss": 0.4401,
      "step": 41437
    },
    {
      "epoch": 1.9016107567344318,
      "grad_norm": 0.47182920575141907,
      "learning_rate": 3.0663556893052216e-06,
      "loss": 0.3173,
      "step": 41438
    },
    {
      "epoch": 1.901656647239686,
      "grad_norm": 0.4812453091144562,
      "learning_rate": 3.0661295809445556e-06,
      "loss": 0.3652,
      "step": 41439
    },
    {
      "epoch": 1.9017025377449406,
      "grad_norm": 0.4881761372089386,
      "learning_rate": 3.065903477234239e-06,
      "loss": 0.366,
      "step": 41440
    },
    {
      "epoch": 1.901748428250195,
      "grad_norm": 0.4897553324699402,
      "learning_rate": 3.065677378174811e-06,
      "loss": 0.3015,
      "step": 41441
    },
    {
      "epoch": 1.9017943187554494,
      "grad_norm": 0.4538304805755615,
      "learning_rate": 3.0654512837668203e-06,
      "loss": 0.358,
      "step": 41442
    },
    {
      "epoch": 1.9018402092607039,
      "grad_norm": 0.5251752138137817,
      "learning_rate": 3.06522519401081e-06,
      "loss": 0.4637,
      "step": 41443
    },
    {
      "epoch": 1.9018860997659583,
      "grad_norm": 0.5094059109687805,
      "learning_rate": 3.0649991089073205e-06,
      "loss": 0.374,
      "step": 41444
    },
    {
      "epoch": 1.9019319902712128,
      "grad_norm": 0.46599093079566956,
      "learning_rate": 3.0647730284568987e-06,
      "loss": 0.3759,
      "step": 41445
    },
    {
      "epoch": 1.9019778807764673,
      "grad_norm": 0.48446768522262573,
      "learning_rate": 3.064546952660087e-06,
      "loss": 0.3129,
      "step": 41446
    },
    {
      "epoch": 1.9020237712817218,
      "grad_norm": 0.4529275894165039,
      "learning_rate": 3.0643208815174265e-06,
      "loss": 0.3102,
      "step": 41447
    },
    {
      "epoch": 1.9020696617869763,
      "grad_norm": 0.46129462122917175,
      "learning_rate": 3.0640948150294657e-06,
      "loss": 0.3313,
      "step": 41448
    },
    {
      "epoch": 1.9021155522922308,
      "grad_norm": 0.4423235356807709,
      "learning_rate": 3.063868753196746e-06,
      "loss": 0.3182,
      "step": 41449
    },
    {
      "epoch": 1.9021614427974853,
      "grad_norm": 0.477204829454422,
      "learning_rate": 3.0636426960198096e-06,
      "loss": 0.3557,
      "step": 41450
    },
    {
      "epoch": 1.9022073333027396,
      "grad_norm": 0.4457971751689911,
      "learning_rate": 3.0634166434992028e-06,
      "loss": 0.3309,
      "step": 41451
    },
    {
      "epoch": 1.902253223807994,
      "grad_norm": 0.501105010509491,
      "learning_rate": 3.063190595635468e-06,
      "loss": 0.451,
      "step": 41452
    },
    {
      "epoch": 1.9022991143132486,
      "grad_norm": 0.4747656285762787,
      "learning_rate": 3.062964552429148e-06,
      "loss": 0.3975,
      "step": 41453
    },
    {
      "epoch": 1.902345004818503,
      "grad_norm": 0.48892056941986084,
      "learning_rate": 3.0627385138807853e-06,
      "loss": 0.3437,
      "step": 41454
    },
    {
      "epoch": 1.9023908953237574,
      "grad_norm": 0.4740126430988312,
      "learning_rate": 3.0625124799909277e-06,
      "loss": 0.3934,
      "step": 41455
    },
    {
      "epoch": 1.9024367858290119,
      "grad_norm": 0.48836010694503784,
      "learning_rate": 3.062286450760115e-06,
      "loss": 0.4109,
      "step": 41456
    },
    {
      "epoch": 1.9024826763342664,
      "grad_norm": 0.5288827419281006,
      "learning_rate": 3.0620604261888897e-06,
      "loss": 0.4777,
      "step": 41457
    },
    {
      "epoch": 1.9025285668395209,
      "grad_norm": 0.4669460654258728,
      "learning_rate": 3.0618344062778007e-06,
      "loss": 0.3591,
      "step": 41458
    },
    {
      "epoch": 1.9025744573447754,
      "grad_norm": 0.49849891662597656,
      "learning_rate": 3.061608391027387e-06,
      "loss": 0.3892,
      "step": 41459
    },
    {
      "epoch": 1.9026203478500299,
      "grad_norm": 0.4655285179615021,
      "learning_rate": 3.0613823804381927e-06,
      "loss": 0.3949,
      "step": 41460
    },
    {
      "epoch": 1.9026662383552844,
      "grad_norm": 0.5049852132797241,
      "learning_rate": 3.061156374510763e-06,
      "loss": 0.3645,
      "step": 41461
    },
    {
      "epoch": 1.9027121288605389,
      "grad_norm": 0.4669059216976166,
      "learning_rate": 3.0609303732456397e-06,
      "loss": 0.3256,
      "step": 41462
    },
    {
      "epoch": 1.9027580193657934,
      "grad_norm": 0.47471845149993896,
      "learning_rate": 3.060704376643365e-06,
      "loss": 0.3812,
      "step": 41463
    },
    {
      "epoch": 1.9028039098710476,
      "grad_norm": 0.4622300863265991,
      "learning_rate": 3.060478384704487e-06,
      "loss": 0.3192,
      "step": 41464
    },
    {
      "epoch": 1.9028498003763021,
      "grad_norm": 0.4324879050254822,
      "learning_rate": 3.0602523974295462e-06,
      "loss": 0.3165,
      "step": 41465
    },
    {
      "epoch": 1.9028956908815566,
      "grad_norm": 0.5104799270629883,
      "learning_rate": 3.060026414819085e-06,
      "loss": 0.3818,
      "step": 41466
    },
    {
      "epoch": 1.902941581386811,
      "grad_norm": 0.4826866686344147,
      "learning_rate": 3.059800436873649e-06,
      "loss": 0.3372,
      "step": 41467
    },
    {
      "epoch": 1.9029874718920654,
      "grad_norm": 0.5172342658042908,
      "learning_rate": 3.0595744635937804e-06,
      "loss": 0.4503,
      "step": 41468
    },
    {
      "epoch": 1.90303336239732,
      "grad_norm": 0.5518525838851929,
      "learning_rate": 3.0593484949800214e-06,
      "loss": 0.4825,
      "step": 41469
    },
    {
      "epoch": 1.9030792529025744,
      "grad_norm": 0.48422640562057495,
      "learning_rate": 3.0591225310329176e-06,
      "loss": 0.3271,
      "step": 41470
    },
    {
      "epoch": 1.903125143407829,
      "grad_norm": 0.4536167085170746,
      "learning_rate": 3.0588965717530123e-06,
      "loss": 0.3303,
      "step": 41471
    },
    {
      "epoch": 1.9031710339130834,
      "grad_norm": 0.4635917544364929,
      "learning_rate": 3.058670617140847e-06,
      "loss": 0.3398,
      "step": 41472
    },
    {
      "epoch": 1.903216924418338,
      "grad_norm": 0.48019567131996155,
      "learning_rate": 3.058444667196967e-06,
      "loss": 0.3531,
      "step": 41473
    },
    {
      "epoch": 1.9032628149235924,
      "grad_norm": 0.42461317777633667,
      "learning_rate": 3.058218721921915e-06,
      "loss": 0.265,
      "step": 41474
    },
    {
      "epoch": 1.903308705428847,
      "grad_norm": 0.4585840106010437,
      "learning_rate": 3.0579927813162323e-06,
      "loss": 0.3601,
      "step": 41475
    },
    {
      "epoch": 1.9033545959341014,
      "grad_norm": 0.45265302062034607,
      "learning_rate": 3.0577668453804665e-06,
      "loss": 0.3083,
      "step": 41476
    },
    {
      "epoch": 1.9034004864393557,
      "grad_norm": 0.49232858419418335,
      "learning_rate": 3.0575409141151587e-06,
      "loss": 0.3907,
      "step": 41477
    },
    {
      "epoch": 1.9034463769446102,
      "grad_norm": 0.49260351061820984,
      "learning_rate": 3.0573149875208492e-06,
      "loss": 0.3712,
      "step": 41478
    },
    {
      "epoch": 1.9034922674498647,
      "grad_norm": 0.4469263255596161,
      "learning_rate": 3.0570890655980857e-06,
      "loss": 0.3363,
      "step": 41479
    },
    {
      "epoch": 1.903538157955119,
      "grad_norm": 0.47698330879211426,
      "learning_rate": 3.0568631483474098e-06,
      "loss": 0.4058,
      "step": 41480
    },
    {
      "epoch": 1.9035840484603734,
      "grad_norm": 0.4693165123462677,
      "learning_rate": 3.056637235769364e-06,
      "loss": 0.3347,
      "step": 41481
    },
    {
      "epoch": 1.903629938965628,
      "grad_norm": 0.4870064854621887,
      "learning_rate": 3.0564113278644935e-06,
      "loss": 0.3602,
      "step": 41482
    },
    {
      "epoch": 1.9036758294708824,
      "grad_norm": 0.48844945430755615,
      "learning_rate": 3.05618542463334e-06,
      "loss": 0.3658,
      "step": 41483
    },
    {
      "epoch": 1.903721719976137,
      "grad_norm": 0.44118231534957886,
      "learning_rate": 3.0559595260764472e-06,
      "loss": 0.3132,
      "step": 41484
    },
    {
      "epoch": 1.9037676104813914,
      "grad_norm": 0.506024956703186,
      "learning_rate": 3.0557336321943566e-06,
      "loss": 0.4547,
      "step": 41485
    },
    {
      "epoch": 1.903813500986646,
      "grad_norm": 0.4778393507003784,
      "learning_rate": 3.0555077429876147e-06,
      "loss": 0.3469,
      "step": 41486
    },
    {
      "epoch": 1.9038593914919004,
      "grad_norm": 0.5090705752372742,
      "learning_rate": 3.0552818584567623e-06,
      "loss": 0.4252,
      "step": 41487
    },
    {
      "epoch": 1.903905281997155,
      "grad_norm": 0.4378887414932251,
      "learning_rate": 3.0550559786023433e-06,
      "loss": 0.263,
      "step": 41488
    },
    {
      "epoch": 1.9039511725024094,
      "grad_norm": 0.4595848321914673,
      "learning_rate": 3.0548301034249014e-06,
      "loss": 0.298,
      "step": 41489
    },
    {
      "epoch": 1.9039970630076637,
      "grad_norm": 0.4499305188655853,
      "learning_rate": 3.054604232924979e-06,
      "loss": 0.3078,
      "step": 41490
    },
    {
      "epoch": 1.9040429535129182,
      "grad_norm": 0.5040925145149231,
      "learning_rate": 3.0543783671031173e-06,
      "loss": 0.4087,
      "step": 41491
    },
    {
      "epoch": 1.9040888440181727,
      "grad_norm": 0.4440968632698059,
      "learning_rate": 3.0541525059598644e-06,
      "loss": 0.2989,
      "step": 41492
    },
    {
      "epoch": 1.904134734523427,
      "grad_norm": 0.4526048004627228,
      "learning_rate": 3.0539266494957596e-06,
      "loss": 0.2973,
      "step": 41493
    },
    {
      "epoch": 1.9041806250286815,
      "grad_norm": 0.44277554750442505,
      "learning_rate": 3.0537007977113462e-06,
      "loss": 0.3152,
      "step": 41494
    },
    {
      "epoch": 1.904226515533936,
      "grad_norm": 0.45494958758354187,
      "learning_rate": 3.0534749506071694e-06,
      "loss": 0.3488,
      "step": 41495
    },
    {
      "epoch": 1.9042724060391905,
      "grad_norm": 0.5186036229133606,
      "learning_rate": 3.053249108183771e-06,
      "loss": 0.3605,
      "step": 41496
    },
    {
      "epoch": 1.904318296544445,
      "grad_norm": 0.44100144505500793,
      "learning_rate": 3.0530232704416923e-06,
      "loss": 0.327,
      "step": 41497
    },
    {
      "epoch": 1.9043641870496995,
      "grad_norm": 0.4666369557380676,
      "learning_rate": 3.0527974373814796e-06,
      "loss": 0.3326,
      "step": 41498
    },
    {
      "epoch": 1.904410077554954,
      "grad_norm": 0.4460645318031311,
      "learning_rate": 3.052571609003676e-06,
      "loss": 0.3105,
      "step": 41499
    },
    {
      "epoch": 1.9044559680602084,
      "grad_norm": 0.486577570438385,
      "learning_rate": 3.05234578530882e-06,
      "loss": 0.3805,
      "step": 41500
    },
    {
      "epoch": 1.904501858565463,
      "grad_norm": 0.46949371695518494,
      "learning_rate": 3.0521199662974595e-06,
      "loss": 0.3586,
      "step": 41501
    },
    {
      "epoch": 1.9045477490707172,
      "grad_norm": 0.4689438045024872,
      "learning_rate": 3.0518941519701356e-06,
      "loss": 0.3397,
      "step": 41502
    },
    {
      "epoch": 1.9045936395759717,
      "grad_norm": 0.5158957839012146,
      "learning_rate": 3.0516683423273906e-06,
      "loss": 0.4308,
      "step": 41503
    },
    {
      "epoch": 1.9046395300812262,
      "grad_norm": 0.9596680402755737,
      "learning_rate": 3.05144253736977e-06,
      "loss": 0.3019,
      "step": 41504
    },
    {
      "epoch": 1.9046854205864805,
      "grad_norm": 0.4527713656425476,
      "learning_rate": 3.0512167370978142e-06,
      "loss": 0.3126,
      "step": 41505
    },
    {
      "epoch": 1.904731311091735,
      "grad_norm": 0.4414074122905731,
      "learning_rate": 3.0509909415120655e-06,
      "loss": 0.31,
      "step": 41506
    },
    {
      "epoch": 1.9047772015969895,
      "grad_norm": 0.4565839171409607,
      "learning_rate": 3.05076515061307e-06,
      "loss": 0.3144,
      "step": 41507
    },
    {
      "epoch": 1.904823092102244,
      "grad_norm": 0.4432392716407776,
      "learning_rate": 3.050539364401369e-06,
      "loss": 0.3499,
      "step": 41508
    },
    {
      "epoch": 1.9048689826074985,
      "grad_norm": 0.49031075835227966,
      "learning_rate": 3.0503135828775056e-06,
      "loss": 0.3893,
      "step": 41509
    },
    {
      "epoch": 1.904914873112753,
      "grad_norm": 0.5095597505569458,
      "learning_rate": 3.0500878060420224e-06,
      "loss": 0.4622,
      "step": 41510
    },
    {
      "epoch": 1.9049607636180075,
      "grad_norm": 0.4964854121208191,
      "learning_rate": 3.0498620338954634e-06,
      "loss": 0.3939,
      "step": 41511
    },
    {
      "epoch": 1.905006654123262,
      "grad_norm": 0.49824872612953186,
      "learning_rate": 3.0496362664383684e-06,
      "loss": 0.3285,
      "step": 41512
    },
    {
      "epoch": 1.9050525446285165,
      "grad_norm": 0.46538910269737244,
      "learning_rate": 3.049410503671285e-06,
      "loss": 0.3906,
      "step": 41513
    },
    {
      "epoch": 1.905098435133771,
      "grad_norm": 0.4694610834121704,
      "learning_rate": 3.049184745594753e-06,
      "loss": 0.3739,
      "step": 41514
    },
    {
      "epoch": 1.9051443256390252,
      "grad_norm": 0.4684291481971741,
      "learning_rate": 3.048958992209316e-06,
      "loss": 0.3693,
      "step": 41515
    },
    {
      "epoch": 1.9051902161442797,
      "grad_norm": 0.5311850905418396,
      "learning_rate": 3.0487332435155175e-06,
      "loss": 0.3912,
      "step": 41516
    },
    {
      "epoch": 1.9052361066495342,
      "grad_norm": 0.4864659607410431,
      "learning_rate": 3.0485074995139e-06,
      "loss": 0.3635,
      "step": 41517
    },
    {
      "epoch": 1.9052819971547885,
      "grad_norm": 0.4855727255344391,
      "learning_rate": 3.0482817602050053e-06,
      "loss": 0.3532,
      "step": 41518
    },
    {
      "epoch": 1.905327887660043,
      "grad_norm": 0.4549660086631775,
      "learning_rate": 3.0480560255893754e-06,
      "loss": 0.3275,
      "step": 41519
    },
    {
      "epoch": 1.9053737781652975,
      "grad_norm": 0.4462898373603821,
      "learning_rate": 3.0478302956675566e-06,
      "loss": 0.3293,
      "step": 41520
    },
    {
      "epoch": 1.905419668670552,
      "grad_norm": 0.47286492586135864,
      "learning_rate": 3.047604570440091e-06,
      "loss": 0.3492,
      "step": 41521
    },
    {
      "epoch": 1.9054655591758065,
      "grad_norm": 0.45767542719841003,
      "learning_rate": 3.0473788499075167e-06,
      "loss": 0.35,
      "step": 41522
    },
    {
      "epoch": 1.905511449681061,
      "grad_norm": 0.48942410945892334,
      "learning_rate": 3.0471531340703824e-06,
      "loss": 0.3771,
      "step": 41523
    },
    {
      "epoch": 1.9055573401863155,
      "grad_norm": 0.4900580048561096,
      "learning_rate": 3.0469274229292288e-06,
      "loss": 0.3848,
      "step": 41524
    },
    {
      "epoch": 1.90560323069157,
      "grad_norm": 0.461749404668808,
      "learning_rate": 3.0467017164845967e-06,
      "loss": 0.3521,
      "step": 41525
    },
    {
      "epoch": 1.9056491211968245,
      "grad_norm": 0.45963889360427856,
      "learning_rate": 3.046476014737032e-06,
      "loss": 0.3239,
      "step": 41526
    },
    {
      "epoch": 1.905695011702079,
      "grad_norm": 0.4606442451477051,
      "learning_rate": 3.0462503176870755e-06,
      "loss": 0.3311,
      "step": 41527
    },
    {
      "epoch": 1.9057409022073333,
      "grad_norm": 0.4645827114582062,
      "learning_rate": 3.0460246253352688e-06,
      "loss": 0.3332,
      "step": 41528
    },
    {
      "epoch": 1.9057867927125878,
      "grad_norm": 0.451529860496521,
      "learning_rate": 3.045798937682158e-06,
      "loss": 0.3052,
      "step": 41529
    },
    {
      "epoch": 1.9058326832178423,
      "grad_norm": 0.4676174223423004,
      "learning_rate": 3.0455732547282835e-06,
      "loss": 0.343,
      "step": 41530
    },
    {
      "epoch": 1.9058785737230965,
      "grad_norm": 0.44746118783950806,
      "learning_rate": 3.045347576474188e-06,
      "loss": 0.3352,
      "step": 41531
    },
    {
      "epoch": 1.905924464228351,
      "grad_norm": 0.45558294653892517,
      "learning_rate": 3.045121902920416e-06,
      "loss": 0.3718,
      "step": 41532
    },
    {
      "epoch": 1.9059703547336055,
      "grad_norm": 0.4658662676811218,
      "learning_rate": 3.044896234067508e-06,
      "loss": 0.3726,
      "step": 41533
    },
    {
      "epoch": 1.90601624523886,
      "grad_norm": 0.46545907855033875,
      "learning_rate": 3.044670569916006e-06,
      "loss": 0.3307,
      "step": 41534
    },
    {
      "epoch": 1.9060621357441145,
      "grad_norm": 0.4706931412220001,
      "learning_rate": 3.0444449104664565e-06,
      "loss": 0.4031,
      "step": 41535
    },
    {
      "epoch": 1.906108026249369,
      "grad_norm": 0.4775809943675995,
      "learning_rate": 3.0442192557193996e-06,
      "loss": 0.3692,
      "step": 41536
    },
    {
      "epoch": 1.9061539167546235,
      "grad_norm": 0.48859256505966187,
      "learning_rate": 3.0439936056753766e-06,
      "loss": 0.3996,
      "step": 41537
    },
    {
      "epoch": 1.906199807259878,
      "grad_norm": 0.5069428086280823,
      "learning_rate": 3.0437679603349336e-06,
      "loss": 0.3975,
      "step": 41538
    },
    {
      "epoch": 1.9062456977651325,
      "grad_norm": 0.4373435378074646,
      "learning_rate": 3.0435423196986104e-06,
      "loss": 0.2881,
      "step": 41539
    },
    {
      "epoch": 1.9062915882703868,
      "grad_norm": 0.462407648563385,
      "learning_rate": 3.0433166837669483e-06,
      "loss": 0.3268,
      "step": 41540
    },
    {
      "epoch": 1.9063374787756413,
      "grad_norm": 0.4256834387779236,
      "learning_rate": 3.0430910525404946e-06,
      "loss": 0.3198,
      "step": 41541
    },
    {
      "epoch": 1.9063833692808958,
      "grad_norm": 0.47236669063568115,
      "learning_rate": 3.042865426019789e-06,
      "loss": 0.3555,
      "step": 41542
    },
    {
      "epoch": 1.9064292597861503,
      "grad_norm": 0.5328798294067383,
      "learning_rate": 3.042639804205374e-06,
      "loss": 0.4622,
      "step": 41543
    },
    {
      "epoch": 1.9064751502914046,
      "grad_norm": 0.44350722432136536,
      "learning_rate": 3.042414187097793e-06,
      "loss": 0.3081,
      "step": 41544
    },
    {
      "epoch": 1.906521040796659,
      "grad_norm": 0.4943241775035858,
      "learning_rate": 3.0421885746975877e-06,
      "loss": 0.429,
      "step": 41545
    },
    {
      "epoch": 1.9065669313019136,
      "grad_norm": 0.49184250831604004,
      "learning_rate": 3.0419629670053e-06,
      "loss": 0.375,
      "step": 41546
    },
    {
      "epoch": 1.906612821807168,
      "grad_norm": 0.4664251506328583,
      "learning_rate": 3.0417373640214753e-06,
      "loss": 0.347,
      "step": 41547
    },
    {
      "epoch": 1.9066587123124226,
      "grad_norm": 0.4462032616138458,
      "learning_rate": 3.041511765746653e-06,
      "loss": 0.3153,
      "step": 41548
    },
    {
      "epoch": 1.906704602817677,
      "grad_norm": 0.4971274137496948,
      "learning_rate": 3.041286172181378e-06,
      "loss": 0.4329,
      "step": 41549
    },
    {
      "epoch": 1.9067504933229316,
      "grad_norm": 0.4704916775226593,
      "learning_rate": 3.0410605833261884e-06,
      "loss": 0.3769,
      "step": 41550
    },
    {
      "epoch": 1.906796383828186,
      "grad_norm": 0.4660366475582123,
      "learning_rate": 3.040834999181633e-06,
      "loss": 0.3592,
      "step": 41551
    },
    {
      "epoch": 1.9068422743334406,
      "grad_norm": 0.45621976256370544,
      "learning_rate": 3.0406094197482504e-06,
      "loss": 0.3179,
      "step": 41552
    },
    {
      "epoch": 1.9068881648386948,
      "grad_norm": 0.49223658442497253,
      "learning_rate": 3.0403838450265827e-06,
      "loss": 0.3954,
      "step": 41553
    },
    {
      "epoch": 1.9069340553439493,
      "grad_norm": 0.4706885516643524,
      "learning_rate": 3.0401582750171745e-06,
      "loss": 0.3353,
      "step": 41554
    },
    {
      "epoch": 1.9069799458492038,
      "grad_norm": 0.4801173210144043,
      "learning_rate": 3.0399327097205674e-06,
      "loss": 0.3773,
      "step": 41555
    },
    {
      "epoch": 1.907025836354458,
      "grad_norm": 0.47125449776649475,
      "learning_rate": 3.039707149137301e-06,
      "loss": 0.3602,
      "step": 41556
    },
    {
      "epoch": 1.9070717268597126,
      "grad_norm": 0.4658561646938324,
      "learning_rate": 3.0394815932679224e-06,
      "loss": 0.3842,
      "step": 41557
    },
    {
      "epoch": 1.907117617364967,
      "grad_norm": 0.4492603540420532,
      "learning_rate": 3.039256042112972e-06,
      "loss": 0.3212,
      "step": 41558
    },
    {
      "epoch": 1.9071635078702216,
      "grad_norm": 0.5101969242095947,
      "learning_rate": 3.039030495672991e-06,
      "loss": 0.4037,
      "step": 41559
    },
    {
      "epoch": 1.907209398375476,
      "grad_norm": 0.4976857900619507,
      "learning_rate": 3.0388049539485233e-06,
      "loss": 0.3637,
      "step": 41560
    },
    {
      "epoch": 1.9072552888807306,
      "grad_norm": 0.6312056183815002,
      "learning_rate": 3.038579416940111e-06,
      "loss": 0.4498,
      "step": 41561
    },
    {
      "epoch": 1.907301179385985,
      "grad_norm": 0.455485075712204,
      "learning_rate": 3.038353884648294e-06,
      "loss": 0.379,
      "step": 41562
    },
    {
      "epoch": 1.9073470698912396,
      "grad_norm": 0.4763627350330353,
      "learning_rate": 3.03812835707362e-06,
      "loss": 0.348,
      "step": 41563
    },
    {
      "epoch": 1.907392960396494,
      "grad_norm": 0.49052444100379944,
      "learning_rate": 3.037902834216626e-06,
      "loss": 0.4205,
      "step": 41564
    },
    {
      "epoch": 1.9074388509017486,
      "grad_norm": 0.5040299296379089,
      "learning_rate": 3.037677316077857e-06,
      "loss": 0.4231,
      "step": 41565
    },
    {
      "epoch": 1.9074847414070029,
      "grad_norm": 0.49479496479034424,
      "learning_rate": 3.0374518026578553e-06,
      "loss": 0.3828,
      "step": 41566
    },
    {
      "epoch": 1.9075306319122574,
      "grad_norm": 0.4606206715106964,
      "learning_rate": 3.0372262939571628e-06,
      "loss": 0.3282,
      "step": 41567
    },
    {
      "epoch": 1.9075765224175119,
      "grad_norm": 0.49075305461883545,
      "learning_rate": 3.03700078997632e-06,
      "loss": 0.3659,
      "step": 41568
    },
    {
      "epoch": 1.9076224129227661,
      "grad_norm": 0.44100430607795715,
      "learning_rate": 3.0367752907158723e-06,
      "loss": 0.3251,
      "step": 41569
    },
    {
      "epoch": 1.9076683034280206,
      "grad_norm": 0.45194581151008606,
      "learning_rate": 3.03654979617636e-06,
      "loss": 0.3443,
      "step": 41570
    },
    {
      "epoch": 1.9077141939332751,
      "grad_norm": 0.47992345690727234,
      "learning_rate": 3.036324306358324e-06,
      "loss": 0.395,
      "step": 41571
    },
    {
      "epoch": 1.9077600844385296,
      "grad_norm": 0.4880546033382416,
      "learning_rate": 3.0360988212623103e-06,
      "loss": 0.3724,
      "step": 41572
    },
    {
      "epoch": 1.9078059749437841,
      "grad_norm": 0.43381908535957336,
      "learning_rate": 3.0358733408888597e-06,
      "loss": 0.2904,
      "step": 41573
    },
    {
      "epoch": 1.9078518654490386,
      "grad_norm": 0.5021708011627197,
      "learning_rate": 3.035647865238512e-06,
      "loss": 0.3332,
      "step": 41574
    },
    {
      "epoch": 1.9078977559542931,
      "grad_norm": 0.4859614074230194,
      "learning_rate": 3.035422394311812e-06,
      "loss": 0.3976,
      "step": 41575
    },
    {
      "epoch": 1.9079436464595476,
      "grad_norm": 0.47483667731285095,
      "learning_rate": 3.035196928109302e-06,
      "loss": 0.3871,
      "step": 41576
    },
    {
      "epoch": 1.9079895369648021,
      "grad_norm": 0.4652535915374756,
      "learning_rate": 3.0349714666315213e-06,
      "loss": 0.3413,
      "step": 41577
    },
    {
      "epoch": 1.9080354274700566,
      "grad_norm": 0.48963284492492676,
      "learning_rate": 3.034746009879015e-06,
      "loss": 0.3906,
      "step": 41578
    },
    {
      "epoch": 1.9080813179753109,
      "grad_norm": 0.4341690242290497,
      "learning_rate": 3.034520557852325e-06,
      "loss": 0.2858,
      "step": 41579
    },
    {
      "epoch": 1.9081272084805654,
      "grad_norm": 0.4465838372707367,
      "learning_rate": 3.034295110551991e-06,
      "loss": 0.3297,
      "step": 41580
    },
    {
      "epoch": 1.9081730989858199,
      "grad_norm": 0.4731297791004181,
      "learning_rate": 3.0340696679785586e-06,
      "loss": 0.3259,
      "step": 41581
    },
    {
      "epoch": 1.9082189894910742,
      "grad_norm": 0.42460453510284424,
      "learning_rate": 3.033844230132569e-06,
      "loss": 0.2947,
      "step": 41582
    },
    {
      "epoch": 1.9082648799963287,
      "grad_norm": 0.4718386232852936,
      "learning_rate": 3.033618797014562e-06,
      "loss": 0.3608,
      "step": 41583
    },
    {
      "epoch": 1.9083107705015832,
      "grad_norm": 0.4700298607349396,
      "learning_rate": 3.0333933686250795e-06,
      "loss": 0.3575,
      "step": 41584
    },
    {
      "epoch": 1.9083566610068377,
      "grad_norm": 0.4496673047542572,
      "learning_rate": 3.033167944964668e-06,
      "loss": 0.303,
      "step": 41585
    },
    {
      "epoch": 1.9084025515120921,
      "grad_norm": 0.5110210180282593,
      "learning_rate": 3.032942526033866e-06,
      "loss": 0.4337,
      "step": 41586
    },
    {
      "epoch": 1.9084484420173466,
      "grad_norm": 0.5274921655654907,
      "learning_rate": 3.0327171118332154e-06,
      "loss": 0.4293,
      "step": 41587
    },
    {
      "epoch": 1.9084943325226011,
      "grad_norm": 0.4800852835178375,
      "learning_rate": 3.032491702363261e-06,
      "loss": 0.3814,
      "step": 41588
    },
    {
      "epoch": 1.9085402230278556,
      "grad_norm": 0.4119489789009094,
      "learning_rate": 3.0322662976245426e-06,
      "loss": 0.2756,
      "step": 41589
    },
    {
      "epoch": 1.9085861135331101,
      "grad_norm": 0.44783815741539,
      "learning_rate": 3.0320408976176017e-06,
      "loss": 0.3424,
      "step": 41590
    },
    {
      "epoch": 1.9086320040383644,
      "grad_norm": 0.4968620538711548,
      "learning_rate": 3.0318155023429827e-06,
      "loss": 0.3651,
      "step": 41591
    },
    {
      "epoch": 1.908677894543619,
      "grad_norm": 0.4413295388221741,
      "learning_rate": 3.031590111801226e-06,
      "loss": 0.3041,
      "step": 41592
    },
    {
      "epoch": 1.9087237850488734,
      "grad_norm": 0.450089693069458,
      "learning_rate": 3.0313647259928714e-06,
      "loss": 0.2814,
      "step": 41593
    },
    {
      "epoch": 1.9087696755541277,
      "grad_norm": 0.4679228365421295,
      "learning_rate": 3.0311393449184654e-06,
      "loss": 0.4171,
      "step": 41594
    },
    {
      "epoch": 1.9088155660593822,
      "grad_norm": 0.4476267993450165,
      "learning_rate": 3.030913968578549e-06,
      "loss": 0.3235,
      "step": 41595
    },
    {
      "epoch": 1.9088614565646367,
      "grad_norm": 0.4413772523403168,
      "learning_rate": 3.030688596973661e-06,
      "loss": 0.3116,
      "step": 41596
    },
    {
      "epoch": 1.9089073470698912,
      "grad_norm": 0.5042796730995178,
      "learning_rate": 3.0304632301043467e-06,
      "loss": 0.4061,
      "step": 41597
    },
    {
      "epoch": 1.9089532375751457,
      "grad_norm": 0.4520305097103119,
      "learning_rate": 3.030237867971146e-06,
      "loss": 0.3337,
      "step": 41598
    },
    {
      "epoch": 1.9089991280804002,
      "grad_norm": 0.47830358147621155,
      "learning_rate": 3.030012510574601e-06,
      "loss": 0.3555,
      "step": 41599
    },
    {
      "epoch": 1.9090450185856547,
      "grad_norm": 0.46305590867996216,
      "learning_rate": 3.0297871579152548e-06,
      "loss": 0.36,
      "step": 41600
    },
    {
      "epoch": 1.9090909090909092,
      "grad_norm": 0.47261208295822144,
      "learning_rate": 3.0295618099936497e-06,
      "loss": 0.3959,
      "step": 41601
    },
    {
      "epoch": 1.9091367995961637,
      "grad_norm": 0.46703213453292847,
      "learning_rate": 3.0293364668103254e-06,
      "loss": 0.3588,
      "step": 41602
    },
    {
      "epoch": 1.9091826901014182,
      "grad_norm": 0.4855630099773407,
      "learning_rate": 3.029111128365825e-06,
      "loss": 0.4047,
      "step": 41603
    },
    {
      "epoch": 1.9092285806066724,
      "grad_norm": 0.41291651129722595,
      "learning_rate": 3.0288857946606915e-06,
      "loss": 0.2298,
      "step": 41604
    },
    {
      "epoch": 1.909274471111927,
      "grad_norm": 0.4846917986869812,
      "learning_rate": 3.028660465695463e-06,
      "loss": 0.4055,
      "step": 41605
    },
    {
      "epoch": 1.9093203616171814,
      "grad_norm": 0.4999304711818695,
      "learning_rate": 3.028435141470686e-06,
      "loss": 0.3673,
      "step": 41606
    },
    {
      "epoch": 1.9093662521224357,
      "grad_norm": 0.4918630123138428,
      "learning_rate": 3.0282098219869e-06,
      "loss": 0.4025,
      "step": 41607
    },
    {
      "epoch": 1.9094121426276902,
      "grad_norm": 0.42343443632125854,
      "learning_rate": 3.0279845072446463e-06,
      "loss": 0.2835,
      "step": 41608
    },
    {
      "epoch": 1.9094580331329447,
      "grad_norm": 0.48952850699424744,
      "learning_rate": 3.0277591972444687e-06,
      "loss": 0.4068,
      "step": 41609
    },
    {
      "epoch": 1.9095039236381992,
      "grad_norm": 0.4705795347690582,
      "learning_rate": 3.027533891986907e-06,
      "loss": 0.3842,
      "step": 41610
    },
    {
      "epoch": 1.9095498141434537,
      "grad_norm": 0.47345244884490967,
      "learning_rate": 3.027308591472502e-06,
      "loss": 0.3762,
      "step": 41611
    },
    {
      "epoch": 1.9095957046487082,
      "grad_norm": 0.4783681035041809,
      "learning_rate": 3.027083295701801e-06,
      "loss": 0.3787,
      "step": 41612
    },
    {
      "epoch": 1.9096415951539627,
      "grad_norm": 0.4678381383419037,
      "learning_rate": 3.0268580046753405e-06,
      "loss": 0.336,
      "step": 41613
    },
    {
      "epoch": 1.9096874856592172,
      "grad_norm": 0.4978208839893341,
      "learning_rate": 3.026632718393662e-06,
      "loss": 0.3297,
      "step": 41614
    },
    {
      "epoch": 1.9097333761644717,
      "grad_norm": 0.4917984902858734,
      "learning_rate": 3.0264074368573102e-06,
      "loss": 0.3473,
      "step": 41615
    },
    {
      "epoch": 1.9097792666697262,
      "grad_norm": 0.49002933502197266,
      "learning_rate": 3.0261821600668263e-06,
      "loss": 0.3713,
      "step": 41616
    },
    {
      "epoch": 1.9098251571749805,
      "grad_norm": 0.4661906957626343,
      "learning_rate": 3.0259568880227515e-06,
      "loss": 0.3526,
      "step": 41617
    },
    {
      "epoch": 1.909871047680235,
      "grad_norm": 0.47650644183158875,
      "learning_rate": 3.025731620725626e-06,
      "loss": 0.3893,
      "step": 41618
    },
    {
      "epoch": 1.9099169381854895,
      "grad_norm": 0.4874667823314667,
      "learning_rate": 3.0255063581759936e-06,
      "loss": 0.3733,
      "step": 41619
    },
    {
      "epoch": 1.9099628286907437,
      "grad_norm": 0.517134428024292,
      "learning_rate": 3.025281100374396e-06,
      "loss": 0.4524,
      "step": 41620
    },
    {
      "epoch": 1.9100087191959982,
      "grad_norm": 0.45107442140579224,
      "learning_rate": 3.0250558473213716e-06,
      "loss": 0.3223,
      "step": 41621
    },
    {
      "epoch": 1.9100546097012527,
      "grad_norm": 0.48389163613319397,
      "learning_rate": 3.0248305990174666e-06,
      "loss": 0.3836,
      "step": 41622
    },
    {
      "epoch": 1.9101005002065072,
      "grad_norm": 0.4673812985420227,
      "learning_rate": 3.0246053554632206e-06,
      "loss": 0.3427,
      "step": 41623
    },
    {
      "epoch": 1.9101463907117617,
      "grad_norm": 0.4295503497123718,
      "learning_rate": 3.0243801166591747e-06,
      "loss": 0.3094,
      "step": 41624
    },
    {
      "epoch": 1.9101922812170162,
      "grad_norm": 0.4560742676258087,
      "learning_rate": 3.0241548826058715e-06,
      "loss": 0.34,
      "step": 41625
    },
    {
      "epoch": 1.9102381717222707,
      "grad_norm": 0.4905715584754944,
      "learning_rate": 3.0239296533038526e-06,
      "loss": 0.3703,
      "step": 41626
    },
    {
      "epoch": 1.9102840622275252,
      "grad_norm": 0.45267176628112793,
      "learning_rate": 3.0237044287536564e-06,
      "loss": 0.3427,
      "step": 41627
    },
    {
      "epoch": 1.9103299527327797,
      "grad_norm": 0.42638611793518066,
      "learning_rate": 3.023479208955831e-06,
      "loss": 0.2823,
      "step": 41628
    },
    {
      "epoch": 1.910375843238034,
      "grad_norm": 0.4465063214302063,
      "learning_rate": 3.023253993910913e-06,
      "loss": 0.3162,
      "step": 41629
    },
    {
      "epoch": 1.9104217337432885,
      "grad_norm": 0.4348711669445038,
      "learning_rate": 3.0230287836194445e-06,
      "loss": 0.2949,
      "step": 41630
    },
    {
      "epoch": 1.910467624248543,
      "grad_norm": 0.4795144498348236,
      "learning_rate": 3.022803578081969e-06,
      "loss": 0.3501,
      "step": 41631
    },
    {
      "epoch": 1.9105135147537975,
      "grad_norm": 0.47305795550346375,
      "learning_rate": 3.022578377299027e-06,
      "loss": 0.3515,
      "step": 41632
    },
    {
      "epoch": 1.9105594052590518,
      "grad_norm": 0.47762468457221985,
      "learning_rate": 3.0223531812711572e-06,
      "loss": 0.3419,
      "step": 41633
    },
    {
      "epoch": 1.9106052957643063,
      "grad_norm": 0.461089551448822,
      "learning_rate": 3.0221279899989077e-06,
      "loss": 0.339,
      "step": 41634
    },
    {
      "epoch": 1.9106511862695608,
      "grad_norm": 0.5005724430084229,
      "learning_rate": 3.0219028034828146e-06,
      "loss": 0.4052,
      "step": 41635
    },
    {
      "epoch": 1.9106970767748153,
      "grad_norm": 0.4699056148529053,
      "learning_rate": 3.021677621723419e-06,
      "loss": 0.3698,
      "step": 41636
    },
    {
      "epoch": 1.9107429672800698,
      "grad_norm": 0.43832463026046753,
      "learning_rate": 3.021452444721267e-06,
      "loss": 0.2763,
      "step": 41637
    },
    {
      "epoch": 1.9107888577853243,
      "grad_norm": 0.4869318902492523,
      "learning_rate": 3.0212272724768967e-06,
      "loss": 0.4142,
      "step": 41638
    },
    {
      "epoch": 1.9108347482905788,
      "grad_norm": 0.4772319197654724,
      "learning_rate": 3.021002104990849e-06,
      "loss": 0.3594,
      "step": 41639
    },
    {
      "epoch": 1.9108806387958333,
      "grad_norm": 0.44488105177879333,
      "learning_rate": 3.0207769422636678e-06,
      "loss": 0.3165,
      "step": 41640
    },
    {
      "epoch": 1.9109265293010878,
      "grad_norm": 0.459980845451355,
      "learning_rate": 3.0205517842958944e-06,
      "loss": 0.2696,
      "step": 41641
    },
    {
      "epoch": 1.910972419806342,
      "grad_norm": 0.4743506610393524,
      "learning_rate": 3.0203266310880663e-06,
      "loss": 0.341,
      "step": 41642
    },
    {
      "epoch": 1.9110183103115965,
      "grad_norm": 0.5056712627410889,
      "learning_rate": 3.02010148264073e-06,
      "loss": 0.4365,
      "step": 41643
    },
    {
      "epoch": 1.911064200816851,
      "grad_norm": 0.5320311784744263,
      "learning_rate": 3.019876338954425e-06,
      "loss": 0.4278,
      "step": 41644
    },
    {
      "epoch": 1.9111100913221053,
      "grad_norm": 0.5276470184326172,
      "learning_rate": 3.019651200029691e-06,
      "loss": 0.3651,
      "step": 41645
    },
    {
      "epoch": 1.9111559818273598,
      "grad_norm": 0.45133671164512634,
      "learning_rate": 3.0194260658670727e-06,
      "loss": 0.3141,
      "step": 41646
    },
    {
      "epoch": 1.9112018723326143,
      "grad_norm": 0.4967919886112213,
      "learning_rate": 3.019200936467109e-06,
      "loss": 0.4217,
      "step": 41647
    },
    {
      "epoch": 1.9112477628378688,
      "grad_norm": 0.45508357882499695,
      "learning_rate": 3.0189758118303416e-06,
      "loss": 0.3468,
      "step": 41648
    },
    {
      "epoch": 1.9112936533431233,
      "grad_norm": 0.4873710572719574,
      "learning_rate": 3.0187506919573105e-06,
      "loss": 0.4072,
      "step": 41649
    },
    {
      "epoch": 1.9113395438483778,
      "grad_norm": 0.4491294324398041,
      "learning_rate": 3.018525576848561e-06,
      "loss": 0.2981,
      "step": 41650
    },
    {
      "epoch": 1.9113854343536323,
      "grad_norm": 0.43335095047950745,
      "learning_rate": 3.018300466504632e-06,
      "loss": 0.2955,
      "step": 41651
    },
    {
      "epoch": 1.9114313248588868,
      "grad_norm": 0.4938606321811676,
      "learning_rate": 3.018075360926063e-06,
      "loss": 0.3983,
      "step": 41652
    },
    {
      "epoch": 1.9114772153641413,
      "grad_norm": 0.47704455256462097,
      "learning_rate": 3.0178502601134e-06,
      "loss": 0.3579,
      "step": 41653
    },
    {
      "epoch": 1.9115231058693958,
      "grad_norm": 0.4614299237728119,
      "learning_rate": 3.01762516406718e-06,
      "loss": 0.3354,
      "step": 41654
    },
    {
      "epoch": 1.91156899637465,
      "grad_norm": 0.4565369784832001,
      "learning_rate": 3.0174000727879443e-06,
      "loss": 0.3175,
      "step": 41655
    },
    {
      "epoch": 1.9116148868799046,
      "grad_norm": 0.45117250084877014,
      "learning_rate": 3.017174986276239e-06,
      "loss": 0.328,
      "step": 41656
    },
    {
      "epoch": 1.911660777385159,
      "grad_norm": 0.4690169394016266,
      "learning_rate": 3.016949904532601e-06,
      "loss": 0.3103,
      "step": 41657
    },
    {
      "epoch": 1.9117066678904133,
      "grad_norm": 0.49757808446884155,
      "learning_rate": 3.01672482755757e-06,
      "loss": 0.4029,
      "step": 41658
    },
    {
      "epoch": 1.9117525583956678,
      "grad_norm": 0.4536018967628479,
      "learning_rate": 3.0164997553516922e-06,
      "loss": 0.2825,
      "step": 41659
    },
    {
      "epoch": 1.9117984489009223,
      "grad_norm": 0.47895774245262146,
      "learning_rate": 3.016274687915507e-06,
      "loss": 0.3648,
      "step": 41660
    },
    {
      "epoch": 1.9118443394061768,
      "grad_norm": 0.4861104190349579,
      "learning_rate": 3.016049625249553e-06,
      "loss": 0.3718,
      "step": 41661
    },
    {
      "epoch": 1.9118902299114313,
      "grad_norm": 0.48282304406166077,
      "learning_rate": 3.015824567354375e-06,
      "loss": 0.3599,
      "step": 41662
    },
    {
      "epoch": 1.9119361204166858,
      "grad_norm": 0.4301174581050873,
      "learning_rate": 3.0155995142305134e-06,
      "loss": 0.2783,
      "step": 41663
    },
    {
      "epoch": 1.9119820109219403,
      "grad_norm": 0.4923028349876404,
      "learning_rate": 3.0153744658785057e-06,
      "loss": 0.3815,
      "step": 41664
    },
    {
      "epoch": 1.9120279014271948,
      "grad_norm": 0.448930561542511,
      "learning_rate": 3.0151494222988985e-06,
      "loss": 0.3544,
      "step": 41665
    },
    {
      "epoch": 1.9120737919324493,
      "grad_norm": 0.5043471455574036,
      "learning_rate": 3.0149243834922305e-06,
      "loss": 0.3809,
      "step": 41666
    },
    {
      "epoch": 1.9121196824377038,
      "grad_norm": 0.4646277129650116,
      "learning_rate": 3.014699349459042e-06,
      "loss": 0.3822,
      "step": 41667
    },
    {
      "epoch": 1.912165572942958,
      "grad_norm": 0.4866785705089569,
      "learning_rate": 3.014474320199876e-06,
      "loss": 0.3686,
      "step": 41668
    },
    {
      "epoch": 1.9122114634482126,
      "grad_norm": 0.4378274977207184,
      "learning_rate": 3.0142492957152724e-06,
      "loss": 0.2878,
      "step": 41669
    },
    {
      "epoch": 1.912257353953467,
      "grad_norm": 0.45598047971725464,
      "learning_rate": 3.0140242760057702e-06,
      "loss": 0.3592,
      "step": 41670
    },
    {
      "epoch": 1.9123032444587214,
      "grad_norm": 0.4920518398284912,
      "learning_rate": 3.0137992610719152e-06,
      "loss": 0.3779,
      "step": 41671
    },
    {
      "epoch": 1.9123491349639758,
      "grad_norm": 0.46314477920532227,
      "learning_rate": 3.013574250914246e-06,
      "loss": 0.3552,
      "step": 41672
    },
    {
      "epoch": 1.9123950254692303,
      "grad_norm": 0.44451725482940674,
      "learning_rate": 3.013349245533303e-06,
      "loss": 0.3237,
      "step": 41673
    },
    {
      "epoch": 1.9124409159744848,
      "grad_norm": 0.43934786319732666,
      "learning_rate": 3.0131242449296287e-06,
      "loss": 0.3014,
      "step": 41674
    },
    {
      "epoch": 1.9124868064797393,
      "grad_norm": 0.4652961790561676,
      "learning_rate": 3.0128992491037638e-06,
      "loss": 0.3463,
      "step": 41675
    },
    {
      "epoch": 1.9125326969849938,
      "grad_norm": 0.486237108707428,
      "learning_rate": 3.012674258056247e-06,
      "loss": 0.3393,
      "step": 41676
    },
    {
      "epoch": 1.9125785874902483,
      "grad_norm": 0.48892784118652344,
      "learning_rate": 3.0124492717876236e-06,
      "loss": 0.3788,
      "step": 41677
    },
    {
      "epoch": 1.9126244779955028,
      "grad_norm": 0.45125612616539,
      "learning_rate": 3.0122242902984334e-06,
      "loss": 0.3546,
      "step": 41678
    },
    {
      "epoch": 1.9126703685007573,
      "grad_norm": 0.49366915225982666,
      "learning_rate": 3.0119993135892127e-06,
      "loss": 0.3437,
      "step": 41679
    },
    {
      "epoch": 1.9127162590060116,
      "grad_norm": 0.47382959723472595,
      "learning_rate": 3.0117743416605088e-06,
      "loss": 0.3591,
      "step": 41680
    },
    {
      "epoch": 1.912762149511266,
      "grad_norm": 0.4751995801925659,
      "learning_rate": 3.01154937451286e-06,
      "loss": 0.3841,
      "step": 41681
    },
    {
      "epoch": 1.9128080400165206,
      "grad_norm": 0.4567111134529114,
      "learning_rate": 3.0113244121468077e-06,
      "loss": 0.3456,
      "step": 41682
    },
    {
      "epoch": 1.9128539305217749,
      "grad_norm": 0.4498870074748993,
      "learning_rate": 3.0110994545628913e-06,
      "loss": 0.3216,
      "step": 41683
    },
    {
      "epoch": 1.9128998210270294,
      "grad_norm": 0.4735798239707947,
      "learning_rate": 3.0108745017616537e-06,
      "loss": 0.3751,
      "step": 41684
    },
    {
      "epoch": 1.9129457115322839,
      "grad_norm": 0.4554518759250641,
      "learning_rate": 3.0106495537436353e-06,
      "loss": 0.3378,
      "step": 41685
    },
    {
      "epoch": 1.9129916020375384,
      "grad_norm": 0.47784242033958435,
      "learning_rate": 3.010424610509375e-06,
      "loss": 0.3506,
      "step": 41686
    },
    {
      "epoch": 1.9130374925427929,
      "grad_norm": 0.4618084132671356,
      "learning_rate": 3.0101996720594177e-06,
      "loss": 0.3247,
      "step": 41687
    },
    {
      "epoch": 1.9130833830480474,
      "grad_norm": 0.4639759063720703,
      "learning_rate": 3.0099747383943023e-06,
      "loss": 0.3454,
      "step": 41688
    },
    {
      "epoch": 1.9131292735533019,
      "grad_norm": 0.48087120056152344,
      "learning_rate": 3.009749809514568e-06,
      "loss": 0.3583,
      "step": 41689
    },
    {
      "epoch": 1.9131751640585564,
      "grad_norm": 0.4728133976459503,
      "learning_rate": 3.0095248854207586e-06,
      "loss": 0.4158,
      "step": 41690
    },
    {
      "epoch": 1.9132210545638109,
      "grad_norm": 0.4833429753780365,
      "learning_rate": 3.0092999661134137e-06,
      "loss": 0.3768,
      "step": 41691
    },
    {
      "epoch": 1.9132669450690654,
      "grad_norm": 0.459972083568573,
      "learning_rate": 3.009075051593072e-06,
      "loss": 0.3203,
      "step": 41692
    },
    {
      "epoch": 1.9133128355743196,
      "grad_norm": 0.4629722535610199,
      "learning_rate": 3.008850141860279e-06,
      "loss": 0.3023,
      "step": 41693
    },
    {
      "epoch": 1.9133587260795741,
      "grad_norm": 0.4597342014312744,
      "learning_rate": 3.0086252369155723e-06,
      "loss": 0.3535,
      "step": 41694
    },
    {
      "epoch": 1.9134046165848286,
      "grad_norm": 0.4700259268283844,
      "learning_rate": 3.0084003367594927e-06,
      "loss": 0.3451,
      "step": 41695
    },
    {
      "epoch": 1.913450507090083,
      "grad_norm": 0.4755796492099762,
      "learning_rate": 3.0081754413925824e-06,
      "loss": 0.3337,
      "step": 41696
    },
    {
      "epoch": 1.9134963975953374,
      "grad_norm": 0.4516092538833618,
      "learning_rate": 3.0079505508153817e-06,
      "loss": 0.3425,
      "step": 41697
    },
    {
      "epoch": 1.913542288100592,
      "grad_norm": 0.5098990201950073,
      "learning_rate": 3.0077256650284292e-06,
      "loss": 0.3783,
      "step": 41698
    },
    {
      "epoch": 1.9135881786058464,
      "grad_norm": 0.4617146849632263,
      "learning_rate": 3.0075007840322697e-06,
      "loss": 0.3262,
      "step": 41699
    },
    {
      "epoch": 1.913634069111101,
      "grad_norm": 0.44453758001327515,
      "learning_rate": 3.0072759078274427e-06,
      "loss": 0.3092,
      "step": 41700
    },
    {
      "epoch": 1.9136799596163554,
      "grad_norm": 0.48229146003723145,
      "learning_rate": 3.0070510364144856e-06,
      "loss": 0.3955,
      "step": 41701
    },
    {
      "epoch": 1.91372585012161,
      "grad_norm": 0.45639634132385254,
      "learning_rate": 3.006826169793944e-06,
      "loss": 0.3374,
      "step": 41702
    },
    {
      "epoch": 1.9137717406268644,
      "grad_norm": 0.47794729471206665,
      "learning_rate": 3.0066013079663557e-06,
      "loss": 0.3892,
      "step": 41703
    },
    {
      "epoch": 1.913817631132119,
      "grad_norm": 0.4468208849430084,
      "learning_rate": 3.0063764509322614e-06,
      "loss": 0.32,
      "step": 41704
    },
    {
      "epoch": 1.9138635216373734,
      "grad_norm": 0.4624806046485901,
      "learning_rate": 3.006151598692204e-06,
      "loss": 0.3383,
      "step": 41705
    },
    {
      "epoch": 1.9139094121426277,
      "grad_norm": 0.47448474168777466,
      "learning_rate": 3.005926751246722e-06,
      "loss": 0.3646,
      "step": 41706
    },
    {
      "epoch": 1.9139553026478822,
      "grad_norm": 0.4930984675884247,
      "learning_rate": 3.0057019085963557e-06,
      "loss": 0.3728,
      "step": 41707
    },
    {
      "epoch": 1.9140011931531367,
      "grad_norm": 0.5671778917312622,
      "learning_rate": 3.005477070741648e-06,
      "loss": 0.3373,
      "step": 41708
    },
    {
      "epoch": 1.914047083658391,
      "grad_norm": 0.4597375690937042,
      "learning_rate": 3.005252237683139e-06,
      "loss": 0.3482,
      "step": 41709
    },
    {
      "epoch": 1.9140929741636454,
      "grad_norm": 0.44170600175857544,
      "learning_rate": 3.0050274094213674e-06,
      "loss": 0.2914,
      "step": 41710
    },
    {
      "epoch": 1.9141388646689,
      "grad_norm": 0.45526987314224243,
      "learning_rate": 3.004802585956876e-06,
      "loss": 0.3254,
      "step": 41711
    },
    {
      "epoch": 1.9141847551741544,
      "grad_norm": 0.43136322498321533,
      "learning_rate": 3.0045777672902053e-06,
      "loss": 0.2934,
      "step": 41712
    },
    {
      "epoch": 1.914230645679409,
      "grad_norm": 0.4659397006034851,
      "learning_rate": 3.0043529534218956e-06,
      "loss": 0.3284,
      "step": 41713
    },
    {
      "epoch": 1.9142765361846634,
      "grad_norm": 0.49269405007362366,
      "learning_rate": 3.004128144352485e-06,
      "loss": 0.3695,
      "step": 41714
    },
    {
      "epoch": 1.914322426689918,
      "grad_norm": 0.4972403943538666,
      "learning_rate": 3.003903340082518e-06,
      "loss": 0.3705,
      "step": 41715
    },
    {
      "epoch": 1.9143683171951724,
      "grad_norm": 0.45791342854499817,
      "learning_rate": 3.0036785406125334e-06,
      "loss": 0.3819,
      "step": 41716
    },
    {
      "epoch": 1.914414207700427,
      "grad_norm": 0.4808557331562042,
      "learning_rate": 3.003453745943071e-06,
      "loss": 0.3487,
      "step": 41717
    },
    {
      "epoch": 1.9144600982056812,
      "grad_norm": 0.4396786689758301,
      "learning_rate": 3.003228956074673e-06,
      "loss": 0.3376,
      "step": 41718
    },
    {
      "epoch": 1.9145059887109357,
      "grad_norm": 0.45059651136398315,
      "learning_rate": 3.0030041710078794e-06,
      "loss": 0.321,
      "step": 41719
    },
    {
      "epoch": 1.9145518792161902,
      "grad_norm": 0.4912230372428894,
      "learning_rate": 3.0027793907432285e-06,
      "loss": 0.3637,
      "step": 41720
    },
    {
      "epoch": 1.9145977697214447,
      "grad_norm": 0.5117926001548767,
      "learning_rate": 3.0025546152812646e-06,
      "loss": 0.4122,
      "step": 41721
    },
    {
      "epoch": 1.914643660226699,
      "grad_norm": 0.4468610882759094,
      "learning_rate": 3.002329844622528e-06,
      "loss": 0.309,
      "step": 41722
    },
    {
      "epoch": 1.9146895507319535,
      "grad_norm": 0.47212472558021545,
      "learning_rate": 3.002105078767554e-06,
      "loss": 0.3654,
      "step": 41723
    },
    {
      "epoch": 1.914735441237208,
      "grad_norm": 0.4743143618106842,
      "learning_rate": 3.0018803177168885e-06,
      "loss": 0.3495,
      "step": 41724
    },
    {
      "epoch": 1.9147813317424625,
      "grad_norm": 0.5651989579200745,
      "learning_rate": 3.0016555614710715e-06,
      "loss": 0.457,
      "step": 41725
    },
    {
      "epoch": 1.914827222247717,
      "grad_norm": 0.4660301208496094,
      "learning_rate": 3.00143081003064e-06,
      "loss": 0.3527,
      "step": 41726
    },
    {
      "epoch": 1.9148731127529715,
      "grad_norm": 0.5289551019668579,
      "learning_rate": 3.001206063396138e-06,
      "loss": 0.4389,
      "step": 41727
    },
    {
      "epoch": 1.914919003258226,
      "grad_norm": 0.4840894937515259,
      "learning_rate": 3.000981321568105e-06,
      "loss": 0.3913,
      "step": 41728
    },
    {
      "epoch": 1.9149648937634804,
      "grad_norm": 0.48058992624282837,
      "learning_rate": 3.000756584547079e-06,
      "loss": 0.3857,
      "step": 41729
    },
    {
      "epoch": 1.915010784268735,
      "grad_norm": 0.4991056025028229,
      "learning_rate": 3.0005318523336038e-06,
      "loss": 0.376,
      "step": 41730
    },
    {
      "epoch": 1.9150566747739892,
      "grad_norm": 0.4563962519168854,
      "learning_rate": 3.0003071249282184e-06,
      "loss": 0.3283,
      "step": 41731
    },
    {
      "epoch": 1.9151025652792437,
      "grad_norm": 0.465788871049881,
      "learning_rate": 3.0000824023314624e-06,
      "loss": 0.3161,
      "step": 41732
    },
    {
      "epoch": 1.9151484557844982,
      "grad_norm": 0.5125126838684082,
      "learning_rate": 2.9998576845438783e-06,
      "loss": 0.4973,
      "step": 41733
    },
    {
      "epoch": 1.9151943462897525,
      "grad_norm": 0.42360371351242065,
      "learning_rate": 2.999632971566005e-06,
      "loss": 0.2756,
      "step": 41734
    },
    {
      "epoch": 1.915240236795007,
      "grad_norm": 0.477347195148468,
      "learning_rate": 2.9994082633983812e-06,
      "loss": 0.3666,
      "step": 41735
    },
    {
      "epoch": 1.9152861273002615,
      "grad_norm": 0.4850514233112335,
      "learning_rate": 2.999183560041551e-06,
      "loss": 0.3948,
      "step": 41736
    },
    {
      "epoch": 1.915332017805516,
      "grad_norm": 0.42634090781211853,
      "learning_rate": 2.998958861496053e-06,
      "loss": 0.284,
      "step": 41737
    },
    {
      "epoch": 1.9153779083107705,
      "grad_norm": 0.4769398868083954,
      "learning_rate": 2.998734167762426e-06,
      "loss": 0.4056,
      "step": 41738
    },
    {
      "epoch": 1.915423798816025,
      "grad_norm": 0.4478975236415863,
      "learning_rate": 2.9985094788412134e-06,
      "loss": 0.3403,
      "step": 41739
    },
    {
      "epoch": 1.9154696893212795,
      "grad_norm": 0.4741719961166382,
      "learning_rate": 2.998284794732953e-06,
      "loss": 0.3343,
      "step": 41740
    },
    {
      "epoch": 1.915515579826534,
      "grad_norm": 0.4893762767314911,
      "learning_rate": 2.998060115438185e-06,
      "loss": 0.3787,
      "step": 41741
    },
    {
      "epoch": 1.9155614703317885,
      "grad_norm": 0.4750825762748718,
      "learning_rate": 2.997835440957452e-06,
      "loss": 0.3804,
      "step": 41742
    },
    {
      "epoch": 1.915607360837043,
      "grad_norm": 0.43534931540489197,
      "learning_rate": 2.9976107712912926e-06,
      "loss": 0.2666,
      "step": 41743
    },
    {
      "epoch": 1.9156532513422972,
      "grad_norm": 0.43634799122810364,
      "learning_rate": 2.9973861064402463e-06,
      "loss": 0.2858,
      "step": 41744
    },
    {
      "epoch": 1.9156991418475517,
      "grad_norm": 0.45257246494293213,
      "learning_rate": 2.997161446404856e-06,
      "loss": 0.3376,
      "step": 41745
    },
    {
      "epoch": 1.9157450323528062,
      "grad_norm": 0.45347413420677185,
      "learning_rate": 2.9969367911856597e-06,
      "loss": 0.3504,
      "step": 41746
    },
    {
      "epoch": 1.9157909228580605,
      "grad_norm": 0.4663369357585907,
      "learning_rate": 2.9967121407831985e-06,
      "loss": 0.3057,
      "step": 41747
    },
    {
      "epoch": 1.915836813363315,
      "grad_norm": 0.4768209457397461,
      "learning_rate": 2.996487495198011e-06,
      "loss": 0.3583,
      "step": 41748
    },
    {
      "epoch": 1.9158827038685695,
      "grad_norm": 0.5064505934715271,
      "learning_rate": 2.9962628544306406e-06,
      "loss": 0.4014,
      "step": 41749
    },
    {
      "epoch": 1.915928594373824,
      "grad_norm": 0.4451788067817688,
      "learning_rate": 2.9960382184816243e-06,
      "loss": 0.3206,
      "step": 41750
    },
    {
      "epoch": 1.9159744848790785,
      "grad_norm": 0.4536818861961365,
      "learning_rate": 2.9958135873515026e-06,
      "loss": 0.327,
      "step": 41751
    },
    {
      "epoch": 1.916020375384333,
      "grad_norm": 0.45812085270881653,
      "learning_rate": 2.9955889610408183e-06,
      "loss": 0.3518,
      "step": 41752
    },
    {
      "epoch": 1.9160662658895875,
      "grad_norm": 0.44672730565071106,
      "learning_rate": 2.9953643395501103e-06,
      "loss": 0.3318,
      "step": 41753
    },
    {
      "epoch": 1.916112156394842,
      "grad_norm": 0.46509045362472534,
      "learning_rate": 2.995139722879917e-06,
      "loss": 0.3783,
      "step": 41754
    },
    {
      "epoch": 1.9161580469000965,
      "grad_norm": 0.46337124705314636,
      "learning_rate": 2.994915111030781e-06,
      "loss": 0.3425,
      "step": 41755
    },
    {
      "epoch": 1.916203937405351,
      "grad_norm": 0.492946058511734,
      "learning_rate": 2.9946905040032413e-06,
      "loss": 0.358,
      "step": 41756
    },
    {
      "epoch": 1.9162498279106053,
      "grad_norm": 0.466013103723526,
      "learning_rate": 2.994465901797836e-06,
      "loss": 0.3318,
      "step": 41757
    },
    {
      "epoch": 1.9162957184158598,
      "grad_norm": 0.4845457077026367,
      "learning_rate": 2.9942413044151096e-06,
      "loss": 0.3806,
      "step": 41758
    },
    {
      "epoch": 1.9163416089211143,
      "grad_norm": 0.47627297043800354,
      "learning_rate": 2.9940167118556e-06,
      "loss": 0.3617,
      "step": 41759
    },
    {
      "epoch": 1.9163874994263685,
      "grad_norm": 0.4811996817588806,
      "learning_rate": 2.9937921241198457e-06,
      "loss": 0.3874,
      "step": 41760
    },
    {
      "epoch": 1.916433389931623,
      "grad_norm": 0.474230021238327,
      "learning_rate": 2.993567541208389e-06,
      "loss": 0.3687,
      "step": 41761
    },
    {
      "epoch": 1.9164792804368775,
      "grad_norm": 0.47917526960372925,
      "learning_rate": 2.9933429631217698e-06,
      "loss": 0.3279,
      "step": 41762
    },
    {
      "epoch": 1.916525170942132,
      "grad_norm": 0.5076526999473572,
      "learning_rate": 2.993118389860525e-06,
      "loss": 0.3134,
      "step": 41763
    },
    {
      "epoch": 1.9165710614473865,
      "grad_norm": 0.4887468218803406,
      "learning_rate": 2.992893821425199e-06,
      "loss": 0.3605,
      "step": 41764
    },
    {
      "epoch": 1.916616951952641,
      "grad_norm": 0.4685856103897095,
      "learning_rate": 2.9926692578163296e-06,
      "loss": 0.3477,
      "step": 41765
    },
    {
      "epoch": 1.9166628424578955,
      "grad_norm": 0.5036227107048035,
      "learning_rate": 2.992444699034457e-06,
      "loss": 0.3543,
      "step": 41766
    },
    {
      "epoch": 1.91670873296315,
      "grad_norm": 0.463027685880661,
      "learning_rate": 2.9922201450801213e-06,
      "loss": 0.3557,
      "step": 41767
    },
    {
      "epoch": 1.9167546234684045,
      "grad_norm": 0.4274784326553345,
      "learning_rate": 2.9919955959538626e-06,
      "loss": 0.3065,
      "step": 41768
    },
    {
      "epoch": 1.9168005139736588,
      "grad_norm": 0.4676278829574585,
      "learning_rate": 2.9917710516562193e-06,
      "loss": 0.3318,
      "step": 41769
    },
    {
      "epoch": 1.9168464044789133,
      "grad_norm": 0.44727200269699097,
      "learning_rate": 2.991546512187736e-06,
      "loss": 0.2994,
      "step": 41770
    },
    {
      "epoch": 1.9168922949841678,
      "grad_norm": 0.44487136602401733,
      "learning_rate": 2.9913219775489477e-06,
      "loss": 0.3059,
      "step": 41771
    },
    {
      "epoch": 1.916938185489422,
      "grad_norm": 0.48411113023757935,
      "learning_rate": 2.9910974477403942e-06,
      "loss": 0.3739,
      "step": 41772
    },
    {
      "epoch": 1.9169840759946766,
      "grad_norm": 0.46533793210983276,
      "learning_rate": 2.99087292276262e-06,
      "loss": 0.3762,
      "step": 41773
    },
    {
      "epoch": 1.917029966499931,
      "grad_norm": 0.49383974075317383,
      "learning_rate": 2.990648402616162e-06,
      "loss": 0.4357,
      "step": 41774
    },
    {
      "epoch": 1.9170758570051856,
      "grad_norm": 0.4654548168182373,
      "learning_rate": 2.990423887301559e-06,
      "loss": 0.3421,
      "step": 41775
    },
    {
      "epoch": 1.91712174751044,
      "grad_norm": 0.44804930686950684,
      "learning_rate": 2.9901993768193536e-06,
      "loss": 0.3477,
      "step": 41776
    },
    {
      "epoch": 1.9171676380156946,
      "grad_norm": 0.4503944218158722,
      "learning_rate": 2.989974871170085e-06,
      "loss": 0.3107,
      "step": 41777
    },
    {
      "epoch": 1.917213528520949,
      "grad_norm": 0.445602148771286,
      "learning_rate": 2.9897503703542913e-06,
      "loss": 0.277,
      "step": 41778
    },
    {
      "epoch": 1.9172594190262036,
      "grad_norm": 0.464530348777771,
      "learning_rate": 2.989525874372512e-06,
      "loss": 0.3392,
      "step": 41779
    },
    {
      "epoch": 1.917305309531458,
      "grad_norm": 0.48807087540626526,
      "learning_rate": 2.9893013832252904e-06,
      "loss": 0.3583,
      "step": 41780
    },
    {
      "epoch": 1.9173512000367126,
      "grad_norm": 0.5158286094665527,
      "learning_rate": 2.9890768969131646e-06,
      "loss": 0.4432,
      "step": 41781
    },
    {
      "epoch": 1.9173970905419668,
      "grad_norm": 0.46204522252082825,
      "learning_rate": 2.9888524154366728e-06,
      "loss": 0.3314,
      "step": 41782
    },
    {
      "epoch": 1.9174429810472213,
      "grad_norm": 0.4104793965816498,
      "learning_rate": 2.988627938796358e-06,
      "loss": 0.2535,
      "step": 41783
    },
    {
      "epoch": 1.9174888715524758,
      "grad_norm": 0.47452694177627563,
      "learning_rate": 2.9884034669927575e-06,
      "loss": 0.3315,
      "step": 41784
    },
    {
      "epoch": 1.91753476205773,
      "grad_norm": 0.4363560974597931,
      "learning_rate": 2.9881790000264097e-06,
      "loss": 0.3186,
      "step": 41785
    },
    {
      "epoch": 1.9175806525629846,
      "grad_norm": 0.4924233853816986,
      "learning_rate": 2.9879545378978593e-06,
      "loss": 0.4069,
      "step": 41786
    },
    {
      "epoch": 1.917626543068239,
      "grad_norm": 0.4382440447807312,
      "learning_rate": 2.9877300806076424e-06,
      "loss": 0.2962,
      "step": 41787
    },
    {
      "epoch": 1.9176724335734936,
      "grad_norm": 0.4669543504714966,
      "learning_rate": 2.9875056281562986e-06,
      "loss": 0.3679,
      "step": 41788
    },
    {
      "epoch": 1.917718324078748,
      "grad_norm": 0.4740910828113556,
      "learning_rate": 2.9872811805443704e-06,
      "loss": 0.3815,
      "step": 41789
    },
    {
      "epoch": 1.9177642145840026,
      "grad_norm": 0.4820399880409241,
      "learning_rate": 2.9870567377723948e-06,
      "loss": 0.3027,
      "step": 41790
    },
    {
      "epoch": 1.917810105089257,
      "grad_norm": 0.4799916744232178,
      "learning_rate": 2.9868322998409105e-06,
      "loss": 0.3652,
      "step": 41791
    },
    {
      "epoch": 1.9178559955945116,
      "grad_norm": 0.4766339063644409,
      "learning_rate": 2.9866078667504626e-06,
      "loss": 0.3562,
      "step": 41792
    },
    {
      "epoch": 1.917901886099766,
      "grad_norm": 0.48064756393432617,
      "learning_rate": 2.986383438501586e-06,
      "loss": 0.35,
      "step": 41793
    },
    {
      "epoch": 1.9179477766050206,
      "grad_norm": 0.4538975954055786,
      "learning_rate": 2.9861590150948194e-06,
      "loss": 0.3556,
      "step": 41794
    },
    {
      "epoch": 1.9179936671102749,
      "grad_norm": 0.46717819571495056,
      "learning_rate": 2.9859345965307074e-06,
      "loss": 0.3627,
      "step": 41795
    },
    {
      "epoch": 1.9180395576155294,
      "grad_norm": 0.47148364782333374,
      "learning_rate": 2.985710182809787e-06,
      "loss": 0.3388,
      "step": 41796
    },
    {
      "epoch": 1.9180854481207839,
      "grad_norm": 0.4607082009315491,
      "learning_rate": 2.9854857739325973e-06,
      "loss": 0.333,
      "step": 41797
    },
    {
      "epoch": 1.9181313386260381,
      "grad_norm": 0.48574763536453247,
      "learning_rate": 2.9852613698996786e-06,
      "loss": 0.3607,
      "step": 41798
    },
    {
      "epoch": 1.9181772291312926,
      "grad_norm": 0.48474326729774475,
      "learning_rate": 2.985036970711571e-06,
      "loss": 0.3352,
      "step": 41799
    },
    {
      "epoch": 1.9182231196365471,
      "grad_norm": 0.47861436009407043,
      "learning_rate": 2.9848125763688114e-06,
      "loss": 0.358,
      "step": 41800
    },
    {
      "epoch": 1.9182690101418016,
      "grad_norm": 0.47306203842163086,
      "learning_rate": 2.9845881868719432e-06,
      "loss": 0.3671,
      "step": 41801
    },
    {
      "epoch": 1.9183149006470561,
      "grad_norm": 0.4566236734390259,
      "learning_rate": 2.984363802221505e-06,
      "loss": 0.3141,
      "step": 41802
    },
    {
      "epoch": 1.9183607911523106,
      "grad_norm": 0.4790041148662567,
      "learning_rate": 2.984139422418034e-06,
      "loss": 0.3489,
      "step": 41803
    },
    {
      "epoch": 1.9184066816575651,
      "grad_norm": 0.49721789360046387,
      "learning_rate": 2.9839150474620725e-06,
      "loss": 0.4003,
      "step": 41804
    },
    {
      "epoch": 1.9184525721628196,
      "grad_norm": 0.44512462615966797,
      "learning_rate": 2.9836906773541596e-06,
      "loss": 0.3274,
      "step": 41805
    },
    {
      "epoch": 1.9184984626680741,
      "grad_norm": 0.41029343008995056,
      "learning_rate": 2.983466312094832e-06,
      "loss": 0.2645,
      "step": 41806
    },
    {
      "epoch": 1.9185443531733284,
      "grad_norm": 0.4615787863731384,
      "learning_rate": 2.9832419516846333e-06,
      "loss": 0.3531,
      "step": 41807
    },
    {
      "epoch": 1.9185902436785829,
      "grad_norm": 0.4936063885688782,
      "learning_rate": 2.9830175961241013e-06,
      "loss": 0.3863,
      "step": 41808
    },
    {
      "epoch": 1.9186361341838374,
      "grad_norm": 0.49395230412483215,
      "learning_rate": 2.9827932454137748e-06,
      "loss": 0.3607,
      "step": 41809
    },
    {
      "epoch": 1.9186820246890919,
      "grad_norm": 0.46430283784866333,
      "learning_rate": 2.9825688995541946e-06,
      "loss": 0.3398,
      "step": 41810
    },
    {
      "epoch": 1.9187279151943462,
      "grad_norm": 0.4747788906097412,
      "learning_rate": 2.9823445585458995e-06,
      "loss": 0.407,
      "step": 41811
    },
    {
      "epoch": 1.9187738056996007,
      "grad_norm": 0.48410215973854065,
      "learning_rate": 2.9821202223894286e-06,
      "loss": 0.3696,
      "step": 41812
    },
    {
      "epoch": 1.9188196962048552,
      "grad_norm": 0.4767077565193176,
      "learning_rate": 2.98189589108532e-06,
      "loss": 0.3338,
      "step": 41813
    },
    {
      "epoch": 1.9188655867101096,
      "grad_norm": 0.47598183155059814,
      "learning_rate": 2.9816715646341176e-06,
      "loss": 0.3367,
      "step": 41814
    },
    {
      "epoch": 1.9189114772153641,
      "grad_norm": 0.4904116094112396,
      "learning_rate": 2.981447243036357e-06,
      "loss": 0.3998,
      "step": 41815
    },
    {
      "epoch": 1.9189573677206186,
      "grad_norm": 0.4456970691680908,
      "learning_rate": 2.9812229262925773e-06,
      "loss": 0.3401,
      "step": 41816
    },
    {
      "epoch": 1.9190032582258731,
      "grad_norm": 0.4662283957004547,
      "learning_rate": 2.980998614403321e-06,
      "loss": 0.3272,
      "step": 41817
    },
    {
      "epoch": 1.9190491487311276,
      "grad_norm": 0.44600093364715576,
      "learning_rate": 2.9807743073691264e-06,
      "loss": 0.2907,
      "step": 41818
    },
    {
      "epoch": 1.9190950392363821,
      "grad_norm": 0.4765022099018097,
      "learning_rate": 2.9805500051905304e-06,
      "loss": 0.3398,
      "step": 41819
    },
    {
      "epoch": 1.9191409297416364,
      "grad_norm": 0.4777287542819977,
      "learning_rate": 2.980325707868076e-06,
      "loss": 0.3532,
      "step": 41820
    },
    {
      "epoch": 1.919186820246891,
      "grad_norm": 0.4977973699569702,
      "learning_rate": 2.9801014154023002e-06,
      "loss": 0.4108,
      "step": 41821
    },
    {
      "epoch": 1.9192327107521454,
      "grad_norm": 0.5042879581451416,
      "learning_rate": 2.9798771277937412e-06,
      "loss": 0.4072,
      "step": 41822
    },
    {
      "epoch": 1.9192786012573997,
      "grad_norm": 0.5028910040855408,
      "learning_rate": 2.9796528450429424e-06,
      "loss": 0.3816,
      "step": 41823
    },
    {
      "epoch": 1.9193244917626542,
      "grad_norm": 0.4690171480178833,
      "learning_rate": 2.979428567150441e-06,
      "loss": 0.2964,
      "step": 41824
    },
    {
      "epoch": 1.9193703822679087,
      "grad_norm": 0.4921649694442749,
      "learning_rate": 2.9792042941167744e-06,
      "loss": 0.3734,
      "step": 41825
    },
    {
      "epoch": 1.9194162727731632,
      "grad_norm": 0.4507807791233063,
      "learning_rate": 2.978980025942485e-06,
      "loss": 0.2937,
      "step": 41826
    },
    {
      "epoch": 1.9194621632784177,
      "grad_norm": 0.45530906319618225,
      "learning_rate": 2.9787557626281118e-06,
      "loss": 0.3335,
      "step": 41827
    },
    {
      "epoch": 1.9195080537836722,
      "grad_norm": 0.47488731145858765,
      "learning_rate": 2.9785315041741897e-06,
      "loss": 0.3433,
      "step": 41828
    },
    {
      "epoch": 1.9195539442889267,
      "grad_norm": 0.46775805950164795,
      "learning_rate": 2.978307250581264e-06,
      "loss": 0.3538,
      "step": 41829
    },
    {
      "epoch": 1.9195998347941812,
      "grad_norm": 0.47717270255088806,
      "learning_rate": 2.978083001849871e-06,
      "loss": 0.3824,
      "step": 41830
    },
    {
      "epoch": 1.9196457252994357,
      "grad_norm": 0.4396953582763672,
      "learning_rate": 2.9778587579805496e-06,
      "loss": 0.3133,
      "step": 41831
    },
    {
      "epoch": 1.9196916158046902,
      "grad_norm": 0.4536491334438324,
      "learning_rate": 2.97763451897384e-06,
      "loss": 0.3469,
      "step": 41832
    },
    {
      "epoch": 1.9197375063099444,
      "grad_norm": 0.5234580636024475,
      "learning_rate": 2.977410284830282e-06,
      "loss": 0.4376,
      "step": 41833
    },
    {
      "epoch": 1.919783396815199,
      "grad_norm": 0.48807504773139954,
      "learning_rate": 2.9771860555504113e-06,
      "loss": 0.3963,
      "step": 41834
    },
    {
      "epoch": 1.9198292873204534,
      "grad_norm": 0.4543924927711487,
      "learning_rate": 2.976961831134772e-06,
      "loss": 0.3082,
      "step": 41835
    },
    {
      "epoch": 1.9198751778257077,
      "grad_norm": 0.4713936746120453,
      "learning_rate": 2.976737611583902e-06,
      "loss": 0.3292,
      "step": 41836
    },
    {
      "epoch": 1.9199210683309622,
      "grad_norm": 0.41817507147789,
      "learning_rate": 2.976513396898336e-06,
      "loss": 0.2741,
      "step": 41837
    },
    {
      "epoch": 1.9199669588362167,
      "grad_norm": 0.44113749265670776,
      "learning_rate": 2.976289187078619e-06,
      "loss": 0.2849,
      "step": 41838
    },
    {
      "epoch": 1.9200128493414712,
      "grad_norm": 0.7265335917472839,
      "learning_rate": 2.9760649821252885e-06,
      "loss": 0.4523,
      "step": 41839
    },
    {
      "epoch": 1.9200587398467257,
      "grad_norm": 0.4962233006954193,
      "learning_rate": 2.9758407820388812e-06,
      "loss": 0.3799,
      "step": 41840
    },
    {
      "epoch": 1.9201046303519802,
      "grad_norm": 0.46318161487579346,
      "learning_rate": 2.9756165868199387e-06,
      "loss": 0.3694,
      "step": 41841
    },
    {
      "epoch": 1.9201505208572347,
      "grad_norm": 0.459846168756485,
      "learning_rate": 2.9753923964690003e-06,
      "loss": 0.3117,
      "step": 41842
    },
    {
      "epoch": 1.9201964113624892,
      "grad_norm": 0.4549297094345093,
      "learning_rate": 2.9751682109866015e-06,
      "loss": 0.3496,
      "step": 41843
    },
    {
      "epoch": 1.9202423018677437,
      "grad_norm": 0.5336573719978333,
      "learning_rate": 2.9749440303732857e-06,
      "loss": 0.3525,
      "step": 41844
    },
    {
      "epoch": 1.9202881923729982,
      "grad_norm": 0.459121972322464,
      "learning_rate": 2.9747198546295914e-06,
      "loss": 0.3433,
      "step": 41845
    },
    {
      "epoch": 1.9203340828782525,
      "grad_norm": 0.47601741552352905,
      "learning_rate": 2.9744956837560563e-06,
      "loss": 0.3341,
      "step": 41846
    },
    {
      "epoch": 1.920379973383507,
      "grad_norm": 0.4690493643283844,
      "learning_rate": 2.974271517753219e-06,
      "loss": 0.3385,
      "step": 41847
    },
    {
      "epoch": 1.9204258638887615,
      "grad_norm": 0.4414122402667999,
      "learning_rate": 2.9740473566216198e-06,
      "loss": 0.3003,
      "step": 41848
    },
    {
      "epoch": 1.9204717543940157,
      "grad_norm": 0.5038902759552002,
      "learning_rate": 2.973823200361797e-06,
      "loss": 0.3752,
      "step": 41849
    },
    {
      "epoch": 1.9205176448992702,
      "grad_norm": 0.4800187051296234,
      "learning_rate": 2.973599048974289e-06,
      "loss": 0.3849,
      "step": 41850
    },
    {
      "epoch": 1.9205635354045247,
      "grad_norm": 0.45802515745162964,
      "learning_rate": 2.973374902459637e-06,
      "loss": 0.3291,
      "step": 41851
    },
    {
      "epoch": 1.9206094259097792,
      "grad_norm": 0.4946646988391876,
      "learning_rate": 2.97315076081838e-06,
      "loss": 0.3672,
      "step": 41852
    },
    {
      "epoch": 1.9206553164150337,
      "grad_norm": 0.4805428087711334,
      "learning_rate": 2.9729266240510528e-06,
      "loss": 0.3404,
      "step": 41853
    },
    {
      "epoch": 1.9207012069202882,
      "grad_norm": 0.502227246761322,
      "learning_rate": 2.972702492158199e-06,
      "loss": 0.3473,
      "step": 41854
    },
    {
      "epoch": 1.9207470974255427,
      "grad_norm": 0.4485898017883301,
      "learning_rate": 2.9724783651403564e-06,
      "loss": 0.3062,
      "step": 41855
    },
    {
      "epoch": 1.9207929879307972,
      "grad_norm": 0.4735080897808075,
      "learning_rate": 2.972254242998061e-06,
      "loss": 0.3378,
      "step": 41856
    },
    {
      "epoch": 1.9208388784360517,
      "grad_norm": 0.459975928068161,
      "learning_rate": 2.9720301257318564e-06,
      "loss": 0.351,
      "step": 41857
    },
    {
      "epoch": 1.920884768941306,
      "grad_norm": 0.5321691632270813,
      "learning_rate": 2.97180601334228e-06,
      "loss": 0.4083,
      "step": 41858
    },
    {
      "epoch": 1.9209306594465605,
      "grad_norm": 0.44991692900657654,
      "learning_rate": 2.9715819058298678e-06,
      "loss": 0.2747,
      "step": 41859
    },
    {
      "epoch": 1.920976549951815,
      "grad_norm": 0.5082188248634338,
      "learning_rate": 2.971357803195162e-06,
      "loss": 0.4257,
      "step": 41860
    },
    {
      "epoch": 1.9210224404570693,
      "grad_norm": 0.4544338881969452,
      "learning_rate": 2.971133705438701e-06,
      "loss": 0.3695,
      "step": 41861
    },
    {
      "epoch": 1.9210683309623238,
      "grad_norm": 0.4720914959907532,
      "learning_rate": 2.9709096125610216e-06,
      "loss": 0.3757,
      "step": 41862
    },
    {
      "epoch": 1.9211142214675783,
      "grad_norm": 0.44792866706848145,
      "learning_rate": 2.970685524562665e-06,
      "loss": 0.3271,
      "step": 41863
    },
    {
      "epoch": 1.9211601119728328,
      "grad_norm": 0.46573078632354736,
      "learning_rate": 2.97046144144417e-06,
      "loss": 0.3257,
      "step": 41864
    },
    {
      "epoch": 1.9212060024780873,
      "grad_norm": 0.49484550952911377,
      "learning_rate": 2.9702373632060726e-06,
      "loss": 0.404,
      "step": 41865
    },
    {
      "epoch": 1.9212518929833418,
      "grad_norm": 0.49919289350509644,
      "learning_rate": 2.9700132898489152e-06,
      "loss": 0.4288,
      "step": 41866
    },
    {
      "epoch": 1.9212977834885963,
      "grad_norm": 0.48735877871513367,
      "learning_rate": 2.969789221373236e-06,
      "loss": 0.3925,
      "step": 41867
    },
    {
      "epoch": 1.9213436739938508,
      "grad_norm": 0.5008126497268677,
      "learning_rate": 2.9695651577795715e-06,
      "loss": 0.385,
      "step": 41868
    },
    {
      "epoch": 1.9213895644991053,
      "grad_norm": 0.49752306938171387,
      "learning_rate": 2.9693410990684625e-06,
      "loss": 0.3901,
      "step": 41869
    },
    {
      "epoch": 1.9214354550043597,
      "grad_norm": 0.47773054242134094,
      "learning_rate": 2.9691170452404485e-06,
      "loss": 0.3294,
      "step": 41870
    },
    {
      "epoch": 1.921481345509614,
      "grad_norm": 0.45417356491088867,
      "learning_rate": 2.9688929962960635e-06,
      "loss": 0.3488,
      "step": 41871
    },
    {
      "epoch": 1.9215272360148685,
      "grad_norm": 0.467722088098526,
      "learning_rate": 2.9686689522358526e-06,
      "loss": 0.2979,
      "step": 41872
    },
    {
      "epoch": 1.921573126520123,
      "grad_norm": 0.6165967583656311,
      "learning_rate": 2.968444913060352e-06,
      "loss": 0.3496,
      "step": 41873
    },
    {
      "epoch": 1.9216190170253773,
      "grad_norm": 0.5154542326927185,
      "learning_rate": 2.968220878770098e-06,
      "loss": 0.4205,
      "step": 41874
    },
    {
      "epoch": 1.9216649075306318,
      "grad_norm": 0.5117013454437256,
      "learning_rate": 2.967996849365634e-06,
      "loss": 0.4294,
      "step": 41875
    },
    {
      "epoch": 1.9217107980358863,
      "grad_norm": 0.4553329348564148,
      "learning_rate": 2.9677728248474957e-06,
      "loss": 0.3397,
      "step": 41876
    },
    {
      "epoch": 1.9217566885411408,
      "grad_norm": 0.4612378180027008,
      "learning_rate": 2.967548805216223e-06,
      "loss": 0.3517,
      "step": 41877
    },
    {
      "epoch": 1.9218025790463953,
      "grad_norm": 0.48482367396354675,
      "learning_rate": 2.967324790472351e-06,
      "loss": 0.3515,
      "step": 41878
    },
    {
      "epoch": 1.9218484695516498,
      "grad_norm": 0.46882227063179016,
      "learning_rate": 2.967100780616423e-06,
      "loss": 0.3729,
      "step": 41879
    },
    {
      "epoch": 1.9218943600569043,
      "grad_norm": 0.4302375316619873,
      "learning_rate": 2.966876775648978e-06,
      "loss": 0.2818,
      "step": 41880
    },
    {
      "epoch": 1.9219402505621588,
      "grad_norm": 0.45599886775016785,
      "learning_rate": 2.9666527755705498e-06,
      "loss": 0.349,
      "step": 41881
    },
    {
      "epoch": 1.9219861410674133,
      "grad_norm": 0.45980632305145264,
      "learning_rate": 2.9664287803816814e-06,
      "loss": 0.349,
      "step": 41882
    },
    {
      "epoch": 1.9220320315726678,
      "grad_norm": 0.43737518787384033,
      "learning_rate": 2.96620479008291e-06,
      "loss": 0.2985,
      "step": 41883
    },
    {
      "epoch": 1.922077922077922,
      "grad_norm": 0.43544867634773254,
      "learning_rate": 2.965980804674773e-06,
      "loss": 0.3108,
      "step": 41884
    },
    {
      "epoch": 1.9221238125831765,
      "grad_norm": 0.4507075548171997,
      "learning_rate": 2.9657568241578117e-06,
      "loss": 0.3197,
      "step": 41885
    },
    {
      "epoch": 1.922169703088431,
      "grad_norm": 0.5068269968032837,
      "learning_rate": 2.965532848532563e-06,
      "loss": 0.3928,
      "step": 41886
    },
    {
      "epoch": 1.9222155935936853,
      "grad_norm": 0.4722825288772583,
      "learning_rate": 2.9653088777995637e-06,
      "loss": 0.369,
      "step": 41887
    },
    {
      "epoch": 1.9222614840989398,
      "grad_norm": 0.45635172724723816,
      "learning_rate": 2.965084911959356e-06,
      "loss": 0.2801,
      "step": 41888
    },
    {
      "epoch": 1.9223073746041943,
      "grad_norm": 0.44466546177864075,
      "learning_rate": 2.9648609510124777e-06,
      "loss": 0.2994,
      "step": 41889
    },
    {
      "epoch": 1.9223532651094488,
      "grad_norm": 0.4614938497543335,
      "learning_rate": 2.964636994959465e-06,
      "loss": 0.3196,
      "step": 41890
    },
    {
      "epoch": 1.9223991556147033,
      "grad_norm": 0.4615688920021057,
      "learning_rate": 2.964413043800859e-06,
      "loss": 0.3464,
      "step": 41891
    },
    {
      "epoch": 1.9224450461199578,
      "grad_norm": 0.47216251492500305,
      "learning_rate": 2.9641890975371973e-06,
      "loss": 0.366,
      "step": 41892
    },
    {
      "epoch": 1.9224909366252123,
      "grad_norm": 0.43224194645881653,
      "learning_rate": 2.9639651561690163e-06,
      "loss": 0.2864,
      "step": 41893
    },
    {
      "epoch": 1.9225368271304668,
      "grad_norm": 0.4847928583621979,
      "learning_rate": 2.9637412196968585e-06,
      "loss": 0.4007,
      "step": 41894
    },
    {
      "epoch": 1.9225827176357213,
      "grad_norm": 0.468631774187088,
      "learning_rate": 2.9635172881212605e-06,
      "loss": 0.3392,
      "step": 41895
    },
    {
      "epoch": 1.9226286081409756,
      "grad_norm": 0.4627031087875366,
      "learning_rate": 2.963293361442759e-06,
      "loss": 0.3608,
      "step": 41896
    },
    {
      "epoch": 1.92267449864623,
      "grad_norm": 0.48067599534988403,
      "learning_rate": 2.963069439661895e-06,
      "loss": 0.3703,
      "step": 41897
    },
    {
      "epoch": 1.9227203891514846,
      "grad_norm": 0.49436983466148376,
      "learning_rate": 2.9628455227792076e-06,
      "loss": 0.3745,
      "step": 41898
    },
    {
      "epoch": 1.922766279656739,
      "grad_norm": 0.4585495889186859,
      "learning_rate": 2.962621610795231e-06,
      "loss": 0.3511,
      "step": 41899
    },
    {
      "epoch": 1.9228121701619934,
      "grad_norm": 0.46161770820617676,
      "learning_rate": 2.962397703710508e-06,
      "loss": 0.3551,
      "step": 41900
    },
    {
      "epoch": 1.9228580606672478,
      "grad_norm": 0.4922403395175934,
      "learning_rate": 2.962173801525576e-06,
      "loss": 0.4351,
      "step": 41901
    },
    {
      "epoch": 1.9229039511725023,
      "grad_norm": 0.43618670105934143,
      "learning_rate": 2.9619499042409717e-06,
      "loss": 0.2918,
      "step": 41902
    },
    {
      "epoch": 1.9229498416777568,
      "grad_norm": 0.4862567186355591,
      "learning_rate": 2.9617260118572354e-06,
      "loss": 0.3236,
      "step": 41903
    },
    {
      "epoch": 1.9229957321830113,
      "grad_norm": 0.4898208677768707,
      "learning_rate": 2.961502124374905e-06,
      "loss": 0.4289,
      "step": 41904
    },
    {
      "epoch": 1.9230416226882658,
      "grad_norm": 0.5007339715957642,
      "learning_rate": 2.9612782417945174e-06,
      "loss": 0.3542,
      "step": 41905
    },
    {
      "epoch": 1.9230875131935203,
      "grad_norm": 0.467084139585495,
      "learning_rate": 2.9610543641166135e-06,
      "loss": 0.3916,
      "step": 41906
    },
    {
      "epoch": 1.9231334036987748,
      "grad_norm": 0.5827847123146057,
      "learning_rate": 2.9608304913417298e-06,
      "loss": 0.3337,
      "step": 41907
    },
    {
      "epoch": 1.9231792942040293,
      "grad_norm": 0.4145813584327698,
      "learning_rate": 2.960606623470403e-06,
      "loss": 0.2769,
      "step": 41908
    },
    {
      "epoch": 1.9232251847092836,
      "grad_norm": 0.4832856059074402,
      "learning_rate": 2.960382760503177e-06,
      "loss": 0.3366,
      "step": 41909
    },
    {
      "epoch": 1.923271075214538,
      "grad_norm": 0.5123738050460815,
      "learning_rate": 2.9601589024405857e-06,
      "loss": 0.4106,
      "step": 41910
    },
    {
      "epoch": 1.9233169657197926,
      "grad_norm": 0.47792401909828186,
      "learning_rate": 2.9599350492831693e-06,
      "loss": 0.3562,
      "step": 41911
    },
    {
      "epoch": 1.9233628562250469,
      "grad_norm": 0.4488510191440582,
      "learning_rate": 2.959711201031463e-06,
      "loss": 0.3299,
      "step": 41912
    },
    {
      "epoch": 1.9234087467303014,
      "grad_norm": 0.4758816361427307,
      "learning_rate": 2.9594873576860094e-06,
      "loss": 0.3644,
      "step": 41913
    },
    {
      "epoch": 1.9234546372355559,
      "grad_norm": 0.48227789998054504,
      "learning_rate": 2.959263519247344e-06,
      "loss": 0.3515,
      "step": 41914
    },
    {
      "epoch": 1.9235005277408104,
      "grad_norm": 0.42213302850723267,
      "learning_rate": 2.9590396857160044e-06,
      "loss": 0.2736,
      "step": 41915
    },
    {
      "epoch": 1.9235464182460649,
      "grad_norm": 0.4177272617816925,
      "learning_rate": 2.9588158570925318e-06,
      "loss": 0.2875,
      "step": 41916
    },
    {
      "epoch": 1.9235923087513194,
      "grad_norm": 0.4797302484512329,
      "learning_rate": 2.9585920333774626e-06,
      "loss": 0.353,
      "step": 41917
    },
    {
      "epoch": 1.9236381992565739,
      "grad_norm": 0.4814547002315521,
      "learning_rate": 2.9583682145713343e-06,
      "loss": 0.3811,
      "step": 41918
    },
    {
      "epoch": 1.9236840897618284,
      "grad_norm": 0.4413779079914093,
      "learning_rate": 2.9581444006746874e-06,
      "loss": 0.3512,
      "step": 41919
    },
    {
      "epoch": 1.9237299802670829,
      "grad_norm": 0.4780769348144531,
      "learning_rate": 2.9579205916880588e-06,
      "loss": 0.3942,
      "step": 41920
    },
    {
      "epoch": 1.9237758707723374,
      "grad_norm": 0.459751695394516,
      "learning_rate": 2.9576967876119842e-06,
      "loss": 0.3266,
      "step": 41921
    },
    {
      "epoch": 1.9238217612775916,
      "grad_norm": 0.5005755424499512,
      "learning_rate": 2.957472988447007e-06,
      "loss": 0.4192,
      "step": 41922
    },
    {
      "epoch": 1.9238676517828461,
      "grad_norm": 0.5024755001068115,
      "learning_rate": 2.957249194193662e-06,
      "loss": 0.3712,
      "step": 41923
    },
    {
      "epoch": 1.9239135422881006,
      "grad_norm": 0.5143392086029053,
      "learning_rate": 2.9570254048524875e-06,
      "loss": 0.4486,
      "step": 41924
    },
    {
      "epoch": 1.923959432793355,
      "grad_norm": 0.49715369939804077,
      "learning_rate": 2.956801620424023e-06,
      "loss": 0.4206,
      "step": 41925
    },
    {
      "epoch": 1.9240053232986094,
      "grad_norm": 0.46908146142959595,
      "learning_rate": 2.9565778409088052e-06,
      "loss": 0.394,
      "step": 41926
    },
    {
      "epoch": 1.924051213803864,
      "grad_norm": 0.45585912466049194,
      "learning_rate": 2.9563540663073724e-06,
      "loss": 0.3059,
      "step": 41927
    },
    {
      "epoch": 1.9240971043091184,
      "grad_norm": 0.4791746437549591,
      "learning_rate": 2.956130296620264e-06,
      "loss": 0.4014,
      "step": 41928
    },
    {
      "epoch": 1.924142994814373,
      "grad_norm": 0.47247791290283203,
      "learning_rate": 2.9559065318480172e-06,
      "loss": 0.3084,
      "step": 41929
    },
    {
      "epoch": 1.9241888853196274,
      "grad_norm": 0.4580712914466858,
      "learning_rate": 2.9556827719911673e-06,
      "loss": 0.3355,
      "step": 41930
    },
    {
      "epoch": 1.924234775824882,
      "grad_norm": 0.5332343578338623,
      "learning_rate": 2.9554590170502585e-06,
      "loss": 0.4631,
      "step": 41931
    },
    {
      "epoch": 1.9242806663301364,
      "grad_norm": 0.4268239140510559,
      "learning_rate": 2.9552352670258246e-06,
      "loss": 0.3028,
      "step": 41932
    },
    {
      "epoch": 1.9243265568353909,
      "grad_norm": 0.4749758541584015,
      "learning_rate": 2.9550115219184034e-06,
      "loss": 0.3715,
      "step": 41933
    },
    {
      "epoch": 1.9243724473406454,
      "grad_norm": 0.5002425909042358,
      "learning_rate": 2.9547877817285353e-06,
      "loss": 0.3831,
      "step": 41934
    },
    {
      "epoch": 1.9244183378458997,
      "grad_norm": 0.6400960087776184,
      "learning_rate": 2.9545640464567573e-06,
      "loss": 0.382,
      "step": 41935
    },
    {
      "epoch": 1.9244642283511542,
      "grad_norm": 0.46105068922042847,
      "learning_rate": 2.9543403161036048e-06,
      "loss": 0.3584,
      "step": 41936
    },
    {
      "epoch": 1.9245101188564087,
      "grad_norm": 0.48778751492500305,
      "learning_rate": 2.9541165906696207e-06,
      "loss": 0.374,
      "step": 41937
    },
    {
      "epoch": 1.924556009361663,
      "grad_norm": 0.4511970579624176,
      "learning_rate": 2.9538928701553403e-06,
      "loss": 0.3096,
      "step": 41938
    },
    {
      "epoch": 1.9246018998669174,
      "grad_norm": 0.4820083677768707,
      "learning_rate": 2.9536691545613007e-06,
      "loss": 0.3392,
      "step": 41939
    },
    {
      "epoch": 1.924647790372172,
      "grad_norm": 0.4615422785282135,
      "learning_rate": 2.9534454438880422e-06,
      "loss": 0.3515,
      "step": 41940
    },
    {
      "epoch": 1.9246936808774264,
      "grad_norm": 0.5101310610771179,
      "learning_rate": 2.9532217381361006e-06,
      "loss": 0.3376,
      "step": 41941
    },
    {
      "epoch": 1.924739571382681,
      "grad_norm": 0.4846838414669037,
      "learning_rate": 2.9529980373060155e-06,
      "loss": 0.3956,
      "step": 41942
    },
    {
      "epoch": 1.9247854618879354,
      "grad_norm": 0.4240683615207672,
      "learning_rate": 2.9527743413983222e-06,
      "loss": 0.2859,
      "step": 41943
    },
    {
      "epoch": 1.92483135239319,
      "grad_norm": 0.46952730417251587,
      "learning_rate": 2.9525506504135624e-06,
      "loss": 0.3716,
      "step": 41944
    },
    {
      "epoch": 1.9248772428984444,
      "grad_norm": 0.49186116456985474,
      "learning_rate": 2.9523269643522718e-06,
      "loss": 0.39,
      "step": 41945
    },
    {
      "epoch": 1.924923133403699,
      "grad_norm": 0.4616750180721283,
      "learning_rate": 2.9521032832149878e-06,
      "loss": 0.2902,
      "step": 41946
    },
    {
      "epoch": 1.9249690239089532,
      "grad_norm": 0.4390864670276642,
      "learning_rate": 2.95187960700225e-06,
      "loss": 0.3061,
      "step": 41947
    },
    {
      "epoch": 1.9250149144142077,
      "grad_norm": 0.4713165760040283,
      "learning_rate": 2.951655935714595e-06,
      "loss": 0.3829,
      "step": 41948
    },
    {
      "epoch": 1.9250608049194622,
      "grad_norm": 0.4831652343273163,
      "learning_rate": 2.9514322693525604e-06,
      "loss": 0.3231,
      "step": 41949
    },
    {
      "epoch": 1.9251066954247165,
      "grad_norm": 0.4459077715873718,
      "learning_rate": 2.9512086079166855e-06,
      "loss": 0.3233,
      "step": 41950
    },
    {
      "epoch": 1.925152585929971,
      "grad_norm": 0.4912387728691101,
      "learning_rate": 2.9509849514075075e-06,
      "loss": 0.3933,
      "step": 41951
    },
    {
      "epoch": 1.9251984764352255,
      "grad_norm": 0.4813597798347473,
      "learning_rate": 2.950761299825562e-06,
      "loss": 0.3937,
      "step": 41952
    },
    {
      "epoch": 1.92524436694048,
      "grad_norm": 0.47595614194869995,
      "learning_rate": 2.9505376531713904e-06,
      "loss": 0.36,
      "step": 41953
    },
    {
      "epoch": 1.9252902574457345,
      "grad_norm": 0.4626908302307129,
      "learning_rate": 2.9503140114455296e-06,
      "loss": 0.3046,
      "step": 41954
    },
    {
      "epoch": 1.925336147950989,
      "grad_norm": 0.4720257818698883,
      "learning_rate": 2.9500903746485155e-06,
      "loss": 0.4229,
      "step": 41955
    },
    {
      "epoch": 1.9253820384562434,
      "grad_norm": 0.47402459383010864,
      "learning_rate": 2.9498667427808875e-06,
      "loss": 0.2719,
      "step": 41956
    },
    {
      "epoch": 1.925427928961498,
      "grad_norm": 0.4400371015071869,
      "learning_rate": 2.949643115843184e-06,
      "loss": 0.3466,
      "step": 41957
    },
    {
      "epoch": 1.9254738194667524,
      "grad_norm": 0.4624444842338562,
      "learning_rate": 2.9494194938359385e-06,
      "loss": 0.3417,
      "step": 41958
    },
    {
      "epoch": 1.925519709972007,
      "grad_norm": 0.46192869544029236,
      "learning_rate": 2.949195876759694e-06,
      "loss": 0.3461,
      "step": 41959
    },
    {
      "epoch": 1.9255656004772612,
      "grad_norm": 0.45404961705207825,
      "learning_rate": 2.9489722646149864e-06,
      "loss": 0.331,
      "step": 41960
    },
    {
      "epoch": 1.9256114909825157,
      "grad_norm": 0.46741676330566406,
      "learning_rate": 2.9487486574023525e-06,
      "loss": 0.3695,
      "step": 41961
    },
    {
      "epoch": 1.9256573814877702,
      "grad_norm": 0.44783446192741394,
      "learning_rate": 2.948525055122332e-06,
      "loss": 0.3213,
      "step": 41962
    },
    {
      "epoch": 1.9257032719930245,
      "grad_norm": 0.47805196046829224,
      "learning_rate": 2.948301457775461e-06,
      "loss": 0.3278,
      "step": 41963
    },
    {
      "epoch": 1.925749162498279,
      "grad_norm": 0.4132208526134491,
      "learning_rate": 2.9480778653622754e-06,
      "loss": 0.2631,
      "step": 41964
    },
    {
      "epoch": 1.9257950530035335,
      "grad_norm": 0.46063268184661865,
      "learning_rate": 2.9478542778833163e-06,
      "loss": 0.3333,
      "step": 41965
    },
    {
      "epoch": 1.925840943508788,
      "grad_norm": 0.4590644836425781,
      "learning_rate": 2.947630695339121e-06,
      "loss": 0.3227,
      "step": 41966
    },
    {
      "epoch": 1.9258868340140425,
      "grad_norm": 0.46503424644470215,
      "learning_rate": 2.9474071177302244e-06,
      "loss": 0.3523,
      "step": 41967
    },
    {
      "epoch": 1.925932724519297,
      "grad_norm": 0.41096875071525574,
      "learning_rate": 2.9471835450571674e-06,
      "loss": 0.2591,
      "step": 41968
    },
    {
      "epoch": 1.9259786150245515,
      "grad_norm": 0.4484167695045471,
      "learning_rate": 2.946959977320486e-06,
      "loss": 0.3343,
      "step": 41969
    },
    {
      "epoch": 1.926024505529806,
      "grad_norm": 0.470647931098938,
      "learning_rate": 2.9467364145207155e-06,
      "loss": 0.316,
      "step": 41970
    },
    {
      "epoch": 1.9260703960350605,
      "grad_norm": 0.46141326427459717,
      "learning_rate": 2.946512856658399e-06,
      "loss": 0.3207,
      "step": 41971
    },
    {
      "epoch": 1.926116286540315,
      "grad_norm": 0.43813079595565796,
      "learning_rate": 2.94628930373407e-06,
      "loss": 0.2788,
      "step": 41972
    },
    {
      "epoch": 1.9261621770455692,
      "grad_norm": 0.46059638261795044,
      "learning_rate": 2.946065755748265e-06,
      "loss": 0.3586,
      "step": 41973
    },
    {
      "epoch": 1.9262080675508237,
      "grad_norm": 0.4555175006389618,
      "learning_rate": 2.9458422127015263e-06,
      "loss": 0.3288,
      "step": 41974
    },
    {
      "epoch": 1.9262539580560782,
      "grad_norm": 0.48617950081825256,
      "learning_rate": 2.9456186745943884e-06,
      "loss": 0.3446,
      "step": 41975
    },
    {
      "epoch": 1.9262998485613325,
      "grad_norm": 0.48549461364746094,
      "learning_rate": 2.9453951414273885e-06,
      "loss": 0.3991,
      "step": 41976
    },
    {
      "epoch": 1.926345739066587,
      "grad_norm": 0.4570874869823456,
      "learning_rate": 2.9451716132010645e-06,
      "loss": 0.3232,
      "step": 41977
    },
    {
      "epoch": 1.9263916295718415,
      "grad_norm": 0.4721803069114685,
      "learning_rate": 2.9449480899159554e-06,
      "loss": 0.3737,
      "step": 41978
    },
    {
      "epoch": 1.926437520077096,
      "grad_norm": 0.5005239248275757,
      "learning_rate": 2.944724571572597e-06,
      "loss": 0.3825,
      "step": 41979
    },
    {
      "epoch": 1.9264834105823505,
      "grad_norm": 0.48354828357696533,
      "learning_rate": 2.9445010581715256e-06,
      "loss": 0.3402,
      "step": 41980
    },
    {
      "epoch": 1.926529301087605,
      "grad_norm": 0.47847437858581543,
      "learning_rate": 2.9442775497132823e-06,
      "loss": 0.3259,
      "step": 41981
    },
    {
      "epoch": 1.9265751915928595,
      "grad_norm": 0.4790976643562317,
      "learning_rate": 2.944054046198403e-06,
      "loss": 0.3718,
      "step": 41982
    },
    {
      "epoch": 1.926621082098114,
      "grad_norm": 0.5011014938354492,
      "learning_rate": 2.943830547627423e-06,
      "loss": 0.3964,
      "step": 41983
    },
    {
      "epoch": 1.9266669726033685,
      "grad_norm": 0.46058952808380127,
      "learning_rate": 2.943607054000884e-06,
      "loss": 0.3353,
      "step": 41984
    },
    {
      "epoch": 1.9267128631086228,
      "grad_norm": 0.4711500406265259,
      "learning_rate": 2.94338356531932e-06,
      "loss": 0.3217,
      "step": 41985
    },
    {
      "epoch": 1.9267587536138773,
      "grad_norm": 0.46995511651039124,
      "learning_rate": 2.9431600815832677e-06,
      "loss": 0.3688,
      "step": 41986
    },
    {
      "epoch": 1.9268046441191318,
      "grad_norm": 0.43752768635749817,
      "learning_rate": 2.942936602793268e-06,
      "loss": 0.3252,
      "step": 41987
    },
    {
      "epoch": 1.9268505346243863,
      "grad_norm": 0.4626831114292145,
      "learning_rate": 2.942713128949857e-06,
      "loss": 0.3409,
      "step": 41988
    },
    {
      "epoch": 1.9268964251296405,
      "grad_norm": 0.4503333270549774,
      "learning_rate": 2.94248966005357e-06,
      "loss": 0.3439,
      "step": 41989
    },
    {
      "epoch": 1.926942315634895,
      "grad_norm": 0.43884652853012085,
      "learning_rate": 2.942266196104948e-06,
      "loss": 0.2921,
      "step": 41990
    },
    {
      "epoch": 1.9269882061401495,
      "grad_norm": 0.44349929690361023,
      "learning_rate": 2.942042737104526e-06,
      "loss": 0.2878,
      "step": 41991
    },
    {
      "epoch": 1.927034096645404,
      "grad_norm": 0.4687816798686981,
      "learning_rate": 2.9418192830528393e-06,
      "loss": 0.3561,
      "step": 41992
    },
    {
      "epoch": 1.9270799871506585,
      "grad_norm": 0.522201657295227,
      "learning_rate": 2.9415958339504306e-06,
      "loss": 0.4591,
      "step": 41993
    },
    {
      "epoch": 1.927125877655913,
      "grad_norm": 0.5221208333969116,
      "learning_rate": 2.9413723897978335e-06,
      "loss": 0.3825,
      "step": 41994
    },
    {
      "epoch": 1.9271717681611675,
      "grad_norm": 0.4742391109466553,
      "learning_rate": 2.9411489505955844e-06,
      "loss": 0.3464,
      "step": 41995
    },
    {
      "epoch": 1.927217658666422,
      "grad_norm": 0.48604616522789,
      "learning_rate": 2.940925516344224e-06,
      "loss": 0.4005,
      "step": 41996
    },
    {
      "epoch": 1.9272635491716765,
      "grad_norm": 0.448048859834671,
      "learning_rate": 2.9407020870442877e-06,
      "loss": 0.3451,
      "step": 41997
    },
    {
      "epoch": 1.9273094396769308,
      "grad_norm": 0.46839645504951477,
      "learning_rate": 2.9404786626963123e-06,
      "loss": 0.3255,
      "step": 41998
    },
    {
      "epoch": 1.9273553301821853,
      "grad_norm": 0.46611514687538147,
      "learning_rate": 2.9402552433008365e-06,
      "loss": 0.3933,
      "step": 41999
    },
    {
      "epoch": 1.9274012206874398,
      "grad_norm": 0.4442223310470581,
      "learning_rate": 2.9400318288583964e-06,
      "loss": 0.3179,
      "step": 42000
    },
    {
      "epoch": 1.927447111192694,
      "grad_norm": 0.45116138458251953,
      "learning_rate": 2.9398084193695282e-06,
      "loss": 0.3284,
      "step": 42001
    },
    {
      "epoch": 1.9274930016979486,
      "grad_norm": 0.6890984773635864,
      "learning_rate": 2.939585014834772e-06,
      "loss": 0.3524,
      "step": 42002
    },
    {
      "epoch": 1.927538892203203,
      "grad_norm": 0.46744847297668457,
      "learning_rate": 2.939361615254664e-06,
      "loss": 0.3175,
      "step": 42003
    },
    {
      "epoch": 1.9275847827084576,
      "grad_norm": 0.48233115673065186,
      "learning_rate": 2.93913822062974e-06,
      "loss": 0.4167,
      "step": 42004
    },
    {
      "epoch": 1.927630673213712,
      "grad_norm": 0.47306227684020996,
      "learning_rate": 2.9389148309605385e-06,
      "loss": 0.3463,
      "step": 42005
    },
    {
      "epoch": 1.9276765637189666,
      "grad_norm": 0.4628267288208008,
      "learning_rate": 2.9386914462475966e-06,
      "loss": 0.3095,
      "step": 42006
    },
    {
      "epoch": 1.927722454224221,
      "grad_norm": 0.49226096272468567,
      "learning_rate": 2.9384680664914522e-06,
      "loss": 0.3859,
      "step": 42007
    },
    {
      "epoch": 1.9277683447294756,
      "grad_norm": 0.4826112389564514,
      "learning_rate": 2.9382446916926384e-06,
      "loss": 0.403,
      "step": 42008
    },
    {
      "epoch": 1.92781423523473,
      "grad_norm": 0.47766876220703125,
      "learning_rate": 2.9380213218516974e-06,
      "loss": 0.3695,
      "step": 42009
    },
    {
      "epoch": 1.9278601257399846,
      "grad_norm": 0.4814397990703583,
      "learning_rate": 2.937797956969165e-06,
      "loss": 0.3894,
      "step": 42010
    },
    {
      "epoch": 1.9279060162452388,
      "grad_norm": 0.4574735760688782,
      "learning_rate": 2.937574597045576e-06,
      "loss": 0.3496,
      "step": 42011
    },
    {
      "epoch": 1.9279519067504933,
      "grad_norm": 0.4615628719329834,
      "learning_rate": 2.937351242081471e-06,
      "loss": 0.3254,
      "step": 42012
    },
    {
      "epoch": 1.9279977972557478,
      "grad_norm": 0.4787447154521942,
      "learning_rate": 2.9371278920773843e-06,
      "loss": 0.3396,
      "step": 42013
    },
    {
      "epoch": 1.928043687761002,
      "grad_norm": 0.45925071835517883,
      "learning_rate": 2.936904547033852e-06,
      "loss": 0.3463,
      "step": 42014
    },
    {
      "epoch": 1.9280895782662566,
      "grad_norm": 0.4711998701095581,
      "learning_rate": 2.936681206951417e-06,
      "loss": 0.3873,
      "step": 42015
    },
    {
      "epoch": 1.928135468771511,
      "grad_norm": 0.44571447372436523,
      "learning_rate": 2.936457871830611e-06,
      "loss": 0.2821,
      "step": 42016
    },
    {
      "epoch": 1.9281813592767656,
      "grad_norm": 0.4529707133769989,
      "learning_rate": 2.9362345416719705e-06,
      "loss": 0.3361,
      "step": 42017
    },
    {
      "epoch": 1.92822724978202,
      "grad_norm": 0.47458595037460327,
      "learning_rate": 2.9360112164760367e-06,
      "loss": 0.3524,
      "step": 42018
    },
    {
      "epoch": 1.9282731402872746,
      "grad_norm": 0.5064071416854858,
      "learning_rate": 2.9357878962433446e-06,
      "loss": 0.387,
      "step": 42019
    },
    {
      "epoch": 1.928319030792529,
      "grad_norm": 0.4533954858779907,
      "learning_rate": 2.93556458097443e-06,
      "loss": 0.3327,
      "step": 42020
    },
    {
      "epoch": 1.9283649212977836,
      "grad_norm": 0.407736212015152,
      "learning_rate": 2.9353412706698316e-06,
      "loss": 0.231,
      "step": 42021
    },
    {
      "epoch": 1.928410811803038,
      "grad_norm": 0.4988231658935547,
      "learning_rate": 2.9351179653300865e-06,
      "loss": 0.3689,
      "step": 42022
    },
    {
      "epoch": 1.9284567023082926,
      "grad_norm": 0.4432070553302765,
      "learning_rate": 2.9348946649557286e-06,
      "loss": 0.3154,
      "step": 42023
    },
    {
      "epoch": 1.9285025928135469,
      "grad_norm": 0.47020474076271057,
      "learning_rate": 2.9346713695472996e-06,
      "loss": 0.3731,
      "step": 42024
    },
    {
      "epoch": 1.9285484833188014,
      "grad_norm": 0.4891517162322998,
      "learning_rate": 2.934448079105334e-06,
      "loss": 0.4265,
      "step": 42025
    },
    {
      "epoch": 1.9285943738240559,
      "grad_norm": 0.47520044445991516,
      "learning_rate": 2.9342247936303677e-06,
      "loss": 0.37,
      "step": 42026
    },
    {
      "epoch": 1.9286402643293101,
      "grad_norm": 0.5230405330657959,
      "learning_rate": 2.9340015131229393e-06,
      "loss": 0.4124,
      "step": 42027
    },
    {
      "epoch": 1.9286861548345646,
      "grad_norm": 0.4929693937301636,
      "learning_rate": 2.933778237583586e-06,
      "loss": 0.4079,
      "step": 42028
    },
    {
      "epoch": 1.9287320453398191,
      "grad_norm": 0.4549710154533386,
      "learning_rate": 2.9335549670128414e-06,
      "loss": 0.375,
      "step": 42029
    },
    {
      "epoch": 1.9287779358450736,
      "grad_norm": 0.46801477670669556,
      "learning_rate": 2.9333317014112473e-06,
      "loss": 0.3408,
      "step": 42030
    },
    {
      "epoch": 1.9288238263503281,
      "grad_norm": 0.5093133449554443,
      "learning_rate": 2.9331084407793385e-06,
      "loss": 0.4043,
      "step": 42031
    },
    {
      "epoch": 1.9288697168555826,
      "grad_norm": 0.4378516674041748,
      "learning_rate": 2.9328851851176498e-06,
      "loss": 0.3051,
      "step": 42032
    },
    {
      "epoch": 1.9289156073608371,
      "grad_norm": 0.48876699805259705,
      "learning_rate": 2.9326619344267217e-06,
      "loss": 0.3963,
      "step": 42033
    },
    {
      "epoch": 1.9289614978660916,
      "grad_norm": 0.4867330491542816,
      "learning_rate": 2.932438688707089e-06,
      "loss": 0.3718,
      "step": 42034
    },
    {
      "epoch": 1.929007388371346,
      "grad_norm": 0.4951794445514679,
      "learning_rate": 2.932215447959286e-06,
      "loss": 0.3927,
      "step": 42035
    },
    {
      "epoch": 1.9290532788766004,
      "grad_norm": 0.4565521776676178,
      "learning_rate": 2.9319922121838546e-06,
      "loss": 0.3338,
      "step": 42036
    },
    {
      "epoch": 1.9290991693818549,
      "grad_norm": 0.4589291512966156,
      "learning_rate": 2.93176898138133e-06,
      "loss": 0.3593,
      "step": 42037
    },
    {
      "epoch": 1.9291450598871094,
      "grad_norm": 0.5452494025230408,
      "learning_rate": 2.931545755552246e-06,
      "loss": 0.4368,
      "step": 42038
    },
    {
      "epoch": 1.9291909503923637,
      "grad_norm": 0.5167415142059326,
      "learning_rate": 2.931322534697143e-06,
      "loss": 0.3852,
      "step": 42039
    },
    {
      "epoch": 1.9292368408976182,
      "grad_norm": 0.4778722822666168,
      "learning_rate": 2.9310993188165563e-06,
      "loss": 0.3808,
      "step": 42040
    },
    {
      "epoch": 1.9292827314028727,
      "grad_norm": 0.4628783166408539,
      "learning_rate": 2.9308761079110233e-06,
      "loss": 0.3551,
      "step": 42041
    },
    {
      "epoch": 1.9293286219081272,
      "grad_norm": 0.5074269771575928,
      "learning_rate": 2.930652901981079e-06,
      "loss": 0.3506,
      "step": 42042
    },
    {
      "epoch": 1.9293745124133816,
      "grad_norm": 0.5508561730384827,
      "learning_rate": 2.930429701027262e-06,
      "loss": 0.4317,
      "step": 42043
    },
    {
      "epoch": 1.9294204029186361,
      "grad_norm": 0.5162907838821411,
      "learning_rate": 2.9302065050501093e-06,
      "loss": 0.4355,
      "step": 42044
    },
    {
      "epoch": 1.9294662934238906,
      "grad_norm": 0.4954761266708374,
      "learning_rate": 2.929983314050154e-06,
      "loss": 0.3991,
      "step": 42045
    },
    {
      "epoch": 1.9295121839291451,
      "grad_norm": 0.4895554780960083,
      "learning_rate": 2.9297601280279374e-06,
      "loss": 0.3984,
      "step": 42046
    },
    {
      "epoch": 1.9295580744343996,
      "grad_norm": 0.45133498311042786,
      "learning_rate": 2.9295369469839944e-06,
      "loss": 0.3221,
      "step": 42047
    },
    {
      "epoch": 1.9296039649396541,
      "grad_norm": 0.4288400411605835,
      "learning_rate": 2.929313770918861e-06,
      "loss": 0.2783,
      "step": 42048
    },
    {
      "epoch": 1.9296498554449084,
      "grad_norm": 0.4393431842327118,
      "learning_rate": 2.9290905998330755e-06,
      "loss": 0.3051,
      "step": 42049
    },
    {
      "epoch": 1.929695745950163,
      "grad_norm": 0.4939703941345215,
      "learning_rate": 2.9288674337271726e-06,
      "loss": 0.3434,
      "step": 42050
    },
    {
      "epoch": 1.9297416364554174,
      "grad_norm": 0.481050044298172,
      "learning_rate": 2.928644272601689e-06,
      "loss": 0.3526,
      "step": 42051
    },
    {
      "epoch": 1.9297875269606717,
      "grad_norm": 0.4991039037704468,
      "learning_rate": 2.9284211164571637e-06,
      "loss": 0.4258,
      "step": 42052
    },
    {
      "epoch": 1.9298334174659262,
      "grad_norm": 0.4867273271083832,
      "learning_rate": 2.9281979652941317e-06,
      "loss": 0.3689,
      "step": 42053
    },
    {
      "epoch": 1.9298793079711807,
      "grad_norm": 0.5187358856201172,
      "learning_rate": 2.9279748191131284e-06,
      "loss": 0.3522,
      "step": 42054
    },
    {
      "epoch": 1.9299251984764352,
      "grad_norm": 0.428314208984375,
      "learning_rate": 2.927751677914693e-06,
      "loss": 0.2793,
      "step": 42055
    },
    {
      "epoch": 1.9299710889816897,
      "grad_norm": 0.4773090183734894,
      "learning_rate": 2.9275285416993617e-06,
      "loss": 0.3876,
      "step": 42056
    },
    {
      "epoch": 1.9300169794869442,
      "grad_norm": 0.4792625904083252,
      "learning_rate": 2.9273054104676666e-06,
      "loss": 0.3763,
      "step": 42057
    },
    {
      "epoch": 1.9300628699921987,
      "grad_norm": 0.44843921065330505,
      "learning_rate": 2.9270822842201506e-06,
      "loss": 0.3027,
      "step": 42058
    },
    {
      "epoch": 1.9301087604974532,
      "grad_norm": 0.42734670639038086,
      "learning_rate": 2.9268591629573486e-06,
      "loss": 0.2932,
      "step": 42059
    },
    {
      "epoch": 1.9301546510027077,
      "grad_norm": 0.4823648929595947,
      "learning_rate": 2.926636046679793e-06,
      "loss": 0.4101,
      "step": 42060
    },
    {
      "epoch": 1.9302005415079622,
      "grad_norm": 0.4839698374271393,
      "learning_rate": 2.9264129353880254e-06,
      "loss": 0.3662,
      "step": 42061
    },
    {
      "epoch": 1.9302464320132164,
      "grad_norm": 0.4839797019958496,
      "learning_rate": 2.9261898290825797e-06,
      "loss": 0.3816,
      "step": 42062
    },
    {
      "epoch": 1.930292322518471,
      "grad_norm": 0.49361205101013184,
      "learning_rate": 2.9259667277639925e-06,
      "loss": 0.3881,
      "step": 42063
    },
    {
      "epoch": 1.9303382130237254,
      "grad_norm": 0.42431896924972534,
      "learning_rate": 2.9257436314328014e-06,
      "loss": 0.2716,
      "step": 42064
    },
    {
      "epoch": 1.9303841035289797,
      "grad_norm": 0.45804792642593384,
      "learning_rate": 2.9255205400895426e-06,
      "loss": 0.3051,
      "step": 42065
    },
    {
      "epoch": 1.9304299940342342,
      "grad_norm": 0.5154539346694946,
      "learning_rate": 2.9252974537347496e-06,
      "loss": 0.4405,
      "step": 42066
    },
    {
      "epoch": 1.9304758845394887,
      "grad_norm": 0.4379839301109314,
      "learning_rate": 2.9250743723689633e-06,
      "loss": 0.2845,
      "step": 42067
    },
    {
      "epoch": 1.9305217750447432,
      "grad_norm": 0.49869784712791443,
      "learning_rate": 2.924851295992719e-06,
      "loss": 0.4407,
      "step": 42068
    },
    {
      "epoch": 1.9305676655499977,
      "grad_norm": 0.4674595892429352,
      "learning_rate": 2.9246282246065516e-06,
      "loss": 0.369,
      "step": 42069
    },
    {
      "epoch": 1.9306135560552522,
      "grad_norm": 0.5784079432487488,
      "learning_rate": 2.9244051582109982e-06,
      "loss": 0.3863,
      "step": 42070
    },
    {
      "epoch": 1.9306594465605067,
      "grad_norm": 0.49187976121902466,
      "learning_rate": 2.9241820968065958e-06,
      "loss": 0.4153,
      "step": 42071
    },
    {
      "epoch": 1.9307053370657612,
      "grad_norm": 0.43748101592063904,
      "learning_rate": 2.9239590403938787e-06,
      "loss": 0.3271,
      "step": 42072
    },
    {
      "epoch": 1.9307512275710157,
      "grad_norm": 0.44551801681518555,
      "learning_rate": 2.9237359889733867e-06,
      "loss": 0.3017,
      "step": 42073
    },
    {
      "epoch": 1.93079711807627,
      "grad_norm": 0.4959730803966522,
      "learning_rate": 2.923512942545654e-06,
      "loss": 0.3916,
      "step": 42074
    },
    {
      "epoch": 1.9308430085815245,
      "grad_norm": 0.48266884684562683,
      "learning_rate": 2.923289901111218e-06,
      "loss": 0.3724,
      "step": 42075
    },
    {
      "epoch": 1.930888899086779,
      "grad_norm": 0.4773469865322113,
      "learning_rate": 2.923066864670613e-06,
      "loss": 0.4333,
      "step": 42076
    },
    {
      "epoch": 1.9309347895920335,
      "grad_norm": 0.506525993347168,
      "learning_rate": 2.922843833224378e-06,
      "loss": 0.3461,
      "step": 42077
    },
    {
      "epoch": 1.9309806800972877,
      "grad_norm": 0.47661933302879333,
      "learning_rate": 2.9226208067730477e-06,
      "loss": 0.3589,
      "step": 42078
    },
    {
      "epoch": 1.9310265706025422,
      "grad_norm": 0.4811803102493286,
      "learning_rate": 2.9223977853171563e-06,
      "loss": 0.3798,
      "step": 42079
    },
    {
      "epoch": 1.9310724611077967,
      "grad_norm": 0.4394163191318512,
      "learning_rate": 2.922174768857245e-06,
      "loss": 0.3513,
      "step": 42080
    },
    {
      "epoch": 1.9311183516130512,
      "grad_norm": 0.4961719214916229,
      "learning_rate": 2.9219517573938482e-06,
      "loss": 0.4458,
      "step": 42081
    },
    {
      "epoch": 1.9311642421183057,
      "grad_norm": 0.4551577568054199,
      "learning_rate": 2.9217287509274996e-06,
      "loss": 0.3326,
      "step": 42082
    },
    {
      "epoch": 1.9312101326235602,
      "grad_norm": 0.4592211842536926,
      "learning_rate": 2.9215057494587385e-06,
      "loss": 0.3372,
      "step": 42083
    },
    {
      "epoch": 1.9312560231288147,
      "grad_norm": 0.46909746527671814,
      "learning_rate": 2.9212827529881003e-06,
      "loss": 0.3431,
      "step": 42084
    },
    {
      "epoch": 1.9313019136340692,
      "grad_norm": 0.49250999093055725,
      "learning_rate": 2.9210597615161195e-06,
      "loss": 0.3983,
      "step": 42085
    },
    {
      "epoch": 1.9313478041393237,
      "grad_norm": 0.5084668397903442,
      "learning_rate": 2.920836775043335e-06,
      "loss": 0.4409,
      "step": 42086
    },
    {
      "epoch": 1.931393694644578,
      "grad_norm": 0.4608757197856903,
      "learning_rate": 2.9206137935702827e-06,
      "loss": 0.3799,
      "step": 42087
    },
    {
      "epoch": 1.9314395851498325,
      "grad_norm": 0.4868035912513733,
      "learning_rate": 2.920390817097496e-06,
      "loss": 0.381,
      "step": 42088
    },
    {
      "epoch": 1.931485475655087,
      "grad_norm": 0.4886019229888916,
      "learning_rate": 2.9201678456255145e-06,
      "loss": 0.4257,
      "step": 42089
    },
    {
      "epoch": 1.9315313661603413,
      "grad_norm": 0.4507116973400116,
      "learning_rate": 2.9199448791548724e-06,
      "loss": 0.3259,
      "step": 42090
    },
    {
      "epoch": 1.9315772566655958,
      "grad_norm": 0.5034294724464417,
      "learning_rate": 2.919721917686106e-06,
      "loss": 0.4467,
      "step": 42091
    },
    {
      "epoch": 1.9316231471708503,
      "grad_norm": 0.47825857996940613,
      "learning_rate": 2.9194989612197526e-06,
      "loss": 0.4231,
      "step": 42092
    },
    {
      "epoch": 1.9316690376761048,
      "grad_norm": 0.4530491530895233,
      "learning_rate": 2.9192760097563477e-06,
      "loss": 0.3012,
      "step": 42093
    },
    {
      "epoch": 1.9317149281813593,
      "grad_norm": 0.4639139473438263,
      "learning_rate": 2.9190530632964253e-06,
      "loss": 0.3418,
      "step": 42094
    },
    {
      "epoch": 1.9317608186866138,
      "grad_norm": 0.4808740019798279,
      "learning_rate": 2.918830121840526e-06,
      "loss": 0.3815,
      "step": 42095
    },
    {
      "epoch": 1.9318067091918683,
      "grad_norm": 0.5169860124588013,
      "learning_rate": 2.918607185389183e-06,
      "loss": 0.389,
      "step": 42096
    },
    {
      "epoch": 1.9318525996971228,
      "grad_norm": 0.4533408582210541,
      "learning_rate": 2.918384253942931e-06,
      "loss": 0.3058,
      "step": 42097
    },
    {
      "epoch": 1.9318984902023772,
      "grad_norm": 0.9880781173706055,
      "learning_rate": 2.918161327502309e-06,
      "loss": 0.3605,
      "step": 42098
    },
    {
      "epoch": 1.9319443807076317,
      "grad_norm": 0.5079941153526306,
      "learning_rate": 2.9179384060678527e-06,
      "loss": 0.3942,
      "step": 42099
    },
    {
      "epoch": 1.931990271212886,
      "grad_norm": 0.5017287135124207,
      "learning_rate": 2.917715489640097e-06,
      "loss": 0.3507,
      "step": 42100
    },
    {
      "epoch": 1.9320361617181405,
      "grad_norm": 0.47653111815452576,
      "learning_rate": 2.917492578219579e-06,
      "loss": 0.3328,
      "step": 42101
    },
    {
      "epoch": 1.932082052223395,
      "grad_norm": 0.45364201068878174,
      "learning_rate": 2.917269671806834e-06,
      "loss": 0.3674,
      "step": 42102
    },
    {
      "epoch": 1.9321279427286493,
      "grad_norm": 0.4628329575061798,
      "learning_rate": 2.917046770402396e-06,
      "loss": 0.374,
      "step": 42103
    },
    {
      "epoch": 1.9321738332339038,
      "grad_norm": 0.4694858193397522,
      "learning_rate": 2.9168238740068044e-06,
      "loss": 0.3445,
      "step": 42104
    },
    {
      "epoch": 1.9322197237391583,
      "grad_norm": 0.45285314321517944,
      "learning_rate": 2.916600982620595e-06,
      "loss": 0.3015,
      "step": 42105
    },
    {
      "epoch": 1.9322656142444128,
      "grad_norm": 0.4394005537033081,
      "learning_rate": 2.9163780962443024e-06,
      "loss": 0.3378,
      "step": 42106
    },
    {
      "epoch": 1.9323115047496673,
      "grad_norm": 0.4401160180568695,
      "learning_rate": 2.91615521487846e-06,
      "loss": 0.2752,
      "step": 42107
    },
    {
      "epoch": 1.9323573952549218,
      "grad_norm": 0.4532160460948944,
      "learning_rate": 2.915932338523609e-06,
      "loss": 0.3035,
      "step": 42108
    },
    {
      "epoch": 1.9324032857601763,
      "grad_norm": 0.4551438093185425,
      "learning_rate": 2.9157094671802842e-06,
      "loss": 0.3206,
      "step": 42109
    },
    {
      "epoch": 1.9324491762654308,
      "grad_norm": 0.5063810348510742,
      "learning_rate": 2.915486600849019e-06,
      "loss": 0.4034,
      "step": 42110
    },
    {
      "epoch": 1.9324950667706853,
      "grad_norm": 0.46457839012145996,
      "learning_rate": 2.915263739530351e-06,
      "loss": 0.3331,
      "step": 42111
    },
    {
      "epoch": 1.9325409572759396,
      "grad_norm": 0.5064998865127563,
      "learning_rate": 2.9150408832248155e-06,
      "loss": 0.3894,
      "step": 42112
    },
    {
      "epoch": 1.932586847781194,
      "grad_norm": 0.4744535982608795,
      "learning_rate": 2.914818031932948e-06,
      "loss": 0.3337,
      "step": 42113
    },
    {
      "epoch": 1.9326327382864485,
      "grad_norm": 0.42398813366889954,
      "learning_rate": 2.9145951856552857e-06,
      "loss": 0.2739,
      "step": 42114
    },
    {
      "epoch": 1.932678628791703,
      "grad_norm": 0.4621594548225403,
      "learning_rate": 2.914372344392364e-06,
      "loss": 0.3178,
      "step": 42115
    },
    {
      "epoch": 1.9327245192969573,
      "grad_norm": 0.4785335958003998,
      "learning_rate": 2.914149508144716e-06,
      "loss": 0.3504,
      "step": 42116
    },
    {
      "epoch": 1.9327704098022118,
      "grad_norm": 0.4840576648712158,
      "learning_rate": 2.9139266769128834e-06,
      "loss": 0.3824,
      "step": 42117
    },
    {
      "epoch": 1.9328163003074663,
      "grad_norm": 0.5077571868896484,
      "learning_rate": 2.913703850697398e-06,
      "loss": 0.4273,
      "step": 42118
    },
    {
      "epoch": 1.9328621908127208,
      "grad_norm": 0.4889448583126068,
      "learning_rate": 2.9134810294987942e-06,
      "loss": 0.3989,
      "step": 42119
    },
    {
      "epoch": 1.9329080813179753,
      "grad_norm": 0.44574835896492004,
      "learning_rate": 2.9132582133176124e-06,
      "loss": 0.3015,
      "step": 42120
    },
    {
      "epoch": 1.9329539718232298,
      "grad_norm": 0.47418269515037537,
      "learning_rate": 2.913035402154386e-06,
      "loss": 0.3664,
      "step": 42121
    },
    {
      "epoch": 1.9329998623284843,
      "grad_norm": 0.4595847725868225,
      "learning_rate": 2.912812596009651e-06,
      "loss": 0.3344,
      "step": 42122
    },
    {
      "epoch": 1.9330457528337388,
      "grad_norm": 0.5350896716117859,
      "learning_rate": 2.9125897948839423e-06,
      "loss": 0.3602,
      "step": 42123
    },
    {
      "epoch": 1.9330916433389933,
      "grad_norm": 0.45919641852378845,
      "learning_rate": 2.912366998777797e-06,
      "loss": 0.3457,
      "step": 42124
    },
    {
      "epoch": 1.9331375338442476,
      "grad_norm": 0.4299490451812744,
      "learning_rate": 2.9121442076917472e-06,
      "loss": 0.3044,
      "step": 42125
    },
    {
      "epoch": 1.933183424349502,
      "grad_norm": 0.5153296589851379,
      "learning_rate": 2.911921421626335e-06,
      "loss": 0.4377,
      "step": 42126
    },
    {
      "epoch": 1.9332293148547566,
      "grad_norm": 0.47546374797821045,
      "learning_rate": 2.9116986405820918e-06,
      "loss": 0.3638,
      "step": 42127
    },
    {
      "epoch": 1.9332752053600109,
      "grad_norm": 0.4855397641658783,
      "learning_rate": 2.9114758645595526e-06,
      "loss": 0.3811,
      "step": 42128
    },
    {
      "epoch": 1.9333210958652653,
      "grad_norm": 0.4485084116458893,
      "learning_rate": 2.9112530935592564e-06,
      "loss": 0.3402,
      "step": 42129
    },
    {
      "epoch": 1.9333669863705198,
      "grad_norm": 0.45368972420692444,
      "learning_rate": 2.9110303275817377e-06,
      "loss": 0.3126,
      "step": 42130
    },
    {
      "epoch": 1.9334128768757743,
      "grad_norm": 0.49389320611953735,
      "learning_rate": 2.9108075666275323e-06,
      "loss": 0.3814,
      "step": 42131
    },
    {
      "epoch": 1.9334587673810288,
      "grad_norm": 0.4625241756439209,
      "learning_rate": 2.9105848106971745e-06,
      "loss": 0.3067,
      "step": 42132
    },
    {
      "epoch": 1.9335046578862833,
      "grad_norm": 0.48028871417045593,
      "learning_rate": 2.9103620597912006e-06,
      "loss": 0.3749,
      "step": 42133
    },
    {
      "epoch": 1.9335505483915378,
      "grad_norm": 0.5282205939292908,
      "learning_rate": 2.9101393139101455e-06,
      "loss": 0.3922,
      "step": 42134
    },
    {
      "epoch": 1.9335964388967923,
      "grad_norm": 0.4563572108745575,
      "learning_rate": 2.9099165730545465e-06,
      "loss": 0.3453,
      "step": 42135
    },
    {
      "epoch": 1.9336423294020468,
      "grad_norm": 0.4855148494243622,
      "learning_rate": 2.9096938372249396e-06,
      "loss": 0.3414,
      "step": 42136
    },
    {
      "epoch": 1.9336882199073013,
      "grad_norm": 0.45987820625305176,
      "learning_rate": 2.9094711064218566e-06,
      "loss": 0.3296,
      "step": 42137
    },
    {
      "epoch": 1.9337341104125556,
      "grad_norm": 0.5211333632469177,
      "learning_rate": 2.9092483806458383e-06,
      "loss": 0.4557,
      "step": 42138
    },
    {
      "epoch": 1.93378000091781,
      "grad_norm": 0.4933227300643921,
      "learning_rate": 2.9090256598974175e-06,
      "loss": 0.3568,
      "step": 42139
    },
    {
      "epoch": 1.9338258914230646,
      "grad_norm": 0.4733366072177887,
      "learning_rate": 2.90880294417713e-06,
      "loss": 0.4067,
      "step": 42140
    },
    {
      "epoch": 1.9338717819283189,
      "grad_norm": 0.46922239661216736,
      "learning_rate": 2.9085802334855097e-06,
      "loss": 0.3291,
      "step": 42141
    },
    {
      "epoch": 1.9339176724335734,
      "grad_norm": 0.4737309515476227,
      "learning_rate": 2.908357527823095e-06,
      "loss": 0.2983,
      "step": 42142
    },
    {
      "epoch": 1.9339635629388279,
      "grad_norm": 0.5814620852470398,
      "learning_rate": 2.9081348271904214e-06,
      "loss": 0.4987,
      "step": 42143
    },
    {
      "epoch": 1.9340094534440824,
      "grad_norm": 0.4887355864048004,
      "learning_rate": 2.907912131588023e-06,
      "loss": 0.4084,
      "step": 42144
    },
    {
      "epoch": 1.9340553439493369,
      "grad_norm": 0.5126840472221375,
      "learning_rate": 2.9076894410164352e-06,
      "loss": 0.3431,
      "step": 42145
    },
    {
      "epoch": 1.9341012344545914,
      "grad_norm": 0.5245996713638306,
      "learning_rate": 2.9074667554761946e-06,
      "loss": 0.4325,
      "step": 42146
    },
    {
      "epoch": 1.9341471249598459,
      "grad_norm": 0.4735458493232727,
      "learning_rate": 2.9072440749678333e-06,
      "loss": 0.3502,
      "step": 42147
    },
    {
      "epoch": 1.9341930154651004,
      "grad_norm": 0.43617480993270874,
      "learning_rate": 2.907021399491892e-06,
      "loss": 0.3062,
      "step": 42148
    },
    {
      "epoch": 1.9342389059703549,
      "grad_norm": 0.47177228331565857,
      "learning_rate": 2.906798729048904e-06,
      "loss": 0.3343,
      "step": 42149
    },
    {
      "epoch": 1.9342847964756094,
      "grad_norm": 0.45170366764068604,
      "learning_rate": 2.9065760636394014e-06,
      "loss": 0.3282,
      "step": 42150
    },
    {
      "epoch": 1.9343306869808636,
      "grad_norm": 0.4893967807292938,
      "learning_rate": 2.9063534032639256e-06,
      "loss": 0.3518,
      "step": 42151
    },
    {
      "epoch": 1.9343765774861181,
      "grad_norm": 0.4224891662597656,
      "learning_rate": 2.9061307479230087e-06,
      "loss": 0.272,
      "step": 42152
    },
    {
      "epoch": 1.9344224679913726,
      "grad_norm": 0.4226889908313751,
      "learning_rate": 2.9059080976171868e-06,
      "loss": 0.2654,
      "step": 42153
    },
    {
      "epoch": 1.934468358496627,
      "grad_norm": 0.47182178497314453,
      "learning_rate": 2.905685452346995e-06,
      "loss": 0.3401,
      "step": 42154
    },
    {
      "epoch": 1.9345142490018814,
      "grad_norm": 0.46092697978019714,
      "learning_rate": 2.905462812112968e-06,
      "loss": 0.3444,
      "step": 42155
    },
    {
      "epoch": 1.934560139507136,
      "grad_norm": 0.4311964511871338,
      "learning_rate": 2.9052401769156403e-06,
      "loss": 0.2945,
      "step": 42156
    },
    {
      "epoch": 1.9346060300123904,
      "grad_norm": 0.4491490125656128,
      "learning_rate": 2.905017546755551e-06,
      "loss": 0.3244,
      "step": 42157
    },
    {
      "epoch": 1.934651920517645,
      "grad_norm": 0.4752199351787567,
      "learning_rate": 2.904794921633233e-06,
      "loss": 0.3521,
      "step": 42158
    },
    {
      "epoch": 1.9346978110228994,
      "grad_norm": 0.4533317983150482,
      "learning_rate": 2.9045723015492207e-06,
      "loss": 0.3309,
      "step": 42159
    },
    {
      "epoch": 1.934743701528154,
      "grad_norm": 0.4792214334011078,
      "learning_rate": 2.9043496865040523e-06,
      "loss": 0.425,
      "step": 42160
    },
    {
      "epoch": 1.9347895920334084,
      "grad_norm": 0.533702552318573,
      "learning_rate": 2.904127076498261e-06,
      "loss": 0.4237,
      "step": 42161
    },
    {
      "epoch": 1.9348354825386629,
      "grad_norm": 0.4997898042201996,
      "learning_rate": 2.9039044715323806e-06,
      "loss": 0.39,
      "step": 42162
    },
    {
      "epoch": 1.9348813730439172,
      "grad_norm": 0.4483516812324524,
      "learning_rate": 2.9036818716069514e-06,
      "loss": 0.3369,
      "step": 42163
    },
    {
      "epoch": 1.9349272635491717,
      "grad_norm": 0.4723590612411499,
      "learning_rate": 2.9034592767225046e-06,
      "loss": 0.37,
      "step": 42164
    },
    {
      "epoch": 1.9349731540544262,
      "grad_norm": 0.5056209564208984,
      "learning_rate": 2.903236686879577e-06,
      "loss": 0.3748,
      "step": 42165
    },
    {
      "epoch": 1.9350190445596807,
      "grad_norm": 0.4600323736667633,
      "learning_rate": 2.9030141020787034e-06,
      "loss": 0.3644,
      "step": 42166
    },
    {
      "epoch": 1.935064935064935,
      "grad_norm": 0.4863569438457489,
      "learning_rate": 2.9027915223204194e-06,
      "loss": 0.3845,
      "step": 42167
    },
    {
      "epoch": 1.9351108255701894,
      "grad_norm": 0.4348300099372864,
      "learning_rate": 2.9025689476052576e-06,
      "loss": 0.2912,
      "step": 42168
    },
    {
      "epoch": 1.935156716075444,
      "grad_norm": 0.488645076751709,
      "learning_rate": 2.902346377933757e-06,
      "loss": 0.35,
      "step": 42169
    },
    {
      "epoch": 1.9352026065806984,
      "grad_norm": 0.41446781158447266,
      "learning_rate": 2.9021238133064523e-06,
      "loss": 0.2736,
      "step": 42170
    },
    {
      "epoch": 1.935248497085953,
      "grad_norm": 0.43505948781967163,
      "learning_rate": 2.901901253723877e-06,
      "loss": 0.2999,
      "step": 42171
    },
    {
      "epoch": 1.9352943875912074,
      "grad_norm": 0.4395577609539032,
      "learning_rate": 2.9016786991865654e-06,
      "loss": 0.3156,
      "step": 42172
    },
    {
      "epoch": 1.935340278096462,
      "grad_norm": 0.44843775033950806,
      "learning_rate": 2.9014561496950555e-06,
      "loss": 0.3446,
      "step": 42173
    },
    {
      "epoch": 1.9353861686017164,
      "grad_norm": 0.4655916392803192,
      "learning_rate": 2.9012336052498824e-06,
      "loss": 0.2908,
      "step": 42174
    },
    {
      "epoch": 1.935432059106971,
      "grad_norm": 0.45491644740104675,
      "learning_rate": 2.901011065851579e-06,
      "loss": 0.2982,
      "step": 42175
    },
    {
      "epoch": 1.9354779496122252,
      "grad_norm": 0.5015460848808289,
      "learning_rate": 2.9007885315006823e-06,
      "loss": 0.4545,
      "step": 42176
    },
    {
      "epoch": 1.9355238401174797,
      "grad_norm": 0.4443851113319397,
      "learning_rate": 2.9005660021977266e-06,
      "loss": 0.3232,
      "step": 42177
    },
    {
      "epoch": 1.9355697306227342,
      "grad_norm": 0.454862505197525,
      "learning_rate": 2.900343477943245e-06,
      "loss": 0.3256,
      "step": 42178
    },
    {
      "epoch": 1.9356156211279885,
      "grad_norm": 0.4940004348754883,
      "learning_rate": 2.900120958737776e-06,
      "loss": 0.3937,
      "step": 42179
    },
    {
      "epoch": 1.935661511633243,
      "grad_norm": 0.40973758697509766,
      "learning_rate": 2.899898444581854e-06,
      "loss": 0.288,
      "step": 42180
    },
    {
      "epoch": 1.9357074021384975,
      "grad_norm": 0.4419251084327698,
      "learning_rate": 2.8996759354760113e-06,
      "loss": 0.2838,
      "step": 42181
    },
    {
      "epoch": 1.935753292643752,
      "grad_norm": 0.47457218170166016,
      "learning_rate": 2.8994534314207866e-06,
      "loss": 0.369,
      "step": 42182
    },
    {
      "epoch": 1.9357991831490065,
      "grad_norm": 0.46906277537345886,
      "learning_rate": 2.8992309324167145e-06,
      "loss": 0.3306,
      "step": 42183
    },
    {
      "epoch": 1.935845073654261,
      "grad_norm": 0.48183688521385193,
      "learning_rate": 2.8990084384643265e-06,
      "loss": 0.357,
      "step": 42184
    },
    {
      "epoch": 1.9358909641595154,
      "grad_norm": 0.49267420172691345,
      "learning_rate": 2.898785949564162e-06,
      "loss": 0.3492,
      "step": 42185
    },
    {
      "epoch": 1.93593685466477,
      "grad_norm": 0.4676329791545868,
      "learning_rate": 2.8985634657167538e-06,
      "loss": 0.3551,
      "step": 42186
    },
    {
      "epoch": 1.9359827451700244,
      "grad_norm": 0.4613717198371887,
      "learning_rate": 2.8983409869226375e-06,
      "loss": 0.3206,
      "step": 42187
    },
    {
      "epoch": 1.936028635675279,
      "grad_norm": 0.512630820274353,
      "learning_rate": 2.898118513182349e-06,
      "loss": 0.426,
      "step": 42188
    },
    {
      "epoch": 1.9360745261805332,
      "grad_norm": 0.4612993597984314,
      "learning_rate": 2.89789604449642e-06,
      "loss": 0.3273,
      "step": 42189
    },
    {
      "epoch": 1.9361204166857877,
      "grad_norm": 0.421529084444046,
      "learning_rate": 2.8976735808653866e-06,
      "loss": 0.2942,
      "step": 42190
    },
    {
      "epoch": 1.9361663071910422,
      "grad_norm": 0.4599722623825073,
      "learning_rate": 2.897451122289787e-06,
      "loss": 0.3489,
      "step": 42191
    },
    {
      "epoch": 1.9362121976962965,
      "grad_norm": 0.4721481502056122,
      "learning_rate": 2.8972286687701535e-06,
      "loss": 0.368,
      "step": 42192
    },
    {
      "epoch": 1.936258088201551,
      "grad_norm": 0.43672046065330505,
      "learning_rate": 2.8970062203070195e-06,
      "loss": 0.3314,
      "step": 42193
    },
    {
      "epoch": 1.9363039787068055,
      "grad_norm": 0.4383222758769989,
      "learning_rate": 2.8967837769009237e-06,
      "loss": 0.3222,
      "step": 42194
    },
    {
      "epoch": 1.93634986921206,
      "grad_norm": 0.48983079195022583,
      "learning_rate": 2.8965613385523994e-06,
      "loss": 0.4209,
      "step": 42195
    },
    {
      "epoch": 1.9363957597173145,
      "grad_norm": 0.4640178680419922,
      "learning_rate": 2.896338905261981e-06,
      "loss": 0.3223,
      "step": 42196
    },
    {
      "epoch": 1.936441650222569,
      "grad_norm": 0.46103790402412415,
      "learning_rate": 2.8961164770302036e-06,
      "loss": 0.3175,
      "step": 42197
    },
    {
      "epoch": 1.9364875407278235,
      "grad_norm": 0.48561206459999084,
      "learning_rate": 2.895894053857602e-06,
      "loss": 0.3811,
      "step": 42198
    },
    {
      "epoch": 1.936533431233078,
      "grad_norm": 0.5179795622825623,
      "learning_rate": 2.895671635744709e-06,
      "loss": 0.3793,
      "step": 42199
    },
    {
      "epoch": 1.9365793217383325,
      "grad_norm": 0.4862802028656006,
      "learning_rate": 2.895449222692064e-06,
      "loss": 0.3763,
      "step": 42200
    },
    {
      "epoch": 1.9366252122435867,
      "grad_norm": 0.4537147879600525,
      "learning_rate": 2.8952268147001993e-06,
      "loss": 0.3344,
      "step": 42201
    },
    {
      "epoch": 1.9366711027488412,
      "grad_norm": 0.4523382782936096,
      "learning_rate": 2.895004411769648e-06,
      "loss": 0.3027,
      "step": 42202
    },
    {
      "epoch": 1.9367169932540957,
      "grad_norm": 0.4754458963871002,
      "learning_rate": 2.894782013900949e-06,
      "loss": 0.3542,
      "step": 42203
    },
    {
      "epoch": 1.9367628837593502,
      "grad_norm": 0.48246800899505615,
      "learning_rate": 2.8945596210946336e-06,
      "loss": 0.4005,
      "step": 42204
    },
    {
      "epoch": 1.9368087742646045,
      "grad_norm": 0.4990043342113495,
      "learning_rate": 2.8943372333512386e-06,
      "loss": 0.3951,
      "step": 42205
    },
    {
      "epoch": 1.936854664769859,
      "grad_norm": 0.4921986162662506,
      "learning_rate": 2.894114850671296e-06,
      "loss": 0.3757,
      "step": 42206
    },
    {
      "epoch": 1.9369005552751135,
      "grad_norm": 0.5044657588005066,
      "learning_rate": 2.8938924730553443e-06,
      "loss": 0.4283,
      "step": 42207
    },
    {
      "epoch": 1.936946445780368,
      "grad_norm": 0.4628879427909851,
      "learning_rate": 2.893670100503917e-06,
      "loss": 0.3071,
      "step": 42208
    },
    {
      "epoch": 1.9369923362856225,
      "grad_norm": 0.5120128989219666,
      "learning_rate": 2.8934477330175475e-06,
      "loss": 0.3984,
      "step": 42209
    },
    {
      "epoch": 1.937038226790877,
      "grad_norm": 0.4499504268169403,
      "learning_rate": 2.8932253705967716e-06,
      "loss": 0.3186,
      "step": 42210
    },
    {
      "epoch": 1.9370841172961315,
      "grad_norm": 0.5084688663482666,
      "learning_rate": 2.8930030132421243e-06,
      "loss": 0.4366,
      "step": 42211
    },
    {
      "epoch": 1.937130007801386,
      "grad_norm": 0.4980149567127228,
      "learning_rate": 2.8927806609541375e-06,
      "loss": 0.4157,
      "step": 42212
    },
    {
      "epoch": 1.9371758983066405,
      "grad_norm": 0.5168014168739319,
      "learning_rate": 2.89255831373335e-06,
      "loss": 0.4309,
      "step": 42213
    },
    {
      "epoch": 1.9372217888118948,
      "grad_norm": 0.46546173095703125,
      "learning_rate": 2.8923359715802946e-06,
      "loss": 0.3502,
      "step": 42214
    },
    {
      "epoch": 1.9372676793171493,
      "grad_norm": 0.47357630729675293,
      "learning_rate": 2.892113634495505e-06,
      "loss": 0.353,
      "step": 42215
    },
    {
      "epoch": 1.9373135698224038,
      "grad_norm": 0.4673360288143158,
      "learning_rate": 2.891891302479517e-06,
      "loss": 0.3532,
      "step": 42216
    },
    {
      "epoch": 1.937359460327658,
      "grad_norm": 0.46716147661209106,
      "learning_rate": 2.8916689755328663e-06,
      "loss": 0.3076,
      "step": 42217
    },
    {
      "epoch": 1.9374053508329125,
      "grad_norm": 0.47978776693344116,
      "learning_rate": 2.891446653656086e-06,
      "loss": 0.3718,
      "step": 42218
    },
    {
      "epoch": 1.937451241338167,
      "grad_norm": 0.4397413432598114,
      "learning_rate": 2.891224336849711e-06,
      "loss": 0.3502,
      "step": 42219
    },
    {
      "epoch": 1.9374971318434215,
      "grad_norm": 0.47159865498542786,
      "learning_rate": 2.891002025114276e-06,
      "loss": 0.3427,
      "step": 42220
    },
    {
      "epoch": 1.937543022348676,
      "grad_norm": 0.4388810396194458,
      "learning_rate": 2.890779718450314e-06,
      "loss": 0.3008,
      "step": 42221
    },
    {
      "epoch": 1.9375889128539305,
      "grad_norm": 0.4507732391357422,
      "learning_rate": 2.890557416858363e-06,
      "loss": 0.3549,
      "step": 42222
    },
    {
      "epoch": 1.937634803359185,
      "grad_norm": 0.47572746872901917,
      "learning_rate": 2.8903351203389546e-06,
      "loss": 0.3611,
      "step": 42223
    },
    {
      "epoch": 1.9376806938644395,
      "grad_norm": 0.49364981055259705,
      "learning_rate": 2.890112828892623e-06,
      "loss": 0.3834,
      "step": 42224
    },
    {
      "epoch": 1.937726584369694,
      "grad_norm": 0.5216743350028992,
      "learning_rate": 2.8898905425199066e-06,
      "loss": 0.3894,
      "step": 42225
    },
    {
      "epoch": 1.9377724748749485,
      "grad_norm": 0.4627477526664734,
      "learning_rate": 2.8896682612213365e-06,
      "loss": 0.3594,
      "step": 42226
    },
    {
      "epoch": 1.9378183653802028,
      "grad_norm": 0.42021045088768005,
      "learning_rate": 2.889445984997447e-06,
      "loss": 0.27,
      "step": 42227
    },
    {
      "epoch": 1.9378642558854573,
      "grad_norm": 0.43709009885787964,
      "learning_rate": 2.8892237138487755e-06,
      "loss": 0.2957,
      "step": 42228
    },
    {
      "epoch": 1.9379101463907118,
      "grad_norm": 0.5263662934303284,
      "learning_rate": 2.8890014477758547e-06,
      "loss": 0.4336,
      "step": 42229
    },
    {
      "epoch": 1.937956036895966,
      "grad_norm": 0.43518704175949097,
      "learning_rate": 2.8887791867792198e-06,
      "loss": 0.2909,
      "step": 42230
    },
    {
      "epoch": 1.9380019274012206,
      "grad_norm": 0.4150220453739166,
      "learning_rate": 2.8885569308594043e-06,
      "loss": 0.2697,
      "step": 42231
    },
    {
      "epoch": 1.938047817906475,
      "grad_norm": 0.46325165033340454,
      "learning_rate": 2.8883346800169427e-06,
      "loss": 0.38,
      "step": 42232
    },
    {
      "epoch": 1.9380937084117296,
      "grad_norm": 0.4877205193042755,
      "learning_rate": 2.888112434252368e-06,
      "loss": 0.3826,
      "step": 42233
    },
    {
      "epoch": 1.938139598916984,
      "grad_norm": 0.4559682309627533,
      "learning_rate": 2.887890193566218e-06,
      "loss": 0.3603,
      "step": 42234
    },
    {
      "epoch": 1.9381854894222386,
      "grad_norm": 0.4480857849121094,
      "learning_rate": 2.887667957959026e-06,
      "loss": 0.3135,
      "step": 42235
    },
    {
      "epoch": 1.938231379927493,
      "grad_norm": 0.4790278673171997,
      "learning_rate": 2.887445727431325e-06,
      "loss": 0.3739,
      "step": 42236
    },
    {
      "epoch": 1.9382772704327476,
      "grad_norm": 0.49222835898399353,
      "learning_rate": 2.8872235019836493e-06,
      "loss": 0.3651,
      "step": 42237
    },
    {
      "epoch": 1.938323160938002,
      "grad_norm": 0.48999857902526855,
      "learning_rate": 2.887001281616536e-06,
      "loss": 0.3783,
      "step": 42238
    },
    {
      "epoch": 1.9383690514432566,
      "grad_norm": 0.47584909200668335,
      "learning_rate": 2.8867790663305175e-06,
      "loss": 0.3516,
      "step": 42239
    },
    {
      "epoch": 1.9384149419485108,
      "grad_norm": 0.4508305788040161,
      "learning_rate": 2.886556856126128e-06,
      "loss": 0.3236,
      "step": 42240
    },
    {
      "epoch": 1.9384608324537653,
      "grad_norm": 0.48666274547576904,
      "learning_rate": 2.886334651003903e-06,
      "loss": 0.3935,
      "step": 42241
    },
    {
      "epoch": 1.9385067229590198,
      "grad_norm": 0.4965948462486267,
      "learning_rate": 2.8861124509643756e-06,
      "loss": 0.3818,
      "step": 42242
    },
    {
      "epoch": 1.938552613464274,
      "grad_norm": 0.5081051588058472,
      "learning_rate": 2.8858902560080794e-06,
      "loss": 0.3848,
      "step": 42243
    },
    {
      "epoch": 1.9385985039695286,
      "grad_norm": 0.46162158250808716,
      "learning_rate": 2.885668066135551e-06,
      "loss": 0.3491,
      "step": 42244
    },
    {
      "epoch": 1.938644394474783,
      "grad_norm": 0.5153376460075378,
      "learning_rate": 2.885445881347324e-06,
      "loss": 0.4676,
      "step": 42245
    },
    {
      "epoch": 1.9386902849800376,
      "grad_norm": 0.6601130962371826,
      "learning_rate": 2.88522370164393e-06,
      "loss": 0.3797,
      "step": 42246
    },
    {
      "epoch": 1.938736175485292,
      "grad_norm": 0.5284818410873413,
      "learning_rate": 2.885001527025908e-06,
      "loss": 0.3569,
      "step": 42247
    },
    {
      "epoch": 1.9387820659905466,
      "grad_norm": 0.4753550887107849,
      "learning_rate": 2.88477935749379e-06,
      "loss": 0.3459,
      "step": 42248
    },
    {
      "epoch": 1.938827956495801,
      "grad_norm": 0.4834121763706207,
      "learning_rate": 2.8845571930481076e-06,
      "loss": 0.3823,
      "step": 42249
    },
    {
      "epoch": 1.9388738470010556,
      "grad_norm": 0.4619996249675751,
      "learning_rate": 2.8843350336893993e-06,
      "loss": 0.317,
      "step": 42250
    },
    {
      "epoch": 1.93891973750631,
      "grad_norm": 0.4547794461250305,
      "learning_rate": 2.884112879418197e-06,
      "loss": 0.3214,
      "step": 42251
    },
    {
      "epoch": 1.9389656280115644,
      "grad_norm": 0.41677114367485046,
      "learning_rate": 2.8838907302350365e-06,
      "loss": 0.261,
      "step": 42252
    },
    {
      "epoch": 1.9390115185168189,
      "grad_norm": 0.5085619688034058,
      "learning_rate": 2.8836685861404514e-06,
      "loss": 0.4129,
      "step": 42253
    },
    {
      "epoch": 1.9390574090220734,
      "grad_norm": 0.43450894951820374,
      "learning_rate": 2.8834464471349745e-06,
      "loss": 0.2874,
      "step": 42254
    },
    {
      "epoch": 1.9391032995273279,
      "grad_norm": 0.48946666717529297,
      "learning_rate": 2.8832243132191386e-06,
      "loss": 0.4237,
      "step": 42255
    },
    {
      "epoch": 1.9391491900325821,
      "grad_norm": 0.43609315156936646,
      "learning_rate": 2.8830021843934824e-06,
      "loss": 0.31,
      "step": 42256
    },
    {
      "epoch": 1.9391950805378366,
      "grad_norm": 0.4435378313064575,
      "learning_rate": 2.8827800606585375e-06,
      "loss": 0.3235,
      "step": 42257
    },
    {
      "epoch": 1.9392409710430911,
      "grad_norm": 0.5002389550209045,
      "learning_rate": 2.8825579420148373e-06,
      "loss": 0.353,
      "step": 42258
    },
    {
      "epoch": 1.9392868615483456,
      "grad_norm": 0.423209547996521,
      "learning_rate": 2.8823358284629186e-06,
      "loss": 0.3048,
      "step": 42259
    },
    {
      "epoch": 1.9393327520536001,
      "grad_norm": 0.4849652051925659,
      "learning_rate": 2.8821137200033133e-06,
      "loss": 0.3664,
      "step": 42260
    },
    {
      "epoch": 1.9393786425588546,
      "grad_norm": 0.40150266885757446,
      "learning_rate": 2.8818916166365567e-06,
      "loss": 0.2702,
      "step": 42261
    },
    {
      "epoch": 1.9394245330641091,
      "grad_norm": 0.45437896251678467,
      "learning_rate": 2.8816695183631828e-06,
      "loss": 0.3259,
      "step": 42262
    },
    {
      "epoch": 1.9394704235693636,
      "grad_norm": 0.47784993052482605,
      "learning_rate": 2.8814474251837245e-06,
      "loss": 0.4095,
      "step": 42263
    },
    {
      "epoch": 1.939516314074618,
      "grad_norm": 0.46958816051483154,
      "learning_rate": 2.881225337098714e-06,
      "loss": 0.3461,
      "step": 42264
    },
    {
      "epoch": 1.9395622045798724,
      "grad_norm": 0.48354607820510864,
      "learning_rate": 2.88100325410869e-06,
      "loss": 0.3589,
      "step": 42265
    },
    {
      "epoch": 1.9396080950851269,
      "grad_norm": 0.46869584918022156,
      "learning_rate": 2.8807811762141845e-06,
      "loss": 0.3217,
      "step": 42266
    },
    {
      "epoch": 1.9396539855903814,
      "grad_norm": 0.4924481213092804,
      "learning_rate": 2.8805591034157297e-06,
      "loss": 0.3991,
      "step": 42267
    },
    {
      "epoch": 1.9396998760956357,
      "grad_norm": 0.44264158606529236,
      "learning_rate": 2.8803370357138637e-06,
      "loss": 0.3226,
      "step": 42268
    },
    {
      "epoch": 1.9397457666008902,
      "grad_norm": 0.47411057353019714,
      "learning_rate": 2.8801149731091175e-06,
      "loss": 0.317,
      "step": 42269
    },
    {
      "epoch": 1.9397916571061447,
      "grad_norm": 0.4653431177139282,
      "learning_rate": 2.879892915602026e-06,
      "loss": 0.3211,
      "step": 42270
    },
    {
      "epoch": 1.9398375476113991,
      "grad_norm": 0.5193920135498047,
      "learning_rate": 2.8796708631931214e-06,
      "loss": 0.3977,
      "step": 42271
    },
    {
      "epoch": 1.9398834381166536,
      "grad_norm": 0.44486159086227417,
      "learning_rate": 2.8794488158829404e-06,
      "loss": 0.3235,
      "step": 42272
    },
    {
      "epoch": 1.9399293286219081,
      "grad_norm": 0.45401468873023987,
      "learning_rate": 2.879226773672016e-06,
      "loss": 0.3565,
      "step": 42273
    },
    {
      "epoch": 1.9399752191271626,
      "grad_norm": 0.47701025009155273,
      "learning_rate": 2.879004736560882e-06,
      "loss": 0.3566,
      "step": 42274
    },
    {
      "epoch": 1.9400211096324171,
      "grad_norm": 0.4512176811695099,
      "learning_rate": 2.8787827045500715e-06,
      "loss": 0.3301,
      "step": 42275
    },
    {
      "epoch": 1.9400670001376716,
      "grad_norm": 0.47241300344467163,
      "learning_rate": 2.87856067764012e-06,
      "loss": 0.3312,
      "step": 42276
    },
    {
      "epoch": 1.9401128906429261,
      "grad_norm": 0.4834900200366974,
      "learning_rate": 2.8783386558315583e-06,
      "loss": 0.3443,
      "step": 42277
    },
    {
      "epoch": 1.9401587811481804,
      "grad_norm": 0.45314115285873413,
      "learning_rate": 2.878116639124924e-06,
      "loss": 0.3138,
      "step": 42278
    },
    {
      "epoch": 1.940204671653435,
      "grad_norm": 0.49480968713760376,
      "learning_rate": 2.8778946275207505e-06,
      "loss": 0.3862,
      "step": 42279
    },
    {
      "epoch": 1.9402505621586894,
      "grad_norm": 0.46920135617256165,
      "learning_rate": 2.877672621019568e-06,
      "loss": 0.3603,
      "step": 42280
    },
    {
      "epoch": 1.9402964526639437,
      "grad_norm": 0.43961387872695923,
      "learning_rate": 2.877450619621915e-06,
      "loss": 0.2966,
      "step": 42281
    },
    {
      "epoch": 1.9403423431691982,
      "grad_norm": 0.45799311995506287,
      "learning_rate": 2.877228623328323e-06,
      "loss": 0.3085,
      "step": 42282
    },
    {
      "epoch": 1.9403882336744527,
      "grad_norm": 0.4498348534107208,
      "learning_rate": 2.877006632139327e-06,
      "loss": 0.3467,
      "step": 42283
    },
    {
      "epoch": 1.9404341241797072,
      "grad_norm": 0.48230624198913574,
      "learning_rate": 2.876784646055459e-06,
      "loss": 0.377,
      "step": 42284
    },
    {
      "epoch": 1.9404800146849617,
      "grad_norm": 0.46398574113845825,
      "learning_rate": 2.876562665077255e-06,
      "loss": 0.3226,
      "step": 42285
    },
    {
      "epoch": 1.9405259051902162,
      "grad_norm": 0.47819042205810547,
      "learning_rate": 2.8763406892052454e-06,
      "loss": 0.3826,
      "step": 42286
    },
    {
      "epoch": 1.9405717956954707,
      "grad_norm": 0.47676002979278564,
      "learning_rate": 2.8761187184399673e-06,
      "loss": 0.3439,
      "step": 42287
    },
    {
      "epoch": 1.9406176862007252,
      "grad_norm": 0.4551865756511688,
      "learning_rate": 2.875896752781955e-06,
      "loss": 0.3104,
      "step": 42288
    },
    {
      "epoch": 1.9406635767059797,
      "grad_norm": 0.4709368646144867,
      "learning_rate": 2.8756747922317375e-06,
      "loss": 0.3958,
      "step": 42289
    },
    {
      "epoch": 1.940709467211234,
      "grad_norm": 0.44992244243621826,
      "learning_rate": 2.875452836789854e-06,
      "loss": 0.3333,
      "step": 42290
    },
    {
      "epoch": 1.9407553577164884,
      "grad_norm": 0.497011661529541,
      "learning_rate": 2.875230886456835e-06,
      "loss": 0.442,
      "step": 42291
    },
    {
      "epoch": 1.940801248221743,
      "grad_norm": 0.4552088677883148,
      "learning_rate": 2.8750089412332137e-06,
      "loss": 0.3703,
      "step": 42292
    },
    {
      "epoch": 1.9408471387269974,
      "grad_norm": 0.5278934240341187,
      "learning_rate": 2.874787001119528e-06,
      "loss": 0.4657,
      "step": 42293
    },
    {
      "epoch": 1.9408930292322517,
      "grad_norm": 0.4521971344947815,
      "learning_rate": 2.874565066116308e-06,
      "loss": 0.3412,
      "step": 42294
    },
    {
      "epoch": 1.9409389197375062,
      "grad_norm": 0.463529497385025,
      "learning_rate": 2.874343136224088e-06,
      "loss": 0.343,
      "step": 42295
    },
    {
      "epoch": 1.9409848102427607,
      "grad_norm": 0.4611636996269226,
      "learning_rate": 2.8741212114434024e-06,
      "loss": 0.3536,
      "step": 42296
    },
    {
      "epoch": 1.9410307007480152,
      "grad_norm": 0.45311644673347473,
      "learning_rate": 2.873899291774784e-06,
      "loss": 0.3092,
      "step": 42297
    },
    {
      "epoch": 1.9410765912532697,
      "grad_norm": 0.536088764667511,
      "learning_rate": 2.873677377218765e-06,
      "loss": 0.4206,
      "step": 42298
    },
    {
      "epoch": 1.9411224817585242,
      "grad_norm": 0.46567708253860474,
      "learning_rate": 2.873455467775883e-06,
      "loss": 0.3675,
      "step": 42299
    },
    {
      "epoch": 1.9411683722637787,
      "grad_norm": 0.4789806604385376,
      "learning_rate": 2.873233563446669e-06,
      "loss": 0.4028,
      "step": 42300
    },
    {
      "epoch": 1.9412142627690332,
      "grad_norm": 0.4266676604747772,
      "learning_rate": 2.8730116642316553e-06,
      "loss": 0.2885,
      "step": 42301
    },
    {
      "epoch": 1.9412601532742877,
      "grad_norm": 0.4976128935813904,
      "learning_rate": 2.872789770131379e-06,
      "loss": 0.3864,
      "step": 42302
    },
    {
      "epoch": 1.941306043779542,
      "grad_norm": 0.48775428533554077,
      "learning_rate": 2.872567881146372e-06,
      "loss": 0.3573,
      "step": 42303
    },
    {
      "epoch": 1.9413519342847965,
      "grad_norm": 0.48267868161201477,
      "learning_rate": 2.872345997277168e-06,
      "loss": 0.4549,
      "step": 42304
    },
    {
      "epoch": 1.941397824790051,
      "grad_norm": 0.4972201883792877,
      "learning_rate": 2.872124118524301e-06,
      "loss": 0.4429,
      "step": 42305
    },
    {
      "epoch": 1.9414437152953052,
      "grad_norm": 0.4835011065006256,
      "learning_rate": 2.8719022448883026e-06,
      "loss": 0.4098,
      "step": 42306
    },
    {
      "epoch": 1.9414896058005597,
      "grad_norm": 0.43651050329208374,
      "learning_rate": 2.8716803763697086e-06,
      "loss": 0.297,
      "step": 42307
    },
    {
      "epoch": 1.9415354963058142,
      "grad_norm": 0.4938250482082367,
      "learning_rate": 2.871458512969049e-06,
      "loss": 0.4041,
      "step": 42308
    },
    {
      "epoch": 1.9415813868110687,
      "grad_norm": 0.4971288740634918,
      "learning_rate": 2.8712366546868627e-06,
      "loss": 0.4305,
      "step": 42309
    },
    {
      "epoch": 1.9416272773163232,
      "grad_norm": 0.4850749969482422,
      "learning_rate": 2.8710148015236793e-06,
      "loss": 0.3926,
      "step": 42310
    },
    {
      "epoch": 1.9416731678215777,
      "grad_norm": 0.4991370439529419,
      "learning_rate": 2.870792953480032e-06,
      "loss": 0.3984,
      "step": 42311
    },
    {
      "epoch": 1.9417190583268322,
      "grad_norm": 0.5606815218925476,
      "learning_rate": 2.8705711105564584e-06,
      "loss": 0.477,
      "step": 42312
    },
    {
      "epoch": 1.9417649488320867,
      "grad_norm": 0.48161765933036804,
      "learning_rate": 2.8703492727534878e-06,
      "loss": 0.3902,
      "step": 42313
    },
    {
      "epoch": 1.9418108393373412,
      "grad_norm": 0.444158136844635,
      "learning_rate": 2.8701274400716538e-06,
      "loss": 0.3331,
      "step": 42314
    },
    {
      "epoch": 1.9418567298425957,
      "grad_norm": 0.42736807465553284,
      "learning_rate": 2.8699056125114934e-06,
      "loss": 0.2705,
      "step": 42315
    },
    {
      "epoch": 1.94190262034785,
      "grad_norm": 0.46261027455329895,
      "learning_rate": 2.869683790073537e-06,
      "loss": 0.3235,
      "step": 42316
    },
    {
      "epoch": 1.9419485108531045,
      "grad_norm": 0.4624198377132416,
      "learning_rate": 2.869461972758319e-06,
      "loss": 0.3406,
      "step": 42317
    },
    {
      "epoch": 1.941994401358359,
      "grad_norm": 0.44138920307159424,
      "learning_rate": 2.8692401605663724e-06,
      "loss": 0.2729,
      "step": 42318
    },
    {
      "epoch": 1.9420402918636133,
      "grad_norm": 0.4355107545852661,
      "learning_rate": 2.8690183534982307e-06,
      "loss": 0.3389,
      "step": 42319
    },
    {
      "epoch": 1.9420861823688678,
      "grad_norm": 0.4525952935218811,
      "learning_rate": 2.8687965515544257e-06,
      "loss": 0.3484,
      "step": 42320
    },
    {
      "epoch": 1.9421320728741223,
      "grad_norm": 0.4638443887233734,
      "learning_rate": 2.868574754735494e-06,
      "loss": 0.3044,
      "step": 42321
    },
    {
      "epoch": 1.9421779633793768,
      "grad_norm": 0.45758339762687683,
      "learning_rate": 2.8683529630419676e-06,
      "loss": 0.2869,
      "step": 42322
    },
    {
      "epoch": 1.9422238538846313,
      "grad_norm": 0.4716130197048187,
      "learning_rate": 2.8681311764743775e-06,
      "loss": 0.342,
      "step": 42323
    },
    {
      "epoch": 1.9422697443898858,
      "grad_norm": 0.5328834056854248,
      "learning_rate": 2.867909395033261e-06,
      "loss": 0.4426,
      "step": 42324
    },
    {
      "epoch": 1.9423156348951403,
      "grad_norm": 0.48337554931640625,
      "learning_rate": 2.8676876187191495e-06,
      "loss": 0.355,
      "step": 42325
    },
    {
      "epoch": 1.9423615254003948,
      "grad_norm": 0.4759368896484375,
      "learning_rate": 2.8674658475325745e-06,
      "loss": 0.3444,
      "step": 42326
    },
    {
      "epoch": 1.9424074159056492,
      "grad_norm": 0.46888524293899536,
      "learning_rate": 2.867244081474075e-06,
      "loss": 0.3444,
      "step": 42327
    },
    {
      "epoch": 1.9424533064109037,
      "grad_norm": 0.4886777997016907,
      "learning_rate": 2.8670223205441783e-06,
      "loss": 0.3675,
      "step": 42328
    },
    {
      "epoch": 1.942499196916158,
      "grad_norm": 0.44057416915893555,
      "learning_rate": 2.8668005647434173e-06,
      "loss": 0.3245,
      "step": 42329
    },
    {
      "epoch": 1.9425450874214125,
      "grad_norm": 0.4584789276123047,
      "learning_rate": 2.8665788140723305e-06,
      "loss": 0.3512,
      "step": 42330
    },
    {
      "epoch": 1.942590977926667,
      "grad_norm": 0.4837767481803894,
      "learning_rate": 2.866357068531448e-06,
      "loss": 0.3538,
      "step": 42331
    },
    {
      "epoch": 1.9426368684319213,
      "grad_norm": 0.4615592360496521,
      "learning_rate": 2.8661353281213013e-06,
      "loss": 0.3466,
      "step": 42332
    },
    {
      "epoch": 1.9426827589371758,
      "grad_norm": 0.4658503532409668,
      "learning_rate": 2.8659135928424273e-06,
      "loss": 0.3485,
      "step": 42333
    },
    {
      "epoch": 1.9427286494424303,
      "grad_norm": 0.4668123126029968,
      "learning_rate": 2.865691862695358e-06,
      "loss": 0.3507,
      "step": 42334
    },
    {
      "epoch": 1.9427745399476848,
      "grad_norm": 0.45850706100463867,
      "learning_rate": 2.865470137680626e-06,
      "loss": 0.3667,
      "step": 42335
    },
    {
      "epoch": 1.9428204304529393,
      "grad_norm": 0.48505502939224243,
      "learning_rate": 2.8652484177987625e-06,
      "loss": 0.334,
      "step": 42336
    },
    {
      "epoch": 1.9428663209581938,
      "grad_norm": 0.46346062421798706,
      "learning_rate": 2.8650267030503054e-06,
      "loss": 0.3583,
      "step": 42337
    },
    {
      "epoch": 1.9429122114634483,
      "grad_norm": 0.4359017312526703,
      "learning_rate": 2.8648049934357848e-06,
      "loss": 0.2702,
      "step": 42338
    },
    {
      "epoch": 1.9429581019687028,
      "grad_norm": 0.4613456428050995,
      "learning_rate": 2.8645832889557344e-06,
      "loss": 0.361,
      "step": 42339
    },
    {
      "epoch": 1.9430039924739573,
      "grad_norm": 0.514443039894104,
      "learning_rate": 2.864361589610687e-06,
      "loss": 0.4548,
      "step": 42340
    },
    {
      "epoch": 1.9430498829792116,
      "grad_norm": 0.461169570684433,
      "learning_rate": 2.864139895401176e-06,
      "loss": 0.3753,
      "step": 42341
    },
    {
      "epoch": 1.943095773484466,
      "grad_norm": 0.5346269607543945,
      "learning_rate": 2.8639182063277327e-06,
      "loss": 0.3142,
      "step": 42342
    },
    {
      "epoch": 1.9431416639897205,
      "grad_norm": 0.4384450614452362,
      "learning_rate": 2.863696522390894e-06,
      "loss": 0.2973,
      "step": 42343
    },
    {
      "epoch": 1.943187554494975,
      "grad_norm": 0.47538575530052185,
      "learning_rate": 2.8634748435911906e-06,
      "loss": 0.413,
      "step": 42344
    },
    {
      "epoch": 1.9432334450002293,
      "grad_norm": 0.45169275999069214,
      "learning_rate": 2.8632531699291545e-06,
      "loss": 0.3447,
      "step": 42345
    },
    {
      "epoch": 1.9432793355054838,
      "grad_norm": 0.40608957409858704,
      "learning_rate": 2.8630315014053215e-06,
      "loss": 0.2677,
      "step": 42346
    },
    {
      "epoch": 1.9433252260107383,
      "grad_norm": 0.45585164427757263,
      "learning_rate": 2.862809838020224e-06,
      "loss": 0.3561,
      "step": 42347
    },
    {
      "epoch": 1.9433711165159928,
      "grad_norm": 0.4664261043071747,
      "learning_rate": 2.8625881797743913e-06,
      "loss": 0.3474,
      "step": 42348
    },
    {
      "epoch": 1.9434170070212473,
      "grad_norm": 0.5196836590766907,
      "learning_rate": 2.8623665266683643e-06,
      "loss": 0.4372,
      "step": 42349
    },
    {
      "epoch": 1.9434628975265018,
      "grad_norm": 0.4719387888908386,
      "learning_rate": 2.862144878702669e-06,
      "loss": 0.3728,
      "step": 42350
    },
    {
      "epoch": 1.9435087880317563,
      "grad_norm": 0.44653886556625366,
      "learning_rate": 2.861923235877838e-06,
      "loss": 0.3161,
      "step": 42351
    },
    {
      "epoch": 1.9435546785370108,
      "grad_norm": 0.4639488160610199,
      "learning_rate": 2.8617015981944094e-06,
      "loss": 0.3667,
      "step": 42352
    },
    {
      "epoch": 1.9436005690422653,
      "grad_norm": 0.5011780261993408,
      "learning_rate": 2.8614799656529134e-06,
      "loss": 0.437,
      "step": 42353
    },
    {
      "epoch": 1.9436464595475196,
      "grad_norm": 0.450440913438797,
      "learning_rate": 2.861258338253881e-06,
      "loss": 0.3125,
      "step": 42354
    },
    {
      "epoch": 1.943692350052774,
      "grad_norm": 0.5109073519706726,
      "learning_rate": 2.86103671599785e-06,
      "loss": 0.4073,
      "step": 42355
    },
    {
      "epoch": 1.9437382405580286,
      "grad_norm": 0.4427860975265503,
      "learning_rate": 2.86081509888535e-06,
      "loss": 0.3189,
      "step": 42356
    },
    {
      "epoch": 1.9437841310632828,
      "grad_norm": 0.5144197940826416,
      "learning_rate": 2.8605934869169124e-06,
      "loss": 0.4971,
      "step": 42357
    },
    {
      "epoch": 1.9438300215685373,
      "grad_norm": 0.4542863070964813,
      "learning_rate": 2.8603718800930748e-06,
      "loss": 0.3344,
      "step": 42358
    },
    {
      "epoch": 1.9438759120737918,
      "grad_norm": 0.4256596565246582,
      "learning_rate": 2.8601502784143664e-06,
      "loss": 0.2804,
      "step": 42359
    },
    {
      "epoch": 1.9439218025790463,
      "grad_norm": 0.46273332834243774,
      "learning_rate": 2.859928681881322e-06,
      "loss": 0.3951,
      "step": 42360
    },
    {
      "epoch": 1.9439676930843008,
      "grad_norm": 0.4489917457103729,
      "learning_rate": 2.859707090494474e-06,
      "loss": 0.3265,
      "step": 42361
    },
    {
      "epoch": 1.9440135835895553,
      "grad_norm": 0.4978906512260437,
      "learning_rate": 2.8594855042543545e-06,
      "loss": 0.3482,
      "step": 42362
    },
    {
      "epoch": 1.9440594740948098,
      "grad_norm": 0.42999300360679626,
      "learning_rate": 2.859263923161495e-06,
      "loss": 0.2905,
      "step": 42363
    },
    {
      "epoch": 1.9441053646000643,
      "grad_norm": 0.4374128580093384,
      "learning_rate": 2.859042347216432e-06,
      "loss": 0.2781,
      "step": 42364
    },
    {
      "epoch": 1.9441512551053188,
      "grad_norm": 0.5311461091041565,
      "learning_rate": 2.858820776419697e-06,
      "loss": 0.4248,
      "step": 42365
    },
    {
      "epoch": 1.9441971456105733,
      "grad_norm": 0.4619891047477722,
      "learning_rate": 2.8585992107718197e-06,
      "loss": 0.3083,
      "step": 42366
    },
    {
      "epoch": 1.9442430361158276,
      "grad_norm": 0.4418337941169739,
      "learning_rate": 2.858377650273337e-06,
      "loss": 0.3057,
      "step": 42367
    },
    {
      "epoch": 1.944288926621082,
      "grad_norm": 0.44293010234832764,
      "learning_rate": 2.858156094924781e-06,
      "loss": 0.3252,
      "step": 42368
    },
    {
      "epoch": 1.9443348171263366,
      "grad_norm": 0.48752227425575256,
      "learning_rate": 2.8579345447266837e-06,
      "loss": 0.4254,
      "step": 42369
    },
    {
      "epoch": 1.9443807076315909,
      "grad_norm": 0.49328041076660156,
      "learning_rate": 2.857712999679575e-06,
      "loss": 0.3912,
      "step": 42370
    },
    {
      "epoch": 1.9444265981368454,
      "grad_norm": 0.4733926057815552,
      "learning_rate": 2.857491459783995e-06,
      "loss": 0.3974,
      "step": 42371
    },
    {
      "epoch": 1.9444724886420999,
      "grad_norm": 0.5088169574737549,
      "learning_rate": 2.8572699250404693e-06,
      "loss": 0.2726,
      "step": 42372
    },
    {
      "epoch": 1.9445183791473544,
      "grad_norm": 0.45664742588996887,
      "learning_rate": 2.857048395449532e-06,
      "loss": 0.3331,
      "step": 42373
    },
    {
      "epoch": 1.9445642696526089,
      "grad_norm": 0.47497960925102234,
      "learning_rate": 2.856826871011718e-06,
      "loss": 0.3564,
      "step": 42374
    },
    {
      "epoch": 1.9446101601578634,
      "grad_norm": 0.44423750042915344,
      "learning_rate": 2.8566053517275593e-06,
      "loss": 0.3015,
      "step": 42375
    },
    {
      "epoch": 1.9446560506631179,
      "grad_norm": 0.5260916352272034,
      "learning_rate": 2.856383837597587e-06,
      "loss": 0.5292,
      "step": 42376
    },
    {
      "epoch": 1.9447019411683724,
      "grad_norm": 0.4879947304725647,
      "learning_rate": 2.8561623286223363e-06,
      "loss": 0.3734,
      "step": 42377
    },
    {
      "epoch": 1.9447478316736269,
      "grad_norm": 0.4590960741043091,
      "learning_rate": 2.8559408248023383e-06,
      "loss": 0.3658,
      "step": 42378
    },
    {
      "epoch": 1.9447937221788811,
      "grad_norm": 0.47383373975753784,
      "learning_rate": 2.8557193261381244e-06,
      "loss": 0.3745,
      "step": 42379
    },
    {
      "epoch": 1.9448396126841356,
      "grad_norm": 0.47204065322875977,
      "learning_rate": 2.855497832630231e-06,
      "loss": 0.3735,
      "step": 42380
    },
    {
      "epoch": 1.9448855031893901,
      "grad_norm": 0.4993223547935486,
      "learning_rate": 2.8552763442791877e-06,
      "loss": 0.3953,
      "step": 42381
    },
    {
      "epoch": 1.9449313936946446,
      "grad_norm": 0.47566714882850647,
      "learning_rate": 2.8550548610855288e-06,
      "loss": 0.3448,
      "step": 42382
    },
    {
      "epoch": 1.944977284199899,
      "grad_norm": 0.4414668679237366,
      "learning_rate": 2.8548333830497855e-06,
      "loss": 0.3053,
      "step": 42383
    },
    {
      "epoch": 1.9450231747051534,
      "grad_norm": 0.4796183109283447,
      "learning_rate": 2.8546119101724906e-06,
      "loss": 0.375,
      "step": 42384
    },
    {
      "epoch": 1.945069065210408,
      "grad_norm": 0.500424861907959,
      "learning_rate": 2.8543904424541757e-06,
      "loss": 0.4537,
      "step": 42385
    },
    {
      "epoch": 1.9451149557156624,
      "grad_norm": 0.45035821199417114,
      "learning_rate": 2.854168979895376e-06,
      "loss": 0.3556,
      "step": 42386
    },
    {
      "epoch": 1.945160846220917,
      "grad_norm": 0.47117337584495544,
      "learning_rate": 2.8539475224966227e-06,
      "loss": 0.3817,
      "step": 42387
    },
    {
      "epoch": 1.9452067367261714,
      "grad_norm": 0.46737900376319885,
      "learning_rate": 2.853726070258447e-06,
      "loss": 0.3282,
      "step": 42388
    },
    {
      "epoch": 1.945252627231426,
      "grad_norm": 0.520505428314209,
      "learning_rate": 2.8535046231813834e-06,
      "loss": 0.4338,
      "step": 42389
    },
    {
      "epoch": 1.9452985177366804,
      "grad_norm": 0.4850632846355438,
      "learning_rate": 2.8532831812659647e-06,
      "loss": 0.3277,
      "step": 42390
    },
    {
      "epoch": 1.9453444082419349,
      "grad_norm": 0.4342845678329468,
      "learning_rate": 2.8530617445127205e-06,
      "loss": 0.3052,
      "step": 42391
    },
    {
      "epoch": 1.9453902987471892,
      "grad_norm": 0.49113643169403076,
      "learning_rate": 2.852840312922187e-06,
      "loss": 0.4142,
      "step": 42392
    },
    {
      "epoch": 1.9454361892524437,
      "grad_norm": 0.4937026798725128,
      "learning_rate": 2.852618886494897e-06,
      "loss": 0.3759,
      "step": 42393
    },
    {
      "epoch": 1.9454820797576982,
      "grad_norm": 0.5197744369506836,
      "learning_rate": 2.8523974652313764e-06,
      "loss": 0.4344,
      "step": 42394
    },
    {
      "epoch": 1.9455279702629524,
      "grad_norm": 0.505089521408081,
      "learning_rate": 2.8521760491321645e-06,
      "loss": 0.413,
      "step": 42395
    },
    {
      "epoch": 1.945573860768207,
      "grad_norm": 0.501385509967804,
      "learning_rate": 2.851954638197791e-06,
      "loss": 0.3664,
      "step": 42396
    },
    {
      "epoch": 1.9456197512734614,
      "grad_norm": 0.5000234842300415,
      "learning_rate": 2.851733232428787e-06,
      "loss": 0.4409,
      "step": 42397
    },
    {
      "epoch": 1.945665641778716,
      "grad_norm": 0.46717020869255066,
      "learning_rate": 2.8515118318256885e-06,
      "loss": 0.3585,
      "step": 42398
    },
    {
      "epoch": 1.9457115322839704,
      "grad_norm": 0.51054447889328,
      "learning_rate": 2.851290436389026e-06,
      "loss": 0.4745,
      "step": 42399
    },
    {
      "epoch": 1.945757422789225,
      "grad_norm": 0.4783273935317993,
      "learning_rate": 2.851069046119331e-06,
      "loss": 0.3521,
      "step": 42400
    },
    {
      "epoch": 1.9458033132944794,
      "grad_norm": 0.4688833951950073,
      "learning_rate": 2.8508476610171353e-06,
      "loss": 0.3685,
      "step": 42401
    },
    {
      "epoch": 1.945849203799734,
      "grad_norm": 0.4221022427082062,
      "learning_rate": 2.8506262810829756e-06,
      "loss": 0.2972,
      "step": 42402
    },
    {
      "epoch": 1.9458950943049884,
      "grad_norm": 0.454321950674057,
      "learning_rate": 2.85040490631738e-06,
      "loss": 0.3027,
      "step": 42403
    },
    {
      "epoch": 1.945940984810243,
      "grad_norm": 0.4578109383583069,
      "learning_rate": 2.850183536720883e-06,
      "loss": 0.3076,
      "step": 42404
    },
    {
      "epoch": 1.9459868753154972,
      "grad_norm": 0.4838213324546814,
      "learning_rate": 2.8499621722940155e-06,
      "loss": 0.417,
      "step": 42405
    },
    {
      "epoch": 1.9460327658207517,
      "grad_norm": 0.46016213297843933,
      "learning_rate": 2.849740813037311e-06,
      "loss": 0.3467,
      "step": 42406
    },
    {
      "epoch": 1.9460786563260062,
      "grad_norm": 0.4736923277378082,
      "learning_rate": 2.8495194589512985e-06,
      "loss": 0.3395,
      "step": 42407
    },
    {
      "epoch": 1.9461245468312605,
      "grad_norm": 0.528885543346405,
      "learning_rate": 2.8492981100365157e-06,
      "loss": 0.4364,
      "step": 42408
    },
    {
      "epoch": 1.946170437336515,
      "grad_norm": 0.4899018108844757,
      "learning_rate": 2.849076766293492e-06,
      "loss": 0.3689,
      "step": 42409
    },
    {
      "epoch": 1.9462163278417695,
      "grad_norm": 0.4459191560745239,
      "learning_rate": 2.8488554277227577e-06,
      "loss": 0.2972,
      "step": 42410
    },
    {
      "epoch": 1.946262218347024,
      "grad_norm": 0.5053806900978088,
      "learning_rate": 2.8486340943248494e-06,
      "loss": 0.4293,
      "step": 42411
    },
    {
      "epoch": 1.9463081088522785,
      "grad_norm": 0.4487350881099701,
      "learning_rate": 2.848412766100297e-06,
      "loss": 0.3316,
      "step": 42412
    },
    {
      "epoch": 1.946353999357533,
      "grad_norm": 0.4759421646595001,
      "learning_rate": 2.8481914430496303e-06,
      "loss": 0.3425,
      "step": 42413
    },
    {
      "epoch": 1.9463998898627874,
      "grad_norm": 0.4829477369785309,
      "learning_rate": 2.8479701251733864e-06,
      "loss": 0.3586,
      "step": 42414
    },
    {
      "epoch": 1.946445780368042,
      "grad_norm": 0.4899613857269287,
      "learning_rate": 2.8477488124720975e-06,
      "loss": 0.3578,
      "step": 42415
    },
    {
      "epoch": 1.9464916708732964,
      "grad_norm": 0.5239880084991455,
      "learning_rate": 2.847527504946289e-06,
      "loss": 0.4435,
      "step": 42416
    },
    {
      "epoch": 1.946537561378551,
      "grad_norm": 0.5071898698806763,
      "learning_rate": 2.8473062025964997e-06,
      "loss": 0.3325,
      "step": 42417
    },
    {
      "epoch": 1.9465834518838052,
      "grad_norm": 0.48911234736442566,
      "learning_rate": 2.84708490542326e-06,
      "loss": 0.3749,
      "step": 42418
    },
    {
      "epoch": 1.9466293423890597,
      "grad_norm": 0.4952391982078552,
      "learning_rate": 2.846863613427099e-06,
      "loss": 0.4127,
      "step": 42419
    },
    {
      "epoch": 1.9466752328943142,
      "grad_norm": 0.5020302534103394,
      "learning_rate": 2.8466423266085534e-06,
      "loss": 0.4434,
      "step": 42420
    },
    {
      "epoch": 1.9467211233995685,
      "grad_norm": 0.46724188327789307,
      "learning_rate": 2.846421044968154e-06,
      "loss": 0.3404,
      "step": 42421
    },
    {
      "epoch": 1.946767013904823,
      "grad_norm": 0.47430843114852905,
      "learning_rate": 2.8461997685064295e-06,
      "loss": 0.3413,
      "step": 42422
    },
    {
      "epoch": 1.9468129044100775,
      "grad_norm": 0.4442247748374939,
      "learning_rate": 2.8459784972239176e-06,
      "loss": 0.3207,
      "step": 42423
    },
    {
      "epoch": 1.946858794915332,
      "grad_norm": 0.49720704555511475,
      "learning_rate": 2.8457572311211468e-06,
      "loss": 0.3965,
      "step": 42424
    },
    {
      "epoch": 1.9469046854205865,
      "grad_norm": 0.483260840177536,
      "learning_rate": 2.8455359701986508e-06,
      "loss": 0.3772,
      "step": 42425
    },
    {
      "epoch": 1.946950575925841,
      "grad_norm": 0.4952370822429657,
      "learning_rate": 2.84531471445696e-06,
      "loss": 0.4414,
      "step": 42426
    },
    {
      "epoch": 1.9469964664310955,
      "grad_norm": 0.4601132869720459,
      "learning_rate": 2.845093463896608e-06,
      "loss": 0.3128,
      "step": 42427
    },
    {
      "epoch": 1.94704235693635,
      "grad_norm": 0.49074140191078186,
      "learning_rate": 2.844872218518123e-06,
      "loss": 0.3817,
      "step": 42428
    },
    {
      "epoch": 1.9470882474416045,
      "grad_norm": 0.491845965385437,
      "learning_rate": 2.844650978322043e-06,
      "loss": 0.4049,
      "step": 42429
    },
    {
      "epoch": 1.9471341379468587,
      "grad_norm": 0.4709020256996155,
      "learning_rate": 2.8444297433088975e-06,
      "loss": 0.3901,
      "step": 42430
    },
    {
      "epoch": 1.9471800284521132,
      "grad_norm": 0.4624183177947998,
      "learning_rate": 2.8442085134792153e-06,
      "loss": 0.3227,
      "step": 42431
    },
    {
      "epoch": 1.9472259189573677,
      "grad_norm": 0.4746169447898865,
      "learning_rate": 2.843987288833534e-06,
      "loss": 0.3905,
      "step": 42432
    },
    {
      "epoch": 1.9472718094626222,
      "grad_norm": 0.46739792823791504,
      "learning_rate": 2.843766069372382e-06,
      "loss": 0.3644,
      "step": 42433
    },
    {
      "epoch": 1.9473176999678765,
      "grad_norm": 0.5035632848739624,
      "learning_rate": 2.843544855096293e-06,
      "loss": 0.3695,
      "step": 42434
    },
    {
      "epoch": 1.947363590473131,
      "grad_norm": 0.508005678653717,
      "learning_rate": 2.8433236460057958e-06,
      "loss": 0.4164,
      "step": 42435
    },
    {
      "epoch": 1.9474094809783855,
      "grad_norm": 0.45017924904823303,
      "learning_rate": 2.8431024421014264e-06,
      "loss": 0.3457,
      "step": 42436
    },
    {
      "epoch": 1.94745537148364,
      "grad_norm": 0.4457870423793793,
      "learning_rate": 2.8428812433837163e-06,
      "loss": 0.3058,
      "step": 42437
    },
    {
      "epoch": 1.9475012619888945,
      "grad_norm": 0.4544684588909149,
      "learning_rate": 2.8426600498531927e-06,
      "loss": 0.3676,
      "step": 42438
    },
    {
      "epoch": 1.947547152494149,
      "grad_norm": 0.4508378505706787,
      "learning_rate": 2.8424388615103928e-06,
      "loss": 0.2974,
      "step": 42439
    },
    {
      "epoch": 1.9475930429994035,
      "grad_norm": 0.44784507155418396,
      "learning_rate": 2.842217678355846e-06,
      "loss": 0.3011,
      "step": 42440
    },
    {
      "epoch": 1.947638933504658,
      "grad_norm": 0.45314615964889526,
      "learning_rate": 2.841996500390083e-06,
      "loss": 0.3105,
      "step": 42441
    },
    {
      "epoch": 1.9476848240099125,
      "grad_norm": 0.4720838963985443,
      "learning_rate": 2.8417753276136394e-06,
      "loss": 0.3412,
      "step": 42442
    },
    {
      "epoch": 1.9477307145151668,
      "grad_norm": 0.46888601779937744,
      "learning_rate": 2.8415541600270447e-06,
      "loss": 0.3345,
      "step": 42443
    },
    {
      "epoch": 1.9477766050204213,
      "grad_norm": 0.4765230715274811,
      "learning_rate": 2.841332997630829e-06,
      "loss": 0.3134,
      "step": 42444
    },
    {
      "epoch": 1.9478224955256758,
      "grad_norm": 0.5001325011253357,
      "learning_rate": 2.841111840425529e-06,
      "loss": 0.3799,
      "step": 42445
    },
    {
      "epoch": 1.94786838603093,
      "grad_norm": 0.5129137635231018,
      "learning_rate": 2.8408906884116727e-06,
      "loss": 0.4493,
      "step": 42446
    },
    {
      "epoch": 1.9479142765361845,
      "grad_norm": 0.521508514881134,
      "learning_rate": 2.8406695415897934e-06,
      "loss": 0.4657,
      "step": 42447
    },
    {
      "epoch": 1.947960167041439,
      "grad_norm": 0.4937654733657837,
      "learning_rate": 2.8404483999604217e-06,
      "loss": 0.3902,
      "step": 42448
    },
    {
      "epoch": 1.9480060575466935,
      "grad_norm": 0.4775790572166443,
      "learning_rate": 2.840227263524091e-06,
      "loss": 0.3627,
      "step": 42449
    },
    {
      "epoch": 1.948051948051948,
      "grad_norm": 0.43531087040901184,
      "learning_rate": 2.840006132281329e-06,
      "loss": 0.3293,
      "step": 42450
    },
    {
      "epoch": 1.9480978385572025,
      "grad_norm": 0.46937450766563416,
      "learning_rate": 2.839785006232673e-06,
      "loss": 0.3652,
      "step": 42451
    },
    {
      "epoch": 1.948143729062457,
      "grad_norm": 0.47242382168769836,
      "learning_rate": 2.839563885378652e-06,
      "loss": 0.3479,
      "step": 42452
    },
    {
      "epoch": 1.9481896195677115,
      "grad_norm": 0.4657030999660492,
      "learning_rate": 2.8393427697197973e-06,
      "loss": 0.3583,
      "step": 42453
    },
    {
      "epoch": 1.948235510072966,
      "grad_norm": 0.439483106136322,
      "learning_rate": 2.8391216592566416e-06,
      "loss": 0.2991,
      "step": 42454
    },
    {
      "epoch": 1.9482814005782205,
      "grad_norm": 0.44933009147644043,
      "learning_rate": 2.8389005539897176e-06,
      "loss": 0.3354,
      "step": 42455
    },
    {
      "epoch": 1.9483272910834748,
      "grad_norm": 0.4766775071620941,
      "learning_rate": 2.838679453919553e-06,
      "loss": 0.3551,
      "step": 42456
    },
    {
      "epoch": 1.9483731815887293,
      "grad_norm": 0.45603442192077637,
      "learning_rate": 2.8384583590466845e-06,
      "loss": 0.304,
      "step": 42457
    },
    {
      "epoch": 1.9484190720939838,
      "grad_norm": 0.4871738851070404,
      "learning_rate": 2.8382372693716414e-06,
      "loss": 0.3833,
      "step": 42458
    },
    {
      "epoch": 1.948464962599238,
      "grad_norm": 0.4615531265735626,
      "learning_rate": 2.838016184894955e-06,
      "loss": 0.3202,
      "step": 42459
    },
    {
      "epoch": 1.9485108531044926,
      "grad_norm": 0.4755474328994751,
      "learning_rate": 2.837795105617158e-06,
      "loss": 0.4256,
      "step": 42460
    },
    {
      "epoch": 1.948556743609747,
      "grad_norm": 0.4822428822517395,
      "learning_rate": 2.8375740315387812e-06,
      "loss": 0.3565,
      "step": 42461
    },
    {
      "epoch": 1.9486026341150016,
      "grad_norm": 0.43045395612716675,
      "learning_rate": 2.8373529626603536e-06,
      "loss": 0.2924,
      "step": 42462
    },
    {
      "epoch": 1.948648524620256,
      "grad_norm": 0.47284722328186035,
      "learning_rate": 2.8371318989824127e-06,
      "loss": 0.3711,
      "step": 42463
    },
    {
      "epoch": 1.9486944151255106,
      "grad_norm": 0.4907500147819519,
      "learning_rate": 2.8369108405054862e-06,
      "loss": 0.3948,
      "step": 42464
    },
    {
      "epoch": 1.948740305630765,
      "grad_norm": 0.5306557416915894,
      "learning_rate": 2.836689787230107e-06,
      "loss": 0.4532,
      "step": 42465
    },
    {
      "epoch": 1.9487861961360196,
      "grad_norm": 0.4576500654220581,
      "learning_rate": 2.8364687391568037e-06,
      "loss": 0.3428,
      "step": 42466
    },
    {
      "epoch": 1.948832086641274,
      "grad_norm": 0.4818820059299469,
      "learning_rate": 2.836247696286112e-06,
      "loss": 0.4032,
      "step": 42467
    },
    {
      "epoch": 1.9488779771465283,
      "grad_norm": 0.4623189866542816,
      "learning_rate": 2.836026658618562e-06,
      "loss": 0.3406,
      "step": 42468
    },
    {
      "epoch": 1.9489238676517828,
      "grad_norm": 0.5096890330314636,
      "learning_rate": 2.835805626154685e-06,
      "loss": 0.4218,
      "step": 42469
    },
    {
      "epoch": 1.9489697581570373,
      "grad_norm": 0.4543008506298065,
      "learning_rate": 2.8355845988950125e-06,
      "loss": 0.3235,
      "step": 42470
    },
    {
      "epoch": 1.9490156486622918,
      "grad_norm": 0.4332923889160156,
      "learning_rate": 2.8353635768400754e-06,
      "loss": 0.2731,
      "step": 42471
    },
    {
      "epoch": 1.949061539167546,
      "grad_norm": 0.6159323453903198,
      "learning_rate": 2.8351425599904044e-06,
      "loss": 0.3947,
      "step": 42472
    },
    {
      "epoch": 1.9491074296728006,
      "grad_norm": 0.4629206359386444,
      "learning_rate": 2.834921548346534e-06,
      "loss": 0.3405,
      "step": 42473
    },
    {
      "epoch": 1.949153320178055,
      "grad_norm": 0.4813768267631531,
      "learning_rate": 2.8347005419089933e-06,
      "loss": 0.4072,
      "step": 42474
    },
    {
      "epoch": 1.9491992106833096,
      "grad_norm": 0.4381105303764343,
      "learning_rate": 2.8344795406783133e-06,
      "loss": 0.2965,
      "step": 42475
    },
    {
      "epoch": 1.949245101188564,
      "grad_norm": 0.4875320494174957,
      "learning_rate": 2.8342585446550285e-06,
      "loss": 0.3751,
      "step": 42476
    },
    {
      "epoch": 1.9492909916938186,
      "grad_norm": 0.5055011510848999,
      "learning_rate": 2.834037553839668e-06,
      "loss": 0.4668,
      "step": 42477
    },
    {
      "epoch": 1.949336882199073,
      "grad_norm": 0.48518139123916626,
      "learning_rate": 2.833816568232762e-06,
      "loss": 0.4158,
      "step": 42478
    },
    {
      "epoch": 1.9493827727043276,
      "grad_norm": 0.4666910171508789,
      "learning_rate": 2.8335955878348455e-06,
      "loss": 0.3554,
      "step": 42479
    },
    {
      "epoch": 1.949428663209582,
      "grad_norm": 1.4019774198532104,
      "learning_rate": 2.833374612646447e-06,
      "loss": 0.3494,
      "step": 42480
    },
    {
      "epoch": 1.9494745537148364,
      "grad_norm": 0.4731694161891937,
      "learning_rate": 2.8331536426680994e-06,
      "loss": 0.3393,
      "step": 42481
    },
    {
      "epoch": 1.9495204442200909,
      "grad_norm": 0.49350517988204956,
      "learning_rate": 2.8329326779003336e-06,
      "loss": 0.3877,
      "step": 42482
    },
    {
      "epoch": 1.9495663347253454,
      "grad_norm": 0.4587632417678833,
      "learning_rate": 2.83271171834368e-06,
      "loss": 0.3364,
      "step": 42483
    },
    {
      "epoch": 1.9496122252305996,
      "grad_norm": 0.4981546103954315,
      "learning_rate": 2.8324907639986697e-06,
      "loss": 0.4214,
      "step": 42484
    },
    {
      "epoch": 1.9496581157358541,
      "grad_norm": 0.48166605830192566,
      "learning_rate": 2.8322698148658366e-06,
      "loss": 0.3567,
      "step": 42485
    },
    {
      "epoch": 1.9497040062411086,
      "grad_norm": 0.49478209018707275,
      "learning_rate": 2.8320488709457104e-06,
      "loss": 0.3692,
      "step": 42486
    },
    {
      "epoch": 1.9497498967463631,
      "grad_norm": 0.43518316745758057,
      "learning_rate": 2.8318279322388203e-06,
      "loss": 0.2934,
      "step": 42487
    },
    {
      "epoch": 1.9497957872516176,
      "grad_norm": 0.44196969270706177,
      "learning_rate": 2.831606998745702e-06,
      "loss": 0.3045,
      "step": 42488
    },
    {
      "epoch": 1.9498416777568721,
      "grad_norm": 0.5177297592163086,
      "learning_rate": 2.8313860704668843e-06,
      "loss": 0.4622,
      "step": 42489
    },
    {
      "epoch": 1.9498875682621266,
      "grad_norm": 0.4825424253940582,
      "learning_rate": 2.831165147402899e-06,
      "loss": 0.3981,
      "step": 42490
    },
    {
      "epoch": 1.9499334587673811,
      "grad_norm": 0.5021687150001526,
      "learning_rate": 2.8309442295542764e-06,
      "loss": 0.3819,
      "step": 42491
    },
    {
      "epoch": 1.9499793492726356,
      "grad_norm": 0.5172713994979858,
      "learning_rate": 2.8307233169215487e-06,
      "loss": 0.3725,
      "step": 42492
    },
    {
      "epoch": 1.95002523977789,
      "grad_norm": 0.4370590150356293,
      "learning_rate": 2.8305024095052446e-06,
      "loss": 0.3105,
      "step": 42493
    },
    {
      "epoch": 1.9500711302831444,
      "grad_norm": 0.4698132574558258,
      "learning_rate": 2.8302815073058998e-06,
      "loss": 0.3446,
      "step": 42494
    },
    {
      "epoch": 1.9501170207883989,
      "grad_norm": 0.4961320757865906,
      "learning_rate": 2.8300606103240437e-06,
      "loss": 0.3915,
      "step": 42495
    },
    {
      "epoch": 1.9501629112936534,
      "grad_norm": 0.5245936512947083,
      "learning_rate": 2.8298397185602046e-06,
      "loss": 0.4078,
      "step": 42496
    },
    {
      "epoch": 1.9502088017989077,
      "grad_norm": 0.4557390511035919,
      "learning_rate": 2.8296188320149176e-06,
      "loss": 0.3221,
      "step": 42497
    },
    {
      "epoch": 1.9502546923041622,
      "grad_norm": 0.46944060921669006,
      "learning_rate": 2.8293979506887127e-06,
      "loss": 0.3528,
      "step": 42498
    },
    {
      "epoch": 1.9503005828094166,
      "grad_norm": 0.40785226225852966,
      "learning_rate": 2.82917707458212e-06,
      "loss": 0.2672,
      "step": 42499
    },
    {
      "epoch": 1.9503464733146711,
      "grad_norm": 0.45082634687423706,
      "learning_rate": 2.82895620369567e-06,
      "loss": 0.3222,
      "step": 42500
    },
    {
      "epoch": 1.9503923638199256,
      "grad_norm": 0.47749245166778564,
      "learning_rate": 2.8287353380298966e-06,
      "loss": 0.3402,
      "step": 42501
    },
    {
      "epoch": 1.9504382543251801,
      "grad_norm": 0.46454060077667236,
      "learning_rate": 2.8285144775853303e-06,
      "loss": 0.3184,
      "step": 42502
    },
    {
      "epoch": 1.9504841448304346,
      "grad_norm": 0.4924120604991913,
      "learning_rate": 2.8282936223625e-06,
      "loss": 0.3561,
      "step": 42503
    },
    {
      "epoch": 1.9505300353356891,
      "grad_norm": 0.4693794548511505,
      "learning_rate": 2.8280727723619393e-06,
      "loss": 0.3846,
      "step": 42504
    },
    {
      "epoch": 1.9505759258409436,
      "grad_norm": 0.47027331590652466,
      "learning_rate": 2.8278519275841777e-06,
      "loss": 0.3835,
      "step": 42505
    },
    {
      "epoch": 1.9506218163461981,
      "grad_norm": 0.4731113612651825,
      "learning_rate": 2.827631088029745e-06,
      "loss": 0.3744,
      "step": 42506
    },
    {
      "epoch": 1.9506677068514524,
      "grad_norm": 0.4881572723388672,
      "learning_rate": 2.8274102536991754e-06,
      "loss": 0.3701,
      "step": 42507
    },
    {
      "epoch": 1.950713597356707,
      "grad_norm": 0.43266406655311584,
      "learning_rate": 2.827189424592999e-06,
      "loss": 0.3083,
      "step": 42508
    },
    {
      "epoch": 1.9507594878619614,
      "grad_norm": 0.5126971006393433,
      "learning_rate": 2.826968600711744e-06,
      "loss": 0.4272,
      "step": 42509
    },
    {
      "epoch": 1.9508053783672157,
      "grad_norm": 0.46197235584259033,
      "learning_rate": 2.826747782055945e-06,
      "loss": 0.3321,
      "step": 42510
    },
    {
      "epoch": 1.9508512688724702,
      "grad_norm": 0.4891059100627899,
      "learning_rate": 2.8265269686261326e-06,
      "loss": 0.3618,
      "step": 42511
    },
    {
      "epoch": 1.9508971593777247,
      "grad_norm": 0.44532984495162964,
      "learning_rate": 2.826306160422837e-06,
      "loss": 0.2911,
      "step": 42512
    },
    {
      "epoch": 1.9509430498829792,
      "grad_norm": 0.4884195029735565,
      "learning_rate": 2.8260853574465884e-06,
      "loss": 0.4022,
      "step": 42513
    },
    {
      "epoch": 1.9509889403882337,
      "grad_norm": 0.49024680256843567,
      "learning_rate": 2.8258645596979185e-06,
      "loss": 0.4169,
      "step": 42514
    },
    {
      "epoch": 1.9510348308934882,
      "grad_norm": 0.4090471565723419,
      "learning_rate": 2.825643767177356e-06,
      "loss": 0.2565,
      "step": 42515
    },
    {
      "epoch": 1.9510807213987427,
      "grad_norm": 0.5053630471229553,
      "learning_rate": 2.825422979885437e-06,
      "loss": 0.3949,
      "step": 42516
    },
    {
      "epoch": 1.9511266119039972,
      "grad_norm": 0.4524213671684265,
      "learning_rate": 2.8252021978226883e-06,
      "loss": 0.2911,
      "step": 42517
    },
    {
      "epoch": 1.9511725024092517,
      "grad_norm": 0.4692447781562805,
      "learning_rate": 2.824981420989641e-06,
      "loss": 0.329,
      "step": 42518
    },
    {
      "epoch": 1.951218392914506,
      "grad_norm": 0.4875037968158722,
      "learning_rate": 2.824760649386828e-06,
      "loss": 0.339,
      "step": 42519
    },
    {
      "epoch": 1.9512642834197604,
      "grad_norm": 0.5145904421806335,
      "learning_rate": 2.824539883014779e-06,
      "loss": 0.4635,
      "step": 42520
    },
    {
      "epoch": 1.951310173925015,
      "grad_norm": 0.451589435338974,
      "learning_rate": 2.8243191218740237e-06,
      "loss": 0.3266,
      "step": 42521
    },
    {
      "epoch": 1.9513560644302694,
      "grad_norm": 0.4614278972148895,
      "learning_rate": 2.8240983659650955e-06,
      "loss": 0.3158,
      "step": 42522
    },
    {
      "epoch": 1.9514019549355237,
      "grad_norm": 0.480217307806015,
      "learning_rate": 2.823877615288525e-06,
      "loss": 0.3556,
      "step": 42523
    },
    {
      "epoch": 1.9514478454407782,
      "grad_norm": 0.43987399339675903,
      "learning_rate": 2.8236568698448407e-06,
      "loss": 0.3067,
      "step": 42524
    },
    {
      "epoch": 1.9514937359460327,
      "grad_norm": 0.4859858453273773,
      "learning_rate": 2.823436129634576e-06,
      "loss": 0.3376,
      "step": 42525
    },
    {
      "epoch": 1.9515396264512872,
      "grad_norm": 0.47993794083595276,
      "learning_rate": 2.82321539465826e-06,
      "loss": 0.3751,
      "step": 42526
    },
    {
      "epoch": 1.9515855169565417,
      "grad_norm": 0.4649876058101654,
      "learning_rate": 2.822994664916422e-06,
      "loss": 0.3209,
      "step": 42527
    },
    {
      "epoch": 1.9516314074617962,
      "grad_norm": 0.4646109640598297,
      "learning_rate": 2.8227739404095966e-06,
      "loss": 0.3221,
      "step": 42528
    },
    {
      "epoch": 1.9516772979670507,
      "grad_norm": 0.4545367658138275,
      "learning_rate": 2.8225532211383128e-06,
      "loss": 0.3629,
      "step": 42529
    },
    {
      "epoch": 1.9517231884723052,
      "grad_norm": 0.48687881231307983,
      "learning_rate": 2.8223325071030995e-06,
      "loss": 0.4337,
      "step": 42530
    },
    {
      "epoch": 1.9517690789775597,
      "grad_norm": 0.5047340393066406,
      "learning_rate": 2.822111798304491e-06,
      "loss": 0.4243,
      "step": 42531
    },
    {
      "epoch": 1.951814969482814,
      "grad_norm": 0.4757983684539795,
      "learning_rate": 2.8218910947430157e-06,
      "loss": 0.3305,
      "step": 42532
    },
    {
      "epoch": 1.9518608599880685,
      "grad_norm": 0.45608747005462646,
      "learning_rate": 2.821670396419206e-06,
      "loss": 0.302,
      "step": 42533
    },
    {
      "epoch": 1.951906750493323,
      "grad_norm": 0.47517067193984985,
      "learning_rate": 2.821449703333591e-06,
      "loss": 0.3433,
      "step": 42534
    },
    {
      "epoch": 1.9519526409985772,
      "grad_norm": 0.48527246713638306,
      "learning_rate": 2.821229015486702e-06,
      "loss": 0.3726,
      "step": 42535
    },
    {
      "epoch": 1.9519985315038317,
      "grad_norm": 0.4840909242630005,
      "learning_rate": 2.821008332879069e-06,
      "loss": 0.4219,
      "step": 42536
    },
    {
      "epoch": 1.9520444220090862,
      "grad_norm": 0.48162558674812317,
      "learning_rate": 2.820787655511221e-06,
      "loss": 0.3283,
      "step": 42537
    },
    {
      "epoch": 1.9520903125143407,
      "grad_norm": 0.4531991481781006,
      "learning_rate": 2.8205669833836944e-06,
      "loss": 0.3301,
      "step": 42538
    },
    {
      "epoch": 1.9521362030195952,
      "grad_norm": 0.4448402523994446,
      "learning_rate": 2.820346316497016e-06,
      "loss": 0.2906,
      "step": 42539
    },
    {
      "epoch": 1.9521820935248497,
      "grad_norm": 0.5169556140899658,
      "learning_rate": 2.8201256548517137e-06,
      "loss": 0.4126,
      "step": 42540
    },
    {
      "epoch": 1.9522279840301042,
      "grad_norm": 0.4678075313568115,
      "learning_rate": 2.819904998448324e-06,
      "loss": 0.3408,
      "step": 42541
    },
    {
      "epoch": 1.9522738745353587,
      "grad_norm": 0.4547345042228699,
      "learning_rate": 2.819684347287375e-06,
      "loss": 0.3387,
      "step": 42542
    },
    {
      "epoch": 1.9523197650406132,
      "grad_norm": 0.4803924262523651,
      "learning_rate": 2.819463701369394e-06,
      "loss": 0.3542,
      "step": 42543
    },
    {
      "epoch": 1.9523656555458677,
      "grad_norm": 0.47113198041915894,
      "learning_rate": 2.8192430606949172e-06,
      "loss": 0.387,
      "step": 42544
    },
    {
      "epoch": 1.952411546051122,
      "grad_norm": 0.47771966457366943,
      "learning_rate": 2.819022425264473e-06,
      "loss": 0.3581,
      "step": 42545
    },
    {
      "epoch": 1.9524574365563765,
      "grad_norm": 0.4476812481880188,
      "learning_rate": 2.8188017950785905e-06,
      "loss": 0.3195,
      "step": 42546
    },
    {
      "epoch": 1.952503327061631,
      "grad_norm": 0.46197864413261414,
      "learning_rate": 2.8185811701378018e-06,
      "loss": 0.3335,
      "step": 42547
    },
    {
      "epoch": 1.9525492175668853,
      "grad_norm": 0.4620197117328644,
      "learning_rate": 2.8183605504426372e-06,
      "loss": 0.3381,
      "step": 42548
    },
    {
      "epoch": 1.9525951080721398,
      "grad_norm": 0.4671401381492615,
      "learning_rate": 2.8181399359936246e-06,
      "loss": 0.362,
      "step": 42549
    },
    {
      "epoch": 1.9526409985773943,
      "grad_norm": 0.5330823063850403,
      "learning_rate": 2.817919326791299e-06,
      "loss": 0.4382,
      "step": 42550
    },
    {
      "epoch": 1.9526868890826488,
      "grad_norm": 0.46532678604125977,
      "learning_rate": 2.8176987228361883e-06,
      "loss": 0.3429,
      "step": 42551
    },
    {
      "epoch": 1.9527327795879033,
      "grad_norm": 0.40732866525650024,
      "learning_rate": 2.817478124128822e-06,
      "loss": 0.2593,
      "step": 42552
    },
    {
      "epoch": 1.9527786700931578,
      "grad_norm": 0.5604402422904968,
      "learning_rate": 2.8172575306697337e-06,
      "loss": 0.4952,
      "step": 42553
    },
    {
      "epoch": 1.9528245605984123,
      "grad_norm": 0.47648921608924866,
      "learning_rate": 2.8170369424594525e-06,
      "loss": 0.3899,
      "step": 42554
    },
    {
      "epoch": 1.9528704511036667,
      "grad_norm": 0.46424171328544617,
      "learning_rate": 2.816816359498508e-06,
      "loss": 0.3295,
      "step": 42555
    },
    {
      "epoch": 1.9529163416089212,
      "grad_norm": 0.4813767075538635,
      "learning_rate": 2.816595781787431e-06,
      "loss": 0.3436,
      "step": 42556
    },
    {
      "epoch": 1.9529622321141755,
      "grad_norm": 0.4569571614265442,
      "learning_rate": 2.8163752093267527e-06,
      "loss": 0.3446,
      "step": 42557
    },
    {
      "epoch": 1.95300812261943,
      "grad_norm": 0.48273661732673645,
      "learning_rate": 2.8161546421170005e-06,
      "loss": 0.4289,
      "step": 42558
    },
    {
      "epoch": 1.9530540131246845,
      "grad_norm": 0.46390098333358765,
      "learning_rate": 2.81593408015871e-06,
      "loss": 0.3394,
      "step": 42559
    },
    {
      "epoch": 1.953099903629939,
      "grad_norm": 0.5205157399177551,
      "learning_rate": 2.8157135234524075e-06,
      "loss": 0.3802,
      "step": 42560
    },
    {
      "epoch": 1.9531457941351933,
      "grad_norm": 0.48730960488319397,
      "learning_rate": 2.815492971998623e-06,
      "loss": 0.3552,
      "step": 42561
    },
    {
      "epoch": 1.9531916846404478,
      "grad_norm": 0.4250916540622711,
      "learning_rate": 2.8152724257978904e-06,
      "loss": 0.2488,
      "step": 42562
    },
    {
      "epoch": 1.9532375751457023,
      "grad_norm": 0.5010707974433899,
      "learning_rate": 2.815051884850738e-06,
      "loss": 0.4288,
      "step": 42563
    },
    {
      "epoch": 1.9532834656509568,
      "grad_norm": 0.49828991293907166,
      "learning_rate": 2.814831349157696e-06,
      "loss": 0.4441,
      "step": 42564
    },
    {
      "epoch": 1.9533293561562113,
      "grad_norm": 0.49227431416511536,
      "learning_rate": 2.814610818719294e-06,
      "loss": 0.3693,
      "step": 42565
    },
    {
      "epoch": 1.9533752466614658,
      "grad_norm": 0.48540937900543213,
      "learning_rate": 2.8143902935360646e-06,
      "loss": 0.3592,
      "step": 42566
    },
    {
      "epoch": 1.9534211371667203,
      "grad_norm": 0.4975406527519226,
      "learning_rate": 2.814169773608536e-06,
      "loss": 0.3813,
      "step": 42567
    },
    {
      "epoch": 1.9534670276719748,
      "grad_norm": 0.47891250252723694,
      "learning_rate": 2.8139492589372407e-06,
      "loss": 0.3517,
      "step": 42568
    },
    {
      "epoch": 1.9535129181772293,
      "grad_norm": 0.45402589440345764,
      "learning_rate": 2.813728749522706e-06,
      "loss": 0.3129,
      "step": 42569
    },
    {
      "epoch": 1.9535588086824835,
      "grad_norm": 0.48088952898979187,
      "learning_rate": 2.813508245365465e-06,
      "loss": 0.3743,
      "step": 42570
    },
    {
      "epoch": 1.953604699187738,
      "grad_norm": 0.48054125905036926,
      "learning_rate": 2.8132877464660435e-06,
      "loss": 0.3847,
      "step": 42571
    },
    {
      "epoch": 1.9536505896929925,
      "grad_norm": 0.4481973350048065,
      "learning_rate": 2.8130672528249775e-06,
      "loss": 0.3266,
      "step": 42572
    },
    {
      "epoch": 1.9536964801982468,
      "grad_norm": 0.5275359153747559,
      "learning_rate": 2.8128467644427947e-06,
      "loss": 0.3731,
      "step": 42573
    },
    {
      "epoch": 1.9537423707035013,
      "grad_norm": 0.48155292868614197,
      "learning_rate": 2.812626281320022e-06,
      "loss": 0.3689,
      "step": 42574
    },
    {
      "epoch": 1.9537882612087558,
      "grad_norm": 0.4975397288799286,
      "learning_rate": 2.812405803457196e-06,
      "loss": 0.3972,
      "step": 42575
    },
    {
      "epoch": 1.9538341517140103,
      "grad_norm": 0.609037458896637,
      "learning_rate": 2.812185330854843e-06,
      "loss": 0.383,
      "step": 42576
    },
    {
      "epoch": 1.9538800422192648,
      "grad_norm": 0.47047191858291626,
      "learning_rate": 2.8119648635134943e-06,
      "loss": 0.3378,
      "step": 42577
    },
    {
      "epoch": 1.9539259327245193,
      "grad_norm": 0.4643189311027527,
      "learning_rate": 2.8117444014336793e-06,
      "loss": 0.3277,
      "step": 42578
    },
    {
      "epoch": 1.9539718232297738,
      "grad_norm": 0.49498069286346436,
      "learning_rate": 2.811523944615928e-06,
      "loss": 0.3649,
      "step": 42579
    },
    {
      "epoch": 1.9540177137350283,
      "grad_norm": 0.5016837120056152,
      "learning_rate": 2.811303493060769e-06,
      "loss": 0.4672,
      "step": 42580
    },
    {
      "epoch": 1.9540636042402828,
      "grad_norm": 0.4253122806549072,
      "learning_rate": 2.811083046768737e-06,
      "loss": 0.2697,
      "step": 42581
    },
    {
      "epoch": 1.9541094947455373,
      "grad_norm": 0.4417501389980316,
      "learning_rate": 2.8108626057403587e-06,
      "loss": 0.2855,
      "step": 42582
    },
    {
      "epoch": 1.9541553852507916,
      "grad_norm": 0.4434540271759033,
      "learning_rate": 2.8106421699761633e-06,
      "loss": 0.3095,
      "step": 42583
    },
    {
      "epoch": 1.954201275756046,
      "grad_norm": 0.46882364153862,
      "learning_rate": 2.8104217394766843e-06,
      "loss": 0.3492,
      "step": 42584
    },
    {
      "epoch": 1.9542471662613006,
      "grad_norm": 0.5064638257026672,
      "learning_rate": 2.81020131424245e-06,
      "loss": 0.3942,
      "step": 42585
    },
    {
      "epoch": 1.9542930567665548,
      "grad_norm": 0.5324026942253113,
      "learning_rate": 2.809980894273988e-06,
      "loss": 0.4332,
      "step": 42586
    },
    {
      "epoch": 1.9543389472718093,
      "grad_norm": 0.4467504620552063,
      "learning_rate": 2.8097604795718337e-06,
      "loss": 0.3007,
      "step": 42587
    },
    {
      "epoch": 1.9543848377770638,
      "grad_norm": 0.48012593388557434,
      "learning_rate": 2.809540070136513e-06,
      "loss": 0.3334,
      "step": 42588
    },
    {
      "epoch": 1.9544307282823183,
      "grad_norm": 0.45491722226142883,
      "learning_rate": 2.8093196659685585e-06,
      "loss": 0.3636,
      "step": 42589
    },
    {
      "epoch": 1.9544766187875728,
      "grad_norm": 0.440604567527771,
      "learning_rate": 2.809099267068498e-06,
      "loss": 0.3178,
      "step": 42590
    },
    {
      "epoch": 1.9545225092928273,
      "grad_norm": 0.44432374835014343,
      "learning_rate": 2.808878873436862e-06,
      "loss": 0.3337,
      "step": 42591
    },
    {
      "epoch": 1.9545683997980818,
      "grad_norm": 0.4582633078098297,
      "learning_rate": 2.8086584850741793e-06,
      "loss": 0.3219,
      "step": 42592
    },
    {
      "epoch": 1.9546142903033363,
      "grad_norm": 0.46526646614074707,
      "learning_rate": 2.8084381019809833e-06,
      "loss": 0.377,
      "step": 42593
    },
    {
      "epoch": 1.9546601808085908,
      "grad_norm": 0.4804357886314392,
      "learning_rate": 2.808217724157802e-06,
      "loss": 0.3765,
      "step": 42594
    },
    {
      "epoch": 1.9547060713138453,
      "grad_norm": 0.46754559874534607,
      "learning_rate": 2.807997351605163e-06,
      "loss": 0.3681,
      "step": 42595
    },
    {
      "epoch": 1.9547519618190996,
      "grad_norm": 0.46250200271606445,
      "learning_rate": 2.807776984323601e-06,
      "loss": 0.3192,
      "step": 42596
    },
    {
      "epoch": 1.954797852324354,
      "grad_norm": 0.4233126938343048,
      "learning_rate": 2.807556622313643e-06,
      "loss": 0.3075,
      "step": 42597
    },
    {
      "epoch": 1.9548437428296086,
      "grad_norm": 0.4820030927658081,
      "learning_rate": 2.8073362655758197e-06,
      "loss": 0.3236,
      "step": 42598
    },
    {
      "epoch": 1.9548896333348629,
      "grad_norm": 0.4851837754249573,
      "learning_rate": 2.8071159141106607e-06,
      "loss": 0.3748,
      "step": 42599
    },
    {
      "epoch": 1.9549355238401174,
      "grad_norm": 0.4749301075935364,
      "learning_rate": 2.8068955679186956e-06,
      "loss": 0.3413,
      "step": 42600
    },
    {
      "epoch": 1.9549814143453719,
      "grad_norm": 0.4721699655056,
      "learning_rate": 2.806675227000454e-06,
      "loss": 0.3423,
      "step": 42601
    },
    {
      "epoch": 1.9550273048506264,
      "grad_norm": 0.4495364725589752,
      "learning_rate": 2.806454891356465e-06,
      "loss": 0.3039,
      "step": 42602
    },
    {
      "epoch": 1.9550731953558809,
      "grad_norm": 0.4818999767303467,
      "learning_rate": 2.8062345609872617e-06,
      "loss": 0.4053,
      "step": 42603
    },
    {
      "epoch": 1.9551190858611354,
      "grad_norm": 0.4635816514492035,
      "learning_rate": 2.8060142358933716e-06,
      "loss": 0.3362,
      "step": 42604
    },
    {
      "epoch": 1.9551649763663899,
      "grad_norm": 0.44836628437042236,
      "learning_rate": 2.8057939160753235e-06,
      "loss": 0.287,
      "step": 42605
    },
    {
      "epoch": 1.9552108668716444,
      "grad_norm": 0.438023179769516,
      "learning_rate": 2.80557360153365e-06,
      "loss": 0.2924,
      "step": 42606
    },
    {
      "epoch": 1.9552567573768989,
      "grad_norm": 0.47951453924179077,
      "learning_rate": 2.8053532922688786e-06,
      "loss": 0.3809,
      "step": 42607
    },
    {
      "epoch": 1.9553026478821531,
      "grad_norm": 0.46520310640335083,
      "learning_rate": 2.805132988281539e-06,
      "loss": 0.3626,
      "step": 42608
    },
    {
      "epoch": 1.9553485383874076,
      "grad_norm": 0.42953282594680786,
      "learning_rate": 2.8049126895721633e-06,
      "loss": 0.2727,
      "step": 42609
    },
    {
      "epoch": 1.9553944288926621,
      "grad_norm": 0.46154528856277466,
      "learning_rate": 2.8046923961412795e-06,
      "loss": 0.3742,
      "step": 42610
    },
    {
      "epoch": 1.9554403193979166,
      "grad_norm": 0.4611739218235016,
      "learning_rate": 2.804472107989419e-06,
      "loss": 0.3808,
      "step": 42611
    },
    {
      "epoch": 1.955486209903171,
      "grad_norm": 0.45480746030807495,
      "learning_rate": 2.804251825117109e-06,
      "loss": 0.3637,
      "step": 42612
    },
    {
      "epoch": 1.9555321004084254,
      "grad_norm": 0.47020718455314636,
      "learning_rate": 2.8040315475248804e-06,
      "loss": 0.2885,
      "step": 42613
    },
    {
      "epoch": 1.95557799091368,
      "grad_norm": 0.45602288842201233,
      "learning_rate": 2.8038112752132614e-06,
      "loss": 0.3294,
      "step": 42614
    },
    {
      "epoch": 1.9556238814189344,
      "grad_norm": 0.4646095335483551,
      "learning_rate": 2.8035910081827843e-06,
      "loss": 0.3617,
      "step": 42615
    },
    {
      "epoch": 1.955669771924189,
      "grad_norm": 0.5059548020362854,
      "learning_rate": 2.8033707464339787e-06,
      "loss": 0.3588,
      "step": 42616
    },
    {
      "epoch": 1.9557156624294434,
      "grad_norm": 0.5013076663017273,
      "learning_rate": 2.80315048996737e-06,
      "loss": 0.3961,
      "step": 42617
    },
    {
      "epoch": 1.9557615529346979,
      "grad_norm": 0.4865446090698242,
      "learning_rate": 2.8029302387834946e-06,
      "loss": 0.3725,
      "step": 42618
    },
    {
      "epoch": 1.9558074434399524,
      "grad_norm": 0.4943789839744568,
      "learning_rate": 2.802709992882877e-06,
      "loss": 0.4118,
      "step": 42619
    },
    {
      "epoch": 1.9558533339452069,
      "grad_norm": 0.5054281949996948,
      "learning_rate": 2.8024897522660488e-06,
      "loss": 0.4323,
      "step": 42620
    },
    {
      "epoch": 1.9558992244504612,
      "grad_norm": 0.5302573442459106,
      "learning_rate": 2.8022695169335394e-06,
      "loss": 0.4339,
      "step": 42621
    },
    {
      "epoch": 1.9559451149557157,
      "grad_norm": 0.5048846006393433,
      "learning_rate": 2.8020492868858786e-06,
      "loss": 0.4098,
      "step": 42622
    },
    {
      "epoch": 1.9559910054609702,
      "grad_norm": 0.4763071537017822,
      "learning_rate": 2.8018290621235933e-06,
      "loss": 0.3609,
      "step": 42623
    },
    {
      "epoch": 1.9560368959662244,
      "grad_norm": 0.4737274646759033,
      "learning_rate": 2.8016088426472167e-06,
      "loss": 0.3575,
      "step": 42624
    },
    {
      "epoch": 1.956082786471479,
      "grad_norm": 0.4714599549770355,
      "learning_rate": 2.8013886284572777e-06,
      "loss": 0.4007,
      "step": 42625
    },
    {
      "epoch": 1.9561286769767334,
      "grad_norm": 0.4745331406593323,
      "learning_rate": 2.8011684195543033e-06,
      "loss": 0.3354,
      "step": 42626
    },
    {
      "epoch": 1.956174567481988,
      "grad_norm": 0.5430697798728943,
      "learning_rate": 2.800948215938826e-06,
      "loss": 0.4104,
      "step": 42627
    },
    {
      "epoch": 1.9562204579872424,
      "grad_norm": 0.4837821424007416,
      "learning_rate": 2.800728017611375e-06,
      "loss": 0.3509,
      "step": 42628
    },
    {
      "epoch": 1.956266348492497,
      "grad_norm": 0.4735223352909088,
      "learning_rate": 2.8005078245724784e-06,
      "loss": 0.3433,
      "step": 42629
    },
    {
      "epoch": 1.9563122389977514,
      "grad_norm": 0.46920254826545715,
      "learning_rate": 2.8002876368226652e-06,
      "loss": 0.3515,
      "step": 42630
    },
    {
      "epoch": 1.956358129503006,
      "grad_norm": 0.43987610936164856,
      "learning_rate": 2.8000674543624674e-06,
      "loss": 0.3203,
      "step": 42631
    },
    {
      "epoch": 1.9564040200082604,
      "grad_norm": 0.4764060080051422,
      "learning_rate": 2.799847277192412e-06,
      "loss": 0.357,
      "step": 42632
    },
    {
      "epoch": 1.956449910513515,
      "grad_norm": 0.4895569384098053,
      "learning_rate": 2.799627105313031e-06,
      "loss": 0.3528,
      "step": 42633
    },
    {
      "epoch": 1.9564958010187692,
      "grad_norm": 0.4754211902618408,
      "learning_rate": 2.7994069387248525e-06,
      "loss": 0.3545,
      "step": 42634
    },
    {
      "epoch": 1.9565416915240237,
      "grad_norm": 0.4861854016780853,
      "learning_rate": 2.7991867774284047e-06,
      "loss": 0.385,
      "step": 42635
    },
    {
      "epoch": 1.9565875820292782,
      "grad_norm": 0.4404165744781494,
      "learning_rate": 2.7989666214242167e-06,
      "loss": 0.3197,
      "step": 42636
    },
    {
      "epoch": 1.9566334725345325,
      "grad_norm": 0.48736467957496643,
      "learning_rate": 2.7987464707128216e-06,
      "loss": 0.3828,
      "step": 42637
    },
    {
      "epoch": 1.956679363039787,
      "grad_norm": 0.4658512473106384,
      "learning_rate": 2.798526325294746e-06,
      "loss": 0.3477,
      "step": 42638
    },
    {
      "epoch": 1.9567252535450415,
      "grad_norm": 0.46470773220062256,
      "learning_rate": 2.7983061851705177e-06,
      "loss": 0.3516,
      "step": 42639
    },
    {
      "epoch": 1.956771144050296,
      "grad_norm": 0.45387718081474304,
      "learning_rate": 2.7980860503406702e-06,
      "loss": 0.3181,
      "step": 42640
    },
    {
      "epoch": 1.9568170345555504,
      "grad_norm": 0.4341164827346802,
      "learning_rate": 2.797865920805731e-06,
      "loss": 0.3402,
      "step": 42641
    },
    {
      "epoch": 1.956862925060805,
      "grad_norm": 0.43490374088287354,
      "learning_rate": 2.7976457965662295e-06,
      "loss": 0.3148,
      "step": 42642
    },
    {
      "epoch": 1.9569088155660594,
      "grad_norm": 0.4563387930393219,
      "learning_rate": 2.7974256776226945e-06,
      "loss": 0.3503,
      "step": 42643
    },
    {
      "epoch": 1.956954706071314,
      "grad_norm": 0.45153698325157166,
      "learning_rate": 2.797205563975656e-06,
      "loss": 0.3418,
      "step": 42644
    },
    {
      "epoch": 1.9570005965765684,
      "grad_norm": 0.4886835813522339,
      "learning_rate": 2.7969854556256396e-06,
      "loss": 0.3655,
      "step": 42645
    },
    {
      "epoch": 1.9570464870818227,
      "grad_norm": 0.5136573910713196,
      "learning_rate": 2.796765352573181e-06,
      "loss": 0.4141,
      "step": 42646
    },
    {
      "epoch": 1.9570923775870772,
      "grad_norm": 0.455084890127182,
      "learning_rate": 2.7965452548188067e-06,
      "loss": 0.3544,
      "step": 42647
    },
    {
      "epoch": 1.9571382680923317,
      "grad_norm": 0.4364270567893982,
      "learning_rate": 2.796325162363043e-06,
      "loss": 0.3198,
      "step": 42648
    },
    {
      "epoch": 1.9571841585975862,
      "grad_norm": 0.4567607641220093,
      "learning_rate": 2.7961050752064233e-06,
      "loss": 0.368,
      "step": 42649
    },
    {
      "epoch": 1.9572300491028405,
      "grad_norm": 0.4805290699005127,
      "learning_rate": 2.795884993349476e-06,
      "loss": 0.3848,
      "step": 42650
    },
    {
      "epoch": 1.957275939608095,
      "grad_norm": 0.4760979115962982,
      "learning_rate": 2.7956649167927275e-06,
      "loss": 0.3482,
      "step": 42651
    },
    {
      "epoch": 1.9573218301133495,
      "grad_norm": 0.4655333459377289,
      "learning_rate": 2.7954448455367113e-06,
      "loss": 0.32,
      "step": 42652
    },
    {
      "epoch": 1.957367720618604,
      "grad_norm": 0.4388354420661926,
      "learning_rate": 2.795224779581954e-06,
      "loss": 0.3027,
      "step": 42653
    },
    {
      "epoch": 1.9574136111238585,
      "grad_norm": 0.47450101375579834,
      "learning_rate": 2.795004718928986e-06,
      "loss": 0.3658,
      "step": 42654
    },
    {
      "epoch": 1.957459501629113,
      "grad_norm": 0.5150549411773682,
      "learning_rate": 2.7947846635783353e-06,
      "loss": 0.4081,
      "step": 42655
    },
    {
      "epoch": 1.9575053921343675,
      "grad_norm": 0.4700637757778168,
      "learning_rate": 2.7945646135305316e-06,
      "loss": 0.3223,
      "step": 42656
    },
    {
      "epoch": 1.957551282639622,
      "grad_norm": 0.47039172053337097,
      "learning_rate": 2.7943445687861016e-06,
      "loss": 0.3644,
      "step": 42657
    },
    {
      "epoch": 1.9575971731448765,
      "grad_norm": 0.4821496307849884,
      "learning_rate": 2.7941245293455793e-06,
      "loss": 0.4421,
      "step": 42658
    },
    {
      "epoch": 1.9576430636501307,
      "grad_norm": 0.4409633278846741,
      "learning_rate": 2.793904495209491e-06,
      "loss": 0.3098,
      "step": 42659
    },
    {
      "epoch": 1.9576889541553852,
      "grad_norm": 0.41578003764152527,
      "learning_rate": 2.7936844663783648e-06,
      "loss": 0.2717,
      "step": 42660
    },
    {
      "epoch": 1.9577348446606397,
      "grad_norm": 0.5190191268920898,
      "learning_rate": 2.7934644428527326e-06,
      "loss": 0.4288,
      "step": 42661
    },
    {
      "epoch": 1.957780735165894,
      "grad_norm": 0.488781213760376,
      "learning_rate": 2.7932444246331216e-06,
      "loss": 0.3919,
      "step": 42662
    },
    {
      "epoch": 1.9578266256711485,
      "grad_norm": 0.47790324687957764,
      "learning_rate": 2.7930244117200618e-06,
      "loss": 0.3739,
      "step": 42663
    },
    {
      "epoch": 1.957872516176403,
      "grad_norm": 0.46007439494132996,
      "learning_rate": 2.792804404114081e-06,
      "loss": 0.328,
      "step": 42664
    },
    {
      "epoch": 1.9579184066816575,
      "grad_norm": 0.5023325085639954,
      "learning_rate": 2.79258440181571e-06,
      "loss": 0.3666,
      "step": 42665
    },
    {
      "epoch": 1.957964297186912,
      "grad_norm": 0.4422638714313507,
      "learning_rate": 2.792364404825476e-06,
      "loss": 0.3325,
      "step": 42666
    },
    {
      "epoch": 1.9580101876921665,
      "grad_norm": 0.5093100666999817,
      "learning_rate": 2.792144413143908e-06,
      "loss": 0.3658,
      "step": 42667
    },
    {
      "epoch": 1.958056078197421,
      "grad_norm": 0.5203491449356079,
      "learning_rate": 2.7919244267715367e-06,
      "loss": 0.4367,
      "step": 42668
    },
    {
      "epoch": 1.9581019687026755,
      "grad_norm": 0.4647260904312134,
      "learning_rate": 2.7917044457088906e-06,
      "loss": 0.3356,
      "step": 42669
    },
    {
      "epoch": 1.95814785920793,
      "grad_norm": 0.4616556763648987,
      "learning_rate": 2.791484469956497e-06,
      "loss": 0.3289,
      "step": 42670
    },
    {
      "epoch": 1.9581937497131845,
      "grad_norm": 0.5037557482719421,
      "learning_rate": 2.7912644995148875e-06,
      "loss": 0.3849,
      "step": 42671
    },
    {
      "epoch": 1.9582396402184388,
      "grad_norm": 0.48734956979751587,
      "learning_rate": 2.79104453438459e-06,
      "loss": 0.3505,
      "step": 42672
    },
    {
      "epoch": 1.9582855307236933,
      "grad_norm": 0.4402637779712677,
      "learning_rate": 2.790824574566131e-06,
      "loss": 0.3008,
      "step": 42673
    },
    {
      "epoch": 1.9583314212289478,
      "grad_norm": 0.45045006275177,
      "learning_rate": 2.790604620060044e-06,
      "loss": 0.315,
      "step": 42674
    },
    {
      "epoch": 1.958377311734202,
      "grad_norm": 0.5239521265029907,
      "learning_rate": 2.7903846708668553e-06,
      "loss": 0.3867,
      "step": 42675
    },
    {
      "epoch": 1.9584232022394565,
      "grad_norm": 0.49133506417274475,
      "learning_rate": 2.7901647269870946e-06,
      "loss": 0.3725,
      "step": 42676
    },
    {
      "epoch": 1.958469092744711,
      "grad_norm": 0.4734187424182892,
      "learning_rate": 2.789944788421289e-06,
      "loss": 0.3462,
      "step": 42677
    },
    {
      "epoch": 1.9585149832499655,
      "grad_norm": 0.4839710295200348,
      "learning_rate": 2.78972485516997e-06,
      "loss": 0.3342,
      "step": 42678
    },
    {
      "epoch": 1.95856087375522,
      "grad_norm": 0.4628952145576477,
      "learning_rate": 2.789504927233663e-06,
      "loss": 0.3446,
      "step": 42679
    },
    {
      "epoch": 1.9586067642604745,
      "grad_norm": 0.4872036278247833,
      "learning_rate": 2.7892850046129007e-06,
      "loss": 0.3317,
      "step": 42680
    },
    {
      "epoch": 1.958652654765729,
      "grad_norm": 0.46467864513397217,
      "learning_rate": 2.7890650873082105e-06,
      "loss": 0.3637,
      "step": 42681
    },
    {
      "epoch": 1.9586985452709835,
      "grad_norm": 0.44176945090293884,
      "learning_rate": 2.7888451753201183e-06,
      "loss": 0.3215,
      "step": 42682
    },
    {
      "epoch": 1.958744435776238,
      "grad_norm": 0.4156891703605652,
      "learning_rate": 2.7886252686491585e-06,
      "loss": 0.306,
      "step": 42683
    },
    {
      "epoch": 1.9587903262814925,
      "grad_norm": 0.4656137228012085,
      "learning_rate": 2.7884053672958567e-06,
      "loss": 0.3298,
      "step": 42684
    },
    {
      "epoch": 1.9588362167867468,
      "grad_norm": 0.4568449556827545,
      "learning_rate": 2.7881854712607394e-06,
      "loss": 0.3013,
      "step": 42685
    },
    {
      "epoch": 1.9588821072920013,
      "grad_norm": 0.5044740438461304,
      "learning_rate": 2.7879655805443427e-06,
      "loss": 0.3948,
      "step": 42686
    },
    {
      "epoch": 1.9589279977972558,
      "grad_norm": 0.4669477343559265,
      "learning_rate": 2.7877456951471883e-06,
      "loss": 0.3705,
      "step": 42687
    },
    {
      "epoch": 1.95897388830251,
      "grad_norm": 0.5089010000228882,
      "learning_rate": 2.787525815069805e-06,
      "loss": 0.4724,
      "step": 42688
    },
    {
      "epoch": 1.9590197788077646,
      "grad_norm": 0.5099437832832336,
      "learning_rate": 2.7873059403127266e-06,
      "loss": 0.3948,
      "step": 42689
    },
    {
      "epoch": 1.959065669313019,
      "grad_norm": 0.4738597869873047,
      "learning_rate": 2.7870860708764784e-06,
      "loss": 0.3779,
      "step": 42690
    },
    {
      "epoch": 1.9591115598182736,
      "grad_norm": 0.4569297134876251,
      "learning_rate": 2.786866206761588e-06,
      "loss": 0.3093,
      "step": 42691
    },
    {
      "epoch": 1.959157450323528,
      "grad_norm": 0.489706814289093,
      "learning_rate": 2.786646347968588e-06,
      "loss": 0.4058,
      "step": 42692
    },
    {
      "epoch": 1.9592033408287826,
      "grad_norm": 0.455205500125885,
      "learning_rate": 2.7864264944980044e-06,
      "loss": 0.3178,
      "step": 42693
    },
    {
      "epoch": 1.959249231334037,
      "grad_norm": 0.49059292674064636,
      "learning_rate": 2.7862066463503674e-06,
      "loss": 0.3892,
      "step": 42694
    },
    {
      "epoch": 1.9592951218392916,
      "grad_norm": 0.49572381377220154,
      "learning_rate": 2.785986803526202e-06,
      "loss": 0.321,
      "step": 42695
    },
    {
      "epoch": 1.959341012344546,
      "grad_norm": 0.4644756019115448,
      "learning_rate": 2.7857669660260423e-06,
      "loss": 0.3564,
      "step": 42696
    },
    {
      "epoch": 1.9593869028498003,
      "grad_norm": 0.42119333148002625,
      "learning_rate": 2.7855471338504135e-06,
      "loss": 0.2977,
      "step": 42697
    },
    {
      "epoch": 1.9594327933550548,
      "grad_norm": 0.48818761110305786,
      "learning_rate": 2.7853273069998453e-06,
      "loss": 0.3316,
      "step": 42698
    },
    {
      "epoch": 1.9594786838603093,
      "grad_norm": 0.49452564120292664,
      "learning_rate": 2.785107485474866e-06,
      "loss": 0.4417,
      "step": 42699
    },
    {
      "epoch": 1.9595245743655636,
      "grad_norm": 0.5207871794700623,
      "learning_rate": 2.7848876692760043e-06,
      "loss": 0.4352,
      "step": 42700
    },
    {
      "epoch": 1.959570464870818,
      "grad_norm": 0.44345036149024963,
      "learning_rate": 2.7846678584037856e-06,
      "loss": 0.3264,
      "step": 42701
    },
    {
      "epoch": 1.9596163553760726,
      "grad_norm": 0.479976624250412,
      "learning_rate": 2.7844480528587447e-06,
      "loss": 0.4187,
      "step": 42702
    },
    {
      "epoch": 1.959662245881327,
      "grad_norm": 0.46240389347076416,
      "learning_rate": 2.7842282526414068e-06,
      "loss": 0.3258,
      "step": 42703
    },
    {
      "epoch": 1.9597081363865816,
      "grad_norm": 0.46940183639526367,
      "learning_rate": 2.7840084577522987e-06,
      "loss": 0.3376,
      "step": 42704
    },
    {
      "epoch": 1.959754026891836,
      "grad_norm": 0.45884016156196594,
      "learning_rate": 2.7837886681919524e-06,
      "loss": 0.3408,
      "step": 42705
    },
    {
      "epoch": 1.9597999173970906,
      "grad_norm": 0.4864921569824219,
      "learning_rate": 2.7835688839608943e-06,
      "loss": 0.3988,
      "step": 42706
    },
    {
      "epoch": 1.959845807902345,
      "grad_norm": 0.47822505235671997,
      "learning_rate": 2.783349105059653e-06,
      "loss": 0.3263,
      "step": 42707
    },
    {
      "epoch": 1.9598916984075996,
      "grad_norm": 0.5204706192016602,
      "learning_rate": 2.7831293314887607e-06,
      "loss": 0.4288,
      "step": 42708
    },
    {
      "epoch": 1.959937588912854,
      "grad_norm": 0.4732123911380768,
      "learning_rate": 2.7829095632487403e-06,
      "loss": 0.3607,
      "step": 42709
    },
    {
      "epoch": 1.9599834794181084,
      "grad_norm": 0.5083494186401367,
      "learning_rate": 2.7826898003401213e-06,
      "loss": 0.4462,
      "step": 42710
    },
    {
      "epoch": 1.9600293699233629,
      "grad_norm": 0.4558075964450836,
      "learning_rate": 2.7824700427634355e-06,
      "loss": 0.3451,
      "step": 42711
    },
    {
      "epoch": 1.9600752604286173,
      "grad_norm": 0.4846417307853699,
      "learning_rate": 2.7822502905192095e-06,
      "loss": 0.3654,
      "step": 42712
    },
    {
      "epoch": 1.9601211509338716,
      "grad_norm": 0.4569319486618042,
      "learning_rate": 2.782030543607969e-06,
      "loss": 0.3349,
      "step": 42713
    },
    {
      "epoch": 1.9601670414391261,
      "grad_norm": 0.5422413349151611,
      "learning_rate": 2.7818108020302478e-06,
      "loss": 0.3994,
      "step": 42714
    },
    {
      "epoch": 1.9602129319443806,
      "grad_norm": 0.4837099611759186,
      "learning_rate": 2.7815910657865707e-06,
      "loss": 0.3697,
      "step": 42715
    },
    {
      "epoch": 1.9602588224496351,
      "grad_norm": 0.48875996470451355,
      "learning_rate": 2.7813713348774653e-06,
      "loss": 0.3743,
      "step": 42716
    },
    {
      "epoch": 1.9603047129548896,
      "grad_norm": 0.48086073994636536,
      "learning_rate": 2.7811516093034642e-06,
      "loss": 0.3826,
      "step": 42717
    },
    {
      "epoch": 1.9603506034601441,
      "grad_norm": 0.43515247106552124,
      "learning_rate": 2.7809318890650927e-06,
      "loss": 0.3076,
      "step": 42718
    },
    {
      "epoch": 1.9603964939653986,
      "grad_norm": 0.46085551381111145,
      "learning_rate": 2.7807121741628796e-06,
      "loss": 0.3831,
      "step": 42719
    },
    {
      "epoch": 1.960442384470653,
      "grad_norm": 0.4880446493625641,
      "learning_rate": 2.780492464597353e-06,
      "loss": 0.3786,
      "step": 42720
    },
    {
      "epoch": 1.9604882749759076,
      "grad_norm": 0.4760649502277374,
      "learning_rate": 2.7802727603690414e-06,
      "loss": 0.3692,
      "step": 42721
    },
    {
      "epoch": 1.960534165481162,
      "grad_norm": 0.5025593638420105,
      "learning_rate": 2.780053061478472e-06,
      "loss": 0.3374,
      "step": 42722
    },
    {
      "epoch": 1.9605800559864164,
      "grad_norm": 0.4647161662578583,
      "learning_rate": 2.7798333679261764e-06,
      "loss": 0.3398,
      "step": 42723
    },
    {
      "epoch": 1.9606259464916709,
      "grad_norm": 0.4864025115966797,
      "learning_rate": 2.7796136797126806e-06,
      "loss": 0.3886,
      "step": 42724
    },
    {
      "epoch": 1.9606718369969254,
      "grad_norm": 0.4592362344264984,
      "learning_rate": 2.779393996838511e-06,
      "loss": 0.3408,
      "step": 42725
    },
    {
      "epoch": 1.9607177275021797,
      "grad_norm": 0.43354520201683044,
      "learning_rate": 2.7791743193041997e-06,
      "loss": 0.2939,
      "step": 42726
    },
    {
      "epoch": 1.9607636180074342,
      "grad_norm": 0.51848965883255,
      "learning_rate": 2.7789546471102734e-06,
      "loss": 0.3389,
      "step": 42727
    },
    {
      "epoch": 1.9608095085126886,
      "grad_norm": 0.5233510732650757,
      "learning_rate": 2.7787349802572604e-06,
      "loss": 0.4447,
      "step": 42728
    },
    {
      "epoch": 1.9608553990179431,
      "grad_norm": 0.47242671251296997,
      "learning_rate": 2.778515318745687e-06,
      "loss": 0.3763,
      "step": 42729
    },
    {
      "epoch": 1.9609012895231976,
      "grad_norm": 0.47130510210990906,
      "learning_rate": 2.778295662576087e-06,
      "loss": 0.3413,
      "step": 42730
    },
    {
      "epoch": 1.9609471800284521,
      "grad_norm": 0.4930626451969147,
      "learning_rate": 2.7780760117489826e-06,
      "loss": 0.3509,
      "step": 42731
    },
    {
      "epoch": 1.9609930705337066,
      "grad_norm": 0.4682520627975464,
      "learning_rate": 2.7778563662649015e-06,
      "loss": 0.3274,
      "step": 42732
    },
    {
      "epoch": 1.9610389610389611,
      "grad_norm": 0.46139806509017944,
      "learning_rate": 2.7776367261243774e-06,
      "loss": 0.3482,
      "step": 42733
    },
    {
      "epoch": 1.9610848515442156,
      "grad_norm": 0.4676688611507416,
      "learning_rate": 2.7774170913279353e-06,
      "loss": 0.3373,
      "step": 42734
    },
    {
      "epoch": 1.96113074204947,
      "grad_norm": 0.46672171354293823,
      "learning_rate": 2.7771974618761017e-06,
      "loss": 0.3485,
      "step": 42735
    },
    {
      "epoch": 1.9611766325547244,
      "grad_norm": 0.48873549699783325,
      "learning_rate": 2.7769778377694085e-06,
      "loss": 0.3589,
      "step": 42736
    },
    {
      "epoch": 1.961222523059979,
      "grad_norm": 0.47027504444122314,
      "learning_rate": 2.776758219008382e-06,
      "loss": 0.3353,
      "step": 42737
    },
    {
      "epoch": 1.9612684135652334,
      "grad_norm": 0.4936540722846985,
      "learning_rate": 2.7765386055935484e-06,
      "loss": 0.3893,
      "step": 42738
    },
    {
      "epoch": 1.9613143040704877,
      "grad_norm": 0.44667407870292664,
      "learning_rate": 2.7763189975254397e-06,
      "loss": 0.3103,
      "step": 42739
    },
    {
      "epoch": 1.9613601945757422,
      "grad_norm": 0.47585251927375793,
      "learning_rate": 2.7760993948045823e-06,
      "loss": 0.3634,
      "step": 42740
    },
    {
      "epoch": 1.9614060850809967,
      "grad_norm": 0.44235143065452576,
      "learning_rate": 2.7758797974315034e-06,
      "loss": 0.322,
      "step": 42741
    },
    {
      "epoch": 1.9614519755862512,
      "grad_norm": 0.4876372516155243,
      "learning_rate": 2.7756602054067316e-06,
      "loss": 0.3582,
      "step": 42742
    },
    {
      "epoch": 1.9614978660915057,
      "grad_norm": 0.4664761424064636,
      "learning_rate": 2.7754406187307954e-06,
      "loss": 0.3807,
      "step": 42743
    },
    {
      "epoch": 1.9615437565967602,
      "grad_norm": 0.4331066310405731,
      "learning_rate": 2.77522103740422e-06,
      "loss": 0.2868,
      "step": 42744
    },
    {
      "epoch": 1.9615896471020147,
      "grad_norm": 0.4685346484184265,
      "learning_rate": 2.7750014614275384e-06,
      "loss": 0.3584,
      "step": 42745
    },
    {
      "epoch": 1.9616355376072692,
      "grad_norm": 0.5147796869277954,
      "learning_rate": 2.7747818908012756e-06,
      "loss": 0.4228,
      "step": 42746
    },
    {
      "epoch": 1.9616814281125237,
      "grad_norm": 0.48538294434547424,
      "learning_rate": 2.7745623255259584e-06,
      "loss": 0.3985,
      "step": 42747
    },
    {
      "epoch": 1.961727318617778,
      "grad_norm": 0.4577445089817047,
      "learning_rate": 2.7743427656021183e-06,
      "loss": 0.3598,
      "step": 42748
    },
    {
      "epoch": 1.9617732091230324,
      "grad_norm": 0.41643044352531433,
      "learning_rate": 2.7741232110302813e-06,
      "loss": 0.2679,
      "step": 42749
    },
    {
      "epoch": 1.961819099628287,
      "grad_norm": 0.47026699781417847,
      "learning_rate": 2.7739036618109738e-06,
      "loss": 0.432,
      "step": 42750
    },
    {
      "epoch": 1.9618649901335412,
      "grad_norm": 0.44650447368621826,
      "learning_rate": 2.7736841179447273e-06,
      "loss": 0.3051,
      "step": 42751
    },
    {
      "epoch": 1.9619108806387957,
      "grad_norm": 0.5170765519142151,
      "learning_rate": 2.773464579432069e-06,
      "loss": 0.4114,
      "step": 42752
    },
    {
      "epoch": 1.9619567711440502,
      "grad_norm": 0.46876320242881775,
      "learning_rate": 2.7732450462735227e-06,
      "loss": 0.3551,
      "step": 42753
    },
    {
      "epoch": 1.9620026616493047,
      "grad_norm": 0.5149762630462646,
      "learning_rate": 2.773025518469621e-06,
      "loss": 0.4218,
      "step": 42754
    },
    {
      "epoch": 1.9620485521545592,
      "grad_norm": 0.5108720064163208,
      "learning_rate": 2.7728059960208905e-06,
      "loss": 0.4239,
      "step": 42755
    },
    {
      "epoch": 1.9620944426598137,
      "grad_norm": 0.4551020562648773,
      "learning_rate": 2.7725864789278558e-06,
      "loss": 0.3172,
      "step": 42756
    },
    {
      "epoch": 1.9621403331650682,
      "grad_norm": 0.46515828371047974,
      "learning_rate": 2.7723669671910505e-06,
      "loss": 0.3478,
      "step": 42757
    },
    {
      "epoch": 1.9621862236703227,
      "grad_norm": 0.48717600107192993,
      "learning_rate": 2.772147460810999e-06,
      "loss": 0.3562,
      "step": 42758
    },
    {
      "epoch": 1.9622321141755772,
      "grad_norm": 0.47209182381629944,
      "learning_rate": 2.7719279597882298e-06,
      "loss": 0.3743,
      "step": 42759
    },
    {
      "epoch": 1.9622780046808317,
      "grad_norm": 0.49100419878959656,
      "learning_rate": 2.771708464123268e-06,
      "loss": 0.4104,
      "step": 42760
    },
    {
      "epoch": 1.962323895186086,
      "grad_norm": 0.4669548273086548,
      "learning_rate": 2.771488973816647e-06,
      "loss": 0.3572,
      "step": 42761
    },
    {
      "epoch": 1.9623697856913405,
      "grad_norm": 0.48151662945747375,
      "learning_rate": 2.771269488868892e-06,
      "loss": 0.3872,
      "step": 42762
    },
    {
      "epoch": 1.962415676196595,
      "grad_norm": 0.522543728351593,
      "learning_rate": 2.77105000928053e-06,
      "loss": 0.4845,
      "step": 42763
    },
    {
      "epoch": 1.9624615667018492,
      "grad_norm": 0.4768964946269989,
      "learning_rate": 2.7708305350520893e-06,
      "loss": 0.3893,
      "step": 42764
    },
    {
      "epoch": 1.9625074572071037,
      "grad_norm": 0.47656336426734924,
      "learning_rate": 2.770611066184097e-06,
      "loss": 0.3584,
      "step": 42765
    },
    {
      "epoch": 1.9625533477123582,
      "grad_norm": 0.44240084290504456,
      "learning_rate": 2.77039160267708e-06,
      "loss": 0.3341,
      "step": 42766
    },
    {
      "epoch": 1.9625992382176127,
      "grad_norm": 0.4717763662338257,
      "learning_rate": 2.77017214453157e-06,
      "loss": 0.3481,
      "step": 42767
    },
    {
      "epoch": 1.9626451287228672,
      "grad_norm": 0.45515933632850647,
      "learning_rate": 2.769952691748091e-06,
      "loss": 0.361,
      "step": 42768
    },
    {
      "epoch": 1.9626910192281217,
      "grad_norm": 0.4609382450580597,
      "learning_rate": 2.7697332443271705e-06,
      "loss": 0.3469,
      "step": 42769
    },
    {
      "epoch": 1.9627369097333762,
      "grad_norm": 0.43735918402671814,
      "learning_rate": 2.76951380226934e-06,
      "loss": 0.3001,
      "step": 42770
    },
    {
      "epoch": 1.9627828002386307,
      "grad_norm": 0.48054012656211853,
      "learning_rate": 2.769294365575125e-06,
      "loss": 0.3722,
      "step": 42771
    },
    {
      "epoch": 1.9628286907438852,
      "grad_norm": 0.46344009041786194,
      "learning_rate": 2.7690749342450507e-06,
      "loss": 0.3657,
      "step": 42772
    },
    {
      "epoch": 1.9628745812491397,
      "grad_norm": 0.4297401010990143,
      "learning_rate": 2.7688555082796486e-06,
      "loss": 0.2913,
      "step": 42773
    },
    {
      "epoch": 1.962920471754394,
      "grad_norm": 0.4592282474040985,
      "learning_rate": 2.7686360876794477e-06,
      "loss": 0.3479,
      "step": 42774
    },
    {
      "epoch": 1.9629663622596485,
      "grad_norm": 0.4723133146762848,
      "learning_rate": 2.7684166724449684e-06,
      "loss": 0.3449,
      "step": 42775
    },
    {
      "epoch": 1.963012252764903,
      "grad_norm": 0.45876574516296387,
      "learning_rate": 2.768197262576745e-06,
      "loss": 0.3234,
      "step": 42776
    },
    {
      "epoch": 1.9630581432701573,
      "grad_norm": 0.45564863085746765,
      "learning_rate": 2.7679778580753024e-06,
      "loss": 0.3264,
      "step": 42777
    },
    {
      "epoch": 1.9631040337754118,
      "grad_norm": 0.480070024728775,
      "learning_rate": 2.7677584589411665e-06,
      "loss": 0.3329,
      "step": 42778
    },
    {
      "epoch": 1.9631499242806663,
      "grad_norm": 0.5161702036857605,
      "learning_rate": 2.767539065174869e-06,
      "loss": 0.476,
      "step": 42779
    },
    {
      "epoch": 1.9631958147859208,
      "grad_norm": 0.43776044249534607,
      "learning_rate": 2.7673196767769363e-06,
      "loss": 0.287,
      "step": 42780
    },
    {
      "epoch": 1.9632417052911753,
      "grad_norm": 0.5098643898963928,
      "learning_rate": 2.767100293747892e-06,
      "loss": 0.4032,
      "step": 42781
    },
    {
      "epoch": 1.9632875957964298,
      "grad_norm": 0.4734732210636139,
      "learning_rate": 2.7668809160882692e-06,
      "loss": 0.3197,
      "step": 42782
    },
    {
      "epoch": 1.9633334863016842,
      "grad_norm": 0.44803106784820557,
      "learning_rate": 2.7666615437985933e-06,
      "loss": 0.3356,
      "step": 42783
    },
    {
      "epoch": 1.9633793768069387,
      "grad_norm": 0.4644460082054138,
      "learning_rate": 2.766442176879391e-06,
      "loss": 0.2991,
      "step": 42784
    },
    {
      "epoch": 1.9634252673121932,
      "grad_norm": 0.464834600687027,
      "learning_rate": 2.7662228153311905e-06,
      "loss": 0.3409,
      "step": 42785
    },
    {
      "epoch": 1.9634711578174475,
      "grad_norm": 0.41341957449913025,
      "learning_rate": 2.7660034591545194e-06,
      "loss": 0.2791,
      "step": 42786
    },
    {
      "epoch": 1.963517048322702,
      "grad_norm": 0.5017837285995483,
      "learning_rate": 2.7657841083499028e-06,
      "loss": 0.3983,
      "step": 42787
    },
    {
      "epoch": 1.9635629388279565,
      "grad_norm": 0.47702300548553467,
      "learning_rate": 2.7655647629178717e-06,
      "loss": 0.3523,
      "step": 42788
    },
    {
      "epoch": 1.9636088293332108,
      "grad_norm": 0.5161416530609131,
      "learning_rate": 2.7653454228589526e-06,
      "loss": 0.411,
      "step": 42789
    },
    {
      "epoch": 1.9636547198384653,
      "grad_norm": 0.479185551404953,
      "learning_rate": 2.7651260881736707e-06,
      "loss": 0.4007,
      "step": 42790
    },
    {
      "epoch": 1.9637006103437198,
      "grad_norm": 0.5051237344741821,
      "learning_rate": 2.764906758862557e-06,
      "loss": 0.3846,
      "step": 42791
    },
    {
      "epoch": 1.9637465008489743,
      "grad_norm": 0.4717367887496948,
      "learning_rate": 2.7646874349261377e-06,
      "loss": 0.3315,
      "step": 42792
    },
    {
      "epoch": 1.9637923913542288,
      "grad_norm": 0.4566437900066376,
      "learning_rate": 2.764468116364939e-06,
      "loss": 0.3276,
      "step": 42793
    },
    {
      "epoch": 1.9638382818594833,
      "grad_norm": 0.462356835603714,
      "learning_rate": 2.7642488031794865e-06,
      "loss": 0.3492,
      "step": 42794
    },
    {
      "epoch": 1.9638841723647378,
      "grad_norm": 0.4651295244693756,
      "learning_rate": 2.7640294953703127e-06,
      "loss": 0.3461,
      "step": 42795
    },
    {
      "epoch": 1.9639300628699923,
      "grad_norm": 0.4714713990688324,
      "learning_rate": 2.763810192937945e-06,
      "loss": 0.3934,
      "step": 42796
    },
    {
      "epoch": 1.9639759533752468,
      "grad_norm": 0.42005255818367004,
      "learning_rate": 2.763590895882903e-06,
      "loss": 0.25,
      "step": 42797
    },
    {
      "epoch": 1.9640218438805013,
      "grad_norm": 0.48238804936408997,
      "learning_rate": 2.7633716042057206e-06,
      "loss": 0.3822,
      "step": 42798
    },
    {
      "epoch": 1.9640677343857555,
      "grad_norm": 0.47966626286506653,
      "learning_rate": 2.7631523179069243e-06,
      "loss": 0.3967,
      "step": 42799
    },
    {
      "epoch": 1.96411362489101,
      "grad_norm": 0.46954798698425293,
      "learning_rate": 2.762933036987039e-06,
      "loss": 0.3814,
      "step": 42800
    },
    {
      "epoch": 1.9641595153962645,
      "grad_norm": 0.48840081691741943,
      "learning_rate": 2.762713761446595e-06,
      "loss": 0.3617,
      "step": 42801
    },
    {
      "epoch": 1.9642054059015188,
      "grad_norm": 0.48907679319381714,
      "learning_rate": 2.7624944912861187e-06,
      "loss": 0.3848,
      "step": 42802
    },
    {
      "epoch": 1.9642512964067733,
      "grad_norm": 0.4763927459716797,
      "learning_rate": 2.7622752265061347e-06,
      "loss": 0.3917,
      "step": 42803
    },
    {
      "epoch": 1.9642971869120278,
      "grad_norm": 0.5006308555603027,
      "learning_rate": 2.7620559671071745e-06,
      "loss": 0.3599,
      "step": 42804
    },
    {
      "epoch": 1.9643430774172823,
      "grad_norm": 0.467944860458374,
      "learning_rate": 2.761836713089764e-06,
      "loss": 0.3595,
      "step": 42805
    },
    {
      "epoch": 1.9643889679225368,
      "grad_norm": 0.4432183504104614,
      "learning_rate": 2.761617464454429e-06,
      "loss": 0.3268,
      "step": 42806
    },
    {
      "epoch": 1.9644348584277913,
      "grad_norm": 0.49462372064590454,
      "learning_rate": 2.7613982212016978e-06,
      "loss": 0.3359,
      "step": 42807
    },
    {
      "epoch": 1.9644807489330458,
      "grad_norm": 0.5255132913589478,
      "learning_rate": 2.761178983332098e-06,
      "loss": 0.4229,
      "step": 42808
    },
    {
      "epoch": 1.9645266394383003,
      "grad_norm": 0.4943297207355499,
      "learning_rate": 2.760959750846153e-06,
      "loss": 0.4114,
      "step": 42809
    },
    {
      "epoch": 1.9645725299435548,
      "grad_norm": 0.4755267798900604,
      "learning_rate": 2.760740523744396e-06,
      "loss": 0.4052,
      "step": 42810
    },
    {
      "epoch": 1.9646184204488093,
      "grad_norm": 0.4742956757545471,
      "learning_rate": 2.760521302027351e-06,
      "loss": 0.3656,
      "step": 42811
    },
    {
      "epoch": 1.9646643109540636,
      "grad_norm": 0.46458613872528076,
      "learning_rate": 2.7603020856955444e-06,
      "loss": 0.3412,
      "step": 42812
    },
    {
      "epoch": 1.964710201459318,
      "grad_norm": 0.46573612093925476,
      "learning_rate": 2.760082874749505e-06,
      "loss": 0.3145,
      "step": 42813
    },
    {
      "epoch": 1.9647560919645726,
      "grad_norm": 0.5091037750244141,
      "learning_rate": 2.7598636691897594e-06,
      "loss": 0.3938,
      "step": 42814
    },
    {
      "epoch": 1.9648019824698268,
      "grad_norm": 0.48971569538116455,
      "learning_rate": 2.759644469016834e-06,
      "loss": 0.3754,
      "step": 42815
    },
    {
      "epoch": 1.9648478729750813,
      "grad_norm": 0.4829268753528595,
      "learning_rate": 2.7594252742312576e-06,
      "loss": 0.3905,
      "step": 42816
    },
    {
      "epoch": 1.9648937634803358,
      "grad_norm": 0.5252099633216858,
      "learning_rate": 2.7592060848335565e-06,
      "loss": 0.4342,
      "step": 42817
    },
    {
      "epoch": 1.9649396539855903,
      "grad_norm": 0.4242521822452545,
      "learning_rate": 2.7589869008242576e-06,
      "loss": 0.3029,
      "step": 42818
    },
    {
      "epoch": 1.9649855444908448,
      "grad_norm": 0.43491846323013306,
      "learning_rate": 2.7587677222038878e-06,
      "loss": 0.2985,
      "step": 42819
    },
    {
      "epoch": 1.9650314349960993,
      "grad_norm": 0.5011933445930481,
      "learning_rate": 2.758548548972974e-06,
      "loss": 0.4069,
      "step": 42820
    },
    {
      "epoch": 1.9650773255013538,
      "grad_norm": 0.5023173689842224,
      "learning_rate": 2.7583293811320424e-06,
      "loss": 0.4158,
      "step": 42821
    },
    {
      "epoch": 1.9651232160066083,
      "grad_norm": 0.43884092569351196,
      "learning_rate": 2.758110218681622e-06,
      "loss": 0.3175,
      "step": 42822
    },
    {
      "epoch": 1.9651691065118628,
      "grad_norm": 0.49482211470603943,
      "learning_rate": 2.7578910616222397e-06,
      "loss": 0.3685,
      "step": 42823
    },
    {
      "epoch": 1.965214997017117,
      "grad_norm": 0.4836224913597107,
      "learning_rate": 2.7576719099544193e-06,
      "loss": 0.4032,
      "step": 42824
    },
    {
      "epoch": 1.9652608875223716,
      "grad_norm": 0.6609395146369934,
      "learning_rate": 2.757452763678693e-06,
      "loss": 0.4138,
      "step": 42825
    },
    {
      "epoch": 1.965306778027626,
      "grad_norm": 0.45624494552612305,
      "learning_rate": 2.7572336227955844e-06,
      "loss": 0.3417,
      "step": 42826
    },
    {
      "epoch": 1.9653526685328806,
      "grad_norm": 0.4524540901184082,
      "learning_rate": 2.7570144873056215e-06,
      "loss": 0.3471,
      "step": 42827
    },
    {
      "epoch": 1.9653985590381349,
      "grad_norm": 0.4653138518333435,
      "learning_rate": 2.7567953572093302e-06,
      "loss": 0.3496,
      "step": 42828
    },
    {
      "epoch": 1.9654444495433894,
      "grad_norm": 0.4477055072784424,
      "learning_rate": 2.7565762325072387e-06,
      "loss": 0.3338,
      "step": 42829
    },
    {
      "epoch": 1.9654903400486439,
      "grad_norm": 0.4896380305290222,
      "learning_rate": 2.7563571131998733e-06,
      "loss": 0.3647,
      "step": 42830
    },
    {
      "epoch": 1.9655362305538984,
      "grad_norm": 0.45300671458244324,
      "learning_rate": 2.7561379992877583e-06,
      "loss": 0.3167,
      "step": 42831
    },
    {
      "epoch": 1.9655821210591529,
      "grad_norm": 0.46967750787734985,
      "learning_rate": 2.7559188907714262e-06,
      "loss": 0.364,
      "step": 42832
    },
    {
      "epoch": 1.9656280115644074,
      "grad_norm": 0.4484383761882782,
      "learning_rate": 2.7556997876514004e-06,
      "loss": 0.3341,
      "step": 42833
    },
    {
      "epoch": 1.9656739020696619,
      "grad_norm": 0.4530675411224365,
      "learning_rate": 2.755480689928206e-06,
      "loss": 0.3555,
      "step": 42834
    },
    {
      "epoch": 1.9657197925749164,
      "grad_norm": 0.5037322044372559,
      "learning_rate": 2.7552615976023743e-06,
      "loss": 0.3818,
      "step": 42835
    },
    {
      "epoch": 1.9657656830801709,
      "grad_norm": 0.4686565101146698,
      "learning_rate": 2.7550425106744306e-06,
      "loss": 0.3738,
      "step": 42836
    },
    {
      "epoch": 1.9658115735854251,
      "grad_norm": 0.4718894958496094,
      "learning_rate": 2.754823429144898e-06,
      "loss": 0.3495,
      "step": 42837
    },
    {
      "epoch": 1.9658574640906796,
      "grad_norm": 0.46795952320098877,
      "learning_rate": 2.7546043530143096e-06,
      "loss": 0.3598,
      "step": 42838
    },
    {
      "epoch": 1.9659033545959341,
      "grad_norm": 0.4499022364616394,
      "learning_rate": 2.7543852822831885e-06,
      "loss": 0.2695,
      "step": 42839
    },
    {
      "epoch": 1.9659492451011884,
      "grad_norm": 0.42825740575790405,
      "learning_rate": 2.754166216952062e-06,
      "loss": 0.2969,
      "step": 42840
    },
    {
      "epoch": 1.965995135606443,
      "grad_norm": 0.485538125038147,
      "learning_rate": 2.753947157021457e-06,
      "loss": 0.384,
      "step": 42841
    },
    {
      "epoch": 1.9660410261116974,
      "grad_norm": 0.5101372599601746,
      "learning_rate": 2.7537281024919005e-06,
      "loss": 0.3958,
      "step": 42842
    },
    {
      "epoch": 1.966086916616952,
      "grad_norm": 0.46135130524635315,
      "learning_rate": 2.753509053363917e-06,
      "loss": 0.3506,
      "step": 42843
    },
    {
      "epoch": 1.9661328071222064,
      "grad_norm": 0.49424290657043457,
      "learning_rate": 2.7532900096380373e-06,
      "loss": 0.3897,
      "step": 42844
    },
    {
      "epoch": 1.966178697627461,
      "grad_norm": 0.4105677306652069,
      "learning_rate": 2.753070971314785e-06,
      "loss": 0.271,
      "step": 42845
    },
    {
      "epoch": 1.9662245881327154,
      "grad_norm": 0.4613577127456665,
      "learning_rate": 2.752851938394687e-06,
      "loss": 0.3429,
      "step": 42846
    },
    {
      "epoch": 1.9662704786379699,
      "grad_norm": 0.45415017008781433,
      "learning_rate": 2.7526329108782722e-06,
      "loss": 0.3163,
      "step": 42847
    },
    {
      "epoch": 1.9663163691432244,
      "grad_norm": 0.47726237773895264,
      "learning_rate": 2.752413888766067e-06,
      "loss": 0.3894,
      "step": 42848
    },
    {
      "epoch": 1.9663622596484789,
      "grad_norm": 0.4523850381374359,
      "learning_rate": 2.7521948720585957e-06,
      "loss": 0.3084,
      "step": 42849
    },
    {
      "epoch": 1.9664081501537332,
      "grad_norm": 0.4916176497936249,
      "learning_rate": 2.7519758607563875e-06,
      "loss": 0.4355,
      "step": 42850
    },
    {
      "epoch": 1.9664540406589877,
      "grad_norm": 0.4534570276737213,
      "learning_rate": 2.751756854859967e-06,
      "loss": 0.3532,
      "step": 42851
    },
    {
      "epoch": 1.9664999311642422,
      "grad_norm": 0.4663563370704651,
      "learning_rate": 2.75153785436986e-06,
      "loss": 0.3324,
      "step": 42852
    },
    {
      "epoch": 1.9665458216694964,
      "grad_norm": 0.5541428923606873,
      "learning_rate": 2.7513188592865963e-06,
      "loss": 0.3516,
      "step": 42853
    },
    {
      "epoch": 1.966591712174751,
      "grad_norm": 0.4353538751602173,
      "learning_rate": 2.7510998696107015e-06,
      "loss": 0.303,
      "step": 42854
    },
    {
      "epoch": 1.9666376026800054,
      "grad_norm": 0.44753772020339966,
      "learning_rate": 2.7508808853427e-06,
      "loss": 0.3383,
      "step": 42855
    },
    {
      "epoch": 1.96668349318526,
      "grad_norm": 0.4872537851333618,
      "learning_rate": 2.750661906483122e-06,
      "loss": 0.3553,
      "step": 42856
    },
    {
      "epoch": 1.9667293836905144,
      "grad_norm": 0.4808294475078583,
      "learning_rate": 2.7504429330324923e-06,
      "loss": 0.3567,
      "step": 42857
    },
    {
      "epoch": 1.966775274195769,
      "grad_norm": 0.43983739614486694,
      "learning_rate": 2.7502239649913365e-06,
      "loss": 0.356,
      "step": 42858
    },
    {
      "epoch": 1.9668211647010234,
      "grad_norm": 0.4941234588623047,
      "learning_rate": 2.750005002360181e-06,
      "loss": 0.3812,
      "step": 42859
    },
    {
      "epoch": 1.966867055206278,
      "grad_norm": 0.5301010012626648,
      "learning_rate": 2.7497860451395555e-06,
      "loss": 0.4144,
      "step": 42860
    },
    {
      "epoch": 1.9669129457115324,
      "grad_norm": 0.4641619324684143,
      "learning_rate": 2.7495670933299836e-06,
      "loss": 0.3789,
      "step": 42861
    },
    {
      "epoch": 1.966958836216787,
      "grad_norm": 0.4496115446090698,
      "learning_rate": 2.7493481469319925e-06,
      "loss": 0.314,
      "step": 42862
    },
    {
      "epoch": 1.9670047267220412,
      "grad_norm": 0.5010966062545776,
      "learning_rate": 2.7491292059461096e-06,
      "loss": 0.3819,
      "step": 42863
    },
    {
      "epoch": 1.9670506172272957,
      "grad_norm": 0.5159322023391724,
      "learning_rate": 2.74891027037286e-06,
      "loss": 0.4192,
      "step": 42864
    },
    {
      "epoch": 1.9670965077325502,
      "grad_norm": 0.4328669607639313,
      "learning_rate": 2.7486913402127687e-06,
      "loss": 0.2913,
      "step": 42865
    },
    {
      "epoch": 1.9671423982378045,
      "grad_norm": 0.49574124813079834,
      "learning_rate": 2.748472415466366e-06,
      "loss": 0.3802,
      "step": 42866
    },
    {
      "epoch": 1.967188288743059,
      "grad_norm": 0.4546925127506256,
      "learning_rate": 2.7482534961341767e-06,
      "loss": 0.3374,
      "step": 42867
    },
    {
      "epoch": 1.9672341792483135,
      "grad_norm": 0.46523427963256836,
      "learning_rate": 2.7480345822167253e-06,
      "loss": 0.3511,
      "step": 42868
    },
    {
      "epoch": 1.967280069753568,
      "grad_norm": 0.45414188504219055,
      "learning_rate": 2.7478156737145424e-06,
      "loss": 0.3448,
      "step": 42869
    },
    {
      "epoch": 1.9673259602588224,
      "grad_norm": 0.5064060091972351,
      "learning_rate": 2.7475967706281513e-06,
      "loss": 0.4747,
      "step": 42870
    },
    {
      "epoch": 1.967371850764077,
      "grad_norm": 0.46521270275115967,
      "learning_rate": 2.7473778729580793e-06,
      "loss": 0.3202,
      "step": 42871
    },
    {
      "epoch": 1.9674177412693314,
      "grad_norm": 0.4995707869529724,
      "learning_rate": 2.7471589807048525e-06,
      "loss": 0.452,
      "step": 42872
    },
    {
      "epoch": 1.967463631774586,
      "grad_norm": 0.46405601501464844,
      "learning_rate": 2.7469400938689973e-06,
      "loss": 0.3924,
      "step": 42873
    },
    {
      "epoch": 1.9675095222798404,
      "grad_norm": 0.4776924252510071,
      "learning_rate": 2.7467212124510385e-06,
      "loss": 0.3277,
      "step": 42874
    },
    {
      "epoch": 1.9675554127850947,
      "grad_norm": 0.43833082914352417,
      "learning_rate": 2.7465023364515053e-06,
      "loss": 0.2984,
      "step": 42875
    },
    {
      "epoch": 1.9676013032903492,
      "grad_norm": 0.46614187955856323,
      "learning_rate": 2.746283465870924e-06,
      "loss": 0.3685,
      "step": 42876
    },
    {
      "epoch": 1.9676471937956037,
      "grad_norm": 0.46277499198913574,
      "learning_rate": 2.7460646007098163e-06,
      "loss": 0.3438,
      "step": 42877
    },
    {
      "epoch": 1.967693084300858,
      "grad_norm": 0.5671554207801819,
      "learning_rate": 2.7458457409687147e-06,
      "loss": 0.3679,
      "step": 42878
    },
    {
      "epoch": 1.9677389748061125,
      "grad_norm": 0.46522852778434753,
      "learning_rate": 2.7456268866481423e-06,
      "loss": 0.3298,
      "step": 42879
    },
    {
      "epoch": 1.967784865311367,
      "grad_norm": 0.470097154378891,
      "learning_rate": 2.7454080377486236e-06,
      "loss": 0.3321,
      "step": 42880
    },
    {
      "epoch": 1.9678307558166215,
      "grad_norm": 0.4826377034187317,
      "learning_rate": 2.7451891942706893e-06,
      "loss": 0.3966,
      "step": 42881
    },
    {
      "epoch": 1.967876646321876,
      "grad_norm": 0.4593963623046875,
      "learning_rate": 2.7449703562148635e-06,
      "loss": 0.3086,
      "step": 42882
    },
    {
      "epoch": 1.9679225368271305,
      "grad_norm": 0.46453648805618286,
      "learning_rate": 2.744751523581672e-06,
      "loss": 0.3349,
      "step": 42883
    },
    {
      "epoch": 1.967968427332385,
      "grad_norm": 0.6076543927192688,
      "learning_rate": 2.7445326963716414e-06,
      "loss": 0.3173,
      "step": 42884
    },
    {
      "epoch": 1.9680143178376395,
      "grad_norm": 0.44654417037963867,
      "learning_rate": 2.7443138745852975e-06,
      "loss": 0.309,
      "step": 42885
    },
    {
      "epoch": 1.968060208342894,
      "grad_norm": 0.4610488712787628,
      "learning_rate": 2.744095058223165e-06,
      "loss": 0.3278,
      "step": 42886
    },
    {
      "epoch": 1.9681060988481485,
      "grad_norm": 0.48547354340553284,
      "learning_rate": 2.7438762472857745e-06,
      "loss": 0.3664,
      "step": 42887
    },
    {
      "epoch": 1.9681519893534027,
      "grad_norm": 0.47164011001586914,
      "learning_rate": 2.7436574417736487e-06,
      "loss": 0.3358,
      "step": 42888
    },
    {
      "epoch": 1.9681978798586572,
      "grad_norm": 0.47939860820770264,
      "learning_rate": 2.743438641687313e-06,
      "loss": 0.3181,
      "step": 42889
    },
    {
      "epoch": 1.9682437703639117,
      "grad_norm": 0.47419288754463196,
      "learning_rate": 2.7432198470272977e-06,
      "loss": 0.3313,
      "step": 42890
    },
    {
      "epoch": 1.968289660869166,
      "grad_norm": 0.47287169098854065,
      "learning_rate": 2.7430010577941257e-06,
      "loss": 0.3234,
      "step": 42891
    },
    {
      "epoch": 1.9683355513744205,
      "grad_norm": 0.4942079484462738,
      "learning_rate": 2.7427822739883237e-06,
      "loss": 0.3512,
      "step": 42892
    },
    {
      "epoch": 1.968381441879675,
      "grad_norm": 0.43587690591812134,
      "learning_rate": 2.7425634956104186e-06,
      "loss": 0.3047,
      "step": 42893
    },
    {
      "epoch": 1.9684273323849295,
      "grad_norm": 0.45775818824768066,
      "learning_rate": 2.742344722660936e-06,
      "loss": 0.3442,
      "step": 42894
    },
    {
      "epoch": 1.968473222890184,
      "grad_norm": 0.4763652980327606,
      "learning_rate": 2.7421259551404006e-06,
      "loss": 0.3419,
      "step": 42895
    },
    {
      "epoch": 1.9685191133954385,
      "grad_norm": 0.47356292605400085,
      "learning_rate": 2.7419071930493385e-06,
      "loss": 0.3449,
      "step": 42896
    },
    {
      "epoch": 1.968565003900693,
      "grad_norm": 0.4498305022716522,
      "learning_rate": 2.741688436388279e-06,
      "loss": 0.3236,
      "step": 42897
    },
    {
      "epoch": 1.9686108944059475,
      "grad_norm": 0.4654752314090729,
      "learning_rate": 2.7414696851577455e-06,
      "loss": 0.3656,
      "step": 42898
    },
    {
      "epoch": 1.968656784911202,
      "grad_norm": 0.43499255180358887,
      "learning_rate": 2.741250939358263e-06,
      "loss": 0.3197,
      "step": 42899
    },
    {
      "epoch": 1.9687026754164565,
      "grad_norm": 0.438260942697525,
      "learning_rate": 2.741032198990361e-06,
      "loss": 0.2696,
      "step": 42900
    },
    {
      "epoch": 1.9687485659217108,
      "grad_norm": 0.4765600562095642,
      "learning_rate": 2.7408134640545635e-06,
      "loss": 0.3606,
      "step": 42901
    },
    {
      "epoch": 1.9687944564269653,
      "grad_norm": 0.5291645526885986,
      "learning_rate": 2.740594734551395e-06,
      "loss": 0.4426,
      "step": 42902
    },
    {
      "epoch": 1.9688403469322198,
      "grad_norm": 0.47119608521461487,
      "learning_rate": 2.740376010481385e-06,
      "loss": 0.3586,
      "step": 42903
    },
    {
      "epoch": 1.968886237437474,
      "grad_norm": 0.49991872906684875,
      "learning_rate": 2.740157291845057e-06,
      "loss": 0.4441,
      "step": 42904
    },
    {
      "epoch": 1.9689321279427285,
      "grad_norm": 0.4392249882221222,
      "learning_rate": 2.7399385786429376e-06,
      "loss": 0.3302,
      "step": 42905
    },
    {
      "epoch": 1.968978018447983,
      "grad_norm": 0.4530896544456482,
      "learning_rate": 2.7397198708755535e-06,
      "loss": 0.3295,
      "step": 42906
    },
    {
      "epoch": 1.9690239089532375,
      "grad_norm": 0.42862504720687866,
      "learning_rate": 2.739501168543429e-06,
      "loss": 0.2902,
      "step": 42907
    },
    {
      "epoch": 1.969069799458492,
      "grad_norm": 0.4913938045501709,
      "learning_rate": 2.739282471647089e-06,
      "loss": 0.3829,
      "step": 42908
    },
    {
      "epoch": 1.9691156899637465,
      "grad_norm": 0.45172184705734253,
      "learning_rate": 2.739063780187063e-06,
      "loss": 0.3301,
      "step": 42909
    },
    {
      "epoch": 1.969161580469001,
      "grad_norm": 0.4399804472923279,
      "learning_rate": 2.738845094163875e-06,
      "loss": 0.3239,
      "step": 42910
    },
    {
      "epoch": 1.9692074709742555,
      "grad_norm": 0.43593159317970276,
      "learning_rate": 2.7386264135780487e-06,
      "loss": 0.3192,
      "step": 42911
    },
    {
      "epoch": 1.96925336147951,
      "grad_norm": 0.47051000595092773,
      "learning_rate": 2.7384077384301142e-06,
      "loss": 0.3845,
      "step": 42912
    },
    {
      "epoch": 1.9692992519847643,
      "grad_norm": 0.5013925433158875,
      "learning_rate": 2.738189068720596e-06,
      "loss": 0.3423,
      "step": 42913
    },
    {
      "epoch": 1.9693451424900188,
      "grad_norm": 0.4304320514202118,
      "learning_rate": 2.737970404450019e-06,
      "loss": 0.274,
      "step": 42914
    },
    {
      "epoch": 1.9693910329952733,
      "grad_norm": 0.437936007976532,
      "learning_rate": 2.737751745618909e-06,
      "loss": 0.3231,
      "step": 42915
    },
    {
      "epoch": 1.9694369235005278,
      "grad_norm": 0.48506858944892883,
      "learning_rate": 2.737533092227792e-06,
      "loss": 0.3503,
      "step": 42916
    },
    {
      "epoch": 1.969482814005782,
      "grad_norm": 0.474704772233963,
      "learning_rate": 2.737314444277192e-06,
      "loss": 0.4048,
      "step": 42917
    },
    {
      "epoch": 1.9695287045110366,
      "grad_norm": 0.4491488039493561,
      "learning_rate": 2.7370958017676384e-06,
      "loss": 0.3358,
      "step": 42918
    },
    {
      "epoch": 1.969574595016291,
      "grad_norm": 0.4897958040237427,
      "learning_rate": 2.7368771646996557e-06,
      "loss": 0.3503,
      "step": 42919
    },
    {
      "epoch": 1.9696204855215456,
      "grad_norm": 0.49196600914001465,
      "learning_rate": 2.736658533073767e-06,
      "loss": 0.3963,
      "step": 42920
    },
    {
      "epoch": 1.9696663760268,
      "grad_norm": 0.4512460231781006,
      "learning_rate": 2.736439906890502e-06,
      "loss": 0.3083,
      "step": 42921
    },
    {
      "epoch": 1.9697122665320546,
      "grad_norm": 0.4670305848121643,
      "learning_rate": 2.7362212861503844e-06,
      "loss": 0.392,
      "step": 42922
    },
    {
      "epoch": 1.969758157037309,
      "grad_norm": 0.47963711619377136,
      "learning_rate": 2.7360026708539404e-06,
      "loss": 0.3647,
      "step": 42923
    },
    {
      "epoch": 1.9698040475425636,
      "grad_norm": 0.4632793068885803,
      "learning_rate": 2.735784061001694e-06,
      "loss": 0.2974,
      "step": 42924
    },
    {
      "epoch": 1.969849938047818,
      "grad_norm": 0.5326076745986938,
      "learning_rate": 2.7355654565941735e-06,
      "loss": 0.426,
      "step": 42925
    },
    {
      "epoch": 1.9698958285530723,
      "grad_norm": 0.48221635818481445,
      "learning_rate": 2.735346857631904e-06,
      "loss": 0.3606,
      "step": 42926
    },
    {
      "epoch": 1.9699417190583268,
      "grad_norm": 0.46529513597488403,
      "learning_rate": 2.73512826411541e-06,
      "loss": 0.3289,
      "step": 42927
    },
    {
      "epoch": 1.9699876095635813,
      "grad_norm": 0.5107858777046204,
      "learning_rate": 2.734909676045218e-06,
      "loss": 0.3612,
      "step": 42928
    },
    {
      "epoch": 1.9700335000688356,
      "grad_norm": 0.4577659070491791,
      "learning_rate": 2.734691093421853e-06,
      "loss": 0.3398,
      "step": 42929
    },
    {
      "epoch": 1.97007939057409,
      "grad_norm": 0.4828402101993561,
      "learning_rate": 2.7344725162458395e-06,
      "loss": 0.3761,
      "step": 42930
    },
    {
      "epoch": 1.9701252810793446,
      "grad_norm": 0.5176158547401428,
      "learning_rate": 2.734253944517706e-06,
      "loss": 0.4263,
      "step": 42931
    },
    {
      "epoch": 1.970171171584599,
      "grad_norm": 0.47431305050849915,
      "learning_rate": 2.7340353782379775e-06,
      "loss": 0.3715,
      "step": 42932
    },
    {
      "epoch": 1.9702170620898536,
      "grad_norm": 0.46877244114875793,
      "learning_rate": 2.7338168174071767e-06,
      "loss": 0.3507,
      "step": 42933
    },
    {
      "epoch": 1.970262952595108,
      "grad_norm": 0.46306392550468445,
      "learning_rate": 2.7335982620258326e-06,
      "loss": 0.3462,
      "step": 42934
    },
    {
      "epoch": 1.9703088431003626,
      "grad_norm": 0.47579166293144226,
      "learning_rate": 2.733379712094469e-06,
      "loss": 0.352,
      "step": 42935
    },
    {
      "epoch": 1.970354733605617,
      "grad_norm": 0.4748784601688385,
      "learning_rate": 2.733161167613613e-06,
      "loss": 0.3949,
      "step": 42936
    },
    {
      "epoch": 1.9704006241108716,
      "grad_norm": 0.45802444219589233,
      "learning_rate": 2.732942628583788e-06,
      "loss": 0.3198,
      "step": 42937
    },
    {
      "epoch": 1.970446514616126,
      "grad_norm": 0.4631478190422058,
      "learning_rate": 2.7327240950055213e-06,
      "loss": 0.3727,
      "step": 42938
    },
    {
      "epoch": 1.9704924051213804,
      "grad_norm": 0.48245882987976074,
      "learning_rate": 2.7325055668793347e-06,
      "loss": 0.3832,
      "step": 42939
    },
    {
      "epoch": 1.9705382956266349,
      "grad_norm": 0.46733686327934265,
      "learning_rate": 2.7322870442057592e-06,
      "loss": 0.3511,
      "step": 42940
    },
    {
      "epoch": 1.9705841861318893,
      "grad_norm": 0.45125386118888855,
      "learning_rate": 2.7320685269853175e-06,
      "loss": 0.3365,
      "step": 42941
    },
    {
      "epoch": 1.9706300766371436,
      "grad_norm": 0.44264450669288635,
      "learning_rate": 2.7318500152185333e-06,
      "loss": 0.3036,
      "step": 42942
    },
    {
      "epoch": 1.9706759671423981,
      "grad_norm": 0.4715102016925812,
      "learning_rate": 2.731631508905936e-06,
      "loss": 0.3491,
      "step": 42943
    },
    {
      "epoch": 1.9707218576476526,
      "grad_norm": 0.4583599865436554,
      "learning_rate": 2.7314130080480483e-06,
      "loss": 0.3461,
      "step": 42944
    },
    {
      "epoch": 1.9707677481529071,
      "grad_norm": 0.47679781913757324,
      "learning_rate": 2.7311945126453953e-06,
      "loss": 0.3365,
      "step": 42945
    },
    {
      "epoch": 1.9708136386581616,
      "grad_norm": 0.4098743796348572,
      "learning_rate": 2.730976022698505e-06,
      "loss": 0.2691,
      "step": 42946
    },
    {
      "epoch": 1.9708595291634161,
      "grad_norm": 0.4181537628173828,
      "learning_rate": 2.730757538207901e-06,
      "loss": 0.2778,
      "step": 42947
    },
    {
      "epoch": 1.9709054196686706,
      "grad_norm": 0.39751628041267395,
      "learning_rate": 2.730539059174109e-06,
      "loss": 0.2417,
      "step": 42948
    },
    {
      "epoch": 1.970951310173925,
      "grad_norm": 0.48492783308029175,
      "learning_rate": 2.7303205855976545e-06,
      "loss": 0.4413,
      "step": 42949
    },
    {
      "epoch": 1.9709972006791796,
      "grad_norm": 0.47156772017478943,
      "learning_rate": 2.730102117479063e-06,
      "loss": 0.3594,
      "step": 42950
    },
    {
      "epoch": 1.971043091184434,
      "grad_norm": 0.4805081784725189,
      "learning_rate": 2.729883654818857e-06,
      "loss": 0.3463,
      "step": 42951
    },
    {
      "epoch": 1.9710889816896884,
      "grad_norm": 0.4399474561214447,
      "learning_rate": 2.729665197617567e-06,
      "loss": 0.3554,
      "step": 42952
    },
    {
      "epoch": 1.9711348721949429,
      "grad_norm": 0.48627540469169617,
      "learning_rate": 2.7294467458757145e-06,
      "loss": 0.3887,
      "step": 42953
    },
    {
      "epoch": 1.9711807627001974,
      "grad_norm": 0.48422372341156006,
      "learning_rate": 2.7292282995938246e-06,
      "loss": 0.3976,
      "step": 42954
    },
    {
      "epoch": 1.9712266532054517,
      "grad_norm": 0.492933064699173,
      "learning_rate": 2.7290098587724263e-06,
      "loss": 0.3084,
      "step": 42955
    },
    {
      "epoch": 1.9712725437107061,
      "grad_norm": 0.48610132932662964,
      "learning_rate": 2.728791423412042e-06,
      "loss": 0.3933,
      "step": 42956
    },
    {
      "epoch": 1.9713184342159606,
      "grad_norm": 0.4766220152378082,
      "learning_rate": 2.7285729935131976e-06,
      "loss": 0.3733,
      "step": 42957
    },
    {
      "epoch": 1.9713643247212151,
      "grad_norm": 0.46368440985679626,
      "learning_rate": 2.728354569076419e-06,
      "loss": 0.3816,
      "step": 42958
    },
    {
      "epoch": 1.9714102152264696,
      "grad_norm": 0.4775179922580719,
      "learning_rate": 2.7281361501022295e-06,
      "loss": 0.37,
      "step": 42959
    },
    {
      "epoch": 1.9714561057317241,
      "grad_norm": 0.4159829616546631,
      "learning_rate": 2.7279177365911567e-06,
      "loss": 0.2667,
      "step": 42960
    },
    {
      "epoch": 1.9715019962369786,
      "grad_norm": 0.47893813252449036,
      "learning_rate": 2.727699328543722e-06,
      "loss": 0.3197,
      "step": 42961
    },
    {
      "epoch": 1.9715478867422331,
      "grad_norm": 0.43894073367118835,
      "learning_rate": 2.727480925960455e-06,
      "loss": 0.3245,
      "step": 42962
    },
    {
      "epoch": 1.9715937772474876,
      "grad_norm": 0.47644418478012085,
      "learning_rate": 2.7272625288418797e-06,
      "loss": 0.3621,
      "step": 42963
    },
    {
      "epoch": 1.971639667752742,
      "grad_norm": 0.4449101984500885,
      "learning_rate": 2.727044137188519e-06,
      "loss": 0.2853,
      "step": 42964
    },
    {
      "epoch": 1.9716855582579964,
      "grad_norm": 0.4864620566368103,
      "learning_rate": 2.726825751000901e-06,
      "loss": 0.3847,
      "step": 42965
    },
    {
      "epoch": 1.971731448763251,
      "grad_norm": 0.4494321942329407,
      "learning_rate": 2.72660737027955e-06,
      "loss": 0.283,
      "step": 42966
    },
    {
      "epoch": 1.9717773392685052,
      "grad_norm": 0.43498343229293823,
      "learning_rate": 2.726388995024989e-06,
      "loss": 0.2821,
      "step": 42967
    },
    {
      "epoch": 1.9718232297737597,
      "grad_norm": 0.48384878039360046,
      "learning_rate": 2.7261706252377464e-06,
      "loss": 0.3454,
      "step": 42968
    },
    {
      "epoch": 1.9718691202790142,
      "grad_norm": 0.479285329580307,
      "learning_rate": 2.7259522609183458e-06,
      "loss": 0.4084,
      "step": 42969
    },
    {
      "epoch": 1.9719150107842687,
      "grad_norm": 0.4740636944770813,
      "learning_rate": 2.7257339020673124e-06,
      "loss": 0.3324,
      "step": 42970
    },
    {
      "epoch": 1.9719609012895232,
      "grad_norm": 0.5093536972999573,
      "learning_rate": 2.7255155486851714e-06,
      "loss": 0.4067,
      "step": 42971
    },
    {
      "epoch": 1.9720067917947777,
      "grad_norm": 0.4903455376625061,
      "learning_rate": 2.7252972007724474e-06,
      "loss": 0.3337,
      "step": 42972
    },
    {
      "epoch": 1.9720526823000322,
      "grad_norm": 0.47710734605789185,
      "learning_rate": 2.725078858329664e-06,
      "loss": 0.37,
      "step": 42973
    },
    {
      "epoch": 1.9720985728052867,
      "grad_norm": 0.5705099701881409,
      "learning_rate": 2.72486052135735e-06,
      "loss": 0.3685,
      "step": 42974
    },
    {
      "epoch": 1.9721444633105412,
      "grad_norm": 0.4722226858139038,
      "learning_rate": 2.724642189856028e-06,
      "loss": 0.3603,
      "step": 42975
    },
    {
      "epoch": 1.9721903538157957,
      "grad_norm": 0.4722960889339447,
      "learning_rate": 2.724423863826222e-06,
      "loss": 0.3548,
      "step": 42976
    },
    {
      "epoch": 1.97223624432105,
      "grad_norm": 0.42023125290870667,
      "learning_rate": 2.72420554326846e-06,
      "loss": 0.3054,
      "step": 42977
    },
    {
      "epoch": 1.9722821348263044,
      "grad_norm": 0.4182262122631073,
      "learning_rate": 2.7239872281832657e-06,
      "loss": 0.2749,
      "step": 42978
    },
    {
      "epoch": 1.972328025331559,
      "grad_norm": 0.4726836085319519,
      "learning_rate": 2.7237689185711634e-06,
      "loss": 0.326,
      "step": 42979
    },
    {
      "epoch": 1.9723739158368132,
      "grad_norm": 0.4718392491340637,
      "learning_rate": 2.7235506144326783e-06,
      "loss": 0.3154,
      "step": 42980
    },
    {
      "epoch": 1.9724198063420677,
      "grad_norm": 0.4731011688709259,
      "learning_rate": 2.7233323157683362e-06,
      "loss": 0.3356,
      "step": 42981
    },
    {
      "epoch": 1.9724656968473222,
      "grad_norm": 0.44952255487442017,
      "learning_rate": 2.723114022578659e-06,
      "loss": 0.3298,
      "step": 42982
    },
    {
      "epoch": 1.9725115873525767,
      "grad_norm": 0.45575496554374695,
      "learning_rate": 2.722895734864176e-06,
      "loss": 0.3662,
      "step": 42983
    },
    {
      "epoch": 1.9725574778578312,
      "grad_norm": 0.48486408591270447,
      "learning_rate": 2.7226774526254095e-06,
      "loss": 0.3429,
      "step": 42984
    },
    {
      "epoch": 1.9726033683630857,
      "grad_norm": 0.474436491727829,
      "learning_rate": 2.722459175862884e-06,
      "loss": 0.367,
      "step": 42985
    },
    {
      "epoch": 1.9726492588683402,
      "grad_norm": 0.4444190561771393,
      "learning_rate": 2.7222409045771264e-06,
      "loss": 0.2982,
      "step": 42986
    },
    {
      "epoch": 1.9726951493735947,
      "grad_norm": 0.4446990191936493,
      "learning_rate": 2.722022638768661e-06,
      "loss": 0.3114,
      "step": 42987
    },
    {
      "epoch": 1.9727410398788492,
      "grad_norm": 0.5016521215438843,
      "learning_rate": 2.721804378438012e-06,
      "loss": 0.441,
      "step": 42988
    },
    {
      "epoch": 1.9727869303841037,
      "grad_norm": 0.5104778409004211,
      "learning_rate": 2.7215861235857035e-06,
      "loss": 0.4052,
      "step": 42989
    },
    {
      "epoch": 1.972832820889358,
      "grad_norm": 0.5073155164718628,
      "learning_rate": 2.721367874212262e-06,
      "loss": 0.3933,
      "step": 42990
    },
    {
      "epoch": 1.9728787113946125,
      "grad_norm": 0.49366697669029236,
      "learning_rate": 2.7211496303182116e-06,
      "loss": 0.4149,
      "step": 42991
    },
    {
      "epoch": 1.972924601899867,
      "grad_norm": 0.4453880488872528,
      "learning_rate": 2.720931391904078e-06,
      "loss": 0.3166,
      "step": 42992
    },
    {
      "epoch": 1.9729704924051212,
      "grad_norm": 0.4908974766731262,
      "learning_rate": 2.7207131589703844e-06,
      "loss": 0.3363,
      "step": 42993
    },
    {
      "epoch": 1.9730163829103757,
      "grad_norm": 0.4505852460861206,
      "learning_rate": 2.7204949315176565e-06,
      "loss": 0.3236,
      "step": 42994
    },
    {
      "epoch": 1.9730622734156302,
      "grad_norm": 0.49948233366012573,
      "learning_rate": 2.7202767095464173e-06,
      "loss": 0.3717,
      "step": 42995
    },
    {
      "epoch": 1.9731081639208847,
      "grad_norm": 0.5156455636024475,
      "learning_rate": 2.7200584930571947e-06,
      "loss": 0.4265,
      "step": 42996
    },
    {
      "epoch": 1.9731540544261392,
      "grad_norm": 0.4576736092567444,
      "learning_rate": 2.719840282050512e-06,
      "loss": 0.3472,
      "step": 42997
    },
    {
      "epoch": 1.9731999449313937,
      "grad_norm": 0.46018362045288086,
      "learning_rate": 2.719622076526891e-06,
      "loss": 0.3723,
      "step": 42998
    },
    {
      "epoch": 1.9732458354366482,
      "grad_norm": 0.463645339012146,
      "learning_rate": 2.7194038764868625e-06,
      "loss": 0.3167,
      "step": 42999
    },
    {
      "epoch": 1.9732917259419027,
      "grad_norm": 0.4825528562068939,
      "learning_rate": 2.719185681930947e-06,
      "loss": 0.3627,
      "step": 43000
    },
    {
      "epoch": 1.9733376164471572,
      "grad_norm": 0.48279833793640137,
      "learning_rate": 2.7189674928596706e-06,
      "loss": 0.3464,
      "step": 43001
    },
    {
      "epoch": 1.9733835069524115,
      "grad_norm": 0.4674633741378784,
      "learning_rate": 2.7187493092735573e-06,
      "loss": 0.3635,
      "step": 43002
    },
    {
      "epoch": 1.973429397457666,
      "grad_norm": 0.4723011553287506,
      "learning_rate": 2.7185311311731317e-06,
      "loss": 0.3763,
      "step": 43003
    },
    {
      "epoch": 1.9734752879629205,
      "grad_norm": 0.4689598083496094,
      "learning_rate": 2.7183129585589165e-06,
      "loss": 0.3596,
      "step": 43004
    },
    {
      "epoch": 1.973521178468175,
      "grad_norm": 0.47768914699554443,
      "learning_rate": 2.718094791431441e-06,
      "loss": 0.3544,
      "step": 43005
    },
    {
      "epoch": 1.9735670689734293,
      "grad_norm": 0.48436489701271057,
      "learning_rate": 2.7178766297912275e-06,
      "loss": 0.4014,
      "step": 43006
    },
    {
      "epoch": 1.9736129594786838,
      "grad_norm": 0.45127004384994507,
      "learning_rate": 2.717658473638798e-06,
      "loss": 0.3176,
      "step": 43007
    },
    {
      "epoch": 1.9736588499839383,
      "grad_norm": 0.4875739514827728,
      "learning_rate": 2.7174403229746816e-06,
      "loss": 0.3697,
      "step": 43008
    },
    {
      "epoch": 1.9737047404891928,
      "grad_norm": 0.4774787127971649,
      "learning_rate": 2.7172221777994008e-06,
      "loss": 0.3856,
      "step": 43009
    },
    {
      "epoch": 1.9737506309944473,
      "grad_norm": 0.5168051719665527,
      "learning_rate": 2.7170040381134783e-06,
      "loss": 0.4139,
      "step": 43010
    },
    {
      "epoch": 1.9737965214997018,
      "grad_norm": 0.4452592432498932,
      "learning_rate": 2.7167859039174428e-06,
      "loss": 0.3949,
      "step": 43011
    },
    {
      "epoch": 1.9738424120049562,
      "grad_norm": 0.48069578409194946,
      "learning_rate": 2.716567775211816e-06,
      "loss": 0.3517,
      "step": 43012
    },
    {
      "epoch": 1.9738883025102107,
      "grad_norm": 0.43767228722572327,
      "learning_rate": 2.7163496519971233e-06,
      "loss": 0.333,
      "step": 43013
    },
    {
      "epoch": 1.9739341930154652,
      "grad_norm": 0.4627382159233093,
      "learning_rate": 2.7161315342738893e-06,
      "loss": 0.3437,
      "step": 43014
    },
    {
      "epoch": 1.9739800835207195,
      "grad_norm": 0.47605419158935547,
      "learning_rate": 2.7159134220426374e-06,
      "loss": 0.3661,
      "step": 43015
    },
    {
      "epoch": 1.974025974025974,
      "grad_norm": 0.46920841932296753,
      "learning_rate": 2.715695315303891e-06,
      "loss": 0.3251,
      "step": 43016
    },
    {
      "epoch": 1.9740718645312285,
      "grad_norm": 0.4719802141189575,
      "learning_rate": 2.715477214058179e-06,
      "loss": 0.3757,
      "step": 43017
    },
    {
      "epoch": 1.9741177550364828,
      "grad_norm": 0.4033859968185425,
      "learning_rate": 2.7152591183060226e-06,
      "loss": 0.2626,
      "step": 43018
    },
    {
      "epoch": 1.9741636455417373,
      "grad_norm": 0.45449164509773254,
      "learning_rate": 2.7150410280479463e-06,
      "loss": 0.3578,
      "step": 43019
    },
    {
      "epoch": 1.9742095360469918,
      "grad_norm": 0.46683114767074585,
      "learning_rate": 2.7148229432844757e-06,
      "loss": 0.3584,
      "step": 43020
    },
    {
      "epoch": 1.9742554265522463,
      "grad_norm": 0.43749991059303284,
      "learning_rate": 2.7146048640161356e-06,
      "loss": 0.3088,
      "step": 43021
    },
    {
      "epoch": 1.9743013170575008,
      "grad_norm": 0.45934537053108215,
      "learning_rate": 2.714386790243449e-06,
      "loss": 0.3608,
      "step": 43022
    },
    {
      "epoch": 1.9743472075627553,
      "grad_norm": 0.44710272550582886,
      "learning_rate": 2.714168721966941e-06,
      "loss": 0.3077,
      "step": 43023
    },
    {
      "epoch": 1.9743930980680098,
      "grad_norm": 0.4379536509513855,
      "learning_rate": 2.7139506591871354e-06,
      "loss": 0.2865,
      "step": 43024
    },
    {
      "epoch": 1.9744389885732643,
      "grad_norm": 0.47337642312049866,
      "learning_rate": 2.7137326019045574e-06,
      "loss": 0.3406,
      "step": 43025
    },
    {
      "epoch": 1.9744848790785188,
      "grad_norm": 0.45020273327827454,
      "learning_rate": 2.713514550119729e-06,
      "loss": 0.3261,
      "step": 43026
    },
    {
      "epoch": 1.9745307695837733,
      "grad_norm": 0.523016631603241,
      "learning_rate": 2.713296503833178e-06,
      "loss": 0.4053,
      "step": 43027
    },
    {
      "epoch": 1.9745766600890275,
      "grad_norm": 0.49811258912086487,
      "learning_rate": 2.7130784630454275e-06,
      "loss": 0.3577,
      "step": 43028
    },
    {
      "epoch": 1.974622550594282,
      "grad_norm": 0.4756976366043091,
      "learning_rate": 2.7128604277569997e-06,
      "loss": 0.3617,
      "step": 43029
    },
    {
      "epoch": 1.9746684410995365,
      "grad_norm": 0.4516562223434448,
      "learning_rate": 2.7126423979684225e-06,
      "loss": 0.3506,
      "step": 43030
    },
    {
      "epoch": 1.9747143316047908,
      "grad_norm": 0.4506928324699402,
      "learning_rate": 2.7124243736802186e-06,
      "loss": 0.3374,
      "step": 43031
    },
    {
      "epoch": 1.9747602221100453,
      "grad_norm": 0.45910584926605225,
      "learning_rate": 2.71220635489291e-06,
      "loss": 0.3467,
      "step": 43032
    },
    {
      "epoch": 1.9748061126152998,
      "grad_norm": 0.455471932888031,
      "learning_rate": 2.711988341607025e-06,
      "loss": 0.3224,
      "step": 43033
    },
    {
      "epoch": 1.9748520031205543,
      "grad_norm": 0.5064855217933655,
      "learning_rate": 2.711770333823086e-06,
      "loss": 0.4672,
      "step": 43034
    },
    {
      "epoch": 1.9748978936258088,
      "grad_norm": 0.48381954431533813,
      "learning_rate": 2.711552331541617e-06,
      "loss": 0.3946,
      "step": 43035
    },
    {
      "epoch": 1.9749437841310633,
      "grad_norm": 0.4522872865200043,
      "learning_rate": 2.711334334763143e-06,
      "loss": 0.3506,
      "step": 43036
    },
    {
      "epoch": 1.9749896746363178,
      "grad_norm": 0.480379194021225,
      "learning_rate": 2.7111163434881864e-06,
      "loss": 0.3453,
      "step": 43037
    },
    {
      "epoch": 1.9750355651415723,
      "grad_norm": 0.4447477459907532,
      "learning_rate": 2.7108983577172713e-06,
      "loss": 0.3484,
      "step": 43038
    },
    {
      "epoch": 1.9750814556468268,
      "grad_norm": 0.4796711206436157,
      "learning_rate": 2.7106803774509256e-06,
      "loss": 0.4088,
      "step": 43039
    },
    {
      "epoch": 1.9751273461520813,
      "grad_norm": 0.4892890453338623,
      "learning_rate": 2.710462402689671e-06,
      "loss": 0.3471,
      "step": 43040
    },
    {
      "epoch": 1.9751732366573356,
      "grad_norm": 0.4526127576828003,
      "learning_rate": 2.710244433434029e-06,
      "loss": 0.3097,
      "step": 43041
    },
    {
      "epoch": 1.97521912716259,
      "grad_norm": 0.48445919156074524,
      "learning_rate": 2.710026469684529e-06,
      "loss": 0.3915,
      "step": 43042
    },
    {
      "epoch": 1.9752650176678446,
      "grad_norm": 0.4925227463245392,
      "learning_rate": 2.709808511441692e-06,
      "loss": 0.3689,
      "step": 43043
    },
    {
      "epoch": 1.9753109081730988,
      "grad_norm": 0.45780596137046814,
      "learning_rate": 2.709590558706043e-06,
      "loss": 0.3546,
      "step": 43044
    },
    {
      "epoch": 1.9753567986783533,
      "grad_norm": 0.44601908326148987,
      "learning_rate": 2.709372611478106e-06,
      "loss": 0.3202,
      "step": 43045
    },
    {
      "epoch": 1.9754026891836078,
      "grad_norm": 0.48589542508125305,
      "learning_rate": 2.7091546697584054e-06,
      "loss": 0.4023,
      "step": 43046
    },
    {
      "epoch": 1.9754485796888623,
      "grad_norm": 0.4720264673233032,
      "learning_rate": 2.7089367335474626e-06,
      "loss": 0.3725,
      "step": 43047
    },
    {
      "epoch": 1.9754944701941168,
      "grad_norm": 0.4534439444541931,
      "learning_rate": 2.7087188028458054e-06,
      "loss": 0.3412,
      "step": 43048
    },
    {
      "epoch": 1.9755403606993713,
      "grad_norm": 0.5026981234550476,
      "learning_rate": 2.7085008776539565e-06,
      "loss": 0.3938,
      "step": 43049
    },
    {
      "epoch": 1.9755862512046258,
      "grad_norm": 0.44207507371902466,
      "learning_rate": 2.7082829579724377e-06,
      "loss": 0.3202,
      "step": 43050
    },
    {
      "epoch": 1.9756321417098803,
      "grad_norm": 0.4496201276779175,
      "learning_rate": 2.7080650438017776e-06,
      "loss": 0.3181,
      "step": 43051
    },
    {
      "epoch": 1.9756780322151348,
      "grad_norm": 0.4767274260520935,
      "learning_rate": 2.7078471351424973e-06,
      "loss": 0.374,
      "step": 43052
    },
    {
      "epoch": 1.975723922720389,
      "grad_norm": 0.45770418643951416,
      "learning_rate": 2.7076292319951193e-06,
      "loss": 0.326,
      "step": 43053
    },
    {
      "epoch": 1.9757698132256436,
      "grad_norm": 0.4409182667732239,
      "learning_rate": 2.7074113343601715e-06,
      "loss": 0.304,
      "step": 43054
    },
    {
      "epoch": 1.975815703730898,
      "grad_norm": 0.5399067997932434,
      "learning_rate": 2.7071934422381753e-06,
      "loss": 0.4476,
      "step": 43055
    },
    {
      "epoch": 1.9758615942361524,
      "grad_norm": 0.43848690390586853,
      "learning_rate": 2.7069755556296564e-06,
      "loss": 0.2867,
      "step": 43056
    },
    {
      "epoch": 1.9759074847414069,
      "grad_norm": 0.474045991897583,
      "learning_rate": 2.7067576745351364e-06,
      "loss": 0.3528,
      "step": 43057
    },
    {
      "epoch": 1.9759533752466614,
      "grad_norm": 0.48616617918014526,
      "learning_rate": 2.706539798955141e-06,
      "loss": 0.3843,
      "step": 43058
    },
    {
      "epoch": 1.9759992657519159,
      "grad_norm": 0.4606216847896576,
      "learning_rate": 2.7063219288901937e-06,
      "loss": 0.3286,
      "step": 43059
    },
    {
      "epoch": 1.9760451562571704,
      "grad_norm": 0.46810126304626465,
      "learning_rate": 2.706104064340816e-06,
      "loss": 0.3659,
      "step": 43060
    },
    {
      "epoch": 1.9760910467624249,
      "grad_norm": 0.48562416434288025,
      "learning_rate": 2.705886205307536e-06,
      "loss": 0.3875,
      "step": 43061
    },
    {
      "epoch": 1.9761369372676794,
      "grad_norm": 0.545442521572113,
      "learning_rate": 2.705668351790876e-06,
      "loss": 0.4408,
      "step": 43062
    },
    {
      "epoch": 1.9761828277729339,
      "grad_norm": 0.5049591064453125,
      "learning_rate": 2.705450503791357e-06,
      "loss": 0.3809,
      "step": 43063
    },
    {
      "epoch": 1.9762287182781884,
      "grad_norm": 0.5137231945991516,
      "learning_rate": 2.705232661309508e-06,
      "loss": 0.4193,
      "step": 43064
    },
    {
      "epoch": 1.9762746087834429,
      "grad_norm": 0.46165013313293457,
      "learning_rate": 2.7050148243458494e-06,
      "loss": 0.3786,
      "step": 43065
    },
    {
      "epoch": 1.9763204992886971,
      "grad_norm": 0.482311874628067,
      "learning_rate": 2.7047969929009064e-06,
      "loss": 0.3431,
      "step": 43066
    },
    {
      "epoch": 1.9763663897939516,
      "grad_norm": 0.48097366094589233,
      "learning_rate": 2.704579166975202e-06,
      "loss": 0.3733,
      "step": 43067
    },
    {
      "epoch": 1.9764122802992061,
      "grad_norm": 0.4603255093097687,
      "learning_rate": 2.704361346569261e-06,
      "loss": 0.3277,
      "step": 43068
    },
    {
      "epoch": 1.9764581708044604,
      "grad_norm": 0.46200206875801086,
      "learning_rate": 2.7041435316836036e-06,
      "loss": 0.3442,
      "step": 43069
    },
    {
      "epoch": 1.976504061309715,
      "grad_norm": 0.46408936381340027,
      "learning_rate": 2.703925722318759e-06,
      "loss": 0.3385,
      "step": 43070
    },
    {
      "epoch": 1.9765499518149694,
      "grad_norm": 0.496676504611969,
      "learning_rate": 2.7037079184752475e-06,
      "loss": 0.3602,
      "step": 43071
    },
    {
      "epoch": 1.976595842320224,
      "grad_norm": 0.4971490204334259,
      "learning_rate": 2.7034901201535923e-06,
      "loss": 0.4083,
      "step": 43072
    },
    {
      "epoch": 1.9766417328254784,
      "grad_norm": 0.4566163718700409,
      "learning_rate": 2.7032723273543204e-06,
      "loss": 0.3546,
      "step": 43073
    },
    {
      "epoch": 1.976687623330733,
      "grad_norm": 0.47319772839546204,
      "learning_rate": 2.703054540077954e-06,
      "loss": 0.3579,
      "step": 43074
    },
    {
      "epoch": 1.9767335138359874,
      "grad_norm": 0.4660305380821228,
      "learning_rate": 2.7028367583250137e-06,
      "loss": 0.3598,
      "step": 43075
    },
    {
      "epoch": 1.9767794043412419,
      "grad_norm": 0.4800879657268524,
      "learning_rate": 2.702618982096029e-06,
      "loss": 0.3909,
      "step": 43076
    },
    {
      "epoch": 1.9768252948464964,
      "grad_norm": 0.4666239619255066,
      "learning_rate": 2.70240121139152e-06,
      "loss": 0.3835,
      "step": 43077
    },
    {
      "epoch": 1.9768711853517509,
      "grad_norm": 0.470390260219574,
      "learning_rate": 2.702183446212011e-06,
      "loss": 0.3607,
      "step": 43078
    },
    {
      "epoch": 1.9769170758570052,
      "grad_norm": 0.485868901014328,
      "learning_rate": 2.7019656865580257e-06,
      "loss": 0.3878,
      "step": 43079
    },
    {
      "epoch": 1.9769629663622597,
      "grad_norm": 0.45943811535835266,
      "learning_rate": 2.701747932430088e-06,
      "loss": 0.3867,
      "step": 43080
    },
    {
      "epoch": 1.9770088568675142,
      "grad_norm": 0.4532267153263092,
      "learning_rate": 2.701530183828718e-06,
      "loss": 0.3523,
      "step": 43081
    },
    {
      "epoch": 1.9770547473727684,
      "grad_norm": 0.47276532649993896,
      "learning_rate": 2.7013124407544456e-06,
      "loss": 0.3539,
      "step": 43082
    },
    {
      "epoch": 1.977100637878023,
      "grad_norm": 0.49072468280792236,
      "learning_rate": 2.701094703207791e-06,
      "loss": 0.3891,
      "step": 43083
    },
    {
      "epoch": 1.9771465283832774,
      "grad_norm": 0.47355905175209045,
      "learning_rate": 2.7008769711892757e-06,
      "loss": 0.3709,
      "step": 43084
    },
    {
      "epoch": 1.977192418888532,
      "grad_norm": 0.42839449644088745,
      "learning_rate": 2.7006592446994284e-06,
      "loss": 0.3004,
      "step": 43085
    },
    {
      "epoch": 1.9772383093937864,
      "grad_norm": 0.47325605154037476,
      "learning_rate": 2.7004415237387694e-06,
      "loss": 0.3378,
      "step": 43086
    },
    {
      "epoch": 1.977284199899041,
      "grad_norm": 0.44909465312957764,
      "learning_rate": 2.700223808307823e-06,
      "loss": 0.3098,
      "step": 43087
    },
    {
      "epoch": 1.9773300904042954,
      "grad_norm": 0.5121114253997803,
      "learning_rate": 2.7000060984071123e-06,
      "loss": 0.4155,
      "step": 43088
    },
    {
      "epoch": 1.97737598090955,
      "grad_norm": 0.49131855368614197,
      "learning_rate": 2.699788394037161e-06,
      "loss": 0.4021,
      "step": 43089
    },
    {
      "epoch": 1.9774218714148044,
      "grad_norm": 0.508849024772644,
      "learning_rate": 2.699570695198493e-06,
      "loss": 0.445,
      "step": 43090
    },
    {
      "epoch": 1.9774677619200587,
      "grad_norm": 0.47688090801239014,
      "learning_rate": 2.6993530018916293e-06,
      "loss": 0.3947,
      "step": 43091
    },
    {
      "epoch": 1.9775136524253132,
      "grad_norm": 0.5060880780220032,
      "learning_rate": 2.6991353141170973e-06,
      "loss": 0.3623,
      "step": 43092
    },
    {
      "epoch": 1.9775595429305677,
      "grad_norm": 0.4545553922653198,
      "learning_rate": 2.698917631875419e-06,
      "loss": 0.3358,
      "step": 43093
    },
    {
      "epoch": 1.9776054334358222,
      "grad_norm": 0.3883197009563446,
      "learning_rate": 2.6986999551671154e-06,
      "loss": 0.2521,
      "step": 43094
    },
    {
      "epoch": 1.9776513239410765,
      "grad_norm": 0.4842115044593811,
      "learning_rate": 2.698482283992714e-06,
      "loss": 0.4033,
      "step": 43095
    },
    {
      "epoch": 1.977697214446331,
      "grad_norm": 0.46912774443626404,
      "learning_rate": 2.698264618352736e-06,
      "loss": 0.3419,
      "step": 43096
    },
    {
      "epoch": 1.9777431049515855,
      "grad_norm": 0.5099304914474487,
      "learning_rate": 2.6980469582477043e-06,
      "loss": 0.4316,
      "step": 43097
    },
    {
      "epoch": 1.97778899545684,
      "grad_norm": 0.47993943095207214,
      "learning_rate": 2.6978293036781442e-06,
      "loss": 0.3801,
      "step": 43098
    },
    {
      "epoch": 1.9778348859620944,
      "grad_norm": 0.48221877217292786,
      "learning_rate": 2.6976116546445785e-06,
      "loss": 0.359,
      "step": 43099
    },
    {
      "epoch": 1.977880776467349,
      "grad_norm": 0.4791015386581421,
      "learning_rate": 2.6973940111475296e-06,
      "loss": 0.354,
      "step": 43100
    },
    {
      "epoch": 1.9779266669726034,
      "grad_norm": 0.44510969519615173,
      "learning_rate": 2.6971763731875216e-06,
      "loss": 0.3102,
      "step": 43101
    },
    {
      "epoch": 1.977972557477858,
      "grad_norm": 0.5093168616294861,
      "learning_rate": 2.696958740765078e-06,
      "loss": 0.4484,
      "step": 43102
    },
    {
      "epoch": 1.9780184479831124,
      "grad_norm": 0.5319295525550842,
      "learning_rate": 2.6967411138807193e-06,
      "loss": 0.4433,
      "step": 43103
    },
    {
      "epoch": 1.9780643384883667,
      "grad_norm": 0.5002061724662781,
      "learning_rate": 2.6965234925349738e-06,
      "loss": 0.4322,
      "step": 43104
    },
    {
      "epoch": 1.9781102289936212,
      "grad_norm": 0.4628775119781494,
      "learning_rate": 2.696305876728361e-06,
      "loss": 0.3644,
      "step": 43105
    },
    {
      "epoch": 1.9781561194988757,
      "grad_norm": 0.479069322347641,
      "learning_rate": 2.696088266461405e-06,
      "loss": 0.3566,
      "step": 43106
    },
    {
      "epoch": 1.97820201000413,
      "grad_norm": 0.5325402021408081,
      "learning_rate": 2.695870661734631e-06,
      "loss": 0.277,
      "step": 43107
    },
    {
      "epoch": 1.9782479005093845,
      "grad_norm": 0.5054181814193726,
      "learning_rate": 2.695653062548561e-06,
      "loss": 0.3988,
      "step": 43108
    },
    {
      "epoch": 1.978293791014639,
      "grad_norm": 0.48243290185928345,
      "learning_rate": 2.695435468903716e-06,
      "loss": 0.3419,
      "step": 43109
    },
    {
      "epoch": 1.9783396815198935,
      "grad_norm": 0.4701647460460663,
      "learning_rate": 2.6952178808006254e-06,
      "loss": 0.3804,
      "step": 43110
    },
    {
      "epoch": 1.978385572025148,
      "grad_norm": 0.49498775601387024,
      "learning_rate": 2.695000298239806e-06,
      "loss": 0.3896,
      "step": 43111
    },
    {
      "epoch": 1.9784314625304025,
      "grad_norm": 0.45390596985816956,
      "learning_rate": 2.694782721221782e-06,
      "loss": 0.3411,
      "step": 43112
    },
    {
      "epoch": 1.978477353035657,
      "grad_norm": 0.4760518968105316,
      "learning_rate": 2.6945651497470798e-06,
      "loss": 0.3457,
      "step": 43113
    },
    {
      "epoch": 1.9785232435409115,
      "grad_norm": 0.44144967198371887,
      "learning_rate": 2.69434758381622e-06,
      "loss": 0.2997,
      "step": 43114
    },
    {
      "epoch": 1.978569134046166,
      "grad_norm": 0.4646770656108856,
      "learning_rate": 2.6941300234297264e-06,
      "loss": 0.33,
      "step": 43115
    },
    {
      "epoch": 1.9786150245514205,
      "grad_norm": 0.43195241689682007,
      "learning_rate": 2.693912468588123e-06,
      "loss": 0.3174,
      "step": 43116
    },
    {
      "epoch": 1.9786609150566747,
      "grad_norm": 0.5053424835205078,
      "learning_rate": 2.6936949192919325e-06,
      "loss": 0.4097,
      "step": 43117
    },
    {
      "epoch": 1.9787068055619292,
      "grad_norm": 0.5600436925888062,
      "learning_rate": 2.6934773755416754e-06,
      "loss": 0.3597,
      "step": 43118
    },
    {
      "epoch": 1.9787526960671837,
      "grad_norm": 0.444132536649704,
      "learning_rate": 2.6932598373378806e-06,
      "loss": 0.3278,
      "step": 43119
    },
    {
      "epoch": 1.978798586572438,
      "grad_norm": 0.5015686750411987,
      "learning_rate": 2.6930423046810666e-06,
      "loss": 0.4451,
      "step": 43120
    },
    {
      "epoch": 1.9788444770776925,
      "grad_norm": 0.49007469415664673,
      "learning_rate": 2.692824777571758e-06,
      "loss": 0.3948,
      "step": 43121
    },
    {
      "epoch": 1.978890367582947,
      "grad_norm": 0.46476617455482483,
      "learning_rate": 2.6926072560104777e-06,
      "loss": 0.3532,
      "step": 43122
    },
    {
      "epoch": 1.9789362580882015,
      "grad_norm": 0.4301423728466034,
      "learning_rate": 2.6923897399977493e-06,
      "loss": 0.2713,
      "step": 43123
    },
    {
      "epoch": 1.978982148593456,
      "grad_norm": 0.46674251556396484,
      "learning_rate": 2.6921722295340947e-06,
      "loss": 0.3618,
      "step": 43124
    },
    {
      "epoch": 1.9790280390987105,
      "grad_norm": 0.43386775255203247,
      "learning_rate": 2.6919547246200356e-06,
      "loss": 0.3063,
      "step": 43125
    },
    {
      "epoch": 1.979073929603965,
      "grad_norm": 0.46225476264953613,
      "learning_rate": 2.691737225256099e-06,
      "loss": 0.3519,
      "step": 43126
    },
    {
      "epoch": 1.9791198201092195,
      "grad_norm": 0.4568866789340973,
      "learning_rate": 2.6915197314428056e-06,
      "loss": 0.3511,
      "step": 43127
    },
    {
      "epoch": 1.979165710614474,
      "grad_norm": 0.46512269973754883,
      "learning_rate": 2.691302243180677e-06,
      "loss": 0.3464,
      "step": 43128
    },
    {
      "epoch": 1.9792116011197285,
      "grad_norm": 0.5248969793319702,
      "learning_rate": 2.6910847604702394e-06,
      "loss": 0.3279,
      "step": 43129
    },
    {
      "epoch": 1.9792574916249828,
      "grad_norm": 0.486942857503891,
      "learning_rate": 2.6908672833120147e-06,
      "loss": 0.3774,
      "step": 43130
    },
    {
      "epoch": 1.9793033821302373,
      "grad_norm": 0.49245813488960266,
      "learning_rate": 2.690649811706524e-06,
      "loss": 0.4404,
      "step": 43131
    },
    {
      "epoch": 1.9793492726354918,
      "grad_norm": 0.47092217206954956,
      "learning_rate": 2.690432345654295e-06,
      "loss": 0.3289,
      "step": 43132
    },
    {
      "epoch": 1.979395163140746,
      "grad_norm": 0.4704309105873108,
      "learning_rate": 2.6902148851558445e-06,
      "loss": 0.3758,
      "step": 43133
    },
    {
      "epoch": 1.9794410536460005,
      "grad_norm": 0.44915661215782166,
      "learning_rate": 2.689997430211697e-06,
      "loss": 0.3137,
      "step": 43134
    },
    {
      "epoch": 1.979486944151255,
      "grad_norm": 0.5048395395278931,
      "learning_rate": 2.689779980822379e-06,
      "loss": 0.4015,
      "step": 43135
    },
    {
      "epoch": 1.9795328346565095,
      "grad_norm": 0.46840277314186096,
      "learning_rate": 2.6895625369884105e-06,
      "loss": 0.3318,
      "step": 43136
    },
    {
      "epoch": 1.979578725161764,
      "grad_norm": 0.47447413206100464,
      "learning_rate": 2.689345098710313e-06,
      "loss": 0.3639,
      "step": 43137
    },
    {
      "epoch": 1.9796246156670185,
      "grad_norm": 0.46133333444595337,
      "learning_rate": 2.689127665988613e-06,
      "loss": 0.3545,
      "step": 43138
    },
    {
      "epoch": 1.979670506172273,
      "grad_norm": 0.4870116412639618,
      "learning_rate": 2.688910238823832e-06,
      "loss": 0.3671,
      "step": 43139
    },
    {
      "epoch": 1.9797163966775275,
      "grad_norm": 0.5232671499252319,
      "learning_rate": 2.6886928172164905e-06,
      "loss": 0.3599,
      "step": 43140
    },
    {
      "epoch": 1.979762287182782,
      "grad_norm": 0.5012676119804382,
      "learning_rate": 2.6884754011671147e-06,
      "loss": 0.3696,
      "step": 43141
    },
    {
      "epoch": 1.9798081776880363,
      "grad_norm": 0.4681413471698761,
      "learning_rate": 2.6882579906762262e-06,
      "loss": 0.3335,
      "step": 43142
    },
    {
      "epoch": 1.9798540681932908,
      "grad_norm": 0.47126173973083496,
      "learning_rate": 2.688040585744348e-06,
      "loss": 0.3586,
      "step": 43143
    },
    {
      "epoch": 1.9798999586985453,
      "grad_norm": 0.4498625099658966,
      "learning_rate": 2.687823186372003e-06,
      "loss": 0.3178,
      "step": 43144
    },
    {
      "epoch": 1.9799458492037996,
      "grad_norm": 0.4618842303752899,
      "learning_rate": 2.6876057925597128e-06,
      "loss": 0.3471,
      "step": 43145
    },
    {
      "epoch": 1.979991739709054,
      "grad_norm": 0.47371867299079895,
      "learning_rate": 2.687388404307999e-06,
      "loss": 0.3455,
      "step": 43146
    },
    {
      "epoch": 1.9800376302143086,
      "grad_norm": 0.4555322825908661,
      "learning_rate": 2.6871710216173884e-06,
      "loss": 0.3203,
      "step": 43147
    },
    {
      "epoch": 1.980083520719563,
      "grad_norm": 0.4681788384914398,
      "learning_rate": 2.6869536444884016e-06,
      "loss": 0.3785,
      "step": 43148
    },
    {
      "epoch": 1.9801294112248176,
      "grad_norm": 0.4780575931072235,
      "learning_rate": 2.6867362729215586e-06,
      "loss": 0.3825,
      "step": 43149
    },
    {
      "epoch": 1.980175301730072,
      "grad_norm": 0.4842798113822937,
      "learning_rate": 2.686518906917387e-06,
      "loss": 0.3566,
      "step": 43150
    },
    {
      "epoch": 1.9802211922353266,
      "grad_norm": 0.4575755000114441,
      "learning_rate": 2.6863015464764077e-06,
      "loss": 0.3232,
      "step": 43151
    },
    {
      "epoch": 1.980267082740581,
      "grad_norm": 0.47339776158332825,
      "learning_rate": 2.6860841915991428e-06,
      "loss": 0.3426,
      "step": 43152
    },
    {
      "epoch": 1.9803129732458356,
      "grad_norm": 0.47130417823791504,
      "learning_rate": 2.6858668422861127e-06,
      "loss": 0.3102,
      "step": 43153
    },
    {
      "epoch": 1.98035886375109,
      "grad_norm": 0.5300039649009705,
      "learning_rate": 2.685649498537847e-06,
      "loss": 0.4439,
      "step": 43154
    },
    {
      "epoch": 1.9804047542563443,
      "grad_norm": 0.45438867807388306,
      "learning_rate": 2.685432160354862e-06,
      "loss": 0.335,
      "step": 43155
    },
    {
      "epoch": 1.9804506447615988,
      "grad_norm": 0.4509146213531494,
      "learning_rate": 2.6852148277376795e-06,
      "loss": 0.331,
      "step": 43156
    },
    {
      "epoch": 1.9804965352668533,
      "grad_norm": 0.4600050747394562,
      "learning_rate": 2.6849975006868268e-06,
      "loss": 0.3281,
      "step": 43157
    },
    {
      "epoch": 1.9805424257721076,
      "grad_norm": 0.47251150012016296,
      "learning_rate": 2.684780179202825e-06,
      "loss": 0.3536,
      "step": 43158
    },
    {
      "epoch": 1.980588316277362,
      "grad_norm": 0.4947102963924408,
      "learning_rate": 2.6845628632861935e-06,
      "loss": 0.3914,
      "step": 43159
    },
    {
      "epoch": 1.9806342067826166,
      "grad_norm": 0.4416678547859192,
      "learning_rate": 2.68434555293746e-06,
      "loss": 0.2755,
      "step": 43160
    },
    {
      "epoch": 1.980680097287871,
      "grad_norm": 0.4619218707084656,
      "learning_rate": 2.6841282481571448e-06,
      "loss": 0.3125,
      "step": 43161
    },
    {
      "epoch": 1.9807259877931256,
      "grad_norm": 0.5089234709739685,
      "learning_rate": 2.6839109489457683e-06,
      "loss": 0.414,
      "step": 43162
    },
    {
      "epoch": 1.98077187829838,
      "grad_norm": 0.4740343689918518,
      "learning_rate": 2.6836936553038563e-06,
      "loss": 0.3339,
      "step": 43163
    },
    {
      "epoch": 1.9808177688036346,
      "grad_norm": 0.5019904375076294,
      "learning_rate": 2.6834763672319305e-06,
      "loss": 0.4203,
      "step": 43164
    },
    {
      "epoch": 1.980863659308889,
      "grad_norm": 0.45936256647109985,
      "learning_rate": 2.683259084730513e-06,
      "loss": 0.3642,
      "step": 43165
    },
    {
      "epoch": 1.9809095498141436,
      "grad_norm": 0.4591171145439148,
      "learning_rate": 2.6830418078001257e-06,
      "loss": 0.3355,
      "step": 43166
    },
    {
      "epoch": 1.980955440319398,
      "grad_norm": 0.44852474331855774,
      "learning_rate": 2.682824536441292e-06,
      "loss": 0.3435,
      "step": 43167
    },
    {
      "epoch": 1.9810013308246524,
      "grad_norm": 0.49362272024154663,
      "learning_rate": 2.682607270654532e-06,
      "loss": 0.3959,
      "step": 43168
    },
    {
      "epoch": 1.9810472213299068,
      "grad_norm": 0.48774832487106323,
      "learning_rate": 2.6823900104403723e-06,
      "loss": 0.3518,
      "step": 43169
    },
    {
      "epoch": 1.9810931118351613,
      "grad_norm": 0.47221601009368896,
      "learning_rate": 2.6821727557993326e-06,
      "loss": 0.3266,
      "step": 43170
    },
    {
      "epoch": 1.9811390023404156,
      "grad_norm": 0.5298958420753479,
      "learning_rate": 2.6819555067319344e-06,
      "loss": 0.4232,
      "step": 43171
    },
    {
      "epoch": 1.9811848928456701,
      "grad_norm": 0.4496658146381378,
      "learning_rate": 2.6817382632387033e-06,
      "loss": 0.3283,
      "step": 43172
    },
    {
      "epoch": 1.9812307833509246,
      "grad_norm": 0.44550344347953796,
      "learning_rate": 2.6815210253201602e-06,
      "loss": 0.3136,
      "step": 43173
    },
    {
      "epoch": 1.9812766738561791,
      "grad_norm": 0.44526734948158264,
      "learning_rate": 2.681303792976825e-06,
      "loss": 0.2867,
      "step": 43174
    },
    {
      "epoch": 1.9813225643614336,
      "grad_norm": 0.4784291684627533,
      "learning_rate": 2.6810865662092246e-06,
      "loss": 0.3343,
      "step": 43175
    },
    {
      "epoch": 1.9813684548666881,
      "grad_norm": 0.46351248025894165,
      "learning_rate": 2.6808693450178814e-06,
      "loss": 0.3449,
      "step": 43176
    },
    {
      "epoch": 1.9814143453719426,
      "grad_norm": 0.48730435967445374,
      "learning_rate": 2.6806521294033112e-06,
      "loss": 0.3964,
      "step": 43177
    },
    {
      "epoch": 1.981460235877197,
      "grad_norm": 0.48592326045036316,
      "learning_rate": 2.6804349193660422e-06,
      "loss": 0.364,
      "step": 43178
    },
    {
      "epoch": 1.9815061263824516,
      "grad_norm": 0.502487063407898,
      "learning_rate": 2.6802177149065957e-06,
      "loss": 0.4822,
      "step": 43179
    },
    {
      "epoch": 1.9815520168877059,
      "grad_norm": 0.46852949261665344,
      "learning_rate": 2.680000516025492e-06,
      "loss": 0.3326,
      "step": 43180
    },
    {
      "epoch": 1.9815979073929604,
      "grad_norm": 0.4312870502471924,
      "learning_rate": 2.6797833227232556e-06,
      "loss": 0.3286,
      "step": 43181
    },
    {
      "epoch": 1.9816437978982149,
      "grad_norm": 0.46828654408454895,
      "learning_rate": 2.679566135000409e-06,
      "loss": 0.3938,
      "step": 43182
    },
    {
      "epoch": 1.9816896884034694,
      "grad_norm": 0.514013946056366,
      "learning_rate": 2.6793489528574705e-06,
      "loss": 0.4195,
      "step": 43183
    },
    {
      "epoch": 1.9817355789087236,
      "grad_norm": 0.48704972863197327,
      "learning_rate": 2.679131776294968e-06,
      "loss": 0.3821,
      "step": 43184
    },
    {
      "epoch": 1.9817814694139781,
      "grad_norm": 0.4998438060283661,
      "learning_rate": 2.6789146053134215e-06,
      "loss": 0.4217,
      "step": 43185
    },
    {
      "epoch": 1.9818273599192326,
      "grad_norm": 0.4347667396068573,
      "learning_rate": 2.678697439913352e-06,
      "loss": 0.3002,
      "step": 43186
    },
    {
      "epoch": 1.9818732504244871,
      "grad_norm": 0.468792587518692,
      "learning_rate": 2.678480280095283e-06,
      "loss": 0.3976,
      "step": 43187
    },
    {
      "epoch": 1.9819191409297416,
      "grad_norm": 0.4715765118598938,
      "learning_rate": 2.6782631258597357e-06,
      "loss": 0.3584,
      "step": 43188
    },
    {
      "epoch": 1.9819650314349961,
      "grad_norm": 0.5024660229682922,
      "learning_rate": 2.6780459772072336e-06,
      "loss": 0.4455,
      "step": 43189
    },
    {
      "epoch": 1.9820109219402506,
      "grad_norm": 0.49241316318511963,
      "learning_rate": 2.6778288341382957e-06,
      "loss": 0.3723,
      "step": 43190
    },
    {
      "epoch": 1.9820568124455051,
      "grad_norm": 0.4418708086013794,
      "learning_rate": 2.6776116966534487e-06,
      "loss": 0.3004,
      "step": 43191
    },
    {
      "epoch": 1.9821027029507596,
      "grad_norm": 0.5096988677978516,
      "learning_rate": 2.6773945647532128e-06,
      "loss": 0.4391,
      "step": 43192
    },
    {
      "epoch": 1.982148593456014,
      "grad_norm": 0.4441571533679962,
      "learning_rate": 2.6771774384381075e-06,
      "loss": 0.3242,
      "step": 43193
    },
    {
      "epoch": 1.9821944839612684,
      "grad_norm": 0.47249630093574524,
      "learning_rate": 2.6769603177086594e-06,
      "loss": 0.4025,
      "step": 43194
    },
    {
      "epoch": 1.982240374466523,
      "grad_norm": 0.4635104238986969,
      "learning_rate": 2.6767432025653885e-06,
      "loss": 0.3248,
      "step": 43195
    },
    {
      "epoch": 1.9822862649717772,
      "grad_norm": 0.4669688940048218,
      "learning_rate": 2.6765260930088153e-06,
      "loss": 0.3523,
      "step": 43196
    },
    {
      "epoch": 1.9823321554770317,
      "grad_norm": 0.49607664346694946,
      "learning_rate": 2.6763089890394654e-06,
      "loss": 0.4208,
      "step": 43197
    },
    {
      "epoch": 1.9823780459822862,
      "grad_norm": 0.4378975033760071,
      "learning_rate": 2.676091890657861e-06,
      "loss": 0.3336,
      "step": 43198
    },
    {
      "epoch": 1.9824239364875407,
      "grad_norm": 0.43017831444740295,
      "learning_rate": 2.6758747978645183e-06,
      "loss": 0.289,
      "step": 43199
    },
    {
      "epoch": 1.9824698269927952,
      "grad_norm": 0.4746286869049072,
      "learning_rate": 2.6756577106599646e-06,
      "loss": 0.3786,
      "step": 43200
    },
    {
      "epoch": 1.9825157174980497,
      "grad_norm": 0.5178020596504211,
      "learning_rate": 2.675440629044721e-06,
      "loss": 0.4205,
      "step": 43201
    },
    {
      "epoch": 1.9825616080033042,
      "grad_norm": 0.4440845847129822,
      "learning_rate": 2.6752235530193073e-06,
      "loss": 0.3186,
      "step": 43202
    },
    {
      "epoch": 1.9826074985085587,
      "grad_norm": 0.4626782536506653,
      "learning_rate": 2.675006482584249e-06,
      "loss": 0.3061,
      "step": 43203
    },
    {
      "epoch": 1.9826533890138132,
      "grad_norm": 0.4835585951805115,
      "learning_rate": 2.6747894177400656e-06,
      "loss": 0.3828,
      "step": 43204
    },
    {
      "epoch": 1.9826992795190677,
      "grad_norm": 0.43964555859565735,
      "learning_rate": 2.6745723584872787e-06,
      "loss": 0.2903,
      "step": 43205
    },
    {
      "epoch": 1.982745170024322,
      "grad_norm": 0.46803954243659973,
      "learning_rate": 2.674355304826413e-06,
      "loss": 0.3393,
      "step": 43206
    },
    {
      "epoch": 1.9827910605295764,
      "grad_norm": 0.45373210310935974,
      "learning_rate": 2.674138256757989e-06,
      "loss": 0.3414,
      "step": 43207
    },
    {
      "epoch": 1.982836951034831,
      "grad_norm": 0.44474294781684875,
      "learning_rate": 2.6739212142825283e-06,
      "loss": 0.2868,
      "step": 43208
    },
    {
      "epoch": 1.9828828415400852,
      "grad_norm": 0.4508303999900818,
      "learning_rate": 2.6737041774005527e-06,
      "loss": 0.3237,
      "step": 43209
    },
    {
      "epoch": 1.9829287320453397,
      "grad_norm": 0.4737163484096527,
      "learning_rate": 2.6734871461125843e-06,
      "loss": 0.3761,
      "step": 43210
    },
    {
      "epoch": 1.9829746225505942,
      "grad_norm": 0.5069913268089294,
      "learning_rate": 2.6732701204191436e-06,
      "loss": 0.4616,
      "step": 43211
    },
    {
      "epoch": 1.9830205130558487,
      "grad_norm": 0.49565333127975464,
      "learning_rate": 2.673053100320756e-06,
      "loss": 0.361,
      "step": 43212
    },
    {
      "epoch": 1.9830664035611032,
      "grad_norm": 0.4690817892551422,
      "learning_rate": 2.6728360858179407e-06,
      "loss": 0.3688,
      "step": 43213
    },
    {
      "epoch": 1.9831122940663577,
      "grad_norm": 0.4544459283351898,
      "learning_rate": 2.6726190769112183e-06,
      "loss": 0.3057,
      "step": 43214
    },
    {
      "epoch": 1.9831581845716122,
      "grad_norm": 0.4465959668159485,
      "learning_rate": 2.6724020736011145e-06,
      "loss": 0.3226,
      "step": 43215
    },
    {
      "epoch": 1.9832040750768667,
      "grad_norm": 0.47844526171684265,
      "learning_rate": 2.6721850758881494e-06,
      "loss": 0.4058,
      "step": 43216
    },
    {
      "epoch": 1.9832499655821212,
      "grad_norm": 0.4553939700126648,
      "learning_rate": 2.6719680837728447e-06,
      "loss": 0.3272,
      "step": 43217
    },
    {
      "epoch": 1.9832958560873757,
      "grad_norm": 0.4971645176410675,
      "learning_rate": 2.671751097255719e-06,
      "loss": 0.4424,
      "step": 43218
    },
    {
      "epoch": 1.98334174659263,
      "grad_norm": 0.44561585783958435,
      "learning_rate": 2.6715341163373e-06,
      "loss": 0.2855,
      "step": 43219
    },
    {
      "epoch": 1.9833876370978845,
      "grad_norm": 0.4681265354156494,
      "learning_rate": 2.6713171410181083e-06,
      "loss": 0.3969,
      "step": 43220
    },
    {
      "epoch": 1.983433527603139,
      "grad_norm": 0.4675227999687195,
      "learning_rate": 2.6711001712986596e-06,
      "loss": 0.3333,
      "step": 43221
    },
    {
      "epoch": 1.9834794181083932,
      "grad_norm": 0.4366927146911621,
      "learning_rate": 2.670883207179482e-06,
      "loss": 0.3134,
      "step": 43222
    },
    {
      "epoch": 1.9835253086136477,
      "grad_norm": 0.43500030040740967,
      "learning_rate": 2.6706662486610957e-06,
      "loss": 0.2955,
      "step": 43223
    },
    {
      "epoch": 1.9835711991189022,
      "grad_norm": 0.43693315982818604,
      "learning_rate": 2.67044929574402e-06,
      "loss": 0.3262,
      "step": 43224
    },
    {
      "epoch": 1.9836170896241567,
      "grad_norm": 0.4820905327796936,
      "learning_rate": 2.6702323484287794e-06,
      "loss": 0.3632,
      "step": 43225
    },
    {
      "epoch": 1.9836629801294112,
      "grad_norm": 0.4821542501449585,
      "learning_rate": 2.6700154067158958e-06,
      "loss": 0.3575,
      "step": 43226
    },
    {
      "epoch": 1.9837088706346657,
      "grad_norm": 0.45242324471473694,
      "learning_rate": 2.6697984706058877e-06,
      "loss": 0.3571,
      "step": 43227
    },
    {
      "epoch": 1.9837547611399202,
      "grad_norm": 0.48843076825141907,
      "learning_rate": 2.669581540099281e-06,
      "loss": 0.3719,
      "step": 43228
    },
    {
      "epoch": 1.9838006516451747,
      "grad_norm": 0.7229228019714355,
      "learning_rate": 2.6693646151965945e-06,
      "loss": 0.3719,
      "step": 43229
    },
    {
      "epoch": 1.9838465421504292,
      "grad_norm": 0.42431607842445374,
      "learning_rate": 2.6691476958983508e-06,
      "loss": 0.2807,
      "step": 43230
    },
    {
      "epoch": 1.9838924326556835,
      "grad_norm": 0.5001864433288574,
      "learning_rate": 2.6689307822050714e-06,
      "loss": 0.4092,
      "step": 43231
    },
    {
      "epoch": 1.983938323160938,
      "grad_norm": 0.46732696890830994,
      "learning_rate": 2.6687138741172773e-06,
      "loss": 0.3503,
      "step": 43232
    },
    {
      "epoch": 1.9839842136661925,
      "grad_norm": 0.48768356442451477,
      "learning_rate": 2.668496971635489e-06,
      "loss": 0.4142,
      "step": 43233
    },
    {
      "epoch": 1.9840301041714468,
      "grad_norm": 0.4879835844039917,
      "learning_rate": 2.668280074760232e-06,
      "loss": 0.3605,
      "step": 43234
    },
    {
      "epoch": 1.9840759946767013,
      "grad_norm": 0.486217200756073,
      "learning_rate": 2.6680631834920244e-06,
      "loss": 0.3798,
      "step": 43235
    },
    {
      "epoch": 1.9841218851819558,
      "grad_norm": 0.46995219588279724,
      "learning_rate": 2.667846297831388e-06,
      "loss": 0.3559,
      "step": 43236
    },
    {
      "epoch": 1.9841677756872103,
      "grad_norm": 0.4792799949645996,
      "learning_rate": 2.6676294177788463e-06,
      "loss": 0.3821,
      "step": 43237
    },
    {
      "epoch": 1.9842136661924648,
      "grad_norm": 0.4595497250556946,
      "learning_rate": 2.66741254333492e-06,
      "loss": 0.3517,
      "step": 43238
    },
    {
      "epoch": 1.9842595566977193,
      "grad_norm": 0.4479900598526001,
      "learning_rate": 2.6671956745001287e-06,
      "loss": 0.3023,
      "step": 43239
    },
    {
      "epoch": 1.9843054472029737,
      "grad_norm": 0.49852603673934937,
      "learning_rate": 2.6669788112749972e-06,
      "loss": 0.3499,
      "step": 43240
    },
    {
      "epoch": 1.9843513377082282,
      "grad_norm": 0.4633050262928009,
      "learning_rate": 2.6667619536600453e-06,
      "loss": 0.3347,
      "step": 43241
    },
    {
      "epoch": 1.9843972282134827,
      "grad_norm": 0.4546201527118683,
      "learning_rate": 2.666545101655795e-06,
      "loss": 0.2997,
      "step": 43242
    },
    {
      "epoch": 1.9844431187187372,
      "grad_norm": 0.4829615354537964,
      "learning_rate": 2.666328255262767e-06,
      "loss": 0.3491,
      "step": 43243
    },
    {
      "epoch": 1.9844890092239915,
      "grad_norm": 0.4754326343536377,
      "learning_rate": 2.666111414481483e-06,
      "loss": 0.3851,
      "step": 43244
    },
    {
      "epoch": 1.984534899729246,
      "grad_norm": 0.49072161316871643,
      "learning_rate": 2.6658945793124625e-06,
      "loss": 0.4123,
      "step": 43245
    },
    {
      "epoch": 1.9845807902345005,
      "grad_norm": 0.45561665296554565,
      "learning_rate": 2.665677749756231e-06,
      "loss": 0.3587,
      "step": 43246
    },
    {
      "epoch": 1.9846266807397548,
      "grad_norm": 0.4680860638618469,
      "learning_rate": 2.6654609258133072e-06,
      "loss": 0.316,
      "step": 43247
    },
    {
      "epoch": 1.9846725712450093,
      "grad_norm": 0.48525574803352356,
      "learning_rate": 2.665244107484212e-06,
      "loss": 0.4066,
      "step": 43248
    },
    {
      "epoch": 1.9847184617502638,
      "grad_norm": 0.4495392143726349,
      "learning_rate": 2.665027294769469e-06,
      "loss": 0.3504,
      "step": 43249
    },
    {
      "epoch": 1.9847643522555183,
      "grad_norm": 0.44399794936180115,
      "learning_rate": 2.664810487669599e-06,
      "loss": 0.3304,
      "step": 43250
    },
    {
      "epoch": 1.9848102427607728,
      "grad_norm": 0.7158249020576477,
      "learning_rate": 2.6645936861851216e-06,
      "loss": 0.3877,
      "step": 43251
    },
    {
      "epoch": 1.9848561332660273,
      "grad_norm": 0.4605919122695923,
      "learning_rate": 2.664376890316561e-06,
      "loss": 0.3327,
      "step": 43252
    },
    {
      "epoch": 1.9849020237712818,
      "grad_norm": 0.4624258279800415,
      "learning_rate": 2.664160100064436e-06,
      "loss": 0.3461,
      "step": 43253
    },
    {
      "epoch": 1.9849479142765363,
      "grad_norm": 0.45508649945259094,
      "learning_rate": 2.6639433154292684e-06,
      "loss": 0.3582,
      "step": 43254
    },
    {
      "epoch": 1.9849938047817908,
      "grad_norm": 0.4862181842327118,
      "learning_rate": 2.663726536411578e-06,
      "loss": 0.3536,
      "step": 43255
    },
    {
      "epoch": 1.9850396952870453,
      "grad_norm": 0.4658057689666748,
      "learning_rate": 2.6635097630118906e-06,
      "loss": 0.3482,
      "step": 43256
    },
    {
      "epoch": 1.9850855857922995,
      "grad_norm": 0.47662732005119324,
      "learning_rate": 2.6632929952307242e-06,
      "loss": 0.3682,
      "step": 43257
    },
    {
      "epoch": 1.985131476297554,
      "grad_norm": 0.4431224763393402,
      "learning_rate": 2.6630762330685987e-06,
      "loss": 0.2737,
      "step": 43258
    },
    {
      "epoch": 1.9851773668028085,
      "grad_norm": 0.420151025056839,
      "learning_rate": 2.6628594765260396e-06,
      "loss": 0.2965,
      "step": 43259
    },
    {
      "epoch": 1.9852232573080628,
      "grad_norm": 0.48970189690589905,
      "learning_rate": 2.662642725603566e-06,
      "loss": 0.4131,
      "step": 43260
    },
    {
      "epoch": 1.9852691478133173,
      "grad_norm": 0.44862979650497437,
      "learning_rate": 2.662425980301696e-06,
      "loss": 0.3202,
      "step": 43261
    },
    {
      "epoch": 1.9853150383185718,
      "grad_norm": 0.4888520836830139,
      "learning_rate": 2.6622092406209566e-06,
      "loss": 0.3494,
      "step": 43262
    },
    {
      "epoch": 1.9853609288238263,
      "grad_norm": 0.496141254901886,
      "learning_rate": 2.6619925065618656e-06,
      "loss": 0.376,
      "step": 43263
    },
    {
      "epoch": 1.9854068193290808,
      "grad_norm": 0.5399676561355591,
      "learning_rate": 2.6617757781249455e-06,
      "loss": 0.3829,
      "step": 43264
    },
    {
      "epoch": 1.9854527098343353,
      "grad_norm": 0.49451279640197754,
      "learning_rate": 2.661559055310716e-06,
      "loss": 0.3751,
      "step": 43265
    },
    {
      "epoch": 1.9854986003395898,
      "grad_norm": 0.43755173683166504,
      "learning_rate": 2.6613423381196996e-06,
      "loss": 0.2939,
      "step": 43266
    },
    {
      "epoch": 1.9855444908448443,
      "grad_norm": 0.4871475398540497,
      "learning_rate": 2.6611256265524144e-06,
      "loss": 0.4042,
      "step": 43267
    },
    {
      "epoch": 1.9855903813500988,
      "grad_norm": 0.4361470341682434,
      "learning_rate": 2.660908920609386e-06,
      "loss": 0.2837,
      "step": 43268
    },
    {
      "epoch": 1.985636271855353,
      "grad_norm": 0.45998796820640564,
      "learning_rate": 2.6606922202911328e-06,
      "loss": 0.3499,
      "step": 43269
    },
    {
      "epoch": 1.9856821623606076,
      "grad_norm": 0.4479124844074249,
      "learning_rate": 2.6604755255981753e-06,
      "loss": 0.3404,
      "step": 43270
    },
    {
      "epoch": 1.985728052865862,
      "grad_norm": 0.5127881765365601,
      "learning_rate": 2.660258836531037e-06,
      "loss": 0.4286,
      "step": 43271
    },
    {
      "epoch": 1.9857739433711166,
      "grad_norm": 0.47999855875968933,
      "learning_rate": 2.6600421530902385e-06,
      "loss": 0.3721,
      "step": 43272
    },
    {
      "epoch": 1.9858198338763708,
      "grad_norm": 0.45671913027763367,
      "learning_rate": 2.6598254752762997e-06,
      "loss": 0.3403,
      "step": 43273
    },
    {
      "epoch": 1.9858657243816253,
      "grad_norm": 0.47982463240623474,
      "learning_rate": 2.659608803089742e-06,
      "loss": 0.3846,
      "step": 43274
    },
    {
      "epoch": 1.9859116148868798,
      "grad_norm": 0.45638489723205566,
      "learning_rate": 2.6593921365310873e-06,
      "loss": 0.3313,
      "step": 43275
    },
    {
      "epoch": 1.9859575053921343,
      "grad_norm": 0.4985358715057373,
      "learning_rate": 2.6591754756008525e-06,
      "loss": 0.4098,
      "step": 43276
    },
    {
      "epoch": 1.9860033958973888,
      "grad_norm": 0.48731914162635803,
      "learning_rate": 2.6589588202995646e-06,
      "loss": 0.3824,
      "step": 43277
    },
    {
      "epoch": 1.9860492864026433,
      "grad_norm": 0.46081170439720154,
      "learning_rate": 2.6587421706277416e-06,
      "loss": 0.3413,
      "step": 43278
    },
    {
      "epoch": 1.9860951769078978,
      "grad_norm": 0.4757758677005768,
      "learning_rate": 2.6585255265859034e-06,
      "loss": 0.3507,
      "step": 43279
    },
    {
      "epoch": 1.9861410674131523,
      "grad_norm": 0.49008744955062866,
      "learning_rate": 2.658308888174574e-06,
      "loss": 0.4068,
      "step": 43280
    },
    {
      "epoch": 1.9861869579184068,
      "grad_norm": 0.48594027757644653,
      "learning_rate": 2.6580922553942722e-06,
      "loss": 0.3861,
      "step": 43281
    },
    {
      "epoch": 1.986232848423661,
      "grad_norm": 0.46818509697914124,
      "learning_rate": 2.657875628245518e-06,
      "loss": 0.3616,
      "step": 43282
    },
    {
      "epoch": 1.9862787389289156,
      "grad_norm": 0.4532407820224762,
      "learning_rate": 2.6576590067288356e-06,
      "loss": 0.3338,
      "step": 43283
    },
    {
      "epoch": 1.98632462943417,
      "grad_norm": 0.4521293342113495,
      "learning_rate": 2.657442390844744e-06,
      "loss": 0.3545,
      "step": 43284
    },
    {
      "epoch": 1.9863705199394244,
      "grad_norm": 0.4783223569393158,
      "learning_rate": 2.657225780593764e-06,
      "loss": 0.3816,
      "step": 43285
    },
    {
      "epoch": 1.9864164104446789,
      "grad_norm": 0.44485804438591003,
      "learning_rate": 2.657009175976416e-06,
      "loss": 0.318,
      "step": 43286
    },
    {
      "epoch": 1.9864623009499334,
      "grad_norm": 0.43985068798065186,
      "learning_rate": 2.6567925769932222e-06,
      "loss": 0.3115,
      "step": 43287
    },
    {
      "epoch": 1.9865081914551879,
      "grad_norm": 0.4451111853122711,
      "learning_rate": 2.6565759836447025e-06,
      "loss": 0.2978,
      "step": 43288
    },
    {
      "epoch": 1.9865540819604424,
      "grad_norm": 0.4707503914833069,
      "learning_rate": 2.656359395931376e-06,
      "loss": 0.4082,
      "step": 43289
    },
    {
      "epoch": 1.9865999724656969,
      "grad_norm": 0.4510821998119354,
      "learning_rate": 2.656142813853768e-06,
      "loss": 0.3402,
      "step": 43290
    },
    {
      "epoch": 1.9866458629709514,
      "grad_norm": 0.4516948461532593,
      "learning_rate": 2.6559262374123962e-06,
      "loss": 0.3123,
      "step": 43291
    },
    {
      "epoch": 1.9866917534762059,
      "grad_norm": 0.4725872278213501,
      "learning_rate": 2.6557096666077804e-06,
      "loss": 0.3472,
      "step": 43292
    },
    {
      "epoch": 1.9867376439814604,
      "grad_norm": 0.4570804834365845,
      "learning_rate": 2.6554931014404445e-06,
      "loss": 0.3058,
      "step": 43293
    },
    {
      "epoch": 1.9867835344867149,
      "grad_norm": 0.47417378425598145,
      "learning_rate": 2.655276541910908e-06,
      "loss": 0.3403,
      "step": 43294
    },
    {
      "epoch": 1.9868294249919691,
      "grad_norm": 0.5003190636634827,
      "learning_rate": 2.6550599880196913e-06,
      "loss": 0.4293,
      "step": 43295
    },
    {
      "epoch": 1.9868753154972236,
      "grad_norm": 0.4518091678619385,
      "learning_rate": 2.654843439767315e-06,
      "loss": 0.3411,
      "step": 43296
    },
    {
      "epoch": 1.9869212060024781,
      "grad_norm": 0.45399075746536255,
      "learning_rate": 2.6546268971543006e-06,
      "loss": 0.3612,
      "step": 43297
    },
    {
      "epoch": 1.9869670965077324,
      "grad_norm": 0.4527285099029541,
      "learning_rate": 2.6544103601811654e-06,
      "loss": 0.364,
      "step": 43298
    },
    {
      "epoch": 1.987012987012987,
      "grad_norm": 0.5471859574317932,
      "learning_rate": 2.654193828848436e-06,
      "loss": 0.4023,
      "step": 43299
    },
    {
      "epoch": 1.9870588775182414,
      "grad_norm": 0.47555187344551086,
      "learning_rate": 2.65397730315663e-06,
      "loss": 0.3597,
      "step": 43300
    },
    {
      "epoch": 1.987104768023496,
      "grad_norm": 0.48820748925209045,
      "learning_rate": 2.6537607831062654e-06,
      "loss": 0.3972,
      "step": 43301
    },
    {
      "epoch": 1.9871506585287504,
      "grad_norm": 0.45449841022491455,
      "learning_rate": 2.6535442686978685e-06,
      "loss": 0.2934,
      "step": 43302
    },
    {
      "epoch": 1.9871965490340049,
      "grad_norm": 0.4821808636188507,
      "learning_rate": 2.653327759931956e-06,
      "loss": 0.367,
      "step": 43303
    },
    {
      "epoch": 1.9872424395392594,
      "grad_norm": 0.49018624424934387,
      "learning_rate": 2.653111256809049e-06,
      "loss": 0.3568,
      "step": 43304
    },
    {
      "epoch": 1.9872883300445139,
      "grad_norm": 0.5007258057594299,
      "learning_rate": 2.6528947593296697e-06,
      "loss": 0.4489,
      "step": 43305
    },
    {
      "epoch": 1.9873342205497684,
      "grad_norm": 0.5081180334091187,
      "learning_rate": 2.6526782674943387e-06,
      "loss": 0.4719,
      "step": 43306
    },
    {
      "epoch": 1.9873801110550229,
      "grad_norm": 0.45810064673423767,
      "learning_rate": 2.6524617813035747e-06,
      "loss": 0.365,
      "step": 43307
    },
    {
      "epoch": 1.9874260015602772,
      "grad_norm": 0.512195348739624,
      "learning_rate": 2.652245300757899e-06,
      "loss": 0.3656,
      "step": 43308
    },
    {
      "epoch": 1.9874718920655317,
      "grad_norm": 0.4660709500312805,
      "learning_rate": 2.652028825857833e-06,
      "loss": 0.3128,
      "step": 43309
    },
    {
      "epoch": 1.9875177825707862,
      "grad_norm": 0.4849500060081482,
      "learning_rate": 2.6518123566038944e-06,
      "loss": 0.3956,
      "step": 43310
    },
    {
      "epoch": 1.9875636730760404,
      "grad_norm": 0.45419880747795105,
      "learning_rate": 2.651595892996608e-06,
      "loss": 0.3112,
      "step": 43311
    },
    {
      "epoch": 1.987609563581295,
      "grad_norm": 0.49454638361930847,
      "learning_rate": 2.6513794350364925e-06,
      "loss": 0.4176,
      "step": 43312
    },
    {
      "epoch": 1.9876554540865494,
      "grad_norm": 0.44968920946121216,
      "learning_rate": 2.651162982724066e-06,
      "loss": 0.3263,
      "step": 43313
    },
    {
      "epoch": 1.987701344591804,
      "grad_norm": 0.47197842597961426,
      "learning_rate": 2.6509465360598536e-06,
      "loss": 0.3456,
      "step": 43314
    },
    {
      "epoch": 1.9877472350970584,
      "grad_norm": 0.4636886417865753,
      "learning_rate": 2.6507300950443726e-06,
      "loss": 0.3183,
      "step": 43315
    },
    {
      "epoch": 1.987793125602313,
      "grad_norm": 0.4490591883659363,
      "learning_rate": 2.650513659678145e-06,
      "loss": 0.341,
      "step": 43316
    },
    {
      "epoch": 1.9878390161075674,
      "grad_norm": 0.4331201910972595,
      "learning_rate": 2.65029722996169e-06,
      "loss": 0.311,
      "step": 43317
    },
    {
      "epoch": 1.987884906612822,
      "grad_norm": 0.4280503988265991,
      "learning_rate": 2.650080805895529e-06,
      "loss": 0.2844,
      "step": 43318
    },
    {
      "epoch": 1.9879307971180764,
      "grad_norm": 0.48602843284606934,
      "learning_rate": 2.649864387480182e-06,
      "loss": 0.3565,
      "step": 43319
    },
    {
      "epoch": 1.9879766876233307,
      "grad_norm": 0.494600385427475,
      "learning_rate": 2.6496479747161674e-06,
      "loss": 0.4133,
      "step": 43320
    },
    {
      "epoch": 1.9880225781285852,
      "grad_norm": 0.41177231073379517,
      "learning_rate": 2.6494315676040093e-06,
      "loss": 0.2651,
      "step": 43321
    },
    {
      "epoch": 1.9880684686338397,
      "grad_norm": 0.4757979214191437,
      "learning_rate": 2.6492151661442266e-06,
      "loss": 0.4093,
      "step": 43322
    },
    {
      "epoch": 1.988114359139094,
      "grad_norm": 0.5252898335456848,
      "learning_rate": 2.6489987703373376e-06,
      "loss": 0.3677,
      "step": 43323
    },
    {
      "epoch": 1.9881602496443485,
      "grad_norm": 0.44714444875717163,
      "learning_rate": 2.6487823801838663e-06,
      "loss": 0.3283,
      "step": 43324
    },
    {
      "epoch": 1.988206140149603,
      "grad_norm": 0.48737475275993347,
      "learning_rate": 2.648565995684332e-06,
      "loss": 0.405,
      "step": 43325
    },
    {
      "epoch": 1.9882520306548574,
      "grad_norm": 0.4801470637321472,
      "learning_rate": 2.648349616839252e-06,
      "loss": 0.353,
      "step": 43326
    },
    {
      "epoch": 1.988297921160112,
      "grad_norm": 0.5045482516288757,
      "learning_rate": 2.64813324364915e-06,
      "loss": 0.4309,
      "step": 43327
    },
    {
      "epoch": 1.9883438116653664,
      "grad_norm": 0.45169174671173096,
      "learning_rate": 2.6479168761145467e-06,
      "loss": 0.3071,
      "step": 43328
    },
    {
      "epoch": 1.988389702170621,
      "grad_norm": 0.5058801174163818,
      "learning_rate": 2.64770051423596e-06,
      "loss": 0.4339,
      "step": 43329
    },
    {
      "epoch": 1.9884355926758754,
      "grad_norm": 0.505872905254364,
      "learning_rate": 2.647484158013912e-06,
      "loss": 0.3621,
      "step": 43330
    },
    {
      "epoch": 1.98848148318113,
      "grad_norm": 0.43194788694381714,
      "learning_rate": 2.647267807448921e-06,
      "loss": 0.3044,
      "step": 43331
    },
    {
      "epoch": 1.9885273736863844,
      "grad_norm": 0.4160498082637787,
      "learning_rate": 2.6470514625415077e-06,
      "loss": 0.2855,
      "step": 43332
    },
    {
      "epoch": 1.9885732641916387,
      "grad_norm": 0.48693475127220154,
      "learning_rate": 2.646835123292194e-06,
      "loss": 0.3922,
      "step": 43333
    },
    {
      "epoch": 1.9886191546968932,
      "grad_norm": 0.43357419967651367,
      "learning_rate": 2.6466187897014996e-06,
      "loss": 0.3389,
      "step": 43334
    },
    {
      "epoch": 1.9886650452021477,
      "grad_norm": 0.4509291648864746,
      "learning_rate": 2.6464024617699425e-06,
      "loss": 0.3421,
      "step": 43335
    },
    {
      "epoch": 1.988710935707402,
      "grad_norm": 0.4928635060787201,
      "learning_rate": 2.646186139498046e-06,
      "loss": 0.3688,
      "step": 43336
    },
    {
      "epoch": 1.9887568262126565,
      "grad_norm": 0.4841051399707794,
      "learning_rate": 2.6459698228863295e-06,
      "loss": 0.405,
      "step": 43337
    },
    {
      "epoch": 1.988802716717911,
      "grad_norm": 0.4398481249809265,
      "learning_rate": 2.6457535119353117e-06,
      "loss": 0.3065,
      "step": 43338
    },
    {
      "epoch": 1.9888486072231655,
      "grad_norm": 0.46194323897361755,
      "learning_rate": 2.645537206645515e-06,
      "loss": 0.3473,
      "step": 43339
    },
    {
      "epoch": 1.98889449772842,
      "grad_norm": 0.42861220240592957,
      "learning_rate": 2.6453209070174574e-06,
      "loss": 0.2902,
      "step": 43340
    },
    {
      "epoch": 1.9889403882336745,
      "grad_norm": 0.4792320728302002,
      "learning_rate": 2.6451046130516576e-06,
      "loss": 0.332,
      "step": 43341
    },
    {
      "epoch": 1.988986278738929,
      "grad_norm": 0.4614847004413605,
      "learning_rate": 2.644888324748641e-06,
      "loss": 0.3356,
      "step": 43342
    },
    {
      "epoch": 1.9890321692441835,
      "grad_norm": 0.5008379220962524,
      "learning_rate": 2.644672042108924e-06,
      "loss": 0.4241,
      "step": 43343
    },
    {
      "epoch": 1.989078059749438,
      "grad_norm": 0.4531208276748657,
      "learning_rate": 2.6444557651330255e-06,
      "loss": 0.319,
      "step": 43344
    },
    {
      "epoch": 1.9891239502546925,
      "grad_norm": 0.412923663854599,
      "learning_rate": 2.6442394938214695e-06,
      "loss": 0.2657,
      "step": 43345
    },
    {
      "epoch": 1.9891698407599467,
      "grad_norm": 0.4707610011100769,
      "learning_rate": 2.6440232281747742e-06,
      "loss": 0.3457,
      "step": 43346
    },
    {
      "epoch": 1.9892157312652012,
      "grad_norm": 0.470540851354599,
      "learning_rate": 2.643806968193457e-06,
      "loss": 0.3329,
      "step": 43347
    },
    {
      "epoch": 1.9892616217704557,
      "grad_norm": 0.4819805920124054,
      "learning_rate": 2.6435907138780425e-06,
      "loss": 0.3919,
      "step": 43348
    },
    {
      "epoch": 1.98930751227571,
      "grad_norm": 0.5174906849861145,
      "learning_rate": 2.6433744652290487e-06,
      "loss": 0.3277,
      "step": 43349
    },
    {
      "epoch": 1.9893534027809645,
      "grad_norm": 0.46744304895401,
      "learning_rate": 2.643158222246996e-06,
      "loss": 0.3159,
      "step": 43350
    },
    {
      "epoch": 1.989399293286219,
      "grad_norm": 0.4820377826690674,
      "learning_rate": 2.6429419849324033e-06,
      "loss": 0.3623,
      "step": 43351
    },
    {
      "epoch": 1.9894451837914735,
      "grad_norm": 0.4900696873664856,
      "learning_rate": 2.642725753285792e-06,
      "loss": 0.3673,
      "step": 43352
    },
    {
      "epoch": 1.989491074296728,
      "grad_norm": 0.4480346143245697,
      "learning_rate": 2.6425095273076805e-06,
      "loss": 0.3205,
      "step": 43353
    },
    {
      "epoch": 1.9895369648019825,
      "grad_norm": 0.5048576593399048,
      "learning_rate": 2.642293306998588e-06,
      "loss": 0.3733,
      "step": 43354
    },
    {
      "epoch": 1.989582855307237,
      "grad_norm": 0.4269384741783142,
      "learning_rate": 2.6420770923590378e-06,
      "loss": 0.2734,
      "step": 43355
    },
    {
      "epoch": 1.9896287458124915,
      "grad_norm": 0.46099716424942017,
      "learning_rate": 2.6418608833895476e-06,
      "loss": 0.35,
      "step": 43356
    },
    {
      "epoch": 1.989674636317746,
      "grad_norm": 0.4357855021953583,
      "learning_rate": 2.641644680090636e-06,
      "loss": 0.3227,
      "step": 43357
    },
    {
      "epoch": 1.9897205268230003,
      "grad_norm": 0.4243274927139282,
      "learning_rate": 2.641428482462827e-06,
      "loss": 0.281,
      "step": 43358
    },
    {
      "epoch": 1.9897664173282548,
      "grad_norm": 0.4360494017601013,
      "learning_rate": 2.6412122905066372e-06,
      "loss": 0.3265,
      "step": 43359
    },
    {
      "epoch": 1.9898123078335093,
      "grad_norm": 0.48708534240722656,
      "learning_rate": 2.6409961042225885e-06,
      "loss": 0.386,
      "step": 43360
    },
    {
      "epoch": 1.9898581983387638,
      "grad_norm": 0.4680542051792145,
      "learning_rate": 2.640779923611199e-06,
      "loss": 0.3539,
      "step": 43361
    },
    {
      "epoch": 1.989904088844018,
      "grad_norm": 0.44694122672080994,
      "learning_rate": 2.6405637486729897e-06,
      "loss": 0.3172,
      "step": 43362
    },
    {
      "epoch": 1.9899499793492725,
      "grad_norm": 0.45567813515663147,
      "learning_rate": 2.6403475794084777e-06,
      "loss": 0.3089,
      "step": 43363
    },
    {
      "epoch": 1.989995869854527,
      "grad_norm": 0.47259414196014404,
      "learning_rate": 2.640131415818187e-06,
      "loss": 0.3464,
      "step": 43364
    },
    {
      "epoch": 1.9900417603597815,
      "grad_norm": 0.4794715940952301,
      "learning_rate": 2.6399152579026357e-06,
      "loss": 0.3805,
      "step": 43365
    },
    {
      "epoch": 1.990087650865036,
      "grad_norm": 0.44594621658325195,
      "learning_rate": 2.6396991056623407e-06,
      "loss": 0.3223,
      "step": 43366
    },
    {
      "epoch": 1.9901335413702905,
      "grad_norm": 0.44707733392715454,
      "learning_rate": 2.6394829590978264e-06,
      "loss": 0.3145,
      "step": 43367
    },
    {
      "epoch": 1.990179431875545,
      "grad_norm": 0.4865078628063202,
      "learning_rate": 2.6392668182096114e-06,
      "loss": 0.3996,
      "step": 43368
    },
    {
      "epoch": 1.9902253223807995,
      "grad_norm": 0.6785011291503906,
      "learning_rate": 2.639050682998212e-06,
      "loss": 0.278,
      "step": 43369
    },
    {
      "epoch": 1.990271212886054,
      "grad_norm": 0.4647110104560852,
      "learning_rate": 2.6388345534641524e-06,
      "loss": 0.3207,
      "step": 43370
    },
    {
      "epoch": 1.9903171033913083,
      "grad_norm": 0.5090306401252747,
      "learning_rate": 2.638618429607951e-06,
      "loss": 0.4192,
      "step": 43371
    },
    {
      "epoch": 1.9903629938965628,
      "grad_norm": 0.44892221689224243,
      "learning_rate": 2.638402311430127e-06,
      "loss": 0.3004,
      "step": 43372
    },
    {
      "epoch": 1.9904088844018173,
      "grad_norm": 0.5025840401649475,
      "learning_rate": 2.6381861989311998e-06,
      "loss": 0.3977,
      "step": 43373
    },
    {
      "epoch": 1.9904547749070716,
      "grad_norm": 0.43405893445014954,
      "learning_rate": 2.6379700921116897e-06,
      "loss": 0.2953,
      "step": 43374
    },
    {
      "epoch": 1.990500665412326,
      "grad_norm": 0.492634117603302,
      "learning_rate": 2.637753990972114e-06,
      "loss": 0.3914,
      "step": 43375
    },
    {
      "epoch": 1.9905465559175806,
      "grad_norm": 0.48303088545799255,
      "learning_rate": 2.637537895512996e-06,
      "loss": 0.4275,
      "step": 43376
    },
    {
      "epoch": 1.990592446422835,
      "grad_norm": 0.5093284249305725,
      "learning_rate": 2.6373218057348535e-06,
      "loss": 0.3523,
      "step": 43377
    },
    {
      "epoch": 1.9906383369280896,
      "grad_norm": 0.4901745617389679,
      "learning_rate": 2.6371057216382055e-06,
      "loss": 0.3881,
      "step": 43378
    },
    {
      "epoch": 1.990684227433344,
      "grad_norm": 0.43879425525665283,
      "learning_rate": 2.6368896432235734e-06,
      "loss": 0.3027,
      "step": 43379
    },
    {
      "epoch": 1.9907301179385986,
      "grad_norm": 0.4827568531036377,
      "learning_rate": 2.636673570491477e-06,
      "loss": 0.3437,
      "step": 43380
    },
    {
      "epoch": 1.990776008443853,
      "grad_norm": 0.4673489034175873,
      "learning_rate": 2.636457503442434e-06,
      "loss": 0.3757,
      "step": 43381
    },
    {
      "epoch": 1.9908218989491075,
      "grad_norm": 0.4985295534133911,
      "learning_rate": 2.636241442076965e-06,
      "loss": 0.3589,
      "step": 43382
    },
    {
      "epoch": 1.990867789454362,
      "grad_norm": 0.4456523060798645,
      "learning_rate": 2.6360253863955888e-06,
      "loss": 0.3113,
      "step": 43383
    },
    {
      "epoch": 1.9909136799596163,
      "grad_norm": 0.46577149629592896,
      "learning_rate": 2.6358093363988257e-06,
      "loss": 0.3301,
      "step": 43384
    },
    {
      "epoch": 1.9909595704648708,
      "grad_norm": 0.4556271731853485,
      "learning_rate": 2.6355932920871936e-06,
      "loss": 0.3593,
      "step": 43385
    },
    {
      "epoch": 1.9910054609701253,
      "grad_norm": 0.44698846340179443,
      "learning_rate": 2.6353772534612147e-06,
      "loss": 0.3361,
      "step": 43386
    },
    {
      "epoch": 1.9910513514753796,
      "grad_norm": 0.4638999104499817,
      "learning_rate": 2.6351612205214078e-06,
      "loss": 0.277,
      "step": 43387
    },
    {
      "epoch": 1.991097241980634,
      "grad_norm": 0.41555649042129517,
      "learning_rate": 2.634945193268289e-06,
      "loss": 0.2616,
      "step": 43388
    },
    {
      "epoch": 1.9911431324858886,
      "grad_norm": 0.4532291889190674,
      "learning_rate": 2.634729171702384e-06,
      "loss": 0.2867,
      "step": 43389
    },
    {
      "epoch": 1.991189022991143,
      "grad_norm": 0.4801202416419983,
      "learning_rate": 2.6345131558242074e-06,
      "loss": 0.3755,
      "step": 43390
    },
    {
      "epoch": 1.9912349134963976,
      "grad_norm": 0.456535279750824,
      "learning_rate": 2.6342971456342782e-06,
      "loss": 0.375,
      "step": 43391
    },
    {
      "epoch": 1.991280804001652,
      "grad_norm": 0.4986515939235687,
      "learning_rate": 2.63408114113312e-06,
      "loss": 0.3966,
      "step": 43392
    },
    {
      "epoch": 1.9913266945069066,
      "grad_norm": 0.46540403366088867,
      "learning_rate": 2.6338651423212502e-06,
      "loss": 0.3744,
      "step": 43393
    },
    {
      "epoch": 1.991372585012161,
      "grad_norm": 0.46242424845695496,
      "learning_rate": 2.633649149199188e-06,
      "loss": 0.3524,
      "step": 43394
    },
    {
      "epoch": 1.9914184755174156,
      "grad_norm": 0.5062050819396973,
      "learning_rate": 2.6334331617674525e-06,
      "loss": 0.4207,
      "step": 43395
    },
    {
      "epoch": 1.99146436602267,
      "grad_norm": 0.4778480529785156,
      "learning_rate": 2.633217180026564e-06,
      "loss": 0.3561,
      "step": 43396
    },
    {
      "epoch": 1.9915102565279243,
      "grad_norm": 0.45920810103416443,
      "learning_rate": 2.633001203977038e-06,
      "loss": 0.2934,
      "step": 43397
    },
    {
      "epoch": 1.9915561470331788,
      "grad_norm": 0.4391583204269409,
      "learning_rate": 2.6327852336193997e-06,
      "loss": 0.304,
      "step": 43398
    },
    {
      "epoch": 1.9916020375384333,
      "grad_norm": 0.4553906321525574,
      "learning_rate": 2.632569268954166e-06,
      "loss": 0.3593,
      "step": 43399
    },
    {
      "epoch": 1.9916479280436876,
      "grad_norm": 0.6995207667350769,
      "learning_rate": 2.6323533099818534e-06,
      "loss": 0.3767,
      "step": 43400
    },
    {
      "epoch": 1.9916938185489421,
      "grad_norm": 0.4738048017024994,
      "learning_rate": 2.6321373567029863e-06,
      "loss": 0.3794,
      "step": 43401
    },
    {
      "epoch": 1.9917397090541966,
      "grad_norm": 0.4616584777832031,
      "learning_rate": 2.6319214091180817e-06,
      "loss": 0.2895,
      "step": 43402
    },
    {
      "epoch": 1.9917855995594511,
      "grad_norm": 0.47007960081100464,
      "learning_rate": 2.631705467227659e-06,
      "loss": 0.2875,
      "step": 43403
    },
    {
      "epoch": 1.9918314900647056,
      "grad_norm": 0.4791438579559326,
      "learning_rate": 2.6314895310322363e-06,
      "loss": 0.3611,
      "step": 43404
    },
    {
      "epoch": 1.99187738056996,
      "grad_norm": 0.5257573127746582,
      "learning_rate": 2.6312736005323336e-06,
      "loss": 0.489,
      "step": 43405
    },
    {
      "epoch": 1.9919232710752146,
      "grad_norm": 0.45210495591163635,
      "learning_rate": 2.631057675728469e-06,
      "loss": 0.3198,
      "step": 43406
    },
    {
      "epoch": 1.991969161580469,
      "grad_norm": 0.4715217053890228,
      "learning_rate": 2.6308417566211654e-06,
      "loss": 0.4221,
      "step": 43407
    },
    {
      "epoch": 1.9920150520857236,
      "grad_norm": 0.47869932651519775,
      "learning_rate": 2.630625843210939e-06,
      "loss": 0.4044,
      "step": 43408
    },
    {
      "epoch": 1.9920609425909779,
      "grad_norm": 0.48531776666641235,
      "learning_rate": 2.6304099354983078e-06,
      "loss": 0.399,
      "step": 43409
    },
    {
      "epoch": 1.9921068330962324,
      "grad_norm": 0.4626838266849518,
      "learning_rate": 2.6301940334837944e-06,
      "loss": 0.3719,
      "step": 43410
    },
    {
      "epoch": 1.9921527236014869,
      "grad_norm": 0.48710039258003235,
      "learning_rate": 2.6299781371679167e-06,
      "loss": 0.3819,
      "step": 43411
    },
    {
      "epoch": 1.9921986141067412,
      "grad_norm": 0.4521794617176056,
      "learning_rate": 2.6297622465511917e-06,
      "loss": 0.3386,
      "step": 43412
    },
    {
      "epoch": 1.9922445046119956,
      "grad_norm": 0.45772430300712585,
      "learning_rate": 2.6295463616341426e-06,
      "loss": 0.3391,
      "step": 43413
    },
    {
      "epoch": 1.9922903951172501,
      "grad_norm": 0.41179922223091125,
      "learning_rate": 2.629330482417286e-06,
      "loss": 0.2784,
      "step": 43414
    },
    {
      "epoch": 1.9923362856225046,
      "grad_norm": 0.4537340998649597,
      "learning_rate": 2.6291146089011423e-06,
      "loss": 0.3486,
      "step": 43415
    },
    {
      "epoch": 1.9923821761277591,
      "grad_norm": 0.4797387719154358,
      "learning_rate": 2.6288987410862287e-06,
      "loss": 0.4118,
      "step": 43416
    },
    {
      "epoch": 1.9924280666330136,
      "grad_norm": 0.541167140007019,
      "learning_rate": 2.628682878973066e-06,
      "loss": 0.4511,
      "step": 43417
    },
    {
      "epoch": 1.9924739571382681,
      "grad_norm": 0.4526997208595276,
      "learning_rate": 2.628467022562171e-06,
      "loss": 0.3294,
      "step": 43418
    },
    {
      "epoch": 1.9925198476435226,
      "grad_norm": 0.5251675248146057,
      "learning_rate": 2.6282511718540636e-06,
      "loss": 0.4303,
      "step": 43419
    },
    {
      "epoch": 1.9925657381487771,
      "grad_norm": 0.47006529569625854,
      "learning_rate": 2.6280353268492654e-06,
      "loss": 0.3575,
      "step": 43420
    },
    {
      "epoch": 1.9926116286540316,
      "grad_norm": 0.4677884876728058,
      "learning_rate": 2.6278194875482943e-06,
      "loss": 0.407,
      "step": 43421
    },
    {
      "epoch": 1.992657519159286,
      "grad_norm": 0.4622874855995178,
      "learning_rate": 2.6276036539516657e-06,
      "loss": 0.3725,
      "step": 43422
    },
    {
      "epoch": 1.9927034096645404,
      "grad_norm": 0.4669547975063324,
      "learning_rate": 2.627387826059904e-06,
      "loss": 0.3081,
      "step": 43423
    },
    {
      "epoch": 1.992749300169795,
      "grad_norm": 0.4693675637245178,
      "learning_rate": 2.6271720038735254e-06,
      "loss": 0.369,
      "step": 43424
    },
    {
      "epoch": 1.9927951906750492,
      "grad_norm": 0.4572746753692627,
      "learning_rate": 2.6269561873930495e-06,
      "loss": 0.2796,
      "step": 43425
    },
    {
      "epoch": 1.9928410811803037,
      "grad_norm": 0.510162353515625,
      "learning_rate": 2.626740376618995e-06,
      "loss": 0.4147,
      "step": 43426
    },
    {
      "epoch": 1.9928869716855582,
      "grad_norm": 0.4567500352859497,
      "learning_rate": 2.626524571551881e-06,
      "loss": 0.3479,
      "step": 43427
    },
    {
      "epoch": 1.9929328621908127,
      "grad_norm": 0.4552430510520935,
      "learning_rate": 2.626308772192224e-06,
      "loss": 0.361,
      "step": 43428
    },
    {
      "epoch": 1.9929787526960672,
      "grad_norm": 0.4946906268596649,
      "learning_rate": 2.6260929785405475e-06,
      "loss": 0.3679,
      "step": 43429
    },
    {
      "epoch": 1.9930246432013217,
      "grad_norm": 0.4993841350078583,
      "learning_rate": 2.625877190597368e-06,
      "loss": 0.3721,
      "step": 43430
    },
    {
      "epoch": 1.9930705337065762,
      "grad_norm": 0.46443378925323486,
      "learning_rate": 2.6256614083632025e-06,
      "loss": 0.3761,
      "step": 43431
    },
    {
      "epoch": 1.9931164242118307,
      "grad_norm": 0.4553373157978058,
      "learning_rate": 2.6254456318385744e-06,
      "loss": 0.3345,
      "step": 43432
    },
    {
      "epoch": 1.9931623147170852,
      "grad_norm": 0.4568401873111725,
      "learning_rate": 2.625229861024e-06,
      "loss": 0.3452,
      "step": 43433
    },
    {
      "epoch": 1.9932082052223397,
      "grad_norm": 0.4471507966518402,
      "learning_rate": 2.625014095919996e-06,
      "loss": 0.3248,
      "step": 43434
    },
    {
      "epoch": 1.993254095727594,
      "grad_norm": 0.42721685767173767,
      "learning_rate": 2.624798336527086e-06,
      "loss": 0.322,
      "step": 43435
    },
    {
      "epoch": 1.9932999862328484,
      "grad_norm": 0.4957304894924164,
      "learning_rate": 2.624582582845786e-06,
      "loss": 0.3853,
      "step": 43436
    },
    {
      "epoch": 1.993345876738103,
      "grad_norm": 0.4773694574832916,
      "learning_rate": 2.6243668348766147e-06,
      "loss": 0.4047,
      "step": 43437
    },
    {
      "epoch": 1.9933917672433572,
      "grad_norm": 0.44989436864852905,
      "learning_rate": 2.624151092620092e-06,
      "loss": 0.3398,
      "step": 43438
    },
    {
      "epoch": 1.9934376577486117,
      "grad_norm": 0.45981255173683167,
      "learning_rate": 2.623935356076736e-06,
      "loss": 0.3559,
      "step": 43439
    },
    {
      "epoch": 1.9934835482538662,
      "grad_norm": 0.47435519099235535,
      "learning_rate": 2.6237196252470637e-06,
      "loss": 0.3368,
      "step": 43440
    },
    {
      "epoch": 1.9935294387591207,
      "grad_norm": 0.49050796031951904,
      "learning_rate": 2.6235039001315977e-06,
      "loss": 0.4256,
      "step": 43441
    },
    {
      "epoch": 1.9935753292643752,
      "grad_norm": 0.4807266592979431,
      "learning_rate": 2.623288180730854e-06,
      "loss": 0.3511,
      "step": 43442
    },
    {
      "epoch": 1.9936212197696297,
      "grad_norm": 0.5053434371948242,
      "learning_rate": 2.623072467045351e-06,
      "loss": 0.3995,
      "step": 43443
    },
    {
      "epoch": 1.9936671102748842,
      "grad_norm": 0.43278032541275024,
      "learning_rate": 2.6228567590756103e-06,
      "loss": 0.3049,
      "step": 43444
    },
    {
      "epoch": 1.9937130007801387,
      "grad_norm": 0.49170348048210144,
      "learning_rate": 2.6226410568221488e-06,
      "loss": 0.3861,
      "step": 43445
    },
    {
      "epoch": 1.9937588912853932,
      "grad_norm": 0.43277403712272644,
      "learning_rate": 2.6224253602854854e-06,
      "loss": 0.3047,
      "step": 43446
    },
    {
      "epoch": 1.9938047817906475,
      "grad_norm": 0.5038232207298279,
      "learning_rate": 2.6222096694661386e-06,
      "loss": 0.3919,
      "step": 43447
    },
    {
      "epoch": 1.993850672295902,
      "grad_norm": 0.45835354924201965,
      "learning_rate": 2.6219939843646268e-06,
      "loss": 0.3982,
      "step": 43448
    },
    {
      "epoch": 1.9938965628011565,
      "grad_norm": 0.49463650584220886,
      "learning_rate": 2.6217783049814695e-06,
      "loss": 0.3655,
      "step": 43449
    },
    {
      "epoch": 1.993942453306411,
      "grad_norm": 0.4506600797176361,
      "learning_rate": 2.621562631317182e-06,
      "loss": 0.2983,
      "step": 43450
    },
    {
      "epoch": 1.9939883438116652,
      "grad_norm": 0.4609457850456238,
      "learning_rate": 2.6213469633722883e-06,
      "loss": 0.3143,
      "step": 43451
    },
    {
      "epoch": 1.9940342343169197,
      "grad_norm": 0.481280654668808,
      "learning_rate": 2.6211313011473042e-06,
      "loss": 0.31,
      "step": 43452
    },
    {
      "epoch": 1.9940801248221742,
      "grad_norm": 0.48059970140457153,
      "learning_rate": 2.6209156446427466e-06,
      "loss": 0.3431,
      "step": 43453
    },
    {
      "epoch": 1.9941260153274287,
      "grad_norm": 0.4950662851333618,
      "learning_rate": 2.6206999938591384e-06,
      "loss": 0.3938,
      "step": 43454
    },
    {
      "epoch": 1.9941719058326832,
      "grad_norm": 0.48106417059898376,
      "learning_rate": 2.620484348796995e-06,
      "loss": 0.3613,
      "step": 43455
    },
    {
      "epoch": 1.9942177963379377,
      "grad_norm": 0.45322123169898987,
      "learning_rate": 2.6202687094568345e-06,
      "loss": 0.3476,
      "step": 43456
    },
    {
      "epoch": 1.9942636868431922,
      "grad_norm": 0.445407897233963,
      "learning_rate": 2.6200530758391782e-06,
      "loss": 0.3112,
      "step": 43457
    },
    {
      "epoch": 1.9943095773484467,
      "grad_norm": 0.48476442694664,
      "learning_rate": 2.6198374479445433e-06,
      "loss": 0.3959,
      "step": 43458
    },
    {
      "epoch": 1.9943554678537012,
      "grad_norm": 0.4721941649913788,
      "learning_rate": 2.619621825773448e-06,
      "loss": 0.4006,
      "step": 43459
    },
    {
      "epoch": 1.9944013583589555,
      "grad_norm": 0.4715723395347595,
      "learning_rate": 2.6194062093264108e-06,
      "loss": 0.3599,
      "step": 43460
    },
    {
      "epoch": 1.99444724886421,
      "grad_norm": 0.48683395981788635,
      "learning_rate": 2.619190598603951e-06,
      "loss": 0.3987,
      "step": 43461
    },
    {
      "epoch": 1.9944931393694645,
      "grad_norm": 0.42631796002388,
      "learning_rate": 2.6189749936065832e-06,
      "loss": 0.2589,
      "step": 43462
    },
    {
      "epoch": 1.9945390298747188,
      "grad_norm": 0.47948041558265686,
      "learning_rate": 2.6187593943348322e-06,
      "loss": 0.4098,
      "step": 43463
    },
    {
      "epoch": 1.9945849203799733,
      "grad_norm": 0.48881423473358154,
      "learning_rate": 2.618543800789213e-06,
      "loss": 0.4405,
      "step": 43464
    },
    {
      "epoch": 1.9946308108852278,
      "grad_norm": 0.4477350115776062,
      "learning_rate": 2.6183282129702426e-06,
      "loss": 0.3176,
      "step": 43465
    },
    {
      "epoch": 1.9946767013904823,
      "grad_norm": 0.530788004398346,
      "learning_rate": 2.618112630878443e-06,
      "loss": 0.4829,
      "step": 43466
    },
    {
      "epoch": 1.9947225918957368,
      "grad_norm": 0.49737969040870667,
      "learning_rate": 2.61789705451433e-06,
      "loss": 0.349,
      "step": 43467
    },
    {
      "epoch": 1.9947684824009912,
      "grad_norm": 0.4586632251739502,
      "learning_rate": 2.6176814838784238e-06,
      "loss": 0.3066,
      "step": 43468
    },
    {
      "epoch": 1.9948143729062457,
      "grad_norm": 0.48012715578079224,
      "learning_rate": 2.6174659189712416e-06,
      "loss": 0.3803,
      "step": 43469
    },
    {
      "epoch": 1.9948602634115002,
      "grad_norm": 0.5049107670783997,
      "learning_rate": 2.6172503597933014e-06,
      "loss": 0.3554,
      "step": 43470
    },
    {
      "epoch": 1.9949061539167547,
      "grad_norm": 0.4434470534324646,
      "learning_rate": 2.61703480634512e-06,
      "loss": 0.2907,
      "step": 43471
    },
    {
      "epoch": 1.9949520444220092,
      "grad_norm": 0.4836183786392212,
      "learning_rate": 2.61681925862722e-06,
      "loss": 0.3841,
      "step": 43472
    },
    {
      "epoch": 1.9949979349272635,
      "grad_norm": 0.47668492794036865,
      "learning_rate": 2.616603716640118e-06,
      "loss": 0.3858,
      "step": 43473
    },
    {
      "epoch": 1.995043825432518,
      "grad_norm": 0.4343286156654358,
      "learning_rate": 2.6163881803843293e-06,
      "loss": 0.2976,
      "step": 43474
    },
    {
      "epoch": 1.9950897159377725,
      "grad_norm": 0.46433010697364807,
      "learning_rate": 2.6161726498603773e-06,
      "loss": 0.3755,
      "step": 43475
    },
    {
      "epoch": 1.9951356064430268,
      "grad_norm": 0.48212745785713196,
      "learning_rate": 2.6159571250687776e-06,
      "loss": 0.3658,
      "step": 43476
    },
    {
      "epoch": 1.9951814969482813,
      "grad_norm": 0.4511442482471466,
      "learning_rate": 2.615741606010046e-06,
      "loss": 0.3159,
      "step": 43477
    },
    {
      "epoch": 1.9952273874535358,
      "grad_norm": 0.4912593960762024,
      "learning_rate": 2.615526092684706e-06,
      "loss": 0.3708,
      "step": 43478
    },
    {
      "epoch": 1.9952732779587903,
      "grad_norm": 0.4846009910106659,
      "learning_rate": 2.6153105850932727e-06,
      "loss": 0.3444,
      "step": 43479
    },
    {
      "epoch": 1.9953191684640448,
      "grad_norm": 0.47422948479652405,
      "learning_rate": 2.6150950832362652e-06,
      "loss": 0.373,
      "step": 43480
    },
    {
      "epoch": 1.9953650589692993,
      "grad_norm": 0.46811941266059875,
      "learning_rate": 2.6148795871142007e-06,
      "loss": 0.3504,
      "step": 43481
    },
    {
      "epoch": 1.9954109494745538,
      "grad_norm": 0.4349003732204437,
      "learning_rate": 2.6146640967275987e-06,
      "loss": 0.3015,
      "step": 43482
    },
    {
      "epoch": 1.9954568399798083,
      "grad_norm": 0.45451727509498596,
      "learning_rate": 2.6144486120769763e-06,
      "loss": 0.3226,
      "step": 43483
    },
    {
      "epoch": 1.9955027304850628,
      "grad_norm": 0.4717479646205902,
      "learning_rate": 2.6142331331628503e-06,
      "loss": 0.4057,
      "step": 43484
    },
    {
      "epoch": 1.9955486209903173,
      "grad_norm": 0.49039867520332336,
      "learning_rate": 2.6140176599857427e-06,
      "loss": 0.3888,
      "step": 43485
    },
    {
      "epoch": 1.9955945114955715,
      "grad_norm": 0.4888279139995575,
      "learning_rate": 2.61380219254617e-06,
      "loss": 0.3511,
      "step": 43486
    },
    {
      "epoch": 1.995640402000826,
      "grad_norm": 0.48961466550827026,
      "learning_rate": 2.613586730844647e-06,
      "loss": 0.4077,
      "step": 43487
    },
    {
      "epoch": 1.9956862925060805,
      "grad_norm": 0.5120751261711121,
      "learning_rate": 2.6133712748816974e-06,
      "loss": 0.4465,
      "step": 43488
    },
    {
      "epoch": 1.9957321830113348,
      "grad_norm": 0.4331657588481903,
      "learning_rate": 2.613155824657836e-06,
      "loss": 0.308,
      "step": 43489
    },
    {
      "epoch": 1.9957780735165893,
      "grad_norm": 0.47769105434417725,
      "learning_rate": 2.612940380173582e-06,
      "loss": 0.3384,
      "step": 43490
    },
    {
      "epoch": 1.9958239640218438,
      "grad_norm": 0.48808202147483826,
      "learning_rate": 2.612724941429453e-06,
      "loss": 0.3737,
      "step": 43491
    },
    {
      "epoch": 1.9958698545270983,
      "grad_norm": 0.44576549530029297,
      "learning_rate": 2.6125095084259667e-06,
      "loss": 0.3172,
      "step": 43492
    },
    {
      "epoch": 1.9959157450323528,
      "grad_norm": 0.47165411710739136,
      "learning_rate": 2.61229408116364e-06,
      "loss": 0.356,
      "step": 43493
    },
    {
      "epoch": 1.9959616355376073,
      "grad_norm": 0.4942033588886261,
      "learning_rate": 2.6120786596429947e-06,
      "loss": 0.3752,
      "step": 43494
    },
    {
      "epoch": 1.9960075260428618,
      "grad_norm": 0.4608750641345978,
      "learning_rate": 2.611863243864546e-06,
      "loss": 0.3586,
      "step": 43495
    },
    {
      "epoch": 1.9960534165481163,
      "grad_norm": 0.4616643786430359,
      "learning_rate": 2.611647833828811e-06,
      "loss": 0.3629,
      "step": 43496
    },
    {
      "epoch": 1.9960993070533708,
      "grad_norm": 0.5115818977355957,
      "learning_rate": 2.6114324295363102e-06,
      "loss": 0.3879,
      "step": 43497
    },
    {
      "epoch": 1.996145197558625,
      "grad_norm": 0.4367506802082062,
      "learning_rate": 2.611217030987562e-06,
      "loss": 0.3051,
      "step": 43498
    },
    {
      "epoch": 1.9961910880638796,
      "grad_norm": 0.4718990921974182,
      "learning_rate": 2.6110016381830793e-06,
      "loss": 0.3518,
      "step": 43499
    },
    {
      "epoch": 1.996236978569134,
      "grad_norm": 0.44062480330467224,
      "learning_rate": 2.6107862511233867e-06,
      "loss": 0.2815,
      "step": 43500
    },
    {
      "epoch": 1.9962828690743883,
      "grad_norm": 0.4524686634540558,
      "learning_rate": 2.6105708698089988e-06,
      "loss": 0.3264,
      "step": 43501
    },
    {
      "epoch": 1.9963287595796428,
      "grad_norm": 0.4581872522830963,
      "learning_rate": 2.6103554942404343e-06,
      "loss": 0.3202,
      "step": 43502
    },
    {
      "epoch": 1.9963746500848973,
      "grad_norm": 0.48497748374938965,
      "learning_rate": 2.61014012441821e-06,
      "loss": 0.3565,
      "step": 43503
    },
    {
      "epoch": 1.9964205405901518,
      "grad_norm": 0.4639185965061188,
      "learning_rate": 2.6099247603428444e-06,
      "loss": 0.3632,
      "step": 43504
    },
    {
      "epoch": 1.9964664310954063,
      "grad_norm": 0.4519108831882477,
      "learning_rate": 2.6097094020148538e-06,
      "loss": 0.3049,
      "step": 43505
    },
    {
      "epoch": 1.9965123216006608,
      "grad_norm": 0.4374118447303772,
      "learning_rate": 2.6094940494347596e-06,
      "loss": 0.3298,
      "step": 43506
    },
    {
      "epoch": 1.9965582121059153,
      "grad_norm": 0.47617262601852417,
      "learning_rate": 2.6092787026030775e-06,
      "loss": 0.3561,
      "step": 43507
    },
    {
      "epoch": 1.9966041026111698,
      "grad_norm": 0.4675796627998352,
      "learning_rate": 2.6090633615203244e-06,
      "loss": 0.3632,
      "step": 43508
    },
    {
      "epoch": 1.9966499931164243,
      "grad_norm": 0.4834240674972534,
      "learning_rate": 2.60884802618702e-06,
      "loss": 0.3542,
      "step": 43509
    },
    {
      "epoch": 1.9966958836216788,
      "grad_norm": 0.49119311571121216,
      "learning_rate": 2.6086326966036824e-06,
      "loss": 0.322,
      "step": 43510
    },
    {
      "epoch": 1.996741774126933,
      "grad_norm": 0.465612530708313,
      "learning_rate": 2.608417372770826e-06,
      "loss": 0.3572,
      "step": 43511
    },
    {
      "epoch": 1.9967876646321876,
      "grad_norm": 0.48576590418815613,
      "learning_rate": 2.6082020546889752e-06,
      "loss": 0.4164,
      "step": 43512
    },
    {
      "epoch": 1.996833555137442,
      "grad_norm": 0.49535486102104187,
      "learning_rate": 2.607986742358642e-06,
      "loss": 0.4019,
      "step": 43513
    },
    {
      "epoch": 1.9968794456426964,
      "grad_norm": 0.4790431261062622,
      "learning_rate": 2.6077714357803453e-06,
      "loss": 0.3295,
      "step": 43514
    },
    {
      "epoch": 1.9969253361479509,
      "grad_norm": 0.5595104098320007,
      "learning_rate": 2.607556134954601e-06,
      "loss": 0.4117,
      "step": 43515
    },
    {
      "epoch": 1.9969712266532054,
      "grad_norm": 0.43628984689712524,
      "learning_rate": 2.6073408398819316e-06,
      "loss": 0.3606,
      "step": 43516
    },
    {
      "epoch": 1.9970171171584599,
      "grad_norm": 0.4653662145137787,
      "learning_rate": 2.6071255505628523e-06,
      "loss": 0.3184,
      "step": 43517
    },
    {
      "epoch": 1.9970630076637144,
      "grad_norm": 0.45600298047065735,
      "learning_rate": 2.6069102669978787e-06,
      "loss": 0.3538,
      "step": 43518
    },
    {
      "epoch": 1.9971088981689689,
      "grad_norm": 0.509855329990387,
      "learning_rate": 2.6066949891875326e-06,
      "loss": 0.4693,
      "step": 43519
    },
    {
      "epoch": 1.9971547886742234,
      "grad_norm": 0.4904026985168457,
      "learning_rate": 2.60647971713233e-06,
      "loss": 0.4098,
      "step": 43520
    },
    {
      "epoch": 1.9972006791794779,
      "grad_norm": 0.47173231840133667,
      "learning_rate": 2.6062644508327862e-06,
      "loss": 0.311,
      "step": 43521
    },
    {
      "epoch": 1.9972465696847324,
      "grad_norm": 0.4544871151447296,
      "learning_rate": 2.6060491902894223e-06,
      "loss": 0.3138,
      "step": 43522
    },
    {
      "epoch": 1.9972924601899869,
      "grad_norm": 0.49053242802619934,
      "learning_rate": 2.605833935502755e-06,
      "loss": 0.3402,
      "step": 43523
    },
    {
      "epoch": 1.9973383506952411,
      "grad_norm": 0.4452246427536011,
      "learning_rate": 2.605618686473301e-06,
      "loss": 0.3184,
      "step": 43524
    },
    {
      "epoch": 1.9973842412004956,
      "grad_norm": 0.47043755650520325,
      "learning_rate": 2.6054034432015794e-06,
      "loss": 0.3605,
      "step": 43525
    },
    {
      "epoch": 1.9974301317057501,
      "grad_norm": 0.4630054235458374,
      "learning_rate": 2.6051882056881062e-06,
      "loss": 0.3691,
      "step": 43526
    },
    {
      "epoch": 1.9974760222110044,
      "grad_norm": 0.4433959424495697,
      "learning_rate": 2.6049729739333974e-06,
      "loss": 0.3093,
      "step": 43527
    },
    {
      "epoch": 1.997521912716259,
      "grad_norm": 0.47992223501205444,
      "learning_rate": 2.604757747937975e-06,
      "loss": 0.3347,
      "step": 43528
    },
    {
      "epoch": 1.9975678032215134,
      "grad_norm": 0.4607185423374176,
      "learning_rate": 2.604542527702354e-06,
      "loss": 0.3418,
      "step": 43529
    },
    {
      "epoch": 1.997613693726768,
      "grad_norm": 0.449396014213562,
      "learning_rate": 2.60432731322705e-06,
      "loss": 0.3384,
      "step": 43530
    },
    {
      "epoch": 1.9976595842320224,
      "grad_norm": 0.45522579550743103,
      "learning_rate": 2.604112104512585e-06,
      "loss": 0.2932,
      "step": 43531
    },
    {
      "epoch": 1.9977054747372769,
      "grad_norm": 0.4637390077114105,
      "learning_rate": 2.603896901559474e-06,
      "loss": 0.3068,
      "step": 43532
    },
    {
      "epoch": 1.9977513652425314,
      "grad_norm": 0.4483066499233246,
      "learning_rate": 2.6036817043682326e-06,
      "loss": 0.3018,
      "step": 43533
    },
    {
      "epoch": 1.9977972557477859,
      "grad_norm": 0.45455294847488403,
      "learning_rate": 2.6034665129393856e-06,
      "loss": 0.3726,
      "step": 43534
    },
    {
      "epoch": 1.9978431462530404,
      "grad_norm": 0.47806546092033386,
      "learning_rate": 2.6032513272734416e-06,
      "loss": 0.3473,
      "step": 43535
    },
    {
      "epoch": 1.9978890367582947,
      "grad_norm": 0.49040573835372925,
      "learning_rate": 2.603036147370921e-06,
      "loss": 0.3532,
      "step": 43536
    },
    {
      "epoch": 1.9979349272635492,
      "grad_norm": 0.4696485996246338,
      "learning_rate": 2.6028209732323433e-06,
      "loss": 0.3334,
      "step": 43537
    },
    {
      "epoch": 1.9979808177688037,
      "grad_norm": 0.45687347650527954,
      "learning_rate": 2.602605804858225e-06,
      "loss": 0.3422,
      "step": 43538
    },
    {
      "epoch": 1.9980267082740581,
      "grad_norm": 0.4481918513774872,
      "learning_rate": 2.6023906422490806e-06,
      "loss": 0.3017,
      "step": 43539
    },
    {
      "epoch": 1.9980725987793124,
      "grad_norm": 0.5024559497833252,
      "learning_rate": 2.6021754854054325e-06,
      "loss": 0.3967,
      "step": 43540
    },
    {
      "epoch": 1.998118489284567,
      "grad_norm": 0.42068129777908325,
      "learning_rate": 2.6019603343277955e-06,
      "loss": 0.2592,
      "step": 43541
    },
    {
      "epoch": 1.9981643797898214,
      "grad_norm": 0.4506835639476776,
      "learning_rate": 2.6017451890166844e-06,
      "loss": 0.3025,
      "step": 43542
    },
    {
      "epoch": 1.998210270295076,
      "grad_norm": 0.49793699383735657,
      "learning_rate": 2.6015300494726216e-06,
      "loss": 0.379,
      "step": 43543
    },
    {
      "epoch": 1.9982561608003304,
      "grad_norm": 0.4432838559150696,
      "learning_rate": 2.6013149156961226e-06,
      "loss": 0.3096,
      "step": 43544
    },
    {
      "epoch": 1.998302051305585,
      "grad_norm": 0.4561251699924469,
      "learning_rate": 2.6010997876877032e-06,
      "loss": 0.3037,
      "step": 43545
    },
    {
      "epoch": 1.9983479418108394,
      "grad_norm": 0.4678066372871399,
      "learning_rate": 2.6008846654478823e-06,
      "loss": 0.3208,
      "step": 43546
    },
    {
      "epoch": 1.998393832316094,
      "grad_norm": 0.4909549653530121,
      "learning_rate": 2.6006695489771766e-06,
      "loss": 0.3553,
      "step": 43547
    },
    {
      "epoch": 1.9984397228213484,
      "grad_norm": 0.48084524273872375,
      "learning_rate": 2.600454438276103e-06,
      "loss": 0.3903,
      "step": 43548
    },
    {
      "epoch": 1.9984856133266027,
      "grad_norm": 0.4458681046962738,
      "learning_rate": 2.600239333345178e-06,
      "loss": 0.2814,
      "step": 43549
    },
    {
      "epoch": 1.9985315038318572,
      "grad_norm": 0.44019216299057007,
      "learning_rate": 2.6000242341849214e-06,
      "loss": 0.297,
      "step": 43550
    },
    {
      "epoch": 1.9985773943371117,
      "grad_norm": 0.4973321557044983,
      "learning_rate": 2.599809140795849e-06,
      "loss": 0.4253,
      "step": 43551
    },
    {
      "epoch": 1.998623284842366,
      "grad_norm": 0.4755401015281677,
      "learning_rate": 2.599594053178476e-06,
      "loss": 0.3982,
      "step": 43552
    },
    {
      "epoch": 1.9986691753476205,
      "grad_norm": 0.4840502142906189,
      "learning_rate": 2.5993789713333245e-06,
      "loss": 0.404,
      "step": 43553
    },
    {
      "epoch": 1.998715065852875,
      "grad_norm": 0.45206618309020996,
      "learning_rate": 2.599163895260908e-06,
      "loss": 0.3312,
      "step": 43554
    },
    {
      "epoch": 1.9987609563581294,
      "grad_norm": 0.5010907649993896,
      "learning_rate": 2.5989488249617435e-06,
      "loss": 0.3994,
      "step": 43555
    },
    {
      "epoch": 1.998806846863384,
      "grad_norm": 0.44600456953048706,
      "learning_rate": 2.5987337604363527e-06,
      "loss": 0.3006,
      "step": 43556
    },
    {
      "epoch": 1.9988527373686384,
      "grad_norm": 0.4460684359073639,
      "learning_rate": 2.598518701685247e-06,
      "loss": 0.3398,
      "step": 43557
    },
    {
      "epoch": 1.998898627873893,
      "grad_norm": 0.4644332826137543,
      "learning_rate": 2.598303648708944e-06,
      "loss": 0.3096,
      "step": 43558
    },
    {
      "epoch": 1.9989445183791474,
      "grad_norm": 0.4907805323600769,
      "learning_rate": 2.5980886015079654e-06,
      "loss": 0.4468,
      "step": 43559
    },
    {
      "epoch": 1.998990408884402,
      "grad_norm": 0.45399779081344604,
      "learning_rate": 2.5978735600828254e-06,
      "loss": 0.3651,
      "step": 43560
    },
    {
      "epoch": 1.9990362993896564,
      "grad_norm": 0.4339732825756073,
      "learning_rate": 2.597658524434039e-06,
      "loss": 0.3163,
      "step": 43561
    },
    {
      "epoch": 1.9990821898949107,
      "grad_norm": 0.43160516023635864,
      "learning_rate": 2.597443494562128e-06,
      "loss": 0.3009,
      "step": 43562
    },
    {
      "epoch": 1.9991280804001652,
      "grad_norm": 0.4887740910053253,
      "learning_rate": 2.597228470467607e-06,
      "loss": 0.3787,
      "step": 43563
    },
    {
      "epoch": 1.9991739709054197,
      "grad_norm": 0.4786919355392456,
      "learning_rate": 2.5970134521509916e-06,
      "loss": 0.3914,
      "step": 43564
    },
    {
      "epoch": 1.999219861410674,
      "grad_norm": 0.42965057492256165,
      "learning_rate": 2.596798439612802e-06,
      "loss": 0.2872,
      "step": 43565
    },
    {
      "epoch": 1.9992657519159285,
      "grad_norm": 0.4697248041629791,
      "learning_rate": 2.5965834328535533e-06,
      "loss": 0.3275,
      "step": 43566
    },
    {
      "epoch": 1.999311642421183,
      "grad_norm": 0.4421330392360687,
      "learning_rate": 2.596368431873764e-06,
      "loss": 0.3179,
      "step": 43567
    },
    {
      "epoch": 1.9993575329264375,
      "grad_norm": 0.4759458601474762,
      "learning_rate": 2.596153436673949e-06,
      "loss": 0.3525,
      "step": 43568
    },
    {
      "epoch": 1.999403423431692,
      "grad_norm": 0.45938190817832947,
      "learning_rate": 2.595938447254627e-06,
      "loss": 0.3649,
      "step": 43569
    },
    {
      "epoch": 1.9994493139369465,
      "grad_norm": 0.4524313509464264,
      "learning_rate": 2.5957234636163114e-06,
      "loss": 0.3086,
      "step": 43570
    },
    {
      "epoch": 1.999495204442201,
      "grad_norm": 0.43937116861343384,
      "learning_rate": 2.595508485759525e-06,
      "loss": 0.3271,
      "step": 43571
    },
    {
      "epoch": 1.9995410949474555,
      "grad_norm": 0.4756285846233368,
      "learning_rate": 2.5952935136847813e-06,
      "loss": 0.3623,
      "step": 43572
    },
    {
      "epoch": 1.99958698545271,
      "grad_norm": 0.5210612416267395,
      "learning_rate": 2.5950785473925953e-06,
      "loss": 0.4454,
      "step": 43573
    },
    {
      "epoch": 1.9996328759579645,
      "grad_norm": 0.4370582401752472,
      "learning_rate": 2.5948635868834894e-06,
      "loss": 0.2792,
      "step": 43574
    },
    {
      "epoch": 1.9996787664632187,
      "grad_norm": 0.4702579379081726,
      "learning_rate": 2.5946486321579766e-06,
      "loss": 0.3379,
      "step": 43575
    },
    {
      "epoch": 1.9997246569684732,
      "grad_norm": 0.4482766389846802,
      "learning_rate": 2.5944336832165736e-06,
      "loss": 0.3035,
      "step": 43576
    },
    {
      "epoch": 1.9997705474737277,
      "grad_norm": 0.5091465711593628,
      "learning_rate": 2.5942187400597996e-06,
      "loss": 0.4438,
      "step": 43577
    },
    {
      "epoch": 1.999816437978982,
      "grad_norm": 0.4840451776981354,
      "learning_rate": 2.5940038026881724e-06,
      "loss": 0.4279,
      "step": 43578
    },
    {
      "epoch": 1.9998623284842365,
      "grad_norm": 0.5300149321556091,
      "learning_rate": 2.593788871102205e-06,
      "loss": 0.302,
      "step": 43579
    },
    {
      "epoch": 1.999908218989491,
      "grad_norm": 0.4825488030910492,
      "learning_rate": 2.5935739453024135e-06,
      "loss": 0.3597,
      "step": 43580
    },
    {
      "epoch": 1.9999541094947455,
      "grad_norm": 0.4886365234851837,
      "learning_rate": 2.5933590252893193e-06,
      "loss": 0.3945,
      "step": 43581
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.44774341583251953,
      "learning_rate": 2.5931441110634375e-06,
      "loss": 0.286,
      "step": 43582
    },
    {
      "epoch": 2.0000458905052545,
      "grad_norm": 0.4844712018966675,
      "learning_rate": 2.5929292026252823e-06,
      "loss": 0.3938,
      "step": 43583
    },
    {
      "epoch": 2.000091781010509,
      "grad_norm": 0.4449591040611267,
      "learning_rate": 2.5927142999753746e-06,
      "loss": 0.2723,
      "step": 43584
    },
    {
      "epoch": 2.0001376715157635,
      "grad_norm": 0.4971478283405304,
      "learning_rate": 2.5924994031142294e-06,
      "loss": 0.3517,
      "step": 43585
    },
    {
      "epoch": 2.000183562021018,
      "grad_norm": 0.4722493290901184,
      "learning_rate": 2.5922845120423613e-06,
      "loss": 0.3557,
      "step": 43586
    },
    {
      "epoch": 2.0002294525262725,
      "grad_norm": 0.4680596590042114,
      "learning_rate": 2.5920696267602907e-06,
      "loss": 0.2794,
      "step": 43587
    },
    {
      "epoch": 2.000275343031527,
      "grad_norm": 0.4398815631866455,
      "learning_rate": 2.591854747268533e-06,
      "loss": 0.3129,
      "step": 43588
    },
    {
      "epoch": 2.000321233536781,
      "grad_norm": 0.4684835970401764,
      "learning_rate": 2.591639873567604e-06,
      "loss": 0.3361,
      "step": 43589
    },
    {
      "epoch": 2.0003671240420355,
      "grad_norm": 0.44333454966545105,
      "learning_rate": 2.5914250056580216e-06,
      "loss": 0.2702,
      "step": 43590
    },
    {
      "epoch": 2.00041301454729,
      "grad_norm": 0.4783194661140442,
      "learning_rate": 2.591210143540302e-06,
      "loss": 0.3146,
      "step": 43591
    },
    {
      "epoch": 2.0004589050525445,
      "grad_norm": 0.4561416208744049,
      "learning_rate": 2.59099528721496e-06,
      "loss": 0.3159,
      "step": 43592
    },
    {
      "epoch": 2.000504795557799,
      "grad_norm": 0.4763440191745758,
      "learning_rate": 2.590780436682515e-06,
      "loss": 0.3386,
      "step": 43593
    },
    {
      "epoch": 2.0005506860630535,
      "grad_norm": 0.43661850690841675,
      "learning_rate": 2.5905655919434834e-06,
      "loss": 0.2594,
      "step": 43594
    },
    {
      "epoch": 2.000596576568308,
      "grad_norm": 0.6111595630645752,
      "learning_rate": 2.590350752998379e-06,
      "loss": 0.3965,
      "step": 43595
    },
    {
      "epoch": 2.0006424670735625,
      "grad_norm": 0.4455806612968445,
      "learning_rate": 2.590135919847723e-06,
      "loss": 0.2996,
      "step": 43596
    },
    {
      "epoch": 2.000688357578817,
      "grad_norm": 0.4861047565937042,
      "learning_rate": 2.589921092492028e-06,
      "loss": 0.3237,
      "step": 43597
    },
    {
      "epoch": 2.0007342480840715,
      "grad_norm": 0.4809759557247162,
      "learning_rate": 2.5897062709318117e-06,
      "loss": 0.3219,
      "step": 43598
    },
    {
      "epoch": 2.000780138589326,
      "grad_norm": 0.5701629519462585,
      "learning_rate": 2.5894914551675926e-06,
      "loss": 0.287,
      "step": 43599
    },
    {
      "epoch": 2.0008260290945805,
      "grad_norm": 0.4957137703895569,
      "learning_rate": 2.5892766451998873e-06,
      "loss": 0.3111,
      "step": 43600
    },
    {
      "epoch": 2.000871919599835,
      "grad_norm": 0.5069689750671387,
      "learning_rate": 2.5890618410292077e-06,
      "loss": 0.3482,
      "step": 43601
    },
    {
      "epoch": 2.000917810105089,
      "grad_norm": 0.5015645623207092,
      "learning_rate": 2.588847042656075e-06,
      "loss": 0.3287,
      "step": 43602
    },
    {
      "epoch": 2.0009637006103436,
      "grad_norm": 0.4182235598564148,
      "learning_rate": 2.588632250081004e-06,
      "loss": 0.2286,
      "step": 43603
    },
    {
      "epoch": 2.001009591115598,
      "grad_norm": 0.5139101147651672,
      "learning_rate": 2.58841746330451e-06,
      "loss": 0.3965,
      "step": 43604
    },
    {
      "epoch": 2.0010554816208526,
      "grad_norm": 0.547021746635437,
      "learning_rate": 2.588202682327112e-06,
      "loss": 0.4359,
      "step": 43605
    },
    {
      "epoch": 2.001101372126107,
      "grad_norm": 0.46635201573371887,
      "learning_rate": 2.5879879071493263e-06,
      "loss": 0.3207,
      "step": 43606
    },
    {
      "epoch": 2.0011472626313616,
      "grad_norm": 0.47318413853645325,
      "learning_rate": 2.587773137771666e-06,
      "loss": 0.3202,
      "step": 43607
    },
    {
      "epoch": 2.001193153136616,
      "grad_norm": 0.4387804865837097,
      "learning_rate": 2.587558374194652e-06,
      "loss": 0.2565,
      "step": 43608
    },
    {
      "epoch": 2.0012390436418706,
      "grad_norm": 0.48904040455818176,
      "learning_rate": 2.587343616418798e-06,
      "loss": 0.351,
      "step": 43609
    },
    {
      "epoch": 2.001284934147125,
      "grad_norm": 0.46633121371269226,
      "learning_rate": 2.5871288644446223e-06,
      "loss": 0.2942,
      "step": 43610
    },
    {
      "epoch": 2.0013308246523795,
      "grad_norm": 0.503555417060852,
      "learning_rate": 2.58691411827264e-06,
      "loss": 0.3625,
      "step": 43611
    },
    {
      "epoch": 2.001376715157634,
      "grad_norm": 0.4855799674987793,
      "learning_rate": 2.5866993779033665e-06,
      "loss": 0.3491,
      "step": 43612
    },
    {
      "epoch": 2.0014226056628885,
      "grad_norm": 0.46129336953163147,
      "learning_rate": 2.5864846433373203e-06,
      "loss": 0.3108,
      "step": 43613
    },
    {
      "epoch": 2.0014684961681426,
      "grad_norm": 0.4918213188648224,
      "learning_rate": 2.5862699145750147e-06,
      "loss": 0.3753,
      "step": 43614
    },
    {
      "epoch": 2.001514386673397,
      "grad_norm": 0.5084760785102844,
      "learning_rate": 2.5860551916169706e-06,
      "loss": 0.3494,
      "step": 43615
    },
    {
      "epoch": 2.0015602771786516,
      "grad_norm": 0.49756646156311035,
      "learning_rate": 2.585840474463701e-06,
      "loss": 0.3891,
      "step": 43616
    },
    {
      "epoch": 2.001606167683906,
      "grad_norm": 0.5025991201400757,
      "learning_rate": 2.5856257631157223e-06,
      "loss": 0.3863,
      "step": 43617
    },
    {
      "epoch": 2.0016520581891606,
      "grad_norm": 0.4629007577896118,
      "learning_rate": 2.5854110575735523e-06,
      "loss": 0.3063,
      "step": 43618
    },
    {
      "epoch": 2.001697948694415,
      "grad_norm": 0.4451930522918701,
      "learning_rate": 2.585196357837707e-06,
      "loss": 0.2831,
      "step": 43619
    },
    {
      "epoch": 2.0017438391996696,
      "grad_norm": 0.4724777936935425,
      "learning_rate": 2.584981663908701e-06,
      "loss": 0.2993,
      "step": 43620
    },
    {
      "epoch": 2.001789729704924,
      "grad_norm": 0.5392200946807861,
      "learning_rate": 2.5847669757870537e-06,
      "loss": 0.4277,
      "step": 43621
    },
    {
      "epoch": 2.0018356202101786,
      "grad_norm": 0.4594961702823639,
      "learning_rate": 2.584552293473281e-06,
      "loss": 0.2804,
      "step": 43622
    },
    {
      "epoch": 2.001881510715433,
      "grad_norm": 0.4628780782222748,
      "learning_rate": 2.584337616967894e-06,
      "loss": 0.3038,
      "step": 43623
    },
    {
      "epoch": 2.0019274012206876,
      "grad_norm": 0.4215259253978729,
      "learning_rate": 2.5841229462714147e-06,
      "loss": 0.2266,
      "step": 43624
    },
    {
      "epoch": 2.001973291725942,
      "grad_norm": 0.4617981016635895,
      "learning_rate": 2.5839082813843575e-06,
      "loss": 0.3087,
      "step": 43625
    },
    {
      "epoch": 2.0020191822311966,
      "grad_norm": 0.5508463382720947,
      "learning_rate": 2.583693622307236e-06,
      "loss": 0.4324,
      "step": 43626
    },
    {
      "epoch": 2.0020650727364506,
      "grad_norm": 0.46040359139442444,
      "learning_rate": 2.5834789690405705e-06,
      "loss": 0.3022,
      "step": 43627
    },
    {
      "epoch": 2.002110963241705,
      "grad_norm": 0.4900420308113098,
      "learning_rate": 2.583264321584876e-06,
      "loss": 0.3211,
      "step": 43628
    },
    {
      "epoch": 2.0021568537469596,
      "grad_norm": 0.49306097626686096,
      "learning_rate": 2.5830496799406656e-06,
      "loss": 0.407,
      "step": 43629
    },
    {
      "epoch": 2.002202744252214,
      "grad_norm": 0.4392761290073395,
      "learning_rate": 2.5828350441084597e-06,
      "loss": 0.3163,
      "step": 43630
    },
    {
      "epoch": 2.0022486347574686,
      "grad_norm": 0.4424388110637665,
      "learning_rate": 2.5826204140887735e-06,
      "loss": 0.2776,
      "step": 43631
    },
    {
      "epoch": 2.002294525262723,
      "grad_norm": 0.5118333697319031,
      "learning_rate": 2.5824057898821215e-06,
      "loss": 0.3852,
      "step": 43632
    },
    {
      "epoch": 2.0023404157679776,
      "grad_norm": 0.4602487087249756,
      "learning_rate": 2.582191171489021e-06,
      "loss": 0.299,
      "step": 43633
    },
    {
      "epoch": 2.002386306273232,
      "grad_norm": 0.43042048811912537,
      "learning_rate": 2.5819765589099873e-06,
      "loss": 0.2499,
      "step": 43634
    },
    {
      "epoch": 2.0024321967784866,
      "grad_norm": 0.49058640003204346,
      "learning_rate": 2.581761952145535e-06,
      "loss": 0.3622,
      "step": 43635
    },
    {
      "epoch": 2.002478087283741,
      "grad_norm": 0.47918495535850525,
      "learning_rate": 2.581547351196184e-06,
      "loss": 0.3376,
      "step": 43636
    },
    {
      "epoch": 2.0025239777889956,
      "grad_norm": 0.42966604232788086,
      "learning_rate": 2.581332756062448e-06,
      "loss": 0.242,
      "step": 43637
    },
    {
      "epoch": 2.00256986829425,
      "grad_norm": 0.4447372257709503,
      "learning_rate": 2.5811181667448416e-06,
      "loss": 0.3097,
      "step": 43638
    },
    {
      "epoch": 2.0026157587995046,
      "grad_norm": 0.4536151587963104,
      "learning_rate": 2.580903583243885e-06,
      "loss": 0.277,
      "step": 43639
    },
    {
      "epoch": 2.0026616493047587,
      "grad_norm": 0.4750029444694519,
      "learning_rate": 2.5806890055600916e-06,
      "loss": 0.2836,
      "step": 43640
    },
    {
      "epoch": 2.002707539810013,
      "grad_norm": 0.4889214336872101,
      "learning_rate": 2.5804744336939754e-06,
      "loss": 0.293,
      "step": 43641
    },
    {
      "epoch": 2.0027534303152676,
      "grad_norm": 0.4424329996109009,
      "learning_rate": 2.5802598676460565e-06,
      "loss": 0.2695,
      "step": 43642
    },
    {
      "epoch": 2.002799320820522,
      "grad_norm": 0.5081387758255005,
      "learning_rate": 2.5800453074168486e-06,
      "loss": 0.3827,
      "step": 43643
    },
    {
      "epoch": 2.0028452113257766,
      "grad_norm": 0.4818970263004303,
      "learning_rate": 2.5798307530068706e-06,
      "loss": 0.3247,
      "step": 43644
    },
    {
      "epoch": 2.002891101831031,
      "grad_norm": 0.45791763067245483,
      "learning_rate": 2.5796162044166315e-06,
      "loss": 0.3392,
      "step": 43645
    },
    {
      "epoch": 2.0029369923362856,
      "grad_norm": 0.5072755217552185,
      "learning_rate": 2.579401661646654e-06,
      "loss": 0.3315,
      "step": 43646
    },
    {
      "epoch": 2.00298288284154,
      "grad_norm": 0.5062328577041626,
      "learning_rate": 2.579187124697451e-06,
      "loss": 0.3721,
      "step": 43647
    },
    {
      "epoch": 2.0030287733467946,
      "grad_norm": 0.49300679564476013,
      "learning_rate": 2.578972593569537e-06,
      "loss": 0.3243,
      "step": 43648
    },
    {
      "epoch": 2.003074663852049,
      "grad_norm": 0.4685254991054535,
      "learning_rate": 2.5787580682634316e-06,
      "loss": 0.2989,
      "step": 43649
    },
    {
      "epoch": 2.0031205543573036,
      "grad_norm": 0.46845701336860657,
      "learning_rate": 2.5785435487796493e-06,
      "loss": 0.3075,
      "step": 43650
    },
    {
      "epoch": 2.003166444862558,
      "grad_norm": 0.48184698820114136,
      "learning_rate": 2.578329035118704e-06,
      "loss": 0.3277,
      "step": 43651
    },
    {
      "epoch": 2.003212335367812,
      "grad_norm": 0.48065608739852905,
      "learning_rate": 2.5781145272811144e-06,
      "loss": 0.3424,
      "step": 43652
    },
    {
      "epoch": 2.0032582258730667,
      "grad_norm": 0.49073874950408936,
      "learning_rate": 2.5779000252673946e-06,
      "loss": 0.3505,
      "step": 43653
    },
    {
      "epoch": 2.003304116378321,
      "grad_norm": 0.45567062497138977,
      "learning_rate": 2.5776855290780615e-06,
      "loss": 0.3347,
      "step": 43654
    },
    {
      "epoch": 2.0033500068835757,
      "grad_norm": 0.5064171552658081,
      "learning_rate": 2.5774710387136304e-06,
      "loss": 0.3519,
      "step": 43655
    },
    {
      "epoch": 2.00339589738883,
      "grad_norm": 0.5021344423294067,
      "learning_rate": 2.5772565541746163e-06,
      "loss": 0.3352,
      "step": 43656
    },
    {
      "epoch": 2.0034417878940847,
      "grad_norm": 0.5424817800521851,
      "learning_rate": 2.5770420754615338e-06,
      "loss": 0.3763,
      "step": 43657
    },
    {
      "epoch": 2.003487678399339,
      "grad_norm": 0.5128730535507202,
      "learning_rate": 2.5768276025749024e-06,
      "loss": 0.2587,
      "step": 43658
    },
    {
      "epoch": 2.0035335689045937,
      "grad_norm": 0.5240526795387268,
      "learning_rate": 2.5766131355152357e-06,
      "loss": 0.4059,
      "step": 43659
    },
    {
      "epoch": 2.003579459409848,
      "grad_norm": 0.49225014448165894,
      "learning_rate": 2.5763986742830476e-06,
      "loss": 0.3222,
      "step": 43660
    },
    {
      "epoch": 2.0036253499151027,
      "grad_norm": 0.4729706645011902,
      "learning_rate": 2.576184218878858e-06,
      "loss": 0.2936,
      "step": 43661
    },
    {
      "epoch": 2.003671240420357,
      "grad_norm": 0.4805581569671631,
      "learning_rate": 2.5759697693031797e-06,
      "loss": 0.319,
      "step": 43662
    },
    {
      "epoch": 2.0037171309256117,
      "grad_norm": 0.48805731534957886,
      "learning_rate": 2.575755325556528e-06,
      "loss": 0.3031,
      "step": 43663
    },
    {
      "epoch": 2.003763021430866,
      "grad_norm": 0.50248122215271,
      "learning_rate": 2.5755408876394213e-06,
      "loss": 0.3661,
      "step": 43664
    },
    {
      "epoch": 2.00380891193612,
      "grad_norm": 0.4822775721549988,
      "learning_rate": 2.575326455552373e-06,
      "loss": 0.3771,
      "step": 43665
    },
    {
      "epoch": 2.0038548024413747,
      "grad_norm": 0.5054476857185364,
      "learning_rate": 2.5751120292958996e-06,
      "loss": 0.3874,
      "step": 43666
    },
    {
      "epoch": 2.003900692946629,
      "grad_norm": 0.4908200800418854,
      "learning_rate": 2.574897608870517e-06,
      "loss": 0.3219,
      "step": 43667
    },
    {
      "epoch": 2.0039465834518837,
      "grad_norm": 0.4589777886867523,
      "learning_rate": 2.5746831942767396e-06,
      "loss": 0.2906,
      "step": 43668
    },
    {
      "epoch": 2.003992473957138,
      "grad_norm": 0.5172182321548462,
      "learning_rate": 2.5744687855150823e-06,
      "loss": 0.333,
      "step": 43669
    },
    {
      "epoch": 2.0040383644623927,
      "grad_norm": 0.4623895585536957,
      "learning_rate": 2.5742543825860633e-06,
      "loss": 0.2911,
      "step": 43670
    },
    {
      "epoch": 2.004084254967647,
      "grad_norm": 0.4870242476463318,
      "learning_rate": 2.574039985490198e-06,
      "loss": 0.3433,
      "step": 43671
    },
    {
      "epoch": 2.0041301454729017,
      "grad_norm": 0.4880375564098358,
      "learning_rate": 2.5738255942279977e-06,
      "loss": 0.3223,
      "step": 43672
    },
    {
      "epoch": 2.004176035978156,
      "grad_norm": 0.6126068830490112,
      "learning_rate": 2.5736112087999843e-06,
      "loss": 0.3578,
      "step": 43673
    },
    {
      "epoch": 2.0042219264834107,
      "grad_norm": 0.462011456489563,
      "learning_rate": 2.5733968292066693e-06,
      "loss": 0.3496,
      "step": 43674
    },
    {
      "epoch": 2.004267816988665,
      "grad_norm": 0.4952201843261719,
      "learning_rate": 2.573182455448569e-06,
      "loss": 0.397,
      "step": 43675
    },
    {
      "epoch": 2.0043137074939197,
      "grad_norm": 0.4475078284740448,
      "learning_rate": 2.572968087526199e-06,
      "loss": 0.2681,
      "step": 43676
    },
    {
      "epoch": 2.004359597999174,
      "grad_norm": 0.4546314775943756,
      "learning_rate": 2.5727537254400755e-06,
      "loss": 0.2892,
      "step": 43677
    },
    {
      "epoch": 2.0044054885044282,
      "grad_norm": 0.4949602782726288,
      "learning_rate": 2.5725393691907123e-06,
      "loss": 0.3274,
      "step": 43678
    },
    {
      "epoch": 2.0044513790096827,
      "grad_norm": 0.4624207317829132,
      "learning_rate": 2.5723250187786243e-06,
      "loss": 0.2685,
      "step": 43679
    },
    {
      "epoch": 2.0044972695149372,
      "grad_norm": 0.45165547728538513,
      "learning_rate": 2.5721106742043305e-06,
      "loss": 0.291,
      "step": 43680
    },
    {
      "epoch": 2.0045431600201917,
      "grad_norm": 0.4655570983886719,
      "learning_rate": 2.5718963354683443e-06,
      "loss": 0.2668,
      "step": 43681
    },
    {
      "epoch": 2.0045890505254462,
      "grad_norm": 0.44164353609085083,
      "learning_rate": 2.5716820025711785e-06,
      "loss": 0.2778,
      "step": 43682
    },
    {
      "epoch": 2.0046349410307007,
      "grad_norm": 0.4908074140548706,
      "learning_rate": 2.5714676755133537e-06,
      "loss": 0.4017,
      "step": 43683
    },
    {
      "epoch": 2.004680831535955,
      "grad_norm": 0.4459596276283264,
      "learning_rate": 2.5712533542953822e-06,
      "loss": 0.2518,
      "step": 43684
    },
    {
      "epoch": 2.0047267220412097,
      "grad_norm": 0.4482170641422272,
      "learning_rate": 2.571039038917779e-06,
      "loss": 0.2797,
      "step": 43685
    },
    {
      "epoch": 2.004772612546464,
      "grad_norm": 0.47261425852775574,
      "learning_rate": 2.5708247293810605e-06,
      "loss": 0.314,
      "step": 43686
    },
    {
      "epoch": 2.0048185030517187,
      "grad_norm": 0.4824172556400299,
      "learning_rate": 2.5706104256857435e-06,
      "loss": 0.3536,
      "step": 43687
    },
    {
      "epoch": 2.004864393556973,
      "grad_norm": 0.4346352517604828,
      "learning_rate": 2.5703961278323408e-06,
      "loss": 0.229,
      "step": 43688
    },
    {
      "epoch": 2.0049102840622277,
      "grad_norm": 0.48954832553863525,
      "learning_rate": 2.570181835821368e-06,
      "loss": 0.3503,
      "step": 43689
    },
    {
      "epoch": 2.004956174567482,
      "grad_norm": 0.5062331557273865,
      "learning_rate": 2.5699675496533422e-06,
      "loss": 0.3631,
      "step": 43690
    },
    {
      "epoch": 2.0050020650727363,
      "grad_norm": 0.44990062713623047,
      "learning_rate": 2.569753269328775e-06,
      "loss": 0.2617,
      "step": 43691
    },
    {
      "epoch": 2.0050479555779908,
      "grad_norm": 0.47855308651924133,
      "learning_rate": 2.5695389948481865e-06,
      "loss": 0.3125,
      "step": 43692
    },
    {
      "epoch": 2.0050938460832453,
      "grad_norm": 0.5123343467712402,
      "learning_rate": 2.5693247262120892e-06,
      "loss": 0.3704,
      "step": 43693
    },
    {
      "epoch": 2.0051397365884998,
      "grad_norm": 0.47852420806884766,
      "learning_rate": 2.5691104634209974e-06,
      "loss": 0.3341,
      "step": 43694
    },
    {
      "epoch": 2.0051856270937543,
      "grad_norm": 0.4693133533000946,
      "learning_rate": 2.5688962064754287e-06,
      "loss": 0.3347,
      "step": 43695
    },
    {
      "epoch": 2.0052315175990088,
      "grad_norm": 0.5039023160934448,
      "learning_rate": 2.568681955375898e-06,
      "loss": 0.3449,
      "step": 43696
    },
    {
      "epoch": 2.0052774081042632,
      "grad_norm": 0.46535900235176086,
      "learning_rate": 2.5684677101229198e-06,
      "loss": 0.2907,
      "step": 43697
    },
    {
      "epoch": 2.0053232986095177,
      "grad_norm": 0.5254426002502441,
      "learning_rate": 2.568253470717009e-06,
      "loss": 0.3968,
      "step": 43698
    },
    {
      "epoch": 2.0053691891147722,
      "grad_norm": 0.5210719108581543,
      "learning_rate": 2.568039237158682e-06,
      "loss": 0.3821,
      "step": 43699
    },
    {
      "epoch": 2.0054150796200267,
      "grad_norm": 0.42893901467323303,
      "learning_rate": 2.5678250094484504e-06,
      "loss": 0.2279,
      "step": 43700
    },
    {
      "epoch": 2.0054609701252812,
      "grad_norm": 0.4883652329444885,
      "learning_rate": 2.5676107875868346e-06,
      "loss": 0.3264,
      "step": 43701
    },
    {
      "epoch": 2.0055068606305357,
      "grad_norm": 0.4708270728588104,
      "learning_rate": 2.5673965715743465e-06,
      "loss": 0.292,
      "step": 43702
    },
    {
      "epoch": 2.00555275113579,
      "grad_norm": 0.5141642689704895,
      "learning_rate": 2.5671823614115e-06,
      "loss": 0.358,
      "step": 43703
    },
    {
      "epoch": 2.0055986416410443,
      "grad_norm": 0.4907875955104828,
      "learning_rate": 2.5669681570988146e-06,
      "loss": 0.3216,
      "step": 43704
    },
    {
      "epoch": 2.005644532146299,
      "grad_norm": 0.49331778287887573,
      "learning_rate": 2.566753958636802e-06,
      "loss": 0.392,
      "step": 43705
    },
    {
      "epoch": 2.0056904226515533,
      "grad_norm": 0.4538797438144684,
      "learning_rate": 2.566539766025977e-06,
      "loss": 0.2693,
      "step": 43706
    },
    {
      "epoch": 2.005736313156808,
      "grad_norm": 0.502259373664856,
      "learning_rate": 2.5663255792668578e-06,
      "loss": 0.3497,
      "step": 43707
    },
    {
      "epoch": 2.0057822036620623,
      "grad_norm": 0.4454059600830078,
      "learning_rate": 2.566111398359957e-06,
      "loss": 0.2641,
      "step": 43708
    },
    {
      "epoch": 2.0058280941673168,
      "grad_norm": 0.46292680501937866,
      "learning_rate": 2.5658972233057904e-06,
      "loss": 0.2933,
      "step": 43709
    },
    {
      "epoch": 2.0058739846725713,
      "grad_norm": 0.5193755626678467,
      "learning_rate": 2.565683054104873e-06,
      "loss": 0.3277,
      "step": 43710
    },
    {
      "epoch": 2.0059198751778258,
      "grad_norm": 0.4787512719631195,
      "learning_rate": 2.565468890757719e-06,
      "loss": 0.3289,
      "step": 43711
    },
    {
      "epoch": 2.0059657656830803,
      "grad_norm": 0.44748401641845703,
      "learning_rate": 2.5652547332648446e-06,
      "loss": 0.275,
      "step": 43712
    },
    {
      "epoch": 2.0060116561883348,
      "grad_norm": 0.4969683885574341,
      "learning_rate": 2.565040581626762e-06,
      "loss": 0.312,
      "step": 43713
    },
    {
      "epoch": 2.0060575466935893,
      "grad_norm": 0.46029752492904663,
      "learning_rate": 2.5648264358439903e-06,
      "loss": 0.2917,
      "step": 43714
    },
    {
      "epoch": 2.0061034371988438,
      "grad_norm": 0.4765859842300415,
      "learning_rate": 2.5646122959170417e-06,
      "loss": 0.3256,
      "step": 43715
    },
    {
      "epoch": 2.006149327704098,
      "grad_norm": 0.4484683573246002,
      "learning_rate": 2.5643981618464307e-06,
      "loss": 0.2923,
      "step": 43716
    },
    {
      "epoch": 2.0061952182093523,
      "grad_norm": 0.5032331943511963,
      "learning_rate": 2.564184033632675e-06,
      "loss": 0.3017,
      "step": 43717
    },
    {
      "epoch": 2.006241108714607,
      "grad_norm": 0.4672977030277252,
      "learning_rate": 2.5639699112762883e-06,
      "loss": 0.337,
      "step": 43718
    },
    {
      "epoch": 2.0062869992198613,
      "grad_norm": 0.47363749146461487,
      "learning_rate": 2.563755794777785e-06,
      "loss": 0.3519,
      "step": 43719
    },
    {
      "epoch": 2.006332889725116,
      "grad_norm": 0.5113139748573303,
      "learning_rate": 2.56354168413768e-06,
      "loss": 0.3747,
      "step": 43720
    },
    {
      "epoch": 2.0063787802303703,
      "grad_norm": 0.4829648733139038,
      "learning_rate": 2.563327579356488e-06,
      "loss": 0.3398,
      "step": 43721
    },
    {
      "epoch": 2.006424670735625,
      "grad_norm": 0.4470955431461334,
      "learning_rate": 2.5631134804347226e-06,
      "loss": 0.2696,
      "step": 43722
    },
    {
      "epoch": 2.0064705612408793,
      "grad_norm": 0.47903206944465637,
      "learning_rate": 2.5628993873729025e-06,
      "loss": 0.3206,
      "step": 43723
    },
    {
      "epoch": 2.006516451746134,
      "grad_norm": 0.49464279413223267,
      "learning_rate": 2.562685300171539e-06,
      "loss": 0.3506,
      "step": 43724
    },
    {
      "epoch": 2.0065623422513883,
      "grad_norm": 0.48042237758636475,
      "learning_rate": 2.5624712188311473e-06,
      "loss": 0.3053,
      "step": 43725
    },
    {
      "epoch": 2.006608232756643,
      "grad_norm": 0.5060098767280579,
      "learning_rate": 2.5622571433522447e-06,
      "loss": 0.3401,
      "step": 43726
    },
    {
      "epoch": 2.0066541232618973,
      "grad_norm": 0.46541112661361694,
      "learning_rate": 2.562043073735344e-06,
      "loss": 0.3133,
      "step": 43727
    },
    {
      "epoch": 2.006700013767152,
      "grad_norm": 0.4658122956752777,
      "learning_rate": 2.5618290099809585e-06,
      "loss": 0.3233,
      "step": 43728
    },
    {
      "epoch": 2.006745904272406,
      "grad_norm": 0.45811259746551514,
      "learning_rate": 2.5616149520896065e-06,
      "loss": 0.3122,
      "step": 43729
    },
    {
      "epoch": 2.0067917947776603,
      "grad_norm": 0.4856727421283722,
      "learning_rate": 2.5614009000618013e-06,
      "loss": 0.3499,
      "step": 43730
    },
    {
      "epoch": 2.006837685282915,
      "grad_norm": 0.5311927199363708,
      "learning_rate": 2.561186853898057e-06,
      "loss": 0.3399,
      "step": 43731
    },
    {
      "epoch": 2.0068835757881693,
      "grad_norm": 0.4651277959346771,
      "learning_rate": 2.560972813598889e-06,
      "loss": 0.3344,
      "step": 43732
    },
    {
      "epoch": 2.006929466293424,
      "grad_norm": 0.4807904362678528,
      "learning_rate": 2.560758779164811e-06,
      "loss": 0.3016,
      "step": 43733
    },
    {
      "epoch": 2.0069753567986783,
      "grad_norm": 0.5431494116783142,
      "learning_rate": 2.5605447505963364e-06,
      "loss": 0.4372,
      "step": 43734
    },
    {
      "epoch": 2.007021247303933,
      "grad_norm": 0.48266854882240295,
      "learning_rate": 2.560330727893985e-06,
      "loss": 0.2769,
      "step": 43735
    },
    {
      "epoch": 2.0070671378091873,
      "grad_norm": 0.4945589601993561,
      "learning_rate": 2.560116711058267e-06,
      "loss": 0.3028,
      "step": 43736
    },
    {
      "epoch": 2.007113028314442,
      "grad_norm": 0.47400084137916565,
      "learning_rate": 2.559902700089696e-06,
      "loss": 0.3185,
      "step": 43737
    },
    {
      "epoch": 2.0071589188196963,
      "grad_norm": 0.4820843040943146,
      "learning_rate": 2.5596886949887916e-06,
      "loss": 0.2952,
      "step": 43738
    },
    {
      "epoch": 2.007204809324951,
      "grad_norm": 0.43015116453170776,
      "learning_rate": 2.559474695756066e-06,
      "loss": 0.2309,
      "step": 43739
    },
    {
      "epoch": 2.0072506998302053,
      "grad_norm": 0.490279883146286,
      "learning_rate": 2.5592607023920324e-06,
      "loss": 0.3398,
      "step": 43740
    },
    {
      "epoch": 2.0072965903354594,
      "grad_norm": 0.4648468494415283,
      "learning_rate": 2.559046714897207e-06,
      "loss": 0.3128,
      "step": 43741
    },
    {
      "epoch": 2.007342480840714,
      "grad_norm": 0.48699691891670227,
      "learning_rate": 2.5588327332721043e-06,
      "loss": 0.3375,
      "step": 43742
    },
    {
      "epoch": 2.0073883713459684,
      "grad_norm": 0.4886507987976074,
      "learning_rate": 2.558618757517238e-06,
      "loss": 0.322,
      "step": 43743
    },
    {
      "epoch": 2.007434261851223,
      "grad_norm": 0.5255902409553528,
      "learning_rate": 2.5584047876331207e-06,
      "loss": 0.3953,
      "step": 43744
    },
    {
      "epoch": 2.0074801523564774,
      "grad_norm": 0.46543624997138977,
      "learning_rate": 2.558190823620272e-06,
      "loss": 0.2943,
      "step": 43745
    },
    {
      "epoch": 2.007526042861732,
      "grad_norm": 0.4834055006504059,
      "learning_rate": 2.557976865479203e-06,
      "loss": 0.3541,
      "step": 43746
    },
    {
      "epoch": 2.0075719333669864,
      "grad_norm": 0.4499962627887726,
      "learning_rate": 2.5577629132104275e-06,
      "loss": 0.3255,
      "step": 43747
    },
    {
      "epoch": 2.007617823872241,
      "grad_norm": 0.4805598855018616,
      "learning_rate": 2.5575489668144633e-06,
      "loss": 0.3478,
      "step": 43748
    },
    {
      "epoch": 2.0076637143774954,
      "grad_norm": 0.4580801725387573,
      "learning_rate": 2.557335026291823e-06,
      "loss": 0.3005,
      "step": 43749
    },
    {
      "epoch": 2.00770960488275,
      "grad_norm": 0.4909159541130066,
      "learning_rate": 2.5571210916430187e-06,
      "loss": 0.3548,
      "step": 43750
    },
    {
      "epoch": 2.0077554953880044,
      "grad_norm": 0.4659135341644287,
      "learning_rate": 2.556907162868569e-06,
      "loss": 0.2711,
      "step": 43751
    },
    {
      "epoch": 2.007801385893259,
      "grad_norm": 0.4905852675437927,
      "learning_rate": 2.5566932399689876e-06,
      "loss": 0.3449,
      "step": 43752
    },
    {
      "epoch": 2.0078472763985133,
      "grad_norm": 0.49415165185928345,
      "learning_rate": 2.556479322944787e-06,
      "loss": 0.3656,
      "step": 43753
    },
    {
      "epoch": 2.0078931669037674,
      "grad_norm": 0.4498327672481537,
      "learning_rate": 2.5562654117964824e-06,
      "loss": 0.2534,
      "step": 43754
    },
    {
      "epoch": 2.007939057409022,
      "grad_norm": 0.46590033173561096,
      "learning_rate": 2.5560515065245883e-06,
      "loss": 0.341,
      "step": 43755
    },
    {
      "epoch": 2.0079849479142764,
      "grad_norm": 0.4729540944099426,
      "learning_rate": 2.5558376071296175e-06,
      "loss": 0.3385,
      "step": 43756
    },
    {
      "epoch": 2.008030838419531,
      "grad_norm": 0.5121253728866577,
      "learning_rate": 2.5556237136120877e-06,
      "loss": 0.339,
      "step": 43757
    },
    {
      "epoch": 2.0080767289247854,
      "grad_norm": 0.4590664803981781,
      "learning_rate": 2.555409825972511e-06,
      "loss": 0.3299,
      "step": 43758
    },
    {
      "epoch": 2.00812261943004,
      "grad_norm": 0.4793297052383423,
      "learning_rate": 2.5551959442114006e-06,
      "loss": 0.326,
      "step": 43759
    },
    {
      "epoch": 2.0081685099352944,
      "grad_norm": 0.5160317420959473,
      "learning_rate": 2.5549820683292737e-06,
      "loss": 0.2978,
      "step": 43760
    },
    {
      "epoch": 2.008214400440549,
      "grad_norm": 0.520672082901001,
      "learning_rate": 2.5547681983266436e-06,
      "loss": 0.3882,
      "step": 43761
    },
    {
      "epoch": 2.0082602909458034,
      "grad_norm": 0.4532833397388458,
      "learning_rate": 2.554554334204024e-06,
      "loss": 0.2522,
      "step": 43762
    },
    {
      "epoch": 2.008306181451058,
      "grad_norm": 0.4701540768146515,
      "learning_rate": 2.5543404759619294e-06,
      "loss": 0.3156,
      "step": 43763
    },
    {
      "epoch": 2.0083520719563124,
      "grad_norm": 0.4838799238204956,
      "learning_rate": 2.5541266236008745e-06,
      "loss": 0.3093,
      "step": 43764
    },
    {
      "epoch": 2.008397962461567,
      "grad_norm": 0.48567995429039,
      "learning_rate": 2.5539127771213712e-06,
      "loss": 0.3696,
      "step": 43765
    },
    {
      "epoch": 2.0084438529668214,
      "grad_norm": 0.48994317650794983,
      "learning_rate": 2.5536989365239375e-06,
      "loss": 0.3378,
      "step": 43766
    },
    {
      "epoch": 2.0084897434720754,
      "grad_norm": 0.5031680464744568,
      "learning_rate": 2.5534851018090857e-06,
      "loss": 0.3772,
      "step": 43767
    },
    {
      "epoch": 2.00853563397733,
      "grad_norm": 0.4770662188529968,
      "learning_rate": 2.553271272977328e-06,
      "loss": 0.3502,
      "step": 43768
    },
    {
      "epoch": 2.0085815244825844,
      "grad_norm": 0.4693060517311096,
      "learning_rate": 2.5530574500291826e-06,
      "loss": 0.2741,
      "step": 43769
    },
    {
      "epoch": 2.008627414987839,
      "grad_norm": 0.5177751779556274,
      "learning_rate": 2.5528436329651625e-06,
      "loss": 0.428,
      "step": 43770
    },
    {
      "epoch": 2.0086733054930934,
      "grad_norm": 0.4525376856327057,
      "learning_rate": 2.552629821785778e-06,
      "loss": 0.2765,
      "step": 43771
    },
    {
      "epoch": 2.008719195998348,
      "grad_norm": 0.5076024532318115,
      "learning_rate": 2.552416016491549e-06,
      "loss": 0.334,
      "step": 43772
    },
    {
      "epoch": 2.0087650865036024,
      "grad_norm": 0.4827359914779663,
      "learning_rate": 2.5522022170829874e-06,
      "loss": 0.3518,
      "step": 43773
    },
    {
      "epoch": 2.008810977008857,
      "grad_norm": 0.4879688620567322,
      "learning_rate": 2.551988423560606e-06,
      "loss": 0.3178,
      "step": 43774
    },
    {
      "epoch": 2.0088568675141114,
      "grad_norm": 0.5137057304382324,
      "learning_rate": 2.551774635924921e-06,
      "loss": 0.3994,
      "step": 43775
    },
    {
      "epoch": 2.008902758019366,
      "grad_norm": 0.4939470589160919,
      "learning_rate": 2.551560854176444e-06,
      "loss": 0.314,
      "step": 43776
    },
    {
      "epoch": 2.0089486485246204,
      "grad_norm": 0.4707486629486084,
      "learning_rate": 2.5513470783156912e-06,
      "loss": 0.3004,
      "step": 43777
    },
    {
      "epoch": 2.008994539029875,
      "grad_norm": 0.48745715618133545,
      "learning_rate": 2.5511333083431734e-06,
      "loss": 0.3689,
      "step": 43778
    },
    {
      "epoch": 2.0090404295351294,
      "grad_norm": 0.45244458317756653,
      "learning_rate": 2.550919544259409e-06,
      "loss": 0.2646,
      "step": 43779
    },
    {
      "epoch": 2.0090863200403835,
      "grad_norm": 0.534142255783081,
      "learning_rate": 2.550705786064911e-06,
      "loss": 0.4078,
      "step": 43780
    },
    {
      "epoch": 2.009132210545638,
      "grad_norm": 0.4892018437385559,
      "learning_rate": 2.5504920337601902e-06,
      "loss": 0.3102,
      "step": 43781
    },
    {
      "epoch": 2.0091781010508925,
      "grad_norm": 0.49181580543518066,
      "learning_rate": 2.550278287345765e-06,
      "loss": 0.3335,
      "step": 43782
    },
    {
      "epoch": 2.009223991556147,
      "grad_norm": 0.5097191333770752,
      "learning_rate": 2.5500645468221474e-06,
      "loss": 0.353,
      "step": 43783
    },
    {
      "epoch": 2.0092698820614014,
      "grad_norm": 0.46487361192703247,
      "learning_rate": 2.549850812189851e-06,
      "loss": 0.2839,
      "step": 43784
    },
    {
      "epoch": 2.009315772566656,
      "grad_norm": 0.4459165334701538,
      "learning_rate": 2.54963708344939e-06,
      "loss": 0.2541,
      "step": 43785
    },
    {
      "epoch": 2.0093616630719104,
      "grad_norm": 0.4816977083683014,
      "learning_rate": 2.5494233606012788e-06,
      "loss": 0.3567,
      "step": 43786
    },
    {
      "epoch": 2.009407553577165,
      "grad_norm": 0.4885375201702118,
      "learning_rate": 2.549209643646029e-06,
      "loss": 0.3169,
      "step": 43787
    },
    {
      "epoch": 2.0094534440824194,
      "grad_norm": 0.5076816082000732,
      "learning_rate": 2.5489959325841585e-06,
      "loss": 0.3181,
      "step": 43788
    },
    {
      "epoch": 2.009499334587674,
      "grad_norm": 0.49200117588043213,
      "learning_rate": 2.5487822274161787e-06,
      "loss": 0.3491,
      "step": 43789
    },
    {
      "epoch": 2.0095452250929284,
      "grad_norm": 0.48720628023147583,
      "learning_rate": 2.5485685281426033e-06,
      "loss": 0.301,
      "step": 43790
    },
    {
      "epoch": 2.009591115598183,
      "grad_norm": 0.5126504898071289,
      "learning_rate": 2.5483548347639474e-06,
      "loss": 0.3331,
      "step": 43791
    },
    {
      "epoch": 2.009637006103437,
      "grad_norm": 0.43541908264160156,
      "learning_rate": 2.548141147280725e-06,
      "loss": 0.2505,
      "step": 43792
    },
    {
      "epoch": 2.0096828966086915,
      "grad_norm": 0.4436420798301697,
      "learning_rate": 2.547927465693447e-06,
      "loss": 0.2772,
      "step": 43793
    },
    {
      "epoch": 2.009728787113946,
      "grad_norm": 0.4911344349384308,
      "learning_rate": 2.5477137900026326e-06,
      "loss": 0.3056,
      "step": 43794
    },
    {
      "epoch": 2.0097746776192005,
      "grad_norm": 0.5293405055999756,
      "learning_rate": 2.5475001202087917e-06,
      "loss": 0.3492,
      "step": 43795
    },
    {
      "epoch": 2.009820568124455,
      "grad_norm": 0.5054666996002197,
      "learning_rate": 2.547286456312439e-06,
      "loss": 0.3635,
      "step": 43796
    },
    {
      "epoch": 2.0098664586297095,
      "grad_norm": 0.49643296003341675,
      "learning_rate": 2.5470727983140886e-06,
      "loss": 0.3043,
      "step": 43797
    },
    {
      "epoch": 2.009912349134964,
      "grad_norm": 0.4807972311973572,
      "learning_rate": 2.5468591462142535e-06,
      "loss": 0.3305,
      "step": 43798
    },
    {
      "epoch": 2.0099582396402185,
      "grad_norm": 0.4591931402683258,
      "learning_rate": 2.546645500013446e-06,
      "loss": 0.2779,
      "step": 43799
    },
    {
      "epoch": 2.010004130145473,
      "grad_norm": 0.44251781702041626,
      "learning_rate": 2.546431859712184e-06,
      "loss": 0.2743,
      "step": 43800
    },
    {
      "epoch": 2.0100500206507275,
      "grad_norm": 0.4940275251865387,
      "learning_rate": 2.546218225310979e-06,
      "loss": 0.3481,
      "step": 43801
    },
    {
      "epoch": 2.010095911155982,
      "grad_norm": 0.46646368503570557,
      "learning_rate": 2.546004596810343e-06,
      "loss": 0.3036,
      "step": 43802
    },
    {
      "epoch": 2.0101418016612365,
      "grad_norm": 0.45650115609169006,
      "learning_rate": 2.5457909742107934e-06,
      "loss": 0.2938,
      "step": 43803
    },
    {
      "epoch": 2.010187692166491,
      "grad_norm": 0.4809924364089966,
      "learning_rate": 2.5455773575128418e-06,
      "loss": 0.3659,
      "step": 43804
    },
    {
      "epoch": 2.010233582671745,
      "grad_norm": 0.42802879214286804,
      "learning_rate": 2.5453637467170013e-06,
      "loss": 0.2596,
      "step": 43805
    },
    {
      "epoch": 2.0102794731769995,
      "grad_norm": 0.5067252516746521,
      "learning_rate": 2.5451501418237866e-06,
      "loss": 0.3875,
      "step": 43806
    },
    {
      "epoch": 2.010325363682254,
      "grad_norm": 0.4658513367176056,
      "learning_rate": 2.5449365428337114e-06,
      "loss": 0.3121,
      "step": 43807
    },
    {
      "epoch": 2.0103712541875085,
      "grad_norm": 0.4886285662651062,
      "learning_rate": 2.544722949747289e-06,
      "loss": 0.3596,
      "step": 43808
    },
    {
      "epoch": 2.010417144692763,
      "grad_norm": 0.4542262554168701,
      "learning_rate": 2.544509362565031e-06,
      "loss": 0.2742,
      "step": 43809
    },
    {
      "epoch": 2.0104630351980175,
      "grad_norm": 0.4549892544746399,
      "learning_rate": 2.544295781287455e-06,
      "loss": 0.2798,
      "step": 43810
    },
    {
      "epoch": 2.010508925703272,
      "grad_norm": 0.48968493938446045,
      "learning_rate": 2.5440822059150726e-06,
      "loss": 0.3819,
      "step": 43811
    },
    {
      "epoch": 2.0105548162085265,
      "grad_norm": 0.513505220413208,
      "learning_rate": 2.5438686364483956e-06,
      "loss": 0.2875,
      "step": 43812
    },
    {
      "epoch": 2.010600706713781,
      "grad_norm": 0.4749179482460022,
      "learning_rate": 2.5436550728879404e-06,
      "loss": 0.3071,
      "step": 43813
    },
    {
      "epoch": 2.0106465972190355,
      "grad_norm": 0.5001460313796997,
      "learning_rate": 2.5434415152342207e-06,
      "loss": 0.3099,
      "step": 43814
    },
    {
      "epoch": 2.01069248772429,
      "grad_norm": 0.5225446820259094,
      "learning_rate": 2.543227963487746e-06,
      "loss": 0.4213,
      "step": 43815
    },
    {
      "epoch": 2.0107383782295445,
      "grad_norm": 0.431170254945755,
      "learning_rate": 2.5430144176490345e-06,
      "loss": 0.2767,
      "step": 43816
    },
    {
      "epoch": 2.010784268734799,
      "grad_norm": 0.4868049621582031,
      "learning_rate": 2.5428008777185988e-06,
      "loss": 0.3152,
      "step": 43817
    },
    {
      "epoch": 2.010830159240053,
      "grad_norm": 0.46672624349594116,
      "learning_rate": 2.5425873436969505e-06,
      "loss": 0.289,
      "step": 43818
    },
    {
      "epoch": 2.0108760497453075,
      "grad_norm": 0.47404828667640686,
      "learning_rate": 2.542373815584605e-06,
      "loss": 0.3301,
      "step": 43819
    },
    {
      "epoch": 2.010921940250562,
      "grad_norm": 0.4752424359321594,
      "learning_rate": 2.5421602933820744e-06,
      "loss": 0.3066,
      "step": 43820
    },
    {
      "epoch": 2.0109678307558165,
      "grad_norm": 0.4813607931137085,
      "learning_rate": 2.541946777089871e-06,
      "loss": 0.344,
      "step": 43821
    },
    {
      "epoch": 2.011013721261071,
      "grad_norm": 0.5086230635643005,
      "learning_rate": 2.541733266708511e-06,
      "loss": 0.3095,
      "step": 43822
    },
    {
      "epoch": 2.0110596117663255,
      "grad_norm": 0.5241252779960632,
      "learning_rate": 2.5415197622385067e-06,
      "loss": 0.3673,
      "step": 43823
    },
    {
      "epoch": 2.01110550227158,
      "grad_norm": 0.520601212978363,
      "learning_rate": 2.54130626368037e-06,
      "loss": 0.367,
      "step": 43824
    },
    {
      "epoch": 2.0111513927768345,
      "grad_norm": 0.5152016878128052,
      "learning_rate": 2.5410927710346176e-06,
      "loss": 0.3981,
      "step": 43825
    },
    {
      "epoch": 2.011197283282089,
      "grad_norm": 0.5219717025756836,
      "learning_rate": 2.5408792843017605e-06,
      "loss": 0.3522,
      "step": 43826
    },
    {
      "epoch": 2.0112431737873435,
      "grad_norm": 0.4979209005832672,
      "learning_rate": 2.5406658034823135e-06,
      "loss": 0.3288,
      "step": 43827
    },
    {
      "epoch": 2.011289064292598,
      "grad_norm": 0.5158754587173462,
      "learning_rate": 2.5404523285767883e-06,
      "loss": 0.3341,
      "step": 43828
    },
    {
      "epoch": 2.0113349547978525,
      "grad_norm": 0.5207123756408691,
      "learning_rate": 2.5402388595857e-06,
      "loss": 0.3377,
      "step": 43829
    },
    {
      "epoch": 2.0113808453031066,
      "grad_norm": 0.49175387620925903,
      "learning_rate": 2.5400253965095577e-06,
      "loss": 0.3231,
      "step": 43830
    },
    {
      "epoch": 2.011426735808361,
      "grad_norm": 0.5107274055480957,
      "learning_rate": 2.5398119393488806e-06,
      "loss": 0.3632,
      "step": 43831
    },
    {
      "epoch": 2.0114726263136156,
      "grad_norm": 0.4979594945907593,
      "learning_rate": 2.539598488104179e-06,
      "loss": 0.3428,
      "step": 43832
    },
    {
      "epoch": 2.01151851681887,
      "grad_norm": 0.48799246549606323,
      "learning_rate": 2.5393850427759647e-06,
      "loss": 0.3225,
      "step": 43833
    },
    {
      "epoch": 2.0115644073241246,
      "grad_norm": 0.48617082834243774,
      "learning_rate": 2.539171603364755e-06,
      "loss": 0.3265,
      "step": 43834
    },
    {
      "epoch": 2.011610297829379,
      "grad_norm": 0.502138614654541,
      "learning_rate": 2.5389581698710607e-06,
      "loss": 0.342,
      "step": 43835
    },
    {
      "epoch": 2.0116561883346336,
      "grad_norm": 0.508787989616394,
      "learning_rate": 2.5387447422953932e-06,
      "loss": 0.3159,
      "step": 43836
    },
    {
      "epoch": 2.011702078839888,
      "grad_norm": 0.4687052369117737,
      "learning_rate": 2.53853132063827e-06,
      "loss": 0.3078,
      "step": 43837
    },
    {
      "epoch": 2.0117479693451426,
      "grad_norm": 0.4818289577960968,
      "learning_rate": 2.5383179049002017e-06,
      "loss": 0.3036,
      "step": 43838
    },
    {
      "epoch": 2.011793859850397,
      "grad_norm": 0.48350799083709717,
      "learning_rate": 2.5381044950817025e-06,
      "loss": 0.3078,
      "step": 43839
    },
    {
      "epoch": 2.0118397503556515,
      "grad_norm": 0.4493600130081177,
      "learning_rate": 2.5378910911832842e-06,
      "loss": 0.2587,
      "step": 43840
    },
    {
      "epoch": 2.011885640860906,
      "grad_norm": 0.4948440492153168,
      "learning_rate": 2.5376776932054615e-06,
      "loss": 0.3782,
      "step": 43841
    },
    {
      "epoch": 2.0119315313661605,
      "grad_norm": 0.5043162703514099,
      "learning_rate": 2.5374643011487464e-06,
      "loss": 0.3751,
      "step": 43842
    },
    {
      "epoch": 2.0119774218714146,
      "grad_norm": 0.4890897274017334,
      "learning_rate": 2.537250915013651e-06,
      "loss": 0.3541,
      "step": 43843
    },
    {
      "epoch": 2.012023312376669,
      "grad_norm": 0.5071358680725098,
      "learning_rate": 2.5370375348006914e-06,
      "loss": 0.3131,
      "step": 43844
    },
    {
      "epoch": 2.0120692028819236,
      "grad_norm": 0.45573458075523376,
      "learning_rate": 2.5368241605103793e-06,
      "loss": 0.2908,
      "step": 43845
    },
    {
      "epoch": 2.012115093387178,
      "grad_norm": 0.4748174250125885,
      "learning_rate": 2.536610792143226e-06,
      "loss": 0.3263,
      "step": 43846
    },
    {
      "epoch": 2.0121609838924326,
      "grad_norm": 0.4881909191608429,
      "learning_rate": 2.5363974296997473e-06,
      "loss": 0.3159,
      "step": 43847
    },
    {
      "epoch": 2.012206874397687,
      "grad_norm": 0.4843316078186035,
      "learning_rate": 2.5361840731804567e-06,
      "loss": 0.3326,
      "step": 43848
    },
    {
      "epoch": 2.0122527649029416,
      "grad_norm": 0.4606104791164398,
      "learning_rate": 2.535970722585864e-06,
      "loss": 0.2609,
      "step": 43849
    },
    {
      "epoch": 2.012298655408196,
      "grad_norm": 0.468860000371933,
      "learning_rate": 2.5357573779164856e-06,
      "loss": 0.2854,
      "step": 43850
    },
    {
      "epoch": 2.0123445459134506,
      "grad_norm": 0.46169668436050415,
      "learning_rate": 2.5355440391728327e-06,
      "loss": 0.3035,
      "step": 43851
    },
    {
      "epoch": 2.012390436418705,
      "grad_norm": 0.4977363348007202,
      "learning_rate": 2.535330706355416e-06,
      "loss": 0.3351,
      "step": 43852
    },
    {
      "epoch": 2.0124363269239596,
      "grad_norm": 0.5179955363273621,
      "learning_rate": 2.535117379464753e-06,
      "loss": 0.3875,
      "step": 43853
    },
    {
      "epoch": 2.012482217429214,
      "grad_norm": 0.5369314551353455,
      "learning_rate": 2.5349040585013556e-06,
      "loss": 0.3844,
      "step": 43854
    },
    {
      "epoch": 2.0125281079344686,
      "grad_norm": 0.501923680305481,
      "learning_rate": 2.5346907434657335e-06,
      "loss": 0.3598,
      "step": 43855
    },
    {
      "epoch": 2.0125739984397226,
      "grad_norm": 0.44475021958351135,
      "learning_rate": 2.5344774343584043e-06,
      "loss": 0.2731,
      "step": 43856
    },
    {
      "epoch": 2.012619888944977,
      "grad_norm": 0.487800270318985,
      "learning_rate": 2.534264131179879e-06,
      "loss": 0.3749,
      "step": 43857
    },
    {
      "epoch": 2.0126657794502316,
      "grad_norm": 0.48556065559387207,
      "learning_rate": 2.5340508339306667e-06,
      "loss": 0.2866,
      "step": 43858
    },
    {
      "epoch": 2.012711669955486,
      "grad_norm": 0.5159242153167725,
      "learning_rate": 2.533837542611287e-06,
      "loss": 0.3447,
      "step": 43859
    },
    {
      "epoch": 2.0127575604607406,
      "grad_norm": 0.5006429553031921,
      "learning_rate": 2.5336242572222504e-06,
      "loss": 0.2717,
      "step": 43860
    },
    {
      "epoch": 2.012803450965995,
      "grad_norm": 0.4680998921394348,
      "learning_rate": 2.5334109777640683e-06,
      "loss": 0.3283,
      "step": 43861
    },
    {
      "epoch": 2.0128493414712496,
      "grad_norm": 0.5020669102668762,
      "learning_rate": 2.5331977042372534e-06,
      "loss": 0.2891,
      "step": 43862
    },
    {
      "epoch": 2.012895231976504,
      "grad_norm": 0.4863584339618683,
      "learning_rate": 2.5329844366423208e-06,
      "loss": 0.3145,
      "step": 43863
    },
    {
      "epoch": 2.0129411224817586,
      "grad_norm": 0.45392343401908875,
      "learning_rate": 2.5327711749797795e-06,
      "loss": 0.2524,
      "step": 43864
    },
    {
      "epoch": 2.012987012987013,
      "grad_norm": 0.47296059131622314,
      "learning_rate": 2.532557919250147e-06,
      "loss": 0.3104,
      "step": 43865
    },
    {
      "epoch": 2.0130329034922676,
      "grad_norm": 0.4889378845691681,
      "learning_rate": 2.532344669453934e-06,
      "loss": 0.2958,
      "step": 43866
    },
    {
      "epoch": 2.013078793997522,
      "grad_norm": 0.5107224583625793,
      "learning_rate": 2.5321314255916506e-06,
      "loss": 0.3204,
      "step": 43867
    },
    {
      "epoch": 2.0131246845027766,
      "grad_norm": 0.453745037317276,
      "learning_rate": 2.5319181876638153e-06,
      "loss": 0.2512,
      "step": 43868
    },
    {
      "epoch": 2.0131705750080306,
      "grad_norm": 0.4848005473613739,
      "learning_rate": 2.531704955670937e-06,
      "loss": 0.3286,
      "step": 43869
    },
    {
      "epoch": 2.013216465513285,
      "grad_norm": 0.4894874095916748,
      "learning_rate": 2.5314917296135278e-06,
      "loss": 0.327,
      "step": 43870
    },
    {
      "epoch": 2.0132623560185396,
      "grad_norm": 0.5080018043518066,
      "learning_rate": 2.5312785094921055e-06,
      "loss": 0.366,
      "step": 43871
    },
    {
      "epoch": 2.013308246523794,
      "grad_norm": 0.475763201713562,
      "learning_rate": 2.5310652953071776e-06,
      "loss": 0.343,
      "step": 43872
    },
    {
      "epoch": 2.0133541370290486,
      "grad_norm": 0.4893595278263092,
      "learning_rate": 2.5308520870592577e-06,
      "loss": 0.3423,
      "step": 43873
    },
    {
      "epoch": 2.013400027534303,
      "grad_norm": 0.4915807247161865,
      "learning_rate": 2.5306388847488584e-06,
      "loss": 0.3504,
      "step": 43874
    },
    {
      "epoch": 2.0134459180395576,
      "grad_norm": 0.5083504915237427,
      "learning_rate": 2.5304256883764942e-06,
      "loss": 0.3468,
      "step": 43875
    },
    {
      "epoch": 2.013491808544812,
      "grad_norm": 0.5098035335540771,
      "learning_rate": 2.530212497942677e-06,
      "loss": 0.3424,
      "step": 43876
    },
    {
      "epoch": 2.0135376990500666,
      "grad_norm": 0.49076545238494873,
      "learning_rate": 2.5299993134479177e-06,
      "loss": 0.313,
      "step": 43877
    },
    {
      "epoch": 2.013583589555321,
      "grad_norm": 0.4642910361289978,
      "learning_rate": 2.5297861348927323e-06,
      "loss": 0.3071,
      "step": 43878
    },
    {
      "epoch": 2.0136294800605756,
      "grad_norm": 0.4723716080188751,
      "learning_rate": 2.5295729622776314e-06,
      "loss": 0.2906,
      "step": 43879
    },
    {
      "epoch": 2.01367537056583,
      "grad_norm": 0.5080165266990662,
      "learning_rate": 2.5293597956031262e-06,
      "loss": 0.3599,
      "step": 43880
    },
    {
      "epoch": 2.013721261071084,
      "grad_norm": 0.49046531319618225,
      "learning_rate": 2.5291466348697326e-06,
      "loss": 0.3306,
      "step": 43881
    },
    {
      "epoch": 2.0137671515763387,
      "grad_norm": 0.49121373891830444,
      "learning_rate": 2.5289334800779617e-06,
      "loss": 0.333,
      "step": 43882
    },
    {
      "epoch": 2.013813042081593,
      "grad_norm": 0.48879274725914,
      "learning_rate": 2.528720331228326e-06,
      "loss": 0.3579,
      "step": 43883
    },
    {
      "epoch": 2.0138589325868477,
      "grad_norm": 0.4825191795825958,
      "learning_rate": 2.5285071883213375e-06,
      "loss": 0.321,
      "step": 43884
    },
    {
      "epoch": 2.013904823092102,
      "grad_norm": 0.44309690594673157,
      "learning_rate": 2.5282940513575094e-06,
      "loss": 0.2659,
      "step": 43885
    },
    {
      "epoch": 2.0139507135973567,
      "grad_norm": 0.4973062574863434,
      "learning_rate": 2.5280809203373523e-06,
      "loss": 0.3438,
      "step": 43886
    },
    {
      "epoch": 2.013996604102611,
      "grad_norm": 0.4765530526638031,
      "learning_rate": 2.5278677952613827e-06,
      "loss": 0.3004,
      "step": 43887
    },
    {
      "epoch": 2.0140424946078657,
      "grad_norm": 0.49563929438591003,
      "learning_rate": 2.52765467613011e-06,
      "loss": 0.3365,
      "step": 43888
    },
    {
      "epoch": 2.01408838511312,
      "grad_norm": 0.4780856966972351,
      "learning_rate": 2.527441562944046e-06,
      "loss": 0.3397,
      "step": 43889
    },
    {
      "epoch": 2.0141342756183747,
      "grad_norm": 0.49740245938301086,
      "learning_rate": 2.527228455703706e-06,
      "loss": 0.339,
      "step": 43890
    },
    {
      "epoch": 2.014180166123629,
      "grad_norm": 0.49825698137283325,
      "learning_rate": 2.527015354409602e-06,
      "loss": 0.317,
      "step": 43891
    },
    {
      "epoch": 2.0142260566288837,
      "grad_norm": 0.5217834115028381,
      "learning_rate": 2.5268022590622433e-06,
      "loss": 0.3928,
      "step": 43892
    },
    {
      "epoch": 2.014271947134138,
      "grad_norm": 0.44545963406562805,
      "learning_rate": 2.5265891696621486e-06,
      "loss": 0.2666,
      "step": 43893
    },
    {
      "epoch": 2.014317837639392,
      "grad_norm": 0.4696104824542999,
      "learning_rate": 2.526376086209824e-06,
      "loss": 0.3241,
      "step": 43894
    },
    {
      "epoch": 2.0143637281446467,
      "grad_norm": 0.4544749855995178,
      "learning_rate": 2.526163008705783e-06,
      "loss": 0.308,
      "step": 43895
    },
    {
      "epoch": 2.014409618649901,
      "grad_norm": 0.43423962593078613,
      "learning_rate": 2.5259499371505402e-06,
      "loss": 0.2677,
      "step": 43896
    },
    {
      "epoch": 2.0144555091551557,
      "grad_norm": 0.46652644872665405,
      "learning_rate": 2.5257368715446073e-06,
      "loss": 0.277,
      "step": 43897
    },
    {
      "epoch": 2.01450139966041,
      "grad_norm": 0.4934419095516205,
      "learning_rate": 2.525523811888495e-06,
      "loss": 0.3391,
      "step": 43898
    },
    {
      "epoch": 2.0145472901656647,
      "grad_norm": 0.47350746393203735,
      "learning_rate": 2.5253107581827186e-06,
      "loss": 0.2624,
      "step": 43899
    },
    {
      "epoch": 2.014593180670919,
      "grad_norm": 0.49032244086265564,
      "learning_rate": 2.5250977104277886e-06,
      "loss": 0.3738,
      "step": 43900
    },
    {
      "epoch": 2.0146390711761737,
      "grad_norm": 0.472483366727829,
      "learning_rate": 2.5248846686242155e-06,
      "loss": 0.2978,
      "step": 43901
    },
    {
      "epoch": 2.014684961681428,
      "grad_norm": 0.5161455273628235,
      "learning_rate": 2.524671632772515e-06,
      "loss": 0.3415,
      "step": 43902
    },
    {
      "epoch": 2.0147308521866827,
      "grad_norm": 0.4800645112991333,
      "learning_rate": 2.5244586028731987e-06,
      "loss": 0.3612,
      "step": 43903
    },
    {
      "epoch": 2.014776742691937,
      "grad_norm": 0.5013259053230286,
      "learning_rate": 2.5242455789267784e-06,
      "loss": 0.3477,
      "step": 43904
    },
    {
      "epoch": 2.0148226331971917,
      "grad_norm": 0.46954140067100525,
      "learning_rate": 2.5240325609337656e-06,
      "loss": 0.3325,
      "step": 43905
    },
    {
      "epoch": 2.014868523702446,
      "grad_norm": 0.4729161560535431,
      "learning_rate": 2.5238195488946735e-06,
      "loss": 0.3248,
      "step": 43906
    },
    {
      "epoch": 2.0149144142077002,
      "grad_norm": 0.500304639339447,
      "learning_rate": 2.523606542810014e-06,
      "loss": 0.4027,
      "step": 43907
    },
    {
      "epoch": 2.0149603047129547,
      "grad_norm": 0.4824456572532654,
      "learning_rate": 2.523393542680297e-06,
      "loss": 0.3178,
      "step": 43908
    },
    {
      "epoch": 2.0150061952182092,
      "grad_norm": 0.4339597821235657,
      "learning_rate": 2.523180548506038e-06,
      "loss": 0.2756,
      "step": 43909
    },
    {
      "epoch": 2.0150520857234637,
      "grad_norm": 0.47673264145851135,
      "learning_rate": 2.522967560287749e-06,
      "loss": 0.2945,
      "step": 43910
    },
    {
      "epoch": 2.0150979762287182,
      "grad_norm": 0.5147387385368347,
      "learning_rate": 2.5227545780259387e-06,
      "loss": 0.3727,
      "step": 43911
    },
    {
      "epoch": 2.0151438667339727,
      "grad_norm": 0.4680587649345398,
      "learning_rate": 2.5225416017211245e-06,
      "loss": 0.2979,
      "step": 43912
    },
    {
      "epoch": 2.015189757239227,
      "grad_norm": 0.5131319165229797,
      "learning_rate": 2.5223286313738153e-06,
      "loss": 0.3345,
      "step": 43913
    },
    {
      "epoch": 2.0152356477444817,
      "grad_norm": 0.49784141778945923,
      "learning_rate": 2.522115666984522e-06,
      "loss": 0.3742,
      "step": 43914
    },
    {
      "epoch": 2.015281538249736,
      "grad_norm": 0.48749786615371704,
      "learning_rate": 2.5219027085537626e-06,
      "loss": 0.3239,
      "step": 43915
    },
    {
      "epoch": 2.0153274287549907,
      "grad_norm": 0.42005592584609985,
      "learning_rate": 2.5216897560820432e-06,
      "loss": 0.2312,
      "step": 43916
    },
    {
      "epoch": 2.015373319260245,
      "grad_norm": 0.48870399594306946,
      "learning_rate": 2.521476809569875e-06,
      "loss": 0.3251,
      "step": 43917
    },
    {
      "epoch": 2.0154192097654997,
      "grad_norm": 0.5034509301185608,
      "learning_rate": 2.5212638690177754e-06,
      "loss": 0.4244,
      "step": 43918
    },
    {
      "epoch": 2.0154651002707538,
      "grad_norm": 0.4646150469779968,
      "learning_rate": 2.5210509344262536e-06,
      "loss": 0.3287,
      "step": 43919
    },
    {
      "epoch": 2.0155109907760083,
      "grad_norm": 0.45048967003822327,
      "learning_rate": 2.5208380057958203e-06,
      "loss": 0.3104,
      "step": 43920
    },
    {
      "epoch": 2.0155568812812628,
      "grad_norm": 0.46446090936660767,
      "learning_rate": 2.5206250831269907e-06,
      "loss": 0.305,
      "step": 43921
    },
    {
      "epoch": 2.0156027717865173,
      "grad_norm": 0.49919232726097107,
      "learning_rate": 2.5204121664202752e-06,
      "loss": 0.3233,
      "step": 43922
    },
    {
      "epoch": 2.0156486622917718,
      "grad_norm": 0.4784594476222992,
      "learning_rate": 2.5201992556761847e-06,
      "loss": 0.3186,
      "step": 43923
    },
    {
      "epoch": 2.0156945527970263,
      "grad_norm": 0.45333245396614075,
      "learning_rate": 2.5199863508952336e-06,
      "loss": 0.2998,
      "step": 43924
    },
    {
      "epoch": 2.0157404433022807,
      "grad_norm": 0.483097106218338,
      "learning_rate": 2.519773452077933e-06,
      "loss": 0.3086,
      "step": 43925
    },
    {
      "epoch": 2.0157863338075352,
      "grad_norm": 0.49029210209846497,
      "learning_rate": 2.5195605592247942e-06,
      "loss": 0.3292,
      "step": 43926
    },
    {
      "epoch": 2.0158322243127897,
      "grad_norm": 0.45336493849754333,
      "learning_rate": 2.5193476723363296e-06,
      "loss": 0.3407,
      "step": 43927
    },
    {
      "epoch": 2.0158781148180442,
      "grad_norm": 0.4453563392162323,
      "learning_rate": 2.5191347914130504e-06,
      "loss": 0.2821,
      "step": 43928
    },
    {
      "epoch": 2.0159240053232987,
      "grad_norm": 0.48070886731147766,
      "learning_rate": 2.518921916455468e-06,
      "loss": 0.2868,
      "step": 43929
    },
    {
      "epoch": 2.0159698958285532,
      "grad_norm": 0.4838741719722748,
      "learning_rate": 2.518709047464096e-06,
      "loss": 0.3588,
      "step": 43930
    },
    {
      "epoch": 2.0160157863338077,
      "grad_norm": 0.4852774739265442,
      "learning_rate": 2.518496184439447e-06,
      "loss": 0.3467,
      "step": 43931
    },
    {
      "epoch": 2.016061676839062,
      "grad_norm": 0.49208828806877136,
      "learning_rate": 2.518283327382029e-06,
      "loss": 0.3444,
      "step": 43932
    },
    {
      "epoch": 2.0161075673443163,
      "grad_norm": 0.507270097732544,
      "learning_rate": 2.5180704762923582e-06,
      "loss": 0.3131,
      "step": 43933
    },
    {
      "epoch": 2.016153457849571,
      "grad_norm": 0.4793279767036438,
      "learning_rate": 2.5178576311709446e-06,
      "loss": 0.3614,
      "step": 43934
    },
    {
      "epoch": 2.0161993483548253,
      "grad_norm": 0.47131043672561646,
      "learning_rate": 2.5176447920182977e-06,
      "loss": 0.3251,
      "step": 43935
    },
    {
      "epoch": 2.01624523886008,
      "grad_norm": 0.473725289106369,
      "learning_rate": 2.5174319588349343e-06,
      "loss": 0.3014,
      "step": 43936
    },
    {
      "epoch": 2.0162911293653343,
      "grad_norm": 0.4771834909915924,
      "learning_rate": 2.517219131621365e-06,
      "loss": 0.3364,
      "step": 43937
    },
    {
      "epoch": 2.0163370198705888,
      "grad_norm": 0.4983333945274353,
      "learning_rate": 2.5170063103780986e-06,
      "loss": 0.3333,
      "step": 43938
    },
    {
      "epoch": 2.0163829103758433,
      "grad_norm": 0.4583309590816498,
      "learning_rate": 2.516793495105646e-06,
      "loss": 0.2918,
      "step": 43939
    },
    {
      "epoch": 2.0164288008810978,
      "grad_norm": 0.4522255063056946,
      "learning_rate": 2.5165806858045232e-06,
      "loss": 0.2851,
      "step": 43940
    },
    {
      "epoch": 2.0164746913863523,
      "grad_norm": 0.5120725631713867,
      "learning_rate": 2.51636788247524e-06,
      "loss": 0.3627,
      "step": 43941
    },
    {
      "epoch": 2.0165205818916068,
      "grad_norm": 0.476701945066452,
      "learning_rate": 2.516155085118307e-06,
      "loss": 0.2979,
      "step": 43942
    },
    {
      "epoch": 2.0165664723968613,
      "grad_norm": 0.4793146848678589,
      "learning_rate": 2.515942293734238e-06,
      "loss": 0.3165,
      "step": 43943
    },
    {
      "epoch": 2.0166123629021158,
      "grad_norm": 0.47328636050224304,
      "learning_rate": 2.5157295083235444e-06,
      "loss": 0.3147,
      "step": 43944
    },
    {
      "epoch": 2.01665825340737,
      "grad_norm": 0.4885639548301697,
      "learning_rate": 2.5155167288867355e-06,
      "loss": 0.3611,
      "step": 43945
    },
    {
      "epoch": 2.0167041439126243,
      "grad_norm": 0.5268436670303345,
      "learning_rate": 2.515303955424326e-06,
      "loss": 0.4061,
      "step": 43946
    },
    {
      "epoch": 2.016750034417879,
      "grad_norm": 0.4444359838962555,
      "learning_rate": 2.5150911879368267e-06,
      "loss": 0.2938,
      "step": 43947
    },
    {
      "epoch": 2.0167959249231333,
      "grad_norm": 0.4631958305835724,
      "learning_rate": 2.5148784264247495e-06,
      "loss": 0.2942,
      "step": 43948
    },
    {
      "epoch": 2.016841815428388,
      "grad_norm": 0.47893357276916504,
      "learning_rate": 2.514665670888604e-06,
      "loss": 0.2898,
      "step": 43949
    },
    {
      "epoch": 2.0168877059336423,
      "grad_norm": 0.5468403100967407,
      "learning_rate": 2.514452921328904e-06,
      "loss": 0.386,
      "step": 43950
    },
    {
      "epoch": 2.016933596438897,
      "grad_norm": 0.48551762104034424,
      "learning_rate": 2.5142401777461585e-06,
      "loss": 0.3431,
      "step": 43951
    },
    {
      "epoch": 2.0169794869441513,
      "grad_norm": 0.46854326128959656,
      "learning_rate": 2.5140274401408828e-06,
      "loss": 0.3349,
      "step": 43952
    },
    {
      "epoch": 2.017025377449406,
      "grad_norm": 0.43519696593284607,
      "learning_rate": 2.5138147085135857e-06,
      "loss": 0.2232,
      "step": 43953
    },
    {
      "epoch": 2.0170712679546603,
      "grad_norm": 0.5051290392875671,
      "learning_rate": 2.5136019828647784e-06,
      "loss": 0.3881,
      "step": 43954
    },
    {
      "epoch": 2.017117158459915,
      "grad_norm": 0.5379789471626282,
      "learning_rate": 2.513389263194975e-06,
      "loss": 0.383,
      "step": 43955
    },
    {
      "epoch": 2.0171630489651693,
      "grad_norm": 0.49520957469940186,
      "learning_rate": 2.5131765495046854e-06,
      "loss": 0.3462,
      "step": 43956
    },
    {
      "epoch": 2.0172089394704233,
      "grad_norm": 0.46914055943489075,
      "learning_rate": 2.5129638417944203e-06,
      "loss": 0.3065,
      "step": 43957
    },
    {
      "epoch": 2.017254829975678,
      "grad_norm": 0.462139755487442,
      "learning_rate": 2.5127511400646944e-06,
      "loss": 0.2956,
      "step": 43958
    },
    {
      "epoch": 2.0173007204809323,
      "grad_norm": 0.4541197121143341,
      "learning_rate": 2.5125384443160176e-06,
      "loss": 0.2667,
      "step": 43959
    },
    {
      "epoch": 2.017346610986187,
      "grad_norm": 0.467947781085968,
      "learning_rate": 2.5123257545488978e-06,
      "loss": 0.3083,
      "step": 43960
    },
    {
      "epoch": 2.0173925014914413,
      "grad_norm": 0.45543619990348816,
      "learning_rate": 2.5121130707638507e-06,
      "loss": 0.2882,
      "step": 43961
    },
    {
      "epoch": 2.017438391996696,
      "grad_norm": 0.4957212209701538,
      "learning_rate": 2.5119003929613872e-06,
      "loss": 0.3748,
      "step": 43962
    },
    {
      "epoch": 2.0174842825019503,
      "grad_norm": 0.4610357880592346,
      "learning_rate": 2.511687721142016e-06,
      "loss": 0.276,
      "step": 43963
    },
    {
      "epoch": 2.017530173007205,
      "grad_norm": 0.4969644844532013,
      "learning_rate": 2.511475055306252e-06,
      "loss": 0.3201,
      "step": 43964
    },
    {
      "epoch": 2.0175760635124593,
      "grad_norm": 0.48030999302864075,
      "learning_rate": 2.5112623954546056e-06,
      "loss": 0.3221,
      "step": 43965
    },
    {
      "epoch": 2.017621954017714,
      "grad_norm": 0.47609302401542664,
      "learning_rate": 2.5110497415875856e-06,
      "loss": 0.3159,
      "step": 43966
    },
    {
      "epoch": 2.0176678445229683,
      "grad_norm": 0.4846997559070587,
      "learning_rate": 2.510837093705707e-06,
      "loss": 0.292,
      "step": 43967
    },
    {
      "epoch": 2.017713735028223,
      "grad_norm": 0.5531144738197327,
      "learning_rate": 2.5106244518094798e-06,
      "loss": 0.3116,
      "step": 43968
    },
    {
      "epoch": 2.0177596255334773,
      "grad_norm": 0.5079323053359985,
      "learning_rate": 2.510411815899415e-06,
      "loss": 0.3403,
      "step": 43969
    },
    {
      "epoch": 2.0178055160387314,
      "grad_norm": 0.5111728310585022,
      "learning_rate": 2.5101991859760242e-06,
      "loss": 0.383,
      "step": 43970
    },
    {
      "epoch": 2.017851406543986,
      "grad_norm": 0.43835774064064026,
      "learning_rate": 2.5099865620398185e-06,
      "loss": 0.2787,
      "step": 43971
    },
    {
      "epoch": 2.0178972970492404,
      "grad_norm": 0.452069491147995,
      "learning_rate": 2.5097739440913093e-06,
      "loss": 0.2785,
      "step": 43972
    },
    {
      "epoch": 2.017943187554495,
      "grad_norm": 0.4641483426094055,
      "learning_rate": 2.509561332131006e-06,
      "loss": 0.2748,
      "step": 43973
    },
    {
      "epoch": 2.0179890780597494,
      "grad_norm": 0.5283618569374084,
      "learning_rate": 2.509348726159423e-06,
      "loss": 0.384,
      "step": 43974
    },
    {
      "epoch": 2.018034968565004,
      "grad_norm": 0.4725908935070038,
      "learning_rate": 2.5091361261770708e-06,
      "loss": 0.3303,
      "step": 43975
    },
    {
      "epoch": 2.0180808590702584,
      "grad_norm": 0.4774918556213379,
      "learning_rate": 2.5089235321844584e-06,
      "loss": 0.3443,
      "step": 43976
    },
    {
      "epoch": 2.018126749575513,
      "grad_norm": 0.4773925244808197,
      "learning_rate": 2.5087109441821e-06,
      "loss": 0.285,
      "step": 43977
    },
    {
      "epoch": 2.0181726400807674,
      "grad_norm": 0.4938560128211975,
      "learning_rate": 2.508498362170506e-06,
      "loss": 0.3341,
      "step": 43978
    },
    {
      "epoch": 2.018218530586022,
      "grad_norm": 0.436746746301651,
      "learning_rate": 2.5082857861501854e-06,
      "loss": 0.2467,
      "step": 43979
    },
    {
      "epoch": 2.0182644210912764,
      "grad_norm": 0.48668918013572693,
      "learning_rate": 2.508073216121652e-06,
      "loss": 0.3313,
      "step": 43980
    },
    {
      "epoch": 2.018310311596531,
      "grad_norm": 0.45892852544784546,
      "learning_rate": 2.5078606520854188e-06,
      "loss": 0.2598,
      "step": 43981
    },
    {
      "epoch": 2.0183562021017853,
      "grad_norm": 0.4988388419151306,
      "learning_rate": 2.50764809404199e-06,
      "loss": 0.3381,
      "step": 43982
    },
    {
      "epoch": 2.0184020926070394,
      "grad_norm": 0.5160019993782043,
      "learning_rate": 2.5074355419918828e-06,
      "loss": 0.3862,
      "step": 43983
    },
    {
      "epoch": 2.018447983112294,
      "grad_norm": 0.4716640114784241,
      "learning_rate": 2.5072229959356063e-06,
      "loss": 0.2799,
      "step": 43984
    },
    {
      "epoch": 2.0184938736175484,
      "grad_norm": 0.4866861402988434,
      "learning_rate": 2.5070104558736707e-06,
      "loss": 0.3226,
      "step": 43985
    },
    {
      "epoch": 2.018539764122803,
      "grad_norm": 0.4635947048664093,
      "learning_rate": 2.506797921806589e-06,
      "loss": 0.3643,
      "step": 43986
    },
    {
      "epoch": 2.0185856546280574,
      "grad_norm": 0.47598734498023987,
      "learning_rate": 2.5065853937348715e-06,
      "loss": 0.33,
      "step": 43987
    },
    {
      "epoch": 2.018631545133312,
      "grad_norm": 0.4753451943397522,
      "learning_rate": 2.5063728716590287e-06,
      "loss": 0.3326,
      "step": 43988
    },
    {
      "epoch": 2.0186774356385664,
      "grad_norm": 0.48025956749916077,
      "learning_rate": 2.5061603555795722e-06,
      "loss": 0.3394,
      "step": 43989
    },
    {
      "epoch": 2.018723326143821,
      "grad_norm": 0.5563918948173523,
      "learning_rate": 2.505947845497014e-06,
      "loss": 0.2202,
      "step": 43990
    },
    {
      "epoch": 2.0187692166490754,
      "grad_norm": 0.49601122736930847,
      "learning_rate": 2.505735341411864e-06,
      "loss": 0.355,
      "step": 43991
    },
    {
      "epoch": 2.01881510715433,
      "grad_norm": 0.4918726980686188,
      "learning_rate": 2.505522843324633e-06,
      "loss": 0.3245,
      "step": 43992
    },
    {
      "epoch": 2.0188609976595844,
      "grad_norm": 0.4753451943397522,
      "learning_rate": 2.5053103512358325e-06,
      "loss": 0.3531,
      "step": 43993
    },
    {
      "epoch": 2.018906888164839,
      "grad_norm": 0.4845462739467621,
      "learning_rate": 2.505097865145971e-06,
      "loss": 0.3318,
      "step": 43994
    },
    {
      "epoch": 2.0189527786700934,
      "grad_norm": 0.49072739481925964,
      "learning_rate": 2.5048853850555643e-06,
      "loss": 0.3097,
      "step": 43995
    },
    {
      "epoch": 2.0189986691753474,
      "grad_norm": 0.5049818158149719,
      "learning_rate": 2.50467291096512e-06,
      "loss": 0.332,
      "step": 43996
    },
    {
      "epoch": 2.019044559680602,
      "grad_norm": 0.49033215641975403,
      "learning_rate": 2.5044604428751483e-06,
      "loss": 0.3206,
      "step": 43997
    },
    {
      "epoch": 2.0190904501858564,
      "grad_norm": 0.45598530769348145,
      "learning_rate": 2.5042479807861635e-06,
      "loss": 0.2627,
      "step": 43998
    },
    {
      "epoch": 2.019136340691111,
      "grad_norm": 0.45843300223350525,
      "learning_rate": 2.504035524698675e-06,
      "loss": 0.2819,
      "step": 43999
    },
    {
      "epoch": 2.0191822311963654,
      "grad_norm": 0.5183503031730652,
      "learning_rate": 2.503823074613191e-06,
      "loss": 0.3487,
      "step": 44000
    }
  ],
  "logging_steps": 1,
  "max_steps": 65373,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 1000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 9.601551861547008e+19,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}